{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "eval_steps": 500, "global_step": 48327, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 6.207709975789931e-05, "grad_norm": 1.4921084877385822, "learning_rate": 2.069108214359611e-08, "loss": 11.9497, "step": 1 }, { "epoch": 0.00012415419951579862, "grad_norm": 1.5655681241910728, "learning_rate": 4.138216428719222e-08, "loss": 11.9501, "step": 2 }, { "epoch": 0.00018623129927369795, "grad_norm": 1.5838071298268475, "learning_rate": 6.207324643078834e-08, "loss": 11.9441, "step": 3 }, { "epoch": 0.00024830839903159724, "grad_norm": 1.5464848498377175, "learning_rate": 8.276432857438443e-08, "loss": 11.954, "step": 4 }, { "epoch": 0.00031038549878949657, "grad_norm": 1.5277707802968699, "learning_rate": 1.0345541071798054e-07, "loss": 11.9506, "step": 5 }, { "epoch": 0.0003724625985473959, "grad_norm": 1.5625742322454526, "learning_rate": 1.2414649286157668e-07, "loss": 11.9481, "step": 6 }, { "epoch": 0.00043453969830529516, "grad_norm": 1.4195963448362139, "learning_rate": 1.448375750051728e-07, "loss": 11.9403, "step": 7 }, { "epoch": 0.0004966167980631945, "grad_norm": 1.541516279497776, "learning_rate": 1.6552865714876887e-07, "loss": 11.959, "step": 8 }, { "epoch": 0.0005586938978210938, "grad_norm": 1.5661094174994818, "learning_rate": 1.86219739292365e-07, "loss": 11.9499, "step": 9 }, { "epoch": 0.0006207709975789931, "grad_norm": 1.6201200571431411, "learning_rate": 2.0691082143596109e-07, "loss": 11.9517, "step": 10 }, { "epoch": 0.0006828480973368925, "grad_norm": 1.5915552622901863, "learning_rate": 2.2760190357955722e-07, "loss": 11.9468, "step": 11 }, { "epoch": 0.0007449251970947918, "grad_norm": 1.5629715780064608, "learning_rate": 2.4829298572315336e-07, "loss": 11.9491, "step": 12 }, { "epoch": 0.000807002296852691, "grad_norm": 1.4619697555095934, "learning_rate": 2.6898406786674944e-07, "loss": 11.9486, "step": 13 }, { "epoch": 0.0008690793966105903, "grad_norm": 1.5142947779561575, "learning_rate": 2.896751500103456e-07, "loss": 11.9485, "step": 14 }, { "epoch": 0.0009311564963684896, "grad_norm": 1.5180403350426919, "learning_rate": 3.103662321539417e-07, "loss": 11.9435, "step": 15 }, { "epoch": 0.000993233596126389, "grad_norm": 1.5623276424712613, "learning_rate": 3.3105731429753774e-07, "loss": 11.9484, "step": 16 }, { "epoch": 0.0010553106958842882, "grad_norm": 1.4937942402565056, "learning_rate": 3.517483964411339e-07, "loss": 11.9547, "step": 17 }, { "epoch": 0.0011173877956421876, "grad_norm": 1.5859898149853626, "learning_rate": 3.7243947858473e-07, "loss": 11.9422, "step": 18 }, { "epoch": 0.0011794648954000868, "grad_norm": 1.5927414040118726, "learning_rate": 3.931305607283261e-07, "loss": 11.944, "step": 19 }, { "epoch": 0.0012415419951579863, "grad_norm": 1.5877107375234834, "learning_rate": 4.1382164287192217e-07, "loss": 11.9458, "step": 20 }, { "epoch": 0.0013036190949158855, "grad_norm": 1.541012530692394, "learning_rate": 4.345127250155183e-07, "loss": 11.9482, "step": 21 }, { "epoch": 0.001365696194673785, "grad_norm": 1.5432976867041488, "learning_rate": 4.5520380715911444e-07, "loss": 11.9455, "step": 22 }, { "epoch": 0.0014277732944316841, "grad_norm": 1.566873989523344, "learning_rate": 4.758948893027105e-07, "loss": 11.9408, "step": 23 }, { "epoch": 0.0014898503941895836, "grad_norm": 1.597730490305004, "learning_rate": 4.965859714463067e-07, "loss": 11.9445, "step": 24 }, { "epoch": 0.0015519274939474828, "grad_norm": 1.595294503195463, "learning_rate": 5.172770535899027e-07, "loss": 11.9372, "step": 25 }, { "epoch": 0.001614004593705382, "grad_norm": 1.458016724504955, "learning_rate": 5.379681357334989e-07, "loss": 11.9465, "step": 26 }, { "epoch": 0.0016760816934632814, "grad_norm": 1.658937109805417, "learning_rate": 5.58659217877095e-07, "loss": 11.9462, "step": 27 }, { "epoch": 0.0017381587932211806, "grad_norm": 1.6020278626732345, "learning_rate": 5.793503000206911e-07, "loss": 11.941, "step": 28 }, { "epoch": 0.00180023589297908, "grad_norm": 1.5286263546414776, "learning_rate": 6.000413821642873e-07, "loss": 11.938, "step": 29 }, { "epoch": 0.0018623129927369793, "grad_norm": 1.506616622183704, "learning_rate": 6.207324643078834e-07, "loss": 11.9373, "step": 30 }, { "epoch": 0.0019243900924948787, "grad_norm": 1.5681913863079902, "learning_rate": 6.414235464514794e-07, "loss": 11.9273, "step": 31 }, { "epoch": 0.001986467192252778, "grad_norm": 1.6622816415667632, "learning_rate": 6.621146285950755e-07, "loss": 11.9283, "step": 32 }, { "epoch": 0.002048544292010677, "grad_norm": 1.4573035100856007, "learning_rate": 6.828057107386716e-07, "loss": 11.9253, "step": 33 }, { "epoch": 0.0021106213917685764, "grad_norm": 1.4629049669690826, "learning_rate": 7.034967928822677e-07, "loss": 11.931, "step": 34 }, { "epoch": 0.002172698491526476, "grad_norm": 1.606547566322026, "learning_rate": 7.241878750258639e-07, "loss": 11.9226, "step": 35 }, { "epoch": 0.0022347755912843752, "grad_norm": 1.6199959220363953, "learning_rate": 7.4487895716946e-07, "loss": 11.9198, "step": 36 }, { "epoch": 0.0022968526910422745, "grad_norm": 1.6239819638937263, "learning_rate": 7.655700393130562e-07, "loss": 11.9217, "step": 37 }, { "epoch": 0.0023589297908001737, "grad_norm": 1.6583320199140144, "learning_rate": 7.862611214566522e-07, "loss": 11.9193, "step": 38 }, { "epoch": 0.0024210068905580733, "grad_norm": 1.4156283986975016, "learning_rate": 8.069522036002483e-07, "loss": 11.9228, "step": 39 }, { "epoch": 0.0024830839903159725, "grad_norm": 1.5275442474749379, "learning_rate": 8.276432857438443e-07, "loss": 11.9109, "step": 40 }, { "epoch": 0.0025451610900738717, "grad_norm": 1.5298238068150434, "learning_rate": 8.483343678874406e-07, "loss": 11.8865, "step": 41 }, { "epoch": 0.002607238189831771, "grad_norm": 1.5904883125496954, "learning_rate": 8.690254500310366e-07, "loss": 11.8771, "step": 42 }, { "epoch": 0.00266931528958967, "grad_norm": 1.4972939719231972, "learning_rate": 8.897165321746329e-07, "loss": 11.8755, "step": 43 }, { "epoch": 0.00273139238934757, "grad_norm": 1.59001192520276, "learning_rate": 9.104076143182289e-07, "loss": 11.8714, "step": 44 }, { "epoch": 0.002793469489105469, "grad_norm": 1.6577359137790204, "learning_rate": 9.31098696461825e-07, "loss": 11.8599, "step": 45 }, { "epoch": 0.0028555465888633683, "grad_norm": 1.6378488212798512, "learning_rate": 9.51789778605421e-07, "loss": 11.8602, "step": 46 }, { "epoch": 0.0029176236886212675, "grad_norm": 1.501211630706496, "learning_rate": 9.724808607490173e-07, "loss": 11.861, "step": 47 }, { "epoch": 0.002979700788379167, "grad_norm": 1.549999587766531, "learning_rate": 9.931719428926134e-07, "loss": 11.868, "step": 48 }, { "epoch": 0.0030417778881370663, "grad_norm": 1.5864341221358524, "learning_rate": 1.0138630250362094e-06, "loss": 11.855, "step": 49 }, { "epoch": 0.0031038549878949656, "grad_norm": 1.5524795419079127, "learning_rate": 1.0345541071798055e-06, "loss": 11.8497, "step": 50 }, { "epoch": 0.0031659320876528648, "grad_norm": 1.5637853294912885, "learning_rate": 1.0552451893234016e-06, "loss": 11.846, "step": 51 }, { "epoch": 0.003228009187410764, "grad_norm": 1.656959885387343, "learning_rate": 1.0759362714669978e-06, "loss": 11.8384, "step": 52 }, { "epoch": 0.0032900862871686636, "grad_norm": 1.5394882929902878, "learning_rate": 1.0966273536105939e-06, "loss": 11.8461, "step": 53 }, { "epoch": 0.003352163386926563, "grad_norm": 1.4955475210931615, "learning_rate": 1.11731843575419e-06, "loss": 11.8348, "step": 54 }, { "epoch": 0.003414240486684462, "grad_norm": 1.6056522162459475, "learning_rate": 1.1380095178977862e-06, "loss": 11.8255, "step": 55 }, { "epoch": 0.0034763175864423613, "grad_norm": 1.5471500190782448, "learning_rate": 1.1587006000413823e-06, "loss": 11.8016, "step": 56 }, { "epoch": 0.003538394686200261, "grad_norm": 1.5933794825080778, "learning_rate": 1.1793916821849782e-06, "loss": 11.7364, "step": 57 }, { "epoch": 0.00360047178595816, "grad_norm": 1.579945184988118, "learning_rate": 1.2000827643285746e-06, "loss": 11.7092, "step": 58 }, { "epoch": 0.0036625488857160594, "grad_norm": 1.6639072542799183, "learning_rate": 1.2207738464721705e-06, "loss": 11.6723, "step": 59 }, { "epoch": 0.0037246259854739586, "grad_norm": 1.5353331330487388, "learning_rate": 1.2414649286157668e-06, "loss": 11.6875, "step": 60 }, { "epoch": 0.003786703085231858, "grad_norm": 1.4948818786549596, "learning_rate": 1.2621560107593628e-06, "loss": 11.6876, "step": 61 }, { "epoch": 0.0038487801849897574, "grad_norm": 1.6576428584563703, "learning_rate": 1.282847092902959e-06, "loss": 11.6415, "step": 62 }, { "epoch": 0.003910857284747656, "grad_norm": 1.5272170694797058, "learning_rate": 1.303538175046555e-06, "loss": 11.6644, "step": 63 }, { "epoch": 0.003972934384505556, "grad_norm": 1.5448684779607542, "learning_rate": 1.324229257190151e-06, "loss": 11.6415, "step": 64 }, { "epoch": 0.0040350114842634555, "grad_norm": 1.5253050483407489, "learning_rate": 1.3449203393337473e-06, "loss": 11.6516, "step": 65 }, { "epoch": 0.004097088584021354, "grad_norm": 1.5332683573652826, "learning_rate": 1.3656114214773432e-06, "loss": 11.6544, "step": 66 }, { "epoch": 0.004159165683779254, "grad_norm": 1.6409905162725529, "learning_rate": 1.3863025036209396e-06, "loss": 11.6067, "step": 67 }, { "epoch": 0.004221242783537153, "grad_norm": 1.6486981081948666, "learning_rate": 1.4069935857645355e-06, "loss": 11.6113, "step": 68 }, { "epoch": 0.004283319883295052, "grad_norm": 1.6750733629928216, "learning_rate": 1.4276846679081316e-06, "loss": 11.5962, "step": 69 }, { "epoch": 0.004345396983052952, "grad_norm": 1.6682088710357417, "learning_rate": 1.4483757500517278e-06, "loss": 11.5914, "step": 70 }, { "epoch": 0.004407474082810851, "grad_norm": 1.6518084874295467, "learning_rate": 1.469066832195324e-06, "loss": 11.5663, "step": 71 }, { "epoch": 0.0044695511825687505, "grad_norm": 1.775960463590703, "learning_rate": 1.48975791433892e-06, "loss": 11.5208, "step": 72 }, { "epoch": 0.00453162828232665, "grad_norm": 1.6911002712693244, "learning_rate": 1.5104489964825162e-06, "loss": 11.5319, "step": 73 }, { "epoch": 0.004593705382084549, "grad_norm": 1.6783342454763217, "learning_rate": 1.5311400786261123e-06, "loss": 11.5131, "step": 74 }, { "epoch": 0.0046557824818424486, "grad_norm": 1.6935683740734364, "learning_rate": 1.5518311607697082e-06, "loss": 11.5012, "step": 75 }, { "epoch": 0.004717859581600347, "grad_norm": 1.6936610743978324, "learning_rate": 1.5725222429133044e-06, "loss": 11.4528, "step": 76 }, { "epoch": 0.004779936681358247, "grad_norm": 1.6828876616712773, "learning_rate": 1.5932133250569005e-06, "loss": 11.4099, "step": 77 }, { "epoch": 0.004842013781116147, "grad_norm": 1.6709982453707437, "learning_rate": 1.6139044072004966e-06, "loss": 11.3569, "step": 78 }, { "epoch": 0.004904090880874045, "grad_norm": 1.6135136249913842, "learning_rate": 1.634595489344093e-06, "loss": 11.3144, "step": 79 }, { "epoch": 0.004966167980631945, "grad_norm": 1.5714592992577419, "learning_rate": 1.6552865714876887e-06, "loss": 11.2799, "step": 80 }, { "epoch": 0.005028245080389844, "grad_norm": 1.4787054103858366, "learning_rate": 1.675977653631285e-06, "loss": 11.2509, "step": 81 }, { "epoch": 0.0050903221801477435, "grad_norm": 1.430687226508294, "learning_rate": 1.6966687357748812e-06, "loss": 11.237, "step": 82 }, { "epoch": 0.005152399279905643, "grad_norm": 1.4970394163682386, "learning_rate": 1.7173598179184773e-06, "loss": 11.1567, "step": 83 }, { "epoch": 0.005214476379663542, "grad_norm": 1.413500173470885, "learning_rate": 1.7380509000620732e-06, "loss": 11.1759, "step": 84 }, { "epoch": 0.005276553479421442, "grad_norm": 1.3616188647370902, "learning_rate": 1.7587419822056694e-06, "loss": 11.1942, "step": 85 }, { "epoch": 0.00533863057917934, "grad_norm": 1.328755128263839, "learning_rate": 1.7794330643492657e-06, "loss": 11.15, "step": 86 }, { "epoch": 0.00540070767893724, "grad_norm": 1.3794370931625026, "learning_rate": 1.8001241464928614e-06, "loss": 11.1046, "step": 87 }, { "epoch": 0.00546278477869514, "grad_norm": 1.307458413522168, "learning_rate": 1.8208152286364578e-06, "loss": 11.1423, "step": 88 }, { "epoch": 0.0055248618784530384, "grad_norm": 1.3003316107869485, "learning_rate": 1.841506310780054e-06, "loss": 11.0795, "step": 89 }, { "epoch": 0.005586938978210938, "grad_norm": 1.319795857307186, "learning_rate": 1.86219739292365e-06, "loss": 11.0402, "step": 90 }, { "epoch": 0.005649016077968837, "grad_norm": 1.321043590854474, "learning_rate": 1.882888475067246e-06, "loss": 11.0182, "step": 91 }, { "epoch": 0.0057110931777267365, "grad_norm": 1.2859661483818055, "learning_rate": 1.903579557210842e-06, "loss": 11.0906, "step": 92 }, { "epoch": 0.005773170277484636, "grad_norm": 1.2648891146313685, "learning_rate": 1.9242706393544382e-06, "loss": 11.0525, "step": 93 }, { "epoch": 0.005835247377242535, "grad_norm": 1.230436585021065, "learning_rate": 1.9449617214980346e-06, "loss": 10.9866, "step": 94 }, { "epoch": 0.005897324477000435, "grad_norm": 1.2293366083027735, "learning_rate": 1.9656528036416305e-06, "loss": 11.0117, "step": 95 }, { "epoch": 0.005959401576758334, "grad_norm": 1.2458137508541642, "learning_rate": 1.986343885785227e-06, "loss": 10.9874, "step": 96 }, { "epoch": 0.006021478676516233, "grad_norm": 1.254896159416497, "learning_rate": 2.0070349679288228e-06, "loss": 10.9154, "step": 97 }, { "epoch": 0.006083555776274133, "grad_norm": 1.2259043488893138, "learning_rate": 2.0277260500724187e-06, "loss": 10.9712, "step": 98 }, { "epoch": 0.0061456328760320315, "grad_norm": 1.2155564845372364, "learning_rate": 2.048417132216015e-06, "loss": 10.8863, "step": 99 }, { "epoch": 0.006207709975789931, "grad_norm": 1.1470109032708844, "learning_rate": 2.069108214359611e-06, "loss": 10.903, "step": 100 }, { "epoch": 0.006269787075547831, "grad_norm": 1.2053151199039374, "learning_rate": 2.0897992965032073e-06, "loss": 10.8867, "step": 101 }, { "epoch": 0.0063318641753057296, "grad_norm": 1.2172539160338953, "learning_rate": 2.1104903786468032e-06, "loss": 10.8741, "step": 102 }, { "epoch": 0.006393941275063629, "grad_norm": 1.1757175770498145, "learning_rate": 2.1311814607903996e-06, "loss": 10.8688, "step": 103 }, { "epoch": 0.006456018374821528, "grad_norm": 1.1822445169645581, "learning_rate": 2.1518725429339955e-06, "loss": 10.8446, "step": 104 }, { "epoch": 0.006518095474579428, "grad_norm": 1.142412535577005, "learning_rate": 2.172563625077592e-06, "loss": 10.8757, "step": 105 }, { "epoch": 0.006580172574337327, "grad_norm": 1.1703715534870411, "learning_rate": 2.1932547072211878e-06, "loss": 10.8103, "step": 106 }, { "epoch": 0.006642249674095226, "grad_norm": 1.1711107432205794, "learning_rate": 2.2139457893647837e-06, "loss": 10.7641, "step": 107 }, { "epoch": 0.006704326773853126, "grad_norm": 1.095750831476619, "learning_rate": 2.23463687150838e-06, "loss": 10.8375, "step": 108 }, { "epoch": 0.0067664038736110245, "grad_norm": 1.1280488979211292, "learning_rate": 2.2553279536519764e-06, "loss": 10.7695, "step": 109 }, { "epoch": 0.006828480973368924, "grad_norm": 1.1216588207856597, "learning_rate": 2.2760190357955723e-06, "loss": 10.7503, "step": 110 }, { "epoch": 0.006890558073126824, "grad_norm": 1.125016212346561, "learning_rate": 2.2967101179391682e-06, "loss": 10.7763, "step": 111 }, { "epoch": 0.006952635172884723, "grad_norm": 1.1097546854523954, "learning_rate": 2.3174012000827646e-06, "loss": 10.7649, "step": 112 }, { "epoch": 0.007014712272642622, "grad_norm": 1.1187487810011034, "learning_rate": 2.3380922822263605e-06, "loss": 10.6933, "step": 113 }, { "epoch": 0.007076789372400522, "grad_norm": 1.116241214371609, "learning_rate": 2.3587833643699564e-06, "loss": 10.683, "step": 114 }, { "epoch": 0.007138866472158421, "grad_norm": 1.1294185967361448, "learning_rate": 2.3794744465135528e-06, "loss": 10.641, "step": 115 }, { "epoch": 0.00720094357191632, "grad_norm": 1.0604604331067409, "learning_rate": 2.400165528657149e-06, "loss": 10.7366, "step": 116 }, { "epoch": 0.007263020671674219, "grad_norm": 1.1145271156370191, "learning_rate": 2.420856610800745e-06, "loss": 10.636, "step": 117 }, { "epoch": 0.007325097771432119, "grad_norm": 1.0828463792661371, "learning_rate": 2.441547692944341e-06, "loss": 10.6517, "step": 118 }, { "epoch": 0.007387174871190018, "grad_norm": 1.060588856914667, "learning_rate": 2.4622387750879373e-06, "loss": 10.7005, "step": 119 }, { "epoch": 0.007449251970947917, "grad_norm": 1.020864148875047, "learning_rate": 2.4829298572315337e-06, "loss": 10.6182, "step": 120 }, { "epoch": 0.007511329070705817, "grad_norm": 1.023787984457747, "learning_rate": 2.503620939375129e-06, "loss": 10.6456, "step": 121 }, { "epoch": 0.007573406170463716, "grad_norm": 1.0052956195679323, "learning_rate": 2.5243120215187255e-06, "loss": 10.6503, "step": 122 }, { "epoch": 0.007635483270221615, "grad_norm": 1.0233126848127814, "learning_rate": 2.545003103662322e-06, "loss": 10.611, "step": 123 }, { "epoch": 0.007697560369979515, "grad_norm": 1.0276779965487508, "learning_rate": 2.565694185805918e-06, "loss": 10.6593, "step": 124 }, { "epoch": 0.007759637469737414, "grad_norm": 0.9860768819667932, "learning_rate": 2.5863852679495137e-06, "loss": 10.616, "step": 125 }, { "epoch": 0.007821714569495312, "grad_norm": 0.9835886463520696, "learning_rate": 2.60707635009311e-06, "loss": 10.5627, "step": 126 }, { "epoch": 0.007883791669253213, "grad_norm": 0.9497584625674442, "learning_rate": 2.6277674322367064e-06, "loss": 10.6027, "step": 127 }, { "epoch": 0.007945868769011112, "grad_norm": 1.0038411753896703, "learning_rate": 2.648458514380302e-06, "loss": 10.5026, "step": 128 }, { "epoch": 0.00800794586876901, "grad_norm": 0.9692593127834679, "learning_rate": 2.6691495965238983e-06, "loss": 10.5636, "step": 129 }, { "epoch": 0.008070022968526911, "grad_norm": 0.9988005119985445, "learning_rate": 2.6898406786674946e-06, "loss": 10.5042, "step": 130 }, { "epoch": 0.00813210006828481, "grad_norm": 0.9599061719420785, "learning_rate": 2.7105317608110905e-06, "loss": 10.4984, "step": 131 }, { "epoch": 0.008194177168042709, "grad_norm": 0.9664313656157655, "learning_rate": 2.7312228429546864e-06, "loss": 10.5414, "step": 132 }, { "epoch": 0.00825625426780061, "grad_norm": 0.9730836166863175, "learning_rate": 2.751913925098283e-06, "loss": 10.529, "step": 133 }, { "epoch": 0.008318331367558508, "grad_norm": 0.9655327749808524, "learning_rate": 2.772605007241879e-06, "loss": 10.5258, "step": 134 }, { "epoch": 0.008380408467316407, "grad_norm": 0.9760954999106549, "learning_rate": 2.7932960893854746e-06, "loss": 10.4386, "step": 135 }, { "epoch": 0.008442485567074305, "grad_norm": 0.944915467422289, "learning_rate": 2.813987171529071e-06, "loss": 10.4761, "step": 136 }, { "epoch": 0.008504562666832206, "grad_norm": 0.9008436077201308, "learning_rate": 2.8346782536726673e-06, "loss": 10.5233, "step": 137 }, { "epoch": 0.008566639766590105, "grad_norm": 0.9117916950768027, "learning_rate": 2.8553693358162633e-06, "loss": 10.4987, "step": 138 }, { "epoch": 0.008628716866348004, "grad_norm": 0.9323878780283165, "learning_rate": 2.876060417959859e-06, "loss": 10.4855, "step": 139 }, { "epoch": 0.008690793966105904, "grad_norm": 0.9229593849772569, "learning_rate": 2.8967515001034555e-06, "loss": 10.4237, "step": 140 }, { "epoch": 0.008752871065863803, "grad_norm": 0.9185517428053048, "learning_rate": 2.917442582247052e-06, "loss": 10.4791, "step": 141 }, { "epoch": 0.008814948165621702, "grad_norm": 0.9289570992606955, "learning_rate": 2.938133664390648e-06, "loss": 10.3981, "step": 142 }, { "epoch": 0.008877025265379602, "grad_norm": 0.9329369121761008, "learning_rate": 2.9588247465342437e-06, "loss": 10.3656, "step": 143 }, { "epoch": 0.008939102365137501, "grad_norm": 0.9020400754205117, "learning_rate": 2.97951582867784e-06, "loss": 10.4574, "step": 144 }, { "epoch": 0.0090011794648954, "grad_norm": 0.9074237391583676, "learning_rate": 3.000206910821436e-06, "loss": 10.3865, "step": 145 }, { "epoch": 0.0090632565646533, "grad_norm": 0.9151084979408398, "learning_rate": 3.0208979929650323e-06, "loss": 10.3987, "step": 146 }, { "epoch": 0.009125333664411199, "grad_norm": 0.908778904346301, "learning_rate": 3.0415890751086283e-06, "loss": 10.3486, "step": 147 }, { "epoch": 0.009187410764169098, "grad_norm": 0.9019753500875328, "learning_rate": 3.0622801572522246e-06, "loss": 10.3854, "step": 148 }, { "epoch": 0.009249487863926997, "grad_norm": 0.8719979968529188, "learning_rate": 3.0829712393958205e-06, "loss": 10.4065, "step": 149 }, { "epoch": 0.009311564963684897, "grad_norm": 0.9116699663331492, "learning_rate": 3.1036623215394165e-06, "loss": 10.3619, "step": 150 }, { "epoch": 0.009373642063442796, "grad_norm": 0.9233432973457193, "learning_rate": 3.124353403683013e-06, "loss": 10.3036, "step": 151 }, { "epoch": 0.009435719163200695, "grad_norm": 0.8816559228681164, "learning_rate": 3.1450444858266087e-06, "loss": 10.3703, "step": 152 }, { "epoch": 0.009497796262958595, "grad_norm": 0.8983623970817559, "learning_rate": 3.165735567970205e-06, "loss": 10.3103, "step": 153 }, { "epoch": 0.009559873362716494, "grad_norm": 0.8598889634620247, "learning_rate": 3.186426650113801e-06, "loss": 10.3656, "step": 154 }, { "epoch": 0.009621950462474393, "grad_norm": 0.891106558996477, "learning_rate": 3.2071177322573973e-06, "loss": 10.3206, "step": 155 }, { "epoch": 0.009684027562232293, "grad_norm": 0.8514003074243989, "learning_rate": 3.2278088144009933e-06, "loss": 10.3587, "step": 156 }, { "epoch": 0.009746104661990192, "grad_norm": 0.8800462577322832, "learning_rate": 3.248499896544589e-06, "loss": 10.3025, "step": 157 }, { "epoch": 0.00980818176174809, "grad_norm": 0.8797884434297634, "learning_rate": 3.269190978688186e-06, "loss": 10.263, "step": 158 }, { "epoch": 0.00987025886150599, "grad_norm": 0.8929197630040153, "learning_rate": 3.2898820608317815e-06, "loss": 10.2433, "step": 159 }, { "epoch": 0.00993233596126389, "grad_norm": 0.8325192806657431, "learning_rate": 3.3105731429753774e-06, "loss": 10.3235, "step": 160 }, { "epoch": 0.009994413061021789, "grad_norm": 0.8493182505666204, "learning_rate": 3.331264225118974e-06, "loss": 10.286, "step": 161 }, { "epoch": 0.010056490160779688, "grad_norm": 0.8544834332698888, "learning_rate": 3.35195530726257e-06, "loss": 10.331, "step": 162 }, { "epoch": 0.010118567260537588, "grad_norm": 0.8566520902368259, "learning_rate": 3.3726463894061656e-06, "loss": 10.277, "step": 163 }, { "epoch": 0.010180644360295487, "grad_norm": 0.852175046062499, "learning_rate": 3.3933374715497623e-06, "loss": 10.2875, "step": 164 }, { "epoch": 0.010242721460053386, "grad_norm": 0.8719918791388416, "learning_rate": 3.4140285536933583e-06, "loss": 10.2124, "step": 165 }, { "epoch": 0.010304798559811286, "grad_norm": 0.8478996876571733, "learning_rate": 3.4347196358369546e-06, "loss": 10.2957, "step": 166 }, { "epoch": 0.010366875659569185, "grad_norm": 0.8504470729360581, "learning_rate": 3.4554107179805505e-06, "loss": 10.2518, "step": 167 }, { "epoch": 0.010428952759327084, "grad_norm": 0.8463704425518359, "learning_rate": 3.4761018001241465e-06, "loss": 10.2161, "step": 168 }, { "epoch": 0.010491029859084984, "grad_norm": 0.8511761480104245, "learning_rate": 3.496792882267743e-06, "loss": 10.2476, "step": 169 }, { "epoch": 0.010553106958842883, "grad_norm": 0.8588367163506812, "learning_rate": 3.5174839644113387e-06, "loss": 10.2174, "step": 170 }, { "epoch": 0.010615184058600782, "grad_norm": 0.8363684853405668, "learning_rate": 3.5381750465549347e-06, "loss": 10.2247, "step": 171 }, { "epoch": 0.01067726115835868, "grad_norm": 0.8664207505439341, "learning_rate": 3.5588661286985314e-06, "loss": 10.1719, "step": 172 }, { "epoch": 0.010739338258116581, "grad_norm": 0.8235588505485962, "learning_rate": 3.579557210842127e-06, "loss": 10.2728, "step": 173 }, { "epoch": 0.01080141535787448, "grad_norm": 0.8274924183408883, "learning_rate": 3.600248292985723e-06, "loss": 10.2579, "step": 174 }, { "epoch": 0.010863492457632379, "grad_norm": 0.8369766977523412, "learning_rate": 3.6209393751293196e-06, "loss": 10.1763, "step": 175 }, { "epoch": 0.01092556955739028, "grad_norm": 0.8208202788642924, "learning_rate": 3.6416304572729155e-06, "loss": 10.2212, "step": 176 }, { "epoch": 0.010987646657148178, "grad_norm": 0.8099057523032113, "learning_rate": 3.662321539416512e-06, "loss": 10.2015, "step": 177 }, { "epoch": 0.011049723756906077, "grad_norm": 0.8167509601518483, "learning_rate": 3.683012621560108e-06, "loss": 10.2189, "step": 178 }, { "epoch": 0.011111800856663977, "grad_norm": 0.8158097711913613, "learning_rate": 3.7037037037037037e-06, "loss": 10.1578, "step": 179 }, { "epoch": 0.011173877956421876, "grad_norm": 0.811593356833007, "learning_rate": 3.7243947858473e-06, "loss": 10.2029, "step": 180 }, { "epoch": 0.011235955056179775, "grad_norm": 0.8115477116485592, "learning_rate": 3.745085867990896e-06, "loss": 10.1284, "step": 181 }, { "epoch": 0.011298032155937674, "grad_norm": 0.8153713182585228, "learning_rate": 3.765776950134492e-06, "loss": 10.1955, "step": 182 }, { "epoch": 0.011360109255695574, "grad_norm": 0.8292031377570135, "learning_rate": 3.7864680322780887e-06, "loss": 10.1724, "step": 183 }, { "epoch": 0.011422186355453473, "grad_norm": 0.7926092601393043, "learning_rate": 3.807159114421684e-06, "loss": 10.1707, "step": 184 }, { "epoch": 0.011484263455211372, "grad_norm": 0.8332980188198564, "learning_rate": 3.8278501965652806e-06, "loss": 10.1415, "step": 185 }, { "epoch": 0.011546340554969272, "grad_norm": 0.8392858382053922, "learning_rate": 3.8485412787088765e-06, "loss": 10.1542, "step": 186 }, { "epoch": 0.011608417654727171, "grad_norm": 0.8658655417702095, "learning_rate": 3.869232360852472e-06, "loss": 10.041, "step": 187 }, { "epoch": 0.01167049475448507, "grad_norm": 0.8435400242380056, "learning_rate": 3.889923442996069e-06, "loss": 10.0777, "step": 188 }, { "epoch": 0.01173257185424297, "grad_norm": 0.7740312089320546, "learning_rate": 3.910614525139665e-06, "loss": 10.1547, "step": 189 }, { "epoch": 0.01179464895400087, "grad_norm": 0.7988978497731114, "learning_rate": 3.931305607283261e-06, "loss": 10.1604, "step": 190 }, { "epoch": 0.011856726053758768, "grad_norm": 0.8384437424878407, "learning_rate": 3.951996689426858e-06, "loss": 9.9974, "step": 191 }, { "epoch": 0.011918803153516669, "grad_norm": 0.8059415951906183, "learning_rate": 3.972687771570454e-06, "loss": 10.058, "step": 192 }, { "epoch": 0.011980880253274567, "grad_norm": 0.8091937432897203, "learning_rate": 3.993378853714049e-06, "loss": 10.0434, "step": 193 }, { "epoch": 0.012042957353032466, "grad_norm": 0.79989470742639, "learning_rate": 4.0140699358576456e-06, "loss": 10.0625, "step": 194 }, { "epoch": 0.012105034452790365, "grad_norm": 0.7947560730276422, "learning_rate": 4.0347610180012415e-06, "loss": 10.0918, "step": 195 }, { "epoch": 0.012167111552548265, "grad_norm": 0.7931099376044809, "learning_rate": 4.055452100144837e-06, "loss": 10.0733, "step": 196 }, { "epoch": 0.012229188652306164, "grad_norm": 0.8042911182881192, "learning_rate": 4.076143182288434e-06, "loss": 10.0349, "step": 197 }, { "epoch": 0.012291265752064063, "grad_norm": 0.7812192529349946, "learning_rate": 4.09683426443203e-06, "loss": 10.1075, "step": 198 }, { "epoch": 0.012353342851821963, "grad_norm": 0.835880883697161, "learning_rate": 4.117525346575626e-06, "loss": 9.9474, "step": 199 }, { "epoch": 0.012415419951579862, "grad_norm": 0.7911833032035218, "learning_rate": 4.138216428719222e-06, "loss": 10.0733, "step": 200 }, { "epoch": 0.012477497051337761, "grad_norm": 0.8068408486653816, "learning_rate": 4.158907510862818e-06, "loss": 10.036, "step": 201 }, { "epoch": 0.012539574151095662, "grad_norm": 0.7847814471945931, "learning_rate": 4.179598593006415e-06, "loss": 10.0635, "step": 202 }, { "epoch": 0.01260165125085356, "grad_norm": 0.8173098149854618, "learning_rate": 4.2002896751500106e-06, "loss": 9.9737, "step": 203 }, { "epoch": 0.012663728350611459, "grad_norm": 0.829706086203809, "learning_rate": 4.2209807572936065e-06, "loss": 10.03, "step": 204 }, { "epoch": 0.012725805450369358, "grad_norm": 0.8102368232882065, "learning_rate": 4.241671839437203e-06, "loss": 9.9813, "step": 205 }, { "epoch": 0.012787882550127258, "grad_norm": 0.8008860449722233, "learning_rate": 4.262362921580799e-06, "loss": 9.9927, "step": 206 }, { "epoch": 0.012849959649885157, "grad_norm": 0.7895354185101906, "learning_rate": 4.283054003724395e-06, "loss": 10.0326, "step": 207 }, { "epoch": 0.012912036749643056, "grad_norm": 0.8078611067826079, "learning_rate": 4.303745085867991e-06, "loss": 9.9595, "step": 208 }, { "epoch": 0.012974113849400956, "grad_norm": 0.8019632687231402, "learning_rate": 4.324436168011587e-06, "loss": 9.9293, "step": 209 }, { "epoch": 0.013036190949158855, "grad_norm": 0.8013236093259066, "learning_rate": 4.345127250155184e-06, "loss": 9.9542, "step": 210 }, { "epoch": 0.013098268048916754, "grad_norm": 0.7964800435575655, "learning_rate": 4.36581833229878e-06, "loss": 9.9372, "step": 211 }, { "epoch": 0.013160345148674655, "grad_norm": 0.7711210143613471, "learning_rate": 4.3865094144423756e-06, "loss": 10.0067, "step": 212 }, { "epoch": 0.013222422248432553, "grad_norm": 0.7634872642658999, "learning_rate": 4.4072004965859715e-06, "loss": 9.9989, "step": 213 }, { "epoch": 0.013284499348190452, "grad_norm": 0.7731028131309706, "learning_rate": 4.427891578729567e-06, "loss": 9.9381, "step": 214 }, { "epoch": 0.013346576447948353, "grad_norm": 0.7840719754743848, "learning_rate": 4.448582660873163e-06, "loss": 9.9029, "step": 215 }, { "epoch": 0.013408653547706251, "grad_norm": 0.7903217804749627, "learning_rate": 4.46927374301676e-06, "loss": 9.8872, "step": 216 }, { "epoch": 0.01347073064746415, "grad_norm": 0.7585201695249384, "learning_rate": 4.489964825160356e-06, "loss": 9.9868, "step": 217 }, { "epoch": 0.013532807747222049, "grad_norm": 0.7688351111089315, "learning_rate": 4.510655907303953e-06, "loss": 9.9477, "step": 218 }, { "epoch": 0.01359488484697995, "grad_norm": 0.7761148678521732, "learning_rate": 4.531346989447549e-06, "loss": 9.9024, "step": 219 }, { "epoch": 0.013656961946737848, "grad_norm": 0.7862784872670955, "learning_rate": 4.552038071591145e-06, "loss": 9.9002, "step": 220 }, { "epoch": 0.013719039046495747, "grad_norm": 0.792349098772769, "learning_rate": 4.5727291537347406e-06, "loss": 9.8417, "step": 221 }, { "epoch": 0.013781116146253648, "grad_norm": 0.7715880119150847, "learning_rate": 4.5934202358783365e-06, "loss": 9.8982, "step": 222 }, { "epoch": 0.013843193246011546, "grad_norm": 0.7416006478561937, "learning_rate": 4.614111318021932e-06, "loss": 9.9497, "step": 223 }, { "epoch": 0.013905270345769445, "grad_norm": 0.7661240566606979, "learning_rate": 4.634802400165529e-06, "loss": 9.8651, "step": 224 }, { "epoch": 0.013967347445527346, "grad_norm": 0.7461846580844257, "learning_rate": 4.655493482309125e-06, "loss": 9.8891, "step": 225 }, { "epoch": 0.014029424545285244, "grad_norm": 0.8071861362187306, "learning_rate": 4.676184564452721e-06, "loss": 9.7929, "step": 226 }, { "epoch": 0.014091501645043143, "grad_norm": 0.780508986954693, "learning_rate": 4.696875646596317e-06, "loss": 9.8673, "step": 227 }, { "epoch": 0.014153578744801044, "grad_norm": 0.7817836845980041, "learning_rate": 4.717566728739913e-06, "loss": 9.8504, "step": 228 }, { "epoch": 0.014215655844558943, "grad_norm": 0.751421931518215, "learning_rate": 4.73825781088351e-06, "loss": 9.8231, "step": 229 }, { "epoch": 0.014277732944316841, "grad_norm": 0.7386471684205802, "learning_rate": 4.7589488930271056e-06, "loss": 9.8814, "step": 230 }, { "epoch": 0.01433981004407474, "grad_norm": 0.7236091311308297, "learning_rate": 4.7796399751707015e-06, "loss": 9.8768, "step": 231 }, { "epoch": 0.01440188714383264, "grad_norm": 0.7404283254374919, "learning_rate": 4.800331057314298e-06, "loss": 9.7771, "step": 232 }, { "epoch": 0.01446396424359054, "grad_norm": 0.7403863836337838, "learning_rate": 4.821022139457894e-06, "loss": 9.876, "step": 233 }, { "epoch": 0.014526041343348438, "grad_norm": 0.7499390418393448, "learning_rate": 4.84171322160149e-06, "loss": 9.796, "step": 234 }, { "epoch": 0.014588118443106339, "grad_norm": 0.7466057024559402, "learning_rate": 4.862404303745086e-06, "loss": 9.8608, "step": 235 }, { "epoch": 0.014650195542864237, "grad_norm": 0.7556110536031352, "learning_rate": 4.883095385888682e-06, "loss": 9.7805, "step": 236 }, { "epoch": 0.014712272642622136, "grad_norm": 0.75197028598772, "learning_rate": 4.903786468032278e-06, "loss": 9.8243, "step": 237 }, { "epoch": 0.014774349742380037, "grad_norm": 0.7545786018862187, "learning_rate": 4.924477550175875e-06, "loss": 9.7253, "step": 238 }, { "epoch": 0.014836426842137936, "grad_norm": 0.7225121844231155, "learning_rate": 4.945168632319471e-06, "loss": 9.8434, "step": 239 }, { "epoch": 0.014898503941895834, "grad_norm": 0.7401966900656411, "learning_rate": 4.965859714463067e-06, "loss": 9.7804, "step": 240 }, { "epoch": 0.014960581041653733, "grad_norm": 0.7377712769162907, "learning_rate": 4.9865507966066624e-06, "loss": 9.772, "step": 241 }, { "epoch": 0.015022658141411634, "grad_norm": 0.770451919597735, "learning_rate": 5.007241878750258e-06, "loss": 9.7302, "step": 242 }, { "epoch": 0.015084735241169532, "grad_norm": 0.7429845445239687, "learning_rate": 5.027932960893855e-06, "loss": 9.7813, "step": 243 }, { "epoch": 0.015146812340927431, "grad_norm": 0.7566684069872545, "learning_rate": 5.048624043037451e-06, "loss": 9.7485, "step": 244 }, { "epoch": 0.015208889440685332, "grad_norm": 0.7278338558328213, "learning_rate": 5.069315125181047e-06, "loss": 9.7709, "step": 245 }, { "epoch": 0.01527096654044323, "grad_norm": 0.7232523753380693, "learning_rate": 5.090006207324644e-06, "loss": 9.7984, "step": 246 }, { "epoch": 0.01533304364020113, "grad_norm": 0.721682949858894, "learning_rate": 5.11069728946824e-06, "loss": 9.7732, "step": 247 }, { "epoch": 0.01539512073995903, "grad_norm": 0.7345417320414015, "learning_rate": 5.131388371611836e-06, "loss": 9.7178, "step": 248 }, { "epoch": 0.015457197839716929, "grad_norm": 0.7341777861363257, "learning_rate": 5.1520794537554315e-06, "loss": 9.6846, "step": 249 }, { "epoch": 0.015519274939474827, "grad_norm": 0.7466983841718798, "learning_rate": 5.1727705358990274e-06, "loss": 9.7466, "step": 250 }, { "epoch": 0.015581352039232728, "grad_norm": 0.7301971420177824, "learning_rate": 5.193461618042624e-06, "loss": 9.6981, "step": 251 }, { "epoch": 0.015643429138990625, "grad_norm": 0.7640685005186395, "learning_rate": 5.21415270018622e-06, "loss": 9.6283, "step": 252 }, { "epoch": 0.015705506238748525, "grad_norm": 0.7402133183768197, "learning_rate": 5.234843782329816e-06, "loss": 9.7311, "step": 253 }, { "epoch": 0.015767583338506426, "grad_norm": 0.7379084862101778, "learning_rate": 5.255534864473413e-06, "loss": 9.6154, "step": 254 }, { "epoch": 0.015829660438264323, "grad_norm": 0.7397714742706644, "learning_rate": 5.276225946617008e-06, "loss": 9.5819, "step": 255 }, { "epoch": 0.015891737538022224, "grad_norm": 0.7352543294183145, "learning_rate": 5.296917028760604e-06, "loss": 9.5875, "step": 256 }, { "epoch": 0.015953814637780124, "grad_norm": 0.7263176618300222, "learning_rate": 5.317608110904201e-06, "loss": 9.6022, "step": 257 }, { "epoch": 0.01601589173753802, "grad_norm": 0.7033607935252933, "learning_rate": 5.3382991930477965e-06, "loss": 9.7023, "step": 258 }, { "epoch": 0.01607796883729592, "grad_norm": 0.7283245173501952, "learning_rate": 5.3589902751913924e-06, "loss": 9.6212, "step": 259 }, { "epoch": 0.016140045937053822, "grad_norm": 0.7095094948201177, "learning_rate": 5.379681357334989e-06, "loss": 9.6876, "step": 260 }, { "epoch": 0.01620212303681172, "grad_norm": 0.7342484141882402, "learning_rate": 5.400372439478585e-06, "loss": 9.5803, "step": 261 }, { "epoch": 0.01626420013656962, "grad_norm": 0.7217936548416575, "learning_rate": 5.421063521622181e-06, "loss": 9.6451, "step": 262 }, { "epoch": 0.01632627723632752, "grad_norm": 0.7413734388577804, "learning_rate": 5.441754603765777e-06, "loss": 9.6054, "step": 263 }, { "epoch": 0.016388354336085417, "grad_norm": 0.7155294829344814, "learning_rate": 5.462445685909373e-06, "loss": 9.6257, "step": 264 }, { "epoch": 0.016450431435843318, "grad_norm": 0.7390644420251473, "learning_rate": 5.48313676805297e-06, "loss": 9.5837, "step": 265 }, { "epoch": 0.01651250853560122, "grad_norm": 0.7029348540162594, "learning_rate": 5.503827850196566e-06, "loss": 9.6083, "step": 266 }, { "epoch": 0.016574585635359115, "grad_norm": 0.7023376719176787, "learning_rate": 5.5245189323401615e-06, "loss": 9.6181, "step": 267 }, { "epoch": 0.016636662735117016, "grad_norm": 0.6983570351216316, "learning_rate": 5.545210014483758e-06, "loss": 9.6248, "step": 268 }, { "epoch": 0.016698739834874916, "grad_norm": 0.6959889582437737, "learning_rate": 5.565901096627354e-06, "loss": 9.5206, "step": 269 }, { "epoch": 0.016760816934632813, "grad_norm": 0.6987500244645607, "learning_rate": 5.586592178770949e-06, "loss": 9.5645, "step": 270 }, { "epoch": 0.016822894034390714, "grad_norm": 0.7102674323778851, "learning_rate": 5.607283260914546e-06, "loss": 9.5045, "step": 271 }, { "epoch": 0.01688497113414861, "grad_norm": 0.6896463615064292, "learning_rate": 5.627974343058142e-06, "loss": 9.6149, "step": 272 }, { "epoch": 0.01694704823390651, "grad_norm": 0.7139183266165745, "learning_rate": 5.648665425201739e-06, "loss": 9.5377, "step": 273 }, { "epoch": 0.017009125333664412, "grad_norm": 0.6719135672567061, "learning_rate": 5.669356507345335e-06, "loss": 9.5956, "step": 274 }, { "epoch": 0.01707120243342231, "grad_norm": 0.6922422624198977, "learning_rate": 5.690047589488931e-06, "loss": 9.5974, "step": 275 }, { "epoch": 0.01713327953318021, "grad_norm": 0.7037844003092186, "learning_rate": 5.7107386716325265e-06, "loss": 9.5299, "step": 276 }, { "epoch": 0.01719535663293811, "grad_norm": 0.715685322980174, "learning_rate": 5.7314297537761224e-06, "loss": 9.4565, "step": 277 }, { "epoch": 0.017257433732696007, "grad_norm": 0.7046492690988608, "learning_rate": 5.752120835919718e-06, "loss": 9.5439, "step": 278 }, { "epoch": 0.017319510832453908, "grad_norm": 0.7105413737616247, "learning_rate": 5.772811918063315e-06, "loss": 9.4222, "step": 279 }, { "epoch": 0.017381587932211808, "grad_norm": 0.6864205034991699, "learning_rate": 5.793503000206911e-06, "loss": 9.5368, "step": 280 }, { "epoch": 0.017443665031969705, "grad_norm": 0.7201410598460571, "learning_rate": 5.814194082350508e-06, "loss": 9.4854, "step": 281 }, { "epoch": 0.017505742131727606, "grad_norm": 0.710582434890064, "learning_rate": 5.834885164494104e-06, "loss": 9.4254, "step": 282 }, { "epoch": 0.017567819231485506, "grad_norm": 0.6874934976443658, "learning_rate": 5.8555762466377e-06, "loss": 9.4764, "step": 283 }, { "epoch": 0.017629896331243403, "grad_norm": 0.6871097063732812, "learning_rate": 5.876267328781296e-06, "loss": 9.5366, "step": 284 }, { "epoch": 0.017691973431001304, "grad_norm": 0.6919742126455954, "learning_rate": 5.8969584109248915e-06, "loss": 9.4923, "step": 285 }, { "epoch": 0.017754050530759204, "grad_norm": 0.6910296777791778, "learning_rate": 5.9176494930684874e-06, "loss": 9.439, "step": 286 }, { "epoch": 0.0178161276305171, "grad_norm": 0.6799859893281078, "learning_rate": 5.938340575212084e-06, "loss": 9.4701, "step": 287 }, { "epoch": 0.017878204730275002, "grad_norm": 0.6767622161765997, "learning_rate": 5.95903165735568e-06, "loss": 9.4741, "step": 288 }, { "epoch": 0.017940281830032902, "grad_norm": 0.6856036257824341, "learning_rate": 5.979722739499276e-06, "loss": 9.4507, "step": 289 }, { "epoch": 0.0180023589297908, "grad_norm": 0.6857105505329252, "learning_rate": 6.000413821642872e-06, "loss": 9.3676, "step": 290 }, { "epoch": 0.0180644360295487, "grad_norm": 0.6961289659148432, "learning_rate": 6.021104903786468e-06, "loss": 9.36, "step": 291 }, { "epoch": 0.0181265131293066, "grad_norm": 0.6826755351396097, "learning_rate": 6.041795985930065e-06, "loss": 9.453, "step": 292 }, { "epoch": 0.018188590229064498, "grad_norm": 0.6603486046080304, "learning_rate": 6.062487068073661e-06, "loss": 9.4461, "step": 293 }, { "epoch": 0.018250667328822398, "grad_norm": 0.6768087172526264, "learning_rate": 6.0831781502172565e-06, "loss": 9.4014, "step": 294 }, { "epoch": 0.018312744428580295, "grad_norm": 0.6537278391143618, "learning_rate": 6.103869232360853e-06, "loss": 9.3949, "step": 295 }, { "epoch": 0.018374821528338196, "grad_norm": 0.6527919257662167, "learning_rate": 6.124560314504449e-06, "loss": 9.3718, "step": 296 }, { "epoch": 0.018436898628096096, "grad_norm": 0.6516539987404373, "learning_rate": 6.145251396648045e-06, "loss": 9.4163, "step": 297 }, { "epoch": 0.018498975727853993, "grad_norm": 0.6957026683499585, "learning_rate": 6.165942478791641e-06, "loss": 9.279, "step": 298 }, { "epoch": 0.018561052827611894, "grad_norm": 0.6723941527172531, "learning_rate": 6.186633560935237e-06, "loss": 9.3223, "step": 299 }, { "epoch": 0.018623129927369794, "grad_norm": 0.6506631870038306, "learning_rate": 6.207324643078833e-06, "loss": 9.3279, "step": 300 }, { "epoch": 0.01868520702712769, "grad_norm": 0.6734884318057051, "learning_rate": 6.22801572522243e-06, "loss": 9.3105, "step": 301 }, { "epoch": 0.018747284126885592, "grad_norm": 0.6828346179024877, "learning_rate": 6.248706807366026e-06, "loss": 9.2711, "step": 302 }, { "epoch": 0.018809361226643492, "grad_norm": 0.6626762731432752, "learning_rate": 6.269397889509622e-06, "loss": 9.3111, "step": 303 }, { "epoch": 0.01887143832640139, "grad_norm": 0.6405520979175987, "learning_rate": 6.2900889716532175e-06, "loss": 9.3893, "step": 304 }, { "epoch": 0.01893351542615929, "grad_norm": 0.6850563968288789, "learning_rate": 6.310780053796813e-06, "loss": 9.2793, "step": 305 }, { "epoch": 0.01899559252591719, "grad_norm": 0.6548302824783164, "learning_rate": 6.33147113594041e-06, "loss": 9.2591, "step": 306 }, { "epoch": 0.019057669625675087, "grad_norm": 0.6500426040038538, "learning_rate": 6.352162218084005e-06, "loss": 9.2721, "step": 307 }, { "epoch": 0.019119746725432988, "grad_norm": 0.6641486841261836, "learning_rate": 6.372853300227602e-06, "loss": 9.295, "step": 308 }, { "epoch": 0.01918182382519089, "grad_norm": 0.6482871237429194, "learning_rate": 6.393544382371199e-06, "loss": 9.3387, "step": 309 }, { "epoch": 0.019243900924948786, "grad_norm": 0.6673258537477661, "learning_rate": 6.414235464514795e-06, "loss": 9.2106, "step": 310 }, { "epoch": 0.019305978024706686, "grad_norm": 0.6409220588271092, "learning_rate": 6.434926546658391e-06, "loss": 9.339, "step": 311 }, { "epoch": 0.019368055124464587, "grad_norm": 0.6244529475757467, "learning_rate": 6.4556176288019865e-06, "loss": 9.3382, "step": 312 }, { "epoch": 0.019430132224222484, "grad_norm": 0.6529603659371578, "learning_rate": 6.476308710945583e-06, "loss": 9.2552, "step": 313 }, { "epoch": 0.019492209323980384, "grad_norm": 0.631739070960834, "learning_rate": 6.496999793089178e-06, "loss": 9.2945, "step": 314 }, { "epoch": 0.019554286423738285, "grad_norm": 0.6237249481353438, "learning_rate": 6.517690875232775e-06, "loss": 9.3027, "step": 315 }, { "epoch": 0.01961636352349618, "grad_norm": 0.6472307949789987, "learning_rate": 6.538381957376372e-06, "loss": 9.2585, "step": 316 }, { "epoch": 0.019678440623254082, "grad_norm": 0.6544529287504859, "learning_rate": 6.559073039519967e-06, "loss": 9.1638, "step": 317 }, { "epoch": 0.01974051772301198, "grad_norm": 0.6235161332564629, "learning_rate": 6.579764121663563e-06, "loss": 9.1597, "step": 318 }, { "epoch": 0.01980259482276988, "grad_norm": 0.6604046777385107, "learning_rate": 6.60045520380716e-06, "loss": 9.1163, "step": 319 }, { "epoch": 0.01986467192252778, "grad_norm": 0.6217495796591792, "learning_rate": 6.621146285950755e-06, "loss": 9.2374, "step": 320 }, { "epoch": 0.019926749022285677, "grad_norm": 0.6241155326594314, "learning_rate": 6.6418373680943515e-06, "loss": 9.2227, "step": 321 }, { "epoch": 0.019988826122043578, "grad_norm": 0.6550280456872252, "learning_rate": 6.662528450237948e-06, "loss": 9.1939, "step": 322 }, { "epoch": 0.02005090322180148, "grad_norm": 0.6195062947388338, "learning_rate": 6.683219532381543e-06, "loss": 9.2431, "step": 323 }, { "epoch": 0.020112980321559375, "grad_norm": 0.644239070441396, "learning_rate": 6.70391061452514e-06, "loss": 9.1503, "step": 324 }, { "epoch": 0.020175057421317276, "grad_norm": 0.6286613271763944, "learning_rate": 6.724601696668736e-06, "loss": 9.1489, "step": 325 }, { "epoch": 0.020237134521075176, "grad_norm": 0.6287925333080252, "learning_rate": 6.745292778812331e-06, "loss": 9.1743, "step": 326 }, { "epoch": 0.020299211620833073, "grad_norm": 0.6109594141159206, "learning_rate": 6.765983860955928e-06, "loss": 9.1764, "step": 327 }, { "epoch": 0.020361288720590974, "grad_norm": 0.6104170130922899, "learning_rate": 6.786674943099525e-06, "loss": 9.1664, "step": 328 }, { "epoch": 0.020423365820348875, "grad_norm": 0.6615309087231722, "learning_rate": 6.8073660252431215e-06, "loss": 9.0687, "step": 329 }, { "epoch": 0.02048544292010677, "grad_norm": 0.6265289202512717, "learning_rate": 6.8280571073867165e-06, "loss": 9.1413, "step": 330 }, { "epoch": 0.020547520019864672, "grad_norm": 0.6335898675576203, "learning_rate": 6.848748189530313e-06, "loss": 9.113, "step": 331 }, { "epoch": 0.020609597119622573, "grad_norm": 0.6368135253017422, "learning_rate": 6.869439271673909e-06, "loss": 9.1288, "step": 332 }, { "epoch": 0.02067167421938047, "grad_norm": 0.5939195792316465, "learning_rate": 6.890130353817504e-06, "loss": 9.1196, "step": 333 }, { "epoch": 0.02073375131913837, "grad_norm": 0.5897177504809525, "learning_rate": 6.910821435961101e-06, "loss": 9.1335, "step": 334 }, { "epoch": 0.02079582841889627, "grad_norm": 0.5908726944835045, "learning_rate": 6.931512518104698e-06, "loss": 9.2086, "step": 335 }, { "epoch": 0.020857905518654168, "grad_norm": 0.5942903870406081, "learning_rate": 6.952203600248293e-06, "loss": 9.1509, "step": 336 }, { "epoch": 0.020919982618412068, "grad_norm": 0.6098006302574757, "learning_rate": 6.97289468239189e-06, "loss": 9.0884, "step": 337 }, { "epoch": 0.02098205971816997, "grad_norm": 0.5864433648032563, "learning_rate": 6.993585764535486e-06, "loss": 9.1407, "step": 338 }, { "epoch": 0.021044136817927866, "grad_norm": 0.5933392760518532, "learning_rate": 7.0142768466790815e-06, "loss": 9.1587, "step": 339 }, { "epoch": 0.021106213917685766, "grad_norm": 0.5926588221040845, "learning_rate": 7.0349679288226775e-06, "loss": 9.1201, "step": 340 }, { "epoch": 0.021168291017443663, "grad_norm": 0.6208664100164878, "learning_rate": 7.055659010966274e-06, "loss": 9.0733, "step": 341 }, { "epoch": 0.021230368117201564, "grad_norm": 0.5812672971387273, "learning_rate": 7.076350093109869e-06, "loss": 9.1115, "step": 342 }, { "epoch": 0.021292445216959464, "grad_norm": 0.5987398135965278, "learning_rate": 7.097041175253466e-06, "loss": 9.034, "step": 343 }, { "epoch": 0.02135452231671736, "grad_norm": 0.601931273146069, "learning_rate": 7.117732257397063e-06, "loss": 9.0385, "step": 344 }, { "epoch": 0.021416599416475262, "grad_norm": 0.6003810874365076, "learning_rate": 7.138423339540658e-06, "loss": 9.0796, "step": 345 }, { "epoch": 0.021478676516233162, "grad_norm": 0.6158654745366782, "learning_rate": 7.159114421684254e-06, "loss": 8.9488, "step": 346 }, { "epoch": 0.02154075361599106, "grad_norm": 0.577189332704056, "learning_rate": 7.179805503827851e-06, "loss": 9.0323, "step": 347 }, { "epoch": 0.02160283071574896, "grad_norm": 0.6150886968680217, "learning_rate": 7.200496585971446e-06, "loss": 9.01, "step": 348 }, { "epoch": 0.02166490781550686, "grad_norm": 0.5870262965383493, "learning_rate": 7.2211876681150425e-06, "loss": 9.063, "step": 349 }, { "epoch": 0.021726984915264758, "grad_norm": 0.5996069395531515, "learning_rate": 7.241878750258639e-06, "loss": 8.9994, "step": 350 }, { "epoch": 0.021789062015022658, "grad_norm": 0.613464595334256, "learning_rate": 7.262569832402236e-06, "loss": 8.972, "step": 351 }, { "epoch": 0.02185113911478056, "grad_norm": 0.5996829804711844, "learning_rate": 7.283260914545831e-06, "loss": 8.971, "step": 352 }, { "epoch": 0.021913216214538456, "grad_norm": 0.5892444875877692, "learning_rate": 7.303951996689427e-06, "loss": 9.008, "step": 353 }, { "epoch": 0.021975293314296356, "grad_norm": 0.58225417508356, "learning_rate": 7.324643078833024e-06, "loss": 8.9505, "step": 354 }, { "epoch": 0.022037370414054257, "grad_norm": 0.6025444290194185, "learning_rate": 7.345334160976619e-06, "loss": 8.9208, "step": 355 }, { "epoch": 0.022099447513812154, "grad_norm": 0.5930954185963652, "learning_rate": 7.366025243120216e-06, "loss": 8.9336, "step": 356 }, { "epoch": 0.022161524613570054, "grad_norm": 0.5614895752357738, "learning_rate": 7.386716325263812e-06, "loss": 8.9655, "step": 357 }, { "epoch": 0.022223601713327955, "grad_norm": 0.5763070675277707, "learning_rate": 7.4074074074074075e-06, "loss": 8.9606, "step": 358 }, { "epoch": 0.022285678813085852, "grad_norm": 0.5580454449486436, "learning_rate": 7.428098489551004e-06, "loss": 8.9736, "step": 359 }, { "epoch": 0.022347755912843752, "grad_norm": 0.5699579103305928, "learning_rate": 7.4487895716946e-06, "loss": 8.9828, "step": 360 }, { "epoch": 0.022409833012601653, "grad_norm": 0.5897016543583323, "learning_rate": 7.469480653838195e-06, "loss": 8.9524, "step": 361 }, { "epoch": 0.02247191011235955, "grad_norm": 0.5840791351088382, "learning_rate": 7.490171735981792e-06, "loss": 8.8379, "step": 362 }, { "epoch": 0.02253398721211745, "grad_norm": 0.5735999425576127, "learning_rate": 7.510862818125389e-06, "loss": 8.9649, "step": 363 }, { "epoch": 0.022596064311875348, "grad_norm": 0.5751298177376796, "learning_rate": 7.531553900268984e-06, "loss": 8.9015, "step": 364 }, { "epoch": 0.022658141411633248, "grad_norm": 0.5605648715755582, "learning_rate": 7.552244982412581e-06, "loss": 8.9405, "step": 365 }, { "epoch": 0.02272021851139115, "grad_norm": 0.589095999577178, "learning_rate": 7.572936064556177e-06, "loss": 8.8428, "step": 366 }, { "epoch": 0.022782295611149046, "grad_norm": 0.5777224866406064, "learning_rate": 7.5936271466997725e-06, "loss": 8.8357, "step": 367 }, { "epoch": 0.022844372710906946, "grad_norm": 0.6066648911967165, "learning_rate": 7.614318228843368e-06, "loss": 8.8354, "step": 368 }, { "epoch": 0.022906449810664847, "grad_norm": 0.5512403914637068, "learning_rate": 7.635009310986966e-06, "loss": 8.9189, "step": 369 }, { "epoch": 0.022968526910422744, "grad_norm": 0.5437930320286312, "learning_rate": 7.655700393130561e-06, "loss": 8.9041, "step": 370 }, { "epoch": 0.023030604010180644, "grad_norm": 0.5835687440470932, "learning_rate": 7.676391475274156e-06, "loss": 8.7437, "step": 371 }, { "epoch": 0.023092681109938545, "grad_norm": 0.5884007113519851, "learning_rate": 7.697082557417753e-06, "loss": 8.7888, "step": 372 }, { "epoch": 0.023154758209696442, "grad_norm": 0.5539423275754102, "learning_rate": 7.71777363956135e-06, "loss": 8.8464, "step": 373 }, { "epoch": 0.023216835309454342, "grad_norm": 0.5706663732163991, "learning_rate": 7.738464721704945e-06, "loss": 8.8652, "step": 374 }, { "epoch": 0.023278912409212243, "grad_norm": 0.5650274133783544, "learning_rate": 7.759155803848542e-06, "loss": 8.7819, "step": 375 }, { "epoch": 0.02334098950897014, "grad_norm": 0.5656517791731531, "learning_rate": 7.779846885992138e-06, "loss": 8.8179, "step": 376 }, { "epoch": 0.02340306660872804, "grad_norm": 0.545207751906302, "learning_rate": 7.800537968135733e-06, "loss": 8.8583, "step": 377 }, { "epoch": 0.02346514370848594, "grad_norm": 0.5747949473015618, "learning_rate": 7.82122905027933e-06, "loss": 8.7519, "step": 378 }, { "epoch": 0.023527220808243838, "grad_norm": 0.5651443631222232, "learning_rate": 7.841920132422927e-06, "loss": 8.8069, "step": 379 }, { "epoch": 0.02358929790800174, "grad_norm": 0.5459738254526275, "learning_rate": 7.862611214566522e-06, "loss": 8.7725, "step": 380 }, { "epoch": 0.02365137500775964, "grad_norm": 0.5443862689195345, "learning_rate": 7.883302296710119e-06, "loss": 8.8331, "step": 381 }, { "epoch": 0.023713452107517536, "grad_norm": 0.573431552367588, "learning_rate": 7.903993378853716e-06, "loss": 8.7699, "step": 382 }, { "epoch": 0.023775529207275437, "grad_norm": 0.5612763234275849, "learning_rate": 7.92468446099731e-06, "loss": 8.7769, "step": 383 }, { "epoch": 0.023837606307033337, "grad_norm": 0.5610049300189658, "learning_rate": 7.945375543140907e-06, "loss": 8.6969, "step": 384 }, { "epoch": 0.023899683406791234, "grad_norm": 0.5926058939372564, "learning_rate": 7.966066625284502e-06, "loss": 8.7603, "step": 385 }, { "epoch": 0.023961760506549135, "grad_norm": 0.5492010783883667, "learning_rate": 7.986757707428098e-06, "loss": 8.7331, "step": 386 }, { "epoch": 0.02402383760630703, "grad_norm": 0.5382358459900835, "learning_rate": 8.007448789571694e-06, "loss": 8.7401, "step": 387 }, { "epoch": 0.024085914706064932, "grad_norm": 0.5632274215734233, "learning_rate": 8.028139871715291e-06, "loss": 8.6728, "step": 388 }, { "epoch": 0.024147991805822833, "grad_norm": 0.5357018077623301, "learning_rate": 8.048830953858886e-06, "loss": 8.6971, "step": 389 }, { "epoch": 0.02421006890558073, "grad_norm": 0.5564335488079813, "learning_rate": 8.069522036002483e-06, "loss": 8.6888, "step": 390 }, { "epoch": 0.02427214600533863, "grad_norm": 0.5465287065827964, "learning_rate": 8.09021311814608e-06, "loss": 8.7043, "step": 391 }, { "epoch": 0.02433422310509653, "grad_norm": 0.5381706434203168, "learning_rate": 8.110904200289675e-06, "loss": 8.7031, "step": 392 }, { "epoch": 0.024396300204854428, "grad_norm": 0.5271941926595805, "learning_rate": 8.131595282433272e-06, "loss": 8.7373, "step": 393 }, { "epoch": 0.02445837730461233, "grad_norm": 0.5257257293363059, "learning_rate": 8.152286364576868e-06, "loss": 8.7015, "step": 394 }, { "epoch": 0.02452045440437023, "grad_norm": 0.5385937795259995, "learning_rate": 8.172977446720465e-06, "loss": 8.6326, "step": 395 }, { "epoch": 0.024582531504128126, "grad_norm": 0.5204007609430712, "learning_rate": 8.19366852886406e-06, "loss": 8.6884, "step": 396 }, { "epoch": 0.024644608603886026, "grad_norm": 0.5191898801929731, "learning_rate": 8.214359611007657e-06, "loss": 8.6549, "step": 397 }, { "epoch": 0.024706685703643927, "grad_norm": 0.5408586901062956, "learning_rate": 8.235050693151252e-06, "loss": 8.6187, "step": 398 }, { "epoch": 0.024768762803401824, "grad_norm": 0.5481591814490234, "learning_rate": 8.255741775294847e-06, "loss": 8.6217, "step": 399 }, { "epoch": 0.024830839903159724, "grad_norm": 0.5086333170089171, "learning_rate": 8.276432857438444e-06, "loss": 8.6922, "step": 400 }, { "epoch": 0.024892917002917625, "grad_norm": 0.5147749801343768, "learning_rate": 8.29712393958204e-06, "loss": 8.6746, "step": 401 }, { "epoch": 0.024954994102675522, "grad_norm": 0.5211204373275917, "learning_rate": 8.317815021725636e-06, "loss": 8.6285, "step": 402 }, { "epoch": 0.025017071202433423, "grad_norm": 0.5219695445325717, "learning_rate": 8.338506103869233e-06, "loss": 8.6581, "step": 403 }, { "epoch": 0.025079148302191323, "grad_norm": 0.5153265580665035, "learning_rate": 8.35919718601283e-06, "loss": 8.6271, "step": 404 }, { "epoch": 0.02514122540194922, "grad_norm": 0.5118472687751511, "learning_rate": 8.379888268156424e-06, "loss": 8.5913, "step": 405 }, { "epoch": 0.02520330250170712, "grad_norm": 0.5079221753746682, "learning_rate": 8.400579350300021e-06, "loss": 8.5965, "step": 406 }, { "epoch": 0.02526537960146502, "grad_norm": 0.5207249401470686, "learning_rate": 8.421270432443618e-06, "loss": 8.6122, "step": 407 }, { "epoch": 0.025327456701222918, "grad_norm": 0.5291084151107642, "learning_rate": 8.441961514587213e-06, "loss": 8.6323, "step": 408 }, { "epoch": 0.02538953380098082, "grad_norm": 0.5201660522930734, "learning_rate": 8.46265259673081e-06, "loss": 8.6332, "step": 409 }, { "epoch": 0.025451610900738716, "grad_norm": 0.4998876177136589, "learning_rate": 8.483343678874407e-06, "loss": 8.558, "step": 410 }, { "epoch": 0.025513688000496616, "grad_norm": 0.5019252784218643, "learning_rate": 8.504034761018002e-06, "loss": 8.6555, "step": 411 }, { "epoch": 0.025575765100254517, "grad_norm": 0.507910322890853, "learning_rate": 8.524725843161598e-06, "loss": 8.5763, "step": 412 }, { "epoch": 0.025637842200012414, "grad_norm": 0.49600732606238096, "learning_rate": 8.545416925305193e-06, "loss": 8.5955, "step": 413 }, { "epoch": 0.025699919299770314, "grad_norm": 0.512789065553757, "learning_rate": 8.56610800744879e-06, "loss": 8.5165, "step": 414 }, { "epoch": 0.025761996399528215, "grad_norm": 0.5011553231625788, "learning_rate": 8.586799089592385e-06, "loss": 8.5285, "step": 415 }, { "epoch": 0.025824073499286112, "grad_norm": 0.48691710078289563, "learning_rate": 8.607490171735982e-06, "loss": 8.6219, "step": 416 }, { "epoch": 0.025886150599044012, "grad_norm": 0.5048512904584875, "learning_rate": 8.628181253879579e-06, "loss": 8.5944, "step": 417 }, { "epoch": 0.025948227698801913, "grad_norm": 0.5082564833922264, "learning_rate": 8.648872336023174e-06, "loss": 8.4726, "step": 418 }, { "epoch": 0.02601030479855981, "grad_norm": 0.5094556779684873, "learning_rate": 8.66956341816677e-06, "loss": 8.4887, "step": 419 }, { "epoch": 0.02607238189831771, "grad_norm": 0.5016473514169486, "learning_rate": 8.690254500310367e-06, "loss": 8.4658, "step": 420 }, { "epoch": 0.02613445899807561, "grad_norm": 0.4896282232956093, "learning_rate": 8.710945582453963e-06, "loss": 8.5259, "step": 421 }, { "epoch": 0.026196536097833508, "grad_norm": 0.48408054815622714, "learning_rate": 8.73163666459756e-06, "loss": 8.5319, "step": 422 }, { "epoch": 0.02625861319759141, "grad_norm": 0.49094262198745353, "learning_rate": 8.752327746741156e-06, "loss": 8.5562, "step": 423 }, { "epoch": 0.02632069029734931, "grad_norm": 0.49224972709989173, "learning_rate": 8.773018828884751e-06, "loss": 8.4805, "step": 424 }, { "epoch": 0.026382767397107206, "grad_norm": 0.4967727669920683, "learning_rate": 8.793709911028348e-06, "loss": 8.407, "step": 425 }, { "epoch": 0.026444844496865107, "grad_norm": 0.5083610506266087, "learning_rate": 8.814400993171943e-06, "loss": 8.4609, "step": 426 }, { "epoch": 0.026506921596623007, "grad_norm": 0.4665539320806104, "learning_rate": 8.835092075315538e-06, "loss": 8.4762, "step": 427 }, { "epoch": 0.026568998696380904, "grad_norm": 0.4753734419388657, "learning_rate": 8.855783157459135e-06, "loss": 8.484, "step": 428 }, { "epoch": 0.026631075796138805, "grad_norm": 0.48288094921135527, "learning_rate": 8.876474239602732e-06, "loss": 8.4651, "step": 429 }, { "epoch": 0.026693152895896705, "grad_norm": 0.4998673471674396, "learning_rate": 8.897165321746327e-06, "loss": 8.4753, "step": 430 }, { "epoch": 0.026755229995654602, "grad_norm": 0.4801896843285203, "learning_rate": 8.917856403889923e-06, "loss": 8.5077, "step": 431 }, { "epoch": 0.026817307095412503, "grad_norm": 0.46376972981663767, "learning_rate": 8.93854748603352e-06, "loss": 8.4925, "step": 432 }, { "epoch": 0.026879384195170403, "grad_norm": 0.4607575760454989, "learning_rate": 8.959238568177115e-06, "loss": 8.4333, "step": 433 }, { "epoch": 0.0269414612949283, "grad_norm": 0.4709150541610171, "learning_rate": 8.979929650320712e-06, "loss": 8.5034, "step": 434 }, { "epoch": 0.0270035383946862, "grad_norm": 0.46071648552307104, "learning_rate": 9.000620732464309e-06, "loss": 8.4978, "step": 435 }, { "epoch": 0.027065615494444098, "grad_norm": 0.47243847834998903, "learning_rate": 9.021311814607906e-06, "loss": 8.4066, "step": 436 }, { "epoch": 0.027127692594202, "grad_norm": 0.4723750294566492, "learning_rate": 9.0420028967515e-06, "loss": 8.4637, "step": 437 }, { "epoch": 0.0271897696939599, "grad_norm": 0.46284236736888235, "learning_rate": 9.062693978895097e-06, "loss": 8.4191, "step": 438 }, { "epoch": 0.027251846793717796, "grad_norm": 0.46445962116905565, "learning_rate": 9.083385061038693e-06, "loss": 8.4345, "step": 439 }, { "epoch": 0.027313923893475697, "grad_norm": 0.480567239730785, "learning_rate": 9.10407614318229e-06, "loss": 8.4079, "step": 440 }, { "epoch": 0.027376000993233597, "grad_norm": 0.47942945454549996, "learning_rate": 9.124767225325884e-06, "loss": 8.4194, "step": 441 }, { "epoch": 0.027438078092991494, "grad_norm": 0.46739715222599904, "learning_rate": 9.145458307469481e-06, "loss": 8.4589, "step": 442 }, { "epoch": 0.027500155192749395, "grad_norm": 0.4840991557595579, "learning_rate": 9.166149389613076e-06, "loss": 8.4556, "step": 443 }, { "epoch": 0.027562232292507295, "grad_norm": 0.4635333496233422, "learning_rate": 9.186840471756673e-06, "loss": 8.4312, "step": 444 }, { "epoch": 0.027624309392265192, "grad_norm": 0.49162554070515485, "learning_rate": 9.20753155390027e-06, "loss": 8.4334, "step": 445 }, { "epoch": 0.027686386492023093, "grad_norm": 0.46131118885448824, "learning_rate": 9.228222636043865e-06, "loss": 8.3456, "step": 446 }, { "epoch": 0.027748463591780993, "grad_norm": 0.4856709555749849, "learning_rate": 9.248913718187462e-06, "loss": 8.4514, "step": 447 }, { "epoch": 0.02781054069153889, "grad_norm": 0.4538536297340652, "learning_rate": 9.269604800331058e-06, "loss": 8.3882, "step": 448 }, { "epoch": 0.02787261779129679, "grad_norm": 0.45640317031045935, "learning_rate": 9.290295882474653e-06, "loss": 8.3941, "step": 449 }, { "epoch": 0.02793469489105469, "grad_norm": 0.5024356823179029, "learning_rate": 9.31098696461825e-06, "loss": 8.3932, "step": 450 }, { "epoch": 0.02799677199081259, "grad_norm": 0.4704881074103701, "learning_rate": 9.331678046761847e-06, "loss": 8.3194, "step": 451 }, { "epoch": 0.02805884909057049, "grad_norm": 0.4716734638948722, "learning_rate": 9.352369128905442e-06, "loss": 8.3473, "step": 452 }, { "epoch": 0.02812092619032839, "grad_norm": 0.4498426665663715, "learning_rate": 9.373060211049039e-06, "loss": 8.3507, "step": 453 }, { "epoch": 0.028183003290086286, "grad_norm": 0.457378956176337, "learning_rate": 9.393751293192634e-06, "loss": 8.2754, "step": 454 }, { "epoch": 0.028245080389844187, "grad_norm": 0.48040919981571767, "learning_rate": 9.41444237533623e-06, "loss": 8.2058, "step": 455 }, { "epoch": 0.028307157489602087, "grad_norm": 0.45386618800159506, "learning_rate": 9.435133457479826e-06, "loss": 8.391, "step": 456 }, { "epoch": 0.028369234589359985, "grad_norm": 0.4637604279386781, "learning_rate": 9.455824539623423e-06, "loss": 8.2379, "step": 457 }, { "epoch": 0.028431311689117885, "grad_norm": 0.4514489747265513, "learning_rate": 9.47651562176702e-06, "loss": 8.3601, "step": 458 }, { "epoch": 0.028493388788875782, "grad_norm": 0.4516636698818145, "learning_rate": 9.497206703910614e-06, "loss": 8.3015, "step": 459 }, { "epoch": 0.028555465888633683, "grad_norm": 0.4424320724042304, "learning_rate": 9.517897786054211e-06, "loss": 8.2941, "step": 460 }, { "epoch": 0.028617542988391583, "grad_norm": 0.4636325764894508, "learning_rate": 9.538588868197808e-06, "loss": 8.2917, "step": 461 }, { "epoch": 0.02867962008814948, "grad_norm": 0.43507803071826384, "learning_rate": 9.559279950341403e-06, "loss": 8.3063, "step": 462 }, { "epoch": 0.02874169718790738, "grad_norm": 0.44802578624526296, "learning_rate": 9.579971032485e-06, "loss": 8.2341, "step": 463 }, { "epoch": 0.02880377428766528, "grad_norm": 0.4443907986822299, "learning_rate": 9.600662114628597e-06, "loss": 8.286, "step": 464 }, { "epoch": 0.02886585138742318, "grad_norm": 0.42834030849192756, "learning_rate": 9.621353196772192e-06, "loss": 8.2883, "step": 465 }, { "epoch": 0.02892792848718108, "grad_norm": 0.43309033011521586, "learning_rate": 9.642044278915788e-06, "loss": 8.2861, "step": 466 }, { "epoch": 0.02899000558693898, "grad_norm": 0.43927634180185454, "learning_rate": 9.662735361059385e-06, "loss": 8.1986, "step": 467 }, { "epoch": 0.029052082686696876, "grad_norm": 0.44347059224522734, "learning_rate": 9.68342644320298e-06, "loss": 8.2023, "step": 468 }, { "epoch": 0.029114159786454777, "grad_norm": 0.4200622501392413, "learning_rate": 9.704117525346575e-06, "loss": 8.2687, "step": 469 }, { "epoch": 0.029176236886212677, "grad_norm": 0.513358655228645, "learning_rate": 9.724808607490172e-06, "loss": 8.1418, "step": 470 }, { "epoch": 0.029238313985970574, "grad_norm": 0.43382487280210724, "learning_rate": 9.745499689633767e-06, "loss": 8.1929, "step": 471 }, { "epoch": 0.029300391085728475, "grad_norm": 0.4506612092283499, "learning_rate": 9.766190771777364e-06, "loss": 8.2968, "step": 472 }, { "epoch": 0.029362468185486375, "grad_norm": 0.422966815844011, "learning_rate": 9.78688185392096e-06, "loss": 8.1536, "step": 473 }, { "epoch": 0.029424545285244273, "grad_norm": 0.4455450856003806, "learning_rate": 9.807572936064556e-06, "loss": 8.2138, "step": 474 }, { "epoch": 0.029486622385002173, "grad_norm": 0.40461653820001287, "learning_rate": 9.828264018208153e-06, "loss": 8.2813, "step": 475 }, { "epoch": 0.029548699484760074, "grad_norm": 0.4569928568828521, "learning_rate": 9.84895510035175e-06, "loss": 8.1998, "step": 476 }, { "epoch": 0.02961077658451797, "grad_norm": 0.45629510525919087, "learning_rate": 9.869646182495346e-06, "loss": 8.247, "step": 477 }, { "epoch": 0.02967285368427587, "grad_norm": 0.4692818167950338, "learning_rate": 9.890337264638941e-06, "loss": 8.1146, "step": 478 }, { "epoch": 0.02973493078403377, "grad_norm": 0.4724295994874802, "learning_rate": 9.911028346782538e-06, "loss": 8.179, "step": 479 }, { "epoch": 0.02979700788379167, "grad_norm": 0.47724601319407794, "learning_rate": 9.931719428926135e-06, "loss": 8.1586, "step": 480 }, { "epoch": 0.02985908498354957, "grad_norm": 0.4182815807114649, "learning_rate": 9.95241051106973e-06, "loss": 8.2344, "step": 481 }, { "epoch": 0.029921162083307466, "grad_norm": 0.4918943319193421, "learning_rate": 9.973101593213325e-06, "loss": 8.1896, "step": 482 }, { "epoch": 0.029983239183065367, "grad_norm": 0.4403451301282902, "learning_rate": 9.993792675356922e-06, "loss": 8.1295, "step": 483 }, { "epoch": 0.030045316282823267, "grad_norm": 0.4814214178195018, "learning_rate": 1.0014483757500517e-05, "loss": 8.1019, "step": 484 }, { "epoch": 0.030107393382581164, "grad_norm": 0.4043421512627875, "learning_rate": 1.0035174839644113e-05, "loss": 8.2173, "step": 485 }, { "epoch": 0.030169470482339065, "grad_norm": 0.4654388145099353, "learning_rate": 1.005586592178771e-05, "loss": 8.1717, "step": 486 }, { "epoch": 0.030231547582096965, "grad_norm": 0.47290795300000044, "learning_rate": 1.0076557003931305e-05, "loss": 8.1225, "step": 487 }, { "epoch": 0.030293624681854862, "grad_norm": 0.42585109872706484, "learning_rate": 1.0097248086074902e-05, "loss": 8.1311, "step": 488 }, { "epoch": 0.030355701781612763, "grad_norm": 0.4315756494285295, "learning_rate": 1.0117939168218499e-05, "loss": 8.0852, "step": 489 }, { "epoch": 0.030417778881370663, "grad_norm": 0.5186359176663459, "learning_rate": 1.0138630250362094e-05, "loss": 8.166, "step": 490 }, { "epoch": 0.03047985598112856, "grad_norm": 0.4598113057560596, "learning_rate": 1.015932133250569e-05, "loss": 7.9981, "step": 491 }, { "epoch": 0.03054193308088646, "grad_norm": 0.4935822479559855, "learning_rate": 1.0180012414649287e-05, "loss": 8.1351, "step": 492 }, { "epoch": 0.03060401018064436, "grad_norm": 0.4128702561073896, "learning_rate": 1.0200703496792883e-05, "loss": 8.1096, "step": 493 }, { "epoch": 0.03066608728040226, "grad_norm": 0.4625270719595297, "learning_rate": 1.022139457893648e-05, "loss": 8.0676, "step": 494 }, { "epoch": 0.03072816438016016, "grad_norm": 0.3841699743473263, "learning_rate": 1.0242085661080076e-05, "loss": 8.1023, "step": 495 }, { "epoch": 0.03079024147991806, "grad_norm": 0.3940357836289235, "learning_rate": 1.0262776743223671e-05, "loss": 8.068, "step": 496 }, { "epoch": 0.030852318579675957, "grad_norm": 0.4187135744621261, "learning_rate": 1.0283467825367266e-05, "loss": 8.1145, "step": 497 }, { "epoch": 0.030914395679433857, "grad_norm": 0.43584329414412215, "learning_rate": 1.0304158907510863e-05, "loss": 7.9827, "step": 498 }, { "epoch": 0.030976472779191758, "grad_norm": 0.3831634275384667, "learning_rate": 1.032484998965446e-05, "loss": 8.0621, "step": 499 }, { "epoch": 0.031038549878949655, "grad_norm": 0.44824132316688864, "learning_rate": 1.0345541071798055e-05, "loss": 8.0646, "step": 500 }, { "epoch": 0.031100626978707555, "grad_norm": 0.40270639020776955, "learning_rate": 1.0366232153941652e-05, "loss": 8.0084, "step": 501 }, { "epoch": 0.031162704078465456, "grad_norm": 0.47273412961359296, "learning_rate": 1.0386923236085248e-05, "loss": 7.9818, "step": 502 }, { "epoch": 0.031224781178223353, "grad_norm": 0.515960150948714, "learning_rate": 1.0407614318228843e-05, "loss": 8.0176, "step": 503 }, { "epoch": 0.03128685827798125, "grad_norm": 0.39424678253767087, "learning_rate": 1.042830540037244e-05, "loss": 8.0614, "step": 504 }, { "epoch": 0.031348935377739154, "grad_norm": 0.45527102921027723, "learning_rate": 1.0448996482516037e-05, "loss": 7.8897, "step": 505 }, { "epoch": 0.03141101247749705, "grad_norm": 0.38107782603547785, "learning_rate": 1.0469687564659632e-05, "loss": 8.0512, "step": 506 }, { "epoch": 0.03147308957725495, "grad_norm": 0.5029203247383954, "learning_rate": 1.0490378646803229e-05, "loss": 8.0033, "step": 507 }, { "epoch": 0.03153516667701285, "grad_norm": 0.46452779209814404, "learning_rate": 1.0511069728946826e-05, "loss": 8.0052, "step": 508 }, { "epoch": 0.03159724377677075, "grad_norm": 0.40888341770702774, "learning_rate": 1.053176081109042e-05, "loss": 7.9962, "step": 509 }, { "epoch": 0.031659320876528646, "grad_norm": 0.6257758093893877, "learning_rate": 1.0552451893234016e-05, "loss": 8.0045, "step": 510 }, { "epoch": 0.03172139797628655, "grad_norm": 0.5762033486256058, "learning_rate": 1.0573142975377613e-05, "loss": 7.9793, "step": 511 }, { "epoch": 0.03178347507604445, "grad_norm": 0.3932564222814177, "learning_rate": 1.0593834057521208e-05, "loss": 8.0213, "step": 512 }, { "epoch": 0.031845552175802344, "grad_norm": 0.5737741526753805, "learning_rate": 1.0614525139664804e-05, "loss": 7.9838, "step": 513 }, { "epoch": 0.03190762927556025, "grad_norm": 0.5295510332875638, "learning_rate": 1.0635216221808401e-05, "loss": 7.9295, "step": 514 }, { "epoch": 0.031969706375318145, "grad_norm": 0.38850811522831663, "learning_rate": 1.0655907303951996e-05, "loss": 8.0403, "step": 515 }, { "epoch": 0.03203178347507604, "grad_norm": 0.36563649200246257, "learning_rate": 1.0676598386095593e-05, "loss": 7.9547, "step": 516 }, { "epoch": 0.032093860574833946, "grad_norm": 0.36658826589478005, "learning_rate": 1.069728946823919e-05, "loss": 7.9698, "step": 517 }, { "epoch": 0.03215593767459184, "grad_norm": 0.37205188935892974, "learning_rate": 1.0717980550382785e-05, "loss": 8.0085, "step": 518 }, { "epoch": 0.03221801477434974, "grad_norm": 0.4187023107218477, "learning_rate": 1.0738671632526382e-05, "loss": 7.867, "step": 519 }, { "epoch": 0.032280091874107644, "grad_norm": 0.4005147601966449, "learning_rate": 1.0759362714669978e-05, "loss": 7.9594, "step": 520 }, { "epoch": 0.03234216897386554, "grad_norm": 0.3825117408907714, "learning_rate": 1.0780053796813575e-05, "loss": 7.9458, "step": 521 }, { "epoch": 0.03240424607362344, "grad_norm": 0.3844157530727373, "learning_rate": 1.080074487895717e-05, "loss": 7.8583, "step": 522 }, { "epoch": 0.03246632317338134, "grad_norm": 0.44327389737560535, "learning_rate": 1.0821435961100767e-05, "loss": 7.9555, "step": 523 }, { "epoch": 0.03252840027313924, "grad_norm": 0.49422910900759104, "learning_rate": 1.0842127043244362e-05, "loss": 7.9041, "step": 524 }, { "epoch": 0.032590477372897136, "grad_norm": 0.43442100788961946, "learning_rate": 1.0862818125387957e-05, "loss": 7.9372, "step": 525 }, { "epoch": 0.03265255447265504, "grad_norm": 0.39972695031431976, "learning_rate": 1.0883509207531554e-05, "loss": 7.8474, "step": 526 }, { "epoch": 0.03271463157241294, "grad_norm": 0.39115909303352875, "learning_rate": 1.090420028967515e-05, "loss": 7.8537, "step": 527 }, { "epoch": 0.032776708672170835, "grad_norm": 0.43505504876436085, "learning_rate": 1.0924891371818746e-05, "loss": 7.8908, "step": 528 }, { "epoch": 0.03283878577192874, "grad_norm": 0.3672732192282661, "learning_rate": 1.0945582453962343e-05, "loss": 7.9635, "step": 529 }, { "epoch": 0.032900862871686636, "grad_norm": 0.3669904220562953, "learning_rate": 1.096627353610594e-05, "loss": 8.0036, "step": 530 }, { "epoch": 0.03296293997144453, "grad_norm": 0.3697960298128576, "learning_rate": 1.0986964618249534e-05, "loss": 7.8523, "step": 531 }, { "epoch": 0.03302501707120244, "grad_norm": 0.42873116621463214, "learning_rate": 1.1007655700393131e-05, "loss": 7.8306, "step": 532 }, { "epoch": 0.033087094170960334, "grad_norm": 0.44797704164078195, "learning_rate": 1.1028346782536728e-05, "loss": 7.8557, "step": 533 }, { "epoch": 0.03314917127071823, "grad_norm": 0.36357258835518425, "learning_rate": 1.1049037864680323e-05, "loss": 7.8533, "step": 534 }, { "epoch": 0.033211248370476135, "grad_norm": 0.39233512379754426, "learning_rate": 1.106972894682392e-05, "loss": 7.9619, "step": 535 }, { "epoch": 0.03327332547023403, "grad_norm": 0.46043386261819663, "learning_rate": 1.1090420028967517e-05, "loss": 7.9078, "step": 536 }, { "epoch": 0.03333540256999193, "grad_norm": 0.49948325873148264, "learning_rate": 1.1111111111111112e-05, "loss": 7.9155, "step": 537 }, { "epoch": 0.03339747966974983, "grad_norm": 0.45177488777159674, "learning_rate": 1.1131802193254708e-05, "loss": 7.9007, "step": 538 }, { "epoch": 0.03345955676950773, "grad_norm": 0.4144632811025528, "learning_rate": 1.1152493275398303e-05, "loss": 7.8806, "step": 539 }, { "epoch": 0.03352163386926563, "grad_norm": 0.4294502990751816, "learning_rate": 1.1173184357541899e-05, "loss": 7.7696, "step": 540 }, { "epoch": 0.033583710969023524, "grad_norm": 0.36573271998576423, "learning_rate": 1.1193875439685495e-05, "loss": 7.9012, "step": 541 }, { "epoch": 0.03364578806878143, "grad_norm": 0.46575423437764457, "learning_rate": 1.1214566521829092e-05, "loss": 7.8058, "step": 542 }, { "epoch": 0.033707865168539325, "grad_norm": 0.49023410143597634, "learning_rate": 1.1235257603972689e-05, "loss": 7.8906, "step": 543 }, { "epoch": 0.03376994226829722, "grad_norm": 0.532127021090339, "learning_rate": 1.1255948686116284e-05, "loss": 7.796, "step": 544 }, { "epoch": 0.033832019368055126, "grad_norm": 0.42568007850814543, "learning_rate": 1.127663976825988e-05, "loss": 7.7486, "step": 545 }, { "epoch": 0.03389409646781302, "grad_norm": 0.4144542748006535, "learning_rate": 1.1297330850403477e-05, "loss": 7.8513, "step": 546 }, { "epoch": 0.03395617356757092, "grad_norm": 0.48668995767484524, "learning_rate": 1.1318021932547073e-05, "loss": 7.7512, "step": 547 }, { "epoch": 0.034018250667328824, "grad_norm": 0.5570920753173513, "learning_rate": 1.133871301469067e-05, "loss": 7.8174, "step": 548 }, { "epoch": 0.03408032776708672, "grad_norm": 0.5818638144707654, "learning_rate": 1.1359404096834266e-05, "loss": 7.8496, "step": 549 }, { "epoch": 0.03414240486684462, "grad_norm": 0.5412018383173935, "learning_rate": 1.1380095178977861e-05, "loss": 7.9034, "step": 550 }, { "epoch": 0.03420448196660252, "grad_norm": 0.40453582189399473, "learning_rate": 1.1400786261121458e-05, "loss": 7.7597, "step": 551 }, { "epoch": 0.03426655906636042, "grad_norm": 0.6051028468854812, "learning_rate": 1.1421477343265053e-05, "loss": 7.7473, "step": 552 }, { "epoch": 0.034328636166118316, "grad_norm": 0.5993311979181878, "learning_rate": 1.1442168425408648e-05, "loss": 7.8328, "step": 553 }, { "epoch": 0.03439071326587622, "grad_norm": 0.41708678483567013, "learning_rate": 1.1462859507552245e-05, "loss": 7.7158, "step": 554 }, { "epoch": 0.03445279036563412, "grad_norm": 0.7937483689899658, "learning_rate": 1.1483550589695842e-05, "loss": 7.8065, "step": 555 }, { "epoch": 0.034514867465392014, "grad_norm": 0.4832778317823618, "learning_rate": 1.1504241671839437e-05, "loss": 7.7851, "step": 556 }, { "epoch": 0.03457694456514992, "grad_norm": 0.5688772300269135, "learning_rate": 1.1524932753983033e-05, "loss": 7.7443, "step": 557 }, { "epoch": 0.034639021664907815, "grad_norm": 0.513816726045406, "learning_rate": 1.154562383612663e-05, "loss": 7.8458, "step": 558 }, { "epoch": 0.03470109876466571, "grad_norm": 0.45583864441113237, "learning_rate": 1.1566314918270225e-05, "loss": 7.7537, "step": 559 }, { "epoch": 0.034763175864423616, "grad_norm": 0.5931791774015994, "learning_rate": 1.1587006000413822e-05, "loss": 7.7372, "step": 560 }, { "epoch": 0.03482525296418151, "grad_norm": 0.6540321611712212, "learning_rate": 1.1607697082557419e-05, "loss": 7.8245, "step": 561 }, { "epoch": 0.03488733006393941, "grad_norm": 0.6419221260132151, "learning_rate": 1.1628388164701016e-05, "loss": 7.6637, "step": 562 }, { "epoch": 0.034949407163697314, "grad_norm": 0.4921767672624984, "learning_rate": 1.164907924684461e-05, "loss": 7.6562, "step": 563 }, { "epoch": 0.03501148426345521, "grad_norm": 0.5818886038053509, "learning_rate": 1.1669770328988208e-05, "loss": 7.7001, "step": 564 }, { "epoch": 0.03507356136321311, "grad_norm": 0.503476869262062, "learning_rate": 1.1690461411131803e-05, "loss": 7.7081, "step": 565 }, { "epoch": 0.03513563846297101, "grad_norm": 0.6120718676110137, "learning_rate": 1.17111524932754e-05, "loss": 7.6849, "step": 566 }, { "epoch": 0.03519771556272891, "grad_norm": 0.39614846417843014, "learning_rate": 1.1731843575418994e-05, "loss": 7.7833, "step": 567 }, { "epoch": 0.03525979266248681, "grad_norm": 0.46758350824939643, "learning_rate": 1.1752534657562591e-05, "loss": 7.6635, "step": 568 }, { "epoch": 0.03532186976224471, "grad_norm": 0.3755300868142452, "learning_rate": 1.1773225739706186e-05, "loss": 7.8112, "step": 569 }, { "epoch": 0.03538394686200261, "grad_norm": 0.505911717597323, "learning_rate": 1.1793916821849783e-05, "loss": 7.7383, "step": 570 }, { "epoch": 0.035446023961760505, "grad_norm": 0.49893159263752085, "learning_rate": 1.181460790399338e-05, "loss": 7.6847, "step": 571 }, { "epoch": 0.03550810106151841, "grad_norm": 0.4537180111383333, "learning_rate": 1.1835298986136975e-05, "loss": 7.7845, "step": 572 }, { "epoch": 0.035570178161276306, "grad_norm": 0.444253381458652, "learning_rate": 1.1855990068280572e-05, "loss": 7.797, "step": 573 }, { "epoch": 0.0356322552610342, "grad_norm": 0.3289082692462616, "learning_rate": 1.1876681150424168e-05, "loss": 7.6546, "step": 574 }, { "epoch": 0.03569433236079211, "grad_norm": 0.3906858778245392, "learning_rate": 1.1897372232567764e-05, "loss": 7.713, "step": 575 }, { "epoch": 0.035756409460550004, "grad_norm": 0.37073529649543524, "learning_rate": 1.191806331471136e-05, "loss": 7.6879, "step": 576 }, { "epoch": 0.0358184865603079, "grad_norm": 0.31770750603281506, "learning_rate": 1.1938754396854957e-05, "loss": 7.7019, "step": 577 }, { "epoch": 0.035880563660065805, "grad_norm": 0.4546307997072711, "learning_rate": 1.1959445478998552e-05, "loss": 7.7469, "step": 578 }, { "epoch": 0.0359426407598237, "grad_norm": 0.5936130315506309, "learning_rate": 1.1980136561142149e-05, "loss": 7.724, "step": 579 }, { "epoch": 0.0360047178595816, "grad_norm": 0.5617452166904985, "learning_rate": 1.2000827643285744e-05, "loss": 7.6264, "step": 580 }, { "epoch": 0.0360667949593395, "grad_norm": 0.45747214024041466, "learning_rate": 1.2021518725429339e-05, "loss": 7.6271, "step": 581 }, { "epoch": 0.0361288720590974, "grad_norm": 0.39524855416918375, "learning_rate": 1.2042209807572936e-05, "loss": 7.6023, "step": 582 }, { "epoch": 0.0361909491588553, "grad_norm": 0.5188247638278163, "learning_rate": 1.2062900889716533e-05, "loss": 7.7334, "step": 583 }, { "epoch": 0.0362530262586132, "grad_norm": 0.6180486104220195, "learning_rate": 1.208359197186013e-05, "loss": 7.6173, "step": 584 }, { "epoch": 0.0363151033583711, "grad_norm": 0.529104528633554, "learning_rate": 1.2104283054003724e-05, "loss": 7.6682, "step": 585 }, { "epoch": 0.036377180458128995, "grad_norm": 0.3358227955361527, "learning_rate": 1.2124974136147321e-05, "loss": 7.6869, "step": 586 }, { "epoch": 0.03643925755788689, "grad_norm": 0.6004444482885151, "learning_rate": 1.2145665218290918e-05, "loss": 7.6768, "step": 587 }, { "epoch": 0.036501334657644796, "grad_norm": 0.6605178700310275, "learning_rate": 1.2166356300434513e-05, "loss": 7.6334, "step": 588 }, { "epoch": 0.03656341175740269, "grad_norm": 0.4963449908272664, "learning_rate": 1.218704738257811e-05, "loss": 7.5361, "step": 589 }, { "epoch": 0.03662548885716059, "grad_norm": 0.6471419893745864, "learning_rate": 1.2207738464721707e-05, "loss": 7.6019, "step": 590 }, { "epoch": 0.036687565956918494, "grad_norm": 0.8475080791302104, "learning_rate": 1.2228429546865302e-05, "loss": 7.5527, "step": 591 }, { "epoch": 0.03674964305667639, "grad_norm": 0.7237013399570063, "learning_rate": 1.2249120629008898e-05, "loss": 7.5943, "step": 592 }, { "epoch": 0.03681172015643429, "grad_norm": 0.3651948795281354, "learning_rate": 1.2269811711152494e-05, "loss": 7.6935, "step": 593 }, { "epoch": 0.03687379725619219, "grad_norm": 0.6547168579438377, "learning_rate": 1.229050279329609e-05, "loss": 7.585, "step": 594 }, { "epoch": 0.03693587435595009, "grad_norm": 0.6628909603875592, "learning_rate": 1.2311193875439685e-05, "loss": 7.687, "step": 595 }, { "epoch": 0.036997951455707986, "grad_norm": 0.4055158878040855, "learning_rate": 1.2331884957583282e-05, "loss": 7.6068, "step": 596 }, { "epoch": 0.03706002855546589, "grad_norm": 0.6597377646996789, "learning_rate": 1.2352576039726877e-05, "loss": 7.6454, "step": 597 }, { "epoch": 0.03712210565522379, "grad_norm": 0.6662040958371969, "learning_rate": 1.2373267121870474e-05, "loss": 7.5386, "step": 598 }, { "epoch": 0.037184182754981684, "grad_norm": 0.5040475809782998, "learning_rate": 1.239395820401407e-05, "loss": 7.4778, "step": 599 }, { "epoch": 0.03724625985473959, "grad_norm": 0.3969024370876985, "learning_rate": 1.2414649286157666e-05, "loss": 7.5916, "step": 600 }, { "epoch": 0.037308336954497485, "grad_norm": 0.46670168939652096, "learning_rate": 1.2435340368301263e-05, "loss": 7.6207, "step": 601 }, { "epoch": 0.03737041405425538, "grad_norm": 0.6701625691865501, "learning_rate": 1.245603145044486e-05, "loss": 7.505, "step": 602 }, { "epoch": 0.037432491154013287, "grad_norm": 0.37653792447507967, "learning_rate": 1.2476722532588454e-05, "loss": 7.5599, "step": 603 }, { "epoch": 0.037494568253771184, "grad_norm": 0.4560041379176168, "learning_rate": 1.2497413614732051e-05, "loss": 7.4664, "step": 604 }, { "epoch": 0.03755664535352908, "grad_norm": 0.38144585862831126, "learning_rate": 1.2518104696875646e-05, "loss": 7.5522, "step": 605 }, { "epoch": 0.037618722453286985, "grad_norm": 0.4812744487398056, "learning_rate": 1.2538795779019245e-05, "loss": 7.5023, "step": 606 }, { "epoch": 0.03768079955304488, "grad_norm": 0.5113841411544702, "learning_rate": 1.255948686116284e-05, "loss": 7.5591, "step": 607 }, { "epoch": 0.03774287665280278, "grad_norm": 0.4531845678440577, "learning_rate": 1.2580177943306435e-05, "loss": 7.5629, "step": 608 }, { "epoch": 0.03780495375256068, "grad_norm": 0.4277325808692128, "learning_rate": 1.2600869025450032e-05, "loss": 7.5219, "step": 609 }, { "epoch": 0.03786703085231858, "grad_norm": 0.43807077640961084, "learning_rate": 1.2621560107593627e-05, "loss": 7.4549, "step": 610 }, { "epoch": 0.03792910795207648, "grad_norm": 0.5301767053769978, "learning_rate": 1.2642251189737222e-05, "loss": 7.4737, "step": 611 }, { "epoch": 0.03799118505183438, "grad_norm": 0.4739182546109505, "learning_rate": 1.266294227188082e-05, "loss": 7.497, "step": 612 }, { "epoch": 0.03805326215159228, "grad_norm": 0.5252963977821652, "learning_rate": 1.2683633354024415e-05, "loss": 7.5383, "step": 613 }, { "epoch": 0.038115339251350175, "grad_norm": 0.5494994540936411, "learning_rate": 1.270432443616801e-05, "loss": 7.5586, "step": 614 }, { "epoch": 0.03817741635110808, "grad_norm": 0.554085848467477, "learning_rate": 1.2725015518311609e-05, "loss": 7.4996, "step": 615 }, { "epoch": 0.038239493450865976, "grad_norm": 0.6278811090404934, "learning_rate": 1.2745706600455204e-05, "loss": 7.5365, "step": 616 }, { "epoch": 0.03830157055062387, "grad_norm": 0.4648440545345559, "learning_rate": 1.2766397682598802e-05, "loss": 7.4635, "step": 617 }, { "epoch": 0.03836364765038178, "grad_norm": 0.614797721797641, "learning_rate": 1.2787088764742398e-05, "loss": 7.5001, "step": 618 }, { "epoch": 0.038425724750139674, "grad_norm": 0.4915211572012886, "learning_rate": 1.2807779846885993e-05, "loss": 7.5369, "step": 619 }, { "epoch": 0.03848780184989757, "grad_norm": 0.41010749163776344, "learning_rate": 1.282847092902959e-05, "loss": 7.4557, "step": 620 }, { "epoch": 0.038549878949655475, "grad_norm": 0.5511431215438598, "learning_rate": 1.2849162011173184e-05, "loss": 7.51, "step": 621 }, { "epoch": 0.03861195604941337, "grad_norm": 0.439368786666153, "learning_rate": 1.2869853093316781e-05, "loss": 7.5036, "step": 622 }, { "epoch": 0.03867403314917127, "grad_norm": 0.5128058323493533, "learning_rate": 1.2890544175460378e-05, "loss": 7.4831, "step": 623 }, { "epoch": 0.03873611024892917, "grad_norm": 0.6400820594572623, "learning_rate": 1.2911235257603973e-05, "loss": 7.4353, "step": 624 }, { "epoch": 0.03879818734868707, "grad_norm": 0.5567638863063821, "learning_rate": 1.2931926339747568e-05, "loss": 7.5458, "step": 625 }, { "epoch": 0.03886026444844497, "grad_norm": 0.552948468804613, "learning_rate": 1.2952617421891167e-05, "loss": 7.5069, "step": 626 }, { "epoch": 0.03892234154820287, "grad_norm": 0.5167486778153777, "learning_rate": 1.2973308504034762e-05, "loss": 7.431, "step": 627 }, { "epoch": 0.03898441864796077, "grad_norm": 0.43516528943738886, "learning_rate": 1.2993999586178357e-05, "loss": 7.4666, "step": 628 }, { "epoch": 0.039046495747718665, "grad_norm": 0.3989327475641078, "learning_rate": 1.3014690668321955e-05, "loss": 7.4767, "step": 629 }, { "epoch": 0.03910857284747657, "grad_norm": 0.5032055440843675, "learning_rate": 1.303538175046555e-05, "loss": 7.3705, "step": 630 }, { "epoch": 0.039170649947234466, "grad_norm": 0.4507014146052595, "learning_rate": 1.3056072832609145e-05, "loss": 7.4903, "step": 631 }, { "epoch": 0.03923272704699236, "grad_norm": 0.4355958645900174, "learning_rate": 1.3076763914752744e-05, "loss": 7.4814, "step": 632 }, { "epoch": 0.03929480414675026, "grad_norm": 0.5561154481406628, "learning_rate": 1.3097454996896339e-05, "loss": 7.3533, "step": 633 }, { "epoch": 0.039356881246508164, "grad_norm": 0.48949575824575214, "learning_rate": 1.3118146079039934e-05, "loss": 7.4321, "step": 634 }, { "epoch": 0.03941895834626606, "grad_norm": 0.5379329955589535, "learning_rate": 1.313883716118353e-05, "loss": 7.4191, "step": 635 }, { "epoch": 0.03948103544602396, "grad_norm": 0.42645920141989285, "learning_rate": 1.3159528243327126e-05, "loss": 7.4454, "step": 636 }, { "epoch": 0.03954311254578186, "grad_norm": 0.4656170895563781, "learning_rate": 1.3180219325470723e-05, "loss": 7.4481, "step": 637 }, { "epoch": 0.03960518964553976, "grad_norm": 0.5895489341172327, "learning_rate": 1.320091040761432e-05, "loss": 7.3612, "step": 638 }, { "epoch": 0.03966726674529766, "grad_norm": 0.3871493013888272, "learning_rate": 1.3221601489757914e-05, "loss": 7.4443, "step": 639 }, { "epoch": 0.03972934384505556, "grad_norm": 0.4974300828762502, "learning_rate": 1.324229257190151e-05, "loss": 7.4609, "step": 640 }, { "epoch": 0.03979142094481346, "grad_norm": 0.49034893931350665, "learning_rate": 1.3262983654045108e-05, "loss": 7.4656, "step": 641 }, { "epoch": 0.039853498044571355, "grad_norm": 0.4389204739051269, "learning_rate": 1.3283674736188703e-05, "loss": 7.3895, "step": 642 }, { "epoch": 0.03991557514432926, "grad_norm": 0.49211904261604367, "learning_rate": 1.3304365818332298e-05, "loss": 7.3263, "step": 643 }, { "epoch": 0.039977652244087156, "grad_norm": 0.3852877100436123, "learning_rate": 1.3325056900475897e-05, "loss": 7.4279, "step": 644 }, { "epoch": 0.04003972934384505, "grad_norm": 0.5624833104518708, "learning_rate": 1.3345747982619492e-05, "loss": 7.4381, "step": 645 }, { "epoch": 0.04010180644360296, "grad_norm": 0.6458758806754565, "learning_rate": 1.3366439064763087e-05, "loss": 7.3613, "step": 646 }, { "epoch": 0.040163883543360854, "grad_norm": 0.7848215860756936, "learning_rate": 1.3387130146906685e-05, "loss": 7.367, "step": 647 }, { "epoch": 0.04022596064311875, "grad_norm": 0.8261324825770506, "learning_rate": 1.340782122905028e-05, "loss": 7.4014, "step": 648 }, { "epoch": 0.040288037742876655, "grad_norm": 0.47044366990776065, "learning_rate": 1.3428512311193875e-05, "loss": 7.3363, "step": 649 }, { "epoch": 0.04035011484263455, "grad_norm": 0.762037591960611, "learning_rate": 1.3449203393337472e-05, "loss": 7.2459, "step": 650 }, { "epoch": 0.04041219194239245, "grad_norm": 0.9249732438934614, "learning_rate": 1.3469894475481067e-05, "loss": 7.4296, "step": 651 }, { "epoch": 0.04047426904215035, "grad_norm": 0.9706477987997915, "learning_rate": 1.3490585557624662e-05, "loss": 7.4111, "step": 652 }, { "epoch": 0.04053634614190825, "grad_norm": 1.2543939135332023, "learning_rate": 1.351127663976826e-05, "loss": 7.3144, "step": 653 }, { "epoch": 0.04059842324166615, "grad_norm": 0.5932337122456391, "learning_rate": 1.3531967721911856e-05, "loss": 7.4341, "step": 654 }, { "epoch": 0.04066050034142405, "grad_norm": 0.9693039571577609, "learning_rate": 1.3552658804055451e-05, "loss": 7.3279, "step": 655 }, { "epoch": 0.04072257744118195, "grad_norm": 1.0613697997143152, "learning_rate": 1.357334988619905e-05, "loss": 7.3467, "step": 656 }, { "epoch": 0.040784654540939845, "grad_norm": 0.9411754721003383, "learning_rate": 1.3594040968342644e-05, "loss": 7.4431, "step": 657 }, { "epoch": 0.04084673164069775, "grad_norm": 0.5827454403093775, "learning_rate": 1.3614732050486243e-05, "loss": 7.4143, "step": 658 }, { "epoch": 0.040908808740455646, "grad_norm": 0.6795125494984758, "learning_rate": 1.3635423132629838e-05, "loss": 7.361, "step": 659 }, { "epoch": 0.04097088584021354, "grad_norm": 0.6846225299592147, "learning_rate": 1.3656114214773433e-05, "loss": 7.3151, "step": 660 }, { "epoch": 0.04103296293997145, "grad_norm": 0.6225419584557479, "learning_rate": 1.3676805296917032e-05, "loss": 7.3004, "step": 661 }, { "epoch": 0.041095040039729344, "grad_norm": 0.5633592136305061, "learning_rate": 1.3697496379060627e-05, "loss": 7.3727, "step": 662 }, { "epoch": 0.04115711713948724, "grad_norm": 0.4518875127196953, "learning_rate": 1.3718187461204222e-05, "loss": 7.374, "step": 663 }, { "epoch": 0.041219194239245145, "grad_norm": 0.5865300297882706, "learning_rate": 1.3738878543347818e-05, "loss": 7.3774, "step": 664 }, { "epoch": 0.04128127133900304, "grad_norm": 0.5440781052842121, "learning_rate": 1.3759569625491414e-05, "loss": 7.2891, "step": 665 }, { "epoch": 0.04134334843876094, "grad_norm": 0.49776247169469834, "learning_rate": 1.3780260707635009e-05, "loss": 7.3869, "step": 666 }, { "epoch": 0.04140542553851884, "grad_norm": 0.47429222530674026, "learning_rate": 1.3800951789778607e-05, "loss": 7.3243, "step": 667 }, { "epoch": 0.04146750263827674, "grad_norm": 0.6231277556822363, "learning_rate": 1.3821642871922202e-05, "loss": 7.3114, "step": 668 }, { "epoch": 0.04152957973803464, "grad_norm": 0.5238031846494748, "learning_rate": 1.3842333954065797e-05, "loss": 7.2777, "step": 669 }, { "epoch": 0.04159165683779254, "grad_norm": 0.435172223494802, "learning_rate": 1.3863025036209396e-05, "loss": 7.2618, "step": 670 }, { "epoch": 0.04165373393755044, "grad_norm": 0.6483197852225244, "learning_rate": 1.388371611835299e-05, "loss": 7.276, "step": 671 }, { "epoch": 0.041715811037308335, "grad_norm": 0.6539016440900431, "learning_rate": 1.3904407200496586e-05, "loss": 7.3753, "step": 672 }, { "epoch": 0.04177788813706624, "grad_norm": 0.4542642281528886, "learning_rate": 1.3925098282640184e-05, "loss": 7.3461, "step": 673 }, { "epoch": 0.041839965236824136, "grad_norm": 0.5331311299551874, "learning_rate": 1.394578936478378e-05, "loss": 7.3088, "step": 674 }, { "epoch": 0.041902042336582034, "grad_norm": 0.6746013241073774, "learning_rate": 1.3966480446927374e-05, "loss": 7.2867, "step": 675 }, { "epoch": 0.04196411943633994, "grad_norm": 0.6712840830640713, "learning_rate": 1.3987171529070971e-05, "loss": 7.2952, "step": 676 }, { "epoch": 0.042026196536097835, "grad_norm": 0.4391500657211527, "learning_rate": 1.4007862611214568e-05, "loss": 7.3785, "step": 677 }, { "epoch": 0.04208827363585573, "grad_norm": 0.4653422463607755, "learning_rate": 1.4028553693358163e-05, "loss": 7.342, "step": 678 }, { "epoch": 0.042150350735613636, "grad_norm": 0.5012039650018894, "learning_rate": 1.404924477550176e-05, "loss": 7.2831, "step": 679 }, { "epoch": 0.04221242783537153, "grad_norm": 0.5140451155856532, "learning_rate": 1.4069935857645355e-05, "loss": 7.2593, "step": 680 }, { "epoch": 0.04227450493512943, "grad_norm": 0.5880838627875252, "learning_rate": 1.409062693978895e-05, "loss": 7.1316, "step": 681 }, { "epoch": 0.04233658203488733, "grad_norm": 0.5931982438152783, "learning_rate": 1.4111318021932548e-05, "loss": 7.3038, "step": 682 }, { "epoch": 0.04239865913464523, "grad_norm": 0.5226509675584371, "learning_rate": 1.4132009104076144e-05, "loss": 7.257, "step": 683 }, { "epoch": 0.04246073623440313, "grad_norm": 0.48644048472898765, "learning_rate": 1.4152700186219739e-05, "loss": 7.2594, "step": 684 }, { "epoch": 0.042522813334161025, "grad_norm": 0.6184769210857758, "learning_rate": 1.4173391268363337e-05, "loss": 7.201, "step": 685 }, { "epoch": 0.04258489043391893, "grad_norm": 0.7089150975536851, "learning_rate": 1.4194082350506932e-05, "loss": 7.2702, "step": 686 }, { "epoch": 0.042646967533676826, "grad_norm": 0.4603780163498605, "learning_rate": 1.4214773432650527e-05, "loss": 7.1988, "step": 687 }, { "epoch": 0.04270904463343472, "grad_norm": 0.5674230984749853, "learning_rate": 1.4235464514794126e-05, "loss": 7.2032, "step": 688 }, { "epoch": 0.04277112173319263, "grad_norm": 0.5469396825412729, "learning_rate": 1.425615559693772e-05, "loss": 7.2625, "step": 689 }, { "epoch": 0.042833198832950524, "grad_norm": 0.408184251850915, "learning_rate": 1.4276846679081316e-05, "loss": 7.1962, "step": 690 }, { "epoch": 0.04289527593270842, "grad_norm": 0.4957517962238053, "learning_rate": 1.4297537761224913e-05, "loss": 7.3117, "step": 691 }, { "epoch": 0.042957353032466325, "grad_norm": 0.5493497168719641, "learning_rate": 1.4318228843368508e-05, "loss": 7.1683, "step": 692 }, { "epoch": 0.04301943013222422, "grad_norm": 0.44148847776176325, "learning_rate": 1.4338919925512104e-05, "loss": 7.2572, "step": 693 }, { "epoch": 0.04308150723198212, "grad_norm": 0.5094865641670764, "learning_rate": 1.4359611007655701e-05, "loss": 7.2676, "step": 694 }, { "epoch": 0.04314358433174002, "grad_norm": 0.45885389380676034, "learning_rate": 1.4380302089799296e-05, "loss": 7.3444, "step": 695 }, { "epoch": 0.04320566143149792, "grad_norm": 0.48868835262873567, "learning_rate": 1.4400993171942891e-05, "loss": 7.2432, "step": 696 }, { "epoch": 0.04326773853125582, "grad_norm": 0.853417397440073, "learning_rate": 1.442168425408649e-05, "loss": 7.2844, "step": 697 }, { "epoch": 0.04332981563101372, "grad_norm": 1.0294692530327259, "learning_rate": 1.4442375336230085e-05, "loss": 7.2779, "step": 698 }, { "epoch": 0.04339189273077162, "grad_norm": 0.9450175242842729, "learning_rate": 1.446306641837368e-05, "loss": 7.1786, "step": 699 }, { "epoch": 0.043453969830529515, "grad_norm": 0.9419768570853335, "learning_rate": 1.4483757500517278e-05, "loss": 7.0828, "step": 700 }, { "epoch": 0.04351604693028742, "grad_norm": 0.8327970964964713, "learning_rate": 1.4504448582660874e-05, "loss": 7.2392, "step": 701 }, { "epoch": 0.043578124030045316, "grad_norm": 0.5550364350482168, "learning_rate": 1.4525139664804472e-05, "loss": 7.2056, "step": 702 }, { "epoch": 0.04364020112980321, "grad_norm": 0.48719266471547634, "learning_rate": 1.4545830746948067e-05, "loss": 7.2688, "step": 703 }, { "epoch": 0.04370227822956112, "grad_norm": 0.775685021462652, "learning_rate": 1.4566521829091662e-05, "loss": 7.0827, "step": 704 }, { "epoch": 0.043764355329319014, "grad_norm": 0.879432080116946, "learning_rate": 1.4587212911235259e-05, "loss": 7.1231, "step": 705 }, { "epoch": 0.04382643242907691, "grad_norm": 0.7051637127356922, "learning_rate": 1.4607903993378854e-05, "loss": 7.3033, "step": 706 }, { "epoch": 0.043888509528834815, "grad_norm": 0.5207575615761044, "learning_rate": 1.4628595075522449e-05, "loss": 7.1679, "step": 707 }, { "epoch": 0.04395058662859271, "grad_norm": 0.7099041817711336, "learning_rate": 1.4649286157666048e-05, "loss": 7.1636, "step": 708 }, { "epoch": 0.04401266372835061, "grad_norm": 0.6594566159110197, "learning_rate": 1.4669977239809643e-05, "loss": 7.1553, "step": 709 }, { "epoch": 0.04407474082810851, "grad_norm": 0.658833911770368, "learning_rate": 1.4690668321953238e-05, "loss": 7.188, "step": 710 }, { "epoch": 0.04413681792786641, "grad_norm": 0.5692856310020259, "learning_rate": 1.4711359404096836e-05, "loss": 7.1522, "step": 711 }, { "epoch": 0.04419889502762431, "grad_norm": 0.76978046235296, "learning_rate": 1.4732050486240431e-05, "loss": 7.1465, "step": 712 }, { "epoch": 0.04426097212738221, "grad_norm": 0.6946492192561556, "learning_rate": 1.4752741568384026e-05, "loss": 7.0861, "step": 713 }, { "epoch": 0.04432304922714011, "grad_norm": 0.6022043148399292, "learning_rate": 1.4773432650527625e-05, "loss": 7.106, "step": 714 }, { "epoch": 0.044385126326898006, "grad_norm": 0.7420891546042127, "learning_rate": 1.479412373267122e-05, "loss": 7.0986, "step": 715 }, { "epoch": 0.04444720342665591, "grad_norm": 0.9279952154385612, "learning_rate": 1.4814814814814815e-05, "loss": 7.2062, "step": 716 }, { "epoch": 0.04450928052641381, "grad_norm": 0.6144086837460361, "learning_rate": 1.4835505896958413e-05, "loss": 7.125, "step": 717 }, { "epoch": 0.044571357626171704, "grad_norm": 0.8806643047996555, "learning_rate": 1.4856196979102009e-05, "loss": 7.1093, "step": 718 }, { "epoch": 0.04463343472592961, "grad_norm": 0.7458736234754183, "learning_rate": 1.4876888061245604e-05, "loss": 7.1374, "step": 719 }, { "epoch": 0.044695511825687505, "grad_norm": 0.643350062161033, "learning_rate": 1.48975791433892e-05, "loss": 7.0355, "step": 720 }, { "epoch": 0.0447575889254454, "grad_norm": 1.0536298747165451, "learning_rate": 1.4918270225532795e-05, "loss": 7.1915, "step": 721 }, { "epoch": 0.044819666025203306, "grad_norm": 0.9052576353798814, "learning_rate": 1.493896130767639e-05, "loss": 7.1028, "step": 722 }, { "epoch": 0.0448817431249612, "grad_norm": 0.8958747469770081, "learning_rate": 1.4959652389819989e-05, "loss": 7.0001, "step": 723 }, { "epoch": 0.0449438202247191, "grad_norm": 0.5077049728405479, "learning_rate": 1.4980343471963584e-05, "loss": 7.1506, "step": 724 }, { "epoch": 0.045005897324477004, "grad_norm": 0.7028716160602714, "learning_rate": 1.5001034554107179e-05, "loss": 7.1262, "step": 725 }, { "epoch": 0.0450679744242349, "grad_norm": 0.6861509612278723, "learning_rate": 1.5021725636250778e-05, "loss": 7.1656, "step": 726 }, { "epoch": 0.0451300515239928, "grad_norm": 0.8025208720645103, "learning_rate": 1.5042416718394373e-05, "loss": 7.1125, "step": 727 }, { "epoch": 0.045192128623750695, "grad_norm": 0.6313391830217977, "learning_rate": 1.5063107800537968e-05, "loss": 7.1193, "step": 728 }, { "epoch": 0.0452542057235086, "grad_norm": 0.6595897614082898, "learning_rate": 1.5083798882681566e-05, "loss": 7.0479, "step": 729 }, { "epoch": 0.045316282823266496, "grad_norm": 0.576016550626002, "learning_rate": 1.5104489964825161e-05, "loss": 7.0417, "step": 730 }, { "epoch": 0.04537835992302439, "grad_norm": 0.5472129594946232, "learning_rate": 1.5125181046968756e-05, "loss": 7.1629, "step": 731 }, { "epoch": 0.0454404370227823, "grad_norm": 0.4948988212870483, "learning_rate": 1.5145872129112355e-05, "loss": 7.0812, "step": 732 }, { "epoch": 0.045502514122540194, "grad_norm": 0.5596170455830827, "learning_rate": 1.516656321125595e-05, "loss": 7.1448, "step": 733 }, { "epoch": 0.04556459122229809, "grad_norm": 0.6722848440611587, "learning_rate": 1.5187254293399545e-05, "loss": 7.0898, "step": 734 }, { "epoch": 0.045626668322055995, "grad_norm": 0.45690873941231624, "learning_rate": 1.5207945375543142e-05, "loss": 7.085, "step": 735 }, { "epoch": 0.04568874542181389, "grad_norm": 0.6039009715808191, "learning_rate": 1.5228636457686737e-05, "loss": 7.1197, "step": 736 }, { "epoch": 0.04575082252157179, "grad_norm": 0.4908964997380279, "learning_rate": 1.5249327539830332e-05, "loss": 7.1146, "step": 737 }, { "epoch": 0.04581289962132969, "grad_norm": 0.578911890164595, "learning_rate": 1.5270018621973932e-05, "loss": 7.1122, "step": 738 }, { "epoch": 0.04587497672108759, "grad_norm": 0.5852773126065124, "learning_rate": 1.5290709704117527e-05, "loss": 7.2051, "step": 739 }, { "epoch": 0.04593705382084549, "grad_norm": 0.6184985083220496, "learning_rate": 1.5311400786261122e-05, "loss": 7.0736, "step": 740 }, { "epoch": 0.04599913092060339, "grad_norm": 0.48772763720698914, "learning_rate": 1.5332091868404717e-05, "loss": 7.0727, "step": 741 }, { "epoch": 0.04606120802036129, "grad_norm": 0.6360744188332823, "learning_rate": 1.5352782950548312e-05, "loss": 7.1504, "step": 742 }, { "epoch": 0.046123285120119185, "grad_norm": 0.6027674558984368, "learning_rate": 1.537347403269191e-05, "loss": 7.0988, "step": 743 }, { "epoch": 0.04618536221987709, "grad_norm": 0.5744540711977059, "learning_rate": 1.5394165114835506e-05, "loss": 7.0217, "step": 744 }, { "epoch": 0.046247439319634986, "grad_norm": 0.7755778588686221, "learning_rate": 1.54148561969791e-05, "loss": 7.0621, "step": 745 }, { "epoch": 0.046309516419392883, "grad_norm": 0.8884198177159989, "learning_rate": 1.54355472791227e-05, "loss": 7.0248, "step": 746 }, { "epoch": 0.04637159351915079, "grad_norm": 0.645758921940776, "learning_rate": 1.5456238361266295e-05, "loss": 7.0713, "step": 747 }, { "epoch": 0.046433670618908685, "grad_norm": 0.6631917293242243, "learning_rate": 1.547692944340989e-05, "loss": 7.0278, "step": 748 }, { "epoch": 0.04649574771866658, "grad_norm": 0.875173755832317, "learning_rate": 1.5497620525553488e-05, "loss": 7.1208, "step": 749 }, { "epoch": 0.046557824818424486, "grad_norm": 0.827008015897066, "learning_rate": 1.5518311607697083e-05, "loss": 7.2045, "step": 750 }, { "epoch": 0.04661990191818238, "grad_norm": 0.8337716539307615, "learning_rate": 1.5539002689840678e-05, "loss": 6.9933, "step": 751 }, { "epoch": 0.04668197901794028, "grad_norm": 0.697635114633727, "learning_rate": 1.5559693771984277e-05, "loss": 7.0017, "step": 752 }, { "epoch": 0.046744056117698184, "grad_norm": 0.7556832278323894, "learning_rate": 1.5580384854127872e-05, "loss": 7.015, "step": 753 }, { "epoch": 0.04680613321745608, "grad_norm": 0.8886177465701145, "learning_rate": 1.5601075936271467e-05, "loss": 7.0295, "step": 754 }, { "epoch": 0.04686821031721398, "grad_norm": 0.7394435956515243, "learning_rate": 1.5621767018415065e-05, "loss": 7.1537, "step": 755 }, { "epoch": 0.04693028741697188, "grad_norm": 0.6106261101266537, "learning_rate": 1.564245810055866e-05, "loss": 7.0865, "step": 756 }, { "epoch": 0.04699236451672978, "grad_norm": 0.8472381123023699, "learning_rate": 1.5663149182702255e-05, "loss": 6.9556, "step": 757 }, { "epoch": 0.047054441616487676, "grad_norm": 1.185766812431538, "learning_rate": 1.5683840264845854e-05, "loss": 6.9694, "step": 758 }, { "epoch": 0.04711651871624558, "grad_norm": 0.7119391375444072, "learning_rate": 1.570453134698945e-05, "loss": 7.0543, "step": 759 }, { "epoch": 0.04717859581600348, "grad_norm": 0.6703224208070958, "learning_rate": 1.5725222429133044e-05, "loss": 7.0615, "step": 760 }, { "epoch": 0.047240672915761374, "grad_norm": 0.8284013754826339, "learning_rate": 1.5745913511276643e-05, "loss": 7.0012, "step": 761 }, { "epoch": 0.04730275001551928, "grad_norm": 0.8530454060520632, "learning_rate": 1.5766604593420238e-05, "loss": 6.9961, "step": 762 }, { "epoch": 0.047364827115277175, "grad_norm": 0.7301435919851602, "learning_rate": 1.5787295675563833e-05, "loss": 7.0266, "step": 763 }, { "epoch": 0.04742690421503507, "grad_norm": 0.8374985353257766, "learning_rate": 1.580798675770743e-05, "loss": 7.0822, "step": 764 }, { "epoch": 0.047488981314792976, "grad_norm": 0.6341844686005461, "learning_rate": 1.5828677839851026e-05, "loss": 6.885, "step": 765 }, { "epoch": 0.04755105841455087, "grad_norm": 0.7077367224140106, "learning_rate": 1.584936892199462e-05, "loss": 6.9717, "step": 766 }, { "epoch": 0.04761313551430877, "grad_norm": 0.6918784431089277, "learning_rate": 1.5870060004138216e-05, "loss": 7.1129, "step": 767 }, { "epoch": 0.047675212614066674, "grad_norm": 0.6447310398134709, "learning_rate": 1.5890751086281815e-05, "loss": 6.9216, "step": 768 }, { "epoch": 0.04773728971382457, "grad_norm": 0.7041670600807249, "learning_rate": 1.591144216842541e-05, "loss": 6.9458, "step": 769 }, { "epoch": 0.04779936681358247, "grad_norm": 0.8470357214523424, "learning_rate": 1.5932133250569005e-05, "loss": 7.025, "step": 770 }, { "epoch": 0.04786144391334037, "grad_norm": 0.6583471848040674, "learning_rate": 1.59528243327126e-05, "loss": 7.0203, "step": 771 }, { "epoch": 0.04792352101309827, "grad_norm": 0.6670607256548092, "learning_rate": 1.5973515414856195e-05, "loss": 7.0101, "step": 772 }, { "epoch": 0.047985598112856166, "grad_norm": 0.6628333890321955, "learning_rate": 1.5994206496999794e-05, "loss": 6.9933, "step": 773 }, { "epoch": 0.04804767521261406, "grad_norm": 0.6377058510252744, "learning_rate": 1.601489757914339e-05, "loss": 7.0321, "step": 774 }, { "epoch": 0.04810975231237197, "grad_norm": 0.5988118089218619, "learning_rate": 1.6035588661286984e-05, "loss": 6.9262, "step": 775 }, { "epoch": 0.048171829412129864, "grad_norm": 0.7050151849694055, "learning_rate": 1.6056279743430582e-05, "loss": 7.0287, "step": 776 }, { "epoch": 0.04823390651188776, "grad_norm": 0.6260236463529147, "learning_rate": 1.6076970825574177e-05, "loss": 6.9304, "step": 777 }, { "epoch": 0.048295983611645665, "grad_norm": 0.6438111646538571, "learning_rate": 1.6097661907717772e-05, "loss": 6.9735, "step": 778 }, { "epoch": 0.04835806071140356, "grad_norm": 0.6463565757720914, "learning_rate": 1.611835298986137e-05, "loss": 7.0876, "step": 779 }, { "epoch": 0.04842013781116146, "grad_norm": 0.5998944527456201, "learning_rate": 1.6139044072004966e-05, "loss": 6.883, "step": 780 }, { "epoch": 0.04848221491091936, "grad_norm": 0.609843074320273, "learning_rate": 1.615973515414856e-05, "loss": 6.9762, "step": 781 }, { "epoch": 0.04854429201067726, "grad_norm": 0.7668786198189388, "learning_rate": 1.618042623629216e-05, "loss": 6.9052, "step": 782 }, { "epoch": 0.04860636911043516, "grad_norm": 0.5769433497877312, "learning_rate": 1.6201117318435755e-05, "loss": 6.9351, "step": 783 }, { "epoch": 0.04866844621019306, "grad_norm": 0.7129833840417652, "learning_rate": 1.622180840057935e-05, "loss": 6.9508, "step": 784 }, { "epoch": 0.04873052330995096, "grad_norm": 0.5791644469111144, "learning_rate": 1.6242499482722948e-05, "loss": 6.9742, "step": 785 }, { "epoch": 0.048792600409708856, "grad_norm": 0.538545664828405, "learning_rate": 1.6263190564866543e-05, "loss": 6.8958, "step": 786 }, { "epoch": 0.04885467750946676, "grad_norm": 0.5150379538207694, "learning_rate": 1.628388164701014e-05, "loss": 7.0141, "step": 787 }, { "epoch": 0.04891675460922466, "grad_norm": 0.5794408875153336, "learning_rate": 1.6304572729153737e-05, "loss": 6.847, "step": 788 }, { "epoch": 0.048978831708982554, "grad_norm": 0.5770527716805847, "learning_rate": 1.6325263811297332e-05, "loss": 6.8944, "step": 789 }, { "epoch": 0.04904090880874046, "grad_norm": 0.697526450219651, "learning_rate": 1.634595489344093e-05, "loss": 6.9433, "step": 790 }, { "epoch": 0.049102985908498355, "grad_norm": 0.7870704508734028, "learning_rate": 1.6366645975584525e-05, "loss": 6.9826, "step": 791 }, { "epoch": 0.04916506300825625, "grad_norm": 0.6649715932715995, "learning_rate": 1.638733705772812e-05, "loss": 7.025, "step": 792 }, { "epoch": 0.049227140108014156, "grad_norm": 0.852292124498596, "learning_rate": 1.640802813987172e-05, "loss": 6.9789, "step": 793 }, { "epoch": 0.04928921720777205, "grad_norm": 0.9055425579279806, "learning_rate": 1.6428719222015314e-05, "loss": 6.9877, "step": 794 }, { "epoch": 0.04935129430752995, "grad_norm": 1.1430554558491397, "learning_rate": 1.644941030415891e-05, "loss": 6.9558, "step": 795 }, { "epoch": 0.049413371407287854, "grad_norm": 1.042459370669848, "learning_rate": 1.6470101386302504e-05, "loss": 6.9299, "step": 796 }, { "epoch": 0.04947544850704575, "grad_norm": 0.7695346580468633, "learning_rate": 1.64907924684461e-05, "loss": 7.0138, "step": 797 }, { "epoch": 0.04953752560680365, "grad_norm": 0.7711673906250207, "learning_rate": 1.6511483550589694e-05, "loss": 6.9273, "step": 798 }, { "epoch": 0.04959960270656155, "grad_norm": 0.7387296737740447, "learning_rate": 1.6532174632733293e-05, "loss": 6.9393, "step": 799 }, { "epoch": 0.04966167980631945, "grad_norm": 0.6974572524777511, "learning_rate": 1.6552865714876888e-05, "loss": 6.9078, "step": 800 }, { "epoch": 0.049723756906077346, "grad_norm": 0.7635380464253768, "learning_rate": 1.6573556797020483e-05, "loss": 6.8484, "step": 801 }, { "epoch": 0.04978583400583525, "grad_norm": 0.8426229225385471, "learning_rate": 1.659424787916408e-05, "loss": 6.8166, "step": 802 }, { "epoch": 0.04984791110559315, "grad_norm": 0.8835893774219526, "learning_rate": 1.6614938961307676e-05, "loss": 6.8344, "step": 803 }, { "epoch": 0.049909988205351044, "grad_norm": 0.7992797917745985, "learning_rate": 1.663563004345127e-05, "loss": 6.8758, "step": 804 }, { "epoch": 0.04997206530510895, "grad_norm": 0.6691689444638159, "learning_rate": 1.665632112559487e-05, "loss": 6.9442, "step": 805 }, { "epoch": 0.050034142404866845, "grad_norm": 0.758868199840296, "learning_rate": 1.6677012207738465e-05, "loss": 6.7892, "step": 806 }, { "epoch": 0.05009621950462474, "grad_norm": 0.940992050540681, "learning_rate": 1.669770328988206e-05, "loss": 6.8638, "step": 807 }, { "epoch": 0.050158296604382646, "grad_norm": 0.7947712598574843, "learning_rate": 1.671839437202566e-05, "loss": 6.8749, "step": 808 }, { "epoch": 0.05022037370414054, "grad_norm": 0.6990092299265378, "learning_rate": 1.6739085454169254e-05, "loss": 6.9509, "step": 809 }, { "epoch": 0.05028245080389844, "grad_norm": 0.6850260524946737, "learning_rate": 1.675977653631285e-05, "loss": 6.9288, "step": 810 }, { "epoch": 0.050344527903656344, "grad_norm": 0.7144076158181696, "learning_rate": 1.6780467618456447e-05, "loss": 6.8236, "step": 811 }, { "epoch": 0.05040660500341424, "grad_norm": 0.6393796865789307, "learning_rate": 1.6801158700600042e-05, "loss": 6.8869, "step": 812 }, { "epoch": 0.05046868210317214, "grad_norm": 0.5729146668370341, "learning_rate": 1.6821849782743637e-05, "loss": 6.897, "step": 813 }, { "epoch": 0.05053075920293004, "grad_norm": 0.545002100791732, "learning_rate": 1.6842540864887236e-05, "loss": 6.9285, "step": 814 }, { "epoch": 0.05059283630268794, "grad_norm": 0.5919178754026279, "learning_rate": 1.686323194703083e-05, "loss": 6.9336, "step": 815 }, { "epoch": 0.050654913402445836, "grad_norm": 0.6333995497955803, "learning_rate": 1.6883923029174426e-05, "loss": 6.8914, "step": 816 }, { "epoch": 0.05071699050220374, "grad_norm": 0.8140609126157898, "learning_rate": 1.6904614111318024e-05, "loss": 6.7859, "step": 817 }, { "epoch": 0.05077906760196164, "grad_norm": 0.6804286817979796, "learning_rate": 1.692530519346162e-05, "loss": 6.7716, "step": 818 }, { "epoch": 0.050841144701719534, "grad_norm": 0.5880234273668032, "learning_rate": 1.6945996275605215e-05, "loss": 6.8163, "step": 819 }, { "epoch": 0.05090322180147743, "grad_norm": 0.674341509308264, "learning_rate": 1.6966687357748813e-05, "loss": 6.8676, "step": 820 }, { "epoch": 0.050965298901235336, "grad_norm": 0.49984473559564724, "learning_rate": 1.6987378439892408e-05, "loss": 6.7887, "step": 821 }, { "epoch": 0.05102737600099323, "grad_norm": 0.7172283568203218, "learning_rate": 1.7008069522036003e-05, "loss": 6.7449, "step": 822 }, { "epoch": 0.05108945310075113, "grad_norm": 0.7141014789651989, "learning_rate": 1.70287606041796e-05, "loss": 6.8317, "step": 823 }, { "epoch": 0.051151530200509034, "grad_norm": 0.6452376915749122, "learning_rate": 1.7049451686323197e-05, "loss": 6.8969, "step": 824 }, { "epoch": 0.05121360730026693, "grad_norm": 0.6124090565097756, "learning_rate": 1.7070142768466792e-05, "loss": 6.8957, "step": 825 }, { "epoch": 0.05127568440002483, "grad_norm": 0.5850875284460828, "learning_rate": 1.7090833850610387e-05, "loss": 6.7627, "step": 826 }, { "epoch": 0.05133776149978273, "grad_norm": 0.712695536635644, "learning_rate": 1.7111524932753982e-05, "loss": 6.8817, "step": 827 }, { "epoch": 0.05139983859954063, "grad_norm": 0.7485212211531084, "learning_rate": 1.713221601489758e-05, "loss": 6.8187, "step": 828 }, { "epoch": 0.051461915699298526, "grad_norm": 0.8287570628559998, "learning_rate": 1.7152907097041175e-05, "loss": 6.8399, "step": 829 }, { "epoch": 0.05152399279905643, "grad_norm": 0.8092404787660153, "learning_rate": 1.717359817918477e-05, "loss": 6.8069, "step": 830 }, { "epoch": 0.05158606989881433, "grad_norm": 0.6959054968489516, "learning_rate": 1.719428926132837e-05, "loss": 6.7855, "step": 831 }, { "epoch": 0.051648146998572224, "grad_norm": 0.5865473053579338, "learning_rate": 1.7214980343471964e-05, "loss": 6.8738, "step": 832 }, { "epoch": 0.05171022409833013, "grad_norm": 0.688769879997508, "learning_rate": 1.723567142561556e-05, "loss": 6.8711, "step": 833 }, { "epoch": 0.051772301198088025, "grad_norm": 0.87538867219965, "learning_rate": 1.7256362507759158e-05, "loss": 6.8183, "step": 834 }, { "epoch": 0.05183437829784592, "grad_norm": 0.8354917032753305, "learning_rate": 1.7277053589902753e-05, "loss": 6.8378, "step": 835 }, { "epoch": 0.051896455397603826, "grad_norm": 0.6494122727448317, "learning_rate": 1.7297744672046348e-05, "loss": 6.7944, "step": 836 }, { "epoch": 0.05195853249736172, "grad_norm": 0.5529273137857197, "learning_rate": 1.7318435754189946e-05, "loss": 6.8362, "step": 837 }, { "epoch": 0.05202060959711962, "grad_norm": 0.8591911119177158, "learning_rate": 1.733912683633354e-05, "loss": 6.8273, "step": 838 }, { "epoch": 0.052082686696877524, "grad_norm": 0.8935551544354996, "learning_rate": 1.7359817918477136e-05, "loss": 6.7357, "step": 839 }, { "epoch": 0.05214476379663542, "grad_norm": 0.5953388788982019, "learning_rate": 1.7380509000620735e-05, "loss": 6.832, "step": 840 }, { "epoch": 0.05220684089639332, "grad_norm": 0.6243154353004899, "learning_rate": 1.740120008276433e-05, "loss": 6.9009, "step": 841 }, { "epoch": 0.05226891799615122, "grad_norm": 0.7216808850764971, "learning_rate": 1.7421891164907925e-05, "loss": 6.8817, "step": 842 }, { "epoch": 0.05233099509590912, "grad_norm": 0.7163350127067187, "learning_rate": 1.7442582247051523e-05, "loss": 6.7581, "step": 843 }, { "epoch": 0.052393072195667016, "grad_norm": 0.7566379214244904, "learning_rate": 1.746327332919512e-05, "loss": 6.7364, "step": 844 }, { "epoch": 0.05245514929542492, "grad_norm": 0.4834752339713694, "learning_rate": 1.7483964411338714e-05, "loss": 6.7969, "step": 845 }, { "epoch": 0.05251722639518282, "grad_norm": 0.6610763052340737, "learning_rate": 1.7504655493482312e-05, "loss": 6.6213, "step": 846 }, { "epoch": 0.052579303494940714, "grad_norm": 0.7447908619340582, "learning_rate": 1.7525346575625907e-05, "loss": 6.7893, "step": 847 }, { "epoch": 0.05264138059469862, "grad_norm": 0.6109955343014588, "learning_rate": 1.7546037657769502e-05, "loss": 6.7298, "step": 848 }, { "epoch": 0.052703457694456515, "grad_norm": 0.6533588802888158, "learning_rate": 1.75667287399131e-05, "loss": 6.9355, "step": 849 }, { "epoch": 0.05276553479421441, "grad_norm": 0.8952504560932275, "learning_rate": 1.7587419822056696e-05, "loss": 6.7256, "step": 850 }, { "epoch": 0.052827611893972316, "grad_norm": 0.8963565111914648, "learning_rate": 1.760811090420029e-05, "loss": 6.7385, "step": 851 }, { "epoch": 0.05288968899373021, "grad_norm": 0.8178532749341572, "learning_rate": 1.7628801986343886e-05, "loss": 6.7819, "step": 852 }, { "epoch": 0.05295176609348811, "grad_norm": 0.7804591563019868, "learning_rate": 1.764949306848748e-05, "loss": 6.8363, "step": 853 }, { "epoch": 0.053013843193246014, "grad_norm": 0.5873326032656023, "learning_rate": 1.7670184150631076e-05, "loss": 6.609, "step": 854 }, { "epoch": 0.05307592029300391, "grad_norm": 0.7644731849373667, "learning_rate": 1.7690875232774675e-05, "loss": 6.7693, "step": 855 }, { "epoch": 0.05313799739276181, "grad_norm": 0.8135734949141916, "learning_rate": 1.771156631491827e-05, "loss": 6.6728, "step": 856 }, { "epoch": 0.05320007449251971, "grad_norm": 0.9472531032717814, "learning_rate": 1.7732257397061865e-05, "loss": 6.73, "step": 857 }, { "epoch": 0.05326215159227761, "grad_norm": 0.9963533250267088, "learning_rate": 1.7752948479205463e-05, "loss": 6.8188, "step": 858 }, { "epoch": 0.05332422869203551, "grad_norm": 0.9725632527332961, "learning_rate": 1.7773639561349058e-05, "loss": 6.8367, "step": 859 }, { "epoch": 0.05338630579179341, "grad_norm": 0.8005784312871421, "learning_rate": 1.7794330643492653e-05, "loss": 6.6726, "step": 860 }, { "epoch": 0.05344838289155131, "grad_norm": 0.8482034209279787, "learning_rate": 1.7815021725636252e-05, "loss": 6.8302, "step": 861 }, { "epoch": 0.053510459991309205, "grad_norm": 0.8398867219089045, "learning_rate": 1.7835712807779847e-05, "loss": 6.7562, "step": 862 }, { "epoch": 0.05357253709106711, "grad_norm": 0.6968069624937833, "learning_rate": 1.7856403889923442e-05, "loss": 6.7116, "step": 863 }, { "epoch": 0.053634614190825006, "grad_norm": 0.8487268534047739, "learning_rate": 1.787709497206704e-05, "loss": 6.7039, "step": 864 }, { "epoch": 0.0536966912905829, "grad_norm": 0.9392048275654536, "learning_rate": 1.7897786054210635e-05, "loss": 6.677, "step": 865 }, { "epoch": 0.05375876839034081, "grad_norm": 0.6361229337979907, "learning_rate": 1.791847713635423e-05, "loss": 6.7727, "step": 866 }, { "epoch": 0.053820845490098704, "grad_norm": 0.8864041413097694, "learning_rate": 1.793916821849783e-05, "loss": 6.6317, "step": 867 }, { "epoch": 0.0538829225898566, "grad_norm": 1.0064473331844952, "learning_rate": 1.7959859300641424e-05, "loss": 6.7273, "step": 868 }, { "epoch": 0.0539449996896145, "grad_norm": 0.8856917159769092, "learning_rate": 1.798055038278502e-05, "loss": 6.7256, "step": 869 }, { "epoch": 0.0540070767893724, "grad_norm": 0.7218476591671154, "learning_rate": 1.8001241464928618e-05, "loss": 6.6649, "step": 870 }, { "epoch": 0.0540691538891303, "grad_norm": 0.8603683625918073, "learning_rate": 1.8021932547072213e-05, "loss": 6.6957, "step": 871 }, { "epoch": 0.054131230988888196, "grad_norm": 0.8324399490578188, "learning_rate": 1.804262362921581e-05, "loss": 6.7356, "step": 872 }, { "epoch": 0.0541933080886461, "grad_norm": 0.6921172069739838, "learning_rate": 1.8063314711359406e-05, "loss": 6.8427, "step": 873 }, { "epoch": 0.054255385188404, "grad_norm": 0.8247186672367772, "learning_rate": 1.8084005793503e-05, "loss": 6.75, "step": 874 }, { "epoch": 0.054317462288161894, "grad_norm": 0.8117966542020175, "learning_rate": 1.81046968756466e-05, "loss": 6.7016, "step": 875 }, { "epoch": 0.0543795393879198, "grad_norm": 0.5010982016260705, "learning_rate": 1.8125387957790195e-05, "loss": 6.6315, "step": 876 }, { "epoch": 0.054441616487677695, "grad_norm": 0.601733270161384, "learning_rate": 1.814607903993379e-05, "loss": 6.7345, "step": 877 }, { "epoch": 0.05450369358743559, "grad_norm": 0.602142524999544, "learning_rate": 1.8166770122077385e-05, "loss": 6.652, "step": 878 }, { "epoch": 0.054565770687193496, "grad_norm": 0.6713529821069258, "learning_rate": 1.8187461204220984e-05, "loss": 6.6866, "step": 879 }, { "epoch": 0.05462784778695139, "grad_norm": 0.4946780241578601, "learning_rate": 1.820815228636458e-05, "loss": 6.7421, "step": 880 }, { "epoch": 0.05468992488670929, "grad_norm": 0.5528792605890318, "learning_rate": 1.8228843368508174e-05, "loss": 6.8446, "step": 881 }, { "epoch": 0.054752001986467194, "grad_norm": 0.5635520844780504, "learning_rate": 1.824953445065177e-05, "loss": 6.7793, "step": 882 }, { "epoch": 0.05481407908622509, "grad_norm": 0.6215112830186672, "learning_rate": 1.8270225532795364e-05, "loss": 6.6972, "step": 883 }, { "epoch": 0.05487615618598299, "grad_norm": 0.591569964023743, "learning_rate": 1.8290916614938962e-05, "loss": 6.6552, "step": 884 }, { "epoch": 0.05493823328574089, "grad_norm": 0.7342950087465853, "learning_rate": 1.8311607697082557e-05, "loss": 6.6814, "step": 885 }, { "epoch": 0.05500031038549879, "grad_norm": 0.8001261194474873, "learning_rate": 1.8332298779226152e-05, "loss": 6.6458, "step": 886 }, { "epoch": 0.055062387485256686, "grad_norm": 0.6846423145254799, "learning_rate": 1.835298986136975e-05, "loss": 6.7007, "step": 887 }, { "epoch": 0.05512446458501459, "grad_norm": 0.5520904078720071, "learning_rate": 1.8373680943513346e-05, "loss": 6.6477, "step": 888 }, { "epoch": 0.05518654168477249, "grad_norm": 0.6259530668565494, "learning_rate": 1.839437202565694e-05, "loss": 6.6568, "step": 889 }, { "epoch": 0.055248618784530384, "grad_norm": 0.7358767393262372, "learning_rate": 1.841506310780054e-05, "loss": 6.7094, "step": 890 }, { "epoch": 0.05531069588428829, "grad_norm": 0.5685205394956546, "learning_rate": 1.8435754189944135e-05, "loss": 6.5652, "step": 891 }, { "epoch": 0.055372772984046185, "grad_norm": 0.6037161287240663, "learning_rate": 1.845644527208773e-05, "loss": 6.7092, "step": 892 }, { "epoch": 0.05543485008380408, "grad_norm": 0.5951512516339625, "learning_rate": 1.8477136354231328e-05, "loss": 6.6956, "step": 893 }, { "epoch": 0.055496927183561987, "grad_norm": 0.608670707998233, "learning_rate": 1.8497827436374923e-05, "loss": 6.5829, "step": 894 }, { "epoch": 0.055559004283319884, "grad_norm": 0.6269484190006646, "learning_rate": 1.8518518518518518e-05, "loss": 6.6702, "step": 895 }, { "epoch": 0.05562108138307778, "grad_norm": 0.6515367736128503, "learning_rate": 1.8539209600662117e-05, "loss": 6.7031, "step": 896 }, { "epoch": 0.055683158482835685, "grad_norm": 0.6917279890368435, "learning_rate": 1.8559900682805712e-05, "loss": 6.7312, "step": 897 }, { "epoch": 0.05574523558259358, "grad_norm": 0.7720610575817279, "learning_rate": 1.8580591764949307e-05, "loss": 6.5083, "step": 898 }, { "epoch": 0.05580731268235148, "grad_norm": 0.7855843283262298, "learning_rate": 1.8601282847092905e-05, "loss": 6.615, "step": 899 }, { "epoch": 0.05586938978210938, "grad_norm": 1.0232242624162138, "learning_rate": 1.86219739292365e-05, "loss": 6.5942, "step": 900 }, { "epoch": 0.05593146688186728, "grad_norm": 1.1739777074578683, "learning_rate": 1.8642665011380096e-05, "loss": 6.6124, "step": 901 }, { "epoch": 0.05599354398162518, "grad_norm": 0.758982237352943, "learning_rate": 1.8663356093523694e-05, "loss": 6.6022, "step": 902 }, { "epoch": 0.05605562108138308, "grad_norm": 0.8076606563507693, "learning_rate": 1.868404717566729e-05, "loss": 6.5921, "step": 903 }, { "epoch": 0.05611769818114098, "grad_norm": 1.1029733591433293, "learning_rate": 1.8704738257810884e-05, "loss": 6.6826, "step": 904 }, { "epoch": 0.056179775280898875, "grad_norm": 1.1235693736131616, "learning_rate": 1.8725429339954483e-05, "loss": 6.6065, "step": 905 }, { "epoch": 0.05624185238065678, "grad_norm": 1.1491216580929395, "learning_rate": 1.8746120422098078e-05, "loss": 6.6035, "step": 906 }, { "epoch": 0.056303929480414676, "grad_norm": 0.7462823839046026, "learning_rate": 1.8766811504241673e-05, "loss": 6.6827, "step": 907 }, { "epoch": 0.05636600658017257, "grad_norm": 0.5938906001212162, "learning_rate": 1.8787502586385268e-05, "loss": 6.5931, "step": 908 }, { "epoch": 0.05642808367993048, "grad_norm": 0.8258157237498193, "learning_rate": 1.8808193668528863e-05, "loss": 6.7055, "step": 909 }, { "epoch": 0.056490160779688374, "grad_norm": 0.8580269383956627, "learning_rate": 1.882888475067246e-05, "loss": 6.4989, "step": 910 }, { "epoch": 0.05655223787944627, "grad_norm": 0.8200265295110077, "learning_rate": 1.8849575832816056e-05, "loss": 6.6727, "step": 911 }, { "epoch": 0.056614314979204175, "grad_norm": 0.797276190202266, "learning_rate": 1.887026691495965e-05, "loss": 6.7445, "step": 912 }, { "epoch": 0.05667639207896207, "grad_norm": 0.7498869413355683, "learning_rate": 1.889095799710325e-05, "loss": 6.6977, "step": 913 }, { "epoch": 0.05673846917871997, "grad_norm": 0.7087369965398181, "learning_rate": 1.8911649079246845e-05, "loss": 6.6384, "step": 914 }, { "epoch": 0.056800546278477866, "grad_norm": 0.8447800812277658, "learning_rate": 1.893234016139044e-05, "loss": 6.5344, "step": 915 }, { "epoch": 0.05686262337823577, "grad_norm": 0.7379014587414807, "learning_rate": 1.895303124353404e-05, "loss": 6.6539, "step": 916 }, { "epoch": 0.05692470047799367, "grad_norm": 0.7114849288534478, "learning_rate": 1.8973722325677634e-05, "loss": 6.6931, "step": 917 }, { "epoch": 0.056986777577751564, "grad_norm": 0.6966289316761525, "learning_rate": 1.899441340782123e-05, "loss": 6.6222, "step": 918 }, { "epoch": 0.05704885467750947, "grad_norm": 0.8195476053967803, "learning_rate": 1.9015104489964827e-05, "loss": 6.5771, "step": 919 }, { "epoch": 0.057110931777267365, "grad_norm": 1.0959392979859766, "learning_rate": 1.9035795572108422e-05, "loss": 6.5109, "step": 920 }, { "epoch": 0.05717300887702526, "grad_norm": 1.0361598493983681, "learning_rate": 1.9056486654252017e-05, "loss": 6.5606, "step": 921 }, { "epoch": 0.057235085976783166, "grad_norm": 0.7937959011953366, "learning_rate": 1.9077177736395616e-05, "loss": 6.6863, "step": 922 }, { "epoch": 0.05729716307654106, "grad_norm": 0.8015805604918922, "learning_rate": 1.909786881853921e-05, "loss": 6.5059, "step": 923 }, { "epoch": 0.05735924017629896, "grad_norm": 0.952783179352787, "learning_rate": 1.9118559900682806e-05, "loss": 6.4717, "step": 924 }, { "epoch": 0.057421317276056864, "grad_norm": 1.0797313388696597, "learning_rate": 1.9139250982826404e-05, "loss": 6.6308, "step": 925 }, { "epoch": 0.05748339437581476, "grad_norm": 0.9435714578056146, "learning_rate": 1.915994206497e-05, "loss": 6.5471, "step": 926 }, { "epoch": 0.05754547147557266, "grad_norm": 0.7887082579459854, "learning_rate": 1.9180633147113595e-05, "loss": 6.6176, "step": 927 }, { "epoch": 0.05760754857533056, "grad_norm": 0.9800043107930565, "learning_rate": 1.9201324229257193e-05, "loss": 6.5856, "step": 928 }, { "epoch": 0.05766962567508846, "grad_norm": 1.087909472317663, "learning_rate": 1.9222015311400788e-05, "loss": 6.6206, "step": 929 }, { "epoch": 0.05773170277484636, "grad_norm": 0.8831914577869218, "learning_rate": 1.9242706393544383e-05, "loss": 6.5456, "step": 930 }, { "epoch": 0.05779377987460426, "grad_norm": 1.1839158529491522, "learning_rate": 1.926339747568798e-05, "loss": 6.61, "step": 931 }, { "epoch": 0.05785585697436216, "grad_norm": 0.8203341889920113, "learning_rate": 1.9284088557831577e-05, "loss": 6.54, "step": 932 }, { "epoch": 0.057917934074120055, "grad_norm": 0.8846377181340555, "learning_rate": 1.9304779639975172e-05, "loss": 6.7028, "step": 933 }, { "epoch": 0.05798001117387796, "grad_norm": 0.874977179638418, "learning_rate": 1.932547072211877e-05, "loss": 6.6129, "step": 934 }, { "epoch": 0.058042088273635856, "grad_norm": 0.6518249504574788, "learning_rate": 1.9346161804262365e-05, "loss": 6.6114, "step": 935 }, { "epoch": 0.05810416537339375, "grad_norm": 0.9108068408469593, "learning_rate": 1.936685288640596e-05, "loss": 6.6765, "step": 936 }, { "epoch": 0.05816624247315166, "grad_norm": 0.9267876221311395, "learning_rate": 1.9387543968549556e-05, "loss": 6.5517, "step": 937 }, { "epoch": 0.058228319572909554, "grad_norm": 0.892612047615092, "learning_rate": 1.940823505069315e-05, "loss": 6.5285, "step": 938 }, { "epoch": 0.05829039667266745, "grad_norm": 1.1182625140029976, "learning_rate": 1.9428926132836746e-05, "loss": 6.5648, "step": 939 }, { "epoch": 0.058352473772425355, "grad_norm": 0.8853158818647248, "learning_rate": 1.9449617214980344e-05, "loss": 6.5541, "step": 940 }, { "epoch": 0.05841455087218325, "grad_norm": 0.9246148918706049, "learning_rate": 1.947030829712394e-05, "loss": 6.6485, "step": 941 }, { "epoch": 0.05847662797194115, "grad_norm": 0.7142298412609301, "learning_rate": 1.9490999379267534e-05, "loss": 6.5434, "step": 942 }, { "epoch": 0.05853870507169905, "grad_norm": 0.8422161217746489, "learning_rate": 1.9511690461411133e-05, "loss": 6.4736, "step": 943 }, { "epoch": 0.05860078217145695, "grad_norm": 0.8426008169260327, "learning_rate": 1.9532381543554728e-05, "loss": 6.561, "step": 944 }, { "epoch": 0.05866285927121485, "grad_norm": 0.9557697311787372, "learning_rate": 1.9553072625698323e-05, "loss": 6.6527, "step": 945 }, { "epoch": 0.05872493637097275, "grad_norm": 0.8350987370870325, "learning_rate": 1.957376370784192e-05, "loss": 6.6779, "step": 946 }, { "epoch": 0.05878701347073065, "grad_norm": 0.634721819026308, "learning_rate": 1.9594454789985516e-05, "loss": 6.5707, "step": 947 }, { "epoch": 0.058849090570488545, "grad_norm": 0.812542694143833, "learning_rate": 1.961514587212911e-05, "loss": 6.5036, "step": 948 }, { "epoch": 0.05891116767024645, "grad_norm": 0.8342956390928961, "learning_rate": 1.963583695427271e-05, "loss": 6.4727, "step": 949 }, { "epoch": 0.058973244770004346, "grad_norm": 0.847950053826931, "learning_rate": 1.9656528036416305e-05, "loss": 6.5488, "step": 950 }, { "epoch": 0.05903532186976224, "grad_norm": 0.8139149010341857, "learning_rate": 1.96772191185599e-05, "loss": 6.5097, "step": 951 }, { "epoch": 0.05909739896952015, "grad_norm": 0.8267434309752365, "learning_rate": 1.96979102007035e-05, "loss": 6.514, "step": 952 }, { "epoch": 0.059159476069278044, "grad_norm": 0.6796870834524698, "learning_rate": 1.9718601282847094e-05, "loss": 6.5517, "step": 953 }, { "epoch": 0.05922155316903594, "grad_norm": 0.723389784225963, "learning_rate": 1.9739292364990692e-05, "loss": 6.5745, "step": 954 }, { "epoch": 0.059283630268793845, "grad_norm": 0.8602139279519362, "learning_rate": 1.9759983447134287e-05, "loss": 6.5158, "step": 955 }, { "epoch": 0.05934570736855174, "grad_norm": 0.6941817443724451, "learning_rate": 1.9780674529277882e-05, "loss": 6.5756, "step": 956 }, { "epoch": 0.05940778446830964, "grad_norm": 0.6189458883757725, "learning_rate": 1.980136561142148e-05, "loss": 6.584, "step": 957 }, { "epoch": 0.05946986156806754, "grad_norm": 0.7661102469914817, "learning_rate": 1.9822056693565076e-05, "loss": 6.4557, "step": 958 }, { "epoch": 0.05953193866782544, "grad_norm": 0.692775598764073, "learning_rate": 1.984274777570867e-05, "loss": 6.4977, "step": 959 }, { "epoch": 0.05959401576758334, "grad_norm": 0.6558294992645852, "learning_rate": 1.986343885785227e-05, "loss": 6.4832, "step": 960 }, { "epoch": 0.059656092867341234, "grad_norm": 0.8142881523659903, "learning_rate": 1.9884129939995864e-05, "loss": 6.4597, "step": 961 }, { "epoch": 0.05971816996709914, "grad_norm": 0.8351838466841573, "learning_rate": 1.990482102213946e-05, "loss": 6.4827, "step": 962 }, { "epoch": 0.059780247066857035, "grad_norm": 0.8787022671819724, "learning_rate": 1.9925512104283055e-05, "loss": 6.3797, "step": 963 }, { "epoch": 0.05984232416661493, "grad_norm": 0.9296642589068312, "learning_rate": 1.994620318642665e-05, "loss": 6.5274, "step": 964 }, { "epoch": 0.059904401266372836, "grad_norm": 0.8649702164037227, "learning_rate": 1.9966894268570248e-05, "loss": 6.5567, "step": 965 }, { "epoch": 0.059966478366130734, "grad_norm": 0.9737807977333242, "learning_rate": 1.9987585350713843e-05, "loss": 6.3938, "step": 966 }, { "epoch": 0.06002855546588863, "grad_norm": 1.050579803329746, "learning_rate": 2.000827643285744e-05, "loss": 6.4623, "step": 967 }, { "epoch": 0.060090632565646535, "grad_norm": 0.8743803350518983, "learning_rate": 2.0028967515001033e-05, "loss": 6.4559, "step": 968 }, { "epoch": 0.06015270966540443, "grad_norm": 0.9256542456705547, "learning_rate": 2.0049658597144632e-05, "loss": 6.4395, "step": 969 }, { "epoch": 0.06021478676516233, "grad_norm": 0.9149833344677204, "learning_rate": 2.0070349679288227e-05, "loss": 6.4551, "step": 970 }, { "epoch": 0.06027686386492023, "grad_norm": 0.7917789120323255, "learning_rate": 2.0091040761431822e-05, "loss": 6.4565, "step": 971 }, { "epoch": 0.06033894096467813, "grad_norm": 0.9104168563830937, "learning_rate": 2.011173184357542e-05, "loss": 6.61, "step": 972 }, { "epoch": 0.06040101806443603, "grad_norm": 0.8728891524634748, "learning_rate": 2.0132422925719016e-05, "loss": 6.5678, "step": 973 }, { "epoch": 0.06046309516419393, "grad_norm": 0.8776094151612558, "learning_rate": 2.015311400786261e-05, "loss": 6.4913, "step": 974 }, { "epoch": 0.06052517226395183, "grad_norm": 0.8230134729995244, "learning_rate": 2.017380509000621e-05, "loss": 6.3794, "step": 975 }, { "epoch": 0.060587249363709725, "grad_norm": 0.885737476946198, "learning_rate": 2.0194496172149804e-05, "loss": 6.4642, "step": 976 }, { "epoch": 0.06064932646346763, "grad_norm": 0.8996342856386566, "learning_rate": 2.02151872542934e-05, "loss": 6.5391, "step": 977 }, { "epoch": 0.060711403563225526, "grad_norm": 0.9153761919496478, "learning_rate": 2.0235878336436998e-05, "loss": 6.4681, "step": 978 }, { "epoch": 0.06077348066298342, "grad_norm": 0.9190728223753434, "learning_rate": 2.0256569418580593e-05, "loss": 6.4988, "step": 979 }, { "epoch": 0.06083555776274133, "grad_norm": 0.9074983548971771, "learning_rate": 2.0277260500724188e-05, "loss": 6.5477, "step": 980 }, { "epoch": 0.060897634862499224, "grad_norm": 1.0427063458271628, "learning_rate": 2.0297951582867786e-05, "loss": 6.405, "step": 981 }, { "epoch": 0.06095971196225712, "grad_norm": 1.2288721777090321, "learning_rate": 2.031864266501138e-05, "loss": 6.4715, "step": 982 }, { "epoch": 0.061021789062015025, "grad_norm": 0.781839148586063, "learning_rate": 2.0339333747154976e-05, "loss": 6.4066, "step": 983 }, { "epoch": 0.06108386616177292, "grad_norm": 1.0455163708412616, "learning_rate": 2.0360024829298575e-05, "loss": 6.5198, "step": 984 }, { "epoch": 0.06114594326153082, "grad_norm": 1.0613323415167726, "learning_rate": 2.038071591144217e-05, "loss": 6.4474, "step": 985 }, { "epoch": 0.06120802036128872, "grad_norm": 1.353906396160349, "learning_rate": 2.0401406993585765e-05, "loss": 6.4324, "step": 986 }, { "epoch": 0.06127009746104662, "grad_norm": 0.8858510952481929, "learning_rate": 2.0422098075729364e-05, "loss": 6.4255, "step": 987 }, { "epoch": 0.06133217456080452, "grad_norm": 1.263158537708713, "learning_rate": 2.044278915787296e-05, "loss": 6.423, "step": 988 }, { "epoch": 0.06139425166056242, "grad_norm": 0.8001474617116217, "learning_rate": 2.0463480240016554e-05, "loss": 6.4614, "step": 989 }, { "epoch": 0.06145632876032032, "grad_norm": 0.8999386395623965, "learning_rate": 2.0484171322160152e-05, "loss": 6.4731, "step": 990 }, { "epoch": 0.061518405860078215, "grad_norm": 0.9277193409554884, "learning_rate": 2.0504862404303747e-05, "loss": 6.371, "step": 991 }, { "epoch": 0.06158048295983612, "grad_norm": 1.0449575257681456, "learning_rate": 2.0525553486447342e-05, "loss": 6.4544, "step": 992 }, { "epoch": 0.061642560059594016, "grad_norm": 1.0055486995563991, "learning_rate": 2.0546244568590937e-05, "loss": 6.4185, "step": 993 }, { "epoch": 0.06170463715935191, "grad_norm": 0.8837463705805774, "learning_rate": 2.0566935650734532e-05, "loss": 6.412, "step": 994 }, { "epoch": 0.06176671425910982, "grad_norm": 0.890959258681872, "learning_rate": 2.0587626732878128e-05, "loss": 6.3335, "step": 995 }, { "epoch": 0.061828791358867714, "grad_norm": 1.043638949853963, "learning_rate": 2.0608317815021726e-05, "loss": 6.4237, "step": 996 }, { "epoch": 0.06189086845862561, "grad_norm": 0.9885003672508904, "learning_rate": 2.062900889716532e-05, "loss": 6.4505, "step": 997 }, { "epoch": 0.061952945558383515, "grad_norm": 1.009895419664513, "learning_rate": 2.064969997930892e-05, "loss": 6.5567, "step": 998 }, { "epoch": 0.06201502265814141, "grad_norm": 0.8819576614537599, "learning_rate": 2.0670391061452515e-05, "loss": 6.3891, "step": 999 }, { "epoch": 0.06207709975789931, "grad_norm": 0.928192443195414, "learning_rate": 2.069108214359611e-05, "loss": 6.3281, "step": 1000 }, { "epoch": 0.06213917685765721, "grad_norm": 0.9322650344426068, "learning_rate": 2.0711773225739708e-05, "loss": 6.4014, "step": 1001 }, { "epoch": 0.06220125395741511, "grad_norm": 0.8146317466794791, "learning_rate": 2.0732464307883303e-05, "loss": 6.472, "step": 1002 }, { "epoch": 0.06226333105717301, "grad_norm": 0.7031921778482529, "learning_rate": 2.07531553900269e-05, "loss": 6.3636, "step": 1003 }, { "epoch": 0.06232540815693091, "grad_norm": 0.870659998171809, "learning_rate": 2.0773846472170497e-05, "loss": 6.3814, "step": 1004 }, { "epoch": 0.06238748525668881, "grad_norm": 0.989087947861585, "learning_rate": 2.0794537554314092e-05, "loss": 6.4729, "step": 1005 }, { "epoch": 0.062449562356446706, "grad_norm": 0.902942706391127, "learning_rate": 2.0815228636457687e-05, "loss": 6.3499, "step": 1006 }, { "epoch": 0.06251163945620461, "grad_norm": 0.7996763096064066, "learning_rate": 2.0835919718601285e-05, "loss": 6.3701, "step": 1007 }, { "epoch": 0.0625737165559625, "grad_norm": 0.8912458430938297, "learning_rate": 2.085661080074488e-05, "loss": 6.4085, "step": 1008 }, { "epoch": 0.0626357936557204, "grad_norm": 1.0100157555210496, "learning_rate": 2.0877301882888476e-05, "loss": 6.3955, "step": 1009 }, { "epoch": 0.06269787075547831, "grad_norm": 0.913547501206479, "learning_rate": 2.0897992965032074e-05, "loss": 6.4328, "step": 1010 }, { "epoch": 0.0627599478552362, "grad_norm": 0.7393468604130229, "learning_rate": 2.091868404717567e-05, "loss": 6.3473, "step": 1011 }, { "epoch": 0.0628220249549941, "grad_norm": 0.9554842220123558, "learning_rate": 2.0939375129319264e-05, "loss": 6.5093, "step": 1012 }, { "epoch": 0.062884102054752, "grad_norm": 0.9249285541470159, "learning_rate": 2.0960066211462863e-05, "loss": 6.3975, "step": 1013 }, { "epoch": 0.0629461791545099, "grad_norm": 1.0260499995840473, "learning_rate": 2.0980757293606458e-05, "loss": 6.3884, "step": 1014 }, { "epoch": 0.0630082562542678, "grad_norm": 0.907012388573759, "learning_rate": 2.1001448375750053e-05, "loss": 6.3754, "step": 1015 }, { "epoch": 0.0630703333540257, "grad_norm": 0.8322918980629559, "learning_rate": 2.102213945789365e-05, "loss": 6.4596, "step": 1016 }, { "epoch": 0.0631324104537836, "grad_norm": 0.8806201076980833, "learning_rate": 2.1042830540037246e-05, "loss": 6.3998, "step": 1017 }, { "epoch": 0.0631944875535415, "grad_norm": 0.8894214528994892, "learning_rate": 2.106352162218084e-05, "loss": 6.4354, "step": 1018 }, { "epoch": 0.0632565646532994, "grad_norm": 0.8391959597473055, "learning_rate": 2.1084212704324436e-05, "loss": 6.3747, "step": 1019 }, { "epoch": 0.06331864175305729, "grad_norm": 0.8273187707402688, "learning_rate": 2.110490378646803e-05, "loss": 6.3839, "step": 1020 }, { "epoch": 0.0633807188528152, "grad_norm": 0.7682933241783904, "learning_rate": 2.112559486861163e-05, "loss": 6.3852, "step": 1021 }, { "epoch": 0.0634427959525731, "grad_norm": 0.7061175770800766, "learning_rate": 2.1146285950755225e-05, "loss": 6.4535, "step": 1022 }, { "epoch": 0.06350487305233099, "grad_norm": 0.7417406845304958, "learning_rate": 2.116697703289882e-05, "loss": 6.4187, "step": 1023 }, { "epoch": 0.0635669501520889, "grad_norm": 0.8400792346210189, "learning_rate": 2.1187668115042415e-05, "loss": 6.371, "step": 1024 }, { "epoch": 0.0636290272518468, "grad_norm": 0.9387317830437621, "learning_rate": 2.1208359197186014e-05, "loss": 6.4404, "step": 1025 }, { "epoch": 0.06369110435160469, "grad_norm": 0.8327212986920165, "learning_rate": 2.122905027932961e-05, "loss": 6.3209, "step": 1026 }, { "epoch": 0.06375318145136259, "grad_norm": 0.9285107706767137, "learning_rate": 2.1249741361473204e-05, "loss": 6.4513, "step": 1027 }, { "epoch": 0.0638152585511205, "grad_norm": 1.060328339880856, "learning_rate": 2.1270432443616802e-05, "loss": 6.3841, "step": 1028 }, { "epoch": 0.06387733565087839, "grad_norm": 0.8845833428506107, "learning_rate": 2.1291123525760397e-05, "loss": 6.3484, "step": 1029 }, { "epoch": 0.06393941275063629, "grad_norm": 0.8792675038825153, "learning_rate": 2.1311814607903992e-05, "loss": 6.3224, "step": 1030 }, { "epoch": 0.0640014898503942, "grad_norm": 0.7638714229629205, "learning_rate": 2.133250569004759e-05, "loss": 6.4659, "step": 1031 }, { "epoch": 0.06406356695015208, "grad_norm": 0.7755619119320227, "learning_rate": 2.1353196772191186e-05, "loss": 6.3538, "step": 1032 }, { "epoch": 0.06412564404990999, "grad_norm": 0.807934294028453, "learning_rate": 2.137388785433478e-05, "loss": 6.4109, "step": 1033 }, { "epoch": 0.06418772114966789, "grad_norm": 0.7870944189728475, "learning_rate": 2.139457893647838e-05, "loss": 6.3064, "step": 1034 }, { "epoch": 0.06424979824942578, "grad_norm": 0.8119171693112467, "learning_rate": 2.1415270018621975e-05, "loss": 6.3873, "step": 1035 }, { "epoch": 0.06431187534918369, "grad_norm": 0.8106482017331973, "learning_rate": 2.143596110076557e-05, "loss": 6.3792, "step": 1036 }, { "epoch": 0.06437395244894159, "grad_norm": 0.8195656055562364, "learning_rate": 2.1456652182909168e-05, "loss": 6.5018, "step": 1037 }, { "epoch": 0.06443602954869948, "grad_norm": 1.1329703385459402, "learning_rate": 2.1477343265052763e-05, "loss": 6.4483, "step": 1038 }, { "epoch": 0.06449810664845738, "grad_norm": 0.8962128670633471, "learning_rate": 2.1498034347196362e-05, "loss": 6.2724, "step": 1039 }, { "epoch": 0.06456018374821529, "grad_norm": 0.6692326729034462, "learning_rate": 2.1518725429339957e-05, "loss": 6.3882, "step": 1040 }, { "epoch": 0.06462226084797318, "grad_norm": 0.6599342830769945, "learning_rate": 2.1539416511483552e-05, "loss": 6.3301, "step": 1041 }, { "epoch": 0.06468433794773108, "grad_norm": 0.6754170577534275, "learning_rate": 2.156010759362715e-05, "loss": 6.4133, "step": 1042 }, { "epoch": 0.06474641504748899, "grad_norm": 0.6380637770752182, "learning_rate": 2.1580798675770745e-05, "loss": 6.3032, "step": 1043 }, { "epoch": 0.06480849214724688, "grad_norm": 0.6691772281681032, "learning_rate": 2.160148975791434e-05, "loss": 6.3591, "step": 1044 }, { "epoch": 0.06487056924700478, "grad_norm": 0.730587424363583, "learning_rate": 2.162218084005794e-05, "loss": 6.3646, "step": 1045 }, { "epoch": 0.06493264634676268, "grad_norm": 0.8472068755700243, "learning_rate": 2.1642871922201534e-05, "loss": 6.2624, "step": 1046 }, { "epoch": 0.06499472344652057, "grad_norm": 0.8260066090312471, "learning_rate": 2.166356300434513e-05, "loss": 6.3927, "step": 1047 }, { "epoch": 0.06505680054627848, "grad_norm": 0.8833629613681917, "learning_rate": 2.1684254086488724e-05, "loss": 6.3787, "step": 1048 }, { "epoch": 0.06511887764603638, "grad_norm": 0.900416442103772, "learning_rate": 2.170494516863232e-05, "loss": 6.3026, "step": 1049 }, { "epoch": 0.06518095474579427, "grad_norm": 0.7877976127649354, "learning_rate": 2.1725636250775914e-05, "loss": 6.401, "step": 1050 }, { "epoch": 0.06524303184555218, "grad_norm": 0.7204920762107497, "learning_rate": 2.1746327332919513e-05, "loss": 6.4884, "step": 1051 }, { "epoch": 0.06530510894531008, "grad_norm": 0.9376003847419223, "learning_rate": 2.1767018415063108e-05, "loss": 6.3362, "step": 1052 }, { "epoch": 0.06536718604506797, "grad_norm": 1.262857071265297, "learning_rate": 2.1787709497206703e-05, "loss": 6.3747, "step": 1053 }, { "epoch": 0.06542926314482587, "grad_norm": 0.9242453086822704, "learning_rate": 2.18084005793503e-05, "loss": 6.3588, "step": 1054 }, { "epoch": 0.06549134024458378, "grad_norm": 1.023534056789673, "learning_rate": 2.1829091661493897e-05, "loss": 6.334, "step": 1055 }, { "epoch": 0.06555341734434167, "grad_norm": 0.9986804063729967, "learning_rate": 2.184978274363749e-05, "loss": 6.3871, "step": 1056 }, { "epoch": 0.06561549444409957, "grad_norm": 0.8293204406178734, "learning_rate": 2.187047382578109e-05, "loss": 6.1558, "step": 1057 }, { "epoch": 0.06567757154385748, "grad_norm": 0.6990397559854165, "learning_rate": 2.1891164907924685e-05, "loss": 6.3003, "step": 1058 }, { "epoch": 0.06573964864361537, "grad_norm": 0.6566021973248681, "learning_rate": 2.191185599006828e-05, "loss": 6.3818, "step": 1059 }, { "epoch": 0.06580172574337327, "grad_norm": 0.7708681373643305, "learning_rate": 2.193254707221188e-05, "loss": 6.306, "step": 1060 }, { "epoch": 0.06586380284313118, "grad_norm": 1.1712103930018714, "learning_rate": 2.1953238154355474e-05, "loss": 6.2416, "step": 1061 }, { "epoch": 0.06592587994288907, "grad_norm": 0.838350112233951, "learning_rate": 2.197392923649907e-05, "loss": 6.2934, "step": 1062 }, { "epoch": 0.06598795704264697, "grad_norm": 0.9639772856655935, "learning_rate": 2.1994620318642667e-05, "loss": 6.2698, "step": 1063 }, { "epoch": 0.06605003414240487, "grad_norm": 1.15652586249408, "learning_rate": 2.2015311400786262e-05, "loss": 6.2268, "step": 1064 }, { "epoch": 0.06611211124216276, "grad_norm": 0.8645169833971232, "learning_rate": 2.2036002482929857e-05, "loss": 6.2505, "step": 1065 }, { "epoch": 0.06617418834192067, "grad_norm": 0.8922412996388988, "learning_rate": 2.2056693565073456e-05, "loss": 6.3077, "step": 1066 }, { "epoch": 0.06623626544167857, "grad_norm": 0.9479902504387429, "learning_rate": 2.207738464721705e-05, "loss": 6.2132, "step": 1067 }, { "epoch": 0.06629834254143646, "grad_norm": 0.9886039538736396, "learning_rate": 2.2098075729360646e-05, "loss": 6.2733, "step": 1068 }, { "epoch": 0.06636041964119437, "grad_norm": 1.2084228000185764, "learning_rate": 2.2118766811504245e-05, "loss": 6.2878, "step": 1069 }, { "epoch": 0.06642249674095227, "grad_norm": 1.0161172114254544, "learning_rate": 2.213945789364784e-05, "loss": 6.2033, "step": 1070 }, { "epoch": 0.06648457384071016, "grad_norm": 1.0243863082922566, "learning_rate": 2.2160148975791435e-05, "loss": 6.3668, "step": 1071 }, { "epoch": 0.06654665094046806, "grad_norm": 0.7238133420222064, "learning_rate": 2.2180840057935033e-05, "loss": 6.3825, "step": 1072 }, { "epoch": 0.06660872804022597, "grad_norm": 0.9828152181954691, "learning_rate": 2.2201531140078628e-05, "loss": 6.2871, "step": 1073 }, { "epoch": 0.06667080513998386, "grad_norm": 0.8791085582591249, "learning_rate": 2.2222222222222223e-05, "loss": 6.4017, "step": 1074 }, { "epoch": 0.06673288223974176, "grad_norm": 0.8502608684326343, "learning_rate": 2.224291330436582e-05, "loss": 6.3267, "step": 1075 }, { "epoch": 0.06679495933949967, "grad_norm": 0.9314791621438502, "learning_rate": 2.2263604386509417e-05, "loss": 6.2507, "step": 1076 }, { "epoch": 0.06685703643925756, "grad_norm": 1.1191426488515186, "learning_rate": 2.2284295468653012e-05, "loss": 6.2781, "step": 1077 }, { "epoch": 0.06691911353901546, "grad_norm": 0.96826969053033, "learning_rate": 2.2304986550796607e-05, "loss": 6.3488, "step": 1078 }, { "epoch": 0.06698119063877335, "grad_norm": 1.0648908642472539, "learning_rate": 2.2325677632940202e-05, "loss": 6.2487, "step": 1079 }, { "epoch": 0.06704326773853125, "grad_norm": 0.9165006219869065, "learning_rate": 2.2346368715083797e-05, "loss": 6.2188, "step": 1080 }, { "epoch": 0.06710534483828916, "grad_norm": 0.891436926191036, "learning_rate": 2.2367059797227396e-05, "loss": 6.2551, "step": 1081 }, { "epoch": 0.06716742193804705, "grad_norm": 0.7511589711747766, "learning_rate": 2.238775087937099e-05, "loss": 6.3191, "step": 1082 }, { "epoch": 0.06722949903780495, "grad_norm": 0.9452374641335819, "learning_rate": 2.240844196151459e-05, "loss": 6.3479, "step": 1083 }, { "epoch": 0.06729157613756286, "grad_norm": 0.8845088493503966, "learning_rate": 2.2429133043658184e-05, "loss": 6.253, "step": 1084 }, { "epoch": 0.06735365323732075, "grad_norm": 1.0423536451324935, "learning_rate": 2.244982412580178e-05, "loss": 6.2151, "step": 1085 }, { "epoch": 0.06741573033707865, "grad_norm": 1.1441633655271721, "learning_rate": 2.2470515207945378e-05, "loss": 6.3495, "step": 1086 }, { "epoch": 0.06747780743683655, "grad_norm": 0.8411178151159843, "learning_rate": 2.2491206290088973e-05, "loss": 6.2252, "step": 1087 }, { "epoch": 0.06753988453659444, "grad_norm": 0.8885394322559903, "learning_rate": 2.2511897372232568e-05, "loss": 6.3172, "step": 1088 }, { "epoch": 0.06760196163635235, "grad_norm": 1.0302492545665995, "learning_rate": 2.2532588454376166e-05, "loss": 6.2808, "step": 1089 }, { "epoch": 0.06766403873611025, "grad_norm": 1.0261782573012115, "learning_rate": 2.255327953651976e-05, "loss": 6.2242, "step": 1090 }, { "epoch": 0.06772611583586814, "grad_norm": 1.2450628051849355, "learning_rate": 2.2573970618663357e-05, "loss": 6.32, "step": 1091 }, { "epoch": 0.06778819293562605, "grad_norm": 0.7629187887826514, "learning_rate": 2.2594661700806955e-05, "loss": 6.3119, "step": 1092 }, { "epoch": 0.06785027003538395, "grad_norm": 1.0688648318343623, "learning_rate": 2.261535278295055e-05, "loss": 6.2128, "step": 1093 }, { "epoch": 0.06791234713514184, "grad_norm": 1.1540312743849734, "learning_rate": 2.2636043865094145e-05, "loss": 6.3232, "step": 1094 }, { "epoch": 0.06797442423489974, "grad_norm": 1.059750308732161, "learning_rate": 2.2656734947237744e-05, "loss": 6.2929, "step": 1095 }, { "epoch": 0.06803650133465765, "grad_norm": 0.9875188608420594, "learning_rate": 2.267742602938134e-05, "loss": 6.2438, "step": 1096 }, { "epoch": 0.06809857843441554, "grad_norm": 1.031117459651935, "learning_rate": 2.2698117111524934e-05, "loss": 6.2756, "step": 1097 }, { "epoch": 0.06816065553417344, "grad_norm": 1.1068896001782322, "learning_rate": 2.2718808193668532e-05, "loss": 6.1609, "step": 1098 }, { "epoch": 0.06822273263393135, "grad_norm": 0.6546895457306358, "learning_rate": 2.2739499275812127e-05, "loss": 6.2707, "step": 1099 }, { "epoch": 0.06828480973368924, "grad_norm": 0.8232688311228233, "learning_rate": 2.2760190357955722e-05, "loss": 6.2739, "step": 1100 }, { "epoch": 0.06834688683344714, "grad_norm": 0.8012730646607612, "learning_rate": 2.278088144009932e-05, "loss": 6.2337, "step": 1101 }, { "epoch": 0.06840896393320504, "grad_norm": 0.8012258272497109, "learning_rate": 2.2801572522242916e-05, "loss": 6.2462, "step": 1102 }, { "epoch": 0.06847104103296293, "grad_norm": 0.8353728767264149, "learning_rate": 2.282226360438651e-05, "loss": 6.2483, "step": 1103 }, { "epoch": 0.06853311813272084, "grad_norm": 0.6895899866441992, "learning_rate": 2.2842954686530106e-05, "loss": 6.3066, "step": 1104 }, { "epoch": 0.06859519523247874, "grad_norm": 0.7145742106249248, "learning_rate": 2.28636457686737e-05, "loss": 6.2293, "step": 1105 }, { "epoch": 0.06865727233223663, "grad_norm": 0.7784548539073016, "learning_rate": 2.2884336850817296e-05, "loss": 6.3697, "step": 1106 }, { "epoch": 0.06871934943199454, "grad_norm": 0.777698033555262, "learning_rate": 2.2905027932960895e-05, "loss": 6.3466, "step": 1107 }, { "epoch": 0.06878142653175244, "grad_norm": 0.7531162926261545, "learning_rate": 2.292571901510449e-05, "loss": 6.164, "step": 1108 }, { "epoch": 0.06884350363151033, "grad_norm": 0.790445343533659, "learning_rate": 2.2946410097248085e-05, "loss": 6.2436, "step": 1109 }, { "epoch": 0.06890558073126823, "grad_norm": 0.8690492046152619, "learning_rate": 2.2967101179391683e-05, "loss": 6.3154, "step": 1110 }, { "epoch": 0.06896765783102614, "grad_norm": 0.985546414151804, "learning_rate": 2.298779226153528e-05, "loss": 6.1719, "step": 1111 }, { "epoch": 0.06902973493078403, "grad_norm": 1.2931529487255586, "learning_rate": 2.3008483343678873e-05, "loss": 6.2684, "step": 1112 }, { "epoch": 0.06909181203054193, "grad_norm": 0.7335249668493036, "learning_rate": 2.3029174425822472e-05, "loss": 6.3679, "step": 1113 }, { "epoch": 0.06915388913029984, "grad_norm": 0.9471781896764778, "learning_rate": 2.3049865507966067e-05, "loss": 6.176, "step": 1114 }, { "epoch": 0.06921596623005773, "grad_norm": 1.287374744063012, "learning_rate": 2.3070556590109662e-05, "loss": 6.1785, "step": 1115 }, { "epoch": 0.06927804332981563, "grad_norm": 0.9445749037682972, "learning_rate": 2.309124767225326e-05, "loss": 6.24, "step": 1116 }, { "epoch": 0.06934012042957353, "grad_norm": 1.054216746069266, "learning_rate": 2.3111938754396856e-05, "loss": 6.2666, "step": 1117 }, { "epoch": 0.06940219752933142, "grad_norm": 1.1030763545130515, "learning_rate": 2.313262983654045e-05, "loss": 6.1584, "step": 1118 }, { "epoch": 0.06946427462908933, "grad_norm": 0.7835967485060518, "learning_rate": 2.315332091868405e-05, "loss": 6.2226, "step": 1119 }, { "epoch": 0.06952635172884723, "grad_norm": 0.9142142601861335, "learning_rate": 2.3174012000827644e-05, "loss": 6.3285, "step": 1120 }, { "epoch": 0.06958842882860512, "grad_norm": 1.0512505940221317, "learning_rate": 2.319470308297124e-05, "loss": 6.1718, "step": 1121 }, { "epoch": 0.06965050592836303, "grad_norm": 0.7975403495076048, "learning_rate": 2.3215394165114838e-05, "loss": 6.2676, "step": 1122 }, { "epoch": 0.06971258302812093, "grad_norm": 0.8067756151944603, "learning_rate": 2.3236085247258433e-05, "loss": 6.2042, "step": 1123 }, { "epoch": 0.06977466012787882, "grad_norm": 0.8369453985163287, "learning_rate": 2.325677632940203e-05, "loss": 6.2039, "step": 1124 }, { "epoch": 0.06983673722763672, "grad_norm": 0.872727406058789, "learning_rate": 2.3277467411545626e-05, "loss": 6.1098, "step": 1125 }, { "epoch": 0.06989881432739463, "grad_norm": 0.8545721921251884, "learning_rate": 2.329815849368922e-05, "loss": 6.1852, "step": 1126 }, { "epoch": 0.06996089142715252, "grad_norm": 0.7849885752022698, "learning_rate": 2.331884957583282e-05, "loss": 6.2544, "step": 1127 }, { "epoch": 0.07002296852691042, "grad_norm": 0.8351972635902141, "learning_rate": 2.3339540657976415e-05, "loss": 6.1323, "step": 1128 }, { "epoch": 0.07008504562666833, "grad_norm": 0.9232142467398634, "learning_rate": 2.336023174012001e-05, "loss": 6.1862, "step": 1129 }, { "epoch": 0.07014712272642622, "grad_norm": 0.7637888630970291, "learning_rate": 2.3380922822263605e-05, "loss": 6.117, "step": 1130 }, { "epoch": 0.07020919982618412, "grad_norm": 0.8898479351049109, "learning_rate": 2.34016139044072e-05, "loss": 6.2423, "step": 1131 }, { "epoch": 0.07027127692594203, "grad_norm": 1.1796317182072487, "learning_rate": 2.34223049865508e-05, "loss": 6.1474, "step": 1132 }, { "epoch": 0.07033335402569992, "grad_norm": 0.8430436321157642, "learning_rate": 2.3442996068694394e-05, "loss": 6.0679, "step": 1133 }, { "epoch": 0.07039543112545782, "grad_norm": 0.8175620334898912, "learning_rate": 2.346368715083799e-05, "loss": 6.1191, "step": 1134 }, { "epoch": 0.07045750822521572, "grad_norm": 1.1074829084743774, "learning_rate": 2.3484378232981584e-05, "loss": 6.1323, "step": 1135 }, { "epoch": 0.07051958532497361, "grad_norm": 0.8459271845560304, "learning_rate": 2.3505069315125182e-05, "loss": 6.0923, "step": 1136 }, { "epoch": 0.07058166242473152, "grad_norm": 0.9863339565595708, "learning_rate": 2.3525760397268777e-05, "loss": 6.1396, "step": 1137 }, { "epoch": 0.07064373952448942, "grad_norm": 1.014187837681168, "learning_rate": 2.3546451479412373e-05, "loss": 6.085, "step": 1138 }, { "epoch": 0.07070581662424731, "grad_norm": 0.925742812505558, "learning_rate": 2.356714256155597e-05, "loss": 6.2216, "step": 1139 }, { "epoch": 0.07076789372400522, "grad_norm": 0.9624996049062735, "learning_rate": 2.3587833643699566e-05, "loss": 6.2402, "step": 1140 }, { "epoch": 0.07082997082376312, "grad_norm": 0.8480377042143616, "learning_rate": 2.360852472584316e-05, "loss": 6.0741, "step": 1141 }, { "epoch": 0.07089204792352101, "grad_norm": 0.9082942137620729, "learning_rate": 2.362921580798676e-05, "loss": 6.1654, "step": 1142 }, { "epoch": 0.07095412502327891, "grad_norm": 1.2250125592424834, "learning_rate": 2.3649906890130355e-05, "loss": 6.1257, "step": 1143 }, { "epoch": 0.07101620212303682, "grad_norm": 0.8012434207211704, "learning_rate": 2.367059797227395e-05, "loss": 6.1435, "step": 1144 }, { "epoch": 0.07107827922279471, "grad_norm": 1.0014705574099565, "learning_rate": 2.3691289054417548e-05, "loss": 6.1897, "step": 1145 }, { "epoch": 0.07114035632255261, "grad_norm": 1.000538561755098, "learning_rate": 2.3711980136561143e-05, "loss": 6.1743, "step": 1146 }, { "epoch": 0.07120243342231052, "grad_norm": 1.0258694938362234, "learning_rate": 2.373267121870474e-05, "loss": 6.1863, "step": 1147 }, { "epoch": 0.0712645105220684, "grad_norm": 1.173611028973478, "learning_rate": 2.3753362300848337e-05, "loss": 6.1251, "step": 1148 }, { "epoch": 0.07132658762182631, "grad_norm": 0.9585963627687147, "learning_rate": 2.3774053382991932e-05, "loss": 6.1927, "step": 1149 }, { "epoch": 0.07138866472158421, "grad_norm": 0.8178888393014218, "learning_rate": 2.3794744465135527e-05, "loss": 6.1862, "step": 1150 }, { "epoch": 0.0714507418213421, "grad_norm": 0.9883837740947586, "learning_rate": 2.3815435547279125e-05, "loss": 6.1386, "step": 1151 }, { "epoch": 0.07151281892110001, "grad_norm": 0.800185935943393, "learning_rate": 2.383612662942272e-05, "loss": 6.2809, "step": 1152 }, { "epoch": 0.07157489602085791, "grad_norm": 0.8601099859367828, "learning_rate": 2.3856817711566316e-05, "loss": 6.1582, "step": 1153 }, { "epoch": 0.0716369731206158, "grad_norm": 0.7904022853014221, "learning_rate": 2.3877508793709914e-05, "loss": 6.1978, "step": 1154 }, { "epoch": 0.0716990502203737, "grad_norm": 0.7334576416500678, "learning_rate": 2.389819987585351e-05, "loss": 6.1057, "step": 1155 }, { "epoch": 0.07176112732013161, "grad_norm": 0.7434755492124565, "learning_rate": 2.3918890957997104e-05, "loss": 6.2433, "step": 1156 }, { "epoch": 0.0718232044198895, "grad_norm": 0.9075812723217357, "learning_rate": 2.3939582040140703e-05, "loss": 6.1429, "step": 1157 }, { "epoch": 0.0718852815196474, "grad_norm": 0.966011051671563, "learning_rate": 2.3960273122284298e-05, "loss": 6.0128, "step": 1158 }, { "epoch": 0.07194735861940531, "grad_norm": 1.1441765974595055, "learning_rate": 2.3980964204427893e-05, "loss": 6.0545, "step": 1159 }, { "epoch": 0.0720094357191632, "grad_norm": 1.1033041422248668, "learning_rate": 2.4001655286571488e-05, "loss": 6.0032, "step": 1160 }, { "epoch": 0.0720715128189211, "grad_norm": 0.8928983685648921, "learning_rate": 2.4022346368715083e-05, "loss": 6.1012, "step": 1161 }, { "epoch": 0.072133589918679, "grad_norm": 0.928170866426965, "learning_rate": 2.4043037450858678e-05, "loss": 6.2295, "step": 1162 }, { "epoch": 0.0721956670184369, "grad_norm": 1.1483122050127632, "learning_rate": 2.4063728533002277e-05, "loss": 6.1752, "step": 1163 }, { "epoch": 0.0722577441181948, "grad_norm": 0.8796097716010325, "learning_rate": 2.408441961514587e-05, "loss": 6.1988, "step": 1164 }, { "epoch": 0.0723198212179527, "grad_norm": 0.8122702787338809, "learning_rate": 2.4105110697289467e-05, "loss": 6.1252, "step": 1165 }, { "epoch": 0.0723818983177106, "grad_norm": 0.9223247013131572, "learning_rate": 2.4125801779433065e-05, "loss": 6.0978, "step": 1166 }, { "epoch": 0.0724439754174685, "grad_norm": 0.9079938421817809, "learning_rate": 2.414649286157666e-05, "loss": 6.1174, "step": 1167 }, { "epoch": 0.0725060525172264, "grad_norm": 0.8779178402178766, "learning_rate": 2.416718394372026e-05, "loss": 6.1716, "step": 1168 }, { "epoch": 0.07256812961698429, "grad_norm": 0.9431526162848621, "learning_rate": 2.4187875025863854e-05, "loss": 6.1351, "step": 1169 }, { "epoch": 0.0726302067167422, "grad_norm": 0.9859530198316061, "learning_rate": 2.420856610800745e-05, "loss": 6.1903, "step": 1170 }, { "epoch": 0.07269228381650009, "grad_norm": 0.778394592646276, "learning_rate": 2.4229257190151047e-05, "loss": 6.1152, "step": 1171 }, { "epoch": 0.07275436091625799, "grad_norm": 0.8167265487307311, "learning_rate": 2.4249948272294642e-05, "loss": 6.084, "step": 1172 }, { "epoch": 0.0728164380160159, "grad_norm": 0.9373344911072564, "learning_rate": 2.4270639354438237e-05, "loss": 6.0913, "step": 1173 }, { "epoch": 0.07287851511577378, "grad_norm": 0.8951550103856655, "learning_rate": 2.4291330436581836e-05, "loss": 6.132, "step": 1174 }, { "epoch": 0.07294059221553169, "grad_norm": 0.8207789230386314, "learning_rate": 2.431202151872543e-05, "loss": 6.1178, "step": 1175 }, { "epoch": 0.07300266931528959, "grad_norm": 1.0560425707046293, "learning_rate": 2.4332712600869026e-05, "loss": 6.1727, "step": 1176 }, { "epoch": 0.07306474641504748, "grad_norm": 0.9755501038898168, "learning_rate": 2.4353403683012625e-05, "loss": 6.1755, "step": 1177 }, { "epoch": 0.07312682351480539, "grad_norm": 1.116800414563788, "learning_rate": 2.437409476515622e-05, "loss": 6.1341, "step": 1178 }, { "epoch": 0.07318890061456329, "grad_norm": 1.0186081138050995, "learning_rate": 2.4394785847299815e-05, "loss": 6.1202, "step": 1179 }, { "epoch": 0.07325097771432118, "grad_norm": 0.9867532248375451, "learning_rate": 2.4415476929443413e-05, "loss": 6.1441, "step": 1180 }, { "epoch": 0.07331305481407908, "grad_norm": 1.134116533747097, "learning_rate": 2.4436168011587008e-05, "loss": 6.0273, "step": 1181 }, { "epoch": 0.07337513191383699, "grad_norm": 0.955140843116613, "learning_rate": 2.4456859093730603e-05, "loss": 6.1112, "step": 1182 }, { "epoch": 0.07343720901359488, "grad_norm": 1.0101389683589772, "learning_rate": 2.4477550175874202e-05, "loss": 6.1975, "step": 1183 }, { "epoch": 0.07349928611335278, "grad_norm": 0.9776569489727598, "learning_rate": 2.4498241258017797e-05, "loss": 6.1308, "step": 1184 }, { "epoch": 0.07356136321311069, "grad_norm": 1.0108758073286783, "learning_rate": 2.4518932340161392e-05, "loss": 6.0989, "step": 1185 }, { "epoch": 0.07362344031286858, "grad_norm": 0.9587993559884116, "learning_rate": 2.4539623422304987e-05, "loss": 6.0841, "step": 1186 }, { "epoch": 0.07368551741262648, "grad_norm": 1.0662981474583322, "learning_rate": 2.4560314504448585e-05, "loss": 5.9865, "step": 1187 }, { "epoch": 0.07374759451238438, "grad_norm": 0.8729238060694933, "learning_rate": 2.458100558659218e-05, "loss": 6.0509, "step": 1188 }, { "epoch": 0.07380967161214227, "grad_norm": 0.8369871305709145, "learning_rate": 2.4601696668735776e-05, "loss": 6.0676, "step": 1189 }, { "epoch": 0.07387174871190018, "grad_norm": 0.95173794236197, "learning_rate": 2.462238775087937e-05, "loss": 6.1143, "step": 1190 }, { "epoch": 0.07393382581165808, "grad_norm": 1.2778172446573546, "learning_rate": 2.4643078833022966e-05, "loss": 6.1161, "step": 1191 }, { "epoch": 0.07399590291141597, "grad_norm": 0.800075238385323, "learning_rate": 2.4663769915166564e-05, "loss": 6.0524, "step": 1192 }, { "epoch": 0.07405798001117388, "grad_norm": 0.8515597702122603, "learning_rate": 2.468446099731016e-05, "loss": 5.974, "step": 1193 }, { "epoch": 0.07412005711093178, "grad_norm": 1.0838892439604786, "learning_rate": 2.4705152079453754e-05, "loss": 6.0535, "step": 1194 }, { "epoch": 0.07418213421068967, "grad_norm": 0.8293484341607594, "learning_rate": 2.4725843161597353e-05, "loss": 6.0809, "step": 1195 }, { "epoch": 0.07424421131044757, "grad_norm": 0.8798529379596446, "learning_rate": 2.4746534243740948e-05, "loss": 6.2134, "step": 1196 }, { "epoch": 0.07430628841020548, "grad_norm": 0.9885747420978376, "learning_rate": 2.4767225325884543e-05, "loss": 6.11, "step": 1197 }, { "epoch": 0.07436836550996337, "grad_norm": 1.1621438414006542, "learning_rate": 2.478791640802814e-05, "loss": 6.1144, "step": 1198 }, { "epoch": 0.07443044260972127, "grad_norm": 0.8383002358141494, "learning_rate": 2.4808607490171737e-05, "loss": 6.2288, "step": 1199 }, { "epoch": 0.07449251970947918, "grad_norm": 1.1242136856399787, "learning_rate": 2.482929857231533e-05, "loss": 6.1042, "step": 1200 }, { "epoch": 0.07455459680923707, "grad_norm": 0.8917077159048519, "learning_rate": 2.484998965445893e-05, "loss": 6.0046, "step": 1201 }, { "epoch": 0.07461667390899497, "grad_norm": 1.0202512573742941, "learning_rate": 2.4870680736602525e-05, "loss": 6.0565, "step": 1202 }, { "epoch": 0.07467875100875287, "grad_norm": 1.0581593766010664, "learning_rate": 2.489137181874612e-05, "loss": 6.0215, "step": 1203 }, { "epoch": 0.07474082810851077, "grad_norm": 0.883054548181444, "learning_rate": 2.491206290088972e-05, "loss": 6.0284, "step": 1204 }, { "epoch": 0.07480290520826867, "grad_norm": 0.8341344479472103, "learning_rate": 2.4932753983033314e-05, "loss": 6.0864, "step": 1205 }, { "epoch": 0.07486498230802657, "grad_norm": 0.9998084421268049, "learning_rate": 2.495344506517691e-05, "loss": 6.0764, "step": 1206 }, { "epoch": 0.07492705940778446, "grad_norm": 0.88715506418775, "learning_rate": 2.4974136147320507e-05, "loss": 6.0895, "step": 1207 }, { "epoch": 0.07498913650754237, "grad_norm": 0.7951378681189265, "learning_rate": 2.4994827229464102e-05, "loss": 6.1077, "step": 1208 }, { "epoch": 0.07505121360730027, "grad_norm": 0.9034905350595205, "learning_rate": 2.50155183116077e-05, "loss": 6.1048, "step": 1209 }, { "epoch": 0.07511329070705816, "grad_norm": 0.8336838978266966, "learning_rate": 2.5036209393751293e-05, "loss": 6.0287, "step": 1210 }, { "epoch": 0.07517536780681607, "grad_norm": 0.8045660871741216, "learning_rate": 2.505690047589489e-05, "loss": 6.0167, "step": 1211 }, { "epoch": 0.07523744490657397, "grad_norm": 0.8567473728171932, "learning_rate": 2.507759155803849e-05, "loss": 6.076, "step": 1212 }, { "epoch": 0.07529952200633186, "grad_norm": 0.8904744405810621, "learning_rate": 2.509828264018208e-05, "loss": 6.0332, "step": 1213 }, { "epoch": 0.07536159910608976, "grad_norm": 1.075207231197422, "learning_rate": 2.511897372232568e-05, "loss": 6.086, "step": 1214 }, { "epoch": 0.07542367620584767, "grad_norm": 1.067140710567359, "learning_rate": 2.5139664804469275e-05, "loss": 6.1172, "step": 1215 }, { "epoch": 0.07548575330560556, "grad_norm": 1.1413037618987039, "learning_rate": 2.516035588661287e-05, "loss": 6.0866, "step": 1216 }, { "epoch": 0.07554783040536346, "grad_norm": 0.7632429084011059, "learning_rate": 2.5181046968756465e-05, "loss": 6.0929, "step": 1217 }, { "epoch": 0.07560990750512137, "grad_norm": 0.7479715334248994, "learning_rate": 2.5201738050900063e-05, "loss": 5.9515, "step": 1218 }, { "epoch": 0.07567198460487926, "grad_norm": 0.8677585886445327, "learning_rate": 2.522242913304366e-05, "loss": 5.9798, "step": 1219 }, { "epoch": 0.07573406170463716, "grad_norm": 0.9202521312453699, "learning_rate": 2.5243120215187254e-05, "loss": 6.0011, "step": 1220 }, { "epoch": 0.07579613880439506, "grad_norm": 0.8773711977271802, "learning_rate": 2.5263811297330852e-05, "loss": 6.0014, "step": 1221 }, { "epoch": 0.07585821590415295, "grad_norm": 0.9447016042551775, "learning_rate": 2.5284502379474444e-05, "loss": 6.096, "step": 1222 }, { "epoch": 0.07592029300391086, "grad_norm": 1.069030773915327, "learning_rate": 2.5305193461618042e-05, "loss": 6.0495, "step": 1223 }, { "epoch": 0.07598237010366876, "grad_norm": 0.8466520205930558, "learning_rate": 2.532588454376164e-05, "loss": 6.0556, "step": 1224 }, { "epoch": 0.07604444720342665, "grad_norm": 0.8235718054788875, "learning_rate": 2.5346575625905232e-05, "loss": 6.114, "step": 1225 }, { "epoch": 0.07610652430318456, "grad_norm": 0.7425397739610681, "learning_rate": 2.536726670804883e-05, "loss": 6.0555, "step": 1226 }, { "epoch": 0.07616860140294246, "grad_norm": 0.7529566818128282, "learning_rate": 2.538795779019243e-05, "loss": 6.0916, "step": 1227 }, { "epoch": 0.07623067850270035, "grad_norm": 0.9440356371788148, "learning_rate": 2.540864887233602e-05, "loss": 5.967, "step": 1228 }, { "epoch": 0.07629275560245825, "grad_norm": 0.9529003441372731, "learning_rate": 2.542933995447962e-05, "loss": 6.0295, "step": 1229 }, { "epoch": 0.07635483270221616, "grad_norm": 0.8316336743022675, "learning_rate": 2.5450031036623218e-05, "loss": 6.0583, "step": 1230 }, { "epoch": 0.07641690980197405, "grad_norm": 0.8415715741427224, "learning_rate": 2.5470722118766816e-05, "loss": 5.98, "step": 1231 }, { "epoch": 0.07647898690173195, "grad_norm": 0.8341382708815057, "learning_rate": 2.5491413200910408e-05, "loss": 5.9633, "step": 1232 }, { "epoch": 0.07654106400148986, "grad_norm": 0.8694004999036912, "learning_rate": 2.5512104283054006e-05, "loss": 5.8972, "step": 1233 }, { "epoch": 0.07660314110124775, "grad_norm": 0.8978893100320973, "learning_rate": 2.5532795365197605e-05, "loss": 6.0841, "step": 1234 }, { "epoch": 0.07666521820100565, "grad_norm": 1.0916293703264968, "learning_rate": 2.5553486447341197e-05, "loss": 5.9875, "step": 1235 }, { "epoch": 0.07672729530076355, "grad_norm": 0.9484069869586512, "learning_rate": 2.5574177529484795e-05, "loss": 6.0358, "step": 1236 }, { "epoch": 0.07678937240052144, "grad_norm": 0.8341058647056601, "learning_rate": 2.5594868611628394e-05, "loss": 5.9799, "step": 1237 }, { "epoch": 0.07685144950027935, "grad_norm": 0.8062928491487503, "learning_rate": 2.5615559693771985e-05, "loss": 6.0264, "step": 1238 }, { "epoch": 0.07691352660003725, "grad_norm": 1.1916433629956293, "learning_rate": 2.5636250775915584e-05, "loss": 6.0218, "step": 1239 }, { "epoch": 0.07697560369979514, "grad_norm": 0.9612441155262221, "learning_rate": 2.565694185805918e-05, "loss": 5.9301, "step": 1240 }, { "epoch": 0.07703768079955305, "grad_norm": 0.8565326850542562, "learning_rate": 2.5677632940202774e-05, "loss": 5.9262, "step": 1241 }, { "epoch": 0.07709975789931095, "grad_norm": 0.8043205016909989, "learning_rate": 2.569832402234637e-05, "loss": 5.9862, "step": 1242 }, { "epoch": 0.07716183499906884, "grad_norm": 0.8383867622234035, "learning_rate": 2.5719015104489967e-05, "loss": 6.026, "step": 1243 }, { "epoch": 0.07722391209882674, "grad_norm": 0.8899158867189207, "learning_rate": 2.5739706186633562e-05, "loss": 5.9832, "step": 1244 }, { "epoch": 0.07728598919858465, "grad_norm": 0.8425644209821251, "learning_rate": 2.5760397268777158e-05, "loss": 6.0053, "step": 1245 }, { "epoch": 0.07734806629834254, "grad_norm": 0.9929204621417881, "learning_rate": 2.5781088350920756e-05, "loss": 5.9829, "step": 1246 }, { "epoch": 0.07741014339810044, "grad_norm": 0.8258281380470397, "learning_rate": 2.5801779433064348e-05, "loss": 6.0779, "step": 1247 }, { "epoch": 0.07747222049785835, "grad_norm": 0.861654033980254, "learning_rate": 2.5822470515207946e-05, "loss": 6.0254, "step": 1248 }, { "epoch": 0.07753429759761624, "grad_norm": 0.9186812537287382, "learning_rate": 2.5843161597351545e-05, "loss": 5.9315, "step": 1249 }, { "epoch": 0.07759637469737414, "grad_norm": 0.9909392487201641, "learning_rate": 2.5863852679495136e-05, "loss": 6.0542, "step": 1250 }, { "epoch": 0.07765845179713204, "grad_norm": 1.205534318486878, "learning_rate": 2.5884543761638735e-05, "loss": 6.0872, "step": 1251 }, { "epoch": 0.07772052889688993, "grad_norm": 0.7508119320827208, "learning_rate": 2.5905234843782333e-05, "loss": 6.0317, "step": 1252 }, { "epoch": 0.07778260599664784, "grad_norm": 0.9959225675805838, "learning_rate": 2.5925925925925925e-05, "loss": 6.0103, "step": 1253 }, { "epoch": 0.07784468309640574, "grad_norm": 1.099176189409314, "learning_rate": 2.5946617008069523e-05, "loss": 5.9865, "step": 1254 }, { "epoch": 0.07790676019616363, "grad_norm": 0.9530250309281771, "learning_rate": 2.5967308090213122e-05, "loss": 6.0201, "step": 1255 }, { "epoch": 0.07796883729592154, "grad_norm": 0.9987632136610687, "learning_rate": 2.5987999172356714e-05, "loss": 6.0251, "step": 1256 }, { "epoch": 0.07803091439567944, "grad_norm": 0.8363924660091145, "learning_rate": 2.6008690254500312e-05, "loss": 5.9289, "step": 1257 }, { "epoch": 0.07809299149543733, "grad_norm": 0.8733871445445464, "learning_rate": 2.602938133664391e-05, "loss": 6.0561, "step": 1258 }, { "epoch": 0.07815506859519523, "grad_norm": 0.9841713164787906, "learning_rate": 2.6050072418787502e-05, "loss": 6.0377, "step": 1259 }, { "epoch": 0.07821714569495314, "grad_norm": 0.9621740619133597, "learning_rate": 2.60707635009311e-05, "loss": 6.027, "step": 1260 }, { "epoch": 0.07827922279471103, "grad_norm": 0.8996224922949825, "learning_rate": 2.60914545830747e-05, "loss": 5.8981, "step": 1261 }, { "epoch": 0.07834129989446893, "grad_norm": 0.8475384608406069, "learning_rate": 2.611214566521829e-05, "loss": 5.9354, "step": 1262 }, { "epoch": 0.07840337699422684, "grad_norm": 0.8815169829551801, "learning_rate": 2.613283674736189e-05, "loss": 5.9873, "step": 1263 }, { "epoch": 0.07846545409398473, "grad_norm": 1.0724015280059946, "learning_rate": 2.6153527829505488e-05, "loss": 5.9259, "step": 1264 }, { "epoch": 0.07852753119374263, "grad_norm": 1.0156467141984626, "learning_rate": 2.617421891164908e-05, "loss": 5.9357, "step": 1265 }, { "epoch": 0.07858960829350052, "grad_norm": 0.9704144545601288, "learning_rate": 2.6194909993792678e-05, "loss": 5.8609, "step": 1266 }, { "epoch": 0.07865168539325842, "grad_norm": 0.9684068625848666, "learning_rate": 2.6215601075936276e-05, "loss": 5.9803, "step": 1267 }, { "epoch": 0.07871376249301633, "grad_norm": 1.0491143374210854, "learning_rate": 2.6236292158079868e-05, "loss": 6.0838, "step": 1268 }, { "epoch": 0.07877583959277422, "grad_norm": 0.8431603172489461, "learning_rate": 2.6256983240223466e-05, "loss": 6.1229, "step": 1269 }, { "epoch": 0.07883791669253212, "grad_norm": 0.8435604447673772, "learning_rate": 2.627767432236706e-05, "loss": 6.0253, "step": 1270 }, { "epoch": 0.07889999379229003, "grad_norm": 1.0154047213605597, "learning_rate": 2.6298365404510657e-05, "loss": 5.9747, "step": 1271 }, { "epoch": 0.07896207089204792, "grad_norm": 0.8688590633076142, "learning_rate": 2.631905648665425e-05, "loss": 5.9397, "step": 1272 }, { "epoch": 0.07902414799180582, "grad_norm": 0.8583415319612901, "learning_rate": 2.633974756879785e-05, "loss": 5.9604, "step": 1273 }, { "epoch": 0.07908622509156372, "grad_norm": 1.0637575449154868, "learning_rate": 2.6360438650941445e-05, "loss": 5.9176, "step": 1274 }, { "epoch": 0.07914830219132162, "grad_norm": 0.8510721177093257, "learning_rate": 2.638112973308504e-05, "loss": 5.9494, "step": 1275 }, { "epoch": 0.07921037929107952, "grad_norm": 0.9335431520166025, "learning_rate": 2.640182081522864e-05, "loss": 5.9452, "step": 1276 }, { "epoch": 0.07927245639083742, "grad_norm": 0.9748591101161316, "learning_rate": 2.642251189737223e-05, "loss": 6.0094, "step": 1277 }, { "epoch": 0.07933453349059531, "grad_norm": 0.7350760826644395, "learning_rate": 2.644320297951583e-05, "loss": 6.0561, "step": 1278 }, { "epoch": 0.07939661059035322, "grad_norm": 0.9465600286170028, "learning_rate": 2.6463894061659427e-05, "loss": 5.972, "step": 1279 }, { "epoch": 0.07945868769011112, "grad_norm": 1.0593246245433359, "learning_rate": 2.648458514380302e-05, "loss": 5.8775, "step": 1280 }, { "epoch": 0.07952076478986901, "grad_norm": 1.077791355862465, "learning_rate": 2.6505276225946618e-05, "loss": 5.9514, "step": 1281 }, { "epoch": 0.07958284188962692, "grad_norm": 1.089525456590238, "learning_rate": 2.6525967308090216e-05, "loss": 5.9547, "step": 1282 }, { "epoch": 0.07964491898938482, "grad_norm": 1.00354205104532, "learning_rate": 2.6546658390233808e-05, "loss": 5.9481, "step": 1283 }, { "epoch": 0.07970699608914271, "grad_norm": 0.931845492504083, "learning_rate": 2.6567349472377406e-05, "loss": 5.9963, "step": 1284 }, { "epoch": 0.07976907318890061, "grad_norm": 0.7828752397381243, "learning_rate": 2.6588040554521005e-05, "loss": 6.0296, "step": 1285 }, { "epoch": 0.07983115028865852, "grad_norm": 0.8907394335663668, "learning_rate": 2.6608731636664596e-05, "loss": 6.0237, "step": 1286 }, { "epoch": 0.07989322738841641, "grad_norm": 0.86454862884974, "learning_rate": 2.6629422718808195e-05, "loss": 6.0357, "step": 1287 }, { "epoch": 0.07995530448817431, "grad_norm": 0.9165946253008644, "learning_rate": 2.6650113800951793e-05, "loss": 5.9342, "step": 1288 }, { "epoch": 0.08001738158793222, "grad_norm": 1.0192980737153554, "learning_rate": 2.6670804883095385e-05, "loss": 5.9292, "step": 1289 }, { "epoch": 0.0800794586876901, "grad_norm": 1.0211462916520693, "learning_rate": 2.6691495965238983e-05, "loss": 5.8776, "step": 1290 }, { "epoch": 0.08014153578744801, "grad_norm": 1.0662712039605626, "learning_rate": 2.6712187047382582e-05, "loss": 6.0127, "step": 1291 }, { "epoch": 0.08020361288720591, "grad_norm": 1.026568624399807, "learning_rate": 2.6732878129526174e-05, "loss": 5.9531, "step": 1292 }, { "epoch": 0.0802656899869638, "grad_norm": 0.9326247609415171, "learning_rate": 2.6753569211669772e-05, "loss": 5.9985, "step": 1293 }, { "epoch": 0.08032776708672171, "grad_norm": 0.8540572623613245, "learning_rate": 2.677426029381337e-05, "loss": 5.8861, "step": 1294 }, { "epoch": 0.08038984418647961, "grad_norm": 1.0071878197575084, "learning_rate": 2.6794951375956962e-05, "loss": 5.9054, "step": 1295 }, { "epoch": 0.0804519212862375, "grad_norm": 1.2887639017770558, "learning_rate": 2.681564245810056e-05, "loss": 5.9023, "step": 1296 }, { "epoch": 0.0805139983859954, "grad_norm": 0.9473493716000622, "learning_rate": 2.6836333540244156e-05, "loss": 5.9216, "step": 1297 }, { "epoch": 0.08057607548575331, "grad_norm": 1.1800413295814107, "learning_rate": 2.685702462238775e-05, "loss": 5.9893, "step": 1298 }, { "epoch": 0.0806381525855112, "grad_norm": 1.2285407568660993, "learning_rate": 2.687771570453135e-05, "loss": 5.9507, "step": 1299 }, { "epoch": 0.0807002296852691, "grad_norm": 0.987110484839277, "learning_rate": 2.6898406786674944e-05, "loss": 6.0497, "step": 1300 }, { "epoch": 0.08076230678502701, "grad_norm": 1.1088721519195408, "learning_rate": 2.691909786881854e-05, "loss": 5.9552, "step": 1301 }, { "epoch": 0.0808243838847849, "grad_norm": 0.9023552105328586, "learning_rate": 2.6939788950962134e-05, "loss": 6.0186, "step": 1302 }, { "epoch": 0.0808864609845428, "grad_norm": 0.9997866820263923, "learning_rate": 2.6960480033105733e-05, "loss": 5.9317, "step": 1303 }, { "epoch": 0.0809485380843007, "grad_norm": 1.247995008366946, "learning_rate": 2.6981171115249325e-05, "loss": 5.8737, "step": 1304 }, { "epoch": 0.0810106151840586, "grad_norm": 0.7650943396248603, "learning_rate": 2.7001862197392923e-05, "loss": 5.87, "step": 1305 }, { "epoch": 0.0810726922838165, "grad_norm": 0.8581653404339191, "learning_rate": 2.702255327953652e-05, "loss": 5.9955, "step": 1306 }, { "epoch": 0.0811347693835744, "grad_norm": 1.0687478874141048, "learning_rate": 2.7043244361680113e-05, "loss": 5.9776, "step": 1307 }, { "epoch": 0.0811968464833323, "grad_norm": 1.019789564359769, "learning_rate": 2.7063935443823712e-05, "loss": 5.8914, "step": 1308 }, { "epoch": 0.0812589235830902, "grad_norm": 0.876759123211665, "learning_rate": 2.708462652596731e-05, "loss": 6.0316, "step": 1309 }, { "epoch": 0.0813210006828481, "grad_norm": 1.1011214218679115, "learning_rate": 2.7105317608110902e-05, "loss": 5.9231, "step": 1310 }, { "epoch": 0.08138307778260599, "grad_norm": 0.8445615750637112, "learning_rate": 2.71260086902545e-05, "loss": 5.9574, "step": 1311 }, { "epoch": 0.0814451548823639, "grad_norm": 0.8116274329800715, "learning_rate": 2.71466997723981e-05, "loss": 5.9779, "step": 1312 }, { "epoch": 0.0815072319821218, "grad_norm": 0.8485182497155868, "learning_rate": 2.716739085454169e-05, "loss": 5.8562, "step": 1313 }, { "epoch": 0.08156930908187969, "grad_norm": 0.8999210548215, "learning_rate": 2.718808193668529e-05, "loss": 5.9168, "step": 1314 }, { "epoch": 0.0816313861816376, "grad_norm": 1.043207662700085, "learning_rate": 2.7208773018828887e-05, "loss": 5.9854, "step": 1315 }, { "epoch": 0.0816934632813955, "grad_norm": 0.9852666298412132, "learning_rate": 2.7229464100972486e-05, "loss": 5.9064, "step": 1316 }, { "epoch": 0.08175554038115339, "grad_norm": 0.8879302902525065, "learning_rate": 2.7250155183116078e-05, "loss": 5.9416, "step": 1317 }, { "epoch": 0.08181761748091129, "grad_norm": 1.0886939762036076, "learning_rate": 2.7270846265259676e-05, "loss": 5.8866, "step": 1318 }, { "epoch": 0.0818796945806692, "grad_norm": 1.0997524134563266, "learning_rate": 2.7291537347403274e-05, "loss": 5.9089, "step": 1319 }, { "epoch": 0.08194177168042709, "grad_norm": 0.7808218736598125, "learning_rate": 2.7312228429546866e-05, "loss": 5.8931, "step": 1320 }, { "epoch": 0.08200384878018499, "grad_norm": 0.649901380760345, "learning_rate": 2.7332919511690465e-05, "loss": 5.9585, "step": 1321 }, { "epoch": 0.0820659258799429, "grad_norm": 0.715197366270554, "learning_rate": 2.7353610593834063e-05, "loss": 6.0451, "step": 1322 }, { "epoch": 0.08212800297970078, "grad_norm": 0.7080978862423292, "learning_rate": 2.7374301675977655e-05, "loss": 5.9161, "step": 1323 }, { "epoch": 0.08219008007945869, "grad_norm": 0.7826780237971992, "learning_rate": 2.7394992758121253e-05, "loss": 5.87, "step": 1324 }, { "epoch": 0.08225215717921659, "grad_norm": 0.9508608519082168, "learning_rate": 2.741568384026485e-05, "loss": 5.914, "step": 1325 }, { "epoch": 0.08231423427897448, "grad_norm": 0.9655503685514015, "learning_rate": 2.7436374922408443e-05, "loss": 5.988, "step": 1326 }, { "epoch": 0.08237631137873239, "grad_norm": 1.0031985151976528, "learning_rate": 2.745706600455204e-05, "loss": 5.8474, "step": 1327 }, { "epoch": 0.08243838847849029, "grad_norm": 0.9168241466729301, "learning_rate": 2.7477757086695637e-05, "loss": 5.7747, "step": 1328 }, { "epoch": 0.08250046557824818, "grad_norm": 0.8540981932075737, "learning_rate": 2.7498448168839232e-05, "loss": 5.993, "step": 1329 }, { "epoch": 0.08256254267800608, "grad_norm": 0.8293545789583369, "learning_rate": 2.7519139250982827e-05, "loss": 5.9494, "step": 1330 }, { "epoch": 0.08262461977776399, "grad_norm": 0.8565879015825009, "learning_rate": 2.7539830333126426e-05, "loss": 5.8974, "step": 1331 }, { "epoch": 0.08268669687752188, "grad_norm": 0.7631736359340169, "learning_rate": 2.7560521415270017e-05, "loss": 5.9208, "step": 1332 }, { "epoch": 0.08274877397727978, "grad_norm": 1.038083813038013, "learning_rate": 2.7581212497413616e-05, "loss": 5.8439, "step": 1333 }, { "epoch": 0.08281085107703769, "grad_norm": 1.091150518997559, "learning_rate": 2.7601903579557214e-05, "loss": 5.9194, "step": 1334 }, { "epoch": 0.08287292817679558, "grad_norm": 0.9037664518006461, "learning_rate": 2.7622594661700806e-05, "loss": 5.7973, "step": 1335 }, { "epoch": 0.08293500527655348, "grad_norm": 0.7653189845528892, "learning_rate": 2.7643285743844404e-05, "loss": 5.8778, "step": 1336 }, { "epoch": 0.08299708237631138, "grad_norm": 0.8440411030198017, "learning_rate": 2.7663976825988003e-05, "loss": 5.8218, "step": 1337 }, { "epoch": 0.08305915947606927, "grad_norm": 0.706635780802877, "learning_rate": 2.7684667908131594e-05, "loss": 5.819, "step": 1338 }, { "epoch": 0.08312123657582718, "grad_norm": 0.7293732940732329, "learning_rate": 2.7705358990275193e-05, "loss": 5.9318, "step": 1339 }, { "epoch": 0.08318331367558508, "grad_norm": 0.8249505216037686, "learning_rate": 2.772605007241879e-05, "loss": 5.8642, "step": 1340 }, { "epoch": 0.08324539077534297, "grad_norm": 0.8041171302428104, "learning_rate": 2.7746741154562383e-05, "loss": 5.8576, "step": 1341 }, { "epoch": 0.08330746787510088, "grad_norm": 0.8712133830077693, "learning_rate": 2.776743223670598e-05, "loss": 5.9027, "step": 1342 }, { "epoch": 0.08336954497485878, "grad_norm": 0.8501231777472836, "learning_rate": 2.778812331884958e-05, "loss": 5.9615, "step": 1343 }, { "epoch": 0.08343162207461667, "grad_norm": 0.9099787592242908, "learning_rate": 2.7808814400993172e-05, "loss": 5.8668, "step": 1344 }, { "epoch": 0.08349369917437457, "grad_norm": 0.8782926324779662, "learning_rate": 2.782950548313677e-05, "loss": 5.8495, "step": 1345 }, { "epoch": 0.08355577627413248, "grad_norm": 0.8972514741385386, "learning_rate": 2.785019656528037e-05, "loss": 5.8983, "step": 1346 }, { "epoch": 0.08361785337389037, "grad_norm": 1.1119751881774675, "learning_rate": 2.787088764742396e-05, "loss": 5.8709, "step": 1347 }, { "epoch": 0.08367993047364827, "grad_norm": 0.8472517252477042, "learning_rate": 2.789157872956756e-05, "loss": 5.854, "step": 1348 }, { "epoch": 0.08374200757340618, "grad_norm": 0.824177637023993, "learning_rate": 2.7912269811711157e-05, "loss": 5.8431, "step": 1349 }, { "epoch": 0.08380408467316407, "grad_norm": 0.8907707496643801, "learning_rate": 2.793296089385475e-05, "loss": 5.8318, "step": 1350 }, { "epoch": 0.08386616177292197, "grad_norm": 0.998491759161641, "learning_rate": 2.7953651975998347e-05, "loss": 5.8595, "step": 1351 }, { "epoch": 0.08392823887267988, "grad_norm": 1.0048300207449068, "learning_rate": 2.7974343058141942e-05, "loss": 5.8236, "step": 1352 }, { "epoch": 0.08399031597243777, "grad_norm": 1.0552865481722757, "learning_rate": 2.7995034140285538e-05, "loss": 5.9434, "step": 1353 }, { "epoch": 0.08405239307219567, "grad_norm": 0.8088045075928626, "learning_rate": 2.8015725222429136e-05, "loss": 5.7961, "step": 1354 }, { "epoch": 0.08411447017195357, "grad_norm": 0.9877818960718534, "learning_rate": 2.803641630457273e-05, "loss": 5.9123, "step": 1355 }, { "epoch": 0.08417654727171146, "grad_norm": 1.1347710876858619, "learning_rate": 2.8057107386716326e-05, "loss": 5.9106, "step": 1356 }, { "epoch": 0.08423862437146937, "grad_norm": 0.8926621612454598, "learning_rate": 2.807779846885992e-05, "loss": 5.9242, "step": 1357 }, { "epoch": 0.08430070147122727, "grad_norm": 0.9519021366131072, "learning_rate": 2.809848955100352e-05, "loss": 5.8972, "step": 1358 }, { "epoch": 0.08436277857098516, "grad_norm": 0.7749953715893759, "learning_rate": 2.811918063314711e-05, "loss": 5.8433, "step": 1359 }, { "epoch": 0.08442485567074307, "grad_norm": 0.8128642586167472, "learning_rate": 2.813987171529071e-05, "loss": 5.7894, "step": 1360 }, { "epoch": 0.08448693277050096, "grad_norm": 0.8369306208820863, "learning_rate": 2.816056279743431e-05, "loss": 5.8367, "step": 1361 }, { "epoch": 0.08454900987025886, "grad_norm": 0.9283853916990358, "learning_rate": 2.81812538795779e-05, "loss": 5.8308, "step": 1362 }, { "epoch": 0.08461108697001676, "grad_norm": 0.9302417161055484, "learning_rate": 2.82019449617215e-05, "loss": 5.9533, "step": 1363 }, { "epoch": 0.08467316406977465, "grad_norm": 0.8452725272326017, "learning_rate": 2.8222636043865097e-05, "loss": 5.8454, "step": 1364 }, { "epoch": 0.08473524116953256, "grad_norm": 0.8943890454397689, "learning_rate": 2.824332712600869e-05, "loss": 5.8151, "step": 1365 }, { "epoch": 0.08479731826929046, "grad_norm": 0.8635366415898376, "learning_rate": 2.8264018208152287e-05, "loss": 5.8335, "step": 1366 }, { "epoch": 0.08485939536904835, "grad_norm": 0.9812751280853405, "learning_rate": 2.8284709290295886e-05, "loss": 5.8092, "step": 1367 }, { "epoch": 0.08492147246880626, "grad_norm": 0.8650185550507704, "learning_rate": 2.8305400372439477e-05, "loss": 5.9604, "step": 1368 }, { "epoch": 0.08498354956856416, "grad_norm": 0.6905494381058296, "learning_rate": 2.8326091454583076e-05, "loss": 5.9145, "step": 1369 }, { "epoch": 0.08504562666832205, "grad_norm": 0.7998868832487174, "learning_rate": 2.8346782536726674e-05, "loss": 5.9017, "step": 1370 }, { "epoch": 0.08510770376807995, "grad_norm": 0.7434537826621442, "learning_rate": 2.8367473618870266e-05, "loss": 5.8187, "step": 1371 }, { "epoch": 0.08516978086783786, "grad_norm": 0.974953539059235, "learning_rate": 2.8388164701013864e-05, "loss": 5.7632, "step": 1372 }, { "epoch": 0.08523185796759575, "grad_norm": 0.999858369334391, "learning_rate": 2.8408855783157463e-05, "loss": 5.9223, "step": 1373 }, { "epoch": 0.08529393506735365, "grad_norm": 1.0922330418425208, "learning_rate": 2.8429546865301055e-05, "loss": 5.8484, "step": 1374 }, { "epoch": 0.08535601216711156, "grad_norm": 1.0603891610186262, "learning_rate": 2.8450237947444653e-05, "loss": 5.8042, "step": 1375 }, { "epoch": 0.08541808926686945, "grad_norm": 0.7451846190971748, "learning_rate": 2.847092902958825e-05, "loss": 5.8056, "step": 1376 }, { "epoch": 0.08548016636662735, "grad_norm": 0.9093419963707413, "learning_rate": 2.8491620111731843e-05, "loss": 5.8515, "step": 1377 }, { "epoch": 0.08554224346638525, "grad_norm": 1.0439916730672893, "learning_rate": 2.851231119387544e-05, "loss": 5.7028, "step": 1378 }, { "epoch": 0.08560432056614314, "grad_norm": 0.7909360115582867, "learning_rate": 2.853300227601904e-05, "loss": 5.8499, "step": 1379 }, { "epoch": 0.08566639766590105, "grad_norm": 0.8085877422902008, "learning_rate": 2.8553693358162632e-05, "loss": 5.8042, "step": 1380 }, { "epoch": 0.08572847476565895, "grad_norm": 0.7197304962340557, "learning_rate": 2.857438444030623e-05, "loss": 5.8267, "step": 1381 }, { "epoch": 0.08579055186541684, "grad_norm": 0.8033074702324523, "learning_rate": 2.8595075522449825e-05, "loss": 5.8063, "step": 1382 }, { "epoch": 0.08585262896517475, "grad_norm": 0.8297818361951976, "learning_rate": 2.861576660459342e-05, "loss": 5.8606, "step": 1383 }, { "epoch": 0.08591470606493265, "grad_norm": 0.9718353300185854, "learning_rate": 2.8636457686737015e-05, "loss": 5.8032, "step": 1384 }, { "epoch": 0.08597678316469054, "grad_norm": 1.0527246931264258, "learning_rate": 2.8657148768880614e-05, "loss": 5.9116, "step": 1385 }, { "epoch": 0.08603886026444844, "grad_norm": 1.0418388732980925, "learning_rate": 2.867783985102421e-05, "loss": 5.7872, "step": 1386 }, { "epoch": 0.08610093736420635, "grad_norm": 1.124954805525988, "learning_rate": 2.8698530933167804e-05, "loss": 5.8547, "step": 1387 }, { "epoch": 0.08616301446396424, "grad_norm": 1.0355026529223812, "learning_rate": 2.8719222015311403e-05, "loss": 5.8435, "step": 1388 }, { "epoch": 0.08622509156372214, "grad_norm": 0.9174041635109045, "learning_rate": 2.8739913097454994e-05, "loss": 5.8557, "step": 1389 }, { "epoch": 0.08628716866348005, "grad_norm": 0.7990273224953677, "learning_rate": 2.8760604179598593e-05, "loss": 5.8203, "step": 1390 }, { "epoch": 0.08634924576323794, "grad_norm": 0.8241074834993241, "learning_rate": 2.878129526174219e-05, "loss": 5.8215, "step": 1391 }, { "epoch": 0.08641132286299584, "grad_norm": 0.8083599323799097, "learning_rate": 2.8801986343885783e-05, "loss": 5.8395, "step": 1392 }, { "epoch": 0.08647339996275374, "grad_norm": 0.9387812124922592, "learning_rate": 2.882267742602938e-05, "loss": 5.8551, "step": 1393 }, { "epoch": 0.08653547706251163, "grad_norm": 1.0200233295054042, "learning_rate": 2.884336850817298e-05, "loss": 5.8068, "step": 1394 }, { "epoch": 0.08659755416226954, "grad_norm": 1.0394814012559959, "learning_rate": 2.886405959031657e-05, "loss": 5.8795, "step": 1395 }, { "epoch": 0.08665963126202744, "grad_norm": 1.098689668667686, "learning_rate": 2.888475067246017e-05, "loss": 5.8935, "step": 1396 }, { "epoch": 0.08672170836178533, "grad_norm": 0.8644421052294998, "learning_rate": 2.890544175460377e-05, "loss": 5.6983, "step": 1397 }, { "epoch": 0.08678378546154324, "grad_norm": 1.1344397071899712, "learning_rate": 2.892613283674736e-05, "loss": 5.8086, "step": 1398 }, { "epoch": 0.08684586256130114, "grad_norm": 1.046609873435585, "learning_rate": 2.894682391889096e-05, "loss": 5.9381, "step": 1399 }, { "epoch": 0.08690793966105903, "grad_norm": 0.8429441489093699, "learning_rate": 2.8967515001034557e-05, "loss": 5.8489, "step": 1400 }, { "epoch": 0.08697001676081693, "grad_norm": 0.934316762174576, "learning_rate": 2.8988206083178155e-05, "loss": 5.8511, "step": 1401 }, { "epoch": 0.08703209386057484, "grad_norm": 1.0256895955518608, "learning_rate": 2.9008897165321747e-05, "loss": 5.9169, "step": 1402 }, { "epoch": 0.08709417096033273, "grad_norm": 0.9758886835529796, "learning_rate": 2.9029588247465346e-05, "loss": 5.7414, "step": 1403 }, { "epoch": 0.08715624806009063, "grad_norm": 0.9962485997131539, "learning_rate": 2.9050279329608944e-05, "loss": 5.7965, "step": 1404 }, { "epoch": 0.08721832515984854, "grad_norm": 0.9105042213978912, "learning_rate": 2.9070970411752536e-05, "loss": 5.8187, "step": 1405 }, { "epoch": 0.08728040225960643, "grad_norm": 0.8341031135169855, "learning_rate": 2.9091661493896134e-05, "loss": 5.7837, "step": 1406 }, { "epoch": 0.08734247935936433, "grad_norm": 0.914099765898246, "learning_rate": 2.911235257603973e-05, "loss": 5.8151, "step": 1407 }, { "epoch": 0.08740455645912223, "grad_norm": 0.9211092532651342, "learning_rate": 2.9133043658183324e-05, "loss": 5.7957, "step": 1408 }, { "epoch": 0.08746663355888012, "grad_norm": 1.0753978569655005, "learning_rate": 2.9153734740326923e-05, "loss": 5.7921, "step": 1409 }, { "epoch": 0.08752871065863803, "grad_norm": 1.1170531505551387, "learning_rate": 2.9174425822470518e-05, "loss": 5.8678, "step": 1410 }, { "epoch": 0.08759078775839593, "grad_norm": 1.0571548987314037, "learning_rate": 2.9195116904614113e-05, "loss": 5.8741, "step": 1411 }, { "epoch": 0.08765286485815382, "grad_norm": 1.1544490913939693, "learning_rate": 2.9215807986757708e-05, "loss": 5.7462, "step": 1412 }, { "epoch": 0.08771494195791173, "grad_norm": 0.9329797169953249, "learning_rate": 2.9236499068901307e-05, "loss": 5.8667, "step": 1413 }, { "epoch": 0.08777701905766963, "grad_norm": 1.175355194054745, "learning_rate": 2.9257190151044898e-05, "loss": 5.842, "step": 1414 }, { "epoch": 0.08783909615742752, "grad_norm": 1.250139467088347, "learning_rate": 2.9277881233188497e-05, "loss": 5.7323, "step": 1415 }, { "epoch": 0.08790117325718542, "grad_norm": 0.9348639940597636, "learning_rate": 2.9298572315332095e-05, "loss": 5.7468, "step": 1416 }, { "epoch": 0.08796325035694333, "grad_norm": 1.1275324927171162, "learning_rate": 2.9319263397475687e-05, "loss": 5.7668, "step": 1417 }, { "epoch": 0.08802532745670122, "grad_norm": 1.1056883361340586, "learning_rate": 2.9339954479619285e-05, "loss": 5.8777, "step": 1418 }, { "epoch": 0.08808740455645912, "grad_norm": 0.814317943597257, "learning_rate": 2.9360645561762884e-05, "loss": 5.7808, "step": 1419 }, { "epoch": 0.08814948165621703, "grad_norm": 1.0612367525385933, "learning_rate": 2.9381336643906475e-05, "loss": 5.7852, "step": 1420 }, { "epoch": 0.08821155875597492, "grad_norm": 1.0450447937157994, "learning_rate": 2.9402027726050074e-05, "loss": 5.793, "step": 1421 }, { "epoch": 0.08827363585573282, "grad_norm": 0.955911752304797, "learning_rate": 2.9422718808193672e-05, "loss": 5.7734, "step": 1422 }, { "epoch": 0.08833571295549072, "grad_norm": 0.984851600679835, "learning_rate": 2.9443409890337264e-05, "loss": 5.7228, "step": 1423 }, { "epoch": 0.08839779005524862, "grad_norm": 1.066891570272636, "learning_rate": 2.9464100972480863e-05, "loss": 5.8261, "step": 1424 }, { "epoch": 0.08845986715500652, "grad_norm": 0.8206113724982502, "learning_rate": 2.948479205462446e-05, "loss": 5.9217, "step": 1425 }, { "epoch": 0.08852194425476442, "grad_norm": 0.9319086548978974, "learning_rate": 2.9505483136768053e-05, "loss": 5.8332, "step": 1426 }, { "epoch": 0.08858402135452231, "grad_norm": 0.7929579635412661, "learning_rate": 2.952617421891165e-05, "loss": 5.7557, "step": 1427 }, { "epoch": 0.08864609845428022, "grad_norm": 0.9859176234737015, "learning_rate": 2.954686530105525e-05, "loss": 5.7757, "step": 1428 }, { "epoch": 0.08870817555403812, "grad_norm": 1.111801288084796, "learning_rate": 2.956755638319884e-05, "loss": 5.751, "step": 1429 }, { "epoch": 0.08877025265379601, "grad_norm": 0.8678624044210622, "learning_rate": 2.958824746534244e-05, "loss": 5.7648, "step": 1430 }, { "epoch": 0.08883232975355392, "grad_norm": 1.0054920896297714, "learning_rate": 2.9608938547486038e-05, "loss": 5.7016, "step": 1431 }, { "epoch": 0.08889440685331182, "grad_norm": 0.9092941785730282, "learning_rate": 2.962962962962963e-05, "loss": 5.8661, "step": 1432 }, { "epoch": 0.08895648395306971, "grad_norm": 1.099431133561283, "learning_rate": 2.965032071177323e-05, "loss": 5.6922, "step": 1433 }, { "epoch": 0.08901856105282761, "grad_norm": 0.8542503843141963, "learning_rate": 2.9671011793916827e-05, "loss": 5.7069, "step": 1434 }, { "epoch": 0.08908063815258552, "grad_norm": 0.9951561797554229, "learning_rate": 2.969170287606042e-05, "loss": 5.7422, "step": 1435 }, { "epoch": 0.08914271525234341, "grad_norm": 0.994997985109369, "learning_rate": 2.9712393958204017e-05, "loss": 5.807, "step": 1436 }, { "epoch": 0.08920479235210131, "grad_norm": 0.8820910586410896, "learning_rate": 2.9733085040347612e-05, "loss": 5.7595, "step": 1437 }, { "epoch": 0.08926686945185922, "grad_norm": 1.1635825684728478, "learning_rate": 2.9753776122491207e-05, "loss": 5.8209, "step": 1438 }, { "epoch": 0.0893289465516171, "grad_norm": 0.9555435138936765, "learning_rate": 2.9774467204634802e-05, "loss": 5.7014, "step": 1439 }, { "epoch": 0.08939102365137501, "grad_norm": 0.84495109264209, "learning_rate": 2.97951582867784e-05, "loss": 5.7297, "step": 1440 }, { "epoch": 0.08945310075113291, "grad_norm": 0.9108346203768456, "learning_rate": 2.9815849368921996e-05, "loss": 5.7859, "step": 1441 }, { "epoch": 0.0895151778508908, "grad_norm": 1.1766342161493015, "learning_rate": 2.983654045106559e-05, "loss": 5.7942, "step": 1442 }, { "epoch": 0.08957725495064871, "grad_norm": 0.8422740106259855, "learning_rate": 2.985723153320919e-05, "loss": 5.8653, "step": 1443 }, { "epoch": 0.08963933205040661, "grad_norm": 0.918443857166449, "learning_rate": 2.987792261535278e-05, "loss": 5.8769, "step": 1444 }, { "epoch": 0.0897014091501645, "grad_norm": 1.1320335604036604, "learning_rate": 2.989861369749638e-05, "loss": 5.7961, "step": 1445 }, { "epoch": 0.0897634862499224, "grad_norm": 0.9532565354434084, "learning_rate": 2.9919304779639978e-05, "loss": 5.8031, "step": 1446 }, { "epoch": 0.08982556334968031, "grad_norm": 0.979503506017917, "learning_rate": 2.993999586178357e-05, "loss": 5.7585, "step": 1447 }, { "epoch": 0.0898876404494382, "grad_norm": 0.969701883822194, "learning_rate": 2.9960686943927168e-05, "loss": 5.7911, "step": 1448 }, { "epoch": 0.0899497175491961, "grad_norm": 1.0232403397839243, "learning_rate": 2.9981378026070767e-05, "loss": 5.7435, "step": 1449 }, { "epoch": 0.09001179464895401, "grad_norm": 0.7892833438435759, "learning_rate": 3.0002069108214358e-05, "loss": 5.7398, "step": 1450 }, { "epoch": 0.0900738717487119, "grad_norm": 0.8769496246471936, "learning_rate": 3.0022760190357957e-05, "loss": 5.813, "step": 1451 }, { "epoch": 0.0901359488484698, "grad_norm": 0.9072897963973579, "learning_rate": 3.0043451272501555e-05, "loss": 5.7063, "step": 1452 }, { "epoch": 0.09019802594822769, "grad_norm": 0.8879852996309608, "learning_rate": 3.0064142354645147e-05, "loss": 5.7728, "step": 1453 }, { "epoch": 0.0902601030479856, "grad_norm": 0.9355994988217098, "learning_rate": 3.0084833436788745e-05, "loss": 5.7664, "step": 1454 }, { "epoch": 0.0903221801477435, "grad_norm": 0.9141498100119906, "learning_rate": 3.0105524518932344e-05, "loss": 5.8612, "step": 1455 }, { "epoch": 0.09038425724750139, "grad_norm": 0.8712331207041416, "learning_rate": 3.0126215601075935e-05, "loss": 5.8818, "step": 1456 }, { "epoch": 0.0904463343472593, "grad_norm": 0.9669429784253056, "learning_rate": 3.0146906683219534e-05, "loss": 5.7436, "step": 1457 }, { "epoch": 0.0905084114470172, "grad_norm": 0.9455579013555048, "learning_rate": 3.0167597765363132e-05, "loss": 5.8379, "step": 1458 }, { "epoch": 0.09057048854677509, "grad_norm": 0.8586831515750668, "learning_rate": 3.0188288847506724e-05, "loss": 5.71, "step": 1459 }, { "epoch": 0.09063256564653299, "grad_norm": 0.8814340818515098, "learning_rate": 3.0208979929650323e-05, "loss": 5.7017, "step": 1460 }, { "epoch": 0.0906946427462909, "grad_norm": 0.8475939116870034, "learning_rate": 3.022967101179392e-05, "loss": 5.7938, "step": 1461 }, { "epoch": 0.09075671984604879, "grad_norm": 0.8468487316716923, "learning_rate": 3.0250362093937513e-05, "loss": 5.7439, "step": 1462 }, { "epoch": 0.09081879694580669, "grad_norm": 1.045339454565648, "learning_rate": 3.027105317608111e-05, "loss": 5.7471, "step": 1463 }, { "epoch": 0.0908808740455646, "grad_norm": 1.1844885437018071, "learning_rate": 3.029174425822471e-05, "loss": 5.7758, "step": 1464 }, { "epoch": 0.09094295114532248, "grad_norm": 0.7536331235441779, "learning_rate": 3.03124353403683e-05, "loss": 5.7382, "step": 1465 }, { "epoch": 0.09100502824508039, "grad_norm": 0.8478415150498523, "learning_rate": 3.03331264225119e-05, "loss": 5.7254, "step": 1466 }, { "epoch": 0.09106710534483829, "grad_norm": 1.0072184741553694, "learning_rate": 3.0353817504655495e-05, "loss": 5.6303, "step": 1467 }, { "epoch": 0.09112918244459618, "grad_norm": 0.9604216330697055, "learning_rate": 3.037450858679909e-05, "loss": 5.7871, "step": 1468 }, { "epoch": 0.09119125954435409, "grad_norm": 1.0388806979955223, "learning_rate": 3.0395199668942685e-05, "loss": 5.8497, "step": 1469 }, { "epoch": 0.09125333664411199, "grad_norm": 0.9856514904306054, "learning_rate": 3.0415890751086283e-05, "loss": 5.6569, "step": 1470 }, { "epoch": 0.09131541374386988, "grad_norm": 1.1544982424690016, "learning_rate": 3.043658183322988e-05, "loss": 5.7372, "step": 1471 }, { "epoch": 0.09137749084362778, "grad_norm": 0.9507479183024774, "learning_rate": 3.0457272915373474e-05, "loss": 5.692, "step": 1472 }, { "epoch": 0.09143956794338569, "grad_norm": 0.9501261828093612, "learning_rate": 3.0477963997517072e-05, "loss": 5.7553, "step": 1473 }, { "epoch": 0.09150164504314358, "grad_norm": 1.0962792716038494, "learning_rate": 3.0498655079660664e-05, "loss": 5.6759, "step": 1474 }, { "epoch": 0.09156372214290148, "grad_norm": 0.8758825211243556, "learning_rate": 3.051934616180426e-05, "loss": 5.7219, "step": 1475 }, { "epoch": 0.09162579924265939, "grad_norm": 0.8691482715056019, "learning_rate": 3.0540037243947864e-05, "loss": 5.7258, "step": 1476 }, { "epoch": 0.09168787634241728, "grad_norm": 0.8278377682612049, "learning_rate": 3.056072832609145e-05, "loss": 5.7584, "step": 1477 }, { "epoch": 0.09174995344217518, "grad_norm": 0.8482389424978045, "learning_rate": 3.0581419408235054e-05, "loss": 5.6536, "step": 1478 }, { "epoch": 0.09181203054193308, "grad_norm": 0.8832582681777983, "learning_rate": 3.060211049037865e-05, "loss": 5.8102, "step": 1479 }, { "epoch": 0.09187410764169097, "grad_norm": 1.0395210802987587, "learning_rate": 3.0622801572522244e-05, "loss": 5.6983, "step": 1480 }, { "epoch": 0.09193618474144888, "grad_norm": 0.8664858273284273, "learning_rate": 3.064349265466584e-05, "loss": 5.6779, "step": 1481 }, { "epoch": 0.09199826184120678, "grad_norm": 1.0283478446093202, "learning_rate": 3.0664183736809435e-05, "loss": 5.7436, "step": 1482 }, { "epoch": 0.09206033894096467, "grad_norm": 1.287503496183815, "learning_rate": 3.068487481895303e-05, "loss": 5.6571, "step": 1483 }, { "epoch": 0.09212241604072258, "grad_norm": 1.103553163683281, "learning_rate": 3.0705565901096625e-05, "loss": 5.7694, "step": 1484 }, { "epoch": 0.09218449314048048, "grad_norm": 0.8924332382171987, "learning_rate": 3.0726256983240227e-05, "loss": 5.717, "step": 1485 }, { "epoch": 0.09224657024023837, "grad_norm": 0.7940525859887938, "learning_rate": 3.074694806538382e-05, "loss": 5.8093, "step": 1486 }, { "epoch": 0.09230864733999627, "grad_norm": 1.016264083501959, "learning_rate": 3.076763914752742e-05, "loss": 5.7496, "step": 1487 }, { "epoch": 0.09237072443975418, "grad_norm": 0.8632220770117139, "learning_rate": 3.078833022967101e-05, "loss": 5.7754, "step": 1488 }, { "epoch": 0.09243280153951207, "grad_norm": 0.858219375421771, "learning_rate": 3.0809021311814614e-05, "loss": 5.7452, "step": 1489 }, { "epoch": 0.09249487863926997, "grad_norm": 0.8925662718138191, "learning_rate": 3.08297123939582e-05, "loss": 5.6798, "step": 1490 }, { "epoch": 0.09255695573902788, "grad_norm": 0.858846918325719, "learning_rate": 3.0850403476101804e-05, "loss": 5.7841, "step": 1491 }, { "epoch": 0.09261903283878577, "grad_norm": 0.7675399964935679, "learning_rate": 3.08710945582454e-05, "loss": 5.765, "step": 1492 }, { "epoch": 0.09268110993854367, "grad_norm": 0.8566105856022944, "learning_rate": 3.0891785640388994e-05, "loss": 5.7086, "step": 1493 }, { "epoch": 0.09274318703830157, "grad_norm": 0.7603617391387871, "learning_rate": 3.091247672253259e-05, "loss": 5.7606, "step": 1494 }, { "epoch": 0.09280526413805947, "grad_norm": 0.9210644407519156, "learning_rate": 3.093316780467619e-05, "loss": 5.7494, "step": 1495 }, { "epoch": 0.09286734123781737, "grad_norm": 1.2689999451633498, "learning_rate": 3.095385888681978e-05, "loss": 5.7823, "step": 1496 }, { "epoch": 0.09292941833757527, "grad_norm": 0.8314907485415131, "learning_rate": 3.097454996896338e-05, "loss": 5.6689, "step": 1497 }, { "epoch": 0.09299149543733316, "grad_norm": 0.8756095602238556, "learning_rate": 3.0995241051106976e-05, "loss": 5.7112, "step": 1498 }, { "epoch": 0.09305357253709107, "grad_norm": 1.2275098798707726, "learning_rate": 3.101593213325057e-05, "loss": 5.7315, "step": 1499 }, { "epoch": 0.09311564963684897, "grad_norm": 0.9344829331131003, "learning_rate": 3.1036623215394166e-05, "loss": 5.5472, "step": 1500 }, { "epoch": 0.09317772673660686, "grad_norm": 0.8708284940691615, "learning_rate": 3.105731429753777e-05, "loss": 5.7183, "step": 1501 }, { "epoch": 0.09323980383636477, "grad_norm": 0.7916473879892788, "learning_rate": 3.1078005379681356e-05, "loss": 5.733, "step": 1502 }, { "epoch": 0.09330188093612267, "grad_norm": 0.8797991138070498, "learning_rate": 3.109869646182496e-05, "loss": 5.614, "step": 1503 }, { "epoch": 0.09336395803588056, "grad_norm": 1.0196489288114945, "learning_rate": 3.111938754396855e-05, "loss": 5.7035, "step": 1504 }, { "epoch": 0.09342603513563846, "grad_norm": 1.0445458992477907, "learning_rate": 3.114007862611215e-05, "loss": 5.7879, "step": 1505 }, { "epoch": 0.09348811223539637, "grad_norm": 0.8967681967380855, "learning_rate": 3.1160769708255743e-05, "loss": 5.634, "step": 1506 }, { "epoch": 0.09355018933515426, "grad_norm": 0.7584489725484559, "learning_rate": 3.118146079039934e-05, "loss": 5.7327, "step": 1507 }, { "epoch": 0.09361226643491216, "grad_norm": 0.875020571875567, "learning_rate": 3.1202151872542934e-05, "loss": 5.7065, "step": 1508 }, { "epoch": 0.09367434353467007, "grad_norm": 0.9083762711400191, "learning_rate": 3.122284295468653e-05, "loss": 5.6661, "step": 1509 }, { "epoch": 0.09373642063442796, "grad_norm": 0.8681696579434243, "learning_rate": 3.124353403683013e-05, "loss": 5.6105, "step": 1510 }, { "epoch": 0.09379849773418586, "grad_norm": 0.851469262294634, "learning_rate": 3.1264225118973726e-05, "loss": 5.6787, "step": 1511 }, { "epoch": 0.09386057483394376, "grad_norm": 0.9327800189166713, "learning_rate": 3.128491620111732e-05, "loss": 5.7515, "step": 1512 }, { "epoch": 0.09392265193370165, "grad_norm": 1.0214354756833626, "learning_rate": 3.1305607283260916e-05, "loss": 5.7322, "step": 1513 }, { "epoch": 0.09398472903345956, "grad_norm": 0.9815797324673342, "learning_rate": 3.132629836540451e-05, "loss": 5.717, "step": 1514 }, { "epoch": 0.09404680613321746, "grad_norm": 1.157114839593085, "learning_rate": 3.1346989447548106e-05, "loss": 5.6337, "step": 1515 }, { "epoch": 0.09410888323297535, "grad_norm": 1.0052269586343043, "learning_rate": 3.136768052969171e-05, "loss": 5.6829, "step": 1516 }, { "epoch": 0.09417096033273326, "grad_norm": 0.9068757231884944, "learning_rate": 3.1388371611835296e-05, "loss": 5.6012, "step": 1517 }, { "epoch": 0.09423303743249116, "grad_norm": 1.027302730307065, "learning_rate": 3.14090626939789e-05, "loss": 5.6041, "step": 1518 }, { "epoch": 0.09429511453224905, "grad_norm": 0.7792406181355673, "learning_rate": 3.142975377612249e-05, "loss": 5.7075, "step": 1519 }, { "epoch": 0.09435719163200695, "grad_norm": 0.7785351694795664, "learning_rate": 3.145044485826609e-05, "loss": 5.6584, "step": 1520 }, { "epoch": 0.09441926873176486, "grad_norm": 1.011111279605204, "learning_rate": 3.147113594040968e-05, "loss": 5.6457, "step": 1521 }, { "epoch": 0.09448134583152275, "grad_norm": 1.145341866026977, "learning_rate": 3.1491827022553285e-05, "loss": 5.7214, "step": 1522 }, { "epoch": 0.09454342293128065, "grad_norm": 1.0427512181342835, "learning_rate": 3.151251810469687e-05, "loss": 5.6835, "step": 1523 }, { "epoch": 0.09460550003103856, "grad_norm": 1.105591783989605, "learning_rate": 3.1533209186840475e-05, "loss": 5.7081, "step": 1524 }, { "epoch": 0.09466757713079645, "grad_norm": 1.021782738489892, "learning_rate": 3.155390026898407e-05, "loss": 5.677, "step": 1525 }, { "epoch": 0.09472965423055435, "grad_norm": 1.0280064523175279, "learning_rate": 3.1574591351127665e-05, "loss": 5.6723, "step": 1526 }, { "epoch": 0.09479173133031225, "grad_norm": 1.0371316738569103, "learning_rate": 3.159528243327126e-05, "loss": 5.6858, "step": 1527 }, { "epoch": 0.09485380843007014, "grad_norm": 0.96098231389446, "learning_rate": 3.161597351541486e-05, "loss": 5.6802, "step": 1528 }, { "epoch": 0.09491588552982805, "grad_norm": 1.0581257478876958, "learning_rate": 3.163666459755845e-05, "loss": 5.6833, "step": 1529 }, { "epoch": 0.09497796262958595, "grad_norm": 0.9723180466309989, "learning_rate": 3.165735567970205e-05, "loss": 5.6489, "step": 1530 }, { "epoch": 0.09504003972934384, "grad_norm": 1.0645758041941735, "learning_rate": 3.167804676184565e-05, "loss": 5.6634, "step": 1531 }, { "epoch": 0.09510211682910175, "grad_norm": 0.9409360225669692, "learning_rate": 3.169873784398924e-05, "loss": 5.5654, "step": 1532 }, { "epoch": 0.09516419392885965, "grad_norm": 0.8789390388306576, "learning_rate": 3.171942892613284e-05, "loss": 5.7052, "step": 1533 }, { "epoch": 0.09522627102861754, "grad_norm": 0.8630420474244338, "learning_rate": 3.174012000827643e-05, "loss": 5.546, "step": 1534 }, { "epoch": 0.09528834812837544, "grad_norm": 0.8035016627528915, "learning_rate": 3.176081109042003e-05, "loss": 5.6077, "step": 1535 }, { "epoch": 0.09535042522813335, "grad_norm": 0.823881432810415, "learning_rate": 3.178150217256363e-05, "loss": 5.6816, "step": 1536 }, { "epoch": 0.09541250232789124, "grad_norm": 0.8559603040929658, "learning_rate": 3.1802193254707225e-05, "loss": 5.6629, "step": 1537 }, { "epoch": 0.09547457942764914, "grad_norm": 0.9401927740690478, "learning_rate": 3.182288433685082e-05, "loss": 5.6793, "step": 1538 }, { "epoch": 0.09553665652740705, "grad_norm": 0.9059561549013156, "learning_rate": 3.1843575418994415e-05, "loss": 5.6467, "step": 1539 }, { "epoch": 0.09559873362716494, "grad_norm": 0.8919565720335774, "learning_rate": 3.186426650113801e-05, "loss": 5.6656, "step": 1540 }, { "epoch": 0.09566081072692284, "grad_norm": 0.8086754209396282, "learning_rate": 3.1884957583281605e-05, "loss": 5.6383, "step": 1541 }, { "epoch": 0.09572288782668074, "grad_norm": 0.8865474585896612, "learning_rate": 3.19056486654252e-05, "loss": 5.6756, "step": 1542 }, { "epoch": 0.09578496492643863, "grad_norm": 0.9278574008683287, "learning_rate": 3.19263397475688e-05, "loss": 5.7055, "step": 1543 }, { "epoch": 0.09584704202619654, "grad_norm": 1.010631433583237, "learning_rate": 3.194703082971239e-05, "loss": 5.6848, "step": 1544 }, { "epoch": 0.09590911912595444, "grad_norm": 1.0875808028095064, "learning_rate": 3.196772191185599e-05, "loss": 5.6624, "step": 1545 }, { "epoch": 0.09597119622571233, "grad_norm": 0.8420918737699133, "learning_rate": 3.198841299399959e-05, "loss": 5.6435, "step": 1546 }, { "epoch": 0.09603327332547024, "grad_norm": 0.8417816868803675, "learning_rate": 3.200910407614318e-05, "loss": 5.7737, "step": 1547 }, { "epoch": 0.09609535042522813, "grad_norm": 0.9255572019754211, "learning_rate": 3.202979515828678e-05, "loss": 5.7171, "step": 1548 }, { "epoch": 0.09615742752498603, "grad_norm": 0.8674546938451734, "learning_rate": 3.205048624043038e-05, "loss": 5.6829, "step": 1549 }, { "epoch": 0.09621950462474393, "grad_norm": 0.906859160271146, "learning_rate": 3.207117732257397e-05, "loss": 5.716, "step": 1550 }, { "epoch": 0.09628158172450182, "grad_norm": 0.965780907300522, "learning_rate": 3.209186840471757e-05, "loss": 5.6349, "step": 1551 }, { "epoch": 0.09634365882425973, "grad_norm": 0.9381389665675863, "learning_rate": 3.2112559486861164e-05, "loss": 5.5603, "step": 1552 }, { "epoch": 0.09640573592401763, "grad_norm": 1.0370465563124018, "learning_rate": 3.213325056900476e-05, "loss": 5.6613, "step": 1553 }, { "epoch": 0.09646781302377552, "grad_norm": 0.8731989717380483, "learning_rate": 3.2153941651148355e-05, "loss": 5.6699, "step": 1554 }, { "epoch": 0.09652989012353343, "grad_norm": 0.9469751600917096, "learning_rate": 3.2174632733291956e-05, "loss": 5.6695, "step": 1555 }, { "epoch": 0.09659196722329133, "grad_norm": 1.0811675993992433, "learning_rate": 3.2195323815435545e-05, "loss": 5.6956, "step": 1556 }, { "epoch": 0.09665404432304922, "grad_norm": 1.024416448046704, "learning_rate": 3.2216014897579147e-05, "loss": 5.6758, "step": 1557 }, { "epoch": 0.09671612142280712, "grad_norm": 0.7561565672480484, "learning_rate": 3.223670597972274e-05, "loss": 5.6938, "step": 1558 }, { "epoch": 0.09677819852256503, "grad_norm": 1.0575173173212467, "learning_rate": 3.225739706186634e-05, "loss": 5.6474, "step": 1559 }, { "epoch": 0.09684027562232292, "grad_norm": 1.0283272100871172, "learning_rate": 3.227808814400993e-05, "loss": 5.6415, "step": 1560 }, { "epoch": 0.09690235272208082, "grad_norm": 0.7682340891190063, "learning_rate": 3.2298779226153534e-05, "loss": 5.6083, "step": 1561 }, { "epoch": 0.09696442982183873, "grad_norm": 0.7598310113929815, "learning_rate": 3.231947030829712e-05, "loss": 5.6335, "step": 1562 }, { "epoch": 0.09702650692159662, "grad_norm": 1.0549648909127993, "learning_rate": 3.2340161390440724e-05, "loss": 5.6042, "step": 1563 }, { "epoch": 0.09708858402135452, "grad_norm": 1.1410913428571485, "learning_rate": 3.236085247258432e-05, "loss": 5.5286, "step": 1564 }, { "epoch": 0.09715066112111242, "grad_norm": 1.0995938721778504, "learning_rate": 3.2381543554727914e-05, "loss": 5.6206, "step": 1565 }, { "epoch": 0.09721273822087032, "grad_norm": 1.015941981220952, "learning_rate": 3.240223463687151e-05, "loss": 5.5737, "step": 1566 }, { "epoch": 0.09727481532062822, "grad_norm": 0.9748273464314049, "learning_rate": 3.2422925719015104e-05, "loss": 5.7369, "step": 1567 }, { "epoch": 0.09733689242038612, "grad_norm": 0.9424257891248837, "learning_rate": 3.24436168011587e-05, "loss": 5.566, "step": 1568 }, { "epoch": 0.09739896952014401, "grad_norm": 1.0638726847958129, "learning_rate": 3.2464307883302294e-05, "loss": 5.6591, "step": 1569 }, { "epoch": 0.09746104661990192, "grad_norm": 1.1653645037716744, "learning_rate": 3.2484998965445896e-05, "loss": 5.6236, "step": 1570 }, { "epoch": 0.09752312371965982, "grad_norm": 0.8738963114601147, "learning_rate": 3.250569004758949e-05, "loss": 5.591, "step": 1571 }, { "epoch": 0.09758520081941771, "grad_norm": 0.8292546754325143, "learning_rate": 3.2526381129733086e-05, "loss": 5.6908, "step": 1572 }, { "epoch": 0.09764727791917562, "grad_norm": 1.0646237297879129, "learning_rate": 3.254707221187668e-05, "loss": 5.5732, "step": 1573 }, { "epoch": 0.09770935501893352, "grad_norm": 1.2587320982219765, "learning_rate": 3.256776329402028e-05, "loss": 5.6015, "step": 1574 }, { "epoch": 0.09777143211869141, "grad_norm": 0.8598170444029859, "learning_rate": 3.258845437616387e-05, "loss": 5.6153, "step": 1575 }, { "epoch": 0.09783350921844931, "grad_norm": 1.1090700509466476, "learning_rate": 3.260914545830747e-05, "loss": 5.7818, "step": 1576 }, { "epoch": 0.09789558631820722, "grad_norm": 1.157100210253329, "learning_rate": 3.262983654045107e-05, "loss": 5.6384, "step": 1577 }, { "epoch": 0.09795766341796511, "grad_norm": 0.7729297088454933, "learning_rate": 3.2650527622594664e-05, "loss": 5.7257, "step": 1578 }, { "epoch": 0.09801974051772301, "grad_norm": 0.9461802763554494, "learning_rate": 3.267121870473826e-05, "loss": 5.6428, "step": 1579 }, { "epoch": 0.09808181761748092, "grad_norm": 1.2526223808545172, "learning_rate": 3.269190978688186e-05, "loss": 5.6089, "step": 1580 }, { "epoch": 0.0981438947172388, "grad_norm": 0.7334083119627987, "learning_rate": 3.271260086902545e-05, "loss": 5.6228, "step": 1581 }, { "epoch": 0.09820597181699671, "grad_norm": 0.7673979491937399, "learning_rate": 3.273329195116905e-05, "loss": 5.6768, "step": 1582 }, { "epoch": 0.09826804891675461, "grad_norm": 0.8852074806087771, "learning_rate": 3.2753983033312646e-05, "loss": 5.5902, "step": 1583 }, { "epoch": 0.0983301260165125, "grad_norm": 0.7905483419776241, "learning_rate": 3.277467411545624e-05, "loss": 5.6521, "step": 1584 }, { "epoch": 0.09839220311627041, "grad_norm": 0.8066657110936308, "learning_rate": 3.2795365197599836e-05, "loss": 5.5318, "step": 1585 }, { "epoch": 0.09845428021602831, "grad_norm": 0.7677177321653629, "learning_rate": 3.281605627974344e-05, "loss": 5.6403, "step": 1586 }, { "epoch": 0.0985163573157862, "grad_norm": 0.8977279514929134, "learning_rate": 3.2836747361887026e-05, "loss": 5.6311, "step": 1587 }, { "epoch": 0.0985784344155441, "grad_norm": 0.72962695080887, "learning_rate": 3.285743844403063e-05, "loss": 5.6736, "step": 1588 }, { "epoch": 0.09864051151530201, "grad_norm": 0.7439701155305781, "learning_rate": 3.287812952617422e-05, "loss": 5.5538, "step": 1589 }, { "epoch": 0.0987025886150599, "grad_norm": 0.9166519134952625, "learning_rate": 3.289882060831782e-05, "loss": 5.6176, "step": 1590 }, { "epoch": 0.0987646657148178, "grad_norm": 1.0953868606727977, "learning_rate": 3.291951169046141e-05, "loss": 5.6596, "step": 1591 }, { "epoch": 0.09882674281457571, "grad_norm": 1.1077183456116682, "learning_rate": 3.294020277260501e-05, "loss": 5.569, "step": 1592 }, { "epoch": 0.0988888199143336, "grad_norm": 1.0419069521649256, "learning_rate": 3.29608938547486e-05, "loss": 5.6504, "step": 1593 }, { "epoch": 0.0989508970140915, "grad_norm": 1.155083196459239, "learning_rate": 3.29815849368922e-05, "loss": 5.6118, "step": 1594 }, { "epoch": 0.0990129741138494, "grad_norm": 0.8674965385126098, "learning_rate": 3.30022760190358e-05, "loss": 5.6149, "step": 1595 }, { "epoch": 0.0990750512136073, "grad_norm": 0.8487281877414407, "learning_rate": 3.302296710117939e-05, "loss": 5.6493, "step": 1596 }, { "epoch": 0.0991371283133652, "grad_norm": 1.0927841826891567, "learning_rate": 3.304365818332299e-05, "loss": 5.6324, "step": 1597 }, { "epoch": 0.0991992054131231, "grad_norm": 0.9424784715556402, "learning_rate": 3.3064349265466585e-05, "loss": 5.6535, "step": 1598 }, { "epoch": 0.099261282512881, "grad_norm": 0.8992889681791536, "learning_rate": 3.308504034761018e-05, "loss": 5.5812, "step": 1599 }, { "epoch": 0.0993233596126389, "grad_norm": 1.1955878744627846, "learning_rate": 3.3105731429753776e-05, "loss": 5.5941, "step": 1600 }, { "epoch": 0.0993854367123968, "grad_norm": 0.9480564552329188, "learning_rate": 3.312642251189738e-05, "loss": 5.6846, "step": 1601 }, { "epoch": 0.09944751381215469, "grad_norm": 0.8999271482546128, "learning_rate": 3.3147113594040966e-05, "loss": 5.7238, "step": 1602 }, { "epoch": 0.0995095909119126, "grad_norm": 0.9280230263841717, "learning_rate": 3.316780467618457e-05, "loss": 5.5827, "step": 1603 }, { "epoch": 0.0995716680116705, "grad_norm": 0.9121949770373668, "learning_rate": 3.318849575832816e-05, "loss": 5.5947, "step": 1604 }, { "epoch": 0.09963374511142839, "grad_norm": 0.8740462145687872, "learning_rate": 3.320918684047176e-05, "loss": 5.6156, "step": 1605 }, { "epoch": 0.0996958222111863, "grad_norm": 0.9731585879155227, "learning_rate": 3.322987792261535e-05, "loss": 5.5826, "step": 1606 }, { "epoch": 0.0997578993109442, "grad_norm": 1.0749944487140062, "learning_rate": 3.3250569004758955e-05, "loss": 5.6125, "step": 1607 }, { "epoch": 0.09981997641070209, "grad_norm": 0.8652438471084055, "learning_rate": 3.327126008690254e-05, "loss": 5.5864, "step": 1608 }, { "epoch": 0.09988205351045999, "grad_norm": 1.0098480250017097, "learning_rate": 3.3291951169046145e-05, "loss": 5.6182, "step": 1609 }, { "epoch": 0.0999441306102179, "grad_norm": 1.3283289584459146, "learning_rate": 3.331264225118974e-05, "loss": 5.5787, "step": 1610 }, { "epoch": 0.10000620770997579, "grad_norm": 1.0885729747206268, "learning_rate": 3.3333333333333335e-05, "loss": 5.6146, "step": 1611 }, { "epoch": 0.10006828480973369, "grad_norm": 0.9291497927503463, "learning_rate": 3.335402441547693e-05, "loss": 5.6065, "step": 1612 }, { "epoch": 0.1001303619094916, "grad_norm": 0.9316379379137743, "learning_rate": 3.337471549762053e-05, "loss": 5.5172, "step": 1613 }, { "epoch": 0.10019243900924948, "grad_norm": 1.0766973579882382, "learning_rate": 3.339540657976412e-05, "loss": 5.6376, "step": 1614 }, { "epoch": 0.10025451610900739, "grad_norm": 1.0026532379956428, "learning_rate": 3.341609766190772e-05, "loss": 5.6171, "step": 1615 }, { "epoch": 0.10031659320876529, "grad_norm": 1.0616374160793338, "learning_rate": 3.343678874405132e-05, "loss": 5.5716, "step": 1616 }, { "epoch": 0.10037867030852318, "grad_norm": 0.8001475362036933, "learning_rate": 3.345747982619491e-05, "loss": 5.561, "step": 1617 }, { "epoch": 0.10044074740828109, "grad_norm": 0.9161183313893989, "learning_rate": 3.347817090833851e-05, "loss": 5.5997, "step": 1618 }, { "epoch": 0.10050282450803899, "grad_norm": 0.7413662030385196, "learning_rate": 3.34988619904821e-05, "loss": 5.567, "step": 1619 }, { "epoch": 0.10056490160779688, "grad_norm": 0.8434057946480765, "learning_rate": 3.35195530726257e-05, "loss": 5.604, "step": 1620 }, { "epoch": 0.10062697870755478, "grad_norm": 0.8521282521527571, "learning_rate": 3.354024415476929e-05, "loss": 5.571, "step": 1621 }, { "epoch": 0.10068905580731269, "grad_norm": 0.7764882503143764, "learning_rate": 3.3560935236912894e-05, "loss": 5.6278, "step": 1622 }, { "epoch": 0.10075113290707058, "grad_norm": 0.8448097849516727, "learning_rate": 3.358162631905649e-05, "loss": 5.5723, "step": 1623 }, { "epoch": 0.10081321000682848, "grad_norm": 0.9005306904512361, "learning_rate": 3.3602317401200084e-05, "loss": 5.6872, "step": 1624 }, { "epoch": 0.10087528710658639, "grad_norm": 0.8506773101395899, "learning_rate": 3.362300848334368e-05, "loss": 5.509, "step": 1625 }, { "epoch": 0.10093736420634428, "grad_norm": 0.7293224213778796, "learning_rate": 3.3643699565487275e-05, "loss": 5.5435, "step": 1626 }, { "epoch": 0.10099944130610218, "grad_norm": 0.8764574650955947, "learning_rate": 3.366439064763087e-05, "loss": 5.5891, "step": 1627 }, { "epoch": 0.10106151840586008, "grad_norm": 1.03241479000182, "learning_rate": 3.368508172977447e-05, "loss": 5.5506, "step": 1628 }, { "epoch": 0.10112359550561797, "grad_norm": 1.0145448663039567, "learning_rate": 3.370577281191806e-05, "loss": 5.6514, "step": 1629 }, { "epoch": 0.10118567260537588, "grad_norm": 0.98961883615085, "learning_rate": 3.372646389406166e-05, "loss": 5.6145, "step": 1630 }, { "epoch": 0.10124774970513378, "grad_norm": 1.1178205070603413, "learning_rate": 3.374715497620526e-05, "loss": 5.5573, "step": 1631 }, { "epoch": 0.10130982680489167, "grad_norm": 0.9479742801191203, "learning_rate": 3.376784605834885e-05, "loss": 5.5508, "step": 1632 }, { "epoch": 0.10137190390464958, "grad_norm": 0.834508190882799, "learning_rate": 3.378853714049245e-05, "loss": 5.5226, "step": 1633 }, { "epoch": 0.10143398100440748, "grad_norm": 0.682397416827801, "learning_rate": 3.380922822263605e-05, "loss": 5.5464, "step": 1634 }, { "epoch": 0.10149605810416537, "grad_norm": 0.9717984074282192, "learning_rate": 3.382991930477964e-05, "loss": 5.4933, "step": 1635 }, { "epoch": 0.10155813520392327, "grad_norm": 1.0967918603585325, "learning_rate": 3.385061038692324e-05, "loss": 5.469, "step": 1636 }, { "epoch": 0.10162021230368118, "grad_norm": 1.1176634088300565, "learning_rate": 3.3871301469066834e-05, "loss": 5.5605, "step": 1637 }, { "epoch": 0.10168228940343907, "grad_norm": 1.0764827659988587, "learning_rate": 3.389199255121043e-05, "loss": 5.587, "step": 1638 }, { "epoch": 0.10174436650319697, "grad_norm": 0.9079392572830913, "learning_rate": 3.3912683633354024e-05, "loss": 5.5615, "step": 1639 }, { "epoch": 0.10180644360295486, "grad_norm": 1.0065228986603578, "learning_rate": 3.3933374715497626e-05, "loss": 5.5321, "step": 1640 }, { "epoch": 0.10186852070271277, "grad_norm": 0.8220538593801885, "learning_rate": 3.3954065797641214e-05, "loss": 5.5453, "step": 1641 }, { "epoch": 0.10193059780247067, "grad_norm": 0.7511467511383898, "learning_rate": 3.3974756879784816e-05, "loss": 5.5861, "step": 1642 }, { "epoch": 0.10199267490222856, "grad_norm": 0.7996298723524099, "learning_rate": 3.399544796192841e-05, "loss": 5.5924, "step": 1643 }, { "epoch": 0.10205475200198647, "grad_norm": 0.8165457939368979, "learning_rate": 3.4016139044072006e-05, "loss": 5.5462, "step": 1644 }, { "epoch": 0.10211682910174437, "grad_norm": 0.9991079464367026, "learning_rate": 3.40368301262156e-05, "loss": 5.5709, "step": 1645 }, { "epoch": 0.10217890620150226, "grad_norm": 0.9814860321251987, "learning_rate": 3.40575212083592e-05, "loss": 5.5733, "step": 1646 }, { "epoch": 0.10224098330126016, "grad_norm": 0.9905254229730341, "learning_rate": 3.407821229050279e-05, "loss": 5.5167, "step": 1647 }, { "epoch": 0.10230306040101807, "grad_norm": 0.9670001808410917, "learning_rate": 3.4098903372646393e-05, "loss": 5.484, "step": 1648 }, { "epoch": 0.10236513750077596, "grad_norm": 0.8507227475568064, "learning_rate": 3.411959445478999e-05, "loss": 5.5494, "step": 1649 }, { "epoch": 0.10242721460053386, "grad_norm": 1.0682433795897892, "learning_rate": 3.4140285536933584e-05, "loss": 5.5721, "step": 1650 }, { "epoch": 0.10248929170029177, "grad_norm": 1.1032117577171743, "learning_rate": 3.416097661907718e-05, "loss": 5.5518, "step": 1651 }, { "epoch": 0.10255136880004966, "grad_norm": 0.7400485084718146, "learning_rate": 3.4181667701220774e-05, "loss": 5.4706, "step": 1652 }, { "epoch": 0.10261344589980756, "grad_norm": 0.8683828071868175, "learning_rate": 3.420235878336437e-05, "loss": 5.6022, "step": 1653 }, { "epoch": 0.10267552299956546, "grad_norm": 1.11885916747346, "learning_rate": 3.4223049865507964e-05, "loss": 5.5976, "step": 1654 }, { "epoch": 0.10273760009932335, "grad_norm": 0.8237227262575149, "learning_rate": 3.4243740947651566e-05, "loss": 5.5693, "step": 1655 }, { "epoch": 0.10279967719908126, "grad_norm": 0.7876807474671574, "learning_rate": 3.426443202979516e-05, "loss": 5.5746, "step": 1656 }, { "epoch": 0.10286175429883916, "grad_norm": 0.910290851381433, "learning_rate": 3.4285123111938756e-05, "loss": 5.559, "step": 1657 }, { "epoch": 0.10292383139859705, "grad_norm": 0.7978288514209473, "learning_rate": 3.430581419408235e-05, "loss": 5.3872, "step": 1658 }, { "epoch": 0.10298590849835496, "grad_norm": 0.8120199399200483, "learning_rate": 3.432650527622595e-05, "loss": 5.5945, "step": 1659 }, { "epoch": 0.10304798559811286, "grad_norm": 0.9083598340102962, "learning_rate": 3.434719635836954e-05, "loss": 5.4831, "step": 1660 }, { "epoch": 0.10311006269787075, "grad_norm": 0.832779096037257, "learning_rate": 3.436788744051314e-05, "loss": 5.5735, "step": 1661 }, { "epoch": 0.10317213979762865, "grad_norm": 0.7260252904664207, "learning_rate": 3.438857852265674e-05, "loss": 5.5272, "step": 1662 }, { "epoch": 0.10323421689738656, "grad_norm": 0.8721551287885545, "learning_rate": 3.440926960480033e-05, "loss": 5.5138, "step": 1663 }, { "epoch": 0.10329629399714445, "grad_norm": 1.160428512941957, "learning_rate": 3.442996068694393e-05, "loss": 5.6151, "step": 1664 }, { "epoch": 0.10335837109690235, "grad_norm": 1.2319834272335108, "learning_rate": 3.445065176908753e-05, "loss": 5.5952, "step": 1665 }, { "epoch": 0.10342044819666026, "grad_norm": 1.077662493143005, "learning_rate": 3.447134285123112e-05, "loss": 5.6063, "step": 1666 }, { "epoch": 0.10348252529641815, "grad_norm": 1.3375374209436615, "learning_rate": 3.449203393337472e-05, "loss": 5.5425, "step": 1667 }, { "epoch": 0.10354460239617605, "grad_norm": 0.9379409071135393, "learning_rate": 3.4512725015518315e-05, "loss": 5.5443, "step": 1668 }, { "epoch": 0.10360667949593395, "grad_norm": 1.2669836216615407, "learning_rate": 3.453341609766191e-05, "loss": 5.5135, "step": 1669 }, { "epoch": 0.10366875659569184, "grad_norm": 1.0248601213238953, "learning_rate": 3.4554107179805505e-05, "loss": 5.5072, "step": 1670 }, { "epoch": 0.10373083369544975, "grad_norm": 1.170970567735686, "learning_rate": 3.457479826194911e-05, "loss": 5.6151, "step": 1671 }, { "epoch": 0.10379291079520765, "grad_norm": 0.955274097642844, "learning_rate": 3.4595489344092696e-05, "loss": 5.6393, "step": 1672 }, { "epoch": 0.10385498789496554, "grad_norm": 1.1770637082964845, "learning_rate": 3.46161804262363e-05, "loss": 5.5186, "step": 1673 }, { "epoch": 0.10391706499472345, "grad_norm": 0.9481038897282161, "learning_rate": 3.463687150837989e-05, "loss": 5.5079, "step": 1674 }, { "epoch": 0.10397914209448135, "grad_norm": 1.0215124997995395, "learning_rate": 3.465756259052349e-05, "loss": 5.4906, "step": 1675 }, { "epoch": 0.10404121919423924, "grad_norm": 0.974620536174584, "learning_rate": 3.467825367266708e-05, "loss": 5.4614, "step": 1676 }, { "epoch": 0.10410329629399714, "grad_norm": 0.9585967047536216, "learning_rate": 3.469894475481068e-05, "loss": 5.5335, "step": 1677 }, { "epoch": 0.10416537339375505, "grad_norm": 1.0869731767041861, "learning_rate": 3.471963583695427e-05, "loss": 5.4667, "step": 1678 }, { "epoch": 0.10422745049351294, "grad_norm": 0.9291443079380914, "learning_rate": 3.474032691909787e-05, "loss": 5.5384, "step": 1679 }, { "epoch": 0.10428952759327084, "grad_norm": 0.9198907089623428, "learning_rate": 3.476101800124147e-05, "loss": 5.505, "step": 1680 }, { "epoch": 0.10435160469302875, "grad_norm": 0.8298468054296965, "learning_rate": 3.478170908338506e-05, "loss": 5.542, "step": 1681 }, { "epoch": 0.10441368179278664, "grad_norm": 0.7786679897324326, "learning_rate": 3.480240016552866e-05, "loss": 5.4417, "step": 1682 }, { "epoch": 0.10447575889254454, "grad_norm": 0.9254465584619797, "learning_rate": 3.4823091247672255e-05, "loss": 5.532, "step": 1683 }, { "epoch": 0.10453783599230244, "grad_norm": 0.935475356647488, "learning_rate": 3.484378232981585e-05, "loss": 5.4949, "step": 1684 }, { "epoch": 0.10459991309206033, "grad_norm": 0.8967307755680411, "learning_rate": 3.4864473411959445e-05, "loss": 5.4766, "step": 1685 }, { "epoch": 0.10466199019181824, "grad_norm": 0.9542985476566229, "learning_rate": 3.488516449410305e-05, "loss": 5.4142, "step": 1686 }, { "epoch": 0.10472406729157614, "grad_norm": 1.0504923143173777, "learning_rate": 3.4905855576246635e-05, "loss": 5.4942, "step": 1687 }, { "epoch": 0.10478614439133403, "grad_norm": 1.1256986673788258, "learning_rate": 3.492654665839024e-05, "loss": 5.4926, "step": 1688 }, { "epoch": 0.10484822149109194, "grad_norm": 0.8393769142202903, "learning_rate": 3.494723774053383e-05, "loss": 5.5425, "step": 1689 }, { "epoch": 0.10491029859084984, "grad_norm": 0.8783751854223878, "learning_rate": 3.496792882267743e-05, "loss": 5.5999, "step": 1690 }, { "epoch": 0.10497237569060773, "grad_norm": 1.0070599252244943, "learning_rate": 3.498861990482102e-05, "loss": 5.5889, "step": 1691 }, { "epoch": 0.10503445279036563, "grad_norm": 1.0690971769871334, "learning_rate": 3.5009310986964624e-05, "loss": 5.4457, "step": 1692 }, { "epoch": 0.10509652989012354, "grad_norm": 0.9853074637640318, "learning_rate": 3.503000206910821e-05, "loss": 5.5239, "step": 1693 }, { "epoch": 0.10515860698988143, "grad_norm": 0.9650504010632059, "learning_rate": 3.5050693151251814e-05, "loss": 5.3704, "step": 1694 }, { "epoch": 0.10522068408963933, "grad_norm": 0.7317595475345846, "learning_rate": 3.507138423339541e-05, "loss": 5.5858, "step": 1695 }, { "epoch": 0.10528276118939724, "grad_norm": 0.9207984327949226, "learning_rate": 3.5092075315539005e-05, "loss": 5.5477, "step": 1696 }, { "epoch": 0.10534483828915513, "grad_norm": 1.0018656136590058, "learning_rate": 3.51127663976826e-05, "loss": 5.489, "step": 1697 }, { "epoch": 0.10540691538891303, "grad_norm": 1.0648336587764975, "learning_rate": 3.51334574798262e-05, "loss": 5.4874, "step": 1698 }, { "epoch": 0.10546899248867093, "grad_norm": 1.0836552300645161, "learning_rate": 3.515414856196979e-05, "loss": 5.5443, "step": 1699 }, { "epoch": 0.10553106958842882, "grad_norm": 0.9146246852230351, "learning_rate": 3.517483964411339e-05, "loss": 5.3855, "step": 1700 }, { "epoch": 0.10559314668818673, "grad_norm": 0.9242751671024422, "learning_rate": 3.519553072625699e-05, "loss": 5.4434, "step": 1701 }, { "epoch": 0.10565522378794463, "grad_norm": 0.9803591359178035, "learning_rate": 3.521622180840058e-05, "loss": 5.4196, "step": 1702 }, { "epoch": 0.10571730088770252, "grad_norm": 1.092032309927503, "learning_rate": 3.523691289054418e-05, "loss": 5.4177, "step": 1703 }, { "epoch": 0.10577937798746043, "grad_norm": 0.9025482672573011, "learning_rate": 3.525760397268777e-05, "loss": 5.4569, "step": 1704 }, { "epoch": 0.10584145508721833, "grad_norm": 0.9262261910924254, "learning_rate": 3.527829505483137e-05, "loss": 5.5826, "step": 1705 }, { "epoch": 0.10590353218697622, "grad_norm": 0.901878468169102, "learning_rate": 3.529898613697496e-05, "loss": 5.4975, "step": 1706 }, { "epoch": 0.10596560928673412, "grad_norm": 0.7738262317062393, "learning_rate": 3.5319677219118564e-05, "loss": 5.5086, "step": 1707 }, { "epoch": 0.10602768638649203, "grad_norm": 1.0291381945255926, "learning_rate": 3.534036830126215e-05, "loss": 5.5343, "step": 1708 }, { "epoch": 0.10608976348624992, "grad_norm": 1.0787842435031954, "learning_rate": 3.5361059383405754e-05, "loss": 5.5684, "step": 1709 }, { "epoch": 0.10615184058600782, "grad_norm": 0.9539689876785904, "learning_rate": 3.538175046554935e-05, "loss": 5.4991, "step": 1710 }, { "epoch": 0.10621391768576573, "grad_norm": 0.9365501996652782, "learning_rate": 3.5402441547692944e-05, "loss": 5.44, "step": 1711 }, { "epoch": 0.10627599478552362, "grad_norm": 0.9415056801359735, "learning_rate": 3.542313262983654e-05, "loss": 5.4348, "step": 1712 }, { "epoch": 0.10633807188528152, "grad_norm": 1.0391137820330882, "learning_rate": 3.544382371198014e-05, "loss": 5.5682, "step": 1713 }, { "epoch": 0.10640014898503942, "grad_norm": 1.0231937380230003, "learning_rate": 3.546451479412373e-05, "loss": 5.5429, "step": 1714 }, { "epoch": 0.10646222608479732, "grad_norm": 1.1963880100979685, "learning_rate": 3.548520587626733e-05, "loss": 5.4927, "step": 1715 }, { "epoch": 0.10652430318455522, "grad_norm": 1.0357101973560991, "learning_rate": 3.5505896958410926e-05, "loss": 5.5092, "step": 1716 }, { "epoch": 0.10658638028431312, "grad_norm": 1.1080508321435776, "learning_rate": 3.552658804055452e-05, "loss": 5.5118, "step": 1717 }, { "epoch": 0.10664845738407101, "grad_norm": 1.0568550724291683, "learning_rate": 3.5547279122698117e-05, "loss": 5.5647, "step": 1718 }, { "epoch": 0.10671053448382892, "grad_norm": 1.0409569929548679, "learning_rate": 3.556797020484172e-05, "loss": 5.5651, "step": 1719 }, { "epoch": 0.10677261158358682, "grad_norm": 0.982798206607485, "learning_rate": 3.558866128698531e-05, "loss": 5.5274, "step": 1720 }, { "epoch": 0.10683468868334471, "grad_norm": 0.8855573467616504, "learning_rate": 3.560935236912891e-05, "loss": 5.5358, "step": 1721 }, { "epoch": 0.10689676578310262, "grad_norm": 0.9717970580711042, "learning_rate": 3.5630043451272504e-05, "loss": 5.5047, "step": 1722 }, { "epoch": 0.10695884288286052, "grad_norm": 0.9261687550196552, "learning_rate": 3.56507345334161e-05, "loss": 5.415, "step": 1723 }, { "epoch": 0.10702091998261841, "grad_norm": 0.8872983945287107, "learning_rate": 3.5671425615559694e-05, "loss": 5.4686, "step": 1724 }, { "epoch": 0.10708299708237631, "grad_norm": 0.831855827070471, "learning_rate": 3.5692116697703296e-05, "loss": 5.3768, "step": 1725 }, { "epoch": 0.10714507418213422, "grad_norm": 0.7975819761422598, "learning_rate": 3.5712807779846884e-05, "loss": 5.4578, "step": 1726 }, { "epoch": 0.10720715128189211, "grad_norm": 0.7291108427931182, "learning_rate": 3.5733498861990486e-05, "loss": 5.5746, "step": 1727 }, { "epoch": 0.10726922838165001, "grad_norm": 0.8055140870115708, "learning_rate": 3.575418994413408e-05, "loss": 5.4972, "step": 1728 }, { "epoch": 0.10733130548140792, "grad_norm": 0.9819695889606224, "learning_rate": 3.5774881026277676e-05, "loss": 5.5473, "step": 1729 }, { "epoch": 0.1073933825811658, "grad_norm": 1.0032234096087955, "learning_rate": 3.579557210842127e-05, "loss": 5.4764, "step": 1730 }, { "epoch": 0.10745545968092371, "grad_norm": 0.9892767012846099, "learning_rate": 3.5816263190564866e-05, "loss": 5.447, "step": 1731 }, { "epoch": 0.10751753678068161, "grad_norm": 1.045669316859277, "learning_rate": 3.583695427270846e-05, "loss": 5.55, "step": 1732 }, { "epoch": 0.1075796138804395, "grad_norm": 1.0576293606467027, "learning_rate": 3.585764535485206e-05, "loss": 5.4982, "step": 1733 }, { "epoch": 0.10764169098019741, "grad_norm": 0.9638392976785979, "learning_rate": 3.587833643699566e-05, "loss": 5.4162, "step": 1734 }, { "epoch": 0.1077037680799553, "grad_norm": 1.1902753121197056, "learning_rate": 3.589902751913925e-05, "loss": 5.4686, "step": 1735 }, { "epoch": 0.1077658451797132, "grad_norm": 1.0935565232406117, "learning_rate": 3.591971860128285e-05, "loss": 5.4548, "step": 1736 }, { "epoch": 0.1078279222794711, "grad_norm": 0.7932189020672871, "learning_rate": 3.594040968342644e-05, "loss": 5.4465, "step": 1737 }, { "epoch": 0.107889999379229, "grad_norm": 0.835180171270687, "learning_rate": 3.596110076557004e-05, "loss": 5.5861, "step": 1738 }, { "epoch": 0.1079520764789869, "grad_norm": 0.8012184996222709, "learning_rate": 3.5981791847713633e-05, "loss": 5.5069, "step": 1739 }, { "epoch": 0.1080141535787448, "grad_norm": 0.9290169613812207, "learning_rate": 3.6002482929857235e-05, "loss": 5.4625, "step": 1740 }, { "epoch": 0.1080762306785027, "grad_norm": 1.0268799057400446, "learning_rate": 3.602317401200083e-05, "loss": 5.4831, "step": 1741 }, { "epoch": 0.1081383077782606, "grad_norm": 1.108987713762063, "learning_rate": 3.6043865094144425e-05, "loss": 5.4439, "step": 1742 }, { "epoch": 0.1082003848780185, "grad_norm": 0.9744079014653516, "learning_rate": 3.606455617628802e-05, "loss": 5.419, "step": 1743 }, { "epoch": 0.10826246197777639, "grad_norm": 0.8646397667969891, "learning_rate": 3.608524725843162e-05, "loss": 5.4015, "step": 1744 }, { "epoch": 0.1083245390775343, "grad_norm": 0.8285610202725162, "learning_rate": 3.610593834057521e-05, "loss": 5.378, "step": 1745 }, { "epoch": 0.1083866161772922, "grad_norm": 0.8293626282149382, "learning_rate": 3.612662942271881e-05, "loss": 5.5137, "step": 1746 }, { "epoch": 0.10844869327705009, "grad_norm": 0.8317388459522049, "learning_rate": 3.614732050486241e-05, "loss": 5.4313, "step": 1747 }, { "epoch": 0.108510770376808, "grad_norm": 0.9145335100603227, "learning_rate": 3.6168011587006e-05, "loss": 5.5343, "step": 1748 }, { "epoch": 0.1085728474765659, "grad_norm": 0.7529913138614303, "learning_rate": 3.61887026691496e-05, "loss": 5.4198, "step": 1749 }, { "epoch": 0.10863492457632379, "grad_norm": 0.8457499572614072, "learning_rate": 3.62093937512932e-05, "loss": 5.3626, "step": 1750 }, { "epoch": 0.10869700167608169, "grad_norm": 1.0163494167521192, "learning_rate": 3.623008483343679e-05, "loss": 5.3764, "step": 1751 }, { "epoch": 0.1087590787758396, "grad_norm": 0.9612713365704305, "learning_rate": 3.625077591558039e-05, "loss": 5.4795, "step": 1752 }, { "epoch": 0.10882115587559749, "grad_norm": 0.9322939007837105, "learning_rate": 3.6271466997723985e-05, "loss": 5.4175, "step": 1753 }, { "epoch": 0.10888323297535539, "grad_norm": 0.9912161089381721, "learning_rate": 3.629215807986758e-05, "loss": 5.4991, "step": 1754 }, { "epoch": 0.1089453100751133, "grad_norm": 0.8730828194611403, "learning_rate": 3.6312849162011175e-05, "loss": 5.4629, "step": 1755 }, { "epoch": 0.10900738717487118, "grad_norm": 0.8019737110942512, "learning_rate": 3.633354024415477e-05, "loss": 5.3714, "step": 1756 }, { "epoch": 0.10906946427462909, "grad_norm": 0.8606420020236508, "learning_rate": 3.6354231326298365e-05, "loss": 5.5878, "step": 1757 }, { "epoch": 0.10913154137438699, "grad_norm": 0.9967193733161954, "learning_rate": 3.637492240844197e-05, "loss": 5.5149, "step": 1758 }, { "epoch": 0.10919361847414488, "grad_norm": 1.1214500685081443, "learning_rate": 3.639561349058556e-05, "loss": 5.5603, "step": 1759 }, { "epoch": 0.10925569557390279, "grad_norm": 0.9446180961571642, "learning_rate": 3.641630457272916e-05, "loss": 5.2773, "step": 1760 }, { "epoch": 0.10931777267366069, "grad_norm": 1.1067545392202074, "learning_rate": 3.643699565487275e-05, "loss": 5.4134, "step": 1761 }, { "epoch": 0.10937984977341858, "grad_norm": 0.8901363169825616, "learning_rate": 3.645768673701635e-05, "loss": 5.4344, "step": 1762 }, { "epoch": 0.10944192687317648, "grad_norm": 1.0253022655382038, "learning_rate": 3.647837781915994e-05, "loss": 5.4621, "step": 1763 }, { "epoch": 0.10950400397293439, "grad_norm": 0.8861763929783087, "learning_rate": 3.649906890130354e-05, "loss": 5.5171, "step": 1764 }, { "epoch": 0.10956608107269228, "grad_norm": 0.9554290750770535, "learning_rate": 3.651975998344714e-05, "loss": 5.4488, "step": 1765 }, { "epoch": 0.10962815817245018, "grad_norm": 0.9668399685589967, "learning_rate": 3.654045106559073e-05, "loss": 5.5251, "step": 1766 }, { "epoch": 0.10969023527220809, "grad_norm": 0.8509792815509298, "learning_rate": 3.656114214773433e-05, "loss": 5.4245, "step": 1767 }, { "epoch": 0.10975231237196598, "grad_norm": 0.8800507278343144, "learning_rate": 3.6581833229877925e-05, "loss": 5.3602, "step": 1768 }, { "epoch": 0.10981438947172388, "grad_norm": 1.103323806700438, "learning_rate": 3.660252431202152e-05, "loss": 5.5352, "step": 1769 }, { "epoch": 0.10987646657148178, "grad_norm": 1.0786601410854257, "learning_rate": 3.6623215394165115e-05, "loss": 5.406, "step": 1770 }, { "epoch": 0.10993854367123967, "grad_norm": 1.0263947139282759, "learning_rate": 3.6643906476308717e-05, "loss": 5.4741, "step": 1771 }, { "epoch": 0.11000062077099758, "grad_norm": 0.6635212206921779, "learning_rate": 3.6664597558452305e-05, "loss": 5.4914, "step": 1772 }, { "epoch": 0.11006269787075548, "grad_norm": 0.9113367635396805, "learning_rate": 3.668528864059591e-05, "loss": 5.4626, "step": 1773 }, { "epoch": 0.11012477497051337, "grad_norm": 1.0387909613289894, "learning_rate": 3.67059797227395e-05, "loss": 5.3992, "step": 1774 }, { "epoch": 0.11018685207027128, "grad_norm": 1.0268139651659172, "learning_rate": 3.67266708048831e-05, "loss": 5.3896, "step": 1775 }, { "epoch": 0.11024892917002918, "grad_norm": 0.9160157551389166, "learning_rate": 3.674736188702669e-05, "loss": 5.5457, "step": 1776 }, { "epoch": 0.11031100626978707, "grad_norm": 1.0670198345146258, "learning_rate": 3.6768052969170294e-05, "loss": 5.5173, "step": 1777 }, { "epoch": 0.11037308336954497, "grad_norm": 1.000108236177377, "learning_rate": 3.678874405131388e-05, "loss": 5.4546, "step": 1778 }, { "epoch": 0.11043516046930288, "grad_norm": 1.0945216590303213, "learning_rate": 3.6809435133457484e-05, "loss": 5.3539, "step": 1779 }, { "epoch": 0.11049723756906077, "grad_norm": 0.9252418201974472, "learning_rate": 3.683012621560108e-05, "loss": 5.3196, "step": 1780 }, { "epoch": 0.11055931466881867, "grad_norm": 0.9875548938994028, "learning_rate": 3.6850817297744674e-05, "loss": 5.4366, "step": 1781 }, { "epoch": 0.11062139176857658, "grad_norm": 1.098328951034332, "learning_rate": 3.687150837988827e-05, "loss": 5.4709, "step": 1782 }, { "epoch": 0.11068346886833447, "grad_norm": 0.9113352265553412, "learning_rate": 3.689219946203187e-05, "loss": 5.3817, "step": 1783 }, { "epoch": 0.11074554596809237, "grad_norm": 0.7794660418089117, "learning_rate": 3.691289054417546e-05, "loss": 5.4101, "step": 1784 }, { "epoch": 0.11080762306785027, "grad_norm": 0.7499859331719159, "learning_rate": 3.693358162631906e-05, "loss": 5.4534, "step": 1785 }, { "epoch": 0.11086970016760817, "grad_norm": 0.7909292668426902, "learning_rate": 3.6954272708462656e-05, "loss": 5.4424, "step": 1786 }, { "epoch": 0.11093177726736607, "grad_norm": 0.9601596071760369, "learning_rate": 3.697496379060625e-05, "loss": 5.4808, "step": 1787 }, { "epoch": 0.11099385436712397, "grad_norm": 1.015832557010328, "learning_rate": 3.6995654872749846e-05, "loss": 5.4529, "step": 1788 }, { "epoch": 0.11105593146688186, "grad_norm": 0.8865583501590261, "learning_rate": 3.701634595489344e-05, "loss": 5.4548, "step": 1789 }, { "epoch": 0.11111800856663977, "grad_norm": 0.9307000552539639, "learning_rate": 3.7037037037037037e-05, "loss": 5.474, "step": 1790 }, { "epoch": 0.11118008566639767, "grad_norm": 1.0107999775900496, "learning_rate": 3.705772811918063e-05, "loss": 5.4108, "step": 1791 }, { "epoch": 0.11124216276615556, "grad_norm": 1.0628874577170515, "learning_rate": 3.7078419201324233e-05, "loss": 5.4173, "step": 1792 }, { "epoch": 0.11130423986591347, "grad_norm": 0.961077207008346, "learning_rate": 3.709911028346782e-05, "loss": 5.4218, "step": 1793 }, { "epoch": 0.11136631696567137, "grad_norm": 0.7546863374246998, "learning_rate": 3.7119801365611424e-05, "loss": 5.4025, "step": 1794 }, { "epoch": 0.11142839406542926, "grad_norm": 0.9529749017314559, "learning_rate": 3.714049244775502e-05, "loss": 5.3792, "step": 1795 }, { "epoch": 0.11149047116518716, "grad_norm": 0.9783781644166377, "learning_rate": 3.7161183529898614e-05, "loss": 5.3698, "step": 1796 }, { "epoch": 0.11155254826494507, "grad_norm": 0.8885903457300032, "learning_rate": 3.718187461204221e-05, "loss": 5.4267, "step": 1797 }, { "epoch": 0.11161462536470296, "grad_norm": 0.9667489396359015, "learning_rate": 3.720256569418581e-05, "loss": 5.415, "step": 1798 }, { "epoch": 0.11167670246446086, "grad_norm": 1.0659214356978088, "learning_rate": 3.72232567763294e-05, "loss": 5.4647, "step": 1799 }, { "epoch": 0.11173877956421877, "grad_norm": 1.0196397511709558, "learning_rate": 3.7243947858473e-05, "loss": 5.4319, "step": 1800 }, { "epoch": 0.11180085666397666, "grad_norm": 0.8592987026677593, "learning_rate": 3.7264638940616596e-05, "loss": 5.2714, "step": 1801 }, { "epoch": 0.11186293376373456, "grad_norm": 0.7690072768875307, "learning_rate": 3.728533002276019e-05, "loss": 5.4787, "step": 1802 }, { "epoch": 0.11192501086349246, "grad_norm": 0.9172199522967573, "learning_rate": 3.7306021104903786e-05, "loss": 5.3836, "step": 1803 }, { "epoch": 0.11198708796325035, "grad_norm": 1.0433300978231586, "learning_rate": 3.732671218704739e-05, "loss": 5.4787, "step": 1804 }, { "epoch": 0.11204916506300826, "grad_norm": 1.0203975344123795, "learning_rate": 3.7347403269190976e-05, "loss": 5.513, "step": 1805 }, { "epoch": 0.11211124216276616, "grad_norm": 0.8296528882639596, "learning_rate": 3.736809435133458e-05, "loss": 5.3809, "step": 1806 }, { "epoch": 0.11217331926252405, "grad_norm": 0.9111785386561495, "learning_rate": 3.738878543347817e-05, "loss": 5.4214, "step": 1807 }, { "epoch": 0.11223539636228196, "grad_norm": 0.9353679573500511, "learning_rate": 3.740947651562177e-05, "loss": 5.33, "step": 1808 }, { "epoch": 0.11229747346203986, "grad_norm": 0.7695239691220441, "learning_rate": 3.743016759776536e-05, "loss": 5.4726, "step": 1809 }, { "epoch": 0.11235955056179775, "grad_norm": 0.8660910264271225, "learning_rate": 3.7450858679908965e-05, "loss": 5.4983, "step": 1810 }, { "epoch": 0.11242162766155565, "grad_norm": 0.7604050482990218, "learning_rate": 3.7471549762052553e-05, "loss": 5.3917, "step": 1811 }, { "epoch": 0.11248370476131356, "grad_norm": 0.8839917029643033, "learning_rate": 3.7492240844196155e-05, "loss": 5.482, "step": 1812 }, { "epoch": 0.11254578186107145, "grad_norm": 0.9830453460397709, "learning_rate": 3.751293192633975e-05, "loss": 5.3836, "step": 1813 }, { "epoch": 0.11260785896082935, "grad_norm": 1.072544693893674, "learning_rate": 3.7533623008483345e-05, "loss": 5.32, "step": 1814 }, { "epoch": 0.11266993606058726, "grad_norm": 0.8280078157249552, "learning_rate": 3.755431409062694e-05, "loss": 5.3858, "step": 1815 }, { "epoch": 0.11273201316034515, "grad_norm": 0.908720627878061, "learning_rate": 3.7575005172770536e-05, "loss": 5.2585, "step": 1816 }, { "epoch": 0.11279409026010305, "grad_norm": 0.9387884822429863, "learning_rate": 3.759569625491413e-05, "loss": 5.3991, "step": 1817 }, { "epoch": 0.11285616735986095, "grad_norm": 0.881273700855327, "learning_rate": 3.7616387337057726e-05, "loss": 5.4242, "step": 1818 }, { "epoch": 0.11291824445961884, "grad_norm": 0.9103722704807581, "learning_rate": 3.763707841920133e-05, "loss": 5.3649, "step": 1819 }, { "epoch": 0.11298032155937675, "grad_norm": 0.939616389373304, "learning_rate": 3.765776950134492e-05, "loss": 5.3012, "step": 1820 }, { "epoch": 0.11304239865913465, "grad_norm": 0.9723772315173688, "learning_rate": 3.767846058348852e-05, "loss": 5.4128, "step": 1821 }, { "epoch": 0.11310447575889254, "grad_norm": 1.0352996726814576, "learning_rate": 3.769915166563211e-05, "loss": 5.3966, "step": 1822 }, { "epoch": 0.11316655285865045, "grad_norm": 1.0729389126641078, "learning_rate": 3.7719842747775715e-05, "loss": 5.3866, "step": 1823 }, { "epoch": 0.11322862995840835, "grad_norm": 0.9333056719688647, "learning_rate": 3.77405338299193e-05, "loss": 5.346, "step": 1824 }, { "epoch": 0.11329070705816624, "grad_norm": 0.8857372414178503, "learning_rate": 3.7761224912062905e-05, "loss": 5.4329, "step": 1825 }, { "epoch": 0.11335278415792414, "grad_norm": 0.9436753338209486, "learning_rate": 3.77819159942065e-05, "loss": 5.3367, "step": 1826 }, { "epoch": 0.11341486125768203, "grad_norm": 0.9684059701213982, "learning_rate": 3.7802607076350095e-05, "loss": 5.4425, "step": 1827 }, { "epoch": 0.11347693835743994, "grad_norm": 0.9116083439495735, "learning_rate": 3.782329815849369e-05, "loss": 5.363, "step": 1828 }, { "epoch": 0.11353901545719784, "grad_norm": 0.9018796247335966, "learning_rate": 3.784398924063729e-05, "loss": 5.4727, "step": 1829 }, { "epoch": 0.11360109255695573, "grad_norm": 0.9195771756043073, "learning_rate": 3.786468032278088e-05, "loss": 5.3655, "step": 1830 }, { "epoch": 0.11366316965671364, "grad_norm": 0.9412608687341963, "learning_rate": 3.788537140492448e-05, "loss": 5.358, "step": 1831 }, { "epoch": 0.11372524675647154, "grad_norm": 1.020575971015603, "learning_rate": 3.790606248706808e-05, "loss": 5.3078, "step": 1832 }, { "epoch": 0.11378732385622943, "grad_norm": 0.9326602945800857, "learning_rate": 3.792675356921167e-05, "loss": 5.4403, "step": 1833 }, { "epoch": 0.11384940095598733, "grad_norm": 0.8708749333777634, "learning_rate": 3.794744465135527e-05, "loss": 5.4189, "step": 1834 }, { "epoch": 0.11391147805574524, "grad_norm": 1.0522923718252912, "learning_rate": 3.796813573349887e-05, "loss": 5.3612, "step": 1835 }, { "epoch": 0.11397355515550313, "grad_norm": 1.1164060941019343, "learning_rate": 3.798882681564246e-05, "loss": 5.3838, "step": 1836 }, { "epoch": 0.11403563225526103, "grad_norm": 1.010333316681737, "learning_rate": 3.800951789778606e-05, "loss": 5.3571, "step": 1837 }, { "epoch": 0.11409770935501894, "grad_norm": 1.1486579008821436, "learning_rate": 3.8030208979929654e-05, "loss": 5.4129, "step": 1838 }, { "epoch": 0.11415978645477683, "grad_norm": 0.8381189777632967, "learning_rate": 3.805090006207325e-05, "loss": 5.3632, "step": 1839 }, { "epoch": 0.11422186355453473, "grad_norm": 0.7474504289169964, "learning_rate": 3.8071591144216845e-05, "loss": 5.411, "step": 1840 }, { "epoch": 0.11428394065429263, "grad_norm": 0.7050538626852665, "learning_rate": 3.809228222636044e-05, "loss": 5.3372, "step": 1841 }, { "epoch": 0.11434601775405052, "grad_norm": 0.767735899417816, "learning_rate": 3.8112973308504035e-05, "loss": 5.4907, "step": 1842 }, { "epoch": 0.11440809485380843, "grad_norm": 0.8926784867972961, "learning_rate": 3.813366439064763e-05, "loss": 5.4481, "step": 1843 }, { "epoch": 0.11447017195356633, "grad_norm": 1.1698165107369294, "learning_rate": 3.815435547279123e-05, "loss": 5.4155, "step": 1844 }, { "epoch": 0.11453224905332422, "grad_norm": 0.8750518034177768, "learning_rate": 3.817504655493483e-05, "loss": 5.3124, "step": 1845 }, { "epoch": 0.11459432615308213, "grad_norm": 0.7770590716163411, "learning_rate": 3.819573763707842e-05, "loss": 5.3668, "step": 1846 }, { "epoch": 0.11465640325284003, "grad_norm": 0.9332635846358949, "learning_rate": 3.821642871922202e-05, "loss": 5.3441, "step": 1847 }, { "epoch": 0.11471848035259792, "grad_norm": 1.0156253521258405, "learning_rate": 3.823711980136561e-05, "loss": 5.3473, "step": 1848 }, { "epoch": 0.11478055745235582, "grad_norm": 0.856724066264724, "learning_rate": 3.825781088350921e-05, "loss": 5.2962, "step": 1849 }, { "epoch": 0.11484263455211373, "grad_norm": 0.8292413420732669, "learning_rate": 3.827850196565281e-05, "loss": 5.2839, "step": 1850 }, { "epoch": 0.11490471165187162, "grad_norm": 0.8619774686829283, "learning_rate": 3.82991930477964e-05, "loss": 5.3199, "step": 1851 }, { "epoch": 0.11496678875162952, "grad_norm": 0.810686398183162, "learning_rate": 3.831988412994e-05, "loss": 5.3664, "step": 1852 }, { "epoch": 0.11502886585138743, "grad_norm": 0.9753625440240958, "learning_rate": 3.8340575212083594e-05, "loss": 5.4439, "step": 1853 }, { "epoch": 0.11509094295114532, "grad_norm": 1.0664586763455306, "learning_rate": 3.836126629422719e-05, "loss": 5.2867, "step": 1854 }, { "epoch": 0.11515302005090322, "grad_norm": 0.8361437177760027, "learning_rate": 3.8381957376370784e-05, "loss": 5.2995, "step": 1855 }, { "epoch": 0.11521509715066112, "grad_norm": 0.9523244528814264, "learning_rate": 3.8402648458514386e-05, "loss": 5.2006, "step": 1856 }, { "epoch": 0.11527717425041902, "grad_norm": 1.0590747711650363, "learning_rate": 3.8423339540657974e-05, "loss": 5.3058, "step": 1857 }, { "epoch": 0.11533925135017692, "grad_norm": 0.7989466050011241, "learning_rate": 3.8444030622801576e-05, "loss": 5.3078, "step": 1858 }, { "epoch": 0.11540132844993482, "grad_norm": 0.835244720597582, "learning_rate": 3.846472170494517e-05, "loss": 5.3683, "step": 1859 }, { "epoch": 0.11546340554969271, "grad_norm": 0.8881207108847062, "learning_rate": 3.8485412787088766e-05, "loss": 5.4174, "step": 1860 }, { "epoch": 0.11552548264945062, "grad_norm": 1.0213030623211243, "learning_rate": 3.850610386923236e-05, "loss": 5.4667, "step": 1861 }, { "epoch": 0.11558755974920852, "grad_norm": 0.9103242776346514, "learning_rate": 3.852679495137596e-05, "loss": 5.3897, "step": 1862 }, { "epoch": 0.11564963684896641, "grad_norm": 0.7951060463694746, "learning_rate": 3.854748603351955e-05, "loss": 5.4175, "step": 1863 }, { "epoch": 0.11571171394872432, "grad_norm": 1.0284875804353406, "learning_rate": 3.8568177115663154e-05, "loss": 5.4282, "step": 1864 }, { "epoch": 0.11577379104848222, "grad_norm": 1.1743820838247634, "learning_rate": 3.858886819780675e-05, "loss": 5.4279, "step": 1865 }, { "epoch": 0.11583586814824011, "grad_norm": 0.9208086603118666, "learning_rate": 3.8609559279950344e-05, "loss": 5.395, "step": 1866 }, { "epoch": 0.11589794524799801, "grad_norm": 0.9343267779202535, "learning_rate": 3.863025036209394e-05, "loss": 5.3587, "step": 1867 }, { "epoch": 0.11596002234775592, "grad_norm": 0.9377975945208177, "learning_rate": 3.865094144423754e-05, "loss": 5.4583, "step": 1868 }, { "epoch": 0.11602209944751381, "grad_norm": 0.8887128883371591, "learning_rate": 3.867163252638113e-05, "loss": 5.3972, "step": 1869 }, { "epoch": 0.11608417654727171, "grad_norm": 0.8210679708076288, "learning_rate": 3.869232360852473e-05, "loss": 5.4044, "step": 1870 }, { "epoch": 0.11614625364702962, "grad_norm": 0.858354204967272, "learning_rate": 3.8713014690668326e-05, "loss": 5.3464, "step": 1871 }, { "epoch": 0.1162083307467875, "grad_norm": 0.9423695933941704, "learning_rate": 3.873370577281192e-05, "loss": 5.307, "step": 1872 }, { "epoch": 0.11627040784654541, "grad_norm": 0.9847725564969202, "learning_rate": 3.8754396854955516e-05, "loss": 5.3901, "step": 1873 }, { "epoch": 0.11633248494630331, "grad_norm": 1.1129048028901756, "learning_rate": 3.877508793709911e-05, "loss": 5.3107, "step": 1874 }, { "epoch": 0.1163945620460612, "grad_norm": 1.0835102866000916, "learning_rate": 3.8795779019242706e-05, "loss": 5.3163, "step": 1875 }, { "epoch": 0.11645663914581911, "grad_norm": 0.95742285365378, "learning_rate": 3.88164701013863e-05, "loss": 5.3822, "step": 1876 }, { "epoch": 0.11651871624557701, "grad_norm": 1.0200300494610877, "learning_rate": 3.88371611835299e-05, "loss": 5.2882, "step": 1877 }, { "epoch": 0.1165807933453349, "grad_norm": 1.0672713463212937, "learning_rate": 3.885785226567349e-05, "loss": 5.3735, "step": 1878 }, { "epoch": 0.1166428704450928, "grad_norm": 0.9162771669031892, "learning_rate": 3.887854334781709e-05, "loss": 5.2966, "step": 1879 }, { "epoch": 0.11670494754485071, "grad_norm": 0.8940070336212381, "learning_rate": 3.889923442996069e-05, "loss": 5.3995, "step": 1880 }, { "epoch": 0.1167670246446086, "grad_norm": 0.993616651789102, "learning_rate": 3.891992551210428e-05, "loss": 5.3438, "step": 1881 }, { "epoch": 0.1168291017443665, "grad_norm": 1.080165866371411, "learning_rate": 3.894061659424788e-05, "loss": 5.2482, "step": 1882 }, { "epoch": 0.11689117884412441, "grad_norm": 0.8937446434020532, "learning_rate": 3.896130767639148e-05, "loss": 5.3963, "step": 1883 }, { "epoch": 0.1169532559438823, "grad_norm": 0.9003175387229922, "learning_rate": 3.898199875853507e-05, "loss": 5.3271, "step": 1884 }, { "epoch": 0.1170153330436402, "grad_norm": 0.8876030499444163, "learning_rate": 3.900268984067867e-05, "loss": 5.349, "step": 1885 }, { "epoch": 0.1170774101433981, "grad_norm": 0.8641374579643096, "learning_rate": 3.9023380922822266e-05, "loss": 5.2432, "step": 1886 }, { "epoch": 0.117139487243156, "grad_norm": 0.9888157360442771, "learning_rate": 3.904407200496586e-05, "loss": 5.3156, "step": 1887 }, { "epoch": 0.1172015643429139, "grad_norm": 1.0035766179135333, "learning_rate": 3.9064763087109456e-05, "loss": 5.2995, "step": 1888 }, { "epoch": 0.1172636414426718, "grad_norm": 1.1071389995894492, "learning_rate": 3.908545416925306e-05, "loss": 5.295, "step": 1889 }, { "epoch": 0.1173257185424297, "grad_norm": 1.1690663057585, "learning_rate": 3.9106145251396646e-05, "loss": 5.3063, "step": 1890 }, { "epoch": 0.1173877956421876, "grad_norm": 0.7764766208128322, "learning_rate": 3.912683633354025e-05, "loss": 5.3699, "step": 1891 }, { "epoch": 0.1174498727419455, "grad_norm": 0.8599747385706247, "learning_rate": 3.914752741568384e-05, "loss": 5.2385, "step": 1892 }, { "epoch": 0.11751194984170339, "grad_norm": 0.9952963535789736, "learning_rate": 3.916821849782744e-05, "loss": 5.28, "step": 1893 }, { "epoch": 0.1175740269414613, "grad_norm": 0.8548672098361795, "learning_rate": 3.918890957997103e-05, "loss": 5.3381, "step": 1894 }, { "epoch": 0.1176361040412192, "grad_norm": 0.9398630243292598, "learning_rate": 3.9209600662114635e-05, "loss": 5.4255, "step": 1895 }, { "epoch": 0.11769818114097709, "grad_norm": 1.013402473063187, "learning_rate": 3.923029174425822e-05, "loss": 5.3373, "step": 1896 }, { "epoch": 0.117760258240735, "grad_norm": 1.0338188312683272, "learning_rate": 3.9250982826401825e-05, "loss": 5.3164, "step": 1897 }, { "epoch": 0.1178223353404929, "grad_norm": 1.0828823777147494, "learning_rate": 3.927167390854542e-05, "loss": 5.4156, "step": 1898 }, { "epoch": 0.11788441244025079, "grad_norm": 0.9894752978864927, "learning_rate": 3.9292364990689015e-05, "loss": 5.2162, "step": 1899 }, { "epoch": 0.11794648954000869, "grad_norm": 0.9302936789620906, "learning_rate": 3.931305607283261e-05, "loss": 5.3645, "step": 1900 }, { "epoch": 0.1180085666397666, "grad_norm": 0.8520482628902611, "learning_rate": 3.9333747154976205e-05, "loss": 5.2902, "step": 1901 }, { "epoch": 0.11807064373952449, "grad_norm": 0.9979899589897304, "learning_rate": 3.93544382371198e-05, "loss": 5.3857, "step": 1902 }, { "epoch": 0.11813272083928239, "grad_norm": 1.0892323518855116, "learning_rate": 3.9375129319263395e-05, "loss": 5.3278, "step": 1903 }, { "epoch": 0.1181947979390403, "grad_norm": 0.8544724467585788, "learning_rate": 3.9395820401407e-05, "loss": 5.3271, "step": 1904 }, { "epoch": 0.11825687503879818, "grad_norm": 0.8612347333797952, "learning_rate": 3.9416511483550586e-05, "loss": 5.2624, "step": 1905 }, { "epoch": 0.11831895213855609, "grad_norm": 0.8049882493435018, "learning_rate": 3.943720256569419e-05, "loss": 5.3648, "step": 1906 }, { "epoch": 0.11838102923831399, "grad_norm": 0.9221053724647277, "learning_rate": 3.945789364783778e-05, "loss": 5.2995, "step": 1907 }, { "epoch": 0.11844310633807188, "grad_norm": 1.0170246163786447, "learning_rate": 3.9478584729981384e-05, "loss": 5.3363, "step": 1908 }, { "epoch": 0.11850518343782979, "grad_norm": 0.9571475834109614, "learning_rate": 3.949927581212497e-05, "loss": 5.4634, "step": 1909 }, { "epoch": 0.11856726053758769, "grad_norm": 1.0176222310236482, "learning_rate": 3.9519966894268574e-05, "loss": 5.3466, "step": 1910 }, { "epoch": 0.11862933763734558, "grad_norm": 0.8847836793653239, "learning_rate": 3.954065797641217e-05, "loss": 5.314, "step": 1911 }, { "epoch": 0.11869141473710348, "grad_norm": 0.7640305869335864, "learning_rate": 3.9561349058555765e-05, "loss": 5.319, "step": 1912 }, { "epoch": 0.11875349183686139, "grad_norm": 0.8258187912534103, "learning_rate": 3.958204014069936e-05, "loss": 5.3468, "step": 1913 }, { "epoch": 0.11881556893661928, "grad_norm": 0.8934478862701797, "learning_rate": 3.960273122284296e-05, "loss": 5.2801, "step": 1914 }, { "epoch": 0.11887764603637718, "grad_norm": 0.9213214440767863, "learning_rate": 3.962342230498655e-05, "loss": 5.3537, "step": 1915 }, { "epoch": 0.11893972313613509, "grad_norm": 0.9295538157027471, "learning_rate": 3.964411338713015e-05, "loss": 5.3243, "step": 1916 }, { "epoch": 0.11900180023589298, "grad_norm": 0.9575686444608946, "learning_rate": 3.966480446927375e-05, "loss": 5.3007, "step": 1917 }, { "epoch": 0.11906387733565088, "grad_norm": 0.9510625506166218, "learning_rate": 3.968549555141734e-05, "loss": 5.3286, "step": 1918 }, { "epoch": 0.11912595443540878, "grad_norm": 1.0097467948294137, "learning_rate": 3.970618663356094e-05, "loss": 5.3044, "step": 1919 }, { "epoch": 0.11918803153516667, "grad_norm": 1.0742229114798654, "learning_rate": 3.972687771570454e-05, "loss": 5.2879, "step": 1920 }, { "epoch": 0.11925010863492458, "grad_norm": 0.9748387191111927, "learning_rate": 3.974756879784813e-05, "loss": 5.2603, "step": 1921 }, { "epoch": 0.11931218573468247, "grad_norm": 1.074080468727554, "learning_rate": 3.976825987999173e-05, "loss": 5.2276, "step": 1922 }, { "epoch": 0.11937426283444037, "grad_norm": 1.2791137564254522, "learning_rate": 3.9788950962135324e-05, "loss": 5.3355, "step": 1923 }, { "epoch": 0.11943633993419828, "grad_norm": 0.8733574914199606, "learning_rate": 3.980964204427892e-05, "loss": 5.318, "step": 1924 }, { "epoch": 0.11949841703395617, "grad_norm": 0.7297333059698209, "learning_rate": 3.9830333126422514e-05, "loss": 5.2681, "step": 1925 }, { "epoch": 0.11956049413371407, "grad_norm": 0.8386218030717668, "learning_rate": 3.985102420856611e-05, "loss": 5.2467, "step": 1926 }, { "epoch": 0.11962257123347197, "grad_norm": 1.0742292368983855, "learning_rate": 3.9871715290709704e-05, "loss": 5.3025, "step": 1927 }, { "epoch": 0.11968464833322986, "grad_norm": 0.8580957080706613, "learning_rate": 3.98924063728533e-05, "loss": 5.3625, "step": 1928 }, { "epoch": 0.11974672543298777, "grad_norm": 0.883227900444809, "learning_rate": 3.99130974549969e-05, "loss": 5.3309, "step": 1929 }, { "epoch": 0.11980880253274567, "grad_norm": 1.008672598571967, "learning_rate": 3.9933788537140496e-05, "loss": 5.3573, "step": 1930 }, { "epoch": 0.11987087963250356, "grad_norm": 1.1003217811702242, "learning_rate": 3.995447961928409e-05, "loss": 5.2809, "step": 1931 }, { "epoch": 0.11993295673226147, "grad_norm": 0.8967884686558981, "learning_rate": 3.9975170701427686e-05, "loss": 5.2758, "step": 1932 }, { "epoch": 0.11999503383201937, "grad_norm": 0.880809977091551, "learning_rate": 3.999586178357128e-05, "loss": 5.3312, "step": 1933 }, { "epoch": 0.12005711093177726, "grad_norm": 0.7134745257641208, "learning_rate": 4.001655286571488e-05, "loss": 5.2977, "step": 1934 }, { "epoch": 0.12011918803153517, "grad_norm": 0.7999103555521428, "learning_rate": 4.003724394785848e-05, "loss": 5.2133, "step": 1935 }, { "epoch": 0.12018126513129307, "grad_norm": 0.9047668422239189, "learning_rate": 4.005793503000207e-05, "loss": 5.3325, "step": 1936 }, { "epoch": 0.12024334223105096, "grad_norm": 0.8987982316001027, "learning_rate": 4.007862611214567e-05, "loss": 5.2739, "step": 1937 }, { "epoch": 0.12030541933080886, "grad_norm": 0.8337868449864552, "learning_rate": 4.0099317194289264e-05, "loss": 5.3427, "step": 1938 }, { "epoch": 0.12036749643056677, "grad_norm": 0.6915504391424812, "learning_rate": 4.012000827643286e-05, "loss": 5.279, "step": 1939 }, { "epoch": 0.12042957353032466, "grad_norm": 0.818541476944762, "learning_rate": 4.0140699358576454e-05, "loss": 5.2186, "step": 1940 }, { "epoch": 0.12049165063008256, "grad_norm": 1.0240793326760667, "learning_rate": 4.0161390440720056e-05, "loss": 5.2692, "step": 1941 }, { "epoch": 0.12055372772984047, "grad_norm": 0.9838616985613665, "learning_rate": 4.0182081522863644e-05, "loss": 5.2445, "step": 1942 }, { "epoch": 0.12061580482959836, "grad_norm": 0.9010873279504745, "learning_rate": 4.0202772605007246e-05, "loss": 5.1509, "step": 1943 }, { "epoch": 0.12067788192935626, "grad_norm": 0.8409073238565122, "learning_rate": 4.022346368715084e-05, "loss": 5.2897, "step": 1944 }, { "epoch": 0.12073995902911416, "grad_norm": 0.8489459725320639, "learning_rate": 4.0244154769294436e-05, "loss": 5.3124, "step": 1945 }, { "epoch": 0.12080203612887205, "grad_norm": 0.9204699914237804, "learning_rate": 4.026484585143803e-05, "loss": 5.2295, "step": 1946 }, { "epoch": 0.12086411322862996, "grad_norm": 0.9666775408220882, "learning_rate": 4.028553693358163e-05, "loss": 5.3255, "step": 1947 }, { "epoch": 0.12092619032838786, "grad_norm": 0.8676937103565812, "learning_rate": 4.030622801572522e-05, "loss": 5.4042, "step": 1948 }, { "epoch": 0.12098826742814575, "grad_norm": 0.8952541844947426, "learning_rate": 4.032691909786882e-05, "loss": 5.253, "step": 1949 }, { "epoch": 0.12105034452790366, "grad_norm": 0.7501090685693037, "learning_rate": 4.034761018001242e-05, "loss": 5.2484, "step": 1950 }, { "epoch": 0.12111242162766156, "grad_norm": 0.7720119169555721, "learning_rate": 4.036830126215601e-05, "loss": 5.2879, "step": 1951 }, { "epoch": 0.12117449872741945, "grad_norm": 0.8698262457151041, "learning_rate": 4.038899234429961e-05, "loss": 5.2906, "step": 1952 }, { "epoch": 0.12123657582717735, "grad_norm": 0.9453695926292205, "learning_rate": 4.0409683426443203e-05, "loss": 5.311, "step": 1953 }, { "epoch": 0.12129865292693526, "grad_norm": 0.9017286970574485, "learning_rate": 4.04303745085868e-05, "loss": 5.2687, "step": 1954 }, { "epoch": 0.12136073002669315, "grad_norm": 0.9571146403202027, "learning_rate": 4.04510655907304e-05, "loss": 5.2921, "step": 1955 }, { "epoch": 0.12142280712645105, "grad_norm": 0.9917845621204852, "learning_rate": 4.0471756672873995e-05, "loss": 5.2357, "step": 1956 }, { "epoch": 0.12148488422620896, "grad_norm": 0.9053965858225727, "learning_rate": 4.049244775501759e-05, "loss": 5.2727, "step": 1957 }, { "epoch": 0.12154696132596685, "grad_norm": 1.0596418646635553, "learning_rate": 4.0513138837161186e-05, "loss": 5.2868, "step": 1958 }, { "epoch": 0.12160903842572475, "grad_norm": 1.1755687230059184, "learning_rate": 4.053382991930478e-05, "loss": 5.2963, "step": 1959 }, { "epoch": 0.12167111552548265, "grad_norm": 0.9892488048026631, "learning_rate": 4.0554521001448376e-05, "loss": 5.2098, "step": 1960 }, { "epoch": 0.12173319262524054, "grad_norm": 0.8335882313625205, "learning_rate": 4.057521208359197e-05, "loss": 5.1651, "step": 1961 }, { "epoch": 0.12179526972499845, "grad_norm": 0.858985118856971, "learning_rate": 4.059590316573557e-05, "loss": 5.2629, "step": 1962 }, { "epoch": 0.12185734682475635, "grad_norm": 0.9112406806835555, "learning_rate": 4.061659424787916e-05, "loss": 5.3888, "step": 1963 }, { "epoch": 0.12191942392451424, "grad_norm": 0.9925035288769615, "learning_rate": 4.063728533002276e-05, "loss": 5.2687, "step": 1964 }, { "epoch": 0.12198150102427215, "grad_norm": 0.7917674318399542, "learning_rate": 4.065797641216636e-05, "loss": 5.3523, "step": 1965 }, { "epoch": 0.12204357812403005, "grad_norm": 0.8832566823331364, "learning_rate": 4.067866749430995e-05, "loss": 5.3228, "step": 1966 }, { "epoch": 0.12210565522378794, "grad_norm": 1.068025805408023, "learning_rate": 4.069935857645355e-05, "loss": 5.1907, "step": 1967 }, { "epoch": 0.12216773232354584, "grad_norm": 1.0306092786081436, "learning_rate": 4.072004965859715e-05, "loss": 5.2087, "step": 1968 }, { "epoch": 0.12222980942330375, "grad_norm": 0.8970055488448822, "learning_rate": 4.074074074074074e-05, "loss": 5.3152, "step": 1969 }, { "epoch": 0.12229188652306164, "grad_norm": 1.1420491877149335, "learning_rate": 4.076143182288434e-05, "loss": 5.2459, "step": 1970 }, { "epoch": 0.12235396362281954, "grad_norm": 0.8229851553378862, "learning_rate": 4.0782122905027935e-05, "loss": 5.2353, "step": 1971 }, { "epoch": 0.12241604072257745, "grad_norm": 0.9225956637705035, "learning_rate": 4.080281398717153e-05, "loss": 5.2223, "step": 1972 }, { "epoch": 0.12247811782233534, "grad_norm": 0.9489002085756757, "learning_rate": 4.0823505069315125e-05, "loss": 5.2838, "step": 1973 }, { "epoch": 0.12254019492209324, "grad_norm": 0.945804121319767, "learning_rate": 4.084419615145873e-05, "loss": 5.1435, "step": 1974 }, { "epoch": 0.12260227202185114, "grad_norm": 0.9774551279349412, "learning_rate": 4.0864887233602315e-05, "loss": 5.255, "step": 1975 }, { "epoch": 0.12266434912160903, "grad_norm": 0.8898022850496872, "learning_rate": 4.088557831574592e-05, "loss": 5.2083, "step": 1976 }, { "epoch": 0.12272642622136694, "grad_norm": 0.8769266511111323, "learning_rate": 4.090626939788951e-05, "loss": 5.2982, "step": 1977 }, { "epoch": 0.12278850332112484, "grad_norm": 0.9722593485843046, "learning_rate": 4.092696048003311e-05, "loss": 5.3267, "step": 1978 }, { "epoch": 0.12285058042088273, "grad_norm": 1.0262697937667675, "learning_rate": 4.09476515621767e-05, "loss": 5.2702, "step": 1979 }, { "epoch": 0.12291265752064064, "grad_norm": 1.1533690465697195, "learning_rate": 4.0968342644320304e-05, "loss": 5.2595, "step": 1980 }, { "epoch": 0.12297473462039854, "grad_norm": 0.960989477721215, "learning_rate": 4.098903372646389e-05, "loss": 5.3619, "step": 1981 }, { "epoch": 0.12303681172015643, "grad_norm": 0.9151638275713335, "learning_rate": 4.1009724808607494e-05, "loss": 5.2482, "step": 1982 }, { "epoch": 0.12309888881991433, "grad_norm": 1.124597212423241, "learning_rate": 4.103041589075109e-05, "loss": 5.1689, "step": 1983 }, { "epoch": 0.12316096591967224, "grad_norm": 1.0459194092201827, "learning_rate": 4.1051106972894685e-05, "loss": 5.2339, "step": 1984 }, { "epoch": 0.12322304301943013, "grad_norm": 1.0203170963201351, "learning_rate": 4.107179805503828e-05, "loss": 5.3043, "step": 1985 }, { "epoch": 0.12328512011918803, "grad_norm": 1.1840942130494505, "learning_rate": 4.1092489137181875e-05, "loss": 5.3111, "step": 1986 }, { "epoch": 0.12334719721894594, "grad_norm": 0.9338033911621585, "learning_rate": 4.111318021932547e-05, "loss": 5.2624, "step": 1987 }, { "epoch": 0.12340927431870383, "grad_norm": 0.9168829482136447, "learning_rate": 4.1133871301469065e-05, "loss": 5.2485, "step": 1988 }, { "epoch": 0.12347135141846173, "grad_norm": 0.8895780533418741, "learning_rate": 4.115456238361267e-05, "loss": 5.2227, "step": 1989 }, { "epoch": 0.12353342851821963, "grad_norm": 0.8850760010235791, "learning_rate": 4.1175253465756255e-05, "loss": 5.2237, "step": 1990 }, { "epoch": 0.12359550561797752, "grad_norm": 0.8851381574433146, "learning_rate": 4.119594454789986e-05, "loss": 5.1936, "step": 1991 }, { "epoch": 0.12365758271773543, "grad_norm": 0.9581798278489498, "learning_rate": 4.121663563004345e-05, "loss": 5.316, "step": 1992 }, { "epoch": 0.12371965981749333, "grad_norm": 1.050577760867773, "learning_rate": 4.1237326712187054e-05, "loss": 5.2182, "step": 1993 }, { "epoch": 0.12378173691725122, "grad_norm": 1.0339490076695346, "learning_rate": 4.125801779433064e-05, "loss": 5.3302, "step": 1994 }, { "epoch": 0.12384381401700913, "grad_norm": 1.0770921597343555, "learning_rate": 4.1278708876474244e-05, "loss": 5.3347, "step": 1995 }, { "epoch": 0.12390589111676703, "grad_norm": 0.8107235267897506, "learning_rate": 4.129939995861784e-05, "loss": 5.2065, "step": 1996 }, { "epoch": 0.12396796821652492, "grad_norm": 0.8222122356270833, "learning_rate": 4.1320091040761434e-05, "loss": 5.2516, "step": 1997 }, { "epoch": 0.12403004531628282, "grad_norm": 0.9915367455418406, "learning_rate": 4.134078212290503e-05, "loss": 5.1975, "step": 1998 }, { "epoch": 0.12409212241604073, "grad_norm": 0.9368433878160627, "learning_rate": 4.136147320504863e-05, "loss": 5.2709, "step": 1999 }, { "epoch": 0.12415419951579862, "grad_norm": 1.0710980950176274, "learning_rate": 4.138216428719222e-05, "loss": 5.2582, "step": 2000 }, { "epoch": 0.12421627661555652, "grad_norm": 1.1300424316339148, "learning_rate": 4.140285536933582e-05, "loss": 5.2039, "step": 2001 }, { "epoch": 0.12427835371531443, "grad_norm": 1.0269191430158557, "learning_rate": 4.1423546451479416e-05, "loss": 5.2227, "step": 2002 }, { "epoch": 0.12434043081507232, "grad_norm": 0.9960764565556112, "learning_rate": 4.144423753362301e-05, "loss": 5.1864, "step": 2003 }, { "epoch": 0.12440250791483022, "grad_norm": 0.8224151910231405, "learning_rate": 4.1464928615766607e-05, "loss": 5.2954, "step": 2004 }, { "epoch": 0.12446458501458812, "grad_norm": 0.8154795008994258, "learning_rate": 4.148561969791021e-05, "loss": 5.2622, "step": 2005 }, { "epoch": 0.12452666211434602, "grad_norm": 0.695737537387555, "learning_rate": 4.15063107800538e-05, "loss": 5.2179, "step": 2006 }, { "epoch": 0.12458873921410392, "grad_norm": 0.7712940609060704, "learning_rate": 4.15270018621974e-05, "loss": 5.1839, "step": 2007 }, { "epoch": 0.12465081631386182, "grad_norm": 0.8538289138939692, "learning_rate": 4.1547692944340994e-05, "loss": 5.2892, "step": 2008 }, { "epoch": 0.12471289341361971, "grad_norm": 0.9036434441057587, "learning_rate": 4.156838402648459e-05, "loss": 5.1842, "step": 2009 }, { "epoch": 0.12477497051337762, "grad_norm": 0.8762698835982067, "learning_rate": 4.1589075108628184e-05, "loss": 5.1975, "step": 2010 }, { "epoch": 0.12483704761313552, "grad_norm": 0.8626658805333093, "learning_rate": 4.160976619077178e-05, "loss": 5.184, "step": 2011 }, { "epoch": 0.12489912471289341, "grad_norm": 0.89102541973746, "learning_rate": 4.1630457272915374e-05, "loss": 5.2819, "step": 2012 }, { "epoch": 0.12496120181265132, "grad_norm": 0.8043084594425873, "learning_rate": 4.165114835505897e-05, "loss": 5.1926, "step": 2013 }, { "epoch": 0.12502327891240922, "grad_norm": 0.7957925363306172, "learning_rate": 4.167183943720257e-05, "loss": 5.2576, "step": 2014 }, { "epoch": 0.1250853560121671, "grad_norm": 1.0277091417219868, "learning_rate": 4.169253051934616e-05, "loss": 5.304, "step": 2015 }, { "epoch": 0.125147433111925, "grad_norm": 1.1111968755735795, "learning_rate": 4.171322160148976e-05, "loss": 5.2989, "step": 2016 }, { "epoch": 0.12520951021168292, "grad_norm": 0.8650880435832151, "learning_rate": 4.1733912683633356e-05, "loss": 5.2786, "step": 2017 }, { "epoch": 0.1252715873114408, "grad_norm": 0.949566592936124, "learning_rate": 4.175460376577695e-05, "loss": 5.1295, "step": 2018 }, { "epoch": 0.1253336644111987, "grad_norm": 1.1208765437554877, "learning_rate": 4.1775294847920546e-05, "loss": 5.1681, "step": 2019 }, { "epoch": 0.12539574151095662, "grad_norm": 0.9442384150740998, "learning_rate": 4.179598593006415e-05, "loss": 5.1072, "step": 2020 }, { "epoch": 0.1254578186107145, "grad_norm": 0.9027127588936735, "learning_rate": 4.1816677012207736e-05, "loss": 5.1884, "step": 2021 }, { "epoch": 0.1255198957104724, "grad_norm": 1.0135547959881201, "learning_rate": 4.183736809435134e-05, "loss": 5.0834, "step": 2022 }, { "epoch": 0.1255819728102303, "grad_norm": 1.0187585467582678, "learning_rate": 4.185805917649493e-05, "loss": 5.1547, "step": 2023 }, { "epoch": 0.1256440499099882, "grad_norm": 0.8652855231535612, "learning_rate": 4.187875025863853e-05, "loss": 5.2981, "step": 2024 }, { "epoch": 0.1257061270097461, "grad_norm": 0.7883215720957369, "learning_rate": 4.1899441340782123e-05, "loss": 5.2236, "step": 2025 }, { "epoch": 0.125768204109504, "grad_norm": 0.9437294623053918, "learning_rate": 4.1920132422925725e-05, "loss": 5.2156, "step": 2026 }, { "epoch": 0.1258302812092619, "grad_norm": 0.8504806085446975, "learning_rate": 4.1940823505069314e-05, "loss": 5.2137, "step": 2027 }, { "epoch": 0.1258923583090198, "grad_norm": 0.8314040779783671, "learning_rate": 4.1961514587212915e-05, "loss": 5.2038, "step": 2028 }, { "epoch": 0.1259544354087777, "grad_norm": 0.9172573173503483, "learning_rate": 4.198220566935651e-05, "loss": 5.204, "step": 2029 }, { "epoch": 0.1260165125085356, "grad_norm": 1.0378098358423347, "learning_rate": 4.2002896751500106e-05, "loss": 5.177, "step": 2030 }, { "epoch": 0.1260785896082935, "grad_norm": 1.1791300688187478, "learning_rate": 4.20235878336437e-05, "loss": 5.157, "step": 2031 }, { "epoch": 0.1261406667080514, "grad_norm": 0.7911859776296689, "learning_rate": 4.20442789157873e-05, "loss": 5.185, "step": 2032 }, { "epoch": 0.1262027438078093, "grad_norm": 0.9161314413224453, "learning_rate": 4.206496999793089e-05, "loss": 5.2574, "step": 2033 }, { "epoch": 0.1262648209075672, "grad_norm": 1.1260800475367585, "learning_rate": 4.208566108007449e-05, "loss": 5.1005, "step": 2034 }, { "epoch": 0.1263268980073251, "grad_norm": 0.995039767472218, "learning_rate": 4.210635216221809e-05, "loss": 5.168, "step": 2035 }, { "epoch": 0.126388975107083, "grad_norm": 0.9381809940184979, "learning_rate": 4.212704324436168e-05, "loss": 5.1079, "step": 2036 }, { "epoch": 0.12645105220684089, "grad_norm": 1.0523883201262716, "learning_rate": 4.214773432650528e-05, "loss": 5.2519, "step": 2037 }, { "epoch": 0.1265131293065988, "grad_norm": 0.9413131414565389, "learning_rate": 4.216842540864887e-05, "loss": 5.3451, "step": 2038 }, { "epoch": 0.1265752064063567, "grad_norm": 1.0186342115071152, "learning_rate": 4.218911649079247e-05, "loss": 5.1726, "step": 2039 }, { "epoch": 0.12663728350611458, "grad_norm": 1.0187257237262537, "learning_rate": 4.220980757293606e-05, "loss": 5.1213, "step": 2040 }, { "epoch": 0.1266993606058725, "grad_norm": 1.06668998101703, "learning_rate": 4.2230498655079665e-05, "loss": 5.2036, "step": 2041 }, { "epoch": 0.1267614377056304, "grad_norm": 0.9270051680341606, "learning_rate": 4.225118973722326e-05, "loss": 5.1253, "step": 2042 }, { "epoch": 0.12682351480538828, "grad_norm": 0.8749191723684812, "learning_rate": 4.2271880819366855e-05, "loss": 5.0103, "step": 2043 }, { "epoch": 0.1268855919051462, "grad_norm": 1.0681133091500508, "learning_rate": 4.229257190151045e-05, "loss": 5.1525, "step": 2044 }, { "epoch": 0.1269476690049041, "grad_norm": 0.9719201791939636, "learning_rate": 4.2313262983654045e-05, "loss": 5.0653, "step": 2045 }, { "epoch": 0.12700974610466198, "grad_norm": 1.0655078946648586, "learning_rate": 4.233395406579764e-05, "loss": 5.1437, "step": 2046 }, { "epoch": 0.1270718232044199, "grad_norm": 1.0214333749442854, "learning_rate": 4.235464514794124e-05, "loss": 5.1135, "step": 2047 }, { "epoch": 0.1271339003041778, "grad_norm": 1.0626010846690166, "learning_rate": 4.237533623008483e-05, "loss": 5.1798, "step": 2048 }, { "epoch": 0.12719597740393568, "grad_norm": 1.043677328629333, "learning_rate": 4.239602731222843e-05, "loss": 5.2542, "step": 2049 }, { "epoch": 0.1272580545036936, "grad_norm": 0.8170680963579465, "learning_rate": 4.241671839437203e-05, "loss": 5.1624, "step": 2050 }, { "epoch": 0.1273201316034515, "grad_norm": 0.8159507685517149, "learning_rate": 4.243740947651562e-05, "loss": 5.2428, "step": 2051 }, { "epoch": 0.12738220870320938, "grad_norm": 0.9196570596378665, "learning_rate": 4.245810055865922e-05, "loss": 5.1444, "step": 2052 }, { "epoch": 0.1274442858029673, "grad_norm": 0.9558127606635441, "learning_rate": 4.247879164080282e-05, "loss": 5.114, "step": 2053 }, { "epoch": 0.12750636290272518, "grad_norm": 0.9343584831079287, "learning_rate": 4.249948272294641e-05, "loss": 5.1249, "step": 2054 }, { "epoch": 0.12756844000248307, "grad_norm": 0.971698948654889, "learning_rate": 4.252017380509001e-05, "loss": 5.1216, "step": 2055 }, { "epoch": 0.127630517102241, "grad_norm": 1.0638829375239733, "learning_rate": 4.2540864887233605e-05, "loss": 5.192, "step": 2056 }, { "epoch": 0.12769259420199888, "grad_norm": 0.9559101934607606, "learning_rate": 4.25615559693772e-05, "loss": 5.1628, "step": 2057 }, { "epoch": 0.12775467130175677, "grad_norm": 0.9238546420076164, "learning_rate": 4.2582247051520795e-05, "loss": 5.0962, "step": 2058 }, { "epoch": 0.1278167484015147, "grad_norm": 0.9353863094532642, "learning_rate": 4.26029381336644e-05, "loss": 5.2089, "step": 2059 }, { "epoch": 0.12787882550127258, "grad_norm": 0.9421086203988648, "learning_rate": 4.2623629215807985e-05, "loss": 5.0297, "step": 2060 }, { "epoch": 0.12794090260103047, "grad_norm": 0.8988906712593616, "learning_rate": 4.264432029795159e-05, "loss": 5.0605, "step": 2061 }, { "epoch": 0.1280029797007884, "grad_norm": 0.8743528629688251, "learning_rate": 4.266501138009518e-05, "loss": 5.1695, "step": 2062 }, { "epoch": 0.12806505680054628, "grad_norm": 0.9430254548602621, "learning_rate": 4.268570246223878e-05, "loss": 5.1971, "step": 2063 }, { "epoch": 0.12812713390030417, "grad_norm": 1.1596575863022074, "learning_rate": 4.270639354438237e-05, "loss": 5.1329, "step": 2064 }, { "epoch": 0.1281892110000621, "grad_norm": 1.0701169023201846, "learning_rate": 4.2727084626525974e-05, "loss": 5.0704, "step": 2065 }, { "epoch": 0.12825128809981998, "grad_norm": 1.0756827625854994, "learning_rate": 4.274777570866956e-05, "loss": 5.1317, "step": 2066 }, { "epoch": 0.12831336519957787, "grad_norm": 0.8850138739128783, "learning_rate": 4.2768466790813164e-05, "loss": 5.1859, "step": 2067 }, { "epoch": 0.12837544229933578, "grad_norm": 0.8967781665909972, "learning_rate": 4.278915787295676e-05, "loss": 5.0813, "step": 2068 }, { "epoch": 0.12843751939909367, "grad_norm": 1.0593094887071437, "learning_rate": 4.2809848955100354e-05, "loss": 5.139, "step": 2069 }, { "epoch": 0.12849959649885156, "grad_norm": 1.1039925683918588, "learning_rate": 4.283054003724395e-05, "loss": 5.1545, "step": 2070 }, { "epoch": 0.12856167359860948, "grad_norm": 0.9491144444805477, "learning_rate": 4.2851231119387544e-05, "loss": 5.2064, "step": 2071 }, { "epoch": 0.12862375069836737, "grad_norm": 0.8751014582531479, "learning_rate": 4.287192220153114e-05, "loss": 5.2328, "step": 2072 }, { "epoch": 0.12868582779812526, "grad_norm": 0.9340193987206218, "learning_rate": 4.2892613283674735e-05, "loss": 5.0334, "step": 2073 }, { "epoch": 0.12874790489788318, "grad_norm": 0.9312396196132446, "learning_rate": 4.2913304365818336e-05, "loss": 5.2202, "step": 2074 }, { "epoch": 0.12880998199764107, "grad_norm": 0.951096236023741, "learning_rate": 4.2933995447961925e-05, "loss": 5.1376, "step": 2075 }, { "epoch": 0.12887205909739896, "grad_norm": 0.9979529887194546, "learning_rate": 4.2954686530105527e-05, "loss": 5.1456, "step": 2076 }, { "epoch": 0.12893413619715688, "grad_norm": 1.0000746222310202, "learning_rate": 4.297537761224912e-05, "loss": 5.1541, "step": 2077 }, { "epoch": 0.12899621329691477, "grad_norm": 0.9478573239094584, "learning_rate": 4.2996068694392723e-05, "loss": 5.1151, "step": 2078 }, { "epoch": 0.12905829039667266, "grad_norm": 0.8855082445843215, "learning_rate": 4.301675977653631e-05, "loss": 5.1843, "step": 2079 }, { "epoch": 0.12912036749643058, "grad_norm": 1.070197440318392, "learning_rate": 4.3037450858679914e-05, "loss": 5.0821, "step": 2080 }, { "epoch": 0.12918244459618847, "grad_norm": 0.9246437069603186, "learning_rate": 4.305814194082351e-05, "loss": 5.1214, "step": 2081 }, { "epoch": 0.12924452169594636, "grad_norm": 0.8839868819419925, "learning_rate": 4.3078833022967104e-05, "loss": 5.071, "step": 2082 }, { "epoch": 0.12930659879570428, "grad_norm": 0.8890240652562429, "learning_rate": 4.30995241051107e-05, "loss": 5.1878, "step": 2083 }, { "epoch": 0.12936867589546217, "grad_norm": 0.8422376005596691, "learning_rate": 4.31202151872543e-05, "loss": 5.0708, "step": 2084 }, { "epoch": 0.12943075299522006, "grad_norm": 0.8069022726447947, "learning_rate": 4.314090626939789e-05, "loss": 5.1732, "step": 2085 }, { "epoch": 0.12949283009497797, "grad_norm": 0.974542252363954, "learning_rate": 4.316159735154149e-05, "loss": 5.0526, "step": 2086 }, { "epoch": 0.12955490719473586, "grad_norm": 1.0432934773442608, "learning_rate": 4.3182288433685086e-05, "loss": 5.1017, "step": 2087 }, { "epoch": 0.12961698429449375, "grad_norm": 1.0377572832175908, "learning_rate": 4.320297951582868e-05, "loss": 5.1879, "step": 2088 }, { "epoch": 0.12967906139425167, "grad_norm": 0.9817482248362595, "learning_rate": 4.3223670597972276e-05, "loss": 5.1233, "step": 2089 }, { "epoch": 0.12974113849400956, "grad_norm": 1.162276619809637, "learning_rate": 4.324436168011588e-05, "loss": 5.2079, "step": 2090 }, { "epoch": 0.12980321559376745, "grad_norm": 1.1421442235009602, "learning_rate": 4.3265052762259466e-05, "loss": 5.0415, "step": 2091 }, { "epoch": 0.12986529269352537, "grad_norm": 1.0176566710515895, "learning_rate": 4.328574384440307e-05, "loss": 5.1578, "step": 2092 }, { "epoch": 0.12992736979328326, "grad_norm": 1.047343590469041, "learning_rate": 4.330643492654666e-05, "loss": 4.9913, "step": 2093 }, { "epoch": 0.12998944689304115, "grad_norm": 1.014797812839838, "learning_rate": 4.332712600869026e-05, "loss": 5.0727, "step": 2094 }, { "epoch": 0.13005152399279907, "grad_norm": 0.9549151634561568, "learning_rate": 4.334781709083385e-05, "loss": 5.1361, "step": 2095 }, { "epoch": 0.13011360109255696, "grad_norm": 1.0052643729024122, "learning_rate": 4.336850817297745e-05, "loss": 5.1327, "step": 2096 }, { "epoch": 0.13017567819231485, "grad_norm": 1.0945086436690594, "learning_rate": 4.3389199255121043e-05, "loss": 5.1374, "step": 2097 }, { "epoch": 0.13023775529207277, "grad_norm": 0.9659269484230893, "learning_rate": 4.340989033726464e-05, "loss": 5.143, "step": 2098 }, { "epoch": 0.13029983239183066, "grad_norm": 0.8580328083456956, "learning_rate": 4.343058141940824e-05, "loss": 5.1524, "step": 2099 }, { "epoch": 0.13036190949158855, "grad_norm": 0.8994612790118087, "learning_rate": 4.345127250155183e-05, "loss": 5.1241, "step": 2100 }, { "epoch": 0.13042398659134646, "grad_norm": 0.9417513139764679, "learning_rate": 4.347196358369543e-05, "loss": 5.1357, "step": 2101 }, { "epoch": 0.13048606369110435, "grad_norm": 1.0008144638184238, "learning_rate": 4.3492654665839026e-05, "loss": 5.0818, "step": 2102 }, { "epoch": 0.13054814079086224, "grad_norm": 1.0422835367269345, "learning_rate": 4.351334574798262e-05, "loss": 5.1656, "step": 2103 }, { "epoch": 0.13061021789062016, "grad_norm": 1.0494861776087774, "learning_rate": 4.3534036830126216e-05, "loss": 5.0677, "step": 2104 }, { "epoch": 0.13067229499037805, "grad_norm": 0.9422333752042784, "learning_rate": 4.355472791226982e-05, "loss": 5.0682, "step": 2105 }, { "epoch": 0.13073437209013594, "grad_norm": 0.8618660628842294, "learning_rate": 4.3575418994413406e-05, "loss": 5.1576, "step": 2106 }, { "epoch": 0.13079644918989386, "grad_norm": 0.8529078504153006, "learning_rate": 4.359611007655701e-05, "loss": 5.0991, "step": 2107 }, { "epoch": 0.13085852628965175, "grad_norm": 0.9481906425067448, "learning_rate": 4.36168011587006e-05, "loss": 5.0876, "step": 2108 }, { "epoch": 0.13092060338940964, "grad_norm": 1.003900075218121, "learning_rate": 4.36374922408442e-05, "loss": 5.1995, "step": 2109 }, { "epoch": 0.13098268048916756, "grad_norm": 0.9446896794855868, "learning_rate": 4.365818332298779e-05, "loss": 5.1611, "step": 2110 }, { "epoch": 0.13104475758892545, "grad_norm": 0.9812464574276095, "learning_rate": 4.3678874405131395e-05, "loss": 5.1553, "step": 2111 }, { "epoch": 0.13110683468868334, "grad_norm": 1.0456720529242645, "learning_rate": 4.369956548727498e-05, "loss": 4.9191, "step": 2112 }, { "epoch": 0.13116891178844126, "grad_norm": 0.8142846388319749, "learning_rate": 4.3720256569418585e-05, "loss": 5.1653, "step": 2113 }, { "epoch": 0.13123098888819915, "grad_norm": 0.9456272074288512, "learning_rate": 4.374094765156218e-05, "loss": 5.0845, "step": 2114 }, { "epoch": 0.13129306598795704, "grad_norm": 0.932639524182365, "learning_rate": 4.3761638733705775e-05, "loss": 5.0766, "step": 2115 }, { "epoch": 0.13135514308771495, "grad_norm": 0.8683895337508585, "learning_rate": 4.378232981584937e-05, "loss": 5.0791, "step": 2116 }, { "epoch": 0.13141722018747284, "grad_norm": 0.8317297805722587, "learning_rate": 4.380302089799297e-05, "loss": 5.0759, "step": 2117 }, { "epoch": 0.13147929728723073, "grad_norm": 0.7786333514757932, "learning_rate": 4.382371198013656e-05, "loss": 5.1877, "step": 2118 }, { "epoch": 0.13154137438698865, "grad_norm": 0.9735059588562932, "learning_rate": 4.384440306228016e-05, "loss": 5.1077, "step": 2119 }, { "epoch": 0.13160345148674654, "grad_norm": 1.0208577847309648, "learning_rate": 4.386509414442376e-05, "loss": 5.1048, "step": 2120 }, { "epoch": 0.13166552858650443, "grad_norm": 0.9083731871503234, "learning_rate": 4.388578522656735e-05, "loss": 4.9271, "step": 2121 }, { "epoch": 0.13172760568626235, "grad_norm": 0.8293936386831231, "learning_rate": 4.390647630871095e-05, "loss": 5.0807, "step": 2122 }, { "epoch": 0.13178968278602024, "grad_norm": 0.9184053397348233, "learning_rate": 4.392716739085454e-05, "loss": 5.1202, "step": 2123 }, { "epoch": 0.13185175988577813, "grad_norm": 0.8260486050892031, "learning_rate": 4.394785847299814e-05, "loss": 5.1284, "step": 2124 }, { "epoch": 0.13191383698553605, "grad_norm": 0.9284325151235832, "learning_rate": 4.396854955514173e-05, "loss": 5.1088, "step": 2125 }, { "epoch": 0.13197591408529394, "grad_norm": 0.8887274756172745, "learning_rate": 4.3989240637285335e-05, "loss": 5.1042, "step": 2126 }, { "epoch": 0.13203799118505183, "grad_norm": 1.0175043178157435, "learning_rate": 4.400993171942892e-05, "loss": 5.0965, "step": 2127 }, { "epoch": 0.13210006828480975, "grad_norm": 1.018379115830431, "learning_rate": 4.4030622801572525e-05, "loss": 5.1106, "step": 2128 }, { "epoch": 0.13216214538456764, "grad_norm": 0.8906460475107739, "learning_rate": 4.405131388371612e-05, "loss": 5.0819, "step": 2129 }, { "epoch": 0.13222422248432553, "grad_norm": 0.8868381907141202, "learning_rate": 4.4072004965859715e-05, "loss": 5.0838, "step": 2130 }, { "epoch": 0.13228629958408344, "grad_norm": 0.8826854243009314, "learning_rate": 4.409269604800331e-05, "loss": 5.1479, "step": 2131 }, { "epoch": 0.13234837668384133, "grad_norm": 0.815111109467372, "learning_rate": 4.411338713014691e-05, "loss": 5.1319, "step": 2132 }, { "epoch": 0.13241045378359922, "grad_norm": 0.9274471440639346, "learning_rate": 4.41340782122905e-05, "loss": 4.9531, "step": 2133 }, { "epoch": 0.13247253088335714, "grad_norm": 0.9497760031791206, "learning_rate": 4.41547692944341e-05, "loss": 5.1279, "step": 2134 }, { "epoch": 0.13253460798311503, "grad_norm": 1.2462893724104869, "learning_rate": 4.41754603765777e-05, "loss": 5.0809, "step": 2135 }, { "epoch": 0.13259668508287292, "grad_norm": 0.8787529366608301, "learning_rate": 4.419615145872129e-05, "loss": 5.0978, "step": 2136 }, { "epoch": 0.13265876218263084, "grad_norm": 0.9473279165649537, "learning_rate": 4.421684254086489e-05, "loss": 5.0745, "step": 2137 }, { "epoch": 0.13272083928238873, "grad_norm": 0.9998792336974351, "learning_rate": 4.423753362300849e-05, "loss": 5.1164, "step": 2138 }, { "epoch": 0.13278291638214662, "grad_norm": 1.0496370687405332, "learning_rate": 4.425822470515208e-05, "loss": 5.1561, "step": 2139 }, { "epoch": 0.13284499348190454, "grad_norm": 0.9879866752826383, "learning_rate": 4.427891578729568e-05, "loss": 5.071, "step": 2140 }, { "epoch": 0.13290707058166243, "grad_norm": 0.9866932712385423, "learning_rate": 4.4299606869439274e-05, "loss": 5.0791, "step": 2141 }, { "epoch": 0.13296914768142032, "grad_norm": 1.1748484209375971, "learning_rate": 4.432029795158287e-05, "loss": 5.0143, "step": 2142 }, { "epoch": 0.13303122478117824, "grad_norm": 0.9646708839401651, "learning_rate": 4.4340989033726464e-05, "loss": 5.1344, "step": 2143 }, { "epoch": 0.13309330188093613, "grad_norm": 0.9476070457320549, "learning_rate": 4.4361680115870066e-05, "loss": 5.0419, "step": 2144 }, { "epoch": 0.13315537898069402, "grad_norm": 0.8567478598130196, "learning_rate": 4.4382371198013655e-05, "loss": 5.0242, "step": 2145 }, { "epoch": 0.13321745608045193, "grad_norm": 0.9156669815386034, "learning_rate": 4.4403062280157256e-05, "loss": 5.1012, "step": 2146 }, { "epoch": 0.13327953318020982, "grad_norm": 1.0217063181299173, "learning_rate": 4.442375336230085e-05, "loss": 5.1293, "step": 2147 }, { "epoch": 0.13334161027996771, "grad_norm": 1.1231751476506724, "learning_rate": 4.4444444444444447e-05, "loss": 5.1396, "step": 2148 }, { "epoch": 0.13340368737972563, "grad_norm": 0.9454147661867258, "learning_rate": 4.446513552658804e-05, "loss": 5.097, "step": 2149 }, { "epoch": 0.13346576447948352, "grad_norm": 1.042404993974517, "learning_rate": 4.448582660873164e-05, "loss": 5.0723, "step": 2150 }, { "epoch": 0.1335278415792414, "grad_norm": 1.050101134561337, "learning_rate": 4.450651769087523e-05, "loss": 5.0815, "step": 2151 }, { "epoch": 0.13358991867899933, "grad_norm": 0.8514307252611164, "learning_rate": 4.4527208773018834e-05, "loss": 4.9861, "step": 2152 }, { "epoch": 0.13365199577875722, "grad_norm": 0.9706321398686759, "learning_rate": 4.454789985516243e-05, "loss": 5.0123, "step": 2153 }, { "epoch": 0.1337140728785151, "grad_norm": 1.018153930121899, "learning_rate": 4.4568590937306024e-05, "loss": 4.9985, "step": 2154 }, { "epoch": 0.133776149978273, "grad_norm": 0.8632412724278842, "learning_rate": 4.458928201944962e-05, "loss": 5.0895, "step": 2155 }, { "epoch": 0.13383822707803092, "grad_norm": 1.0795676489195802, "learning_rate": 4.4609973101593214e-05, "loss": 5.068, "step": 2156 }, { "epoch": 0.1339003041777888, "grad_norm": 1.1883772571109334, "learning_rate": 4.463066418373681e-05, "loss": 5.088, "step": 2157 }, { "epoch": 0.1339623812775467, "grad_norm": 1.0494468185450008, "learning_rate": 4.4651355265880404e-05, "loss": 4.9372, "step": 2158 }, { "epoch": 0.13402445837730462, "grad_norm": 0.9839854302122387, "learning_rate": 4.4672046348024006e-05, "loss": 4.9867, "step": 2159 }, { "epoch": 0.1340865354770625, "grad_norm": 0.9674655335962862, "learning_rate": 4.4692737430167594e-05, "loss": 5.1266, "step": 2160 }, { "epoch": 0.1341486125768204, "grad_norm": 1.0784830797943517, "learning_rate": 4.4713428512311196e-05, "loss": 5.0979, "step": 2161 }, { "epoch": 0.13421068967657832, "grad_norm": 0.9864277704345539, "learning_rate": 4.473411959445479e-05, "loss": 4.9944, "step": 2162 }, { "epoch": 0.1342727667763362, "grad_norm": 1.1333129303629015, "learning_rate": 4.475481067659839e-05, "loss": 4.962, "step": 2163 }, { "epoch": 0.1343348438760941, "grad_norm": 1.0259573396733364, "learning_rate": 4.477550175874198e-05, "loss": 5.1376, "step": 2164 }, { "epoch": 0.134396920975852, "grad_norm": 1.0036105896892424, "learning_rate": 4.479619284088558e-05, "loss": 5.1694, "step": 2165 }, { "epoch": 0.1344589980756099, "grad_norm": 1.0623797180308006, "learning_rate": 4.481688392302918e-05, "loss": 5.0524, "step": 2166 }, { "epoch": 0.1345210751753678, "grad_norm": 1.0114549565414388, "learning_rate": 4.483757500517277e-05, "loss": 5.0232, "step": 2167 }, { "epoch": 0.1345831522751257, "grad_norm": 0.8753371270449325, "learning_rate": 4.485826608731637e-05, "loss": 5.0902, "step": 2168 }, { "epoch": 0.1346452293748836, "grad_norm": 0.8598888594870042, "learning_rate": 4.487895716945997e-05, "loss": 5.0584, "step": 2169 }, { "epoch": 0.1347073064746415, "grad_norm": 0.8207817188888306, "learning_rate": 4.489964825160356e-05, "loss": 5.0587, "step": 2170 }, { "epoch": 0.1347693835743994, "grad_norm": 0.8596412593007796, "learning_rate": 4.492033933374716e-05, "loss": 4.9613, "step": 2171 }, { "epoch": 0.1348314606741573, "grad_norm": 0.9104300811500924, "learning_rate": 4.4941030415890756e-05, "loss": 5.0733, "step": 2172 }, { "epoch": 0.1348935377739152, "grad_norm": 0.9809992293627995, "learning_rate": 4.496172149803435e-05, "loss": 4.9336, "step": 2173 }, { "epoch": 0.1349556148736731, "grad_norm": 0.8739457250389598, "learning_rate": 4.4982412580177946e-05, "loss": 4.9653, "step": 2174 }, { "epoch": 0.135017691973431, "grad_norm": 0.7931288008001354, "learning_rate": 4.500310366232154e-05, "loss": 5.108, "step": 2175 }, { "epoch": 0.1350797690731889, "grad_norm": 0.8777584464792125, "learning_rate": 4.5023794744465136e-05, "loss": 5.0052, "step": 2176 }, { "epoch": 0.1351418461729468, "grad_norm": 0.943189491155109, "learning_rate": 4.504448582660874e-05, "loss": 4.9896, "step": 2177 }, { "epoch": 0.1352039232727047, "grad_norm": 0.8648406913448234, "learning_rate": 4.506517690875233e-05, "loss": 5.0054, "step": 2178 }, { "epoch": 0.13526600037246259, "grad_norm": 0.8368065852945481, "learning_rate": 4.508586799089593e-05, "loss": 5.1095, "step": 2179 }, { "epoch": 0.1353280774722205, "grad_norm": 0.9293535418107952, "learning_rate": 4.510655907303952e-05, "loss": 5.0636, "step": 2180 }, { "epoch": 0.1353901545719784, "grad_norm": 1.0099046268378677, "learning_rate": 4.512725015518312e-05, "loss": 5.1208, "step": 2181 }, { "epoch": 0.13545223167173628, "grad_norm": 1.2366774127387063, "learning_rate": 4.514794123732671e-05, "loss": 5.1937, "step": 2182 }, { "epoch": 0.1355143087714942, "grad_norm": 1.0635758731035132, "learning_rate": 4.516863231947031e-05, "loss": 5.0363, "step": 2183 }, { "epoch": 0.1355763858712521, "grad_norm": 1.0007870080619934, "learning_rate": 4.518932340161391e-05, "loss": 5.0873, "step": 2184 }, { "epoch": 0.13563846297100998, "grad_norm": 1.1330932269347007, "learning_rate": 4.52100144837575e-05, "loss": 5.0462, "step": 2185 }, { "epoch": 0.1357005400707679, "grad_norm": 0.9086399791499352, "learning_rate": 4.52307055659011e-05, "loss": 5.0044, "step": 2186 }, { "epoch": 0.1357626171705258, "grad_norm": 1.0412472579938268, "learning_rate": 4.5251396648044695e-05, "loss": 4.9954, "step": 2187 }, { "epoch": 0.13582469427028368, "grad_norm": 1.07184385471304, "learning_rate": 4.527208773018829e-05, "loss": 5.0528, "step": 2188 }, { "epoch": 0.1358867713700416, "grad_norm": 1.053533700031929, "learning_rate": 4.5292778812331885e-05, "loss": 5.0659, "step": 2189 }, { "epoch": 0.1359488484697995, "grad_norm": 1.0916242377696668, "learning_rate": 4.531346989447549e-05, "loss": 5.0163, "step": 2190 }, { "epoch": 0.13601092556955738, "grad_norm": 0.9294836958921942, "learning_rate": 4.5334160976619076e-05, "loss": 4.9868, "step": 2191 }, { "epoch": 0.1360730026693153, "grad_norm": 0.8487475002740783, "learning_rate": 4.535485205876268e-05, "loss": 5.0899, "step": 2192 }, { "epoch": 0.1361350797690732, "grad_norm": 0.941639471646005, "learning_rate": 4.537554314090627e-05, "loss": 5.096, "step": 2193 }, { "epoch": 0.13619715686883108, "grad_norm": 1.0184207875793436, "learning_rate": 4.539623422304987e-05, "loss": 5.0203, "step": 2194 }, { "epoch": 0.136259233968589, "grad_norm": 0.926362092858464, "learning_rate": 4.541692530519346e-05, "loss": 5.0765, "step": 2195 }, { "epoch": 0.13632131106834688, "grad_norm": 1.0121077448818163, "learning_rate": 4.5437616387337064e-05, "loss": 5.0743, "step": 2196 }, { "epoch": 0.13638338816810477, "grad_norm": 1.1052509538136879, "learning_rate": 4.545830746948065e-05, "loss": 5.082, "step": 2197 }, { "epoch": 0.1364454652678627, "grad_norm": 1.0417907005215765, "learning_rate": 4.5478998551624255e-05, "loss": 4.9804, "step": 2198 }, { "epoch": 0.13650754236762058, "grad_norm": 1.1094686979027288, "learning_rate": 4.549968963376785e-05, "loss": 4.9663, "step": 2199 }, { "epoch": 0.13656961946737847, "grad_norm": 1.0572443170076782, "learning_rate": 4.5520380715911445e-05, "loss": 4.9787, "step": 2200 }, { "epoch": 0.1366316965671364, "grad_norm": 0.7494002169199239, "learning_rate": 4.554107179805504e-05, "loss": 5.0953, "step": 2201 }, { "epoch": 0.13669377366689428, "grad_norm": 0.8136997901204581, "learning_rate": 4.556176288019864e-05, "loss": 5.1121, "step": 2202 }, { "epoch": 0.13675585076665217, "grad_norm": 1.0395953884203006, "learning_rate": 4.558245396234223e-05, "loss": 5.0593, "step": 2203 }, { "epoch": 0.1368179278664101, "grad_norm": 0.9526790451118088, "learning_rate": 4.560314504448583e-05, "loss": 5.0342, "step": 2204 }, { "epoch": 0.13688000496616798, "grad_norm": 1.0658050631088796, "learning_rate": 4.562383612662943e-05, "loss": 5.0073, "step": 2205 }, { "epoch": 0.13694208206592587, "grad_norm": 1.0061625495962478, "learning_rate": 4.564452720877302e-05, "loss": 4.9925, "step": 2206 }, { "epoch": 0.1370041591656838, "grad_norm": 1.205346570649785, "learning_rate": 4.566521829091662e-05, "loss": 5.0627, "step": 2207 }, { "epoch": 0.13706623626544168, "grad_norm": 0.9220689472559264, "learning_rate": 4.568590937306021e-05, "loss": 5.0053, "step": 2208 }, { "epoch": 0.13712831336519957, "grad_norm": 0.7311701951128369, "learning_rate": 4.570660045520381e-05, "loss": 4.9999, "step": 2209 }, { "epoch": 0.13719039046495748, "grad_norm": 0.8488849576979157, "learning_rate": 4.57272915373474e-05, "loss": 4.9245, "step": 2210 }, { "epoch": 0.13725246756471537, "grad_norm": 1.0455339867108835, "learning_rate": 4.5747982619491004e-05, "loss": 4.9843, "step": 2211 }, { "epoch": 0.13731454466447326, "grad_norm": 0.9274500682274077, "learning_rate": 4.576867370163459e-05, "loss": 4.8677, "step": 2212 }, { "epoch": 0.13737662176423118, "grad_norm": 1.0565662181555273, "learning_rate": 4.5789364783778194e-05, "loss": 5.0778, "step": 2213 }, { "epoch": 0.13743869886398907, "grad_norm": 1.075767759683168, "learning_rate": 4.581005586592179e-05, "loss": 4.9669, "step": 2214 }, { "epoch": 0.13750077596374696, "grad_norm": 0.9994028215668892, "learning_rate": 4.5830746948065384e-05, "loss": 4.9952, "step": 2215 }, { "epoch": 0.13756285306350488, "grad_norm": 1.1368739490808262, "learning_rate": 4.585143803020898e-05, "loss": 4.8766, "step": 2216 }, { "epoch": 0.13762493016326277, "grad_norm": 0.8392576079604708, "learning_rate": 4.587212911235258e-05, "loss": 4.952, "step": 2217 }, { "epoch": 0.13768700726302066, "grad_norm": 1.168713335492308, "learning_rate": 4.589282019449617e-05, "loss": 4.9265, "step": 2218 }, { "epoch": 0.13774908436277858, "grad_norm": 1.061733361958381, "learning_rate": 4.591351127663977e-05, "loss": 4.8954, "step": 2219 }, { "epoch": 0.13781116146253647, "grad_norm": 1.0441683610342718, "learning_rate": 4.5934202358783367e-05, "loss": 5.0439, "step": 2220 }, { "epoch": 0.13787323856229436, "grad_norm": 1.2869434394441857, "learning_rate": 4.595489344092696e-05, "loss": 5.0229, "step": 2221 }, { "epoch": 0.13793531566205228, "grad_norm": 0.8726431576695262, "learning_rate": 4.597558452307056e-05, "loss": 5.0504, "step": 2222 }, { "epoch": 0.13799739276181017, "grad_norm": 0.9478803390137257, "learning_rate": 4.599627560521416e-05, "loss": 4.8946, "step": 2223 }, { "epoch": 0.13805946986156806, "grad_norm": 1.1526571591148602, "learning_rate": 4.601696668735775e-05, "loss": 4.9547, "step": 2224 }, { "epoch": 0.13812154696132597, "grad_norm": 0.9714105741402465, "learning_rate": 4.603765776950135e-05, "loss": 5.0413, "step": 2225 }, { "epoch": 0.13818362406108387, "grad_norm": 1.067531564857814, "learning_rate": 4.6058348851644944e-05, "loss": 5.0778, "step": 2226 }, { "epoch": 0.13824570116084176, "grad_norm": 1.0677022918282135, "learning_rate": 4.607903993378854e-05, "loss": 4.9374, "step": 2227 }, { "epoch": 0.13830777826059967, "grad_norm": 1.0089113617404648, "learning_rate": 4.6099731015932134e-05, "loss": 4.9197, "step": 2228 }, { "epoch": 0.13836985536035756, "grad_norm": 1.111258613544461, "learning_rate": 4.6120422098075736e-05, "loss": 4.9433, "step": 2229 }, { "epoch": 0.13843193246011545, "grad_norm": 0.9557125426188345, "learning_rate": 4.6141113180219324e-05, "loss": 5.0301, "step": 2230 }, { "epoch": 0.13849400955987337, "grad_norm": 1.1372188241559262, "learning_rate": 4.6161804262362926e-05, "loss": 5.0922, "step": 2231 }, { "epoch": 0.13855608665963126, "grad_norm": 0.9171591902202888, "learning_rate": 4.618249534450652e-05, "loss": 5.0017, "step": 2232 }, { "epoch": 0.13861816375938915, "grad_norm": 0.9500562638637163, "learning_rate": 4.6203186426650116e-05, "loss": 4.9725, "step": 2233 }, { "epoch": 0.13868024085914707, "grad_norm": 0.9015641250463481, "learning_rate": 4.622387750879371e-05, "loss": 5.07, "step": 2234 }, { "epoch": 0.13874231795890496, "grad_norm": 0.9809599478086309, "learning_rate": 4.6244568590937306e-05, "loss": 4.9179, "step": 2235 }, { "epoch": 0.13880439505866285, "grad_norm": 0.9990779679108962, "learning_rate": 4.62652596730809e-05, "loss": 4.9465, "step": 2236 }, { "epoch": 0.13886647215842077, "grad_norm": 0.965256822634566, "learning_rate": 4.6285950755224496e-05, "loss": 4.9369, "step": 2237 }, { "epoch": 0.13892854925817866, "grad_norm": 0.9964142526846884, "learning_rate": 4.63066418373681e-05, "loss": 4.9494, "step": 2238 }, { "epoch": 0.13899062635793655, "grad_norm": 1.0192669639195577, "learning_rate": 4.6327332919511693e-05, "loss": 4.8415, "step": 2239 }, { "epoch": 0.13905270345769447, "grad_norm": 0.9839648648789052, "learning_rate": 4.634802400165529e-05, "loss": 4.9831, "step": 2240 }, { "epoch": 0.13911478055745236, "grad_norm": 1.0660288495745813, "learning_rate": 4.6368715083798884e-05, "loss": 4.9553, "step": 2241 }, { "epoch": 0.13917685765721025, "grad_norm": 0.8834593774429866, "learning_rate": 4.638940616594248e-05, "loss": 4.9396, "step": 2242 }, { "epoch": 0.13923893475696816, "grad_norm": 0.9452115154710264, "learning_rate": 4.6410097248086074e-05, "loss": 4.8984, "step": 2243 }, { "epoch": 0.13930101185672605, "grad_norm": 0.9443983591903206, "learning_rate": 4.6430788330229676e-05, "loss": 4.9385, "step": 2244 }, { "epoch": 0.13936308895648394, "grad_norm": 1.0712806054531732, "learning_rate": 4.6451479412373264e-05, "loss": 4.9812, "step": 2245 }, { "epoch": 0.13942516605624186, "grad_norm": 1.0478999685748154, "learning_rate": 4.6472170494516866e-05, "loss": 4.972, "step": 2246 }, { "epoch": 0.13948724315599975, "grad_norm": 1.0624312210261004, "learning_rate": 4.649286157666046e-05, "loss": 5.0453, "step": 2247 }, { "epoch": 0.13954932025575764, "grad_norm": 1.0101112349792225, "learning_rate": 4.651355265880406e-05, "loss": 4.8932, "step": 2248 }, { "epoch": 0.13961139735551556, "grad_norm": 0.9537022827930695, "learning_rate": 4.653424374094765e-05, "loss": 5.0222, "step": 2249 }, { "epoch": 0.13967347445527345, "grad_norm": 1.0055408751280799, "learning_rate": 4.655493482309125e-05, "loss": 4.842, "step": 2250 }, { "epoch": 0.13973555155503134, "grad_norm": 1.0399818655413695, "learning_rate": 4.657562590523485e-05, "loss": 4.9219, "step": 2251 }, { "epoch": 0.13979762865478926, "grad_norm": 1.047663946500381, "learning_rate": 4.659631698737844e-05, "loss": 4.9108, "step": 2252 }, { "epoch": 0.13985970575454715, "grad_norm": 0.9549050827690866, "learning_rate": 4.661700806952204e-05, "loss": 4.9672, "step": 2253 }, { "epoch": 0.13992178285430504, "grad_norm": 1.0065371701861936, "learning_rate": 4.663769915166564e-05, "loss": 4.9956, "step": 2254 }, { "epoch": 0.13998385995406296, "grad_norm": 1.2348518416303607, "learning_rate": 4.665839023380923e-05, "loss": 4.9607, "step": 2255 }, { "epoch": 0.14004593705382085, "grad_norm": 0.8392544475343257, "learning_rate": 4.667908131595283e-05, "loss": 4.8778, "step": 2256 }, { "epoch": 0.14010801415357874, "grad_norm": 1.1753069152131348, "learning_rate": 4.6699772398096425e-05, "loss": 5.0087, "step": 2257 }, { "epoch": 0.14017009125333665, "grad_norm": 1.0172175319474839, "learning_rate": 4.672046348024002e-05, "loss": 4.9136, "step": 2258 }, { "epoch": 0.14023216835309454, "grad_norm": 0.9928124629696992, "learning_rate": 4.6741154562383615e-05, "loss": 4.9735, "step": 2259 }, { "epoch": 0.14029424545285243, "grad_norm": 1.0908438083232193, "learning_rate": 4.676184564452721e-05, "loss": 4.9815, "step": 2260 }, { "epoch": 0.14035632255261035, "grad_norm": 0.8909122271584315, "learning_rate": 4.6782536726670805e-05, "loss": 4.9867, "step": 2261 }, { "epoch": 0.14041839965236824, "grad_norm": 1.0550104282641157, "learning_rate": 4.68032278088144e-05, "loss": 5.0803, "step": 2262 }, { "epoch": 0.14048047675212613, "grad_norm": 1.1273542777750913, "learning_rate": 4.6823918890958e-05, "loss": 4.9336, "step": 2263 }, { "epoch": 0.14054255385188405, "grad_norm": 1.1169931736244882, "learning_rate": 4.68446099731016e-05, "loss": 4.9181, "step": 2264 }, { "epoch": 0.14060463095164194, "grad_norm": 0.764267632967029, "learning_rate": 4.686530105524519e-05, "loss": 5.0551, "step": 2265 }, { "epoch": 0.14066670805139983, "grad_norm": 0.9132850797000208, "learning_rate": 4.688599213738879e-05, "loss": 4.9205, "step": 2266 }, { "epoch": 0.14072878515115775, "grad_norm": 1.0379760339502322, "learning_rate": 4.690668321953238e-05, "loss": 4.9247, "step": 2267 }, { "epoch": 0.14079086225091564, "grad_norm": 1.1311313124299374, "learning_rate": 4.692737430167598e-05, "loss": 4.9292, "step": 2268 }, { "epoch": 0.14085293935067353, "grad_norm": 0.9086374208380933, "learning_rate": 4.694806538381958e-05, "loss": 4.9375, "step": 2269 }, { "epoch": 0.14091501645043145, "grad_norm": 1.0575272934910838, "learning_rate": 4.696875646596317e-05, "loss": 4.98, "step": 2270 }, { "epoch": 0.14097709355018934, "grad_norm": 1.0448027640932682, "learning_rate": 4.698944754810677e-05, "loss": 4.9424, "step": 2271 }, { "epoch": 0.14103917064994723, "grad_norm": 0.9196527496419565, "learning_rate": 4.7010138630250365e-05, "loss": 4.9858, "step": 2272 }, { "epoch": 0.14110124774970514, "grad_norm": 0.9098567879299667, "learning_rate": 4.703082971239396e-05, "loss": 4.8653, "step": 2273 }, { "epoch": 0.14116332484946303, "grad_norm": 0.9363318795330603, "learning_rate": 4.7051520794537555e-05, "loss": 4.9056, "step": 2274 }, { "epoch": 0.14122540194922092, "grad_norm": 1.1181534545928316, "learning_rate": 4.707221187668116e-05, "loss": 4.8779, "step": 2275 }, { "epoch": 0.14128747904897884, "grad_norm": 0.8853065235220329, "learning_rate": 4.7092902958824745e-05, "loss": 4.9787, "step": 2276 }, { "epoch": 0.14134955614873673, "grad_norm": 1.0468983434095767, "learning_rate": 4.711359404096835e-05, "loss": 4.9695, "step": 2277 }, { "epoch": 0.14141163324849462, "grad_norm": 1.0600510592129173, "learning_rate": 4.713428512311194e-05, "loss": 4.9528, "step": 2278 }, { "epoch": 0.14147371034825254, "grad_norm": 1.0328215559708107, "learning_rate": 4.715497620525554e-05, "loss": 4.9312, "step": 2279 }, { "epoch": 0.14153578744801043, "grad_norm": 1.0072865259512083, "learning_rate": 4.717566728739913e-05, "loss": 4.9522, "step": 2280 }, { "epoch": 0.14159786454776832, "grad_norm": 1.043893240048355, "learning_rate": 4.7196358369542734e-05, "loss": 4.9152, "step": 2281 }, { "epoch": 0.14165994164752624, "grad_norm": 1.1202734627832263, "learning_rate": 4.721704945168632e-05, "loss": 4.9148, "step": 2282 }, { "epoch": 0.14172201874728413, "grad_norm": 1.070753800126448, "learning_rate": 4.7237740533829924e-05, "loss": 4.9482, "step": 2283 }, { "epoch": 0.14178409584704202, "grad_norm": 0.9442576678719432, "learning_rate": 4.725843161597352e-05, "loss": 4.9304, "step": 2284 }, { "epoch": 0.14184617294679994, "grad_norm": 1.0812878861156832, "learning_rate": 4.7279122698117114e-05, "loss": 4.9804, "step": 2285 }, { "epoch": 0.14190825004655783, "grad_norm": 0.9888327043862344, "learning_rate": 4.729981378026071e-05, "loss": 4.9484, "step": 2286 }, { "epoch": 0.14197032714631572, "grad_norm": 1.1063847605203931, "learning_rate": 4.732050486240431e-05, "loss": 4.9457, "step": 2287 }, { "epoch": 0.14203240424607363, "grad_norm": 0.8113806790635173, "learning_rate": 4.73411959445479e-05, "loss": 4.916, "step": 2288 }, { "epoch": 0.14209448134583152, "grad_norm": 0.8484642994057132, "learning_rate": 4.73618870266915e-05, "loss": 4.859, "step": 2289 }, { "epoch": 0.14215655844558941, "grad_norm": 1.0831481457420262, "learning_rate": 4.7382578108835096e-05, "loss": 4.9141, "step": 2290 }, { "epoch": 0.14221863554534733, "grad_norm": 0.9946715730140434, "learning_rate": 4.740326919097869e-05, "loss": 4.9045, "step": 2291 }, { "epoch": 0.14228071264510522, "grad_norm": 1.1306301273295056, "learning_rate": 4.742396027312229e-05, "loss": 4.9343, "step": 2292 }, { "epoch": 0.1423427897448631, "grad_norm": 1.1423304101798781, "learning_rate": 4.744465135526588e-05, "loss": 4.9112, "step": 2293 }, { "epoch": 0.14240486684462103, "grad_norm": 1.1048133443889703, "learning_rate": 4.746534243740948e-05, "loss": 4.8692, "step": 2294 }, { "epoch": 0.14246694394437892, "grad_norm": 1.0938789836578664, "learning_rate": 4.748603351955307e-05, "loss": 4.9872, "step": 2295 }, { "epoch": 0.1425290210441368, "grad_norm": 1.0911515022561507, "learning_rate": 4.7506724601696674e-05, "loss": 4.9628, "step": 2296 }, { "epoch": 0.14259109814389473, "grad_norm": 1.312835332630613, "learning_rate": 4.752741568384026e-05, "loss": 4.849, "step": 2297 }, { "epoch": 0.14265317524365262, "grad_norm": 0.8790010189250123, "learning_rate": 4.7548106765983864e-05, "loss": 4.8761, "step": 2298 }, { "epoch": 0.1427152523434105, "grad_norm": 1.06121012986845, "learning_rate": 4.756879784812746e-05, "loss": 4.9144, "step": 2299 }, { "epoch": 0.14277732944316843, "grad_norm": 1.0742940980321787, "learning_rate": 4.7589488930271054e-05, "loss": 5.0341, "step": 2300 }, { "epoch": 0.14283940654292632, "grad_norm": 1.1845085712894057, "learning_rate": 4.761018001241465e-05, "loss": 4.9955, "step": 2301 }, { "epoch": 0.1429014836426842, "grad_norm": 0.8952529527923881, "learning_rate": 4.763087109455825e-05, "loss": 4.9819, "step": 2302 }, { "epoch": 0.14296356074244213, "grad_norm": 1.1705053845346503, "learning_rate": 4.765156217670184e-05, "loss": 5.0307, "step": 2303 }, { "epoch": 0.14302563784220002, "grad_norm": 1.0577469144743255, "learning_rate": 4.767225325884544e-05, "loss": 4.8457, "step": 2304 }, { "epoch": 0.1430877149419579, "grad_norm": 0.9447802156359363, "learning_rate": 4.7692944340989036e-05, "loss": 4.8927, "step": 2305 }, { "epoch": 0.14314979204171582, "grad_norm": 1.2394736051745916, "learning_rate": 4.771363542313263e-05, "loss": 4.9705, "step": 2306 }, { "epoch": 0.1432118691414737, "grad_norm": 0.9476715476616712, "learning_rate": 4.7734326505276226e-05, "loss": 4.8951, "step": 2307 }, { "epoch": 0.1432739462412316, "grad_norm": 0.9767120551986108, "learning_rate": 4.775501758741983e-05, "loss": 4.9581, "step": 2308 }, { "epoch": 0.14333602334098952, "grad_norm": 0.9834479553423274, "learning_rate": 4.7775708669563416e-05, "loss": 4.8656, "step": 2309 }, { "epoch": 0.1433981004407474, "grad_norm": 1.0180508911999728, "learning_rate": 4.779639975170702e-05, "loss": 4.7698, "step": 2310 }, { "epoch": 0.1434601775405053, "grad_norm": 1.1024652664180137, "learning_rate": 4.7817090833850613e-05, "loss": 4.9392, "step": 2311 }, { "epoch": 0.14352225464026322, "grad_norm": 0.7769929870843765, "learning_rate": 4.783778191599421e-05, "loss": 4.96, "step": 2312 }, { "epoch": 0.1435843317400211, "grad_norm": 0.8889774006909338, "learning_rate": 4.7858472998137804e-05, "loss": 4.9155, "step": 2313 }, { "epoch": 0.143646408839779, "grad_norm": 1.1239037470721416, "learning_rate": 4.7879164080281405e-05, "loss": 4.906, "step": 2314 }, { "epoch": 0.14370848593953692, "grad_norm": 1.0465296275232128, "learning_rate": 4.7899855162424994e-05, "loss": 4.9319, "step": 2315 }, { "epoch": 0.1437705630392948, "grad_norm": 1.0572675442253614, "learning_rate": 4.7920546244568596e-05, "loss": 5.004, "step": 2316 }, { "epoch": 0.1438326401390527, "grad_norm": 1.2729890185749262, "learning_rate": 4.794123732671219e-05, "loss": 4.8988, "step": 2317 }, { "epoch": 0.14389471723881062, "grad_norm": 0.8919437416173156, "learning_rate": 4.7961928408855786e-05, "loss": 4.8252, "step": 2318 }, { "epoch": 0.1439567943385685, "grad_norm": 1.0341629732994833, "learning_rate": 4.798261949099938e-05, "loss": 4.9437, "step": 2319 }, { "epoch": 0.1440188714383264, "grad_norm": 1.3419395493083146, "learning_rate": 4.8003310573142976e-05, "loss": 4.929, "step": 2320 }, { "epoch": 0.1440809485380843, "grad_norm": 0.8633920589809079, "learning_rate": 4.802400165528657e-05, "loss": 4.9583, "step": 2321 }, { "epoch": 0.1441430256378422, "grad_norm": 0.9711783030888476, "learning_rate": 4.8044692737430166e-05, "loss": 4.8823, "step": 2322 }, { "epoch": 0.1442051027376001, "grad_norm": 1.2129152943533383, "learning_rate": 4.806538381957377e-05, "loss": 4.9865, "step": 2323 }, { "epoch": 0.144267179837358, "grad_norm": 0.8423186983810427, "learning_rate": 4.8086074901717356e-05, "loss": 4.9281, "step": 2324 }, { "epoch": 0.1443292569371159, "grad_norm": 1.0412367823793371, "learning_rate": 4.810676598386096e-05, "loss": 4.785, "step": 2325 }, { "epoch": 0.1443913340368738, "grad_norm": 1.0155434795821467, "learning_rate": 4.812745706600455e-05, "loss": 4.8896, "step": 2326 }, { "epoch": 0.1444534111366317, "grad_norm": 1.0560227595463128, "learning_rate": 4.814814814814815e-05, "loss": 4.9869, "step": 2327 }, { "epoch": 0.1445154882363896, "grad_norm": 0.9918315879873941, "learning_rate": 4.816883923029174e-05, "loss": 4.9914, "step": 2328 }, { "epoch": 0.1445775653361475, "grad_norm": 1.071441055406774, "learning_rate": 4.8189530312435345e-05, "loss": 4.8886, "step": 2329 }, { "epoch": 0.1446396424359054, "grad_norm": 1.204864384075389, "learning_rate": 4.8210221394578933e-05, "loss": 4.9375, "step": 2330 }, { "epoch": 0.1447017195356633, "grad_norm": 0.8064734260239341, "learning_rate": 4.8230912476722535e-05, "loss": 4.8737, "step": 2331 }, { "epoch": 0.1447637966354212, "grad_norm": 1.2360772093697032, "learning_rate": 4.825160355886613e-05, "loss": 4.8428, "step": 2332 }, { "epoch": 0.1448258737351791, "grad_norm": 0.9790758368116145, "learning_rate": 4.827229464100973e-05, "loss": 4.8212, "step": 2333 }, { "epoch": 0.144887950834937, "grad_norm": 1.0969492308449236, "learning_rate": 4.829298572315332e-05, "loss": 4.841, "step": 2334 }, { "epoch": 0.14495002793469489, "grad_norm": 1.0545762745171228, "learning_rate": 4.831367680529692e-05, "loss": 4.8855, "step": 2335 }, { "epoch": 0.1450121050344528, "grad_norm": 1.0983456113319103, "learning_rate": 4.833436788744052e-05, "loss": 4.8229, "step": 2336 }, { "epoch": 0.1450741821342107, "grad_norm": 0.8210066628481961, "learning_rate": 4.835505896958411e-05, "loss": 4.7758, "step": 2337 }, { "epoch": 0.14513625923396858, "grad_norm": 0.8925666057085422, "learning_rate": 4.837575005172771e-05, "loss": 4.8658, "step": 2338 }, { "epoch": 0.1451983363337265, "grad_norm": 1.1707965211522713, "learning_rate": 4.839644113387131e-05, "loss": 4.8898, "step": 2339 }, { "epoch": 0.1452604134334844, "grad_norm": 1.0018900652518428, "learning_rate": 4.84171322160149e-05, "loss": 4.9347, "step": 2340 }, { "epoch": 0.14532249053324228, "grad_norm": 1.1268159197713576, "learning_rate": 4.84378232981585e-05, "loss": 4.9266, "step": 2341 }, { "epoch": 0.14538456763300017, "grad_norm": 1.1472140691886732, "learning_rate": 4.8458514380302095e-05, "loss": 4.8957, "step": 2342 }, { "epoch": 0.1454466447327581, "grad_norm": 1.0818063109666751, "learning_rate": 4.847920546244569e-05, "loss": 4.8826, "step": 2343 }, { "epoch": 0.14550872183251598, "grad_norm": 1.1585860495848388, "learning_rate": 4.8499896544589285e-05, "loss": 4.8677, "step": 2344 }, { "epoch": 0.14557079893227387, "grad_norm": 1.2731863136864614, "learning_rate": 4.852058762673288e-05, "loss": 4.8613, "step": 2345 }, { "epoch": 0.1456328760320318, "grad_norm": 1.1310889977804415, "learning_rate": 4.8541278708876475e-05, "loss": 4.9695, "step": 2346 }, { "epoch": 0.14569495313178968, "grad_norm": 1.3484605683334907, "learning_rate": 4.856196979102007e-05, "loss": 4.8724, "step": 2347 }, { "epoch": 0.14575703023154757, "grad_norm": 0.7752641627578174, "learning_rate": 4.858266087316367e-05, "loss": 4.8424, "step": 2348 }, { "epoch": 0.1458191073313055, "grad_norm": 1.1356704439970555, "learning_rate": 4.860335195530727e-05, "loss": 4.886, "step": 2349 }, { "epoch": 0.14588118443106338, "grad_norm": 1.1305166723379276, "learning_rate": 4.862404303745086e-05, "loss": 4.7297, "step": 2350 }, { "epoch": 0.14594326153082127, "grad_norm": 0.9668859267403377, "learning_rate": 4.864473411959446e-05, "loss": 4.8412, "step": 2351 }, { "epoch": 0.14600533863057918, "grad_norm": 0.9275335154685813, "learning_rate": 4.866542520173805e-05, "loss": 4.8241, "step": 2352 }, { "epoch": 0.14606741573033707, "grad_norm": 0.9777208096412441, "learning_rate": 4.868611628388165e-05, "loss": 4.8387, "step": 2353 }, { "epoch": 0.14612949283009496, "grad_norm": 1.1219546637767925, "learning_rate": 4.870680736602525e-05, "loss": 4.9247, "step": 2354 }, { "epoch": 0.14619156992985288, "grad_norm": 1.0908129357999512, "learning_rate": 4.872749844816884e-05, "loss": 4.8283, "step": 2355 }, { "epoch": 0.14625364702961077, "grad_norm": 1.0935096476634676, "learning_rate": 4.874818953031244e-05, "loss": 4.8803, "step": 2356 }, { "epoch": 0.14631572412936866, "grad_norm": 0.9968496110128503, "learning_rate": 4.8768880612456034e-05, "loss": 4.8929, "step": 2357 }, { "epoch": 0.14637780122912658, "grad_norm": 0.9367891477677011, "learning_rate": 4.878957169459963e-05, "loss": 4.7385, "step": 2358 }, { "epoch": 0.14643987832888447, "grad_norm": 0.8543720967509258, "learning_rate": 4.8810262776743225e-05, "loss": 4.8163, "step": 2359 }, { "epoch": 0.14650195542864236, "grad_norm": 0.9366577179459786, "learning_rate": 4.8830953858886826e-05, "loss": 4.816, "step": 2360 }, { "epoch": 0.14656403252840028, "grad_norm": 1.0256575173494844, "learning_rate": 4.8851644941030415e-05, "loss": 4.7929, "step": 2361 }, { "epoch": 0.14662610962815817, "grad_norm": 1.0268888445412552, "learning_rate": 4.8872336023174017e-05, "loss": 4.9356, "step": 2362 }, { "epoch": 0.14668818672791606, "grad_norm": 1.231371884195012, "learning_rate": 4.889302710531761e-05, "loss": 4.8318, "step": 2363 }, { "epoch": 0.14675026382767398, "grad_norm": 0.9413564201874792, "learning_rate": 4.891371818746121e-05, "loss": 4.87, "step": 2364 }, { "epoch": 0.14681234092743187, "grad_norm": 0.9105845086410235, "learning_rate": 4.89344092696048e-05, "loss": 4.8166, "step": 2365 }, { "epoch": 0.14687441802718976, "grad_norm": 0.8850404762990519, "learning_rate": 4.8955100351748404e-05, "loss": 4.8148, "step": 2366 }, { "epoch": 0.14693649512694767, "grad_norm": 0.8165047326076356, "learning_rate": 4.897579143389199e-05, "loss": 4.8379, "step": 2367 }, { "epoch": 0.14699857222670557, "grad_norm": 0.9852859278615669, "learning_rate": 4.8996482516035594e-05, "loss": 4.8126, "step": 2368 }, { "epoch": 0.14706064932646346, "grad_norm": 1.0472483751981492, "learning_rate": 4.901717359817919e-05, "loss": 4.8249, "step": 2369 }, { "epoch": 0.14712272642622137, "grad_norm": 1.0573218894109087, "learning_rate": 4.9037864680322784e-05, "loss": 4.8558, "step": 2370 }, { "epoch": 0.14718480352597926, "grad_norm": 1.207385557449806, "learning_rate": 4.905855576246638e-05, "loss": 4.8748, "step": 2371 }, { "epoch": 0.14724688062573715, "grad_norm": 1.199702857898724, "learning_rate": 4.9079246844609974e-05, "loss": 4.9809, "step": 2372 }, { "epoch": 0.14730895772549507, "grad_norm": 0.9539479002111653, "learning_rate": 4.909993792675357e-05, "loss": 4.9407, "step": 2373 }, { "epoch": 0.14737103482525296, "grad_norm": 1.1343494906328953, "learning_rate": 4.912062900889717e-05, "loss": 4.907, "step": 2374 }, { "epoch": 0.14743311192501085, "grad_norm": 1.0277722991476583, "learning_rate": 4.9141320091040766e-05, "loss": 4.9632, "step": 2375 }, { "epoch": 0.14749518902476877, "grad_norm": 1.1891548772511296, "learning_rate": 4.916201117318436e-05, "loss": 4.7684, "step": 2376 }, { "epoch": 0.14755726612452666, "grad_norm": 0.8772500264082714, "learning_rate": 4.9182702255327956e-05, "loss": 4.8224, "step": 2377 }, { "epoch": 0.14761934322428455, "grad_norm": 1.1152273201260001, "learning_rate": 4.920339333747155e-05, "loss": 4.9884, "step": 2378 }, { "epoch": 0.14768142032404247, "grad_norm": 1.1266024091743918, "learning_rate": 4.9224084419615146e-05, "loss": 4.9684, "step": 2379 }, { "epoch": 0.14774349742380036, "grad_norm": 1.0928406886456885, "learning_rate": 4.924477550175874e-05, "loss": 4.9532, "step": 2380 }, { "epoch": 0.14780557452355825, "grad_norm": 1.0790507999407832, "learning_rate": 4.926546658390234e-05, "loss": 4.7579, "step": 2381 }, { "epoch": 0.14786765162331617, "grad_norm": 1.0102281820673682, "learning_rate": 4.928615766604593e-05, "loss": 4.8112, "step": 2382 }, { "epoch": 0.14792972872307406, "grad_norm": 0.9812333063054095, "learning_rate": 4.9306848748189533e-05, "loss": 4.8169, "step": 2383 }, { "epoch": 0.14799180582283195, "grad_norm": 1.1651550388726346, "learning_rate": 4.932753983033313e-05, "loss": 4.9244, "step": 2384 }, { "epoch": 0.14805388292258986, "grad_norm": 1.1873418301098575, "learning_rate": 4.9348230912476724e-05, "loss": 4.7273, "step": 2385 }, { "epoch": 0.14811596002234775, "grad_norm": 1.0135307434267071, "learning_rate": 4.936892199462032e-05, "loss": 4.7816, "step": 2386 }, { "epoch": 0.14817803712210564, "grad_norm": 1.2145315179955813, "learning_rate": 4.938961307676392e-05, "loss": 4.8263, "step": 2387 }, { "epoch": 0.14824011422186356, "grad_norm": 1.093773487383835, "learning_rate": 4.941030415890751e-05, "loss": 4.9118, "step": 2388 }, { "epoch": 0.14830219132162145, "grad_norm": 0.8671760558326895, "learning_rate": 4.943099524105111e-05, "loss": 4.835, "step": 2389 }, { "epoch": 0.14836426842137934, "grad_norm": 0.9131145941985412, "learning_rate": 4.9451686323194706e-05, "loss": 4.837, "step": 2390 }, { "epoch": 0.14842634552113726, "grad_norm": 0.9619094771067739, "learning_rate": 4.94723774053383e-05, "loss": 4.9298, "step": 2391 }, { "epoch": 0.14848842262089515, "grad_norm": 1.2374073398463035, "learning_rate": 4.9493068487481896e-05, "loss": 4.9151, "step": 2392 }, { "epoch": 0.14855049972065304, "grad_norm": 1.0903927093855463, "learning_rate": 4.95137595696255e-05, "loss": 4.8198, "step": 2393 }, { "epoch": 0.14861257682041096, "grad_norm": 1.1340589308925628, "learning_rate": 4.9534450651769086e-05, "loss": 4.7935, "step": 2394 }, { "epoch": 0.14867465392016885, "grad_norm": 1.0455120950999879, "learning_rate": 4.955514173391269e-05, "loss": 4.7577, "step": 2395 }, { "epoch": 0.14873673101992674, "grad_norm": 0.951598461455023, "learning_rate": 4.957583281605628e-05, "loss": 4.8082, "step": 2396 }, { "epoch": 0.14879880811968466, "grad_norm": 0.9303418909772946, "learning_rate": 4.959652389819988e-05, "loss": 4.7649, "step": 2397 }, { "epoch": 0.14886088521944255, "grad_norm": 0.9805976813299631, "learning_rate": 4.961721498034347e-05, "loss": 4.8924, "step": 2398 }, { "epoch": 0.14892296231920044, "grad_norm": 0.9644715676257405, "learning_rate": 4.9637906062487075e-05, "loss": 4.8036, "step": 2399 }, { "epoch": 0.14898503941895835, "grad_norm": 1.0420665482047926, "learning_rate": 4.965859714463066e-05, "loss": 4.6818, "step": 2400 }, { "epoch": 0.14904711651871624, "grad_norm": 0.9960694852379274, "learning_rate": 4.9679288226774265e-05, "loss": 4.7161, "step": 2401 }, { "epoch": 0.14910919361847413, "grad_norm": 0.9318001408638472, "learning_rate": 4.969997930891786e-05, "loss": 4.8441, "step": 2402 }, { "epoch": 0.14917127071823205, "grad_norm": 1.1377609037557013, "learning_rate": 4.9720670391061455e-05, "loss": 4.8351, "step": 2403 }, { "epoch": 0.14923334781798994, "grad_norm": 0.9707501590010488, "learning_rate": 4.974136147320505e-05, "loss": 4.8665, "step": 2404 }, { "epoch": 0.14929542491774783, "grad_norm": 0.9942335523780086, "learning_rate": 4.9762052555348645e-05, "loss": 4.7345, "step": 2405 }, { "epoch": 0.14935750201750575, "grad_norm": 1.0331891641232473, "learning_rate": 4.978274363749224e-05, "loss": 4.8223, "step": 2406 }, { "epoch": 0.14941957911726364, "grad_norm": 1.3008223976752473, "learning_rate": 4.9803434719635836e-05, "loss": 4.8007, "step": 2407 }, { "epoch": 0.14948165621702153, "grad_norm": 0.838471395892938, "learning_rate": 4.982412580177944e-05, "loss": 4.794, "step": 2408 }, { "epoch": 0.14954373331677945, "grad_norm": 1.0729887422715882, "learning_rate": 4.9844816883923026e-05, "loss": 4.7453, "step": 2409 }, { "epoch": 0.14960581041653734, "grad_norm": 1.1972299791415693, "learning_rate": 4.986550796606663e-05, "loss": 4.8008, "step": 2410 }, { "epoch": 0.14966788751629523, "grad_norm": 0.796708931167586, "learning_rate": 4.988619904821022e-05, "loss": 4.8476, "step": 2411 }, { "epoch": 0.14972996461605315, "grad_norm": 1.0382589230171202, "learning_rate": 4.990689013035382e-05, "loss": 4.7755, "step": 2412 }, { "epoch": 0.14979204171581104, "grad_norm": 1.206025133272757, "learning_rate": 4.992758121249741e-05, "loss": 4.7115, "step": 2413 }, { "epoch": 0.14985411881556893, "grad_norm": 0.8279983495635939, "learning_rate": 4.9948272294641015e-05, "loss": 4.9264, "step": 2414 }, { "epoch": 0.14991619591532684, "grad_norm": 1.0522749257397115, "learning_rate": 4.99689633767846e-05, "loss": 4.7404, "step": 2415 }, { "epoch": 0.14997827301508473, "grad_norm": 1.1409064690270672, "learning_rate": 4.9989654458928205e-05, "loss": 4.7462, "step": 2416 }, { "epoch": 0.15004035011484262, "grad_norm": 0.9465713631034589, "learning_rate": 5.00103455410718e-05, "loss": 4.6525, "step": 2417 }, { "epoch": 0.15010242721460054, "grad_norm": 1.1631608069318222, "learning_rate": 5.00310366232154e-05, "loss": 4.6364, "step": 2418 }, { "epoch": 0.15016450431435843, "grad_norm": 0.969150614196885, "learning_rate": 5.0051727705359e-05, "loss": 4.9927, "step": 2419 }, { "epoch": 0.15022658141411632, "grad_norm": 1.1344252583112917, "learning_rate": 5.0072418787502585e-05, "loss": 4.7583, "step": 2420 }, { "epoch": 0.15028865851387424, "grad_norm": 1.0600400946789539, "learning_rate": 5.009310986964618e-05, "loss": 4.8573, "step": 2421 }, { "epoch": 0.15035073561363213, "grad_norm": 0.9200003685639514, "learning_rate": 5.011380095178978e-05, "loss": 4.6914, "step": 2422 }, { "epoch": 0.15041281271339002, "grad_norm": 1.010980697723017, "learning_rate": 5.013449203393338e-05, "loss": 4.7943, "step": 2423 }, { "epoch": 0.15047488981314794, "grad_norm": 1.0171310995143366, "learning_rate": 5.015518311607698e-05, "loss": 4.74, "step": 2424 }, { "epoch": 0.15053696691290583, "grad_norm": 0.9960023725895026, "learning_rate": 5.0175874198220574e-05, "loss": 4.7347, "step": 2425 }, { "epoch": 0.15059904401266372, "grad_norm": 0.960104418380642, "learning_rate": 5.019656528036416e-05, "loss": 4.6544, "step": 2426 }, { "epoch": 0.15066112111242164, "grad_norm": 0.9935696204240434, "learning_rate": 5.021725636250776e-05, "loss": 4.7647, "step": 2427 }, { "epoch": 0.15072319821217953, "grad_norm": 1.0450630449141114, "learning_rate": 5.023794744465136e-05, "loss": 4.7971, "step": 2428 }, { "epoch": 0.15078527531193742, "grad_norm": 1.0452193077524559, "learning_rate": 5.0258638526794954e-05, "loss": 4.8409, "step": 2429 }, { "epoch": 0.15084735241169533, "grad_norm": 1.0843204011830982, "learning_rate": 5.027932960893855e-05, "loss": 4.8751, "step": 2430 }, { "epoch": 0.15090942951145322, "grad_norm": 1.1491279343146708, "learning_rate": 5.030002069108215e-05, "loss": 4.7841, "step": 2431 }, { "epoch": 0.15097150661121111, "grad_norm": 0.8346990877444705, "learning_rate": 5.032071177322574e-05, "loss": 4.7674, "step": 2432 }, { "epoch": 0.15103358371096903, "grad_norm": 0.8587904589365992, "learning_rate": 5.0341402855369335e-05, "loss": 4.6809, "step": 2433 }, { "epoch": 0.15109566081072692, "grad_norm": 0.9460446698588374, "learning_rate": 5.036209393751293e-05, "loss": 4.8395, "step": 2434 }, { "epoch": 0.1511577379104848, "grad_norm": 1.0656953894672314, "learning_rate": 5.038278501965653e-05, "loss": 4.8285, "step": 2435 }, { "epoch": 0.15121981501024273, "grad_norm": 1.0228774307298547, "learning_rate": 5.040347610180013e-05, "loss": 4.9091, "step": 2436 }, { "epoch": 0.15128189211000062, "grad_norm": 1.0384199407185228, "learning_rate": 5.042416718394373e-05, "loss": 4.8008, "step": 2437 }, { "epoch": 0.1513439692097585, "grad_norm": 1.1489222016220002, "learning_rate": 5.044485826608732e-05, "loss": 4.7208, "step": 2438 }, { "epoch": 0.15140604630951643, "grad_norm": 0.9882869795682172, "learning_rate": 5.046554934823091e-05, "loss": 4.8213, "step": 2439 }, { "epoch": 0.15146812340927432, "grad_norm": 1.1391485927369325, "learning_rate": 5.048624043037451e-05, "loss": 4.7931, "step": 2440 }, { "epoch": 0.1515302005090322, "grad_norm": 1.0671586955750503, "learning_rate": 5.050693151251811e-05, "loss": 4.8246, "step": 2441 }, { "epoch": 0.15159227760879013, "grad_norm": 1.1901204659690428, "learning_rate": 5.0527622594661704e-05, "loss": 4.8276, "step": 2442 }, { "epoch": 0.15165435470854802, "grad_norm": 1.1270539818759522, "learning_rate": 5.0548313676805306e-05, "loss": 4.8276, "step": 2443 }, { "epoch": 0.1517164318083059, "grad_norm": 1.2246271402874331, "learning_rate": 5.056900475894889e-05, "loss": 4.7573, "step": 2444 }, { "epoch": 0.15177850890806383, "grad_norm": 1.010591858819979, "learning_rate": 5.058969584109249e-05, "loss": 4.6929, "step": 2445 }, { "epoch": 0.15184058600782172, "grad_norm": 1.0063390680545417, "learning_rate": 5.0610386923236084e-05, "loss": 4.7709, "step": 2446 }, { "epoch": 0.1519026631075796, "grad_norm": 0.9136331642173362, "learning_rate": 5.0631078005379686e-05, "loss": 4.7698, "step": 2447 }, { "epoch": 0.15196474020733752, "grad_norm": 1.0018260495988238, "learning_rate": 5.065176908752328e-05, "loss": 4.7798, "step": 2448 }, { "epoch": 0.1520268173070954, "grad_norm": 1.473453134107928, "learning_rate": 5.067246016966688e-05, "loss": 4.7583, "step": 2449 }, { "epoch": 0.1520888944068533, "grad_norm": 1.0289242267124417, "learning_rate": 5.0693151251810465e-05, "loss": 4.6997, "step": 2450 }, { "epoch": 0.15215097150661122, "grad_norm": 0.9717877045214127, "learning_rate": 5.0713842333954066e-05, "loss": 4.7272, "step": 2451 }, { "epoch": 0.1522130486063691, "grad_norm": 1.123553883091026, "learning_rate": 5.073453341609766e-05, "loss": 4.7748, "step": 2452 }, { "epoch": 0.152275125706127, "grad_norm": 1.004453992033999, "learning_rate": 5.075522449824126e-05, "loss": 4.6891, "step": 2453 }, { "epoch": 0.15233720280588492, "grad_norm": 1.6147587178454588, "learning_rate": 5.077591558038486e-05, "loss": 4.738, "step": 2454 }, { "epoch": 0.1523992799056428, "grad_norm": 0.914466018944348, "learning_rate": 5.0796606662528453e-05, "loss": 4.6734, "step": 2455 }, { "epoch": 0.1524613570054007, "grad_norm": 1.1437654504618358, "learning_rate": 5.081729774467204e-05, "loss": 4.7019, "step": 2456 }, { "epoch": 0.15252343410515862, "grad_norm": 0.9014608183696416, "learning_rate": 5.0837988826815644e-05, "loss": 4.8093, "step": 2457 }, { "epoch": 0.1525855112049165, "grad_norm": 1.0576443514589267, "learning_rate": 5.085867990895924e-05, "loss": 4.7289, "step": 2458 }, { "epoch": 0.1526475883046744, "grad_norm": 1.1033886861793971, "learning_rate": 5.0879370991102834e-05, "loss": 4.6923, "step": 2459 }, { "epoch": 0.15270966540443232, "grad_norm": 0.9502812471177453, "learning_rate": 5.0900062073246436e-05, "loss": 4.7915, "step": 2460 }, { "epoch": 0.1527717425041902, "grad_norm": 1.1174733789844626, "learning_rate": 5.092075315539003e-05, "loss": 4.8192, "step": 2461 }, { "epoch": 0.1528338196039481, "grad_norm": 1.1445589029662535, "learning_rate": 5.094144423753363e-05, "loss": 4.8114, "step": 2462 }, { "epoch": 0.152895896703706, "grad_norm": 0.9716830919675755, "learning_rate": 5.096213531967722e-05, "loss": 4.7571, "step": 2463 }, { "epoch": 0.1529579738034639, "grad_norm": 0.9694578599472181, "learning_rate": 5.0982826401820816e-05, "loss": 4.6145, "step": 2464 }, { "epoch": 0.1530200509032218, "grad_norm": 1.0959966201869336, "learning_rate": 5.100351748396441e-05, "loss": 4.6556, "step": 2465 }, { "epoch": 0.1530821280029797, "grad_norm": 1.0294681529621288, "learning_rate": 5.102420856610801e-05, "loss": 4.7494, "step": 2466 }, { "epoch": 0.1531442051027376, "grad_norm": 0.8854596108886651, "learning_rate": 5.104489964825161e-05, "loss": 4.8331, "step": 2467 }, { "epoch": 0.1532062822024955, "grad_norm": 1.0378613520619315, "learning_rate": 5.106559073039521e-05, "loss": 4.7236, "step": 2468 }, { "epoch": 0.1532683593022534, "grad_norm": 1.2119998824651428, "learning_rate": 5.108628181253879e-05, "loss": 4.7884, "step": 2469 }, { "epoch": 0.1533304364020113, "grad_norm": 0.995160342432832, "learning_rate": 5.110697289468239e-05, "loss": 4.7866, "step": 2470 }, { "epoch": 0.1533925135017692, "grad_norm": 1.2644221865663967, "learning_rate": 5.112766397682599e-05, "loss": 4.8146, "step": 2471 }, { "epoch": 0.1534545906015271, "grad_norm": 0.9919446813268019, "learning_rate": 5.114835505896959e-05, "loss": 4.726, "step": 2472 }, { "epoch": 0.153516667701285, "grad_norm": 1.1478366967030975, "learning_rate": 5.1169046141113185e-05, "loss": 4.8354, "step": 2473 }, { "epoch": 0.1535787448010429, "grad_norm": 0.9487854396347741, "learning_rate": 5.118973722325679e-05, "loss": 4.711, "step": 2474 }, { "epoch": 0.1536408219008008, "grad_norm": 0.9074856785361406, "learning_rate": 5.121042830540037e-05, "loss": 4.7062, "step": 2475 }, { "epoch": 0.1537028990005587, "grad_norm": 1.0585268569588002, "learning_rate": 5.123111938754397e-05, "loss": 4.7721, "step": 2476 }, { "epoch": 0.15376497610031659, "grad_norm": 0.9336544320404528, "learning_rate": 5.1251810469687566e-05, "loss": 4.7271, "step": 2477 }, { "epoch": 0.1538270532000745, "grad_norm": 0.8547342360862359, "learning_rate": 5.127250155183117e-05, "loss": 4.8051, "step": 2478 }, { "epoch": 0.1538891302998324, "grad_norm": 0.9922502753155731, "learning_rate": 5.129319263397476e-05, "loss": 4.7738, "step": 2479 }, { "epoch": 0.15395120739959028, "grad_norm": 1.072152609506214, "learning_rate": 5.131388371611836e-05, "loss": 4.8173, "step": 2480 }, { "epoch": 0.1540132844993482, "grad_norm": 1.1179050195470885, "learning_rate": 5.1334574798261946e-05, "loss": 4.7469, "step": 2481 }, { "epoch": 0.1540753615991061, "grad_norm": 1.0320391669825202, "learning_rate": 5.135526588040555e-05, "loss": 4.7616, "step": 2482 }, { "epoch": 0.15413743869886398, "grad_norm": 0.8746051237915362, "learning_rate": 5.137595696254914e-05, "loss": 4.8124, "step": 2483 }, { "epoch": 0.1541995157986219, "grad_norm": 1.0674316730117985, "learning_rate": 5.139664804469274e-05, "loss": 4.6199, "step": 2484 }, { "epoch": 0.1542615928983798, "grad_norm": 1.2887864250711334, "learning_rate": 5.141733912683634e-05, "loss": 4.5966, "step": 2485 }, { "epoch": 0.15432366999813768, "grad_norm": 0.9387068292314313, "learning_rate": 5.1438030208979935e-05, "loss": 4.7556, "step": 2486 }, { "epoch": 0.1543857470978956, "grad_norm": 0.9462575365922898, "learning_rate": 5.145872129112352e-05, "loss": 4.6988, "step": 2487 }, { "epoch": 0.1544478241976535, "grad_norm": 1.0524585482845576, "learning_rate": 5.1479412373267125e-05, "loss": 4.6989, "step": 2488 }, { "epoch": 0.15450990129741138, "grad_norm": 1.000430610450467, "learning_rate": 5.150010345541072e-05, "loss": 4.64, "step": 2489 }, { "epoch": 0.1545719783971693, "grad_norm": 1.1843862366961087, "learning_rate": 5.1520794537554315e-05, "loss": 4.7552, "step": 2490 }, { "epoch": 0.1546340554969272, "grad_norm": 1.1293391153517083, "learning_rate": 5.154148561969792e-05, "loss": 4.7861, "step": 2491 }, { "epoch": 0.15469613259668508, "grad_norm": 1.043749570560938, "learning_rate": 5.156217670184151e-05, "loss": 4.7624, "step": 2492 }, { "epoch": 0.154758209696443, "grad_norm": 1.1059119747996442, "learning_rate": 5.15828677839851e-05, "loss": 4.8628, "step": 2493 }, { "epoch": 0.15482028679620088, "grad_norm": 0.8944604504148939, "learning_rate": 5.1603558866128695e-05, "loss": 4.8011, "step": 2494 }, { "epoch": 0.15488236389595877, "grad_norm": 0.9143146264984187, "learning_rate": 5.16242499482723e-05, "loss": 4.6863, "step": 2495 }, { "epoch": 0.1549444409957167, "grad_norm": 0.8750407005108428, "learning_rate": 5.164494103041589e-05, "loss": 4.7554, "step": 2496 }, { "epoch": 0.15500651809547458, "grad_norm": 1.257713550916769, "learning_rate": 5.1665632112559494e-05, "loss": 4.7639, "step": 2497 }, { "epoch": 0.15506859519523247, "grad_norm": 0.8300584499046911, "learning_rate": 5.168632319470309e-05, "loss": 4.7554, "step": 2498 }, { "epoch": 0.1551306722949904, "grad_norm": 0.9962287063428981, "learning_rate": 5.170701427684668e-05, "loss": 4.7191, "step": 2499 }, { "epoch": 0.15519274939474828, "grad_norm": 1.113315247133725, "learning_rate": 5.172770535899027e-05, "loss": 4.7579, "step": 2500 }, { "epoch": 0.15525482649450617, "grad_norm": 0.8731019346212036, "learning_rate": 5.1748396441133874e-05, "loss": 4.5951, "step": 2501 }, { "epoch": 0.1553169035942641, "grad_norm": 0.9654751460759773, "learning_rate": 5.176908752327747e-05, "loss": 4.7368, "step": 2502 }, { "epoch": 0.15537898069402198, "grad_norm": 0.980903103733665, "learning_rate": 5.178977860542107e-05, "loss": 4.7301, "step": 2503 }, { "epoch": 0.15544105779377987, "grad_norm": 0.9625903855032077, "learning_rate": 5.1810469687564666e-05, "loss": 4.7803, "step": 2504 }, { "epoch": 0.1555031348935378, "grad_norm": 1.0854357651234114, "learning_rate": 5.1831160769708255e-05, "loss": 4.7688, "step": 2505 }, { "epoch": 0.15556521199329568, "grad_norm": 1.0857048059798964, "learning_rate": 5.185185185185185e-05, "loss": 4.8796, "step": 2506 }, { "epoch": 0.15562728909305357, "grad_norm": 0.9639282207239016, "learning_rate": 5.187254293399545e-05, "loss": 4.645, "step": 2507 }, { "epoch": 0.15568936619281148, "grad_norm": 1.025204654359615, "learning_rate": 5.189323401613905e-05, "loss": 4.7329, "step": 2508 }, { "epoch": 0.15575144329256937, "grad_norm": 1.0428733639192738, "learning_rate": 5.191392509828265e-05, "loss": 4.6318, "step": 2509 }, { "epoch": 0.15581352039232726, "grad_norm": 0.974387103403808, "learning_rate": 5.1934616180426244e-05, "loss": 4.6871, "step": 2510 }, { "epoch": 0.15587559749208518, "grad_norm": 0.9853146019649135, "learning_rate": 5.195530726256983e-05, "loss": 4.6346, "step": 2511 }, { "epoch": 0.15593767459184307, "grad_norm": 1.114078651086439, "learning_rate": 5.197599834471343e-05, "loss": 4.6994, "step": 2512 }, { "epoch": 0.15599975169160096, "grad_norm": 1.020706902025069, "learning_rate": 5.199668942685703e-05, "loss": 4.7261, "step": 2513 }, { "epoch": 0.15606182879135888, "grad_norm": 1.2712007783496075, "learning_rate": 5.2017380509000624e-05, "loss": 4.8021, "step": 2514 }, { "epoch": 0.15612390589111677, "grad_norm": 1.0642732520759077, "learning_rate": 5.203807159114422e-05, "loss": 4.7637, "step": 2515 }, { "epoch": 0.15618598299087466, "grad_norm": 1.111972454446848, "learning_rate": 5.205876267328782e-05, "loss": 4.5918, "step": 2516 }, { "epoch": 0.15624806009063258, "grad_norm": 0.8700297404829324, "learning_rate": 5.207945375543141e-05, "loss": 4.7203, "step": 2517 }, { "epoch": 0.15631013719039047, "grad_norm": 1.105648174485079, "learning_rate": 5.2100144837575004e-05, "loss": 4.7656, "step": 2518 }, { "epoch": 0.15637221429014836, "grad_norm": 1.1250904894569826, "learning_rate": 5.21208359197186e-05, "loss": 4.7915, "step": 2519 }, { "epoch": 0.15643429138990628, "grad_norm": 0.9652387605796736, "learning_rate": 5.21415270018622e-05, "loss": 4.6972, "step": 2520 }, { "epoch": 0.15649636848966417, "grad_norm": 1.0707906504237474, "learning_rate": 5.2162218084005796e-05, "loss": 4.7042, "step": 2521 }, { "epoch": 0.15655844558942206, "grad_norm": 1.1939369215093965, "learning_rate": 5.21829091661494e-05, "loss": 4.8117, "step": 2522 }, { "epoch": 0.15662052268917998, "grad_norm": 1.040518522113297, "learning_rate": 5.2203600248292986e-05, "loss": 4.6421, "step": 2523 }, { "epoch": 0.15668259978893787, "grad_norm": 1.0888714572155638, "learning_rate": 5.222429133043658e-05, "loss": 4.7687, "step": 2524 }, { "epoch": 0.15674467688869576, "grad_norm": 1.0781852041932873, "learning_rate": 5.2244982412580177e-05, "loss": 4.7209, "step": 2525 }, { "epoch": 0.15680675398845367, "grad_norm": 1.2530406685704059, "learning_rate": 5.226567349472378e-05, "loss": 4.6229, "step": 2526 }, { "epoch": 0.15686883108821156, "grad_norm": 1.077639318341195, "learning_rate": 5.2286364576867374e-05, "loss": 4.7042, "step": 2527 }, { "epoch": 0.15693090818796945, "grad_norm": 0.9713109836064566, "learning_rate": 5.2307055659010975e-05, "loss": 4.664, "step": 2528 }, { "epoch": 0.15699298528772734, "grad_norm": 1.0189217100943402, "learning_rate": 5.232774674115456e-05, "loss": 4.6952, "step": 2529 }, { "epoch": 0.15705506238748526, "grad_norm": 1.0700099545309896, "learning_rate": 5.234843782329816e-05, "loss": 4.7266, "step": 2530 }, { "epoch": 0.15711713948724315, "grad_norm": 1.2068300048563771, "learning_rate": 5.2369128905441754e-05, "loss": 4.8248, "step": 2531 }, { "epoch": 0.15717921658700104, "grad_norm": 1.3009498189122515, "learning_rate": 5.2389819987585356e-05, "loss": 4.7497, "step": 2532 }, { "epoch": 0.15724129368675896, "grad_norm": 1.0354156744867147, "learning_rate": 5.241051106972895e-05, "loss": 4.7191, "step": 2533 }, { "epoch": 0.15730337078651685, "grad_norm": 1.1473957449066965, "learning_rate": 5.243120215187255e-05, "loss": 4.7092, "step": 2534 }, { "epoch": 0.15736544788627474, "grad_norm": 0.9154511035964, "learning_rate": 5.2451893234016134e-05, "loss": 4.647, "step": 2535 }, { "epoch": 0.15742752498603266, "grad_norm": 1.2433514689758314, "learning_rate": 5.2472584316159736e-05, "loss": 4.6817, "step": 2536 }, { "epoch": 0.15748960208579055, "grad_norm": 1.105722890157161, "learning_rate": 5.249327539830333e-05, "loss": 4.6974, "step": 2537 }, { "epoch": 0.15755167918554844, "grad_norm": 0.9021500218491683, "learning_rate": 5.251396648044693e-05, "loss": 4.659, "step": 2538 }, { "epoch": 0.15761375628530636, "grad_norm": 1.1006937050374004, "learning_rate": 5.253465756259053e-05, "loss": 4.6847, "step": 2539 }, { "epoch": 0.15767583338506425, "grad_norm": 1.02868616853977, "learning_rate": 5.255534864473412e-05, "loss": 4.6306, "step": 2540 }, { "epoch": 0.15773791048482214, "grad_norm": 0.8826810013050297, "learning_rate": 5.257603972687771e-05, "loss": 4.6339, "step": 2541 }, { "epoch": 0.15779998758458005, "grad_norm": 0.8595303828476563, "learning_rate": 5.259673080902131e-05, "loss": 4.6537, "step": 2542 }, { "epoch": 0.15786206468433794, "grad_norm": 0.9695037555100494, "learning_rate": 5.261742189116491e-05, "loss": 4.7149, "step": 2543 }, { "epoch": 0.15792414178409583, "grad_norm": 0.9516462204658038, "learning_rate": 5.26381129733085e-05, "loss": 4.7911, "step": 2544 }, { "epoch": 0.15798621888385375, "grad_norm": 1.0457089891009232, "learning_rate": 5.2658804055452105e-05, "loss": 4.7319, "step": 2545 }, { "epoch": 0.15804829598361164, "grad_norm": 1.1542196231816533, "learning_rate": 5.26794951375957e-05, "loss": 4.5696, "step": 2546 }, { "epoch": 0.15811037308336953, "grad_norm": 0.9926462089310542, "learning_rate": 5.27001862197393e-05, "loss": 4.7057, "step": 2547 }, { "epoch": 0.15817245018312745, "grad_norm": 1.1740100484772165, "learning_rate": 5.272087730188289e-05, "loss": 4.6765, "step": 2548 }, { "epoch": 0.15823452728288534, "grad_norm": 1.2638845367962217, "learning_rate": 5.2741568384026486e-05, "loss": 4.7492, "step": 2549 }, { "epoch": 0.15829660438264323, "grad_norm": 0.972880724393454, "learning_rate": 5.276225946617008e-05, "loss": 4.67, "step": 2550 }, { "epoch": 0.15835868148240115, "grad_norm": 0.974362665114221, "learning_rate": 5.278295054831368e-05, "loss": 4.6387, "step": 2551 }, { "epoch": 0.15842075858215904, "grad_norm": 1.0769240836516065, "learning_rate": 5.280364163045728e-05, "loss": 4.7011, "step": 2552 }, { "epoch": 0.15848283568191693, "grad_norm": 1.0446067834105457, "learning_rate": 5.282433271260088e-05, "loss": 4.8487, "step": 2553 }, { "epoch": 0.15854491278167485, "grad_norm": 1.1431643815971333, "learning_rate": 5.284502379474446e-05, "loss": 4.6798, "step": 2554 }, { "epoch": 0.15860698988143274, "grad_norm": 0.9180077402478357, "learning_rate": 5.286571487688806e-05, "loss": 4.7127, "step": 2555 }, { "epoch": 0.15866906698119063, "grad_norm": 1.1595155121825422, "learning_rate": 5.288640595903166e-05, "loss": 4.7028, "step": 2556 }, { "epoch": 0.15873114408094854, "grad_norm": 1.070511124272923, "learning_rate": 5.290709704117526e-05, "loss": 4.6708, "step": 2557 }, { "epoch": 0.15879322118070643, "grad_norm": 1.0277350083178458, "learning_rate": 5.2927788123318855e-05, "loss": 4.7106, "step": 2558 }, { "epoch": 0.15885529828046432, "grad_norm": 1.1083440691683724, "learning_rate": 5.294847920546246e-05, "loss": 4.698, "step": 2559 }, { "epoch": 0.15891737538022224, "grad_norm": 0.9447994258496109, "learning_rate": 5.296917028760604e-05, "loss": 4.6749, "step": 2560 }, { "epoch": 0.15897945247998013, "grad_norm": 0.9794982118887204, "learning_rate": 5.298986136974964e-05, "loss": 4.5837, "step": 2561 }, { "epoch": 0.15904152957973802, "grad_norm": 1.0386807880686293, "learning_rate": 5.3010552451893235e-05, "loss": 4.5342, "step": 2562 }, { "epoch": 0.15910360667949594, "grad_norm": 0.9818389558641378, "learning_rate": 5.303124353403684e-05, "loss": 4.6461, "step": 2563 }, { "epoch": 0.15916568377925383, "grad_norm": 1.3088069002011566, "learning_rate": 5.305193461618043e-05, "loss": 4.629, "step": 2564 }, { "epoch": 0.15922776087901172, "grad_norm": 1.133663673268686, "learning_rate": 5.307262569832403e-05, "loss": 4.6632, "step": 2565 }, { "epoch": 0.15928983797876964, "grad_norm": 0.9305334009471773, "learning_rate": 5.3093316780467615e-05, "loss": 4.6784, "step": 2566 }, { "epoch": 0.15935191507852753, "grad_norm": 1.108845006577317, "learning_rate": 5.311400786261122e-05, "loss": 4.6731, "step": 2567 }, { "epoch": 0.15941399217828542, "grad_norm": 1.0443500415826283, "learning_rate": 5.313469894475481e-05, "loss": 4.6207, "step": 2568 }, { "epoch": 0.15947606927804334, "grad_norm": 1.0985627147776136, "learning_rate": 5.315539002689841e-05, "loss": 4.6828, "step": 2569 }, { "epoch": 0.15953814637780123, "grad_norm": 0.9038547893570531, "learning_rate": 5.317608110904201e-05, "loss": 4.7082, "step": 2570 }, { "epoch": 0.15960022347755912, "grad_norm": 1.153206092030037, "learning_rate": 5.3196772191185604e-05, "loss": 4.6246, "step": 2571 }, { "epoch": 0.15966230057731703, "grad_norm": 0.985354133353079, "learning_rate": 5.321746327332919e-05, "loss": 4.6755, "step": 2572 }, { "epoch": 0.15972437767707492, "grad_norm": 0.9710275579896777, "learning_rate": 5.3238154355472794e-05, "loss": 4.6411, "step": 2573 }, { "epoch": 0.15978645477683281, "grad_norm": 1.1530405297910213, "learning_rate": 5.325884543761639e-05, "loss": 4.7022, "step": 2574 }, { "epoch": 0.15984853187659073, "grad_norm": 0.9742954948928794, "learning_rate": 5.3279536519759985e-05, "loss": 4.6052, "step": 2575 }, { "epoch": 0.15991060897634862, "grad_norm": 1.1086686800580792, "learning_rate": 5.3300227601903586e-05, "loss": 4.7571, "step": 2576 }, { "epoch": 0.1599726860761065, "grad_norm": 0.9304034577722003, "learning_rate": 5.332091868404718e-05, "loss": 4.7312, "step": 2577 }, { "epoch": 0.16003476317586443, "grad_norm": 1.1127085500970622, "learning_rate": 5.334160976619077e-05, "loss": 4.6744, "step": 2578 }, { "epoch": 0.16009684027562232, "grad_norm": 1.0605372082461146, "learning_rate": 5.3362300848334365e-05, "loss": 4.5464, "step": 2579 }, { "epoch": 0.1601589173753802, "grad_norm": 1.02057386850733, "learning_rate": 5.338299193047797e-05, "loss": 4.6549, "step": 2580 }, { "epoch": 0.16022099447513813, "grad_norm": 1.12706529928412, "learning_rate": 5.340368301262156e-05, "loss": 4.6686, "step": 2581 }, { "epoch": 0.16028307157489602, "grad_norm": 1.0137464903783666, "learning_rate": 5.3424374094765164e-05, "loss": 4.6433, "step": 2582 }, { "epoch": 0.1603451486746539, "grad_norm": 0.7846982008896747, "learning_rate": 5.344506517690876e-05, "loss": 4.6098, "step": 2583 }, { "epoch": 0.16040722577441183, "grad_norm": 0.8331212886604583, "learning_rate": 5.346575625905235e-05, "loss": 4.6414, "step": 2584 }, { "epoch": 0.16046930287416972, "grad_norm": 1.0666945071362117, "learning_rate": 5.348644734119594e-05, "loss": 4.7181, "step": 2585 }, { "epoch": 0.1605313799739276, "grad_norm": 1.34446023313501, "learning_rate": 5.3507138423339544e-05, "loss": 4.6567, "step": 2586 }, { "epoch": 0.16059345707368552, "grad_norm": 0.9190198035052192, "learning_rate": 5.352782950548314e-05, "loss": 4.5759, "step": 2587 }, { "epoch": 0.16065553417344342, "grad_norm": 1.1769921540360992, "learning_rate": 5.354852058762674e-05, "loss": 4.6046, "step": 2588 }, { "epoch": 0.1607176112732013, "grad_norm": 0.9843036383541739, "learning_rate": 5.3569211669770336e-05, "loss": 4.569, "step": 2589 }, { "epoch": 0.16077968837295922, "grad_norm": 0.9705734933916771, "learning_rate": 5.3589902751913924e-05, "loss": 4.6605, "step": 2590 }, { "epoch": 0.1608417654727171, "grad_norm": 1.0824061546436876, "learning_rate": 5.361059383405752e-05, "loss": 4.5712, "step": 2591 }, { "epoch": 0.160903842572475, "grad_norm": 1.1044774008114417, "learning_rate": 5.363128491620112e-05, "loss": 4.7071, "step": 2592 }, { "epoch": 0.16096591967223292, "grad_norm": 0.9792844755576892, "learning_rate": 5.3651975998344716e-05, "loss": 4.6821, "step": 2593 }, { "epoch": 0.1610279967719908, "grad_norm": 1.0226865736251893, "learning_rate": 5.367266708048831e-05, "loss": 4.6729, "step": 2594 }, { "epoch": 0.1610900738717487, "grad_norm": 1.0887981581083046, "learning_rate": 5.369335816263191e-05, "loss": 4.6268, "step": 2595 }, { "epoch": 0.16115215097150662, "grad_norm": 0.9323582154857623, "learning_rate": 5.37140492447755e-05, "loss": 4.721, "step": 2596 }, { "epoch": 0.1612142280712645, "grad_norm": 0.9456509386275695, "learning_rate": 5.37347403269191e-05, "loss": 4.7021, "step": 2597 }, { "epoch": 0.1612763051710224, "grad_norm": 0.9938185794126905, "learning_rate": 5.37554314090627e-05, "loss": 4.6392, "step": 2598 }, { "epoch": 0.16133838227078032, "grad_norm": 0.9445130301485964, "learning_rate": 5.3776122491206294e-05, "loss": 4.6346, "step": 2599 }, { "epoch": 0.1614004593705382, "grad_norm": 0.9421998790225083, "learning_rate": 5.379681357334989e-05, "loss": 4.7513, "step": 2600 }, { "epoch": 0.1614625364702961, "grad_norm": 1.0133662415376883, "learning_rate": 5.381750465549349e-05, "loss": 4.6661, "step": 2601 }, { "epoch": 0.16152461357005402, "grad_norm": 0.9998923482170825, "learning_rate": 5.383819573763708e-05, "loss": 4.6539, "step": 2602 }, { "epoch": 0.1615866906698119, "grad_norm": 1.0236088267185413, "learning_rate": 5.3858886819780674e-05, "loss": 4.6748, "step": 2603 }, { "epoch": 0.1616487677695698, "grad_norm": 0.8934232688109301, "learning_rate": 5.387957790192427e-05, "loss": 4.6705, "step": 2604 }, { "epoch": 0.1617108448693277, "grad_norm": 1.0086163412647406, "learning_rate": 5.390026898406787e-05, "loss": 4.6389, "step": 2605 }, { "epoch": 0.1617729219690856, "grad_norm": 1.217071159395066, "learning_rate": 5.3920960066211466e-05, "loss": 4.6876, "step": 2606 }, { "epoch": 0.1618349990688435, "grad_norm": 1.3368758982637088, "learning_rate": 5.394165114835507e-05, "loss": 4.6176, "step": 2607 }, { "epoch": 0.1618970761686014, "grad_norm": 1.16195226296479, "learning_rate": 5.396234223049865e-05, "loss": 4.7357, "step": 2608 }, { "epoch": 0.1619591532683593, "grad_norm": 1.0800833676091965, "learning_rate": 5.398303331264225e-05, "loss": 4.6079, "step": 2609 }, { "epoch": 0.1620212303681172, "grad_norm": 1.1400141713449485, "learning_rate": 5.4003724394785846e-05, "loss": 4.5957, "step": 2610 }, { "epoch": 0.1620833074678751, "grad_norm": 1.1220112943607816, "learning_rate": 5.402441547692945e-05, "loss": 4.6123, "step": 2611 }, { "epoch": 0.162145384567633, "grad_norm": 1.0999189520368329, "learning_rate": 5.404510655907304e-05, "loss": 4.5822, "step": 2612 }, { "epoch": 0.1622074616673909, "grad_norm": 1.026934408000614, "learning_rate": 5.4065797641216645e-05, "loss": 4.6493, "step": 2613 }, { "epoch": 0.1622695387671488, "grad_norm": 1.135145901700428, "learning_rate": 5.4086488723360226e-05, "loss": 4.5727, "step": 2614 }, { "epoch": 0.1623316158669067, "grad_norm": 0.8478661905646794, "learning_rate": 5.410717980550383e-05, "loss": 4.6153, "step": 2615 }, { "epoch": 0.1623936929666646, "grad_norm": 0.956703597052674, "learning_rate": 5.4127870887647423e-05, "loss": 4.5589, "step": 2616 }, { "epoch": 0.1624557700664225, "grad_norm": 1.1886266583816139, "learning_rate": 5.4148561969791025e-05, "loss": 4.6944, "step": 2617 }, { "epoch": 0.1625178471661804, "grad_norm": 0.9309806165285602, "learning_rate": 5.416925305193462e-05, "loss": 4.6619, "step": 2618 }, { "epoch": 0.16257992426593829, "grad_norm": 1.1220722248270258, "learning_rate": 5.4189944134078215e-05, "loss": 4.5509, "step": 2619 }, { "epoch": 0.1626420013656962, "grad_norm": 1.1613482378909459, "learning_rate": 5.4210635216221804e-05, "loss": 4.5801, "step": 2620 }, { "epoch": 0.1627040784654541, "grad_norm": 0.9904436420773313, "learning_rate": 5.4231326298365406e-05, "loss": 4.4541, "step": 2621 }, { "epoch": 0.16276615556521198, "grad_norm": 1.2197713729164923, "learning_rate": 5.4252017380509e-05, "loss": 4.5591, "step": 2622 }, { "epoch": 0.1628282326649699, "grad_norm": 1.2182787448710035, "learning_rate": 5.42727084626526e-05, "loss": 4.6486, "step": 2623 }, { "epoch": 0.1628903097647278, "grad_norm": 0.8860201667552856, "learning_rate": 5.42933995447962e-05, "loss": 4.5915, "step": 2624 }, { "epoch": 0.16295238686448568, "grad_norm": 0.9950789724093142, "learning_rate": 5.431409062693979e-05, "loss": 4.6767, "step": 2625 }, { "epoch": 0.1630144639642436, "grad_norm": 1.2613308431313137, "learning_rate": 5.433478170908338e-05, "loss": 4.5776, "step": 2626 }, { "epoch": 0.1630765410640015, "grad_norm": 0.9540695450563583, "learning_rate": 5.435547279122698e-05, "loss": 4.639, "step": 2627 }, { "epoch": 0.16313861816375938, "grad_norm": 1.0061484504779148, "learning_rate": 5.437616387337058e-05, "loss": 4.66, "step": 2628 }, { "epoch": 0.1632006952635173, "grad_norm": 1.0175361844313928, "learning_rate": 5.439685495551417e-05, "loss": 4.539, "step": 2629 }, { "epoch": 0.1632627723632752, "grad_norm": 1.1366196946444316, "learning_rate": 5.4417546037657775e-05, "loss": 4.6304, "step": 2630 }, { "epoch": 0.16332484946303308, "grad_norm": 0.908794415685881, "learning_rate": 5.443823711980137e-05, "loss": 4.6648, "step": 2631 }, { "epoch": 0.163386926562791, "grad_norm": 0.9576977024701264, "learning_rate": 5.445892820194497e-05, "loss": 4.5638, "step": 2632 }, { "epoch": 0.1634490036625489, "grad_norm": 1.0429157143584804, "learning_rate": 5.447961928408855e-05, "loss": 4.6321, "step": 2633 }, { "epoch": 0.16351108076230678, "grad_norm": 1.1028777582972213, "learning_rate": 5.4500310366232155e-05, "loss": 4.7636, "step": 2634 }, { "epoch": 0.1635731578620647, "grad_norm": 1.0244607832939547, "learning_rate": 5.452100144837575e-05, "loss": 4.6722, "step": 2635 }, { "epoch": 0.16363523496182258, "grad_norm": 1.3955638492510565, "learning_rate": 5.454169253051935e-05, "loss": 4.653, "step": 2636 }, { "epoch": 0.16369731206158047, "grad_norm": 1.0109526926399581, "learning_rate": 5.456238361266295e-05, "loss": 4.5554, "step": 2637 }, { "epoch": 0.1637593891613384, "grad_norm": 1.0745226066167395, "learning_rate": 5.458307469480655e-05, "loss": 4.6154, "step": 2638 }, { "epoch": 0.16382146626109628, "grad_norm": 1.3570260987214224, "learning_rate": 5.460376577695013e-05, "loss": 4.6177, "step": 2639 }, { "epoch": 0.16388354336085417, "grad_norm": 1.1478222606814104, "learning_rate": 5.462445685909373e-05, "loss": 4.6103, "step": 2640 }, { "epoch": 0.1639456204606121, "grad_norm": 1.3068771262313499, "learning_rate": 5.464514794123733e-05, "loss": 4.5627, "step": 2641 }, { "epoch": 0.16400769756036998, "grad_norm": 0.7663851876042923, "learning_rate": 5.466583902338093e-05, "loss": 4.6556, "step": 2642 }, { "epoch": 0.16406977466012787, "grad_norm": 1.0614652644047482, "learning_rate": 5.4686530105524524e-05, "loss": 4.669, "step": 2643 }, { "epoch": 0.1641318517598858, "grad_norm": 1.4461384655568466, "learning_rate": 5.4707221187668126e-05, "loss": 4.6535, "step": 2644 }, { "epoch": 0.16419392885964368, "grad_norm": 0.8624508346144885, "learning_rate": 5.472791226981171e-05, "loss": 4.6796, "step": 2645 }, { "epoch": 0.16425600595940157, "grad_norm": 1.3370475841691118, "learning_rate": 5.474860335195531e-05, "loss": 4.6534, "step": 2646 }, { "epoch": 0.1643180830591595, "grad_norm": 1.0377928929370621, "learning_rate": 5.4769294434098905e-05, "loss": 4.6602, "step": 2647 }, { "epoch": 0.16438016015891738, "grad_norm": 1.013893353149015, "learning_rate": 5.4789985516242507e-05, "loss": 4.5422, "step": 2648 }, { "epoch": 0.16444223725867527, "grad_norm": 0.9069474266149911, "learning_rate": 5.48106765983861e-05, "loss": 4.5785, "step": 2649 }, { "epoch": 0.16450431435843318, "grad_norm": 0.9568729503268285, "learning_rate": 5.48313676805297e-05, "loss": 4.5358, "step": 2650 }, { "epoch": 0.16456639145819107, "grad_norm": 1.0906842999882715, "learning_rate": 5.4852058762673285e-05, "loss": 4.6142, "step": 2651 }, { "epoch": 0.16462846855794896, "grad_norm": 1.045788044094539, "learning_rate": 5.487274984481689e-05, "loss": 4.6338, "step": 2652 }, { "epoch": 0.16469054565770688, "grad_norm": 1.0932540859213227, "learning_rate": 5.489344092696048e-05, "loss": 4.5968, "step": 2653 }, { "epoch": 0.16475262275746477, "grad_norm": 0.917155193429241, "learning_rate": 5.491413200910408e-05, "loss": 4.5281, "step": 2654 }, { "epoch": 0.16481469985722266, "grad_norm": 1.24521780766842, "learning_rate": 5.493482309124768e-05, "loss": 4.5946, "step": 2655 }, { "epoch": 0.16487677695698058, "grad_norm": 1.0949102514754399, "learning_rate": 5.4955514173391274e-05, "loss": 4.4058, "step": 2656 }, { "epoch": 0.16493885405673847, "grad_norm": 0.9579004505594072, "learning_rate": 5.497620525553486e-05, "loss": 4.6475, "step": 2657 }, { "epoch": 0.16500093115649636, "grad_norm": 1.1788034221076547, "learning_rate": 5.4996896337678464e-05, "loss": 4.6471, "step": 2658 }, { "epoch": 0.16506300825625428, "grad_norm": 1.1393405004688484, "learning_rate": 5.501758741982206e-05, "loss": 4.645, "step": 2659 }, { "epoch": 0.16512508535601217, "grad_norm": 0.8864057551449329, "learning_rate": 5.5038278501965654e-05, "loss": 4.666, "step": 2660 }, { "epoch": 0.16518716245577006, "grad_norm": 1.0495522270928825, "learning_rate": 5.5058969584109256e-05, "loss": 4.6082, "step": 2661 }, { "epoch": 0.16524923955552798, "grad_norm": 1.222915037025176, "learning_rate": 5.507966066625285e-05, "loss": 4.6063, "step": 2662 }, { "epoch": 0.16531131665528587, "grad_norm": 0.8720530885232165, "learning_rate": 5.510035174839644e-05, "loss": 4.6818, "step": 2663 }, { "epoch": 0.16537339375504376, "grad_norm": 1.04433634386827, "learning_rate": 5.5121042830540035e-05, "loss": 4.5405, "step": 2664 }, { "epoch": 0.16543547085480168, "grad_norm": 0.9683775955092535, "learning_rate": 5.5141733912683636e-05, "loss": 4.6145, "step": 2665 }, { "epoch": 0.16549754795455957, "grad_norm": 0.9092630087986735, "learning_rate": 5.516242499482723e-05, "loss": 4.5116, "step": 2666 }, { "epoch": 0.16555962505431746, "grad_norm": 1.0353806162544488, "learning_rate": 5.518311607697083e-05, "loss": 4.592, "step": 2667 }, { "epoch": 0.16562170215407537, "grad_norm": 0.8929622833471696, "learning_rate": 5.520380715911443e-05, "loss": 4.7253, "step": 2668 }, { "epoch": 0.16568377925383326, "grad_norm": 0.8991848366686588, "learning_rate": 5.522449824125802e-05, "loss": 4.5239, "step": 2669 }, { "epoch": 0.16574585635359115, "grad_norm": 0.91923254147639, "learning_rate": 5.524518932340161e-05, "loss": 4.6442, "step": 2670 }, { "epoch": 0.16580793345334907, "grad_norm": 1.0540468884196095, "learning_rate": 5.5265880405545214e-05, "loss": 4.6696, "step": 2671 }, { "epoch": 0.16587001055310696, "grad_norm": 1.1811046551290418, "learning_rate": 5.528657148768881e-05, "loss": 4.552, "step": 2672 }, { "epoch": 0.16593208765286485, "grad_norm": 0.9384099359113146, "learning_rate": 5.530726256983241e-05, "loss": 4.5825, "step": 2673 }, { "epoch": 0.16599416475262277, "grad_norm": 1.228295577921303, "learning_rate": 5.5327953651976006e-05, "loss": 4.6463, "step": 2674 }, { "epoch": 0.16605624185238066, "grad_norm": 0.9951825031561624, "learning_rate": 5.5348644734119594e-05, "loss": 4.6432, "step": 2675 }, { "epoch": 0.16611831895213855, "grad_norm": 0.9200990348358117, "learning_rate": 5.536933581626319e-05, "loss": 4.6206, "step": 2676 }, { "epoch": 0.16618039605189647, "grad_norm": 0.9909324517868834, "learning_rate": 5.539002689840679e-05, "loss": 4.6004, "step": 2677 }, { "epoch": 0.16624247315165436, "grad_norm": 1.1323480640580281, "learning_rate": 5.5410717980550386e-05, "loss": 4.6308, "step": 2678 }, { "epoch": 0.16630455025141225, "grad_norm": 1.0717402493046881, "learning_rate": 5.543140906269398e-05, "loss": 4.6737, "step": 2679 }, { "epoch": 0.16636662735117017, "grad_norm": 1.0565565713954743, "learning_rate": 5.545210014483758e-05, "loss": 4.5776, "step": 2680 }, { "epoch": 0.16642870445092806, "grad_norm": 1.083166470270295, "learning_rate": 5.547279122698117e-05, "loss": 4.5133, "step": 2681 }, { "epoch": 0.16649078155068595, "grad_norm": 1.0261781411329967, "learning_rate": 5.5493482309124766e-05, "loss": 4.5934, "step": 2682 }, { "epoch": 0.16655285865044386, "grad_norm": 1.0143118952820402, "learning_rate": 5.551417339126837e-05, "loss": 4.5762, "step": 2683 }, { "epoch": 0.16661493575020175, "grad_norm": 0.896619134219798, "learning_rate": 5.553486447341196e-05, "loss": 4.6303, "step": 2684 }, { "epoch": 0.16667701284995964, "grad_norm": 1.025588534193324, "learning_rate": 5.555555555555556e-05, "loss": 4.575, "step": 2685 }, { "epoch": 0.16673908994971756, "grad_norm": 1.0746048666867445, "learning_rate": 5.557624663769916e-05, "loss": 4.6014, "step": 2686 }, { "epoch": 0.16680116704947545, "grad_norm": 0.9029513538623574, "learning_rate": 5.559693771984275e-05, "loss": 4.5997, "step": 2687 }, { "epoch": 0.16686324414923334, "grad_norm": 1.0692513210314336, "learning_rate": 5.5617628801986343e-05, "loss": 4.6016, "step": 2688 }, { "epoch": 0.16692532124899126, "grad_norm": 1.0771682474592121, "learning_rate": 5.563831988412994e-05, "loss": 4.6966, "step": 2689 }, { "epoch": 0.16698739834874915, "grad_norm": 1.3536188435796273, "learning_rate": 5.565901096627354e-05, "loss": 4.5637, "step": 2690 }, { "epoch": 0.16704947544850704, "grad_norm": 0.9429674617881794, "learning_rate": 5.5679702048417135e-05, "loss": 4.6327, "step": 2691 }, { "epoch": 0.16711155254826496, "grad_norm": 1.0539083357805181, "learning_rate": 5.570039313056074e-05, "loss": 4.4475, "step": 2692 }, { "epoch": 0.16717362964802285, "grad_norm": 1.035307617636561, "learning_rate": 5.572108421270432e-05, "loss": 4.5768, "step": 2693 }, { "epoch": 0.16723570674778074, "grad_norm": 1.2288325981870214, "learning_rate": 5.574177529484792e-05, "loss": 4.6566, "step": 2694 }, { "epoch": 0.16729778384753866, "grad_norm": 0.903351060498646, "learning_rate": 5.5762466376991516e-05, "loss": 4.7275, "step": 2695 }, { "epoch": 0.16735986094729655, "grad_norm": 1.2784703970187037, "learning_rate": 5.578315745913512e-05, "loss": 4.6397, "step": 2696 }, { "epoch": 0.16742193804705444, "grad_norm": 1.0115445845051272, "learning_rate": 5.580384854127871e-05, "loss": 4.6521, "step": 2697 }, { "epoch": 0.16748401514681235, "grad_norm": 0.9230201203952267, "learning_rate": 5.5824539623422315e-05, "loss": 4.5763, "step": 2698 }, { "epoch": 0.16754609224657024, "grad_norm": 1.2034879855232574, "learning_rate": 5.5845230705565896e-05, "loss": 4.7012, "step": 2699 }, { "epoch": 0.16760816934632813, "grad_norm": 0.8811458282584947, "learning_rate": 5.58659217877095e-05, "loss": 4.5592, "step": 2700 }, { "epoch": 0.16767024644608605, "grad_norm": 1.0466963202736455, "learning_rate": 5.588661286985309e-05, "loss": 4.5739, "step": 2701 }, { "epoch": 0.16773232354584394, "grad_norm": 0.9056793915794519, "learning_rate": 5.5907303951996695e-05, "loss": 4.6457, "step": 2702 }, { "epoch": 0.16779440064560183, "grad_norm": 1.0864547485604523, "learning_rate": 5.592799503414029e-05, "loss": 4.6215, "step": 2703 }, { "epoch": 0.16785647774535975, "grad_norm": 1.029590253622731, "learning_rate": 5.5948686116283885e-05, "loss": 4.5478, "step": 2704 }, { "epoch": 0.16791855484511764, "grad_norm": 1.050960539656857, "learning_rate": 5.596937719842747e-05, "loss": 4.6541, "step": 2705 }, { "epoch": 0.16798063194487553, "grad_norm": 1.141812111762351, "learning_rate": 5.5990068280571075e-05, "loss": 4.5364, "step": 2706 }, { "epoch": 0.16804270904463345, "grad_norm": 1.0822089975388038, "learning_rate": 5.601075936271467e-05, "loss": 4.6002, "step": 2707 }, { "epoch": 0.16810478614439134, "grad_norm": 1.0547754886799752, "learning_rate": 5.603145044485827e-05, "loss": 4.5865, "step": 2708 }, { "epoch": 0.16816686324414923, "grad_norm": 0.99194384008495, "learning_rate": 5.605214152700187e-05, "loss": 4.5372, "step": 2709 }, { "epoch": 0.16822894034390715, "grad_norm": 0.9997240818362405, "learning_rate": 5.607283260914546e-05, "loss": 4.5412, "step": 2710 }, { "epoch": 0.16829101744366504, "grad_norm": 1.030124917556987, "learning_rate": 5.609352369128905e-05, "loss": 4.4425, "step": 2711 }, { "epoch": 0.16835309454342293, "grad_norm": 1.0201881017399799, "learning_rate": 5.611421477343265e-05, "loss": 4.6887, "step": 2712 }, { "epoch": 0.16841517164318084, "grad_norm": 0.8989324450310616, "learning_rate": 5.613490585557625e-05, "loss": 4.6237, "step": 2713 }, { "epoch": 0.16847724874293873, "grad_norm": 0.9682278610365699, "learning_rate": 5.615559693771984e-05, "loss": 4.5987, "step": 2714 }, { "epoch": 0.16853932584269662, "grad_norm": 1.0019809057239992, "learning_rate": 5.6176288019863444e-05, "loss": 4.4801, "step": 2715 }, { "epoch": 0.16860140294245454, "grad_norm": 0.8771674015472869, "learning_rate": 5.619697910200704e-05, "loss": 4.6238, "step": 2716 }, { "epoch": 0.16866348004221243, "grad_norm": 0.9445173213616428, "learning_rate": 5.621767018415064e-05, "loss": 4.6113, "step": 2717 }, { "epoch": 0.16872555714197032, "grad_norm": 0.9760196244971535, "learning_rate": 5.623836126629422e-05, "loss": 4.6215, "step": 2718 }, { "epoch": 0.1687876342417282, "grad_norm": 0.871086428487484, "learning_rate": 5.6259052348437825e-05, "loss": 4.6316, "step": 2719 }, { "epoch": 0.16884971134148613, "grad_norm": 0.886608839575349, "learning_rate": 5.627974343058142e-05, "loss": 4.4992, "step": 2720 }, { "epoch": 0.16891178844124402, "grad_norm": 0.9835980997754451, "learning_rate": 5.630043451272502e-05, "loss": 4.6226, "step": 2721 }, { "epoch": 0.1689738655410019, "grad_norm": 1.1133642333038933, "learning_rate": 5.632112559486862e-05, "loss": 4.4812, "step": 2722 }, { "epoch": 0.16903594264075983, "grad_norm": 0.939513809242696, "learning_rate": 5.634181667701222e-05, "loss": 4.6203, "step": 2723 }, { "epoch": 0.16909801974051772, "grad_norm": 1.4247467886839522, "learning_rate": 5.63625077591558e-05, "loss": 4.5802, "step": 2724 }, { "epoch": 0.1691600968402756, "grad_norm": 0.9504047723572454, "learning_rate": 5.63831988412994e-05, "loss": 4.5196, "step": 2725 }, { "epoch": 0.16922217394003353, "grad_norm": 1.1706210230135026, "learning_rate": 5.6403889923443e-05, "loss": 4.6643, "step": 2726 }, { "epoch": 0.16928425103979142, "grad_norm": 0.9702190827198854, "learning_rate": 5.64245810055866e-05, "loss": 4.6239, "step": 2727 }, { "epoch": 0.1693463281395493, "grad_norm": 1.1785160336542733, "learning_rate": 5.6445272087730194e-05, "loss": 4.5461, "step": 2728 }, { "epoch": 0.16940840523930722, "grad_norm": 0.9589669717425446, "learning_rate": 5.646596316987379e-05, "loss": 4.5465, "step": 2729 }, { "epoch": 0.16947048233906511, "grad_norm": 1.0241203068751745, "learning_rate": 5.648665425201738e-05, "loss": 4.5687, "step": 2730 }, { "epoch": 0.169532559438823, "grad_norm": 1.0683229987665732, "learning_rate": 5.650734533416098e-05, "loss": 4.5846, "step": 2731 }, { "epoch": 0.16959463653858092, "grad_norm": 0.9756733317747577, "learning_rate": 5.6528036416304574e-05, "loss": 4.5932, "step": 2732 }, { "epoch": 0.1696567136383388, "grad_norm": 1.1348209334796118, "learning_rate": 5.6548727498448176e-05, "loss": 4.5393, "step": 2733 }, { "epoch": 0.1697187907380967, "grad_norm": 1.0176224653134902, "learning_rate": 5.656941858059177e-05, "loss": 4.5921, "step": 2734 }, { "epoch": 0.16978086783785462, "grad_norm": 1.0570249299327759, "learning_rate": 5.6590109662735366e-05, "loss": 4.5226, "step": 2735 }, { "epoch": 0.1698429449376125, "grad_norm": 1.1394336174363149, "learning_rate": 5.6610800744878955e-05, "loss": 4.5501, "step": 2736 }, { "epoch": 0.1699050220373704, "grad_norm": 1.1314673494220802, "learning_rate": 5.6631491827022556e-05, "loss": 4.6368, "step": 2737 }, { "epoch": 0.16996709913712832, "grad_norm": 0.9727695506338806, "learning_rate": 5.665218290916615e-05, "loss": 4.4803, "step": 2738 }, { "epoch": 0.1700291762368862, "grad_norm": 0.9107813182705788, "learning_rate": 5.6672873991309747e-05, "loss": 4.5645, "step": 2739 }, { "epoch": 0.1700912533366441, "grad_norm": 0.7814437244554557, "learning_rate": 5.669356507345335e-05, "loss": 4.4917, "step": 2740 }, { "epoch": 0.17015333043640202, "grad_norm": 0.9271729065274072, "learning_rate": 5.6714256155596943e-05, "loss": 4.5629, "step": 2741 }, { "epoch": 0.1702154075361599, "grad_norm": 1.0893114765881857, "learning_rate": 5.673494723774053e-05, "loss": 4.5513, "step": 2742 }, { "epoch": 0.1702774846359178, "grad_norm": 0.9818341599875744, "learning_rate": 5.675563831988413e-05, "loss": 4.4561, "step": 2743 }, { "epoch": 0.17033956173567572, "grad_norm": 0.8897632311260321, "learning_rate": 5.677632940202773e-05, "loss": 4.5089, "step": 2744 }, { "epoch": 0.1704016388354336, "grad_norm": 1.0613200423600146, "learning_rate": 5.6797020484171324e-05, "loss": 4.5387, "step": 2745 }, { "epoch": 0.1704637159351915, "grad_norm": 1.193085884184412, "learning_rate": 5.6817711566314926e-05, "loss": 4.4762, "step": 2746 }, { "epoch": 0.1705257930349494, "grad_norm": 0.959403698022907, "learning_rate": 5.683840264845852e-05, "loss": 4.4078, "step": 2747 }, { "epoch": 0.1705878701347073, "grad_norm": 0.9897309477615933, "learning_rate": 5.685909373060211e-05, "loss": 4.5284, "step": 2748 }, { "epoch": 0.1706499472344652, "grad_norm": 1.205392014538961, "learning_rate": 5.6879784812745704e-05, "loss": 4.4598, "step": 2749 }, { "epoch": 0.1707120243342231, "grad_norm": 0.9326479282605435, "learning_rate": 5.6900475894889306e-05, "loss": 4.4872, "step": 2750 }, { "epoch": 0.170774101433981, "grad_norm": 0.9413074742332408, "learning_rate": 5.69211669770329e-05, "loss": 4.4844, "step": 2751 }, { "epoch": 0.1708361785337389, "grad_norm": 1.0404324785914436, "learning_rate": 5.69418580591765e-05, "loss": 4.6359, "step": 2752 }, { "epoch": 0.1708982556334968, "grad_norm": 1.0178164387932127, "learning_rate": 5.69625491413201e-05, "loss": 4.5111, "step": 2753 }, { "epoch": 0.1709603327332547, "grad_norm": 1.2847150607668398, "learning_rate": 5.6983240223463686e-05, "loss": 4.5273, "step": 2754 }, { "epoch": 0.1710224098330126, "grad_norm": 0.8738334236103743, "learning_rate": 5.700393130560728e-05, "loss": 4.4994, "step": 2755 }, { "epoch": 0.1710844869327705, "grad_norm": 1.1077962037516786, "learning_rate": 5.702462238775088e-05, "loss": 4.5095, "step": 2756 }, { "epoch": 0.1711465640325284, "grad_norm": 0.9255312811512783, "learning_rate": 5.704531346989448e-05, "loss": 4.3221, "step": 2757 }, { "epoch": 0.1712086411322863, "grad_norm": 0.902601230100744, "learning_rate": 5.706600455203808e-05, "loss": 4.6695, "step": 2758 }, { "epoch": 0.1712707182320442, "grad_norm": 0.9286602018438924, "learning_rate": 5.7086695634181675e-05, "loss": 4.5791, "step": 2759 }, { "epoch": 0.1713327953318021, "grad_norm": 1.0808176885299543, "learning_rate": 5.7107386716325263e-05, "loss": 4.411, "step": 2760 }, { "epoch": 0.17139487243155999, "grad_norm": 1.0002288556484835, "learning_rate": 5.712807779846886e-05, "loss": 4.4623, "step": 2761 }, { "epoch": 0.1714569495313179, "grad_norm": 1.1837249783829147, "learning_rate": 5.714876888061246e-05, "loss": 4.5559, "step": 2762 }, { "epoch": 0.1715190266310758, "grad_norm": 1.0276507364977487, "learning_rate": 5.7169459962756055e-05, "loss": 4.6004, "step": 2763 }, { "epoch": 0.17158110373083368, "grad_norm": 1.1274899372742913, "learning_rate": 5.719015104489965e-05, "loss": 4.5757, "step": 2764 }, { "epoch": 0.1716431808305916, "grad_norm": 1.0214037890767151, "learning_rate": 5.721084212704325e-05, "loss": 4.5616, "step": 2765 }, { "epoch": 0.1717052579303495, "grad_norm": 1.1373442406960534, "learning_rate": 5.723153320918684e-05, "loss": 4.5197, "step": 2766 }, { "epoch": 0.17176733503010738, "grad_norm": 0.890091016540948, "learning_rate": 5.7252224291330436e-05, "loss": 4.4798, "step": 2767 }, { "epoch": 0.1718294121298653, "grad_norm": 1.0957736776284202, "learning_rate": 5.727291537347403e-05, "loss": 4.5261, "step": 2768 }, { "epoch": 0.1718914892296232, "grad_norm": 0.8702966096204721, "learning_rate": 5.729360645561763e-05, "loss": 4.567, "step": 2769 }, { "epoch": 0.17195356632938108, "grad_norm": 0.9486101813441334, "learning_rate": 5.731429753776123e-05, "loss": 4.5883, "step": 2770 }, { "epoch": 0.172015643429139, "grad_norm": 0.9885975629293758, "learning_rate": 5.733498861990483e-05, "loss": 4.5415, "step": 2771 }, { "epoch": 0.1720777205288969, "grad_norm": 1.030072551505352, "learning_rate": 5.735567970204842e-05, "loss": 4.5638, "step": 2772 }, { "epoch": 0.17213979762865478, "grad_norm": 1.0289931020959624, "learning_rate": 5.737637078419201e-05, "loss": 4.4782, "step": 2773 }, { "epoch": 0.1722018747284127, "grad_norm": 1.0784002312118044, "learning_rate": 5.739706186633561e-05, "loss": 4.4744, "step": 2774 }, { "epoch": 0.1722639518281706, "grad_norm": 0.9039215561371202, "learning_rate": 5.741775294847921e-05, "loss": 4.5544, "step": 2775 }, { "epoch": 0.17232602892792848, "grad_norm": 0.9743272759163024, "learning_rate": 5.7438444030622805e-05, "loss": 4.495, "step": 2776 }, { "epoch": 0.1723881060276864, "grad_norm": 1.105561970261367, "learning_rate": 5.745913511276641e-05, "loss": 4.5265, "step": 2777 }, { "epoch": 0.17245018312744428, "grad_norm": 1.1190422507517714, "learning_rate": 5.747982619490999e-05, "loss": 4.4819, "step": 2778 }, { "epoch": 0.17251226022720217, "grad_norm": 0.8803692318880236, "learning_rate": 5.750051727705359e-05, "loss": 4.5155, "step": 2779 }, { "epoch": 0.1725743373269601, "grad_norm": 1.0636212659459365, "learning_rate": 5.7521208359197185e-05, "loss": 4.5566, "step": 2780 }, { "epoch": 0.17263641442671798, "grad_norm": 1.0533689380515234, "learning_rate": 5.754189944134079e-05, "loss": 4.5458, "step": 2781 }, { "epoch": 0.17269849152647587, "grad_norm": 0.8554918120598255, "learning_rate": 5.756259052348438e-05, "loss": 4.6003, "step": 2782 }, { "epoch": 0.1727605686262338, "grad_norm": 1.006863938690557, "learning_rate": 5.7583281605627984e-05, "loss": 4.53, "step": 2783 }, { "epoch": 0.17282264572599168, "grad_norm": 0.9664384582135852, "learning_rate": 5.7603972687771566e-05, "loss": 4.4103, "step": 2784 }, { "epoch": 0.17288472282574957, "grad_norm": 0.9753032958413177, "learning_rate": 5.762466376991517e-05, "loss": 4.4255, "step": 2785 }, { "epoch": 0.1729467999255075, "grad_norm": 0.953492172059051, "learning_rate": 5.764535485205876e-05, "loss": 4.4205, "step": 2786 }, { "epoch": 0.17300887702526538, "grad_norm": 1.083605285891162, "learning_rate": 5.7666045934202364e-05, "loss": 4.5117, "step": 2787 }, { "epoch": 0.17307095412502327, "grad_norm": 1.090061999928747, "learning_rate": 5.768673701634596e-05, "loss": 4.5168, "step": 2788 }, { "epoch": 0.1731330312247812, "grad_norm": 1.0776582896360012, "learning_rate": 5.7707428098489555e-05, "loss": 4.5339, "step": 2789 }, { "epoch": 0.17319510832453908, "grad_norm": 0.8016818680845346, "learning_rate": 5.772811918063314e-05, "loss": 4.5029, "step": 2790 }, { "epoch": 0.17325718542429697, "grad_norm": 0.8025357634160218, "learning_rate": 5.7748810262776745e-05, "loss": 4.5099, "step": 2791 }, { "epoch": 0.17331926252405488, "grad_norm": 0.9412359503093874, "learning_rate": 5.776950134492034e-05, "loss": 4.482, "step": 2792 }, { "epoch": 0.17338133962381277, "grad_norm": 1.0035575528293434, "learning_rate": 5.779019242706394e-05, "loss": 4.5718, "step": 2793 }, { "epoch": 0.17344341672357066, "grad_norm": 1.0610943078635182, "learning_rate": 5.781088350920754e-05, "loss": 4.4299, "step": 2794 }, { "epoch": 0.17350549382332858, "grad_norm": 0.7998328198153525, "learning_rate": 5.783157459135113e-05, "loss": 4.5171, "step": 2795 }, { "epoch": 0.17356757092308647, "grad_norm": 0.9402775309618863, "learning_rate": 5.785226567349472e-05, "loss": 4.5882, "step": 2796 }, { "epoch": 0.17362964802284436, "grad_norm": 1.116919532063132, "learning_rate": 5.787295675563832e-05, "loss": 4.5479, "step": 2797 }, { "epoch": 0.17369172512260228, "grad_norm": 0.9812080362179024, "learning_rate": 5.789364783778192e-05, "loss": 4.5025, "step": 2798 }, { "epoch": 0.17375380222236017, "grad_norm": 0.9438312684998202, "learning_rate": 5.791433891992551e-05, "loss": 4.4345, "step": 2799 }, { "epoch": 0.17381587932211806, "grad_norm": 0.9392454112014182, "learning_rate": 5.7935030002069114e-05, "loss": 4.5531, "step": 2800 }, { "epoch": 0.17387795642187598, "grad_norm": 0.9312215378271287, "learning_rate": 5.795572108421271e-05, "loss": 4.4529, "step": 2801 }, { "epoch": 0.17394003352163387, "grad_norm": 0.8548409235354457, "learning_rate": 5.797641216635631e-05, "loss": 4.4168, "step": 2802 }, { "epoch": 0.17400211062139176, "grad_norm": 0.8060057129527303, "learning_rate": 5.799710324849989e-05, "loss": 4.5473, "step": 2803 }, { "epoch": 0.17406418772114968, "grad_norm": 0.8253158210476691, "learning_rate": 5.8017794330643494e-05, "loss": 4.5034, "step": 2804 }, { "epoch": 0.17412626482090757, "grad_norm": 0.9016650401069604, "learning_rate": 5.803848541278709e-05, "loss": 4.5163, "step": 2805 }, { "epoch": 0.17418834192066546, "grad_norm": 1.012022230665576, "learning_rate": 5.805917649493069e-05, "loss": 4.4878, "step": 2806 }, { "epoch": 0.17425041902042337, "grad_norm": 1.1346361411408055, "learning_rate": 5.8079867577074286e-05, "loss": 4.6599, "step": 2807 }, { "epoch": 0.17431249612018127, "grad_norm": 0.9452312529146842, "learning_rate": 5.810055865921789e-05, "loss": 4.5204, "step": 2808 }, { "epoch": 0.17437457321993916, "grad_norm": 1.0548324485357605, "learning_rate": 5.812124974136147e-05, "loss": 4.48, "step": 2809 }, { "epoch": 0.17443665031969707, "grad_norm": 1.0826794166855558, "learning_rate": 5.814194082350507e-05, "loss": 4.4759, "step": 2810 }, { "epoch": 0.17449872741945496, "grad_norm": 0.7347127665235953, "learning_rate": 5.8162631905648667e-05, "loss": 4.5435, "step": 2811 }, { "epoch": 0.17456080451921285, "grad_norm": 0.846469347036598, "learning_rate": 5.818332298779227e-05, "loss": 4.4886, "step": 2812 }, { "epoch": 0.17462288161897077, "grad_norm": 1.1721988993298658, "learning_rate": 5.8204014069935864e-05, "loss": 4.4255, "step": 2813 }, { "epoch": 0.17468495871872866, "grad_norm": 0.769719676848682, "learning_rate": 5.822470515207946e-05, "loss": 4.4147, "step": 2814 }, { "epoch": 0.17474703581848655, "grad_norm": 1.0142960289789493, "learning_rate": 5.824539623422305e-05, "loss": 4.432, "step": 2815 }, { "epoch": 0.17480911291824447, "grad_norm": 1.0832508801602614, "learning_rate": 5.826608731636665e-05, "loss": 4.5829, "step": 2816 }, { "epoch": 0.17487119001800236, "grad_norm": 1.081842784787609, "learning_rate": 5.8286778398510244e-05, "loss": 4.465, "step": 2817 }, { "epoch": 0.17493326711776025, "grad_norm": 1.0591529410908795, "learning_rate": 5.8307469480653846e-05, "loss": 4.4381, "step": 2818 }, { "epoch": 0.17499534421751817, "grad_norm": 0.990603376327035, "learning_rate": 5.832816056279744e-05, "loss": 4.4659, "step": 2819 }, { "epoch": 0.17505742131727606, "grad_norm": 1.1085310936413841, "learning_rate": 5.8348851644941036e-05, "loss": 4.5844, "step": 2820 }, { "epoch": 0.17511949841703395, "grad_norm": 0.9687590137185405, "learning_rate": 5.8369542727084624e-05, "loss": 4.4996, "step": 2821 }, { "epoch": 0.17518157551679187, "grad_norm": 0.9965487768520845, "learning_rate": 5.8390233809228226e-05, "loss": 4.4795, "step": 2822 }, { "epoch": 0.17524365261654976, "grad_norm": 1.1786260819251637, "learning_rate": 5.841092489137182e-05, "loss": 4.4485, "step": 2823 }, { "epoch": 0.17530572971630765, "grad_norm": 0.9311664582229453, "learning_rate": 5.8431615973515416e-05, "loss": 4.4648, "step": 2824 }, { "epoch": 0.17536780681606556, "grad_norm": 0.8319761589329937, "learning_rate": 5.845230705565902e-05, "loss": 4.6096, "step": 2825 }, { "epoch": 0.17542988391582345, "grad_norm": 1.0596582894783606, "learning_rate": 5.847299813780261e-05, "loss": 4.3999, "step": 2826 }, { "epoch": 0.17549196101558134, "grad_norm": 0.9478615056571056, "learning_rate": 5.84936892199462e-05, "loss": 4.4892, "step": 2827 }, { "epoch": 0.17555403811533926, "grad_norm": 0.8519560578167653, "learning_rate": 5.8514380302089796e-05, "loss": 4.4354, "step": 2828 }, { "epoch": 0.17561611521509715, "grad_norm": 0.9675373978989174, "learning_rate": 5.85350713842334e-05, "loss": 4.4365, "step": 2829 }, { "epoch": 0.17567819231485504, "grad_norm": 0.9212554207361541, "learning_rate": 5.855576246637699e-05, "loss": 4.4366, "step": 2830 }, { "epoch": 0.17574026941461296, "grad_norm": 0.9090638051002342, "learning_rate": 5.8576453548520595e-05, "loss": 4.6085, "step": 2831 }, { "epoch": 0.17580234651437085, "grad_norm": 0.902854147558176, "learning_rate": 5.859714463066419e-05, "loss": 4.6093, "step": 2832 }, { "epoch": 0.17586442361412874, "grad_norm": 1.0273508862149665, "learning_rate": 5.861783571280778e-05, "loss": 4.5111, "step": 2833 }, { "epoch": 0.17592650071388666, "grad_norm": 1.0344640180655107, "learning_rate": 5.8638526794951374e-05, "loss": 4.5133, "step": 2834 }, { "epoch": 0.17598857781364455, "grad_norm": 1.068237186118455, "learning_rate": 5.8659217877094976e-05, "loss": 4.4937, "step": 2835 }, { "epoch": 0.17605065491340244, "grad_norm": 1.1336382782890657, "learning_rate": 5.867990895923857e-05, "loss": 4.3838, "step": 2836 }, { "epoch": 0.17611273201316036, "grad_norm": 1.1004061750966025, "learning_rate": 5.870060004138217e-05, "loss": 4.5314, "step": 2837 }, { "epoch": 0.17617480911291825, "grad_norm": 1.0226169820521789, "learning_rate": 5.872129112352577e-05, "loss": 4.4375, "step": 2838 }, { "epoch": 0.17623688621267614, "grad_norm": 1.0896456411913507, "learning_rate": 5.8741982205669356e-05, "loss": 4.4704, "step": 2839 }, { "epoch": 0.17629896331243405, "grad_norm": 0.9940744614083853, "learning_rate": 5.876267328781295e-05, "loss": 4.5359, "step": 2840 }, { "epoch": 0.17636104041219194, "grad_norm": 0.9709811819788997, "learning_rate": 5.878336436995655e-05, "loss": 4.5396, "step": 2841 }, { "epoch": 0.17642311751194983, "grad_norm": 1.0327502233338328, "learning_rate": 5.880405545210015e-05, "loss": 4.6207, "step": 2842 }, { "epoch": 0.17648519461170775, "grad_norm": 0.9253026312572666, "learning_rate": 5.882474653424375e-05, "loss": 4.507, "step": 2843 }, { "epoch": 0.17654727171146564, "grad_norm": 0.8918157280605462, "learning_rate": 5.8845437616387345e-05, "loss": 4.4227, "step": 2844 }, { "epoch": 0.17660934881122353, "grad_norm": 0.9674710167873335, "learning_rate": 5.886612869853093e-05, "loss": 4.3466, "step": 2845 }, { "epoch": 0.17667142591098145, "grad_norm": 0.9040622612814089, "learning_rate": 5.888681978067453e-05, "loss": 4.3062, "step": 2846 }, { "epoch": 0.17673350301073934, "grad_norm": 0.8325205335862734, "learning_rate": 5.890751086281813e-05, "loss": 4.6143, "step": 2847 }, { "epoch": 0.17679558011049723, "grad_norm": 0.933199587815656, "learning_rate": 5.8928201944961725e-05, "loss": 4.4061, "step": 2848 }, { "epoch": 0.17685765721025515, "grad_norm": 0.9932221252213144, "learning_rate": 5.894889302710532e-05, "loss": 4.5104, "step": 2849 }, { "epoch": 0.17691973431001304, "grad_norm": 1.164562809080313, "learning_rate": 5.896958410924892e-05, "loss": 4.508, "step": 2850 }, { "epoch": 0.17698181140977093, "grad_norm": 0.9763984542391402, "learning_rate": 5.899027519139251e-05, "loss": 4.3681, "step": 2851 }, { "epoch": 0.17704388850952885, "grad_norm": 0.9445966421860381, "learning_rate": 5.9010966273536105e-05, "loss": 4.5765, "step": 2852 }, { "epoch": 0.17710596560928674, "grad_norm": 1.013815985341869, "learning_rate": 5.90316573556797e-05, "loss": 4.3623, "step": 2853 }, { "epoch": 0.17716804270904463, "grad_norm": 1.0304985198260639, "learning_rate": 5.90523484378233e-05, "loss": 4.4638, "step": 2854 }, { "epoch": 0.17723011980880254, "grad_norm": 0.995197476338517, "learning_rate": 5.90730395199669e-05, "loss": 4.5102, "step": 2855 }, { "epoch": 0.17729219690856043, "grad_norm": 0.8984405849237967, "learning_rate": 5.90937306021105e-05, "loss": 4.4433, "step": 2856 }, { "epoch": 0.17735427400831832, "grad_norm": 0.8621247428903616, "learning_rate": 5.911442168425409e-05, "loss": 4.4732, "step": 2857 }, { "epoch": 0.17741635110807624, "grad_norm": 0.9693559474290182, "learning_rate": 5.913511276639768e-05, "loss": 4.4973, "step": 2858 }, { "epoch": 0.17747842820783413, "grad_norm": 0.9809388633249464, "learning_rate": 5.915580384854128e-05, "loss": 4.5383, "step": 2859 }, { "epoch": 0.17754050530759202, "grad_norm": 1.0761801274531875, "learning_rate": 5.917649493068488e-05, "loss": 4.4803, "step": 2860 }, { "epoch": 0.17760258240734994, "grad_norm": 1.1064766646250346, "learning_rate": 5.9197186012828475e-05, "loss": 4.5078, "step": 2861 }, { "epoch": 0.17766465950710783, "grad_norm": 0.8421872112436454, "learning_rate": 5.9217877094972076e-05, "loss": 4.475, "step": 2862 }, { "epoch": 0.17772673660686572, "grad_norm": 0.8047249424195313, "learning_rate": 5.923856817711566e-05, "loss": 4.3433, "step": 2863 }, { "epoch": 0.17778881370662364, "grad_norm": 0.8980949412222207, "learning_rate": 5.925925925925926e-05, "loss": 4.4587, "step": 2864 }, { "epoch": 0.17785089080638153, "grad_norm": 1.0009350577332674, "learning_rate": 5.9279950341402855e-05, "loss": 4.4382, "step": 2865 }, { "epoch": 0.17791296790613942, "grad_norm": 0.9022645461425884, "learning_rate": 5.930064142354646e-05, "loss": 4.474, "step": 2866 }, { "epoch": 0.17797504500589734, "grad_norm": 0.8092379744905728, "learning_rate": 5.932133250569005e-05, "loss": 4.4865, "step": 2867 }, { "epoch": 0.17803712210565523, "grad_norm": 0.8776472101871665, "learning_rate": 5.9342023587833654e-05, "loss": 4.5178, "step": 2868 }, { "epoch": 0.17809919920541312, "grad_norm": 0.7925023553614708, "learning_rate": 5.9362714669977235e-05, "loss": 4.5016, "step": 2869 }, { "epoch": 0.17816127630517103, "grad_norm": 0.7420860622754933, "learning_rate": 5.938340575212084e-05, "loss": 4.3497, "step": 2870 }, { "epoch": 0.17822335340492892, "grad_norm": 0.81506050569805, "learning_rate": 5.940409683426443e-05, "loss": 4.5287, "step": 2871 }, { "epoch": 0.17828543050468681, "grad_norm": 0.8401676351872982, "learning_rate": 5.9424787916408034e-05, "loss": 4.5238, "step": 2872 }, { "epoch": 0.17834750760444473, "grad_norm": 0.8376344942876385, "learning_rate": 5.944547899855163e-05, "loss": 4.3125, "step": 2873 }, { "epoch": 0.17840958470420262, "grad_norm": 0.8999997364149768, "learning_rate": 5.9466170080695224e-05, "loss": 4.5466, "step": 2874 }, { "epoch": 0.1784716618039605, "grad_norm": 1.0187176494493564, "learning_rate": 5.948686116283881e-05, "loss": 4.4532, "step": 2875 }, { "epoch": 0.17853373890371843, "grad_norm": 0.9728556355864219, "learning_rate": 5.9507552244982414e-05, "loss": 4.4096, "step": 2876 }, { "epoch": 0.17859581600347632, "grad_norm": 1.181128121199023, "learning_rate": 5.952824332712601e-05, "loss": 4.4397, "step": 2877 }, { "epoch": 0.1786578931032342, "grad_norm": 1.0206492056840772, "learning_rate": 5.9548934409269604e-05, "loss": 4.4788, "step": 2878 }, { "epoch": 0.17871997020299213, "grad_norm": 1.0903221365671443, "learning_rate": 5.9569625491413206e-05, "loss": 4.4481, "step": 2879 }, { "epoch": 0.17878204730275002, "grad_norm": 0.9184314616739131, "learning_rate": 5.95903165735568e-05, "loss": 4.3832, "step": 2880 }, { "epoch": 0.1788441244025079, "grad_norm": 1.0685676218322748, "learning_rate": 5.961100765570039e-05, "loss": 4.5614, "step": 2881 }, { "epoch": 0.17890620150226583, "grad_norm": 1.0610314486316377, "learning_rate": 5.963169873784399e-05, "loss": 4.3787, "step": 2882 }, { "epoch": 0.17896827860202372, "grad_norm": 1.0189418916240855, "learning_rate": 5.965238981998759e-05, "loss": 4.362, "step": 2883 }, { "epoch": 0.1790303557017816, "grad_norm": 1.0666221428062503, "learning_rate": 5.967308090213118e-05, "loss": 4.6497, "step": 2884 }, { "epoch": 0.17909243280153953, "grad_norm": 1.0255198370039276, "learning_rate": 5.9693771984274784e-05, "loss": 4.4525, "step": 2885 }, { "epoch": 0.17915450990129742, "grad_norm": 1.0777415270842845, "learning_rate": 5.971446306641838e-05, "loss": 4.3841, "step": 2886 }, { "epoch": 0.1792165870010553, "grad_norm": 1.0190495196992357, "learning_rate": 5.973515414856198e-05, "loss": 4.5021, "step": 2887 }, { "epoch": 0.17927866410081322, "grad_norm": 1.019060281881558, "learning_rate": 5.975584523070556e-05, "loss": 4.4146, "step": 2888 }, { "epoch": 0.1793407412005711, "grad_norm": 0.9305186363190577, "learning_rate": 5.9776536312849164e-05, "loss": 4.4947, "step": 2889 }, { "epoch": 0.179402818300329, "grad_norm": 0.9355570049971857, "learning_rate": 5.979722739499276e-05, "loss": 4.4063, "step": 2890 }, { "epoch": 0.17946489540008692, "grad_norm": 1.067095299866794, "learning_rate": 5.981791847713636e-05, "loss": 4.4897, "step": 2891 }, { "epoch": 0.1795269724998448, "grad_norm": 0.9363893924126984, "learning_rate": 5.9838609559279956e-05, "loss": 4.4529, "step": 2892 }, { "epoch": 0.1795890495996027, "grad_norm": 0.9648344525958767, "learning_rate": 5.985930064142356e-05, "loss": 4.4207, "step": 2893 }, { "epoch": 0.17965112669936062, "grad_norm": 1.0868241786207558, "learning_rate": 5.987999172356714e-05, "loss": 4.2968, "step": 2894 }, { "epoch": 0.1797132037991185, "grad_norm": 1.1113335956244417, "learning_rate": 5.990068280571074e-05, "loss": 4.3276, "step": 2895 }, { "epoch": 0.1797752808988764, "grad_norm": 0.9517562293171216, "learning_rate": 5.9921373887854336e-05, "loss": 4.5113, "step": 2896 }, { "epoch": 0.17983735799863432, "grad_norm": 0.9181175273697634, "learning_rate": 5.994206496999794e-05, "loss": 4.364, "step": 2897 }, { "epoch": 0.1798994350983922, "grad_norm": 1.1312736319081027, "learning_rate": 5.996275605214153e-05, "loss": 4.5629, "step": 2898 }, { "epoch": 0.1799615121981501, "grad_norm": 1.015655986606628, "learning_rate": 5.998344713428513e-05, "loss": 4.401, "step": 2899 }, { "epoch": 0.18002358929790802, "grad_norm": 1.2405881842620472, "learning_rate": 6.0004138216428716e-05, "loss": 4.5059, "step": 2900 }, { "epoch": 0.1800856663976659, "grad_norm": 0.9188925781685898, "learning_rate": 6.002482929857232e-05, "loss": 4.3781, "step": 2901 }, { "epoch": 0.1801477434974238, "grad_norm": 1.0149425267639405, "learning_rate": 6.0045520380715913e-05, "loss": 4.4145, "step": 2902 }, { "epoch": 0.1802098205971817, "grad_norm": 1.1167137368563254, "learning_rate": 6.006621146285951e-05, "loss": 4.4312, "step": 2903 }, { "epoch": 0.1802718976969396, "grad_norm": 0.8943430940067217, "learning_rate": 6.008690254500311e-05, "loss": 4.4425, "step": 2904 }, { "epoch": 0.1803339747966975, "grad_norm": 0.9525037523130663, "learning_rate": 6.0107593627146705e-05, "loss": 4.5109, "step": 2905 }, { "epoch": 0.18039605189645538, "grad_norm": 0.9397818452229846, "learning_rate": 6.0128284709290294e-05, "loss": 4.5075, "step": 2906 }, { "epoch": 0.1804581289962133, "grad_norm": 0.9269266246121405, "learning_rate": 6.0148975791433896e-05, "loss": 4.4965, "step": 2907 }, { "epoch": 0.1805202060959712, "grad_norm": 0.9714389521981813, "learning_rate": 6.016966687357749e-05, "loss": 4.475, "step": 2908 }, { "epoch": 0.18058228319572908, "grad_norm": 1.0400198634708278, "learning_rate": 6.0190357955721086e-05, "loss": 4.414, "step": 2909 }, { "epoch": 0.180644360295487, "grad_norm": 1.027143451615466, "learning_rate": 6.021104903786469e-05, "loss": 4.5254, "step": 2910 }, { "epoch": 0.1807064373952449, "grad_norm": 0.8291394839471701, "learning_rate": 6.023174012000828e-05, "loss": 4.2773, "step": 2911 }, { "epoch": 0.18076851449500278, "grad_norm": 0.8519604654183933, "learning_rate": 6.025243120215187e-05, "loss": 4.3745, "step": 2912 }, { "epoch": 0.1808305915947607, "grad_norm": 0.9418299817500585, "learning_rate": 6.0273122284295466e-05, "loss": 4.3421, "step": 2913 }, { "epoch": 0.1808926686945186, "grad_norm": 1.0159417465435678, "learning_rate": 6.029381336643907e-05, "loss": 4.4547, "step": 2914 }, { "epoch": 0.18095474579427648, "grad_norm": 0.9836633925745928, "learning_rate": 6.031450444858266e-05, "loss": 4.439, "step": 2915 }, { "epoch": 0.1810168228940344, "grad_norm": 0.8781527192434507, "learning_rate": 6.0335195530726265e-05, "loss": 4.4533, "step": 2916 }, { "epoch": 0.18107889999379229, "grad_norm": 0.9032187904838582, "learning_rate": 6.035588661286986e-05, "loss": 4.4432, "step": 2917 }, { "epoch": 0.18114097709355018, "grad_norm": 0.8589624281538987, "learning_rate": 6.037657769501345e-05, "loss": 4.4119, "step": 2918 }, { "epoch": 0.1812030541933081, "grad_norm": 0.725279259337054, "learning_rate": 6.039726877715704e-05, "loss": 4.3774, "step": 2919 }, { "epoch": 0.18126513129306598, "grad_norm": 0.808991891631715, "learning_rate": 6.0417959859300645e-05, "loss": 4.3142, "step": 2920 }, { "epoch": 0.18132720839282387, "grad_norm": 0.7750246351695079, "learning_rate": 6.043865094144424e-05, "loss": 4.449, "step": 2921 }, { "epoch": 0.1813892854925818, "grad_norm": 0.8073701307661089, "learning_rate": 6.045934202358784e-05, "loss": 4.4243, "step": 2922 }, { "epoch": 0.18145136259233968, "grad_norm": 0.8815494718403684, "learning_rate": 6.048003310573144e-05, "loss": 4.4209, "step": 2923 }, { "epoch": 0.18151343969209757, "grad_norm": 0.8232613376910587, "learning_rate": 6.0500724187875025e-05, "loss": 4.4879, "step": 2924 }, { "epoch": 0.1815755167918555, "grad_norm": 0.7585975670448389, "learning_rate": 6.052141527001862e-05, "loss": 4.4012, "step": 2925 }, { "epoch": 0.18163759389161338, "grad_norm": 0.6547737092134778, "learning_rate": 6.054210635216222e-05, "loss": 4.4707, "step": 2926 }, { "epoch": 0.18169967099137127, "grad_norm": 0.8316385838029788, "learning_rate": 6.056279743430582e-05, "loss": 4.4919, "step": 2927 }, { "epoch": 0.1817617480911292, "grad_norm": 0.9068392449342685, "learning_rate": 6.058348851644942e-05, "loss": 4.3815, "step": 2928 }, { "epoch": 0.18182382519088708, "grad_norm": 0.9588008479693486, "learning_rate": 6.0604179598593014e-05, "loss": 4.344, "step": 2929 }, { "epoch": 0.18188590229064497, "grad_norm": 0.9661365141851852, "learning_rate": 6.06248706807366e-05, "loss": 4.4054, "step": 2930 }, { "epoch": 0.1819479793904029, "grad_norm": 1.0643293676265624, "learning_rate": 6.06455617628802e-05, "loss": 4.3732, "step": 2931 }, { "epoch": 0.18201005649016078, "grad_norm": 1.0233475157803893, "learning_rate": 6.06662528450238e-05, "loss": 4.4786, "step": 2932 }, { "epoch": 0.18207213358991867, "grad_norm": 1.1200397068886996, "learning_rate": 6.0686943927167395e-05, "loss": 4.3835, "step": 2933 }, { "epoch": 0.18213421068967658, "grad_norm": 0.8717329476966639, "learning_rate": 6.070763500931099e-05, "loss": 4.4342, "step": 2934 }, { "epoch": 0.18219628778943447, "grad_norm": 0.8346995876041142, "learning_rate": 6.072832609145459e-05, "loss": 4.5164, "step": 2935 }, { "epoch": 0.18225836488919236, "grad_norm": 0.9100818521765681, "learning_rate": 6.074901717359818e-05, "loss": 4.3476, "step": 2936 }, { "epoch": 0.18232044198895028, "grad_norm": 1.048374070637997, "learning_rate": 6.0769708255741775e-05, "loss": 4.3967, "step": 2937 }, { "epoch": 0.18238251908870817, "grad_norm": 1.16077398943097, "learning_rate": 6.079039933788537e-05, "loss": 4.5139, "step": 2938 }, { "epoch": 0.18244459618846606, "grad_norm": 0.8486868223292527, "learning_rate": 6.081109042002897e-05, "loss": 4.3713, "step": 2939 }, { "epoch": 0.18250667328822398, "grad_norm": 0.9574615445395199, "learning_rate": 6.083178150217257e-05, "loss": 4.2476, "step": 2940 }, { "epoch": 0.18256875038798187, "grad_norm": 1.0503711680195194, "learning_rate": 6.085247258431617e-05, "loss": 4.3634, "step": 2941 }, { "epoch": 0.18263082748773976, "grad_norm": 1.1833716457952437, "learning_rate": 6.087316366645976e-05, "loss": 4.4282, "step": 2942 }, { "epoch": 0.18269290458749768, "grad_norm": 0.8818318144067163, "learning_rate": 6.089385474860335e-05, "loss": 4.2684, "step": 2943 }, { "epoch": 0.18275498168725557, "grad_norm": 0.9528019467048072, "learning_rate": 6.091454583074695e-05, "loss": 4.465, "step": 2944 }, { "epoch": 0.18281705878701346, "grad_norm": 0.913320843676168, "learning_rate": 6.093523691289055e-05, "loss": 4.3177, "step": 2945 }, { "epoch": 0.18287913588677138, "grad_norm": 0.9531573618413397, "learning_rate": 6.0955927995034144e-05, "loss": 4.4827, "step": 2946 }, { "epoch": 0.18294121298652927, "grad_norm": 1.0891973841586382, "learning_rate": 6.0976619077177746e-05, "loss": 4.3494, "step": 2947 }, { "epoch": 0.18300329008628716, "grad_norm": 0.950430453806852, "learning_rate": 6.099731015932133e-05, "loss": 4.4943, "step": 2948 }, { "epoch": 0.18306536718604507, "grad_norm": 0.937918792009311, "learning_rate": 6.101800124146493e-05, "loss": 4.3833, "step": 2949 }, { "epoch": 0.18312744428580296, "grad_norm": 1.018967395826144, "learning_rate": 6.103869232360852e-05, "loss": 4.4874, "step": 2950 }, { "epoch": 0.18318952138556086, "grad_norm": 1.0907337560779666, "learning_rate": 6.105938340575212e-05, "loss": 4.3547, "step": 2951 }, { "epoch": 0.18325159848531877, "grad_norm": 0.9196741049876952, "learning_rate": 6.108007448789573e-05, "loss": 4.2783, "step": 2952 }, { "epoch": 0.18331367558507666, "grad_norm": 0.8738171893523924, "learning_rate": 6.110076557003932e-05, "loss": 4.4287, "step": 2953 }, { "epoch": 0.18337575268483455, "grad_norm": 0.9561025044501286, "learning_rate": 6.11214566521829e-05, "loss": 4.4367, "step": 2954 }, { "epoch": 0.18343782978459247, "grad_norm": 1.418949345099225, "learning_rate": 6.11421477343265e-05, "loss": 4.3593, "step": 2955 }, { "epoch": 0.18349990688435036, "grad_norm": 1.0292426136632977, "learning_rate": 6.116283881647011e-05, "loss": 4.4891, "step": 2956 }, { "epoch": 0.18356198398410825, "grad_norm": 1.102939367479823, "learning_rate": 6.11835298986137e-05, "loss": 4.4629, "step": 2957 }, { "epoch": 0.18362406108386617, "grad_norm": 1.077464689309011, "learning_rate": 6.12042209807573e-05, "loss": 4.4481, "step": 2958 }, { "epoch": 0.18368613818362406, "grad_norm": 1.316709896577101, "learning_rate": 6.12249120629009e-05, "loss": 4.3791, "step": 2959 }, { "epoch": 0.18374821528338195, "grad_norm": 0.7705105586903493, "learning_rate": 6.124560314504449e-05, "loss": 4.3634, "step": 2960 }, { "epoch": 0.18381029238313987, "grad_norm": 0.961211250823838, "learning_rate": 6.126629422718808e-05, "loss": 4.477, "step": 2961 }, { "epoch": 0.18387236948289776, "grad_norm": 1.1799698043451197, "learning_rate": 6.128698530933168e-05, "loss": 4.485, "step": 2962 }, { "epoch": 0.18393444658265565, "grad_norm": 0.8647072867889265, "learning_rate": 6.130767639147527e-05, "loss": 4.4644, "step": 2963 }, { "epoch": 0.18399652368241357, "grad_norm": 0.9554112639331961, "learning_rate": 6.132836747361887e-05, "loss": 4.476, "step": 2964 }, { "epoch": 0.18405860078217146, "grad_norm": 1.0507664448996217, "learning_rate": 6.134905855576248e-05, "loss": 4.5105, "step": 2965 }, { "epoch": 0.18412067788192935, "grad_norm": 1.032075551516647, "learning_rate": 6.136974963790606e-05, "loss": 4.3253, "step": 2966 }, { "epoch": 0.18418275498168726, "grad_norm": 0.8323387330027385, "learning_rate": 6.139044072004965e-05, "loss": 4.3963, "step": 2967 }, { "epoch": 0.18424483208144515, "grad_norm": 1.0761898752358339, "learning_rate": 6.141113180219325e-05, "loss": 4.4387, "step": 2968 }, { "epoch": 0.18430690918120304, "grad_norm": 0.9462667015748458, "learning_rate": 6.143182288433686e-05, "loss": 4.3526, "step": 2969 }, { "epoch": 0.18436898628096096, "grad_norm": 0.7890736040901469, "learning_rate": 6.145251396648045e-05, "loss": 4.4939, "step": 2970 }, { "epoch": 0.18443106338071885, "grad_norm": 0.9615435666168847, "learning_rate": 6.147320504862405e-05, "loss": 4.4407, "step": 2971 }, { "epoch": 0.18449314048047674, "grad_norm": 1.0692694377688767, "learning_rate": 6.149389613076764e-05, "loss": 4.4151, "step": 2972 }, { "epoch": 0.18455521758023466, "grad_norm": 0.8737205961025017, "learning_rate": 6.151458721291124e-05, "loss": 4.3874, "step": 2973 }, { "epoch": 0.18461729467999255, "grad_norm": 1.0298938503936284, "learning_rate": 6.153527829505483e-05, "loss": 4.4937, "step": 2974 }, { "epoch": 0.18467937177975044, "grad_norm": 1.1322965828310105, "learning_rate": 6.155596937719843e-05, "loss": 4.4682, "step": 2975 }, { "epoch": 0.18474144887950836, "grad_norm": 0.927697656808328, "learning_rate": 6.157666045934202e-05, "loss": 4.3145, "step": 2976 }, { "epoch": 0.18480352597926625, "grad_norm": 1.091198206018696, "learning_rate": 6.159735154148563e-05, "loss": 4.374, "step": 2977 }, { "epoch": 0.18486560307902414, "grad_norm": 1.0774952251567502, "learning_rate": 6.161804262362923e-05, "loss": 4.4102, "step": 2978 }, { "epoch": 0.18492768017878206, "grad_norm": 1.0075854260030088, "learning_rate": 6.163873370577281e-05, "loss": 4.3003, "step": 2979 }, { "epoch": 0.18498975727853995, "grad_norm": 1.0167413884725163, "learning_rate": 6.16594247879164e-05, "loss": 4.4325, "step": 2980 }, { "epoch": 0.18505183437829784, "grad_norm": 1.0785823902215503, "learning_rate": 6.168011587006001e-05, "loss": 4.4184, "step": 2981 }, { "epoch": 0.18511391147805575, "grad_norm": 1.0715114811198958, "learning_rate": 6.170080695220361e-05, "loss": 4.3684, "step": 2982 }, { "epoch": 0.18517598857781364, "grad_norm": 0.9791525711235414, "learning_rate": 6.17214980343472e-05, "loss": 4.3015, "step": 2983 }, { "epoch": 0.18523806567757153, "grad_norm": 0.8337972819749063, "learning_rate": 6.17421891164908e-05, "loss": 4.3445, "step": 2984 }, { "epoch": 0.18530014277732945, "grad_norm": 0.8889787751851249, "learning_rate": 6.176288019863439e-05, "loss": 4.4141, "step": 2985 }, { "epoch": 0.18536221987708734, "grad_norm": 1.0600663531426633, "learning_rate": 6.178357128077799e-05, "loss": 4.401, "step": 2986 }, { "epoch": 0.18542429697684523, "grad_norm": 0.9005097746882496, "learning_rate": 6.180426236292158e-05, "loss": 4.2979, "step": 2987 }, { "epoch": 0.18548637407660315, "grad_norm": 0.9181712150498003, "learning_rate": 6.182495344506518e-05, "loss": 4.4144, "step": 2988 }, { "epoch": 0.18554845117636104, "grad_norm": 0.8827089232051974, "learning_rate": 6.184564452720877e-05, "loss": 4.3863, "step": 2989 }, { "epoch": 0.18561052827611893, "grad_norm": 0.9157388102133907, "learning_rate": 6.186633560935238e-05, "loss": 4.2159, "step": 2990 }, { "epoch": 0.18567260537587685, "grad_norm": 1.0512221875406835, "learning_rate": 6.188702669149596e-05, "loss": 4.3344, "step": 2991 }, { "epoch": 0.18573468247563474, "grad_norm": 1.0191007560280239, "learning_rate": 6.190771777363956e-05, "loss": 4.4622, "step": 2992 }, { "epoch": 0.18579675957539263, "grad_norm": 1.1468535418672823, "learning_rate": 6.192840885578315e-05, "loss": 4.3742, "step": 2993 }, { "epoch": 0.18585883667515055, "grad_norm": 0.9224730669137248, "learning_rate": 6.194909993792676e-05, "loss": 4.3935, "step": 2994 }, { "epoch": 0.18592091377490844, "grad_norm": 0.7854872807566258, "learning_rate": 6.196979102007036e-05, "loss": 4.4082, "step": 2995 }, { "epoch": 0.18598299087466633, "grad_norm": 0.9853197438572353, "learning_rate": 6.199048210221395e-05, "loss": 4.3584, "step": 2996 }, { "epoch": 0.18604506797442424, "grad_norm": 1.137186275438539, "learning_rate": 6.201117318435755e-05, "loss": 4.4093, "step": 2997 }, { "epoch": 0.18610714507418213, "grad_norm": 0.8616437960663276, "learning_rate": 6.203186426650114e-05, "loss": 4.3074, "step": 2998 }, { "epoch": 0.18616922217394002, "grad_norm": 0.9686689958238089, "learning_rate": 6.205255534864474e-05, "loss": 4.3736, "step": 2999 }, { "epoch": 0.18623129927369794, "grad_norm": 0.9701979184030265, "learning_rate": 6.207324643078833e-05, "loss": 4.3452, "step": 3000 }, { "epoch": 0.18629337637345583, "grad_norm": 0.907051192663212, "learning_rate": 6.209393751293193e-05, "loss": 4.474, "step": 3001 }, { "epoch": 0.18635545347321372, "grad_norm": 0.9811550337400606, "learning_rate": 6.211462859507554e-05, "loss": 4.3542, "step": 3002 }, { "epoch": 0.18641753057297164, "grad_norm": 0.9633933838163936, "learning_rate": 6.213531967721912e-05, "loss": 4.3345, "step": 3003 }, { "epoch": 0.18647960767272953, "grad_norm": 0.8978811448779292, "learning_rate": 6.215601075936271e-05, "loss": 4.4094, "step": 3004 }, { "epoch": 0.18654168477248742, "grad_norm": 0.9660769161242776, "learning_rate": 6.217670184150631e-05, "loss": 4.3229, "step": 3005 }, { "epoch": 0.18660376187224534, "grad_norm": 0.9433872068734226, "learning_rate": 6.219739292364992e-05, "loss": 4.3591, "step": 3006 }, { "epoch": 0.18666583897200323, "grad_norm": 0.94020890826233, "learning_rate": 6.221808400579351e-05, "loss": 4.3678, "step": 3007 }, { "epoch": 0.18672791607176112, "grad_norm": 0.9513603509364811, "learning_rate": 6.22387750879371e-05, "loss": 4.3742, "step": 3008 }, { "epoch": 0.18678999317151904, "grad_norm": 1.0372820062156198, "learning_rate": 6.225946617008069e-05, "loss": 4.3802, "step": 3009 }, { "epoch": 0.18685207027127693, "grad_norm": 1.110316776747322, "learning_rate": 6.22801572522243e-05, "loss": 4.3705, "step": 3010 }, { "epoch": 0.18691414737103482, "grad_norm": 0.7854746082968458, "learning_rate": 6.230084833436789e-05, "loss": 4.3578, "step": 3011 }, { "epoch": 0.18697622447079273, "grad_norm": 0.834021145369597, "learning_rate": 6.232153941651149e-05, "loss": 4.3979, "step": 3012 }, { "epoch": 0.18703830157055062, "grad_norm": 0.9280188194701503, "learning_rate": 6.234223049865508e-05, "loss": 4.4878, "step": 3013 }, { "epoch": 0.18710037867030851, "grad_norm": 0.9037638796931625, "learning_rate": 6.236292158079868e-05, "loss": 4.2586, "step": 3014 }, { "epoch": 0.18716245577006643, "grad_norm": 0.9052653389478974, "learning_rate": 6.238361266294227e-05, "loss": 4.2694, "step": 3015 }, { "epoch": 0.18722453286982432, "grad_norm": 1.0454255503376737, "learning_rate": 6.240430374508587e-05, "loss": 4.3723, "step": 3016 }, { "epoch": 0.1872866099695822, "grad_norm": 0.9752391546331357, "learning_rate": 6.242499482722946e-05, "loss": 4.3835, "step": 3017 }, { "epoch": 0.18734868706934013, "grad_norm": 1.1439921701169993, "learning_rate": 6.244568590937306e-05, "loss": 4.4142, "step": 3018 }, { "epoch": 0.18741076416909802, "grad_norm": 0.926927331950591, "learning_rate": 6.246637699151667e-05, "loss": 4.4302, "step": 3019 }, { "epoch": 0.1874728412688559, "grad_norm": 0.8898277730068437, "learning_rate": 6.248706807366026e-05, "loss": 4.3853, "step": 3020 }, { "epoch": 0.18753491836861383, "grad_norm": 0.8907791472769668, "learning_rate": 6.250775915580384e-05, "loss": 4.3672, "step": 3021 }, { "epoch": 0.18759699546837172, "grad_norm": 0.8630142144129896, "learning_rate": 6.252845023794745e-05, "loss": 4.3355, "step": 3022 }, { "epoch": 0.1876590725681296, "grad_norm": 0.9566858407889204, "learning_rate": 6.254914132009105e-05, "loss": 4.4771, "step": 3023 }, { "epoch": 0.18772114966788753, "grad_norm": 0.8790656390305286, "learning_rate": 6.256983240223464e-05, "loss": 4.3698, "step": 3024 }, { "epoch": 0.18778322676764542, "grad_norm": 0.8452859955473613, "learning_rate": 6.259052348437824e-05, "loss": 4.3761, "step": 3025 }, { "epoch": 0.1878453038674033, "grad_norm": 0.8918460707059365, "learning_rate": 6.261121456652183e-05, "loss": 4.3119, "step": 3026 }, { "epoch": 0.18790738096716122, "grad_norm": 1.0840908238836406, "learning_rate": 6.263190564866543e-05, "loss": 4.2935, "step": 3027 }, { "epoch": 0.18796945806691912, "grad_norm": 0.9185337357075202, "learning_rate": 6.265259673080902e-05, "loss": 4.4788, "step": 3028 }, { "epoch": 0.188031535166677, "grad_norm": 1.2215236500134226, "learning_rate": 6.267328781295262e-05, "loss": 4.4629, "step": 3029 }, { "epoch": 0.18809361226643492, "grad_norm": 0.9459639267243531, "learning_rate": 6.269397889509621e-05, "loss": 4.3401, "step": 3030 }, { "epoch": 0.1881556893661928, "grad_norm": 0.9371568687650877, "learning_rate": 6.271466997723982e-05, "loss": 4.4178, "step": 3031 }, { "epoch": 0.1882177664659507, "grad_norm": 1.0653691541222008, "learning_rate": 6.273536105938342e-05, "loss": 4.4901, "step": 3032 }, { "epoch": 0.18827984356570862, "grad_norm": 0.9989879672707167, "learning_rate": 6.2756052141527e-05, "loss": 4.3868, "step": 3033 }, { "epoch": 0.1883419206654665, "grad_norm": 0.9834672284428605, "learning_rate": 6.277674322367059e-05, "loss": 4.446, "step": 3034 }, { "epoch": 0.1884039977652244, "grad_norm": 1.1083930599401488, "learning_rate": 6.27974343058142e-05, "loss": 4.4534, "step": 3035 }, { "epoch": 0.18846607486498232, "grad_norm": 1.045404910805383, "learning_rate": 6.28181253879578e-05, "loss": 4.3457, "step": 3036 }, { "epoch": 0.1885281519647402, "grad_norm": 1.024847849741577, "learning_rate": 6.283881647010139e-05, "loss": 4.3714, "step": 3037 }, { "epoch": 0.1885902290644981, "grad_norm": 0.8253320343698604, "learning_rate": 6.285950755224499e-05, "loss": 4.3562, "step": 3038 }, { "epoch": 0.18865230616425602, "grad_norm": 1.136400682972532, "learning_rate": 6.288019863438858e-05, "loss": 4.3779, "step": 3039 }, { "epoch": 0.1887143832640139, "grad_norm": 1.0473013052576827, "learning_rate": 6.290088971653218e-05, "loss": 4.4027, "step": 3040 }, { "epoch": 0.1887764603637718, "grad_norm": 0.8923790373694156, "learning_rate": 6.292158079867577e-05, "loss": 4.3141, "step": 3041 }, { "epoch": 0.18883853746352972, "grad_norm": 0.892110689668182, "learning_rate": 6.294227188081937e-05, "loss": 4.3366, "step": 3042 }, { "epoch": 0.1889006145632876, "grad_norm": 0.9728473337752057, "learning_rate": 6.296296296296296e-05, "loss": 4.3888, "step": 3043 }, { "epoch": 0.1889626916630455, "grad_norm": 0.8399136181534534, "learning_rate": 6.298365404510657e-05, "loss": 4.2783, "step": 3044 }, { "epoch": 0.1890247687628034, "grad_norm": 0.9461852214515425, "learning_rate": 6.300434512725015e-05, "loss": 4.2317, "step": 3045 }, { "epoch": 0.1890868458625613, "grad_norm": 1.3281967592487913, "learning_rate": 6.302503620939375e-05, "loss": 4.284, "step": 3046 }, { "epoch": 0.1891489229623192, "grad_norm": 0.733771502295104, "learning_rate": 6.304572729153736e-05, "loss": 4.3413, "step": 3047 }, { "epoch": 0.1892110000620771, "grad_norm": 0.7720766522248329, "learning_rate": 6.306641837368095e-05, "loss": 4.3108, "step": 3048 }, { "epoch": 0.189273077161835, "grad_norm": 1.1256390981516766, "learning_rate": 6.308710945582455e-05, "loss": 4.4393, "step": 3049 }, { "epoch": 0.1893351542615929, "grad_norm": 1.0271482680546007, "learning_rate": 6.310780053796814e-05, "loss": 4.3677, "step": 3050 }, { "epoch": 0.1893972313613508, "grad_norm": 0.9431790639601038, "learning_rate": 6.312849162011174e-05, "loss": 4.3897, "step": 3051 }, { "epoch": 0.1894593084611087, "grad_norm": 0.9794158449401326, "learning_rate": 6.314918270225533e-05, "loss": 4.4024, "step": 3052 }, { "epoch": 0.1895213855608666, "grad_norm": 1.0289303092489785, "learning_rate": 6.316987378439893e-05, "loss": 4.4447, "step": 3053 }, { "epoch": 0.1895834626606245, "grad_norm": 1.1880806457010367, "learning_rate": 6.319056486654252e-05, "loss": 4.3722, "step": 3054 }, { "epoch": 0.1896455397603824, "grad_norm": 0.8461580176350665, "learning_rate": 6.321125594868612e-05, "loss": 4.4408, "step": 3055 }, { "epoch": 0.1897076168601403, "grad_norm": 0.8169182445050798, "learning_rate": 6.323194703082972e-05, "loss": 4.4143, "step": 3056 }, { "epoch": 0.1897696939598982, "grad_norm": 0.7570801568546396, "learning_rate": 6.325263811297332e-05, "loss": 4.3747, "step": 3057 }, { "epoch": 0.1898317710596561, "grad_norm": 0.742094817146777, "learning_rate": 6.32733291951169e-05, "loss": 4.3331, "step": 3058 }, { "epoch": 0.18989384815941399, "grad_norm": 0.6878554768981419, "learning_rate": 6.32940202772605e-05, "loss": 4.3536, "step": 3059 }, { "epoch": 0.1899559252591719, "grad_norm": 0.8017088193246793, "learning_rate": 6.33147113594041e-05, "loss": 4.2287, "step": 3060 }, { "epoch": 0.1900180023589298, "grad_norm": 0.855082699139151, "learning_rate": 6.33354024415477e-05, "loss": 4.3841, "step": 3061 }, { "epoch": 0.19008007945868768, "grad_norm": 0.7195668347368211, "learning_rate": 6.33560935236913e-05, "loss": 4.3101, "step": 3062 }, { "epoch": 0.1901421565584456, "grad_norm": 0.8390628821800115, "learning_rate": 6.337678460583489e-05, "loss": 4.272, "step": 3063 }, { "epoch": 0.1902042336582035, "grad_norm": 0.7516539139381242, "learning_rate": 6.339747568797849e-05, "loss": 4.2949, "step": 3064 }, { "epoch": 0.19026631075796138, "grad_norm": 0.6796520267744566, "learning_rate": 6.341816677012208e-05, "loss": 4.4105, "step": 3065 }, { "epoch": 0.1903283878577193, "grad_norm": 0.7308368265495491, "learning_rate": 6.343885785226568e-05, "loss": 4.4423, "step": 3066 }, { "epoch": 0.1903904649574772, "grad_norm": 0.6818733801678265, "learning_rate": 6.345954893440927e-05, "loss": 4.2462, "step": 3067 }, { "epoch": 0.19045254205723508, "grad_norm": 0.6762127243035133, "learning_rate": 6.348024001655287e-05, "loss": 4.3574, "step": 3068 }, { "epoch": 0.190514619156993, "grad_norm": 0.867530754819422, "learning_rate": 6.350093109869647e-05, "loss": 4.2453, "step": 3069 }, { "epoch": 0.1905766962567509, "grad_norm": 0.9119304342591602, "learning_rate": 6.352162218084006e-05, "loss": 4.2948, "step": 3070 }, { "epoch": 0.19063877335650878, "grad_norm": 1.0662499617188639, "learning_rate": 6.354231326298365e-05, "loss": 4.4221, "step": 3071 }, { "epoch": 0.1907008504562667, "grad_norm": 1.0074131492304896, "learning_rate": 6.356300434512726e-05, "loss": 4.3515, "step": 3072 }, { "epoch": 0.1907629275560246, "grad_norm": 0.9359085562534061, "learning_rate": 6.358369542727085e-05, "loss": 4.3021, "step": 3073 }, { "epoch": 0.19082500465578248, "grad_norm": 0.8934324420627006, "learning_rate": 6.360438650941445e-05, "loss": 4.2813, "step": 3074 }, { "epoch": 0.1908870817555404, "grad_norm": 0.8800376222436053, "learning_rate": 6.362507759155804e-05, "loss": 4.3552, "step": 3075 }, { "epoch": 0.19094915885529828, "grad_norm": 1.0807779814331697, "learning_rate": 6.364576867370164e-05, "loss": 4.4171, "step": 3076 }, { "epoch": 0.19101123595505617, "grad_norm": 1.0883979641669264, "learning_rate": 6.366645975584523e-05, "loss": 4.2206, "step": 3077 }, { "epoch": 0.1910733130548141, "grad_norm": 1.0382021446892242, "learning_rate": 6.368715083798883e-05, "loss": 4.3552, "step": 3078 }, { "epoch": 0.19113539015457198, "grad_norm": 0.9431254726785394, "learning_rate": 6.370784192013242e-05, "loss": 4.3956, "step": 3079 }, { "epoch": 0.19119746725432987, "grad_norm": 0.9584193225650547, "learning_rate": 6.372853300227602e-05, "loss": 4.3982, "step": 3080 }, { "epoch": 0.1912595443540878, "grad_norm": 1.1470507598826547, "learning_rate": 6.374922408441963e-05, "loss": 4.3781, "step": 3081 }, { "epoch": 0.19132162145384568, "grad_norm": 0.8863802024097952, "learning_rate": 6.376991516656321e-05, "loss": 4.2946, "step": 3082 }, { "epoch": 0.19138369855360357, "grad_norm": 0.8890898675175813, "learning_rate": 6.37906062487068e-05, "loss": 4.3734, "step": 3083 }, { "epoch": 0.1914457756533615, "grad_norm": 1.0659025910229787, "learning_rate": 6.38112973308504e-05, "loss": 4.3531, "step": 3084 }, { "epoch": 0.19150785275311938, "grad_norm": 0.9825381140074388, "learning_rate": 6.383198841299401e-05, "loss": 4.3476, "step": 3085 }, { "epoch": 0.19156992985287727, "grad_norm": 0.9831848821815529, "learning_rate": 6.38526794951376e-05, "loss": 4.3908, "step": 3086 }, { "epoch": 0.1916320069526352, "grad_norm": 1.0128681620244973, "learning_rate": 6.38733705772812e-05, "loss": 4.2951, "step": 3087 }, { "epoch": 0.19169408405239308, "grad_norm": 1.2409092306098262, "learning_rate": 6.389406165942478e-05, "loss": 4.4331, "step": 3088 }, { "epoch": 0.19175616115215097, "grad_norm": 0.9344280457509848, "learning_rate": 6.391475274156839e-05, "loss": 4.2441, "step": 3089 }, { "epoch": 0.19181823825190888, "grad_norm": 0.9162896891138596, "learning_rate": 6.393544382371198e-05, "loss": 4.2502, "step": 3090 }, { "epoch": 0.19188031535166677, "grad_norm": 1.2038996295883073, "learning_rate": 6.395613490585558e-05, "loss": 4.2404, "step": 3091 }, { "epoch": 0.19194239245142466, "grad_norm": 0.8916735583481702, "learning_rate": 6.397682598799917e-05, "loss": 4.3242, "step": 3092 }, { "epoch": 0.19200446955118255, "grad_norm": 0.9587875133073717, "learning_rate": 6.399751707014277e-05, "loss": 4.278, "step": 3093 }, { "epoch": 0.19206654665094047, "grad_norm": 0.9332432108748895, "learning_rate": 6.401820815228636e-05, "loss": 4.3918, "step": 3094 }, { "epoch": 0.19212862375069836, "grad_norm": 0.9321953103904889, "learning_rate": 6.403889923442996e-05, "loss": 4.3121, "step": 3095 }, { "epoch": 0.19219070085045625, "grad_norm": 0.9772716236410705, "learning_rate": 6.405959031657355e-05, "loss": 4.3235, "step": 3096 }, { "epoch": 0.19225277795021417, "grad_norm": 0.8565311889047171, "learning_rate": 6.408028139871716e-05, "loss": 4.3783, "step": 3097 }, { "epoch": 0.19231485504997206, "grad_norm": 1.0578360573245862, "learning_rate": 6.410097248086076e-05, "loss": 4.3858, "step": 3098 }, { "epoch": 0.19237693214972995, "grad_norm": 1.0499514455013914, "learning_rate": 6.412166356300435e-05, "loss": 4.2815, "step": 3099 }, { "epoch": 0.19243900924948787, "grad_norm": 1.0713048080080065, "learning_rate": 6.414235464514794e-05, "loss": 4.2712, "step": 3100 }, { "epoch": 0.19250108634924576, "grad_norm": 0.9255595203243249, "learning_rate": 6.416304572729154e-05, "loss": 4.3486, "step": 3101 }, { "epoch": 0.19256316344900365, "grad_norm": 0.8925449356820131, "learning_rate": 6.418373680943514e-05, "loss": 4.2782, "step": 3102 }, { "epoch": 0.19262524054876157, "grad_norm": 0.9348235071341056, "learning_rate": 6.420442789157873e-05, "loss": 4.3682, "step": 3103 }, { "epoch": 0.19268731764851946, "grad_norm": 1.035669221273496, "learning_rate": 6.422511897372233e-05, "loss": 4.3625, "step": 3104 }, { "epoch": 0.19274939474827735, "grad_norm": 0.8436119531800257, "learning_rate": 6.424581005586592e-05, "loss": 4.3267, "step": 3105 }, { "epoch": 0.19281147184803527, "grad_norm": 0.8216411547991459, "learning_rate": 6.426650113800952e-05, "loss": 4.3716, "step": 3106 }, { "epoch": 0.19287354894779316, "grad_norm": 0.7820095947194097, "learning_rate": 6.428719222015311e-05, "loss": 4.3868, "step": 3107 }, { "epoch": 0.19293562604755105, "grad_norm": 0.8815255701442677, "learning_rate": 6.430788330229671e-05, "loss": 4.2735, "step": 3108 }, { "epoch": 0.19299770314730896, "grad_norm": 0.866187025426097, "learning_rate": 6.43285743844403e-05, "loss": 4.2857, "step": 3109 }, { "epoch": 0.19305978024706685, "grad_norm": 0.7463259510386776, "learning_rate": 6.434926546658391e-05, "loss": 4.3473, "step": 3110 }, { "epoch": 0.19312185734682474, "grad_norm": 0.9322180088433233, "learning_rate": 6.436995654872751e-05, "loss": 4.2984, "step": 3111 }, { "epoch": 0.19318393444658266, "grad_norm": 0.9051451031396744, "learning_rate": 6.439064763087109e-05, "loss": 4.3513, "step": 3112 }, { "epoch": 0.19324601154634055, "grad_norm": 0.8036658087815092, "learning_rate": 6.441133871301468e-05, "loss": 4.4183, "step": 3113 }, { "epoch": 0.19330808864609844, "grad_norm": 0.8756180351682821, "learning_rate": 6.443202979515829e-05, "loss": 4.1646, "step": 3114 }, { "epoch": 0.19337016574585636, "grad_norm": 0.9871343358170686, "learning_rate": 6.445272087730189e-05, "loss": 4.3768, "step": 3115 }, { "epoch": 0.19343224284561425, "grad_norm": 0.9447337499080533, "learning_rate": 6.447341195944548e-05, "loss": 4.2761, "step": 3116 }, { "epoch": 0.19349431994537214, "grad_norm": 1.0054610861935978, "learning_rate": 6.449410304158908e-05, "loss": 4.4334, "step": 3117 }, { "epoch": 0.19355639704513006, "grad_norm": 1.0532378229668289, "learning_rate": 6.451479412373267e-05, "loss": 4.3217, "step": 3118 }, { "epoch": 0.19361847414488795, "grad_norm": 0.8465554257277196, "learning_rate": 6.453548520587627e-05, "loss": 4.3096, "step": 3119 }, { "epoch": 0.19368055124464584, "grad_norm": 0.9158457454144603, "learning_rate": 6.455617628801986e-05, "loss": 4.3167, "step": 3120 }, { "epoch": 0.19374262834440376, "grad_norm": 1.0750378624214352, "learning_rate": 6.457686737016346e-05, "loss": 4.2679, "step": 3121 }, { "epoch": 0.19380470544416165, "grad_norm": 1.1967500506883766, "learning_rate": 6.459755845230707e-05, "loss": 4.3923, "step": 3122 }, { "epoch": 0.19386678254391954, "grad_norm": 0.855574022199759, "learning_rate": 6.461824953445066e-05, "loss": 4.2262, "step": 3123 }, { "epoch": 0.19392885964367745, "grad_norm": 0.8924403846021409, "learning_rate": 6.463894061659424e-05, "loss": 4.3356, "step": 3124 }, { "epoch": 0.19399093674343534, "grad_norm": 1.2570810501352694, "learning_rate": 6.465963169873784e-05, "loss": 4.3239, "step": 3125 }, { "epoch": 0.19405301384319323, "grad_norm": 1.0973632530055561, "learning_rate": 6.468032278088145e-05, "loss": 4.2404, "step": 3126 }, { "epoch": 0.19411509094295115, "grad_norm": 1.020110101180394, "learning_rate": 6.470101386302504e-05, "loss": 4.3074, "step": 3127 }, { "epoch": 0.19417716804270904, "grad_norm": 0.973506387443728, "learning_rate": 6.472170494516864e-05, "loss": 4.3009, "step": 3128 }, { "epoch": 0.19423924514246693, "grad_norm": 1.1445180221678541, "learning_rate": 6.474239602731223e-05, "loss": 4.3744, "step": 3129 }, { "epoch": 0.19430132224222485, "grad_norm": 0.9077447534386951, "learning_rate": 6.476308710945583e-05, "loss": 4.2411, "step": 3130 }, { "epoch": 0.19436339934198274, "grad_norm": 1.048215150635493, "learning_rate": 6.478377819159942e-05, "loss": 4.3951, "step": 3131 }, { "epoch": 0.19442547644174063, "grad_norm": 1.0033028418215642, "learning_rate": 6.480446927374302e-05, "loss": 4.3579, "step": 3132 }, { "epoch": 0.19448755354149855, "grad_norm": 0.9635756649858186, "learning_rate": 6.482516035588661e-05, "loss": 4.474, "step": 3133 }, { "epoch": 0.19454963064125644, "grad_norm": 0.8686609894963924, "learning_rate": 6.484585143803021e-05, "loss": 4.3959, "step": 3134 }, { "epoch": 0.19461170774101433, "grad_norm": 0.8828194879575997, "learning_rate": 6.486654252017382e-05, "loss": 4.3246, "step": 3135 }, { "epoch": 0.19467378484077225, "grad_norm": 0.9489325888722151, "learning_rate": 6.48872336023174e-05, "loss": 4.3546, "step": 3136 }, { "epoch": 0.19473586194053014, "grad_norm": 0.9129930744785899, "learning_rate": 6.4907924684461e-05, "loss": 4.2697, "step": 3137 }, { "epoch": 0.19479793904028803, "grad_norm": 0.9344236763105216, "learning_rate": 6.492861576660459e-05, "loss": 4.2526, "step": 3138 }, { "epoch": 0.19486001614004594, "grad_norm": 1.0603052359280731, "learning_rate": 6.49493068487482e-05, "loss": 4.3402, "step": 3139 }, { "epoch": 0.19492209323980383, "grad_norm": 0.9393903747031369, "learning_rate": 6.496999793089179e-05, "loss": 4.2194, "step": 3140 }, { "epoch": 0.19498417033956172, "grad_norm": 0.8072701646135771, "learning_rate": 6.499068901303539e-05, "loss": 4.338, "step": 3141 }, { "epoch": 0.19504624743931964, "grad_norm": 0.8958853920884335, "learning_rate": 6.501138009517898e-05, "loss": 4.291, "step": 3142 }, { "epoch": 0.19510832453907753, "grad_norm": 0.8526985569839374, "learning_rate": 6.503207117732258e-05, "loss": 4.3356, "step": 3143 }, { "epoch": 0.19517040163883542, "grad_norm": 0.83644794316725, "learning_rate": 6.505276225946617e-05, "loss": 4.3376, "step": 3144 }, { "epoch": 0.19523247873859334, "grad_norm": 0.913864603383405, "learning_rate": 6.507345334160977e-05, "loss": 4.289, "step": 3145 }, { "epoch": 0.19529455583835123, "grad_norm": 0.7688824004621961, "learning_rate": 6.509414442375336e-05, "loss": 4.3339, "step": 3146 }, { "epoch": 0.19535663293810912, "grad_norm": 0.6910347883587339, "learning_rate": 6.511483550589697e-05, "loss": 4.2313, "step": 3147 }, { "epoch": 0.19541871003786704, "grad_norm": 0.7124253100013186, "learning_rate": 6.513552658804057e-05, "loss": 4.2783, "step": 3148 }, { "epoch": 0.19548078713762493, "grad_norm": 0.7973890329607104, "learning_rate": 6.515621767018415e-05, "loss": 4.2883, "step": 3149 }, { "epoch": 0.19554286423738282, "grad_norm": 0.8947563054858373, "learning_rate": 6.517690875232774e-05, "loss": 4.2247, "step": 3150 }, { "epoch": 0.19560494133714074, "grad_norm": 1.0487445819337404, "learning_rate": 6.519759983447135e-05, "loss": 4.2995, "step": 3151 }, { "epoch": 0.19566701843689863, "grad_norm": 1.0256308428628713, "learning_rate": 6.521829091661495e-05, "loss": 4.2381, "step": 3152 }, { "epoch": 0.19572909553665652, "grad_norm": 0.9244675147526695, "learning_rate": 6.523898199875854e-05, "loss": 4.2573, "step": 3153 }, { "epoch": 0.19579117263641443, "grad_norm": 0.984251741230931, "learning_rate": 6.525967308090214e-05, "loss": 4.2728, "step": 3154 }, { "epoch": 0.19585324973617232, "grad_norm": 1.3628322003853466, "learning_rate": 6.528036416304573e-05, "loss": 4.394, "step": 3155 }, { "epoch": 0.19591532683593021, "grad_norm": 0.8706901539058436, "learning_rate": 6.530105524518933e-05, "loss": 4.278, "step": 3156 }, { "epoch": 0.19597740393568813, "grad_norm": 0.8615950951344531, "learning_rate": 6.532174632733292e-05, "loss": 4.2963, "step": 3157 }, { "epoch": 0.19603948103544602, "grad_norm": 1.0793525231367462, "learning_rate": 6.534243740947652e-05, "loss": 4.2926, "step": 3158 }, { "epoch": 0.1961015581352039, "grad_norm": 0.9465128316155587, "learning_rate": 6.536312849162011e-05, "loss": 4.3566, "step": 3159 }, { "epoch": 0.19616363523496183, "grad_norm": 0.929698463182963, "learning_rate": 6.538381957376372e-05, "loss": 4.3058, "step": 3160 }, { "epoch": 0.19622571233471972, "grad_norm": 0.8226187122326674, "learning_rate": 6.54045106559073e-05, "loss": 4.2868, "step": 3161 }, { "epoch": 0.1962877894344776, "grad_norm": 0.9426800669161004, "learning_rate": 6.54252017380509e-05, "loss": 4.3729, "step": 3162 }, { "epoch": 0.19634986653423553, "grad_norm": 0.8259775641167594, "learning_rate": 6.544589282019449e-05, "loss": 4.3577, "step": 3163 }, { "epoch": 0.19641194363399342, "grad_norm": 0.8209854998312867, "learning_rate": 6.54665839023381e-05, "loss": 4.2843, "step": 3164 }, { "epoch": 0.1964740207337513, "grad_norm": 0.9503610589900086, "learning_rate": 6.54872749844817e-05, "loss": 4.2193, "step": 3165 }, { "epoch": 0.19653609783350923, "grad_norm": 0.9269865214058253, "learning_rate": 6.550796606662529e-05, "loss": 4.2275, "step": 3166 }, { "epoch": 0.19659817493326712, "grad_norm": 1.010415434426936, "learning_rate": 6.552865714876887e-05, "loss": 4.3842, "step": 3167 }, { "epoch": 0.196660252033025, "grad_norm": 1.0615408438875902, "learning_rate": 6.554934823091248e-05, "loss": 4.2977, "step": 3168 }, { "epoch": 0.19672232913278292, "grad_norm": 0.853701189413557, "learning_rate": 6.557003931305608e-05, "loss": 4.3062, "step": 3169 }, { "epoch": 0.19678440623254081, "grad_norm": 0.7551157445165143, "learning_rate": 6.559073039519967e-05, "loss": 4.3075, "step": 3170 }, { "epoch": 0.1968464833322987, "grad_norm": 0.8497117605670702, "learning_rate": 6.561142147734327e-05, "loss": 4.3371, "step": 3171 }, { "epoch": 0.19690856043205662, "grad_norm": 0.9599410993169671, "learning_rate": 6.563211255948688e-05, "loss": 4.2512, "step": 3172 }, { "epoch": 0.1969706375318145, "grad_norm": 1.0835481883002311, "learning_rate": 6.565280364163046e-05, "loss": 4.2479, "step": 3173 }, { "epoch": 0.1970327146315724, "grad_norm": 0.8549950738396863, "learning_rate": 6.567349472377405e-05, "loss": 4.1885, "step": 3174 }, { "epoch": 0.19709479173133032, "grad_norm": 0.8099479208793261, "learning_rate": 6.569418580591765e-05, "loss": 4.3036, "step": 3175 }, { "epoch": 0.1971568688310882, "grad_norm": 0.907652361824217, "learning_rate": 6.571487688806126e-05, "loss": 4.3693, "step": 3176 }, { "epoch": 0.1972189459308461, "grad_norm": 0.946094032243455, "learning_rate": 6.573556797020485e-05, "loss": 4.301, "step": 3177 }, { "epoch": 0.19728102303060402, "grad_norm": 0.9194630249952938, "learning_rate": 6.575625905234845e-05, "loss": 4.3524, "step": 3178 }, { "epoch": 0.1973431001303619, "grad_norm": 0.8072724904069585, "learning_rate": 6.577695013449203e-05, "loss": 4.3106, "step": 3179 }, { "epoch": 0.1974051772301198, "grad_norm": 0.9372997387936048, "learning_rate": 6.579764121663564e-05, "loss": 4.2949, "step": 3180 }, { "epoch": 0.19746725432987772, "grad_norm": 0.985949694864874, "learning_rate": 6.581833229877923e-05, "loss": 4.3755, "step": 3181 }, { "epoch": 0.1975293314296356, "grad_norm": 0.992715409778352, "learning_rate": 6.583902338092283e-05, "loss": 4.216, "step": 3182 }, { "epoch": 0.1975914085293935, "grad_norm": 0.9540856944768618, "learning_rate": 6.585971446306642e-05, "loss": 4.2199, "step": 3183 }, { "epoch": 0.19765348562915142, "grad_norm": 0.8186577592097475, "learning_rate": 6.588040554521002e-05, "loss": 4.347, "step": 3184 }, { "epoch": 0.1977155627289093, "grad_norm": 0.9285798727427093, "learning_rate": 6.590109662735361e-05, "loss": 4.1658, "step": 3185 }, { "epoch": 0.1977776398286672, "grad_norm": 1.0303227851222418, "learning_rate": 6.59217877094972e-05, "loss": 4.2361, "step": 3186 }, { "epoch": 0.1978397169284251, "grad_norm": 0.9804538558497399, "learning_rate": 6.59424787916408e-05, "loss": 4.2732, "step": 3187 }, { "epoch": 0.197901794028183, "grad_norm": 0.9692638019113361, "learning_rate": 6.59631698737844e-05, "loss": 4.2872, "step": 3188 }, { "epoch": 0.1979638711279409, "grad_norm": 0.8966569921321415, "learning_rate": 6.5983860955928e-05, "loss": 4.2808, "step": 3189 }, { "epoch": 0.1980259482276988, "grad_norm": 0.8008951617841251, "learning_rate": 6.60045520380716e-05, "loss": 4.2376, "step": 3190 }, { "epoch": 0.1980880253274567, "grad_norm": 0.8498880214420198, "learning_rate": 6.602524312021518e-05, "loss": 4.2969, "step": 3191 }, { "epoch": 0.1981501024272146, "grad_norm": 0.891714166255471, "learning_rate": 6.604593420235878e-05, "loss": 4.1827, "step": 3192 }, { "epoch": 0.1982121795269725, "grad_norm": 0.8162589784511032, "learning_rate": 6.606662528450239e-05, "loss": 4.1972, "step": 3193 }, { "epoch": 0.1982742566267304, "grad_norm": 0.9467153622355259, "learning_rate": 6.608731636664598e-05, "loss": 4.2796, "step": 3194 }, { "epoch": 0.1983363337264883, "grad_norm": 0.9115260871046034, "learning_rate": 6.610800744878958e-05, "loss": 4.1682, "step": 3195 }, { "epoch": 0.1983984108262462, "grad_norm": 0.9545393596285319, "learning_rate": 6.612869853093317e-05, "loss": 4.3244, "step": 3196 }, { "epoch": 0.1984604879260041, "grad_norm": 0.9519072711405989, "learning_rate": 6.614938961307677e-05, "loss": 4.2978, "step": 3197 }, { "epoch": 0.198522565025762, "grad_norm": 1.0205405782157675, "learning_rate": 6.617008069522036e-05, "loss": 4.2246, "step": 3198 }, { "epoch": 0.1985846421255199, "grad_norm": 0.9057255904814607, "learning_rate": 6.619077177736396e-05, "loss": 4.1139, "step": 3199 }, { "epoch": 0.1986467192252778, "grad_norm": 0.8058186700611776, "learning_rate": 6.621146285950755e-05, "loss": 4.2575, "step": 3200 }, { "epoch": 0.19870879632503569, "grad_norm": 0.9302411714733091, "learning_rate": 6.623215394165116e-05, "loss": 4.2968, "step": 3201 }, { "epoch": 0.1987708734247936, "grad_norm": 0.9345310884573471, "learning_rate": 6.625284502379475e-05, "loss": 4.2611, "step": 3202 }, { "epoch": 0.1988329505245515, "grad_norm": 1.0410186023204375, "learning_rate": 6.627353610593834e-05, "loss": 4.2222, "step": 3203 }, { "epoch": 0.19889502762430938, "grad_norm": 0.806945669210106, "learning_rate": 6.629422718808193e-05, "loss": 4.3491, "step": 3204 }, { "epoch": 0.1989571047240673, "grad_norm": 0.7511719448270086, "learning_rate": 6.631491827022554e-05, "loss": 4.2493, "step": 3205 }, { "epoch": 0.1990191818238252, "grad_norm": 1.024143586939801, "learning_rate": 6.633560935236914e-05, "loss": 4.2005, "step": 3206 }, { "epoch": 0.19908125892358308, "grad_norm": 0.9970160010898976, "learning_rate": 6.635630043451273e-05, "loss": 4.2265, "step": 3207 }, { "epoch": 0.199143336023341, "grad_norm": 0.8589098972231323, "learning_rate": 6.637699151665633e-05, "loss": 4.1283, "step": 3208 }, { "epoch": 0.1992054131230989, "grad_norm": 0.9690843128356086, "learning_rate": 6.639768259879992e-05, "loss": 4.2766, "step": 3209 }, { "epoch": 0.19926749022285678, "grad_norm": 1.06207154554907, "learning_rate": 6.641837368094352e-05, "loss": 4.3325, "step": 3210 }, { "epoch": 0.1993295673226147, "grad_norm": 1.0465757597080159, "learning_rate": 6.643906476308711e-05, "loss": 4.3141, "step": 3211 }, { "epoch": 0.1993916444223726, "grad_norm": 1.0352517137953154, "learning_rate": 6.64597558452307e-05, "loss": 4.2019, "step": 3212 }, { "epoch": 0.19945372152213048, "grad_norm": 0.9391063914776431, "learning_rate": 6.64804469273743e-05, "loss": 4.2102, "step": 3213 }, { "epoch": 0.1995157986218884, "grad_norm": 0.912382293320812, "learning_rate": 6.650113800951791e-05, "loss": 4.2967, "step": 3214 }, { "epoch": 0.1995778757216463, "grad_norm": 1.1297077119108692, "learning_rate": 6.652182909166149e-05, "loss": 4.3254, "step": 3215 }, { "epoch": 0.19963995282140418, "grad_norm": 0.9617980577258486, "learning_rate": 6.654252017380509e-05, "loss": 4.1893, "step": 3216 }, { "epoch": 0.1997020299211621, "grad_norm": 0.8862324864388189, "learning_rate": 6.656321125594868e-05, "loss": 4.1723, "step": 3217 }, { "epoch": 0.19976410702091998, "grad_norm": 0.9555887054975872, "learning_rate": 6.658390233809229e-05, "loss": 4.3089, "step": 3218 }, { "epoch": 0.19982618412067787, "grad_norm": 0.9120155629306664, "learning_rate": 6.660459342023588e-05, "loss": 4.3163, "step": 3219 }, { "epoch": 0.1998882612204358, "grad_norm": 0.7268264865183912, "learning_rate": 6.662528450237948e-05, "loss": 4.295, "step": 3220 }, { "epoch": 0.19995033832019368, "grad_norm": 0.885883021487778, "learning_rate": 6.664597558452307e-05, "loss": 4.2858, "step": 3221 }, { "epoch": 0.20001241541995157, "grad_norm": 1.0446664087903408, "learning_rate": 6.666666666666667e-05, "loss": 4.3439, "step": 3222 }, { "epoch": 0.2000744925197095, "grad_norm": 1.0219478686188554, "learning_rate": 6.668735774881026e-05, "loss": 4.1847, "step": 3223 }, { "epoch": 0.20013656961946738, "grad_norm": 1.0387525741600339, "learning_rate": 6.670804883095386e-05, "loss": 4.3755, "step": 3224 }, { "epoch": 0.20019864671922527, "grad_norm": 1.0086512660758407, "learning_rate": 6.672873991309746e-05, "loss": 4.3315, "step": 3225 }, { "epoch": 0.2002607238189832, "grad_norm": 0.8374383946435408, "learning_rate": 6.674943099524106e-05, "loss": 4.2438, "step": 3226 }, { "epoch": 0.20032280091874108, "grad_norm": 0.8475990451900078, "learning_rate": 6.677012207738466e-05, "loss": 4.3215, "step": 3227 }, { "epoch": 0.20038487801849897, "grad_norm": 0.9052675446582914, "learning_rate": 6.679081315952824e-05, "loss": 4.2993, "step": 3228 }, { "epoch": 0.2004469551182569, "grad_norm": 1.0345297014887147, "learning_rate": 6.681150424167184e-05, "loss": 4.2295, "step": 3229 }, { "epoch": 0.20050903221801478, "grad_norm": 1.011407988552825, "learning_rate": 6.683219532381544e-05, "loss": 4.1635, "step": 3230 }, { "epoch": 0.20057110931777267, "grad_norm": 1.0468220483853585, "learning_rate": 6.685288640595904e-05, "loss": 4.2129, "step": 3231 }, { "epoch": 0.20063318641753058, "grad_norm": 0.9002281919580045, "learning_rate": 6.687357748810263e-05, "loss": 4.2956, "step": 3232 }, { "epoch": 0.20069526351728847, "grad_norm": 0.7976674272462136, "learning_rate": 6.689426857024623e-05, "loss": 4.3384, "step": 3233 }, { "epoch": 0.20075734061704636, "grad_norm": 0.7966157079969093, "learning_rate": 6.691495965238982e-05, "loss": 4.3507, "step": 3234 }, { "epoch": 0.20081941771680428, "grad_norm": 0.8353290660969376, "learning_rate": 6.693565073453342e-05, "loss": 4.2626, "step": 3235 }, { "epoch": 0.20088149481656217, "grad_norm": 0.8067611345693643, "learning_rate": 6.695634181667701e-05, "loss": 4.3116, "step": 3236 }, { "epoch": 0.20094357191632006, "grad_norm": 0.8651129506074957, "learning_rate": 6.697703289882061e-05, "loss": 4.14, "step": 3237 }, { "epoch": 0.20100564901607798, "grad_norm": 0.7481454729483329, "learning_rate": 6.69977239809642e-05, "loss": 4.2629, "step": 3238 }, { "epoch": 0.20106772611583587, "grad_norm": 0.7386131146246664, "learning_rate": 6.701841506310781e-05, "loss": 4.2221, "step": 3239 }, { "epoch": 0.20112980321559376, "grad_norm": 0.8368849688072265, "learning_rate": 6.70391061452514e-05, "loss": 4.1801, "step": 3240 }, { "epoch": 0.20119188031535168, "grad_norm": 1.1115787270481954, "learning_rate": 6.705979722739499e-05, "loss": 4.2112, "step": 3241 }, { "epoch": 0.20125395741510957, "grad_norm": 1.0439289830414284, "learning_rate": 6.708048830953858e-05, "loss": 4.2352, "step": 3242 }, { "epoch": 0.20131603451486746, "grad_norm": 0.925529252534411, "learning_rate": 6.71011793916822e-05, "loss": 4.1664, "step": 3243 }, { "epoch": 0.20137811161462538, "grad_norm": 0.7472796377492357, "learning_rate": 6.712187047382579e-05, "loss": 4.2388, "step": 3244 }, { "epoch": 0.20144018871438327, "grad_norm": 1.0283812299256612, "learning_rate": 6.714256155596938e-05, "loss": 4.2201, "step": 3245 }, { "epoch": 0.20150226581414116, "grad_norm": 1.2030166478103088, "learning_rate": 6.716325263811298e-05, "loss": 4.2448, "step": 3246 }, { "epoch": 0.20156434291389907, "grad_norm": 0.7477600503932879, "learning_rate": 6.718394372025657e-05, "loss": 4.3101, "step": 3247 }, { "epoch": 0.20162642001365697, "grad_norm": 0.9246147307098598, "learning_rate": 6.720463480240017e-05, "loss": 4.2599, "step": 3248 }, { "epoch": 0.20168849711341486, "grad_norm": 0.981018003749614, "learning_rate": 6.722532588454376e-05, "loss": 4.215, "step": 3249 }, { "epoch": 0.20175057421317277, "grad_norm": 0.8652364760948141, "learning_rate": 6.724601696668736e-05, "loss": 4.2866, "step": 3250 }, { "epoch": 0.20181265131293066, "grad_norm": 0.8420084888682732, "learning_rate": 6.726670804883097e-05, "loss": 4.3053, "step": 3251 }, { "epoch": 0.20187472841268855, "grad_norm": 0.909656764159558, "learning_rate": 6.728739913097455e-05, "loss": 4.2543, "step": 3252 }, { "epoch": 0.20193680551244647, "grad_norm": 0.8146568427776048, "learning_rate": 6.730809021311814e-05, "loss": 4.2386, "step": 3253 }, { "epoch": 0.20199888261220436, "grad_norm": 0.6864873841370506, "learning_rate": 6.732878129526174e-05, "loss": 4.2807, "step": 3254 }, { "epoch": 0.20206095971196225, "grad_norm": 0.8399889443441999, "learning_rate": 6.734947237740535e-05, "loss": 4.2912, "step": 3255 }, { "epoch": 0.20212303681172017, "grad_norm": 0.7865717260875221, "learning_rate": 6.737016345954894e-05, "loss": 4.3138, "step": 3256 }, { "epoch": 0.20218511391147806, "grad_norm": 0.7206849300082974, "learning_rate": 6.739085454169254e-05, "loss": 4.1899, "step": 3257 }, { "epoch": 0.20224719101123595, "grad_norm": 0.6907609633129003, "learning_rate": 6.741154562383612e-05, "loss": 4.3106, "step": 3258 }, { "epoch": 0.20230926811099387, "grad_norm": 0.7465114522721089, "learning_rate": 6.743223670597973e-05, "loss": 4.246, "step": 3259 }, { "epoch": 0.20237134521075176, "grad_norm": 0.8112433692766949, "learning_rate": 6.745292778812332e-05, "loss": 4.2699, "step": 3260 }, { "epoch": 0.20243342231050965, "grad_norm": 1.0276902923351172, "learning_rate": 6.747361887026692e-05, "loss": 4.2442, "step": 3261 }, { "epoch": 0.20249549941026757, "grad_norm": 1.020001182555467, "learning_rate": 6.749430995241051e-05, "loss": 4.2418, "step": 3262 }, { "epoch": 0.20255757651002546, "grad_norm": 0.9736642172444254, "learning_rate": 6.751500103455411e-05, "loss": 4.2868, "step": 3263 }, { "epoch": 0.20261965360978335, "grad_norm": 0.963126782286305, "learning_rate": 6.75356921166977e-05, "loss": 4.3182, "step": 3264 }, { "epoch": 0.20268173070954126, "grad_norm": 1.0530178275886033, "learning_rate": 6.75563831988413e-05, "loss": 4.2138, "step": 3265 }, { "epoch": 0.20274380780929915, "grad_norm": 0.9991335393320466, "learning_rate": 6.75770742809849e-05, "loss": 4.2208, "step": 3266 }, { "epoch": 0.20280588490905704, "grad_norm": 0.9388963473009343, "learning_rate": 6.759776536312849e-05, "loss": 4.1639, "step": 3267 }, { "epoch": 0.20286796200881496, "grad_norm": 0.9656299368722877, "learning_rate": 6.76184564452721e-05, "loss": 4.1012, "step": 3268 }, { "epoch": 0.20293003910857285, "grad_norm": 1.0056723411517134, "learning_rate": 6.763914752741569e-05, "loss": 4.2964, "step": 3269 }, { "epoch": 0.20299211620833074, "grad_norm": 1.011833920790677, "learning_rate": 6.765983860955927e-05, "loss": 4.2118, "step": 3270 }, { "epoch": 0.20305419330808866, "grad_norm": 0.9341362689673516, "learning_rate": 6.768052969170288e-05, "loss": 4.2129, "step": 3271 }, { "epoch": 0.20311627040784655, "grad_norm": 0.9383750646288052, "learning_rate": 6.770122077384648e-05, "loss": 4.2874, "step": 3272 }, { "epoch": 0.20317834750760444, "grad_norm": 0.7853250654181512, "learning_rate": 6.772191185599007e-05, "loss": 4.2388, "step": 3273 }, { "epoch": 0.20324042460736236, "grad_norm": 0.8632126517680297, "learning_rate": 6.774260293813367e-05, "loss": 4.3206, "step": 3274 }, { "epoch": 0.20330250170712025, "grad_norm": 0.7818455524220254, "learning_rate": 6.776329402027726e-05, "loss": 4.2148, "step": 3275 }, { "epoch": 0.20336457880687814, "grad_norm": 0.8914053325999401, "learning_rate": 6.778398510242086e-05, "loss": 4.2269, "step": 3276 }, { "epoch": 0.20342665590663606, "grad_norm": 0.8005729218220877, "learning_rate": 6.780467618456445e-05, "loss": 4.1463, "step": 3277 }, { "epoch": 0.20348873300639395, "grad_norm": 0.9209467206738652, "learning_rate": 6.782536726670805e-05, "loss": 4.2618, "step": 3278 }, { "epoch": 0.20355081010615184, "grad_norm": 0.9631543833157337, "learning_rate": 6.784605834885164e-05, "loss": 4.2459, "step": 3279 }, { "epoch": 0.20361288720590973, "grad_norm": 1.0078824632593766, "learning_rate": 6.786674943099525e-05, "loss": 4.3128, "step": 3280 }, { "epoch": 0.20367496430566764, "grad_norm": 1.0932185380457469, "learning_rate": 6.788744051313885e-05, "loss": 4.2218, "step": 3281 }, { "epoch": 0.20373704140542553, "grad_norm": 1.010067450093509, "learning_rate": 6.790813159528243e-05, "loss": 4.2036, "step": 3282 }, { "epoch": 0.20379911850518342, "grad_norm": 0.9790205880358674, "learning_rate": 6.792882267742602e-05, "loss": 4.3251, "step": 3283 }, { "epoch": 0.20386119560494134, "grad_norm": 1.1686918641997364, "learning_rate": 6.794951375956963e-05, "loss": 4.2799, "step": 3284 }, { "epoch": 0.20392327270469923, "grad_norm": 0.9107999694544435, "learning_rate": 6.797020484171323e-05, "loss": 4.208, "step": 3285 }, { "epoch": 0.20398534980445712, "grad_norm": 0.9549574824550117, "learning_rate": 6.799089592385682e-05, "loss": 4.3392, "step": 3286 }, { "epoch": 0.20404742690421504, "grad_norm": 0.9857830996759789, "learning_rate": 6.801158700600042e-05, "loss": 4.1796, "step": 3287 }, { "epoch": 0.20410950400397293, "grad_norm": 0.8609550515824487, "learning_rate": 6.803227808814401e-05, "loss": 4.2121, "step": 3288 }, { "epoch": 0.20417158110373082, "grad_norm": 0.8329415155215079, "learning_rate": 6.805296917028761e-05, "loss": 4.1602, "step": 3289 }, { "epoch": 0.20423365820348874, "grad_norm": 0.9929676142940435, "learning_rate": 6.80736602524312e-05, "loss": 4.2982, "step": 3290 }, { "epoch": 0.20429573530324663, "grad_norm": 1.112922691025181, "learning_rate": 6.80943513345748e-05, "loss": 4.155, "step": 3291 }, { "epoch": 0.20435781240300452, "grad_norm": 0.8274511798851844, "learning_rate": 6.81150424167184e-05, "loss": 4.0917, "step": 3292 }, { "epoch": 0.20441988950276244, "grad_norm": 0.9186168249755076, "learning_rate": 6.8135733498862e-05, "loss": 4.2051, "step": 3293 }, { "epoch": 0.20448196660252033, "grad_norm": 0.9119844534891559, "learning_rate": 6.815642458100558e-05, "loss": 4.2226, "step": 3294 }, { "epoch": 0.20454404370227822, "grad_norm": 0.9102350618351487, "learning_rate": 6.817711566314918e-05, "loss": 4.2916, "step": 3295 }, { "epoch": 0.20460612080203613, "grad_norm": 0.9712708499688313, "learning_rate": 6.819780674529279e-05, "loss": 4.2693, "step": 3296 }, { "epoch": 0.20466819790179402, "grad_norm": 0.9685490615167472, "learning_rate": 6.821849782743638e-05, "loss": 4.2764, "step": 3297 }, { "epoch": 0.20473027500155191, "grad_norm": 0.9733146369512325, "learning_rate": 6.823918890957998e-05, "loss": 4.097, "step": 3298 }, { "epoch": 0.20479235210130983, "grad_norm": 1.0194707976794466, "learning_rate": 6.825987999172357e-05, "loss": 4.0259, "step": 3299 }, { "epoch": 0.20485442920106772, "grad_norm": 0.9039879883098976, "learning_rate": 6.828057107386717e-05, "loss": 4.1941, "step": 3300 }, { "epoch": 0.2049165063008256, "grad_norm": 1.0578999515598704, "learning_rate": 6.830126215601076e-05, "loss": 4.1695, "step": 3301 }, { "epoch": 0.20497858340058353, "grad_norm": 0.8370507216632107, "learning_rate": 6.832195323815436e-05, "loss": 4.1303, "step": 3302 }, { "epoch": 0.20504066050034142, "grad_norm": 0.8021314821199836, "learning_rate": 6.834264432029795e-05, "loss": 4.2095, "step": 3303 }, { "epoch": 0.2051027376000993, "grad_norm": 0.8016673326258229, "learning_rate": 6.836333540244155e-05, "loss": 4.2644, "step": 3304 }, { "epoch": 0.20516481469985723, "grad_norm": 0.7684216317504331, "learning_rate": 6.838402648458516e-05, "loss": 4.3282, "step": 3305 }, { "epoch": 0.20522689179961512, "grad_norm": 0.8937061285708785, "learning_rate": 6.840471756672874e-05, "loss": 4.2341, "step": 3306 }, { "epoch": 0.205288968899373, "grad_norm": 0.9023842740557938, "learning_rate": 6.842540864887233e-05, "loss": 4.2685, "step": 3307 }, { "epoch": 0.20535104599913093, "grad_norm": 0.9062190214979057, "learning_rate": 6.844609973101593e-05, "loss": 4.3061, "step": 3308 }, { "epoch": 0.20541312309888882, "grad_norm": 0.8899291147442077, "learning_rate": 6.846679081315954e-05, "loss": 4.2677, "step": 3309 }, { "epoch": 0.2054752001986467, "grad_norm": 0.9898828187140682, "learning_rate": 6.848748189530313e-05, "loss": 4.1838, "step": 3310 }, { "epoch": 0.20553727729840462, "grad_norm": 1.2064323544404632, "learning_rate": 6.850817297744673e-05, "loss": 4.1627, "step": 3311 }, { "epoch": 0.20559935439816251, "grad_norm": 0.7970217027449465, "learning_rate": 6.852886405959032e-05, "loss": 4.2141, "step": 3312 }, { "epoch": 0.2056614314979204, "grad_norm": 1.076085802018154, "learning_rate": 6.854955514173392e-05, "loss": 4.2393, "step": 3313 }, { "epoch": 0.20572350859767832, "grad_norm": 1.158514383058, "learning_rate": 6.857024622387751e-05, "loss": 4.2614, "step": 3314 }, { "epoch": 0.2057855856974362, "grad_norm": 0.7587220544083598, "learning_rate": 6.859093730602111e-05, "loss": 4.1997, "step": 3315 }, { "epoch": 0.2058476627971941, "grad_norm": 1.0124895848809965, "learning_rate": 6.86116283881647e-05, "loss": 4.253, "step": 3316 }, { "epoch": 0.20590973989695202, "grad_norm": 1.115912930814468, "learning_rate": 6.863231947030831e-05, "loss": 4.1203, "step": 3317 }, { "epoch": 0.2059718169967099, "grad_norm": 0.9283614117629647, "learning_rate": 6.86530105524519e-05, "loss": 4.244, "step": 3318 }, { "epoch": 0.2060338940964678, "grad_norm": 0.8521474176820873, "learning_rate": 6.867370163459549e-05, "loss": 4.197, "step": 3319 }, { "epoch": 0.20609597119622572, "grad_norm": 0.9225450763699397, "learning_rate": 6.869439271673908e-05, "loss": 4.1832, "step": 3320 }, { "epoch": 0.2061580482959836, "grad_norm": 0.9203770639152135, "learning_rate": 6.871508379888269e-05, "loss": 4.2508, "step": 3321 }, { "epoch": 0.2062201253957415, "grad_norm": 0.8247704490006715, "learning_rate": 6.873577488102629e-05, "loss": 4.1301, "step": 3322 }, { "epoch": 0.20628220249549942, "grad_norm": 0.6890871627442255, "learning_rate": 6.875646596316988e-05, "loss": 4.2248, "step": 3323 }, { "epoch": 0.2063442795952573, "grad_norm": 0.9572554967866389, "learning_rate": 6.877715704531348e-05, "loss": 4.1136, "step": 3324 }, { "epoch": 0.2064063566950152, "grad_norm": 1.1017109114860795, "learning_rate": 6.879784812745707e-05, "loss": 4.1258, "step": 3325 }, { "epoch": 0.20646843379477312, "grad_norm": 0.893174954799867, "learning_rate": 6.881853920960067e-05, "loss": 4.2867, "step": 3326 }, { "epoch": 0.206530510894531, "grad_norm": 1.0720489227390566, "learning_rate": 6.883923029174426e-05, "loss": 4.2914, "step": 3327 }, { "epoch": 0.2065925879942889, "grad_norm": 0.9836491224535648, "learning_rate": 6.885992137388786e-05, "loss": 4.2034, "step": 3328 }, { "epoch": 0.2066546650940468, "grad_norm": 0.9903048879584253, "learning_rate": 6.888061245603145e-05, "loss": 4.2097, "step": 3329 }, { "epoch": 0.2067167421938047, "grad_norm": 0.7982707795766671, "learning_rate": 6.890130353817506e-05, "loss": 4.2633, "step": 3330 }, { "epoch": 0.2067788192935626, "grad_norm": 0.892708967038749, "learning_rate": 6.892199462031864e-05, "loss": 4.191, "step": 3331 }, { "epoch": 0.2068408963933205, "grad_norm": 0.8659578489062305, "learning_rate": 6.894268570246224e-05, "loss": 4.2411, "step": 3332 }, { "epoch": 0.2069029734930784, "grad_norm": 0.8683385341132208, "learning_rate": 6.896337678460583e-05, "loss": 4.1271, "step": 3333 }, { "epoch": 0.2069650505928363, "grad_norm": 0.8495997636624095, "learning_rate": 6.898406786674944e-05, "loss": 4.226, "step": 3334 }, { "epoch": 0.2070271276925942, "grad_norm": 0.8726150844498598, "learning_rate": 6.900475894889304e-05, "loss": 4.2564, "step": 3335 }, { "epoch": 0.2070892047923521, "grad_norm": 0.8889616776671587, "learning_rate": 6.902545003103663e-05, "loss": 4.2047, "step": 3336 }, { "epoch": 0.20715128189211, "grad_norm": 0.9096269173402068, "learning_rate": 6.904614111318021e-05, "loss": 4.2238, "step": 3337 }, { "epoch": 0.2072133589918679, "grad_norm": 0.9420187449483399, "learning_rate": 6.906683219532382e-05, "loss": 4.2111, "step": 3338 }, { "epoch": 0.2072754360916258, "grad_norm": 1.1798298225697275, "learning_rate": 6.908752327746742e-05, "loss": 4.3516, "step": 3339 }, { "epoch": 0.2073375131913837, "grad_norm": 0.9030092435788241, "learning_rate": 6.910821435961101e-05, "loss": 4.1595, "step": 3340 }, { "epoch": 0.2073995902911416, "grad_norm": 0.9429381952652992, "learning_rate": 6.91289054417546e-05, "loss": 4.2471, "step": 3341 }, { "epoch": 0.2074616673908995, "grad_norm": 0.9646251909691539, "learning_rate": 6.914959652389821e-05, "loss": 4.2537, "step": 3342 }, { "epoch": 0.20752374449065739, "grad_norm": 1.0544219071152927, "learning_rate": 6.91702876060418e-05, "loss": 4.1833, "step": 3343 }, { "epoch": 0.2075858215904153, "grad_norm": 0.8631565124703566, "learning_rate": 6.919097868818539e-05, "loss": 4.1918, "step": 3344 }, { "epoch": 0.2076478986901732, "grad_norm": 0.9537748091584489, "learning_rate": 6.921166977032899e-05, "loss": 4.2268, "step": 3345 }, { "epoch": 0.20770997578993108, "grad_norm": 0.9001876039479122, "learning_rate": 6.92323608524726e-05, "loss": 4.119, "step": 3346 }, { "epoch": 0.207772052889689, "grad_norm": 0.8255809717906649, "learning_rate": 6.925305193461619e-05, "loss": 4.1219, "step": 3347 }, { "epoch": 0.2078341299894469, "grad_norm": 0.6972671209581073, "learning_rate": 6.927374301675979e-05, "loss": 4.1419, "step": 3348 }, { "epoch": 0.20789620708920478, "grad_norm": 0.8026440052110633, "learning_rate": 6.929443409890337e-05, "loss": 4.1949, "step": 3349 }, { "epoch": 0.2079582841889627, "grad_norm": 0.7556879997652449, "learning_rate": 6.931512518104698e-05, "loss": 4.1754, "step": 3350 }, { "epoch": 0.2080203612887206, "grad_norm": 0.7165865472617329, "learning_rate": 6.933581626319057e-05, "loss": 4.2333, "step": 3351 }, { "epoch": 0.20808243838847848, "grad_norm": 0.77962641811196, "learning_rate": 6.935650734533417e-05, "loss": 4.2074, "step": 3352 }, { "epoch": 0.2081445154882364, "grad_norm": 0.785882945660542, "learning_rate": 6.937719842747776e-05, "loss": 4.2134, "step": 3353 }, { "epoch": 0.2082065925879943, "grad_norm": 0.8997448241563779, "learning_rate": 6.939788950962136e-05, "loss": 4.264, "step": 3354 }, { "epoch": 0.20826866968775218, "grad_norm": 0.9864036304542423, "learning_rate": 6.941858059176495e-05, "loss": 4.2269, "step": 3355 }, { "epoch": 0.2083307467875101, "grad_norm": 1.0413292847085351, "learning_rate": 6.943927167390855e-05, "loss": 4.1324, "step": 3356 }, { "epoch": 0.20839282388726799, "grad_norm": 0.8235885959233108, "learning_rate": 6.945996275605214e-05, "loss": 4.09, "step": 3357 }, { "epoch": 0.20845490098702588, "grad_norm": 0.7442236425994608, "learning_rate": 6.948065383819574e-05, "loss": 4.2792, "step": 3358 }, { "epoch": 0.2085169780867838, "grad_norm": 0.9554493500556155, "learning_rate": 6.950134492033934e-05, "loss": 4.0683, "step": 3359 }, { "epoch": 0.20857905518654168, "grad_norm": 0.9009819157952211, "learning_rate": 6.952203600248294e-05, "loss": 4.1363, "step": 3360 }, { "epoch": 0.20864113228629957, "grad_norm": 0.9069214339564025, "learning_rate": 6.954272708462652e-05, "loss": 4.2743, "step": 3361 }, { "epoch": 0.2087032093860575, "grad_norm": 0.8632063336899237, "learning_rate": 6.956341816677012e-05, "loss": 4.24, "step": 3362 }, { "epoch": 0.20876528648581538, "grad_norm": 0.817516794469667, "learning_rate": 6.958410924891372e-05, "loss": 4.315, "step": 3363 }, { "epoch": 0.20882736358557327, "grad_norm": 0.9123310416058918, "learning_rate": 6.960480033105732e-05, "loss": 4.1336, "step": 3364 }, { "epoch": 0.2088894406853312, "grad_norm": 1.0865002827334573, "learning_rate": 6.962549141320091e-05, "loss": 4.2128, "step": 3365 }, { "epoch": 0.20895151778508908, "grad_norm": 1.0412426212594303, "learning_rate": 6.964618249534451e-05, "loss": 4.2946, "step": 3366 }, { "epoch": 0.20901359488484697, "grad_norm": 0.8482510286274909, "learning_rate": 6.96668735774881e-05, "loss": 4.1812, "step": 3367 }, { "epoch": 0.2090756719846049, "grad_norm": 1.1318141352583604, "learning_rate": 6.96875646596317e-05, "loss": 4.1568, "step": 3368 }, { "epoch": 0.20913774908436278, "grad_norm": 0.8302328526518916, "learning_rate": 6.97082557417753e-05, "loss": 4.2327, "step": 3369 }, { "epoch": 0.20919982618412067, "grad_norm": 0.9278054942929436, "learning_rate": 6.972894682391889e-05, "loss": 4.1394, "step": 3370 }, { "epoch": 0.2092619032838786, "grad_norm": 1.0333906555171162, "learning_rate": 6.97496379060625e-05, "loss": 4.2237, "step": 3371 }, { "epoch": 0.20932398038363648, "grad_norm": 0.8441354789382852, "learning_rate": 6.97703289882061e-05, "loss": 4.1413, "step": 3372 }, { "epoch": 0.20938605748339437, "grad_norm": 0.9143441247322939, "learning_rate": 6.979102007034968e-05, "loss": 4.0402, "step": 3373 }, { "epoch": 0.20944813458315228, "grad_norm": 1.148514154691581, "learning_rate": 6.981171115249327e-05, "loss": 4.1124, "step": 3374 }, { "epoch": 0.20951021168291017, "grad_norm": 0.869712658489804, "learning_rate": 6.983240223463688e-05, "loss": 4.3002, "step": 3375 }, { "epoch": 0.20957228878266806, "grad_norm": 0.9272364836361034, "learning_rate": 6.985309331678047e-05, "loss": 4.1519, "step": 3376 }, { "epoch": 0.20963436588242598, "grad_norm": 0.9902861692588153, "learning_rate": 6.987378439892407e-05, "loss": 4.3342, "step": 3377 }, { "epoch": 0.20969644298218387, "grad_norm": 1.0533397399027469, "learning_rate": 6.989447548106766e-05, "loss": 4.1156, "step": 3378 }, { "epoch": 0.20975852008194176, "grad_norm": 0.8876885307900672, "learning_rate": 6.991516656321126e-05, "loss": 4.1792, "step": 3379 }, { "epoch": 0.20982059718169968, "grad_norm": 0.8921906279126601, "learning_rate": 6.993585764535485e-05, "loss": 4.1227, "step": 3380 }, { "epoch": 0.20988267428145757, "grad_norm": 0.7607218160582763, "learning_rate": 6.995654872749845e-05, "loss": 4.1561, "step": 3381 }, { "epoch": 0.20994475138121546, "grad_norm": 0.7427714008521045, "learning_rate": 6.997723980964204e-05, "loss": 4.1999, "step": 3382 }, { "epoch": 0.21000682848097338, "grad_norm": 0.7603825513598643, "learning_rate": 6.999793089178564e-05, "loss": 4.1631, "step": 3383 }, { "epoch": 0.21006890558073127, "grad_norm": 0.7464412259536937, "learning_rate": 7.001862197392925e-05, "loss": 4.1192, "step": 3384 }, { "epoch": 0.21013098268048916, "grad_norm": 0.7779081586329261, "learning_rate": 7.003931305607283e-05, "loss": 4.1092, "step": 3385 }, { "epoch": 0.21019305978024708, "grad_norm": 0.736150543088427, "learning_rate": 7.006000413821643e-05, "loss": 4.1323, "step": 3386 }, { "epoch": 0.21025513688000497, "grad_norm": 0.891136758573438, "learning_rate": 7.008069522036002e-05, "loss": 4.2397, "step": 3387 }, { "epoch": 0.21031721397976286, "grad_norm": 0.86417060024086, "learning_rate": 7.010138630250363e-05, "loss": 4.1607, "step": 3388 }, { "epoch": 0.21037929107952077, "grad_norm": 0.7637220595473178, "learning_rate": 7.012207738464722e-05, "loss": 4.1633, "step": 3389 }, { "epoch": 0.21044136817927867, "grad_norm": 0.6880266599868061, "learning_rate": 7.014276846679082e-05, "loss": 4.1976, "step": 3390 }, { "epoch": 0.21050344527903656, "grad_norm": 0.8110678963021264, "learning_rate": 7.01634595489344e-05, "loss": 4.1788, "step": 3391 }, { "epoch": 0.21056552237879447, "grad_norm": 0.77640783815304, "learning_rate": 7.018415063107801e-05, "loss": 4.1817, "step": 3392 }, { "epoch": 0.21062759947855236, "grad_norm": 0.7998330433794585, "learning_rate": 7.02048417132216e-05, "loss": 4.0643, "step": 3393 }, { "epoch": 0.21068967657831025, "grad_norm": 0.7586509156614483, "learning_rate": 7.02255327953652e-05, "loss": 4.0445, "step": 3394 }, { "epoch": 0.21075175367806817, "grad_norm": 0.9190588140170555, "learning_rate": 7.02462238775088e-05, "loss": 4.1517, "step": 3395 }, { "epoch": 0.21081383077782606, "grad_norm": 0.9744229797823862, "learning_rate": 7.02669149596524e-05, "loss": 4.2102, "step": 3396 }, { "epoch": 0.21087590787758395, "grad_norm": 0.9402884340613702, "learning_rate": 7.0287606041796e-05, "loss": 4.382, "step": 3397 }, { "epoch": 0.21093798497734187, "grad_norm": 0.901733588475194, "learning_rate": 7.030829712393958e-05, "loss": 4.2623, "step": 3398 }, { "epoch": 0.21100006207709976, "grad_norm": 0.9175237348017546, "learning_rate": 7.032898820608317e-05, "loss": 4.1891, "step": 3399 }, { "epoch": 0.21106213917685765, "grad_norm": 0.9492828912609974, "learning_rate": 7.034967928822678e-05, "loss": 4.1535, "step": 3400 }, { "epoch": 0.21112421627661557, "grad_norm": 0.9995168472404456, "learning_rate": 7.037037037037038e-05, "loss": 4.1015, "step": 3401 }, { "epoch": 0.21118629337637346, "grad_norm": 0.8662775093653703, "learning_rate": 7.039106145251397e-05, "loss": 4.0955, "step": 3402 }, { "epoch": 0.21124837047613135, "grad_norm": 0.8320532424240978, "learning_rate": 7.041175253465757e-05, "loss": 4.2096, "step": 3403 }, { "epoch": 0.21131044757588927, "grad_norm": 0.8759514200095958, "learning_rate": 7.043244361680116e-05, "loss": 4.227, "step": 3404 }, { "epoch": 0.21137252467564716, "grad_norm": 0.9222666668108461, "learning_rate": 7.045313469894476e-05, "loss": 4.2209, "step": 3405 }, { "epoch": 0.21143460177540505, "grad_norm": 0.783100834966104, "learning_rate": 7.047382578108835e-05, "loss": 4.1257, "step": 3406 }, { "epoch": 0.21149667887516296, "grad_norm": 0.7432361221968192, "learning_rate": 7.049451686323195e-05, "loss": 4.2505, "step": 3407 }, { "epoch": 0.21155875597492085, "grad_norm": 0.926126890405308, "learning_rate": 7.051520794537554e-05, "loss": 4.1102, "step": 3408 }, { "epoch": 0.21162083307467874, "grad_norm": 0.8897630636526854, "learning_rate": 7.053589902751915e-05, "loss": 4.1399, "step": 3409 }, { "epoch": 0.21168291017443666, "grad_norm": 0.768362949597684, "learning_rate": 7.055659010966273e-05, "loss": 4.0427, "step": 3410 }, { "epoch": 0.21174498727419455, "grad_norm": 0.8282813518403931, "learning_rate": 7.057728119180633e-05, "loss": 4.1958, "step": 3411 }, { "epoch": 0.21180706437395244, "grad_norm": 0.7444238799451102, "learning_rate": 7.059797227394992e-05, "loss": 4.2064, "step": 3412 }, { "epoch": 0.21186914147371036, "grad_norm": 0.7424182332417316, "learning_rate": 7.061866335609353e-05, "loss": 4.1459, "step": 3413 }, { "epoch": 0.21193121857346825, "grad_norm": 0.7400847512762678, "learning_rate": 7.063935443823713e-05, "loss": 4.0918, "step": 3414 }, { "epoch": 0.21199329567322614, "grad_norm": 0.8330154329154846, "learning_rate": 7.066004552038072e-05, "loss": 4.141, "step": 3415 }, { "epoch": 0.21205537277298406, "grad_norm": 0.7875984130401664, "learning_rate": 7.06807366025243e-05, "loss": 4.1031, "step": 3416 }, { "epoch": 0.21211744987274195, "grad_norm": 0.6098596406313348, "learning_rate": 7.070142768466791e-05, "loss": 4.1747, "step": 3417 }, { "epoch": 0.21217952697249984, "grad_norm": 0.6117154926393583, "learning_rate": 7.072211876681151e-05, "loss": 4.167, "step": 3418 }, { "epoch": 0.21224160407225776, "grad_norm": 0.6241303592591603, "learning_rate": 7.07428098489551e-05, "loss": 4.0714, "step": 3419 }, { "epoch": 0.21230368117201565, "grad_norm": 0.6491292687548799, "learning_rate": 7.07635009310987e-05, "loss": 4.2409, "step": 3420 }, { "epoch": 0.21236575827177354, "grad_norm": 0.6560690948367097, "learning_rate": 7.078419201324231e-05, "loss": 4.2947, "step": 3421 }, { "epoch": 0.21242783537153145, "grad_norm": 0.6135803513779774, "learning_rate": 7.080488309538589e-05, "loss": 4.0964, "step": 3422 }, { "epoch": 0.21248991247128934, "grad_norm": 0.7811145665080427, "learning_rate": 7.082557417752948e-05, "loss": 4.1879, "step": 3423 }, { "epoch": 0.21255198957104723, "grad_norm": 0.8042344975564879, "learning_rate": 7.084626525967308e-05, "loss": 4.1495, "step": 3424 }, { "epoch": 0.21261406667080515, "grad_norm": 0.7798727866506863, "learning_rate": 7.086695634181669e-05, "loss": 4.201, "step": 3425 }, { "epoch": 0.21267614377056304, "grad_norm": 0.8583922228626181, "learning_rate": 7.088764742396028e-05, "loss": 4.1955, "step": 3426 }, { "epoch": 0.21273822087032093, "grad_norm": 0.9710960897600094, "learning_rate": 7.090833850610388e-05, "loss": 4.1504, "step": 3427 }, { "epoch": 0.21280029797007885, "grad_norm": 1.1027803655338326, "learning_rate": 7.092902958824746e-05, "loss": 4.152, "step": 3428 }, { "epoch": 0.21286237506983674, "grad_norm": 0.9170166922353136, "learning_rate": 7.094972067039107e-05, "loss": 4.1106, "step": 3429 }, { "epoch": 0.21292445216959463, "grad_norm": 0.7769653702926944, "learning_rate": 7.097041175253466e-05, "loss": 4.1434, "step": 3430 }, { "epoch": 0.21298652926935255, "grad_norm": 0.9860682985626487, "learning_rate": 7.099110283467826e-05, "loss": 4.0981, "step": 3431 }, { "epoch": 0.21304860636911044, "grad_norm": 1.1168596546931266, "learning_rate": 7.101179391682185e-05, "loss": 4.2854, "step": 3432 }, { "epoch": 0.21311068346886833, "grad_norm": 0.8655425269985819, "learning_rate": 7.103248499896545e-05, "loss": 4.1546, "step": 3433 }, { "epoch": 0.21317276056862625, "grad_norm": 1.0095168495302462, "learning_rate": 7.105317608110904e-05, "loss": 4.1218, "step": 3434 }, { "epoch": 0.21323483766838414, "grad_norm": 0.9518661941451626, "learning_rate": 7.107386716325264e-05, "loss": 4.2042, "step": 3435 }, { "epoch": 0.21329691476814203, "grad_norm": 1.0692209957132108, "learning_rate": 7.109455824539623e-05, "loss": 4.1523, "step": 3436 }, { "epoch": 0.21335899186789994, "grad_norm": 0.8762261790957162, "learning_rate": 7.111524932753983e-05, "loss": 4.108, "step": 3437 }, { "epoch": 0.21342106896765783, "grad_norm": 0.7254677595418938, "learning_rate": 7.113594040968344e-05, "loss": 4.1122, "step": 3438 }, { "epoch": 0.21348314606741572, "grad_norm": 0.7149987391480753, "learning_rate": 7.115663149182703e-05, "loss": 4.1218, "step": 3439 }, { "epoch": 0.21354522316717364, "grad_norm": 0.7585011135940575, "learning_rate": 7.117732257397061e-05, "loss": 4.2063, "step": 3440 }, { "epoch": 0.21360730026693153, "grad_norm": 0.8905534882864223, "learning_rate": 7.119801365611422e-05, "loss": 4.2134, "step": 3441 }, { "epoch": 0.21366937736668942, "grad_norm": 0.7917041936981348, "learning_rate": 7.121870473825782e-05, "loss": 4.2428, "step": 3442 }, { "epoch": 0.21373145446644734, "grad_norm": 0.7580165784879869, "learning_rate": 7.123939582040141e-05, "loss": 4.1194, "step": 3443 }, { "epoch": 0.21379353156620523, "grad_norm": 0.7815629713689274, "learning_rate": 7.126008690254501e-05, "loss": 4.1764, "step": 3444 }, { "epoch": 0.21385560866596312, "grad_norm": 0.9112225618143044, "learning_rate": 7.12807779846886e-05, "loss": 4.1701, "step": 3445 }, { "epoch": 0.21391768576572104, "grad_norm": 0.8977599865074578, "learning_rate": 7.13014690668322e-05, "loss": 4.0781, "step": 3446 }, { "epoch": 0.21397976286547893, "grad_norm": 0.9782913773382163, "learning_rate": 7.132216014897579e-05, "loss": 4.2033, "step": 3447 }, { "epoch": 0.21404183996523682, "grad_norm": 0.9751945680492821, "learning_rate": 7.134285123111939e-05, "loss": 4.151, "step": 3448 }, { "epoch": 0.21410391706499474, "grad_norm": 0.8646955340708071, "learning_rate": 7.136354231326298e-05, "loss": 4.2119, "step": 3449 }, { "epoch": 0.21416599416475263, "grad_norm": 0.8162547431796305, "learning_rate": 7.138423339540659e-05, "loss": 4.1792, "step": 3450 }, { "epoch": 0.21422807126451052, "grad_norm": 0.9702480485184389, "learning_rate": 7.140492447755019e-05, "loss": 4.0792, "step": 3451 }, { "epoch": 0.21429014836426843, "grad_norm": 0.8622366682898348, "learning_rate": 7.142561555969377e-05, "loss": 4.0708, "step": 3452 }, { "epoch": 0.21435222546402632, "grad_norm": 0.887206493791114, "learning_rate": 7.144630664183736e-05, "loss": 4.1551, "step": 3453 }, { "epoch": 0.21441430256378421, "grad_norm": 0.8261726325281877, "learning_rate": 7.146699772398097e-05, "loss": 4.1157, "step": 3454 }, { "epoch": 0.21447637966354213, "grad_norm": 0.7174595152834126, "learning_rate": 7.148768880612457e-05, "loss": 4.2812, "step": 3455 }, { "epoch": 0.21453845676330002, "grad_norm": 0.8605523800598074, "learning_rate": 7.150837988826816e-05, "loss": 4.2272, "step": 3456 }, { "epoch": 0.2146005338630579, "grad_norm": 0.9706515753277217, "learning_rate": 7.152907097041176e-05, "loss": 4.1589, "step": 3457 }, { "epoch": 0.21466261096281583, "grad_norm": 0.9168150449365506, "learning_rate": 7.154976205255535e-05, "loss": 4.136, "step": 3458 }, { "epoch": 0.21472468806257372, "grad_norm": 0.8055753531891554, "learning_rate": 7.157045313469895e-05, "loss": 4.1296, "step": 3459 }, { "epoch": 0.2147867651623316, "grad_norm": 0.7431409233190355, "learning_rate": 7.159114421684254e-05, "loss": 4.0536, "step": 3460 }, { "epoch": 0.21484884226208953, "grad_norm": 0.9335059599520135, "learning_rate": 7.161183529898614e-05, "loss": 4.1752, "step": 3461 }, { "epoch": 0.21491091936184742, "grad_norm": 1.0057543653468806, "learning_rate": 7.163252638112973e-05, "loss": 4.188, "step": 3462 }, { "epoch": 0.2149729964616053, "grad_norm": 0.9423101683768255, "learning_rate": 7.165321746327334e-05, "loss": 4.1581, "step": 3463 }, { "epoch": 0.21503507356136323, "grad_norm": 0.8970054823964114, "learning_rate": 7.167390854541692e-05, "loss": 4.1658, "step": 3464 }, { "epoch": 0.21509715066112112, "grad_norm": 0.9027163904462756, "learning_rate": 7.169459962756052e-05, "loss": 4.1205, "step": 3465 }, { "epoch": 0.215159227760879, "grad_norm": 0.8947563054858373, "learning_rate": 7.171529070970413e-05, "loss": 4.2162, "step": 3466 }, { "epoch": 0.2152213048606369, "grad_norm": 0.8827915695544952, "learning_rate": 7.173598179184772e-05, "loss": 4.2258, "step": 3467 }, { "epoch": 0.21528338196039482, "grad_norm": 0.9045752144344789, "learning_rate": 7.175667287399132e-05, "loss": 4.0839, "step": 3468 }, { "epoch": 0.2153454590601527, "grad_norm": 0.8795242849275126, "learning_rate": 7.177736395613491e-05, "loss": 4.1344, "step": 3469 }, { "epoch": 0.2154075361599106, "grad_norm": 0.807851330766469, "learning_rate": 7.17980550382785e-05, "loss": 4.1849, "step": 3470 }, { "epoch": 0.2154696132596685, "grad_norm": 0.7311248688537094, "learning_rate": 7.18187461204221e-05, "loss": 4.1012, "step": 3471 }, { "epoch": 0.2155316903594264, "grad_norm": 0.6647372912511593, "learning_rate": 7.18394372025657e-05, "loss": 4.2164, "step": 3472 }, { "epoch": 0.2155937674591843, "grad_norm": 0.8782730194992873, "learning_rate": 7.186012828470929e-05, "loss": 4.1424, "step": 3473 }, { "epoch": 0.2156558445589422, "grad_norm": 1.0013369325576518, "learning_rate": 7.188081936685289e-05, "loss": 4.2131, "step": 3474 }, { "epoch": 0.2157179216587001, "grad_norm": 0.9973170230230047, "learning_rate": 7.19015104489965e-05, "loss": 4.1748, "step": 3475 }, { "epoch": 0.215779998758458, "grad_norm": 1.0007468653183575, "learning_rate": 7.192220153114008e-05, "loss": 4.0705, "step": 3476 }, { "epoch": 0.2158420758582159, "grad_norm": 0.9893073383071086, "learning_rate": 7.194289261328367e-05, "loss": 4.1119, "step": 3477 }, { "epoch": 0.2159041529579738, "grad_norm": 1.0917105597973238, "learning_rate": 7.196358369542727e-05, "loss": 4.2285, "step": 3478 }, { "epoch": 0.2159662300577317, "grad_norm": 0.8696765404688165, "learning_rate": 7.198427477757088e-05, "loss": 4.0136, "step": 3479 }, { "epoch": 0.2160283071574896, "grad_norm": 0.7735444197300513, "learning_rate": 7.200496585971447e-05, "loss": 4.0581, "step": 3480 }, { "epoch": 0.2160903842572475, "grad_norm": 0.948318525516055, "learning_rate": 7.202565694185807e-05, "loss": 4.1876, "step": 3481 }, { "epoch": 0.2161524613570054, "grad_norm": 0.8786425724809457, "learning_rate": 7.204634802400166e-05, "loss": 4.3157, "step": 3482 }, { "epoch": 0.2162145384567633, "grad_norm": 0.8854593752864062, "learning_rate": 7.206703910614526e-05, "loss": 4.0572, "step": 3483 }, { "epoch": 0.2162766155565212, "grad_norm": 0.8429656798525135, "learning_rate": 7.208773018828885e-05, "loss": 4.1638, "step": 3484 }, { "epoch": 0.21633869265627909, "grad_norm": 0.8683205839991816, "learning_rate": 7.210842127043245e-05, "loss": 4.1771, "step": 3485 }, { "epoch": 0.216400769756037, "grad_norm": 0.9142799444846758, "learning_rate": 7.212911235257604e-05, "loss": 4.1641, "step": 3486 }, { "epoch": 0.2164628468557949, "grad_norm": 0.9203089327006347, "learning_rate": 7.214980343471964e-05, "loss": 4.1755, "step": 3487 }, { "epoch": 0.21652492395555278, "grad_norm": 0.9977393824414561, "learning_rate": 7.217049451686324e-05, "loss": 4.0537, "step": 3488 }, { "epoch": 0.2165870010553107, "grad_norm": 0.9365215918052394, "learning_rate": 7.219118559900683e-05, "loss": 4.2802, "step": 3489 }, { "epoch": 0.2166490781550686, "grad_norm": 0.9407854684717077, "learning_rate": 7.221187668115042e-05, "loss": 4.1726, "step": 3490 }, { "epoch": 0.21671115525482648, "grad_norm": 1.0358026177928052, "learning_rate": 7.223256776329403e-05, "loss": 4.1626, "step": 3491 }, { "epoch": 0.2167732323545844, "grad_norm": 0.8859475769331044, "learning_rate": 7.225325884543763e-05, "loss": 4.0509, "step": 3492 }, { "epoch": 0.2168353094543423, "grad_norm": 0.8019533836486964, "learning_rate": 7.227394992758122e-05, "loss": 4.0714, "step": 3493 }, { "epoch": 0.21689738655410018, "grad_norm": 0.8813623316169127, "learning_rate": 7.229464100972482e-05, "loss": 4.1269, "step": 3494 }, { "epoch": 0.2169594636538581, "grad_norm": 0.8882159058920212, "learning_rate": 7.231533209186841e-05, "loss": 4.0382, "step": 3495 }, { "epoch": 0.217021540753616, "grad_norm": 0.7779149396237323, "learning_rate": 7.2336023174012e-05, "loss": 4.0988, "step": 3496 }, { "epoch": 0.21708361785337388, "grad_norm": 0.6537886966244584, "learning_rate": 7.23567142561556e-05, "loss": 4.1782, "step": 3497 }, { "epoch": 0.2171456949531318, "grad_norm": 0.6718196180716949, "learning_rate": 7.23774053382992e-05, "loss": 4.1401, "step": 3498 }, { "epoch": 0.21720777205288969, "grad_norm": 0.7346188566461523, "learning_rate": 7.239809642044279e-05, "loss": 4.1027, "step": 3499 }, { "epoch": 0.21726984915264758, "grad_norm": 0.7325313232026064, "learning_rate": 7.24187875025864e-05, "loss": 4.0923, "step": 3500 }, { "epoch": 0.2173319262524055, "grad_norm": 0.819384094932704, "learning_rate": 7.243947858472998e-05, "loss": 4.0866, "step": 3501 }, { "epoch": 0.21739400335216338, "grad_norm": 0.8339925582448527, "learning_rate": 7.246016966687358e-05, "loss": 4.1908, "step": 3502 }, { "epoch": 0.21745608045192127, "grad_norm": 0.8659835912744525, "learning_rate": 7.248086074901717e-05, "loss": 4.192, "step": 3503 }, { "epoch": 0.2175181575516792, "grad_norm": 0.7107458589924964, "learning_rate": 7.250155183116078e-05, "loss": 4.0382, "step": 3504 }, { "epoch": 0.21758023465143708, "grad_norm": 0.7992079375161156, "learning_rate": 7.252224291330437e-05, "loss": 4.1654, "step": 3505 }, { "epoch": 0.21764231175119497, "grad_norm": 0.6974912860626842, "learning_rate": 7.254293399544797e-05, "loss": 4.039, "step": 3506 }, { "epoch": 0.2177043888509529, "grad_norm": 0.7338341283578037, "learning_rate": 7.256362507759155e-05, "loss": 4.165, "step": 3507 }, { "epoch": 0.21776646595071078, "grad_norm": 0.8140022255517964, "learning_rate": 7.258431615973516e-05, "loss": 4.1337, "step": 3508 }, { "epoch": 0.21782854305046867, "grad_norm": 0.7283439536233297, "learning_rate": 7.260500724187875e-05, "loss": 4.1044, "step": 3509 }, { "epoch": 0.2178906201502266, "grad_norm": 0.7560161260469367, "learning_rate": 7.262569832402235e-05, "loss": 4.0844, "step": 3510 }, { "epoch": 0.21795269724998448, "grad_norm": 0.6872619953864069, "learning_rate": 7.264638940616595e-05, "loss": 4.0823, "step": 3511 }, { "epoch": 0.21801477434974237, "grad_norm": 0.7549464506144049, "learning_rate": 7.266708048830954e-05, "loss": 4.1471, "step": 3512 }, { "epoch": 0.2180768514495003, "grad_norm": 0.7372368989817211, "learning_rate": 7.268777157045314e-05, "loss": 4.1554, "step": 3513 }, { "epoch": 0.21813892854925818, "grad_norm": 0.683349434926514, "learning_rate": 7.270846265259673e-05, "loss": 4.1733, "step": 3514 }, { "epoch": 0.21820100564901607, "grad_norm": 0.765434747529783, "learning_rate": 7.272915373474033e-05, "loss": 4.1978, "step": 3515 }, { "epoch": 0.21826308274877398, "grad_norm": 0.9442282204008655, "learning_rate": 7.274984481688393e-05, "loss": 4.1791, "step": 3516 }, { "epoch": 0.21832515984853187, "grad_norm": 1.235486411473427, "learning_rate": 7.277053589902753e-05, "loss": 4.1914, "step": 3517 }, { "epoch": 0.21838723694828976, "grad_norm": 0.6786499654703139, "learning_rate": 7.279122698117112e-05, "loss": 4.0398, "step": 3518 }, { "epoch": 0.21844931404804768, "grad_norm": 0.7465966811404994, "learning_rate": 7.28119180633147e-05, "loss": 4.044, "step": 3519 }, { "epoch": 0.21851139114780557, "grad_norm": 1.0949873873014722, "learning_rate": 7.283260914545831e-05, "loss": 4.0974, "step": 3520 }, { "epoch": 0.21857346824756346, "grad_norm": 1.0088269945751454, "learning_rate": 7.285330022760191e-05, "loss": 4.1494, "step": 3521 }, { "epoch": 0.21863554534732138, "grad_norm": 0.9527569982791667, "learning_rate": 7.28739913097455e-05, "loss": 4.0658, "step": 3522 }, { "epoch": 0.21869762244707927, "grad_norm": 0.9893456558435227, "learning_rate": 7.28946823918891e-05, "loss": 4.0395, "step": 3523 }, { "epoch": 0.21875969954683716, "grad_norm": 1.122023353224827, "learning_rate": 7.29153734740327e-05, "loss": 4.1352, "step": 3524 }, { "epoch": 0.21882177664659508, "grad_norm": 0.8689982779873203, "learning_rate": 7.293606455617629e-05, "loss": 4.2004, "step": 3525 }, { "epoch": 0.21888385374635297, "grad_norm": 0.9498540063408778, "learning_rate": 7.295675563831988e-05, "loss": 4.1774, "step": 3526 }, { "epoch": 0.21894593084611086, "grad_norm": 1.0800654874479643, "learning_rate": 7.297744672046348e-05, "loss": 4.1521, "step": 3527 }, { "epoch": 0.21900800794586878, "grad_norm": 0.8216824672281094, "learning_rate": 7.299813780260707e-05, "loss": 4.1503, "step": 3528 }, { "epoch": 0.21907008504562667, "grad_norm": 0.9765152881654144, "learning_rate": 7.301882888475068e-05, "loss": 4.1851, "step": 3529 }, { "epoch": 0.21913216214538456, "grad_norm": 0.8730827511919615, "learning_rate": 7.303951996689428e-05, "loss": 4.1251, "step": 3530 }, { "epoch": 0.21919423924514247, "grad_norm": 0.9240496895979013, "learning_rate": 7.306021104903786e-05, "loss": 4.1481, "step": 3531 }, { "epoch": 0.21925631634490036, "grad_norm": 0.96418690427501, "learning_rate": 7.308090213118146e-05, "loss": 4.0501, "step": 3532 }, { "epoch": 0.21931839344465826, "grad_norm": 0.8680510474193919, "learning_rate": 7.310159321332506e-05, "loss": 4.1762, "step": 3533 }, { "epoch": 0.21938047054441617, "grad_norm": 0.8671318242526415, "learning_rate": 7.312228429546866e-05, "loss": 4.0831, "step": 3534 }, { "epoch": 0.21944254764417406, "grad_norm": 1.001039739336321, "learning_rate": 7.314297537761225e-05, "loss": 4.1566, "step": 3535 }, { "epoch": 0.21950462474393195, "grad_norm": 0.9758467225465515, "learning_rate": 7.316366645975585e-05, "loss": 4.1352, "step": 3536 }, { "epoch": 0.21956670184368987, "grad_norm": 0.8295854969146529, "learning_rate": 7.318435754189944e-05, "loss": 4.1065, "step": 3537 }, { "epoch": 0.21962877894344776, "grad_norm": 0.7465395169973859, "learning_rate": 7.320504862404304e-05, "loss": 4.198, "step": 3538 }, { "epoch": 0.21969085604320565, "grad_norm": 0.780583860597595, "learning_rate": 7.322573970618663e-05, "loss": 4.0685, "step": 3539 }, { "epoch": 0.21975293314296357, "grad_norm": 0.672086527151271, "learning_rate": 7.324643078833023e-05, "loss": 4.058, "step": 3540 }, { "epoch": 0.21981501024272146, "grad_norm": 0.8666332294970788, "learning_rate": 7.326712187047384e-05, "loss": 4.0621, "step": 3541 }, { "epoch": 0.21987708734247935, "grad_norm": 0.8039065154346736, "learning_rate": 7.328781295261743e-05, "loss": 4.0921, "step": 3542 }, { "epoch": 0.21993916444223727, "grad_norm": 0.812465923768501, "learning_rate": 7.330850403476101e-05, "loss": 4.12, "step": 3543 }, { "epoch": 0.22000124154199516, "grad_norm": 0.9308453250261707, "learning_rate": 7.332919511690461e-05, "loss": 4.1678, "step": 3544 }, { "epoch": 0.22006331864175305, "grad_norm": 0.9004849862324997, "learning_rate": 7.334988619904822e-05, "loss": 4.1153, "step": 3545 }, { "epoch": 0.22012539574151097, "grad_norm": 0.8936489368199113, "learning_rate": 7.337057728119181e-05, "loss": 3.9803, "step": 3546 }, { "epoch": 0.22018747284126886, "grad_norm": 0.995207059056999, "learning_rate": 7.339126836333541e-05, "loss": 4.2069, "step": 3547 }, { "epoch": 0.22024954994102675, "grad_norm": 0.8826519305285058, "learning_rate": 7.3411959445479e-05, "loss": 4.1689, "step": 3548 }, { "epoch": 0.22031162704078466, "grad_norm": 0.7261079986861961, "learning_rate": 7.34326505276226e-05, "loss": 4.1624, "step": 3549 }, { "epoch": 0.22037370414054255, "grad_norm": 0.7201360937481396, "learning_rate": 7.34533416097662e-05, "loss": 4.0636, "step": 3550 }, { "epoch": 0.22043578124030044, "grad_norm": 0.7878305241097768, "learning_rate": 7.347403269190979e-05, "loss": 4.1046, "step": 3551 }, { "epoch": 0.22049785834005836, "grad_norm": 0.7596231832076094, "learning_rate": 7.349472377405338e-05, "loss": 4.0857, "step": 3552 }, { "epoch": 0.22055993543981625, "grad_norm": 0.7136651417248092, "learning_rate": 7.351541485619698e-05, "loss": 4.0649, "step": 3553 }, { "epoch": 0.22062201253957414, "grad_norm": 0.6497579710861501, "learning_rate": 7.353610593834059e-05, "loss": 4.0571, "step": 3554 }, { "epoch": 0.22068408963933206, "grad_norm": 0.7805065432267531, "learning_rate": 7.355679702048417e-05, "loss": 4.1146, "step": 3555 }, { "epoch": 0.22074616673908995, "grad_norm": 0.8302932281037885, "learning_rate": 7.357748810262776e-05, "loss": 4.0008, "step": 3556 }, { "epoch": 0.22080824383884784, "grad_norm": 0.886601511729247, "learning_rate": 7.359817918477136e-05, "loss": 4.1056, "step": 3557 }, { "epoch": 0.22087032093860576, "grad_norm": 0.9639905794542075, "learning_rate": 7.361887026691497e-05, "loss": 4.0763, "step": 3558 }, { "epoch": 0.22093239803836365, "grad_norm": 0.9584225564692825, "learning_rate": 7.363956134905856e-05, "loss": 4.1077, "step": 3559 }, { "epoch": 0.22099447513812154, "grad_norm": 1.0588939283379366, "learning_rate": 7.366025243120216e-05, "loss": 4.0624, "step": 3560 }, { "epoch": 0.22105655223787946, "grad_norm": 0.7480554405897282, "learning_rate": 7.368094351334574e-05, "loss": 4.0939, "step": 3561 }, { "epoch": 0.22111862933763735, "grad_norm": 0.790888081381668, "learning_rate": 7.370163459548935e-05, "loss": 4.1852, "step": 3562 }, { "epoch": 0.22118070643739524, "grad_norm": 1.0717221743346412, "learning_rate": 7.372232567763294e-05, "loss": 4.1439, "step": 3563 }, { "epoch": 0.22124278353715315, "grad_norm": 1.0784850141415245, "learning_rate": 7.374301675977654e-05, "loss": 4.0281, "step": 3564 }, { "epoch": 0.22130486063691104, "grad_norm": 1.057645084087161, "learning_rate": 7.376370784192013e-05, "loss": 4.0494, "step": 3565 }, { "epoch": 0.22136693773666893, "grad_norm": 0.9066098255216639, "learning_rate": 7.378439892406374e-05, "loss": 4.107, "step": 3566 }, { "epoch": 0.22142901483642685, "grad_norm": 0.9626370728719809, "learning_rate": 7.380509000620734e-05, "loss": 4.1291, "step": 3567 }, { "epoch": 0.22149109193618474, "grad_norm": 0.9396606660188724, "learning_rate": 7.382578108835092e-05, "loss": 4.1138, "step": 3568 }, { "epoch": 0.22155316903594263, "grad_norm": 0.8837833973402683, "learning_rate": 7.384647217049451e-05, "loss": 4.1548, "step": 3569 }, { "epoch": 0.22161524613570055, "grad_norm": 0.7948967182199655, "learning_rate": 7.386716325263812e-05, "loss": 4.1078, "step": 3570 }, { "epoch": 0.22167732323545844, "grad_norm": 0.7280150629308765, "learning_rate": 7.388785433478172e-05, "loss": 4.1275, "step": 3571 }, { "epoch": 0.22173940033521633, "grad_norm": 0.7897367335114238, "learning_rate": 7.390854541692531e-05, "loss": 4.0686, "step": 3572 }, { "epoch": 0.22180147743497425, "grad_norm": 0.7533954214009095, "learning_rate": 7.392923649906891e-05, "loss": 4.066, "step": 3573 }, { "epoch": 0.22186355453473214, "grad_norm": 0.6740695907722077, "learning_rate": 7.39499275812125e-05, "loss": 4.0865, "step": 3574 }, { "epoch": 0.22192563163449003, "grad_norm": 0.7840272368083927, "learning_rate": 7.39706186633561e-05, "loss": 4.144, "step": 3575 }, { "epoch": 0.22198770873424795, "grad_norm": 0.7799904779317598, "learning_rate": 7.399130974549969e-05, "loss": 4.1625, "step": 3576 }, { "epoch": 0.22204978583400584, "grad_norm": 0.7050289020633655, "learning_rate": 7.401200082764329e-05, "loss": 4.1229, "step": 3577 }, { "epoch": 0.22211186293376373, "grad_norm": 0.7443697919431176, "learning_rate": 7.403269190978688e-05, "loss": 4.1372, "step": 3578 }, { "epoch": 0.22217394003352164, "grad_norm": 0.6668383531234506, "learning_rate": 7.405338299193049e-05, "loss": 4.189, "step": 3579 }, { "epoch": 0.22223601713327953, "grad_norm": 0.6567839993017608, "learning_rate": 7.407407407407407e-05, "loss": 4.06, "step": 3580 }, { "epoch": 0.22229809423303742, "grad_norm": 0.6493435470042468, "learning_rate": 7.409476515621767e-05, "loss": 4.0882, "step": 3581 }, { "epoch": 0.22236017133279534, "grad_norm": 0.6279225920043439, "learning_rate": 7.411545623836126e-05, "loss": 3.8841, "step": 3582 }, { "epoch": 0.22242224843255323, "grad_norm": 0.7687048177699897, "learning_rate": 7.413614732050487e-05, "loss": 4.0934, "step": 3583 }, { "epoch": 0.22248432553231112, "grad_norm": 0.7652478359532033, "learning_rate": 7.415683840264847e-05, "loss": 4.158, "step": 3584 }, { "epoch": 0.22254640263206904, "grad_norm": 0.7420411216543485, "learning_rate": 7.417752948479206e-05, "loss": 4.1516, "step": 3585 }, { "epoch": 0.22260847973182693, "grad_norm": 0.7449168564657904, "learning_rate": 7.419822056693564e-05, "loss": 4.062, "step": 3586 }, { "epoch": 0.22267055683158482, "grad_norm": 0.7080583646419343, "learning_rate": 7.421891164907925e-05, "loss": 4.0525, "step": 3587 }, { "epoch": 0.22273263393134274, "grad_norm": 0.7785206612262819, "learning_rate": 7.423960273122285e-05, "loss": 4.1403, "step": 3588 }, { "epoch": 0.22279471103110063, "grad_norm": 0.918449178740189, "learning_rate": 7.426029381336644e-05, "loss": 4.1861, "step": 3589 }, { "epoch": 0.22285678813085852, "grad_norm": 0.8747830462562849, "learning_rate": 7.428098489551004e-05, "loss": 4.0541, "step": 3590 }, { "epoch": 0.22291886523061644, "grad_norm": 0.7894179799270887, "learning_rate": 7.430167597765365e-05, "loss": 4.1358, "step": 3591 }, { "epoch": 0.22298094233037433, "grad_norm": 0.8018873809310583, "learning_rate": 7.432236705979723e-05, "loss": 4.1558, "step": 3592 }, { "epoch": 0.22304301943013222, "grad_norm": 0.8614223714115996, "learning_rate": 7.434305814194082e-05, "loss": 4.1268, "step": 3593 }, { "epoch": 0.22310509652989013, "grad_norm": 0.7151186888559136, "learning_rate": 7.436374922408442e-05, "loss": 3.9949, "step": 3594 }, { "epoch": 0.22316717362964802, "grad_norm": 0.6932992828548106, "learning_rate": 7.438444030622803e-05, "loss": 4.0744, "step": 3595 }, { "epoch": 0.22322925072940591, "grad_norm": 0.7065748564741839, "learning_rate": 7.440513138837162e-05, "loss": 4.0354, "step": 3596 }, { "epoch": 0.22329132782916383, "grad_norm": 0.7896795975742971, "learning_rate": 7.442582247051522e-05, "loss": 4.1841, "step": 3597 }, { "epoch": 0.22335340492892172, "grad_norm": 0.9478675109917564, "learning_rate": 7.44465135526588e-05, "loss": 4.1061, "step": 3598 }, { "epoch": 0.2234154820286796, "grad_norm": 1.0180349075238098, "learning_rate": 7.44672046348024e-05, "loss": 4.0799, "step": 3599 }, { "epoch": 0.22347755912843753, "grad_norm": 0.9592775724065695, "learning_rate": 7.4487895716946e-05, "loss": 4.1108, "step": 3600 }, { "epoch": 0.22353963622819542, "grad_norm": 0.9985571525408105, "learning_rate": 7.45085867990896e-05, "loss": 4.0063, "step": 3601 }, { "epoch": 0.2236017133279533, "grad_norm": 0.8769793602092538, "learning_rate": 7.452927788123319e-05, "loss": 4.1796, "step": 3602 }, { "epoch": 0.22366379042771123, "grad_norm": 0.8650219314198402, "learning_rate": 7.454996896337679e-05, "loss": 3.9817, "step": 3603 }, { "epoch": 0.22372586752746912, "grad_norm": 0.9892500701028926, "learning_rate": 7.457066004552038e-05, "loss": 4.1231, "step": 3604 }, { "epoch": 0.223787944627227, "grad_norm": 1.0333943469482627, "learning_rate": 7.459135112766398e-05, "loss": 4.1976, "step": 3605 }, { "epoch": 0.22385002172698493, "grad_norm": 0.8384356026871491, "learning_rate": 7.461204220980757e-05, "loss": 3.9696, "step": 3606 }, { "epoch": 0.22391209882674282, "grad_norm": 0.8532949201560737, "learning_rate": 7.463273329195117e-05, "loss": 4.1355, "step": 3607 }, { "epoch": 0.2239741759265007, "grad_norm": 0.8238543387339813, "learning_rate": 7.465342437409478e-05, "loss": 4.1309, "step": 3608 }, { "epoch": 0.22403625302625862, "grad_norm": 0.755003807171955, "learning_rate": 7.467411545623837e-05, "loss": 4.0269, "step": 3609 }, { "epoch": 0.22409833012601652, "grad_norm": 0.7643848807818382, "learning_rate": 7.469480653838195e-05, "loss": 4.1101, "step": 3610 }, { "epoch": 0.2241604072257744, "grad_norm": 0.7681009002331629, "learning_rate": 7.471549762052555e-05, "loss": 4.045, "step": 3611 }, { "epoch": 0.22422248432553232, "grad_norm": 0.7443485320003422, "learning_rate": 7.473618870266916e-05, "loss": 4.0259, "step": 3612 }, { "epoch": 0.2242845614252902, "grad_norm": 0.7780591270415954, "learning_rate": 7.475687978481275e-05, "loss": 4.0709, "step": 3613 }, { "epoch": 0.2243466385250481, "grad_norm": 0.8379935532597316, "learning_rate": 7.477757086695635e-05, "loss": 4.0982, "step": 3614 }, { "epoch": 0.22440871562480602, "grad_norm": 0.8698714366280178, "learning_rate": 7.479826194909994e-05, "loss": 4.1166, "step": 3615 }, { "epoch": 0.2244707927245639, "grad_norm": 0.8773420529166543, "learning_rate": 7.481895303124354e-05, "loss": 4.1307, "step": 3616 }, { "epoch": 0.2245328698243218, "grad_norm": 0.995164145727185, "learning_rate": 7.483964411338713e-05, "loss": 4.0323, "step": 3617 }, { "epoch": 0.22459494692407972, "grad_norm": 0.8648495130338311, "learning_rate": 7.486033519553073e-05, "loss": 4.1663, "step": 3618 }, { "epoch": 0.2246570240238376, "grad_norm": 0.7897219781930769, "learning_rate": 7.488102627767432e-05, "loss": 4.144, "step": 3619 }, { "epoch": 0.2247191011235955, "grad_norm": 0.8811370682557577, "learning_rate": 7.490171735981793e-05, "loss": 4.0147, "step": 3620 }, { "epoch": 0.22478117822335342, "grad_norm": 0.8370761425063263, "learning_rate": 7.492240844196153e-05, "loss": 4.0063, "step": 3621 }, { "epoch": 0.2248432553231113, "grad_norm": 0.8720986766410159, "learning_rate": 7.494309952410511e-05, "loss": 4.0525, "step": 3622 }, { "epoch": 0.2249053324228692, "grad_norm": 0.9752370155015748, "learning_rate": 7.49637906062487e-05, "loss": 4.1523, "step": 3623 }, { "epoch": 0.22496740952262712, "grad_norm": 1.2176052488231437, "learning_rate": 7.498448168839231e-05, "loss": 4.0729, "step": 3624 }, { "epoch": 0.225029486622385, "grad_norm": 0.8121988031672998, "learning_rate": 7.50051727705359e-05, "loss": 4.0557, "step": 3625 }, { "epoch": 0.2250915637221429, "grad_norm": 0.8833512206935583, "learning_rate": 7.50258638526795e-05, "loss": 4.1193, "step": 3626 }, { "epoch": 0.2251536408219008, "grad_norm": 1.0232558343992675, "learning_rate": 7.50465549348231e-05, "loss": 4.0008, "step": 3627 }, { "epoch": 0.2252157179216587, "grad_norm": 0.9189563389764207, "learning_rate": 7.506724601696669e-05, "loss": 4.0355, "step": 3628 }, { "epoch": 0.2252777950214166, "grad_norm": 0.7477968759335284, "learning_rate": 7.508793709911029e-05, "loss": 4.2136, "step": 3629 }, { "epoch": 0.2253398721211745, "grad_norm": 0.8075474405672924, "learning_rate": 7.510862818125388e-05, "loss": 4.0117, "step": 3630 }, { "epoch": 0.2254019492209324, "grad_norm": 0.8015283216659201, "learning_rate": 7.512931926339748e-05, "loss": 4.0865, "step": 3631 }, { "epoch": 0.2254640263206903, "grad_norm": 0.7642772645763348, "learning_rate": 7.515001034554107e-05, "loss": 4.0743, "step": 3632 }, { "epoch": 0.2255261034204482, "grad_norm": 0.8464313217598975, "learning_rate": 7.517070142768468e-05, "loss": 4.1186, "step": 3633 }, { "epoch": 0.2255881805202061, "grad_norm": 0.7108182702326304, "learning_rate": 7.519139250982826e-05, "loss": 4.1722, "step": 3634 }, { "epoch": 0.225650257619964, "grad_norm": 0.7757215586398875, "learning_rate": 7.521208359197186e-05, "loss": 4.0986, "step": 3635 }, { "epoch": 0.2257123347197219, "grad_norm": 0.889980082664077, "learning_rate": 7.523277467411545e-05, "loss": 4.0345, "step": 3636 }, { "epoch": 0.2257744118194798, "grad_norm": 0.7917602801605604, "learning_rate": 7.525346575625906e-05, "loss": 4.1219, "step": 3637 }, { "epoch": 0.2258364889192377, "grad_norm": 0.7677892341240559, "learning_rate": 7.527415683840266e-05, "loss": 4.0812, "step": 3638 }, { "epoch": 0.2258985660189956, "grad_norm": 0.7494041141966175, "learning_rate": 7.529484792054625e-05, "loss": 4.1185, "step": 3639 }, { "epoch": 0.2259606431187535, "grad_norm": 0.89634460822415, "learning_rate": 7.531553900268985e-05, "loss": 4.076, "step": 3640 }, { "epoch": 0.22602272021851139, "grad_norm": 0.9048938799123024, "learning_rate": 7.533623008483344e-05, "loss": 4.1072, "step": 3641 }, { "epoch": 0.2260847973182693, "grad_norm": 0.9582848571182003, "learning_rate": 7.535692116697704e-05, "loss": 4.0112, "step": 3642 }, { "epoch": 0.2261468744180272, "grad_norm": 0.9449061000937525, "learning_rate": 7.537761224912063e-05, "loss": 4.1635, "step": 3643 }, { "epoch": 0.22620895151778508, "grad_norm": 0.8699296091915254, "learning_rate": 7.539830333126423e-05, "loss": 4.0205, "step": 3644 }, { "epoch": 0.226271028617543, "grad_norm": 0.7672387457550769, "learning_rate": 7.541899441340783e-05, "loss": 4.1061, "step": 3645 }, { "epoch": 0.2263331057173009, "grad_norm": 0.7860901246416616, "learning_rate": 7.543968549555143e-05, "loss": 4.1591, "step": 3646 }, { "epoch": 0.22639518281705878, "grad_norm": 0.8391230840139735, "learning_rate": 7.546037657769501e-05, "loss": 4.1018, "step": 3647 }, { "epoch": 0.2264572599168167, "grad_norm": 0.7601598584676565, "learning_rate": 7.54810676598386e-05, "loss": 4.0048, "step": 3648 }, { "epoch": 0.2265193370165746, "grad_norm": 0.6941044846863456, "learning_rate": 7.550175874198221e-05, "loss": 4.1077, "step": 3649 }, { "epoch": 0.22658141411633248, "grad_norm": 0.6553021579593967, "learning_rate": 7.552244982412581e-05, "loss": 3.9986, "step": 3650 }, { "epoch": 0.2266434912160904, "grad_norm": 0.6990747996217257, "learning_rate": 7.55431409062694e-05, "loss": 4.0251, "step": 3651 }, { "epoch": 0.2267055683158483, "grad_norm": 0.7681578953328349, "learning_rate": 7.5563831988413e-05, "loss": 4.0846, "step": 3652 }, { "epoch": 0.22676764541560618, "grad_norm": 0.7217894846148394, "learning_rate": 7.55845230705566e-05, "loss": 4.0707, "step": 3653 }, { "epoch": 0.22682972251536407, "grad_norm": 0.688635170705502, "learning_rate": 7.560521415270019e-05, "loss": 4.1195, "step": 3654 }, { "epoch": 0.226891799615122, "grad_norm": 0.7605913867278177, "learning_rate": 7.562590523484379e-05, "loss": 4.1288, "step": 3655 }, { "epoch": 0.22695387671487988, "grad_norm": 0.7236886562776672, "learning_rate": 7.564659631698738e-05, "loss": 3.9936, "step": 3656 }, { "epoch": 0.22701595381463777, "grad_norm": 0.7089705452739271, "learning_rate": 7.566728739913098e-05, "loss": 4.0204, "step": 3657 }, { "epoch": 0.22707803091439568, "grad_norm": 0.6948441846368801, "learning_rate": 7.568797848127458e-05, "loss": 4.0785, "step": 3658 }, { "epoch": 0.22714010801415357, "grad_norm": 0.6949482510189073, "learning_rate": 7.570866956341817e-05, "loss": 4.037, "step": 3659 }, { "epoch": 0.22720218511391146, "grad_norm": 0.7214235669479099, "learning_rate": 7.572936064556176e-05, "loss": 4.057, "step": 3660 }, { "epoch": 0.22726426221366938, "grad_norm": 0.8377628184481647, "learning_rate": 7.575005172770536e-05, "loss": 4.1334, "step": 3661 }, { "epoch": 0.22732633931342727, "grad_norm": 0.9415083074056322, "learning_rate": 7.577074280984896e-05, "loss": 4.0875, "step": 3662 }, { "epoch": 0.22738841641318516, "grad_norm": 1.0360411118925814, "learning_rate": 7.579143389199256e-05, "loss": 4.0538, "step": 3663 }, { "epoch": 0.22745049351294308, "grad_norm": 1.046325083673854, "learning_rate": 7.581212497413615e-05, "loss": 4.0105, "step": 3664 }, { "epoch": 0.22751257061270097, "grad_norm": 1.1186746156324459, "learning_rate": 7.583281605627975e-05, "loss": 4.0404, "step": 3665 }, { "epoch": 0.22757464771245886, "grad_norm": 0.9117822489326544, "learning_rate": 7.585350713842334e-05, "loss": 4.004, "step": 3666 }, { "epoch": 0.22763672481221678, "grad_norm": 0.82085193880815, "learning_rate": 7.587419822056694e-05, "loss": 4.1933, "step": 3667 }, { "epoch": 0.22769880191197467, "grad_norm": 0.8412794040568851, "learning_rate": 7.589488930271053e-05, "loss": 4.1287, "step": 3668 }, { "epoch": 0.22776087901173256, "grad_norm": 0.9194933628259673, "learning_rate": 7.591558038485413e-05, "loss": 4.2393, "step": 3669 }, { "epoch": 0.22782295611149048, "grad_norm": 0.970835747668057, "learning_rate": 7.593627146699774e-05, "loss": 4.1329, "step": 3670 }, { "epoch": 0.22788503321124837, "grad_norm": 0.759808615127727, "learning_rate": 7.595696254914132e-05, "loss": 4.069, "step": 3671 }, { "epoch": 0.22794711031100626, "grad_norm": 0.7908321591707872, "learning_rate": 7.597765363128491e-05, "loss": 4.0935, "step": 3672 }, { "epoch": 0.22800918741076417, "grad_norm": 0.8501908691065795, "learning_rate": 7.599834471342851e-05, "loss": 4.1571, "step": 3673 }, { "epoch": 0.22807126451052206, "grad_norm": 0.6545336849907708, "learning_rate": 7.601903579557212e-05, "loss": 4.113, "step": 3674 }, { "epoch": 0.22813334161027995, "grad_norm": 0.7105891034295497, "learning_rate": 7.603972687771571e-05, "loss": 4.078, "step": 3675 }, { "epoch": 0.22819541871003787, "grad_norm": 0.7847772698988641, "learning_rate": 7.606041795985931e-05, "loss": 4.0637, "step": 3676 }, { "epoch": 0.22825749580979576, "grad_norm": 0.9150871663187131, "learning_rate": 7.608110904200289e-05, "loss": 4.1638, "step": 3677 }, { "epoch": 0.22831957290955365, "grad_norm": 0.8480359470782629, "learning_rate": 7.61018001241465e-05, "loss": 4.0971, "step": 3678 }, { "epoch": 0.22838165000931157, "grad_norm": 0.849095928220653, "learning_rate": 7.61224912062901e-05, "loss": 4.1525, "step": 3679 }, { "epoch": 0.22844372710906946, "grad_norm": 0.8118508019486301, "learning_rate": 7.614318228843369e-05, "loss": 4.1331, "step": 3680 }, { "epoch": 0.22850580420882735, "grad_norm": 0.8781832964134592, "learning_rate": 7.616387337057728e-05, "loss": 4.1599, "step": 3681 }, { "epoch": 0.22856788130858527, "grad_norm": 0.7065315166565516, "learning_rate": 7.618456445272088e-05, "loss": 4.091, "step": 3682 }, { "epoch": 0.22862995840834316, "grad_norm": 0.7599631398447753, "learning_rate": 7.620525553486447e-05, "loss": 4.0967, "step": 3683 }, { "epoch": 0.22869203550810105, "grad_norm": 0.8166900944431007, "learning_rate": 7.622594661700807e-05, "loss": 4.0427, "step": 3684 }, { "epoch": 0.22875411260785897, "grad_norm": 0.862762329731962, "learning_rate": 7.624663769915166e-05, "loss": 4.0709, "step": 3685 }, { "epoch": 0.22881618970761686, "grad_norm": 0.7445850797169133, "learning_rate": 7.626732878129526e-05, "loss": 4.0137, "step": 3686 }, { "epoch": 0.22887826680737475, "grad_norm": 0.6674743964610405, "learning_rate": 7.628801986343887e-05, "loss": 4.0037, "step": 3687 }, { "epoch": 0.22894034390713267, "grad_norm": 0.7625356744019031, "learning_rate": 7.630871094558246e-05, "loss": 4.1031, "step": 3688 }, { "epoch": 0.22900242100689056, "grad_norm": 0.8598553442168997, "learning_rate": 7.632940202772604e-05, "loss": 4.093, "step": 3689 }, { "epoch": 0.22906449810664845, "grad_norm": 0.8909666510355179, "learning_rate": 7.635009310986965e-05, "loss": 4.0718, "step": 3690 }, { "epoch": 0.22912657520640636, "grad_norm": 0.7892768493184094, "learning_rate": 7.637078419201325e-05, "loss": 4.0473, "step": 3691 }, { "epoch": 0.22918865230616425, "grad_norm": 0.8762683871376649, "learning_rate": 7.639147527415684e-05, "loss": 4.1445, "step": 3692 }, { "epoch": 0.22925072940592214, "grad_norm": 1.0126499912637725, "learning_rate": 7.641216635630044e-05, "loss": 4.0946, "step": 3693 }, { "epoch": 0.22931280650568006, "grad_norm": 1.1420104613651718, "learning_rate": 7.643285743844403e-05, "loss": 4.0611, "step": 3694 }, { "epoch": 0.22937488360543795, "grad_norm": 0.9027893485444467, "learning_rate": 7.645354852058763e-05, "loss": 3.9552, "step": 3695 }, { "epoch": 0.22943696070519584, "grad_norm": 0.9422959678046261, "learning_rate": 7.647423960273122e-05, "loss": 4.0224, "step": 3696 }, { "epoch": 0.22949903780495376, "grad_norm": 1.1618040565308267, "learning_rate": 7.649493068487482e-05, "loss": 4.0516, "step": 3697 }, { "epoch": 0.22956111490471165, "grad_norm": 0.8327409466606649, "learning_rate": 7.651562176701841e-05, "loss": 3.9906, "step": 3698 }, { "epoch": 0.22962319200446954, "grad_norm": 0.7873197137245216, "learning_rate": 7.653631284916202e-05, "loss": 4.0149, "step": 3699 }, { "epoch": 0.22968526910422746, "grad_norm": 1.0043957181264256, "learning_rate": 7.655700393130562e-05, "loss": 4.1264, "step": 3700 }, { "epoch": 0.22974734620398535, "grad_norm": 0.9487598393046892, "learning_rate": 7.65776950134492e-05, "loss": 4.1587, "step": 3701 }, { "epoch": 0.22980942330374324, "grad_norm": 0.776733887639447, "learning_rate": 7.65983860955928e-05, "loss": 4.116, "step": 3702 }, { "epoch": 0.22987150040350116, "grad_norm": 0.8216271900755449, "learning_rate": 7.66190771777364e-05, "loss": 4.0334, "step": 3703 }, { "epoch": 0.22993357750325905, "grad_norm": 0.8941466658335252, "learning_rate": 7.663976825988e-05, "loss": 4.0436, "step": 3704 }, { "epoch": 0.22999565460301694, "grad_norm": 0.8266579931583483, "learning_rate": 7.666045934202359e-05, "loss": 3.9746, "step": 3705 }, { "epoch": 0.23005773170277485, "grad_norm": 0.7559160709064053, "learning_rate": 7.668115042416719e-05, "loss": 4.1124, "step": 3706 }, { "epoch": 0.23011980880253274, "grad_norm": 0.7991224646716786, "learning_rate": 7.670184150631078e-05, "loss": 4.177, "step": 3707 }, { "epoch": 0.23018188590229063, "grad_norm": 0.8637756726966873, "learning_rate": 7.672253258845438e-05, "loss": 4.1028, "step": 3708 }, { "epoch": 0.23024396300204855, "grad_norm": 0.8187545514162243, "learning_rate": 7.674322367059797e-05, "loss": 3.9373, "step": 3709 }, { "epoch": 0.23030604010180644, "grad_norm": 0.6202643993186934, "learning_rate": 7.676391475274157e-05, "loss": 4.0524, "step": 3710 }, { "epoch": 0.23036811720156433, "grad_norm": 0.7770194906100115, "learning_rate": 7.678460583488518e-05, "loss": 3.9499, "step": 3711 }, { "epoch": 0.23043019430132225, "grad_norm": 0.7399782729181101, "learning_rate": 7.680529691702877e-05, "loss": 4.0488, "step": 3712 }, { "epoch": 0.23049227140108014, "grad_norm": 0.7234822681655758, "learning_rate": 7.682598799917235e-05, "loss": 4.0277, "step": 3713 }, { "epoch": 0.23055434850083803, "grad_norm": 0.7014801495313596, "learning_rate": 7.684667908131595e-05, "loss": 4.0291, "step": 3714 }, { "epoch": 0.23061642560059595, "grad_norm": 0.7241480919402252, "learning_rate": 7.686737016345956e-05, "loss": 3.9326, "step": 3715 }, { "epoch": 0.23067850270035384, "grad_norm": 0.7704848756265001, "learning_rate": 7.688806124560315e-05, "loss": 3.9679, "step": 3716 }, { "epoch": 0.23074057980011173, "grad_norm": 0.8627118610628383, "learning_rate": 7.690875232774675e-05, "loss": 4.148, "step": 3717 }, { "epoch": 0.23080265689986965, "grad_norm": 0.8736808233204519, "learning_rate": 7.692944340989034e-05, "loss": 4.032, "step": 3718 }, { "epoch": 0.23086473399962754, "grad_norm": 0.911821242245379, "learning_rate": 7.695013449203394e-05, "loss": 4.0407, "step": 3719 }, { "epoch": 0.23092681109938543, "grad_norm": 0.7862092354157558, "learning_rate": 7.697082557417753e-05, "loss": 4.1416, "step": 3720 }, { "epoch": 0.23098888819914334, "grad_norm": 0.7898189582292003, "learning_rate": 7.699151665632113e-05, "loss": 4.0274, "step": 3721 }, { "epoch": 0.23105096529890123, "grad_norm": 0.8367499565623705, "learning_rate": 7.701220773846472e-05, "loss": 4.0618, "step": 3722 }, { "epoch": 0.23111304239865912, "grad_norm": 0.7838988983131956, "learning_rate": 7.703289882060832e-05, "loss": 4.0255, "step": 3723 }, { "epoch": 0.23117511949841704, "grad_norm": 0.7812054812144799, "learning_rate": 7.705358990275193e-05, "loss": 3.9788, "step": 3724 }, { "epoch": 0.23123719659817493, "grad_norm": 0.7495752562391118, "learning_rate": 7.707428098489551e-05, "loss": 3.96, "step": 3725 }, { "epoch": 0.23129927369793282, "grad_norm": 0.7871179698541094, "learning_rate": 7.70949720670391e-05, "loss": 3.9966, "step": 3726 }, { "epoch": 0.23136135079769074, "grad_norm": 0.8131102324426848, "learning_rate": 7.71156631491827e-05, "loss": 4.0614, "step": 3727 }, { "epoch": 0.23142342789744863, "grad_norm": 0.8766617322705527, "learning_rate": 7.713635423132631e-05, "loss": 4.0888, "step": 3728 }, { "epoch": 0.23148550499720652, "grad_norm": 0.9216825316640498, "learning_rate": 7.71570453134699e-05, "loss": 3.9609, "step": 3729 }, { "epoch": 0.23154758209696444, "grad_norm": 0.7930250852520876, "learning_rate": 7.71777363956135e-05, "loss": 4.0167, "step": 3730 }, { "epoch": 0.23160965919672233, "grad_norm": 0.7983286818911034, "learning_rate": 7.719842747775709e-05, "loss": 4.0239, "step": 3731 }, { "epoch": 0.23167173629648022, "grad_norm": 0.7466053831192014, "learning_rate": 7.721911855990069e-05, "loss": 4.0577, "step": 3732 }, { "epoch": 0.23173381339623814, "grad_norm": 0.8988274432946065, "learning_rate": 7.723980964204428e-05, "loss": 4.0542, "step": 3733 }, { "epoch": 0.23179589049599603, "grad_norm": 0.7497467964640703, "learning_rate": 7.726050072418788e-05, "loss": 4.0652, "step": 3734 }, { "epoch": 0.23185796759575392, "grad_norm": 0.8321717551070911, "learning_rate": 7.728119180633147e-05, "loss": 4.0035, "step": 3735 }, { "epoch": 0.23192004469551183, "grad_norm": 0.8528850679145785, "learning_rate": 7.730188288847508e-05, "loss": 3.9997, "step": 3736 }, { "epoch": 0.23198212179526972, "grad_norm": 0.9288819293021783, "learning_rate": 7.732257397061868e-05, "loss": 4.0343, "step": 3737 }, { "epoch": 0.23204419889502761, "grad_norm": 0.8889376064864511, "learning_rate": 7.734326505276226e-05, "loss": 3.9908, "step": 3738 }, { "epoch": 0.23210627599478553, "grad_norm": 0.9622152563727125, "learning_rate": 7.736395613490585e-05, "loss": 4.1049, "step": 3739 }, { "epoch": 0.23216835309454342, "grad_norm": 1.1415697130862466, "learning_rate": 7.738464721704946e-05, "loss": 4.1035, "step": 3740 }, { "epoch": 0.2322304301943013, "grad_norm": 0.8749798023072579, "learning_rate": 7.740533829919306e-05, "loss": 4.0396, "step": 3741 }, { "epoch": 0.23229250729405923, "grad_norm": 0.7337854331979967, "learning_rate": 7.742602938133665e-05, "loss": 4.0832, "step": 3742 }, { "epoch": 0.23235458439381712, "grad_norm": 0.7633408319182694, "learning_rate": 7.744672046348025e-05, "loss": 4.0329, "step": 3743 }, { "epoch": 0.232416661493575, "grad_norm": 0.801682760278019, "learning_rate": 7.746741154562384e-05, "loss": 4.0177, "step": 3744 }, { "epoch": 0.23247873859333293, "grad_norm": 0.6864700839281982, "learning_rate": 7.748810262776744e-05, "loss": 4.076, "step": 3745 }, { "epoch": 0.23254081569309082, "grad_norm": 0.6951660473493904, "learning_rate": 7.750879370991103e-05, "loss": 4.0768, "step": 3746 }, { "epoch": 0.2326028927928487, "grad_norm": 0.6860175622604919, "learning_rate": 7.752948479205463e-05, "loss": 4.1064, "step": 3747 }, { "epoch": 0.23266496989260663, "grad_norm": 0.7505517757760453, "learning_rate": 7.755017587419822e-05, "loss": 4.0276, "step": 3748 }, { "epoch": 0.23272704699236452, "grad_norm": 0.7217062814894526, "learning_rate": 7.757086695634183e-05, "loss": 4.0106, "step": 3749 }, { "epoch": 0.2327891240921224, "grad_norm": 0.7119123880081473, "learning_rate": 7.759155803848541e-05, "loss": 4.0169, "step": 3750 }, { "epoch": 0.23285120119188032, "grad_norm": 0.7163150841666588, "learning_rate": 7.761224912062901e-05, "loss": 4.0447, "step": 3751 }, { "epoch": 0.23291327829163821, "grad_norm": 0.7182332959700719, "learning_rate": 7.76329402027726e-05, "loss": 4.0554, "step": 3752 }, { "epoch": 0.2329753553913961, "grad_norm": 0.8384037536244958, "learning_rate": 7.765363128491621e-05, "loss": 4.0802, "step": 3753 }, { "epoch": 0.23303743249115402, "grad_norm": 0.8781598121557516, "learning_rate": 7.76743223670598e-05, "loss": 4.043, "step": 3754 }, { "epoch": 0.2330995095909119, "grad_norm": 0.8733676602688638, "learning_rate": 7.76950134492034e-05, "loss": 4.0737, "step": 3755 }, { "epoch": 0.2331615866906698, "grad_norm": 0.7521747213053956, "learning_rate": 7.771570453134698e-05, "loss": 4.0077, "step": 3756 }, { "epoch": 0.23322366379042772, "grad_norm": 0.6451810624424952, "learning_rate": 7.773639561349059e-05, "loss": 4.0595, "step": 3757 }, { "epoch": 0.2332857408901856, "grad_norm": 0.7088677176348835, "learning_rate": 7.775708669563419e-05, "loss": 3.98, "step": 3758 }, { "epoch": 0.2333478179899435, "grad_norm": 0.7352799865823345, "learning_rate": 7.777777777777778e-05, "loss": 4.0178, "step": 3759 }, { "epoch": 0.23340989508970142, "grad_norm": 0.7255311023636263, "learning_rate": 7.779846885992138e-05, "loss": 4.075, "step": 3760 }, { "epoch": 0.2334719721894593, "grad_norm": 0.6672856814754903, "learning_rate": 7.781915994206499e-05, "loss": 4.0162, "step": 3761 }, { "epoch": 0.2335340492892172, "grad_norm": 0.6925559533578074, "learning_rate": 7.783985102420857e-05, "loss": 4.0928, "step": 3762 }, { "epoch": 0.23359612638897512, "grad_norm": 0.5762032969037202, "learning_rate": 7.786054210635216e-05, "loss": 4.0654, "step": 3763 }, { "epoch": 0.233658203488733, "grad_norm": 0.657909520259197, "learning_rate": 7.788123318849576e-05, "loss": 4.1086, "step": 3764 }, { "epoch": 0.2337202805884909, "grad_norm": 0.6687921947465835, "learning_rate": 7.790192427063937e-05, "loss": 3.9955, "step": 3765 }, { "epoch": 0.23378235768824882, "grad_norm": 0.9160601964977426, "learning_rate": 7.792261535278296e-05, "loss": 4.0073, "step": 3766 }, { "epoch": 0.2338444347880067, "grad_norm": 1.0324412314844444, "learning_rate": 7.794330643492656e-05, "loss": 4.0734, "step": 3767 }, { "epoch": 0.2339065118877646, "grad_norm": 0.987110545222227, "learning_rate": 7.796399751707014e-05, "loss": 4.0562, "step": 3768 }, { "epoch": 0.2339685889875225, "grad_norm": 0.8277937748520321, "learning_rate": 7.798468859921375e-05, "loss": 4.0107, "step": 3769 }, { "epoch": 0.2340306660872804, "grad_norm": 0.7267140876392447, "learning_rate": 7.800537968135734e-05, "loss": 3.9919, "step": 3770 }, { "epoch": 0.2340927431870383, "grad_norm": 0.8282525666067861, "learning_rate": 7.802607076350094e-05, "loss": 4.1439, "step": 3771 }, { "epoch": 0.2341548202867962, "grad_norm": 0.7859941253588171, "learning_rate": 7.804676184564453e-05, "loss": 4.0142, "step": 3772 }, { "epoch": 0.2342168973865541, "grad_norm": 0.7491587848552818, "learning_rate": 7.806745292778813e-05, "loss": 4.0159, "step": 3773 }, { "epoch": 0.234278974486312, "grad_norm": 0.723973162016641, "learning_rate": 7.808814400993172e-05, "loss": 4.0419, "step": 3774 }, { "epoch": 0.2343410515860699, "grad_norm": 0.8017735357752369, "learning_rate": 7.810883509207532e-05, "loss": 3.9822, "step": 3775 }, { "epoch": 0.2344031286858278, "grad_norm": 0.8667510025243075, "learning_rate": 7.812952617421891e-05, "loss": 4.1095, "step": 3776 }, { "epoch": 0.2344652057855857, "grad_norm": 0.8012246369797617, "learning_rate": 7.81502172563625e-05, "loss": 4.1324, "step": 3777 }, { "epoch": 0.2345272828853436, "grad_norm": 0.8328307702670973, "learning_rate": 7.817090833850612e-05, "loss": 3.933, "step": 3778 }, { "epoch": 0.2345893599851015, "grad_norm": 0.9995984522948356, "learning_rate": 7.819159942064971e-05, "loss": 3.9831, "step": 3779 }, { "epoch": 0.2346514370848594, "grad_norm": 1.0093043913813464, "learning_rate": 7.821229050279329e-05, "loss": 3.9682, "step": 3780 }, { "epoch": 0.2347135141846173, "grad_norm": 1.0853360932672012, "learning_rate": 7.823298158493689e-05, "loss": 3.9402, "step": 3781 }, { "epoch": 0.2347755912843752, "grad_norm": 0.8411964346169352, "learning_rate": 7.82536726670805e-05, "loss": 3.94, "step": 3782 }, { "epoch": 0.23483766838413309, "grad_norm": 0.8530049487593139, "learning_rate": 7.827436374922409e-05, "loss": 3.9556, "step": 3783 }, { "epoch": 0.234899745483891, "grad_norm": 0.757603095312754, "learning_rate": 7.829505483136769e-05, "loss": 4.0268, "step": 3784 }, { "epoch": 0.2349618225836489, "grad_norm": 0.8127935686147049, "learning_rate": 7.831574591351128e-05, "loss": 3.9572, "step": 3785 }, { "epoch": 0.23502389968340678, "grad_norm": 0.9018764524388663, "learning_rate": 7.833643699565488e-05, "loss": 4.0312, "step": 3786 }, { "epoch": 0.2350859767831647, "grad_norm": 0.9962027695778727, "learning_rate": 7.835712807779847e-05, "loss": 4.0853, "step": 3787 }, { "epoch": 0.2351480538829226, "grad_norm": 1.0042928700791949, "learning_rate": 7.837781915994207e-05, "loss": 3.9317, "step": 3788 }, { "epoch": 0.23521013098268048, "grad_norm": 0.8779871903747213, "learning_rate": 7.839851024208566e-05, "loss": 4.0412, "step": 3789 }, { "epoch": 0.2352722080824384, "grad_norm": 0.7836634931020058, "learning_rate": 7.841920132422927e-05, "loss": 3.9491, "step": 3790 }, { "epoch": 0.2353342851821963, "grad_norm": 0.8399447358208117, "learning_rate": 7.843989240637286e-05, "loss": 3.9858, "step": 3791 }, { "epoch": 0.23539636228195418, "grad_norm": 0.8159594613739035, "learning_rate": 7.846058348851645e-05, "loss": 4.0328, "step": 3792 }, { "epoch": 0.2354584393817121, "grad_norm": 0.7406668091896027, "learning_rate": 7.848127457066004e-05, "loss": 3.9767, "step": 3793 }, { "epoch": 0.23552051648147, "grad_norm": 0.7241106810692466, "learning_rate": 7.850196565280365e-05, "loss": 4.0504, "step": 3794 }, { "epoch": 0.23558259358122788, "grad_norm": 0.6662621587177043, "learning_rate": 7.852265673494724e-05, "loss": 3.9656, "step": 3795 }, { "epoch": 0.2356446706809858, "grad_norm": 0.8220064747014718, "learning_rate": 7.854334781709084e-05, "loss": 3.9753, "step": 3796 }, { "epoch": 0.2357067477807437, "grad_norm": 1.018738537121273, "learning_rate": 7.856403889923444e-05, "loss": 4.1467, "step": 3797 }, { "epoch": 0.23576882488050158, "grad_norm": 1.2121806009149871, "learning_rate": 7.858472998137803e-05, "loss": 4.0243, "step": 3798 }, { "epoch": 0.2358309019802595, "grad_norm": 0.6353290111762324, "learning_rate": 7.860542106352163e-05, "loss": 4.02, "step": 3799 }, { "epoch": 0.23589297908001738, "grad_norm": 0.9475417845578757, "learning_rate": 7.862611214566522e-05, "loss": 3.9666, "step": 3800 }, { "epoch": 0.23595505617977527, "grad_norm": 1.28945588553895, "learning_rate": 7.864680322780882e-05, "loss": 4.0456, "step": 3801 }, { "epoch": 0.2360171332795332, "grad_norm": 0.7610258031598559, "learning_rate": 7.866749430995241e-05, "loss": 4.0842, "step": 3802 }, { "epoch": 0.23607921037929108, "grad_norm": 0.8684321569914627, "learning_rate": 7.868818539209602e-05, "loss": 4.0176, "step": 3803 }, { "epoch": 0.23614128747904897, "grad_norm": 0.822600199210924, "learning_rate": 7.87088764742396e-05, "loss": 3.903, "step": 3804 }, { "epoch": 0.2362033645788069, "grad_norm": 0.8194346862865698, "learning_rate": 7.87295675563832e-05, "loss": 4.0508, "step": 3805 }, { "epoch": 0.23626544167856478, "grad_norm": 0.9258099644066822, "learning_rate": 7.875025863852679e-05, "loss": 4.0477, "step": 3806 }, { "epoch": 0.23632751877832267, "grad_norm": 0.9526208576111829, "learning_rate": 7.87709497206704e-05, "loss": 4.0437, "step": 3807 }, { "epoch": 0.2363895958780806, "grad_norm": 0.8861326053275257, "learning_rate": 7.8791640802814e-05, "loss": 3.9886, "step": 3808 }, { "epoch": 0.23645167297783848, "grad_norm": 1.0489858157141683, "learning_rate": 7.881233188495759e-05, "loss": 4.0454, "step": 3809 }, { "epoch": 0.23651375007759637, "grad_norm": 1.0767557599099846, "learning_rate": 7.883302296710117e-05, "loss": 4.001, "step": 3810 }, { "epoch": 0.2365758271773543, "grad_norm": 0.8601575929386975, "learning_rate": 7.885371404924478e-05, "loss": 4.1183, "step": 3811 }, { "epoch": 0.23663790427711218, "grad_norm": 0.8470177772912767, "learning_rate": 7.887440513138837e-05, "loss": 3.9931, "step": 3812 }, { "epoch": 0.23669998137687007, "grad_norm": 0.9379892979725099, "learning_rate": 7.889509621353197e-05, "loss": 3.9289, "step": 3813 }, { "epoch": 0.23676205847662798, "grad_norm": 0.8598391579878488, "learning_rate": 7.891578729567556e-05, "loss": 3.8656, "step": 3814 }, { "epoch": 0.23682413557638587, "grad_norm": 0.7979971894021123, "learning_rate": 7.893647837781917e-05, "loss": 4.03, "step": 3815 }, { "epoch": 0.23688621267614376, "grad_norm": 0.8210865183764969, "learning_rate": 7.895716945996277e-05, "loss": 3.9499, "step": 3816 }, { "epoch": 0.23694828977590168, "grad_norm": 0.8262081273559435, "learning_rate": 7.897786054210635e-05, "loss": 4.0056, "step": 3817 }, { "epoch": 0.23701036687565957, "grad_norm": 0.6208264237431133, "learning_rate": 7.899855162424995e-05, "loss": 4.0178, "step": 3818 }, { "epoch": 0.23707244397541746, "grad_norm": 0.7814043274085198, "learning_rate": 7.901924270639355e-05, "loss": 3.996, "step": 3819 }, { "epoch": 0.23713452107517538, "grad_norm": 0.7965325012185231, "learning_rate": 7.903993378853715e-05, "loss": 4.1342, "step": 3820 }, { "epoch": 0.23719659817493327, "grad_norm": 0.8625348346696943, "learning_rate": 7.906062487068074e-05, "loss": 4.0156, "step": 3821 }, { "epoch": 0.23725867527469116, "grad_norm": 0.7527301291329681, "learning_rate": 7.908131595282434e-05, "loss": 4.1071, "step": 3822 }, { "epoch": 0.23732075237444908, "grad_norm": 0.7345728911943114, "learning_rate": 7.910200703496793e-05, "loss": 4.017, "step": 3823 }, { "epoch": 0.23738282947420697, "grad_norm": 0.6968624541517563, "learning_rate": 7.912269811711153e-05, "loss": 3.9822, "step": 3824 }, { "epoch": 0.23744490657396486, "grad_norm": 0.7400227347309684, "learning_rate": 7.914338919925512e-05, "loss": 3.9299, "step": 3825 }, { "epoch": 0.23750698367372278, "grad_norm": 0.8365069793642584, "learning_rate": 7.916408028139872e-05, "loss": 3.9917, "step": 3826 }, { "epoch": 0.23756906077348067, "grad_norm": 0.9223740730038011, "learning_rate": 7.918477136354231e-05, "loss": 4.0662, "step": 3827 }, { "epoch": 0.23763113787323856, "grad_norm": 0.9221046614273941, "learning_rate": 7.920546244568592e-05, "loss": 3.9711, "step": 3828 }, { "epoch": 0.23769321497299647, "grad_norm": 0.9071480477874343, "learning_rate": 7.92261535278295e-05, "loss": 3.9033, "step": 3829 }, { "epoch": 0.23775529207275437, "grad_norm": 0.8434718697927532, "learning_rate": 7.92468446099731e-05, "loss": 4.11, "step": 3830 }, { "epoch": 0.23781736917251226, "grad_norm": 0.7121333868903191, "learning_rate": 7.92675356921167e-05, "loss": 3.9624, "step": 3831 }, { "epoch": 0.23787944627227017, "grad_norm": 0.7132474659401061, "learning_rate": 7.92882267742603e-05, "loss": 4.0592, "step": 3832 }, { "epoch": 0.23794152337202806, "grad_norm": 0.8072886600233304, "learning_rate": 7.93089178564039e-05, "loss": 4.0449, "step": 3833 }, { "epoch": 0.23800360047178595, "grad_norm": 0.7447934628622809, "learning_rate": 7.93296089385475e-05, "loss": 4.0021, "step": 3834 }, { "epoch": 0.23806567757154387, "grad_norm": 0.7390623048294165, "learning_rate": 7.935030002069108e-05, "loss": 4.0181, "step": 3835 }, { "epoch": 0.23812775467130176, "grad_norm": 0.8217239226522808, "learning_rate": 7.937099110283468e-05, "loss": 3.9912, "step": 3836 }, { "epoch": 0.23818983177105965, "grad_norm": 0.7144447422109366, "learning_rate": 7.939168218497828e-05, "loss": 4.026, "step": 3837 }, { "epoch": 0.23825190887081757, "grad_norm": 0.7423866807796196, "learning_rate": 7.941237326712187e-05, "loss": 4.1156, "step": 3838 }, { "epoch": 0.23831398597057546, "grad_norm": 0.8003420113600546, "learning_rate": 7.943306434926547e-05, "loss": 4.0645, "step": 3839 }, { "epoch": 0.23837606307033335, "grad_norm": 0.7640494269440342, "learning_rate": 7.945375543140908e-05, "loss": 4.0433, "step": 3840 }, { "epoch": 0.23843814017009124, "grad_norm": 0.8303914634362278, "learning_rate": 7.947444651355266e-05, "loss": 3.9649, "step": 3841 }, { "epoch": 0.23850021726984916, "grad_norm": 0.7789669629217775, "learning_rate": 7.949513759569625e-05, "loss": 3.9255, "step": 3842 }, { "epoch": 0.23856229436960705, "grad_norm": 0.7454477679262769, "learning_rate": 7.951582867783985e-05, "loss": 3.915, "step": 3843 }, { "epoch": 0.23862437146936494, "grad_norm": 0.8696820576449866, "learning_rate": 7.953651975998346e-05, "loss": 4.0723, "step": 3844 }, { "epoch": 0.23868644856912286, "grad_norm": 0.8686485354772655, "learning_rate": 7.955721084212705e-05, "loss": 3.9923, "step": 3845 }, { "epoch": 0.23874852566888075, "grad_norm": 0.9002520605612968, "learning_rate": 7.957790192427065e-05, "loss": 3.8483, "step": 3846 }, { "epoch": 0.23881060276863864, "grad_norm": 0.9384165416184813, "learning_rate": 7.959859300641423e-05, "loss": 4.0117, "step": 3847 }, { "epoch": 0.23887267986839655, "grad_norm": 1.110815644053025, "learning_rate": 7.961928408855784e-05, "loss": 4.0249, "step": 3848 }, { "epoch": 0.23893475696815444, "grad_norm": 0.8361005178486662, "learning_rate": 7.963997517070143e-05, "loss": 4.1178, "step": 3849 }, { "epoch": 0.23899683406791233, "grad_norm": 0.6682068695923243, "learning_rate": 7.966066625284503e-05, "loss": 3.9549, "step": 3850 }, { "epoch": 0.23905891116767025, "grad_norm": 1.090332960580241, "learning_rate": 7.968135733498862e-05, "loss": 3.8888, "step": 3851 }, { "epoch": 0.23912098826742814, "grad_norm": 0.9800887568335405, "learning_rate": 7.970204841713222e-05, "loss": 4.0735, "step": 3852 }, { "epoch": 0.23918306536718603, "grad_norm": 0.6836013139578734, "learning_rate": 7.972273949927581e-05, "loss": 4.0178, "step": 3853 }, { "epoch": 0.23924514246694395, "grad_norm": 0.8435811474292048, "learning_rate": 7.974343058141941e-05, "loss": 4.05, "step": 3854 }, { "epoch": 0.23930721956670184, "grad_norm": 0.8342811003007321, "learning_rate": 7.9764121663563e-05, "loss": 3.9867, "step": 3855 }, { "epoch": 0.23936929666645973, "grad_norm": 0.808173950612741, "learning_rate": 7.97848127457066e-05, "loss": 4.0539, "step": 3856 }, { "epoch": 0.23943137376621765, "grad_norm": 0.8119223815623916, "learning_rate": 7.980550382785021e-05, "loss": 4.0178, "step": 3857 }, { "epoch": 0.23949345086597554, "grad_norm": 0.8602859870436294, "learning_rate": 7.98261949099938e-05, "loss": 3.9337, "step": 3858 }, { "epoch": 0.23955552796573343, "grad_norm": 0.8555948869910889, "learning_rate": 7.984688599213738e-05, "loss": 3.9574, "step": 3859 }, { "epoch": 0.23961760506549135, "grad_norm": 0.7446736506792289, "learning_rate": 7.986757707428099e-05, "loss": 3.9789, "step": 3860 }, { "epoch": 0.23967968216524924, "grad_norm": 0.69959557890801, "learning_rate": 7.988826815642459e-05, "loss": 4.0187, "step": 3861 }, { "epoch": 0.23974175926500713, "grad_norm": 0.9221087014049515, "learning_rate": 7.990895923856818e-05, "loss": 4.0182, "step": 3862 }, { "epoch": 0.23980383636476504, "grad_norm": 0.9187347787128153, "learning_rate": 7.992965032071178e-05, "loss": 4.0143, "step": 3863 }, { "epoch": 0.23986591346452293, "grad_norm": 0.8330592578953926, "learning_rate": 7.995034140285537e-05, "loss": 4.0384, "step": 3864 }, { "epoch": 0.23992799056428082, "grad_norm": 0.8941674304708849, "learning_rate": 7.997103248499897e-05, "loss": 3.9263, "step": 3865 }, { "epoch": 0.23999006766403874, "grad_norm": 0.835861666082859, "learning_rate": 7.999172356714256e-05, "loss": 4.1329, "step": 3866 }, { "epoch": 0.24005214476379663, "grad_norm": 0.8049371943052268, "learning_rate": 8.001241464928616e-05, "loss": 4.0471, "step": 3867 }, { "epoch": 0.24011422186355452, "grad_norm": 0.7731244002565788, "learning_rate": 8.003310573142975e-05, "loss": 4.0415, "step": 3868 }, { "epoch": 0.24017629896331244, "grad_norm": 0.8515679945462251, "learning_rate": 8.005379681357336e-05, "loss": 3.9276, "step": 3869 }, { "epoch": 0.24023837606307033, "grad_norm": 0.8672315569159439, "learning_rate": 8.007448789571696e-05, "loss": 3.9393, "step": 3870 }, { "epoch": 0.24030045316282822, "grad_norm": 0.7911388160004261, "learning_rate": 8.009517897786054e-05, "loss": 3.9132, "step": 3871 }, { "epoch": 0.24036253026258614, "grad_norm": 0.7702957847901812, "learning_rate": 8.011587006000413e-05, "loss": 3.9692, "step": 3872 }, { "epoch": 0.24042460736234403, "grad_norm": 0.8842467750898186, "learning_rate": 8.013656114214774e-05, "loss": 3.9205, "step": 3873 }, { "epoch": 0.24048668446210192, "grad_norm": 0.7989205304635275, "learning_rate": 8.015725222429134e-05, "loss": 4.0541, "step": 3874 }, { "epoch": 0.24054876156185984, "grad_norm": 0.7435196816431032, "learning_rate": 8.017794330643493e-05, "loss": 4.0269, "step": 3875 }, { "epoch": 0.24061083866161773, "grad_norm": 0.8347671730123545, "learning_rate": 8.019863438857853e-05, "loss": 4.0469, "step": 3876 }, { "epoch": 0.24067291576137562, "grad_norm": 0.8243890378507314, "learning_rate": 8.021932547072212e-05, "loss": 4.0727, "step": 3877 }, { "epoch": 0.24073499286113353, "grad_norm": 0.8103925443130282, "learning_rate": 8.024001655286572e-05, "loss": 3.9719, "step": 3878 }, { "epoch": 0.24079706996089142, "grad_norm": 0.8450959384786761, "learning_rate": 8.026070763500931e-05, "loss": 3.8916, "step": 3879 }, { "epoch": 0.24085914706064931, "grad_norm": 0.8200225317580867, "learning_rate": 8.028139871715291e-05, "loss": 4.0843, "step": 3880 }, { "epoch": 0.24092122416040723, "grad_norm": 1.0803405001550805, "learning_rate": 8.03020897992965e-05, "loss": 3.9122, "step": 3881 }, { "epoch": 0.24098330126016512, "grad_norm": 0.8392447176083253, "learning_rate": 8.032278088144011e-05, "loss": 4.0342, "step": 3882 }, { "epoch": 0.241045378359923, "grad_norm": 0.6506154585330154, "learning_rate": 8.034347196358369e-05, "loss": 3.9502, "step": 3883 }, { "epoch": 0.24110745545968093, "grad_norm": 0.7762275234291601, "learning_rate": 8.036416304572729e-05, "loss": 4.0695, "step": 3884 }, { "epoch": 0.24116953255943882, "grad_norm": 0.9132367504233286, "learning_rate": 8.03848541278709e-05, "loss": 4.013, "step": 3885 }, { "epoch": 0.2412316096591967, "grad_norm": 1.0518820289925483, "learning_rate": 8.040554521001449e-05, "loss": 4.0394, "step": 3886 }, { "epoch": 0.24129368675895463, "grad_norm": 0.9280891462908235, "learning_rate": 8.042623629215809e-05, "loss": 3.9607, "step": 3887 }, { "epoch": 0.24135576385871252, "grad_norm": 1.0069651627144771, "learning_rate": 8.044692737430168e-05, "loss": 4.1815, "step": 3888 }, { "epoch": 0.2414178409584704, "grad_norm": 0.8176108057131066, "learning_rate": 8.046761845644528e-05, "loss": 3.8981, "step": 3889 }, { "epoch": 0.24147991805822833, "grad_norm": 0.8076723165273801, "learning_rate": 8.048830953858887e-05, "loss": 3.9925, "step": 3890 }, { "epoch": 0.24154199515798622, "grad_norm": 0.8496545538454798, "learning_rate": 8.050900062073247e-05, "loss": 3.9094, "step": 3891 }, { "epoch": 0.2416040722577441, "grad_norm": 0.781285895476639, "learning_rate": 8.052969170287606e-05, "loss": 3.9715, "step": 3892 }, { "epoch": 0.24166614935750202, "grad_norm": 0.7356221181315559, "learning_rate": 8.055038278501966e-05, "loss": 3.9795, "step": 3893 }, { "epoch": 0.24172822645725991, "grad_norm": 0.7013219603677329, "learning_rate": 8.057107386716327e-05, "loss": 4.0635, "step": 3894 }, { "epoch": 0.2417903035570178, "grad_norm": 0.7467760491363051, "learning_rate": 8.059176494930685e-05, "loss": 4.0124, "step": 3895 }, { "epoch": 0.24185238065677572, "grad_norm": 0.7896739365892655, "learning_rate": 8.061245603145044e-05, "loss": 4.032, "step": 3896 }, { "epoch": 0.2419144577565336, "grad_norm": 0.7997139121162641, "learning_rate": 8.063314711359404e-05, "loss": 3.9755, "step": 3897 }, { "epoch": 0.2419765348562915, "grad_norm": 0.8814008786518451, "learning_rate": 8.065383819573765e-05, "loss": 3.9531, "step": 3898 }, { "epoch": 0.24203861195604942, "grad_norm": 0.779370832638585, "learning_rate": 8.067452927788124e-05, "loss": 3.9318, "step": 3899 }, { "epoch": 0.2421006890558073, "grad_norm": 0.7861557477365724, "learning_rate": 8.069522036002484e-05, "loss": 4.0705, "step": 3900 }, { "epoch": 0.2421627661555652, "grad_norm": 0.9273850275123046, "learning_rate": 8.071591144216843e-05, "loss": 3.968, "step": 3901 }, { "epoch": 0.24222484325532312, "grad_norm": 0.9283975900901439, "learning_rate": 8.073660252431203e-05, "loss": 4.0702, "step": 3902 }, { "epoch": 0.242286920355081, "grad_norm": 0.8327240186570347, "learning_rate": 8.075729360645562e-05, "loss": 4.0021, "step": 3903 }, { "epoch": 0.2423489974548389, "grad_norm": 0.8606774950050528, "learning_rate": 8.077798468859922e-05, "loss": 3.951, "step": 3904 }, { "epoch": 0.24241107455459682, "grad_norm": 0.8876984011824016, "learning_rate": 8.079867577074281e-05, "loss": 4.0297, "step": 3905 }, { "epoch": 0.2424731516543547, "grad_norm": 0.9279138331262233, "learning_rate": 8.081936685288641e-05, "loss": 4.017, "step": 3906 }, { "epoch": 0.2425352287541126, "grad_norm": 0.8401412791698483, "learning_rate": 8.084005793503002e-05, "loss": 3.9883, "step": 3907 }, { "epoch": 0.24259730585387052, "grad_norm": 0.6790545574834973, "learning_rate": 8.08607490171736e-05, "loss": 3.9143, "step": 3908 }, { "epoch": 0.2426593829536284, "grad_norm": 0.7147589878212335, "learning_rate": 8.088144009931719e-05, "loss": 3.9864, "step": 3909 }, { "epoch": 0.2427214600533863, "grad_norm": 0.7217654124385924, "learning_rate": 8.09021311814608e-05, "loss": 3.996, "step": 3910 }, { "epoch": 0.2427835371531442, "grad_norm": 0.7286190337969032, "learning_rate": 8.09228222636044e-05, "loss": 4.0015, "step": 3911 }, { "epoch": 0.2428456142529021, "grad_norm": 0.7880747816768693, "learning_rate": 8.094351334574799e-05, "loss": 3.973, "step": 3912 }, { "epoch": 0.24290769135266, "grad_norm": 0.7480980678800087, "learning_rate": 8.096420442789159e-05, "loss": 3.891, "step": 3913 }, { "epoch": 0.2429697684524179, "grad_norm": 0.7571519155166097, "learning_rate": 8.098489551003518e-05, "loss": 3.9934, "step": 3914 }, { "epoch": 0.2430318455521758, "grad_norm": 0.8940740358047784, "learning_rate": 8.100558659217878e-05, "loss": 3.9752, "step": 3915 }, { "epoch": 0.2430939226519337, "grad_norm": 1.0107964395162221, "learning_rate": 8.102627767432237e-05, "loss": 3.9538, "step": 3916 }, { "epoch": 0.2431559997516916, "grad_norm": 0.8667281368965222, "learning_rate": 8.104696875646597e-05, "loss": 3.9996, "step": 3917 }, { "epoch": 0.2432180768514495, "grad_norm": 0.8041274705205793, "learning_rate": 8.106765983860956e-05, "loss": 3.9905, "step": 3918 }, { "epoch": 0.2432801539512074, "grad_norm": 0.8179359888549753, "learning_rate": 8.108835092075317e-05, "loss": 3.9379, "step": 3919 }, { "epoch": 0.2433422310509653, "grad_norm": 0.6664261210118658, "learning_rate": 8.110904200289675e-05, "loss": 3.979, "step": 3920 }, { "epoch": 0.2434043081507232, "grad_norm": 0.7257879907728212, "learning_rate": 8.112973308504035e-05, "loss": 4.0056, "step": 3921 }, { "epoch": 0.2434663852504811, "grad_norm": 0.6828137552699373, "learning_rate": 8.115042416718394e-05, "loss": 3.9943, "step": 3922 }, { "epoch": 0.243528462350239, "grad_norm": 0.7642871300260218, "learning_rate": 8.117111524932755e-05, "loss": 3.9939, "step": 3923 }, { "epoch": 0.2435905394499969, "grad_norm": 0.7517204420078212, "learning_rate": 8.119180633147115e-05, "loss": 3.9096, "step": 3924 }, { "epoch": 0.24365261654975479, "grad_norm": 0.705565265799792, "learning_rate": 8.121249741361474e-05, "loss": 3.9804, "step": 3925 }, { "epoch": 0.2437146936495127, "grad_norm": 0.6692406440981494, "learning_rate": 8.123318849575832e-05, "loss": 3.9865, "step": 3926 }, { "epoch": 0.2437767707492706, "grad_norm": 0.6442306048560562, "learning_rate": 8.125387957790193e-05, "loss": 3.8781, "step": 3927 }, { "epoch": 0.24383884784902848, "grad_norm": 0.7491164168399705, "learning_rate": 8.127457066004553e-05, "loss": 3.8364, "step": 3928 }, { "epoch": 0.2439009249487864, "grad_norm": 0.7551205595025157, "learning_rate": 8.129526174218912e-05, "loss": 3.865, "step": 3929 }, { "epoch": 0.2439630020485443, "grad_norm": 0.7280461330023205, "learning_rate": 8.131595282433272e-05, "loss": 3.976, "step": 3930 }, { "epoch": 0.24402507914830218, "grad_norm": 0.7005507286717066, "learning_rate": 8.133664390647631e-05, "loss": 3.9459, "step": 3931 }, { "epoch": 0.2440871562480601, "grad_norm": 0.6678180636155, "learning_rate": 8.13573349886199e-05, "loss": 3.997, "step": 3932 }, { "epoch": 0.244149233347818, "grad_norm": 0.6832882225645659, "learning_rate": 8.13780260707635e-05, "loss": 3.9298, "step": 3933 }, { "epoch": 0.24421131044757588, "grad_norm": 0.6579335959778805, "learning_rate": 8.13987171529071e-05, "loss": 3.9557, "step": 3934 }, { "epoch": 0.2442733875473338, "grad_norm": 0.7110716305475357, "learning_rate": 8.14194082350507e-05, "loss": 3.9969, "step": 3935 }, { "epoch": 0.2443354646470917, "grad_norm": 0.8087677330773324, "learning_rate": 8.14400993171943e-05, "loss": 3.8765, "step": 3936 }, { "epoch": 0.24439754174684958, "grad_norm": 0.8921555201048346, "learning_rate": 8.14607903993379e-05, "loss": 3.8994, "step": 3937 }, { "epoch": 0.2444596188466075, "grad_norm": 0.839665948543668, "learning_rate": 8.148148148148148e-05, "loss": 3.9946, "step": 3938 }, { "epoch": 0.24452169594636539, "grad_norm": 0.8180917015322526, "learning_rate": 8.150217256362508e-05, "loss": 4.0474, "step": 3939 }, { "epoch": 0.24458377304612328, "grad_norm": 0.8303095595911097, "learning_rate": 8.152286364576868e-05, "loss": 3.9996, "step": 3940 }, { "epoch": 0.2446458501458812, "grad_norm": 0.780904998896755, "learning_rate": 8.154355472791228e-05, "loss": 3.9217, "step": 3941 }, { "epoch": 0.24470792724563908, "grad_norm": 0.739703911815602, "learning_rate": 8.156424581005587e-05, "loss": 3.8837, "step": 3942 }, { "epoch": 0.24477000434539697, "grad_norm": 0.7564530517749547, "learning_rate": 8.158493689219947e-05, "loss": 4.035, "step": 3943 }, { "epoch": 0.2448320814451549, "grad_norm": 0.7541901444281038, "learning_rate": 8.160562797434306e-05, "loss": 4.0047, "step": 3944 }, { "epoch": 0.24489415854491278, "grad_norm": 0.699418897060803, "learning_rate": 8.162631905648666e-05, "loss": 4.0496, "step": 3945 }, { "epoch": 0.24495623564467067, "grad_norm": 0.7175091522909068, "learning_rate": 8.164701013863025e-05, "loss": 3.9445, "step": 3946 }, { "epoch": 0.2450183127444286, "grad_norm": 0.6601568368762415, "learning_rate": 8.166770122077385e-05, "loss": 3.9194, "step": 3947 }, { "epoch": 0.24508038984418648, "grad_norm": 0.687864423747817, "learning_rate": 8.168839230291745e-05, "loss": 3.9214, "step": 3948 }, { "epoch": 0.24514246694394437, "grad_norm": 0.815311628920944, "learning_rate": 8.170908338506105e-05, "loss": 3.9686, "step": 3949 }, { "epoch": 0.2452045440437023, "grad_norm": 0.8543018761284799, "learning_rate": 8.172977446720463e-05, "loss": 4.0111, "step": 3950 }, { "epoch": 0.24526662114346018, "grad_norm": 0.7292679080614771, "learning_rate": 8.175046554934823e-05, "loss": 3.9683, "step": 3951 }, { "epoch": 0.24532869824321807, "grad_norm": 0.6809979663768098, "learning_rate": 8.177115663149183e-05, "loss": 3.9833, "step": 3952 }, { "epoch": 0.245390775342976, "grad_norm": 0.8097141998494127, "learning_rate": 8.179184771363543e-05, "loss": 3.9025, "step": 3953 }, { "epoch": 0.24545285244273388, "grad_norm": 0.9964332450809831, "learning_rate": 8.181253879577902e-05, "loss": 3.993, "step": 3954 }, { "epoch": 0.24551492954249177, "grad_norm": 0.8648077125932343, "learning_rate": 8.183322987792262e-05, "loss": 3.8509, "step": 3955 }, { "epoch": 0.24557700664224968, "grad_norm": 0.8063228249003759, "learning_rate": 8.185392096006621e-05, "loss": 4.032, "step": 3956 }, { "epoch": 0.24563908374200757, "grad_norm": 0.8191450985707673, "learning_rate": 8.187461204220981e-05, "loss": 3.9795, "step": 3957 }, { "epoch": 0.24570116084176546, "grad_norm": 0.8407173758310041, "learning_rate": 8.18953031243534e-05, "loss": 4.0031, "step": 3958 }, { "epoch": 0.24576323794152338, "grad_norm": 0.9103812074775677, "learning_rate": 8.1915994206497e-05, "loss": 4.0045, "step": 3959 }, { "epoch": 0.24582531504128127, "grad_norm": 0.8386660102600092, "learning_rate": 8.193668528864061e-05, "loss": 3.9343, "step": 3960 }, { "epoch": 0.24588739214103916, "grad_norm": 0.8727769882566314, "learning_rate": 8.19573763707842e-05, "loss": 3.8607, "step": 3961 }, { "epoch": 0.24594946924079708, "grad_norm": 0.7733566164822452, "learning_rate": 8.197806745292779e-05, "loss": 3.8193, "step": 3962 }, { "epoch": 0.24601154634055497, "grad_norm": 0.6907917891489866, "learning_rate": 8.199875853507138e-05, "loss": 4.0478, "step": 3963 }, { "epoch": 0.24607362344031286, "grad_norm": 0.6609884580623094, "learning_rate": 8.201944961721499e-05, "loss": 3.8465, "step": 3964 }, { "epoch": 0.24613570054007078, "grad_norm": 0.7108498403981144, "learning_rate": 8.204014069935858e-05, "loss": 3.9815, "step": 3965 }, { "epoch": 0.24619777763982867, "grad_norm": 0.7850258182726313, "learning_rate": 8.206083178150218e-05, "loss": 3.9115, "step": 3966 }, { "epoch": 0.24625985473958656, "grad_norm": 0.7876788935237671, "learning_rate": 8.208152286364577e-05, "loss": 3.9158, "step": 3967 }, { "epoch": 0.24632193183934448, "grad_norm": 0.8608475293678112, "learning_rate": 8.210221394578937e-05, "loss": 3.9722, "step": 3968 }, { "epoch": 0.24638400893910237, "grad_norm": 0.8577857494526, "learning_rate": 8.212290502793296e-05, "loss": 3.9708, "step": 3969 }, { "epoch": 0.24644608603886026, "grad_norm": 0.9766352206810636, "learning_rate": 8.214359611007656e-05, "loss": 3.9961, "step": 3970 }, { "epoch": 0.24650816313861817, "grad_norm": 0.9084445755469442, "learning_rate": 8.216428719222015e-05, "loss": 3.9679, "step": 3971 }, { "epoch": 0.24657024023837606, "grad_norm": 1.0061907472402278, "learning_rate": 8.218497827436375e-05, "loss": 3.9432, "step": 3972 }, { "epoch": 0.24663231733813396, "grad_norm": 0.9708406285733588, "learning_rate": 8.220566935650736e-05, "loss": 3.952, "step": 3973 }, { "epoch": 0.24669439443789187, "grad_norm": 0.9494529403828222, "learning_rate": 8.222636043865094e-05, "loss": 3.7696, "step": 3974 }, { "epoch": 0.24675647153764976, "grad_norm": 0.8315699556712468, "learning_rate": 8.224705152079453e-05, "loss": 4.0095, "step": 3975 }, { "epoch": 0.24681854863740765, "grad_norm": 0.9503621565525312, "learning_rate": 8.226774260293813e-05, "loss": 4.0521, "step": 3976 }, { "epoch": 0.24688062573716557, "grad_norm": 1.080920313587957, "learning_rate": 8.228843368508174e-05, "loss": 3.9125, "step": 3977 }, { "epoch": 0.24694270283692346, "grad_norm": 0.9508689068874464, "learning_rate": 8.230912476722533e-05, "loss": 3.9585, "step": 3978 }, { "epoch": 0.24700477993668135, "grad_norm": 0.8114491783347129, "learning_rate": 8.232981584936893e-05, "loss": 3.884, "step": 3979 }, { "epoch": 0.24706685703643927, "grad_norm": 0.8391684367613573, "learning_rate": 8.235050693151251e-05, "loss": 3.8948, "step": 3980 }, { "epoch": 0.24712893413619716, "grad_norm": 0.9295193496073135, "learning_rate": 8.237119801365612e-05, "loss": 3.9222, "step": 3981 }, { "epoch": 0.24719101123595505, "grad_norm": 0.7637548377200685, "learning_rate": 8.239188909579971e-05, "loss": 3.935, "step": 3982 }, { "epoch": 0.24725308833571297, "grad_norm": 0.7921456260072571, "learning_rate": 8.241258017794331e-05, "loss": 4.0406, "step": 3983 }, { "epoch": 0.24731516543547086, "grad_norm": 0.7305095885585504, "learning_rate": 8.24332712600869e-05, "loss": 3.8547, "step": 3984 }, { "epoch": 0.24737724253522875, "grad_norm": 0.6807808249110868, "learning_rate": 8.245396234223051e-05, "loss": 4.0312, "step": 3985 }, { "epoch": 0.24743931963498667, "grad_norm": 0.7497289883199639, "learning_rate": 8.247465342437411e-05, "loss": 3.9989, "step": 3986 }, { "epoch": 0.24750139673474456, "grad_norm": 0.791790655532296, "learning_rate": 8.249534450651769e-05, "loss": 3.9185, "step": 3987 }, { "epoch": 0.24756347383450245, "grad_norm": 0.7077213102297674, "learning_rate": 8.251603558866128e-05, "loss": 3.927, "step": 3988 }, { "epoch": 0.24762555093426036, "grad_norm": 0.8101937133602602, "learning_rate": 8.253672667080489e-05, "loss": 3.9586, "step": 3989 }, { "epoch": 0.24768762803401825, "grad_norm": 0.8610319636146081, "learning_rate": 8.255741775294849e-05, "loss": 3.9449, "step": 3990 }, { "epoch": 0.24774970513377614, "grad_norm": 0.942864582467969, "learning_rate": 8.257810883509208e-05, "loss": 3.8928, "step": 3991 }, { "epoch": 0.24781178223353406, "grad_norm": 0.9404646093553388, "learning_rate": 8.259879991723568e-05, "loss": 3.8976, "step": 3992 }, { "epoch": 0.24787385933329195, "grad_norm": 0.8266083486493058, "learning_rate": 8.261949099937927e-05, "loss": 4.0381, "step": 3993 }, { "epoch": 0.24793593643304984, "grad_norm": 0.7073230878754292, "learning_rate": 8.264018208152287e-05, "loss": 3.9827, "step": 3994 }, { "epoch": 0.24799801353280776, "grad_norm": 0.7609986251312348, "learning_rate": 8.266087316366646e-05, "loss": 3.9599, "step": 3995 }, { "epoch": 0.24806009063256565, "grad_norm": 0.8496221782552723, "learning_rate": 8.268156424581006e-05, "loss": 4.0837, "step": 3996 }, { "epoch": 0.24812216773232354, "grad_norm": 0.6782832053436869, "learning_rate": 8.270225532795365e-05, "loss": 3.8437, "step": 3997 }, { "epoch": 0.24818424483208146, "grad_norm": 0.6895455144517098, "learning_rate": 8.272294641009726e-05, "loss": 3.8421, "step": 3998 }, { "epoch": 0.24824632193183935, "grad_norm": 0.7602479477528222, "learning_rate": 8.274363749224084e-05, "loss": 3.9225, "step": 3999 }, { "epoch": 0.24830839903159724, "grad_norm": 0.7961885450113597, "learning_rate": 8.276432857438444e-05, "loss": 4.036, "step": 4000 }, { "epoch": 0.24837047613135516, "grad_norm": 0.7643738859079622, "learning_rate": 8.278501965652803e-05, "loss": 3.915, "step": 4001 }, { "epoch": 0.24843255323111305, "grad_norm": 0.9060363188762878, "learning_rate": 8.280571073867164e-05, "loss": 3.8737, "step": 4002 }, { "epoch": 0.24849463033087094, "grad_norm": 0.8249820591680881, "learning_rate": 8.282640182081524e-05, "loss": 3.9203, "step": 4003 }, { "epoch": 0.24855670743062885, "grad_norm": 0.7305759208517534, "learning_rate": 8.284709290295883e-05, "loss": 3.9081, "step": 4004 }, { "epoch": 0.24861878453038674, "grad_norm": 0.8428476594883265, "learning_rate": 8.286778398510241e-05, "loss": 3.898, "step": 4005 }, { "epoch": 0.24868086163014463, "grad_norm": 0.7357615914658214, "learning_rate": 8.288847506724602e-05, "loss": 4.028, "step": 4006 }, { "epoch": 0.24874293872990255, "grad_norm": 0.6507659608068462, "learning_rate": 8.290916614938962e-05, "loss": 3.9443, "step": 4007 }, { "epoch": 0.24880501582966044, "grad_norm": 0.7941645358304887, "learning_rate": 8.292985723153321e-05, "loss": 3.9716, "step": 4008 }, { "epoch": 0.24886709292941833, "grad_norm": 0.7544618051896065, "learning_rate": 8.295054831367681e-05, "loss": 3.8564, "step": 4009 }, { "epoch": 0.24892917002917625, "grad_norm": 0.6388950249704378, "learning_rate": 8.297123939582042e-05, "loss": 3.9555, "step": 4010 }, { "epoch": 0.24899124712893414, "grad_norm": 0.6668932996418147, "learning_rate": 8.2991930477964e-05, "loss": 3.9632, "step": 4011 }, { "epoch": 0.24905332422869203, "grad_norm": 0.722921209476793, "learning_rate": 8.30126215601076e-05, "loss": 4.0142, "step": 4012 }, { "epoch": 0.24911540132844995, "grad_norm": 0.7889912261499229, "learning_rate": 8.303331264225119e-05, "loss": 4.0189, "step": 4013 }, { "epoch": 0.24917747842820784, "grad_norm": 0.7954006018863902, "learning_rate": 8.30540037243948e-05, "loss": 4.0155, "step": 4014 }, { "epoch": 0.24923955552796573, "grad_norm": 0.8440125727937381, "learning_rate": 8.307469480653839e-05, "loss": 4.0798, "step": 4015 }, { "epoch": 0.24930163262772365, "grad_norm": 0.765543213193303, "learning_rate": 8.309538588868199e-05, "loss": 4.098, "step": 4016 }, { "epoch": 0.24936370972748154, "grad_norm": 0.7473310426991137, "learning_rate": 8.311607697082557e-05, "loss": 3.9973, "step": 4017 }, { "epoch": 0.24942578682723943, "grad_norm": 0.6636948633731181, "learning_rate": 8.313676805296918e-05, "loss": 4.0484, "step": 4018 }, { "epoch": 0.24948786392699734, "grad_norm": 0.6926934066158538, "learning_rate": 8.315745913511277e-05, "loss": 3.9708, "step": 4019 }, { "epoch": 0.24954994102675523, "grad_norm": 0.6885076204694054, "learning_rate": 8.317815021725637e-05, "loss": 3.8817, "step": 4020 }, { "epoch": 0.24961201812651312, "grad_norm": 0.6919915690514851, "learning_rate": 8.319884129939996e-05, "loss": 3.9032, "step": 4021 }, { "epoch": 0.24967409522627104, "grad_norm": 0.6846867787415237, "learning_rate": 8.321953238154356e-05, "loss": 3.9699, "step": 4022 }, { "epoch": 0.24973617232602893, "grad_norm": 0.7183861848309935, "learning_rate": 8.324022346368715e-05, "loss": 3.9792, "step": 4023 }, { "epoch": 0.24979824942578682, "grad_norm": 0.6897830983993796, "learning_rate": 8.326091454583075e-05, "loss": 3.897, "step": 4024 }, { "epoch": 0.24986032652554474, "grad_norm": 0.728777718123202, "learning_rate": 8.328160562797434e-05, "loss": 3.9055, "step": 4025 }, { "epoch": 0.24992240362530263, "grad_norm": 0.9259175711272501, "learning_rate": 8.330229671011794e-05, "loss": 4.0008, "step": 4026 }, { "epoch": 0.24998448072506052, "grad_norm": 1.1366143457242068, "learning_rate": 8.332298779226155e-05, "loss": 3.9765, "step": 4027 }, { "epoch": 0.25004655782481844, "grad_norm": 1.1399051014401622, "learning_rate": 8.334367887440514e-05, "loss": 3.9236, "step": 4028 }, { "epoch": 0.2501086349245763, "grad_norm": 0.7194679240671151, "learning_rate": 8.336436995654872e-05, "loss": 3.848, "step": 4029 }, { "epoch": 0.2501707120243342, "grad_norm": 0.9316752045489849, "learning_rate": 8.338506103869232e-05, "loss": 3.93, "step": 4030 }, { "epoch": 0.25023278912409214, "grad_norm": 1.1914420200668983, "learning_rate": 8.340575212083593e-05, "loss": 3.9038, "step": 4031 }, { "epoch": 0.25029486622385, "grad_norm": 0.7738068209415782, "learning_rate": 8.342644320297952e-05, "loss": 3.9045, "step": 4032 }, { "epoch": 0.2503569433236079, "grad_norm": 0.8333155073802437, "learning_rate": 8.344713428512312e-05, "loss": 3.8887, "step": 4033 }, { "epoch": 0.25041902042336583, "grad_norm": 1.048791071345589, "learning_rate": 8.346782536726671e-05, "loss": 4.0104, "step": 4034 }, { "epoch": 0.2504810975231237, "grad_norm": 0.8942608153830545, "learning_rate": 8.348851644941031e-05, "loss": 3.9454, "step": 4035 }, { "epoch": 0.2505431746228816, "grad_norm": 0.8580918182159032, "learning_rate": 8.35092075315539e-05, "loss": 3.9499, "step": 4036 }, { "epoch": 0.25060525172263953, "grad_norm": 0.8639595508467208, "learning_rate": 8.35298986136975e-05, "loss": 3.902, "step": 4037 }, { "epoch": 0.2506673288223974, "grad_norm": 0.7919437943994851, "learning_rate": 8.355058969584109e-05, "loss": 3.8568, "step": 4038 }, { "epoch": 0.2507294059221553, "grad_norm": 0.7842325119145696, "learning_rate": 8.35712807779847e-05, "loss": 3.9436, "step": 4039 }, { "epoch": 0.25079148302191323, "grad_norm": 0.8010578329160623, "learning_rate": 8.35919718601283e-05, "loss": 3.9936, "step": 4040 }, { "epoch": 0.2508535601216711, "grad_norm": 0.848880885205338, "learning_rate": 8.361266294227188e-05, "loss": 3.9393, "step": 4041 }, { "epoch": 0.250915637221429, "grad_norm": 0.8166611925901822, "learning_rate": 8.363335402441547e-05, "loss": 3.9422, "step": 4042 }, { "epoch": 0.25097771432118693, "grad_norm": 0.7303771507235435, "learning_rate": 8.365404510655908e-05, "loss": 3.9191, "step": 4043 }, { "epoch": 0.2510397914209448, "grad_norm": 0.7291165652546647, "learning_rate": 8.367473618870268e-05, "loss": 4.0496, "step": 4044 }, { "epoch": 0.2511018685207027, "grad_norm": 0.8220397929052401, "learning_rate": 8.369542727084627e-05, "loss": 3.9412, "step": 4045 }, { "epoch": 0.2511639456204606, "grad_norm": 0.7797614221179652, "learning_rate": 8.371611835298987e-05, "loss": 3.8154, "step": 4046 }, { "epoch": 0.2512260227202185, "grad_norm": 0.7276456818766461, "learning_rate": 8.373680943513346e-05, "loss": 3.9049, "step": 4047 }, { "epoch": 0.2512880998199764, "grad_norm": 0.7257740295510892, "learning_rate": 8.375750051727706e-05, "loss": 3.966, "step": 4048 }, { "epoch": 0.2513501769197343, "grad_norm": 0.7211924806663036, "learning_rate": 8.377819159942065e-05, "loss": 4.0148, "step": 4049 }, { "epoch": 0.2514122540194922, "grad_norm": 0.8063182417503464, "learning_rate": 8.379888268156425e-05, "loss": 3.9584, "step": 4050 }, { "epoch": 0.2514743311192501, "grad_norm": 0.8706576705575445, "learning_rate": 8.381957376370784e-05, "loss": 3.9893, "step": 4051 }, { "epoch": 0.251536408219008, "grad_norm": 0.7728879883853365, "learning_rate": 8.384026484585145e-05, "loss": 4.0549, "step": 4052 }, { "epoch": 0.2515984853187659, "grad_norm": 0.626041950494466, "learning_rate": 8.386095592799503e-05, "loss": 3.8674, "step": 4053 }, { "epoch": 0.2516605624185238, "grad_norm": 0.7878663467350843, "learning_rate": 8.388164701013863e-05, "loss": 3.8676, "step": 4054 }, { "epoch": 0.2517226395182817, "grad_norm": 0.9760482349534366, "learning_rate": 8.390233809228222e-05, "loss": 3.8831, "step": 4055 }, { "epoch": 0.2517847166180396, "grad_norm": 1.095787765851265, "learning_rate": 8.392302917442583e-05, "loss": 3.9042, "step": 4056 }, { "epoch": 0.2518467937177975, "grad_norm": 0.8735591059941761, "learning_rate": 8.394372025656943e-05, "loss": 3.9508, "step": 4057 }, { "epoch": 0.2519088708175554, "grad_norm": 1.057422623565375, "learning_rate": 8.396441133871302e-05, "loss": 3.9125, "step": 4058 }, { "epoch": 0.2519709479173133, "grad_norm": 1.0272658866123636, "learning_rate": 8.398510242085662e-05, "loss": 3.8445, "step": 4059 }, { "epoch": 0.2520330250170712, "grad_norm": 0.9491773368686717, "learning_rate": 8.400579350300021e-05, "loss": 3.8684, "step": 4060 }, { "epoch": 0.2520951021168291, "grad_norm": 1.0125675249541919, "learning_rate": 8.40264845851438e-05, "loss": 3.9273, "step": 4061 }, { "epoch": 0.252157179216587, "grad_norm": 0.910332494870357, "learning_rate": 8.40471756672874e-05, "loss": 4.0231, "step": 4062 }, { "epoch": 0.2522192563163449, "grad_norm": 0.9256070773947915, "learning_rate": 8.4067866749431e-05, "loss": 3.8589, "step": 4063 }, { "epoch": 0.2522813334161028, "grad_norm": 0.9236928828729369, "learning_rate": 8.40885578315746e-05, "loss": 3.9378, "step": 4064 }, { "epoch": 0.2523434105158607, "grad_norm": 0.8305768837063277, "learning_rate": 8.410924891371819e-05, "loss": 3.9955, "step": 4065 }, { "epoch": 0.2524054876156186, "grad_norm": 0.7869468502240254, "learning_rate": 8.412993999586178e-05, "loss": 3.8796, "step": 4066 }, { "epoch": 0.2524675647153765, "grad_norm": 0.7434468877840229, "learning_rate": 8.415063107800538e-05, "loss": 3.9811, "step": 4067 }, { "epoch": 0.2525296418151344, "grad_norm": 0.6690028924594383, "learning_rate": 8.417132216014899e-05, "loss": 3.8447, "step": 4068 }, { "epoch": 0.2525917189148923, "grad_norm": 0.6761799997494998, "learning_rate": 8.419201324229258e-05, "loss": 3.9083, "step": 4069 }, { "epoch": 0.2526537960146502, "grad_norm": 0.6642116379160794, "learning_rate": 8.421270432443618e-05, "loss": 4.1089, "step": 4070 }, { "epoch": 0.2527158731144081, "grad_norm": 0.682500717421214, "learning_rate": 8.423339540657977e-05, "loss": 3.9779, "step": 4071 }, { "epoch": 0.252777950214166, "grad_norm": 0.6643285106352135, "learning_rate": 8.425408648872337e-05, "loss": 3.8827, "step": 4072 }, { "epoch": 0.2528400273139239, "grad_norm": 0.7140486417464841, "learning_rate": 8.427477757086696e-05, "loss": 3.9165, "step": 4073 }, { "epoch": 0.25290210441368177, "grad_norm": 0.7261864704308967, "learning_rate": 8.429546865301056e-05, "loss": 4.0185, "step": 4074 }, { "epoch": 0.2529641815134397, "grad_norm": 0.7423044614485896, "learning_rate": 8.431615973515415e-05, "loss": 3.8983, "step": 4075 }, { "epoch": 0.2530262586131976, "grad_norm": 0.6787178533188006, "learning_rate": 8.433685081729775e-05, "loss": 3.9359, "step": 4076 }, { "epoch": 0.25308833571295547, "grad_norm": 0.5564510357215299, "learning_rate": 8.435754189944135e-05, "loss": 3.9625, "step": 4077 }, { "epoch": 0.2531504128127134, "grad_norm": 0.631424429371477, "learning_rate": 8.437823298158494e-05, "loss": 3.8695, "step": 4078 }, { "epoch": 0.2532124899124713, "grad_norm": 0.6289802886788457, "learning_rate": 8.439892406372853e-05, "loss": 3.8344, "step": 4079 }, { "epoch": 0.25327456701222917, "grad_norm": 0.7068727642330899, "learning_rate": 8.441961514587213e-05, "loss": 3.8587, "step": 4080 }, { "epoch": 0.2533366441119871, "grad_norm": 0.7811182292437847, "learning_rate": 8.444030622801573e-05, "loss": 3.9323, "step": 4081 }, { "epoch": 0.253398721211745, "grad_norm": 0.946864060694386, "learning_rate": 8.446099731015933e-05, "loss": 3.9737, "step": 4082 }, { "epoch": 0.25346079831150287, "grad_norm": 1.0103239954239285, "learning_rate": 8.448168839230293e-05, "loss": 3.8146, "step": 4083 }, { "epoch": 0.2535228754112608, "grad_norm": 0.9779746564670599, "learning_rate": 8.450237947444652e-05, "loss": 3.9131, "step": 4084 }, { "epoch": 0.2535849525110187, "grad_norm": 0.8154078680534158, "learning_rate": 8.452307055659012e-05, "loss": 3.9325, "step": 4085 }, { "epoch": 0.25364702961077656, "grad_norm": 0.9426515818562026, "learning_rate": 8.454376163873371e-05, "loss": 3.9283, "step": 4086 }, { "epoch": 0.2537091067105345, "grad_norm": 1.016360735332321, "learning_rate": 8.45644527208773e-05, "loss": 3.9193, "step": 4087 }, { "epoch": 0.2537711838102924, "grad_norm": 0.750605815353248, "learning_rate": 8.45851438030209e-05, "loss": 3.9174, "step": 4088 }, { "epoch": 0.25383326091005026, "grad_norm": 0.7373329814004087, "learning_rate": 8.460583488516451e-05, "loss": 3.9498, "step": 4089 }, { "epoch": 0.2538953380098082, "grad_norm": 0.7767735599324855, "learning_rate": 8.462652596730809e-05, "loss": 3.862, "step": 4090 }, { "epoch": 0.2539574151095661, "grad_norm": 0.8616179932165928, "learning_rate": 8.464721704945169e-05, "loss": 3.9338, "step": 4091 }, { "epoch": 0.25401949220932396, "grad_norm": 0.8630715715361279, "learning_rate": 8.466790813159528e-05, "loss": 3.9451, "step": 4092 }, { "epoch": 0.2540815693090819, "grad_norm": 0.7537923934427233, "learning_rate": 8.468859921373889e-05, "loss": 3.9675, "step": 4093 }, { "epoch": 0.2541436464088398, "grad_norm": 0.6952953979446324, "learning_rate": 8.470929029588248e-05, "loss": 3.8186, "step": 4094 }, { "epoch": 0.25420572350859766, "grad_norm": 0.7156899451891202, "learning_rate": 8.472998137802608e-05, "loss": 3.838, "step": 4095 }, { "epoch": 0.2542678006083556, "grad_norm": 0.7526278948535756, "learning_rate": 8.475067246016966e-05, "loss": 3.9076, "step": 4096 }, { "epoch": 0.2543298777081135, "grad_norm": 0.8637642868339943, "learning_rate": 8.477136354231327e-05, "loss": 3.9016, "step": 4097 }, { "epoch": 0.25439195480787136, "grad_norm": 0.8589854311101718, "learning_rate": 8.479205462445686e-05, "loss": 4.0582, "step": 4098 }, { "epoch": 0.2544540319076293, "grad_norm": 0.8262228803469002, "learning_rate": 8.481274570660046e-05, "loss": 3.8838, "step": 4099 }, { "epoch": 0.2545161090073872, "grad_norm": 0.7775563471697642, "learning_rate": 8.483343678874405e-05, "loss": 3.8607, "step": 4100 }, { "epoch": 0.25457818610714505, "grad_norm": 0.6886172969177463, "learning_rate": 8.485412787088765e-05, "loss": 3.9965, "step": 4101 }, { "epoch": 0.254640263206903, "grad_norm": 0.7940545001146799, "learning_rate": 8.487481895303125e-05, "loss": 3.863, "step": 4102 }, { "epoch": 0.2547023403066609, "grad_norm": 0.814982582912509, "learning_rate": 8.489551003517484e-05, "loss": 3.8872, "step": 4103 }, { "epoch": 0.25476441740641875, "grad_norm": 0.6841071244728537, "learning_rate": 8.491620111731844e-05, "loss": 4.0738, "step": 4104 }, { "epoch": 0.25482649450617667, "grad_norm": 0.8028155654723969, "learning_rate": 8.493689219946203e-05, "loss": 3.9225, "step": 4105 }, { "epoch": 0.2548885716059346, "grad_norm": 0.9006133228209453, "learning_rate": 8.495758328160564e-05, "loss": 3.8586, "step": 4106 }, { "epoch": 0.25495064870569245, "grad_norm": 0.978066042259696, "learning_rate": 8.497827436374923e-05, "loss": 3.9889, "step": 4107 }, { "epoch": 0.25501272580545037, "grad_norm": 0.9424768588718623, "learning_rate": 8.499896544589282e-05, "loss": 3.9172, "step": 4108 }, { "epoch": 0.2550748029052083, "grad_norm": 0.7341573676806046, "learning_rate": 8.501965652803642e-05, "loss": 3.8973, "step": 4109 }, { "epoch": 0.25513688000496615, "grad_norm": 0.6581372644091351, "learning_rate": 8.504034761018002e-05, "loss": 3.7979, "step": 4110 }, { "epoch": 0.25519895710472407, "grad_norm": 0.7033225311862958, "learning_rate": 8.506103869232361e-05, "loss": 3.7673, "step": 4111 }, { "epoch": 0.255261034204482, "grad_norm": 0.6552196770572003, "learning_rate": 8.508172977446721e-05, "loss": 3.9877, "step": 4112 }, { "epoch": 0.25532311130423985, "grad_norm": 0.7285332562560468, "learning_rate": 8.51024208566108e-05, "loss": 3.8806, "step": 4113 }, { "epoch": 0.25538518840399776, "grad_norm": 0.7330925477529162, "learning_rate": 8.51231119387544e-05, "loss": 3.8906, "step": 4114 }, { "epoch": 0.2554472655037557, "grad_norm": 0.6786833393880297, "learning_rate": 8.5143803020898e-05, "loss": 3.9678, "step": 4115 }, { "epoch": 0.25550934260351355, "grad_norm": 0.8200498978040655, "learning_rate": 8.516449410304159e-05, "loss": 3.9208, "step": 4116 }, { "epoch": 0.25557141970327146, "grad_norm": 0.758261400171666, "learning_rate": 8.518518518518518e-05, "loss": 3.9464, "step": 4117 }, { "epoch": 0.2556334968030294, "grad_norm": 0.6277090251482842, "learning_rate": 8.52058762673288e-05, "loss": 4.0318, "step": 4118 }, { "epoch": 0.25569557390278724, "grad_norm": 0.7222391522901113, "learning_rate": 8.522656734947239e-05, "loss": 3.9721, "step": 4119 }, { "epoch": 0.25575765100254516, "grad_norm": 0.7128619044541025, "learning_rate": 8.524725843161597e-05, "loss": 3.8571, "step": 4120 }, { "epoch": 0.2558197281023031, "grad_norm": 0.8277907506692834, "learning_rate": 8.526794951375957e-05, "loss": 3.9229, "step": 4121 }, { "epoch": 0.25588180520206094, "grad_norm": 0.7596895233717367, "learning_rate": 8.528864059590317e-05, "loss": 3.973, "step": 4122 }, { "epoch": 0.25594388230181886, "grad_norm": 0.6551402790971461, "learning_rate": 8.530933167804677e-05, "loss": 3.9238, "step": 4123 }, { "epoch": 0.2560059594015768, "grad_norm": 0.6043016321331597, "learning_rate": 8.533002276019036e-05, "loss": 3.7984, "step": 4124 }, { "epoch": 0.25606803650133464, "grad_norm": 0.7044916223281019, "learning_rate": 8.535071384233396e-05, "loss": 3.8368, "step": 4125 }, { "epoch": 0.25613011360109256, "grad_norm": 0.7317905400746132, "learning_rate": 8.537140492447755e-05, "loss": 3.829, "step": 4126 }, { "epoch": 0.2561921907008505, "grad_norm": 0.7925732006598734, "learning_rate": 8.539209600662115e-05, "loss": 3.932, "step": 4127 }, { "epoch": 0.25625426780060834, "grad_norm": 0.7141479691454455, "learning_rate": 8.541278708876474e-05, "loss": 3.8999, "step": 4128 }, { "epoch": 0.25631634490036626, "grad_norm": 0.6513219586044214, "learning_rate": 8.543347817090834e-05, "loss": 3.8663, "step": 4129 }, { "epoch": 0.2563784220001242, "grad_norm": 0.6940867946792029, "learning_rate": 8.545416925305195e-05, "loss": 3.8354, "step": 4130 }, { "epoch": 0.25644049909988204, "grad_norm": 0.7430998565167527, "learning_rate": 8.547486033519554e-05, "loss": 3.7752, "step": 4131 }, { "epoch": 0.25650257619963995, "grad_norm": 0.7289400884004559, "learning_rate": 8.549555141733912e-05, "loss": 3.8567, "step": 4132 }, { "epoch": 0.25656465329939787, "grad_norm": 0.7856759017077344, "learning_rate": 8.551624249948272e-05, "loss": 3.8895, "step": 4133 }, { "epoch": 0.25662673039915573, "grad_norm": 0.8792576744258785, "learning_rate": 8.553693358162633e-05, "loss": 3.9291, "step": 4134 }, { "epoch": 0.25668880749891365, "grad_norm": 0.9212253673888664, "learning_rate": 8.555762466376992e-05, "loss": 3.9395, "step": 4135 }, { "epoch": 0.25675088459867157, "grad_norm": 0.8538397690702162, "learning_rate": 8.557831574591352e-05, "loss": 3.9719, "step": 4136 }, { "epoch": 0.25681296169842943, "grad_norm": 0.9845141055199478, "learning_rate": 8.559900682805711e-05, "loss": 4.0149, "step": 4137 }, { "epoch": 0.25687503879818735, "grad_norm": 1.0915812107058458, "learning_rate": 8.561969791020071e-05, "loss": 3.9722, "step": 4138 }, { "epoch": 0.25693711589794527, "grad_norm": 0.826327370153405, "learning_rate": 8.56403889923443e-05, "loss": 3.9049, "step": 4139 }, { "epoch": 0.25699919299770313, "grad_norm": 0.7578630332222217, "learning_rate": 8.56610800744879e-05, "loss": 3.9246, "step": 4140 }, { "epoch": 0.25706127009746105, "grad_norm": 0.746223198810369, "learning_rate": 8.56817711566315e-05, "loss": 3.8776, "step": 4141 }, { "epoch": 0.25712334719721897, "grad_norm": 0.7590128250797605, "learning_rate": 8.570246223877509e-05, "loss": 3.8954, "step": 4142 }, { "epoch": 0.25718542429697683, "grad_norm": 0.6892329396629449, "learning_rate": 8.57231533209187e-05, "loss": 3.9336, "step": 4143 }, { "epoch": 0.25724750139673475, "grad_norm": 0.7262440465528811, "learning_rate": 8.574384440306228e-05, "loss": 3.8769, "step": 4144 }, { "epoch": 0.25730957849649266, "grad_norm": 0.7043007028045137, "learning_rate": 8.576453548520587e-05, "loss": 3.8733, "step": 4145 }, { "epoch": 0.2573716555962505, "grad_norm": 0.6982938734011692, "learning_rate": 8.578522656734947e-05, "loss": 3.9993, "step": 4146 }, { "epoch": 0.25743373269600844, "grad_norm": 0.7995447055324796, "learning_rate": 8.580591764949308e-05, "loss": 3.858, "step": 4147 }, { "epoch": 0.25749580979576636, "grad_norm": 0.8505902821268921, "learning_rate": 8.582660873163667e-05, "loss": 3.9452, "step": 4148 }, { "epoch": 0.2575578868955242, "grad_norm": 0.8044304298206955, "learning_rate": 8.584729981378027e-05, "loss": 3.8361, "step": 4149 }, { "epoch": 0.25761996399528214, "grad_norm": 0.8944445239008466, "learning_rate": 8.586799089592385e-05, "loss": 3.9629, "step": 4150 }, { "epoch": 0.25768204109504006, "grad_norm": 0.8377447823885933, "learning_rate": 8.588868197806746e-05, "loss": 3.8846, "step": 4151 }, { "epoch": 0.2577441181947979, "grad_norm": 0.8220213756102356, "learning_rate": 8.590937306021105e-05, "loss": 3.893, "step": 4152 }, { "epoch": 0.25780619529455584, "grad_norm": 0.9274202477543132, "learning_rate": 8.593006414235465e-05, "loss": 3.8256, "step": 4153 }, { "epoch": 0.25786827239431376, "grad_norm": 0.8715749873883786, "learning_rate": 8.595075522449824e-05, "loss": 3.9909, "step": 4154 }, { "epoch": 0.2579303494940716, "grad_norm": 0.9046954600983768, "learning_rate": 8.597144630664185e-05, "loss": 3.8605, "step": 4155 }, { "epoch": 0.25799242659382954, "grad_norm": 0.8666089851795732, "learning_rate": 8.599213738878545e-05, "loss": 3.8951, "step": 4156 }, { "epoch": 0.25805450369358746, "grad_norm": 0.8625767798042294, "learning_rate": 8.601282847092903e-05, "loss": 3.8943, "step": 4157 }, { "epoch": 0.2581165807933453, "grad_norm": 0.9146993651824413, "learning_rate": 8.603351955307262e-05, "loss": 3.8621, "step": 4158 }, { "epoch": 0.25817865789310324, "grad_norm": 0.8827739470849355, "learning_rate": 8.605421063521623e-05, "loss": 3.8749, "step": 4159 }, { "epoch": 0.25824073499286115, "grad_norm": 0.8395702892706888, "learning_rate": 8.607490171735983e-05, "loss": 3.8146, "step": 4160 }, { "epoch": 0.258302812092619, "grad_norm": 0.8594726853771051, "learning_rate": 8.609559279950342e-05, "loss": 3.9236, "step": 4161 }, { "epoch": 0.25836488919237693, "grad_norm": 0.8514487864313243, "learning_rate": 8.611628388164702e-05, "loss": 3.8073, "step": 4162 }, { "epoch": 0.25842696629213485, "grad_norm": 0.8076532026053701, "learning_rate": 8.613697496379061e-05, "loss": 3.8529, "step": 4163 }, { "epoch": 0.2584890433918927, "grad_norm": 0.8479659398139372, "learning_rate": 8.615766604593421e-05, "loss": 3.9159, "step": 4164 }, { "epoch": 0.25855112049165063, "grad_norm": 0.9418235265567436, "learning_rate": 8.61783571280778e-05, "loss": 4.0154, "step": 4165 }, { "epoch": 0.25861319759140855, "grad_norm": 0.8909851149269343, "learning_rate": 8.61990482102214e-05, "loss": 3.8846, "step": 4166 }, { "epoch": 0.2586752746911664, "grad_norm": 0.6983144869419756, "learning_rate": 8.621973929236499e-05, "loss": 3.8679, "step": 4167 }, { "epoch": 0.25873735179092433, "grad_norm": 0.915209287201086, "learning_rate": 8.62404303745086e-05, "loss": 3.82, "step": 4168 }, { "epoch": 0.25879942889068225, "grad_norm": 0.8502622354126776, "learning_rate": 8.626112145665218e-05, "loss": 3.932, "step": 4169 }, { "epoch": 0.2588615059904401, "grad_norm": 0.8683460847195361, "learning_rate": 8.628181253879578e-05, "loss": 3.8445, "step": 4170 }, { "epoch": 0.25892358309019803, "grad_norm": 0.7491223047450818, "learning_rate": 8.630250362093937e-05, "loss": 3.8658, "step": 4171 }, { "epoch": 0.25898566018995595, "grad_norm": 0.7083478019209035, "learning_rate": 8.632319470308298e-05, "loss": 3.8694, "step": 4172 }, { "epoch": 0.2590477372897138, "grad_norm": 0.7493412383096415, "learning_rate": 8.634388578522658e-05, "loss": 3.9213, "step": 4173 }, { "epoch": 0.2591098143894717, "grad_norm": 0.7665551821533949, "learning_rate": 8.636457686737017e-05, "loss": 3.9492, "step": 4174 }, { "epoch": 0.25917189148922964, "grad_norm": 0.7625126931431615, "learning_rate": 8.638526794951375e-05, "loss": 3.817, "step": 4175 }, { "epoch": 0.2592339685889875, "grad_norm": 0.7991585269734885, "learning_rate": 8.640595903165736e-05, "loss": 3.9527, "step": 4176 }, { "epoch": 0.2592960456887454, "grad_norm": 0.665794380558062, "learning_rate": 8.642665011380096e-05, "loss": 3.9106, "step": 4177 }, { "epoch": 0.25935812278850334, "grad_norm": 0.7567026722389607, "learning_rate": 8.644734119594455e-05, "loss": 3.7982, "step": 4178 }, { "epoch": 0.2594201998882612, "grad_norm": 0.8612800290612675, "learning_rate": 8.646803227808815e-05, "loss": 3.8093, "step": 4179 }, { "epoch": 0.2594822769880191, "grad_norm": 0.9564325756730708, "learning_rate": 8.648872336023176e-05, "loss": 3.8131, "step": 4180 }, { "epoch": 0.25954435408777704, "grad_norm": 0.7622040987416562, "learning_rate": 8.650941444237534e-05, "loss": 3.9219, "step": 4181 }, { "epoch": 0.2596064311875349, "grad_norm": 0.6480121596352266, "learning_rate": 8.653010552451893e-05, "loss": 3.876, "step": 4182 }, { "epoch": 0.2596685082872928, "grad_norm": 0.6846513251285464, "learning_rate": 8.655079660666253e-05, "loss": 3.9492, "step": 4183 }, { "epoch": 0.25973058538705074, "grad_norm": 0.6752670404472463, "learning_rate": 8.657148768880614e-05, "loss": 3.9635, "step": 4184 }, { "epoch": 0.2597926624868086, "grad_norm": 0.7271292742085002, "learning_rate": 8.659217877094973e-05, "loss": 3.9257, "step": 4185 }, { "epoch": 0.2598547395865665, "grad_norm": 0.6334705933939276, "learning_rate": 8.661286985309333e-05, "loss": 3.9262, "step": 4186 }, { "epoch": 0.25991681668632444, "grad_norm": 0.7032421226439987, "learning_rate": 8.663356093523691e-05, "loss": 4.0046, "step": 4187 }, { "epoch": 0.2599788937860823, "grad_norm": 0.792960030644026, "learning_rate": 8.665425201738052e-05, "loss": 3.9089, "step": 4188 }, { "epoch": 0.2600409708858402, "grad_norm": 0.7277765692837544, "learning_rate": 8.667494309952411e-05, "loss": 3.8857, "step": 4189 }, { "epoch": 0.26010304798559813, "grad_norm": 0.7602954968277075, "learning_rate": 8.66956341816677e-05, "loss": 3.9504, "step": 4190 }, { "epoch": 0.260165125085356, "grad_norm": 0.8633645824413058, "learning_rate": 8.67163252638113e-05, "loss": 3.8777, "step": 4191 }, { "epoch": 0.2602272021851139, "grad_norm": 0.7952567854816354, "learning_rate": 8.67370163459549e-05, "loss": 3.8418, "step": 4192 }, { "epoch": 0.26028927928487183, "grad_norm": 0.7791985853732621, "learning_rate": 8.675770742809849e-05, "loss": 3.8856, "step": 4193 }, { "epoch": 0.2603513563846297, "grad_norm": 0.8051001592486936, "learning_rate": 8.677839851024209e-05, "loss": 3.9609, "step": 4194 }, { "epoch": 0.2604134334843876, "grad_norm": 0.7306747961900171, "learning_rate": 8.679908959238568e-05, "loss": 3.8525, "step": 4195 }, { "epoch": 0.26047551058414553, "grad_norm": 0.7147926770919052, "learning_rate": 8.681978067452928e-05, "loss": 3.8762, "step": 4196 }, { "epoch": 0.2605375876839034, "grad_norm": 0.7108693771421652, "learning_rate": 8.684047175667289e-05, "loss": 3.9037, "step": 4197 }, { "epoch": 0.2605996647836613, "grad_norm": 0.7386307872881838, "learning_rate": 8.686116283881648e-05, "loss": 3.9375, "step": 4198 }, { "epoch": 0.26066174188341923, "grad_norm": 0.7297161938010444, "learning_rate": 8.688185392096006e-05, "loss": 3.9227, "step": 4199 }, { "epoch": 0.2607238189831771, "grad_norm": 0.5931323754240952, "learning_rate": 8.690254500310366e-05, "loss": 3.806, "step": 4200 }, { "epoch": 0.260785896082935, "grad_norm": 0.6616374897068972, "learning_rate": 8.692323608524727e-05, "loss": 3.8405, "step": 4201 }, { "epoch": 0.2608479731826929, "grad_norm": 0.7016047254739256, "learning_rate": 8.694392716739086e-05, "loss": 3.8467, "step": 4202 }, { "epoch": 0.2609100502824508, "grad_norm": 0.6902619067639675, "learning_rate": 8.696461824953446e-05, "loss": 3.7873, "step": 4203 }, { "epoch": 0.2609721273822087, "grad_norm": 0.7249133765331656, "learning_rate": 8.698530933167805e-05, "loss": 3.9052, "step": 4204 }, { "epoch": 0.2610342044819666, "grad_norm": 0.674894092339917, "learning_rate": 8.700600041382165e-05, "loss": 3.8926, "step": 4205 }, { "epoch": 0.2610962815817245, "grad_norm": 0.7077987046402641, "learning_rate": 8.702669149596524e-05, "loss": 3.8843, "step": 4206 }, { "epoch": 0.2611583586814824, "grad_norm": 0.5763369824759306, "learning_rate": 8.704738257810884e-05, "loss": 3.8843, "step": 4207 }, { "epoch": 0.2612204357812403, "grad_norm": 0.6454271043886116, "learning_rate": 8.706807366025243e-05, "loss": 3.8724, "step": 4208 }, { "epoch": 0.2612825128809982, "grad_norm": 0.6139814910646201, "learning_rate": 8.708876474239604e-05, "loss": 3.8029, "step": 4209 }, { "epoch": 0.2613445899807561, "grad_norm": 0.6084516572407292, "learning_rate": 8.710945582453964e-05, "loss": 3.8178, "step": 4210 }, { "epoch": 0.261406667080514, "grad_norm": 0.6109899249596313, "learning_rate": 8.713014690668322e-05, "loss": 3.7973, "step": 4211 }, { "epoch": 0.2614687441802719, "grad_norm": 0.6034453514690665, "learning_rate": 8.715083798882681e-05, "loss": 4.0515, "step": 4212 }, { "epoch": 0.2615308212800298, "grad_norm": 0.5939810203047093, "learning_rate": 8.717152907097042e-05, "loss": 3.8896, "step": 4213 }, { "epoch": 0.2615928983797877, "grad_norm": 0.7118514756894971, "learning_rate": 8.719222015311402e-05, "loss": 3.7794, "step": 4214 }, { "epoch": 0.2616549754795456, "grad_norm": 0.681616424743057, "learning_rate": 8.721291123525761e-05, "loss": 3.9284, "step": 4215 }, { "epoch": 0.2617170525793035, "grad_norm": 0.8610450123641087, "learning_rate": 8.72336023174012e-05, "loss": 3.8591, "step": 4216 }, { "epoch": 0.2617791296790614, "grad_norm": 0.891921889389768, "learning_rate": 8.72542933995448e-05, "loss": 3.9205, "step": 4217 }, { "epoch": 0.2618412067788193, "grad_norm": 0.845051362321935, "learning_rate": 8.72749844816884e-05, "loss": 3.8954, "step": 4218 }, { "epoch": 0.2619032838785772, "grad_norm": 0.7474369919417373, "learning_rate": 8.729567556383199e-05, "loss": 3.8489, "step": 4219 }, { "epoch": 0.2619653609783351, "grad_norm": 0.8618044412014669, "learning_rate": 8.731636664597559e-05, "loss": 3.981, "step": 4220 }, { "epoch": 0.262027438078093, "grad_norm": 0.9634760069599015, "learning_rate": 8.733705772811918e-05, "loss": 3.9219, "step": 4221 }, { "epoch": 0.2620895151778509, "grad_norm": 1.0179764156682367, "learning_rate": 8.735774881026279e-05, "loss": 3.9325, "step": 4222 }, { "epoch": 0.2621515922776088, "grad_norm": 0.7735449205804946, "learning_rate": 8.737843989240637e-05, "loss": 3.8542, "step": 4223 }, { "epoch": 0.2622136693773667, "grad_norm": 0.8640258477768624, "learning_rate": 8.739913097454997e-05, "loss": 3.9049, "step": 4224 }, { "epoch": 0.2622757464771246, "grad_norm": 0.8484503195340357, "learning_rate": 8.741982205669356e-05, "loss": 3.9824, "step": 4225 }, { "epoch": 0.2623378235768825, "grad_norm": 0.78608053282346, "learning_rate": 8.744051313883717e-05, "loss": 3.8716, "step": 4226 }, { "epoch": 0.2623999006766404, "grad_norm": 0.8750413135584585, "learning_rate": 8.746120422098077e-05, "loss": 3.9693, "step": 4227 }, { "epoch": 0.2624619777763983, "grad_norm": 0.856107777945155, "learning_rate": 8.748189530312436e-05, "loss": 3.9362, "step": 4228 }, { "epoch": 0.2625240548761562, "grad_norm": 0.7837343768641891, "learning_rate": 8.750258638526794e-05, "loss": 3.9218, "step": 4229 }, { "epoch": 0.26258613197591407, "grad_norm": 0.8468263844404716, "learning_rate": 8.752327746741155e-05, "loss": 3.8658, "step": 4230 }, { "epoch": 0.262648209075672, "grad_norm": 0.8368877108508167, "learning_rate": 8.754396854955515e-05, "loss": 3.8516, "step": 4231 }, { "epoch": 0.2627102861754299, "grad_norm": 0.6417958794166426, "learning_rate": 8.756465963169874e-05, "loss": 3.8322, "step": 4232 }, { "epoch": 0.26277236327518777, "grad_norm": 0.6710824949943864, "learning_rate": 8.758535071384234e-05, "loss": 3.8668, "step": 4233 }, { "epoch": 0.2628344403749457, "grad_norm": 0.7806826247851809, "learning_rate": 8.760604179598594e-05, "loss": 3.835, "step": 4234 }, { "epoch": 0.2628965174747036, "grad_norm": 0.6526880326590794, "learning_rate": 8.762673287812953e-05, "loss": 3.8769, "step": 4235 }, { "epoch": 0.26295859457446147, "grad_norm": 0.6867744345107442, "learning_rate": 8.764742396027312e-05, "loss": 3.83, "step": 4236 }, { "epoch": 0.2630206716742194, "grad_norm": 0.6680513074395703, "learning_rate": 8.766811504241672e-05, "loss": 3.7805, "step": 4237 }, { "epoch": 0.2630827487739773, "grad_norm": 0.6839104273117044, "learning_rate": 8.768880612456032e-05, "loss": 3.8777, "step": 4238 }, { "epoch": 0.26314482587373517, "grad_norm": 0.6760291378073061, "learning_rate": 8.770949720670392e-05, "loss": 3.9089, "step": 4239 }, { "epoch": 0.2632069029734931, "grad_norm": 0.7571017678311334, "learning_rate": 8.773018828884751e-05, "loss": 3.9213, "step": 4240 }, { "epoch": 0.263268980073251, "grad_norm": 0.6381602906902952, "learning_rate": 8.775087937099111e-05, "loss": 3.9478, "step": 4241 }, { "epoch": 0.26333105717300886, "grad_norm": 0.5902467323691025, "learning_rate": 8.77715704531347e-05, "loss": 3.813, "step": 4242 }, { "epoch": 0.2633931342727668, "grad_norm": 0.7745695210648599, "learning_rate": 8.77922615352783e-05, "loss": 3.8321, "step": 4243 }, { "epoch": 0.2634552113725247, "grad_norm": 0.9289855871817422, "learning_rate": 8.78129526174219e-05, "loss": 3.8234, "step": 4244 }, { "epoch": 0.26351728847228256, "grad_norm": 0.9249135711603915, "learning_rate": 8.783364369956549e-05, "loss": 3.949, "step": 4245 }, { "epoch": 0.2635793655720405, "grad_norm": 0.7602176057335964, "learning_rate": 8.785433478170909e-05, "loss": 3.9954, "step": 4246 }, { "epoch": 0.2636414426717984, "grad_norm": 0.8924041845676659, "learning_rate": 8.78750258638527e-05, "loss": 3.8819, "step": 4247 }, { "epoch": 0.26370351977155626, "grad_norm": 1.0391028833787797, "learning_rate": 8.789571694599628e-05, "loss": 3.8816, "step": 4248 }, { "epoch": 0.2637655968713142, "grad_norm": 0.7807445416878078, "learning_rate": 8.791640802813987e-05, "loss": 3.9022, "step": 4249 }, { "epoch": 0.2638276739710721, "grad_norm": 0.6996854577795009, "learning_rate": 8.793709911028347e-05, "loss": 3.8969, "step": 4250 }, { "epoch": 0.26388975107082996, "grad_norm": 0.8895304127362075, "learning_rate": 8.795779019242707e-05, "loss": 3.8919, "step": 4251 }, { "epoch": 0.2639518281705879, "grad_norm": 0.8240630016245815, "learning_rate": 8.797848127457067e-05, "loss": 3.8578, "step": 4252 }, { "epoch": 0.2640139052703458, "grad_norm": 0.7600361555684712, "learning_rate": 8.799917235671426e-05, "loss": 3.8544, "step": 4253 }, { "epoch": 0.26407598237010366, "grad_norm": 0.9080321120179843, "learning_rate": 8.801986343885785e-05, "loss": 3.9321, "step": 4254 }, { "epoch": 0.2641380594698616, "grad_norm": 0.8691880116778361, "learning_rate": 8.804055452100145e-05, "loss": 3.6985, "step": 4255 }, { "epoch": 0.2642001365696195, "grad_norm": 0.7452495252447142, "learning_rate": 8.806124560314505e-05, "loss": 3.8434, "step": 4256 }, { "epoch": 0.26426221366937735, "grad_norm": 0.7147738730041963, "learning_rate": 8.808193668528864e-05, "loss": 3.747, "step": 4257 }, { "epoch": 0.2643242907691353, "grad_norm": 0.6231876082806181, "learning_rate": 8.810262776743224e-05, "loss": 3.9612, "step": 4258 }, { "epoch": 0.2643863678688932, "grad_norm": 0.8403763257886763, "learning_rate": 8.812331884957585e-05, "loss": 3.7346, "step": 4259 }, { "epoch": 0.26444844496865105, "grad_norm": 0.800347969255262, "learning_rate": 8.814400993171943e-05, "loss": 3.8721, "step": 4260 }, { "epoch": 0.26451052206840897, "grad_norm": 0.6848137350639618, "learning_rate": 8.816470101386302e-05, "loss": 3.975, "step": 4261 }, { "epoch": 0.2645725991681669, "grad_norm": 0.7125962577179815, "learning_rate": 8.818539209600662e-05, "loss": 3.8952, "step": 4262 }, { "epoch": 0.26463467626792475, "grad_norm": 0.7165390091317523, "learning_rate": 8.820608317815023e-05, "loss": 3.8767, "step": 4263 }, { "epoch": 0.26469675336768267, "grad_norm": 0.725789469004105, "learning_rate": 8.822677426029382e-05, "loss": 3.8054, "step": 4264 }, { "epoch": 0.2647588304674406, "grad_norm": 0.7196255202776055, "learning_rate": 8.824746534243742e-05, "loss": 3.8697, "step": 4265 }, { "epoch": 0.26482090756719845, "grad_norm": 0.7150197047072615, "learning_rate": 8.8268156424581e-05, "loss": 3.898, "step": 4266 }, { "epoch": 0.26488298466695637, "grad_norm": 0.691086711429405, "learning_rate": 8.828884750672461e-05, "loss": 3.7304, "step": 4267 }, { "epoch": 0.2649450617667143, "grad_norm": 0.7306059030300107, "learning_rate": 8.83095385888682e-05, "loss": 3.964, "step": 4268 }, { "epoch": 0.26500713886647215, "grad_norm": 0.6972466687956872, "learning_rate": 8.83302296710118e-05, "loss": 3.9293, "step": 4269 }, { "epoch": 0.26506921596623007, "grad_norm": 0.7795431565357777, "learning_rate": 8.83509207531554e-05, "loss": 3.8789, "step": 4270 }, { "epoch": 0.265131293065988, "grad_norm": 0.9425120843168282, "learning_rate": 8.837161183529899e-05, "loss": 3.8787, "step": 4271 }, { "epoch": 0.26519337016574585, "grad_norm": 0.9515224805412333, "learning_rate": 8.839230291744258e-05, "loss": 3.8185, "step": 4272 }, { "epoch": 0.26525544726550376, "grad_norm": 0.7895360602030935, "learning_rate": 8.841299399958618e-05, "loss": 4.0436, "step": 4273 }, { "epoch": 0.2653175243652617, "grad_norm": 0.7677604711073797, "learning_rate": 8.843368508172977e-05, "loss": 3.809, "step": 4274 }, { "epoch": 0.26537960146501954, "grad_norm": 0.8255373274221374, "learning_rate": 8.845437616387337e-05, "loss": 3.8993, "step": 4275 }, { "epoch": 0.26544167856477746, "grad_norm": 0.8353760875128015, "learning_rate": 8.847506724601698e-05, "loss": 3.8374, "step": 4276 }, { "epoch": 0.2655037556645354, "grad_norm": 0.741685918672412, "learning_rate": 8.849575832816057e-05, "loss": 3.8023, "step": 4277 }, { "epoch": 0.26556583276429324, "grad_norm": 0.7902933473027307, "learning_rate": 8.851644941030415e-05, "loss": 3.7543, "step": 4278 }, { "epoch": 0.26562790986405116, "grad_norm": 0.8975083895851687, "learning_rate": 8.853714049244776e-05, "loss": 3.8, "step": 4279 }, { "epoch": 0.2656899869638091, "grad_norm": 0.8400802634130475, "learning_rate": 8.855783157459136e-05, "loss": 3.9668, "step": 4280 }, { "epoch": 0.26575206406356694, "grad_norm": 0.7191251936484092, "learning_rate": 8.857852265673495e-05, "loss": 3.9026, "step": 4281 }, { "epoch": 0.26581414116332486, "grad_norm": 0.7611600342880555, "learning_rate": 8.859921373887855e-05, "loss": 3.9188, "step": 4282 }, { "epoch": 0.2658762182630828, "grad_norm": 0.881583379358267, "learning_rate": 8.861990482102214e-05, "loss": 3.7805, "step": 4283 }, { "epoch": 0.26593829536284064, "grad_norm": 0.8804014923213043, "learning_rate": 8.864059590316574e-05, "loss": 3.9287, "step": 4284 }, { "epoch": 0.26600037246259856, "grad_norm": 0.7154123931635674, "learning_rate": 8.866128698530933e-05, "loss": 3.927, "step": 4285 }, { "epoch": 0.2660624495623565, "grad_norm": 0.7816686652862223, "learning_rate": 8.868197806745293e-05, "loss": 3.8276, "step": 4286 }, { "epoch": 0.26612452666211434, "grad_norm": 0.8095137296811759, "learning_rate": 8.870266914959652e-05, "loss": 3.8567, "step": 4287 }, { "epoch": 0.26618660376187225, "grad_norm": 0.8295030464268182, "learning_rate": 8.872336023174013e-05, "loss": 3.8314, "step": 4288 }, { "epoch": 0.26624868086163017, "grad_norm": 0.7286118349256018, "learning_rate": 8.874405131388373e-05, "loss": 3.8934, "step": 4289 }, { "epoch": 0.26631075796138803, "grad_norm": 0.6842644153591508, "learning_rate": 8.876474239602731e-05, "loss": 3.8094, "step": 4290 }, { "epoch": 0.26637283506114595, "grad_norm": 0.7328223595840464, "learning_rate": 8.87854334781709e-05, "loss": 3.8761, "step": 4291 }, { "epoch": 0.26643491216090387, "grad_norm": 0.6022159445774836, "learning_rate": 8.880612456031451e-05, "loss": 3.8588, "step": 4292 }, { "epoch": 0.26649698926066173, "grad_norm": 0.6731222906385776, "learning_rate": 8.882681564245811e-05, "loss": 3.8595, "step": 4293 }, { "epoch": 0.26655906636041965, "grad_norm": 0.6884758482584393, "learning_rate": 8.88475067246017e-05, "loss": 3.8428, "step": 4294 }, { "epoch": 0.26662114346017757, "grad_norm": 0.6243520954730815, "learning_rate": 8.88681978067453e-05, "loss": 3.7766, "step": 4295 }, { "epoch": 0.26668322055993543, "grad_norm": 0.579958071631905, "learning_rate": 8.888888888888889e-05, "loss": 3.8752, "step": 4296 }, { "epoch": 0.26674529765969335, "grad_norm": 0.6719136781426943, "learning_rate": 8.890957997103249e-05, "loss": 3.8602, "step": 4297 }, { "epoch": 0.26680737475945127, "grad_norm": 0.7034332659274927, "learning_rate": 8.893027105317608e-05, "loss": 3.8108, "step": 4298 }, { "epoch": 0.26686945185920913, "grad_norm": 0.8398299814914264, "learning_rate": 8.895096213531968e-05, "loss": 3.8616, "step": 4299 }, { "epoch": 0.26693152895896705, "grad_norm": 0.8615749636894564, "learning_rate": 8.897165321746327e-05, "loss": 3.9686, "step": 4300 }, { "epoch": 0.26699360605872496, "grad_norm": 0.7268012280145704, "learning_rate": 8.899234429960688e-05, "loss": 3.9324, "step": 4301 }, { "epoch": 0.2670556831584828, "grad_norm": 0.7843443055245347, "learning_rate": 8.901303538175046e-05, "loss": 3.9255, "step": 4302 }, { "epoch": 0.26711776025824074, "grad_norm": 0.8645290832657846, "learning_rate": 8.903372646389406e-05, "loss": 3.9525, "step": 4303 }, { "epoch": 0.26717983735799866, "grad_norm": 0.8110992387950751, "learning_rate": 8.905441754603767e-05, "loss": 3.8616, "step": 4304 }, { "epoch": 0.2672419144577565, "grad_norm": 0.7173417850002994, "learning_rate": 8.907510862818126e-05, "loss": 3.8073, "step": 4305 }, { "epoch": 0.26730399155751444, "grad_norm": 0.6949858165912896, "learning_rate": 8.909579971032486e-05, "loss": 3.9358, "step": 4306 }, { "epoch": 0.26736606865727236, "grad_norm": 0.6605907567730266, "learning_rate": 8.911649079246845e-05, "loss": 3.876, "step": 4307 }, { "epoch": 0.2674281457570302, "grad_norm": 0.7975574264060079, "learning_rate": 8.913718187461205e-05, "loss": 3.8335, "step": 4308 }, { "epoch": 0.26749022285678814, "grad_norm": 0.7350802991431307, "learning_rate": 8.915787295675564e-05, "loss": 3.9367, "step": 4309 }, { "epoch": 0.267552299956546, "grad_norm": 0.7256204383831572, "learning_rate": 8.917856403889924e-05, "loss": 3.7723, "step": 4310 }, { "epoch": 0.2676143770563039, "grad_norm": 0.6473140175869655, "learning_rate": 8.919925512104283e-05, "loss": 3.8095, "step": 4311 }, { "epoch": 0.26767645415606184, "grad_norm": 0.7140752277867652, "learning_rate": 8.921994620318643e-05, "loss": 3.9258, "step": 4312 }, { "epoch": 0.2677385312558197, "grad_norm": 0.7525832669393221, "learning_rate": 8.924063728533004e-05, "loss": 3.8163, "step": 4313 }, { "epoch": 0.2678006083555776, "grad_norm": 0.7486590079549622, "learning_rate": 8.926132836747362e-05, "loss": 3.7768, "step": 4314 }, { "epoch": 0.26786268545533554, "grad_norm": 0.7433737660717381, "learning_rate": 8.928201944961721e-05, "loss": 3.8586, "step": 4315 }, { "epoch": 0.2679247625550934, "grad_norm": 0.7960843296261089, "learning_rate": 8.930271053176081e-05, "loss": 3.8325, "step": 4316 }, { "epoch": 0.2679868396548513, "grad_norm": 0.7696722066136629, "learning_rate": 8.932340161390442e-05, "loss": 3.8775, "step": 4317 }, { "epoch": 0.26804891675460923, "grad_norm": 0.7468673131060366, "learning_rate": 8.934409269604801e-05, "loss": 3.8811, "step": 4318 }, { "epoch": 0.2681109938543671, "grad_norm": 0.7557176166250348, "learning_rate": 8.936478377819161e-05, "loss": 3.8857, "step": 4319 }, { "epoch": 0.268173070954125, "grad_norm": 0.7808982820345173, "learning_rate": 8.938547486033519e-05, "loss": 3.8416, "step": 4320 }, { "epoch": 0.26823514805388293, "grad_norm": 0.6889470217729476, "learning_rate": 8.94061659424788e-05, "loss": 3.7955, "step": 4321 }, { "epoch": 0.2682972251536408, "grad_norm": 0.6678073308777671, "learning_rate": 8.942685702462239e-05, "loss": 3.8585, "step": 4322 }, { "epoch": 0.2683593022533987, "grad_norm": 0.7497166257211797, "learning_rate": 8.944754810676599e-05, "loss": 3.8704, "step": 4323 }, { "epoch": 0.26842137935315663, "grad_norm": 0.6813055269552708, "learning_rate": 8.946823918890958e-05, "loss": 3.732, "step": 4324 }, { "epoch": 0.2684834564529145, "grad_norm": 0.6231006371911307, "learning_rate": 8.948893027105318e-05, "loss": 3.8738, "step": 4325 }, { "epoch": 0.2685455335526724, "grad_norm": 0.6957318830677441, "learning_rate": 8.950962135319679e-05, "loss": 3.9058, "step": 4326 }, { "epoch": 0.26860761065243033, "grad_norm": 0.738103814979827, "learning_rate": 8.953031243534037e-05, "loss": 3.8559, "step": 4327 }, { "epoch": 0.2686696877521882, "grad_norm": 0.8451231626330691, "learning_rate": 8.955100351748396e-05, "loss": 3.8541, "step": 4328 }, { "epoch": 0.2687317648519461, "grad_norm": 0.8741958192437783, "learning_rate": 8.957169459962757e-05, "loss": 3.8975, "step": 4329 }, { "epoch": 0.268793841951704, "grad_norm": 0.8729395106990214, "learning_rate": 8.959238568177117e-05, "loss": 3.961, "step": 4330 }, { "epoch": 0.2688559190514619, "grad_norm": 0.7919256179998394, "learning_rate": 8.961307676391476e-05, "loss": 3.7885, "step": 4331 }, { "epoch": 0.2689179961512198, "grad_norm": 0.6757021510407953, "learning_rate": 8.963376784605836e-05, "loss": 3.7979, "step": 4332 }, { "epoch": 0.2689800732509777, "grad_norm": 0.6469003543399908, "learning_rate": 8.965445892820195e-05, "loss": 3.9156, "step": 4333 }, { "epoch": 0.2690421503507356, "grad_norm": 0.7097735255992587, "learning_rate": 8.967515001034555e-05, "loss": 3.9397, "step": 4334 }, { "epoch": 0.2691042274504935, "grad_norm": 0.850283966595116, "learning_rate": 8.969584109248914e-05, "loss": 3.7366, "step": 4335 }, { "epoch": 0.2691663045502514, "grad_norm": 0.7224776588602482, "learning_rate": 8.971653217463274e-05, "loss": 3.8145, "step": 4336 }, { "epoch": 0.2692283816500093, "grad_norm": 0.7052776234073912, "learning_rate": 8.973722325677633e-05, "loss": 3.875, "step": 4337 }, { "epoch": 0.2692904587497672, "grad_norm": 0.6704891238898606, "learning_rate": 8.975791433891994e-05, "loss": 3.8957, "step": 4338 }, { "epoch": 0.2693525358495251, "grad_norm": 0.6414403378980374, "learning_rate": 8.977860542106352e-05, "loss": 3.8629, "step": 4339 }, { "epoch": 0.269414612949283, "grad_norm": 0.7926865249287317, "learning_rate": 8.979929650320712e-05, "loss": 3.8025, "step": 4340 }, { "epoch": 0.2694766900490409, "grad_norm": 0.7348562652151579, "learning_rate": 8.981998758535071e-05, "loss": 3.8308, "step": 4341 }, { "epoch": 0.2695387671487988, "grad_norm": 0.6415780816506339, "learning_rate": 8.984067866749432e-05, "loss": 3.8265, "step": 4342 }, { "epoch": 0.2696008442485567, "grad_norm": 0.7323313339219416, "learning_rate": 8.986136974963792e-05, "loss": 3.8875, "step": 4343 }, { "epoch": 0.2696629213483146, "grad_norm": 0.8195145858946872, "learning_rate": 8.988206083178151e-05, "loss": 3.8576, "step": 4344 }, { "epoch": 0.2697249984480725, "grad_norm": 0.7488231565243972, "learning_rate": 8.990275191392509e-05, "loss": 3.8028, "step": 4345 }, { "epoch": 0.2697870755478304, "grad_norm": 0.6650633282372626, "learning_rate": 8.99234429960687e-05, "loss": 3.8581, "step": 4346 }, { "epoch": 0.2698491526475883, "grad_norm": 0.7907086188686421, "learning_rate": 8.99441340782123e-05, "loss": 3.8364, "step": 4347 }, { "epoch": 0.2699112297473462, "grad_norm": 0.7709285617120021, "learning_rate": 8.996482516035589e-05, "loss": 3.7809, "step": 4348 }, { "epoch": 0.2699733068471041, "grad_norm": 0.6782957934279368, "learning_rate": 8.998551624249949e-05, "loss": 3.9523, "step": 4349 }, { "epoch": 0.270035383946862, "grad_norm": 0.5992818428653669, "learning_rate": 9.000620732464308e-05, "loss": 3.8111, "step": 4350 }, { "epoch": 0.2700974610466199, "grad_norm": 0.5863150842323196, "learning_rate": 9.002689840678668e-05, "loss": 3.7881, "step": 4351 }, { "epoch": 0.2701595381463778, "grad_norm": 0.7644789935507402, "learning_rate": 9.004758948893027e-05, "loss": 3.8694, "step": 4352 }, { "epoch": 0.2702216152461357, "grad_norm": 0.876751237157553, "learning_rate": 9.006828057107387e-05, "loss": 3.7558, "step": 4353 }, { "epoch": 0.2702836923458936, "grad_norm": 0.8149377126821443, "learning_rate": 9.008897165321748e-05, "loss": 3.9293, "step": 4354 }, { "epoch": 0.2703457694456515, "grad_norm": 0.7801839039487504, "learning_rate": 9.010966273536107e-05, "loss": 3.9359, "step": 4355 }, { "epoch": 0.2704078465454094, "grad_norm": 0.7899968120655491, "learning_rate": 9.013035381750467e-05, "loss": 3.8114, "step": 4356 }, { "epoch": 0.2704699236451673, "grad_norm": 0.7820644329702299, "learning_rate": 9.015104489964825e-05, "loss": 3.8346, "step": 4357 }, { "epoch": 0.27053200074492517, "grad_norm": 0.8823701881378003, "learning_rate": 9.017173598179186e-05, "loss": 3.8845, "step": 4358 }, { "epoch": 0.2705940778446831, "grad_norm": 0.7863001674906622, "learning_rate": 9.019242706393545e-05, "loss": 3.8651, "step": 4359 }, { "epoch": 0.270656154944441, "grad_norm": 0.8348578138414485, "learning_rate": 9.021311814607905e-05, "loss": 3.8445, "step": 4360 }, { "epoch": 0.27071823204419887, "grad_norm": 0.8143089404653805, "learning_rate": 9.023380922822264e-05, "loss": 3.8135, "step": 4361 }, { "epoch": 0.2707803091439568, "grad_norm": 0.7377526949500743, "learning_rate": 9.025450031036624e-05, "loss": 3.8213, "step": 4362 }, { "epoch": 0.2708423862437147, "grad_norm": 0.7725983122576855, "learning_rate": 9.027519139250983e-05, "loss": 3.8796, "step": 4363 }, { "epoch": 0.27090446334347257, "grad_norm": 0.8204051737434814, "learning_rate": 9.029588247465343e-05, "loss": 3.7939, "step": 4364 }, { "epoch": 0.2709665404432305, "grad_norm": 0.7232706705614709, "learning_rate": 9.031657355679702e-05, "loss": 3.8446, "step": 4365 }, { "epoch": 0.2710286175429884, "grad_norm": 0.8272188375119262, "learning_rate": 9.033726463894062e-05, "loss": 3.9235, "step": 4366 }, { "epoch": 0.27109069464274627, "grad_norm": 0.9291446286882374, "learning_rate": 9.035795572108422e-05, "loss": 3.8535, "step": 4367 }, { "epoch": 0.2711527717425042, "grad_norm": 0.8964566494512705, "learning_rate": 9.037864680322782e-05, "loss": 3.832, "step": 4368 }, { "epoch": 0.2712148488422621, "grad_norm": 0.7690519592728449, "learning_rate": 9.03993378853714e-05, "loss": 3.79, "step": 4369 }, { "epoch": 0.27127692594201996, "grad_norm": 0.7271123466787563, "learning_rate": 9.0420028967515e-05, "loss": 3.9114, "step": 4370 }, { "epoch": 0.2713390030417779, "grad_norm": 0.8061055401455398, "learning_rate": 9.04407200496586e-05, "loss": 3.8659, "step": 4371 }, { "epoch": 0.2714010801415358, "grad_norm": 0.8413070351026624, "learning_rate": 9.04614111318022e-05, "loss": 3.8146, "step": 4372 }, { "epoch": 0.27146315724129366, "grad_norm": 0.8572020368355152, "learning_rate": 9.04821022139458e-05, "loss": 3.8851, "step": 4373 }, { "epoch": 0.2715252343410516, "grad_norm": 0.8355679631079944, "learning_rate": 9.050279329608939e-05, "loss": 3.7696, "step": 4374 }, { "epoch": 0.2715873114408095, "grad_norm": 0.7910346511624836, "learning_rate": 9.052348437823299e-05, "loss": 3.8582, "step": 4375 }, { "epoch": 0.27164938854056736, "grad_norm": 0.8431726175580688, "learning_rate": 9.054417546037658e-05, "loss": 3.809, "step": 4376 }, { "epoch": 0.2717114656403253, "grad_norm": 0.9738935085217773, "learning_rate": 9.056486654252018e-05, "loss": 3.8004, "step": 4377 }, { "epoch": 0.2717735427400832, "grad_norm": 1.2005887236468964, "learning_rate": 9.058555762466377e-05, "loss": 3.8478, "step": 4378 }, { "epoch": 0.27183561983984106, "grad_norm": 0.7774894999442106, "learning_rate": 9.060624870680738e-05, "loss": 3.7308, "step": 4379 }, { "epoch": 0.271897696939599, "grad_norm": 0.691191904065173, "learning_rate": 9.062693978895097e-05, "loss": 3.7561, "step": 4380 }, { "epoch": 0.2719597740393569, "grad_norm": 0.6665957656550181, "learning_rate": 9.064763087109456e-05, "loss": 3.9044, "step": 4381 }, { "epoch": 0.27202185113911476, "grad_norm": 0.7092718377070982, "learning_rate": 9.066832195323815e-05, "loss": 3.8842, "step": 4382 }, { "epoch": 0.2720839282388727, "grad_norm": 0.6828616772713176, "learning_rate": 9.068901303538176e-05, "loss": 3.789, "step": 4383 }, { "epoch": 0.2721460053386306, "grad_norm": 0.6221098355803317, "learning_rate": 9.070970411752535e-05, "loss": 3.7101, "step": 4384 }, { "epoch": 0.27220808243838845, "grad_norm": 0.6101604681321016, "learning_rate": 9.073039519966895e-05, "loss": 3.8539, "step": 4385 }, { "epoch": 0.2722701595381464, "grad_norm": 0.6218825555467007, "learning_rate": 9.075108628181254e-05, "loss": 3.7507, "step": 4386 }, { "epoch": 0.2723322366379043, "grad_norm": 0.6003677343945116, "learning_rate": 9.077177736395614e-05, "loss": 3.8108, "step": 4387 }, { "epoch": 0.27239431373766215, "grad_norm": 0.728778127058784, "learning_rate": 9.079246844609974e-05, "loss": 3.8217, "step": 4388 }, { "epoch": 0.27245639083742007, "grad_norm": 0.6926240917599845, "learning_rate": 9.081315952824333e-05, "loss": 3.8435, "step": 4389 }, { "epoch": 0.272518467937178, "grad_norm": 0.6965146120774621, "learning_rate": 9.083385061038693e-05, "loss": 3.7994, "step": 4390 }, { "epoch": 0.27258054503693585, "grad_norm": 0.7431460164068397, "learning_rate": 9.085454169253052e-05, "loss": 3.8221, "step": 4391 }, { "epoch": 0.27264262213669377, "grad_norm": 0.7328749819148342, "learning_rate": 9.087523277467413e-05, "loss": 3.879, "step": 4392 }, { "epoch": 0.2727046992364517, "grad_norm": 0.6688267512263362, "learning_rate": 9.089592385681771e-05, "loss": 3.727, "step": 4393 }, { "epoch": 0.27276677633620955, "grad_norm": 0.6188105235698651, "learning_rate": 9.09166149389613e-05, "loss": 3.7779, "step": 4394 }, { "epoch": 0.27282885343596747, "grad_norm": 0.6679164799098822, "learning_rate": 9.09373060211049e-05, "loss": 3.8391, "step": 4395 }, { "epoch": 0.2728909305357254, "grad_norm": 0.7873304638703998, "learning_rate": 9.095799710324851e-05, "loss": 3.9307, "step": 4396 }, { "epoch": 0.27295300763548325, "grad_norm": 0.7501457787934829, "learning_rate": 9.09786881853921e-05, "loss": 3.9024, "step": 4397 }, { "epoch": 0.27301508473524116, "grad_norm": 0.7140716802562015, "learning_rate": 9.09993792675357e-05, "loss": 3.8249, "step": 4398 }, { "epoch": 0.2730771618349991, "grad_norm": 0.7086122234926756, "learning_rate": 9.102007034967928e-05, "loss": 3.7278, "step": 4399 }, { "epoch": 0.27313923893475694, "grad_norm": 0.6757522753198432, "learning_rate": 9.104076143182289e-05, "loss": 3.8841, "step": 4400 }, { "epoch": 0.27320131603451486, "grad_norm": 0.7570049663189302, "learning_rate": 9.106145251396648e-05, "loss": 3.8275, "step": 4401 }, { "epoch": 0.2732633931342728, "grad_norm": 0.756592860410373, "learning_rate": 9.108214359611008e-05, "loss": 3.814, "step": 4402 }, { "epoch": 0.27332547023403064, "grad_norm": 0.6680628615442683, "learning_rate": 9.110283467825367e-05, "loss": 3.7376, "step": 4403 }, { "epoch": 0.27338754733378856, "grad_norm": 0.5945335037248881, "learning_rate": 9.112352576039728e-05, "loss": 3.7376, "step": 4404 }, { "epoch": 0.2734496244335465, "grad_norm": 0.6027414733112735, "learning_rate": 9.114421684254086e-05, "loss": 3.9014, "step": 4405 }, { "epoch": 0.27351170153330434, "grad_norm": 0.7466752347780972, "learning_rate": 9.116490792468446e-05, "loss": 3.8541, "step": 4406 }, { "epoch": 0.27357377863306226, "grad_norm": 0.7029379065929535, "learning_rate": 9.118559900682806e-05, "loss": 3.861, "step": 4407 }, { "epoch": 0.2736358557328202, "grad_norm": 0.6225439212042423, "learning_rate": 9.120629008897166e-05, "loss": 3.8288, "step": 4408 }, { "epoch": 0.27369793283257804, "grad_norm": 0.6258923359354297, "learning_rate": 9.122698117111526e-05, "loss": 3.8373, "step": 4409 }, { "epoch": 0.27376000993233596, "grad_norm": 0.6297792807961704, "learning_rate": 9.124767225325885e-05, "loss": 3.8087, "step": 4410 }, { "epoch": 0.2738220870320939, "grad_norm": 0.7001444054563467, "learning_rate": 9.126836333540245e-05, "loss": 3.8788, "step": 4411 }, { "epoch": 0.27388416413185174, "grad_norm": 0.7524045307646623, "learning_rate": 9.128905441754604e-05, "loss": 3.7603, "step": 4412 }, { "epoch": 0.27394624123160966, "grad_norm": 0.7162237968777987, "learning_rate": 9.130974549968964e-05, "loss": 3.7892, "step": 4413 }, { "epoch": 0.2740083183313676, "grad_norm": 0.7904535627894944, "learning_rate": 9.133043658183323e-05, "loss": 3.8283, "step": 4414 }, { "epoch": 0.27407039543112544, "grad_norm": 0.9377443313101863, "learning_rate": 9.135112766397683e-05, "loss": 3.759, "step": 4415 }, { "epoch": 0.27413247253088335, "grad_norm": 1.1537555386956908, "learning_rate": 9.137181874612042e-05, "loss": 3.7572, "step": 4416 }, { "epoch": 0.27419454963064127, "grad_norm": 0.789634836944521, "learning_rate": 9.139250982826403e-05, "loss": 3.8149, "step": 4417 }, { "epoch": 0.27425662673039913, "grad_norm": 0.8373093089035558, "learning_rate": 9.141320091040761e-05, "loss": 3.8248, "step": 4418 }, { "epoch": 0.27431870383015705, "grad_norm": 1.0845182186828253, "learning_rate": 9.143389199255121e-05, "loss": 3.8614, "step": 4419 }, { "epoch": 0.27438078092991497, "grad_norm": 0.8369966018908579, "learning_rate": 9.14545830746948e-05, "loss": 3.8971, "step": 4420 }, { "epoch": 0.27444285802967283, "grad_norm": 0.8667858671475294, "learning_rate": 9.147527415683841e-05, "loss": 3.7164, "step": 4421 }, { "epoch": 0.27450493512943075, "grad_norm": 0.900542604276277, "learning_rate": 9.149596523898201e-05, "loss": 3.7726, "step": 4422 }, { "epoch": 0.27456701222918867, "grad_norm": 0.7498751377436884, "learning_rate": 9.15166563211256e-05, "loss": 3.685, "step": 4423 }, { "epoch": 0.27462908932894653, "grad_norm": 0.8281328812709986, "learning_rate": 9.153734740326918e-05, "loss": 3.8124, "step": 4424 }, { "epoch": 0.27469116642870445, "grad_norm": 0.7484315127330367, "learning_rate": 9.15580384854128e-05, "loss": 3.7182, "step": 4425 }, { "epoch": 0.27475324352846237, "grad_norm": 0.7715423562196365, "learning_rate": 9.157872956755639e-05, "loss": 3.8897, "step": 4426 }, { "epoch": 0.2748153206282202, "grad_norm": 0.7964478264056659, "learning_rate": 9.159942064969998e-05, "loss": 3.6983, "step": 4427 }, { "epoch": 0.27487739772797815, "grad_norm": 0.7127044852772786, "learning_rate": 9.162011173184358e-05, "loss": 3.7748, "step": 4428 }, { "epoch": 0.27493947482773606, "grad_norm": 0.6330157530468123, "learning_rate": 9.164080281398719e-05, "loss": 3.9309, "step": 4429 }, { "epoch": 0.2750015519274939, "grad_norm": 0.5864803088753422, "learning_rate": 9.166149389613077e-05, "loss": 3.7874, "step": 4430 }, { "epoch": 0.27506362902725184, "grad_norm": 0.6657128862556889, "learning_rate": 9.168218497827436e-05, "loss": 3.7955, "step": 4431 }, { "epoch": 0.27512570612700976, "grad_norm": 0.631957595173681, "learning_rate": 9.170287606041796e-05, "loss": 3.7872, "step": 4432 }, { "epoch": 0.2751877832267676, "grad_norm": 0.6519988741630944, "learning_rate": 9.172356714256157e-05, "loss": 3.8845, "step": 4433 }, { "epoch": 0.27524986032652554, "grad_norm": 0.718641521726422, "learning_rate": 9.174425822470516e-05, "loss": 3.8355, "step": 4434 }, { "epoch": 0.27531193742628346, "grad_norm": 0.6795275708289945, "learning_rate": 9.176494930684876e-05, "loss": 3.7806, "step": 4435 }, { "epoch": 0.2753740145260413, "grad_norm": 0.7412674270433973, "learning_rate": 9.178564038899234e-05, "loss": 3.8503, "step": 4436 }, { "epoch": 0.27543609162579924, "grad_norm": 0.7325704601888036, "learning_rate": 9.180633147113595e-05, "loss": 3.8521, "step": 4437 }, { "epoch": 0.27549816872555716, "grad_norm": 0.7035078489900064, "learning_rate": 9.182702255327954e-05, "loss": 3.8426, "step": 4438 }, { "epoch": 0.275560245825315, "grad_norm": 0.7026860032283236, "learning_rate": 9.184771363542314e-05, "loss": 3.9016, "step": 4439 }, { "epoch": 0.27562232292507294, "grad_norm": 0.715530649153857, "learning_rate": 9.186840471756673e-05, "loss": 3.9433, "step": 4440 }, { "epoch": 0.27568440002483086, "grad_norm": 0.658811384377613, "learning_rate": 9.188909579971033e-05, "loss": 3.7966, "step": 4441 }, { "epoch": 0.2757464771245887, "grad_norm": 0.640749988923536, "learning_rate": 9.190978688185392e-05, "loss": 3.838, "step": 4442 }, { "epoch": 0.27580855422434664, "grad_norm": 0.7105446453174842, "learning_rate": 9.193047796399752e-05, "loss": 3.779, "step": 4443 }, { "epoch": 0.27587063132410455, "grad_norm": 0.7823118622903646, "learning_rate": 9.195116904614111e-05, "loss": 3.7763, "step": 4444 }, { "epoch": 0.2759327084238624, "grad_norm": 0.6922714079208613, "learning_rate": 9.197186012828471e-05, "loss": 3.715, "step": 4445 }, { "epoch": 0.27599478552362033, "grad_norm": 0.6931322182364613, "learning_rate": 9.199255121042832e-05, "loss": 3.8195, "step": 4446 }, { "epoch": 0.27605686262337825, "grad_norm": 0.7996543584636732, "learning_rate": 9.201324229257191e-05, "loss": 3.7894, "step": 4447 }, { "epoch": 0.2761189397231361, "grad_norm": 0.8727928321116598, "learning_rate": 9.20339333747155e-05, "loss": 3.7196, "step": 4448 }, { "epoch": 0.27618101682289403, "grad_norm": 0.9708904492408941, "learning_rate": 9.205462445685909e-05, "loss": 3.8194, "step": 4449 }, { "epoch": 0.27624309392265195, "grad_norm": 0.9260703814373829, "learning_rate": 9.20753155390027e-05, "loss": 3.7308, "step": 4450 }, { "epoch": 0.2763051710224098, "grad_norm": 0.7499160322233347, "learning_rate": 9.209600662114629e-05, "loss": 3.7649, "step": 4451 }, { "epoch": 0.27636724812216773, "grad_norm": 0.7552363467227748, "learning_rate": 9.211669770328989e-05, "loss": 3.9752, "step": 4452 }, { "epoch": 0.27642932522192565, "grad_norm": 0.8364677530014043, "learning_rate": 9.213738878543348e-05, "loss": 3.9077, "step": 4453 }, { "epoch": 0.2764914023216835, "grad_norm": 0.849822234750911, "learning_rate": 9.215807986757708e-05, "loss": 3.8859, "step": 4454 }, { "epoch": 0.27655347942144143, "grad_norm": 0.809887831833412, "learning_rate": 9.217877094972067e-05, "loss": 3.8687, "step": 4455 }, { "epoch": 0.27661555652119935, "grad_norm": 0.7860182400254138, "learning_rate": 9.219946203186427e-05, "loss": 3.8701, "step": 4456 }, { "epoch": 0.2766776336209572, "grad_norm": 0.8273950274519294, "learning_rate": 9.222015311400786e-05, "loss": 3.7994, "step": 4457 }, { "epoch": 0.2767397107207151, "grad_norm": 0.9483468403605757, "learning_rate": 9.224084419615147e-05, "loss": 3.9456, "step": 4458 }, { "epoch": 0.27680178782047304, "grad_norm": 0.8898721826044969, "learning_rate": 9.226153527829507e-05, "loss": 3.7856, "step": 4459 }, { "epoch": 0.2768638649202309, "grad_norm": 0.8640900357430659, "learning_rate": 9.228222636043865e-05, "loss": 3.8636, "step": 4460 }, { "epoch": 0.2769259420199888, "grad_norm": 0.8886620489562365, "learning_rate": 9.230291744258224e-05, "loss": 3.8266, "step": 4461 }, { "epoch": 0.27698801911974674, "grad_norm": 0.7865439530475609, "learning_rate": 9.232360852472585e-05, "loss": 3.7937, "step": 4462 }, { "epoch": 0.2770500962195046, "grad_norm": 0.8075008285447349, "learning_rate": 9.234429960686945e-05, "loss": 3.8345, "step": 4463 }, { "epoch": 0.2771121733192625, "grad_norm": 0.727809492197533, "learning_rate": 9.236499068901304e-05, "loss": 3.8075, "step": 4464 }, { "epoch": 0.27717425041902044, "grad_norm": 0.83133955308193, "learning_rate": 9.238568177115664e-05, "loss": 3.7569, "step": 4465 }, { "epoch": 0.2772363275187783, "grad_norm": 0.7743755393530543, "learning_rate": 9.240637285330023e-05, "loss": 3.7929, "step": 4466 }, { "epoch": 0.2772984046185362, "grad_norm": 0.7456515534653791, "learning_rate": 9.242706393544383e-05, "loss": 3.8488, "step": 4467 }, { "epoch": 0.27736048171829414, "grad_norm": 0.7159356288477515, "learning_rate": 9.244775501758742e-05, "loss": 3.6992, "step": 4468 }, { "epoch": 0.277422558818052, "grad_norm": 0.8604566529398444, "learning_rate": 9.246844609973102e-05, "loss": 3.8506, "step": 4469 }, { "epoch": 0.2774846359178099, "grad_norm": 0.6885716365884276, "learning_rate": 9.248913718187461e-05, "loss": 3.7816, "step": 4470 }, { "epoch": 0.27754671301756784, "grad_norm": 0.5929591031100242, "learning_rate": 9.250982826401822e-05, "loss": 3.7551, "step": 4471 }, { "epoch": 0.2776087901173257, "grad_norm": 0.7611884594300552, "learning_rate": 9.25305193461618e-05, "loss": 3.8739, "step": 4472 }, { "epoch": 0.2776708672170836, "grad_norm": 0.699507946540517, "learning_rate": 9.25512104283054e-05, "loss": 3.7982, "step": 4473 }, { "epoch": 0.27773294431684153, "grad_norm": 0.7257872105939854, "learning_rate": 9.257190151044899e-05, "loss": 3.7363, "step": 4474 }, { "epoch": 0.2777950214165994, "grad_norm": 0.6355225495675467, "learning_rate": 9.25925925925926e-05, "loss": 3.8437, "step": 4475 }, { "epoch": 0.2778570985163573, "grad_norm": 0.5988906628625288, "learning_rate": 9.26132836747362e-05, "loss": 3.734, "step": 4476 }, { "epoch": 0.27791917561611523, "grad_norm": 0.6361513242291466, "learning_rate": 9.263397475687979e-05, "loss": 3.7321, "step": 4477 }, { "epoch": 0.2779812527158731, "grad_norm": 0.6145313424140169, "learning_rate": 9.265466583902339e-05, "loss": 3.662, "step": 4478 }, { "epoch": 0.278043329815631, "grad_norm": 0.6158800158489013, "learning_rate": 9.267535692116698e-05, "loss": 3.9414, "step": 4479 }, { "epoch": 0.27810540691538893, "grad_norm": 0.6262961061526856, "learning_rate": 9.269604800331058e-05, "loss": 3.7726, "step": 4480 }, { "epoch": 0.2781674840151468, "grad_norm": 0.680434353787957, "learning_rate": 9.271673908545417e-05, "loss": 3.7366, "step": 4481 }, { "epoch": 0.2782295611149047, "grad_norm": 0.7544052765213709, "learning_rate": 9.273743016759777e-05, "loss": 3.8195, "step": 4482 }, { "epoch": 0.27829163821466263, "grad_norm": 0.7177753889615266, "learning_rate": 9.275812124974138e-05, "loss": 3.791, "step": 4483 }, { "epoch": 0.2783537153144205, "grad_norm": 0.6204133055138815, "learning_rate": 9.277881233188496e-05, "loss": 3.7432, "step": 4484 }, { "epoch": 0.2784157924141784, "grad_norm": 0.7363897646898744, "learning_rate": 9.279950341402855e-05, "loss": 3.7272, "step": 4485 }, { "epoch": 0.2784778695139363, "grad_norm": 0.8415528052019843, "learning_rate": 9.282019449617215e-05, "loss": 3.7302, "step": 4486 }, { "epoch": 0.2785399466136942, "grad_norm": 0.905864238849329, "learning_rate": 9.284088557831576e-05, "loss": 3.8005, "step": 4487 }, { "epoch": 0.2786020237134521, "grad_norm": 0.9842052918934311, "learning_rate": 9.286157666045935e-05, "loss": 3.8693, "step": 4488 }, { "epoch": 0.27866410081321, "grad_norm": 0.9493249590444979, "learning_rate": 9.288226774260295e-05, "loss": 3.8438, "step": 4489 }, { "epoch": 0.2787261779129679, "grad_norm": 0.893919523207604, "learning_rate": 9.290295882474653e-05, "loss": 3.717, "step": 4490 }, { "epoch": 0.2787882550127258, "grad_norm": 0.7812029252174797, "learning_rate": 9.292364990689014e-05, "loss": 3.7135, "step": 4491 }, { "epoch": 0.2788503321124837, "grad_norm": 0.7646788765218776, "learning_rate": 9.294434098903373e-05, "loss": 3.8384, "step": 4492 }, { "epoch": 0.2789124092122416, "grad_norm": 0.7554101520838854, "learning_rate": 9.296503207117733e-05, "loss": 3.7429, "step": 4493 }, { "epoch": 0.2789744863119995, "grad_norm": 0.7194115455286388, "learning_rate": 9.298572315332092e-05, "loss": 3.8303, "step": 4494 }, { "epoch": 0.2790365634117574, "grad_norm": 0.8044349496312125, "learning_rate": 9.300641423546452e-05, "loss": 3.7775, "step": 4495 }, { "epoch": 0.2790986405115153, "grad_norm": 0.7365072422072926, "learning_rate": 9.302710531760813e-05, "loss": 3.7673, "step": 4496 }, { "epoch": 0.2791607176112732, "grad_norm": 0.7038850702789199, "learning_rate": 9.304779639975171e-05, "loss": 3.8837, "step": 4497 }, { "epoch": 0.2792227947110311, "grad_norm": 0.690483209890631, "learning_rate": 9.30684874818953e-05, "loss": 3.7641, "step": 4498 }, { "epoch": 0.279284871810789, "grad_norm": 0.737087798902039, "learning_rate": 9.30891785640389e-05, "loss": 3.8031, "step": 4499 }, { "epoch": 0.2793469489105469, "grad_norm": 0.7760111826258065, "learning_rate": 9.31098696461825e-05, "loss": 3.8327, "step": 4500 }, { "epoch": 0.2794090260103048, "grad_norm": 0.8133847847738819, "learning_rate": 9.31305607283261e-05, "loss": 3.8211, "step": 4501 }, { "epoch": 0.2794711031100627, "grad_norm": 0.7948109692337355, "learning_rate": 9.31512518104697e-05, "loss": 3.836, "step": 4502 }, { "epoch": 0.2795331802098206, "grad_norm": 0.6974918842527158, "learning_rate": 9.317194289261329e-05, "loss": 3.743, "step": 4503 }, { "epoch": 0.2795952573095785, "grad_norm": 0.7639812418472326, "learning_rate": 9.319263397475689e-05, "loss": 3.8487, "step": 4504 }, { "epoch": 0.2796573344093364, "grad_norm": 0.8938009287588843, "learning_rate": 9.321332505690048e-05, "loss": 3.8343, "step": 4505 }, { "epoch": 0.2797194115090943, "grad_norm": 0.8646541742617495, "learning_rate": 9.323401613904408e-05, "loss": 3.767, "step": 4506 }, { "epoch": 0.2797814886088522, "grad_norm": 0.6817137670761093, "learning_rate": 9.325470722118767e-05, "loss": 3.8529, "step": 4507 }, { "epoch": 0.2798435657086101, "grad_norm": 0.7424849867667599, "learning_rate": 9.327539830333128e-05, "loss": 3.7991, "step": 4508 }, { "epoch": 0.279905642808368, "grad_norm": 0.8472927386393792, "learning_rate": 9.329608938547486e-05, "loss": 3.818, "step": 4509 }, { "epoch": 0.2799677199081259, "grad_norm": 0.8019708496751223, "learning_rate": 9.331678046761846e-05, "loss": 3.8386, "step": 4510 }, { "epoch": 0.2800297970078838, "grad_norm": 0.6996196471311621, "learning_rate": 9.333747154976205e-05, "loss": 3.7046, "step": 4511 }, { "epoch": 0.2800918741076417, "grad_norm": 0.6435578364988398, "learning_rate": 9.335816263190566e-05, "loss": 3.7854, "step": 4512 }, { "epoch": 0.2801539512073996, "grad_norm": 0.7525033498290266, "learning_rate": 9.337885371404926e-05, "loss": 4.0069, "step": 4513 }, { "epoch": 0.28021602830715747, "grad_norm": 0.7625581078197688, "learning_rate": 9.339954479619285e-05, "loss": 3.8857, "step": 4514 }, { "epoch": 0.2802781054069154, "grad_norm": 0.6491742831539519, "learning_rate": 9.342023587833643e-05, "loss": 3.7971, "step": 4515 }, { "epoch": 0.2803401825066733, "grad_norm": 0.628671234953488, "learning_rate": 9.344092696048004e-05, "loss": 3.9143, "step": 4516 }, { "epoch": 0.28040225960643117, "grad_norm": 0.6336628886740145, "learning_rate": 9.346161804262364e-05, "loss": 3.8921, "step": 4517 }, { "epoch": 0.2804643367061891, "grad_norm": 0.6994983604185618, "learning_rate": 9.348230912476723e-05, "loss": 3.9628, "step": 4518 }, { "epoch": 0.280526413805947, "grad_norm": 0.7176347455823308, "learning_rate": 9.350300020691083e-05, "loss": 3.8081, "step": 4519 }, { "epoch": 0.28058849090570487, "grad_norm": 0.7132968111210526, "learning_rate": 9.352369128905442e-05, "loss": 3.7861, "step": 4520 }, { "epoch": 0.2806505680054628, "grad_norm": 0.7273500747487622, "learning_rate": 9.354438237119802e-05, "loss": 3.7134, "step": 4521 }, { "epoch": 0.2807126451052207, "grad_norm": 0.6795363422552978, "learning_rate": 9.356507345334161e-05, "loss": 3.8321, "step": 4522 }, { "epoch": 0.28077472220497857, "grad_norm": 0.68552368211775, "learning_rate": 9.35857645354852e-05, "loss": 3.7564, "step": 4523 }, { "epoch": 0.2808367993047365, "grad_norm": 0.6836147414117998, "learning_rate": 9.36064556176288e-05, "loss": 3.8335, "step": 4524 }, { "epoch": 0.2808988764044944, "grad_norm": 0.7151514860548353, "learning_rate": 9.362714669977241e-05, "loss": 3.8095, "step": 4525 }, { "epoch": 0.28096095350425226, "grad_norm": 0.8156147455615858, "learning_rate": 9.3647837781916e-05, "loss": 3.7867, "step": 4526 }, { "epoch": 0.2810230306040102, "grad_norm": 0.904360116020211, "learning_rate": 9.366852886405959e-05, "loss": 3.8436, "step": 4527 }, { "epoch": 0.2810851077037681, "grad_norm": 0.9397060188724237, "learning_rate": 9.36892199462032e-05, "loss": 3.8405, "step": 4528 }, { "epoch": 0.28114718480352596, "grad_norm": 0.7000757431876915, "learning_rate": 9.370991102834679e-05, "loss": 3.7591, "step": 4529 }, { "epoch": 0.2812092619032839, "grad_norm": 0.6883909131606316, "learning_rate": 9.373060211049038e-05, "loss": 3.6731, "step": 4530 }, { "epoch": 0.2812713390030418, "grad_norm": 0.7563569552870094, "learning_rate": 9.375129319263398e-05, "loss": 3.8397, "step": 4531 }, { "epoch": 0.28133341610279966, "grad_norm": 0.7376358602138368, "learning_rate": 9.377198427477758e-05, "loss": 3.7766, "step": 4532 }, { "epoch": 0.2813954932025576, "grad_norm": 0.708929222631001, "learning_rate": 9.379267535692117e-05, "loss": 3.8742, "step": 4533 }, { "epoch": 0.2814575703023155, "grad_norm": 0.6855276816960766, "learning_rate": 9.381336643906477e-05, "loss": 3.8433, "step": 4534 }, { "epoch": 0.28151964740207336, "grad_norm": 0.664050225537123, "learning_rate": 9.383405752120836e-05, "loss": 3.8015, "step": 4535 }, { "epoch": 0.2815817245018313, "grad_norm": 0.6008129652206634, "learning_rate": 9.385474860335196e-05, "loss": 3.8089, "step": 4536 }, { "epoch": 0.2816438016015892, "grad_norm": 0.6272061511784222, "learning_rate": 9.387543968549556e-05, "loss": 3.807, "step": 4537 }, { "epoch": 0.28170587870134706, "grad_norm": 0.6259828230950945, "learning_rate": 9.389613076763916e-05, "loss": 3.8062, "step": 4538 }, { "epoch": 0.281767955801105, "grad_norm": 0.6217469673052098, "learning_rate": 9.391682184978274e-05, "loss": 3.7662, "step": 4539 }, { "epoch": 0.2818300329008629, "grad_norm": 0.712383434732346, "learning_rate": 9.393751293192634e-05, "loss": 3.8364, "step": 4540 }, { "epoch": 0.28189211000062075, "grad_norm": 0.6502409066749103, "learning_rate": 9.395820401406994e-05, "loss": 3.7388, "step": 4541 }, { "epoch": 0.2819541871003787, "grad_norm": 0.6420502044858029, "learning_rate": 9.397889509621354e-05, "loss": 3.7599, "step": 4542 }, { "epoch": 0.2820162642001366, "grad_norm": 0.7295341020055709, "learning_rate": 9.399958617835713e-05, "loss": 3.8074, "step": 4543 }, { "epoch": 0.28207834129989445, "grad_norm": 0.6891179769591628, "learning_rate": 9.402027726050073e-05, "loss": 3.7777, "step": 4544 }, { "epoch": 0.28214041839965237, "grad_norm": 0.7108507627457465, "learning_rate": 9.404096834264432e-05, "loss": 3.7048, "step": 4545 }, { "epoch": 0.2822024954994103, "grad_norm": 0.6906787255910117, "learning_rate": 9.406165942478792e-05, "loss": 3.7932, "step": 4546 }, { "epoch": 0.28226457259916815, "grad_norm": 0.7056429389180289, "learning_rate": 9.408235050693151e-05, "loss": 3.7373, "step": 4547 }, { "epoch": 0.28232664969892607, "grad_norm": 0.6455940500616278, "learning_rate": 9.410304158907511e-05, "loss": 3.7337, "step": 4548 }, { "epoch": 0.282388726798684, "grad_norm": 0.6358155004651223, "learning_rate": 9.412373267121872e-05, "loss": 3.7785, "step": 4549 }, { "epoch": 0.28245080389844185, "grad_norm": 0.7672006392276396, "learning_rate": 9.414442375336231e-05, "loss": 3.8495, "step": 4550 }, { "epoch": 0.28251288099819977, "grad_norm": 0.697111716501112, "learning_rate": 9.41651148355059e-05, "loss": 3.8107, "step": 4551 }, { "epoch": 0.2825749580979577, "grad_norm": 0.7411013234423287, "learning_rate": 9.418580591764949e-05, "loss": 3.7215, "step": 4552 }, { "epoch": 0.28263703519771555, "grad_norm": 0.7227868039535734, "learning_rate": 9.42064969997931e-05, "loss": 3.7943, "step": 4553 }, { "epoch": 0.28269911229747346, "grad_norm": 0.7964917551608907, "learning_rate": 9.42271880819367e-05, "loss": 3.7079, "step": 4554 }, { "epoch": 0.2827611893972314, "grad_norm": 0.7730902846120035, "learning_rate": 9.424787916408029e-05, "loss": 3.7694, "step": 4555 }, { "epoch": 0.28282326649698925, "grad_norm": 0.7381616323943511, "learning_rate": 9.426857024622388e-05, "loss": 3.8519, "step": 4556 }, { "epoch": 0.28288534359674716, "grad_norm": 0.7880210424314945, "learning_rate": 9.428926132836748e-05, "loss": 3.7108, "step": 4557 }, { "epoch": 0.2829474206965051, "grad_norm": 0.6917436067653149, "learning_rate": 9.430995241051107e-05, "loss": 3.8321, "step": 4558 }, { "epoch": 0.28300949779626294, "grad_norm": 0.6740359663040689, "learning_rate": 9.433064349265467e-05, "loss": 3.7957, "step": 4559 }, { "epoch": 0.28307157489602086, "grad_norm": 0.705752317229523, "learning_rate": 9.435133457479826e-05, "loss": 3.7738, "step": 4560 }, { "epoch": 0.2831336519957788, "grad_norm": 0.7641387058940268, "learning_rate": 9.437202565694186e-05, "loss": 3.7245, "step": 4561 }, { "epoch": 0.28319572909553664, "grad_norm": 0.7237232476068369, "learning_rate": 9.439271673908547e-05, "loss": 3.7341, "step": 4562 }, { "epoch": 0.28325780619529456, "grad_norm": 0.7782205208973951, "learning_rate": 9.441340782122905e-05, "loss": 3.6986, "step": 4563 }, { "epoch": 0.2833198832950525, "grad_norm": 0.7913415308074873, "learning_rate": 9.443409890337264e-05, "loss": 3.7562, "step": 4564 }, { "epoch": 0.28338196039481034, "grad_norm": 0.8788417538184561, "learning_rate": 9.445478998551624e-05, "loss": 3.7341, "step": 4565 }, { "epoch": 0.28344403749456826, "grad_norm": 0.9679530772863869, "learning_rate": 9.447548106765985e-05, "loss": 3.7326, "step": 4566 }, { "epoch": 0.2835061145943262, "grad_norm": 1.0584288746719157, "learning_rate": 9.449617214980344e-05, "loss": 3.7293, "step": 4567 }, { "epoch": 0.28356819169408404, "grad_norm": 0.7106847628221278, "learning_rate": 9.451686323194704e-05, "loss": 3.742, "step": 4568 }, { "epoch": 0.28363026879384196, "grad_norm": 0.7773429065488907, "learning_rate": 9.453755431409062e-05, "loss": 3.8392, "step": 4569 }, { "epoch": 0.2836923458935999, "grad_norm": 0.8527188979879291, "learning_rate": 9.455824539623423e-05, "loss": 3.8004, "step": 4570 }, { "epoch": 0.28375442299335774, "grad_norm": 0.6904163925854867, "learning_rate": 9.457893647837782e-05, "loss": 3.8155, "step": 4571 }, { "epoch": 0.28381650009311565, "grad_norm": 0.7829288373376404, "learning_rate": 9.459962756052142e-05, "loss": 3.7416, "step": 4572 }, { "epoch": 0.28387857719287357, "grad_norm": 0.6848975907993371, "learning_rate": 9.462031864266501e-05, "loss": 3.9016, "step": 4573 }, { "epoch": 0.28394065429263143, "grad_norm": 0.5895266406791294, "learning_rate": 9.464100972480862e-05, "loss": 3.806, "step": 4574 }, { "epoch": 0.28400273139238935, "grad_norm": 0.7369288819392855, "learning_rate": 9.46617008069522e-05, "loss": 3.7753, "step": 4575 }, { "epoch": 0.28406480849214727, "grad_norm": 0.6773722227947367, "learning_rate": 9.46823918890958e-05, "loss": 3.7942, "step": 4576 }, { "epoch": 0.28412688559190513, "grad_norm": 0.7533362412888882, "learning_rate": 9.47030829712394e-05, "loss": 3.7432, "step": 4577 }, { "epoch": 0.28418896269166305, "grad_norm": 0.6974844282045949, "learning_rate": 9.4723774053383e-05, "loss": 3.7776, "step": 4578 }, { "epoch": 0.28425103979142097, "grad_norm": 0.7191612891316255, "learning_rate": 9.47444651355266e-05, "loss": 3.8384, "step": 4579 }, { "epoch": 0.28431311689117883, "grad_norm": 0.6581080563312226, "learning_rate": 9.476515621767019e-05, "loss": 3.7716, "step": 4580 }, { "epoch": 0.28437519399093675, "grad_norm": 0.7823499946505008, "learning_rate": 9.478584729981379e-05, "loss": 3.8524, "step": 4581 }, { "epoch": 0.28443727109069467, "grad_norm": 0.7468490771659078, "learning_rate": 9.480653838195738e-05, "loss": 3.7604, "step": 4582 }, { "epoch": 0.28449934819045253, "grad_norm": 0.7267241349448289, "learning_rate": 9.482722946410098e-05, "loss": 3.7246, "step": 4583 }, { "epoch": 0.28456142529021045, "grad_norm": 0.6701123158770635, "learning_rate": 9.484792054624457e-05, "loss": 3.8452, "step": 4584 }, { "epoch": 0.28462350238996836, "grad_norm": 0.6981195728975573, "learning_rate": 9.486861162838817e-05, "loss": 3.8426, "step": 4585 }, { "epoch": 0.2846855794897262, "grad_norm": 0.8659153103747291, "learning_rate": 9.488930271053176e-05, "loss": 3.7936, "step": 4586 }, { "epoch": 0.28474765658948414, "grad_norm": 0.9196648694286181, "learning_rate": 9.490999379267537e-05, "loss": 3.8151, "step": 4587 }, { "epoch": 0.28480973368924206, "grad_norm": 0.7677421103456856, "learning_rate": 9.493068487481895e-05, "loss": 3.7966, "step": 4588 }, { "epoch": 0.2848718107889999, "grad_norm": 0.6804145563664817, "learning_rate": 9.495137595696255e-05, "loss": 3.8016, "step": 4589 }, { "epoch": 0.28493388788875784, "grad_norm": 0.6633827095927908, "learning_rate": 9.497206703910614e-05, "loss": 3.7384, "step": 4590 }, { "epoch": 0.28499596498851576, "grad_norm": 0.7599954135028577, "learning_rate": 9.499275812124975e-05, "loss": 3.8375, "step": 4591 }, { "epoch": 0.2850580420882736, "grad_norm": 0.7489313618592202, "learning_rate": 9.501344920339335e-05, "loss": 3.7958, "step": 4592 }, { "epoch": 0.28512011918803154, "grad_norm": 0.7406927214617022, "learning_rate": 9.503414028553694e-05, "loss": 3.8046, "step": 4593 }, { "epoch": 0.28518219628778946, "grad_norm": 0.6441507314394631, "learning_rate": 9.505483136768052e-05, "loss": 3.7774, "step": 4594 }, { "epoch": 0.2852442733875473, "grad_norm": 0.6444305803691244, "learning_rate": 9.507552244982413e-05, "loss": 3.7971, "step": 4595 }, { "epoch": 0.28530635048730524, "grad_norm": 0.6135721185075417, "learning_rate": 9.509621353196773e-05, "loss": 3.7854, "step": 4596 }, { "epoch": 0.28536842758706316, "grad_norm": 0.572586308224476, "learning_rate": 9.511690461411132e-05, "loss": 3.8237, "step": 4597 }, { "epoch": 0.285430504686821, "grad_norm": 0.7223464714261428, "learning_rate": 9.513759569625492e-05, "loss": 3.7514, "step": 4598 }, { "epoch": 0.28549258178657894, "grad_norm": 0.6668082523995482, "learning_rate": 9.515828677839853e-05, "loss": 3.7643, "step": 4599 }, { "epoch": 0.28555465888633685, "grad_norm": 0.6410725821064576, "learning_rate": 9.517897786054211e-05, "loss": 3.7033, "step": 4600 }, { "epoch": 0.2856167359860947, "grad_norm": 0.6549733323208534, "learning_rate": 9.51996689426857e-05, "loss": 3.8071, "step": 4601 }, { "epoch": 0.28567881308585263, "grad_norm": 0.5843151775889996, "learning_rate": 9.52203600248293e-05, "loss": 3.6937, "step": 4602 }, { "epoch": 0.28574089018561055, "grad_norm": 0.5838439415323594, "learning_rate": 9.524105110697291e-05, "loss": 3.7185, "step": 4603 }, { "epoch": 0.2858029672853684, "grad_norm": 0.5850403847978413, "learning_rate": 9.52617421891165e-05, "loss": 3.758, "step": 4604 }, { "epoch": 0.28586504438512633, "grad_norm": 0.6305937783685375, "learning_rate": 9.52824332712601e-05, "loss": 3.8465, "step": 4605 }, { "epoch": 0.28592712148488425, "grad_norm": 0.7148269484583091, "learning_rate": 9.530312435340368e-05, "loss": 3.8287, "step": 4606 }, { "epoch": 0.2859891985846421, "grad_norm": 0.7440677316762899, "learning_rate": 9.532381543554729e-05, "loss": 3.8936, "step": 4607 }, { "epoch": 0.28605127568440003, "grad_norm": 0.704170170799731, "learning_rate": 9.534450651769088e-05, "loss": 3.6673, "step": 4608 }, { "epoch": 0.28611335278415795, "grad_norm": 0.7019783206550997, "learning_rate": 9.536519759983448e-05, "loss": 3.7976, "step": 4609 }, { "epoch": 0.2861754298839158, "grad_norm": 0.7935134820664114, "learning_rate": 9.538588868197807e-05, "loss": 3.7555, "step": 4610 }, { "epoch": 0.28623750698367373, "grad_norm": 1.0772490329060846, "learning_rate": 9.540657976412167e-05, "loss": 3.8605, "step": 4611 }, { "epoch": 0.28629958408343165, "grad_norm": 0.8322585963879235, "learning_rate": 9.542727084626526e-05, "loss": 3.864, "step": 4612 }, { "epoch": 0.2863616611831895, "grad_norm": 0.7712492379257418, "learning_rate": 9.544796192840886e-05, "loss": 3.7582, "step": 4613 }, { "epoch": 0.2864237382829474, "grad_norm": 0.7874107824983045, "learning_rate": 9.546865301055245e-05, "loss": 3.7862, "step": 4614 }, { "epoch": 0.28648581538270534, "grad_norm": 0.7921642111942121, "learning_rate": 9.548934409269605e-05, "loss": 3.7124, "step": 4615 }, { "epoch": 0.2865478924824632, "grad_norm": 0.739378582904612, "learning_rate": 9.551003517483966e-05, "loss": 3.8464, "step": 4616 }, { "epoch": 0.2866099695822211, "grad_norm": 0.8550699427924643, "learning_rate": 9.553072625698325e-05, "loss": 3.7835, "step": 4617 }, { "epoch": 0.28667204668197904, "grad_norm": 0.7285980096210243, "learning_rate": 9.555141733912683e-05, "loss": 3.8348, "step": 4618 }, { "epoch": 0.2867341237817369, "grad_norm": 0.6743699921924946, "learning_rate": 9.557210842127043e-05, "loss": 3.7539, "step": 4619 }, { "epoch": 0.2867962008814948, "grad_norm": 0.76363682971358, "learning_rate": 9.559279950341404e-05, "loss": 3.7105, "step": 4620 }, { "epoch": 0.28685827798125274, "grad_norm": 0.6446178493698626, "learning_rate": 9.561349058555763e-05, "loss": 3.7586, "step": 4621 }, { "epoch": 0.2869203550810106, "grad_norm": 0.7429596347831312, "learning_rate": 9.563418166770123e-05, "loss": 3.776, "step": 4622 }, { "epoch": 0.2869824321807685, "grad_norm": 0.6666364712632125, "learning_rate": 9.565487274984482e-05, "loss": 3.7427, "step": 4623 }, { "epoch": 0.28704450928052644, "grad_norm": 0.6523004848729305, "learning_rate": 9.567556383198842e-05, "loss": 3.7161, "step": 4624 }, { "epoch": 0.2871065863802843, "grad_norm": 0.6214694919878112, "learning_rate": 9.569625491413201e-05, "loss": 3.6955, "step": 4625 }, { "epoch": 0.2871686634800422, "grad_norm": 0.6063224680346314, "learning_rate": 9.571694599627561e-05, "loss": 3.6737, "step": 4626 }, { "epoch": 0.28723074057980014, "grad_norm": 0.6978617879063421, "learning_rate": 9.57376370784192e-05, "loss": 3.7887, "step": 4627 }, { "epoch": 0.287292817679558, "grad_norm": 0.6295496806355605, "learning_rate": 9.575832816056281e-05, "loss": 3.7207, "step": 4628 }, { "epoch": 0.2873548947793159, "grad_norm": 0.7455387148050276, "learning_rate": 9.57790192427064e-05, "loss": 3.7355, "step": 4629 }, { "epoch": 0.28741697187907383, "grad_norm": 0.7644026983799359, "learning_rate": 9.579971032484999e-05, "loss": 3.6941, "step": 4630 }, { "epoch": 0.2874790489788317, "grad_norm": 0.8152641813699438, "learning_rate": 9.582040140699358e-05, "loss": 3.8034, "step": 4631 }, { "epoch": 0.2875411260785896, "grad_norm": 0.7929447339559996, "learning_rate": 9.584109248913719e-05, "loss": 3.7915, "step": 4632 }, { "epoch": 0.28760320317834753, "grad_norm": 0.6544090516449534, "learning_rate": 9.586178357128079e-05, "loss": 3.7602, "step": 4633 }, { "epoch": 0.2876652802781054, "grad_norm": 0.5918915927740385, "learning_rate": 9.588247465342438e-05, "loss": 3.79, "step": 4634 }, { "epoch": 0.2877273573778633, "grad_norm": 0.6168140053861653, "learning_rate": 9.590316573556798e-05, "loss": 3.832, "step": 4635 }, { "epoch": 0.28778943447762123, "grad_norm": 0.6538585504159339, "learning_rate": 9.592385681771157e-05, "loss": 3.7264, "step": 4636 }, { "epoch": 0.2878515115773791, "grad_norm": 0.7072404583543193, "learning_rate": 9.594454789985517e-05, "loss": 3.7317, "step": 4637 }, { "epoch": 0.287913588677137, "grad_norm": 0.8024466711370425, "learning_rate": 9.596523898199876e-05, "loss": 3.7422, "step": 4638 }, { "epoch": 0.28797566577689493, "grad_norm": 0.8338137950253808, "learning_rate": 9.598593006414236e-05, "loss": 3.7446, "step": 4639 }, { "epoch": 0.2880377428766528, "grad_norm": 0.8982658885757752, "learning_rate": 9.600662114628595e-05, "loss": 3.7508, "step": 4640 }, { "epoch": 0.2880998199764107, "grad_norm": 0.7977057595724961, "learning_rate": 9.602731222842956e-05, "loss": 3.7021, "step": 4641 }, { "epoch": 0.2881618970761686, "grad_norm": 0.8207973319031664, "learning_rate": 9.604800331057314e-05, "loss": 3.7807, "step": 4642 }, { "epoch": 0.2882239741759265, "grad_norm": 0.7731678039740867, "learning_rate": 9.606869439271674e-05, "loss": 3.8786, "step": 4643 }, { "epoch": 0.2882860512756844, "grad_norm": 0.7131202641429331, "learning_rate": 9.608938547486033e-05, "loss": 3.8648, "step": 4644 }, { "epoch": 0.2883481283754423, "grad_norm": 0.7318038163721824, "learning_rate": 9.611007655700394e-05, "loss": 3.7668, "step": 4645 }, { "epoch": 0.2884102054752002, "grad_norm": 0.6882632310567561, "learning_rate": 9.613076763914754e-05, "loss": 3.7542, "step": 4646 }, { "epoch": 0.2884722825749581, "grad_norm": 0.7000994117854219, "learning_rate": 9.615145872129113e-05, "loss": 3.7504, "step": 4647 }, { "epoch": 0.288534359674716, "grad_norm": 0.6620054810518936, "learning_rate": 9.617214980343471e-05, "loss": 3.6925, "step": 4648 }, { "epoch": 0.2885964367744739, "grad_norm": 0.6281382924661418, "learning_rate": 9.619284088557832e-05, "loss": 3.7557, "step": 4649 }, { "epoch": 0.2886585138742318, "grad_norm": 0.6018897318398786, "learning_rate": 9.621353196772192e-05, "loss": 3.6424, "step": 4650 }, { "epoch": 0.2887205909739897, "grad_norm": 0.7082695838814046, "learning_rate": 9.623422304986551e-05, "loss": 3.8601, "step": 4651 }, { "epoch": 0.2887826680737476, "grad_norm": 0.7347635499315504, "learning_rate": 9.62549141320091e-05, "loss": 3.8216, "step": 4652 }, { "epoch": 0.2888447451735055, "grad_norm": 0.7269244369265884, "learning_rate": 9.627560521415271e-05, "loss": 3.7926, "step": 4653 }, { "epoch": 0.2889068222732634, "grad_norm": 0.6306921911597233, "learning_rate": 9.62962962962963e-05, "loss": 3.7794, "step": 4654 }, { "epoch": 0.2889688993730213, "grad_norm": 0.7154836656690587, "learning_rate": 9.631698737843989e-05, "loss": 3.713, "step": 4655 }, { "epoch": 0.2890309764727792, "grad_norm": 0.6962823212071213, "learning_rate": 9.633767846058349e-05, "loss": 3.8169, "step": 4656 }, { "epoch": 0.2890930535725371, "grad_norm": 0.6472722349594833, "learning_rate": 9.63583695427271e-05, "loss": 3.7599, "step": 4657 }, { "epoch": 0.289155130672295, "grad_norm": 0.6052541044679844, "learning_rate": 9.637906062487069e-05, "loss": 3.8092, "step": 4658 }, { "epoch": 0.2892172077720529, "grad_norm": 0.7291898087961716, "learning_rate": 9.639975170701429e-05, "loss": 3.7976, "step": 4659 }, { "epoch": 0.2892792848718108, "grad_norm": 0.7730079768313346, "learning_rate": 9.642044278915787e-05, "loss": 3.8489, "step": 4660 }, { "epoch": 0.2893413619715687, "grad_norm": 0.8009195347920712, "learning_rate": 9.644113387130148e-05, "loss": 3.7689, "step": 4661 }, { "epoch": 0.2894034390713266, "grad_norm": 0.7941149615988276, "learning_rate": 9.646182495344507e-05, "loss": 3.8521, "step": 4662 }, { "epoch": 0.2894655161710845, "grad_norm": 0.8431667855373203, "learning_rate": 9.648251603558867e-05, "loss": 3.8908, "step": 4663 }, { "epoch": 0.2895275932708424, "grad_norm": 0.7330027807008632, "learning_rate": 9.650320711773226e-05, "loss": 3.8382, "step": 4664 }, { "epoch": 0.2895896703706003, "grad_norm": 0.77254692978264, "learning_rate": 9.652389819987586e-05, "loss": 3.8214, "step": 4665 }, { "epoch": 0.2896517474703582, "grad_norm": 0.8566490287733843, "learning_rate": 9.654458928201946e-05, "loss": 3.8309, "step": 4666 }, { "epoch": 0.2897138245701161, "grad_norm": 0.8962607509973766, "learning_rate": 9.656528036416305e-05, "loss": 3.8904, "step": 4667 }, { "epoch": 0.289775901669874, "grad_norm": 0.9716473291271168, "learning_rate": 9.658597144630664e-05, "loss": 3.7707, "step": 4668 }, { "epoch": 0.2898379787696319, "grad_norm": 0.9476953849039293, "learning_rate": 9.660666252845024e-05, "loss": 3.7005, "step": 4669 }, { "epoch": 0.28990005586938977, "grad_norm": 0.9977551834652348, "learning_rate": 9.662735361059384e-05, "loss": 3.8267, "step": 4670 }, { "epoch": 0.2899621329691477, "grad_norm": 0.9008076130432654, "learning_rate": 9.664804469273744e-05, "loss": 3.7749, "step": 4671 }, { "epoch": 0.2900242100689056, "grad_norm": 1.0910717461803832, "learning_rate": 9.666873577488103e-05, "loss": 3.7288, "step": 4672 }, { "epoch": 0.29008628716866347, "grad_norm": 0.9188003020237487, "learning_rate": 9.668942685702462e-05, "loss": 3.7865, "step": 4673 }, { "epoch": 0.2901483642684214, "grad_norm": 0.9398112734592352, "learning_rate": 9.671011793916823e-05, "loss": 3.7429, "step": 4674 }, { "epoch": 0.2902104413681793, "grad_norm": 0.8831856875615295, "learning_rate": 9.673080902131182e-05, "loss": 3.7561, "step": 4675 }, { "epoch": 0.29027251846793717, "grad_norm": 0.8541189893736868, "learning_rate": 9.675150010345542e-05, "loss": 3.7885, "step": 4676 }, { "epoch": 0.2903345955676951, "grad_norm": 0.6977101039284989, "learning_rate": 9.677219118559901e-05, "loss": 3.845, "step": 4677 }, { "epoch": 0.290396672667453, "grad_norm": 0.7943633645778992, "learning_rate": 9.679288226774262e-05, "loss": 3.7924, "step": 4678 }, { "epoch": 0.29045874976721087, "grad_norm": 0.6733341783889655, "learning_rate": 9.68135733498862e-05, "loss": 3.7595, "step": 4679 }, { "epoch": 0.2905208268669688, "grad_norm": 0.6327576083641857, "learning_rate": 9.68342644320298e-05, "loss": 3.7588, "step": 4680 }, { "epoch": 0.2905829039667267, "grad_norm": 0.6851626022896941, "learning_rate": 9.685495551417339e-05, "loss": 3.714, "step": 4681 }, { "epoch": 0.29064498106648456, "grad_norm": 0.6006373685819619, "learning_rate": 9.6875646596317e-05, "loss": 3.7293, "step": 4682 }, { "epoch": 0.2907070581662425, "grad_norm": 0.5779355744026277, "learning_rate": 9.68963376784606e-05, "loss": 3.7661, "step": 4683 }, { "epoch": 0.29076913526600034, "grad_norm": 0.6219357235652841, "learning_rate": 9.691702876060419e-05, "loss": 3.73, "step": 4684 }, { "epoch": 0.29083121236575826, "grad_norm": 0.6243948390890368, "learning_rate": 9.693771984274777e-05, "loss": 3.8649, "step": 4685 }, { "epoch": 0.2908932894655162, "grad_norm": 0.6356366329197629, "learning_rate": 9.695841092489138e-05, "loss": 3.7284, "step": 4686 }, { "epoch": 0.29095536656527404, "grad_norm": 0.5386742285441214, "learning_rate": 9.697910200703497e-05, "loss": 3.8032, "step": 4687 }, { "epoch": 0.29101744366503196, "grad_norm": 0.5572179644563447, "learning_rate": 9.699979308917857e-05, "loss": 3.9062, "step": 4688 }, { "epoch": 0.2910795207647899, "grad_norm": 0.5659138084436434, "learning_rate": 9.702048417132216e-05, "loss": 3.7526, "step": 4689 }, { "epoch": 0.29114159786454774, "grad_norm": 0.5717307966849619, "learning_rate": 9.704117525346576e-05, "loss": 3.7259, "step": 4690 }, { "epoch": 0.29120367496430566, "grad_norm": 0.5553724063209141, "learning_rate": 9.706186633560935e-05, "loss": 3.7838, "step": 4691 }, { "epoch": 0.2912657520640636, "grad_norm": 0.648311786644967, "learning_rate": 9.708255741775295e-05, "loss": 3.6477, "step": 4692 }, { "epoch": 0.29132782916382144, "grad_norm": 0.7251378109278186, "learning_rate": 9.710324849989655e-05, "loss": 3.7442, "step": 4693 }, { "epoch": 0.29138990626357936, "grad_norm": 0.6805241795419682, "learning_rate": 9.712393958204014e-05, "loss": 3.8024, "step": 4694 }, { "epoch": 0.2914519833633373, "grad_norm": 0.5615165909059503, "learning_rate": 9.714463066418375e-05, "loss": 3.7377, "step": 4695 }, { "epoch": 0.29151406046309514, "grad_norm": 0.6167787332989346, "learning_rate": 9.716532174632734e-05, "loss": 3.775, "step": 4696 }, { "epoch": 0.29157613756285305, "grad_norm": 0.7001855442530907, "learning_rate": 9.718601282847093e-05, "loss": 3.6959, "step": 4697 }, { "epoch": 0.291638214662611, "grad_norm": 0.7570929503123098, "learning_rate": 9.720670391061453e-05, "loss": 3.6952, "step": 4698 }, { "epoch": 0.29170029176236884, "grad_norm": 0.6422057768733853, "learning_rate": 9.722739499275813e-05, "loss": 3.7094, "step": 4699 }, { "epoch": 0.29176236886212675, "grad_norm": 0.6561670932170879, "learning_rate": 9.724808607490172e-05, "loss": 3.7987, "step": 4700 }, { "epoch": 0.29182444596188467, "grad_norm": 0.6978594604694388, "learning_rate": 9.726877715704532e-05, "loss": 3.7478, "step": 4701 }, { "epoch": 0.29188652306164253, "grad_norm": 0.7076660594230822, "learning_rate": 9.728946823918891e-05, "loss": 3.8633, "step": 4702 }, { "epoch": 0.29194860016140045, "grad_norm": 0.730366541601818, "learning_rate": 9.731015932133251e-05, "loss": 3.8433, "step": 4703 }, { "epoch": 0.29201067726115837, "grad_norm": 0.7683828469566606, "learning_rate": 9.73308504034761e-05, "loss": 3.8003, "step": 4704 }, { "epoch": 0.29207275436091623, "grad_norm": 0.720949663209271, "learning_rate": 9.73515414856197e-05, "loss": 3.7839, "step": 4705 }, { "epoch": 0.29213483146067415, "grad_norm": 0.7835571934700684, "learning_rate": 9.73722325677633e-05, "loss": 3.7352, "step": 4706 }, { "epoch": 0.29219690856043207, "grad_norm": 0.8810464190985363, "learning_rate": 9.73929236499069e-05, "loss": 3.7233, "step": 4707 }, { "epoch": 0.29225898566018993, "grad_norm": 0.8145457336320608, "learning_rate": 9.74136147320505e-05, "loss": 3.7557, "step": 4708 }, { "epoch": 0.29232106275994785, "grad_norm": 0.6916751231017465, "learning_rate": 9.743430581419408e-05, "loss": 3.7626, "step": 4709 }, { "epoch": 0.29238313985970577, "grad_norm": 0.7281829237921417, "learning_rate": 9.745499689633767e-05, "loss": 3.7681, "step": 4710 }, { "epoch": 0.2924452169594636, "grad_norm": 0.7028561183951606, "learning_rate": 9.747568797848128e-05, "loss": 3.7221, "step": 4711 }, { "epoch": 0.29250729405922155, "grad_norm": 0.7099148448059919, "learning_rate": 9.749637906062488e-05, "loss": 3.7273, "step": 4712 }, { "epoch": 0.29256937115897946, "grad_norm": 0.8199259619317918, "learning_rate": 9.751707014276847e-05, "loss": 3.6527, "step": 4713 }, { "epoch": 0.2926314482587373, "grad_norm": 0.7241765706475553, "learning_rate": 9.753776122491207e-05, "loss": 3.7234, "step": 4714 }, { "epoch": 0.29269352535849524, "grad_norm": 0.6197190335667289, "learning_rate": 9.755845230705566e-05, "loss": 3.6698, "step": 4715 }, { "epoch": 0.29275560245825316, "grad_norm": 0.6656279568315479, "learning_rate": 9.757914338919926e-05, "loss": 3.7254, "step": 4716 }, { "epoch": 0.292817679558011, "grad_norm": 0.6875092982617067, "learning_rate": 9.759983447134285e-05, "loss": 3.8195, "step": 4717 }, { "epoch": 0.29287975665776894, "grad_norm": 0.7786039555918263, "learning_rate": 9.762052555348645e-05, "loss": 3.7042, "step": 4718 }, { "epoch": 0.29294183375752686, "grad_norm": 0.7311669750846291, "learning_rate": 9.764121663563004e-05, "loss": 3.6961, "step": 4719 }, { "epoch": 0.2930039108572847, "grad_norm": 0.6790783443436968, "learning_rate": 9.766190771777365e-05, "loss": 3.7088, "step": 4720 }, { "epoch": 0.29306598795704264, "grad_norm": 0.6146337332237931, "learning_rate": 9.768259879991723e-05, "loss": 3.7448, "step": 4721 }, { "epoch": 0.29312806505680056, "grad_norm": 0.5912144882957512, "learning_rate": 9.770328988206083e-05, "loss": 3.7587, "step": 4722 }, { "epoch": 0.2931901421565584, "grad_norm": 0.7007181723964474, "learning_rate": 9.772398096420444e-05, "loss": 3.7327, "step": 4723 }, { "epoch": 0.29325221925631634, "grad_norm": 0.6799443954285039, "learning_rate": 9.774467204634803e-05, "loss": 3.8454, "step": 4724 }, { "epoch": 0.29331429635607426, "grad_norm": 0.6647489029572936, "learning_rate": 9.776536312849163e-05, "loss": 3.6223, "step": 4725 }, { "epoch": 0.2933763734558321, "grad_norm": 0.583342132047871, "learning_rate": 9.778605421063522e-05, "loss": 3.789, "step": 4726 }, { "epoch": 0.29343845055559004, "grad_norm": 0.6827576674174354, "learning_rate": 9.780674529277882e-05, "loss": 3.7252, "step": 4727 }, { "epoch": 0.29350052765534795, "grad_norm": 0.5810596369684815, "learning_rate": 9.782743637492241e-05, "loss": 3.6707, "step": 4728 }, { "epoch": 0.2935626047551058, "grad_norm": 0.5610820594617142, "learning_rate": 9.784812745706601e-05, "loss": 3.7617, "step": 4729 }, { "epoch": 0.29362468185486373, "grad_norm": 0.5880820130743352, "learning_rate": 9.78688185392096e-05, "loss": 3.759, "step": 4730 }, { "epoch": 0.29368675895462165, "grad_norm": 0.6286562073069428, "learning_rate": 9.78895096213532e-05, "loss": 3.8236, "step": 4731 }, { "epoch": 0.2937488360543795, "grad_norm": 0.6510386403331354, "learning_rate": 9.791020070349681e-05, "loss": 3.7026, "step": 4732 }, { "epoch": 0.29381091315413743, "grad_norm": 0.6903625196620086, "learning_rate": 9.793089178564039e-05, "loss": 3.8466, "step": 4733 }, { "epoch": 0.29387299025389535, "grad_norm": 0.6914869573870532, "learning_rate": 9.795158286778398e-05, "loss": 3.7528, "step": 4734 }, { "epoch": 0.2939350673536532, "grad_norm": 0.6828863570993475, "learning_rate": 9.797227394992758e-05, "loss": 3.7605, "step": 4735 }, { "epoch": 0.29399714445341113, "grad_norm": 0.6281919984528345, "learning_rate": 9.799296503207119e-05, "loss": 3.8362, "step": 4736 }, { "epoch": 0.29405922155316905, "grad_norm": 0.6779876367699261, "learning_rate": 9.801365611421478e-05, "loss": 3.7596, "step": 4737 }, { "epoch": 0.2941212986529269, "grad_norm": 0.7142891739012178, "learning_rate": 9.803434719635838e-05, "loss": 3.6805, "step": 4738 }, { "epoch": 0.29418337575268483, "grad_norm": 0.6545615728352242, "learning_rate": 9.805503827850196e-05, "loss": 3.6851, "step": 4739 }, { "epoch": 0.29424545285244275, "grad_norm": 0.5997951604943003, "learning_rate": 9.807572936064557e-05, "loss": 3.6434, "step": 4740 }, { "epoch": 0.2943075299522006, "grad_norm": 0.6455431075842, "learning_rate": 9.809642044278916e-05, "loss": 3.779, "step": 4741 }, { "epoch": 0.2943696070519585, "grad_norm": 0.6798705369761123, "learning_rate": 9.811711152493276e-05, "loss": 3.8058, "step": 4742 }, { "epoch": 0.29443168415171644, "grad_norm": 0.6737434223398925, "learning_rate": 9.813780260707635e-05, "loss": 3.765, "step": 4743 }, { "epoch": 0.2944937612514743, "grad_norm": 0.5872646560447812, "learning_rate": 9.815849368921995e-05, "loss": 3.8034, "step": 4744 }, { "epoch": 0.2945558383512322, "grad_norm": 0.5650656785178324, "learning_rate": 9.817918477136354e-05, "loss": 3.7183, "step": 4745 }, { "epoch": 0.29461791545099014, "grad_norm": 0.5544797750160692, "learning_rate": 9.819987585350714e-05, "loss": 3.732, "step": 4746 }, { "epoch": 0.294679992550748, "grad_norm": 0.5892206399665247, "learning_rate": 9.822056693565073e-05, "loss": 3.6518, "step": 4747 }, { "epoch": 0.2947420696505059, "grad_norm": 0.7299022824061622, "learning_rate": 9.824125801779434e-05, "loss": 3.7321, "step": 4748 }, { "epoch": 0.29480414675026384, "grad_norm": 0.7199672260375091, "learning_rate": 9.826194909993794e-05, "loss": 3.7205, "step": 4749 }, { "epoch": 0.2948662238500217, "grad_norm": 0.6921882732722869, "learning_rate": 9.828264018208153e-05, "loss": 3.8327, "step": 4750 }, { "epoch": 0.2949283009497796, "grad_norm": 0.7569928012843247, "learning_rate": 9.830333126422513e-05, "loss": 3.7149, "step": 4751 }, { "epoch": 0.29499037804953754, "grad_norm": 0.666114978707028, "learning_rate": 9.832402234636872e-05, "loss": 3.7565, "step": 4752 }, { "epoch": 0.2950524551492954, "grad_norm": 0.6456661751496954, "learning_rate": 9.834471342851232e-05, "loss": 3.7289, "step": 4753 }, { "epoch": 0.2951145322490533, "grad_norm": 0.6377406666445439, "learning_rate": 9.836540451065591e-05, "loss": 3.7267, "step": 4754 }, { "epoch": 0.29517660934881124, "grad_norm": 0.6440306366613943, "learning_rate": 9.838609559279951e-05, "loss": 3.6991, "step": 4755 }, { "epoch": 0.2952386864485691, "grad_norm": 0.6709997109015989, "learning_rate": 9.84067866749431e-05, "loss": 3.7139, "step": 4756 }, { "epoch": 0.295300763548327, "grad_norm": 0.6146165682540345, "learning_rate": 9.842747775708671e-05, "loss": 3.7612, "step": 4757 }, { "epoch": 0.29536284064808493, "grad_norm": 0.5938008939112522, "learning_rate": 9.844816883923029e-05, "loss": 3.7275, "step": 4758 }, { "epoch": 0.2954249177478428, "grad_norm": 0.6194205869571752, "learning_rate": 9.846885992137389e-05, "loss": 3.7344, "step": 4759 }, { "epoch": 0.2954869948476007, "grad_norm": 0.5775936881359802, "learning_rate": 9.848955100351748e-05, "loss": 3.6907, "step": 4760 }, { "epoch": 0.29554907194735863, "grad_norm": 0.6159533946378124, "learning_rate": 9.851024208566109e-05, "loss": 3.6842, "step": 4761 }, { "epoch": 0.2956111490471165, "grad_norm": 0.7439774862569569, "learning_rate": 9.853093316780469e-05, "loss": 3.6868, "step": 4762 }, { "epoch": 0.2956732261468744, "grad_norm": 0.7982928060970845, "learning_rate": 9.855162424994828e-05, "loss": 3.7864, "step": 4763 }, { "epoch": 0.29573530324663233, "grad_norm": 0.9056854791764134, "learning_rate": 9.857231533209186e-05, "loss": 3.7024, "step": 4764 }, { "epoch": 0.2957973803463902, "grad_norm": 0.7588120779820702, "learning_rate": 9.859300641423547e-05, "loss": 3.7887, "step": 4765 }, { "epoch": 0.2958594574461481, "grad_norm": 0.6761472295349904, "learning_rate": 9.861369749637907e-05, "loss": 3.7272, "step": 4766 }, { "epoch": 0.29592153454590603, "grad_norm": 0.7103681695070376, "learning_rate": 9.863438857852266e-05, "loss": 3.8069, "step": 4767 }, { "epoch": 0.2959836116456639, "grad_norm": 0.9707402120524596, "learning_rate": 9.865507966066626e-05, "loss": 3.7698, "step": 4768 }, { "epoch": 0.2960456887454218, "grad_norm": 1.0029146514627212, "learning_rate": 9.867577074280985e-05, "loss": 3.7272, "step": 4769 }, { "epoch": 0.2961077658451797, "grad_norm": 0.8625053267383562, "learning_rate": 9.869646182495345e-05, "loss": 3.6969, "step": 4770 }, { "epoch": 0.2961698429449376, "grad_norm": 0.8107741439428536, "learning_rate": 9.871715290709704e-05, "loss": 3.8572, "step": 4771 }, { "epoch": 0.2962319200446955, "grad_norm": 0.8387436870217986, "learning_rate": 9.873784398924064e-05, "loss": 3.6667, "step": 4772 }, { "epoch": 0.2962939971444534, "grad_norm": 0.892638490334284, "learning_rate": 9.875853507138425e-05, "loss": 3.8559, "step": 4773 }, { "epoch": 0.2963560742442113, "grad_norm": 0.8424045997736769, "learning_rate": 9.877922615352784e-05, "loss": 3.7856, "step": 4774 }, { "epoch": 0.2964181513439692, "grad_norm": 0.8844034830635562, "learning_rate": 9.879991723567144e-05, "loss": 3.6647, "step": 4775 }, { "epoch": 0.2964802284437271, "grad_norm": 0.8362803602765908, "learning_rate": 9.882060831781502e-05, "loss": 3.7188, "step": 4776 }, { "epoch": 0.296542305543485, "grad_norm": 0.7627476321358375, "learning_rate": 9.884129939995863e-05, "loss": 3.6903, "step": 4777 }, { "epoch": 0.2966043826432429, "grad_norm": 0.6426012446641667, "learning_rate": 9.886199048210222e-05, "loss": 3.8578, "step": 4778 }, { "epoch": 0.2966664597430008, "grad_norm": 0.6685718914184378, "learning_rate": 9.888268156424582e-05, "loss": 3.7174, "step": 4779 }, { "epoch": 0.2967285368427587, "grad_norm": 0.7287931348359022, "learning_rate": 9.890337264638941e-05, "loss": 3.7301, "step": 4780 }, { "epoch": 0.2967906139425166, "grad_norm": 0.6717863689780694, "learning_rate": 9.8924063728533e-05, "loss": 3.7618, "step": 4781 }, { "epoch": 0.2968526910422745, "grad_norm": 0.6610890631801479, "learning_rate": 9.89447548106766e-05, "loss": 3.7021, "step": 4782 }, { "epoch": 0.2969147681420324, "grad_norm": 0.6205057443710986, "learning_rate": 9.89654458928202e-05, "loss": 3.8357, "step": 4783 }, { "epoch": 0.2969768452417903, "grad_norm": 0.6154772566676929, "learning_rate": 9.898613697496379e-05, "loss": 3.705, "step": 4784 }, { "epoch": 0.2970389223415482, "grad_norm": 0.613198827562499, "learning_rate": 9.900682805710739e-05, "loss": 3.7175, "step": 4785 }, { "epoch": 0.2971009994413061, "grad_norm": 0.6340385145949788, "learning_rate": 9.9027519139251e-05, "loss": 3.7328, "step": 4786 }, { "epoch": 0.297163076541064, "grad_norm": 0.661793658847985, "learning_rate": 9.904821022139459e-05, "loss": 3.6355, "step": 4787 }, { "epoch": 0.2972251536408219, "grad_norm": 0.5895697355106436, "learning_rate": 9.906890130353817e-05, "loss": 3.7492, "step": 4788 }, { "epoch": 0.2972872307405798, "grad_norm": 0.6919174890471569, "learning_rate": 9.908959238568177e-05, "loss": 3.7046, "step": 4789 }, { "epoch": 0.2973493078403377, "grad_norm": 0.7631703556838317, "learning_rate": 9.911028346782538e-05, "loss": 3.8273, "step": 4790 }, { "epoch": 0.2974113849400956, "grad_norm": 0.9048619986531636, "learning_rate": 9.913097454996897e-05, "loss": 3.8831, "step": 4791 }, { "epoch": 0.2974734620398535, "grad_norm": 0.8268398173049208, "learning_rate": 9.915166563211257e-05, "loss": 3.7799, "step": 4792 }, { "epoch": 0.2975355391396114, "grad_norm": 0.7034817955953152, "learning_rate": 9.917235671425616e-05, "loss": 3.7063, "step": 4793 }, { "epoch": 0.2975976162393693, "grad_norm": 0.7229824670166805, "learning_rate": 9.919304779639976e-05, "loss": 3.792, "step": 4794 }, { "epoch": 0.2976596933391272, "grad_norm": 0.7142953071631272, "learning_rate": 9.921373887854335e-05, "loss": 3.8154, "step": 4795 }, { "epoch": 0.2977217704388851, "grad_norm": 0.7232892949531868, "learning_rate": 9.923442996068695e-05, "loss": 3.741, "step": 4796 }, { "epoch": 0.297783847538643, "grad_norm": 0.5824070683624156, "learning_rate": 9.925512104283054e-05, "loss": 3.6881, "step": 4797 }, { "epoch": 0.29784592463840087, "grad_norm": 0.5931158694980443, "learning_rate": 9.927581212497415e-05, "loss": 3.6304, "step": 4798 }, { "epoch": 0.2979080017381588, "grad_norm": 0.6493960729416232, "learning_rate": 9.929650320711775e-05, "loss": 3.6458, "step": 4799 }, { "epoch": 0.2979700788379167, "grad_norm": 0.6917670865487091, "learning_rate": 9.931719428926133e-05, "loss": 3.7783, "step": 4800 }, { "epoch": 0.29803215593767457, "grad_norm": 0.6864560828126206, "learning_rate": 9.933788537140492e-05, "loss": 3.7627, "step": 4801 }, { "epoch": 0.2980942330374325, "grad_norm": 0.7441596882298206, "learning_rate": 9.935857645354853e-05, "loss": 3.6405, "step": 4802 }, { "epoch": 0.2981563101371904, "grad_norm": 0.7472357273850945, "learning_rate": 9.937926753569213e-05, "loss": 3.783, "step": 4803 }, { "epoch": 0.29821838723694827, "grad_norm": 0.6109635847528044, "learning_rate": 9.939995861783572e-05, "loss": 3.7482, "step": 4804 }, { "epoch": 0.2982804643367062, "grad_norm": 0.7848581535814521, "learning_rate": 9.942064969997932e-05, "loss": 3.6332, "step": 4805 }, { "epoch": 0.2983425414364641, "grad_norm": 0.7437230337687895, "learning_rate": 9.944134078212291e-05, "loss": 3.7722, "step": 4806 }, { "epoch": 0.29840461853622197, "grad_norm": 0.7921721868897195, "learning_rate": 9.94620318642665e-05, "loss": 3.7948, "step": 4807 }, { "epoch": 0.2984666956359799, "grad_norm": 0.7919531646785706, "learning_rate": 9.94827229464101e-05, "loss": 3.773, "step": 4808 }, { "epoch": 0.2985287727357378, "grad_norm": 0.8312743778526206, "learning_rate": 9.95034140285537e-05, "loss": 3.7498, "step": 4809 }, { "epoch": 0.29859084983549566, "grad_norm": 0.8281107487441935, "learning_rate": 9.952410511069729e-05, "loss": 3.761, "step": 4810 }, { "epoch": 0.2986529269352536, "grad_norm": 0.7915548111714975, "learning_rate": 9.95447961928409e-05, "loss": 3.6919, "step": 4811 }, { "epoch": 0.2987150040350115, "grad_norm": 0.7278824578253144, "learning_rate": 9.956548727498448e-05, "loss": 3.8301, "step": 4812 }, { "epoch": 0.29877708113476936, "grad_norm": 0.7745822565099637, "learning_rate": 9.958617835712808e-05, "loss": 3.7612, "step": 4813 }, { "epoch": 0.2988391582345273, "grad_norm": 0.7378918059598601, "learning_rate": 9.960686943927167e-05, "loss": 3.7371, "step": 4814 }, { "epoch": 0.2989012353342852, "grad_norm": 0.6738879191087729, "learning_rate": 9.962756052141528e-05, "loss": 3.6699, "step": 4815 }, { "epoch": 0.29896331243404306, "grad_norm": 0.7422106689049177, "learning_rate": 9.964825160355887e-05, "loss": 3.8067, "step": 4816 }, { "epoch": 0.299025389533801, "grad_norm": 0.7207033731103486, "learning_rate": 9.966894268570247e-05, "loss": 3.7241, "step": 4817 }, { "epoch": 0.2990874666335589, "grad_norm": 0.6387136603168997, "learning_rate": 9.968963376784605e-05, "loss": 3.7787, "step": 4818 }, { "epoch": 0.29914954373331676, "grad_norm": 0.7037813725770988, "learning_rate": 9.971032484998966e-05, "loss": 3.724, "step": 4819 }, { "epoch": 0.2992116208330747, "grad_norm": 0.6380209215644205, "learning_rate": 9.973101593213326e-05, "loss": 3.8054, "step": 4820 }, { "epoch": 0.2992736979328326, "grad_norm": 0.7160121353655416, "learning_rate": 9.975170701427685e-05, "loss": 3.7191, "step": 4821 }, { "epoch": 0.29933577503259046, "grad_norm": 0.630302135853248, "learning_rate": 9.977239809642045e-05, "loss": 3.7444, "step": 4822 }, { "epoch": 0.2993978521323484, "grad_norm": 0.730613041462533, "learning_rate": 9.979308917856405e-05, "loss": 3.651, "step": 4823 }, { "epoch": 0.2994599292321063, "grad_norm": 0.6933016685864672, "learning_rate": 9.981378026070764e-05, "loss": 3.7974, "step": 4824 }, { "epoch": 0.29952200633186415, "grad_norm": 0.7387370564445109, "learning_rate": 9.983447134285123e-05, "loss": 3.7801, "step": 4825 }, { "epoch": 0.2995840834316221, "grad_norm": 0.7040168616025797, "learning_rate": 9.985516242499483e-05, "loss": 3.781, "step": 4826 }, { "epoch": 0.29964616053138, "grad_norm": 0.7314868706380816, "learning_rate": 9.987585350713843e-05, "loss": 3.7594, "step": 4827 }, { "epoch": 0.29970823763113785, "grad_norm": 0.7976674272462136, "learning_rate": 9.989654458928203e-05, "loss": 3.7902, "step": 4828 }, { "epoch": 0.29977031473089577, "grad_norm": 0.8356040574775564, "learning_rate": 9.991723567142562e-05, "loss": 3.8399, "step": 4829 }, { "epoch": 0.2998323918306537, "grad_norm": 0.6749969738433256, "learning_rate": 9.99379267535692e-05, "loss": 3.744, "step": 4830 }, { "epoch": 0.29989446893041155, "grad_norm": 0.5930073259393228, "learning_rate": 9.995861783571281e-05, "loss": 3.7186, "step": 4831 }, { "epoch": 0.29995654603016947, "grad_norm": 0.6331706740614088, "learning_rate": 9.997930891785641e-05, "loss": 3.7731, "step": 4832 }, { "epoch": 0.3000186231299274, "grad_norm": 0.6905799497177233, "learning_rate": 0.0001, "loss": 3.7311, "step": 4833 }, { "epoch": 0.30008070022968525, "grad_norm": 0.6860787917408895, "learning_rate": 9.999999986956894e-05, "loss": 3.6924, "step": 4834 }, { "epoch": 0.30014277732944317, "grad_norm": 0.6780665570226595, "learning_rate": 9.999999947827573e-05, "loss": 3.6645, "step": 4835 }, { "epoch": 0.3002048544292011, "grad_norm": 0.6743680256084305, "learning_rate": 9.99999988261204e-05, "loss": 3.7412, "step": 4836 }, { "epoch": 0.30026693152895895, "grad_norm": 0.7248284350662639, "learning_rate": 9.999999791310292e-05, "loss": 3.7025, "step": 4837 }, { "epoch": 0.30032900862871686, "grad_norm": 0.6730570707694642, "learning_rate": 9.99999967392233e-05, "loss": 3.6841, "step": 4838 }, { "epoch": 0.3003910857284748, "grad_norm": 0.5857354387922948, "learning_rate": 9.999999530448159e-05, "loss": 3.7745, "step": 4839 }, { "epoch": 0.30045316282823264, "grad_norm": 0.6455507249750819, "learning_rate": 9.999999360887775e-05, "loss": 3.7846, "step": 4840 }, { "epoch": 0.30051523992799056, "grad_norm": 0.6659324124167413, "learning_rate": 9.999999165241181e-05, "loss": 3.7554, "step": 4841 }, { "epoch": 0.3005773170277485, "grad_norm": 0.6510558749873743, "learning_rate": 9.999998943508378e-05, "loss": 3.6678, "step": 4842 }, { "epoch": 0.30063939412750634, "grad_norm": 0.5008840494083343, "learning_rate": 9.999998695689368e-05, "loss": 3.5514, "step": 4843 }, { "epoch": 0.30070147122726426, "grad_norm": 0.5410798237853879, "learning_rate": 9.999998421784149e-05, "loss": 3.6934, "step": 4844 }, { "epoch": 0.3007635483270222, "grad_norm": 0.6267923642171765, "learning_rate": 9.999998121792724e-05, "loss": 3.7661, "step": 4845 }, { "epoch": 0.30082562542678004, "grad_norm": 0.6980215937941379, "learning_rate": 9.999997795715096e-05, "loss": 3.71, "step": 4846 }, { "epoch": 0.30088770252653796, "grad_norm": 0.7509133023329287, "learning_rate": 9.999997443551266e-05, "loss": 3.8007, "step": 4847 }, { "epoch": 0.3009497796262959, "grad_norm": 0.6590822274957687, "learning_rate": 9.999997065301234e-05, "loss": 3.6686, "step": 4848 }, { "epoch": 0.30101185672605374, "grad_norm": 0.583486925635675, "learning_rate": 9.999996660965006e-05, "loss": 3.7117, "step": 4849 }, { "epoch": 0.30107393382581166, "grad_norm": 0.6247679518035839, "learning_rate": 9.999996230542579e-05, "loss": 3.7181, "step": 4850 }, { "epoch": 0.3011360109255696, "grad_norm": 0.597179946093091, "learning_rate": 9.99999577403396e-05, "loss": 3.7485, "step": 4851 }, { "epoch": 0.30119808802532744, "grad_norm": 0.7040079295297023, "learning_rate": 9.999995291439148e-05, "loss": 3.6963, "step": 4852 }, { "epoch": 0.30126016512508536, "grad_norm": 0.7706565955028128, "learning_rate": 9.999994782758147e-05, "loss": 3.6531, "step": 4853 }, { "epoch": 0.3013222422248433, "grad_norm": 0.7128810934187284, "learning_rate": 9.99999424799096e-05, "loss": 3.7291, "step": 4854 }, { "epoch": 0.30138431932460114, "grad_norm": 0.6950106662165041, "learning_rate": 9.999993687137588e-05, "loss": 3.7232, "step": 4855 }, { "epoch": 0.30144639642435905, "grad_norm": 0.6577407163474477, "learning_rate": 9.999993100198036e-05, "loss": 3.6337, "step": 4856 }, { "epoch": 0.30150847352411697, "grad_norm": 0.6665496251561407, "learning_rate": 9.999992487172307e-05, "loss": 3.6183, "step": 4857 }, { "epoch": 0.30157055062387483, "grad_norm": 0.7249735613639068, "learning_rate": 9.999991848060402e-05, "loss": 3.7736, "step": 4858 }, { "epoch": 0.30163262772363275, "grad_norm": 0.7509865629921748, "learning_rate": 9.999991182862326e-05, "loss": 3.5863, "step": 4859 }, { "epoch": 0.30169470482339067, "grad_norm": 0.7667372273717428, "learning_rate": 9.999990491578083e-05, "loss": 3.6957, "step": 4860 }, { "epoch": 0.30175678192314853, "grad_norm": 0.6701599899048888, "learning_rate": 9.999989774207674e-05, "loss": 3.7893, "step": 4861 }, { "epoch": 0.30181885902290645, "grad_norm": 0.6962723054514658, "learning_rate": 9.999989030751107e-05, "loss": 3.6934, "step": 4862 }, { "epoch": 0.30188093612266437, "grad_norm": 0.7851532513765228, "learning_rate": 9.999988261208382e-05, "loss": 3.7129, "step": 4863 }, { "epoch": 0.30194301322242223, "grad_norm": 0.7244367779512668, "learning_rate": 9.999987465579506e-05, "loss": 3.6807, "step": 4864 }, { "epoch": 0.30200509032218015, "grad_norm": 0.6886715227408544, "learning_rate": 9.999986643864478e-05, "loss": 3.7346, "step": 4865 }, { "epoch": 0.30206716742193807, "grad_norm": 0.6973728556951311, "learning_rate": 9.99998579606331e-05, "loss": 3.7259, "step": 4866 }, { "epoch": 0.3021292445216959, "grad_norm": 0.7243187828071265, "learning_rate": 9.999984922176e-05, "loss": 3.6273, "step": 4867 }, { "epoch": 0.30219132162145385, "grad_norm": 0.7038649374368419, "learning_rate": 9.999984022202556e-05, "loss": 3.7415, "step": 4868 }, { "epoch": 0.30225339872121176, "grad_norm": 0.6733592738058419, "learning_rate": 9.99998309614298e-05, "loss": 3.66, "step": 4869 }, { "epoch": 0.3023154758209696, "grad_norm": 0.5938992563672412, "learning_rate": 9.99998214399728e-05, "loss": 3.7372, "step": 4870 }, { "epoch": 0.30237755292072754, "grad_norm": 0.6422289563462656, "learning_rate": 9.999981165765458e-05, "loss": 3.6212, "step": 4871 }, { "epoch": 0.30243963002048546, "grad_norm": 0.530457494557319, "learning_rate": 9.999980161447521e-05, "loss": 3.7997, "step": 4872 }, { "epoch": 0.3025017071202433, "grad_norm": 0.5560131618910981, "learning_rate": 9.999979131043476e-05, "loss": 3.8109, "step": 4873 }, { "epoch": 0.30256378422000124, "grad_norm": 0.7855098933805509, "learning_rate": 9.999978074553322e-05, "loss": 3.7599, "step": 4874 }, { "epoch": 0.30262586131975916, "grad_norm": 0.7836474825234794, "learning_rate": 9.999976991977071e-05, "loss": 3.688, "step": 4875 }, { "epoch": 0.302687938419517, "grad_norm": 0.7811244100714043, "learning_rate": 9.999975883314727e-05, "loss": 3.7161, "step": 4876 }, { "epoch": 0.30275001551927494, "grad_norm": 0.7066131746728843, "learning_rate": 9.999974748566293e-05, "loss": 3.6658, "step": 4877 }, { "epoch": 0.30281209261903286, "grad_norm": 0.7152997422986744, "learning_rate": 9.999973587731778e-05, "loss": 3.7343, "step": 4878 }, { "epoch": 0.3028741697187907, "grad_norm": 0.8311187683574088, "learning_rate": 9.999972400811186e-05, "loss": 3.7207, "step": 4879 }, { "epoch": 0.30293624681854864, "grad_norm": 0.8191505195045559, "learning_rate": 9.999971187804527e-05, "loss": 3.7164, "step": 4880 }, { "epoch": 0.30299832391830656, "grad_norm": 0.7592031951711445, "learning_rate": 9.999969948711801e-05, "loss": 3.7682, "step": 4881 }, { "epoch": 0.3030604010180644, "grad_norm": 0.7221924814393303, "learning_rate": 9.99996868353302e-05, "loss": 3.6798, "step": 4882 }, { "epoch": 0.30312247811782234, "grad_norm": 0.7228813027152152, "learning_rate": 9.999967392268188e-05, "loss": 3.7191, "step": 4883 }, { "epoch": 0.30318455521758025, "grad_norm": 0.7342719553625314, "learning_rate": 9.999966074917312e-05, "loss": 3.8785, "step": 4884 }, { "epoch": 0.3032466323173381, "grad_norm": 0.6978516026434686, "learning_rate": 9.9999647314804e-05, "loss": 3.7321, "step": 4885 }, { "epoch": 0.30330870941709603, "grad_norm": 0.6329652578332411, "learning_rate": 9.999963361957459e-05, "loss": 3.7269, "step": 4886 }, { "epoch": 0.30337078651685395, "grad_norm": 0.5998328969638881, "learning_rate": 9.999961966348493e-05, "loss": 3.6303, "step": 4887 }, { "epoch": 0.3034328636166118, "grad_norm": 0.587629672207381, "learning_rate": 9.999960544653513e-05, "loss": 3.6735, "step": 4888 }, { "epoch": 0.30349494071636973, "grad_norm": 0.6170626163890184, "learning_rate": 9.999959096872525e-05, "loss": 3.6059, "step": 4889 }, { "epoch": 0.30355701781612765, "grad_norm": 0.6074740274002765, "learning_rate": 9.999957623005537e-05, "loss": 3.6362, "step": 4890 }, { "epoch": 0.3036190949158855, "grad_norm": 0.5408222114416683, "learning_rate": 9.999956123052556e-05, "loss": 3.5989, "step": 4891 }, { "epoch": 0.30368117201564343, "grad_norm": 0.5059068281974035, "learning_rate": 9.99995459701359e-05, "loss": 3.653, "step": 4892 }, { "epoch": 0.30374324911540135, "grad_norm": 0.6089741048381188, "learning_rate": 9.999953044888646e-05, "loss": 3.6956, "step": 4893 }, { "epoch": 0.3038053262151592, "grad_norm": 0.5764471398707387, "learning_rate": 9.999951466677735e-05, "loss": 3.6775, "step": 4894 }, { "epoch": 0.30386740331491713, "grad_norm": 0.6045633006134458, "learning_rate": 9.999949862380863e-05, "loss": 3.7085, "step": 4895 }, { "epoch": 0.30392948041467505, "grad_norm": 0.5567236587432199, "learning_rate": 9.999948231998037e-05, "loss": 3.7131, "step": 4896 }, { "epoch": 0.3039915575144329, "grad_norm": 0.5488717547142434, "learning_rate": 9.99994657552927e-05, "loss": 3.7211, "step": 4897 }, { "epoch": 0.3040536346141908, "grad_norm": 0.5521546683619403, "learning_rate": 9.999944892974566e-05, "loss": 3.7029, "step": 4898 }, { "epoch": 0.30411571171394874, "grad_norm": 0.6618882431708195, "learning_rate": 9.999943184333937e-05, "loss": 3.741, "step": 4899 }, { "epoch": 0.3041777888137066, "grad_norm": 0.7468035053206664, "learning_rate": 9.99994144960739e-05, "loss": 3.6707, "step": 4900 }, { "epoch": 0.3042398659134645, "grad_norm": 0.7571453028251242, "learning_rate": 9.999939688794935e-05, "loss": 3.6917, "step": 4901 }, { "epoch": 0.30430194301322244, "grad_norm": 0.6156588104568121, "learning_rate": 9.99993790189658e-05, "loss": 3.7005, "step": 4902 }, { "epoch": 0.3043640201129803, "grad_norm": 0.6248046092742556, "learning_rate": 9.999936088912336e-05, "loss": 3.7216, "step": 4903 }, { "epoch": 0.3044260972127382, "grad_norm": 0.6469250931440595, "learning_rate": 9.999934249842211e-05, "loss": 3.7337, "step": 4904 }, { "epoch": 0.30448817431249614, "grad_norm": 0.7041361849204795, "learning_rate": 9.999932384686217e-05, "loss": 3.7097, "step": 4905 }, { "epoch": 0.304550251412254, "grad_norm": 0.6447483968518825, "learning_rate": 9.999930493444361e-05, "loss": 3.6729, "step": 4906 }, { "epoch": 0.3046123285120119, "grad_norm": 0.630248893399385, "learning_rate": 9.999928576116652e-05, "loss": 3.7828, "step": 4907 }, { "epoch": 0.30467440561176984, "grad_norm": 0.6495580757333261, "learning_rate": 9.999926632703104e-05, "loss": 3.7784, "step": 4908 }, { "epoch": 0.3047364827115277, "grad_norm": 0.6514248800964866, "learning_rate": 9.999924663203724e-05, "loss": 3.734, "step": 4909 }, { "epoch": 0.3047985598112856, "grad_norm": 0.6159476127162066, "learning_rate": 9.999922667618524e-05, "loss": 3.7778, "step": 4910 }, { "epoch": 0.30486063691104354, "grad_norm": 0.6029717917858621, "learning_rate": 9.999920645947514e-05, "loss": 3.6926, "step": 4911 }, { "epoch": 0.3049227140108014, "grad_norm": 0.6249264435399049, "learning_rate": 9.999918598190703e-05, "loss": 3.6486, "step": 4912 }, { "epoch": 0.3049847911105593, "grad_norm": 0.7411083607818811, "learning_rate": 9.999916524348102e-05, "loss": 3.693, "step": 4913 }, { "epoch": 0.30504686821031723, "grad_norm": 0.8386188179364807, "learning_rate": 9.999914424419725e-05, "loss": 3.742, "step": 4914 }, { "epoch": 0.3051089453100751, "grad_norm": 0.9060019778531109, "learning_rate": 9.999912298405579e-05, "loss": 3.6251, "step": 4915 }, { "epoch": 0.305171022409833, "grad_norm": 0.7702978740177978, "learning_rate": 9.999910146305678e-05, "loss": 3.8245, "step": 4916 }, { "epoch": 0.30523309950959093, "grad_norm": 0.7103034744998336, "learning_rate": 9.999907968120032e-05, "loss": 3.5851, "step": 4917 }, { "epoch": 0.3052951766093488, "grad_norm": 0.7571805304861073, "learning_rate": 9.999905763848652e-05, "loss": 3.783, "step": 4918 }, { "epoch": 0.3053572537091067, "grad_norm": 0.7830571445254856, "learning_rate": 9.99990353349155e-05, "loss": 3.5547, "step": 4919 }, { "epoch": 0.30541933080886463, "grad_norm": 0.7750377092108826, "learning_rate": 9.999901277048736e-05, "loss": 3.6585, "step": 4920 }, { "epoch": 0.3054814079086225, "grad_norm": 0.6557925310089402, "learning_rate": 9.999898994520224e-05, "loss": 3.7756, "step": 4921 }, { "epoch": 0.3055434850083804, "grad_norm": 0.7097935538289704, "learning_rate": 9.999896685906028e-05, "loss": 3.7565, "step": 4922 }, { "epoch": 0.30560556210813833, "grad_norm": 0.6619098779135786, "learning_rate": 9.999894351206154e-05, "loss": 3.5876, "step": 4923 }, { "epoch": 0.3056676392078962, "grad_norm": 0.7198540459491284, "learning_rate": 9.999891990420618e-05, "loss": 3.7751, "step": 4924 }, { "epoch": 0.3057297163076541, "grad_norm": 0.769029870300159, "learning_rate": 9.999889603549431e-05, "loss": 3.7984, "step": 4925 }, { "epoch": 0.305791793407412, "grad_norm": 0.6693604009836773, "learning_rate": 9.999887190592607e-05, "loss": 3.6477, "step": 4926 }, { "epoch": 0.3058538705071699, "grad_norm": 0.7054209206083512, "learning_rate": 9.999884751550159e-05, "loss": 3.6689, "step": 4927 }, { "epoch": 0.3059159476069278, "grad_norm": 0.596736724175269, "learning_rate": 9.999882286422098e-05, "loss": 3.6394, "step": 4928 }, { "epoch": 0.3059780247066857, "grad_norm": 0.6022323990352384, "learning_rate": 9.999879795208436e-05, "loss": 3.6311, "step": 4929 }, { "epoch": 0.3060401018064436, "grad_norm": 0.7011160230573688, "learning_rate": 9.999877277909188e-05, "loss": 3.6598, "step": 4930 }, { "epoch": 0.3061021789062015, "grad_norm": 0.6350698358325877, "learning_rate": 9.999874734524367e-05, "loss": 3.6446, "step": 4931 }, { "epoch": 0.3061642560059594, "grad_norm": 0.6843683704072533, "learning_rate": 9.999872165053985e-05, "loss": 3.6477, "step": 4932 }, { "epoch": 0.3062263331057173, "grad_norm": 0.572803257785955, "learning_rate": 9.999869569498057e-05, "loss": 3.6458, "step": 4933 }, { "epoch": 0.3062884102054752, "grad_norm": 0.5986297009933284, "learning_rate": 9.999866947856596e-05, "loss": 3.6958, "step": 4934 }, { "epoch": 0.3063504873052331, "grad_norm": 0.6280638460973754, "learning_rate": 9.999864300129613e-05, "loss": 3.7531, "step": 4935 }, { "epoch": 0.306412564404991, "grad_norm": 0.5928167738933867, "learning_rate": 9.999861626317126e-05, "loss": 3.5695, "step": 4936 }, { "epoch": 0.3064746415047489, "grad_norm": 0.5557518003145437, "learning_rate": 9.999858926419148e-05, "loss": 3.6459, "step": 4937 }, { "epoch": 0.3065367186045068, "grad_norm": 0.6173327552706537, "learning_rate": 9.999856200435691e-05, "loss": 3.5698, "step": 4938 }, { "epoch": 0.3065987957042647, "grad_norm": 0.6466275428295744, "learning_rate": 9.999853448366769e-05, "loss": 3.6384, "step": 4939 }, { "epoch": 0.3066608728040226, "grad_norm": 0.5865036327512975, "learning_rate": 9.999850670212401e-05, "loss": 3.6894, "step": 4940 }, { "epoch": 0.3067229499037805, "grad_norm": 0.544014561626542, "learning_rate": 9.999847865972596e-05, "loss": 3.7251, "step": 4941 }, { "epoch": 0.3067850270035384, "grad_norm": 0.6913106388859864, "learning_rate": 9.999845035647371e-05, "loss": 3.6557, "step": 4942 }, { "epoch": 0.3068471041032963, "grad_norm": 0.6460847800945627, "learning_rate": 9.999842179236742e-05, "loss": 3.724, "step": 4943 }, { "epoch": 0.3069091812030542, "grad_norm": 0.6070073912263239, "learning_rate": 9.999839296740722e-05, "loss": 3.7736, "step": 4944 }, { "epoch": 0.3069712583028121, "grad_norm": 0.6160922654105248, "learning_rate": 9.999836388159327e-05, "loss": 3.639, "step": 4945 }, { "epoch": 0.30703333540257, "grad_norm": 0.6329594194322012, "learning_rate": 9.999833453492571e-05, "loss": 3.7301, "step": 4946 }, { "epoch": 0.3070954125023279, "grad_norm": 0.6505653902609532, "learning_rate": 9.99983049274047e-05, "loss": 3.6759, "step": 4947 }, { "epoch": 0.3071574896020858, "grad_norm": 0.7528846420488954, "learning_rate": 9.99982750590304e-05, "loss": 3.7321, "step": 4948 }, { "epoch": 0.3072195667018437, "grad_norm": 0.7991216069135588, "learning_rate": 9.999824492980296e-05, "loss": 3.6746, "step": 4949 }, { "epoch": 0.3072816438016016, "grad_norm": 0.6826938482413883, "learning_rate": 9.999821453972255e-05, "loss": 3.6535, "step": 4950 }, { "epoch": 0.3073437209013595, "grad_norm": 0.5752231351538436, "learning_rate": 9.99981838887893e-05, "loss": 3.7582, "step": 4951 }, { "epoch": 0.3074057980011174, "grad_norm": 0.6360979154371226, "learning_rate": 9.999815297700339e-05, "loss": 3.6718, "step": 4952 }, { "epoch": 0.3074678751008753, "grad_norm": 0.6144933688595138, "learning_rate": 9.999812180436499e-05, "loss": 3.78, "step": 4953 }, { "epoch": 0.30752995220063317, "grad_norm": 0.5679572869026903, "learning_rate": 9.999809037087424e-05, "loss": 3.7064, "step": 4954 }, { "epoch": 0.3075920293003911, "grad_norm": 0.6583147039272438, "learning_rate": 9.999805867653133e-05, "loss": 3.6307, "step": 4955 }, { "epoch": 0.307654106400149, "grad_norm": 0.6933246872512595, "learning_rate": 9.99980267213364e-05, "loss": 3.723, "step": 4956 }, { "epoch": 0.30771618349990687, "grad_norm": 0.8393632090657138, "learning_rate": 9.999799450528961e-05, "loss": 3.7427, "step": 4957 }, { "epoch": 0.3077782605996648, "grad_norm": 0.7797055809967084, "learning_rate": 9.999796202839118e-05, "loss": 3.7251, "step": 4958 }, { "epoch": 0.3078403376994227, "grad_norm": 0.6548693074450249, "learning_rate": 9.999792929064122e-05, "loss": 3.6511, "step": 4959 }, { "epoch": 0.30790241479918057, "grad_norm": 0.6786454422999144, "learning_rate": 9.999789629203993e-05, "loss": 3.6634, "step": 4960 }, { "epoch": 0.3079644918989385, "grad_norm": 0.7799813842385974, "learning_rate": 9.999786303258749e-05, "loss": 3.6831, "step": 4961 }, { "epoch": 0.3080265689986964, "grad_norm": 0.7395516823098589, "learning_rate": 9.999782951228404e-05, "loss": 3.7077, "step": 4962 }, { "epoch": 0.30808864609845427, "grad_norm": 0.7283357290742115, "learning_rate": 9.999779573112979e-05, "loss": 3.7143, "step": 4963 }, { "epoch": 0.3081507231982122, "grad_norm": 0.7581447777365791, "learning_rate": 9.99977616891249e-05, "loss": 3.7515, "step": 4964 }, { "epoch": 0.3082128002979701, "grad_norm": 0.7835684896792293, "learning_rate": 9.999772738626956e-05, "loss": 3.6448, "step": 4965 }, { "epoch": 0.30827487739772796, "grad_norm": 0.7052985398952288, "learning_rate": 9.999769282256393e-05, "loss": 3.7758, "step": 4966 }, { "epoch": 0.3083369544974859, "grad_norm": 0.6004081400870962, "learning_rate": 9.999765799800818e-05, "loss": 3.8125, "step": 4967 }, { "epoch": 0.3083990315972438, "grad_norm": 0.7748446908811629, "learning_rate": 9.999762291260253e-05, "loss": 3.6744, "step": 4968 }, { "epoch": 0.30846110869700166, "grad_norm": 0.7587807359079015, "learning_rate": 9.999758756634714e-05, "loss": 3.6655, "step": 4969 }, { "epoch": 0.3085231857967596, "grad_norm": 0.8067019164512211, "learning_rate": 9.999755195924221e-05, "loss": 3.7099, "step": 4970 }, { "epoch": 0.3085852628965175, "grad_norm": 0.8125513867487584, "learning_rate": 9.999751609128791e-05, "loss": 3.7248, "step": 4971 }, { "epoch": 0.30864733999627536, "grad_norm": 0.6605859294875792, "learning_rate": 9.999747996248441e-05, "loss": 3.678, "step": 4972 }, { "epoch": 0.3087094170960333, "grad_norm": 0.6458311670533632, "learning_rate": 9.999744357283193e-05, "loss": 3.7048, "step": 4973 }, { "epoch": 0.3087714941957912, "grad_norm": 0.6480679435593332, "learning_rate": 9.999740692233067e-05, "loss": 3.7049, "step": 4974 }, { "epoch": 0.30883357129554906, "grad_norm": 0.6056550262377398, "learning_rate": 9.999737001098078e-05, "loss": 3.7017, "step": 4975 }, { "epoch": 0.308895648395307, "grad_norm": 0.6468623920262414, "learning_rate": 9.99973328387825e-05, "loss": 3.691, "step": 4976 }, { "epoch": 0.3089577254950649, "grad_norm": 0.6874098285148876, "learning_rate": 9.999729540573596e-05, "loss": 3.7388, "step": 4977 }, { "epoch": 0.30901980259482276, "grad_norm": 0.5895794156172388, "learning_rate": 9.999725771184141e-05, "loss": 3.5744, "step": 4978 }, { "epoch": 0.3090818796945807, "grad_norm": 0.6065332960107739, "learning_rate": 9.999721975709903e-05, "loss": 3.5915, "step": 4979 }, { "epoch": 0.3091439567943386, "grad_norm": 0.6451072200105603, "learning_rate": 9.999718154150901e-05, "loss": 3.6656, "step": 4980 }, { "epoch": 0.30920603389409645, "grad_norm": 0.6470556817673332, "learning_rate": 9.999714306507157e-05, "loss": 3.646, "step": 4981 }, { "epoch": 0.3092681109938544, "grad_norm": 0.6225629978398145, "learning_rate": 9.999710432778689e-05, "loss": 3.6354, "step": 4982 }, { "epoch": 0.3093301880936123, "grad_norm": 0.6699258120229271, "learning_rate": 9.999706532965518e-05, "loss": 3.588, "step": 4983 }, { "epoch": 0.30939226519337015, "grad_norm": 0.7159649753501858, "learning_rate": 9.999702607067665e-05, "loss": 3.6946, "step": 4984 }, { "epoch": 0.30945434229312807, "grad_norm": 0.7224854963490218, "learning_rate": 9.999698655085148e-05, "loss": 3.6657, "step": 4985 }, { "epoch": 0.309516419392886, "grad_norm": 0.5745381096499649, "learning_rate": 9.99969467701799e-05, "loss": 3.5765, "step": 4986 }, { "epoch": 0.30957849649264385, "grad_norm": 0.6475871151606019, "learning_rate": 9.999690672866212e-05, "loss": 3.6746, "step": 4987 }, { "epoch": 0.30964057359240177, "grad_norm": 0.8161753277702397, "learning_rate": 9.999686642629832e-05, "loss": 3.7188, "step": 4988 }, { "epoch": 0.3097026506921597, "grad_norm": 0.8046482039550845, "learning_rate": 9.999682586308875e-05, "loss": 3.5727, "step": 4989 }, { "epoch": 0.30976472779191755, "grad_norm": 0.7390723052392598, "learning_rate": 9.999678503903361e-05, "loss": 3.7284, "step": 4990 }, { "epoch": 0.30982680489167547, "grad_norm": 0.6501105911368565, "learning_rate": 9.99967439541331e-05, "loss": 3.7816, "step": 4991 }, { "epoch": 0.3098888819914334, "grad_norm": 0.7976876023843997, "learning_rate": 9.999670260838743e-05, "loss": 3.7784, "step": 4992 }, { "epoch": 0.30995095909119125, "grad_norm": 0.8308495740218275, "learning_rate": 9.999666100179683e-05, "loss": 3.6889, "step": 4993 }, { "epoch": 0.31001303619094916, "grad_norm": 0.7412949264446403, "learning_rate": 9.999661913436152e-05, "loss": 3.6364, "step": 4994 }, { "epoch": 0.3100751132907071, "grad_norm": 0.681534286117945, "learning_rate": 9.999657700608169e-05, "loss": 3.7128, "step": 4995 }, { "epoch": 0.31013719039046495, "grad_norm": 0.699361498875394, "learning_rate": 9.999653461695759e-05, "loss": 3.7544, "step": 4996 }, { "epoch": 0.31019926749022286, "grad_norm": 0.6942123753823595, "learning_rate": 9.999649196698944e-05, "loss": 3.7029, "step": 4997 }, { "epoch": 0.3102613445899808, "grad_norm": 0.693851350134212, "learning_rate": 9.999644905617745e-05, "loss": 3.7692, "step": 4998 }, { "epoch": 0.31032342168973864, "grad_norm": 0.632016776618228, "learning_rate": 9.999640588452184e-05, "loss": 3.7266, "step": 4999 }, { "epoch": 0.31038549878949656, "grad_norm": 0.5991100754847798, "learning_rate": 9.999636245202286e-05, "loss": 3.6791, "step": 5000 }, { "epoch": 0.3104475758892545, "grad_norm": 0.6368741974905122, "learning_rate": 9.999631875868072e-05, "loss": 3.7166, "step": 5001 }, { "epoch": 0.31050965298901234, "grad_norm": 0.6259650647421173, "learning_rate": 9.999627480449563e-05, "loss": 3.69, "step": 5002 }, { "epoch": 0.31057173008877026, "grad_norm": 0.5470284927765027, "learning_rate": 9.999623058946786e-05, "loss": 3.8069, "step": 5003 }, { "epoch": 0.3106338071885282, "grad_norm": 0.6520374972612984, "learning_rate": 9.99961861135976e-05, "loss": 3.7447, "step": 5004 }, { "epoch": 0.31069588428828604, "grad_norm": 0.6484526896708412, "learning_rate": 9.99961413768851e-05, "loss": 3.5982, "step": 5005 }, { "epoch": 0.31075796138804396, "grad_norm": 0.5979410378100352, "learning_rate": 9.999609637933059e-05, "loss": 3.6195, "step": 5006 }, { "epoch": 0.3108200384878019, "grad_norm": 0.6779772628316545, "learning_rate": 9.999605112093433e-05, "loss": 3.5818, "step": 5007 }, { "epoch": 0.31088211558755974, "grad_norm": 0.6284424629565561, "learning_rate": 9.999600560169651e-05, "loss": 3.7035, "step": 5008 }, { "epoch": 0.31094419268731766, "grad_norm": 0.6986238943484709, "learning_rate": 9.99959598216174e-05, "loss": 3.5811, "step": 5009 }, { "epoch": 0.3110062697870756, "grad_norm": 0.63541107227385, "learning_rate": 9.999591378069723e-05, "loss": 3.6386, "step": 5010 }, { "epoch": 0.31106834688683344, "grad_norm": 0.612554666454092, "learning_rate": 9.999586747893625e-05, "loss": 3.6807, "step": 5011 }, { "epoch": 0.31113042398659135, "grad_norm": 0.6006683789247177, "learning_rate": 9.999582091633469e-05, "loss": 3.5659, "step": 5012 }, { "epoch": 0.31119250108634927, "grad_norm": 0.5329752005234544, "learning_rate": 9.99957740928928e-05, "loss": 3.7062, "step": 5013 }, { "epoch": 0.31125457818610713, "grad_norm": 0.5515252357408744, "learning_rate": 9.999572700861079e-05, "loss": 3.6915, "step": 5014 }, { "epoch": 0.31131665528586505, "grad_norm": 0.5761766336535481, "learning_rate": 9.999567966348898e-05, "loss": 3.7131, "step": 5015 }, { "epoch": 0.31137873238562297, "grad_norm": 0.6666781354950817, "learning_rate": 9.999563205752754e-05, "loss": 3.5847, "step": 5016 }, { "epoch": 0.31144080948538083, "grad_norm": 0.6463681488205169, "learning_rate": 9.999558419072677e-05, "loss": 3.6655, "step": 5017 }, { "epoch": 0.31150288658513875, "grad_norm": 0.6301340947417121, "learning_rate": 9.999553606308689e-05, "loss": 3.604, "step": 5018 }, { "epoch": 0.31156496368489667, "grad_norm": 0.6207357130384561, "learning_rate": 9.999548767460816e-05, "loss": 3.649, "step": 5019 }, { "epoch": 0.31162704078465453, "grad_norm": 0.6111840514040794, "learning_rate": 9.999543902529083e-05, "loss": 3.6744, "step": 5020 }, { "epoch": 0.31168911788441245, "grad_norm": 0.6000810935929498, "learning_rate": 9.999539011513516e-05, "loss": 3.66, "step": 5021 }, { "epoch": 0.31175119498417037, "grad_norm": 0.577897955154482, "learning_rate": 9.99953409441414e-05, "loss": 3.5956, "step": 5022 }, { "epoch": 0.31181327208392823, "grad_norm": 0.6504070749579566, "learning_rate": 9.999529151230982e-05, "loss": 3.6677, "step": 5023 }, { "epoch": 0.31187534918368615, "grad_norm": 0.7748249594611386, "learning_rate": 9.999524181964067e-05, "loss": 3.6869, "step": 5024 }, { "epoch": 0.31193742628344406, "grad_norm": 0.6399199397332956, "learning_rate": 9.99951918661342e-05, "loss": 3.7479, "step": 5025 }, { "epoch": 0.3119995033832019, "grad_norm": 0.6245007428240424, "learning_rate": 9.999514165179067e-05, "loss": 3.6571, "step": 5026 }, { "epoch": 0.31206158048295984, "grad_norm": 0.6901276182592438, "learning_rate": 9.999509117661036e-05, "loss": 3.6065, "step": 5027 }, { "epoch": 0.31212365758271776, "grad_norm": 0.67834818634749, "learning_rate": 9.999504044059351e-05, "loss": 3.7804, "step": 5028 }, { "epoch": 0.3121857346824756, "grad_norm": 0.665256391912443, "learning_rate": 9.999498944374042e-05, "loss": 3.6406, "step": 5029 }, { "epoch": 0.31224781178223354, "grad_norm": 0.5528965095529204, "learning_rate": 9.999493818605133e-05, "loss": 3.7054, "step": 5030 }, { "epoch": 0.31230988888199146, "grad_norm": 0.6106114436961145, "learning_rate": 9.999488666752649e-05, "loss": 3.7137, "step": 5031 }, { "epoch": 0.3123719659817493, "grad_norm": 0.682575644782984, "learning_rate": 9.99948348881662e-05, "loss": 3.7082, "step": 5032 }, { "epoch": 0.31243404308150724, "grad_norm": 0.5634980883527823, "learning_rate": 9.999478284797073e-05, "loss": 3.6365, "step": 5033 }, { "epoch": 0.31249612018126516, "grad_norm": 0.5199279131348847, "learning_rate": 9.999473054694032e-05, "loss": 3.6724, "step": 5034 }, { "epoch": 0.312558197281023, "grad_norm": 0.49785131467182253, "learning_rate": 9.999467798507527e-05, "loss": 3.6121, "step": 5035 }, { "epoch": 0.31262027438078094, "grad_norm": 0.4810496284692779, "learning_rate": 9.999462516237585e-05, "loss": 3.6204, "step": 5036 }, { "epoch": 0.31268235148053886, "grad_norm": 0.501103583995599, "learning_rate": 9.999457207884235e-05, "loss": 3.6664, "step": 5037 }, { "epoch": 0.3127444285802967, "grad_norm": 0.5121302515309532, "learning_rate": 9.9994518734475e-05, "loss": 3.5821, "step": 5038 }, { "epoch": 0.31280650568005464, "grad_norm": 0.5546869627184348, "learning_rate": 9.999446512927415e-05, "loss": 3.6749, "step": 5039 }, { "epoch": 0.31286858277981255, "grad_norm": 0.5368838859950709, "learning_rate": 9.999441126324001e-05, "loss": 3.5718, "step": 5040 }, { "epoch": 0.3129306598795704, "grad_norm": 0.5573895686303759, "learning_rate": 9.99943571363729e-05, "loss": 3.6749, "step": 5041 }, { "epoch": 0.31299273697932833, "grad_norm": 0.6512011955630437, "learning_rate": 9.999430274867309e-05, "loss": 3.6133, "step": 5042 }, { "epoch": 0.31305481407908625, "grad_norm": 0.6750323826474971, "learning_rate": 9.999424810014086e-05, "loss": 3.6364, "step": 5043 }, { "epoch": 0.3131168911788441, "grad_norm": 0.6242556669660129, "learning_rate": 9.999419319077649e-05, "loss": 3.7033, "step": 5044 }, { "epoch": 0.31317896827860203, "grad_norm": 0.5829136962578626, "learning_rate": 9.999413802058031e-05, "loss": 3.7075, "step": 5045 }, { "epoch": 0.31324104537835995, "grad_norm": 0.5960394238355675, "learning_rate": 9.999408258955257e-05, "loss": 3.669, "step": 5046 }, { "epoch": 0.3133031224781178, "grad_norm": 0.649007317075641, "learning_rate": 9.999402689769354e-05, "loss": 3.7479, "step": 5047 }, { "epoch": 0.31336519957787573, "grad_norm": 0.7157093081724077, "learning_rate": 9.999397094500355e-05, "loss": 3.6994, "step": 5048 }, { "epoch": 0.31342727667763365, "grad_norm": 0.7839400708006535, "learning_rate": 9.999391473148288e-05, "loss": 3.6128, "step": 5049 }, { "epoch": 0.3134893537773915, "grad_norm": 0.7260590316323157, "learning_rate": 9.999385825713182e-05, "loss": 3.6705, "step": 5050 }, { "epoch": 0.31355143087714943, "grad_norm": 0.7605018089042183, "learning_rate": 9.999380152195066e-05, "loss": 3.6441, "step": 5051 }, { "epoch": 0.31361350797690735, "grad_norm": 0.8356437523505996, "learning_rate": 9.99937445259397e-05, "loss": 3.7408, "step": 5052 }, { "epoch": 0.3136755850766652, "grad_norm": 0.6297701002898242, "learning_rate": 9.999368726909923e-05, "loss": 3.6066, "step": 5053 }, { "epoch": 0.3137376621764231, "grad_norm": 0.6301578837817385, "learning_rate": 9.999362975142957e-05, "loss": 3.7159, "step": 5054 }, { "epoch": 0.31379973927618104, "grad_norm": 0.760211960573151, "learning_rate": 9.9993571972931e-05, "loss": 3.7064, "step": 5055 }, { "epoch": 0.3138618163759389, "grad_norm": 0.8294198691125726, "learning_rate": 9.999351393360383e-05, "loss": 3.6548, "step": 5056 }, { "epoch": 0.3139238934756968, "grad_norm": 0.8683967064071533, "learning_rate": 9.999345563344837e-05, "loss": 3.6831, "step": 5057 }, { "epoch": 0.3139859705754547, "grad_norm": 0.7500003576277834, "learning_rate": 9.999339707246492e-05, "loss": 3.6453, "step": 5058 }, { "epoch": 0.3140480476752126, "grad_norm": 0.7386923558680497, "learning_rate": 9.999333825065377e-05, "loss": 3.6974, "step": 5059 }, { "epoch": 0.3141101247749705, "grad_norm": 0.9020711313586812, "learning_rate": 9.999327916801525e-05, "loss": 3.6802, "step": 5060 }, { "epoch": 0.3141722018747284, "grad_norm": 0.8306825838013114, "learning_rate": 9.999321982454964e-05, "loss": 3.6649, "step": 5061 }, { "epoch": 0.3142342789744863, "grad_norm": 0.7221944209600362, "learning_rate": 9.999316022025727e-05, "loss": 3.6078, "step": 5062 }, { "epoch": 0.3142963560742442, "grad_norm": 0.7080653515816482, "learning_rate": 9.999310035513844e-05, "loss": 3.6289, "step": 5063 }, { "epoch": 0.3143584331740021, "grad_norm": 0.7121600444300417, "learning_rate": 9.999304022919347e-05, "loss": 3.7087, "step": 5064 }, { "epoch": 0.31442051027376, "grad_norm": 0.8600523966529923, "learning_rate": 9.99929798424227e-05, "loss": 3.6114, "step": 5065 }, { "epoch": 0.3144825873735179, "grad_norm": 0.687918427320877, "learning_rate": 9.999291919482639e-05, "loss": 3.7225, "step": 5066 }, { "epoch": 0.3145446644732758, "grad_norm": 0.6838610099417153, "learning_rate": 9.99928582864049e-05, "loss": 3.6889, "step": 5067 }, { "epoch": 0.3146067415730337, "grad_norm": 0.6893355185511292, "learning_rate": 9.999279711715852e-05, "loss": 3.6756, "step": 5068 }, { "epoch": 0.3146688186727916, "grad_norm": 0.7173233801081492, "learning_rate": 9.999273568708759e-05, "loss": 3.6185, "step": 5069 }, { "epoch": 0.3147308957725495, "grad_norm": 0.5840281849634317, "learning_rate": 9.999267399619241e-05, "loss": 3.6771, "step": 5070 }, { "epoch": 0.3147929728723074, "grad_norm": 0.5645191193646455, "learning_rate": 9.999261204447332e-05, "loss": 3.7439, "step": 5071 }, { "epoch": 0.3148550499720653, "grad_norm": 0.5943954623497748, "learning_rate": 9.999254983193065e-05, "loss": 3.6631, "step": 5072 }, { "epoch": 0.3149171270718232, "grad_norm": 0.6048658999873812, "learning_rate": 9.99924873585647e-05, "loss": 3.6105, "step": 5073 }, { "epoch": 0.3149792041715811, "grad_norm": 0.5849421630292, "learning_rate": 9.999242462437581e-05, "loss": 3.6781, "step": 5074 }, { "epoch": 0.315041281271339, "grad_norm": 0.6227509324898739, "learning_rate": 9.99923616293643e-05, "loss": 3.7469, "step": 5075 }, { "epoch": 0.3151033583710969, "grad_norm": 0.536282488943598, "learning_rate": 9.99922983735305e-05, "loss": 3.6399, "step": 5076 }, { "epoch": 0.3151654354708548, "grad_norm": 0.5582528040824469, "learning_rate": 9.999223485687474e-05, "loss": 3.5864, "step": 5077 }, { "epoch": 0.3152275125706127, "grad_norm": 0.5203852697124658, "learning_rate": 9.999217107939738e-05, "loss": 3.7024, "step": 5078 }, { "epoch": 0.3152895896703706, "grad_norm": 0.5655461061369413, "learning_rate": 9.999210704109871e-05, "loss": 3.6078, "step": 5079 }, { "epoch": 0.3153516667701285, "grad_norm": 0.5886259074954467, "learning_rate": 9.999204274197909e-05, "loss": 3.7301, "step": 5080 }, { "epoch": 0.3154137438698864, "grad_norm": 0.6200511506730624, "learning_rate": 9.999197818203883e-05, "loss": 3.7333, "step": 5081 }, { "epoch": 0.31547582096964427, "grad_norm": 0.6576388965872102, "learning_rate": 9.999191336127829e-05, "loss": 3.6177, "step": 5082 }, { "epoch": 0.3155378980694022, "grad_norm": 0.5706802123350905, "learning_rate": 9.999184827969781e-05, "loss": 3.7339, "step": 5083 }, { "epoch": 0.3155999751691601, "grad_norm": 0.5885394245431328, "learning_rate": 9.999178293729772e-05, "loss": 3.6312, "step": 5084 }, { "epoch": 0.31566205226891797, "grad_norm": 0.5527514936943697, "learning_rate": 9.999171733407835e-05, "loss": 3.6476, "step": 5085 }, { "epoch": 0.3157241293686759, "grad_norm": 0.5574758587729374, "learning_rate": 9.999165147004008e-05, "loss": 3.6851, "step": 5086 }, { "epoch": 0.3157862064684338, "grad_norm": 0.5690991807456145, "learning_rate": 9.99915853451832e-05, "loss": 3.5734, "step": 5087 }, { "epoch": 0.31584828356819167, "grad_norm": 0.6144506397427439, "learning_rate": 9.99915189595081e-05, "loss": 3.7186, "step": 5088 }, { "epoch": 0.3159103606679496, "grad_norm": 0.6784281189426761, "learning_rate": 9.99914523130151e-05, "loss": 3.6322, "step": 5089 }, { "epoch": 0.3159724377677075, "grad_norm": 0.7889469930879557, "learning_rate": 9.999138540570457e-05, "loss": 3.6484, "step": 5090 }, { "epoch": 0.31603451486746537, "grad_norm": 0.7513951597071554, "learning_rate": 9.999131823757683e-05, "loss": 3.6804, "step": 5091 }, { "epoch": 0.3160965919672233, "grad_norm": 0.6270449086918153, "learning_rate": 9.999125080863227e-05, "loss": 3.6246, "step": 5092 }, { "epoch": 0.3161586690669812, "grad_norm": 0.5603435027805017, "learning_rate": 9.999118311887121e-05, "loss": 3.584, "step": 5093 }, { "epoch": 0.31622074616673906, "grad_norm": 0.6513587230492708, "learning_rate": 9.999111516829401e-05, "loss": 3.7203, "step": 5094 }, { "epoch": 0.316282823266497, "grad_norm": 0.6803139397707113, "learning_rate": 9.999104695690101e-05, "loss": 3.6066, "step": 5095 }, { "epoch": 0.3163449003662549, "grad_norm": 0.7177314590323448, "learning_rate": 9.99909784846926e-05, "loss": 3.6307, "step": 5096 }, { "epoch": 0.31640697746601276, "grad_norm": 0.6855618293196233, "learning_rate": 9.999090975166911e-05, "loss": 3.6995, "step": 5097 }, { "epoch": 0.3164690545657707, "grad_norm": 0.730140612332942, "learning_rate": 9.999084075783092e-05, "loss": 3.6523, "step": 5098 }, { "epoch": 0.3165311316655286, "grad_norm": 0.7203467089955614, "learning_rate": 9.999077150317837e-05, "loss": 3.743, "step": 5099 }, { "epoch": 0.31659320876528646, "grad_norm": 0.7722232805064914, "learning_rate": 9.999070198771184e-05, "loss": 3.7276, "step": 5100 }, { "epoch": 0.3166552858650444, "grad_norm": 0.7789262545206297, "learning_rate": 9.999063221143165e-05, "loss": 3.704, "step": 5101 }, { "epoch": 0.3167173629648023, "grad_norm": 0.6765384070707564, "learning_rate": 9.999056217433823e-05, "loss": 3.6105, "step": 5102 }, { "epoch": 0.31677944006456016, "grad_norm": 0.768557208007402, "learning_rate": 9.99904918764319e-05, "loss": 3.6922, "step": 5103 }, { "epoch": 0.3168415171643181, "grad_norm": 0.7533911887531357, "learning_rate": 9.999042131771304e-05, "loss": 3.669, "step": 5104 }, { "epoch": 0.316903594264076, "grad_norm": 0.7287937482262133, "learning_rate": 9.999035049818202e-05, "loss": 3.6902, "step": 5105 }, { "epoch": 0.31696567136383386, "grad_norm": 0.6911085317550103, "learning_rate": 9.99902794178392e-05, "loss": 3.7159, "step": 5106 }, { "epoch": 0.3170277484635918, "grad_norm": 0.5813334589691566, "learning_rate": 9.999020807668496e-05, "loss": 3.6216, "step": 5107 }, { "epoch": 0.3170898255633497, "grad_norm": 0.7021264509172815, "learning_rate": 9.999013647471966e-05, "loss": 3.7298, "step": 5108 }, { "epoch": 0.31715190266310755, "grad_norm": 0.663064183817964, "learning_rate": 9.99900646119437e-05, "loss": 3.6509, "step": 5109 }, { "epoch": 0.3172139797628655, "grad_norm": 0.6372955377724497, "learning_rate": 9.998999248835742e-05, "loss": 3.5526, "step": 5110 }, { "epoch": 0.3172760568626234, "grad_norm": 0.6073005775945144, "learning_rate": 9.998992010396122e-05, "loss": 3.6328, "step": 5111 }, { "epoch": 0.31733813396238125, "grad_norm": 0.6166480157153843, "learning_rate": 9.998984745875548e-05, "loss": 3.6308, "step": 5112 }, { "epoch": 0.31740021106213917, "grad_norm": 0.60759896888997, "learning_rate": 9.998977455274056e-05, "loss": 3.6049, "step": 5113 }, { "epoch": 0.3174622881618971, "grad_norm": 0.6316886624085596, "learning_rate": 9.998970138591686e-05, "loss": 3.679, "step": 5114 }, { "epoch": 0.31752436526165495, "grad_norm": 0.6962265906858767, "learning_rate": 9.998962795828473e-05, "loss": 3.6614, "step": 5115 }, { "epoch": 0.31758644236141287, "grad_norm": 0.6850729061923692, "learning_rate": 9.99895542698446e-05, "loss": 3.6153, "step": 5116 }, { "epoch": 0.3176485194611708, "grad_norm": 0.5700863690359815, "learning_rate": 9.998948032059682e-05, "loss": 3.5761, "step": 5117 }, { "epoch": 0.31771059656092865, "grad_norm": 0.564199634905362, "learning_rate": 9.998940611054178e-05, "loss": 3.6206, "step": 5118 }, { "epoch": 0.31777267366068657, "grad_norm": 0.5931136586237507, "learning_rate": 9.998933163967988e-05, "loss": 3.6477, "step": 5119 }, { "epoch": 0.3178347507604445, "grad_norm": 0.5884710595867184, "learning_rate": 9.998925690801151e-05, "loss": 3.6251, "step": 5120 }, { "epoch": 0.31789682786020235, "grad_norm": 0.5666582007570474, "learning_rate": 9.998918191553703e-05, "loss": 3.6986, "step": 5121 }, { "epoch": 0.31795890495996026, "grad_norm": 0.688721114208879, "learning_rate": 9.998910666225687e-05, "loss": 3.7339, "step": 5122 }, { "epoch": 0.3180209820597182, "grad_norm": 0.6097705242148336, "learning_rate": 9.99890311481714e-05, "loss": 3.6483, "step": 5123 }, { "epoch": 0.31808305915947604, "grad_norm": 0.6463307777330981, "learning_rate": 9.998895537328102e-05, "loss": 3.6496, "step": 5124 }, { "epoch": 0.31814513625923396, "grad_norm": 0.5538887397785469, "learning_rate": 9.998887933758612e-05, "loss": 3.6881, "step": 5125 }, { "epoch": 0.3182072133589919, "grad_norm": 0.5469957490856965, "learning_rate": 9.998880304108711e-05, "loss": 3.5875, "step": 5126 }, { "epoch": 0.31826929045874974, "grad_norm": 0.5444433100222535, "learning_rate": 9.998872648378438e-05, "loss": 3.6535, "step": 5127 }, { "epoch": 0.31833136755850766, "grad_norm": 0.6026868839800751, "learning_rate": 9.998864966567833e-05, "loss": 3.6241, "step": 5128 }, { "epoch": 0.3183934446582656, "grad_norm": 0.5778554597340382, "learning_rate": 9.998857258676935e-05, "loss": 3.5296, "step": 5129 }, { "epoch": 0.31845552175802344, "grad_norm": 0.6356442518141333, "learning_rate": 9.998849524705785e-05, "loss": 3.5289, "step": 5130 }, { "epoch": 0.31851759885778136, "grad_norm": 0.5653538628371069, "learning_rate": 9.998841764654424e-05, "loss": 3.7278, "step": 5131 }, { "epoch": 0.3185796759575393, "grad_norm": 0.6142969164508105, "learning_rate": 9.998833978522893e-05, "loss": 3.5806, "step": 5132 }, { "epoch": 0.31864175305729714, "grad_norm": 0.6320685263676211, "learning_rate": 9.99882616631123e-05, "loss": 3.6276, "step": 5133 }, { "epoch": 0.31870383015705506, "grad_norm": 0.6554013396053575, "learning_rate": 9.998818328019476e-05, "loss": 3.6458, "step": 5134 }, { "epoch": 0.318765907256813, "grad_norm": 0.7329110508353551, "learning_rate": 9.998810463647676e-05, "loss": 3.6483, "step": 5135 }, { "epoch": 0.31882798435657084, "grad_norm": 0.7918728551557077, "learning_rate": 9.998802573195868e-05, "loss": 3.6858, "step": 5136 }, { "epoch": 0.31889006145632875, "grad_norm": 0.6745193695301629, "learning_rate": 9.998794656664095e-05, "loss": 3.5371, "step": 5137 }, { "epoch": 0.3189521385560867, "grad_norm": 0.717781949154996, "learning_rate": 9.998786714052393e-05, "loss": 3.6593, "step": 5138 }, { "epoch": 0.31901421565584454, "grad_norm": 0.5707426147584659, "learning_rate": 9.99877874536081e-05, "loss": 3.5806, "step": 5139 }, { "epoch": 0.31907629275560245, "grad_norm": 0.599231961538602, "learning_rate": 9.998770750589384e-05, "loss": 3.6834, "step": 5140 }, { "epoch": 0.31913836985536037, "grad_norm": 0.5346954477991422, "learning_rate": 9.998762729738156e-05, "loss": 3.6341, "step": 5141 }, { "epoch": 0.31920044695511823, "grad_norm": 0.5475577588293555, "learning_rate": 9.998754682807172e-05, "loss": 3.622, "step": 5142 }, { "epoch": 0.31926252405487615, "grad_norm": 0.6296590008214327, "learning_rate": 9.99874660979647e-05, "loss": 3.5607, "step": 5143 }, { "epoch": 0.31932460115463407, "grad_norm": 0.6429735955279403, "learning_rate": 9.998738510706094e-05, "loss": 3.6195, "step": 5144 }, { "epoch": 0.31938667825439193, "grad_norm": 0.53097069634833, "learning_rate": 9.998730385536084e-05, "loss": 3.6376, "step": 5145 }, { "epoch": 0.31944875535414985, "grad_norm": 0.4861138725959209, "learning_rate": 9.998722234286487e-05, "loss": 3.6018, "step": 5146 }, { "epoch": 0.31951083245390777, "grad_norm": 0.5393794137148655, "learning_rate": 9.998714056957339e-05, "loss": 3.5715, "step": 5147 }, { "epoch": 0.31957290955366563, "grad_norm": 0.5795124776828233, "learning_rate": 9.998705853548689e-05, "loss": 3.7052, "step": 5148 }, { "epoch": 0.31963498665342355, "grad_norm": 0.5283738988752513, "learning_rate": 9.998697624060576e-05, "loss": 3.6812, "step": 5149 }, { "epoch": 0.31969706375318147, "grad_norm": 0.5699493872190037, "learning_rate": 9.998689368493044e-05, "loss": 3.7005, "step": 5150 }, { "epoch": 0.3197591408529393, "grad_norm": 0.5836371726104334, "learning_rate": 9.998681086846137e-05, "loss": 3.6711, "step": 5151 }, { "epoch": 0.31982121795269725, "grad_norm": 0.6924404448505044, "learning_rate": 9.998672779119896e-05, "loss": 3.5838, "step": 5152 }, { "epoch": 0.31988329505245516, "grad_norm": 0.7323458619409788, "learning_rate": 9.998664445314367e-05, "loss": 3.5789, "step": 5153 }, { "epoch": 0.319945372152213, "grad_norm": 0.6646192572838657, "learning_rate": 9.998656085429591e-05, "loss": 3.7386, "step": 5154 }, { "epoch": 0.32000744925197094, "grad_norm": 0.5763357155822152, "learning_rate": 9.998647699465614e-05, "loss": 3.6825, "step": 5155 }, { "epoch": 0.32006952635172886, "grad_norm": 0.6803537589140284, "learning_rate": 9.998639287422477e-05, "loss": 3.6401, "step": 5156 }, { "epoch": 0.3201316034514867, "grad_norm": 0.7692273696952674, "learning_rate": 9.998630849300226e-05, "loss": 3.7016, "step": 5157 }, { "epoch": 0.32019368055124464, "grad_norm": 0.7373542011132251, "learning_rate": 9.998622385098903e-05, "loss": 3.6087, "step": 5158 }, { "epoch": 0.32025575765100256, "grad_norm": 0.6421744287278983, "learning_rate": 9.998613894818556e-05, "loss": 3.6766, "step": 5159 }, { "epoch": 0.3203178347507604, "grad_norm": 0.6892516622595932, "learning_rate": 9.998605378459226e-05, "loss": 3.6273, "step": 5160 }, { "epoch": 0.32037991185051834, "grad_norm": 0.6136055107339861, "learning_rate": 9.998596836020958e-05, "loss": 3.6167, "step": 5161 }, { "epoch": 0.32044198895027626, "grad_norm": 0.5842718992213919, "learning_rate": 9.998588267503797e-05, "loss": 3.6238, "step": 5162 }, { "epoch": 0.3205040660500341, "grad_norm": 0.6008779420663584, "learning_rate": 9.998579672907788e-05, "loss": 3.6081, "step": 5163 }, { "epoch": 0.32056614314979204, "grad_norm": 0.5766677791340338, "learning_rate": 9.998571052232975e-05, "loss": 3.7202, "step": 5164 }, { "epoch": 0.32062822024954996, "grad_norm": 0.5845075324415906, "learning_rate": 9.998562405479403e-05, "loss": 3.6407, "step": 5165 }, { "epoch": 0.3206902973493078, "grad_norm": 0.5492232994010546, "learning_rate": 9.998553732647119e-05, "loss": 3.4989, "step": 5166 }, { "epoch": 0.32075237444906574, "grad_norm": 0.5912125223552531, "learning_rate": 9.998545033736164e-05, "loss": 3.6466, "step": 5167 }, { "epoch": 0.32081445154882365, "grad_norm": 0.5600035689444575, "learning_rate": 9.99853630874659e-05, "loss": 3.69, "step": 5168 }, { "epoch": 0.3208765286485815, "grad_norm": 0.6880804342655652, "learning_rate": 9.998527557678435e-05, "loss": 3.676, "step": 5169 }, { "epoch": 0.32093860574833943, "grad_norm": 0.6296513095044503, "learning_rate": 9.99851878053175e-05, "loss": 3.6164, "step": 5170 }, { "epoch": 0.32100068284809735, "grad_norm": 0.7712679402445747, "learning_rate": 9.998509977306578e-05, "loss": 3.627, "step": 5171 }, { "epoch": 0.3210627599478552, "grad_norm": 0.7279127965983282, "learning_rate": 9.998501148002967e-05, "loss": 3.5916, "step": 5172 }, { "epoch": 0.32112483704761313, "grad_norm": 0.6856606543910517, "learning_rate": 9.998492292620962e-05, "loss": 3.6792, "step": 5173 }, { "epoch": 0.32118691414737105, "grad_norm": 0.7545042680350412, "learning_rate": 9.998483411160609e-05, "loss": 3.6966, "step": 5174 }, { "epoch": 0.3212489912471289, "grad_norm": 0.7988878521590506, "learning_rate": 9.998474503621954e-05, "loss": 3.7172, "step": 5175 }, { "epoch": 0.32131106834688683, "grad_norm": 0.96499765381009, "learning_rate": 9.998465570005045e-05, "loss": 3.6704, "step": 5176 }, { "epoch": 0.32137314544664475, "grad_norm": 1.0157107390379574, "learning_rate": 9.998456610309928e-05, "loss": 3.634, "step": 5177 }, { "epoch": 0.3214352225464026, "grad_norm": 0.7297081889256194, "learning_rate": 9.998447624536649e-05, "loss": 3.6764, "step": 5178 }, { "epoch": 0.32149729964616053, "grad_norm": 0.7056039555995597, "learning_rate": 9.998438612685255e-05, "loss": 3.6209, "step": 5179 }, { "epoch": 0.32155937674591845, "grad_norm": 0.7909158526004997, "learning_rate": 9.998429574755794e-05, "loss": 3.6624, "step": 5180 }, { "epoch": 0.3216214538456763, "grad_norm": 0.6672298962882921, "learning_rate": 9.998420510748312e-05, "loss": 3.5134, "step": 5181 }, { "epoch": 0.3216835309454342, "grad_norm": 0.7918245300371546, "learning_rate": 9.998411420662857e-05, "loss": 3.564, "step": 5182 }, { "epoch": 0.32174560804519214, "grad_norm": 1.0683376721508773, "learning_rate": 9.998402304499474e-05, "loss": 3.5293, "step": 5183 }, { "epoch": 0.32180768514495, "grad_norm": 1.1158775172033728, "learning_rate": 9.998393162258216e-05, "loss": 3.609, "step": 5184 }, { "epoch": 0.3218697622447079, "grad_norm": 0.8996528366529225, "learning_rate": 9.998383993939127e-05, "loss": 3.6167, "step": 5185 }, { "epoch": 0.32193183934446584, "grad_norm": 0.7463094988542884, "learning_rate": 9.998374799542254e-05, "loss": 3.6176, "step": 5186 }, { "epoch": 0.3219939164442237, "grad_norm": 0.7703512247809335, "learning_rate": 9.998365579067648e-05, "loss": 3.7339, "step": 5187 }, { "epoch": 0.3220559935439816, "grad_norm": 0.9726389688082671, "learning_rate": 9.998356332515353e-05, "loss": 3.6481, "step": 5188 }, { "epoch": 0.32211807064373954, "grad_norm": 0.8305474603769484, "learning_rate": 9.998347059885422e-05, "loss": 3.6807, "step": 5189 }, { "epoch": 0.3221801477434974, "grad_norm": 0.7014391503931973, "learning_rate": 9.9983377611779e-05, "loss": 3.6476, "step": 5190 }, { "epoch": 0.3222422248432553, "grad_norm": 0.7901869589458201, "learning_rate": 9.998328436392837e-05, "loss": 3.7363, "step": 5191 }, { "epoch": 0.32230430194301324, "grad_norm": 0.6035869766080453, "learning_rate": 9.998319085530279e-05, "loss": 3.6283, "step": 5192 }, { "epoch": 0.3223663790427711, "grad_norm": 0.6835823712083103, "learning_rate": 9.998309708590279e-05, "loss": 3.6639, "step": 5193 }, { "epoch": 0.322428456142529, "grad_norm": 0.8869014333343709, "learning_rate": 9.998300305572883e-05, "loss": 3.6866, "step": 5194 }, { "epoch": 0.32249053324228694, "grad_norm": 0.7953086493198429, "learning_rate": 9.998290876478142e-05, "loss": 3.7618, "step": 5195 }, { "epoch": 0.3225526103420448, "grad_norm": 0.6598684146145668, "learning_rate": 9.998281421306103e-05, "loss": 3.703, "step": 5196 }, { "epoch": 0.3226146874418027, "grad_norm": 0.6385727551739995, "learning_rate": 9.998271940056818e-05, "loss": 3.6255, "step": 5197 }, { "epoch": 0.32267676454156063, "grad_norm": 0.6195390303942185, "learning_rate": 9.998262432730334e-05, "loss": 3.6613, "step": 5198 }, { "epoch": 0.3227388416413185, "grad_norm": 0.6192331335229478, "learning_rate": 9.9982528993267e-05, "loss": 3.6305, "step": 5199 }, { "epoch": 0.3228009187410764, "grad_norm": 0.571671786447552, "learning_rate": 9.998243339845969e-05, "loss": 3.6788, "step": 5200 }, { "epoch": 0.32286299584083433, "grad_norm": 0.6760424952411325, "learning_rate": 9.998233754288189e-05, "loss": 3.6143, "step": 5201 }, { "epoch": 0.3229250729405922, "grad_norm": 0.6273873984994183, "learning_rate": 9.99822414265341e-05, "loss": 3.5132, "step": 5202 }, { "epoch": 0.3229871500403501, "grad_norm": 0.5122467985616332, "learning_rate": 9.99821450494168e-05, "loss": 3.6045, "step": 5203 }, { "epoch": 0.32304922714010803, "grad_norm": 0.5308282243741289, "learning_rate": 9.998204841153055e-05, "loss": 3.6183, "step": 5204 }, { "epoch": 0.3231113042398659, "grad_norm": 0.5424100072991789, "learning_rate": 9.998195151287582e-05, "loss": 3.6283, "step": 5205 }, { "epoch": 0.3231733813396238, "grad_norm": 0.556261944374686, "learning_rate": 9.998185435345309e-05, "loss": 3.6225, "step": 5206 }, { "epoch": 0.32323545843938173, "grad_norm": 0.5796474823415889, "learning_rate": 9.99817569332629e-05, "loss": 3.7422, "step": 5207 }, { "epoch": 0.3232975355391396, "grad_norm": 0.6070772524258365, "learning_rate": 9.998165925230576e-05, "loss": 3.5856, "step": 5208 }, { "epoch": 0.3233596126388975, "grad_norm": 0.5283889302875932, "learning_rate": 9.998156131058217e-05, "loss": 3.7084, "step": 5209 }, { "epoch": 0.3234216897386554, "grad_norm": 0.4768151567189832, "learning_rate": 9.998146310809265e-05, "loss": 3.588, "step": 5210 }, { "epoch": 0.3234837668384133, "grad_norm": 0.5999567473239344, "learning_rate": 9.998136464483767e-05, "loss": 3.5915, "step": 5211 }, { "epoch": 0.3235458439381712, "grad_norm": 0.565500548144074, "learning_rate": 9.99812659208178e-05, "loss": 3.5767, "step": 5212 }, { "epoch": 0.3236079210379291, "grad_norm": 0.5072120931777396, "learning_rate": 9.998116693603354e-05, "loss": 3.6417, "step": 5213 }, { "epoch": 0.323669998137687, "grad_norm": 0.4846770821244317, "learning_rate": 9.998106769048538e-05, "loss": 3.5098, "step": 5214 }, { "epoch": 0.3237320752374449, "grad_norm": 0.5498062865537221, "learning_rate": 9.998096818417387e-05, "loss": 3.607, "step": 5215 }, { "epoch": 0.3237941523372028, "grad_norm": 0.5821762160448021, "learning_rate": 9.99808684170995e-05, "loss": 3.6574, "step": 5216 }, { "epoch": 0.3238562294369607, "grad_norm": 0.5665621838829139, "learning_rate": 9.998076838926282e-05, "loss": 3.6097, "step": 5217 }, { "epoch": 0.3239183065367186, "grad_norm": 0.5923798714151897, "learning_rate": 9.998066810066434e-05, "loss": 3.6464, "step": 5218 }, { "epoch": 0.3239803836364765, "grad_norm": 0.6211810020723321, "learning_rate": 9.998056755130458e-05, "loss": 3.636, "step": 5219 }, { "epoch": 0.3240424607362344, "grad_norm": 0.5962999449710542, "learning_rate": 9.998046674118406e-05, "loss": 3.5933, "step": 5220 }, { "epoch": 0.3241045378359923, "grad_norm": 0.5513633465742916, "learning_rate": 9.998036567030331e-05, "loss": 3.6634, "step": 5221 }, { "epoch": 0.3241666149357502, "grad_norm": 0.48140123641422294, "learning_rate": 9.998026433866287e-05, "loss": 3.6073, "step": 5222 }, { "epoch": 0.3242286920355081, "grad_norm": 0.49471176071102113, "learning_rate": 9.998016274626324e-05, "loss": 3.6361, "step": 5223 }, { "epoch": 0.324290769135266, "grad_norm": 0.5467961390719671, "learning_rate": 9.998006089310499e-05, "loss": 3.5573, "step": 5224 }, { "epoch": 0.3243528462350239, "grad_norm": 0.55493515825814, "learning_rate": 9.997995877918862e-05, "loss": 3.6732, "step": 5225 }, { "epoch": 0.3244149233347818, "grad_norm": 0.6418379488970313, "learning_rate": 9.997985640451466e-05, "loss": 3.6411, "step": 5226 }, { "epoch": 0.3244770004345397, "grad_norm": 0.5844634269680408, "learning_rate": 9.997975376908368e-05, "loss": 3.602, "step": 5227 }, { "epoch": 0.3245390775342976, "grad_norm": 0.6121032237393214, "learning_rate": 9.997965087289617e-05, "loss": 3.7184, "step": 5228 }, { "epoch": 0.3246011546340555, "grad_norm": 0.6151714958877095, "learning_rate": 9.997954771595271e-05, "loss": 3.6658, "step": 5229 }, { "epoch": 0.3246632317338134, "grad_norm": 0.6639511014961248, "learning_rate": 9.997944429825379e-05, "loss": 3.5925, "step": 5230 }, { "epoch": 0.3247253088335713, "grad_norm": 0.6317261448648371, "learning_rate": 9.997934061979998e-05, "loss": 3.6694, "step": 5231 }, { "epoch": 0.3247873859333292, "grad_norm": 0.5939065827145708, "learning_rate": 9.997923668059183e-05, "loss": 3.6515, "step": 5232 }, { "epoch": 0.3248494630330871, "grad_norm": 0.6277651890216425, "learning_rate": 9.997913248062987e-05, "loss": 3.5999, "step": 5233 }, { "epoch": 0.324911540132845, "grad_norm": 0.714369194227441, "learning_rate": 9.997902801991463e-05, "loss": 3.6791, "step": 5234 }, { "epoch": 0.3249736172326029, "grad_norm": 0.666029272750678, "learning_rate": 9.997892329844669e-05, "loss": 3.5405, "step": 5235 }, { "epoch": 0.3250356943323608, "grad_norm": 0.6087141610783409, "learning_rate": 9.997881831622656e-05, "loss": 3.5437, "step": 5236 }, { "epoch": 0.3250977714321187, "grad_norm": 0.6130244209316373, "learning_rate": 9.997871307325479e-05, "loss": 3.6687, "step": 5237 }, { "epoch": 0.32515984853187657, "grad_norm": 0.5767673325003234, "learning_rate": 9.997860756953197e-05, "loss": 3.54, "step": 5238 }, { "epoch": 0.3252219256316345, "grad_norm": 0.565984371525564, "learning_rate": 9.99785018050586e-05, "loss": 3.6407, "step": 5239 }, { "epoch": 0.3252840027313924, "grad_norm": 0.6354808827037615, "learning_rate": 9.997839577983525e-05, "loss": 3.6279, "step": 5240 }, { "epoch": 0.32534607983115027, "grad_norm": 0.5738730395496927, "learning_rate": 9.99782894938625e-05, "loss": 3.6444, "step": 5241 }, { "epoch": 0.3254081569309082, "grad_norm": 0.5708367798926266, "learning_rate": 9.997818294714086e-05, "loss": 3.636, "step": 5242 }, { "epoch": 0.3254702340306661, "grad_norm": 0.6962928718366388, "learning_rate": 9.997807613967091e-05, "loss": 3.6227, "step": 5243 }, { "epoch": 0.32553231113042397, "grad_norm": 0.6624465578944407, "learning_rate": 9.99779690714532e-05, "loss": 3.5776, "step": 5244 }, { "epoch": 0.3255943882301819, "grad_norm": 0.6020081095798785, "learning_rate": 9.997786174248832e-05, "loss": 3.5989, "step": 5245 }, { "epoch": 0.3256564653299398, "grad_norm": 0.60624426121306, "learning_rate": 9.997775415277678e-05, "loss": 3.5547, "step": 5246 }, { "epoch": 0.32571854242969767, "grad_norm": 0.5400516998302408, "learning_rate": 9.997764630231917e-05, "loss": 3.6771, "step": 5247 }, { "epoch": 0.3257806195294556, "grad_norm": 0.5714625690130319, "learning_rate": 9.997753819111604e-05, "loss": 3.6675, "step": 5248 }, { "epoch": 0.3258426966292135, "grad_norm": 0.5666500224679707, "learning_rate": 9.997742981916795e-05, "loss": 3.6365, "step": 5249 }, { "epoch": 0.32590477372897136, "grad_norm": 0.5804688906765713, "learning_rate": 9.99773211864755e-05, "loss": 3.544, "step": 5250 }, { "epoch": 0.3259668508287293, "grad_norm": 0.6218135429570658, "learning_rate": 9.997721229303922e-05, "loss": 3.6237, "step": 5251 }, { "epoch": 0.3260289279284872, "grad_norm": 0.6825562151090424, "learning_rate": 9.997710313885971e-05, "loss": 3.5824, "step": 5252 }, { "epoch": 0.32609100502824506, "grad_norm": 0.5817682068667865, "learning_rate": 9.99769937239375e-05, "loss": 3.5163, "step": 5253 }, { "epoch": 0.326153082128003, "grad_norm": 0.559885332467427, "learning_rate": 9.997688404827319e-05, "loss": 3.5848, "step": 5254 }, { "epoch": 0.3262151592277609, "grad_norm": 0.6557310868392435, "learning_rate": 9.997677411186733e-05, "loss": 3.6754, "step": 5255 }, { "epoch": 0.32627723632751876, "grad_norm": 0.6558697825203034, "learning_rate": 9.997666391472052e-05, "loss": 3.6533, "step": 5256 }, { "epoch": 0.3263393134272767, "grad_norm": 0.6172015031602198, "learning_rate": 9.997655345683332e-05, "loss": 3.7016, "step": 5257 }, { "epoch": 0.3264013905270346, "grad_norm": 0.5968322059509202, "learning_rate": 9.99764427382063e-05, "loss": 3.6024, "step": 5258 }, { "epoch": 0.32646346762679246, "grad_norm": 0.6554071372427265, "learning_rate": 9.997633175884006e-05, "loss": 3.5739, "step": 5259 }, { "epoch": 0.3265255447265504, "grad_norm": 0.8527134807576389, "learning_rate": 9.997622051873517e-05, "loss": 3.7325, "step": 5260 }, { "epoch": 0.3265876218263083, "grad_norm": 0.7708802767369338, "learning_rate": 9.997610901789219e-05, "loss": 3.6334, "step": 5261 }, { "epoch": 0.32664969892606616, "grad_norm": 0.6470044167838063, "learning_rate": 9.997599725631173e-05, "loss": 3.653, "step": 5262 }, { "epoch": 0.3267117760258241, "grad_norm": 0.7316128751479049, "learning_rate": 9.997588523399435e-05, "loss": 3.6294, "step": 5263 }, { "epoch": 0.326773853125582, "grad_norm": 0.763690216546644, "learning_rate": 9.997577295094064e-05, "loss": 3.5717, "step": 5264 }, { "epoch": 0.32683593022533985, "grad_norm": 0.7534197487751372, "learning_rate": 9.99756604071512e-05, "loss": 3.6394, "step": 5265 }, { "epoch": 0.3268980073250978, "grad_norm": 0.7291941819222012, "learning_rate": 9.997554760262659e-05, "loss": 3.6447, "step": 5266 }, { "epoch": 0.3269600844248557, "grad_norm": 0.6262231540446068, "learning_rate": 9.997543453736744e-05, "loss": 3.6461, "step": 5267 }, { "epoch": 0.32702216152461355, "grad_norm": 0.6345546195642597, "learning_rate": 9.99753212113743e-05, "loss": 3.5862, "step": 5268 }, { "epoch": 0.32708423862437147, "grad_norm": 0.7691204698051952, "learning_rate": 9.99752076246478e-05, "loss": 3.5732, "step": 5269 }, { "epoch": 0.3271463157241294, "grad_norm": 0.6847153971084512, "learning_rate": 9.997509377718849e-05, "loss": 3.5374, "step": 5270 }, { "epoch": 0.32720839282388725, "grad_norm": 0.7202910198724322, "learning_rate": 9.997497966899698e-05, "loss": 3.6319, "step": 5271 }, { "epoch": 0.32727046992364517, "grad_norm": 0.6100309460638561, "learning_rate": 9.997486530007388e-05, "loss": 3.5824, "step": 5272 }, { "epoch": 0.3273325470234031, "grad_norm": 0.6239938027474577, "learning_rate": 9.997475067041977e-05, "loss": 3.6485, "step": 5273 }, { "epoch": 0.32739462412316095, "grad_norm": 0.6831342624270202, "learning_rate": 9.997463578003527e-05, "loss": 3.6789, "step": 5274 }, { "epoch": 0.32745670122291887, "grad_norm": 0.6203038693075053, "learning_rate": 9.997452062892094e-05, "loss": 3.6682, "step": 5275 }, { "epoch": 0.3275187783226768, "grad_norm": 0.5374162109626632, "learning_rate": 9.997440521707741e-05, "loss": 3.6139, "step": 5276 }, { "epoch": 0.32758085542243465, "grad_norm": 0.5804944069943676, "learning_rate": 9.997428954450529e-05, "loss": 3.6134, "step": 5277 }, { "epoch": 0.32764293252219256, "grad_norm": 0.5881969882960622, "learning_rate": 9.997417361120515e-05, "loss": 3.6061, "step": 5278 }, { "epoch": 0.3277050096219505, "grad_norm": 0.6112861009480088, "learning_rate": 9.997405741717764e-05, "loss": 3.4903, "step": 5279 }, { "epoch": 0.32776708672170835, "grad_norm": 0.5527624385974921, "learning_rate": 9.997394096242332e-05, "loss": 3.681, "step": 5280 }, { "epoch": 0.32782916382146626, "grad_norm": 0.5199991571437901, "learning_rate": 9.997382424694282e-05, "loss": 3.6938, "step": 5281 }, { "epoch": 0.3278912409212242, "grad_norm": 0.5738011612864111, "learning_rate": 9.997370727073676e-05, "loss": 3.7361, "step": 5282 }, { "epoch": 0.32795331802098204, "grad_norm": 0.5741481088796019, "learning_rate": 9.997359003380574e-05, "loss": 3.5698, "step": 5283 }, { "epoch": 0.32801539512073996, "grad_norm": 0.6349807374765977, "learning_rate": 9.997347253615035e-05, "loss": 3.7917, "step": 5284 }, { "epoch": 0.3280774722204979, "grad_norm": 0.6317097509867405, "learning_rate": 9.997335477777123e-05, "loss": 3.6706, "step": 5285 }, { "epoch": 0.32813954932025574, "grad_norm": 0.6883987274541624, "learning_rate": 9.997323675866898e-05, "loss": 3.61, "step": 5286 }, { "epoch": 0.32820162642001366, "grad_norm": 0.6916478914753649, "learning_rate": 9.997311847884423e-05, "loss": 3.6152, "step": 5287 }, { "epoch": 0.3282637035197716, "grad_norm": 0.655789077199452, "learning_rate": 9.99729999382976e-05, "loss": 3.6494, "step": 5288 }, { "epoch": 0.32832578061952944, "grad_norm": 0.5716697011687295, "learning_rate": 9.997288113702968e-05, "loss": 3.6838, "step": 5289 }, { "epoch": 0.32838785771928736, "grad_norm": 0.6075995574816461, "learning_rate": 9.997276207504112e-05, "loss": 3.6258, "step": 5290 }, { "epoch": 0.3284499348190453, "grad_norm": 0.5555744869927245, "learning_rate": 9.997264275233253e-05, "loss": 3.7129, "step": 5291 }, { "epoch": 0.32851201191880314, "grad_norm": 0.5873693138215891, "learning_rate": 9.997252316890452e-05, "loss": 3.5997, "step": 5292 }, { "epoch": 0.32857408901856106, "grad_norm": 0.5604503694726218, "learning_rate": 9.997240332475772e-05, "loss": 3.5231, "step": 5293 }, { "epoch": 0.328636166118319, "grad_norm": 0.6094539053278313, "learning_rate": 9.997228321989278e-05, "loss": 3.6103, "step": 5294 }, { "epoch": 0.32869824321807684, "grad_norm": 0.6170563135879233, "learning_rate": 9.99721628543103e-05, "loss": 3.6243, "step": 5295 }, { "epoch": 0.32876032031783475, "grad_norm": 0.7515144158285861, "learning_rate": 9.997204222801091e-05, "loss": 3.5786, "step": 5296 }, { "epoch": 0.32882239741759267, "grad_norm": 0.6973482398235414, "learning_rate": 9.997192134099526e-05, "loss": 3.5763, "step": 5297 }, { "epoch": 0.32888447451735053, "grad_norm": 0.6655235822183685, "learning_rate": 9.997180019326395e-05, "loss": 3.6689, "step": 5298 }, { "epoch": 0.32894655161710845, "grad_norm": 0.7234341944707571, "learning_rate": 9.997167878481763e-05, "loss": 3.5145, "step": 5299 }, { "epoch": 0.32900862871686637, "grad_norm": 0.6451688444941245, "learning_rate": 9.997155711565693e-05, "loss": 3.6379, "step": 5300 }, { "epoch": 0.32907070581662423, "grad_norm": 0.6310921348859005, "learning_rate": 9.997143518578249e-05, "loss": 3.6031, "step": 5301 }, { "epoch": 0.32913278291638215, "grad_norm": 0.6268929188184036, "learning_rate": 9.997131299519493e-05, "loss": 3.6997, "step": 5302 }, { "epoch": 0.32919486001614007, "grad_norm": 0.6097046131839198, "learning_rate": 9.997119054389492e-05, "loss": 3.5477, "step": 5303 }, { "epoch": 0.32925693711589793, "grad_norm": 0.5764282949093793, "learning_rate": 9.997106783188306e-05, "loss": 3.6812, "step": 5304 }, { "epoch": 0.32931901421565585, "grad_norm": 0.6184433099937375, "learning_rate": 9.997094485916002e-05, "loss": 3.592, "step": 5305 }, { "epoch": 0.32938109131541377, "grad_norm": 0.606622151070804, "learning_rate": 9.997082162572642e-05, "loss": 3.651, "step": 5306 }, { "epoch": 0.3294431684151716, "grad_norm": 0.6741853292410283, "learning_rate": 9.997069813158293e-05, "loss": 3.5102, "step": 5307 }, { "epoch": 0.32950524551492955, "grad_norm": 0.6844054498386944, "learning_rate": 9.997057437673016e-05, "loss": 3.64, "step": 5308 }, { "epoch": 0.32956732261468746, "grad_norm": 0.6836766655908517, "learning_rate": 9.997045036116876e-05, "loss": 3.5993, "step": 5309 }, { "epoch": 0.3296293997144453, "grad_norm": 0.7026235487752769, "learning_rate": 9.997032608489942e-05, "loss": 3.6324, "step": 5310 }, { "epoch": 0.32969147681420324, "grad_norm": 0.7253538764350941, "learning_rate": 9.997020154792274e-05, "loss": 3.6955, "step": 5311 }, { "epoch": 0.32975355391396116, "grad_norm": 0.7245306913115795, "learning_rate": 9.997007675023939e-05, "loss": 3.6072, "step": 5312 }, { "epoch": 0.329815631013719, "grad_norm": 0.7524576055427912, "learning_rate": 9.996995169185003e-05, "loss": 3.644, "step": 5313 }, { "epoch": 0.32987770811347694, "grad_norm": 0.7663006817732642, "learning_rate": 9.996982637275529e-05, "loss": 3.639, "step": 5314 }, { "epoch": 0.32993978521323486, "grad_norm": 0.6910917137783549, "learning_rate": 9.996970079295584e-05, "loss": 3.5748, "step": 5315 }, { "epoch": 0.3300018623129927, "grad_norm": 0.6715925865125986, "learning_rate": 9.996957495245232e-05, "loss": 3.656, "step": 5316 }, { "epoch": 0.33006393941275064, "grad_norm": 0.7137230598456207, "learning_rate": 9.996944885124541e-05, "loss": 3.6834, "step": 5317 }, { "epoch": 0.33012601651250856, "grad_norm": 0.6527025754934175, "learning_rate": 9.996932248933575e-05, "loss": 3.4823, "step": 5318 }, { "epoch": 0.3301880936122664, "grad_norm": 0.527970005744012, "learning_rate": 9.996919586672401e-05, "loss": 3.6541, "step": 5319 }, { "epoch": 0.33025017071202434, "grad_norm": 0.7015662824023264, "learning_rate": 9.996906898341084e-05, "loss": 3.7593, "step": 5320 }, { "epoch": 0.33031224781178226, "grad_norm": 0.6688805319055259, "learning_rate": 9.996894183939691e-05, "loss": 3.7279, "step": 5321 }, { "epoch": 0.3303743249115401, "grad_norm": 0.6507649990951553, "learning_rate": 9.996881443468286e-05, "loss": 3.6799, "step": 5322 }, { "epoch": 0.33043640201129804, "grad_norm": 0.6335419347909471, "learning_rate": 9.99686867692694e-05, "loss": 3.7452, "step": 5323 }, { "epoch": 0.33049847911105595, "grad_norm": 0.5847175104379779, "learning_rate": 9.996855884315716e-05, "loss": 3.5602, "step": 5324 }, { "epoch": 0.3305605562108138, "grad_norm": 0.6393026800636791, "learning_rate": 9.996843065634682e-05, "loss": 3.5949, "step": 5325 }, { "epoch": 0.33062263331057173, "grad_norm": 0.7451636223679295, "learning_rate": 9.996830220883905e-05, "loss": 3.5912, "step": 5326 }, { "epoch": 0.33068471041032965, "grad_norm": 0.8097357678648097, "learning_rate": 9.996817350063452e-05, "loss": 3.5968, "step": 5327 }, { "epoch": 0.3307467875100875, "grad_norm": 0.7556339686818311, "learning_rate": 9.996804453173388e-05, "loss": 3.5773, "step": 5328 }, { "epoch": 0.33080886460984543, "grad_norm": 0.6947231366639048, "learning_rate": 9.996791530213783e-05, "loss": 3.6384, "step": 5329 }, { "epoch": 0.33087094170960335, "grad_norm": 0.6805270260931444, "learning_rate": 9.996778581184704e-05, "loss": 3.5883, "step": 5330 }, { "epoch": 0.3309330188093612, "grad_norm": 0.619246368523839, "learning_rate": 9.996765606086218e-05, "loss": 3.5836, "step": 5331 }, { "epoch": 0.33099509590911913, "grad_norm": 0.538365424762547, "learning_rate": 9.996752604918394e-05, "loss": 3.6302, "step": 5332 }, { "epoch": 0.33105717300887705, "grad_norm": 0.6060110309402601, "learning_rate": 9.996739577681297e-05, "loss": 3.5282, "step": 5333 }, { "epoch": 0.3311192501086349, "grad_norm": 0.6547880691816266, "learning_rate": 9.996726524374998e-05, "loss": 3.627, "step": 5334 }, { "epoch": 0.33118132720839283, "grad_norm": 0.5707933410733957, "learning_rate": 9.996713444999564e-05, "loss": 3.5853, "step": 5335 }, { "epoch": 0.33124340430815075, "grad_norm": 0.6565582822857579, "learning_rate": 9.99670033955506e-05, "loss": 3.6227, "step": 5336 }, { "epoch": 0.3313054814079086, "grad_norm": 0.6365033352847639, "learning_rate": 9.99668720804156e-05, "loss": 3.6276, "step": 5337 }, { "epoch": 0.3313675585076665, "grad_norm": 0.6003221878628321, "learning_rate": 9.996674050459129e-05, "loss": 3.4915, "step": 5338 }, { "epoch": 0.33142963560742444, "grad_norm": 0.5449681416220544, "learning_rate": 9.996660866807837e-05, "loss": 3.6596, "step": 5339 }, { "epoch": 0.3314917127071823, "grad_norm": 0.5771331785092244, "learning_rate": 9.996647657087753e-05, "loss": 3.6002, "step": 5340 }, { "epoch": 0.3315537898069402, "grad_norm": 0.585693308491166, "learning_rate": 9.996634421298942e-05, "loss": 3.6635, "step": 5341 }, { "epoch": 0.33161586690669814, "grad_norm": 0.6968803944400301, "learning_rate": 9.99662115944148e-05, "loss": 3.6191, "step": 5342 }, { "epoch": 0.331677944006456, "grad_norm": 0.7472921204244319, "learning_rate": 9.996607871515429e-05, "loss": 3.6679, "step": 5343 }, { "epoch": 0.3317400211062139, "grad_norm": 0.7014935533930197, "learning_rate": 9.996594557520864e-05, "loss": 3.5668, "step": 5344 }, { "epoch": 0.33180209820597184, "grad_norm": 0.6938330523291568, "learning_rate": 9.996581217457852e-05, "loss": 3.7338, "step": 5345 }, { "epoch": 0.3318641753057297, "grad_norm": 0.677896793482932, "learning_rate": 9.996567851326463e-05, "loss": 3.6349, "step": 5346 }, { "epoch": 0.3319262524054876, "grad_norm": 0.5745740038100746, "learning_rate": 9.996554459126765e-05, "loss": 3.5071, "step": 5347 }, { "epoch": 0.33198832950524554, "grad_norm": 0.5599414067609504, "learning_rate": 9.996541040858831e-05, "loss": 3.6171, "step": 5348 }, { "epoch": 0.3320504066050034, "grad_norm": 0.5807312650314761, "learning_rate": 9.99652759652273e-05, "loss": 3.5751, "step": 5349 }, { "epoch": 0.3321124837047613, "grad_norm": 0.5909460769345585, "learning_rate": 9.996514126118531e-05, "loss": 3.5851, "step": 5350 }, { "epoch": 0.33217456080451924, "grad_norm": 0.533549549064914, "learning_rate": 9.996500629646303e-05, "loss": 3.6481, "step": 5351 }, { "epoch": 0.3322366379042771, "grad_norm": 0.542037082221774, "learning_rate": 9.996487107106121e-05, "loss": 3.5947, "step": 5352 }, { "epoch": 0.332298715004035, "grad_norm": 0.6155298401535179, "learning_rate": 9.99647355849805e-05, "loss": 3.4965, "step": 5353 }, { "epoch": 0.33236079210379293, "grad_norm": 0.5872689442085957, "learning_rate": 9.996459983822165e-05, "loss": 3.5951, "step": 5354 }, { "epoch": 0.3324228692035508, "grad_norm": 0.6064384570632675, "learning_rate": 9.996446383078535e-05, "loss": 3.5894, "step": 5355 }, { "epoch": 0.3324849463033087, "grad_norm": 0.495713411048602, "learning_rate": 9.996432756267232e-05, "loss": 3.7061, "step": 5356 }, { "epoch": 0.33254702340306663, "grad_norm": 0.544945665065787, "learning_rate": 9.996419103388326e-05, "loss": 3.6102, "step": 5357 }, { "epoch": 0.3326091005028245, "grad_norm": 0.4909128304541369, "learning_rate": 9.996405424441889e-05, "loss": 3.5624, "step": 5358 }, { "epoch": 0.3326711776025824, "grad_norm": 0.4975667247164, "learning_rate": 9.996391719427991e-05, "loss": 3.5401, "step": 5359 }, { "epoch": 0.33273325470234033, "grad_norm": 0.5620970077161507, "learning_rate": 9.996377988346706e-05, "loss": 3.594, "step": 5360 }, { "epoch": 0.3327953318020982, "grad_norm": 0.5770872958307758, "learning_rate": 9.996364231198103e-05, "loss": 3.5388, "step": 5361 }, { "epoch": 0.3328574089018561, "grad_norm": 0.4956904596825495, "learning_rate": 9.996350447982255e-05, "loss": 3.5167, "step": 5362 }, { "epoch": 0.33291948600161403, "grad_norm": 0.5479558073744665, "learning_rate": 9.996336638699234e-05, "loss": 3.5652, "step": 5363 }, { "epoch": 0.3329815631013719, "grad_norm": 0.5493083495502921, "learning_rate": 9.996322803349112e-05, "loss": 3.6316, "step": 5364 }, { "epoch": 0.3330436402011298, "grad_norm": 0.5843450905800656, "learning_rate": 9.99630894193196e-05, "loss": 3.6457, "step": 5365 }, { "epoch": 0.3331057173008877, "grad_norm": 0.6671177088307068, "learning_rate": 9.996295054447853e-05, "loss": 3.5605, "step": 5366 }, { "epoch": 0.3331677944006456, "grad_norm": 0.6797488831612271, "learning_rate": 9.996281140896861e-05, "loss": 3.5287, "step": 5367 }, { "epoch": 0.3332298715004035, "grad_norm": 0.6227217875490942, "learning_rate": 9.996267201279056e-05, "loss": 3.659, "step": 5368 }, { "epoch": 0.3332919486001614, "grad_norm": 0.5361122447030969, "learning_rate": 9.996253235594515e-05, "loss": 3.564, "step": 5369 }, { "epoch": 0.3333540256999193, "grad_norm": 0.5755834646078007, "learning_rate": 9.996239243843306e-05, "loss": 3.4525, "step": 5370 }, { "epoch": 0.3334161027996772, "grad_norm": 0.5796157072703184, "learning_rate": 9.996225226025505e-05, "loss": 3.673, "step": 5371 }, { "epoch": 0.3334781798994351, "grad_norm": 0.5697172272041278, "learning_rate": 9.996211182141185e-05, "loss": 3.6788, "step": 5372 }, { "epoch": 0.333540256999193, "grad_norm": 0.5628204227802761, "learning_rate": 9.996197112190415e-05, "loss": 3.541, "step": 5373 }, { "epoch": 0.3336023340989509, "grad_norm": 0.6205834745746136, "learning_rate": 9.996183016173273e-05, "loss": 3.6858, "step": 5374 }, { "epoch": 0.3336644111987088, "grad_norm": 0.6143509107428825, "learning_rate": 9.996168894089834e-05, "loss": 3.6456, "step": 5375 }, { "epoch": 0.3337264882984667, "grad_norm": 0.6215179482244528, "learning_rate": 9.996154745940168e-05, "loss": 3.5953, "step": 5376 }, { "epoch": 0.3337885653982246, "grad_norm": 0.6164179966415299, "learning_rate": 9.996140571724347e-05, "loss": 3.6329, "step": 5377 }, { "epoch": 0.3338506424979825, "grad_norm": 0.6322078641967133, "learning_rate": 9.99612637144245e-05, "loss": 3.5698, "step": 5378 }, { "epoch": 0.3339127195977404, "grad_norm": 0.5842829932746725, "learning_rate": 9.996112145094549e-05, "loss": 3.6319, "step": 5379 }, { "epoch": 0.3339747966974983, "grad_norm": 0.6780192191598556, "learning_rate": 9.996097892680718e-05, "loss": 3.6134, "step": 5380 }, { "epoch": 0.3340368737972562, "grad_norm": 0.6474072878605037, "learning_rate": 9.996083614201031e-05, "loss": 3.5439, "step": 5381 }, { "epoch": 0.3340989508970141, "grad_norm": 0.8922272374377844, "learning_rate": 9.996069309655562e-05, "loss": 3.6573, "step": 5382 }, { "epoch": 0.334161027996772, "grad_norm": 0.8967792300354325, "learning_rate": 9.99605497904439e-05, "loss": 3.553, "step": 5383 }, { "epoch": 0.3342231050965299, "grad_norm": 0.7355159661863535, "learning_rate": 9.996040622367582e-05, "loss": 3.6371, "step": 5384 }, { "epoch": 0.3342851821962878, "grad_norm": 0.7715447510863673, "learning_rate": 9.99602623962522e-05, "loss": 3.5755, "step": 5385 }, { "epoch": 0.3343472592960457, "grad_norm": 0.6357274676375292, "learning_rate": 9.996011830817376e-05, "loss": 3.5736, "step": 5386 }, { "epoch": 0.3344093363958036, "grad_norm": 0.5529728028126538, "learning_rate": 9.995997395944125e-05, "loss": 3.5549, "step": 5387 }, { "epoch": 0.3344714134955615, "grad_norm": 0.5933269198031744, "learning_rate": 9.995982935005542e-05, "loss": 3.5547, "step": 5388 }, { "epoch": 0.3345334905953194, "grad_norm": 0.5719945480380294, "learning_rate": 9.995968448001704e-05, "loss": 3.5954, "step": 5389 }, { "epoch": 0.3345955676950773, "grad_norm": 0.6086353067452607, "learning_rate": 9.995953934932686e-05, "loss": 3.5648, "step": 5390 }, { "epoch": 0.3346576447948352, "grad_norm": 0.6021015674861467, "learning_rate": 9.995939395798564e-05, "loss": 3.5909, "step": 5391 }, { "epoch": 0.3347197218945931, "grad_norm": 0.5294317657769011, "learning_rate": 9.995924830599413e-05, "loss": 3.5535, "step": 5392 }, { "epoch": 0.334781798994351, "grad_norm": 0.6234749546129478, "learning_rate": 9.995910239335308e-05, "loss": 3.5716, "step": 5393 }, { "epoch": 0.33484387609410887, "grad_norm": 0.5252658738429802, "learning_rate": 9.995895622006328e-05, "loss": 3.6003, "step": 5394 }, { "epoch": 0.3349059531938668, "grad_norm": 0.5070082531432211, "learning_rate": 9.995880978612546e-05, "loss": 3.593, "step": 5395 }, { "epoch": 0.3349680302936247, "grad_norm": 0.5932861071938585, "learning_rate": 9.995866309154041e-05, "loss": 3.7743, "step": 5396 }, { "epoch": 0.33503010739338257, "grad_norm": 0.52211517586483, "learning_rate": 9.99585161363089e-05, "loss": 3.6303, "step": 5397 }, { "epoch": 0.3350921844931405, "grad_norm": 0.6341327505361806, "learning_rate": 9.995836892043166e-05, "loss": 3.6036, "step": 5398 }, { "epoch": 0.3351542615928984, "grad_norm": 0.6528906899279062, "learning_rate": 9.995822144390952e-05, "loss": 3.5762, "step": 5399 }, { "epoch": 0.33521633869265627, "grad_norm": 0.5294269810076188, "learning_rate": 9.995807370674318e-05, "loss": 3.5148, "step": 5400 }, { "epoch": 0.3352784157924142, "grad_norm": 0.567076740011677, "learning_rate": 9.995792570893344e-05, "loss": 3.5884, "step": 5401 }, { "epoch": 0.3353404928921721, "grad_norm": 0.6929702741569608, "learning_rate": 9.995777745048109e-05, "loss": 3.4875, "step": 5402 }, { "epoch": 0.33540256999192997, "grad_norm": 0.7133751045976946, "learning_rate": 9.995762893138687e-05, "loss": 3.6982, "step": 5403 }, { "epoch": 0.3354646470916879, "grad_norm": 0.7256506253103315, "learning_rate": 9.995748015165158e-05, "loss": 3.59, "step": 5404 }, { "epoch": 0.3355267241914458, "grad_norm": 0.6077990572112122, "learning_rate": 9.9957331111276e-05, "loss": 3.587, "step": 5405 }, { "epoch": 0.33558880129120366, "grad_norm": 0.5399783274487076, "learning_rate": 9.995718181026088e-05, "loss": 3.5127, "step": 5406 }, { "epoch": 0.3356508783909616, "grad_norm": 0.616782961224008, "learning_rate": 9.995703224860702e-05, "loss": 3.6254, "step": 5407 }, { "epoch": 0.3357129554907195, "grad_norm": 0.7855729472928284, "learning_rate": 9.99568824263152e-05, "loss": 3.6861, "step": 5408 }, { "epoch": 0.33577503259047736, "grad_norm": 0.7143447468409948, "learning_rate": 9.99567323433862e-05, "loss": 3.6645, "step": 5409 }, { "epoch": 0.3358371096902353, "grad_norm": 0.6261627110925008, "learning_rate": 9.995658199982079e-05, "loss": 3.6177, "step": 5410 }, { "epoch": 0.3358991867899932, "grad_norm": 0.6945877155225977, "learning_rate": 9.995643139561976e-05, "loss": 3.6251, "step": 5411 }, { "epoch": 0.33596126388975106, "grad_norm": 0.6365339795082695, "learning_rate": 9.995628053078392e-05, "loss": 3.5153, "step": 5412 }, { "epoch": 0.336023340989509, "grad_norm": 0.6466572233703481, "learning_rate": 9.995612940531401e-05, "loss": 3.636, "step": 5413 }, { "epoch": 0.3360854180892669, "grad_norm": 0.5575459663617404, "learning_rate": 9.995597801921087e-05, "loss": 3.6223, "step": 5414 }, { "epoch": 0.33614749518902476, "grad_norm": 0.48883819859900784, "learning_rate": 9.995582637247525e-05, "loss": 3.4658, "step": 5415 }, { "epoch": 0.3362095722887827, "grad_norm": 0.6624004883074458, "learning_rate": 9.995567446510796e-05, "loss": 3.5704, "step": 5416 }, { "epoch": 0.3362716493885406, "grad_norm": 0.7185558803306855, "learning_rate": 9.99555222971098e-05, "loss": 3.6445, "step": 5417 }, { "epoch": 0.33633372648829846, "grad_norm": 0.6899666103286839, "learning_rate": 9.995536986848155e-05, "loss": 3.6349, "step": 5418 }, { "epoch": 0.3363958035880564, "grad_norm": 0.6521285153133913, "learning_rate": 9.995521717922399e-05, "loss": 3.6316, "step": 5419 }, { "epoch": 0.3364578806878143, "grad_norm": 0.6110952012803926, "learning_rate": 9.995506422933795e-05, "loss": 3.5754, "step": 5420 }, { "epoch": 0.33651995778757215, "grad_norm": 0.6057450187185459, "learning_rate": 9.995491101882422e-05, "loss": 3.5642, "step": 5421 }, { "epoch": 0.3365820348873301, "grad_norm": 0.6259872268982379, "learning_rate": 9.995475754768357e-05, "loss": 3.593, "step": 5422 }, { "epoch": 0.336644111987088, "grad_norm": 0.6046789370775757, "learning_rate": 9.995460381591683e-05, "loss": 3.5642, "step": 5423 }, { "epoch": 0.33670618908684585, "grad_norm": 0.5654607312998139, "learning_rate": 9.995444982352482e-05, "loss": 3.6333, "step": 5424 }, { "epoch": 0.33676826618660377, "grad_norm": 0.5961031961754998, "learning_rate": 9.99542955705083e-05, "loss": 3.5706, "step": 5425 }, { "epoch": 0.3368303432863617, "grad_norm": 0.6221907183239642, "learning_rate": 9.995414105686809e-05, "loss": 3.6274, "step": 5426 }, { "epoch": 0.33689242038611955, "grad_norm": 0.6631339818434272, "learning_rate": 9.9953986282605e-05, "loss": 3.6215, "step": 5427 }, { "epoch": 0.33695449748587747, "grad_norm": 0.5976115440652242, "learning_rate": 9.995383124771984e-05, "loss": 3.5681, "step": 5428 }, { "epoch": 0.3370165745856354, "grad_norm": 0.5906833458394036, "learning_rate": 9.995367595221342e-05, "loss": 3.6395, "step": 5429 }, { "epoch": 0.33707865168539325, "grad_norm": 0.6499728673994075, "learning_rate": 9.995352039608653e-05, "loss": 3.6004, "step": 5430 }, { "epoch": 0.33714072878515117, "grad_norm": 0.581771434167764, "learning_rate": 9.995336457934001e-05, "loss": 3.6291, "step": 5431 }, { "epoch": 0.3372028058849091, "grad_norm": 0.5938454852372671, "learning_rate": 9.995320850197465e-05, "loss": 3.6102, "step": 5432 }, { "epoch": 0.33726488298466695, "grad_norm": 0.5637573653203088, "learning_rate": 9.995305216399129e-05, "loss": 3.6535, "step": 5433 }, { "epoch": 0.33732696008442487, "grad_norm": 0.5193886238450461, "learning_rate": 9.995289556539072e-05, "loss": 3.6281, "step": 5434 }, { "epoch": 0.3373890371841827, "grad_norm": 0.5872080950249179, "learning_rate": 9.995273870617377e-05, "loss": 3.583, "step": 5435 }, { "epoch": 0.33745111428394065, "grad_norm": 0.5489718971772425, "learning_rate": 9.995258158634127e-05, "loss": 3.623, "step": 5436 }, { "epoch": 0.33751319138369856, "grad_norm": 0.5871481785543029, "learning_rate": 9.995242420589398e-05, "loss": 3.57, "step": 5437 }, { "epoch": 0.3375752684834564, "grad_norm": 0.6077400429104958, "learning_rate": 9.995226656483282e-05, "loss": 3.5929, "step": 5438 }, { "epoch": 0.33763734558321434, "grad_norm": 0.5836687798125035, "learning_rate": 9.995210866315854e-05, "loss": 3.5582, "step": 5439 }, { "epoch": 0.33769942268297226, "grad_norm": 0.651539678196797, "learning_rate": 9.995195050087197e-05, "loss": 3.6474, "step": 5440 }, { "epoch": 0.3377614997827301, "grad_norm": 0.6277564775413013, "learning_rate": 9.995179207797396e-05, "loss": 3.5491, "step": 5441 }, { "epoch": 0.33782357688248804, "grad_norm": 0.5680363054778247, "learning_rate": 9.995163339446531e-05, "loss": 3.603, "step": 5442 }, { "epoch": 0.33788565398224596, "grad_norm": 0.6488153896195783, "learning_rate": 9.99514744503469e-05, "loss": 3.4817, "step": 5443 }, { "epoch": 0.3379477310820038, "grad_norm": 0.6103447753528882, "learning_rate": 9.995131524561949e-05, "loss": 3.5399, "step": 5444 }, { "epoch": 0.33800980818176174, "grad_norm": 0.5606377026278775, "learning_rate": 9.995115578028394e-05, "loss": 3.6073, "step": 5445 }, { "epoch": 0.33807188528151966, "grad_norm": 0.6106871879950853, "learning_rate": 9.995099605434109e-05, "loss": 3.5781, "step": 5446 }, { "epoch": 0.3381339623812775, "grad_norm": 0.6625948001403648, "learning_rate": 9.995083606779177e-05, "loss": 3.6575, "step": 5447 }, { "epoch": 0.33819603948103544, "grad_norm": 0.5817754810754353, "learning_rate": 9.995067582063682e-05, "loss": 3.5358, "step": 5448 }, { "epoch": 0.33825811658079336, "grad_norm": 0.580034016688656, "learning_rate": 9.995051531287706e-05, "loss": 3.4642, "step": 5449 }, { "epoch": 0.3383201936805512, "grad_norm": 0.6204596827701149, "learning_rate": 9.995035454451334e-05, "loss": 3.6347, "step": 5450 }, { "epoch": 0.33838227078030914, "grad_norm": 0.700425624877115, "learning_rate": 9.995019351554649e-05, "loss": 3.4482, "step": 5451 }, { "epoch": 0.33844434788006705, "grad_norm": 0.6870151674228807, "learning_rate": 9.995003222597736e-05, "loss": 3.499, "step": 5452 }, { "epoch": 0.3385064249798249, "grad_norm": 0.7320000412841952, "learning_rate": 9.994987067580678e-05, "loss": 3.6303, "step": 5453 }, { "epoch": 0.33856850207958283, "grad_norm": 0.7291476701350926, "learning_rate": 9.99497088650356e-05, "loss": 3.5338, "step": 5454 }, { "epoch": 0.33863057917934075, "grad_norm": 0.6549309462262987, "learning_rate": 9.994954679366466e-05, "loss": 3.61, "step": 5455 }, { "epoch": 0.3386926562790986, "grad_norm": 0.5824191445811774, "learning_rate": 9.994938446169482e-05, "loss": 3.6043, "step": 5456 }, { "epoch": 0.33875473337885653, "grad_norm": 0.5721843778982357, "learning_rate": 9.99492218691269e-05, "loss": 3.4925, "step": 5457 }, { "epoch": 0.33881681047861445, "grad_norm": 0.7171316586365342, "learning_rate": 9.994905901596178e-05, "loss": 3.6747, "step": 5458 }, { "epoch": 0.3388788875783723, "grad_norm": 0.598434813769813, "learning_rate": 9.994889590220029e-05, "loss": 3.5477, "step": 5459 }, { "epoch": 0.33894096467813023, "grad_norm": 0.589567536613103, "learning_rate": 9.994873252784327e-05, "loss": 3.6595, "step": 5460 }, { "epoch": 0.33900304177788815, "grad_norm": 0.6799612042285245, "learning_rate": 9.994856889289159e-05, "loss": 3.789, "step": 5461 }, { "epoch": 0.339065118877646, "grad_norm": 0.6384873429838478, "learning_rate": 9.994840499734612e-05, "loss": 3.5721, "step": 5462 }, { "epoch": 0.33912719597740393, "grad_norm": 0.6212154005382436, "learning_rate": 9.994824084120768e-05, "loss": 3.5689, "step": 5463 }, { "epoch": 0.33918927307716185, "grad_norm": 0.7408518769639949, "learning_rate": 9.994807642447714e-05, "loss": 3.6635, "step": 5464 }, { "epoch": 0.3392513501769197, "grad_norm": 0.7021414553490101, "learning_rate": 9.994791174715536e-05, "loss": 3.4929, "step": 5465 }, { "epoch": 0.3393134272766776, "grad_norm": 0.6882638372671577, "learning_rate": 9.994774680924321e-05, "loss": 3.5626, "step": 5466 }, { "epoch": 0.33937550437643554, "grad_norm": 0.6297720641722792, "learning_rate": 9.994758161074154e-05, "loss": 3.541, "step": 5467 }, { "epoch": 0.3394375814761934, "grad_norm": 0.6032851930463246, "learning_rate": 9.994741615165121e-05, "loss": 3.6318, "step": 5468 }, { "epoch": 0.3394996585759513, "grad_norm": 0.7265093794241516, "learning_rate": 9.994725043197309e-05, "loss": 3.5981, "step": 5469 }, { "epoch": 0.33956173567570924, "grad_norm": 0.6408089396947926, "learning_rate": 9.994708445170802e-05, "loss": 3.6205, "step": 5470 }, { "epoch": 0.3396238127754671, "grad_norm": 0.6510008506696098, "learning_rate": 9.994691821085688e-05, "loss": 3.58, "step": 5471 }, { "epoch": 0.339685889875225, "grad_norm": 0.611811924801524, "learning_rate": 9.994675170942058e-05, "loss": 3.4853, "step": 5472 }, { "epoch": 0.33974796697498294, "grad_norm": 0.5436380501305896, "learning_rate": 9.994658494739992e-05, "loss": 3.5686, "step": 5473 }, { "epoch": 0.3398100440747408, "grad_norm": 0.6519135752864729, "learning_rate": 9.994641792479581e-05, "loss": 3.5797, "step": 5474 }, { "epoch": 0.3398721211744987, "grad_norm": 0.6297719458662806, "learning_rate": 9.994625064160912e-05, "loss": 3.5907, "step": 5475 }, { "epoch": 0.33993419827425664, "grad_norm": 0.5786180970840595, "learning_rate": 9.994608309784071e-05, "loss": 3.5521, "step": 5476 }, { "epoch": 0.3399962753740145, "grad_norm": 0.6139721956900488, "learning_rate": 9.994591529349145e-05, "loss": 3.431, "step": 5477 }, { "epoch": 0.3400583524737724, "grad_norm": 0.563376485575056, "learning_rate": 9.994574722856223e-05, "loss": 3.6369, "step": 5478 }, { "epoch": 0.34012042957353034, "grad_norm": 0.47806867629260835, "learning_rate": 9.994557890305393e-05, "loss": 3.5688, "step": 5479 }, { "epoch": 0.3401825066732882, "grad_norm": 0.5329902140003496, "learning_rate": 9.994541031696742e-05, "loss": 3.5511, "step": 5480 }, { "epoch": 0.3402445837730461, "grad_norm": 0.6097923463566307, "learning_rate": 9.994524147030357e-05, "loss": 3.6895, "step": 5481 }, { "epoch": 0.34030666087280403, "grad_norm": 0.5540852568159808, "learning_rate": 9.994507236306327e-05, "loss": 3.5825, "step": 5482 }, { "epoch": 0.3403687379725619, "grad_norm": 0.5221755629886703, "learning_rate": 9.994490299524741e-05, "loss": 3.6072, "step": 5483 }, { "epoch": 0.3404308150723198, "grad_norm": 0.5265872474084655, "learning_rate": 9.994473336685688e-05, "loss": 3.5441, "step": 5484 }, { "epoch": 0.34049289217207773, "grad_norm": 0.5265664482534654, "learning_rate": 9.994456347789254e-05, "loss": 3.6056, "step": 5485 }, { "epoch": 0.3405549692718356, "grad_norm": 0.6784301396501761, "learning_rate": 9.994439332835528e-05, "loss": 3.5944, "step": 5486 }, { "epoch": 0.3406170463715935, "grad_norm": 0.7882755245307047, "learning_rate": 9.9944222918246e-05, "loss": 3.5819, "step": 5487 }, { "epoch": 0.34067912347135143, "grad_norm": 0.6290623488902154, "learning_rate": 9.994405224756557e-05, "loss": 3.5176, "step": 5488 }, { "epoch": 0.3407412005711093, "grad_norm": 0.5899503750459533, "learning_rate": 9.994388131631492e-05, "loss": 3.5967, "step": 5489 }, { "epoch": 0.3408032776708672, "grad_norm": 0.7120488461766024, "learning_rate": 9.994371012449492e-05, "loss": 3.5393, "step": 5490 }, { "epoch": 0.34086535477062513, "grad_norm": 0.7363205158032728, "learning_rate": 9.994353867210643e-05, "loss": 3.6371, "step": 5491 }, { "epoch": 0.340927431870383, "grad_norm": 0.7487307378236808, "learning_rate": 9.99433669591504e-05, "loss": 3.6361, "step": 5492 }, { "epoch": 0.3409895089701409, "grad_norm": 0.7385245431942452, "learning_rate": 9.99431949856277e-05, "loss": 3.6067, "step": 5493 }, { "epoch": 0.3410515860698988, "grad_norm": 0.6904532982918105, "learning_rate": 9.994302275153922e-05, "loss": 3.6034, "step": 5494 }, { "epoch": 0.3411136631696567, "grad_norm": 0.6959486633556529, "learning_rate": 9.994285025688585e-05, "loss": 3.583, "step": 5495 }, { "epoch": 0.3411757402694146, "grad_norm": 0.7173046008149828, "learning_rate": 9.994267750166854e-05, "loss": 3.5025, "step": 5496 }, { "epoch": 0.3412378173691725, "grad_norm": 0.6572327067979298, "learning_rate": 9.994250448588814e-05, "loss": 3.5287, "step": 5497 }, { "epoch": 0.3412998944689304, "grad_norm": 0.6584672256516533, "learning_rate": 9.994233120954558e-05, "loss": 3.6346, "step": 5498 }, { "epoch": 0.3413619715686883, "grad_norm": 0.6531570070810878, "learning_rate": 9.994215767264172e-05, "loss": 3.6148, "step": 5499 }, { "epoch": 0.3414240486684462, "grad_norm": 0.5867146742027727, "learning_rate": 9.994198387517752e-05, "loss": 3.701, "step": 5500 }, { "epoch": 0.3414861257682041, "grad_norm": 0.5950736300005827, "learning_rate": 9.994180981715388e-05, "loss": 3.6099, "step": 5501 }, { "epoch": 0.341548202867962, "grad_norm": 0.5391920182912194, "learning_rate": 9.994163549857168e-05, "loss": 3.542, "step": 5502 }, { "epoch": 0.3416102799677199, "grad_norm": 0.5627208382173781, "learning_rate": 9.994146091943186e-05, "loss": 3.6223, "step": 5503 }, { "epoch": 0.3416723570674778, "grad_norm": 0.57409931419915, "learning_rate": 9.99412860797353e-05, "loss": 3.5621, "step": 5504 }, { "epoch": 0.3417344341672357, "grad_norm": 0.5696346748958531, "learning_rate": 9.994111097948294e-05, "loss": 3.5472, "step": 5505 }, { "epoch": 0.3417965112669936, "grad_norm": 0.5602408713635841, "learning_rate": 9.994093561867566e-05, "loss": 3.6046, "step": 5506 }, { "epoch": 0.3418585883667515, "grad_norm": 0.5010851766473203, "learning_rate": 9.99407599973144e-05, "loss": 3.6067, "step": 5507 }, { "epoch": 0.3419206654665094, "grad_norm": 0.5613391872476043, "learning_rate": 9.994058411540009e-05, "loss": 3.5617, "step": 5508 }, { "epoch": 0.3419827425662673, "grad_norm": 0.4796125956152397, "learning_rate": 9.99404079729336e-05, "loss": 3.5757, "step": 5509 }, { "epoch": 0.3420448196660252, "grad_norm": 0.539160070020285, "learning_rate": 9.99402315699159e-05, "loss": 3.6527, "step": 5510 }, { "epoch": 0.3421068967657831, "grad_norm": 0.46239009788595803, "learning_rate": 9.994005490634788e-05, "loss": 3.593, "step": 5511 }, { "epoch": 0.342168973865541, "grad_norm": 0.47132519023480474, "learning_rate": 9.993987798223048e-05, "loss": 3.5441, "step": 5512 }, { "epoch": 0.3422310509652989, "grad_norm": 0.5556910127831864, "learning_rate": 9.993970079756461e-05, "loss": 3.5848, "step": 5513 }, { "epoch": 0.3422931280650568, "grad_norm": 0.5359388045581477, "learning_rate": 9.99395233523512e-05, "loss": 3.562, "step": 5514 }, { "epoch": 0.3423552051648147, "grad_norm": 0.5338851144395681, "learning_rate": 9.993934564659115e-05, "loss": 3.6453, "step": 5515 }, { "epoch": 0.3424172822645726, "grad_norm": 0.52206163210062, "learning_rate": 9.993916768028543e-05, "loss": 3.5807, "step": 5516 }, { "epoch": 0.3424793593643305, "grad_norm": 0.5597747642266153, "learning_rate": 9.993898945343495e-05, "loss": 3.5812, "step": 5517 }, { "epoch": 0.3425414364640884, "grad_norm": 0.6152567299308512, "learning_rate": 9.993881096604064e-05, "loss": 3.4955, "step": 5518 }, { "epoch": 0.3426035135638463, "grad_norm": 0.5874447319737433, "learning_rate": 9.993863221810342e-05, "loss": 3.6327, "step": 5519 }, { "epoch": 0.3426655906636042, "grad_norm": 0.5862022056292763, "learning_rate": 9.993845320962424e-05, "loss": 3.5844, "step": 5520 }, { "epoch": 0.3427276677633621, "grad_norm": 0.5994832406346094, "learning_rate": 9.993827394060402e-05, "loss": 3.6726, "step": 5521 }, { "epoch": 0.34278974486311997, "grad_norm": 0.6680523111824603, "learning_rate": 9.993809441104371e-05, "loss": 3.5903, "step": 5522 }, { "epoch": 0.3428518219628779, "grad_norm": 0.7819257484024359, "learning_rate": 9.993791462094424e-05, "loss": 3.5621, "step": 5523 }, { "epoch": 0.3429138990626358, "grad_norm": 0.7274050185870825, "learning_rate": 9.993773457030653e-05, "loss": 3.4905, "step": 5524 }, { "epoch": 0.34297597616239367, "grad_norm": 0.5560948959300193, "learning_rate": 9.993755425913155e-05, "loss": 3.5696, "step": 5525 }, { "epoch": 0.3430380532621516, "grad_norm": 0.619811407677766, "learning_rate": 9.993737368742023e-05, "loss": 3.6574, "step": 5526 }, { "epoch": 0.3431001303619095, "grad_norm": 0.7288054025440359, "learning_rate": 9.993719285517351e-05, "loss": 3.6108, "step": 5527 }, { "epoch": 0.34316220746166737, "grad_norm": 0.7385517815532584, "learning_rate": 9.993701176239232e-05, "loss": 3.5945, "step": 5528 }, { "epoch": 0.3432242845614253, "grad_norm": 0.5957535018882008, "learning_rate": 9.993683040907762e-05, "loss": 3.6224, "step": 5529 }, { "epoch": 0.3432863616611832, "grad_norm": 0.5904039665674308, "learning_rate": 9.993664879523036e-05, "loss": 3.6433, "step": 5530 }, { "epoch": 0.34334843876094107, "grad_norm": 0.6112847846027335, "learning_rate": 9.993646692085148e-05, "loss": 3.5999, "step": 5531 }, { "epoch": 0.343410515860699, "grad_norm": 0.554974924758266, "learning_rate": 9.993628478594193e-05, "loss": 3.6104, "step": 5532 }, { "epoch": 0.3434725929604569, "grad_norm": 0.6488488971983916, "learning_rate": 9.993610239050265e-05, "loss": 3.5896, "step": 5533 }, { "epoch": 0.34353467006021476, "grad_norm": 0.6502212899663049, "learning_rate": 9.99359197345346e-05, "loss": 3.5046, "step": 5534 }, { "epoch": 0.3435967471599727, "grad_norm": 0.6681754478414552, "learning_rate": 9.993573681803875e-05, "loss": 3.5515, "step": 5535 }, { "epoch": 0.3436588242597306, "grad_norm": 0.6906369989310783, "learning_rate": 9.993555364101603e-05, "loss": 3.4567, "step": 5536 }, { "epoch": 0.34372090135948846, "grad_norm": 0.7605194431681133, "learning_rate": 9.993537020346741e-05, "loss": 3.5901, "step": 5537 }, { "epoch": 0.3437829784592464, "grad_norm": 0.67533297361648, "learning_rate": 9.993518650539384e-05, "loss": 3.5652, "step": 5538 }, { "epoch": 0.3438450555590043, "grad_norm": 0.6743159200899994, "learning_rate": 9.993500254679627e-05, "loss": 3.6547, "step": 5539 }, { "epoch": 0.34390713265876216, "grad_norm": 0.7308816813618115, "learning_rate": 9.993481832767568e-05, "loss": 3.5153, "step": 5540 }, { "epoch": 0.3439692097585201, "grad_norm": 0.6744898546076571, "learning_rate": 9.993463384803302e-05, "loss": 3.5433, "step": 5541 }, { "epoch": 0.344031286858278, "grad_norm": 0.6516847079634204, "learning_rate": 9.993444910786924e-05, "loss": 3.5682, "step": 5542 }, { "epoch": 0.34409336395803586, "grad_norm": 0.5772660292163114, "learning_rate": 9.993426410718532e-05, "loss": 3.6742, "step": 5543 }, { "epoch": 0.3441554410577938, "grad_norm": 0.591599812446794, "learning_rate": 9.993407884598223e-05, "loss": 3.5512, "step": 5544 }, { "epoch": 0.3442175181575517, "grad_norm": 0.6323637371896352, "learning_rate": 9.993389332426093e-05, "loss": 3.57, "step": 5545 }, { "epoch": 0.34427959525730956, "grad_norm": 0.6195384290935864, "learning_rate": 9.993370754202239e-05, "loss": 3.6905, "step": 5546 }, { "epoch": 0.3443416723570675, "grad_norm": 0.5999260479334467, "learning_rate": 9.993352149926756e-05, "loss": 3.5088, "step": 5547 }, { "epoch": 0.3444037494568254, "grad_norm": 0.5444329368689983, "learning_rate": 9.993333519599743e-05, "loss": 3.5653, "step": 5548 }, { "epoch": 0.34446582655658325, "grad_norm": 0.49876341792153356, "learning_rate": 9.993314863221299e-05, "loss": 3.6081, "step": 5549 }, { "epoch": 0.3445279036563412, "grad_norm": 0.5851067311988695, "learning_rate": 9.993296180791518e-05, "loss": 3.6936, "step": 5550 }, { "epoch": 0.3445899807560991, "grad_norm": 0.5637237429795015, "learning_rate": 9.993277472310498e-05, "loss": 3.6334, "step": 5551 }, { "epoch": 0.34465205785585695, "grad_norm": 0.5603869539015908, "learning_rate": 9.993258737778337e-05, "loss": 3.6838, "step": 5552 }, { "epoch": 0.34471413495561487, "grad_norm": 0.6153226516606908, "learning_rate": 9.993239977195135e-05, "loss": 3.669, "step": 5553 }, { "epoch": 0.3447762120553728, "grad_norm": 0.5924959743137654, "learning_rate": 9.993221190560986e-05, "loss": 3.6943, "step": 5554 }, { "epoch": 0.34483828915513065, "grad_norm": 0.5812402252687954, "learning_rate": 9.99320237787599e-05, "loss": 3.6181, "step": 5555 }, { "epoch": 0.34490036625488857, "grad_norm": 0.47762024858273305, "learning_rate": 9.993183539140245e-05, "loss": 3.6292, "step": 5556 }, { "epoch": 0.3449624433546465, "grad_norm": 0.5728771080462737, "learning_rate": 9.99316467435385e-05, "loss": 3.5739, "step": 5557 }, { "epoch": 0.34502452045440435, "grad_norm": 0.5761720560364519, "learning_rate": 9.993145783516903e-05, "loss": 3.5975, "step": 5558 }, { "epoch": 0.34508659755416227, "grad_norm": 0.5610570944803694, "learning_rate": 9.993126866629502e-05, "loss": 3.6125, "step": 5559 }, { "epoch": 0.3451486746539202, "grad_norm": 0.6079477076441668, "learning_rate": 9.993107923691746e-05, "loss": 3.4862, "step": 5560 }, { "epoch": 0.34521075175367805, "grad_norm": 0.6463174287456659, "learning_rate": 9.993088954703734e-05, "loss": 3.6416, "step": 5561 }, { "epoch": 0.34527282885343596, "grad_norm": 0.5930514241560311, "learning_rate": 9.993069959665565e-05, "loss": 3.5242, "step": 5562 }, { "epoch": 0.3453349059531939, "grad_norm": 0.5673812104562016, "learning_rate": 9.993050938577338e-05, "loss": 3.5039, "step": 5563 }, { "epoch": 0.34539698305295174, "grad_norm": 0.5556691308620042, "learning_rate": 9.99303189143915e-05, "loss": 3.5415, "step": 5564 }, { "epoch": 0.34545906015270966, "grad_norm": 0.6364493941216751, "learning_rate": 9.993012818251105e-05, "loss": 3.591, "step": 5565 }, { "epoch": 0.3455211372524676, "grad_norm": 0.6235246649974512, "learning_rate": 9.992993719013299e-05, "loss": 3.6539, "step": 5566 }, { "epoch": 0.34558321435222544, "grad_norm": 0.686471755360125, "learning_rate": 9.992974593725832e-05, "loss": 3.5625, "step": 5567 }, { "epoch": 0.34564529145198336, "grad_norm": 0.7991135886953004, "learning_rate": 9.992955442388806e-05, "loss": 3.5912, "step": 5568 }, { "epoch": 0.3457073685517413, "grad_norm": 0.7532439806647337, "learning_rate": 9.992936265002318e-05, "loss": 3.6379, "step": 5569 }, { "epoch": 0.34576944565149914, "grad_norm": 0.7280074896558171, "learning_rate": 9.992917061566472e-05, "loss": 3.5704, "step": 5570 }, { "epoch": 0.34583152275125706, "grad_norm": 0.6594922490304336, "learning_rate": 9.992897832081363e-05, "loss": 3.6152, "step": 5571 }, { "epoch": 0.345893599851015, "grad_norm": 0.5651962773600022, "learning_rate": 9.992878576547093e-05, "loss": 3.5015, "step": 5572 }, { "epoch": 0.34595567695077284, "grad_norm": 0.6354947641458625, "learning_rate": 9.992859294963766e-05, "loss": 3.667, "step": 5573 }, { "epoch": 0.34601775405053076, "grad_norm": 0.6517730317869498, "learning_rate": 9.99283998733148e-05, "loss": 3.5759, "step": 5574 }, { "epoch": 0.3460798311502887, "grad_norm": 0.6213131642995686, "learning_rate": 9.992820653650335e-05, "loss": 3.5076, "step": 5575 }, { "epoch": 0.34614190825004654, "grad_norm": 0.6047393096493038, "learning_rate": 9.992801293920431e-05, "loss": 3.6148, "step": 5576 }, { "epoch": 0.34620398534980446, "grad_norm": 0.6491020887383878, "learning_rate": 9.992781908141873e-05, "loss": 3.5864, "step": 5577 }, { "epoch": 0.3462660624495624, "grad_norm": 0.6159476127162066, "learning_rate": 9.992762496314758e-05, "loss": 3.5194, "step": 5578 }, { "epoch": 0.34632813954932024, "grad_norm": 0.7532364632206453, "learning_rate": 9.99274305843919e-05, "loss": 3.6107, "step": 5579 }, { "epoch": 0.34639021664907815, "grad_norm": 0.7222389459711479, "learning_rate": 9.992723594515269e-05, "loss": 3.5159, "step": 5580 }, { "epoch": 0.34645229374883607, "grad_norm": 0.6779372161825252, "learning_rate": 9.992704104543098e-05, "loss": 3.5449, "step": 5581 }, { "epoch": 0.34651437084859393, "grad_norm": 0.6266266635375787, "learning_rate": 9.992684588522777e-05, "loss": 3.5605, "step": 5582 }, { "epoch": 0.34657644794835185, "grad_norm": 0.6268532931980475, "learning_rate": 9.992665046454408e-05, "loss": 3.5875, "step": 5583 }, { "epoch": 0.34663852504810977, "grad_norm": 0.7223125568365775, "learning_rate": 9.992645478338094e-05, "loss": 3.6492, "step": 5584 }, { "epoch": 0.34670060214786763, "grad_norm": 0.6431540137698263, "learning_rate": 9.992625884173936e-05, "loss": 3.4843, "step": 5585 }, { "epoch": 0.34676267924762555, "grad_norm": 0.6137844863805499, "learning_rate": 9.992606263962037e-05, "loss": 3.574, "step": 5586 }, { "epoch": 0.34682475634738347, "grad_norm": 0.7651629318684031, "learning_rate": 9.992586617702499e-05, "loss": 3.4926, "step": 5587 }, { "epoch": 0.34688683344714133, "grad_norm": 0.7802150737910987, "learning_rate": 9.992566945395426e-05, "loss": 3.6215, "step": 5588 }, { "epoch": 0.34694891054689925, "grad_norm": 0.8401528432846881, "learning_rate": 9.992547247040918e-05, "loss": 3.5583, "step": 5589 }, { "epoch": 0.34701098764665717, "grad_norm": 0.7627987370232027, "learning_rate": 9.992527522639078e-05, "loss": 3.5761, "step": 5590 }, { "epoch": 0.347073064746415, "grad_norm": 0.6215460228210374, "learning_rate": 9.992507772190011e-05, "loss": 3.5698, "step": 5591 }, { "epoch": 0.34713514184617295, "grad_norm": 0.6943871416185793, "learning_rate": 9.99248799569382e-05, "loss": 3.625, "step": 5592 }, { "epoch": 0.34719721894593086, "grad_norm": 0.6628464503678156, "learning_rate": 9.992468193150606e-05, "loss": 3.5972, "step": 5593 }, { "epoch": 0.3472592960456887, "grad_norm": 0.6421968900267844, "learning_rate": 9.992448364560474e-05, "loss": 3.585, "step": 5594 }, { "epoch": 0.34732137314544664, "grad_norm": 0.6231831368708806, "learning_rate": 9.992428509923526e-05, "loss": 3.5602, "step": 5595 }, { "epoch": 0.34738345024520456, "grad_norm": 0.6488874551837612, "learning_rate": 9.992408629239867e-05, "loss": 3.5614, "step": 5596 }, { "epoch": 0.3474455273449624, "grad_norm": 0.540417557536469, "learning_rate": 9.9923887225096e-05, "loss": 3.4943, "step": 5597 }, { "epoch": 0.34750760444472034, "grad_norm": 0.5463452361776754, "learning_rate": 9.992368789732829e-05, "loss": 3.5217, "step": 5598 }, { "epoch": 0.34756968154447826, "grad_norm": 0.5626495480422968, "learning_rate": 9.99234883090966e-05, "loss": 3.4792, "step": 5599 }, { "epoch": 0.3476317586442361, "grad_norm": 0.5643703307574361, "learning_rate": 9.992328846040193e-05, "loss": 3.4689, "step": 5600 }, { "epoch": 0.34769383574399404, "grad_norm": 0.5617267008551654, "learning_rate": 9.992308835124536e-05, "loss": 3.4898, "step": 5601 }, { "epoch": 0.34775591284375196, "grad_norm": 0.6122032458708143, "learning_rate": 9.99228879816279e-05, "loss": 3.4896, "step": 5602 }, { "epoch": 0.3478179899435098, "grad_norm": 0.5790395204911878, "learning_rate": 9.992268735155063e-05, "loss": 3.6138, "step": 5603 }, { "epoch": 0.34788006704326774, "grad_norm": 0.5813481207034644, "learning_rate": 9.992248646101456e-05, "loss": 3.5531, "step": 5604 }, { "epoch": 0.34794214414302566, "grad_norm": 0.5603304455123154, "learning_rate": 9.992228531002079e-05, "loss": 3.5678, "step": 5605 }, { "epoch": 0.3480042212427835, "grad_norm": 0.5877399948644494, "learning_rate": 9.992208389857031e-05, "loss": 3.5864, "step": 5606 }, { "epoch": 0.34806629834254144, "grad_norm": 0.5669510950328392, "learning_rate": 9.992188222666422e-05, "loss": 3.5246, "step": 5607 }, { "epoch": 0.34812837544229935, "grad_norm": 0.5158714514634116, "learning_rate": 9.992168029430355e-05, "loss": 3.595, "step": 5608 }, { "epoch": 0.3481904525420572, "grad_norm": 0.541191458102016, "learning_rate": 9.992147810148936e-05, "loss": 3.593, "step": 5609 }, { "epoch": 0.34825252964181513, "grad_norm": 0.4757342567475023, "learning_rate": 9.992127564822267e-05, "loss": 3.624, "step": 5610 }, { "epoch": 0.34831460674157305, "grad_norm": 0.576829308904243, "learning_rate": 9.99210729345046e-05, "loss": 3.6375, "step": 5611 }, { "epoch": 0.3483766838413309, "grad_norm": 0.6274362289274481, "learning_rate": 9.992086996033616e-05, "loss": 3.6139, "step": 5612 }, { "epoch": 0.34843876094108883, "grad_norm": 0.5383033659312976, "learning_rate": 9.992066672571841e-05, "loss": 3.5751, "step": 5613 }, { "epoch": 0.34850083804084675, "grad_norm": 0.4997725416417159, "learning_rate": 9.992046323065242e-05, "loss": 3.6046, "step": 5614 }, { "epoch": 0.3485629151406046, "grad_norm": 0.5424773098555783, "learning_rate": 9.992025947513928e-05, "loss": 3.4646, "step": 5615 }, { "epoch": 0.34862499224036253, "grad_norm": 0.5233795646375496, "learning_rate": 9.992005545918002e-05, "loss": 3.5813, "step": 5616 }, { "epoch": 0.34868706934012045, "grad_norm": 0.5543801167885237, "learning_rate": 9.991985118277571e-05, "loss": 3.5769, "step": 5617 }, { "epoch": 0.3487491464398783, "grad_norm": 0.4974028219028319, "learning_rate": 9.991964664592741e-05, "loss": 3.5444, "step": 5618 }, { "epoch": 0.34881122353963623, "grad_norm": 0.566424191947105, "learning_rate": 9.991944184863621e-05, "loss": 3.6539, "step": 5619 }, { "epoch": 0.34887330063939415, "grad_norm": 0.5457364355473188, "learning_rate": 9.991923679090316e-05, "loss": 3.6051, "step": 5620 }, { "epoch": 0.348935377739152, "grad_norm": 0.5477648443720348, "learning_rate": 9.991903147272932e-05, "loss": 3.4758, "step": 5621 }, { "epoch": 0.3489974548389099, "grad_norm": 0.6001198301261469, "learning_rate": 9.991882589411578e-05, "loss": 3.6684, "step": 5622 }, { "epoch": 0.34905953193866784, "grad_norm": 0.703816772938481, "learning_rate": 9.991862005506361e-05, "loss": 3.5653, "step": 5623 }, { "epoch": 0.3491216090384257, "grad_norm": 0.7462594612199135, "learning_rate": 9.991841395557388e-05, "loss": 3.5857, "step": 5624 }, { "epoch": 0.3491836861381836, "grad_norm": 0.5843076034685114, "learning_rate": 9.991820759564768e-05, "loss": 3.5583, "step": 5625 }, { "epoch": 0.34924576323794154, "grad_norm": 0.5274797617388148, "learning_rate": 9.991800097528606e-05, "loss": 3.608, "step": 5626 }, { "epoch": 0.3493078403376994, "grad_norm": 0.6635516277226293, "learning_rate": 9.991779409449011e-05, "loss": 3.5597, "step": 5627 }, { "epoch": 0.3493699174374573, "grad_norm": 0.8525016225872092, "learning_rate": 9.991758695326091e-05, "loss": 3.6035, "step": 5628 }, { "epoch": 0.34943199453721524, "grad_norm": 0.8548300113456003, "learning_rate": 9.991737955159954e-05, "loss": 3.5635, "step": 5629 }, { "epoch": 0.3494940716369731, "grad_norm": 0.6898536707336482, "learning_rate": 9.991717188950709e-05, "loss": 3.5192, "step": 5630 }, { "epoch": 0.349556148736731, "grad_norm": 0.5640127714855163, "learning_rate": 9.991696396698463e-05, "loss": 3.5584, "step": 5631 }, { "epoch": 0.34961822583648894, "grad_norm": 0.6383065628392685, "learning_rate": 9.991675578403325e-05, "loss": 3.637, "step": 5632 }, { "epoch": 0.3496803029362468, "grad_norm": 0.6427680897149749, "learning_rate": 9.991654734065405e-05, "loss": 3.4197, "step": 5633 }, { "epoch": 0.3497423800360047, "grad_norm": 0.5468866891974098, "learning_rate": 9.991633863684809e-05, "loss": 3.6855, "step": 5634 }, { "epoch": 0.34980445713576264, "grad_norm": 0.6102366346874895, "learning_rate": 9.991612967261647e-05, "loss": 3.5683, "step": 5635 }, { "epoch": 0.3498665342355205, "grad_norm": 0.5486286927744487, "learning_rate": 9.99159204479603e-05, "loss": 3.5774, "step": 5636 }, { "epoch": 0.3499286113352784, "grad_norm": 0.5642710197278943, "learning_rate": 9.991571096288066e-05, "loss": 3.6062, "step": 5637 }, { "epoch": 0.34999068843503633, "grad_norm": 0.6523917860059582, "learning_rate": 9.991550121737863e-05, "loss": 3.6295, "step": 5638 }, { "epoch": 0.3500527655347942, "grad_norm": 0.582955771828064, "learning_rate": 9.991529121145531e-05, "loss": 3.5709, "step": 5639 }, { "epoch": 0.3501148426345521, "grad_norm": 0.5925426505226183, "learning_rate": 9.99150809451118e-05, "loss": 3.5775, "step": 5640 }, { "epoch": 0.35017691973431003, "grad_norm": 0.5530613717158075, "learning_rate": 9.991487041834919e-05, "loss": 3.6519, "step": 5641 }, { "epoch": 0.3502389968340679, "grad_norm": 0.5760471790962824, "learning_rate": 9.991465963116858e-05, "loss": 3.5811, "step": 5642 }, { "epoch": 0.3503010739338258, "grad_norm": 0.5272181043351288, "learning_rate": 9.991444858357108e-05, "loss": 3.5404, "step": 5643 }, { "epoch": 0.35036315103358373, "grad_norm": 0.5398492738300374, "learning_rate": 9.99142372755578e-05, "loss": 3.5616, "step": 5644 }, { "epoch": 0.3504252281333416, "grad_norm": 0.5213949576879215, "learning_rate": 9.99140257071298e-05, "loss": 3.5028, "step": 5645 }, { "epoch": 0.3504873052330995, "grad_norm": 0.5429992118359039, "learning_rate": 9.991381387828823e-05, "loss": 3.5748, "step": 5646 }, { "epoch": 0.35054938233285743, "grad_norm": 0.5200956333692213, "learning_rate": 9.991360178903418e-05, "loss": 3.5798, "step": 5647 }, { "epoch": 0.3506114594326153, "grad_norm": 0.5444263680153908, "learning_rate": 9.991338943936874e-05, "loss": 3.6074, "step": 5648 }, { "epoch": 0.3506735365323732, "grad_norm": 0.5765772541091564, "learning_rate": 9.991317682929302e-05, "loss": 3.5766, "step": 5649 }, { "epoch": 0.3507356136321311, "grad_norm": 0.5298607838175395, "learning_rate": 9.991296395880816e-05, "loss": 3.5668, "step": 5650 }, { "epoch": 0.350797690731889, "grad_norm": 0.4857373306773989, "learning_rate": 9.991275082791524e-05, "loss": 3.5421, "step": 5651 }, { "epoch": 0.3508597678316469, "grad_norm": 0.5693200241665421, "learning_rate": 9.991253743661538e-05, "loss": 3.5346, "step": 5652 }, { "epoch": 0.3509218449314048, "grad_norm": 0.6947859581327246, "learning_rate": 9.99123237849097e-05, "loss": 3.5877, "step": 5653 }, { "epoch": 0.3509839220311627, "grad_norm": 0.6949443914272627, "learning_rate": 9.991210987279931e-05, "loss": 3.555, "step": 5654 }, { "epoch": 0.3510459991309206, "grad_norm": 0.5594944394852063, "learning_rate": 9.991189570028533e-05, "loss": 3.4908, "step": 5655 }, { "epoch": 0.3511080762306785, "grad_norm": 0.5478383434965199, "learning_rate": 9.991168126736888e-05, "loss": 3.6047, "step": 5656 }, { "epoch": 0.3511701533304364, "grad_norm": 0.6016392596924917, "learning_rate": 9.991146657405105e-05, "loss": 3.6173, "step": 5657 }, { "epoch": 0.3512322304301943, "grad_norm": 0.6015213791305231, "learning_rate": 9.9911251620333e-05, "loss": 3.5135, "step": 5658 }, { "epoch": 0.3512943075299522, "grad_norm": 0.5800555960136766, "learning_rate": 9.991103640621582e-05, "loss": 3.5696, "step": 5659 }, { "epoch": 0.3513563846297101, "grad_norm": 0.5895950854268082, "learning_rate": 9.991082093170065e-05, "loss": 3.5571, "step": 5660 }, { "epoch": 0.351418461729468, "grad_norm": 0.5781599755919464, "learning_rate": 9.991060519678864e-05, "loss": 3.6222, "step": 5661 }, { "epoch": 0.3514805388292259, "grad_norm": 0.5781422947539145, "learning_rate": 9.991038920148086e-05, "loss": 3.549, "step": 5662 }, { "epoch": 0.3515426159289838, "grad_norm": 0.5611802354285705, "learning_rate": 9.991017294577846e-05, "loss": 3.5348, "step": 5663 }, { "epoch": 0.3516046930287417, "grad_norm": 0.6329590663013789, "learning_rate": 9.990995642968257e-05, "loss": 3.5833, "step": 5664 }, { "epoch": 0.3516667701284996, "grad_norm": 0.6259266183243424, "learning_rate": 9.990973965319433e-05, "loss": 3.6232, "step": 5665 }, { "epoch": 0.3517288472282575, "grad_norm": 0.6352641099701151, "learning_rate": 9.990952261631487e-05, "loss": 3.5205, "step": 5666 }, { "epoch": 0.3517909243280154, "grad_norm": 0.639123319609097, "learning_rate": 9.99093053190453e-05, "loss": 3.4609, "step": 5667 }, { "epoch": 0.3518530014277733, "grad_norm": 0.6190524113625919, "learning_rate": 9.990908776138678e-05, "loss": 3.573, "step": 5668 }, { "epoch": 0.3519150785275312, "grad_norm": 0.6417285207149008, "learning_rate": 9.990886994334042e-05, "loss": 3.5522, "step": 5669 }, { "epoch": 0.3519771556272891, "grad_norm": 0.6253488044170401, "learning_rate": 9.990865186490738e-05, "loss": 3.5336, "step": 5670 }, { "epoch": 0.352039232727047, "grad_norm": 0.576667236491383, "learning_rate": 9.990843352608879e-05, "loss": 3.4655, "step": 5671 }, { "epoch": 0.3521013098268049, "grad_norm": 0.5956366079746415, "learning_rate": 9.990821492688578e-05, "loss": 3.6262, "step": 5672 }, { "epoch": 0.3521633869265628, "grad_norm": 0.5852940141291277, "learning_rate": 9.99079960672995e-05, "loss": 3.5673, "step": 5673 }, { "epoch": 0.3522254640263207, "grad_norm": 0.5646916713676638, "learning_rate": 9.990777694733109e-05, "loss": 3.548, "step": 5674 }, { "epoch": 0.3522875411260786, "grad_norm": 0.5989118862110818, "learning_rate": 9.990755756698168e-05, "loss": 3.5759, "step": 5675 }, { "epoch": 0.3523496182258365, "grad_norm": 0.6677970888736425, "learning_rate": 9.990733792625244e-05, "loss": 3.5675, "step": 5676 }, { "epoch": 0.3524116953255944, "grad_norm": 0.6693660109314418, "learning_rate": 9.99071180251445e-05, "loss": 3.569, "step": 5677 }, { "epoch": 0.35247377242535227, "grad_norm": 0.5013108355926479, "learning_rate": 9.990689786365901e-05, "loss": 3.5124, "step": 5678 }, { "epoch": 0.3525358495251102, "grad_norm": 0.4942889236044588, "learning_rate": 9.990667744179712e-05, "loss": 3.5949, "step": 5679 }, { "epoch": 0.3525979266248681, "grad_norm": 0.6135720699357069, "learning_rate": 9.990645675955996e-05, "loss": 3.5028, "step": 5680 }, { "epoch": 0.35266000372462597, "grad_norm": 0.5491962487941445, "learning_rate": 9.990623581694872e-05, "loss": 3.5882, "step": 5681 }, { "epoch": 0.3527220808243839, "grad_norm": 0.5923987120152651, "learning_rate": 9.990601461396454e-05, "loss": 3.5708, "step": 5682 }, { "epoch": 0.3527841579241418, "grad_norm": 0.5214903470386769, "learning_rate": 9.990579315060855e-05, "loss": 3.5782, "step": 5683 }, { "epoch": 0.35284623502389967, "grad_norm": 0.4815196972192119, "learning_rate": 9.990557142688193e-05, "loss": 3.4494, "step": 5684 }, { "epoch": 0.3529083121236576, "grad_norm": 0.6039001326356007, "learning_rate": 9.990534944278582e-05, "loss": 3.522, "step": 5685 }, { "epoch": 0.3529703892234155, "grad_norm": 0.5950500911175679, "learning_rate": 9.990512719832139e-05, "loss": 3.4929, "step": 5686 }, { "epoch": 0.35303246632317337, "grad_norm": 0.6172621235218879, "learning_rate": 9.990490469348981e-05, "loss": 3.5661, "step": 5687 }, { "epoch": 0.3530945434229313, "grad_norm": 0.5547540651596226, "learning_rate": 9.99046819282922e-05, "loss": 3.5215, "step": 5688 }, { "epoch": 0.3531566205226892, "grad_norm": 0.5556449417552017, "learning_rate": 9.990445890272977e-05, "loss": 3.5256, "step": 5689 }, { "epoch": 0.35321869762244706, "grad_norm": 0.5324162136568936, "learning_rate": 9.990423561680367e-05, "loss": 3.4677, "step": 5690 }, { "epoch": 0.353280774722205, "grad_norm": 0.5193017151378749, "learning_rate": 9.990401207051504e-05, "loss": 3.5127, "step": 5691 }, { "epoch": 0.3533428518219629, "grad_norm": 0.5881013207528291, "learning_rate": 9.990378826386508e-05, "loss": 3.5375, "step": 5692 }, { "epoch": 0.35340492892172076, "grad_norm": 0.6182239617951065, "learning_rate": 9.990356419685492e-05, "loss": 3.6019, "step": 5693 }, { "epoch": 0.3534670060214787, "grad_norm": 0.588501039863841, "learning_rate": 9.990333986948577e-05, "loss": 3.5514, "step": 5694 }, { "epoch": 0.3535290831212366, "grad_norm": 0.5340306773339196, "learning_rate": 9.990311528175877e-05, "loss": 3.6211, "step": 5695 }, { "epoch": 0.35359116022099446, "grad_norm": 0.5410160381439505, "learning_rate": 9.990289043367511e-05, "loss": 3.5543, "step": 5696 }, { "epoch": 0.3536532373207524, "grad_norm": 0.536115524485754, "learning_rate": 9.990266532523596e-05, "loss": 3.4975, "step": 5697 }, { "epoch": 0.3537153144205103, "grad_norm": 0.6480346255441011, "learning_rate": 9.990243995644249e-05, "loss": 3.4916, "step": 5698 }, { "epoch": 0.35377739152026816, "grad_norm": 0.6505623438968081, "learning_rate": 9.990221432729587e-05, "loss": 3.6331, "step": 5699 }, { "epoch": 0.3538394686200261, "grad_norm": 0.5370558001068045, "learning_rate": 9.990198843779729e-05, "loss": 3.564, "step": 5700 }, { "epoch": 0.353901545719784, "grad_norm": 0.5043442766647855, "learning_rate": 9.990176228794792e-05, "loss": 3.549, "step": 5701 }, { "epoch": 0.35396362281954186, "grad_norm": 0.4978089306470516, "learning_rate": 9.990153587774895e-05, "loss": 3.6433, "step": 5702 }, { "epoch": 0.3540256999192998, "grad_norm": 0.5979530245883956, "learning_rate": 9.990130920720153e-05, "loss": 3.5405, "step": 5703 }, { "epoch": 0.3540877770190577, "grad_norm": 0.5865417162587226, "learning_rate": 9.990108227630688e-05, "loss": 3.5718, "step": 5704 }, { "epoch": 0.35414985411881555, "grad_norm": 0.6438211864629416, "learning_rate": 9.990085508506616e-05, "loss": 3.539, "step": 5705 }, { "epoch": 0.3542119312185735, "grad_norm": 0.612360293161737, "learning_rate": 9.990062763348059e-05, "loss": 3.5947, "step": 5706 }, { "epoch": 0.3542740083183314, "grad_norm": 0.5649705514997804, "learning_rate": 9.990039992155131e-05, "loss": 3.474, "step": 5707 }, { "epoch": 0.35433608541808925, "grad_norm": 0.5532832032594458, "learning_rate": 9.990017194927954e-05, "loss": 3.5123, "step": 5708 }, { "epoch": 0.35439816251784717, "grad_norm": 0.5465535990544791, "learning_rate": 9.989994371666647e-05, "loss": 3.6121, "step": 5709 }, { "epoch": 0.3544602396176051, "grad_norm": 0.607026219684668, "learning_rate": 9.989971522371326e-05, "loss": 3.517, "step": 5710 }, { "epoch": 0.35452231671736295, "grad_norm": 0.6128229019939592, "learning_rate": 9.989948647042113e-05, "loss": 3.5469, "step": 5711 }, { "epoch": 0.35458439381712087, "grad_norm": 0.5867989628764442, "learning_rate": 9.989925745679126e-05, "loss": 3.5252, "step": 5712 }, { "epoch": 0.3546464709168788, "grad_norm": 0.5646961309452094, "learning_rate": 9.989902818282486e-05, "loss": 3.5819, "step": 5713 }, { "epoch": 0.35470854801663665, "grad_norm": 0.5783519814789181, "learning_rate": 9.989879864852311e-05, "loss": 3.6073, "step": 5714 }, { "epoch": 0.35477062511639457, "grad_norm": 0.5853490799503422, "learning_rate": 9.989856885388721e-05, "loss": 3.5116, "step": 5715 }, { "epoch": 0.3548327022161525, "grad_norm": 0.5769955453535441, "learning_rate": 9.989833879891836e-05, "loss": 3.6383, "step": 5716 }, { "epoch": 0.35489477931591035, "grad_norm": 0.6204515651992704, "learning_rate": 9.989810848361778e-05, "loss": 3.5748, "step": 5717 }, { "epoch": 0.35495685641566826, "grad_norm": 0.6616907062015052, "learning_rate": 9.989787790798665e-05, "loss": 3.6207, "step": 5718 }, { "epoch": 0.3550189335154262, "grad_norm": 0.5966978179624517, "learning_rate": 9.989764707202616e-05, "loss": 3.6386, "step": 5719 }, { "epoch": 0.35508101061518405, "grad_norm": 0.5811504637697888, "learning_rate": 9.989741597573756e-05, "loss": 3.6638, "step": 5720 }, { "epoch": 0.35514308771494196, "grad_norm": 0.6142361732637661, "learning_rate": 9.989718461912201e-05, "loss": 3.547, "step": 5721 }, { "epoch": 0.3552051648146999, "grad_norm": 0.5845892335000787, "learning_rate": 9.989695300218073e-05, "loss": 3.4932, "step": 5722 }, { "epoch": 0.35526724191445774, "grad_norm": 0.6549677356024308, "learning_rate": 9.989672112491494e-05, "loss": 3.512, "step": 5723 }, { "epoch": 0.35532931901421566, "grad_norm": 0.681807642371186, "learning_rate": 9.989648898732586e-05, "loss": 3.5539, "step": 5724 }, { "epoch": 0.3553913961139736, "grad_norm": 0.5787616265943754, "learning_rate": 9.989625658941467e-05, "loss": 3.6055, "step": 5725 }, { "epoch": 0.35545347321373144, "grad_norm": 0.545175062857013, "learning_rate": 9.989602393118259e-05, "loss": 3.5666, "step": 5726 }, { "epoch": 0.35551555031348936, "grad_norm": 0.514994032927082, "learning_rate": 9.989579101263085e-05, "loss": 3.542, "step": 5727 }, { "epoch": 0.3555776274132473, "grad_norm": 0.543567492011768, "learning_rate": 9.989555783376066e-05, "loss": 3.5042, "step": 5728 }, { "epoch": 0.35563970451300514, "grad_norm": 0.5861631594679427, "learning_rate": 9.98953243945732e-05, "loss": 3.5572, "step": 5729 }, { "epoch": 0.35570178161276306, "grad_norm": 0.4898862020297574, "learning_rate": 9.989509069506976e-05, "loss": 3.5019, "step": 5730 }, { "epoch": 0.355763858712521, "grad_norm": 0.494752979439221, "learning_rate": 9.98948567352515e-05, "loss": 3.4853, "step": 5731 }, { "epoch": 0.35582593581227884, "grad_norm": 0.6386641289010859, "learning_rate": 9.989462251511967e-05, "loss": 3.584, "step": 5732 }, { "epoch": 0.35588801291203676, "grad_norm": 0.5959714691610544, "learning_rate": 9.989438803467548e-05, "loss": 3.5565, "step": 5733 }, { "epoch": 0.3559500900117947, "grad_norm": 0.5290703095041662, "learning_rate": 9.989415329392014e-05, "loss": 3.5937, "step": 5734 }, { "epoch": 0.35601216711155254, "grad_norm": 0.6120731822635618, "learning_rate": 9.989391829285491e-05, "loss": 3.5852, "step": 5735 }, { "epoch": 0.35607424421131045, "grad_norm": 0.5856037206457974, "learning_rate": 9.989368303148098e-05, "loss": 3.5561, "step": 5736 }, { "epoch": 0.35613632131106837, "grad_norm": 0.5858458892413683, "learning_rate": 9.98934475097996e-05, "loss": 3.5833, "step": 5737 }, { "epoch": 0.35619839841082623, "grad_norm": 0.6336900489711467, "learning_rate": 9.989321172781199e-05, "loss": 3.4944, "step": 5738 }, { "epoch": 0.35626047551058415, "grad_norm": 0.636912170238895, "learning_rate": 9.989297568551938e-05, "loss": 3.4835, "step": 5739 }, { "epoch": 0.35632255261034207, "grad_norm": 0.6528055075804634, "learning_rate": 9.989273938292302e-05, "loss": 3.5197, "step": 5740 }, { "epoch": 0.35638462971009993, "grad_norm": 0.6606329601822387, "learning_rate": 9.989250282002411e-05, "loss": 3.5283, "step": 5741 }, { "epoch": 0.35644670680985785, "grad_norm": 0.6412466800893764, "learning_rate": 9.98922659968239e-05, "loss": 3.6511, "step": 5742 }, { "epoch": 0.35650878390961577, "grad_norm": 0.649446782026425, "learning_rate": 9.989202891332364e-05, "loss": 3.5301, "step": 5743 }, { "epoch": 0.35657086100937363, "grad_norm": 0.5847018373363628, "learning_rate": 9.989179156952455e-05, "loss": 3.5427, "step": 5744 }, { "epoch": 0.35663293810913155, "grad_norm": 0.5951097127997752, "learning_rate": 9.989155396542786e-05, "loss": 3.4885, "step": 5745 }, { "epoch": 0.35669501520888947, "grad_norm": 0.5760189305992791, "learning_rate": 9.989131610103484e-05, "loss": 3.598, "step": 5746 }, { "epoch": 0.35675709230864733, "grad_norm": 0.5793935673196279, "learning_rate": 9.98910779763467e-05, "loss": 3.5008, "step": 5747 }, { "epoch": 0.35681916940840525, "grad_norm": 0.5959737944490241, "learning_rate": 9.989083959136469e-05, "loss": 3.5741, "step": 5748 }, { "epoch": 0.35688124650816316, "grad_norm": 0.5940931483052944, "learning_rate": 9.989060094609007e-05, "loss": 3.5276, "step": 5749 }, { "epoch": 0.356943323607921, "grad_norm": 0.5170238477973058, "learning_rate": 9.989036204052407e-05, "loss": 3.3505, "step": 5750 }, { "epoch": 0.35700540070767894, "grad_norm": 0.5223283260714457, "learning_rate": 9.989012287466794e-05, "loss": 3.5526, "step": 5751 }, { "epoch": 0.35706747780743686, "grad_norm": 0.6200814304405489, "learning_rate": 9.988988344852292e-05, "loss": 3.5929, "step": 5752 }, { "epoch": 0.3571295549071947, "grad_norm": 0.699120647858159, "learning_rate": 9.988964376209029e-05, "loss": 3.4584, "step": 5753 }, { "epoch": 0.35719163200695264, "grad_norm": 0.6435990267632442, "learning_rate": 9.988940381537126e-05, "loss": 3.5934, "step": 5754 }, { "epoch": 0.35725370910671056, "grad_norm": 0.6020089511617748, "learning_rate": 9.98891636083671e-05, "loss": 3.4963, "step": 5755 }, { "epoch": 0.3573157862064684, "grad_norm": 0.5939934131196095, "learning_rate": 9.988892314107906e-05, "loss": 3.561, "step": 5756 }, { "epoch": 0.35737786330622634, "grad_norm": 0.5840530355144864, "learning_rate": 9.98886824135084e-05, "loss": 3.5387, "step": 5757 }, { "epoch": 0.35743994040598426, "grad_norm": 0.5432377430234864, "learning_rate": 9.988844142565637e-05, "loss": 3.4967, "step": 5758 }, { "epoch": 0.3575020175057421, "grad_norm": 0.4980267240150246, "learning_rate": 9.988820017752423e-05, "loss": 3.5101, "step": 5759 }, { "epoch": 0.35756409460550004, "grad_norm": 0.588229769091745, "learning_rate": 9.988795866911325e-05, "loss": 3.5711, "step": 5760 }, { "epoch": 0.35762617170525796, "grad_norm": 0.6134969004138473, "learning_rate": 9.988771690042467e-05, "loss": 3.4933, "step": 5761 }, { "epoch": 0.3576882488050158, "grad_norm": 0.5497614841866304, "learning_rate": 9.988747487145976e-05, "loss": 3.521, "step": 5762 }, { "epoch": 0.35775032590477374, "grad_norm": 0.5447773895678321, "learning_rate": 9.988723258221977e-05, "loss": 3.5562, "step": 5763 }, { "epoch": 0.35781240300453165, "grad_norm": 0.6092392696927442, "learning_rate": 9.9886990032706e-05, "loss": 3.5457, "step": 5764 }, { "epoch": 0.3578744801042895, "grad_norm": 0.6656368442609896, "learning_rate": 9.988674722291969e-05, "loss": 3.5943, "step": 5765 }, { "epoch": 0.35793655720404743, "grad_norm": 0.6196026687538122, "learning_rate": 9.988650415286209e-05, "loss": 3.4465, "step": 5766 }, { "epoch": 0.35799863430380535, "grad_norm": 0.5392819662739805, "learning_rate": 9.988626082253451e-05, "loss": 3.4619, "step": 5767 }, { "epoch": 0.3580607114035632, "grad_norm": 0.5059474146756829, "learning_rate": 9.988601723193818e-05, "loss": 3.6362, "step": 5768 }, { "epoch": 0.35812278850332113, "grad_norm": 0.5146270400056209, "learning_rate": 9.98857733810744e-05, "loss": 3.5625, "step": 5769 }, { "epoch": 0.35818486560307905, "grad_norm": 0.5627596308961461, "learning_rate": 9.988552926994444e-05, "loss": 3.5219, "step": 5770 }, { "epoch": 0.3582469427028369, "grad_norm": 0.6738907052433449, "learning_rate": 9.988528489854953e-05, "loss": 3.593, "step": 5771 }, { "epoch": 0.35830901980259483, "grad_norm": 0.8644328656690523, "learning_rate": 9.9885040266891e-05, "loss": 3.6146, "step": 5772 }, { "epoch": 0.35837109690235275, "grad_norm": 1.1291838835826564, "learning_rate": 9.98847953749701e-05, "loss": 3.5321, "step": 5773 }, { "epoch": 0.3584331740021106, "grad_norm": 0.8581023068910821, "learning_rate": 9.988455022278812e-05, "loss": 3.4988, "step": 5774 }, { "epoch": 0.35849525110186853, "grad_norm": 0.8212732773941032, "learning_rate": 9.988430481034632e-05, "loss": 3.654, "step": 5775 }, { "epoch": 0.35855732820162645, "grad_norm": 0.80518388715172, "learning_rate": 9.9884059137646e-05, "loss": 3.5276, "step": 5776 }, { "epoch": 0.3586194053013843, "grad_norm": 0.8963263544839699, "learning_rate": 9.988381320468844e-05, "loss": 3.5749, "step": 5777 }, { "epoch": 0.3586814824011422, "grad_norm": 0.9249425058281606, "learning_rate": 9.98835670114749e-05, "loss": 3.6446, "step": 5778 }, { "epoch": 0.35874355950090014, "grad_norm": 0.6665710862505843, "learning_rate": 9.988332055800669e-05, "loss": 3.5201, "step": 5779 }, { "epoch": 0.358805636600658, "grad_norm": 0.580167590091339, "learning_rate": 9.988307384428509e-05, "loss": 3.4644, "step": 5780 }, { "epoch": 0.3588677137004159, "grad_norm": 0.5950773360377487, "learning_rate": 9.988282687031139e-05, "loss": 3.4129, "step": 5781 }, { "epoch": 0.35892979080017384, "grad_norm": 0.6106530993428012, "learning_rate": 9.988257963608685e-05, "loss": 3.5658, "step": 5782 }, { "epoch": 0.3589918678999317, "grad_norm": 0.5954682934773561, "learning_rate": 9.98823321416128e-05, "loss": 3.5751, "step": 5783 }, { "epoch": 0.3590539449996896, "grad_norm": 0.6173204689202857, "learning_rate": 9.988208438689052e-05, "loss": 3.5583, "step": 5784 }, { "epoch": 0.35911602209944754, "grad_norm": 0.5820824517380144, "learning_rate": 9.988183637192128e-05, "loss": 3.4452, "step": 5785 }, { "epoch": 0.3591780991992054, "grad_norm": 0.5778484972009462, "learning_rate": 9.98815880967064e-05, "loss": 3.5388, "step": 5786 }, { "epoch": 0.3592401762989633, "grad_norm": 0.6201074794700234, "learning_rate": 9.988133956124716e-05, "loss": 3.5261, "step": 5787 }, { "epoch": 0.35930225339872124, "grad_norm": 0.518016873939101, "learning_rate": 9.988109076554486e-05, "loss": 3.4957, "step": 5788 }, { "epoch": 0.3593643304984791, "grad_norm": 0.6061402565642268, "learning_rate": 9.988084170960081e-05, "loss": 3.5845, "step": 5789 }, { "epoch": 0.359426407598237, "grad_norm": 0.6916359126672578, "learning_rate": 9.988059239341629e-05, "loss": 3.4959, "step": 5790 }, { "epoch": 0.35948848469799494, "grad_norm": 0.5949412492388882, "learning_rate": 9.988034281699263e-05, "loss": 3.508, "step": 5791 }, { "epoch": 0.3595505617977528, "grad_norm": 0.6170205481985681, "learning_rate": 9.98800929803311e-05, "loss": 3.5975, "step": 5792 }, { "epoch": 0.3596126388975107, "grad_norm": 0.6538223139809699, "learning_rate": 9.987984288343302e-05, "loss": 3.5846, "step": 5793 }, { "epoch": 0.35967471599726863, "grad_norm": 0.5836613760172282, "learning_rate": 9.987959252629968e-05, "loss": 3.5068, "step": 5794 }, { "epoch": 0.3597367930970265, "grad_norm": 0.5308674107856347, "learning_rate": 9.98793419089324e-05, "loss": 3.6223, "step": 5795 }, { "epoch": 0.3597988701967844, "grad_norm": 0.572913834535939, "learning_rate": 9.987909103133248e-05, "loss": 3.593, "step": 5796 }, { "epoch": 0.35986094729654233, "grad_norm": 0.5916483475531078, "learning_rate": 9.987883989350125e-05, "loss": 3.5726, "step": 5797 }, { "epoch": 0.3599230243963002, "grad_norm": 0.5535757448098458, "learning_rate": 9.987858849544e-05, "loss": 3.5818, "step": 5798 }, { "epoch": 0.3599851014960581, "grad_norm": 0.5514390953271767, "learning_rate": 9.987833683715003e-05, "loss": 3.4255, "step": 5799 }, { "epoch": 0.36004717859581603, "grad_norm": 0.5083419680513453, "learning_rate": 9.987808491863268e-05, "loss": 3.5586, "step": 5800 }, { "epoch": 0.3601092556955739, "grad_norm": 0.5806158893309789, "learning_rate": 9.987783273988926e-05, "loss": 3.4177, "step": 5801 }, { "epoch": 0.3601713327953318, "grad_norm": 0.6709597808023788, "learning_rate": 9.987758030092106e-05, "loss": 3.5776, "step": 5802 }, { "epoch": 0.36023340989508973, "grad_norm": 0.6919196857195249, "learning_rate": 9.987732760172943e-05, "loss": 3.5383, "step": 5803 }, { "epoch": 0.3602954869948476, "grad_norm": 0.6748346744394327, "learning_rate": 9.987707464231568e-05, "loss": 3.5593, "step": 5804 }, { "epoch": 0.3603575640946055, "grad_norm": 0.7357694089677553, "learning_rate": 9.98768214226811e-05, "loss": 3.497, "step": 5805 }, { "epoch": 0.3604196411943634, "grad_norm": 0.6910080921717807, "learning_rate": 9.987656794282705e-05, "loss": 3.5652, "step": 5806 }, { "epoch": 0.3604817182941213, "grad_norm": 0.6864210896284862, "learning_rate": 9.987631420275483e-05, "loss": 3.5168, "step": 5807 }, { "epoch": 0.3605437953938792, "grad_norm": 0.7774897682645168, "learning_rate": 9.987606020246578e-05, "loss": 3.4661, "step": 5808 }, { "epoch": 0.36060587249363707, "grad_norm": 0.8763524910578928, "learning_rate": 9.987580594196121e-05, "loss": 3.531, "step": 5809 }, { "epoch": 0.360667949593395, "grad_norm": 0.7407896027962069, "learning_rate": 9.987555142124245e-05, "loss": 3.5269, "step": 5810 }, { "epoch": 0.3607300266931529, "grad_norm": 0.6757761784181308, "learning_rate": 9.987529664031083e-05, "loss": 3.6086, "step": 5811 }, { "epoch": 0.36079210379291077, "grad_norm": 0.7648646413192371, "learning_rate": 9.98750415991677e-05, "loss": 3.4828, "step": 5812 }, { "epoch": 0.3608541808926687, "grad_norm": 0.7207584928421724, "learning_rate": 9.987478629781435e-05, "loss": 3.5915, "step": 5813 }, { "epoch": 0.3609162579924266, "grad_norm": 0.6759287061397018, "learning_rate": 9.987453073625212e-05, "loss": 3.5089, "step": 5814 }, { "epoch": 0.36097833509218447, "grad_norm": 0.6703213537738527, "learning_rate": 9.987427491448237e-05, "loss": 3.5909, "step": 5815 }, { "epoch": 0.3610404121919424, "grad_norm": 0.604032819544681, "learning_rate": 9.987401883250645e-05, "loss": 3.4529, "step": 5816 }, { "epoch": 0.3611024892917003, "grad_norm": 0.6543740525917402, "learning_rate": 9.987376249032562e-05, "loss": 3.543, "step": 5817 }, { "epoch": 0.36116456639145816, "grad_norm": 0.6218577071042694, "learning_rate": 9.987350588794129e-05, "loss": 3.4829, "step": 5818 }, { "epoch": 0.3612266434912161, "grad_norm": 0.5163549689196316, "learning_rate": 9.987324902535476e-05, "loss": 3.4373, "step": 5819 }, { "epoch": 0.361288720590974, "grad_norm": 0.5518673138283036, "learning_rate": 9.987299190256739e-05, "loss": 3.6155, "step": 5820 }, { "epoch": 0.36135079769073186, "grad_norm": 0.6585194990447139, "learning_rate": 9.987273451958052e-05, "loss": 3.5454, "step": 5821 }, { "epoch": 0.3614128747904898, "grad_norm": 0.7050714466063996, "learning_rate": 9.987247687639548e-05, "loss": 3.5867, "step": 5822 }, { "epoch": 0.3614749518902477, "grad_norm": 0.6801341113206031, "learning_rate": 9.987221897301363e-05, "loss": 3.4959, "step": 5823 }, { "epoch": 0.36153702899000556, "grad_norm": 0.6925070238273631, "learning_rate": 9.987196080943628e-05, "loss": 3.5735, "step": 5824 }, { "epoch": 0.3615991060897635, "grad_norm": 0.6076861723410905, "learning_rate": 9.987170238566483e-05, "loss": 3.5381, "step": 5825 }, { "epoch": 0.3616611831895214, "grad_norm": 0.6308815541812208, "learning_rate": 9.987144370170059e-05, "loss": 3.4601, "step": 5826 }, { "epoch": 0.36172326028927926, "grad_norm": 0.6561759725325335, "learning_rate": 9.987118475754492e-05, "loss": 3.4829, "step": 5827 }, { "epoch": 0.3617853373890372, "grad_norm": 0.5817639293847268, "learning_rate": 9.987092555319918e-05, "loss": 3.4784, "step": 5828 }, { "epoch": 0.3618474144887951, "grad_norm": 0.6051420499404709, "learning_rate": 9.987066608866471e-05, "loss": 3.538, "step": 5829 }, { "epoch": 0.36190949158855296, "grad_norm": 0.5800981614476795, "learning_rate": 9.987040636394287e-05, "loss": 3.6026, "step": 5830 }, { "epoch": 0.3619715686883109, "grad_norm": 0.5525984849796519, "learning_rate": 9.987014637903501e-05, "loss": 3.4823, "step": 5831 }, { "epoch": 0.3620336457880688, "grad_norm": 0.5927473185030917, "learning_rate": 9.98698861339425e-05, "loss": 3.5467, "step": 5832 }, { "epoch": 0.36209572288782665, "grad_norm": 0.5611821472595844, "learning_rate": 9.986962562866669e-05, "loss": 3.5166, "step": 5833 }, { "epoch": 0.36215779998758457, "grad_norm": 0.6885847507180453, "learning_rate": 9.986936486320892e-05, "loss": 3.5499, "step": 5834 }, { "epoch": 0.3622198770873425, "grad_norm": 0.7280823592769288, "learning_rate": 9.986910383757058e-05, "loss": 3.4847, "step": 5835 }, { "epoch": 0.36228195418710035, "grad_norm": 0.6225451898066842, "learning_rate": 9.9868842551753e-05, "loss": 3.5394, "step": 5836 }, { "epoch": 0.36234403128685827, "grad_norm": 0.5699359748251841, "learning_rate": 9.986858100575758e-05, "loss": 3.586, "step": 5837 }, { "epoch": 0.3624061083866162, "grad_norm": 0.5550752278627498, "learning_rate": 9.986831919958565e-05, "loss": 3.5256, "step": 5838 }, { "epoch": 0.36246818548637405, "grad_norm": 0.6764604980540633, "learning_rate": 9.986805713323862e-05, "loss": 3.5705, "step": 5839 }, { "epoch": 0.36253026258613197, "grad_norm": 0.6339038574459442, "learning_rate": 9.986779480671782e-05, "loss": 3.5554, "step": 5840 }, { "epoch": 0.3625923396858899, "grad_norm": 0.5522764216155689, "learning_rate": 9.986753222002463e-05, "loss": 3.5402, "step": 5841 }, { "epoch": 0.36265441678564775, "grad_norm": 0.5651963828183034, "learning_rate": 9.986726937316041e-05, "loss": 3.5432, "step": 5842 }, { "epoch": 0.36271649388540567, "grad_norm": 0.5375112244077218, "learning_rate": 9.986700626612654e-05, "loss": 3.5058, "step": 5843 }, { "epoch": 0.3627785709851636, "grad_norm": 0.5510624647872756, "learning_rate": 9.98667428989244e-05, "loss": 3.5068, "step": 5844 }, { "epoch": 0.36284064808492145, "grad_norm": 0.5738723904004648, "learning_rate": 9.986647927155534e-05, "loss": 3.6983, "step": 5845 }, { "epoch": 0.36290272518467936, "grad_norm": 0.5760350727671946, "learning_rate": 9.986621538402078e-05, "loss": 3.4916, "step": 5846 }, { "epoch": 0.3629648022844373, "grad_norm": 0.5557273734819631, "learning_rate": 9.986595123632206e-05, "loss": 3.6325, "step": 5847 }, { "epoch": 0.36302687938419514, "grad_norm": 0.6219301170575637, "learning_rate": 9.986568682846055e-05, "loss": 3.5136, "step": 5848 }, { "epoch": 0.36308895648395306, "grad_norm": 0.5624885557917855, "learning_rate": 9.986542216043766e-05, "loss": 3.4739, "step": 5849 }, { "epoch": 0.363151033583711, "grad_norm": 0.5303407910813854, "learning_rate": 9.986515723225476e-05, "loss": 3.5113, "step": 5850 }, { "epoch": 0.36321311068346884, "grad_norm": 0.6467151057234447, "learning_rate": 9.986489204391323e-05, "loss": 3.4309, "step": 5851 }, { "epoch": 0.36327518778322676, "grad_norm": 0.6124676452576928, "learning_rate": 9.986462659541445e-05, "loss": 3.5517, "step": 5852 }, { "epoch": 0.3633372648829847, "grad_norm": 0.5506177618999373, "learning_rate": 9.986436088675978e-05, "loss": 3.5715, "step": 5853 }, { "epoch": 0.36339934198274254, "grad_norm": 0.5896223802296041, "learning_rate": 9.986409491795067e-05, "loss": 3.4311, "step": 5854 }, { "epoch": 0.36346141908250046, "grad_norm": 0.5476787110863813, "learning_rate": 9.986382868898847e-05, "loss": 3.4884, "step": 5855 }, { "epoch": 0.3635234961822584, "grad_norm": 0.6181569031214929, "learning_rate": 9.986356219987456e-05, "loss": 3.4078, "step": 5856 }, { "epoch": 0.36358557328201624, "grad_norm": 0.5895773178530588, "learning_rate": 9.986329545061034e-05, "loss": 3.5362, "step": 5857 }, { "epoch": 0.36364765038177416, "grad_norm": 0.5723165808375233, "learning_rate": 9.986302844119722e-05, "loss": 3.5538, "step": 5858 }, { "epoch": 0.3637097274815321, "grad_norm": 0.5775189960955497, "learning_rate": 9.986276117163656e-05, "loss": 3.5852, "step": 5859 }, { "epoch": 0.36377180458128994, "grad_norm": 0.5606032020898607, "learning_rate": 9.986249364192977e-05, "loss": 3.6548, "step": 5860 }, { "epoch": 0.36383388168104785, "grad_norm": 0.5686054811161151, "learning_rate": 9.986222585207825e-05, "loss": 3.5063, "step": 5861 }, { "epoch": 0.3638959587808058, "grad_norm": 0.5900703651688711, "learning_rate": 9.986195780208338e-05, "loss": 3.599, "step": 5862 }, { "epoch": 0.36395803588056364, "grad_norm": 0.5408045773681363, "learning_rate": 9.986168949194658e-05, "loss": 3.607, "step": 5863 }, { "epoch": 0.36402011298032155, "grad_norm": 0.5445376172781231, "learning_rate": 9.986142092166926e-05, "loss": 3.5046, "step": 5864 }, { "epoch": 0.36408219008007947, "grad_norm": 0.6011378783688601, "learning_rate": 9.986115209125278e-05, "loss": 3.4655, "step": 5865 }, { "epoch": 0.36414426717983733, "grad_norm": 0.5807473018762558, "learning_rate": 9.986088300069858e-05, "loss": 3.5637, "step": 5866 }, { "epoch": 0.36420634427959525, "grad_norm": 0.5965218592808899, "learning_rate": 9.986061365000804e-05, "loss": 3.595, "step": 5867 }, { "epoch": 0.36426842137935317, "grad_norm": 0.5602324664033258, "learning_rate": 9.986034403918258e-05, "loss": 3.5983, "step": 5868 }, { "epoch": 0.36433049847911103, "grad_norm": 0.605054381306718, "learning_rate": 9.98600741682236e-05, "loss": 3.4502, "step": 5869 }, { "epoch": 0.36439257557886895, "grad_norm": 0.5379504467645075, "learning_rate": 9.985980403713251e-05, "loss": 3.5366, "step": 5870 }, { "epoch": 0.36445465267862687, "grad_norm": 0.4776798812965392, "learning_rate": 9.985953364591073e-05, "loss": 3.5353, "step": 5871 }, { "epoch": 0.36451672977838473, "grad_norm": 0.5701783950113827, "learning_rate": 9.985926299455964e-05, "loss": 3.5456, "step": 5872 }, { "epoch": 0.36457880687814265, "grad_norm": 0.6212979586796893, "learning_rate": 9.985899208308068e-05, "loss": 3.5838, "step": 5873 }, { "epoch": 0.36464088397790057, "grad_norm": 0.6434705850062694, "learning_rate": 9.985872091147523e-05, "loss": 3.4806, "step": 5874 }, { "epoch": 0.3647029610776584, "grad_norm": 0.6073536236556369, "learning_rate": 9.985844947974476e-05, "loss": 3.4687, "step": 5875 }, { "epoch": 0.36476503817741635, "grad_norm": 0.5007290888860917, "learning_rate": 9.985817778789064e-05, "loss": 3.5527, "step": 5876 }, { "epoch": 0.36482711527717426, "grad_norm": 0.6458638835162211, "learning_rate": 9.985790583591431e-05, "loss": 3.4928, "step": 5877 }, { "epoch": 0.3648891923769321, "grad_norm": 0.6119862143469201, "learning_rate": 9.985763362381718e-05, "loss": 3.5138, "step": 5878 }, { "epoch": 0.36495126947669004, "grad_norm": 0.5630411617856724, "learning_rate": 9.985736115160067e-05, "loss": 3.539, "step": 5879 }, { "epoch": 0.36501334657644796, "grad_norm": 0.5818951082698901, "learning_rate": 9.985708841926622e-05, "loss": 3.5632, "step": 5880 }, { "epoch": 0.3650754236762058, "grad_norm": 0.5502861178224692, "learning_rate": 9.985681542681522e-05, "loss": 3.5448, "step": 5881 }, { "epoch": 0.36513750077596374, "grad_norm": 0.5767096902760449, "learning_rate": 9.985654217424911e-05, "loss": 3.481, "step": 5882 }, { "epoch": 0.36519957787572166, "grad_norm": 0.5395380769502621, "learning_rate": 9.985626866156932e-05, "loss": 3.4983, "step": 5883 }, { "epoch": 0.3652616549754795, "grad_norm": 0.5123516275045173, "learning_rate": 9.985599488877728e-05, "loss": 3.5702, "step": 5884 }, { "epoch": 0.36532373207523744, "grad_norm": 0.5575032827682666, "learning_rate": 9.985572085587441e-05, "loss": 3.5383, "step": 5885 }, { "epoch": 0.36538580917499536, "grad_norm": 0.6008628640962014, "learning_rate": 9.985544656286213e-05, "loss": 3.6087, "step": 5886 }, { "epoch": 0.3654478862747532, "grad_norm": 0.5821403298795372, "learning_rate": 9.985517200974188e-05, "loss": 3.6171, "step": 5887 }, { "epoch": 0.36550996337451114, "grad_norm": 0.6126774530844896, "learning_rate": 9.98548971965151e-05, "loss": 3.4882, "step": 5888 }, { "epoch": 0.36557204047426906, "grad_norm": 0.6300887370058422, "learning_rate": 9.985462212318323e-05, "loss": 3.5696, "step": 5889 }, { "epoch": 0.3656341175740269, "grad_norm": 0.6092572709824854, "learning_rate": 9.985434678974769e-05, "loss": 3.5485, "step": 5890 }, { "epoch": 0.36569619467378484, "grad_norm": 0.588798936523055, "learning_rate": 9.985407119620991e-05, "loss": 3.5063, "step": 5891 }, { "epoch": 0.36575827177354275, "grad_norm": 0.7094257336594272, "learning_rate": 9.985379534257135e-05, "loss": 3.5273, "step": 5892 }, { "epoch": 0.3658203488733006, "grad_norm": 0.7605900153183563, "learning_rate": 9.985351922883344e-05, "loss": 3.5289, "step": 5893 }, { "epoch": 0.36588242597305853, "grad_norm": 0.6652455954465021, "learning_rate": 9.98532428549976e-05, "loss": 3.591, "step": 5894 }, { "epoch": 0.36594450307281645, "grad_norm": 0.6374946734261976, "learning_rate": 9.985296622106531e-05, "loss": 3.461, "step": 5895 }, { "epoch": 0.3660065801725743, "grad_norm": 0.6600855050186297, "learning_rate": 9.985268932703797e-05, "loss": 3.4584, "step": 5896 }, { "epoch": 0.36606865727233223, "grad_norm": 0.5791027461485321, "learning_rate": 9.985241217291706e-05, "loss": 3.5726, "step": 5897 }, { "epoch": 0.36613073437209015, "grad_norm": 0.5348544139729493, "learning_rate": 9.985213475870401e-05, "loss": 3.5703, "step": 5898 }, { "epoch": 0.366192811471848, "grad_norm": 0.630751963822146, "learning_rate": 9.985185708440028e-05, "loss": 3.5209, "step": 5899 }, { "epoch": 0.36625488857160593, "grad_norm": 0.6438183859269172, "learning_rate": 9.98515791500073e-05, "loss": 3.5527, "step": 5900 }, { "epoch": 0.36631696567136385, "grad_norm": 0.6469803489371668, "learning_rate": 9.985130095552654e-05, "loss": 3.5367, "step": 5901 }, { "epoch": 0.3663790427711217, "grad_norm": 0.6001408610298901, "learning_rate": 9.985102250095943e-05, "loss": 3.5214, "step": 5902 }, { "epoch": 0.36644111987087963, "grad_norm": 0.6010538898814024, "learning_rate": 9.985074378630745e-05, "loss": 3.5132, "step": 5903 }, { "epoch": 0.36650319697063755, "grad_norm": 0.6379725275761685, "learning_rate": 9.985046481157202e-05, "loss": 3.5511, "step": 5904 }, { "epoch": 0.3665652740703954, "grad_norm": 0.6485006232732926, "learning_rate": 9.985018557675462e-05, "loss": 3.4388, "step": 5905 }, { "epoch": 0.3666273511701533, "grad_norm": 0.5262283680019477, "learning_rate": 9.984990608185669e-05, "loss": 3.4942, "step": 5906 }, { "epoch": 0.36668942826991124, "grad_norm": 0.6379201355442619, "learning_rate": 9.984962632687972e-05, "loss": 3.4889, "step": 5907 }, { "epoch": 0.3667515053696691, "grad_norm": 0.6798895612259537, "learning_rate": 9.984934631182515e-05, "loss": 3.4836, "step": 5908 }, { "epoch": 0.366813582469427, "grad_norm": 0.5938588343808614, "learning_rate": 9.984906603669442e-05, "loss": 3.4504, "step": 5909 }, { "epoch": 0.36687565956918494, "grad_norm": 0.6635611717395508, "learning_rate": 9.9848785501489e-05, "loss": 3.4931, "step": 5910 }, { "epoch": 0.3669377366689428, "grad_norm": 0.6215610545560161, "learning_rate": 9.984850470621039e-05, "loss": 3.429, "step": 5911 }, { "epoch": 0.3669998137687007, "grad_norm": 0.6069109322451904, "learning_rate": 9.984822365086002e-05, "loss": 3.6474, "step": 5912 }, { "epoch": 0.36706189086845864, "grad_norm": 0.6068052737785755, "learning_rate": 9.984794233543937e-05, "loss": 3.4704, "step": 5913 }, { "epoch": 0.3671239679682165, "grad_norm": 0.5967376730756002, "learning_rate": 9.98476607599499e-05, "loss": 3.5744, "step": 5914 }, { "epoch": 0.3671860450679744, "grad_norm": 0.5694286342948527, "learning_rate": 9.984737892439309e-05, "loss": 3.5461, "step": 5915 }, { "epoch": 0.36724812216773234, "grad_norm": 0.5184280327041161, "learning_rate": 9.984709682877039e-05, "loss": 3.5316, "step": 5916 }, { "epoch": 0.3673101992674902, "grad_norm": 0.5422816972503726, "learning_rate": 9.98468144730833e-05, "loss": 3.5499, "step": 5917 }, { "epoch": 0.3673722763672481, "grad_norm": 0.5222759168776331, "learning_rate": 9.984653185733327e-05, "loss": 3.4363, "step": 5918 }, { "epoch": 0.36743435346700604, "grad_norm": 0.6840651058741637, "learning_rate": 9.984624898152178e-05, "loss": 3.4731, "step": 5919 }, { "epoch": 0.3674964305667639, "grad_norm": 0.7297564618934401, "learning_rate": 9.984596584565032e-05, "loss": 3.4725, "step": 5920 }, { "epoch": 0.3675585076665218, "grad_norm": 0.601898198774957, "learning_rate": 9.984568244972034e-05, "loss": 3.6006, "step": 5921 }, { "epoch": 0.36762058476627973, "grad_norm": 0.6753347608718144, "learning_rate": 9.984539879373335e-05, "loss": 3.5094, "step": 5922 }, { "epoch": 0.3676826618660376, "grad_norm": 0.6592026317983122, "learning_rate": 9.984511487769079e-05, "loss": 3.4446, "step": 5923 }, { "epoch": 0.3677447389657955, "grad_norm": 0.7820185886085972, "learning_rate": 9.98448307015942e-05, "loss": 3.542, "step": 5924 }, { "epoch": 0.36780681606555343, "grad_norm": 0.7214192293390467, "learning_rate": 9.9844546265445e-05, "loss": 3.5354, "step": 5925 }, { "epoch": 0.3678688931653113, "grad_norm": 0.6411163260277922, "learning_rate": 9.984426156924471e-05, "loss": 3.5978, "step": 5926 }, { "epoch": 0.3679309702650692, "grad_norm": 0.697565009119535, "learning_rate": 9.98439766129948e-05, "loss": 3.4757, "step": 5927 }, { "epoch": 0.36799304736482713, "grad_norm": 0.5727477662780975, "learning_rate": 9.984369139669678e-05, "loss": 3.5597, "step": 5928 }, { "epoch": 0.368055124464585, "grad_norm": 0.6611981944178363, "learning_rate": 9.984340592035211e-05, "loss": 3.5289, "step": 5929 }, { "epoch": 0.3681172015643429, "grad_norm": 0.6194667258220663, "learning_rate": 9.984312018396231e-05, "loss": 3.6387, "step": 5930 }, { "epoch": 0.36817927866410083, "grad_norm": 0.5881475096478218, "learning_rate": 9.984283418752884e-05, "loss": 3.5622, "step": 5931 }, { "epoch": 0.3682413557638587, "grad_norm": 0.6406988473523118, "learning_rate": 9.98425479310532e-05, "loss": 3.5532, "step": 5932 }, { "epoch": 0.3683034328636166, "grad_norm": 0.632028871586006, "learning_rate": 9.984226141453689e-05, "loss": 3.5748, "step": 5933 }, { "epoch": 0.3683655099633745, "grad_norm": 0.6799042893118592, "learning_rate": 9.98419746379814e-05, "loss": 3.4615, "step": 5934 }, { "epoch": 0.3684275870631324, "grad_norm": 0.6102834434643707, "learning_rate": 9.984168760138823e-05, "loss": 3.6843, "step": 5935 }, { "epoch": 0.3684896641628903, "grad_norm": 0.6955271400309319, "learning_rate": 9.984140030475889e-05, "loss": 3.4198, "step": 5936 }, { "epoch": 0.3685517412626482, "grad_norm": 0.6446295460971014, "learning_rate": 9.984111274809485e-05, "loss": 3.5299, "step": 5937 }, { "epoch": 0.3686138183624061, "grad_norm": 0.6383139397644043, "learning_rate": 9.984082493139761e-05, "loss": 3.4649, "step": 5938 }, { "epoch": 0.368675895462164, "grad_norm": 0.7157452844537031, "learning_rate": 9.984053685466872e-05, "loss": 3.6054, "step": 5939 }, { "epoch": 0.3687379725619219, "grad_norm": 0.6677978029182396, "learning_rate": 9.984024851790963e-05, "loss": 3.5527, "step": 5940 }, { "epoch": 0.3688000496616798, "grad_norm": 0.5557333797401197, "learning_rate": 9.983995992112187e-05, "loss": 3.5028, "step": 5941 }, { "epoch": 0.3688621267614377, "grad_norm": 0.5879949432791394, "learning_rate": 9.983967106430694e-05, "loss": 3.5209, "step": 5942 }, { "epoch": 0.3689242038611956, "grad_norm": 0.6333926802569688, "learning_rate": 9.983938194746635e-05, "loss": 3.5648, "step": 5943 }, { "epoch": 0.3689862809609535, "grad_norm": 0.5425766826085912, "learning_rate": 9.98390925706016e-05, "loss": 3.5438, "step": 5944 }, { "epoch": 0.3690483580607114, "grad_norm": 0.46299884030174204, "learning_rate": 9.983880293371422e-05, "loss": 3.5019, "step": 5945 }, { "epoch": 0.3691104351604693, "grad_norm": 0.517822265625, "learning_rate": 9.98385130368057e-05, "loss": 3.471, "step": 5946 }, { "epoch": 0.3691725122602272, "grad_norm": 0.49021818886590607, "learning_rate": 9.983822287987755e-05, "loss": 3.5732, "step": 5947 }, { "epoch": 0.3692345893599851, "grad_norm": 0.48468264684946716, "learning_rate": 9.983793246293128e-05, "loss": 3.5661, "step": 5948 }, { "epoch": 0.369296666459743, "grad_norm": 0.5107954883662222, "learning_rate": 9.983764178596844e-05, "loss": 3.5064, "step": 5949 }, { "epoch": 0.3693587435595009, "grad_norm": 0.5023100359624745, "learning_rate": 9.983735084899051e-05, "loss": 3.5383, "step": 5950 }, { "epoch": 0.3694208206592588, "grad_norm": 0.5473282434443281, "learning_rate": 9.983705965199904e-05, "loss": 3.4004, "step": 5951 }, { "epoch": 0.3694828977590167, "grad_norm": 0.5705645017527151, "learning_rate": 9.983676819499551e-05, "loss": 3.5014, "step": 5952 }, { "epoch": 0.3695449748587746, "grad_norm": 0.5744202707857268, "learning_rate": 9.983647647798147e-05, "loss": 3.4554, "step": 5953 }, { "epoch": 0.3696070519585325, "grad_norm": 0.5953982464170277, "learning_rate": 9.983618450095844e-05, "loss": 3.5579, "step": 5954 }, { "epoch": 0.3696691290582904, "grad_norm": 0.5839689373923035, "learning_rate": 9.983589226392793e-05, "loss": 3.5679, "step": 5955 }, { "epoch": 0.3697312061580483, "grad_norm": 0.5913810654068911, "learning_rate": 9.983559976689149e-05, "loss": 3.4859, "step": 5956 }, { "epoch": 0.3697932832578062, "grad_norm": 0.5666422122232004, "learning_rate": 9.983530700985062e-05, "loss": 3.4517, "step": 5957 }, { "epoch": 0.3698553603575641, "grad_norm": 0.5193885090857915, "learning_rate": 9.983501399280684e-05, "loss": 3.4923, "step": 5958 }, { "epoch": 0.369917437457322, "grad_norm": 0.63225585096585, "learning_rate": 9.98347207157617e-05, "loss": 3.4661, "step": 5959 }, { "epoch": 0.3699795145570799, "grad_norm": 0.561744394366923, "learning_rate": 9.983442717871672e-05, "loss": 3.5604, "step": 5960 }, { "epoch": 0.3700415916568378, "grad_norm": 0.5833408803678676, "learning_rate": 9.983413338167346e-05, "loss": 3.548, "step": 5961 }, { "epoch": 0.37010366875659567, "grad_norm": 0.6684731926646682, "learning_rate": 9.98338393246334e-05, "loss": 3.4378, "step": 5962 }, { "epoch": 0.3701657458563536, "grad_norm": 0.6178715211020605, "learning_rate": 9.983354500759811e-05, "loss": 3.5094, "step": 5963 }, { "epoch": 0.3702278229561115, "grad_norm": 0.5435072883338365, "learning_rate": 9.983325043056914e-05, "loss": 3.5528, "step": 5964 }, { "epoch": 0.37028990005586937, "grad_norm": 0.5376607665559855, "learning_rate": 9.983295559354798e-05, "loss": 3.5055, "step": 5965 }, { "epoch": 0.3703519771556273, "grad_norm": 0.6195659680638396, "learning_rate": 9.98326604965362e-05, "loss": 3.4922, "step": 5966 }, { "epoch": 0.3704140542553852, "grad_norm": 0.6133020482515336, "learning_rate": 9.983236513953534e-05, "loss": 3.4464, "step": 5967 }, { "epoch": 0.37047613135514307, "grad_norm": 0.5748895373180816, "learning_rate": 9.983206952254693e-05, "loss": 3.5374, "step": 5968 }, { "epoch": 0.370538208454901, "grad_norm": 0.5608272582297118, "learning_rate": 9.983177364557252e-05, "loss": 3.5407, "step": 5969 }, { "epoch": 0.3706002855546589, "grad_norm": 0.5831057467864854, "learning_rate": 9.983147750861364e-05, "loss": 3.5157, "step": 5970 }, { "epoch": 0.37066236265441677, "grad_norm": 0.5739096503776268, "learning_rate": 9.983118111167184e-05, "loss": 3.4767, "step": 5971 }, { "epoch": 0.3707244397541747, "grad_norm": 0.5086258756508629, "learning_rate": 9.983088445474868e-05, "loss": 3.5132, "step": 5972 }, { "epoch": 0.3707865168539326, "grad_norm": 0.5523077190900624, "learning_rate": 9.983058753784571e-05, "loss": 3.5073, "step": 5973 }, { "epoch": 0.37084859395369046, "grad_norm": 0.6256946279934995, "learning_rate": 9.983029036096445e-05, "loss": 3.547, "step": 5974 }, { "epoch": 0.3709106710534484, "grad_norm": 0.5761165527681255, "learning_rate": 9.982999292410647e-05, "loss": 3.5057, "step": 5975 }, { "epoch": 0.3709727481532063, "grad_norm": 0.6050813235253153, "learning_rate": 9.982969522727333e-05, "loss": 3.5569, "step": 5976 }, { "epoch": 0.37103482525296416, "grad_norm": 0.5097037676116082, "learning_rate": 9.982939727046656e-05, "loss": 3.5903, "step": 5977 }, { "epoch": 0.3710969023527221, "grad_norm": 0.5068565467363499, "learning_rate": 9.982909905368775e-05, "loss": 3.4876, "step": 5978 }, { "epoch": 0.37115897945248, "grad_norm": 0.6110832040608781, "learning_rate": 9.982880057693842e-05, "loss": 3.5523, "step": 5979 }, { "epoch": 0.37122105655223786, "grad_norm": 0.5784353505758767, "learning_rate": 9.982850184022014e-05, "loss": 3.5315, "step": 5980 }, { "epoch": 0.3712831336519958, "grad_norm": 0.5525107318819763, "learning_rate": 9.982820284353447e-05, "loss": 3.5094, "step": 5981 }, { "epoch": 0.3713452107517537, "grad_norm": 0.4771422783559115, "learning_rate": 9.982790358688296e-05, "loss": 3.4194, "step": 5982 }, { "epoch": 0.37140728785151156, "grad_norm": 0.5165153676689078, "learning_rate": 9.982760407026721e-05, "loss": 3.4942, "step": 5983 }, { "epoch": 0.3714693649512695, "grad_norm": 0.5330671199849294, "learning_rate": 9.982730429368872e-05, "loss": 3.4057, "step": 5984 }, { "epoch": 0.3715314420510274, "grad_norm": 0.6139894757518239, "learning_rate": 9.982700425714911e-05, "loss": 3.4321, "step": 5985 }, { "epoch": 0.37159351915078526, "grad_norm": 0.6353431304737989, "learning_rate": 9.982670396064992e-05, "loss": 3.436, "step": 5986 }, { "epoch": 0.3716555962505432, "grad_norm": 0.735913025596866, "learning_rate": 9.982640340419272e-05, "loss": 3.5984, "step": 5987 }, { "epoch": 0.3717176733503011, "grad_norm": 0.7823228717240462, "learning_rate": 9.982610258777907e-05, "loss": 3.5534, "step": 5988 }, { "epoch": 0.37177975045005895, "grad_norm": 0.6817579195816672, "learning_rate": 9.982580151141055e-05, "loss": 3.5288, "step": 5989 }, { "epoch": 0.3718418275498169, "grad_norm": 0.6301904917521778, "learning_rate": 9.982550017508873e-05, "loss": 3.5247, "step": 5990 }, { "epoch": 0.3719039046495748, "grad_norm": 0.621530942772657, "learning_rate": 9.982519857881518e-05, "loss": 3.5313, "step": 5991 }, { "epoch": 0.37196598174933265, "grad_norm": 0.5711876151496662, "learning_rate": 9.982489672259147e-05, "loss": 3.5326, "step": 5992 }, { "epoch": 0.37202805884909057, "grad_norm": 0.5891275921381519, "learning_rate": 9.982459460641918e-05, "loss": 3.5597, "step": 5993 }, { "epoch": 0.3720901359488485, "grad_norm": 0.6358190159000217, "learning_rate": 9.982429223029988e-05, "loss": 3.489, "step": 5994 }, { "epoch": 0.37215221304860635, "grad_norm": 0.5631070040792072, "learning_rate": 9.982398959423515e-05, "loss": 3.4507, "step": 5995 }, { "epoch": 0.37221429014836427, "grad_norm": 0.5063731114491897, "learning_rate": 9.98236866982266e-05, "loss": 3.3987, "step": 5996 }, { "epoch": 0.3722763672481222, "grad_norm": 0.5831553976838627, "learning_rate": 9.982338354227573e-05, "loss": 3.519, "step": 5997 }, { "epoch": 0.37233844434788005, "grad_norm": 0.6190964595794497, "learning_rate": 9.982308012638422e-05, "loss": 3.5368, "step": 5998 }, { "epoch": 0.37240052144763797, "grad_norm": 0.564408006811832, "learning_rate": 9.982277645055357e-05, "loss": 3.4457, "step": 5999 }, { "epoch": 0.3724625985473959, "grad_norm": 0.5657640697164255, "learning_rate": 9.982247251478543e-05, "loss": 3.4694, "step": 6000 }, { "epoch": 0.37252467564715375, "grad_norm": 0.6743020422839924, "learning_rate": 9.982216831908135e-05, "loss": 3.3639, "step": 6001 }, { "epoch": 0.37258675274691166, "grad_norm": 0.7020957407057835, "learning_rate": 9.982186386344292e-05, "loss": 3.477, "step": 6002 }, { "epoch": 0.3726488298466696, "grad_norm": 0.6372462468850504, "learning_rate": 9.982155914787173e-05, "loss": 3.4526, "step": 6003 }, { "epoch": 0.37271090694642744, "grad_norm": 0.5989236793997442, "learning_rate": 9.982125417236935e-05, "loss": 3.5008, "step": 6004 }, { "epoch": 0.37277298404618536, "grad_norm": 0.4817585574123989, "learning_rate": 9.982094893693743e-05, "loss": 3.4478, "step": 6005 }, { "epoch": 0.3728350611459433, "grad_norm": 0.5203294287324712, "learning_rate": 9.982064344157752e-05, "loss": 3.6591, "step": 6006 }, { "epoch": 0.37289713824570114, "grad_norm": 0.6025740553032113, "learning_rate": 9.98203376862912e-05, "loss": 3.5336, "step": 6007 }, { "epoch": 0.37295921534545906, "grad_norm": 0.6455744537257252, "learning_rate": 9.982003167108008e-05, "loss": 3.4933, "step": 6008 }, { "epoch": 0.373021292445217, "grad_norm": 0.6477003158469031, "learning_rate": 9.981972539594577e-05, "loss": 3.4284, "step": 6009 }, { "epoch": 0.37308336954497484, "grad_norm": 0.6771268952857061, "learning_rate": 9.981941886088986e-05, "loss": 3.5088, "step": 6010 }, { "epoch": 0.37314544664473276, "grad_norm": 0.6169818583870975, "learning_rate": 9.981911206591395e-05, "loss": 3.4286, "step": 6011 }, { "epoch": 0.3732075237444907, "grad_norm": 0.5095624262944101, "learning_rate": 9.981880501101964e-05, "loss": 3.6042, "step": 6012 }, { "epoch": 0.37326960084424854, "grad_norm": 0.5381470245745499, "learning_rate": 9.981849769620852e-05, "loss": 3.5971, "step": 6013 }, { "epoch": 0.37333167794400646, "grad_norm": 0.5326539169981073, "learning_rate": 9.981819012148221e-05, "loss": 3.496, "step": 6014 }, { "epoch": 0.3733937550437644, "grad_norm": 0.481781059019149, "learning_rate": 9.981788228684231e-05, "loss": 3.5586, "step": 6015 }, { "epoch": 0.37345583214352224, "grad_norm": 0.5239264354008468, "learning_rate": 9.981757419229043e-05, "loss": 3.6096, "step": 6016 }, { "epoch": 0.37351790924328016, "grad_norm": 0.5631657739965057, "learning_rate": 9.981726583782816e-05, "loss": 3.3756, "step": 6017 }, { "epoch": 0.3735799863430381, "grad_norm": 0.5682507178406517, "learning_rate": 9.981695722345712e-05, "loss": 3.4565, "step": 6018 }, { "epoch": 0.37364206344279594, "grad_norm": 0.5430747827419982, "learning_rate": 9.981664834917893e-05, "loss": 3.4988, "step": 6019 }, { "epoch": 0.37370414054255385, "grad_norm": 0.5258756759671853, "learning_rate": 9.981633921499519e-05, "loss": 3.4957, "step": 6020 }, { "epoch": 0.37376621764231177, "grad_norm": 0.5081089621590039, "learning_rate": 9.981602982090751e-05, "loss": 3.42, "step": 6021 }, { "epoch": 0.37382829474206963, "grad_norm": 0.5832025404030999, "learning_rate": 9.981572016691752e-05, "loss": 3.4863, "step": 6022 }, { "epoch": 0.37389037184182755, "grad_norm": 0.6607448955573985, "learning_rate": 9.981541025302681e-05, "loss": 3.5006, "step": 6023 }, { "epoch": 0.37395244894158547, "grad_norm": 0.646719575721525, "learning_rate": 9.981510007923702e-05, "loss": 3.5937, "step": 6024 }, { "epoch": 0.37401452604134333, "grad_norm": 0.5552547670406215, "learning_rate": 9.981478964554976e-05, "loss": 3.4408, "step": 6025 }, { "epoch": 0.37407660314110125, "grad_norm": 0.5536760591693617, "learning_rate": 9.981447895196665e-05, "loss": 3.5382, "step": 6026 }, { "epoch": 0.37413868024085917, "grad_norm": 0.5613522475971889, "learning_rate": 9.981416799848931e-05, "loss": 3.4608, "step": 6027 }, { "epoch": 0.37420075734061703, "grad_norm": 0.6825036430610988, "learning_rate": 9.981385678511935e-05, "loss": 3.4592, "step": 6028 }, { "epoch": 0.37426283444037495, "grad_norm": 0.6426793862186515, "learning_rate": 9.981354531185843e-05, "loss": 3.4504, "step": 6029 }, { "epoch": 0.37432491154013287, "grad_norm": 0.5789977265232629, "learning_rate": 9.981323357870813e-05, "loss": 3.6235, "step": 6030 }, { "epoch": 0.3743869886398907, "grad_norm": 0.5822110762355317, "learning_rate": 9.981292158567011e-05, "loss": 3.5301, "step": 6031 }, { "epoch": 0.37444906573964865, "grad_norm": 0.5810010356335757, "learning_rate": 9.981260933274598e-05, "loss": 3.4564, "step": 6032 }, { "epoch": 0.37451114283940656, "grad_norm": 0.6608228759932377, "learning_rate": 9.981229681993738e-05, "loss": 3.4801, "step": 6033 }, { "epoch": 0.3745732199391644, "grad_norm": 0.5849146243502363, "learning_rate": 9.981198404724592e-05, "loss": 3.4671, "step": 6034 }, { "epoch": 0.37463529703892234, "grad_norm": 0.5286993334778808, "learning_rate": 9.981167101467325e-05, "loss": 3.466, "step": 6035 }, { "epoch": 0.37469737413868026, "grad_norm": 0.6243065086044413, "learning_rate": 9.981135772222102e-05, "loss": 3.3755, "step": 6036 }, { "epoch": 0.3747594512384381, "grad_norm": 0.5522124181542769, "learning_rate": 9.981104416989083e-05, "loss": 3.4517, "step": 6037 }, { "epoch": 0.37482152833819604, "grad_norm": 0.6741284792905967, "learning_rate": 9.981073035768433e-05, "loss": 3.5648, "step": 6038 }, { "epoch": 0.37488360543795396, "grad_norm": 0.7168690076077016, "learning_rate": 9.981041628560316e-05, "loss": 3.4775, "step": 6039 }, { "epoch": 0.3749456825377118, "grad_norm": 0.627472588027774, "learning_rate": 9.981010195364897e-05, "loss": 3.5252, "step": 6040 }, { "epoch": 0.37500775963746974, "grad_norm": 0.6070888624504589, "learning_rate": 9.980978736182338e-05, "loss": 3.5527, "step": 6041 }, { "epoch": 0.37506983673722766, "grad_norm": 0.5943995737236646, "learning_rate": 9.980947251012802e-05, "loss": 3.488, "step": 6042 }, { "epoch": 0.3751319138369855, "grad_norm": 0.6216200032588544, "learning_rate": 9.980915739856457e-05, "loss": 3.5236, "step": 6043 }, { "epoch": 0.37519399093674344, "grad_norm": 0.5742107572453403, "learning_rate": 9.980884202713464e-05, "loss": 3.5547, "step": 6044 }, { "epoch": 0.37525606803650136, "grad_norm": 0.5327548423736769, "learning_rate": 9.980852639583991e-05, "loss": 3.4747, "step": 6045 }, { "epoch": 0.3753181451362592, "grad_norm": 0.6076614300013423, "learning_rate": 9.980821050468198e-05, "loss": 3.4922, "step": 6046 }, { "epoch": 0.37538022223601714, "grad_norm": 0.5834037675796045, "learning_rate": 9.980789435366253e-05, "loss": 3.5402, "step": 6047 }, { "epoch": 0.37544229933577505, "grad_norm": 0.5867032959549793, "learning_rate": 9.980757794278322e-05, "loss": 3.5488, "step": 6048 }, { "epoch": 0.3755043764355329, "grad_norm": 0.55899041770372, "learning_rate": 9.980726127204567e-05, "loss": 3.5168, "step": 6049 }, { "epoch": 0.37556645353529083, "grad_norm": 0.517600667210684, "learning_rate": 9.980694434145155e-05, "loss": 3.5661, "step": 6050 }, { "epoch": 0.37562853063504875, "grad_norm": 0.5673953923260515, "learning_rate": 9.980662715100252e-05, "loss": 3.5021, "step": 6051 }, { "epoch": 0.3756906077348066, "grad_norm": 0.574688207015017, "learning_rate": 9.980630970070022e-05, "loss": 3.4889, "step": 6052 }, { "epoch": 0.37575268483456453, "grad_norm": 0.5522748836767081, "learning_rate": 9.98059919905463e-05, "loss": 3.4694, "step": 6053 }, { "epoch": 0.37581476193432245, "grad_norm": 0.5486368409358933, "learning_rate": 9.980567402054243e-05, "loss": 3.5748, "step": 6054 }, { "epoch": 0.3758768390340803, "grad_norm": 0.550859459268501, "learning_rate": 9.98053557906903e-05, "loss": 3.5486, "step": 6055 }, { "epoch": 0.37593891613383823, "grad_norm": 0.6212662990972528, "learning_rate": 9.980503730099152e-05, "loss": 3.4306, "step": 6056 }, { "epoch": 0.37600099323359615, "grad_norm": 0.6616777796937087, "learning_rate": 9.980471855144775e-05, "loss": 3.5702, "step": 6057 }, { "epoch": 0.376063070333354, "grad_norm": 0.631423650594794, "learning_rate": 9.980439954206069e-05, "loss": 3.5304, "step": 6058 }, { "epoch": 0.37612514743311193, "grad_norm": 0.6013449486055101, "learning_rate": 9.980408027283198e-05, "loss": 3.4999, "step": 6059 }, { "epoch": 0.37618722453286985, "grad_norm": 0.5391234764220876, "learning_rate": 9.980376074376329e-05, "loss": 3.4995, "step": 6060 }, { "epoch": 0.3762493016326277, "grad_norm": 0.5356702076852847, "learning_rate": 9.98034409548563e-05, "loss": 3.469, "step": 6061 }, { "epoch": 0.3763113787323856, "grad_norm": 0.5658633820453629, "learning_rate": 9.980312090611267e-05, "loss": 3.4198, "step": 6062 }, { "epoch": 0.37637345583214354, "grad_norm": 0.6462196199813499, "learning_rate": 9.980280059753406e-05, "loss": 3.5316, "step": 6063 }, { "epoch": 0.3764355329319014, "grad_norm": 0.49480473517844337, "learning_rate": 9.980248002912216e-05, "loss": 3.4506, "step": 6064 }, { "epoch": 0.3764976100316593, "grad_norm": 0.5932479542834814, "learning_rate": 9.980215920087861e-05, "loss": 3.4927, "step": 6065 }, { "epoch": 0.37655968713141724, "grad_norm": 0.6573380805030175, "learning_rate": 9.980183811280512e-05, "loss": 3.4415, "step": 6066 }, { "epoch": 0.3766217642311751, "grad_norm": 0.6398743907406159, "learning_rate": 9.980151676490336e-05, "loss": 3.4965, "step": 6067 }, { "epoch": 0.376683841330933, "grad_norm": 0.6448303453453934, "learning_rate": 9.980119515717498e-05, "loss": 3.4926, "step": 6068 }, { "epoch": 0.37674591843069094, "grad_norm": 0.6057838112435301, "learning_rate": 9.980087328962169e-05, "loss": 3.514, "step": 6069 }, { "epoch": 0.3768079955304488, "grad_norm": 0.6204915035843401, "learning_rate": 9.980055116224514e-05, "loss": 3.4962, "step": 6070 }, { "epoch": 0.3768700726302067, "grad_norm": 0.6517992088242254, "learning_rate": 9.980022877504703e-05, "loss": 3.5089, "step": 6071 }, { "epoch": 0.37693214972996464, "grad_norm": 0.577062661756693, "learning_rate": 9.979990612802905e-05, "loss": 3.4375, "step": 6072 }, { "epoch": 0.3769942268297225, "grad_norm": 0.5370530532437587, "learning_rate": 9.979958322119285e-05, "loss": 3.528, "step": 6073 }, { "epoch": 0.3770563039294804, "grad_norm": 0.558423449825723, "learning_rate": 9.979926005454014e-05, "loss": 3.4699, "step": 6074 }, { "epoch": 0.37711838102923834, "grad_norm": 0.6223888928865707, "learning_rate": 9.979893662807261e-05, "loss": 3.5325, "step": 6075 }, { "epoch": 0.3771804581289962, "grad_norm": 0.6637018907572656, "learning_rate": 9.979861294179194e-05, "loss": 3.5471, "step": 6076 }, { "epoch": 0.3772425352287541, "grad_norm": 0.5970149620088613, "learning_rate": 9.97982889956998e-05, "loss": 3.5351, "step": 6077 }, { "epoch": 0.37730461232851203, "grad_norm": 0.5881538942234221, "learning_rate": 9.979796478979793e-05, "loss": 3.5281, "step": 6078 }, { "epoch": 0.3773666894282699, "grad_norm": 0.531707538595274, "learning_rate": 9.979764032408796e-05, "loss": 3.4564, "step": 6079 }, { "epoch": 0.3774287665280278, "grad_norm": 0.5735155847588658, "learning_rate": 9.979731559857162e-05, "loss": 3.5601, "step": 6080 }, { "epoch": 0.37749084362778573, "grad_norm": 0.5670077320140186, "learning_rate": 9.97969906132506e-05, "loss": 3.5843, "step": 6081 }, { "epoch": 0.3775529207275436, "grad_norm": 0.5886912929316738, "learning_rate": 9.979666536812658e-05, "loss": 3.6144, "step": 6082 }, { "epoch": 0.3776149978273015, "grad_norm": 0.5462949946834602, "learning_rate": 9.979633986320127e-05, "loss": 3.5165, "step": 6083 }, { "epoch": 0.37767707492705943, "grad_norm": 0.5815444036637419, "learning_rate": 9.979601409847638e-05, "loss": 3.3922, "step": 6084 }, { "epoch": 0.3777391520268173, "grad_norm": 0.5767464052409903, "learning_rate": 9.979568807395358e-05, "loss": 3.5068, "step": 6085 }, { "epoch": 0.3778012291265752, "grad_norm": 0.5840797474020357, "learning_rate": 9.97953617896346e-05, "loss": 3.5074, "step": 6086 }, { "epoch": 0.37786330622633313, "grad_norm": 0.6229859564485737, "learning_rate": 9.979503524552112e-05, "loss": 3.557, "step": 6087 }, { "epoch": 0.377925383326091, "grad_norm": 0.6069120862100856, "learning_rate": 9.979470844161486e-05, "loss": 3.5076, "step": 6088 }, { "epoch": 0.3779874604258489, "grad_norm": 0.5685998728968357, "learning_rate": 9.979438137791751e-05, "loss": 3.4811, "step": 6089 }, { "epoch": 0.3780495375256068, "grad_norm": 0.557251524742166, "learning_rate": 9.979405405443079e-05, "loss": 3.4476, "step": 6090 }, { "epoch": 0.3781116146253647, "grad_norm": 0.5087950956898472, "learning_rate": 9.97937264711564e-05, "loss": 3.4694, "step": 6091 }, { "epoch": 0.3781736917251226, "grad_norm": 0.5574558913311441, "learning_rate": 9.979339862809605e-05, "loss": 3.5494, "step": 6092 }, { "epoch": 0.3782357688248805, "grad_norm": 0.48736837884397693, "learning_rate": 9.979307052525146e-05, "loss": 3.4047, "step": 6093 }, { "epoch": 0.3782978459246384, "grad_norm": 0.5233229896207567, "learning_rate": 9.979274216262433e-05, "loss": 3.5149, "step": 6094 }, { "epoch": 0.3783599230243963, "grad_norm": 0.5933142368006323, "learning_rate": 9.979241354021636e-05, "loss": 3.5728, "step": 6095 }, { "epoch": 0.3784220001241542, "grad_norm": 0.5636302930016966, "learning_rate": 9.979208465802931e-05, "loss": 3.4491, "step": 6096 }, { "epoch": 0.3784840772239121, "grad_norm": 0.5645717509396622, "learning_rate": 9.979175551606484e-05, "loss": 3.4474, "step": 6097 }, { "epoch": 0.37854615432367, "grad_norm": 0.5501781706057476, "learning_rate": 9.97914261143247e-05, "loss": 3.3851, "step": 6098 }, { "epoch": 0.3786082314234279, "grad_norm": 0.6226165864150448, "learning_rate": 9.979109645281062e-05, "loss": 3.5748, "step": 6099 }, { "epoch": 0.3786703085231858, "grad_norm": 0.5888014419824368, "learning_rate": 9.979076653152428e-05, "loss": 3.4961, "step": 6100 }, { "epoch": 0.3787323856229437, "grad_norm": 0.58257884949021, "learning_rate": 9.979043635046744e-05, "loss": 3.4749, "step": 6101 }, { "epoch": 0.3787944627227016, "grad_norm": 0.6220318170011648, "learning_rate": 9.97901059096418e-05, "loss": 3.5122, "step": 6102 }, { "epoch": 0.3788565398224595, "grad_norm": 0.6249359813327995, "learning_rate": 9.97897752090491e-05, "loss": 3.5259, "step": 6103 }, { "epoch": 0.3789186169222174, "grad_norm": 0.6564469950052008, "learning_rate": 9.978944424869105e-05, "loss": 3.538, "step": 6104 }, { "epoch": 0.3789806940219753, "grad_norm": 0.5230732689623045, "learning_rate": 9.978911302856938e-05, "loss": 3.4708, "step": 6105 }, { "epoch": 0.3790427711217332, "grad_norm": 0.5224496715436193, "learning_rate": 9.978878154868582e-05, "loss": 3.4309, "step": 6106 }, { "epoch": 0.3791048482214911, "grad_norm": 0.5649929698996222, "learning_rate": 9.978844980904209e-05, "loss": 3.4909, "step": 6107 }, { "epoch": 0.379166925321249, "grad_norm": 0.624516848736199, "learning_rate": 9.978811780963995e-05, "loss": 3.5474, "step": 6108 }, { "epoch": 0.3792290024210069, "grad_norm": 0.5558357174758832, "learning_rate": 9.978778555048111e-05, "loss": 3.4793, "step": 6109 }, { "epoch": 0.3792910795207648, "grad_norm": 0.5234178283183332, "learning_rate": 9.97874530315673e-05, "loss": 3.4663, "step": 6110 }, { "epoch": 0.3793531566205227, "grad_norm": 0.5204339753360625, "learning_rate": 9.978712025290026e-05, "loss": 3.5022, "step": 6111 }, { "epoch": 0.3794152337202806, "grad_norm": 0.5554458377367298, "learning_rate": 9.978678721448172e-05, "loss": 3.4888, "step": 6112 }, { "epoch": 0.3794773108200385, "grad_norm": 0.5672530585618865, "learning_rate": 9.978645391631344e-05, "loss": 3.5879, "step": 6113 }, { "epoch": 0.3795393879197964, "grad_norm": 0.5760869367676154, "learning_rate": 9.978612035839714e-05, "loss": 3.4814, "step": 6114 }, { "epoch": 0.3796014650195543, "grad_norm": 0.5653436097820771, "learning_rate": 9.978578654073456e-05, "loss": 3.516, "step": 6115 }, { "epoch": 0.3796635421193122, "grad_norm": 0.4771156540240894, "learning_rate": 9.978545246332745e-05, "loss": 3.5468, "step": 6116 }, { "epoch": 0.3797256192190701, "grad_norm": 0.4739586721845261, "learning_rate": 9.978511812617755e-05, "loss": 3.5518, "step": 6117 }, { "epoch": 0.37978769631882797, "grad_norm": 0.4730374241584939, "learning_rate": 9.97847835292866e-05, "loss": 3.5921, "step": 6118 }, { "epoch": 0.3798497734185859, "grad_norm": 0.4910250666643017, "learning_rate": 9.978444867265635e-05, "loss": 3.4728, "step": 6119 }, { "epoch": 0.3799118505183438, "grad_norm": 0.5256450278223442, "learning_rate": 9.978411355628854e-05, "loss": 3.4786, "step": 6120 }, { "epoch": 0.37997392761810167, "grad_norm": 0.4952237016520932, "learning_rate": 9.978377818018491e-05, "loss": 3.556, "step": 6121 }, { "epoch": 0.3800360047178596, "grad_norm": 0.5149759774021799, "learning_rate": 9.978344254434727e-05, "loss": 3.5591, "step": 6122 }, { "epoch": 0.3800980818176175, "grad_norm": 0.5678580521244676, "learning_rate": 9.978310664877728e-05, "loss": 3.564, "step": 6123 }, { "epoch": 0.38016015891737537, "grad_norm": 0.6343331374839458, "learning_rate": 9.978277049347675e-05, "loss": 3.5929, "step": 6124 }, { "epoch": 0.3802222360171333, "grad_norm": 0.6554181639801143, "learning_rate": 9.978243407844742e-05, "loss": 3.538, "step": 6125 }, { "epoch": 0.3802843131168912, "grad_norm": 0.6005917680658927, "learning_rate": 9.978209740369107e-05, "loss": 3.4564, "step": 6126 }, { "epoch": 0.38034639021664907, "grad_norm": 0.5531125611452945, "learning_rate": 9.978176046920941e-05, "loss": 3.4599, "step": 6127 }, { "epoch": 0.380408467316407, "grad_norm": 0.6718604507645272, "learning_rate": 9.978142327500421e-05, "loss": 3.5413, "step": 6128 }, { "epoch": 0.3804705444161649, "grad_norm": 0.6787485455309492, "learning_rate": 9.978108582107727e-05, "loss": 3.4279, "step": 6129 }, { "epoch": 0.38053262151592276, "grad_norm": 0.5432430370409476, "learning_rate": 9.97807481074303e-05, "loss": 3.3267, "step": 6130 }, { "epoch": 0.3805946986156807, "grad_norm": 0.5742986714284795, "learning_rate": 9.97804101340651e-05, "loss": 3.5696, "step": 6131 }, { "epoch": 0.3806567757154386, "grad_norm": 0.7077977783154245, "learning_rate": 9.97800719009834e-05, "loss": 3.6192, "step": 6132 }, { "epoch": 0.38071885281519646, "grad_norm": 0.5868264385196819, "learning_rate": 9.977973340818699e-05, "loss": 3.4643, "step": 6133 }, { "epoch": 0.3807809299149544, "grad_norm": 0.5963480475721192, "learning_rate": 9.97793946556776e-05, "loss": 3.4523, "step": 6134 }, { "epoch": 0.3808430070147123, "grad_norm": 0.5933178784825744, "learning_rate": 9.977905564345705e-05, "loss": 3.458, "step": 6135 }, { "epoch": 0.38090508411447016, "grad_norm": 0.5409031003859998, "learning_rate": 9.977871637152709e-05, "loss": 3.422, "step": 6136 }, { "epoch": 0.3809671612142281, "grad_norm": 0.4980259311236657, "learning_rate": 9.977837683988946e-05, "loss": 3.517, "step": 6137 }, { "epoch": 0.381029238313986, "grad_norm": 0.5766412148849267, "learning_rate": 9.977803704854596e-05, "loss": 3.4989, "step": 6138 }, { "epoch": 0.38109131541374386, "grad_norm": 0.5034871687442675, "learning_rate": 9.977769699749836e-05, "loss": 3.4637, "step": 6139 }, { "epoch": 0.3811533925135018, "grad_norm": 0.5077139832228323, "learning_rate": 9.977735668674843e-05, "loss": 3.5302, "step": 6140 }, { "epoch": 0.3812154696132597, "grad_norm": 0.5276629859193895, "learning_rate": 9.977701611629796e-05, "loss": 3.3827, "step": 6141 }, { "epoch": 0.38127754671301756, "grad_norm": 0.49846145727765656, "learning_rate": 9.97766752861487e-05, "loss": 3.4269, "step": 6142 }, { "epoch": 0.3813396238127755, "grad_norm": 0.5504972388469388, "learning_rate": 9.977633419630243e-05, "loss": 3.5514, "step": 6143 }, { "epoch": 0.3814017009125334, "grad_norm": 0.5277242629905765, "learning_rate": 9.977599284676095e-05, "loss": 3.3896, "step": 6144 }, { "epoch": 0.38146377801229125, "grad_norm": 0.4982333712992253, "learning_rate": 9.977565123752604e-05, "loss": 3.3427, "step": 6145 }, { "epoch": 0.3815258551120492, "grad_norm": 0.564894190392018, "learning_rate": 9.977530936859946e-05, "loss": 3.453, "step": 6146 }, { "epoch": 0.3815879322118071, "grad_norm": 0.4888375127353355, "learning_rate": 9.977496723998302e-05, "loss": 3.4798, "step": 6147 }, { "epoch": 0.38165000931156495, "grad_norm": 0.5479905059385337, "learning_rate": 9.977462485167848e-05, "loss": 3.4498, "step": 6148 }, { "epoch": 0.38171208641132287, "grad_norm": 0.5373378475804567, "learning_rate": 9.977428220368765e-05, "loss": 3.5882, "step": 6149 }, { "epoch": 0.3817741635110808, "grad_norm": 0.5706178553155887, "learning_rate": 9.97739392960123e-05, "loss": 3.5224, "step": 6150 }, { "epoch": 0.38183624061083865, "grad_norm": 0.6239164974174544, "learning_rate": 9.977359612865423e-05, "loss": 3.5346, "step": 6151 }, { "epoch": 0.38189831771059657, "grad_norm": 0.5763407314308154, "learning_rate": 9.977325270161524e-05, "loss": 3.466, "step": 6152 }, { "epoch": 0.3819603948103545, "grad_norm": 0.5950236964115204, "learning_rate": 9.977290901489709e-05, "loss": 3.5123, "step": 6153 }, { "epoch": 0.38202247191011235, "grad_norm": 0.6033787248021124, "learning_rate": 9.977256506850159e-05, "loss": 3.5368, "step": 6154 }, { "epoch": 0.38208454900987027, "grad_norm": 0.6166620794546641, "learning_rate": 9.977222086243054e-05, "loss": 3.5549, "step": 6155 }, { "epoch": 0.3821466261096282, "grad_norm": 0.6240293356268968, "learning_rate": 9.977187639668572e-05, "loss": 3.494, "step": 6156 }, { "epoch": 0.38220870320938605, "grad_norm": 0.5614768896297945, "learning_rate": 9.977153167126896e-05, "loss": 3.4072, "step": 6157 }, { "epoch": 0.38227078030914396, "grad_norm": 0.5221293600465206, "learning_rate": 9.977118668618202e-05, "loss": 3.5411, "step": 6158 }, { "epoch": 0.3823328574089019, "grad_norm": 0.51843952976448, "learning_rate": 9.977084144142672e-05, "loss": 3.4392, "step": 6159 }, { "epoch": 0.38239493450865975, "grad_norm": 0.5349100757944999, "learning_rate": 9.977049593700487e-05, "loss": 3.4863, "step": 6160 }, { "epoch": 0.38245701160841766, "grad_norm": 0.573977361272513, "learning_rate": 9.977015017291824e-05, "loss": 3.502, "step": 6161 }, { "epoch": 0.3825190887081756, "grad_norm": 0.5214793744365378, "learning_rate": 9.976980414916867e-05, "loss": 3.4085, "step": 6162 }, { "epoch": 0.38258116580793344, "grad_norm": 0.48337570291097637, "learning_rate": 9.976945786575795e-05, "loss": 3.461, "step": 6163 }, { "epoch": 0.38264324290769136, "grad_norm": 0.5367669418443621, "learning_rate": 9.976911132268787e-05, "loss": 3.398, "step": 6164 }, { "epoch": 0.3827053200074493, "grad_norm": 0.6234125481299386, "learning_rate": 9.976876451996026e-05, "loss": 3.5297, "step": 6165 }, { "epoch": 0.38276739710720714, "grad_norm": 0.6553470439474305, "learning_rate": 9.976841745757694e-05, "loss": 3.4507, "step": 6166 }, { "epoch": 0.38282947420696506, "grad_norm": 0.6347363509250302, "learning_rate": 9.97680701355397e-05, "loss": 3.406, "step": 6167 }, { "epoch": 0.382891551306723, "grad_norm": 0.7007049238301168, "learning_rate": 9.976772255385034e-05, "loss": 3.5024, "step": 6168 }, { "epoch": 0.38295362840648084, "grad_norm": 0.6347978084915181, "learning_rate": 9.97673747125107e-05, "loss": 3.5329, "step": 6169 }, { "epoch": 0.38301570550623876, "grad_norm": 0.5628248971847747, "learning_rate": 9.976702661152258e-05, "loss": 3.4983, "step": 6170 }, { "epoch": 0.3830777826059967, "grad_norm": 0.5215177489613267, "learning_rate": 9.976667825088782e-05, "loss": 3.4432, "step": 6171 }, { "epoch": 0.38313985970575454, "grad_norm": 0.49326834001101777, "learning_rate": 9.976632963060819e-05, "loss": 3.6274, "step": 6172 }, { "epoch": 0.38320193680551246, "grad_norm": 0.5347366915915337, "learning_rate": 9.976598075068556e-05, "loss": 3.4567, "step": 6173 }, { "epoch": 0.3832640139052704, "grad_norm": 0.5043947675913047, "learning_rate": 9.976563161112171e-05, "loss": 3.511, "step": 6174 }, { "epoch": 0.38332609100502824, "grad_norm": 0.5044140290353581, "learning_rate": 9.97652822119185e-05, "loss": 3.4236, "step": 6175 }, { "epoch": 0.38338816810478615, "grad_norm": 0.5596307585392065, "learning_rate": 9.97649325530777e-05, "loss": 3.3603, "step": 6176 }, { "epoch": 0.38345024520454407, "grad_norm": 0.5830060233595087, "learning_rate": 9.97645826346012e-05, "loss": 3.4621, "step": 6177 }, { "epoch": 0.38351232230430193, "grad_norm": 0.5442822712208996, "learning_rate": 9.976423245649077e-05, "loss": 3.4846, "step": 6178 }, { "epoch": 0.38357439940405985, "grad_norm": 0.5237939389627909, "learning_rate": 9.976388201874827e-05, "loss": 3.5388, "step": 6179 }, { "epoch": 0.38363647650381777, "grad_norm": 0.6022641684756652, "learning_rate": 9.97635313213755e-05, "loss": 3.4615, "step": 6180 }, { "epoch": 0.38369855360357563, "grad_norm": 0.6069905261269792, "learning_rate": 9.976318036437433e-05, "loss": 3.4531, "step": 6181 }, { "epoch": 0.38376063070333355, "grad_norm": 0.5383806200586613, "learning_rate": 9.976282914774655e-05, "loss": 3.4303, "step": 6182 }, { "epoch": 0.3838227078030914, "grad_norm": 0.546047210893231, "learning_rate": 9.976247767149403e-05, "loss": 3.3662, "step": 6183 }, { "epoch": 0.38388478490284933, "grad_norm": 0.6004637555151159, "learning_rate": 9.976212593561855e-05, "loss": 3.4734, "step": 6184 }, { "epoch": 0.38394686200260725, "grad_norm": 0.6458691899795221, "learning_rate": 9.9761773940122e-05, "loss": 3.3505, "step": 6185 }, { "epoch": 0.3840089391023651, "grad_norm": 0.7379561016057705, "learning_rate": 9.976142168500619e-05, "loss": 3.5201, "step": 6186 }, { "epoch": 0.38407101620212303, "grad_norm": 0.7825204818531291, "learning_rate": 9.976106917027297e-05, "loss": 3.4916, "step": 6187 }, { "epoch": 0.38413309330188095, "grad_norm": 0.8105708842570796, "learning_rate": 9.976071639592417e-05, "loss": 3.3924, "step": 6188 }, { "epoch": 0.3841951704016388, "grad_norm": 0.6764152727871313, "learning_rate": 9.976036336196163e-05, "loss": 3.3926, "step": 6189 }, { "epoch": 0.3842572475013967, "grad_norm": 0.6897180064635793, "learning_rate": 9.97600100683872e-05, "loss": 3.5195, "step": 6190 }, { "epoch": 0.38431932460115464, "grad_norm": 0.7600180003268148, "learning_rate": 9.975965651520272e-05, "loss": 3.4153, "step": 6191 }, { "epoch": 0.3843814017009125, "grad_norm": 0.5946251793749452, "learning_rate": 9.975930270241003e-05, "loss": 3.4811, "step": 6192 }, { "epoch": 0.3844434788006704, "grad_norm": 0.6153118024272849, "learning_rate": 9.975894863001096e-05, "loss": 3.461, "step": 6193 }, { "epoch": 0.38450555590042834, "grad_norm": 0.5932680985221526, "learning_rate": 9.97585942980074e-05, "loss": 3.5072, "step": 6194 }, { "epoch": 0.3845676330001862, "grad_norm": 0.6215434096115166, "learning_rate": 9.975823970640117e-05, "loss": 3.4399, "step": 6195 }, { "epoch": 0.3846297100999441, "grad_norm": 0.5923169309139007, "learning_rate": 9.975788485519411e-05, "loss": 3.4981, "step": 6196 }, { "epoch": 0.38469178719970204, "grad_norm": 0.5621274668185801, "learning_rate": 9.97575297443881e-05, "loss": 3.5742, "step": 6197 }, { "epoch": 0.3847538642994599, "grad_norm": 0.6319721906344139, "learning_rate": 9.975717437398498e-05, "loss": 3.5858, "step": 6198 }, { "epoch": 0.3848159413992178, "grad_norm": 0.6828229427630262, "learning_rate": 9.975681874398659e-05, "loss": 3.5501, "step": 6199 }, { "epoch": 0.38487801849897574, "grad_norm": 0.5589357142990953, "learning_rate": 9.97564628543948e-05, "loss": 3.4077, "step": 6200 }, { "epoch": 0.3849400955987336, "grad_norm": 0.5175722229819005, "learning_rate": 9.975610670521148e-05, "loss": 3.4256, "step": 6201 }, { "epoch": 0.3850021726984915, "grad_norm": 0.5341666322887646, "learning_rate": 9.975575029643845e-05, "loss": 3.396, "step": 6202 }, { "epoch": 0.38506424979824944, "grad_norm": 0.5465902677170678, "learning_rate": 9.97553936280776e-05, "loss": 3.5491, "step": 6203 }, { "epoch": 0.3851263268980073, "grad_norm": 0.463217768573577, "learning_rate": 9.975503670013078e-05, "loss": 3.5001, "step": 6204 }, { "epoch": 0.3851884039977652, "grad_norm": 0.521554292142269, "learning_rate": 9.975467951259987e-05, "loss": 3.4761, "step": 6205 }, { "epoch": 0.38525048109752313, "grad_norm": 0.5121976053336104, "learning_rate": 9.97543220654867e-05, "loss": 3.4389, "step": 6206 }, { "epoch": 0.385312558197281, "grad_norm": 0.5783033353664172, "learning_rate": 9.975396435879316e-05, "loss": 3.4134, "step": 6207 }, { "epoch": 0.3853746352970389, "grad_norm": 0.5964631532142776, "learning_rate": 9.975360639252111e-05, "loss": 3.4623, "step": 6208 }, { "epoch": 0.38543671239679683, "grad_norm": 0.535611147261861, "learning_rate": 9.975324816667242e-05, "loss": 3.4534, "step": 6209 }, { "epoch": 0.3854987894965547, "grad_norm": 0.5732267985317256, "learning_rate": 9.975288968124896e-05, "loss": 3.4967, "step": 6210 }, { "epoch": 0.3855608665963126, "grad_norm": 0.5796301296666099, "learning_rate": 9.975253093625258e-05, "loss": 3.4786, "step": 6211 }, { "epoch": 0.38562294369607053, "grad_norm": 0.5656684542661231, "learning_rate": 9.975217193168519e-05, "loss": 3.4677, "step": 6212 }, { "epoch": 0.3856850207958284, "grad_norm": 0.6032564662297489, "learning_rate": 9.975181266754863e-05, "loss": 3.4758, "step": 6213 }, { "epoch": 0.3857470978955863, "grad_norm": 0.612816579903381, "learning_rate": 9.97514531438448e-05, "loss": 3.4692, "step": 6214 }, { "epoch": 0.38580917499534423, "grad_norm": 0.5794558028629468, "learning_rate": 9.975109336057554e-05, "loss": 3.4215, "step": 6215 }, { "epoch": 0.3858712520951021, "grad_norm": 0.6046339863994015, "learning_rate": 9.975073331774276e-05, "loss": 3.4187, "step": 6216 }, { "epoch": 0.38593332919486, "grad_norm": 0.6512228878842949, "learning_rate": 9.975037301534832e-05, "loss": 3.3857, "step": 6217 }, { "epoch": 0.3859954062946179, "grad_norm": 0.6811478153201285, "learning_rate": 9.975001245339411e-05, "loss": 3.5584, "step": 6218 }, { "epoch": 0.3860574833943758, "grad_norm": 0.5973545883915691, "learning_rate": 9.974965163188201e-05, "loss": 3.4903, "step": 6219 }, { "epoch": 0.3861195604941337, "grad_norm": 0.5830541237691418, "learning_rate": 9.97492905508139e-05, "loss": 3.4404, "step": 6220 }, { "epoch": 0.3861816375938916, "grad_norm": 0.6383510332002523, "learning_rate": 9.974892921019166e-05, "loss": 3.4295, "step": 6221 }, { "epoch": 0.3862437146936495, "grad_norm": 0.5573375154215511, "learning_rate": 9.974856761001718e-05, "loss": 3.4788, "step": 6222 }, { "epoch": 0.3863057917934074, "grad_norm": 0.5305405816727417, "learning_rate": 9.974820575029235e-05, "loss": 3.4397, "step": 6223 }, { "epoch": 0.3863678688931653, "grad_norm": 0.5524784209926543, "learning_rate": 9.974784363101905e-05, "loss": 3.4532, "step": 6224 }, { "epoch": 0.3864299459929232, "grad_norm": 0.5269213927762546, "learning_rate": 9.974748125219918e-05, "loss": 3.4773, "step": 6225 }, { "epoch": 0.3864920230926811, "grad_norm": 0.5683826297577614, "learning_rate": 9.974711861383461e-05, "loss": 3.4403, "step": 6226 }, { "epoch": 0.386554100192439, "grad_norm": 0.5089664552685703, "learning_rate": 9.974675571592726e-05, "loss": 3.4811, "step": 6227 }, { "epoch": 0.3866161772921969, "grad_norm": 0.5106814988441609, "learning_rate": 9.9746392558479e-05, "loss": 3.4283, "step": 6228 }, { "epoch": 0.3866782543919548, "grad_norm": 0.5594930012872871, "learning_rate": 9.974602914149174e-05, "loss": 3.3914, "step": 6229 }, { "epoch": 0.3867403314917127, "grad_norm": 0.6427668378427974, "learning_rate": 9.974566546496737e-05, "loss": 3.4922, "step": 6230 }, { "epoch": 0.3868024085914706, "grad_norm": 0.748279385997516, "learning_rate": 9.974530152890777e-05, "loss": 3.4841, "step": 6231 }, { "epoch": 0.3868644856912285, "grad_norm": 0.7506915321689624, "learning_rate": 9.974493733331489e-05, "loss": 3.5858, "step": 6232 }, { "epoch": 0.3869265627909864, "grad_norm": 0.6501008267330352, "learning_rate": 9.974457287819056e-05, "loss": 3.4777, "step": 6233 }, { "epoch": 0.3869886398907443, "grad_norm": 0.6007760840145483, "learning_rate": 9.974420816353674e-05, "loss": 3.5722, "step": 6234 }, { "epoch": 0.3870507169905022, "grad_norm": 0.6608663272638539, "learning_rate": 9.97438431893553e-05, "loss": 3.4698, "step": 6235 }, { "epoch": 0.3871127940902601, "grad_norm": 0.7330759205369854, "learning_rate": 9.974347795564816e-05, "loss": 3.4479, "step": 6236 }, { "epoch": 0.387174871190018, "grad_norm": 0.7168279739826409, "learning_rate": 9.974311246241721e-05, "loss": 3.5401, "step": 6237 }, { "epoch": 0.3872369482897759, "grad_norm": 0.6508677568217629, "learning_rate": 9.974274670966436e-05, "loss": 3.478, "step": 6238 }, { "epoch": 0.3872990253895338, "grad_norm": 0.6333143104866913, "learning_rate": 9.974238069739156e-05, "loss": 3.4855, "step": 6239 }, { "epoch": 0.3873611024892917, "grad_norm": 0.6731207410204789, "learning_rate": 9.974201442560066e-05, "loss": 3.4615, "step": 6240 }, { "epoch": 0.3874231795890496, "grad_norm": 0.6351907333641555, "learning_rate": 9.97416478942936e-05, "loss": 3.4426, "step": 6241 }, { "epoch": 0.3874852566888075, "grad_norm": 0.7148357453445271, "learning_rate": 9.974128110347228e-05, "loss": 3.3723, "step": 6242 }, { "epoch": 0.3875473337885654, "grad_norm": 0.5690023973075902, "learning_rate": 9.974091405313863e-05, "loss": 3.4383, "step": 6243 }, { "epoch": 0.3876094108883233, "grad_norm": 0.5289351856875665, "learning_rate": 9.974054674329456e-05, "loss": 3.5034, "step": 6244 }, { "epoch": 0.3876714879880812, "grad_norm": 0.6195631059920519, "learning_rate": 9.974017917394198e-05, "loss": 3.4487, "step": 6245 }, { "epoch": 0.38773356508783907, "grad_norm": 0.5341311751798106, "learning_rate": 9.97398113450828e-05, "loss": 3.5213, "step": 6246 }, { "epoch": 0.387795642187597, "grad_norm": 0.5895721871453588, "learning_rate": 9.973944325671896e-05, "loss": 3.3401, "step": 6247 }, { "epoch": 0.3878577192873549, "grad_norm": 0.551462820409782, "learning_rate": 9.973907490885237e-05, "loss": 3.4675, "step": 6248 }, { "epoch": 0.38791979638711277, "grad_norm": 0.5949246933107002, "learning_rate": 9.973870630148494e-05, "loss": 3.6005, "step": 6249 }, { "epoch": 0.3879818734868707, "grad_norm": 0.6696474497501379, "learning_rate": 9.973833743461862e-05, "loss": 3.5562, "step": 6250 }, { "epoch": 0.3880439505866286, "grad_norm": 0.6727347307134487, "learning_rate": 9.97379683082553e-05, "loss": 3.4804, "step": 6251 }, { "epoch": 0.38810602768638647, "grad_norm": 0.6429951251009804, "learning_rate": 9.973759892239694e-05, "loss": 3.4162, "step": 6252 }, { "epoch": 0.3881681047861444, "grad_norm": 0.6071441849726094, "learning_rate": 9.973722927704545e-05, "loss": 3.4109, "step": 6253 }, { "epoch": 0.3882301818859023, "grad_norm": 0.5387300212368532, "learning_rate": 9.973685937220276e-05, "loss": 3.5027, "step": 6254 }, { "epoch": 0.38829225898566017, "grad_norm": 0.5400255970716145, "learning_rate": 9.97364892078708e-05, "loss": 3.417, "step": 6255 }, { "epoch": 0.3883543360854181, "grad_norm": 0.46163306046242164, "learning_rate": 9.973611878405152e-05, "loss": 3.4544, "step": 6256 }, { "epoch": 0.388416413185176, "grad_norm": 0.49388891149231656, "learning_rate": 9.97357481007468e-05, "loss": 3.4735, "step": 6257 }, { "epoch": 0.38847849028493386, "grad_norm": 0.5199298333548443, "learning_rate": 9.973537715795863e-05, "loss": 3.4441, "step": 6258 }, { "epoch": 0.3885405673846918, "grad_norm": 0.5082354618137941, "learning_rate": 9.973500595568892e-05, "loss": 3.5195, "step": 6259 }, { "epoch": 0.3886026444844497, "grad_norm": 0.463217800742379, "learning_rate": 9.973463449393961e-05, "loss": 3.4713, "step": 6260 }, { "epoch": 0.38866472158420756, "grad_norm": 0.5604177718405716, "learning_rate": 9.973426277271264e-05, "loss": 3.4473, "step": 6261 }, { "epoch": 0.3887267986839655, "grad_norm": 0.5672659827545009, "learning_rate": 9.973389079200994e-05, "loss": 3.5043, "step": 6262 }, { "epoch": 0.3887888757837234, "grad_norm": 0.5063159016701054, "learning_rate": 9.973351855183346e-05, "loss": 3.4325, "step": 6263 }, { "epoch": 0.38885095288348126, "grad_norm": 0.5667749717471106, "learning_rate": 9.973314605218516e-05, "loss": 3.4963, "step": 6264 }, { "epoch": 0.3889130299832392, "grad_norm": 0.5849584920128348, "learning_rate": 9.973277329306694e-05, "loss": 3.4453, "step": 6265 }, { "epoch": 0.3889751070829971, "grad_norm": 0.5659297122345768, "learning_rate": 9.973240027448077e-05, "loss": 3.4783, "step": 6266 }, { "epoch": 0.38903718418275496, "grad_norm": 0.6206018911479173, "learning_rate": 9.973202699642861e-05, "loss": 3.5624, "step": 6267 }, { "epoch": 0.3890992612825129, "grad_norm": 0.6081322444315683, "learning_rate": 9.973165345891237e-05, "loss": 3.4851, "step": 6268 }, { "epoch": 0.3891613383822708, "grad_norm": 0.558041386166767, "learning_rate": 9.973127966193406e-05, "loss": 3.417, "step": 6269 }, { "epoch": 0.38922341548202866, "grad_norm": 0.6318974413409909, "learning_rate": 9.973090560549555e-05, "loss": 3.4594, "step": 6270 }, { "epoch": 0.3892854925817866, "grad_norm": 0.5946922856116766, "learning_rate": 9.973053128959884e-05, "loss": 3.368, "step": 6271 }, { "epoch": 0.3893475696815445, "grad_norm": 0.5999330274635954, "learning_rate": 9.97301567142459e-05, "loss": 3.4488, "step": 6272 }, { "epoch": 0.38940964678130235, "grad_norm": 0.5999566976498142, "learning_rate": 9.972978187943864e-05, "loss": 3.4628, "step": 6273 }, { "epoch": 0.38947172388106027, "grad_norm": 0.5660739055190835, "learning_rate": 9.972940678517903e-05, "loss": 3.4384, "step": 6274 }, { "epoch": 0.3895338009808182, "grad_norm": 0.4935347721744936, "learning_rate": 9.972903143146906e-05, "loss": 3.422, "step": 6275 }, { "epoch": 0.38959587808057605, "grad_norm": 0.5750109598939696, "learning_rate": 9.972865581831063e-05, "loss": 3.4256, "step": 6276 }, { "epoch": 0.38965795518033397, "grad_norm": 0.5954285786315647, "learning_rate": 9.972827994570575e-05, "loss": 3.5134, "step": 6277 }, { "epoch": 0.3897200322800919, "grad_norm": 0.5655585423833889, "learning_rate": 9.972790381365635e-05, "loss": 3.4676, "step": 6278 }, { "epoch": 0.38978210937984975, "grad_norm": 0.5910890078574976, "learning_rate": 9.972752742216443e-05, "loss": 3.4967, "step": 6279 }, { "epoch": 0.38984418647960767, "grad_norm": 0.5380671613847802, "learning_rate": 9.972715077123189e-05, "loss": 3.5014, "step": 6280 }, { "epoch": 0.3899062635793656, "grad_norm": 0.5415037961775699, "learning_rate": 9.972677386086075e-05, "loss": 3.4255, "step": 6281 }, { "epoch": 0.38996834067912345, "grad_norm": 0.5164344964547724, "learning_rate": 9.972639669105297e-05, "loss": 3.4432, "step": 6282 }, { "epoch": 0.39003041777888137, "grad_norm": 0.6143498435162363, "learning_rate": 9.97260192618105e-05, "loss": 3.4683, "step": 6283 }, { "epoch": 0.3900924948786393, "grad_norm": 0.5701516852244464, "learning_rate": 9.972564157313531e-05, "loss": 3.4552, "step": 6284 }, { "epoch": 0.39015457197839715, "grad_norm": 0.5901261468369956, "learning_rate": 9.972526362502938e-05, "loss": 3.4987, "step": 6285 }, { "epoch": 0.39021664907815506, "grad_norm": 0.5287110863112208, "learning_rate": 9.972488541749468e-05, "loss": 3.4408, "step": 6286 }, { "epoch": 0.390278726177913, "grad_norm": 0.5171777580242641, "learning_rate": 9.972450695053317e-05, "loss": 3.5087, "step": 6287 }, { "epoch": 0.39034080327767084, "grad_norm": 0.5726500642242036, "learning_rate": 9.972412822414685e-05, "loss": 3.5414, "step": 6288 }, { "epoch": 0.39040288037742876, "grad_norm": 0.5070004940256274, "learning_rate": 9.972374923833768e-05, "loss": 3.4418, "step": 6289 }, { "epoch": 0.3904649574771867, "grad_norm": 0.5694645626757994, "learning_rate": 9.972336999310762e-05, "loss": 3.4687, "step": 6290 }, { "epoch": 0.39052703457694454, "grad_norm": 0.5447883579112341, "learning_rate": 9.97229904884587e-05, "loss": 3.4853, "step": 6291 }, { "epoch": 0.39058911167670246, "grad_norm": 0.6032446589385627, "learning_rate": 9.972261072439286e-05, "loss": 3.5048, "step": 6292 }, { "epoch": 0.3906511887764604, "grad_norm": 0.6501071988253935, "learning_rate": 9.972223070091207e-05, "loss": 3.5613, "step": 6293 }, { "epoch": 0.39071326587621824, "grad_norm": 0.5584427155868592, "learning_rate": 9.972185041801836e-05, "loss": 3.3863, "step": 6294 }, { "epoch": 0.39077534297597616, "grad_norm": 0.5558010261026515, "learning_rate": 9.972146987571367e-05, "loss": 3.497, "step": 6295 }, { "epoch": 0.3908374200757341, "grad_norm": 0.5483100319127082, "learning_rate": 9.9721089074e-05, "loss": 3.5505, "step": 6296 }, { "epoch": 0.39089949717549194, "grad_norm": 0.6707100194133445, "learning_rate": 9.972070801287934e-05, "loss": 3.4601, "step": 6297 }, { "epoch": 0.39096157427524986, "grad_norm": 0.5629503778302044, "learning_rate": 9.972032669235367e-05, "loss": 3.4677, "step": 6298 }, { "epoch": 0.3910236513750078, "grad_norm": 0.5082913122110481, "learning_rate": 9.971994511242499e-05, "loss": 3.4446, "step": 6299 }, { "epoch": 0.39108572847476564, "grad_norm": 0.5893599439702318, "learning_rate": 9.971956327309528e-05, "loss": 3.4535, "step": 6300 }, { "epoch": 0.39114780557452355, "grad_norm": 0.5727840067701231, "learning_rate": 9.971918117436656e-05, "loss": 3.5177, "step": 6301 }, { "epoch": 0.3912098826742815, "grad_norm": 0.5185005748841236, "learning_rate": 9.971879881624077e-05, "loss": 3.4071, "step": 6302 }, { "epoch": 0.39127195977403934, "grad_norm": 0.5696773650355444, "learning_rate": 9.971841619871997e-05, "loss": 3.4487, "step": 6303 }, { "epoch": 0.39133403687379725, "grad_norm": 0.5658581943228408, "learning_rate": 9.971803332180611e-05, "loss": 3.4357, "step": 6304 }, { "epoch": 0.39139611397355517, "grad_norm": 0.5942544300201487, "learning_rate": 9.97176501855012e-05, "loss": 3.4257, "step": 6305 }, { "epoch": 0.39145819107331303, "grad_norm": 0.5837153705070428, "learning_rate": 9.971726678980724e-05, "loss": 3.4402, "step": 6306 }, { "epoch": 0.39152026817307095, "grad_norm": 0.5977770116136187, "learning_rate": 9.971688313472623e-05, "loss": 3.4837, "step": 6307 }, { "epoch": 0.39158234527282887, "grad_norm": 0.5824047912507339, "learning_rate": 9.971649922026016e-05, "loss": 3.3817, "step": 6308 }, { "epoch": 0.39164442237258673, "grad_norm": 0.5475589018103596, "learning_rate": 9.971611504641106e-05, "loss": 3.3907, "step": 6309 }, { "epoch": 0.39170649947234465, "grad_norm": 0.47972368607110266, "learning_rate": 9.97157306131809e-05, "loss": 3.4748, "step": 6310 }, { "epoch": 0.39176857657210257, "grad_norm": 0.48969325563107113, "learning_rate": 9.971534592057172e-05, "loss": 3.4593, "step": 6311 }, { "epoch": 0.39183065367186043, "grad_norm": 0.5623111142807851, "learning_rate": 9.971496096858552e-05, "loss": 3.4734, "step": 6312 }, { "epoch": 0.39189273077161835, "grad_norm": 0.5786725363992192, "learning_rate": 9.971457575722429e-05, "loss": 3.3787, "step": 6313 }, { "epoch": 0.39195480787137627, "grad_norm": 0.5662300098755891, "learning_rate": 9.971419028649004e-05, "loss": 3.5313, "step": 6314 }, { "epoch": 0.3920168849711341, "grad_norm": 0.5369507711495743, "learning_rate": 9.971380455638481e-05, "loss": 3.3452, "step": 6315 }, { "epoch": 0.39207896207089205, "grad_norm": 0.5540052166806785, "learning_rate": 9.971341856691058e-05, "loss": 3.5757, "step": 6316 }, { "epoch": 0.39214103917064996, "grad_norm": 0.7242464457363129, "learning_rate": 9.971303231806939e-05, "loss": 3.5132, "step": 6317 }, { "epoch": 0.3922031162704078, "grad_norm": 0.7586066027887229, "learning_rate": 9.971264580986323e-05, "loss": 3.4596, "step": 6318 }, { "epoch": 0.39226519337016574, "grad_norm": 0.7708142424272759, "learning_rate": 9.971225904229414e-05, "loss": 3.5336, "step": 6319 }, { "epoch": 0.39232727046992366, "grad_norm": 0.6470964651582768, "learning_rate": 9.97118720153641e-05, "loss": 3.4246, "step": 6320 }, { "epoch": 0.3923893475696815, "grad_norm": 0.5965457647343202, "learning_rate": 9.971148472907516e-05, "loss": 3.4827, "step": 6321 }, { "epoch": 0.39245142466943944, "grad_norm": 0.6084503592536562, "learning_rate": 9.971109718342938e-05, "loss": 3.4423, "step": 6322 }, { "epoch": 0.39251350176919736, "grad_norm": 0.5786690600586689, "learning_rate": 9.971070937842868e-05, "loss": 3.4229, "step": 6323 }, { "epoch": 0.3925755788689552, "grad_norm": 0.5513663194293473, "learning_rate": 9.971032131407518e-05, "loss": 3.4358, "step": 6324 }, { "epoch": 0.39263765596871314, "grad_norm": 0.6932059534677896, "learning_rate": 9.970993299037086e-05, "loss": 3.5976, "step": 6325 }, { "epoch": 0.39269973306847106, "grad_norm": 0.6710768327812261, "learning_rate": 9.970954440731775e-05, "loss": 3.4463, "step": 6326 }, { "epoch": 0.3927618101682289, "grad_norm": 0.5763121870499454, "learning_rate": 9.970915556491787e-05, "loss": 3.4519, "step": 6327 }, { "epoch": 0.39282388726798684, "grad_norm": 0.5436512342180949, "learning_rate": 9.970876646317326e-05, "loss": 3.429, "step": 6328 }, { "epoch": 0.39288596436774476, "grad_norm": 0.6102237659177281, "learning_rate": 9.970837710208596e-05, "loss": 3.399, "step": 6329 }, { "epoch": 0.3929480414675026, "grad_norm": 0.5595923081580374, "learning_rate": 9.970798748165797e-05, "loss": 3.4321, "step": 6330 }, { "epoch": 0.39301011856726054, "grad_norm": 0.5044987174869413, "learning_rate": 9.970759760189137e-05, "loss": 3.4275, "step": 6331 }, { "epoch": 0.39307219566701845, "grad_norm": 0.5779198462965593, "learning_rate": 9.970720746278815e-05, "loss": 3.4426, "step": 6332 }, { "epoch": 0.3931342727667763, "grad_norm": 0.5471063124698836, "learning_rate": 9.970681706435035e-05, "loss": 3.4795, "step": 6333 }, { "epoch": 0.39319634986653423, "grad_norm": 0.593986388883798, "learning_rate": 9.970642640658003e-05, "loss": 3.4548, "step": 6334 }, { "epoch": 0.39325842696629215, "grad_norm": 0.5491241796909859, "learning_rate": 9.970603548947923e-05, "loss": 3.4974, "step": 6335 }, { "epoch": 0.39332050406605, "grad_norm": 0.5749382390149087, "learning_rate": 9.970564431304996e-05, "loss": 3.4172, "step": 6336 }, { "epoch": 0.39338258116580793, "grad_norm": 0.5058762536491694, "learning_rate": 9.970525287729428e-05, "loss": 3.5147, "step": 6337 }, { "epoch": 0.39344465826556585, "grad_norm": 0.5463184249237186, "learning_rate": 9.970486118221422e-05, "loss": 3.2756, "step": 6338 }, { "epoch": 0.3935067353653237, "grad_norm": 0.568109621316142, "learning_rate": 9.970446922781186e-05, "loss": 3.3859, "step": 6339 }, { "epoch": 0.39356881246508163, "grad_norm": 0.6036662435386195, "learning_rate": 9.970407701408919e-05, "loss": 3.4666, "step": 6340 }, { "epoch": 0.39363088956483955, "grad_norm": 0.6253741336628176, "learning_rate": 9.970368454104831e-05, "loss": 3.4797, "step": 6341 }, { "epoch": 0.3936929666645974, "grad_norm": 0.6147348949815202, "learning_rate": 9.970329180869123e-05, "loss": 3.3544, "step": 6342 }, { "epoch": 0.39375504376435533, "grad_norm": 0.6827718534963559, "learning_rate": 9.970289881702e-05, "loss": 3.3956, "step": 6343 }, { "epoch": 0.39381712086411325, "grad_norm": 0.7086194573126926, "learning_rate": 9.970250556603668e-05, "loss": 3.3836, "step": 6344 }, { "epoch": 0.3938791979638711, "grad_norm": 0.7353719883243842, "learning_rate": 9.970211205574333e-05, "loss": 3.4194, "step": 6345 }, { "epoch": 0.393941275063629, "grad_norm": 0.7263303816625811, "learning_rate": 9.9701718286142e-05, "loss": 3.4606, "step": 6346 }, { "epoch": 0.39400335216338694, "grad_norm": 0.5686332593306688, "learning_rate": 9.970132425723475e-05, "loss": 3.315, "step": 6347 }, { "epoch": 0.3940654292631448, "grad_norm": 0.5974277233956437, "learning_rate": 9.970092996902361e-05, "loss": 3.4807, "step": 6348 }, { "epoch": 0.3941275063629027, "grad_norm": 0.7107987320838745, "learning_rate": 9.970053542151066e-05, "loss": 3.5388, "step": 6349 }, { "epoch": 0.39418958346266064, "grad_norm": 0.7804352517332747, "learning_rate": 9.970014061469794e-05, "loss": 3.5548, "step": 6350 }, { "epoch": 0.3942516605624185, "grad_norm": 0.7511053681749701, "learning_rate": 9.969974554858754e-05, "loss": 3.4076, "step": 6351 }, { "epoch": 0.3943137376621764, "grad_norm": 0.6908429324108623, "learning_rate": 9.969935022318149e-05, "loss": 3.4836, "step": 6352 }, { "epoch": 0.39437581476193434, "grad_norm": 0.6886447565773035, "learning_rate": 9.969895463848186e-05, "loss": 3.4618, "step": 6353 }, { "epoch": 0.3944378918616922, "grad_norm": 0.7015338484047187, "learning_rate": 9.969855879449072e-05, "loss": 3.444, "step": 6354 }, { "epoch": 0.3944999689614501, "grad_norm": 0.6701617242505475, "learning_rate": 9.969816269121015e-05, "loss": 3.4111, "step": 6355 }, { "epoch": 0.39456204606120804, "grad_norm": 0.6141995158224952, "learning_rate": 9.969776632864219e-05, "loss": 3.5152, "step": 6356 }, { "epoch": 0.3946241231609659, "grad_norm": 0.5487794413304624, "learning_rate": 9.969736970678892e-05, "loss": 3.3974, "step": 6357 }, { "epoch": 0.3946862002607238, "grad_norm": 0.6175314270933523, "learning_rate": 9.969697282565241e-05, "loss": 3.5154, "step": 6358 }, { "epoch": 0.39474827736048174, "grad_norm": 0.5193077409714205, "learning_rate": 9.969657568523473e-05, "loss": 3.4723, "step": 6359 }, { "epoch": 0.3948103544602396, "grad_norm": 0.47983492160875224, "learning_rate": 9.969617828553795e-05, "loss": 3.4807, "step": 6360 }, { "epoch": 0.3948724315599975, "grad_norm": 0.5520762107197731, "learning_rate": 9.969578062656412e-05, "loss": 3.3609, "step": 6361 }, { "epoch": 0.39493450865975543, "grad_norm": 0.5448620397606456, "learning_rate": 9.969538270831538e-05, "loss": 3.3792, "step": 6362 }, { "epoch": 0.3949965857595133, "grad_norm": 0.5979888339690647, "learning_rate": 9.969498453079373e-05, "loss": 3.4922, "step": 6363 }, { "epoch": 0.3950586628592712, "grad_norm": 0.5958790753775157, "learning_rate": 9.969458609400129e-05, "loss": 3.5336, "step": 6364 }, { "epoch": 0.39512073995902913, "grad_norm": 0.7075008287189065, "learning_rate": 9.969418739794014e-05, "loss": 3.5512, "step": 6365 }, { "epoch": 0.395182817058787, "grad_norm": 0.6932130471143595, "learning_rate": 9.969378844261234e-05, "loss": 3.4936, "step": 6366 }, { "epoch": 0.3952448941585449, "grad_norm": 0.5883346097741405, "learning_rate": 9.969338922802e-05, "loss": 3.5371, "step": 6367 }, { "epoch": 0.39530697125830283, "grad_norm": 0.5560054166327577, "learning_rate": 9.969298975416516e-05, "loss": 3.427, "step": 6368 }, { "epoch": 0.3953690483580607, "grad_norm": 0.5676985369429114, "learning_rate": 9.969259002104994e-05, "loss": 3.3975, "step": 6369 }, { "epoch": 0.3954311254578186, "grad_norm": 0.6294846570082048, "learning_rate": 9.969219002867642e-05, "loss": 3.5313, "step": 6370 }, { "epoch": 0.39549320255757653, "grad_norm": 0.584951996137906, "learning_rate": 9.969178977704668e-05, "loss": 3.5254, "step": 6371 }, { "epoch": 0.3955552796573344, "grad_norm": 0.5577226261857698, "learning_rate": 9.969138926616282e-05, "loss": 3.467, "step": 6372 }, { "epoch": 0.3956173567570923, "grad_norm": 0.549363143628046, "learning_rate": 9.96909884960269e-05, "loss": 3.3941, "step": 6373 }, { "epoch": 0.3956794338568502, "grad_norm": 0.5778383111217608, "learning_rate": 9.969058746664102e-05, "loss": 3.5034, "step": 6374 }, { "epoch": 0.3957415109566081, "grad_norm": 0.6646432244709862, "learning_rate": 9.969018617800731e-05, "loss": 3.4525, "step": 6375 }, { "epoch": 0.395803588056366, "grad_norm": 0.6126493126015031, "learning_rate": 9.968978463012782e-05, "loss": 3.5164, "step": 6376 }, { "epoch": 0.3958656651561239, "grad_norm": 0.5274476408170597, "learning_rate": 9.968938282300466e-05, "loss": 3.4442, "step": 6377 }, { "epoch": 0.3959277422558818, "grad_norm": 0.47806612038684665, "learning_rate": 9.968898075663995e-05, "loss": 3.4358, "step": 6378 }, { "epoch": 0.3959898193556397, "grad_norm": 0.5396929385336974, "learning_rate": 9.968857843103573e-05, "loss": 3.4662, "step": 6379 }, { "epoch": 0.3960518964553976, "grad_norm": 0.5257649559455018, "learning_rate": 9.968817584619417e-05, "loss": 3.4229, "step": 6380 }, { "epoch": 0.3961139735551555, "grad_norm": 0.44700173334877985, "learning_rate": 9.968777300211731e-05, "loss": 3.585, "step": 6381 }, { "epoch": 0.3961760506549134, "grad_norm": 0.5722223467296889, "learning_rate": 9.968736989880729e-05, "loss": 3.5149, "step": 6382 }, { "epoch": 0.3962381277546713, "grad_norm": 0.577974428724951, "learning_rate": 9.968696653626619e-05, "loss": 3.4171, "step": 6383 }, { "epoch": 0.3963002048544292, "grad_norm": 0.5455831801390628, "learning_rate": 9.968656291449612e-05, "loss": 3.4385, "step": 6384 }, { "epoch": 0.3963622819541871, "grad_norm": 0.5688301312935531, "learning_rate": 9.96861590334992e-05, "loss": 3.5012, "step": 6385 }, { "epoch": 0.396424359053945, "grad_norm": 0.5586735595029055, "learning_rate": 9.968575489327752e-05, "loss": 3.3731, "step": 6386 }, { "epoch": 0.3964864361537029, "grad_norm": 0.47160905897618205, "learning_rate": 9.968535049383318e-05, "loss": 3.479, "step": 6387 }, { "epoch": 0.3965485132534608, "grad_norm": 0.5309861874499077, "learning_rate": 9.968494583516833e-05, "loss": 3.4508, "step": 6388 }, { "epoch": 0.3966105903532187, "grad_norm": 0.47553932854518166, "learning_rate": 9.968454091728505e-05, "loss": 3.4278, "step": 6389 }, { "epoch": 0.3966726674529766, "grad_norm": 0.4543209407398562, "learning_rate": 9.968413574018545e-05, "loss": 3.3186, "step": 6390 }, { "epoch": 0.3967347445527345, "grad_norm": 0.5785239363954863, "learning_rate": 9.968373030387166e-05, "loss": 3.4005, "step": 6391 }, { "epoch": 0.3967968216524924, "grad_norm": 0.6360897397482053, "learning_rate": 9.968332460834579e-05, "loss": 3.5944, "step": 6392 }, { "epoch": 0.3968588987522503, "grad_norm": 0.6464190493306133, "learning_rate": 9.968291865360994e-05, "loss": 3.5357, "step": 6393 }, { "epoch": 0.3969209758520082, "grad_norm": 0.6615535687549633, "learning_rate": 9.968251243966626e-05, "loss": 3.4844, "step": 6394 }, { "epoch": 0.3969830529517661, "grad_norm": 0.5843120918480597, "learning_rate": 9.968210596651686e-05, "loss": 3.4065, "step": 6395 }, { "epoch": 0.397045130051524, "grad_norm": 0.559263534735109, "learning_rate": 9.968169923416383e-05, "loss": 3.4425, "step": 6396 }, { "epoch": 0.3971072071512819, "grad_norm": 0.6028775048295831, "learning_rate": 9.968129224260931e-05, "loss": 3.4228, "step": 6397 }, { "epoch": 0.3971692842510398, "grad_norm": 0.5920573501693152, "learning_rate": 9.968088499185545e-05, "loss": 3.3941, "step": 6398 }, { "epoch": 0.3972313613507977, "grad_norm": 0.5909917913599426, "learning_rate": 9.968047748190434e-05, "loss": 3.5579, "step": 6399 }, { "epoch": 0.3972934384505556, "grad_norm": 0.5749515865510713, "learning_rate": 9.968006971275813e-05, "loss": 3.4349, "step": 6400 }, { "epoch": 0.3973555155503135, "grad_norm": 0.6074492273581208, "learning_rate": 9.967966168441892e-05, "loss": 3.3889, "step": 6401 }, { "epoch": 0.39741759265007137, "grad_norm": 0.530859523208968, "learning_rate": 9.967925339688886e-05, "loss": 3.457, "step": 6402 }, { "epoch": 0.3974796697498293, "grad_norm": 0.5507582531179473, "learning_rate": 9.967884485017008e-05, "loss": 3.3763, "step": 6403 }, { "epoch": 0.3975417468495872, "grad_norm": 0.5113842577099774, "learning_rate": 9.96784360442647e-05, "loss": 3.4889, "step": 6404 }, { "epoch": 0.39760382394934507, "grad_norm": 0.5284756980187962, "learning_rate": 9.967802697917488e-05, "loss": 3.333, "step": 6405 }, { "epoch": 0.397665901049103, "grad_norm": 0.5836918841559968, "learning_rate": 9.967761765490272e-05, "loss": 3.4651, "step": 6406 }, { "epoch": 0.3977279781488609, "grad_norm": 0.5500163997025795, "learning_rate": 9.967720807145036e-05, "loss": 3.4247, "step": 6407 }, { "epoch": 0.39779005524861877, "grad_norm": 0.4689935210589755, "learning_rate": 9.967679822881994e-05, "loss": 3.4909, "step": 6408 }, { "epoch": 0.3978521323483767, "grad_norm": 0.5449278909768163, "learning_rate": 9.967638812701361e-05, "loss": 3.4345, "step": 6409 }, { "epoch": 0.3979142094481346, "grad_norm": 0.5245996640486084, "learning_rate": 9.967597776603353e-05, "loss": 3.5827, "step": 6410 }, { "epoch": 0.39797628654789247, "grad_norm": 0.5364280957585781, "learning_rate": 9.967556714588178e-05, "loss": 3.483, "step": 6411 }, { "epoch": 0.3980383636476504, "grad_norm": 0.5257628019590905, "learning_rate": 9.967515626656055e-05, "loss": 3.4154, "step": 6412 }, { "epoch": 0.3981004407474083, "grad_norm": 0.49515381325832075, "learning_rate": 9.967474512807198e-05, "loss": 3.5272, "step": 6413 }, { "epoch": 0.39816251784716616, "grad_norm": 0.5706669215652233, "learning_rate": 9.967433373041819e-05, "loss": 3.4907, "step": 6414 }, { "epoch": 0.3982245949469241, "grad_norm": 0.6190320229574374, "learning_rate": 9.967392207360136e-05, "loss": 3.5076, "step": 6415 }, { "epoch": 0.398286672046682, "grad_norm": 0.5683236127266347, "learning_rate": 9.96735101576236e-05, "loss": 3.4532, "step": 6416 }, { "epoch": 0.39834874914643986, "grad_norm": 0.57166891918721, "learning_rate": 9.967309798248708e-05, "loss": 3.4589, "step": 6417 }, { "epoch": 0.3984108262461978, "grad_norm": 0.6219890068254131, "learning_rate": 9.967268554819395e-05, "loss": 3.3802, "step": 6418 }, { "epoch": 0.3984729033459557, "grad_norm": 0.6492062114142844, "learning_rate": 9.967227285474638e-05, "loss": 3.3816, "step": 6419 }, { "epoch": 0.39853498044571356, "grad_norm": 0.5670988649413095, "learning_rate": 9.967185990214649e-05, "loss": 3.4159, "step": 6420 }, { "epoch": 0.3985970575454715, "grad_norm": 0.47485503757889935, "learning_rate": 9.967144669039645e-05, "loss": 3.3936, "step": 6421 }, { "epoch": 0.3986591346452294, "grad_norm": 0.5756193453565897, "learning_rate": 9.967103321949842e-05, "loss": 3.5179, "step": 6422 }, { "epoch": 0.39872121174498726, "grad_norm": 0.6059507604823222, "learning_rate": 9.967061948945455e-05, "loss": 3.5015, "step": 6423 }, { "epoch": 0.3987832888447452, "grad_norm": 0.5845251736991738, "learning_rate": 9.9670205500267e-05, "loss": 3.3445, "step": 6424 }, { "epoch": 0.3988453659445031, "grad_norm": 0.6860320501834156, "learning_rate": 9.966979125193791e-05, "loss": 3.4826, "step": 6425 }, { "epoch": 0.39890744304426096, "grad_norm": 0.6916266052486696, "learning_rate": 9.96693767444695e-05, "loss": 3.3625, "step": 6426 }, { "epoch": 0.3989695201440189, "grad_norm": 0.6293783133986837, "learning_rate": 9.966896197786386e-05, "loss": 3.536, "step": 6427 }, { "epoch": 0.3990315972437768, "grad_norm": 0.5804010641853966, "learning_rate": 9.96685469521232e-05, "loss": 3.5308, "step": 6428 }, { "epoch": 0.39909367434353465, "grad_norm": 0.6472999522212244, "learning_rate": 9.966813166724969e-05, "loss": 3.4382, "step": 6429 }, { "epoch": 0.3991557514432926, "grad_norm": 0.6236627817046522, "learning_rate": 9.966771612324545e-05, "loss": 3.4769, "step": 6430 }, { "epoch": 0.3992178285430505, "grad_norm": 0.5160596345071423, "learning_rate": 9.966730032011268e-05, "loss": 3.3449, "step": 6431 }, { "epoch": 0.39927990564280835, "grad_norm": 0.6443911775913276, "learning_rate": 9.966688425785358e-05, "loss": 3.4563, "step": 6432 }, { "epoch": 0.39934198274256627, "grad_norm": 0.7096229388805276, "learning_rate": 9.966646793647026e-05, "loss": 3.458, "step": 6433 }, { "epoch": 0.3994040598423242, "grad_norm": 0.5573917073331353, "learning_rate": 9.966605135596492e-05, "loss": 3.5368, "step": 6434 }, { "epoch": 0.39946613694208205, "grad_norm": 0.5830869125814994, "learning_rate": 9.966563451633974e-05, "loss": 3.4217, "step": 6435 }, { "epoch": 0.39952821404183997, "grad_norm": 0.5948449879766938, "learning_rate": 9.96652174175969e-05, "loss": 3.5966, "step": 6436 }, { "epoch": 0.3995902911415979, "grad_norm": 0.5499506700674844, "learning_rate": 9.966480005973854e-05, "loss": 3.3843, "step": 6437 }, { "epoch": 0.39965236824135575, "grad_norm": 0.49655949514258896, "learning_rate": 9.966438244276687e-05, "loss": 3.4116, "step": 6438 }, { "epoch": 0.39971444534111367, "grad_norm": 0.531221557304576, "learning_rate": 9.966396456668406e-05, "loss": 3.435, "step": 6439 }, { "epoch": 0.3997765224408716, "grad_norm": 0.5706028917506762, "learning_rate": 9.96635464314923e-05, "loss": 3.5939, "step": 6440 }, { "epoch": 0.39983859954062945, "grad_norm": 0.5633649269203914, "learning_rate": 9.966312803719376e-05, "loss": 3.3943, "step": 6441 }, { "epoch": 0.39990067664038736, "grad_norm": 0.5634722783869581, "learning_rate": 9.966270938379061e-05, "loss": 3.4746, "step": 6442 }, { "epoch": 0.3999627537401453, "grad_norm": 0.5323699198147035, "learning_rate": 9.966229047128505e-05, "loss": 3.548, "step": 6443 }, { "epoch": 0.40002483083990314, "grad_norm": 0.6300857098874342, "learning_rate": 9.966187129967929e-05, "loss": 3.4062, "step": 6444 }, { "epoch": 0.40008690793966106, "grad_norm": 0.6898077682148954, "learning_rate": 9.966145186897546e-05, "loss": 3.4858, "step": 6445 }, { "epoch": 0.400148985039419, "grad_norm": 0.5829259153572648, "learning_rate": 9.966103217917581e-05, "loss": 3.4977, "step": 6446 }, { "epoch": 0.40021106213917684, "grad_norm": 0.5599048140640952, "learning_rate": 9.966061223028248e-05, "loss": 3.5182, "step": 6447 }, { "epoch": 0.40027313923893476, "grad_norm": 0.5961861575033851, "learning_rate": 9.966019202229768e-05, "loss": 3.4258, "step": 6448 }, { "epoch": 0.4003352163386927, "grad_norm": 0.582272831493328, "learning_rate": 9.965977155522362e-05, "loss": 3.3395, "step": 6449 }, { "epoch": 0.40039729343845054, "grad_norm": 0.6138539889667831, "learning_rate": 9.965935082906246e-05, "loss": 3.3892, "step": 6450 }, { "epoch": 0.40045937053820846, "grad_norm": 0.5747796776813867, "learning_rate": 9.965892984381642e-05, "loss": 3.4622, "step": 6451 }, { "epoch": 0.4005214476379664, "grad_norm": 0.5707156702812938, "learning_rate": 9.965850859948768e-05, "loss": 3.4619, "step": 6452 }, { "epoch": 0.40058352473772424, "grad_norm": 0.5816887737607387, "learning_rate": 9.965808709607845e-05, "loss": 3.4016, "step": 6453 }, { "epoch": 0.40064560183748216, "grad_norm": 0.566548717602568, "learning_rate": 9.965766533359093e-05, "loss": 3.4305, "step": 6454 }, { "epoch": 0.4007076789372401, "grad_norm": 0.5419908678193394, "learning_rate": 9.96572433120273e-05, "loss": 3.4219, "step": 6455 }, { "epoch": 0.40076975603699794, "grad_norm": 0.5571315808779467, "learning_rate": 9.965682103138979e-05, "loss": 3.4993, "step": 6456 }, { "epoch": 0.40083183313675586, "grad_norm": 0.6059589493542114, "learning_rate": 9.965639849168058e-05, "loss": 3.4117, "step": 6457 }, { "epoch": 0.4008939102365138, "grad_norm": 0.7448698860834077, "learning_rate": 9.96559756929019e-05, "loss": 3.4355, "step": 6458 }, { "epoch": 0.40095598733627164, "grad_norm": 0.688324758689466, "learning_rate": 9.965555263505593e-05, "loss": 3.4077, "step": 6459 }, { "epoch": 0.40101806443602955, "grad_norm": 0.6184046850996705, "learning_rate": 9.965512931814489e-05, "loss": 3.4934, "step": 6460 }, { "epoch": 0.40108014153578747, "grad_norm": 0.6173711334543049, "learning_rate": 9.965470574217098e-05, "loss": 3.4984, "step": 6461 }, { "epoch": 0.40114221863554533, "grad_norm": 0.5665548458518651, "learning_rate": 9.965428190713642e-05, "loss": 3.4315, "step": 6462 }, { "epoch": 0.40120429573530325, "grad_norm": 0.530370770087132, "learning_rate": 9.965385781304342e-05, "loss": 3.4162, "step": 6463 }, { "epoch": 0.40126637283506117, "grad_norm": 0.49493106752212174, "learning_rate": 9.96534334598942e-05, "loss": 3.4448, "step": 6464 }, { "epoch": 0.40132844993481903, "grad_norm": 0.5585966043466054, "learning_rate": 9.965300884769095e-05, "loss": 3.4829, "step": 6465 }, { "epoch": 0.40139052703457695, "grad_norm": 0.5705041171551383, "learning_rate": 9.965258397643588e-05, "loss": 3.4778, "step": 6466 }, { "epoch": 0.40145260413433487, "grad_norm": 0.5946225480948693, "learning_rate": 9.965215884613126e-05, "loss": 3.4033, "step": 6467 }, { "epoch": 0.40151468123409273, "grad_norm": 0.5933899791828625, "learning_rate": 9.965173345677926e-05, "loss": 3.4514, "step": 6468 }, { "epoch": 0.40157675833385065, "grad_norm": 0.5793592834810031, "learning_rate": 9.965130780838212e-05, "loss": 3.4321, "step": 6469 }, { "epoch": 0.40163883543360857, "grad_norm": 0.5876643610391739, "learning_rate": 9.965088190094205e-05, "loss": 3.5178, "step": 6470 }, { "epoch": 0.4017009125333664, "grad_norm": 0.5310791806868175, "learning_rate": 9.965045573446127e-05, "loss": 3.3107, "step": 6471 }, { "epoch": 0.40176298963312435, "grad_norm": 0.5377964411038624, "learning_rate": 9.9650029308942e-05, "loss": 3.4073, "step": 6472 }, { "epoch": 0.40182506673288226, "grad_norm": 0.6321523779005099, "learning_rate": 9.964960262438649e-05, "loss": 3.5463, "step": 6473 }, { "epoch": 0.4018871438326401, "grad_norm": 0.5733523417341767, "learning_rate": 9.964917568079696e-05, "loss": 3.4934, "step": 6474 }, { "epoch": 0.40194922093239804, "grad_norm": 0.49872247089331956, "learning_rate": 9.964874847817562e-05, "loss": 3.4032, "step": 6475 }, { "epoch": 0.40201129803215596, "grad_norm": 0.637237010265207, "learning_rate": 9.964832101652469e-05, "loss": 3.5353, "step": 6476 }, { "epoch": 0.4020733751319138, "grad_norm": 0.688320104257894, "learning_rate": 9.964789329584642e-05, "loss": 3.5069, "step": 6477 }, { "epoch": 0.40213545223167174, "grad_norm": 0.5611264625134256, "learning_rate": 9.964746531614307e-05, "loss": 3.3815, "step": 6478 }, { "epoch": 0.40219752933142966, "grad_norm": 0.5046768801490055, "learning_rate": 9.964703707741681e-05, "loss": 3.4267, "step": 6479 }, { "epoch": 0.4022596064311875, "grad_norm": 0.5671730113215011, "learning_rate": 9.964660857966991e-05, "loss": 3.5064, "step": 6480 }, { "epoch": 0.40232168353094544, "grad_norm": 0.5356640321003698, "learning_rate": 9.96461798229046e-05, "loss": 3.5198, "step": 6481 }, { "epoch": 0.40238376063070336, "grad_norm": 0.5670987335606645, "learning_rate": 9.964575080712312e-05, "loss": 3.4625, "step": 6482 }, { "epoch": 0.4024458377304612, "grad_norm": 0.6274613789151533, "learning_rate": 9.964532153232771e-05, "loss": 3.4258, "step": 6483 }, { "epoch": 0.40250791483021914, "grad_norm": 0.679000682869794, "learning_rate": 9.96448919985206e-05, "loss": 3.4037, "step": 6484 }, { "epoch": 0.40256999192997706, "grad_norm": 0.5612756597094936, "learning_rate": 9.964446220570405e-05, "loss": 3.4199, "step": 6485 }, { "epoch": 0.4026320690297349, "grad_norm": 0.5570480729354206, "learning_rate": 9.964403215388028e-05, "loss": 3.3574, "step": 6486 }, { "epoch": 0.40269414612949284, "grad_norm": 0.5159303454637221, "learning_rate": 9.964360184305155e-05, "loss": 3.4516, "step": 6487 }, { "epoch": 0.40275622322925075, "grad_norm": 0.5209672342474808, "learning_rate": 9.964317127322009e-05, "loss": 3.4058, "step": 6488 }, { "epoch": 0.4028183003290086, "grad_norm": 0.5450185601110099, "learning_rate": 9.964274044438815e-05, "loss": 3.4534, "step": 6489 }, { "epoch": 0.40288037742876653, "grad_norm": 0.48227032895052, "learning_rate": 9.964230935655798e-05, "loss": 3.4845, "step": 6490 }, { "epoch": 0.40294245452852445, "grad_norm": 0.5020288788745115, "learning_rate": 9.964187800973183e-05, "loss": 3.4285, "step": 6491 }, { "epoch": 0.4030045316282823, "grad_norm": 0.4869568678566546, "learning_rate": 9.964144640391198e-05, "loss": 3.4806, "step": 6492 }, { "epoch": 0.40306660872804023, "grad_norm": 0.5140190836651529, "learning_rate": 9.964101453910061e-05, "loss": 3.4086, "step": 6493 }, { "epoch": 0.40312868582779815, "grad_norm": 0.5806288753745646, "learning_rate": 9.964058241530006e-05, "loss": 3.4457, "step": 6494 }, { "epoch": 0.403190762927556, "grad_norm": 0.5271639764735014, "learning_rate": 9.96401500325125e-05, "loss": 3.406, "step": 6495 }, { "epoch": 0.40325284002731393, "grad_norm": 0.5538737816165513, "learning_rate": 9.963971739074025e-05, "loss": 3.4049, "step": 6496 }, { "epoch": 0.40331491712707185, "grad_norm": 0.522384695033877, "learning_rate": 9.963928448998554e-05, "loss": 3.4736, "step": 6497 }, { "epoch": 0.4033769942268297, "grad_norm": 0.4768615628520444, "learning_rate": 9.963885133025064e-05, "loss": 3.5143, "step": 6498 }, { "epoch": 0.40343907132658763, "grad_norm": 0.5322489602511935, "learning_rate": 9.963841791153778e-05, "loss": 3.4401, "step": 6499 }, { "epoch": 0.40350114842634555, "grad_norm": 0.5428301682394108, "learning_rate": 9.963798423384927e-05, "loss": 3.3977, "step": 6500 }, { "epoch": 0.4035632255261034, "grad_norm": 0.5298129447997245, "learning_rate": 9.963755029718733e-05, "loss": 3.5405, "step": 6501 }, { "epoch": 0.4036253026258613, "grad_norm": 0.5813785195548608, "learning_rate": 9.963711610155424e-05, "loss": 3.4187, "step": 6502 }, { "epoch": 0.40368737972561924, "grad_norm": 0.45536438284388914, "learning_rate": 9.963668164695228e-05, "loss": 3.5182, "step": 6503 }, { "epoch": 0.4037494568253771, "grad_norm": 0.5004376046180735, "learning_rate": 9.963624693338369e-05, "loss": 3.4867, "step": 6504 }, { "epoch": 0.403811533925135, "grad_norm": 0.5323828511479844, "learning_rate": 9.963581196085075e-05, "loss": 3.3864, "step": 6505 }, { "epoch": 0.40387361102489294, "grad_norm": 0.48601132534316155, "learning_rate": 9.963537672935574e-05, "loss": 3.4872, "step": 6506 }, { "epoch": 0.4039356881246508, "grad_norm": 0.5224315028945659, "learning_rate": 9.963494123890091e-05, "loss": 3.4826, "step": 6507 }, { "epoch": 0.4039977652244087, "grad_norm": 0.5991995339626084, "learning_rate": 9.963450548948854e-05, "loss": 3.4469, "step": 6508 }, { "epoch": 0.40405984232416664, "grad_norm": 0.5346201139311084, "learning_rate": 9.96340694811209e-05, "loss": 3.4028, "step": 6509 }, { "epoch": 0.4041219194239245, "grad_norm": 0.5495650750087635, "learning_rate": 9.96336332138003e-05, "loss": 3.394, "step": 6510 }, { "epoch": 0.4041839965236824, "grad_norm": 0.5389348860355102, "learning_rate": 9.963319668752895e-05, "loss": 3.3534, "step": 6511 }, { "epoch": 0.40424607362344034, "grad_norm": 0.5097231207772401, "learning_rate": 9.963275990230918e-05, "loss": 3.4666, "step": 6512 }, { "epoch": 0.4043081507231982, "grad_norm": 0.5521743957225377, "learning_rate": 9.963232285814325e-05, "loss": 3.4089, "step": 6513 }, { "epoch": 0.4043702278229561, "grad_norm": 0.49863337969869115, "learning_rate": 9.963188555503346e-05, "loss": 3.4884, "step": 6514 }, { "epoch": 0.40443230492271404, "grad_norm": 0.5378421570460239, "learning_rate": 9.963144799298205e-05, "loss": 3.4612, "step": 6515 }, { "epoch": 0.4044943820224719, "grad_norm": 0.5240401030402362, "learning_rate": 9.963101017199135e-05, "loss": 3.4313, "step": 6516 }, { "epoch": 0.4045564591222298, "grad_norm": 0.4803714498606597, "learning_rate": 9.963057209206362e-05, "loss": 3.3815, "step": 6517 }, { "epoch": 0.40461853622198773, "grad_norm": 0.5338169239978793, "learning_rate": 9.963013375320112e-05, "loss": 3.4001, "step": 6518 }, { "epoch": 0.4046806133217456, "grad_norm": 0.5335229047490043, "learning_rate": 9.962969515540618e-05, "loss": 3.4704, "step": 6519 }, { "epoch": 0.4047426904215035, "grad_norm": 0.5925142076454786, "learning_rate": 9.962925629868108e-05, "loss": 3.4357, "step": 6520 }, { "epoch": 0.40480476752126143, "grad_norm": 0.4833479575570084, "learning_rate": 9.96288171830281e-05, "loss": 3.3706, "step": 6521 }, { "epoch": 0.4048668446210193, "grad_norm": 0.49407908844289145, "learning_rate": 9.962837780844953e-05, "loss": 3.4346, "step": 6522 }, { "epoch": 0.4049289217207772, "grad_norm": 0.6032948506799927, "learning_rate": 9.962793817494767e-05, "loss": 3.4012, "step": 6523 }, { "epoch": 0.40499099882053513, "grad_norm": 0.5711412548341086, "learning_rate": 9.96274982825248e-05, "loss": 3.4328, "step": 6524 }, { "epoch": 0.405053075920293, "grad_norm": 0.5520815819286661, "learning_rate": 9.962705813118323e-05, "loss": 3.4428, "step": 6525 }, { "epoch": 0.4051151530200509, "grad_norm": 0.6318845184846853, "learning_rate": 9.962661772092523e-05, "loss": 3.3944, "step": 6526 }, { "epoch": 0.40517723011980883, "grad_norm": 0.717718171567622, "learning_rate": 9.962617705175314e-05, "loss": 3.4029, "step": 6527 }, { "epoch": 0.4052393072195667, "grad_norm": 0.7163454968100367, "learning_rate": 9.962573612366923e-05, "loss": 3.5716, "step": 6528 }, { "epoch": 0.4053013843193246, "grad_norm": 0.7224215152046939, "learning_rate": 9.962529493667581e-05, "loss": 3.4251, "step": 6529 }, { "epoch": 0.4053634614190825, "grad_norm": 0.671803892054389, "learning_rate": 9.962485349077519e-05, "loss": 3.5044, "step": 6530 }, { "epoch": 0.4054255385188404, "grad_norm": 0.7158099456078049, "learning_rate": 9.962441178596964e-05, "loss": 3.5717, "step": 6531 }, { "epoch": 0.4054876156185983, "grad_norm": 0.7900528311266417, "learning_rate": 9.96239698222615e-05, "loss": 3.4899, "step": 6532 }, { "epoch": 0.4055496927183562, "grad_norm": 0.7101809490008807, "learning_rate": 9.962352759965307e-05, "loss": 3.41, "step": 6533 }, { "epoch": 0.4056117698181141, "grad_norm": 0.7378237078541273, "learning_rate": 9.962308511814664e-05, "loss": 3.3863, "step": 6534 }, { "epoch": 0.405673846917872, "grad_norm": 0.7031554533403486, "learning_rate": 9.962264237774452e-05, "loss": 3.4005, "step": 6535 }, { "epoch": 0.4057359240176299, "grad_norm": 0.6775942390959827, "learning_rate": 9.962219937844904e-05, "loss": 3.449, "step": 6536 }, { "epoch": 0.4057980011173878, "grad_norm": 0.6815369754018857, "learning_rate": 9.96217561202625e-05, "loss": 3.4743, "step": 6537 }, { "epoch": 0.4058600782171457, "grad_norm": 0.6294921609876759, "learning_rate": 9.96213126031872e-05, "loss": 3.4641, "step": 6538 }, { "epoch": 0.4059221553169036, "grad_norm": 0.5743676336436647, "learning_rate": 9.96208688272255e-05, "loss": 3.4204, "step": 6539 }, { "epoch": 0.4059842324166615, "grad_norm": 0.5312347409917965, "learning_rate": 9.962042479237966e-05, "loss": 3.3114, "step": 6540 }, { "epoch": 0.4060463095164194, "grad_norm": 0.5713540325552946, "learning_rate": 9.961998049865199e-05, "loss": 3.4299, "step": 6541 }, { "epoch": 0.4061083866161773, "grad_norm": 0.5956422118014737, "learning_rate": 9.961953594604488e-05, "loss": 3.5297, "step": 6542 }, { "epoch": 0.4061704637159352, "grad_norm": 0.5719497902747823, "learning_rate": 9.961909113456059e-05, "loss": 3.3643, "step": 6543 }, { "epoch": 0.4062325408156931, "grad_norm": 0.6179907919018726, "learning_rate": 9.961864606420145e-05, "loss": 3.4113, "step": 6544 }, { "epoch": 0.406294617915451, "grad_norm": 0.5082360482015501, "learning_rate": 9.961820073496979e-05, "loss": 3.5132, "step": 6545 }, { "epoch": 0.4063566950152089, "grad_norm": 0.5170872789991133, "learning_rate": 9.961775514686793e-05, "loss": 3.4906, "step": 6546 }, { "epoch": 0.4064187721149668, "grad_norm": 0.6277478371525569, "learning_rate": 9.961730929989821e-05, "loss": 3.4427, "step": 6547 }, { "epoch": 0.4064808492147247, "grad_norm": 0.7627264155689862, "learning_rate": 9.961686319406293e-05, "loss": 3.4389, "step": 6548 }, { "epoch": 0.4065429263144826, "grad_norm": 0.5259629433019956, "learning_rate": 9.961641682936442e-05, "loss": 3.3911, "step": 6549 }, { "epoch": 0.4066050034142405, "grad_norm": 0.5113483573576852, "learning_rate": 9.961597020580505e-05, "loss": 3.4942, "step": 6550 }, { "epoch": 0.4066670805139984, "grad_norm": 0.5627417045093691, "learning_rate": 9.961552332338708e-05, "loss": 3.4559, "step": 6551 }, { "epoch": 0.4067291576137563, "grad_norm": 0.44864777476529333, "learning_rate": 9.961507618211289e-05, "loss": 3.3935, "step": 6552 }, { "epoch": 0.4067912347135142, "grad_norm": 0.5699740150033219, "learning_rate": 9.961462878198482e-05, "loss": 3.4773, "step": 6553 }, { "epoch": 0.4068533118132721, "grad_norm": 0.5969111965847821, "learning_rate": 9.961418112300518e-05, "loss": 3.4779, "step": 6554 }, { "epoch": 0.40691538891303, "grad_norm": 0.5688456392047796, "learning_rate": 9.961373320517632e-05, "loss": 3.3419, "step": 6555 }, { "epoch": 0.4069774660127879, "grad_norm": 0.5602393020926032, "learning_rate": 9.961328502850055e-05, "loss": 3.4758, "step": 6556 }, { "epoch": 0.40703954311254575, "grad_norm": 0.5149251928441064, "learning_rate": 9.961283659298026e-05, "loss": 3.4232, "step": 6557 }, { "epoch": 0.40710162021230367, "grad_norm": 0.564693149101403, "learning_rate": 9.961238789861773e-05, "loss": 3.4935, "step": 6558 }, { "epoch": 0.4071636973120616, "grad_norm": 0.5699248368079739, "learning_rate": 9.961193894541533e-05, "loss": 3.4031, "step": 6559 }, { "epoch": 0.40722577441181945, "grad_norm": 0.4777311629080535, "learning_rate": 9.961148973337541e-05, "loss": 3.3791, "step": 6560 }, { "epoch": 0.40728785151157737, "grad_norm": 0.5200520250081306, "learning_rate": 9.96110402625003e-05, "loss": 3.3277, "step": 6561 }, { "epoch": 0.4073499286113353, "grad_norm": 0.5217079511988932, "learning_rate": 9.961059053279236e-05, "loss": 3.544, "step": 6562 }, { "epoch": 0.40741200571109315, "grad_norm": 0.5022578223793235, "learning_rate": 9.96101405442539e-05, "loss": 3.3791, "step": 6563 }, { "epoch": 0.40747408281085107, "grad_norm": 0.546727596580971, "learning_rate": 9.960969029688732e-05, "loss": 3.4432, "step": 6564 }, { "epoch": 0.407536159910609, "grad_norm": 0.5475005523034363, "learning_rate": 9.960923979069494e-05, "loss": 3.4262, "step": 6565 }, { "epoch": 0.40759823701036685, "grad_norm": 0.5547568855408443, "learning_rate": 9.960878902567909e-05, "loss": 3.4992, "step": 6566 }, { "epoch": 0.40766031411012477, "grad_norm": 0.6442846115312513, "learning_rate": 9.960833800184217e-05, "loss": 3.5021, "step": 6567 }, { "epoch": 0.4077223912098827, "grad_norm": 0.623744657566225, "learning_rate": 9.96078867191865e-05, "loss": 3.4589, "step": 6568 }, { "epoch": 0.40778446830964055, "grad_norm": 0.5754840430558876, "learning_rate": 9.960743517771443e-05, "loss": 3.3273, "step": 6569 }, { "epoch": 0.40784654540939846, "grad_norm": 0.543816268653556, "learning_rate": 9.960698337742835e-05, "loss": 3.4108, "step": 6570 }, { "epoch": 0.4079086225091564, "grad_norm": 0.5604624668063729, "learning_rate": 9.960653131833058e-05, "loss": 3.4503, "step": 6571 }, { "epoch": 0.40797069960891424, "grad_norm": 0.578935029914213, "learning_rate": 9.96060790004235e-05, "loss": 3.385, "step": 6572 }, { "epoch": 0.40803277670867216, "grad_norm": 0.5468255156880886, "learning_rate": 9.960562642370945e-05, "loss": 3.5207, "step": 6573 }, { "epoch": 0.4080948538084301, "grad_norm": 0.4824436352721816, "learning_rate": 9.960517358819082e-05, "loss": 3.3262, "step": 6574 }, { "epoch": 0.40815693090818794, "grad_norm": 0.5662467995188962, "learning_rate": 9.960472049386995e-05, "loss": 3.3706, "step": 6575 }, { "epoch": 0.40821900800794586, "grad_norm": 0.5775763511553078, "learning_rate": 9.960426714074922e-05, "loss": 3.4325, "step": 6576 }, { "epoch": 0.4082810851077038, "grad_norm": 0.5510624107057045, "learning_rate": 9.960381352883099e-05, "loss": 3.4464, "step": 6577 }, { "epoch": 0.40834316220746164, "grad_norm": 0.6118099763354261, "learning_rate": 9.960335965811761e-05, "loss": 3.4164, "step": 6578 }, { "epoch": 0.40840523930721956, "grad_norm": 0.6015615587103279, "learning_rate": 9.960290552861147e-05, "loss": 3.4494, "step": 6579 }, { "epoch": 0.4084673164069775, "grad_norm": 0.5426939124633856, "learning_rate": 9.960245114031492e-05, "loss": 3.4571, "step": 6580 }, { "epoch": 0.40852939350673534, "grad_norm": 0.516118045088365, "learning_rate": 9.960199649323034e-05, "loss": 3.4476, "step": 6581 }, { "epoch": 0.40859147060649326, "grad_norm": 0.596518287119371, "learning_rate": 9.960154158736012e-05, "loss": 3.3651, "step": 6582 }, { "epoch": 0.4086535477062512, "grad_norm": 0.6524772735818014, "learning_rate": 9.960108642270661e-05, "loss": 3.4177, "step": 6583 }, { "epoch": 0.40871562480600904, "grad_norm": 0.6583949412318407, "learning_rate": 9.960063099927217e-05, "loss": 3.4328, "step": 6584 }, { "epoch": 0.40877770190576695, "grad_norm": 0.5804853968331887, "learning_rate": 9.960017531705922e-05, "loss": 3.5246, "step": 6585 }, { "epoch": 0.4088397790055249, "grad_norm": 0.6284776968978061, "learning_rate": 9.959971937607011e-05, "loss": 3.5007, "step": 6586 }, { "epoch": 0.40890185610528273, "grad_norm": 0.6205305508970188, "learning_rate": 9.959926317630721e-05, "loss": 3.3837, "step": 6587 }, { "epoch": 0.40896393320504065, "grad_norm": 0.5549856915648552, "learning_rate": 9.959880671777294e-05, "loss": 3.392, "step": 6588 }, { "epoch": 0.40902601030479857, "grad_norm": 0.5784854796061901, "learning_rate": 9.959835000046964e-05, "loss": 3.4035, "step": 6589 }, { "epoch": 0.40908808740455643, "grad_norm": 0.6053562182786314, "learning_rate": 9.95978930243997e-05, "loss": 3.408, "step": 6590 }, { "epoch": 0.40915016450431435, "grad_norm": 0.6152462428325838, "learning_rate": 9.959743578956551e-05, "loss": 3.3823, "step": 6591 }, { "epoch": 0.40921224160407227, "grad_norm": 0.498717182341731, "learning_rate": 9.959697829596948e-05, "loss": 3.4322, "step": 6592 }, { "epoch": 0.40927431870383013, "grad_norm": 0.4495363191034396, "learning_rate": 9.959652054361395e-05, "loss": 3.3314, "step": 6593 }, { "epoch": 0.40933639580358805, "grad_norm": 0.5483083741421878, "learning_rate": 9.959606253250134e-05, "loss": 3.4037, "step": 6594 }, { "epoch": 0.40939847290334597, "grad_norm": 0.705727339081785, "learning_rate": 9.959560426263404e-05, "loss": 3.4234, "step": 6595 }, { "epoch": 0.40946055000310383, "grad_norm": 0.5450661854408856, "learning_rate": 9.959514573401443e-05, "loss": 3.3935, "step": 6596 }, { "epoch": 0.40952262710286175, "grad_norm": 0.5951523282499488, "learning_rate": 9.95946869466449e-05, "loss": 3.4268, "step": 6597 }, { "epoch": 0.40958470420261966, "grad_norm": 0.6281140473582804, "learning_rate": 9.959422790052786e-05, "loss": 3.4412, "step": 6598 }, { "epoch": 0.4096467813023775, "grad_norm": 0.7311654262052503, "learning_rate": 9.959376859566568e-05, "loss": 3.4043, "step": 6599 }, { "epoch": 0.40970885840213545, "grad_norm": 0.5824172512922448, "learning_rate": 9.959330903206077e-05, "loss": 3.1766, "step": 6600 }, { "epoch": 0.40977093550189336, "grad_norm": 0.6560152178807714, "learning_rate": 9.959284920971553e-05, "loss": 3.4066, "step": 6601 }, { "epoch": 0.4098330126016512, "grad_norm": 0.5909106729309539, "learning_rate": 9.959238912863236e-05, "loss": 3.3, "step": 6602 }, { "epoch": 0.40989508970140914, "grad_norm": 0.5461228511625708, "learning_rate": 9.959192878881366e-05, "loss": 3.4292, "step": 6603 }, { "epoch": 0.40995716680116706, "grad_norm": 0.5460372502646731, "learning_rate": 9.959146819026182e-05, "loss": 3.3324, "step": 6604 }, { "epoch": 0.4100192439009249, "grad_norm": 0.6434606966836832, "learning_rate": 9.959100733297926e-05, "loss": 3.5236, "step": 6605 }, { "epoch": 0.41008132100068284, "grad_norm": 0.570841896266708, "learning_rate": 9.959054621696837e-05, "loss": 3.4181, "step": 6606 }, { "epoch": 0.41014339810044076, "grad_norm": 0.5756519363926738, "learning_rate": 9.959008484223156e-05, "loss": 3.4618, "step": 6607 }, { "epoch": 0.4102054752001986, "grad_norm": 0.5583959109078054, "learning_rate": 9.958962320877125e-05, "loss": 3.4711, "step": 6608 }, { "epoch": 0.41026755229995654, "grad_norm": 0.565145734204539, "learning_rate": 9.958916131658983e-05, "loss": 3.3909, "step": 6609 }, { "epoch": 0.41032962939971446, "grad_norm": 0.5542444689590406, "learning_rate": 9.958869916568971e-05, "loss": 3.3919, "step": 6610 }, { "epoch": 0.4103917064994723, "grad_norm": 0.5466416542312857, "learning_rate": 9.958823675607332e-05, "loss": 3.4023, "step": 6611 }, { "epoch": 0.41045378359923024, "grad_norm": 0.5279450555809205, "learning_rate": 9.958777408774306e-05, "loss": 3.4079, "step": 6612 }, { "epoch": 0.41051586069898816, "grad_norm": 0.5989986129706473, "learning_rate": 9.958731116070135e-05, "loss": 3.4207, "step": 6613 }, { "epoch": 0.410577937798746, "grad_norm": 0.5523726287155547, "learning_rate": 9.95868479749506e-05, "loss": 3.3834, "step": 6614 }, { "epoch": 0.41064001489850394, "grad_norm": 0.5074484473855426, "learning_rate": 9.95863845304932e-05, "loss": 3.498, "step": 6615 }, { "epoch": 0.41070209199826185, "grad_norm": 0.5722442987523817, "learning_rate": 9.958592082733161e-05, "loss": 3.43, "step": 6616 }, { "epoch": 0.4107641690980197, "grad_norm": 0.5243744222480946, "learning_rate": 9.958545686546824e-05, "loss": 3.3056, "step": 6617 }, { "epoch": 0.41082624619777763, "grad_norm": 0.5210065331168413, "learning_rate": 9.958499264490551e-05, "loss": 3.3892, "step": 6618 }, { "epoch": 0.41088832329753555, "grad_norm": 0.543498871350279, "learning_rate": 9.958452816564582e-05, "loss": 3.5024, "step": 6619 }, { "epoch": 0.4109504003972934, "grad_norm": 0.544112804657383, "learning_rate": 9.958406342769162e-05, "loss": 3.4108, "step": 6620 }, { "epoch": 0.41101247749705133, "grad_norm": 0.5017076181360656, "learning_rate": 9.958359843104532e-05, "loss": 3.4194, "step": 6621 }, { "epoch": 0.41107455459680925, "grad_norm": 0.45897895322803406, "learning_rate": 9.958313317570935e-05, "loss": 3.4056, "step": 6622 }, { "epoch": 0.4111366316965671, "grad_norm": 0.48690709347410394, "learning_rate": 9.958266766168613e-05, "loss": 3.365, "step": 6623 }, { "epoch": 0.41119870879632503, "grad_norm": 0.5076739782523624, "learning_rate": 9.95822018889781e-05, "loss": 3.5451, "step": 6624 }, { "epoch": 0.41126078589608295, "grad_norm": 0.5801873152491234, "learning_rate": 9.95817358575877e-05, "loss": 3.4093, "step": 6625 }, { "epoch": 0.4113228629958408, "grad_norm": 0.6086901949462393, "learning_rate": 9.958126956751735e-05, "loss": 3.3747, "step": 6626 }, { "epoch": 0.41138494009559873, "grad_norm": 0.6564281085687385, "learning_rate": 9.958080301876946e-05, "loss": 3.486, "step": 6627 }, { "epoch": 0.41144701719535665, "grad_norm": 0.644402508449112, "learning_rate": 9.958033621134647e-05, "loss": 3.4139, "step": 6628 }, { "epoch": 0.4115090942951145, "grad_norm": 0.6163510799497215, "learning_rate": 9.957986914525086e-05, "loss": 3.4892, "step": 6629 }, { "epoch": 0.4115711713948724, "grad_norm": 0.6170747871330867, "learning_rate": 9.957940182048504e-05, "loss": 3.4079, "step": 6630 }, { "epoch": 0.41163324849463034, "grad_norm": 0.6622360243397278, "learning_rate": 9.957893423705144e-05, "loss": 3.4795, "step": 6631 }, { "epoch": 0.4116953255943882, "grad_norm": 0.6522865270312437, "learning_rate": 9.95784663949525e-05, "loss": 3.5352, "step": 6632 }, { "epoch": 0.4117574026941461, "grad_norm": 0.6164755759277772, "learning_rate": 9.957799829419065e-05, "loss": 3.3443, "step": 6633 }, { "epoch": 0.41181947979390404, "grad_norm": 0.6772388768653869, "learning_rate": 9.957752993476836e-05, "loss": 3.4719, "step": 6634 }, { "epoch": 0.4118815568936619, "grad_norm": 0.6329664584654886, "learning_rate": 9.957706131668807e-05, "loss": 3.3952, "step": 6635 }, { "epoch": 0.4119436339934198, "grad_norm": 0.7241014206542434, "learning_rate": 9.957659243995221e-05, "loss": 3.5191, "step": 6636 }, { "epoch": 0.41200571109317774, "grad_norm": 0.6301003014102319, "learning_rate": 9.957612330456323e-05, "loss": 3.4332, "step": 6637 }, { "epoch": 0.4120677881929356, "grad_norm": 0.5679564473368961, "learning_rate": 9.957565391052358e-05, "loss": 3.3586, "step": 6638 }, { "epoch": 0.4121298652926935, "grad_norm": 0.6448380866979394, "learning_rate": 9.957518425783572e-05, "loss": 3.5125, "step": 6639 }, { "epoch": 0.41219194239245144, "grad_norm": 0.6374317927429972, "learning_rate": 9.957471434650207e-05, "loss": 3.4258, "step": 6640 }, { "epoch": 0.4122540194922093, "grad_norm": 0.6137171368628483, "learning_rate": 9.95742441765251e-05, "loss": 3.4303, "step": 6641 }, { "epoch": 0.4123160965919672, "grad_norm": 0.6132229576835446, "learning_rate": 9.957377374790729e-05, "loss": 3.4464, "step": 6642 }, { "epoch": 0.41237817369172514, "grad_norm": 0.6433344509019516, "learning_rate": 9.957330306065104e-05, "loss": 3.4836, "step": 6643 }, { "epoch": 0.412440250791483, "grad_norm": 0.6204240895560172, "learning_rate": 9.957283211475884e-05, "loss": 3.4618, "step": 6644 }, { "epoch": 0.4125023278912409, "grad_norm": 0.5836524913389246, "learning_rate": 9.957236091023314e-05, "loss": 3.5173, "step": 6645 }, { "epoch": 0.41256440499099883, "grad_norm": 0.6433962683415008, "learning_rate": 9.95718894470764e-05, "loss": 3.3401, "step": 6646 }, { "epoch": 0.4126264820907567, "grad_norm": 0.5827717774510272, "learning_rate": 9.957141772529108e-05, "loss": 3.5092, "step": 6647 }, { "epoch": 0.4126885591905146, "grad_norm": 0.5552531568410859, "learning_rate": 9.957094574487965e-05, "loss": 3.4417, "step": 6648 }, { "epoch": 0.41275063629027253, "grad_norm": 0.5767605376833185, "learning_rate": 9.957047350584457e-05, "loss": 3.4574, "step": 6649 }, { "epoch": 0.4128127133900304, "grad_norm": 0.561245074728444, "learning_rate": 9.957000100818827e-05, "loss": 3.4019, "step": 6650 }, { "epoch": 0.4128747904897883, "grad_norm": 0.5113527284768911, "learning_rate": 9.956952825191325e-05, "loss": 3.3311, "step": 6651 }, { "epoch": 0.41293686758954623, "grad_norm": 0.5514882197085954, "learning_rate": 9.956905523702197e-05, "loss": 3.4384, "step": 6652 }, { "epoch": 0.4129989446893041, "grad_norm": 0.5291878563350453, "learning_rate": 9.956858196351688e-05, "loss": 3.3246, "step": 6653 }, { "epoch": 0.413061021789062, "grad_norm": 0.5152621871232493, "learning_rate": 9.95681084314005e-05, "loss": 3.4738, "step": 6654 }, { "epoch": 0.41312309888881993, "grad_norm": 0.6021085960429672, "learning_rate": 9.956763464067523e-05, "loss": 3.4189, "step": 6655 }, { "epoch": 0.4131851759885778, "grad_norm": 0.5918003245841732, "learning_rate": 9.956716059134359e-05, "loss": 3.4836, "step": 6656 }, { "epoch": 0.4132472530883357, "grad_norm": 0.5536156897055925, "learning_rate": 9.956668628340803e-05, "loss": 3.4709, "step": 6657 }, { "epoch": 0.4133093301880936, "grad_norm": 0.4617659994822234, "learning_rate": 9.956621171687106e-05, "loss": 3.4278, "step": 6658 }, { "epoch": 0.4133714072878515, "grad_norm": 0.573352575639951, "learning_rate": 9.956573689173512e-05, "loss": 3.4966, "step": 6659 }, { "epoch": 0.4134334843876094, "grad_norm": 0.6500109460715978, "learning_rate": 9.95652618080027e-05, "loss": 3.2944, "step": 6660 }, { "epoch": 0.4134955614873673, "grad_norm": 0.6026200003647023, "learning_rate": 9.956478646567626e-05, "loss": 3.3695, "step": 6661 }, { "epoch": 0.4135576385871252, "grad_norm": 0.59066409733456, "learning_rate": 9.956431086475833e-05, "loss": 3.4392, "step": 6662 }, { "epoch": 0.4136197156868831, "grad_norm": 0.6206714465998324, "learning_rate": 9.956383500525133e-05, "loss": 3.3432, "step": 6663 }, { "epoch": 0.413681792786641, "grad_norm": 0.5647709357681734, "learning_rate": 9.956335888715778e-05, "loss": 3.3628, "step": 6664 }, { "epoch": 0.4137438698863989, "grad_norm": 0.6316850768084618, "learning_rate": 9.956288251048015e-05, "loss": 3.3787, "step": 6665 }, { "epoch": 0.4138059469861568, "grad_norm": 0.6460403807806778, "learning_rate": 9.956240587522093e-05, "loss": 3.4816, "step": 6666 }, { "epoch": 0.4138680240859147, "grad_norm": 0.5687200968080981, "learning_rate": 9.956192898138261e-05, "loss": 3.4628, "step": 6667 }, { "epoch": 0.4139301011856726, "grad_norm": 0.49205807089123516, "learning_rate": 9.956145182896768e-05, "loss": 3.3088, "step": 6668 }, { "epoch": 0.4139921782854305, "grad_norm": 0.6245231716844661, "learning_rate": 9.956097441797861e-05, "loss": 3.4178, "step": 6669 }, { "epoch": 0.4140542553851884, "grad_norm": 0.655938437844197, "learning_rate": 9.956049674841791e-05, "loss": 3.434, "step": 6670 }, { "epoch": 0.4141163324849463, "grad_norm": 0.5509448518657409, "learning_rate": 9.956001882028808e-05, "loss": 3.4084, "step": 6671 }, { "epoch": 0.4141784095847042, "grad_norm": 0.5022499602048639, "learning_rate": 9.955954063359159e-05, "loss": 3.4425, "step": 6672 }, { "epoch": 0.4142404866844621, "grad_norm": 0.5873438170996991, "learning_rate": 9.955906218833096e-05, "loss": 3.4122, "step": 6673 }, { "epoch": 0.41430256378422, "grad_norm": 0.5584302542909639, "learning_rate": 9.955858348450865e-05, "loss": 3.4076, "step": 6674 }, { "epoch": 0.4143646408839779, "grad_norm": 0.5387144208873879, "learning_rate": 9.95581045221272e-05, "loss": 3.395, "step": 6675 }, { "epoch": 0.4144267179837358, "grad_norm": 0.6029248355117879, "learning_rate": 9.955762530118907e-05, "loss": 3.3704, "step": 6676 }, { "epoch": 0.4144887950834937, "grad_norm": 0.5579339245194171, "learning_rate": 9.955714582169677e-05, "loss": 3.5117, "step": 6677 }, { "epoch": 0.4145508721832516, "grad_norm": 0.5615972321160215, "learning_rate": 9.955666608365284e-05, "loss": 3.4577, "step": 6678 }, { "epoch": 0.4146129492830095, "grad_norm": 0.5986926498502217, "learning_rate": 9.955618608705972e-05, "loss": 3.4637, "step": 6679 }, { "epoch": 0.4146750263827674, "grad_norm": 0.6321540986617136, "learning_rate": 9.955570583191998e-05, "loss": 3.4692, "step": 6680 }, { "epoch": 0.4147371034825253, "grad_norm": 0.5953024094895376, "learning_rate": 9.955522531823607e-05, "loss": 3.4238, "step": 6681 }, { "epoch": 0.4147991805822832, "grad_norm": 0.5229996515313003, "learning_rate": 9.955474454601053e-05, "loss": 3.3828, "step": 6682 }, { "epoch": 0.4148612576820411, "grad_norm": 0.5276176871412254, "learning_rate": 9.955426351524585e-05, "loss": 3.3933, "step": 6683 }, { "epoch": 0.414923334781799, "grad_norm": 0.5831015813362845, "learning_rate": 9.955378222594455e-05, "loss": 3.3814, "step": 6684 }, { "epoch": 0.4149854118815569, "grad_norm": 0.5550643285638108, "learning_rate": 9.955330067810915e-05, "loss": 3.358, "step": 6685 }, { "epoch": 0.41504748898131477, "grad_norm": 0.5118441537265873, "learning_rate": 9.955281887174214e-05, "loss": 3.3925, "step": 6686 }, { "epoch": 0.4151095660810727, "grad_norm": 0.4890983763384499, "learning_rate": 9.955233680684604e-05, "loss": 3.4508, "step": 6687 }, { "epoch": 0.4151716431808306, "grad_norm": 0.5273855969814862, "learning_rate": 9.955185448342337e-05, "loss": 3.4594, "step": 6688 }, { "epoch": 0.41523372028058847, "grad_norm": 0.6035360684682214, "learning_rate": 9.955137190147665e-05, "loss": 3.4042, "step": 6689 }, { "epoch": 0.4152957973803464, "grad_norm": 0.5978492475064161, "learning_rate": 9.95508890610084e-05, "loss": 3.4626, "step": 6690 }, { "epoch": 0.4153578744801043, "grad_norm": 0.5858910097116402, "learning_rate": 9.955040596202113e-05, "loss": 3.4108, "step": 6691 }, { "epoch": 0.41541995157986217, "grad_norm": 0.5357540161295575, "learning_rate": 9.954992260451737e-05, "loss": 3.4368, "step": 6692 }, { "epoch": 0.4154820286796201, "grad_norm": 0.5502366965495548, "learning_rate": 9.954943898849962e-05, "loss": 3.4587, "step": 6693 }, { "epoch": 0.415544105779378, "grad_norm": 0.5994660892822259, "learning_rate": 9.954895511397042e-05, "loss": 3.3546, "step": 6694 }, { "epoch": 0.41560618287913587, "grad_norm": 0.5888541300830012, "learning_rate": 9.954847098093229e-05, "loss": 3.4978, "step": 6695 }, { "epoch": 0.4156682599788938, "grad_norm": 0.5519715561863728, "learning_rate": 9.954798658938778e-05, "loss": 3.3927, "step": 6696 }, { "epoch": 0.4157303370786517, "grad_norm": 0.5382444836997828, "learning_rate": 9.954750193933938e-05, "loss": 3.36, "step": 6697 }, { "epoch": 0.41579241417840956, "grad_norm": 0.5196938511224668, "learning_rate": 9.954701703078963e-05, "loss": 3.4066, "step": 6698 }, { "epoch": 0.4158544912781675, "grad_norm": 0.6248162476309979, "learning_rate": 9.954653186374108e-05, "loss": 3.5233, "step": 6699 }, { "epoch": 0.4159165683779254, "grad_norm": 0.5734840155920674, "learning_rate": 9.954604643819622e-05, "loss": 3.4093, "step": 6700 }, { "epoch": 0.41597864547768326, "grad_norm": 0.5582044618776087, "learning_rate": 9.954556075415764e-05, "loss": 3.462, "step": 6701 }, { "epoch": 0.4160407225774412, "grad_norm": 0.5836494531564039, "learning_rate": 9.954507481162782e-05, "loss": 3.4148, "step": 6702 }, { "epoch": 0.4161027996771991, "grad_norm": 0.5742952205056551, "learning_rate": 9.954458861060933e-05, "loss": 3.4408, "step": 6703 }, { "epoch": 0.41616487677695696, "grad_norm": 0.5535544791681206, "learning_rate": 9.954410215110468e-05, "loss": 3.4506, "step": 6704 }, { "epoch": 0.4162269538767149, "grad_norm": 0.5919817645125404, "learning_rate": 9.954361543311643e-05, "loss": 3.3556, "step": 6705 }, { "epoch": 0.4162890309764728, "grad_norm": 0.6178560860772077, "learning_rate": 9.95431284566471e-05, "loss": 3.4382, "step": 6706 }, { "epoch": 0.41635110807623066, "grad_norm": 0.5985173025283239, "learning_rate": 9.954264122169928e-05, "loss": 3.3344, "step": 6707 }, { "epoch": 0.4164131851759886, "grad_norm": 0.5485312313831122, "learning_rate": 9.954215372827542e-05, "loss": 3.3808, "step": 6708 }, { "epoch": 0.4164752622757465, "grad_norm": 0.5008306458105422, "learning_rate": 9.954166597637814e-05, "loss": 3.3818, "step": 6709 }, { "epoch": 0.41653733937550436, "grad_norm": 0.4962294385479464, "learning_rate": 9.954117796600999e-05, "loss": 3.3968, "step": 6710 }, { "epoch": 0.4165994164752623, "grad_norm": 0.5284635451906827, "learning_rate": 9.954068969717345e-05, "loss": 3.3742, "step": 6711 }, { "epoch": 0.4166614935750202, "grad_norm": 0.538005538998775, "learning_rate": 9.954020116987111e-05, "loss": 3.4145, "step": 6712 }, { "epoch": 0.41672357067477805, "grad_norm": 0.4454862690574556, "learning_rate": 9.953971238410553e-05, "loss": 3.4539, "step": 6713 }, { "epoch": 0.41678564777453597, "grad_norm": 0.5339541892035873, "learning_rate": 9.953922333987923e-05, "loss": 3.4127, "step": 6714 }, { "epoch": 0.4168477248742939, "grad_norm": 0.4524258448499553, "learning_rate": 9.953873403719479e-05, "loss": 3.4653, "step": 6715 }, { "epoch": 0.41690980197405175, "grad_norm": 0.4366776708072974, "learning_rate": 9.953824447605474e-05, "loss": 3.4044, "step": 6716 }, { "epoch": 0.41697187907380967, "grad_norm": 0.5295431542430304, "learning_rate": 9.953775465646165e-05, "loss": 3.3804, "step": 6717 }, { "epoch": 0.4170339561735676, "grad_norm": 0.581909166883817, "learning_rate": 9.953726457841806e-05, "loss": 3.4217, "step": 6718 }, { "epoch": 0.41709603327332545, "grad_norm": 0.5378480582753724, "learning_rate": 9.953677424192653e-05, "loss": 3.341, "step": 6719 }, { "epoch": 0.41715811037308337, "grad_norm": 0.5811517714489757, "learning_rate": 9.953628364698963e-05, "loss": 3.3706, "step": 6720 }, { "epoch": 0.4172201874728413, "grad_norm": 0.6034502901433578, "learning_rate": 9.953579279360991e-05, "loss": 3.3392, "step": 6721 }, { "epoch": 0.41728226457259915, "grad_norm": 0.5540421721316584, "learning_rate": 9.953530168178994e-05, "loss": 3.4623, "step": 6722 }, { "epoch": 0.41734434167235707, "grad_norm": 0.6159307988700797, "learning_rate": 9.953481031153228e-05, "loss": 3.4297, "step": 6723 }, { "epoch": 0.417406418772115, "grad_norm": 0.6855370501822249, "learning_rate": 9.953431868283949e-05, "loss": 3.5206, "step": 6724 }, { "epoch": 0.41746849587187285, "grad_norm": 0.6163508865383491, "learning_rate": 9.953382679571412e-05, "loss": 3.3587, "step": 6725 }, { "epoch": 0.41753057297163076, "grad_norm": 0.5016851340609134, "learning_rate": 9.953333465015875e-05, "loss": 3.4826, "step": 6726 }, { "epoch": 0.4175926500713887, "grad_norm": 0.5105593118706988, "learning_rate": 9.953284224617596e-05, "loss": 3.429, "step": 6727 }, { "epoch": 0.41765472717114654, "grad_norm": 0.5975041912841635, "learning_rate": 9.953234958376831e-05, "loss": 3.3917, "step": 6728 }, { "epoch": 0.41771680427090446, "grad_norm": 0.5188984853731591, "learning_rate": 9.953185666293836e-05, "loss": 3.3484, "step": 6729 }, { "epoch": 0.4177788813706624, "grad_norm": 0.504049591258636, "learning_rate": 9.953136348368868e-05, "loss": 3.3494, "step": 6730 }, { "epoch": 0.41784095847042024, "grad_norm": 0.5114614990302148, "learning_rate": 9.953087004602187e-05, "loss": 3.4838, "step": 6731 }, { "epoch": 0.41790303557017816, "grad_norm": 0.5095092596498569, "learning_rate": 9.953037634994049e-05, "loss": 3.4098, "step": 6732 }, { "epoch": 0.4179651126699361, "grad_norm": 0.4527854962981467, "learning_rate": 9.952988239544709e-05, "loss": 3.3336, "step": 6733 }, { "epoch": 0.41802718976969394, "grad_norm": 0.5456505282483113, "learning_rate": 9.95293881825443e-05, "loss": 3.4089, "step": 6734 }, { "epoch": 0.41808926686945186, "grad_norm": 0.6404489182181882, "learning_rate": 9.952889371123464e-05, "loss": 3.439, "step": 6735 }, { "epoch": 0.4181513439692098, "grad_norm": 0.6840542577296641, "learning_rate": 9.952839898152073e-05, "loss": 3.4469, "step": 6736 }, { "epoch": 0.41821342106896764, "grad_norm": 0.5191449041025176, "learning_rate": 9.952790399340514e-05, "loss": 3.3894, "step": 6737 }, { "epoch": 0.41827549816872556, "grad_norm": 0.4738073597608943, "learning_rate": 9.952740874689046e-05, "loss": 3.4275, "step": 6738 }, { "epoch": 0.4183375752684835, "grad_norm": 0.5625981139340266, "learning_rate": 9.952691324197924e-05, "loss": 3.4574, "step": 6739 }, { "epoch": 0.41839965236824134, "grad_norm": 0.570820934504415, "learning_rate": 9.952641747867412e-05, "loss": 3.4137, "step": 6740 }, { "epoch": 0.41846172946799925, "grad_norm": 0.6095167386445441, "learning_rate": 9.952592145697764e-05, "loss": 3.2845, "step": 6741 }, { "epoch": 0.4185238065677572, "grad_norm": 0.5424504997154069, "learning_rate": 9.952542517689241e-05, "loss": 3.461, "step": 6742 }, { "epoch": 0.41858588366751504, "grad_norm": 0.5018255168040415, "learning_rate": 9.9524928638421e-05, "loss": 3.4074, "step": 6743 }, { "epoch": 0.41864796076727295, "grad_norm": 0.5831735653294324, "learning_rate": 9.952443184156604e-05, "loss": 3.367, "step": 6744 }, { "epoch": 0.41871003786703087, "grad_norm": 0.5509655420931421, "learning_rate": 9.952393478633008e-05, "loss": 3.3469, "step": 6745 }, { "epoch": 0.41877211496678873, "grad_norm": 0.5343576584617774, "learning_rate": 9.952343747271573e-05, "loss": 3.4649, "step": 6746 }, { "epoch": 0.41883419206654665, "grad_norm": 0.5367325449229909, "learning_rate": 9.952293990072558e-05, "loss": 3.3978, "step": 6747 }, { "epoch": 0.41889626916630457, "grad_norm": 0.5273299886532717, "learning_rate": 9.952244207036225e-05, "loss": 3.388, "step": 6748 }, { "epoch": 0.41895834626606243, "grad_norm": 0.5493227809959971, "learning_rate": 9.952194398162831e-05, "loss": 3.4692, "step": 6749 }, { "epoch": 0.41902042336582035, "grad_norm": 0.4998373124571714, "learning_rate": 9.952144563452636e-05, "loss": 3.493, "step": 6750 }, { "epoch": 0.41908250046557827, "grad_norm": 0.49643792956363786, "learning_rate": 9.952094702905901e-05, "loss": 3.4231, "step": 6751 }, { "epoch": 0.41914457756533613, "grad_norm": 0.5334539417399031, "learning_rate": 9.952044816522885e-05, "loss": 3.4521, "step": 6752 }, { "epoch": 0.41920665466509405, "grad_norm": 0.5651093730305796, "learning_rate": 9.95199490430385e-05, "loss": 3.3915, "step": 6753 }, { "epoch": 0.41926873176485197, "grad_norm": 0.5533409699186884, "learning_rate": 9.951944966249055e-05, "loss": 3.4203, "step": 6754 }, { "epoch": 0.4193308088646098, "grad_norm": 0.5368583231190973, "learning_rate": 9.951895002358761e-05, "loss": 3.4253, "step": 6755 }, { "epoch": 0.41939288596436775, "grad_norm": 0.5524762902415523, "learning_rate": 9.951845012633229e-05, "loss": 3.3907, "step": 6756 }, { "epoch": 0.41945496306412566, "grad_norm": 0.5320246322153033, "learning_rate": 9.95179499707272e-05, "loss": 3.3671, "step": 6757 }, { "epoch": 0.4195170401638835, "grad_norm": 0.555043925359948, "learning_rate": 9.951744955677493e-05, "loss": 3.4492, "step": 6758 }, { "epoch": 0.41957911726364144, "grad_norm": 0.5022329003464572, "learning_rate": 9.951694888447811e-05, "loss": 3.3245, "step": 6759 }, { "epoch": 0.41964119436339936, "grad_norm": 0.4279813184281886, "learning_rate": 9.951644795383935e-05, "loss": 3.3318, "step": 6760 }, { "epoch": 0.4197032714631572, "grad_norm": 0.5625544892516204, "learning_rate": 9.951594676486125e-05, "loss": 3.4456, "step": 6761 }, { "epoch": 0.41976534856291514, "grad_norm": 0.5618365932148606, "learning_rate": 9.951544531754644e-05, "loss": 3.386, "step": 6762 }, { "epoch": 0.41982742566267306, "grad_norm": 0.5546694470543744, "learning_rate": 9.951494361189755e-05, "loss": 3.444, "step": 6763 }, { "epoch": 0.4198895027624309, "grad_norm": 0.5488987670179889, "learning_rate": 9.951444164791717e-05, "loss": 3.4435, "step": 6764 }, { "epoch": 0.41995157986218884, "grad_norm": 0.541097448888788, "learning_rate": 9.951393942560792e-05, "loss": 3.5363, "step": 6765 }, { "epoch": 0.42001365696194676, "grad_norm": 0.5175725108867839, "learning_rate": 9.951343694497241e-05, "loss": 3.4419, "step": 6766 }, { "epoch": 0.4200757340617046, "grad_norm": 0.49887064705490636, "learning_rate": 9.951293420601331e-05, "loss": 3.4643, "step": 6767 }, { "epoch": 0.42013781116146254, "grad_norm": 0.5752721453452918, "learning_rate": 9.951243120873321e-05, "loss": 3.3678, "step": 6768 }, { "epoch": 0.42019988826122046, "grad_norm": 0.5437341961811171, "learning_rate": 9.951192795313473e-05, "loss": 3.3725, "step": 6769 }, { "epoch": 0.4202619653609783, "grad_norm": 0.4876421751651451, "learning_rate": 9.951142443922051e-05, "loss": 3.5293, "step": 6770 }, { "epoch": 0.42032404246073624, "grad_norm": 0.5598474851329833, "learning_rate": 9.951092066699315e-05, "loss": 3.4023, "step": 6771 }, { "epoch": 0.42038611956049415, "grad_norm": 0.5160914536464604, "learning_rate": 9.95104166364553e-05, "loss": 3.3393, "step": 6772 }, { "epoch": 0.420448196660252, "grad_norm": 0.47415214460467786, "learning_rate": 9.95099123476096e-05, "loss": 3.4409, "step": 6773 }, { "epoch": 0.42051027376000993, "grad_norm": 0.5361911762759383, "learning_rate": 9.950940780045867e-05, "loss": 3.4671, "step": 6774 }, { "epoch": 0.42057235085976785, "grad_norm": 0.5928605846717123, "learning_rate": 9.950890299500513e-05, "loss": 3.4382, "step": 6775 }, { "epoch": 0.4206344279595257, "grad_norm": 0.6470136060890729, "learning_rate": 9.950839793125164e-05, "loss": 3.2969, "step": 6776 }, { "epoch": 0.42069650505928363, "grad_norm": 0.5658270669888845, "learning_rate": 9.950789260920081e-05, "loss": 3.4889, "step": 6777 }, { "epoch": 0.42075858215904155, "grad_norm": 0.6327004510084763, "learning_rate": 9.950738702885527e-05, "loss": 3.3461, "step": 6778 }, { "epoch": 0.4208206592587994, "grad_norm": 0.5886233506583621, "learning_rate": 9.95068811902177e-05, "loss": 3.4586, "step": 6779 }, { "epoch": 0.42088273635855733, "grad_norm": 0.5126470924589659, "learning_rate": 9.95063750932907e-05, "loss": 3.4118, "step": 6780 }, { "epoch": 0.42094481345831525, "grad_norm": 0.5309624735328956, "learning_rate": 9.950586873807692e-05, "loss": 3.3833, "step": 6781 }, { "epoch": 0.4210068905580731, "grad_norm": 0.5465905130752977, "learning_rate": 9.950536212457901e-05, "loss": 3.382, "step": 6782 }, { "epoch": 0.42106896765783103, "grad_norm": 0.6249165002358301, "learning_rate": 9.95048552527996e-05, "loss": 3.3999, "step": 6783 }, { "epoch": 0.42113104475758895, "grad_norm": 0.5243273331229517, "learning_rate": 9.950434812274137e-05, "loss": 3.4017, "step": 6784 }, { "epoch": 0.4211931218573468, "grad_norm": 0.5791099766446458, "learning_rate": 9.950384073440691e-05, "loss": 3.3713, "step": 6785 }, { "epoch": 0.4212551989571047, "grad_norm": 0.5893142799579547, "learning_rate": 9.95033330877989e-05, "loss": 3.465, "step": 6786 }, { "epoch": 0.42131727605686264, "grad_norm": 0.7292005576477387, "learning_rate": 9.950282518291998e-05, "loss": 3.4675, "step": 6787 }, { "epoch": 0.4213793531566205, "grad_norm": 0.7843824150677234, "learning_rate": 9.950231701977282e-05, "loss": 3.3942, "step": 6788 }, { "epoch": 0.4214414302563784, "grad_norm": 0.6737977173067096, "learning_rate": 9.950180859836004e-05, "loss": 3.3417, "step": 6789 }, { "epoch": 0.42150350735613634, "grad_norm": 0.70873909903188, "learning_rate": 9.950129991868431e-05, "loss": 3.4525, "step": 6790 }, { "epoch": 0.4215655844558942, "grad_norm": 0.5986066504482144, "learning_rate": 9.95007909807483e-05, "loss": 3.4579, "step": 6791 }, { "epoch": 0.4216276615556521, "grad_norm": 0.603461673633648, "learning_rate": 9.950028178455461e-05, "loss": 3.4331, "step": 6792 }, { "epoch": 0.42168973865541004, "grad_norm": 0.555823492633157, "learning_rate": 9.949977233010596e-05, "loss": 3.3906, "step": 6793 }, { "epoch": 0.4217518157551679, "grad_norm": 0.4903390932134898, "learning_rate": 9.949926261740499e-05, "loss": 3.254, "step": 6794 }, { "epoch": 0.4218138928549258, "grad_norm": 0.5333430068310997, "learning_rate": 9.949875264645433e-05, "loss": 3.3744, "step": 6795 }, { "epoch": 0.42187596995468374, "grad_norm": 0.5708238582335194, "learning_rate": 9.949824241725667e-05, "loss": 3.4326, "step": 6796 }, { "epoch": 0.4219380470544416, "grad_norm": 0.5362291926902159, "learning_rate": 9.949773192981466e-05, "loss": 3.5271, "step": 6797 }, { "epoch": 0.4220001241541995, "grad_norm": 0.526338990540851, "learning_rate": 9.949722118413096e-05, "loss": 3.4918, "step": 6798 }, { "epoch": 0.42206220125395744, "grad_norm": 0.5218886048148315, "learning_rate": 9.949671018020826e-05, "loss": 3.3942, "step": 6799 }, { "epoch": 0.4221242783537153, "grad_norm": 0.5248425542801333, "learning_rate": 9.94961989180492e-05, "loss": 3.4862, "step": 6800 }, { "epoch": 0.4221863554534732, "grad_norm": 0.591474766459863, "learning_rate": 9.949568739765645e-05, "loss": 3.4052, "step": 6801 }, { "epoch": 0.42224843255323113, "grad_norm": 0.5889906366361776, "learning_rate": 9.949517561903267e-05, "loss": 3.3701, "step": 6802 }, { "epoch": 0.422310509652989, "grad_norm": 0.5196020608701795, "learning_rate": 9.949466358218056e-05, "loss": 3.4112, "step": 6803 }, { "epoch": 0.4223725867527469, "grad_norm": 0.5315619842867674, "learning_rate": 9.949415128710278e-05, "loss": 3.3518, "step": 6804 }, { "epoch": 0.42243466385250483, "grad_norm": 0.45945584046486193, "learning_rate": 9.949363873380199e-05, "loss": 3.3143, "step": 6805 }, { "epoch": 0.4224967409522627, "grad_norm": 0.5116122477116735, "learning_rate": 9.949312592228087e-05, "loss": 3.4513, "step": 6806 }, { "epoch": 0.4225588180520206, "grad_norm": 0.5678906949859672, "learning_rate": 9.94926128525421e-05, "loss": 3.4059, "step": 6807 }, { "epoch": 0.42262089515177853, "grad_norm": 0.4595007082152279, "learning_rate": 9.949209952458836e-05, "loss": 3.449, "step": 6808 }, { "epoch": 0.4226829722515364, "grad_norm": 0.528415692475548, "learning_rate": 9.949158593842231e-05, "loss": 3.4061, "step": 6809 }, { "epoch": 0.4227450493512943, "grad_norm": 0.5144960297207987, "learning_rate": 9.949107209404665e-05, "loss": 3.4397, "step": 6810 }, { "epoch": 0.42280712645105223, "grad_norm": 0.5342747465979418, "learning_rate": 9.949055799146404e-05, "loss": 3.4405, "step": 6811 }, { "epoch": 0.4228692035508101, "grad_norm": 0.6129385847360389, "learning_rate": 9.949004363067718e-05, "loss": 3.2971, "step": 6812 }, { "epoch": 0.422931280650568, "grad_norm": 0.4986937865508711, "learning_rate": 9.948952901168875e-05, "loss": 3.4619, "step": 6813 }, { "epoch": 0.4229933577503259, "grad_norm": 0.5791283226584801, "learning_rate": 9.948901413450144e-05, "loss": 3.4426, "step": 6814 }, { "epoch": 0.4230554348500838, "grad_norm": 0.5486109836862579, "learning_rate": 9.948849899911791e-05, "loss": 3.4089, "step": 6815 }, { "epoch": 0.4231175119498417, "grad_norm": 0.5243347789993452, "learning_rate": 9.94879836055409e-05, "loss": 3.419, "step": 6816 }, { "epoch": 0.4231795890495996, "grad_norm": 0.6201466230479975, "learning_rate": 9.948746795377304e-05, "loss": 3.4028, "step": 6817 }, { "epoch": 0.4232416661493575, "grad_norm": 0.5767201288289981, "learning_rate": 9.948695204381706e-05, "loss": 3.482, "step": 6818 }, { "epoch": 0.4233037432491154, "grad_norm": 0.5262793925995354, "learning_rate": 9.948643587567563e-05, "loss": 3.2915, "step": 6819 }, { "epoch": 0.4233658203488733, "grad_norm": 0.4932262570282922, "learning_rate": 9.948591944935147e-05, "loss": 3.4126, "step": 6820 }, { "epoch": 0.4234278974486312, "grad_norm": 0.5183374844422685, "learning_rate": 9.948540276484722e-05, "loss": 3.489, "step": 6821 }, { "epoch": 0.4234899745483891, "grad_norm": 0.5428538303749226, "learning_rate": 9.948488582216563e-05, "loss": 3.4187, "step": 6822 }, { "epoch": 0.423552051648147, "grad_norm": 0.5675669911205526, "learning_rate": 9.948436862130938e-05, "loss": 3.4641, "step": 6823 }, { "epoch": 0.4236141287479049, "grad_norm": 0.5242511631230089, "learning_rate": 9.948385116228116e-05, "loss": 3.3867, "step": 6824 }, { "epoch": 0.4236762058476628, "grad_norm": 0.48525204780024667, "learning_rate": 9.948333344508367e-05, "loss": 3.4782, "step": 6825 }, { "epoch": 0.4237382829474207, "grad_norm": 0.5156008685851938, "learning_rate": 9.948281546971963e-05, "loss": 3.46, "step": 6826 }, { "epoch": 0.4238003600471786, "grad_norm": 0.5608503204860351, "learning_rate": 9.948229723619172e-05, "loss": 3.3934, "step": 6827 }, { "epoch": 0.4238624371469365, "grad_norm": 0.5422112924154083, "learning_rate": 9.948177874450266e-05, "loss": 3.3069, "step": 6828 }, { "epoch": 0.4239245142466944, "grad_norm": 0.46047661255853867, "learning_rate": 9.948125999465514e-05, "loss": 3.4057, "step": 6829 }, { "epoch": 0.4239865913464523, "grad_norm": 0.4811921759954329, "learning_rate": 9.948074098665188e-05, "loss": 3.4642, "step": 6830 }, { "epoch": 0.4240486684462102, "grad_norm": 0.4666241441949272, "learning_rate": 9.948022172049559e-05, "loss": 3.4054, "step": 6831 }, { "epoch": 0.4241107455459681, "grad_norm": 0.5101682454169612, "learning_rate": 9.947970219618896e-05, "loss": 3.3212, "step": 6832 }, { "epoch": 0.424172822645726, "grad_norm": 0.5706160012127881, "learning_rate": 9.947918241373471e-05, "loss": 3.3807, "step": 6833 }, { "epoch": 0.4242348997454839, "grad_norm": 0.5882509971023325, "learning_rate": 9.947866237313557e-05, "loss": 3.3518, "step": 6834 }, { "epoch": 0.4242969768452418, "grad_norm": 0.5542020150672845, "learning_rate": 9.947814207439423e-05, "loss": 3.4391, "step": 6835 }, { "epoch": 0.4243590539449997, "grad_norm": 0.5997679808156579, "learning_rate": 9.947762151751341e-05, "loss": 3.362, "step": 6836 }, { "epoch": 0.4244211310447576, "grad_norm": 0.6228518524599881, "learning_rate": 9.947710070249583e-05, "loss": 3.3466, "step": 6837 }, { "epoch": 0.4244832081445155, "grad_norm": 0.669531639036691, "learning_rate": 9.947657962934419e-05, "loss": 3.3993, "step": 6838 }, { "epoch": 0.4245452852442734, "grad_norm": 0.6259604703408853, "learning_rate": 9.947605829806124e-05, "loss": 3.3505, "step": 6839 }, { "epoch": 0.4246073623440313, "grad_norm": 0.5803611398915576, "learning_rate": 9.947553670864967e-05, "loss": 3.3754, "step": 6840 }, { "epoch": 0.4246694394437892, "grad_norm": 0.5104219442860239, "learning_rate": 9.947501486111222e-05, "loss": 3.3364, "step": 6841 }, { "epoch": 0.42473151654354707, "grad_norm": 0.47969624203845534, "learning_rate": 9.947449275545163e-05, "loss": 3.3865, "step": 6842 }, { "epoch": 0.424793593643305, "grad_norm": 0.5343861573153198, "learning_rate": 9.947397039167057e-05, "loss": 3.4905, "step": 6843 }, { "epoch": 0.4248556707430629, "grad_norm": 0.575824749363222, "learning_rate": 9.94734477697718e-05, "loss": 3.3625, "step": 6844 }, { "epoch": 0.42491774784282077, "grad_norm": 0.6229210609952969, "learning_rate": 9.947292488975804e-05, "loss": 3.3962, "step": 6845 }, { "epoch": 0.4249798249425787, "grad_norm": 0.580516687932031, "learning_rate": 9.947240175163204e-05, "loss": 3.3387, "step": 6846 }, { "epoch": 0.4250419020423366, "grad_norm": 0.6231833759845128, "learning_rate": 9.94718783553965e-05, "loss": 3.4416, "step": 6847 }, { "epoch": 0.42510397914209447, "grad_norm": 0.6630300237034167, "learning_rate": 9.947135470105416e-05, "loss": 3.3416, "step": 6848 }, { "epoch": 0.4251660562418524, "grad_norm": 0.5765495226466957, "learning_rate": 9.947083078860774e-05, "loss": 3.3611, "step": 6849 }, { "epoch": 0.4252281333416103, "grad_norm": 0.5871638624905924, "learning_rate": 9.947030661806001e-05, "loss": 3.3261, "step": 6850 }, { "epoch": 0.42529021044136817, "grad_norm": 0.5566021286887045, "learning_rate": 9.946978218941367e-05, "loss": 3.3556, "step": 6851 }, { "epoch": 0.4253522875411261, "grad_norm": 0.49529809299057975, "learning_rate": 9.946925750267147e-05, "loss": 3.3606, "step": 6852 }, { "epoch": 0.425414364640884, "grad_norm": 0.5911759748350727, "learning_rate": 9.946873255783613e-05, "loss": 3.3428, "step": 6853 }, { "epoch": 0.42547644174064186, "grad_norm": 0.6089610380995102, "learning_rate": 9.946820735491043e-05, "loss": 3.3673, "step": 6854 }, { "epoch": 0.4255385188403998, "grad_norm": 0.5269812291528422, "learning_rate": 9.946768189389706e-05, "loss": 3.4016, "step": 6855 }, { "epoch": 0.4256005959401577, "grad_norm": 0.6038461415923032, "learning_rate": 9.94671561747988e-05, "loss": 3.4585, "step": 6856 }, { "epoch": 0.42566267303991556, "grad_norm": 0.6775908524311822, "learning_rate": 9.946663019761837e-05, "loss": 3.3517, "step": 6857 }, { "epoch": 0.4257247501396735, "grad_norm": 0.6256223202932858, "learning_rate": 9.946610396235851e-05, "loss": 3.4346, "step": 6858 }, { "epoch": 0.4257868272394314, "grad_norm": 0.546470519756637, "learning_rate": 9.9465577469022e-05, "loss": 3.4205, "step": 6859 }, { "epoch": 0.42584890433918926, "grad_norm": 0.48576730157821196, "learning_rate": 9.946505071761156e-05, "loss": 3.3823, "step": 6860 }, { "epoch": 0.4259109814389472, "grad_norm": 0.5316141227041722, "learning_rate": 9.946452370812993e-05, "loss": 3.3872, "step": 6861 }, { "epoch": 0.4259730585387051, "grad_norm": 0.48899852712093467, "learning_rate": 9.946399644057989e-05, "loss": 3.4676, "step": 6862 }, { "epoch": 0.42603513563846296, "grad_norm": 0.5049570940222149, "learning_rate": 9.946346891496417e-05, "loss": 3.3485, "step": 6863 }, { "epoch": 0.4260972127382209, "grad_norm": 0.566391648729161, "learning_rate": 9.94629411312855e-05, "loss": 3.4277, "step": 6864 }, { "epoch": 0.4261592898379788, "grad_norm": 0.4871577956935972, "learning_rate": 9.946241308954668e-05, "loss": 3.3217, "step": 6865 }, { "epoch": 0.42622136693773666, "grad_norm": 0.5524160324346623, "learning_rate": 9.946188478975043e-05, "loss": 3.3746, "step": 6866 }, { "epoch": 0.4262834440374946, "grad_norm": 0.48581185571766944, "learning_rate": 9.946135623189954e-05, "loss": 3.4461, "step": 6867 }, { "epoch": 0.4263455211372525, "grad_norm": 0.5342564501518055, "learning_rate": 9.946082741599673e-05, "loss": 3.4943, "step": 6868 }, { "epoch": 0.42640759823701035, "grad_norm": 0.5233670372294575, "learning_rate": 9.946029834204478e-05, "loss": 3.4161, "step": 6869 }, { "epoch": 0.4264696753367683, "grad_norm": 0.5290640005506133, "learning_rate": 9.945976901004645e-05, "loss": 3.4538, "step": 6870 }, { "epoch": 0.4265317524365262, "grad_norm": 0.5610007067164221, "learning_rate": 9.94592394200045e-05, "loss": 3.4256, "step": 6871 }, { "epoch": 0.42659382953628405, "grad_norm": 0.5467045382091363, "learning_rate": 9.94587095719217e-05, "loss": 3.3889, "step": 6872 }, { "epoch": 0.42665590663604197, "grad_norm": 0.5103951144696143, "learning_rate": 9.945817946580078e-05, "loss": 3.3738, "step": 6873 }, { "epoch": 0.4267179837357999, "grad_norm": 0.527744960043445, "learning_rate": 9.945764910164455e-05, "loss": 3.3748, "step": 6874 }, { "epoch": 0.42678006083555775, "grad_norm": 0.47519772956631673, "learning_rate": 9.945711847945576e-05, "loss": 3.4136, "step": 6875 }, { "epoch": 0.42684213793531567, "grad_norm": 0.5114126090479857, "learning_rate": 9.945658759923716e-05, "loss": 3.3398, "step": 6876 }, { "epoch": 0.4269042150350736, "grad_norm": 0.4582881110451943, "learning_rate": 9.945605646099154e-05, "loss": 3.3762, "step": 6877 }, { "epoch": 0.42696629213483145, "grad_norm": 0.4781569557481412, "learning_rate": 9.945552506472168e-05, "loss": 3.4062, "step": 6878 }, { "epoch": 0.42702836923458937, "grad_norm": 0.5261580807379347, "learning_rate": 9.945499341043035e-05, "loss": 3.3656, "step": 6879 }, { "epoch": 0.4270904463343473, "grad_norm": 0.5935420374826184, "learning_rate": 9.945446149812029e-05, "loss": 3.4612, "step": 6880 }, { "epoch": 0.42715252343410515, "grad_norm": 0.6699912032574742, "learning_rate": 9.945392932779431e-05, "loss": 3.5281, "step": 6881 }, { "epoch": 0.42721460053386306, "grad_norm": 0.6208287999501795, "learning_rate": 9.945339689945517e-05, "loss": 3.3386, "step": 6882 }, { "epoch": 0.427276677633621, "grad_norm": 0.5719740453530303, "learning_rate": 9.945286421310566e-05, "loss": 3.386, "step": 6883 }, { "epoch": 0.42733875473337884, "grad_norm": 0.7069860201536651, "learning_rate": 9.945233126874854e-05, "loss": 3.4089, "step": 6884 }, { "epoch": 0.42740083183313676, "grad_norm": 0.8089327631336217, "learning_rate": 9.945179806638661e-05, "loss": 3.3729, "step": 6885 }, { "epoch": 0.4274629089328947, "grad_norm": 0.6909325695853992, "learning_rate": 9.945126460602266e-05, "loss": 3.3901, "step": 6886 }, { "epoch": 0.42752498603265254, "grad_norm": 0.6626006472807514, "learning_rate": 9.945073088765944e-05, "loss": 3.3787, "step": 6887 }, { "epoch": 0.42758706313241046, "grad_norm": 0.8907982841700425, "learning_rate": 9.945019691129976e-05, "loss": 3.3627, "step": 6888 }, { "epoch": 0.4276491402321684, "grad_norm": 0.8665402376745489, "learning_rate": 9.944966267694637e-05, "loss": 3.3937, "step": 6889 }, { "epoch": 0.42771121733192624, "grad_norm": 0.7543683785161371, "learning_rate": 9.944912818460212e-05, "loss": 3.4375, "step": 6890 }, { "epoch": 0.42777329443168416, "grad_norm": 0.6290243286740528, "learning_rate": 9.944859343426976e-05, "loss": 3.3649, "step": 6891 }, { "epoch": 0.4278353715314421, "grad_norm": 0.6146096100012113, "learning_rate": 9.944805842595208e-05, "loss": 3.2668, "step": 6892 }, { "epoch": 0.42789744863119994, "grad_norm": 0.6600394738198246, "learning_rate": 9.944752315965187e-05, "loss": 3.3252, "step": 6893 }, { "epoch": 0.42795952573095786, "grad_norm": 0.6589864489981186, "learning_rate": 9.944698763537192e-05, "loss": 3.4701, "step": 6894 }, { "epoch": 0.4280216028307158, "grad_norm": 0.5967896605194272, "learning_rate": 9.944645185311505e-05, "loss": 3.4862, "step": 6895 }, { "epoch": 0.42808367993047364, "grad_norm": 0.6252334635521879, "learning_rate": 9.944591581288403e-05, "loss": 3.4574, "step": 6896 }, { "epoch": 0.42814575703023156, "grad_norm": 0.5179989813128143, "learning_rate": 9.944537951468166e-05, "loss": 3.3405, "step": 6897 }, { "epoch": 0.4282078341299895, "grad_norm": 0.5825153104464514, "learning_rate": 9.944484295851075e-05, "loss": 3.5145, "step": 6898 }, { "epoch": 0.42826991122974734, "grad_norm": 0.6031186671739862, "learning_rate": 9.944430614437407e-05, "loss": 3.3278, "step": 6899 }, { "epoch": 0.42833198832950525, "grad_norm": 0.5402498851212887, "learning_rate": 9.944376907227447e-05, "loss": 3.2767, "step": 6900 }, { "epoch": 0.42839406542926317, "grad_norm": 0.5485219950130181, "learning_rate": 9.944323174221471e-05, "loss": 3.3732, "step": 6901 }, { "epoch": 0.42845614252902103, "grad_norm": 0.5059372241824684, "learning_rate": 9.944269415419761e-05, "loss": 3.4154, "step": 6902 }, { "epoch": 0.42851821962877895, "grad_norm": 0.4749113621327077, "learning_rate": 9.944215630822598e-05, "loss": 3.3416, "step": 6903 }, { "epoch": 0.42858029672853687, "grad_norm": 0.5054733688874561, "learning_rate": 9.94416182043026e-05, "loss": 3.3773, "step": 6904 }, { "epoch": 0.42864237382829473, "grad_norm": 0.48569065276170575, "learning_rate": 9.944107984243031e-05, "loss": 3.2998, "step": 6905 }, { "epoch": 0.42870445092805265, "grad_norm": 0.42326831323308695, "learning_rate": 9.94405412226119e-05, "loss": 3.3566, "step": 6906 }, { "epoch": 0.42876652802781057, "grad_norm": 0.47222027493835816, "learning_rate": 9.944000234485018e-05, "loss": 3.3079, "step": 6907 }, { "epoch": 0.42882860512756843, "grad_norm": 0.4630987608834339, "learning_rate": 9.943946320914797e-05, "loss": 3.4335, "step": 6908 }, { "epoch": 0.42889068222732635, "grad_norm": 0.5532481632934779, "learning_rate": 9.943892381550808e-05, "loss": 3.3226, "step": 6909 }, { "epoch": 0.42895275932708427, "grad_norm": 0.5792287651453665, "learning_rate": 9.943838416393332e-05, "loss": 3.3809, "step": 6910 }, { "epoch": 0.4290148364268421, "grad_norm": 0.5543286680710344, "learning_rate": 9.94378442544265e-05, "loss": 3.412, "step": 6911 }, { "epoch": 0.42907691352660005, "grad_norm": 0.5263532307602216, "learning_rate": 9.943730408699045e-05, "loss": 3.3602, "step": 6912 }, { "epoch": 0.42913899062635796, "grad_norm": 0.5454418750306411, "learning_rate": 9.9436763661628e-05, "loss": 3.3139, "step": 6913 }, { "epoch": 0.4292010677261158, "grad_norm": 0.5132301374792331, "learning_rate": 9.943622297834194e-05, "loss": 3.444, "step": 6914 }, { "epoch": 0.42926314482587374, "grad_norm": 0.5224130769002747, "learning_rate": 9.94356820371351e-05, "loss": 3.4541, "step": 6915 }, { "epoch": 0.42932522192563166, "grad_norm": 0.5002430682638579, "learning_rate": 9.943514083801031e-05, "loss": 3.4692, "step": 6916 }, { "epoch": 0.4293872990253895, "grad_norm": 0.46016699838309727, "learning_rate": 9.943459938097039e-05, "loss": 3.4322, "step": 6917 }, { "epoch": 0.42944937612514744, "grad_norm": 0.4477715774759734, "learning_rate": 9.943405766601817e-05, "loss": 3.4028, "step": 6918 }, { "epoch": 0.42951145322490536, "grad_norm": 0.5510989415979025, "learning_rate": 9.943351569315647e-05, "loss": 3.4598, "step": 6919 }, { "epoch": 0.4295735303246632, "grad_norm": 0.5461275442251435, "learning_rate": 9.94329734623881e-05, "loss": 3.4734, "step": 6920 }, { "epoch": 0.42963560742442114, "grad_norm": 0.5050338430678586, "learning_rate": 9.943243097371592e-05, "loss": 3.3934, "step": 6921 }, { "epoch": 0.42969768452417906, "grad_norm": 0.4827893050958453, "learning_rate": 9.943188822714275e-05, "loss": 3.3176, "step": 6922 }, { "epoch": 0.4297597616239369, "grad_norm": 0.5522030004928266, "learning_rate": 9.943134522267142e-05, "loss": 3.3475, "step": 6923 }, { "epoch": 0.42982183872369484, "grad_norm": 0.5225377676971219, "learning_rate": 9.943080196030474e-05, "loss": 3.3374, "step": 6924 }, { "epoch": 0.42988391582345276, "grad_norm": 0.4834479412586179, "learning_rate": 9.943025844004558e-05, "loss": 3.4541, "step": 6925 }, { "epoch": 0.4299459929232106, "grad_norm": 0.5347348524090996, "learning_rate": 9.942971466189679e-05, "loss": 3.3798, "step": 6926 }, { "epoch": 0.43000807002296854, "grad_norm": 0.5452140653680858, "learning_rate": 9.942917062586115e-05, "loss": 3.4276, "step": 6927 }, { "epoch": 0.43007014712272645, "grad_norm": 0.5251792000157043, "learning_rate": 9.942862633194152e-05, "loss": 3.3098, "step": 6928 }, { "epoch": 0.4301322242224843, "grad_norm": 0.5286152240429238, "learning_rate": 9.942808178014078e-05, "loss": 3.485, "step": 6929 }, { "epoch": 0.43019430132224223, "grad_norm": 0.5886773456605325, "learning_rate": 9.942753697046172e-05, "loss": 3.3566, "step": 6930 }, { "epoch": 0.4302563784220001, "grad_norm": 0.5150995756381195, "learning_rate": 9.942699190290719e-05, "loss": 3.4483, "step": 6931 }, { "epoch": 0.430318455521758, "grad_norm": 0.49830091392782366, "learning_rate": 9.942644657748005e-05, "loss": 3.2408, "step": 6932 }, { "epoch": 0.43038053262151593, "grad_norm": 0.5027151001834322, "learning_rate": 9.942590099418314e-05, "loss": 3.3003, "step": 6933 }, { "epoch": 0.4304426097212738, "grad_norm": 0.4789302664330864, "learning_rate": 9.942535515301932e-05, "loss": 3.3884, "step": 6934 }, { "epoch": 0.4305046868210317, "grad_norm": 0.5336701300761015, "learning_rate": 9.94248090539914e-05, "loss": 3.3809, "step": 6935 }, { "epoch": 0.43056676392078963, "grad_norm": 0.5037112191278234, "learning_rate": 9.942426269710227e-05, "loss": 3.2656, "step": 6936 }, { "epoch": 0.4306288410205475, "grad_norm": 0.5101574090123703, "learning_rate": 9.942371608235476e-05, "loss": 3.4488, "step": 6937 }, { "epoch": 0.4306909181203054, "grad_norm": 0.48605677683245624, "learning_rate": 9.942316920975172e-05, "loss": 3.3777, "step": 6938 }, { "epoch": 0.43075299522006333, "grad_norm": 0.5694641440041517, "learning_rate": 9.942262207929602e-05, "loss": 3.4377, "step": 6939 }, { "epoch": 0.4308150723198212, "grad_norm": 0.5556664491947453, "learning_rate": 9.942207469099049e-05, "loss": 3.3432, "step": 6940 }, { "epoch": 0.4308771494195791, "grad_norm": 0.4849655335508511, "learning_rate": 9.9421527044838e-05, "loss": 3.2779, "step": 6941 }, { "epoch": 0.430939226519337, "grad_norm": 0.4887025922578502, "learning_rate": 9.942097914084142e-05, "loss": 3.3594, "step": 6942 }, { "epoch": 0.4310013036190949, "grad_norm": 0.5094394151644369, "learning_rate": 9.942043097900358e-05, "loss": 3.4529, "step": 6943 }, { "epoch": 0.4310633807188528, "grad_norm": 0.6361867167959612, "learning_rate": 9.941988255932736e-05, "loss": 3.5318, "step": 6944 }, { "epoch": 0.4311254578186107, "grad_norm": 0.7132713241972859, "learning_rate": 9.941933388181562e-05, "loss": 3.4009, "step": 6945 }, { "epoch": 0.4311875349183686, "grad_norm": 0.630554432583209, "learning_rate": 9.941878494647121e-05, "loss": 3.3995, "step": 6946 }, { "epoch": 0.4312496120181265, "grad_norm": 0.531372617146837, "learning_rate": 9.9418235753297e-05, "loss": 3.4116, "step": 6947 }, { "epoch": 0.4313116891178844, "grad_norm": 0.47293612178270467, "learning_rate": 9.941768630229586e-05, "loss": 3.4142, "step": 6948 }, { "epoch": 0.4313737662176423, "grad_norm": 0.6274566054822482, "learning_rate": 9.941713659347067e-05, "loss": 3.3989, "step": 6949 }, { "epoch": 0.4314358433174002, "grad_norm": 0.6798840819459027, "learning_rate": 9.941658662682426e-05, "loss": 3.3342, "step": 6950 }, { "epoch": 0.4314979204171581, "grad_norm": 0.6246657908466405, "learning_rate": 9.941603640235952e-05, "loss": 3.351, "step": 6951 }, { "epoch": 0.431559997516916, "grad_norm": 0.5307383036427489, "learning_rate": 9.941548592007934e-05, "loss": 3.3621, "step": 6952 }, { "epoch": 0.4316220746166739, "grad_norm": 0.5799878239586783, "learning_rate": 9.941493517998657e-05, "loss": 3.3747, "step": 6953 }, { "epoch": 0.4316841517164318, "grad_norm": 0.6882249088154682, "learning_rate": 9.941438418208407e-05, "loss": 3.4049, "step": 6954 }, { "epoch": 0.4317462288161897, "grad_norm": 0.6596852038409933, "learning_rate": 9.941383292637475e-05, "loss": 3.3912, "step": 6955 }, { "epoch": 0.4318083059159476, "grad_norm": 0.5013708455539668, "learning_rate": 9.941328141286146e-05, "loss": 3.3197, "step": 6956 }, { "epoch": 0.4318703830157055, "grad_norm": 0.6313710219447294, "learning_rate": 9.941272964154708e-05, "loss": 3.3539, "step": 6957 }, { "epoch": 0.4319324601154634, "grad_norm": 0.6867894488885548, "learning_rate": 9.941217761243451e-05, "loss": 3.4335, "step": 6958 }, { "epoch": 0.4319945372152213, "grad_norm": 0.6451091141041096, "learning_rate": 9.941162532552662e-05, "loss": 3.3768, "step": 6959 }, { "epoch": 0.4320566143149792, "grad_norm": 0.7291022590165857, "learning_rate": 9.941107278082626e-05, "loss": 3.487, "step": 6960 }, { "epoch": 0.4321186914147371, "grad_norm": 0.6796407793145154, "learning_rate": 9.941051997833635e-05, "loss": 3.3786, "step": 6961 }, { "epoch": 0.432180768514495, "grad_norm": 0.5263153786720379, "learning_rate": 9.940996691805978e-05, "loss": 3.3738, "step": 6962 }, { "epoch": 0.4322428456142529, "grad_norm": 0.6902450249826658, "learning_rate": 9.940941359999938e-05, "loss": 3.3855, "step": 6963 }, { "epoch": 0.4323049227140108, "grad_norm": 0.6246209664190695, "learning_rate": 9.94088600241581e-05, "loss": 3.3999, "step": 6964 }, { "epoch": 0.4323669998137687, "grad_norm": 0.5738640552591328, "learning_rate": 9.94083061905388e-05, "loss": 3.4395, "step": 6965 }, { "epoch": 0.4324290769135266, "grad_norm": 0.6255199891867382, "learning_rate": 9.940775209914438e-05, "loss": 3.3844, "step": 6966 }, { "epoch": 0.4324911540132845, "grad_norm": 0.6144675668424607, "learning_rate": 9.94071977499777e-05, "loss": 3.4125, "step": 6967 }, { "epoch": 0.4325532311130424, "grad_norm": 0.558456910985562, "learning_rate": 9.940664314304169e-05, "loss": 3.35, "step": 6968 }, { "epoch": 0.4326153082128003, "grad_norm": 0.607027422525915, "learning_rate": 9.940608827833923e-05, "loss": 3.3391, "step": 6969 }, { "epoch": 0.43267738531255817, "grad_norm": 0.7392996973949109, "learning_rate": 9.940553315587321e-05, "loss": 3.4473, "step": 6970 }, { "epoch": 0.4327394624123161, "grad_norm": 0.7271167732863296, "learning_rate": 9.940497777564653e-05, "loss": 3.3172, "step": 6971 }, { "epoch": 0.432801539512074, "grad_norm": 0.6277446088449254, "learning_rate": 9.940442213766209e-05, "loss": 3.3619, "step": 6972 }, { "epoch": 0.43286361661183187, "grad_norm": 0.5081917740230777, "learning_rate": 9.940386624192278e-05, "loss": 3.3686, "step": 6973 }, { "epoch": 0.4329256937115898, "grad_norm": 0.5523912422880902, "learning_rate": 9.94033100884315e-05, "loss": 3.3657, "step": 6974 }, { "epoch": 0.4329877708113477, "grad_norm": 0.48348725398856335, "learning_rate": 9.940275367719118e-05, "loss": 3.403, "step": 6975 }, { "epoch": 0.43304984791110557, "grad_norm": 0.4361601817951502, "learning_rate": 9.940219700820471e-05, "loss": 3.3126, "step": 6976 }, { "epoch": 0.4331119250108635, "grad_norm": 0.4838568007442305, "learning_rate": 9.940164008147494e-05, "loss": 3.3992, "step": 6977 }, { "epoch": 0.4331740021106214, "grad_norm": 0.5262633099084713, "learning_rate": 9.940108289700486e-05, "loss": 3.4284, "step": 6978 }, { "epoch": 0.43323607921037927, "grad_norm": 0.49030432638986154, "learning_rate": 9.940052545479733e-05, "loss": 3.2952, "step": 6979 }, { "epoch": 0.4332981563101372, "grad_norm": 0.472501366830292, "learning_rate": 9.939996775485524e-05, "loss": 3.5094, "step": 6980 }, { "epoch": 0.4333602334098951, "grad_norm": 0.5718733188208378, "learning_rate": 9.939940979718157e-05, "loss": 3.3707, "step": 6981 }, { "epoch": 0.43342231050965296, "grad_norm": 0.5262395247461122, "learning_rate": 9.939885158177915e-05, "loss": 3.2889, "step": 6982 }, { "epoch": 0.4334843876094109, "grad_norm": 0.5131171244250377, "learning_rate": 9.939829310865096e-05, "loss": 3.4219, "step": 6983 }, { "epoch": 0.4335464647091688, "grad_norm": 0.48681354426209406, "learning_rate": 9.939773437779986e-05, "loss": 3.4129, "step": 6984 }, { "epoch": 0.43360854180892666, "grad_norm": 0.5384232974490148, "learning_rate": 9.939717538922879e-05, "loss": 3.3659, "step": 6985 }, { "epoch": 0.4336706189086846, "grad_norm": 0.4617272577938088, "learning_rate": 9.939661614294067e-05, "loss": 3.3409, "step": 6986 }, { "epoch": 0.4337326960084425, "grad_norm": 0.5597199777437722, "learning_rate": 9.939605663893842e-05, "loss": 3.4403, "step": 6987 }, { "epoch": 0.43379477310820036, "grad_norm": 0.5508135522338693, "learning_rate": 9.939549687722493e-05, "loss": 3.3375, "step": 6988 }, { "epoch": 0.4338568502079583, "grad_norm": 0.5264683274182326, "learning_rate": 9.939493685780317e-05, "loss": 3.3818, "step": 6989 }, { "epoch": 0.4339189273077162, "grad_norm": 0.5063728760312448, "learning_rate": 9.9394376580676e-05, "loss": 3.357, "step": 6990 }, { "epoch": 0.43398100440747406, "grad_norm": 0.5856530579936476, "learning_rate": 9.93938160458464e-05, "loss": 3.3983, "step": 6991 }, { "epoch": 0.434043081507232, "grad_norm": 0.5827261086013267, "learning_rate": 9.939325525331726e-05, "loss": 3.3538, "step": 6992 }, { "epoch": 0.4341051586069899, "grad_norm": 0.5173554427272333, "learning_rate": 9.939269420309154e-05, "loss": 3.3431, "step": 6993 }, { "epoch": 0.43416723570674776, "grad_norm": 0.5164763329485258, "learning_rate": 9.939213289517212e-05, "loss": 3.3917, "step": 6994 }, { "epoch": 0.4342293128065057, "grad_norm": 0.5921676280127224, "learning_rate": 9.939157132956196e-05, "loss": 3.3523, "step": 6995 }, { "epoch": 0.4342913899062636, "grad_norm": 0.6101216608029183, "learning_rate": 9.939100950626398e-05, "loss": 3.2669, "step": 6996 }, { "epoch": 0.43435346700602145, "grad_norm": 0.5330780217892671, "learning_rate": 9.93904474252811e-05, "loss": 3.398, "step": 6997 }, { "epoch": 0.43441554410577937, "grad_norm": 0.5713285253213789, "learning_rate": 9.938988508661628e-05, "loss": 3.4054, "step": 6998 }, { "epoch": 0.4344776212055373, "grad_norm": 0.6021329726201554, "learning_rate": 9.938932249027244e-05, "loss": 3.3879, "step": 6999 }, { "epoch": 0.43453969830529515, "grad_norm": 0.5611189471721616, "learning_rate": 9.938875963625252e-05, "loss": 3.3591, "step": 7000 }, { "epoch": 0.43460177540505307, "grad_norm": 0.5825475412622855, "learning_rate": 9.938819652455943e-05, "loss": 3.3609, "step": 7001 }, { "epoch": 0.434663852504811, "grad_norm": 0.5170549159370458, "learning_rate": 9.938763315519616e-05, "loss": 3.319, "step": 7002 }, { "epoch": 0.43472592960456885, "grad_norm": 0.4395281918307766, "learning_rate": 9.938706952816562e-05, "loss": 3.4102, "step": 7003 }, { "epoch": 0.43478800670432677, "grad_norm": 0.6031497228822765, "learning_rate": 9.938650564347071e-05, "loss": 3.4702, "step": 7004 }, { "epoch": 0.4348500838040847, "grad_norm": 0.569581150752111, "learning_rate": 9.938594150111444e-05, "loss": 3.3248, "step": 7005 }, { "epoch": 0.43491216090384255, "grad_norm": 0.5603265894380558, "learning_rate": 9.938537710109972e-05, "loss": 3.3497, "step": 7006 }, { "epoch": 0.43497423800360047, "grad_norm": 0.5045248271383276, "learning_rate": 9.93848124434295e-05, "loss": 3.3799, "step": 7007 }, { "epoch": 0.4350363151033584, "grad_norm": 0.46053115261377375, "learning_rate": 9.93842475281067e-05, "loss": 3.4423, "step": 7008 }, { "epoch": 0.43509839220311625, "grad_norm": 0.47932129590186956, "learning_rate": 9.938368235513433e-05, "loss": 3.4528, "step": 7009 }, { "epoch": 0.43516046930287416, "grad_norm": 0.5315375391388302, "learning_rate": 9.938311692451528e-05, "loss": 3.3945, "step": 7010 }, { "epoch": 0.4352225464026321, "grad_norm": 0.476616559323801, "learning_rate": 9.938255123625253e-05, "loss": 3.3667, "step": 7011 }, { "epoch": 0.43528462350238994, "grad_norm": 0.5206252318126442, "learning_rate": 9.938198529034901e-05, "loss": 3.4182, "step": 7012 }, { "epoch": 0.43534670060214786, "grad_norm": 0.5510770125371516, "learning_rate": 9.938141908680769e-05, "loss": 3.3907, "step": 7013 }, { "epoch": 0.4354087777019058, "grad_norm": 0.5177976034965031, "learning_rate": 9.938085262563153e-05, "loss": 3.446, "step": 7014 }, { "epoch": 0.43547085480166364, "grad_norm": 0.5341421947537084, "learning_rate": 9.938028590682347e-05, "loss": 3.3357, "step": 7015 }, { "epoch": 0.43553293190142156, "grad_norm": 0.5582051292474215, "learning_rate": 9.937971893038647e-05, "loss": 3.3615, "step": 7016 }, { "epoch": 0.4355950090011795, "grad_norm": 0.5355960681450461, "learning_rate": 9.937915169632348e-05, "loss": 3.3456, "step": 7017 }, { "epoch": 0.43565708610093734, "grad_norm": 0.5236717383242108, "learning_rate": 9.937858420463748e-05, "loss": 3.427, "step": 7018 }, { "epoch": 0.43571916320069526, "grad_norm": 0.6074417209143008, "learning_rate": 9.93780164553314e-05, "loss": 3.3493, "step": 7019 }, { "epoch": 0.4357812403004532, "grad_norm": 0.5246419855294374, "learning_rate": 9.937744844840824e-05, "loss": 3.431, "step": 7020 }, { "epoch": 0.43584331740021104, "grad_norm": 0.4549946851734334, "learning_rate": 9.937688018387093e-05, "loss": 3.4105, "step": 7021 }, { "epoch": 0.43590539449996896, "grad_norm": 0.5105191504167851, "learning_rate": 9.937631166172246e-05, "loss": 3.3318, "step": 7022 }, { "epoch": 0.4359674715997269, "grad_norm": 0.5924670764773303, "learning_rate": 9.937574288196578e-05, "loss": 3.466, "step": 7023 }, { "epoch": 0.43602954869948474, "grad_norm": 0.595555546725073, "learning_rate": 9.937517384460388e-05, "loss": 3.4064, "step": 7024 }, { "epoch": 0.43609162579924265, "grad_norm": 0.5804825987774691, "learning_rate": 9.937460454963969e-05, "loss": 3.3713, "step": 7025 }, { "epoch": 0.4361537028990006, "grad_norm": 0.5611417318897283, "learning_rate": 9.937403499707622e-05, "loss": 3.4216, "step": 7026 }, { "epoch": 0.43621577999875844, "grad_norm": 0.5201512702664566, "learning_rate": 9.937346518691641e-05, "loss": 3.4424, "step": 7027 }, { "epoch": 0.43627785709851635, "grad_norm": 0.5394968411509582, "learning_rate": 9.937289511916325e-05, "loss": 3.3509, "step": 7028 }, { "epoch": 0.43633993419827427, "grad_norm": 0.5776508552066162, "learning_rate": 9.93723247938197e-05, "loss": 3.3899, "step": 7029 }, { "epoch": 0.43640201129803213, "grad_norm": 0.5437637656212316, "learning_rate": 9.937175421088876e-05, "loss": 3.3026, "step": 7030 }, { "epoch": 0.43646408839779005, "grad_norm": 0.5558592012878275, "learning_rate": 9.93711833703734e-05, "loss": 3.3138, "step": 7031 }, { "epoch": 0.43652616549754797, "grad_norm": 0.5980494084918119, "learning_rate": 9.937061227227657e-05, "loss": 3.4113, "step": 7032 }, { "epoch": 0.43658824259730583, "grad_norm": 0.6375771101896897, "learning_rate": 9.937004091660129e-05, "loss": 3.3354, "step": 7033 }, { "epoch": 0.43665031969706375, "grad_norm": 0.6137847291556376, "learning_rate": 9.936946930335051e-05, "loss": 3.3475, "step": 7034 }, { "epoch": 0.43671239679682167, "grad_norm": 0.5505345380431221, "learning_rate": 9.936889743252723e-05, "loss": 3.4404, "step": 7035 }, { "epoch": 0.43677447389657953, "grad_norm": 0.5859858429357394, "learning_rate": 9.936832530413442e-05, "loss": 3.4965, "step": 7036 }, { "epoch": 0.43683655099633745, "grad_norm": 0.608799271165998, "learning_rate": 9.936775291817508e-05, "loss": 3.2952, "step": 7037 }, { "epoch": 0.43689862809609536, "grad_norm": 0.5216117446169565, "learning_rate": 9.936718027465219e-05, "loss": 3.3752, "step": 7038 }, { "epoch": 0.4369607051958532, "grad_norm": 0.5029907782033495, "learning_rate": 9.936660737356873e-05, "loss": 3.435, "step": 7039 }, { "epoch": 0.43702278229561115, "grad_norm": 0.48986020950334397, "learning_rate": 9.93660342149277e-05, "loss": 3.3376, "step": 7040 }, { "epoch": 0.43708485939536906, "grad_norm": 0.5100642531722063, "learning_rate": 9.936546079873208e-05, "loss": 3.3959, "step": 7041 }, { "epoch": 0.4371469364951269, "grad_norm": 0.4895640041377399, "learning_rate": 9.936488712498486e-05, "loss": 3.2591, "step": 7042 }, { "epoch": 0.43720901359488484, "grad_norm": 0.5366893999051221, "learning_rate": 9.936431319368905e-05, "loss": 3.4213, "step": 7043 }, { "epoch": 0.43727109069464276, "grad_norm": 0.5157305002654365, "learning_rate": 9.936373900484764e-05, "loss": 3.3776, "step": 7044 }, { "epoch": 0.4373331677944006, "grad_norm": 0.4995766277780564, "learning_rate": 9.936316455846362e-05, "loss": 3.3096, "step": 7045 }, { "epoch": 0.43739524489415854, "grad_norm": 0.49749156436765224, "learning_rate": 9.936258985453997e-05, "loss": 3.4297, "step": 7046 }, { "epoch": 0.43745732199391646, "grad_norm": 0.5162650384167359, "learning_rate": 9.936201489307972e-05, "loss": 3.3733, "step": 7047 }, { "epoch": 0.4375193990936743, "grad_norm": 0.6048058849831144, "learning_rate": 9.936143967408585e-05, "loss": 3.332, "step": 7048 }, { "epoch": 0.43758147619343224, "grad_norm": 0.595494568384416, "learning_rate": 9.936086419756137e-05, "loss": 3.2974, "step": 7049 }, { "epoch": 0.43764355329319016, "grad_norm": 0.5546330707333386, "learning_rate": 9.936028846350928e-05, "loss": 3.3316, "step": 7050 }, { "epoch": 0.437705630392948, "grad_norm": 0.5921921118403193, "learning_rate": 9.935971247193257e-05, "loss": 3.3725, "step": 7051 }, { "epoch": 0.43776770749270594, "grad_norm": 0.5714059563402105, "learning_rate": 9.935913622283426e-05, "loss": 3.2979, "step": 7052 }, { "epoch": 0.43782978459246386, "grad_norm": 0.575968276436734, "learning_rate": 9.935855971621736e-05, "loss": 3.4646, "step": 7053 }, { "epoch": 0.4378918616922217, "grad_norm": 0.5909164476603604, "learning_rate": 9.935798295208488e-05, "loss": 3.3138, "step": 7054 }, { "epoch": 0.43795393879197964, "grad_norm": 0.5756793745952393, "learning_rate": 9.935740593043982e-05, "loss": 3.3859, "step": 7055 }, { "epoch": 0.43801601589173755, "grad_norm": 0.5281350591931743, "learning_rate": 9.935682865128518e-05, "loss": 3.3854, "step": 7056 }, { "epoch": 0.4380780929914954, "grad_norm": 0.5200614231769172, "learning_rate": 9.935625111462398e-05, "loss": 3.3383, "step": 7057 }, { "epoch": 0.43814017009125333, "grad_norm": 0.4898269145451347, "learning_rate": 9.935567332045926e-05, "loss": 3.4315, "step": 7058 }, { "epoch": 0.43820224719101125, "grad_norm": 0.493508730259425, "learning_rate": 9.935509526879398e-05, "loss": 3.4691, "step": 7059 }, { "epoch": 0.4382643242907691, "grad_norm": 0.543589120940005, "learning_rate": 9.93545169596312e-05, "loss": 3.3915, "step": 7060 }, { "epoch": 0.43832640139052703, "grad_norm": 0.5152354936681581, "learning_rate": 9.935393839297393e-05, "loss": 3.2704, "step": 7061 }, { "epoch": 0.43838847849028495, "grad_norm": 0.45720475318774956, "learning_rate": 9.935335956882519e-05, "loss": 3.2553, "step": 7062 }, { "epoch": 0.4384505555900428, "grad_norm": 0.5449776022797099, "learning_rate": 9.935278048718797e-05, "loss": 3.3491, "step": 7063 }, { "epoch": 0.43851263268980073, "grad_norm": 0.7355585098359336, "learning_rate": 9.935220114806533e-05, "loss": 3.3137, "step": 7064 }, { "epoch": 0.43857470978955865, "grad_norm": 0.7847252037381796, "learning_rate": 9.935162155146028e-05, "loss": 3.3703, "step": 7065 }, { "epoch": 0.4386367868893165, "grad_norm": 0.6961138318865248, "learning_rate": 9.935104169737581e-05, "loss": 3.3989, "step": 7066 }, { "epoch": 0.43869886398907443, "grad_norm": 0.5787164652441635, "learning_rate": 9.9350461585815e-05, "loss": 3.2628, "step": 7067 }, { "epoch": 0.43876094108883235, "grad_norm": 0.6034792052702551, "learning_rate": 9.934988121678083e-05, "loss": 3.3376, "step": 7068 }, { "epoch": 0.4388230181885902, "grad_norm": 0.5614254542568952, "learning_rate": 9.934930059027638e-05, "loss": 3.4591, "step": 7069 }, { "epoch": 0.4388850952883481, "grad_norm": 0.5962036281523286, "learning_rate": 9.934871970630464e-05, "loss": 3.3556, "step": 7070 }, { "epoch": 0.43894717238810604, "grad_norm": 0.5268325305714576, "learning_rate": 9.934813856486864e-05, "loss": 3.4128, "step": 7071 }, { "epoch": 0.4390092494878639, "grad_norm": 0.6046605776922189, "learning_rate": 9.934755716597142e-05, "loss": 3.3864, "step": 7072 }, { "epoch": 0.4390713265876218, "grad_norm": 0.5912191762813271, "learning_rate": 9.934697550961602e-05, "loss": 3.3721, "step": 7073 }, { "epoch": 0.43913340368737974, "grad_norm": 0.5560336635247574, "learning_rate": 9.934639359580548e-05, "loss": 3.3909, "step": 7074 }, { "epoch": 0.4391954807871376, "grad_norm": 0.5024713890983361, "learning_rate": 9.93458114245428e-05, "loss": 3.4658, "step": 7075 }, { "epoch": 0.4392575578868955, "grad_norm": 0.5607929819261076, "learning_rate": 9.934522899583106e-05, "loss": 3.3169, "step": 7076 }, { "epoch": 0.43931963498665344, "grad_norm": 0.5109324930394841, "learning_rate": 9.934464630967328e-05, "loss": 3.3265, "step": 7077 }, { "epoch": 0.4393817120864113, "grad_norm": 0.5408131464975597, "learning_rate": 9.934406336607251e-05, "loss": 3.2293, "step": 7078 }, { "epoch": 0.4394437891861692, "grad_norm": 0.48437526149127424, "learning_rate": 9.934348016503176e-05, "loss": 3.4965, "step": 7079 }, { "epoch": 0.43950586628592714, "grad_norm": 0.49100834515621655, "learning_rate": 9.934289670655413e-05, "loss": 3.2919, "step": 7080 }, { "epoch": 0.439567943385685, "grad_norm": 0.4784866947070132, "learning_rate": 9.93423129906426e-05, "loss": 3.3372, "step": 7081 }, { "epoch": 0.4396300204854429, "grad_norm": 0.49950954104906226, "learning_rate": 9.934172901730025e-05, "loss": 3.4394, "step": 7082 }, { "epoch": 0.43969209758520084, "grad_norm": 0.5296471763273408, "learning_rate": 9.934114478653012e-05, "loss": 3.4584, "step": 7083 }, { "epoch": 0.4397541746849587, "grad_norm": 0.5751003582075235, "learning_rate": 9.934056029833526e-05, "loss": 3.3568, "step": 7084 }, { "epoch": 0.4398162517847166, "grad_norm": 0.5492547165881285, "learning_rate": 9.933997555271872e-05, "loss": 3.2967, "step": 7085 }, { "epoch": 0.43987832888447453, "grad_norm": 0.5161511886375315, "learning_rate": 9.933939054968355e-05, "loss": 3.3289, "step": 7086 }, { "epoch": 0.4399404059842324, "grad_norm": 0.5401049775497937, "learning_rate": 9.93388052892328e-05, "loss": 3.3794, "step": 7087 }, { "epoch": 0.4400024830839903, "grad_norm": 0.48966067967975696, "learning_rate": 9.933821977136955e-05, "loss": 3.343, "step": 7088 }, { "epoch": 0.44006456018374823, "grad_norm": 0.4968301009039205, "learning_rate": 9.93376339960968e-05, "loss": 3.3001, "step": 7089 }, { "epoch": 0.4401266372835061, "grad_norm": 0.5099718108053481, "learning_rate": 9.933704796341765e-05, "loss": 3.2928, "step": 7090 }, { "epoch": 0.440188714383264, "grad_norm": 0.5515288021178788, "learning_rate": 9.933646167333514e-05, "loss": 3.337, "step": 7091 }, { "epoch": 0.44025079148302193, "grad_norm": 0.6210480439264856, "learning_rate": 9.933587512585234e-05, "loss": 3.3924, "step": 7092 }, { "epoch": 0.4403128685827798, "grad_norm": 0.5501146272939537, "learning_rate": 9.933528832097229e-05, "loss": 3.4166, "step": 7093 }, { "epoch": 0.4403749456825377, "grad_norm": 0.5061112001666509, "learning_rate": 9.933470125869806e-05, "loss": 3.3922, "step": 7094 }, { "epoch": 0.44043702278229563, "grad_norm": 0.48895600045970544, "learning_rate": 9.933411393903274e-05, "loss": 3.3982, "step": 7095 }, { "epoch": 0.4404990998820535, "grad_norm": 0.5572867942906042, "learning_rate": 9.933352636197936e-05, "loss": 3.4052, "step": 7096 }, { "epoch": 0.4405611769818114, "grad_norm": 0.5876005857678968, "learning_rate": 9.933293852754099e-05, "loss": 3.2997, "step": 7097 }, { "epoch": 0.4406232540815693, "grad_norm": 0.5700275283992297, "learning_rate": 9.933235043572072e-05, "loss": 3.3646, "step": 7098 }, { "epoch": 0.4406853311813272, "grad_norm": 0.5042485693408678, "learning_rate": 9.933176208652158e-05, "loss": 3.421, "step": 7099 }, { "epoch": 0.4407474082810851, "grad_norm": 0.5475405321838331, "learning_rate": 9.933117347994668e-05, "loss": 3.3039, "step": 7100 }, { "epoch": 0.440809485380843, "grad_norm": 0.5169781932658953, "learning_rate": 9.933058461599906e-05, "loss": 3.3505, "step": 7101 }, { "epoch": 0.4408715624806009, "grad_norm": 0.544111681823599, "learning_rate": 9.93299954946818e-05, "loss": 3.4654, "step": 7102 }, { "epoch": 0.4409336395803588, "grad_norm": 0.5208742666053964, "learning_rate": 9.932940611599799e-05, "loss": 3.3312, "step": 7103 }, { "epoch": 0.4409957166801167, "grad_norm": 0.5856044331280527, "learning_rate": 9.932881647995069e-05, "loss": 3.3878, "step": 7104 }, { "epoch": 0.4410577937798746, "grad_norm": 0.6062603113929594, "learning_rate": 9.932822658654298e-05, "loss": 3.4243, "step": 7105 }, { "epoch": 0.4411198708796325, "grad_norm": 0.5131906786573852, "learning_rate": 9.932763643577793e-05, "loss": 3.3549, "step": 7106 }, { "epoch": 0.4411819479793904, "grad_norm": 0.5545638846615818, "learning_rate": 9.932704602765865e-05, "loss": 3.2763, "step": 7107 }, { "epoch": 0.4412440250791483, "grad_norm": 0.5892426415179389, "learning_rate": 9.932645536218817e-05, "loss": 3.3772, "step": 7108 }, { "epoch": 0.4413061021789062, "grad_norm": 0.5441291813355642, "learning_rate": 9.932586443936962e-05, "loss": 3.332, "step": 7109 }, { "epoch": 0.4413681792786641, "grad_norm": 0.47142860885564236, "learning_rate": 9.932527325920604e-05, "loss": 3.3696, "step": 7110 }, { "epoch": 0.441430256378422, "grad_norm": 0.550809737747107, "learning_rate": 9.932468182170054e-05, "loss": 3.4032, "step": 7111 }, { "epoch": 0.4414923334781799, "grad_norm": 0.5613860385143322, "learning_rate": 9.932409012685621e-05, "loss": 3.4139, "step": 7112 }, { "epoch": 0.4415544105779378, "grad_norm": 0.4944410051923906, "learning_rate": 9.932349817467612e-05, "loss": 3.2572, "step": 7113 }, { "epoch": 0.4416164876776957, "grad_norm": 0.5106205403332801, "learning_rate": 9.932290596516337e-05, "loss": 3.2318, "step": 7114 }, { "epoch": 0.4416785647774536, "grad_norm": 0.48552000933565437, "learning_rate": 9.932231349832103e-05, "loss": 3.3337, "step": 7115 }, { "epoch": 0.4417406418772115, "grad_norm": 0.5379185078582259, "learning_rate": 9.932172077415224e-05, "loss": 3.3843, "step": 7116 }, { "epoch": 0.4418027189769694, "grad_norm": 0.5581308594382216, "learning_rate": 9.932112779266005e-05, "loss": 3.4413, "step": 7117 }, { "epoch": 0.4418647960767273, "grad_norm": 0.5389135956699087, "learning_rate": 9.932053455384755e-05, "loss": 3.3384, "step": 7118 }, { "epoch": 0.4419268731764852, "grad_norm": 0.49827408937062195, "learning_rate": 9.931994105771785e-05, "loss": 3.4273, "step": 7119 }, { "epoch": 0.4419889502762431, "grad_norm": 0.5748949027418248, "learning_rate": 9.931934730427405e-05, "loss": 3.2756, "step": 7120 }, { "epoch": 0.442051027376001, "grad_norm": 0.5080271340655323, "learning_rate": 9.931875329351922e-05, "loss": 3.3387, "step": 7121 }, { "epoch": 0.4421131044757589, "grad_norm": 0.5042812814820665, "learning_rate": 9.931815902545651e-05, "loss": 3.3682, "step": 7122 }, { "epoch": 0.4421751815755168, "grad_norm": 0.5117994637934092, "learning_rate": 9.931756450008899e-05, "loss": 3.4386, "step": 7123 }, { "epoch": 0.4422372586752747, "grad_norm": 0.5982049156382367, "learning_rate": 9.931696971741976e-05, "loss": 3.4865, "step": 7124 }, { "epoch": 0.4422993357750326, "grad_norm": 0.6064128283463552, "learning_rate": 9.931637467745192e-05, "loss": 3.324, "step": 7125 }, { "epoch": 0.44236141287479047, "grad_norm": 0.583756801189853, "learning_rate": 9.93157793801886e-05, "loss": 3.2687, "step": 7126 }, { "epoch": 0.4424234899745484, "grad_norm": 0.6265024484600255, "learning_rate": 9.931518382563286e-05, "loss": 3.4561, "step": 7127 }, { "epoch": 0.4424855670743063, "grad_norm": 0.6539100984120579, "learning_rate": 9.931458801378784e-05, "loss": 3.3665, "step": 7128 }, { "epoch": 0.44254764417406417, "grad_norm": 0.5449387469123744, "learning_rate": 9.931399194465665e-05, "loss": 3.2845, "step": 7129 }, { "epoch": 0.4426097212738221, "grad_norm": 0.4946657940891355, "learning_rate": 9.93133956182424e-05, "loss": 3.3793, "step": 7130 }, { "epoch": 0.44267179837358, "grad_norm": 0.5733597746932262, "learning_rate": 9.931279903454819e-05, "loss": 3.3509, "step": 7131 }, { "epoch": 0.44273387547333787, "grad_norm": 0.520444798188972, "learning_rate": 9.931220219357714e-05, "loss": 3.4077, "step": 7132 }, { "epoch": 0.4427959525730958, "grad_norm": 0.517888015968379, "learning_rate": 9.931160509533235e-05, "loss": 3.452, "step": 7133 }, { "epoch": 0.4428580296728537, "grad_norm": 0.5273840429665912, "learning_rate": 9.931100773981694e-05, "loss": 3.3633, "step": 7134 }, { "epoch": 0.44292010677261157, "grad_norm": 0.5088214827329052, "learning_rate": 9.931041012703404e-05, "loss": 3.3365, "step": 7135 }, { "epoch": 0.4429821838723695, "grad_norm": 0.5506814636308333, "learning_rate": 9.930981225698676e-05, "loss": 3.3547, "step": 7136 }, { "epoch": 0.4430442609721274, "grad_norm": 0.6095390343578491, "learning_rate": 9.930921412967822e-05, "loss": 3.2462, "step": 7137 }, { "epoch": 0.44310633807188526, "grad_norm": 0.5134389951362596, "learning_rate": 9.930861574511155e-05, "loss": 3.3185, "step": 7138 }, { "epoch": 0.4431684151716432, "grad_norm": 0.5091885921868771, "learning_rate": 9.930801710328985e-05, "loss": 3.309, "step": 7139 }, { "epoch": 0.4432304922714011, "grad_norm": 0.6145634701854132, "learning_rate": 9.930741820421626e-05, "loss": 3.426, "step": 7140 }, { "epoch": 0.44329256937115896, "grad_norm": 0.6293635867541484, "learning_rate": 9.930681904789387e-05, "loss": 3.3243, "step": 7141 }, { "epoch": 0.4433546464709169, "grad_norm": 0.5622813806609456, "learning_rate": 9.930621963432586e-05, "loss": 3.2981, "step": 7142 }, { "epoch": 0.4434167235706748, "grad_norm": 0.552471111672972, "learning_rate": 9.930561996351534e-05, "loss": 3.3868, "step": 7143 }, { "epoch": 0.44347880067043266, "grad_norm": 0.5456722384581786, "learning_rate": 9.930502003546542e-05, "loss": 3.4239, "step": 7144 }, { "epoch": 0.4435408777701906, "grad_norm": 0.5576448984826511, "learning_rate": 9.930441985017925e-05, "loss": 3.3858, "step": 7145 }, { "epoch": 0.4436029548699485, "grad_norm": 0.6415605459355374, "learning_rate": 9.930381940765995e-05, "loss": 3.4145, "step": 7146 }, { "epoch": 0.44366503196970636, "grad_norm": 0.6325275344660352, "learning_rate": 9.930321870791065e-05, "loss": 3.4305, "step": 7147 }, { "epoch": 0.4437271090694643, "grad_norm": 0.5820709573398984, "learning_rate": 9.93026177509345e-05, "loss": 3.3708, "step": 7148 }, { "epoch": 0.4437891861692222, "grad_norm": 0.6666842150365403, "learning_rate": 9.93020165367346e-05, "loss": 3.4136, "step": 7149 }, { "epoch": 0.44385126326898006, "grad_norm": 0.5491234198759614, "learning_rate": 9.930141506531414e-05, "loss": 3.2575, "step": 7150 }, { "epoch": 0.443913340368738, "grad_norm": 0.5874592158182441, "learning_rate": 9.930081333667622e-05, "loss": 3.4175, "step": 7151 }, { "epoch": 0.4439754174684959, "grad_norm": 0.6335759678989976, "learning_rate": 9.9300211350824e-05, "loss": 3.3367, "step": 7152 }, { "epoch": 0.44403749456825375, "grad_norm": 0.5390284707206019, "learning_rate": 9.929960910776057e-05, "loss": 3.2543, "step": 7153 }, { "epoch": 0.4440995716680117, "grad_norm": 0.5547268275973549, "learning_rate": 9.929900660748915e-05, "loss": 3.3568, "step": 7154 }, { "epoch": 0.4441616487677696, "grad_norm": 0.6407560121504986, "learning_rate": 9.929840385001284e-05, "loss": 3.4169, "step": 7155 }, { "epoch": 0.44422372586752745, "grad_norm": 0.7486938387994699, "learning_rate": 9.929780083533477e-05, "loss": 3.3487, "step": 7156 }, { "epoch": 0.44428580296728537, "grad_norm": 0.7575288221963108, "learning_rate": 9.929719756345813e-05, "loss": 3.4413, "step": 7157 }, { "epoch": 0.4443478800670433, "grad_norm": 0.697956800493026, "learning_rate": 9.929659403438602e-05, "loss": 3.4518, "step": 7158 }, { "epoch": 0.44440995716680115, "grad_norm": 0.7280988549380525, "learning_rate": 9.929599024812162e-05, "loss": 3.3512, "step": 7159 }, { "epoch": 0.44447203426655907, "grad_norm": 0.7985270709355868, "learning_rate": 9.929538620466808e-05, "loss": 3.3758, "step": 7160 }, { "epoch": 0.444534111366317, "grad_norm": 0.6164364893159584, "learning_rate": 9.929478190402852e-05, "loss": 3.3208, "step": 7161 }, { "epoch": 0.44459618846607485, "grad_norm": 0.5735012944048082, "learning_rate": 9.929417734620614e-05, "loss": 3.3989, "step": 7162 }, { "epoch": 0.44465826556583277, "grad_norm": 0.6309737583075181, "learning_rate": 9.929357253120406e-05, "loss": 3.3879, "step": 7163 }, { "epoch": 0.4447203426655907, "grad_norm": 0.5840697975580441, "learning_rate": 9.929296745902546e-05, "loss": 3.3588, "step": 7164 }, { "epoch": 0.44478241976534855, "grad_norm": 0.5384496439285511, "learning_rate": 9.929236212967345e-05, "loss": 3.3817, "step": 7165 }, { "epoch": 0.44484449686510646, "grad_norm": 0.5768970903938517, "learning_rate": 9.929175654315125e-05, "loss": 3.3928, "step": 7166 }, { "epoch": 0.4449065739648644, "grad_norm": 0.6105321754252907, "learning_rate": 9.929115069946198e-05, "loss": 3.298, "step": 7167 }, { "epoch": 0.44496865106462224, "grad_norm": 0.6003081136579979, "learning_rate": 9.929054459860881e-05, "loss": 3.3037, "step": 7168 }, { "epoch": 0.44503072816438016, "grad_norm": 0.4586498096261066, "learning_rate": 9.928993824059492e-05, "loss": 3.3962, "step": 7169 }, { "epoch": 0.4450928052641381, "grad_norm": 0.5703057066630398, "learning_rate": 9.928933162542343e-05, "loss": 3.3233, "step": 7170 }, { "epoch": 0.44515488236389594, "grad_norm": 0.6078974588439259, "learning_rate": 9.928872475309754e-05, "loss": 3.4462, "step": 7171 }, { "epoch": 0.44521695946365386, "grad_norm": 0.6633334702063504, "learning_rate": 9.928811762362043e-05, "loss": 3.4287, "step": 7172 }, { "epoch": 0.4452790365634118, "grad_norm": 0.6064488999019548, "learning_rate": 9.928751023699522e-05, "loss": 3.346, "step": 7173 }, { "epoch": 0.44534111366316964, "grad_norm": 0.5255311118696964, "learning_rate": 9.92869025932251e-05, "loss": 3.2373, "step": 7174 }, { "epoch": 0.44540319076292756, "grad_norm": 0.570400832141962, "learning_rate": 9.928629469231326e-05, "loss": 3.3437, "step": 7175 }, { "epoch": 0.4454652678626855, "grad_norm": 0.5726609931114458, "learning_rate": 9.928568653426284e-05, "loss": 3.3404, "step": 7176 }, { "epoch": 0.44552734496244334, "grad_norm": 0.5522012734543881, "learning_rate": 9.928507811907705e-05, "loss": 3.3924, "step": 7177 }, { "epoch": 0.44558942206220126, "grad_norm": 0.6380381575327877, "learning_rate": 9.928446944675905e-05, "loss": 3.3068, "step": 7178 }, { "epoch": 0.4456514991619592, "grad_norm": 0.6331617780737528, "learning_rate": 9.928386051731198e-05, "loss": 3.402, "step": 7179 }, { "epoch": 0.44571357626171704, "grad_norm": 0.46355583583592797, "learning_rate": 9.928325133073906e-05, "loss": 3.4094, "step": 7180 }, { "epoch": 0.44577565336147496, "grad_norm": 0.6156759221660573, "learning_rate": 9.928264188704345e-05, "loss": 3.396, "step": 7181 }, { "epoch": 0.4458377304612329, "grad_norm": 0.6719733986769374, "learning_rate": 9.928203218622834e-05, "loss": 3.3493, "step": 7182 }, { "epoch": 0.44589980756099074, "grad_norm": 0.5802840050525048, "learning_rate": 9.92814222282969e-05, "loss": 3.3427, "step": 7183 }, { "epoch": 0.44596188466074865, "grad_norm": 0.5369070885941852, "learning_rate": 9.92808120132523e-05, "loss": 3.2662, "step": 7184 }, { "epoch": 0.44602396176050657, "grad_norm": 0.5796567111822996, "learning_rate": 9.928020154109776e-05, "loss": 3.4107, "step": 7185 }, { "epoch": 0.44608603886026443, "grad_norm": 0.5719929328595253, "learning_rate": 9.927959081183645e-05, "loss": 3.2453, "step": 7186 }, { "epoch": 0.44614811596002235, "grad_norm": 0.6501307383516732, "learning_rate": 9.927897982547154e-05, "loss": 3.3367, "step": 7187 }, { "epoch": 0.44621019305978027, "grad_norm": 0.6397272891298867, "learning_rate": 9.927836858200623e-05, "loss": 3.3569, "step": 7188 }, { "epoch": 0.44627227015953813, "grad_norm": 0.5314045008221456, "learning_rate": 9.927775708144371e-05, "loss": 3.3287, "step": 7189 }, { "epoch": 0.44633434725929605, "grad_norm": 0.5728396768569248, "learning_rate": 9.927714532378718e-05, "loss": 3.3108, "step": 7190 }, { "epoch": 0.44639642435905397, "grad_norm": 0.5896521756134874, "learning_rate": 9.927653330903981e-05, "loss": 3.3558, "step": 7191 }, { "epoch": 0.44645850145881183, "grad_norm": 0.5272871764177869, "learning_rate": 9.92759210372048e-05, "loss": 3.2683, "step": 7192 }, { "epoch": 0.44652057855856975, "grad_norm": 0.5015633104739667, "learning_rate": 9.927530850828535e-05, "loss": 3.3539, "step": 7193 }, { "epoch": 0.44658265565832767, "grad_norm": 0.5194659945012093, "learning_rate": 9.927469572228465e-05, "loss": 3.4281, "step": 7194 }, { "epoch": 0.4466447327580855, "grad_norm": 0.5018878107585526, "learning_rate": 9.92740826792059e-05, "loss": 3.3773, "step": 7195 }, { "epoch": 0.44670680985784345, "grad_norm": 0.5295773146812811, "learning_rate": 9.92734693790523e-05, "loss": 3.3076, "step": 7196 }, { "epoch": 0.44676888695760136, "grad_norm": 0.5481209865932053, "learning_rate": 9.927285582182707e-05, "loss": 3.3181, "step": 7197 }, { "epoch": 0.4468309640573592, "grad_norm": 0.46160707494835884, "learning_rate": 9.927224200753338e-05, "loss": 3.3564, "step": 7198 }, { "epoch": 0.44689304115711714, "grad_norm": 0.5491838490234657, "learning_rate": 9.927162793617443e-05, "loss": 3.3581, "step": 7199 }, { "epoch": 0.44695511825687506, "grad_norm": 0.5128337854760059, "learning_rate": 9.927101360775344e-05, "loss": 3.4132, "step": 7200 }, { "epoch": 0.4470171953566329, "grad_norm": 0.5440118772914105, "learning_rate": 9.927039902227361e-05, "loss": 3.3215, "step": 7201 }, { "epoch": 0.44707927245639084, "grad_norm": 0.5394789704050357, "learning_rate": 9.926978417973815e-05, "loss": 3.314, "step": 7202 }, { "epoch": 0.44714134955614876, "grad_norm": 0.5753936166679058, "learning_rate": 9.926916908015027e-05, "loss": 3.3816, "step": 7203 }, { "epoch": 0.4472034266559066, "grad_norm": 0.5251389081294059, "learning_rate": 9.926855372351315e-05, "loss": 3.4515, "step": 7204 }, { "epoch": 0.44726550375566454, "grad_norm": 0.48330411665130296, "learning_rate": 9.926793810983006e-05, "loss": 3.4308, "step": 7205 }, { "epoch": 0.44732758085542246, "grad_norm": 0.5206601490735415, "learning_rate": 9.926732223910416e-05, "loss": 3.3951, "step": 7206 }, { "epoch": 0.4473896579551803, "grad_norm": 0.49848794293242493, "learning_rate": 9.926670611133868e-05, "loss": 3.3752, "step": 7207 }, { "epoch": 0.44745173505493824, "grad_norm": 0.5432535426326706, "learning_rate": 9.926608972653684e-05, "loss": 3.2955, "step": 7208 }, { "epoch": 0.44751381215469616, "grad_norm": 0.5601495537563795, "learning_rate": 9.926547308470184e-05, "loss": 3.3621, "step": 7209 }, { "epoch": 0.447575889254454, "grad_norm": 0.4627814828188314, "learning_rate": 9.926485618583691e-05, "loss": 3.3415, "step": 7210 }, { "epoch": 0.44763796635421194, "grad_norm": 0.45337771899068247, "learning_rate": 9.926423902994527e-05, "loss": 3.3554, "step": 7211 }, { "epoch": 0.44770004345396985, "grad_norm": 0.47358723843942585, "learning_rate": 9.926362161703014e-05, "loss": 3.4035, "step": 7212 }, { "epoch": 0.4477621205537277, "grad_norm": 0.5338924549297721, "learning_rate": 9.926300394709474e-05, "loss": 3.3787, "step": 7213 }, { "epoch": 0.44782419765348563, "grad_norm": 0.5120601534375873, "learning_rate": 9.926238602014226e-05, "loss": 3.257, "step": 7214 }, { "epoch": 0.44788627475324355, "grad_norm": 0.5386495528876215, "learning_rate": 9.926176783617597e-05, "loss": 3.3904, "step": 7215 }, { "epoch": 0.4479483518530014, "grad_norm": 0.6079673648250803, "learning_rate": 9.926114939519909e-05, "loss": 3.38, "step": 7216 }, { "epoch": 0.44801042895275933, "grad_norm": 0.523335802817542, "learning_rate": 9.926053069721483e-05, "loss": 3.4004, "step": 7217 }, { "epoch": 0.44807250605251725, "grad_norm": 0.5321926001115669, "learning_rate": 9.925991174222641e-05, "loss": 3.3761, "step": 7218 }, { "epoch": 0.4481345831522751, "grad_norm": 0.4779156095475886, "learning_rate": 9.925929253023709e-05, "loss": 3.4505, "step": 7219 }, { "epoch": 0.44819666025203303, "grad_norm": 0.47655888853111766, "learning_rate": 9.925867306125008e-05, "loss": 3.2862, "step": 7220 }, { "epoch": 0.44825873735179095, "grad_norm": 0.5202107968753916, "learning_rate": 9.92580533352686e-05, "loss": 3.3456, "step": 7221 }, { "epoch": 0.4483208144515488, "grad_norm": 0.4910287538167719, "learning_rate": 9.925743335229592e-05, "loss": 3.417, "step": 7222 }, { "epoch": 0.44838289155130673, "grad_norm": 0.5311723820700331, "learning_rate": 9.925681311233523e-05, "loss": 3.2921, "step": 7223 }, { "epoch": 0.44844496865106465, "grad_norm": 0.5495448201325932, "learning_rate": 9.925619261538981e-05, "loss": 3.3515, "step": 7224 }, { "epoch": 0.4485070457508225, "grad_norm": 0.5917866268466649, "learning_rate": 9.925557186146286e-05, "loss": 3.5039, "step": 7225 }, { "epoch": 0.4485691228505804, "grad_norm": 0.5327734980473727, "learning_rate": 9.925495085055764e-05, "loss": 3.348, "step": 7226 }, { "epoch": 0.44863119995033834, "grad_norm": 0.48242505647791717, "learning_rate": 9.925432958267737e-05, "loss": 3.3804, "step": 7227 }, { "epoch": 0.4486932770500962, "grad_norm": 0.5946083640417141, "learning_rate": 9.925370805782532e-05, "loss": 3.4622, "step": 7228 }, { "epoch": 0.4487553541498541, "grad_norm": 0.6841506869099029, "learning_rate": 9.925308627600474e-05, "loss": 3.2446, "step": 7229 }, { "epoch": 0.44881743124961204, "grad_norm": 0.6407104063091278, "learning_rate": 9.925246423721881e-05, "loss": 3.2474, "step": 7230 }, { "epoch": 0.4488795083493699, "grad_norm": 0.48116379374465756, "learning_rate": 9.925184194147085e-05, "loss": 3.3627, "step": 7231 }, { "epoch": 0.4489415854491278, "grad_norm": 0.5102006656303685, "learning_rate": 9.925121938876405e-05, "loss": 3.4045, "step": 7232 }, { "epoch": 0.44900366254888574, "grad_norm": 0.6439453930286361, "learning_rate": 9.92505965791017e-05, "loss": 3.3957, "step": 7233 }, { "epoch": 0.4490657396486436, "grad_norm": 0.6179481117091283, "learning_rate": 9.924997351248703e-05, "loss": 3.4175, "step": 7234 }, { "epoch": 0.4491278167484015, "grad_norm": 0.6179645571946224, "learning_rate": 9.924935018892328e-05, "loss": 3.4357, "step": 7235 }, { "epoch": 0.44918989384815944, "grad_norm": 0.6472970286125652, "learning_rate": 9.924872660841372e-05, "loss": 3.395, "step": 7236 }, { "epoch": 0.4492519709479173, "grad_norm": 0.5693307552461672, "learning_rate": 9.92481027709616e-05, "loss": 3.3937, "step": 7237 }, { "epoch": 0.4493140480476752, "grad_norm": 0.5373747846110929, "learning_rate": 9.924747867657018e-05, "loss": 3.3358, "step": 7238 }, { "epoch": 0.44937612514743314, "grad_norm": 0.4810179541096038, "learning_rate": 9.92468543252427e-05, "loss": 3.2499, "step": 7239 }, { "epoch": 0.449438202247191, "grad_norm": 0.43256851508041627, "learning_rate": 9.924622971698243e-05, "loss": 3.3569, "step": 7240 }, { "epoch": 0.4495002793469489, "grad_norm": 0.5125216235274431, "learning_rate": 9.924560485179262e-05, "loss": 3.3622, "step": 7241 }, { "epoch": 0.44956235644670683, "grad_norm": 0.558095643231533, "learning_rate": 9.924497972967652e-05, "loss": 3.4154, "step": 7242 }, { "epoch": 0.4496244335464647, "grad_norm": 0.5253523631858809, "learning_rate": 9.924435435063744e-05, "loss": 3.4473, "step": 7243 }, { "epoch": 0.4496865106462226, "grad_norm": 0.5870124112462903, "learning_rate": 9.924372871467859e-05, "loss": 3.3599, "step": 7244 }, { "epoch": 0.44974858774598053, "grad_norm": 0.5866721824319874, "learning_rate": 9.924310282180325e-05, "loss": 3.431, "step": 7245 }, { "epoch": 0.4498106648457384, "grad_norm": 0.4749233165128593, "learning_rate": 9.92424766720147e-05, "loss": 3.3843, "step": 7246 }, { "epoch": 0.4498727419454963, "grad_norm": 0.5837156768445088, "learning_rate": 9.924185026531618e-05, "loss": 3.4508, "step": 7247 }, { "epoch": 0.44993481904525423, "grad_norm": 0.5768187173270019, "learning_rate": 9.924122360171098e-05, "loss": 3.3159, "step": 7248 }, { "epoch": 0.4499968961450121, "grad_norm": 0.49077945209745877, "learning_rate": 9.924059668120236e-05, "loss": 3.3761, "step": 7249 }, { "epoch": 0.45005897324477, "grad_norm": 0.6216640373366987, "learning_rate": 9.92399695037936e-05, "loss": 3.3114, "step": 7250 }, { "epoch": 0.45012105034452793, "grad_norm": 0.6355964505320973, "learning_rate": 9.923934206948796e-05, "loss": 3.3427, "step": 7251 }, { "epoch": 0.4501831274442858, "grad_norm": 0.6232649083839044, "learning_rate": 9.923871437828872e-05, "loss": 3.2737, "step": 7252 }, { "epoch": 0.4502452045440437, "grad_norm": 0.5709180882687294, "learning_rate": 9.923808643019917e-05, "loss": 3.2848, "step": 7253 }, { "epoch": 0.4503072816438016, "grad_norm": 0.539664333356526, "learning_rate": 9.923745822522255e-05, "loss": 3.4296, "step": 7254 }, { "epoch": 0.4503693587435595, "grad_norm": 0.5668703216982154, "learning_rate": 9.923682976336216e-05, "loss": 3.4415, "step": 7255 }, { "epoch": 0.4504314358433174, "grad_norm": 0.6402670162461462, "learning_rate": 9.923620104462128e-05, "loss": 3.3412, "step": 7256 }, { "epoch": 0.4504935129430753, "grad_norm": 0.502577664033977, "learning_rate": 9.923557206900318e-05, "loss": 3.3795, "step": 7257 }, { "epoch": 0.4505555900428332, "grad_norm": 0.5422030202089885, "learning_rate": 9.923494283651115e-05, "loss": 3.3634, "step": 7258 }, { "epoch": 0.4506176671425911, "grad_norm": 0.6363947927678535, "learning_rate": 9.923431334714849e-05, "loss": 3.486, "step": 7259 }, { "epoch": 0.450679744242349, "grad_norm": 0.5252378447232363, "learning_rate": 9.923368360091844e-05, "loss": 3.3458, "step": 7260 }, { "epoch": 0.4507418213421069, "grad_norm": 0.5072627099119529, "learning_rate": 9.923305359782433e-05, "loss": 3.338, "step": 7261 }, { "epoch": 0.4508038984418648, "grad_norm": 0.5220010890657019, "learning_rate": 9.923242333786941e-05, "loss": 3.3841, "step": 7262 }, { "epoch": 0.4508659755416227, "grad_norm": 0.5891580701449942, "learning_rate": 9.923179282105701e-05, "loss": 3.403, "step": 7263 }, { "epoch": 0.4509280526413806, "grad_norm": 0.6306234574086402, "learning_rate": 9.923116204739038e-05, "loss": 3.4391, "step": 7264 }, { "epoch": 0.4509901297411385, "grad_norm": 0.5118960881343263, "learning_rate": 9.923053101687283e-05, "loss": 3.3184, "step": 7265 }, { "epoch": 0.4510522068408964, "grad_norm": 0.5619332849571258, "learning_rate": 9.922989972950765e-05, "loss": 3.3907, "step": 7266 }, { "epoch": 0.4511142839406543, "grad_norm": 0.4916889905576029, "learning_rate": 9.922926818529811e-05, "loss": 3.3988, "step": 7267 }, { "epoch": 0.4511763610404122, "grad_norm": 0.5118123325650883, "learning_rate": 9.922863638424755e-05, "loss": 3.394, "step": 7268 }, { "epoch": 0.4512384381401701, "grad_norm": 0.5488817182011156, "learning_rate": 9.922800432635924e-05, "loss": 3.3056, "step": 7269 }, { "epoch": 0.451300515239928, "grad_norm": 0.46262031420767413, "learning_rate": 9.922737201163648e-05, "loss": 3.408, "step": 7270 }, { "epoch": 0.4513625923396859, "grad_norm": 0.5837270112177293, "learning_rate": 9.922673944008256e-05, "loss": 3.297, "step": 7271 }, { "epoch": 0.4514246694394438, "grad_norm": 0.5938111073018204, "learning_rate": 9.92261066117008e-05, "loss": 3.3921, "step": 7272 }, { "epoch": 0.4514867465392017, "grad_norm": 0.5584072521964346, "learning_rate": 9.922547352649449e-05, "loss": 3.3636, "step": 7273 }, { "epoch": 0.4515488236389596, "grad_norm": 0.4953125938632046, "learning_rate": 9.922484018446693e-05, "loss": 3.2819, "step": 7274 }, { "epoch": 0.4516109007387175, "grad_norm": 0.4736691962356234, "learning_rate": 9.922420658562145e-05, "loss": 3.3805, "step": 7275 }, { "epoch": 0.4516729778384754, "grad_norm": 0.5075002443848571, "learning_rate": 9.922357272996133e-05, "loss": 3.3966, "step": 7276 }, { "epoch": 0.4517350549382333, "grad_norm": 0.6033939127734814, "learning_rate": 9.922293861748986e-05, "loss": 3.3129, "step": 7277 }, { "epoch": 0.4517971320379912, "grad_norm": 0.6247788037838126, "learning_rate": 9.922230424821037e-05, "loss": 3.3201, "step": 7278 }, { "epoch": 0.4518592091377491, "grad_norm": 0.6110726697227077, "learning_rate": 9.922166962212619e-05, "loss": 3.3974, "step": 7279 }, { "epoch": 0.451921286237507, "grad_norm": 0.615592948931256, "learning_rate": 9.922103473924061e-05, "loss": 3.3414, "step": 7280 }, { "epoch": 0.4519833633372649, "grad_norm": 0.5773097811605261, "learning_rate": 9.922039959955692e-05, "loss": 3.4404, "step": 7281 }, { "epoch": 0.45204544043702277, "grad_norm": 0.5460650849902791, "learning_rate": 9.921976420307848e-05, "loss": 3.4617, "step": 7282 }, { "epoch": 0.4521075175367807, "grad_norm": 0.6003107448395345, "learning_rate": 9.921912854980857e-05, "loss": 3.4308, "step": 7283 }, { "epoch": 0.4521695946365386, "grad_norm": 0.5663521313298918, "learning_rate": 9.921849263975051e-05, "loss": 3.367, "step": 7284 }, { "epoch": 0.45223167173629647, "grad_norm": 0.5222997113604828, "learning_rate": 9.921785647290765e-05, "loss": 3.317, "step": 7285 }, { "epoch": 0.4522937488360544, "grad_norm": 0.5286365627233875, "learning_rate": 9.921722004928326e-05, "loss": 3.4198, "step": 7286 }, { "epoch": 0.4523558259358123, "grad_norm": 0.5646739118721873, "learning_rate": 9.92165833688807e-05, "loss": 3.4045, "step": 7287 }, { "epoch": 0.45241790303557017, "grad_norm": 0.595937913995528, "learning_rate": 9.921594643170326e-05, "loss": 3.3189, "step": 7288 }, { "epoch": 0.4524799801353281, "grad_norm": 0.5292590926645676, "learning_rate": 9.92153092377543e-05, "loss": 3.4296, "step": 7289 }, { "epoch": 0.452542057235086, "grad_norm": 0.5128518292527119, "learning_rate": 9.921467178703712e-05, "loss": 3.3321, "step": 7290 }, { "epoch": 0.45260413433484387, "grad_norm": 0.5250505672806685, "learning_rate": 9.921403407955506e-05, "loss": 3.3422, "step": 7291 }, { "epoch": 0.4526662114346018, "grad_norm": 0.5433203018806112, "learning_rate": 9.921339611531142e-05, "loss": 3.2953, "step": 7292 }, { "epoch": 0.4527282885343597, "grad_norm": 0.4970550234978316, "learning_rate": 9.921275789430955e-05, "loss": 3.3072, "step": 7293 }, { "epoch": 0.45279036563411756, "grad_norm": 0.47539680997262085, "learning_rate": 9.921211941655278e-05, "loss": 3.3735, "step": 7294 }, { "epoch": 0.4528524427338755, "grad_norm": 0.4794766643054722, "learning_rate": 9.921148068204442e-05, "loss": 3.2806, "step": 7295 }, { "epoch": 0.4529145198336334, "grad_norm": 0.5419470141031798, "learning_rate": 9.921084169078782e-05, "loss": 3.3406, "step": 7296 }, { "epoch": 0.45297659693339126, "grad_norm": 0.50988440755707, "learning_rate": 9.921020244278634e-05, "loss": 3.2988, "step": 7297 }, { "epoch": 0.4530386740331492, "grad_norm": 0.5049961339217152, "learning_rate": 9.920956293804327e-05, "loss": 3.4111, "step": 7298 }, { "epoch": 0.4531007511329071, "grad_norm": 0.47432962657968375, "learning_rate": 9.920892317656195e-05, "loss": 3.2923, "step": 7299 }, { "epoch": 0.45316282823266496, "grad_norm": 0.5708963984669342, "learning_rate": 9.920828315834575e-05, "loss": 3.3327, "step": 7300 }, { "epoch": 0.4532249053324229, "grad_norm": 0.5191226585844252, "learning_rate": 9.920764288339799e-05, "loss": 3.2995, "step": 7301 }, { "epoch": 0.4532869824321808, "grad_norm": 0.5124431171483909, "learning_rate": 9.9207002351722e-05, "loss": 3.2953, "step": 7302 }, { "epoch": 0.45334905953193866, "grad_norm": 0.5037854367705649, "learning_rate": 9.920636156332115e-05, "loss": 3.3866, "step": 7303 }, { "epoch": 0.4534111366316966, "grad_norm": 0.5698936701195316, "learning_rate": 9.920572051819876e-05, "loss": 3.3949, "step": 7304 }, { "epoch": 0.4534732137314545, "grad_norm": 0.5959629430275438, "learning_rate": 9.920507921635817e-05, "loss": 3.2512, "step": 7305 }, { "epoch": 0.45353529083121236, "grad_norm": 0.5545195473010889, "learning_rate": 9.920443765780275e-05, "loss": 3.2615, "step": 7306 }, { "epoch": 0.4535973679309703, "grad_norm": 0.449543412686762, "learning_rate": 9.920379584253583e-05, "loss": 3.278, "step": 7307 }, { "epoch": 0.45365944503072814, "grad_norm": 0.5247884369503807, "learning_rate": 9.920315377056077e-05, "loss": 3.3756, "step": 7308 }, { "epoch": 0.45372152213048605, "grad_norm": 0.5067863366583948, "learning_rate": 9.92025114418809e-05, "loss": 3.3589, "step": 7309 }, { "epoch": 0.453783599230244, "grad_norm": 0.5199509266475766, "learning_rate": 9.920186885649958e-05, "loss": 3.2955, "step": 7310 }, { "epoch": 0.45384567633000183, "grad_norm": 0.5455107702823138, "learning_rate": 9.920122601442018e-05, "loss": 3.2504, "step": 7311 }, { "epoch": 0.45390775342975975, "grad_norm": 0.6115089124808302, "learning_rate": 9.920058291564604e-05, "loss": 3.3777, "step": 7312 }, { "epoch": 0.45396983052951767, "grad_norm": 0.5278805014404687, "learning_rate": 9.91999395601805e-05, "loss": 3.3817, "step": 7313 }, { "epoch": 0.45403190762927553, "grad_norm": 0.499641364343303, "learning_rate": 9.919929594802695e-05, "loss": 3.3279, "step": 7314 }, { "epoch": 0.45409398472903345, "grad_norm": 0.519783618408561, "learning_rate": 9.919865207918871e-05, "loss": 3.3378, "step": 7315 }, { "epoch": 0.45415606182879137, "grad_norm": 0.49717351595275916, "learning_rate": 9.919800795366918e-05, "loss": 3.4014, "step": 7316 }, { "epoch": 0.45421813892854923, "grad_norm": 0.5293640994916707, "learning_rate": 9.919736357147168e-05, "loss": 3.4087, "step": 7317 }, { "epoch": 0.45428021602830715, "grad_norm": 0.5741474081342863, "learning_rate": 9.919671893259961e-05, "loss": 3.4399, "step": 7318 }, { "epoch": 0.45434229312806507, "grad_norm": 0.5669726203803593, "learning_rate": 9.91960740370563e-05, "loss": 3.3506, "step": 7319 }, { "epoch": 0.45440437022782293, "grad_norm": 0.4875550935865097, "learning_rate": 9.919542888484512e-05, "loss": 3.3567, "step": 7320 }, { "epoch": 0.45446644732758085, "grad_norm": 0.5012756109472807, "learning_rate": 9.919478347596947e-05, "loss": 3.4451, "step": 7321 }, { "epoch": 0.45452852442733876, "grad_norm": 0.5531349213567056, "learning_rate": 9.919413781043267e-05, "loss": 3.4009, "step": 7322 }, { "epoch": 0.4545906015270966, "grad_norm": 0.5692545863768012, "learning_rate": 9.919349188823812e-05, "loss": 3.4342, "step": 7323 }, { "epoch": 0.45465267862685455, "grad_norm": 0.5454931785076561, "learning_rate": 9.919284570938918e-05, "loss": 3.3469, "step": 7324 }, { "epoch": 0.45471475572661246, "grad_norm": 0.5396644438041412, "learning_rate": 9.919219927388923e-05, "loss": 3.2938, "step": 7325 }, { "epoch": 0.4547768328263703, "grad_norm": 0.5495396952865488, "learning_rate": 9.919155258174161e-05, "loss": 3.3598, "step": 7326 }, { "epoch": 0.45483890992612824, "grad_norm": 0.5871173677896584, "learning_rate": 9.919090563294975e-05, "loss": 3.3526, "step": 7327 }, { "epoch": 0.45490098702588616, "grad_norm": 0.513102226448916, "learning_rate": 9.919025842751699e-05, "loss": 3.3093, "step": 7328 }, { "epoch": 0.454963064125644, "grad_norm": 0.5720913983096293, "learning_rate": 9.91896109654467e-05, "loss": 3.3489, "step": 7329 }, { "epoch": 0.45502514122540194, "grad_norm": 0.5338546628872186, "learning_rate": 9.918896324674226e-05, "loss": 3.3302, "step": 7330 }, { "epoch": 0.45508721832515986, "grad_norm": 0.54622660754853, "learning_rate": 9.918831527140707e-05, "loss": 3.3169, "step": 7331 }, { "epoch": 0.4551492954249177, "grad_norm": 0.5263617520591345, "learning_rate": 9.918766703944448e-05, "loss": 3.305, "step": 7332 }, { "epoch": 0.45521137252467564, "grad_norm": 0.4829820322945788, "learning_rate": 9.918701855085792e-05, "loss": 3.4136, "step": 7333 }, { "epoch": 0.45527344962443356, "grad_norm": 0.5010409843561784, "learning_rate": 9.918636980565072e-05, "loss": 3.3177, "step": 7334 }, { "epoch": 0.4553355267241914, "grad_norm": 0.49237471758836066, "learning_rate": 9.918572080382631e-05, "loss": 3.4434, "step": 7335 }, { "epoch": 0.45539760382394934, "grad_norm": 0.5596555475158901, "learning_rate": 9.918507154538803e-05, "loss": 3.4271, "step": 7336 }, { "epoch": 0.45545968092370726, "grad_norm": 0.5131945695075383, "learning_rate": 9.918442203033931e-05, "loss": 3.3933, "step": 7337 }, { "epoch": 0.4555217580234651, "grad_norm": 0.5059594014960284, "learning_rate": 9.918377225868352e-05, "loss": 3.3639, "step": 7338 }, { "epoch": 0.45558383512322304, "grad_norm": 0.5004523436506788, "learning_rate": 9.918312223042404e-05, "loss": 3.2822, "step": 7339 }, { "epoch": 0.45564591222298095, "grad_norm": 0.5330929205613135, "learning_rate": 9.918247194556427e-05, "loss": 3.3921, "step": 7340 }, { "epoch": 0.4557079893227388, "grad_norm": 0.5523589244090831, "learning_rate": 9.918182140410761e-05, "loss": 3.3106, "step": 7341 }, { "epoch": 0.45577006642249673, "grad_norm": 0.5013019479198191, "learning_rate": 9.918117060605745e-05, "loss": 3.416, "step": 7342 }, { "epoch": 0.45583214352225465, "grad_norm": 0.526841270389805, "learning_rate": 9.918051955141718e-05, "loss": 3.3046, "step": 7343 }, { "epoch": 0.4558942206220125, "grad_norm": 0.5755609926967662, "learning_rate": 9.917986824019021e-05, "loss": 3.3592, "step": 7344 }, { "epoch": 0.45595629772177043, "grad_norm": 0.5185501472001386, "learning_rate": 9.917921667237993e-05, "loss": 3.3905, "step": 7345 }, { "epoch": 0.45601837482152835, "grad_norm": 0.5458025907881331, "learning_rate": 9.917856484798972e-05, "loss": 3.3063, "step": 7346 }, { "epoch": 0.4560804519212862, "grad_norm": 0.5338525136286256, "learning_rate": 9.9177912767023e-05, "loss": 3.3342, "step": 7347 }, { "epoch": 0.45614252902104413, "grad_norm": 0.4817823425842751, "learning_rate": 9.917726042948318e-05, "loss": 3.3267, "step": 7348 }, { "epoch": 0.45620460612080205, "grad_norm": 0.5657905915802566, "learning_rate": 9.917660783537365e-05, "loss": 3.3373, "step": 7349 }, { "epoch": 0.4562666832205599, "grad_norm": 0.6153835539625258, "learning_rate": 9.917595498469783e-05, "loss": 3.4122, "step": 7350 }, { "epoch": 0.45632876032031783, "grad_norm": 0.5630937727251709, "learning_rate": 9.917530187745912e-05, "loss": 3.37, "step": 7351 }, { "epoch": 0.45639083742007575, "grad_norm": 0.5405033316443396, "learning_rate": 9.91746485136609e-05, "loss": 3.4348, "step": 7352 }, { "epoch": 0.4564529145198336, "grad_norm": 0.5982887063168267, "learning_rate": 9.917399489330661e-05, "loss": 3.3554, "step": 7353 }, { "epoch": 0.4565149916195915, "grad_norm": 0.5909558605728786, "learning_rate": 9.917334101639964e-05, "loss": 3.3104, "step": 7354 }, { "epoch": 0.45657706871934944, "grad_norm": 0.5522828701196048, "learning_rate": 9.917268688294343e-05, "loss": 3.3357, "step": 7355 }, { "epoch": 0.4566391458191073, "grad_norm": 0.5320481587550457, "learning_rate": 9.917203249294138e-05, "loss": 3.3869, "step": 7356 }, { "epoch": 0.4567012229188652, "grad_norm": 0.5181657154849159, "learning_rate": 9.917137784639687e-05, "loss": 3.3017, "step": 7357 }, { "epoch": 0.45676330001862314, "grad_norm": 0.5141842393587425, "learning_rate": 9.917072294331338e-05, "loss": 3.3636, "step": 7358 }, { "epoch": 0.456825377118381, "grad_norm": 0.5882589764070248, "learning_rate": 9.917006778369427e-05, "loss": 3.4578, "step": 7359 }, { "epoch": 0.4568874542181389, "grad_norm": 0.6817884967945289, "learning_rate": 9.916941236754298e-05, "loss": 3.3701, "step": 7360 }, { "epoch": 0.45694953131789684, "grad_norm": 0.5606038666040976, "learning_rate": 9.916875669486295e-05, "loss": 3.3837, "step": 7361 }, { "epoch": 0.4570116084176547, "grad_norm": 0.5399786310029776, "learning_rate": 9.916810076565756e-05, "loss": 3.3099, "step": 7362 }, { "epoch": 0.4570736855174126, "grad_norm": 0.5273954860598964, "learning_rate": 9.916744457993025e-05, "loss": 3.3443, "step": 7363 }, { "epoch": 0.45713576261717054, "grad_norm": 0.5557304570601074, "learning_rate": 9.916678813768445e-05, "loss": 3.2975, "step": 7364 }, { "epoch": 0.4571978397169284, "grad_norm": 0.5301843782670513, "learning_rate": 9.91661314389236e-05, "loss": 3.4312, "step": 7365 }, { "epoch": 0.4572599168166863, "grad_norm": 0.5389242686076695, "learning_rate": 9.916547448365108e-05, "loss": 3.273, "step": 7366 }, { "epoch": 0.45732199391644424, "grad_norm": 0.5091198159133493, "learning_rate": 9.916481727187035e-05, "loss": 3.4019, "step": 7367 }, { "epoch": 0.4573840710162021, "grad_norm": 0.5627970176232084, "learning_rate": 9.916415980358485e-05, "loss": 3.352, "step": 7368 }, { "epoch": 0.45744614811596, "grad_norm": 0.501239313596743, "learning_rate": 9.916350207879798e-05, "loss": 3.274, "step": 7369 }, { "epoch": 0.45750822521571793, "grad_norm": 0.5423151926727993, "learning_rate": 9.916284409751318e-05, "loss": 3.3477, "step": 7370 }, { "epoch": 0.4575703023154758, "grad_norm": 0.5209396603453109, "learning_rate": 9.91621858597339e-05, "loss": 3.3671, "step": 7371 }, { "epoch": 0.4576323794152337, "grad_norm": 0.543739732009578, "learning_rate": 9.916152736546356e-05, "loss": 3.3914, "step": 7372 }, { "epoch": 0.45769445651499163, "grad_norm": 0.516712659921623, "learning_rate": 9.91608686147056e-05, "loss": 3.2772, "step": 7373 }, { "epoch": 0.4577565336147495, "grad_norm": 0.4707460660856175, "learning_rate": 9.916020960746345e-05, "loss": 3.2638, "step": 7374 }, { "epoch": 0.4578186107145074, "grad_norm": 0.5056251955916093, "learning_rate": 9.915955034374054e-05, "loss": 3.3949, "step": 7375 }, { "epoch": 0.45788068781426533, "grad_norm": 0.47060219090118316, "learning_rate": 9.915889082354033e-05, "loss": 3.3052, "step": 7376 }, { "epoch": 0.4579427649140232, "grad_norm": 0.4645915108404368, "learning_rate": 9.915823104686625e-05, "loss": 3.3568, "step": 7377 }, { "epoch": 0.4580048420137811, "grad_norm": 0.4411636631172487, "learning_rate": 9.915757101372177e-05, "loss": 3.4332, "step": 7378 }, { "epoch": 0.45806691911353903, "grad_norm": 0.496476949040472, "learning_rate": 9.915691072411027e-05, "loss": 3.3437, "step": 7379 }, { "epoch": 0.4581289962132969, "grad_norm": 0.5060243079939557, "learning_rate": 9.915625017803525e-05, "loss": 3.2805, "step": 7380 }, { "epoch": 0.4581910733130548, "grad_norm": 0.5207179291345586, "learning_rate": 9.915558937550014e-05, "loss": 3.3016, "step": 7381 }, { "epoch": 0.4582531504128127, "grad_norm": 0.4915600364705781, "learning_rate": 9.915492831650839e-05, "loss": 3.3121, "step": 7382 }, { "epoch": 0.4583152275125706, "grad_norm": 0.446196131139574, "learning_rate": 9.915426700106345e-05, "loss": 3.4079, "step": 7383 }, { "epoch": 0.4583773046123285, "grad_norm": 0.48869952787591536, "learning_rate": 9.915360542916876e-05, "loss": 3.3023, "step": 7384 }, { "epoch": 0.4584393817120864, "grad_norm": 0.49594077808087567, "learning_rate": 9.915294360082778e-05, "loss": 3.2945, "step": 7385 }, { "epoch": 0.4585014588118443, "grad_norm": 0.4844045630016449, "learning_rate": 9.915228151604397e-05, "loss": 3.3165, "step": 7386 }, { "epoch": 0.4585635359116022, "grad_norm": 0.559852781784534, "learning_rate": 9.915161917482076e-05, "loss": 3.3518, "step": 7387 }, { "epoch": 0.4586256130113601, "grad_norm": 0.5283102432955519, "learning_rate": 9.915095657716162e-05, "loss": 3.277, "step": 7388 }, { "epoch": 0.458687690111118, "grad_norm": 0.6122720274664101, "learning_rate": 9.915029372307002e-05, "loss": 3.3536, "step": 7389 }, { "epoch": 0.4587497672108759, "grad_norm": 0.6694926007042122, "learning_rate": 9.91496306125494e-05, "loss": 3.3471, "step": 7390 }, { "epoch": 0.4588118443106338, "grad_norm": 0.6265620024719546, "learning_rate": 9.914896724560323e-05, "loss": 3.4503, "step": 7391 }, { "epoch": 0.4588739214103917, "grad_norm": 0.5571963297026109, "learning_rate": 9.914830362223495e-05, "loss": 3.349, "step": 7392 }, { "epoch": 0.4589359985101496, "grad_norm": 0.487053188947203, "learning_rate": 9.914763974244807e-05, "loss": 3.3873, "step": 7393 }, { "epoch": 0.4589980756099075, "grad_norm": 0.5693258085053231, "learning_rate": 9.914697560624599e-05, "loss": 3.3617, "step": 7394 }, { "epoch": 0.4590601527096654, "grad_norm": 0.5279641069996643, "learning_rate": 9.914631121363222e-05, "loss": 3.4041, "step": 7395 }, { "epoch": 0.4591222298094233, "grad_norm": 0.6072981239180573, "learning_rate": 9.914564656461022e-05, "loss": 3.3491, "step": 7396 }, { "epoch": 0.4591843069091812, "grad_norm": 0.6504254947513702, "learning_rate": 9.914498165918345e-05, "loss": 3.3035, "step": 7397 }, { "epoch": 0.4592463840089391, "grad_norm": 0.5434975004937076, "learning_rate": 9.914431649735537e-05, "loss": 3.4348, "step": 7398 }, { "epoch": 0.459308461108697, "grad_norm": 0.5337554809413972, "learning_rate": 9.914365107912946e-05, "loss": 3.3851, "step": 7399 }, { "epoch": 0.4593705382084549, "grad_norm": 0.5366671041769215, "learning_rate": 9.914298540450922e-05, "loss": 3.3785, "step": 7400 }, { "epoch": 0.4594326153082128, "grad_norm": 0.5726491274527349, "learning_rate": 9.914231947349805e-05, "loss": 3.3189, "step": 7401 }, { "epoch": 0.4594946924079707, "grad_norm": 0.46976507270916285, "learning_rate": 9.91416532860995e-05, "loss": 3.2697, "step": 7402 }, { "epoch": 0.4595567695077286, "grad_norm": 0.518956662571853, "learning_rate": 9.914098684231701e-05, "loss": 3.3405, "step": 7403 }, { "epoch": 0.4596188466074865, "grad_norm": 0.5271513976628458, "learning_rate": 9.914032014215407e-05, "loss": 3.4574, "step": 7404 }, { "epoch": 0.4596809237072444, "grad_norm": 0.4965000741368879, "learning_rate": 9.913965318561413e-05, "loss": 3.3816, "step": 7405 }, { "epoch": 0.4597430008070023, "grad_norm": 0.4996279434917174, "learning_rate": 9.913898597270071e-05, "loss": 3.315, "step": 7406 }, { "epoch": 0.4598050779067602, "grad_norm": 0.549833225844389, "learning_rate": 9.913831850341726e-05, "loss": 3.3596, "step": 7407 }, { "epoch": 0.4598671550065181, "grad_norm": 0.5444364675950636, "learning_rate": 9.913765077776728e-05, "loss": 3.2716, "step": 7408 }, { "epoch": 0.459929232106276, "grad_norm": 0.5314000983528503, "learning_rate": 9.913698279575425e-05, "loss": 3.3275, "step": 7409 }, { "epoch": 0.45999130920603387, "grad_norm": 0.6067539971325664, "learning_rate": 9.913631455738166e-05, "loss": 3.387, "step": 7410 }, { "epoch": 0.4600533863057918, "grad_norm": 0.5930510221358181, "learning_rate": 9.913564606265297e-05, "loss": 3.3926, "step": 7411 }, { "epoch": 0.4601154634055497, "grad_norm": 0.48812749349174595, "learning_rate": 9.913497731157168e-05, "loss": 3.3335, "step": 7412 }, { "epoch": 0.46017754050530757, "grad_norm": 0.5012446647226919, "learning_rate": 9.91343083041413e-05, "loss": 3.3075, "step": 7413 }, { "epoch": 0.4602396176050655, "grad_norm": 0.5546558532283575, "learning_rate": 9.91336390403653e-05, "loss": 3.3799, "step": 7414 }, { "epoch": 0.4603016947048234, "grad_norm": 0.46444362762211777, "learning_rate": 9.913296952024719e-05, "loss": 3.3002, "step": 7415 }, { "epoch": 0.46036377180458127, "grad_norm": 0.5087124699145728, "learning_rate": 9.913229974379043e-05, "loss": 3.3824, "step": 7416 }, { "epoch": 0.4604258489043392, "grad_norm": 0.636835871496819, "learning_rate": 9.913162971099857e-05, "loss": 3.3903, "step": 7417 }, { "epoch": 0.4604879260040971, "grad_norm": 0.655195046751573, "learning_rate": 9.913095942187503e-05, "loss": 3.34, "step": 7418 }, { "epoch": 0.46055000310385497, "grad_norm": 0.6305216543697615, "learning_rate": 9.913028887642337e-05, "loss": 3.386, "step": 7419 }, { "epoch": 0.4606120802036129, "grad_norm": 0.5880799859847056, "learning_rate": 9.912961807464707e-05, "loss": 3.2856, "step": 7420 }, { "epoch": 0.4606741573033708, "grad_norm": 0.5532205822359485, "learning_rate": 9.91289470165496e-05, "loss": 3.1799, "step": 7421 }, { "epoch": 0.46073623440312866, "grad_norm": 0.6118854261528965, "learning_rate": 9.91282757021345e-05, "loss": 3.3115, "step": 7422 }, { "epoch": 0.4607983115028866, "grad_norm": 0.5809089028972365, "learning_rate": 9.912760413140528e-05, "loss": 3.3719, "step": 7423 }, { "epoch": 0.4608603886026445, "grad_norm": 0.6256743607689002, "learning_rate": 9.912693230436541e-05, "loss": 3.3101, "step": 7424 }, { "epoch": 0.46092246570240236, "grad_norm": 0.617754277605463, "learning_rate": 9.91262602210184e-05, "loss": 3.2174, "step": 7425 }, { "epoch": 0.4609845428021603, "grad_norm": 0.5397090904012402, "learning_rate": 9.912558788136777e-05, "loss": 3.3657, "step": 7426 }, { "epoch": 0.4610466199019182, "grad_norm": 0.6531968163973939, "learning_rate": 9.912491528541702e-05, "loss": 3.3536, "step": 7427 }, { "epoch": 0.46110869700167606, "grad_norm": 0.5488225861945265, "learning_rate": 9.912424243316967e-05, "loss": 3.3307, "step": 7428 }, { "epoch": 0.461170774101434, "grad_norm": 0.5509223216233068, "learning_rate": 9.912356932462921e-05, "loss": 3.2897, "step": 7429 }, { "epoch": 0.4612328512011919, "grad_norm": 0.5612751818319819, "learning_rate": 9.912289595979915e-05, "loss": 3.3186, "step": 7430 }, { "epoch": 0.46129492830094976, "grad_norm": 0.5882516557157068, "learning_rate": 9.912222233868305e-05, "loss": 3.203, "step": 7431 }, { "epoch": 0.4613570054007077, "grad_norm": 0.5394438624708994, "learning_rate": 9.912154846128438e-05, "loss": 3.3829, "step": 7432 }, { "epoch": 0.4614190825004656, "grad_norm": 0.5954765764541415, "learning_rate": 9.912087432760667e-05, "loss": 3.3271, "step": 7433 }, { "epoch": 0.46148115960022346, "grad_norm": 0.6602860723750248, "learning_rate": 9.912019993765343e-05, "loss": 3.4193, "step": 7434 }, { "epoch": 0.4615432366999814, "grad_norm": 0.5725163245892942, "learning_rate": 9.911952529142816e-05, "loss": 3.3136, "step": 7435 }, { "epoch": 0.4616053137997393, "grad_norm": 0.6164391000017776, "learning_rate": 9.911885038893442e-05, "loss": 3.4595, "step": 7436 }, { "epoch": 0.46166739089949715, "grad_norm": 0.5885841866271496, "learning_rate": 9.911817523017573e-05, "loss": 3.4746, "step": 7437 }, { "epoch": 0.46172946799925507, "grad_norm": 0.5980371246440047, "learning_rate": 9.911749981515558e-05, "loss": 3.3429, "step": 7438 }, { "epoch": 0.461791545099013, "grad_norm": 0.516677677770409, "learning_rate": 9.911682414387751e-05, "loss": 3.3055, "step": 7439 }, { "epoch": 0.46185362219877085, "grad_norm": 0.5195739843245812, "learning_rate": 9.911614821634504e-05, "loss": 3.3702, "step": 7440 }, { "epoch": 0.46191569929852877, "grad_norm": 0.5717597000957213, "learning_rate": 9.911547203256172e-05, "loss": 3.3442, "step": 7441 }, { "epoch": 0.4619777763982867, "grad_norm": 0.5194494139960025, "learning_rate": 9.911479559253105e-05, "loss": 3.4022, "step": 7442 }, { "epoch": 0.46203985349804455, "grad_norm": 0.5968294845323088, "learning_rate": 9.911411889625656e-05, "loss": 3.3609, "step": 7443 }, { "epoch": 0.46210193059780247, "grad_norm": 0.5968753724821036, "learning_rate": 9.911344194374179e-05, "loss": 3.3104, "step": 7444 }, { "epoch": 0.4621640076975604, "grad_norm": 0.5712099199821417, "learning_rate": 9.911276473499028e-05, "loss": 3.269, "step": 7445 }, { "epoch": 0.46222608479731825, "grad_norm": 0.6418267584987067, "learning_rate": 9.911208727000555e-05, "loss": 3.3993, "step": 7446 }, { "epoch": 0.46228816189707617, "grad_norm": 0.6493552962940389, "learning_rate": 9.911140954879113e-05, "loss": 3.3747, "step": 7447 }, { "epoch": 0.4623502389968341, "grad_norm": 0.7142862115585776, "learning_rate": 9.911073157135058e-05, "loss": 3.3743, "step": 7448 }, { "epoch": 0.46241231609659195, "grad_norm": 0.6424427527591378, "learning_rate": 9.911005333768741e-05, "loss": 3.3801, "step": 7449 }, { "epoch": 0.46247439319634986, "grad_norm": 0.6798722684678762, "learning_rate": 9.910937484780518e-05, "loss": 3.3635, "step": 7450 }, { "epoch": 0.4625364702961078, "grad_norm": 0.6248521868438316, "learning_rate": 9.91086961017074e-05, "loss": 3.276, "step": 7451 }, { "epoch": 0.46259854739586564, "grad_norm": 0.5226849227361875, "learning_rate": 9.910801709939765e-05, "loss": 3.2812, "step": 7452 }, { "epoch": 0.46266062449562356, "grad_norm": 0.6802140531158256, "learning_rate": 9.910733784087944e-05, "loss": 3.2985, "step": 7453 }, { "epoch": 0.4627227015953815, "grad_norm": 0.6522574910223159, "learning_rate": 9.910665832615635e-05, "loss": 3.339, "step": 7454 }, { "epoch": 0.46278477869513934, "grad_norm": 0.6189656777611153, "learning_rate": 9.910597855523187e-05, "loss": 3.3758, "step": 7455 }, { "epoch": 0.46284685579489726, "grad_norm": 0.5023621254699993, "learning_rate": 9.91052985281096e-05, "loss": 3.326, "step": 7456 }, { "epoch": 0.4629089328946552, "grad_norm": 0.6043115940779602, "learning_rate": 9.910461824479307e-05, "loss": 3.4085, "step": 7457 }, { "epoch": 0.46297100999441304, "grad_norm": 0.6000979542246514, "learning_rate": 9.91039377052858e-05, "loss": 3.3412, "step": 7458 }, { "epoch": 0.46303308709417096, "grad_norm": 0.5219855311405456, "learning_rate": 9.91032569095914e-05, "loss": 3.3992, "step": 7459 }, { "epoch": 0.4630951641939289, "grad_norm": 0.5535460803605281, "learning_rate": 9.910257585771338e-05, "loss": 3.3752, "step": 7460 }, { "epoch": 0.46315724129368674, "grad_norm": 0.540296303598446, "learning_rate": 9.910189454965528e-05, "loss": 3.423, "step": 7461 }, { "epoch": 0.46321931839344466, "grad_norm": 0.5544606403058117, "learning_rate": 9.91012129854207e-05, "loss": 3.3001, "step": 7462 }, { "epoch": 0.4632813954932026, "grad_norm": 0.5870602087778048, "learning_rate": 9.910053116501316e-05, "loss": 3.2964, "step": 7463 }, { "epoch": 0.46334347259296044, "grad_norm": 0.5260925425738535, "learning_rate": 9.909984908843623e-05, "loss": 3.3658, "step": 7464 }, { "epoch": 0.46340554969271835, "grad_norm": 0.44613163882355766, "learning_rate": 9.909916675569349e-05, "loss": 3.3405, "step": 7465 }, { "epoch": 0.4634676267924763, "grad_norm": 0.5138397232656047, "learning_rate": 9.909848416678844e-05, "loss": 3.3047, "step": 7466 }, { "epoch": 0.46352970389223414, "grad_norm": 0.4799648837317488, "learning_rate": 9.90978013217247e-05, "loss": 3.3081, "step": 7467 }, { "epoch": 0.46359178099199205, "grad_norm": 0.45033720349779, "learning_rate": 9.909711822050582e-05, "loss": 3.3224, "step": 7468 }, { "epoch": 0.46365385809174997, "grad_norm": 0.4966758556870691, "learning_rate": 9.909643486313533e-05, "loss": 3.3308, "step": 7469 }, { "epoch": 0.46371593519150783, "grad_norm": 0.602840898239401, "learning_rate": 9.909575124961685e-05, "loss": 3.3483, "step": 7470 }, { "epoch": 0.46377801229126575, "grad_norm": 0.6065299547869406, "learning_rate": 9.90950673799539e-05, "loss": 3.3881, "step": 7471 }, { "epoch": 0.46384008939102367, "grad_norm": 0.532129989401698, "learning_rate": 9.909438325415007e-05, "loss": 3.3359, "step": 7472 }, { "epoch": 0.46390216649078153, "grad_norm": 0.49354823794753255, "learning_rate": 9.909369887220892e-05, "loss": 3.2385, "step": 7473 }, { "epoch": 0.46396424359053945, "grad_norm": 0.4676807765729759, "learning_rate": 9.909301423413403e-05, "loss": 3.348, "step": 7474 }, { "epoch": 0.46402632069029737, "grad_norm": 0.4982038661884874, "learning_rate": 9.909232933992895e-05, "loss": 3.3068, "step": 7475 }, { "epoch": 0.46408839779005523, "grad_norm": 0.510867364251868, "learning_rate": 9.90916441895973e-05, "loss": 3.3617, "step": 7476 }, { "epoch": 0.46415047488981315, "grad_norm": 0.5441070809195047, "learning_rate": 9.90909587831426e-05, "loss": 3.3291, "step": 7477 }, { "epoch": 0.46421255198957107, "grad_norm": 0.63346351291595, "learning_rate": 9.909027312056847e-05, "loss": 3.3097, "step": 7478 }, { "epoch": 0.4642746290893289, "grad_norm": 0.6290004967625389, "learning_rate": 9.908958720187846e-05, "loss": 3.2667, "step": 7479 }, { "epoch": 0.46433670618908685, "grad_norm": 0.5344040868327347, "learning_rate": 9.908890102707615e-05, "loss": 3.4783, "step": 7480 }, { "epoch": 0.46439878328884476, "grad_norm": 0.5553826019828041, "learning_rate": 9.908821459616514e-05, "loss": 3.3229, "step": 7481 }, { "epoch": 0.4644608603886026, "grad_norm": 0.5368689814042584, "learning_rate": 9.908752790914898e-05, "loss": 3.2611, "step": 7482 }, { "epoch": 0.46452293748836054, "grad_norm": 0.60858959544261, "learning_rate": 9.908684096603128e-05, "loss": 3.3103, "step": 7483 }, { "epoch": 0.46458501458811846, "grad_norm": 0.5577099350526381, "learning_rate": 9.908615376681563e-05, "loss": 3.2735, "step": 7484 }, { "epoch": 0.4646470916878763, "grad_norm": 0.5123183254065716, "learning_rate": 9.908546631150557e-05, "loss": 3.3322, "step": 7485 }, { "epoch": 0.46470916878763424, "grad_norm": 0.48664733644475955, "learning_rate": 9.908477860010473e-05, "loss": 3.3471, "step": 7486 }, { "epoch": 0.46477124588739216, "grad_norm": 0.5239648297806458, "learning_rate": 9.908409063261669e-05, "loss": 3.4373, "step": 7487 }, { "epoch": 0.46483332298715, "grad_norm": 0.6231771590002587, "learning_rate": 9.908340240904504e-05, "loss": 3.3822, "step": 7488 }, { "epoch": 0.46489540008690794, "grad_norm": 0.562389018448545, "learning_rate": 9.908271392939334e-05, "loss": 3.2995, "step": 7489 }, { "epoch": 0.46495747718666586, "grad_norm": 0.5079393375157432, "learning_rate": 9.908202519366521e-05, "loss": 3.2971, "step": 7490 }, { "epoch": 0.4650195542864237, "grad_norm": 0.49637582223997506, "learning_rate": 9.908133620186426e-05, "loss": 3.3673, "step": 7491 }, { "epoch": 0.46508163138618164, "grad_norm": 0.6086593240024832, "learning_rate": 9.908064695399405e-05, "loss": 3.3445, "step": 7492 }, { "epoch": 0.46514370848593956, "grad_norm": 0.685588650639649, "learning_rate": 9.907995745005819e-05, "loss": 3.4043, "step": 7493 }, { "epoch": 0.4652057855856974, "grad_norm": 0.5951765891495805, "learning_rate": 9.907926769006028e-05, "loss": 3.2342, "step": 7494 }, { "epoch": 0.46526786268545534, "grad_norm": 0.5027127288687321, "learning_rate": 9.907857767400393e-05, "loss": 3.2828, "step": 7495 }, { "epoch": 0.46532993978521325, "grad_norm": 0.5817801938944583, "learning_rate": 9.90778874018927e-05, "loss": 3.3237, "step": 7496 }, { "epoch": 0.4653920168849711, "grad_norm": 0.6247317215198364, "learning_rate": 9.907719687373022e-05, "loss": 3.4245, "step": 7497 }, { "epoch": 0.46545409398472903, "grad_norm": 0.5919685240735397, "learning_rate": 9.90765060895201e-05, "loss": 3.2818, "step": 7498 }, { "epoch": 0.46551617108448695, "grad_norm": 0.5662692200217732, "learning_rate": 9.907581504926595e-05, "loss": 3.3028, "step": 7499 }, { "epoch": 0.4655782481842448, "grad_norm": 0.5726078038975952, "learning_rate": 9.907512375297134e-05, "loss": 3.4044, "step": 7500 }, { "epoch": 0.46564032528400273, "grad_norm": 0.533017639762889, "learning_rate": 9.90744322006399e-05, "loss": 3.3145, "step": 7501 }, { "epoch": 0.46570240238376065, "grad_norm": 0.5431300410836777, "learning_rate": 9.907374039227523e-05, "loss": 3.4045, "step": 7502 }, { "epoch": 0.4657644794835185, "grad_norm": 0.6104008525019375, "learning_rate": 9.907304832788094e-05, "loss": 3.3129, "step": 7503 }, { "epoch": 0.46582655658327643, "grad_norm": 0.6094735383790033, "learning_rate": 9.907235600746066e-05, "loss": 3.2791, "step": 7504 }, { "epoch": 0.46588863368303435, "grad_norm": 0.4445423527086972, "learning_rate": 9.907166343101797e-05, "loss": 3.3188, "step": 7505 }, { "epoch": 0.4659507107827922, "grad_norm": 0.5117084997184257, "learning_rate": 9.90709705985565e-05, "loss": 3.3189, "step": 7506 }, { "epoch": 0.46601278788255013, "grad_norm": 0.4979594584023984, "learning_rate": 9.907027751007988e-05, "loss": 3.3504, "step": 7507 }, { "epoch": 0.46607486498230805, "grad_norm": 0.5619874049146472, "learning_rate": 9.906958416559168e-05, "loss": 3.3456, "step": 7508 }, { "epoch": 0.4661369420820659, "grad_norm": 0.6114578353212615, "learning_rate": 9.906889056509558e-05, "loss": 3.2808, "step": 7509 }, { "epoch": 0.4661990191818238, "grad_norm": 0.5399561399290723, "learning_rate": 9.906819670859515e-05, "loss": 3.4782, "step": 7510 }, { "epoch": 0.46626109628158174, "grad_norm": 0.5565418088773364, "learning_rate": 9.906750259609403e-05, "loss": 3.3838, "step": 7511 }, { "epoch": 0.4663231733813396, "grad_norm": 0.5665122886580726, "learning_rate": 9.906680822759584e-05, "loss": 3.3373, "step": 7512 }, { "epoch": 0.4663852504810975, "grad_norm": 0.47948650038195884, "learning_rate": 9.906611360310421e-05, "loss": 3.3312, "step": 7513 }, { "epoch": 0.46644732758085544, "grad_norm": 0.5317913270663447, "learning_rate": 9.906541872262272e-05, "loss": 3.3733, "step": 7514 }, { "epoch": 0.4665094046806133, "grad_norm": 0.5363928159192122, "learning_rate": 9.906472358615506e-05, "loss": 3.3371, "step": 7515 }, { "epoch": 0.4665714817803712, "grad_norm": 0.5627154626386447, "learning_rate": 9.906402819370482e-05, "loss": 3.2672, "step": 7516 }, { "epoch": 0.46663355888012914, "grad_norm": 0.562453268017271, "learning_rate": 9.906333254527563e-05, "loss": 3.2611, "step": 7517 }, { "epoch": 0.466695635979887, "grad_norm": 0.5400985491861154, "learning_rate": 9.906263664087112e-05, "loss": 3.3598, "step": 7518 }, { "epoch": 0.4667577130796449, "grad_norm": 0.5265774846420996, "learning_rate": 9.906194048049493e-05, "loss": 3.2855, "step": 7519 }, { "epoch": 0.46681979017940284, "grad_norm": 0.5574657815858153, "learning_rate": 9.906124406415068e-05, "loss": 3.2708, "step": 7520 }, { "epoch": 0.4668818672791607, "grad_norm": 0.5446292269115657, "learning_rate": 9.906054739184202e-05, "loss": 3.3722, "step": 7521 }, { "epoch": 0.4669439443789186, "grad_norm": 0.47411238783161413, "learning_rate": 9.905985046357256e-05, "loss": 3.2996, "step": 7522 }, { "epoch": 0.46700602147867654, "grad_norm": 0.5027137366788464, "learning_rate": 9.905915327934596e-05, "loss": 3.2888, "step": 7523 }, { "epoch": 0.4670680985784344, "grad_norm": 0.5289314387985109, "learning_rate": 9.905845583916584e-05, "loss": 3.3585, "step": 7524 }, { "epoch": 0.4671301756781923, "grad_norm": 0.5118435132469628, "learning_rate": 9.905775814303586e-05, "loss": 3.3653, "step": 7525 }, { "epoch": 0.46719225277795023, "grad_norm": 0.49915262778361486, "learning_rate": 9.905706019095964e-05, "loss": 3.2798, "step": 7526 }, { "epoch": 0.4672543298777081, "grad_norm": 0.475011615862111, "learning_rate": 9.905636198294084e-05, "loss": 3.2364, "step": 7527 }, { "epoch": 0.467316406977466, "grad_norm": 0.5673244006346104, "learning_rate": 9.905566351898308e-05, "loss": 3.3493, "step": 7528 }, { "epoch": 0.46737848407722393, "grad_norm": 0.5718637037985524, "learning_rate": 9.905496479909e-05, "loss": 3.2981, "step": 7529 }, { "epoch": 0.4674405611769818, "grad_norm": 0.46531168041579785, "learning_rate": 9.905426582326527e-05, "loss": 3.4488, "step": 7530 }, { "epoch": 0.4675026382767397, "grad_norm": 0.5113084910260894, "learning_rate": 9.905356659151253e-05, "loss": 3.3632, "step": 7531 }, { "epoch": 0.46756471537649763, "grad_norm": 0.6092127559549948, "learning_rate": 9.905286710383543e-05, "loss": 3.388, "step": 7532 }, { "epoch": 0.4676267924762555, "grad_norm": 0.5748801801066166, "learning_rate": 9.90521673602376e-05, "loss": 3.371, "step": 7533 }, { "epoch": 0.4676888695760134, "grad_norm": 0.5588657810489178, "learning_rate": 9.905146736072272e-05, "loss": 3.357, "step": 7534 }, { "epoch": 0.46775094667577133, "grad_norm": 0.5037660033947309, "learning_rate": 9.905076710529443e-05, "loss": 3.4179, "step": 7535 }, { "epoch": 0.4678130237755292, "grad_norm": 0.4442954914842197, "learning_rate": 9.905006659395636e-05, "loss": 3.2159, "step": 7536 }, { "epoch": 0.4678751008752871, "grad_norm": 0.5349821935346442, "learning_rate": 9.90493658267122e-05, "loss": 3.2813, "step": 7537 }, { "epoch": 0.467937177975045, "grad_norm": 0.5166239742616835, "learning_rate": 9.904866480356561e-05, "loss": 3.3792, "step": 7538 }, { "epoch": 0.4679992550748029, "grad_norm": 0.5572970618821663, "learning_rate": 9.90479635245202e-05, "loss": 3.3588, "step": 7539 }, { "epoch": 0.4680613321745608, "grad_norm": 0.5431026869866663, "learning_rate": 9.904726198957965e-05, "loss": 3.3354, "step": 7540 }, { "epoch": 0.4681234092743187, "grad_norm": 0.5216542513754276, "learning_rate": 9.904656019874765e-05, "loss": 3.2639, "step": 7541 }, { "epoch": 0.4681854863740766, "grad_norm": 0.6071577080344857, "learning_rate": 9.904585815202784e-05, "loss": 3.248, "step": 7542 }, { "epoch": 0.4682475634738345, "grad_norm": 0.6474785205415567, "learning_rate": 9.904515584942386e-05, "loss": 3.3805, "step": 7543 }, { "epoch": 0.4683096405735924, "grad_norm": 0.5748230226413613, "learning_rate": 9.904445329093941e-05, "loss": 3.2956, "step": 7544 }, { "epoch": 0.4683717176733503, "grad_norm": 0.5392344103926431, "learning_rate": 9.904375047657815e-05, "loss": 3.3112, "step": 7545 }, { "epoch": 0.4684337947731082, "grad_norm": 0.5454596870082502, "learning_rate": 9.904304740634372e-05, "loss": 3.3485, "step": 7546 }, { "epoch": 0.4684958718728661, "grad_norm": 0.5828250106409648, "learning_rate": 9.904234408023982e-05, "loss": 3.3476, "step": 7547 }, { "epoch": 0.468557948972624, "grad_norm": 0.5150357549718112, "learning_rate": 9.904164049827009e-05, "loss": 3.2742, "step": 7548 }, { "epoch": 0.4686200260723819, "grad_norm": 0.5397446780240559, "learning_rate": 9.904093666043823e-05, "loss": 3.3199, "step": 7549 }, { "epoch": 0.4686821031721398, "grad_norm": 0.5645992261566443, "learning_rate": 9.904023256674791e-05, "loss": 3.4004, "step": 7550 }, { "epoch": 0.4687441802718977, "grad_norm": 0.554239979055869, "learning_rate": 9.903952821720276e-05, "loss": 3.3492, "step": 7551 }, { "epoch": 0.4688062573716556, "grad_norm": 0.5636844615099399, "learning_rate": 9.90388236118065e-05, "loss": 3.2775, "step": 7552 }, { "epoch": 0.4688683344714135, "grad_norm": 0.5226436117380509, "learning_rate": 9.903811875056282e-05, "loss": 3.3988, "step": 7553 }, { "epoch": 0.4689304115711714, "grad_norm": 0.609041146989865, "learning_rate": 9.903741363347532e-05, "loss": 3.3972, "step": 7554 }, { "epoch": 0.4689924886709293, "grad_norm": 0.5857637529260578, "learning_rate": 9.903670826054776e-05, "loss": 3.3194, "step": 7555 }, { "epoch": 0.4690545657706872, "grad_norm": 0.5605961050286807, "learning_rate": 9.903600263178378e-05, "loss": 3.2896, "step": 7556 }, { "epoch": 0.4691166428704451, "grad_norm": 0.570778486535969, "learning_rate": 9.903529674718707e-05, "loss": 3.3069, "step": 7557 }, { "epoch": 0.469178719970203, "grad_norm": 0.4887267712343943, "learning_rate": 9.903459060676132e-05, "loss": 3.4331, "step": 7558 }, { "epoch": 0.4692407970699609, "grad_norm": 0.5256364949043344, "learning_rate": 9.90338842105102e-05, "loss": 3.3243, "step": 7559 }, { "epoch": 0.4693028741697188, "grad_norm": 0.5536212343891902, "learning_rate": 9.903317755843742e-05, "loss": 3.4055, "step": 7560 }, { "epoch": 0.4693649512694767, "grad_norm": 0.5397470522893266, "learning_rate": 9.903247065054664e-05, "loss": 3.275, "step": 7561 }, { "epoch": 0.4694270283692346, "grad_norm": 0.5006384945121054, "learning_rate": 9.903176348684156e-05, "loss": 3.2962, "step": 7562 }, { "epoch": 0.4694891054689925, "grad_norm": 0.45090941023758485, "learning_rate": 9.903105606732585e-05, "loss": 3.3263, "step": 7563 }, { "epoch": 0.4695511825687504, "grad_norm": 0.4806738159975781, "learning_rate": 9.903034839200323e-05, "loss": 3.2366, "step": 7564 }, { "epoch": 0.4696132596685083, "grad_norm": 0.5295131846633425, "learning_rate": 9.902964046087737e-05, "loss": 3.3027, "step": 7565 }, { "epoch": 0.46967533676826617, "grad_norm": 0.4944598407021079, "learning_rate": 9.9028932273952e-05, "loss": 3.3044, "step": 7566 }, { "epoch": 0.4697374138680241, "grad_norm": 0.4654711168221967, "learning_rate": 9.902822383123076e-05, "loss": 3.3842, "step": 7567 }, { "epoch": 0.469799490967782, "grad_norm": 0.43695286189289106, "learning_rate": 9.902751513271738e-05, "loss": 3.2969, "step": 7568 }, { "epoch": 0.46986156806753987, "grad_norm": 0.5593853123742549, "learning_rate": 9.902680617841555e-05, "loss": 3.3693, "step": 7569 }, { "epoch": 0.4699236451672978, "grad_norm": 0.5173933455078, "learning_rate": 9.902609696832898e-05, "loss": 3.3414, "step": 7570 }, { "epoch": 0.4699857222670557, "grad_norm": 0.467207851413179, "learning_rate": 9.902538750246136e-05, "loss": 3.3316, "step": 7571 }, { "epoch": 0.47004779936681357, "grad_norm": 0.480574572935055, "learning_rate": 9.90246777808164e-05, "loss": 3.3994, "step": 7572 }, { "epoch": 0.4701098764665715, "grad_norm": 0.5736507279193189, "learning_rate": 9.902396780339778e-05, "loss": 3.3118, "step": 7573 }, { "epoch": 0.4701719535663294, "grad_norm": 0.5458444967899772, "learning_rate": 9.902325757020922e-05, "loss": 3.3624, "step": 7574 }, { "epoch": 0.47023403066608727, "grad_norm": 0.5056353923733973, "learning_rate": 9.902254708125442e-05, "loss": 3.2029, "step": 7575 }, { "epoch": 0.4702961077658452, "grad_norm": 0.4543249257711028, "learning_rate": 9.902183633653709e-05, "loss": 3.3261, "step": 7576 }, { "epoch": 0.4703581848656031, "grad_norm": 0.46991711595873126, "learning_rate": 9.902112533606096e-05, "loss": 3.2912, "step": 7577 }, { "epoch": 0.47042026196536096, "grad_norm": 0.5042225635952945, "learning_rate": 9.902041407982972e-05, "loss": 3.3165, "step": 7578 }, { "epoch": 0.4704823390651189, "grad_norm": 0.51535885617963, "learning_rate": 9.901970256784705e-05, "loss": 3.2945, "step": 7579 }, { "epoch": 0.4705444161648768, "grad_norm": 0.5237464846854948, "learning_rate": 9.901899080011672e-05, "loss": 3.3409, "step": 7580 }, { "epoch": 0.47060649326463466, "grad_norm": 0.47472896560566696, "learning_rate": 9.90182787766424e-05, "loss": 3.3352, "step": 7581 }, { "epoch": 0.4706685703643926, "grad_norm": 0.5317188606183624, "learning_rate": 9.901756649742782e-05, "loss": 3.3289, "step": 7582 }, { "epoch": 0.4707306474641505, "grad_norm": 0.5541067172252081, "learning_rate": 9.90168539624767e-05, "loss": 3.3026, "step": 7583 }, { "epoch": 0.47079272456390836, "grad_norm": 0.5219481616789852, "learning_rate": 9.901614117179275e-05, "loss": 3.3781, "step": 7584 }, { "epoch": 0.4708548016636663, "grad_norm": 0.45205766660663516, "learning_rate": 9.90154281253797e-05, "loss": 3.2693, "step": 7585 }, { "epoch": 0.4709168787634242, "grad_norm": 0.5506376525689567, "learning_rate": 9.901471482324126e-05, "loss": 3.412, "step": 7586 }, { "epoch": 0.47097895586318206, "grad_norm": 0.5710511323445153, "learning_rate": 9.901400126538115e-05, "loss": 3.339, "step": 7587 }, { "epoch": 0.47104103296294, "grad_norm": 0.5097930433127156, "learning_rate": 9.901328745180309e-05, "loss": 3.3461, "step": 7588 }, { "epoch": 0.4711031100626979, "grad_norm": 0.4542277171139186, "learning_rate": 9.901257338251083e-05, "loss": 3.2957, "step": 7589 }, { "epoch": 0.47116518716245576, "grad_norm": 0.4853849188473676, "learning_rate": 9.901185905750808e-05, "loss": 3.4038, "step": 7590 }, { "epoch": 0.4712272642622137, "grad_norm": 0.5893725098024352, "learning_rate": 9.901114447679853e-05, "loss": 3.3193, "step": 7591 }, { "epoch": 0.4712893413619716, "grad_norm": 0.5230612754976095, "learning_rate": 9.901042964038596e-05, "loss": 3.3689, "step": 7592 }, { "epoch": 0.47135141846172945, "grad_norm": 0.5279437290120881, "learning_rate": 9.900971454827409e-05, "loss": 3.2377, "step": 7593 }, { "epoch": 0.4714134955614874, "grad_norm": 0.4852184980574466, "learning_rate": 9.900899920046663e-05, "loss": 3.3355, "step": 7594 }, { "epoch": 0.4714755726612453, "grad_norm": 0.5191403115575115, "learning_rate": 9.900828359696733e-05, "loss": 3.3625, "step": 7595 }, { "epoch": 0.47153764976100315, "grad_norm": 0.5555687472313737, "learning_rate": 9.900756773777992e-05, "loss": 3.2417, "step": 7596 }, { "epoch": 0.47159972686076107, "grad_norm": 0.4876044656624543, "learning_rate": 9.900685162290812e-05, "loss": 3.2625, "step": 7597 }, { "epoch": 0.471661803960519, "grad_norm": 0.5184128849378281, "learning_rate": 9.90061352523557e-05, "loss": 3.3835, "step": 7598 }, { "epoch": 0.47172388106027685, "grad_norm": 0.5184727260953781, "learning_rate": 9.900541862612635e-05, "loss": 3.3059, "step": 7599 }, { "epoch": 0.47178595816003477, "grad_norm": 0.5153217869113196, "learning_rate": 9.900470174422385e-05, "loss": 3.1582, "step": 7600 }, { "epoch": 0.4718480352597927, "grad_norm": 0.5372317920242505, "learning_rate": 9.900398460665192e-05, "loss": 3.3792, "step": 7601 }, { "epoch": 0.47191011235955055, "grad_norm": 0.5843945851843423, "learning_rate": 9.90032672134143e-05, "loss": 3.28, "step": 7602 }, { "epoch": 0.47197218945930847, "grad_norm": 0.49717158277101325, "learning_rate": 9.900254956451474e-05, "loss": 3.3291, "step": 7603 }, { "epoch": 0.4720342665590664, "grad_norm": 0.5165107228937678, "learning_rate": 9.9001831659957e-05, "loss": 3.3522, "step": 7604 }, { "epoch": 0.47209634365882425, "grad_norm": 0.475137408985937, "learning_rate": 9.90011134997448e-05, "loss": 3.4423, "step": 7605 }, { "epoch": 0.47215842075858216, "grad_norm": 0.5052753155013177, "learning_rate": 9.900039508388188e-05, "loss": 3.3149, "step": 7606 }, { "epoch": 0.4722204978583401, "grad_norm": 0.5007037038765529, "learning_rate": 9.899967641237203e-05, "loss": 3.362, "step": 7607 }, { "epoch": 0.47228257495809794, "grad_norm": 0.4660031261830394, "learning_rate": 9.899895748521895e-05, "loss": 3.2546, "step": 7608 }, { "epoch": 0.47234465205785586, "grad_norm": 0.5035513581824373, "learning_rate": 9.899823830242643e-05, "loss": 3.2961, "step": 7609 }, { "epoch": 0.4724067291576138, "grad_norm": 0.5577092670902628, "learning_rate": 9.89975188639982e-05, "loss": 3.3478, "step": 7610 }, { "epoch": 0.47246880625737164, "grad_norm": 0.5733109129173962, "learning_rate": 9.8996799169938e-05, "loss": 3.3194, "step": 7611 }, { "epoch": 0.47253088335712956, "grad_norm": 0.5050167297198238, "learning_rate": 9.899607922024962e-05, "loss": 3.3181, "step": 7612 }, { "epoch": 0.4725929604568875, "grad_norm": 0.5721378899927263, "learning_rate": 9.899535901493682e-05, "loss": 3.3768, "step": 7613 }, { "epoch": 0.47265503755664534, "grad_norm": 0.47361329028609256, "learning_rate": 9.899463855400332e-05, "loss": 3.2749, "step": 7614 }, { "epoch": 0.47271711465640326, "grad_norm": 0.5203108137406984, "learning_rate": 9.89939178374529e-05, "loss": 3.3521, "step": 7615 }, { "epoch": 0.4727791917561612, "grad_norm": 0.5878337693206941, "learning_rate": 9.899319686528932e-05, "loss": 3.3574, "step": 7616 }, { "epoch": 0.47284126885591904, "grad_norm": 0.6075078283990768, "learning_rate": 9.899247563751633e-05, "loss": 3.3615, "step": 7617 }, { "epoch": 0.47290334595567696, "grad_norm": 0.5659964558874584, "learning_rate": 9.89917541541377e-05, "loss": 3.2667, "step": 7618 }, { "epoch": 0.4729654230554349, "grad_norm": 0.5057176958610685, "learning_rate": 9.89910324151572e-05, "loss": 3.3601, "step": 7619 }, { "epoch": 0.47302750015519274, "grad_norm": 0.5618935866034528, "learning_rate": 9.899031042057859e-05, "loss": 3.3805, "step": 7620 }, { "epoch": 0.47308957725495066, "grad_norm": 0.4844125456256002, "learning_rate": 9.898958817040565e-05, "loss": 3.2373, "step": 7621 }, { "epoch": 0.4731516543547086, "grad_norm": 0.5313134716488594, "learning_rate": 9.898886566464212e-05, "loss": 3.3656, "step": 7622 }, { "epoch": 0.47321373145446644, "grad_norm": 0.44224327193986834, "learning_rate": 9.898814290329179e-05, "loss": 3.3701, "step": 7623 }, { "epoch": 0.47327580855422435, "grad_norm": 0.4998464646642817, "learning_rate": 9.898741988635842e-05, "loss": 3.4043, "step": 7624 }, { "epoch": 0.47333788565398227, "grad_norm": 0.5448051792557477, "learning_rate": 9.89866966138458e-05, "loss": 3.3277, "step": 7625 }, { "epoch": 0.47339996275374013, "grad_norm": 0.6119938598384008, "learning_rate": 9.898597308575767e-05, "loss": 3.469, "step": 7626 }, { "epoch": 0.47346203985349805, "grad_norm": 0.6798051534965771, "learning_rate": 9.898524930209786e-05, "loss": 3.3549, "step": 7627 }, { "epoch": 0.47352411695325597, "grad_norm": 0.5596344596484205, "learning_rate": 9.89845252628701e-05, "loss": 3.388, "step": 7628 }, { "epoch": 0.47358619405301383, "grad_norm": 0.5551989709042073, "learning_rate": 9.898380096807818e-05, "loss": 3.2945, "step": 7629 }, { "epoch": 0.47364827115277175, "grad_norm": 0.5642171980467554, "learning_rate": 9.898307641772586e-05, "loss": 3.3375, "step": 7630 }, { "epoch": 0.47371034825252967, "grad_norm": 0.5594829337983364, "learning_rate": 9.898235161181696e-05, "loss": 3.3035, "step": 7631 }, { "epoch": 0.47377242535228753, "grad_norm": 0.5466646607389322, "learning_rate": 9.898162655035524e-05, "loss": 3.2013, "step": 7632 }, { "epoch": 0.47383450245204545, "grad_norm": 0.4887649124338847, "learning_rate": 9.898090123334447e-05, "loss": 3.3591, "step": 7633 }, { "epoch": 0.47389657955180337, "grad_norm": 0.5038623938586423, "learning_rate": 9.898017566078845e-05, "loss": 3.3148, "step": 7634 }, { "epoch": 0.4739586566515612, "grad_norm": 0.5190203169206863, "learning_rate": 9.897944983269098e-05, "loss": 3.2682, "step": 7635 }, { "epoch": 0.47402073375131915, "grad_norm": 0.5591996381573605, "learning_rate": 9.897872374905582e-05, "loss": 3.2983, "step": 7636 }, { "epoch": 0.47408281085107706, "grad_norm": 0.6887162893668388, "learning_rate": 9.897799740988676e-05, "loss": 3.4118, "step": 7637 }, { "epoch": 0.4741448879508349, "grad_norm": 0.6149215629614648, "learning_rate": 9.897727081518762e-05, "loss": 3.3517, "step": 7638 }, { "epoch": 0.47420696505059284, "grad_norm": 0.611573873734787, "learning_rate": 9.897654396496214e-05, "loss": 3.3364, "step": 7639 }, { "epoch": 0.47426904215035076, "grad_norm": 0.6195835971724823, "learning_rate": 9.897581685921416e-05, "loss": 3.375, "step": 7640 }, { "epoch": 0.4743311192501086, "grad_norm": 0.5348597631082801, "learning_rate": 9.897508949794744e-05, "loss": 3.328, "step": 7641 }, { "epoch": 0.47439319634986654, "grad_norm": 0.5907843208384516, "learning_rate": 9.897436188116581e-05, "loss": 3.4443, "step": 7642 }, { "epoch": 0.47445527344962446, "grad_norm": 0.552254323451127, "learning_rate": 9.897363400887303e-05, "loss": 3.275, "step": 7643 }, { "epoch": 0.4745173505493823, "grad_norm": 0.5115888590887349, "learning_rate": 9.897290588107291e-05, "loss": 3.3905, "step": 7644 }, { "epoch": 0.47457942764914024, "grad_norm": 0.5557056002519521, "learning_rate": 9.897217749776926e-05, "loss": 3.3582, "step": 7645 }, { "epoch": 0.47464150474889816, "grad_norm": 0.5627521903233924, "learning_rate": 9.897144885896588e-05, "loss": 3.3679, "step": 7646 }, { "epoch": 0.474703581848656, "grad_norm": 0.6448064505546687, "learning_rate": 9.897071996466656e-05, "loss": 3.3891, "step": 7647 }, { "epoch": 0.47476565894841394, "grad_norm": 0.6725081188747036, "learning_rate": 9.896999081487509e-05, "loss": 3.3587, "step": 7648 }, { "epoch": 0.47482773604817186, "grad_norm": 0.625495499651886, "learning_rate": 9.89692614095953e-05, "loss": 3.395, "step": 7649 }, { "epoch": 0.4748898131479297, "grad_norm": 0.5895137242538796, "learning_rate": 9.896853174883098e-05, "loss": 3.3834, "step": 7650 }, { "epoch": 0.47495189024768764, "grad_norm": 0.6118957516791599, "learning_rate": 9.896780183258594e-05, "loss": 3.3812, "step": 7651 }, { "epoch": 0.47501396734744555, "grad_norm": 0.5835506397752717, "learning_rate": 9.896707166086401e-05, "loss": 3.2257, "step": 7652 }, { "epoch": 0.4750760444472034, "grad_norm": 0.5330372926269226, "learning_rate": 9.896634123366895e-05, "loss": 3.3552, "step": 7653 }, { "epoch": 0.47513812154696133, "grad_norm": 0.5264347294944334, "learning_rate": 9.896561055100462e-05, "loss": 3.3342, "step": 7654 }, { "epoch": 0.47520019864671925, "grad_norm": 0.5773588981241375, "learning_rate": 9.896487961287482e-05, "loss": 3.3038, "step": 7655 }, { "epoch": 0.4752622757464771, "grad_norm": 0.6352727419547931, "learning_rate": 9.896414841928334e-05, "loss": 3.3906, "step": 7656 }, { "epoch": 0.47532435284623503, "grad_norm": 0.6112997030165412, "learning_rate": 9.896341697023401e-05, "loss": 3.3311, "step": 7657 }, { "epoch": 0.47538642994599295, "grad_norm": 0.5063654308822734, "learning_rate": 9.896268526573067e-05, "loss": 3.3372, "step": 7658 }, { "epoch": 0.4754485070457508, "grad_norm": 0.5720854335569471, "learning_rate": 9.89619533057771e-05, "loss": 3.3931, "step": 7659 }, { "epoch": 0.47551058414550873, "grad_norm": 0.4725920263116579, "learning_rate": 9.896122109037713e-05, "loss": 3.3213, "step": 7660 }, { "epoch": 0.47557266124526665, "grad_norm": 0.5205053218474212, "learning_rate": 9.896048861953459e-05, "loss": 3.2868, "step": 7661 }, { "epoch": 0.4756347383450245, "grad_norm": 0.5130787605332591, "learning_rate": 9.895975589325328e-05, "loss": 3.26, "step": 7662 }, { "epoch": 0.47569681544478243, "grad_norm": 0.48759292915019414, "learning_rate": 9.895902291153706e-05, "loss": 3.3373, "step": 7663 }, { "epoch": 0.47575889254454035, "grad_norm": 0.47195333436943304, "learning_rate": 9.895828967438971e-05, "loss": 3.2868, "step": 7664 }, { "epoch": 0.4758209696442982, "grad_norm": 0.46232792901045894, "learning_rate": 9.89575561818151e-05, "loss": 3.3024, "step": 7665 }, { "epoch": 0.4758830467440561, "grad_norm": 0.47882068796928445, "learning_rate": 9.895682243381703e-05, "loss": 3.2978, "step": 7666 }, { "epoch": 0.47594512384381404, "grad_norm": 0.49429490768067647, "learning_rate": 9.895608843039933e-05, "loss": 3.3034, "step": 7667 }, { "epoch": 0.4760072009435719, "grad_norm": 0.4722333545245277, "learning_rate": 9.895535417156583e-05, "loss": 3.2814, "step": 7668 }, { "epoch": 0.4760692780433298, "grad_norm": 0.4599777510175879, "learning_rate": 9.895461965732036e-05, "loss": 3.389, "step": 7669 }, { "epoch": 0.47613135514308774, "grad_norm": 0.48196118864232323, "learning_rate": 9.895388488766676e-05, "loss": 3.2299, "step": 7670 }, { "epoch": 0.4761934322428456, "grad_norm": 0.5647051819321903, "learning_rate": 9.895314986260887e-05, "loss": 3.3384, "step": 7671 }, { "epoch": 0.4762555093426035, "grad_norm": 0.5599477403813328, "learning_rate": 9.89524145821505e-05, "loss": 3.3713, "step": 7672 }, { "epoch": 0.47631758644236144, "grad_norm": 0.5196000247263737, "learning_rate": 9.895167904629549e-05, "loss": 3.3461, "step": 7673 }, { "epoch": 0.4763796635421193, "grad_norm": 0.5227943568664266, "learning_rate": 9.895094325504768e-05, "loss": 3.3332, "step": 7674 }, { "epoch": 0.4764417406418772, "grad_norm": 0.5752248189770434, "learning_rate": 9.895020720841094e-05, "loss": 3.4054, "step": 7675 }, { "epoch": 0.47650381774163514, "grad_norm": 0.5207791363810291, "learning_rate": 9.894947090638908e-05, "loss": 3.3299, "step": 7676 }, { "epoch": 0.476565894841393, "grad_norm": 0.5377836953605806, "learning_rate": 9.894873434898595e-05, "loss": 3.314, "step": 7677 }, { "epoch": 0.4766279719411509, "grad_norm": 0.6287952111827457, "learning_rate": 9.894799753620538e-05, "loss": 3.3275, "step": 7678 }, { "epoch": 0.47669004904090884, "grad_norm": 0.6430094236763088, "learning_rate": 9.894726046805123e-05, "loss": 3.3515, "step": 7679 }, { "epoch": 0.4767521261406667, "grad_norm": 0.6846911314219878, "learning_rate": 9.894652314452734e-05, "loss": 3.3653, "step": 7680 }, { "epoch": 0.4768142032404246, "grad_norm": 0.60173839623199, "learning_rate": 9.894578556563756e-05, "loss": 3.3214, "step": 7681 }, { "epoch": 0.4768762803401825, "grad_norm": 0.47175485395088324, "learning_rate": 9.894504773138573e-05, "loss": 3.3143, "step": 7682 }, { "epoch": 0.4769383574399404, "grad_norm": 0.5023385138112691, "learning_rate": 9.89443096417757e-05, "loss": 3.32, "step": 7683 }, { "epoch": 0.4770004345396983, "grad_norm": 0.5027637985855001, "learning_rate": 9.894357129681134e-05, "loss": 3.3226, "step": 7684 }, { "epoch": 0.4770625116394562, "grad_norm": 0.5239043360264759, "learning_rate": 9.894283269649647e-05, "loss": 3.2872, "step": 7685 }, { "epoch": 0.4771245887392141, "grad_norm": 0.5047864397696459, "learning_rate": 9.894209384083497e-05, "loss": 3.3511, "step": 7686 }, { "epoch": 0.477186665838972, "grad_norm": 0.5473573192642661, "learning_rate": 9.894135472983069e-05, "loss": 3.3708, "step": 7687 }, { "epoch": 0.4772487429387299, "grad_norm": 0.5331366081807578, "learning_rate": 9.894061536348746e-05, "loss": 3.3671, "step": 7688 }, { "epoch": 0.4773108200384878, "grad_norm": 0.4702084898395585, "learning_rate": 9.893987574180918e-05, "loss": 3.3401, "step": 7689 }, { "epoch": 0.4773728971382457, "grad_norm": 0.47221839738057275, "learning_rate": 9.893913586479969e-05, "loss": 3.3205, "step": 7690 }, { "epoch": 0.4774349742380036, "grad_norm": 0.4483535063233715, "learning_rate": 9.893839573246284e-05, "loss": 3.2512, "step": 7691 }, { "epoch": 0.4774970513377615, "grad_norm": 0.5147660933466529, "learning_rate": 9.89376553448025e-05, "loss": 3.3546, "step": 7692 }, { "epoch": 0.4775591284375194, "grad_norm": 0.5481872074403462, "learning_rate": 9.893691470182252e-05, "loss": 3.3566, "step": 7693 }, { "epoch": 0.47762120553727727, "grad_norm": 0.5755458469622469, "learning_rate": 9.89361738035268e-05, "loss": 3.326, "step": 7694 }, { "epoch": 0.4776832826370352, "grad_norm": 0.5555580185464728, "learning_rate": 9.893543264991916e-05, "loss": 3.2618, "step": 7695 }, { "epoch": 0.4777453597367931, "grad_norm": 0.5622019773000675, "learning_rate": 9.89346912410035e-05, "loss": 3.2973, "step": 7696 }, { "epoch": 0.47780743683655097, "grad_norm": 0.5436112972364896, "learning_rate": 9.893394957678367e-05, "loss": 3.2656, "step": 7697 }, { "epoch": 0.4778695139363089, "grad_norm": 0.4990800948555723, "learning_rate": 9.893320765726355e-05, "loss": 3.3246, "step": 7698 }, { "epoch": 0.4779315910360668, "grad_norm": 0.5247614329847736, "learning_rate": 9.893246548244701e-05, "loss": 3.3432, "step": 7699 }, { "epoch": 0.47799366813582467, "grad_norm": 0.5085869385809606, "learning_rate": 9.893172305233791e-05, "loss": 3.2318, "step": 7700 }, { "epoch": 0.4780557452355826, "grad_norm": 0.5264661197013585, "learning_rate": 9.893098036694014e-05, "loss": 3.3292, "step": 7701 }, { "epoch": 0.4781178223353405, "grad_norm": 0.6199923652517251, "learning_rate": 9.893023742625756e-05, "loss": 3.3584, "step": 7702 }, { "epoch": 0.47817989943509837, "grad_norm": 0.6147657032233462, "learning_rate": 9.892949423029405e-05, "loss": 3.2665, "step": 7703 }, { "epoch": 0.4782419765348563, "grad_norm": 0.5753756695349206, "learning_rate": 9.892875077905351e-05, "loss": 3.3465, "step": 7704 }, { "epoch": 0.4783040536346142, "grad_norm": 0.6166140391042388, "learning_rate": 9.892800707253979e-05, "loss": 3.3755, "step": 7705 }, { "epoch": 0.47836613073437206, "grad_norm": 0.6372471354642039, "learning_rate": 9.892726311075677e-05, "loss": 3.3497, "step": 7706 }, { "epoch": 0.47842820783413, "grad_norm": 0.5985064474282797, "learning_rate": 9.892651889370834e-05, "loss": 3.3172, "step": 7707 }, { "epoch": 0.4784902849338879, "grad_norm": 0.6376194346502307, "learning_rate": 9.892577442139839e-05, "loss": 3.3232, "step": 7708 }, { "epoch": 0.47855236203364576, "grad_norm": 0.5648065800130201, "learning_rate": 9.89250296938308e-05, "loss": 3.3041, "step": 7709 }, { "epoch": 0.4786144391334037, "grad_norm": 0.5521265198952445, "learning_rate": 9.892428471100945e-05, "loss": 3.3043, "step": 7710 }, { "epoch": 0.4786765162331616, "grad_norm": 0.5994141102303255, "learning_rate": 9.892353947293825e-05, "loss": 3.4019, "step": 7711 }, { "epoch": 0.47873859333291946, "grad_norm": 0.42370960670417646, "learning_rate": 9.892279397962104e-05, "loss": 3.305, "step": 7712 }, { "epoch": 0.4788006704326774, "grad_norm": 0.5234960267099256, "learning_rate": 9.892204823106175e-05, "loss": 3.3235, "step": 7713 }, { "epoch": 0.4788627475324353, "grad_norm": 0.498744879387978, "learning_rate": 9.892130222726426e-05, "loss": 3.2724, "step": 7714 }, { "epoch": 0.47892482463219316, "grad_norm": 0.42329267439240537, "learning_rate": 9.892055596823244e-05, "loss": 3.3329, "step": 7715 }, { "epoch": 0.4789869017319511, "grad_norm": 0.4870639122103348, "learning_rate": 9.891980945397022e-05, "loss": 3.2426, "step": 7716 }, { "epoch": 0.479048978831709, "grad_norm": 0.5091463324622165, "learning_rate": 9.891906268448147e-05, "loss": 3.3908, "step": 7717 }, { "epoch": 0.47911105593146686, "grad_norm": 0.5835603431219741, "learning_rate": 9.891831565977011e-05, "loss": 3.3859, "step": 7718 }, { "epoch": 0.4791731330312248, "grad_norm": 0.4760527620615451, "learning_rate": 9.891756837984001e-05, "loss": 3.3569, "step": 7719 }, { "epoch": 0.4792352101309827, "grad_norm": 0.5402243711878955, "learning_rate": 9.891682084469508e-05, "loss": 3.2913, "step": 7720 }, { "epoch": 0.47929728723074055, "grad_norm": 0.5045660268376084, "learning_rate": 9.891607305433923e-05, "loss": 3.2644, "step": 7721 }, { "epoch": 0.47935936433049847, "grad_norm": 0.507950309247376, "learning_rate": 9.891532500877635e-05, "loss": 3.286, "step": 7722 }, { "epoch": 0.4794214414302564, "grad_norm": 0.5568857289146645, "learning_rate": 9.891457670801035e-05, "loss": 3.276, "step": 7723 }, { "epoch": 0.47948351853001425, "grad_norm": 0.5036233802953046, "learning_rate": 9.89138281520451e-05, "loss": 3.3211, "step": 7724 }, { "epoch": 0.47954559562977217, "grad_norm": 0.5061318977961835, "learning_rate": 9.891307934088456e-05, "loss": 3.3367, "step": 7725 }, { "epoch": 0.4796076727295301, "grad_norm": 0.46860915293218774, "learning_rate": 9.891233027453262e-05, "loss": 3.2446, "step": 7726 }, { "epoch": 0.47966974982928795, "grad_norm": 0.5058218157754234, "learning_rate": 9.891158095299315e-05, "loss": 3.2764, "step": 7727 }, { "epoch": 0.47973182692904587, "grad_norm": 0.5378177202803696, "learning_rate": 9.891083137627011e-05, "loss": 3.2801, "step": 7728 }, { "epoch": 0.4797939040288038, "grad_norm": 0.5590681449131778, "learning_rate": 9.891008154436738e-05, "loss": 3.3085, "step": 7729 }, { "epoch": 0.47985598112856165, "grad_norm": 0.5229056488892663, "learning_rate": 9.890933145728889e-05, "loss": 3.1723, "step": 7730 }, { "epoch": 0.47991805822831957, "grad_norm": 0.4640572088434431, "learning_rate": 9.890858111503853e-05, "loss": 3.2937, "step": 7731 }, { "epoch": 0.4799801353280775, "grad_norm": 0.48450142225798826, "learning_rate": 9.890783051762025e-05, "loss": 3.3886, "step": 7732 }, { "epoch": 0.48004221242783535, "grad_norm": 0.645365065865214, "learning_rate": 9.890707966503793e-05, "loss": 3.2892, "step": 7733 }, { "epoch": 0.48010428952759326, "grad_norm": 0.7130197486469795, "learning_rate": 9.89063285572955e-05, "loss": 3.3817, "step": 7734 }, { "epoch": 0.4801663666273512, "grad_norm": 0.5833365377837676, "learning_rate": 9.89055771943969e-05, "loss": 3.3008, "step": 7735 }, { "epoch": 0.48022844372710904, "grad_norm": 0.5517678859094971, "learning_rate": 9.890482557634602e-05, "loss": 3.2796, "step": 7736 }, { "epoch": 0.48029052082686696, "grad_norm": 0.6337204059861548, "learning_rate": 9.890407370314678e-05, "loss": 3.2933, "step": 7737 }, { "epoch": 0.4803525979266249, "grad_norm": 0.5892804974370462, "learning_rate": 9.890332157480313e-05, "loss": 3.3147, "step": 7738 }, { "epoch": 0.48041467502638274, "grad_norm": 0.5813614236004734, "learning_rate": 9.890256919131898e-05, "loss": 3.2517, "step": 7739 }, { "epoch": 0.48047675212614066, "grad_norm": 0.5862590161994652, "learning_rate": 9.890181655269825e-05, "loss": 3.2358, "step": 7740 }, { "epoch": 0.4805388292258986, "grad_norm": 0.5535926759834864, "learning_rate": 9.890106365894488e-05, "loss": 3.25, "step": 7741 }, { "epoch": 0.48060090632565644, "grad_norm": 0.6075009113665993, "learning_rate": 9.890031051006278e-05, "loss": 3.3327, "step": 7742 }, { "epoch": 0.48066298342541436, "grad_norm": 0.5456402872834499, "learning_rate": 9.88995571060559e-05, "loss": 3.3234, "step": 7743 }, { "epoch": 0.4807250605251723, "grad_norm": 0.6126058710813258, "learning_rate": 9.889880344692814e-05, "loss": 3.3984, "step": 7744 }, { "epoch": 0.48078713762493014, "grad_norm": 0.6119745998347443, "learning_rate": 9.889804953268347e-05, "loss": 3.3398, "step": 7745 }, { "epoch": 0.48084921472468806, "grad_norm": 0.5243722341325856, "learning_rate": 9.88972953633258e-05, "loss": 3.2524, "step": 7746 }, { "epoch": 0.480911291824446, "grad_norm": 0.5087884767496028, "learning_rate": 9.889654093885908e-05, "loss": 3.2596, "step": 7747 }, { "epoch": 0.48097336892420384, "grad_norm": 0.45933091088298905, "learning_rate": 9.889578625928722e-05, "loss": 3.3003, "step": 7748 }, { "epoch": 0.48103544602396175, "grad_norm": 0.5125934027249877, "learning_rate": 9.889503132461417e-05, "loss": 3.3561, "step": 7749 }, { "epoch": 0.4810975231237197, "grad_norm": 0.5341538836228649, "learning_rate": 9.889427613484388e-05, "loss": 3.3368, "step": 7750 }, { "epoch": 0.48115960022347753, "grad_norm": 0.5184114477453622, "learning_rate": 9.889352068998028e-05, "loss": 3.3367, "step": 7751 }, { "epoch": 0.48122167732323545, "grad_norm": 0.4830391058866961, "learning_rate": 9.889276499002731e-05, "loss": 3.324, "step": 7752 }, { "epoch": 0.48128375442299337, "grad_norm": 0.52576427574022, "learning_rate": 9.889200903498893e-05, "loss": 3.2088, "step": 7753 }, { "epoch": 0.48134583152275123, "grad_norm": 0.5958154791266598, "learning_rate": 9.889125282486906e-05, "loss": 3.3039, "step": 7754 }, { "epoch": 0.48140790862250915, "grad_norm": 0.568334651029109, "learning_rate": 9.889049635967166e-05, "loss": 3.2703, "step": 7755 }, { "epoch": 0.48146998572226707, "grad_norm": 0.5097372405235879, "learning_rate": 9.888973963940067e-05, "loss": 3.3859, "step": 7756 }, { "epoch": 0.48153206282202493, "grad_norm": 0.505825763307432, "learning_rate": 9.888898266406003e-05, "loss": 3.298, "step": 7757 }, { "epoch": 0.48159413992178285, "grad_norm": 0.543242927320902, "learning_rate": 9.888822543365371e-05, "loss": 3.2943, "step": 7758 }, { "epoch": 0.48165621702154077, "grad_norm": 0.58694457928476, "learning_rate": 9.888746794818564e-05, "loss": 3.3056, "step": 7759 }, { "epoch": 0.48171829412129863, "grad_norm": 0.5421270804036555, "learning_rate": 9.88867102076598e-05, "loss": 3.1998, "step": 7760 }, { "epoch": 0.48178037122105655, "grad_norm": 0.47445300944725927, "learning_rate": 9.888595221208011e-05, "loss": 3.3364, "step": 7761 }, { "epoch": 0.48184244832081446, "grad_norm": 0.5684778673330941, "learning_rate": 9.888519396145055e-05, "loss": 3.3287, "step": 7762 }, { "epoch": 0.4819045254205723, "grad_norm": 0.609771135147682, "learning_rate": 9.888443545577506e-05, "loss": 3.2542, "step": 7763 }, { "epoch": 0.48196660252033025, "grad_norm": 0.4933478588895574, "learning_rate": 9.888367669505761e-05, "loss": 3.2395, "step": 7764 }, { "epoch": 0.48202867962008816, "grad_norm": 0.5739649516680052, "learning_rate": 9.888291767930213e-05, "loss": 3.329, "step": 7765 }, { "epoch": 0.482090756719846, "grad_norm": 0.5665102106909801, "learning_rate": 9.888215840851263e-05, "loss": 3.3347, "step": 7766 }, { "epoch": 0.48215283381960394, "grad_norm": 0.594167763154818, "learning_rate": 9.888139888269302e-05, "loss": 3.2003, "step": 7767 }, { "epoch": 0.48221491091936186, "grad_norm": 0.6983364228367651, "learning_rate": 9.88806391018473e-05, "loss": 3.3711, "step": 7768 }, { "epoch": 0.4822769880191197, "grad_norm": 0.6080071429363207, "learning_rate": 9.887987906597942e-05, "loss": 3.25, "step": 7769 }, { "epoch": 0.48233906511887764, "grad_norm": 0.5881406689541967, "learning_rate": 9.887911877509333e-05, "loss": 3.2228, "step": 7770 }, { "epoch": 0.48240114221863556, "grad_norm": 0.5430593620821638, "learning_rate": 9.887835822919302e-05, "loss": 3.2139, "step": 7771 }, { "epoch": 0.4824632193183934, "grad_norm": 0.523603384577607, "learning_rate": 9.887759742828246e-05, "loss": 3.3117, "step": 7772 }, { "epoch": 0.48252529641815134, "grad_norm": 0.7165455806446565, "learning_rate": 9.88768363723656e-05, "loss": 3.2955, "step": 7773 }, { "epoch": 0.48258737351790926, "grad_norm": 0.659053626545643, "learning_rate": 9.887607506144642e-05, "loss": 3.3226, "step": 7774 }, { "epoch": 0.4826494506176671, "grad_norm": 0.6107380368220081, "learning_rate": 9.887531349552888e-05, "loss": 3.4013, "step": 7775 }, { "epoch": 0.48271152771742504, "grad_norm": 0.4762059738555104, "learning_rate": 9.887455167461697e-05, "loss": 3.316, "step": 7776 }, { "epoch": 0.48277360481718296, "grad_norm": 0.5455142940323299, "learning_rate": 9.887378959871467e-05, "loss": 3.2618, "step": 7777 }, { "epoch": 0.4828356819169408, "grad_norm": 0.6507879424008638, "learning_rate": 9.887302726782595e-05, "loss": 3.3834, "step": 7778 }, { "epoch": 0.48289775901669874, "grad_norm": 0.6376719448604313, "learning_rate": 9.887226468195476e-05, "loss": 3.3529, "step": 7779 }, { "epoch": 0.48295983611645665, "grad_norm": 0.598005828345803, "learning_rate": 9.887150184110511e-05, "loss": 3.2298, "step": 7780 }, { "epoch": 0.4830219132162145, "grad_norm": 0.5181136042373605, "learning_rate": 9.887073874528099e-05, "loss": 3.2661, "step": 7781 }, { "epoch": 0.48308399031597243, "grad_norm": 0.5796994601350001, "learning_rate": 9.886997539448634e-05, "loss": 3.3588, "step": 7782 }, { "epoch": 0.48314606741573035, "grad_norm": 0.5164035064116528, "learning_rate": 9.886921178872517e-05, "loss": 3.3003, "step": 7783 }, { "epoch": 0.4832081445154882, "grad_norm": 0.5711758754142972, "learning_rate": 9.886844792800146e-05, "loss": 3.3061, "step": 7784 }, { "epoch": 0.48327022161524613, "grad_norm": 0.5935276267588286, "learning_rate": 9.886768381231919e-05, "loss": 3.2826, "step": 7785 }, { "epoch": 0.48333229871500405, "grad_norm": 0.5881634962919421, "learning_rate": 9.886691944168236e-05, "loss": 3.3444, "step": 7786 }, { "epoch": 0.4833943758147619, "grad_norm": 0.5604249243381965, "learning_rate": 9.886615481609493e-05, "loss": 3.3477, "step": 7787 }, { "epoch": 0.48345645291451983, "grad_norm": 0.5626316181480956, "learning_rate": 9.886538993556093e-05, "loss": 3.3753, "step": 7788 }, { "epoch": 0.48351853001427775, "grad_norm": 0.6092773016837745, "learning_rate": 9.886462480008432e-05, "loss": 3.376, "step": 7789 }, { "epoch": 0.4835806071140356, "grad_norm": 0.5681569633682931, "learning_rate": 9.886385940966909e-05, "loss": 3.3169, "step": 7790 }, { "epoch": 0.48364268421379353, "grad_norm": 0.4588985119311092, "learning_rate": 9.886309376431926e-05, "loss": 3.2562, "step": 7791 }, { "epoch": 0.48370476131355145, "grad_norm": 0.5074586368955271, "learning_rate": 9.886232786403879e-05, "loss": 3.2538, "step": 7792 }, { "epoch": 0.4837668384133093, "grad_norm": 0.6002767619943721, "learning_rate": 9.886156170883171e-05, "loss": 3.3984, "step": 7793 }, { "epoch": 0.4838289155130672, "grad_norm": 0.5891395811854, "learning_rate": 9.886079529870199e-05, "loss": 3.2851, "step": 7794 }, { "epoch": 0.48389099261282514, "grad_norm": 0.5656914772043077, "learning_rate": 9.886002863365365e-05, "loss": 3.3127, "step": 7795 }, { "epoch": 0.483953069712583, "grad_norm": 0.5174144558070096, "learning_rate": 9.885926171369066e-05, "loss": 3.264, "step": 7796 }, { "epoch": 0.4840151468123409, "grad_norm": 0.562428602349814, "learning_rate": 9.885849453881706e-05, "loss": 3.3488, "step": 7797 }, { "epoch": 0.48407722391209884, "grad_norm": 0.546208738737135, "learning_rate": 9.885772710903682e-05, "loss": 3.3672, "step": 7798 }, { "epoch": 0.4841393010118567, "grad_norm": 0.5226083422772745, "learning_rate": 9.885695942435397e-05, "loss": 3.2737, "step": 7799 }, { "epoch": 0.4842013781116146, "grad_norm": 0.5343183375874226, "learning_rate": 9.88561914847725e-05, "loss": 3.3156, "step": 7800 }, { "epoch": 0.48426345521137254, "grad_norm": 0.5274754112623536, "learning_rate": 9.885542329029642e-05, "loss": 3.2796, "step": 7801 }, { "epoch": 0.4843255323111304, "grad_norm": 0.5227875731295192, "learning_rate": 9.885465484092973e-05, "loss": 3.3262, "step": 7802 }, { "epoch": 0.4843876094108883, "grad_norm": 0.5150268148181857, "learning_rate": 9.885388613667646e-05, "loss": 3.3063, "step": 7803 }, { "epoch": 0.48444968651064624, "grad_norm": 0.5037807337848969, "learning_rate": 9.885311717754058e-05, "loss": 3.198, "step": 7804 }, { "epoch": 0.4845117636104041, "grad_norm": 0.4891628850951019, "learning_rate": 9.885234796352614e-05, "loss": 3.2023, "step": 7805 }, { "epoch": 0.484573840710162, "grad_norm": 0.5788644494990028, "learning_rate": 9.885157849463716e-05, "loss": 3.3667, "step": 7806 }, { "epoch": 0.48463591780991994, "grad_norm": 0.585432598014892, "learning_rate": 9.88508087708776e-05, "loss": 3.2705, "step": 7807 }, { "epoch": 0.4846979949096778, "grad_norm": 0.6050204678690857, "learning_rate": 9.885003879225155e-05, "loss": 3.3841, "step": 7808 }, { "epoch": 0.4847600720094357, "grad_norm": 0.48961353885188164, "learning_rate": 9.884926855876296e-05, "loss": 3.3118, "step": 7809 }, { "epoch": 0.48482214910919363, "grad_norm": 0.5155014988352967, "learning_rate": 9.884849807041588e-05, "loss": 3.2955, "step": 7810 }, { "epoch": 0.4848842262089515, "grad_norm": 0.5465468648393974, "learning_rate": 9.884772732721435e-05, "loss": 3.3199, "step": 7811 }, { "epoch": 0.4849463033087094, "grad_norm": 0.49729942632416174, "learning_rate": 9.884695632916234e-05, "loss": 3.3094, "step": 7812 }, { "epoch": 0.48500838040846733, "grad_norm": 0.4881027658407607, "learning_rate": 9.884618507626391e-05, "loss": 3.3176, "step": 7813 }, { "epoch": 0.4850704575082252, "grad_norm": 0.5070928315793592, "learning_rate": 9.884541356852308e-05, "loss": 3.2673, "step": 7814 }, { "epoch": 0.4851325346079831, "grad_norm": 0.4959004844211613, "learning_rate": 9.884464180594387e-05, "loss": 3.3692, "step": 7815 }, { "epoch": 0.48519461170774103, "grad_norm": 0.5508063560878163, "learning_rate": 9.88438697885303e-05, "loss": 3.2858, "step": 7816 }, { "epoch": 0.4852566888074989, "grad_norm": 0.5353387117844098, "learning_rate": 9.884309751628642e-05, "loss": 3.3058, "step": 7817 }, { "epoch": 0.4853187659072568, "grad_norm": 0.595294599689485, "learning_rate": 9.884232498921622e-05, "loss": 3.2968, "step": 7818 }, { "epoch": 0.48538084300701473, "grad_norm": 0.6121420271586944, "learning_rate": 9.884155220732376e-05, "loss": 3.3006, "step": 7819 }, { "epoch": 0.4854429201067726, "grad_norm": 0.46868115555339906, "learning_rate": 9.884077917061308e-05, "loss": 3.2643, "step": 7820 }, { "epoch": 0.4855049972065305, "grad_norm": 0.4928567967798665, "learning_rate": 9.884000587908819e-05, "loss": 3.2183, "step": 7821 }, { "epoch": 0.4855670743062884, "grad_norm": 0.5074041632188656, "learning_rate": 9.883923233275313e-05, "loss": 3.3156, "step": 7822 }, { "epoch": 0.4856291514060463, "grad_norm": 0.5338625341240375, "learning_rate": 9.883845853161195e-05, "loss": 3.2082, "step": 7823 }, { "epoch": 0.4856912285058042, "grad_norm": 0.5435875035974489, "learning_rate": 9.883768447566865e-05, "loss": 3.3222, "step": 7824 }, { "epoch": 0.4857533056055621, "grad_norm": 0.5642979022368844, "learning_rate": 9.883691016492733e-05, "loss": 3.2951, "step": 7825 }, { "epoch": 0.48581538270532, "grad_norm": 0.5616559477650179, "learning_rate": 9.883613559939197e-05, "loss": 3.3164, "step": 7826 }, { "epoch": 0.4858774598050779, "grad_norm": 0.45918167418874006, "learning_rate": 9.883536077906663e-05, "loss": 3.22, "step": 7827 }, { "epoch": 0.4859395369048358, "grad_norm": 0.5563225088115928, "learning_rate": 9.883458570395537e-05, "loss": 3.2726, "step": 7828 }, { "epoch": 0.4860016140045937, "grad_norm": 0.6434975860686771, "learning_rate": 9.883381037406223e-05, "loss": 3.314, "step": 7829 }, { "epoch": 0.4860636911043516, "grad_norm": 0.5827702944203068, "learning_rate": 9.883303478939124e-05, "loss": 3.3528, "step": 7830 }, { "epoch": 0.4861257682041095, "grad_norm": 0.5381796421216747, "learning_rate": 9.883225894994644e-05, "loss": 3.383, "step": 7831 }, { "epoch": 0.4861878453038674, "grad_norm": 0.5400575493240315, "learning_rate": 9.88314828557319e-05, "loss": 3.3228, "step": 7832 }, { "epoch": 0.4862499224036253, "grad_norm": 0.6067315253976521, "learning_rate": 9.883070650675166e-05, "loss": 3.3471, "step": 7833 }, { "epoch": 0.4863119995033832, "grad_norm": 0.5803780599115387, "learning_rate": 9.882992990300976e-05, "loss": 3.1762, "step": 7834 }, { "epoch": 0.4863740766031411, "grad_norm": 0.514185717346729, "learning_rate": 9.882915304451028e-05, "loss": 3.3077, "step": 7835 }, { "epoch": 0.486436153702899, "grad_norm": 0.5461511726368202, "learning_rate": 9.882837593125724e-05, "loss": 3.2252, "step": 7836 }, { "epoch": 0.4864982308026569, "grad_norm": 0.47098031330980095, "learning_rate": 9.882759856325471e-05, "loss": 3.2706, "step": 7837 }, { "epoch": 0.4865603079024148, "grad_norm": 0.4812352958513585, "learning_rate": 9.882682094050674e-05, "loss": 3.3057, "step": 7838 }, { "epoch": 0.4866223850021727, "grad_norm": 0.49595332223210403, "learning_rate": 9.882604306301742e-05, "loss": 3.3981, "step": 7839 }, { "epoch": 0.4866844621019306, "grad_norm": 0.6229301749926199, "learning_rate": 9.882526493079076e-05, "loss": 3.2995, "step": 7840 }, { "epoch": 0.4867465392016885, "grad_norm": 0.6025060462252568, "learning_rate": 9.882448654383085e-05, "loss": 3.2738, "step": 7841 }, { "epoch": 0.4868086163014464, "grad_norm": 0.4616236671100032, "learning_rate": 9.882370790214173e-05, "loss": 3.248, "step": 7842 }, { "epoch": 0.4868706934012043, "grad_norm": 0.599836523912273, "learning_rate": 9.882292900572747e-05, "loss": 3.2793, "step": 7843 }, { "epoch": 0.4869327705009622, "grad_norm": 0.5356525152788489, "learning_rate": 9.882214985459215e-05, "loss": 3.3611, "step": 7844 }, { "epoch": 0.4869948476007201, "grad_norm": 0.492976192533048, "learning_rate": 9.882137044873982e-05, "loss": 3.1565, "step": 7845 }, { "epoch": 0.487056924700478, "grad_norm": 0.5253628578080802, "learning_rate": 9.882059078817455e-05, "loss": 3.2553, "step": 7846 }, { "epoch": 0.4871190018002359, "grad_norm": 0.5251759938027493, "learning_rate": 9.88198108729004e-05, "loss": 3.2633, "step": 7847 }, { "epoch": 0.4871810788999938, "grad_norm": 0.4652801997167583, "learning_rate": 9.881903070292145e-05, "loss": 3.3168, "step": 7848 }, { "epoch": 0.4872431559997517, "grad_norm": 0.5383885635127048, "learning_rate": 9.881825027824175e-05, "loss": 3.3315, "step": 7849 }, { "epoch": 0.48730523309950957, "grad_norm": 0.5804236567811843, "learning_rate": 9.88174695988654e-05, "loss": 3.321, "step": 7850 }, { "epoch": 0.4873673101992675, "grad_norm": 0.4856099413840815, "learning_rate": 9.881668866479645e-05, "loss": 3.2401, "step": 7851 }, { "epoch": 0.4874293872990254, "grad_norm": 0.5449835082717435, "learning_rate": 9.881590747603901e-05, "loss": 3.2274, "step": 7852 }, { "epoch": 0.48749146439878327, "grad_norm": 0.5392908359075065, "learning_rate": 9.88151260325971e-05, "loss": 3.3812, "step": 7853 }, { "epoch": 0.4875535414985412, "grad_norm": 0.4848354212506262, "learning_rate": 9.881434433447485e-05, "loss": 3.3978, "step": 7854 }, { "epoch": 0.4876156185982991, "grad_norm": 0.49475532866774913, "learning_rate": 9.881356238167632e-05, "loss": 3.3436, "step": 7855 }, { "epoch": 0.48767769569805697, "grad_norm": 0.5394287799974143, "learning_rate": 9.881278017420557e-05, "loss": 3.3665, "step": 7856 }, { "epoch": 0.4877397727978149, "grad_norm": 0.5378540148496632, "learning_rate": 9.881199771206669e-05, "loss": 3.2376, "step": 7857 }, { "epoch": 0.4878018498975728, "grad_norm": 0.4897880498194667, "learning_rate": 9.881121499526378e-05, "loss": 3.3368, "step": 7858 }, { "epoch": 0.48786392699733067, "grad_norm": 0.5133379006440647, "learning_rate": 9.88104320238009e-05, "loss": 3.3399, "step": 7859 }, { "epoch": 0.4879260040970886, "grad_norm": 0.4751879615074003, "learning_rate": 9.880964879768216e-05, "loss": 3.3488, "step": 7860 }, { "epoch": 0.4879880811968465, "grad_norm": 0.4640059414187226, "learning_rate": 9.880886531691162e-05, "loss": 3.2795, "step": 7861 }, { "epoch": 0.48805015829660436, "grad_norm": 0.5366240094342565, "learning_rate": 9.88080815814934e-05, "loss": 3.3262, "step": 7862 }, { "epoch": 0.4881122353963623, "grad_norm": 0.49594831963258396, "learning_rate": 9.880729759143156e-05, "loss": 3.285, "step": 7863 }, { "epoch": 0.4881743124961202, "grad_norm": 0.5172422074716888, "learning_rate": 9.88065133467302e-05, "loss": 3.3549, "step": 7864 }, { "epoch": 0.48823638959587806, "grad_norm": 0.567238268931417, "learning_rate": 9.880572884739341e-05, "loss": 3.2473, "step": 7865 }, { "epoch": 0.488298466695636, "grad_norm": 0.5683317144945121, "learning_rate": 9.880494409342529e-05, "loss": 3.2029, "step": 7866 }, { "epoch": 0.4883605437953939, "grad_norm": 0.5619544456590513, "learning_rate": 9.880415908482992e-05, "loss": 3.3452, "step": 7867 }, { "epoch": 0.48842262089515176, "grad_norm": 0.5452240683656484, "learning_rate": 9.880337382161142e-05, "loss": 3.3051, "step": 7868 }, { "epoch": 0.4884846979949097, "grad_norm": 0.6134516485353917, "learning_rate": 9.880258830377385e-05, "loss": 3.3277, "step": 7869 }, { "epoch": 0.4885467750946676, "grad_norm": 0.5775903342963883, "learning_rate": 9.880180253132134e-05, "loss": 3.299, "step": 7870 }, { "epoch": 0.48860885219442546, "grad_norm": 0.6343663999604073, "learning_rate": 9.8801016504258e-05, "loss": 3.4518, "step": 7871 }, { "epoch": 0.4886709292941834, "grad_norm": 0.5701064429948349, "learning_rate": 9.880023022258788e-05, "loss": 3.3115, "step": 7872 }, { "epoch": 0.4887330063939413, "grad_norm": 0.5037481666228675, "learning_rate": 9.879944368631512e-05, "loss": 3.3117, "step": 7873 }, { "epoch": 0.48879508349369916, "grad_norm": 0.5041552675025166, "learning_rate": 9.879865689544381e-05, "loss": 3.2817, "step": 7874 }, { "epoch": 0.4888571605934571, "grad_norm": 0.548620463008551, "learning_rate": 9.879786984997808e-05, "loss": 3.4191, "step": 7875 }, { "epoch": 0.488919237693215, "grad_norm": 0.5453615773149444, "learning_rate": 9.8797082549922e-05, "loss": 3.2938, "step": 7876 }, { "epoch": 0.48898131479297285, "grad_norm": 0.5348584536894722, "learning_rate": 9.879629499527969e-05, "loss": 3.2859, "step": 7877 }, { "epoch": 0.48904339189273077, "grad_norm": 0.4830553783375963, "learning_rate": 9.879550718605529e-05, "loss": 3.1606, "step": 7878 }, { "epoch": 0.4891054689924887, "grad_norm": 0.5457308926682162, "learning_rate": 9.879471912225286e-05, "loss": 3.3488, "step": 7879 }, { "epoch": 0.48916754609224655, "grad_norm": 0.5267754211990391, "learning_rate": 9.879393080387654e-05, "loss": 3.3347, "step": 7880 }, { "epoch": 0.48922962319200447, "grad_norm": 0.5247768802182619, "learning_rate": 9.879314223093043e-05, "loss": 3.3077, "step": 7881 }, { "epoch": 0.4892917002917624, "grad_norm": 0.6254140197838521, "learning_rate": 9.879235340341867e-05, "loss": 3.284, "step": 7882 }, { "epoch": 0.48935377739152025, "grad_norm": 0.7320203977775832, "learning_rate": 9.879156432134536e-05, "loss": 3.3086, "step": 7883 }, { "epoch": 0.48941585449127817, "grad_norm": 0.6928815887925198, "learning_rate": 9.87907749847146e-05, "loss": 3.2423, "step": 7884 }, { "epoch": 0.4894779315910361, "grad_norm": 0.6061931337644815, "learning_rate": 9.878998539353053e-05, "loss": 3.257, "step": 7885 }, { "epoch": 0.48954000869079395, "grad_norm": 0.6231589859214253, "learning_rate": 9.878919554779724e-05, "loss": 3.347, "step": 7886 }, { "epoch": 0.48960208579055187, "grad_norm": 0.5605030109110262, "learning_rate": 9.87884054475189e-05, "loss": 3.3584, "step": 7887 }, { "epoch": 0.4896641628903098, "grad_norm": 0.6432548136881269, "learning_rate": 9.87876150926996e-05, "loss": 3.3584, "step": 7888 }, { "epoch": 0.48972623999006765, "grad_norm": 0.5236966644479535, "learning_rate": 9.878682448334347e-05, "loss": 3.1742, "step": 7889 }, { "epoch": 0.48978831708982556, "grad_norm": 0.5148466767154104, "learning_rate": 9.878603361945462e-05, "loss": 3.2642, "step": 7890 }, { "epoch": 0.4898503941895835, "grad_norm": 0.5425400722663728, "learning_rate": 9.878524250103719e-05, "loss": 3.2038, "step": 7891 }, { "epoch": 0.48991247128934134, "grad_norm": 0.46953252008301244, "learning_rate": 9.878445112809531e-05, "loss": 3.235, "step": 7892 }, { "epoch": 0.48997454838909926, "grad_norm": 0.457675154202094, "learning_rate": 9.87836595006331e-05, "loss": 3.3103, "step": 7893 }, { "epoch": 0.4900366254888572, "grad_norm": 0.4766426019107055, "learning_rate": 9.878286761865471e-05, "loss": 3.2736, "step": 7894 }, { "epoch": 0.49009870258861504, "grad_norm": 0.5010132060490003, "learning_rate": 9.878207548216426e-05, "loss": 3.2751, "step": 7895 }, { "epoch": 0.49016077968837296, "grad_norm": 0.46923273978399827, "learning_rate": 9.878128309116588e-05, "loss": 3.2826, "step": 7896 }, { "epoch": 0.4902228567881309, "grad_norm": 0.4754470841820928, "learning_rate": 9.878049044566368e-05, "loss": 3.3319, "step": 7897 }, { "epoch": 0.49028493388788874, "grad_norm": 0.5836197853664274, "learning_rate": 9.877969754566185e-05, "loss": 3.3074, "step": 7898 }, { "epoch": 0.49034701098764666, "grad_norm": 0.49409469571975184, "learning_rate": 9.877890439116448e-05, "loss": 3.3551, "step": 7899 }, { "epoch": 0.4904090880874046, "grad_norm": 0.47038400683065834, "learning_rate": 9.877811098217573e-05, "loss": 3.3232, "step": 7900 }, { "epoch": 0.49047116518716244, "grad_norm": 0.46207034993427515, "learning_rate": 9.877731731869972e-05, "loss": 3.3058, "step": 7901 }, { "epoch": 0.49053324228692036, "grad_norm": 0.4477225555792835, "learning_rate": 9.877652340074061e-05, "loss": 3.2117, "step": 7902 }, { "epoch": 0.4905953193866783, "grad_norm": 0.4717865027247512, "learning_rate": 9.877572922830256e-05, "loss": 3.3073, "step": 7903 }, { "epoch": 0.49065739648643614, "grad_norm": 0.48912966452874246, "learning_rate": 9.877493480138966e-05, "loss": 3.2855, "step": 7904 }, { "epoch": 0.49071947358619405, "grad_norm": 0.4471435880127274, "learning_rate": 9.877414012000609e-05, "loss": 3.1462, "step": 7905 }, { "epoch": 0.490781550685952, "grad_norm": 0.5147983976773429, "learning_rate": 9.8773345184156e-05, "loss": 3.3243, "step": 7906 }, { "epoch": 0.49084362778570984, "grad_norm": 0.5771059385295227, "learning_rate": 9.877254999384353e-05, "loss": 3.2074, "step": 7907 }, { "epoch": 0.49090570488546775, "grad_norm": 0.6240241538746621, "learning_rate": 9.877175454907282e-05, "loss": 3.2465, "step": 7908 }, { "epoch": 0.49096778198522567, "grad_norm": 0.7312817460692522, "learning_rate": 9.877095884984802e-05, "loss": 3.357, "step": 7909 }, { "epoch": 0.49102985908498353, "grad_norm": 0.7222028805107198, "learning_rate": 9.877016289617331e-05, "loss": 3.3096, "step": 7910 }, { "epoch": 0.49109193618474145, "grad_norm": 0.5494483619721067, "learning_rate": 9.87693666880528e-05, "loss": 3.3092, "step": 7911 }, { "epoch": 0.49115401328449937, "grad_norm": 0.5384531585400226, "learning_rate": 9.876857022549066e-05, "loss": 3.4246, "step": 7912 }, { "epoch": 0.49121609038425723, "grad_norm": 0.69757216524596, "learning_rate": 9.876777350849106e-05, "loss": 3.4073, "step": 7913 }, { "epoch": 0.49127816748401515, "grad_norm": 0.6286876132569447, "learning_rate": 9.876697653705815e-05, "loss": 3.3248, "step": 7914 }, { "epoch": 0.49134024458377307, "grad_norm": 0.5724526057552967, "learning_rate": 9.87661793111961e-05, "loss": 3.3245, "step": 7915 }, { "epoch": 0.49140232168353093, "grad_norm": 0.6688450647683498, "learning_rate": 9.876538183090902e-05, "loss": 3.3036, "step": 7916 }, { "epoch": 0.49146439878328885, "grad_norm": 0.5545673240177516, "learning_rate": 9.876458409620113e-05, "loss": 3.3598, "step": 7917 }, { "epoch": 0.49152647588304677, "grad_norm": 0.5512578538953646, "learning_rate": 9.876378610707656e-05, "loss": 3.3083, "step": 7918 }, { "epoch": 0.4915885529828046, "grad_norm": 0.5270636203029745, "learning_rate": 9.876298786353948e-05, "loss": 3.2423, "step": 7919 }, { "epoch": 0.49165063008256255, "grad_norm": 0.525457611918764, "learning_rate": 9.876218936559405e-05, "loss": 3.3242, "step": 7920 }, { "epoch": 0.49171270718232046, "grad_norm": 0.5377124521757173, "learning_rate": 9.876139061324445e-05, "loss": 3.3575, "step": 7921 }, { "epoch": 0.4917747842820783, "grad_norm": 0.5459259244655686, "learning_rate": 9.876059160649483e-05, "loss": 3.3426, "step": 7922 }, { "epoch": 0.49183686138183624, "grad_norm": 0.5670166672623415, "learning_rate": 9.875979234534936e-05, "loss": 3.3072, "step": 7923 }, { "epoch": 0.49189893848159416, "grad_norm": 0.543731976352883, "learning_rate": 9.875899282981222e-05, "loss": 3.2662, "step": 7924 }, { "epoch": 0.491961015581352, "grad_norm": 0.4384014345644622, "learning_rate": 9.87581930598876e-05, "loss": 3.2809, "step": 7925 }, { "epoch": 0.49202309268110994, "grad_norm": 0.5621299055961473, "learning_rate": 9.875739303557963e-05, "loss": 3.2681, "step": 7926 }, { "epoch": 0.49208516978086786, "grad_norm": 0.49391060402147885, "learning_rate": 9.875659275689251e-05, "loss": 3.2698, "step": 7927 }, { "epoch": 0.4921472468806257, "grad_norm": 0.46541821221129437, "learning_rate": 9.87557922238304e-05, "loss": 3.2745, "step": 7928 }, { "epoch": 0.49220932398038364, "grad_norm": 0.4270199596868005, "learning_rate": 9.87549914363975e-05, "loss": 3.1839, "step": 7929 }, { "epoch": 0.49227140108014156, "grad_norm": 0.5352497715225695, "learning_rate": 9.875419039459797e-05, "loss": 3.3178, "step": 7930 }, { "epoch": 0.4923334781798994, "grad_norm": 0.5453556207697539, "learning_rate": 9.875338909843599e-05, "loss": 3.2094, "step": 7931 }, { "epoch": 0.49239555527965734, "grad_norm": 0.5608889502976859, "learning_rate": 9.875258754791575e-05, "loss": 3.3674, "step": 7932 }, { "epoch": 0.49245763237941526, "grad_norm": 0.5165666305183, "learning_rate": 9.875178574304141e-05, "loss": 3.3239, "step": 7933 }, { "epoch": 0.4925197094791731, "grad_norm": 0.5800295465852412, "learning_rate": 9.875098368381717e-05, "loss": 3.3264, "step": 7934 }, { "epoch": 0.49258178657893104, "grad_norm": 0.5394451055130911, "learning_rate": 9.875018137024723e-05, "loss": 3.3638, "step": 7935 }, { "epoch": 0.49264386367868895, "grad_norm": 0.46173382523094, "learning_rate": 9.874937880233574e-05, "loss": 3.1291, "step": 7936 }, { "epoch": 0.4927059407784468, "grad_norm": 0.5419750863891347, "learning_rate": 9.874857598008693e-05, "loss": 3.3564, "step": 7937 }, { "epoch": 0.49276801787820473, "grad_norm": 0.4996970868458397, "learning_rate": 9.874777290350495e-05, "loss": 3.2915, "step": 7938 }, { "epoch": 0.49283009497796265, "grad_norm": 0.49817020160915776, "learning_rate": 9.874696957259401e-05, "loss": 3.1547, "step": 7939 }, { "epoch": 0.4928921720777205, "grad_norm": 0.5471283189717776, "learning_rate": 9.874616598735829e-05, "loss": 3.3268, "step": 7940 }, { "epoch": 0.49295424917747843, "grad_norm": 0.5708814944070556, "learning_rate": 9.874536214780199e-05, "loss": 3.3251, "step": 7941 }, { "epoch": 0.49301632627723635, "grad_norm": 0.47292599192708534, "learning_rate": 9.874455805392931e-05, "loss": 3.2923, "step": 7942 }, { "epoch": 0.4930784033769942, "grad_norm": 0.48509227803921195, "learning_rate": 9.874375370574442e-05, "loss": 3.3162, "step": 7943 }, { "epoch": 0.49314048047675213, "grad_norm": 0.47719564746928556, "learning_rate": 9.874294910325154e-05, "loss": 3.2407, "step": 7944 }, { "epoch": 0.49320255757651005, "grad_norm": 0.5113058098454554, "learning_rate": 9.874214424645488e-05, "loss": 3.3479, "step": 7945 }, { "epoch": 0.4932646346762679, "grad_norm": 0.5625269936336504, "learning_rate": 9.874133913535861e-05, "loss": 3.2424, "step": 7946 }, { "epoch": 0.49332671177602583, "grad_norm": 0.5710586213502764, "learning_rate": 9.874053376996694e-05, "loss": 3.2627, "step": 7947 }, { "epoch": 0.49338878887578375, "grad_norm": 0.572591304869392, "learning_rate": 9.873972815028408e-05, "loss": 3.3188, "step": 7948 }, { "epoch": 0.4934508659755416, "grad_norm": 0.4258789160469752, "learning_rate": 9.87389222763142e-05, "loss": 3.1581, "step": 7949 }, { "epoch": 0.4935129430752995, "grad_norm": 0.5700201304126888, "learning_rate": 9.873811614806156e-05, "loss": 3.3762, "step": 7950 }, { "epoch": 0.49357502017505744, "grad_norm": 0.621278411476532, "learning_rate": 9.873730976553032e-05, "loss": 3.148, "step": 7951 }, { "epoch": 0.4936370972748153, "grad_norm": 0.5754556633369888, "learning_rate": 9.873650312872472e-05, "loss": 3.3155, "step": 7952 }, { "epoch": 0.4936991743745732, "grad_norm": 0.5179641148056295, "learning_rate": 9.873569623764894e-05, "loss": 3.1677, "step": 7953 }, { "epoch": 0.49376125147433114, "grad_norm": 0.5143890885547725, "learning_rate": 9.87348890923072e-05, "loss": 3.2575, "step": 7954 }, { "epoch": 0.493823328574089, "grad_norm": 0.5573785809155706, "learning_rate": 9.873408169270372e-05, "loss": 3.245, "step": 7955 }, { "epoch": 0.4938854056738469, "grad_norm": 0.5633332121395606, "learning_rate": 9.873327403884269e-05, "loss": 3.3058, "step": 7956 }, { "epoch": 0.49394748277360484, "grad_norm": 0.5968998129914832, "learning_rate": 9.873246613072837e-05, "loss": 3.2454, "step": 7957 }, { "epoch": 0.4940095598733627, "grad_norm": 0.5751264754394363, "learning_rate": 9.873165796836492e-05, "loss": 3.2898, "step": 7958 }, { "epoch": 0.4940716369731206, "grad_norm": 0.6070638257568612, "learning_rate": 9.873084955175658e-05, "loss": 3.223, "step": 7959 }, { "epoch": 0.49413371407287854, "grad_norm": 0.557338050147947, "learning_rate": 9.873004088090758e-05, "loss": 3.3734, "step": 7960 }, { "epoch": 0.4941957911726364, "grad_norm": 0.5849584410650768, "learning_rate": 9.872923195582211e-05, "loss": 3.2949, "step": 7961 }, { "epoch": 0.4942578682723943, "grad_norm": 0.5061864789506662, "learning_rate": 9.872842277650443e-05, "loss": 3.3652, "step": 7962 }, { "epoch": 0.49431994537215224, "grad_norm": 0.5388857233745417, "learning_rate": 9.872761334295873e-05, "loss": 3.3405, "step": 7963 }, { "epoch": 0.4943820224719101, "grad_norm": 0.5346791723462312, "learning_rate": 9.872680365518922e-05, "loss": 3.2153, "step": 7964 }, { "epoch": 0.494444099571668, "grad_norm": 0.5728318729652093, "learning_rate": 9.872599371320017e-05, "loss": 3.3275, "step": 7965 }, { "epoch": 0.49450617667142593, "grad_norm": 0.5660111989861162, "learning_rate": 9.872518351699578e-05, "loss": 3.2295, "step": 7966 }, { "epoch": 0.4945682537711838, "grad_norm": 0.5420545937291328, "learning_rate": 9.872437306658027e-05, "loss": 3.3087, "step": 7967 }, { "epoch": 0.4946303308709417, "grad_norm": 0.5298423068252749, "learning_rate": 9.872356236195789e-05, "loss": 3.2514, "step": 7968 }, { "epoch": 0.49469240797069963, "grad_norm": 0.4594055353772409, "learning_rate": 9.872275140313285e-05, "loss": 3.1797, "step": 7969 }, { "epoch": 0.4947544850704575, "grad_norm": 0.5322448727319108, "learning_rate": 9.872194019010938e-05, "loss": 3.3109, "step": 7970 }, { "epoch": 0.4948165621702154, "grad_norm": 0.6356520112711511, "learning_rate": 9.872112872289174e-05, "loss": 3.3171, "step": 7971 }, { "epoch": 0.49487863926997333, "grad_norm": 0.5261504057742995, "learning_rate": 9.872031700148413e-05, "loss": 3.3526, "step": 7972 }, { "epoch": 0.4949407163697312, "grad_norm": 0.5372102954725804, "learning_rate": 9.87195050258908e-05, "loss": 3.227, "step": 7973 }, { "epoch": 0.4950027934694891, "grad_norm": 0.5157601728411468, "learning_rate": 9.871869279611598e-05, "loss": 3.257, "step": 7974 }, { "epoch": 0.49506487056924703, "grad_norm": 0.5242326305388009, "learning_rate": 9.871788031216394e-05, "loss": 3.3241, "step": 7975 }, { "epoch": 0.4951269476690049, "grad_norm": 0.5707464787846211, "learning_rate": 9.871706757403886e-05, "loss": 3.2745, "step": 7976 }, { "epoch": 0.4951890247687628, "grad_norm": 0.6071954286679039, "learning_rate": 9.871625458174503e-05, "loss": 3.3263, "step": 7977 }, { "epoch": 0.4952511018685207, "grad_norm": 0.5551794852181026, "learning_rate": 9.871544133528667e-05, "loss": 3.3426, "step": 7978 }, { "epoch": 0.4953131789682786, "grad_norm": 0.6064805467478134, "learning_rate": 9.871462783466803e-05, "loss": 3.2175, "step": 7979 }, { "epoch": 0.4953752560680365, "grad_norm": 0.4563919878503789, "learning_rate": 9.871381407989333e-05, "loss": 3.2757, "step": 7980 }, { "epoch": 0.4954373331677944, "grad_norm": 0.4751358095317883, "learning_rate": 9.871300007096686e-05, "loss": 3.3006, "step": 7981 }, { "epoch": 0.4954994102675523, "grad_norm": 0.49476835463635416, "learning_rate": 9.871218580789283e-05, "loss": 3.257, "step": 7982 }, { "epoch": 0.4955614873673102, "grad_norm": 0.4908233539260886, "learning_rate": 9.871137129067552e-05, "loss": 3.3172, "step": 7983 }, { "epoch": 0.4956235644670681, "grad_norm": 0.47909709867690947, "learning_rate": 9.871055651931914e-05, "loss": 3.1841, "step": 7984 }, { "epoch": 0.495685641566826, "grad_norm": 0.4651704489655432, "learning_rate": 9.870974149382797e-05, "loss": 3.3597, "step": 7985 }, { "epoch": 0.4957477186665839, "grad_norm": 0.4991581803950047, "learning_rate": 9.870892621420625e-05, "loss": 3.2049, "step": 7986 }, { "epoch": 0.4958097957663418, "grad_norm": 0.5017388327970974, "learning_rate": 9.870811068045824e-05, "loss": 3.2952, "step": 7987 }, { "epoch": 0.4958718728660997, "grad_norm": 0.4691181803217021, "learning_rate": 9.870729489258819e-05, "loss": 3.3112, "step": 7988 }, { "epoch": 0.4959339499658576, "grad_norm": 0.5335825593593135, "learning_rate": 9.870647885060035e-05, "loss": 3.2568, "step": 7989 }, { "epoch": 0.4959960270656155, "grad_norm": 0.5419620264957462, "learning_rate": 9.8705662554499e-05, "loss": 3.3039, "step": 7990 }, { "epoch": 0.4960581041653734, "grad_norm": 0.4731630650404862, "learning_rate": 9.870484600428839e-05, "loss": 3.2413, "step": 7991 }, { "epoch": 0.4961201812651313, "grad_norm": 0.4971019682331675, "learning_rate": 9.870402919997276e-05, "loss": 3.2921, "step": 7992 }, { "epoch": 0.4961822583648892, "grad_norm": 0.4622666221196579, "learning_rate": 9.87032121415564e-05, "loss": 3.4366, "step": 7993 }, { "epoch": 0.4962443354646471, "grad_norm": 0.550405495891467, "learning_rate": 9.870239482904354e-05, "loss": 3.2629, "step": 7994 }, { "epoch": 0.496306412564405, "grad_norm": 0.6300110205942842, "learning_rate": 9.870157726243847e-05, "loss": 3.2747, "step": 7995 }, { "epoch": 0.4963684896641629, "grad_norm": 0.5878105235738346, "learning_rate": 9.870075944174547e-05, "loss": 3.2887, "step": 7996 }, { "epoch": 0.4964305667639208, "grad_norm": 0.5328456812396217, "learning_rate": 9.869994136696876e-05, "loss": 3.241, "step": 7997 }, { "epoch": 0.4964926438636787, "grad_norm": 0.4943707348751691, "learning_rate": 9.869912303811264e-05, "loss": 3.3918, "step": 7998 }, { "epoch": 0.4965547209634366, "grad_norm": 0.583951585544446, "learning_rate": 9.869830445518139e-05, "loss": 3.2939, "step": 7999 }, { "epoch": 0.4966167980631945, "grad_norm": 0.6181407520304376, "learning_rate": 9.869748561817925e-05, "loss": 3.2854, "step": 8000 }, { "epoch": 0.4966788751629524, "grad_norm": 0.5750965493454389, "learning_rate": 9.869666652711049e-05, "loss": 3.2263, "step": 8001 }, { "epoch": 0.4967409522627103, "grad_norm": 0.5068608389968259, "learning_rate": 9.869584718197941e-05, "loss": 3.2549, "step": 8002 }, { "epoch": 0.4968030293624682, "grad_norm": 0.5780690913616997, "learning_rate": 9.869502758279028e-05, "loss": 3.2352, "step": 8003 }, { "epoch": 0.4968651064622261, "grad_norm": 0.6769900893538233, "learning_rate": 9.869420772954738e-05, "loss": 3.2283, "step": 8004 }, { "epoch": 0.496927183561984, "grad_norm": 0.5944321127406673, "learning_rate": 9.869338762225496e-05, "loss": 3.3938, "step": 8005 }, { "epoch": 0.49698926066174187, "grad_norm": 0.5061587180741447, "learning_rate": 9.869256726091731e-05, "loss": 3.2412, "step": 8006 }, { "epoch": 0.4970513377614998, "grad_norm": 0.5219405390080315, "learning_rate": 9.869174664553874e-05, "loss": 3.4199, "step": 8007 }, { "epoch": 0.4971134148612577, "grad_norm": 0.5466399914011514, "learning_rate": 9.869092577612349e-05, "loss": 3.3287, "step": 8008 }, { "epoch": 0.49717549196101557, "grad_norm": 0.6094772302021652, "learning_rate": 9.869010465267588e-05, "loss": 3.2922, "step": 8009 }, { "epoch": 0.4972375690607735, "grad_norm": 0.6356641542796452, "learning_rate": 9.868928327520015e-05, "loss": 3.3209, "step": 8010 }, { "epoch": 0.4972996461605314, "grad_norm": 0.5445437469580846, "learning_rate": 9.868846164370062e-05, "loss": 3.2766, "step": 8011 }, { "epoch": 0.49736172326028927, "grad_norm": 0.5928680997925917, "learning_rate": 9.868763975818156e-05, "loss": 3.3085, "step": 8012 }, { "epoch": 0.4974238003600472, "grad_norm": 0.5380694876647026, "learning_rate": 9.868681761864726e-05, "loss": 3.2582, "step": 8013 }, { "epoch": 0.4974858774598051, "grad_norm": 0.5590043592846019, "learning_rate": 9.868599522510202e-05, "loss": 3.276, "step": 8014 }, { "epoch": 0.49754795455956297, "grad_norm": 0.4855697724643139, "learning_rate": 9.868517257755015e-05, "loss": 3.2292, "step": 8015 }, { "epoch": 0.4976100316593209, "grad_norm": 0.47102878116235763, "learning_rate": 9.868434967599587e-05, "loss": 3.3317, "step": 8016 }, { "epoch": 0.4976721087590788, "grad_norm": 0.5122695462866689, "learning_rate": 9.868352652044355e-05, "loss": 3.2357, "step": 8017 }, { "epoch": 0.49773418585883666, "grad_norm": 0.5155271957579348, "learning_rate": 9.868270311089744e-05, "loss": 3.2626, "step": 8018 }, { "epoch": 0.4977962629585946, "grad_norm": 0.5257183315046638, "learning_rate": 9.868187944736186e-05, "loss": 3.3286, "step": 8019 }, { "epoch": 0.4978583400583525, "grad_norm": 0.4955500078410484, "learning_rate": 9.868105552984109e-05, "loss": 3.3052, "step": 8020 }, { "epoch": 0.49792041715811036, "grad_norm": 0.5308163780556728, "learning_rate": 9.868023135833944e-05, "loss": 3.2368, "step": 8021 }, { "epoch": 0.4979824942578683, "grad_norm": 0.5500612896668735, "learning_rate": 9.86794069328612e-05, "loss": 3.2597, "step": 8022 }, { "epoch": 0.4980445713576262, "grad_norm": 0.4467460579679534, "learning_rate": 9.867858225341068e-05, "loss": 3.2755, "step": 8023 }, { "epoch": 0.49810664845738406, "grad_norm": 0.47833651869186217, "learning_rate": 9.867775731999218e-05, "loss": 3.3184, "step": 8024 }, { "epoch": 0.498168725557142, "grad_norm": 0.540159463825988, "learning_rate": 9.867693213261e-05, "loss": 3.3054, "step": 8025 }, { "epoch": 0.4982308026568999, "grad_norm": 0.5444557082812423, "learning_rate": 9.867610669126845e-05, "loss": 3.2402, "step": 8026 }, { "epoch": 0.49829287975665776, "grad_norm": 0.4742034934793256, "learning_rate": 9.867528099597182e-05, "loss": 3.2241, "step": 8027 }, { "epoch": 0.4983549568564157, "grad_norm": 0.6304914978537095, "learning_rate": 9.867445504672446e-05, "loss": 3.204, "step": 8028 }, { "epoch": 0.4984170339561736, "grad_norm": 0.6025636194996424, "learning_rate": 9.867362884353064e-05, "loss": 3.3268, "step": 8029 }, { "epoch": 0.49847911105593146, "grad_norm": 0.5834380436196474, "learning_rate": 9.867280238639468e-05, "loss": 3.2885, "step": 8030 }, { "epoch": 0.4985411881556894, "grad_norm": 0.46161609741373527, "learning_rate": 9.86719756753209e-05, "loss": 3.2475, "step": 8031 }, { "epoch": 0.4986032652554473, "grad_norm": 0.551945855159628, "learning_rate": 9.86711487103136e-05, "loss": 3.2341, "step": 8032 }, { "epoch": 0.49866534235520515, "grad_norm": 0.6470435913353402, "learning_rate": 9.867032149137711e-05, "loss": 3.233, "step": 8033 }, { "epoch": 0.4987274194549631, "grad_norm": 0.5220554382511545, "learning_rate": 9.866949401851572e-05, "loss": 3.3182, "step": 8034 }, { "epoch": 0.498789496554721, "grad_norm": 0.501865394139315, "learning_rate": 9.866866629173379e-05, "loss": 3.3495, "step": 8035 }, { "epoch": 0.49885157365447885, "grad_norm": 0.5134554504952945, "learning_rate": 9.86678383110356e-05, "loss": 3.2825, "step": 8036 }, { "epoch": 0.49891365075423677, "grad_norm": 0.581967164678237, "learning_rate": 9.866701007642548e-05, "loss": 3.2199, "step": 8037 }, { "epoch": 0.4989757278539947, "grad_norm": 0.4729197374599876, "learning_rate": 9.866618158790776e-05, "loss": 3.3076, "step": 8038 }, { "epoch": 0.49903780495375255, "grad_norm": 0.5593643741557429, "learning_rate": 9.866535284548676e-05, "loss": 3.3027, "step": 8039 }, { "epoch": 0.49909988205351047, "grad_norm": 0.5658793399190177, "learning_rate": 9.866452384916679e-05, "loss": 3.3872, "step": 8040 }, { "epoch": 0.4991619591532684, "grad_norm": 0.5423409379530671, "learning_rate": 9.86636945989522e-05, "loss": 3.2919, "step": 8041 }, { "epoch": 0.49922403625302625, "grad_norm": 0.6028194918704107, "learning_rate": 9.866286509484728e-05, "loss": 3.2685, "step": 8042 }, { "epoch": 0.49928611335278417, "grad_norm": 0.5772406541506413, "learning_rate": 9.866203533685639e-05, "loss": 3.2884, "step": 8043 }, { "epoch": 0.4993481904525421, "grad_norm": 0.5375507552557369, "learning_rate": 9.866120532498386e-05, "loss": 3.3216, "step": 8044 }, { "epoch": 0.49941026755229995, "grad_norm": 0.5817924111877469, "learning_rate": 9.8660375059234e-05, "loss": 3.3181, "step": 8045 }, { "epoch": 0.49947234465205786, "grad_norm": 0.4917642196224548, "learning_rate": 9.865954453961115e-05, "loss": 3.2176, "step": 8046 }, { "epoch": 0.4995344217518158, "grad_norm": 0.4493188331901606, "learning_rate": 9.865871376611963e-05, "loss": 3.2441, "step": 8047 }, { "epoch": 0.49959649885157364, "grad_norm": 0.5250964280536768, "learning_rate": 9.865788273876381e-05, "loss": 3.2512, "step": 8048 }, { "epoch": 0.49965857595133156, "grad_norm": 0.49439716841598996, "learning_rate": 9.8657051457548e-05, "loss": 3.2377, "step": 8049 }, { "epoch": 0.4997206530510895, "grad_norm": 0.4518645942513458, "learning_rate": 9.865621992247654e-05, "loss": 3.3431, "step": 8050 }, { "epoch": 0.49978273015084734, "grad_norm": 0.4594537323673815, "learning_rate": 9.865538813355375e-05, "loss": 3.2684, "step": 8051 }, { "epoch": 0.49984480725060526, "grad_norm": 0.4588334342640273, "learning_rate": 9.8654556090784e-05, "loss": 3.2879, "step": 8052 }, { "epoch": 0.4999068843503632, "grad_norm": 0.5919126894591372, "learning_rate": 9.865372379417163e-05, "loss": 3.2036, "step": 8053 }, { "epoch": 0.49996896145012104, "grad_norm": 0.5439596670780416, "learning_rate": 9.865289124372097e-05, "loss": 3.3513, "step": 8054 }, { "epoch": 0.500031038549879, "grad_norm": 0.49096855720951826, "learning_rate": 9.865205843943634e-05, "loss": 3.3121, "step": 8055 }, { "epoch": 0.5000931156496369, "grad_norm": 0.5274917394886192, "learning_rate": 9.865122538132214e-05, "loss": 3.3007, "step": 8056 }, { "epoch": 0.5001551927493948, "grad_norm": 0.5282500497066043, "learning_rate": 9.865039206938267e-05, "loss": 3.3381, "step": 8057 }, { "epoch": 0.5002172698491526, "grad_norm": 0.5761250363724627, "learning_rate": 9.864955850362231e-05, "loss": 3.3643, "step": 8058 }, { "epoch": 0.5002793469489105, "grad_norm": 0.6334941865749122, "learning_rate": 9.864872468404538e-05, "loss": 3.319, "step": 8059 }, { "epoch": 0.5003414240486684, "grad_norm": 0.5846645260185903, "learning_rate": 9.864789061065625e-05, "loss": 3.3172, "step": 8060 }, { "epoch": 0.5004035011484264, "grad_norm": 0.4942606753748241, "learning_rate": 9.864705628345928e-05, "loss": 3.2849, "step": 8061 }, { "epoch": 0.5004655782481843, "grad_norm": 0.5648911040784181, "learning_rate": 9.864622170245878e-05, "loss": 3.3282, "step": 8062 }, { "epoch": 0.5005276553479422, "grad_norm": 0.5247801456597673, "learning_rate": 9.864538686765914e-05, "loss": 3.2336, "step": 8063 }, { "epoch": 0.5005897324477, "grad_norm": 0.5195503231362197, "learning_rate": 9.864455177906471e-05, "loss": 3.3104, "step": 8064 }, { "epoch": 0.5006518095474579, "grad_norm": 0.5501367572341956, "learning_rate": 9.864371643667985e-05, "loss": 3.2699, "step": 8065 }, { "epoch": 0.5007138866472158, "grad_norm": 0.48110247121837124, "learning_rate": 9.86428808405089e-05, "loss": 3.2528, "step": 8066 }, { "epoch": 0.5007759637469738, "grad_norm": 0.5334662322768804, "learning_rate": 9.864204499055624e-05, "loss": 3.1471, "step": 8067 }, { "epoch": 0.5008380408467317, "grad_norm": 0.5548391269243081, "learning_rate": 9.864120888682622e-05, "loss": 3.2532, "step": 8068 }, { "epoch": 0.5009001179464896, "grad_norm": 0.45935911747989994, "learning_rate": 9.864037252932322e-05, "loss": 3.2611, "step": 8069 }, { "epoch": 0.5009621950462474, "grad_norm": 0.6787459330147142, "learning_rate": 9.863953591805157e-05, "loss": 3.2905, "step": 8070 }, { "epoch": 0.5010242721460053, "grad_norm": 0.5906468664894955, "learning_rate": 9.863869905301566e-05, "loss": 3.2475, "step": 8071 }, { "epoch": 0.5010863492457632, "grad_norm": 0.42672923186595835, "learning_rate": 9.863786193421985e-05, "loss": 3.1676, "step": 8072 }, { "epoch": 0.5011484263455211, "grad_norm": 0.48884197845264793, "learning_rate": 9.86370245616685e-05, "loss": 3.2563, "step": 8073 }, { "epoch": 0.5012105034452791, "grad_norm": 0.5399851711762609, "learning_rate": 9.8636186935366e-05, "loss": 3.255, "step": 8074 }, { "epoch": 0.501272580545037, "grad_norm": 0.584013156755731, "learning_rate": 9.86353490553167e-05, "loss": 3.2756, "step": 8075 }, { "epoch": 0.5013346576447948, "grad_norm": 0.5721729971747248, "learning_rate": 9.863451092152497e-05, "loss": 3.2243, "step": 8076 }, { "epoch": 0.5013967347445527, "grad_norm": 0.5009168920246219, "learning_rate": 9.86336725339952e-05, "loss": 3.3476, "step": 8077 }, { "epoch": 0.5014588118443106, "grad_norm": 0.5353590866278252, "learning_rate": 9.863283389273175e-05, "loss": 3.1959, "step": 8078 }, { "epoch": 0.5015208889440685, "grad_norm": 0.5763872164455596, "learning_rate": 9.8631994997739e-05, "loss": 3.24, "step": 8079 }, { "epoch": 0.5015829660438265, "grad_norm": 0.557015784478271, "learning_rate": 9.863115584902132e-05, "loss": 3.2345, "step": 8080 }, { "epoch": 0.5016450431435844, "grad_norm": 0.49784957867388374, "learning_rate": 9.86303164465831e-05, "loss": 3.2702, "step": 8081 }, { "epoch": 0.5017071202433422, "grad_norm": 0.6111767371178226, "learning_rate": 9.86294767904287e-05, "loss": 3.2736, "step": 8082 }, { "epoch": 0.5017691973431001, "grad_norm": 0.5824900361465479, "learning_rate": 9.862863688056255e-05, "loss": 3.3033, "step": 8083 }, { "epoch": 0.501831274442858, "grad_norm": 0.5325978515159083, "learning_rate": 9.862779671698896e-05, "loss": 3.2087, "step": 8084 }, { "epoch": 0.5018933515426159, "grad_norm": 0.5973575818143999, "learning_rate": 9.862695629971237e-05, "loss": 3.3096, "step": 8085 }, { "epoch": 0.5019554286423739, "grad_norm": 0.585269394492518, "learning_rate": 9.862611562873714e-05, "loss": 3.234, "step": 8086 }, { "epoch": 0.5020175057421318, "grad_norm": 0.5906776192855545, "learning_rate": 9.862527470406766e-05, "loss": 3.2448, "step": 8087 }, { "epoch": 0.5020795828418896, "grad_norm": 0.5519409145940316, "learning_rate": 9.862443352570832e-05, "loss": 3.344, "step": 8088 }, { "epoch": 0.5021416599416475, "grad_norm": 0.5575929760060889, "learning_rate": 9.86235920936635e-05, "loss": 3.3057, "step": 8089 }, { "epoch": 0.5022037370414054, "grad_norm": 0.6274080853947195, "learning_rate": 9.86227504079376e-05, "loss": 3.3302, "step": 8090 }, { "epoch": 0.5022658141411633, "grad_norm": 0.5793346688513439, "learning_rate": 9.862190846853501e-05, "loss": 3.3454, "step": 8091 }, { "epoch": 0.5023278912409213, "grad_norm": 0.4359297673076243, "learning_rate": 9.86210662754601e-05, "loss": 3.3257, "step": 8092 }, { "epoch": 0.5023899683406792, "grad_norm": 0.55345303888978, "learning_rate": 9.862022382871732e-05, "loss": 3.3848, "step": 8093 }, { "epoch": 0.502452045440437, "grad_norm": 0.5974893523907532, "learning_rate": 9.8619381128311e-05, "loss": 3.2378, "step": 8094 }, { "epoch": 0.5025141225401949, "grad_norm": 0.6029713222412222, "learning_rate": 9.861853817424558e-05, "loss": 3.3301, "step": 8095 }, { "epoch": 0.5025761996399528, "grad_norm": 0.5835755361960069, "learning_rate": 9.861769496652544e-05, "loss": 3.2305, "step": 8096 }, { "epoch": 0.5026382767397107, "grad_norm": 0.5419653533616192, "learning_rate": 9.861685150515498e-05, "loss": 3.2049, "step": 8097 }, { "epoch": 0.5027003538394686, "grad_norm": 0.542636989533986, "learning_rate": 9.86160077901386e-05, "loss": 3.2282, "step": 8098 }, { "epoch": 0.5027624309392266, "grad_norm": 0.6132608641133716, "learning_rate": 9.861516382148073e-05, "loss": 3.2953, "step": 8099 }, { "epoch": 0.5028245080389844, "grad_norm": 0.515228812857937, "learning_rate": 9.861431959918573e-05, "loss": 3.3193, "step": 8100 }, { "epoch": 0.5028865851387423, "grad_norm": 0.5023147527177378, "learning_rate": 9.861347512325801e-05, "loss": 3.246, "step": 8101 }, { "epoch": 0.5029486622385002, "grad_norm": 0.5065520268651317, "learning_rate": 9.861263039370201e-05, "loss": 3.283, "step": 8102 }, { "epoch": 0.5030107393382581, "grad_norm": 0.6117214607188453, "learning_rate": 9.861178541052211e-05, "loss": 3.329, "step": 8103 }, { "epoch": 0.503072816438016, "grad_norm": 0.5006970672550484, "learning_rate": 9.861094017372272e-05, "loss": 3.2447, "step": 8104 }, { "epoch": 0.503134893537774, "grad_norm": 0.4988425985341057, "learning_rate": 9.861009468330826e-05, "loss": 3.249, "step": 8105 }, { "epoch": 0.5031969706375318, "grad_norm": 0.5553092963749361, "learning_rate": 9.860924893928313e-05, "loss": 3.3335, "step": 8106 }, { "epoch": 0.5032590477372897, "grad_norm": 0.532987306395387, "learning_rate": 9.860840294165174e-05, "loss": 3.2667, "step": 8107 }, { "epoch": 0.5033211248370476, "grad_norm": 0.5021224571222517, "learning_rate": 9.860755669041853e-05, "loss": 3.2941, "step": 8108 }, { "epoch": 0.5033832019368055, "grad_norm": 0.5255135317982962, "learning_rate": 9.86067101855879e-05, "loss": 3.2412, "step": 8109 }, { "epoch": 0.5034452790365634, "grad_norm": 0.5051091349110841, "learning_rate": 9.860586342716425e-05, "loss": 3.2923, "step": 8110 }, { "epoch": 0.5035073561363214, "grad_norm": 0.5746842139162057, "learning_rate": 9.8605016415152e-05, "loss": 3.2974, "step": 8111 }, { "epoch": 0.5035694332360792, "grad_norm": 0.5424340998269092, "learning_rate": 9.86041691495556e-05, "loss": 3.2464, "step": 8112 }, { "epoch": 0.5036315103358371, "grad_norm": 0.5149173215127364, "learning_rate": 9.860332163037944e-05, "loss": 3.1541, "step": 8113 }, { "epoch": 0.503693587435595, "grad_norm": 0.509605177844944, "learning_rate": 9.860247385762797e-05, "loss": 3.33, "step": 8114 }, { "epoch": 0.5037556645353529, "grad_norm": 0.4945778853376966, "learning_rate": 9.860162583130557e-05, "loss": 3.3572, "step": 8115 }, { "epoch": 0.5038177416351108, "grad_norm": 0.48904571199700836, "learning_rate": 9.860077755141671e-05, "loss": 3.1891, "step": 8116 }, { "epoch": 0.5038798187348688, "grad_norm": 0.5302775964169897, "learning_rate": 9.859992901796578e-05, "loss": 3.2344, "step": 8117 }, { "epoch": 0.5039418958346266, "grad_norm": 0.5957914693392639, "learning_rate": 9.859908023095724e-05, "loss": 3.2864, "step": 8118 }, { "epoch": 0.5040039729343845, "grad_norm": 0.6079648648198913, "learning_rate": 9.859823119039549e-05, "loss": 3.2799, "step": 8119 }, { "epoch": 0.5040660500341424, "grad_norm": 0.5354332311519524, "learning_rate": 9.859738189628497e-05, "loss": 3.2863, "step": 8120 }, { "epoch": 0.5041281271339003, "grad_norm": 0.5349198536244806, "learning_rate": 9.85965323486301e-05, "loss": 3.2125, "step": 8121 }, { "epoch": 0.5041902042336582, "grad_norm": 0.521283085868479, "learning_rate": 9.859568254743535e-05, "loss": 3.3194, "step": 8122 }, { "epoch": 0.5042522813334162, "grad_norm": 0.5587350093187536, "learning_rate": 9.85948324927051e-05, "loss": 3.2789, "step": 8123 }, { "epoch": 0.504314358433174, "grad_norm": 0.5786036237079718, "learning_rate": 9.859398218444383e-05, "loss": 3.261, "step": 8124 }, { "epoch": 0.5043764355329319, "grad_norm": 0.5423474221639631, "learning_rate": 9.859313162265595e-05, "loss": 3.3984, "step": 8125 }, { "epoch": 0.5044385126326898, "grad_norm": 0.45413463841234153, "learning_rate": 9.85922808073459e-05, "loss": 3.1864, "step": 8126 }, { "epoch": 0.5045005897324477, "grad_norm": 0.5113799451385156, "learning_rate": 9.859142973851813e-05, "loss": 3.3338, "step": 8127 }, { "epoch": 0.5045626668322056, "grad_norm": 0.526797598134404, "learning_rate": 9.859057841617709e-05, "loss": 3.3087, "step": 8128 }, { "epoch": 0.5046247439319635, "grad_norm": 0.48142460590890584, "learning_rate": 9.858972684032718e-05, "loss": 3.2773, "step": 8129 }, { "epoch": 0.5046868210317214, "grad_norm": 0.5121100000163789, "learning_rate": 9.85888750109729e-05, "loss": 3.3151, "step": 8130 }, { "epoch": 0.5047488981314793, "grad_norm": 0.5002170330129433, "learning_rate": 9.858802292811864e-05, "loss": 3.2705, "step": 8131 }, { "epoch": 0.5048109752312372, "grad_norm": 0.49440407044195694, "learning_rate": 9.858717059176888e-05, "loss": 3.2392, "step": 8132 }, { "epoch": 0.5048730523309951, "grad_norm": 0.5359892939683334, "learning_rate": 9.858631800192806e-05, "loss": 3.2816, "step": 8133 }, { "epoch": 0.504935129430753, "grad_norm": 0.5648205890893071, "learning_rate": 9.858546515860061e-05, "loss": 3.3304, "step": 8134 }, { "epoch": 0.504997206530511, "grad_norm": 0.5406491246656555, "learning_rate": 9.858461206179101e-05, "loss": 3.2917, "step": 8135 }, { "epoch": 0.5050592836302688, "grad_norm": 0.5925565319615739, "learning_rate": 9.858375871150369e-05, "loss": 3.3277, "step": 8136 }, { "epoch": 0.5051213607300267, "grad_norm": 0.5462969586071422, "learning_rate": 9.85829051077431e-05, "loss": 3.3182, "step": 8137 }, { "epoch": 0.5051834378297846, "grad_norm": 0.5268990513661812, "learning_rate": 9.858205125051369e-05, "loss": 3.2025, "step": 8138 }, { "epoch": 0.5052455149295425, "grad_norm": 0.4684135342066634, "learning_rate": 9.858119713981994e-05, "loss": 3.3511, "step": 8139 }, { "epoch": 0.5053075920293004, "grad_norm": 0.5256252403122651, "learning_rate": 9.858034277566628e-05, "loss": 3.2499, "step": 8140 }, { "epoch": 0.5053696691290583, "grad_norm": 0.47202170156843043, "learning_rate": 9.85794881580572e-05, "loss": 3.3363, "step": 8141 }, { "epoch": 0.5054317462288161, "grad_norm": 0.5036018990174833, "learning_rate": 9.857863328699712e-05, "loss": 3.2942, "step": 8142 }, { "epoch": 0.5054938233285741, "grad_norm": 0.5259867409713898, "learning_rate": 9.857777816249051e-05, "loss": 3.3557, "step": 8143 }, { "epoch": 0.505555900428332, "grad_norm": 0.5498029529279921, "learning_rate": 9.857692278454187e-05, "loss": 3.3172, "step": 8144 }, { "epoch": 0.5056179775280899, "grad_norm": 0.5426002184857928, "learning_rate": 9.857606715315561e-05, "loss": 3.3246, "step": 8145 }, { "epoch": 0.5056800546278478, "grad_norm": 0.5003722414545337, "learning_rate": 9.857521126833622e-05, "loss": 3.3356, "step": 8146 }, { "epoch": 0.5057421317276057, "grad_norm": 0.541577869864147, "learning_rate": 9.857435513008815e-05, "loss": 3.295, "step": 8147 }, { "epoch": 0.5058042088273635, "grad_norm": 0.5106877722846241, "learning_rate": 9.857349873841587e-05, "loss": 3.3082, "step": 8148 }, { "epoch": 0.5058662859271215, "grad_norm": 0.48853254336830904, "learning_rate": 9.857264209332388e-05, "loss": 3.2947, "step": 8149 }, { "epoch": 0.5059283630268794, "grad_norm": 0.5216014316352697, "learning_rate": 9.857178519481663e-05, "loss": 3.2988, "step": 8150 }, { "epoch": 0.5059904401266373, "grad_norm": 0.4999863801531686, "learning_rate": 9.857092804289856e-05, "loss": 3.2355, "step": 8151 }, { "epoch": 0.5060525172263952, "grad_norm": 0.5243559508563539, "learning_rate": 9.857007063757417e-05, "loss": 3.3246, "step": 8152 }, { "epoch": 0.5061145943261531, "grad_norm": 0.537660960559558, "learning_rate": 9.856921297884796e-05, "loss": 3.2425, "step": 8153 }, { "epoch": 0.5061766714259109, "grad_norm": 0.5421069324447727, "learning_rate": 9.856835506672435e-05, "loss": 3.2937, "step": 8154 }, { "epoch": 0.5062387485256689, "grad_norm": 0.6382671555418551, "learning_rate": 9.856749690120784e-05, "loss": 3.2376, "step": 8155 }, { "epoch": 0.5063008256254268, "grad_norm": 0.5228760113306226, "learning_rate": 9.856663848230292e-05, "loss": 3.2109, "step": 8156 }, { "epoch": 0.5063629027251847, "grad_norm": 0.5053870746831635, "learning_rate": 9.856577981001405e-05, "loss": 3.332, "step": 8157 }, { "epoch": 0.5064249798249426, "grad_norm": 0.5691275632428971, "learning_rate": 9.856492088434571e-05, "loss": 3.3146, "step": 8158 }, { "epoch": 0.5064870569247005, "grad_norm": 0.5425642688758858, "learning_rate": 9.856406170530241e-05, "loss": 3.2672, "step": 8159 }, { "epoch": 0.5065491340244583, "grad_norm": 0.5227685610952707, "learning_rate": 9.85632022728886e-05, "loss": 3.3031, "step": 8160 }, { "epoch": 0.5066112111242163, "grad_norm": 0.4973623770468339, "learning_rate": 9.856234258710876e-05, "loss": 3.195, "step": 8161 }, { "epoch": 0.5066732882239742, "grad_norm": 0.4995000395989617, "learning_rate": 9.85614826479674e-05, "loss": 3.2779, "step": 8162 }, { "epoch": 0.5067353653237321, "grad_norm": 0.5290861097228822, "learning_rate": 9.8560622455469e-05, "loss": 3.268, "step": 8163 }, { "epoch": 0.50679744242349, "grad_norm": 0.5016292907831716, "learning_rate": 9.855976200961805e-05, "loss": 3.2664, "step": 8164 }, { "epoch": 0.5068595195232479, "grad_norm": 0.5093264679518871, "learning_rate": 9.855890131041901e-05, "loss": 3.2087, "step": 8165 }, { "epoch": 0.5069215966230057, "grad_norm": 0.48460519612041997, "learning_rate": 9.855804035787641e-05, "loss": 3.2931, "step": 8166 }, { "epoch": 0.5069836737227637, "grad_norm": 0.5455023022628439, "learning_rate": 9.855717915199472e-05, "loss": 3.3435, "step": 8167 }, { "epoch": 0.5070457508225216, "grad_norm": 0.46958124843474186, "learning_rate": 9.855631769277844e-05, "loss": 3.2019, "step": 8168 }, { "epoch": 0.5071078279222795, "grad_norm": 0.5637484841421517, "learning_rate": 9.855545598023208e-05, "loss": 3.3155, "step": 8169 }, { "epoch": 0.5071699050220374, "grad_norm": 0.5287153702526474, "learning_rate": 9.85545940143601e-05, "loss": 3.2578, "step": 8170 }, { "epoch": 0.5072319821217953, "grad_norm": 0.5103718160529491, "learning_rate": 9.855373179516703e-05, "loss": 3.3519, "step": 8171 }, { "epoch": 0.5072940592215531, "grad_norm": 0.5239977898223508, "learning_rate": 9.855286932265733e-05, "loss": 3.2326, "step": 8172 }, { "epoch": 0.507356136321311, "grad_norm": 0.460034520445132, "learning_rate": 9.855200659683555e-05, "loss": 3.3118, "step": 8173 }, { "epoch": 0.507418213421069, "grad_norm": 0.47251831753359314, "learning_rate": 9.855114361770616e-05, "loss": 3.2219, "step": 8174 }, { "epoch": 0.5074802905208269, "grad_norm": 0.4801205404492128, "learning_rate": 9.855028038527366e-05, "loss": 3.2196, "step": 8175 }, { "epoch": 0.5075423676205848, "grad_norm": 0.5342277770215567, "learning_rate": 9.854941689954255e-05, "loss": 3.2111, "step": 8176 }, { "epoch": 0.5076044447203427, "grad_norm": 0.5140637255714904, "learning_rate": 9.854855316051737e-05, "loss": 3.2381, "step": 8177 }, { "epoch": 0.5076665218201005, "grad_norm": 0.5068269997104515, "learning_rate": 9.85476891682026e-05, "loss": 3.3502, "step": 8178 }, { "epoch": 0.5077285989198584, "grad_norm": 0.47901558740276434, "learning_rate": 9.854682492260275e-05, "loss": 3.3074, "step": 8179 }, { "epoch": 0.5077906760196164, "grad_norm": 0.5816363076382198, "learning_rate": 9.854596042372231e-05, "loss": 3.2997, "step": 8180 }, { "epoch": 0.5078527531193743, "grad_norm": 0.5384829626301451, "learning_rate": 9.854509567156583e-05, "loss": 3.2301, "step": 8181 }, { "epoch": 0.5079148302191322, "grad_norm": 0.4798051081072243, "learning_rate": 9.854423066613779e-05, "loss": 3.1601, "step": 8182 }, { "epoch": 0.5079769073188901, "grad_norm": 0.688493725718303, "learning_rate": 9.854336540744272e-05, "loss": 3.359, "step": 8183 }, { "epoch": 0.5080389844186479, "grad_norm": 0.7509928330741075, "learning_rate": 9.854249989548512e-05, "loss": 3.2918, "step": 8184 }, { "epoch": 0.5081010615184058, "grad_norm": 0.5674515385458909, "learning_rate": 9.854163413026951e-05, "loss": 3.2134, "step": 8185 }, { "epoch": 0.5081631386181638, "grad_norm": 0.5703337155929227, "learning_rate": 9.854076811180042e-05, "loss": 3.3344, "step": 8186 }, { "epoch": 0.5082252157179217, "grad_norm": 0.6325977337758938, "learning_rate": 9.853990184008236e-05, "loss": 3.3027, "step": 8187 }, { "epoch": 0.5082872928176796, "grad_norm": 0.5913553132832319, "learning_rate": 9.853903531511986e-05, "loss": 3.3559, "step": 8188 }, { "epoch": 0.5083493699174375, "grad_norm": 0.631729777407585, "learning_rate": 9.853816853691741e-05, "loss": 3.2541, "step": 8189 }, { "epoch": 0.5084114470171953, "grad_norm": 0.5680933588699043, "learning_rate": 9.853730150547955e-05, "loss": 3.3038, "step": 8190 }, { "epoch": 0.5084735241169532, "grad_norm": 0.5119515682891272, "learning_rate": 9.853643422081083e-05, "loss": 3.2705, "step": 8191 }, { "epoch": 0.5085356012167112, "grad_norm": 0.5779330734045454, "learning_rate": 9.853556668291573e-05, "loss": 3.2373, "step": 8192 }, { "epoch": 0.5085976783164691, "grad_norm": 0.5804098959411721, "learning_rate": 9.853469889179879e-05, "loss": 3.2475, "step": 8193 }, { "epoch": 0.508659755416227, "grad_norm": 0.5728925584286815, "learning_rate": 9.853383084746456e-05, "loss": 3.3405, "step": 8194 }, { "epoch": 0.5087218325159849, "grad_norm": 0.5128709183543765, "learning_rate": 9.853296254991755e-05, "loss": 3.2588, "step": 8195 }, { "epoch": 0.5087839096157427, "grad_norm": 0.5392103130749443, "learning_rate": 9.853209399916228e-05, "loss": 3.2533, "step": 8196 }, { "epoch": 0.5088459867155006, "grad_norm": 0.5455117809726309, "learning_rate": 9.85312251952033e-05, "loss": 3.2156, "step": 8197 }, { "epoch": 0.5089080638152585, "grad_norm": 0.4858181742527316, "learning_rate": 9.853035613804514e-05, "loss": 3.2453, "step": 8198 }, { "epoch": 0.5089701409150165, "grad_norm": 0.5104434012852747, "learning_rate": 9.852948682769234e-05, "loss": 3.3133, "step": 8199 }, { "epoch": 0.5090322180147744, "grad_norm": 0.5866017962844768, "learning_rate": 9.852861726414941e-05, "loss": 3.2328, "step": 8200 }, { "epoch": 0.5090942951145323, "grad_norm": 0.512617355959422, "learning_rate": 9.852774744742092e-05, "loss": 3.3332, "step": 8201 }, { "epoch": 0.5091563722142901, "grad_norm": 0.5161380815998492, "learning_rate": 9.852687737751137e-05, "loss": 3.3509, "step": 8202 }, { "epoch": 0.509218449314048, "grad_norm": 0.5553386251457721, "learning_rate": 9.852600705442533e-05, "loss": 3.246, "step": 8203 }, { "epoch": 0.509280526413806, "grad_norm": 0.5467138325398391, "learning_rate": 9.852513647816735e-05, "loss": 3.3197, "step": 8204 }, { "epoch": 0.5093426035135639, "grad_norm": 0.4952641707226688, "learning_rate": 9.852426564874193e-05, "loss": 3.2914, "step": 8205 }, { "epoch": 0.5094046806133218, "grad_norm": 0.5514426352391406, "learning_rate": 9.852339456615364e-05, "loss": 3.3092, "step": 8206 }, { "epoch": 0.5094667577130797, "grad_norm": 0.5306791716872203, "learning_rate": 9.852252323040703e-05, "loss": 3.253, "step": 8207 }, { "epoch": 0.5095288348128375, "grad_norm": 0.4574616190960881, "learning_rate": 9.852165164150662e-05, "loss": 3.2115, "step": 8208 }, { "epoch": 0.5095909119125954, "grad_norm": 0.5183987715352626, "learning_rate": 9.852077979945698e-05, "loss": 3.3383, "step": 8209 }, { "epoch": 0.5096529890123533, "grad_norm": 0.4984629519909022, "learning_rate": 9.851990770426266e-05, "loss": 3.3374, "step": 8210 }, { "epoch": 0.5097150661121113, "grad_norm": 0.4866830381025037, "learning_rate": 9.85190353559282e-05, "loss": 3.2798, "step": 8211 }, { "epoch": 0.5097771432118692, "grad_norm": 0.5352972638422097, "learning_rate": 9.851816275445815e-05, "loss": 3.2222, "step": 8212 }, { "epoch": 0.5098392203116271, "grad_norm": 0.5646755215968763, "learning_rate": 9.851728989985708e-05, "loss": 3.2381, "step": 8213 }, { "epoch": 0.5099012974113849, "grad_norm": 0.5093733056163106, "learning_rate": 9.851641679212951e-05, "loss": 3.3548, "step": 8214 }, { "epoch": 0.5099633745111428, "grad_norm": 0.5530133840586483, "learning_rate": 9.851554343128004e-05, "loss": 3.3173, "step": 8215 }, { "epoch": 0.5100254516109007, "grad_norm": 0.570694860546057, "learning_rate": 9.851466981731319e-05, "loss": 3.2655, "step": 8216 }, { "epoch": 0.5100875287106587, "grad_norm": 0.5109754799075411, "learning_rate": 9.851379595023354e-05, "loss": 3.2867, "step": 8217 }, { "epoch": 0.5101496058104166, "grad_norm": 0.4900772025149673, "learning_rate": 9.851292183004561e-05, "loss": 3.2436, "step": 8218 }, { "epoch": 0.5102116829101745, "grad_norm": 0.4728889523030231, "learning_rate": 9.851204745675402e-05, "loss": 3.3134, "step": 8219 }, { "epoch": 0.5102737600099323, "grad_norm": 0.5439956066501148, "learning_rate": 9.851117283036328e-05, "loss": 3.1685, "step": 8220 }, { "epoch": 0.5103358371096902, "grad_norm": 0.5284427352907622, "learning_rate": 9.851029795087799e-05, "loss": 3.2415, "step": 8221 }, { "epoch": 0.5103979142094481, "grad_norm": 0.5788825716229022, "learning_rate": 9.85094228183027e-05, "loss": 3.2973, "step": 8222 }, { "epoch": 0.510459991309206, "grad_norm": 0.4615738081273235, "learning_rate": 9.850854743264197e-05, "loss": 3.3597, "step": 8223 }, { "epoch": 0.510522068408964, "grad_norm": 0.4417680253678179, "learning_rate": 9.850767179390037e-05, "loss": 3.3373, "step": 8224 }, { "epoch": 0.5105841455087219, "grad_norm": 0.506935830093886, "learning_rate": 9.850679590208248e-05, "loss": 3.34, "step": 8225 }, { "epoch": 0.5106462226084797, "grad_norm": 0.4984082575145361, "learning_rate": 9.850591975719286e-05, "loss": 3.2871, "step": 8226 }, { "epoch": 0.5107082997082376, "grad_norm": 0.5230595946812995, "learning_rate": 9.850504335923608e-05, "loss": 3.2962, "step": 8227 }, { "epoch": 0.5107703768079955, "grad_norm": 0.5010516610473867, "learning_rate": 9.85041667082167e-05, "loss": 3.3133, "step": 8228 }, { "epoch": 0.5108324539077534, "grad_norm": 0.4873382465067698, "learning_rate": 9.85032898041393e-05, "loss": 3.2935, "step": 8229 }, { "epoch": 0.5108945310075114, "grad_norm": 0.5097474427470092, "learning_rate": 9.850241264700848e-05, "loss": 3.1886, "step": 8230 }, { "epoch": 0.5109566081072693, "grad_norm": 0.4959969463667777, "learning_rate": 9.850153523682881e-05, "loss": 3.2302, "step": 8231 }, { "epoch": 0.5110186852070271, "grad_norm": 0.5821040831189518, "learning_rate": 9.850065757360484e-05, "loss": 3.3643, "step": 8232 }, { "epoch": 0.511080762306785, "grad_norm": 0.5436171358390091, "learning_rate": 9.849977965734116e-05, "loss": 3.2356, "step": 8233 }, { "epoch": 0.5111428394065429, "grad_norm": 0.5672888095217326, "learning_rate": 9.849890148804237e-05, "loss": 3.2999, "step": 8234 }, { "epoch": 0.5112049165063008, "grad_norm": 0.49726396247244237, "learning_rate": 9.849802306571303e-05, "loss": 3.3179, "step": 8235 }, { "epoch": 0.5112669936060588, "grad_norm": 0.4078353429659577, "learning_rate": 9.849714439035775e-05, "loss": 3.1864, "step": 8236 }, { "epoch": 0.5113290707058167, "grad_norm": 0.5184380351610508, "learning_rate": 9.849626546198108e-05, "loss": 3.2168, "step": 8237 }, { "epoch": 0.5113911478055745, "grad_norm": 0.5531719349490262, "learning_rate": 9.84953862805876e-05, "loss": 3.3647, "step": 8238 }, { "epoch": 0.5114532249053324, "grad_norm": 0.5384356959529151, "learning_rate": 9.849450684618195e-05, "loss": 3.2349, "step": 8239 }, { "epoch": 0.5115153020050903, "grad_norm": 0.547927687960788, "learning_rate": 9.849362715876866e-05, "loss": 3.1753, "step": 8240 }, { "epoch": 0.5115773791048482, "grad_norm": 0.47398541097180386, "learning_rate": 9.849274721835236e-05, "loss": 3.2178, "step": 8241 }, { "epoch": 0.5116394562046062, "grad_norm": 0.5213108702779224, "learning_rate": 9.849186702493762e-05, "loss": 3.3679, "step": 8242 }, { "epoch": 0.511701533304364, "grad_norm": 0.5431161035614618, "learning_rate": 9.849098657852904e-05, "loss": 3.2682, "step": 8243 }, { "epoch": 0.5117636104041219, "grad_norm": 0.4221911482014508, "learning_rate": 9.849010587913121e-05, "loss": 3.2678, "step": 8244 }, { "epoch": 0.5118256875038798, "grad_norm": 0.4944738387881638, "learning_rate": 9.848922492674872e-05, "loss": 3.3098, "step": 8245 }, { "epoch": 0.5118877646036377, "grad_norm": 0.5559851015671162, "learning_rate": 9.848834372138617e-05, "loss": 3.323, "step": 8246 }, { "epoch": 0.5119498417033956, "grad_norm": 0.49598949568435013, "learning_rate": 9.848746226304819e-05, "loss": 3.2956, "step": 8247 }, { "epoch": 0.5120119188031536, "grad_norm": 0.4884942162521769, "learning_rate": 9.848658055173931e-05, "loss": 3.2854, "step": 8248 }, { "epoch": 0.5120739959029114, "grad_norm": 0.5166074756074779, "learning_rate": 9.848569858746419e-05, "loss": 3.3177, "step": 8249 }, { "epoch": 0.5121360730026693, "grad_norm": 0.5270038781068626, "learning_rate": 9.848481637022741e-05, "loss": 3.2923, "step": 8250 }, { "epoch": 0.5121981501024272, "grad_norm": 0.5541264288737786, "learning_rate": 9.848393390003356e-05, "loss": 3.2783, "step": 8251 }, { "epoch": 0.5122602272021851, "grad_norm": 0.5177462610619049, "learning_rate": 9.848305117688727e-05, "loss": 3.3226, "step": 8252 }, { "epoch": 0.512322304301943, "grad_norm": 0.5182669033195898, "learning_rate": 9.848216820079315e-05, "loss": 3.2387, "step": 8253 }, { "epoch": 0.512384381401701, "grad_norm": 0.5841308716197858, "learning_rate": 9.848128497175575e-05, "loss": 3.2682, "step": 8254 }, { "epoch": 0.5124464585014588, "grad_norm": 0.6070565600132577, "learning_rate": 9.848040148977974e-05, "loss": 3.3057, "step": 8255 }, { "epoch": 0.5125085356012167, "grad_norm": 0.5792470559468074, "learning_rate": 9.84795177548697e-05, "loss": 3.1449, "step": 8256 }, { "epoch": 0.5125706127009746, "grad_norm": 0.5993433557934029, "learning_rate": 9.847863376703024e-05, "loss": 3.2539, "step": 8257 }, { "epoch": 0.5126326898007325, "grad_norm": 0.5777406703768474, "learning_rate": 9.847774952626599e-05, "loss": 3.24, "step": 8258 }, { "epoch": 0.5126947669004904, "grad_norm": 0.4897666919214089, "learning_rate": 9.847686503258156e-05, "loss": 3.178, "step": 8259 }, { "epoch": 0.5127568440002483, "grad_norm": 0.5567806937686017, "learning_rate": 9.847598028598155e-05, "loss": 3.2957, "step": 8260 }, { "epoch": 0.5128189211000062, "grad_norm": 0.7522059820525077, "learning_rate": 9.847509528647056e-05, "loss": 3.3151, "step": 8261 }, { "epoch": 0.5128809981997641, "grad_norm": 0.6352970656937408, "learning_rate": 9.847421003405327e-05, "loss": 3.3622, "step": 8262 }, { "epoch": 0.512943075299522, "grad_norm": 0.5324330340620292, "learning_rate": 9.847332452873424e-05, "loss": 3.3643, "step": 8263 }, { "epoch": 0.5130051523992799, "grad_norm": 0.5692001102232274, "learning_rate": 9.84724387705181e-05, "loss": 3.3188, "step": 8264 }, { "epoch": 0.5130672294990378, "grad_norm": 0.4927127397056581, "learning_rate": 9.847155275940949e-05, "loss": 3.2737, "step": 8265 }, { "epoch": 0.5131293065987957, "grad_norm": 0.46891835686149846, "learning_rate": 9.847066649541302e-05, "loss": 3.2578, "step": 8266 }, { "epoch": 0.5131913836985535, "grad_norm": 0.5008191015545617, "learning_rate": 9.846977997853331e-05, "loss": 3.2637, "step": 8267 }, { "epoch": 0.5132534607983115, "grad_norm": 0.5284451321367106, "learning_rate": 9.846889320877499e-05, "loss": 3.301, "step": 8268 }, { "epoch": 0.5133155378980694, "grad_norm": 0.5152823436696998, "learning_rate": 9.84680061861427e-05, "loss": 3.2243, "step": 8269 }, { "epoch": 0.5133776149978273, "grad_norm": 0.5250251945626392, "learning_rate": 9.846711891064105e-05, "loss": 3.3303, "step": 8270 }, { "epoch": 0.5134396920975852, "grad_norm": 0.49976885936621324, "learning_rate": 9.846623138227467e-05, "loss": 3.1544, "step": 8271 }, { "epoch": 0.5135017691973431, "grad_norm": 0.4691681268589115, "learning_rate": 9.84653436010482e-05, "loss": 3.2306, "step": 8272 }, { "epoch": 0.513563846297101, "grad_norm": 0.48696438024039246, "learning_rate": 9.846445556696627e-05, "loss": 3.306, "step": 8273 }, { "epoch": 0.5136259233968589, "grad_norm": 0.4690810147417516, "learning_rate": 9.84635672800335e-05, "loss": 3.2198, "step": 8274 }, { "epoch": 0.5136880004966168, "grad_norm": 0.4745332947336206, "learning_rate": 9.846267874025456e-05, "loss": 3.2863, "step": 8275 }, { "epoch": 0.5137500775963747, "grad_norm": 0.4761574227421962, "learning_rate": 9.846178994763404e-05, "loss": 3.3011, "step": 8276 }, { "epoch": 0.5138121546961326, "grad_norm": 0.5119204524055123, "learning_rate": 9.846090090217659e-05, "loss": 3.2744, "step": 8277 }, { "epoch": 0.5138742317958905, "grad_norm": 0.547587040161759, "learning_rate": 9.846001160388687e-05, "loss": 3.2851, "step": 8278 }, { "epoch": 0.5139363088956483, "grad_norm": 0.4852798838810211, "learning_rate": 9.84591220527695e-05, "loss": 3.3059, "step": 8279 }, { "epoch": 0.5139983859954063, "grad_norm": 0.5133641993183191, "learning_rate": 9.845823224882913e-05, "loss": 3.3225, "step": 8280 }, { "epoch": 0.5140604630951642, "grad_norm": 0.5695437122546102, "learning_rate": 9.845734219207038e-05, "loss": 3.2336, "step": 8281 }, { "epoch": 0.5141225401949221, "grad_norm": 0.5432701097848739, "learning_rate": 9.845645188249794e-05, "loss": 3.3008, "step": 8282 }, { "epoch": 0.51418461729468, "grad_norm": 0.41507155347318103, "learning_rate": 9.845556132011642e-05, "loss": 3.2278, "step": 8283 }, { "epoch": 0.5142466943944379, "grad_norm": 0.4924257700791927, "learning_rate": 9.845467050493047e-05, "loss": 3.2812, "step": 8284 }, { "epoch": 0.5143087714941957, "grad_norm": 0.45198443344727196, "learning_rate": 9.845377943694474e-05, "loss": 3.2712, "step": 8285 }, { "epoch": 0.5143708485939537, "grad_norm": 0.49724763057061744, "learning_rate": 9.845288811616389e-05, "loss": 3.3229, "step": 8286 }, { "epoch": 0.5144329256937116, "grad_norm": 0.546543075102926, "learning_rate": 9.845199654259254e-05, "loss": 3.1966, "step": 8287 }, { "epoch": 0.5144950027934695, "grad_norm": 0.5894380146615317, "learning_rate": 9.845110471623538e-05, "loss": 3.3215, "step": 8288 }, { "epoch": 0.5145570798932274, "grad_norm": 0.7009843665895206, "learning_rate": 9.845021263709704e-05, "loss": 3.255, "step": 8289 }, { "epoch": 0.5146191569929853, "grad_norm": 0.7192082809955936, "learning_rate": 9.844932030518216e-05, "loss": 3.2751, "step": 8290 }, { "epoch": 0.5146812340927431, "grad_norm": 0.5956601987287286, "learning_rate": 9.844842772049545e-05, "loss": 3.3672, "step": 8291 }, { "epoch": 0.514743311192501, "grad_norm": 0.5396798234138488, "learning_rate": 9.84475348830415e-05, "loss": 3.2162, "step": 8292 }, { "epoch": 0.514805388292259, "grad_norm": 0.5724124915279449, "learning_rate": 9.8446641792825e-05, "loss": 3.1429, "step": 8293 }, { "epoch": 0.5148674653920169, "grad_norm": 0.6285330097266977, "learning_rate": 9.844574844985064e-05, "loss": 3.3542, "step": 8294 }, { "epoch": 0.5149295424917748, "grad_norm": 0.5303048533763347, "learning_rate": 9.844485485412303e-05, "loss": 3.2594, "step": 8295 }, { "epoch": 0.5149916195915327, "grad_norm": 0.49029376517511025, "learning_rate": 9.844396100564685e-05, "loss": 3.2188, "step": 8296 }, { "epoch": 0.5150536966912905, "grad_norm": 0.5104612084501, "learning_rate": 9.844306690442677e-05, "loss": 3.2491, "step": 8297 }, { "epoch": 0.5151157737910484, "grad_norm": 0.5038739571093303, "learning_rate": 9.844217255046743e-05, "loss": 3.2016, "step": 8298 }, { "epoch": 0.5151778508908064, "grad_norm": 0.5048573116746624, "learning_rate": 9.844127794377355e-05, "loss": 3.2731, "step": 8299 }, { "epoch": 0.5152399279905643, "grad_norm": 0.5589255567910724, "learning_rate": 9.844038308434975e-05, "loss": 3.2888, "step": 8300 }, { "epoch": 0.5153020050903222, "grad_norm": 0.6077449221663693, "learning_rate": 9.84394879722007e-05, "loss": 3.2872, "step": 8301 }, { "epoch": 0.5153640821900801, "grad_norm": 0.4662748314583363, "learning_rate": 9.843859260733111e-05, "loss": 3.3344, "step": 8302 }, { "epoch": 0.5154261592898379, "grad_norm": 0.5025503561305373, "learning_rate": 9.843769698974559e-05, "loss": 3.2065, "step": 8303 }, { "epoch": 0.5154882363895958, "grad_norm": 0.5271528392945766, "learning_rate": 9.843680111944886e-05, "loss": 3.2811, "step": 8304 }, { "epoch": 0.5155503134893538, "grad_norm": 0.5099275120008196, "learning_rate": 9.843590499644558e-05, "loss": 3.2099, "step": 8305 }, { "epoch": 0.5156123905891117, "grad_norm": 0.5047733918787904, "learning_rate": 9.843500862074044e-05, "loss": 3.2993, "step": 8306 }, { "epoch": 0.5156744676888696, "grad_norm": 0.5271798903589349, "learning_rate": 9.843411199233808e-05, "loss": 3.1629, "step": 8307 }, { "epoch": 0.5157365447886275, "grad_norm": 0.5146359871040547, "learning_rate": 9.843321511124321e-05, "loss": 3.323, "step": 8308 }, { "epoch": 0.5157986218883853, "grad_norm": 0.4998966795743327, "learning_rate": 9.843231797746049e-05, "loss": 3.2224, "step": 8309 }, { "epoch": 0.5158606989881432, "grad_norm": 0.5397955291896459, "learning_rate": 9.843142059099462e-05, "loss": 3.3028, "step": 8310 }, { "epoch": 0.5159227760879012, "grad_norm": 0.5287002917522522, "learning_rate": 9.843052295185027e-05, "loss": 3.2194, "step": 8311 }, { "epoch": 0.5159848531876591, "grad_norm": 0.5538965953421306, "learning_rate": 9.842962506003212e-05, "loss": 3.2936, "step": 8312 }, { "epoch": 0.516046930287417, "grad_norm": 0.4995234871202365, "learning_rate": 9.842872691554486e-05, "loss": 3.2333, "step": 8313 }, { "epoch": 0.5161090073871749, "grad_norm": 0.49061068495744387, "learning_rate": 9.842782851839319e-05, "loss": 3.2419, "step": 8314 }, { "epoch": 0.5161710844869327, "grad_norm": 0.5469057619434775, "learning_rate": 9.842692986858177e-05, "loss": 3.3152, "step": 8315 }, { "epoch": 0.5162331615866906, "grad_norm": 0.5250936186289201, "learning_rate": 9.84260309661153e-05, "loss": 3.2053, "step": 8316 }, { "epoch": 0.5162952386864486, "grad_norm": 0.5392330839627951, "learning_rate": 9.842513181099847e-05, "loss": 3.2051, "step": 8317 }, { "epoch": 0.5163573157862065, "grad_norm": 0.545627642840335, "learning_rate": 9.842423240323597e-05, "loss": 3.1704, "step": 8318 }, { "epoch": 0.5164193928859644, "grad_norm": 0.4640136005827464, "learning_rate": 9.84233327428325e-05, "loss": 3.1817, "step": 8319 }, { "epoch": 0.5164814699857223, "grad_norm": 0.4897611088984936, "learning_rate": 9.842243282979276e-05, "loss": 3.2036, "step": 8320 }, { "epoch": 0.5165435470854801, "grad_norm": 0.5136155362291319, "learning_rate": 9.842153266412141e-05, "loss": 3.1713, "step": 8321 }, { "epoch": 0.516605624185238, "grad_norm": 0.49553537859715074, "learning_rate": 9.842063224582318e-05, "loss": 3.1905, "step": 8322 }, { "epoch": 0.516667701284996, "grad_norm": 0.4662430003103551, "learning_rate": 9.841973157490276e-05, "loss": 3.2297, "step": 8323 }, { "epoch": 0.5167297783847539, "grad_norm": 0.5126053794821169, "learning_rate": 9.841883065136484e-05, "loss": 3.2184, "step": 8324 }, { "epoch": 0.5167918554845118, "grad_norm": 0.570498240285127, "learning_rate": 9.841792947521412e-05, "loss": 3.1984, "step": 8325 }, { "epoch": 0.5168539325842697, "grad_norm": 0.4730748458896324, "learning_rate": 9.841702804645531e-05, "loss": 3.2726, "step": 8326 }, { "epoch": 0.5169160096840275, "grad_norm": 0.47038584419257695, "learning_rate": 9.841612636509311e-05, "loss": 3.1998, "step": 8327 }, { "epoch": 0.5169780867837854, "grad_norm": 0.47613075898729357, "learning_rate": 9.841522443113225e-05, "loss": 3.2206, "step": 8328 }, { "epoch": 0.5170401638835433, "grad_norm": 0.5243460612899463, "learning_rate": 9.84143222445774e-05, "loss": 3.1985, "step": 8329 }, { "epoch": 0.5171022409833013, "grad_norm": 0.5450249304450163, "learning_rate": 9.841341980543326e-05, "loss": 3.2698, "step": 8330 }, { "epoch": 0.5171643180830592, "grad_norm": 0.49508997975971164, "learning_rate": 9.841251711370458e-05, "loss": 3.1741, "step": 8331 }, { "epoch": 0.5172263951828171, "grad_norm": 0.5283879714485424, "learning_rate": 9.841161416939602e-05, "loss": 3.248, "step": 8332 }, { "epoch": 0.5172884722825749, "grad_norm": 0.5282085533184763, "learning_rate": 9.841071097251234e-05, "loss": 3.2545, "step": 8333 }, { "epoch": 0.5173505493823328, "grad_norm": 0.5039497474999812, "learning_rate": 9.840980752305821e-05, "loss": 3.347, "step": 8334 }, { "epoch": 0.5174126264820907, "grad_norm": 0.5281075773807569, "learning_rate": 9.840890382103838e-05, "loss": 3.3133, "step": 8335 }, { "epoch": 0.5174747035818487, "grad_norm": 0.5069397689510828, "learning_rate": 9.840799986645753e-05, "loss": 3.3019, "step": 8336 }, { "epoch": 0.5175367806816066, "grad_norm": 0.5119662960086583, "learning_rate": 9.84070956593204e-05, "loss": 3.1891, "step": 8337 }, { "epoch": 0.5175988577813645, "grad_norm": 0.48893273188896696, "learning_rate": 9.840619119963171e-05, "loss": 3.281, "step": 8338 }, { "epoch": 0.5176609348811223, "grad_norm": 0.5962128006629147, "learning_rate": 9.840528648739615e-05, "loss": 3.3296, "step": 8339 }, { "epoch": 0.5177230119808802, "grad_norm": 0.5831643921445346, "learning_rate": 9.840438152261849e-05, "loss": 3.3285, "step": 8340 }, { "epoch": 0.5177850890806381, "grad_norm": 0.47062685653300074, "learning_rate": 9.840347630530338e-05, "loss": 3.2264, "step": 8341 }, { "epoch": 0.5178471661803961, "grad_norm": 0.5208365313113897, "learning_rate": 9.840257083545562e-05, "loss": 3.2171, "step": 8342 }, { "epoch": 0.517909243280154, "grad_norm": 0.5857388986340645, "learning_rate": 9.840166511307988e-05, "loss": 3.2225, "step": 8343 }, { "epoch": 0.5179713203799119, "grad_norm": 0.5636355805426553, "learning_rate": 9.84007591381809e-05, "loss": 3.2852, "step": 8344 }, { "epoch": 0.5180333974796697, "grad_norm": 0.4785356167100977, "learning_rate": 9.839985291076341e-05, "loss": 3.2478, "step": 8345 }, { "epoch": 0.5180954745794276, "grad_norm": 0.584030863975125, "learning_rate": 9.839894643083216e-05, "loss": 3.2359, "step": 8346 }, { "epoch": 0.5181575516791855, "grad_norm": 0.6386131702939489, "learning_rate": 9.839803969839183e-05, "loss": 3.2488, "step": 8347 }, { "epoch": 0.5182196287789435, "grad_norm": 0.6797486639455143, "learning_rate": 9.839713271344719e-05, "loss": 3.2562, "step": 8348 }, { "epoch": 0.5182817058787014, "grad_norm": 0.5070126028996261, "learning_rate": 9.839622547600295e-05, "loss": 3.2726, "step": 8349 }, { "epoch": 0.5183437829784593, "grad_norm": 0.4983231470120564, "learning_rate": 9.839531798606387e-05, "loss": 3.3552, "step": 8350 }, { "epoch": 0.5184058600782171, "grad_norm": 0.5859602605849539, "learning_rate": 9.839441024363467e-05, "loss": 3.2445, "step": 8351 }, { "epoch": 0.518467937177975, "grad_norm": 0.4997260565133844, "learning_rate": 9.839350224872008e-05, "loss": 3.1975, "step": 8352 }, { "epoch": 0.5185300142777329, "grad_norm": 0.48394718042638685, "learning_rate": 9.839259400132483e-05, "loss": 3.3216, "step": 8353 }, { "epoch": 0.5185920913774908, "grad_norm": 0.5449812388471913, "learning_rate": 9.839168550145367e-05, "loss": 3.3231, "step": 8354 }, { "epoch": 0.5186541684772488, "grad_norm": 0.5324829883723612, "learning_rate": 9.839077674911135e-05, "loss": 3.2205, "step": 8355 }, { "epoch": 0.5187162455770067, "grad_norm": 0.45285180505058187, "learning_rate": 9.838986774430258e-05, "loss": 3.2644, "step": 8356 }, { "epoch": 0.5187783226767645, "grad_norm": 0.539939774669294, "learning_rate": 9.838895848703213e-05, "loss": 3.2497, "step": 8357 }, { "epoch": 0.5188403997765224, "grad_norm": 0.5816947168902824, "learning_rate": 9.838804897730476e-05, "loss": 3.254, "step": 8358 }, { "epoch": 0.5189024768762803, "grad_norm": 0.5083983636387798, "learning_rate": 9.838713921512516e-05, "loss": 3.2135, "step": 8359 }, { "epoch": 0.5189645539760382, "grad_norm": 0.5042150571420139, "learning_rate": 9.838622920049814e-05, "loss": 3.3284, "step": 8360 }, { "epoch": 0.5190266310757962, "grad_norm": 0.5182927793888567, "learning_rate": 9.838531893342839e-05, "loss": 3.2223, "step": 8361 }, { "epoch": 0.5190887081755541, "grad_norm": 0.6137779314168854, "learning_rate": 9.83844084139207e-05, "loss": 3.2436, "step": 8362 }, { "epoch": 0.5191507852753119, "grad_norm": 0.5544065919323191, "learning_rate": 9.83834976419798e-05, "loss": 3.1724, "step": 8363 }, { "epoch": 0.5192128623750698, "grad_norm": 0.45027661365200616, "learning_rate": 9.838258661761044e-05, "loss": 3.2952, "step": 8364 }, { "epoch": 0.5192749394748277, "grad_norm": 0.5285972955028355, "learning_rate": 9.838167534081739e-05, "loss": 3.3191, "step": 8365 }, { "epoch": 0.5193370165745856, "grad_norm": 0.48359414194611916, "learning_rate": 9.83807638116054e-05, "loss": 3.3179, "step": 8366 }, { "epoch": 0.5193990936743436, "grad_norm": 0.47854095703335886, "learning_rate": 9.837985202997922e-05, "loss": 3.3198, "step": 8367 }, { "epoch": 0.5194611707741015, "grad_norm": 0.5013607700889752, "learning_rate": 9.837893999594359e-05, "loss": 3.3149, "step": 8368 }, { "epoch": 0.5195232478738593, "grad_norm": 0.4779239343968041, "learning_rate": 9.83780277095033e-05, "loss": 3.3134, "step": 8369 }, { "epoch": 0.5195853249736172, "grad_norm": 0.43931648673293106, "learning_rate": 9.837711517066308e-05, "loss": 3.1536, "step": 8370 }, { "epoch": 0.5196474020733751, "grad_norm": 0.4456121373560619, "learning_rate": 9.837620237942773e-05, "loss": 3.2863, "step": 8371 }, { "epoch": 0.519709479173133, "grad_norm": 0.500478217791356, "learning_rate": 9.837528933580197e-05, "loss": 3.3321, "step": 8372 }, { "epoch": 0.519771556272891, "grad_norm": 0.5025735427407984, "learning_rate": 9.837437603979058e-05, "loss": 3.2798, "step": 8373 }, { "epoch": 0.5198336333726489, "grad_norm": 0.47879930767692064, "learning_rate": 9.837346249139834e-05, "loss": 3.2763, "step": 8374 }, { "epoch": 0.5198957104724067, "grad_norm": 0.5189949365104238, "learning_rate": 9.837254869063e-05, "loss": 3.3249, "step": 8375 }, { "epoch": 0.5199577875721646, "grad_norm": 0.601442993976197, "learning_rate": 9.837163463749033e-05, "loss": 3.1403, "step": 8376 }, { "epoch": 0.5200198646719225, "grad_norm": 0.6221246620761351, "learning_rate": 9.837072033198409e-05, "loss": 3.25, "step": 8377 }, { "epoch": 0.5200819417716804, "grad_norm": 0.5064318742304257, "learning_rate": 9.836980577411607e-05, "loss": 3.2229, "step": 8378 }, { "epoch": 0.5201440188714384, "grad_norm": 0.47497572711363073, "learning_rate": 9.836889096389103e-05, "loss": 3.2842, "step": 8379 }, { "epoch": 0.5202060959711963, "grad_norm": 0.5742668598807259, "learning_rate": 9.836797590131374e-05, "loss": 3.2528, "step": 8380 }, { "epoch": 0.5202681730709541, "grad_norm": 0.6186425645592767, "learning_rate": 9.836706058638898e-05, "loss": 3.2423, "step": 8381 }, { "epoch": 0.520330250170712, "grad_norm": 0.5208191933301933, "learning_rate": 9.836614501912153e-05, "loss": 3.2039, "step": 8382 }, { "epoch": 0.5203923272704699, "grad_norm": 0.513291250564023, "learning_rate": 9.836522919951615e-05, "loss": 3.3042, "step": 8383 }, { "epoch": 0.5204544043702278, "grad_norm": 0.49842891625932584, "learning_rate": 9.836431312757763e-05, "loss": 3.2956, "step": 8384 }, { "epoch": 0.5205164814699857, "grad_norm": 0.5433160782435862, "learning_rate": 9.836339680331074e-05, "loss": 3.3388, "step": 8385 }, { "epoch": 0.5205785585697437, "grad_norm": 0.551573460931986, "learning_rate": 9.83624802267203e-05, "loss": 3.1436, "step": 8386 }, { "epoch": 0.5206406356695015, "grad_norm": 0.48781320646382703, "learning_rate": 9.836156339781103e-05, "loss": 3.2485, "step": 8387 }, { "epoch": 0.5207027127692594, "grad_norm": 0.5660290744781596, "learning_rate": 9.836064631658776e-05, "loss": 3.2465, "step": 8388 }, { "epoch": 0.5207647898690173, "grad_norm": 0.4817792496504525, "learning_rate": 9.835972898305525e-05, "loss": 3.2411, "step": 8389 }, { "epoch": 0.5208268669687752, "grad_norm": 0.550054381673099, "learning_rate": 9.835881139721829e-05, "loss": 3.32, "step": 8390 }, { "epoch": 0.5208889440685331, "grad_norm": 0.6059592444466281, "learning_rate": 9.835789355908167e-05, "loss": 3.2614, "step": 8391 }, { "epoch": 0.5209510211682911, "grad_norm": 0.5075844472708927, "learning_rate": 9.83569754686502e-05, "loss": 3.2478, "step": 8392 }, { "epoch": 0.5210130982680489, "grad_norm": 0.49061724541239676, "learning_rate": 9.835605712592863e-05, "loss": 3.2201, "step": 8393 }, { "epoch": 0.5210751753678068, "grad_norm": 0.5033027524479109, "learning_rate": 9.835513853092178e-05, "loss": 3.3575, "step": 8394 }, { "epoch": 0.5211372524675647, "grad_norm": 0.5135184519423128, "learning_rate": 9.835421968363443e-05, "loss": 3.2134, "step": 8395 }, { "epoch": 0.5211993295673226, "grad_norm": 0.5013494460946182, "learning_rate": 9.835330058407137e-05, "loss": 3.1727, "step": 8396 }, { "epoch": 0.5212614066670805, "grad_norm": 0.5470098874089402, "learning_rate": 9.835238123223741e-05, "loss": 3.2535, "step": 8397 }, { "epoch": 0.5213234837668385, "grad_norm": 0.48941498932837074, "learning_rate": 9.835146162813735e-05, "loss": 3.2388, "step": 8398 }, { "epoch": 0.5213855608665963, "grad_norm": 0.5295714057021349, "learning_rate": 9.835054177177597e-05, "loss": 3.2797, "step": 8399 }, { "epoch": 0.5214476379663542, "grad_norm": 0.5627902130001234, "learning_rate": 9.834962166315806e-05, "loss": 3.2629, "step": 8400 }, { "epoch": 0.5215097150661121, "grad_norm": 0.5501481875278731, "learning_rate": 9.834870130228845e-05, "loss": 3.3251, "step": 8401 }, { "epoch": 0.52157179216587, "grad_norm": 0.575976606998332, "learning_rate": 9.834778068917193e-05, "loss": 3.2991, "step": 8402 }, { "epoch": 0.5216338692656279, "grad_norm": 0.4832418631319057, "learning_rate": 9.83468598238133e-05, "loss": 3.2853, "step": 8403 }, { "epoch": 0.5216959463653859, "grad_norm": 0.5503597134423759, "learning_rate": 9.834593870621735e-05, "loss": 3.2319, "step": 8404 }, { "epoch": 0.5217580234651437, "grad_norm": 0.4905321045419915, "learning_rate": 9.834501733638892e-05, "loss": 3.3371, "step": 8405 }, { "epoch": 0.5218201005649016, "grad_norm": 0.5127938893327529, "learning_rate": 9.83440957143328e-05, "loss": 3.2637, "step": 8406 }, { "epoch": 0.5218821776646595, "grad_norm": 0.5599989921696994, "learning_rate": 9.834317384005379e-05, "loss": 3.2244, "step": 8407 }, { "epoch": 0.5219442547644174, "grad_norm": 0.5732783187579171, "learning_rate": 9.834225171355671e-05, "loss": 3.2228, "step": 8408 }, { "epoch": 0.5220063318641753, "grad_norm": 0.5506451215325802, "learning_rate": 9.834132933484637e-05, "loss": 3.2174, "step": 8409 }, { "epoch": 0.5220684089639333, "grad_norm": 0.5253049645913123, "learning_rate": 9.834040670392757e-05, "loss": 3.1992, "step": 8410 }, { "epoch": 0.5221304860636911, "grad_norm": 0.5559530194042465, "learning_rate": 9.833948382080514e-05, "loss": 3.2905, "step": 8411 }, { "epoch": 0.522192563163449, "grad_norm": 0.5482736141600429, "learning_rate": 9.833856068548389e-05, "loss": 3.2277, "step": 8412 }, { "epoch": 0.5222546402632069, "grad_norm": 0.4792037842375447, "learning_rate": 9.833763729796863e-05, "loss": 3.2341, "step": 8413 }, { "epoch": 0.5223167173629648, "grad_norm": 0.4788275966806309, "learning_rate": 9.833671365826417e-05, "loss": 3.2404, "step": 8414 }, { "epoch": 0.5223787944627227, "grad_norm": 0.5545333056977435, "learning_rate": 9.833578976637536e-05, "loss": 3.2743, "step": 8415 }, { "epoch": 0.5224408715624806, "grad_norm": 0.5576817463449013, "learning_rate": 9.8334865622307e-05, "loss": 3.2074, "step": 8416 }, { "epoch": 0.5225029486622385, "grad_norm": 0.5240243781639753, "learning_rate": 9.83339412260639e-05, "loss": 3.2656, "step": 8417 }, { "epoch": 0.5225650257619964, "grad_norm": 0.5961389417003898, "learning_rate": 9.833301657765091e-05, "loss": 3.2096, "step": 8418 }, { "epoch": 0.5226271028617543, "grad_norm": 0.5757171388127416, "learning_rate": 9.833209167707283e-05, "loss": 3.2429, "step": 8419 }, { "epoch": 0.5226891799615122, "grad_norm": 0.575736291752827, "learning_rate": 9.833116652433449e-05, "loss": 3.2993, "step": 8420 }, { "epoch": 0.5227512570612701, "grad_norm": 0.6415523934067635, "learning_rate": 9.833024111944071e-05, "loss": 3.3923, "step": 8421 }, { "epoch": 0.522813334161028, "grad_norm": 0.5721344260367938, "learning_rate": 9.832931546239636e-05, "loss": 3.2285, "step": 8422 }, { "epoch": 0.5228754112607858, "grad_norm": 0.5348016162516923, "learning_rate": 9.832838955320622e-05, "loss": 3.2394, "step": 8423 }, { "epoch": 0.5229374883605438, "grad_norm": 0.4917862937219188, "learning_rate": 9.832746339187515e-05, "loss": 3.1415, "step": 8424 }, { "epoch": 0.5229995654603017, "grad_norm": 0.5653190702243173, "learning_rate": 9.832653697840797e-05, "loss": 3.2026, "step": 8425 }, { "epoch": 0.5230616425600596, "grad_norm": 0.47413399518705307, "learning_rate": 9.83256103128095e-05, "loss": 3.1641, "step": 8426 }, { "epoch": 0.5231237196598175, "grad_norm": 0.5070567743257185, "learning_rate": 9.832468339508461e-05, "loss": 3.1632, "step": 8427 }, { "epoch": 0.5231857967595754, "grad_norm": 0.5696313265142978, "learning_rate": 9.83237562252381e-05, "loss": 3.2209, "step": 8428 }, { "epoch": 0.5232478738593332, "grad_norm": 0.5571985226330886, "learning_rate": 9.832282880327483e-05, "loss": 3.3151, "step": 8429 }, { "epoch": 0.5233099509590912, "grad_norm": 0.5623453245488048, "learning_rate": 9.832190112919963e-05, "loss": 3.1995, "step": 8430 }, { "epoch": 0.5233720280588491, "grad_norm": 0.4588950536944717, "learning_rate": 9.832097320301734e-05, "loss": 3.2295, "step": 8431 }, { "epoch": 0.523434105158607, "grad_norm": 0.480053093849728, "learning_rate": 9.83200450247328e-05, "loss": 3.2678, "step": 8432 }, { "epoch": 0.5234961822583649, "grad_norm": 0.4666546880371095, "learning_rate": 9.831911659435085e-05, "loss": 3.1977, "step": 8433 }, { "epoch": 0.5235582593581228, "grad_norm": 0.48179536361791303, "learning_rate": 9.831818791187636e-05, "loss": 3.2972, "step": 8434 }, { "epoch": 0.5236203364578806, "grad_norm": 0.5143122579379549, "learning_rate": 9.831725897731414e-05, "loss": 3.2974, "step": 8435 }, { "epoch": 0.5236824135576386, "grad_norm": 0.5481205516186658, "learning_rate": 9.831632979066904e-05, "loss": 3.2081, "step": 8436 }, { "epoch": 0.5237444906573965, "grad_norm": 0.4790247952478474, "learning_rate": 9.831540035194594e-05, "loss": 3.2497, "step": 8437 }, { "epoch": 0.5238065677571544, "grad_norm": 0.4477340876895034, "learning_rate": 9.831447066114966e-05, "loss": 3.296, "step": 8438 }, { "epoch": 0.5238686448569123, "grad_norm": 0.501143637714269, "learning_rate": 9.831354071828505e-05, "loss": 3.3132, "step": 8439 }, { "epoch": 0.5239307219566702, "grad_norm": 0.5305465360305326, "learning_rate": 9.831261052335698e-05, "loss": 3.3383, "step": 8440 }, { "epoch": 0.523992799056428, "grad_norm": 0.49724954847312436, "learning_rate": 9.831168007637029e-05, "loss": 3.2608, "step": 8441 }, { "epoch": 0.524054876156186, "grad_norm": 0.5276046390161034, "learning_rate": 9.831074937732983e-05, "loss": 3.2007, "step": 8442 }, { "epoch": 0.5241169532559439, "grad_norm": 0.48504527691754606, "learning_rate": 9.830981842624047e-05, "loss": 3.2749, "step": 8443 }, { "epoch": 0.5241790303557018, "grad_norm": 0.4498223934423924, "learning_rate": 9.830888722310704e-05, "loss": 3.3116, "step": 8444 }, { "epoch": 0.5242411074554597, "grad_norm": 0.4843438815303697, "learning_rate": 9.830795576793446e-05, "loss": 3.2536, "step": 8445 }, { "epoch": 0.5243031845552176, "grad_norm": 0.5057125099286036, "learning_rate": 9.830702406072751e-05, "loss": 3.2863, "step": 8446 }, { "epoch": 0.5243652616549754, "grad_norm": 0.49975675388064184, "learning_rate": 9.830609210149111e-05, "loss": 3.2093, "step": 8447 }, { "epoch": 0.5244273387547334, "grad_norm": 0.403833187881115, "learning_rate": 9.830515989023009e-05, "loss": 3.2492, "step": 8448 }, { "epoch": 0.5244894158544913, "grad_norm": 0.4736503676474373, "learning_rate": 9.830422742694933e-05, "loss": 3.1955, "step": 8449 }, { "epoch": 0.5245514929542492, "grad_norm": 0.4684436591814197, "learning_rate": 9.830329471165366e-05, "loss": 3.212, "step": 8450 }, { "epoch": 0.5246135700540071, "grad_norm": 0.4526928123110806, "learning_rate": 9.830236174434801e-05, "loss": 3.2424, "step": 8451 }, { "epoch": 0.524675647153765, "grad_norm": 0.5496274889554964, "learning_rate": 9.830142852503721e-05, "loss": 3.1919, "step": 8452 }, { "epoch": 0.5247377242535228, "grad_norm": 0.5748795839356282, "learning_rate": 9.830049505372613e-05, "loss": 3.2073, "step": 8453 }, { "epoch": 0.5247998013532807, "grad_norm": 0.4404273360838786, "learning_rate": 9.829956133041965e-05, "loss": 3.2006, "step": 8454 }, { "epoch": 0.5248618784530387, "grad_norm": 0.5091466836655175, "learning_rate": 9.82986273551226e-05, "loss": 3.2426, "step": 8455 }, { "epoch": 0.5249239555527966, "grad_norm": 0.5266270040514246, "learning_rate": 9.829769312783992e-05, "loss": 3.2345, "step": 8456 }, { "epoch": 0.5249860326525545, "grad_norm": 0.5464552767289143, "learning_rate": 9.829675864857644e-05, "loss": 3.1886, "step": 8457 }, { "epoch": 0.5250481097523124, "grad_norm": 0.5283958959190511, "learning_rate": 9.829582391733703e-05, "loss": 3.299, "step": 8458 }, { "epoch": 0.5251101868520702, "grad_norm": 0.46610876465212997, "learning_rate": 9.829488893412659e-05, "loss": 3.2744, "step": 8459 }, { "epoch": 0.5251722639518281, "grad_norm": 0.5644268306956779, "learning_rate": 9.829395369895e-05, "loss": 3.2437, "step": 8460 }, { "epoch": 0.5252343410515861, "grad_norm": 0.5291531075470067, "learning_rate": 9.829301821181213e-05, "loss": 3.1942, "step": 8461 }, { "epoch": 0.525296418151344, "grad_norm": 0.530826680437654, "learning_rate": 9.829208247271786e-05, "loss": 3.1435, "step": 8462 }, { "epoch": 0.5253584952511019, "grad_norm": 0.5251286076651298, "learning_rate": 9.829114648167207e-05, "loss": 3.1996, "step": 8463 }, { "epoch": 0.5254205723508598, "grad_norm": 0.646149286284182, "learning_rate": 9.829021023867964e-05, "loss": 3.1946, "step": 8464 }, { "epoch": 0.5254826494506176, "grad_norm": 0.6685777308610367, "learning_rate": 9.828927374374547e-05, "loss": 3.1867, "step": 8465 }, { "epoch": 0.5255447265503755, "grad_norm": 0.5363323069810184, "learning_rate": 9.828833699687443e-05, "loss": 3.2704, "step": 8466 }, { "epoch": 0.5256068036501335, "grad_norm": 0.5859496051809461, "learning_rate": 9.828739999807141e-05, "loss": 3.2756, "step": 8467 }, { "epoch": 0.5256688807498914, "grad_norm": 0.578925969744224, "learning_rate": 9.828646274734131e-05, "loss": 3.3232, "step": 8468 }, { "epoch": 0.5257309578496493, "grad_norm": 0.5902724318792075, "learning_rate": 9.828552524468901e-05, "loss": 3.2212, "step": 8469 }, { "epoch": 0.5257930349494072, "grad_norm": 0.5231742766402971, "learning_rate": 9.82845874901194e-05, "loss": 3.2212, "step": 8470 }, { "epoch": 0.525855112049165, "grad_norm": 0.4746172398045068, "learning_rate": 9.828364948363737e-05, "loss": 3.1571, "step": 8471 }, { "epoch": 0.5259171891489229, "grad_norm": 0.5413496453005449, "learning_rate": 9.828271122524784e-05, "loss": 3.3742, "step": 8472 }, { "epoch": 0.5259792662486809, "grad_norm": 0.6198123212519614, "learning_rate": 9.828177271495566e-05, "loss": 3.2659, "step": 8473 }, { "epoch": 0.5260413433484388, "grad_norm": 0.554648760665132, "learning_rate": 9.828083395276577e-05, "loss": 3.2053, "step": 8474 }, { "epoch": 0.5261034204481967, "grad_norm": 0.5607637789883291, "learning_rate": 9.827989493868304e-05, "loss": 3.2772, "step": 8475 }, { "epoch": 0.5261654975479546, "grad_norm": 0.6196249141820033, "learning_rate": 9.827895567271236e-05, "loss": 3.3093, "step": 8476 }, { "epoch": 0.5262275746477124, "grad_norm": 0.6023566713443889, "learning_rate": 9.827801615485867e-05, "loss": 3.343, "step": 8477 }, { "epoch": 0.5262896517474703, "grad_norm": 0.5163700616249963, "learning_rate": 9.827707638512687e-05, "loss": 3.2832, "step": 8478 }, { "epoch": 0.5263517288472283, "grad_norm": 0.5335615860501224, "learning_rate": 9.827613636352181e-05, "loss": 3.3156, "step": 8479 }, { "epoch": 0.5264138059469862, "grad_norm": 0.5191924632579027, "learning_rate": 9.827519609004843e-05, "loss": 3.3433, "step": 8480 }, { "epoch": 0.5264758830467441, "grad_norm": 0.5593115189829625, "learning_rate": 9.827425556471163e-05, "loss": 3.3373, "step": 8481 }, { "epoch": 0.526537960146502, "grad_norm": 0.5221339548396499, "learning_rate": 9.827331478751634e-05, "loss": 3.2769, "step": 8482 }, { "epoch": 0.5266000372462598, "grad_norm": 0.5070553931085345, "learning_rate": 9.827237375846744e-05, "loss": 3.234, "step": 8483 }, { "epoch": 0.5266621143460177, "grad_norm": 0.4972165834946574, "learning_rate": 9.827143247756985e-05, "loss": 3.2249, "step": 8484 }, { "epoch": 0.5267241914457756, "grad_norm": 0.46788764148830375, "learning_rate": 9.827049094482846e-05, "loss": 3.3277, "step": 8485 }, { "epoch": 0.5267862685455336, "grad_norm": 0.5744554978789481, "learning_rate": 9.826954916024822e-05, "loss": 3.3195, "step": 8486 }, { "epoch": 0.5268483456452915, "grad_norm": 0.48291600361708026, "learning_rate": 9.826860712383401e-05, "loss": 3.2142, "step": 8487 }, { "epoch": 0.5269104227450494, "grad_norm": 0.4917046130887378, "learning_rate": 9.826766483559078e-05, "loss": 3.2444, "step": 8488 }, { "epoch": 0.5269724998448072, "grad_norm": 0.46059047437595674, "learning_rate": 9.826672229552341e-05, "loss": 3.3277, "step": 8489 }, { "epoch": 0.5270345769445651, "grad_norm": 0.4698324262639131, "learning_rate": 9.826577950363682e-05, "loss": 3.1505, "step": 8490 }, { "epoch": 0.527096654044323, "grad_norm": 0.517779818408835, "learning_rate": 9.826483645993596e-05, "loss": 3.2216, "step": 8491 }, { "epoch": 0.527158731144081, "grad_norm": 0.46858501709842393, "learning_rate": 9.826389316442574e-05, "loss": 3.2622, "step": 8492 }, { "epoch": 0.5272208082438389, "grad_norm": 0.6297104947267121, "learning_rate": 9.826294961711105e-05, "loss": 3.2694, "step": 8493 }, { "epoch": 0.5272828853435968, "grad_norm": 0.6157997803028231, "learning_rate": 9.826200581799684e-05, "loss": 3.2114, "step": 8494 }, { "epoch": 0.5273449624433546, "grad_norm": 0.47508514168263616, "learning_rate": 9.826106176708805e-05, "loss": 3.3083, "step": 8495 }, { "epoch": 0.5274070395431125, "grad_norm": 0.494740931938152, "learning_rate": 9.826011746438957e-05, "loss": 3.1976, "step": 8496 }, { "epoch": 0.5274691166428704, "grad_norm": 0.5792402644930111, "learning_rate": 9.825917290990633e-05, "loss": 3.1245, "step": 8497 }, { "epoch": 0.5275311937426284, "grad_norm": 0.5518569452134043, "learning_rate": 9.82582281036433e-05, "loss": 3.2149, "step": 8498 }, { "epoch": 0.5275932708423863, "grad_norm": 0.5376965451737369, "learning_rate": 9.825728304560535e-05, "loss": 3.297, "step": 8499 }, { "epoch": 0.5276553479421442, "grad_norm": 0.5154485111478165, "learning_rate": 9.825633773579746e-05, "loss": 3.2142, "step": 8500 }, { "epoch": 0.527717425041902, "grad_norm": 0.5881844733527894, "learning_rate": 9.825539217422452e-05, "loss": 3.2774, "step": 8501 }, { "epoch": 0.5277795021416599, "grad_norm": 0.5837613448566071, "learning_rate": 9.82544463608915e-05, "loss": 3.2794, "step": 8502 }, { "epoch": 0.5278415792414178, "grad_norm": 0.5371405575110246, "learning_rate": 9.825350029580332e-05, "loss": 3.2667, "step": 8503 }, { "epoch": 0.5279036563411758, "grad_norm": 0.5823056642991556, "learning_rate": 9.825255397896491e-05, "loss": 3.2415, "step": 8504 }, { "epoch": 0.5279657334409337, "grad_norm": 0.559085416158899, "learning_rate": 9.825160741038121e-05, "loss": 3.1972, "step": 8505 }, { "epoch": 0.5280278105406916, "grad_norm": 0.5287281936856407, "learning_rate": 9.825066059005716e-05, "loss": 3.2411, "step": 8506 }, { "epoch": 0.5280898876404494, "grad_norm": 0.7423342910963691, "learning_rate": 9.82497135179977e-05, "loss": 3.2316, "step": 8507 }, { "epoch": 0.5281519647402073, "grad_norm": 0.6128855844245427, "learning_rate": 9.824876619420777e-05, "loss": 3.2409, "step": 8508 }, { "epoch": 0.5282140418399652, "grad_norm": 0.5960932970357751, "learning_rate": 9.824781861869232e-05, "loss": 3.3861, "step": 8509 }, { "epoch": 0.5282761189397231, "grad_norm": 0.630008276930434, "learning_rate": 9.82468707914563e-05, "loss": 3.1729, "step": 8510 }, { "epoch": 0.5283381960394811, "grad_norm": 0.5356260590336945, "learning_rate": 9.824592271250462e-05, "loss": 3.1677, "step": 8511 }, { "epoch": 0.528400273139239, "grad_norm": 0.5869750944829549, "learning_rate": 9.824497438184226e-05, "loss": 3.2307, "step": 8512 }, { "epoch": 0.5284623502389968, "grad_norm": 0.5920521654475214, "learning_rate": 9.824402579947416e-05, "loss": 3.2146, "step": 8513 }, { "epoch": 0.5285244273387547, "grad_norm": 0.7939662203317359, "learning_rate": 9.824307696540527e-05, "loss": 3.2718, "step": 8514 }, { "epoch": 0.5285865044385126, "grad_norm": 0.9666254373231493, "learning_rate": 9.824212787964051e-05, "loss": 3.2923, "step": 8515 }, { "epoch": 0.5286485815382705, "grad_norm": 0.7104890268191179, "learning_rate": 9.824117854218489e-05, "loss": 3.2116, "step": 8516 }, { "epoch": 0.5287106586380285, "grad_norm": 0.5599761609555093, "learning_rate": 9.82402289530433e-05, "loss": 3.3466, "step": 8517 }, { "epoch": 0.5287727357377864, "grad_norm": 0.5372265497186975, "learning_rate": 9.823927911222075e-05, "loss": 3.1176, "step": 8518 }, { "epoch": 0.5288348128375442, "grad_norm": 0.5480582107600575, "learning_rate": 9.823832901972217e-05, "loss": 3.3061, "step": 8519 }, { "epoch": 0.5288968899373021, "grad_norm": 0.5644631302418702, "learning_rate": 9.823737867555249e-05, "loss": 3.394, "step": 8520 }, { "epoch": 0.52895896703706, "grad_norm": 0.6427338709979378, "learning_rate": 9.823642807971671e-05, "loss": 3.2161, "step": 8521 }, { "epoch": 0.5290210441368179, "grad_norm": 0.63065712819521, "learning_rate": 9.82354772322198e-05, "loss": 3.3438, "step": 8522 }, { "epoch": 0.5290831212365759, "grad_norm": 0.5323157279541637, "learning_rate": 9.823452613306666e-05, "loss": 3.2926, "step": 8523 }, { "epoch": 0.5291451983363338, "grad_norm": 0.5487906826727682, "learning_rate": 9.82335747822623e-05, "loss": 3.2477, "step": 8524 }, { "epoch": 0.5292072754360916, "grad_norm": 0.6273713662801772, "learning_rate": 9.823262317981167e-05, "loss": 3.256, "step": 8525 }, { "epoch": 0.5292693525358495, "grad_norm": 0.5925515779346279, "learning_rate": 9.823167132571975e-05, "loss": 3.121, "step": 8526 }, { "epoch": 0.5293314296356074, "grad_norm": 0.6768196815295006, "learning_rate": 9.823071921999148e-05, "loss": 3.2521, "step": 8527 }, { "epoch": 0.5293935067353653, "grad_norm": 0.5764688276165836, "learning_rate": 9.822976686263184e-05, "loss": 3.2636, "step": 8528 }, { "epoch": 0.5294555838351233, "grad_norm": 0.5897476673815862, "learning_rate": 9.82288142536458e-05, "loss": 3.2175, "step": 8529 }, { "epoch": 0.5295176609348812, "grad_norm": 0.5251206622653972, "learning_rate": 9.822786139303833e-05, "loss": 3.2114, "step": 8530 }, { "epoch": 0.529579738034639, "grad_norm": 0.469723009504494, "learning_rate": 9.822690828081439e-05, "loss": 3.2732, "step": 8531 }, { "epoch": 0.5296418151343969, "grad_norm": 0.5429692988770269, "learning_rate": 9.822595491697898e-05, "loss": 3.2913, "step": 8532 }, { "epoch": 0.5297038922341548, "grad_norm": 0.5783063758683321, "learning_rate": 9.822500130153703e-05, "loss": 3.3007, "step": 8533 }, { "epoch": 0.5297659693339127, "grad_norm": 0.5123995554582602, "learning_rate": 9.822404743449356e-05, "loss": 3.2048, "step": 8534 }, { "epoch": 0.5298280464336707, "grad_norm": 0.5610501093984085, "learning_rate": 9.822309331585352e-05, "loss": 3.3568, "step": 8535 }, { "epoch": 0.5298901235334286, "grad_norm": 0.5488278263403463, "learning_rate": 9.82221389456219e-05, "loss": 3.3095, "step": 8536 }, { "epoch": 0.5299522006331864, "grad_norm": 0.5868953505086355, "learning_rate": 9.822118432380368e-05, "loss": 3.3022, "step": 8537 }, { "epoch": 0.5300142777329443, "grad_norm": 0.5574617987746319, "learning_rate": 9.822022945040382e-05, "loss": 3.1741, "step": 8538 }, { "epoch": 0.5300763548327022, "grad_norm": 0.509220840669866, "learning_rate": 9.821927432542733e-05, "loss": 3.2106, "step": 8539 }, { "epoch": 0.5301384319324601, "grad_norm": 0.522569249414666, "learning_rate": 9.821831894887918e-05, "loss": 3.2752, "step": 8540 }, { "epoch": 0.530200509032218, "grad_norm": 0.5538266984337616, "learning_rate": 9.821736332076434e-05, "loss": 3.2496, "step": 8541 }, { "epoch": 0.530262586131976, "grad_norm": 0.5945968110558136, "learning_rate": 9.821640744108782e-05, "loss": 3.2074, "step": 8542 }, { "epoch": 0.5303246632317338, "grad_norm": 0.5060742485243951, "learning_rate": 9.821545130985461e-05, "loss": 3.2574, "step": 8543 }, { "epoch": 0.5303867403314917, "grad_norm": 0.5382896079423583, "learning_rate": 9.821449492706967e-05, "loss": 3.3037, "step": 8544 }, { "epoch": 0.5304488174312496, "grad_norm": 0.5852395287448362, "learning_rate": 9.821353829273801e-05, "loss": 3.2783, "step": 8545 }, { "epoch": 0.5305108945310075, "grad_norm": 0.5266923908347425, "learning_rate": 9.821258140686462e-05, "loss": 3.3094, "step": 8546 }, { "epoch": 0.5305729716307654, "grad_norm": 0.5415348907494487, "learning_rate": 9.821162426945447e-05, "loss": 3.3286, "step": 8547 }, { "epoch": 0.5306350487305234, "grad_norm": 0.6024794093052308, "learning_rate": 9.821066688051259e-05, "loss": 3.2654, "step": 8548 }, { "epoch": 0.5306971258302812, "grad_norm": 0.5455976828065172, "learning_rate": 9.820970924004395e-05, "loss": 3.1273, "step": 8549 }, { "epoch": 0.5307592029300391, "grad_norm": 0.4634376223378139, "learning_rate": 9.820875134805356e-05, "loss": 3.1999, "step": 8550 }, { "epoch": 0.530821280029797, "grad_norm": 0.515435097143096, "learning_rate": 9.820779320454643e-05, "loss": 3.2382, "step": 8551 }, { "epoch": 0.5308833571295549, "grad_norm": 0.5227667653201659, "learning_rate": 9.820683480952751e-05, "loss": 3.2306, "step": 8552 }, { "epoch": 0.5309454342293128, "grad_norm": 0.5984818982419986, "learning_rate": 9.820587616300184e-05, "loss": 3.2388, "step": 8553 }, { "epoch": 0.5310075113290708, "grad_norm": 0.46019971930421927, "learning_rate": 9.820491726497441e-05, "loss": 3.2599, "step": 8554 }, { "epoch": 0.5310695884288286, "grad_norm": 0.5252143819516901, "learning_rate": 9.820395811545023e-05, "loss": 3.1564, "step": 8555 }, { "epoch": 0.5311316655285865, "grad_norm": 0.48442354266667076, "learning_rate": 9.820299871443429e-05, "loss": 3.1202, "step": 8556 }, { "epoch": 0.5311937426283444, "grad_norm": 0.5245396980375768, "learning_rate": 9.820203906193162e-05, "loss": 3.1443, "step": 8557 }, { "epoch": 0.5312558197281023, "grad_norm": 0.45242759046203446, "learning_rate": 9.820107915794718e-05, "loss": 3.2838, "step": 8558 }, { "epoch": 0.5313178968278602, "grad_norm": 0.4826091294829734, "learning_rate": 9.820011900248604e-05, "loss": 3.2629, "step": 8559 }, { "epoch": 0.5313799739276182, "grad_norm": 0.4509304110577196, "learning_rate": 9.819915859555317e-05, "loss": 3.193, "step": 8560 }, { "epoch": 0.531442051027376, "grad_norm": 0.5043602014991793, "learning_rate": 9.819819793715356e-05, "loss": 3.3085, "step": 8561 }, { "epoch": 0.5315041281271339, "grad_norm": 0.5629613362040402, "learning_rate": 9.819723702729229e-05, "loss": 3.3139, "step": 8562 }, { "epoch": 0.5315662052268918, "grad_norm": 0.4837650026661348, "learning_rate": 9.819627586597433e-05, "loss": 3.2232, "step": 8563 }, { "epoch": 0.5316282823266497, "grad_norm": 0.4708321738721136, "learning_rate": 9.819531445320469e-05, "loss": 3.2984, "step": 8564 }, { "epoch": 0.5316903594264076, "grad_norm": 0.45965557936696566, "learning_rate": 9.819435278898839e-05, "loss": 3.333, "step": 8565 }, { "epoch": 0.5317524365261656, "grad_norm": 0.4700041048144604, "learning_rate": 9.819339087333045e-05, "loss": 3.2863, "step": 8566 }, { "epoch": 0.5318145136259234, "grad_norm": 0.6328403325494609, "learning_rate": 9.81924287062359e-05, "loss": 3.2236, "step": 8567 }, { "epoch": 0.5318765907256813, "grad_norm": 0.5741114613477175, "learning_rate": 9.819146628770975e-05, "loss": 3.3891, "step": 8568 }, { "epoch": 0.5319386678254392, "grad_norm": 0.5467080542648453, "learning_rate": 9.819050361775701e-05, "loss": 3.2705, "step": 8569 }, { "epoch": 0.5320007449251971, "grad_norm": 0.626378232539173, "learning_rate": 9.818954069638272e-05, "loss": 3.1295, "step": 8570 }, { "epoch": 0.532062822024955, "grad_norm": 0.7558640669645031, "learning_rate": 9.81885775235919e-05, "loss": 3.3693, "step": 8571 }, { "epoch": 0.532124899124713, "grad_norm": 0.6833902546582704, "learning_rate": 9.818761409938957e-05, "loss": 3.2302, "step": 8572 }, { "epoch": 0.5321869762244708, "grad_norm": 0.535046225048181, "learning_rate": 9.818665042378076e-05, "loss": 3.29, "step": 8573 }, { "epoch": 0.5322490533242287, "grad_norm": 0.5983853349783286, "learning_rate": 9.81856864967705e-05, "loss": 3.347, "step": 8574 }, { "epoch": 0.5323111304239866, "grad_norm": 0.6766508163708246, "learning_rate": 9.818472231836382e-05, "loss": 3.2682, "step": 8575 }, { "epoch": 0.5323732075237445, "grad_norm": 0.6158544896904818, "learning_rate": 9.818375788856574e-05, "loss": 3.3268, "step": 8576 }, { "epoch": 0.5324352846235024, "grad_norm": 0.6390869704880153, "learning_rate": 9.818279320738129e-05, "loss": 3.3278, "step": 8577 }, { "epoch": 0.5324973617232603, "grad_norm": 0.7534946642422861, "learning_rate": 9.818182827481553e-05, "loss": 3.1694, "step": 8578 }, { "epoch": 0.5325594388230181, "grad_norm": 0.6129774568217956, "learning_rate": 9.818086309087347e-05, "loss": 3.3689, "step": 8579 }, { "epoch": 0.5326215159227761, "grad_norm": 0.6062954580823631, "learning_rate": 9.817989765556014e-05, "loss": 3.2402, "step": 8580 }, { "epoch": 0.532683593022534, "grad_norm": 0.5230656911757415, "learning_rate": 9.817893196888059e-05, "loss": 3.2264, "step": 8581 }, { "epoch": 0.5327456701222919, "grad_norm": 0.5781270362199545, "learning_rate": 9.817796603083987e-05, "loss": 3.2407, "step": 8582 }, { "epoch": 0.5328077472220498, "grad_norm": 0.6301233822862912, "learning_rate": 9.8176999841443e-05, "loss": 3.2569, "step": 8583 }, { "epoch": 0.5328698243218077, "grad_norm": 0.5766113932484588, "learning_rate": 9.817603340069501e-05, "loss": 3.218, "step": 8584 }, { "epoch": 0.5329319014215655, "grad_norm": 0.530194215143145, "learning_rate": 9.817506670860097e-05, "loss": 3.3089, "step": 8585 }, { "epoch": 0.5329939785213235, "grad_norm": 0.6345126073259623, "learning_rate": 9.817409976516592e-05, "loss": 3.3043, "step": 8586 }, { "epoch": 0.5330560556210814, "grad_norm": 0.4643272126563011, "learning_rate": 9.817313257039488e-05, "loss": 3.1456, "step": 8587 }, { "epoch": 0.5331181327208393, "grad_norm": 0.5227765422433263, "learning_rate": 9.817216512429292e-05, "loss": 3.2125, "step": 8588 }, { "epoch": 0.5331802098205972, "grad_norm": 0.5679371369810845, "learning_rate": 9.817119742686509e-05, "loss": 3.229, "step": 8589 }, { "epoch": 0.5332422869203551, "grad_norm": 0.48843375304011927, "learning_rate": 9.817022947811643e-05, "loss": 3.1413, "step": 8590 }, { "epoch": 0.5333043640201129, "grad_norm": 0.46623035992939954, "learning_rate": 9.816926127805197e-05, "loss": 3.0356, "step": 8591 }, { "epoch": 0.5333664411198709, "grad_norm": 0.4742986030528283, "learning_rate": 9.81682928266768e-05, "loss": 3.2221, "step": 8592 }, { "epoch": 0.5334285182196288, "grad_norm": 0.45646423640134653, "learning_rate": 9.816732412399594e-05, "loss": 3.2749, "step": 8593 }, { "epoch": 0.5334905953193867, "grad_norm": 0.47545706630245227, "learning_rate": 9.816635517001447e-05, "loss": 3.3427, "step": 8594 }, { "epoch": 0.5335526724191446, "grad_norm": 0.46894585968074626, "learning_rate": 9.816538596473742e-05, "loss": 3.2598, "step": 8595 }, { "epoch": 0.5336147495189025, "grad_norm": 0.47493404696002806, "learning_rate": 9.816441650816986e-05, "loss": 3.3114, "step": 8596 }, { "epoch": 0.5336768266186603, "grad_norm": 0.46775868816655974, "learning_rate": 9.816344680031687e-05, "loss": 3.2598, "step": 8597 }, { "epoch": 0.5337389037184183, "grad_norm": 0.4902115774743084, "learning_rate": 9.816247684118345e-05, "loss": 3.1157, "step": 8598 }, { "epoch": 0.5338009808181762, "grad_norm": 0.5280228939331008, "learning_rate": 9.816150663077471e-05, "loss": 3.2613, "step": 8599 }, { "epoch": 0.5338630579179341, "grad_norm": 0.49254704013547357, "learning_rate": 9.81605361690957e-05, "loss": 3.2094, "step": 8600 }, { "epoch": 0.533925135017692, "grad_norm": 0.438904856825261, "learning_rate": 9.815956545615149e-05, "loss": 3.1861, "step": 8601 }, { "epoch": 0.5339872121174499, "grad_norm": 0.5292419742954159, "learning_rate": 9.815859449194712e-05, "loss": 3.1917, "step": 8602 }, { "epoch": 0.5340492892172077, "grad_norm": 0.5025281173279155, "learning_rate": 9.815762327648767e-05, "loss": 3.2852, "step": 8603 }, { "epoch": 0.5341113663169657, "grad_norm": 0.45948332592760777, "learning_rate": 9.815665180977823e-05, "loss": 3.2676, "step": 8604 }, { "epoch": 0.5341734434167236, "grad_norm": 0.5009907502959464, "learning_rate": 9.815568009182384e-05, "loss": 3.2206, "step": 8605 }, { "epoch": 0.5342355205164815, "grad_norm": 0.5231617158335701, "learning_rate": 9.815470812262957e-05, "loss": 3.2972, "step": 8606 }, { "epoch": 0.5342975976162394, "grad_norm": 0.4465582638773866, "learning_rate": 9.81537359022005e-05, "loss": 3.2571, "step": 8607 }, { "epoch": 0.5343596747159973, "grad_norm": 0.46166010966734256, "learning_rate": 9.815276343054169e-05, "loss": 3.2613, "step": 8608 }, { "epoch": 0.5344217518157551, "grad_norm": 0.4941761471491348, "learning_rate": 9.815179070765825e-05, "loss": 3.1473, "step": 8609 }, { "epoch": 0.534483828915513, "grad_norm": 0.5129160959698714, "learning_rate": 9.81508177335552e-05, "loss": 3.2904, "step": 8610 }, { "epoch": 0.534545906015271, "grad_norm": 0.5312623976214536, "learning_rate": 9.814984450823766e-05, "loss": 3.2855, "step": 8611 }, { "epoch": 0.5346079831150289, "grad_norm": 0.5567302698597113, "learning_rate": 9.814887103171069e-05, "loss": 3.269, "step": 8612 }, { "epoch": 0.5346700602147868, "grad_norm": 0.5342359495783241, "learning_rate": 9.814789730397938e-05, "loss": 3.2601, "step": 8613 }, { "epoch": 0.5347321373145447, "grad_norm": 0.4940014217149686, "learning_rate": 9.814692332504877e-05, "loss": 3.321, "step": 8614 }, { "epoch": 0.5347942144143025, "grad_norm": 0.4822196073096192, "learning_rate": 9.8145949094924e-05, "loss": 3.2891, "step": 8615 }, { "epoch": 0.5348562915140604, "grad_norm": 0.5507047884197983, "learning_rate": 9.814497461361013e-05, "loss": 3.218, "step": 8616 }, { "epoch": 0.5349183686138184, "grad_norm": 0.5003269437936384, "learning_rate": 9.814399988111223e-05, "loss": 3.2453, "step": 8617 }, { "epoch": 0.5349804457135763, "grad_norm": 0.49540635793091947, "learning_rate": 9.814302489743539e-05, "loss": 3.2752, "step": 8618 }, { "epoch": 0.5350425228133342, "grad_norm": 0.5744644988589102, "learning_rate": 9.814204966258471e-05, "loss": 3.2821, "step": 8619 }, { "epoch": 0.535104599913092, "grad_norm": 0.5692206081095129, "learning_rate": 9.814107417656528e-05, "loss": 3.2885, "step": 8620 }, { "epoch": 0.5351666770128499, "grad_norm": 0.5074314448369704, "learning_rate": 9.814009843938217e-05, "loss": 3.2524, "step": 8621 }, { "epoch": 0.5352287541126078, "grad_norm": 0.45479792863056234, "learning_rate": 9.813912245104047e-05, "loss": 3.1419, "step": 8622 }, { "epoch": 0.5352908312123658, "grad_norm": 0.5083691407187975, "learning_rate": 9.813814621154529e-05, "loss": 3.2576, "step": 8623 }, { "epoch": 0.5353529083121237, "grad_norm": 0.5180018004492912, "learning_rate": 9.81371697209017e-05, "loss": 3.1694, "step": 8624 }, { "epoch": 0.5354149854118816, "grad_norm": 0.4920193824051584, "learning_rate": 9.813619297911482e-05, "loss": 3.2311, "step": 8625 }, { "epoch": 0.5354770625116394, "grad_norm": 0.486924521972348, "learning_rate": 9.813521598618976e-05, "loss": 3.2385, "step": 8626 }, { "epoch": 0.5355391396113973, "grad_norm": 0.4981778290565707, "learning_rate": 9.813423874213157e-05, "loss": 3.2663, "step": 8627 }, { "epoch": 0.5356012167111552, "grad_norm": 0.47436484171428517, "learning_rate": 9.813326124694537e-05, "loss": 3.2546, "step": 8628 }, { "epoch": 0.5356632938109132, "grad_norm": 0.5345085289527132, "learning_rate": 9.813228350063627e-05, "loss": 3.2634, "step": 8629 }, { "epoch": 0.5357253709106711, "grad_norm": 0.48726276224997855, "learning_rate": 9.813130550320936e-05, "loss": 3.282, "step": 8630 }, { "epoch": 0.535787448010429, "grad_norm": 0.4508976288614472, "learning_rate": 9.813032725466974e-05, "loss": 3.1471, "step": 8631 }, { "epoch": 0.5358495251101868, "grad_norm": 0.5514378522992643, "learning_rate": 9.812934875502252e-05, "loss": 3.2119, "step": 8632 }, { "epoch": 0.5359116022099447, "grad_norm": 0.4879574426812201, "learning_rate": 9.81283700042728e-05, "loss": 3.26, "step": 8633 }, { "epoch": 0.5359736793097026, "grad_norm": 0.5060920622026943, "learning_rate": 9.81273910024257e-05, "loss": 3.2398, "step": 8634 }, { "epoch": 0.5360357564094606, "grad_norm": 0.45919877584187063, "learning_rate": 9.812641174948632e-05, "loss": 3.2445, "step": 8635 }, { "epoch": 0.5360978335092185, "grad_norm": 0.5144756396210651, "learning_rate": 9.812543224545976e-05, "loss": 3.1394, "step": 8636 }, { "epoch": 0.5361599106089764, "grad_norm": 0.4813547107181915, "learning_rate": 9.812445249035113e-05, "loss": 3.318, "step": 8637 }, { "epoch": 0.5362219877087342, "grad_norm": 0.5200397899424599, "learning_rate": 9.812347248416556e-05, "loss": 3.2657, "step": 8638 }, { "epoch": 0.5362840648084921, "grad_norm": 0.5006790616280671, "learning_rate": 9.812249222690815e-05, "loss": 3.2631, "step": 8639 }, { "epoch": 0.53634614190825, "grad_norm": 0.492125325590166, "learning_rate": 9.812151171858402e-05, "loss": 3.2792, "step": 8640 }, { "epoch": 0.536408219008008, "grad_norm": 0.48614466320534844, "learning_rate": 9.812053095919827e-05, "loss": 3.1883, "step": 8641 }, { "epoch": 0.5364702961077659, "grad_norm": 0.43981247514619226, "learning_rate": 9.811954994875602e-05, "loss": 3.172, "step": 8642 }, { "epoch": 0.5365323732075238, "grad_norm": 0.49531754271530565, "learning_rate": 9.811856868726243e-05, "loss": 3.3002, "step": 8643 }, { "epoch": 0.5365944503072816, "grad_norm": 0.5475029473682657, "learning_rate": 9.811758717472256e-05, "loss": 3.2652, "step": 8644 }, { "epoch": 0.5366565274070395, "grad_norm": 0.5208780142374435, "learning_rate": 9.811660541114155e-05, "loss": 3.2561, "step": 8645 }, { "epoch": 0.5367186045067974, "grad_norm": 0.442353961332195, "learning_rate": 9.811562339652457e-05, "loss": 3.3459, "step": 8646 }, { "epoch": 0.5367806816065553, "grad_norm": 0.5327898317140712, "learning_rate": 9.811464113087666e-05, "loss": 3.323, "step": 8647 }, { "epoch": 0.5368427587063133, "grad_norm": 0.5850797860438693, "learning_rate": 9.811365861420301e-05, "loss": 3.1761, "step": 8648 }, { "epoch": 0.5369048358060712, "grad_norm": 0.5118350413728812, "learning_rate": 9.81126758465087e-05, "loss": 3.2458, "step": 8649 }, { "epoch": 0.536966912905829, "grad_norm": 0.49148562507427973, "learning_rate": 9.811169282779891e-05, "loss": 3.2277, "step": 8650 }, { "epoch": 0.5370289900055869, "grad_norm": 0.5255042311355597, "learning_rate": 9.811070955807871e-05, "loss": 3.2661, "step": 8651 }, { "epoch": 0.5370910671053448, "grad_norm": 0.4682517900896046, "learning_rate": 9.810972603735326e-05, "loss": 3.1866, "step": 8652 }, { "epoch": 0.5371531442051027, "grad_norm": 0.5692642454677652, "learning_rate": 9.810874226562772e-05, "loss": 3.3107, "step": 8653 }, { "epoch": 0.5372152213048607, "grad_norm": 0.643046199912703, "learning_rate": 9.810775824290717e-05, "loss": 3.2861, "step": 8654 }, { "epoch": 0.5372772984046186, "grad_norm": 0.6611216106057444, "learning_rate": 9.810677396919677e-05, "loss": 3.2455, "step": 8655 }, { "epoch": 0.5373393755043764, "grad_norm": 0.6625276217910299, "learning_rate": 9.810578944450166e-05, "loss": 3.2475, "step": 8656 }, { "epoch": 0.5374014526041343, "grad_norm": 0.7258547956079773, "learning_rate": 9.810480466882697e-05, "loss": 3.1899, "step": 8657 }, { "epoch": 0.5374635297038922, "grad_norm": 0.6650186723224439, "learning_rate": 9.810381964217783e-05, "loss": 3.2296, "step": 8658 }, { "epoch": 0.5375256068036501, "grad_norm": 0.6095021922077564, "learning_rate": 9.810283436455937e-05, "loss": 3.2172, "step": 8659 }, { "epoch": 0.537587683903408, "grad_norm": 0.6678186660717567, "learning_rate": 9.810184883597676e-05, "loss": 3.1553, "step": 8660 }, { "epoch": 0.537649761003166, "grad_norm": 0.6656537009485388, "learning_rate": 9.810086305643513e-05, "loss": 3.1579, "step": 8661 }, { "epoch": 0.5377118381029238, "grad_norm": 0.6343510139327937, "learning_rate": 9.809987702593961e-05, "loss": 3.2186, "step": 8662 }, { "epoch": 0.5377739152026817, "grad_norm": 0.5768819539073594, "learning_rate": 9.809889074449537e-05, "loss": 3.1347, "step": 8663 }, { "epoch": 0.5378359923024396, "grad_norm": 0.5383714309690995, "learning_rate": 9.809790421210754e-05, "loss": 3.2336, "step": 8664 }, { "epoch": 0.5378980694021975, "grad_norm": 0.6055268536652357, "learning_rate": 9.809691742878125e-05, "loss": 3.2029, "step": 8665 }, { "epoch": 0.5379601465019554, "grad_norm": 0.5789510135615736, "learning_rate": 9.809593039452167e-05, "loss": 3.1652, "step": 8666 }, { "epoch": 0.5380222236017134, "grad_norm": 0.5528227125553952, "learning_rate": 9.809494310933395e-05, "loss": 3.236, "step": 8667 }, { "epoch": 0.5380843007014712, "grad_norm": 0.516311217781731, "learning_rate": 9.809395557322325e-05, "loss": 3.2532, "step": 8668 }, { "epoch": 0.5381463778012291, "grad_norm": 0.5197524553491414, "learning_rate": 9.80929677861947e-05, "loss": 3.2315, "step": 8669 }, { "epoch": 0.538208454900987, "grad_norm": 0.5755880986831007, "learning_rate": 9.809197974825345e-05, "loss": 3.2287, "step": 8670 }, { "epoch": 0.5382705320007449, "grad_norm": 0.5452093371113725, "learning_rate": 9.809099145940468e-05, "loss": 3.2809, "step": 8671 }, { "epoch": 0.5383326091005028, "grad_norm": 0.4905633620401469, "learning_rate": 9.809000291965354e-05, "loss": 3.2386, "step": 8672 }, { "epoch": 0.5383946862002608, "grad_norm": 0.554085364389027, "learning_rate": 9.808901412900515e-05, "loss": 3.2599, "step": 8673 }, { "epoch": 0.5384567633000186, "grad_norm": 0.4968994536470345, "learning_rate": 9.808802508746473e-05, "loss": 3.2977, "step": 8674 }, { "epoch": 0.5385188403997765, "grad_norm": 0.46890450156628544, "learning_rate": 9.80870357950374e-05, "loss": 3.2455, "step": 8675 }, { "epoch": 0.5385809174995344, "grad_norm": 0.6199375883729804, "learning_rate": 9.808604625172833e-05, "loss": 3.3101, "step": 8676 }, { "epoch": 0.5386429945992923, "grad_norm": 0.5345006114717131, "learning_rate": 9.808505645754268e-05, "loss": 3.1693, "step": 8677 }, { "epoch": 0.5387050716990502, "grad_norm": 0.5304122096901044, "learning_rate": 9.808406641248562e-05, "loss": 3.2102, "step": 8678 }, { "epoch": 0.5387671487988082, "grad_norm": 0.5050698972239863, "learning_rate": 9.808307611656232e-05, "loss": 3.0798, "step": 8679 }, { "epoch": 0.538829225898566, "grad_norm": 0.48804997879929485, "learning_rate": 9.808208556977792e-05, "loss": 3.1941, "step": 8680 }, { "epoch": 0.5388913029983239, "grad_norm": 0.5119351810202075, "learning_rate": 9.808109477213763e-05, "loss": 3.1766, "step": 8681 }, { "epoch": 0.5389533800980818, "grad_norm": 0.4250236210149255, "learning_rate": 9.808010372364657e-05, "loss": 3.2166, "step": 8682 }, { "epoch": 0.5390154571978397, "grad_norm": 0.4877263081836867, "learning_rate": 9.807911242430995e-05, "loss": 3.2794, "step": 8683 }, { "epoch": 0.5390775342975976, "grad_norm": 0.5214457979827982, "learning_rate": 9.807812087413293e-05, "loss": 3.2283, "step": 8684 }, { "epoch": 0.5391396113973556, "grad_norm": 0.49982058763275095, "learning_rate": 9.80771290731207e-05, "loss": 3.1512, "step": 8685 }, { "epoch": 0.5392016884971134, "grad_norm": 0.5363924547750003, "learning_rate": 9.807613702127838e-05, "loss": 3.2675, "step": 8686 }, { "epoch": 0.5392637655968713, "grad_norm": 0.544872350048772, "learning_rate": 9.807514471861121e-05, "loss": 3.2333, "step": 8687 }, { "epoch": 0.5393258426966292, "grad_norm": 0.545306736388257, "learning_rate": 9.807415216512433e-05, "loss": 3.1961, "step": 8688 }, { "epoch": 0.5393879197963871, "grad_norm": 0.46118336925111947, "learning_rate": 9.807315936082293e-05, "loss": 3.2722, "step": 8689 }, { "epoch": 0.539449996896145, "grad_norm": 0.5194198947971221, "learning_rate": 9.807216630571218e-05, "loss": 3.2079, "step": 8690 }, { "epoch": 0.539512073995903, "grad_norm": 0.47557249578949534, "learning_rate": 9.807117299979728e-05, "loss": 3.207, "step": 8691 }, { "epoch": 0.5395741510956608, "grad_norm": 0.46454394313229264, "learning_rate": 9.807017944308341e-05, "loss": 3.278, "step": 8692 }, { "epoch": 0.5396362281954187, "grad_norm": 0.48287422193049295, "learning_rate": 9.806918563557572e-05, "loss": 3.3405, "step": 8693 }, { "epoch": 0.5396983052951766, "grad_norm": 0.5593761220313386, "learning_rate": 9.806819157727943e-05, "loss": 3.282, "step": 8694 }, { "epoch": 0.5397603823949345, "grad_norm": 0.5690717133096187, "learning_rate": 9.806719726819972e-05, "loss": 3.2967, "step": 8695 }, { "epoch": 0.5398224594946924, "grad_norm": 0.5213920997389112, "learning_rate": 9.806620270834178e-05, "loss": 3.2754, "step": 8696 }, { "epoch": 0.5398845365944503, "grad_norm": 0.4809662174180235, "learning_rate": 9.806520789771077e-05, "loss": 3.1495, "step": 8697 }, { "epoch": 0.5399466136942082, "grad_norm": 0.5128168452495472, "learning_rate": 9.80642128363119e-05, "loss": 3.2464, "step": 8698 }, { "epoch": 0.5400086907939661, "grad_norm": 0.543796786126085, "learning_rate": 9.806321752415038e-05, "loss": 3.2721, "step": 8699 }, { "epoch": 0.540070767893724, "grad_norm": 0.5058731312888101, "learning_rate": 9.806222196123139e-05, "loss": 3.0827, "step": 8700 }, { "epoch": 0.5401328449934819, "grad_norm": 0.49545311300720474, "learning_rate": 9.80612261475601e-05, "loss": 3.1219, "step": 8701 }, { "epoch": 0.5401949220932398, "grad_norm": 0.5956018580677366, "learning_rate": 9.806023008314175e-05, "loss": 3.2614, "step": 8702 }, { "epoch": 0.5402569991929977, "grad_norm": 0.5715018634114962, "learning_rate": 9.80592337679815e-05, "loss": 3.1967, "step": 8703 }, { "epoch": 0.5403190762927556, "grad_norm": 0.43985568787549567, "learning_rate": 9.805823720208456e-05, "loss": 3.2881, "step": 8704 }, { "epoch": 0.5403811533925135, "grad_norm": 0.5441662050520772, "learning_rate": 9.805724038545614e-05, "loss": 3.2751, "step": 8705 }, { "epoch": 0.5404432304922714, "grad_norm": 0.5375937757083492, "learning_rate": 9.805624331810141e-05, "loss": 3.2558, "step": 8706 }, { "epoch": 0.5405053075920293, "grad_norm": 0.5156541584902513, "learning_rate": 9.805524600002562e-05, "loss": 3.3017, "step": 8707 }, { "epoch": 0.5405673846917872, "grad_norm": 0.5105680967676302, "learning_rate": 9.805424843123392e-05, "loss": 3.1982, "step": 8708 }, { "epoch": 0.5406294617915451, "grad_norm": 0.4895945168969631, "learning_rate": 9.805325061173157e-05, "loss": 3.2761, "step": 8709 }, { "epoch": 0.540691538891303, "grad_norm": 0.5085967244004146, "learning_rate": 9.805225254152372e-05, "loss": 3.1542, "step": 8710 }, { "epoch": 0.5407536159910609, "grad_norm": 0.5177170764893515, "learning_rate": 9.80512542206156e-05, "loss": 3.2316, "step": 8711 }, { "epoch": 0.5408156930908188, "grad_norm": 0.46787920176265024, "learning_rate": 9.805025564901244e-05, "loss": 3.1927, "step": 8712 }, { "epoch": 0.5408777701905767, "grad_norm": 0.4331614598978575, "learning_rate": 9.804925682671942e-05, "loss": 3.188, "step": 8713 }, { "epoch": 0.5409398472903346, "grad_norm": 0.4681139286142817, "learning_rate": 9.804825775374177e-05, "loss": 3.2019, "step": 8714 }, { "epoch": 0.5410019243900925, "grad_norm": 0.5205562204034189, "learning_rate": 9.804725843008468e-05, "loss": 3.1029, "step": 8715 }, { "epoch": 0.5410640014898503, "grad_norm": 0.5063728171767414, "learning_rate": 9.80462588557534e-05, "loss": 3.1945, "step": 8716 }, { "epoch": 0.5411260785896083, "grad_norm": 0.46724092439600706, "learning_rate": 9.804525903075312e-05, "loss": 3.3097, "step": 8717 }, { "epoch": 0.5411881556893662, "grad_norm": 0.44847848684966113, "learning_rate": 9.804425895508905e-05, "loss": 3.1851, "step": 8718 }, { "epoch": 0.5412502327891241, "grad_norm": 0.5950474116324938, "learning_rate": 9.804325862876642e-05, "loss": 3.2617, "step": 8719 }, { "epoch": 0.541312309888882, "grad_norm": 0.5141943823280883, "learning_rate": 9.804225805179046e-05, "loss": 3.2609, "step": 8720 }, { "epoch": 0.5413743869886399, "grad_norm": 0.5325786580836709, "learning_rate": 9.804125722416637e-05, "loss": 3.2642, "step": 8721 }, { "epoch": 0.5414364640883977, "grad_norm": 0.5436846726735737, "learning_rate": 9.804025614589938e-05, "loss": 3.2197, "step": 8722 }, { "epoch": 0.5414985411881557, "grad_norm": 0.5413191181655215, "learning_rate": 9.80392548169947e-05, "loss": 3.2273, "step": 8723 }, { "epoch": 0.5415606182879136, "grad_norm": 0.4787468954687828, "learning_rate": 9.803825323745757e-05, "loss": 3.2126, "step": 8724 }, { "epoch": 0.5416226953876715, "grad_norm": 0.4381612820914154, "learning_rate": 9.803725140729321e-05, "loss": 3.2416, "step": 8725 }, { "epoch": 0.5416847724874294, "grad_norm": 0.49243793474069986, "learning_rate": 9.803624932650686e-05, "loss": 3.2927, "step": 8726 }, { "epoch": 0.5417468495871873, "grad_norm": 0.4851904434710387, "learning_rate": 9.803524699510374e-05, "loss": 3.1781, "step": 8727 }, { "epoch": 0.5418089266869451, "grad_norm": 0.49254226009478475, "learning_rate": 9.803424441308906e-05, "loss": 3.2822, "step": 8728 }, { "epoch": 0.541871003786703, "grad_norm": 0.3849929486595046, "learning_rate": 9.803324158046806e-05, "loss": 3.2776, "step": 8729 }, { "epoch": 0.541933080886461, "grad_norm": 0.4378826987972465, "learning_rate": 9.803223849724599e-05, "loss": 3.2219, "step": 8730 }, { "epoch": 0.5419951579862189, "grad_norm": 0.4950644109462963, "learning_rate": 9.803123516342807e-05, "loss": 3.2967, "step": 8731 }, { "epoch": 0.5420572350859768, "grad_norm": 0.49514907342989134, "learning_rate": 9.803023157901953e-05, "loss": 3.1672, "step": 8732 }, { "epoch": 0.5421193121857347, "grad_norm": 0.4805797975810394, "learning_rate": 9.802922774402563e-05, "loss": 3.2637, "step": 8733 }, { "epoch": 0.5421813892854925, "grad_norm": 0.5232317505892361, "learning_rate": 9.802822365845157e-05, "loss": 3.2493, "step": 8734 }, { "epoch": 0.5422434663852504, "grad_norm": 0.49032947484155154, "learning_rate": 9.802721932230262e-05, "loss": 3.3023, "step": 8735 }, { "epoch": 0.5423055434850084, "grad_norm": 0.4547581018121479, "learning_rate": 9.8026214735584e-05, "loss": 3.3268, "step": 8736 }, { "epoch": 0.5423676205847663, "grad_norm": 0.5166050238394202, "learning_rate": 9.802520989830097e-05, "loss": 3.2581, "step": 8737 }, { "epoch": 0.5424296976845242, "grad_norm": 0.5040145580198737, "learning_rate": 9.802420481045875e-05, "loss": 3.1398, "step": 8738 }, { "epoch": 0.5424917747842821, "grad_norm": 0.4239659590664064, "learning_rate": 9.80231994720626e-05, "loss": 3.2642, "step": 8739 }, { "epoch": 0.5425538518840399, "grad_norm": 0.39351392586943135, "learning_rate": 9.802219388311775e-05, "loss": 3.1911, "step": 8740 }, { "epoch": 0.5426159289837978, "grad_norm": 0.5318638676913913, "learning_rate": 9.802118804362947e-05, "loss": 3.2554, "step": 8741 }, { "epoch": 0.5426780060835558, "grad_norm": 0.5137575317443303, "learning_rate": 9.802018195360298e-05, "loss": 3.2371, "step": 8742 }, { "epoch": 0.5427400831833137, "grad_norm": 0.44977009980250293, "learning_rate": 9.801917561304357e-05, "loss": 3.2543, "step": 8743 }, { "epoch": 0.5428021602830716, "grad_norm": 0.6243566301104534, "learning_rate": 9.801816902195645e-05, "loss": 3.1928, "step": 8744 }, { "epoch": 0.5428642373828295, "grad_norm": 0.5412526627050013, "learning_rate": 9.801716218034688e-05, "loss": 3.2375, "step": 8745 }, { "epoch": 0.5429263144825873, "grad_norm": 0.4985564971749909, "learning_rate": 9.801615508822012e-05, "loss": 3.197, "step": 8746 }, { "epoch": 0.5429883915823452, "grad_norm": 0.5423772320513526, "learning_rate": 9.801514774558142e-05, "loss": 3.2922, "step": 8747 }, { "epoch": 0.5430504686821032, "grad_norm": 0.5920420978968844, "learning_rate": 9.801414015243604e-05, "loss": 3.2546, "step": 8748 }, { "epoch": 0.5431125457818611, "grad_norm": 0.5070091936342546, "learning_rate": 9.801313230878923e-05, "loss": 3.1861, "step": 8749 }, { "epoch": 0.543174622881619, "grad_norm": 0.4981363403149443, "learning_rate": 9.801212421464627e-05, "loss": 3.2051, "step": 8750 }, { "epoch": 0.5432366999813769, "grad_norm": 0.4921073091363368, "learning_rate": 9.801111587001239e-05, "loss": 3.1913, "step": 8751 }, { "epoch": 0.5432987770811347, "grad_norm": 0.43155484620198437, "learning_rate": 9.801010727489287e-05, "loss": 3.2826, "step": 8752 }, { "epoch": 0.5433608541808926, "grad_norm": 0.5769909742333681, "learning_rate": 9.800909842929296e-05, "loss": 3.1865, "step": 8753 }, { "epoch": 0.5434229312806506, "grad_norm": 0.6520115126239897, "learning_rate": 9.800808933321792e-05, "loss": 3.2489, "step": 8754 }, { "epoch": 0.5434850083804085, "grad_norm": 0.5963218852361019, "learning_rate": 9.800707998667305e-05, "loss": 3.277, "step": 8755 }, { "epoch": 0.5435470854801664, "grad_norm": 0.4960029398788726, "learning_rate": 9.800607038966356e-05, "loss": 3.2705, "step": 8756 }, { "epoch": 0.5436091625799243, "grad_norm": 0.527523462261889, "learning_rate": 9.800506054219476e-05, "loss": 3.2093, "step": 8757 }, { "epoch": 0.5436712396796821, "grad_norm": 0.5935857947314269, "learning_rate": 9.800405044427191e-05, "loss": 3.2176, "step": 8758 }, { "epoch": 0.54373331677944, "grad_norm": 0.6821479767391132, "learning_rate": 9.800304009590028e-05, "loss": 3.187, "step": 8759 }, { "epoch": 0.543795393879198, "grad_norm": 0.6210252016255063, "learning_rate": 9.800202949708513e-05, "loss": 3.2217, "step": 8760 }, { "epoch": 0.5438574709789559, "grad_norm": 0.5835983122988163, "learning_rate": 9.800101864783173e-05, "loss": 3.28, "step": 8761 }, { "epoch": 0.5439195480787138, "grad_norm": 0.5236804455296677, "learning_rate": 9.800000754814537e-05, "loss": 3.1821, "step": 8762 }, { "epoch": 0.5439816251784717, "grad_norm": 0.5805361958849679, "learning_rate": 9.799899619803133e-05, "loss": 3.3019, "step": 8763 }, { "epoch": 0.5440437022782295, "grad_norm": 0.5310810886455134, "learning_rate": 9.799798459749485e-05, "loss": 3.1443, "step": 8764 }, { "epoch": 0.5441057793779874, "grad_norm": 0.5019674929926252, "learning_rate": 9.799697274654125e-05, "loss": 3.1136, "step": 8765 }, { "epoch": 0.5441678564777453, "grad_norm": 0.48920679492652774, "learning_rate": 9.799596064517578e-05, "loss": 3.1744, "step": 8766 }, { "epoch": 0.5442299335775033, "grad_norm": 0.4918702480614961, "learning_rate": 9.799494829340375e-05, "loss": 3.218, "step": 8767 }, { "epoch": 0.5442920106772612, "grad_norm": 0.4748199843308523, "learning_rate": 9.799393569123041e-05, "loss": 3.2005, "step": 8768 }, { "epoch": 0.5443540877770191, "grad_norm": 0.5279288260523083, "learning_rate": 9.799292283866106e-05, "loss": 3.2774, "step": 8769 }, { "epoch": 0.5444161648767769, "grad_norm": 0.5031121969074873, "learning_rate": 9.799190973570099e-05, "loss": 3.1918, "step": 8770 }, { "epoch": 0.5444782419765348, "grad_norm": 0.5145502159591931, "learning_rate": 9.799089638235546e-05, "loss": 3.2381, "step": 8771 }, { "epoch": 0.5445403190762927, "grad_norm": 0.5148076891390087, "learning_rate": 9.798988277862978e-05, "loss": 3.367, "step": 8772 }, { "epoch": 0.5446023961760507, "grad_norm": 0.5484193521741194, "learning_rate": 9.798886892452924e-05, "loss": 3.1872, "step": 8773 }, { "epoch": 0.5446644732758086, "grad_norm": 0.5587027650107533, "learning_rate": 9.798785482005909e-05, "loss": 3.1778, "step": 8774 }, { "epoch": 0.5447265503755665, "grad_norm": 0.5388763493322134, "learning_rate": 9.798684046522469e-05, "loss": 3.2894, "step": 8775 }, { "epoch": 0.5447886274753243, "grad_norm": 0.5230841511554938, "learning_rate": 9.798582586003127e-05, "loss": 3.1833, "step": 8776 }, { "epoch": 0.5448507045750822, "grad_norm": 0.44240918617705005, "learning_rate": 9.798481100448415e-05, "loss": 3.2524, "step": 8777 }, { "epoch": 0.5449127816748401, "grad_norm": 0.5309044050440256, "learning_rate": 9.798379589858864e-05, "loss": 3.1604, "step": 8778 }, { "epoch": 0.5449748587745981, "grad_norm": 0.5239973063854698, "learning_rate": 9.798278054234999e-05, "loss": 3.2686, "step": 8779 }, { "epoch": 0.545036935874356, "grad_norm": 0.4848706416774772, "learning_rate": 9.798176493577356e-05, "loss": 3.1382, "step": 8780 }, { "epoch": 0.5450990129741139, "grad_norm": 0.6035451295424344, "learning_rate": 9.798074907886459e-05, "loss": 3.2051, "step": 8781 }, { "epoch": 0.5451610900738717, "grad_norm": 0.6786468695171308, "learning_rate": 9.797973297162841e-05, "loss": 3.2912, "step": 8782 }, { "epoch": 0.5452231671736296, "grad_norm": 0.5733724832706176, "learning_rate": 9.797871661407032e-05, "loss": 3.1772, "step": 8783 }, { "epoch": 0.5452852442733875, "grad_norm": 0.6304947120965083, "learning_rate": 9.797770000619561e-05, "loss": 3.2365, "step": 8784 }, { "epoch": 0.5453473213731455, "grad_norm": 0.6323395362677791, "learning_rate": 9.797668314800961e-05, "loss": 3.3, "step": 8785 }, { "epoch": 0.5454093984729034, "grad_norm": 0.5852245571060001, "learning_rate": 9.797566603951759e-05, "loss": 3.2122, "step": 8786 }, { "epoch": 0.5454714755726613, "grad_norm": 0.6048656043615606, "learning_rate": 9.797464868072488e-05, "loss": 3.1737, "step": 8787 }, { "epoch": 0.5455335526724191, "grad_norm": 0.5074125622371208, "learning_rate": 9.797363107163678e-05, "loss": 3.2106, "step": 8788 }, { "epoch": 0.545595629772177, "grad_norm": 0.5456830249708181, "learning_rate": 9.797261321225859e-05, "loss": 3.1122, "step": 8789 }, { "epoch": 0.5456577068719349, "grad_norm": 0.5000997682216559, "learning_rate": 9.797159510259565e-05, "loss": 3.2754, "step": 8790 }, { "epoch": 0.5457197839716929, "grad_norm": 0.5045952337622023, "learning_rate": 9.797057674265324e-05, "loss": 3.2156, "step": 8791 }, { "epoch": 0.5457818610714508, "grad_norm": 0.4903227129979402, "learning_rate": 9.79695581324367e-05, "loss": 3.2091, "step": 8792 }, { "epoch": 0.5458439381712087, "grad_norm": 0.5347420697705388, "learning_rate": 9.796853927195131e-05, "loss": 3.287, "step": 8793 }, { "epoch": 0.5459060152709665, "grad_norm": 0.5228155341006182, "learning_rate": 9.796752016120242e-05, "loss": 3.2888, "step": 8794 }, { "epoch": 0.5459680923707244, "grad_norm": 0.446116892119161, "learning_rate": 9.796650080019533e-05, "loss": 3.1313, "step": 8795 }, { "epoch": 0.5460301694704823, "grad_norm": 0.5271590863198521, "learning_rate": 9.796548118893536e-05, "loss": 3.1946, "step": 8796 }, { "epoch": 0.5460922465702402, "grad_norm": 0.4818913094349439, "learning_rate": 9.796446132742782e-05, "loss": 3.1597, "step": 8797 }, { "epoch": 0.5461543236699982, "grad_norm": 0.4572594390873058, "learning_rate": 9.796344121567806e-05, "loss": 3.3404, "step": 8798 }, { "epoch": 0.5462164007697561, "grad_norm": 0.5135146506001526, "learning_rate": 9.796242085369137e-05, "loss": 3.2185, "step": 8799 }, { "epoch": 0.5462784778695139, "grad_norm": 0.4777414871896636, "learning_rate": 9.79614002414731e-05, "loss": 3.1496, "step": 8800 }, { "epoch": 0.5463405549692718, "grad_norm": 0.4709184715917305, "learning_rate": 9.796037937902855e-05, "loss": 3.2702, "step": 8801 }, { "epoch": 0.5464026320690297, "grad_norm": 0.4561906403635864, "learning_rate": 9.795935826636307e-05, "loss": 3.1626, "step": 8802 }, { "epoch": 0.5464647091687876, "grad_norm": 0.4572961316799499, "learning_rate": 9.795833690348196e-05, "loss": 3.26, "step": 8803 }, { "epoch": 0.5465267862685456, "grad_norm": 0.5248357686250471, "learning_rate": 9.795731529039058e-05, "loss": 3.2183, "step": 8804 }, { "epoch": 0.5465888633683035, "grad_norm": 0.51440965589023, "learning_rate": 9.795629342709422e-05, "loss": 3.1608, "step": 8805 }, { "epoch": 0.5466509404680613, "grad_norm": 0.37529548288813985, "learning_rate": 9.795527131359826e-05, "loss": 3.2199, "step": 8806 }, { "epoch": 0.5467130175678192, "grad_norm": 0.521280227306125, "learning_rate": 9.7954248949908e-05, "loss": 3.1308, "step": 8807 }, { "epoch": 0.5467750946675771, "grad_norm": 0.4805883708384682, "learning_rate": 9.795322633602879e-05, "loss": 3.2597, "step": 8808 }, { "epoch": 0.546837171767335, "grad_norm": 0.49473861276052356, "learning_rate": 9.795220347196595e-05, "loss": 3.2559, "step": 8809 }, { "epoch": 0.546899248867093, "grad_norm": 0.5007540560047206, "learning_rate": 9.795118035772482e-05, "loss": 3.2533, "step": 8810 }, { "epoch": 0.5469613259668509, "grad_norm": 0.4945537815113717, "learning_rate": 9.795015699331074e-05, "loss": 3.2125, "step": 8811 }, { "epoch": 0.5470234030666087, "grad_norm": 0.4731699776277775, "learning_rate": 9.794913337872906e-05, "loss": 3.1584, "step": 8812 }, { "epoch": 0.5470854801663666, "grad_norm": 0.45740892830109064, "learning_rate": 9.79481095139851e-05, "loss": 3.2352, "step": 8813 }, { "epoch": 0.5471475572661245, "grad_norm": 0.5031572438208854, "learning_rate": 9.794708539908421e-05, "loss": 3.2878, "step": 8814 }, { "epoch": 0.5472096343658824, "grad_norm": 0.4591530185679997, "learning_rate": 9.794606103403175e-05, "loss": 3.296, "step": 8815 }, { "epoch": 0.5472717114656404, "grad_norm": 0.4803553811900471, "learning_rate": 9.794503641883305e-05, "loss": 3.2064, "step": 8816 }, { "epoch": 0.5473337885653983, "grad_norm": 0.493429478899633, "learning_rate": 9.794401155349346e-05, "loss": 3.268, "step": 8817 }, { "epoch": 0.5473958656651561, "grad_norm": 0.5861601342987056, "learning_rate": 9.79429864380183e-05, "loss": 3.1363, "step": 8818 }, { "epoch": 0.547457942764914, "grad_norm": 0.5217158914483729, "learning_rate": 9.794196107241295e-05, "loss": 3.2339, "step": 8819 }, { "epoch": 0.5475200198646719, "grad_norm": 0.5551047837515296, "learning_rate": 9.794093545668277e-05, "loss": 3.2383, "step": 8820 }, { "epoch": 0.5475820969644298, "grad_norm": 0.536094483497627, "learning_rate": 9.793990959083307e-05, "loss": 3.2113, "step": 8821 }, { "epoch": 0.5476441740641877, "grad_norm": 0.5576802500324403, "learning_rate": 9.793888347486924e-05, "loss": 3.1611, "step": 8822 }, { "epoch": 0.5477062511639457, "grad_norm": 0.5817327566046213, "learning_rate": 9.793785710879661e-05, "loss": 3.1689, "step": 8823 }, { "epoch": 0.5477683282637035, "grad_norm": 0.4747462918938752, "learning_rate": 9.793683049262053e-05, "loss": 3.2687, "step": 8824 }, { "epoch": 0.5478304053634614, "grad_norm": 0.5221019331421543, "learning_rate": 9.793580362634638e-05, "loss": 3.2143, "step": 8825 }, { "epoch": 0.5478924824632193, "grad_norm": 0.5108276645766957, "learning_rate": 9.793477650997951e-05, "loss": 3.2056, "step": 8826 }, { "epoch": 0.5479545595629772, "grad_norm": 0.498700763571426, "learning_rate": 9.793374914352526e-05, "loss": 3.2523, "step": 8827 }, { "epoch": 0.5480166366627351, "grad_norm": 0.49425333418852707, "learning_rate": 9.793272152698902e-05, "loss": 3.1289, "step": 8828 }, { "epoch": 0.5480787137624931, "grad_norm": 0.4770468767332377, "learning_rate": 9.793169366037614e-05, "loss": 3.2954, "step": 8829 }, { "epoch": 0.5481407908622509, "grad_norm": 0.45636627531527796, "learning_rate": 9.793066554369195e-05, "loss": 3.2807, "step": 8830 }, { "epoch": 0.5482028679620088, "grad_norm": 0.4545158383355405, "learning_rate": 9.792963717694186e-05, "loss": 3.1961, "step": 8831 }, { "epoch": 0.5482649450617667, "grad_norm": 0.4788922132018717, "learning_rate": 9.792860856013122e-05, "loss": 3.2754, "step": 8832 }, { "epoch": 0.5483270221615246, "grad_norm": 0.5220485878251486, "learning_rate": 9.79275796932654e-05, "loss": 3.17, "step": 8833 }, { "epoch": 0.5483890992612825, "grad_norm": 0.4700488216325727, "learning_rate": 9.792655057634974e-05, "loss": 3.2744, "step": 8834 }, { "epoch": 0.5484511763610405, "grad_norm": 0.48994011425435713, "learning_rate": 9.792552120938965e-05, "loss": 3.3153, "step": 8835 }, { "epoch": 0.5485132534607983, "grad_norm": 0.46837631271316804, "learning_rate": 9.792449159239047e-05, "loss": 3.1761, "step": 8836 }, { "epoch": 0.5485753305605562, "grad_norm": 0.45365327426780316, "learning_rate": 9.79234617253576e-05, "loss": 3.1683, "step": 8837 }, { "epoch": 0.5486374076603141, "grad_norm": 0.49035188702721705, "learning_rate": 9.792243160829639e-05, "loss": 3.2019, "step": 8838 }, { "epoch": 0.548699484760072, "grad_norm": 0.48832890087314196, "learning_rate": 9.792140124121221e-05, "loss": 3.2516, "step": 8839 }, { "epoch": 0.5487615618598299, "grad_norm": 0.49007568222548736, "learning_rate": 9.792037062411046e-05, "loss": 3.3, "step": 8840 }, { "epoch": 0.5488236389595879, "grad_norm": 0.46512090610854345, "learning_rate": 9.791933975699649e-05, "loss": 3.1672, "step": 8841 }, { "epoch": 0.5488857160593457, "grad_norm": 0.5321221485433422, "learning_rate": 9.79183086398757e-05, "loss": 3.1612, "step": 8842 }, { "epoch": 0.5489477931591036, "grad_norm": 0.47645672421613866, "learning_rate": 9.791727727275345e-05, "loss": 3.2599, "step": 8843 }, { "epoch": 0.5490098702588615, "grad_norm": 0.49114844226110527, "learning_rate": 9.791624565563515e-05, "loss": 3.3069, "step": 8844 }, { "epoch": 0.5490719473586194, "grad_norm": 0.4872420429335521, "learning_rate": 9.791521378852615e-05, "loss": 3.2274, "step": 8845 }, { "epoch": 0.5491340244583773, "grad_norm": 0.4854260700226414, "learning_rate": 9.791418167143184e-05, "loss": 3.2778, "step": 8846 }, { "epoch": 0.5491961015581353, "grad_norm": 0.49445611886226976, "learning_rate": 9.791314930435764e-05, "loss": 3.1673, "step": 8847 }, { "epoch": 0.5492581786578931, "grad_norm": 0.5201190692038166, "learning_rate": 9.791211668730888e-05, "loss": 3.2349, "step": 8848 }, { "epoch": 0.549320255757651, "grad_norm": 0.5364395959291621, "learning_rate": 9.7911083820291e-05, "loss": 3.2104, "step": 8849 }, { "epoch": 0.5493823328574089, "grad_norm": 0.5591233950309821, "learning_rate": 9.791005070330937e-05, "loss": 3.2284, "step": 8850 }, { "epoch": 0.5494444099571668, "grad_norm": 0.5901997735959934, "learning_rate": 9.790901733636935e-05, "loss": 3.2811, "step": 8851 }, { "epoch": 0.5495064870569247, "grad_norm": 0.5718680032187432, "learning_rate": 9.790798371947636e-05, "loss": 3.2319, "step": 8852 }, { "epoch": 0.5495685641566826, "grad_norm": 0.5925136795154852, "learning_rate": 9.79069498526358e-05, "loss": 3.3604, "step": 8853 }, { "epoch": 0.5496306412564405, "grad_norm": 0.5933332486456193, "learning_rate": 9.790591573585305e-05, "loss": 3.2822, "step": 8854 }, { "epoch": 0.5496927183561984, "grad_norm": 0.49362623256974786, "learning_rate": 9.790488136913351e-05, "loss": 3.2738, "step": 8855 }, { "epoch": 0.5497547954559563, "grad_norm": 0.5182045654387276, "learning_rate": 9.790384675248256e-05, "loss": 3.1859, "step": 8856 }, { "epoch": 0.5498168725557142, "grad_norm": 0.5062655729088528, "learning_rate": 9.790281188590563e-05, "loss": 3.1781, "step": 8857 }, { "epoch": 0.5498789496554721, "grad_norm": 0.5566945635445004, "learning_rate": 9.790177676940811e-05, "loss": 3.1766, "step": 8858 }, { "epoch": 0.54994102675523, "grad_norm": 0.5632304640902526, "learning_rate": 9.790074140299536e-05, "loss": 3.1287, "step": 8859 }, { "epoch": 0.5500031038549879, "grad_norm": 0.4869899153585648, "learning_rate": 9.789970578667283e-05, "loss": 3.2522, "step": 8860 }, { "epoch": 0.5500651809547458, "grad_norm": 0.5176400489471149, "learning_rate": 9.789866992044592e-05, "loss": 3.2241, "step": 8861 }, { "epoch": 0.5501272580545037, "grad_norm": 0.5332815650561229, "learning_rate": 9.789763380432e-05, "loss": 3.3087, "step": 8862 }, { "epoch": 0.5501893351542616, "grad_norm": 0.528593010604481, "learning_rate": 9.78965974383005e-05, "loss": 3.2379, "step": 8863 }, { "epoch": 0.5502514122540195, "grad_norm": 0.49929531925050363, "learning_rate": 9.789556082239284e-05, "loss": 3.0643, "step": 8864 }, { "epoch": 0.5503134893537774, "grad_norm": 0.4997339882139031, "learning_rate": 9.78945239566024e-05, "loss": 3.2105, "step": 8865 }, { "epoch": 0.5503755664535352, "grad_norm": 0.5450497822379579, "learning_rate": 9.78934868409346e-05, "loss": 3.2833, "step": 8866 }, { "epoch": 0.5504376435532932, "grad_norm": 0.5167191485186281, "learning_rate": 9.789244947539487e-05, "loss": 3.2903, "step": 8867 }, { "epoch": 0.5504997206530511, "grad_norm": 0.4743424281018884, "learning_rate": 9.789141185998858e-05, "loss": 3.1883, "step": 8868 }, { "epoch": 0.550561797752809, "grad_norm": 0.44291654814069326, "learning_rate": 9.789037399472119e-05, "loss": 3.1338, "step": 8869 }, { "epoch": 0.5506238748525669, "grad_norm": 0.5367726050486518, "learning_rate": 9.788933587959807e-05, "loss": 3.2294, "step": 8870 }, { "epoch": 0.5506859519523248, "grad_norm": 0.4960487225140322, "learning_rate": 9.788829751462467e-05, "loss": 3.2632, "step": 8871 }, { "epoch": 0.5507480290520826, "grad_norm": 0.5283350351929957, "learning_rate": 9.78872588998064e-05, "loss": 3.1899, "step": 8872 }, { "epoch": 0.5508101061518406, "grad_norm": 0.4875513037559988, "learning_rate": 9.788622003514866e-05, "loss": 3.2812, "step": 8873 }, { "epoch": 0.5508721832515985, "grad_norm": 0.47840070317624955, "learning_rate": 9.78851809206569e-05, "loss": 3.2084, "step": 8874 }, { "epoch": 0.5509342603513564, "grad_norm": 0.4427751902043704, "learning_rate": 9.788414155633654e-05, "loss": 3.1754, "step": 8875 }, { "epoch": 0.5509963374511143, "grad_norm": 0.5668379092463154, "learning_rate": 9.788310194219297e-05, "loss": 3.257, "step": 8876 }, { "epoch": 0.5510584145508722, "grad_norm": 0.6273367352883258, "learning_rate": 9.788206207823164e-05, "loss": 3.213, "step": 8877 }, { "epoch": 0.55112049165063, "grad_norm": 0.5847303289388603, "learning_rate": 9.788102196445797e-05, "loss": 3.2355, "step": 8878 }, { "epoch": 0.551182568750388, "grad_norm": 0.5118452308950574, "learning_rate": 9.787998160087737e-05, "loss": 3.2083, "step": 8879 }, { "epoch": 0.5512446458501459, "grad_norm": 0.481339789349689, "learning_rate": 9.787894098749531e-05, "loss": 3.2742, "step": 8880 }, { "epoch": 0.5513067229499038, "grad_norm": 0.5643654989581316, "learning_rate": 9.787790012431716e-05, "loss": 3.1469, "step": 8881 }, { "epoch": 0.5513688000496617, "grad_norm": 0.522097052657498, "learning_rate": 9.78768590113484e-05, "loss": 3.2894, "step": 8882 }, { "epoch": 0.5514308771494196, "grad_norm": 0.4587324871649494, "learning_rate": 9.787581764859444e-05, "loss": 3.2022, "step": 8883 }, { "epoch": 0.5514929542491774, "grad_norm": 0.5366780716807664, "learning_rate": 9.787477603606072e-05, "loss": 3.2811, "step": 8884 }, { "epoch": 0.5515550313489354, "grad_norm": 0.5736665211203946, "learning_rate": 9.787373417375268e-05, "loss": 3.2262, "step": 8885 }, { "epoch": 0.5516171084486933, "grad_norm": 0.4891492985978606, "learning_rate": 9.787269206167574e-05, "loss": 3.2009, "step": 8886 }, { "epoch": 0.5516791855484512, "grad_norm": 0.4718713469711275, "learning_rate": 9.787164969983534e-05, "loss": 3.2137, "step": 8887 }, { "epoch": 0.5517412626482091, "grad_norm": 0.48975196599054566, "learning_rate": 9.787060708823692e-05, "loss": 3.2189, "step": 8888 }, { "epoch": 0.551803339747967, "grad_norm": 0.44250754309545165, "learning_rate": 9.786956422688592e-05, "loss": 3.1985, "step": 8889 }, { "epoch": 0.5518654168477248, "grad_norm": 0.4703675177701603, "learning_rate": 9.786852111578779e-05, "loss": 3.138, "step": 8890 }, { "epoch": 0.5519274939474828, "grad_norm": 0.518122203529459, "learning_rate": 9.786747775494795e-05, "loss": 3.1549, "step": 8891 }, { "epoch": 0.5519895710472407, "grad_norm": 0.5589808210024024, "learning_rate": 9.786643414437189e-05, "loss": 3.3204, "step": 8892 }, { "epoch": 0.5520516481469986, "grad_norm": 0.5755109714221215, "learning_rate": 9.786539028406501e-05, "loss": 3.3373, "step": 8893 }, { "epoch": 0.5521137252467565, "grad_norm": 0.528792982496222, "learning_rate": 9.786434617403275e-05, "loss": 3.22, "step": 8894 }, { "epoch": 0.5521758023465144, "grad_norm": 0.4979095119962511, "learning_rate": 9.78633018142806e-05, "loss": 3.2251, "step": 8895 }, { "epoch": 0.5522378794462722, "grad_norm": 0.5263998555819479, "learning_rate": 9.786225720481397e-05, "loss": 3.2538, "step": 8896 }, { "epoch": 0.5522999565460301, "grad_norm": 0.5540136085277965, "learning_rate": 9.786121234563832e-05, "loss": 3.3299, "step": 8897 }, { "epoch": 0.5523620336457881, "grad_norm": 0.46446764179768174, "learning_rate": 9.786016723675912e-05, "loss": 3.1492, "step": 8898 }, { "epoch": 0.552424110745546, "grad_norm": 0.5067700175964543, "learning_rate": 9.785912187818181e-05, "loss": 3.1271, "step": 8899 }, { "epoch": 0.5524861878453039, "grad_norm": 0.5435333883791438, "learning_rate": 9.785807626991184e-05, "loss": 3.3001, "step": 8900 }, { "epoch": 0.5525482649450618, "grad_norm": 0.5675919585566978, "learning_rate": 9.785703041195466e-05, "loss": 3.2016, "step": 8901 }, { "epoch": 0.5526103420448196, "grad_norm": 0.5893148868118809, "learning_rate": 9.785598430431575e-05, "loss": 3.2475, "step": 8902 }, { "epoch": 0.5526724191445775, "grad_norm": 0.5019196970455105, "learning_rate": 9.785493794700053e-05, "loss": 3.1763, "step": 8903 }, { "epoch": 0.5527344962443355, "grad_norm": 0.48437344643128344, "learning_rate": 9.78538913400145e-05, "loss": 3.1944, "step": 8904 }, { "epoch": 0.5527965733440934, "grad_norm": 0.5392368974398138, "learning_rate": 9.78528444833631e-05, "loss": 3.2348, "step": 8905 }, { "epoch": 0.5528586504438513, "grad_norm": 0.5927485251808563, "learning_rate": 9.785179737705179e-05, "loss": 3.1106, "step": 8906 }, { "epoch": 0.5529207275436092, "grad_norm": 0.4837956655619213, "learning_rate": 9.785075002108604e-05, "loss": 3.1454, "step": 8907 }, { "epoch": 0.552982804643367, "grad_norm": 0.5551671385616579, "learning_rate": 9.78497024154713e-05, "loss": 3.1905, "step": 8908 }, { "epoch": 0.5530448817431249, "grad_norm": 0.555260134338683, "learning_rate": 9.784865456021307e-05, "loss": 3.271, "step": 8909 }, { "epoch": 0.5531069588428829, "grad_norm": 0.5173047189324362, "learning_rate": 9.784760645531679e-05, "loss": 3.1989, "step": 8910 }, { "epoch": 0.5531690359426408, "grad_norm": 0.49241879493250007, "learning_rate": 9.784655810078791e-05, "loss": 3.2388, "step": 8911 }, { "epoch": 0.5532311130423987, "grad_norm": 0.5227859484399674, "learning_rate": 9.784550949663193e-05, "loss": 3.226, "step": 8912 }, { "epoch": 0.5532931901421566, "grad_norm": 0.5754625771383325, "learning_rate": 9.784446064285431e-05, "loss": 3.2841, "step": 8913 }, { "epoch": 0.5533552672419144, "grad_norm": 0.5361703605913944, "learning_rate": 9.784341153946053e-05, "loss": 3.2323, "step": 8914 }, { "epoch": 0.5534173443416723, "grad_norm": 0.5438206802119436, "learning_rate": 9.784236218645605e-05, "loss": 3.377, "step": 8915 }, { "epoch": 0.5534794214414303, "grad_norm": 0.5384306314199658, "learning_rate": 9.784131258384637e-05, "loss": 3.1739, "step": 8916 }, { "epoch": 0.5535414985411882, "grad_norm": 0.6419325951474366, "learning_rate": 9.784026273163693e-05, "loss": 3.2416, "step": 8917 }, { "epoch": 0.5536035756409461, "grad_norm": 0.5935674185860877, "learning_rate": 9.783921262983324e-05, "loss": 3.2441, "step": 8918 }, { "epoch": 0.553665652740704, "grad_norm": 0.49265072232383833, "learning_rate": 9.783816227844076e-05, "loss": 3.2199, "step": 8919 }, { "epoch": 0.5537277298404618, "grad_norm": 0.5264070739920812, "learning_rate": 9.783711167746497e-05, "loss": 3.2709, "step": 8920 }, { "epoch": 0.5537898069402197, "grad_norm": 0.5243001633030683, "learning_rate": 9.783606082691134e-05, "loss": 3.0944, "step": 8921 }, { "epoch": 0.5538518840399776, "grad_norm": 0.476359370981974, "learning_rate": 9.783500972678539e-05, "loss": 3.1563, "step": 8922 }, { "epoch": 0.5539139611397356, "grad_norm": 0.5090358377292432, "learning_rate": 9.783395837709258e-05, "loss": 3.1814, "step": 8923 }, { "epoch": 0.5539760382394935, "grad_norm": 0.5669730408918304, "learning_rate": 9.78329067778384e-05, "loss": 3.1654, "step": 8924 }, { "epoch": 0.5540381153392514, "grad_norm": 0.5623562417175828, "learning_rate": 9.783185492902832e-05, "loss": 3.2182, "step": 8925 }, { "epoch": 0.5541001924390092, "grad_norm": 0.5796068377040144, "learning_rate": 9.783080283066786e-05, "loss": 3.2351, "step": 8926 }, { "epoch": 0.5541622695387671, "grad_norm": 0.5937916590481577, "learning_rate": 9.782975048276246e-05, "loss": 3.1952, "step": 8927 }, { "epoch": 0.554224346638525, "grad_norm": 0.5393194608947565, "learning_rate": 9.782869788531766e-05, "loss": 3.2855, "step": 8928 }, { "epoch": 0.554286423738283, "grad_norm": 0.5210183164796167, "learning_rate": 9.782764503833894e-05, "loss": 3.1659, "step": 8929 }, { "epoch": 0.5543485008380409, "grad_norm": 0.5483423438486527, "learning_rate": 9.782659194183177e-05, "loss": 3.3627, "step": 8930 }, { "epoch": 0.5544105779377988, "grad_norm": 0.5424616249856405, "learning_rate": 9.782553859580167e-05, "loss": 3.197, "step": 8931 }, { "epoch": 0.5544726550375566, "grad_norm": 0.5081099299393181, "learning_rate": 9.782448500025411e-05, "loss": 3.2055, "step": 8932 }, { "epoch": 0.5545347321373145, "grad_norm": 0.5286370419176835, "learning_rate": 9.782343115519462e-05, "loss": 3.242, "step": 8933 }, { "epoch": 0.5545968092370724, "grad_norm": 0.5979775207290969, "learning_rate": 9.782237706062868e-05, "loss": 3.2061, "step": 8934 }, { "epoch": 0.5546588863368304, "grad_norm": 0.4944869625756607, "learning_rate": 9.782132271656178e-05, "loss": 3.1698, "step": 8935 }, { "epoch": 0.5547209634365883, "grad_norm": 0.5771383165224048, "learning_rate": 9.782026812299942e-05, "loss": 3.2098, "step": 8936 }, { "epoch": 0.5547830405363462, "grad_norm": 0.6736202195678002, "learning_rate": 9.781921327994714e-05, "loss": 3.2356, "step": 8937 }, { "epoch": 0.554845117636104, "grad_norm": 0.654491361494007, "learning_rate": 9.78181581874104e-05, "loss": 3.2089, "step": 8938 }, { "epoch": 0.5549071947358619, "grad_norm": 0.5549598347675814, "learning_rate": 9.781710284539472e-05, "loss": 3.294, "step": 8939 }, { "epoch": 0.5549692718356198, "grad_norm": 0.5916054796987762, "learning_rate": 9.781604725390561e-05, "loss": 3.2116, "step": 8940 }, { "epoch": 0.5550313489353778, "grad_norm": 0.5237402254071076, "learning_rate": 9.781499141294857e-05, "loss": 3.2562, "step": 8941 }, { "epoch": 0.5550934260351357, "grad_norm": 0.5796024157246042, "learning_rate": 9.781393532252911e-05, "loss": 3.2463, "step": 8942 }, { "epoch": 0.5551555031348936, "grad_norm": 0.5831184473926815, "learning_rate": 9.781287898265275e-05, "loss": 3.1707, "step": 8943 }, { "epoch": 0.5552175802346514, "grad_norm": 0.4931417479046168, "learning_rate": 9.781182239332499e-05, "loss": 3.2121, "step": 8944 }, { "epoch": 0.5552796573344093, "grad_norm": 0.46846279246509465, "learning_rate": 9.781076555455135e-05, "loss": 3.1673, "step": 8945 }, { "epoch": 0.5553417344341672, "grad_norm": 0.4905487663425777, "learning_rate": 9.780970846633734e-05, "loss": 3.224, "step": 8946 }, { "epoch": 0.5554038115339252, "grad_norm": 0.508970144194406, "learning_rate": 9.780865112868847e-05, "loss": 3.1701, "step": 8947 }, { "epoch": 0.5554658886336831, "grad_norm": 0.4820133749987769, "learning_rate": 9.780759354161026e-05, "loss": 3.215, "step": 8948 }, { "epoch": 0.555527965733441, "grad_norm": 0.4863933994349251, "learning_rate": 9.780653570510822e-05, "loss": 3.1921, "step": 8949 }, { "epoch": 0.5555900428331988, "grad_norm": 0.5173701606260567, "learning_rate": 9.78054776191879e-05, "loss": 3.2473, "step": 8950 }, { "epoch": 0.5556521199329567, "grad_norm": 0.5507987810956645, "learning_rate": 9.780441928385477e-05, "loss": 3.2392, "step": 8951 }, { "epoch": 0.5557141970327146, "grad_norm": 0.514997071054148, "learning_rate": 9.78033606991144e-05, "loss": 3.1703, "step": 8952 }, { "epoch": 0.5557762741324725, "grad_norm": 0.4606236181147847, "learning_rate": 9.78023018649723e-05, "loss": 3.1829, "step": 8953 }, { "epoch": 0.5558383512322305, "grad_norm": 0.4723393347228119, "learning_rate": 9.780124278143396e-05, "loss": 3.2425, "step": 8954 }, { "epoch": 0.5559004283319884, "grad_norm": 0.523885961834329, "learning_rate": 9.780018344850495e-05, "loss": 3.2524, "step": 8955 }, { "epoch": 0.5559625054317462, "grad_norm": 0.4633399295794745, "learning_rate": 9.779912386619079e-05, "loss": 3.2125, "step": 8956 }, { "epoch": 0.5560245825315041, "grad_norm": 0.47633057563480435, "learning_rate": 9.779806403449699e-05, "loss": 3.2881, "step": 8957 }, { "epoch": 0.556086659631262, "grad_norm": 0.48107180702392016, "learning_rate": 9.779700395342907e-05, "loss": 3.2225, "step": 8958 }, { "epoch": 0.5561487367310199, "grad_norm": 0.45406205200257227, "learning_rate": 9.779594362299258e-05, "loss": 3.1414, "step": 8959 }, { "epoch": 0.5562108138307779, "grad_norm": 0.45955017603423726, "learning_rate": 9.779488304319306e-05, "loss": 3.192, "step": 8960 }, { "epoch": 0.5562728909305358, "grad_norm": 0.4720822308716298, "learning_rate": 9.779382221403604e-05, "loss": 3.2539, "step": 8961 }, { "epoch": 0.5563349680302936, "grad_norm": 0.4958334562825069, "learning_rate": 9.779276113552702e-05, "loss": 3.1642, "step": 8962 }, { "epoch": 0.5563970451300515, "grad_norm": 0.47794253227205563, "learning_rate": 9.779169980767158e-05, "loss": 3.2876, "step": 8963 }, { "epoch": 0.5564591222298094, "grad_norm": 0.4256261315519777, "learning_rate": 9.779063823047525e-05, "loss": 3.1756, "step": 8964 }, { "epoch": 0.5565211993295673, "grad_norm": 0.47947559211432245, "learning_rate": 9.778957640394355e-05, "loss": 3.2402, "step": 8965 }, { "epoch": 0.5565832764293253, "grad_norm": 0.5354931460083506, "learning_rate": 9.778851432808203e-05, "loss": 3.1638, "step": 8966 }, { "epoch": 0.5566453535290832, "grad_norm": 0.4696512300671178, "learning_rate": 9.77874520028962e-05, "loss": 3.2098, "step": 8967 }, { "epoch": 0.556707430628841, "grad_norm": 0.4807449260400269, "learning_rate": 9.778638942839166e-05, "loss": 3.1676, "step": 8968 }, { "epoch": 0.5567695077285989, "grad_norm": 0.5227344118071808, "learning_rate": 9.778532660457394e-05, "loss": 3.2219, "step": 8969 }, { "epoch": 0.5568315848283568, "grad_norm": 0.529962128644701, "learning_rate": 9.778426353144856e-05, "loss": 3.3088, "step": 8970 }, { "epoch": 0.5568936619281147, "grad_norm": 0.4876377595759059, "learning_rate": 9.778320020902107e-05, "loss": 3.2293, "step": 8971 }, { "epoch": 0.5569557390278727, "grad_norm": 0.576638398175862, "learning_rate": 9.778213663729702e-05, "loss": 3.1901, "step": 8972 }, { "epoch": 0.5570178161276306, "grad_norm": 0.6353906459379175, "learning_rate": 9.778107281628197e-05, "loss": 3.2492, "step": 8973 }, { "epoch": 0.5570798932273884, "grad_norm": 0.5411532395718536, "learning_rate": 9.778000874598147e-05, "loss": 3.2039, "step": 8974 }, { "epoch": 0.5571419703271463, "grad_norm": 0.4878639574299043, "learning_rate": 9.777894442640105e-05, "loss": 3.1458, "step": 8975 }, { "epoch": 0.5572040474269042, "grad_norm": 0.5552772288372478, "learning_rate": 9.77778798575463e-05, "loss": 3.2331, "step": 8976 }, { "epoch": 0.5572661245266621, "grad_norm": 0.5667303276299264, "learning_rate": 9.777681503942273e-05, "loss": 3.2986, "step": 8977 }, { "epoch": 0.55732820162642, "grad_norm": 0.6085538956369916, "learning_rate": 9.777574997203594e-05, "loss": 3.1792, "step": 8978 }, { "epoch": 0.557390278726178, "grad_norm": 0.4634108537407583, "learning_rate": 9.777468465539145e-05, "loss": 3.2693, "step": 8979 }, { "epoch": 0.5574523558259358, "grad_norm": 0.6497646676052241, "learning_rate": 9.777361908949483e-05, "loss": 3.1288, "step": 8980 }, { "epoch": 0.5575144329256937, "grad_norm": 0.6498542383791721, "learning_rate": 9.777255327435164e-05, "loss": 3.2119, "step": 8981 }, { "epoch": 0.5575765100254516, "grad_norm": 0.5511510432772213, "learning_rate": 9.777148720996746e-05, "loss": 3.2636, "step": 8982 }, { "epoch": 0.5576385871252095, "grad_norm": 0.6435163657279207, "learning_rate": 9.777042089634783e-05, "loss": 3.2278, "step": 8983 }, { "epoch": 0.5577006642249674, "grad_norm": 0.6985134424828916, "learning_rate": 9.77693543334983e-05, "loss": 3.2503, "step": 8984 }, { "epoch": 0.5577627413247254, "grad_norm": 0.6461539216257669, "learning_rate": 9.776828752142446e-05, "loss": 3.2611, "step": 8985 }, { "epoch": 0.5578248184244832, "grad_norm": 0.5183284287460863, "learning_rate": 9.776722046013187e-05, "loss": 3.2799, "step": 8986 }, { "epoch": 0.5578868955242411, "grad_norm": 0.6032074570497177, "learning_rate": 9.776615314962609e-05, "loss": 3.2163, "step": 8987 }, { "epoch": 0.557948972623999, "grad_norm": 0.558546771228102, "learning_rate": 9.77650855899127e-05, "loss": 3.233, "step": 8988 }, { "epoch": 0.5580110497237569, "grad_norm": 0.6068742251955973, "learning_rate": 9.776401778099728e-05, "loss": 3.2554, "step": 8989 }, { "epoch": 0.5580731268235148, "grad_norm": 0.49160588428899776, "learning_rate": 9.776294972288535e-05, "loss": 3.2088, "step": 8990 }, { "epoch": 0.5581352039232728, "grad_norm": 0.527608705998808, "learning_rate": 9.776188141558253e-05, "loss": 3.1019, "step": 8991 }, { "epoch": 0.5581972810230306, "grad_norm": 0.5110996957151447, "learning_rate": 9.776081285909439e-05, "loss": 3.2488, "step": 8992 }, { "epoch": 0.5582593581227885, "grad_norm": 0.5907853297446086, "learning_rate": 9.77597440534265e-05, "loss": 3.2333, "step": 8993 }, { "epoch": 0.5583214352225464, "grad_norm": 0.5077146289118396, "learning_rate": 9.77586749985844e-05, "loss": 3.1763, "step": 8994 }, { "epoch": 0.5583835123223043, "grad_norm": 0.4394960170582733, "learning_rate": 9.775760569457373e-05, "loss": 3.247, "step": 8995 }, { "epoch": 0.5584455894220622, "grad_norm": 0.47238644867827717, "learning_rate": 9.775653614140002e-05, "loss": 3.2391, "step": 8996 }, { "epoch": 0.55850766652182, "grad_norm": 0.49596119408833467, "learning_rate": 9.775546633906888e-05, "loss": 3.3012, "step": 8997 }, { "epoch": 0.558569743621578, "grad_norm": 0.4911625043831593, "learning_rate": 9.775439628758588e-05, "loss": 3.1768, "step": 8998 }, { "epoch": 0.5586318207213359, "grad_norm": 0.5372028893645859, "learning_rate": 9.77533259869566e-05, "loss": 3.2695, "step": 8999 }, { "epoch": 0.5586938978210938, "grad_norm": 0.4669911606736264, "learning_rate": 9.77522554371866e-05, "loss": 3.3194, "step": 9000 }, { "epoch": 0.5587559749208517, "grad_norm": 0.4451269465904447, "learning_rate": 9.775118463828152e-05, "loss": 3.2638, "step": 9001 }, { "epoch": 0.5588180520206096, "grad_norm": 0.41345640123430943, "learning_rate": 9.77501135902469e-05, "loss": 3.1523, "step": 9002 }, { "epoch": 0.5588801291203674, "grad_norm": 0.45226222145221917, "learning_rate": 9.774904229308836e-05, "loss": 3.1395, "step": 9003 }, { "epoch": 0.5589422062201254, "grad_norm": 0.4066767468564609, "learning_rate": 9.774797074681146e-05, "loss": 3.2357, "step": 9004 }, { "epoch": 0.5590042833198833, "grad_norm": 0.4721743436302139, "learning_rate": 9.774689895142181e-05, "loss": 3.1695, "step": 9005 }, { "epoch": 0.5590663604196412, "grad_norm": 0.5275644193880334, "learning_rate": 9.7745826906925e-05, "loss": 3.1757, "step": 9006 }, { "epoch": 0.5591284375193991, "grad_norm": 0.5767369231340128, "learning_rate": 9.774475461332664e-05, "loss": 3.1341, "step": 9007 }, { "epoch": 0.559190514619157, "grad_norm": 0.5124324451675978, "learning_rate": 9.774368207063228e-05, "loss": 3.1957, "step": 9008 }, { "epoch": 0.5592525917189148, "grad_norm": 0.4760677239390709, "learning_rate": 9.774260927884755e-05, "loss": 3.1893, "step": 9009 }, { "epoch": 0.5593146688186728, "grad_norm": 0.5054975121169932, "learning_rate": 9.774153623797804e-05, "loss": 3.2635, "step": 9010 }, { "epoch": 0.5593767459184307, "grad_norm": 0.6360719121652852, "learning_rate": 9.774046294802933e-05, "loss": 3.2777, "step": 9011 }, { "epoch": 0.5594388230181886, "grad_norm": 0.5121264398846486, "learning_rate": 9.773938940900704e-05, "loss": 3.1369, "step": 9012 }, { "epoch": 0.5595009001179465, "grad_norm": 0.5360913147700076, "learning_rate": 9.773831562091678e-05, "loss": 3.1959, "step": 9013 }, { "epoch": 0.5595629772177044, "grad_norm": 0.5438653691990283, "learning_rate": 9.773724158376413e-05, "loss": 3.0657, "step": 9014 }, { "epoch": 0.5596250543174622, "grad_norm": 0.5313823759419105, "learning_rate": 9.77361672975547e-05, "loss": 3.1537, "step": 9015 }, { "epoch": 0.5596871314172202, "grad_norm": 0.4947404199128575, "learning_rate": 9.77350927622941e-05, "loss": 3.2714, "step": 9016 }, { "epoch": 0.5597492085169781, "grad_norm": 0.5746156008841471, "learning_rate": 9.773401797798794e-05, "loss": 3.3176, "step": 9017 }, { "epoch": 0.559811285616736, "grad_norm": 0.5142332425389146, "learning_rate": 9.773294294464181e-05, "loss": 3.213, "step": 9018 }, { "epoch": 0.5598733627164939, "grad_norm": 0.5175881438815099, "learning_rate": 9.773186766226132e-05, "loss": 3.241, "step": 9019 }, { "epoch": 0.5599354398162518, "grad_norm": 0.5840767624666354, "learning_rate": 9.773079213085211e-05, "loss": 3.151, "step": 9020 }, { "epoch": 0.5599975169160096, "grad_norm": 0.5898441036804352, "learning_rate": 9.772971635041976e-05, "loss": 3.2085, "step": 9021 }, { "epoch": 0.5600595940157675, "grad_norm": 0.45368832069151427, "learning_rate": 9.77286403209699e-05, "loss": 3.2244, "step": 9022 }, { "epoch": 0.5601216711155255, "grad_norm": 0.545333870630225, "learning_rate": 9.772756404250813e-05, "loss": 3.1874, "step": 9023 }, { "epoch": 0.5601837482152834, "grad_norm": 0.5529908572539712, "learning_rate": 9.772648751504008e-05, "loss": 3.1556, "step": 9024 }, { "epoch": 0.5602458253150413, "grad_norm": 0.5107101226135831, "learning_rate": 9.772541073857135e-05, "loss": 3.2808, "step": 9025 }, { "epoch": 0.5603079024147992, "grad_norm": 0.5390753260759, "learning_rate": 9.772433371310756e-05, "loss": 3.1902, "step": 9026 }, { "epoch": 0.560369979514557, "grad_norm": 0.5098474078875771, "learning_rate": 9.772325643865434e-05, "loss": 3.1927, "step": 9027 }, { "epoch": 0.5604320566143149, "grad_norm": 0.5009761460806322, "learning_rate": 9.772217891521731e-05, "loss": 3.2835, "step": 9028 }, { "epoch": 0.5604941337140729, "grad_norm": 0.4658732354192373, "learning_rate": 9.772110114280208e-05, "loss": 3.1183, "step": 9029 }, { "epoch": 0.5605562108138308, "grad_norm": 0.5195440133039981, "learning_rate": 9.772002312141431e-05, "loss": 3.2033, "step": 9030 }, { "epoch": 0.5606182879135887, "grad_norm": 0.5340160000525724, "learning_rate": 9.771894485105956e-05, "loss": 3.2153, "step": 9031 }, { "epoch": 0.5606803650133466, "grad_norm": 0.5230333846434361, "learning_rate": 9.771786633174351e-05, "loss": 3.149, "step": 9032 }, { "epoch": 0.5607424421131044, "grad_norm": 0.5336436034744301, "learning_rate": 9.771678756347175e-05, "loss": 3.1089, "step": 9033 }, { "epoch": 0.5608045192128623, "grad_norm": 0.47740417947062347, "learning_rate": 9.771570854624993e-05, "loss": 3.1719, "step": 9034 }, { "epoch": 0.5608665963126203, "grad_norm": 0.47721279052286675, "learning_rate": 9.77146292800837e-05, "loss": 3.2123, "step": 9035 }, { "epoch": 0.5609286734123782, "grad_norm": 0.5001473507719572, "learning_rate": 9.771354976497864e-05, "loss": 3.2306, "step": 9036 }, { "epoch": 0.5609907505121361, "grad_norm": 0.458061166351996, "learning_rate": 9.771247000094042e-05, "loss": 3.2189, "step": 9037 }, { "epoch": 0.561052827611894, "grad_norm": 0.4645348492228563, "learning_rate": 9.771138998797465e-05, "loss": 3.2359, "step": 9038 }, { "epoch": 0.5611149047116518, "grad_norm": 0.4876851677746551, "learning_rate": 9.771030972608699e-05, "loss": 3.2277, "step": 9039 }, { "epoch": 0.5611769818114097, "grad_norm": 0.4593727903248068, "learning_rate": 9.770922921528305e-05, "loss": 3.065, "step": 9040 }, { "epoch": 0.5612390589111677, "grad_norm": 0.4383527925140224, "learning_rate": 9.770814845556848e-05, "loss": 3.2044, "step": 9041 }, { "epoch": 0.5613011360109256, "grad_norm": 0.502380248741642, "learning_rate": 9.770706744694892e-05, "loss": 3.2186, "step": 9042 }, { "epoch": 0.5613632131106835, "grad_norm": 0.47206655870933, "learning_rate": 9.770598618943002e-05, "loss": 3.2439, "step": 9043 }, { "epoch": 0.5614252902104414, "grad_norm": 0.45769372838048883, "learning_rate": 9.770490468301741e-05, "loss": 3.1994, "step": 9044 }, { "epoch": 0.5614873673101992, "grad_norm": 0.4653937187815612, "learning_rate": 9.770382292771672e-05, "loss": 3.2866, "step": 9045 }, { "epoch": 0.5615494444099571, "grad_norm": 0.5411930000033627, "learning_rate": 9.77027409235336e-05, "loss": 3.2105, "step": 9046 }, { "epoch": 0.561611521509715, "grad_norm": 0.4847314661180528, "learning_rate": 9.77016586704737e-05, "loss": 3.1257, "step": 9047 }, { "epoch": 0.561673598609473, "grad_norm": 0.5046682879685613, "learning_rate": 9.770057616854268e-05, "loss": 3.1971, "step": 9048 }, { "epoch": 0.5617356757092309, "grad_norm": 0.5150190607728667, "learning_rate": 9.769949341774617e-05, "loss": 3.1182, "step": 9049 }, { "epoch": 0.5617977528089888, "grad_norm": 0.4993751018366513, "learning_rate": 9.769841041808983e-05, "loss": 3.1577, "step": 9050 }, { "epoch": 0.5618598299087466, "grad_norm": 0.539528852336471, "learning_rate": 9.76973271695793e-05, "loss": 3.2885, "step": 9051 }, { "epoch": 0.5619219070085045, "grad_norm": 0.5840721702285401, "learning_rate": 9.769624367222021e-05, "loss": 3.2607, "step": 9052 }, { "epoch": 0.5619839841082624, "grad_norm": 0.5275578382159785, "learning_rate": 9.769515992601825e-05, "loss": 3.2572, "step": 9053 }, { "epoch": 0.5620460612080204, "grad_norm": 0.47899944213902396, "learning_rate": 9.769407593097908e-05, "loss": 3.171, "step": 9054 }, { "epoch": 0.5621081383077783, "grad_norm": 0.4733427752417218, "learning_rate": 9.769299168710833e-05, "loss": 3.2533, "step": 9055 }, { "epoch": 0.5621702154075362, "grad_norm": 0.5410443515243376, "learning_rate": 9.769190719441166e-05, "loss": 3.2671, "step": 9056 }, { "epoch": 0.562232292507294, "grad_norm": 0.5187847539451245, "learning_rate": 9.769082245289474e-05, "loss": 3.1741, "step": 9057 }, { "epoch": 0.5622943696070519, "grad_norm": 0.4790655129543846, "learning_rate": 9.768973746256322e-05, "loss": 3.2934, "step": 9058 }, { "epoch": 0.5623564467068098, "grad_norm": 0.5273150118308735, "learning_rate": 9.768865222342276e-05, "loss": 3.2553, "step": 9059 }, { "epoch": 0.5624185238065678, "grad_norm": 0.4856032058803637, "learning_rate": 9.768756673547902e-05, "loss": 3.0755, "step": 9060 }, { "epoch": 0.5624806009063257, "grad_norm": 0.4216768364502933, "learning_rate": 9.768648099873768e-05, "loss": 3.1252, "step": 9061 }, { "epoch": 0.5625426780060836, "grad_norm": 0.5372855157735527, "learning_rate": 9.768539501320437e-05, "loss": 3.2837, "step": 9062 }, { "epoch": 0.5626047551058414, "grad_norm": 0.4840610317210419, "learning_rate": 9.76843087788848e-05, "loss": 3.2791, "step": 9063 }, { "epoch": 0.5626668322055993, "grad_norm": 0.4964112294697488, "learning_rate": 9.768322229578461e-05, "loss": 3.1659, "step": 9064 }, { "epoch": 0.5627289093053572, "grad_norm": 0.47377938431069966, "learning_rate": 9.768213556390945e-05, "loss": 3.2391, "step": 9065 }, { "epoch": 0.5627909864051152, "grad_norm": 0.5230060336383094, "learning_rate": 9.768104858326506e-05, "loss": 3.2047, "step": 9066 }, { "epoch": 0.5628530635048731, "grad_norm": 0.48356539222510075, "learning_rate": 9.767996135385703e-05, "loss": 3.2433, "step": 9067 }, { "epoch": 0.562915140604631, "grad_norm": 0.45157727280049925, "learning_rate": 9.767887387569108e-05, "loss": 3.1756, "step": 9068 }, { "epoch": 0.5629772177043888, "grad_norm": 0.542423138820973, "learning_rate": 9.767778614877286e-05, "loss": 3.1587, "step": 9069 }, { "epoch": 0.5630392948041467, "grad_norm": 0.5192090232688819, "learning_rate": 9.767669817310808e-05, "loss": 3.1314, "step": 9070 }, { "epoch": 0.5631013719039046, "grad_norm": 0.5780252937735548, "learning_rate": 9.767560994870236e-05, "loss": 3.1789, "step": 9071 }, { "epoch": 0.5631634490036626, "grad_norm": 0.6080444432738955, "learning_rate": 9.767452147556142e-05, "loss": 3.2755, "step": 9072 }, { "epoch": 0.5632255261034205, "grad_norm": 0.5889927870886813, "learning_rate": 9.767343275369094e-05, "loss": 3.2193, "step": 9073 }, { "epoch": 0.5632876032031784, "grad_norm": 0.5280221037536779, "learning_rate": 9.767234378309658e-05, "loss": 3.1597, "step": 9074 }, { "epoch": 0.5633496803029362, "grad_norm": 0.5228320363573536, "learning_rate": 9.767125456378403e-05, "loss": 3.1697, "step": 9075 }, { "epoch": 0.5634117574026941, "grad_norm": 0.6089652713620097, "learning_rate": 9.767016509575897e-05, "loss": 3.2324, "step": 9076 }, { "epoch": 0.563473834502452, "grad_norm": 0.608116023123841, "learning_rate": 9.766907537902709e-05, "loss": 3.2428, "step": 9077 }, { "epoch": 0.56353591160221, "grad_norm": 0.5727585632186927, "learning_rate": 9.766798541359407e-05, "loss": 3.243, "step": 9078 }, { "epoch": 0.5635979887019679, "grad_norm": 0.5966051869933877, "learning_rate": 9.766689519946559e-05, "loss": 3.1709, "step": 9079 }, { "epoch": 0.5636600658017258, "grad_norm": 0.567483233233565, "learning_rate": 9.766580473664736e-05, "loss": 3.2358, "step": 9080 }, { "epoch": 0.5637221429014836, "grad_norm": 0.5445341966677917, "learning_rate": 9.766471402514506e-05, "loss": 3.2636, "step": 9081 }, { "epoch": 0.5637842200012415, "grad_norm": 0.5665949540014205, "learning_rate": 9.766362306496434e-05, "loss": 3.1562, "step": 9082 }, { "epoch": 0.5638462971009994, "grad_norm": 0.4875941515750181, "learning_rate": 9.766253185611097e-05, "loss": 3.2705, "step": 9083 }, { "epoch": 0.5639083742007573, "grad_norm": 0.49559205887415336, "learning_rate": 9.766144039859057e-05, "loss": 3.2596, "step": 9084 }, { "epoch": 0.5639704513005153, "grad_norm": 0.5336586819150084, "learning_rate": 9.76603486924089e-05, "loss": 3.161, "step": 9085 }, { "epoch": 0.5640325284002732, "grad_norm": 0.4560006904084433, "learning_rate": 9.765925673757159e-05, "loss": 3.2368, "step": 9086 }, { "epoch": 0.564094605500031, "grad_norm": 0.4801050996457508, "learning_rate": 9.765816453408439e-05, "loss": 3.1617, "step": 9087 }, { "epoch": 0.5641566825997889, "grad_norm": 0.5409963170541431, "learning_rate": 9.765707208195296e-05, "loss": 3.2483, "step": 9088 }, { "epoch": 0.5642187596995468, "grad_norm": 0.5064805094724264, "learning_rate": 9.765597938118302e-05, "loss": 3.2459, "step": 9089 }, { "epoch": 0.5642808367993047, "grad_norm": 0.543313198472244, "learning_rate": 9.765488643178027e-05, "loss": 3.297, "step": 9090 }, { "epoch": 0.5643429138990627, "grad_norm": 0.6048396133600683, "learning_rate": 9.76537932337504e-05, "loss": 3.3308, "step": 9091 }, { "epoch": 0.5644049909988206, "grad_norm": 0.6678190230826153, "learning_rate": 9.765269978709912e-05, "loss": 3.249, "step": 9092 }, { "epoch": 0.5644670680985784, "grad_norm": 0.5771948058322248, "learning_rate": 9.765160609183215e-05, "loss": 3.2348, "step": 9093 }, { "epoch": 0.5645291451983363, "grad_norm": 0.5010029151930674, "learning_rate": 9.765051214795518e-05, "loss": 3.2901, "step": 9094 }, { "epoch": 0.5645912222980942, "grad_norm": 0.5400005692461334, "learning_rate": 9.764941795547391e-05, "loss": 3.1869, "step": 9095 }, { "epoch": 0.5646532993978521, "grad_norm": 0.5162683865595789, "learning_rate": 9.764832351439408e-05, "loss": 3.1271, "step": 9096 }, { "epoch": 0.56471537649761, "grad_norm": 0.5159628368235465, "learning_rate": 9.764722882472137e-05, "loss": 3.192, "step": 9097 }, { "epoch": 0.564777453597368, "grad_norm": 0.4795016969565236, "learning_rate": 9.76461338864615e-05, "loss": 3.2331, "step": 9098 }, { "epoch": 0.5648395306971258, "grad_norm": 0.5729207269929288, "learning_rate": 9.764503869962017e-05, "loss": 3.2173, "step": 9099 }, { "epoch": 0.5649016077968837, "grad_norm": 0.4919833562224167, "learning_rate": 9.764394326420311e-05, "loss": 3.2554, "step": 9100 }, { "epoch": 0.5649636848966416, "grad_norm": 0.4876518312624987, "learning_rate": 9.764284758021605e-05, "loss": 3.2348, "step": 9101 }, { "epoch": 0.5650257619963995, "grad_norm": 0.566196192233259, "learning_rate": 9.764175164766466e-05, "loss": 3.1453, "step": 9102 }, { "epoch": 0.5650878390961575, "grad_norm": 0.5463104331153426, "learning_rate": 9.764065546655472e-05, "loss": 3.3122, "step": 9103 }, { "epoch": 0.5651499161959154, "grad_norm": 0.5623146387500291, "learning_rate": 9.76395590368919e-05, "loss": 3.1948, "step": 9104 }, { "epoch": 0.5652119932956732, "grad_norm": 0.46599022347607494, "learning_rate": 9.763846235868194e-05, "loss": 3.2378, "step": 9105 }, { "epoch": 0.5652740703954311, "grad_norm": 0.5401383596673655, "learning_rate": 9.763736543193055e-05, "loss": 3.1637, "step": 9106 }, { "epoch": 0.565336147495189, "grad_norm": 0.5241402420859048, "learning_rate": 9.763626825664346e-05, "loss": 3.1842, "step": 9107 }, { "epoch": 0.5653982245949469, "grad_norm": 0.5378761228602429, "learning_rate": 9.763517083282641e-05, "loss": 3.2618, "step": 9108 }, { "epoch": 0.5654603016947048, "grad_norm": 0.5423300300166171, "learning_rate": 9.763407316048508e-05, "loss": 3.2792, "step": 9109 }, { "epoch": 0.5655223787944628, "grad_norm": 0.5096066106323327, "learning_rate": 9.763297523962525e-05, "loss": 3.1719, "step": 9110 }, { "epoch": 0.5655844558942206, "grad_norm": 0.47970088604568295, "learning_rate": 9.763187707025262e-05, "loss": 3.1776, "step": 9111 }, { "epoch": 0.5656465329939785, "grad_norm": 0.50579209045917, "learning_rate": 9.763077865237292e-05, "loss": 3.1139, "step": 9112 }, { "epoch": 0.5657086100937364, "grad_norm": 0.4582994585812704, "learning_rate": 9.762967998599188e-05, "loss": 3.1739, "step": 9113 }, { "epoch": 0.5657706871934943, "grad_norm": 0.4648626467926951, "learning_rate": 9.762858107111525e-05, "loss": 3.2482, "step": 9114 }, { "epoch": 0.5658327642932522, "grad_norm": 0.5126323842984398, "learning_rate": 9.762748190774875e-05, "loss": 3.188, "step": 9115 }, { "epoch": 0.5658948413930102, "grad_norm": 0.5486444729373546, "learning_rate": 9.76263824958981e-05, "loss": 3.1532, "step": 9116 }, { "epoch": 0.565956918492768, "grad_norm": 0.6184461772477653, "learning_rate": 9.762528283556907e-05, "loss": 3.309, "step": 9117 }, { "epoch": 0.5660189955925259, "grad_norm": 0.5958688224102551, "learning_rate": 9.762418292676736e-05, "loss": 3.241, "step": 9118 }, { "epoch": 0.5660810726922838, "grad_norm": 0.6453937193072885, "learning_rate": 9.762308276949874e-05, "loss": 3.2484, "step": 9119 }, { "epoch": 0.5661431497920417, "grad_norm": 0.4991850917734669, "learning_rate": 9.762198236376894e-05, "loss": 3.2019, "step": 9120 }, { "epoch": 0.5662052268917996, "grad_norm": 0.5906984313851023, "learning_rate": 9.762088170958368e-05, "loss": 3.2141, "step": 9121 }, { "epoch": 0.5662673039915576, "grad_norm": 0.5428714253314275, "learning_rate": 9.761978080694873e-05, "loss": 3.21, "step": 9122 }, { "epoch": 0.5663293810913154, "grad_norm": 0.5477114955665229, "learning_rate": 9.761867965586983e-05, "loss": 3.1273, "step": 9123 }, { "epoch": 0.5663914581910733, "grad_norm": 0.5209600262728685, "learning_rate": 9.761757825635271e-05, "loss": 3.2674, "step": 9124 }, { "epoch": 0.5664535352908312, "grad_norm": 0.4644688127978743, "learning_rate": 9.761647660840312e-05, "loss": 3.235, "step": 9125 }, { "epoch": 0.5665156123905891, "grad_norm": 0.49863630832234135, "learning_rate": 9.761537471202682e-05, "loss": 3.1885, "step": 9126 }, { "epoch": 0.566577689490347, "grad_norm": 0.4765698635595544, "learning_rate": 9.761427256722955e-05, "loss": 3.1099, "step": 9127 }, { "epoch": 0.566639766590105, "grad_norm": 0.4481720381221335, "learning_rate": 9.761317017401705e-05, "loss": 3.2166, "step": 9128 }, { "epoch": 0.5667018436898628, "grad_norm": 0.42324178543388585, "learning_rate": 9.761206753239511e-05, "loss": 3.2017, "step": 9129 }, { "epoch": 0.5667639207896207, "grad_norm": 0.46443948878105146, "learning_rate": 9.761096464236942e-05, "loss": 3.2238, "step": 9130 }, { "epoch": 0.5668259978893786, "grad_norm": 0.4889175236719758, "learning_rate": 9.76098615039458e-05, "loss": 3.2378, "step": 9131 }, { "epoch": 0.5668880749891365, "grad_norm": 0.46771249394028247, "learning_rate": 9.760875811712995e-05, "loss": 3.1676, "step": 9132 }, { "epoch": 0.5669501520888944, "grad_norm": 0.46656507858003277, "learning_rate": 9.760765448192768e-05, "loss": 3.1836, "step": 9133 }, { "epoch": 0.5670122291886524, "grad_norm": 0.48576941818400043, "learning_rate": 9.76065505983447e-05, "loss": 3.1852, "step": 9134 }, { "epoch": 0.5670743062884102, "grad_norm": 0.48231773951384654, "learning_rate": 9.760544646638678e-05, "loss": 3.1779, "step": 9135 }, { "epoch": 0.5671363833881681, "grad_norm": 0.43303388242657137, "learning_rate": 9.760434208605971e-05, "loss": 3.1781, "step": 9136 }, { "epoch": 0.567198460487926, "grad_norm": 0.5490260191659607, "learning_rate": 9.760323745736922e-05, "loss": 3.3106, "step": 9137 }, { "epoch": 0.5672605375876839, "grad_norm": 0.5669654716376312, "learning_rate": 9.760213258032109e-05, "loss": 3.1864, "step": 9138 }, { "epoch": 0.5673226146874418, "grad_norm": 0.5227303639139592, "learning_rate": 9.760102745492108e-05, "loss": 3.2188, "step": 9139 }, { "epoch": 0.5673846917871997, "grad_norm": 0.4671244888170164, "learning_rate": 9.759992208117495e-05, "loss": 3.181, "step": 9140 }, { "epoch": 0.5674467688869576, "grad_norm": 0.5372369233319962, "learning_rate": 9.759881645908846e-05, "loss": 3.1447, "step": 9141 }, { "epoch": 0.5675088459867155, "grad_norm": 0.47811561525397034, "learning_rate": 9.75977105886674e-05, "loss": 3.1958, "step": 9142 }, { "epoch": 0.5675709230864734, "grad_norm": 0.5253458677606239, "learning_rate": 9.759660446991753e-05, "loss": 3.2965, "step": 9143 }, { "epoch": 0.5676330001862313, "grad_norm": 0.5684688764136556, "learning_rate": 9.759549810284462e-05, "loss": 3.2345, "step": 9144 }, { "epoch": 0.5676950772859892, "grad_norm": 0.6426839770293137, "learning_rate": 9.759439148745443e-05, "loss": 3.2152, "step": 9145 }, { "epoch": 0.5677571543857471, "grad_norm": 0.5865207312563092, "learning_rate": 9.759328462375276e-05, "loss": 3.1809, "step": 9146 }, { "epoch": 0.567819231485505, "grad_norm": 0.516333295831172, "learning_rate": 9.759217751174536e-05, "loss": 3.2088, "step": 9147 }, { "epoch": 0.5678813085852629, "grad_norm": 0.4857617339406884, "learning_rate": 9.759107015143804e-05, "loss": 3.2145, "step": 9148 }, { "epoch": 0.5679433856850208, "grad_norm": 0.5161755541402865, "learning_rate": 9.758996254283654e-05, "loss": 3.1415, "step": 9149 }, { "epoch": 0.5680054627847787, "grad_norm": 0.4797697332268738, "learning_rate": 9.758885468594664e-05, "loss": 3.1967, "step": 9150 }, { "epoch": 0.5680675398845366, "grad_norm": 0.525585351190201, "learning_rate": 9.758774658077414e-05, "loss": 3.1791, "step": 9151 }, { "epoch": 0.5681296169842945, "grad_norm": 0.4689469400589744, "learning_rate": 9.758663822732483e-05, "loss": 3.1863, "step": 9152 }, { "epoch": 0.5681916940840523, "grad_norm": 0.5055215952398675, "learning_rate": 9.758552962560445e-05, "loss": 3.15, "step": 9153 }, { "epoch": 0.5682537711838103, "grad_norm": 0.5019078809482341, "learning_rate": 9.758442077561882e-05, "loss": 3.0645, "step": 9154 }, { "epoch": 0.5683158482835682, "grad_norm": 0.49108197925429925, "learning_rate": 9.758331167737372e-05, "loss": 3.1331, "step": 9155 }, { "epoch": 0.5683779253833261, "grad_norm": 0.5106233418413225, "learning_rate": 9.758220233087494e-05, "loss": 3.3231, "step": 9156 }, { "epoch": 0.568440002483084, "grad_norm": 0.557685380229878, "learning_rate": 9.758109273612824e-05, "loss": 3.3, "step": 9157 }, { "epoch": 0.5685020795828419, "grad_norm": 0.5409110067966006, "learning_rate": 9.757998289313945e-05, "loss": 3.2704, "step": 9158 }, { "epoch": 0.5685641566825997, "grad_norm": 0.4719799184053028, "learning_rate": 9.757887280191431e-05, "loss": 3.157, "step": 9159 }, { "epoch": 0.5686262337823577, "grad_norm": 0.5014823160219868, "learning_rate": 9.757776246245867e-05, "loss": 3.1712, "step": 9160 }, { "epoch": 0.5686883108821156, "grad_norm": 0.4887944233259769, "learning_rate": 9.757665187477827e-05, "loss": 3.1431, "step": 9161 }, { "epoch": 0.5687503879818735, "grad_norm": 0.4886111556400069, "learning_rate": 9.757554103887894e-05, "loss": 3.2159, "step": 9162 }, { "epoch": 0.5688124650816314, "grad_norm": 0.5126947021480742, "learning_rate": 9.757442995476646e-05, "loss": 3.1534, "step": 9163 }, { "epoch": 0.5688745421813893, "grad_norm": 0.5849187259377857, "learning_rate": 9.757331862244663e-05, "loss": 3.1006, "step": 9164 }, { "epoch": 0.5689366192811471, "grad_norm": 0.587543727383743, "learning_rate": 9.757220704192525e-05, "loss": 3.1353, "step": 9165 }, { "epoch": 0.5689986963809051, "grad_norm": 0.5640807722096182, "learning_rate": 9.757109521320812e-05, "loss": 3.2044, "step": 9166 }, { "epoch": 0.569060773480663, "grad_norm": 0.5570712113993426, "learning_rate": 9.756998313630103e-05, "loss": 3.1532, "step": 9167 }, { "epoch": 0.5691228505804209, "grad_norm": 0.5155060370801221, "learning_rate": 9.75688708112098e-05, "loss": 3.1391, "step": 9168 }, { "epoch": 0.5691849276801788, "grad_norm": 0.6056986956641689, "learning_rate": 9.75677582379402e-05, "loss": 3.2648, "step": 9169 }, { "epoch": 0.5692470047799367, "grad_norm": 0.5680511005629012, "learning_rate": 9.75666454164981e-05, "loss": 3.221, "step": 9170 }, { "epoch": 0.5693090818796945, "grad_norm": 0.47731805535570476, "learning_rate": 9.756553234688923e-05, "loss": 3.2507, "step": 9171 }, { "epoch": 0.5693711589794525, "grad_norm": 0.5503466088351399, "learning_rate": 9.756441902911943e-05, "loss": 3.1785, "step": 9172 }, { "epoch": 0.5694332360792104, "grad_norm": 0.5885480582191195, "learning_rate": 9.756330546319452e-05, "loss": 3.1641, "step": 9173 }, { "epoch": 0.5694953131789683, "grad_norm": 0.5297396732618164, "learning_rate": 9.75621916491203e-05, "loss": 3.1722, "step": 9174 }, { "epoch": 0.5695573902787262, "grad_norm": 0.48965194573075704, "learning_rate": 9.756107758690258e-05, "loss": 3.2223, "step": 9175 }, { "epoch": 0.5696194673784841, "grad_norm": 0.4890569704858352, "learning_rate": 9.755996327654717e-05, "loss": 3.1265, "step": 9176 }, { "epoch": 0.5696815444782419, "grad_norm": 0.5513428334375174, "learning_rate": 9.755884871805989e-05, "loss": 3.1518, "step": 9177 }, { "epoch": 0.5697436215779998, "grad_norm": 0.42964038590580184, "learning_rate": 9.755773391144654e-05, "loss": 3.2584, "step": 9178 }, { "epoch": 0.5698056986777578, "grad_norm": 0.4966429126369374, "learning_rate": 9.755661885671296e-05, "loss": 3.2753, "step": 9179 }, { "epoch": 0.5698677757775157, "grad_norm": 0.5099704374833247, "learning_rate": 9.755550355386493e-05, "loss": 3.2512, "step": 9180 }, { "epoch": 0.5699298528772736, "grad_norm": 0.4552674943232465, "learning_rate": 9.755438800290832e-05, "loss": 3.0279, "step": 9181 }, { "epoch": 0.5699919299770315, "grad_norm": 0.4907983522249929, "learning_rate": 9.755327220384892e-05, "loss": 3.1962, "step": 9182 }, { "epoch": 0.5700540070767893, "grad_norm": 0.49333194103667394, "learning_rate": 9.755215615669255e-05, "loss": 3.2058, "step": 9183 }, { "epoch": 0.5701160841765472, "grad_norm": 0.5055463846245847, "learning_rate": 9.755103986144505e-05, "loss": 3.2965, "step": 9184 }, { "epoch": 0.5701781612763052, "grad_norm": 0.46210470963071515, "learning_rate": 9.754992331811222e-05, "loss": 3.2127, "step": 9185 }, { "epoch": 0.5702402383760631, "grad_norm": 0.4677349863196514, "learning_rate": 9.754880652669989e-05, "loss": 3.2677, "step": 9186 }, { "epoch": 0.570302315475821, "grad_norm": 0.5262079227959509, "learning_rate": 9.754768948721388e-05, "loss": 3.3061, "step": 9187 }, { "epoch": 0.5703643925755789, "grad_norm": 0.48870396436293617, "learning_rate": 9.754657219966007e-05, "loss": 3.1935, "step": 9188 }, { "epoch": 0.5704264696753367, "grad_norm": 0.48013449105970035, "learning_rate": 9.754545466404424e-05, "loss": 3.2484, "step": 9189 }, { "epoch": 0.5704885467750946, "grad_norm": 0.5331594987285873, "learning_rate": 9.754433688037222e-05, "loss": 3.2521, "step": 9190 }, { "epoch": 0.5705506238748526, "grad_norm": 0.4640286937434868, "learning_rate": 9.754321884864985e-05, "loss": 3.2124, "step": 9191 }, { "epoch": 0.5706127009746105, "grad_norm": 0.5473511938688236, "learning_rate": 9.754210056888298e-05, "loss": 3.1945, "step": 9192 }, { "epoch": 0.5706747780743684, "grad_norm": 0.4920084491388045, "learning_rate": 9.754098204107741e-05, "loss": 3.2054, "step": 9193 }, { "epoch": 0.5707368551741263, "grad_norm": 0.49326986556499236, "learning_rate": 9.753986326523901e-05, "loss": 3.2695, "step": 9194 }, { "epoch": 0.5707989322738841, "grad_norm": 0.4278803711213318, "learning_rate": 9.753874424137361e-05, "loss": 3.2002, "step": 9195 }, { "epoch": 0.570861009373642, "grad_norm": 0.4349543214251117, "learning_rate": 9.753762496948702e-05, "loss": 3.2038, "step": 9196 }, { "epoch": 0.5709230864734, "grad_norm": 0.5368227939735727, "learning_rate": 9.753650544958511e-05, "loss": 3.178, "step": 9197 }, { "epoch": 0.5709851635731579, "grad_norm": 0.4938101134344555, "learning_rate": 9.753538568167371e-05, "loss": 3.2346, "step": 9198 }, { "epoch": 0.5710472406729158, "grad_norm": 0.4860026791149438, "learning_rate": 9.753426566575867e-05, "loss": 3.1917, "step": 9199 }, { "epoch": 0.5711093177726737, "grad_norm": 0.5429559884527034, "learning_rate": 9.753314540184581e-05, "loss": 3.1909, "step": 9200 }, { "epoch": 0.5711713948724315, "grad_norm": 0.5423841004567553, "learning_rate": 9.753202488994102e-05, "loss": 3.2432, "step": 9201 }, { "epoch": 0.5712334719721894, "grad_norm": 0.4303718146207107, "learning_rate": 9.753090413005011e-05, "loss": 3.2008, "step": 9202 }, { "epoch": 0.5712955490719474, "grad_norm": 0.4759083463790603, "learning_rate": 9.752978312217891e-05, "loss": 3.2169, "step": 9203 }, { "epoch": 0.5713576261717053, "grad_norm": 0.4277170604528348, "learning_rate": 9.752866186633331e-05, "loss": 3.1921, "step": 9204 }, { "epoch": 0.5714197032714632, "grad_norm": 0.5455327045731361, "learning_rate": 9.752754036251914e-05, "loss": 3.2209, "step": 9205 }, { "epoch": 0.5714817803712211, "grad_norm": 0.5376599905409952, "learning_rate": 9.752641861074225e-05, "loss": 3.2009, "step": 9206 }, { "epoch": 0.5715438574709789, "grad_norm": 0.5242890221330674, "learning_rate": 9.752529661100851e-05, "loss": 3.069, "step": 9207 }, { "epoch": 0.5716059345707368, "grad_norm": 0.47393598783307894, "learning_rate": 9.752417436332376e-05, "loss": 3.2122, "step": 9208 }, { "epoch": 0.5716680116704947, "grad_norm": 0.5183465399802443, "learning_rate": 9.752305186769384e-05, "loss": 3.1832, "step": 9209 }, { "epoch": 0.5717300887702527, "grad_norm": 0.5569865707733862, "learning_rate": 9.752192912412463e-05, "loss": 3.2117, "step": 9210 }, { "epoch": 0.5717921658700106, "grad_norm": 0.496219393821696, "learning_rate": 9.752080613262197e-05, "loss": 3.1593, "step": 9211 }, { "epoch": 0.5718542429697685, "grad_norm": 0.505330195591872, "learning_rate": 9.751968289319176e-05, "loss": 3.1095, "step": 9212 }, { "epoch": 0.5719163200695263, "grad_norm": 0.46496042021612305, "learning_rate": 9.75185594058398e-05, "loss": 3.1788, "step": 9213 }, { "epoch": 0.5719783971692842, "grad_norm": 0.5536079916547412, "learning_rate": 9.751743567057199e-05, "loss": 3.2862, "step": 9214 }, { "epoch": 0.5720404742690421, "grad_norm": 0.5008936584763496, "learning_rate": 9.751631168739418e-05, "loss": 3.1704, "step": 9215 }, { "epoch": 0.5721025513688001, "grad_norm": 0.44815471518727007, "learning_rate": 9.751518745631222e-05, "loss": 3.2427, "step": 9216 }, { "epoch": 0.572164628468558, "grad_norm": 0.4651947299262696, "learning_rate": 9.751406297733202e-05, "loss": 3.2776, "step": 9217 }, { "epoch": 0.5722267055683159, "grad_norm": 0.4575192381248227, "learning_rate": 9.751293825045942e-05, "loss": 3.1775, "step": 9218 }, { "epoch": 0.5722887826680737, "grad_norm": 0.49330472545328513, "learning_rate": 9.751181327570027e-05, "loss": 3.1425, "step": 9219 }, { "epoch": 0.5723508597678316, "grad_norm": 0.4291781267315004, "learning_rate": 9.751068805306046e-05, "loss": 3.2038, "step": 9220 }, { "epoch": 0.5724129368675895, "grad_norm": 0.49374332966705065, "learning_rate": 9.750956258254586e-05, "loss": 3.1075, "step": 9221 }, { "epoch": 0.5724750139673475, "grad_norm": 0.5191662302010275, "learning_rate": 9.750843686416233e-05, "loss": 3.2019, "step": 9222 }, { "epoch": 0.5725370910671054, "grad_norm": 0.4868707655207135, "learning_rate": 9.750731089791576e-05, "loss": 3.1952, "step": 9223 }, { "epoch": 0.5725991681668633, "grad_norm": 0.5548852111485274, "learning_rate": 9.750618468381201e-05, "loss": 3.1483, "step": 9224 }, { "epoch": 0.5726612452666211, "grad_norm": 0.47259610952070685, "learning_rate": 9.750505822185696e-05, "loss": 3.1775, "step": 9225 }, { "epoch": 0.572723322366379, "grad_norm": 0.5141464477914197, "learning_rate": 9.75039315120565e-05, "loss": 3.1729, "step": 9226 }, { "epoch": 0.5727853994661369, "grad_norm": 0.56562237923368, "learning_rate": 9.75028045544165e-05, "loss": 3.2867, "step": 9227 }, { "epoch": 0.5728474765658949, "grad_norm": 0.608674771910736, "learning_rate": 9.750167734894282e-05, "loss": 3.2102, "step": 9228 }, { "epoch": 0.5729095536656528, "grad_norm": 0.5318029555265511, "learning_rate": 9.750054989564136e-05, "loss": 3.117, "step": 9229 }, { "epoch": 0.5729716307654107, "grad_norm": 0.5300714378525465, "learning_rate": 9.749942219451802e-05, "loss": 3.2353, "step": 9230 }, { "epoch": 0.5730337078651685, "grad_norm": 0.5570092837589007, "learning_rate": 9.749829424557866e-05, "loss": 3.2697, "step": 9231 }, { "epoch": 0.5730957849649264, "grad_norm": 0.633382258177705, "learning_rate": 9.749716604882915e-05, "loss": 3.3126, "step": 9232 }, { "epoch": 0.5731578620646843, "grad_norm": 0.6250693759560999, "learning_rate": 9.749603760427541e-05, "loss": 3.2122, "step": 9233 }, { "epoch": 0.5732199391644422, "grad_norm": 0.5307607080472783, "learning_rate": 9.74949089119233e-05, "loss": 3.1949, "step": 9234 }, { "epoch": 0.5732820162642002, "grad_norm": 0.5359114448731124, "learning_rate": 9.749377997177873e-05, "loss": 3.1106, "step": 9235 }, { "epoch": 0.5733440933639581, "grad_norm": 0.6182535838892095, "learning_rate": 9.749265078384757e-05, "loss": 3.2054, "step": 9236 }, { "epoch": 0.5734061704637159, "grad_norm": 0.6346567383321526, "learning_rate": 9.749152134813573e-05, "loss": 3.2443, "step": 9237 }, { "epoch": 0.5734682475634738, "grad_norm": 0.5689784869583776, "learning_rate": 9.749039166464911e-05, "loss": 3.1661, "step": 9238 }, { "epoch": 0.5735303246632317, "grad_norm": 0.5476961782486828, "learning_rate": 9.748926173339357e-05, "loss": 3.2405, "step": 9239 }, { "epoch": 0.5735924017629896, "grad_norm": 0.5446710043251575, "learning_rate": 9.748813155437502e-05, "loss": 3.2466, "step": 9240 }, { "epoch": 0.5736544788627476, "grad_norm": 0.5844372425488764, "learning_rate": 9.748700112759935e-05, "loss": 3.0878, "step": 9241 }, { "epoch": 0.5737165559625055, "grad_norm": 0.4609578823173712, "learning_rate": 9.748587045307249e-05, "loss": 3.193, "step": 9242 }, { "epoch": 0.5737786330622633, "grad_norm": 0.5349930563146317, "learning_rate": 9.74847395308003e-05, "loss": 3.1538, "step": 9243 }, { "epoch": 0.5738407101620212, "grad_norm": 0.5424528621375503, "learning_rate": 9.748360836078871e-05, "loss": 3.2186, "step": 9244 }, { "epoch": 0.5739027872617791, "grad_norm": 0.5602514039838665, "learning_rate": 9.74824769430436e-05, "loss": 3.1477, "step": 9245 }, { "epoch": 0.573964864361537, "grad_norm": 0.5274839426823817, "learning_rate": 9.748134527757088e-05, "loss": 3.1633, "step": 9246 }, { "epoch": 0.574026941461295, "grad_norm": 0.4829629034196324, "learning_rate": 9.748021336437645e-05, "loss": 3.2397, "step": 9247 }, { "epoch": 0.5740890185610529, "grad_norm": 0.5550489456915183, "learning_rate": 9.747908120346623e-05, "loss": 3.1505, "step": 9248 }, { "epoch": 0.5741510956608107, "grad_norm": 0.5048944409210018, "learning_rate": 9.747794879484612e-05, "loss": 3.2187, "step": 9249 }, { "epoch": 0.5742131727605686, "grad_norm": 0.4874660094955866, "learning_rate": 9.747681613852203e-05, "loss": 3.2151, "step": 9250 }, { "epoch": 0.5742752498603265, "grad_norm": 0.5073664246553657, "learning_rate": 9.747568323449984e-05, "loss": 3.1993, "step": 9251 }, { "epoch": 0.5743373269600844, "grad_norm": 0.517955253966097, "learning_rate": 9.74745500827855e-05, "loss": 3.3289, "step": 9252 }, { "epoch": 0.5743994040598424, "grad_norm": 0.5445197204069786, "learning_rate": 9.747341668338492e-05, "loss": 3.2542, "step": 9253 }, { "epoch": 0.5744614811596003, "grad_norm": 0.5088783522617061, "learning_rate": 9.747228303630398e-05, "loss": 3.2402, "step": 9254 }, { "epoch": 0.5745235582593581, "grad_norm": 0.4927953720502862, "learning_rate": 9.747114914154863e-05, "loss": 3.1341, "step": 9255 }, { "epoch": 0.574585635359116, "grad_norm": 0.4139797379895778, "learning_rate": 9.747001499912476e-05, "loss": 3.1637, "step": 9256 }, { "epoch": 0.5746477124588739, "grad_norm": 0.49186712767664376, "learning_rate": 9.746888060903831e-05, "loss": 3.2012, "step": 9257 }, { "epoch": 0.5747097895586318, "grad_norm": 0.4978493092943507, "learning_rate": 9.746774597129517e-05, "loss": 3.1567, "step": 9258 }, { "epoch": 0.5747718666583898, "grad_norm": 0.44546098074130647, "learning_rate": 9.74666110859013e-05, "loss": 3.183, "step": 9259 }, { "epoch": 0.5748339437581477, "grad_norm": 0.5379805556881854, "learning_rate": 9.746547595286257e-05, "loss": 3.2907, "step": 9260 }, { "epoch": 0.5748960208579055, "grad_norm": 0.5459811125849078, "learning_rate": 9.746434057218494e-05, "loss": 3.2361, "step": 9261 }, { "epoch": 0.5749580979576634, "grad_norm": 0.5255391077721168, "learning_rate": 9.746320494387433e-05, "loss": 3.1449, "step": 9262 }, { "epoch": 0.5750201750574213, "grad_norm": 0.46355604478060825, "learning_rate": 9.746206906793664e-05, "loss": 3.0156, "step": 9263 }, { "epoch": 0.5750822521571792, "grad_norm": 0.43779783669530203, "learning_rate": 9.746093294437783e-05, "loss": 3.2066, "step": 9264 }, { "epoch": 0.5751443292569371, "grad_norm": 0.5481353949297328, "learning_rate": 9.74597965732038e-05, "loss": 3.2909, "step": 9265 }, { "epoch": 0.5752064063566951, "grad_norm": 0.5332062270524892, "learning_rate": 9.745865995442051e-05, "loss": 3.2059, "step": 9266 }, { "epoch": 0.5752684834564529, "grad_norm": 0.5111471288568863, "learning_rate": 9.745752308803387e-05, "loss": 3.1252, "step": 9267 }, { "epoch": 0.5753305605562108, "grad_norm": 0.48895499476834536, "learning_rate": 9.74563859740498e-05, "loss": 3.2026, "step": 9268 }, { "epoch": 0.5753926376559687, "grad_norm": 0.527722173474917, "learning_rate": 9.745524861247424e-05, "loss": 3.2134, "step": 9269 }, { "epoch": 0.5754547147557266, "grad_norm": 0.5807554612529013, "learning_rate": 9.745411100331311e-05, "loss": 3.2002, "step": 9270 }, { "epoch": 0.5755167918554845, "grad_norm": 0.5418624860205146, "learning_rate": 9.74529731465724e-05, "loss": 3.1269, "step": 9271 }, { "epoch": 0.5755788689552425, "grad_norm": 0.5214232219600801, "learning_rate": 9.745183504225798e-05, "loss": 3.1557, "step": 9272 }, { "epoch": 0.5756409460550003, "grad_norm": 0.5735951105702806, "learning_rate": 9.745069669037584e-05, "loss": 3.1882, "step": 9273 }, { "epoch": 0.5757030231547582, "grad_norm": 0.5028516097710696, "learning_rate": 9.744955809093187e-05, "loss": 3.3198, "step": 9274 }, { "epoch": 0.5757651002545161, "grad_norm": 0.5411440149545332, "learning_rate": 9.744841924393207e-05, "loss": 3.2255, "step": 9275 }, { "epoch": 0.575827177354274, "grad_norm": 0.5467684778284009, "learning_rate": 9.744728014938233e-05, "loss": 3.2616, "step": 9276 }, { "epoch": 0.5758892544540319, "grad_norm": 0.5934330194514041, "learning_rate": 9.744614080728863e-05, "loss": 3.2511, "step": 9277 }, { "epoch": 0.5759513315537899, "grad_norm": 0.5678864179332833, "learning_rate": 9.744500121765686e-05, "loss": 3.3066, "step": 9278 }, { "epoch": 0.5760134086535477, "grad_norm": 0.573276577231535, "learning_rate": 9.744386138049301e-05, "loss": 3.2071, "step": 9279 }, { "epoch": 0.5760754857533056, "grad_norm": 0.5006189389760826, "learning_rate": 9.744272129580304e-05, "loss": 3.2103, "step": 9280 }, { "epoch": 0.5761375628530635, "grad_norm": 0.5602278649001498, "learning_rate": 9.744158096359286e-05, "loss": 3.164, "step": 9281 }, { "epoch": 0.5761996399528214, "grad_norm": 0.6264044716957982, "learning_rate": 9.744044038386844e-05, "loss": 3.2199, "step": 9282 }, { "epoch": 0.5762617170525793, "grad_norm": 0.5637828714975298, "learning_rate": 9.743929955663574e-05, "loss": 3.0758, "step": 9283 }, { "epoch": 0.5763237941523373, "grad_norm": 0.6035388830943951, "learning_rate": 9.743815848190068e-05, "loss": 3.2695, "step": 9284 }, { "epoch": 0.5763858712520951, "grad_norm": 0.6329914828898106, "learning_rate": 9.743701715966922e-05, "loss": 3.2311, "step": 9285 }, { "epoch": 0.576447948351853, "grad_norm": 0.4718727838072618, "learning_rate": 9.743587558994734e-05, "loss": 3.2687, "step": 9286 }, { "epoch": 0.5765100254516109, "grad_norm": 0.5397444847694463, "learning_rate": 9.743473377274098e-05, "loss": 3.225, "step": 9287 }, { "epoch": 0.5765721025513688, "grad_norm": 0.6287059109437392, "learning_rate": 9.74335917080561e-05, "loss": 3.141, "step": 9288 }, { "epoch": 0.5766341796511267, "grad_norm": 0.60712266034906, "learning_rate": 9.743244939589867e-05, "loss": 3.1731, "step": 9289 }, { "epoch": 0.5766962567508847, "grad_norm": 0.6393737203038927, "learning_rate": 9.743130683627462e-05, "loss": 3.1827, "step": 9290 }, { "epoch": 0.5767583338506425, "grad_norm": 0.6174553639962862, "learning_rate": 9.743016402918992e-05, "loss": 3.18, "step": 9291 }, { "epoch": 0.5768204109504004, "grad_norm": 0.5571756301531255, "learning_rate": 9.742902097465056e-05, "loss": 3.1436, "step": 9292 }, { "epoch": 0.5768824880501583, "grad_norm": 0.5389856200750535, "learning_rate": 9.742787767266247e-05, "loss": 3.2379, "step": 9293 }, { "epoch": 0.5769445651499162, "grad_norm": 0.6382325320754103, "learning_rate": 9.742673412323163e-05, "loss": 3.2533, "step": 9294 }, { "epoch": 0.5770066422496741, "grad_norm": 0.5595210721119025, "learning_rate": 9.742559032636401e-05, "loss": 3.1649, "step": 9295 }, { "epoch": 0.577068719349432, "grad_norm": 0.5104693820226684, "learning_rate": 9.742444628206558e-05, "loss": 3.2228, "step": 9296 }, { "epoch": 0.5771307964491899, "grad_norm": 0.5161195464101314, "learning_rate": 9.742330199034228e-05, "loss": 3.209, "step": 9297 }, { "epoch": 0.5771928735489478, "grad_norm": 0.5096117276972654, "learning_rate": 9.742215745120011e-05, "loss": 3.1254, "step": 9298 }, { "epoch": 0.5772549506487057, "grad_norm": 0.464470384820888, "learning_rate": 9.742101266464505e-05, "loss": 3.1519, "step": 9299 }, { "epoch": 0.5773170277484636, "grad_norm": 0.5139907601338674, "learning_rate": 9.741986763068303e-05, "loss": 3.1451, "step": 9300 }, { "epoch": 0.5773791048482215, "grad_norm": 0.5248788659811219, "learning_rate": 9.741872234932006e-05, "loss": 3.1408, "step": 9301 }, { "epoch": 0.5774411819479794, "grad_norm": 0.5058443516448246, "learning_rate": 9.741757682056212e-05, "loss": 3.2726, "step": 9302 }, { "epoch": 0.5775032590477372, "grad_norm": 0.4625430029131093, "learning_rate": 9.741643104441514e-05, "loss": 3.2459, "step": 9303 }, { "epoch": 0.5775653361474952, "grad_norm": 0.4710052596910155, "learning_rate": 9.741528502088515e-05, "loss": 3.2697, "step": 9304 }, { "epoch": 0.5776274132472531, "grad_norm": 0.5166130136760294, "learning_rate": 9.741413874997809e-05, "loss": 3.2701, "step": 9305 }, { "epoch": 0.577689490347011, "grad_norm": 0.5405601760279741, "learning_rate": 9.741299223169997e-05, "loss": 3.2994, "step": 9306 }, { "epoch": 0.5777515674467689, "grad_norm": 0.5393354305459181, "learning_rate": 9.741184546605678e-05, "loss": 3.2317, "step": 9307 }, { "epoch": 0.5778136445465268, "grad_norm": 0.45847236807540076, "learning_rate": 9.741069845305444e-05, "loss": 3.1866, "step": 9308 }, { "epoch": 0.5778757216462846, "grad_norm": 0.44573765252363545, "learning_rate": 9.740955119269902e-05, "loss": 3.2185, "step": 9309 }, { "epoch": 0.5779377987460426, "grad_norm": 0.4846074869210427, "learning_rate": 9.740840368499641e-05, "loss": 3.17, "step": 9310 }, { "epoch": 0.5779998758458005, "grad_norm": 0.47233105341144127, "learning_rate": 9.74072559299527e-05, "loss": 3.2138, "step": 9311 }, { "epoch": 0.5780619529455584, "grad_norm": 0.4267040541809619, "learning_rate": 9.74061079275738e-05, "loss": 3.1012, "step": 9312 }, { "epoch": 0.5781240300453163, "grad_norm": 0.47676015100839353, "learning_rate": 9.740495967786573e-05, "loss": 3.1799, "step": 9313 }, { "epoch": 0.5781861071450742, "grad_norm": 0.46764921637438545, "learning_rate": 9.740381118083449e-05, "loss": 3.2861, "step": 9314 }, { "epoch": 0.578248184244832, "grad_norm": 0.4822462743019006, "learning_rate": 9.740266243648604e-05, "loss": 3.2038, "step": 9315 }, { "epoch": 0.57831026134459, "grad_norm": 0.43878608093362614, "learning_rate": 9.740151344482641e-05, "loss": 3.2138, "step": 9316 }, { "epoch": 0.5783723384443479, "grad_norm": 0.4829632890899173, "learning_rate": 9.740036420586156e-05, "loss": 3.2186, "step": 9317 }, { "epoch": 0.5784344155441058, "grad_norm": 0.5166520380697437, "learning_rate": 9.739921471959752e-05, "loss": 3.2366, "step": 9318 }, { "epoch": 0.5784964926438637, "grad_norm": 0.43690723043803387, "learning_rate": 9.739806498604026e-05, "loss": 3.1457, "step": 9319 }, { "epoch": 0.5785585697436216, "grad_norm": 0.45305868189662274, "learning_rate": 9.739691500519579e-05, "loss": 3.1927, "step": 9320 }, { "epoch": 0.5786206468433794, "grad_norm": 0.5386682257137934, "learning_rate": 9.739576477707011e-05, "loss": 3.2646, "step": 9321 }, { "epoch": 0.5786827239431374, "grad_norm": 0.6009839731577138, "learning_rate": 9.739461430166923e-05, "loss": 3.1975, "step": 9322 }, { "epoch": 0.5787448010428953, "grad_norm": 0.5608028133282313, "learning_rate": 9.739346357899912e-05, "loss": 3.2053, "step": 9323 }, { "epoch": 0.5788068781426532, "grad_norm": 0.4763269467657557, "learning_rate": 9.739231260906582e-05, "loss": 3.1582, "step": 9324 }, { "epoch": 0.5788689552424111, "grad_norm": 0.49153170720176814, "learning_rate": 9.739116139187532e-05, "loss": 3.3101, "step": 9325 }, { "epoch": 0.578931032342169, "grad_norm": 0.5177184292623098, "learning_rate": 9.739000992743363e-05, "loss": 3.1638, "step": 9326 }, { "epoch": 0.5789931094419268, "grad_norm": 0.4644589955541997, "learning_rate": 9.738885821574674e-05, "loss": 3.1718, "step": 9327 }, { "epoch": 0.5790551865416848, "grad_norm": 0.4977952508518235, "learning_rate": 9.738770625682068e-05, "loss": 3.2238, "step": 9328 }, { "epoch": 0.5791172636414427, "grad_norm": 0.6401808292138568, "learning_rate": 9.738655405066146e-05, "loss": 3.0692, "step": 9329 }, { "epoch": 0.5791793407412006, "grad_norm": 0.5411931376729122, "learning_rate": 9.738540159727508e-05, "loss": 3.2613, "step": 9330 }, { "epoch": 0.5792414178409585, "grad_norm": 0.4782504690011452, "learning_rate": 9.738424889666755e-05, "loss": 3.0335, "step": 9331 }, { "epoch": 0.5793034949407164, "grad_norm": 0.5250809050845252, "learning_rate": 9.73830959488449e-05, "loss": 3.1272, "step": 9332 }, { "epoch": 0.5793655720404742, "grad_norm": 0.5014179211497748, "learning_rate": 9.738194275381312e-05, "loss": 3.166, "step": 9333 }, { "epoch": 0.5794276491402321, "grad_norm": 0.5814436692434712, "learning_rate": 9.738078931157826e-05, "loss": 3.2671, "step": 9334 }, { "epoch": 0.5794897262399901, "grad_norm": 0.5899797750093814, "learning_rate": 9.73796356221463e-05, "loss": 3.2234, "step": 9335 }, { "epoch": 0.579551803339748, "grad_norm": 0.5377294394474295, "learning_rate": 9.73784816855233e-05, "loss": 3.189, "step": 9336 }, { "epoch": 0.5796138804395059, "grad_norm": 0.47059669716016855, "learning_rate": 9.737732750171524e-05, "loss": 3.1737, "step": 9337 }, { "epoch": 0.5796759575392638, "grad_norm": 0.4992548784725363, "learning_rate": 9.737617307072817e-05, "loss": 3.1894, "step": 9338 }, { "epoch": 0.5797380346390216, "grad_norm": 0.5284590054314529, "learning_rate": 9.737501839256812e-05, "loss": 3.2082, "step": 9339 }, { "epoch": 0.5798001117387795, "grad_norm": 0.5358210701327701, "learning_rate": 9.737386346724109e-05, "loss": 3.2088, "step": 9340 }, { "epoch": 0.5798621888385375, "grad_norm": 0.5551547379477149, "learning_rate": 9.73727082947531e-05, "loss": 3.1547, "step": 9341 }, { "epoch": 0.5799242659382954, "grad_norm": 0.4377075111905559, "learning_rate": 9.73715528751102e-05, "loss": 3.1437, "step": 9342 }, { "epoch": 0.5799863430380533, "grad_norm": 0.49610686097128787, "learning_rate": 9.73703972083184e-05, "loss": 3.2408, "step": 9343 }, { "epoch": 0.5800484201378112, "grad_norm": 0.48843842074941396, "learning_rate": 9.736924129438375e-05, "loss": 3.217, "step": 9344 }, { "epoch": 0.580110497237569, "grad_norm": 0.4588784927634268, "learning_rate": 9.736808513331227e-05, "loss": 3.1912, "step": 9345 }, { "epoch": 0.5801725743373269, "grad_norm": 0.4961222692987784, "learning_rate": 9.736692872510999e-05, "loss": 3.2013, "step": 9346 }, { "epoch": 0.5802346514370849, "grad_norm": 0.4918175927025871, "learning_rate": 9.736577206978293e-05, "loss": 3.1748, "step": 9347 }, { "epoch": 0.5802967285368428, "grad_norm": 0.5185956920783954, "learning_rate": 9.736461516733715e-05, "loss": 3.1494, "step": 9348 }, { "epoch": 0.5803588056366007, "grad_norm": 0.5014012490484065, "learning_rate": 9.736345801777867e-05, "loss": 3.1926, "step": 9349 }, { "epoch": 0.5804208827363586, "grad_norm": 0.5546605009573427, "learning_rate": 9.736230062111353e-05, "loss": 3.2579, "step": 9350 }, { "epoch": 0.5804829598361164, "grad_norm": 0.5684744597127233, "learning_rate": 9.736114297734778e-05, "loss": 3.1967, "step": 9351 }, { "epoch": 0.5805450369358743, "grad_norm": 0.5944073200764708, "learning_rate": 9.735998508648745e-05, "loss": 3.148, "step": 9352 }, { "epoch": 0.5806071140356323, "grad_norm": 0.5620393191138274, "learning_rate": 9.735882694853858e-05, "loss": 3.2654, "step": 9353 }, { "epoch": 0.5806691911353902, "grad_norm": 0.5588650078150069, "learning_rate": 9.73576685635072e-05, "loss": 3.2557, "step": 9354 }, { "epoch": 0.5807312682351481, "grad_norm": 0.525115724706425, "learning_rate": 9.735650993139937e-05, "loss": 3.2997, "step": 9355 }, { "epoch": 0.580793345334906, "grad_norm": 0.5402694955527133, "learning_rate": 9.735535105222114e-05, "loss": 3.1846, "step": 9356 }, { "epoch": 0.5808554224346638, "grad_norm": 0.5070975332327642, "learning_rate": 9.735419192597854e-05, "loss": 3.1908, "step": 9357 }, { "epoch": 0.5809174995344217, "grad_norm": 0.568897189492753, "learning_rate": 9.735303255267765e-05, "loss": 3.2199, "step": 9358 }, { "epoch": 0.5809795766341797, "grad_norm": 0.5064356404656049, "learning_rate": 9.735187293232446e-05, "loss": 3.2217, "step": 9359 }, { "epoch": 0.5810416537339376, "grad_norm": 0.4533521313165026, "learning_rate": 9.735071306492507e-05, "loss": 3.2258, "step": 9360 }, { "epoch": 0.5811037308336955, "grad_norm": 0.5202042945406622, "learning_rate": 9.734955295048551e-05, "loss": 3.2597, "step": 9361 }, { "epoch": 0.5811658079334534, "grad_norm": 0.5070962402824234, "learning_rate": 9.734839258901185e-05, "loss": 3.158, "step": 9362 }, { "epoch": 0.5812278850332112, "grad_norm": 0.5104264109193903, "learning_rate": 9.734723198051011e-05, "loss": 3.2487, "step": 9363 }, { "epoch": 0.5812899621329691, "grad_norm": 0.44512816847027303, "learning_rate": 9.734607112498638e-05, "loss": 3.2008, "step": 9364 }, { "epoch": 0.581352039232727, "grad_norm": 0.5167402575238393, "learning_rate": 9.73449100224467e-05, "loss": 3.2887, "step": 9365 }, { "epoch": 0.581414116332485, "grad_norm": 0.44291350341425695, "learning_rate": 9.734374867289713e-05, "loss": 3.2722, "step": 9366 }, { "epoch": 0.5814761934322429, "grad_norm": 0.4476164229220275, "learning_rate": 9.734258707634373e-05, "loss": 3.172, "step": 9367 }, { "epoch": 0.5815382705320007, "grad_norm": 0.47238564429463104, "learning_rate": 9.734142523279257e-05, "loss": 3.2351, "step": 9368 }, { "epoch": 0.5816003476317586, "grad_norm": 0.4530039658530743, "learning_rate": 9.73402631422497e-05, "loss": 3.1843, "step": 9369 }, { "epoch": 0.5816624247315165, "grad_norm": 0.5237206220012373, "learning_rate": 9.733910080472118e-05, "loss": 3.2719, "step": 9370 }, { "epoch": 0.5817245018312744, "grad_norm": 0.5918817490753829, "learning_rate": 9.733793822021307e-05, "loss": 3.1973, "step": 9371 }, { "epoch": 0.5817865789310324, "grad_norm": 0.5317060252372883, "learning_rate": 9.733677538873146e-05, "loss": 3.1904, "step": 9372 }, { "epoch": 0.5818486560307903, "grad_norm": 0.48040991322098364, "learning_rate": 9.73356123102824e-05, "loss": 3.3132, "step": 9373 }, { "epoch": 0.5819107331305481, "grad_norm": 0.5164170395363128, "learning_rate": 9.733444898487195e-05, "loss": 3.2606, "step": 9374 }, { "epoch": 0.581972810230306, "grad_norm": 0.46108770347775013, "learning_rate": 9.73332854125062e-05, "loss": 3.1604, "step": 9375 }, { "epoch": 0.5820348873300639, "grad_norm": 0.48643660963129665, "learning_rate": 9.73321215931912e-05, "loss": 3.1766, "step": 9376 }, { "epoch": 0.5820969644298218, "grad_norm": 0.5645123091032284, "learning_rate": 9.733095752693303e-05, "loss": 3.1767, "step": 9377 }, { "epoch": 0.5821590415295798, "grad_norm": 0.5945043540214442, "learning_rate": 9.732979321373776e-05, "loss": 3.2316, "step": 9378 }, { "epoch": 0.5822211186293377, "grad_norm": 0.6277514215040805, "learning_rate": 9.73286286536115e-05, "loss": 3.1479, "step": 9379 }, { "epoch": 0.5822831957290955, "grad_norm": 0.579944736544508, "learning_rate": 9.732746384656027e-05, "loss": 3.2315, "step": 9380 }, { "epoch": 0.5823452728288534, "grad_norm": 0.47397225397927095, "learning_rate": 9.732629879259017e-05, "loss": 3.23, "step": 9381 }, { "epoch": 0.5824073499286113, "grad_norm": 0.529731150041087, "learning_rate": 9.732513349170729e-05, "loss": 3.181, "step": 9382 }, { "epoch": 0.5824694270283692, "grad_norm": 0.47430963038190826, "learning_rate": 9.732396794391769e-05, "loss": 3.1869, "step": 9383 }, { "epoch": 0.5825315041281272, "grad_norm": 0.5313398902240002, "learning_rate": 9.732280214922747e-05, "loss": 3.1843, "step": 9384 }, { "epoch": 0.5825935812278851, "grad_norm": 0.47845721764176835, "learning_rate": 9.732163610764269e-05, "loss": 3.1019, "step": 9385 }, { "epoch": 0.5826556583276429, "grad_norm": 0.5008074082611198, "learning_rate": 9.732046981916946e-05, "loss": 3.1872, "step": 9386 }, { "epoch": 0.5827177354274008, "grad_norm": 0.5147100190990166, "learning_rate": 9.731930328381384e-05, "loss": 3.2052, "step": 9387 }, { "epoch": 0.5827798125271587, "grad_norm": 0.5447074167327572, "learning_rate": 9.731813650158195e-05, "loss": 3.2336, "step": 9388 }, { "epoch": 0.5828418896269166, "grad_norm": 0.6153443979855898, "learning_rate": 9.731696947247984e-05, "loss": 3.1448, "step": 9389 }, { "epoch": 0.5829039667266745, "grad_norm": 0.615316209951427, "learning_rate": 9.73158021965136e-05, "loss": 3.2234, "step": 9390 }, { "epoch": 0.5829660438264325, "grad_norm": 0.5518839464078096, "learning_rate": 9.731463467368935e-05, "loss": 3.2077, "step": 9391 }, { "epoch": 0.5830281209261903, "grad_norm": 0.4729189024744944, "learning_rate": 9.731346690401317e-05, "loss": 3.1831, "step": 9392 }, { "epoch": 0.5830901980259482, "grad_norm": 0.5106145287120598, "learning_rate": 9.731229888749114e-05, "loss": 3.2665, "step": 9393 }, { "epoch": 0.5831522751257061, "grad_norm": 0.5322500521174701, "learning_rate": 9.731113062412936e-05, "loss": 3.2172, "step": 9394 }, { "epoch": 0.583214352225464, "grad_norm": 0.418037533670853, "learning_rate": 9.730996211393393e-05, "loss": 3.2562, "step": 9395 }, { "epoch": 0.583276429325222, "grad_norm": 0.5268049808100909, "learning_rate": 9.730879335691094e-05, "loss": 3.1659, "step": 9396 }, { "epoch": 0.5833385064249799, "grad_norm": 0.4827784868870322, "learning_rate": 9.730762435306649e-05, "loss": 3.1874, "step": 9397 }, { "epoch": 0.5834005835247377, "grad_norm": 0.530976309118503, "learning_rate": 9.730645510240668e-05, "loss": 3.2672, "step": 9398 }, { "epoch": 0.5834626606244956, "grad_norm": 0.5192616559145161, "learning_rate": 9.730528560493761e-05, "loss": 3.2642, "step": 9399 }, { "epoch": 0.5835247377242535, "grad_norm": 0.5074325901042507, "learning_rate": 9.730411586066538e-05, "loss": 3.164, "step": 9400 }, { "epoch": 0.5835868148240114, "grad_norm": 0.44536352701797577, "learning_rate": 9.730294586959609e-05, "loss": 3.1935, "step": 9401 }, { "epoch": 0.5836488919237693, "grad_norm": 0.44727238812393205, "learning_rate": 9.730177563173586e-05, "loss": 3.1624, "step": 9402 }, { "epoch": 0.5837109690235273, "grad_norm": 0.48431351486790886, "learning_rate": 9.730060514709077e-05, "loss": 3.162, "step": 9403 }, { "epoch": 0.5837730461232851, "grad_norm": 0.5139435024567377, "learning_rate": 9.729943441566696e-05, "loss": 3.2448, "step": 9404 }, { "epoch": 0.583835123223043, "grad_norm": 0.4880176748324859, "learning_rate": 9.729826343747049e-05, "loss": 3.143, "step": 9405 }, { "epoch": 0.5838972003228009, "grad_norm": 0.4411121671576848, "learning_rate": 9.729709221250751e-05, "loss": 3.253, "step": 9406 }, { "epoch": 0.5839592774225588, "grad_norm": 0.49119364592375164, "learning_rate": 9.729592074078414e-05, "loss": 3.1342, "step": 9407 }, { "epoch": 0.5840213545223167, "grad_norm": 0.49344346091769103, "learning_rate": 9.729474902230644e-05, "loss": 3.1518, "step": 9408 }, { "epoch": 0.5840834316220747, "grad_norm": 0.4855149759630719, "learning_rate": 9.729357705708056e-05, "loss": 3.1392, "step": 9409 }, { "epoch": 0.5841455087218325, "grad_norm": 0.4173098566880837, "learning_rate": 9.729240484511262e-05, "loss": 3.0882, "step": 9410 }, { "epoch": 0.5842075858215904, "grad_norm": 0.45514372324181224, "learning_rate": 9.729123238640871e-05, "loss": 3.1937, "step": 9411 }, { "epoch": 0.5842696629213483, "grad_norm": 0.470806094734863, "learning_rate": 9.729005968097498e-05, "loss": 3.2807, "step": 9412 }, { "epoch": 0.5843317400211062, "grad_norm": 0.46647382275464344, "learning_rate": 9.728888672881751e-05, "loss": 3.2475, "step": 9413 }, { "epoch": 0.5843938171208641, "grad_norm": 0.496308708177394, "learning_rate": 9.728771352994244e-05, "loss": 3.1787, "step": 9414 }, { "epoch": 0.584455894220622, "grad_norm": 0.496287766011664, "learning_rate": 9.728654008435589e-05, "loss": 3.2489, "step": 9415 }, { "epoch": 0.5845179713203799, "grad_norm": 0.5127305372944023, "learning_rate": 9.728536639206399e-05, "loss": 3.1707, "step": 9416 }, { "epoch": 0.5845800484201378, "grad_norm": 0.4744515647215543, "learning_rate": 9.728419245307283e-05, "loss": 3.1684, "step": 9417 }, { "epoch": 0.5846421255198957, "grad_norm": 0.46343906924495876, "learning_rate": 9.728301826738858e-05, "loss": 3.1897, "step": 9418 }, { "epoch": 0.5847042026196536, "grad_norm": 0.5672727862253998, "learning_rate": 9.728184383501736e-05, "loss": 3.1657, "step": 9419 }, { "epoch": 0.5847662797194115, "grad_norm": 0.5242400209116542, "learning_rate": 9.728066915596525e-05, "loss": 3.1455, "step": 9420 }, { "epoch": 0.5848283568191694, "grad_norm": 0.493496803358367, "learning_rate": 9.727949423023843e-05, "loss": 3.2083, "step": 9421 }, { "epoch": 0.5848904339189273, "grad_norm": 0.5033527557728833, "learning_rate": 9.7278319057843e-05, "loss": 3.2267, "step": 9422 }, { "epoch": 0.5849525110186852, "grad_norm": 0.5384793928678703, "learning_rate": 9.727714363878511e-05, "loss": 3.1771, "step": 9423 }, { "epoch": 0.5850145881184431, "grad_norm": 0.553926160317337, "learning_rate": 9.727596797307089e-05, "loss": 3.2086, "step": 9424 }, { "epoch": 0.585076665218201, "grad_norm": 0.5206972389432241, "learning_rate": 9.727479206070646e-05, "loss": 3.1583, "step": 9425 }, { "epoch": 0.5851387423179589, "grad_norm": 0.45440512734378025, "learning_rate": 9.727361590169797e-05, "loss": 3.1559, "step": 9426 }, { "epoch": 0.5852008194177168, "grad_norm": 0.4798186486354467, "learning_rate": 9.727243949605153e-05, "loss": 3.1926, "step": 9427 }, { "epoch": 0.5852628965174747, "grad_norm": 0.467017149726737, "learning_rate": 9.727126284377333e-05, "loss": 3.1954, "step": 9428 }, { "epoch": 0.5853249736172326, "grad_norm": 0.4374053546165682, "learning_rate": 9.727008594486945e-05, "loss": 3.0857, "step": 9429 }, { "epoch": 0.5853870507169905, "grad_norm": 0.4383305091851541, "learning_rate": 9.726890879934607e-05, "loss": 3.0404, "step": 9430 }, { "epoch": 0.5854491278167484, "grad_norm": 0.44759291954845054, "learning_rate": 9.726773140720931e-05, "loss": 3.2308, "step": 9431 }, { "epoch": 0.5855112049165063, "grad_norm": 0.5185102885576464, "learning_rate": 9.726655376846531e-05, "loss": 3.2266, "step": 9432 }, { "epoch": 0.5855732820162642, "grad_norm": 0.5220217846694809, "learning_rate": 9.726537588312025e-05, "loss": 3.1929, "step": 9433 }, { "epoch": 0.585635359116022, "grad_norm": 0.4916326634992807, "learning_rate": 9.726419775118024e-05, "loss": 3.1399, "step": 9434 }, { "epoch": 0.58569743621578, "grad_norm": 0.5670899835411966, "learning_rate": 9.726301937265143e-05, "loss": 3.1966, "step": 9435 }, { "epoch": 0.5857595133155379, "grad_norm": 0.5848232610376367, "learning_rate": 9.726184074753999e-05, "loss": 3.2276, "step": 9436 }, { "epoch": 0.5858215904152958, "grad_norm": 0.5319684163771348, "learning_rate": 9.726066187585203e-05, "loss": 3.1757, "step": 9437 }, { "epoch": 0.5858836675150537, "grad_norm": 0.5282369325739347, "learning_rate": 9.725948275759375e-05, "loss": 3.224, "step": 9438 }, { "epoch": 0.5859457446148116, "grad_norm": 0.5321009776486579, "learning_rate": 9.725830339277127e-05, "loss": 3.1685, "step": 9439 }, { "epoch": 0.5860078217145694, "grad_norm": 0.49234941634849305, "learning_rate": 9.725712378139073e-05, "loss": 3.1541, "step": 9440 }, { "epoch": 0.5860698988143274, "grad_norm": 0.5024274670267892, "learning_rate": 9.725594392345832e-05, "loss": 3.2631, "step": 9441 }, { "epoch": 0.5861319759140853, "grad_norm": 0.6271244183469785, "learning_rate": 9.725476381898018e-05, "loss": 3.2672, "step": 9442 }, { "epoch": 0.5861940530138432, "grad_norm": 0.5675958440315975, "learning_rate": 9.725358346796245e-05, "loss": 3.2332, "step": 9443 }, { "epoch": 0.5862561301136011, "grad_norm": 0.5756156434780481, "learning_rate": 9.72524028704113e-05, "loss": 3.2433, "step": 9444 }, { "epoch": 0.586318207213359, "grad_norm": 0.6178784426207952, "learning_rate": 9.725122202633293e-05, "loss": 3.1574, "step": 9445 }, { "epoch": 0.5863802843131168, "grad_norm": 0.6699745001851077, "learning_rate": 9.725004093573342e-05, "loss": 3.2145, "step": 9446 }, { "epoch": 0.5864423614128748, "grad_norm": 0.5986803792005739, "learning_rate": 9.7248859598619e-05, "loss": 3.1811, "step": 9447 }, { "epoch": 0.5865044385126327, "grad_norm": 0.692666301073517, "learning_rate": 9.724767801499582e-05, "loss": 3.1657, "step": 9448 }, { "epoch": 0.5865665156123906, "grad_norm": 0.6309804416382095, "learning_rate": 9.724649618487e-05, "loss": 3.2187, "step": 9449 }, { "epoch": 0.5866285927121485, "grad_norm": 0.6662052813009992, "learning_rate": 9.724531410824778e-05, "loss": 3.1868, "step": 9450 }, { "epoch": 0.5866906698119064, "grad_norm": 0.707147114570264, "learning_rate": 9.724413178513526e-05, "loss": 3.2726, "step": 9451 }, { "epoch": 0.5867527469116642, "grad_norm": 0.5790601332771038, "learning_rate": 9.724294921553865e-05, "loss": 3.2343, "step": 9452 }, { "epoch": 0.5868148240114222, "grad_norm": 0.5504158106308007, "learning_rate": 9.72417663994641e-05, "loss": 3.1243, "step": 9453 }, { "epoch": 0.5868769011111801, "grad_norm": 0.464423125559857, "learning_rate": 9.724058333691776e-05, "loss": 3.1619, "step": 9454 }, { "epoch": 0.586938978210938, "grad_norm": 0.5506343239764456, "learning_rate": 9.723940002790587e-05, "loss": 3.1726, "step": 9455 }, { "epoch": 0.5870010553106959, "grad_norm": 0.6065267855208542, "learning_rate": 9.723821647243453e-05, "loss": 3.2323, "step": 9456 }, { "epoch": 0.5870631324104538, "grad_norm": 0.5068895315305069, "learning_rate": 9.723703267050998e-05, "loss": 3.1784, "step": 9457 }, { "epoch": 0.5871252095102116, "grad_norm": 0.4883641287084541, "learning_rate": 9.723584862213835e-05, "loss": 3.1436, "step": 9458 }, { "epoch": 0.5871872866099695, "grad_norm": 0.4965598552482183, "learning_rate": 9.723466432732581e-05, "loss": 3.1055, "step": 9459 }, { "epoch": 0.5872493637097275, "grad_norm": 0.5097422101135123, "learning_rate": 9.723347978607858e-05, "loss": 3.1307, "step": 9460 }, { "epoch": 0.5873114408094854, "grad_norm": 0.5339699844396983, "learning_rate": 9.723229499840283e-05, "loss": 3.213, "step": 9461 }, { "epoch": 0.5873735179092433, "grad_norm": 0.537794418429985, "learning_rate": 9.723110996430472e-05, "loss": 3.2198, "step": 9462 }, { "epoch": 0.5874355950090012, "grad_norm": 0.5158699205341212, "learning_rate": 9.722992468379043e-05, "loss": 3.1878, "step": 9463 }, { "epoch": 0.587497672108759, "grad_norm": 0.5492035745664849, "learning_rate": 9.722873915686616e-05, "loss": 3.0842, "step": 9464 }, { "epoch": 0.587559749208517, "grad_norm": 0.5663232676786746, "learning_rate": 9.72275533835381e-05, "loss": 3.1718, "step": 9465 }, { "epoch": 0.5876218263082749, "grad_norm": 0.5703496267697236, "learning_rate": 9.722636736381243e-05, "loss": 3.124, "step": 9466 }, { "epoch": 0.5876839034080328, "grad_norm": 0.5289115206795318, "learning_rate": 9.722518109769534e-05, "loss": 3.2371, "step": 9467 }, { "epoch": 0.5877459805077907, "grad_norm": 0.5178450274107271, "learning_rate": 9.722399458519303e-05, "loss": 3.1233, "step": 9468 }, { "epoch": 0.5878080576075486, "grad_norm": 0.5821613703959231, "learning_rate": 9.722280782631165e-05, "loss": 3.2063, "step": 9469 }, { "epoch": 0.5878701347073064, "grad_norm": 0.6114759418826688, "learning_rate": 9.722162082105743e-05, "loss": 3.1635, "step": 9470 }, { "epoch": 0.5879322118070643, "grad_norm": 0.5354567749002341, "learning_rate": 9.722043356943653e-05, "loss": 3.218, "step": 9471 }, { "epoch": 0.5879942889068223, "grad_norm": 0.5948884990945618, "learning_rate": 9.72192460714552e-05, "loss": 3.1739, "step": 9472 }, { "epoch": 0.5880563660065802, "grad_norm": 0.656560120648626, "learning_rate": 9.721805832711957e-05, "loss": 3.1774, "step": 9473 }, { "epoch": 0.5881184431063381, "grad_norm": 0.642328439918405, "learning_rate": 9.721687033643588e-05, "loss": 3.1686, "step": 9474 }, { "epoch": 0.588180520206096, "grad_norm": 0.509969064157603, "learning_rate": 9.721568209941032e-05, "loss": 3.1422, "step": 9475 }, { "epoch": 0.5882425973058538, "grad_norm": 0.504558673159441, "learning_rate": 9.721449361604909e-05, "loss": 3.1864, "step": 9476 }, { "epoch": 0.5883046744056117, "grad_norm": 0.6476948863470319, "learning_rate": 9.721330488635835e-05, "loss": 3.2418, "step": 9477 }, { "epoch": 0.5883667515053697, "grad_norm": 0.6263008170534737, "learning_rate": 9.721211591034438e-05, "loss": 3.2087, "step": 9478 }, { "epoch": 0.5884288286051276, "grad_norm": 0.4981219216446664, "learning_rate": 9.721092668801331e-05, "loss": 3.185, "step": 9479 }, { "epoch": 0.5884909057048855, "grad_norm": 0.6063360585746875, "learning_rate": 9.720973721937139e-05, "loss": 3.1122, "step": 9480 }, { "epoch": 0.5885529828046434, "grad_norm": 0.5781793053355255, "learning_rate": 9.720854750442481e-05, "loss": 3.1773, "step": 9481 }, { "epoch": 0.5886150599044012, "grad_norm": 0.5479028851138101, "learning_rate": 9.720735754317978e-05, "loss": 3.252, "step": 9482 }, { "epoch": 0.5886771370041591, "grad_norm": 0.4910364466753738, "learning_rate": 9.72061673356425e-05, "loss": 3.1923, "step": 9483 }, { "epoch": 0.588739214103917, "grad_norm": 0.5325375829337667, "learning_rate": 9.720497688181917e-05, "loss": 3.2543, "step": 9484 }, { "epoch": 0.588801291203675, "grad_norm": 0.49468597655263563, "learning_rate": 9.720378618171604e-05, "loss": 3.2426, "step": 9485 }, { "epoch": 0.5888633683034329, "grad_norm": 0.47110704052145447, "learning_rate": 9.720259523533928e-05, "loss": 3.0987, "step": 9486 }, { "epoch": 0.5889254454031908, "grad_norm": 0.5287016727916733, "learning_rate": 9.720140404269513e-05, "loss": 3.2004, "step": 9487 }, { "epoch": 0.5889875225029486, "grad_norm": 0.5496744438646386, "learning_rate": 9.72002126037898e-05, "loss": 3.0922, "step": 9488 }, { "epoch": 0.5890495996027065, "grad_norm": 0.46982551214547175, "learning_rate": 9.719902091862952e-05, "loss": 3.0632, "step": 9489 }, { "epoch": 0.5891116767024644, "grad_norm": 0.515688950011966, "learning_rate": 9.719782898722045e-05, "loss": 3.1109, "step": 9490 }, { "epoch": 0.5891737538022224, "grad_norm": 0.5630907294697648, "learning_rate": 9.719663680956887e-05, "loss": 3.1556, "step": 9491 }, { "epoch": 0.5892358309019803, "grad_norm": 0.5165805343646049, "learning_rate": 9.719544438568098e-05, "loss": 3.1606, "step": 9492 }, { "epoch": 0.5892979080017382, "grad_norm": 0.523816896413549, "learning_rate": 9.7194251715563e-05, "loss": 3.1436, "step": 9493 }, { "epoch": 0.589359985101496, "grad_norm": 0.5271986585158617, "learning_rate": 9.719305879922115e-05, "loss": 3.3376, "step": 9494 }, { "epoch": 0.5894220622012539, "grad_norm": 0.5596642007669301, "learning_rate": 9.719186563666165e-05, "loss": 3.2749, "step": 9495 }, { "epoch": 0.5894841393010118, "grad_norm": 0.6017704641367333, "learning_rate": 9.719067222789074e-05, "loss": 3.0849, "step": 9496 }, { "epoch": 0.5895462164007698, "grad_norm": 0.5666955669363641, "learning_rate": 9.718947857291462e-05, "loss": 3.0815, "step": 9497 }, { "epoch": 0.5896082935005277, "grad_norm": 0.47302061806596657, "learning_rate": 9.718828467173956e-05, "loss": 3.2265, "step": 9498 }, { "epoch": 0.5896703706002856, "grad_norm": 0.584243181222309, "learning_rate": 9.718709052437175e-05, "loss": 3.2743, "step": 9499 }, { "epoch": 0.5897324477000434, "grad_norm": 0.5541716312482448, "learning_rate": 9.718589613081743e-05, "loss": 3.0911, "step": 9500 }, { "epoch": 0.5897945247998013, "grad_norm": 0.5270512087334113, "learning_rate": 9.718470149108286e-05, "loss": 3.2005, "step": 9501 }, { "epoch": 0.5898566018995592, "grad_norm": 0.6048750396810488, "learning_rate": 9.718350660517421e-05, "loss": 3.1402, "step": 9502 }, { "epoch": 0.5899186789993172, "grad_norm": 0.520581009461429, "learning_rate": 9.718231147309778e-05, "loss": 3.2189, "step": 9503 }, { "epoch": 0.5899807560990751, "grad_norm": 0.537723481493168, "learning_rate": 9.718111609485976e-05, "loss": 3.2102, "step": 9504 }, { "epoch": 0.590042833198833, "grad_norm": 0.5814169643924945, "learning_rate": 9.717992047046641e-05, "loss": 3.1494, "step": 9505 }, { "epoch": 0.5901049102985908, "grad_norm": 0.5288127641874446, "learning_rate": 9.717872459992396e-05, "loss": 3.11, "step": 9506 }, { "epoch": 0.5901669873983487, "grad_norm": 0.5476996879383819, "learning_rate": 9.717752848323864e-05, "loss": 3.2471, "step": 9507 }, { "epoch": 0.5902290644981066, "grad_norm": 0.5083895412491787, "learning_rate": 9.717633212041672e-05, "loss": 3.1405, "step": 9508 }, { "epoch": 0.5902911415978646, "grad_norm": 0.5244508868879605, "learning_rate": 9.717513551146443e-05, "loss": 3.0857, "step": 9509 }, { "epoch": 0.5903532186976225, "grad_norm": 0.4986475743978639, "learning_rate": 9.717393865638798e-05, "loss": 3.2504, "step": 9510 }, { "epoch": 0.5904152957973804, "grad_norm": 0.46359032650549103, "learning_rate": 9.717274155519365e-05, "loss": 3.1491, "step": 9511 }, { "epoch": 0.5904773728971382, "grad_norm": 0.4514134573567399, "learning_rate": 9.717154420788768e-05, "loss": 3.1864, "step": 9512 }, { "epoch": 0.5905394499968961, "grad_norm": 0.5034900987308835, "learning_rate": 9.717034661447632e-05, "loss": 3.1552, "step": 9513 }, { "epoch": 0.590601527096654, "grad_norm": 0.5203789413691412, "learning_rate": 9.71691487749658e-05, "loss": 3.1932, "step": 9514 }, { "epoch": 0.590663604196412, "grad_norm": 0.5070850149840924, "learning_rate": 9.716795068936237e-05, "loss": 3.228, "step": 9515 }, { "epoch": 0.5907256812961699, "grad_norm": 0.4710952581631383, "learning_rate": 9.716675235767231e-05, "loss": 3.1835, "step": 9516 }, { "epoch": 0.5907877583959278, "grad_norm": 0.47259262539473057, "learning_rate": 9.716555377990184e-05, "loss": 3.2021, "step": 9517 }, { "epoch": 0.5908498354956856, "grad_norm": 0.468043860392508, "learning_rate": 9.716435495605723e-05, "loss": 3.1007, "step": 9518 }, { "epoch": 0.5909119125954435, "grad_norm": 0.5629542424018479, "learning_rate": 9.716315588614474e-05, "loss": 3.192, "step": 9519 }, { "epoch": 0.5909739896952014, "grad_norm": 0.4974180403007849, "learning_rate": 9.71619565701706e-05, "loss": 3.1796, "step": 9520 }, { "epoch": 0.5910360667949593, "grad_norm": 0.517617450875545, "learning_rate": 9.71607570081411e-05, "loss": 3.1413, "step": 9521 }, { "epoch": 0.5910981438947173, "grad_norm": 0.514084218966473, "learning_rate": 9.715955720006247e-05, "loss": 3.1916, "step": 9522 }, { "epoch": 0.5911602209944752, "grad_norm": 0.48484453393625065, "learning_rate": 9.715835714594099e-05, "loss": 3.1016, "step": 9523 }, { "epoch": 0.591222298094233, "grad_norm": 0.49995177751696884, "learning_rate": 9.715715684578291e-05, "loss": 3.0688, "step": 9524 }, { "epoch": 0.5912843751939909, "grad_norm": 0.4792496015122199, "learning_rate": 9.715595629959449e-05, "loss": 3.0656, "step": 9525 }, { "epoch": 0.5913464522937488, "grad_norm": 0.5060726879574166, "learning_rate": 9.7154755507382e-05, "loss": 3.2793, "step": 9526 }, { "epoch": 0.5914085293935067, "grad_norm": 0.6690406230216664, "learning_rate": 9.715355446915171e-05, "loss": 3.2998, "step": 9527 }, { "epoch": 0.5914706064932647, "grad_norm": 0.7062287386504293, "learning_rate": 9.715235318490987e-05, "loss": 3.2709, "step": 9528 }, { "epoch": 0.5915326835930226, "grad_norm": 0.6212710241525091, "learning_rate": 9.715115165466275e-05, "loss": 3.1072, "step": 9529 }, { "epoch": 0.5915947606927804, "grad_norm": 0.5688801899394316, "learning_rate": 9.714994987841665e-05, "loss": 3.2106, "step": 9530 }, { "epoch": 0.5916568377925383, "grad_norm": 0.5305367056869501, "learning_rate": 9.714874785617781e-05, "loss": 3.1693, "step": 9531 }, { "epoch": 0.5917189148922962, "grad_norm": 0.6288723906885804, "learning_rate": 9.71475455879525e-05, "loss": 3.2448, "step": 9532 }, { "epoch": 0.5917809919920541, "grad_norm": 0.60654462163992, "learning_rate": 9.714634307374699e-05, "loss": 3.1302, "step": 9533 }, { "epoch": 0.5918430690918121, "grad_norm": 0.49702820665134706, "learning_rate": 9.714514031356756e-05, "loss": 3.1979, "step": 9534 }, { "epoch": 0.59190514619157, "grad_norm": 0.5546848673153577, "learning_rate": 9.714393730742051e-05, "loss": 3.2367, "step": 9535 }, { "epoch": 0.5919672232913278, "grad_norm": 0.5348399264627739, "learning_rate": 9.714273405531209e-05, "loss": 3.2366, "step": 9536 }, { "epoch": 0.5920293003910857, "grad_norm": 0.44964912364789056, "learning_rate": 9.714153055724858e-05, "loss": 3.1311, "step": 9537 }, { "epoch": 0.5920913774908436, "grad_norm": 0.5383629890467839, "learning_rate": 9.714032681323623e-05, "loss": 3.1599, "step": 9538 }, { "epoch": 0.5921534545906015, "grad_norm": 0.49261477256172687, "learning_rate": 9.713912282328139e-05, "loss": 3.1427, "step": 9539 }, { "epoch": 0.5922155316903595, "grad_norm": 0.4703383714270667, "learning_rate": 9.713791858739029e-05, "loss": 3.1569, "step": 9540 }, { "epoch": 0.5922776087901174, "grad_norm": 0.5460081041489175, "learning_rate": 9.713671410556923e-05, "loss": 3.1361, "step": 9541 }, { "epoch": 0.5923396858898752, "grad_norm": 0.48025050284971477, "learning_rate": 9.713550937782449e-05, "loss": 3.1769, "step": 9542 }, { "epoch": 0.5924017629896331, "grad_norm": 0.48856792426834417, "learning_rate": 9.713430440416234e-05, "loss": 3.1467, "step": 9543 }, { "epoch": 0.592463840089391, "grad_norm": 0.4470465844199979, "learning_rate": 9.713309918458911e-05, "loss": 3.1984, "step": 9544 }, { "epoch": 0.5925259171891489, "grad_norm": 0.506407187042264, "learning_rate": 9.713189371911102e-05, "loss": 3.2234, "step": 9545 }, { "epoch": 0.5925879942889068, "grad_norm": 0.5584529085600485, "learning_rate": 9.713068800773444e-05, "loss": 3.2312, "step": 9546 }, { "epoch": 0.5926500713886648, "grad_norm": 0.5408945051325645, "learning_rate": 9.71294820504656e-05, "loss": 3.205, "step": 9547 }, { "epoch": 0.5927121484884226, "grad_norm": 0.4816750216364429, "learning_rate": 9.71282758473108e-05, "loss": 3.2254, "step": 9548 }, { "epoch": 0.5927742255881805, "grad_norm": 0.4968239674062349, "learning_rate": 9.712706939827636e-05, "loss": 3.1807, "step": 9549 }, { "epoch": 0.5928363026879384, "grad_norm": 0.5594161055821711, "learning_rate": 9.712586270336856e-05, "loss": 3.2068, "step": 9550 }, { "epoch": 0.5928983797876963, "grad_norm": 0.5031291973341726, "learning_rate": 9.712465576259368e-05, "loss": 3.2017, "step": 9551 }, { "epoch": 0.5929604568874542, "grad_norm": 0.44679090135359323, "learning_rate": 9.712344857595805e-05, "loss": 3.2163, "step": 9552 }, { "epoch": 0.5930225339872122, "grad_norm": 0.5632132405825052, "learning_rate": 9.712224114346793e-05, "loss": 3.1755, "step": 9553 }, { "epoch": 0.59308461108697, "grad_norm": 0.4790539107824891, "learning_rate": 9.712103346512964e-05, "loss": 3.1581, "step": 9554 }, { "epoch": 0.5931466881867279, "grad_norm": 0.5249837554961592, "learning_rate": 9.71198255409495e-05, "loss": 3.0899, "step": 9555 }, { "epoch": 0.5932087652864858, "grad_norm": 0.5438272289810724, "learning_rate": 9.711861737093377e-05, "loss": 3.3034, "step": 9556 }, { "epoch": 0.5932708423862437, "grad_norm": 0.5060506333925255, "learning_rate": 9.711740895508879e-05, "loss": 3.1258, "step": 9557 }, { "epoch": 0.5933329194860016, "grad_norm": 0.49868196871948506, "learning_rate": 9.711620029342083e-05, "loss": 3.1905, "step": 9558 }, { "epoch": 0.5933949965857596, "grad_norm": 0.5111350887774113, "learning_rate": 9.711499138593623e-05, "loss": 3.2187, "step": 9559 }, { "epoch": 0.5934570736855174, "grad_norm": 0.49089196164517773, "learning_rate": 9.711378223264128e-05, "loss": 3.172, "step": 9560 }, { "epoch": 0.5935191507852753, "grad_norm": 0.4961521985322941, "learning_rate": 9.71125728335423e-05, "loss": 3.1874, "step": 9561 }, { "epoch": 0.5935812278850332, "grad_norm": 0.5309409476758216, "learning_rate": 9.711136318864558e-05, "loss": 3.2076, "step": 9562 }, { "epoch": 0.5936433049847911, "grad_norm": 0.5485312857142408, "learning_rate": 9.711015329795743e-05, "loss": 3.2734, "step": 9563 }, { "epoch": 0.593705382084549, "grad_norm": 0.5194038865999764, "learning_rate": 9.71089431614842e-05, "loss": 3.0948, "step": 9564 }, { "epoch": 0.593767459184307, "grad_norm": 0.4758422912627405, "learning_rate": 9.710773277923215e-05, "loss": 3.0931, "step": 9565 }, { "epoch": 0.5938295362840648, "grad_norm": 0.466075387529702, "learning_rate": 9.710652215120764e-05, "loss": 3.0164, "step": 9566 }, { "epoch": 0.5938916133838227, "grad_norm": 0.5395649213362076, "learning_rate": 9.710531127741697e-05, "loss": 3.2171, "step": 9567 }, { "epoch": 0.5939536904835806, "grad_norm": 0.49107953659398984, "learning_rate": 9.710410015786644e-05, "loss": 3.1914, "step": 9568 }, { "epoch": 0.5940157675833385, "grad_norm": 0.5079047192687961, "learning_rate": 9.71028887925624e-05, "loss": 3.1828, "step": 9569 }, { "epoch": 0.5940778446830964, "grad_norm": 0.5466425537930205, "learning_rate": 9.710167718151115e-05, "loss": 3.2026, "step": 9570 }, { "epoch": 0.5941399217828544, "grad_norm": 0.5228399025347589, "learning_rate": 9.710046532471901e-05, "loss": 3.2481, "step": 9571 }, { "epoch": 0.5942019988826122, "grad_norm": 0.5054399968596582, "learning_rate": 9.709925322219232e-05, "loss": 3.0763, "step": 9572 }, { "epoch": 0.5942640759823701, "grad_norm": 0.5439851975671022, "learning_rate": 9.709804087393738e-05, "loss": 3.2048, "step": 9573 }, { "epoch": 0.594326153082128, "grad_norm": 0.4749521344311305, "learning_rate": 9.709682827996054e-05, "loss": 3.1495, "step": 9574 }, { "epoch": 0.5943882301818859, "grad_norm": 0.49164649961081486, "learning_rate": 9.70956154402681e-05, "loss": 3.2381, "step": 9575 }, { "epoch": 0.5944503072816438, "grad_norm": 0.5141341591373235, "learning_rate": 9.709440235486643e-05, "loss": 3.1656, "step": 9576 }, { "epoch": 0.5945123843814017, "grad_norm": 0.4825543519028687, "learning_rate": 9.709318902376179e-05, "loss": 3.2014, "step": 9577 }, { "epoch": 0.5945744614811596, "grad_norm": 0.4751241270808486, "learning_rate": 9.709197544696057e-05, "loss": 3.151, "step": 9578 }, { "epoch": 0.5946365385809175, "grad_norm": 0.5340923677310173, "learning_rate": 9.709076162446908e-05, "loss": 3.2177, "step": 9579 }, { "epoch": 0.5946986156806754, "grad_norm": 0.47065049189796465, "learning_rate": 9.708954755629366e-05, "loss": 3.2008, "step": 9580 }, { "epoch": 0.5947606927804333, "grad_norm": 0.46651983619212806, "learning_rate": 9.708833324244063e-05, "loss": 3.0901, "step": 9581 }, { "epoch": 0.5948227698801912, "grad_norm": 0.530575042996955, "learning_rate": 9.708711868291635e-05, "loss": 3.2037, "step": 9582 }, { "epoch": 0.5948848469799491, "grad_norm": 0.48886119733699923, "learning_rate": 9.708590387772712e-05, "loss": 3.2399, "step": 9583 }, { "epoch": 0.594946924079707, "grad_norm": 0.49380195078790845, "learning_rate": 9.70846888268793e-05, "loss": 3.1237, "step": 9584 }, { "epoch": 0.5950090011794649, "grad_norm": 0.5201993103171635, "learning_rate": 9.708347353037926e-05, "loss": 3.2552, "step": 9585 }, { "epoch": 0.5950710782792228, "grad_norm": 0.5327458359525605, "learning_rate": 9.708225798823328e-05, "loss": 3.1211, "step": 9586 }, { "epoch": 0.5951331553789807, "grad_norm": 0.5343586902471802, "learning_rate": 9.708104220044774e-05, "loss": 3.1182, "step": 9587 }, { "epoch": 0.5951952324787386, "grad_norm": 0.5174024751699113, "learning_rate": 9.707982616702896e-05, "loss": 3.2093, "step": 9588 }, { "epoch": 0.5952573095784965, "grad_norm": 0.6429779524857258, "learning_rate": 9.707860988798332e-05, "loss": 3.195, "step": 9589 }, { "epoch": 0.5953193866782543, "grad_norm": 0.5889577210833248, "learning_rate": 9.707739336331713e-05, "loss": 3.1519, "step": 9590 }, { "epoch": 0.5953814637780123, "grad_norm": 0.5456712826809523, "learning_rate": 9.707617659303675e-05, "loss": 3.267, "step": 9591 }, { "epoch": 0.5954435408777702, "grad_norm": 0.5030794087479581, "learning_rate": 9.707495957714853e-05, "loss": 3.113, "step": 9592 }, { "epoch": 0.5955056179775281, "grad_norm": 0.5362677065720348, "learning_rate": 9.707374231565882e-05, "loss": 3.094, "step": 9593 }, { "epoch": 0.595567695077286, "grad_norm": 0.5271236949539946, "learning_rate": 9.707252480857397e-05, "loss": 3.2533, "step": 9594 }, { "epoch": 0.5956297721770439, "grad_norm": 0.5602550477988223, "learning_rate": 9.707130705590032e-05, "loss": 3.1378, "step": 9595 }, { "epoch": 0.5956918492768017, "grad_norm": 0.5250536891187552, "learning_rate": 9.707008905764425e-05, "loss": 3.2355, "step": 9596 }, { "epoch": 0.5957539263765597, "grad_norm": 0.5525876177276183, "learning_rate": 9.706887081381209e-05, "loss": 3.2328, "step": 9597 }, { "epoch": 0.5958160034763176, "grad_norm": 0.6012094625975793, "learning_rate": 9.70676523244102e-05, "loss": 3.2294, "step": 9598 }, { "epoch": 0.5958780805760755, "grad_norm": 0.5994020780904727, "learning_rate": 9.706643358944496e-05, "loss": 3.1842, "step": 9599 }, { "epoch": 0.5959401576758334, "grad_norm": 0.547120747527691, "learning_rate": 9.706521460892268e-05, "loss": 3.1986, "step": 9600 }, { "epoch": 0.5960022347755913, "grad_norm": 0.5106012795493906, "learning_rate": 9.706399538284978e-05, "loss": 3.0991, "step": 9601 }, { "epoch": 0.5960643118753491, "grad_norm": 0.6229564637024325, "learning_rate": 9.706277591123258e-05, "loss": 3.2155, "step": 9602 }, { "epoch": 0.5961263889751071, "grad_norm": 0.6350510175740804, "learning_rate": 9.706155619407745e-05, "loss": 3.1811, "step": 9603 }, { "epoch": 0.596188466074865, "grad_norm": 0.5530036566873822, "learning_rate": 9.706033623139075e-05, "loss": 3.1593, "step": 9604 }, { "epoch": 0.5962505431746229, "grad_norm": 0.5222451022673105, "learning_rate": 9.705911602317886e-05, "loss": 3.0678, "step": 9605 }, { "epoch": 0.5963126202743808, "grad_norm": 0.48100625962568044, "learning_rate": 9.705789556944814e-05, "loss": 3.1025, "step": 9606 }, { "epoch": 0.5963746973741387, "grad_norm": 0.5749624976077302, "learning_rate": 9.705667487020495e-05, "loss": 3.1537, "step": 9607 }, { "epoch": 0.5964367744738965, "grad_norm": 0.5646948115472372, "learning_rate": 9.705545392545568e-05, "loss": 3.2348, "step": 9608 }, { "epoch": 0.5964988515736545, "grad_norm": 0.4943893620783626, "learning_rate": 9.705423273520666e-05, "loss": 3.1853, "step": 9609 }, { "epoch": 0.5965609286734124, "grad_norm": 0.5141328548997187, "learning_rate": 9.705301129946429e-05, "loss": 3.141, "step": 9610 }, { "epoch": 0.5966230057731703, "grad_norm": 0.5074482418313626, "learning_rate": 9.705178961823495e-05, "loss": 3.1704, "step": 9611 }, { "epoch": 0.5966850828729282, "grad_norm": 0.49157041891307274, "learning_rate": 9.7050567691525e-05, "loss": 3.1458, "step": 9612 }, { "epoch": 0.5967471599726861, "grad_norm": 0.43843362168916694, "learning_rate": 9.704934551934081e-05, "loss": 3.2659, "step": 9613 }, { "epoch": 0.5968092370724439, "grad_norm": 0.49307766903581984, "learning_rate": 9.704812310168876e-05, "loss": 3.2189, "step": 9614 }, { "epoch": 0.5968713141722019, "grad_norm": 0.5005186192691459, "learning_rate": 9.704690043857523e-05, "loss": 3.2783, "step": 9615 }, { "epoch": 0.5969333912719598, "grad_norm": 0.48148847158243807, "learning_rate": 9.70456775300066e-05, "loss": 3.2081, "step": 9616 }, { "epoch": 0.5969954683717177, "grad_norm": 0.4365868916998484, "learning_rate": 9.704445437598925e-05, "loss": 3.0705, "step": 9617 }, { "epoch": 0.5970575454714756, "grad_norm": 0.41474672680782454, "learning_rate": 9.704323097652958e-05, "loss": 3.0359, "step": 9618 }, { "epoch": 0.5971196225712335, "grad_norm": 0.40234600918330005, "learning_rate": 9.704200733163393e-05, "loss": 3.1813, "step": 9619 }, { "epoch": 0.5971816996709913, "grad_norm": 0.4991889276088878, "learning_rate": 9.704078344130871e-05, "loss": 3.1958, "step": 9620 }, { "epoch": 0.5972437767707492, "grad_norm": 0.4616596739230215, "learning_rate": 9.703955930556033e-05, "loss": 3.2136, "step": 9621 }, { "epoch": 0.5973058538705072, "grad_norm": 0.4222503334685523, "learning_rate": 9.703833492439512e-05, "loss": 3.1595, "step": 9622 }, { "epoch": 0.5973679309702651, "grad_norm": 0.5409730143929611, "learning_rate": 9.70371102978195e-05, "loss": 3.0926, "step": 9623 }, { "epoch": 0.597430008070023, "grad_norm": 0.46467813378484774, "learning_rate": 9.703588542583987e-05, "loss": 3.1971, "step": 9624 }, { "epoch": 0.5974920851697809, "grad_norm": 0.4847976932170173, "learning_rate": 9.70346603084626e-05, "loss": 3.2202, "step": 9625 }, { "epoch": 0.5975541622695387, "grad_norm": 0.5202523010529312, "learning_rate": 9.703343494569409e-05, "loss": 3.2464, "step": 9626 }, { "epoch": 0.5976162393692966, "grad_norm": 0.5336794838790817, "learning_rate": 9.703220933754074e-05, "loss": 3.1918, "step": 9627 }, { "epoch": 0.5976783164690546, "grad_norm": 0.5599408212966256, "learning_rate": 9.703098348400892e-05, "loss": 3.2443, "step": 9628 }, { "epoch": 0.5977403935688125, "grad_norm": 0.5796199491888331, "learning_rate": 9.702975738510504e-05, "loss": 3.2301, "step": 9629 }, { "epoch": 0.5978024706685704, "grad_norm": 0.5993783362795965, "learning_rate": 9.702853104083551e-05, "loss": 3.2293, "step": 9630 }, { "epoch": 0.5978645477683283, "grad_norm": 0.6344500182570946, "learning_rate": 9.702730445120672e-05, "loss": 3.168, "step": 9631 }, { "epoch": 0.5979266248680861, "grad_norm": 0.5568497649703447, "learning_rate": 9.702607761622507e-05, "loss": 3.2179, "step": 9632 }, { "epoch": 0.597988701967844, "grad_norm": 0.48930963177095527, "learning_rate": 9.702485053589695e-05, "loss": 3.0709, "step": 9633 }, { "epoch": 0.598050779067602, "grad_norm": 0.5973121550094888, "learning_rate": 9.702362321022878e-05, "loss": 3.2229, "step": 9634 }, { "epoch": 0.5981128561673599, "grad_norm": 0.5917667091663278, "learning_rate": 9.702239563922693e-05, "loss": 3.0311, "step": 9635 }, { "epoch": 0.5981749332671178, "grad_norm": 0.5131211319939075, "learning_rate": 9.702116782289782e-05, "loss": 3.1531, "step": 9636 }, { "epoch": 0.5982370103668757, "grad_norm": 0.5840585718895358, "learning_rate": 9.70199397612479e-05, "loss": 3.1954, "step": 9637 }, { "epoch": 0.5982990874666335, "grad_norm": 0.5687962062636747, "learning_rate": 9.701871145428352e-05, "loss": 3.1578, "step": 9638 }, { "epoch": 0.5983611645663914, "grad_norm": 0.5940822374495452, "learning_rate": 9.701748290201111e-05, "loss": 3.1403, "step": 9639 }, { "epoch": 0.5984232416661494, "grad_norm": 0.526777684194401, "learning_rate": 9.701625410443707e-05, "loss": 3.1819, "step": 9640 }, { "epoch": 0.5984853187659073, "grad_norm": 0.4834889182750836, "learning_rate": 9.701502506156783e-05, "loss": 3.1925, "step": 9641 }, { "epoch": 0.5985473958656652, "grad_norm": 0.4921153182127922, "learning_rate": 9.701379577340979e-05, "loss": 3.1338, "step": 9642 }, { "epoch": 0.5986094729654231, "grad_norm": 0.5534568351586139, "learning_rate": 9.701256623996935e-05, "loss": 3.119, "step": 9643 }, { "epoch": 0.5986715500651809, "grad_norm": 0.5330161580808919, "learning_rate": 9.701133646125295e-05, "loss": 3.1646, "step": 9644 }, { "epoch": 0.5987336271649388, "grad_norm": 0.4745858896983186, "learning_rate": 9.7010106437267e-05, "loss": 3.1362, "step": 9645 }, { "epoch": 0.5987957042646967, "grad_norm": 0.5774300754621841, "learning_rate": 9.70088761680179e-05, "loss": 3.0798, "step": 9646 }, { "epoch": 0.5988577813644547, "grad_norm": 0.48549863248803377, "learning_rate": 9.70076456535121e-05, "loss": 3.0886, "step": 9647 }, { "epoch": 0.5989198584642126, "grad_norm": 0.5265389694706661, "learning_rate": 9.700641489375599e-05, "loss": 3.1134, "step": 9648 }, { "epoch": 0.5989819355639705, "grad_norm": 0.5085317946052342, "learning_rate": 9.7005183888756e-05, "loss": 3.1935, "step": 9649 }, { "epoch": 0.5990440126637283, "grad_norm": 0.5078559563456022, "learning_rate": 9.700395263851856e-05, "loss": 3.2023, "step": 9650 }, { "epoch": 0.5991060897634862, "grad_norm": 0.4156454525841414, "learning_rate": 9.700272114305009e-05, "loss": 3.1237, "step": 9651 }, { "epoch": 0.5991681668632441, "grad_norm": 0.5404365001451726, "learning_rate": 9.700148940235701e-05, "loss": 3.231, "step": 9652 }, { "epoch": 0.5992302439630021, "grad_norm": 0.5184039455360069, "learning_rate": 9.700025741644574e-05, "loss": 3.2281, "step": 9653 }, { "epoch": 0.59929232106276, "grad_norm": 0.5855745081269333, "learning_rate": 9.699902518532274e-05, "loss": 3.18, "step": 9654 }, { "epoch": 0.5993543981625179, "grad_norm": 0.5811025391437368, "learning_rate": 9.69977927089944e-05, "loss": 3.1286, "step": 9655 }, { "epoch": 0.5994164752622757, "grad_norm": 0.5296925828627951, "learning_rate": 9.699655998746717e-05, "loss": 3.0903, "step": 9656 }, { "epoch": 0.5994785523620336, "grad_norm": 0.44073536620186, "learning_rate": 9.699532702074748e-05, "loss": 3.1245, "step": 9657 }, { "epoch": 0.5995406294617915, "grad_norm": 0.48477002924932594, "learning_rate": 9.699409380884174e-05, "loss": 3.208, "step": 9658 }, { "epoch": 0.5996027065615495, "grad_norm": 0.465155713199899, "learning_rate": 9.699286035175643e-05, "loss": 3.1798, "step": 9659 }, { "epoch": 0.5996647836613074, "grad_norm": 0.4822490861467767, "learning_rate": 9.699162664949795e-05, "loss": 3.1765, "step": 9660 }, { "epoch": 0.5997268607610653, "grad_norm": 0.4533590337215272, "learning_rate": 9.699039270207276e-05, "loss": 3.1871, "step": 9661 }, { "epoch": 0.5997889378608231, "grad_norm": 0.503286527684752, "learning_rate": 9.698915850948725e-05, "loss": 3.1306, "step": 9662 }, { "epoch": 0.599851014960581, "grad_norm": 0.46173038823003837, "learning_rate": 9.698792407174792e-05, "loss": 3.2973, "step": 9663 }, { "epoch": 0.5999130920603389, "grad_norm": 0.5412817070023455, "learning_rate": 9.698668938886118e-05, "loss": 3.1457, "step": 9664 }, { "epoch": 0.5999751691600969, "grad_norm": 0.4899431708769625, "learning_rate": 9.698545446083347e-05, "loss": 3.1927, "step": 9665 }, { "epoch": 0.6000372462598548, "grad_norm": 0.45645930701737675, "learning_rate": 9.698421928767124e-05, "loss": 3.1888, "step": 9666 }, { "epoch": 0.6000993233596127, "grad_norm": 0.4747347567995281, "learning_rate": 9.698298386938093e-05, "loss": 3.1526, "step": 9667 }, { "epoch": 0.6001614004593705, "grad_norm": 0.5183172167125606, "learning_rate": 9.6981748205969e-05, "loss": 3.1633, "step": 9668 }, { "epoch": 0.6002234775591284, "grad_norm": 0.4684953951797624, "learning_rate": 9.698051229744188e-05, "loss": 3.158, "step": 9669 }, { "epoch": 0.6002855546588863, "grad_norm": 0.4794231295142003, "learning_rate": 9.6979276143806e-05, "loss": 3.1587, "step": 9670 }, { "epoch": 0.6003476317586443, "grad_norm": 0.47912549450115083, "learning_rate": 9.697803974506785e-05, "loss": 3.1866, "step": 9671 }, { "epoch": 0.6004097088584022, "grad_norm": 0.4746873422638742, "learning_rate": 9.697680310123387e-05, "loss": 3.203, "step": 9672 }, { "epoch": 0.6004717859581601, "grad_norm": 0.5361750851834798, "learning_rate": 9.69755662123105e-05, "loss": 3.1335, "step": 9673 }, { "epoch": 0.6005338630579179, "grad_norm": 0.47547189022635594, "learning_rate": 9.69743290783042e-05, "loss": 3.1166, "step": 9674 }, { "epoch": 0.6005959401576758, "grad_norm": 0.4296487270548123, "learning_rate": 9.69730916992214e-05, "loss": 3.0454, "step": 9675 }, { "epoch": 0.6006580172574337, "grad_norm": 0.49981656285884135, "learning_rate": 9.697185407506859e-05, "loss": 3.1106, "step": 9676 }, { "epoch": 0.6007200943571916, "grad_norm": 0.5023836004371002, "learning_rate": 9.697061620585221e-05, "loss": 3.1744, "step": 9677 }, { "epoch": 0.6007821714569496, "grad_norm": 0.4807663592839686, "learning_rate": 9.696937809157873e-05, "loss": 3.2719, "step": 9678 }, { "epoch": 0.6008442485567075, "grad_norm": 0.48177468753682035, "learning_rate": 9.69681397322546e-05, "loss": 3.216, "step": 9679 }, { "epoch": 0.6009063256564653, "grad_norm": 0.44270239059349087, "learning_rate": 9.696690112788629e-05, "loss": 3.2085, "step": 9680 }, { "epoch": 0.6009684027562232, "grad_norm": 0.5307955761790537, "learning_rate": 9.696566227848026e-05, "loss": 3.2224, "step": 9681 }, { "epoch": 0.6010304798559811, "grad_norm": 0.5526609338287415, "learning_rate": 9.696442318404295e-05, "loss": 3.1868, "step": 9682 }, { "epoch": 0.601092556955739, "grad_norm": 0.5391883979507774, "learning_rate": 9.696318384458085e-05, "loss": 3.0899, "step": 9683 }, { "epoch": 0.601154634055497, "grad_norm": 0.49281838265380995, "learning_rate": 9.696194426010043e-05, "loss": 3.2348, "step": 9684 }, { "epoch": 0.6012167111552549, "grad_norm": 0.4492287676150355, "learning_rate": 9.696070443060814e-05, "loss": 3.1688, "step": 9685 }, { "epoch": 0.6012787882550127, "grad_norm": 0.4919001179283778, "learning_rate": 9.695946435611045e-05, "loss": 3.1767, "step": 9686 }, { "epoch": 0.6013408653547706, "grad_norm": 0.4680955132440339, "learning_rate": 9.695822403661383e-05, "loss": 3.1638, "step": 9687 }, { "epoch": 0.6014029424545285, "grad_norm": 0.435849909716386, "learning_rate": 9.695698347212477e-05, "loss": 3.1522, "step": 9688 }, { "epoch": 0.6014650195542864, "grad_norm": 0.42378730401549725, "learning_rate": 9.695574266264972e-05, "loss": 3.2124, "step": 9689 }, { "epoch": 0.6015270966540444, "grad_norm": 0.4430158183070139, "learning_rate": 9.695450160819516e-05, "loss": 3.1292, "step": 9690 }, { "epoch": 0.6015891737538023, "grad_norm": 0.517066068886201, "learning_rate": 9.695326030876757e-05, "loss": 3.2324, "step": 9691 }, { "epoch": 0.6016512508535601, "grad_norm": 0.46235090894142555, "learning_rate": 9.695201876437343e-05, "loss": 3.1506, "step": 9692 }, { "epoch": 0.601713327953318, "grad_norm": 0.5041324492210859, "learning_rate": 9.695077697501919e-05, "loss": 3.2325, "step": 9693 }, { "epoch": 0.6017754050530759, "grad_norm": 0.4849855973622112, "learning_rate": 9.694953494071137e-05, "loss": 3.166, "step": 9694 }, { "epoch": 0.6018374821528338, "grad_norm": 0.4369080660344138, "learning_rate": 9.694829266145642e-05, "loss": 3.2688, "step": 9695 }, { "epoch": 0.6018995592525918, "grad_norm": 0.4890398160408606, "learning_rate": 9.694705013726082e-05, "loss": 3.2074, "step": 9696 }, { "epoch": 0.6019616363523497, "grad_norm": 0.5187613440043093, "learning_rate": 9.694580736813108e-05, "loss": 3.208, "step": 9697 }, { "epoch": 0.6020237134521075, "grad_norm": 0.43528234890857903, "learning_rate": 9.694456435407366e-05, "loss": 3.13, "step": 9698 }, { "epoch": 0.6020857905518654, "grad_norm": 0.49501308874203753, "learning_rate": 9.694332109509504e-05, "loss": 3.2749, "step": 9699 }, { "epoch": 0.6021478676516233, "grad_norm": 0.5328200364908278, "learning_rate": 9.694207759120173e-05, "loss": 3.1883, "step": 9700 }, { "epoch": 0.6022099447513812, "grad_norm": 0.44761569054091355, "learning_rate": 9.69408338424002e-05, "loss": 3.1008, "step": 9701 }, { "epoch": 0.6022720218511391, "grad_norm": 0.5484340787166813, "learning_rate": 9.693958984869696e-05, "loss": 3.2086, "step": 9702 }, { "epoch": 0.6023340989508971, "grad_norm": 0.5270058008196828, "learning_rate": 9.693834561009847e-05, "loss": 3.1487, "step": 9703 }, { "epoch": 0.6023961760506549, "grad_norm": 0.4558085853767383, "learning_rate": 9.693710112661123e-05, "loss": 3.1725, "step": 9704 }, { "epoch": 0.6024582531504128, "grad_norm": 0.5330063732850467, "learning_rate": 9.693585639824176e-05, "loss": 3.0788, "step": 9705 }, { "epoch": 0.6025203302501707, "grad_norm": 0.5760885404677674, "learning_rate": 9.69346114249965e-05, "loss": 3.1319, "step": 9706 }, { "epoch": 0.6025824073499286, "grad_norm": 0.555333795262603, "learning_rate": 9.6933366206882e-05, "loss": 3.1731, "step": 9707 }, { "epoch": 0.6026444844496865, "grad_norm": 0.5085962556229382, "learning_rate": 9.693212074390474e-05, "loss": 3.182, "step": 9708 }, { "epoch": 0.6027065615494445, "grad_norm": 0.4738822516135794, "learning_rate": 9.693087503607121e-05, "loss": 3.1922, "step": 9709 }, { "epoch": 0.6027686386492023, "grad_norm": 0.505639724469894, "learning_rate": 9.692962908338791e-05, "loss": 3.2438, "step": 9710 }, { "epoch": 0.6028307157489602, "grad_norm": 0.5606655566750348, "learning_rate": 9.692838288586134e-05, "loss": 3.1948, "step": 9711 }, { "epoch": 0.6028927928487181, "grad_norm": 0.5591465275639331, "learning_rate": 9.692713644349801e-05, "loss": 3.2634, "step": 9712 }, { "epoch": 0.602954869948476, "grad_norm": 0.49721911588186285, "learning_rate": 9.692588975630441e-05, "loss": 3.0462, "step": 9713 }, { "epoch": 0.6030169470482339, "grad_norm": 0.547412526402133, "learning_rate": 9.692464282428706e-05, "loss": 3.1921, "step": 9714 }, { "epoch": 0.6030790241479919, "grad_norm": 0.5265361111421228, "learning_rate": 9.692339564745246e-05, "loss": 3.1049, "step": 9715 }, { "epoch": 0.6031411012477497, "grad_norm": 0.4862964420645773, "learning_rate": 9.69221482258071e-05, "loss": 3.2065, "step": 9716 }, { "epoch": 0.6032031783475076, "grad_norm": 0.49713607982539976, "learning_rate": 9.692090055935751e-05, "loss": 3.1275, "step": 9717 }, { "epoch": 0.6032652554472655, "grad_norm": 0.4855860521710567, "learning_rate": 9.691965264811018e-05, "loss": 3.1743, "step": 9718 }, { "epoch": 0.6033273325470234, "grad_norm": 0.5059128957589133, "learning_rate": 9.691840449207164e-05, "loss": 3.1249, "step": 9719 }, { "epoch": 0.6033894096467813, "grad_norm": 0.5013456416506629, "learning_rate": 9.69171560912484e-05, "loss": 3.113, "step": 9720 }, { "epoch": 0.6034514867465393, "grad_norm": 0.5101859453563844, "learning_rate": 9.691590744564696e-05, "loss": 3.0523, "step": 9721 }, { "epoch": 0.6035135638462971, "grad_norm": 0.5227921621376479, "learning_rate": 9.691465855527385e-05, "loss": 3.1631, "step": 9722 }, { "epoch": 0.603575640946055, "grad_norm": 0.48853150630327574, "learning_rate": 9.691340942013557e-05, "loss": 3.2582, "step": 9723 }, { "epoch": 0.6036377180458129, "grad_norm": 0.4727226242246607, "learning_rate": 9.691216004023864e-05, "loss": 3.2083, "step": 9724 }, { "epoch": 0.6036997951455708, "grad_norm": 0.4984564200610147, "learning_rate": 9.69109104155896e-05, "loss": 3.2293, "step": 9725 }, { "epoch": 0.6037618722453287, "grad_norm": 0.5655463959678311, "learning_rate": 9.690966054619493e-05, "loss": 3.1812, "step": 9726 }, { "epoch": 0.6038239493450867, "grad_norm": 0.5443895535981732, "learning_rate": 9.69084104320612e-05, "loss": 3.1934, "step": 9727 }, { "epoch": 0.6038860264448445, "grad_norm": 0.5130844819012321, "learning_rate": 9.690716007319488e-05, "loss": 3.1513, "step": 9728 }, { "epoch": 0.6039481035446024, "grad_norm": 0.48656779438879527, "learning_rate": 9.690590946960253e-05, "loss": 3.12, "step": 9729 }, { "epoch": 0.6040101806443603, "grad_norm": 0.43928429645438005, "learning_rate": 9.690465862129065e-05, "loss": 3.2048, "step": 9730 }, { "epoch": 0.6040722577441182, "grad_norm": 0.517665754853326, "learning_rate": 9.69034075282658e-05, "loss": 3.2454, "step": 9731 }, { "epoch": 0.6041343348438761, "grad_norm": 0.46109282575906996, "learning_rate": 9.690215619053448e-05, "loss": 3.1388, "step": 9732 }, { "epoch": 0.604196411943634, "grad_norm": 0.42569675832136494, "learning_rate": 9.690090460810321e-05, "loss": 3.187, "step": 9733 }, { "epoch": 0.6042584890433919, "grad_norm": 0.4466961896389272, "learning_rate": 9.689965278097857e-05, "loss": 3.2143, "step": 9734 }, { "epoch": 0.6043205661431498, "grad_norm": 0.5251815833825627, "learning_rate": 9.689840070916703e-05, "loss": 3.1675, "step": 9735 }, { "epoch": 0.6043826432429077, "grad_norm": 0.4923764426255525, "learning_rate": 9.689714839267515e-05, "loss": 3.1418, "step": 9736 }, { "epoch": 0.6044447203426656, "grad_norm": 0.508151660046188, "learning_rate": 9.689589583150946e-05, "loss": 3.1501, "step": 9737 }, { "epoch": 0.6045067974424235, "grad_norm": 0.6131522170470265, "learning_rate": 9.68946430256765e-05, "loss": 3.1572, "step": 9738 }, { "epoch": 0.6045688745421814, "grad_norm": 0.6361619820259484, "learning_rate": 9.689338997518281e-05, "loss": 3.1697, "step": 9739 }, { "epoch": 0.6046309516419393, "grad_norm": 0.5129336720376416, "learning_rate": 9.689213668003492e-05, "loss": 3.1363, "step": 9740 }, { "epoch": 0.6046930287416972, "grad_norm": 0.49950093456272526, "learning_rate": 9.689088314023937e-05, "loss": 3.1207, "step": 9741 }, { "epoch": 0.6047551058414551, "grad_norm": 0.574555227086945, "learning_rate": 9.688962935580268e-05, "loss": 3.1482, "step": 9742 }, { "epoch": 0.604817182941213, "grad_norm": 0.6192561381793898, "learning_rate": 9.688837532673143e-05, "loss": 3.1602, "step": 9743 }, { "epoch": 0.6048792600409709, "grad_norm": 0.5298303822141915, "learning_rate": 9.688712105303215e-05, "loss": 3.1022, "step": 9744 }, { "epoch": 0.6049413371407287, "grad_norm": 0.5195342329126068, "learning_rate": 9.688586653471137e-05, "loss": 3.1276, "step": 9745 }, { "epoch": 0.6050034142404866, "grad_norm": 0.5700327827412512, "learning_rate": 9.688461177177563e-05, "loss": 3.1639, "step": 9746 }, { "epoch": 0.6050654913402446, "grad_norm": 0.538311670382206, "learning_rate": 9.688335676423151e-05, "loss": 3.1429, "step": 9747 }, { "epoch": 0.6051275684400025, "grad_norm": 0.6761414995428267, "learning_rate": 9.68821015120855e-05, "loss": 3.262, "step": 9748 }, { "epoch": 0.6051896455397604, "grad_norm": 0.6300876254874914, "learning_rate": 9.688084601534421e-05, "loss": 3.1641, "step": 9749 }, { "epoch": 0.6052517226395183, "grad_norm": 0.5386536747967754, "learning_rate": 9.687959027401418e-05, "loss": 3.2244, "step": 9750 }, { "epoch": 0.6053137997392761, "grad_norm": 0.6056805886108603, "learning_rate": 9.687833428810194e-05, "loss": 3.1359, "step": 9751 }, { "epoch": 0.605375876839034, "grad_norm": 0.6277072447235496, "learning_rate": 9.687707805761403e-05, "loss": 3.1241, "step": 9752 }, { "epoch": 0.605437953938792, "grad_norm": 0.5751625919588278, "learning_rate": 9.687582158255704e-05, "loss": 3.1986, "step": 9753 }, { "epoch": 0.6055000310385499, "grad_norm": 0.5437225762390533, "learning_rate": 9.68745648629375e-05, "loss": 3.212, "step": 9754 }, { "epoch": 0.6055621081383078, "grad_norm": 0.5414348314015415, "learning_rate": 9.6873307898762e-05, "loss": 3.1999, "step": 9755 }, { "epoch": 0.6056241852380657, "grad_norm": 0.5619742267475198, "learning_rate": 9.687205069003704e-05, "loss": 3.1412, "step": 9756 }, { "epoch": 0.6056862623378235, "grad_norm": 0.4800914896011898, "learning_rate": 9.687079323676922e-05, "loss": 3.1191, "step": 9757 }, { "epoch": 0.6057483394375814, "grad_norm": 0.5584871683943362, "learning_rate": 9.68695355389651e-05, "loss": 3.1113, "step": 9758 }, { "epoch": 0.6058104165373394, "grad_norm": 0.5205294549103444, "learning_rate": 9.686827759663124e-05, "loss": 3.1109, "step": 9759 }, { "epoch": 0.6058724936370973, "grad_norm": 0.5460323108170917, "learning_rate": 9.68670194097742e-05, "loss": 3.1554, "step": 9760 }, { "epoch": 0.6059345707368552, "grad_norm": 0.489635968591629, "learning_rate": 9.686576097840053e-05, "loss": 3.1098, "step": 9761 }, { "epoch": 0.6059966478366131, "grad_norm": 0.5788676157632566, "learning_rate": 9.686450230251683e-05, "loss": 3.1332, "step": 9762 }, { "epoch": 0.6060587249363709, "grad_norm": 0.4924986930287636, "learning_rate": 9.686324338212962e-05, "loss": 3.2157, "step": 9763 }, { "epoch": 0.6061208020361288, "grad_norm": 0.49495023060585275, "learning_rate": 9.686198421724551e-05, "loss": 3.1545, "step": 9764 }, { "epoch": 0.6061828791358868, "grad_norm": 0.47380971849231684, "learning_rate": 9.686072480787103e-05, "loss": 3.1547, "step": 9765 }, { "epoch": 0.6062449562356447, "grad_norm": 0.470106007308732, "learning_rate": 9.685946515401279e-05, "loss": 3.0954, "step": 9766 }, { "epoch": 0.6063070333354026, "grad_norm": 0.47136750562421126, "learning_rate": 9.685820525567735e-05, "loss": 3.1779, "step": 9767 }, { "epoch": 0.6063691104351605, "grad_norm": 0.4703252866660682, "learning_rate": 9.685694511287126e-05, "loss": 3.2259, "step": 9768 }, { "epoch": 0.6064311875349183, "grad_norm": 0.4908686027041996, "learning_rate": 9.685568472560114e-05, "loss": 3.1616, "step": 9769 }, { "epoch": 0.6064932646346762, "grad_norm": 0.44209069381911903, "learning_rate": 9.685442409387351e-05, "loss": 3.1144, "step": 9770 }, { "epoch": 0.6065553417344342, "grad_norm": 0.496861367068611, "learning_rate": 9.685316321769499e-05, "loss": 3.1803, "step": 9771 }, { "epoch": 0.6066174188341921, "grad_norm": 0.47366616043199783, "learning_rate": 9.685190209707213e-05, "loss": 3.1071, "step": 9772 }, { "epoch": 0.60667949593395, "grad_norm": 0.4781328499841472, "learning_rate": 9.685064073201154e-05, "loss": 3.1141, "step": 9773 }, { "epoch": 0.6067415730337079, "grad_norm": 0.4676992879246333, "learning_rate": 9.684937912251976e-05, "loss": 3.2355, "step": 9774 }, { "epoch": 0.6068036501334657, "grad_norm": 0.47954655356550363, "learning_rate": 9.684811726860343e-05, "loss": 3.0751, "step": 9775 }, { "epoch": 0.6068657272332236, "grad_norm": 0.41743456493339065, "learning_rate": 9.684685517026909e-05, "loss": 3.2398, "step": 9776 }, { "epoch": 0.6069278043329815, "grad_norm": 0.49536430617443594, "learning_rate": 9.684559282752332e-05, "loss": 3.1914, "step": 9777 }, { "epoch": 0.6069898814327395, "grad_norm": 0.44622328125281085, "learning_rate": 9.684433024037271e-05, "loss": 3.2521, "step": 9778 }, { "epoch": 0.6070519585324974, "grad_norm": 0.5274078895275294, "learning_rate": 9.684306740882388e-05, "loss": 3.1878, "step": 9779 }, { "epoch": 0.6071140356322553, "grad_norm": 0.5381781469632569, "learning_rate": 9.68418043328834e-05, "loss": 3.1065, "step": 9780 }, { "epoch": 0.6071761127320131, "grad_norm": 0.6056830242378554, "learning_rate": 9.684054101255783e-05, "loss": 3.1723, "step": 9781 }, { "epoch": 0.607238189831771, "grad_norm": 0.5936672755385883, "learning_rate": 9.683927744785382e-05, "loss": 3.1677, "step": 9782 }, { "epoch": 0.6073002669315289, "grad_norm": 0.6069535047314139, "learning_rate": 9.68380136387779e-05, "loss": 3.1723, "step": 9783 }, { "epoch": 0.6073623440312869, "grad_norm": 0.5171541600847553, "learning_rate": 9.68367495853367e-05, "loss": 3.1937, "step": 9784 }, { "epoch": 0.6074244211310448, "grad_norm": 0.48629874021926667, "learning_rate": 9.68354852875368e-05, "loss": 3.1275, "step": 9785 }, { "epoch": 0.6074864982308027, "grad_norm": 0.5001817313379634, "learning_rate": 9.683422074538482e-05, "loss": 3.1273, "step": 9786 }, { "epoch": 0.6075485753305605, "grad_norm": 0.5547958053375288, "learning_rate": 9.683295595888734e-05, "loss": 3.1789, "step": 9787 }, { "epoch": 0.6076106524303184, "grad_norm": 0.5614733599018487, "learning_rate": 9.683169092805096e-05, "loss": 3.2329, "step": 9788 }, { "epoch": 0.6076727295300763, "grad_norm": 0.5140847696970744, "learning_rate": 9.683042565288228e-05, "loss": 3.211, "step": 9789 }, { "epoch": 0.6077348066298343, "grad_norm": 0.5540090091657917, "learning_rate": 9.682916013338789e-05, "loss": 3.2487, "step": 9790 }, { "epoch": 0.6077968837295922, "grad_norm": 0.6413533094816648, "learning_rate": 9.68278943695744e-05, "loss": 3.1693, "step": 9791 }, { "epoch": 0.6078589608293501, "grad_norm": 0.5222720366193537, "learning_rate": 9.682662836144843e-05, "loss": 3.1336, "step": 9792 }, { "epoch": 0.6079210379291079, "grad_norm": 0.4871624144530498, "learning_rate": 9.682536210901655e-05, "loss": 3.1657, "step": 9793 }, { "epoch": 0.6079831150288658, "grad_norm": 0.5698424453239265, "learning_rate": 9.682409561228543e-05, "loss": 3.1978, "step": 9794 }, { "epoch": 0.6080451921286237, "grad_norm": 0.5221487948890959, "learning_rate": 9.68228288712616e-05, "loss": 3.1641, "step": 9795 }, { "epoch": 0.6081072692283817, "grad_norm": 0.5005456807786304, "learning_rate": 9.682156188595173e-05, "loss": 3.106, "step": 9796 }, { "epoch": 0.6081693463281396, "grad_norm": 0.5316636494647585, "learning_rate": 9.682029465636239e-05, "loss": 3.1686, "step": 9797 }, { "epoch": 0.6082314234278975, "grad_norm": 0.5366466123919171, "learning_rate": 9.68190271825002e-05, "loss": 3.1694, "step": 9798 }, { "epoch": 0.6082935005276553, "grad_norm": 0.5563871910989604, "learning_rate": 9.68177594643718e-05, "loss": 3.172, "step": 9799 }, { "epoch": 0.6083555776274132, "grad_norm": 0.4231812599171649, "learning_rate": 9.681649150198377e-05, "loss": 3.1425, "step": 9800 }, { "epoch": 0.6084176547271711, "grad_norm": 0.5600043406062841, "learning_rate": 9.681522329534273e-05, "loss": 3.1738, "step": 9801 }, { "epoch": 0.608479731826929, "grad_norm": 0.5622760008808259, "learning_rate": 9.681395484445532e-05, "loss": 3.1875, "step": 9802 }, { "epoch": 0.608541808926687, "grad_norm": 0.550401028818386, "learning_rate": 9.681268614932815e-05, "loss": 3.1689, "step": 9803 }, { "epoch": 0.6086038860264449, "grad_norm": 0.6044812178348246, "learning_rate": 9.681141720996782e-05, "loss": 3.092, "step": 9804 }, { "epoch": 0.6086659631262027, "grad_norm": 0.4675296474684554, "learning_rate": 9.681014802638096e-05, "loss": 3.1398, "step": 9805 }, { "epoch": 0.6087280402259606, "grad_norm": 0.701052791078302, "learning_rate": 9.680887859857421e-05, "loss": 3.2487, "step": 9806 }, { "epoch": 0.6087901173257185, "grad_norm": 0.7440302809557111, "learning_rate": 9.680760892655416e-05, "loss": 3.0437, "step": 9807 }, { "epoch": 0.6088521944254764, "grad_norm": 0.6458635374410132, "learning_rate": 9.680633901032745e-05, "loss": 3.1811, "step": 9808 }, { "epoch": 0.6089142715252344, "grad_norm": 0.6274073016327024, "learning_rate": 9.680506884990072e-05, "loss": 3.2195, "step": 9809 }, { "epoch": 0.6089763486249923, "grad_norm": 0.6063765334780523, "learning_rate": 9.680379844528058e-05, "loss": 3.1591, "step": 9810 }, { "epoch": 0.6090384257247501, "grad_norm": 0.7254010833988771, "learning_rate": 9.680252779647365e-05, "loss": 3.196, "step": 9811 }, { "epoch": 0.609100502824508, "grad_norm": 0.4755337508314391, "learning_rate": 9.680125690348659e-05, "loss": 3.0677, "step": 9812 }, { "epoch": 0.6091625799242659, "grad_norm": 0.6254030358669327, "learning_rate": 9.6799985766326e-05, "loss": 3.2002, "step": 9813 }, { "epoch": 0.6092246570240238, "grad_norm": 0.5810559953947334, "learning_rate": 9.679871438499852e-05, "loss": 3.2335, "step": 9814 }, { "epoch": 0.6092867341237818, "grad_norm": 0.48634398101328574, "learning_rate": 9.679744275951079e-05, "loss": 3.1582, "step": 9815 }, { "epoch": 0.6093488112235397, "grad_norm": 0.582282376997199, "learning_rate": 9.679617088986945e-05, "loss": 3.1843, "step": 9816 }, { "epoch": 0.6094108883232975, "grad_norm": 0.5704840310692734, "learning_rate": 9.679489877608112e-05, "loss": 3.2006, "step": 9817 }, { "epoch": 0.6094729654230554, "grad_norm": 0.5307273818545393, "learning_rate": 9.679362641815242e-05, "loss": 3.1096, "step": 9818 }, { "epoch": 0.6095350425228133, "grad_norm": 0.6871447512317584, "learning_rate": 9.679235381609004e-05, "loss": 3.1831, "step": 9819 }, { "epoch": 0.6095971196225712, "grad_norm": 0.5512537451376819, "learning_rate": 9.67910809699006e-05, "loss": 3.1894, "step": 9820 }, { "epoch": 0.6096591967223292, "grad_norm": 0.6139152554157113, "learning_rate": 9.67898078795907e-05, "loss": 3.1368, "step": 9821 }, { "epoch": 0.6097212738220871, "grad_norm": 0.6330394102577047, "learning_rate": 9.678853454516704e-05, "loss": 3.2009, "step": 9822 }, { "epoch": 0.6097833509218449, "grad_norm": 0.6145923231297139, "learning_rate": 9.678726096663622e-05, "loss": 3.2396, "step": 9823 }, { "epoch": 0.6098454280216028, "grad_norm": 0.5436551537500214, "learning_rate": 9.678598714400491e-05, "loss": 3.1195, "step": 9824 }, { "epoch": 0.6099075051213607, "grad_norm": 0.5849959883593087, "learning_rate": 9.678471307727975e-05, "loss": 3.1978, "step": 9825 }, { "epoch": 0.6099695822211186, "grad_norm": 0.4735375062676111, "learning_rate": 9.678343876646736e-05, "loss": 3.054, "step": 9826 }, { "epoch": 0.6100316593208766, "grad_norm": 0.5244901804318461, "learning_rate": 9.678216421157443e-05, "loss": 3.1726, "step": 9827 }, { "epoch": 0.6100937364206345, "grad_norm": 0.563549995595854, "learning_rate": 9.67808894126076e-05, "loss": 3.162, "step": 9828 }, { "epoch": 0.6101558135203923, "grad_norm": 0.5267214741688739, "learning_rate": 9.67796143695735e-05, "loss": 3.1944, "step": 9829 }, { "epoch": 0.6102178906201502, "grad_norm": 0.4443928944060418, "learning_rate": 9.677833908247881e-05, "loss": 3.1771, "step": 9830 }, { "epoch": 0.6102799677199081, "grad_norm": 0.6034103104141966, "learning_rate": 9.677706355133016e-05, "loss": 3.2271, "step": 9831 }, { "epoch": 0.610342044819666, "grad_norm": 0.5458054028225116, "learning_rate": 9.677578777613422e-05, "loss": 3.1305, "step": 9832 }, { "epoch": 0.610404121919424, "grad_norm": 0.5452111956217927, "learning_rate": 9.677451175689763e-05, "loss": 3.2168, "step": 9833 }, { "epoch": 0.6104661990191819, "grad_norm": 0.5049092859525283, "learning_rate": 9.677323549362707e-05, "loss": 3.0952, "step": 9834 }, { "epoch": 0.6105282761189397, "grad_norm": 0.519094527403075, "learning_rate": 9.677195898632918e-05, "loss": 3.1758, "step": 9835 }, { "epoch": 0.6105903532186976, "grad_norm": 0.46444151008013357, "learning_rate": 9.677068223501063e-05, "loss": 3.1162, "step": 9836 }, { "epoch": 0.6106524303184555, "grad_norm": 0.5260968761617457, "learning_rate": 9.676940523967806e-05, "loss": 3.0715, "step": 9837 }, { "epoch": 0.6107145074182134, "grad_norm": 0.49131711566883873, "learning_rate": 9.676812800033816e-05, "loss": 3.1458, "step": 9838 }, { "epoch": 0.6107765845179713, "grad_norm": 0.4838386457874096, "learning_rate": 9.676685051699758e-05, "loss": 3.1199, "step": 9839 }, { "epoch": 0.6108386616177293, "grad_norm": 0.5278098695462473, "learning_rate": 9.6765572789663e-05, "loss": 3.0864, "step": 9840 }, { "epoch": 0.6109007387174871, "grad_norm": 0.49096957395158, "learning_rate": 9.676429481834106e-05, "loss": 3.0653, "step": 9841 }, { "epoch": 0.610962815817245, "grad_norm": 0.5154819000962677, "learning_rate": 9.676301660303846e-05, "loss": 3.0917, "step": 9842 }, { "epoch": 0.6110248929170029, "grad_norm": 0.5110104733306424, "learning_rate": 9.676173814376184e-05, "loss": 3.1886, "step": 9843 }, { "epoch": 0.6110869700167608, "grad_norm": 0.5106281568973592, "learning_rate": 9.676045944051787e-05, "loss": 3.1868, "step": 9844 }, { "epoch": 0.6111490471165187, "grad_norm": 0.4321764041637516, "learning_rate": 9.675918049331323e-05, "loss": 3.2, "step": 9845 }, { "epoch": 0.6112111242162767, "grad_norm": 0.4873867080556031, "learning_rate": 9.675790130215461e-05, "loss": 3.1017, "step": 9846 }, { "epoch": 0.6112732013160345, "grad_norm": 0.4928569479511499, "learning_rate": 9.675662186704865e-05, "loss": 3.2824, "step": 9847 }, { "epoch": 0.6113352784157924, "grad_norm": 0.5643747928741515, "learning_rate": 9.675534218800206e-05, "loss": 3.1445, "step": 9848 }, { "epoch": 0.6113973555155503, "grad_norm": 0.48379320151353034, "learning_rate": 9.675406226502151e-05, "loss": 3.1298, "step": 9849 }, { "epoch": 0.6114594326153082, "grad_norm": 0.47419950267060973, "learning_rate": 9.675278209811365e-05, "loss": 3.1105, "step": 9850 }, { "epoch": 0.6115215097150661, "grad_norm": 0.4618066578169107, "learning_rate": 9.675150168728518e-05, "loss": 3.0078, "step": 9851 }, { "epoch": 0.611583586814824, "grad_norm": 0.4852028816651754, "learning_rate": 9.675022103254277e-05, "loss": 3.1743, "step": 9852 }, { "epoch": 0.6116456639145819, "grad_norm": 0.5135946759713753, "learning_rate": 9.67489401338931e-05, "loss": 3.0843, "step": 9853 }, { "epoch": 0.6117077410143398, "grad_norm": 0.47376400418581616, "learning_rate": 9.674765899134289e-05, "loss": 3.1173, "step": 9854 }, { "epoch": 0.6117698181140977, "grad_norm": 0.48488345685243184, "learning_rate": 9.674637760489878e-05, "loss": 3.1732, "step": 9855 }, { "epoch": 0.6118318952138556, "grad_norm": 0.4346281548912998, "learning_rate": 9.674509597456746e-05, "loss": 3.2113, "step": 9856 }, { "epoch": 0.6118939723136135, "grad_norm": 0.4759679744229007, "learning_rate": 9.674381410035563e-05, "loss": 3.1092, "step": 9857 }, { "epoch": 0.6119560494133715, "grad_norm": 0.4836371862966896, "learning_rate": 9.674253198226998e-05, "loss": 3.1175, "step": 9858 }, { "epoch": 0.6120181265131293, "grad_norm": 0.5150718611977588, "learning_rate": 9.674124962031719e-05, "loss": 3.1773, "step": 9859 }, { "epoch": 0.6120802036128872, "grad_norm": 0.5563735856771223, "learning_rate": 9.673996701450396e-05, "loss": 3.0899, "step": 9860 }, { "epoch": 0.6121422807126451, "grad_norm": 0.4718280661528242, "learning_rate": 9.673868416483697e-05, "loss": 3.201, "step": 9861 }, { "epoch": 0.612204357812403, "grad_norm": 0.4871054262981648, "learning_rate": 9.673740107132291e-05, "loss": 3.131, "step": 9862 }, { "epoch": 0.6122664349121609, "grad_norm": 0.45705359966699666, "learning_rate": 9.673611773396849e-05, "loss": 3.2131, "step": 9863 }, { "epoch": 0.6123285120119188, "grad_norm": 0.5585672332231897, "learning_rate": 9.67348341527804e-05, "loss": 3.1793, "step": 9864 }, { "epoch": 0.6123905891116767, "grad_norm": 0.5527399285243039, "learning_rate": 9.673355032776534e-05, "loss": 3.1377, "step": 9865 }, { "epoch": 0.6124526662114346, "grad_norm": 0.5383288325081252, "learning_rate": 9.673226625893e-05, "loss": 3.0047, "step": 9866 }, { "epoch": 0.6125147433111925, "grad_norm": 0.5136006527104048, "learning_rate": 9.673098194628109e-05, "loss": 3.1672, "step": 9867 }, { "epoch": 0.6125768204109504, "grad_norm": 0.5543975340812443, "learning_rate": 9.672969738982529e-05, "loss": 3.1987, "step": 9868 }, { "epoch": 0.6126388975107083, "grad_norm": 0.5852162308758132, "learning_rate": 9.672841258956932e-05, "loss": 3.1701, "step": 9869 }, { "epoch": 0.6127009746104662, "grad_norm": 0.5237427860300348, "learning_rate": 9.672712754551987e-05, "loss": 3.241, "step": 9870 }, { "epoch": 0.612763051710224, "grad_norm": 0.5279063579162007, "learning_rate": 9.672584225768366e-05, "loss": 3.1986, "step": 9871 }, { "epoch": 0.612825128809982, "grad_norm": 0.4984693194191123, "learning_rate": 9.672455672606738e-05, "loss": 3.0206, "step": 9872 }, { "epoch": 0.6128872059097399, "grad_norm": 0.4883509624307678, "learning_rate": 9.672327095067776e-05, "loss": 3.0583, "step": 9873 }, { "epoch": 0.6129492830094978, "grad_norm": 0.4740853576078273, "learning_rate": 9.672198493152148e-05, "loss": 3.1362, "step": 9874 }, { "epoch": 0.6130113601092557, "grad_norm": 0.5052700070624876, "learning_rate": 9.672069866860527e-05, "loss": 3.1337, "step": 9875 }, { "epoch": 0.6130734372090136, "grad_norm": 0.5403726661414655, "learning_rate": 9.671941216193583e-05, "loss": 3.192, "step": 9876 }, { "epoch": 0.6131355143087714, "grad_norm": 0.4954921950005598, "learning_rate": 9.671812541151988e-05, "loss": 3.1443, "step": 9877 }, { "epoch": 0.6131975914085294, "grad_norm": 0.4807802446644326, "learning_rate": 9.671683841736411e-05, "loss": 3.1515, "step": 9878 }, { "epoch": 0.6132596685082873, "grad_norm": 0.5597746843668402, "learning_rate": 9.671555117947527e-05, "loss": 3.1695, "step": 9879 }, { "epoch": 0.6133217456080452, "grad_norm": 0.5943533692700947, "learning_rate": 9.671426369786005e-05, "loss": 3.228, "step": 9880 }, { "epoch": 0.6133838227078031, "grad_norm": 0.5392619330757545, "learning_rate": 9.671297597252519e-05, "loss": 3.1878, "step": 9881 }, { "epoch": 0.613445899807561, "grad_norm": 0.5494764578030844, "learning_rate": 9.671168800347738e-05, "loss": 3.2167, "step": 9882 }, { "epoch": 0.6135079769073188, "grad_norm": 0.4867045313195668, "learning_rate": 9.671039979072335e-05, "loss": 3.1548, "step": 9883 }, { "epoch": 0.6135700540070768, "grad_norm": 0.5051030576943798, "learning_rate": 9.670911133426986e-05, "loss": 3.1999, "step": 9884 }, { "epoch": 0.6136321311068347, "grad_norm": 0.4637231864272246, "learning_rate": 9.670782263412356e-05, "loss": 3.1621, "step": 9885 }, { "epoch": 0.6136942082065926, "grad_norm": 0.45368070070600575, "learning_rate": 9.670653369029121e-05, "loss": 3.1311, "step": 9886 }, { "epoch": 0.6137562853063505, "grad_norm": 0.4827870673977878, "learning_rate": 9.670524450277956e-05, "loss": 3.2077, "step": 9887 }, { "epoch": 0.6138183624061084, "grad_norm": 0.5189803221138192, "learning_rate": 9.67039550715953e-05, "loss": 3.1254, "step": 9888 }, { "epoch": 0.6138804395058662, "grad_norm": 0.5002010954344208, "learning_rate": 9.670266539674516e-05, "loss": 3.1292, "step": 9889 }, { "epoch": 0.6139425166056242, "grad_norm": 0.49395015498480443, "learning_rate": 9.670137547823589e-05, "loss": 3.2088, "step": 9890 }, { "epoch": 0.6140045937053821, "grad_norm": 0.5158615725567378, "learning_rate": 9.67000853160742e-05, "loss": 3.1557, "step": 9891 }, { "epoch": 0.61406667080514, "grad_norm": 0.6063243112541222, "learning_rate": 9.669879491026683e-05, "loss": 3.2198, "step": 9892 }, { "epoch": 0.6141287479048979, "grad_norm": 0.5736930932390408, "learning_rate": 9.669750426082052e-05, "loss": 3.1755, "step": 9893 }, { "epoch": 0.6141908250046558, "grad_norm": 0.5140787985864445, "learning_rate": 9.669621336774197e-05, "loss": 3.1373, "step": 9894 }, { "epoch": 0.6142529021044136, "grad_norm": 0.5944804667415232, "learning_rate": 9.669492223103795e-05, "loss": 3.1495, "step": 9895 }, { "epoch": 0.6143149792041716, "grad_norm": 0.5473586260064222, "learning_rate": 9.669363085071519e-05, "loss": 3.1384, "step": 9896 }, { "epoch": 0.6143770563039295, "grad_norm": 0.5578548906818811, "learning_rate": 9.669233922678043e-05, "loss": 3.1614, "step": 9897 }, { "epoch": 0.6144391334036874, "grad_norm": 0.6660064517645168, "learning_rate": 9.669104735924039e-05, "loss": 3.1021, "step": 9898 }, { "epoch": 0.6145012105034453, "grad_norm": 0.5263125191292047, "learning_rate": 9.668975524810182e-05, "loss": 3.0618, "step": 9899 }, { "epoch": 0.6145632876032032, "grad_norm": 0.48837116176580064, "learning_rate": 9.668846289337147e-05, "loss": 3.1503, "step": 9900 }, { "epoch": 0.614625364702961, "grad_norm": 0.5333255445636803, "learning_rate": 9.668717029505607e-05, "loss": 3.2117, "step": 9901 }, { "epoch": 0.614687441802719, "grad_norm": 0.4664848114507394, "learning_rate": 9.668587745316238e-05, "loss": 3.1063, "step": 9902 }, { "epoch": 0.6147495189024769, "grad_norm": 0.5445843817076433, "learning_rate": 9.668458436769712e-05, "loss": 3.2269, "step": 9903 }, { "epoch": 0.6148115960022348, "grad_norm": 0.4748424224734539, "learning_rate": 9.668329103866704e-05, "loss": 3.111, "step": 9904 }, { "epoch": 0.6148736731019927, "grad_norm": 0.5081335959933231, "learning_rate": 9.668199746607892e-05, "loss": 3.1525, "step": 9905 }, { "epoch": 0.6149357502017506, "grad_norm": 0.5242161723725894, "learning_rate": 9.668070364993949e-05, "loss": 3.1905, "step": 9906 }, { "epoch": 0.6149978273015084, "grad_norm": 0.5595316182816824, "learning_rate": 9.667940959025547e-05, "loss": 3.1763, "step": 9907 }, { "epoch": 0.6150599044012663, "grad_norm": 0.5339187458955041, "learning_rate": 9.667811528703366e-05, "loss": 3.2198, "step": 9908 }, { "epoch": 0.6151219815010243, "grad_norm": 0.5208393923089142, "learning_rate": 9.667682074028079e-05, "loss": 3.1501, "step": 9909 }, { "epoch": 0.6151840586007822, "grad_norm": 0.4672066234897652, "learning_rate": 9.667552595000359e-05, "loss": 3.0293, "step": 9910 }, { "epoch": 0.6152461357005401, "grad_norm": 0.5423676985605045, "learning_rate": 9.667423091620887e-05, "loss": 3.2446, "step": 9911 }, { "epoch": 0.615308212800298, "grad_norm": 0.5424259958463689, "learning_rate": 9.667293563890335e-05, "loss": 3.2254, "step": 9912 }, { "epoch": 0.6153702899000558, "grad_norm": 0.5192645255891492, "learning_rate": 9.66716401180938e-05, "loss": 3.1924, "step": 9913 }, { "epoch": 0.6154323669998137, "grad_norm": 0.520805717371909, "learning_rate": 9.667034435378696e-05, "loss": 3.1943, "step": 9914 }, { "epoch": 0.6154944440995717, "grad_norm": 0.44649198830989534, "learning_rate": 9.66690483459896e-05, "loss": 3.0856, "step": 9915 }, { "epoch": 0.6155565211993296, "grad_norm": 0.4866475814050226, "learning_rate": 9.666775209470851e-05, "loss": 3.2701, "step": 9916 }, { "epoch": 0.6156185982990875, "grad_norm": 0.5643631754552123, "learning_rate": 9.66664555999504e-05, "loss": 3.1562, "step": 9917 }, { "epoch": 0.6156806753988454, "grad_norm": 0.4892527414390313, "learning_rate": 9.666515886172209e-05, "loss": 3.0961, "step": 9918 }, { "epoch": 0.6157427524986032, "grad_norm": 0.4668079678163124, "learning_rate": 9.66638618800303e-05, "loss": 3.1482, "step": 9919 }, { "epoch": 0.6158048295983611, "grad_norm": 0.5524299241414096, "learning_rate": 9.666256465488182e-05, "loss": 3.1496, "step": 9920 }, { "epoch": 0.615866906698119, "grad_norm": 0.5859660586708852, "learning_rate": 9.66612671862834e-05, "loss": 3.1323, "step": 9921 }, { "epoch": 0.615928983797877, "grad_norm": 0.4796924833107221, "learning_rate": 9.665996947424184e-05, "loss": 3.1236, "step": 9922 }, { "epoch": 0.6159910608976349, "grad_norm": 0.42784333256586965, "learning_rate": 9.665867151876388e-05, "loss": 3.1379, "step": 9923 }, { "epoch": 0.6160531379973928, "grad_norm": 0.4611167236003155, "learning_rate": 9.66573733198563e-05, "loss": 3.0711, "step": 9924 }, { "epoch": 0.6161152150971506, "grad_norm": 0.4580948021109821, "learning_rate": 9.665607487752588e-05, "loss": 3.1824, "step": 9925 }, { "epoch": 0.6161772921969085, "grad_norm": 0.4628931037417975, "learning_rate": 9.66547761917794e-05, "loss": 3.2063, "step": 9926 }, { "epoch": 0.6162393692966665, "grad_norm": 0.4644795120738729, "learning_rate": 9.665347726262363e-05, "loss": 3.2286, "step": 9927 }, { "epoch": 0.6163014463964244, "grad_norm": 0.5844628150767925, "learning_rate": 9.665217809006533e-05, "loss": 3.1241, "step": 9928 }, { "epoch": 0.6163635234961823, "grad_norm": 0.5568583547842436, "learning_rate": 9.66508786741113e-05, "loss": 3.2435, "step": 9929 }, { "epoch": 0.6164256005959402, "grad_norm": 0.505195447095409, "learning_rate": 9.66495790147683e-05, "loss": 3.0786, "step": 9930 }, { "epoch": 0.616487677695698, "grad_norm": 0.5822690695493506, "learning_rate": 9.664827911204314e-05, "loss": 3.2004, "step": 9931 }, { "epoch": 0.6165497547954559, "grad_norm": 0.5902264092726459, "learning_rate": 9.664697896594257e-05, "loss": 3.1286, "step": 9932 }, { "epoch": 0.6166118318952138, "grad_norm": 0.4707352876557179, "learning_rate": 9.66456785764734e-05, "loss": 3.1081, "step": 9933 }, { "epoch": 0.6166739089949718, "grad_norm": 0.4871529321843243, "learning_rate": 9.664437794364238e-05, "loss": 3.156, "step": 9934 }, { "epoch": 0.6167359860947297, "grad_norm": 0.533937584167138, "learning_rate": 9.664307706745634e-05, "loss": 3.1849, "step": 9935 }, { "epoch": 0.6167980631944876, "grad_norm": 0.5249660151698953, "learning_rate": 9.664177594792203e-05, "loss": 3.2641, "step": 9936 }, { "epoch": 0.6168601402942454, "grad_norm": 0.5420724070525657, "learning_rate": 9.664047458504625e-05, "loss": 3.181, "step": 9937 }, { "epoch": 0.6169222173940033, "grad_norm": 0.5103085137309085, "learning_rate": 9.66391729788358e-05, "loss": 3.1747, "step": 9938 }, { "epoch": 0.6169842944937612, "grad_norm": 0.5290748440180639, "learning_rate": 9.663787112929746e-05, "loss": 3.1117, "step": 9939 }, { "epoch": 0.6170463715935192, "grad_norm": 0.5377789849007452, "learning_rate": 9.663656903643803e-05, "loss": 3.1408, "step": 9940 }, { "epoch": 0.6171084486932771, "grad_norm": 0.5964836385406365, "learning_rate": 9.663526670026429e-05, "loss": 3.2445, "step": 9941 }, { "epoch": 0.617170525793035, "grad_norm": 0.5716630282253814, "learning_rate": 9.663396412078305e-05, "loss": 3.1256, "step": 9942 }, { "epoch": 0.6172326028927928, "grad_norm": 0.48770383644828175, "learning_rate": 9.663266129800109e-05, "loss": 3.0962, "step": 9943 }, { "epoch": 0.6172946799925507, "grad_norm": 0.5773272551992249, "learning_rate": 9.663135823192522e-05, "loss": 3.2293, "step": 9944 }, { "epoch": 0.6173567570923086, "grad_norm": 0.6163511283025551, "learning_rate": 9.663005492256223e-05, "loss": 3.1161, "step": 9945 }, { "epoch": 0.6174188341920666, "grad_norm": 0.5643345005294091, "learning_rate": 9.662875136991893e-05, "loss": 3.1152, "step": 9946 }, { "epoch": 0.6174809112918245, "grad_norm": 0.5861238310499813, "learning_rate": 9.66274475740021e-05, "loss": 3.0395, "step": 9947 }, { "epoch": 0.6175429883915824, "grad_norm": 0.5383592800947966, "learning_rate": 9.662614353481859e-05, "loss": 3.0574, "step": 9948 }, { "epoch": 0.6176050654913402, "grad_norm": 0.47437293045939466, "learning_rate": 9.662483925237513e-05, "loss": 3.1444, "step": 9949 }, { "epoch": 0.6176671425910981, "grad_norm": 0.5714834811659294, "learning_rate": 9.662353472667859e-05, "loss": 3.1779, "step": 9950 }, { "epoch": 0.617729219690856, "grad_norm": 0.5785123712936281, "learning_rate": 9.662222995773574e-05, "loss": 3.2478, "step": 9951 }, { "epoch": 0.617791296790614, "grad_norm": 0.4948931907765789, "learning_rate": 9.662092494555341e-05, "loss": 3.1344, "step": 9952 }, { "epoch": 0.6178533738903719, "grad_norm": 0.5055257219756426, "learning_rate": 9.661961969013839e-05, "loss": 3.1875, "step": 9953 }, { "epoch": 0.6179154509901298, "grad_norm": 0.5185885948115714, "learning_rate": 9.661831419149749e-05, "loss": 3.0812, "step": 9954 }, { "epoch": 0.6179775280898876, "grad_norm": 0.5264534959116715, "learning_rate": 9.661700844963754e-05, "loss": 3.1289, "step": 9955 }, { "epoch": 0.6180396051896455, "grad_norm": 0.5536152590482316, "learning_rate": 9.661570246456533e-05, "loss": 3.137, "step": 9956 }, { "epoch": 0.6181016822894034, "grad_norm": 0.5080126441335676, "learning_rate": 9.661439623628768e-05, "loss": 3.1579, "step": 9957 }, { "epoch": 0.6181637593891613, "grad_norm": 0.5078359745175209, "learning_rate": 9.661308976481142e-05, "loss": 3.1788, "step": 9958 }, { "epoch": 0.6182258364889193, "grad_norm": 0.5654878734634189, "learning_rate": 9.661178305014335e-05, "loss": 3.0701, "step": 9959 }, { "epoch": 0.6182879135886772, "grad_norm": 0.5427248015684405, "learning_rate": 9.66104760922903e-05, "loss": 3.0971, "step": 9960 }, { "epoch": 0.618349990688435, "grad_norm": 0.5485883304303257, "learning_rate": 9.660916889125907e-05, "loss": 3.1321, "step": 9961 }, { "epoch": 0.6184120677881929, "grad_norm": 0.5645773463852602, "learning_rate": 9.66078614470565e-05, "loss": 3.1678, "step": 9962 }, { "epoch": 0.6184741448879508, "grad_norm": 0.4974978543719312, "learning_rate": 9.66065537596894e-05, "loss": 3.0492, "step": 9963 }, { "epoch": 0.6185362219877087, "grad_norm": 0.5048661957889495, "learning_rate": 9.660524582916459e-05, "loss": 3.2318, "step": 9964 }, { "epoch": 0.6185982990874667, "grad_norm": 0.6295239986438528, "learning_rate": 9.660393765548891e-05, "loss": 3.0935, "step": 9965 }, { "epoch": 0.6186603761872246, "grad_norm": 0.5707359309757879, "learning_rate": 9.660262923866917e-05, "loss": 3.1115, "step": 9966 }, { "epoch": 0.6187224532869824, "grad_norm": 0.5490142126675622, "learning_rate": 9.660132057871218e-05, "loss": 3.1448, "step": 9967 }, { "epoch": 0.6187845303867403, "grad_norm": 0.5751826182909239, "learning_rate": 9.660001167562481e-05, "loss": 3.1947, "step": 9968 }, { "epoch": 0.6188466074864982, "grad_norm": 0.5962715813591758, "learning_rate": 9.659870252941386e-05, "loss": 3.1205, "step": 9969 }, { "epoch": 0.6189086845862561, "grad_norm": 0.4840623400240314, "learning_rate": 9.659739314008618e-05, "loss": 3.2018, "step": 9970 }, { "epoch": 0.6189707616860141, "grad_norm": 0.49502543063777166, "learning_rate": 9.659608350764857e-05, "loss": 3.0942, "step": 9971 }, { "epoch": 0.619032838785772, "grad_norm": 0.4730349198208025, "learning_rate": 9.659477363210788e-05, "loss": 3.1565, "step": 9972 }, { "epoch": 0.6190949158855298, "grad_norm": 0.6200042862128718, "learning_rate": 9.659346351347096e-05, "loss": 3.1718, "step": 9973 }, { "epoch": 0.6191569929852877, "grad_norm": 0.5359715286884805, "learning_rate": 9.659215315174461e-05, "loss": 3.0377, "step": 9974 }, { "epoch": 0.6192190700850456, "grad_norm": 0.49071181555772964, "learning_rate": 9.65908425469357e-05, "loss": 3.0262, "step": 9975 }, { "epoch": 0.6192811471848035, "grad_norm": 0.5678742425885398, "learning_rate": 9.658953169905105e-05, "loss": 3.0831, "step": 9976 }, { "epoch": 0.6193432242845615, "grad_norm": 0.557178732460177, "learning_rate": 9.658822060809751e-05, "loss": 3.2179, "step": 9977 }, { "epoch": 0.6194053013843194, "grad_norm": 0.5411588018024232, "learning_rate": 9.65869092740819e-05, "loss": 3.2347, "step": 9978 }, { "epoch": 0.6194673784840772, "grad_norm": 0.5007599479441338, "learning_rate": 9.658559769701109e-05, "loss": 3.2566, "step": 9979 }, { "epoch": 0.6195294555838351, "grad_norm": 0.5240848011965381, "learning_rate": 9.65842858768919e-05, "loss": 3.0852, "step": 9980 }, { "epoch": 0.619591532683593, "grad_norm": 0.5453880530978267, "learning_rate": 9.658297381373118e-05, "loss": 3.2563, "step": 9981 }, { "epoch": 0.6196536097833509, "grad_norm": 0.5941346328102123, "learning_rate": 9.658166150753578e-05, "loss": 3.1731, "step": 9982 }, { "epoch": 0.6197156868831089, "grad_norm": 0.5083266956125274, "learning_rate": 9.658034895831254e-05, "loss": 3.1101, "step": 9983 }, { "epoch": 0.6197777639828668, "grad_norm": 0.4574356897940888, "learning_rate": 9.657903616606832e-05, "loss": 3.1966, "step": 9984 }, { "epoch": 0.6198398410826246, "grad_norm": 0.5048843767448115, "learning_rate": 9.657772313080995e-05, "loss": 3.1585, "step": 9985 }, { "epoch": 0.6199019181823825, "grad_norm": 0.592865159104462, "learning_rate": 9.65764098525443e-05, "loss": 3.1981, "step": 9986 }, { "epoch": 0.6199639952821404, "grad_norm": 0.5581546471422552, "learning_rate": 9.657509633127822e-05, "loss": 3.2032, "step": 9987 }, { "epoch": 0.6200260723818983, "grad_norm": 0.5477836145161399, "learning_rate": 9.657378256701855e-05, "loss": 3.0978, "step": 9988 }, { "epoch": 0.6200881494816562, "grad_norm": 0.5116453627749845, "learning_rate": 9.657246855977213e-05, "loss": 3.1242, "step": 9989 }, { "epoch": 0.6201502265814142, "grad_norm": 0.5651375076607484, "learning_rate": 9.657115430954586e-05, "loss": 3.167, "step": 9990 }, { "epoch": 0.620212303681172, "grad_norm": 0.5767432531682302, "learning_rate": 9.656983981634657e-05, "loss": 3.2058, "step": 9991 }, { "epoch": 0.6202743807809299, "grad_norm": 0.5739402095427887, "learning_rate": 9.65685250801811e-05, "loss": 3.1987, "step": 9992 }, { "epoch": 0.6203364578806878, "grad_norm": 0.5552243603175833, "learning_rate": 9.656721010105634e-05, "loss": 3.1419, "step": 9993 }, { "epoch": 0.6203985349804457, "grad_norm": 0.5451646489605855, "learning_rate": 9.656589487897915e-05, "loss": 3.1146, "step": 9994 }, { "epoch": 0.6204606120802036, "grad_norm": 0.5508342473930974, "learning_rate": 9.656457941395638e-05, "loss": 3.0644, "step": 9995 }, { "epoch": 0.6205226891799616, "grad_norm": 0.489853319503542, "learning_rate": 9.656326370599488e-05, "loss": 3.1266, "step": 9996 }, { "epoch": 0.6205847662797194, "grad_norm": 0.5065477614050099, "learning_rate": 9.656194775510156e-05, "loss": 3.2082, "step": 9997 }, { "epoch": 0.6206468433794773, "grad_norm": 0.5144765374975793, "learning_rate": 9.656063156128323e-05, "loss": 3.1435, "step": 9998 }, { "epoch": 0.6207089204792352, "grad_norm": 0.4856056607444748, "learning_rate": 9.655931512454679e-05, "loss": 3.1359, "step": 9999 }, { "epoch": 0.6207709975789931, "grad_norm": 0.4813373746449124, "learning_rate": 9.65579984448991e-05, "loss": 3.1898, "step": 10000 }, { "epoch": 0.620833074678751, "grad_norm": 0.4772387069027441, "learning_rate": 9.655668152234702e-05, "loss": 3.1499, "step": 10001 }, { "epoch": 0.620895151778509, "grad_norm": 0.44004309153331406, "learning_rate": 9.655536435689745e-05, "loss": 3.0924, "step": 10002 }, { "epoch": 0.6209572288782668, "grad_norm": 0.4555769548084166, "learning_rate": 9.655404694855724e-05, "loss": 3.1095, "step": 10003 }, { "epoch": 0.6210193059780247, "grad_norm": 0.4824964334425277, "learning_rate": 9.655272929733325e-05, "loss": 3.1647, "step": 10004 }, { "epoch": 0.6210813830777826, "grad_norm": 0.47088203354137653, "learning_rate": 9.655141140323238e-05, "loss": 3.1953, "step": 10005 }, { "epoch": 0.6211434601775405, "grad_norm": 0.5089468976595678, "learning_rate": 9.65500932662615e-05, "loss": 3.1711, "step": 10006 }, { "epoch": 0.6212055372772984, "grad_norm": 0.4950860369210981, "learning_rate": 9.654877488642747e-05, "loss": 3.2469, "step": 10007 }, { "epoch": 0.6212676143770564, "grad_norm": 0.45006229976039863, "learning_rate": 9.65474562637372e-05, "loss": 3.1544, "step": 10008 }, { "epoch": 0.6213296914768142, "grad_norm": 0.511283922725528, "learning_rate": 9.654613739819754e-05, "loss": 3.1804, "step": 10009 }, { "epoch": 0.6213917685765721, "grad_norm": 0.5069114614089111, "learning_rate": 9.654481828981536e-05, "loss": 3.1044, "step": 10010 }, { "epoch": 0.62145384567633, "grad_norm": 0.49973410748663427, "learning_rate": 9.65434989385976e-05, "loss": 3.0955, "step": 10011 }, { "epoch": 0.6215159227760879, "grad_norm": 0.47771686138199726, "learning_rate": 9.654217934455109e-05, "loss": 3.0085, "step": 10012 }, { "epoch": 0.6215779998758458, "grad_norm": 0.48860993575956635, "learning_rate": 9.654085950768273e-05, "loss": 3.1845, "step": 10013 }, { "epoch": 0.6216400769756038, "grad_norm": 0.5320543763096066, "learning_rate": 9.653953942799941e-05, "loss": 3.2108, "step": 10014 }, { "epoch": 0.6217021540753616, "grad_norm": 0.5254179936764583, "learning_rate": 9.653821910550802e-05, "loss": 3.191, "step": 10015 }, { "epoch": 0.6217642311751195, "grad_norm": 0.4843663553266361, "learning_rate": 9.653689854021543e-05, "loss": 3.2246, "step": 10016 }, { "epoch": 0.6218263082748774, "grad_norm": 0.5239860450236842, "learning_rate": 9.653557773212856e-05, "loss": 3.1382, "step": 10017 }, { "epoch": 0.6218883853746353, "grad_norm": 0.5188983130716694, "learning_rate": 9.653425668125429e-05, "loss": 3.1729, "step": 10018 }, { "epoch": 0.6219504624743932, "grad_norm": 0.4511787942463119, "learning_rate": 9.653293538759949e-05, "loss": 3.1617, "step": 10019 }, { "epoch": 0.6220125395741511, "grad_norm": 0.5410702949747659, "learning_rate": 9.653161385117107e-05, "loss": 3.1198, "step": 10020 }, { "epoch": 0.622074616673909, "grad_norm": 0.5127689853596115, "learning_rate": 9.653029207197592e-05, "loss": 3.1775, "step": 10021 }, { "epoch": 0.6221366937736669, "grad_norm": 0.4914215576747221, "learning_rate": 9.652897005002095e-05, "loss": 3.1457, "step": 10022 }, { "epoch": 0.6221987708734248, "grad_norm": 0.4858986670380605, "learning_rate": 9.652764778531304e-05, "loss": 3.0438, "step": 10023 }, { "epoch": 0.6222608479731827, "grad_norm": 0.48203013032013425, "learning_rate": 9.652632527785912e-05, "loss": 3.2283, "step": 10024 }, { "epoch": 0.6223229250729406, "grad_norm": 0.5028574475006475, "learning_rate": 9.652500252766604e-05, "loss": 3.1974, "step": 10025 }, { "epoch": 0.6223850021726985, "grad_norm": 0.428828681643822, "learning_rate": 9.652367953474074e-05, "loss": 3.047, "step": 10026 }, { "epoch": 0.6224470792724563, "grad_norm": 0.49336335338398973, "learning_rate": 9.65223562990901e-05, "loss": 3.1229, "step": 10027 }, { "epoch": 0.6225091563722143, "grad_norm": 0.47817423573222123, "learning_rate": 9.652103282072104e-05, "loss": 3.0826, "step": 10028 }, { "epoch": 0.6225712334719722, "grad_norm": 0.49361201422489526, "learning_rate": 9.651970909964047e-05, "loss": 3.1856, "step": 10029 }, { "epoch": 0.6226333105717301, "grad_norm": 0.46223288696883075, "learning_rate": 9.651838513585527e-05, "loss": 3.22, "step": 10030 }, { "epoch": 0.622695387671488, "grad_norm": 0.45152292177138026, "learning_rate": 9.651706092937236e-05, "loss": 3.1054, "step": 10031 }, { "epoch": 0.6227574647712459, "grad_norm": 0.4498240332143233, "learning_rate": 9.651573648019866e-05, "loss": 3.1721, "step": 10032 }, { "epoch": 0.6228195418710037, "grad_norm": 0.4850661515072778, "learning_rate": 9.651441178834106e-05, "loss": 3.1856, "step": 10033 }, { "epoch": 0.6228816189707617, "grad_norm": 0.44680050649166003, "learning_rate": 9.651308685380648e-05, "loss": 3.1601, "step": 10034 }, { "epoch": 0.6229436960705196, "grad_norm": 0.47403012944084716, "learning_rate": 9.651176167660185e-05, "loss": 3.1393, "step": 10035 }, { "epoch": 0.6230057731702775, "grad_norm": 0.5579948415774155, "learning_rate": 9.651043625673406e-05, "loss": 3.144, "step": 10036 }, { "epoch": 0.6230678502700354, "grad_norm": 0.5248593618848274, "learning_rate": 9.650911059421004e-05, "loss": 3.1352, "step": 10037 }, { "epoch": 0.6231299273697933, "grad_norm": 0.5078805144318344, "learning_rate": 9.650778468903669e-05, "loss": 3.2313, "step": 10038 }, { "epoch": 0.6231920044695511, "grad_norm": 0.4217542369507598, "learning_rate": 9.650645854122096e-05, "loss": 3.1758, "step": 10039 }, { "epoch": 0.6232540815693091, "grad_norm": 0.5479494439197513, "learning_rate": 9.650513215076972e-05, "loss": 3.1456, "step": 10040 }, { "epoch": 0.623316158669067, "grad_norm": 0.5729827293685895, "learning_rate": 9.650380551768991e-05, "loss": 3.1278, "step": 10041 }, { "epoch": 0.6233782357688249, "grad_norm": 0.4627704061855025, "learning_rate": 9.650247864198848e-05, "loss": 3.219, "step": 10042 }, { "epoch": 0.6234403128685828, "grad_norm": 0.4549298351809256, "learning_rate": 9.650115152367232e-05, "loss": 3.1401, "step": 10043 }, { "epoch": 0.6235023899683407, "grad_norm": 0.5588289579291678, "learning_rate": 9.649982416274837e-05, "loss": 3.1727, "step": 10044 }, { "epoch": 0.6235644670680985, "grad_norm": 0.5050890444160097, "learning_rate": 9.649849655922355e-05, "loss": 3.2199, "step": 10045 }, { "epoch": 0.6236265441678565, "grad_norm": 0.4768395947203375, "learning_rate": 9.649716871310477e-05, "loss": 3.103, "step": 10046 }, { "epoch": 0.6236886212676144, "grad_norm": 0.4741378765494815, "learning_rate": 9.649584062439898e-05, "loss": 3.0653, "step": 10047 }, { "epoch": 0.6237506983673723, "grad_norm": 0.47376894223534, "learning_rate": 9.649451229311311e-05, "loss": 3.0913, "step": 10048 }, { "epoch": 0.6238127754671302, "grad_norm": 0.519156528800495, "learning_rate": 9.649318371925407e-05, "loss": 3.1544, "step": 10049 }, { "epoch": 0.6238748525668881, "grad_norm": 0.42672139237157397, "learning_rate": 9.649185490282882e-05, "loss": 3.0995, "step": 10050 }, { "epoch": 0.6239369296666459, "grad_norm": 0.4908630322142982, "learning_rate": 9.649052584384426e-05, "loss": 3.1327, "step": 10051 }, { "epoch": 0.6239990067664039, "grad_norm": 0.4946175787665189, "learning_rate": 9.648919654230735e-05, "loss": 3.1872, "step": 10052 }, { "epoch": 0.6240610838661618, "grad_norm": 0.573990705192222, "learning_rate": 9.6487866998225e-05, "loss": 3.1965, "step": 10053 }, { "epoch": 0.6241231609659197, "grad_norm": 0.4907497721150239, "learning_rate": 9.648653721160419e-05, "loss": 3.1555, "step": 10054 }, { "epoch": 0.6241852380656776, "grad_norm": 0.48727768572837843, "learning_rate": 9.648520718245181e-05, "loss": 3.2404, "step": 10055 }, { "epoch": 0.6242473151654355, "grad_norm": 0.48074379468534356, "learning_rate": 9.648387691077483e-05, "loss": 3.2, "step": 10056 }, { "epoch": 0.6243093922651933, "grad_norm": 0.5836465426137785, "learning_rate": 9.648254639658019e-05, "loss": 3.1068, "step": 10057 }, { "epoch": 0.6243714693649512, "grad_norm": 0.5158347367938808, "learning_rate": 9.64812156398748e-05, "loss": 3.0816, "step": 10058 }, { "epoch": 0.6244335464647092, "grad_norm": 0.4198486264239856, "learning_rate": 9.647988464066564e-05, "loss": 3.1317, "step": 10059 }, { "epoch": 0.6244956235644671, "grad_norm": 0.4936169348449336, "learning_rate": 9.647855339895961e-05, "loss": 3.2124, "step": 10060 }, { "epoch": 0.624557700664225, "grad_norm": 0.5513656167559625, "learning_rate": 9.647722191476371e-05, "loss": 3.1534, "step": 10061 }, { "epoch": 0.6246197777639829, "grad_norm": 0.48931697099479726, "learning_rate": 9.647589018808487e-05, "loss": 3.1646, "step": 10062 }, { "epoch": 0.6246818548637407, "grad_norm": 0.4824869675679828, "learning_rate": 9.647455821893001e-05, "loss": 3.0396, "step": 10063 }, { "epoch": 0.6247439319634986, "grad_norm": 0.5302591339486488, "learning_rate": 9.64732260073061e-05, "loss": 3.2439, "step": 10064 }, { "epoch": 0.6248060090632566, "grad_norm": 0.5544934269057151, "learning_rate": 9.647189355322008e-05, "loss": 3.1405, "step": 10065 }, { "epoch": 0.6248680861630145, "grad_norm": 0.4452336392092846, "learning_rate": 9.647056085667893e-05, "loss": 3.0874, "step": 10066 }, { "epoch": 0.6249301632627724, "grad_norm": 0.5166467888402533, "learning_rate": 9.646922791768958e-05, "loss": 3.1399, "step": 10067 }, { "epoch": 0.6249922403625303, "grad_norm": 0.5289589623197963, "learning_rate": 9.646789473625897e-05, "loss": 3.1114, "step": 10068 }, { "epoch": 0.6250543174622881, "grad_norm": 0.44720536871397515, "learning_rate": 9.646656131239407e-05, "loss": 3.1879, "step": 10069 }, { "epoch": 0.625116394562046, "grad_norm": 0.531454271425371, "learning_rate": 9.646522764610186e-05, "loss": 3.0516, "step": 10070 }, { "epoch": 0.625178471661804, "grad_norm": 0.6222412018374256, "learning_rate": 9.646389373738926e-05, "loss": 3.1569, "step": 10071 }, { "epoch": 0.6252405487615619, "grad_norm": 0.4934580011922581, "learning_rate": 9.646255958626327e-05, "loss": 3.1919, "step": 10072 }, { "epoch": 0.6253026258613198, "grad_norm": 0.48439919503603424, "learning_rate": 9.64612251927308e-05, "loss": 3.1231, "step": 10073 }, { "epoch": 0.6253647029610777, "grad_norm": 0.4939222644935789, "learning_rate": 9.645989055679886e-05, "loss": 3.1039, "step": 10074 }, { "epoch": 0.6254267800608355, "grad_norm": 0.5254288273078922, "learning_rate": 9.645855567847439e-05, "loss": 3.1451, "step": 10075 }, { "epoch": 0.6254888571605934, "grad_norm": 0.4922035305500706, "learning_rate": 9.645722055776436e-05, "loss": 3.1674, "step": 10076 }, { "epoch": 0.6255509342603514, "grad_norm": 0.5729893349309085, "learning_rate": 9.645588519467571e-05, "loss": 3.1488, "step": 10077 }, { "epoch": 0.6256130113601093, "grad_norm": 0.4990024300948009, "learning_rate": 9.645454958921545e-05, "loss": 3.148, "step": 10078 }, { "epoch": 0.6256750884598672, "grad_norm": 0.5329000988319321, "learning_rate": 9.645321374139052e-05, "loss": 3.0989, "step": 10079 }, { "epoch": 0.6257371655596251, "grad_norm": 0.563443901926951, "learning_rate": 9.64518776512079e-05, "loss": 3.0958, "step": 10080 }, { "epoch": 0.6257992426593829, "grad_norm": 0.5335889545186089, "learning_rate": 9.645054131867455e-05, "loss": 3.144, "step": 10081 }, { "epoch": 0.6258613197591408, "grad_norm": 0.4786771540586494, "learning_rate": 9.644920474379746e-05, "loss": 3.1131, "step": 10082 }, { "epoch": 0.6259233968588988, "grad_norm": 0.5041509817641813, "learning_rate": 9.644786792658358e-05, "loss": 3.105, "step": 10083 }, { "epoch": 0.6259854739586567, "grad_norm": 0.5136479129206342, "learning_rate": 9.644653086703992e-05, "loss": 3.1694, "step": 10084 }, { "epoch": 0.6260475510584146, "grad_norm": 0.5921466913983287, "learning_rate": 9.644519356517341e-05, "loss": 3.0982, "step": 10085 }, { "epoch": 0.6261096281581725, "grad_norm": 0.5407578442922109, "learning_rate": 9.644385602099107e-05, "loss": 3.1733, "step": 10086 }, { "epoch": 0.6261717052579303, "grad_norm": 0.5414801300855515, "learning_rate": 9.644251823449984e-05, "loss": 3.2405, "step": 10087 }, { "epoch": 0.6262337823576882, "grad_norm": 0.5291252843440153, "learning_rate": 9.644118020570673e-05, "loss": 3.0846, "step": 10088 }, { "epoch": 0.6262958594574461, "grad_norm": 0.4974282106044287, "learning_rate": 9.64398419346187e-05, "loss": 3.1609, "step": 10089 }, { "epoch": 0.6263579365572041, "grad_norm": 0.5738679761662412, "learning_rate": 9.643850342124275e-05, "loss": 3.2514, "step": 10090 }, { "epoch": 0.626420013656962, "grad_norm": 0.570795168494178, "learning_rate": 9.643716466558584e-05, "loss": 3.1376, "step": 10091 }, { "epoch": 0.6264820907567199, "grad_norm": 0.49764666465738494, "learning_rate": 9.6435825667655e-05, "loss": 3.0772, "step": 10092 }, { "epoch": 0.6265441678564777, "grad_norm": 0.515192197008899, "learning_rate": 9.643448642745714e-05, "loss": 3.2261, "step": 10093 }, { "epoch": 0.6266062449562356, "grad_norm": 0.43965413900344835, "learning_rate": 9.643314694499932e-05, "loss": 3.1439, "step": 10094 }, { "epoch": 0.6266683220559935, "grad_norm": 0.5556769344401026, "learning_rate": 9.643180722028851e-05, "loss": 3.157, "step": 10095 }, { "epoch": 0.6267303991557515, "grad_norm": 0.4344340935997355, "learning_rate": 9.643046725333167e-05, "loss": 3.0331, "step": 10096 }, { "epoch": 0.6267924762555094, "grad_norm": 0.558781918930371, "learning_rate": 9.642912704413581e-05, "loss": 3.1792, "step": 10097 }, { "epoch": 0.6268545533552673, "grad_norm": 0.5300126532177825, "learning_rate": 9.642778659270794e-05, "loss": 3.2154, "step": 10098 }, { "epoch": 0.6269166304550251, "grad_norm": 0.47808340365410606, "learning_rate": 9.642644589905503e-05, "loss": 3.1499, "step": 10099 }, { "epoch": 0.626978707554783, "grad_norm": 0.4962729183064773, "learning_rate": 9.642510496318407e-05, "loss": 3.0418, "step": 10100 }, { "epoch": 0.6270407846545409, "grad_norm": 0.46493896342715263, "learning_rate": 9.642376378510207e-05, "loss": 2.983, "step": 10101 }, { "epoch": 0.6271028617542989, "grad_norm": 0.48007136092608244, "learning_rate": 9.642242236481603e-05, "loss": 3.0892, "step": 10102 }, { "epoch": 0.6271649388540568, "grad_norm": 0.4890749164991693, "learning_rate": 9.642108070233295e-05, "loss": 3.1849, "step": 10103 }, { "epoch": 0.6272270159538147, "grad_norm": 0.46470806802305226, "learning_rate": 9.641973879765982e-05, "loss": 3.1838, "step": 10104 }, { "epoch": 0.6272890930535725, "grad_norm": 0.5012010811125047, "learning_rate": 9.641839665080363e-05, "loss": 3.0755, "step": 10105 }, { "epoch": 0.6273511701533304, "grad_norm": 0.5164883062178148, "learning_rate": 9.641705426177141e-05, "loss": 3.1294, "step": 10106 }, { "epoch": 0.6274132472530883, "grad_norm": 0.5816196804233779, "learning_rate": 9.641571163057015e-05, "loss": 3.0721, "step": 10107 }, { "epoch": 0.6274753243528463, "grad_norm": 0.5529813720090411, "learning_rate": 9.641436875720685e-05, "loss": 3.1009, "step": 10108 }, { "epoch": 0.6275374014526042, "grad_norm": 0.479745087288834, "learning_rate": 9.641302564168852e-05, "loss": 3.1124, "step": 10109 }, { "epoch": 0.6275994785523621, "grad_norm": 0.5432482487175876, "learning_rate": 9.641168228402218e-05, "loss": 3.1735, "step": 10110 }, { "epoch": 0.6276615556521199, "grad_norm": 0.566510026566946, "learning_rate": 9.641033868421481e-05, "loss": 3.2065, "step": 10111 }, { "epoch": 0.6277236327518778, "grad_norm": 0.5295442798279983, "learning_rate": 9.640899484227344e-05, "loss": 3.1415, "step": 10112 }, { "epoch": 0.6277857098516357, "grad_norm": 0.5236412904217408, "learning_rate": 9.640765075820508e-05, "loss": 3.1571, "step": 10113 }, { "epoch": 0.6278477869513936, "grad_norm": 0.5106336430876867, "learning_rate": 9.640630643201674e-05, "loss": 3.2093, "step": 10114 }, { "epoch": 0.6279098640511516, "grad_norm": 0.6060006051453947, "learning_rate": 9.640496186371543e-05, "loss": 3.1722, "step": 10115 }, { "epoch": 0.6279719411509094, "grad_norm": 0.5571716452678626, "learning_rate": 9.640361705330818e-05, "loss": 3.1591, "step": 10116 }, { "epoch": 0.6280340182506673, "grad_norm": 0.5501970750850222, "learning_rate": 9.640227200080198e-05, "loss": 3.0262, "step": 10117 }, { "epoch": 0.6280960953504252, "grad_norm": 0.4861283255738232, "learning_rate": 9.640092670620388e-05, "loss": 3.1388, "step": 10118 }, { "epoch": 0.6281581724501831, "grad_norm": 0.5791548500182022, "learning_rate": 9.639958116952087e-05, "loss": 3.0163, "step": 10119 }, { "epoch": 0.628220249549941, "grad_norm": 0.5453640090968018, "learning_rate": 9.639823539075998e-05, "loss": 3.049, "step": 10120 }, { "epoch": 0.628282326649699, "grad_norm": 0.48750460084553643, "learning_rate": 9.639688936992824e-05, "loss": 3.1803, "step": 10121 }, { "epoch": 0.6283444037494568, "grad_norm": 0.5802280219508277, "learning_rate": 9.639554310703266e-05, "loss": 3.1907, "step": 10122 }, { "epoch": 0.6284064808492147, "grad_norm": 0.5746294745502765, "learning_rate": 9.639419660208026e-05, "loss": 3.1551, "step": 10123 }, { "epoch": 0.6284685579489726, "grad_norm": 0.5119874554490578, "learning_rate": 9.639284985507808e-05, "loss": 3.0488, "step": 10124 }, { "epoch": 0.6285306350487305, "grad_norm": 0.49868934929429887, "learning_rate": 9.639150286603316e-05, "loss": 3.1389, "step": 10125 }, { "epoch": 0.6285927121484884, "grad_norm": 0.5471869804950646, "learning_rate": 9.639015563495248e-05, "loss": 3.0982, "step": 10126 }, { "epoch": 0.6286547892482464, "grad_norm": 0.5089319361668695, "learning_rate": 9.63888081618431e-05, "loss": 3.156, "step": 10127 }, { "epoch": 0.6287168663480042, "grad_norm": 0.48711201866674503, "learning_rate": 9.638746044671205e-05, "loss": 3.1072, "step": 10128 }, { "epoch": 0.6287789434477621, "grad_norm": 0.4562942071949089, "learning_rate": 9.638611248956638e-05, "loss": 3.0723, "step": 10129 }, { "epoch": 0.62884102054752, "grad_norm": 0.5243554109135765, "learning_rate": 9.638476429041308e-05, "loss": 3.1387, "step": 10130 }, { "epoch": 0.6289030976472779, "grad_norm": 0.48065191359746673, "learning_rate": 9.638341584925919e-05, "loss": 3.1591, "step": 10131 }, { "epoch": 0.6289651747470358, "grad_norm": 0.4764256515380773, "learning_rate": 9.638206716611178e-05, "loss": 3.0379, "step": 10132 }, { "epoch": 0.6290272518467938, "grad_norm": 0.4294348927889102, "learning_rate": 9.638071824097786e-05, "loss": 3.1369, "step": 10133 }, { "epoch": 0.6290893289465516, "grad_norm": 0.48634272480693425, "learning_rate": 9.637936907386448e-05, "loss": 3.2093, "step": 10134 }, { "epoch": 0.6291514060463095, "grad_norm": 0.4985879688890126, "learning_rate": 9.637801966477867e-05, "loss": 3.1782, "step": 10135 }, { "epoch": 0.6292134831460674, "grad_norm": 0.5544195199441282, "learning_rate": 9.637667001372747e-05, "loss": 3.0634, "step": 10136 }, { "epoch": 0.6292755602458253, "grad_norm": 0.5276338979987037, "learning_rate": 9.637532012071792e-05, "loss": 3.1799, "step": 10137 }, { "epoch": 0.6293376373455832, "grad_norm": 0.5286682731739103, "learning_rate": 9.637396998575707e-05, "loss": 3.1556, "step": 10138 }, { "epoch": 0.6293997144453412, "grad_norm": 0.5223842956800321, "learning_rate": 9.637261960885197e-05, "loss": 3.1131, "step": 10139 }, { "epoch": 0.629461791545099, "grad_norm": 0.4451152130257241, "learning_rate": 9.637126899000964e-05, "loss": 3.0651, "step": 10140 }, { "epoch": 0.6295238686448569, "grad_norm": 0.5603014844161575, "learning_rate": 9.636991812923716e-05, "loss": 3.1658, "step": 10141 }, { "epoch": 0.6295859457446148, "grad_norm": 0.4851910116424122, "learning_rate": 9.636856702654156e-05, "loss": 3.0881, "step": 10142 }, { "epoch": 0.6296480228443727, "grad_norm": 0.43886104115968366, "learning_rate": 9.636721568192988e-05, "loss": 3.1485, "step": 10143 }, { "epoch": 0.6297100999441306, "grad_norm": 0.5136758202230862, "learning_rate": 9.636586409540917e-05, "loss": 3.1613, "step": 10144 }, { "epoch": 0.6297721770438885, "grad_norm": 0.5200232277322405, "learning_rate": 9.636451226698651e-05, "loss": 3.2011, "step": 10145 }, { "epoch": 0.6298342541436464, "grad_norm": 0.5588492762721081, "learning_rate": 9.636316019666892e-05, "loss": 3.1707, "step": 10146 }, { "epoch": 0.6298963312434043, "grad_norm": 0.528196930360234, "learning_rate": 9.636180788446348e-05, "loss": 3.1309, "step": 10147 }, { "epoch": 0.6299584083431622, "grad_norm": 0.46472871783583103, "learning_rate": 9.636045533037724e-05, "loss": 3.0462, "step": 10148 }, { "epoch": 0.6300204854429201, "grad_norm": 0.46160686512148436, "learning_rate": 9.635910253441723e-05, "loss": 3.0817, "step": 10149 }, { "epoch": 0.630082562542678, "grad_norm": 0.4941072415017892, "learning_rate": 9.635774949659054e-05, "loss": 3.0056, "step": 10150 }, { "epoch": 0.6301446396424359, "grad_norm": 0.5623919330208965, "learning_rate": 9.63563962169042e-05, "loss": 3.1334, "step": 10151 }, { "epoch": 0.6302067167421938, "grad_norm": 0.5943509874998836, "learning_rate": 9.63550426953653e-05, "loss": 3.1563, "step": 10152 }, { "epoch": 0.6302687938419517, "grad_norm": 0.5874175389227865, "learning_rate": 9.635368893198088e-05, "loss": 3.1125, "step": 10153 }, { "epoch": 0.6303308709417096, "grad_norm": 0.6204052593915215, "learning_rate": 9.635233492675804e-05, "loss": 3.1855, "step": 10154 }, { "epoch": 0.6303929480414675, "grad_norm": 0.5364055669320722, "learning_rate": 9.63509806797038e-05, "loss": 3.1538, "step": 10155 }, { "epoch": 0.6304550251412254, "grad_norm": 0.5177987833917145, "learning_rate": 9.634962619082523e-05, "loss": 3.1607, "step": 10156 }, { "epoch": 0.6305171022409833, "grad_norm": 0.590585482249175, "learning_rate": 9.634827146012942e-05, "loss": 3.18, "step": 10157 }, { "epoch": 0.6305791793407411, "grad_norm": 0.5114448047047757, "learning_rate": 9.634691648762342e-05, "loss": 3.0966, "step": 10158 }, { "epoch": 0.6306412564404991, "grad_norm": 0.5608214659472079, "learning_rate": 9.634556127331431e-05, "loss": 3.2204, "step": 10159 }, { "epoch": 0.630703333540257, "grad_norm": 0.5071343809317492, "learning_rate": 9.634420581720917e-05, "loss": 3.109, "step": 10160 }, { "epoch": 0.6307654106400149, "grad_norm": 0.49400603681709754, "learning_rate": 9.634285011931504e-05, "loss": 3.098, "step": 10161 }, { "epoch": 0.6308274877397728, "grad_norm": 0.5005776227899771, "learning_rate": 9.634149417963901e-05, "loss": 3.1033, "step": 10162 }, { "epoch": 0.6308895648395307, "grad_norm": 0.5081613662769011, "learning_rate": 9.634013799818817e-05, "loss": 3.1374, "step": 10163 }, { "epoch": 0.6309516419392885, "grad_norm": 0.4581090168682377, "learning_rate": 9.633878157496956e-05, "loss": 3.2077, "step": 10164 }, { "epoch": 0.6310137190390465, "grad_norm": 0.46460682577200674, "learning_rate": 9.633742490999029e-05, "loss": 3.0698, "step": 10165 }, { "epoch": 0.6310757961388044, "grad_norm": 0.47279560777500723, "learning_rate": 9.633606800325742e-05, "loss": 3.1305, "step": 10166 }, { "epoch": 0.6311378732385623, "grad_norm": 0.5625076293427919, "learning_rate": 9.633471085477805e-05, "loss": 3.1209, "step": 10167 }, { "epoch": 0.6311999503383202, "grad_norm": 0.5507487564782828, "learning_rate": 9.633335346455923e-05, "loss": 3.1062, "step": 10168 }, { "epoch": 0.6312620274380781, "grad_norm": 0.4926183570688823, "learning_rate": 9.633199583260806e-05, "loss": 3.1077, "step": 10169 }, { "epoch": 0.6313241045378359, "grad_norm": 0.5238969409182704, "learning_rate": 9.633063795893162e-05, "loss": 3.1888, "step": 10170 }, { "epoch": 0.6313861816375939, "grad_norm": 0.47077974511028514, "learning_rate": 9.632927984353701e-05, "loss": 3.0146, "step": 10171 }, { "epoch": 0.6314482587373518, "grad_norm": 0.5919724760986949, "learning_rate": 9.632792148643128e-05, "loss": 3.1321, "step": 10172 }, { "epoch": 0.6315103358371097, "grad_norm": 0.48340683746746077, "learning_rate": 9.632656288762154e-05, "loss": 3.1034, "step": 10173 }, { "epoch": 0.6315724129368676, "grad_norm": 0.5081043871725648, "learning_rate": 9.632520404711487e-05, "loss": 3.1206, "step": 10174 }, { "epoch": 0.6316344900366255, "grad_norm": 0.5822727803105798, "learning_rate": 9.632384496491838e-05, "loss": 3.1544, "step": 10175 }, { "epoch": 0.6316965671363833, "grad_norm": 0.6755847976114715, "learning_rate": 9.632248564103915e-05, "loss": 3.103, "step": 10176 }, { "epoch": 0.6317586442361413, "grad_norm": 0.56194842634248, "learning_rate": 9.632112607548424e-05, "loss": 3.0931, "step": 10177 }, { "epoch": 0.6318207213358992, "grad_norm": 0.5448129743368524, "learning_rate": 9.63197662682608e-05, "loss": 3.0909, "step": 10178 }, { "epoch": 0.6318827984356571, "grad_norm": 0.6867894488885548, "learning_rate": 9.631840621937587e-05, "loss": 3.1898, "step": 10179 }, { "epoch": 0.631944875535415, "grad_norm": 0.503001680128655, "learning_rate": 9.631704592883659e-05, "loss": 3.0103, "step": 10180 }, { "epoch": 0.6320069526351729, "grad_norm": 0.5480656604986758, "learning_rate": 9.631568539665003e-05, "loss": 3.1559, "step": 10181 }, { "epoch": 0.6320690297349307, "grad_norm": 0.6003128299185995, "learning_rate": 9.631432462282328e-05, "loss": 3.1769, "step": 10182 }, { "epoch": 0.6321311068346886, "grad_norm": 0.5301361749357658, "learning_rate": 9.631296360736348e-05, "loss": 3.0864, "step": 10183 }, { "epoch": 0.6321931839344466, "grad_norm": 0.5377227055686847, "learning_rate": 9.63116023502777e-05, "loss": 3.1027, "step": 10184 }, { "epoch": 0.6322552610342045, "grad_norm": 0.5485238966313397, "learning_rate": 9.631024085157302e-05, "loss": 3.1288, "step": 10185 }, { "epoch": 0.6323173381339624, "grad_norm": 0.5887578606536352, "learning_rate": 9.630887911125661e-05, "loss": 3.1328, "step": 10186 }, { "epoch": 0.6323794152337203, "grad_norm": 0.4767963586060954, "learning_rate": 9.630751712933553e-05, "loss": 3.1631, "step": 10187 }, { "epoch": 0.6324414923334781, "grad_norm": 0.523873816310207, "learning_rate": 9.630615490581688e-05, "loss": 3.1182, "step": 10188 }, { "epoch": 0.632503569433236, "grad_norm": 0.5158378566313548, "learning_rate": 9.630479244070778e-05, "loss": 3.1872, "step": 10189 }, { "epoch": 0.632565646532994, "grad_norm": 0.552096777578318, "learning_rate": 9.630342973401534e-05, "loss": 3.1466, "step": 10190 }, { "epoch": 0.6326277236327519, "grad_norm": 0.5469417531281094, "learning_rate": 9.630206678574665e-05, "loss": 3.1297, "step": 10191 }, { "epoch": 0.6326898007325098, "grad_norm": 0.52831630740525, "learning_rate": 9.630070359590885e-05, "loss": 3.1433, "step": 10192 }, { "epoch": 0.6327518778322677, "grad_norm": 0.5998136439597218, "learning_rate": 9.629934016450905e-05, "loss": 3.1873, "step": 10193 }, { "epoch": 0.6328139549320255, "grad_norm": 0.5322378734821593, "learning_rate": 9.629797649155434e-05, "loss": 3.065, "step": 10194 }, { "epoch": 0.6328760320317834, "grad_norm": 0.5340863971040015, "learning_rate": 9.629661257705187e-05, "loss": 3.1762, "step": 10195 }, { "epoch": 0.6329381091315414, "grad_norm": 0.5539301147485115, "learning_rate": 9.62952484210087e-05, "loss": 3.1329, "step": 10196 }, { "epoch": 0.6330001862312993, "grad_norm": 0.4923646093542268, "learning_rate": 9.629388402343201e-05, "loss": 3.1488, "step": 10197 }, { "epoch": 0.6330622633310572, "grad_norm": 0.5520993956142628, "learning_rate": 9.629251938432886e-05, "loss": 3.1558, "step": 10198 }, { "epoch": 0.6331243404308151, "grad_norm": 0.6161910354616528, "learning_rate": 9.629115450370642e-05, "loss": 3.2574, "step": 10199 }, { "epoch": 0.6331864175305729, "grad_norm": 0.5010309914808523, "learning_rate": 9.628978938157179e-05, "loss": 3.1652, "step": 10200 }, { "epoch": 0.6332484946303308, "grad_norm": 0.4914633858703862, "learning_rate": 9.628842401793208e-05, "loss": 3.1601, "step": 10201 }, { "epoch": 0.6333105717300888, "grad_norm": 0.5142200576362222, "learning_rate": 9.628705841279441e-05, "loss": 3.1791, "step": 10202 }, { "epoch": 0.6333726488298467, "grad_norm": 0.5040082902131893, "learning_rate": 9.628569256616595e-05, "loss": 3.1428, "step": 10203 }, { "epoch": 0.6334347259296046, "grad_norm": 0.5084428835948989, "learning_rate": 9.62843264780538e-05, "loss": 3.1513, "step": 10204 }, { "epoch": 0.6334968030293625, "grad_norm": 0.5398302002019235, "learning_rate": 9.628296014846505e-05, "loss": 3.1444, "step": 10205 }, { "epoch": 0.6335588801291203, "grad_norm": 0.5247005198841634, "learning_rate": 9.628159357740688e-05, "loss": 3.1472, "step": 10206 }, { "epoch": 0.6336209572288782, "grad_norm": 0.49949601224218193, "learning_rate": 9.628022676488639e-05, "loss": 3.0732, "step": 10207 }, { "epoch": 0.6336830343286362, "grad_norm": 0.43909767186484394, "learning_rate": 9.627885971091075e-05, "loss": 3.0629, "step": 10208 }, { "epoch": 0.6337451114283941, "grad_norm": 0.4837217386501995, "learning_rate": 9.627749241548705e-05, "loss": 3.1172, "step": 10209 }, { "epoch": 0.633807188528152, "grad_norm": 0.5498807322625449, "learning_rate": 9.627612487862243e-05, "loss": 3.1885, "step": 10210 }, { "epoch": 0.6338692656279099, "grad_norm": 0.5044273816293108, "learning_rate": 9.627475710032404e-05, "loss": 3.1253, "step": 10211 }, { "epoch": 0.6339313427276677, "grad_norm": 0.4656952984118132, "learning_rate": 9.6273389080599e-05, "loss": 3.1855, "step": 10212 }, { "epoch": 0.6339934198274256, "grad_norm": 0.4839191445149859, "learning_rate": 9.627202081945445e-05, "loss": 3.1092, "step": 10213 }, { "epoch": 0.6340554969271835, "grad_norm": 0.45398101847413336, "learning_rate": 9.627065231689756e-05, "loss": 3.1179, "step": 10214 }, { "epoch": 0.6341175740269415, "grad_norm": 0.459796495462679, "learning_rate": 9.626928357293542e-05, "loss": 3.1323, "step": 10215 }, { "epoch": 0.6341796511266994, "grad_norm": 0.44936675245863217, "learning_rate": 9.626791458757521e-05, "loss": 3.1127, "step": 10216 }, { "epoch": 0.6342417282264573, "grad_norm": 0.49939586977742184, "learning_rate": 9.626654536082405e-05, "loss": 3.1441, "step": 10217 }, { "epoch": 0.6343038053262151, "grad_norm": 0.5115698677888648, "learning_rate": 9.626517589268909e-05, "loss": 3.105, "step": 10218 }, { "epoch": 0.634365882425973, "grad_norm": 0.501264047211807, "learning_rate": 9.626380618317748e-05, "loss": 3.1028, "step": 10219 }, { "epoch": 0.634427959525731, "grad_norm": 0.4748709785761085, "learning_rate": 9.626243623229636e-05, "loss": 3.1368, "step": 10220 }, { "epoch": 0.6344900366254889, "grad_norm": 0.4607894142335676, "learning_rate": 9.626106604005288e-05, "loss": 3.1165, "step": 10221 }, { "epoch": 0.6345521137252468, "grad_norm": 0.4542592420923725, "learning_rate": 9.62596956064542e-05, "loss": 3.1301, "step": 10222 }, { "epoch": 0.6346141908250047, "grad_norm": 0.45718910881901476, "learning_rate": 9.625832493150744e-05, "loss": 3.1125, "step": 10223 }, { "epoch": 0.6346762679247625, "grad_norm": 0.46134260124213905, "learning_rate": 9.625695401521978e-05, "loss": 3.0997, "step": 10224 }, { "epoch": 0.6347383450245204, "grad_norm": 0.46820739494628694, "learning_rate": 9.625558285759836e-05, "loss": 3.1754, "step": 10225 }, { "epoch": 0.6348004221242783, "grad_norm": 0.4430962844874077, "learning_rate": 9.625421145865033e-05, "loss": 3.1003, "step": 10226 }, { "epoch": 0.6348624992240363, "grad_norm": 0.5104986890426676, "learning_rate": 9.625283981838286e-05, "loss": 3.1685, "step": 10227 }, { "epoch": 0.6349245763237942, "grad_norm": 0.5564629522160716, "learning_rate": 9.625146793680309e-05, "loss": 3.2298, "step": 10228 }, { "epoch": 0.6349866534235521, "grad_norm": 0.5875218397504569, "learning_rate": 9.625009581391818e-05, "loss": 3.1001, "step": 10229 }, { "epoch": 0.6350487305233099, "grad_norm": 0.4884190326843377, "learning_rate": 9.62487234497353e-05, "loss": 3.2587, "step": 10230 }, { "epoch": 0.6351108076230678, "grad_norm": 0.5083810118166644, "learning_rate": 9.624735084426159e-05, "loss": 3.1579, "step": 10231 }, { "epoch": 0.6351728847228257, "grad_norm": 0.5723525362124036, "learning_rate": 9.624597799750424e-05, "loss": 3.1266, "step": 10232 }, { "epoch": 0.6352349618225837, "grad_norm": 0.5125900596481625, "learning_rate": 9.62446049094704e-05, "loss": 3.1678, "step": 10233 }, { "epoch": 0.6352970389223416, "grad_norm": 0.46364522341560466, "learning_rate": 9.624323158016721e-05, "loss": 3.1471, "step": 10234 }, { "epoch": 0.6353591160220995, "grad_norm": 0.48446399117340766, "learning_rate": 9.624185800960186e-05, "loss": 3.1646, "step": 10235 }, { "epoch": 0.6354211931218573, "grad_norm": 0.4659234658474479, "learning_rate": 9.624048419778152e-05, "loss": 3.1787, "step": 10236 }, { "epoch": 0.6354832702216152, "grad_norm": 0.500262072545703, "learning_rate": 9.623911014471333e-05, "loss": 3.1519, "step": 10237 }, { "epoch": 0.6355453473213731, "grad_norm": 0.4161124497329006, "learning_rate": 9.62377358504045e-05, "loss": 3.119, "step": 10238 }, { "epoch": 0.635607424421131, "grad_norm": 0.4929019193493511, "learning_rate": 9.623636131486218e-05, "loss": 3.074, "step": 10239 }, { "epoch": 0.635669501520889, "grad_norm": 0.4947188541310316, "learning_rate": 9.623498653809353e-05, "loss": 3.1057, "step": 10240 }, { "epoch": 0.6357315786206469, "grad_norm": 0.4844573781503944, "learning_rate": 9.623361152010572e-05, "loss": 3.1351, "step": 10241 }, { "epoch": 0.6357936557204047, "grad_norm": 0.4786726557702069, "learning_rate": 9.623223626090594e-05, "loss": 3.0594, "step": 10242 }, { "epoch": 0.6358557328201626, "grad_norm": 0.5018212111687759, "learning_rate": 9.623086076050136e-05, "loss": 3.0971, "step": 10243 }, { "epoch": 0.6359178099199205, "grad_norm": 0.46613120654580814, "learning_rate": 9.622948501889916e-05, "loss": 3.1941, "step": 10244 }, { "epoch": 0.6359798870196784, "grad_norm": 0.6013001204158904, "learning_rate": 9.622810903610653e-05, "loss": 3.1317, "step": 10245 }, { "epoch": 0.6360419641194364, "grad_norm": 0.5419402776380136, "learning_rate": 9.622673281213062e-05, "loss": 3.1108, "step": 10246 }, { "epoch": 0.6361040412191943, "grad_norm": 0.5062096756310716, "learning_rate": 9.622535634697861e-05, "loss": 3.2753, "step": 10247 }, { "epoch": 0.6361661183189521, "grad_norm": 0.5386542004070847, "learning_rate": 9.62239796406577e-05, "loss": 3.0682, "step": 10248 }, { "epoch": 0.63622819541871, "grad_norm": 0.5589236105807983, "learning_rate": 9.622260269317509e-05, "loss": 3.0801, "step": 10249 }, { "epoch": 0.6362902725184679, "grad_norm": 0.5295342901777994, "learning_rate": 9.622122550453792e-05, "loss": 3.1932, "step": 10250 }, { "epoch": 0.6363523496182258, "grad_norm": 0.49539342393672486, "learning_rate": 9.621984807475339e-05, "loss": 3.0917, "step": 10251 }, { "epoch": 0.6364144267179838, "grad_norm": 0.4625392014411696, "learning_rate": 9.62184704038287e-05, "loss": 3.1521, "step": 10252 }, { "epoch": 0.6364765038177417, "grad_norm": 0.4926087529509466, "learning_rate": 9.621709249177104e-05, "loss": 3.0615, "step": 10253 }, { "epoch": 0.6365385809174995, "grad_norm": 0.46980562555871735, "learning_rate": 9.621571433858758e-05, "loss": 3.1716, "step": 10254 }, { "epoch": 0.6366006580172574, "grad_norm": 0.5017923773074464, "learning_rate": 9.621433594428552e-05, "loss": 3.1319, "step": 10255 }, { "epoch": 0.6366627351170153, "grad_norm": 0.5399069322635954, "learning_rate": 9.621295730887205e-05, "loss": 3.1421, "step": 10256 }, { "epoch": 0.6367248122167732, "grad_norm": 0.4386279008518852, "learning_rate": 9.621157843235436e-05, "loss": 3.053, "step": 10257 }, { "epoch": 0.6367868893165312, "grad_norm": 0.5530086147040889, "learning_rate": 9.621019931473964e-05, "loss": 3.1242, "step": 10258 }, { "epoch": 0.6368489664162891, "grad_norm": 0.6031453005679971, "learning_rate": 9.62088199560351e-05, "loss": 3.1541, "step": 10259 }, { "epoch": 0.6369110435160469, "grad_norm": 0.5341573986027057, "learning_rate": 9.620744035624794e-05, "loss": 3.0578, "step": 10260 }, { "epoch": 0.6369731206158048, "grad_norm": 0.4842594531893756, "learning_rate": 9.620606051538534e-05, "loss": 3.0833, "step": 10261 }, { "epoch": 0.6370351977155627, "grad_norm": 0.5524365326948142, "learning_rate": 9.62046804334545e-05, "loss": 3.1718, "step": 10262 }, { "epoch": 0.6370972748153206, "grad_norm": 0.6305994968909553, "learning_rate": 9.620330011046263e-05, "loss": 3.2147, "step": 10263 }, { "epoch": 0.6371593519150786, "grad_norm": 0.5363650904486597, "learning_rate": 9.620191954641693e-05, "loss": 3.1056, "step": 10264 }, { "epoch": 0.6372214290148365, "grad_norm": 0.5637034152929206, "learning_rate": 9.620053874132459e-05, "loss": 3.1226, "step": 10265 }, { "epoch": 0.6372835061145943, "grad_norm": 0.5044581916865388, "learning_rate": 9.619915769519283e-05, "loss": 3.1185, "step": 10266 }, { "epoch": 0.6373455832143522, "grad_norm": 0.5010228126399329, "learning_rate": 9.619777640802885e-05, "loss": 3.1911, "step": 10267 }, { "epoch": 0.6374076603141101, "grad_norm": 0.5074062483028039, "learning_rate": 9.619639487983984e-05, "loss": 3.084, "step": 10268 }, { "epoch": 0.637469737413868, "grad_norm": 0.5200977821755353, "learning_rate": 9.619501311063303e-05, "loss": 3.0888, "step": 10269 }, { "epoch": 0.637531814513626, "grad_norm": 0.5064274017897716, "learning_rate": 9.619363110041564e-05, "loss": 3.2312, "step": 10270 }, { "epoch": 0.6375938916133839, "grad_norm": 0.5025302226434035, "learning_rate": 9.619224884919485e-05, "loss": 3.1234, "step": 10271 }, { "epoch": 0.6376559687131417, "grad_norm": 0.5095010414281352, "learning_rate": 9.61908663569779e-05, "loss": 3.0253, "step": 10272 }, { "epoch": 0.6377180458128996, "grad_norm": 0.5209835662336701, "learning_rate": 9.618948362377195e-05, "loss": 3.1557, "step": 10273 }, { "epoch": 0.6377801229126575, "grad_norm": 0.47253194072787186, "learning_rate": 9.618810064958428e-05, "loss": 3.1654, "step": 10274 }, { "epoch": 0.6378422000124154, "grad_norm": 0.5139816278539912, "learning_rate": 9.618671743442208e-05, "loss": 3.1479, "step": 10275 }, { "epoch": 0.6379042771121733, "grad_norm": 0.4776051637423503, "learning_rate": 9.618533397829255e-05, "loss": 3.0599, "step": 10276 }, { "epoch": 0.6379663542119313, "grad_norm": 0.5619904276297344, "learning_rate": 9.618395028120293e-05, "loss": 3.212, "step": 10277 }, { "epoch": 0.6380284313116891, "grad_norm": 0.4848528166503146, "learning_rate": 9.618256634316042e-05, "loss": 3.1435, "step": 10278 }, { "epoch": 0.638090508411447, "grad_norm": 0.5445149040231625, "learning_rate": 9.618118216417226e-05, "loss": 3.2261, "step": 10279 }, { "epoch": 0.6381525855112049, "grad_norm": 0.5055080652055757, "learning_rate": 9.617979774424566e-05, "loss": 3.033, "step": 10280 }, { "epoch": 0.6382146626109628, "grad_norm": 0.5627505221387762, "learning_rate": 9.617841308338784e-05, "loss": 3.0376, "step": 10281 }, { "epoch": 0.6382767397107207, "grad_norm": 0.5617610262903597, "learning_rate": 9.617702818160603e-05, "loss": 3.2128, "step": 10282 }, { "epoch": 0.6383388168104787, "grad_norm": 0.5433422423238862, "learning_rate": 9.617564303890745e-05, "loss": 3.1293, "step": 10283 }, { "epoch": 0.6384008939102365, "grad_norm": 0.5383258983769851, "learning_rate": 9.617425765529932e-05, "loss": 3.2006, "step": 10284 }, { "epoch": 0.6384629710099944, "grad_norm": 0.5196804893511667, "learning_rate": 9.61728720307889e-05, "loss": 3.1954, "step": 10285 }, { "epoch": 0.6385250481097523, "grad_norm": 0.4766007705080349, "learning_rate": 9.617148616538338e-05, "loss": 3.1223, "step": 10286 }, { "epoch": 0.6385871252095102, "grad_norm": 0.47964593166700525, "learning_rate": 9.617010005909002e-05, "loss": 3.098, "step": 10287 }, { "epoch": 0.6386492023092681, "grad_norm": 0.5397157166686061, "learning_rate": 9.616871371191604e-05, "loss": 3.0929, "step": 10288 }, { "epoch": 0.6387112794090261, "grad_norm": 0.4555783612644182, "learning_rate": 9.616732712386867e-05, "loss": 2.9788, "step": 10289 }, { "epoch": 0.6387733565087839, "grad_norm": 0.5322941448455076, "learning_rate": 9.616594029495516e-05, "loss": 3.2504, "step": 10290 }, { "epoch": 0.6388354336085418, "grad_norm": 0.5187950654734879, "learning_rate": 9.616455322518272e-05, "loss": 3.1061, "step": 10291 }, { "epoch": 0.6388975107082997, "grad_norm": 0.4813131959714713, "learning_rate": 9.616316591455859e-05, "loss": 3.0775, "step": 10292 }, { "epoch": 0.6389595878080576, "grad_norm": 0.44209457000572244, "learning_rate": 9.616177836309004e-05, "loss": 3.0248, "step": 10293 }, { "epoch": 0.6390216649078155, "grad_norm": 0.5690482772231262, "learning_rate": 9.616039057078427e-05, "loss": 3.1383, "step": 10294 }, { "epoch": 0.6390837420075735, "grad_norm": 0.5620715045697596, "learning_rate": 9.615900253764853e-05, "loss": 3.2041, "step": 10295 }, { "epoch": 0.6391458191073313, "grad_norm": 0.4922296717908583, "learning_rate": 9.615761426369008e-05, "loss": 3.2154, "step": 10296 }, { "epoch": 0.6392078962070892, "grad_norm": 0.5259817832123903, "learning_rate": 9.615622574891615e-05, "loss": 3.2818, "step": 10297 }, { "epoch": 0.6392699733068471, "grad_norm": 0.5744940169332453, "learning_rate": 9.615483699333399e-05, "loss": 3.0852, "step": 10298 }, { "epoch": 0.639332050406605, "grad_norm": 0.4598900321199987, "learning_rate": 9.615344799695083e-05, "loss": 3.1772, "step": 10299 }, { "epoch": 0.6393941275063629, "grad_norm": 0.5317935967375852, "learning_rate": 9.615205875977394e-05, "loss": 3.1288, "step": 10300 }, { "epoch": 0.6394562046061208, "grad_norm": 0.4643538963379496, "learning_rate": 9.615066928181053e-05, "loss": 3.0391, "step": 10301 }, { "epoch": 0.6395182817058787, "grad_norm": 0.5399303361320527, "learning_rate": 9.614927956306789e-05, "loss": 3.157, "step": 10302 }, { "epoch": 0.6395803588056366, "grad_norm": 0.58178949135482, "learning_rate": 9.614788960355327e-05, "loss": 3.2384, "step": 10303 }, { "epoch": 0.6396424359053945, "grad_norm": 0.5852013859574677, "learning_rate": 9.61464994032739e-05, "loss": 3.19, "step": 10304 }, { "epoch": 0.6397045130051524, "grad_norm": 0.536985875583281, "learning_rate": 9.614510896223703e-05, "loss": 3.0787, "step": 10305 }, { "epoch": 0.6397665901049103, "grad_norm": 0.5970719416992002, "learning_rate": 9.614371828044993e-05, "loss": 3.1591, "step": 10306 }, { "epoch": 0.6398286672046682, "grad_norm": 0.535351571406991, "learning_rate": 9.614232735791985e-05, "loss": 3.2022, "step": 10307 }, { "epoch": 0.639890744304426, "grad_norm": 0.5811441560994426, "learning_rate": 9.614093619465405e-05, "loss": 3.1541, "step": 10308 }, { "epoch": 0.639952821404184, "grad_norm": 0.5483014712405603, "learning_rate": 9.613954479065979e-05, "loss": 3.1377, "step": 10309 }, { "epoch": 0.6400148985039419, "grad_norm": 0.4754250977053053, "learning_rate": 9.61381531459443e-05, "loss": 3.137, "step": 10310 }, { "epoch": 0.6400769756036998, "grad_norm": 0.5317056889349288, "learning_rate": 9.613676126051489e-05, "loss": 3.1005, "step": 10311 }, { "epoch": 0.6401390527034577, "grad_norm": 0.4716842998650329, "learning_rate": 9.613536913437879e-05, "loss": 3.0449, "step": 10312 }, { "epoch": 0.6402011298032156, "grad_norm": 0.5293217895950239, "learning_rate": 9.613397676754325e-05, "loss": 3.1789, "step": 10313 }, { "epoch": 0.6402632069029734, "grad_norm": 0.4955242221928366, "learning_rate": 9.613258416001559e-05, "loss": 3.1407, "step": 10314 }, { "epoch": 0.6403252840027314, "grad_norm": 0.5111449424228506, "learning_rate": 9.613119131180302e-05, "loss": 3.1297, "step": 10315 }, { "epoch": 0.6403873611024893, "grad_norm": 0.5052666155307034, "learning_rate": 9.612979822291282e-05, "loss": 3.1886, "step": 10316 }, { "epoch": 0.6404494382022472, "grad_norm": 0.483987129948667, "learning_rate": 9.612840489335227e-05, "loss": 3.185, "step": 10317 }, { "epoch": 0.6405115153020051, "grad_norm": 0.49254560310280143, "learning_rate": 9.612701132312863e-05, "loss": 3.0305, "step": 10318 }, { "epoch": 0.640573592401763, "grad_norm": 0.5085967829975688, "learning_rate": 9.612561751224917e-05, "loss": 3.1154, "step": 10319 }, { "epoch": 0.6406356695015208, "grad_norm": 0.5413324137839993, "learning_rate": 9.612422346072119e-05, "loss": 3.0938, "step": 10320 }, { "epoch": 0.6406977466012788, "grad_norm": 0.5212701650419338, "learning_rate": 9.61228291685519e-05, "loss": 3.1338, "step": 10321 }, { "epoch": 0.6407598237010367, "grad_norm": 0.59477496772302, "learning_rate": 9.612143463574865e-05, "loss": 3.1887, "step": 10322 }, { "epoch": 0.6408219008007946, "grad_norm": 0.5431770089975138, "learning_rate": 9.612003986231865e-05, "loss": 3.1463, "step": 10323 }, { "epoch": 0.6408839779005525, "grad_norm": 0.5131463383037451, "learning_rate": 9.611864484826924e-05, "loss": 3.2209, "step": 10324 }, { "epoch": 0.6409460550003104, "grad_norm": 0.5507202114615514, "learning_rate": 9.611724959360764e-05, "loss": 3.1898, "step": 10325 }, { "epoch": 0.6410081321000682, "grad_norm": 0.5453355920924778, "learning_rate": 9.611585409834116e-05, "loss": 3.1984, "step": 10326 }, { "epoch": 0.6410702091998262, "grad_norm": 0.566267483253974, "learning_rate": 9.611445836247708e-05, "loss": 3.2083, "step": 10327 }, { "epoch": 0.6411322862995841, "grad_norm": 0.48972296921664715, "learning_rate": 9.611306238602266e-05, "loss": 3.2219, "step": 10328 }, { "epoch": 0.641194363399342, "grad_norm": 0.5404580889463981, "learning_rate": 9.611166616898521e-05, "loss": 3.0666, "step": 10329 }, { "epoch": 0.6412564404990999, "grad_norm": 0.5199953458651092, "learning_rate": 9.6110269711372e-05, "loss": 3.1431, "step": 10330 }, { "epoch": 0.6413185175988578, "grad_norm": 0.5305533329044452, "learning_rate": 9.610887301319033e-05, "loss": 3.0621, "step": 10331 }, { "epoch": 0.6413805946986156, "grad_norm": 0.49677608153128666, "learning_rate": 9.610747607444745e-05, "loss": 3.0815, "step": 10332 }, { "epoch": 0.6414426717983736, "grad_norm": 0.4730224924395151, "learning_rate": 9.610607889515071e-05, "loss": 3.1039, "step": 10333 }, { "epoch": 0.6415047488981315, "grad_norm": 0.5042365714021768, "learning_rate": 9.610468147530734e-05, "loss": 3.0812, "step": 10334 }, { "epoch": 0.6415668259978894, "grad_norm": 0.500772416011597, "learning_rate": 9.610328381492466e-05, "loss": 3.1455, "step": 10335 }, { "epoch": 0.6416289030976473, "grad_norm": 0.5217128924474838, "learning_rate": 9.610188591400995e-05, "loss": 3.1367, "step": 10336 }, { "epoch": 0.6416909801974052, "grad_norm": 0.5563780047974687, "learning_rate": 9.610048777257052e-05, "loss": 3.1656, "step": 10337 }, { "epoch": 0.641753057297163, "grad_norm": 0.5168935603138611, "learning_rate": 9.609908939061365e-05, "loss": 3.2319, "step": 10338 }, { "epoch": 0.641815134396921, "grad_norm": 0.5646768146510899, "learning_rate": 9.609769076814665e-05, "loss": 3.1786, "step": 10339 }, { "epoch": 0.6418772114966789, "grad_norm": 0.5119542460878282, "learning_rate": 9.60962919051768e-05, "loss": 3.1798, "step": 10340 }, { "epoch": 0.6419392885964368, "grad_norm": 0.4798013502323397, "learning_rate": 9.609489280171139e-05, "loss": 3.164, "step": 10341 }, { "epoch": 0.6420013656961947, "grad_norm": 0.5160357833032057, "learning_rate": 9.609349345775774e-05, "loss": 3.1215, "step": 10342 }, { "epoch": 0.6420634427959526, "grad_norm": 0.5744313475788096, "learning_rate": 9.609209387332316e-05, "loss": 3.0213, "step": 10343 }, { "epoch": 0.6421255198957104, "grad_norm": 0.54064989639023, "learning_rate": 9.609069404841493e-05, "loss": 3.212, "step": 10344 }, { "epoch": 0.6421875969954683, "grad_norm": 0.5525336288557143, "learning_rate": 9.608929398304035e-05, "loss": 3.1631, "step": 10345 }, { "epoch": 0.6422496740952263, "grad_norm": 0.5328432202921854, "learning_rate": 9.608789367720674e-05, "loss": 3.1633, "step": 10346 }, { "epoch": 0.6423117511949842, "grad_norm": 0.4953209722859318, "learning_rate": 9.608649313092141e-05, "loss": 3.1553, "step": 10347 }, { "epoch": 0.6423738282947421, "grad_norm": 0.5075070269337768, "learning_rate": 9.608509234419165e-05, "loss": 3.1486, "step": 10348 }, { "epoch": 0.6424359053945, "grad_norm": 0.5237362990943042, "learning_rate": 9.608369131702477e-05, "loss": 3.0896, "step": 10349 }, { "epoch": 0.6424979824942578, "grad_norm": 0.5129112733401928, "learning_rate": 9.60822900494281e-05, "loss": 3.003, "step": 10350 }, { "epoch": 0.6425600595940157, "grad_norm": 0.5187877411499154, "learning_rate": 9.608088854140894e-05, "loss": 3.1413, "step": 10351 }, { "epoch": 0.6426221366937737, "grad_norm": 0.5245794110180557, "learning_rate": 9.607948679297457e-05, "loss": 3.1329, "step": 10352 }, { "epoch": 0.6426842137935316, "grad_norm": 0.5351195218367885, "learning_rate": 9.607808480413234e-05, "loss": 3.1132, "step": 10353 }, { "epoch": 0.6427462908932895, "grad_norm": 0.4488292996197794, "learning_rate": 9.607668257488957e-05, "loss": 3.1288, "step": 10354 }, { "epoch": 0.6428083679930474, "grad_norm": 0.4875284113307714, "learning_rate": 9.607528010525356e-05, "loss": 3.0318, "step": 10355 }, { "epoch": 0.6428704450928052, "grad_norm": 0.5017174490332925, "learning_rate": 9.607387739523161e-05, "loss": 3.1723, "step": 10356 }, { "epoch": 0.6429325221925631, "grad_norm": 0.49685674849763417, "learning_rate": 9.607247444483108e-05, "loss": 3.1339, "step": 10357 }, { "epoch": 0.6429945992923211, "grad_norm": 0.47269646024300627, "learning_rate": 9.607107125405925e-05, "loss": 3.1348, "step": 10358 }, { "epoch": 0.643056676392079, "grad_norm": 0.5235193957130523, "learning_rate": 9.606966782292346e-05, "loss": 3.1389, "step": 10359 }, { "epoch": 0.6431187534918369, "grad_norm": 0.5207011595543686, "learning_rate": 9.606826415143104e-05, "loss": 3.0925, "step": 10360 }, { "epoch": 0.6431808305915948, "grad_norm": 0.482401426553752, "learning_rate": 9.606686023958929e-05, "loss": 3.0148, "step": 10361 }, { "epoch": 0.6432429076913526, "grad_norm": 0.4714694294935038, "learning_rate": 9.606545608740555e-05, "loss": 3.037, "step": 10362 }, { "epoch": 0.6433049847911105, "grad_norm": 0.5162896005807067, "learning_rate": 9.606405169488715e-05, "loss": 3.129, "step": 10363 }, { "epoch": 0.6433670618908685, "grad_norm": 0.5104481596540722, "learning_rate": 9.60626470620414e-05, "loss": 3.1475, "step": 10364 }, { "epoch": 0.6434291389906264, "grad_norm": 0.5144059770047016, "learning_rate": 9.606124218887564e-05, "loss": 3.119, "step": 10365 }, { "epoch": 0.6434912160903843, "grad_norm": 0.5015255186523688, "learning_rate": 9.60598370753972e-05, "loss": 3.0321, "step": 10366 }, { "epoch": 0.6435532931901422, "grad_norm": 0.5535990014878265, "learning_rate": 9.605843172161341e-05, "loss": 3.0939, "step": 10367 }, { "epoch": 0.6436153702899, "grad_norm": 0.5174744124237208, "learning_rate": 9.605702612753161e-05, "loss": 3.1897, "step": 10368 }, { "epoch": 0.6436774473896579, "grad_norm": 0.4797333151841615, "learning_rate": 9.60556202931591e-05, "loss": 3.0406, "step": 10369 }, { "epoch": 0.6437395244894158, "grad_norm": 0.48886773554808943, "learning_rate": 9.605421421850325e-05, "loss": 3.1479, "step": 10370 }, { "epoch": 0.6438016015891738, "grad_norm": 0.5516368091874293, "learning_rate": 9.605280790357139e-05, "loss": 3.1898, "step": 10371 }, { "epoch": 0.6438636786889317, "grad_norm": 0.5020817395207393, "learning_rate": 9.605140134837086e-05, "loss": 3.1049, "step": 10372 }, { "epoch": 0.6439257557886896, "grad_norm": 0.47552878410430505, "learning_rate": 9.604999455290896e-05, "loss": 3.0541, "step": 10373 }, { "epoch": 0.6439878328884474, "grad_norm": 0.4938408466252075, "learning_rate": 9.604858751719308e-05, "loss": 3.1572, "step": 10374 }, { "epoch": 0.6440499099882053, "grad_norm": 0.5132943278000953, "learning_rate": 9.604718024123053e-05, "loss": 3.0949, "step": 10375 }, { "epoch": 0.6441119870879632, "grad_norm": 0.5261212059041892, "learning_rate": 9.604577272502868e-05, "loss": 3.2231, "step": 10376 }, { "epoch": 0.6441740641877212, "grad_norm": 0.5253295862980804, "learning_rate": 9.604436496859483e-05, "loss": 3.1684, "step": 10377 }, { "epoch": 0.6442361412874791, "grad_norm": 0.5132211368377303, "learning_rate": 9.604295697193637e-05, "loss": 3.156, "step": 10378 }, { "epoch": 0.644298218387237, "grad_norm": 0.49014603628182285, "learning_rate": 9.604154873506063e-05, "loss": 3.1284, "step": 10379 }, { "epoch": 0.6443602954869948, "grad_norm": 0.553602689088922, "learning_rate": 9.604014025797495e-05, "loss": 3.1484, "step": 10380 }, { "epoch": 0.6444223725867527, "grad_norm": 0.48028564065304713, "learning_rate": 9.603873154068666e-05, "loss": 3.103, "step": 10381 }, { "epoch": 0.6444844496865106, "grad_norm": 0.5022424539385567, "learning_rate": 9.603732258320315e-05, "loss": 3.138, "step": 10382 }, { "epoch": 0.6445465267862686, "grad_norm": 0.543189106967047, "learning_rate": 9.603591338553174e-05, "loss": 3.1966, "step": 10383 }, { "epoch": 0.6446086038860265, "grad_norm": 0.5410584801181624, "learning_rate": 9.60345039476798e-05, "loss": 3.1438, "step": 10384 }, { "epoch": 0.6446706809857844, "grad_norm": 0.5560569781895021, "learning_rate": 9.60330942696547e-05, "loss": 3.0771, "step": 10385 }, { "epoch": 0.6447327580855422, "grad_norm": 0.6109810962390503, "learning_rate": 9.603168435146375e-05, "loss": 3.1589, "step": 10386 }, { "epoch": 0.6447948351853001, "grad_norm": 0.4990239749913018, "learning_rate": 9.603027419311433e-05, "loss": 3.1294, "step": 10387 }, { "epoch": 0.644856912285058, "grad_norm": 0.5425687455423265, "learning_rate": 9.60288637946138e-05, "loss": 3.102, "step": 10388 }, { "epoch": 0.644918989384816, "grad_norm": 0.5644302363523375, "learning_rate": 9.60274531559695e-05, "loss": 3.1701, "step": 10389 }, { "epoch": 0.6449810664845739, "grad_norm": 0.5517416622532663, "learning_rate": 9.602604227718884e-05, "loss": 3.0763, "step": 10390 }, { "epoch": 0.6450431435843318, "grad_norm": 0.4270938972779362, "learning_rate": 9.602463115827911e-05, "loss": 3.0834, "step": 10391 }, { "epoch": 0.6451052206840896, "grad_norm": 0.4809483561241736, "learning_rate": 9.602321979924772e-05, "loss": 3.2124, "step": 10392 }, { "epoch": 0.6451672977838475, "grad_norm": 0.5002756848868964, "learning_rate": 9.6021808200102e-05, "loss": 3.1485, "step": 10393 }, { "epoch": 0.6452293748836054, "grad_norm": 0.4745325096899954, "learning_rate": 9.602039636084935e-05, "loss": 3.0815, "step": 10394 }, { "epoch": 0.6452914519833634, "grad_norm": 0.4595635837981571, "learning_rate": 9.601898428149712e-05, "loss": 3.1502, "step": 10395 }, { "epoch": 0.6453535290831213, "grad_norm": 0.5260104525782185, "learning_rate": 9.601757196205267e-05, "loss": 3.0813, "step": 10396 }, { "epoch": 0.6454156061828792, "grad_norm": 0.556507295401327, "learning_rate": 9.60161594025234e-05, "loss": 2.9821, "step": 10397 }, { "epoch": 0.645477683282637, "grad_norm": 0.4470938140171587, "learning_rate": 9.601474660291662e-05, "loss": 3.1883, "step": 10398 }, { "epoch": 0.6455397603823949, "grad_norm": 0.5046186512379275, "learning_rate": 9.601333356323976e-05, "loss": 3.1782, "step": 10399 }, { "epoch": 0.6456018374821528, "grad_norm": 0.5145009243829873, "learning_rate": 9.601192028350016e-05, "loss": 3.1065, "step": 10400 }, { "epoch": 0.6456639145819107, "grad_norm": 0.44446390845308975, "learning_rate": 9.601050676370522e-05, "loss": 3.0764, "step": 10401 }, { "epoch": 0.6457259916816687, "grad_norm": 0.5483613116171314, "learning_rate": 9.600909300386227e-05, "loss": 3.0893, "step": 10402 }, { "epoch": 0.6457880687814266, "grad_norm": 0.5150600575177512, "learning_rate": 9.600767900397874e-05, "loss": 3.0507, "step": 10403 }, { "epoch": 0.6458501458811844, "grad_norm": 0.48541474266926143, "learning_rate": 9.600626476406195e-05, "loss": 3.1403, "step": 10404 }, { "epoch": 0.6459122229809423, "grad_norm": 0.5491535674439747, "learning_rate": 9.600485028411933e-05, "loss": 3.1685, "step": 10405 }, { "epoch": 0.6459743000807002, "grad_norm": 0.6130215040109791, "learning_rate": 9.600343556415822e-05, "loss": 3.1719, "step": 10406 }, { "epoch": 0.6460363771804581, "grad_norm": 0.5711796843326602, "learning_rate": 9.600202060418603e-05, "loss": 3.1161, "step": 10407 }, { "epoch": 0.6460984542802161, "grad_norm": 0.5469781233424582, "learning_rate": 9.600060540421014e-05, "loss": 3.0743, "step": 10408 }, { "epoch": 0.646160531379974, "grad_norm": 0.549467155964072, "learning_rate": 9.599918996423792e-05, "loss": 3.1782, "step": 10409 }, { "epoch": 0.6462226084797318, "grad_norm": 0.5280763976175546, "learning_rate": 9.599777428427676e-05, "loss": 3.0668, "step": 10410 }, { "epoch": 0.6462846855794897, "grad_norm": 0.5568966193235769, "learning_rate": 9.599635836433404e-05, "loss": 3.1225, "step": 10411 }, { "epoch": 0.6463467626792476, "grad_norm": 0.48306099259157004, "learning_rate": 9.599494220441715e-05, "loss": 3.22, "step": 10412 }, { "epoch": 0.6464088397790055, "grad_norm": 0.48453277663919464, "learning_rate": 9.599352580453349e-05, "loss": 3.1527, "step": 10413 }, { "epoch": 0.6464709168787635, "grad_norm": 0.5827767890441568, "learning_rate": 9.599210916469043e-05, "loss": 3.166, "step": 10414 }, { "epoch": 0.6465329939785214, "grad_norm": 0.4899218045336369, "learning_rate": 9.599069228489536e-05, "loss": 3.0425, "step": 10415 }, { "epoch": 0.6465950710782792, "grad_norm": 0.45611408508488693, "learning_rate": 9.598927516515571e-05, "loss": 3.1331, "step": 10416 }, { "epoch": 0.6466571481780371, "grad_norm": 0.5477145426550445, "learning_rate": 9.598785780547883e-05, "loss": 3.1071, "step": 10417 }, { "epoch": 0.646719225277795, "grad_norm": 0.4906597799377405, "learning_rate": 9.598644020587215e-05, "loss": 3.163, "step": 10418 }, { "epoch": 0.6467813023775529, "grad_norm": 0.488781817393269, "learning_rate": 9.598502236634304e-05, "loss": 3.2077, "step": 10419 }, { "epoch": 0.6468433794773109, "grad_norm": 0.5340885454214096, "learning_rate": 9.59836042868989e-05, "loss": 3.1597, "step": 10420 }, { "epoch": 0.6469054565770688, "grad_norm": 0.48423266626863837, "learning_rate": 9.598218596754714e-05, "loss": 3.1559, "step": 10421 }, { "epoch": 0.6469675336768266, "grad_norm": 0.496009113571248, "learning_rate": 9.598076740829514e-05, "loss": 3.2283, "step": 10422 }, { "epoch": 0.6470296107765845, "grad_norm": 0.5192628037862724, "learning_rate": 9.597934860915032e-05, "loss": 3.165, "step": 10423 }, { "epoch": 0.6470916878763424, "grad_norm": 0.48762845462200205, "learning_rate": 9.597792957012009e-05, "loss": 3.0949, "step": 10424 }, { "epoch": 0.6471537649761003, "grad_norm": 0.4661147908222874, "learning_rate": 9.597651029121182e-05, "loss": 3.1454, "step": 10425 }, { "epoch": 0.6472158420758582, "grad_norm": 0.508790028986681, "learning_rate": 9.597509077243293e-05, "loss": 3.1692, "step": 10426 }, { "epoch": 0.6472779191756162, "grad_norm": 0.47125673780000804, "learning_rate": 9.597367101379086e-05, "loss": 3.1509, "step": 10427 }, { "epoch": 0.647339996275374, "grad_norm": 0.5437486659504979, "learning_rate": 9.597225101529296e-05, "loss": 3.234, "step": 10428 }, { "epoch": 0.6474020733751319, "grad_norm": 0.5522600976600184, "learning_rate": 9.597083077694668e-05, "loss": 3.1148, "step": 10429 }, { "epoch": 0.6474641504748898, "grad_norm": 0.5586431521464525, "learning_rate": 9.596941029875942e-05, "loss": 3.0934, "step": 10430 }, { "epoch": 0.6475262275746477, "grad_norm": 0.57225966206331, "learning_rate": 9.596798958073859e-05, "loss": 3.0192, "step": 10431 }, { "epoch": 0.6475883046744056, "grad_norm": 0.4899590619658137, "learning_rate": 9.596656862289158e-05, "loss": 3.0971, "step": 10432 }, { "epoch": 0.6476503817741636, "grad_norm": 0.5202058413589344, "learning_rate": 9.596514742522584e-05, "loss": 3.135, "step": 10433 }, { "epoch": 0.6477124588739214, "grad_norm": 0.5977511361788199, "learning_rate": 9.596372598774875e-05, "loss": 3.1205, "step": 10434 }, { "epoch": 0.6477745359736793, "grad_norm": 0.5032395084460509, "learning_rate": 9.596230431046775e-05, "loss": 3.1564, "step": 10435 }, { "epoch": 0.6478366130734372, "grad_norm": 0.5659665472790165, "learning_rate": 9.596088239339027e-05, "loss": 3.1825, "step": 10436 }, { "epoch": 0.6478986901731951, "grad_norm": 0.5355792079656061, "learning_rate": 9.595946023652369e-05, "loss": 3.1431, "step": 10437 }, { "epoch": 0.647960767272953, "grad_norm": 0.4803452837098521, "learning_rate": 9.595803783987544e-05, "loss": 3.1122, "step": 10438 }, { "epoch": 0.648022844372711, "grad_norm": 0.5590633739040276, "learning_rate": 9.595661520345299e-05, "loss": 2.9743, "step": 10439 }, { "epoch": 0.6480849214724688, "grad_norm": 0.4866750774108555, "learning_rate": 9.595519232726369e-05, "loss": 3.091, "step": 10440 }, { "epoch": 0.6481469985722267, "grad_norm": 0.5621308068808315, "learning_rate": 9.595376921131499e-05, "loss": 3.0826, "step": 10441 }, { "epoch": 0.6482090756719846, "grad_norm": 0.5716862789252363, "learning_rate": 9.595234585561432e-05, "loss": 3.1387, "step": 10442 }, { "epoch": 0.6482711527717425, "grad_norm": 0.5292087777220013, "learning_rate": 9.595092226016913e-05, "loss": 3.0963, "step": 10443 }, { "epoch": 0.6483332298715004, "grad_norm": 0.5221591541351754, "learning_rate": 9.594949842498682e-05, "loss": 3.1654, "step": 10444 }, { "epoch": 0.6483953069712584, "grad_norm": 0.5706058688262733, "learning_rate": 9.59480743500748e-05, "loss": 3.11, "step": 10445 }, { "epoch": 0.6484573840710162, "grad_norm": 0.5220302910383472, "learning_rate": 9.594665003544053e-05, "loss": 3.1784, "step": 10446 }, { "epoch": 0.6485194611707741, "grad_norm": 0.5437194245610124, "learning_rate": 9.594522548109144e-05, "loss": 3.0804, "step": 10447 }, { "epoch": 0.648581538270532, "grad_norm": 0.46727463282870274, "learning_rate": 9.594380068703496e-05, "loss": 3.1717, "step": 10448 }, { "epoch": 0.6486436153702899, "grad_norm": 0.5189218029803117, "learning_rate": 9.59423756532785e-05, "loss": 3.1622, "step": 10449 }, { "epoch": 0.6487056924700478, "grad_norm": 0.5417243364396499, "learning_rate": 9.594095037982953e-05, "loss": 3.1518, "step": 10450 }, { "epoch": 0.6487677695698058, "grad_norm": 0.5194754606422891, "learning_rate": 9.593952486669547e-05, "loss": 3.1008, "step": 10451 }, { "epoch": 0.6488298466695636, "grad_norm": 0.5707542328662951, "learning_rate": 9.593809911388374e-05, "loss": 3.1876, "step": 10452 }, { "epoch": 0.6488919237693215, "grad_norm": 0.48044786175238036, "learning_rate": 9.593667312140181e-05, "loss": 3.1148, "step": 10453 }, { "epoch": 0.6489540008690794, "grad_norm": 0.5051276021209279, "learning_rate": 9.593524688925711e-05, "loss": 3.1314, "step": 10454 }, { "epoch": 0.6490160779688373, "grad_norm": 0.5633445333838405, "learning_rate": 9.593382041745705e-05, "loss": 3.0921, "step": 10455 }, { "epoch": 0.6490781550685952, "grad_norm": 0.5194988384031907, "learning_rate": 9.593239370600912e-05, "loss": 3.0184, "step": 10456 }, { "epoch": 0.6491402321683531, "grad_norm": 0.5135698978744986, "learning_rate": 9.593096675492073e-05, "loss": 3.1394, "step": 10457 }, { "epoch": 0.649202309268111, "grad_norm": 0.5636398898518497, "learning_rate": 9.592953956419935e-05, "loss": 3.1606, "step": 10458 }, { "epoch": 0.6492643863678689, "grad_norm": 0.5360282698745096, "learning_rate": 9.59281121338524e-05, "loss": 3.1787, "step": 10459 }, { "epoch": 0.6493264634676268, "grad_norm": 0.5720105693197443, "learning_rate": 9.592668446388735e-05, "loss": 3.1018, "step": 10460 }, { "epoch": 0.6493885405673847, "grad_norm": 0.4590407965861536, "learning_rate": 9.592525655431164e-05, "loss": 3.1431, "step": 10461 }, { "epoch": 0.6494506176671426, "grad_norm": 0.5484772235797641, "learning_rate": 9.59238284051327e-05, "loss": 3.2086, "step": 10462 }, { "epoch": 0.6495126947669005, "grad_norm": 0.4952574612095414, "learning_rate": 9.5922400016358e-05, "loss": 3.0617, "step": 10463 }, { "epoch": 0.6495747718666584, "grad_norm": 0.5044335851478406, "learning_rate": 9.5920971387995e-05, "loss": 3.1438, "step": 10464 }, { "epoch": 0.6496368489664163, "grad_norm": 0.5377690928007889, "learning_rate": 9.591954252005116e-05, "loss": 2.9692, "step": 10465 }, { "epoch": 0.6496989260661742, "grad_norm": 0.4326782011316493, "learning_rate": 9.591811341253389e-05, "loss": 3.0353, "step": 10466 }, { "epoch": 0.6497610031659321, "grad_norm": 0.48152837752463895, "learning_rate": 9.59166840654507e-05, "loss": 3.1, "step": 10467 }, { "epoch": 0.64982308026569, "grad_norm": 0.43316692961271047, "learning_rate": 9.5915254478809e-05, "loss": 3.1235, "step": 10468 }, { "epoch": 0.6498851573654479, "grad_norm": 0.4029560060175857, "learning_rate": 9.59138246526163e-05, "loss": 3.213, "step": 10469 }, { "epoch": 0.6499472344652057, "grad_norm": 0.5123939427682589, "learning_rate": 9.591239458687999e-05, "loss": 3.1655, "step": 10470 }, { "epoch": 0.6500093115649637, "grad_norm": 0.5030323997721874, "learning_rate": 9.59109642816076e-05, "loss": 3.0665, "step": 10471 }, { "epoch": 0.6500713886647216, "grad_norm": 0.5298571559657842, "learning_rate": 9.590953373680656e-05, "loss": 3.1509, "step": 10472 }, { "epoch": 0.6501334657644795, "grad_norm": 0.4959819247149782, "learning_rate": 9.590810295248434e-05, "loss": 3.1244, "step": 10473 }, { "epoch": 0.6501955428642374, "grad_norm": 0.4823736251609804, "learning_rate": 9.59066719286484e-05, "loss": 3.0912, "step": 10474 }, { "epoch": 0.6502576199639953, "grad_norm": 0.5110731638882448, "learning_rate": 9.59052406653062e-05, "loss": 3.1226, "step": 10475 }, { "epoch": 0.6503196970637531, "grad_norm": 0.5451947421181419, "learning_rate": 9.590380916246523e-05, "loss": 3.1438, "step": 10476 }, { "epoch": 0.6503817741635111, "grad_norm": 0.435613327773042, "learning_rate": 9.590237742013294e-05, "loss": 3.1466, "step": 10477 }, { "epoch": 0.650443851263269, "grad_norm": 0.49733126222858476, "learning_rate": 9.59009454383168e-05, "loss": 3.037, "step": 10478 }, { "epoch": 0.6505059283630269, "grad_norm": 0.5068999674488689, "learning_rate": 9.589951321702428e-05, "loss": 3.182, "step": 10479 }, { "epoch": 0.6505680054627848, "grad_norm": 0.5420955525027049, "learning_rate": 9.589808075626286e-05, "loss": 3.2071, "step": 10480 }, { "epoch": 0.6506300825625427, "grad_norm": 0.45985317385184743, "learning_rate": 9.589664805604002e-05, "loss": 3.0455, "step": 10481 }, { "epoch": 0.6506921596623005, "grad_norm": 0.5015828588994403, "learning_rate": 9.589521511636322e-05, "loss": 3.1917, "step": 10482 }, { "epoch": 0.6507542367620585, "grad_norm": 0.47427415979689086, "learning_rate": 9.589378193723993e-05, "loss": 3.0955, "step": 10483 }, { "epoch": 0.6508163138618164, "grad_norm": 0.4514056669279659, "learning_rate": 9.589234851867766e-05, "loss": 3.108, "step": 10484 }, { "epoch": 0.6508783909615743, "grad_norm": 0.558484793753985, "learning_rate": 9.589091486068385e-05, "loss": 3.1845, "step": 10485 }, { "epoch": 0.6509404680613322, "grad_norm": 0.4947481755333367, "learning_rate": 9.5889480963266e-05, "loss": 3.08, "step": 10486 }, { "epoch": 0.6510025451610901, "grad_norm": 0.47809794355643154, "learning_rate": 9.588804682643158e-05, "loss": 3.1612, "step": 10487 }, { "epoch": 0.6510646222608479, "grad_norm": 0.5071746047490719, "learning_rate": 9.588661245018811e-05, "loss": 3.1857, "step": 10488 }, { "epoch": 0.6511266993606059, "grad_norm": 0.4607582390266425, "learning_rate": 9.588517783454301e-05, "loss": 3.0702, "step": 10489 }, { "epoch": 0.6511887764603638, "grad_norm": 0.4794364632665042, "learning_rate": 9.588374297950382e-05, "loss": 3.0963, "step": 10490 }, { "epoch": 0.6512508535601217, "grad_norm": 0.5217520494736109, "learning_rate": 9.588230788507799e-05, "loss": 3.1299, "step": 10491 }, { "epoch": 0.6513129306598796, "grad_norm": 0.490668890745346, "learning_rate": 9.588087255127303e-05, "loss": 3.0816, "step": 10492 }, { "epoch": 0.6513750077596374, "grad_norm": 0.43709591528629677, "learning_rate": 9.587943697809642e-05, "loss": 3.174, "step": 10493 }, { "epoch": 0.6514370848593953, "grad_norm": 0.4700000307915048, "learning_rate": 9.587800116555564e-05, "loss": 3.117, "step": 10494 }, { "epoch": 0.6514991619591533, "grad_norm": 0.5431403019442862, "learning_rate": 9.58765651136582e-05, "loss": 3.1363, "step": 10495 }, { "epoch": 0.6515612390589112, "grad_norm": 0.511114593752569, "learning_rate": 9.587512882241159e-05, "loss": 3.1316, "step": 10496 }, { "epoch": 0.6516233161586691, "grad_norm": 0.5442997926250835, "learning_rate": 9.587369229182328e-05, "loss": 3.1921, "step": 10497 }, { "epoch": 0.651685393258427, "grad_norm": 0.5216453675206896, "learning_rate": 9.587225552190078e-05, "loss": 3.0885, "step": 10498 }, { "epoch": 0.6517474703581848, "grad_norm": 0.4768939350963565, "learning_rate": 9.587081851265161e-05, "loss": 3.288, "step": 10499 }, { "epoch": 0.6518095474579427, "grad_norm": 0.4848994983926571, "learning_rate": 9.586938126408322e-05, "loss": 3.0487, "step": 10500 }, { "epoch": 0.6518716245577006, "grad_norm": 0.4881763803399734, "learning_rate": 9.586794377620314e-05, "loss": 3.0569, "step": 10501 }, { "epoch": 0.6519337016574586, "grad_norm": 0.4015928130824243, "learning_rate": 9.586650604901887e-05, "loss": 3.067, "step": 10502 }, { "epoch": 0.6519957787572165, "grad_norm": 0.4172042557569715, "learning_rate": 9.58650680825379e-05, "loss": 3.1997, "step": 10503 }, { "epoch": 0.6520578558569744, "grad_norm": 0.44889435042992065, "learning_rate": 9.586362987676773e-05, "loss": 3.1116, "step": 10504 }, { "epoch": 0.6521199329567322, "grad_norm": 0.417802278397923, "learning_rate": 9.586219143171587e-05, "loss": 3.0803, "step": 10505 }, { "epoch": 0.6521820100564901, "grad_norm": 0.4600570481107785, "learning_rate": 9.586075274738983e-05, "loss": 3.1502, "step": 10506 }, { "epoch": 0.652244087156248, "grad_norm": 0.5628873233314483, "learning_rate": 9.58593138237971e-05, "loss": 3.1771, "step": 10507 }, { "epoch": 0.652306164256006, "grad_norm": 0.5635631104747643, "learning_rate": 9.585787466094521e-05, "loss": 3.1667, "step": 10508 }, { "epoch": 0.6523682413557639, "grad_norm": 0.5133188869664662, "learning_rate": 9.585643525884165e-05, "loss": 3.1661, "step": 10509 }, { "epoch": 0.6524303184555218, "grad_norm": 0.5213996161112306, "learning_rate": 9.585499561749392e-05, "loss": 3.1138, "step": 10510 }, { "epoch": 0.6524923955552796, "grad_norm": 0.5729508967722067, "learning_rate": 9.585355573690957e-05, "loss": 3.1705, "step": 10511 }, { "epoch": 0.6525544726550375, "grad_norm": 0.5764954773526155, "learning_rate": 9.585211561709609e-05, "loss": 3.0808, "step": 10512 }, { "epoch": 0.6526165497547954, "grad_norm": 0.4959968261953685, "learning_rate": 9.585067525806097e-05, "loss": 3.1293, "step": 10513 }, { "epoch": 0.6526786268545534, "grad_norm": 0.5441372325651804, "learning_rate": 9.584923465981175e-05, "loss": 3.1295, "step": 10514 }, { "epoch": 0.6527407039543113, "grad_norm": 0.5452275939685897, "learning_rate": 9.584779382235596e-05, "loss": 3.1042, "step": 10515 }, { "epoch": 0.6528027810540692, "grad_norm": 0.6142532275685281, "learning_rate": 9.584635274570109e-05, "loss": 3.2075, "step": 10516 }, { "epoch": 0.652864858153827, "grad_norm": 0.5949497900141348, "learning_rate": 9.584491142985466e-05, "loss": 3.1951, "step": 10517 }, { "epoch": 0.6529269352535849, "grad_norm": 0.54373784106461, "learning_rate": 9.58434698748242e-05, "loss": 3.1547, "step": 10518 }, { "epoch": 0.6529890123533428, "grad_norm": 0.5149532624212184, "learning_rate": 9.584202808061724e-05, "loss": 3.0696, "step": 10519 }, { "epoch": 0.6530510894531008, "grad_norm": 0.5113724563300389, "learning_rate": 9.584058604724128e-05, "loss": 3.1219, "step": 10520 }, { "epoch": 0.6531131665528587, "grad_norm": 0.5437976081877619, "learning_rate": 9.583914377470387e-05, "loss": 3.1732, "step": 10521 }, { "epoch": 0.6531752436526166, "grad_norm": 0.5249202281655255, "learning_rate": 9.58377012630125e-05, "loss": 3.1204, "step": 10522 }, { "epoch": 0.6532373207523744, "grad_norm": 0.4634044226162078, "learning_rate": 9.583625851217473e-05, "loss": 3.2372, "step": 10523 }, { "epoch": 0.6532993978521323, "grad_norm": 0.5133012079687301, "learning_rate": 9.583481552219806e-05, "loss": 3.1365, "step": 10524 }, { "epoch": 0.6533614749518902, "grad_norm": 0.548727739052026, "learning_rate": 9.583337229309004e-05, "loss": 3.1675, "step": 10525 }, { "epoch": 0.6534235520516481, "grad_norm": 0.5114251961885915, "learning_rate": 9.583192882485818e-05, "loss": 3.0333, "step": 10526 }, { "epoch": 0.6534856291514061, "grad_norm": 0.4998429468976623, "learning_rate": 9.583048511751003e-05, "loss": 3.1105, "step": 10527 }, { "epoch": 0.653547706251164, "grad_norm": 0.5870660467650884, "learning_rate": 9.582904117105312e-05, "loss": 3.1277, "step": 10528 }, { "epoch": 0.6536097833509218, "grad_norm": 0.5805179457013987, "learning_rate": 9.582759698549496e-05, "loss": 3.0474, "step": 10529 }, { "epoch": 0.6536718604506797, "grad_norm": 0.5100243156844745, "learning_rate": 9.582615256084311e-05, "loss": 3.1549, "step": 10530 }, { "epoch": 0.6537339375504376, "grad_norm": 0.5733588910597845, "learning_rate": 9.58247078971051e-05, "loss": 3.1723, "step": 10531 }, { "epoch": 0.6537960146501955, "grad_norm": 0.5939737450500073, "learning_rate": 9.582326299428845e-05, "loss": 3.0787, "step": 10532 }, { "epoch": 0.6538580917499535, "grad_norm": 0.47254818082823963, "learning_rate": 9.582181785240073e-05, "loss": 3.08, "step": 10533 }, { "epoch": 0.6539201688497114, "grad_norm": 0.5209851965439413, "learning_rate": 9.582037247144946e-05, "loss": 3.1369, "step": 10534 }, { "epoch": 0.6539822459494692, "grad_norm": 0.6215967025283976, "learning_rate": 9.581892685144218e-05, "loss": 3.1716, "step": 10535 }, { "epoch": 0.6540443230492271, "grad_norm": 0.4721320847738969, "learning_rate": 9.581748099238643e-05, "loss": 3.1282, "step": 10536 }, { "epoch": 0.654106400148985, "grad_norm": 0.5040225996199557, "learning_rate": 9.581603489428976e-05, "loss": 3.0923, "step": 10537 }, { "epoch": 0.6541684772487429, "grad_norm": 0.4789810252970807, "learning_rate": 9.581458855715973e-05, "loss": 3.0563, "step": 10538 }, { "epoch": 0.6542305543485009, "grad_norm": 0.5186576381866345, "learning_rate": 9.581314198100386e-05, "loss": 3.2137, "step": 10539 }, { "epoch": 0.6542926314482588, "grad_norm": 0.5476987356992139, "learning_rate": 9.581169516582971e-05, "loss": 3.0754, "step": 10540 }, { "epoch": 0.6543547085480166, "grad_norm": 0.43661433218726897, "learning_rate": 9.58102481116448e-05, "loss": 3.1345, "step": 10541 }, { "epoch": 0.6544167856477745, "grad_norm": 0.5675385043235806, "learning_rate": 9.580880081845673e-05, "loss": 3.0808, "step": 10542 }, { "epoch": 0.6544788627475324, "grad_norm": 0.43317926201825363, "learning_rate": 9.580735328627304e-05, "loss": 3.0761, "step": 10543 }, { "epoch": 0.6545409398472903, "grad_norm": 0.519242830455654, "learning_rate": 9.580590551510125e-05, "loss": 3.0451, "step": 10544 }, { "epoch": 0.6546030169470483, "grad_norm": 0.5175460517590447, "learning_rate": 9.580445750494892e-05, "loss": 3.1314, "step": 10545 }, { "epoch": 0.6546650940468062, "grad_norm": 0.4958736952881589, "learning_rate": 9.580300925582362e-05, "loss": 3.0651, "step": 10546 }, { "epoch": 0.654727171146564, "grad_norm": 0.5095476583384918, "learning_rate": 9.580156076773291e-05, "loss": 3.1089, "step": 10547 }, { "epoch": 0.6547892482463219, "grad_norm": 0.5310133238519891, "learning_rate": 9.580011204068433e-05, "loss": 3.1605, "step": 10548 }, { "epoch": 0.6548513253460798, "grad_norm": 0.5391228683503833, "learning_rate": 9.579866307468546e-05, "loss": 3.0454, "step": 10549 }, { "epoch": 0.6549134024458377, "grad_norm": 0.5078222713997537, "learning_rate": 9.579721386974385e-05, "loss": 3.1405, "step": 10550 }, { "epoch": 0.6549754795455957, "grad_norm": 0.49825101670383015, "learning_rate": 9.579576442586703e-05, "loss": 3.0655, "step": 10551 }, { "epoch": 0.6550375566453536, "grad_norm": 0.5385962695313772, "learning_rate": 9.579431474306262e-05, "loss": 3.0883, "step": 10552 }, { "epoch": 0.6550996337451114, "grad_norm": 0.43482396299060544, "learning_rate": 9.579286482133814e-05, "loss": 3.1214, "step": 10553 }, { "epoch": 0.6551617108448693, "grad_norm": 0.5507481071294186, "learning_rate": 9.579141466070117e-05, "loss": 3.2032, "step": 10554 }, { "epoch": 0.6552237879446272, "grad_norm": 0.5122525583113033, "learning_rate": 9.578996426115927e-05, "loss": 3.1002, "step": 10555 }, { "epoch": 0.6552858650443851, "grad_norm": 0.4879598246261084, "learning_rate": 9.578851362272001e-05, "loss": 3.1561, "step": 10556 }, { "epoch": 0.655347942144143, "grad_norm": 0.5632830838029405, "learning_rate": 9.578706274539095e-05, "loss": 3.1326, "step": 10557 }, { "epoch": 0.655410019243901, "grad_norm": 0.5052202819774565, "learning_rate": 9.57856116291797e-05, "loss": 3.0525, "step": 10558 }, { "epoch": 0.6554720963436588, "grad_norm": 0.4879366765152755, "learning_rate": 9.578416027409376e-05, "loss": 3.1256, "step": 10559 }, { "epoch": 0.6555341734434167, "grad_norm": 0.5254496998528253, "learning_rate": 9.578270868014076e-05, "loss": 3.2176, "step": 10560 }, { "epoch": 0.6555962505431746, "grad_norm": 0.5068604862096646, "learning_rate": 9.578125684732826e-05, "loss": 3.13, "step": 10561 }, { "epoch": 0.6556583276429325, "grad_norm": 0.5235662444337006, "learning_rate": 9.577980477566381e-05, "loss": 3.0796, "step": 10562 }, { "epoch": 0.6557204047426904, "grad_norm": 0.5134004231007984, "learning_rate": 9.577835246515503e-05, "loss": 3.2796, "step": 10563 }, { "epoch": 0.6557824818424484, "grad_norm": 0.5822492869541884, "learning_rate": 9.577689991580947e-05, "loss": 3.1102, "step": 10564 }, { "epoch": 0.6558445589422062, "grad_norm": 0.47219007533019325, "learning_rate": 9.57754471276347e-05, "loss": 3.1251, "step": 10565 }, { "epoch": 0.6559066360419641, "grad_norm": 0.5012583098352982, "learning_rate": 9.577399410063831e-05, "loss": 3.1563, "step": 10566 }, { "epoch": 0.655968713141722, "grad_norm": 0.5598802490431192, "learning_rate": 9.577254083482787e-05, "loss": 3.2161, "step": 10567 }, { "epoch": 0.6560307902414799, "grad_norm": 0.522604464487203, "learning_rate": 9.577108733021099e-05, "loss": 3.124, "step": 10568 }, { "epoch": 0.6560928673412378, "grad_norm": 0.5298601369930772, "learning_rate": 9.576963358679522e-05, "loss": 3.1352, "step": 10569 }, { "epoch": 0.6561549444409958, "grad_norm": 0.5359517887979958, "learning_rate": 9.57681796045882e-05, "loss": 3.022, "step": 10570 }, { "epoch": 0.6562170215407536, "grad_norm": 0.6483284272528839, "learning_rate": 9.576672538359743e-05, "loss": 3.1229, "step": 10571 }, { "epoch": 0.6562790986405115, "grad_norm": 0.5525273720619371, "learning_rate": 9.576527092383056e-05, "loss": 3.0984, "step": 10572 }, { "epoch": 0.6563411757402694, "grad_norm": 0.5294622184126748, "learning_rate": 9.576381622529518e-05, "loss": 3.099, "step": 10573 }, { "epoch": 0.6564032528400273, "grad_norm": 0.5356974684255433, "learning_rate": 9.576236128799884e-05, "loss": 3.1182, "step": 10574 }, { "epoch": 0.6564653299397852, "grad_norm": 0.5871496759061156, "learning_rate": 9.576090611194915e-05, "loss": 3.1935, "step": 10575 }, { "epoch": 0.6565274070395432, "grad_norm": 0.5357305967027641, "learning_rate": 9.575945069715371e-05, "loss": 3.1827, "step": 10576 }, { "epoch": 0.656589484139301, "grad_norm": 0.53404421021118, "learning_rate": 9.57579950436201e-05, "loss": 3.132, "step": 10577 }, { "epoch": 0.6566515612390589, "grad_norm": 0.5352934501362678, "learning_rate": 9.575653915135593e-05, "loss": 3.1087, "step": 10578 }, { "epoch": 0.6567136383388168, "grad_norm": 0.4879475483241438, "learning_rate": 9.57550830203688e-05, "loss": 3.1682, "step": 10579 }, { "epoch": 0.6567757154385747, "grad_norm": 0.5255429072008417, "learning_rate": 9.575362665066627e-05, "loss": 3.0981, "step": 10580 }, { "epoch": 0.6568377925383326, "grad_norm": 0.4582600824024215, "learning_rate": 9.575217004225598e-05, "loss": 3.1048, "step": 10581 }, { "epoch": 0.6568998696380905, "grad_norm": 0.4822634695583657, "learning_rate": 9.57507131951455e-05, "loss": 3.0614, "step": 10582 }, { "epoch": 0.6569619467378484, "grad_norm": 0.4920842047184855, "learning_rate": 9.574925610934245e-05, "loss": 3.1638, "step": 10583 }, { "epoch": 0.6570240238376063, "grad_norm": 0.5013717966198498, "learning_rate": 9.574779878485443e-05, "loss": 3.1591, "step": 10584 }, { "epoch": 0.6570861009373642, "grad_norm": 0.5019046151436387, "learning_rate": 9.574634122168902e-05, "loss": 3.1043, "step": 10585 }, { "epoch": 0.6571481780371221, "grad_norm": 0.4407865174150183, "learning_rate": 9.574488341985385e-05, "loss": 3.1488, "step": 10586 }, { "epoch": 0.65721025513688, "grad_norm": 0.49652836002265327, "learning_rate": 9.574342537935653e-05, "loss": 3.0973, "step": 10587 }, { "epoch": 0.657272332236638, "grad_norm": 0.5138853666539955, "learning_rate": 9.574196710020465e-05, "loss": 3.0479, "step": 10588 }, { "epoch": 0.6573344093363958, "grad_norm": 0.5506003603580375, "learning_rate": 9.574050858240582e-05, "loss": 3.1681, "step": 10589 }, { "epoch": 0.6573964864361537, "grad_norm": 0.4572061220445449, "learning_rate": 9.573904982596766e-05, "loss": 3.2065, "step": 10590 }, { "epoch": 0.6574585635359116, "grad_norm": 0.5121001067439488, "learning_rate": 9.573759083089778e-05, "loss": 3.1519, "step": 10591 }, { "epoch": 0.6575206406356695, "grad_norm": 0.5212390908579377, "learning_rate": 9.573613159720376e-05, "loss": 3.0876, "step": 10592 }, { "epoch": 0.6575827177354274, "grad_norm": 0.5043459016709237, "learning_rate": 9.573467212489325e-05, "loss": 3.0513, "step": 10593 }, { "epoch": 0.6576447948351853, "grad_norm": 0.5070676182192704, "learning_rate": 9.573321241397385e-05, "loss": 3.1399, "step": 10594 }, { "epoch": 0.6577068719349431, "grad_norm": 0.4533362060811143, "learning_rate": 9.573175246445318e-05, "loss": 3.0476, "step": 10595 }, { "epoch": 0.6577689490347011, "grad_norm": 0.488995068444307, "learning_rate": 9.573029227633887e-05, "loss": 3.1366, "step": 10596 }, { "epoch": 0.657831026134459, "grad_norm": 0.4499893544845242, "learning_rate": 9.57288318496385e-05, "loss": 3.0269, "step": 10597 }, { "epoch": 0.6578931032342169, "grad_norm": 0.4732575650615559, "learning_rate": 9.572737118435972e-05, "loss": 3.1819, "step": 10598 }, { "epoch": 0.6579551803339748, "grad_norm": 0.4655135802276346, "learning_rate": 9.572591028051017e-05, "loss": 3.0228, "step": 10599 }, { "epoch": 0.6580172574337327, "grad_norm": 0.47061047097179853, "learning_rate": 9.572444913809742e-05, "loss": 2.995, "step": 10600 }, { "epoch": 0.6580793345334905, "grad_norm": 0.486721033302205, "learning_rate": 9.572298775712912e-05, "loss": 3.1793, "step": 10601 }, { "epoch": 0.6581414116332485, "grad_norm": 0.4692260867627217, "learning_rate": 9.572152613761289e-05, "loss": 3.0235, "step": 10602 }, { "epoch": 0.6582034887330064, "grad_norm": 0.4810343723754617, "learning_rate": 9.572006427955638e-05, "loss": 3.1525, "step": 10603 }, { "epoch": 0.6582655658327643, "grad_norm": 0.5713830593411168, "learning_rate": 9.571860218296718e-05, "loss": 3.1759, "step": 10604 }, { "epoch": 0.6583276429325222, "grad_norm": 0.5866215337069294, "learning_rate": 9.571713984785293e-05, "loss": 3.1922, "step": 10605 }, { "epoch": 0.6583897200322801, "grad_norm": 0.5994853285918293, "learning_rate": 9.571567727422127e-05, "loss": 3.0724, "step": 10606 }, { "epoch": 0.6584517971320379, "grad_norm": 0.5458449062790081, "learning_rate": 9.571421446207982e-05, "loss": 3.165, "step": 10607 }, { "epoch": 0.6585138742317959, "grad_norm": 0.5365901587527546, "learning_rate": 9.571275141143623e-05, "loss": 3.1367, "step": 10608 }, { "epoch": 0.6585759513315538, "grad_norm": 0.5489000157958627, "learning_rate": 9.571128812229811e-05, "loss": 3.0597, "step": 10609 }, { "epoch": 0.6586380284313117, "grad_norm": 0.49925053574404865, "learning_rate": 9.570982459467309e-05, "loss": 3.1774, "step": 10610 }, { "epoch": 0.6587001055310696, "grad_norm": 0.5470483777552725, "learning_rate": 9.570836082856882e-05, "loss": 3.1584, "step": 10611 }, { "epoch": 0.6587621826308275, "grad_norm": 0.5869148495584755, "learning_rate": 9.570689682399295e-05, "loss": 3.1288, "step": 10612 }, { "epoch": 0.6588242597305853, "grad_norm": 0.5054590121102119, "learning_rate": 9.57054325809531e-05, "loss": 3.0075, "step": 10613 }, { "epoch": 0.6588863368303433, "grad_norm": 0.45761918292585385, "learning_rate": 9.570396809945692e-05, "loss": 3.1927, "step": 10614 }, { "epoch": 0.6589484139301012, "grad_norm": 0.6007989521171763, "learning_rate": 9.570250337951204e-05, "loss": 3.1471, "step": 10615 }, { "epoch": 0.6590104910298591, "grad_norm": 0.5914949710870789, "learning_rate": 9.570103842112609e-05, "loss": 3.1257, "step": 10616 }, { "epoch": 0.659072568129617, "grad_norm": 0.5076789386876942, "learning_rate": 9.569957322430673e-05, "loss": 3.1537, "step": 10617 }, { "epoch": 0.6591346452293749, "grad_norm": 0.501721607058744, "learning_rate": 9.56981077890616e-05, "loss": 3.1542, "step": 10618 }, { "epoch": 0.6591967223291327, "grad_norm": 0.5732216254580565, "learning_rate": 9.569664211539837e-05, "loss": 3.0305, "step": 10619 }, { "epoch": 0.6592587994288907, "grad_norm": 0.549757770819949, "learning_rate": 9.569517620332466e-05, "loss": 3.1081, "step": 10620 }, { "epoch": 0.6593208765286486, "grad_norm": 0.5102294039889318, "learning_rate": 9.569371005284811e-05, "loss": 3.1512, "step": 10621 }, { "epoch": 0.6593829536284065, "grad_norm": 0.4948954640652545, "learning_rate": 9.569224366397638e-05, "loss": 3.0769, "step": 10622 }, { "epoch": 0.6594450307281644, "grad_norm": 0.4878759151345578, "learning_rate": 9.569077703671714e-05, "loss": 3.096, "step": 10623 }, { "epoch": 0.6595071078279223, "grad_norm": 0.5999407272069261, "learning_rate": 9.568931017107802e-05, "loss": 3.0859, "step": 10624 }, { "epoch": 0.6595691849276801, "grad_norm": 0.507361343684948, "learning_rate": 9.568784306706666e-05, "loss": 3.0921, "step": 10625 }, { "epoch": 0.659631262027438, "grad_norm": 0.49166494210258666, "learning_rate": 9.568637572469074e-05, "loss": 3.1085, "step": 10626 }, { "epoch": 0.659693339127196, "grad_norm": 0.5065140777167926, "learning_rate": 9.568490814395792e-05, "loss": 2.9976, "step": 10627 }, { "epoch": 0.6597554162269539, "grad_norm": 0.5189044584560969, "learning_rate": 9.568344032487584e-05, "loss": 3.1616, "step": 10628 }, { "epoch": 0.6598174933267118, "grad_norm": 0.5099579605372133, "learning_rate": 9.568197226745214e-05, "loss": 2.9933, "step": 10629 }, { "epoch": 0.6598795704264697, "grad_norm": 0.5031579842032615, "learning_rate": 9.568050397169453e-05, "loss": 3.1262, "step": 10630 }, { "epoch": 0.6599416475262275, "grad_norm": 0.48468784258478037, "learning_rate": 9.567903543761063e-05, "loss": 3.0606, "step": 10631 }, { "epoch": 0.6600037246259854, "grad_norm": 0.5232951696668277, "learning_rate": 9.567756666520811e-05, "loss": 3.1042, "step": 10632 }, { "epoch": 0.6600658017257434, "grad_norm": 0.43950008565274773, "learning_rate": 9.567609765449464e-05, "loss": 3.0802, "step": 10633 }, { "epoch": 0.6601278788255013, "grad_norm": 0.5280689480622992, "learning_rate": 9.567462840547787e-05, "loss": 3.0622, "step": 10634 }, { "epoch": 0.6601899559252592, "grad_norm": 0.4721192706616166, "learning_rate": 9.56731589181655e-05, "loss": 3.1892, "step": 10635 }, { "epoch": 0.6602520330250171, "grad_norm": 0.453862970491651, "learning_rate": 9.567168919256515e-05, "loss": 3.0686, "step": 10636 }, { "epoch": 0.6603141101247749, "grad_norm": 0.4459036032686169, "learning_rate": 9.56702192286845e-05, "loss": 3.0901, "step": 10637 }, { "epoch": 0.6603761872245328, "grad_norm": 0.5024282974594307, "learning_rate": 9.566874902653126e-05, "loss": 3.1498, "step": 10638 }, { "epoch": 0.6604382643242908, "grad_norm": 0.496976112777974, "learning_rate": 9.566727858611306e-05, "loss": 3.0836, "step": 10639 }, { "epoch": 0.6605003414240487, "grad_norm": 0.41312316293928375, "learning_rate": 9.566580790743758e-05, "loss": 3.0434, "step": 10640 }, { "epoch": 0.6605624185238066, "grad_norm": 0.42375163077994304, "learning_rate": 9.566433699051249e-05, "loss": 3.0572, "step": 10641 }, { "epoch": 0.6606244956235645, "grad_norm": 0.4050505243387232, "learning_rate": 9.566286583534547e-05, "loss": 3.1655, "step": 10642 }, { "epoch": 0.6606865727233223, "grad_norm": 0.461630074623698, "learning_rate": 9.56613944419442e-05, "loss": 3.0691, "step": 10643 }, { "epoch": 0.6607486498230802, "grad_norm": 0.474202456500763, "learning_rate": 9.565992281031635e-05, "loss": 3.1542, "step": 10644 }, { "epoch": 0.6608107269228382, "grad_norm": 0.4640996410778641, "learning_rate": 9.565845094046959e-05, "loss": 3.0852, "step": 10645 }, { "epoch": 0.6608728040225961, "grad_norm": 0.4026173707475483, "learning_rate": 9.56569788324116e-05, "loss": 3.0775, "step": 10646 }, { "epoch": 0.660934881122354, "grad_norm": 0.4384029810959855, "learning_rate": 9.565550648615008e-05, "loss": 3.0646, "step": 10647 }, { "epoch": 0.6609969582221119, "grad_norm": 0.447251348809031, "learning_rate": 9.56540339016927e-05, "loss": 3.1324, "step": 10648 }, { "epoch": 0.6610590353218697, "grad_norm": 0.44333136045702937, "learning_rate": 9.565256107904714e-05, "loss": 3.0563, "step": 10649 }, { "epoch": 0.6611211124216276, "grad_norm": 0.46996602638196594, "learning_rate": 9.565108801822109e-05, "loss": 3.0362, "step": 10650 }, { "epoch": 0.6611831895213856, "grad_norm": 0.5123419715731817, "learning_rate": 9.564961471922223e-05, "loss": 3.1877, "step": 10651 }, { "epoch": 0.6612452666211435, "grad_norm": 0.4967533892432877, "learning_rate": 9.564814118205825e-05, "loss": 3.0818, "step": 10652 }, { "epoch": 0.6613073437209014, "grad_norm": 0.4980343835797572, "learning_rate": 9.564666740673682e-05, "loss": 3.0797, "step": 10653 }, { "epoch": 0.6613694208206593, "grad_norm": 0.5287651403437359, "learning_rate": 9.564519339326566e-05, "loss": 3.0439, "step": 10654 }, { "epoch": 0.6614314979204171, "grad_norm": 0.4362713728208057, "learning_rate": 9.564371914165244e-05, "loss": 3.1727, "step": 10655 }, { "epoch": 0.661493575020175, "grad_norm": 0.47179493572242714, "learning_rate": 9.564224465190487e-05, "loss": 2.9891, "step": 10656 }, { "epoch": 0.661555652119933, "grad_norm": 0.46204484052024086, "learning_rate": 9.564076992403062e-05, "loss": 3.1083, "step": 10657 }, { "epoch": 0.6616177292196909, "grad_norm": 0.5235115966732974, "learning_rate": 9.563929495803739e-05, "loss": 3.1444, "step": 10658 }, { "epoch": 0.6616798063194488, "grad_norm": 0.4726860257692813, "learning_rate": 9.563781975393287e-05, "loss": 3.0929, "step": 10659 }, { "epoch": 0.6617418834192067, "grad_norm": 0.4689298920358078, "learning_rate": 9.563634431172479e-05, "loss": 3.0956, "step": 10660 }, { "epoch": 0.6618039605189645, "grad_norm": 0.44185037128816645, "learning_rate": 9.56348686314208e-05, "loss": 3.058, "step": 10661 }, { "epoch": 0.6618660376187224, "grad_norm": 0.4558599248129333, "learning_rate": 9.563339271302865e-05, "loss": 3.0804, "step": 10662 }, { "epoch": 0.6619281147184803, "grad_norm": 0.5006769782905888, "learning_rate": 9.563191655655599e-05, "loss": 3.0964, "step": 10663 }, { "epoch": 0.6619901918182383, "grad_norm": 0.5012309003220954, "learning_rate": 9.563044016201055e-05, "loss": 3.1308, "step": 10664 }, { "epoch": 0.6620522689179962, "grad_norm": 0.42210641447953356, "learning_rate": 9.562896352940001e-05, "loss": 3.0293, "step": 10665 }, { "epoch": 0.6621143460177541, "grad_norm": 0.47399217009716854, "learning_rate": 9.562748665873211e-05, "loss": 2.9972, "step": 10666 }, { "epoch": 0.6621764231175119, "grad_norm": 0.4832481998522562, "learning_rate": 9.562600955001453e-05, "loss": 3.1431, "step": 10667 }, { "epoch": 0.6622385002172698, "grad_norm": 0.5291116819825676, "learning_rate": 9.562453220325498e-05, "loss": 3.1211, "step": 10668 }, { "epoch": 0.6623005773170277, "grad_norm": 0.5197058936262928, "learning_rate": 9.562305461846118e-05, "loss": 3.1985, "step": 10669 }, { "epoch": 0.6623626544167857, "grad_norm": 0.5010486870615535, "learning_rate": 9.562157679564082e-05, "loss": 3.1873, "step": 10670 }, { "epoch": 0.6624247315165436, "grad_norm": 0.5130753044475843, "learning_rate": 9.562009873480162e-05, "loss": 3.132, "step": 10671 }, { "epoch": 0.6624868086163015, "grad_norm": 0.5053705630092322, "learning_rate": 9.561862043595129e-05, "loss": 3.107, "step": 10672 }, { "epoch": 0.6625488857160593, "grad_norm": 0.5009199262881147, "learning_rate": 9.561714189909754e-05, "loss": 3.1686, "step": 10673 }, { "epoch": 0.6626109628158172, "grad_norm": 0.5749324593060231, "learning_rate": 9.56156631242481e-05, "loss": 3.089, "step": 10674 }, { "epoch": 0.6626730399155751, "grad_norm": 0.6309214227828241, "learning_rate": 9.561418411141066e-05, "loss": 3.1407, "step": 10675 }, { "epoch": 0.662735117015333, "grad_norm": 0.5415124367952678, "learning_rate": 9.561270486059295e-05, "loss": 3.1208, "step": 10676 }, { "epoch": 0.662797194115091, "grad_norm": 0.4765309964382636, "learning_rate": 9.56112253718027e-05, "loss": 3.161, "step": 10677 }, { "epoch": 0.6628592712148489, "grad_norm": 0.506025309208663, "learning_rate": 9.560974564504759e-05, "loss": 3.0877, "step": 10678 }, { "epoch": 0.6629213483146067, "grad_norm": 0.5313531831504017, "learning_rate": 9.560826568033537e-05, "loss": 3.042, "step": 10679 }, { "epoch": 0.6629834254143646, "grad_norm": 0.5319837943939528, "learning_rate": 9.560678547767377e-05, "loss": 3.196, "step": 10680 }, { "epoch": 0.6630455025141225, "grad_norm": 0.46733335368008555, "learning_rate": 9.560530503707048e-05, "loss": 3.1149, "step": 10681 }, { "epoch": 0.6631075796138804, "grad_norm": 0.49962053204472034, "learning_rate": 9.560382435853326e-05, "loss": 3.1236, "step": 10682 }, { "epoch": 0.6631696567136384, "grad_norm": 0.478532845323517, "learning_rate": 9.560234344206981e-05, "loss": 3.0779, "step": 10683 }, { "epoch": 0.6632317338133963, "grad_norm": 0.4261374252479508, "learning_rate": 9.560086228768786e-05, "loss": 3.2204, "step": 10684 }, { "epoch": 0.6632938109131541, "grad_norm": 0.5219939239137362, "learning_rate": 9.559938089539516e-05, "loss": 3.1739, "step": 10685 }, { "epoch": 0.663355888012912, "grad_norm": 0.48956328885324, "learning_rate": 9.55978992651994e-05, "loss": 3.1661, "step": 10686 }, { "epoch": 0.6634179651126699, "grad_norm": 0.4317270252871912, "learning_rate": 9.559641739710834e-05, "loss": 3.0746, "step": 10687 }, { "epoch": 0.6634800422124278, "grad_norm": 0.5284542400521703, "learning_rate": 9.559493529112968e-05, "loss": 3.0123, "step": 10688 }, { "epoch": 0.6635421193121858, "grad_norm": 0.4984756120323397, "learning_rate": 9.55934529472712e-05, "loss": 3.1309, "step": 10689 }, { "epoch": 0.6636041964119437, "grad_norm": 0.5068080357862905, "learning_rate": 9.55919703655406e-05, "loss": 3.1817, "step": 10690 }, { "epoch": 0.6636662735117015, "grad_norm": 0.506646387307156, "learning_rate": 9.559048754594563e-05, "loss": 3.141, "step": 10691 }, { "epoch": 0.6637283506114594, "grad_norm": 0.5268180487409893, "learning_rate": 9.5589004488494e-05, "loss": 3.0421, "step": 10692 }, { "epoch": 0.6637904277112173, "grad_norm": 0.5011726575623823, "learning_rate": 9.558752119319348e-05, "loss": 3.1194, "step": 10693 }, { "epoch": 0.6638525048109752, "grad_norm": 0.6027292603666965, "learning_rate": 9.558603766005181e-05, "loss": 3.0741, "step": 10694 }, { "epoch": 0.6639145819107332, "grad_norm": 0.5672380325043528, "learning_rate": 9.558455388907669e-05, "loss": 3.1423, "step": 10695 }, { "epoch": 0.6639766590104911, "grad_norm": 0.4962323062874868, "learning_rate": 9.55830698802759e-05, "loss": 3.0549, "step": 10696 }, { "epoch": 0.6640387361102489, "grad_norm": 0.5288459010890544, "learning_rate": 9.558158563365716e-05, "loss": 3.2385, "step": 10697 }, { "epoch": 0.6641008132100068, "grad_norm": 0.526358496429655, "learning_rate": 9.558010114922822e-05, "loss": 3.1811, "step": 10698 }, { "epoch": 0.6641628903097647, "grad_norm": 0.48436810888495246, "learning_rate": 9.557861642699683e-05, "loss": 3.0329, "step": 10699 }, { "epoch": 0.6642249674095226, "grad_norm": 0.5155306932153011, "learning_rate": 9.557713146697074e-05, "loss": 3.1435, "step": 10700 }, { "epoch": 0.6642870445092806, "grad_norm": 0.4563243648594226, "learning_rate": 9.557564626915769e-05, "loss": 3.0956, "step": 10701 }, { "epoch": 0.6643491216090385, "grad_norm": 0.5030814525168263, "learning_rate": 9.557416083356542e-05, "loss": 3.1492, "step": 10702 }, { "epoch": 0.6644111987087963, "grad_norm": 0.4757563698813755, "learning_rate": 9.557267516020171e-05, "loss": 3.0395, "step": 10703 }, { "epoch": 0.6644732758085542, "grad_norm": 0.5025868552855209, "learning_rate": 9.557118924907427e-05, "loss": 3.1485, "step": 10704 }, { "epoch": 0.6645353529083121, "grad_norm": 0.4277042395718977, "learning_rate": 9.55697031001909e-05, "loss": 3.092, "step": 10705 }, { "epoch": 0.66459743000807, "grad_norm": 0.46957469555115267, "learning_rate": 9.556821671355929e-05, "loss": 3.2178, "step": 10706 }, { "epoch": 0.664659507107828, "grad_norm": 0.47787234605795115, "learning_rate": 9.556673008918725e-05, "loss": 3.0927, "step": 10707 }, { "epoch": 0.6647215842075859, "grad_norm": 0.4785945126648548, "learning_rate": 9.556524322708252e-05, "loss": 3.1512, "step": 10708 }, { "epoch": 0.6647836613073437, "grad_norm": 0.5199355939721728, "learning_rate": 9.556375612725284e-05, "loss": 3.1262, "step": 10709 }, { "epoch": 0.6648457384071016, "grad_norm": 0.5236561162103522, "learning_rate": 9.5562268789706e-05, "loss": 3.1769, "step": 10710 }, { "epoch": 0.6649078155068595, "grad_norm": 0.5311713721487137, "learning_rate": 9.556078121444974e-05, "loss": 3.0993, "step": 10711 }, { "epoch": 0.6649698926066174, "grad_norm": 0.5182612679154106, "learning_rate": 9.555929340149181e-05, "loss": 3.1142, "step": 10712 }, { "epoch": 0.6650319697063753, "grad_norm": 0.5036629969431407, "learning_rate": 9.555780535084e-05, "loss": 3.1312, "step": 10713 }, { "epoch": 0.6650940468061333, "grad_norm": 0.5074226937349805, "learning_rate": 9.555631706250206e-05, "loss": 3.1643, "step": 10714 }, { "epoch": 0.6651561239058911, "grad_norm": 0.4954237581322474, "learning_rate": 9.555482853648576e-05, "loss": 3.1199, "step": 10715 }, { "epoch": 0.665218201005649, "grad_norm": 0.5205423368702441, "learning_rate": 9.555333977279884e-05, "loss": 3.0447, "step": 10716 }, { "epoch": 0.6652802781054069, "grad_norm": 0.4974740417935162, "learning_rate": 9.55518507714491e-05, "loss": 3.1006, "step": 10717 }, { "epoch": 0.6653423552051648, "grad_norm": 0.4305027943076296, "learning_rate": 9.55503615324443e-05, "loss": 3.1635, "step": 10718 }, { "epoch": 0.6654044323049227, "grad_norm": 0.49603697684724046, "learning_rate": 9.55488720557922e-05, "loss": 3.1467, "step": 10719 }, { "epoch": 0.6654665094046807, "grad_norm": 0.48643279576983006, "learning_rate": 9.554738234150058e-05, "loss": 3.0928, "step": 10720 }, { "epoch": 0.6655285865044385, "grad_norm": 0.4842062933206757, "learning_rate": 9.554589238957721e-05, "loss": 3.1205, "step": 10721 }, { "epoch": 0.6655906636041964, "grad_norm": 0.4510273061878692, "learning_rate": 9.554440220002986e-05, "loss": 3.1882, "step": 10722 }, { "epoch": 0.6656527407039543, "grad_norm": 0.5028222126612422, "learning_rate": 9.554291177286632e-05, "loss": 3.0883, "step": 10723 }, { "epoch": 0.6657148178037122, "grad_norm": 0.47726065666346507, "learning_rate": 9.554142110809434e-05, "loss": 3.1386, "step": 10724 }, { "epoch": 0.6657768949034701, "grad_norm": 0.5184741631179826, "learning_rate": 9.553993020572172e-05, "loss": 3.1573, "step": 10725 }, { "epoch": 0.6658389720032281, "grad_norm": 0.5416716306410039, "learning_rate": 9.553843906575621e-05, "loss": 3.1121, "step": 10726 }, { "epoch": 0.6659010491029859, "grad_norm": 0.4997864654665788, "learning_rate": 9.553694768820562e-05, "loss": 3.1884, "step": 10727 }, { "epoch": 0.6659631262027438, "grad_norm": 0.6565392628715087, "learning_rate": 9.553545607307773e-05, "loss": 3.1408, "step": 10728 }, { "epoch": 0.6660252033025017, "grad_norm": 0.598197866125009, "learning_rate": 9.55339642203803e-05, "loss": 3.0561, "step": 10729 }, { "epoch": 0.6660872804022596, "grad_norm": 0.5461189766270317, "learning_rate": 9.553247213012114e-05, "loss": 3.1784, "step": 10730 }, { "epoch": 0.6661493575020175, "grad_norm": 0.4939635340179387, "learning_rate": 9.553097980230799e-05, "loss": 3.0901, "step": 10731 }, { "epoch": 0.6662114346017755, "grad_norm": 0.4686082943665403, "learning_rate": 9.552948723694868e-05, "loss": 2.9967, "step": 10732 }, { "epoch": 0.6662735117015333, "grad_norm": 0.4754798191122184, "learning_rate": 9.552799443405098e-05, "loss": 3.1694, "step": 10733 }, { "epoch": 0.6663355888012912, "grad_norm": 0.4575324611404175, "learning_rate": 9.552650139362269e-05, "loss": 3.1044, "step": 10734 }, { "epoch": 0.6663976659010491, "grad_norm": 0.42941096684101643, "learning_rate": 9.552500811567159e-05, "loss": 3.1146, "step": 10735 }, { "epoch": 0.666459743000807, "grad_norm": 0.5466657783289801, "learning_rate": 9.552351460020547e-05, "loss": 3.1135, "step": 10736 }, { "epoch": 0.6665218201005649, "grad_norm": 0.45141225248989203, "learning_rate": 9.552202084723211e-05, "loss": 3.1045, "step": 10737 }, { "epoch": 0.6665838972003229, "grad_norm": 0.4806725604729448, "learning_rate": 9.552052685675933e-05, "loss": 3.0103, "step": 10738 }, { "epoch": 0.6666459743000807, "grad_norm": 0.45207736154939493, "learning_rate": 9.551903262879491e-05, "loss": 3.0273, "step": 10739 }, { "epoch": 0.6667080513998386, "grad_norm": 0.4430110420073681, "learning_rate": 9.551753816334664e-05, "loss": 3.1216, "step": 10740 }, { "epoch": 0.6667701284995965, "grad_norm": 0.449695533166408, "learning_rate": 9.551604346042234e-05, "loss": 3.1066, "step": 10741 }, { "epoch": 0.6668322055993544, "grad_norm": 0.39629571870541064, "learning_rate": 9.551454852002977e-05, "loss": 3.0919, "step": 10742 }, { "epoch": 0.6668942826991123, "grad_norm": 0.46547359782880077, "learning_rate": 9.551305334217675e-05, "loss": 3.1914, "step": 10743 }, { "epoch": 0.6669563597988702, "grad_norm": 0.4476524912092174, "learning_rate": 9.55115579268711e-05, "loss": 3.0204, "step": 10744 }, { "epoch": 0.667018436898628, "grad_norm": 0.4484708281835635, "learning_rate": 9.551006227412059e-05, "loss": 3.0709, "step": 10745 }, { "epoch": 0.667080513998386, "grad_norm": 0.5083499411960274, "learning_rate": 9.550856638393303e-05, "loss": 3.0476, "step": 10746 }, { "epoch": 0.6671425910981439, "grad_norm": 0.5437652454212119, "learning_rate": 9.550707025631627e-05, "loss": 3.0771, "step": 10747 }, { "epoch": 0.6672046681979018, "grad_norm": 0.536082447792309, "learning_rate": 9.550557389127805e-05, "loss": 3.1264, "step": 10748 }, { "epoch": 0.6672667452976597, "grad_norm": 0.4982566242227474, "learning_rate": 9.55040772888262e-05, "loss": 3.1567, "step": 10749 }, { "epoch": 0.6673288223974176, "grad_norm": 0.6009075761189528, "learning_rate": 9.550258044896853e-05, "loss": 3.1245, "step": 10750 }, { "epoch": 0.6673908994971754, "grad_norm": 0.74209891342723, "learning_rate": 9.550108337171286e-05, "loss": 3.0826, "step": 10751 }, { "epoch": 0.6674529765969334, "grad_norm": 0.7730154562141429, "learning_rate": 9.5499586057067e-05, "loss": 3.0972, "step": 10752 }, { "epoch": 0.6675150536966913, "grad_norm": 0.5261430705568225, "learning_rate": 9.549808850503875e-05, "loss": 3.1482, "step": 10753 }, { "epoch": 0.6675771307964492, "grad_norm": 0.6176413304065946, "learning_rate": 9.549659071563592e-05, "loss": 3.0776, "step": 10754 }, { "epoch": 0.6676392078962071, "grad_norm": 0.5325952215529699, "learning_rate": 9.549509268886634e-05, "loss": 3.0422, "step": 10755 }, { "epoch": 0.667701284995965, "grad_norm": 0.5248833231473862, "learning_rate": 9.549359442473781e-05, "loss": 3.1455, "step": 10756 }, { "epoch": 0.6677633620957228, "grad_norm": 0.5407157645355579, "learning_rate": 9.549209592325816e-05, "loss": 3.1166, "step": 10757 }, { "epoch": 0.6678254391954808, "grad_norm": 0.47289642032971313, "learning_rate": 9.54905971844352e-05, "loss": 3.2376, "step": 10758 }, { "epoch": 0.6678875162952387, "grad_norm": 0.5039227505133648, "learning_rate": 9.548909820827674e-05, "loss": 3.025, "step": 10759 }, { "epoch": 0.6679495933949966, "grad_norm": 0.5219773666126479, "learning_rate": 9.548759899479063e-05, "loss": 3.0513, "step": 10760 }, { "epoch": 0.6680116704947545, "grad_norm": 0.5192789311161803, "learning_rate": 9.548609954398467e-05, "loss": 3.1134, "step": 10761 }, { "epoch": 0.6680737475945124, "grad_norm": 0.5391915484774201, "learning_rate": 9.548459985586668e-05, "loss": 3.2315, "step": 10762 }, { "epoch": 0.6681358246942702, "grad_norm": 0.5236809861683702, "learning_rate": 9.548309993044449e-05, "loss": 3.1151, "step": 10763 }, { "epoch": 0.6681979017940282, "grad_norm": 0.5115212794750321, "learning_rate": 9.548159976772592e-05, "loss": 3.1478, "step": 10764 }, { "epoch": 0.6682599788937861, "grad_norm": 0.5308446178953545, "learning_rate": 9.548009936771882e-05, "loss": 3.1587, "step": 10765 }, { "epoch": 0.668322055993544, "grad_norm": 0.5018510232236, "learning_rate": 9.5478598730431e-05, "loss": 3.0842, "step": 10766 }, { "epoch": 0.6683841330933019, "grad_norm": 0.5019590325508854, "learning_rate": 9.547709785587027e-05, "loss": 3.0924, "step": 10767 }, { "epoch": 0.6684462101930598, "grad_norm": 0.5462715634382507, "learning_rate": 9.54755967440445e-05, "loss": 3.0406, "step": 10768 }, { "epoch": 0.6685082872928176, "grad_norm": 0.535175156085336, "learning_rate": 9.547409539496148e-05, "loss": 3.1375, "step": 10769 }, { "epoch": 0.6685703643925756, "grad_norm": 0.4778061882185964, "learning_rate": 9.54725938086291e-05, "loss": 3.0927, "step": 10770 }, { "epoch": 0.6686324414923335, "grad_norm": 0.5013695378354319, "learning_rate": 9.547109198505513e-05, "loss": 3.0968, "step": 10771 }, { "epoch": 0.6686945185920914, "grad_norm": 0.48267469075134206, "learning_rate": 9.546958992424744e-05, "loss": 3.0785, "step": 10772 }, { "epoch": 0.6687565956918493, "grad_norm": 0.47139933845629783, "learning_rate": 9.546808762621386e-05, "loss": 3.0604, "step": 10773 }, { "epoch": 0.6688186727916072, "grad_norm": 0.4947698153305241, "learning_rate": 9.546658509096225e-05, "loss": 3.0976, "step": 10774 }, { "epoch": 0.668880749891365, "grad_norm": 0.5313183796576034, "learning_rate": 9.54650823185004e-05, "loss": 3.1495, "step": 10775 }, { "epoch": 0.668942826991123, "grad_norm": 0.5269461934600173, "learning_rate": 9.54635793088362e-05, "loss": 3.0733, "step": 10776 }, { "epoch": 0.6690049040908809, "grad_norm": 0.5280890390466018, "learning_rate": 9.546207606197745e-05, "loss": 3.1193, "step": 10777 }, { "epoch": 0.6690669811906388, "grad_norm": 0.5568585420995814, "learning_rate": 9.546057257793205e-05, "loss": 3.1365, "step": 10778 }, { "epoch": 0.6691290582903967, "grad_norm": 0.4711260814548063, "learning_rate": 9.545906885670777e-05, "loss": 3.0659, "step": 10779 }, { "epoch": 0.6691911353901546, "grad_norm": 0.5164070267895094, "learning_rate": 9.54575648983125e-05, "loss": 3.0935, "step": 10780 }, { "epoch": 0.6692532124899124, "grad_norm": 0.5592769366356614, "learning_rate": 9.545606070275408e-05, "loss": 3.1869, "step": 10781 }, { "epoch": 0.6693152895896703, "grad_norm": 0.5380645304608437, "learning_rate": 9.545455627004036e-05, "loss": 2.9931, "step": 10782 }, { "epoch": 0.6693773666894283, "grad_norm": 0.5241050733403344, "learning_rate": 9.545305160017919e-05, "loss": 3.0992, "step": 10783 }, { "epoch": 0.6694394437891862, "grad_norm": 0.5354927842573882, "learning_rate": 9.545154669317839e-05, "loss": 3.1463, "step": 10784 }, { "epoch": 0.6695015208889441, "grad_norm": 0.5320628903205056, "learning_rate": 9.545004154904586e-05, "loss": 3.0734, "step": 10785 }, { "epoch": 0.669563597988702, "grad_norm": 0.48881843009798087, "learning_rate": 9.544853616778943e-05, "loss": 2.974, "step": 10786 }, { "epoch": 0.6696256750884598, "grad_norm": 0.48544161784288303, "learning_rate": 9.544703054941696e-05, "loss": 3.1729, "step": 10787 }, { "epoch": 0.6696877521882177, "grad_norm": 0.552002196077802, "learning_rate": 9.544552469393628e-05, "loss": 3.1818, "step": 10788 }, { "epoch": 0.6697498292879757, "grad_norm": 0.5374724392145778, "learning_rate": 9.544401860135528e-05, "loss": 3.0674, "step": 10789 }, { "epoch": 0.6698119063877336, "grad_norm": 0.5589046013487574, "learning_rate": 9.544251227168179e-05, "loss": 3.0616, "step": 10790 }, { "epoch": 0.6698739834874915, "grad_norm": 0.5093776644303547, "learning_rate": 9.544100570492369e-05, "loss": 3.087, "step": 10791 }, { "epoch": 0.6699360605872494, "grad_norm": 0.537795471329665, "learning_rate": 9.543949890108884e-05, "loss": 3.0839, "step": 10792 }, { "epoch": 0.6699981376870072, "grad_norm": 0.5063023634308984, "learning_rate": 9.543799186018509e-05, "loss": 3.0634, "step": 10793 }, { "epoch": 0.6700602147867651, "grad_norm": 0.543954380035158, "learning_rate": 9.54364845822203e-05, "loss": 3.1242, "step": 10794 }, { "epoch": 0.6701222918865231, "grad_norm": 0.5499340332091924, "learning_rate": 9.543497706720234e-05, "loss": 3.131, "step": 10795 }, { "epoch": 0.670184368986281, "grad_norm": 0.5066232106224026, "learning_rate": 9.543346931513908e-05, "loss": 3.1767, "step": 10796 }, { "epoch": 0.6702464460860389, "grad_norm": 0.5466975605432438, "learning_rate": 9.543196132603837e-05, "loss": 3.1409, "step": 10797 }, { "epoch": 0.6703085231857968, "grad_norm": 0.5820700613305088, "learning_rate": 9.54304530999081e-05, "loss": 3.0274, "step": 10798 }, { "epoch": 0.6703706002855546, "grad_norm": 0.5265277624699058, "learning_rate": 9.542894463675611e-05, "loss": 3.2178, "step": 10799 }, { "epoch": 0.6704326773853125, "grad_norm": 0.511602810836822, "learning_rate": 9.542743593659031e-05, "loss": 3.0099, "step": 10800 }, { "epoch": 0.6704947544850705, "grad_norm": 0.49398115095190187, "learning_rate": 9.542592699941854e-05, "loss": 3.2117, "step": 10801 }, { "epoch": 0.6705568315848284, "grad_norm": 0.5544605865556955, "learning_rate": 9.54244178252487e-05, "loss": 3.1264, "step": 10802 }, { "epoch": 0.6706189086845863, "grad_norm": 0.5027962813242141, "learning_rate": 9.54229084140886e-05, "loss": 3.149, "step": 10803 }, { "epoch": 0.6706809857843442, "grad_norm": 0.5267918842684032, "learning_rate": 9.54213987659462e-05, "loss": 3.0644, "step": 10804 }, { "epoch": 0.670743062884102, "grad_norm": 0.39752907133858595, "learning_rate": 9.541988888082932e-05, "loss": 3.1377, "step": 10805 }, { "epoch": 0.6708051399838599, "grad_norm": 0.48362985331372027, "learning_rate": 9.541837875874586e-05, "loss": 3.1101, "step": 10806 }, { "epoch": 0.6708672170836179, "grad_norm": 0.42097019371616, "learning_rate": 9.541686839970369e-05, "loss": 3.128, "step": 10807 }, { "epoch": 0.6709292941833758, "grad_norm": 0.5242161155213788, "learning_rate": 9.54153578037107e-05, "loss": 3.16, "step": 10808 }, { "epoch": 0.6709913712831337, "grad_norm": 0.4897439791176384, "learning_rate": 9.541384697077474e-05, "loss": 3.1455, "step": 10809 }, { "epoch": 0.6710534483828916, "grad_norm": 0.474309269091713, "learning_rate": 9.541233590090373e-05, "loss": 3.1274, "step": 10810 }, { "epoch": 0.6711155254826494, "grad_norm": 0.4478557142289009, "learning_rate": 9.541082459410555e-05, "loss": 3.1105, "step": 10811 }, { "epoch": 0.6711776025824073, "grad_norm": 0.4863892482355866, "learning_rate": 9.540931305038806e-05, "loss": 3.1185, "step": 10812 }, { "epoch": 0.6712396796821652, "grad_norm": 0.44137728435710627, "learning_rate": 9.540780126975915e-05, "loss": 3.1184, "step": 10813 }, { "epoch": 0.6713017567819232, "grad_norm": 0.4765156738539399, "learning_rate": 9.540628925222675e-05, "loss": 3.1608, "step": 10814 }, { "epoch": 0.6713638338816811, "grad_norm": 0.5107672194697833, "learning_rate": 9.540477699779869e-05, "loss": 3.1098, "step": 10815 }, { "epoch": 0.671425910981439, "grad_norm": 0.4929430778094009, "learning_rate": 9.540326450648289e-05, "loss": 3.154, "step": 10816 }, { "epoch": 0.6714879880811968, "grad_norm": 0.4930382445330613, "learning_rate": 9.540175177828724e-05, "loss": 3.0413, "step": 10817 }, { "epoch": 0.6715500651809547, "grad_norm": 0.5120391716110047, "learning_rate": 9.540023881321962e-05, "loss": 3.1181, "step": 10818 }, { "epoch": 0.6716121422807126, "grad_norm": 0.5213321363562593, "learning_rate": 9.539872561128796e-05, "loss": 3.0767, "step": 10819 }, { "epoch": 0.6716742193804706, "grad_norm": 0.5461844853236031, "learning_rate": 9.53972121725001e-05, "loss": 3.1538, "step": 10820 }, { "epoch": 0.6717362964802285, "grad_norm": 0.5158410631113184, "learning_rate": 9.539569849686398e-05, "loss": 3.0473, "step": 10821 }, { "epoch": 0.6717983735799864, "grad_norm": 0.49590706504116816, "learning_rate": 9.539418458438746e-05, "loss": 3.0404, "step": 10822 }, { "epoch": 0.6718604506797442, "grad_norm": 0.4777194035602291, "learning_rate": 9.539267043507849e-05, "loss": 3.1332, "step": 10823 }, { "epoch": 0.6719225277795021, "grad_norm": 0.5372263278210377, "learning_rate": 9.539115604894492e-05, "loss": 3.0878, "step": 10824 }, { "epoch": 0.67198460487926, "grad_norm": 0.5439316150303536, "learning_rate": 9.538964142599467e-05, "loss": 3.0759, "step": 10825 }, { "epoch": 0.672046681979018, "grad_norm": 0.5253879305972529, "learning_rate": 9.538812656623564e-05, "loss": 3.1425, "step": 10826 }, { "epoch": 0.6721087590787759, "grad_norm": 0.5786314627450683, "learning_rate": 9.538661146967574e-05, "loss": 3.1707, "step": 10827 }, { "epoch": 0.6721708361785338, "grad_norm": 0.551318346401448, "learning_rate": 9.538509613632287e-05, "loss": 3.1657, "step": 10828 }, { "epoch": 0.6722329132782916, "grad_norm": 0.48235163003498394, "learning_rate": 9.538358056618495e-05, "loss": 3.0926, "step": 10829 }, { "epoch": 0.6722949903780495, "grad_norm": 0.5262592891589094, "learning_rate": 9.538206475926985e-05, "loss": 3.1673, "step": 10830 }, { "epoch": 0.6723570674778074, "grad_norm": 0.4818814142149091, "learning_rate": 9.53805487155855e-05, "loss": 3.1106, "step": 10831 }, { "epoch": 0.6724191445775654, "grad_norm": 0.5383097603698414, "learning_rate": 9.537903243513984e-05, "loss": 3.1314, "step": 10832 }, { "epoch": 0.6724812216773233, "grad_norm": 0.5196941951978773, "learning_rate": 9.537751591794073e-05, "loss": 3.0911, "step": 10833 }, { "epoch": 0.6725432987770812, "grad_norm": 0.5085472952960848, "learning_rate": 9.53759991639961e-05, "loss": 3.141, "step": 10834 }, { "epoch": 0.672605375876839, "grad_norm": 0.5854440264117641, "learning_rate": 9.537448217331387e-05, "loss": 3.0521, "step": 10835 }, { "epoch": 0.6726674529765969, "grad_norm": 0.567328445534919, "learning_rate": 9.537296494590196e-05, "loss": 3.0654, "step": 10836 }, { "epoch": 0.6727295300763548, "grad_norm": 0.5429240421029381, "learning_rate": 9.537144748176827e-05, "loss": 3.1122, "step": 10837 }, { "epoch": 0.6727916071761127, "grad_norm": 0.5371722652722715, "learning_rate": 9.536992978092071e-05, "loss": 3.1403, "step": 10838 }, { "epoch": 0.6728536842758707, "grad_norm": 0.5068504904713872, "learning_rate": 9.536841184336724e-05, "loss": 3.0904, "step": 10839 }, { "epoch": 0.6729157613756286, "grad_norm": 0.5592314541299102, "learning_rate": 9.536689366911574e-05, "loss": 3.1469, "step": 10840 }, { "epoch": 0.6729778384753864, "grad_norm": 0.4955097124772264, "learning_rate": 9.536537525817415e-05, "loss": 3.1353, "step": 10841 }, { "epoch": 0.6730399155751443, "grad_norm": 0.4724806309569994, "learning_rate": 9.536385661055037e-05, "loss": 3.1072, "step": 10842 }, { "epoch": 0.6731019926749022, "grad_norm": 0.5222623358475277, "learning_rate": 9.536233772625235e-05, "loss": 3.1469, "step": 10843 }, { "epoch": 0.6731640697746601, "grad_norm": 0.4855751428331802, "learning_rate": 9.536081860528799e-05, "loss": 3.0004, "step": 10844 }, { "epoch": 0.6732261468744181, "grad_norm": 0.5207358714149817, "learning_rate": 9.535929924766524e-05, "loss": 3.1798, "step": 10845 }, { "epoch": 0.673288223974176, "grad_norm": 0.5135492678912172, "learning_rate": 9.5357779653392e-05, "loss": 3.0041, "step": 10846 }, { "epoch": 0.6733503010739338, "grad_norm": 0.5693995863864416, "learning_rate": 9.535625982247622e-05, "loss": 3.1325, "step": 10847 }, { "epoch": 0.6734123781736917, "grad_norm": 0.5383934347018118, "learning_rate": 9.535473975492584e-05, "loss": 3.0807, "step": 10848 }, { "epoch": 0.6734744552734496, "grad_norm": 0.47553912486577454, "learning_rate": 9.535321945074876e-05, "loss": 3.1926, "step": 10849 }, { "epoch": 0.6735365323732075, "grad_norm": 0.5057061453024396, "learning_rate": 9.53516989099529e-05, "loss": 3.0401, "step": 10850 }, { "epoch": 0.6735986094729655, "grad_norm": 0.5718845752268819, "learning_rate": 9.535017813254626e-05, "loss": 3.096, "step": 10851 }, { "epoch": 0.6736606865727234, "grad_norm": 0.5441911234429979, "learning_rate": 9.534865711853669e-05, "loss": 3.0492, "step": 10852 }, { "epoch": 0.6737227636724812, "grad_norm": 0.48795286198529264, "learning_rate": 9.534713586793219e-05, "loss": 3.0648, "step": 10853 }, { "epoch": 0.6737848407722391, "grad_norm": 0.47354987293298495, "learning_rate": 9.534561438074068e-05, "loss": 3.1336, "step": 10854 }, { "epoch": 0.673846917871997, "grad_norm": 0.5465108203828887, "learning_rate": 9.534409265697007e-05, "loss": 3.0952, "step": 10855 }, { "epoch": 0.6739089949717549, "grad_norm": 0.623629091695551, "learning_rate": 9.534257069662832e-05, "loss": 3.164, "step": 10856 }, { "epoch": 0.6739710720715129, "grad_norm": 0.5132290632174138, "learning_rate": 9.534104849972341e-05, "loss": 3.1458, "step": 10857 }, { "epoch": 0.6740331491712708, "grad_norm": 0.5483891370701882, "learning_rate": 9.533952606626321e-05, "loss": 3.1274, "step": 10858 }, { "epoch": 0.6740952262710286, "grad_norm": 0.5733629453667183, "learning_rate": 9.533800339625569e-05, "loss": 3.0071, "step": 10859 }, { "epoch": 0.6741573033707865, "grad_norm": 0.5031697117148002, "learning_rate": 9.533648048970882e-05, "loss": 2.9944, "step": 10860 }, { "epoch": 0.6742193804705444, "grad_norm": 0.467156387492397, "learning_rate": 9.533495734663051e-05, "loss": 3.0788, "step": 10861 }, { "epoch": 0.6742814575703023, "grad_norm": 0.4982678241182972, "learning_rate": 9.533343396702874e-05, "loss": 3.0712, "step": 10862 }, { "epoch": 0.6743435346700603, "grad_norm": 0.4896532847435023, "learning_rate": 9.533191035091142e-05, "loss": 3.0684, "step": 10863 }, { "epoch": 0.6744056117698182, "grad_norm": 0.5293263782569577, "learning_rate": 9.533038649828654e-05, "loss": 3.0759, "step": 10864 }, { "epoch": 0.674467688869576, "grad_norm": 0.6108567733309188, "learning_rate": 9.532886240916201e-05, "loss": 3.1542, "step": 10865 }, { "epoch": 0.6745297659693339, "grad_norm": 0.6188727680906494, "learning_rate": 9.532733808354581e-05, "loss": 3.197, "step": 10866 }, { "epoch": 0.6745918430690918, "grad_norm": 0.5237249183111715, "learning_rate": 9.53258135214459e-05, "loss": 3.136, "step": 10867 }, { "epoch": 0.6746539201688497, "grad_norm": 0.5025907096313058, "learning_rate": 9.53242887228702e-05, "loss": 3.1007, "step": 10868 }, { "epoch": 0.6747159972686076, "grad_norm": 0.6050663010468281, "learning_rate": 9.532276368782667e-05, "loss": 3.1569, "step": 10869 }, { "epoch": 0.6747780743683655, "grad_norm": 0.5482226524229354, "learning_rate": 9.53212384163233e-05, "loss": 3.1483, "step": 10870 }, { "epoch": 0.6748401514681234, "grad_norm": 0.5369220476330352, "learning_rate": 9.531971290836801e-05, "loss": 3.1894, "step": 10871 }, { "epoch": 0.6749022285678813, "grad_norm": 0.5271682164543994, "learning_rate": 9.531818716396878e-05, "loss": 3.1442, "step": 10872 }, { "epoch": 0.6749643056676392, "grad_norm": 0.4829796566529254, "learning_rate": 9.531666118313359e-05, "loss": 3.0886, "step": 10873 }, { "epoch": 0.6750263827673971, "grad_norm": 0.5313468452143145, "learning_rate": 9.531513496587035e-05, "loss": 3.1364, "step": 10874 }, { "epoch": 0.675088459867155, "grad_norm": 0.532102153831572, "learning_rate": 9.531360851218706e-05, "loss": 3.0014, "step": 10875 }, { "epoch": 0.6751505369669129, "grad_norm": 0.508009241579503, "learning_rate": 9.531208182209168e-05, "loss": 3.0672, "step": 10876 }, { "epoch": 0.6752126140666708, "grad_norm": 0.518397880452144, "learning_rate": 9.531055489559217e-05, "loss": 3.2173, "step": 10877 }, { "epoch": 0.6752746911664287, "grad_norm": 0.48353799674194464, "learning_rate": 9.530902773269647e-05, "loss": 3.1084, "step": 10878 }, { "epoch": 0.6753367682661866, "grad_norm": 0.507913110009692, "learning_rate": 9.53075003334126e-05, "loss": 3.1434, "step": 10879 }, { "epoch": 0.6753988453659445, "grad_norm": 0.5865864783676615, "learning_rate": 9.53059726977485e-05, "loss": 3.058, "step": 10880 }, { "epoch": 0.6754609224657024, "grad_norm": 0.5715256942653365, "learning_rate": 9.530444482571213e-05, "loss": 3.1269, "step": 10881 }, { "epoch": 0.6755229995654602, "grad_norm": 0.5624412399966277, "learning_rate": 9.530291671731148e-05, "loss": 3.1437, "step": 10882 }, { "epoch": 0.6755850766652182, "grad_norm": 0.5170070448622336, "learning_rate": 9.530138837255452e-05, "loss": 3.1581, "step": 10883 }, { "epoch": 0.6756471537649761, "grad_norm": 0.5330058421052749, "learning_rate": 9.529985979144922e-05, "loss": 3.0821, "step": 10884 }, { "epoch": 0.675709230864734, "grad_norm": 0.49620305756957256, "learning_rate": 9.529833097400354e-05, "loss": 2.9859, "step": 10885 }, { "epoch": 0.6757713079644919, "grad_norm": 0.5456876125875166, "learning_rate": 9.529680192022549e-05, "loss": 3.1758, "step": 10886 }, { "epoch": 0.6758333850642498, "grad_norm": 0.4861201105722835, "learning_rate": 9.529527263012301e-05, "loss": 3.1457, "step": 10887 }, { "epoch": 0.6758954621640076, "grad_norm": 0.5018460051829553, "learning_rate": 9.529374310370411e-05, "loss": 2.9864, "step": 10888 }, { "epoch": 0.6759575392637656, "grad_norm": 0.5127901988553101, "learning_rate": 9.529221334097676e-05, "loss": 3.078, "step": 10889 }, { "epoch": 0.6760196163635235, "grad_norm": 0.4997585428871223, "learning_rate": 9.529068334194892e-05, "loss": 3.0978, "step": 10890 }, { "epoch": 0.6760816934632814, "grad_norm": 0.5234838721387886, "learning_rate": 9.52891531066286e-05, "loss": 3.0062, "step": 10891 }, { "epoch": 0.6761437705630393, "grad_norm": 0.4504921156122383, "learning_rate": 9.528762263502378e-05, "loss": 3.0556, "step": 10892 }, { "epoch": 0.6762058476627972, "grad_norm": 0.47457659572856925, "learning_rate": 9.528609192714243e-05, "loss": 3.0662, "step": 10893 }, { "epoch": 0.676267924762555, "grad_norm": 0.47727551823270636, "learning_rate": 9.528456098299254e-05, "loss": 3.2099, "step": 10894 }, { "epoch": 0.676330001862313, "grad_norm": 0.45135736987117925, "learning_rate": 9.528302980258213e-05, "loss": 3.1176, "step": 10895 }, { "epoch": 0.6763920789620709, "grad_norm": 0.4778364226924729, "learning_rate": 9.528149838591913e-05, "loss": 3.0634, "step": 10896 }, { "epoch": 0.6764541560618288, "grad_norm": 0.5904378867076917, "learning_rate": 9.527996673301156e-05, "loss": 3.1362, "step": 10897 }, { "epoch": 0.6765162331615867, "grad_norm": 0.6485878184172983, "learning_rate": 9.527843484386742e-05, "loss": 3.1896, "step": 10898 }, { "epoch": 0.6765783102613446, "grad_norm": 0.5329098016934494, "learning_rate": 9.527690271849471e-05, "loss": 3.0503, "step": 10899 }, { "epoch": 0.6766403873611024, "grad_norm": 0.5367799893331262, "learning_rate": 9.52753703569014e-05, "loss": 3.0132, "step": 10900 }, { "epoch": 0.6767024644608604, "grad_norm": 0.5626479854893112, "learning_rate": 9.527383775909547e-05, "loss": 2.999, "step": 10901 }, { "epoch": 0.6767645415606183, "grad_norm": 0.5033779183855659, "learning_rate": 9.527230492508497e-05, "loss": 3.0671, "step": 10902 }, { "epoch": 0.6768266186603762, "grad_norm": 0.45314415529474816, "learning_rate": 9.527077185487784e-05, "loss": 3.0958, "step": 10903 }, { "epoch": 0.6768886957601341, "grad_norm": 0.5313143130250069, "learning_rate": 9.526923854848211e-05, "loss": 3.0441, "step": 10904 }, { "epoch": 0.676950772859892, "grad_norm": 0.4973253596312649, "learning_rate": 9.526770500590577e-05, "loss": 3.1331, "step": 10905 }, { "epoch": 0.6770128499596498, "grad_norm": 0.48924642157476445, "learning_rate": 9.526617122715684e-05, "loss": 2.9471, "step": 10906 }, { "epoch": 0.6770749270594077, "grad_norm": 0.5389030054735224, "learning_rate": 9.526463721224329e-05, "loss": 3.1233, "step": 10907 }, { "epoch": 0.6771370041591657, "grad_norm": 0.46437180226638175, "learning_rate": 9.526310296117313e-05, "loss": 3.1553, "step": 10908 }, { "epoch": 0.6771990812589236, "grad_norm": 0.545541691095141, "learning_rate": 9.52615684739544e-05, "loss": 3.2017, "step": 10909 }, { "epoch": 0.6772611583586815, "grad_norm": 0.5150476170720164, "learning_rate": 9.526003375059507e-05, "loss": 3.0838, "step": 10910 }, { "epoch": 0.6773232354584394, "grad_norm": 0.5138717958540184, "learning_rate": 9.525849879110315e-05, "loss": 3.0559, "step": 10911 }, { "epoch": 0.6773853125581972, "grad_norm": 0.4876233360577014, "learning_rate": 9.525696359548666e-05, "loss": 3.0832, "step": 10912 }, { "epoch": 0.6774473896579551, "grad_norm": 0.5156025159152954, "learning_rate": 9.525542816375359e-05, "loss": 3.0745, "step": 10913 }, { "epoch": 0.6775094667577131, "grad_norm": 0.5607554881592214, "learning_rate": 9.525389249591199e-05, "loss": 3.0879, "step": 10914 }, { "epoch": 0.677571543857471, "grad_norm": 0.5547644602080154, "learning_rate": 9.525235659196982e-05, "loss": 3.0788, "step": 10915 }, { "epoch": 0.6776336209572289, "grad_norm": 0.528620016159771, "learning_rate": 9.525082045193513e-05, "loss": 3.1684, "step": 10916 }, { "epoch": 0.6776956980569868, "grad_norm": 0.49496486209171175, "learning_rate": 9.524928407581595e-05, "loss": 3.1824, "step": 10917 }, { "epoch": 0.6777577751567446, "grad_norm": 0.5693462495213476, "learning_rate": 9.524774746362025e-05, "loss": 3.062, "step": 10918 }, { "epoch": 0.6778198522565025, "grad_norm": 0.5018408386271441, "learning_rate": 9.524621061535606e-05, "loss": 3.0912, "step": 10919 }, { "epoch": 0.6778819293562605, "grad_norm": 0.43960573725384183, "learning_rate": 9.524467353103142e-05, "loss": 3.0399, "step": 10920 }, { "epoch": 0.6779440064560184, "grad_norm": 0.5464730829431523, "learning_rate": 9.524313621065432e-05, "loss": 3.1458, "step": 10921 }, { "epoch": 0.6780060835557763, "grad_norm": 0.5603590594136219, "learning_rate": 9.524159865423281e-05, "loss": 3.0175, "step": 10922 }, { "epoch": 0.6780681606555342, "grad_norm": 0.4476976432103897, "learning_rate": 9.524006086177488e-05, "loss": 3.1142, "step": 10923 }, { "epoch": 0.678130237755292, "grad_norm": 0.49967894081004227, "learning_rate": 9.523852283328858e-05, "loss": 3.1123, "step": 10924 }, { "epoch": 0.6781923148550499, "grad_norm": 0.47577780861698304, "learning_rate": 9.523698456878194e-05, "loss": 3.0524, "step": 10925 }, { "epoch": 0.6782543919548079, "grad_norm": 0.48363154792050317, "learning_rate": 9.523544606826295e-05, "loss": 3.1315, "step": 10926 }, { "epoch": 0.6783164690545658, "grad_norm": 0.484189413264601, "learning_rate": 9.523390733173966e-05, "loss": 3.047, "step": 10927 }, { "epoch": 0.6783785461543237, "grad_norm": 0.5425782205734004, "learning_rate": 9.523236835922009e-05, "loss": 3.024, "step": 10928 }, { "epoch": 0.6784406232540816, "grad_norm": 0.48137576082496875, "learning_rate": 9.523082915071229e-05, "loss": 3.1309, "step": 10929 }, { "epoch": 0.6785027003538394, "grad_norm": 0.5141830511692413, "learning_rate": 9.522928970622425e-05, "loss": 3.1574, "step": 10930 }, { "epoch": 0.6785647774535973, "grad_norm": 0.4985160263530991, "learning_rate": 9.522775002576405e-05, "loss": 3.1674, "step": 10931 }, { "epoch": 0.6786268545533553, "grad_norm": 0.48404186852497916, "learning_rate": 9.522621010933969e-05, "loss": 3.1142, "step": 10932 }, { "epoch": 0.6786889316531132, "grad_norm": 0.4938157261186747, "learning_rate": 9.52246699569592e-05, "loss": 3.1244, "step": 10933 }, { "epoch": 0.6787510087528711, "grad_norm": 0.5264117163682587, "learning_rate": 9.522312956863064e-05, "loss": 3.1965, "step": 10934 }, { "epoch": 0.678813085852629, "grad_norm": 0.4691630292207139, "learning_rate": 9.522158894436204e-05, "loss": 3.0907, "step": 10935 }, { "epoch": 0.6788751629523868, "grad_norm": 0.4847347400303503, "learning_rate": 9.522004808416142e-05, "loss": 3.0558, "step": 10936 }, { "epoch": 0.6789372400521447, "grad_norm": 0.4689804305495827, "learning_rate": 9.521850698803685e-05, "loss": 3.0284, "step": 10937 }, { "epoch": 0.6789993171519026, "grad_norm": 0.538668806635781, "learning_rate": 9.521696565599633e-05, "loss": 3.0709, "step": 10938 }, { "epoch": 0.6790613942516606, "grad_norm": 0.4768907948349265, "learning_rate": 9.521542408804794e-05, "loss": 3.0478, "step": 10939 }, { "epoch": 0.6791234713514185, "grad_norm": 0.4786846096062624, "learning_rate": 9.521388228419969e-05, "loss": 3.1521, "step": 10940 }, { "epoch": 0.6791855484511764, "grad_norm": 0.457591031898596, "learning_rate": 9.521234024445965e-05, "loss": 3.0398, "step": 10941 }, { "epoch": 0.6792476255509342, "grad_norm": 0.5189801498394994, "learning_rate": 9.521079796883586e-05, "loss": 3.1379, "step": 10942 }, { "epoch": 0.6793097026506921, "grad_norm": 0.48345368965426394, "learning_rate": 9.520925545733636e-05, "loss": 3.0363, "step": 10943 }, { "epoch": 0.67937177975045, "grad_norm": 0.4524683304763726, "learning_rate": 9.520771270996919e-05, "loss": 3.0713, "step": 10944 }, { "epoch": 0.679433856850208, "grad_norm": 0.4938571403245994, "learning_rate": 9.520616972674241e-05, "loss": 3.1415, "step": 10945 }, { "epoch": 0.6794959339499659, "grad_norm": 0.47926946933184394, "learning_rate": 9.520462650766407e-05, "loss": 3.1065, "step": 10946 }, { "epoch": 0.6795580110497238, "grad_norm": 0.48265775713032316, "learning_rate": 9.520308305274223e-05, "loss": 3.1739, "step": 10947 }, { "epoch": 0.6796200881494816, "grad_norm": 0.5567241940592172, "learning_rate": 9.520153936198492e-05, "loss": 3.2478, "step": 10948 }, { "epoch": 0.6796821652492395, "grad_norm": 0.5236447336845419, "learning_rate": 9.519999543540021e-05, "loss": 3.0429, "step": 10949 }, { "epoch": 0.6797442423489974, "grad_norm": 0.5310701739068905, "learning_rate": 9.519845127299616e-05, "loss": 3.2268, "step": 10950 }, { "epoch": 0.6798063194487554, "grad_norm": 0.5137327324965628, "learning_rate": 9.51969068747808e-05, "loss": 3.2126, "step": 10951 }, { "epoch": 0.6798683965485133, "grad_norm": 0.5428330231225393, "learning_rate": 9.519536224076222e-05, "loss": 3.0961, "step": 10952 }, { "epoch": 0.6799304736482712, "grad_norm": 0.5151131719477702, "learning_rate": 9.519381737094845e-05, "loss": 3.0546, "step": 10953 }, { "epoch": 0.679992550748029, "grad_norm": 0.5166333194194798, "learning_rate": 9.519227226534758e-05, "loss": 3.1878, "step": 10954 }, { "epoch": 0.6800546278477869, "grad_norm": 0.5128293688671185, "learning_rate": 9.519072692396764e-05, "loss": 3.1211, "step": 10955 }, { "epoch": 0.6801167049475448, "grad_norm": 0.5529853601432092, "learning_rate": 9.518918134681671e-05, "loss": 3.1663, "step": 10956 }, { "epoch": 0.6801787820473028, "grad_norm": 0.5307211206766521, "learning_rate": 9.518763553390286e-05, "loss": 3.0729, "step": 10957 }, { "epoch": 0.6802408591470607, "grad_norm": 0.5088228591559781, "learning_rate": 9.518608948523413e-05, "loss": 3.1036, "step": 10958 }, { "epoch": 0.6803029362468186, "grad_norm": 0.4878626135060796, "learning_rate": 9.518454320081862e-05, "loss": 3.0337, "step": 10959 }, { "epoch": 0.6803650133465764, "grad_norm": 0.5482422222902554, "learning_rate": 9.518299668066438e-05, "loss": 3.0366, "step": 10960 }, { "epoch": 0.6804270904463343, "grad_norm": 0.5348735536010613, "learning_rate": 9.518144992477945e-05, "loss": 3.1286, "step": 10961 }, { "epoch": 0.6804891675460922, "grad_norm": 0.48146608009967307, "learning_rate": 9.517990293317193e-05, "loss": 3.1458, "step": 10962 }, { "epoch": 0.6805512446458502, "grad_norm": 0.5515088355133921, "learning_rate": 9.517835570584992e-05, "loss": 3.1295, "step": 10963 }, { "epoch": 0.6806133217456081, "grad_norm": 0.566009171832822, "learning_rate": 9.517680824282145e-05, "loss": 3.0672, "step": 10964 }, { "epoch": 0.680675398845366, "grad_norm": 0.5029237320926978, "learning_rate": 9.51752605440946e-05, "loss": 3.1092, "step": 10965 }, { "epoch": 0.6807374759451238, "grad_norm": 0.5227293376825279, "learning_rate": 9.517371260967745e-05, "loss": 3.1071, "step": 10966 }, { "epoch": 0.6807995530448817, "grad_norm": 0.5310129590491613, "learning_rate": 9.517216443957807e-05, "loss": 3.1468, "step": 10967 }, { "epoch": 0.6808616301446396, "grad_norm": 0.5529183664513768, "learning_rate": 9.517061603380455e-05, "loss": 3.1162, "step": 10968 }, { "epoch": 0.6809237072443975, "grad_norm": 0.5391232829448015, "learning_rate": 9.516906739236495e-05, "loss": 3.0852, "step": 10969 }, { "epoch": 0.6809857843441555, "grad_norm": 0.610232239312774, "learning_rate": 9.516751851526738e-05, "loss": 3.1811, "step": 10970 }, { "epoch": 0.6810478614439134, "grad_norm": 0.6288927207107577, "learning_rate": 9.516596940251987e-05, "loss": 3.0747, "step": 10971 }, { "epoch": 0.6811099385436712, "grad_norm": 0.5872389010293609, "learning_rate": 9.516442005413056e-05, "loss": 3.1193, "step": 10972 }, { "epoch": 0.6811720156434291, "grad_norm": 0.5883342551862647, "learning_rate": 9.516287047010749e-05, "loss": 3.0507, "step": 10973 }, { "epoch": 0.681234092743187, "grad_norm": 0.635541986916753, "learning_rate": 9.516132065045877e-05, "loss": 3.051, "step": 10974 }, { "epoch": 0.6812961698429449, "grad_norm": 0.5915551274117791, "learning_rate": 9.515977059519248e-05, "loss": 3.0556, "step": 10975 }, { "epoch": 0.6813582469427029, "grad_norm": 0.5787622960063563, "learning_rate": 9.515822030431669e-05, "loss": 3.0301, "step": 10976 }, { "epoch": 0.6814203240424608, "grad_norm": 0.4911734109729835, "learning_rate": 9.515666977783952e-05, "loss": 3.0898, "step": 10977 }, { "epoch": 0.6814824011422186, "grad_norm": 0.6018996346761112, "learning_rate": 9.515511901576903e-05, "loss": 2.9931, "step": 10978 }, { "epoch": 0.6815444782419765, "grad_norm": 0.574462890625, "learning_rate": 9.515356801811333e-05, "loss": 3.1813, "step": 10979 }, { "epoch": 0.6816065553417344, "grad_norm": 0.5391392584068802, "learning_rate": 9.515201678488049e-05, "loss": 3.1001, "step": 10980 }, { "epoch": 0.6816686324414923, "grad_norm": 0.5948859441255079, "learning_rate": 9.515046531607863e-05, "loss": 3.0403, "step": 10981 }, { "epoch": 0.6817307095412503, "grad_norm": 0.5498547438064191, "learning_rate": 9.514891361171583e-05, "loss": 3.0892, "step": 10982 }, { "epoch": 0.6817927866410082, "grad_norm": 0.5534010462330206, "learning_rate": 9.514736167180018e-05, "loss": 3.1182, "step": 10983 }, { "epoch": 0.681854863740766, "grad_norm": 0.5770397309817056, "learning_rate": 9.51458094963398e-05, "loss": 3.1455, "step": 10984 }, { "epoch": 0.6819169408405239, "grad_norm": 0.4912618379404159, "learning_rate": 9.514425708534276e-05, "loss": 3.0898, "step": 10985 }, { "epoch": 0.6819790179402818, "grad_norm": 0.5028675818753114, "learning_rate": 9.514270443881718e-05, "loss": 3.0402, "step": 10986 }, { "epoch": 0.6820410950400397, "grad_norm": 0.47704148844762445, "learning_rate": 9.514115155677115e-05, "loss": 3.1445, "step": 10987 }, { "epoch": 0.6821031721397977, "grad_norm": 0.5061348124726087, "learning_rate": 9.513959843921276e-05, "loss": 3.1652, "step": 10988 }, { "epoch": 0.6821652492395556, "grad_norm": 0.5543769719412395, "learning_rate": 9.513804508615014e-05, "loss": 3.0206, "step": 10989 }, { "epoch": 0.6822273263393134, "grad_norm": 0.5850337624792139, "learning_rate": 9.513649149759138e-05, "loss": 3.0557, "step": 10990 }, { "epoch": 0.6822894034390713, "grad_norm": 0.5465589154814602, "learning_rate": 9.513493767354458e-05, "loss": 3.1001, "step": 10991 }, { "epoch": 0.6823514805388292, "grad_norm": 0.44099135015291496, "learning_rate": 9.513338361401787e-05, "loss": 3.1026, "step": 10992 }, { "epoch": 0.6824135576385871, "grad_norm": 0.5027843672787157, "learning_rate": 9.513182931901932e-05, "loss": 3.0733, "step": 10993 }, { "epoch": 0.682475634738345, "grad_norm": 0.4929895677720142, "learning_rate": 9.513027478855707e-05, "loss": 3.0676, "step": 10994 }, { "epoch": 0.682537711838103, "grad_norm": 0.5453445818626228, "learning_rate": 9.512872002263922e-05, "loss": 3.2235, "step": 10995 }, { "epoch": 0.6825997889378608, "grad_norm": 0.4791200829404637, "learning_rate": 9.512716502127388e-05, "loss": 3.0912, "step": 10996 }, { "epoch": 0.6826618660376187, "grad_norm": 0.4812417364030277, "learning_rate": 9.512560978446917e-05, "loss": 3.0968, "step": 10997 }, { "epoch": 0.6827239431373766, "grad_norm": 0.5211722193347844, "learning_rate": 9.51240543122332e-05, "loss": 3.1869, "step": 10998 }, { "epoch": 0.6827860202371345, "grad_norm": 0.5431768169640788, "learning_rate": 9.512249860457407e-05, "loss": 3.0995, "step": 10999 }, { "epoch": 0.6828480973368924, "grad_norm": 0.4722783493208355, "learning_rate": 9.512094266149994e-05, "loss": 3.0502, "step": 11000 }, { "epoch": 0.6829101744366504, "grad_norm": 0.45810168184834776, "learning_rate": 9.511938648301888e-05, "loss": 3.1764, "step": 11001 }, { "epoch": 0.6829722515364082, "grad_norm": 0.5784052607673003, "learning_rate": 9.511783006913903e-05, "loss": 3.2097, "step": 11002 }, { "epoch": 0.6830343286361661, "grad_norm": 0.5338871519241544, "learning_rate": 9.51162734198685e-05, "loss": 2.9922, "step": 11003 }, { "epoch": 0.683096405735924, "grad_norm": 0.5105277900660284, "learning_rate": 9.511471653521542e-05, "loss": 2.9949, "step": 11004 }, { "epoch": 0.6831584828356819, "grad_norm": 0.5141642136517852, "learning_rate": 9.511315941518793e-05, "loss": 3.1369, "step": 11005 }, { "epoch": 0.6832205599354398, "grad_norm": 0.6205546360522275, "learning_rate": 9.511160205979411e-05, "loss": 3.1027, "step": 11006 }, { "epoch": 0.6832826370351978, "grad_norm": 0.5748431126476176, "learning_rate": 9.511004446904213e-05, "loss": 3.1885, "step": 11007 }, { "epoch": 0.6833447141349556, "grad_norm": 0.46844881236214586, "learning_rate": 9.51084866429401e-05, "loss": 3.1241, "step": 11008 }, { "epoch": 0.6834067912347135, "grad_norm": 0.546906361361859, "learning_rate": 9.510692858149613e-05, "loss": 3.1177, "step": 11009 }, { "epoch": 0.6834688683344714, "grad_norm": 0.5818744422590326, "learning_rate": 9.510537028471837e-05, "loss": 3.049, "step": 11010 }, { "epoch": 0.6835309454342293, "grad_norm": 0.5522632005978015, "learning_rate": 9.510381175261493e-05, "loss": 3.1459, "step": 11011 }, { "epoch": 0.6835930225339872, "grad_norm": 0.47755411822528654, "learning_rate": 9.510225298519398e-05, "loss": 3.0867, "step": 11012 }, { "epoch": 0.6836550996337452, "grad_norm": 0.5844933323720529, "learning_rate": 9.510069398246361e-05, "loss": 3.1735, "step": 11013 }, { "epoch": 0.683717176733503, "grad_norm": 0.644873025616446, "learning_rate": 9.509913474443197e-05, "loss": 3.0921, "step": 11014 }, { "epoch": 0.6837792538332609, "grad_norm": 0.570605398762738, "learning_rate": 9.509757527110721e-05, "loss": 3.1336, "step": 11015 }, { "epoch": 0.6838413309330188, "grad_norm": 0.5672401866139642, "learning_rate": 9.509601556249744e-05, "loss": 3.1024, "step": 11016 }, { "epoch": 0.6839034080327767, "grad_norm": 0.504381177786392, "learning_rate": 9.509445561861081e-05, "loss": 3.0474, "step": 11017 }, { "epoch": 0.6839654851325346, "grad_norm": 0.5605738031917334, "learning_rate": 9.509289543945547e-05, "loss": 3.1497, "step": 11018 }, { "epoch": 0.6840275622322926, "grad_norm": 0.5411310726890388, "learning_rate": 9.509133502503953e-05, "loss": 3.1579, "step": 11019 }, { "epoch": 0.6840896393320504, "grad_norm": 0.5625835727490922, "learning_rate": 9.508977437537117e-05, "loss": 3.1416, "step": 11020 }, { "epoch": 0.6841517164318083, "grad_norm": 0.7023197437295404, "learning_rate": 9.508821349045849e-05, "loss": 3.1688, "step": 11021 }, { "epoch": 0.6842137935315662, "grad_norm": 0.6480750944076509, "learning_rate": 9.508665237030967e-05, "loss": 3.0987, "step": 11022 }, { "epoch": 0.6842758706313241, "grad_norm": 0.5277902478981278, "learning_rate": 9.508509101493283e-05, "loss": 3.1489, "step": 11023 }, { "epoch": 0.684337947731082, "grad_norm": 0.599564541311695, "learning_rate": 9.508352942433613e-05, "loss": 3.1531, "step": 11024 }, { "epoch": 0.68440002483084, "grad_norm": 0.5660629284286267, "learning_rate": 9.508196759852772e-05, "loss": 3.0684, "step": 11025 }, { "epoch": 0.6844621019305978, "grad_norm": 0.5863276644913951, "learning_rate": 9.508040553751571e-05, "loss": 3.1015, "step": 11026 }, { "epoch": 0.6845241790303557, "grad_norm": 0.49709510366675536, "learning_rate": 9.507884324130831e-05, "loss": 3.067, "step": 11027 }, { "epoch": 0.6845862561301136, "grad_norm": 0.6112626744537938, "learning_rate": 9.507728070991364e-05, "loss": 3.1632, "step": 11028 }, { "epoch": 0.6846483332298715, "grad_norm": 0.46817281473737243, "learning_rate": 9.507571794333985e-05, "loss": 3.0853, "step": 11029 }, { "epoch": 0.6847104103296294, "grad_norm": 0.5218980841185413, "learning_rate": 9.507415494159508e-05, "loss": 3.0927, "step": 11030 }, { "epoch": 0.6847724874293873, "grad_norm": 0.5229271065732417, "learning_rate": 9.507259170468751e-05, "loss": 3.1323, "step": 11031 }, { "epoch": 0.6848345645291452, "grad_norm": 0.4586592963717197, "learning_rate": 9.50710282326253e-05, "loss": 3.0749, "step": 11032 }, { "epoch": 0.6848966416289031, "grad_norm": 0.5066495049021226, "learning_rate": 9.50694645254166e-05, "loss": 3.0556, "step": 11033 }, { "epoch": 0.684958718728661, "grad_norm": 0.49293485546116095, "learning_rate": 9.506790058306955e-05, "loss": 3.1361, "step": 11034 }, { "epoch": 0.6850207958284189, "grad_norm": 0.4741071390598924, "learning_rate": 9.506633640559232e-05, "loss": 3.1152, "step": 11035 }, { "epoch": 0.6850828729281768, "grad_norm": 0.48161941073214215, "learning_rate": 9.506477199299307e-05, "loss": 3.0679, "step": 11036 }, { "epoch": 0.6851449500279347, "grad_norm": 0.5059186098037037, "learning_rate": 9.506320734527998e-05, "loss": 3.0907, "step": 11037 }, { "epoch": 0.6852070271276925, "grad_norm": 0.5012893146688748, "learning_rate": 9.506164246246119e-05, "loss": 3.1271, "step": 11038 }, { "epoch": 0.6852691042274505, "grad_norm": 0.5149082056661154, "learning_rate": 9.506007734454486e-05, "loss": 3.0946, "step": 11039 }, { "epoch": 0.6853311813272084, "grad_norm": 0.5168131519386866, "learning_rate": 9.505851199153918e-05, "loss": 3.1163, "step": 11040 }, { "epoch": 0.6853932584269663, "grad_norm": 0.47482197713187463, "learning_rate": 9.50569464034523e-05, "loss": 3.115, "step": 11041 }, { "epoch": 0.6854553355267242, "grad_norm": 0.479657535051959, "learning_rate": 9.50553805802924e-05, "loss": 3.0625, "step": 11042 }, { "epoch": 0.6855174126264821, "grad_norm": 0.4973618077994646, "learning_rate": 9.505381452206763e-05, "loss": 3.0844, "step": 11043 }, { "epoch": 0.6855794897262399, "grad_norm": 0.49351060230379845, "learning_rate": 9.505224822878618e-05, "loss": 3.0813, "step": 11044 }, { "epoch": 0.6856415668259979, "grad_norm": 0.5269670624582945, "learning_rate": 9.505068170045621e-05, "loss": 3.0109, "step": 11045 }, { "epoch": 0.6857036439257558, "grad_norm": 0.4293236665982557, "learning_rate": 9.504911493708591e-05, "loss": 3.1377, "step": 11046 }, { "epoch": 0.6857657210255137, "grad_norm": 0.46486484255339594, "learning_rate": 9.504754793868343e-05, "loss": 3.1683, "step": 11047 }, { "epoch": 0.6858277981252716, "grad_norm": 0.43601017417979654, "learning_rate": 9.504598070525696e-05, "loss": 3.2052, "step": 11048 }, { "epoch": 0.6858898752250295, "grad_norm": 0.4944460682479159, "learning_rate": 9.504441323681467e-05, "loss": 3.016, "step": 11049 }, { "epoch": 0.6859519523247873, "grad_norm": 0.48708122802270687, "learning_rate": 9.504284553336475e-05, "loss": 3.0912, "step": 11050 }, { "epoch": 0.6860140294245453, "grad_norm": 0.48734381142163663, "learning_rate": 9.504127759491536e-05, "loss": 3.1265, "step": 11051 }, { "epoch": 0.6860761065243032, "grad_norm": 0.5092654935458406, "learning_rate": 9.503970942147469e-05, "loss": 3.0737, "step": 11052 }, { "epoch": 0.6861381836240611, "grad_norm": 0.5019883614725941, "learning_rate": 9.503814101305093e-05, "loss": 3.0652, "step": 11053 }, { "epoch": 0.686200260723819, "grad_norm": 0.5284980009739512, "learning_rate": 9.503657236965225e-05, "loss": 3.0915, "step": 11054 }, { "epoch": 0.6862623378235769, "grad_norm": 0.44912597693409506, "learning_rate": 9.503500349128685e-05, "loss": 3.1054, "step": 11055 }, { "epoch": 0.6863244149233347, "grad_norm": 0.5109190479644713, "learning_rate": 9.503343437796288e-05, "loss": 3.0162, "step": 11056 }, { "epoch": 0.6863864920230927, "grad_norm": 0.5542520775143063, "learning_rate": 9.503186502968858e-05, "loss": 3.0749, "step": 11057 }, { "epoch": 0.6864485691228506, "grad_norm": 0.5134925964568351, "learning_rate": 9.503029544647209e-05, "loss": 3.077, "step": 11058 }, { "epoch": 0.6865106462226085, "grad_norm": 0.4956125944780995, "learning_rate": 9.502872562832162e-05, "loss": 3.2054, "step": 11059 }, { "epoch": 0.6865727233223664, "grad_norm": 0.5509619179787201, "learning_rate": 9.502715557524535e-05, "loss": 3.0466, "step": 11060 }, { "epoch": 0.6866348004221243, "grad_norm": 0.5225584990776317, "learning_rate": 9.502558528725149e-05, "loss": 3.032, "step": 11061 }, { "epoch": 0.6866968775218821, "grad_norm": 0.48900562723506547, "learning_rate": 9.502401476434821e-05, "loss": 3.088, "step": 11062 }, { "epoch": 0.68675895462164, "grad_norm": 0.4974504377607236, "learning_rate": 9.502244400654372e-05, "loss": 3.1754, "step": 11063 }, { "epoch": 0.686821031721398, "grad_norm": 0.5200991001056822, "learning_rate": 9.502087301384623e-05, "loss": 3.0327, "step": 11064 }, { "epoch": 0.6868831088211559, "grad_norm": 0.5276683514771561, "learning_rate": 9.501930178626389e-05, "loss": 3.1409, "step": 11065 }, { "epoch": 0.6869451859209138, "grad_norm": 0.4461383857375597, "learning_rate": 9.501773032380493e-05, "loss": 3.0542, "step": 11066 }, { "epoch": 0.6870072630206717, "grad_norm": 0.5019648212895864, "learning_rate": 9.501615862647755e-05, "loss": 3.1178, "step": 11067 }, { "epoch": 0.6870693401204295, "grad_norm": 0.4872491074784366, "learning_rate": 9.501458669428993e-05, "loss": 3.103, "step": 11068 }, { "epoch": 0.6871314172201874, "grad_norm": 0.43921614297034073, "learning_rate": 9.50130145272503e-05, "loss": 3.0926, "step": 11069 }, { "epoch": 0.6871934943199454, "grad_norm": 0.46327939990029643, "learning_rate": 9.501144212536684e-05, "loss": 3.091, "step": 11070 }, { "epoch": 0.6872555714197033, "grad_norm": 0.45149087568845236, "learning_rate": 9.500986948864775e-05, "loss": 3.0851, "step": 11071 }, { "epoch": 0.6873176485194612, "grad_norm": 0.4590371608817123, "learning_rate": 9.500829661710124e-05, "loss": 3.0242, "step": 11072 }, { "epoch": 0.6873797256192191, "grad_norm": 0.45218676410818165, "learning_rate": 9.500672351073554e-05, "loss": 3.063, "step": 11073 }, { "epoch": 0.6874418027189769, "grad_norm": 0.45735733919119675, "learning_rate": 9.500515016955884e-05, "loss": 3.1008, "step": 11074 }, { "epoch": 0.6875038798187348, "grad_norm": 0.47791129116980746, "learning_rate": 9.500357659357932e-05, "loss": 3.031, "step": 11075 }, { "epoch": 0.6875659569184928, "grad_norm": 0.5171133869924373, "learning_rate": 9.500200278280525e-05, "loss": 3.1229, "step": 11076 }, { "epoch": 0.6876280340182507, "grad_norm": 0.473333535566533, "learning_rate": 9.500042873724477e-05, "loss": 3.0671, "step": 11077 }, { "epoch": 0.6876901111180086, "grad_norm": 0.44124862084429856, "learning_rate": 9.499885445690615e-05, "loss": 3.1274, "step": 11078 }, { "epoch": 0.6877521882177665, "grad_norm": 0.5009077891238627, "learning_rate": 9.499727994179759e-05, "loss": 3.1477, "step": 11079 }, { "epoch": 0.6878142653175243, "grad_norm": 0.5431764054636326, "learning_rate": 9.499570519192728e-05, "loss": 3.1048, "step": 11080 }, { "epoch": 0.6878763424172822, "grad_norm": 0.4737834258353203, "learning_rate": 9.499413020730345e-05, "loss": 3.0993, "step": 11081 }, { "epoch": 0.6879384195170402, "grad_norm": 0.5127822656918398, "learning_rate": 9.499255498793434e-05, "loss": 2.9855, "step": 11082 }, { "epoch": 0.6880004966167981, "grad_norm": 0.5658984835070324, "learning_rate": 9.499097953382814e-05, "loss": 3.1344, "step": 11083 }, { "epoch": 0.688062573716556, "grad_norm": 0.5118493357595293, "learning_rate": 9.498940384499307e-05, "loss": 3.0469, "step": 11084 }, { "epoch": 0.6881246508163139, "grad_norm": 0.5545042836803773, "learning_rate": 9.498782792143737e-05, "loss": 3.041, "step": 11085 }, { "epoch": 0.6881867279160717, "grad_norm": 0.5685871362681514, "learning_rate": 9.498625176316925e-05, "loss": 3.1335, "step": 11086 }, { "epoch": 0.6882488050158296, "grad_norm": 0.5819171051132267, "learning_rate": 9.498467537019691e-05, "loss": 3.0595, "step": 11087 }, { "epoch": 0.6883108821155876, "grad_norm": 0.4566057621004551, "learning_rate": 9.498309874252862e-05, "loss": 3.091, "step": 11088 }, { "epoch": 0.6883729592153455, "grad_norm": 0.7137280288093805, "learning_rate": 9.498152188017259e-05, "loss": 3.1156, "step": 11089 }, { "epoch": 0.6884350363151034, "grad_norm": 0.608993410798304, "learning_rate": 9.497994478313703e-05, "loss": 3.0493, "step": 11090 }, { "epoch": 0.6884971134148613, "grad_norm": 0.47582348603193286, "learning_rate": 9.497836745143019e-05, "loss": 3.1226, "step": 11091 }, { "epoch": 0.6885591905146191, "grad_norm": 0.5807241830198563, "learning_rate": 9.497678988506027e-05, "loss": 3.1266, "step": 11092 }, { "epoch": 0.688621267614377, "grad_norm": 0.559611081007983, "learning_rate": 9.497521208403552e-05, "loss": 3.0528, "step": 11093 }, { "epoch": 0.688683344714135, "grad_norm": 0.5029778614860201, "learning_rate": 9.497363404836417e-05, "loss": 3.0565, "step": 11094 }, { "epoch": 0.6887454218138929, "grad_norm": 0.5736338432596289, "learning_rate": 9.497205577805448e-05, "loss": 3.2009, "step": 11095 }, { "epoch": 0.6888074989136508, "grad_norm": 0.5518287545567897, "learning_rate": 9.497047727311463e-05, "loss": 3.1282, "step": 11096 }, { "epoch": 0.6888695760134087, "grad_norm": 0.5486565318264677, "learning_rate": 9.496889853355289e-05, "loss": 3.1425, "step": 11097 }, { "epoch": 0.6889316531131665, "grad_norm": 0.5286077820812329, "learning_rate": 9.49673195593775e-05, "loss": 3.0725, "step": 11098 }, { "epoch": 0.6889937302129244, "grad_norm": 0.4700727396588869, "learning_rate": 9.496574035059666e-05, "loss": 3.1694, "step": 11099 }, { "epoch": 0.6890558073126823, "grad_norm": 0.5309014860156472, "learning_rate": 9.496416090721867e-05, "loss": 3.0753, "step": 11100 }, { "epoch": 0.6891178844124403, "grad_norm": 0.5922208469634145, "learning_rate": 9.496258122925173e-05, "loss": 3.0748, "step": 11101 }, { "epoch": 0.6891799615121982, "grad_norm": 0.5246063959487328, "learning_rate": 9.496100131670407e-05, "loss": 3.124, "step": 11102 }, { "epoch": 0.6892420386119561, "grad_norm": 0.5453350455971133, "learning_rate": 9.495942116958396e-05, "loss": 3.1359, "step": 11103 }, { "epoch": 0.6893041157117139, "grad_norm": 0.48279350268077753, "learning_rate": 9.495784078789965e-05, "loss": 3.0929, "step": 11104 }, { "epoch": 0.6893661928114718, "grad_norm": 0.45277610040164146, "learning_rate": 9.495626017165935e-05, "loss": 3.1722, "step": 11105 }, { "epoch": 0.6894282699112297, "grad_norm": 0.532991276390514, "learning_rate": 9.495467932087134e-05, "loss": 3.0828, "step": 11106 }, { "epoch": 0.6894903470109877, "grad_norm": 0.4531132269020462, "learning_rate": 9.495309823554386e-05, "loss": 3.0477, "step": 11107 }, { "epoch": 0.6895524241107456, "grad_norm": 0.5054918522692243, "learning_rate": 9.495151691568513e-05, "loss": 3.1924, "step": 11108 }, { "epoch": 0.6896145012105035, "grad_norm": 0.5497480671594078, "learning_rate": 9.494993536130345e-05, "loss": 3.0802, "step": 11109 }, { "epoch": 0.6896765783102613, "grad_norm": 0.5463309988292386, "learning_rate": 9.494835357240703e-05, "loss": 3.0522, "step": 11110 }, { "epoch": 0.6897386554100192, "grad_norm": 0.5197020801969394, "learning_rate": 9.494677154900414e-05, "loss": 3.091, "step": 11111 }, { "epoch": 0.6898007325097771, "grad_norm": 0.43610326015117173, "learning_rate": 9.494518929110304e-05, "loss": 3.0714, "step": 11112 }, { "epoch": 0.689862809609535, "grad_norm": 0.4998312903923539, "learning_rate": 9.494360679871197e-05, "loss": 3.0572, "step": 11113 }, { "epoch": 0.689924886709293, "grad_norm": 0.5017596513915675, "learning_rate": 9.494202407183918e-05, "loss": 3.0749, "step": 11114 }, { "epoch": 0.6899869638090509, "grad_norm": 0.48852828833746736, "learning_rate": 9.494044111049296e-05, "loss": 3.1536, "step": 11115 }, { "epoch": 0.6900490409088087, "grad_norm": 0.5253805563717057, "learning_rate": 9.493885791468155e-05, "loss": 3.1435, "step": 11116 }, { "epoch": 0.6901111180085666, "grad_norm": 0.5738736886981863, "learning_rate": 9.49372744844132e-05, "loss": 3.0601, "step": 11117 }, { "epoch": 0.6901731951083245, "grad_norm": 0.5717060621281493, "learning_rate": 9.493569081969618e-05, "loss": 3.1108, "step": 11118 }, { "epoch": 0.6902352722080825, "grad_norm": 0.5516096338119615, "learning_rate": 9.493410692053874e-05, "loss": 3.1709, "step": 11119 }, { "epoch": 0.6902973493078404, "grad_norm": 0.5288800502131157, "learning_rate": 9.493252278694918e-05, "loss": 3.0803, "step": 11120 }, { "epoch": 0.6903594264075983, "grad_norm": 0.5657622787215227, "learning_rate": 9.493093841893574e-05, "loss": 3.1887, "step": 11121 }, { "epoch": 0.6904215035073561, "grad_norm": 0.5182366266783957, "learning_rate": 9.492935381650668e-05, "loss": 3.1532, "step": 11122 }, { "epoch": 0.690483580607114, "grad_norm": 0.5831472207810477, "learning_rate": 9.492776897967026e-05, "loss": 3.0359, "step": 11123 }, { "epoch": 0.6905456577068719, "grad_norm": 0.5436229195571425, "learning_rate": 9.492618390843478e-05, "loss": 3.1176, "step": 11124 }, { "epoch": 0.6906077348066298, "grad_norm": 0.4984099915665562, "learning_rate": 9.492459860280849e-05, "loss": 3.0938, "step": 11125 }, { "epoch": 0.6906698119063878, "grad_norm": 0.4969009980411836, "learning_rate": 9.492301306279965e-05, "loss": 3.1011, "step": 11126 }, { "epoch": 0.6907318890061457, "grad_norm": 0.4645821131560492, "learning_rate": 9.492142728841656e-05, "loss": 3.0952, "step": 11127 }, { "epoch": 0.6907939661059035, "grad_norm": 0.5324002043917125, "learning_rate": 9.491984127966747e-05, "loss": 3.0743, "step": 11128 }, { "epoch": 0.6908560432056614, "grad_norm": 0.4915923804667275, "learning_rate": 9.491825503656067e-05, "loss": 3.108, "step": 11129 }, { "epoch": 0.6909181203054193, "grad_norm": 0.4720715302975232, "learning_rate": 9.491666855910442e-05, "loss": 3.0745, "step": 11130 }, { "epoch": 0.6909801974051772, "grad_norm": 0.4251197260625116, "learning_rate": 9.491508184730702e-05, "loss": 3.0979, "step": 11131 }, { "epoch": 0.6910422745049352, "grad_norm": 0.45835965434992904, "learning_rate": 9.491349490117674e-05, "loss": 3.0163, "step": 11132 }, { "epoch": 0.6911043516046931, "grad_norm": 0.4485880197061374, "learning_rate": 9.491190772072183e-05, "loss": 3.0943, "step": 11133 }, { "epoch": 0.6911664287044509, "grad_norm": 0.43936285151349513, "learning_rate": 9.491032030595061e-05, "loss": 3.123, "step": 11134 }, { "epoch": 0.6912285058042088, "grad_norm": 0.4835672719482587, "learning_rate": 9.490873265687135e-05, "loss": 3.073, "step": 11135 }, { "epoch": 0.6912905829039667, "grad_norm": 0.5063097506454254, "learning_rate": 9.490714477349231e-05, "loss": 3.1087, "step": 11136 }, { "epoch": 0.6913526600037246, "grad_norm": 0.4138671666172773, "learning_rate": 9.490555665582182e-05, "loss": 2.9661, "step": 11137 }, { "epoch": 0.6914147371034826, "grad_norm": 0.486550980947923, "learning_rate": 9.490396830386811e-05, "loss": 3.1093, "step": 11138 }, { "epoch": 0.6914768142032405, "grad_norm": 0.4747452717947878, "learning_rate": 9.490237971763953e-05, "loss": 3.0683, "step": 11139 }, { "epoch": 0.6915388913029983, "grad_norm": 0.4568658513145977, "learning_rate": 9.490079089714431e-05, "loss": 3.0522, "step": 11140 }, { "epoch": 0.6916009684027562, "grad_norm": 0.4578073384120942, "learning_rate": 9.489920184239077e-05, "loss": 3.1063, "step": 11141 }, { "epoch": 0.6916630455025141, "grad_norm": 0.5052287467948611, "learning_rate": 9.489761255338719e-05, "loss": 3.065, "step": 11142 }, { "epoch": 0.691725122602272, "grad_norm": 0.48565852936868303, "learning_rate": 9.489602303014187e-05, "loss": 3.093, "step": 11143 }, { "epoch": 0.69178719970203, "grad_norm": 0.5106968759309353, "learning_rate": 9.489443327266311e-05, "loss": 3.0812, "step": 11144 }, { "epoch": 0.6918492768017879, "grad_norm": 0.5239652279298451, "learning_rate": 9.489284328095918e-05, "loss": 3.1644, "step": 11145 }, { "epoch": 0.6919113539015457, "grad_norm": 0.5431524556005874, "learning_rate": 9.48912530550384e-05, "loss": 3.0888, "step": 11146 }, { "epoch": 0.6919734310013036, "grad_norm": 0.5798787503278484, "learning_rate": 9.488966259490904e-05, "loss": 2.9826, "step": 11147 }, { "epoch": 0.6920355081010615, "grad_norm": 0.5340067079529938, "learning_rate": 9.488807190057942e-05, "loss": 3.0789, "step": 11148 }, { "epoch": 0.6920975852008194, "grad_norm": 0.5252250495569625, "learning_rate": 9.488648097205783e-05, "loss": 3.0617, "step": 11149 }, { "epoch": 0.6921596623005773, "grad_norm": 0.5764341371717969, "learning_rate": 9.488488980935258e-05, "loss": 3.069, "step": 11150 }, { "epoch": 0.6922217394003353, "grad_norm": 0.5286393815071261, "learning_rate": 9.488329841247197e-05, "loss": 3.0207, "step": 11151 }, { "epoch": 0.6922838165000931, "grad_norm": 0.5224585417220187, "learning_rate": 9.488170678142427e-05, "loss": 3.0496, "step": 11152 }, { "epoch": 0.692345893599851, "grad_norm": 0.5337877247720672, "learning_rate": 9.488011491621782e-05, "loss": 2.9927, "step": 11153 }, { "epoch": 0.6924079706996089, "grad_norm": 0.5383900304095167, "learning_rate": 9.487852281686094e-05, "loss": 3.0777, "step": 11154 }, { "epoch": 0.6924700477993668, "grad_norm": 0.5184530672262049, "learning_rate": 9.487693048336189e-05, "loss": 3.0359, "step": 11155 }, { "epoch": 0.6925321248991247, "grad_norm": 0.5257734584372737, "learning_rate": 9.4875337915729e-05, "loss": 3.0788, "step": 11156 }, { "epoch": 0.6925942019988827, "grad_norm": 0.5223480387838825, "learning_rate": 9.487374511397058e-05, "loss": 3.0242, "step": 11157 }, { "epoch": 0.6926562790986405, "grad_norm": 0.5466647697722081, "learning_rate": 9.487215207809493e-05, "loss": 3.1527, "step": 11158 }, { "epoch": 0.6927183561983984, "grad_norm": 0.5774892455912045, "learning_rate": 9.487055880811037e-05, "loss": 3.02, "step": 11159 }, { "epoch": 0.6927804332981563, "grad_norm": 0.5719699812011273, "learning_rate": 9.486896530402523e-05, "loss": 3.1071, "step": 11160 }, { "epoch": 0.6928425103979142, "grad_norm": 0.6129935981017923, "learning_rate": 9.486737156584779e-05, "loss": 3.0987, "step": 11161 }, { "epoch": 0.6929045874976721, "grad_norm": 0.5967743294386472, "learning_rate": 9.48657775935864e-05, "loss": 3.2232, "step": 11162 }, { "epoch": 0.6929666645974301, "grad_norm": 0.5979677522968455, "learning_rate": 9.486418338724933e-05, "loss": 3.1179, "step": 11163 }, { "epoch": 0.6930287416971879, "grad_norm": 0.6004048640511591, "learning_rate": 9.486258894684495e-05, "loss": 3.1303, "step": 11164 }, { "epoch": 0.6930908187969458, "grad_norm": 0.563318187333197, "learning_rate": 9.486099427238153e-05, "loss": 3.1096, "step": 11165 }, { "epoch": 0.6931528958967037, "grad_norm": 0.5097003471116621, "learning_rate": 9.485939936386742e-05, "loss": 3.0262, "step": 11166 }, { "epoch": 0.6932149729964616, "grad_norm": 0.5699687078318969, "learning_rate": 9.485780422131094e-05, "loss": 3.0858, "step": 11167 }, { "epoch": 0.6932770500962195, "grad_norm": 0.5798899541243564, "learning_rate": 9.48562088447204e-05, "loss": 3.0746, "step": 11168 }, { "epoch": 0.6933391271959775, "grad_norm": 0.48598799243857804, "learning_rate": 9.485461323410413e-05, "loss": 3.011, "step": 11169 }, { "epoch": 0.6934012042957353, "grad_norm": 0.6494693818587838, "learning_rate": 9.485301738947046e-05, "loss": 3.1062, "step": 11170 }, { "epoch": 0.6934632813954932, "grad_norm": 0.6392923077146637, "learning_rate": 9.485142131082769e-05, "loss": 3.0639, "step": 11171 }, { "epoch": 0.6935253584952511, "grad_norm": 0.5905836151398326, "learning_rate": 9.484982499818419e-05, "loss": 3.152, "step": 11172 }, { "epoch": 0.693587435595009, "grad_norm": 0.5907327634401108, "learning_rate": 9.484822845154826e-05, "loss": 3.113, "step": 11173 }, { "epoch": 0.6936495126947669, "grad_norm": 0.5654850539012364, "learning_rate": 9.484663167092822e-05, "loss": 3.0066, "step": 11174 }, { "epoch": 0.6937115897945249, "grad_norm": 0.5987813992320693, "learning_rate": 9.484503465633243e-05, "loss": 3.1111, "step": 11175 }, { "epoch": 0.6937736668942827, "grad_norm": 0.6644660845282677, "learning_rate": 9.484343740776921e-05, "loss": 3.0864, "step": 11176 }, { "epoch": 0.6938357439940406, "grad_norm": 0.5570583449270485, "learning_rate": 9.484183992524689e-05, "loss": 3.0906, "step": 11177 }, { "epoch": 0.6938978210937985, "grad_norm": 0.5639335590619297, "learning_rate": 9.484024220877378e-05, "loss": 3.0526, "step": 11178 }, { "epoch": 0.6939598981935564, "grad_norm": 0.5858579200109006, "learning_rate": 9.483864425835828e-05, "loss": 3.0991, "step": 11179 }, { "epoch": 0.6940219752933143, "grad_norm": 0.5597386398253051, "learning_rate": 9.483704607400868e-05, "loss": 3.1146, "step": 11180 }, { "epoch": 0.6940840523930722, "grad_norm": 0.4925221561879749, "learning_rate": 9.48354476557333e-05, "loss": 3.0436, "step": 11181 }, { "epoch": 0.69414612949283, "grad_norm": 0.5729020261338839, "learning_rate": 9.483384900354053e-05, "loss": 3.1185, "step": 11182 }, { "epoch": 0.694208206592588, "grad_norm": 0.5577603505438187, "learning_rate": 9.483225011743867e-05, "loss": 3.0714, "step": 11183 }, { "epoch": 0.6942702836923459, "grad_norm": 0.515226123156636, "learning_rate": 9.483065099743609e-05, "loss": 3.1159, "step": 11184 }, { "epoch": 0.6943323607921038, "grad_norm": 0.6808687446899374, "learning_rate": 9.482905164354111e-05, "loss": 3.0818, "step": 11185 }, { "epoch": 0.6943944378918617, "grad_norm": 0.5292659342263222, "learning_rate": 9.482745205576209e-05, "loss": 3.1031, "step": 11186 }, { "epoch": 0.6944565149916196, "grad_norm": 0.6484537697095303, "learning_rate": 9.482585223410738e-05, "loss": 3.0801, "step": 11187 }, { "epoch": 0.6945185920913775, "grad_norm": 0.529270410755902, "learning_rate": 9.48242521785853e-05, "loss": 3.0417, "step": 11188 }, { "epoch": 0.6945806691911354, "grad_norm": 0.4599955033517899, "learning_rate": 9.482265188920423e-05, "loss": 3.107, "step": 11189 }, { "epoch": 0.6946427462908933, "grad_norm": 0.5578857683633573, "learning_rate": 9.48210513659725e-05, "loss": 3.1146, "step": 11190 }, { "epoch": 0.6947048233906512, "grad_norm": 0.48246046830837763, "learning_rate": 9.481945060889845e-05, "loss": 3.1551, "step": 11191 }, { "epoch": 0.6947669004904091, "grad_norm": 0.5487374064449392, "learning_rate": 9.481784961799046e-05, "loss": 3.1173, "step": 11192 }, { "epoch": 0.694828977590167, "grad_norm": 0.44480457115959077, "learning_rate": 9.481624839325687e-05, "loss": 3.1302, "step": 11193 }, { "epoch": 0.6948910546899248, "grad_norm": 0.49184858672926013, "learning_rate": 9.481464693470603e-05, "loss": 3.1384, "step": 11194 }, { "epoch": 0.6949531317896828, "grad_norm": 0.5534774584052525, "learning_rate": 9.481304524234629e-05, "loss": 3.1724, "step": 11195 }, { "epoch": 0.6950152088894407, "grad_norm": 0.5143014798747026, "learning_rate": 9.481144331618603e-05, "loss": 3.166, "step": 11196 }, { "epoch": 0.6950772859891986, "grad_norm": 0.5397749629270989, "learning_rate": 9.480984115623357e-05, "loss": 3.1268, "step": 11197 }, { "epoch": 0.6951393630889565, "grad_norm": 0.5408455757034917, "learning_rate": 9.480823876249731e-05, "loss": 3.1157, "step": 11198 }, { "epoch": 0.6952014401887144, "grad_norm": 0.531008890077558, "learning_rate": 9.480663613498559e-05, "loss": 3.1155, "step": 11199 }, { "epoch": 0.6952635172884722, "grad_norm": 0.43974909670196455, "learning_rate": 9.480503327370676e-05, "loss": 3.0633, "step": 11200 }, { "epoch": 0.6953255943882302, "grad_norm": 0.5246224442229551, "learning_rate": 9.48034301786692e-05, "loss": 3.1415, "step": 11201 }, { "epoch": 0.6953876714879881, "grad_norm": 0.43396372089286217, "learning_rate": 9.480182684988126e-05, "loss": 3.0452, "step": 11202 }, { "epoch": 0.695449748587746, "grad_norm": 0.5048987793749574, "learning_rate": 9.480022328735134e-05, "loss": 3.0609, "step": 11203 }, { "epoch": 0.6955118256875039, "grad_norm": 0.48736194281882905, "learning_rate": 9.479861949108775e-05, "loss": 3.1106, "step": 11204 }, { "epoch": 0.6955739027872618, "grad_norm": 0.5162404173545859, "learning_rate": 9.479701546109891e-05, "loss": 3.1067, "step": 11205 }, { "epoch": 0.6956359798870196, "grad_norm": 0.5543211680943286, "learning_rate": 9.479541119739314e-05, "loss": 3.1012, "step": 11206 }, { "epoch": 0.6956980569867776, "grad_norm": 0.5441186378785348, "learning_rate": 9.479380669997886e-05, "loss": 3.0539, "step": 11207 }, { "epoch": 0.6957601340865355, "grad_norm": 0.4689761411041607, "learning_rate": 9.47922019688644e-05, "loss": 3.063, "step": 11208 }, { "epoch": 0.6958222111862934, "grad_norm": 0.5233329554675898, "learning_rate": 9.479059700405816e-05, "loss": 3.0857, "step": 11209 }, { "epoch": 0.6958842882860513, "grad_norm": 0.5097460103554695, "learning_rate": 9.47889918055685e-05, "loss": 3.0758, "step": 11210 }, { "epoch": 0.6959463653858092, "grad_norm": 0.5055298192014064, "learning_rate": 9.478738637340378e-05, "loss": 3.0263, "step": 11211 }, { "epoch": 0.696008442485567, "grad_norm": 0.5207566745291856, "learning_rate": 9.478578070757241e-05, "loss": 3.0641, "step": 11212 }, { "epoch": 0.696070519585325, "grad_norm": 0.560009955079285, "learning_rate": 9.478417480808274e-05, "loss": 3.2183, "step": 11213 }, { "epoch": 0.6961325966850829, "grad_norm": 0.5545798452436245, "learning_rate": 9.478256867494317e-05, "loss": 3.1122, "step": 11214 }, { "epoch": 0.6961946737848408, "grad_norm": 0.5749655298598678, "learning_rate": 9.478096230816207e-05, "loss": 3.1218, "step": 11215 }, { "epoch": 0.6962567508845987, "grad_norm": 0.5104391099575825, "learning_rate": 9.477935570774781e-05, "loss": 3.0222, "step": 11216 }, { "epoch": 0.6963188279843566, "grad_norm": 0.4853552006483441, "learning_rate": 9.477774887370878e-05, "loss": 3.0851, "step": 11217 }, { "epoch": 0.6963809050841144, "grad_norm": 0.5224230886279534, "learning_rate": 9.477614180605337e-05, "loss": 3.1161, "step": 11218 }, { "epoch": 0.6964429821838723, "grad_norm": 0.47691219819703035, "learning_rate": 9.477453450478994e-05, "loss": 3.035, "step": 11219 }, { "epoch": 0.6965050592836303, "grad_norm": 0.47917184792357237, "learning_rate": 9.477292696992692e-05, "loss": 3.05, "step": 11220 }, { "epoch": 0.6965671363833882, "grad_norm": 0.4555181905043316, "learning_rate": 9.477131920147266e-05, "loss": 3.1189, "step": 11221 }, { "epoch": 0.6966292134831461, "grad_norm": 0.49615638818058366, "learning_rate": 9.476971119943557e-05, "loss": 3.032, "step": 11222 }, { "epoch": 0.696691290582904, "grad_norm": 0.5430832611744596, "learning_rate": 9.476810296382402e-05, "loss": 3.0662, "step": 11223 }, { "epoch": 0.6967533676826618, "grad_norm": 0.5001253030647822, "learning_rate": 9.476649449464638e-05, "loss": 3.1267, "step": 11224 }, { "epoch": 0.6968154447824197, "grad_norm": 0.51783878312662, "learning_rate": 9.476488579191111e-05, "loss": 3.0716, "step": 11225 }, { "epoch": 0.6968775218821777, "grad_norm": 0.4631254992566011, "learning_rate": 9.476327685562655e-05, "loss": 3.1454, "step": 11226 }, { "epoch": 0.6969395989819356, "grad_norm": 0.5167937759456429, "learning_rate": 9.476166768580111e-05, "loss": 3.235, "step": 11227 }, { "epoch": 0.6970016760816935, "grad_norm": 0.6374470576573952, "learning_rate": 9.476005828244317e-05, "loss": 2.9867, "step": 11228 }, { "epoch": 0.6970637531814514, "grad_norm": 0.496885298976086, "learning_rate": 9.475844864556116e-05, "loss": 3.022, "step": 11229 }, { "epoch": 0.6971258302812092, "grad_norm": 0.5916894746019926, "learning_rate": 9.475683877516345e-05, "loss": 3.0813, "step": 11230 }, { "epoch": 0.6971879073809671, "grad_norm": 0.6354856427594542, "learning_rate": 9.475522867125844e-05, "loss": 3.1644, "step": 11231 }, { "epoch": 0.6972499844807251, "grad_norm": 0.5263661683591131, "learning_rate": 9.475361833385454e-05, "loss": 3.1703, "step": 11232 }, { "epoch": 0.697312061580483, "grad_norm": 0.5306695684400771, "learning_rate": 9.475200776296016e-05, "loss": 3.1215, "step": 11233 }, { "epoch": 0.6973741386802409, "grad_norm": 0.5611552482604961, "learning_rate": 9.475039695858369e-05, "loss": 3.1449, "step": 11234 }, { "epoch": 0.6974362157799988, "grad_norm": 0.5629263427753598, "learning_rate": 9.474878592073353e-05, "loss": 3.0689, "step": 11235 }, { "epoch": 0.6974982928797566, "grad_norm": 0.5329747811465196, "learning_rate": 9.474717464941809e-05, "loss": 3.0927, "step": 11236 }, { "epoch": 0.6975603699795145, "grad_norm": 0.5498238758377296, "learning_rate": 9.474556314464577e-05, "loss": 3.119, "step": 11237 }, { "epoch": 0.6976224470792725, "grad_norm": 0.506819855245587, "learning_rate": 9.474395140642501e-05, "loss": 3.0933, "step": 11238 }, { "epoch": 0.6976845241790304, "grad_norm": 0.4896552628238121, "learning_rate": 9.474233943476416e-05, "loss": 3.097, "step": 11239 }, { "epoch": 0.6977466012787883, "grad_norm": 0.5004019016064312, "learning_rate": 9.474072722967168e-05, "loss": 3.0797, "step": 11240 }, { "epoch": 0.6978086783785461, "grad_norm": 0.5166179459683506, "learning_rate": 9.473911479115597e-05, "loss": 3.204, "step": 11241 }, { "epoch": 0.697870755478304, "grad_norm": 0.489430395208979, "learning_rate": 9.473750211922543e-05, "loss": 3.0072, "step": 11242 }, { "epoch": 0.6979328325780619, "grad_norm": 0.4500367884698085, "learning_rate": 9.473588921388848e-05, "loss": 3.1913, "step": 11243 }, { "epoch": 0.6979949096778199, "grad_norm": 0.5048661367588058, "learning_rate": 9.473427607515353e-05, "loss": 3.1459, "step": 11244 }, { "epoch": 0.6980569867775778, "grad_norm": 0.5166311850464832, "learning_rate": 9.473266270302902e-05, "loss": 3.129, "step": 11245 }, { "epoch": 0.6981190638773357, "grad_norm": 0.4940147390329001, "learning_rate": 9.473104909752335e-05, "loss": 3.151, "step": 11246 }, { "epoch": 0.6981811409770935, "grad_norm": 0.41795707401117227, "learning_rate": 9.472943525864493e-05, "loss": 3.048, "step": 11247 }, { "epoch": 0.6982432180768514, "grad_norm": 0.5026617725414008, "learning_rate": 9.472782118640218e-05, "loss": 3.1266, "step": 11248 }, { "epoch": 0.6983052951766093, "grad_norm": 0.4705490398278015, "learning_rate": 9.472620688080353e-05, "loss": 3.0276, "step": 11249 }, { "epoch": 0.6983673722763672, "grad_norm": 0.5077533687496705, "learning_rate": 9.47245923418574e-05, "loss": 3.0773, "step": 11250 }, { "epoch": 0.6984294493761252, "grad_norm": 0.507818867570129, "learning_rate": 9.472297756957223e-05, "loss": 3.1483, "step": 11251 }, { "epoch": 0.6984915264758831, "grad_norm": 0.5173372103854701, "learning_rate": 9.472136256395641e-05, "loss": 3.0684, "step": 11252 }, { "epoch": 0.6985536035756409, "grad_norm": 0.4171613756494027, "learning_rate": 9.471974732501841e-05, "loss": 2.9582, "step": 11253 }, { "epoch": 0.6986156806753988, "grad_norm": 0.5023490739365254, "learning_rate": 9.471813185276662e-05, "loss": 3.0713, "step": 11254 }, { "epoch": 0.6986777577751567, "grad_norm": 0.4903395186663429, "learning_rate": 9.471651614720946e-05, "loss": 3.103, "step": 11255 }, { "epoch": 0.6987398348749146, "grad_norm": 0.5315622085489756, "learning_rate": 9.471490020835541e-05, "loss": 3.0852, "step": 11256 }, { "epoch": 0.6988019119746726, "grad_norm": 0.4864658482425149, "learning_rate": 9.471328403621284e-05, "loss": 3.0906, "step": 11257 }, { "epoch": 0.6988639890744305, "grad_norm": 0.5003513651327839, "learning_rate": 9.471166763079025e-05, "loss": 3.094, "step": 11258 }, { "epoch": 0.6989260661741883, "grad_norm": 0.5021550110216433, "learning_rate": 9.4710050992096e-05, "loss": 3.1217, "step": 11259 }, { "epoch": 0.6989881432739462, "grad_norm": 0.5253890083600099, "learning_rate": 9.470843412013859e-05, "loss": 3.0746, "step": 11260 }, { "epoch": 0.6990502203737041, "grad_norm": 0.5259748139407234, "learning_rate": 9.470681701492641e-05, "loss": 3.1115, "step": 11261 }, { "epoch": 0.699112297473462, "grad_norm": 0.5322522638397307, "learning_rate": 9.470519967646791e-05, "loss": 3.0645, "step": 11262 }, { "epoch": 0.69917437457322, "grad_norm": 0.51646381120853, "learning_rate": 9.470358210477153e-05, "loss": 3.1463, "step": 11263 }, { "epoch": 0.6992364516729779, "grad_norm": 0.5334146659893809, "learning_rate": 9.470196429984573e-05, "loss": 3.1, "step": 11264 }, { "epoch": 0.6992985287727357, "grad_norm": 0.4632117207991126, "learning_rate": 9.47003462616989e-05, "loss": 3.2375, "step": 11265 }, { "epoch": 0.6993606058724936, "grad_norm": 0.566208061534661, "learning_rate": 9.469872799033954e-05, "loss": 3.0944, "step": 11266 }, { "epoch": 0.6994226829722515, "grad_norm": 0.4657235514911567, "learning_rate": 9.469710948577605e-05, "loss": 3.0847, "step": 11267 }, { "epoch": 0.6994847600720094, "grad_norm": 0.5001655245025027, "learning_rate": 9.469549074801689e-05, "loss": 3.1501, "step": 11268 }, { "epoch": 0.6995468371717674, "grad_norm": 0.5048997533084768, "learning_rate": 9.469387177707052e-05, "loss": 3.0894, "step": 11269 }, { "epoch": 0.6996089142715253, "grad_norm": 0.5174364579852953, "learning_rate": 9.469225257294535e-05, "loss": 3.0808, "step": 11270 }, { "epoch": 0.6996709913712831, "grad_norm": 0.4585112385386442, "learning_rate": 9.469063313564985e-05, "loss": 3.1205, "step": 11271 }, { "epoch": 0.699733068471041, "grad_norm": 0.5159230093535556, "learning_rate": 9.468901346519248e-05, "loss": 3.1025, "step": 11272 }, { "epoch": 0.6997951455707989, "grad_norm": 0.49822786819774784, "learning_rate": 9.468739356158168e-05, "loss": 3.2137, "step": 11273 }, { "epoch": 0.6998572226705568, "grad_norm": 0.5199254770247302, "learning_rate": 9.46857734248259e-05, "loss": 3.1801, "step": 11274 }, { "epoch": 0.6999192997703148, "grad_norm": 0.5093197681462186, "learning_rate": 9.468415305493357e-05, "loss": 3.0735, "step": 11275 }, { "epoch": 0.6999813768700727, "grad_norm": 0.5378760674528309, "learning_rate": 9.468253245191318e-05, "loss": 3.0659, "step": 11276 }, { "epoch": 0.7000434539698305, "grad_norm": 0.5257596276472419, "learning_rate": 9.468091161577317e-05, "loss": 3.0139, "step": 11277 }, { "epoch": 0.7001055310695884, "grad_norm": 0.4768307821839341, "learning_rate": 9.4679290546522e-05, "loss": 3.0037, "step": 11278 }, { "epoch": 0.7001676081693463, "grad_norm": 0.5624508306417979, "learning_rate": 9.467766924416813e-05, "loss": 2.9941, "step": 11279 }, { "epoch": 0.7002296852691042, "grad_norm": 0.46605941745761265, "learning_rate": 9.467604770872e-05, "loss": 2.9975, "step": 11280 }, { "epoch": 0.7002917623688621, "grad_norm": 0.5845232617388895, "learning_rate": 9.46744259401861e-05, "loss": 3.1988, "step": 11281 }, { "epoch": 0.7003538394686201, "grad_norm": 0.5545763253567625, "learning_rate": 9.467280393857487e-05, "loss": 3.0349, "step": 11282 }, { "epoch": 0.7004159165683779, "grad_norm": 0.48839032290058465, "learning_rate": 9.467118170389477e-05, "loss": 3.0348, "step": 11283 }, { "epoch": 0.7004779936681358, "grad_norm": 0.4757335519919157, "learning_rate": 9.466955923615428e-05, "loss": 3.0196, "step": 11284 }, { "epoch": 0.7005400707678937, "grad_norm": 0.49280321875657385, "learning_rate": 9.466793653536184e-05, "loss": 3.1497, "step": 11285 }, { "epoch": 0.7006021478676516, "grad_norm": 0.4809771228602062, "learning_rate": 9.466631360152595e-05, "loss": 3.1093, "step": 11286 }, { "epoch": 0.7006642249674095, "grad_norm": 0.4371806750470365, "learning_rate": 9.466469043465505e-05, "loss": 3.0727, "step": 11287 }, { "epoch": 0.7007263020671675, "grad_norm": 0.5115225321094239, "learning_rate": 9.466306703475763e-05, "loss": 3.1033, "step": 11288 }, { "epoch": 0.7007883791669253, "grad_norm": 0.4682849961781081, "learning_rate": 9.466144340184212e-05, "loss": 3.1615, "step": 11289 }, { "epoch": 0.7008504562666832, "grad_norm": 0.5026773949546182, "learning_rate": 9.465981953591702e-05, "loss": 3.0658, "step": 11290 }, { "epoch": 0.7009125333664411, "grad_norm": 0.5283927374254772, "learning_rate": 9.465819543699082e-05, "loss": 3.0321, "step": 11291 }, { "epoch": 0.700974610466199, "grad_norm": 0.5462651803637424, "learning_rate": 9.465657110507198e-05, "loss": 3.0515, "step": 11292 }, { "epoch": 0.7010366875659569, "grad_norm": 0.4425486070800865, "learning_rate": 9.465494654016895e-05, "loss": 3.0966, "step": 11293 }, { "epoch": 0.7010987646657149, "grad_norm": 0.5413207973403233, "learning_rate": 9.465332174229023e-05, "loss": 3.0879, "step": 11294 }, { "epoch": 0.7011608417654727, "grad_norm": 0.520238037050657, "learning_rate": 9.465169671144429e-05, "loss": 3.0286, "step": 11295 }, { "epoch": 0.7012229188652306, "grad_norm": 0.4732518187647641, "learning_rate": 9.46500714476396e-05, "loss": 3.0382, "step": 11296 }, { "epoch": 0.7012849959649885, "grad_norm": 0.49421534519910176, "learning_rate": 9.464844595088465e-05, "loss": 3.0804, "step": 11297 }, { "epoch": 0.7013470730647464, "grad_norm": 0.49309170635952354, "learning_rate": 9.464682022118791e-05, "loss": 3.0393, "step": 11298 }, { "epoch": 0.7014091501645043, "grad_norm": 0.5070208908457, "learning_rate": 9.464519425855789e-05, "loss": 3.0362, "step": 11299 }, { "epoch": 0.7014712272642623, "grad_norm": 0.4810977013609356, "learning_rate": 9.464356806300306e-05, "loss": 3.0547, "step": 11300 }, { "epoch": 0.7015333043640201, "grad_norm": 0.5633226048794755, "learning_rate": 9.464194163453188e-05, "loss": 3.0703, "step": 11301 }, { "epoch": 0.701595381463778, "grad_norm": 0.5849591543332848, "learning_rate": 9.464031497315287e-05, "loss": 3.1055, "step": 11302 }, { "epoch": 0.7016574585635359, "grad_norm": 0.5636024532804167, "learning_rate": 9.463868807887449e-05, "loss": 3.1426, "step": 11303 }, { "epoch": 0.7017195356632938, "grad_norm": 0.5885686924448514, "learning_rate": 9.463706095170521e-05, "loss": 3.0491, "step": 11304 }, { "epoch": 0.7017816127630517, "grad_norm": 0.4790893540691108, "learning_rate": 9.46354335916536e-05, "loss": 3.0337, "step": 11305 }, { "epoch": 0.7018436898628096, "grad_norm": 0.5319483878745094, "learning_rate": 9.463380599872806e-05, "loss": 3.105, "step": 11306 }, { "epoch": 0.7019057669625675, "grad_norm": 0.5431829071342346, "learning_rate": 9.463217817293712e-05, "loss": 3.0518, "step": 11307 }, { "epoch": 0.7019678440623254, "grad_norm": 0.5646594769227669, "learning_rate": 9.463055011428928e-05, "loss": 3.0776, "step": 11308 }, { "epoch": 0.7020299211620833, "grad_norm": 0.5465763093786203, "learning_rate": 9.462892182279304e-05, "loss": 3.0611, "step": 11309 }, { "epoch": 0.7020919982618412, "grad_norm": 0.4179281589370643, "learning_rate": 9.462729329845685e-05, "loss": 3.0839, "step": 11310 }, { "epoch": 0.7021540753615991, "grad_norm": 0.5488011363714203, "learning_rate": 9.462566454128928e-05, "loss": 3.1628, "step": 11311 }, { "epoch": 0.702216152461357, "grad_norm": 0.6167533409792028, "learning_rate": 9.462403555129876e-05, "loss": 3.0997, "step": 11312 }, { "epoch": 0.7022782295611149, "grad_norm": 0.6040204846906297, "learning_rate": 9.462240632849379e-05, "loss": 3.1243, "step": 11313 }, { "epoch": 0.7023403066608728, "grad_norm": 0.5375281071521958, "learning_rate": 9.462077687288292e-05, "loss": 3.1709, "step": 11314 }, { "epoch": 0.7024023837606307, "grad_norm": 0.6129667119389282, "learning_rate": 9.461914718447462e-05, "loss": 3.1452, "step": 11315 }, { "epoch": 0.7024644608603886, "grad_norm": 0.6358227656758242, "learning_rate": 9.46175172632774e-05, "loss": 3.0888, "step": 11316 }, { "epoch": 0.7025265379601465, "grad_norm": 0.5464889526175466, "learning_rate": 9.461588710929976e-05, "loss": 3.0926, "step": 11317 }, { "epoch": 0.7025886150599044, "grad_norm": 0.4503128163206219, "learning_rate": 9.461425672255021e-05, "loss": 3.1707, "step": 11318 }, { "epoch": 0.7026506921596622, "grad_norm": 0.5313031787061386, "learning_rate": 9.461262610303725e-05, "loss": 3.0677, "step": 11319 }, { "epoch": 0.7027127692594202, "grad_norm": 0.5072881485688442, "learning_rate": 9.461099525076939e-05, "loss": 3.0754, "step": 11320 }, { "epoch": 0.7027748463591781, "grad_norm": 0.5259353479895533, "learning_rate": 9.460936416575513e-05, "loss": 3.1111, "step": 11321 }, { "epoch": 0.702836923458936, "grad_norm": 0.516991682524339, "learning_rate": 9.4607732848003e-05, "loss": 3.0259, "step": 11322 }, { "epoch": 0.7028990005586939, "grad_norm": 0.4720575465858242, "learning_rate": 9.460610129752149e-05, "loss": 3.0262, "step": 11323 }, { "epoch": 0.7029610776584518, "grad_norm": 0.5153530154890472, "learning_rate": 9.460446951431911e-05, "loss": 3.0317, "step": 11324 }, { "epoch": 0.7030231547582096, "grad_norm": 0.5021479187839935, "learning_rate": 9.460283749840442e-05, "loss": 2.9896, "step": 11325 }, { "epoch": 0.7030852318579676, "grad_norm": 0.5353447241139369, "learning_rate": 9.460120524978586e-05, "loss": 3.0338, "step": 11326 }, { "epoch": 0.7031473089577255, "grad_norm": 0.528574883972568, "learning_rate": 9.459957276847201e-05, "loss": 3.0882, "step": 11327 }, { "epoch": 0.7032093860574834, "grad_norm": 0.45804911347944594, "learning_rate": 9.459794005447136e-05, "loss": 2.929, "step": 11328 }, { "epoch": 0.7032714631572413, "grad_norm": 0.530800994290209, "learning_rate": 9.459630710779242e-05, "loss": 3.1337, "step": 11329 }, { "epoch": 0.7033335402569992, "grad_norm": 0.5242258938389299, "learning_rate": 9.459467392844373e-05, "loss": 3.0694, "step": 11330 }, { "epoch": 0.703395617356757, "grad_norm": 0.47252771505685154, "learning_rate": 9.459304051643379e-05, "loss": 3.1297, "step": 11331 }, { "epoch": 0.703457694456515, "grad_norm": 0.5272793765867041, "learning_rate": 9.459140687177113e-05, "loss": 3.0427, "step": 11332 }, { "epoch": 0.7035197715562729, "grad_norm": 0.4446683876611004, "learning_rate": 9.45897729944643e-05, "loss": 3.0792, "step": 11333 }, { "epoch": 0.7035818486560308, "grad_norm": 0.4727102201560595, "learning_rate": 9.458813888452176e-05, "loss": 3.119, "step": 11334 }, { "epoch": 0.7036439257557887, "grad_norm": 0.4673975188591411, "learning_rate": 9.45865045419521e-05, "loss": 3.0797, "step": 11335 }, { "epoch": 0.7037060028555466, "grad_norm": 0.458575419590886, "learning_rate": 9.458486996676383e-05, "loss": 3.0622, "step": 11336 }, { "epoch": 0.7037680799553044, "grad_norm": 0.46338533772868107, "learning_rate": 9.458323515896546e-05, "loss": 3.0797, "step": 11337 }, { "epoch": 0.7038301570550624, "grad_norm": 0.458274340806728, "learning_rate": 9.458160011856551e-05, "loss": 3.1082, "step": 11338 }, { "epoch": 0.7038922341548203, "grad_norm": 0.4755800626732645, "learning_rate": 9.457996484557257e-05, "loss": 3.0215, "step": 11339 }, { "epoch": 0.7039543112545782, "grad_norm": 0.5478613269500126, "learning_rate": 9.45783293399951e-05, "loss": 3.1381, "step": 11340 }, { "epoch": 0.7040163883543361, "grad_norm": 0.5307073065161877, "learning_rate": 9.457669360184168e-05, "loss": 3.0935, "step": 11341 }, { "epoch": 0.704078465454094, "grad_norm": 0.5226752866462764, "learning_rate": 9.457505763112082e-05, "loss": 3.1412, "step": 11342 }, { "epoch": 0.7041405425538518, "grad_norm": 0.5849138600760855, "learning_rate": 9.457342142784106e-05, "loss": 3.1327, "step": 11343 }, { "epoch": 0.7042026196536098, "grad_norm": 0.5501948000817829, "learning_rate": 9.457178499201094e-05, "loss": 3.2164, "step": 11344 }, { "epoch": 0.7042646967533677, "grad_norm": 0.5626754751949107, "learning_rate": 9.457014832363901e-05, "loss": 3.099, "step": 11345 }, { "epoch": 0.7043267738531256, "grad_norm": 0.5201025668190354, "learning_rate": 9.456851142273379e-05, "loss": 3.0237, "step": 11346 }, { "epoch": 0.7043888509528835, "grad_norm": 0.46811450159603024, "learning_rate": 9.456687428930383e-05, "loss": 3.2092, "step": 11347 }, { "epoch": 0.7044509280526414, "grad_norm": 0.5060874100959621, "learning_rate": 9.456523692335766e-05, "loss": 2.9995, "step": 11348 }, { "epoch": 0.7045130051523992, "grad_norm": 0.5331348752787561, "learning_rate": 9.456359932490383e-05, "loss": 3.1198, "step": 11349 }, { "epoch": 0.7045750822521571, "grad_norm": 0.5547503852120876, "learning_rate": 9.45619614939509e-05, "loss": 3.1239, "step": 11350 }, { "epoch": 0.7046371593519151, "grad_norm": 0.47012009661368326, "learning_rate": 9.456032343050737e-05, "loss": 3.1222, "step": 11351 }, { "epoch": 0.704699236451673, "grad_norm": 0.5535977363927406, "learning_rate": 9.455868513458184e-05, "loss": 3.0408, "step": 11352 }, { "epoch": 0.7047613135514309, "grad_norm": 0.6619045199474318, "learning_rate": 9.45570466061828e-05, "loss": 3.1544, "step": 11353 }, { "epoch": 0.7048233906511888, "grad_norm": 0.6654648727013839, "learning_rate": 9.455540784531887e-05, "loss": 3.0605, "step": 11354 }, { "epoch": 0.7048854677509466, "grad_norm": 0.45034143885664746, "learning_rate": 9.455376885199855e-05, "loss": 3.0415, "step": 11355 }, { "epoch": 0.7049475448507045, "grad_norm": 0.5294825942444942, "learning_rate": 9.455212962623039e-05, "loss": 3.0396, "step": 11356 }, { "epoch": 0.7050096219504625, "grad_norm": 0.5926079809806252, "learning_rate": 9.455049016802297e-05, "loss": 3.0889, "step": 11357 }, { "epoch": 0.7050716990502204, "grad_norm": 0.45918646076874525, "learning_rate": 9.454885047738482e-05, "loss": 3.0583, "step": 11358 }, { "epoch": 0.7051337761499783, "grad_norm": 0.5522779325688754, "learning_rate": 9.45472105543245e-05, "loss": 3.1281, "step": 11359 }, { "epoch": 0.7051958532497362, "grad_norm": 0.5352581233495471, "learning_rate": 9.454557039885059e-05, "loss": 3.0017, "step": 11360 }, { "epoch": 0.705257930349494, "grad_norm": 0.4380869334358803, "learning_rate": 9.454393001097159e-05, "loss": 3.039, "step": 11361 }, { "epoch": 0.7053200074492519, "grad_norm": 0.5518565671873055, "learning_rate": 9.454228939069612e-05, "loss": 3.1077, "step": 11362 }, { "epoch": 0.7053820845490099, "grad_norm": 0.5134716731637107, "learning_rate": 9.454064853803271e-05, "loss": 3.0602, "step": 11363 }, { "epoch": 0.7054441616487678, "grad_norm": 0.4714739332962854, "learning_rate": 9.453900745298991e-05, "loss": 2.9865, "step": 11364 }, { "epoch": 0.7055062387485257, "grad_norm": 0.5453922060451916, "learning_rate": 9.45373661355763e-05, "loss": 3.039, "step": 11365 }, { "epoch": 0.7055683158482836, "grad_norm": 0.4864409595412063, "learning_rate": 9.453572458580045e-05, "loss": 3.0679, "step": 11366 }, { "epoch": 0.7056303929480414, "grad_norm": 0.49951106245756616, "learning_rate": 9.453408280367092e-05, "loss": 3.151, "step": 11367 }, { "epoch": 0.7056924700477993, "grad_norm": 0.5025507119433805, "learning_rate": 9.453244078919624e-05, "loss": 3.0597, "step": 11368 }, { "epoch": 0.7057545471475573, "grad_norm": 0.49077895112032105, "learning_rate": 9.453079854238503e-05, "loss": 3.1951, "step": 11369 }, { "epoch": 0.7058166242473152, "grad_norm": 0.4609035867402914, "learning_rate": 9.452915606324581e-05, "loss": 3.0093, "step": 11370 }, { "epoch": 0.7058787013470731, "grad_norm": 0.4972432852953844, "learning_rate": 9.45275133517872e-05, "loss": 3.0199, "step": 11371 }, { "epoch": 0.705940778446831, "grad_norm": 0.4811830251089019, "learning_rate": 9.452587040801772e-05, "loss": 3.1196, "step": 11372 }, { "epoch": 0.7060028555465888, "grad_norm": 0.4775055949647272, "learning_rate": 9.452422723194598e-05, "loss": 3.0662, "step": 11373 }, { "epoch": 0.7060649326463467, "grad_norm": 0.478775047513545, "learning_rate": 9.452258382358053e-05, "loss": 3.0332, "step": 11374 }, { "epoch": 0.7061270097461047, "grad_norm": 0.4572186534120949, "learning_rate": 9.452094018292996e-05, "loss": 2.9871, "step": 11375 }, { "epoch": 0.7061890868458626, "grad_norm": 0.4807806940729487, "learning_rate": 9.451929631000284e-05, "loss": 3.016, "step": 11376 }, { "epoch": 0.7062511639456205, "grad_norm": 0.4808629752426419, "learning_rate": 9.451765220480773e-05, "loss": 3.0841, "step": 11377 }, { "epoch": 0.7063132410453784, "grad_norm": 0.44445750491337244, "learning_rate": 9.451600786735323e-05, "loss": 2.9857, "step": 11378 }, { "epoch": 0.7063753181451362, "grad_norm": 0.4710628986536632, "learning_rate": 9.451436329764792e-05, "loss": 3.117, "step": 11379 }, { "epoch": 0.7064373952448941, "grad_norm": 0.527705598240012, "learning_rate": 9.451271849570035e-05, "loss": 3.0996, "step": 11380 }, { "epoch": 0.706499472344652, "grad_norm": 0.48847522697281415, "learning_rate": 9.451107346151913e-05, "loss": 3.1606, "step": 11381 }, { "epoch": 0.70656154944441, "grad_norm": 0.45391157544372035, "learning_rate": 9.450942819511284e-05, "loss": 3.0639, "step": 11382 }, { "epoch": 0.7066236265441679, "grad_norm": 0.5130000779912194, "learning_rate": 9.450778269649005e-05, "loss": 3.0381, "step": 11383 }, { "epoch": 0.7066857036439258, "grad_norm": 0.5138671851801537, "learning_rate": 9.450613696565935e-05, "loss": 3.1053, "step": 11384 }, { "epoch": 0.7067477807436836, "grad_norm": 0.3951888128388845, "learning_rate": 9.450449100262934e-05, "loss": 2.9757, "step": 11385 }, { "epoch": 0.7068098578434415, "grad_norm": 0.46260850896460176, "learning_rate": 9.45028448074086e-05, "loss": 3.0653, "step": 11386 }, { "epoch": 0.7068719349431994, "grad_norm": 0.4175683180082955, "learning_rate": 9.450119838000571e-05, "loss": 3.1318, "step": 11387 }, { "epoch": 0.7069340120429574, "grad_norm": 0.46181133652257983, "learning_rate": 9.449955172042926e-05, "loss": 2.9865, "step": 11388 }, { "epoch": 0.7069960891427153, "grad_norm": 0.484334374846848, "learning_rate": 9.449790482868785e-05, "loss": 3.0711, "step": 11389 }, { "epoch": 0.7070581662424732, "grad_norm": 0.5220382834733962, "learning_rate": 9.449625770479005e-05, "loss": 3.0813, "step": 11390 }, { "epoch": 0.707120243342231, "grad_norm": 0.4774218456360064, "learning_rate": 9.44946103487445e-05, "loss": 3.1388, "step": 11391 }, { "epoch": 0.7071823204419889, "grad_norm": 0.46054694229914955, "learning_rate": 9.449296276055975e-05, "loss": 3.1228, "step": 11392 }, { "epoch": 0.7072443975417468, "grad_norm": 0.45239328630353876, "learning_rate": 9.449131494024442e-05, "loss": 3.1294, "step": 11393 }, { "epoch": 0.7073064746415048, "grad_norm": 0.46991768674181084, "learning_rate": 9.44896668878071e-05, "loss": 3.0252, "step": 11394 }, { "epoch": 0.7073685517412627, "grad_norm": 0.4762195072024116, "learning_rate": 9.448801860325638e-05, "loss": 3.0541, "step": 11395 }, { "epoch": 0.7074306288410206, "grad_norm": 0.4477376321312362, "learning_rate": 9.448637008660087e-05, "loss": 3.1223, "step": 11396 }, { "epoch": 0.7074927059407784, "grad_norm": 0.45285042303174444, "learning_rate": 9.448472133784916e-05, "loss": 3.0512, "step": 11397 }, { "epoch": 0.7075547830405363, "grad_norm": 0.42460841763975415, "learning_rate": 9.448307235700988e-05, "loss": 3.0515, "step": 11398 }, { "epoch": 0.7076168601402942, "grad_norm": 0.42843092200351224, "learning_rate": 9.44814231440916e-05, "loss": 3.0992, "step": 11399 }, { "epoch": 0.7076789372400522, "grad_norm": 0.4528978205792443, "learning_rate": 9.447977369910294e-05, "loss": 3.0708, "step": 11400 }, { "epoch": 0.7077410143398101, "grad_norm": 0.45134064791545103, "learning_rate": 9.44781240220525e-05, "loss": 3.1138, "step": 11401 }, { "epoch": 0.707803091439568, "grad_norm": 0.44617489078486405, "learning_rate": 9.447647411294889e-05, "loss": 3.1721, "step": 11402 }, { "epoch": 0.7078651685393258, "grad_norm": 0.5140702765899816, "learning_rate": 9.447482397180072e-05, "loss": 3.153, "step": 11403 }, { "epoch": 0.7079272456390837, "grad_norm": 0.49897613599816004, "learning_rate": 9.44731735986166e-05, "loss": 3.0818, "step": 11404 }, { "epoch": 0.7079893227388416, "grad_norm": 0.4941669653103051, "learning_rate": 9.447152299340513e-05, "loss": 3.1001, "step": 11405 }, { "epoch": 0.7080513998385995, "grad_norm": 0.5073003680705535, "learning_rate": 9.446987215617494e-05, "loss": 3.0146, "step": 11406 }, { "epoch": 0.7081134769383575, "grad_norm": 0.5602620959867742, "learning_rate": 9.446822108693461e-05, "loss": 2.9989, "step": 11407 }, { "epoch": 0.7081755540381154, "grad_norm": 0.46449726879910136, "learning_rate": 9.446656978569279e-05, "loss": 2.9567, "step": 11408 }, { "epoch": 0.7082376311378732, "grad_norm": 0.46589660019846924, "learning_rate": 9.446491825245809e-05, "loss": 3.0673, "step": 11409 }, { "epoch": 0.7082997082376311, "grad_norm": 0.49879580268306045, "learning_rate": 9.44632664872391e-05, "loss": 3.0733, "step": 11410 }, { "epoch": 0.708361785337389, "grad_norm": 0.4419771064616183, "learning_rate": 9.446161449004446e-05, "loss": 3.0339, "step": 11411 }, { "epoch": 0.708423862437147, "grad_norm": 0.42125437261649507, "learning_rate": 9.445996226088279e-05, "loss": 3.1293, "step": 11412 }, { "epoch": 0.7084859395369049, "grad_norm": 0.46875871014449877, "learning_rate": 9.445830979976269e-05, "loss": 3.058, "step": 11413 }, { "epoch": 0.7085480166366628, "grad_norm": 0.4894894110768546, "learning_rate": 9.445665710669281e-05, "loss": 3.0264, "step": 11414 }, { "epoch": 0.7086100937364206, "grad_norm": 0.5026938172336756, "learning_rate": 9.445500418168175e-05, "loss": 3.0353, "step": 11415 }, { "epoch": 0.7086721708361785, "grad_norm": 0.45510485983973537, "learning_rate": 9.445335102473815e-05, "loss": 3.023, "step": 11416 }, { "epoch": 0.7087342479359364, "grad_norm": 0.5427113753201194, "learning_rate": 9.44516976358706e-05, "loss": 3.0753, "step": 11417 }, { "epoch": 0.7087963250356943, "grad_norm": 0.531630323816152, "learning_rate": 9.445004401508779e-05, "loss": 3.084, "step": 11418 }, { "epoch": 0.7088584021354523, "grad_norm": 0.48847293905800265, "learning_rate": 9.444839016239828e-05, "loss": 3.0803, "step": 11419 }, { "epoch": 0.7089204792352102, "grad_norm": 0.5376182226531248, "learning_rate": 9.444673607781073e-05, "loss": 3.1364, "step": 11420 }, { "epoch": 0.708982556334968, "grad_norm": 0.6289939345369074, "learning_rate": 9.44450817613338e-05, "loss": 3.0865, "step": 11421 }, { "epoch": 0.7090446334347259, "grad_norm": 0.5140293748380642, "learning_rate": 9.444342721297606e-05, "loss": 3.0319, "step": 11422 }, { "epoch": 0.7091067105344838, "grad_norm": 0.5527490944132467, "learning_rate": 9.444177243274618e-05, "loss": 3.0409, "step": 11423 }, { "epoch": 0.7091687876342417, "grad_norm": 0.5628865291491446, "learning_rate": 9.444011742065277e-05, "loss": 3.0757, "step": 11424 }, { "epoch": 0.7092308647339997, "grad_norm": 0.5196487179230432, "learning_rate": 9.44384621767045e-05, "loss": 3.0306, "step": 11425 }, { "epoch": 0.7092929418337576, "grad_norm": 0.5813017504224897, "learning_rate": 9.443680670090997e-05, "loss": 3.0657, "step": 11426 }, { "epoch": 0.7093550189335154, "grad_norm": 0.6371906380718761, "learning_rate": 9.443515099327784e-05, "loss": 3.0816, "step": 11427 }, { "epoch": 0.7094170960332733, "grad_norm": 0.631371139951094, "learning_rate": 9.443349505381676e-05, "loss": 3.0749, "step": 11428 }, { "epoch": 0.7094791731330312, "grad_norm": 0.50444014307019, "learning_rate": 9.443183888253531e-05, "loss": 3.1154, "step": 11429 }, { "epoch": 0.7095412502327891, "grad_norm": 0.5729203888743314, "learning_rate": 9.44301824794422e-05, "loss": 3.0986, "step": 11430 }, { "epoch": 0.709603327332547, "grad_norm": 0.46590253316256525, "learning_rate": 9.442852584454602e-05, "loss": 3.0632, "step": 11431 }, { "epoch": 0.709665404432305, "grad_norm": 0.5162185373948394, "learning_rate": 9.442686897785546e-05, "loss": 3.0168, "step": 11432 }, { "epoch": 0.7097274815320628, "grad_norm": 0.4662433199109942, "learning_rate": 9.442521187937912e-05, "loss": 3.1052, "step": 11433 }, { "epoch": 0.7097895586318207, "grad_norm": 0.5448973181875751, "learning_rate": 9.442355454912567e-05, "loss": 2.958, "step": 11434 }, { "epoch": 0.7098516357315786, "grad_norm": 0.4637747742803741, "learning_rate": 9.442189698710376e-05, "loss": 3.1095, "step": 11435 }, { "epoch": 0.7099137128313365, "grad_norm": 0.5162426688017789, "learning_rate": 9.442023919332202e-05, "loss": 3.0149, "step": 11436 }, { "epoch": 0.7099757899310944, "grad_norm": 0.48142260948645865, "learning_rate": 9.441858116778911e-05, "loss": 3.0194, "step": 11437 }, { "epoch": 0.7100378670308524, "grad_norm": 0.5714496354282413, "learning_rate": 9.441692291051368e-05, "loss": 3.0199, "step": 11438 }, { "epoch": 0.7100999441306102, "grad_norm": 0.5686829684653736, "learning_rate": 9.441526442150438e-05, "loss": 3.1784, "step": 11439 }, { "epoch": 0.7101620212303681, "grad_norm": 0.5325867999885262, "learning_rate": 9.441360570076987e-05, "loss": 3.0787, "step": 11440 }, { "epoch": 0.710224098330126, "grad_norm": 0.4790480473796203, "learning_rate": 9.44119467483188e-05, "loss": 3.0411, "step": 11441 }, { "epoch": 0.7102861754298839, "grad_norm": 0.5984096640146274, "learning_rate": 9.44102875641598e-05, "loss": 3.0456, "step": 11442 }, { "epoch": 0.7103482525296418, "grad_norm": 0.6363729228156467, "learning_rate": 9.440862814830156e-05, "loss": 3.1181, "step": 11443 }, { "epoch": 0.7104103296293998, "grad_norm": 0.615554217805434, "learning_rate": 9.440696850075273e-05, "loss": 3.1108, "step": 11444 }, { "epoch": 0.7104724067291576, "grad_norm": 0.5713832418948519, "learning_rate": 9.440530862152198e-05, "loss": 2.9959, "step": 11445 }, { "epoch": 0.7105344838289155, "grad_norm": 0.5005678588126903, "learning_rate": 9.440364851061793e-05, "loss": 3.0989, "step": 11446 }, { "epoch": 0.7105965609286734, "grad_norm": 0.6114524008112737, "learning_rate": 9.440198816804927e-05, "loss": 3.087, "step": 11447 }, { "epoch": 0.7106586380284313, "grad_norm": 0.5936661460289804, "learning_rate": 9.440032759382467e-05, "loss": 3.0466, "step": 11448 }, { "epoch": 0.7107207151281892, "grad_norm": 0.5934943602502081, "learning_rate": 9.439866678795277e-05, "loss": 2.9287, "step": 11449 }, { "epoch": 0.7107827922279472, "grad_norm": 0.5530071865842693, "learning_rate": 9.439700575044226e-05, "loss": 3.0374, "step": 11450 }, { "epoch": 0.710844869327705, "grad_norm": 0.5797540549349888, "learning_rate": 9.439534448130178e-05, "loss": 3.0206, "step": 11451 }, { "epoch": 0.7109069464274629, "grad_norm": 0.5826989510710221, "learning_rate": 9.439368298054003e-05, "loss": 3.0445, "step": 11452 }, { "epoch": 0.7109690235272208, "grad_norm": 0.5919604940812886, "learning_rate": 9.439202124816563e-05, "loss": 3.0942, "step": 11453 }, { "epoch": 0.7110311006269787, "grad_norm": 0.550553213759398, "learning_rate": 9.439035928418732e-05, "loss": 3.0642, "step": 11454 }, { "epoch": 0.7110931777267366, "grad_norm": 0.5642178054839695, "learning_rate": 9.438869708861368e-05, "loss": 3.1378, "step": 11455 }, { "epoch": 0.7111552548264946, "grad_norm": 0.5066994426294351, "learning_rate": 9.438703466145347e-05, "loss": 3.1793, "step": 11456 }, { "epoch": 0.7112173319262524, "grad_norm": 0.591764090356974, "learning_rate": 9.43853720027153e-05, "loss": 3.1197, "step": 11457 }, { "epoch": 0.7112794090260103, "grad_norm": 0.5649141059418292, "learning_rate": 9.438370911240789e-05, "loss": 3.0476, "step": 11458 }, { "epoch": 0.7113414861257682, "grad_norm": 0.46032643627919917, "learning_rate": 9.438204599053989e-05, "loss": 3.1395, "step": 11459 }, { "epoch": 0.7114035632255261, "grad_norm": 0.5566245896556138, "learning_rate": 9.438038263711997e-05, "loss": 3.1097, "step": 11460 }, { "epoch": 0.711465640325284, "grad_norm": 0.4878540611768174, "learning_rate": 9.437871905215683e-05, "loss": 3.1102, "step": 11461 }, { "epoch": 0.711527717425042, "grad_norm": 0.5248320776461112, "learning_rate": 9.437705523565914e-05, "loss": 3.0868, "step": 11462 }, { "epoch": 0.7115897945247998, "grad_norm": 0.4793569991898625, "learning_rate": 9.437539118763558e-05, "loss": 3.0436, "step": 11463 }, { "epoch": 0.7116518716245577, "grad_norm": 0.5050191787344949, "learning_rate": 9.437372690809483e-05, "loss": 2.9857, "step": 11464 }, { "epoch": 0.7117139487243156, "grad_norm": 0.4961477835958133, "learning_rate": 9.437206239704556e-05, "loss": 3.0252, "step": 11465 }, { "epoch": 0.7117760258240735, "grad_norm": 0.4885055179547855, "learning_rate": 9.43703976544965e-05, "loss": 3.158, "step": 11466 }, { "epoch": 0.7118381029238314, "grad_norm": 0.5270078083505829, "learning_rate": 9.436873268045628e-05, "loss": 3.1848, "step": 11467 }, { "epoch": 0.7119001800235893, "grad_norm": 0.5506067202388306, "learning_rate": 9.436706747493361e-05, "loss": 3.0877, "step": 11468 }, { "epoch": 0.7119622571233472, "grad_norm": 0.4791482130588368, "learning_rate": 9.436540203793718e-05, "loss": 3.1082, "step": 11469 }, { "epoch": 0.7120243342231051, "grad_norm": 0.4920915934052186, "learning_rate": 9.436373636947568e-05, "loss": 3.1051, "step": 11470 }, { "epoch": 0.712086411322863, "grad_norm": 0.5845429928682734, "learning_rate": 9.43620704695578e-05, "loss": 3.0213, "step": 11471 }, { "epoch": 0.7121484884226209, "grad_norm": 0.5232842064332868, "learning_rate": 9.436040433819224e-05, "loss": 3.0717, "step": 11472 }, { "epoch": 0.7122105655223788, "grad_norm": 0.5259752388989566, "learning_rate": 9.435873797538767e-05, "loss": 3.1177, "step": 11473 }, { "epoch": 0.7122726426221367, "grad_norm": 0.6278347341076109, "learning_rate": 9.43570713811528e-05, "loss": 3.1373, "step": 11474 }, { "epoch": 0.7123347197218945, "grad_norm": 0.5572128299554614, "learning_rate": 9.435540455549632e-05, "loss": 3.0631, "step": 11475 }, { "epoch": 0.7123967968216525, "grad_norm": 0.5038766778356335, "learning_rate": 9.435373749842693e-05, "loss": 3.0892, "step": 11476 }, { "epoch": 0.7124588739214104, "grad_norm": 0.5480934465867385, "learning_rate": 9.435207020995331e-05, "loss": 3.1399, "step": 11477 }, { "epoch": 0.7125209510211683, "grad_norm": 0.5152774564302761, "learning_rate": 9.435040269008419e-05, "loss": 3.0203, "step": 11478 }, { "epoch": 0.7125830281209262, "grad_norm": 0.46533995678647877, "learning_rate": 9.434873493882823e-05, "loss": 3.0916, "step": 11479 }, { "epoch": 0.7126451052206841, "grad_norm": 0.5492859420806012, "learning_rate": 9.434706695619418e-05, "loss": 3.1669, "step": 11480 }, { "epoch": 0.712707182320442, "grad_norm": 0.5381686221528561, "learning_rate": 9.43453987421907e-05, "loss": 3.0532, "step": 11481 }, { "epoch": 0.7127692594201999, "grad_norm": 0.45306673990802965, "learning_rate": 9.434373029682651e-05, "loss": 3.0576, "step": 11482 }, { "epoch": 0.7128313365199578, "grad_norm": 0.5717947001962469, "learning_rate": 9.434206162011033e-05, "loss": 3.1488, "step": 11483 }, { "epoch": 0.7128934136197157, "grad_norm": 0.5354903354752162, "learning_rate": 9.434039271205083e-05, "loss": 3.1176, "step": 11484 }, { "epoch": 0.7129554907194736, "grad_norm": 0.49370979857658603, "learning_rate": 9.433872357265674e-05, "loss": 3.0668, "step": 11485 }, { "epoch": 0.7130175678192315, "grad_norm": 0.5052128493379191, "learning_rate": 9.433705420193676e-05, "loss": 3.0681, "step": 11486 }, { "epoch": 0.7130796449189893, "grad_norm": 0.4847930980339389, "learning_rate": 9.433538459989964e-05, "loss": 3.1205, "step": 11487 }, { "epoch": 0.7131417220187473, "grad_norm": 0.510977987849249, "learning_rate": 9.433371476655403e-05, "loss": 3.0242, "step": 11488 }, { "epoch": 0.7132037991185052, "grad_norm": 0.5073796995518132, "learning_rate": 9.433204470190866e-05, "loss": 3.0638, "step": 11489 }, { "epoch": 0.7132658762182631, "grad_norm": 0.5245144994729835, "learning_rate": 9.433037440597226e-05, "loss": 3.1408, "step": 11490 }, { "epoch": 0.713327953318021, "grad_norm": 0.5516563119617613, "learning_rate": 9.432870387875355e-05, "loss": 3.1154, "step": 11491 }, { "epoch": 0.7133900304177789, "grad_norm": 0.5459727064461724, "learning_rate": 9.43270331202612e-05, "loss": 3.1004, "step": 11492 }, { "epoch": 0.7134521075175367, "grad_norm": 0.5621824162726752, "learning_rate": 9.432536213050398e-05, "loss": 3.1178, "step": 11493 }, { "epoch": 0.7135141846172947, "grad_norm": 0.52579205007358, "learning_rate": 9.432369090949059e-05, "loss": 3.0689, "step": 11494 }, { "epoch": 0.7135762617170526, "grad_norm": 0.5388761834184912, "learning_rate": 9.432201945722971e-05, "loss": 3.0892, "step": 11495 }, { "epoch": 0.7136383388168105, "grad_norm": 0.5647140744515872, "learning_rate": 9.432034777373013e-05, "loss": 3.0687, "step": 11496 }, { "epoch": 0.7137004159165684, "grad_norm": 0.4731497749534071, "learning_rate": 9.431867585900052e-05, "loss": 3.1421, "step": 11497 }, { "epoch": 0.7137624930163263, "grad_norm": 0.4922358625356998, "learning_rate": 9.431700371304963e-05, "loss": 3.1933, "step": 11498 }, { "epoch": 0.7138245701160841, "grad_norm": 0.49657729005003715, "learning_rate": 9.431533133588617e-05, "loss": 3.1431, "step": 11499 }, { "epoch": 0.713886647215842, "grad_norm": 0.5066673277654016, "learning_rate": 9.431365872751886e-05, "loss": 3.1178, "step": 11500 }, { "epoch": 0.7139487243156, "grad_norm": 0.49434065256993764, "learning_rate": 9.431198588795645e-05, "loss": 3.0275, "step": 11501 }, { "epoch": 0.7140108014153579, "grad_norm": 0.4800480962786868, "learning_rate": 9.431031281720764e-05, "loss": 3.0256, "step": 11502 }, { "epoch": 0.7140728785151158, "grad_norm": 0.4683868272895092, "learning_rate": 9.430863951528118e-05, "loss": 3.0968, "step": 11503 }, { "epoch": 0.7141349556148737, "grad_norm": 0.4715062013897887, "learning_rate": 9.430696598218578e-05, "loss": 3.0498, "step": 11504 }, { "epoch": 0.7141970327146315, "grad_norm": 0.4282215601212186, "learning_rate": 9.430529221793019e-05, "loss": 3.0929, "step": 11505 }, { "epoch": 0.7142591098143894, "grad_norm": 0.46592109917392804, "learning_rate": 9.430361822252312e-05, "loss": 3.01, "step": 11506 }, { "epoch": 0.7143211869141474, "grad_norm": 0.41938314256566095, "learning_rate": 9.430194399597334e-05, "loss": 3.1349, "step": 11507 }, { "epoch": 0.7143832640139053, "grad_norm": 0.4459576367360884, "learning_rate": 9.430026953828956e-05, "loss": 3.0518, "step": 11508 }, { "epoch": 0.7144453411136632, "grad_norm": 0.44058382166542065, "learning_rate": 9.429859484948052e-05, "loss": 3.1072, "step": 11509 }, { "epoch": 0.7145074182134211, "grad_norm": 0.42305283898881035, "learning_rate": 9.429691992955494e-05, "loss": 3.1451, "step": 11510 }, { "epoch": 0.7145694953131789, "grad_norm": 0.5202561104492165, "learning_rate": 9.429524477852158e-05, "loss": 2.9704, "step": 11511 }, { "epoch": 0.7146315724129368, "grad_norm": 0.4780313026022709, "learning_rate": 9.42935693963892e-05, "loss": 2.9613, "step": 11512 }, { "epoch": 0.7146936495126948, "grad_norm": 0.4446452311454074, "learning_rate": 9.429189378316649e-05, "loss": 3.1099, "step": 11513 }, { "epoch": 0.7147557266124527, "grad_norm": 0.48010405989541094, "learning_rate": 9.429021793886224e-05, "loss": 3.0804, "step": 11514 }, { "epoch": 0.7148178037122106, "grad_norm": 0.5349959251658619, "learning_rate": 9.428854186348514e-05, "loss": 3.0847, "step": 11515 }, { "epoch": 0.7148798808119685, "grad_norm": 0.47678026320727945, "learning_rate": 9.428686555704399e-05, "loss": 3.0115, "step": 11516 }, { "epoch": 0.7149419579117263, "grad_norm": 0.512728619174831, "learning_rate": 9.42851890195475e-05, "loss": 3.1492, "step": 11517 }, { "epoch": 0.7150040350114842, "grad_norm": 0.5873947840685356, "learning_rate": 9.428351225100444e-05, "loss": 3.0397, "step": 11518 }, { "epoch": 0.7150661121112422, "grad_norm": 0.47709212031316917, "learning_rate": 9.428183525142353e-05, "loss": 3.0301, "step": 11519 }, { "epoch": 0.7151281892110001, "grad_norm": 0.5048965068561061, "learning_rate": 9.428015802081354e-05, "loss": 3.1724, "step": 11520 }, { "epoch": 0.715190266310758, "grad_norm": 0.508245811458278, "learning_rate": 9.427848055918322e-05, "loss": 3.1363, "step": 11521 }, { "epoch": 0.7152523434105159, "grad_norm": 0.5408232859994379, "learning_rate": 9.427680286654132e-05, "loss": 3.0598, "step": 11522 }, { "epoch": 0.7153144205102737, "grad_norm": 0.4342392421140941, "learning_rate": 9.427512494289659e-05, "loss": 3.0348, "step": 11523 }, { "epoch": 0.7153764976100316, "grad_norm": 0.5264124806579726, "learning_rate": 9.42734467882578e-05, "loss": 3.0149, "step": 11524 }, { "epoch": 0.7154385747097896, "grad_norm": 0.520261638320536, "learning_rate": 9.427176840263367e-05, "loss": 3.0874, "step": 11525 }, { "epoch": 0.7155006518095475, "grad_norm": 0.4616919985797694, "learning_rate": 9.427008978603298e-05, "loss": 3.0575, "step": 11526 }, { "epoch": 0.7155627289093054, "grad_norm": 0.5170540801766449, "learning_rate": 9.426841093846448e-05, "loss": 3.0348, "step": 11527 }, { "epoch": 0.7156248060090633, "grad_norm": 0.5397725611782275, "learning_rate": 9.426673185993695e-05, "loss": 3.0993, "step": 11528 }, { "epoch": 0.7156868831088211, "grad_norm": 0.5131578956216614, "learning_rate": 9.426505255045912e-05, "loss": 3.0749, "step": 11529 }, { "epoch": 0.715748960208579, "grad_norm": 0.4943129348376758, "learning_rate": 9.426337301003977e-05, "loss": 3.0483, "step": 11530 }, { "epoch": 0.715811037308337, "grad_norm": 0.5506034455859797, "learning_rate": 9.426169323868766e-05, "loss": 3.0602, "step": 11531 }, { "epoch": 0.7158731144080949, "grad_norm": 0.5078170776160865, "learning_rate": 9.426001323641154e-05, "loss": 3.1223, "step": 11532 }, { "epoch": 0.7159351915078528, "grad_norm": 0.5937067317255242, "learning_rate": 9.42583330032202e-05, "loss": 3.0114, "step": 11533 }, { "epoch": 0.7159972686076107, "grad_norm": 0.5242207773001707, "learning_rate": 9.425665253912239e-05, "loss": 3.1374, "step": 11534 }, { "epoch": 0.7160593457073685, "grad_norm": 0.6367503433346431, "learning_rate": 9.425497184412688e-05, "loss": 3.1279, "step": 11535 }, { "epoch": 0.7161214228071264, "grad_norm": 0.5618494828723803, "learning_rate": 9.425329091824244e-05, "loss": 3.0593, "step": 11536 }, { "epoch": 0.7161834999068843, "grad_norm": 0.4836978331981642, "learning_rate": 9.425160976147782e-05, "loss": 3.1432, "step": 11537 }, { "epoch": 0.7162455770066423, "grad_norm": 0.6375100537049635, "learning_rate": 9.424992837384183e-05, "loss": 3.015, "step": 11538 }, { "epoch": 0.7163076541064002, "grad_norm": 0.5950271523314871, "learning_rate": 9.424824675534322e-05, "loss": 3.0277, "step": 11539 }, { "epoch": 0.7163697312061581, "grad_norm": 0.46642833190989846, "learning_rate": 9.424656490599075e-05, "loss": 3.0287, "step": 11540 }, { "epoch": 0.7164318083059159, "grad_norm": 0.49609926175637536, "learning_rate": 9.424488282579322e-05, "loss": 2.9306, "step": 11541 }, { "epoch": 0.7164938854056738, "grad_norm": 0.507155301240723, "learning_rate": 9.42432005147594e-05, "loss": 3.1459, "step": 11542 }, { "epoch": 0.7165559625054317, "grad_norm": 0.4858955389715782, "learning_rate": 9.424151797289806e-05, "loss": 3.069, "step": 11543 }, { "epoch": 0.7166180396051897, "grad_norm": 0.5169489365065432, "learning_rate": 9.423983520021798e-05, "loss": 3.0617, "step": 11544 }, { "epoch": 0.7166801167049476, "grad_norm": 0.6070877824575348, "learning_rate": 9.423815219672793e-05, "loss": 3.1429, "step": 11545 }, { "epoch": 0.7167421938047055, "grad_norm": 0.5907656053168115, "learning_rate": 9.423646896243672e-05, "loss": 3.1174, "step": 11546 }, { "epoch": 0.7168042709044633, "grad_norm": 0.4656060643794168, "learning_rate": 9.42347854973531e-05, "loss": 3.0333, "step": 11547 }, { "epoch": 0.7168663480042212, "grad_norm": 0.5429732233316038, "learning_rate": 9.423310180148586e-05, "loss": 3.0876, "step": 11548 }, { "epoch": 0.7169284251039791, "grad_norm": 0.5838510367462448, "learning_rate": 9.42314178748438e-05, "loss": 3.0103, "step": 11549 }, { "epoch": 0.7169905022037371, "grad_norm": 0.4810445638095497, "learning_rate": 9.422973371743571e-05, "loss": 3.1216, "step": 11550 }, { "epoch": 0.717052579303495, "grad_norm": 0.49046283884852204, "learning_rate": 9.422804932927034e-05, "loss": 3.057, "step": 11551 }, { "epoch": 0.7171146564032529, "grad_norm": 0.5663016913358937, "learning_rate": 9.422636471035651e-05, "loss": 3.0858, "step": 11552 }, { "epoch": 0.7171767335030107, "grad_norm": 0.5823593750851555, "learning_rate": 9.422467986070298e-05, "loss": 3.0097, "step": 11553 }, { "epoch": 0.7172388106027686, "grad_norm": 0.511580936423577, "learning_rate": 9.422299478031858e-05, "loss": 3.0802, "step": 11554 }, { "epoch": 0.7173008877025265, "grad_norm": 0.5361184429195527, "learning_rate": 9.422130946921209e-05, "loss": 3.0143, "step": 11555 }, { "epoch": 0.7173629648022845, "grad_norm": 0.5522013813944486, "learning_rate": 9.421962392739227e-05, "loss": 3.1046, "step": 11556 }, { "epoch": 0.7174250419020424, "grad_norm": 0.5534179827513591, "learning_rate": 9.421793815486794e-05, "loss": 3.0298, "step": 11557 }, { "epoch": 0.7174871190018003, "grad_norm": 0.5580038410348531, "learning_rate": 9.421625215164789e-05, "loss": 3.0154, "step": 11558 }, { "epoch": 0.7175491961015581, "grad_norm": 0.49702507367774174, "learning_rate": 9.421456591774094e-05, "loss": 3.1088, "step": 11559 }, { "epoch": 0.717611273201316, "grad_norm": 0.5530909813371346, "learning_rate": 9.421287945315585e-05, "loss": 3.0669, "step": 11560 }, { "epoch": 0.7176733503010739, "grad_norm": 0.48359093734588837, "learning_rate": 9.421119275790142e-05, "loss": 3.1058, "step": 11561 }, { "epoch": 0.7177354274008318, "grad_norm": 0.5495986959128893, "learning_rate": 9.420950583198649e-05, "loss": 3.0446, "step": 11562 }, { "epoch": 0.7177975045005898, "grad_norm": 0.48343584320534083, "learning_rate": 9.420781867541983e-05, "loss": 3.1896, "step": 11563 }, { "epoch": 0.7178595816003477, "grad_norm": 0.5068633085000795, "learning_rate": 9.420613128821026e-05, "loss": 3.0531, "step": 11564 }, { "epoch": 0.7179216587001055, "grad_norm": 0.49508824912830884, "learning_rate": 9.420444367036654e-05, "loss": 3.1153, "step": 11565 }, { "epoch": 0.7179837357998634, "grad_norm": 0.5644283883241419, "learning_rate": 9.420275582189753e-05, "loss": 3.1315, "step": 11566 }, { "epoch": 0.7180458128996213, "grad_norm": 0.5190572081772084, "learning_rate": 9.420106774281201e-05, "loss": 3.1293, "step": 11567 }, { "epoch": 0.7181078899993792, "grad_norm": 0.5307859188904245, "learning_rate": 9.419937943311878e-05, "loss": 3.132, "step": 11568 }, { "epoch": 0.7181699670991372, "grad_norm": 0.48889737738961, "learning_rate": 9.419769089282667e-05, "loss": 2.9893, "step": 11569 }, { "epoch": 0.7182320441988951, "grad_norm": 0.5060751318620614, "learning_rate": 9.419600212194446e-05, "loss": 3.0323, "step": 11570 }, { "epoch": 0.7182941212986529, "grad_norm": 0.5014623476614484, "learning_rate": 9.419431312048099e-05, "loss": 3.1486, "step": 11571 }, { "epoch": 0.7183561983984108, "grad_norm": 0.5209684927725041, "learning_rate": 9.419262388844507e-05, "loss": 3.0308, "step": 11572 }, { "epoch": 0.7184182754981687, "grad_norm": 0.487800773733815, "learning_rate": 9.419093442584547e-05, "loss": 3.0615, "step": 11573 }, { "epoch": 0.7184803525979266, "grad_norm": 0.5226143584975682, "learning_rate": 9.418924473269108e-05, "loss": 3.0306, "step": 11574 }, { "epoch": 0.7185424296976846, "grad_norm": 0.5506875519830949, "learning_rate": 9.418755480899064e-05, "loss": 3.0183, "step": 11575 }, { "epoch": 0.7186045067974425, "grad_norm": 0.49171975025697107, "learning_rate": 9.418586465475301e-05, "loss": 3.0547, "step": 11576 }, { "epoch": 0.7186665838972003, "grad_norm": 0.47915631435102524, "learning_rate": 9.4184174269987e-05, "loss": 3.121, "step": 11577 }, { "epoch": 0.7187286609969582, "grad_norm": 0.5133259409778859, "learning_rate": 9.418248365470144e-05, "loss": 2.9587, "step": 11578 }, { "epoch": 0.7187907380967161, "grad_norm": 0.5377717528791678, "learning_rate": 9.418079280890511e-05, "loss": 3.0357, "step": 11579 }, { "epoch": 0.718852815196474, "grad_norm": 0.559525359846323, "learning_rate": 9.417910173260687e-05, "loss": 3.121, "step": 11580 }, { "epoch": 0.718914892296232, "grad_norm": 0.5821528467747143, "learning_rate": 9.417741042581554e-05, "loss": 3.0569, "step": 11581 }, { "epoch": 0.7189769693959899, "grad_norm": 0.5655283998226897, "learning_rate": 9.417571888853994e-05, "loss": 3.1144, "step": 11582 }, { "epoch": 0.7190390464957477, "grad_norm": 0.6061164099205948, "learning_rate": 9.417402712078886e-05, "loss": 3.0115, "step": 11583 }, { "epoch": 0.7191011235955056, "grad_norm": 0.6165528958206286, "learning_rate": 9.417233512257118e-05, "loss": 3.0295, "step": 11584 }, { "epoch": 0.7191632006952635, "grad_norm": 0.5465940843881775, "learning_rate": 9.41706428938957e-05, "loss": 3.128, "step": 11585 }, { "epoch": 0.7192252777950214, "grad_norm": 0.5679270092765702, "learning_rate": 9.416895043477125e-05, "loss": 3.0279, "step": 11586 }, { "epoch": 0.7192873548947794, "grad_norm": 0.544022011956693, "learning_rate": 9.416725774520667e-05, "loss": 3.0493, "step": 11587 }, { "epoch": 0.7193494319945373, "grad_norm": 0.5507640159562002, "learning_rate": 9.416556482521079e-05, "loss": 3.1412, "step": 11588 }, { "epoch": 0.7194115090942951, "grad_norm": 0.5334039386813393, "learning_rate": 9.416387167479242e-05, "loss": 3.0136, "step": 11589 }, { "epoch": 0.719473586194053, "grad_norm": 0.5321809521651417, "learning_rate": 9.416217829396042e-05, "loss": 3.0492, "step": 11590 }, { "epoch": 0.7195356632938109, "grad_norm": 0.5424517358678117, "learning_rate": 9.416048468272361e-05, "loss": 3.1238, "step": 11591 }, { "epoch": 0.7195977403935688, "grad_norm": 0.5277314068126998, "learning_rate": 9.415879084109084e-05, "loss": 3.0659, "step": 11592 }, { "epoch": 0.7196598174933267, "grad_norm": 0.5736006439717454, "learning_rate": 9.415709676907092e-05, "loss": 3.1606, "step": 11593 }, { "epoch": 0.7197218945930847, "grad_norm": 0.5031251990276441, "learning_rate": 9.415540246667272e-05, "loss": 3.0832, "step": 11594 }, { "epoch": 0.7197839716928425, "grad_norm": 0.548778708191444, "learning_rate": 9.415370793390508e-05, "loss": 3.1045, "step": 11595 }, { "epoch": 0.7198460487926004, "grad_norm": 0.45985602541011417, "learning_rate": 9.41520131707768e-05, "loss": 3.0293, "step": 11596 }, { "epoch": 0.7199081258923583, "grad_norm": 0.5804462484976048, "learning_rate": 9.415031817729676e-05, "loss": 3.067, "step": 11597 }, { "epoch": 0.7199702029921162, "grad_norm": 0.5546134576944104, "learning_rate": 9.414862295347379e-05, "loss": 3.0784, "step": 11598 }, { "epoch": 0.7200322800918741, "grad_norm": 0.579320084768604, "learning_rate": 9.414692749931675e-05, "loss": 3.1291, "step": 11599 }, { "epoch": 0.7200943571916321, "grad_norm": 0.6422188864878526, "learning_rate": 9.414523181483447e-05, "loss": 3.0469, "step": 11600 }, { "epoch": 0.7201564342913899, "grad_norm": 0.5018535173789808, "learning_rate": 9.414353590003578e-05, "loss": 3.0331, "step": 11601 }, { "epoch": 0.7202185113911478, "grad_norm": 0.661620147817013, "learning_rate": 9.414183975492957e-05, "loss": 3.0981, "step": 11602 }, { "epoch": 0.7202805884909057, "grad_norm": 0.609424833558681, "learning_rate": 9.414014337952465e-05, "loss": 3.1279, "step": 11603 }, { "epoch": 0.7203426655906636, "grad_norm": 0.5056193603465164, "learning_rate": 9.413844677382989e-05, "loss": 3.0744, "step": 11604 }, { "epoch": 0.7204047426904215, "grad_norm": 0.612283465977443, "learning_rate": 9.413674993785413e-05, "loss": 3.1266, "step": 11605 }, { "epoch": 0.7204668197901795, "grad_norm": 0.5688607013827438, "learning_rate": 9.413505287160624e-05, "loss": 3.0757, "step": 11606 }, { "epoch": 0.7205288968899373, "grad_norm": 0.5567548400418436, "learning_rate": 9.413335557509507e-05, "loss": 3.1004, "step": 11607 }, { "epoch": 0.7205909739896952, "grad_norm": 0.5534719930505005, "learning_rate": 9.413165804832946e-05, "loss": 3.089, "step": 11608 }, { "epoch": 0.7206530510894531, "grad_norm": 0.5023157613263227, "learning_rate": 9.412996029131829e-05, "loss": 2.9879, "step": 11609 }, { "epoch": 0.720715128189211, "grad_norm": 0.48284562994117153, "learning_rate": 9.41282623040704e-05, "loss": 3.096, "step": 11610 }, { "epoch": 0.7207772052889689, "grad_norm": 0.5027090829505606, "learning_rate": 9.412656408659465e-05, "loss": 3.1165, "step": 11611 }, { "epoch": 0.7208392823887269, "grad_norm": 0.5067490813816742, "learning_rate": 9.412486563889991e-05, "loss": 3.1049, "step": 11612 }, { "epoch": 0.7209013594884847, "grad_norm": 0.48711409883877693, "learning_rate": 9.412316696099502e-05, "loss": 3.0967, "step": 11613 }, { "epoch": 0.7209634365882426, "grad_norm": 0.4831360229922428, "learning_rate": 9.412146805288886e-05, "loss": 3.0148, "step": 11614 }, { "epoch": 0.7210255136880005, "grad_norm": 0.5028761751908868, "learning_rate": 9.41197689145903e-05, "loss": 3.0492, "step": 11615 }, { "epoch": 0.7210875907877584, "grad_norm": 0.5833642531329319, "learning_rate": 9.411806954610818e-05, "loss": 2.9831, "step": 11616 }, { "epoch": 0.7211496678875163, "grad_norm": 0.5947776484354593, "learning_rate": 9.41163699474514e-05, "loss": 3.1025, "step": 11617 }, { "epoch": 0.7212117449872741, "grad_norm": 0.5965343242084241, "learning_rate": 9.41146701186288e-05, "loss": 3.0494, "step": 11618 }, { "epoch": 0.7212738220870321, "grad_norm": 0.49320361290395004, "learning_rate": 9.411297005964925e-05, "loss": 3.0014, "step": 11619 }, { "epoch": 0.72133589918679, "grad_norm": 0.5297717395964586, "learning_rate": 9.411126977052163e-05, "loss": 3.002, "step": 11620 }, { "epoch": 0.7213979762865479, "grad_norm": 0.5085442186369957, "learning_rate": 9.41095692512548e-05, "loss": 3.0588, "step": 11621 }, { "epoch": 0.7214600533863058, "grad_norm": 0.4738067150389286, "learning_rate": 9.410786850185767e-05, "loss": 2.9916, "step": 11622 }, { "epoch": 0.7215221304860637, "grad_norm": 0.5114425321366615, "learning_rate": 9.410616752233906e-05, "loss": 3.0635, "step": 11623 }, { "epoch": 0.7215842075858215, "grad_norm": 0.46271389190870915, "learning_rate": 9.410446631270787e-05, "loss": 3.0879, "step": 11624 }, { "epoch": 0.7216462846855795, "grad_norm": 0.4593087693562652, "learning_rate": 9.410276487297296e-05, "loss": 3.1255, "step": 11625 }, { "epoch": 0.7217083617853374, "grad_norm": 0.5229208374870303, "learning_rate": 9.410106320314323e-05, "loss": 3.1116, "step": 11626 }, { "epoch": 0.7217704388850953, "grad_norm": 0.4655470616442208, "learning_rate": 9.409936130322754e-05, "loss": 3.0259, "step": 11627 }, { "epoch": 0.7218325159848532, "grad_norm": 0.45386019619180473, "learning_rate": 9.409765917323479e-05, "loss": 3.09, "step": 11628 }, { "epoch": 0.7218945930846111, "grad_norm": 0.5326387821375898, "learning_rate": 9.409595681317383e-05, "loss": 3.1339, "step": 11629 }, { "epoch": 0.7219566701843689, "grad_norm": 0.5566640480868869, "learning_rate": 9.409425422305358e-05, "loss": 2.9779, "step": 11630 }, { "epoch": 0.7220187472841268, "grad_norm": 0.4650416393761373, "learning_rate": 9.409255140288289e-05, "loss": 3.1226, "step": 11631 }, { "epoch": 0.7220808243838848, "grad_norm": 0.5524380162367538, "learning_rate": 9.409084835267065e-05, "loss": 3.1341, "step": 11632 }, { "epoch": 0.7221429014836427, "grad_norm": 0.5043749736232822, "learning_rate": 9.408914507242577e-05, "loss": 3.1245, "step": 11633 }, { "epoch": 0.7222049785834006, "grad_norm": 0.5045613015948688, "learning_rate": 9.40874415621571e-05, "loss": 3.1946, "step": 11634 }, { "epoch": 0.7222670556831585, "grad_norm": 0.5430635328376318, "learning_rate": 9.408573782187355e-05, "loss": 3.0951, "step": 11635 }, { "epoch": 0.7223291327829163, "grad_norm": 0.557523248512757, "learning_rate": 9.4084033851584e-05, "loss": 3.1202, "step": 11636 }, { "epoch": 0.7223912098826742, "grad_norm": 0.5563065446552663, "learning_rate": 9.408232965129735e-05, "loss": 3.0578, "step": 11637 }, { "epoch": 0.7224532869824322, "grad_norm": 0.5376722126469956, "learning_rate": 9.408062522102246e-05, "loss": 3.0643, "step": 11638 }, { "epoch": 0.7225153640821901, "grad_norm": 0.4762040024919621, "learning_rate": 9.407892056076828e-05, "loss": 3.0929, "step": 11639 }, { "epoch": 0.722577441181948, "grad_norm": 0.581925632187407, "learning_rate": 9.407721567054367e-05, "loss": 3.1434, "step": 11640 }, { "epoch": 0.7226395182817059, "grad_norm": 0.48686009921678064, "learning_rate": 9.407551055035751e-05, "loss": 3.0286, "step": 11641 }, { "epoch": 0.7227015953814637, "grad_norm": 0.5185205480684696, "learning_rate": 9.407380520021871e-05, "loss": 3.0849, "step": 11642 }, { "epoch": 0.7227636724812216, "grad_norm": 0.586007889652195, "learning_rate": 9.407209962013617e-05, "loss": 3.0548, "step": 11643 }, { "epoch": 0.7228257495809796, "grad_norm": 0.5278010328367303, "learning_rate": 9.407039381011879e-05, "loss": 3.0676, "step": 11644 }, { "epoch": 0.7228878266807375, "grad_norm": 0.5563570873176932, "learning_rate": 9.406868777017548e-05, "loss": 3.0295, "step": 11645 }, { "epoch": 0.7229499037804954, "grad_norm": 0.603842908885812, "learning_rate": 9.406698150031512e-05, "loss": 3.0894, "step": 11646 }, { "epoch": 0.7230119808802533, "grad_norm": 0.5602678142627924, "learning_rate": 9.40652750005466e-05, "loss": 3.13, "step": 11647 }, { "epoch": 0.7230740579800111, "grad_norm": 0.4618737362443001, "learning_rate": 9.406356827087888e-05, "loss": 3.1139, "step": 11648 }, { "epoch": 0.723136135079769, "grad_norm": 0.5842190274250426, "learning_rate": 9.40618613113208e-05, "loss": 3.173, "step": 11649 }, { "epoch": 0.723198212179527, "grad_norm": 0.5909178850296508, "learning_rate": 9.406015412188129e-05, "loss": 3.0331, "step": 11650 }, { "epoch": 0.7232602892792849, "grad_norm": 0.5136412404719604, "learning_rate": 9.405844670256928e-05, "loss": 3.1572, "step": 11651 }, { "epoch": 0.7233223663790428, "grad_norm": 0.5724343061027969, "learning_rate": 9.405673905339365e-05, "loss": 3.116, "step": 11652 }, { "epoch": 0.7233844434788007, "grad_norm": 0.5257078439787295, "learning_rate": 9.405503117436332e-05, "loss": 2.9549, "step": 11653 }, { "epoch": 0.7234465205785585, "grad_norm": 0.575108286778301, "learning_rate": 9.405332306548719e-05, "loss": 3.0651, "step": 11654 }, { "epoch": 0.7235085976783164, "grad_norm": 0.5289843153710557, "learning_rate": 9.405161472677419e-05, "loss": 3.1504, "step": 11655 }, { "epoch": 0.7235706747780744, "grad_norm": 0.5520442792886442, "learning_rate": 9.404990615823321e-05, "loss": 3.1637, "step": 11656 }, { "epoch": 0.7236327518778323, "grad_norm": 0.6118305810502567, "learning_rate": 9.404819735987319e-05, "loss": 2.9467, "step": 11657 }, { "epoch": 0.7236948289775902, "grad_norm": 0.5243151694027598, "learning_rate": 9.404648833170302e-05, "loss": 3.1005, "step": 11658 }, { "epoch": 0.7237569060773481, "grad_norm": 0.599266053363107, "learning_rate": 9.404477907373163e-05, "loss": 3.0823, "step": 11659 }, { "epoch": 0.7238189831771059, "grad_norm": 0.5910092642795447, "learning_rate": 9.404306958596794e-05, "loss": 3.1119, "step": 11660 }, { "epoch": 0.7238810602768638, "grad_norm": 0.5483548713452401, "learning_rate": 9.404135986842086e-05, "loss": 3.155, "step": 11661 }, { "epoch": 0.7239431373766217, "grad_norm": 0.531364204248614, "learning_rate": 9.40396499210993e-05, "loss": 2.9797, "step": 11662 }, { "epoch": 0.7240052144763797, "grad_norm": 0.5082266952361777, "learning_rate": 9.403793974401222e-05, "loss": 3.0671, "step": 11663 }, { "epoch": 0.7240672915761376, "grad_norm": 0.5647532843063251, "learning_rate": 9.40362293371685e-05, "loss": 3.0664, "step": 11664 }, { "epoch": 0.7241293686758955, "grad_norm": 0.543836106684436, "learning_rate": 9.403451870057709e-05, "loss": 3.0499, "step": 11665 }, { "epoch": 0.7241914457756533, "grad_norm": 0.5734377629749828, "learning_rate": 9.403280783424691e-05, "loss": 3.1104, "step": 11666 }, { "epoch": 0.7242535228754112, "grad_norm": 0.4869649157419365, "learning_rate": 9.403109673818687e-05, "loss": 3.115, "step": 11667 }, { "epoch": 0.7243155999751691, "grad_norm": 0.5694273781997372, "learning_rate": 9.402938541240592e-05, "loss": 3.056, "step": 11668 }, { "epoch": 0.7243776770749271, "grad_norm": 0.5456395226170286, "learning_rate": 9.402767385691298e-05, "loss": 3.0476, "step": 11669 }, { "epoch": 0.724439754174685, "grad_norm": 0.5489922817622248, "learning_rate": 9.402596207171697e-05, "loss": 3.074, "step": 11670 }, { "epoch": 0.7245018312744429, "grad_norm": 0.5024621957346357, "learning_rate": 9.402425005682683e-05, "loss": 3.0826, "step": 11671 }, { "epoch": 0.7245639083742007, "grad_norm": 0.47108021743594647, "learning_rate": 9.402253781225148e-05, "loss": 3.0988, "step": 11672 }, { "epoch": 0.7246259854739586, "grad_norm": 0.5291668778016078, "learning_rate": 9.402082533799987e-05, "loss": 3.1228, "step": 11673 }, { "epoch": 0.7246880625737165, "grad_norm": 0.5164329960485969, "learning_rate": 9.401911263408094e-05, "loss": 2.9953, "step": 11674 }, { "epoch": 0.7247501396734745, "grad_norm": 0.4252597344745958, "learning_rate": 9.40173997005036e-05, "loss": 3.0946, "step": 11675 }, { "epoch": 0.7248122167732324, "grad_norm": 0.5128218431104009, "learning_rate": 9.401568653727682e-05, "loss": 3.1692, "step": 11676 }, { "epoch": 0.7248742938729903, "grad_norm": 0.5025356489792795, "learning_rate": 9.401397314440949e-05, "loss": 3.2107, "step": 11677 }, { "epoch": 0.7249363709727481, "grad_norm": 0.487605565818904, "learning_rate": 9.40122595219106e-05, "loss": 3.0392, "step": 11678 }, { "epoch": 0.724998448072506, "grad_norm": 0.4601535758182041, "learning_rate": 9.401054566978905e-05, "loss": 3.1297, "step": 11679 }, { "epoch": 0.7250605251722639, "grad_norm": 0.5792770505819714, "learning_rate": 9.40088315880538e-05, "loss": 3.0588, "step": 11680 }, { "epoch": 0.7251226022720219, "grad_norm": 0.5877118773580596, "learning_rate": 9.400711727671381e-05, "loss": 3.077, "step": 11681 }, { "epoch": 0.7251846793717798, "grad_norm": 0.45139736469168845, "learning_rate": 9.400540273577799e-05, "loss": 3.1026, "step": 11682 }, { "epoch": 0.7252467564715377, "grad_norm": 0.5418930100436642, "learning_rate": 9.400368796525532e-05, "loss": 3.0069, "step": 11683 }, { "epoch": 0.7253088335712955, "grad_norm": 0.48208118099742175, "learning_rate": 9.40019729651547e-05, "loss": 3.0165, "step": 11684 }, { "epoch": 0.7253709106710534, "grad_norm": 0.4840172554035695, "learning_rate": 9.400025773548512e-05, "loss": 3.0414, "step": 11685 }, { "epoch": 0.7254329877708113, "grad_norm": 0.49479740206422673, "learning_rate": 9.39985422762555e-05, "loss": 3.0657, "step": 11686 }, { "epoch": 0.7254950648705693, "grad_norm": 0.5270642140153781, "learning_rate": 9.399682658747484e-05, "loss": 3.1426, "step": 11687 }, { "epoch": 0.7255571419703272, "grad_norm": 0.4804200248512664, "learning_rate": 9.399511066915204e-05, "loss": 3.0709, "step": 11688 }, { "epoch": 0.7256192190700851, "grad_norm": 0.5173611168004433, "learning_rate": 9.399339452129605e-05, "loss": 3.0492, "step": 11689 }, { "epoch": 0.7256812961698429, "grad_norm": 0.5400403041721853, "learning_rate": 9.399167814391585e-05, "loss": 3.088, "step": 11690 }, { "epoch": 0.7257433732696008, "grad_norm": 0.5133146777365182, "learning_rate": 9.398996153702037e-05, "loss": 3.0068, "step": 11691 }, { "epoch": 0.7258054503693587, "grad_norm": 0.5851290911698738, "learning_rate": 9.398824470061861e-05, "loss": 3.0894, "step": 11692 }, { "epoch": 0.7258675274691166, "grad_norm": 0.48936497776686827, "learning_rate": 9.398652763471949e-05, "loss": 3.1102, "step": 11693 }, { "epoch": 0.7259296045688746, "grad_norm": 0.5282893145311511, "learning_rate": 9.398481033933197e-05, "loss": 2.9489, "step": 11694 }, { "epoch": 0.7259916816686325, "grad_norm": 0.4229736328125, "learning_rate": 9.398309281446503e-05, "loss": 3.0575, "step": 11695 }, { "epoch": 0.7260537587683903, "grad_norm": 0.5126750542407622, "learning_rate": 9.39813750601276e-05, "loss": 3.0914, "step": 11696 }, { "epoch": 0.7261158358681482, "grad_norm": 0.38946194117447497, "learning_rate": 9.397965707632868e-05, "loss": 2.9594, "step": 11697 }, { "epoch": 0.7261779129679061, "grad_norm": 0.4945948175732229, "learning_rate": 9.397793886307719e-05, "loss": 2.9928, "step": 11698 }, { "epoch": 0.726239990067664, "grad_norm": 0.49270387839735547, "learning_rate": 9.397622042038212e-05, "loss": 3.1068, "step": 11699 }, { "epoch": 0.726302067167422, "grad_norm": 0.5107341349971687, "learning_rate": 9.397450174825245e-05, "loss": 2.9779, "step": 11700 }, { "epoch": 0.7263641442671799, "grad_norm": 0.4220484271150006, "learning_rate": 9.397278284669712e-05, "loss": 3.0835, "step": 11701 }, { "epoch": 0.7264262213669377, "grad_norm": 0.5422613077193809, "learning_rate": 9.39710637157251e-05, "loss": 3.1164, "step": 11702 }, { "epoch": 0.7264882984666956, "grad_norm": 0.4639446796137665, "learning_rate": 9.396934435534538e-05, "loss": 3.0108, "step": 11703 }, { "epoch": 0.7265503755664535, "grad_norm": 0.48193915923363173, "learning_rate": 9.396762476556691e-05, "loss": 3.0714, "step": 11704 }, { "epoch": 0.7266124526662114, "grad_norm": 0.5147471903205066, "learning_rate": 9.396590494639867e-05, "loss": 3.116, "step": 11705 }, { "epoch": 0.7266745297659694, "grad_norm": 0.48025116994894995, "learning_rate": 9.396418489784963e-05, "loss": 3.0301, "step": 11706 }, { "epoch": 0.7267366068657273, "grad_norm": 0.5596915974086297, "learning_rate": 9.396246461992876e-05, "loss": 3.0525, "step": 11707 }, { "epoch": 0.7267986839654851, "grad_norm": 0.5214774599224044, "learning_rate": 9.396074411264504e-05, "loss": 3.0678, "step": 11708 }, { "epoch": 0.726860761065243, "grad_norm": 0.4628958078086202, "learning_rate": 9.395902337600746e-05, "loss": 3.0342, "step": 11709 }, { "epoch": 0.7269228381650009, "grad_norm": 0.5540392405301505, "learning_rate": 9.395730241002495e-05, "loss": 3.1043, "step": 11710 }, { "epoch": 0.7269849152647588, "grad_norm": 0.5057618036028844, "learning_rate": 9.395558121470656e-05, "loss": 3.0128, "step": 11711 }, { "epoch": 0.7270469923645168, "grad_norm": 0.46793835625634483, "learning_rate": 9.395385979006122e-05, "loss": 3.0217, "step": 11712 }, { "epoch": 0.7271090694642747, "grad_norm": 0.4620789602529648, "learning_rate": 9.395213813609791e-05, "loss": 3.1533, "step": 11713 }, { "epoch": 0.7271711465640325, "grad_norm": 0.5283697249738302, "learning_rate": 9.395041625282563e-05, "loss": 3.0736, "step": 11714 }, { "epoch": 0.7272332236637904, "grad_norm": 0.5312027910237497, "learning_rate": 9.394869414025338e-05, "loss": 3.0787, "step": 11715 }, { "epoch": 0.7272953007635483, "grad_norm": 0.44326534198323136, "learning_rate": 9.39469717983901e-05, "loss": 3.0056, "step": 11716 }, { "epoch": 0.7273573778633062, "grad_norm": 0.5138756525494388, "learning_rate": 9.394524922724481e-05, "loss": 3.0913, "step": 11717 }, { "epoch": 0.7274194549630641, "grad_norm": 0.596246840190712, "learning_rate": 9.394352642682649e-05, "loss": 3.0587, "step": 11718 }, { "epoch": 0.7274815320628221, "grad_norm": 0.5793858774227848, "learning_rate": 9.394180339714411e-05, "loss": 3.125, "step": 11719 }, { "epoch": 0.7275436091625799, "grad_norm": 0.5015782838043211, "learning_rate": 9.39400801382067e-05, "loss": 3.0068, "step": 11720 }, { "epoch": 0.7276056862623378, "grad_norm": 0.4518248716489801, "learning_rate": 9.39383566500232e-05, "loss": 3.0808, "step": 11721 }, { "epoch": 0.7276677633620957, "grad_norm": 0.6183899621874888, "learning_rate": 9.393663293260264e-05, "loss": 3.1083, "step": 11722 }, { "epoch": 0.7277298404618536, "grad_norm": 0.5593228949888324, "learning_rate": 9.3934908985954e-05, "loss": 3.0826, "step": 11723 }, { "epoch": 0.7277919175616115, "grad_norm": 0.5013154131285478, "learning_rate": 9.393318481008626e-05, "loss": 2.9389, "step": 11724 }, { "epoch": 0.7278539946613695, "grad_norm": 0.5719911353167956, "learning_rate": 9.393146040500845e-05, "loss": 2.9592, "step": 11725 }, { "epoch": 0.7279160717611273, "grad_norm": 0.4837029008830026, "learning_rate": 9.392973577072953e-05, "loss": 3.0624, "step": 11726 }, { "epoch": 0.7279781488608852, "grad_norm": 0.49787683015008594, "learning_rate": 9.392801090725852e-05, "loss": 3.1728, "step": 11727 }, { "epoch": 0.7280402259606431, "grad_norm": 0.4693405405137124, "learning_rate": 9.392628581460441e-05, "loss": 3.0353, "step": 11728 }, { "epoch": 0.728102303060401, "grad_norm": 0.5251980964130062, "learning_rate": 9.392456049277622e-05, "loss": 3.0606, "step": 11729 }, { "epoch": 0.7281643801601589, "grad_norm": 0.5166911460348218, "learning_rate": 9.392283494178292e-05, "loss": 3.0897, "step": 11730 }, { "epoch": 0.7282264572599169, "grad_norm": 0.5149121414165171, "learning_rate": 9.392110916163352e-05, "loss": 3.0609, "step": 11731 }, { "epoch": 0.7282885343596747, "grad_norm": 0.506639299118907, "learning_rate": 9.391938315233705e-05, "loss": 3.0196, "step": 11732 }, { "epoch": 0.7283506114594326, "grad_norm": 0.5658826315036376, "learning_rate": 9.391765691390248e-05, "loss": 2.9774, "step": 11733 }, { "epoch": 0.7284126885591905, "grad_norm": 0.5347297249580035, "learning_rate": 9.391593044633886e-05, "loss": 3.108, "step": 11734 }, { "epoch": 0.7284747656589484, "grad_norm": 0.49881466792870716, "learning_rate": 9.391420374965514e-05, "loss": 3.1286, "step": 11735 }, { "epoch": 0.7285368427587063, "grad_norm": 0.4863383127394032, "learning_rate": 9.391247682386038e-05, "loss": 3.0417, "step": 11736 }, { "epoch": 0.7285989198584643, "grad_norm": 0.5001598639031425, "learning_rate": 9.391074966896354e-05, "loss": 3.1141, "step": 11737 }, { "epoch": 0.7286609969582221, "grad_norm": 0.551759351880112, "learning_rate": 9.390902228497368e-05, "loss": 3.1116, "step": 11738 }, { "epoch": 0.72872307405798, "grad_norm": 0.48859092047945263, "learning_rate": 9.390729467189978e-05, "loss": 3.053, "step": 11739 }, { "epoch": 0.7287851511577379, "grad_norm": 0.5508353024196009, "learning_rate": 9.390556682975088e-05, "loss": 3.0321, "step": 11740 }, { "epoch": 0.7288472282574958, "grad_norm": 0.5222174531665903, "learning_rate": 9.390383875853597e-05, "loss": 3.0827, "step": 11741 }, { "epoch": 0.7289093053572537, "grad_norm": 0.5056115504434088, "learning_rate": 9.390211045826409e-05, "loss": 2.9719, "step": 11742 }, { "epoch": 0.7289713824570117, "grad_norm": 0.5244359414990476, "learning_rate": 9.390038192894422e-05, "loss": 3.0716, "step": 11743 }, { "epoch": 0.7290334595567695, "grad_norm": 0.555302909849839, "learning_rate": 9.389865317058541e-05, "loss": 3.1915, "step": 11744 }, { "epoch": 0.7290955366565274, "grad_norm": 0.5555489526462235, "learning_rate": 9.389692418319667e-05, "loss": 3.1151, "step": 11745 }, { "epoch": 0.7291576137562853, "grad_norm": 0.4648773117131039, "learning_rate": 9.389519496678701e-05, "loss": 3.1507, "step": 11746 }, { "epoch": 0.7292196908560432, "grad_norm": 0.608986730858514, "learning_rate": 9.389346552136548e-05, "loss": 3.1429, "step": 11747 }, { "epoch": 0.7292817679558011, "grad_norm": 0.4753340224309655, "learning_rate": 9.389173584694106e-05, "loss": 2.9732, "step": 11748 }, { "epoch": 0.729343845055559, "grad_norm": 0.5789029068534989, "learning_rate": 9.389000594352282e-05, "loss": 3.0747, "step": 11749 }, { "epoch": 0.7294059221553169, "grad_norm": 0.5598759107963442, "learning_rate": 9.388827581111976e-05, "loss": 3.0044, "step": 11750 }, { "epoch": 0.7294679992550748, "grad_norm": 0.4702074757411687, "learning_rate": 9.388654544974092e-05, "loss": 3.0453, "step": 11751 }, { "epoch": 0.7295300763548327, "grad_norm": 0.5166619307034328, "learning_rate": 9.38848148593953e-05, "loss": 3.1061, "step": 11752 }, { "epoch": 0.7295921534545906, "grad_norm": 0.5099017666662105, "learning_rate": 9.388308404009197e-05, "loss": 3.0565, "step": 11753 }, { "epoch": 0.7296542305543485, "grad_norm": 0.5200183850486194, "learning_rate": 9.38813529918399e-05, "loss": 3.0354, "step": 11754 }, { "epoch": 0.7297163076541064, "grad_norm": 0.45426955858468837, "learning_rate": 9.387962171464819e-05, "loss": 3.0904, "step": 11755 }, { "epoch": 0.7297783847538643, "grad_norm": 0.5056732895923908, "learning_rate": 9.387789020852584e-05, "loss": 3.071, "step": 11756 }, { "epoch": 0.7298404618536222, "grad_norm": 0.49431195511777654, "learning_rate": 9.387615847348188e-05, "loss": 3.0468, "step": 11757 }, { "epoch": 0.7299025389533801, "grad_norm": 0.5466170928979858, "learning_rate": 9.387442650952535e-05, "loss": 3.0381, "step": 11758 }, { "epoch": 0.729964616053138, "grad_norm": 0.5076960797005803, "learning_rate": 9.387269431666528e-05, "loss": 3.039, "step": 11759 }, { "epoch": 0.7300266931528959, "grad_norm": 0.5687161666113021, "learning_rate": 9.387096189491072e-05, "loss": 3.1068, "step": 11760 }, { "epoch": 0.7300887702526538, "grad_norm": 0.5900690520013405, "learning_rate": 9.386922924427071e-05, "loss": 3.1039, "step": 11761 }, { "epoch": 0.7301508473524116, "grad_norm": 0.5282804858414574, "learning_rate": 9.386749636475427e-05, "loss": 3.0253, "step": 11762 }, { "epoch": 0.7302129244521696, "grad_norm": 0.5109800000261381, "learning_rate": 9.386576325637044e-05, "loss": 3.0444, "step": 11763 }, { "epoch": 0.7302750015519275, "grad_norm": 0.4742688813962873, "learning_rate": 9.38640299191283e-05, "loss": 3.0585, "step": 11764 }, { "epoch": 0.7303370786516854, "grad_norm": 0.4859313725272227, "learning_rate": 9.386229635303685e-05, "loss": 3.0947, "step": 11765 }, { "epoch": 0.7303991557514433, "grad_norm": 0.4960778601797498, "learning_rate": 9.386056255810516e-05, "loss": 3.1288, "step": 11766 }, { "epoch": 0.7304612328512012, "grad_norm": 0.516209184728065, "learning_rate": 9.385882853434227e-05, "loss": 3.1053, "step": 11767 }, { "epoch": 0.730523309950959, "grad_norm": 0.4943029566764961, "learning_rate": 9.385709428175721e-05, "loss": 3.0479, "step": 11768 }, { "epoch": 0.730585387050717, "grad_norm": 0.49523679052372516, "learning_rate": 9.385535980035905e-05, "loss": 3.0088, "step": 11769 }, { "epoch": 0.7306474641504749, "grad_norm": 0.49997264071951786, "learning_rate": 9.385362509015684e-05, "loss": 3.0613, "step": 11770 }, { "epoch": 0.7307095412502328, "grad_norm": 0.5426846042016739, "learning_rate": 9.385189015115962e-05, "loss": 3.0413, "step": 11771 }, { "epoch": 0.7307716183499907, "grad_norm": 0.4569017275651719, "learning_rate": 9.385015498337644e-05, "loss": 3.0133, "step": 11772 }, { "epoch": 0.7308336954497486, "grad_norm": 0.49033629737127127, "learning_rate": 9.384841958681635e-05, "loss": 2.9963, "step": 11773 }, { "epoch": 0.7308957725495064, "grad_norm": 0.5104815253745412, "learning_rate": 9.384668396148841e-05, "loss": 3.039, "step": 11774 }, { "epoch": 0.7309578496492644, "grad_norm": 0.5062043769906428, "learning_rate": 9.384494810740169e-05, "loss": 3.0118, "step": 11775 }, { "epoch": 0.7310199267490223, "grad_norm": 0.4821440016137522, "learning_rate": 9.384321202456523e-05, "loss": 3.04, "step": 11776 }, { "epoch": 0.7310820038487802, "grad_norm": 0.5035519500248461, "learning_rate": 9.384147571298807e-05, "loss": 3.1156, "step": 11777 }, { "epoch": 0.7311440809485381, "grad_norm": 0.57826865189978, "learning_rate": 9.38397391726793e-05, "loss": 3.0848, "step": 11778 }, { "epoch": 0.731206158048296, "grad_norm": 0.5124647512176967, "learning_rate": 9.383800240364798e-05, "loss": 3.0143, "step": 11779 }, { "epoch": 0.7312682351480538, "grad_norm": 0.502408870907768, "learning_rate": 9.383626540590315e-05, "loss": 3.0814, "step": 11780 }, { "epoch": 0.7313303122478118, "grad_norm": 0.5561954257210477, "learning_rate": 9.383452817945388e-05, "loss": 3.0429, "step": 11781 }, { "epoch": 0.7313923893475697, "grad_norm": 0.40479943925532225, "learning_rate": 9.383279072430924e-05, "loss": 3.1142, "step": 11782 }, { "epoch": 0.7314544664473276, "grad_norm": 0.4829873388804848, "learning_rate": 9.383105304047828e-05, "loss": 3.0473, "step": 11783 }, { "epoch": 0.7315165435470855, "grad_norm": 0.49069585774443675, "learning_rate": 9.382931512797009e-05, "loss": 3.1752, "step": 11784 }, { "epoch": 0.7315786206468434, "grad_norm": 0.5600619993630108, "learning_rate": 9.38275769867937e-05, "loss": 3.086, "step": 11785 }, { "epoch": 0.7316406977466012, "grad_norm": 0.47325277910963176, "learning_rate": 9.382583861695822e-05, "loss": 3.0995, "step": 11786 }, { "epoch": 0.7317027748463591, "grad_norm": 0.4995948371830041, "learning_rate": 9.38241000184727e-05, "loss": 2.9735, "step": 11787 }, { "epoch": 0.7317648519461171, "grad_norm": 0.5367294632510176, "learning_rate": 9.382236119134622e-05, "loss": 3.1298, "step": 11788 }, { "epoch": 0.731826929045875, "grad_norm": 0.5003806989461048, "learning_rate": 9.382062213558782e-05, "loss": 3.0232, "step": 11789 }, { "epoch": 0.7318890061456329, "grad_norm": 0.48704982353777443, "learning_rate": 9.381888285120661e-05, "loss": 3.0259, "step": 11790 }, { "epoch": 0.7319510832453908, "grad_norm": 0.537789652647757, "learning_rate": 9.381714333821166e-05, "loss": 3.0339, "step": 11791 }, { "epoch": 0.7320131603451486, "grad_norm": 0.47021761662664857, "learning_rate": 9.381540359661202e-05, "loss": 2.9825, "step": 11792 }, { "epoch": 0.7320752374449065, "grad_norm": 0.46066280843880564, "learning_rate": 9.381366362641679e-05, "loss": 3.0253, "step": 11793 }, { "epoch": 0.7321373145446645, "grad_norm": 0.5456209791282202, "learning_rate": 9.381192342763504e-05, "loss": 3.08, "step": 11794 }, { "epoch": 0.7321993916444224, "grad_norm": 0.5618825543652148, "learning_rate": 9.381018300027585e-05, "loss": 3.0921, "step": 11795 }, { "epoch": 0.7322614687441803, "grad_norm": 0.5290871799523633, "learning_rate": 9.380844234434829e-05, "loss": 3.0359, "step": 11796 }, { "epoch": 0.7323235458439382, "grad_norm": 0.5340303145922841, "learning_rate": 9.380670145986147e-05, "loss": 2.9992, "step": 11797 }, { "epoch": 0.732385622943696, "grad_norm": 0.4987789682470314, "learning_rate": 9.380496034682444e-05, "loss": 3.0305, "step": 11798 }, { "epoch": 0.7324477000434539, "grad_norm": 0.6247503258779469, "learning_rate": 9.380321900524631e-05, "loss": 2.9701, "step": 11799 }, { "epoch": 0.7325097771432119, "grad_norm": 0.5278892521362609, "learning_rate": 9.380147743513614e-05, "loss": 3.0091, "step": 11800 }, { "epoch": 0.7325718542429698, "grad_norm": 0.48293757192088593, "learning_rate": 9.379973563650304e-05, "loss": 3.0414, "step": 11801 }, { "epoch": 0.7326339313427277, "grad_norm": 0.5834746416689445, "learning_rate": 9.379799360935608e-05, "loss": 3.095, "step": 11802 }, { "epoch": 0.7326960084424856, "grad_norm": 0.5122836831106667, "learning_rate": 9.379625135370435e-05, "loss": 3.0309, "step": 11803 }, { "epoch": 0.7327580855422434, "grad_norm": 0.5158969855387389, "learning_rate": 9.379450886955696e-05, "loss": 3.0521, "step": 11804 }, { "epoch": 0.7328201626420013, "grad_norm": 0.4726053163182204, "learning_rate": 9.379276615692296e-05, "loss": 3.0385, "step": 11805 }, { "epoch": 0.7328822397417593, "grad_norm": 0.5298180073340144, "learning_rate": 9.379102321581149e-05, "loss": 3.1401, "step": 11806 }, { "epoch": 0.7329443168415172, "grad_norm": 0.5444552430095138, "learning_rate": 9.378928004623163e-05, "loss": 3.0422, "step": 11807 }, { "epoch": 0.7330063939412751, "grad_norm": 0.468181901638835, "learning_rate": 9.378753664819244e-05, "loss": 3.1002, "step": 11808 }, { "epoch": 0.733068471041033, "grad_norm": 0.504422714169837, "learning_rate": 9.378579302170304e-05, "loss": 3.0352, "step": 11809 }, { "epoch": 0.7331305481407908, "grad_norm": 0.5550461536410622, "learning_rate": 9.378404916677255e-05, "loss": 3.1016, "step": 11810 }, { "epoch": 0.7331926252405487, "grad_norm": 0.5731067139964623, "learning_rate": 9.378230508341004e-05, "loss": 3.0031, "step": 11811 }, { "epoch": 0.7332547023403067, "grad_norm": 0.4706677467062061, "learning_rate": 9.378056077162462e-05, "loss": 3.1133, "step": 11812 }, { "epoch": 0.7333167794400646, "grad_norm": 0.5193079131370658, "learning_rate": 9.377881623142537e-05, "loss": 3.009, "step": 11813 }, { "epoch": 0.7333788565398225, "grad_norm": 0.47485211918878056, "learning_rate": 9.377707146282143e-05, "loss": 3.1184, "step": 11814 }, { "epoch": 0.7334409336395804, "grad_norm": 0.49100077325154035, "learning_rate": 9.377532646582185e-05, "loss": 3.0485, "step": 11815 }, { "epoch": 0.7335030107393382, "grad_norm": 0.4928219052072211, "learning_rate": 9.37735812404358e-05, "loss": 3.0194, "step": 11816 }, { "epoch": 0.7335650878390961, "grad_norm": 0.4696853524448945, "learning_rate": 9.377183578667233e-05, "loss": 2.9246, "step": 11817 }, { "epoch": 0.733627164938854, "grad_norm": 0.4876461629120705, "learning_rate": 9.377009010454056e-05, "loss": 2.9975, "step": 11818 }, { "epoch": 0.733689242038612, "grad_norm": 0.4870161069707099, "learning_rate": 9.376834419404963e-05, "loss": 2.9737, "step": 11819 }, { "epoch": 0.7337513191383699, "grad_norm": 0.520993891445893, "learning_rate": 9.376659805520861e-05, "loss": 3.0651, "step": 11820 }, { "epoch": 0.7338133962381278, "grad_norm": 0.47413223520657477, "learning_rate": 9.376485168802663e-05, "loss": 3.0618, "step": 11821 }, { "epoch": 0.7338754733378856, "grad_norm": 0.4970506165804836, "learning_rate": 9.376310509251278e-05, "loss": 3.0474, "step": 11822 }, { "epoch": 0.7339375504376435, "grad_norm": 0.48987012606583635, "learning_rate": 9.376135826867619e-05, "loss": 3.0896, "step": 11823 }, { "epoch": 0.7339996275374014, "grad_norm": 0.4333749167346242, "learning_rate": 9.375961121652597e-05, "loss": 3.0138, "step": 11824 }, { "epoch": 0.7340617046371594, "grad_norm": 0.5177208181931565, "learning_rate": 9.375786393607125e-05, "loss": 3.1199, "step": 11825 }, { "epoch": 0.7341237817369173, "grad_norm": 0.497231328100628, "learning_rate": 9.375611642732111e-05, "loss": 3.0401, "step": 11826 }, { "epoch": 0.7341858588366752, "grad_norm": 0.4640865249066244, "learning_rate": 9.375436869028469e-05, "loss": 3.0941, "step": 11827 }, { "epoch": 0.734247935936433, "grad_norm": 0.4938579549954577, "learning_rate": 9.375262072497113e-05, "loss": 3.0618, "step": 11828 }, { "epoch": 0.7343100130361909, "grad_norm": 0.5727356162244455, "learning_rate": 9.375087253138951e-05, "loss": 3.1018, "step": 11829 }, { "epoch": 0.7343720901359488, "grad_norm": 0.45158872297235797, "learning_rate": 9.374912410954897e-05, "loss": 2.9748, "step": 11830 }, { "epoch": 0.7344341672357068, "grad_norm": 0.5035139227365584, "learning_rate": 9.374737545945862e-05, "loss": 3.0305, "step": 11831 }, { "epoch": 0.7344962443354647, "grad_norm": 0.4860967675620883, "learning_rate": 9.37456265811276e-05, "loss": 2.9573, "step": 11832 }, { "epoch": 0.7345583214352226, "grad_norm": 0.49362200635290615, "learning_rate": 9.374387747456503e-05, "loss": 3.1298, "step": 11833 }, { "epoch": 0.7346203985349804, "grad_norm": 0.5141201601475373, "learning_rate": 9.374212813978002e-05, "loss": 3.0164, "step": 11834 }, { "epoch": 0.7346824756347383, "grad_norm": 0.49323704245649613, "learning_rate": 9.374037857678173e-05, "loss": 3.0191, "step": 11835 }, { "epoch": 0.7347445527344962, "grad_norm": 0.47375162738329296, "learning_rate": 9.373862878557926e-05, "loss": 3.0037, "step": 11836 }, { "epoch": 0.7348066298342542, "grad_norm": 0.5193812218211926, "learning_rate": 9.373687876618174e-05, "loss": 3.043, "step": 11837 }, { "epoch": 0.7348687069340121, "grad_norm": 0.4852389966885577, "learning_rate": 9.37351285185983e-05, "loss": 2.9959, "step": 11838 }, { "epoch": 0.73493078403377, "grad_norm": 0.501418485792344, "learning_rate": 9.373337804283809e-05, "loss": 2.9694, "step": 11839 }, { "epoch": 0.7349928611335278, "grad_norm": 0.46076781172121756, "learning_rate": 9.37316273389102e-05, "loss": 3.0956, "step": 11840 }, { "epoch": 0.7350549382332857, "grad_norm": 0.4553713529170062, "learning_rate": 9.372987640682382e-05, "loss": 3.1707, "step": 11841 }, { "epoch": 0.7351170153330436, "grad_norm": 0.46941971620473993, "learning_rate": 9.372812524658808e-05, "loss": 2.9993, "step": 11842 }, { "epoch": 0.7351790924328016, "grad_norm": 0.4952548134842813, "learning_rate": 9.372637385821205e-05, "loss": 3.1655, "step": 11843 }, { "epoch": 0.7352411695325595, "grad_norm": 0.49800944468148256, "learning_rate": 9.372462224170494e-05, "loss": 3.0716, "step": 11844 }, { "epoch": 0.7353032466323174, "grad_norm": 0.4892435738020162, "learning_rate": 9.372287039707586e-05, "loss": 2.9918, "step": 11845 }, { "epoch": 0.7353653237320752, "grad_norm": 0.5266120355324554, "learning_rate": 9.372111832433394e-05, "loss": 3.0538, "step": 11846 }, { "epoch": 0.7354274008318331, "grad_norm": 0.44814312739930645, "learning_rate": 9.371936602348835e-05, "loss": 3.0599, "step": 11847 }, { "epoch": 0.735489477931591, "grad_norm": 0.5234713472269222, "learning_rate": 9.37176134945482e-05, "loss": 3.1414, "step": 11848 }, { "epoch": 0.735551555031349, "grad_norm": 0.44277786569017447, "learning_rate": 9.371586073752264e-05, "loss": 3.0985, "step": 11849 }, { "epoch": 0.7356136321311069, "grad_norm": 0.46908253954141804, "learning_rate": 9.371410775242084e-05, "loss": 3.1164, "step": 11850 }, { "epoch": 0.7356757092308648, "grad_norm": 0.4786579932355234, "learning_rate": 9.37123545392519e-05, "loss": 3.0037, "step": 11851 }, { "epoch": 0.7357377863306226, "grad_norm": 0.47311487882464043, "learning_rate": 9.371060109802503e-05, "loss": 3.0181, "step": 11852 }, { "epoch": 0.7357998634303805, "grad_norm": 0.46565729963679553, "learning_rate": 9.370884742874933e-05, "loss": 3.0631, "step": 11853 }, { "epoch": 0.7358619405301384, "grad_norm": 0.516558813046165, "learning_rate": 9.370709353143397e-05, "loss": 2.951, "step": 11854 }, { "epoch": 0.7359240176298963, "grad_norm": 0.5488883151787339, "learning_rate": 9.370533940608807e-05, "loss": 3.1326, "step": 11855 }, { "epoch": 0.7359860947296543, "grad_norm": 0.5343933793861336, "learning_rate": 9.370358505272082e-05, "loss": 3.0534, "step": 11856 }, { "epoch": 0.7360481718294122, "grad_norm": 0.5225313798707945, "learning_rate": 9.370183047134136e-05, "loss": 3.1261, "step": 11857 }, { "epoch": 0.73611024892917, "grad_norm": 0.5375818567267056, "learning_rate": 9.370007566195884e-05, "loss": 2.9919, "step": 11858 }, { "epoch": 0.7361723260289279, "grad_norm": 0.5017822509136054, "learning_rate": 9.369832062458241e-05, "loss": 3.0686, "step": 11859 }, { "epoch": 0.7362344031286858, "grad_norm": 0.5228405010434122, "learning_rate": 9.369656535922124e-05, "loss": 3.0024, "step": 11860 }, { "epoch": 0.7362964802284437, "grad_norm": 0.4538127842334718, "learning_rate": 9.369480986588447e-05, "loss": 3.1078, "step": 11861 }, { "epoch": 0.7363585573282017, "grad_norm": 0.502374702081452, "learning_rate": 9.369305414458127e-05, "loss": 3.0333, "step": 11862 }, { "epoch": 0.7364206344279596, "grad_norm": 0.4546192294906475, "learning_rate": 9.36912981953208e-05, "loss": 3.0427, "step": 11863 }, { "epoch": 0.7364827115277174, "grad_norm": 0.44729039485924116, "learning_rate": 9.368954201811224e-05, "loss": 3.0683, "step": 11864 }, { "epoch": 0.7365447886274753, "grad_norm": 0.4810087069716682, "learning_rate": 9.368778561296472e-05, "loss": 3.0102, "step": 11865 }, { "epoch": 0.7366068657272332, "grad_norm": 0.47273550078510623, "learning_rate": 9.36860289798874e-05, "loss": 3.0502, "step": 11866 }, { "epoch": 0.7366689428269911, "grad_norm": 0.39421902000729464, "learning_rate": 9.36842721188895e-05, "loss": 2.984, "step": 11867 }, { "epoch": 0.736731019926749, "grad_norm": 0.461563590325454, "learning_rate": 9.368251502998012e-05, "loss": 3.0699, "step": 11868 }, { "epoch": 0.736793097026507, "grad_norm": 0.4379102962815747, "learning_rate": 9.368075771316847e-05, "loss": 2.9858, "step": 11869 }, { "epoch": 0.7368551741262648, "grad_norm": 0.42848387246734837, "learning_rate": 9.36790001684637e-05, "loss": 3.132, "step": 11870 }, { "epoch": 0.7369172512260227, "grad_norm": 0.4936799929277824, "learning_rate": 9.367724239587497e-05, "loss": 3.086, "step": 11871 }, { "epoch": 0.7369793283257806, "grad_norm": 0.4990111646244151, "learning_rate": 9.367548439541148e-05, "loss": 3.0842, "step": 11872 }, { "epoch": 0.7370414054255385, "grad_norm": 0.5213046675083265, "learning_rate": 9.367372616708237e-05, "loss": 3.041, "step": 11873 }, { "epoch": 0.7371034825252964, "grad_norm": 0.5492162723403978, "learning_rate": 9.367196771089683e-05, "loss": 3.1025, "step": 11874 }, { "epoch": 0.7371655596250544, "grad_norm": 0.5022463109273513, "learning_rate": 9.367020902686403e-05, "loss": 3.1038, "step": 11875 }, { "epoch": 0.7372276367248122, "grad_norm": 0.5227154548537966, "learning_rate": 9.366845011499315e-05, "loss": 3.0344, "step": 11876 }, { "epoch": 0.7372897138245701, "grad_norm": 0.5333859196705758, "learning_rate": 9.366669097529337e-05, "loss": 3.028, "step": 11877 }, { "epoch": 0.737351790924328, "grad_norm": 0.4827395948901683, "learning_rate": 9.366493160777388e-05, "loss": 3.0474, "step": 11878 }, { "epoch": 0.7374138680240859, "grad_norm": 0.5435977558230713, "learning_rate": 9.366317201244381e-05, "loss": 3.1512, "step": 11879 }, { "epoch": 0.7374759451238438, "grad_norm": 0.5591955078104194, "learning_rate": 9.36614121893124e-05, "loss": 3.0656, "step": 11880 }, { "epoch": 0.7375380222236018, "grad_norm": 0.5331095519069337, "learning_rate": 9.365965213838879e-05, "loss": 2.9627, "step": 11881 }, { "epoch": 0.7376000993233596, "grad_norm": 0.5220766169163455, "learning_rate": 9.365789185968216e-05, "loss": 3.069, "step": 11882 }, { "epoch": 0.7376621764231175, "grad_norm": 0.5432150851430391, "learning_rate": 9.365613135320174e-05, "loss": 3.0818, "step": 11883 }, { "epoch": 0.7377242535228754, "grad_norm": 0.5284970987234898, "learning_rate": 9.365437061895666e-05, "loss": 2.9744, "step": 11884 }, { "epoch": 0.7377863306226333, "grad_norm": 0.5327901113957842, "learning_rate": 9.365260965695615e-05, "loss": 3.0677, "step": 11885 }, { "epoch": 0.7378484077223912, "grad_norm": 0.5671785548321828, "learning_rate": 9.365084846720937e-05, "loss": 3.0566, "step": 11886 }, { "epoch": 0.7379104848221492, "grad_norm": 0.5392106170615761, "learning_rate": 9.36490870497255e-05, "loss": 3.0904, "step": 11887 }, { "epoch": 0.737972561921907, "grad_norm": 0.48318558447540416, "learning_rate": 9.364732540451378e-05, "loss": 3.0773, "step": 11888 }, { "epoch": 0.7380346390216649, "grad_norm": 0.6027500024476717, "learning_rate": 9.364556353158335e-05, "loss": 3.0468, "step": 11889 }, { "epoch": 0.7380967161214228, "grad_norm": 0.5614548881816677, "learning_rate": 9.364380143094342e-05, "loss": 3.1472, "step": 11890 }, { "epoch": 0.7381587932211807, "grad_norm": 0.531940292223779, "learning_rate": 9.364203910260319e-05, "loss": 3.0431, "step": 11891 }, { "epoch": 0.7382208703209386, "grad_norm": 0.5733386970655396, "learning_rate": 9.364027654657183e-05, "loss": 3.0501, "step": 11892 }, { "epoch": 0.7382829474206966, "grad_norm": 0.5549002226072365, "learning_rate": 9.363851376285858e-05, "loss": 3.1044, "step": 11893 }, { "epoch": 0.7383450245204544, "grad_norm": 0.4899932450685252, "learning_rate": 9.363675075147258e-05, "loss": 3.0407, "step": 11894 }, { "epoch": 0.7384071016202123, "grad_norm": 0.510044328642376, "learning_rate": 9.363498751242308e-05, "loss": 3.0439, "step": 11895 }, { "epoch": 0.7384691787199702, "grad_norm": 0.5472453906656756, "learning_rate": 9.363322404571925e-05, "loss": 3.0653, "step": 11896 }, { "epoch": 0.7385312558197281, "grad_norm": 0.488353815408665, "learning_rate": 9.363146035137029e-05, "loss": 3.0917, "step": 11897 }, { "epoch": 0.738593332919486, "grad_norm": 0.516664555247925, "learning_rate": 9.362969642938541e-05, "loss": 3.0502, "step": 11898 }, { "epoch": 0.738655410019244, "grad_norm": 0.5009394998207082, "learning_rate": 9.362793227977384e-05, "loss": 3.1105, "step": 11899 }, { "epoch": 0.7387174871190018, "grad_norm": 0.47253528339608897, "learning_rate": 9.362616790254472e-05, "loss": 3.001, "step": 11900 }, { "epoch": 0.7387795642187597, "grad_norm": 0.5230290826530472, "learning_rate": 9.362440329770732e-05, "loss": 3.0444, "step": 11901 }, { "epoch": 0.7388416413185176, "grad_norm": 0.48914772972914433, "learning_rate": 9.362263846527081e-05, "loss": 3.0054, "step": 11902 }, { "epoch": 0.7389037184182755, "grad_norm": 0.4990522521895082, "learning_rate": 9.36208734052444e-05, "loss": 3.0853, "step": 11903 }, { "epoch": 0.7389657955180334, "grad_norm": 0.47113982397589593, "learning_rate": 9.36191081176373e-05, "loss": 3.0934, "step": 11904 }, { "epoch": 0.7390278726177913, "grad_norm": 0.4789429294298841, "learning_rate": 9.361734260245874e-05, "loss": 2.9975, "step": 11905 }, { "epoch": 0.7390899497175492, "grad_norm": 0.543584597851276, "learning_rate": 9.361557685971789e-05, "loss": 3.1704, "step": 11906 }, { "epoch": 0.7391520268173071, "grad_norm": 0.5142727952187879, "learning_rate": 9.361381088942401e-05, "loss": 3.1821, "step": 11907 }, { "epoch": 0.739214103917065, "grad_norm": 0.4905011640363796, "learning_rate": 9.361204469158627e-05, "loss": 3.0405, "step": 11908 }, { "epoch": 0.7392761810168229, "grad_norm": 0.5208092938279492, "learning_rate": 9.361027826621392e-05, "loss": 2.9455, "step": 11909 }, { "epoch": 0.7393382581165808, "grad_norm": 0.5087335596544661, "learning_rate": 9.360851161331617e-05, "loss": 3.0206, "step": 11910 }, { "epoch": 0.7394003352163387, "grad_norm": 0.5189506900897806, "learning_rate": 9.36067447329022e-05, "loss": 3.0665, "step": 11911 }, { "epoch": 0.7394624123160966, "grad_norm": 0.6147222415820451, "learning_rate": 9.360497762498129e-05, "loss": 3.0817, "step": 11912 }, { "epoch": 0.7395244894158545, "grad_norm": 0.562420362557138, "learning_rate": 9.360321028956261e-05, "loss": 3.0519, "step": 11913 }, { "epoch": 0.7395865665156124, "grad_norm": 0.602042192705683, "learning_rate": 9.36014427266554e-05, "loss": 3.0521, "step": 11914 }, { "epoch": 0.7396486436153703, "grad_norm": 0.4931404636893216, "learning_rate": 9.359967493626887e-05, "loss": 3.0934, "step": 11915 }, { "epoch": 0.7397107207151282, "grad_norm": 0.5624758132396984, "learning_rate": 9.359790691841225e-05, "loss": 3.0513, "step": 11916 }, { "epoch": 0.7397727978148861, "grad_norm": 0.5081021876451907, "learning_rate": 9.359613867309478e-05, "loss": 3.1195, "step": 11917 }, { "epoch": 0.739834874914644, "grad_norm": 0.540899326204997, "learning_rate": 9.359437020032567e-05, "loss": 3.0243, "step": 11918 }, { "epoch": 0.7398969520144019, "grad_norm": 0.4812819414561714, "learning_rate": 9.359260150011416e-05, "loss": 3.0237, "step": 11919 }, { "epoch": 0.7399590291141598, "grad_norm": 0.5355841603505941, "learning_rate": 9.359083257246944e-05, "loss": 3.0606, "step": 11920 }, { "epoch": 0.7400211062139177, "grad_norm": 0.528905153484701, "learning_rate": 9.358906341740077e-05, "loss": 3.0783, "step": 11921 }, { "epoch": 0.7400831833136756, "grad_norm": 0.5212394053253799, "learning_rate": 9.358729403491737e-05, "loss": 3.0381, "step": 11922 }, { "epoch": 0.7401452604134335, "grad_norm": 0.5641890175231559, "learning_rate": 9.358552442502849e-05, "loss": 3.0974, "step": 11923 }, { "epoch": 0.7402073375131913, "grad_norm": 0.5308866380170811, "learning_rate": 9.358375458774334e-05, "loss": 3.0849, "step": 11924 }, { "epoch": 0.7402694146129493, "grad_norm": 0.6473811173754903, "learning_rate": 9.358198452307117e-05, "loss": 3.0584, "step": 11925 }, { "epoch": 0.7403314917127072, "grad_norm": 0.5440010302512848, "learning_rate": 9.35802142310212e-05, "loss": 2.9263, "step": 11926 }, { "epoch": 0.7403935688124651, "grad_norm": 0.524609861281362, "learning_rate": 9.357844371160267e-05, "loss": 2.9905, "step": 11927 }, { "epoch": 0.740455645912223, "grad_norm": 0.6161060274174496, "learning_rate": 9.357667296482482e-05, "loss": 3.0813, "step": 11928 }, { "epoch": 0.7405177230119809, "grad_norm": 0.5941802273103899, "learning_rate": 9.357490199069689e-05, "loss": 3.148, "step": 11929 }, { "epoch": 0.7405798001117387, "grad_norm": 0.46811327605088054, "learning_rate": 9.357313078922814e-05, "loss": 3.067, "step": 11930 }, { "epoch": 0.7406418772114967, "grad_norm": 0.5618712301838554, "learning_rate": 9.357135936042776e-05, "loss": 2.9921, "step": 11931 }, { "epoch": 0.7407039543112546, "grad_norm": 0.655972649230897, "learning_rate": 9.356958770430503e-05, "loss": 3.0697, "step": 11932 }, { "epoch": 0.7407660314110125, "grad_norm": 0.585200953081128, "learning_rate": 9.356781582086917e-05, "loss": 3.1133, "step": 11933 }, { "epoch": 0.7408281085107704, "grad_norm": 0.5692412361438736, "learning_rate": 9.356604371012947e-05, "loss": 2.9967, "step": 11934 }, { "epoch": 0.7408901856105283, "grad_norm": 0.6028691010824259, "learning_rate": 9.356427137209512e-05, "loss": 3.0514, "step": 11935 }, { "epoch": 0.7409522627102861, "grad_norm": 0.5484427188288074, "learning_rate": 9.35624988067754e-05, "loss": 3.1473, "step": 11936 }, { "epoch": 0.741014339810044, "grad_norm": 0.6394930585301046, "learning_rate": 9.356072601417954e-05, "loss": 3.0699, "step": 11937 }, { "epoch": 0.741076416909802, "grad_norm": 0.5966528154431257, "learning_rate": 9.35589529943168e-05, "loss": 3.146, "step": 11938 }, { "epoch": 0.7411384940095599, "grad_norm": 0.5316352849575819, "learning_rate": 9.355717974719644e-05, "loss": 3.0814, "step": 11939 }, { "epoch": 0.7412005711093178, "grad_norm": 0.6091344309608467, "learning_rate": 9.35554062728277e-05, "loss": 3.0472, "step": 11940 }, { "epoch": 0.7412626482090757, "grad_norm": 0.5769130013576421, "learning_rate": 9.355363257121981e-05, "loss": 3.0385, "step": 11941 }, { "epoch": 0.7413247253088335, "grad_norm": 0.5434418956366038, "learning_rate": 9.355185864238206e-05, "loss": 3.0334, "step": 11942 }, { "epoch": 0.7413868024085914, "grad_norm": 0.5383787656465925, "learning_rate": 9.35500844863237e-05, "loss": 3.1187, "step": 11943 }, { "epoch": 0.7414488795083494, "grad_norm": 0.5340793661866104, "learning_rate": 9.354831010305396e-05, "loss": 2.9859, "step": 11944 }, { "epoch": 0.7415109566081073, "grad_norm": 0.44951954597986205, "learning_rate": 9.354653549258212e-05, "loss": 2.9845, "step": 11945 }, { "epoch": 0.7415730337078652, "grad_norm": 0.5241287279221621, "learning_rate": 9.354476065491744e-05, "loss": 3.0089, "step": 11946 }, { "epoch": 0.7416351108076231, "grad_norm": 0.5411490816294816, "learning_rate": 9.354298559006916e-05, "loss": 3.107, "step": 11947 }, { "epoch": 0.7416971879073809, "grad_norm": 0.4849293827970961, "learning_rate": 9.354121029804657e-05, "loss": 3.0322, "step": 11948 }, { "epoch": 0.7417592650071388, "grad_norm": 0.5170477687010295, "learning_rate": 9.35394347788589e-05, "loss": 3.0201, "step": 11949 }, { "epoch": 0.7418213421068968, "grad_norm": 0.4450221369894962, "learning_rate": 9.353765903251543e-05, "loss": 3.0312, "step": 11950 }, { "epoch": 0.7418834192066547, "grad_norm": 0.5099872385092655, "learning_rate": 9.353588305902543e-05, "loss": 3.0037, "step": 11951 }, { "epoch": 0.7419454963064126, "grad_norm": 0.48786536243721723, "learning_rate": 9.353410685839816e-05, "loss": 2.9856, "step": 11952 }, { "epoch": 0.7420075734061705, "grad_norm": 0.44726071084333846, "learning_rate": 9.353233043064289e-05, "loss": 3.0026, "step": 11953 }, { "epoch": 0.7420696505059283, "grad_norm": 0.5095811415048386, "learning_rate": 9.353055377576889e-05, "loss": 3.0664, "step": 11954 }, { "epoch": 0.7421317276056862, "grad_norm": 0.524143881077645, "learning_rate": 9.35287768937854e-05, "loss": 2.9725, "step": 11955 }, { "epoch": 0.7421938047054442, "grad_norm": 0.5758016140156286, "learning_rate": 9.352699978470174e-05, "loss": 3.0506, "step": 11956 }, { "epoch": 0.7422558818052021, "grad_norm": 0.48482083752382094, "learning_rate": 9.352522244852712e-05, "loss": 2.8815, "step": 11957 }, { "epoch": 0.74231795890496, "grad_norm": 0.5196882025185778, "learning_rate": 9.352344488527088e-05, "loss": 3.0718, "step": 11958 }, { "epoch": 0.7423800360047179, "grad_norm": 0.4793921714005054, "learning_rate": 9.352166709494226e-05, "loss": 3.118, "step": 11959 }, { "epoch": 0.7424421131044757, "grad_norm": 0.5251565574910684, "learning_rate": 9.351988907755054e-05, "loss": 3.075, "step": 11960 }, { "epoch": 0.7425041902042336, "grad_norm": 0.4827756318271578, "learning_rate": 9.351811083310498e-05, "loss": 2.9925, "step": 11961 }, { "epoch": 0.7425662673039916, "grad_norm": 0.49101266973876195, "learning_rate": 9.351633236161488e-05, "loss": 3.0392, "step": 11962 }, { "epoch": 0.7426283444037495, "grad_norm": 0.4855986490059588, "learning_rate": 9.35145536630895e-05, "loss": 3.0307, "step": 11963 }, { "epoch": 0.7426904215035074, "grad_norm": 0.47459278156173923, "learning_rate": 9.351277473753813e-05, "loss": 3.0983, "step": 11964 }, { "epoch": 0.7427524986032653, "grad_norm": 0.5211037950820074, "learning_rate": 9.351099558497006e-05, "loss": 3.0614, "step": 11965 }, { "epoch": 0.7428145757030231, "grad_norm": 0.5450547852671713, "learning_rate": 9.350921620539457e-05, "loss": 3.0469, "step": 11966 }, { "epoch": 0.742876652802781, "grad_norm": 0.48991248212888316, "learning_rate": 9.350743659882092e-05, "loss": 2.9647, "step": 11967 }, { "epoch": 0.742938729902539, "grad_norm": 0.5272998932862748, "learning_rate": 9.350565676525842e-05, "loss": 3.0312, "step": 11968 }, { "epoch": 0.7430008070022969, "grad_norm": 0.49309626953692026, "learning_rate": 9.350387670471634e-05, "loss": 3.0208, "step": 11969 }, { "epoch": 0.7430628841020548, "grad_norm": 0.4675041809507059, "learning_rate": 9.350209641720399e-05, "loss": 2.9128, "step": 11970 }, { "epoch": 0.7431249612018127, "grad_norm": 0.4750065495641803, "learning_rate": 9.350031590273063e-05, "loss": 3.0778, "step": 11971 }, { "epoch": 0.7431870383015705, "grad_norm": 0.4889271850590166, "learning_rate": 9.349853516130556e-05, "loss": 3.0457, "step": 11972 }, { "epoch": 0.7432491154013284, "grad_norm": 0.47553334346775955, "learning_rate": 9.349675419293808e-05, "loss": 3.0677, "step": 11973 }, { "epoch": 0.7433111925010863, "grad_norm": 0.4589749274397522, "learning_rate": 9.349497299763748e-05, "loss": 3.0848, "step": 11974 }, { "epoch": 0.7433732696008443, "grad_norm": 0.4384590434289451, "learning_rate": 9.349319157541303e-05, "loss": 3.0825, "step": 11975 }, { "epoch": 0.7434353467006022, "grad_norm": 0.47949040057127096, "learning_rate": 9.349140992627405e-05, "loss": 2.9622, "step": 11976 }, { "epoch": 0.7434974238003601, "grad_norm": 0.48723491712001216, "learning_rate": 9.348962805022982e-05, "loss": 3.1156, "step": 11977 }, { "epoch": 0.7435595009001179, "grad_norm": 0.49286878451869653, "learning_rate": 9.348784594728966e-05, "loss": 2.9976, "step": 11978 }, { "epoch": 0.7436215779998758, "grad_norm": 0.5162985188681093, "learning_rate": 9.348606361746284e-05, "loss": 2.96, "step": 11979 }, { "epoch": 0.7436836550996337, "grad_norm": 0.5299466357266428, "learning_rate": 9.348428106075867e-05, "loss": 3.0389, "step": 11980 }, { "epoch": 0.7437457321993917, "grad_norm": 0.47279256636206546, "learning_rate": 9.348249827718645e-05, "loss": 3.0551, "step": 11981 }, { "epoch": 0.7438078092991496, "grad_norm": 0.48794128790199043, "learning_rate": 9.348071526675549e-05, "loss": 3.0461, "step": 11982 }, { "epoch": 0.7438698863989075, "grad_norm": 0.4935077791373857, "learning_rate": 9.347893202947506e-05, "loss": 2.9411, "step": 11983 }, { "epoch": 0.7439319634986653, "grad_norm": 0.45480942877408354, "learning_rate": 9.34771485653545e-05, "loss": 2.9693, "step": 11984 }, { "epoch": 0.7439940405984232, "grad_norm": 0.4788530989208962, "learning_rate": 9.34753648744031e-05, "loss": 3.2068, "step": 11985 }, { "epoch": 0.7440561176981811, "grad_norm": 0.497170743557993, "learning_rate": 9.347358095663017e-05, "loss": 3.1121, "step": 11986 }, { "epoch": 0.7441181947979391, "grad_norm": 0.4858656526544284, "learning_rate": 9.347179681204502e-05, "loss": 3.1152, "step": 11987 }, { "epoch": 0.744180271897697, "grad_norm": 0.4875595863312927, "learning_rate": 9.347001244065694e-05, "loss": 2.9818, "step": 11988 }, { "epoch": 0.7442423489974548, "grad_norm": 0.5292313031897534, "learning_rate": 9.346822784247527e-05, "loss": 3.0981, "step": 11989 }, { "epoch": 0.7443044260972127, "grad_norm": 0.5316209900178055, "learning_rate": 9.346644301750929e-05, "loss": 3.0874, "step": 11990 }, { "epoch": 0.7443665031969706, "grad_norm": 0.5061255090022713, "learning_rate": 9.346465796576831e-05, "loss": 2.9835, "step": 11991 }, { "epoch": 0.7444285802967285, "grad_norm": 0.48901787697502397, "learning_rate": 9.346287268726168e-05, "loss": 3.0355, "step": 11992 }, { "epoch": 0.7444906573964865, "grad_norm": 0.49846854217871256, "learning_rate": 9.346108718199869e-05, "loss": 3.0179, "step": 11993 }, { "epoch": 0.7445527344962444, "grad_norm": 0.4823886998928056, "learning_rate": 9.345930144998864e-05, "loss": 3.0874, "step": 11994 }, { "epoch": 0.7446148115960022, "grad_norm": 0.5016892626522204, "learning_rate": 9.345751549124089e-05, "loss": 3.1057, "step": 11995 }, { "epoch": 0.7446768886957601, "grad_norm": 0.45976131505250584, "learning_rate": 9.345572930576471e-05, "loss": 3.0663, "step": 11996 }, { "epoch": 0.744738965795518, "grad_norm": 0.4741724145615763, "learning_rate": 9.345394289356946e-05, "loss": 2.9557, "step": 11997 }, { "epoch": 0.7448010428952759, "grad_norm": 0.48693812464244457, "learning_rate": 9.345215625466444e-05, "loss": 3.1212, "step": 11998 }, { "epoch": 0.7448631199950339, "grad_norm": 0.5995834543794886, "learning_rate": 9.345036938905895e-05, "loss": 3.1153, "step": 11999 }, { "epoch": 0.7449251970947918, "grad_norm": 0.5131020522010221, "learning_rate": 9.344858229676237e-05, "loss": 3.1024, "step": 12000 }, { "epoch": 0.7449872741945496, "grad_norm": 0.506132133326215, "learning_rate": 9.344679497778396e-05, "loss": 3.0229, "step": 12001 }, { "epoch": 0.7450493512943075, "grad_norm": 0.46213537483239503, "learning_rate": 9.344500743213307e-05, "loss": 3.028, "step": 12002 }, { "epoch": 0.7451114283940654, "grad_norm": 0.5499954841168284, "learning_rate": 9.344321965981905e-05, "loss": 2.9914, "step": 12003 }, { "epoch": 0.7451735054938233, "grad_norm": 0.46881858006114846, "learning_rate": 9.34414316608512e-05, "loss": 3.1101, "step": 12004 }, { "epoch": 0.7452355825935812, "grad_norm": 0.4894323589672136, "learning_rate": 9.343964343523886e-05, "loss": 3.0691, "step": 12005 }, { "epoch": 0.7452976596933392, "grad_norm": 0.5893239389754287, "learning_rate": 9.343785498299134e-05, "loss": 3.1509, "step": 12006 }, { "epoch": 0.745359736793097, "grad_norm": 0.5221926561850951, "learning_rate": 9.3436066304118e-05, "loss": 3.0892, "step": 12007 }, { "epoch": 0.7454218138928549, "grad_norm": 0.48508865327817174, "learning_rate": 9.343427739862816e-05, "loss": 3.0451, "step": 12008 }, { "epoch": 0.7454838909926128, "grad_norm": 0.5722538813404796, "learning_rate": 9.343248826653113e-05, "loss": 3.0475, "step": 12009 }, { "epoch": 0.7455459680923707, "grad_norm": 0.5192646977691228, "learning_rate": 9.343069890783627e-05, "loss": 3.0247, "step": 12010 }, { "epoch": 0.7456080451921286, "grad_norm": 0.4959265660004261, "learning_rate": 9.342890932255293e-05, "loss": 3.1074, "step": 12011 }, { "epoch": 0.7456701222918866, "grad_norm": 0.5321200482938071, "learning_rate": 9.34271195106904e-05, "loss": 2.913, "step": 12012 }, { "epoch": 0.7457321993916444, "grad_norm": 0.5814925907840522, "learning_rate": 9.342532947225808e-05, "loss": 3.0702, "step": 12013 }, { "epoch": 0.7457942764914023, "grad_norm": 0.5476934303363988, "learning_rate": 9.342353920726525e-05, "loss": 3.0296, "step": 12014 }, { "epoch": 0.7458563535911602, "grad_norm": 0.47918728770237284, "learning_rate": 9.342174871572127e-05, "loss": 3.0783, "step": 12015 }, { "epoch": 0.7459184306909181, "grad_norm": 0.5468231449065559, "learning_rate": 9.341995799763549e-05, "loss": 3.0564, "step": 12016 }, { "epoch": 0.745980507790676, "grad_norm": 0.4842122173576457, "learning_rate": 9.341816705301727e-05, "loss": 2.9812, "step": 12017 }, { "epoch": 0.746042584890434, "grad_norm": 0.5850011592022096, "learning_rate": 9.34163758818759e-05, "loss": 3.0151, "step": 12018 }, { "epoch": 0.7461046619901918, "grad_norm": 0.5483528332687593, "learning_rate": 9.341458448422079e-05, "loss": 3.1037, "step": 12019 }, { "epoch": 0.7461667390899497, "grad_norm": 0.5628325221234998, "learning_rate": 9.341279286006123e-05, "loss": 3.0291, "step": 12020 }, { "epoch": 0.7462288161897076, "grad_norm": 0.5809856726427747, "learning_rate": 9.34110010094066e-05, "loss": 3.1224, "step": 12021 }, { "epoch": 0.7462908932894655, "grad_norm": 0.5023957019478551, "learning_rate": 9.340920893226624e-05, "loss": 3.0445, "step": 12022 }, { "epoch": 0.7463529703892234, "grad_norm": 0.48412453420780627, "learning_rate": 9.340741662864949e-05, "loss": 3.0881, "step": 12023 }, { "epoch": 0.7464150474889814, "grad_norm": 0.5531216130988098, "learning_rate": 9.340562409856572e-05, "loss": 3.0645, "step": 12024 }, { "epoch": 0.7464771245887392, "grad_norm": 0.4551325753180565, "learning_rate": 9.340383134202429e-05, "loss": 3.024, "step": 12025 }, { "epoch": 0.7465392016884971, "grad_norm": 0.5034539018418217, "learning_rate": 9.340203835903449e-05, "loss": 3.0284, "step": 12026 }, { "epoch": 0.746601278788255, "grad_norm": 0.5182641143772212, "learning_rate": 9.340024514960574e-05, "loss": 3.1028, "step": 12027 }, { "epoch": 0.7466633558880129, "grad_norm": 0.42740785860602787, "learning_rate": 9.33984517137474e-05, "loss": 3.0451, "step": 12028 }, { "epoch": 0.7467254329877708, "grad_norm": 0.48448919672874474, "learning_rate": 9.339665805146878e-05, "loss": 3.0651, "step": 12029 }, { "epoch": 0.7467875100875287, "grad_norm": 0.5368103859757983, "learning_rate": 9.339486416277925e-05, "loss": 3.0135, "step": 12030 }, { "epoch": 0.7468495871872866, "grad_norm": 0.48780434779430565, "learning_rate": 9.33930700476882e-05, "loss": 3.0034, "step": 12031 }, { "epoch": 0.7469116642870445, "grad_norm": 0.46455183399940003, "learning_rate": 9.339127570620495e-05, "loss": 3.1078, "step": 12032 }, { "epoch": 0.7469737413868024, "grad_norm": 0.5822465485601142, "learning_rate": 9.33894811383389e-05, "loss": 3.0735, "step": 12033 }, { "epoch": 0.7470358184865603, "grad_norm": 0.5245449535044581, "learning_rate": 9.338768634409938e-05, "loss": 3.0307, "step": 12034 }, { "epoch": 0.7470978955863182, "grad_norm": 0.5038926470199423, "learning_rate": 9.338589132349577e-05, "loss": 3.1293, "step": 12035 }, { "epoch": 0.7471599726860761, "grad_norm": 0.5173213393721398, "learning_rate": 9.338409607653743e-05, "loss": 3.0337, "step": 12036 }, { "epoch": 0.747222049785834, "grad_norm": 0.5307375736587723, "learning_rate": 9.338230060323375e-05, "loss": 2.9848, "step": 12037 }, { "epoch": 0.7472841268855919, "grad_norm": 0.5235960421373995, "learning_rate": 9.338050490359405e-05, "loss": 2.9877, "step": 12038 }, { "epoch": 0.7473462039853498, "grad_norm": 0.5889857285151708, "learning_rate": 9.337870897762773e-05, "loss": 3.0679, "step": 12039 }, { "epoch": 0.7474082810851077, "grad_norm": 0.4942596201802183, "learning_rate": 9.337691282534416e-05, "loss": 3.0614, "step": 12040 }, { "epoch": 0.7474703581848656, "grad_norm": 0.512348399186279, "learning_rate": 9.33751164467527e-05, "loss": 2.9568, "step": 12041 }, { "epoch": 0.7475324352846235, "grad_norm": 0.4356272157204978, "learning_rate": 9.337331984186274e-05, "loss": 2.854, "step": 12042 }, { "epoch": 0.7475945123843813, "grad_norm": 0.5377883780709335, "learning_rate": 9.337152301068364e-05, "loss": 3.0809, "step": 12043 }, { "epoch": 0.7476565894841393, "grad_norm": 0.5132948213173715, "learning_rate": 9.336972595322476e-05, "loss": 2.9953, "step": 12044 }, { "epoch": 0.7477186665838972, "grad_norm": 0.4613130139000592, "learning_rate": 9.336792866949551e-05, "loss": 2.9196, "step": 12045 }, { "epoch": 0.7477807436836551, "grad_norm": 0.5262265273995963, "learning_rate": 9.336613115950523e-05, "loss": 3.0877, "step": 12046 }, { "epoch": 0.747842820783413, "grad_norm": 0.5566348962321709, "learning_rate": 9.336433342326333e-05, "loss": 3.1123, "step": 12047 }, { "epoch": 0.7479048978831709, "grad_norm": 0.506619504607482, "learning_rate": 9.336253546077918e-05, "loss": 2.9628, "step": 12048 }, { "epoch": 0.7479669749829287, "grad_norm": 0.486424647178469, "learning_rate": 9.336073727206214e-05, "loss": 3.1147, "step": 12049 }, { "epoch": 0.7480290520826867, "grad_norm": 0.5616413291240103, "learning_rate": 9.335893885712161e-05, "loss": 3.0305, "step": 12050 }, { "epoch": 0.7480911291824446, "grad_norm": 0.5157374635061263, "learning_rate": 9.335714021596698e-05, "loss": 3.0772, "step": 12051 }, { "epoch": 0.7481532062822025, "grad_norm": 0.4676452811573767, "learning_rate": 9.335534134860761e-05, "loss": 3.1447, "step": 12052 }, { "epoch": 0.7482152833819604, "grad_norm": 0.5370289690348288, "learning_rate": 9.335354225505291e-05, "loss": 3.1359, "step": 12053 }, { "epoch": 0.7482773604817183, "grad_norm": 0.5121507059116446, "learning_rate": 9.335174293531226e-05, "loss": 3.0293, "step": 12054 }, { "epoch": 0.7483394375814761, "grad_norm": 0.5065191377675119, "learning_rate": 9.334994338939504e-05, "loss": 3.0421, "step": 12055 }, { "epoch": 0.7484015146812341, "grad_norm": 0.5367354044766531, "learning_rate": 9.334814361731063e-05, "loss": 2.9201, "step": 12056 }, { "epoch": 0.748463591780992, "grad_norm": 0.5078894043423511, "learning_rate": 9.334634361906844e-05, "loss": 3.0828, "step": 12057 }, { "epoch": 0.7485256688807499, "grad_norm": 0.5564680668579303, "learning_rate": 9.334454339467787e-05, "loss": 3.1048, "step": 12058 }, { "epoch": 0.7485877459805078, "grad_norm": 0.4552709146541834, "learning_rate": 9.334274294414826e-05, "loss": 3.0215, "step": 12059 }, { "epoch": 0.7486498230802657, "grad_norm": 0.5397390183961155, "learning_rate": 9.334094226748906e-05, "loss": 2.9813, "step": 12060 }, { "epoch": 0.7487119001800235, "grad_norm": 0.5227587270131375, "learning_rate": 9.333914136470965e-05, "loss": 3.0071, "step": 12061 }, { "epoch": 0.7487739772797815, "grad_norm": 0.530075738911329, "learning_rate": 9.33373402358194e-05, "loss": 3.1245, "step": 12062 }, { "epoch": 0.7488360543795394, "grad_norm": 0.518573595409903, "learning_rate": 9.333553888082774e-05, "loss": 2.9878, "step": 12063 }, { "epoch": 0.7488981314792973, "grad_norm": 0.49457979852858863, "learning_rate": 9.333373729974404e-05, "loss": 3.1568, "step": 12064 }, { "epoch": 0.7489602085790552, "grad_norm": 0.5084540789151445, "learning_rate": 9.333193549257772e-05, "loss": 3.1751, "step": 12065 }, { "epoch": 0.7490222856788131, "grad_norm": 0.47400707127028696, "learning_rate": 9.333013345933816e-05, "loss": 2.9925, "step": 12066 }, { "epoch": 0.7490843627785709, "grad_norm": 0.48172343424729136, "learning_rate": 9.332833120003479e-05, "loss": 2.9395, "step": 12067 }, { "epoch": 0.7491464398783289, "grad_norm": 0.49653268154393365, "learning_rate": 9.3326528714677e-05, "loss": 3.1392, "step": 12068 }, { "epoch": 0.7492085169780868, "grad_norm": 0.47751277234935535, "learning_rate": 9.332472600327418e-05, "loss": 3.0527, "step": 12069 }, { "epoch": 0.7492705940778447, "grad_norm": 0.45910537423432624, "learning_rate": 9.332292306583576e-05, "loss": 3.0959, "step": 12070 }, { "epoch": 0.7493326711776026, "grad_norm": 0.5201125657223277, "learning_rate": 9.332111990237114e-05, "loss": 3.0332, "step": 12071 }, { "epoch": 0.7493947482773605, "grad_norm": 0.5550838719821468, "learning_rate": 9.33193165128897e-05, "loss": 3.0457, "step": 12072 }, { "epoch": 0.7494568253771183, "grad_norm": 0.5692856310020259, "learning_rate": 9.331751289740088e-05, "loss": 2.9855, "step": 12073 }, { "epoch": 0.7495189024768762, "grad_norm": 0.508992979806929, "learning_rate": 9.331570905591406e-05, "loss": 3.0789, "step": 12074 }, { "epoch": 0.7495809795766342, "grad_norm": 0.5974540118950565, "learning_rate": 9.33139049884387e-05, "loss": 3.0333, "step": 12075 }, { "epoch": 0.7496430566763921, "grad_norm": 0.5169559121600833, "learning_rate": 9.331210069498417e-05, "loss": 2.9906, "step": 12076 }, { "epoch": 0.74970513377615, "grad_norm": 0.4900322607669899, "learning_rate": 9.331029617555991e-05, "loss": 3.1199, "step": 12077 }, { "epoch": 0.7497672108759079, "grad_norm": 0.4709964170856224, "learning_rate": 9.33084914301753e-05, "loss": 2.9233, "step": 12078 }, { "epoch": 0.7498292879756657, "grad_norm": 0.4996899597221941, "learning_rate": 9.330668645883978e-05, "loss": 3.003, "step": 12079 }, { "epoch": 0.7498913650754236, "grad_norm": 0.5111420854682828, "learning_rate": 9.330488126156279e-05, "loss": 3.0152, "step": 12080 }, { "epoch": 0.7499534421751816, "grad_norm": 0.503535437360602, "learning_rate": 9.330307583835369e-05, "loss": 3.0596, "step": 12081 }, { "epoch": 0.7500155192749395, "grad_norm": 0.4800044817268237, "learning_rate": 9.330127018922194e-05, "loss": 3.0203, "step": 12082 }, { "epoch": 0.7500775963746974, "grad_norm": 0.49203893144898075, "learning_rate": 9.329946431417695e-05, "loss": 2.9812, "step": 12083 }, { "epoch": 0.7501396734744553, "grad_norm": 0.49097273039113315, "learning_rate": 9.329765821322814e-05, "loss": 3.0874, "step": 12084 }, { "epoch": 0.7502017505742131, "grad_norm": 0.4621342946521912, "learning_rate": 9.329585188638494e-05, "loss": 3.1103, "step": 12085 }, { "epoch": 0.750263827673971, "grad_norm": 0.4911896566403592, "learning_rate": 9.329404533365678e-05, "loss": 3.0551, "step": 12086 }, { "epoch": 0.750325904773729, "grad_norm": 0.4720522118326998, "learning_rate": 9.329223855505307e-05, "loss": 3.0155, "step": 12087 }, { "epoch": 0.7503879818734869, "grad_norm": 0.5002350254820169, "learning_rate": 9.329043155058324e-05, "loss": 3.1036, "step": 12088 }, { "epoch": 0.7504500589732448, "grad_norm": 0.5488077886214328, "learning_rate": 9.328862432025673e-05, "loss": 3.067, "step": 12089 }, { "epoch": 0.7505121360730027, "grad_norm": 0.5536279094685914, "learning_rate": 9.328681686408294e-05, "loss": 2.9103, "step": 12090 }, { "epoch": 0.7505742131727605, "grad_norm": 0.5264472971238596, "learning_rate": 9.328500918207132e-05, "loss": 2.991, "step": 12091 }, { "epoch": 0.7506362902725184, "grad_norm": 0.5246273296178139, "learning_rate": 9.328320127423131e-05, "loss": 3.0418, "step": 12092 }, { "epoch": 0.7506983673722764, "grad_norm": 0.5421941982152357, "learning_rate": 9.328139314057233e-05, "loss": 3.076, "step": 12093 }, { "epoch": 0.7507604444720343, "grad_norm": 0.565369729666274, "learning_rate": 9.327958478110382e-05, "loss": 3.0346, "step": 12094 }, { "epoch": 0.7508225215717922, "grad_norm": 0.531347041523204, "learning_rate": 9.32777761958352e-05, "loss": 3.0199, "step": 12095 }, { "epoch": 0.7508845986715501, "grad_norm": 0.4815461863601926, "learning_rate": 9.327596738477591e-05, "loss": 2.9673, "step": 12096 }, { "epoch": 0.7509466757713079, "grad_norm": 0.5258597226128274, "learning_rate": 9.32741583479354e-05, "loss": 3.0881, "step": 12097 }, { "epoch": 0.7510087528710658, "grad_norm": 0.4835457163236699, "learning_rate": 9.32723490853231e-05, "loss": 3.0154, "step": 12098 }, { "epoch": 0.7510708299708238, "grad_norm": 0.505825763307432, "learning_rate": 9.327053959694846e-05, "loss": 2.9932, "step": 12099 }, { "epoch": 0.7511329070705817, "grad_norm": 0.5331624612917638, "learning_rate": 9.326872988282089e-05, "loss": 3.0076, "step": 12100 }, { "epoch": 0.7511949841703396, "grad_norm": 0.4678738193675419, "learning_rate": 9.326691994294987e-05, "loss": 2.9982, "step": 12101 }, { "epoch": 0.7512570612700975, "grad_norm": 0.4833943530594095, "learning_rate": 9.326510977734482e-05, "loss": 3.0205, "step": 12102 }, { "epoch": 0.7513191383698553, "grad_norm": 0.4670197980082254, "learning_rate": 9.32632993860152e-05, "loss": 3.0744, "step": 12103 }, { "epoch": 0.7513812154696132, "grad_norm": 0.5122580561937572, "learning_rate": 9.326148876897043e-05, "loss": 3.0218, "step": 12104 }, { "epoch": 0.7514432925693711, "grad_norm": 0.45224722983807175, "learning_rate": 9.325967792621997e-05, "loss": 2.9473, "step": 12105 }, { "epoch": 0.7515053696691291, "grad_norm": 0.5760236905163333, "learning_rate": 9.325786685777327e-05, "loss": 3.1498, "step": 12106 }, { "epoch": 0.751567446768887, "grad_norm": 0.43076350995030593, "learning_rate": 9.325605556363979e-05, "loss": 2.9936, "step": 12107 }, { "epoch": 0.7516295238686449, "grad_norm": 0.48730462634252497, "learning_rate": 9.325424404382894e-05, "loss": 3.1233, "step": 12108 }, { "epoch": 0.7516916009684027, "grad_norm": 0.4702840973905191, "learning_rate": 9.325243229835023e-05, "loss": 2.9958, "step": 12109 }, { "epoch": 0.7517536780681606, "grad_norm": 0.4758018927645146, "learning_rate": 9.325062032721306e-05, "loss": 3.0863, "step": 12110 }, { "epoch": 0.7518157551679185, "grad_norm": 0.4331074298526098, "learning_rate": 9.324880813042692e-05, "loss": 3.0563, "step": 12111 }, { "epoch": 0.7518778322676765, "grad_norm": 0.47920468601103283, "learning_rate": 9.324699570800123e-05, "loss": 3.0751, "step": 12112 }, { "epoch": 0.7519399093674344, "grad_norm": 0.46077672126332203, "learning_rate": 9.324518305994548e-05, "loss": 3.1194, "step": 12113 }, { "epoch": 0.7520019864671923, "grad_norm": 0.48704707000367253, "learning_rate": 9.32433701862691e-05, "loss": 2.9842, "step": 12114 }, { "epoch": 0.7520640635669501, "grad_norm": 0.47683498534661184, "learning_rate": 9.324155708698156e-05, "loss": 2.9817, "step": 12115 }, { "epoch": 0.752126140666708, "grad_norm": 0.4766434772670598, "learning_rate": 9.323974376209234e-05, "loss": 3.066, "step": 12116 }, { "epoch": 0.7521882177664659, "grad_norm": 0.5705393509862968, "learning_rate": 9.323793021161087e-05, "loss": 3.0839, "step": 12117 }, { "epoch": 0.7522502948662239, "grad_norm": 0.46277773160496627, "learning_rate": 9.323611643554662e-05, "loss": 2.9106, "step": 12118 }, { "epoch": 0.7523123719659818, "grad_norm": 0.5226139308064691, "learning_rate": 9.323430243390905e-05, "loss": 2.9741, "step": 12119 }, { "epoch": 0.7523744490657397, "grad_norm": 0.5144284844199987, "learning_rate": 9.323248820670762e-05, "loss": 2.9488, "step": 12120 }, { "epoch": 0.7524365261654975, "grad_norm": 0.5311249978596756, "learning_rate": 9.323067375395183e-05, "loss": 3.093, "step": 12121 }, { "epoch": 0.7524986032652554, "grad_norm": 0.5408837333240374, "learning_rate": 9.322885907565109e-05, "loss": 3.0524, "step": 12122 }, { "epoch": 0.7525606803650133, "grad_norm": 0.4302655406751923, "learning_rate": 9.322704417181492e-05, "loss": 3.0877, "step": 12123 }, { "epoch": 0.7526227574647713, "grad_norm": 0.5307203064373288, "learning_rate": 9.322522904245277e-05, "loss": 3.0085, "step": 12124 }, { "epoch": 0.7526848345645292, "grad_norm": 0.513357929558026, "learning_rate": 9.322341368757409e-05, "loss": 3.0264, "step": 12125 }, { "epoch": 0.7527469116642871, "grad_norm": 0.4656132811793367, "learning_rate": 9.322159810718838e-05, "loss": 2.9686, "step": 12126 }, { "epoch": 0.7528089887640449, "grad_norm": 0.5467922693073037, "learning_rate": 9.321978230130509e-05, "loss": 3.0047, "step": 12127 }, { "epoch": 0.7528710658638028, "grad_norm": 0.5373874291373285, "learning_rate": 9.321796626993371e-05, "loss": 3.0119, "step": 12128 }, { "epoch": 0.7529331429635607, "grad_norm": 0.5063501576846512, "learning_rate": 9.32161500130837e-05, "loss": 3.0853, "step": 12129 }, { "epoch": 0.7529952200633186, "grad_norm": 0.5380107737139599, "learning_rate": 9.321433353076456e-05, "loss": 3.0715, "step": 12130 }, { "epoch": 0.7530572971630766, "grad_norm": 0.6105876496526319, "learning_rate": 9.321251682298575e-05, "loss": 3.0712, "step": 12131 }, { "epoch": 0.7531193742628345, "grad_norm": 0.533049341179584, "learning_rate": 9.321069988975674e-05, "loss": 2.9848, "step": 12132 }, { "epoch": 0.7531814513625923, "grad_norm": 0.5371435813415205, "learning_rate": 9.320888273108702e-05, "loss": 3.0139, "step": 12133 }, { "epoch": 0.7532435284623502, "grad_norm": 0.5922710923351503, "learning_rate": 9.320706534698607e-05, "loss": 3.0611, "step": 12134 }, { "epoch": 0.7533056055621081, "grad_norm": 0.5552311503127627, "learning_rate": 9.320524773746337e-05, "loss": 3.1739, "step": 12135 }, { "epoch": 0.753367682661866, "grad_norm": 0.5581287235678315, "learning_rate": 9.320342990252842e-05, "loss": 3.0209, "step": 12136 }, { "epoch": 0.753429759761624, "grad_norm": 0.6663174658392501, "learning_rate": 9.320161184219067e-05, "loss": 2.9813, "step": 12137 }, { "epoch": 0.7534918368613819, "grad_norm": 0.576791953353004, "learning_rate": 9.319979355645963e-05, "loss": 2.9758, "step": 12138 }, { "epoch": 0.7535539139611397, "grad_norm": 0.6003206489005933, "learning_rate": 9.319797504534478e-05, "loss": 2.9452, "step": 12139 }, { "epoch": 0.7536159910608976, "grad_norm": 0.6207214535215149, "learning_rate": 9.31961563088556e-05, "loss": 3.0262, "step": 12140 }, { "epoch": 0.7536780681606555, "grad_norm": 0.5960672485140018, "learning_rate": 9.319433734700159e-05, "loss": 2.9962, "step": 12141 }, { "epoch": 0.7537401452604134, "grad_norm": 0.6138700343766014, "learning_rate": 9.319251815979224e-05, "loss": 3.0414, "step": 12142 }, { "epoch": 0.7538022223601714, "grad_norm": 0.6105936287559102, "learning_rate": 9.319069874723703e-05, "loss": 3.0263, "step": 12143 }, { "epoch": 0.7538642994599293, "grad_norm": 0.5495407527983762, "learning_rate": 9.318887910934546e-05, "loss": 3.0086, "step": 12144 }, { "epoch": 0.7539263765596871, "grad_norm": 0.6079555754948579, "learning_rate": 9.318705924612703e-05, "loss": 3.0412, "step": 12145 }, { "epoch": 0.753988453659445, "grad_norm": 0.5605611766093958, "learning_rate": 9.318523915759123e-05, "loss": 3.0093, "step": 12146 }, { "epoch": 0.7540505307592029, "grad_norm": 0.4803684719236714, "learning_rate": 9.318341884374755e-05, "loss": 2.98, "step": 12147 }, { "epoch": 0.7541126078589608, "grad_norm": 0.5572699218738363, "learning_rate": 9.318159830460549e-05, "loss": 3.0788, "step": 12148 }, { "epoch": 0.7541746849587188, "grad_norm": 0.4572820708789443, "learning_rate": 9.317977754017455e-05, "loss": 2.9844, "step": 12149 }, { "epoch": 0.7542367620584767, "grad_norm": 0.4987414733616772, "learning_rate": 9.317795655046424e-05, "loss": 3.0735, "step": 12150 }, { "epoch": 0.7542988391582345, "grad_norm": 0.5067198221605258, "learning_rate": 9.317613533548403e-05, "loss": 2.9966, "step": 12151 }, { "epoch": 0.7543609162579924, "grad_norm": 0.5143184581193139, "learning_rate": 9.317431389524344e-05, "loss": 3.0793, "step": 12152 }, { "epoch": 0.7544229933577503, "grad_norm": 0.5955478653492325, "learning_rate": 9.317249222975199e-05, "loss": 3.0429, "step": 12153 }, { "epoch": 0.7544850704575082, "grad_norm": 0.5876342112864613, "learning_rate": 9.317067033901916e-05, "loss": 3.0486, "step": 12154 }, { "epoch": 0.7545471475572662, "grad_norm": 0.5203131907690464, "learning_rate": 9.316884822305446e-05, "loss": 3.1338, "step": 12155 }, { "epoch": 0.7546092246570241, "grad_norm": 0.5522214848628367, "learning_rate": 9.316702588186741e-05, "loss": 3.0944, "step": 12156 }, { "epoch": 0.7546713017567819, "grad_norm": 0.5219994048356358, "learning_rate": 9.316520331546748e-05, "loss": 3.0874, "step": 12157 }, { "epoch": 0.7547333788565398, "grad_norm": 0.5823938149358272, "learning_rate": 9.316338052386422e-05, "loss": 3.0647, "step": 12158 }, { "epoch": 0.7547954559562977, "grad_norm": 0.47038202690676395, "learning_rate": 9.316155750706715e-05, "loss": 3.0662, "step": 12159 }, { "epoch": 0.7548575330560556, "grad_norm": 0.550465297033052, "learning_rate": 9.315973426508573e-05, "loss": 3.1123, "step": 12160 }, { "epoch": 0.7549196101558135, "grad_norm": 0.4457365493210562, "learning_rate": 9.315791079792951e-05, "loss": 3.0365, "step": 12161 }, { "epoch": 0.7549816872555715, "grad_norm": 0.521614544227765, "learning_rate": 9.315608710560796e-05, "loss": 3.0552, "step": 12162 }, { "epoch": 0.7550437643553293, "grad_norm": 0.4812236685794745, "learning_rate": 9.315426318813067e-05, "loss": 2.9833, "step": 12163 }, { "epoch": 0.7551058414550872, "grad_norm": 0.4759315472566837, "learning_rate": 9.31524390455071e-05, "loss": 3.0957, "step": 12164 }, { "epoch": 0.7551679185548451, "grad_norm": 0.49479164993553804, "learning_rate": 9.315061467774677e-05, "loss": 3.0537, "step": 12165 }, { "epoch": 0.755229995654603, "grad_norm": 0.5356353508606805, "learning_rate": 9.314879008485921e-05, "loss": 2.9872, "step": 12166 }, { "epoch": 0.7552920727543609, "grad_norm": 0.4795705416614283, "learning_rate": 9.314696526685396e-05, "loss": 2.9751, "step": 12167 }, { "epoch": 0.7553541498541189, "grad_norm": 0.5417868560326711, "learning_rate": 9.314514022374049e-05, "loss": 2.9598, "step": 12168 }, { "epoch": 0.7554162269538767, "grad_norm": 0.49646651913447876, "learning_rate": 9.314331495552836e-05, "loss": 3.0241, "step": 12169 }, { "epoch": 0.7554783040536346, "grad_norm": 0.4677214304896329, "learning_rate": 9.314148946222708e-05, "loss": 2.9572, "step": 12170 }, { "epoch": 0.7555403811533925, "grad_norm": 0.5194422136463989, "learning_rate": 9.313966374384618e-05, "loss": 2.9827, "step": 12171 }, { "epoch": 0.7556024582531504, "grad_norm": 0.46837803069336237, "learning_rate": 9.313783780039518e-05, "loss": 2.9619, "step": 12172 }, { "epoch": 0.7556645353529083, "grad_norm": 0.5170135585862707, "learning_rate": 9.313601163188362e-05, "loss": 3.0532, "step": 12173 }, { "epoch": 0.7557266124526663, "grad_norm": 0.5025019632367551, "learning_rate": 9.313418523832099e-05, "loss": 2.9286, "step": 12174 }, { "epoch": 0.7557886895524241, "grad_norm": 0.456735189026517, "learning_rate": 9.313235861971685e-05, "loss": 3.043, "step": 12175 }, { "epoch": 0.755850766652182, "grad_norm": 0.5401352698389958, "learning_rate": 9.313053177608076e-05, "loss": 3.0089, "step": 12176 }, { "epoch": 0.7559128437519399, "grad_norm": 0.51764894396839, "learning_rate": 9.312870470742217e-05, "loss": 2.9394, "step": 12177 }, { "epoch": 0.7559749208516978, "grad_norm": 0.49822232017250034, "learning_rate": 9.312687741375068e-05, "loss": 3.0828, "step": 12178 }, { "epoch": 0.7560369979514557, "grad_norm": 0.5024317971247679, "learning_rate": 9.31250498950758e-05, "loss": 3.0572, "step": 12179 }, { "epoch": 0.7560990750512137, "grad_norm": 0.5254045789489328, "learning_rate": 9.312322215140705e-05, "loss": 3.0624, "step": 12180 }, { "epoch": 0.7561611521509715, "grad_norm": 0.46933857206255236, "learning_rate": 9.3121394182754e-05, "loss": 3.0498, "step": 12181 }, { "epoch": 0.7562232292507294, "grad_norm": 0.5385298101077977, "learning_rate": 9.311956598912617e-05, "loss": 3.1748, "step": 12182 }, { "epoch": 0.7562853063504873, "grad_norm": 0.48191145485655396, "learning_rate": 9.311773757053309e-05, "loss": 3.0777, "step": 12183 }, { "epoch": 0.7563473834502452, "grad_norm": 0.46066164393741504, "learning_rate": 9.31159089269843e-05, "loss": 3.0354, "step": 12184 }, { "epoch": 0.7564094605500031, "grad_norm": 0.587413962131176, "learning_rate": 9.311408005848936e-05, "loss": 3.0916, "step": 12185 }, { "epoch": 0.756471537649761, "grad_norm": 0.5378589739937497, "learning_rate": 9.311225096505778e-05, "loss": 3.1153, "step": 12186 }, { "epoch": 0.7565336147495189, "grad_norm": 0.5681446102270402, "learning_rate": 9.311042164669913e-05, "loss": 3.0548, "step": 12187 }, { "epoch": 0.7565956918492768, "grad_norm": 0.4545166415597538, "learning_rate": 9.310859210342295e-05, "loss": 2.9962, "step": 12188 }, { "epoch": 0.7566577689490347, "grad_norm": 0.5009557708530439, "learning_rate": 9.310676233523877e-05, "loss": 3.0398, "step": 12189 }, { "epoch": 0.7567198460487926, "grad_norm": 0.48611673870310873, "learning_rate": 9.310493234215617e-05, "loss": 2.9618, "step": 12190 }, { "epoch": 0.7567819231485505, "grad_norm": 0.5382993797493553, "learning_rate": 9.310310212418464e-05, "loss": 2.9851, "step": 12191 }, { "epoch": 0.7568440002483084, "grad_norm": 0.5556880362489537, "learning_rate": 9.310127168133378e-05, "loss": 3.0548, "step": 12192 }, { "epoch": 0.7569060773480663, "grad_norm": 0.5400855542773776, "learning_rate": 9.309944101361312e-05, "loss": 3.0773, "step": 12193 }, { "epoch": 0.7569681544478242, "grad_norm": 0.449218633900503, "learning_rate": 9.309761012103222e-05, "loss": 3.0514, "step": 12194 }, { "epoch": 0.7570302315475821, "grad_norm": 0.5177847683651353, "learning_rate": 9.309577900360061e-05, "loss": 3.0187, "step": 12195 }, { "epoch": 0.75709230864734, "grad_norm": 0.49222597849768635, "learning_rate": 9.309394766132788e-05, "loss": 3.0601, "step": 12196 }, { "epoch": 0.7571543857470979, "grad_norm": 0.43722041597172306, "learning_rate": 9.309211609422355e-05, "loss": 3.0429, "step": 12197 }, { "epoch": 0.7572164628468558, "grad_norm": 0.4564986589930163, "learning_rate": 9.309028430229718e-05, "loss": 3.0344, "step": 12198 }, { "epoch": 0.7572785399466136, "grad_norm": 0.5246278124741376, "learning_rate": 9.308845228555836e-05, "loss": 3.0324, "step": 12199 }, { "epoch": 0.7573406170463716, "grad_norm": 0.6019203063217171, "learning_rate": 9.30866200440166e-05, "loss": 3.0815, "step": 12200 }, { "epoch": 0.7574026941461295, "grad_norm": 0.6003246948735148, "learning_rate": 9.30847875776815e-05, "loss": 3.0606, "step": 12201 }, { "epoch": 0.7574647712458874, "grad_norm": 0.45321447212574417, "learning_rate": 9.30829548865626e-05, "loss": 3.111, "step": 12202 }, { "epoch": 0.7575268483456453, "grad_norm": 0.57019888386291, "learning_rate": 9.308112197066947e-05, "loss": 3.0426, "step": 12203 }, { "epoch": 0.7575889254454032, "grad_norm": 0.5890745996502433, "learning_rate": 9.307928883001167e-05, "loss": 3.071, "step": 12204 }, { "epoch": 0.757651002545161, "grad_norm": 0.49970045653836304, "learning_rate": 9.307745546459875e-05, "loss": 3.0341, "step": 12205 }, { "epoch": 0.757713079644919, "grad_norm": 0.5405144694241548, "learning_rate": 9.30756218744403e-05, "loss": 3.0823, "step": 12206 }, { "epoch": 0.7577751567446769, "grad_norm": 0.567214179134021, "learning_rate": 9.307378805954585e-05, "loss": 2.9936, "step": 12207 }, { "epoch": 0.7578372338444348, "grad_norm": 0.5662065614371333, "learning_rate": 9.307195401992501e-05, "loss": 3.0783, "step": 12208 }, { "epoch": 0.7578993109441927, "grad_norm": 0.5677565428769615, "learning_rate": 9.307011975558734e-05, "loss": 3.039, "step": 12209 }, { "epoch": 0.7579613880439506, "grad_norm": 0.49264815132992906, "learning_rate": 9.306828526654239e-05, "loss": 3.0491, "step": 12210 }, { "epoch": 0.7580234651437084, "grad_norm": 0.5391866568623667, "learning_rate": 9.306645055279973e-05, "loss": 3.0216, "step": 12211 }, { "epoch": 0.7580855422434664, "grad_norm": 0.49602761916684235, "learning_rate": 9.306461561436896e-05, "loss": 2.9867, "step": 12212 }, { "epoch": 0.7581476193432243, "grad_norm": 0.49957814898225533, "learning_rate": 9.306278045125962e-05, "loss": 3.0042, "step": 12213 }, { "epoch": 0.7582096964429822, "grad_norm": 0.583151104824181, "learning_rate": 9.306094506348131e-05, "loss": 3.0882, "step": 12214 }, { "epoch": 0.7582717735427401, "grad_norm": 0.5316984023315722, "learning_rate": 9.30591094510436e-05, "loss": 3.0983, "step": 12215 }, { "epoch": 0.758333850642498, "grad_norm": 0.5597413818476249, "learning_rate": 9.305727361395605e-05, "loss": 3.0412, "step": 12216 }, { "epoch": 0.7583959277422558, "grad_norm": 0.49369915929834385, "learning_rate": 9.305543755222825e-05, "loss": 3.0956, "step": 12217 }, { "epoch": 0.7584580048420138, "grad_norm": 0.5611814303237175, "learning_rate": 9.30536012658698e-05, "loss": 3.0548, "step": 12218 }, { "epoch": 0.7585200819417717, "grad_norm": 0.5640156776670626, "learning_rate": 9.305176475489025e-05, "loss": 3.1611, "step": 12219 }, { "epoch": 0.7585821590415296, "grad_norm": 0.5134365572604384, "learning_rate": 9.30499280192992e-05, "loss": 3.019, "step": 12220 }, { "epoch": 0.7586442361412875, "grad_norm": 0.5502799708702311, "learning_rate": 9.304809105910621e-05, "loss": 3.0314, "step": 12221 }, { "epoch": 0.7587063132410454, "grad_norm": 0.5069938222651714, "learning_rate": 9.304625387432089e-05, "loss": 2.9648, "step": 12222 }, { "epoch": 0.7587683903408032, "grad_norm": 0.5828746087573835, "learning_rate": 9.30444164649528e-05, "loss": 3.0482, "step": 12223 }, { "epoch": 0.7588304674405612, "grad_norm": 0.4597215292631819, "learning_rate": 9.304257883101155e-05, "loss": 3.0496, "step": 12224 }, { "epoch": 0.7588925445403191, "grad_norm": 0.4673124522624566, "learning_rate": 9.304074097250672e-05, "loss": 3.0333, "step": 12225 }, { "epoch": 0.758954621640077, "grad_norm": 0.4910366894460702, "learning_rate": 9.303890288944787e-05, "loss": 3.0784, "step": 12226 }, { "epoch": 0.7590166987398349, "grad_norm": 0.4235656206619157, "learning_rate": 9.303706458184464e-05, "loss": 3.016, "step": 12227 }, { "epoch": 0.7590787758395928, "grad_norm": 0.5209209813437818, "learning_rate": 9.30352260497066e-05, "loss": 3.0523, "step": 12228 }, { "epoch": 0.7591408529393506, "grad_norm": 0.438737243782081, "learning_rate": 9.303338729304334e-05, "loss": 3.033, "step": 12229 }, { "epoch": 0.7592029300391085, "grad_norm": 0.4886866755573106, "learning_rate": 9.303154831186444e-05, "loss": 3.0056, "step": 12230 }, { "epoch": 0.7592650071388665, "grad_norm": 0.45378362538929873, "learning_rate": 9.30297091061795e-05, "loss": 3.0993, "step": 12231 }, { "epoch": 0.7593270842386244, "grad_norm": 0.4636058512909226, "learning_rate": 9.302786967599813e-05, "loss": 2.9692, "step": 12232 }, { "epoch": 0.7593891613383823, "grad_norm": 0.5447694025055043, "learning_rate": 9.302603002132992e-05, "loss": 3.098, "step": 12233 }, { "epoch": 0.7594512384381402, "grad_norm": 0.4916355277427198, "learning_rate": 9.302419014218448e-05, "loss": 2.9947, "step": 12234 }, { "epoch": 0.759513315537898, "grad_norm": 0.49685599872556796, "learning_rate": 9.302235003857138e-05, "loss": 2.9918, "step": 12235 }, { "epoch": 0.7595753926376559, "grad_norm": 0.5294046890418619, "learning_rate": 9.302050971050023e-05, "loss": 3.1203, "step": 12236 }, { "epoch": 0.7596374697374139, "grad_norm": 0.5792024727142884, "learning_rate": 9.301866915798065e-05, "loss": 3.1089, "step": 12237 }, { "epoch": 0.7596995468371718, "grad_norm": 0.5505574360074468, "learning_rate": 9.301682838102223e-05, "loss": 2.962, "step": 12238 }, { "epoch": 0.7597616239369297, "grad_norm": 0.4999202128169736, "learning_rate": 9.301498737963457e-05, "loss": 3.0974, "step": 12239 }, { "epoch": 0.7598237010366876, "grad_norm": 0.5098829755502994, "learning_rate": 9.301314615382727e-05, "loss": 2.9658, "step": 12240 }, { "epoch": 0.7598857781364454, "grad_norm": 0.5618901125325655, "learning_rate": 9.301130470360995e-05, "loss": 2.9822, "step": 12241 }, { "epoch": 0.7599478552362033, "grad_norm": 0.592632346039538, "learning_rate": 9.300946302899221e-05, "loss": 3.038, "step": 12242 }, { "epoch": 0.7600099323359613, "grad_norm": 0.44243070835764264, "learning_rate": 9.300762112998367e-05, "loss": 3.0426, "step": 12243 }, { "epoch": 0.7600720094357192, "grad_norm": 0.5670896156695527, "learning_rate": 9.300577900659393e-05, "loss": 2.9551, "step": 12244 }, { "epoch": 0.7601340865354771, "grad_norm": 0.4851959408850517, "learning_rate": 9.300393665883258e-05, "loss": 2.9386, "step": 12245 }, { "epoch": 0.760196163635235, "grad_norm": 0.5728095792622042, "learning_rate": 9.300209408670928e-05, "loss": 2.9875, "step": 12246 }, { "epoch": 0.7602582407349928, "grad_norm": 0.46173658449483984, "learning_rate": 9.30002512902336e-05, "loss": 3.028, "step": 12247 }, { "epoch": 0.7603203178347507, "grad_norm": 0.5656284121415717, "learning_rate": 9.299840826941518e-05, "loss": 3.0427, "step": 12248 }, { "epoch": 0.7603823949345087, "grad_norm": 0.46059351548211497, "learning_rate": 9.299656502426361e-05, "loss": 2.9938, "step": 12249 }, { "epoch": 0.7604444720342666, "grad_norm": 0.5039473228571206, "learning_rate": 9.299472155478855e-05, "loss": 3.0385, "step": 12250 }, { "epoch": 0.7605065491340245, "grad_norm": 0.5269871388988723, "learning_rate": 9.299287786099957e-05, "loss": 3.0583, "step": 12251 }, { "epoch": 0.7605686262337824, "grad_norm": 0.5256514061589584, "learning_rate": 9.299103394290631e-05, "loss": 3.0523, "step": 12252 }, { "epoch": 0.7606307033335402, "grad_norm": 0.4721119481555861, "learning_rate": 9.298918980051839e-05, "loss": 3.0887, "step": 12253 }, { "epoch": 0.7606927804332981, "grad_norm": 0.5245458625528472, "learning_rate": 9.298734543384544e-05, "loss": 3.0218, "step": 12254 }, { "epoch": 0.760754857533056, "grad_norm": 0.5023928248994073, "learning_rate": 9.298550084289705e-05, "loss": 3.0986, "step": 12255 }, { "epoch": 0.760816934632814, "grad_norm": 0.4713391640595658, "learning_rate": 9.298365602768289e-05, "loss": 3.0704, "step": 12256 }, { "epoch": 0.7608790117325719, "grad_norm": 0.4861674828616773, "learning_rate": 9.298181098821256e-05, "loss": 3.0893, "step": 12257 }, { "epoch": 0.7609410888323298, "grad_norm": 0.44736667807471275, "learning_rate": 9.297996572449567e-05, "loss": 2.956, "step": 12258 }, { "epoch": 0.7610031659320876, "grad_norm": 0.48075792867874007, "learning_rate": 9.297812023654189e-05, "loss": 3.0386, "step": 12259 }, { "epoch": 0.7610652430318455, "grad_norm": 0.4757390020746041, "learning_rate": 9.297627452436082e-05, "loss": 3.0566, "step": 12260 }, { "epoch": 0.7611273201316034, "grad_norm": 0.5244780204884172, "learning_rate": 9.297442858796208e-05, "loss": 3.0549, "step": 12261 }, { "epoch": 0.7611893972313614, "grad_norm": 0.5861009242717808, "learning_rate": 9.297258242735532e-05, "loss": 2.999, "step": 12262 }, { "epoch": 0.7612514743311193, "grad_norm": 0.575428188571398, "learning_rate": 9.297073604255015e-05, "loss": 3.0737, "step": 12263 }, { "epoch": 0.7613135514308772, "grad_norm": 0.546130600151185, "learning_rate": 9.296888943355623e-05, "loss": 3.0045, "step": 12264 }, { "epoch": 0.761375628530635, "grad_norm": 0.5202524156216474, "learning_rate": 9.296704260038319e-05, "loss": 3.075, "step": 12265 }, { "epoch": 0.7614377056303929, "grad_norm": 0.4760760654511459, "learning_rate": 9.296519554304066e-05, "loss": 2.9761, "step": 12266 }, { "epoch": 0.7614997827301508, "grad_norm": 0.513363009230818, "learning_rate": 9.296334826153826e-05, "loss": 2.9377, "step": 12267 }, { "epoch": 0.7615618598299088, "grad_norm": 0.4632658101897548, "learning_rate": 9.296150075588566e-05, "loss": 3.0917, "step": 12268 }, { "epoch": 0.7616239369296667, "grad_norm": 0.5292031180475174, "learning_rate": 9.295965302609246e-05, "loss": 3.0039, "step": 12269 }, { "epoch": 0.7616860140294246, "grad_norm": 0.46585493938734013, "learning_rate": 9.295780507216834e-05, "loss": 2.9406, "step": 12270 }, { "epoch": 0.7617480911291824, "grad_norm": 0.4896552628238121, "learning_rate": 9.29559568941229e-05, "loss": 3.0, "step": 12271 }, { "epoch": 0.7618101682289403, "grad_norm": 0.4436571238934301, "learning_rate": 9.295410849196584e-05, "loss": 2.9956, "step": 12272 }, { "epoch": 0.7618722453286982, "grad_norm": 0.5672277872369071, "learning_rate": 9.295225986570675e-05, "loss": 3.0781, "step": 12273 }, { "epoch": 0.7619343224284562, "grad_norm": 0.6062358549768501, "learning_rate": 9.295041101535529e-05, "loss": 3.1202, "step": 12274 }, { "epoch": 0.7619963995282141, "grad_norm": 0.5002382426102706, "learning_rate": 9.294856194092111e-05, "loss": 3.0523, "step": 12275 }, { "epoch": 0.762058476627972, "grad_norm": 0.5607607496611362, "learning_rate": 9.294671264241386e-05, "loss": 2.9999, "step": 12276 }, { "epoch": 0.7621205537277298, "grad_norm": 0.5799573779065954, "learning_rate": 9.294486311984317e-05, "loss": 2.9479, "step": 12277 }, { "epoch": 0.7621826308274877, "grad_norm": 0.48064916991437245, "learning_rate": 9.294301337321872e-05, "loss": 3.0341, "step": 12278 }, { "epoch": 0.7622447079272456, "grad_norm": 0.7066896572361819, "learning_rate": 9.294116340255015e-05, "loss": 3.0747, "step": 12279 }, { "epoch": 0.7623067850270036, "grad_norm": 0.6712030110881588, "learning_rate": 9.293931320784708e-05, "loss": 3.0407, "step": 12280 }, { "epoch": 0.7623688621267615, "grad_norm": 0.46805871216602235, "learning_rate": 9.293746278911921e-05, "loss": 3.004, "step": 12281 }, { "epoch": 0.7624309392265194, "grad_norm": 0.5521050095051583, "learning_rate": 9.293561214637617e-05, "loss": 3.1008, "step": 12282 }, { "epoch": 0.7624930163262772, "grad_norm": 0.582386318172952, "learning_rate": 9.293376127962762e-05, "loss": 3.0327, "step": 12283 }, { "epoch": 0.7625550934260351, "grad_norm": 0.5160453989938367, "learning_rate": 9.293191018888322e-05, "loss": 2.995, "step": 12284 }, { "epoch": 0.762617170525793, "grad_norm": 0.4755246007323701, "learning_rate": 9.29300588741526e-05, "loss": 3.0409, "step": 12285 }, { "epoch": 0.762679247625551, "grad_norm": 0.48117378114328024, "learning_rate": 9.292820733544546e-05, "loss": 3.0275, "step": 12286 }, { "epoch": 0.7627413247253089, "grad_norm": 0.5074508846644727, "learning_rate": 9.292635557277143e-05, "loss": 3.0631, "step": 12287 }, { "epoch": 0.7628034018250668, "grad_norm": 0.5785988077534465, "learning_rate": 9.29245035861402e-05, "loss": 3.0585, "step": 12288 }, { "epoch": 0.7628654789248246, "grad_norm": 0.48976900421743896, "learning_rate": 9.29226513755614e-05, "loss": 3.0403, "step": 12289 }, { "epoch": 0.7629275560245825, "grad_norm": 0.45474153767807585, "learning_rate": 9.292079894104471e-05, "loss": 3.159, "step": 12290 }, { "epoch": 0.7629896331243404, "grad_norm": 0.5171374766426295, "learning_rate": 9.29189462825998e-05, "loss": 3.0406, "step": 12291 }, { "epoch": 0.7630517102240983, "grad_norm": 0.5024788326341029, "learning_rate": 9.291709340023632e-05, "loss": 3.1159, "step": 12292 }, { "epoch": 0.7631137873238563, "grad_norm": 0.475490928731847, "learning_rate": 9.291524029396394e-05, "loss": 3.0296, "step": 12293 }, { "epoch": 0.7631758644236142, "grad_norm": 0.42675460011126476, "learning_rate": 9.291338696379235e-05, "loss": 3.0564, "step": 12294 }, { "epoch": 0.763237941523372, "grad_norm": 0.4587921876504536, "learning_rate": 9.29115334097312e-05, "loss": 3.082, "step": 12295 }, { "epoch": 0.7633000186231299, "grad_norm": 0.4131363641788894, "learning_rate": 9.290967963179015e-05, "loss": 2.9669, "step": 12296 }, { "epoch": 0.7633620957228878, "grad_norm": 0.5263999688126749, "learning_rate": 9.290782562997888e-05, "loss": 3.1098, "step": 12297 }, { "epoch": 0.7634241728226457, "grad_norm": 0.4470896478807256, "learning_rate": 9.290597140430708e-05, "loss": 3.1717, "step": 12298 }, { "epoch": 0.7634862499224037, "grad_norm": 0.4335265236366423, "learning_rate": 9.290411695478442e-05, "loss": 2.9829, "step": 12299 }, { "epoch": 0.7635483270221616, "grad_norm": 0.46793834033420334, "learning_rate": 9.290226228142057e-05, "loss": 2.9821, "step": 12300 }, { "epoch": 0.7636104041219194, "grad_norm": 0.4538011603242222, "learning_rate": 9.290040738422517e-05, "loss": 3.0035, "step": 12301 }, { "epoch": 0.7636724812216773, "grad_norm": 0.4585556462849239, "learning_rate": 9.289855226320796e-05, "loss": 3.0131, "step": 12302 }, { "epoch": 0.7637345583214352, "grad_norm": 0.4431791569445366, "learning_rate": 9.289669691837857e-05, "loss": 3.0669, "step": 12303 }, { "epoch": 0.7637966354211931, "grad_norm": 0.5055141965078144, "learning_rate": 9.28948413497467e-05, "loss": 3.0914, "step": 12304 }, { "epoch": 0.763858712520951, "grad_norm": 0.5011147351297202, "learning_rate": 9.289298555732204e-05, "loss": 3.0304, "step": 12305 }, { "epoch": 0.763920789620709, "grad_norm": 0.44864661229141783, "learning_rate": 9.289112954111426e-05, "loss": 3.0188, "step": 12306 }, { "epoch": 0.7639828667204668, "grad_norm": 0.48556500045814077, "learning_rate": 9.288927330113304e-05, "loss": 3.03, "step": 12307 }, { "epoch": 0.7640449438202247, "grad_norm": 0.4432388007270866, "learning_rate": 9.288741683738809e-05, "loss": 2.9853, "step": 12308 }, { "epoch": 0.7641070209199826, "grad_norm": 0.4526064964500783, "learning_rate": 9.288556014988905e-05, "loss": 3.07, "step": 12309 }, { "epoch": 0.7641690980197405, "grad_norm": 0.526555128640216, "learning_rate": 9.288370323864563e-05, "loss": 3.1113, "step": 12310 }, { "epoch": 0.7642311751194985, "grad_norm": 0.45355638161535783, "learning_rate": 9.288184610366755e-05, "loss": 3.0828, "step": 12311 }, { "epoch": 0.7642932522192564, "grad_norm": 0.47377628630934077, "learning_rate": 9.287998874496445e-05, "loss": 3.078, "step": 12312 }, { "epoch": 0.7643553293190142, "grad_norm": 0.4478387617036096, "learning_rate": 9.287813116254602e-05, "loss": 3.0327, "step": 12313 }, { "epoch": 0.7644174064187721, "grad_norm": 0.50759827423381, "learning_rate": 9.2876273356422e-05, "loss": 3.0024, "step": 12314 }, { "epoch": 0.76447948351853, "grad_norm": 0.46343615934932997, "learning_rate": 9.287441532660204e-05, "loss": 3.0293, "step": 12315 }, { "epoch": 0.7645415606182879, "grad_norm": 0.4620950517667685, "learning_rate": 9.287255707309585e-05, "loss": 2.9807, "step": 12316 }, { "epoch": 0.7646036377180458, "grad_norm": 0.4488338480054159, "learning_rate": 9.28706985959131e-05, "loss": 3.0673, "step": 12317 }, { "epoch": 0.7646657148178038, "grad_norm": 0.5205826124083806, "learning_rate": 9.286883989506354e-05, "loss": 3.0473, "step": 12318 }, { "epoch": 0.7647277919175616, "grad_norm": 0.4812371691937864, "learning_rate": 9.286698097055684e-05, "loss": 3.0546, "step": 12319 }, { "epoch": 0.7647898690173195, "grad_norm": 0.4263563384538156, "learning_rate": 9.286512182240268e-05, "loss": 3.0726, "step": 12320 }, { "epoch": 0.7648519461170774, "grad_norm": 0.4761761365516739, "learning_rate": 9.286326245061077e-05, "loss": 3.0823, "step": 12321 }, { "epoch": 0.7649140232168353, "grad_norm": 0.4868696024917502, "learning_rate": 9.28614028551908e-05, "loss": 3.1172, "step": 12322 }, { "epoch": 0.7649761003165932, "grad_norm": 0.48533839125386224, "learning_rate": 9.285954303615252e-05, "loss": 3.0827, "step": 12323 }, { "epoch": 0.7650381774163512, "grad_norm": 0.44262143198431925, "learning_rate": 9.285768299350557e-05, "loss": 3.015, "step": 12324 }, { "epoch": 0.765100254516109, "grad_norm": 0.46675329922744674, "learning_rate": 9.28558227272597e-05, "loss": 3.0137, "step": 12325 }, { "epoch": 0.7651623316158669, "grad_norm": 0.5033048249164817, "learning_rate": 9.28539622374246e-05, "loss": 3.0412, "step": 12326 }, { "epoch": 0.7652244087156248, "grad_norm": 0.476247167228424, "learning_rate": 9.285210152400995e-05, "loss": 2.9992, "step": 12327 }, { "epoch": 0.7652864858153827, "grad_norm": 0.4428454373802452, "learning_rate": 9.285024058702553e-05, "loss": 2.9797, "step": 12328 }, { "epoch": 0.7653485629151406, "grad_norm": 0.5318090358490847, "learning_rate": 9.284837942648097e-05, "loss": 3.0799, "step": 12329 }, { "epoch": 0.7654106400148986, "grad_norm": 0.5171168449059426, "learning_rate": 9.284651804238602e-05, "loss": 3.054, "step": 12330 }, { "epoch": 0.7654727171146564, "grad_norm": 0.5046849111957113, "learning_rate": 9.284465643475038e-05, "loss": 3.0447, "step": 12331 }, { "epoch": 0.7655347942144143, "grad_norm": 0.5033113383335761, "learning_rate": 9.284279460358378e-05, "loss": 3.0518, "step": 12332 }, { "epoch": 0.7655968713141722, "grad_norm": 0.5080159880007036, "learning_rate": 9.28409325488959e-05, "loss": 3.0523, "step": 12333 }, { "epoch": 0.7656589484139301, "grad_norm": 0.4937427713372032, "learning_rate": 9.28390702706965e-05, "loss": 3.0193, "step": 12334 }, { "epoch": 0.765721025513688, "grad_norm": 0.5242812061199519, "learning_rate": 9.283720776899525e-05, "loss": 3.08, "step": 12335 }, { "epoch": 0.765783102613446, "grad_norm": 0.45326209460773964, "learning_rate": 9.28353450438019e-05, "loss": 3.0196, "step": 12336 }, { "epoch": 0.7658451797132038, "grad_norm": 0.5335704669927314, "learning_rate": 9.283348209512615e-05, "loss": 3.0897, "step": 12337 }, { "epoch": 0.7659072568129617, "grad_norm": 0.512940353690532, "learning_rate": 9.283161892297771e-05, "loss": 3.0127, "step": 12338 }, { "epoch": 0.7659693339127196, "grad_norm": 0.5080646769058981, "learning_rate": 9.282975552736635e-05, "loss": 3.0206, "step": 12339 }, { "epoch": 0.7660314110124775, "grad_norm": 0.5041143298241254, "learning_rate": 9.282789190830173e-05, "loss": 2.9543, "step": 12340 }, { "epoch": 0.7660934881122354, "grad_norm": 0.48691772814555595, "learning_rate": 9.282602806579361e-05, "loss": 2.9564, "step": 12341 }, { "epoch": 0.7661555652119934, "grad_norm": 0.5196910411647547, "learning_rate": 9.282416399985172e-05, "loss": 2.994, "step": 12342 }, { "epoch": 0.7662176423117512, "grad_norm": 0.4538741167680262, "learning_rate": 9.282229971048575e-05, "loss": 3.0104, "step": 12343 }, { "epoch": 0.7662797194115091, "grad_norm": 0.459004844046777, "learning_rate": 9.282043519770545e-05, "loss": 3.0413, "step": 12344 }, { "epoch": 0.766341796511267, "grad_norm": 0.5190706721149998, "learning_rate": 9.281857046152056e-05, "loss": 3.0153, "step": 12345 }, { "epoch": 0.7664038736110249, "grad_norm": 0.5399721735389807, "learning_rate": 9.281670550194078e-05, "loss": 3.1099, "step": 12346 }, { "epoch": 0.7664659507107828, "grad_norm": 0.5091518053527969, "learning_rate": 9.281484031897584e-05, "loss": 3.0154, "step": 12347 }, { "epoch": 0.7665280278105407, "grad_norm": 0.4754920569156549, "learning_rate": 9.281297491263552e-05, "loss": 3.0417, "step": 12348 }, { "epoch": 0.7665901049102986, "grad_norm": 0.5435067948335118, "learning_rate": 9.281110928292948e-05, "loss": 3.1289, "step": 12349 }, { "epoch": 0.7666521820100565, "grad_norm": 0.5092554572313484, "learning_rate": 9.280924342986752e-05, "loss": 2.9694, "step": 12350 }, { "epoch": 0.7667142591098144, "grad_norm": 0.512060211638407, "learning_rate": 9.280737735345933e-05, "loss": 2.9862, "step": 12351 }, { "epoch": 0.7667763362095723, "grad_norm": 0.5244079248506054, "learning_rate": 9.280551105371466e-05, "loss": 3.2224, "step": 12352 }, { "epoch": 0.7668384133093302, "grad_norm": 0.5780066807271699, "learning_rate": 9.280364453064325e-05, "loss": 3.0416, "step": 12353 }, { "epoch": 0.7669004904090881, "grad_norm": 0.5208955505085122, "learning_rate": 9.280177778425484e-05, "loss": 3.0939, "step": 12354 }, { "epoch": 0.766962567508846, "grad_norm": 0.5805789827439588, "learning_rate": 9.279991081455917e-05, "loss": 2.9899, "step": 12355 }, { "epoch": 0.7670246446086039, "grad_norm": 0.5778326377846128, "learning_rate": 9.279804362156596e-05, "loss": 3.0227, "step": 12356 }, { "epoch": 0.7670867217083618, "grad_norm": 0.49226599778798585, "learning_rate": 9.279617620528497e-05, "loss": 3.0609, "step": 12357 }, { "epoch": 0.7671487988081197, "grad_norm": 0.5696219090856487, "learning_rate": 9.279430856572594e-05, "loss": 3.0304, "step": 12358 }, { "epoch": 0.7672108759078776, "grad_norm": 0.5460125252890468, "learning_rate": 9.279244070289861e-05, "loss": 3.0469, "step": 12359 }, { "epoch": 0.7672729530076355, "grad_norm": 0.5235010364744461, "learning_rate": 9.279057261681274e-05, "loss": 3.0198, "step": 12360 }, { "epoch": 0.7673350301073933, "grad_norm": 0.49484477176446345, "learning_rate": 9.278870430747805e-05, "loss": 2.9248, "step": 12361 }, { "epoch": 0.7673971072071513, "grad_norm": 0.5321816241689806, "learning_rate": 9.278683577490432e-05, "loss": 2.9414, "step": 12362 }, { "epoch": 0.7674591843069092, "grad_norm": 0.5286237652561925, "learning_rate": 9.278496701910127e-05, "loss": 3.0912, "step": 12363 }, { "epoch": 0.7675212614066671, "grad_norm": 0.5737280273385555, "learning_rate": 9.278309804007867e-05, "loss": 3.1296, "step": 12364 }, { "epoch": 0.767583338506425, "grad_norm": 0.48398474384629403, "learning_rate": 9.278122883784626e-05, "loss": 3.2223, "step": 12365 }, { "epoch": 0.7676454156061828, "grad_norm": 0.5144759871863532, "learning_rate": 9.277935941241379e-05, "loss": 2.9897, "step": 12366 }, { "epoch": 0.7677074927059407, "grad_norm": 0.4721298754690059, "learning_rate": 9.2777489763791e-05, "loss": 2.9061, "step": 12367 }, { "epoch": 0.7677695698056987, "grad_norm": 0.5727250530133162, "learning_rate": 9.277561989198768e-05, "loss": 3.0967, "step": 12368 }, { "epoch": 0.7678316469054566, "grad_norm": 0.5556902887628136, "learning_rate": 9.277374979701357e-05, "loss": 3.0605, "step": 12369 }, { "epoch": 0.7678937240052145, "grad_norm": 0.45969892032526377, "learning_rate": 9.277187947887839e-05, "loss": 3.0382, "step": 12370 }, { "epoch": 0.7679558011049724, "grad_norm": 0.5682733739433504, "learning_rate": 9.277000893759198e-05, "loss": 3.0705, "step": 12371 }, { "epoch": 0.7680178782047302, "grad_norm": 0.49008614172171633, "learning_rate": 9.276813817316404e-05, "loss": 3.0167, "step": 12372 }, { "epoch": 0.7680799553044881, "grad_norm": 0.5099556521226188, "learning_rate": 9.276626718560432e-05, "loss": 3.0724, "step": 12373 }, { "epoch": 0.7681420324042461, "grad_norm": 0.5395928137792046, "learning_rate": 9.276439597492262e-05, "loss": 3.0955, "step": 12374 }, { "epoch": 0.768204109504004, "grad_norm": 0.4736538282628096, "learning_rate": 9.276252454112866e-05, "loss": 2.9815, "step": 12375 }, { "epoch": 0.7682661866037619, "grad_norm": 0.5171145684488188, "learning_rate": 9.276065288423225e-05, "loss": 2.9484, "step": 12376 }, { "epoch": 0.7683282637035198, "grad_norm": 0.5429183881717224, "learning_rate": 9.275878100424314e-05, "loss": 3.0538, "step": 12377 }, { "epoch": 0.7683903408032776, "grad_norm": 0.5179022583782058, "learning_rate": 9.275690890117106e-05, "loss": 2.9964, "step": 12378 }, { "epoch": 0.7684524179030355, "grad_norm": 0.4895093807884097, "learning_rate": 9.275503657502582e-05, "loss": 3.0698, "step": 12379 }, { "epoch": 0.7685144950027935, "grad_norm": 0.561143457966997, "learning_rate": 9.275316402581719e-05, "loss": 3.0978, "step": 12380 }, { "epoch": 0.7685765721025514, "grad_norm": 0.48751143234248306, "learning_rate": 9.27512912535549e-05, "loss": 3.0285, "step": 12381 }, { "epoch": 0.7686386492023093, "grad_norm": 0.5233941700779403, "learning_rate": 9.274941825824877e-05, "loss": 3.0977, "step": 12382 }, { "epoch": 0.7687007263020672, "grad_norm": 0.6067737421012434, "learning_rate": 9.274754503990855e-05, "loss": 3.0292, "step": 12383 }, { "epoch": 0.768762803401825, "grad_norm": 0.4909600589845649, "learning_rate": 9.2745671598544e-05, "loss": 2.9809, "step": 12384 }, { "epoch": 0.7688248805015829, "grad_norm": 0.5189109483625055, "learning_rate": 9.274379793416489e-05, "loss": 3.027, "step": 12385 }, { "epoch": 0.7688869576013408, "grad_norm": 0.6218443599417859, "learning_rate": 9.274192404678101e-05, "loss": 3.0971, "step": 12386 }, { "epoch": 0.7689490347010988, "grad_norm": 0.5403316595013151, "learning_rate": 9.274004993640215e-05, "loss": 3.0829, "step": 12387 }, { "epoch": 0.7690111118008567, "grad_norm": 0.6011698791589754, "learning_rate": 9.273817560303808e-05, "loss": 3.1099, "step": 12388 }, { "epoch": 0.7690731889006146, "grad_norm": 0.5052931278551869, "learning_rate": 9.273630104669856e-05, "loss": 2.9764, "step": 12389 }, { "epoch": 0.7691352660003724, "grad_norm": 0.5030994314011189, "learning_rate": 9.27344262673934e-05, "loss": 2.9225, "step": 12390 }, { "epoch": 0.7691973431001303, "grad_norm": 0.4722245664709782, "learning_rate": 9.273255126513234e-05, "loss": 2.9978, "step": 12391 }, { "epoch": 0.7692594201998882, "grad_norm": 0.5454122326434849, "learning_rate": 9.27306760399252e-05, "loss": 3.0725, "step": 12392 }, { "epoch": 0.7693214972996462, "grad_norm": 0.497933273823231, "learning_rate": 9.272880059178175e-05, "loss": 3.011, "step": 12393 }, { "epoch": 0.7693835743994041, "grad_norm": 0.4576376128512576, "learning_rate": 9.272692492071176e-05, "loss": 3.0866, "step": 12394 }, { "epoch": 0.769445651499162, "grad_norm": 0.5720610529605081, "learning_rate": 9.272504902672504e-05, "loss": 3.1195, "step": 12395 }, { "epoch": 0.7695077285989198, "grad_norm": 0.4970040567604764, "learning_rate": 9.272317290983136e-05, "loss": 3.0498, "step": 12396 }, { "epoch": 0.7695698056986777, "grad_norm": 0.6504204316511875, "learning_rate": 9.272129657004053e-05, "loss": 3.0084, "step": 12397 }, { "epoch": 0.7696318827984356, "grad_norm": 0.532409076718771, "learning_rate": 9.271942000736234e-05, "loss": 3.0069, "step": 12398 }, { "epoch": 0.7696939598981936, "grad_norm": 0.511899057319028, "learning_rate": 9.271754322180654e-05, "loss": 3.0219, "step": 12399 }, { "epoch": 0.7697560369979515, "grad_norm": 0.5602837187422053, "learning_rate": 9.271566621338295e-05, "loss": 2.9997, "step": 12400 }, { "epoch": 0.7698181140977094, "grad_norm": 0.5818282932011107, "learning_rate": 9.271378898210136e-05, "loss": 3.0018, "step": 12401 }, { "epoch": 0.7698801911974672, "grad_norm": 0.49246114359709353, "learning_rate": 9.271191152797157e-05, "loss": 2.97, "step": 12402 }, { "epoch": 0.7699422682972251, "grad_norm": 0.567348669603892, "learning_rate": 9.271003385100337e-05, "loss": 2.9882, "step": 12403 }, { "epoch": 0.770004345396983, "grad_norm": 0.5738787000998414, "learning_rate": 9.270815595120655e-05, "loss": 3.1082, "step": 12404 }, { "epoch": 0.770066422496741, "grad_norm": 0.550930922712942, "learning_rate": 9.270627782859089e-05, "loss": 3.0035, "step": 12405 }, { "epoch": 0.7701284995964989, "grad_norm": 0.5241773131319825, "learning_rate": 9.270439948316625e-05, "loss": 3.05, "step": 12406 }, { "epoch": 0.7701905766962568, "grad_norm": 0.5018553879873816, "learning_rate": 9.270252091494237e-05, "loss": 3.018, "step": 12407 }, { "epoch": 0.7702526537960146, "grad_norm": 0.5197536881459561, "learning_rate": 9.270064212392908e-05, "loss": 3.0141, "step": 12408 }, { "epoch": 0.7703147308957725, "grad_norm": 0.5358913412874858, "learning_rate": 9.269876311013617e-05, "loss": 3.016, "step": 12409 }, { "epoch": 0.7703768079955304, "grad_norm": 0.512239526057923, "learning_rate": 9.269688387357345e-05, "loss": 3.0361, "step": 12410 }, { "epoch": 0.7704388850952884, "grad_norm": 0.4974803320193458, "learning_rate": 9.269500441425072e-05, "loss": 2.9699, "step": 12411 }, { "epoch": 0.7705009621950463, "grad_norm": 0.5602158954952153, "learning_rate": 9.269312473217777e-05, "loss": 2.9186, "step": 12412 }, { "epoch": 0.7705630392948042, "grad_norm": 0.4556924169346028, "learning_rate": 9.269124482736445e-05, "loss": 2.986, "step": 12413 }, { "epoch": 0.770625116394562, "grad_norm": 0.47018719331406045, "learning_rate": 9.268936469982053e-05, "loss": 2.9569, "step": 12414 }, { "epoch": 0.7706871934943199, "grad_norm": 0.5142038006204741, "learning_rate": 9.268748434955584e-05, "loss": 3.0229, "step": 12415 }, { "epoch": 0.7707492705940778, "grad_norm": 0.543249318476608, "learning_rate": 9.268560377658018e-05, "loss": 3.0181, "step": 12416 }, { "epoch": 0.7708113476938357, "grad_norm": 0.46233909681943114, "learning_rate": 9.268372298090336e-05, "loss": 3.0483, "step": 12417 }, { "epoch": 0.7708734247935937, "grad_norm": 0.4613803255811979, "learning_rate": 9.26818419625352e-05, "loss": 2.9964, "step": 12418 }, { "epoch": 0.7709355018933516, "grad_norm": 0.5173652643069747, "learning_rate": 9.267996072148549e-05, "loss": 3.1514, "step": 12419 }, { "epoch": 0.7709975789931094, "grad_norm": 0.5375356196643991, "learning_rate": 9.267807925776408e-05, "loss": 3.1044, "step": 12420 }, { "epoch": 0.7710596560928673, "grad_norm": 0.44364473007625066, "learning_rate": 9.267619757138077e-05, "loss": 2.9807, "step": 12421 }, { "epoch": 0.7711217331926252, "grad_norm": 0.549439954639168, "learning_rate": 9.267431566234537e-05, "loss": 3.0709, "step": 12422 }, { "epoch": 0.7711838102923831, "grad_norm": 0.5668416684496964, "learning_rate": 9.267243353066771e-05, "loss": 3.1082, "step": 12423 }, { "epoch": 0.7712458873921411, "grad_norm": 0.5024828360838339, "learning_rate": 9.267055117635761e-05, "loss": 3.0367, "step": 12424 }, { "epoch": 0.771307964491899, "grad_norm": 0.5249431647462768, "learning_rate": 9.266866859942488e-05, "loss": 2.9803, "step": 12425 }, { "epoch": 0.7713700415916568, "grad_norm": 0.5345876136909714, "learning_rate": 9.266678579987935e-05, "loss": 2.9161, "step": 12426 }, { "epoch": 0.7714321186914147, "grad_norm": 0.531937771061714, "learning_rate": 9.266490277773083e-05, "loss": 3.0354, "step": 12427 }, { "epoch": 0.7714941957911726, "grad_norm": 0.4626593515020971, "learning_rate": 9.266301953298918e-05, "loss": 3.1364, "step": 12428 }, { "epoch": 0.7715562728909305, "grad_norm": 0.5073100318430384, "learning_rate": 9.266113606566417e-05, "loss": 3.0051, "step": 12429 }, { "epoch": 0.7716183499906885, "grad_norm": 0.4914057593654009, "learning_rate": 9.265925237576568e-05, "loss": 3.0528, "step": 12430 }, { "epoch": 0.7716804270904464, "grad_norm": 0.51166726355668, "learning_rate": 9.26573684633035e-05, "loss": 3.0948, "step": 12431 }, { "epoch": 0.7717425041902042, "grad_norm": 0.5441903567402911, "learning_rate": 9.265548432828748e-05, "loss": 3.0152, "step": 12432 }, { "epoch": 0.7718045812899621, "grad_norm": 0.4825412587082117, "learning_rate": 9.265359997072745e-05, "loss": 2.976, "step": 12433 }, { "epoch": 0.77186665838972, "grad_norm": 0.5267977112796396, "learning_rate": 9.265171539063323e-05, "loss": 2.9247, "step": 12434 }, { "epoch": 0.7719287354894779, "grad_norm": 0.49517991911923015, "learning_rate": 9.264983058801465e-05, "loss": 3.0142, "step": 12435 }, { "epoch": 0.7719908125892359, "grad_norm": 0.5029002356933175, "learning_rate": 9.264794556288156e-05, "loss": 3.0965, "step": 12436 }, { "epoch": 0.7720528896889938, "grad_norm": 0.48649508509647715, "learning_rate": 9.264606031524376e-05, "loss": 3.0303, "step": 12437 }, { "epoch": 0.7721149667887516, "grad_norm": 0.5377344551623959, "learning_rate": 9.264417484511114e-05, "loss": 3.0696, "step": 12438 }, { "epoch": 0.7721770438885095, "grad_norm": 0.5066426520501897, "learning_rate": 9.26422891524935e-05, "loss": 3.0941, "step": 12439 }, { "epoch": 0.7722391209882674, "grad_norm": 0.4991105781810328, "learning_rate": 9.264040323740069e-05, "loss": 3.114, "step": 12440 }, { "epoch": 0.7723011980880253, "grad_norm": 0.5405600657633756, "learning_rate": 9.263851709984252e-05, "loss": 3.1318, "step": 12441 }, { "epoch": 0.7723632751877832, "grad_norm": 0.49933308589001385, "learning_rate": 9.26366307398289e-05, "loss": 3.0412, "step": 12442 }, { "epoch": 0.7724253522875412, "grad_norm": 0.4473337513042398, "learning_rate": 9.263474415736959e-05, "loss": 3.0762, "step": 12443 }, { "epoch": 0.772487429387299, "grad_norm": 0.47107044307818846, "learning_rate": 9.263285735247447e-05, "loss": 2.9543, "step": 12444 }, { "epoch": 0.7725495064870569, "grad_norm": 0.5168999889841667, "learning_rate": 9.26309703251534e-05, "loss": 3.0652, "step": 12445 }, { "epoch": 0.7726115835868148, "grad_norm": 0.5143786597324485, "learning_rate": 9.26290830754162e-05, "loss": 3.0551, "step": 12446 }, { "epoch": 0.7726736606865727, "grad_norm": 0.4409433655379116, "learning_rate": 9.262719560327273e-05, "loss": 3.0292, "step": 12447 }, { "epoch": 0.7727357377863306, "grad_norm": 0.4501242095338618, "learning_rate": 9.262530790873283e-05, "loss": 3.037, "step": 12448 }, { "epoch": 0.7727978148860886, "grad_norm": 0.4428637082326313, "learning_rate": 9.262341999180634e-05, "loss": 2.9558, "step": 12449 }, { "epoch": 0.7728598919858464, "grad_norm": 0.4455950827540852, "learning_rate": 9.262153185250315e-05, "loss": 3.0057, "step": 12450 }, { "epoch": 0.7729219690856043, "grad_norm": 0.5668507114661998, "learning_rate": 9.261964349083305e-05, "loss": 3.0549, "step": 12451 }, { "epoch": 0.7729840461853622, "grad_norm": 0.45086148967653517, "learning_rate": 9.261775490680594e-05, "loss": 3.0654, "step": 12452 }, { "epoch": 0.7730461232851201, "grad_norm": 0.5033596830176158, "learning_rate": 9.261586610043166e-05, "loss": 2.9175, "step": 12453 }, { "epoch": 0.773108200384878, "grad_norm": 0.481624685922496, "learning_rate": 9.261397707172006e-05, "loss": 3.1, "step": 12454 }, { "epoch": 0.773170277484636, "grad_norm": 0.47775602190283706, "learning_rate": 9.261208782068098e-05, "loss": 3.0758, "step": 12455 }, { "epoch": 0.7732323545843938, "grad_norm": 0.5679958531184667, "learning_rate": 9.261019834732432e-05, "loss": 3.0486, "step": 12456 }, { "epoch": 0.7732944316841517, "grad_norm": 0.5075905241353821, "learning_rate": 9.260830865165989e-05, "loss": 2.9922, "step": 12457 }, { "epoch": 0.7733565087839096, "grad_norm": 0.6038688440816503, "learning_rate": 9.260641873369758e-05, "loss": 3.0869, "step": 12458 }, { "epoch": 0.7734185858836675, "grad_norm": 0.6137300780611659, "learning_rate": 9.260452859344723e-05, "loss": 3.116, "step": 12459 }, { "epoch": 0.7734806629834254, "grad_norm": 0.6359655922348371, "learning_rate": 9.260263823091872e-05, "loss": 3.0214, "step": 12460 }, { "epoch": 0.7735427400831834, "grad_norm": 0.5705506075930634, "learning_rate": 9.26007476461219e-05, "loss": 3.0311, "step": 12461 }, { "epoch": 0.7736048171829412, "grad_norm": 0.513445321950461, "learning_rate": 9.259885683906664e-05, "loss": 3.1069, "step": 12462 }, { "epoch": 0.7736668942826991, "grad_norm": 0.5382545054841159, "learning_rate": 9.25969658097628e-05, "loss": 2.957, "step": 12463 }, { "epoch": 0.773728971382457, "grad_norm": 0.5455505955271651, "learning_rate": 9.259507455822025e-05, "loss": 3.025, "step": 12464 }, { "epoch": 0.7737910484822149, "grad_norm": 0.4944202551642658, "learning_rate": 9.259318308444886e-05, "loss": 3.0405, "step": 12465 }, { "epoch": 0.7738531255819728, "grad_norm": 0.5148524073797608, "learning_rate": 9.25912913884585e-05, "loss": 3.1514, "step": 12466 }, { "epoch": 0.7739152026817308, "grad_norm": 0.5394997689137923, "learning_rate": 9.258939947025901e-05, "loss": 2.9686, "step": 12467 }, { "epoch": 0.7739772797814886, "grad_norm": 0.48712694676328117, "learning_rate": 9.258750732986032e-05, "loss": 3.066, "step": 12468 }, { "epoch": 0.7740393568812465, "grad_norm": 0.5621172079318598, "learning_rate": 9.258561496727224e-05, "loss": 3.0617, "step": 12469 }, { "epoch": 0.7741014339810044, "grad_norm": 0.47867425897128196, "learning_rate": 9.258372238250468e-05, "loss": 3.065, "step": 12470 }, { "epoch": 0.7741635110807623, "grad_norm": 0.5737161058291935, "learning_rate": 9.258182957556749e-05, "loss": 2.9936, "step": 12471 }, { "epoch": 0.7742255881805202, "grad_norm": 0.5867060897416173, "learning_rate": 9.257993654647057e-05, "loss": 2.9671, "step": 12472 }, { "epoch": 0.7742876652802781, "grad_norm": 0.6022019393765512, "learning_rate": 9.257804329522378e-05, "loss": 2.9743, "step": 12473 }, { "epoch": 0.774349742380036, "grad_norm": 0.5383405688727662, "learning_rate": 9.2576149821837e-05, "loss": 2.9868, "step": 12474 }, { "epoch": 0.7744118194797939, "grad_norm": 0.589958003012267, "learning_rate": 9.257425612632012e-05, "loss": 3.1054, "step": 12475 }, { "epoch": 0.7744738965795518, "grad_norm": 0.49107125269899066, "learning_rate": 9.257236220868299e-05, "loss": 3.0425, "step": 12476 }, { "epoch": 0.7745359736793097, "grad_norm": 0.5415356061787622, "learning_rate": 9.257046806893553e-05, "loss": 2.9928, "step": 12477 }, { "epoch": 0.7745980507790676, "grad_norm": 0.4834994432136462, "learning_rate": 9.256857370708759e-05, "loss": 2.9602, "step": 12478 }, { "epoch": 0.7746601278788255, "grad_norm": 0.43894291399702473, "learning_rate": 9.256667912314907e-05, "loss": 3.016, "step": 12479 }, { "epoch": 0.7747222049785834, "grad_norm": 0.5110797823959804, "learning_rate": 9.256478431712986e-05, "loss": 2.9675, "step": 12480 }, { "epoch": 0.7747842820783413, "grad_norm": 0.46172177141107873, "learning_rate": 9.256288928903984e-05, "loss": 2.9597, "step": 12481 }, { "epoch": 0.7748463591780992, "grad_norm": 0.507027033243146, "learning_rate": 9.256099403888888e-05, "loss": 3.0347, "step": 12482 }, { "epoch": 0.7749084362778571, "grad_norm": 0.45341503792250365, "learning_rate": 9.25590985666869e-05, "loss": 3.0133, "step": 12483 }, { "epoch": 0.774970513377615, "grad_norm": 0.4541998151907719, "learning_rate": 9.255720287244375e-05, "loss": 2.9961, "step": 12484 }, { "epoch": 0.7750325904773729, "grad_norm": 0.5044624157375395, "learning_rate": 9.255530695616936e-05, "loss": 3.0609, "step": 12485 }, { "epoch": 0.7750946675771307, "grad_norm": 0.44847008058442156, "learning_rate": 9.255341081787358e-05, "loss": 2.893, "step": 12486 }, { "epoch": 0.7751567446768887, "grad_norm": 0.49016939925656, "learning_rate": 9.255151445756635e-05, "loss": 3.0075, "step": 12487 }, { "epoch": 0.7752188217766466, "grad_norm": 0.47742944563149026, "learning_rate": 9.254961787525753e-05, "loss": 3.0445, "step": 12488 }, { "epoch": 0.7752808988764045, "grad_norm": 0.5139406900530714, "learning_rate": 9.254772107095702e-05, "loss": 3.1293, "step": 12489 }, { "epoch": 0.7753429759761624, "grad_norm": 0.49947157891496335, "learning_rate": 9.254582404467473e-05, "loss": 3.0031, "step": 12490 }, { "epoch": 0.7754050530759203, "grad_norm": 0.5043573356569934, "learning_rate": 9.254392679642054e-05, "loss": 3.0249, "step": 12491 }, { "epoch": 0.7754671301756781, "grad_norm": 0.5035460611619059, "learning_rate": 9.254202932620436e-05, "loss": 3.0331, "step": 12492 }, { "epoch": 0.7755292072754361, "grad_norm": 0.4719376263522945, "learning_rate": 9.254013163403607e-05, "loss": 2.9843, "step": 12493 }, { "epoch": 0.775591284375194, "grad_norm": 0.5875805008298389, "learning_rate": 9.25382337199256e-05, "loss": 3.0686, "step": 12494 }, { "epoch": 0.7756533614749519, "grad_norm": 0.5131463383037451, "learning_rate": 9.253633558388285e-05, "loss": 3.0439, "step": 12495 }, { "epoch": 0.7757154385747098, "grad_norm": 0.5050099432315728, "learning_rate": 9.253443722591771e-05, "loss": 3.02, "step": 12496 }, { "epoch": 0.7757775156744677, "grad_norm": 0.46734470478293205, "learning_rate": 9.253253864604006e-05, "loss": 3.1029, "step": 12497 }, { "epoch": 0.7758395927742255, "grad_norm": 0.5053153629257263, "learning_rate": 9.253063984425985e-05, "loss": 3.0869, "step": 12498 }, { "epoch": 0.7759016698739835, "grad_norm": 0.4885676497712236, "learning_rate": 9.252874082058696e-05, "loss": 2.96, "step": 12499 }, { "epoch": 0.7759637469737414, "grad_norm": 0.47310513075475646, "learning_rate": 9.25268415750313e-05, "loss": 3.0127, "step": 12500 }, { "epoch": 0.7760258240734993, "grad_norm": 0.4779613632474356, "learning_rate": 9.252494210760281e-05, "loss": 3.0615, "step": 12501 }, { "epoch": 0.7760879011732572, "grad_norm": 0.5048742828534051, "learning_rate": 9.252304241831134e-05, "loss": 2.9417, "step": 12502 }, { "epoch": 0.7761499782730151, "grad_norm": 0.443905212900628, "learning_rate": 9.252114250716685e-05, "loss": 3.0346, "step": 12503 }, { "epoch": 0.7762120553727729, "grad_norm": 0.4661970714205857, "learning_rate": 9.251924237417924e-05, "loss": 3.0994, "step": 12504 }, { "epoch": 0.7762741324725309, "grad_norm": 0.46918913616638563, "learning_rate": 9.251734201935843e-05, "loss": 3.0237, "step": 12505 }, { "epoch": 0.7763362095722888, "grad_norm": 0.44446000264010077, "learning_rate": 9.251544144271431e-05, "loss": 3.0782, "step": 12506 }, { "epoch": 0.7763982866720467, "grad_norm": 0.4761817693269117, "learning_rate": 9.251354064425681e-05, "loss": 3.0922, "step": 12507 }, { "epoch": 0.7764603637718046, "grad_norm": 0.47298891010046507, "learning_rate": 9.251163962399585e-05, "loss": 3.058, "step": 12508 }, { "epoch": 0.7765224408715625, "grad_norm": 0.5440699436263564, "learning_rate": 9.250973838194134e-05, "loss": 2.9702, "step": 12509 }, { "epoch": 0.7765845179713203, "grad_norm": 0.5395635957188673, "learning_rate": 9.250783691810322e-05, "loss": 3.0249, "step": 12510 }, { "epoch": 0.7766465950710782, "grad_norm": 0.4792420925678012, "learning_rate": 9.25059352324914e-05, "loss": 3.0194, "step": 12511 }, { "epoch": 0.7767086721708362, "grad_norm": 0.5235286462482597, "learning_rate": 9.250403332511579e-05, "loss": 3.0519, "step": 12512 }, { "epoch": 0.7767707492705941, "grad_norm": 0.6026507603318891, "learning_rate": 9.25021311959863e-05, "loss": 3.0149, "step": 12513 }, { "epoch": 0.776832826370352, "grad_norm": 0.5463395085533705, "learning_rate": 9.250022884511289e-05, "loss": 3.0318, "step": 12514 }, { "epoch": 0.7768949034701099, "grad_norm": 0.4999562482765216, "learning_rate": 9.249832627250548e-05, "loss": 3.0749, "step": 12515 }, { "epoch": 0.7769569805698677, "grad_norm": 0.4872947951786272, "learning_rate": 9.249642347817397e-05, "loss": 3.0801, "step": 12516 }, { "epoch": 0.7770190576696256, "grad_norm": 0.5584540292420838, "learning_rate": 9.24945204621283e-05, "loss": 3.0561, "step": 12517 }, { "epoch": 0.7770811347693836, "grad_norm": 0.5070969455284725, "learning_rate": 9.249261722437841e-05, "loss": 3.1284, "step": 12518 }, { "epoch": 0.7771432118691415, "grad_norm": 0.501964376004364, "learning_rate": 9.24907137649342e-05, "loss": 3.0972, "step": 12519 }, { "epoch": 0.7772052889688994, "grad_norm": 0.42879737208605545, "learning_rate": 9.248881008380566e-05, "loss": 2.9395, "step": 12520 }, { "epoch": 0.7772673660686573, "grad_norm": 0.5076593901823874, "learning_rate": 9.248690618100267e-05, "loss": 3.0408, "step": 12521 }, { "epoch": 0.7773294431684151, "grad_norm": 0.44352939051489154, "learning_rate": 9.248500205653517e-05, "loss": 2.9872, "step": 12522 }, { "epoch": 0.777391520268173, "grad_norm": 0.4902533569308544, "learning_rate": 9.248309771041309e-05, "loss": 3.0532, "step": 12523 }, { "epoch": 0.777453597367931, "grad_norm": 0.415692951986246, "learning_rate": 9.248119314264638e-05, "loss": 3.0305, "step": 12524 }, { "epoch": 0.7775156744676889, "grad_norm": 0.488151105702836, "learning_rate": 9.247928835324497e-05, "loss": 3.0036, "step": 12525 }, { "epoch": 0.7775777515674468, "grad_norm": 0.42449280262519273, "learning_rate": 9.247738334221881e-05, "loss": 3.0139, "step": 12526 }, { "epoch": 0.7776398286672047, "grad_norm": 0.43410369166773116, "learning_rate": 9.247547810957785e-05, "loss": 3.0629, "step": 12527 }, { "epoch": 0.7777019057669625, "grad_norm": 0.4866261162939822, "learning_rate": 9.247357265533197e-05, "loss": 3.021, "step": 12528 }, { "epoch": 0.7777639828667204, "grad_norm": 0.5610344922466017, "learning_rate": 9.247166697949117e-05, "loss": 3.0198, "step": 12529 }, { "epoch": 0.7778260599664784, "grad_norm": 0.4831820996088412, "learning_rate": 9.246976108206537e-05, "loss": 3.042, "step": 12530 }, { "epoch": 0.7778881370662363, "grad_norm": 0.5489882917632243, "learning_rate": 9.246785496306452e-05, "loss": 3.0074, "step": 12531 }, { "epoch": 0.7779502141659942, "grad_norm": 0.5608102797644228, "learning_rate": 9.246594862249855e-05, "loss": 3.0559, "step": 12532 }, { "epoch": 0.7780122912657521, "grad_norm": 0.5202542487176747, "learning_rate": 9.246404206037743e-05, "loss": 3.0419, "step": 12533 }, { "epoch": 0.7780743683655099, "grad_norm": 0.482905126531962, "learning_rate": 9.24621352767111e-05, "loss": 2.9203, "step": 12534 }, { "epoch": 0.7781364454652678, "grad_norm": 0.5407755901103309, "learning_rate": 9.246022827150949e-05, "loss": 2.9059, "step": 12535 }, { "epoch": 0.7781985225650258, "grad_norm": 0.5298650865862581, "learning_rate": 9.245832104478256e-05, "loss": 2.9812, "step": 12536 }, { "epoch": 0.7782605996647837, "grad_norm": 0.43269073687781795, "learning_rate": 9.245641359654027e-05, "loss": 3.053, "step": 12537 }, { "epoch": 0.7783226767645416, "grad_norm": 0.5176145432813264, "learning_rate": 9.245450592679256e-05, "loss": 3.0663, "step": 12538 }, { "epoch": 0.7783847538642995, "grad_norm": 0.4956293410391308, "learning_rate": 9.245259803554937e-05, "loss": 2.9485, "step": 12539 }, { "epoch": 0.7784468309640573, "grad_norm": 0.4623360671951766, "learning_rate": 9.24506899228207e-05, "loss": 3.0436, "step": 12540 }, { "epoch": 0.7785089080638152, "grad_norm": 0.4663906963828478, "learning_rate": 9.244878158861645e-05, "loss": 2.934, "step": 12541 }, { "epoch": 0.7785709851635731, "grad_norm": 0.5567358638200399, "learning_rate": 9.244687303294661e-05, "loss": 3.0738, "step": 12542 }, { "epoch": 0.7786330622633311, "grad_norm": 0.5268332376818926, "learning_rate": 9.244496425582114e-05, "loss": 2.9739, "step": 12543 }, { "epoch": 0.778695139363089, "grad_norm": 0.4967787811396482, "learning_rate": 9.244305525724998e-05, "loss": 3.1398, "step": 12544 }, { "epoch": 0.7787572164628469, "grad_norm": 0.53760563899046, "learning_rate": 9.24411460372431e-05, "loss": 3.0327, "step": 12545 }, { "epoch": 0.7788192935626047, "grad_norm": 0.5419671405125407, "learning_rate": 9.243923659581045e-05, "loss": 3.0438, "step": 12546 }, { "epoch": 0.7788813706623626, "grad_norm": 0.5022858878549136, "learning_rate": 9.243732693296201e-05, "loss": 3.08, "step": 12547 }, { "epoch": 0.7789434477621205, "grad_norm": 0.5699026647079403, "learning_rate": 9.243541704870773e-05, "loss": 3.0459, "step": 12548 }, { "epoch": 0.7790055248618785, "grad_norm": 0.4686851774535349, "learning_rate": 9.243350694305757e-05, "loss": 3.0214, "step": 12549 }, { "epoch": 0.7790676019616364, "grad_norm": 0.5684056214070041, "learning_rate": 9.243159661602152e-05, "loss": 3.0756, "step": 12550 }, { "epoch": 0.7791296790613943, "grad_norm": 0.5417038709139178, "learning_rate": 9.242968606760951e-05, "loss": 2.9404, "step": 12551 }, { "epoch": 0.7791917561611521, "grad_norm": 0.5678324140920645, "learning_rate": 9.242777529783154e-05, "loss": 3.0123, "step": 12552 }, { "epoch": 0.77925383326091, "grad_norm": 0.4651052877221119, "learning_rate": 9.242586430669757e-05, "loss": 3.0334, "step": 12553 }, { "epoch": 0.7793159103606679, "grad_norm": 0.4847139126524775, "learning_rate": 9.242395309421755e-05, "loss": 3.0062, "step": 12554 }, { "epoch": 0.7793779874604259, "grad_norm": 0.4763892593682013, "learning_rate": 9.242204166040148e-05, "loss": 2.9458, "step": 12555 }, { "epoch": 0.7794400645601838, "grad_norm": 0.44724658442451637, "learning_rate": 9.242013000525932e-05, "loss": 2.9844, "step": 12556 }, { "epoch": 0.7795021416599417, "grad_norm": 0.4588184949740181, "learning_rate": 9.241821812880105e-05, "loss": 3.0098, "step": 12557 }, { "epoch": 0.7795642187596995, "grad_norm": 0.5251452926132746, "learning_rate": 9.241630603103663e-05, "loss": 2.9177, "step": 12558 }, { "epoch": 0.7796262958594574, "grad_norm": 0.42596020568759047, "learning_rate": 9.241439371197604e-05, "loss": 2.9598, "step": 12559 }, { "epoch": 0.7796883729592153, "grad_norm": 0.4973028272450614, "learning_rate": 9.241248117162926e-05, "loss": 3.0073, "step": 12560 }, { "epoch": 0.7797504500589733, "grad_norm": 0.48498462952512794, "learning_rate": 9.241056841000628e-05, "loss": 3.0653, "step": 12561 }, { "epoch": 0.7798125271587312, "grad_norm": 0.45066448266207415, "learning_rate": 9.240865542711708e-05, "loss": 2.9428, "step": 12562 }, { "epoch": 0.7798746042584891, "grad_norm": 0.4521057240556427, "learning_rate": 9.240674222297161e-05, "loss": 3.0188, "step": 12563 }, { "epoch": 0.7799366813582469, "grad_norm": 0.504596414997597, "learning_rate": 9.240482879757987e-05, "loss": 3.1052, "step": 12564 }, { "epoch": 0.7799987584580048, "grad_norm": 0.49964865618818693, "learning_rate": 9.240291515095185e-05, "loss": 3.0244, "step": 12565 }, { "epoch": 0.7800608355577627, "grad_norm": 0.45819961338372406, "learning_rate": 9.240100128309753e-05, "loss": 3.0214, "step": 12566 }, { "epoch": 0.7801229126575207, "grad_norm": 0.5092393050795978, "learning_rate": 9.23990871940269e-05, "loss": 3.1008, "step": 12567 }, { "epoch": 0.7801849897572786, "grad_norm": 0.46077372987547405, "learning_rate": 9.239717288374992e-05, "loss": 2.9961, "step": 12568 }, { "epoch": 0.7802470668570365, "grad_norm": 0.5499526751271712, "learning_rate": 9.239525835227661e-05, "loss": 3.1173, "step": 12569 }, { "epoch": 0.7803091439567943, "grad_norm": 0.5636987099501983, "learning_rate": 9.239334359961695e-05, "loss": 3.1008, "step": 12570 }, { "epoch": 0.7803712210565522, "grad_norm": 0.4782001155863901, "learning_rate": 9.239142862578092e-05, "loss": 3.0073, "step": 12571 }, { "epoch": 0.7804332981563101, "grad_norm": 0.5393240473827652, "learning_rate": 9.238951343077851e-05, "loss": 3.0362, "step": 12572 }, { "epoch": 0.780495375256068, "grad_norm": 0.5008706321040333, "learning_rate": 9.238759801461971e-05, "loss": 3.0591, "step": 12573 }, { "epoch": 0.780557452355826, "grad_norm": 0.5350879987645867, "learning_rate": 9.238568237731456e-05, "loss": 3.04, "step": 12574 }, { "epoch": 0.7806195294555839, "grad_norm": 0.5119289228661331, "learning_rate": 9.238376651887298e-05, "loss": 3.0614, "step": 12575 }, { "epoch": 0.7806816065553417, "grad_norm": 0.5597701057206833, "learning_rate": 9.2381850439305e-05, "loss": 3.1155, "step": 12576 }, { "epoch": 0.7807436836550996, "grad_norm": 0.5237404814699637, "learning_rate": 9.237993413862063e-05, "loss": 3.0802, "step": 12577 }, { "epoch": 0.7808057607548575, "grad_norm": 0.5505512379527151, "learning_rate": 9.237801761682985e-05, "loss": 3.0549, "step": 12578 }, { "epoch": 0.7808678378546154, "grad_norm": 0.5317649869631693, "learning_rate": 9.237610087394267e-05, "loss": 3.039, "step": 12579 }, { "epoch": 0.7809299149543734, "grad_norm": 0.5684741189495627, "learning_rate": 9.237418390996907e-05, "loss": 2.9601, "step": 12580 }, { "epoch": 0.7809919920541313, "grad_norm": 0.5464541314395326, "learning_rate": 9.23722667249191e-05, "loss": 3.0759, "step": 12581 }, { "epoch": 0.7810540691538891, "grad_norm": 0.4634167220640341, "learning_rate": 9.237034931880271e-05, "loss": 3.0867, "step": 12582 }, { "epoch": 0.781116146253647, "grad_norm": 0.5360646855995906, "learning_rate": 9.23684316916299e-05, "loss": 3.0044, "step": 12583 }, { "epoch": 0.7811782233534049, "grad_norm": 0.430256362971058, "learning_rate": 9.236651384341074e-05, "loss": 2.971, "step": 12584 }, { "epoch": 0.7812403004531628, "grad_norm": 0.5086772305231999, "learning_rate": 9.236459577415515e-05, "loss": 2.9537, "step": 12585 }, { "epoch": 0.7813023775529208, "grad_norm": 0.5193448126566442, "learning_rate": 9.23626774838732e-05, "loss": 3.0699, "step": 12586 }, { "epoch": 0.7813644546526787, "grad_norm": 0.541333487329007, "learning_rate": 9.236075897257488e-05, "loss": 2.9557, "step": 12587 }, { "epoch": 0.7814265317524365, "grad_norm": 0.516679754270897, "learning_rate": 9.23588402402702e-05, "loss": 2.9793, "step": 12588 }, { "epoch": 0.7814886088521944, "grad_norm": 0.5796138562416012, "learning_rate": 9.235692128696914e-05, "loss": 3.0835, "step": 12589 }, { "epoch": 0.7815506859519523, "grad_norm": 0.6491776114787027, "learning_rate": 9.235500211268177e-05, "loss": 3.0696, "step": 12590 }, { "epoch": 0.7816127630517102, "grad_norm": 0.6252091534651661, "learning_rate": 9.235308271741805e-05, "loss": 3.0622, "step": 12591 }, { "epoch": 0.7816748401514682, "grad_norm": 0.5361861461244799, "learning_rate": 9.235116310118803e-05, "loss": 2.9893, "step": 12592 }, { "epoch": 0.7817369172512261, "grad_norm": 0.5326964377799572, "learning_rate": 9.234924326400171e-05, "loss": 2.975, "step": 12593 }, { "epoch": 0.7817989943509839, "grad_norm": 0.5088220977309085, "learning_rate": 9.23473232058691e-05, "loss": 3.0491, "step": 12594 }, { "epoch": 0.7818610714507418, "grad_norm": 0.5421981282800262, "learning_rate": 9.234540292680023e-05, "loss": 3.1187, "step": 12595 }, { "epoch": 0.7819231485504997, "grad_norm": 0.6064283826249375, "learning_rate": 9.234348242680512e-05, "loss": 3.0646, "step": 12596 }, { "epoch": 0.7819852256502576, "grad_norm": 0.5375453220155721, "learning_rate": 9.234156170589377e-05, "loss": 3.0118, "step": 12597 }, { "epoch": 0.7820473027500155, "grad_norm": 0.5603489808920277, "learning_rate": 9.233964076407622e-05, "loss": 3.0285, "step": 12598 }, { "epoch": 0.7821093798497735, "grad_norm": 0.563534130415057, "learning_rate": 9.233771960136249e-05, "loss": 3.1327, "step": 12599 }, { "epoch": 0.7821714569495313, "grad_norm": 0.5569036029872424, "learning_rate": 9.233579821776259e-05, "loss": 2.9684, "step": 12600 }, { "epoch": 0.7822335340492892, "grad_norm": 0.547886322061158, "learning_rate": 9.233387661328656e-05, "loss": 3.0175, "step": 12601 }, { "epoch": 0.7822956111490471, "grad_norm": 0.5004220611679594, "learning_rate": 9.233195478794441e-05, "loss": 3.0333, "step": 12602 }, { "epoch": 0.782357688248805, "grad_norm": 0.5016894408640299, "learning_rate": 9.233003274174618e-05, "loss": 3.0838, "step": 12603 }, { "epoch": 0.782419765348563, "grad_norm": 0.4644970923580163, "learning_rate": 9.232811047470188e-05, "loss": 3.0285, "step": 12604 }, { "epoch": 0.7824818424483209, "grad_norm": 0.537069367739408, "learning_rate": 9.232618798682158e-05, "loss": 3.0499, "step": 12605 }, { "epoch": 0.7825439195480787, "grad_norm": 0.45436933272551633, "learning_rate": 9.232426527811528e-05, "loss": 3.0258, "step": 12606 }, { "epoch": 0.7826059966478366, "grad_norm": 0.5184879870719868, "learning_rate": 9.2322342348593e-05, "loss": 3.1491, "step": 12607 }, { "epoch": 0.7826680737475945, "grad_norm": 0.4565802247864761, "learning_rate": 9.232041919826477e-05, "loss": 2.9377, "step": 12608 }, { "epoch": 0.7827301508473524, "grad_norm": 0.4610980772502441, "learning_rate": 9.231849582714068e-05, "loss": 2.9588, "step": 12609 }, { "epoch": 0.7827922279471103, "grad_norm": 0.44936564158535486, "learning_rate": 9.231657223523069e-05, "loss": 3.0193, "step": 12610 }, { "epoch": 0.7828543050468683, "grad_norm": 0.47134011249419655, "learning_rate": 9.231464842254489e-05, "loss": 2.9542, "step": 12611 }, { "epoch": 0.7829163821466261, "grad_norm": 0.43506873190698764, "learning_rate": 9.231272438909329e-05, "loss": 3.0808, "step": 12612 }, { "epoch": 0.782978459246384, "grad_norm": 0.43840097570241127, "learning_rate": 9.231080013488594e-05, "loss": 3.0431, "step": 12613 }, { "epoch": 0.7830405363461419, "grad_norm": 0.4195609915081281, "learning_rate": 9.230887565993285e-05, "loss": 2.9523, "step": 12614 }, { "epoch": 0.7831026134458998, "grad_norm": 0.4642526731419265, "learning_rate": 9.230695096424411e-05, "loss": 2.9614, "step": 12615 }, { "epoch": 0.7831646905456577, "grad_norm": 0.4669709459656651, "learning_rate": 9.230502604782974e-05, "loss": 3.0369, "step": 12616 }, { "epoch": 0.7832267676454157, "grad_norm": 0.47712302466231715, "learning_rate": 9.230310091069976e-05, "loss": 3.0169, "step": 12617 }, { "epoch": 0.7832888447451735, "grad_norm": 0.4290581863142023, "learning_rate": 9.230117555286423e-05, "loss": 3.0131, "step": 12618 }, { "epoch": 0.7833509218449314, "grad_norm": 0.46991854291513024, "learning_rate": 9.229924997433322e-05, "loss": 3.0739, "step": 12619 }, { "epoch": 0.7834129989446893, "grad_norm": 0.4982948882472538, "learning_rate": 9.229732417511674e-05, "loss": 2.9176, "step": 12620 }, { "epoch": 0.7834750760444472, "grad_norm": 0.4753829867674812, "learning_rate": 9.229539815522485e-05, "loss": 3.1003, "step": 12621 }, { "epoch": 0.7835371531442051, "grad_norm": 0.531716954948182, "learning_rate": 9.229347191466761e-05, "loss": 3.0166, "step": 12622 }, { "epoch": 0.783599230243963, "grad_norm": 0.4749475381139025, "learning_rate": 9.229154545345507e-05, "loss": 3.0052, "step": 12623 }, { "epoch": 0.7836613073437209, "grad_norm": 0.5120131832191351, "learning_rate": 9.228961877159726e-05, "loss": 3.0709, "step": 12624 }, { "epoch": 0.7837233844434788, "grad_norm": 0.48038820041225166, "learning_rate": 9.228769186910424e-05, "loss": 3.0025, "step": 12625 }, { "epoch": 0.7837854615432367, "grad_norm": 0.6289848136402818, "learning_rate": 9.228576474598605e-05, "loss": 3.0155, "step": 12626 }, { "epoch": 0.7838475386429946, "grad_norm": 0.4951186772611879, "learning_rate": 9.228383740225278e-05, "loss": 2.9645, "step": 12627 }, { "epoch": 0.7839096157427525, "grad_norm": 0.5241279603022523, "learning_rate": 9.228190983791446e-05, "loss": 3.0865, "step": 12628 }, { "epoch": 0.7839716928425104, "grad_norm": 0.5820167590903986, "learning_rate": 9.227998205298116e-05, "loss": 3.0393, "step": 12629 }, { "epoch": 0.7840337699422683, "grad_norm": 0.5385870841211132, "learning_rate": 9.227805404746293e-05, "loss": 3.1313, "step": 12630 }, { "epoch": 0.7840958470420262, "grad_norm": 0.4918630984125504, "learning_rate": 9.227612582136981e-05, "loss": 3.0078, "step": 12631 }, { "epoch": 0.7841579241417841, "grad_norm": 0.5064402305268607, "learning_rate": 9.227419737471189e-05, "loss": 3.0652, "step": 12632 }, { "epoch": 0.784220001241542, "grad_norm": 0.5816841114355774, "learning_rate": 9.227226870749922e-05, "loss": 3.0136, "step": 12633 }, { "epoch": 0.7842820783412999, "grad_norm": 0.5372911180441892, "learning_rate": 9.227033981974187e-05, "loss": 3.0702, "step": 12634 }, { "epoch": 0.7843441554410578, "grad_norm": 0.5826602072055822, "learning_rate": 9.226841071144988e-05, "loss": 3.0304, "step": 12635 }, { "epoch": 0.7844062325408157, "grad_norm": 0.4939268501665685, "learning_rate": 9.226648138263334e-05, "loss": 2.9964, "step": 12636 }, { "epoch": 0.7844683096405736, "grad_norm": 0.5261282298813004, "learning_rate": 9.22645518333023e-05, "loss": 2.9784, "step": 12637 }, { "epoch": 0.7845303867403315, "grad_norm": 0.5992997951483667, "learning_rate": 9.226262206346684e-05, "loss": 3.0948, "step": 12638 }, { "epoch": 0.7845924638400894, "grad_norm": 0.550227867953513, "learning_rate": 9.226069207313703e-05, "loss": 3.103, "step": 12639 }, { "epoch": 0.7846545409398473, "grad_norm": 0.5983369976939182, "learning_rate": 9.225876186232291e-05, "loss": 2.9954, "step": 12640 }, { "epoch": 0.7847166180396052, "grad_norm": 0.6130152325845637, "learning_rate": 9.225683143103458e-05, "loss": 2.9923, "step": 12641 }, { "epoch": 0.784778695139363, "grad_norm": 0.5467822677464172, "learning_rate": 9.22549007792821e-05, "loss": 3.049, "step": 12642 }, { "epoch": 0.784840772239121, "grad_norm": 0.5375123610294987, "learning_rate": 9.225296990707555e-05, "loss": 3.0315, "step": 12643 }, { "epoch": 0.7849028493388789, "grad_norm": 0.5127393721767546, "learning_rate": 9.225103881442499e-05, "loss": 3.0638, "step": 12644 }, { "epoch": 0.7849649264386368, "grad_norm": 0.5384694306154163, "learning_rate": 9.224910750134052e-05, "loss": 3.0765, "step": 12645 }, { "epoch": 0.7850270035383947, "grad_norm": 0.4694809618533915, "learning_rate": 9.224717596783218e-05, "loss": 3.0618, "step": 12646 }, { "epoch": 0.7850890806381526, "grad_norm": 0.49023619875625657, "learning_rate": 9.224524421391008e-05, "loss": 2.9921, "step": 12647 }, { "epoch": 0.7851511577379104, "grad_norm": 0.47299108388876693, "learning_rate": 9.224331223958429e-05, "loss": 2.9786, "step": 12648 }, { "epoch": 0.7852132348376684, "grad_norm": 0.4517393960262373, "learning_rate": 9.224138004486487e-05, "loss": 3.0299, "step": 12649 }, { "epoch": 0.7852753119374263, "grad_norm": 0.5019198751755019, "learning_rate": 9.223944762976192e-05, "loss": 3.0492, "step": 12650 }, { "epoch": 0.7853373890371842, "grad_norm": 0.5887926854819536, "learning_rate": 9.223751499428553e-05, "loss": 3.0929, "step": 12651 }, { "epoch": 0.7853994661369421, "grad_norm": 0.4942837082043889, "learning_rate": 9.223558213844575e-05, "loss": 3.0765, "step": 12652 }, { "epoch": 0.7854615432367, "grad_norm": 0.47867519287234594, "learning_rate": 9.22336490622527e-05, "loss": 3.0493, "step": 12653 }, { "epoch": 0.7855236203364578, "grad_norm": 0.5872620425331346, "learning_rate": 9.223171576571645e-05, "loss": 3.0315, "step": 12654 }, { "epoch": 0.7855856974362158, "grad_norm": 0.5836284662872148, "learning_rate": 9.222978224884708e-05, "loss": 2.9551, "step": 12655 }, { "epoch": 0.7856477745359737, "grad_norm": 0.5724289436471383, "learning_rate": 9.222784851165469e-05, "loss": 3.0414, "step": 12656 }, { "epoch": 0.7857098516357316, "grad_norm": 0.4691458620006212, "learning_rate": 9.222591455414935e-05, "loss": 3.0221, "step": 12657 }, { "epoch": 0.7857719287354895, "grad_norm": 0.4919950320838147, "learning_rate": 9.222398037634119e-05, "loss": 2.9787, "step": 12658 }, { "epoch": 0.7858340058352474, "grad_norm": 0.4873947488433432, "learning_rate": 9.222204597824024e-05, "loss": 3.0286, "step": 12659 }, { "epoch": 0.7858960829350052, "grad_norm": 0.5205056653864164, "learning_rate": 9.222011135985665e-05, "loss": 3.02, "step": 12660 }, { "epoch": 0.7859581600347632, "grad_norm": 0.4449187595971731, "learning_rate": 9.221817652120048e-05, "loss": 2.9521, "step": 12661 }, { "epoch": 0.7860202371345211, "grad_norm": 0.5667172071647836, "learning_rate": 9.221624146228185e-05, "loss": 3.0558, "step": 12662 }, { "epoch": 0.786082314234279, "grad_norm": 0.5683918055559776, "learning_rate": 9.221430618311082e-05, "loss": 3.0765, "step": 12663 }, { "epoch": 0.7861443913340369, "grad_norm": 0.5391790844441958, "learning_rate": 9.221237068369753e-05, "loss": 3.0056, "step": 12664 }, { "epoch": 0.7862064684337948, "grad_norm": 0.4663079706157895, "learning_rate": 9.221043496405203e-05, "loss": 3.0111, "step": 12665 }, { "epoch": 0.7862685455335526, "grad_norm": 0.5512267401063332, "learning_rate": 9.220849902418445e-05, "loss": 2.9356, "step": 12666 }, { "epoch": 0.7863306226333105, "grad_norm": 0.518814452889713, "learning_rate": 9.22065628641049e-05, "loss": 3.0103, "step": 12667 }, { "epoch": 0.7863926997330685, "grad_norm": 0.4732414437672702, "learning_rate": 9.220462648382344e-05, "loss": 3.0539, "step": 12668 }, { "epoch": 0.7864547768328264, "grad_norm": 0.5400128487541853, "learning_rate": 9.220268988335022e-05, "loss": 3.0556, "step": 12669 }, { "epoch": 0.7865168539325843, "grad_norm": 0.5573564176881753, "learning_rate": 9.220075306269531e-05, "loss": 3.024, "step": 12670 }, { "epoch": 0.7865789310323422, "grad_norm": 0.4887640740302394, "learning_rate": 9.219881602186882e-05, "loss": 3.0085, "step": 12671 }, { "epoch": 0.7866410081321, "grad_norm": 0.5263637903560365, "learning_rate": 9.219687876088087e-05, "loss": 3.0595, "step": 12672 }, { "epoch": 0.786703085231858, "grad_norm": 0.5293049266245011, "learning_rate": 9.219494127974157e-05, "loss": 3.0028, "step": 12673 }, { "epoch": 0.7867651623316159, "grad_norm": 0.5241532911657112, "learning_rate": 9.219300357846102e-05, "loss": 3.1555, "step": 12674 }, { "epoch": 0.7868272394313738, "grad_norm": 0.534400406172164, "learning_rate": 9.219106565704931e-05, "loss": 3.0086, "step": 12675 }, { "epoch": 0.7868893165311317, "grad_norm": 0.4889090050351186, "learning_rate": 9.218912751551659e-05, "loss": 3.0884, "step": 12676 }, { "epoch": 0.7869513936308896, "grad_norm": 0.5769666459675828, "learning_rate": 9.218718915387292e-05, "loss": 3.0874, "step": 12677 }, { "epoch": 0.7870134707306474, "grad_norm": 0.5560630612803034, "learning_rate": 9.218525057212846e-05, "loss": 3.1011, "step": 12678 }, { "epoch": 0.7870755478304053, "grad_norm": 0.5000135300713012, "learning_rate": 9.21833117702933e-05, "loss": 3.0343, "step": 12679 }, { "epoch": 0.7871376249301633, "grad_norm": 0.48872696941778393, "learning_rate": 9.218137274837758e-05, "loss": 2.9179, "step": 12680 }, { "epoch": 0.7871997020299212, "grad_norm": 0.5169553644882103, "learning_rate": 9.217943350639139e-05, "loss": 2.9992, "step": 12681 }, { "epoch": 0.7872617791296791, "grad_norm": 0.5297141031882062, "learning_rate": 9.217749404434484e-05, "loss": 2.9978, "step": 12682 }, { "epoch": 0.787323856229437, "grad_norm": 0.5286538134443932, "learning_rate": 9.217555436224809e-05, "loss": 3.134, "step": 12683 }, { "epoch": 0.7873859333291948, "grad_norm": 0.5036268716561009, "learning_rate": 9.217361446011121e-05, "loss": 2.9523, "step": 12684 }, { "epoch": 0.7874480104289527, "grad_norm": 0.48913362491645423, "learning_rate": 9.217167433794437e-05, "loss": 3.0388, "step": 12685 }, { "epoch": 0.7875100875287107, "grad_norm": 0.5427423183419424, "learning_rate": 9.216973399575765e-05, "loss": 3.0219, "step": 12686 }, { "epoch": 0.7875721646284686, "grad_norm": 0.44825518553926263, "learning_rate": 9.21677934335612e-05, "loss": 2.9841, "step": 12687 }, { "epoch": 0.7876342417282265, "grad_norm": 0.6392801403851475, "learning_rate": 9.216585265136513e-05, "loss": 2.9046, "step": 12688 }, { "epoch": 0.7876963188279844, "grad_norm": 0.5751150752062912, "learning_rate": 9.216391164917957e-05, "loss": 3.0752, "step": 12689 }, { "epoch": 0.7877583959277422, "grad_norm": 0.5257049811312453, "learning_rate": 9.216197042701464e-05, "loss": 2.9825, "step": 12690 }, { "epoch": 0.7878204730275001, "grad_norm": 0.5726431685094582, "learning_rate": 9.216002898488048e-05, "loss": 3.0531, "step": 12691 }, { "epoch": 0.787882550127258, "grad_norm": 0.6220854513341368, "learning_rate": 9.215808732278722e-05, "loss": 3.1076, "step": 12692 }, { "epoch": 0.787944627227016, "grad_norm": 0.562335970592401, "learning_rate": 9.2156145440745e-05, "loss": 3.0557, "step": 12693 }, { "epoch": 0.7880067043267739, "grad_norm": 0.502966810911813, "learning_rate": 9.21542033387639e-05, "loss": 3.0651, "step": 12694 }, { "epoch": 0.7880687814265318, "grad_norm": 0.5621644184660589, "learning_rate": 9.215226101685411e-05, "loss": 3.0387, "step": 12695 }, { "epoch": 0.7881308585262896, "grad_norm": 0.5364055113727608, "learning_rate": 9.215031847502573e-05, "loss": 3.1308, "step": 12696 }, { "epoch": 0.7881929356260475, "grad_norm": 0.5440435406461553, "learning_rate": 9.214837571328892e-05, "loss": 3.039, "step": 12697 }, { "epoch": 0.7882550127258054, "grad_norm": 0.4932563015631463, "learning_rate": 9.21464327316538e-05, "loss": 2.9558, "step": 12698 }, { "epoch": 0.7883170898255634, "grad_norm": 0.50732498242318, "learning_rate": 9.21444895301305e-05, "loss": 3.0069, "step": 12699 }, { "epoch": 0.7883791669253213, "grad_norm": 0.5316973373581023, "learning_rate": 9.214254610872916e-05, "loss": 3.0318, "step": 12700 }, { "epoch": 0.7884412440250792, "grad_norm": 0.5231514618902117, "learning_rate": 9.214060246745995e-05, "loss": 2.9704, "step": 12701 }, { "epoch": 0.788503321124837, "grad_norm": 0.47741034397538024, "learning_rate": 9.213865860633297e-05, "loss": 3.0522, "step": 12702 }, { "epoch": 0.7885653982245949, "grad_norm": 0.5074298297120539, "learning_rate": 9.21367145253584e-05, "loss": 3.0802, "step": 12703 }, { "epoch": 0.7886274753243528, "grad_norm": 0.4478245703327633, "learning_rate": 9.213477022454633e-05, "loss": 3.0572, "step": 12704 }, { "epoch": 0.7886895524241108, "grad_norm": 0.5251282955266947, "learning_rate": 9.213282570390696e-05, "loss": 3.1098, "step": 12705 }, { "epoch": 0.7887516295238687, "grad_norm": 0.5047753992685671, "learning_rate": 9.21308809634504e-05, "loss": 3.0734, "step": 12706 }, { "epoch": 0.7888137066236266, "grad_norm": 0.5473884896591247, "learning_rate": 9.212893600318679e-05, "loss": 2.9863, "step": 12707 }, { "epoch": 0.7888757837233844, "grad_norm": 0.499821183892796, "learning_rate": 9.212699082312631e-05, "loss": 3.1388, "step": 12708 }, { "epoch": 0.7889378608231423, "grad_norm": 0.5379865107954679, "learning_rate": 9.212504542327909e-05, "loss": 3.0796, "step": 12709 }, { "epoch": 0.7889999379229002, "grad_norm": 0.537286763809197, "learning_rate": 9.212309980365529e-05, "loss": 3.0561, "step": 12710 }, { "epoch": 0.7890620150226582, "grad_norm": 0.4689956021621227, "learning_rate": 9.212115396426505e-05, "loss": 3.0186, "step": 12711 }, { "epoch": 0.7891240921224161, "grad_norm": 0.5027017021084363, "learning_rate": 9.211920790511853e-05, "loss": 2.9803, "step": 12712 }, { "epoch": 0.789186169222174, "grad_norm": 0.5155006027440155, "learning_rate": 9.211726162622586e-05, "loss": 3.0388, "step": 12713 }, { "epoch": 0.7892482463219318, "grad_norm": 0.5980825462143011, "learning_rate": 9.211531512759722e-05, "loss": 3.0422, "step": 12714 }, { "epoch": 0.7893103234216897, "grad_norm": 0.5138858596036595, "learning_rate": 9.211336840924275e-05, "loss": 3.0366, "step": 12715 }, { "epoch": 0.7893724005214476, "grad_norm": 0.5058127716866917, "learning_rate": 9.211142147117262e-05, "loss": 3.0905, "step": 12716 }, { "epoch": 0.7894344776212056, "grad_norm": 0.5244625928647267, "learning_rate": 9.210947431339698e-05, "loss": 2.9985, "step": 12717 }, { "epoch": 0.7894965547209635, "grad_norm": 0.4614529233346991, "learning_rate": 9.2107526935926e-05, "loss": 3.0391, "step": 12718 }, { "epoch": 0.7895586318207214, "grad_norm": 0.4917046736989492, "learning_rate": 9.210557933876981e-05, "loss": 3.0481, "step": 12719 }, { "epoch": 0.7896207089204792, "grad_norm": 0.4938200412034888, "learning_rate": 9.210363152193862e-05, "loss": 2.9898, "step": 12720 }, { "epoch": 0.7896827860202371, "grad_norm": 0.5697767798543888, "learning_rate": 9.210168348544255e-05, "loss": 3.0194, "step": 12721 }, { "epoch": 0.789744863119995, "grad_norm": 0.4854483862842704, "learning_rate": 9.209973522929176e-05, "loss": 3.0229, "step": 12722 }, { "epoch": 0.789806940219753, "grad_norm": 0.5390127960731516, "learning_rate": 9.209778675349645e-05, "loss": 3.0782, "step": 12723 }, { "epoch": 0.7898690173195109, "grad_norm": 0.5137177653537329, "learning_rate": 9.209583805806677e-05, "loss": 3.0161, "step": 12724 }, { "epoch": 0.7899310944192688, "grad_norm": 0.5016134871951885, "learning_rate": 9.209388914301287e-05, "loss": 3.0452, "step": 12725 }, { "epoch": 0.7899931715190266, "grad_norm": 0.46668350566842076, "learning_rate": 9.209194000834494e-05, "loss": 2.9711, "step": 12726 }, { "epoch": 0.7900552486187845, "grad_norm": 0.4786829597456058, "learning_rate": 9.208999065407313e-05, "loss": 2.9669, "step": 12727 }, { "epoch": 0.7901173257185424, "grad_norm": 0.4838352734176095, "learning_rate": 9.208804108020763e-05, "loss": 2.9583, "step": 12728 }, { "epoch": 0.7901794028183003, "grad_norm": 0.48210195211439816, "learning_rate": 9.20860912867586e-05, "loss": 3.0353, "step": 12729 }, { "epoch": 0.7902414799180583, "grad_norm": 0.4798679006368869, "learning_rate": 9.208414127373621e-05, "loss": 3.0127, "step": 12730 }, { "epoch": 0.7903035570178162, "grad_norm": 0.4920623256819194, "learning_rate": 9.208219104115066e-05, "loss": 3.0123, "step": 12731 }, { "epoch": 0.790365634117574, "grad_norm": 0.46237035880398236, "learning_rate": 9.208024058901208e-05, "loss": 3.0856, "step": 12732 }, { "epoch": 0.7904277112173319, "grad_norm": 0.49087092498785867, "learning_rate": 9.207828991733067e-05, "loss": 3.0053, "step": 12733 }, { "epoch": 0.7904897883170898, "grad_norm": 0.5506683666804922, "learning_rate": 9.207633902611663e-05, "loss": 3.1176, "step": 12734 }, { "epoch": 0.7905518654168477, "grad_norm": 0.500429445855798, "learning_rate": 9.20743879153801e-05, "loss": 3.1006, "step": 12735 }, { "epoch": 0.7906139425166057, "grad_norm": 0.5130531733303303, "learning_rate": 9.207243658513127e-05, "loss": 2.9453, "step": 12736 }, { "epoch": 0.7906760196163636, "grad_norm": 0.6164526367139155, "learning_rate": 9.207048503538033e-05, "loss": 3.0704, "step": 12737 }, { "epoch": 0.7907380967161214, "grad_norm": 0.4909888461267854, "learning_rate": 9.206853326613745e-05, "loss": 3.1562, "step": 12738 }, { "epoch": 0.7908001738158793, "grad_norm": 0.48692794945259305, "learning_rate": 9.206658127741284e-05, "loss": 2.9937, "step": 12739 }, { "epoch": 0.7908622509156372, "grad_norm": 0.520446659238884, "learning_rate": 9.206462906921663e-05, "loss": 2.9665, "step": 12740 }, { "epoch": 0.7909243280153951, "grad_norm": 0.5274719082552294, "learning_rate": 9.206267664155907e-05, "loss": 3.0825, "step": 12741 }, { "epoch": 0.7909864051151531, "grad_norm": 0.4988884341974294, "learning_rate": 9.20607239944503e-05, "loss": 2.9727, "step": 12742 }, { "epoch": 0.7910484822149109, "grad_norm": 0.5335055600925003, "learning_rate": 9.205877112790051e-05, "loss": 3.0901, "step": 12743 }, { "epoch": 0.7911105593146688, "grad_norm": 0.5297222047312471, "learning_rate": 9.205681804191992e-05, "loss": 2.9899, "step": 12744 }, { "epoch": 0.7911726364144267, "grad_norm": 0.5119390523051257, "learning_rate": 9.20548647365187e-05, "loss": 2.995, "step": 12745 }, { "epoch": 0.7912347135141846, "grad_norm": 0.4824802347937863, "learning_rate": 9.205291121170703e-05, "loss": 3.0171, "step": 12746 }, { "epoch": 0.7912967906139425, "grad_norm": 0.5520003334369206, "learning_rate": 9.205095746749513e-05, "loss": 3.1223, "step": 12747 }, { "epoch": 0.7913588677137005, "grad_norm": 0.5581601734334237, "learning_rate": 9.204900350389317e-05, "loss": 3.019, "step": 12748 }, { "epoch": 0.7914209448134583, "grad_norm": 0.5195210678135342, "learning_rate": 9.204704932091135e-05, "loss": 3.1007, "step": 12749 }, { "epoch": 0.7914830219132162, "grad_norm": 0.5436152170563665, "learning_rate": 9.204509491855986e-05, "loss": 3.0384, "step": 12750 }, { "epoch": 0.7915450990129741, "grad_norm": 0.503832227607689, "learning_rate": 9.204314029684891e-05, "loss": 3.0508, "step": 12751 }, { "epoch": 0.791607176112732, "grad_norm": 0.5380933868391027, "learning_rate": 9.204118545578869e-05, "loss": 3.0152, "step": 12752 }, { "epoch": 0.7916692532124899, "grad_norm": 0.49145668510860285, "learning_rate": 9.20392303953894e-05, "loss": 2.9482, "step": 12753 }, { "epoch": 0.7917313303122478, "grad_norm": 0.5347725822492159, "learning_rate": 9.203727511566122e-05, "loss": 3.0266, "step": 12754 }, { "epoch": 0.7917934074120057, "grad_norm": 0.4754662490412901, "learning_rate": 9.20353196166144e-05, "loss": 3.0002, "step": 12755 }, { "epoch": 0.7918554845117636, "grad_norm": 0.5428868513295911, "learning_rate": 9.203336389825909e-05, "loss": 3.0719, "step": 12756 }, { "epoch": 0.7919175616115215, "grad_norm": 0.5289797800811296, "learning_rate": 9.203140796060554e-05, "loss": 3.0212, "step": 12757 }, { "epoch": 0.7919796387112794, "grad_norm": 0.4650762601584661, "learning_rate": 9.202945180366391e-05, "loss": 3.0109, "step": 12758 }, { "epoch": 0.7920417158110373, "grad_norm": 0.44985613178836886, "learning_rate": 9.202749542744445e-05, "loss": 2.993, "step": 12759 }, { "epoch": 0.7921037929107952, "grad_norm": 0.5037642581994868, "learning_rate": 9.202553883195733e-05, "loss": 2.9888, "step": 12760 }, { "epoch": 0.792165870010553, "grad_norm": 0.46631600738543866, "learning_rate": 9.202358201721277e-05, "loss": 2.9718, "step": 12761 }, { "epoch": 0.792227947110311, "grad_norm": 0.4568330873137077, "learning_rate": 9.202162498322098e-05, "loss": 2.9623, "step": 12762 }, { "epoch": 0.7922900242100689, "grad_norm": 0.4709361121012304, "learning_rate": 9.201966772999218e-05, "loss": 3.0409, "step": 12763 }, { "epoch": 0.7923521013098268, "grad_norm": 0.5073697434120394, "learning_rate": 9.201771025753657e-05, "loss": 3.0844, "step": 12764 }, { "epoch": 0.7924141784095847, "grad_norm": 0.5182859367317405, "learning_rate": 9.201575256586435e-05, "loss": 3.0476, "step": 12765 }, { "epoch": 0.7924762555093426, "grad_norm": 0.47294124176714936, "learning_rate": 9.201379465498576e-05, "loss": 2.9722, "step": 12766 }, { "epoch": 0.7925383326091004, "grad_norm": 0.4036419119695695, "learning_rate": 9.201183652491101e-05, "loss": 3.0267, "step": 12767 }, { "epoch": 0.7926004097088584, "grad_norm": 0.4707141574184037, "learning_rate": 9.200987817565031e-05, "loss": 3.049, "step": 12768 }, { "epoch": 0.7926624868086163, "grad_norm": 0.4574912438049782, "learning_rate": 9.200791960721388e-05, "loss": 3.032, "step": 12769 }, { "epoch": 0.7927245639083742, "grad_norm": 0.4135739124931994, "learning_rate": 9.200596081961193e-05, "loss": 3.0572, "step": 12770 }, { "epoch": 0.7927866410081321, "grad_norm": 0.4685491926499933, "learning_rate": 9.200400181285468e-05, "loss": 3.0019, "step": 12771 }, { "epoch": 0.79284871810789, "grad_norm": 0.5060150613880302, "learning_rate": 9.200204258695235e-05, "loss": 3.0306, "step": 12772 }, { "epoch": 0.7929107952076478, "grad_norm": 0.5011423294028409, "learning_rate": 9.20000831419152e-05, "loss": 3.1061, "step": 12773 }, { "epoch": 0.7929728723074058, "grad_norm": 0.4405674856268832, "learning_rate": 9.199812347775338e-05, "loss": 3.0115, "step": 12774 }, { "epoch": 0.7930349494071637, "grad_norm": 0.5323588915463825, "learning_rate": 9.199616359447718e-05, "loss": 3.0231, "step": 12775 }, { "epoch": 0.7930970265069216, "grad_norm": 0.45496427560824465, "learning_rate": 9.199420349209679e-05, "loss": 3.0465, "step": 12776 }, { "epoch": 0.7931591036066795, "grad_norm": 0.4312987894295522, "learning_rate": 9.199224317062246e-05, "loss": 2.9843, "step": 12777 }, { "epoch": 0.7932211807064374, "grad_norm": 0.48178834281926647, "learning_rate": 9.199028263006438e-05, "loss": 3.0308, "step": 12778 }, { "epoch": 0.7932832578061952, "grad_norm": 0.4846242294236522, "learning_rate": 9.19883218704328e-05, "loss": 2.9784, "step": 12779 }, { "epoch": 0.7933453349059532, "grad_norm": 0.450546177571992, "learning_rate": 9.198636089173797e-05, "loss": 3.0198, "step": 12780 }, { "epoch": 0.7934074120057111, "grad_norm": 0.465016869756829, "learning_rate": 9.19843996939901e-05, "loss": 3.0147, "step": 12781 }, { "epoch": 0.793469489105469, "grad_norm": 0.4851174355919478, "learning_rate": 9.198243827719942e-05, "loss": 2.9679, "step": 12782 }, { "epoch": 0.7935315662052269, "grad_norm": 0.5384102344306834, "learning_rate": 9.198047664137617e-05, "loss": 3.0113, "step": 12783 }, { "epoch": 0.7935936433049848, "grad_norm": 0.462572012269643, "learning_rate": 9.197851478653059e-05, "loss": 2.9825, "step": 12784 }, { "epoch": 0.7936557204047426, "grad_norm": 0.46531778095597043, "learning_rate": 9.197655271267289e-05, "loss": 3.0173, "step": 12785 }, { "epoch": 0.7937177975045006, "grad_norm": 0.48341870506261064, "learning_rate": 9.197459041981334e-05, "loss": 3.0974, "step": 12786 }, { "epoch": 0.7937798746042585, "grad_norm": 0.4765377663619551, "learning_rate": 9.197262790796216e-05, "loss": 2.998, "step": 12787 }, { "epoch": 0.7938419517040164, "grad_norm": 0.518155333916011, "learning_rate": 9.197066517712958e-05, "loss": 2.9571, "step": 12788 }, { "epoch": 0.7939040288037743, "grad_norm": 0.4570013346214339, "learning_rate": 9.196870222732586e-05, "loss": 3.052, "step": 12789 }, { "epoch": 0.7939661059035322, "grad_norm": 0.5336749047243756, "learning_rate": 9.196673905856124e-05, "loss": 3.0479, "step": 12790 }, { "epoch": 0.79402818300329, "grad_norm": 0.4557179549518371, "learning_rate": 9.196477567084594e-05, "loss": 3.0322, "step": 12791 }, { "epoch": 0.794090260103048, "grad_norm": 0.5344710313942018, "learning_rate": 9.196281206419022e-05, "loss": 2.9942, "step": 12792 }, { "epoch": 0.7941523372028059, "grad_norm": 0.5055987301522316, "learning_rate": 9.196084823860434e-05, "loss": 3.0693, "step": 12793 }, { "epoch": 0.7942144143025638, "grad_norm": 0.4667668991360073, "learning_rate": 9.19588841940985e-05, "loss": 3.0517, "step": 12794 }, { "epoch": 0.7942764914023217, "grad_norm": 0.5030081974711497, "learning_rate": 9.1956919930683e-05, "loss": 3.0575, "step": 12795 }, { "epoch": 0.7943385685020796, "grad_norm": 0.4841227182077954, "learning_rate": 9.195495544836805e-05, "loss": 2.9898, "step": 12796 }, { "epoch": 0.7944006456018374, "grad_norm": 0.4551719110467991, "learning_rate": 9.195299074716392e-05, "loss": 2.8907, "step": 12797 }, { "epoch": 0.7944627227015953, "grad_norm": 0.4857663506363769, "learning_rate": 9.195102582708085e-05, "loss": 3.0333, "step": 12798 }, { "epoch": 0.7945247998013533, "grad_norm": 0.5301658001260752, "learning_rate": 9.194906068812908e-05, "loss": 3.0426, "step": 12799 }, { "epoch": 0.7945868769011112, "grad_norm": 0.5121326956251064, "learning_rate": 9.19470953303189e-05, "loss": 3.0461, "step": 12800 }, { "epoch": 0.7946489540008691, "grad_norm": 0.4619063201319948, "learning_rate": 9.194512975366052e-05, "loss": 3.1078, "step": 12801 }, { "epoch": 0.794711031100627, "grad_norm": 0.49083042764316354, "learning_rate": 9.194316395816424e-05, "loss": 3.0687, "step": 12802 }, { "epoch": 0.7947731082003848, "grad_norm": 0.5613154283121604, "learning_rate": 9.194119794384027e-05, "loss": 3.0027, "step": 12803 }, { "epoch": 0.7948351853001427, "grad_norm": 0.5004266468337158, "learning_rate": 9.193923171069889e-05, "loss": 2.9785, "step": 12804 }, { "epoch": 0.7948972623999007, "grad_norm": 0.5272848590887155, "learning_rate": 9.193726525875037e-05, "loss": 2.9642, "step": 12805 }, { "epoch": 0.7949593394996586, "grad_norm": 0.5333711408337377, "learning_rate": 9.193529858800495e-05, "loss": 3.0837, "step": 12806 }, { "epoch": 0.7950214165994165, "grad_norm": 0.4981266630999325, "learning_rate": 9.193333169847289e-05, "loss": 3.0938, "step": 12807 }, { "epoch": 0.7950834936991744, "grad_norm": 0.5700068503938579, "learning_rate": 9.193136459016447e-05, "loss": 3.0888, "step": 12808 }, { "epoch": 0.7951455707989322, "grad_norm": 0.5068283227484485, "learning_rate": 9.192939726308993e-05, "loss": 3.0018, "step": 12809 }, { "epoch": 0.7952076478986901, "grad_norm": 0.5670511717672294, "learning_rate": 9.192742971725957e-05, "loss": 2.9721, "step": 12810 }, { "epoch": 0.7952697249984481, "grad_norm": 0.5714802739901662, "learning_rate": 9.192546195268361e-05, "loss": 3.0357, "step": 12811 }, { "epoch": 0.795331802098206, "grad_norm": 0.5382285093649612, "learning_rate": 9.192349396937234e-05, "loss": 3.0037, "step": 12812 }, { "epoch": 0.7953938791979639, "grad_norm": 0.5557792289467636, "learning_rate": 9.192152576733602e-05, "loss": 2.9275, "step": 12813 }, { "epoch": 0.7954559562977218, "grad_norm": 0.4718378248160396, "learning_rate": 9.191955734658493e-05, "loss": 3.0381, "step": 12814 }, { "epoch": 0.7955180333974796, "grad_norm": 0.6185165051406664, "learning_rate": 9.191758870712933e-05, "loss": 3.1616, "step": 12815 }, { "epoch": 0.7955801104972375, "grad_norm": 0.5264097914855438, "learning_rate": 9.19156198489795e-05, "loss": 3.044, "step": 12816 }, { "epoch": 0.7956421875969955, "grad_norm": 0.4715016820934222, "learning_rate": 9.191365077214571e-05, "loss": 3.0018, "step": 12817 }, { "epoch": 0.7957042646967534, "grad_norm": 0.5490365498054828, "learning_rate": 9.191168147663822e-05, "loss": 3.0155, "step": 12818 }, { "epoch": 0.7957663417965113, "grad_norm": 0.50339884680054, "learning_rate": 9.190971196246731e-05, "loss": 2.9597, "step": 12819 }, { "epoch": 0.7958284188962692, "grad_norm": 0.45782265245032294, "learning_rate": 9.190774222964327e-05, "loss": 3.1364, "step": 12820 }, { "epoch": 0.795890495996027, "grad_norm": 0.5677284592960301, "learning_rate": 9.190577227817635e-05, "loss": 2.9827, "step": 12821 }, { "epoch": 0.7959525730957849, "grad_norm": 0.4965779502198333, "learning_rate": 9.190380210807685e-05, "loss": 3.0455, "step": 12822 }, { "epoch": 0.7960146501955428, "grad_norm": 0.47332886056846335, "learning_rate": 9.190183171935505e-05, "loss": 3.0985, "step": 12823 }, { "epoch": 0.7960767272953008, "grad_norm": 0.5559269127526503, "learning_rate": 9.189986111202122e-05, "loss": 3.0506, "step": 12824 }, { "epoch": 0.7961388043950587, "grad_norm": 0.4994384622413017, "learning_rate": 9.189789028608563e-05, "loss": 3.0757, "step": 12825 }, { "epoch": 0.7962008814948166, "grad_norm": 0.5562160010073527, "learning_rate": 9.189591924155858e-05, "loss": 3.0925, "step": 12826 }, { "epoch": 0.7962629585945744, "grad_norm": 0.5064290789596455, "learning_rate": 9.189394797845036e-05, "loss": 3.0057, "step": 12827 }, { "epoch": 0.7963250356943323, "grad_norm": 0.5076165921779076, "learning_rate": 9.189197649677124e-05, "loss": 3.0144, "step": 12828 }, { "epoch": 0.7963871127940902, "grad_norm": 0.47426555094663214, "learning_rate": 9.189000479653149e-05, "loss": 3.0502, "step": 12829 }, { "epoch": 0.7964491898938482, "grad_norm": 0.5721612516140102, "learning_rate": 9.188803287774143e-05, "loss": 3.0197, "step": 12830 }, { "epoch": 0.7965112669936061, "grad_norm": 0.5096740055695185, "learning_rate": 9.188606074041134e-05, "loss": 3.0687, "step": 12831 }, { "epoch": 0.796573344093364, "grad_norm": 0.5355820736771086, "learning_rate": 9.18840883845515e-05, "loss": 3.0354, "step": 12832 }, { "epoch": 0.7966354211931218, "grad_norm": 0.5787466161146992, "learning_rate": 9.18821158101722e-05, "loss": 3.0055, "step": 12833 }, { "epoch": 0.7966974982928797, "grad_norm": 0.44020105081802835, "learning_rate": 9.188014301728373e-05, "loss": 2.9282, "step": 12834 }, { "epoch": 0.7967595753926376, "grad_norm": 0.5414841754094545, "learning_rate": 9.187817000589641e-05, "loss": 3.0762, "step": 12835 }, { "epoch": 0.7968216524923956, "grad_norm": 0.4946013703735655, "learning_rate": 9.187619677602049e-05, "loss": 2.9923, "step": 12836 }, { "epoch": 0.7968837295921535, "grad_norm": 0.4711553530167418, "learning_rate": 9.187422332766628e-05, "loss": 3.07, "step": 12837 }, { "epoch": 0.7969458066919114, "grad_norm": 0.43189960150200174, "learning_rate": 9.187224966084409e-05, "loss": 2.9327, "step": 12838 }, { "epoch": 0.7970078837916692, "grad_norm": 0.49948319906517025, "learning_rate": 9.187027577556421e-05, "loss": 3.0884, "step": 12839 }, { "epoch": 0.7970699608914271, "grad_norm": 0.48983803339638643, "learning_rate": 9.186830167183693e-05, "loss": 2.9776, "step": 12840 }, { "epoch": 0.797132037991185, "grad_norm": 0.4798029962472753, "learning_rate": 9.186632734967256e-05, "loss": 3.0019, "step": 12841 }, { "epoch": 0.797194115090943, "grad_norm": 0.4962372159392205, "learning_rate": 9.18643528090814e-05, "loss": 3.0771, "step": 12842 }, { "epoch": 0.7972561921907009, "grad_norm": 0.5331761838932642, "learning_rate": 9.186237805007374e-05, "loss": 3.0665, "step": 12843 }, { "epoch": 0.7973182692904588, "grad_norm": 0.48983868743860026, "learning_rate": 9.18604030726599e-05, "loss": 3.0372, "step": 12844 }, { "epoch": 0.7973803463902166, "grad_norm": 0.5350633248170302, "learning_rate": 9.185842787685019e-05, "loss": 3.109, "step": 12845 }, { "epoch": 0.7974424234899745, "grad_norm": 0.5032526257157937, "learning_rate": 9.185645246265488e-05, "loss": 3.008, "step": 12846 }, { "epoch": 0.7975045005897324, "grad_norm": 0.4925616371245003, "learning_rate": 9.18544768300843e-05, "loss": 2.9348, "step": 12847 }, { "epoch": 0.7975665776894904, "grad_norm": 0.48719477512880954, "learning_rate": 9.185250097914875e-05, "loss": 3.0163, "step": 12848 }, { "epoch": 0.7976286547892483, "grad_norm": 0.5378329864625351, "learning_rate": 9.185052490985855e-05, "loss": 3.0276, "step": 12849 }, { "epoch": 0.7976907318890062, "grad_norm": 0.47837924181490665, "learning_rate": 9.184854862222399e-05, "loss": 3.0579, "step": 12850 }, { "epoch": 0.797752808988764, "grad_norm": 0.45449441303516463, "learning_rate": 9.18465721162554e-05, "loss": 3.035, "step": 12851 }, { "epoch": 0.7978148860885219, "grad_norm": 0.5070612412151076, "learning_rate": 9.184459539196309e-05, "loss": 3.117, "step": 12852 }, { "epoch": 0.7978769631882798, "grad_norm": 0.45724218342660955, "learning_rate": 9.184261844935738e-05, "loss": 2.9893, "step": 12853 }, { "epoch": 0.7979390402880377, "grad_norm": 0.46809303021733756, "learning_rate": 9.184064128844853e-05, "loss": 2.9959, "step": 12854 }, { "epoch": 0.7980011173877957, "grad_norm": 0.49302893571974793, "learning_rate": 9.183866390924694e-05, "loss": 2.9908, "step": 12855 }, { "epoch": 0.7980631944875536, "grad_norm": 0.5127729665893986, "learning_rate": 9.183668631176287e-05, "loss": 3.0756, "step": 12856 }, { "epoch": 0.7981252715873114, "grad_norm": 0.43619543805533156, "learning_rate": 9.183470849600664e-05, "loss": 2.8905, "step": 12857 }, { "epoch": 0.7981873486870693, "grad_norm": 0.512675461157721, "learning_rate": 9.18327304619886e-05, "loss": 2.9408, "step": 12858 }, { "epoch": 0.7982494257868272, "grad_norm": 0.5543767837874175, "learning_rate": 9.183075220971903e-05, "loss": 3.0267, "step": 12859 }, { "epoch": 0.7983115028865851, "grad_norm": 0.4779979007058352, "learning_rate": 9.182877373920827e-05, "loss": 2.9303, "step": 12860 }, { "epoch": 0.7983735799863431, "grad_norm": 0.5256381674805732, "learning_rate": 9.182679505046665e-05, "loss": 3.0984, "step": 12861 }, { "epoch": 0.798435657086101, "grad_norm": 0.5750564899891308, "learning_rate": 9.182481614350448e-05, "loss": 3.0052, "step": 12862 }, { "epoch": 0.7984977341858588, "grad_norm": 0.555207478902516, "learning_rate": 9.182283701833212e-05, "loss": 3.1753, "step": 12863 }, { "epoch": 0.7985598112856167, "grad_norm": 0.4945376463562293, "learning_rate": 9.182085767495985e-05, "loss": 3.0227, "step": 12864 }, { "epoch": 0.7986218883853746, "grad_norm": 0.4936914777478378, "learning_rate": 9.1818878113398e-05, "loss": 2.9879, "step": 12865 }, { "epoch": 0.7986839654851325, "grad_norm": 0.5200223394327053, "learning_rate": 9.181689833365692e-05, "loss": 3.0598, "step": 12866 }, { "epoch": 0.7987460425848905, "grad_norm": 0.631928851269438, "learning_rate": 9.181491833574694e-05, "loss": 2.9526, "step": 12867 }, { "epoch": 0.7988081196846484, "grad_norm": 0.49391793521717986, "learning_rate": 9.181293811967836e-05, "loss": 2.922, "step": 12868 }, { "epoch": 0.7988701967844062, "grad_norm": 0.54365230318415, "learning_rate": 9.181095768546154e-05, "loss": 3.0582, "step": 12869 }, { "epoch": 0.7989322738841641, "grad_norm": 0.5610682491837842, "learning_rate": 9.18089770331068e-05, "loss": 3.0879, "step": 12870 }, { "epoch": 0.798994350983922, "grad_norm": 0.49788579392854004, "learning_rate": 9.18069961626245e-05, "loss": 2.9589, "step": 12871 }, { "epoch": 0.7990564280836799, "grad_norm": 0.6011711928653132, "learning_rate": 9.180501507402492e-05, "loss": 3.1028, "step": 12872 }, { "epoch": 0.7991185051834379, "grad_norm": 0.5305439239855784, "learning_rate": 9.180303376731845e-05, "loss": 2.9334, "step": 12873 }, { "epoch": 0.7991805822831958, "grad_norm": 0.5058431733235879, "learning_rate": 9.180105224251539e-05, "loss": 2.9175, "step": 12874 }, { "epoch": 0.7992426593829536, "grad_norm": 0.508369023472004, "learning_rate": 9.179907049962611e-05, "loss": 3.076, "step": 12875 }, { "epoch": 0.7993047364827115, "grad_norm": 0.5225050577804129, "learning_rate": 9.179708853866092e-05, "loss": 3.0206, "step": 12876 }, { "epoch": 0.7993668135824694, "grad_norm": 0.4774394643318445, "learning_rate": 9.179510635963017e-05, "loss": 3.1024, "step": 12877 }, { "epoch": 0.7994288906822273, "grad_norm": 0.45064487476936765, "learning_rate": 9.17931239625442e-05, "loss": 3.0186, "step": 12878 }, { "epoch": 0.7994909677819853, "grad_norm": 0.5117358430526477, "learning_rate": 9.179114134741338e-05, "loss": 3.0004, "step": 12879 }, { "epoch": 0.7995530448817432, "grad_norm": 0.45544055700970965, "learning_rate": 9.1789158514248e-05, "loss": 3.0009, "step": 12880 }, { "epoch": 0.799615121981501, "grad_norm": 0.5073951473347613, "learning_rate": 9.178717546305846e-05, "loss": 2.9972, "step": 12881 }, { "epoch": 0.7996771990812589, "grad_norm": 0.5132615513846484, "learning_rate": 9.178519219385507e-05, "loss": 3.0349, "step": 12882 }, { "epoch": 0.7997392761810168, "grad_norm": 0.4797122551311434, "learning_rate": 9.17832087066482e-05, "loss": 3.0413, "step": 12883 }, { "epoch": 0.7998013532807747, "grad_norm": 0.48756701302270383, "learning_rate": 9.178122500144817e-05, "loss": 3.0385, "step": 12884 }, { "epoch": 0.7998634303805326, "grad_norm": 0.4224847519679363, "learning_rate": 9.177924107826536e-05, "loss": 3.0183, "step": 12885 }, { "epoch": 0.7999255074802906, "grad_norm": 0.47624835619805467, "learning_rate": 9.17772569371101e-05, "loss": 3.0156, "step": 12886 }, { "epoch": 0.7999875845800484, "grad_norm": 0.5037010425603035, "learning_rate": 9.177527257799274e-05, "loss": 2.9184, "step": 12887 }, { "epoch": 0.8000496616798063, "grad_norm": 0.49403669749568463, "learning_rate": 9.177328800092366e-05, "loss": 2.9458, "step": 12888 }, { "epoch": 0.8001117387795642, "grad_norm": 0.5304555843556611, "learning_rate": 9.177130320591319e-05, "loss": 3.0449, "step": 12889 }, { "epoch": 0.8001738158793221, "grad_norm": 0.5223674084380358, "learning_rate": 9.176931819297168e-05, "loss": 3.0635, "step": 12890 }, { "epoch": 0.80023589297908, "grad_norm": 0.5768148939781759, "learning_rate": 9.176733296210951e-05, "loss": 2.9608, "step": 12891 }, { "epoch": 0.800297970078838, "grad_norm": 0.5079381933910453, "learning_rate": 9.176534751333704e-05, "loss": 3.0657, "step": 12892 }, { "epoch": 0.8003600471785958, "grad_norm": 0.5510469702000903, "learning_rate": 9.176336184666458e-05, "loss": 3.0288, "step": 12893 }, { "epoch": 0.8004221242783537, "grad_norm": 0.5365135078080666, "learning_rate": 9.176137596210252e-05, "loss": 3.0069, "step": 12894 }, { "epoch": 0.8004842013781116, "grad_norm": 0.46246635533724645, "learning_rate": 9.175938985966125e-05, "loss": 3.035, "step": 12895 }, { "epoch": 0.8005462784778695, "grad_norm": 0.530051590684775, "learning_rate": 9.17574035393511e-05, "loss": 3.0792, "step": 12896 }, { "epoch": 0.8006083555776274, "grad_norm": 0.5071291801030245, "learning_rate": 9.175541700118244e-05, "loss": 3.0701, "step": 12897 }, { "epoch": 0.8006704326773854, "grad_norm": 0.5451383536701315, "learning_rate": 9.175343024516561e-05, "loss": 3.0183, "step": 12898 }, { "epoch": 0.8007325097771432, "grad_norm": 0.4659424787545749, "learning_rate": 9.175144327131102e-05, "loss": 2.9516, "step": 12899 }, { "epoch": 0.8007945868769011, "grad_norm": 0.5191796339088247, "learning_rate": 9.174945607962901e-05, "loss": 3.0852, "step": 12900 }, { "epoch": 0.800856663976659, "grad_norm": 0.49642655330798585, "learning_rate": 9.174746867012995e-05, "loss": 2.9381, "step": 12901 }, { "epoch": 0.8009187410764169, "grad_norm": 0.5473600688639287, "learning_rate": 9.174548104282422e-05, "loss": 3.132, "step": 12902 }, { "epoch": 0.8009808181761748, "grad_norm": 0.5005267170359456, "learning_rate": 9.174349319772217e-05, "loss": 3.0222, "step": 12903 }, { "epoch": 0.8010428952759328, "grad_norm": 0.49476153298689696, "learning_rate": 9.174150513483419e-05, "loss": 3.0496, "step": 12904 }, { "epoch": 0.8011049723756906, "grad_norm": 0.5208520377297322, "learning_rate": 9.173951685417065e-05, "loss": 3.052, "step": 12905 }, { "epoch": 0.8011670494754485, "grad_norm": 0.5618196187291368, "learning_rate": 9.173752835574191e-05, "loss": 2.9748, "step": 12906 }, { "epoch": 0.8012291265752064, "grad_norm": 0.5351800843660257, "learning_rate": 9.173553963955836e-05, "loss": 2.9683, "step": 12907 }, { "epoch": 0.8012912036749643, "grad_norm": 0.5680176012105912, "learning_rate": 9.173355070563036e-05, "loss": 3.0028, "step": 12908 }, { "epoch": 0.8013532807747222, "grad_norm": 0.5206120084352899, "learning_rate": 9.173156155396829e-05, "loss": 3.0146, "step": 12909 }, { "epoch": 0.8014153578744801, "grad_norm": 0.5307380509561014, "learning_rate": 9.172957218458254e-05, "loss": 2.9508, "step": 12910 }, { "epoch": 0.801477434974238, "grad_norm": 0.5075890563003821, "learning_rate": 9.172758259748349e-05, "loss": 3.0172, "step": 12911 }, { "epoch": 0.8015395120739959, "grad_norm": 0.5791541810599525, "learning_rate": 9.172559279268149e-05, "loss": 3.0474, "step": 12912 }, { "epoch": 0.8016015891737538, "grad_norm": 0.4740343417189373, "learning_rate": 9.172360277018696e-05, "loss": 3.0179, "step": 12913 }, { "epoch": 0.8016636662735117, "grad_norm": 0.4920377806360036, "learning_rate": 9.172161253001026e-05, "loss": 3.051, "step": 12914 }, { "epoch": 0.8017257433732696, "grad_norm": 0.5930726554612971, "learning_rate": 9.171962207216179e-05, "loss": 3.031, "step": 12915 }, { "epoch": 0.8017878204730275, "grad_norm": 0.48075011783901916, "learning_rate": 9.17176313966519e-05, "loss": 3.0594, "step": 12916 }, { "epoch": 0.8018498975727854, "grad_norm": 0.4991711063962396, "learning_rate": 9.171564050349102e-05, "loss": 2.9753, "step": 12917 }, { "epoch": 0.8019119746725433, "grad_norm": 0.4760520734282475, "learning_rate": 9.171364939268952e-05, "loss": 3.0643, "step": 12918 }, { "epoch": 0.8019740517723012, "grad_norm": 0.49887559048056873, "learning_rate": 9.171165806425777e-05, "loss": 2.9739, "step": 12919 }, { "epoch": 0.8020361288720591, "grad_norm": 0.44137961383224145, "learning_rate": 9.170966651820617e-05, "loss": 2.8954, "step": 12920 }, { "epoch": 0.802098205971817, "grad_norm": 0.4473668612720178, "learning_rate": 9.170767475454512e-05, "loss": 3.0882, "step": 12921 }, { "epoch": 0.8021602830715749, "grad_norm": 0.5488373019174104, "learning_rate": 9.170568277328502e-05, "loss": 3.0789, "step": 12922 }, { "epoch": 0.8022223601713327, "grad_norm": 0.5715528612695442, "learning_rate": 9.170369057443624e-05, "loss": 3.0559, "step": 12923 }, { "epoch": 0.8022844372710907, "grad_norm": 0.5031123153793473, "learning_rate": 9.170169815800917e-05, "loss": 3.0994, "step": 12924 }, { "epoch": 0.8023465143708486, "grad_norm": 0.43466927762371216, "learning_rate": 9.169970552401423e-05, "loss": 2.9755, "step": 12925 }, { "epoch": 0.8024085914706065, "grad_norm": 0.5528252732443877, "learning_rate": 9.169771267246182e-05, "loss": 3.0036, "step": 12926 }, { "epoch": 0.8024706685703644, "grad_norm": 0.5200173248100931, "learning_rate": 9.169571960336229e-05, "loss": 3.0599, "step": 12927 }, { "epoch": 0.8025327456701223, "grad_norm": 0.4617485088433775, "learning_rate": 9.169372631672609e-05, "loss": 2.9916, "step": 12928 }, { "epoch": 0.8025948227698801, "grad_norm": 0.6098058840005792, "learning_rate": 9.169173281256357e-05, "loss": 3.1231, "step": 12929 }, { "epoch": 0.8026568998696381, "grad_norm": 0.5542850377523931, "learning_rate": 9.168973909088518e-05, "loss": 3.0431, "step": 12930 }, { "epoch": 0.802718976969396, "grad_norm": 0.4665397350789936, "learning_rate": 9.168774515170129e-05, "loss": 3.0548, "step": 12931 }, { "epoch": 0.8027810540691539, "grad_norm": 0.4747811622707427, "learning_rate": 9.168575099502233e-05, "loss": 2.9691, "step": 12932 }, { "epoch": 0.8028431311689118, "grad_norm": 0.5307995625667308, "learning_rate": 9.168375662085868e-05, "loss": 3.0844, "step": 12933 }, { "epoch": 0.8029052082686697, "grad_norm": 0.5033018642444824, "learning_rate": 9.168176202922075e-05, "loss": 3.055, "step": 12934 }, { "epoch": 0.8029672853684275, "grad_norm": 0.4866882584084499, "learning_rate": 9.167976722011895e-05, "loss": 2.9772, "step": 12935 }, { "epoch": 0.8030293624681855, "grad_norm": 0.4797458327422014, "learning_rate": 9.16777721935637e-05, "loss": 3.015, "step": 12936 }, { "epoch": 0.8030914395679434, "grad_norm": 0.5682651926771285, "learning_rate": 9.167577694956537e-05, "loss": 3.0022, "step": 12937 }, { "epoch": 0.8031535166677013, "grad_norm": 0.47031575008549364, "learning_rate": 9.167378148813442e-05, "loss": 3.0577, "step": 12938 }, { "epoch": 0.8032155937674592, "grad_norm": 0.5582210924955436, "learning_rate": 9.167178580928122e-05, "loss": 2.9995, "step": 12939 }, { "epoch": 0.8032776708672171, "grad_norm": 0.5189514653677778, "learning_rate": 9.166978991301619e-05, "loss": 3.0386, "step": 12940 }, { "epoch": 0.8033397479669749, "grad_norm": 0.45518913061228855, "learning_rate": 9.166779379934975e-05, "loss": 2.9562, "step": 12941 }, { "epoch": 0.8034018250667329, "grad_norm": 0.549899511477331, "learning_rate": 9.166579746829234e-05, "loss": 2.9654, "step": 12942 }, { "epoch": 0.8034639021664908, "grad_norm": 0.5368617093680461, "learning_rate": 9.166380091985433e-05, "loss": 2.9764, "step": 12943 }, { "epoch": 0.8035259792662487, "grad_norm": 0.5201031111768845, "learning_rate": 9.166180415404616e-05, "loss": 2.9057, "step": 12944 }, { "epoch": 0.8035880563660066, "grad_norm": 0.4589952832812805, "learning_rate": 9.165980717087824e-05, "loss": 3.0219, "step": 12945 }, { "epoch": 0.8036501334657645, "grad_norm": 0.5491078977103367, "learning_rate": 9.1657809970361e-05, "loss": 2.9674, "step": 12946 }, { "epoch": 0.8037122105655223, "grad_norm": 0.4935527667124077, "learning_rate": 9.165581255250484e-05, "loss": 3.0204, "step": 12947 }, { "epoch": 0.8037742876652803, "grad_norm": 0.5611978665120987, "learning_rate": 9.165381491732021e-05, "loss": 3.0123, "step": 12948 }, { "epoch": 0.8038363647650382, "grad_norm": 0.4967525343251962, "learning_rate": 9.165181706481752e-05, "loss": 3.0511, "step": 12949 }, { "epoch": 0.8038984418647961, "grad_norm": 0.5426395433678689, "learning_rate": 9.164981899500718e-05, "loss": 3.0042, "step": 12950 }, { "epoch": 0.803960518964554, "grad_norm": 0.5062362563069053, "learning_rate": 9.164782070789962e-05, "loss": 2.9952, "step": 12951 }, { "epoch": 0.8040225960643119, "grad_norm": 0.5134467440218211, "learning_rate": 9.164582220350527e-05, "loss": 2.9947, "step": 12952 }, { "epoch": 0.8040846731640697, "grad_norm": 0.5045548043138455, "learning_rate": 9.164382348183455e-05, "loss": 2.9847, "step": 12953 }, { "epoch": 0.8041467502638276, "grad_norm": 0.5045793163455718, "learning_rate": 9.164182454289789e-05, "loss": 3.0393, "step": 12954 }, { "epoch": 0.8042088273635856, "grad_norm": 0.5054484579974664, "learning_rate": 9.163982538670574e-05, "loss": 2.9878, "step": 12955 }, { "epoch": 0.8042709044633435, "grad_norm": 0.541581639316506, "learning_rate": 9.16378260132685e-05, "loss": 2.9906, "step": 12956 }, { "epoch": 0.8043329815631014, "grad_norm": 0.502121507478191, "learning_rate": 9.163582642259663e-05, "loss": 3.0234, "step": 12957 }, { "epoch": 0.8043950586628593, "grad_norm": 0.5208420816322578, "learning_rate": 9.163382661470053e-05, "loss": 2.9426, "step": 12958 }, { "epoch": 0.8044571357626171, "grad_norm": 0.5367757697544367, "learning_rate": 9.163182658959064e-05, "loss": 2.9689, "step": 12959 }, { "epoch": 0.804519212862375, "grad_norm": 0.4745154739234435, "learning_rate": 9.162982634727743e-05, "loss": 2.9721, "step": 12960 }, { "epoch": 0.804581289962133, "grad_norm": 0.5737250404912755, "learning_rate": 9.162782588777131e-05, "loss": 3.0095, "step": 12961 }, { "epoch": 0.8046433670618909, "grad_norm": 0.5505947581894206, "learning_rate": 9.16258252110827e-05, "loss": 3.0395, "step": 12962 }, { "epoch": 0.8047054441616488, "grad_norm": 0.513409188407687, "learning_rate": 9.162382431722205e-05, "loss": 3.0987, "step": 12963 }, { "epoch": 0.8047675212614067, "grad_norm": 0.4812818485719718, "learning_rate": 9.16218232061998e-05, "loss": 3.0145, "step": 12964 }, { "epoch": 0.8048295983611645, "grad_norm": 0.50136674406168, "learning_rate": 9.161982187802643e-05, "loss": 3.0176, "step": 12965 }, { "epoch": 0.8048916754609224, "grad_norm": 0.4865126661825765, "learning_rate": 9.161782033271231e-05, "loss": 3.0438, "step": 12966 }, { "epoch": 0.8049537525606804, "grad_norm": 0.4539626206342468, "learning_rate": 9.161581857026793e-05, "loss": 3.0264, "step": 12967 }, { "epoch": 0.8050158296604383, "grad_norm": 0.4932472083374774, "learning_rate": 9.161381659070373e-05, "loss": 3.0693, "step": 12968 }, { "epoch": 0.8050779067601962, "grad_norm": 0.46910005849863506, "learning_rate": 9.161181439403015e-05, "loss": 2.9972, "step": 12969 }, { "epoch": 0.8051399838599541, "grad_norm": 0.48728326662490684, "learning_rate": 9.160981198025762e-05, "loss": 2.9963, "step": 12970 }, { "epoch": 0.8052020609597119, "grad_norm": 0.4879151306839419, "learning_rate": 9.16078093493966e-05, "loss": 2.9846, "step": 12971 }, { "epoch": 0.8052641380594698, "grad_norm": 0.5233985829475908, "learning_rate": 9.160580650145754e-05, "loss": 3.0308, "step": 12972 }, { "epoch": 0.8053262151592278, "grad_norm": 0.5194589378110873, "learning_rate": 9.160380343645088e-05, "loss": 2.9625, "step": 12973 }, { "epoch": 0.8053882922589857, "grad_norm": 0.5041139751149123, "learning_rate": 9.160180015438709e-05, "loss": 3.0148, "step": 12974 }, { "epoch": 0.8054503693587436, "grad_norm": 0.5240500552497415, "learning_rate": 9.15997966552766e-05, "loss": 2.974, "step": 12975 }, { "epoch": 0.8055124464585015, "grad_norm": 0.5295771458542204, "learning_rate": 9.15977929391299e-05, "loss": 3.0565, "step": 12976 }, { "epoch": 0.8055745235582593, "grad_norm": 0.5653806938594816, "learning_rate": 9.159578900595738e-05, "loss": 3.0258, "step": 12977 }, { "epoch": 0.8056366006580172, "grad_norm": 0.5054360757965479, "learning_rate": 9.159378485576954e-05, "loss": 2.9569, "step": 12978 }, { "epoch": 0.8056986777577752, "grad_norm": 0.5146077264829146, "learning_rate": 9.159178048857684e-05, "loss": 3.0041, "step": 12979 }, { "epoch": 0.8057607548575331, "grad_norm": 0.5842021676357975, "learning_rate": 9.158977590438971e-05, "loss": 3.0072, "step": 12980 }, { "epoch": 0.805822831957291, "grad_norm": 0.5030680938193586, "learning_rate": 9.158777110321863e-05, "loss": 3.0765, "step": 12981 }, { "epoch": 0.8058849090570489, "grad_norm": 0.5386213072789422, "learning_rate": 9.158576608507405e-05, "loss": 2.9339, "step": 12982 }, { "epoch": 0.8059469861568067, "grad_norm": 0.5160403168493407, "learning_rate": 9.158376084996645e-05, "loss": 2.9992, "step": 12983 }, { "epoch": 0.8060090632565646, "grad_norm": 0.5234218708950354, "learning_rate": 9.158175539790626e-05, "loss": 3.0199, "step": 12984 }, { "epoch": 0.8060711403563225, "grad_norm": 0.5257382288842649, "learning_rate": 9.157974972890396e-05, "loss": 3.0982, "step": 12985 }, { "epoch": 0.8061332174560805, "grad_norm": 0.506064826160016, "learning_rate": 9.157774384297e-05, "loss": 3.0089, "step": 12986 }, { "epoch": 0.8061952945558384, "grad_norm": 0.5030892424582618, "learning_rate": 9.157573774011486e-05, "loss": 3.1376, "step": 12987 }, { "epoch": 0.8062573716555963, "grad_norm": 0.4673612048410539, "learning_rate": 9.157373142034902e-05, "loss": 2.9618, "step": 12988 }, { "epoch": 0.8063194487553541, "grad_norm": 0.47680151526943193, "learning_rate": 9.157172488368291e-05, "loss": 2.9835, "step": 12989 }, { "epoch": 0.806381525855112, "grad_norm": 0.47718325037834547, "learning_rate": 9.156971813012702e-05, "loss": 3.0269, "step": 12990 }, { "epoch": 0.8064436029548699, "grad_norm": 0.5663488424739761, "learning_rate": 9.156771115969184e-05, "loss": 3.0819, "step": 12991 }, { "epoch": 0.8065056800546279, "grad_norm": 0.6049873899690478, "learning_rate": 9.156570397238781e-05, "loss": 2.9637, "step": 12992 }, { "epoch": 0.8065677571543858, "grad_norm": 0.575774776883016, "learning_rate": 9.156369656822541e-05, "loss": 3.0391, "step": 12993 }, { "epoch": 0.8066298342541437, "grad_norm": 0.540601606359011, "learning_rate": 9.156168894721511e-05, "loss": 3.0211, "step": 12994 }, { "epoch": 0.8066919113539015, "grad_norm": 0.511034967301077, "learning_rate": 9.15596811093674e-05, "loss": 3.0407, "step": 12995 }, { "epoch": 0.8067539884536594, "grad_norm": 0.5299312548318329, "learning_rate": 9.155767305469274e-05, "loss": 3.0046, "step": 12996 }, { "epoch": 0.8068160655534173, "grad_norm": 0.5396357266630425, "learning_rate": 9.15556647832016e-05, "loss": 2.915, "step": 12997 }, { "epoch": 0.8068781426531753, "grad_norm": 0.4468678533876272, "learning_rate": 9.155365629490449e-05, "loss": 3.0125, "step": 12998 }, { "epoch": 0.8069402197529332, "grad_norm": 0.5231905112498306, "learning_rate": 9.155164758981186e-05, "loss": 3.0509, "step": 12999 }, { "epoch": 0.8070022968526911, "grad_norm": 0.4686316499713947, "learning_rate": 9.15496386679342e-05, "loss": 3.0019, "step": 13000 }, { "epoch": 0.8070643739524489, "grad_norm": 0.4774247639237349, "learning_rate": 9.154762952928197e-05, "loss": 3.0368, "step": 13001 }, { "epoch": 0.8071264510522068, "grad_norm": 0.5348043468157456, "learning_rate": 9.154562017386569e-05, "loss": 2.9848, "step": 13002 }, { "epoch": 0.8071885281519647, "grad_norm": 0.49322377966973896, "learning_rate": 9.154361060169582e-05, "loss": 3.0985, "step": 13003 }, { "epoch": 0.8072506052517227, "grad_norm": 0.5206373386363482, "learning_rate": 9.154160081278285e-05, "loss": 2.9878, "step": 13004 }, { "epoch": 0.8073126823514806, "grad_norm": 0.51077019521677, "learning_rate": 9.153959080713725e-05, "loss": 3.0059, "step": 13005 }, { "epoch": 0.8073747594512385, "grad_norm": 0.4922126884675735, "learning_rate": 9.153758058476954e-05, "loss": 3.0163, "step": 13006 }, { "epoch": 0.8074368365509963, "grad_norm": 0.4967033215598908, "learning_rate": 9.153557014569019e-05, "loss": 3.0297, "step": 13007 }, { "epoch": 0.8074989136507542, "grad_norm": 0.4939540013054017, "learning_rate": 9.153355948990968e-05, "loss": 2.989, "step": 13008 }, { "epoch": 0.8075609907505121, "grad_norm": 0.4760067466460073, "learning_rate": 9.15315486174385e-05, "loss": 3.0195, "step": 13009 }, { "epoch": 0.80762306785027, "grad_norm": 0.4577743651094795, "learning_rate": 9.152953752828715e-05, "loss": 3.0312, "step": 13010 }, { "epoch": 0.807685144950028, "grad_norm": 0.5299735723060486, "learning_rate": 9.152752622246612e-05, "loss": 2.999, "step": 13011 }, { "epoch": 0.8077472220497859, "grad_norm": 0.4745098056664364, "learning_rate": 9.152551469998591e-05, "loss": 3.055, "step": 13012 }, { "epoch": 0.8078092991495437, "grad_norm": 0.5258326603098742, "learning_rate": 9.152350296085702e-05, "loss": 3.1349, "step": 13013 }, { "epoch": 0.8078713762493016, "grad_norm": 0.4783536831612001, "learning_rate": 9.15214910050899e-05, "loss": 3.025, "step": 13014 }, { "epoch": 0.8079334533490595, "grad_norm": 0.4932052141903976, "learning_rate": 9.151947883269512e-05, "loss": 3.0556, "step": 13015 }, { "epoch": 0.8079955304488174, "grad_norm": 0.471628901114912, "learning_rate": 9.151746644368311e-05, "loss": 2.9246, "step": 13016 }, { "epoch": 0.8080576075485754, "grad_norm": 0.5633682596460674, "learning_rate": 9.151545383806441e-05, "loss": 3.0658, "step": 13017 }, { "epoch": 0.8081196846483333, "grad_norm": 0.4982423137094898, "learning_rate": 9.151344101584952e-05, "loss": 2.9873, "step": 13018 }, { "epoch": 0.8081817617480911, "grad_norm": 0.5644905579506281, "learning_rate": 9.151142797704893e-05, "loss": 3.0623, "step": 13019 }, { "epoch": 0.808243838847849, "grad_norm": 0.5906125799013554, "learning_rate": 9.150941472167313e-05, "loss": 2.9973, "step": 13020 }, { "epoch": 0.8083059159476069, "grad_norm": 0.519784507116065, "learning_rate": 9.150740124973265e-05, "loss": 3.0875, "step": 13021 }, { "epoch": 0.8083679930473648, "grad_norm": 0.6515474313141563, "learning_rate": 9.150538756123797e-05, "loss": 3.0061, "step": 13022 }, { "epoch": 0.8084300701471228, "grad_norm": 0.5221340689954861, "learning_rate": 9.15033736561996e-05, "loss": 3.0235, "step": 13023 }, { "epoch": 0.8084921472468807, "grad_norm": 0.6057686831922466, "learning_rate": 9.150135953462808e-05, "loss": 2.9995, "step": 13024 }, { "epoch": 0.8085542243466385, "grad_norm": 0.5827509123959921, "learning_rate": 9.149934519653387e-05, "loss": 2.9536, "step": 13025 }, { "epoch": 0.8086163014463964, "grad_norm": 0.6150345250650098, "learning_rate": 9.149733064192751e-05, "loss": 3.0214, "step": 13026 }, { "epoch": 0.8086783785461543, "grad_norm": 0.5168690556803625, "learning_rate": 9.14953158708195e-05, "loss": 2.9801, "step": 13027 }, { "epoch": 0.8087404556459122, "grad_norm": 0.5669204482562183, "learning_rate": 9.149330088322037e-05, "loss": 2.9892, "step": 13028 }, { "epoch": 0.8088025327456702, "grad_norm": 0.6169410406626886, "learning_rate": 9.149128567914059e-05, "loss": 2.9711, "step": 13029 }, { "epoch": 0.8088646098454281, "grad_norm": 0.5328579578417167, "learning_rate": 9.148927025859071e-05, "loss": 2.9947, "step": 13030 }, { "epoch": 0.8089266869451859, "grad_norm": 0.563545341855779, "learning_rate": 9.148725462158124e-05, "loss": 3.092, "step": 13031 }, { "epoch": 0.8089887640449438, "grad_norm": 0.5203099259321446, "learning_rate": 9.14852387681227e-05, "loss": 2.9783, "step": 13032 }, { "epoch": 0.8090508411447017, "grad_norm": 0.5851093543376589, "learning_rate": 9.148322269822557e-05, "loss": 3.0971, "step": 13033 }, { "epoch": 0.8091129182444596, "grad_norm": 0.5262227612239521, "learning_rate": 9.148120641190042e-05, "loss": 3.049, "step": 13034 }, { "epoch": 0.8091749953442176, "grad_norm": 0.47187752059869625, "learning_rate": 9.147918990915775e-05, "loss": 3.019, "step": 13035 }, { "epoch": 0.8092370724439755, "grad_norm": 0.5417322583710764, "learning_rate": 9.147717319000806e-05, "loss": 2.9361, "step": 13036 }, { "epoch": 0.8092991495437333, "grad_norm": 0.4462176042439599, "learning_rate": 9.14751562544619e-05, "loss": 3.0104, "step": 13037 }, { "epoch": 0.8093612266434912, "grad_norm": 0.5711600133441389, "learning_rate": 9.147313910252977e-05, "loss": 3.0188, "step": 13038 }, { "epoch": 0.8094233037432491, "grad_norm": 0.547849821744115, "learning_rate": 9.147112173422222e-05, "loss": 3.0714, "step": 13039 }, { "epoch": 0.809485380843007, "grad_norm": 0.46320893815295544, "learning_rate": 9.146910414954974e-05, "loss": 3.001, "step": 13040 }, { "epoch": 0.809547457942765, "grad_norm": 0.4901216992989707, "learning_rate": 9.14670863485229e-05, "loss": 3.0022, "step": 13041 }, { "epoch": 0.8096095350425229, "grad_norm": 0.5313154348577975, "learning_rate": 9.14650683311522e-05, "loss": 2.96, "step": 13042 }, { "epoch": 0.8096716121422807, "grad_norm": 0.5003549388873133, "learning_rate": 9.146305009744815e-05, "loss": 3.0981, "step": 13043 }, { "epoch": 0.8097336892420386, "grad_norm": 0.5464809633029896, "learning_rate": 9.146103164742132e-05, "loss": 2.9554, "step": 13044 }, { "epoch": 0.8097957663417965, "grad_norm": 0.5796971209773887, "learning_rate": 9.145901298108224e-05, "loss": 2.9355, "step": 13045 }, { "epoch": 0.8098578434415544, "grad_norm": 0.5649706042500006, "learning_rate": 9.14569940984414e-05, "loss": 2.9055, "step": 13046 }, { "epoch": 0.8099199205413123, "grad_norm": 0.5207908676637384, "learning_rate": 9.145497499950937e-05, "loss": 2.9842, "step": 13047 }, { "epoch": 0.8099819976410703, "grad_norm": 0.5202460283772158, "learning_rate": 9.145295568429666e-05, "loss": 3.0154, "step": 13048 }, { "epoch": 0.8100440747408281, "grad_norm": 0.5504339218707717, "learning_rate": 9.145093615281381e-05, "loss": 2.9626, "step": 13049 }, { "epoch": 0.810106151840586, "grad_norm": 0.5751290922792462, "learning_rate": 9.144891640507138e-05, "loss": 2.9396, "step": 13050 }, { "epoch": 0.8101682289403439, "grad_norm": 0.4168655238868923, "learning_rate": 9.144689644107989e-05, "loss": 3.0155, "step": 13051 }, { "epoch": 0.8102303060401018, "grad_norm": 0.5601728567218183, "learning_rate": 9.144487626084988e-05, "loss": 3.0574, "step": 13052 }, { "epoch": 0.8102923831398597, "grad_norm": 0.5275306652972692, "learning_rate": 9.144285586439189e-05, "loss": 3.044, "step": 13053 }, { "epoch": 0.8103544602396177, "grad_norm": 0.5278622374241747, "learning_rate": 9.144083525171646e-05, "loss": 3.0406, "step": 13054 }, { "epoch": 0.8104165373393755, "grad_norm": 0.5537159430060624, "learning_rate": 9.143881442283413e-05, "loss": 3.0031, "step": 13055 }, { "epoch": 0.8104786144391334, "grad_norm": 0.47300445719266127, "learning_rate": 9.143679337775543e-05, "loss": 3.0069, "step": 13056 }, { "epoch": 0.8105406915388913, "grad_norm": 0.5204759197949462, "learning_rate": 9.143477211649093e-05, "loss": 2.9942, "step": 13057 }, { "epoch": 0.8106027686386492, "grad_norm": 0.5154281298175503, "learning_rate": 9.143275063905119e-05, "loss": 2.9452, "step": 13058 }, { "epoch": 0.8106648457384071, "grad_norm": 0.5081647091654035, "learning_rate": 9.143072894544671e-05, "loss": 3.0644, "step": 13059 }, { "epoch": 0.810726922838165, "grad_norm": 0.535555474833438, "learning_rate": 9.142870703568806e-05, "loss": 3.0111, "step": 13060 }, { "epoch": 0.8107889999379229, "grad_norm": 0.5950278535301903, "learning_rate": 9.142668490978579e-05, "loss": 3.0568, "step": 13061 }, { "epoch": 0.8108510770376808, "grad_norm": 0.5027193981628166, "learning_rate": 9.142466256775045e-05, "loss": 2.9902, "step": 13062 }, { "epoch": 0.8109131541374387, "grad_norm": 0.5452112502837679, "learning_rate": 9.142264000959257e-05, "loss": 3.0134, "step": 13063 }, { "epoch": 0.8109752312371966, "grad_norm": 0.532063842536938, "learning_rate": 9.142061723532275e-05, "loss": 3.0027, "step": 13064 }, { "epoch": 0.8110373083369545, "grad_norm": 0.491447861794231, "learning_rate": 9.141859424495149e-05, "loss": 3.0489, "step": 13065 }, { "epoch": 0.8110993854367125, "grad_norm": 0.47873097459125863, "learning_rate": 9.141657103848938e-05, "loss": 2.9954, "step": 13066 }, { "epoch": 0.8111614625364703, "grad_norm": 0.47472507338147885, "learning_rate": 9.141454761594698e-05, "loss": 3.0314, "step": 13067 }, { "epoch": 0.8112235396362282, "grad_norm": 0.5449875549330377, "learning_rate": 9.141252397733481e-05, "loss": 2.9723, "step": 13068 }, { "epoch": 0.8112856167359861, "grad_norm": 0.4819324342661536, "learning_rate": 9.141050012266345e-05, "loss": 2.9935, "step": 13069 }, { "epoch": 0.811347693835744, "grad_norm": 0.5074233985273059, "learning_rate": 9.140847605194347e-05, "loss": 3.0357, "step": 13070 }, { "epoch": 0.8114097709355019, "grad_norm": 0.48355231099847096, "learning_rate": 9.140645176518541e-05, "loss": 3.0557, "step": 13071 }, { "epoch": 0.8114718480352598, "grad_norm": 0.5588440767664925, "learning_rate": 9.140442726239985e-05, "loss": 2.9994, "step": 13072 }, { "epoch": 0.8115339251350177, "grad_norm": 0.5096415810917999, "learning_rate": 9.140240254359734e-05, "loss": 2.9538, "step": 13073 }, { "epoch": 0.8115960022347756, "grad_norm": 0.49816393505026907, "learning_rate": 9.140037760878843e-05, "loss": 2.9461, "step": 13074 }, { "epoch": 0.8116580793345335, "grad_norm": 0.5279159549757514, "learning_rate": 9.139835245798372e-05, "loss": 2.9191, "step": 13075 }, { "epoch": 0.8117201564342914, "grad_norm": 0.4959548996178852, "learning_rate": 9.139632709119375e-05, "loss": 3.0827, "step": 13076 }, { "epoch": 0.8117822335340493, "grad_norm": 0.4740206359604455, "learning_rate": 9.139430150842908e-05, "loss": 3.0015, "step": 13077 }, { "epoch": 0.8118443106338072, "grad_norm": 0.5255983360412974, "learning_rate": 9.139227570970029e-05, "loss": 2.9797, "step": 13078 }, { "epoch": 0.811906387733565, "grad_norm": 0.5132668642672864, "learning_rate": 9.139024969501796e-05, "loss": 2.9583, "step": 13079 }, { "epoch": 0.811968464833323, "grad_norm": 0.47839420880202155, "learning_rate": 9.138822346439265e-05, "loss": 2.9576, "step": 13080 }, { "epoch": 0.8120305419330809, "grad_norm": 0.5020239219759953, "learning_rate": 9.138619701783493e-05, "loss": 3.0493, "step": 13081 }, { "epoch": 0.8120926190328388, "grad_norm": 0.5062182416491225, "learning_rate": 9.138417035535536e-05, "loss": 3.0232, "step": 13082 }, { "epoch": 0.8121546961325967, "grad_norm": 0.4710568408836179, "learning_rate": 9.138214347696455e-05, "loss": 2.973, "step": 13083 }, { "epoch": 0.8122167732323546, "grad_norm": 0.5799950434209881, "learning_rate": 9.138011638267303e-05, "loss": 3.012, "step": 13084 }, { "epoch": 0.8122788503321124, "grad_norm": 0.6403121416901703, "learning_rate": 9.13780890724914e-05, "loss": 3.0121, "step": 13085 }, { "epoch": 0.8123409274318704, "grad_norm": 0.5590107568343934, "learning_rate": 9.137606154643025e-05, "loss": 3.0105, "step": 13086 }, { "epoch": 0.8124030045316283, "grad_norm": 0.5421815008881429, "learning_rate": 9.137403380450013e-05, "loss": 2.9706, "step": 13087 }, { "epoch": 0.8124650816313862, "grad_norm": 0.5649406413317937, "learning_rate": 9.137200584671164e-05, "loss": 2.9228, "step": 13088 }, { "epoch": 0.8125271587311441, "grad_norm": 0.5193377830125399, "learning_rate": 9.136997767307534e-05, "loss": 2.9977, "step": 13089 }, { "epoch": 0.812589235830902, "grad_norm": 0.5290561424184329, "learning_rate": 9.136794928360183e-05, "loss": 3.0077, "step": 13090 }, { "epoch": 0.8126513129306598, "grad_norm": 0.5846092172938787, "learning_rate": 9.136592067830168e-05, "loss": 2.9823, "step": 13091 }, { "epoch": 0.8127133900304178, "grad_norm": 0.6012332560216228, "learning_rate": 9.136389185718549e-05, "loss": 3.0043, "step": 13092 }, { "epoch": 0.8127754671301757, "grad_norm": 0.513903315516323, "learning_rate": 9.136186282026384e-05, "loss": 3.0591, "step": 13093 }, { "epoch": 0.8128375442299336, "grad_norm": 0.525254554570826, "learning_rate": 9.13598335675473e-05, "loss": 2.9797, "step": 13094 }, { "epoch": 0.8128996213296915, "grad_norm": 0.5558325004380756, "learning_rate": 9.135780409904646e-05, "loss": 3.0169, "step": 13095 }, { "epoch": 0.8129616984294494, "grad_norm": 0.5898269499350421, "learning_rate": 9.135577441477193e-05, "loss": 3.0438, "step": 13096 }, { "epoch": 0.8130237755292072, "grad_norm": 0.5423448669545182, "learning_rate": 9.135374451473429e-05, "loss": 3.012, "step": 13097 }, { "epoch": 0.8130858526289652, "grad_norm": 0.52256597015758, "learning_rate": 9.135171439894411e-05, "loss": 2.9458, "step": 13098 }, { "epoch": 0.8131479297287231, "grad_norm": 0.5162571009223768, "learning_rate": 9.1349684067412e-05, "loss": 2.9936, "step": 13099 }, { "epoch": 0.813210006828481, "grad_norm": 0.5150692574680695, "learning_rate": 9.134765352014855e-05, "loss": 2.9642, "step": 13100 }, { "epoch": 0.8132720839282389, "grad_norm": 0.4755618580826252, "learning_rate": 9.134562275716436e-05, "loss": 3.0677, "step": 13101 }, { "epoch": 0.8133341610279968, "grad_norm": 0.5076906204681881, "learning_rate": 9.134359177847003e-05, "loss": 2.9719, "step": 13102 }, { "epoch": 0.8133962381277546, "grad_norm": 0.546058344750642, "learning_rate": 9.134156058407613e-05, "loss": 3.008, "step": 13103 }, { "epoch": 0.8134583152275126, "grad_norm": 0.44270971148407295, "learning_rate": 9.133952917399328e-05, "loss": 2.9674, "step": 13104 }, { "epoch": 0.8135203923272705, "grad_norm": 0.5339349608032191, "learning_rate": 9.133749754823206e-05, "loss": 2.9476, "step": 13105 }, { "epoch": 0.8135824694270284, "grad_norm": 0.5667626147757241, "learning_rate": 9.13354657068031e-05, "loss": 3.022, "step": 13106 }, { "epoch": 0.8136445465267863, "grad_norm": 0.5144185488355546, "learning_rate": 9.133343364971695e-05, "loss": 2.9484, "step": 13107 }, { "epoch": 0.8137066236265442, "grad_norm": 0.5785769165458299, "learning_rate": 9.133140137698426e-05, "loss": 3.0638, "step": 13108 }, { "epoch": 0.813768700726302, "grad_norm": 0.5183942298698281, "learning_rate": 9.132936888861562e-05, "loss": 2.9679, "step": 13109 }, { "epoch": 0.81383077782606, "grad_norm": 0.5487507667188567, "learning_rate": 9.132733618462162e-05, "loss": 3.0397, "step": 13110 }, { "epoch": 0.8138928549258179, "grad_norm": 0.6072516494138075, "learning_rate": 9.132530326501288e-05, "loss": 3.0879, "step": 13111 }, { "epoch": 0.8139549320255758, "grad_norm": 0.49914074618185367, "learning_rate": 9.13232701298e-05, "loss": 2.977, "step": 13112 }, { "epoch": 0.8140170091253337, "grad_norm": 0.5225351441350444, "learning_rate": 9.13212367789936e-05, "loss": 3.0144, "step": 13113 }, { "epoch": 0.8140790862250915, "grad_norm": 0.5950767350603168, "learning_rate": 9.131920321260425e-05, "loss": 3.0905, "step": 13114 }, { "epoch": 0.8141411633248494, "grad_norm": 0.5484535595316772, "learning_rate": 9.131716943064262e-05, "loss": 3.01, "step": 13115 }, { "epoch": 0.8142032404246073, "grad_norm": 0.47041091712983574, "learning_rate": 9.131513543311926e-05, "loss": 2.9455, "step": 13116 }, { "epoch": 0.8142653175243653, "grad_norm": 0.528178564436403, "learning_rate": 9.131310122004482e-05, "loss": 2.8535, "step": 13117 }, { "epoch": 0.8143273946241232, "grad_norm": 0.492342001265093, "learning_rate": 9.131106679142988e-05, "loss": 3.0136, "step": 13118 }, { "epoch": 0.8143894717238811, "grad_norm": 0.5997934462730805, "learning_rate": 9.13090321472851e-05, "loss": 3.0825, "step": 13119 }, { "epoch": 0.8144515488236389, "grad_norm": 0.5524197818811938, "learning_rate": 9.130699728762108e-05, "loss": 2.9611, "step": 13120 }, { "epoch": 0.8145136259233968, "grad_norm": 0.5082648390082802, "learning_rate": 9.130496221244841e-05, "loss": 2.9294, "step": 13121 }, { "epoch": 0.8145757030231547, "grad_norm": 0.5729725868215225, "learning_rate": 9.130292692177774e-05, "loss": 3.1002, "step": 13122 }, { "epoch": 0.8146377801229127, "grad_norm": 0.5777598078186996, "learning_rate": 9.130089141561966e-05, "loss": 3.0315, "step": 13123 }, { "epoch": 0.8146998572226706, "grad_norm": 0.5422991459033113, "learning_rate": 9.129885569398479e-05, "loss": 3.0197, "step": 13124 }, { "epoch": 0.8147619343224285, "grad_norm": 0.5372168416103692, "learning_rate": 9.129681975688378e-05, "loss": 2.9528, "step": 13125 }, { "epoch": 0.8148240114221863, "grad_norm": 0.5409154971455905, "learning_rate": 9.129478360432724e-05, "loss": 3.0804, "step": 13126 }, { "epoch": 0.8148860885219442, "grad_norm": 0.5026841832902523, "learning_rate": 9.129274723632579e-05, "loss": 3.0351, "step": 13127 }, { "epoch": 0.8149481656217021, "grad_norm": 0.5769575807144958, "learning_rate": 9.129071065289004e-05, "loss": 3.0143, "step": 13128 }, { "epoch": 0.81501024272146, "grad_norm": 0.47881626883137524, "learning_rate": 9.128867385403064e-05, "loss": 2.979, "step": 13129 }, { "epoch": 0.815072319821218, "grad_norm": 0.5212044125113996, "learning_rate": 9.12866368397582e-05, "loss": 3.0256, "step": 13130 }, { "epoch": 0.8151343969209759, "grad_norm": 0.5133701497144418, "learning_rate": 9.128459961008336e-05, "loss": 3.0076, "step": 13131 }, { "epoch": 0.8151964740207337, "grad_norm": 0.5357292337829836, "learning_rate": 9.128256216501674e-05, "loss": 3.0756, "step": 13132 }, { "epoch": 0.8152585511204916, "grad_norm": 0.5449337428178144, "learning_rate": 9.128052450456896e-05, "loss": 3.0908, "step": 13133 }, { "epoch": 0.8153206282202495, "grad_norm": 0.4883703837102092, "learning_rate": 9.127848662875069e-05, "loss": 2.9869, "step": 13134 }, { "epoch": 0.8153827053200075, "grad_norm": 0.5260461454547033, "learning_rate": 9.12764485375725e-05, "loss": 3.0366, "step": 13135 }, { "epoch": 0.8154447824197654, "grad_norm": 0.5430797216507889, "learning_rate": 9.127441023104508e-05, "loss": 2.9789, "step": 13136 }, { "epoch": 0.8155068595195233, "grad_norm": 0.4829125630904063, "learning_rate": 9.127237170917904e-05, "loss": 3.0594, "step": 13137 }, { "epoch": 0.8155689366192811, "grad_norm": 0.5254973122430915, "learning_rate": 9.127033297198501e-05, "loss": 3.0531, "step": 13138 }, { "epoch": 0.815631013719039, "grad_norm": 0.433658028039478, "learning_rate": 9.126829401947364e-05, "loss": 3.0353, "step": 13139 }, { "epoch": 0.8156930908187969, "grad_norm": 0.5244824242388096, "learning_rate": 9.126625485165557e-05, "loss": 3.0376, "step": 13140 }, { "epoch": 0.8157551679185548, "grad_norm": 0.43554116146975924, "learning_rate": 9.12642154685414e-05, "loss": 3.0254, "step": 13141 }, { "epoch": 0.8158172450183128, "grad_norm": 0.4694549504900219, "learning_rate": 9.126217587014184e-05, "loss": 3.0356, "step": 13142 }, { "epoch": 0.8158793221180707, "grad_norm": 0.525175937055441, "learning_rate": 9.126013605646746e-05, "loss": 3.0159, "step": 13143 }, { "epoch": 0.8159413992178285, "grad_norm": 0.5170941951521791, "learning_rate": 9.125809602752895e-05, "loss": 2.983, "step": 13144 }, { "epoch": 0.8160034763175864, "grad_norm": 0.5048975398204884, "learning_rate": 9.125605578333693e-05, "loss": 3.0173, "step": 13145 }, { "epoch": 0.8160655534173443, "grad_norm": 0.4960372171705757, "learning_rate": 9.125401532390205e-05, "loss": 2.9852, "step": 13146 }, { "epoch": 0.8161276305171022, "grad_norm": 0.4898389003823884, "learning_rate": 9.125197464923497e-05, "loss": 2.9755, "step": 13147 }, { "epoch": 0.8161897076168602, "grad_norm": 0.4666399352756565, "learning_rate": 9.124993375934631e-05, "loss": 2.9699, "step": 13148 }, { "epoch": 0.8162517847166181, "grad_norm": 0.5265414598828864, "learning_rate": 9.124789265424675e-05, "loss": 3.0166, "step": 13149 }, { "epoch": 0.8163138618163759, "grad_norm": 0.49959820756518153, "learning_rate": 9.124585133394691e-05, "loss": 3.003, "step": 13150 }, { "epoch": 0.8163759389161338, "grad_norm": 0.49243006708350473, "learning_rate": 9.124380979845745e-05, "loss": 3.0718, "step": 13151 }, { "epoch": 0.8164380160158917, "grad_norm": 0.46520192107573133, "learning_rate": 9.124176804778903e-05, "loss": 3.0477, "step": 13152 }, { "epoch": 0.8165000931156496, "grad_norm": 0.4802643100636007, "learning_rate": 9.123972608195228e-05, "loss": 2.9956, "step": 13153 }, { "epoch": 0.8165621702154076, "grad_norm": 0.47592085495494474, "learning_rate": 9.123768390095787e-05, "loss": 3.052, "step": 13154 }, { "epoch": 0.8166242473151655, "grad_norm": 0.5104879764127319, "learning_rate": 9.123564150481646e-05, "loss": 3.0626, "step": 13155 }, { "epoch": 0.8166863244149233, "grad_norm": 0.5364763725976047, "learning_rate": 9.123359889353868e-05, "loss": 2.9922, "step": 13156 }, { "epoch": 0.8167484015146812, "grad_norm": 0.5207443701741474, "learning_rate": 9.123155606713523e-05, "loss": 2.9825, "step": 13157 }, { "epoch": 0.8168104786144391, "grad_norm": 0.5193874762513584, "learning_rate": 9.122951302561673e-05, "loss": 3.0562, "step": 13158 }, { "epoch": 0.816872555714197, "grad_norm": 0.4700616287854336, "learning_rate": 9.122746976899386e-05, "loss": 2.898, "step": 13159 }, { "epoch": 0.816934632813955, "grad_norm": 0.4837633393294251, "learning_rate": 9.122542629727726e-05, "loss": 2.9236, "step": 13160 }, { "epoch": 0.8169967099137129, "grad_norm": 0.4583455936641627, "learning_rate": 9.122338261047762e-05, "loss": 2.9867, "step": 13161 }, { "epoch": 0.8170587870134707, "grad_norm": 0.45237556503758136, "learning_rate": 9.122133870860557e-05, "loss": 3.048, "step": 13162 }, { "epoch": 0.8171208641132286, "grad_norm": 0.46403285230470037, "learning_rate": 9.121929459167178e-05, "loss": 3.0314, "step": 13163 }, { "epoch": 0.8171829412129865, "grad_norm": 0.43187386265562155, "learning_rate": 9.121725025968695e-05, "loss": 2.979, "step": 13164 }, { "epoch": 0.8172450183127444, "grad_norm": 0.5433859285932165, "learning_rate": 9.121520571266171e-05, "loss": 2.9819, "step": 13165 }, { "epoch": 0.8173070954125023, "grad_norm": 0.4857539882231141, "learning_rate": 9.121316095060672e-05, "loss": 3.0123, "step": 13166 }, { "epoch": 0.8173691725122603, "grad_norm": 0.49810847481152326, "learning_rate": 9.121111597353269e-05, "loss": 2.947, "step": 13167 }, { "epoch": 0.8174312496120181, "grad_norm": 0.5131065826270336, "learning_rate": 9.120907078145024e-05, "loss": 3.0199, "step": 13168 }, { "epoch": 0.817493326711776, "grad_norm": 0.4989443154512838, "learning_rate": 9.12070253743701e-05, "loss": 3.0212, "step": 13169 }, { "epoch": 0.8175554038115339, "grad_norm": 0.5560352714640558, "learning_rate": 9.120497975230286e-05, "loss": 3.0488, "step": 13170 }, { "epoch": 0.8176174809112918, "grad_norm": 0.48988813354337535, "learning_rate": 9.120293391525927e-05, "loss": 3.0711, "step": 13171 }, { "epoch": 0.8176795580110497, "grad_norm": 0.5411109427205519, "learning_rate": 9.120088786324996e-05, "loss": 3.0056, "step": 13172 }, { "epoch": 0.8177416351108077, "grad_norm": 0.5101355310412582, "learning_rate": 9.11988415962856e-05, "loss": 2.9696, "step": 13173 }, { "epoch": 0.8178037122105655, "grad_norm": 0.5019096622872388, "learning_rate": 9.119679511437691e-05, "loss": 2.9682, "step": 13174 }, { "epoch": 0.8178657893103234, "grad_norm": 0.5281673640126883, "learning_rate": 9.119474841753452e-05, "loss": 2.9078, "step": 13175 }, { "epoch": 0.8179278664100813, "grad_norm": 0.4179049469923151, "learning_rate": 9.119270150576914e-05, "loss": 2.9622, "step": 13176 }, { "epoch": 0.8179899435098392, "grad_norm": 0.49449074445207014, "learning_rate": 9.119065437909143e-05, "loss": 3.0699, "step": 13177 }, { "epoch": 0.8180520206095971, "grad_norm": 0.4959714693768013, "learning_rate": 9.118860703751208e-05, "loss": 3.0811, "step": 13178 }, { "epoch": 0.8181140977093551, "grad_norm": 0.44219050322142966, "learning_rate": 9.118655948104175e-05, "loss": 2.994, "step": 13179 }, { "epoch": 0.8181761748091129, "grad_norm": 0.4432316398626314, "learning_rate": 9.118451170969117e-05, "loss": 3.0113, "step": 13180 }, { "epoch": 0.8182382519088708, "grad_norm": 0.5079315339474179, "learning_rate": 9.118246372347098e-05, "loss": 3.0082, "step": 13181 }, { "epoch": 0.8183003290086287, "grad_norm": 0.4898349304865424, "learning_rate": 9.118041552239188e-05, "loss": 3.0393, "step": 13182 }, { "epoch": 0.8183624061083866, "grad_norm": 0.49432683157851853, "learning_rate": 9.117836710646456e-05, "loss": 2.9903, "step": 13183 }, { "epoch": 0.8184244832081445, "grad_norm": 0.5296737061451483, "learning_rate": 9.117631847569969e-05, "loss": 3.1318, "step": 13184 }, { "epoch": 0.8184865603079025, "grad_norm": 0.5006099794875248, "learning_rate": 9.117426963010797e-05, "loss": 2.9387, "step": 13185 }, { "epoch": 0.8185486374076603, "grad_norm": 0.5125494467909718, "learning_rate": 9.117222056970012e-05, "loss": 3.0198, "step": 13186 }, { "epoch": 0.8186107145074182, "grad_norm": 0.509456701688784, "learning_rate": 9.117017129448677e-05, "loss": 2.8659, "step": 13187 }, { "epoch": 0.8186727916071761, "grad_norm": 0.4858125151802668, "learning_rate": 9.116812180447865e-05, "loss": 3.0779, "step": 13188 }, { "epoch": 0.818734868706934, "grad_norm": 0.619383683013872, "learning_rate": 9.116607209968644e-05, "loss": 3.0113, "step": 13189 }, { "epoch": 0.8187969458066919, "grad_norm": 0.5657018293184937, "learning_rate": 9.116402218012084e-05, "loss": 2.9357, "step": 13190 }, { "epoch": 0.8188590229064499, "grad_norm": 0.5607836286484256, "learning_rate": 9.116197204579255e-05, "loss": 3.0805, "step": 13191 }, { "epoch": 0.8189211000062077, "grad_norm": 0.6162650542519287, "learning_rate": 9.115992169671225e-05, "loss": 2.9678, "step": 13192 }, { "epoch": 0.8189831771059656, "grad_norm": 0.5596581568179854, "learning_rate": 9.115787113289067e-05, "loss": 2.9717, "step": 13193 }, { "epoch": 0.8190452542057235, "grad_norm": 0.5490265619873684, "learning_rate": 9.115582035433847e-05, "loss": 3.0749, "step": 13194 }, { "epoch": 0.8191073313054814, "grad_norm": 0.6157881651235749, "learning_rate": 9.115376936106635e-05, "loss": 3.0398, "step": 13195 }, { "epoch": 0.8191694084052393, "grad_norm": 0.5166748802496121, "learning_rate": 9.115171815308504e-05, "loss": 3.1011, "step": 13196 }, { "epoch": 0.8192314855049972, "grad_norm": 0.5649735318794994, "learning_rate": 9.114966673040522e-05, "loss": 3.0059, "step": 13197 }, { "epoch": 0.819293562604755, "grad_norm": 0.47823160267135173, "learning_rate": 9.11476150930376e-05, "loss": 2.9076, "step": 13198 }, { "epoch": 0.819355639704513, "grad_norm": 0.533997722645879, "learning_rate": 9.114556324099289e-05, "loss": 3.0778, "step": 13199 }, { "epoch": 0.8194177168042709, "grad_norm": 0.5369356186453818, "learning_rate": 9.11435111742818e-05, "loss": 3.0728, "step": 13200 }, { "epoch": 0.8194797939040288, "grad_norm": 0.5116550900995933, "learning_rate": 9.1141458892915e-05, "loss": 2.9559, "step": 13201 }, { "epoch": 0.8195418710037867, "grad_norm": 0.4681738968995124, "learning_rate": 9.113940639690324e-05, "loss": 2.8746, "step": 13202 }, { "epoch": 0.8196039481035446, "grad_norm": 0.5209495287663783, "learning_rate": 9.11373536862572e-05, "loss": 2.9587, "step": 13203 }, { "epoch": 0.8196660252033025, "grad_norm": 0.45028520129422156, "learning_rate": 9.11353007609876e-05, "loss": 3.0426, "step": 13204 }, { "epoch": 0.8197281023030604, "grad_norm": 0.4968628516001646, "learning_rate": 9.113324762110514e-05, "loss": 2.8964, "step": 13205 }, { "epoch": 0.8197901794028183, "grad_norm": 0.5254302736658677, "learning_rate": 9.113119426662055e-05, "loss": 2.9147, "step": 13206 }, { "epoch": 0.8198522565025762, "grad_norm": 0.5262862445826926, "learning_rate": 9.112914069754452e-05, "loss": 3.0788, "step": 13207 }, { "epoch": 0.8199143336023341, "grad_norm": 0.5557501379296891, "learning_rate": 9.11270869138878e-05, "loss": 3.0871, "step": 13208 }, { "epoch": 0.819976410702092, "grad_norm": 0.5276643696737187, "learning_rate": 9.112503291566107e-05, "loss": 2.917, "step": 13209 }, { "epoch": 0.8200384878018498, "grad_norm": 0.5455880690286037, "learning_rate": 9.112297870287509e-05, "loss": 3.0112, "step": 13210 }, { "epoch": 0.8201005649016078, "grad_norm": 0.5041197391087003, "learning_rate": 9.112092427554053e-05, "loss": 3.0002, "step": 13211 }, { "epoch": 0.8201626420013657, "grad_norm": 0.5508564025253365, "learning_rate": 9.111886963366811e-05, "loss": 3.0114, "step": 13212 }, { "epoch": 0.8202247191011236, "grad_norm": 0.5750751466988802, "learning_rate": 9.111681477726858e-05, "loss": 2.9483, "step": 13213 }, { "epoch": 0.8202867962008815, "grad_norm": 0.5096168447108177, "learning_rate": 9.111475970635263e-05, "loss": 3.0297, "step": 13214 }, { "epoch": 0.8203488733006394, "grad_norm": 0.5568249848893251, "learning_rate": 9.1112704420931e-05, "loss": 2.9982, "step": 13215 }, { "epoch": 0.8204109504003972, "grad_norm": 0.497893814802165, "learning_rate": 9.111064892101443e-05, "loss": 3.0717, "step": 13216 }, { "epoch": 0.8204730275001552, "grad_norm": 0.5044626520470403, "learning_rate": 9.110859320661363e-05, "loss": 3.0157, "step": 13217 }, { "epoch": 0.8205351045999131, "grad_norm": 0.5319521415254284, "learning_rate": 9.110653727773928e-05, "loss": 3.0743, "step": 13218 }, { "epoch": 0.820597181699671, "grad_norm": 0.5206937761891225, "learning_rate": 9.110448113440218e-05, "loss": 2.9656, "step": 13219 }, { "epoch": 0.8206592587994289, "grad_norm": 0.5129198436447574, "learning_rate": 9.110242477661301e-05, "loss": 3.011, "step": 13220 }, { "epoch": 0.8207213358991868, "grad_norm": 0.47023639253258626, "learning_rate": 9.110036820438253e-05, "loss": 2.9331, "step": 13221 }, { "epoch": 0.8207834129989446, "grad_norm": 0.483434194147881, "learning_rate": 9.109831141772143e-05, "loss": 3.0579, "step": 13222 }, { "epoch": 0.8208454900987026, "grad_norm": 0.5228226309898886, "learning_rate": 9.109625441664048e-05, "loss": 2.9811, "step": 13223 }, { "epoch": 0.8209075671984605, "grad_norm": 0.5704171594190827, "learning_rate": 9.109419720115037e-05, "loss": 2.9593, "step": 13224 }, { "epoch": 0.8209696442982184, "grad_norm": 0.538939365200145, "learning_rate": 9.109213977126188e-05, "loss": 3.016, "step": 13225 }, { "epoch": 0.8210317213979763, "grad_norm": 0.477279608205223, "learning_rate": 9.109008212698571e-05, "loss": 3.0493, "step": 13226 }, { "epoch": 0.8210937984977342, "grad_norm": 0.4956005528398822, "learning_rate": 9.10880242683326e-05, "loss": 2.9908, "step": 13227 }, { "epoch": 0.821155875597492, "grad_norm": 0.4953563797103366, "learning_rate": 9.10859661953133e-05, "loss": 3.0038, "step": 13228 }, { "epoch": 0.82121795269725, "grad_norm": 0.4857263020248197, "learning_rate": 9.108390790793855e-05, "loss": 2.9837, "step": 13229 }, { "epoch": 0.8212800297970079, "grad_norm": 0.5182288343981216, "learning_rate": 9.108184940621907e-05, "loss": 3.0001, "step": 13230 }, { "epoch": 0.8213421068967658, "grad_norm": 0.4770599020766392, "learning_rate": 9.107979069016562e-05, "loss": 3.0759, "step": 13231 }, { "epoch": 0.8214041839965237, "grad_norm": 0.5891456514960299, "learning_rate": 9.107773175978891e-05, "loss": 2.9484, "step": 13232 }, { "epoch": 0.8214662610962816, "grad_norm": 0.5499809890409186, "learning_rate": 9.10756726150997e-05, "loss": 2.9847, "step": 13233 }, { "epoch": 0.8215283381960394, "grad_norm": 0.5065220502156447, "learning_rate": 9.107361325610875e-05, "loss": 3.0556, "step": 13234 }, { "epoch": 0.8215904152957973, "grad_norm": 0.5697593096298486, "learning_rate": 9.107155368282678e-05, "loss": 3.0503, "step": 13235 }, { "epoch": 0.8216524923955553, "grad_norm": 0.5618245519919156, "learning_rate": 9.106949389526456e-05, "loss": 2.9941, "step": 13236 }, { "epoch": 0.8217145694953132, "grad_norm": 0.5547155722378317, "learning_rate": 9.10674338934328e-05, "loss": 3.0111, "step": 13237 }, { "epoch": 0.8217766465950711, "grad_norm": 0.4983733958095798, "learning_rate": 9.106537367734228e-05, "loss": 3.0754, "step": 13238 }, { "epoch": 0.821838723694829, "grad_norm": 0.5210848931928199, "learning_rate": 9.106331324700373e-05, "loss": 3.0023, "step": 13239 }, { "epoch": 0.8219008007945868, "grad_norm": 0.560676586282607, "learning_rate": 9.106125260242791e-05, "loss": 2.9961, "step": 13240 }, { "epoch": 0.8219628778943447, "grad_norm": 0.6148787181948886, "learning_rate": 9.105919174362556e-05, "loss": 2.9413, "step": 13241 }, { "epoch": 0.8220249549941027, "grad_norm": 0.5472833471267295, "learning_rate": 9.105713067060745e-05, "loss": 3.0704, "step": 13242 }, { "epoch": 0.8220870320938606, "grad_norm": 0.6127946952400355, "learning_rate": 9.105506938338432e-05, "loss": 3.0922, "step": 13243 }, { "epoch": 0.8221491091936185, "grad_norm": 0.5821777005888688, "learning_rate": 9.105300788196693e-05, "loss": 3.0029, "step": 13244 }, { "epoch": 0.8222111862933764, "grad_norm": 0.5604544905716863, "learning_rate": 9.105094616636603e-05, "loss": 3.1307, "step": 13245 }, { "epoch": 0.8222732633931342, "grad_norm": 0.629214288370339, "learning_rate": 9.104888423659237e-05, "loss": 3.1029, "step": 13246 }, { "epoch": 0.8223353404928921, "grad_norm": 0.6478155209461222, "learning_rate": 9.104682209265671e-05, "loss": 2.983, "step": 13247 }, { "epoch": 0.8223974175926501, "grad_norm": 0.5305902367247558, "learning_rate": 9.104475973456983e-05, "loss": 3.05, "step": 13248 }, { "epoch": 0.822459494692408, "grad_norm": 0.5650248552442292, "learning_rate": 9.104269716234248e-05, "loss": 2.985, "step": 13249 }, { "epoch": 0.8225215717921659, "grad_norm": 0.5467144321686521, "learning_rate": 9.10406343759854e-05, "loss": 3.0239, "step": 13250 }, { "epoch": 0.8225836488919238, "grad_norm": 0.6145615546899014, "learning_rate": 9.103857137550937e-05, "loss": 3.0441, "step": 13251 }, { "epoch": 0.8226457259916816, "grad_norm": 0.5641769736835553, "learning_rate": 9.103650816092515e-05, "loss": 3.0374, "step": 13252 }, { "epoch": 0.8227078030914395, "grad_norm": 0.49675109445982163, "learning_rate": 9.103444473224351e-05, "loss": 2.9181, "step": 13253 }, { "epoch": 0.8227698801911975, "grad_norm": 0.5368297612022942, "learning_rate": 9.103238108947521e-05, "loss": 2.9764, "step": 13254 }, { "epoch": 0.8228319572909554, "grad_norm": 0.4916421502946064, "learning_rate": 9.103031723263101e-05, "loss": 2.8717, "step": 13255 }, { "epoch": 0.8228940343907133, "grad_norm": 0.6421121919753267, "learning_rate": 9.10282531617217e-05, "loss": 3.0098, "step": 13256 }, { "epoch": 0.8229561114904712, "grad_norm": 0.6108921190270842, "learning_rate": 9.102618887675802e-05, "loss": 3.0509, "step": 13257 }, { "epoch": 0.823018188590229, "grad_norm": 0.5198486907146316, "learning_rate": 9.102412437775076e-05, "loss": 3.0366, "step": 13258 }, { "epoch": 0.8230802656899869, "grad_norm": 0.6028735748565236, "learning_rate": 9.102205966471068e-05, "loss": 2.96, "step": 13259 }, { "epoch": 0.8231423427897449, "grad_norm": 0.5226965542293812, "learning_rate": 9.101999473764855e-05, "loss": 2.972, "step": 13260 }, { "epoch": 0.8232044198895028, "grad_norm": 0.49772611274382167, "learning_rate": 9.101792959657516e-05, "loss": 3.0421, "step": 13261 }, { "epoch": 0.8232664969892607, "grad_norm": 0.537637347039017, "learning_rate": 9.101586424150128e-05, "loss": 2.9571, "step": 13262 }, { "epoch": 0.8233285740890186, "grad_norm": 0.5676955708693634, "learning_rate": 9.101379867243766e-05, "loss": 3.0156, "step": 13263 }, { "epoch": 0.8233906511887764, "grad_norm": 0.5108279562828418, "learning_rate": 9.101173288939511e-05, "loss": 3.0903, "step": 13264 }, { "epoch": 0.8234527282885343, "grad_norm": 0.5419221575059593, "learning_rate": 9.10096668923844e-05, "loss": 3.0002, "step": 13265 }, { "epoch": 0.8235148053882922, "grad_norm": 0.5771700214410027, "learning_rate": 9.100760068141627e-05, "loss": 3.0688, "step": 13266 }, { "epoch": 0.8235768824880502, "grad_norm": 0.5640202746872947, "learning_rate": 9.100553425650156e-05, "loss": 3.0063, "step": 13267 }, { "epoch": 0.8236389595878081, "grad_norm": 0.5790476267304934, "learning_rate": 9.100346761765102e-05, "loss": 2.9722, "step": 13268 }, { "epoch": 0.823701036687566, "grad_norm": 0.5193427181195771, "learning_rate": 9.100140076487543e-05, "loss": 3.0179, "step": 13269 }, { "epoch": 0.8237631137873238, "grad_norm": 0.48901068562259065, "learning_rate": 9.099933369818558e-05, "loss": 3.0871, "step": 13270 }, { "epoch": 0.8238251908870817, "grad_norm": 0.4913626374024087, "learning_rate": 9.099726641759225e-05, "loss": 2.917, "step": 13271 }, { "epoch": 0.8238872679868396, "grad_norm": 0.5365596385472936, "learning_rate": 9.099519892310624e-05, "loss": 2.9315, "step": 13272 }, { "epoch": 0.8239493450865976, "grad_norm": 0.5033946138249913, "learning_rate": 9.099313121473831e-05, "loss": 3.0519, "step": 13273 }, { "epoch": 0.8240114221863555, "grad_norm": 0.5120947817564311, "learning_rate": 9.099106329249929e-05, "loss": 3.067, "step": 13274 }, { "epoch": 0.8240734992861134, "grad_norm": 0.4538477034243921, "learning_rate": 9.098899515639992e-05, "loss": 3.0291, "step": 13275 }, { "epoch": 0.8241355763858712, "grad_norm": 0.4827334212691786, "learning_rate": 9.098692680645101e-05, "loss": 2.9926, "step": 13276 }, { "epoch": 0.8241976534856291, "grad_norm": 0.5224152732216356, "learning_rate": 9.098485824266336e-05, "loss": 3.0457, "step": 13277 }, { "epoch": 0.824259730585387, "grad_norm": 0.49328484897309577, "learning_rate": 9.098278946504774e-05, "loss": 2.9829, "step": 13278 }, { "epoch": 0.824321807685145, "grad_norm": 0.5033118712457898, "learning_rate": 9.098072047361499e-05, "loss": 2.9305, "step": 13279 }, { "epoch": 0.8243838847849029, "grad_norm": 0.45510191302688485, "learning_rate": 9.097865126837585e-05, "loss": 3.0576, "step": 13280 }, { "epoch": 0.8244459618846608, "grad_norm": 0.43877431364698566, "learning_rate": 9.097658184934115e-05, "loss": 2.9338, "step": 13281 }, { "epoch": 0.8245080389844186, "grad_norm": 0.503059148329182, "learning_rate": 9.097451221652167e-05, "loss": 2.9869, "step": 13282 }, { "epoch": 0.8245701160841765, "grad_norm": 0.5080909259158557, "learning_rate": 9.09724423699282e-05, "loss": 2.9846, "step": 13283 }, { "epoch": 0.8246321931839344, "grad_norm": 0.4335243753831184, "learning_rate": 9.097037230957158e-05, "loss": 2.9023, "step": 13284 }, { "epoch": 0.8246942702836924, "grad_norm": 0.4753107453614201, "learning_rate": 9.096830203546255e-05, "loss": 2.9949, "step": 13285 }, { "epoch": 0.8247563473834503, "grad_norm": 0.5053416957689626, "learning_rate": 9.096623154761196e-05, "loss": 2.9227, "step": 13286 }, { "epoch": 0.8248184244832082, "grad_norm": 0.433490809687485, "learning_rate": 9.096416084603059e-05, "loss": 3.018, "step": 13287 }, { "epoch": 0.824880501582966, "grad_norm": 0.5197725237708632, "learning_rate": 9.096208993072926e-05, "loss": 3.0603, "step": 13288 }, { "epoch": 0.8249425786827239, "grad_norm": 0.4927843804038355, "learning_rate": 9.096001880171874e-05, "loss": 2.9918, "step": 13289 }, { "epoch": 0.8250046557824818, "grad_norm": 0.5083425543162433, "learning_rate": 9.095794745900988e-05, "loss": 3.052, "step": 13290 }, { "epoch": 0.8250667328822398, "grad_norm": 0.5053722731719329, "learning_rate": 9.095587590261345e-05, "loss": 3.1286, "step": 13291 }, { "epoch": 0.8251288099819977, "grad_norm": 0.5541639946990263, "learning_rate": 9.09538041325403e-05, "loss": 3.0513, "step": 13292 }, { "epoch": 0.8251908870817556, "grad_norm": 0.569113424537419, "learning_rate": 9.095173214880117e-05, "loss": 3.0115, "step": 13293 }, { "epoch": 0.8252529641815134, "grad_norm": 0.45264267737490105, "learning_rate": 9.094965995140694e-05, "loss": 2.9494, "step": 13294 }, { "epoch": 0.8253150412812713, "grad_norm": 0.44446188011672283, "learning_rate": 9.094758754036839e-05, "loss": 2.9199, "step": 13295 }, { "epoch": 0.8253771183810292, "grad_norm": 0.4451485214821133, "learning_rate": 9.094551491569634e-05, "loss": 3.0366, "step": 13296 }, { "epoch": 0.8254391954807871, "grad_norm": 0.5184351896541501, "learning_rate": 9.094344207740158e-05, "loss": 3.0021, "step": 13297 }, { "epoch": 0.8255012725805451, "grad_norm": 0.4982547401034311, "learning_rate": 9.094136902549496e-05, "loss": 3.0117, "step": 13298 }, { "epoch": 0.825563349680303, "grad_norm": 0.457838031069916, "learning_rate": 9.093929575998728e-05, "loss": 3.0422, "step": 13299 }, { "epoch": 0.8256254267800608, "grad_norm": 0.4432211841203608, "learning_rate": 9.093722228088935e-05, "loss": 2.9347, "step": 13300 }, { "epoch": 0.8256875038798187, "grad_norm": 0.47102060335293083, "learning_rate": 9.093514858821199e-05, "loss": 3.0078, "step": 13301 }, { "epoch": 0.8257495809795766, "grad_norm": 0.4399493658126906, "learning_rate": 9.093307468196603e-05, "loss": 2.9619, "step": 13302 }, { "epoch": 0.8258116580793345, "grad_norm": 0.4101440972843859, "learning_rate": 9.093100056216228e-05, "loss": 3.0002, "step": 13303 }, { "epoch": 0.8258737351790925, "grad_norm": 0.43202451317830765, "learning_rate": 9.092892622881155e-05, "loss": 3.0442, "step": 13304 }, { "epoch": 0.8259358122788504, "grad_norm": 0.4950668188955854, "learning_rate": 9.09268516819247e-05, "loss": 2.9763, "step": 13305 }, { "epoch": 0.8259978893786082, "grad_norm": 0.45773113491137585, "learning_rate": 9.092477692151252e-05, "loss": 3.0535, "step": 13306 }, { "epoch": 0.8260599664783661, "grad_norm": 0.48751166158568693, "learning_rate": 9.092270194758583e-05, "loss": 3.0226, "step": 13307 }, { "epoch": 0.826122043578124, "grad_norm": 0.4600948130812901, "learning_rate": 9.09206267601555e-05, "loss": 3.0289, "step": 13308 }, { "epoch": 0.8261841206778819, "grad_norm": 0.4580748616738057, "learning_rate": 9.091855135923229e-05, "loss": 3.0225, "step": 13309 }, { "epoch": 0.8262461977776399, "grad_norm": 0.48111492218182506, "learning_rate": 9.091647574482709e-05, "loss": 3.0848, "step": 13310 }, { "epoch": 0.8263082748773978, "grad_norm": 0.5759534260063496, "learning_rate": 9.09143999169507e-05, "loss": 2.959, "step": 13311 }, { "epoch": 0.8263703519771556, "grad_norm": 0.521345913110645, "learning_rate": 9.091232387561394e-05, "loss": 3.0356, "step": 13312 }, { "epoch": 0.8264324290769135, "grad_norm": 0.5872001014400805, "learning_rate": 9.091024762082766e-05, "loss": 3.0006, "step": 13313 }, { "epoch": 0.8264945061766714, "grad_norm": 0.5559950983874293, "learning_rate": 9.090817115260269e-05, "loss": 3.1322, "step": 13314 }, { "epoch": 0.8265565832764293, "grad_norm": 0.5425462794450449, "learning_rate": 9.090609447094986e-05, "loss": 2.9988, "step": 13315 }, { "epoch": 0.8266186603761873, "grad_norm": 0.5025222757725559, "learning_rate": 9.090401757588001e-05, "loss": 3.046, "step": 13316 }, { "epoch": 0.8266807374759452, "grad_norm": 0.5918067704685913, "learning_rate": 9.090194046740396e-05, "loss": 3.0691, "step": 13317 }, { "epoch": 0.826742814575703, "grad_norm": 0.5600531127980771, "learning_rate": 9.089986314553256e-05, "loss": 3.0523, "step": 13318 }, { "epoch": 0.8268048916754609, "grad_norm": 0.5846203559232649, "learning_rate": 9.089778561027666e-05, "loss": 3.0241, "step": 13319 }, { "epoch": 0.8268669687752188, "grad_norm": 0.5408738980071565, "learning_rate": 9.089570786164707e-05, "loss": 2.9811, "step": 13320 }, { "epoch": 0.8269290458749767, "grad_norm": 0.5575706342247687, "learning_rate": 9.089362989965464e-05, "loss": 2.9526, "step": 13321 }, { "epoch": 0.8269911229747346, "grad_norm": 0.5525178249211016, "learning_rate": 9.089155172431022e-05, "loss": 2.9135, "step": 13322 }, { "epoch": 0.8270532000744926, "grad_norm": 0.5199978962727321, "learning_rate": 9.088947333562466e-05, "loss": 2.9371, "step": 13323 }, { "epoch": 0.8271152771742504, "grad_norm": 0.5649625070335532, "learning_rate": 9.088739473360877e-05, "loss": 3.0194, "step": 13324 }, { "epoch": 0.8271773542740083, "grad_norm": 0.5378534884573831, "learning_rate": 9.088531591827343e-05, "loss": 2.9571, "step": 13325 }, { "epoch": 0.8272394313737662, "grad_norm": 0.5312113747860376, "learning_rate": 9.088323688962947e-05, "loss": 3.0321, "step": 13326 }, { "epoch": 0.8273015084735241, "grad_norm": 0.5768386086653755, "learning_rate": 9.088115764768774e-05, "loss": 2.9885, "step": 13327 }, { "epoch": 0.827363585573282, "grad_norm": 0.5570670385219705, "learning_rate": 9.08790781924591e-05, "loss": 2.9978, "step": 13328 }, { "epoch": 0.82742566267304, "grad_norm": 0.639732483445515, "learning_rate": 9.087699852395435e-05, "loss": 3.0947, "step": 13329 }, { "epoch": 0.8274877397727978, "grad_norm": 0.5890556527303903, "learning_rate": 9.087491864218439e-05, "loss": 3.0151, "step": 13330 }, { "epoch": 0.8275498168725557, "grad_norm": 0.5969261497031157, "learning_rate": 9.087283854716006e-05, "loss": 3.0352, "step": 13331 }, { "epoch": 0.8276118939723136, "grad_norm": 0.5470647210322631, "learning_rate": 9.087075823889222e-05, "loss": 2.9621, "step": 13332 }, { "epoch": 0.8276739710720715, "grad_norm": 0.5507128246901865, "learning_rate": 9.08686777173917e-05, "loss": 3.0755, "step": 13333 }, { "epoch": 0.8277360481718294, "grad_norm": 0.5389072360469107, "learning_rate": 9.086659698266937e-05, "loss": 3.0211, "step": 13334 }, { "epoch": 0.8277981252715874, "grad_norm": 0.5107776636818692, "learning_rate": 9.086451603473608e-05, "loss": 3.0309, "step": 13335 }, { "epoch": 0.8278602023713452, "grad_norm": 0.553774229600281, "learning_rate": 9.086243487360269e-05, "loss": 2.9553, "step": 13336 }, { "epoch": 0.8279222794711031, "grad_norm": 0.4842948078099707, "learning_rate": 9.086035349928005e-05, "loss": 3.0548, "step": 13337 }, { "epoch": 0.827984356570861, "grad_norm": 0.5690816111657514, "learning_rate": 9.085827191177905e-05, "loss": 3.0252, "step": 13338 }, { "epoch": 0.8280464336706189, "grad_norm": 0.5632905967243804, "learning_rate": 9.08561901111105e-05, "loss": 2.9317, "step": 13339 }, { "epoch": 0.8281085107703768, "grad_norm": 0.5107145867265849, "learning_rate": 9.085410809728532e-05, "loss": 2.9812, "step": 13340 }, { "epoch": 0.8281705878701348, "grad_norm": 0.5332589871274732, "learning_rate": 9.085202587031432e-05, "loss": 3.041, "step": 13341 }, { "epoch": 0.8282326649698926, "grad_norm": 0.5471342834528705, "learning_rate": 9.084994343020838e-05, "loss": 3.0605, "step": 13342 }, { "epoch": 0.8282947420696505, "grad_norm": 0.5624496914300743, "learning_rate": 9.084786077697837e-05, "loss": 3.0339, "step": 13343 }, { "epoch": 0.8283568191694084, "grad_norm": 0.5177341729770011, "learning_rate": 9.084577791063516e-05, "loss": 3.0088, "step": 13344 }, { "epoch": 0.8284188962691663, "grad_norm": 0.4769784800345494, "learning_rate": 9.084369483118961e-05, "loss": 3.0197, "step": 13345 }, { "epoch": 0.8284809733689242, "grad_norm": 0.49894434531666215, "learning_rate": 9.084161153865258e-05, "loss": 3.0146, "step": 13346 }, { "epoch": 0.8285430504686822, "grad_norm": 0.5055173505622104, "learning_rate": 9.083952803303497e-05, "loss": 3.0517, "step": 13347 }, { "epoch": 0.82860512756844, "grad_norm": 0.44645956442742274, "learning_rate": 9.083744431434761e-05, "loss": 2.9807, "step": 13348 }, { "epoch": 0.8286672046681979, "grad_norm": 0.5359141697824205, "learning_rate": 9.083536038260139e-05, "loss": 3.0214, "step": 13349 }, { "epoch": 0.8287292817679558, "grad_norm": 0.502552402050945, "learning_rate": 9.083327623780718e-05, "loss": 3.0207, "step": 13350 }, { "epoch": 0.8287913588677137, "grad_norm": 0.5115343300270312, "learning_rate": 9.083119187997587e-05, "loss": 2.9279, "step": 13351 }, { "epoch": 0.8288534359674716, "grad_norm": 0.4062216822218094, "learning_rate": 9.08291073091183e-05, "loss": 3.0044, "step": 13352 }, { "epoch": 0.8289155130672295, "grad_norm": 0.5025420241077745, "learning_rate": 9.082702252524537e-05, "loss": 3.0415, "step": 13353 }, { "epoch": 0.8289775901669874, "grad_norm": 0.449929959753956, "learning_rate": 9.082493752836796e-05, "loss": 2.9854, "step": 13354 }, { "epoch": 0.8290396672667453, "grad_norm": 0.4765737876180686, "learning_rate": 9.082285231849694e-05, "loss": 2.9396, "step": 13355 }, { "epoch": 0.8291017443665032, "grad_norm": 0.46755858645811027, "learning_rate": 9.08207668956432e-05, "loss": 3.0231, "step": 13356 }, { "epoch": 0.8291638214662611, "grad_norm": 0.5073369073244813, "learning_rate": 9.081868125981758e-05, "loss": 2.9609, "step": 13357 }, { "epoch": 0.829225898566019, "grad_norm": 0.3958542425087542, "learning_rate": 9.081659541103102e-05, "loss": 2.9967, "step": 13358 }, { "epoch": 0.8292879756657769, "grad_norm": 0.4934399428014221, "learning_rate": 9.081450934929437e-05, "loss": 3.0672, "step": 13359 }, { "epoch": 0.8293500527655348, "grad_norm": 0.4182946725885463, "learning_rate": 9.081242307461851e-05, "loss": 3.0572, "step": 13360 }, { "epoch": 0.8294121298652927, "grad_norm": 0.44799056293774475, "learning_rate": 9.081033658701432e-05, "loss": 2.9869, "step": 13361 }, { "epoch": 0.8294742069650506, "grad_norm": 0.5470346217853761, "learning_rate": 9.08082498864927e-05, "loss": 3.0389, "step": 13362 }, { "epoch": 0.8295362840648085, "grad_norm": 0.5043740873080366, "learning_rate": 9.080616297306455e-05, "loss": 3.0283, "step": 13363 }, { "epoch": 0.8295983611645664, "grad_norm": 0.5356860079568182, "learning_rate": 9.080407584674073e-05, "loss": 2.9932, "step": 13364 }, { "epoch": 0.8296604382643243, "grad_norm": 0.5382357906197024, "learning_rate": 9.080198850753214e-05, "loss": 2.9998, "step": 13365 }, { "epoch": 0.8297225153640821, "grad_norm": 0.46979891721398503, "learning_rate": 9.079990095544969e-05, "loss": 2.9519, "step": 13366 }, { "epoch": 0.8297845924638401, "grad_norm": 0.5302925738680605, "learning_rate": 9.079781319050422e-05, "loss": 3.0307, "step": 13367 }, { "epoch": 0.829846669563598, "grad_norm": 0.5156452579564693, "learning_rate": 9.079572521270667e-05, "loss": 3.0687, "step": 13368 }, { "epoch": 0.8299087466633559, "grad_norm": 0.5481906052565508, "learning_rate": 9.079363702206792e-05, "loss": 3.0118, "step": 13369 }, { "epoch": 0.8299708237631138, "grad_norm": 0.484315360919981, "learning_rate": 9.079154861859885e-05, "loss": 3.0173, "step": 13370 }, { "epoch": 0.8300329008628717, "grad_norm": 0.49690928970634524, "learning_rate": 9.078946000231039e-05, "loss": 3.0052, "step": 13371 }, { "epoch": 0.8300949779626295, "grad_norm": 0.5255993566710165, "learning_rate": 9.07873711732134e-05, "loss": 3.0715, "step": 13372 }, { "epoch": 0.8301570550623875, "grad_norm": 0.5276321752657251, "learning_rate": 9.07852821313188e-05, "loss": 2.9577, "step": 13373 }, { "epoch": 0.8302191321621454, "grad_norm": 0.546486498574309, "learning_rate": 9.078319287663747e-05, "loss": 3.0112, "step": 13374 }, { "epoch": 0.8302812092619033, "grad_norm": 0.6011412247744591, "learning_rate": 9.078110340918033e-05, "loss": 2.9905, "step": 13375 }, { "epoch": 0.8303432863616612, "grad_norm": 0.623719931089857, "learning_rate": 9.077901372895828e-05, "loss": 2.9905, "step": 13376 }, { "epoch": 0.8304053634614191, "grad_norm": 0.5010196005495308, "learning_rate": 9.077692383598221e-05, "loss": 3.0808, "step": 13377 }, { "epoch": 0.8304674405611769, "grad_norm": 0.5670276784482159, "learning_rate": 9.077483373026303e-05, "loss": 2.9979, "step": 13378 }, { "epoch": 0.8305295176609349, "grad_norm": 0.560647297546547, "learning_rate": 9.077274341181166e-05, "loss": 2.9543, "step": 13379 }, { "epoch": 0.8305915947606928, "grad_norm": 0.5713143367556275, "learning_rate": 9.077065288063897e-05, "loss": 3.0065, "step": 13380 }, { "epoch": 0.8306536718604507, "grad_norm": 0.5391719817509323, "learning_rate": 9.076856213675588e-05, "loss": 3.0423, "step": 13381 }, { "epoch": 0.8307157489602086, "grad_norm": 0.5688706029103013, "learning_rate": 9.076647118017332e-05, "loss": 3.0207, "step": 13382 }, { "epoch": 0.8307778260599665, "grad_norm": 0.5196172026470625, "learning_rate": 9.076438001090217e-05, "loss": 2.9364, "step": 13383 }, { "epoch": 0.8308399031597243, "grad_norm": 0.5571075622589016, "learning_rate": 9.076228862895338e-05, "loss": 3.067, "step": 13384 }, { "epoch": 0.8309019802594823, "grad_norm": 0.5047828088293624, "learning_rate": 9.076019703433781e-05, "loss": 3.0881, "step": 13385 }, { "epoch": 0.8309640573592402, "grad_norm": 0.6001342563626431, "learning_rate": 9.07581052270664e-05, "loss": 2.97, "step": 13386 }, { "epoch": 0.8310261344589981, "grad_norm": 0.6332230590092356, "learning_rate": 9.075601320715006e-05, "loss": 2.9156, "step": 13387 }, { "epoch": 0.831088211558756, "grad_norm": 0.5163105251217689, "learning_rate": 9.075392097459971e-05, "loss": 3.0193, "step": 13388 }, { "epoch": 0.8311502886585139, "grad_norm": 0.581348171967621, "learning_rate": 9.075182852942627e-05, "loss": 2.9407, "step": 13389 }, { "epoch": 0.8312123657582717, "grad_norm": 0.5846299650508732, "learning_rate": 9.074973587164064e-05, "loss": 2.9168, "step": 13390 }, { "epoch": 0.8312744428580296, "grad_norm": 0.6017410706862756, "learning_rate": 9.074764300125374e-05, "loss": 3.1028, "step": 13391 }, { "epoch": 0.8313365199577876, "grad_norm": 0.5365398647486151, "learning_rate": 9.074554991827651e-05, "loss": 3.0248, "step": 13392 }, { "epoch": 0.8313985970575455, "grad_norm": 0.5426629117627638, "learning_rate": 9.074345662271984e-05, "loss": 3.0281, "step": 13393 }, { "epoch": 0.8314606741573034, "grad_norm": 0.5195355522723355, "learning_rate": 9.074136311459467e-05, "loss": 2.9844, "step": 13394 }, { "epoch": 0.8315227512570613, "grad_norm": 0.5464939970051499, "learning_rate": 9.073926939391191e-05, "loss": 2.9964, "step": 13395 }, { "epoch": 0.8315848283568191, "grad_norm": 0.5207734709352405, "learning_rate": 9.07371754606825e-05, "loss": 2.9633, "step": 13396 }, { "epoch": 0.831646905456577, "grad_norm": 0.5606440283821746, "learning_rate": 9.073508131491736e-05, "loss": 3.1135, "step": 13397 }, { "epoch": 0.831708982556335, "grad_norm": 0.5925300010296599, "learning_rate": 9.073298695662743e-05, "loss": 3.0611, "step": 13398 }, { "epoch": 0.8317710596560929, "grad_norm": 0.5626190907189829, "learning_rate": 9.073089238582357e-05, "loss": 3.0078, "step": 13399 }, { "epoch": 0.8318331367558508, "grad_norm": 0.5472528242401675, "learning_rate": 9.072879760251679e-05, "loss": 2.9776, "step": 13400 }, { "epoch": 0.8318952138556087, "grad_norm": 0.621700518298351, "learning_rate": 9.072670260671799e-05, "loss": 2.9866, "step": 13401 }, { "epoch": 0.8319572909553665, "grad_norm": 0.5171260370802444, "learning_rate": 9.072460739843807e-05, "loss": 2.9678, "step": 13402 }, { "epoch": 0.8320193680551244, "grad_norm": 0.578802459308419, "learning_rate": 9.0722511977688e-05, "loss": 3.0494, "step": 13403 }, { "epoch": 0.8320814451548824, "grad_norm": 0.5156625965167789, "learning_rate": 9.072041634447871e-05, "loss": 3.0173, "step": 13404 }, { "epoch": 0.8321435222546403, "grad_norm": 0.5318289855435421, "learning_rate": 9.071832049882112e-05, "loss": 2.9871, "step": 13405 }, { "epoch": 0.8322055993543982, "grad_norm": 0.5292925957651049, "learning_rate": 9.071622444072615e-05, "loss": 2.9458, "step": 13406 }, { "epoch": 0.8322676764541561, "grad_norm": 0.4912071000360712, "learning_rate": 9.071412817020477e-05, "loss": 3.0774, "step": 13407 }, { "epoch": 0.8323297535539139, "grad_norm": 0.5052600388433259, "learning_rate": 9.071203168726789e-05, "loss": 2.9509, "step": 13408 }, { "epoch": 0.8323918306536718, "grad_norm": 0.5661972186350042, "learning_rate": 9.070993499192645e-05, "loss": 3.0536, "step": 13409 }, { "epoch": 0.8324539077534298, "grad_norm": 0.5632109123259925, "learning_rate": 9.070783808419143e-05, "loss": 3.0775, "step": 13410 }, { "epoch": 0.8325159848531877, "grad_norm": 0.583711337048748, "learning_rate": 9.07057409640737e-05, "loss": 3.0222, "step": 13411 }, { "epoch": 0.8325780619529456, "grad_norm": 0.5135809234048031, "learning_rate": 9.070364363158426e-05, "loss": 3.0772, "step": 13412 }, { "epoch": 0.8326401390527035, "grad_norm": 0.5654326654596261, "learning_rate": 9.070154608673403e-05, "loss": 2.9657, "step": 13413 }, { "epoch": 0.8327022161524613, "grad_norm": 0.525294553914243, "learning_rate": 9.069944832953393e-05, "loss": 2.9525, "step": 13414 }, { "epoch": 0.8327642932522192, "grad_norm": 0.5341338533756447, "learning_rate": 9.069735035999495e-05, "loss": 3.0208, "step": 13415 }, { "epoch": 0.8328263703519772, "grad_norm": 0.4853091768297146, "learning_rate": 9.069525217812801e-05, "loss": 3.0165, "step": 13416 }, { "epoch": 0.8328884474517351, "grad_norm": 0.5247853987186841, "learning_rate": 9.069315378394406e-05, "loss": 2.9508, "step": 13417 }, { "epoch": 0.832950524551493, "grad_norm": 0.4714173402831455, "learning_rate": 9.069105517745404e-05, "loss": 2.8917, "step": 13418 }, { "epoch": 0.8330126016512509, "grad_norm": 0.5588225849633754, "learning_rate": 9.06889563586689e-05, "loss": 3.0194, "step": 13419 }, { "epoch": 0.8330746787510087, "grad_norm": 0.5074280383856792, "learning_rate": 9.068685732759961e-05, "loss": 3.0743, "step": 13420 }, { "epoch": 0.8331367558507666, "grad_norm": 0.48653579016294557, "learning_rate": 9.068475808425711e-05, "loss": 2.9883, "step": 13421 }, { "epoch": 0.8331988329505245, "grad_norm": 0.484439076523079, "learning_rate": 9.068265862865235e-05, "loss": 3.0408, "step": 13422 }, { "epoch": 0.8332609100502825, "grad_norm": 0.5356697347826508, "learning_rate": 9.068055896079627e-05, "loss": 3.0164, "step": 13423 }, { "epoch": 0.8333229871500404, "grad_norm": 0.5025725939490938, "learning_rate": 9.067845908069988e-05, "loss": 2.9562, "step": 13424 }, { "epoch": 0.8333850642497983, "grad_norm": 0.47651404775484874, "learning_rate": 9.067635898837406e-05, "loss": 3.0249, "step": 13425 }, { "epoch": 0.8334471413495561, "grad_norm": 0.5480836047088306, "learning_rate": 9.067425868382979e-05, "loss": 3.1113, "step": 13426 }, { "epoch": 0.833509218449314, "grad_norm": 0.6298052599574914, "learning_rate": 9.067215816707806e-05, "loss": 3.0084, "step": 13427 }, { "epoch": 0.8335712955490719, "grad_norm": 0.5231449676221317, "learning_rate": 9.06700574381298e-05, "loss": 3.0128, "step": 13428 }, { "epoch": 0.8336333726488299, "grad_norm": 0.49723416009379323, "learning_rate": 9.066795649699598e-05, "loss": 3.0468, "step": 13429 }, { "epoch": 0.8336954497485878, "grad_norm": 0.5369396149517399, "learning_rate": 9.066585534368756e-05, "loss": 3.0203, "step": 13430 }, { "epoch": 0.8337575268483457, "grad_norm": 0.49094107402566545, "learning_rate": 9.06637539782155e-05, "loss": 2.9299, "step": 13431 }, { "epoch": 0.8338196039481035, "grad_norm": 0.5308678037579198, "learning_rate": 9.066165240059076e-05, "loss": 3.1077, "step": 13432 }, { "epoch": 0.8338816810478614, "grad_norm": 0.5275062310723108, "learning_rate": 9.065955061082431e-05, "loss": 2.9281, "step": 13433 }, { "epoch": 0.8339437581476193, "grad_norm": 0.4759255984215256, "learning_rate": 9.065744860892712e-05, "loss": 2.9507, "step": 13434 }, { "epoch": 0.8340058352473773, "grad_norm": 0.508210774275428, "learning_rate": 9.065534639491013e-05, "loss": 2.9898, "step": 13435 }, { "epoch": 0.8340679123471352, "grad_norm": 0.4865151624002624, "learning_rate": 9.065324396878434e-05, "loss": 2.9619, "step": 13436 }, { "epoch": 0.8341299894468931, "grad_norm": 0.5508108739799688, "learning_rate": 9.065114133056072e-05, "loss": 3.0051, "step": 13437 }, { "epoch": 0.8341920665466509, "grad_norm": 0.5347374161162104, "learning_rate": 9.064903848025021e-05, "loss": 2.9462, "step": 13438 }, { "epoch": 0.8342541436464088, "grad_norm": 0.5399130593177541, "learning_rate": 9.064693541786381e-05, "loss": 3.0029, "step": 13439 }, { "epoch": 0.8343162207461667, "grad_norm": 0.5876196556797696, "learning_rate": 9.064483214341248e-05, "loss": 2.9815, "step": 13440 }, { "epoch": 0.8343782978459247, "grad_norm": 0.5224504416293858, "learning_rate": 9.064272865690718e-05, "loss": 3.0761, "step": 13441 }, { "epoch": 0.8344403749456826, "grad_norm": 0.5868905772050232, "learning_rate": 9.064062495835892e-05, "loss": 3.0141, "step": 13442 }, { "epoch": 0.8345024520454405, "grad_norm": 0.5606951102196467, "learning_rate": 9.063852104777864e-05, "loss": 3.0411, "step": 13443 }, { "epoch": 0.8345645291451983, "grad_norm": 0.5390677520938381, "learning_rate": 9.063641692517733e-05, "loss": 3.0272, "step": 13444 }, { "epoch": 0.8346266062449562, "grad_norm": 0.5251280685168072, "learning_rate": 9.063431259056597e-05, "loss": 3.0015, "step": 13445 }, { "epoch": 0.8346886833447141, "grad_norm": 0.4548619948160972, "learning_rate": 9.063220804395554e-05, "loss": 2.9263, "step": 13446 }, { "epoch": 0.834750760444472, "grad_norm": 0.5532455506936637, "learning_rate": 9.063010328535701e-05, "loss": 2.8732, "step": 13447 }, { "epoch": 0.83481283754423, "grad_norm": 0.46874898274629206, "learning_rate": 9.062799831478137e-05, "loss": 3.025, "step": 13448 }, { "epoch": 0.8348749146439879, "grad_norm": 0.5201417877762736, "learning_rate": 9.06258931322396e-05, "loss": 2.9948, "step": 13449 }, { "epoch": 0.8349369917437457, "grad_norm": 0.4367704098195483, "learning_rate": 9.062378773774269e-05, "loss": 2.9622, "step": 13450 }, { "epoch": 0.8349990688435036, "grad_norm": 0.4820196969550516, "learning_rate": 9.062168213130162e-05, "loss": 2.9397, "step": 13451 }, { "epoch": 0.8350611459432615, "grad_norm": 0.5017346749144007, "learning_rate": 9.061957631292737e-05, "loss": 2.9152, "step": 13452 }, { "epoch": 0.8351232230430194, "grad_norm": 0.4918383919399614, "learning_rate": 9.061747028263093e-05, "loss": 3.013, "step": 13453 }, { "epoch": 0.8351853001427774, "grad_norm": 0.4650616655940776, "learning_rate": 9.061536404042327e-05, "loss": 2.9422, "step": 13454 }, { "epoch": 0.8352473772425353, "grad_norm": 0.45185372818281205, "learning_rate": 9.06132575863154e-05, "loss": 2.9816, "step": 13455 }, { "epoch": 0.8353094543422931, "grad_norm": 0.4665311911183847, "learning_rate": 9.061115092031832e-05, "loss": 2.9391, "step": 13456 }, { "epoch": 0.835371531442051, "grad_norm": 0.42711854611199357, "learning_rate": 9.0609044042443e-05, "loss": 3.0082, "step": 13457 }, { "epoch": 0.8354336085418089, "grad_norm": 0.44720553531705576, "learning_rate": 9.060693695270044e-05, "loss": 3.0602, "step": 13458 }, { "epoch": 0.8354956856415668, "grad_norm": 0.4450575952279733, "learning_rate": 9.060482965110165e-05, "loss": 2.9548, "step": 13459 }, { "epoch": 0.8355577627413248, "grad_norm": 0.45061397324719943, "learning_rate": 9.060272213765758e-05, "loss": 2.9831, "step": 13460 }, { "epoch": 0.8356198398410827, "grad_norm": 0.5270785194549066, "learning_rate": 9.060061441237926e-05, "loss": 3.0059, "step": 13461 }, { "epoch": 0.8356819169408405, "grad_norm": 0.44870644257058223, "learning_rate": 9.059850647527768e-05, "loss": 2.9966, "step": 13462 }, { "epoch": 0.8357439940405984, "grad_norm": 0.4660956250932134, "learning_rate": 9.059639832636383e-05, "loss": 3.016, "step": 13463 }, { "epoch": 0.8358060711403563, "grad_norm": 0.5724549224547917, "learning_rate": 9.059428996564872e-05, "loss": 3.0829, "step": 13464 }, { "epoch": 0.8358681482401142, "grad_norm": 0.49016430722619636, "learning_rate": 9.059218139314336e-05, "loss": 3.0583, "step": 13465 }, { "epoch": 0.8359302253398722, "grad_norm": 0.5100915385822731, "learning_rate": 9.059007260885872e-05, "loss": 3.0155, "step": 13466 }, { "epoch": 0.8359923024396301, "grad_norm": 0.5485884662439999, "learning_rate": 9.058796361280582e-05, "loss": 3.0579, "step": 13467 }, { "epoch": 0.8360543795393879, "grad_norm": 0.486618782419208, "learning_rate": 9.058585440499565e-05, "loss": 2.9701, "step": 13468 }, { "epoch": 0.8361164566391458, "grad_norm": 0.5586090085806603, "learning_rate": 9.058374498543925e-05, "loss": 2.9961, "step": 13469 }, { "epoch": 0.8361785337389037, "grad_norm": 0.5289437780772818, "learning_rate": 9.058163535414758e-05, "loss": 3.0285, "step": 13470 }, { "epoch": 0.8362406108386616, "grad_norm": 0.4408061750762218, "learning_rate": 9.057952551113167e-05, "loss": 3.0294, "step": 13471 }, { "epoch": 0.8363026879384196, "grad_norm": 0.479276185028281, "learning_rate": 9.057741545640253e-05, "loss": 2.994, "step": 13472 }, { "epoch": 0.8363647650381775, "grad_norm": 0.5018137578783366, "learning_rate": 9.057530518997115e-05, "loss": 2.9656, "step": 13473 }, { "epoch": 0.8364268421379353, "grad_norm": 0.4353070817816947, "learning_rate": 9.057319471184856e-05, "loss": 2.9083, "step": 13474 }, { "epoch": 0.8364889192376932, "grad_norm": 0.4579452602072051, "learning_rate": 9.057108402204577e-05, "loss": 2.9725, "step": 13475 }, { "epoch": 0.8365509963374511, "grad_norm": 0.47112432605212423, "learning_rate": 9.056897312057378e-05, "loss": 2.9504, "step": 13476 }, { "epoch": 0.836613073437209, "grad_norm": 0.4475314587185777, "learning_rate": 9.056686200744361e-05, "loss": 2.9956, "step": 13477 }, { "epoch": 0.836675150536967, "grad_norm": 0.5068756264378149, "learning_rate": 9.056475068266628e-05, "loss": 2.9474, "step": 13478 }, { "epoch": 0.8367372276367249, "grad_norm": 0.43321945599936856, "learning_rate": 9.056263914625279e-05, "loss": 2.8343, "step": 13479 }, { "epoch": 0.8367993047364827, "grad_norm": 0.462550557440456, "learning_rate": 9.056052739821417e-05, "loss": 2.9678, "step": 13480 }, { "epoch": 0.8368613818362406, "grad_norm": 0.5135219630806831, "learning_rate": 9.055841543856142e-05, "loss": 2.9542, "step": 13481 }, { "epoch": 0.8369234589359985, "grad_norm": 0.4659772244769391, "learning_rate": 9.055630326730557e-05, "loss": 3.033, "step": 13482 }, { "epoch": 0.8369855360357564, "grad_norm": 0.5643291403217922, "learning_rate": 9.055419088445767e-05, "loss": 3.0613, "step": 13483 }, { "epoch": 0.8370476131355143, "grad_norm": 0.48117757475199724, "learning_rate": 9.055207829002867e-05, "loss": 2.9856, "step": 13484 }, { "epoch": 0.8371096902352723, "grad_norm": 0.433297666203614, "learning_rate": 9.054996548402967e-05, "loss": 2.9747, "step": 13485 }, { "epoch": 0.8371717673350301, "grad_norm": 0.4846301329780961, "learning_rate": 9.054785246647164e-05, "loss": 3.0526, "step": 13486 }, { "epoch": 0.837233844434788, "grad_norm": 0.5271839606183808, "learning_rate": 9.054573923736562e-05, "loss": 2.9738, "step": 13487 }, { "epoch": 0.8372959215345459, "grad_norm": 0.5354600586977537, "learning_rate": 9.054362579672263e-05, "loss": 2.9682, "step": 13488 }, { "epoch": 0.8373579986343038, "grad_norm": 0.48088076223972215, "learning_rate": 9.054151214455371e-05, "loss": 2.9853, "step": 13489 }, { "epoch": 0.8374200757340617, "grad_norm": 0.5392407661570577, "learning_rate": 9.053939828086988e-05, "loss": 3.0384, "step": 13490 }, { "epoch": 0.8374821528338195, "grad_norm": 0.5651090829755513, "learning_rate": 9.053728420568217e-05, "loss": 2.9561, "step": 13491 }, { "epoch": 0.8375442299335775, "grad_norm": 0.4752875612075526, "learning_rate": 9.05351699190016e-05, "loss": 2.9484, "step": 13492 }, { "epoch": 0.8376063070333354, "grad_norm": 0.5155734412107167, "learning_rate": 9.05330554208392e-05, "loss": 2.9534, "step": 13493 }, { "epoch": 0.8376683841330933, "grad_norm": 0.5456458857013908, "learning_rate": 9.053094071120602e-05, "loss": 2.9973, "step": 13494 }, { "epoch": 0.8377304612328512, "grad_norm": 0.5285120138536958, "learning_rate": 9.052882579011309e-05, "loss": 3.112, "step": 13495 }, { "epoch": 0.8377925383326091, "grad_norm": 0.5010848495316237, "learning_rate": 9.052671065757144e-05, "loss": 3.0483, "step": 13496 }, { "epoch": 0.837854615432367, "grad_norm": 0.6082897303470484, "learning_rate": 9.052459531359209e-05, "loss": 2.9775, "step": 13497 }, { "epoch": 0.8379166925321249, "grad_norm": 0.5635696149184222, "learning_rate": 9.052247975818608e-05, "loss": 3.0751, "step": 13498 }, { "epoch": 0.8379787696318828, "grad_norm": 0.4907572264379752, "learning_rate": 9.052036399136448e-05, "loss": 2.9417, "step": 13499 }, { "epoch": 0.8380408467316407, "grad_norm": 0.5213564026359402, "learning_rate": 9.051824801313828e-05, "loss": 2.9272, "step": 13500 }, { "epoch": 0.8381029238313986, "grad_norm": 0.47506511455248523, "learning_rate": 9.051613182351857e-05, "loss": 2.9541, "step": 13501 }, { "epoch": 0.8381650009311565, "grad_norm": 0.533664350181562, "learning_rate": 9.051401542251634e-05, "loss": 2.9874, "step": 13502 }, { "epoch": 0.8382270780309143, "grad_norm": 0.4775368314039538, "learning_rate": 9.051189881014267e-05, "loss": 2.971, "step": 13503 }, { "epoch": 0.8382891551306723, "grad_norm": 0.47758185696122674, "learning_rate": 9.05097819864086e-05, "loss": 2.9698, "step": 13504 }, { "epoch": 0.8383512322304302, "grad_norm": 0.5134087240243213, "learning_rate": 9.050766495132514e-05, "loss": 3.0377, "step": 13505 }, { "epoch": 0.8384133093301881, "grad_norm": 0.4965721737043585, "learning_rate": 9.050554770490336e-05, "loss": 3.0263, "step": 13506 }, { "epoch": 0.838475386429946, "grad_norm": 0.46309428824912097, "learning_rate": 9.050343024715433e-05, "loss": 3.0378, "step": 13507 }, { "epoch": 0.8385374635297039, "grad_norm": 0.5199451948592236, "learning_rate": 9.050131257808904e-05, "loss": 2.965, "step": 13508 }, { "epoch": 0.8385995406294617, "grad_norm": 0.5274251239299651, "learning_rate": 9.049919469771859e-05, "loss": 3.1161, "step": 13509 }, { "epoch": 0.8386616177292197, "grad_norm": 0.49004468249446664, "learning_rate": 9.0497076606054e-05, "loss": 3.068, "step": 13510 }, { "epoch": 0.8387236948289776, "grad_norm": 0.4586826875088269, "learning_rate": 9.049495830310635e-05, "loss": 3.0435, "step": 13511 }, { "epoch": 0.8387857719287355, "grad_norm": 0.4635394574944934, "learning_rate": 9.049283978888665e-05, "loss": 2.9022, "step": 13512 }, { "epoch": 0.8388478490284934, "grad_norm": 0.4900004220007033, "learning_rate": 9.049072106340599e-05, "loss": 3.0748, "step": 13513 }, { "epoch": 0.8389099261282513, "grad_norm": 0.5363574504128215, "learning_rate": 9.04886021266754e-05, "loss": 3.0435, "step": 13514 }, { "epoch": 0.8389720032280091, "grad_norm": 0.5670052879362758, "learning_rate": 9.048648297870594e-05, "loss": 3.0434, "step": 13515 }, { "epoch": 0.839034080327767, "grad_norm": 0.5352271095754952, "learning_rate": 9.048436361950867e-05, "loss": 2.9745, "step": 13516 }, { "epoch": 0.839096157427525, "grad_norm": 0.5569947839386554, "learning_rate": 9.048224404909466e-05, "loss": 3.0293, "step": 13517 }, { "epoch": 0.8391582345272829, "grad_norm": 0.5211479730719096, "learning_rate": 9.048012426747495e-05, "loss": 2.9991, "step": 13518 }, { "epoch": 0.8392203116270408, "grad_norm": 0.5057866106759776, "learning_rate": 9.04780042746606e-05, "loss": 3.0726, "step": 13519 }, { "epoch": 0.8392823887267987, "grad_norm": 0.4920104177504037, "learning_rate": 9.047588407066268e-05, "loss": 2.9396, "step": 13520 }, { "epoch": 0.8393444658265565, "grad_norm": 0.49889936603082263, "learning_rate": 9.047376365549223e-05, "loss": 2.8871, "step": 13521 }, { "epoch": 0.8394065429263144, "grad_norm": 0.49933718916557, "learning_rate": 9.047164302916036e-05, "loss": 3.0617, "step": 13522 }, { "epoch": 0.8394686200260724, "grad_norm": 0.5181167103526445, "learning_rate": 9.046952219167809e-05, "loss": 3.0844, "step": 13523 }, { "epoch": 0.8395306971258303, "grad_norm": 0.47046340283079363, "learning_rate": 9.046740114305649e-05, "loss": 3.0257, "step": 13524 }, { "epoch": 0.8395927742255882, "grad_norm": 0.4924603720034093, "learning_rate": 9.046527988330664e-05, "loss": 2.8703, "step": 13525 }, { "epoch": 0.8396548513253461, "grad_norm": 0.5468034152454495, "learning_rate": 9.04631584124396e-05, "loss": 3.0811, "step": 13526 }, { "epoch": 0.8397169284251039, "grad_norm": 0.527954341469418, "learning_rate": 9.046103673046645e-05, "loss": 2.997, "step": 13527 }, { "epoch": 0.8397790055248618, "grad_norm": 0.5133145325900422, "learning_rate": 9.045891483739824e-05, "loss": 3.0316, "step": 13528 }, { "epoch": 0.8398410826246198, "grad_norm": 0.49289259284535575, "learning_rate": 9.045679273324603e-05, "loss": 3.0288, "step": 13529 }, { "epoch": 0.8399031597243777, "grad_norm": 0.5420862339664451, "learning_rate": 9.045467041802094e-05, "loss": 2.9851, "step": 13530 }, { "epoch": 0.8399652368241356, "grad_norm": 0.5148923466023069, "learning_rate": 9.045254789173401e-05, "loss": 3.0107, "step": 13531 }, { "epoch": 0.8400273139238935, "grad_norm": 0.5652238803988597, "learning_rate": 9.045042515439631e-05, "loss": 2.9914, "step": 13532 }, { "epoch": 0.8400893910236513, "grad_norm": 0.5160079458792968, "learning_rate": 9.044830220601892e-05, "loss": 3.1478, "step": 13533 }, { "epoch": 0.8401514681234092, "grad_norm": 0.5526537887005063, "learning_rate": 9.044617904661293e-05, "loss": 3.0315, "step": 13534 }, { "epoch": 0.8402135452231672, "grad_norm": 0.5050534636929819, "learning_rate": 9.044405567618938e-05, "loss": 2.9405, "step": 13535 }, { "epoch": 0.8402756223229251, "grad_norm": 0.4772931267199697, "learning_rate": 9.044193209475939e-05, "loss": 2.9992, "step": 13536 }, { "epoch": 0.840337699422683, "grad_norm": 0.5371677158922267, "learning_rate": 9.043980830233403e-05, "loss": 3.0286, "step": 13537 }, { "epoch": 0.8403997765224409, "grad_norm": 0.4997320500280307, "learning_rate": 9.043768429892437e-05, "loss": 3.0041, "step": 13538 }, { "epoch": 0.8404618536221987, "grad_norm": 0.5411496874243983, "learning_rate": 9.04355600845415e-05, "loss": 3.087, "step": 13539 }, { "epoch": 0.8405239307219566, "grad_norm": 0.6363012899474156, "learning_rate": 9.043343565919648e-05, "loss": 3.0278, "step": 13540 }, { "epoch": 0.8405860078217146, "grad_norm": 0.6400244153849276, "learning_rate": 9.043131102290043e-05, "loss": 3.0175, "step": 13541 }, { "epoch": 0.8406480849214725, "grad_norm": 0.5282668335196244, "learning_rate": 9.04291861756644e-05, "loss": 2.9601, "step": 13542 }, { "epoch": 0.8407101620212304, "grad_norm": 0.6746062057235128, "learning_rate": 9.04270611174995e-05, "loss": 3.0557, "step": 13543 }, { "epoch": 0.8407722391209883, "grad_norm": 0.5640000105293084, "learning_rate": 9.04249358484168e-05, "loss": 3.0559, "step": 13544 }, { "epoch": 0.8408343162207461, "grad_norm": 0.5560357002470835, "learning_rate": 9.04228103684274e-05, "loss": 2.9324, "step": 13545 }, { "epoch": 0.840896393320504, "grad_norm": 0.5770158179642116, "learning_rate": 9.042068467754238e-05, "loss": 2.9971, "step": 13546 }, { "epoch": 0.840958470420262, "grad_norm": 0.5118086932393959, "learning_rate": 9.041855877577285e-05, "loss": 2.919, "step": 13547 }, { "epoch": 0.8410205475200199, "grad_norm": 0.6097498498864046, "learning_rate": 9.041643266312987e-05, "loss": 2.9722, "step": 13548 }, { "epoch": 0.8410826246197778, "grad_norm": 0.5424579989967026, "learning_rate": 9.041430633962456e-05, "loss": 2.9871, "step": 13549 }, { "epoch": 0.8411447017195357, "grad_norm": 0.6340163988450459, "learning_rate": 9.041217980526799e-05, "loss": 2.9977, "step": 13550 }, { "epoch": 0.8412067788192935, "grad_norm": 0.5664979794643829, "learning_rate": 9.041005306007128e-05, "loss": 2.9919, "step": 13551 }, { "epoch": 0.8412688559190514, "grad_norm": 0.5823463764545975, "learning_rate": 9.040792610404552e-05, "loss": 2.9663, "step": 13552 }, { "epoch": 0.8413309330188093, "grad_norm": 0.5322883220151966, "learning_rate": 9.04057989372018e-05, "loss": 2.9662, "step": 13553 }, { "epoch": 0.8413930101185673, "grad_norm": 0.5185492563771463, "learning_rate": 9.040367155955121e-05, "loss": 2.8662, "step": 13554 }, { "epoch": 0.8414550872183252, "grad_norm": 0.477488431216197, "learning_rate": 9.040154397110485e-05, "loss": 3.0288, "step": 13555 }, { "epoch": 0.8415171643180831, "grad_norm": 0.5131774088943282, "learning_rate": 9.039941617187383e-05, "loss": 3.017, "step": 13556 }, { "epoch": 0.8415792414178409, "grad_norm": 0.538756656959633, "learning_rate": 9.039728816186925e-05, "loss": 2.9863, "step": 13557 }, { "epoch": 0.8416413185175988, "grad_norm": 0.5484768160560032, "learning_rate": 9.039515994110221e-05, "loss": 2.9906, "step": 13558 }, { "epoch": 0.8417033956173567, "grad_norm": 0.469794540086348, "learning_rate": 9.039303150958383e-05, "loss": 2.9781, "step": 13559 }, { "epoch": 0.8417654727171147, "grad_norm": 0.5946698091233977, "learning_rate": 9.039090286732518e-05, "loss": 3.0884, "step": 13560 }, { "epoch": 0.8418275498168726, "grad_norm": 0.5183241452029328, "learning_rate": 9.038877401433738e-05, "loss": 2.8936, "step": 13561 }, { "epoch": 0.8418896269166305, "grad_norm": 0.5696408222638765, "learning_rate": 9.038664495063155e-05, "loss": 2.9744, "step": 13562 }, { "epoch": 0.8419517040163883, "grad_norm": 0.49544336834402813, "learning_rate": 9.03845156762188e-05, "loss": 2.9697, "step": 13563 }, { "epoch": 0.8420137811161462, "grad_norm": 0.4622229094027695, "learning_rate": 9.038238619111022e-05, "loss": 2.9063, "step": 13564 }, { "epoch": 0.8420758582159041, "grad_norm": 0.524220464621183, "learning_rate": 9.038025649531693e-05, "loss": 2.9333, "step": 13565 }, { "epoch": 0.8421379353156621, "grad_norm": 0.520743483104758, "learning_rate": 9.037812658885003e-05, "loss": 3.0453, "step": 13566 }, { "epoch": 0.84220001241542, "grad_norm": 0.5376968500161766, "learning_rate": 9.037599647172066e-05, "loss": 3.0269, "step": 13567 }, { "epoch": 0.8422620895151779, "grad_norm": 0.48389592628687783, "learning_rate": 9.03738661439399e-05, "loss": 3.0657, "step": 13568 }, { "epoch": 0.8423241666149357, "grad_norm": 0.5432503333742441, "learning_rate": 9.037173560551889e-05, "loss": 2.9297, "step": 13569 }, { "epoch": 0.8423862437146936, "grad_norm": 0.5833827974146967, "learning_rate": 9.036960485646872e-05, "loss": 3.0496, "step": 13570 }, { "epoch": 0.8424483208144515, "grad_norm": 0.4807076829247501, "learning_rate": 9.036747389680053e-05, "loss": 2.9244, "step": 13571 }, { "epoch": 0.8425103979142095, "grad_norm": 0.5010167453407742, "learning_rate": 9.036534272652542e-05, "loss": 3.0593, "step": 13572 }, { "epoch": 0.8425724750139674, "grad_norm": 0.5645046804570258, "learning_rate": 9.036321134565453e-05, "loss": 2.9333, "step": 13573 }, { "epoch": 0.8426345521137253, "grad_norm": 0.5084446713458226, "learning_rate": 9.036107975419896e-05, "loss": 2.9578, "step": 13574 }, { "epoch": 0.8426966292134831, "grad_norm": 0.5135511249138084, "learning_rate": 9.035894795216984e-05, "loss": 3.0016, "step": 13575 }, { "epoch": 0.842758706313241, "grad_norm": 0.4961344634643506, "learning_rate": 9.03568159395783e-05, "loss": 3.0861, "step": 13576 }, { "epoch": 0.8428207834129989, "grad_norm": 0.5287216269994676, "learning_rate": 9.035468371643545e-05, "loss": 2.922, "step": 13577 }, { "epoch": 0.8428828605127568, "grad_norm": 0.49081793472193763, "learning_rate": 9.03525512827524e-05, "loss": 2.9718, "step": 13578 }, { "epoch": 0.8429449376125148, "grad_norm": 0.6073623333662376, "learning_rate": 9.035041863854032e-05, "loss": 3.0331, "step": 13579 }, { "epoch": 0.8430070147122727, "grad_norm": 0.5642689335071435, "learning_rate": 9.03482857838103e-05, "loss": 3.0203, "step": 13580 }, { "epoch": 0.8430690918120305, "grad_norm": 0.5563758621980478, "learning_rate": 9.034615271857347e-05, "loss": 3.0341, "step": 13581 }, { "epoch": 0.8431311689117884, "grad_norm": 0.59533447365244, "learning_rate": 9.034401944284099e-05, "loss": 3.0302, "step": 13582 }, { "epoch": 0.8431932460115463, "grad_norm": 0.5463175793795572, "learning_rate": 9.034188595662394e-05, "loss": 2.9938, "step": 13583 }, { "epoch": 0.8432553231113042, "grad_norm": 0.5519986057643974, "learning_rate": 9.03397522599335e-05, "loss": 2.9772, "step": 13584 }, { "epoch": 0.8433174002110622, "grad_norm": 0.5129766945480809, "learning_rate": 9.033761835278076e-05, "loss": 2.9337, "step": 13585 }, { "epoch": 0.8433794773108201, "grad_norm": 0.5529298200913692, "learning_rate": 9.033548423517688e-05, "loss": 3.033, "step": 13586 }, { "epoch": 0.8434415544105779, "grad_norm": 0.5036984096358099, "learning_rate": 9.0333349907133e-05, "loss": 3.02, "step": 13587 }, { "epoch": 0.8435036315103358, "grad_norm": 0.5422882920828347, "learning_rate": 9.033121536866023e-05, "loss": 3.0413, "step": 13588 }, { "epoch": 0.8435657086100937, "grad_norm": 0.5362019034026417, "learning_rate": 9.032908061976973e-05, "loss": 3.0067, "step": 13589 }, { "epoch": 0.8436277857098516, "grad_norm": 0.5230841511554938, "learning_rate": 9.032694566047259e-05, "loss": 2.9236, "step": 13590 }, { "epoch": 0.8436898628096096, "grad_norm": 0.43703696748050636, "learning_rate": 9.032481049078004e-05, "loss": 3.0259, "step": 13591 }, { "epoch": 0.8437519399093675, "grad_norm": 0.7255049772006403, "learning_rate": 9.032267511070312e-05, "loss": 2.935, "step": 13592 }, { "epoch": 0.8438140170091253, "grad_norm": 0.7081916517789616, "learning_rate": 9.032053952025303e-05, "loss": 2.9942, "step": 13593 }, { "epoch": 0.8438760941088832, "grad_norm": 0.5655659987305276, "learning_rate": 9.03184037194409e-05, "loss": 2.9262, "step": 13594 }, { "epoch": 0.8439381712086411, "grad_norm": 0.5188276646397632, "learning_rate": 9.031626770827786e-05, "loss": 3.0453, "step": 13595 }, { "epoch": 0.844000248308399, "grad_norm": 0.5988122319590282, "learning_rate": 9.031413148677508e-05, "loss": 3.008, "step": 13596 }, { "epoch": 0.844062325408157, "grad_norm": 0.5709762629821163, "learning_rate": 9.031199505494368e-05, "loss": 2.9638, "step": 13597 }, { "epoch": 0.8441244025079149, "grad_norm": 0.612848748939733, "learning_rate": 9.030985841279482e-05, "loss": 3.057, "step": 13598 }, { "epoch": 0.8441864796076727, "grad_norm": 0.5974818953959423, "learning_rate": 9.030772156033964e-05, "loss": 3.002, "step": 13599 }, { "epoch": 0.8442485567074306, "grad_norm": 0.6676127282510923, "learning_rate": 9.03055844975893e-05, "loss": 3.032, "step": 13600 }, { "epoch": 0.8443106338071885, "grad_norm": 0.6165819215108326, "learning_rate": 9.030344722455492e-05, "loss": 2.9528, "step": 13601 }, { "epoch": 0.8443727109069464, "grad_norm": 0.559143782624468, "learning_rate": 9.030130974124768e-05, "loss": 2.9536, "step": 13602 }, { "epoch": 0.8444347880067044, "grad_norm": 0.548338593696399, "learning_rate": 9.029917204767875e-05, "loss": 2.9449, "step": 13603 }, { "epoch": 0.8444968651064623, "grad_norm": 0.528867709447478, "learning_rate": 9.029703414385924e-05, "loss": 3.0283, "step": 13604 }, { "epoch": 0.8445589422062201, "grad_norm": 0.533906465775399, "learning_rate": 9.02948960298003e-05, "loss": 2.9938, "step": 13605 }, { "epoch": 0.844621019305978, "grad_norm": 0.5300809676024555, "learning_rate": 9.029275770551313e-05, "loss": 3.0403, "step": 13606 }, { "epoch": 0.8446830964057359, "grad_norm": 0.522331663876811, "learning_rate": 9.029061917100885e-05, "loss": 2.9325, "step": 13607 }, { "epoch": 0.8447451735054938, "grad_norm": 0.5214690016862901, "learning_rate": 9.028848042629862e-05, "loss": 3.0204, "step": 13608 }, { "epoch": 0.8448072506052517, "grad_norm": 0.5163867987173834, "learning_rate": 9.028634147139364e-05, "loss": 2.9459, "step": 13609 }, { "epoch": 0.8448693277050097, "grad_norm": 0.5114759203909958, "learning_rate": 9.028420230630501e-05, "loss": 2.9661, "step": 13610 }, { "epoch": 0.8449314048047675, "grad_norm": 0.557218258619025, "learning_rate": 9.028206293104392e-05, "loss": 3.033, "step": 13611 }, { "epoch": 0.8449934819045254, "grad_norm": 0.5735078420294113, "learning_rate": 9.027992334562153e-05, "loss": 2.9294, "step": 13612 }, { "epoch": 0.8450555590042833, "grad_norm": 0.6026185167253082, "learning_rate": 9.027778355004901e-05, "loss": 3.0408, "step": 13613 }, { "epoch": 0.8451176361040412, "grad_norm": 0.47721502313388275, "learning_rate": 9.02756435443375e-05, "loss": 2.938, "step": 13614 }, { "epoch": 0.8451797132037991, "grad_norm": 0.5652961111988006, "learning_rate": 9.02735033284982e-05, "loss": 2.98, "step": 13615 }, { "epoch": 0.8452417903035571, "grad_norm": 0.5004753594940143, "learning_rate": 9.027136290254224e-05, "loss": 3.0677, "step": 13616 }, { "epoch": 0.8453038674033149, "grad_norm": 0.43233708254157355, "learning_rate": 9.02692222664808e-05, "loss": 2.8972, "step": 13617 }, { "epoch": 0.8453659445030728, "grad_norm": 0.496268624549574, "learning_rate": 9.026708142032507e-05, "loss": 2.9892, "step": 13618 }, { "epoch": 0.8454280216028307, "grad_norm": 0.4412982437193063, "learning_rate": 9.026494036408619e-05, "loss": 2.9344, "step": 13619 }, { "epoch": 0.8454900987025886, "grad_norm": 0.5146453394063683, "learning_rate": 9.026279909777535e-05, "loss": 3.0247, "step": 13620 }, { "epoch": 0.8455521758023465, "grad_norm": 0.46654465377413257, "learning_rate": 9.02606576214037e-05, "loss": 2.9821, "step": 13621 }, { "epoch": 0.8456142529021045, "grad_norm": 0.5072824499505798, "learning_rate": 9.025851593498243e-05, "loss": 3.0036, "step": 13622 }, { "epoch": 0.8456763300018623, "grad_norm": 0.45859688168029905, "learning_rate": 9.025637403852271e-05, "loss": 2.985, "step": 13623 }, { "epoch": 0.8457384071016202, "grad_norm": 0.46415129945452493, "learning_rate": 9.025423193203573e-05, "loss": 2.9997, "step": 13624 }, { "epoch": 0.8458004842013781, "grad_norm": 0.4701501756290269, "learning_rate": 9.025208961553262e-05, "loss": 3.0119, "step": 13625 }, { "epoch": 0.845862561301136, "grad_norm": 0.5055228922175983, "learning_rate": 9.02499470890246e-05, "loss": 3.0257, "step": 13626 }, { "epoch": 0.8459246384008939, "grad_norm": 0.45520193028547745, "learning_rate": 9.024780435252285e-05, "loss": 2.977, "step": 13627 }, { "epoch": 0.8459867155006519, "grad_norm": 0.4745537368124074, "learning_rate": 9.024566140603851e-05, "loss": 3.0729, "step": 13628 }, { "epoch": 0.8460487926004097, "grad_norm": 0.4268831640722297, "learning_rate": 9.02435182495828e-05, "loss": 2.9585, "step": 13629 }, { "epoch": 0.8461108697001676, "grad_norm": 0.48493582037885463, "learning_rate": 9.024137488316689e-05, "loss": 3.0514, "step": 13630 }, { "epoch": 0.8461729467999255, "grad_norm": 0.4870909717206722, "learning_rate": 9.023923130680195e-05, "loss": 3.0603, "step": 13631 }, { "epoch": 0.8462350238996834, "grad_norm": 0.5072094197216833, "learning_rate": 9.023708752049917e-05, "loss": 3.034, "step": 13632 }, { "epoch": 0.8462971009994413, "grad_norm": 0.4535174150456235, "learning_rate": 9.023494352426973e-05, "loss": 2.9909, "step": 13633 }, { "epoch": 0.8463591780991992, "grad_norm": 0.5602104160822915, "learning_rate": 9.023279931812484e-05, "loss": 2.9779, "step": 13634 }, { "epoch": 0.8464212551989571, "grad_norm": 0.46260979740959607, "learning_rate": 9.023065490207565e-05, "loss": 2.9343, "step": 13635 }, { "epoch": 0.846483332298715, "grad_norm": 0.5268786604676072, "learning_rate": 9.022851027613338e-05, "loss": 2.9832, "step": 13636 }, { "epoch": 0.8465454093984729, "grad_norm": 0.5012424648222726, "learning_rate": 9.02263654403092e-05, "loss": 3.0187, "step": 13637 }, { "epoch": 0.8466074864982308, "grad_norm": 0.4986727801994082, "learning_rate": 9.022422039461431e-05, "loss": 3.0376, "step": 13638 }, { "epoch": 0.8466695635979887, "grad_norm": 0.5008194585973942, "learning_rate": 9.022207513905991e-05, "loss": 2.9308, "step": 13639 }, { "epoch": 0.8467316406977466, "grad_norm": 0.45642912562164173, "learning_rate": 9.021992967365716e-05, "loss": 2.9697, "step": 13640 }, { "epoch": 0.8467937177975045, "grad_norm": 0.4811309809710674, "learning_rate": 9.021778399841729e-05, "loss": 2.9694, "step": 13641 }, { "epoch": 0.8468557948972624, "grad_norm": 0.5055355375760935, "learning_rate": 9.021563811335145e-05, "loss": 3.0609, "step": 13642 }, { "epoch": 0.8469178719970203, "grad_norm": 0.4464992637521413, "learning_rate": 9.021349201847089e-05, "loss": 2.9429, "step": 13643 }, { "epoch": 0.8469799490967782, "grad_norm": 0.5466915367834827, "learning_rate": 9.021134571378677e-05, "loss": 2.998, "step": 13644 }, { "epoch": 0.8470420261965361, "grad_norm": 0.546439706041564, "learning_rate": 9.02091991993103e-05, "loss": 2.9842, "step": 13645 }, { "epoch": 0.847104103296294, "grad_norm": 0.5475156301505653, "learning_rate": 9.020705247505268e-05, "loss": 2.9945, "step": 13646 }, { "epoch": 0.8471661803960518, "grad_norm": 0.4798738316528931, "learning_rate": 9.020490554102509e-05, "loss": 2.9156, "step": 13647 }, { "epoch": 0.8472282574958098, "grad_norm": 0.487079591307068, "learning_rate": 9.020275839723877e-05, "loss": 2.9947, "step": 13648 }, { "epoch": 0.8472903345955677, "grad_norm": 0.5483641648779681, "learning_rate": 9.020061104370488e-05, "loss": 2.9881, "step": 13649 }, { "epoch": 0.8473524116953256, "grad_norm": 0.45692675773527547, "learning_rate": 9.019846348043466e-05, "loss": 2.9833, "step": 13650 }, { "epoch": 0.8474144887950835, "grad_norm": 0.553427460513177, "learning_rate": 9.019631570743928e-05, "loss": 3.0224, "step": 13651 }, { "epoch": 0.8474765658948414, "grad_norm": 0.5382394173676329, "learning_rate": 9.019416772472998e-05, "loss": 2.9677, "step": 13652 }, { "epoch": 0.8475386429945992, "grad_norm": 0.4914893845675932, "learning_rate": 9.019201953231794e-05, "loss": 2.9661, "step": 13653 }, { "epoch": 0.8476007200943572, "grad_norm": 0.5530779953289443, "learning_rate": 9.018987113021438e-05, "loss": 3.0664, "step": 13654 }, { "epoch": 0.8476627971941151, "grad_norm": 0.5216856721558341, "learning_rate": 9.018772251843051e-05, "loss": 3.0466, "step": 13655 }, { "epoch": 0.847724874293873, "grad_norm": 0.5252056150606904, "learning_rate": 9.018557369697753e-05, "loss": 3.0171, "step": 13656 }, { "epoch": 0.8477869513936309, "grad_norm": 0.545567803066813, "learning_rate": 9.018342466586665e-05, "loss": 3.0419, "step": 13657 }, { "epoch": 0.8478490284933888, "grad_norm": 0.49304861095922553, "learning_rate": 9.018127542510911e-05, "loss": 2.924, "step": 13658 }, { "epoch": 0.8479111055931466, "grad_norm": 0.5354288339778526, "learning_rate": 9.017912597471607e-05, "loss": 3.0173, "step": 13659 }, { "epoch": 0.8479731826929046, "grad_norm": 0.4983525254573769, "learning_rate": 9.01769763146988e-05, "loss": 2.966, "step": 13660 }, { "epoch": 0.8480352597926625, "grad_norm": 0.5227875731295192, "learning_rate": 9.017482644506849e-05, "loss": 3.0321, "step": 13661 }, { "epoch": 0.8480973368924204, "grad_norm": 0.5019597450146357, "learning_rate": 9.017267636583635e-05, "loss": 3.1, "step": 13662 }, { "epoch": 0.8481594139921783, "grad_norm": 0.4797801689325101, "learning_rate": 9.01705260770136e-05, "loss": 2.9958, "step": 13663 }, { "epoch": 0.8482214910919362, "grad_norm": 0.48600405884252706, "learning_rate": 9.016837557861149e-05, "loss": 2.9989, "step": 13664 }, { "epoch": 0.848283568191694, "grad_norm": 0.5009595187673234, "learning_rate": 9.01662248706412e-05, "loss": 2.8933, "step": 13665 }, { "epoch": 0.848345645291452, "grad_norm": 0.5076180012231508, "learning_rate": 9.016407395311395e-05, "loss": 3.005, "step": 13666 }, { "epoch": 0.8484077223912099, "grad_norm": 0.5191594277586995, "learning_rate": 9.016192282604097e-05, "loss": 3.0233, "step": 13667 }, { "epoch": 0.8484697994909678, "grad_norm": 0.5881351203331485, "learning_rate": 9.015977148943351e-05, "loss": 3.1209, "step": 13668 }, { "epoch": 0.8485318765907257, "grad_norm": 0.5145300596926, "learning_rate": 9.015761994330277e-05, "loss": 3.0629, "step": 13669 }, { "epoch": 0.8485939536904836, "grad_norm": 0.5062890014224578, "learning_rate": 9.015546818765998e-05, "loss": 3.0297, "step": 13670 }, { "epoch": 0.8486560307902414, "grad_norm": 0.488122044357749, "learning_rate": 9.015331622251635e-05, "loss": 2.9328, "step": 13671 }, { "epoch": 0.8487181078899994, "grad_norm": 0.4674104464786911, "learning_rate": 9.015116404788313e-05, "loss": 3.0613, "step": 13672 }, { "epoch": 0.8487801849897573, "grad_norm": 0.5447588440858603, "learning_rate": 9.014901166377156e-05, "loss": 3.0705, "step": 13673 }, { "epoch": 0.8488422620895152, "grad_norm": 0.57752441449597, "learning_rate": 9.014685907019282e-05, "loss": 3.0076, "step": 13674 }, { "epoch": 0.8489043391892731, "grad_norm": 0.5061626924181433, "learning_rate": 9.014470626715818e-05, "loss": 3.0273, "step": 13675 }, { "epoch": 0.848966416289031, "grad_norm": 0.5207780490780307, "learning_rate": 9.014255325467886e-05, "loss": 2.951, "step": 13676 }, { "epoch": 0.8490284933887888, "grad_norm": 0.5273499947637339, "learning_rate": 9.014040003276611e-05, "loss": 2.9927, "step": 13677 }, { "epoch": 0.8490905704885467, "grad_norm": 0.4881542651040432, "learning_rate": 9.013824660143113e-05, "loss": 2.942, "step": 13678 }, { "epoch": 0.8491526475883047, "grad_norm": 0.5361288101797331, "learning_rate": 9.013609296068519e-05, "loss": 2.983, "step": 13679 }, { "epoch": 0.8492147246880626, "grad_norm": 0.6311396870678729, "learning_rate": 9.01339391105395e-05, "loss": 3.0074, "step": 13680 }, { "epoch": 0.8492768017878205, "grad_norm": 0.5810532000867255, "learning_rate": 9.01317850510053e-05, "loss": 2.9541, "step": 13681 }, { "epoch": 0.8493388788875784, "grad_norm": 0.4971819829005301, "learning_rate": 9.012963078209384e-05, "loss": 2.9354, "step": 13682 }, { "epoch": 0.8494009559873362, "grad_norm": 0.5688778062942748, "learning_rate": 9.012747630381636e-05, "loss": 3.052, "step": 13683 }, { "epoch": 0.8494630330870941, "grad_norm": 0.504825168174376, "learning_rate": 9.01253216161841e-05, "loss": 2.9756, "step": 13684 }, { "epoch": 0.8495251101868521, "grad_norm": 0.5320829705237354, "learning_rate": 9.012316671920828e-05, "loss": 3.0176, "step": 13685 }, { "epoch": 0.84958718728661, "grad_norm": 0.5007320052525998, "learning_rate": 9.012101161290016e-05, "loss": 2.9948, "step": 13686 }, { "epoch": 0.8496492643863679, "grad_norm": 0.48471921564910697, "learning_rate": 9.0118856297271e-05, "loss": 3.0823, "step": 13687 }, { "epoch": 0.8497113414861258, "grad_norm": 0.5720857982161012, "learning_rate": 9.011670077233201e-05, "loss": 2.9817, "step": 13688 }, { "epoch": 0.8497734185858836, "grad_norm": 0.48019880759775524, "learning_rate": 9.011454503809446e-05, "loss": 2.9947, "step": 13689 }, { "epoch": 0.8498354956856415, "grad_norm": 0.5281568686897704, "learning_rate": 9.011238909456959e-05, "loss": 3.0198, "step": 13690 }, { "epoch": 0.8498975727853995, "grad_norm": 0.5511542065314288, "learning_rate": 9.011023294176866e-05, "loss": 2.9707, "step": 13691 }, { "epoch": 0.8499596498851574, "grad_norm": 0.5175789599141998, "learning_rate": 9.010807657970288e-05, "loss": 3.103, "step": 13692 }, { "epoch": 0.8500217269849153, "grad_norm": 0.5287461177872789, "learning_rate": 9.010592000838355e-05, "loss": 2.9473, "step": 13693 }, { "epoch": 0.8500838040846732, "grad_norm": 0.5411831152381813, "learning_rate": 9.01037632278219e-05, "loss": 3.0349, "step": 13694 }, { "epoch": 0.850145881184431, "grad_norm": 0.4590690210085696, "learning_rate": 9.010160623802916e-05, "loss": 3.0555, "step": 13695 }, { "epoch": 0.8502079582841889, "grad_norm": 0.49962391716594234, "learning_rate": 9.009944903901664e-05, "loss": 2.9607, "step": 13696 }, { "epoch": 0.8502700353839469, "grad_norm": 0.46093633619258534, "learning_rate": 9.009729163079554e-05, "loss": 2.8851, "step": 13697 }, { "epoch": 0.8503321124837048, "grad_norm": 0.5013828823484769, "learning_rate": 9.009513401337712e-05, "loss": 3.0468, "step": 13698 }, { "epoch": 0.8503941895834627, "grad_norm": 0.4312419862970115, "learning_rate": 9.009297618677268e-05, "loss": 3.0028, "step": 13699 }, { "epoch": 0.8504562666832206, "grad_norm": 0.4621335207901943, "learning_rate": 9.009081815099345e-05, "loss": 3.0431, "step": 13700 }, { "epoch": 0.8505183437829784, "grad_norm": 0.4803401806014832, "learning_rate": 9.008865990605067e-05, "loss": 3.0143, "step": 13701 }, { "epoch": 0.8505804208827363, "grad_norm": 0.4330211329357783, "learning_rate": 9.008650145195563e-05, "loss": 3.0199, "step": 13702 }, { "epoch": 0.8506424979824943, "grad_norm": 0.4622428159645483, "learning_rate": 9.008434278871958e-05, "loss": 2.9187, "step": 13703 }, { "epoch": 0.8507045750822522, "grad_norm": 0.4687966482475146, "learning_rate": 9.008218391635377e-05, "loss": 3.0294, "step": 13704 }, { "epoch": 0.8507666521820101, "grad_norm": 0.4620621748202921, "learning_rate": 9.008002483486949e-05, "loss": 2.9577, "step": 13705 }, { "epoch": 0.850828729281768, "grad_norm": 0.4679922177603049, "learning_rate": 9.0077865544278e-05, "loss": 2.9347, "step": 13706 }, { "epoch": 0.8508908063815258, "grad_norm": 0.4406438505590942, "learning_rate": 9.007570604459053e-05, "loss": 2.9658, "step": 13707 }, { "epoch": 0.8509528834812837, "grad_norm": 0.46051570214144805, "learning_rate": 9.00735463358184e-05, "loss": 2.9839, "step": 13708 }, { "epoch": 0.8510149605810416, "grad_norm": 0.48415974479373736, "learning_rate": 9.007138641797284e-05, "loss": 3.0266, "step": 13709 }, { "epoch": 0.8510770376807996, "grad_norm": 0.505675146070578, "learning_rate": 9.006922629106512e-05, "loss": 2.884, "step": 13710 }, { "epoch": 0.8511391147805575, "grad_norm": 0.49911948994748506, "learning_rate": 9.006706595510652e-05, "loss": 2.9665, "step": 13711 }, { "epoch": 0.8512011918803154, "grad_norm": 0.47261364011775575, "learning_rate": 9.006490541010832e-05, "loss": 3.0236, "step": 13712 }, { "epoch": 0.8512632689800732, "grad_norm": 0.5163669450030518, "learning_rate": 9.006274465608177e-05, "loss": 2.999, "step": 13713 }, { "epoch": 0.8513253460798311, "grad_norm": 0.4631757381254813, "learning_rate": 9.006058369303819e-05, "loss": 2.9832, "step": 13714 }, { "epoch": 0.851387423179589, "grad_norm": 0.4906388851800774, "learning_rate": 9.005842252098877e-05, "loss": 2.9894, "step": 13715 }, { "epoch": 0.851449500279347, "grad_norm": 0.504704043468576, "learning_rate": 9.005626113994488e-05, "loss": 2.9819, "step": 13716 }, { "epoch": 0.8515115773791049, "grad_norm": 0.48411271469579115, "learning_rate": 9.005409954991772e-05, "loss": 2.983, "step": 13717 }, { "epoch": 0.8515736544788628, "grad_norm": 0.52880614219372, "learning_rate": 9.00519377509186e-05, "loss": 2.9494, "step": 13718 }, { "epoch": 0.8516357315786206, "grad_norm": 0.5539016260726182, "learning_rate": 9.004977574295882e-05, "loss": 3.0766, "step": 13719 }, { "epoch": 0.8516978086783785, "grad_norm": 0.5465314876161738, "learning_rate": 9.004761352604963e-05, "loss": 3.0538, "step": 13720 }, { "epoch": 0.8517598857781364, "grad_norm": 0.5152793650618398, "learning_rate": 9.00454511002023e-05, "loss": 2.9687, "step": 13721 }, { "epoch": 0.8518219628778944, "grad_norm": 0.5516556366694647, "learning_rate": 9.004328846542816e-05, "loss": 3.0801, "step": 13722 }, { "epoch": 0.8518840399776523, "grad_norm": 0.5027582561568082, "learning_rate": 9.004112562173845e-05, "loss": 3.0815, "step": 13723 }, { "epoch": 0.8519461170774102, "grad_norm": 0.4756877093136109, "learning_rate": 9.003896256914446e-05, "loss": 3.0157, "step": 13724 }, { "epoch": 0.852008194177168, "grad_norm": 0.5480435013072446, "learning_rate": 9.003679930765749e-05, "loss": 2.9787, "step": 13725 }, { "epoch": 0.8520702712769259, "grad_norm": 0.4529541778852244, "learning_rate": 9.003463583728881e-05, "loss": 2.8791, "step": 13726 }, { "epoch": 0.8521323483766838, "grad_norm": 0.518000362116271, "learning_rate": 9.003247215804971e-05, "loss": 3.0754, "step": 13727 }, { "epoch": 0.8521944254764418, "grad_norm": 0.5395055415317014, "learning_rate": 9.003030826995151e-05, "loss": 2.919, "step": 13728 }, { "epoch": 0.8522565025761997, "grad_norm": 0.497262179474382, "learning_rate": 9.002814417300547e-05, "loss": 2.98, "step": 13729 }, { "epoch": 0.8523185796759576, "grad_norm": 0.5990303797859073, "learning_rate": 9.002597986722286e-05, "loss": 2.8934, "step": 13730 }, { "epoch": 0.8523806567757154, "grad_norm": 0.44122905043885297, "learning_rate": 9.002381535261502e-05, "loss": 2.8651, "step": 13731 }, { "epoch": 0.8524427338754733, "grad_norm": 0.5292634284794001, "learning_rate": 9.00216506291932e-05, "loss": 2.9831, "step": 13732 }, { "epoch": 0.8525048109752312, "grad_norm": 0.5098207524333075, "learning_rate": 9.001948569696873e-05, "loss": 2.9504, "step": 13733 }, { "epoch": 0.8525668880749891, "grad_norm": 0.5460075856177914, "learning_rate": 9.001732055595289e-05, "loss": 2.9831, "step": 13734 }, { "epoch": 0.8526289651747471, "grad_norm": 0.5251126883683714, "learning_rate": 9.001515520615695e-05, "loss": 3.0551, "step": 13735 }, { "epoch": 0.852691042274505, "grad_norm": 0.46893904370777445, "learning_rate": 9.001298964759226e-05, "loss": 3.0058, "step": 13736 }, { "epoch": 0.8527531193742628, "grad_norm": 0.47995355167761766, "learning_rate": 9.001082388027008e-05, "loss": 2.8696, "step": 13737 }, { "epoch": 0.8528151964740207, "grad_norm": 0.48089811477359035, "learning_rate": 9.000865790420171e-05, "loss": 3.0122, "step": 13738 }, { "epoch": 0.8528772735737786, "grad_norm": 0.43748847060679413, "learning_rate": 9.000649171939847e-05, "loss": 2.9994, "step": 13739 }, { "epoch": 0.8529393506735365, "grad_norm": 0.5156589843647912, "learning_rate": 9.000432532587164e-05, "loss": 2.9771, "step": 13740 }, { "epoch": 0.8530014277732945, "grad_norm": 0.45463889540612623, "learning_rate": 9.000215872363254e-05, "loss": 2.998, "step": 13741 }, { "epoch": 0.8530635048730524, "grad_norm": 0.5263525796253953, "learning_rate": 8.999999191269248e-05, "loss": 2.8675, "step": 13742 }, { "epoch": 0.8531255819728102, "grad_norm": 0.5348277789402346, "learning_rate": 8.999782489306272e-05, "loss": 2.9767, "step": 13743 }, { "epoch": 0.8531876590725681, "grad_norm": 0.498588536736452, "learning_rate": 8.999565766475461e-05, "loss": 2.9608, "step": 13744 }, { "epoch": 0.853249736172326, "grad_norm": 0.4722010256725435, "learning_rate": 8.999349022777946e-05, "loss": 3.0331, "step": 13745 }, { "epoch": 0.8533118132720839, "grad_norm": 0.46680798377700927, "learning_rate": 8.999132258214855e-05, "loss": 3.0323, "step": 13746 }, { "epoch": 0.8533738903718419, "grad_norm": 0.4814523228256807, "learning_rate": 8.998915472787321e-05, "loss": 2.9943, "step": 13747 }, { "epoch": 0.8534359674715998, "grad_norm": 0.49044185974256177, "learning_rate": 8.998698666496474e-05, "loss": 3.0174, "step": 13748 }, { "epoch": 0.8534980445713576, "grad_norm": 0.4644752773074449, "learning_rate": 8.998481839343444e-05, "loss": 3.0602, "step": 13749 }, { "epoch": 0.8535601216711155, "grad_norm": 0.5593266514177069, "learning_rate": 8.998264991329365e-05, "loss": 2.9807, "step": 13750 }, { "epoch": 0.8536221987708734, "grad_norm": 0.5240149373244128, "learning_rate": 8.998048122455367e-05, "loss": 2.9399, "step": 13751 }, { "epoch": 0.8536842758706313, "grad_norm": 0.47556198341781747, "learning_rate": 8.997831232722582e-05, "loss": 3.0353, "step": 13752 }, { "epoch": 0.8537463529703893, "grad_norm": 0.5467336744531998, "learning_rate": 8.997614322132139e-05, "loss": 3.0465, "step": 13753 }, { "epoch": 0.8538084300701472, "grad_norm": 0.4793279176093504, "learning_rate": 8.997397390685174e-05, "loss": 3.0389, "step": 13754 }, { "epoch": 0.853870507169905, "grad_norm": 0.49096735836169564, "learning_rate": 8.997180438382814e-05, "loss": 2.9659, "step": 13755 }, { "epoch": 0.8539325842696629, "grad_norm": 0.48177805886563696, "learning_rate": 8.996963465226194e-05, "loss": 2.9345, "step": 13756 }, { "epoch": 0.8539946613694208, "grad_norm": 0.5186366934148771, "learning_rate": 8.996746471216446e-05, "loss": 3.0224, "step": 13757 }, { "epoch": 0.8540567384691787, "grad_norm": 0.5100683431553606, "learning_rate": 8.996529456354701e-05, "loss": 2.935, "step": 13758 }, { "epoch": 0.8541188155689367, "grad_norm": 0.47612661220110336, "learning_rate": 8.996312420642091e-05, "loss": 3.0628, "step": 13759 }, { "epoch": 0.8541808926686946, "grad_norm": 0.5631951698845225, "learning_rate": 8.996095364079751e-05, "loss": 2.9795, "step": 13760 }, { "epoch": 0.8542429697684524, "grad_norm": 0.5592584723034197, "learning_rate": 8.995878286668808e-05, "loss": 3.0016, "step": 13761 }, { "epoch": 0.8543050468682103, "grad_norm": 0.5374109704560696, "learning_rate": 8.9956611884104e-05, "loss": 2.9644, "step": 13762 }, { "epoch": 0.8543671239679682, "grad_norm": 0.406934491626398, "learning_rate": 8.995444069305659e-05, "loss": 3.0283, "step": 13763 }, { "epoch": 0.8544292010677261, "grad_norm": 0.5087098336356102, "learning_rate": 8.995226929355715e-05, "loss": 2.9424, "step": 13764 }, { "epoch": 0.854491278167484, "grad_norm": 0.42910954880781216, "learning_rate": 8.995009768561702e-05, "loss": 2.9155, "step": 13765 }, { "epoch": 0.854553355267242, "grad_norm": 0.4541540792307418, "learning_rate": 8.994792586924753e-05, "loss": 2.9882, "step": 13766 }, { "epoch": 0.8546154323669998, "grad_norm": 0.479344284958581, "learning_rate": 8.994575384446002e-05, "loss": 2.8772, "step": 13767 }, { "epoch": 0.8546775094667577, "grad_norm": 0.5148346942110971, "learning_rate": 8.994358161126581e-05, "loss": 3.0559, "step": 13768 }, { "epoch": 0.8547395865665156, "grad_norm": 0.48727225767180765, "learning_rate": 8.994140916967623e-05, "loss": 2.937, "step": 13769 }, { "epoch": 0.8548016636662735, "grad_norm": 0.48037484654749957, "learning_rate": 8.993923651970263e-05, "loss": 3.0186, "step": 13770 }, { "epoch": 0.8548637407660314, "grad_norm": 0.4816417332100475, "learning_rate": 8.993706366135634e-05, "loss": 3.0074, "step": 13771 }, { "epoch": 0.8549258178657894, "grad_norm": 0.45883505807058905, "learning_rate": 8.99348905946487e-05, "loss": 3.0079, "step": 13772 }, { "epoch": 0.8549878949655472, "grad_norm": 0.5361642741437581, "learning_rate": 8.993271731959103e-05, "loss": 2.9869, "step": 13773 }, { "epoch": 0.8550499720653051, "grad_norm": 0.49131556888694794, "learning_rate": 8.993054383619469e-05, "loss": 3.0012, "step": 13774 }, { "epoch": 0.855112049165063, "grad_norm": 0.4408614924053995, "learning_rate": 8.992837014447099e-05, "loss": 3.0507, "step": 13775 }, { "epoch": 0.8551741262648209, "grad_norm": 0.4875302299272199, "learning_rate": 8.992619624443132e-05, "loss": 2.9928, "step": 13776 }, { "epoch": 0.8552362033645788, "grad_norm": 0.5079712252877789, "learning_rate": 8.992402213608698e-05, "loss": 3.0073, "step": 13777 }, { "epoch": 0.8552982804643368, "grad_norm": 0.4825942468999424, "learning_rate": 8.992184781944931e-05, "loss": 3.0209, "step": 13778 }, { "epoch": 0.8553603575640946, "grad_norm": 0.5168698340807036, "learning_rate": 8.991967329452969e-05, "loss": 3.0308, "step": 13779 }, { "epoch": 0.8554224346638525, "grad_norm": 0.4230674738637377, "learning_rate": 8.991749856133942e-05, "loss": 3.037, "step": 13780 }, { "epoch": 0.8554845117636104, "grad_norm": 0.4952705491889852, "learning_rate": 8.991532361988987e-05, "loss": 3.0733, "step": 13781 }, { "epoch": 0.8555465888633683, "grad_norm": 0.5101594244257961, "learning_rate": 8.99131484701924e-05, "loss": 2.931, "step": 13782 }, { "epoch": 0.8556086659631262, "grad_norm": 0.465563593405972, "learning_rate": 8.991097311225836e-05, "loss": 3.0112, "step": 13783 }, { "epoch": 0.8556707430628842, "grad_norm": 0.5048193236817052, "learning_rate": 8.990879754609906e-05, "loss": 2.9987, "step": 13784 }, { "epoch": 0.855732820162642, "grad_norm": 0.4742516319274977, "learning_rate": 8.990662177172588e-05, "loss": 3.0894, "step": 13785 }, { "epoch": 0.8557948972623999, "grad_norm": 0.4611388429459875, "learning_rate": 8.990444578915018e-05, "loss": 2.9501, "step": 13786 }, { "epoch": 0.8558569743621578, "grad_norm": 0.4731498694340166, "learning_rate": 8.990226959838328e-05, "loss": 3.0125, "step": 13787 }, { "epoch": 0.8559190514619157, "grad_norm": 0.5005426442464808, "learning_rate": 8.990009319943657e-05, "loss": 2.9862, "step": 13788 }, { "epoch": 0.8559811285616736, "grad_norm": 0.534058998324107, "learning_rate": 8.989791659232137e-05, "loss": 3.0007, "step": 13789 }, { "epoch": 0.8560432056614315, "grad_norm": 0.44063380685757214, "learning_rate": 8.989573977704907e-05, "loss": 2.8982, "step": 13790 }, { "epoch": 0.8561052827611894, "grad_norm": 0.5102202920017153, "learning_rate": 8.989356275363101e-05, "loss": 2.9679, "step": 13791 }, { "epoch": 0.8561673598609473, "grad_norm": 0.48582705380267655, "learning_rate": 8.989138552207855e-05, "loss": 3.0527, "step": 13792 }, { "epoch": 0.8562294369607052, "grad_norm": 0.48210824200081803, "learning_rate": 8.988920808240303e-05, "loss": 3.0037, "step": 13793 }, { "epoch": 0.8562915140604631, "grad_norm": 0.4722031399745263, "learning_rate": 8.988703043461583e-05, "loss": 2.9501, "step": 13794 }, { "epoch": 0.856353591160221, "grad_norm": 0.47541799850455624, "learning_rate": 8.988485257872831e-05, "loss": 2.9402, "step": 13795 }, { "epoch": 0.856415668259979, "grad_norm": 0.46383911047839993, "learning_rate": 8.988267451475185e-05, "loss": 2.9592, "step": 13796 }, { "epoch": 0.8564777453597368, "grad_norm": 0.4983428972925536, "learning_rate": 8.988049624269779e-05, "loss": 2.9727, "step": 13797 }, { "epoch": 0.8565398224594947, "grad_norm": 0.5748925181152321, "learning_rate": 8.987831776257749e-05, "loss": 3.0169, "step": 13798 }, { "epoch": 0.8566018995592526, "grad_norm": 0.4824191568274891, "learning_rate": 8.987613907440234e-05, "loss": 2.9395, "step": 13799 }, { "epoch": 0.8566639766590105, "grad_norm": 0.4287116500396716, "learning_rate": 8.987396017818368e-05, "loss": 2.9159, "step": 13800 }, { "epoch": 0.8567260537587684, "grad_norm": 0.47236794750786615, "learning_rate": 8.987178107393287e-05, "loss": 3.0113, "step": 13801 }, { "epoch": 0.8567881308585263, "grad_norm": 0.46664557139650464, "learning_rate": 8.986960176166134e-05, "loss": 2.9618, "step": 13802 }, { "epoch": 0.8568502079582841, "grad_norm": 0.4698693738419403, "learning_rate": 8.98674222413804e-05, "loss": 2.9818, "step": 13803 }, { "epoch": 0.8569122850580421, "grad_norm": 0.47939067939272345, "learning_rate": 8.986524251310143e-05, "loss": 2.9403, "step": 13804 }, { "epoch": 0.8569743621578, "grad_norm": 0.5004209296335818, "learning_rate": 8.986306257683583e-05, "loss": 2.9834, "step": 13805 }, { "epoch": 0.8570364392575579, "grad_norm": 0.5618263555395377, "learning_rate": 8.986088243259494e-05, "loss": 2.9972, "step": 13806 }, { "epoch": 0.8570985163573158, "grad_norm": 0.5267608246453727, "learning_rate": 8.985870208039016e-05, "loss": 2.9432, "step": 13807 }, { "epoch": 0.8571605934570737, "grad_norm": 0.5621001623908451, "learning_rate": 8.985652152023284e-05, "loss": 3.0733, "step": 13808 }, { "epoch": 0.8572226705568315, "grad_norm": 0.512923184572151, "learning_rate": 8.985434075213439e-05, "loss": 3.002, "step": 13809 }, { "epoch": 0.8572847476565895, "grad_norm": 0.5723011409427535, "learning_rate": 8.985215977610617e-05, "loss": 2.9247, "step": 13810 }, { "epoch": 0.8573468247563474, "grad_norm": 0.5343781822511384, "learning_rate": 8.984997859215955e-05, "loss": 3.0542, "step": 13811 }, { "epoch": 0.8574089018561053, "grad_norm": 0.619105389191061, "learning_rate": 8.984779720030591e-05, "loss": 2.9753, "step": 13812 }, { "epoch": 0.8574709789558632, "grad_norm": 0.6176322589965735, "learning_rate": 8.984561560055666e-05, "loss": 3.1099, "step": 13813 }, { "epoch": 0.8575330560556211, "grad_norm": 0.5796547574566937, "learning_rate": 8.984343379292315e-05, "loss": 3.0334, "step": 13814 }, { "epoch": 0.8575951331553789, "grad_norm": 0.5623572486302414, "learning_rate": 8.984125177741677e-05, "loss": 2.9631, "step": 13815 }, { "epoch": 0.8576572102551369, "grad_norm": 0.5023852021243788, "learning_rate": 8.98390695540489e-05, "loss": 3.0425, "step": 13816 }, { "epoch": 0.8577192873548948, "grad_norm": 0.5567476136302465, "learning_rate": 8.983688712283096e-05, "loss": 2.9883, "step": 13817 }, { "epoch": 0.8577813644546527, "grad_norm": 0.5738469951064031, "learning_rate": 8.983470448377429e-05, "loss": 2.9259, "step": 13818 }, { "epoch": 0.8578434415544106, "grad_norm": 0.5291687926552585, "learning_rate": 8.983252163689029e-05, "loss": 2.8808, "step": 13819 }, { "epoch": 0.8579055186541685, "grad_norm": 0.6332540502196559, "learning_rate": 8.983033858219036e-05, "loss": 2.9968, "step": 13820 }, { "epoch": 0.8579675957539263, "grad_norm": 0.6248283388908756, "learning_rate": 8.982815531968589e-05, "loss": 3.0502, "step": 13821 }, { "epoch": 0.8580296728536843, "grad_norm": 0.5742440769552725, "learning_rate": 8.982597184938825e-05, "loss": 3.0957, "step": 13822 }, { "epoch": 0.8580917499534422, "grad_norm": 0.46679893397431793, "learning_rate": 8.982378817130888e-05, "loss": 2.9968, "step": 13823 }, { "epoch": 0.8581538270532001, "grad_norm": 0.5776623601696064, "learning_rate": 8.982160428545912e-05, "loss": 3.0419, "step": 13824 }, { "epoch": 0.858215904152958, "grad_norm": 0.5387830147949685, "learning_rate": 8.981942019185036e-05, "loss": 3.0438, "step": 13825 }, { "epoch": 0.8582779812527159, "grad_norm": 0.5070983266324776, "learning_rate": 8.981723589049405e-05, "loss": 3.0176, "step": 13826 }, { "epoch": 0.8583400583524737, "grad_norm": 0.6108026409331275, "learning_rate": 8.981505138140155e-05, "loss": 2.9845, "step": 13827 }, { "epoch": 0.8584021354522317, "grad_norm": 0.5117502858003192, "learning_rate": 8.981286666458426e-05, "loss": 3.0894, "step": 13828 }, { "epoch": 0.8584642125519896, "grad_norm": 0.5812911377982483, "learning_rate": 8.981068174005357e-05, "loss": 3.0521, "step": 13829 }, { "epoch": 0.8585262896517475, "grad_norm": 0.5766638514233191, "learning_rate": 8.98084966078209e-05, "loss": 2.8833, "step": 13830 }, { "epoch": 0.8585883667515054, "grad_norm": 0.4615226038514824, "learning_rate": 8.980631126789762e-05, "loss": 2.9359, "step": 13831 }, { "epoch": 0.8586504438512633, "grad_norm": 0.5398440845449821, "learning_rate": 8.980412572029517e-05, "loss": 3.0436, "step": 13832 }, { "epoch": 0.8587125209510211, "grad_norm": 0.5274084263452832, "learning_rate": 8.980193996502493e-05, "loss": 2.9879, "step": 13833 }, { "epoch": 0.858774598050779, "grad_norm": 0.5688148063375671, "learning_rate": 8.97997540020983e-05, "loss": 2.9308, "step": 13834 }, { "epoch": 0.858836675150537, "grad_norm": 0.4919693478518444, "learning_rate": 8.97975678315267e-05, "loss": 2.9416, "step": 13835 }, { "epoch": 0.8588987522502949, "grad_norm": 0.500012516818732, "learning_rate": 8.979538145332154e-05, "loss": 3.0028, "step": 13836 }, { "epoch": 0.8589608293500528, "grad_norm": 0.48983517385273073, "learning_rate": 8.97931948674942e-05, "loss": 3.0704, "step": 13837 }, { "epoch": 0.8590229064498107, "grad_norm": 0.58660822308568, "learning_rate": 8.97910080740561e-05, "loss": 2.9623, "step": 13838 }, { "epoch": 0.8590849835495685, "grad_norm": 0.5372775837459158, "learning_rate": 8.978882107301865e-05, "loss": 3.0718, "step": 13839 }, { "epoch": 0.8591470606493264, "grad_norm": 0.5569036029872424, "learning_rate": 8.978663386439325e-05, "loss": 3.0277, "step": 13840 }, { "epoch": 0.8592091377490844, "grad_norm": 0.5443334102400854, "learning_rate": 8.978444644819136e-05, "loss": 2.9501, "step": 13841 }, { "epoch": 0.8592712148488423, "grad_norm": 0.5326039787196154, "learning_rate": 8.978225882442431e-05, "loss": 2.9851, "step": 13842 }, { "epoch": 0.8593332919486002, "grad_norm": 0.48874035423302187, "learning_rate": 8.978007099310358e-05, "loss": 3.0034, "step": 13843 }, { "epoch": 0.8593953690483581, "grad_norm": 0.4953527999752974, "learning_rate": 8.977788295424057e-05, "loss": 2.9583, "step": 13844 }, { "epoch": 0.8594574461481159, "grad_norm": 0.5920171044623738, "learning_rate": 8.977569470784668e-05, "loss": 2.9849, "step": 13845 }, { "epoch": 0.8595195232478738, "grad_norm": 0.5675182870195534, "learning_rate": 8.977350625393333e-05, "loss": 3.022, "step": 13846 }, { "epoch": 0.8595816003476318, "grad_norm": 0.5198314057820256, "learning_rate": 8.977131759251195e-05, "loss": 2.9566, "step": 13847 }, { "epoch": 0.8596436774473897, "grad_norm": 0.5800536950101473, "learning_rate": 8.976912872359396e-05, "loss": 3.0741, "step": 13848 }, { "epoch": 0.8597057545471476, "grad_norm": 0.5070869250654886, "learning_rate": 8.976693964719075e-05, "loss": 3.0767, "step": 13849 }, { "epoch": 0.8597678316469055, "grad_norm": 0.5037140886460391, "learning_rate": 8.976475036331378e-05, "loss": 3.0391, "step": 13850 }, { "epoch": 0.8598299087466633, "grad_norm": 0.530817388652396, "learning_rate": 8.976256087197443e-05, "loss": 2.961, "step": 13851 }, { "epoch": 0.8598919858464212, "grad_norm": 0.46604020149035885, "learning_rate": 8.976037117318417e-05, "loss": 2.9912, "step": 13852 }, { "epoch": 0.8599540629461792, "grad_norm": 0.4919337572244494, "learning_rate": 8.975818126695439e-05, "loss": 3.0137, "step": 13853 }, { "epoch": 0.8600161400459371, "grad_norm": 0.5509763872483714, "learning_rate": 8.975599115329653e-05, "loss": 2.977, "step": 13854 }, { "epoch": 0.860078217145695, "grad_norm": 0.4840336796829308, "learning_rate": 8.975380083222202e-05, "loss": 2.8496, "step": 13855 }, { "epoch": 0.8601402942454529, "grad_norm": 0.4566461456722176, "learning_rate": 8.975161030374228e-05, "loss": 2.8886, "step": 13856 }, { "epoch": 0.8602023713452107, "grad_norm": 0.5273807088828198, "learning_rate": 8.974941956786873e-05, "loss": 3.0147, "step": 13857 }, { "epoch": 0.8602644484449686, "grad_norm": 0.5094357588870112, "learning_rate": 8.974722862461282e-05, "loss": 3.077, "step": 13858 }, { "epoch": 0.8603265255447266, "grad_norm": 0.4691902953827224, "learning_rate": 8.974503747398597e-05, "loss": 2.9468, "step": 13859 }, { "epoch": 0.8603886026444845, "grad_norm": 0.4738582743010141, "learning_rate": 8.97428461159996e-05, "loss": 2.9462, "step": 13860 }, { "epoch": 0.8604506797442424, "grad_norm": 0.4811306557742881, "learning_rate": 8.974065455066515e-05, "loss": 2.9192, "step": 13861 }, { "epoch": 0.8605127568440002, "grad_norm": 0.5132910473496707, "learning_rate": 8.973846277799407e-05, "loss": 2.9202, "step": 13862 }, { "epoch": 0.8605748339437581, "grad_norm": 0.4546723910571026, "learning_rate": 8.973627079799778e-05, "loss": 2.9589, "step": 13863 }, { "epoch": 0.860636911043516, "grad_norm": 0.4773446059648547, "learning_rate": 8.973407861068771e-05, "loss": 3.0236, "step": 13864 }, { "epoch": 0.860698988143274, "grad_norm": 0.5502738238493272, "learning_rate": 8.973188621607531e-05, "loss": 2.9756, "step": 13865 }, { "epoch": 0.8607610652430319, "grad_norm": 0.4647407577923331, "learning_rate": 8.972969361417202e-05, "loss": 2.8614, "step": 13866 }, { "epoch": 0.8608231423427898, "grad_norm": 0.5347626345364848, "learning_rate": 8.972750080498927e-05, "loss": 3.0382, "step": 13867 }, { "epoch": 0.8608852194425476, "grad_norm": 0.6342198299559472, "learning_rate": 8.97253077885385e-05, "loss": 2.9616, "step": 13868 }, { "epoch": 0.8609472965423055, "grad_norm": 0.5102536726225542, "learning_rate": 8.972311456483117e-05, "loss": 2.9881, "step": 13869 }, { "epoch": 0.8610093736420634, "grad_norm": 0.5068613387782174, "learning_rate": 8.972092113387869e-05, "loss": 3.0415, "step": 13870 }, { "epoch": 0.8610714507418213, "grad_norm": 0.5482791041552084, "learning_rate": 8.971872749569254e-05, "loss": 3.077, "step": 13871 }, { "epoch": 0.8611335278415793, "grad_norm": 0.513552952913863, "learning_rate": 8.971653365028413e-05, "loss": 2.9936, "step": 13872 }, { "epoch": 0.8611956049413372, "grad_norm": 0.4742253951937845, "learning_rate": 8.971433959766493e-05, "loss": 2.9687, "step": 13873 }, { "epoch": 0.861257682041095, "grad_norm": 0.5940960076688352, "learning_rate": 8.971214533784637e-05, "loss": 2.967, "step": 13874 }, { "epoch": 0.8613197591408529, "grad_norm": 0.5036538253302618, "learning_rate": 8.970995087083993e-05, "loss": 3.0951, "step": 13875 }, { "epoch": 0.8613818362406108, "grad_norm": 0.5308718738109068, "learning_rate": 8.970775619665702e-05, "loss": 2.957, "step": 13876 }, { "epoch": 0.8614439133403687, "grad_norm": 0.5091105084528539, "learning_rate": 8.97055613153091e-05, "loss": 2.9783, "step": 13877 }, { "epoch": 0.8615059904401267, "grad_norm": 0.47071966561661005, "learning_rate": 8.970336622680763e-05, "loss": 3.0184, "step": 13878 }, { "epoch": 0.8615680675398846, "grad_norm": 0.545228960475227, "learning_rate": 8.970117093116408e-05, "loss": 2.9146, "step": 13879 }, { "epoch": 0.8616301446396424, "grad_norm": 0.501627924327183, "learning_rate": 8.969897542838987e-05, "loss": 3.0451, "step": 13880 }, { "epoch": 0.8616922217394003, "grad_norm": 0.6175261184249837, "learning_rate": 8.969677971849648e-05, "loss": 2.9587, "step": 13881 }, { "epoch": 0.8617542988391582, "grad_norm": 0.48529947412011554, "learning_rate": 8.969458380149533e-05, "loss": 3.0022, "step": 13882 }, { "epoch": 0.8618163759389161, "grad_norm": 0.526084668381619, "learning_rate": 8.969238767739791e-05, "loss": 2.9185, "step": 13883 }, { "epoch": 0.861878453038674, "grad_norm": 0.6033776381683681, "learning_rate": 8.969019134621568e-05, "loss": 2.9786, "step": 13884 }, { "epoch": 0.861940530138432, "grad_norm": 0.5627466561681167, "learning_rate": 8.968799480796006e-05, "loss": 2.9916, "step": 13885 }, { "epoch": 0.8620026072381898, "grad_norm": 0.5463539365968563, "learning_rate": 8.968579806264256e-05, "loss": 2.9628, "step": 13886 }, { "epoch": 0.8620646843379477, "grad_norm": 0.6161451833618897, "learning_rate": 8.968360111027461e-05, "loss": 3.033, "step": 13887 }, { "epoch": 0.8621267614377056, "grad_norm": 0.5245180790497266, "learning_rate": 8.968140395086768e-05, "loss": 3.0809, "step": 13888 }, { "epoch": 0.8621888385374635, "grad_norm": 0.47707077187516267, "learning_rate": 8.967920658443324e-05, "loss": 3.0394, "step": 13889 }, { "epoch": 0.8622509156372214, "grad_norm": 0.586919724219715, "learning_rate": 8.967700901098273e-05, "loss": 3.0224, "step": 13890 }, { "epoch": 0.8623129927369794, "grad_norm": 0.5312296078077954, "learning_rate": 8.967481123052765e-05, "loss": 2.9856, "step": 13891 }, { "epoch": 0.8623750698367372, "grad_norm": 0.5447959891221347, "learning_rate": 8.967261324307944e-05, "loss": 2.9034, "step": 13892 }, { "epoch": 0.8624371469364951, "grad_norm": 0.5713997497272311, "learning_rate": 8.967041504864957e-05, "loss": 3.0976, "step": 13893 }, { "epoch": 0.862499224036253, "grad_norm": 0.582187887530523, "learning_rate": 8.966821664724952e-05, "loss": 3.0227, "step": 13894 }, { "epoch": 0.8625613011360109, "grad_norm": 0.47738281377354264, "learning_rate": 8.966601803889075e-05, "loss": 3.0682, "step": 13895 }, { "epoch": 0.8626233782357688, "grad_norm": 0.5559119558228628, "learning_rate": 8.966381922358475e-05, "loss": 2.9021, "step": 13896 }, { "epoch": 0.8626854553355268, "grad_norm": 0.5160875268896467, "learning_rate": 8.966162020134297e-05, "loss": 2.9072, "step": 13897 }, { "epoch": 0.8627475324352846, "grad_norm": 0.5199037234880137, "learning_rate": 8.965942097217688e-05, "loss": 3.0078, "step": 13898 }, { "epoch": 0.8628096095350425, "grad_norm": 0.5081188744883102, "learning_rate": 8.965722153609796e-05, "loss": 3.0049, "step": 13899 }, { "epoch": 0.8628716866348004, "grad_norm": 0.5118052868071008, "learning_rate": 8.96550218931177e-05, "loss": 3.0196, "step": 13900 }, { "epoch": 0.8629337637345583, "grad_norm": 0.5021764651725855, "learning_rate": 8.965282204324756e-05, "loss": 2.9939, "step": 13901 }, { "epoch": 0.8629958408343162, "grad_norm": 0.45403691313574673, "learning_rate": 8.965062198649902e-05, "loss": 2.9734, "step": 13902 }, { "epoch": 0.8630579179340742, "grad_norm": 0.42392608284033306, "learning_rate": 8.964842172288356e-05, "loss": 2.8359, "step": 13903 }, { "epoch": 0.863119995033832, "grad_norm": 0.49173503847541983, "learning_rate": 8.964622125241266e-05, "loss": 2.9895, "step": 13904 }, { "epoch": 0.8631820721335899, "grad_norm": 0.5059126601268148, "learning_rate": 8.96440205750978e-05, "loss": 3.0533, "step": 13905 }, { "epoch": 0.8632441492333478, "grad_norm": 0.4527751459914381, "learning_rate": 8.964181969095046e-05, "loss": 2.9928, "step": 13906 }, { "epoch": 0.8633062263331057, "grad_norm": 0.5476444281201366, "learning_rate": 8.963961859998212e-05, "loss": 2.8819, "step": 13907 }, { "epoch": 0.8633683034328636, "grad_norm": 0.4979954711715796, "learning_rate": 8.963741730220428e-05, "loss": 2.9882, "step": 13908 }, { "epoch": 0.8634303805326216, "grad_norm": 0.5606454104702515, "learning_rate": 8.963521579762839e-05, "loss": 2.9438, "step": 13909 }, { "epoch": 0.8634924576323794, "grad_norm": 0.570892744267592, "learning_rate": 8.963301408626597e-05, "loss": 3.0233, "step": 13910 }, { "epoch": 0.8635545347321373, "grad_norm": 0.47764721915231106, "learning_rate": 8.96308121681285e-05, "loss": 2.9865, "step": 13911 }, { "epoch": 0.8636166118318952, "grad_norm": 0.5095422189439028, "learning_rate": 8.962861004322744e-05, "loss": 2.9486, "step": 13912 }, { "epoch": 0.8636786889316531, "grad_norm": 0.554719279278839, "learning_rate": 8.962640771157433e-05, "loss": 2.9096, "step": 13913 }, { "epoch": 0.863740766031411, "grad_norm": 0.5125258683444912, "learning_rate": 8.96242051731806e-05, "loss": 2.9741, "step": 13914 }, { "epoch": 0.863802843131169, "grad_norm": 0.5649845037613943, "learning_rate": 8.96220024280578e-05, "loss": 3.0365, "step": 13915 }, { "epoch": 0.8638649202309268, "grad_norm": 0.5323295003754591, "learning_rate": 8.961979947621739e-05, "loss": 2.9885, "step": 13916 }, { "epoch": 0.8639269973306847, "grad_norm": 0.5737319491752841, "learning_rate": 8.961759631767085e-05, "loss": 2.9218, "step": 13917 }, { "epoch": 0.8639890744304426, "grad_norm": 0.49219506878558517, "learning_rate": 8.961539295242971e-05, "loss": 3.0025, "step": 13918 }, { "epoch": 0.8640511515302005, "grad_norm": 0.5484156025484608, "learning_rate": 8.961318938050544e-05, "loss": 3.0229, "step": 13919 }, { "epoch": 0.8641132286299584, "grad_norm": 0.610443181600149, "learning_rate": 8.961098560190955e-05, "loss": 3.0635, "step": 13920 }, { "epoch": 0.8641753057297163, "grad_norm": 0.5612148067000858, "learning_rate": 8.960878161665353e-05, "loss": 2.9113, "step": 13921 }, { "epoch": 0.8642373828294742, "grad_norm": 0.4887646685347908, "learning_rate": 8.960657742474888e-05, "loss": 2.9586, "step": 13922 }, { "epoch": 0.8642994599292321, "grad_norm": 0.6317502749359052, "learning_rate": 8.96043730262071e-05, "loss": 3.0381, "step": 13923 }, { "epoch": 0.86436153702899, "grad_norm": 0.5517134657060897, "learning_rate": 8.96021684210397e-05, "loss": 3.0272, "step": 13924 }, { "epoch": 0.8644236141287479, "grad_norm": 0.5526021792573512, "learning_rate": 8.959996360925818e-05, "loss": 2.9707, "step": 13925 }, { "epoch": 0.8644856912285058, "grad_norm": 0.5030648059258945, "learning_rate": 8.9597758590874e-05, "loss": 2.9293, "step": 13926 }, { "epoch": 0.8645477683282637, "grad_norm": 0.482754318657537, "learning_rate": 8.959555336589873e-05, "loss": 3.0056, "step": 13927 }, { "epoch": 0.8646098454280216, "grad_norm": 0.4615337265746837, "learning_rate": 8.959334793434385e-05, "loss": 3.01, "step": 13928 }, { "epoch": 0.8646719225277795, "grad_norm": 0.5218938584261081, "learning_rate": 8.959114229622086e-05, "loss": 3.022, "step": 13929 }, { "epoch": 0.8647339996275374, "grad_norm": 0.4827542877905699, "learning_rate": 8.958893645154125e-05, "loss": 2.9231, "step": 13930 }, { "epoch": 0.8647960767272953, "grad_norm": 0.46502935086186725, "learning_rate": 8.958673040031656e-05, "loss": 3.0055, "step": 13931 }, { "epoch": 0.8648581538270532, "grad_norm": 0.5046755809970882, "learning_rate": 8.95845241425583e-05, "loss": 2.9551, "step": 13932 }, { "epoch": 0.8649202309268111, "grad_norm": 0.5396581758662155, "learning_rate": 8.958231767827795e-05, "loss": 3.0753, "step": 13933 }, { "epoch": 0.864982308026569, "grad_norm": 0.5452410675744394, "learning_rate": 8.958011100748705e-05, "loss": 2.9906, "step": 13934 }, { "epoch": 0.8650443851263269, "grad_norm": 0.5112302938641355, "learning_rate": 8.957790413019709e-05, "loss": 3.074, "step": 13935 }, { "epoch": 0.8651064622260848, "grad_norm": 0.5313556790425382, "learning_rate": 8.95756970464196e-05, "loss": 3.0602, "step": 13936 }, { "epoch": 0.8651685393258427, "grad_norm": 0.4894159484042632, "learning_rate": 8.95734897561661e-05, "loss": 3.0087, "step": 13937 }, { "epoch": 0.8652306164256006, "grad_norm": 0.5182597727978941, "learning_rate": 8.957128225944809e-05, "loss": 2.9691, "step": 13938 }, { "epoch": 0.8652926935253585, "grad_norm": 0.516090327594253, "learning_rate": 8.95690745562771e-05, "loss": 2.9738, "step": 13939 }, { "epoch": 0.8653547706251163, "grad_norm": 0.5123722475438152, "learning_rate": 8.956686664666463e-05, "loss": 2.9325, "step": 13940 }, { "epoch": 0.8654168477248743, "grad_norm": 0.5015125344840786, "learning_rate": 8.956465853062223e-05, "loss": 2.9561, "step": 13941 }, { "epoch": 0.8654789248246322, "grad_norm": 0.4828990476003653, "learning_rate": 8.956245020816138e-05, "loss": 3.0231, "step": 13942 }, { "epoch": 0.8655410019243901, "grad_norm": 0.5099576683334119, "learning_rate": 8.956024167929365e-05, "loss": 2.9369, "step": 13943 }, { "epoch": 0.865603079024148, "grad_norm": 0.47997123266979175, "learning_rate": 8.955803294403052e-05, "loss": 2.9638, "step": 13944 }, { "epoch": 0.8656651561239059, "grad_norm": 0.5662822192964625, "learning_rate": 8.955582400238353e-05, "loss": 3.0147, "step": 13945 }, { "epoch": 0.8657272332236637, "grad_norm": 0.4922901379575925, "learning_rate": 8.95536148543642e-05, "loss": 3.0371, "step": 13946 }, { "epoch": 0.8657893103234217, "grad_norm": 0.5541648820499376, "learning_rate": 8.955140549998407e-05, "loss": 2.9633, "step": 13947 }, { "epoch": 0.8658513874231796, "grad_norm": 0.47385358875658984, "learning_rate": 8.954919593925464e-05, "loss": 2.8819, "step": 13948 }, { "epoch": 0.8659134645229375, "grad_norm": 0.49162367663773987, "learning_rate": 8.954698617218746e-05, "loss": 2.9908, "step": 13949 }, { "epoch": 0.8659755416226954, "grad_norm": 0.5161142340211907, "learning_rate": 8.954477619879406e-05, "loss": 3.0286, "step": 13950 }, { "epoch": 0.8660376187224533, "grad_norm": 0.5415520057607972, "learning_rate": 8.954256601908596e-05, "loss": 3.0037, "step": 13951 }, { "epoch": 0.8660996958222111, "grad_norm": 0.4941718653276249, "learning_rate": 8.954035563307469e-05, "loss": 3.0481, "step": 13952 }, { "epoch": 0.866161772921969, "grad_norm": 0.5087593933944121, "learning_rate": 8.953814504077179e-05, "loss": 3.0022, "step": 13953 }, { "epoch": 0.866223850021727, "grad_norm": 0.5525067403218206, "learning_rate": 8.95359342421888e-05, "loss": 2.8833, "step": 13954 }, { "epoch": 0.8662859271214849, "grad_norm": 0.5183902918112617, "learning_rate": 8.953372323733722e-05, "loss": 2.9747, "step": 13955 }, { "epoch": 0.8663480042212428, "grad_norm": 0.550821668075258, "learning_rate": 8.953151202622862e-05, "loss": 2.8993, "step": 13956 }, { "epoch": 0.8664100813210007, "grad_norm": 0.4980261555270086, "learning_rate": 8.952930060887452e-05, "loss": 2.9976, "step": 13957 }, { "epoch": 0.8664721584207585, "grad_norm": 0.6270334067593937, "learning_rate": 8.952708898528648e-05, "loss": 2.9915, "step": 13958 }, { "epoch": 0.8665342355205164, "grad_norm": 0.48885828635707557, "learning_rate": 8.9524877155476e-05, "loss": 2.9792, "step": 13959 }, { "epoch": 0.8665963126202744, "grad_norm": 0.5225543927897299, "learning_rate": 8.952266511945464e-05, "loss": 3.1018, "step": 13960 }, { "epoch": 0.8666583897200323, "grad_norm": 0.5609393722800459, "learning_rate": 8.952045287723396e-05, "loss": 3.0231, "step": 13961 }, { "epoch": 0.8667204668197902, "grad_norm": 0.5594831735025595, "learning_rate": 8.951824042882547e-05, "loss": 3.0507, "step": 13962 }, { "epoch": 0.8667825439195481, "grad_norm": 0.494524147259898, "learning_rate": 8.951602777424073e-05, "loss": 2.9653, "step": 13963 }, { "epoch": 0.8668446210193059, "grad_norm": 0.5176346945887683, "learning_rate": 8.95138149134913e-05, "loss": 2.965, "step": 13964 }, { "epoch": 0.8669066981190638, "grad_norm": 0.4878626135060796, "learning_rate": 8.951160184658867e-05, "loss": 3.013, "step": 13965 }, { "epoch": 0.8669687752188218, "grad_norm": 0.4797367163891956, "learning_rate": 8.950938857354443e-05, "loss": 2.9308, "step": 13966 }, { "epoch": 0.8670308523185797, "grad_norm": 0.4598223564519528, "learning_rate": 8.950717509437014e-05, "loss": 2.9519, "step": 13967 }, { "epoch": 0.8670929294183376, "grad_norm": 0.46711600340258497, "learning_rate": 8.950496140907731e-05, "loss": 2.9778, "step": 13968 }, { "epoch": 0.8671550065180955, "grad_norm": 0.45698108560811246, "learning_rate": 8.950274751767752e-05, "loss": 2.9187, "step": 13969 }, { "epoch": 0.8672170836178533, "grad_norm": 0.501783082414056, "learning_rate": 8.950053342018229e-05, "loss": 2.9813, "step": 13970 }, { "epoch": 0.8672791607176112, "grad_norm": 0.47958338971417996, "learning_rate": 8.94983191166032e-05, "loss": 2.9978, "step": 13971 }, { "epoch": 0.8673412378173692, "grad_norm": 0.5048038265932625, "learning_rate": 8.949610460695177e-05, "loss": 2.9949, "step": 13972 }, { "epoch": 0.8674033149171271, "grad_norm": 0.49980623245745825, "learning_rate": 8.949388989123959e-05, "loss": 3.0339, "step": 13973 }, { "epoch": 0.867465392016885, "grad_norm": 0.5326607149563094, "learning_rate": 8.94916749694782e-05, "loss": 2.8829, "step": 13974 }, { "epoch": 0.8675274691166429, "grad_norm": 0.553201619461224, "learning_rate": 8.948945984167915e-05, "loss": 3.0511, "step": 13975 }, { "epoch": 0.8675895462164007, "grad_norm": 0.538198580435444, "learning_rate": 8.948724450785399e-05, "loss": 2.9449, "step": 13976 }, { "epoch": 0.8676516233161586, "grad_norm": 0.5586318956701961, "learning_rate": 8.948502896801432e-05, "loss": 2.9422, "step": 13977 }, { "epoch": 0.8677137004159166, "grad_norm": 0.4898346110682368, "learning_rate": 8.948281322217165e-05, "loss": 2.8958, "step": 13978 }, { "epoch": 0.8677757775156745, "grad_norm": 0.5620170215187495, "learning_rate": 8.948059727033756e-05, "loss": 3.012, "step": 13979 }, { "epoch": 0.8678378546154324, "grad_norm": 0.5307875190928366, "learning_rate": 8.947838111252362e-05, "loss": 2.9558, "step": 13980 }, { "epoch": 0.8678999317151903, "grad_norm": 0.5192241765076497, "learning_rate": 8.947616474874136e-05, "loss": 3.0686, "step": 13981 }, { "epoch": 0.8679620088149481, "grad_norm": 0.47833792053187424, "learning_rate": 8.947394817900238e-05, "loss": 2.9409, "step": 13982 }, { "epoch": 0.868024085914706, "grad_norm": 0.5183714634535751, "learning_rate": 8.947173140331822e-05, "loss": 3.0416, "step": 13983 }, { "epoch": 0.868086163014464, "grad_norm": 0.5136995199328602, "learning_rate": 8.946951442170048e-05, "loss": 2.9464, "step": 13984 }, { "epoch": 0.8681482401142219, "grad_norm": 0.48731323418165035, "learning_rate": 8.946729723416067e-05, "loss": 2.9649, "step": 13985 }, { "epoch": 0.8682103172139798, "grad_norm": 0.5261228202943248, "learning_rate": 8.946507984071042e-05, "loss": 3.0408, "step": 13986 }, { "epoch": 0.8682723943137377, "grad_norm": 0.4567901268700858, "learning_rate": 8.946286224136124e-05, "loss": 2.9887, "step": 13987 }, { "epoch": 0.8683344714134955, "grad_norm": 0.45827030882860503, "learning_rate": 8.946064443612474e-05, "loss": 2.9081, "step": 13988 }, { "epoch": 0.8683965485132534, "grad_norm": 0.4682745270473989, "learning_rate": 8.945842642501247e-05, "loss": 2.9279, "step": 13989 }, { "epoch": 0.8684586256130113, "grad_norm": 0.44769338284016635, "learning_rate": 8.945620820803604e-05, "loss": 2.9567, "step": 13990 }, { "epoch": 0.8685207027127693, "grad_norm": 0.47621110562075997, "learning_rate": 8.945398978520696e-05, "loss": 3.0544, "step": 13991 }, { "epoch": 0.8685827798125272, "grad_norm": 0.4459163185975645, "learning_rate": 8.945177115653684e-05, "loss": 2.9247, "step": 13992 }, { "epoch": 0.8686448569122851, "grad_norm": 0.4763838949241398, "learning_rate": 8.944955232203727e-05, "loss": 3.0107, "step": 13993 }, { "epoch": 0.8687069340120429, "grad_norm": 0.5592853559556553, "learning_rate": 8.944733328171981e-05, "loss": 3.0035, "step": 13994 }, { "epoch": 0.8687690111118008, "grad_norm": 0.4951381641285737, "learning_rate": 8.944511403559602e-05, "loss": 3.0872, "step": 13995 }, { "epoch": 0.8688310882115587, "grad_norm": 0.4711680668401992, "learning_rate": 8.944289458367751e-05, "loss": 2.9113, "step": 13996 }, { "epoch": 0.8688931653113167, "grad_norm": 0.5561014877256906, "learning_rate": 8.944067492597584e-05, "loss": 3.0208, "step": 13997 }, { "epoch": 0.8689552424110746, "grad_norm": 0.5479209434381044, "learning_rate": 8.943845506250258e-05, "loss": 2.8967, "step": 13998 }, { "epoch": 0.8690173195108325, "grad_norm": 0.4699177184519616, "learning_rate": 8.943623499326934e-05, "loss": 2.9143, "step": 13999 }, { "epoch": 0.8690793966105903, "grad_norm": 0.6132518736977457, "learning_rate": 8.943401471828768e-05, "loss": 3.0881, "step": 14000 }, { "epoch": 0.8691414737103482, "grad_norm": 0.6269958576600955, "learning_rate": 8.94317942375692e-05, "loss": 2.9585, "step": 14001 }, { "epoch": 0.8692035508101061, "grad_norm": 0.5256723832535405, "learning_rate": 8.942957355112547e-05, "loss": 3.0192, "step": 14002 }, { "epoch": 0.8692656279098641, "grad_norm": 0.5924193378863513, "learning_rate": 8.942735265896809e-05, "loss": 2.9355, "step": 14003 }, { "epoch": 0.869327705009622, "grad_norm": 0.5991253965389628, "learning_rate": 8.942513156110864e-05, "loss": 3.0329, "step": 14004 }, { "epoch": 0.8693897821093799, "grad_norm": 0.527011342717387, "learning_rate": 8.942291025755869e-05, "loss": 2.9369, "step": 14005 }, { "epoch": 0.8694518592091377, "grad_norm": 0.5370765259841316, "learning_rate": 8.942068874832987e-05, "loss": 3.0219, "step": 14006 }, { "epoch": 0.8695139363088956, "grad_norm": 0.5275820440968425, "learning_rate": 8.941846703343373e-05, "loss": 2.9311, "step": 14007 }, { "epoch": 0.8695760134086535, "grad_norm": 0.5236579658474271, "learning_rate": 8.94162451128819e-05, "loss": 3.0368, "step": 14008 }, { "epoch": 0.8696380905084115, "grad_norm": 0.5822404575065504, "learning_rate": 8.941402298668593e-05, "loss": 3.0117, "step": 14009 }, { "epoch": 0.8697001676081694, "grad_norm": 0.5573817622956633, "learning_rate": 8.941180065485744e-05, "loss": 3.0781, "step": 14010 }, { "epoch": 0.8697622447079273, "grad_norm": 0.5673695758056884, "learning_rate": 8.940957811740803e-05, "loss": 2.9292, "step": 14011 }, { "epoch": 0.8698243218076851, "grad_norm": 0.6230131995209602, "learning_rate": 8.940735537434926e-05, "loss": 2.9533, "step": 14012 }, { "epoch": 0.869886398907443, "grad_norm": 0.5050702217587119, "learning_rate": 8.940513242569277e-05, "loss": 2.8892, "step": 14013 }, { "epoch": 0.8699484760072009, "grad_norm": 0.5543175928008991, "learning_rate": 8.940290927145013e-05, "loss": 2.9778, "step": 14014 }, { "epoch": 0.8700105531069589, "grad_norm": 0.5451377523076978, "learning_rate": 8.940068591163296e-05, "loss": 3.0093, "step": 14015 }, { "epoch": 0.8700726302067168, "grad_norm": 0.5514427973719688, "learning_rate": 8.939846234625282e-05, "loss": 2.9987, "step": 14016 }, { "epoch": 0.8701347073064747, "grad_norm": 0.55063462165627, "learning_rate": 8.939623857532136e-05, "loss": 2.9118, "step": 14017 }, { "epoch": 0.8701967844062325, "grad_norm": 0.49897622558857596, "learning_rate": 8.939401459885016e-05, "loss": 3.0736, "step": 14018 }, { "epoch": 0.8702588615059904, "grad_norm": 0.5348126220114532, "learning_rate": 8.939179041685081e-05, "loss": 2.9694, "step": 14019 }, { "epoch": 0.8703209386057483, "grad_norm": 0.505396421230901, "learning_rate": 8.938956602933493e-05, "loss": 3.0419, "step": 14020 }, { "epoch": 0.8703830157055062, "grad_norm": 0.47705310832679126, "learning_rate": 8.938734143631412e-05, "loss": 2.8855, "step": 14021 }, { "epoch": 0.8704450928052642, "grad_norm": 0.5285601679604011, "learning_rate": 8.93851166378e-05, "loss": 2.9767, "step": 14022 }, { "epoch": 0.8705071699050221, "grad_norm": 0.5076257509020766, "learning_rate": 8.938289163380415e-05, "loss": 3.02, "step": 14023 }, { "epoch": 0.8705692470047799, "grad_norm": 0.4998064113406841, "learning_rate": 8.93806664243382e-05, "loss": 2.9504, "step": 14024 }, { "epoch": 0.8706313241045378, "grad_norm": 0.5367927033894226, "learning_rate": 8.937844100941376e-05, "loss": 3.0646, "step": 14025 }, { "epoch": 0.8706934012042957, "grad_norm": 0.5302817272009009, "learning_rate": 8.937621538904242e-05, "loss": 3.0658, "step": 14026 }, { "epoch": 0.8707554783040536, "grad_norm": 0.5287329002302762, "learning_rate": 8.937398956323581e-05, "loss": 3.0363, "step": 14027 }, { "epoch": 0.8708175554038116, "grad_norm": 0.49970549612459936, "learning_rate": 8.937176353200556e-05, "loss": 2.9284, "step": 14028 }, { "epoch": 0.8708796325035695, "grad_norm": 0.5070996783476486, "learning_rate": 8.936953729536323e-05, "loss": 3.0194, "step": 14029 }, { "epoch": 0.8709417096033273, "grad_norm": 0.5500541920406739, "learning_rate": 8.936731085332049e-05, "loss": 2.9794, "step": 14030 }, { "epoch": 0.8710037867030852, "grad_norm": 0.4708640903650099, "learning_rate": 8.936508420588892e-05, "loss": 2.9405, "step": 14031 }, { "epoch": 0.8710658638028431, "grad_norm": 0.5672175417844081, "learning_rate": 8.936285735308016e-05, "loss": 3.0277, "step": 14032 }, { "epoch": 0.871127940902601, "grad_norm": 0.4926421321270051, "learning_rate": 8.93606302949058e-05, "loss": 3.0165, "step": 14033 }, { "epoch": 0.871190018002359, "grad_norm": 0.514788932371063, "learning_rate": 8.935840303137748e-05, "loss": 2.9192, "step": 14034 }, { "epoch": 0.8712520951021169, "grad_norm": 0.4741701204859086, "learning_rate": 8.935617556250684e-05, "loss": 3.0213, "step": 14035 }, { "epoch": 0.8713141722018747, "grad_norm": 0.5279872500205955, "learning_rate": 8.935394788830547e-05, "loss": 2.9862, "step": 14036 }, { "epoch": 0.8713762493016326, "grad_norm": 0.46965667141205003, "learning_rate": 8.935172000878498e-05, "loss": 2.9365, "step": 14037 }, { "epoch": 0.8714383264013905, "grad_norm": 0.4815359127074663, "learning_rate": 8.934949192395703e-05, "loss": 2.9495, "step": 14038 }, { "epoch": 0.8715004035011484, "grad_norm": 0.5012410973117402, "learning_rate": 8.934726363383323e-05, "loss": 2.9875, "step": 14039 }, { "epoch": 0.8715624806009064, "grad_norm": 0.5297628512279655, "learning_rate": 8.934503513842518e-05, "loss": 2.9539, "step": 14040 }, { "epoch": 0.8716245577006643, "grad_norm": 0.4716337667348545, "learning_rate": 8.934280643774456e-05, "loss": 2.9929, "step": 14041 }, { "epoch": 0.8716866348004221, "grad_norm": 0.49150728724404635, "learning_rate": 8.934057753180296e-05, "loss": 2.9946, "step": 14042 }, { "epoch": 0.87174871190018, "grad_norm": 0.5034737024248953, "learning_rate": 8.9338348420612e-05, "loss": 3.0314, "step": 14043 }, { "epoch": 0.8718107889999379, "grad_norm": 0.4715978737083902, "learning_rate": 8.933611910418334e-05, "loss": 2.9697, "step": 14044 }, { "epoch": 0.8718728660996958, "grad_norm": 0.4368263234974852, "learning_rate": 8.933388958252859e-05, "loss": 2.9364, "step": 14045 }, { "epoch": 0.8719349431994537, "grad_norm": 0.46637261235840843, "learning_rate": 8.933165985565941e-05, "loss": 2.9305, "step": 14046 }, { "epoch": 0.8719970202992117, "grad_norm": 0.44780468834760695, "learning_rate": 8.93294299235874e-05, "loss": 3.0109, "step": 14047 }, { "epoch": 0.8720590973989695, "grad_norm": 0.4904269869725093, "learning_rate": 8.93271997863242e-05, "loss": 3.0127, "step": 14048 }, { "epoch": 0.8721211744987274, "grad_norm": 0.5090322956470478, "learning_rate": 8.932496944388147e-05, "loss": 2.9627, "step": 14049 }, { "epoch": 0.8721832515984853, "grad_norm": 0.4961882523765102, "learning_rate": 8.932273889627081e-05, "loss": 2.9002, "step": 14050 }, { "epoch": 0.8722453286982432, "grad_norm": 0.5263691408478511, "learning_rate": 8.932050814350388e-05, "loss": 3.0087, "step": 14051 }, { "epoch": 0.8723074057980011, "grad_norm": 0.5476387685034496, "learning_rate": 8.931827718559231e-05, "loss": 3.0399, "step": 14052 }, { "epoch": 0.8723694828977591, "grad_norm": 0.48255487685845455, "learning_rate": 8.931604602254775e-05, "loss": 3.0018, "step": 14053 }, { "epoch": 0.8724315599975169, "grad_norm": 0.5945405966867421, "learning_rate": 8.931381465438183e-05, "loss": 3.055, "step": 14054 }, { "epoch": 0.8724936370972748, "grad_norm": 0.506224923631296, "learning_rate": 8.931158308110618e-05, "loss": 2.9681, "step": 14055 }, { "epoch": 0.8725557141970327, "grad_norm": 0.47071649999333404, "learning_rate": 8.930935130273248e-05, "loss": 3.1039, "step": 14056 }, { "epoch": 0.8726177912967906, "grad_norm": 0.5618227749613901, "learning_rate": 8.930711931927236e-05, "loss": 2.9538, "step": 14057 }, { "epoch": 0.8726798683965485, "grad_norm": 0.4700436542914083, "learning_rate": 8.930488713073744e-05, "loss": 2.9069, "step": 14058 }, { "epoch": 0.8727419454963065, "grad_norm": 0.4948136492078975, "learning_rate": 8.930265473713938e-05, "loss": 2.9607, "step": 14059 }, { "epoch": 0.8728040225960643, "grad_norm": 0.5609422412555055, "learning_rate": 8.930042213848984e-05, "loss": 3.086, "step": 14060 }, { "epoch": 0.8728660996958222, "grad_norm": 0.5744832266707033, "learning_rate": 8.929818933480044e-05, "loss": 3.0446, "step": 14061 }, { "epoch": 0.8729281767955801, "grad_norm": 0.5050140151354672, "learning_rate": 8.929595632608285e-05, "loss": 2.9737, "step": 14062 }, { "epoch": 0.872990253895338, "grad_norm": 0.5582537116266009, "learning_rate": 8.929372311234874e-05, "loss": 3.0148, "step": 14063 }, { "epoch": 0.8730523309950959, "grad_norm": 0.5505565428422907, "learning_rate": 8.929148969360971e-05, "loss": 2.943, "step": 14064 }, { "epoch": 0.8731144080948539, "grad_norm": 0.4548340171367574, "learning_rate": 8.928925606987745e-05, "loss": 2.9002, "step": 14065 }, { "epoch": 0.8731764851946117, "grad_norm": 0.5618290873727037, "learning_rate": 8.928702224116361e-05, "loss": 2.9928, "step": 14066 }, { "epoch": 0.8732385622943696, "grad_norm": 0.4913067885327599, "learning_rate": 8.928478820747983e-05, "loss": 2.9376, "step": 14067 }, { "epoch": 0.8733006393941275, "grad_norm": 0.5234674758298974, "learning_rate": 8.928255396883778e-05, "loss": 3.0259, "step": 14068 }, { "epoch": 0.8733627164938854, "grad_norm": 0.5349596873808812, "learning_rate": 8.928031952524909e-05, "loss": 3.0299, "step": 14069 }, { "epoch": 0.8734247935936433, "grad_norm": 0.4720287255753485, "learning_rate": 8.927808487672546e-05, "loss": 3.0101, "step": 14070 }, { "epoch": 0.8734868706934013, "grad_norm": 0.5406579443094345, "learning_rate": 8.927585002327852e-05, "loss": 2.9501, "step": 14071 }, { "epoch": 0.8735489477931591, "grad_norm": 0.4909574943127823, "learning_rate": 8.927361496491993e-05, "loss": 2.9944, "step": 14072 }, { "epoch": 0.873611024892917, "grad_norm": 0.529097853964774, "learning_rate": 8.927137970166137e-05, "loss": 2.9618, "step": 14073 }, { "epoch": 0.8736731019926749, "grad_norm": 0.5035111408611059, "learning_rate": 8.926914423351447e-05, "loss": 2.8804, "step": 14074 }, { "epoch": 0.8737351790924328, "grad_norm": 0.5609178013650067, "learning_rate": 8.926690856049092e-05, "loss": 3.0106, "step": 14075 }, { "epoch": 0.8737972561921907, "grad_norm": 0.5853754272280169, "learning_rate": 8.926467268260237e-05, "loss": 2.9787, "step": 14076 }, { "epoch": 0.8738593332919486, "grad_norm": 0.4798985486056921, "learning_rate": 8.926243659986051e-05, "loss": 2.9159, "step": 14077 }, { "epoch": 0.8739214103917065, "grad_norm": 0.5586801742174127, "learning_rate": 8.926020031227697e-05, "loss": 3.0164, "step": 14078 }, { "epoch": 0.8739834874914644, "grad_norm": 0.5355690248278666, "learning_rate": 8.925796381986344e-05, "loss": 2.973, "step": 14079 }, { "epoch": 0.8740455645912223, "grad_norm": 0.49749102522072786, "learning_rate": 8.925572712263157e-05, "loss": 3.0204, "step": 14080 }, { "epoch": 0.8741076416909802, "grad_norm": 0.522349522199583, "learning_rate": 8.925349022059306e-05, "loss": 2.8822, "step": 14081 }, { "epoch": 0.8741697187907381, "grad_norm": 0.4609072723822195, "learning_rate": 8.925125311375954e-05, "loss": 3.0614, "step": 14082 }, { "epoch": 0.874231795890496, "grad_norm": 0.5213076974418658, "learning_rate": 8.924901580214273e-05, "loss": 3.0422, "step": 14083 }, { "epoch": 0.8742938729902539, "grad_norm": 0.5222145997188906, "learning_rate": 8.924677828575428e-05, "loss": 2.936, "step": 14084 }, { "epoch": 0.8743559500900118, "grad_norm": 0.4884946280596961, "learning_rate": 8.924454056460584e-05, "loss": 2.9926, "step": 14085 }, { "epoch": 0.8744180271897697, "grad_norm": 0.5785877592357147, "learning_rate": 8.924230263870911e-05, "loss": 2.9892, "step": 14086 }, { "epoch": 0.8744801042895276, "grad_norm": 0.4945645078591544, "learning_rate": 8.924006450807575e-05, "loss": 2.9718, "step": 14087 }, { "epoch": 0.8745421813892855, "grad_norm": 0.4937967000677997, "learning_rate": 8.923782617271748e-05, "loss": 2.9791, "step": 14088 }, { "epoch": 0.8746042584890434, "grad_norm": 0.5341441754631404, "learning_rate": 8.923558763264592e-05, "loss": 2.894, "step": 14089 }, { "epoch": 0.8746663355888012, "grad_norm": 0.4801353290144998, "learning_rate": 8.923334888787277e-05, "loss": 2.9224, "step": 14090 }, { "epoch": 0.8747284126885592, "grad_norm": 0.5019534218635127, "learning_rate": 8.923110993840973e-05, "loss": 2.9963, "step": 14091 }, { "epoch": 0.8747904897883171, "grad_norm": 0.659107571648347, "learning_rate": 8.922887078426844e-05, "loss": 2.9675, "step": 14092 }, { "epoch": 0.874852566888075, "grad_norm": 0.5115161523813629, "learning_rate": 8.922663142546064e-05, "loss": 2.9939, "step": 14093 }, { "epoch": 0.8749146439878329, "grad_norm": 0.5586528079951169, "learning_rate": 8.922439186199798e-05, "loss": 2.9788, "step": 14094 }, { "epoch": 0.8749767210875908, "grad_norm": 0.5510892885923623, "learning_rate": 8.922215209389213e-05, "loss": 3.0153, "step": 14095 }, { "epoch": 0.8750387981873486, "grad_norm": 0.5904493948839763, "learning_rate": 8.92199121211548e-05, "loss": 2.9861, "step": 14096 }, { "epoch": 0.8751008752871066, "grad_norm": 0.48934244425167184, "learning_rate": 8.921767194379766e-05, "loss": 3.024, "step": 14097 }, { "epoch": 0.8751629523868645, "grad_norm": 0.616778854097196, "learning_rate": 8.92154315618324e-05, "loss": 2.9361, "step": 14098 }, { "epoch": 0.8752250294866224, "grad_norm": 0.49555197741849394, "learning_rate": 8.921319097527073e-05, "loss": 2.9221, "step": 14099 }, { "epoch": 0.8752871065863803, "grad_norm": 0.47000456452689576, "learning_rate": 8.921095018412432e-05, "loss": 2.9639, "step": 14100 }, { "epoch": 0.8753491836861382, "grad_norm": 0.5078241200218, "learning_rate": 8.920870918840485e-05, "loss": 2.933, "step": 14101 }, { "epoch": 0.875411260785896, "grad_norm": 0.4379222058793687, "learning_rate": 8.920646798812404e-05, "loss": 2.9597, "step": 14102 }, { "epoch": 0.875473337885654, "grad_norm": 0.4482032409901874, "learning_rate": 8.920422658329357e-05, "loss": 2.8893, "step": 14103 }, { "epoch": 0.8755354149854119, "grad_norm": 0.44835292470532606, "learning_rate": 8.920198497392511e-05, "loss": 3.0675, "step": 14104 }, { "epoch": 0.8755974920851698, "grad_norm": 0.5324382675905109, "learning_rate": 8.91997431600304e-05, "loss": 2.9828, "step": 14105 }, { "epoch": 0.8756595691849277, "grad_norm": 0.47247072787665323, "learning_rate": 8.919750114162111e-05, "loss": 2.9631, "step": 14106 }, { "epoch": 0.8757216462846856, "grad_norm": 0.4428432502111164, "learning_rate": 8.919525891870894e-05, "loss": 2.9911, "step": 14107 }, { "epoch": 0.8757837233844434, "grad_norm": 0.5187572363842802, "learning_rate": 8.919301649130558e-05, "loss": 3.0315, "step": 14108 }, { "epoch": 0.8758458004842014, "grad_norm": 0.6064076434981689, "learning_rate": 8.919077385942276e-05, "loss": 2.9848, "step": 14109 }, { "epoch": 0.8759078775839593, "grad_norm": 0.5257789566420806, "learning_rate": 8.918853102307214e-05, "loss": 3.023, "step": 14110 }, { "epoch": 0.8759699546837172, "grad_norm": 0.4993166276859275, "learning_rate": 8.918628798226544e-05, "loss": 3.0813, "step": 14111 }, { "epoch": 0.8760320317834751, "grad_norm": 0.556542692437064, "learning_rate": 8.918404473701438e-05, "loss": 2.9459, "step": 14112 }, { "epoch": 0.876094108883233, "grad_norm": 0.45561778932786534, "learning_rate": 8.918180128733063e-05, "loss": 2.9909, "step": 14113 }, { "epoch": 0.8761561859829908, "grad_norm": 0.5128431415876192, "learning_rate": 8.917955763322591e-05, "loss": 3.0709, "step": 14114 }, { "epoch": 0.8762182630827487, "grad_norm": 0.5829228733878443, "learning_rate": 8.917731377471195e-05, "loss": 2.8593, "step": 14115 }, { "epoch": 0.8762803401825067, "grad_norm": 0.4996495807089755, "learning_rate": 8.91750697118004e-05, "loss": 2.9858, "step": 14116 }, { "epoch": 0.8763424172822646, "grad_norm": 0.5529104430886408, "learning_rate": 8.917282544450303e-05, "loss": 3.0382, "step": 14117 }, { "epoch": 0.8764044943820225, "grad_norm": 0.4526429407377275, "learning_rate": 8.91705809728315e-05, "loss": 2.967, "step": 14118 }, { "epoch": 0.8764665714817804, "grad_norm": 0.49803239389946313, "learning_rate": 8.916833629679755e-05, "loss": 2.9699, "step": 14119 }, { "epoch": 0.8765286485815382, "grad_norm": 0.45169476350734805, "learning_rate": 8.916609141641288e-05, "loss": 3.0238, "step": 14120 }, { "epoch": 0.8765907256812961, "grad_norm": 0.5546741483988805, "learning_rate": 8.91638463316892e-05, "loss": 2.9797, "step": 14121 }, { "epoch": 0.8766528027810541, "grad_norm": 0.4398454398466616, "learning_rate": 8.916160104263825e-05, "loss": 3.0505, "step": 14122 }, { "epoch": 0.876714879880812, "grad_norm": 0.5201696617939009, "learning_rate": 8.915935554927168e-05, "loss": 2.9813, "step": 14123 }, { "epoch": 0.8767769569805699, "grad_norm": 0.5396659900683819, "learning_rate": 8.915710985160126e-05, "loss": 2.976, "step": 14124 }, { "epoch": 0.8768390340803278, "grad_norm": 0.4599201645972418, "learning_rate": 8.91548639496387e-05, "loss": 3.038, "step": 14125 }, { "epoch": 0.8769011111800856, "grad_norm": 0.543242762740792, "learning_rate": 8.91526178433957e-05, "loss": 2.9665, "step": 14126 }, { "epoch": 0.8769631882798435, "grad_norm": 0.5299051878353217, "learning_rate": 8.9150371532884e-05, "loss": 2.988, "step": 14127 }, { "epoch": 0.8770252653796015, "grad_norm": 0.5734390622567248, "learning_rate": 8.914812501811532e-05, "loss": 2.9492, "step": 14128 }, { "epoch": 0.8770873424793594, "grad_norm": 0.5592150134387851, "learning_rate": 8.914587829910135e-05, "loss": 2.9795, "step": 14129 }, { "epoch": 0.8771494195791173, "grad_norm": 0.5500577137750446, "learning_rate": 8.914363137585382e-05, "loss": 2.9939, "step": 14130 }, { "epoch": 0.8772114966788752, "grad_norm": 0.537309172489219, "learning_rate": 8.914138424838447e-05, "loss": 2.9353, "step": 14131 }, { "epoch": 0.877273573778633, "grad_norm": 0.5521409586456264, "learning_rate": 8.913913691670503e-05, "loss": 2.9689, "step": 14132 }, { "epoch": 0.8773356508783909, "grad_norm": 0.5429861491317769, "learning_rate": 8.91368893808272e-05, "loss": 2.8351, "step": 14133 }, { "epoch": 0.8773977279781489, "grad_norm": 0.5276989341146762, "learning_rate": 8.91346416407627e-05, "loss": 2.9526, "step": 14134 }, { "epoch": 0.8774598050779068, "grad_norm": 0.524118237019418, "learning_rate": 8.913239369652331e-05, "loss": 3.0054, "step": 14135 }, { "epoch": 0.8775218821776647, "grad_norm": 0.5118369337280518, "learning_rate": 8.91301455481207e-05, "loss": 2.9822, "step": 14136 }, { "epoch": 0.8775839592774226, "grad_norm": 0.5740956544309299, "learning_rate": 8.912789719556662e-05, "loss": 2.9886, "step": 14137 }, { "epoch": 0.8776460363771804, "grad_norm": 0.5288672586378019, "learning_rate": 8.912564863887282e-05, "loss": 2.9166, "step": 14138 }, { "epoch": 0.8777081134769383, "grad_norm": 0.596970906965681, "learning_rate": 8.9123399878051e-05, "loss": 3.0247, "step": 14139 }, { "epoch": 0.8777701905766963, "grad_norm": 0.5154037000689846, "learning_rate": 8.91211509131129e-05, "loss": 3.0004, "step": 14140 }, { "epoch": 0.8778322676764542, "grad_norm": 0.5655418640496493, "learning_rate": 8.911890174407027e-05, "loss": 3.0637, "step": 14141 }, { "epoch": 0.8778943447762121, "grad_norm": 0.5450676070285613, "learning_rate": 8.911665237093483e-05, "loss": 3.0033, "step": 14142 }, { "epoch": 0.87795642187597, "grad_norm": 0.5507364187189423, "learning_rate": 8.911440279371832e-05, "loss": 3.0222, "step": 14143 }, { "epoch": 0.8780184989757278, "grad_norm": 0.6096831545144551, "learning_rate": 8.911215301243246e-05, "loss": 2.918, "step": 14144 }, { "epoch": 0.8780805760754857, "grad_norm": 0.45512905572594425, "learning_rate": 8.910990302708901e-05, "loss": 2.8666, "step": 14145 }, { "epoch": 0.8781426531752436, "grad_norm": 0.5485016744513325, "learning_rate": 8.910765283769972e-05, "loss": 2.9846, "step": 14146 }, { "epoch": 0.8782047302750016, "grad_norm": 0.53452467021444, "learning_rate": 8.91054024442763e-05, "loss": 3.0541, "step": 14147 }, { "epoch": 0.8782668073747595, "grad_norm": 0.49065492077117695, "learning_rate": 8.91031518468305e-05, "loss": 3.0281, "step": 14148 }, { "epoch": 0.8783288844745174, "grad_norm": 0.4597856061899275, "learning_rate": 8.910090104537407e-05, "loss": 2.9492, "step": 14149 }, { "epoch": 0.8783909615742752, "grad_norm": 0.4983708543377804, "learning_rate": 8.909865003991874e-05, "loss": 3.044, "step": 14150 }, { "epoch": 0.8784530386740331, "grad_norm": 0.4686605368118816, "learning_rate": 8.909639883047626e-05, "loss": 2.9268, "step": 14151 }, { "epoch": 0.878515115773791, "grad_norm": 0.46341410142474043, "learning_rate": 8.909414741705838e-05, "loss": 2.971, "step": 14152 }, { "epoch": 0.878577192873549, "grad_norm": 0.5033836316035776, "learning_rate": 8.909189579967683e-05, "loss": 3.0157, "step": 14153 }, { "epoch": 0.8786392699733069, "grad_norm": 0.49258986178519304, "learning_rate": 8.908964397834339e-05, "loss": 2.9802, "step": 14154 }, { "epoch": 0.8787013470730648, "grad_norm": 0.5132350442140872, "learning_rate": 8.908739195306978e-05, "loss": 3.0653, "step": 14155 }, { "epoch": 0.8787634241728226, "grad_norm": 0.5434536860963703, "learning_rate": 8.908513972386776e-05, "loss": 2.9102, "step": 14156 }, { "epoch": 0.8788255012725805, "grad_norm": 0.41943123120673936, "learning_rate": 8.908288729074908e-05, "loss": 2.9915, "step": 14157 }, { "epoch": 0.8788875783723384, "grad_norm": 0.567464589514029, "learning_rate": 8.908063465372547e-05, "loss": 3.0016, "step": 14158 }, { "epoch": 0.8789496554720964, "grad_norm": 0.4921893467944178, "learning_rate": 8.907838181280872e-05, "loss": 2.9864, "step": 14159 }, { "epoch": 0.8790117325718543, "grad_norm": 0.46799775799495463, "learning_rate": 8.907612876801056e-05, "loss": 2.9145, "step": 14160 }, { "epoch": 0.8790738096716122, "grad_norm": 0.47799652904283774, "learning_rate": 8.907387551934277e-05, "loss": 2.9712, "step": 14161 }, { "epoch": 0.87913588677137, "grad_norm": 0.48648784113663507, "learning_rate": 8.907162206681705e-05, "loss": 2.9487, "step": 14162 }, { "epoch": 0.8791979638711279, "grad_norm": 0.4877813145006147, "learning_rate": 8.906936841044521e-05, "loss": 3.0714, "step": 14163 }, { "epoch": 0.8792600409708858, "grad_norm": 0.4762392980519541, "learning_rate": 8.9067114550239e-05, "loss": 2.8961, "step": 14164 }, { "epoch": 0.8793221180706438, "grad_norm": 0.4973335543350559, "learning_rate": 8.906486048621016e-05, "loss": 3.0213, "step": 14165 }, { "epoch": 0.8793841951704017, "grad_norm": 0.5664487362416114, "learning_rate": 8.906260621837046e-05, "loss": 2.9941, "step": 14166 }, { "epoch": 0.8794462722701596, "grad_norm": 0.46939557443116436, "learning_rate": 8.906035174673167e-05, "loss": 2.9838, "step": 14167 }, { "epoch": 0.8795083493699174, "grad_norm": 0.4831833794521364, "learning_rate": 8.905809707130553e-05, "loss": 2.9278, "step": 14168 }, { "epoch": 0.8795704264696753, "grad_norm": 0.49782240053761695, "learning_rate": 8.905584219210381e-05, "loss": 3.0797, "step": 14169 }, { "epoch": 0.8796325035694332, "grad_norm": 0.572704732593787, "learning_rate": 8.905358710913829e-05, "loss": 2.9493, "step": 14170 }, { "epoch": 0.8796945806691912, "grad_norm": 0.5010095775054273, "learning_rate": 8.905133182242072e-05, "loss": 2.9556, "step": 14171 }, { "epoch": 0.8797566577689491, "grad_norm": 0.5855374305579286, "learning_rate": 8.904907633196288e-05, "loss": 2.9719, "step": 14172 }, { "epoch": 0.879818734868707, "grad_norm": 0.5579711271783915, "learning_rate": 8.904682063777652e-05, "loss": 2.981, "step": 14173 }, { "epoch": 0.8798808119684648, "grad_norm": 0.5377482827751943, "learning_rate": 8.904456473987343e-05, "loss": 3.0505, "step": 14174 }, { "epoch": 0.8799428890682227, "grad_norm": 0.5944514899069049, "learning_rate": 8.904230863826535e-05, "loss": 2.9948, "step": 14175 }, { "epoch": 0.8800049661679806, "grad_norm": 0.4469021815589671, "learning_rate": 8.904005233296408e-05, "loss": 2.8701, "step": 14176 }, { "epoch": 0.8800670432677385, "grad_norm": 0.594982148672032, "learning_rate": 8.90377958239814e-05, "loss": 2.9617, "step": 14177 }, { "epoch": 0.8801291203674965, "grad_norm": 0.5474554251462387, "learning_rate": 8.903553911132903e-05, "loss": 2.907, "step": 14178 }, { "epoch": 0.8801911974672544, "grad_norm": 0.5060098785031689, "learning_rate": 8.903328219501879e-05, "loss": 2.8959, "step": 14179 }, { "epoch": 0.8802532745670122, "grad_norm": 0.5294282475684019, "learning_rate": 8.903102507506244e-05, "loss": 3.0276, "step": 14180 }, { "epoch": 0.8803153516667701, "grad_norm": 0.4581707337256647, "learning_rate": 8.902876775147175e-05, "loss": 2.9238, "step": 14181 }, { "epoch": 0.880377428766528, "grad_norm": 0.5006976029497339, "learning_rate": 8.902651022425852e-05, "loss": 3.0485, "step": 14182 }, { "epoch": 0.8804395058662859, "grad_norm": 0.46928316630063466, "learning_rate": 8.902425249343449e-05, "loss": 2.9962, "step": 14183 }, { "epoch": 0.8805015829660439, "grad_norm": 0.4789884761148795, "learning_rate": 8.902199455901148e-05, "loss": 2.8801, "step": 14184 }, { "epoch": 0.8805636600658018, "grad_norm": 0.5009502679122206, "learning_rate": 8.901973642100124e-05, "loss": 2.9311, "step": 14185 }, { "epoch": 0.8806257371655596, "grad_norm": 0.45698165624476744, "learning_rate": 8.901747807941557e-05, "loss": 3.0583, "step": 14186 }, { "epoch": 0.8806878142653175, "grad_norm": 0.5661619777769764, "learning_rate": 8.901521953426624e-05, "loss": 2.9465, "step": 14187 }, { "epoch": 0.8807498913650754, "grad_norm": 0.5115026935159683, "learning_rate": 8.901296078556503e-05, "loss": 3.0206, "step": 14188 }, { "epoch": 0.8808119684648333, "grad_norm": 0.4941385442516561, "learning_rate": 8.901070183332375e-05, "loss": 2.9785, "step": 14189 }, { "epoch": 0.8808740455645913, "grad_norm": 0.5214822319072449, "learning_rate": 8.900844267755417e-05, "loss": 2.8981, "step": 14190 }, { "epoch": 0.8809361226643492, "grad_norm": 0.5853935005201875, "learning_rate": 8.900618331826807e-05, "loss": 3.0276, "step": 14191 }, { "epoch": 0.880998199764107, "grad_norm": 0.563209113212095, "learning_rate": 8.900392375547724e-05, "loss": 2.9496, "step": 14192 }, { "epoch": 0.8810602768638649, "grad_norm": 0.5463124515353187, "learning_rate": 8.900166398919346e-05, "loss": 3.0027, "step": 14193 }, { "epoch": 0.8811223539636228, "grad_norm": 0.5388441611137254, "learning_rate": 8.899940401942854e-05, "loss": 2.9477, "step": 14194 }, { "epoch": 0.8811844310633807, "grad_norm": 0.5559834934827231, "learning_rate": 8.899714384619426e-05, "loss": 2.9476, "step": 14195 }, { "epoch": 0.8812465081631387, "grad_norm": 0.552523326683419, "learning_rate": 8.899488346950243e-05, "loss": 2.9817, "step": 14196 }, { "epoch": 0.8813085852628966, "grad_norm": 0.4773096418915798, "learning_rate": 8.899262288936481e-05, "loss": 2.9652, "step": 14197 }, { "epoch": 0.8813706623626544, "grad_norm": 0.5694437072201224, "learning_rate": 8.899036210579321e-05, "loss": 2.9455, "step": 14198 }, { "epoch": 0.8814327394624123, "grad_norm": 0.5890290147082331, "learning_rate": 8.898810111879943e-05, "loss": 2.9686, "step": 14199 }, { "epoch": 0.8814948165621702, "grad_norm": 0.5075175088765448, "learning_rate": 8.898583992839527e-05, "loss": 2.9606, "step": 14200 }, { "epoch": 0.8815568936619281, "grad_norm": 0.552701591894412, "learning_rate": 8.898357853459252e-05, "loss": 2.934, "step": 14201 }, { "epoch": 0.881618970761686, "grad_norm": 0.5286564912019204, "learning_rate": 8.898131693740295e-05, "loss": 3.0332, "step": 14202 }, { "epoch": 0.881681047861444, "grad_norm": 0.5928531197248905, "learning_rate": 8.89790551368384e-05, "loss": 2.9164, "step": 14203 }, { "epoch": 0.8817431249612018, "grad_norm": 0.5608206954097058, "learning_rate": 8.897679313291068e-05, "loss": 3.0179, "step": 14204 }, { "epoch": 0.8818052020609597, "grad_norm": 0.5865885614246248, "learning_rate": 8.897453092563154e-05, "loss": 2.8851, "step": 14205 }, { "epoch": 0.8818672791607176, "grad_norm": 0.5436100363094019, "learning_rate": 8.897226851501281e-05, "loss": 2.9422, "step": 14206 }, { "epoch": 0.8819293562604755, "grad_norm": 0.5019500376090605, "learning_rate": 8.897000590106631e-05, "loss": 2.9559, "step": 14207 }, { "epoch": 0.8819914333602334, "grad_norm": 0.6114892473047426, "learning_rate": 8.896774308380382e-05, "loss": 3.0524, "step": 14208 }, { "epoch": 0.8820535104599914, "grad_norm": 0.44302996190641236, "learning_rate": 8.896548006323717e-05, "loss": 3.0277, "step": 14209 }, { "epoch": 0.8821155875597492, "grad_norm": 0.5484990392431366, "learning_rate": 8.896321683937814e-05, "loss": 3.0278, "step": 14210 }, { "epoch": 0.8821776646595071, "grad_norm": 0.5037599691346395, "learning_rate": 8.896095341223855e-05, "loss": 2.9981, "step": 14211 }, { "epoch": 0.882239741759265, "grad_norm": 0.49247482027274897, "learning_rate": 8.895868978183019e-05, "loss": 2.9115, "step": 14212 }, { "epoch": 0.8823018188590229, "grad_norm": 0.48599421670007975, "learning_rate": 8.895642594816492e-05, "loss": 2.9835, "step": 14213 }, { "epoch": 0.8823638959587808, "grad_norm": 0.5503411394646199, "learning_rate": 8.89541619112545e-05, "loss": 3.0426, "step": 14214 }, { "epoch": 0.8824259730585388, "grad_norm": 0.4966177988440025, "learning_rate": 8.895189767111077e-05, "loss": 2.9164, "step": 14215 }, { "epoch": 0.8824880501582966, "grad_norm": 0.4756847333806326, "learning_rate": 8.894963322774553e-05, "loss": 2.9966, "step": 14216 }, { "epoch": 0.8825501272580545, "grad_norm": 0.6450632154181517, "learning_rate": 8.894736858117059e-05, "loss": 3.0264, "step": 14217 }, { "epoch": 0.8826122043578124, "grad_norm": 0.5086783729856478, "learning_rate": 8.894510373139778e-05, "loss": 2.9423, "step": 14218 }, { "epoch": 0.8826742814575703, "grad_norm": 0.5469090314903039, "learning_rate": 8.894283867843892e-05, "loss": 3.017, "step": 14219 }, { "epoch": 0.8827363585573282, "grad_norm": 0.4999602123165527, "learning_rate": 8.89405734223058e-05, "loss": 2.9694, "step": 14220 }, { "epoch": 0.8827984356570862, "grad_norm": 0.49402258145308015, "learning_rate": 8.893830796301024e-05, "loss": 2.912, "step": 14221 }, { "epoch": 0.882860512756844, "grad_norm": 0.502138986576483, "learning_rate": 8.893604230056411e-05, "loss": 3.0554, "step": 14222 }, { "epoch": 0.8829225898566019, "grad_norm": 0.48744389260027027, "learning_rate": 8.893377643497917e-05, "loss": 2.9737, "step": 14223 }, { "epoch": 0.8829846669563598, "grad_norm": 0.49950171019668993, "learning_rate": 8.893151036626728e-05, "loss": 3.005, "step": 14224 }, { "epoch": 0.8830467440561177, "grad_norm": 0.4661654587207092, "learning_rate": 8.892924409444026e-05, "loss": 3.0611, "step": 14225 }, { "epoch": 0.8831088211558756, "grad_norm": 0.5000958946776544, "learning_rate": 8.89269776195099e-05, "loss": 3.0263, "step": 14226 }, { "epoch": 0.8831708982556336, "grad_norm": 0.5448629149120425, "learning_rate": 8.892471094148807e-05, "loss": 2.9115, "step": 14227 }, { "epoch": 0.8832329753553914, "grad_norm": 0.46808376650150396, "learning_rate": 8.892244406038655e-05, "loss": 2.9752, "step": 14228 }, { "epoch": 0.8832950524551493, "grad_norm": 0.4420218437113636, "learning_rate": 8.89201769762172e-05, "loss": 2.9733, "step": 14229 }, { "epoch": 0.8833571295549072, "grad_norm": 0.489504510202174, "learning_rate": 8.891790968899185e-05, "loss": 3.0197, "step": 14230 }, { "epoch": 0.8834192066546651, "grad_norm": 0.44247139218630827, "learning_rate": 8.89156421987223e-05, "loss": 2.9212, "step": 14231 }, { "epoch": 0.883481283754423, "grad_norm": 0.4965262742749669, "learning_rate": 8.891337450542042e-05, "loss": 2.9322, "step": 14232 }, { "epoch": 0.883543360854181, "grad_norm": 0.5022663371237867, "learning_rate": 8.8911106609098e-05, "loss": 3.0122, "step": 14233 }, { "epoch": 0.8836054379539388, "grad_norm": 0.4332415378831556, "learning_rate": 8.89088385097669e-05, "loss": 2.9413, "step": 14234 }, { "epoch": 0.8836675150536967, "grad_norm": 0.483337105607516, "learning_rate": 8.890657020743893e-05, "loss": 2.987, "step": 14235 }, { "epoch": 0.8837295921534546, "grad_norm": 0.4618647672064254, "learning_rate": 8.890430170212596e-05, "loss": 3.0003, "step": 14236 }, { "epoch": 0.8837916692532125, "grad_norm": 0.4783242913575611, "learning_rate": 8.89020329938398e-05, "loss": 2.9793, "step": 14237 }, { "epoch": 0.8838537463529704, "grad_norm": 0.4874783437760438, "learning_rate": 8.889976408259226e-05, "loss": 2.9935, "step": 14238 }, { "epoch": 0.8839158234527282, "grad_norm": 0.4999895392754152, "learning_rate": 8.889749496839524e-05, "loss": 2.9619, "step": 14239 }, { "epoch": 0.8839779005524862, "grad_norm": 0.5405866388812126, "learning_rate": 8.889522565126053e-05, "loss": 3.099, "step": 14240 }, { "epoch": 0.8840399776522441, "grad_norm": 0.5391613966297598, "learning_rate": 8.889295613119999e-05, "loss": 2.9961, "step": 14241 }, { "epoch": 0.884102054752002, "grad_norm": 0.5540871124481024, "learning_rate": 8.889068640822546e-05, "loss": 3.0515, "step": 14242 }, { "epoch": 0.8841641318517599, "grad_norm": 0.5404693379416884, "learning_rate": 8.888841648234878e-05, "loss": 2.9746, "step": 14243 }, { "epoch": 0.8842262089515178, "grad_norm": 0.4997595268379568, "learning_rate": 8.888614635358178e-05, "loss": 2.9357, "step": 14244 }, { "epoch": 0.8842882860512756, "grad_norm": 0.5108328569211911, "learning_rate": 8.888387602193632e-05, "loss": 3.045, "step": 14245 }, { "epoch": 0.8843503631510335, "grad_norm": 0.503144686769857, "learning_rate": 8.888160548742424e-05, "loss": 2.9423, "step": 14246 }, { "epoch": 0.8844124402507915, "grad_norm": 0.5325989146887292, "learning_rate": 8.887933475005739e-05, "loss": 2.9293, "step": 14247 }, { "epoch": 0.8844745173505494, "grad_norm": 0.4980494031654966, "learning_rate": 8.88770638098476e-05, "loss": 3.0317, "step": 14248 }, { "epoch": 0.8845365944503073, "grad_norm": 0.4679643245481402, "learning_rate": 8.887479266680673e-05, "loss": 2.9227, "step": 14249 }, { "epoch": 0.8845986715500652, "grad_norm": 0.4640604520039691, "learning_rate": 8.887252132094666e-05, "loss": 2.9707, "step": 14250 }, { "epoch": 0.884660748649823, "grad_norm": 0.48091480050358093, "learning_rate": 8.887024977227917e-05, "loss": 2.9653, "step": 14251 }, { "epoch": 0.8847228257495809, "grad_norm": 0.520388963596754, "learning_rate": 8.886797802081616e-05, "loss": 2.9872, "step": 14252 }, { "epoch": 0.8847849028493389, "grad_norm": 0.5490054458389753, "learning_rate": 8.88657060665695e-05, "loss": 3.0351, "step": 14253 }, { "epoch": 0.8848469799490968, "grad_norm": 0.4614748328239361, "learning_rate": 8.886343390955098e-05, "loss": 3.0273, "step": 14254 }, { "epoch": 0.8849090570488547, "grad_norm": 0.5207958176135082, "learning_rate": 8.886116154977252e-05, "loss": 2.9511, "step": 14255 }, { "epoch": 0.8849711341486126, "grad_norm": 0.4914739674303181, "learning_rate": 8.885888898724594e-05, "loss": 2.9051, "step": 14256 }, { "epoch": 0.8850332112483704, "grad_norm": 0.46227493867296615, "learning_rate": 8.885661622198309e-05, "loss": 2.956, "step": 14257 }, { "epoch": 0.8850952883481283, "grad_norm": 0.5098323851384631, "learning_rate": 8.885434325399585e-05, "loss": 3.0038, "step": 14258 }, { "epoch": 0.8851573654478863, "grad_norm": 0.4610717222406563, "learning_rate": 8.885207008329607e-05, "loss": 2.9044, "step": 14259 }, { "epoch": 0.8852194425476442, "grad_norm": 0.48882198147588735, "learning_rate": 8.884979670989561e-05, "loss": 2.97, "step": 14260 }, { "epoch": 0.8852815196474021, "grad_norm": 0.49107366505646155, "learning_rate": 8.884752313380632e-05, "loss": 2.975, "step": 14261 }, { "epoch": 0.88534359674716, "grad_norm": 0.44484108517072707, "learning_rate": 8.884524935504008e-05, "loss": 2.8773, "step": 14262 }, { "epoch": 0.8854056738469178, "grad_norm": 0.4825608057289534, "learning_rate": 8.884297537360874e-05, "loss": 2.9166, "step": 14263 }, { "epoch": 0.8854677509466757, "grad_norm": 0.48019056873465565, "learning_rate": 8.884070118952416e-05, "loss": 2.9928, "step": 14264 }, { "epoch": 0.8855298280464337, "grad_norm": 0.46444850434875484, "learning_rate": 8.883842680279824e-05, "loss": 2.9333, "step": 14265 }, { "epoch": 0.8855919051461916, "grad_norm": 0.49082734618666923, "learning_rate": 8.883615221344279e-05, "loss": 3.0325, "step": 14266 }, { "epoch": 0.8856539822459495, "grad_norm": 0.5015389778587196, "learning_rate": 8.883387742146972e-05, "loss": 2.9758, "step": 14267 }, { "epoch": 0.8857160593457074, "grad_norm": 0.5006062289563317, "learning_rate": 8.883160242689089e-05, "loss": 2.9766, "step": 14268 }, { "epoch": 0.8857781364454652, "grad_norm": 0.5938401153843467, "learning_rate": 8.882932722971813e-05, "loss": 2.9261, "step": 14269 }, { "epoch": 0.8858402135452231, "grad_norm": 0.5213576602215351, "learning_rate": 8.882705182996338e-05, "loss": 2.9349, "step": 14270 }, { "epoch": 0.885902290644981, "grad_norm": 0.49723378549244834, "learning_rate": 8.882477622763846e-05, "loss": 2.9734, "step": 14271 }, { "epoch": 0.885964367744739, "grad_norm": 0.5721322903542638, "learning_rate": 8.882250042275526e-05, "loss": 3.0337, "step": 14272 }, { "epoch": 0.8860264448444969, "grad_norm": 0.4748207688991822, "learning_rate": 8.882022441532564e-05, "loss": 2.9688, "step": 14273 }, { "epoch": 0.8860885219442548, "grad_norm": 0.5313274944107045, "learning_rate": 8.881794820536152e-05, "loss": 2.9042, "step": 14274 }, { "epoch": 0.8861505990440126, "grad_norm": 0.5295349936804606, "learning_rate": 8.881567179287472e-05, "loss": 2.9754, "step": 14275 }, { "epoch": 0.8862126761437705, "grad_norm": 0.4932185681126485, "learning_rate": 8.881339517787712e-05, "loss": 2.8894, "step": 14276 }, { "epoch": 0.8862747532435284, "grad_norm": 0.532176780122321, "learning_rate": 8.881111836038063e-05, "loss": 2.8893, "step": 14277 }, { "epoch": 0.8863368303432864, "grad_norm": 0.4884310073079322, "learning_rate": 8.880884134039711e-05, "loss": 2.9836, "step": 14278 }, { "epoch": 0.8863989074430443, "grad_norm": 0.5666851277943857, "learning_rate": 8.880656411793845e-05, "loss": 3.0102, "step": 14279 }, { "epoch": 0.8864609845428022, "grad_norm": 0.5182755575709853, "learning_rate": 8.880428669301652e-05, "loss": 3.0183, "step": 14280 }, { "epoch": 0.88652306164256, "grad_norm": 0.527862660862929, "learning_rate": 8.880200906564322e-05, "loss": 3.0119, "step": 14281 }, { "epoch": 0.8865851387423179, "grad_norm": 0.5815755096735039, "learning_rate": 8.87997312358304e-05, "loss": 3.0385, "step": 14282 }, { "epoch": 0.8866472158420758, "grad_norm": 0.4952138020228595, "learning_rate": 8.879745320358997e-05, "loss": 2.9439, "step": 14283 }, { "epoch": 0.8867092929418338, "grad_norm": 0.49909102249092896, "learning_rate": 8.879517496893381e-05, "loss": 3.0117, "step": 14284 }, { "epoch": 0.8867713700415917, "grad_norm": 0.5690727345249621, "learning_rate": 8.87928965318738e-05, "loss": 2.9581, "step": 14285 }, { "epoch": 0.8868334471413496, "grad_norm": 0.5252240849419917, "learning_rate": 8.879061789242183e-05, "loss": 3.0253, "step": 14286 }, { "epoch": 0.8868955242411074, "grad_norm": 0.5298367664143536, "learning_rate": 8.87883390505898e-05, "loss": 3.0071, "step": 14287 }, { "epoch": 0.8869576013408653, "grad_norm": 0.5060369408225708, "learning_rate": 8.878606000638959e-05, "loss": 2.9998, "step": 14288 }, { "epoch": 0.8870196784406232, "grad_norm": 0.5052895005558937, "learning_rate": 8.878378075983307e-05, "loss": 2.981, "step": 14289 }, { "epoch": 0.8870817555403812, "grad_norm": 0.5761990556932952, "learning_rate": 8.878150131093217e-05, "loss": 2.9773, "step": 14290 }, { "epoch": 0.8871438326401391, "grad_norm": 0.49046036272269417, "learning_rate": 8.877922165969876e-05, "loss": 2.9936, "step": 14291 }, { "epoch": 0.887205909739897, "grad_norm": 0.5176384944616876, "learning_rate": 8.877694180614473e-05, "loss": 3.0013, "step": 14292 }, { "epoch": 0.8872679868396548, "grad_norm": 0.5097411869666061, "learning_rate": 8.877466175028197e-05, "loss": 3.0838, "step": 14293 }, { "epoch": 0.8873300639394127, "grad_norm": 0.5108778355843869, "learning_rate": 8.877238149212241e-05, "loss": 2.9635, "step": 14294 }, { "epoch": 0.8873921410391706, "grad_norm": 0.47541633730761895, "learning_rate": 8.877010103167792e-05, "loss": 2.9361, "step": 14295 }, { "epoch": 0.8874542181389286, "grad_norm": 0.5103123681558227, "learning_rate": 8.876782036896038e-05, "loss": 2.9581, "step": 14296 }, { "epoch": 0.8875162952386865, "grad_norm": 0.5040613275882072, "learning_rate": 8.876553950398173e-05, "loss": 3.0326, "step": 14297 }, { "epoch": 0.8875783723384444, "grad_norm": 0.5206117508336946, "learning_rate": 8.876325843675385e-05, "loss": 3.0563, "step": 14298 }, { "epoch": 0.8876404494382022, "grad_norm": 0.4719004302088156, "learning_rate": 8.876097716728864e-05, "loss": 3.0118, "step": 14299 }, { "epoch": 0.8877025265379601, "grad_norm": 0.4447704328729085, "learning_rate": 8.875869569559799e-05, "loss": 2.9659, "step": 14300 }, { "epoch": 0.887764603637718, "grad_norm": 0.5103936255048733, "learning_rate": 8.875641402169383e-05, "loss": 2.9828, "step": 14301 }, { "epoch": 0.887826680737476, "grad_norm": 0.48575019967864796, "learning_rate": 8.875413214558805e-05, "loss": 3.0309, "step": 14302 }, { "epoch": 0.8878887578372339, "grad_norm": 0.4749790055253092, "learning_rate": 8.875185006729254e-05, "loss": 3.0385, "step": 14303 }, { "epoch": 0.8879508349369918, "grad_norm": 0.45804556749974784, "learning_rate": 8.874956778681922e-05, "loss": 3.0171, "step": 14304 }, { "epoch": 0.8880129120367496, "grad_norm": 0.456904597543523, "learning_rate": 8.874728530418002e-05, "loss": 2.9183, "step": 14305 }, { "epoch": 0.8880749891365075, "grad_norm": 0.4417241732328863, "learning_rate": 8.874500261938681e-05, "loss": 2.9028, "step": 14306 }, { "epoch": 0.8881370662362654, "grad_norm": 0.4905151080109759, "learning_rate": 8.874271973245152e-05, "loss": 2.9842, "step": 14307 }, { "epoch": 0.8881991433360233, "grad_norm": 0.43213987174911384, "learning_rate": 8.874043664338605e-05, "loss": 2.9908, "step": 14308 }, { "epoch": 0.8882612204357813, "grad_norm": 0.47136094596020084, "learning_rate": 8.873815335220232e-05, "loss": 2.909, "step": 14309 }, { "epoch": 0.8883232975355392, "grad_norm": 0.4715418485868262, "learning_rate": 8.873586985891225e-05, "loss": 2.9241, "step": 14310 }, { "epoch": 0.888385374635297, "grad_norm": 0.45346725626503565, "learning_rate": 8.873358616352773e-05, "loss": 2.9458, "step": 14311 }, { "epoch": 0.8884474517350549, "grad_norm": 0.5022795094754369, "learning_rate": 8.87313022660607e-05, "loss": 2.9945, "step": 14312 }, { "epoch": 0.8885095288348128, "grad_norm": 0.46170424682249556, "learning_rate": 8.872901816652305e-05, "loss": 3.0092, "step": 14313 }, { "epoch": 0.8885716059345707, "grad_norm": 0.48928612115095244, "learning_rate": 8.872673386492672e-05, "loss": 2.8852, "step": 14314 }, { "epoch": 0.8886336830343287, "grad_norm": 0.46749219622703075, "learning_rate": 8.872444936128362e-05, "loss": 2.9917, "step": 14315 }, { "epoch": 0.8886957601340866, "grad_norm": 0.4714475577832996, "learning_rate": 8.872216465560566e-05, "loss": 2.8024, "step": 14316 }, { "epoch": 0.8887578372338444, "grad_norm": 0.4554617087865216, "learning_rate": 8.871987974790476e-05, "loss": 2.9564, "step": 14317 }, { "epoch": 0.8888199143336023, "grad_norm": 0.5436728598148172, "learning_rate": 8.871759463819286e-05, "loss": 2.8894, "step": 14318 }, { "epoch": 0.8888819914333602, "grad_norm": 0.5215892329144683, "learning_rate": 8.871530932648186e-05, "loss": 3.0746, "step": 14319 }, { "epoch": 0.8889440685331181, "grad_norm": 0.46386599891911867, "learning_rate": 8.871302381278371e-05, "loss": 3.1026, "step": 14320 }, { "epoch": 0.889006145632876, "grad_norm": 0.4381324420247925, "learning_rate": 8.871073809711031e-05, "loss": 3.0003, "step": 14321 }, { "epoch": 0.889068222732634, "grad_norm": 0.48681989570981715, "learning_rate": 8.870845217947358e-05, "loss": 2.9506, "step": 14322 }, { "epoch": 0.8891302998323918, "grad_norm": 0.44692475435404166, "learning_rate": 8.870616605988547e-05, "loss": 2.9583, "step": 14323 }, { "epoch": 0.8891923769321497, "grad_norm": 0.43416183647768647, "learning_rate": 8.87038797383579e-05, "loss": 2.9141, "step": 14324 }, { "epoch": 0.8892544540319076, "grad_norm": 0.4965420297060795, "learning_rate": 8.87015932149028e-05, "loss": 2.8648, "step": 14325 }, { "epoch": 0.8893165311316655, "grad_norm": 0.4463706912257645, "learning_rate": 8.869930648953206e-05, "loss": 2.9008, "step": 14326 }, { "epoch": 0.8893786082314235, "grad_norm": 0.4962106702098594, "learning_rate": 8.869701956225767e-05, "loss": 2.9678, "step": 14327 }, { "epoch": 0.8894406853311814, "grad_norm": 0.4690950871837917, "learning_rate": 8.869473243309151e-05, "loss": 2.9644, "step": 14328 }, { "epoch": 0.8895027624309392, "grad_norm": 0.5271692340447406, "learning_rate": 8.869244510204558e-05, "loss": 2.9841, "step": 14329 }, { "epoch": 0.8895648395306971, "grad_norm": 0.49169000580991523, "learning_rate": 8.869015756913174e-05, "loss": 2.9797, "step": 14330 }, { "epoch": 0.889626916630455, "grad_norm": 0.5642636519011406, "learning_rate": 8.868786983436195e-05, "loss": 2.9698, "step": 14331 }, { "epoch": 0.8896889937302129, "grad_norm": 0.5579415628843594, "learning_rate": 8.868558189774818e-05, "loss": 3.0611, "step": 14332 }, { "epoch": 0.8897510708299708, "grad_norm": 0.5476459246436175, "learning_rate": 8.868329375930232e-05, "loss": 2.943, "step": 14333 }, { "epoch": 0.8898131479297288, "grad_norm": 0.4533315220818847, "learning_rate": 8.868100541903633e-05, "loss": 2.928, "step": 14334 }, { "epoch": 0.8898752250294866, "grad_norm": 0.4929095527423638, "learning_rate": 8.867871687696214e-05, "loss": 3.0346, "step": 14335 }, { "epoch": 0.8899373021292445, "grad_norm": 0.5723475114478528, "learning_rate": 8.86764281330917e-05, "loss": 2.9344, "step": 14336 }, { "epoch": 0.8899993792290024, "grad_norm": 0.5450946711636483, "learning_rate": 8.867413918743694e-05, "loss": 2.9898, "step": 14337 }, { "epoch": 0.8900614563287603, "grad_norm": 0.47304698462134553, "learning_rate": 8.86718500400098e-05, "loss": 3.0245, "step": 14338 }, { "epoch": 0.8901235334285182, "grad_norm": 0.5197683094665753, "learning_rate": 8.866956069082224e-05, "loss": 3.0441, "step": 14339 }, { "epoch": 0.8901856105282762, "grad_norm": 0.547768544045534, "learning_rate": 8.86672711398862e-05, "loss": 2.9523, "step": 14340 }, { "epoch": 0.890247687628034, "grad_norm": 0.4552494921579309, "learning_rate": 8.86649813872136e-05, "loss": 2.9287, "step": 14341 }, { "epoch": 0.8903097647277919, "grad_norm": 0.5334481036356404, "learning_rate": 8.866269143281642e-05, "loss": 2.9767, "step": 14342 }, { "epoch": 0.8903718418275498, "grad_norm": 0.4762237001150905, "learning_rate": 8.86604012767066e-05, "loss": 3.0081, "step": 14343 }, { "epoch": 0.8904339189273077, "grad_norm": 0.5313683826713803, "learning_rate": 8.865811091889606e-05, "loss": 3.0107, "step": 14344 }, { "epoch": 0.8904959960270656, "grad_norm": 0.5284347551253106, "learning_rate": 8.865582035939678e-05, "loss": 3.0108, "step": 14345 }, { "epoch": 0.8905580731268236, "grad_norm": 0.5075513608433797, "learning_rate": 8.865352959822073e-05, "loss": 2.9321, "step": 14346 }, { "epoch": 0.8906201502265814, "grad_norm": 0.5371737354905115, "learning_rate": 8.865123863537981e-05, "loss": 2.9277, "step": 14347 }, { "epoch": 0.8906822273263393, "grad_norm": 0.5180487166826317, "learning_rate": 8.864894747088599e-05, "loss": 2.979, "step": 14348 }, { "epoch": 0.8907443044260972, "grad_norm": 0.5184369716906025, "learning_rate": 8.864665610475122e-05, "loss": 3.0239, "step": 14349 }, { "epoch": 0.8908063815258551, "grad_norm": 0.5138800601659054, "learning_rate": 8.864436453698748e-05, "loss": 2.9685, "step": 14350 }, { "epoch": 0.890868458625613, "grad_norm": 0.5299976397407806, "learning_rate": 8.86420727676067e-05, "loss": 2.9625, "step": 14351 }, { "epoch": 0.890930535725371, "grad_norm": 0.5134200432492023, "learning_rate": 8.863978079662086e-05, "loss": 2.9951, "step": 14352 }, { "epoch": 0.8909926128251288, "grad_norm": 0.4993415759028716, "learning_rate": 8.863748862404189e-05, "loss": 2.9282, "step": 14353 }, { "epoch": 0.8910546899248867, "grad_norm": 0.5435451768542703, "learning_rate": 8.863519624988176e-05, "loss": 2.9734, "step": 14354 }, { "epoch": 0.8911167670246446, "grad_norm": 0.49268317618741725, "learning_rate": 8.863290367415245e-05, "loss": 2.9334, "step": 14355 }, { "epoch": 0.8911788441244025, "grad_norm": 0.531800685895253, "learning_rate": 8.863061089686589e-05, "loss": 2.9532, "step": 14356 }, { "epoch": 0.8912409212241604, "grad_norm": 0.5116305966922765, "learning_rate": 8.862831791803405e-05, "loss": 2.9525, "step": 14357 }, { "epoch": 0.8913029983239183, "grad_norm": 0.5283067176183801, "learning_rate": 8.86260247376689e-05, "loss": 2.8881, "step": 14358 }, { "epoch": 0.8913650754236762, "grad_norm": 0.4859834085102931, "learning_rate": 8.862373135578241e-05, "loss": 2.9522, "step": 14359 }, { "epoch": 0.8914271525234341, "grad_norm": 0.5242500830210491, "learning_rate": 8.862143777238654e-05, "loss": 2.8789, "step": 14360 }, { "epoch": 0.891489229623192, "grad_norm": 0.4344177321128188, "learning_rate": 8.861914398749325e-05, "loss": 2.9874, "step": 14361 }, { "epoch": 0.8915513067229499, "grad_norm": 0.45757383759105064, "learning_rate": 8.86168500011145e-05, "loss": 3.0837, "step": 14362 }, { "epoch": 0.8916133838227078, "grad_norm": 0.4543965683741174, "learning_rate": 8.861455581326228e-05, "loss": 2.8933, "step": 14363 }, { "epoch": 0.8916754609224657, "grad_norm": 0.5091753059201612, "learning_rate": 8.861226142394854e-05, "loss": 2.9857, "step": 14364 }, { "epoch": 0.8917375380222236, "grad_norm": 0.48368253740189293, "learning_rate": 8.860996683318527e-05, "loss": 2.9994, "step": 14365 }, { "epoch": 0.8917996151219815, "grad_norm": 0.5410765465544244, "learning_rate": 8.860767204098441e-05, "loss": 2.9475, "step": 14366 }, { "epoch": 0.8918616922217394, "grad_norm": 0.4929274340695286, "learning_rate": 8.860537704735797e-05, "loss": 2.9781, "step": 14367 }, { "epoch": 0.8919237693214973, "grad_norm": 0.4863446703934412, "learning_rate": 8.86030818523179e-05, "loss": 2.9055, "step": 14368 }, { "epoch": 0.8919858464212552, "grad_norm": 0.49040504919318473, "learning_rate": 8.86007864558762e-05, "loss": 2.9445, "step": 14369 }, { "epoch": 0.8920479235210131, "grad_norm": 0.4818705910853298, "learning_rate": 8.85984908580448e-05, "loss": 2.9476, "step": 14370 }, { "epoch": 0.892110000620771, "grad_norm": 0.5338572308212051, "learning_rate": 8.859619505883572e-05, "loss": 2.9676, "step": 14371 }, { "epoch": 0.8921720777205289, "grad_norm": 0.479361320086552, "learning_rate": 8.85938990582609e-05, "loss": 2.9181, "step": 14372 }, { "epoch": 0.8922341548202868, "grad_norm": 0.5169786256193821, "learning_rate": 8.859160285633235e-05, "loss": 2.9485, "step": 14373 }, { "epoch": 0.8922962319200447, "grad_norm": 0.44430709577582767, "learning_rate": 8.858930645306205e-05, "loss": 2.943, "step": 14374 }, { "epoch": 0.8923583090198026, "grad_norm": 0.5658740996767992, "learning_rate": 8.858700984846197e-05, "loss": 2.9937, "step": 14375 }, { "epoch": 0.8924203861195605, "grad_norm": 0.48190862558109704, "learning_rate": 8.858471304254408e-05, "loss": 3.0407, "step": 14376 }, { "epoch": 0.8924824632193183, "grad_norm": 0.48656574250637186, "learning_rate": 8.858241603532038e-05, "loss": 2.9254, "step": 14377 }, { "epoch": 0.8925445403190763, "grad_norm": 0.45160388492897424, "learning_rate": 8.858011882680284e-05, "loss": 2.9725, "step": 14378 }, { "epoch": 0.8926066174188342, "grad_norm": 0.45681196641597255, "learning_rate": 8.857782141700348e-05, "loss": 2.8943, "step": 14379 }, { "epoch": 0.8926686945185921, "grad_norm": 0.43153533687088635, "learning_rate": 8.857552380593424e-05, "loss": 2.9109, "step": 14380 }, { "epoch": 0.89273077161835, "grad_norm": 0.4294269118377115, "learning_rate": 8.857322599360714e-05, "loss": 2.9778, "step": 14381 }, { "epoch": 0.8927928487181079, "grad_norm": 0.47707047514544515, "learning_rate": 8.857092798003415e-05, "loss": 2.9333, "step": 14382 }, { "epoch": 0.8928549258178657, "grad_norm": 0.4940974401367927, "learning_rate": 8.856862976522727e-05, "loss": 2.9454, "step": 14383 }, { "epoch": 0.8929170029176237, "grad_norm": 0.522500152542238, "learning_rate": 8.856633134919848e-05, "loss": 3.0402, "step": 14384 }, { "epoch": 0.8929790800173816, "grad_norm": 0.4791242815704493, "learning_rate": 8.856403273195976e-05, "loss": 2.8461, "step": 14385 }, { "epoch": 0.8930411571171395, "grad_norm": 0.5055032308571459, "learning_rate": 8.856173391352314e-05, "loss": 2.9631, "step": 14386 }, { "epoch": 0.8931032342168974, "grad_norm": 0.5182588239710793, "learning_rate": 8.85594348939006e-05, "loss": 2.8625, "step": 14387 }, { "epoch": 0.8931653113166553, "grad_norm": 0.5114023234924387, "learning_rate": 8.855713567310411e-05, "loss": 3.0533, "step": 14388 }, { "epoch": 0.8932273884164131, "grad_norm": 0.6189735018569477, "learning_rate": 8.855483625114569e-05, "loss": 3.0266, "step": 14389 }, { "epoch": 0.893289465516171, "grad_norm": 0.4950484278860155, "learning_rate": 8.855253662803732e-05, "loss": 3.0004, "step": 14390 }, { "epoch": 0.893351542615929, "grad_norm": 0.5580652310875664, "learning_rate": 8.855023680379102e-05, "loss": 3.082, "step": 14391 }, { "epoch": 0.8934136197156869, "grad_norm": 0.4811859205858972, "learning_rate": 8.854793677841877e-05, "loss": 2.906, "step": 14392 }, { "epoch": 0.8934756968154448, "grad_norm": 0.5225949979964633, "learning_rate": 8.854563655193257e-05, "loss": 3.0483, "step": 14393 }, { "epoch": 0.8935377739152027, "grad_norm": 0.49636942795637884, "learning_rate": 8.854333612434445e-05, "loss": 2.7857, "step": 14394 }, { "epoch": 0.8935998510149605, "grad_norm": 0.5913937646714522, "learning_rate": 8.854103549566637e-05, "loss": 3.0136, "step": 14395 }, { "epoch": 0.8936619281147185, "grad_norm": 0.5347600709502742, "learning_rate": 8.853873466591037e-05, "loss": 2.9568, "step": 14396 }, { "epoch": 0.8937240052144764, "grad_norm": 0.487472765112882, "learning_rate": 8.853643363508842e-05, "loss": 2.9612, "step": 14397 }, { "epoch": 0.8937860823142343, "grad_norm": 0.6046549588695883, "learning_rate": 8.853413240321253e-05, "loss": 2.9426, "step": 14398 }, { "epoch": 0.8938481594139922, "grad_norm": 0.50897377453976, "learning_rate": 8.853183097029474e-05, "loss": 3.0139, "step": 14399 }, { "epoch": 0.8939102365137501, "grad_norm": 0.4962561484227414, "learning_rate": 8.852952933634702e-05, "loss": 2.9548, "step": 14400 }, { "epoch": 0.8939723136135079, "grad_norm": 0.5205302278369285, "learning_rate": 8.852722750138138e-05, "loss": 2.9878, "step": 14401 }, { "epoch": 0.8940343907132658, "grad_norm": 0.4742739869930393, "learning_rate": 8.852492546540987e-05, "loss": 3.0504, "step": 14402 }, { "epoch": 0.8940964678130238, "grad_norm": 0.5545617350531438, "learning_rate": 8.852262322844446e-05, "loss": 2.9434, "step": 14403 }, { "epoch": 0.8941585449127817, "grad_norm": 0.5012208815034084, "learning_rate": 8.852032079049717e-05, "loss": 2.8856, "step": 14404 }, { "epoch": 0.8942206220125396, "grad_norm": 0.5529278527771408, "learning_rate": 8.851801815158002e-05, "loss": 2.9045, "step": 14405 }, { "epoch": 0.8942826991122975, "grad_norm": 0.4921044476388529, "learning_rate": 8.851571531170501e-05, "loss": 2.9532, "step": 14406 }, { "epoch": 0.8943447762120553, "grad_norm": 0.6095822544266528, "learning_rate": 8.851341227088416e-05, "loss": 2.9455, "step": 14407 }, { "epoch": 0.8944068533118132, "grad_norm": 0.4937220825077049, "learning_rate": 8.851110902912951e-05, "loss": 2.9233, "step": 14408 }, { "epoch": 0.8944689304115712, "grad_norm": 0.5623590239718529, "learning_rate": 8.850880558645304e-05, "loss": 3.0334, "step": 14409 }, { "epoch": 0.8945310075113291, "grad_norm": 0.5928391195543403, "learning_rate": 8.850650194286678e-05, "loss": 2.9842, "step": 14410 }, { "epoch": 0.894593084611087, "grad_norm": 0.5135129095180114, "learning_rate": 8.850419809838277e-05, "loss": 3.0361, "step": 14411 }, { "epoch": 0.8946551617108449, "grad_norm": 0.5680059795954762, "learning_rate": 8.850189405301299e-05, "loss": 2.9331, "step": 14412 }, { "epoch": 0.8947172388106027, "grad_norm": 0.5280986892195624, "learning_rate": 8.849958980676949e-05, "loss": 2.9698, "step": 14413 }, { "epoch": 0.8947793159103606, "grad_norm": 0.43070930018462084, "learning_rate": 8.849728535966429e-05, "loss": 2.954, "step": 14414 }, { "epoch": 0.8948413930101186, "grad_norm": 0.5478714720020373, "learning_rate": 8.84949807117094e-05, "loss": 2.9511, "step": 14415 }, { "epoch": 0.8949034701098765, "grad_norm": 0.5190718204090963, "learning_rate": 8.849267586291685e-05, "loss": 3.0138, "step": 14416 }, { "epoch": 0.8949655472096344, "grad_norm": 0.5090231915045138, "learning_rate": 8.849037081329866e-05, "loss": 2.9858, "step": 14417 }, { "epoch": 0.8950276243093923, "grad_norm": 0.5962163246689687, "learning_rate": 8.848806556286687e-05, "loss": 2.9078, "step": 14418 }, { "epoch": 0.8950897014091501, "grad_norm": 0.480113743453196, "learning_rate": 8.84857601116335e-05, "loss": 2.9664, "step": 14419 }, { "epoch": 0.895151778508908, "grad_norm": 0.6535674398964602, "learning_rate": 8.848345445961057e-05, "loss": 2.9521, "step": 14420 }, { "epoch": 0.895213855608666, "grad_norm": 0.5080028763956201, "learning_rate": 8.848114860681012e-05, "loss": 3.0189, "step": 14421 }, { "epoch": 0.8952759327084239, "grad_norm": 0.5728670936869587, "learning_rate": 8.847884255324419e-05, "loss": 2.9253, "step": 14422 }, { "epoch": 0.8953380098081818, "grad_norm": 0.5630157543403803, "learning_rate": 8.847653629892478e-05, "loss": 2.9851, "step": 14423 }, { "epoch": 0.8954000869079397, "grad_norm": 0.5102902632176676, "learning_rate": 8.847422984386395e-05, "loss": 2.9963, "step": 14424 }, { "epoch": 0.8954621640076975, "grad_norm": 0.610977340335813, "learning_rate": 8.847192318807372e-05, "loss": 2.9336, "step": 14425 }, { "epoch": 0.8955242411074554, "grad_norm": 0.6113852818493546, "learning_rate": 8.84696163315661e-05, "loss": 2.9748, "step": 14426 }, { "epoch": 0.8955863182072133, "grad_norm": 0.5694355951050621, "learning_rate": 8.846730927435319e-05, "loss": 2.999, "step": 14427 }, { "epoch": 0.8956483953069713, "grad_norm": 0.5461149929209852, "learning_rate": 8.846500201644697e-05, "loss": 2.9487, "step": 14428 }, { "epoch": 0.8957104724067292, "grad_norm": 0.5091482348105325, "learning_rate": 8.84626945578595e-05, "loss": 2.878, "step": 14429 }, { "epoch": 0.8957725495064871, "grad_norm": 0.5958293593436432, "learning_rate": 8.846038689860283e-05, "loss": 2.8934, "step": 14430 }, { "epoch": 0.8958346266062449, "grad_norm": 0.5425237300183102, "learning_rate": 8.845807903868896e-05, "loss": 2.861, "step": 14431 }, { "epoch": 0.8958967037060028, "grad_norm": 0.54313802481816, "learning_rate": 8.845577097812998e-05, "loss": 2.9877, "step": 14432 }, { "epoch": 0.8959587808057607, "grad_norm": 0.5576686801951956, "learning_rate": 8.845346271693789e-05, "loss": 3.0121, "step": 14433 }, { "epoch": 0.8960208579055187, "grad_norm": 0.5580526011500435, "learning_rate": 8.845115425512476e-05, "loss": 2.9434, "step": 14434 }, { "epoch": 0.8960829350052766, "grad_norm": 0.5384831563374985, "learning_rate": 8.84488455927026e-05, "loss": 3.0288, "step": 14435 }, { "epoch": 0.8961450121050345, "grad_norm": 0.5492486394831494, "learning_rate": 8.844653672968351e-05, "loss": 2.9736, "step": 14436 }, { "epoch": 0.8962070892047923, "grad_norm": 0.5387865272317955, "learning_rate": 8.844422766607949e-05, "loss": 2.9338, "step": 14437 }, { "epoch": 0.8962691663045502, "grad_norm": 0.48146205664063957, "learning_rate": 8.844191840190261e-05, "loss": 2.97, "step": 14438 }, { "epoch": 0.8963312434043081, "grad_norm": 0.5568102126533232, "learning_rate": 8.843960893716489e-05, "loss": 2.9699, "step": 14439 }, { "epoch": 0.8963933205040661, "grad_norm": 0.5656034370653807, "learning_rate": 8.843729927187843e-05, "loss": 2.9305, "step": 14440 }, { "epoch": 0.896455397603824, "grad_norm": 0.5196867688529996, "learning_rate": 8.843498940605524e-05, "loss": 2.9676, "step": 14441 }, { "epoch": 0.8965174747035819, "grad_norm": 0.46528237749391377, "learning_rate": 8.843267933970735e-05, "loss": 2.9771, "step": 14442 }, { "epoch": 0.8965795518033397, "grad_norm": 0.5638632150212132, "learning_rate": 8.843036907284687e-05, "loss": 2.9305, "step": 14443 }, { "epoch": 0.8966416289030976, "grad_norm": 0.5851229537230505, "learning_rate": 8.842805860548582e-05, "loss": 2.9648, "step": 14444 }, { "epoch": 0.8967037060028555, "grad_norm": 0.5494956847204514, "learning_rate": 8.842574793763626e-05, "loss": 3.036, "step": 14445 }, { "epoch": 0.8967657831026135, "grad_norm": 0.5639147187123178, "learning_rate": 8.842343706931025e-05, "loss": 3.017, "step": 14446 }, { "epoch": 0.8968278602023714, "grad_norm": 0.591683153361546, "learning_rate": 8.842112600051983e-05, "loss": 2.8776, "step": 14447 }, { "epoch": 0.8968899373021293, "grad_norm": 0.5934318894959245, "learning_rate": 8.841881473127707e-05, "loss": 2.9002, "step": 14448 }, { "epoch": 0.8969520144018871, "grad_norm": 0.5481232430180888, "learning_rate": 8.841650326159403e-05, "loss": 3.0485, "step": 14449 }, { "epoch": 0.897014091501645, "grad_norm": 0.6099000038592928, "learning_rate": 8.841419159148276e-05, "loss": 2.9246, "step": 14450 }, { "epoch": 0.8970761686014029, "grad_norm": 0.5882169001872988, "learning_rate": 8.841187972095535e-05, "loss": 2.9717, "step": 14451 }, { "epoch": 0.8971382457011609, "grad_norm": 0.5451310279371203, "learning_rate": 8.840956765002382e-05, "loss": 2.955, "step": 14452 }, { "epoch": 0.8972003228009188, "grad_norm": 0.5955109583345262, "learning_rate": 8.840725537870025e-05, "loss": 2.9974, "step": 14453 }, { "epoch": 0.8972623999006767, "grad_norm": 0.5848551863159355, "learning_rate": 8.84049429069967e-05, "loss": 2.9723, "step": 14454 }, { "epoch": 0.8973244770004345, "grad_norm": 0.5499936959598963, "learning_rate": 8.840263023492525e-05, "loss": 3.0214, "step": 14455 }, { "epoch": 0.8973865541001924, "grad_norm": 0.5191464540772874, "learning_rate": 8.840031736249797e-05, "loss": 2.9926, "step": 14456 }, { "epoch": 0.8974486311999503, "grad_norm": 0.5963078165713027, "learning_rate": 8.839800428972689e-05, "loss": 2.9415, "step": 14457 }, { "epoch": 0.8975107082997082, "grad_norm": 0.5064032146171035, "learning_rate": 8.83956910166241e-05, "loss": 3.0351, "step": 14458 }, { "epoch": 0.8975727853994662, "grad_norm": 0.519447147761273, "learning_rate": 8.839337754320168e-05, "loss": 2.9623, "step": 14459 }, { "epoch": 0.8976348624992241, "grad_norm": 0.4964794251730201, "learning_rate": 8.83910638694717e-05, "loss": 2.8855, "step": 14460 }, { "epoch": 0.8976969395989819, "grad_norm": 0.4998644853770513, "learning_rate": 8.83887499954462e-05, "loss": 2.9384, "step": 14461 }, { "epoch": 0.8977590166987398, "grad_norm": 0.5333756108464887, "learning_rate": 8.838643592113727e-05, "loss": 3.0307, "step": 14462 }, { "epoch": 0.8978210937984977, "grad_norm": 0.5171153464795959, "learning_rate": 8.838412164655699e-05, "loss": 2.9055, "step": 14463 }, { "epoch": 0.8978831708982556, "grad_norm": 0.4705139033080497, "learning_rate": 8.838180717171742e-05, "loss": 2.8588, "step": 14464 }, { "epoch": 0.8979452479980136, "grad_norm": 0.5507905567005156, "learning_rate": 8.837949249663066e-05, "loss": 2.9596, "step": 14465 }, { "epoch": 0.8980073250977715, "grad_norm": 0.5420249585391076, "learning_rate": 8.837717762130877e-05, "loss": 2.9864, "step": 14466 }, { "epoch": 0.8980694021975293, "grad_norm": 0.5078726512789843, "learning_rate": 8.837486254576382e-05, "loss": 2.9178, "step": 14467 }, { "epoch": 0.8981314792972872, "grad_norm": 0.5663551570604665, "learning_rate": 8.83725472700079e-05, "loss": 2.9367, "step": 14468 }, { "epoch": 0.8981935563970451, "grad_norm": 0.43768722070502303, "learning_rate": 8.837023179405309e-05, "loss": 3.0336, "step": 14469 }, { "epoch": 0.898255633496803, "grad_norm": 0.5824713611111397, "learning_rate": 8.836791611791146e-05, "loss": 2.9505, "step": 14470 }, { "epoch": 0.898317710596561, "grad_norm": 0.45781397835554655, "learning_rate": 8.83656002415951e-05, "loss": 2.9065, "step": 14471 }, { "epoch": 0.8983797876963189, "grad_norm": 0.5372055799849171, "learning_rate": 8.83632841651161e-05, "loss": 2.9258, "step": 14472 }, { "epoch": 0.8984418647960767, "grad_norm": 0.5174906242829713, "learning_rate": 8.836096788848651e-05, "loss": 3.0217, "step": 14473 }, { "epoch": 0.8985039418958346, "grad_norm": 0.4424898300241454, "learning_rate": 8.835865141171847e-05, "loss": 2.892, "step": 14474 }, { "epoch": 0.8985660189955925, "grad_norm": 0.4958325396726438, "learning_rate": 8.8356334734824e-05, "loss": 2.95, "step": 14475 }, { "epoch": 0.8986280960953504, "grad_norm": 0.4780598240728964, "learning_rate": 8.835401785781523e-05, "loss": 2.8423, "step": 14476 }, { "epoch": 0.8986901731951084, "grad_norm": 0.5070065191083046, "learning_rate": 8.835170078070425e-05, "loss": 3.0397, "step": 14477 }, { "epoch": 0.8987522502948663, "grad_norm": 0.4950165806266789, "learning_rate": 8.834938350350313e-05, "loss": 2.9528, "step": 14478 }, { "epoch": 0.8988143273946241, "grad_norm": 0.5165263880241735, "learning_rate": 8.834706602622397e-05, "loss": 2.9535, "step": 14479 }, { "epoch": 0.898876404494382, "grad_norm": 0.4954283599847761, "learning_rate": 8.834474834887884e-05, "loss": 3.0048, "step": 14480 }, { "epoch": 0.8989384815941399, "grad_norm": 0.498867839286796, "learning_rate": 8.834243047147988e-05, "loss": 3.0098, "step": 14481 }, { "epoch": 0.8990005586938978, "grad_norm": 0.5404667738530694, "learning_rate": 8.834011239403913e-05, "loss": 2.9747, "step": 14482 }, { "epoch": 0.8990626357936558, "grad_norm": 0.5137069168032098, "learning_rate": 8.83377941165687e-05, "loss": 3.0104, "step": 14483 }, { "epoch": 0.8991247128934137, "grad_norm": 0.5549073925166595, "learning_rate": 8.833547563908071e-05, "loss": 2.9932, "step": 14484 }, { "epoch": 0.8991867899931715, "grad_norm": 0.4992217175749211, "learning_rate": 8.833315696158724e-05, "loss": 2.9512, "step": 14485 }, { "epoch": 0.8992488670929294, "grad_norm": 0.4949064840981147, "learning_rate": 8.833083808410037e-05, "loss": 2.8996, "step": 14486 }, { "epoch": 0.8993109441926873, "grad_norm": 0.5156631455616658, "learning_rate": 8.832851900663221e-05, "loss": 2.9636, "step": 14487 }, { "epoch": 0.8993730212924452, "grad_norm": 0.500343026272948, "learning_rate": 8.832619972919487e-05, "loss": 2.9848, "step": 14488 }, { "epoch": 0.8994350983922031, "grad_norm": 0.5080713052755039, "learning_rate": 8.832388025180045e-05, "loss": 2.9721, "step": 14489 }, { "epoch": 0.8994971754919611, "grad_norm": 0.4427536679411583, "learning_rate": 8.832156057446103e-05, "loss": 2.9695, "step": 14490 }, { "epoch": 0.8995592525917189, "grad_norm": 0.5230197947915448, "learning_rate": 8.831924069718874e-05, "loss": 2.9659, "step": 14491 }, { "epoch": 0.8996213296914768, "grad_norm": 0.46128245554820035, "learning_rate": 8.831692061999566e-05, "loss": 2.9497, "step": 14492 }, { "epoch": 0.8996834067912347, "grad_norm": 0.502901302388451, "learning_rate": 8.83146003428939e-05, "loss": 3.0339, "step": 14493 }, { "epoch": 0.8997454838909926, "grad_norm": 0.5076248702627024, "learning_rate": 8.831227986589556e-05, "loss": 2.8995, "step": 14494 }, { "epoch": 0.8998075609907505, "grad_norm": 0.5019003695658957, "learning_rate": 8.830995918901279e-05, "loss": 2.9398, "step": 14495 }, { "epoch": 0.8998696380905085, "grad_norm": 0.5566264368223316, "learning_rate": 8.830763831225763e-05, "loss": 2.9477, "step": 14496 }, { "epoch": 0.8999317151902663, "grad_norm": 0.4740580900938936, "learning_rate": 8.830531723564223e-05, "loss": 2.9658, "step": 14497 }, { "epoch": 0.8999937922900242, "grad_norm": 0.50199174546868, "learning_rate": 8.830299595917869e-05, "loss": 2.8883, "step": 14498 }, { "epoch": 0.9000558693897821, "grad_norm": 0.4930392267839551, "learning_rate": 8.830067448287913e-05, "loss": 3.0244, "step": 14499 }, { "epoch": 0.90011794648954, "grad_norm": 0.5326280671796599, "learning_rate": 8.829835280675564e-05, "loss": 2.9193, "step": 14500 }, { "epoch": 0.9001800235892979, "grad_norm": 0.4681410171508674, "learning_rate": 8.829603093082035e-05, "loss": 3.0042, "step": 14501 }, { "epoch": 0.9002421006890559, "grad_norm": 0.5375619542101708, "learning_rate": 8.829370885508538e-05, "loss": 2.9196, "step": 14502 }, { "epoch": 0.9003041777888137, "grad_norm": 0.48688840955354656, "learning_rate": 8.829138657956283e-05, "loss": 2.935, "step": 14503 }, { "epoch": 0.9003662548885716, "grad_norm": 0.523093637280572, "learning_rate": 8.828906410426482e-05, "loss": 2.9195, "step": 14504 }, { "epoch": 0.9004283319883295, "grad_norm": 0.5655932939189191, "learning_rate": 8.828674142920346e-05, "loss": 2.8751, "step": 14505 }, { "epoch": 0.9004904090880874, "grad_norm": 0.5161483305233295, "learning_rate": 8.828441855439089e-05, "loss": 2.959, "step": 14506 }, { "epoch": 0.9005524861878453, "grad_norm": 0.47759613132962664, "learning_rate": 8.82820954798392e-05, "loss": 2.8433, "step": 14507 }, { "epoch": 0.9006145632876033, "grad_norm": 0.5232186785261976, "learning_rate": 8.827977220556054e-05, "loss": 2.9577, "step": 14508 }, { "epoch": 0.9006766403873611, "grad_norm": 0.44905928393588085, "learning_rate": 8.827744873156701e-05, "loss": 2.9868, "step": 14509 }, { "epoch": 0.900738717487119, "grad_norm": 0.6220170362046826, "learning_rate": 8.827512505787072e-05, "loss": 2.9027, "step": 14510 }, { "epoch": 0.9008007945868769, "grad_norm": 0.44786032240993356, "learning_rate": 8.827280118448383e-05, "loss": 2.9332, "step": 14511 }, { "epoch": 0.9008628716866348, "grad_norm": 0.5681554684164396, "learning_rate": 8.827047711141845e-05, "loss": 3.0157, "step": 14512 }, { "epoch": 0.9009249487863927, "grad_norm": 0.6087647585871604, "learning_rate": 8.826815283868668e-05, "loss": 2.8742, "step": 14513 }, { "epoch": 0.9009870258861506, "grad_norm": 0.46251404006370456, "learning_rate": 8.826582836630068e-05, "loss": 2.8719, "step": 14514 }, { "epoch": 0.9010491029859085, "grad_norm": 0.5548382943652452, "learning_rate": 8.826350369427256e-05, "loss": 3.022, "step": 14515 }, { "epoch": 0.9011111800856664, "grad_norm": 0.5395203732858811, "learning_rate": 8.826117882261446e-05, "loss": 3.0186, "step": 14516 }, { "epoch": 0.9011732571854243, "grad_norm": 0.5308870029067112, "learning_rate": 8.825885375133849e-05, "loss": 2.9504, "step": 14517 }, { "epoch": 0.9012353342851822, "grad_norm": 0.5027780544983937, "learning_rate": 8.82565284804568e-05, "loss": 2.9363, "step": 14518 }, { "epoch": 0.9012974113849401, "grad_norm": 0.46629213633630245, "learning_rate": 8.82542030099815e-05, "loss": 2.8987, "step": 14519 }, { "epoch": 0.901359488484698, "grad_norm": 0.48241216055166425, "learning_rate": 8.825187733992476e-05, "loss": 2.9546, "step": 14520 }, { "epoch": 0.9014215655844559, "grad_norm": 0.49614906002706427, "learning_rate": 8.824955147029868e-05, "loss": 2.9579, "step": 14521 }, { "epoch": 0.9014836426842138, "grad_norm": 0.48899456563906013, "learning_rate": 8.824722540111541e-05, "loss": 2.9661, "step": 14522 }, { "epoch": 0.9015457197839717, "grad_norm": 0.4344665403443659, "learning_rate": 8.824489913238707e-05, "loss": 2.9193, "step": 14523 }, { "epoch": 0.9016077968837296, "grad_norm": 0.47952870154177496, "learning_rate": 8.824257266412581e-05, "loss": 3.0002, "step": 14524 }, { "epoch": 0.9016698739834875, "grad_norm": 0.5015702327239138, "learning_rate": 8.824024599634376e-05, "loss": 2.9549, "step": 14525 }, { "epoch": 0.9017319510832454, "grad_norm": 0.47731605736651983, "learning_rate": 8.823791912905308e-05, "loss": 2.9703, "step": 14526 }, { "epoch": 0.9017940281830032, "grad_norm": 0.49628090519602164, "learning_rate": 8.823559206226587e-05, "loss": 2.939, "step": 14527 }, { "epoch": 0.9018561052827612, "grad_norm": 0.5817667212777795, "learning_rate": 8.82332647959943e-05, "loss": 2.9882, "step": 14528 }, { "epoch": 0.9019181823825191, "grad_norm": 0.5122303334654938, "learning_rate": 8.823093733025051e-05, "loss": 2.9635, "step": 14529 }, { "epoch": 0.901980259482277, "grad_norm": 0.5020817395207393, "learning_rate": 8.822860966504665e-05, "loss": 2.9899, "step": 14530 }, { "epoch": 0.9020423365820349, "grad_norm": 0.5575895285886733, "learning_rate": 8.822628180039485e-05, "loss": 3.0134, "step": 14531 }, { "epoch": 0.9021044136817928, "grad_norm": 0.4945278233896052, "learning_rate": 8.822395373630725e-05, "loss": 3.0181, "step": 14532 }, { "epoch": 0.9021664907815506, "grad_norm": 0.5302643889327828, "learning_rate": 8.8221625472796e-05, "loss": 3.0371, "step": 14533 }, { "epoch": 0.9022285678813086, "grad_norm": 0.5212060135411015, "learning_rate": 8.821929700987326e-05, "loss": 2.9593, "step": 14534 }, { "epoch": 0.9022906449810665, "grad_norm": 0.5289005611682842, "learning_rate": 8.821696834755118e-05, "loss": 2.9874, "step": 14535 }, { "epoch": 0.9023527220808244, "grad_norm": 0.5539655150652983, "learning_rate": 8.821463948584188e-05, "loss": 2.9778, "step": 14536 }, { "epoch": 0.9024147991805823, "grad_norm": 0.47781591834572623, "learning_rate": 8.821231042475753e-05, "loss": 2.9806, "step": 14537 }, { "epoch": 0.9024768762803402, "grad_norm": 0.5266857704739881, "learning_rate": 8.820998116431029e-05, "loss": 2.9784, "step": 14538 }, { "epoch": 0.902538953380098, "grad_norm": 0.5622051843954434, "learning_rate": 8.82076517045123e-05, "loss": 2.9678, "step": 14539 }, { "epoch": 0.902601030479856, "grad_norm": 0.606611662085481, "learning_rate": 8.82053220453757e-05, "loss": 3.0677, "step": 14540 }, { "epoch": 0.9026631075796139, "grad_norm": 0.5576537165393181, "learning_rate": 8.820299218691269e-05, "loss": 2.9675, "step": 14541 }, { "epoch": 0.9027251846793718, "grad_norm": 0.4897015324074931, "learning_rate": 8.820066212913536e-05, "loss": 3.0833, "step": 14542 }, { "epoch": 0.9027872617791297, "grad_norm": 0.5571975063980866, "learning_rate": 8.819833187205593e-05, "loss": 3.0673, "step": 14543 }, { "epoch": 0.9028493388788876, "grad_norm": 0.5236333509113029, "learning_rate": 8.819600141568652e-05, "loss": 2.9635, "step": 14544 }, { "epoch": 0.9029114159786454, "grad_norm": 0.4721984222279921, "learning_rate": 8.819367076003931e-05, "loss": 3.0106, "step": 14545 }, { "epoch": 0.9029734930784034, "grad_norm": 0.5187316994713677, "learning_rate": 8.819133990512642e-05, "loss": 2.8863, "step": 14546 }, { "epoch": 0.9030355701781613, "grad_norm": 0.49534469283266125, "learning_rate": 8.818900885096006e-05, "loss": 2.9404, "step": 14547 }, { "epoch": 0.9030976472779192, "grad_norm": 0.47899446468400253, "learning_rate": 8.818667759755237e-05, "loss": 2.9666, "step": 14548 }, { "epoch": 0.9031597243776771, "grad_norm": 0.5417523377477661, "learning_rate": 8.818434614491551e-05, "loss": 3.044, "step": 14549 }, { "epoch": 0.903221801477435, "grad_norm": 0.48699488758952086, "learning_rate": 8.818201449306165e-05, "loss": 2.9433, "step": 14550 }, { "epoch": 0.9032838785771928, "grad_norm": 0.5725689497493748, "learning_rate": 8.817968264200294e-05, "loss": 3.0227, "step": 14551 }, { "epoch": 0.9033459556769508, "grad_norm": 0.47710901722911003, "learning_rate": 8.817735059175156e-05, "loss": 2.9017, "step": 14552 }, { "epoch": 0.9034080327767087, "grad_norm": 0.5638944507610631, "learning_rate": 8.817501834231968e-05, "loss": 2.9938, "step": 14553 }, { "epoch": 0.9034701098764666, "grad_norm": 0.5109943183314315, "learning_rate": 8.817268589371945e-05, "loss": 2.9886, "step": 14554 }, { "epoch": 0.9035321869762245, "grad_norm": 0.47593012267398277, "learning_rate": 8.817035324596308e-05, "loss": 2.9616, "step": 14555 }, { "epoch": 0.9035942640759824, "grad_norm": 0.5218639349763771, "learning_rate": 8.816802039906269e-05, "loss": 2.9487, "step": 14556 }, { "epoch": 0.9036563411757402, "grad_norm": 0.5116350527739673, "learning_rate": 8.816568735303047e-05, "loss": 2.956, "step": 14557 }, { "epoch": 0.9037184182754981, "grad_norm": 0.44418963025238306, "learning_rate": 8.816335410787859e-05, "loss": 2.9524, "step": 14558 }, { "epoch": 0.9037804953752561, "grad_norm": 0.5320337908847985, "learning_rate": 8.816102066361924e-05, "loss": 2.9879, "step": 14559 }, { "epoch": 0.903842572475014, "grad_norm": 0.45607450389081255, "learning_rate": 8.815868702026458e-05, "loss": 2.932, "step": 14560 }, { "epoch": 0.9039046495747719, "grad_norm": 0.4519994832284893, "learning_rate": 8.815635317782677e-05, "loss": 2.8901, "step": 14561 }, { "epoch": 0.9039667266745298, "grad_norm": 0.47725850232195666, "learning_rate": 8.815401913631803e-05, "loss": 2.9796, "step": 14562 }, { "epoch": 0.9040288037742876, "grad_norm": 0.47194182573270144, "learning_rate": 8.815168489575048e-05, "loss": 2.9955, "step": 14563 }, { "epoch": 0.9040908808740455, "grad_norm": 0.42836382462679284, "learning_rate": 8.814935045613635e-05, "loss": 2.9676, "step": 14564 }, { "epoch": 0.9041529579738035, "grad_norm": 0.496223958266602, "learning_rate": 8.814701581748779e-05, "loss": 2.957, "step": 14565 }, { "epoch": 0.9042150350735614, "grad_norm": 0.47301620774596587, "learning_rate": 8.814468097981698e-05, "loss": 2.9345, "step": 14566 }, { "epoch": 0.9042771121733193, "grad_norm": 0.5243043980224272, "learning_rate": 8.814234594313613e-05, "loss": 3.0782, "step": 14567 }, { "epoch": 0.9043391892730772, "grad_norm": 0.4596889363829826, "learning_rate": 8.814001070745739e-05, "loss": 2.9412, "step": 14568 }, { "epoch": 0.904401266372835, "grad_norm": 0.47600695012532296, "learning_rate": 8.813767527279295e-05, "loss": 3.0022, "step": 14569 }, { "epoch": 0.9044633434725929, "grad_norm": 0.5214070180602071, "learning_rate": 8.813533963915501e-05, "loss": 2.9635, "step": 14570 }, { "epoch": 0.9045254205723509, "grad_norm": 0.45671969170592375, "learning_rate": 8.813300380655574e-05, "loss": 2.9637, "step": 14571 }, { "epoch": 0.9045874976721088, "grad_norm": 0.48667621028645996, "learning_rate": 8.813066777500734e-05, "loss": 2.934, "step": 14572 }, { "epoch": 0.9046495747718667, "grad_norm": 0.5106372324249933, "learning_rate": 8.812833154452199e-05, "loss": 2.9012, "step": 14573 }, { "epoch": 0.9047116518716246, "grad_norm": 0.5175721078199024, "learning_rate": 8.812599511511185e-05, "loss": 2.9477, "step": 14574 }, { "epoch": 0.9047737289713824, "grad_norm": 0.513899517021019, "learning_rate": 8.812365848678917e-05, "loss": 3.0028, "step": 14575 }, { "epoch": 0.9048358060711403, "grad_norm": 0.45719730588314433, "learning_rate": 8.81213216595661e-05, "loss": 3.0132, "step": 14576 }, { "epoch": 0.9048978831708983, "grad_norm": 0.48985433856039207, "learning_rate": 8.811898463345482e-05, "loss": 2.8693, "step": 14577 }, { "epoch": 0.9049599602706562, "grad_norm": 0.49788389344141376, "learning_rate": 8.811664740846757e-05, "loss": 2.9198, "step": 14578 }, { "epoch": 0.9050220373704141, "grad_norm": 0.5077406905826707, "learning_rate": 8.811430998461651e-05, "loss": 2.9353, "step": 14579 }, { "epoch": 0.905084114470172, "grad_norm": 0.5152516313718598, "learning_rate": 8.811197236191384e-05, "loss": 3.0868, "step": 14580 }, { "epoch": 0.9051461915699298, "grad_norm": 0.49311301089895193, "learning_rate": 8.810963454037175e-05, "loss": 2.9564, "step": 14581 }, { "epoch": 0.9052082686696877, "grad_norm": 0.4904618514378807, "learning_rate": 8.810729652000247e-05, "loss": 3.005, "step": 14582 }, { "epoch": 0.9052703457694457, "grad_norm": 0.4990981581674006, "learning_rate": 8.810495830081816e-05, "loss": 2.9912, "step": 14583 }, { "epoch": 0.9053324228692036, "grad_norm": 0.4871108562137993, "learning_rate": 8.810261988283102e-05, "loss": 2.9593, "step": 14584 }, { "epoch": 0.9053944999689615, "grad_norm": 0.5407289923168713, "learning_rate": 8.810028126605328e-05, "loss": 2.9719, "step": 14585 }, { "epoch": 0.9054565770687194, "grad_norm": 0.48662630002220253, "learning_rate": 8.809794245049711e-05, "loss": 2.84, "step": 14586 }, { "epoch": 0.9055186541684772, "grad_norm": 0.5502771004637294, "learning_rate": 8.809560343617473e-05, "loss": 3.0156, "step": 14587 }, { "epoch": 0.9055807312682351, "grad_norm": 0.5777460866982921, "learning_rate": 8.809326422309836e-05, "loss": 2.9984, "step": 14588 }, { "epoch": 0.905642808367993, "grad_norm": 0.4907681875788976, "learning_rate": 8.809092481128015e-05, "loss": 2.8914, "step": 14589 }, { "epoch": 0.905704885467751, "grad_norm": 0.49794139867141896, "learning_rate": 8.808858520073236e-05, "loss": 2.9771, "step": 14590 }, { "epoch": 0.9057669625675089, "grad_norm": 0.5513027779189323, "learning_rate": 8.808624539146715e-05, "loss": 3.0158, "step": 14591 }, { "epoch": 0.9058290396672668, "grad_norm": 0.5151313672985512, "learning_rate": 8.808390538349677e-05, "loss": 3.0386, "step": 14592 }, { "epoch": 0.9058911167670246, "grad_norm": 0.5952108381479246, "learning_rate": 8.808156517683342e-05, "loss": 2.9507, "step": 14593 }, { "epoch": 0.9059531938667825, "grad_norm": 0.5987804535704752, "learning_rate": 8.807922477148928e-05, "loss": 2.947, "step": 14594 }, { "epoch": 0.9060152709665404, "grad_norm": 0.516235885565719, "learning_rate": 8.80768841674766e-05, "loss": 2.9374, "step": 14595 }, { "epoch": 0.9060773480662984, "grad_norm": 0.5194886268971578, "learning_rate": 8.807454336480756e-05, "loss": 2.9485, "step": 14596 }, { "epoch": 0.9061394251660563, "grad_norm": 0.5289639766863342, "learning_rate": 8.807220236349438e-05, "loss": 2.9591, "step": 14597 }, { "epoch": 0.9062015022658142, "grad_norm": 0.487797734234417, "learning_rate": 8.806986116354928e-05, "loss": 2.9276, "step": 14598 }, { "epoch": 0.906263579365572, "grad_norm": 0.5037710023103589, "learning_rate": 8.806751976498448e-05, "loss": 2.8848, "step": 14599 }, { "epoch": 0.9063256564653299, "grad_norm": 0.47865453766832017, "learning_rate": 8.806517816781219e-05, "loss": 2.9519, "step": 14600 }, { "epoch": 0.9063877335650878, "grad_norm": 0.5570085079478179, "learning_rate": 8.806283637204462e-05, "loss": 2.9402, "step": 14601 }, { "epoch": 0.9064498106648458, "grad_norm": 0.49326071217037154, "learning_rate": 8.8060494377694e-05, "loss": 2.9904, "step": 14602 }, { "epoch": 0.9065118877646037, "grad_norm": 0.5365443361348312, "learning_rate": 8.805815218477252e-05, "loss": 2.8924, "step": 14603 }, { "epoch": 0.9065739648643616, "grad_norm": 0.5374276622621986, "learning_rate": 8.805580979329244e-05, "loss": 3.0137, "step": 14604 }, { "epoch": 0.9066360419641194, "grad_norm": 0.5152806085604187, "learning_rate": 8.805346720326595e-05, "loss": 2.9868, "step": 14605 }, { "epoch": 0.9066981190638773, "grad_norm": 0.5476678550459277, "learning_rate": 8.805112441470529e-05, "loss": 2.9822, "step": 14606 }, { "epoch": 0.9067601961636352, "grad_norm": 0.5246511026632431, "learning_rate": 8.804878142762268e-05, "loss": 2.9938, "step": 14607 }, { "epoch": 0.9068222732633932, "grad_norm": 0.4790381711939758, "learning_rate": 8.804643824203033e-05, "loss": 3.0173, "step": 14608 }, { "epoch": 0.9068843503631511, "grad_norm": 0.5697629972517856, "learning_rate": 8.804409485794047e-05, "loss": 2.9784, "step": 14609 }, { "epoch": 0.906946427462909, "grad_norm": 0.5024230479157177, "learning_rate": 8.804175127536536e-05, "loss": 2.8987, "step": 14610 }, { "epoch": 0.9070085045626668, "grad_norm": 0.511457157975327, "learning_rate": 8.803940749431718e-05, "loss": 2.9531, "step": 14611 }, { "epoch": 0.9070705816624247, "grad_norm": 0.4756114725703112, "learning_rate": 8.803706351480818e-05, "loss": 2.9747, "step": 14612 }, { "epoch": 0.9071326587621826, "grad_norm": 0.5200722824348414, "learning_rate": 8.803471933685059e-05, "loss": 3.0296, "step": 14613 }, { "epoch": 0.9071947358619405, "grad_norm": 0.5554073391002601, "learning_rate": 8.803237496045662e-05, "loss": 2.9493, "step": 14614 }, { "epoch": 0.9072568129616985, "grad_norm": 0.5350356697063512, "learning_rate": 8.803003038563853e-05, "loss": 2.9714, "step": 14615 }, { "epoch": 0.9073188900614563, "grad_norm": 0.571253770569559, "learning_rate": 8.802768561240854e-05, "loss": 2.897, "step": 14616 }, { "epoch": 0.9073809671612142, "grad_norm": 0.552047896297521, "learning_rate": 8.802534064077887e-05, "loss": 2.8564, "step": 14617 }, { "epoch": 0.9074430442609721, "grad_norm": 0.5187069082171865, "learning_rate": 8.802299547076178e-05, "loss": 2.9405, "step": 14618 }, { "epoch": 0.90750512136073, "grad_norm": 0.5340086891649644, "learning_rate": 8.802065010236951e-05, "loss": 3.0483, "step": 14619 }, { "epoch": 0.907567198460488, "grad_norm": 0.5287699029211234, "learning_rate": 8.801830453561425e-05, "loss": 2.9708, "step": 14620 }, { "epoch": 0.9076292755602459, "grad_norm": 0.6011330446389958, "learning_rate": 8.801595877050827e-05, "loss": 3.0156, "step": 14621 }, { "epoch": 0.9076913526600037, "grad_norm": 0.5517396907024752, "learning_rate": 8.801361280706381e-05, "loss": 3.0182, "step": 14622 }, { "epoch": 0.9077534297597616, "grad_norm": 0.5117758507938458, "learning_rate": 8.80112666452931e-05, "loss": 3.0594, "step": 14623 }, { "epoch": 0.9078155068595195, "grad_norm": 0.5715345067364191, "learning_rate": 8.800892028520839e-05, "loss": 3.0068, "step": 14624 }, { "epoch": 0.9078775839592774, "grad_norm": 0.5840123913018647, "learning_rate": 8.800657372682191e-05, "loss": 3.0009, "step": 14625 }, { "epoch": 0.9079396610590353, "grad_norm": 0.5814344943945933, "learning_rate": 8.800422697014592e-05, "loss": 2.949, "step": 14626 }, { "epoch": 0.9080017381587933, "grad_norm": 0.570799397645551, "learning_rate": 8.800188001519263e-05, "loss": 2.8746, "step": 14627 }, { "epoch": 0.9080638152585511, "grad_norm": 0.5250984996400814, "learning_rate": 8.799953286197433e-05, "loss": 2.9167, "step": 14628 }, { "epoch": 0.908125892358309, "grad_norm": 0.5694643271730354, "learning_rate": 8.799718551050323e-05, "loss": 3.1144, "step": 14629 }, { "epoch": 0.9081879694580669, "grad_norm": 0.4956909406651038, "learning_rate": 8.799483796079159e-05, "loss": 2.9123, "step": 14630 }, { "epoch": 0.9082500465578248, "grad_norm": 0.5754115632411183, "learning_rate": 8.799249021285166e-05, "loss": 2.9029, "step": 14631 }, { "epoch": 0.9083121236575827, "grad_norm": 0.5406717522752782, "learning_rate": 8.799014226669569e-05, "loss": 2.9135, "step": 14632 }, { "epoch": 0.9083742007573407, "grad_norm": 0.5479759578268856, "learning_rate": 8.79877941223359e-05, "loss": 3.0633, "step": 14633 }, { "epoch": 0.9084362778570985, "grad_norm": 0.581068869025009, "learning_rate": 8.798544577978459e-05, "loss": 3.0444, "step": 14634 }, { "epoch": 0.9084983549568564, "grad_norm": 0.5662078246774208, "learning_rate": 8.798309723905398e-05, "loss": 2.8579, "step": 14635 }, { "epoch": 0.9085604320566143, "grad_norm": 0.5899034684625841, "learning_rate": 8.798074850015634e-05, "loss": 2.9915, "step": 14636 }, { "epoch": 0.9086225091563722, "grad_norm": 0.5457022763179441, "learning_rate": 8.797839956310392e-05, "loss": 2.9982, "step": 14637 }, { "epoch": 0.9086845862561301, "grad_norm": 0.6197057846149613, "learning_rate": 8.797605042790895e-05, "loss": 3.0157, "step": 14638 }, { "epoch": 0.908746663355888, "grad_norm": 0.4685062092376723, "learning_rate": 8.797370109458372e-05, "loss": 2.9185, "step": 14639 }, { "epoch": 0.9088087404556459, "grad_norm": 0.5288012672261969, "learning_rate": 8.797135156314047e-05, "loss": 2.9877, "step": 14640 }, { "epoch": 0.9088708175554038, "grad_norm": 0.5954219717461685, "learning_rate": 8.796900183359145e-05, "loss": 2.9866, "step": 14641 }, { "epoch": 0.9089328946551617, "grad_norm": 0.5883613805416311, "learning_rate": 8.796665190594895e-05, "loss": 2.9948, "step": 14642 }, { "epoch": 0.9089949717549196, "grad_norm": 0.5056331821057833, "learning_rate": 8.796430178022519e-05, "loss": 3.0155, "step": 14643 }, { "epoch": 0.9090570488546775, "grad_norm": 0.5440940174261848, "learning_rate": 8.796195145643247e-05, "loss": 2.9307, "step": 14644 }, { "epoch": 0.9091191259544354, "grad_norm": 0.5836992365066321, "learning_rate": 8.795960093458302e-05, "loss": 2.9437, "step": 14645 }, { "epoch": 0.9091812030541933, "grad_norm": 0.5363912046585434, "learning_rate": 8.795725021468912e-05, "loss": 2.949, "step": 14646 }, { "epoch": 0.9092432801539512, "grad_norm": 0.4919317731605137, "learning_rate": 8.795489929676304e-05, "loss": 2.968, "step": 14647 }, { "epoch": 0.9093053572537091, "grad_norm": 0.5800806938140196, "learning_rate": 8.795254818081702e-05, "loss": 2.9292, "step": 14648 }, { "epoch": 0.909367434353467, "grad_norm": 0.5168453571532853, "learning_rate": 8.795019686686334e-05, "loss": 3.0328, "step": 14649 }, { "epoch": 0.9094295114532249, "grad_norm": 0.5567723436278605, "learning_rate": 8.794784535491429e-05, "loss": 2.9054, "step": 14650 }, { "epoch": 0.9094915885529828, "grad_norm": 0.598301532926577, "learning_rate": 8.794549364498211e-05, "loss": 2.9438, "step": 14651 }, { "epoch": 0.9095536656527407, "grad_norm": 0.5328855022636633, "learning_rate": 8.794314173707909e-05, "loss": 2.9895, "step": 14652 }, { "epoch": 0.9096157427524986, "grad_norm": 0.5311400916903986, "learning_rate": 8.794078963121747e-05, "loss": 2.8839, "step": 14653 }, { "epoch": 0.9096778198522565, "grad_norm": 0.5726464992801558, "learning_rate": 8.793843732740954e-05, "loss": 3.034, "step": 14654 }, { "epoch": 0.9097398969520144, "grad_norm": 0.5391744414490733, "learning_rate": 8.793608482566759e-05, "loss": 2.9305, "step": 14655 }, { "epoch": 0.9098019740517723, "grad_norm": 0.5928101630358363, "learning_rate": 8.793373212600386e-05, "loss": 2.942, "step": 14656 }, { "epoch": 0.9098640511515302, "grad_norm": 0.6474475197530892, "learning_rate": 8.793137922843065e-05, "loss": 2.9105, "step": 14657 }, { "epoch": 0.909926128251288, "grad_norm": 0.48757046654848907, "learning_rate": 8.792902613296022e-05, "loss": 2.9402, "step": 14658 }, { "epoch": 0.909988205351046, "grad_norm": 0.5825799749180552, "learning_rate": 8.792667283960486e-05, "loss": 2.9271, "step": 14659 }, { "epoch": 0.9100502824508039, "grad_norm": 0.5682077369413607, "learning_rate": 8.792431934837684e-05, "loss": 2.9772, "step": 14660 }, { "epoch": 0.9101123595505618, "grad_norm": 0.5229520681893075, "learning_rate": 8.792196565928843e-05, "loss": 2.9361, "step": 14661 }, { "epoch": 0.9101744366503197, "grad_norm": 0.5687366033380625, "learning_rate": 8.791961177235193e-05, "loss": 2.9406, "step": 14662 }, { "epoch": 0.9102365137500776, "grad_norm": 0.5720655853294976, "learning_rate": 8.791725768757961e-05, "loss": 3.0139, "step": 14663 }, { "epoch": 0.9102985908498354, "grad_norm": 0.5523757849741889, "learning_rate": 8.791490340498374e-05, "loss": 2.8507, "step": 14664 }, { "epoch": 0.9103606679495934, "grad_norm": 0.61988025854903, "learning_rate": 8.791254892457663e-05, "loss": 2.9498, "step": 14665 }, { "epoch": 0.9104227450493513, "grad_norm": 0.5242310671773465, "learning_rate": 8.791019424637054e-05, "loss": 2.952, "step": 14666 }, { "epoch": 0.9104848221491092, "grad_norm": 0.60178106226544, "learning_rate": 8.790783937037777e-05, "loss": 2.9273, "step": 14667 }, { "epoch": 0.9105468992488671, "grad_norm": 0.558191541440781, "learning_rate": 8.790548429661058e-05, "loss": 3.0019, "step": 14668 }, { "epoch": 0.910608976348625, "grad_norm": 0.5000287286134485, "learning_rate": 8.79031290250813e-05, "loss": 2.9514, "step": 14669 }, { "epoch": 0.9106710534483828, "grad_norm": 0.5612350121176972, "learning_rate": 8.790077355580219e-05, "loss": 2.9808, "step": 14670 }, { "epoch": 0.9107331305481408, "grad_norm": 0.46170652215195773, "learning_rate": 8.789841788878552e-05, "loss": 2.8955, "step": 14671 }, { "epoch": 0.9107952076478987, "grad_norm": 0.5542312142299022, "learning_rate": 8.789606202404363e-05, "loss": 2.9599, "step": 14672 }, { "epoch": 0.9108572847476566, "grad_norm": 0.5877918654718421, "learning_rate": 8.78937059615888e-05, "loss": 2.8977, "step": 14673 }, { "epoch": 0.9109193618474145, "grad_norm": 0.45665270460612856, "learning_rate": 8.789134970143327e-05, "loss": 2.948, "step": 14674 }, { "epoch": 0.9109814389471724, "grad_norm": 0.5297080832113992, "learning_rate": 8.78889932435894e-05, "loss": 2.8822, "step": 14675 }, { "epoch": 0.9110435160469302, "grad_norm": 0.48688397182576865, "learning_rate": 8.788663658806941e-05, "loss": 2.9975, "step": 14676 }, { "epoch": 0.9111055931466882, "grad_norm": 0.5146715133676376, "learning_rate": 8.788427973488569e-05, "loss": 2.9757, "step": 14677 }, { "epoch": 0.9111676702464461, "grad_norm": 0.5049291179975032, "learning_rate": 8.788192268405046e-05, "loss": 2.8819, "step": 14678 }, { "epoch": 0.911229747346204, "grad_norm": 0.5155521688176503, "learning_rate": 8.787956543557605e-05, "loss": 2.891, "step": 14679 }, { "epoch": 0.9112918244459619, "grad_norm": 0.5246651898855699, "learning_rate": 8.787720798947474e-05, "loss": 2.9386, "step": 14680 }, { "epoch": 0.9113539015457198, "grad_norm": 0.5253932343031744, "learning_rate": 8.787485034575886e-05, "loss": 3.0016, "step": 14681 }, { "epoch": 0.9114159786454776, "grad_norm": 0.49179658048952724, "learning_rate": 8.787249250444067e-05, "loss": 2.9644, "step": 14682 }, { "epoch": 0.9114780557452355, "grad_norm": 0.49066737228915974, "learning_rate": 8.787013446553251e-05, "loss": 2.9418, "step": 14683 }, { "epoch": 0.9115401328449935, "grad_norm": 0.503612817302118, "learning_rate": 8.786777622904666e-05, "loss": 2.9549, "step": 14684 }, { "epoch": 0.9116022099447514, "grad_norm": 0.5128632188783173, "learning_rate": 8.786541779499543e-05, "loss": 3.0124, "step": 14685 }, { "epoch": 0.9116642870445093, "grad_norm": 0.623981002773669, "learning_rate": 8.786305916339113e-05, "loss": 2.9759, "step": 14686 }, { "epoch": 0.9117263641442672, "grad_norm": 0.4452987969950427, "learning_rate": 8.786070033424604e-05, "loss": 2.9788, "step": 14687 }, { "epoch": 0.911788441244025, "grad_norm": 0.5733610481625553, "learning_rate": 8.78583413075725e-05, "loss": 2.9711, "step": 14688 }, { "epoch": 0.911850518343783, "grad_norm": 0.5335012587538709, "learning_rate": 8.785598208338279e-05, "loss": 2.93, "step": 14689 }, { "epoch": 0.9119125954435409, "grad_norm": 0.49841541791450233, "learning_rate": 8.785362266168925e-05, "loss": 3.0356, "step": 14690 }, { "epoch": 0.9119746725432988, "grad_norm": 0.5462409294562514, "learning_rate": 8.785126304250416e-05, "loss": 2.8732, "step": 14691 }, { "epoch": 0.9120367496430567, "grad_norm": 0.5882993272331731, "learning_rate": 8.784890322583984e-05, "loss": 2.9181, "step": 14692 }, { "epoch": 0.9120988267428146, "grad_norm": 0.505390111593026, "learning_rate": 8.78465432117086e-05, "loss": 2.9536, "step": 14693 }, { "epoch": 0.9121609038425724, "grad_norm": 0.5567581588068512, "learning_rate": 8.784418300012277e-05, "loss": 2.9797, "step": 14694 }, { "epoch": 0.9122229809423303, "grad_norm": 0.4839414378925662, "learning_rate": 8.784182259109465e-05, "loss": 2.9333, "step": 14695 }, { "epoch": 0.9122850580420883, "grad_norm": 0.5234272514592553, "learning_rate": 8.783946198463655e-05, "loss": 2.9524, "step": 14696 }, { "epoch": 0.9123471351418462, "grad_norm": 0.5242621914051346, "learning_rate": 8.783710118076078e-05, "loss": 2.9925, "step": 14697 }, { "epoch": 0.9124092122416041, "grad_norm": 0.5800269261637747, "learning_rate": 8.783474017947968e-05, "loss": 2.9823, "step": 14698 }, { "epoch": 0.912471289341362, "grad_norm": 0.5381862595322037, "learning_rate": 8.783237898080557e-05, "loss": 3.0703, "step": 14699 }, { "epoch": 0.9125333664411198, "grad_norm": 0.5816483486218662, "learning_rate": 8.783001758475073e-05, "loss": 3.0167, "step": 14700 }, { "epoch": 0.9125954435408777, "grad_norm": 0.4484405577286204, "learning_rate": 8.782765599132751e-05, "loss": 3.0198, "step": 14701 }, { "epoch": 0.9126575206406357, "grad_norm": 0.48481095599279883, "learning_rate": 8.782529420054823e-05, "loss": 2.9212, "step": 14702 }, { "epoch": 0.9127195977403936, "grad_norm": 0.49532360461569946, "learning_rate": 8.782293221242521e-05, "loss": 2.9745, "step": 14703 }, { "epoch": 0.9127816748401515, "grad_norm": 0.47297645000520006, "learning_rate": 8.782057002697076e-05, "loss": 2.9714, "step": 14704 }, { "epoch": 0.9128437519399094, "grad_norm": 0.5291817177361666, "learning_rate": 8.781820764419721e-05, "loss": 2.9137, "step": 14705 }, { "epoch": 0.9129058290396672, "grad_norm": 0.511288702416233, "learning_rate": 8.781584506411691e-05, "loss": 3.0686, "step": 14706 }, { "epoch": 0.9129679061394251, "grad_norm": 0.526196312283781, "learning_rate": 8.781348228674217e-05, "loss": 3.0015, "step": 14707 }, { "epoch": 0.913029983239183, "grad_norm": 0.46906217670365996, "learning_rate": 8.78111193120853e-05, "loss": 2.8555, "step": 14708 }, { "epoch": 0.913092060338941, "grad_norm": 0.46285930157360716, "learning_rate": 8.780875614015865e-05, "loss": 2.9483, "step": 14709 }, { "epoch": 0.9131541374386989, "grad_norm": 0.4988178045987103, "learning_rate": 8.780639277097454e-05, "loss": 2.9903, "step": 14710 }, { "epoch": 0.9132162145384568, "grad_norm": 0.48900734892103076, "learning_rate": 8.780402920454529e-05, "loss": 2.9936, "step": 14711 }, { "epoch": 0.9132782916382146, "grad_norm": 0.44989828046218694, "learning_rate": 8.780166544088325e-05, "loss": 2.9487, "step": 14712 }, { "epoch": 0.9133403687379725, "grad_norm": 0.5592032621785817, "learning_rate": 8.779930148000073e-05, "loss": 2.9963, "step": 14713 }, { "epoch": 0.9134024458377304, "grad_norm": 0.5201298126625121, "learning_rate": 8.77969373219101e-05, "loss": 2.9606, "step": 14714 }, { "epoch": 0.9134645229374884, "grad_norm": 0.5044010305955298, "learning_rate": 8.779457296662366e-05, "loss": 2.9909, "step": 14715 }, { "epoch": 0.9135266000372463, "grad_norm": 0.501734377921461, "learning_rate": 8.779220841415375e-05, "loss": 2.9663, "step": 14716 }, { "epoch": 0.9135886771370042, "grad_norm": 0.5000405295158056, "learning_rate": 8.778984366451273e-05, "loss": 3.0041, "step": 14717 }, { "epoch": 0.913650754236762, "grad_norm": 0.5306269414032736, "learning_rate": 8.778747871771292e-05, "loss": 2.856, "step": 14718 }, { "epoch": 0.9137128313365199, "grad_norm": 0.44403993577426804, "learning_rate": 8.778511357376667e-05, "loss": 2.9083, "step": 14719 }, { "epoch": 0.9137749084362778, "grad_norm": 0.4776510251713767, "learning_rate": 8.778274823268628e-05, "loss": 2.8952, "step": 14720 }, { "epoch": 0.9138369855360358, "grad_norm": 0.4701569740439431, "learning_rate": 8.778038269448414e-05, "loss": 2.9508, "step": 14721 }, { "epoch": 0.9138990626357937, "grad_norm": 0.4969246431658698, "learning_rate": 8.777801695917258e-05, "loss": 2.905, "step": 14722 }, { "epoch": 0.9139611397355516, "grad_norm": 0.5225648865724544, "learning_rate": 8.777565102676392e-05, "loss": 2.9831, "step": 14723 }, { "epoch": 0.9140232168353094, "grad_norm": 0.4704720970903854, "learning_rate": 8.777328489727053e-05, "loss": 2.9143, "step": 14724 }, { "epoch": 0.9140852939350673, "grad_norm": 0.5142530337472759, "learning_rate": 8.777091857070472e-05, "loss": 3.0104, "step": 14725 }, { "epoch": 0.9141473710348252, "grad_norm": 0.49517484852080784, "learning_rate": 8.776855204707888e-05, "loss": 2.8718, "step": 14726 }, { "epoch": 0.9142094481345832, "grad_norm": 0.5402452237466632, "learning_rate": 8.776618532640533e-05, "loss": 2.9603, "step": 14727 }, { "epoch": 0.9142715252343411, "grad_norm": 0.5578059530007539, "learning_rate": 8.776381840869642e-05, "loss": 3.0839, "step": 14728 }, { "epoch": 0.914333602334099, "grad_norm": 0.5795012151608304, "learning_rate": 8.776145129396451e-05, "loss": 2.9674, "step": 14729 }, { "epoch": 0.9143956794338568, "grad_norm": 0.5201062627111284, "learning_rate": 8.775908398222192e-05, "loss": 3.0285, "step": 14730 }, { "epoch": 0.9144577565336147, "grad_norm": 0.4864475455983747, "learning_rate": 8.775671647348105e-05, "loss": 3.0005, "step": 14731 }, { "epoch": 0.9145198336333726, "grad_norm": 0.4818340225707413, "learning_rate": 8.77543487677542e-05, "loss": 2.9806, "step": 14732 }, { "epoch": 0.9145819107331306, "grad_norm": 0.561147680195168, "learning_rate": 8.775198086505376e-05, "loss": 2.9826, "step": 14733 }, { "epoch": 0.9146439878328885, "grad_norm": 0.4918447996850958, "learning_rate": 8.774961276539209e-05, "loss": 3.0294, "step": 14734 }, { "epoch": 0.9147060649326464, "grad_norm": 0.514548188784185, "learning_rate": 8.77472444687815e-05, "loss": 2.9968, "step": 14735 }, { "epoch": 0.9147681420324042, "grad_norm": 0.5514515524739012, "learning_rate": 8.774487597523437e-05, "loss": 2.9818, "step": 14736 }, { "epoch": 0.9148302191321621, "grad_norm": 0.4082201173886915, "learning_rate": 8.774250728476308e-05, "loss": 2.9122, "step": 14737 }, { "epoch": 0.91489229623192, "grad_norm": 0.5023728037451817, "learning_rate": 8.774013839737995e-05, "loss": 3.0445, "step": 14738 }, { "epoch": 0.914954373331678, "grad_norm": 0.5056266101863388, "learning_rate": 8.773776931309736e-05, "loss": 2.9174, "step": 14739 }, { "epoch": 0.9150164504314359, "grad_norm": 0.5045885301989628, "learning_rate": 8.773540003192767e-05, "loss": 2.9632, "step": 14740 }, { "epoch": 0.9150785275311938, "grad_norm": 0.5008576607098396, "learning_rate": 8.773303055388324e-05, "loss": 2.9386, "step": 14741 }, { "epoch": 0.9151406046309516, "grad_norm": 0.5087115032805395, "learning_rate": 8.773066087897644e-05, "loss": 3.0411, "step": 14742 }, { "epoch": 0.9152026817307095, "grad_norm": 0.4504575151993501, "learning_rate": 8.77282910072196e-05, "loss": 2.9708, "step": 14743 }, { "epoch": 0.9152647588304674, "grad_norm": 0.5633345347265641, "learning_rate": 8.77259209386251e-05, "loss": 3.0308, "step": 14744 }, { "epoch": 0.9153268359302253, "grad_norm": 0.44883346620839376, "learning_rate": 8.772355067320535e-05, "loss": 3.0153, "step": 14745 }, { "epoch": 0.9153889130299833, "grad_norm": 0.44281067693707904, "learning_rate": 8.772118021097265e-05, "loss": 2.9617, "step": 14746 }, { "epoch": 0.9154509901297412, "grad_norm": 0.5076211715606478, "learning_rate": 8.771880955193939e-05, "loss": 3.0013, "step": 14747 }, { "epoch": 0.915513067229499, "grad_norm": 0.5183079881645363, "learning_rate": 8.771643869611796e-05, "loss": 2.9813, "step": 14748 }, { "epoch": 0.9155751443292569, "grad_norm": 0.46704867287404267, "learning_rate": 8.77140676435207e-05, "loss": 2.9495, "step": 14749 }, { "epoch": 0.9156372214290148, "grad_norm": 0.47381022168683395, "learning_rate": 8.771169639416002e-05, "loss": 3.0017, "step": 14750 }, { "epoch": 0.9156992985287727, "grad_norm": 0.5009687991962904, "learning_rate": 8.770932494804824e-05, "loss": 2.9829, "step": 14751 }, { "epoch": 0.9157613756285307, "grad_norm": 0.5051467471256923, "learning_rate": 8.770695330519777e-05, "loss": 2.9851, "step": 14752 }, { "epoch": 0.9158234527282886, "grad_norm": 0.5118711113694709, "learning_rate": 8.770458146562096e-05, "loss": 2.9509, "step": 14753 }, { "epoch": 0.9158855298280464, "grad_norm": 0.5122374897383815, "learning_rate": 8.770220942933021e-05, "loss": 3.008, "step": 14754 }, { "epoch": 0.9159476069278043, "grad_norm": 0.49564162249308563, "learning_rate": 8.769983719633788e-05, "loss": 2.965, "step": 14755 }, { "epoch": 0.9160096840275622, "grad_norm": 0.5385962695313772, "learning_rate": 8.769746476665632e-05, "loss": 3.0541, "step": 14756 }, { "epoch": 0.9160717611273201, "grad_norm": 0.5102808018947047, "learning_rate": 8.769509214029796e-05, "loss": 2.9125, "step": 14757 }, { "epoch": 0.9161338382270781, "grad_norm": 0.5118809799571642, "learning_rate": 8.769271931727515e-05, "loss": 2.9545, "step": 14758 }, { "epoch": 0.916195915326836, "grad_norm": 1.3610017185410546, "learning_rate": 8.769034629760026e-05, "loss": 2.9838, "step": 14759 }, { "epoch": 0.9162579924265938, "grad_norm": 0.8699111437638825, "learning_rate": 8.768797308128572e-05, "loss": 2.954, "step": 14760 }, { "epoch": 0.9163200695263517, "grad_norm": 0.7184426023510332, "learning_rate": 8.768559966834384e-05, "loss": 2.9942, "step": 14761 }, { "epoch": 0.9163821466261096, "grad_norm": 0.6720090998405293, "learning_rate": 8.768322605878705e-05, "loss": 2.9585, "step": 14762 }, { "epoch": 0.9164442237258675, "grad_norm": 0.6197223036818387, "learning_rate": 8.768085225262772e-05, "loss": 2.8878, "step": 14763 }, { "epoch": 0.9165063008256255, "grad_norm": 0.6249366966613984, "learning_rate": 8.767847824987823e-05, "loss": 3.014, "step": 14764 }, { "epoch": 0.9165683779253834, "grad_norm": 0.618326849693518, "learning_rate": 8.767610405055098e-05, "loss": 3.0511, "step": 14765 }, { "epoch": 0.9166304550251412, "grad_norm": 0.596156413733015, "learning_rate": 8.767372965465835e-05, "loss": 2.9319, "step": 14766 }, { "epoch": 0.9166925321248991, "grad_norm": 0.6071617084469663, "learning_rate": 8.767135506221273e-05, "loss": 2.9572, "step": 14767 }, { "epoch": 0.916754609224657, "grad_norm": 0.6024192554929938, "learning_rate": 8.76689802732265e-05, "loss": 2.9632, "step": 14768 }, { "epoch": 0.9168166863244149, "grad_norm": 0.5684586795428452, "learning_rate": 8.766660528771204e-05, "loss": 2.9074, "step": 14769 }, { "epoch": 0.9168787634241728, "grad_norm": 0.521224510795484, "learning_rate": 8.766423010568179e-05, "loss": 2.929, "step": 14770 }, { "epoch": 0.9169408405239308, "grad_norm": 0.5902032830098234, "learning_rate": 8.766185472714808e-05, "loss": 2.9601, "step": 14771 }, { "epoch": 0.9170029176236886, "grad_norm": 0.5063859709881554, "learning_rate": 8.765947915212334e-05, "loss": 3.0102, "step": 14772 }, { "epoch": 0.9170649947234465, "grad_norm": 0.5848994151068306, "learning_rate": 8.765710338061996e-05, "loss": 3.0057, "step": 14773 }, { "epoch": 0.9171270718232044, "grad_norm": 0.5514931102903354, "learning_rate": 8.765472741265032e-05, "loss": 2.9083, "step": 14774 }, { "epoch": 0.9171891489229623, "grad_norm": 0.5675090445835448, "learning_rate": 8.765235124822684e-05, "loss": 2.9319, "step": 14775 }, { "epoch": 0.9172512260227202, "grad_norm": 0.5277716984175624, "learning_rate": 8.764997488736189e-05, "loss": 2.9707, "step": 14776 }, { "epoch": 0.9173133031224782, "grad_norm": 0.6102278194873867, "learning_rate": 8.764759833006789e-05, "loss": 2.9136, "step": 14777 }, { "epoch": 0.917375380222236, "grad_norm": 0.5176667047671721, "learning_rate": 8.764522157635723e-05, "loss": 2.8665, "step": 14778 }, { "epoch": 0.9174374573219939, "grad_norm": 0.5778532678345202, "learning_rate": 8.76428446262423e-05, "loss": 2.9195, "step": 14779 }, { "epoch": 0.9174995344217518, "grad_norm": 0.5533946107586801, "learning_rate": 8.764046747973554e-05, "loss": 3.0441, "step": 14780 }, { "epoch": 0.9175616115215097, "grad_norm": 0.5601226583754108, "learning_rate": 8.76380901368493e-05, "loss": 3.0199, "step": 14781 }, { "epoch": 0.9176236886212676, "grad_norm": 0.5371936523901827, "learning_rate": 8.7635712597596e-05, "loss": 2.8613, "step": 14782 }, { "epoch": 0.9176857657210256, "grad_norm": 0.472159510855284, "learning_rate": 8.763333486198805e-05, "loss": 2.9546, "step": 14783 }, { "epoch": 0.9177478428207834, "grad_norm": 0.579088439278081, "learning_rate": 8.763095693003788e-05, "loss": 2.957, "step": 14784 }, { "epoch": 0.9178099199205413, "grad_norm": 0.52939497823297, "learning_rate": 8.762857880175786e-05, "loss": 3.0496, "step": 14785 }, { "epoch": 0.9178719970202992, "grad_norm": 0.5359802862943924, "learning_rate": 8.76262004771604e-05, "loss": 2.9701, "step": 14786 }, { "epoch": 0.9179340741200571, "grad_norm": 0.5580698237212524, "learning_rate": 8.762382195625794e-05, "loss": 2.9912, "step": 14787 }, { "epoch": 0.917996151219815, "grad_norm": 0.5050650291780794, "learning_rate": 8.762144323906285e-05, "loss": 2.9569, "step": 14788 }, { "epoch": 0.918058228319573, "grad_norm": 0.5171220317328459, "learning_rate": 8.761906432558756e-05, "loss": 2.8838, "step": 14789 }, { "epoch": 0.9181203054193308, "grad_norm": 0.4759198373734467, "learning_rate": 8.761668521584447e-05, "loss": 2.9568, "step": 14790 }, { "epoch": 0.9181823825190887, "grad_norm": 0.5106638744409052, "learning_rate": 8.761430590984603e-05, "loss": 2.9645, "step": 14791 }, { "epoch": 0.9182444596188466, "grad_norm": 0.5480603043099722, "learning_rate": 8.76119264076046e-05, "loss": 2.9217, "step": 14792 }, { "epoch": 0.9183065367186045, "grad_norm": 0.4972709445713784, "learning_rate": 8.760954670913262e-05, "loss": 2.9404, "step": 14793 }, { "epoch": 0.9183686138183624, "grad_norm": 0.5445413662423662, "learning_rate": 8.760716681444253e-05, "loss": 2.8415, "step": 14794 }, { "epoch": 0.9184306909181204, "grad_norm": 0.50237161728137, "learning_rate": 8.760478672354668e-05, "loss": 2.9664, "step": 14795 }, { "epoch": 0.9184927680178782, "grad_norm": 0.5099750249488189, "learning_rate": 8.760240643645756e-05, "loss": 2.8536, "step": 14796 }, { "epoch": 0.9185548451176361, "grad_norm": 0.5467910702197375, "learning_rate": 8.760002595318754e-05, "loss": 3.1054, "step": 14797 }, { "epoch": 0.918616922217394, "grad_norm": 0.44991833488872, "learning_rate": 8.759764527374907e-05, "loss": 2.8572, "step": 14798 }, { "epoch": 0.9186789993171519, "grad_norm": 0.5093413593440453, "learning_rate": 8.759526439815455e-05, "loss": 2.9789, "step": 14799 }, { "epoch": 0.9187410764169098, "grad_norm": 0.5259469075985016, "learning_rate": 8.759288332641642e-05, "loss": 2.9229, "step": 14800 }, { "epoch": 0.9188031535166677, "grad_norm": 0.4444842080699857, "learning_rate": 8.759050205854709e-05, "loss": 2.9284, "step": 14801 }, { "epoch": 0.9188652306164256, "grad_norm": 0.4805448673400467, "learning_rate": 8.758812059455896e-05, "loss": 2.9046, "step": 14802 }, { "epoch": 0.9189273077161835, "grad_norm": 0.4958602776405746, "learning_rate": 8.758573893446452e-05, "loss": 3.0159, "step": 14803 }, { "epoch": 0.9189893848159414, "grad_norm": 0.517296192468435, "learning_rate": 8.758335707827612e-05, "loss": 3.0733, "step": 14804 }, { "epoch": 0.9190514619156993, "grad_norm": 0.4868473360725493, "learning_rate": 8.758097502600623e-05, "loss": 3.0238, "step": 14805 }, { "epoch": 0.9191135390154572, "grad_norm": 0.5066843559550613, "learning_rate": 8.757859277766729e-05, "loss": 2.9501, "step": 14806 }, { "epoch": 0.9191756161152151, "grad_norm": 0.4994367019242131, "learning_rate": 8.75762103332717e-05, "loss": 3.0032, "step": 14807 }, { "epoch": 0.919237693214973, "grad_norm": 0.49859437955017416, "learning_rate": 8.757382769283189e-05, "loss": 2.9671, "step": 14808 }, { "epoch": 0.9192997703147309, "grad_norm": 0.5106241589449396, "learning_rate": 8.75714448563603e-05, "loss": 2.9205, "step": 14809 }, { "epoch": 0.9193618474144888, "grad_norm": 0.4599339017596402, "learning_rate": 8.756906182386938e-05, "loss": 3.0042, "step": 14810 }, { "epoch": 0.9194239245142467, "grad_norm": 0.5415403114787741, "learning_rate": 8.756667859537153e-05, "loss": 2.9337, "step": 14811 }, { "epoch": 0.9194860016140046, "grad_norm": 0.47034915895262147, "learning_rate": 8.756429517087918e-05, "loss": 2.8891, "step": 14812 }, { "epoch": 0.9195480787137625, "grad_norm": 0.5142769096818605, "learning_rate": 8.75619115504048e-05, "loss": 2.9785, "step": 14813 }, { "epoch": 0.9196101558135203, "grad_norm": 0.43576738736933845, "learning_rate": 8.755952773396083e-05, "loss": 2.9151, "step": 14814 }, { "epoch": 0.9196722329132783, "grad_norm": 0.5498862875083194, "learning_rate": 8.755714372155965e-05, "loss": 3.0156, "step": 14815 }, { "epoch": 0.9197343100130362, "grad_norm": 0.49582960951499366, "learning_rate": 8.755475951321376e-05, "loss": 2.8813, "step": 14816 }, { "epoch": 0.9197963871127941, "grad_norm": 0.5251126883683714, "learning_rate": 8.755237510893556e-05, "loss": 2.8837, "step": 14817 }, { "epoch": 0.919858464212552, "grad_norm": 0.5054285578837081, "learning_rate": 8.754999050873752e-05, "loss": 2.9558, "step": 14818 }, { "epoch": 0.9199205413123099, "grad_norm": 0.43822026567945754, "learning_rate": 8.754760571263204e-05, "loss": 2.9067, "step": 14819 }, { "epoch": 0.9199826184120677, "grad_norm": 0.5456239013398635, "learning_rate": 8.754522072063161e-05, "loss": 2.9576, "step": 14820 }, { "epoch": 0.9200446955118257, "grad_norm": 0.5284013668364607, "learning_rate": 8.754283553274863e-05, "loss": 2.9653, "step": 14821 }, { "epoch": 0.9201067726115836, "grad_norm": 0.4988650613730074, "learning_rate": 8.754045014899557e-05, "loss": 2.86, "step": 14822 }, { "epoch": 0.9201688497113415, "grad_norm": 0.49736463904967876, "learning_rate": 8.753806456938486e-05, "loss": 2.9558, "step": 14823 }, { "epoch": 0.9202309268110994, "grad_norm": 0.4575131475925383, "learning_rate": 8.753567879392896e-05, "loss": 2.9674, "step": 14824 }, { "epoch": 0.9202930039108573, "grad_norm": 0.5075467808307665, "learning_rate": 8.753329282264031e-05, "loss": 2.9884, "step": 14825 }, { "epoch": 0.9203550810106151, "grad_norm": 0.47276183596278165, "learning_rate": 8.753090665553137e-05, "loss": 2.939, "step": 14826 }, { "epoch": 0.9204171581103731, "grad_norm": 0.5319670438185053, "learning_rate": 8.752852029261457e-05, "loss": 2.9864, "step": 14827 }, { "epoch": 0.920479235210131, "grad_norm": 0.49206158374597864, "learning_rate": 8.752613373390237e-05, "loss": 2.9414, "step": 14828 }, { "epoch": 0.9205413123098889, "grad_norm": 0.5122209661604669, "learning_rate": 8.752374697940721e-05, "loss": 3.0031, "step": 14829 }, { "epoch": 0.9206033894096468, "grad_norm": 0.46906486109089146, "learning_rate": 8.752136002914157e-05, "loss": 3.0238, "step": 14830 }, { "epoch": 0.9206654665094047, "grad_norm": 0.5198770103901842, "learning_rate": 8.751897288311788e-05, "loss": 2.9109, "step": 14831 }, { "epoch": 0.9207275436091625, "grad_norm": 0.494397665726759, "learning_rate": 8.751658554134861e-05, "loss": 2.9599, "step": 14832 }, { "epoch": 0.9207896207089205, "grad_norm": 0.49543810494314877, "learning_rate": 8.751419800384619e-05, "loss": 3.0042, "step": 14833 }, { "epoch": 0.9208516978086784, "grad_norm": 0.5128951782117426, "learning_rate": 8.75118102706231e-05, "loss": 3.0147, "step": 14834 }, { "epoch": 0.9209137749084363, "grad_norm": 0.4713004504905618, "learning_rate": 8.750942234169179e-05, "loss": 2.9156, "step": 14835 }, { "epoch": 0.9209758520081942, "grad_norm": 0.4878737618530576, "learning_rate": 8.750703421706472e-05, "loss": 3.0267, "step": 14836 }, { "epoch": 0.9210379291079521, "grad_norm": 0.48830348154956044, "learning_rate": 8.750464589675433e-05, "loss": 2.8393, "step": 14837 }, { "epoch": 0.9211000062077099, "grad_norm": 0.46890070399989936, "learning_rate": 8.750225738077311e-05, "loss": 2.9601, "step": 14838 }, { "epoch": 0.9211620833074678, "grad_norm": 0.5189429659605544, "learning_rate": 8.749986866913352e-05, "loss": 3.036, "step": 14839 }, { "epoch": 0.9212241604072258, "grad_norm": 0.5523874656719189, "learning_rate": 8.7497479761848e-05, "loss": 2.9121, "step": 14840 }, { "epoch": 0.9212862375069837, "grad_norm": 0.5079941058042208, "learning_rate": 8.749509065892903e-05, "loss": 3.0326, "step": 14841 }, { "epoch": 0.9213483146067416, "grad_norm": 0.4968650409012305, "learning_rate": 8.749270136038907e-05, "loss": 2.9801, "step": 14842 }, { "epoch": 0.9214103917064995, "grad_norm": 0.47910660042429154, "learning_rate": 8.749031186624059e-05, "loss": 3.0468, "step": 14843 }, { "epoch": 0.9214724688062573, "grad_norm": 0.5627765770291926, "learning_rate": 8.748792217649604e-05, "loss": 2.9646, "step": 14844 }, { "epoch": 0.9215345459060152, "grad_norm": 0.48825842231795297, "learning_rate": 8.74855322911679e-05, "loss": 3.0413, "step": 14845 }, { "epoch": 0.9215966230057732, "grad_norm": 0.5471295717905522, "learning_rate": 8.748314221026864e-05, "loss": 2.9765, "step": 14846 }, { "epoch": 0.9216587001055311, "grad_norm": 0.4513232154836691, "learning_rate": 8.748075193381073e-05, "loss": 2.9425, "step": 14847 }, { "epoch": 0.921720777205289, "grad_norm": 0.496794258611094, "learning_rate": 8.747836146180663e-05, "loss": 2.984, "step": 14848 }, { "epoch": 0.9217828543050469, "grad_norm": 0.4636458179894111, "learning_rate": 8.747597079426884e-05, "loss": 2.9498, "step": 14849 }, { "epoch": 0.9218449314048047, "grad_norm": 0.44624495340480297, "learning_rate": 8.747357993120979e-05, "loss": 2.9913, "step": 14850 }, { "epoch": 0.9219070085045626, "grad_norm": 0.495888464798474, "learning_rate": 8.747118887264199e-05, "loss": 2.9233, "step": 14851 }, { "epoch": 0.9219690856043206, "grad_norm": 0.46664405460013836, "learning_rate": 8.74687976185779e-05, "loss": 2.9181, "step": 14852 }, { "epoch": 0.9220311627040785, "grad_norm": 0.5938748680119417, "learning_rate": 8.746640616903e-05, "loss": 3.0543, "step": 14853 }, { "epoch": 0.9220932398038364, "grad_norm": 0.4458766509192093, "learning_rate": 8.746401452401077e-05, "loss": 2.975, "step": 14854 }, { "epoch": 0.9221553169035943, "grad_norm": 0.5535164950841124, "learning_rate": 8.746162268353266e-05, "loss": 3.0108, "step": 14855 }, { "epoch": 0.9222173940033521, "grad_norm": 0.4751444810176879, "learning_rate": 8.745923064760818e-05, "loss": 2.9687, "step": 14856 }, { "epoch": 0.92227947110311, "grad_norm": 0.48152825374241714, "learning_rate": 8.745683841624981e-05, "loss": 2.8915, "step": 14857 }, { "epoch": 0.922341548202868, "grad_norm": 0.4921938729252048, "learning_rate": 8.745444598947001e-05, "loss": 2.9476, "step": 14858 }, { "epoch": 0.9224036253026259, "grad_norm": 0.536768607498885, "learning_rate": 8.745205336728127e-05, "loss": 2.9578, "step": 14859 }, { "epoch": 0.9224657024023838, "grad_norm": 0.5075892911542673, "learning_rate": 8.744966054969607e-05, "loss": 3.0454, "step": 14860 }, { "epoch": 0.9225277795021417, "grad_norm": 0.5163174228191098, "learning_rate": 8.744726753672692e-05, "loss": 2.9142, "step": 14861 }, { "epoch": 0.9225898566018995, "grad_norm": 0.4554717526737649, "learning_rate": 8.744487432838629e-05, "loss": 3.0201, "step": 14862 }, { "epoch": 0.9226519337016574, "grad_norm": 0.548467931962763, "learning_rate": 8.744248092468664e-05, "loss": 2.9462, "step": 14863 }, { "epoch": 0.9227140108014154, "grad_norm": 0.4654198610690786, "learning_rate": 8.744008732564048e-05, "loss": 2.962, "step": 14864 }, { "epoch": 0.9227760879011733, "grad_norm": 0.4824776404887449, "learning_rate": 8.74376935312603e-05, "loss": 2.9856, "step": 14865 }, { "epoch": 0.9228381650009312, "grad_norm": 0.4918259397709577, "learning_rate": 8.743529954155858e-05, "loss": 3.0573, "step": 14866 }, { "epoch": 0.9229002421006891, "grad_norm": 0.5466730017217445, "learning_rate": 8.743290535654783e-05, "loss": 2.8924, "step": 14867 }, { "epoch": 0.9229623192004469, "grad_norm": 0.49870130141049995, "learning_rate": 8.743051097624051e-05, "loss": 2.9242, "step": 14868 }, { "epoch": 0.9230243963002048, "grad_norm": 0.4572699322962821, "learning_rate": 8.742811640064912e-05, "loss": 3.0483, "step": 14869 }, { "epoch": 0.9230864733999627, "grad_norm": 0.4630345149069611, "learning_rate": 8.742572162978617e-05, "loss": 2.9756, "step": 14870 }, { "epoch": 0.9231485504997207, "grad_norm": 0.45952188388238396, "learning_rate": 8.742332666366416e-05, "loss": 2.9683, "step": 14871 }, { "epoch": 0.9232106275994786, "grad_norm": 0.4673623207659825, "learning_rate": 8.742093150229556e-05, "loss": 2.9921, "step": 14872 }, { "epoch": 0.9232727046992365, "grad_norm": 0.5364468181324813, "learning_rate": 8.741853614569286e-05, "loss": 2.9926, "step": 14873 }, { "epoch": 0.9233347817989943, "grad_norm": 0.4875276624949679, "learning_rate": 8.741614059386858e-05, "loss": 3.0113, "step": 14874 }, { "epoch": 0.9233968588987522, "grad_norm": 0.4669157858348987, "learning_rate": 8.741374484683522e-05, "loss": 2.9404, "step": 14875 }, { "epoch": 0.9234589359985101, "grad_norm": 0.5881155603930267, "learning_rate": 8.741134890460527e-05, "loss": 2.9243, "step": 14876 }, { "epoch": 0.9235210130982681, "grad_norm": 0.46145061446279273, "learning_rate": 8.740895276719123e-05, "loss": 2.9643, "step": 14877 }, { "epoch": 0.923583090198026, "grad_norm": 0.5574917092583529, "learning_rate": 8.74065564346056e-05, "loss": 2.9987, "step": 14878 }, { "epoch": 0.9236451672977839, "grad_norm": 0.5285466356335846, "learning_rate": 8.74041599068609e-05, "loss": 2.9094, "step": 14879 }, { "epoch": 0.9237072443975417, "grad_norm": 0.5483634583576201, "learning_rate": 8.740176318396959e-05, "loss": 2.9866, "step": 14880 }, { "epoch": 0.9237693214972996, "grad_norm": 0.5230051219135048, "learning_rate": 8.739936626594423e-05, "loss": 2.956, "step": 14881 }, { "epoch": 0.9238313985970575, "grad_norm": 0.6149110216809681, "learning_rate": 8.739696915279727e-05, "loss": 2.9751, "step": 14882 }, { "epoch": 0.9238934756968155, "grad_norm": 0.5578898816971154, "learning_rate": 8.739457184454127e-05, "loss": 2.9848, "step": 14883 }, { "epoch": 0.9239555527965734, "grad_norm": 0.5075617537956263, "learning_rate": 8.739217434118868e-05, "loss": 2.9743, "step": 14884 }, { "epoch": 0.9240176298963313, "grad_norm": 0.505208720046482, "learning_rate": 8.738977664275205e-05, "loss": 2.9545, "step": 14885 }, { "epoch": 0.9240797069960891, "grad_norm": 0.4712171004014949, "learning_rate": 8.73873787492439e-05, "loss": 2.8553, "step": 14886 }, { "epoch": 0.924141784095847, "grad_norm": 0.5177790989345037, "learning_rate": 8.73849806606767e-05, "loss": 3.0015, "step": 14887 }, { "epoch": 0.9242038611956049, "grad_norm": 0.5727834604480704, "learning_rate": 8.738258237706298e-05, "loss": 3.0175, "step": 14888 }, { "epoch": 0.9242659382953629, "grad_norm": 0.4754502809907021, "learning_rate": 8.738018389841527e-05, "loss": 2.8372, "step": 14889 }, { "epoch": 0.9243280153951208, "grad_norm": 0.5016440839162531, "learning_rate": 8.737778522474605e-05, "loss": 2.951, "step": 14890 }, { "epoch": 0.9243900924948787, "grad_norm": 0.5210480597051842, "learning_rate": 8.737538635606785e-05, "loss": 2.9578, "step": 14891 }, { "epoch": 0.9244521695946365, "grad_norm": 0.5394159070765955, "learning_rate": 8.737298729239319e-05, "loss": 2.9527, "step": 14892 }, { "epoch": 0.9245142466943944, "grad_norm": 0.5193000508477157, "learning_rate": 8.737058803373456e-05, "loss": 3.0156, "step": 14893 }, { "epoch": 0.9245763237941523, "grad_norm": 0.5133543592436413, "learning_rate": 8.736818858010454e-05, "loss": 3.0134, "step": 14894 }, { "epoch": 0.9246384008939103, "grad_norm": 0.4863155164194878, "learning_rate": 8.736578893151559e-05, "loss": 2.8881, "step": 14895 }, { "epoch": 0.9247004779936682, "grad_norm": 0.5690538286515205, "learning_rate": 8.736338908798026e-05, "loss": 2.9637, "step": 14896 }, { "epoch": 0.9247625550934261, "grad_norm": 0.5272063747480069, "learning_rate": 8.736098904951105e-05, "loss": 2.9241, "step": 14897 }, { "epoch": 0.9248246321931839, "grad_norm": 0.5647273733884308, "learning_rate": 8.735858881612048e-05, "loss": 2.9574, "step": 14898 }, { "epoch": 0.9248867092929418, "grad_norm": 0.5416575730611952, "learning_rate": 8.73561883878211e-05, "loss": 2.9198, "step": 14899 }, { "epoch": 0.9249487863926997, "grad_norm": 0.49989113516592687, "learning_rate": 8.73537877646254e-05, "loss": 2.9818, "step": 14900 }, { "epoch": 0.9250108634924576, "grad_norm": 0.5719727948478281, "learning_rate": 8.735138694654592e-05, "loss": 2.9925, "step": 14901 }, { "epoch": 0.9250729405922156, "grad_norm": 0.46679454467702525, "learning_rate": 8.73489859335952e-05, "loss": 3.0297, "step": 14902 }, { "epoch": 0.9251350176919735, "grad_norm": 0.5727196932799622, "learning_rate": 8.734658472578576e-05, "loss": 2.9333, "step": 14903 }, { "epoch": 0.9251970947917313, "grad_norm": 0.5178743486411832, "learning_rate": 8.73441833231301e-05, "loss": 2.9078, "step": 14904 }, { "epoch": 0.9252591718914892, "grad_norm": 0.5163255902662612, "learning_rate": 8.734178172564079e-05, "loss": 2.8961, "step": 14905 }, { "epoch": 0.9253212489912471, "grad_norm": 0.4835447610128529, "learning_rate": 8.733937993333033e-05, "loss": 2.9812, "step": 14906 }, { "epoch": 0.925383326091005, "grad_norm": 0.6173972485648739, "learning_rate": 8.733697794621127e-05, "loss": 3.0186, "step": 14907 }, { "epoch": 0.925445403190763, "grad_norm": 0.4725083048660832, "learning_rate": 8.733457576429612e-05, "loss": 2.9819, "step": 14908 }, { "epoch": 0.9255074802905209, "grad_norm": 0.5641089581407007, "learning_rate": 8.733217338759742e-05, "loss": 2.8579, "step": 14909 }, { "epoch": 0.9255695573902787, "grad_norm": 0.5213278489102834, "learning_rate": 8.732977081612773e-05, "loss": 2.9517, "step": 14910 }, { "epoch": 0.9256316344900366, "grad_norm": 0.5602987983249549, "learning_rate": 8.732736804989954e-05, "loss": 3.0082, "step": 14911 }, { "epoch": 0.9256937115897945, "grad_norm": 0.50829779104623, "learning_rate": 8.732496508892542e-05, "loss": 2.943, "step": 14912 }, { "epoch": 0.9257557886895524, "grad_norm": 0.4857149356549186, "learning_rate": 8.732256193321791e-05, "loss": 2.8987, "step": 14913 }, { "epoch": 0.9258178657893104, "grad_norm": 0.5451300165408737, "learning_rate": 8.732015858278953e-05, "loss": 3.0022, "step": 14914 }, { "epoch": 0.9258799428890683, "grad_norm": 0.46914830769291893, "learning_rate": 8.73177550376528e-05, "loss": 2.8855, "step": 14915 }, { "epoch": 0.9259420199888261, "grad_norm": 0.5404276768874806, "learning_rate": 8.731535129782029e-05, "loss": 2.9695, "step": 14916 }, { "epoch": 0.926004097088584, "grad_norm": 0.5225541076296497, "learning_rate": 8.731294736330454e-05, "loss": 2.8889, "step": 14917 }, { "epoch": 0.9260661741883419, "grad_norm": 0.5038445013492923, "learning_rate": 8.731054323411809e-05, "loss": 2.9209, "step": 14918 }, { "epoch": 0.9261282512880998, "grad_norm": 0.5511641557938319, "learning_rate": 8.730813891027347e-05, "loss": 2.9401, "step": 14919 }, { "epoch": 0.9261903283878578, "grad_norm": 0.4757079764883233, "learning_rate": 8.730573439178324e-05, "loss": 2.9153, "step": 14920 }, { "epoch": 0.9262524054876157, "grad_norm": 0.49635949113624533, "learning_rate": 8.730332967865993e-05, "loss": 3.0077, "step": 14921 }, { "epoch": 0.9263144825873735, "grad_norm": 0.49547869183620874, "learning_rate": 8.730092477091609e-05, "loss": 2.9334, "step": 14922 }, { "epoch": 0.9263765596871314, "grad_norm": 0.513401380906516, "learning_rate": 8.729851966856429e-05, "loss": 2.9916, "step": 14923 }, { "epoch": 0.9264386367868893, "grad_norm": 0.45350295779389765, "learning_rate": 8.729611437161705e-05, "loss": 2.998, "step": 14924 }, { "epoch": 0.9265007138866472, "grad_norm": 0.48901385470618774, "learning_rate": 8.729370888008692e-05, "loss": 3.0371, "step": 14925 }, { "epoch": 0.9265627909864051, "grad_norm": 0.47196220640536124, "learning_rate": 8.729130319398647e-05, "loss": 3.0214, "step": 14926 }, { "epoch": 0.9266248680861631, "grad_norm": 0.5528895291776688, "learning_rate": 8.728889731332821e-05, "loss": 3.0443, "step": 14927 }, { "epoch": 0.9266869451859209, "grad_norm": 0.5195080744801501, "learning_rate": 8.728649123812474e-05, "loss": 2.9152, "step": 14928 }, { "epoch": 0.9267490222856788, "grad_norm": 0.5009089195584431, "learning_rate": 8.728408496838861e-05, "loss": 2.902, "step": 14929 }, { "epoch": 0.9268110993854367, "grad_norm": 0.5252184958144963, "learning_rate": 8.728167850413233e-05, "loss": 2.9139, "step": 14930 }, { "epoch": 0.9268731764851946, "grad_norm": 0.5364497347639517, "learning_rate": 8.727927184536851e-05, "loss": 2.9604, "step": 14931 }, { "epoch": 0.9269352535849525, "grad_norm": 0.4833681809945891, "learning_rate": 8.727686499210965e-05, "loss": 2.8602, "step": 14932 }, { "epoch": 0.9269973306847105, "grad_norm": 0.5054334519102601, "learning_rate": 8.727445794436834e-05, "loss": 2.9006, "step": 14933 }, { "epoch": 0.9270594077844683, "grad_norm": 0.5541554706800119, "learning_rate": 8.727205070215715e-05, "loss": 2.8893, "step": 14934 }, { "epoch": 0.9271214848842262, "grad_norm": 0.5719461949124917, "learning_rate": 8.726964326548863e-05, "loss": 3.0191, "step": 14935 }, { "epoch": 0.9271835619839841, "grad_norm": 0.4655669060891767, "learning_rate": 8.726723563437531e-05, "loss": 2.9036, "step": 14936 }, { "epoch": 0.927245639083742, "grad_norm": 0.5532853039709075, "learning_rate": 8.726482780882978e-05, "loss": 2.9349, "step": 14937 }, { "epoch": 0.9273077161834999, "grad_norm": 0.6358280153247959, "learning_rate": 8.72624197888646e-05, "loss": 2.9844, "step": 14938 }, { "epoch": 0.9273697932832579, "grad_norm": 0.49608733706196606, "learning_rate": 8.726001157449234e-05, "loss": 2.9406, "step": 14939 }, { "epoch": 0.9274318703830157, "grad_norm": 0.5664073812539215, "learning_rate": 8.725760316572554e-05, "loss": 2.9265, "step": 14940 }, { "epoch": 0.9274939474827736, "grad_norm": 0.5472097462260905, "learning_rate": 8.725519456257679e-05, "loss": 2.9087, "step": 14941 }, { "epoch": 0.9275560245825315, "grad_norm": 0.5036900670201179, "learning_rate": 8.725278576505865e-05, "loss": 2.93, "step": 14942 }, { "epoch": 0.9276181016822894, "grad_norm": 0.5559279849185399, "learning_rate": 8.725037677318366e-05, "loss": 3.001, "step": 14943 }, { "epoch": 0.9276801787820473, "grad_norm": 0.5187512041560526, "learning_rate": 8.724796758696444e-05, "loss": 2.9827, "step": 14944 }, { "epoch": 0.9277422558818053, "grad_norm": 0.48973583997981923, "learning_rate": 8.724555820641352e-05, "loss": 3.0031, "step": 14945 }, { "epoch": 0.9278043329815631, "grad_norm": 0.528248977781102, "learning_rate": 8.724314863154347e-05, "loss": 2.8855, "step": 14946 }, { "epoch": 0.927866410081321, "grad_norm": 0.5162500003694622, "learning_rate": 8.724073886236688e-05, "loss": 2.9121, "step": 14947 }, { "epoch": 0.9279284871810789, "grad_norm": 0.48057195283897786, "learning_rate": 8.723832889889631e-05, "loss": 3.0106, "step": 14948 }, { "epoch": 0.9279905642808368, "grad_norm": 0.46974175759718245, "learning_rate": 8.723591874114435e-05, "loss": 2.8603, "step": 14949 }, { "epoch": 0.9280526413805947, "grad_norm": 0.4828634519141676, "learning_rate": 8.723350838912356e-05, "loss": 2.9828, "step": 14950 }, { "epoch": 0.9281147184803527, "grad_norm": 0.5253615247190461, "learning_rate": 8.723109784284651e-05, "loss": 2.9372, "step": 14951 }, { "epoch": 0.9281767955801105, "grad_norm": 0.47315952210352774, "learning_rate": 8.722868710232579e-05, "loss": 2.8012, "step": 14952 }, { "epoch": 0.9282388726798684, "grad_norm": 0.5041255325950195, "learning_rate": 8.722627616757398e-05, "loss": 2.8346, "step": 14953 }, { "epoch": 0.9283009497796263, "grad_norm": 0.4551315603687052, "learning_rate": 8.722386503860363e-05, "loss": 2.9196, "step": 14954 }, { "epoch": 0.9283630268793842, "grad_norm": 0.5293083330441506, "learning_rate": 8.722145371542735e-05, "loss": 2.9644, "step": 14955 }, { "epoch": 0.9284251039791421, "grad_norm": 0.4623190493623231, "learning_rate": 8.721904219805772e-05, "loss": 2.9447, "step": 14956 }, { "epoch": 0.9284871810789, "grad_norm": 0.43865898432690964, "learning_rate": 8.72166304865073e-05, "loss": 2.8714, "step": 14957 }, { "epoch": 0.9285492581786579, "grad_norm": 0.4575298393659756, "learning_rate": 8.721421858078869e-05, "loss": 2.909, "step": 14958 }, { "epoch": 0.9286113352784158, "grad_norm": 0.45612246481456703, "learning_rate": 8.721180648091447e-05, "loss": 2.9899, "step": 14959 }, { "epoch": 0.9286734123781737, "grad_norm": 0.49498081774583214, "learning_rate": 8.72093941868972e-05, "loss": 2.9935, "step": 14960 }, { "epoch": 0.9287354894779316, "grad_norm": 0.4703757702924801, "learning_rate": 8.720698169874952e-05, "loss": 2.9224, "step": 14961 }, { "epoch": 0.9287975665776895, "grad_norm": 0.4696451699527855, "learning_rate": 8.720456901648396e-05, "loss": 2.9285, "step": 14962 }, { "epoch": 0.9288596436774474, "grad_norm": 0.5257632270886865, "learning_rate": 8.720215614011315e-05, "loss": 2.9617, "step": 14963 }, { "epoch": 0.9289217207772053, "grad_norm": 0.5017059548836454, "learning_rate": 8.719974306964964e-05, "loss": 2.9437, "step": 14964 }, { "epoch": 0.9289837978769632, "grad_norm": 0.559250878569962, "learning_rate": 8.719732980510605e-05, "loss": 2.9609, "step": 14965 }, { "epoch": 0.9290458749767211, "grad_norm": 0.5141372023455347, "learning_rate": 8.719491634649497e-05, "loss": 3.0497, "step": 14966 }, { "epoch": 0.929107952076479, "grad_norm": 0.5326547003063554, "learning_rate": 8.719250269382897e-05, "loss": 2.946, "step": 14967 }, { "epoch": 0.9291700291762369, "grad_norm": 0.5229497601446631, "learning_rate": 8.719008884712066e-05, "loss": 3.0033, "step": 14968 }, { "epoch": 0.9292321062759948, "grad_norm": 0.5468189210749077, "learning_rate": 8.718767480638261e-05, "loss": 2.9734, "step": 14969 }, { "epoch": 0.9292941833757526, "grad_norm": 0.5262124536580367, "learning_rate": 8.718526057162745e-05, "loss": 2.8669, "step": 14970 }, { "epoch": 0.9293562604755106, "grad_norm": 0.5241542577512921, "learning_rate": 8.718284614286777e-05, "loss": 2.9534, "step": 14971 }, { "epoch": 0.9294183375752685, "grad_norm": 0.5521918016319056, "learning_rate": 8.718043152011615e-05, "loss": 2.9357, "step": 14972 }, { "epoch": 0.9294804146750264, "grad_norm": 0.5485507359125021, "learning_rate": 8.717801670338517e-05, "loss": 3.0087, "step": 14973 }, { "epoch": 0.9295424917747843, "grad_norm": 0.5543878578750664, "learning_rate": 8.717560169268747e-05, "loss": 2.9581, "step": 14974 }, { "epoch": 0.9296045688745422, "grad_norm": 0.5568668909477885, "learning_rate": 8.717318648803563e-05, "loss": 2.9661, "step": 14975 }, { "epoch": 0.9296666459743, "grad_norm": 0.5227286820336142, "learning_rate": 8.717077108944225e-05, "loss": 2.8844, "step": 14976 }, { "epoch": 0.929728723074058, "grad_norm": 0.563824656844391, "learning_rate": 8.716835549691992e-05, "loss": 2.9849, "step": 14977 }, { "epoch": 0.9297908001738159, "grad_norm": 0.5464872620556083, "learning_rate": 8.716593971048129e-05, "loss": 2.9508, "step": 14978 }, { "epoch": 0.9298528772735738, "grad_norm": 0.5152589192014094, "learning_rate": 8.716352373013891e-05, "loss": 2.8808, "step": 14979 }, { "epoch": 0.9299149543733317, "grad_norm": 0.5415758613131244, "learning_rate": 8.716110755590541e-05, "loss": 2.9276, "step": 14980 }, { "epoch": 0.9299770314730896, "grad_norm": 0.5198342723114316, "learning_rate": 8.715869118779338e-05, "loss": 2.9289, "step": 14981 }, { "epoch": 0.9300391085728474, "grad_norm": 0.5687810899809619, "learning_rate": 8.715627462581543e-05, "loss": 2.907, "step": 14982 }, { "epoch": 0.9301011856726054, "grad_norm": 0.5363240830171201, "learning_rate": 8.715385786998421e-05, "loss": 2.8964, "step": 14983 }, { "epoch": 0.9301632627723633, "grad_norm": 0.5528299094090177, "learning_rate": 8.715144092031227e-05, "loss": 2.9226, "step": 14984 }, { "epoch": 0.9302253398721212, "grad_norm": 0.5198634813449284, "learning_rate": 8.714902377681224e-05, "loss": 2.9089, "step": 14985 }, { "epoch": 0.9302874169718791, "grad_norm": 0.6258770272900368, "learning_rate": 8.714660643949675e-05, "loss": 2.9374, "step": 14986 }, { "epoch": 0.9303494940716369, "grad_norm": 0.5646954184706873, "learning_rate": 8.714418890837838e-05, "loss": 2.9558, "step": 14987 }, { "epoch": 0.9304115711713948, "grad_norm": 0.5419710172350516, "learning_rate": 8.714177118346977e-05, "loss": 2.9629, "step": 14988 }, { "epoch": 0.9304736482711528, "grad_norm": 0.6151368563431293, "learning_rate": 8.713935326478352e-05, "loss": 3.0026, "step": 14989 }, { "epoch": 0.9305357253709107, "grad_norm": 0.5858424809023408, "learning_rate": 8.713693515233224e-05, "loss": 2.9854, "step": 14990 }, { "epoch": 0.9305978024706686, "grad_norm": 0.5203568917889236, "learning_rate": 8.713451684612857e-05, "loss": 2.9076, "step": 14991 }, { "epoch": 0.9306598795704265, "grad_norm": 0.5457209262661425, "learning_rate": 8.71320983461851e-05, "loss": 3.0104, "step": 14992 }, { "epoch": 0.9307219566701843, "grad_norm": 0.5241414929921652, "learning_rate": 8.712967965251446e-05, "loss": 2.9275, "step": 14993 }, { "epoch": 0.9307840337699422, "grad_norm": 0.4895089241730088, "learning_rate": 8.712726076512926e-05, "loss": 2.9788, "step": 14994 }, { "epoch": 0.9308461108697001, "grad_norm": 0.5383057465536568, "learning_rate": 8.712484168404215e-05, "loss": 2.9473, "step": 14995 }, { "epoch": 0.9309081879694581, "grad_norm": 0.5198043449654716, "learning_rate": 8.712242240926571e-05, "loss": 3.0564, "step": 14996 }, { "epoch": 0.930970265069216, "grad_norm": 0.5255897173113022, "learning_rate": 8.71200029408126e-05, "loss": 3.0048, "step": 14997 }, { "epoch": 0.9310323421689739, "grad_norm": 0.4882224238245927, "learning_rate": 8.711758327869542e-05, "loss": 2.8846, "step": 14998 }, { "epoch": 0.9310944192687317, "grad_norm": 0.7047221587596272, "learning_rate": 8.711516342292677e-05, "loss": 2.9205, "step": 14999 }, { "epoch": 0.9311564963684896, "grad_norm": 0.5274135402133321, "learning_rate": 8.711274337351932e-05, "loss": 2.9305, "step": 15000 }, { "epoch": 0.9312185734682475, "grad_norm": 0.5251429090817101, "learning_rate": 8.711032313048568e-05, "loss": 2.9876, "step": 15001 }, { "epoch": 0.9312806505680055, "grad_norm": 0.49729001748789475, "learning_rate": 8.710790269383848e-05, "loss": 2.8909, "step": 15002 }, { "epoch": 0.9313427276677634, "grad_norm": 0.508687424712803, "learning_rate": 8.710548206359034e-05, "loss": 2.8015, "step": 15003 }, { "epoch": 0.9314048047675213, "grad_norm": 0.5001901324688002, "learning_rate": 8.710306123975389e-05, "loss": 3.0446, "step": 15004 }, { "epoch": 0.9314668818672791, "grad_norm": 0.6017611039519561, "learning_rate": 8.710064022234176e-05, "loss": 2.9523, "step": 15005 }, { "epoch": 0.931528958967037, "grad_norm": 0.5421670717463997, "learning_rate": 8.709821901136659e-05, "loss": 2.9513, "step": 15006 }, { "epoch": 0.9315910360667949, "grad_norm": 0.534837307524822, "learning_rate": 8.709579760684099e-05, "loss": 2.8909, "step": 15007 }, { "epoch": 0.9316531131665529, "grad_norm": 0.5530578960512024, "learning_rate": 8.709337600877762e-05, "loss": 2.963, "step": 15008 }, { "epoch": 0.9317151902663108, "grad_norm": 0.5834345701335127, "learning_rate": 8.709095421718912e-05, "loss": 3.0359, "step": 15009 }, { "epoch": 0.9317772673660687, "grad_norm": 0.5853204148017593, "learning_rate": 8.708853223208808e-05, "loss": 2.9505, "step": 15010 }, { "epoch": 0.9318393444658265, "grad_norm": 0.46986750274688727, "learning_rate": 8.708611005348717e-05, "loss": 2.9899, "step": 15011 }, { "epoch": 0.9319014215655844, "grad_norm": 0.6111633273656681, "learning_rate": 8.708368768139903e-05, "loss": 3.053, "step": 15012 }, { "epoch": 0.9319634986653423, "grad_norm": 0.64077831387632, "learning_rate": 8.708126511583627e-05, "loss": 2.9127, "step": 15013 }, { "epoch": 0.9320255757651003, "grad_norm": 0.48383759866140336, "learning_rate": 8.707884235681158e-05, "loss": 3.0174, "step": 15014 }, { "epoch": 0.9320876528648582, "grad_norm": 0.5471000753176398, "learning_rate": 8.707641940433754e-05, "loss": 2.9786, "step": 15015 }, { "epoch": 0.9321497299646161, "grad_norm": 0.5177976034965031, "learning_rate": 8.707399625842683e-05, "loss": 3.1389, "step": 15016 }, { "epoch": 0.9322118070643739, "grad_norm": 0.5860019394090606, "learning_rate": 8.707157291909208e-05, "loss": 2.9714, "step": 15017 }, { "epoch": 0.9322738841641318, "grad_norm": 0.5705598529926016, "learning_rate": 8.706914938634593e-05, "loss": 2.9808, "step": 15018 }, { "epoch": 0.9323359612638897, "grad_norm": 0.5029314059626284, "learning_rate": 8.706672566020105e-05, "loss": 2.9904, "step": 15019 }, { "epoch": 0.9323980383636477, "grad_norm": 0.4739234740324708, "learning_rate": 8.706430174067004e-05, "loss": 2.9819, "step": 15020 }, { "epoch": 0.9324601154634056, "grad_norm": 0.5433135550161905, "learning_rate": 8.706187762776558e-05, "loss": 2.9152, "step": 15021 }, { "epoch": 0.9325221925631635, "grad_norm": 0.5024985530225123, "learning_rate": 8.705945332150031e-05, "loss": 2.896, "step": 15022 }, { "epoch": 0.9325842696629213, "grad_norm": 0.5435259861826925, "learning_rate": 8.705702882188687e-05, "loss": 2.8971, "step": 15023 }, { "epoch": 0.9326463467626792, "grad_norm": 0.5607687215396698, "learning_rate": 8.705460412893791e-05, "loss": 2.967, "step": 15024 }, { "epoch": 0.9327084238624371, "grad_norm": 0.5884712874830994, "learning_rate": 8.705217924266609e-05, "loss": 2.976, "step": 15025 }, { "epoch": 0.932770500962195, "grad_norm": 0.5006077767980072, "learning_rate": 8.704975416308406e-05, "loss": 2.9416, "step": 15026 }, { "epoch": 0.932832578061953, "grad_norm": 0.5020375755901729, "learning_rate": 8.704732889020445e-05, "loss": 2.9184, "step": 15027 }, { "epoch": 0.9328946551617109, "grad_norm": 0.5796253736575799, "learning_rate": 8.704490342403997e-05, "loss": 2.8969, "step": 15028 }, { "epoch": 0.9329567322614687, "grad_norm": 0.5411834181166759, "learning_rate": 8.704247776460321e-05, "loss": 3.0456, "step": 15029 }, { "epoch": 0.9330188093612266, "grad_norm": 0.5196483164665783, "learning_rate": 8.704005191190685e-05, "loss": 2.8736, "step": 15030 }, { "epoch": 0.9330808864609845, "grad_norm": 0.46832636127452265, "learning_rate": 8.703762586596356e-05, "loss": 2.8788, "step": 15031 }, { "epoch": 0.9331429635607424, "grad_norm": 0.510486662859529, "learning_rate": 8.703519962678596e-05, "loss": 2.9962, "step": 15032 }, { "epoch": 0.9332050406605004, "grad_norm": 0.4987642395166638, "learning_rate": 8.703277319438676e-05, "loss": 2.8431, "step": 15033 }, { "epoch": 0.9332671177602583, "grad_norm": 0.5018380177839943, "learning_rate": 8.703034656877857e-05, "loss": 3.0127, "step": 15034 }, { "epoch": 0.9333291948600161, "grad_norm": 0.5642477803776073, "learning_rate": 8.702791974997409e-05, "loss": 2.9112, "step": 15035 }, { "epoch": 0.933391271959774, "grad_norm": 0.4868974073050339, "learning_rate": 8.702549273798597e-05, "loss": 2.9215, "step": 15036 }, { "epoch": 0.9334533490595319, "grad_norm": 0.5722529959993942, "learning_rate": 8.702306553282684e-05, "loss": 2.9604, "step": 15037 }, { "epoch": 0.9335154261592898, "grad_norm": 0.4682672398851305, "learning_rate": 8.70206381345094e-05, "loss": 2.9273, "step": 15038 }, { "epoch": 0.9335775032590478, "grad_norm": 0.5526299529631317, "learning_rate": 8.701821054304632e-05, "loss": 3.0371, "step": 15039 }, { "epoch": 0.9336395803588057, "grad_norm": 0.4990803635707866, "learning_rate": 8.701578275845024e-05, "loss": 2.9627, "step": 15040 }, { "epoch": 0.9337016574585635, "grad_norm": 0.5420231165953361, "learning_rate": 8.701335478073382e-05, "loss": 3.0466, "step": 15041 }, { "epoch": 0.9337637345583214, "grad_norm": 0.5558173533015089, "learning_rate": 8.701092660990976e-05, "loss": 2.979, "step": 15042 }, { "epoch": 0.9338258116580793, "grad_norm": 0.5756892623816411, "learning_rate": 8.700849824599071e-05, "loss": 2.8295, "step": 15043 }, { "epoch": 0.9338878887578372, "grad_norm": 0.5353952138835745, "learning_rate": 8.700606968898933e-05, "loss": 2.9884, "step": 15044 }, { "epoch": 0.9339499658575952, "grad_norm": 0.5316022098068365, "learning_rate": 8.70036409389183e-05, "loss": 2.9575, "step": 15045 }, { "epoch": 0.9340120429573531, "grad_norm": 0.53840655348057, "learning_rate": 8.700121199579031e-05, "loss": 3.0587, "step": 15046 }, { "epoch": 0.9340741200571109, "grad_norm": 0.5819761775030887, "learning_rate": 8.6998782859618e-05, "loss": 2.9602, "step": 15047 }, { "epoch": 0.9341361971568688, "grad_norm": 0.5181586698444887, "learning_rate": 8.699635353041407e-05, "loss": 2.9512, "step": 15048 }, { "epoch": 0.9341982742566267, "grad_norm": 0.5668588342853578, "learning_rate": 8.699392400819118e-05, "loss": 2.9092, "step": 15049 }, { "epoch": 0.9342603513563846, "grad_norm": 0.504239142413072, "learning_rate": 8.6991494292962e-05, "loss": 2.9524, "step": 15050 }, { "epoch": 0.9343224284561426, "grad_norm": 0.5996482414049965, "learning_rate": 8.698906438473922e-05, "loss": 3.0528, "step": 15051 }, { "epoch": 0.9343845055559005, "grad_norm": 0.532083278582318, "learning_rate": 8.698663428353551e-05, "loss": 3.0906, "step": 15052 }, { "epoch": 0.9344465826556583, "grad_norm": 0.586480791622916, "learning_rate": 8.698420398936353e-05, "loss": 2.9869, "step": 15053 }, { "epoch": 0.9345086597554162, "grad_norm": 0.5048862066099497, "learning_rate": 8.6981773502236e-05, "loss": 2.9394, "step": 15054 }, { "epoch": 0.9345707368551741, "grad_norm": 0.4860216730204758, "learning_rate": 8.697934282216558e-05, "loss": 2.9498, "step": 15055 }, { "epoch": 0.934632813954932, "grad_norm": 0.49760294558406076, "learning_rate": 8.697691194916495e-05, "loss": 2.944, "step": 15056 }, { "epoch": 0.93469489105469, "grad_norm": 0.5202037216438385, "learning_rate": 8.69744808832468e-05, "loss": 2.9765, "step": 15057 }, { "epoch": 0.9347569681544479, "grad_norm": 0.46827544986771874, "learning_rate": 8.69720496244238e-05, "loss": 2.8641, "step": 15058 }, { "epoch": 0.9348190452542057, "grad_norm": 0.5653569729815667, "learning_rate": 8.696961817270863e-05, "loss": 2.8551, "step": 15059 }, { "epoch": 0.9348811223539636, "grad_norm": 0.4733063506720912, "learning_rate": 8.696718652811401e-05, "loss": 3.0046, "step": 15060 }, { "epoch": 0.9349431994537215, "grad_norm": 0.5349425007323095, "learning_rate": 8.69647546906526e-05, "loss": 2.9528, "step": 15061 }, { "epoch": 0.9350052765534794, "grad_norm": 0.5667055851764272, "learning_rate": 8.696232266033709e-05, "loss": 3.0975, "step": 15062 }, { "epoch": 0.9350673536532373, "grad_norm": 0.5700160784712381, "learning_rate": 8.695989043718016e-05, "loss": 2.9992, "step": 15063 }, { "epoch": 0.9351294307529953, "grad_norm": 0.5247081024596332, "learning_rate": 8.69574580211945e-05, "loss": 2.9616, "step": 15064 }, { "epoch": 0.9351915078527531, "grad_norm": 0.519463298053699, "learning_rate": 8.695502541239284e-05, "loss": 2.9629, "step": 15065 }, { "epoch": 0.935253584952511, "grad_norm": 0.5045555131122047, "learning_rate": 8.695259261078782e-05, "loss": 2.9088, "step": 15066 }, { "epoch": 0.9353156620522689, "grad_norm": 0.4523107678078584, "learning_rate": 8.695015961639215e-05, "loss": 2.869, "step": 15067 }, { "epoch": 0.9353777391520268, "grad_norm": 0.5229933263305484, "learning_rate": 8.694772642921853e-05, "loss": 3.0359, "step": 15068 }, { "epoch": 0.9354398162517847, "grad_norm": 0.4669340243429103, "learning_rate": 8.694529304927967e-05, "loss": 2.8912, "step": 15069 }, { "epoch": 0.9355018933515427, "grad_norm": 0.507654546967876, "learning_rate": 8.694285947658823e-05, "loss": 3.0141, "step": 15070 }, { "epoch": 0.9355639704513005, "grad_norm": 0.5273724866030942, "learning_rate": 8.694042571115694e-05, "loss": 3.0865, "step": 15071 }, { "epoch": 0.9356260475510584, "grad_norm": 0.4534618178195209, "learning_rate": 8.693799175299846e-05, "loss": 3.0019, "step": 15072 }, { "epoch": 0.9356881246508163, "grad_norm": 0.5786374887802691, "learning_rate": 8.693555760212553e-05, "loss": 2.9064, "step": 15073 }, { "epoch": 0.9357502017505742, "grad_norm": 0.45138219579804434, "learning_rate": 8.693312325855082e-05, "loss": 2.8921, "step": 15074 }, { "epoch": 0.9358122788503321, "grad_norm": 0.5225081377901502, "learning_rate": 8.693068872228704e-05, "loss": 2.9244, "step": 15075 }, { "epoch": 0.93587435595009, "grad_norm": 0.46004143595660196, "learning_rate": 8.69282539933469e-05, "loss": 2.8731, "step": 15076 }, { "epoch": 0.9359364330498479, "grad_norm": 0.5393430007689773, "learning_rate": 8.692581907174307e-05, "loss": 2.9428, "step": 15077 }, { "epoch": 0.9359985101496058, "grad_norm": 0.5701896587440932, "learning_rate": 8.69233839574883e-05, "loss": 2.9141, "step": 15078 }, { "epoch": 0.9360605872493637, "grad_norm": 0.46853364083842786, "learning_rate": 8.692094865059527e-05, "loss": 2.9062, "step": 15079 }, { "epoch": 0.9361226643491216, "grad_norm": 0.5215999460920206, "learning_rate": 8.691851315107668e-05, "loss": 3.0176, "step": 15080 }, { "epoch": 0.9361847414488795, "grad_norm": 0.5705490144441125, "learning_rate": 8.691607745894526e-05, "loss": 2.9931, "step": 15081 }, { "epoch": 0.9362468185486374, "grad_norm": 0.4865539363580327, "learning_rate": 8.691364157421369e-05, "loss": 2.943, "step": 15082 }, { "epoch": 0.9363088956483953, "grad_norm": 0.5612502785407991, "learning_rate": 8.691120549689468e-05, "loss": 2.9806, "step": 15083 }, { "epoch": 0.9363709727481532, "grad_norm": 0.555231928607301, "learning_rate": 8.690876922700097e-05, "loss": 2.9898, "step": 15084 }, { "epoch": 0.9364330498479111, "grad_norm": 0.5833442777787713, "learning_rate": 8.690633276454525e-05, "loss": 2.9245, "step": 15085 }, { "epoch": 0.936495126947669, "grad_norm": 0.4833463390288647, "learning_rate": 8.690389610954021e-05, "loss": 2.8759, "step": 15086 }, { "epoch": 0.9365572040474269, "grad_norm": 0.5028116624729159, "learning_rate": 8.690145926199862e-05, "loss": 2.9168, "step": 15087 }, { "epoch": 0.9366192811471848, "grad_norm": 0.5379303085770102, "learning_rate": 8.689902222193314e-05, "loss": 3.0197, "step": 15088 }, { "epoch": 0.9366813582469427, "grad_norm": 0.5857947366362111, "learning_rate": 8.689658498935649e-05, "loss": 3.0705, "step": 15089 }, { "epoch": 0.9367434353467006, "grad_norm": 0.47044718578610617, "learning_rate": 8.689414756428143e-05, "loss": 2.9362, "step": 15090 }, { "epoch": 0.9368055124464585, "grad_norm": 0.5201448244837885, "learning_rate": 8.689170994672063e-05, "loss": 2.9002, "step": 15091 }, { "epoch": 0.9368675895462164, "grad_norm": 0.4735215675825743, "learning_rate": 8.68892721366868e-05, "loss": 2.8677, "step": 15092 }, { "epoch": 0.9369296666459743, "grad_norm": 0.508411289188942, "learning_rate": 8.688683413419272e-05, "loss": 2.8927, "step": 15093 }, { "epoch": 0.9369917437457322, "grad_norm": 0.5064439084305372, "learning_rate": 8.688439593925105e-05, "loss": 2.9435, "step": 15094 }, { "epoch": 0.93705382084549, "grad_norm": 0.4505786380912398, "learning_rate": 8.688195755187454e-05, "loss": 2.847, "step": 15095 }, { "epoch": 0.937115897945248, "grad_norm": 0.44537568898861013, "learning_rate": 8.687951897207592e-05, "loss": 2.9435, "step": 15096 }, { "epoch": 0.9371779750450059, "grad_norm": 0.4840626786430527, "learning_rate": 8.687708019986787e-05, "loss": 2.9366, "step": 15097 }, { "epoch": 0.9372400521447638, "grad_norm": 0.4516933779483808, "learning_rate": 8.687464123526314e-05, "loss": 2.9764, "step": 15098 }, { "epoch": 0.9373021292445217, "grad_norm": 0.4506079877956843, "learning_rate": 8.687220207827446e-05, "loss": 2.9954, "step": 15099 }, { "epoch": 0.9373642063442796, "grad_norm": 0.4768882326162517, "learning_rate": 8.686976272891456e-05, "loss": 3.0186, "step": 15100 }, { "epoch": 0.9374262834440374, "grad_norm": 0.4865288683979833, "learning_rate": 8.686732318719616e-05, "loss": 2.9327, "step": 15101 }, { "epoch": 0.9374883605437954, "grad_norm": 0.5040657618985286, "learning_rate": 8.686488345313197e-05, "loss": 2.9305, "step": 15102 }, { "epoch": 0.9375504376435533, "grad_norm": 0.46958158162976943, "learning_rate": 8.686244352673474e-05, "loss": 2.9648, "step": 15103 }, { "epoch": 0.9376125147433112, "grad_norm": 0.5416945731481492, "learning_rate": 8.686000340801721e-05, "loss": 3.0452, "step": 15104 }, { "epoch": 0.9376745918430691, "grad_norm": 0.4851146096559533, "learning_rate": 8.685756309699206e-05, "loss": 2.94, "step": 15105 }, { "epoch": 0.937736668942827, "grad_norm": 0.4659768407371369, "learning_rate": 8.685512259367208e-05, "loss": 2.8715, "step": 15106 }, { "epoch": 0.9377987460425848, "grad_norm": 0.4734836146743189, "learning_rate": 8.685268189807e-05, "loss": 3.0505, "step": 15107 }, { "epoch": 0.9378608231423428, "grad_norm": 0.4525934093705581, "learning_rate": 8.685024101019849e-05, "loss": 2.8821, "step": 15108 }, { "epoch": 0.9379229002421007, "grad_norm": 0.5236626895063095, "learning_rate": 8.684779993007033e-05, "loss": 2.921, "step": 15109 }, { "epoch": 0.9379849773418586, "grad_norm": 0.45068570980678746, "learning_rate": 8.684535865769827e-05, "loss": 2.9484, "step": 15110 }, { "epoch": 0.9380470544416165, "grad_norm": 0.4844627147132053, "learning_rate": 8.684291719309503e-05, "loss": 2.8839, "step": 15111 }, { "epoch": 0.9381091315413744, "grad_norm": 0.45829793041943073, "learning_rate": 8.684047553627335e-05, "loss": 2.9957, "step": 15112 }, { "epoch": 0.9381712086411322, "grad_norm": 0.4889198247452002, "learning_rate": 8.683803368724594e-05, "loss": 3.0239, "step": 15113 }, { "epoch": 0.9382332857408902, "grad_norm": 0.539353665209505, "learning_rate": 8.683559164602559e-05, "loss": 2.9074, "step": 15114 }, { "epoch": 0.9382953628406481, "grad_norm": 0.4394013946603529, "learning_rate": 8.683314941262501e-05, "loss": 2.8968, "step": 15115 }, { "epoch": 0.938357439940406, "grad_norm": 0.46572721498689496, "learning_rate": 8.683070698705696e-05, "loss": 2.9574, "step": 15116 }, { "epoch": 0.9384195170401639, "grad_norm": 0.5937439767632401, "learning_rate": 8.682826436933415e-05, "loss": 2.9456, "step": 15117 }, { "epoch": 0.9384815941399218, "grad_norm": 0.524904898703811, "learning_rate": 8.682582155946935e-05, "loss": 2.9302, "step": 15118 }, { "epoch": 0.9385436712396796, "grad_norm": 0.5688917151157804, "learning_rate": 8.682337855747532e-05, "loss": 2.8839, "step": 15119 }, { "epoch": 0.9386057483394376, "grad_norm": 0.6441090212275639, "learning_rate": 8.682093536336475e-05, "loss": 2.9467, "step": 15120 }, { "epoch": 0.9386678254391955, "grad_norm": 0.5135878867756747, "learning_rate": 8.681849197715045e-05, "loss": 2.9675, "step": 15121 }, { "epoch": 0.9387299025389534, "grad_norm": 0.5832011606703303, "learning_rate": 8.681604839884512e-05, "loss": 2.9826, "step": 15122 }, { "epoch": 0.9387919796387113, "grad_norm": 0.49562227568292566, "learning_rate": 8.681360462846153e-05, "loss": 2.976, "step": 15123 }, { "epoch": 0.9388540567384692, "grad_norm": 0.5279159549757514, "learning_rate": 8.681116066601244e-05, "loss": 2.9473, "step": 15124 }, { "epoch": 0.938916133838227, "grad_norm": 0.5423021959251557, "learning_rate": 8.680871651151058e-05, "loss": 2.987, "step": 15125 }, { "epoch": 0.938978210937985, "grad_norm": 0.5236508233666264, "learning_rate": 8.68062721649687e-05, "loss": 2.9382, "step": 15126 }, { "epoch": 0.9390402880377429, "grad_norm": 0.4534332115080961, "learning_rate": 8.680382762639959e-05, "loss": 3.0475, "step": 15127 }, { "epoch": 0.9391023651375008, "grad_norm": 0.5167463420633257, "learning_rate": 8.680138289581596e-05, "loss": 2.9911, "step": 15128 }, { "epoch": 0.9391644422372587, "grad_norm": 0.5186759103008124, "learning_rate": 8.679893797323058e-05, "loss": 2.8595, "step": 15129 }, { "epoch": 0.9392265193370166, "grad_norm": 0.5054449497380031, "learning_rate": 8.67964928586562e-05, "loss": 2.9714, "step": 15130 }, { "epoch": 0.9392885964367744, "grad_norm": 0.5769944090332668, "learning_rate": 8.67940475521056e-05, "loss": 3.0106, "step": 15131 }, { "epoch": 0.9393506735365323, "grad_norm": 0.5303470005553698, "learning_rate": 8.679160205359153e-05, "loss": 2.9926, "step": 15132 }, { "epoch": 0.9394127506362903, "grad_norm": 0.4998061728297021, "learning_rate": 8.678915636312672e-05, "loss": 2.9351, "step": 15133 }, { "epoch": 0.9394748277360482, "grad_norm": 0.5236205449956345, "learning_rate": 8.678671048072397e-05, "loss": 3.017, "step": 15134 }, { "epoch": 0.9395369048358061, "grad_norm": 0.5397291898280399, "learning_rate": 8.6784264406396e-05, "loss": 2.9088, "step": 15135 }, { "epoch": 0.939598981935564, "grad_norm": 0.4711780289248243, "learning_rate": 8.67818181401556e-05, "loss": 3.0201, "step": 15136 }, { "epoch": 0.9396610590353218, "grad_norm": 0.5562744006687855, "learning_rate": 8.677937168201553e-05, "loss": 2.9485, "step": 15137 }, { "epoch": 0.9397231361350797, "grad_norm": 0.5042625764253419, "learning_rate": 8.677692503198856e-05, "loss": 2.9481, "step": 15138 }, { "epoch": 0.9397852132348377, "grad_norm": 0.5462719180513143, "learning_rate": 8.677447819008741e-05, "loss": 3.0674, "step": 15139 }, { "epoch": 0.9398472903345956, "grad_norm": 0.5226149572645188, "learning_rate": 8.67720311563249e-05, "loss": 2.9753, "step": 15140 }, { "epoch": 0.9399093674343535, "grad_norm": 0.5067192340181564, "learning_rate": 8.67695839307138e-05, "loss": 2.9834, "step": 15141 }, { "epoch": 0.9399714445341114, "grad_norm": 0.5312015286940673, "learning_rate": 8.676713651326683e-05, "loss": 2.9289, "step": 15142 }, { "epoch": 0.9400335216338692, "grad_norm": 0.4966289456842293, "learning_rate": 8.676468890399678e-05, "loss": 2.8856, "step": 15143 }, { "epoch": 0.9400955987336271, "grad_norm": 0.5762576540440705, "learning_rate": 8.676224110291643e-05, "loss": 2.8921, "step": 15144 }, { "epoch": 0.940157675833385, "grad_norm": 0.5645831793143237, "learning_rate": 8.675979311003854e-05, "loss": 2.9036, "step": 15145 }, { "epoch": 0.940219752933143, "grad_norm": 0.5371379775320977, "learning_rate": 8.67573449253759e-05, "loss": 3.0106, "step": 15146 }, { "epoch": 0.9402818300329009, "grad_norm": 0.4989931877508326, "learning_rate": 8.675489654894125e-05, "loss": 2.9637, "step": 15147 }, { "epoch": 0.9403439071326588, "grad_norm": 0.5171819934378954, "learning_rate": 8.675244798074739e-05, "loss": 2.969, "step": 15148 }, { "epoch": 0.9404059842324166, "grad_norm": 0.5135716677782065, "learning_rate": 8.674999922080708e-05, "loss": 3.0329, "step": 15149 }, { "epoch": 0.9404680613321745, "grad_norm": 0.5037695825050466, "learning_rate": 8.674755026913312e-05, "loss": 2.8323, "step": 15150 }, { "epoch": 0.9405301384319324, "grad_norm": 0.5264595814086145, "learning_rate": 8.674510112573826e-05, "loss": 2.9619, "step": 15151 }, { "epoch": 0.9405922155316904, "grad_norm": 0.5183361907810715, "learning_rate": 8.674265179063528e-05, "loss": 2.9517, "step": 15152 }, { "epoch": 0.9406542926314483, "grad_norm": 0.5083474496017478, "learning_rate": 8.674020226383697e-05, "loss": 2.8465, "step": 15153 }, { "epoch": 0.9407163697312062, "grad_norm": 0.49577450429978065, "learning_rate": 8.67377525453561e-05, "loss": 2.8925, "step": 15154 }, { "epoch": 0.940778446830964, "grad_norm": 0.460732899532193, "learning_rate": 8.673530263520547e-05, "loss": 2.8806, "step": 15155 }, { "epoch": 0.9408405239307219, "grad_norm": 0.4854633194957615, "learning_rate": 8.673285253339785e-05, "loss": 2.9636, "step": 15156 }, { "epoch": 0.9409026010304798, "grad_norm": 0.5562300121273753, "learning_rate": 8.673040223994601e-05, "loss": 2.9857, "step": 15157 }, { "epoch": 0.9409646781302378, "grad_norm": 0.5139768732059518, "learning_rate": 8.672795175486275e-05, "loss": 2.9958, "step": 15158 }, { "epoch": 0.9410267552299957, "grad_norm": 0.5435247524734831, "learning_rate": 8.672550107816084e-05, "loss": 2.9111, "step": 15159 }, { "epoch": 0.9410888323297536, "grad_norm": 0.45931898865507287, "learning_rate": 8.672305020985308e-05, "loss": 2.9508, "step": 15160 }, { "epoch": 0.9411509094295114, "grad_norm": 0.5195420916580533, "learning_rate": 8.672059914995226e-05, "loss": 2.834, "step": 15161 }, { "epoch": 0.9412129865292693, "grad_norm": 0.4527865329612998, "learning_rate": 8.671814789847116e-05, "loss": 2.9574, "step": 15162 }, { "epoch": 0.9412750636290272, "grad_norm": 0.5282037292576315, "learning_rate": 8.671569645542257e-05, "loss": 2.8767, "step": 15163 }, { "epoch": 0.9413371407287852, "grad_norm": 0.48537903982323793, "learning_rate": 8.671324482081927e-05, "loss": 2.9141, "step": 15164 }, { "epoch": 0.9413992178285431, "grad_norm": 0.5014162569363981, "learning_rate": 8.671079299467406e-05, "loss": 2.9619, "step": 15165 }, { "epoch": 0.941461294928301, "grad_norm": 0.518405727679862, "learning_rate": 8.670834097699974e-05, "loss": 2.9305, "step": 15166 }, { "epoch": 0.9415233720280588, "grad_norm": 0.4630969428754167, "learning_rate": 8.670588876780907e-05, "loss": 2.9283, "step": 15167 }, { "epoch": 0.9415854491278167, "grad_norm": 0.4853056304504404, "learning_rate": 8.670343636711489e-05, "loss": 2.9631, "step": 15168 }, { "epoch": 0.9416475262275746, "grad_norm": 0.49829384159555856, "learning_rate": 8.670098377492996e-05, "loss": 3.0964, "step": 15169 }, { "epoch": 0.9417096033273326, "grad_norm": 0.436283941943305, "learning_rate": 8.66985309912671e-05, "loss": 2.9712, "step": 15170 }, { "epoch": 0.9417716804270905, "grad_norm": 0.4859540488807519, "learning_rate": 8.669607801613909e-05, "loss": 3.0242, "step": 15171 }, { "epoch": 0.9418337575268484, "grad_norm": 0.5046666935272327, "learning_rate": 8.669362484955874e-05, "loss": 2.953, "step": 15172 }, { "epoch": 0.9418958346266062, "grad_norm": 0.4850403000615435, "learning_rate": 8.669117149153882e-05, "loss": 2.9761, "step": 15173 }, { "epoch": 0.9419579117263641, "grad_norm": 0.49847236858128513, "learning_rate": 8.668871794209218e-05, "loss": 2.918, "step": 15174 }, { "epoch": 0.942019988826122, "grad_norm": 0.49832405904073346, "learning_rate": 8.668626420123159e-05, "loss": 2.9819, "step": 15175 }, { "epoch": 0.94208206592588, "grad_norm": 0.4727416000860588, "learning_rate": 8.668381026896984e-05, "loss": 2.9402, "step": 15176 }, { "epoch": 0.9421441430256379, "grad_norm": 0.4998752468162895, "learning_rate": 8.668135614531973e-05, "loss": 2.8779, "step": 15177 }, { "epoch": 0.9422062201253958, "grad_norm": 0.5856532360992872, "learning_rate": 8.66789018302941e-05, "loss": 2.905, "step": 15178 }, { "epoch": 0.9422682972251536, "grad_norm": 0.5517196777128525, "learning_rate": 8.667644732390574e-05, "loss": 2.9207, "step": 15179 }, { "epoch": 0.9423303743249115, "grad_norm": 0.5222719224936738, "learning_rate": 8.667399262616745e-05, "loss": 2.9476, "step": 15180 }, { "epoch": 0.9423924514246694, "grad_norm": 0.5606271506856196, "learning_rate": 8.667153773709202e-05, "loss": 2.9239, "step": 15181 }, { "epoch": 0.9424545285244273, "grad_norm": 0.5755893672223488, "learning_rate": 8.666908265669229e-05, "loss": 2.9312, "step": 15182 }, { "epoch": 0.9425166056241853, "grad_norm": 0.4858101073706367, "learning_rate": 8.666662738498106e-05, "loss": 2.8857, "step": 15183 }, { "epoch": 0.9425786827239432, "grad_norm": 0.5819057098790968, "learning_rate": 8.666417192197112e-05, "loss": 2.9809, "step": 15184 }, { "epoch": 0.942640759823701, "grad_norm": 0.5612748897955244, "learning_rate": 8.66617162676753e-05, "loss": 3.0453, "step": 15185 }, { "epoch": 0.9427028369234589, "grad_norm": 0.5977027974670692, "learning_rate": 8.66592604221064e-05, "loss": 2.9843, "step": 15186 }, { "epoch": 0.9427649140232168, "grad_norm": 0.5893808531647508, "learning_rate": 8.665680438527724e-05, "loss": 2.9918, "step": 15187 }, { "epoch": 0.9428269911229747, "grad_norm": 0.6146100221647233, "learning_rate": 8.665434815720063e-05, "loss": 3.0144, "step": 15188 }, { "epoch": 0.9428890682227327, "grad_norm": 0.5531909526147869, "learning_rate": 8.665189173788942e-05, "loss": 2.9643, "step": 15189 }, { "epoch": 0.9429511453224906, "grad_norm": 0.5176903081272244, "learning_rate": 8.664943512735635e-05, "loss": 2.975, "step": 15190 }, { "epoch": 0.9430132224222484, "grad_norm": 0.5911887793268504, "learning_rate": 8.66469783256143e-05, "loss": 2.9443, "step": 15191 }, { "epoch": 0.9430752995220063, "grad_norm": 0.5225771765772317, "learning_rate": 8.664452133267606e-05, "loss": 3.0756, "step": 15192 }, { "epoch": 0.9431373766217642, "grad_norm": 0.5209607413541711, "learning_rate": 8.664206414855447e-05, "loss": 2.9181, "step": 15193 }, { "epoch": 0.9431994537215221, "grad_norm": 0.5252717745197111, "learning_rate": 8.663960677326233e-05, "loss": 2.919, "step": 15194 }, { "epoch": 0.9432615308212801, "grad_norm": 0.5513406982977418, "learning_rate": 8.663714920681246e-05, "loss": 3.0143, "step": 15195 }, { "epoch": 0.943323607921038, "grad_norm": 0.49212122274145353, "learning_rate": 8.66346914492177e-05, "loss": 2.9513, "step": 15196 }, { "epoch": 0.9433856850207958, "grad_norm": 0.5352889404635242, "learning_rate": 8.663223350049085e-05, "loss": 2.9134, "step": 15197 }, { "epoch": 0.9434477621205537, "grad_norm": 0.5302504223628972, "learning_rate": 8.662977536064476e-05, "loss": 2.9071, "step": 15198 }, { "epoch": 0.9435098392203116, "grad_norm": 0.4952811847980126, "learning_rate": 8.662731702969222e-05, "loss": 3.0047, "step": 15199 }, { "epoch": 0.9435719163200695, "grad_norm": 0.5053590634890573, "learning_rate": 8.662485850764609e-05, "loss": 2.8946, "step": 15200 }, { "epoch": 0.9436339934198275, "grad_norm": 0.5221583265445574, "learning_rate": 8.662239979451918e-05, "loss": 2.9813, "step": 15201 }, { "epoch": 0.9436960705195854, "grad_norm": 0.5317073984697155, "learning_rate": 8.661994089032433e-05, "loss": 2.9768, "step": 15202 }, { "epoch": 0.9437581476193432, "grad_norm": 0.5591393054127363, "learning_rate": 8.661748179507432e-05, "loss": 3.0055, "step": 15203 }, { "epoch": 0.9438202247191011, "grad_norm": 0.4968368041754289, "learning_rate": 8.661502250878207e-05, "loss": 2.8398, "step": 15204 }, { "epoch": 0.943882301818859, "grad_norm": 0.47711360833947947, "learning_rate": 8.661256303146033e-05, "loss": 3.0047, "step": 15205 }, { "epoch": 0.9439443789186169, "grad_norm": 0.5488026025885547, "learning_rate": 8.661010336312195e-05, "loss": 2.8962, "step": 15206 }, { "epoch": 0.9440064560183749, "grad_norm": 0.508871265880253, "learning_rate": 8.66076435037798e-05, "loss": 2.9496, "step": 15207 }, { "epoch": 0.9440685331181328, "grad_norm": 0.5156786341660391, "learning_rate": 8.660518345344668e-05, "loss": 2.9226, "step": 15208 }, { "epoch": 0.9441306102178906, "grad_norm": 0.5948872466601222, "learning_rate": 8.660272321213541e-05, "loss": 2.9978, "step": 15209 }, { "epoch": 0.9441926873176485, "grad_norm": 0.552660232801379, "learning_rate": 8.660026277985887e-05, "loss": 2.9623, "step": 15210 }, { "epoch": 0.9442547644174064, "grad_norm": 0.544683096469012, "learning_rate": 8.659780215662987e-05, "loss": 2.9742, "step": 15211 }, { "epoch": 0.9443168415171643, "grad_norm": 0.48058864989310196, "learning_rate": 8.659534134246123e-05, "loss": 2.8595, "step": 15212 }, { "epoch": 0.9443789186169222, "grad_norm": 0.5906308966044601, "learning_rate": 8.659288033736583e-05, "loss": 2.8938, "step": 15213 }, { "epoch": 0.9444409957166802, "grad_norm": 0.5169344660490413, "learning_rate": 8.65904191413565e-05, "loss": 2.9486, "step": 15214 }, { "epoch": 0.944503072816438, "grad_norm": 0.5335855475000525, "learning_rate": 8.658795775444605e-05, "loss": 2.97, "step": 15215 }, { "epoch": 0.9445651499161959, "grad_norm": 0.5070612999897057, "learning_rate": 8.658549617664736e-05, "loss": 3.0021, "step": 15216 }, { "epoch": 0.9446272270159538, "grad_norm": 0.5279624982397679, "learning_rate": 8.658303440797326e-05, "loss": 2.9705, "step": 15217 }, { "epoch": 0.9446893041157117, "grad_norm": 0.5491145733807954, "learning_rate": 8.658057244843657e-05, "loss": 2.8542, "step": 15218 }, { "epoch": 0.9447513812154696, "grad_norm": 0.4096551468016841, "learning_rate": 8.657811029805016e-05, "loss": 2.936, "step": 15219 }, { "epoch": 0.9448134583152276, "grad_norm": 0.46929219994995486, "learning_rate": 8.657564795682687e-05, "loss": 2.942, "step": 15220 }, { "epoch": 0.9448755354149854, "grad_norm": 0.4232585789385047, "learning_rate": 8.657318542477955e-05, "loss": 2.9154, "step": 15221 }, { "epoch": 0.9449376125147433, "grad_norm": 0.4542354591427287, "learning_rate": 8.657072270192103e-05, "loss": 3.0236, "step": 15222 }, { "epoch": 0.9449996896145012, "grad_norm": 0.437395423924702, "learning_rate": 8.656825978826417e-05, "loss": 2.9381, "step": 15223 }, { "epoch": 0.9450617667142591, "grad_norm": 0.5163230794393714, "learning_rate": 8.656579668382183e-05, "loss": 2.9658, "step": 15224 }, { "epoch": 0.945123843814017, "grad_norm": 0.4642016520091125, "learning_rate": 8.656333338860686e-05, "loss": 2.9185, "step": 15225 }, { "epoch": 0.945185920913775, "grad_norm": 0.5615725553840358, "learning_rate": 8.65608699026321e-05, "loss": 2.8667, "step": 15226 }, { "epoch": 0.9452479980135328, "grad_norm": 0.4734788310073175, "learning_rate": 8.65584062259104e-05, "loss": 2.8384, "step": 15227 }, { "epoch": 0.9453100751132907, "grad_norm": 0.48305195422758956, "learning_rate": 8.655594235845462e-05, "loss": 3.0065, "step": 15228 }, { "epoch": 0.9453721522130486, "grad_norm": 0.44865572932713255, "learning_rate": 8.655347830027762e-05, "loss": 3.0652, "step": 15229 }, { "epoch": 0.9454342293128065, "grad_norm": 0.507876025407992, "learning_rate": 8.655101405139226e-05, "loss": 2.9496, "step": 15230 }, { "epoch": 0.9454963064125644, "grad_norm": 0.4811283019624755, "learning_rate": 8.654854961181136e-05, "loss": 2.9327, "step": 15231 }, { "epoch": 0.9455583835123224, "grad_norm": 0.4818244509029431, "learning_rate": 8.654608498154782e-05, "loss": 2.9173, "step": 15232 }, { "epoch": 0.9456204606120802, "grad_norm": 0.5489762401615965, "learning_rate": 8.654362016061449e-05, "loss": 2.9949, "step": 15233 }, { "epoch": 0.9456825377118381, "grad_norm": 0.5036719908787384, "learning_rate": 8.65411551490242e-05, "loss": 2.9485, "step": 15234 }, { "epoch": 0.945744614811596, "grad_norm": 0.4587913756720252, "learning_rate": 8.653868994678985e-05, "loss": 2.9266, "step": 15235 }, { "epoch": 0.9458066919113539, "grad_norm": 0.4671026688684681, "learning_rate": 8.653622455392427e-05, "loss": 2.93, "step": 15236 }, { "epoch": 0.9458687690111118, "grad_norm": 0.5173378728672734, "learning_rate": 8.653375897044036e-05, "loss": 2.9456, "step": 15237 }, { "epoch": 0.9459308461108697, "grad_norm": 0.4892459799426435, "learning_rate": 8.653129319635093e-05, "loss": 2.9787, "step": 15238 }, { "epoch": 0.9459929232106276, "grad_norm": 0.5056552843471638, "learning_rate": 8.65288272316689e-05, "loss": 2.9446, "step": 15239 }, { "epoch": 0.9460550003103855, "grad_norm": 0.46032353907838985, "learning_rate": 8.65263610764071e-05, "loss": 2.8814, "step": 15240 }, { "epoch": 0.9461170774101434, "grad_norm": 0.49162294919488014, "learning_rate": 8.652389473057843e-05, "loss": 2.9214, "step": 15241 }, { "epoch": 0.9461791545099013, "grad_norm": 0.4865878840043834, "learning_rate": 8.652142819419571e-05, "loss": 2.9871, "step": 15242 }, { "epoch": 0.9462412316096592, "grad_norm": 0.5100534437731374, "learning_rate": 8.651896146727183e-05, "loss": 2.92, "step": 15243 }, { "epoch": 0.9463033087094171, "grad_norm": 0.4813467857335942, "learning_rate": 8.651649454981967e-05, "loss": 2.902, "step": 15244 }, { "epoch": 0.946365385809175, "grad_norm": 0.46976432727791173, "learning_rate": 8.651402744185211e-05, "loss": 2.9422, "step": 15245 }, { "epoch": 0.9464274629089329, "grad_norm": 0.5420122847396547, "learning_rate": 8.651156014338198e-05, "loss": 2.9744, "step": 15246 }, { "epoch": 0.9464895400086908, "grad_norm": 0.6244196342957075, "learning_rate": 8.65090926544222e-05, "loss": 2.9489, "step": 15247 }, { "epoch": 0.9465516171084487, "grad_norm": 0.4897863308778038, "learning_rate": 8.65066249749856e-05, "loss": 3.0418, "step": 15248 }, { "epoch": 0.9466136942082066, "grad_norm": 0.58968788260168, "learning_rate": 8.650415710508508e-05, "loss": 2.8195, "step": 15249 }, { "epoch": 0.9466757713079645, "grad_norm": 0.4987811192625329, "learning_rate": 8.650168904473352e-05, "loss": 2.8773, "step": 15250 }, { "epoch": 0.9467378484077223, "grad_norm": 0.527135143698956, "learning_rate": 8.649922079394377e-05, "loss": 2.9533, "step": 15251 }, { "epoch": 0.9467999255074803, "grad_norm": 0.5776957387430598, "learning_rate": 8.649675235272873e-05, "loss": 2.9002, "step": 15252 }, { "epoch": 0.9468620026072382, "grad_norm": 0.49971511294546817, "learning_rate": 8.649428372110129e-05, "loss": 2.8966, "step": 15253 }, { "epoch": 0.9469240797069961, "grad_norm": 0.5096341836947219, "learning_rate": 8.64918148990743e-05, "loss": 3.0091, "step": 15254 }, { "epoch": 0.946986156806754, "grad_norm": 0.493281903681837, "learning_rate": 8.648934588666064e-05, "loss": 3.0113, "step": 15255 }, { "epoch": 0.9470482339065119, "grad_norm": 0.5275338289523117, "learning_rate": 8.648687668387323e-05, "loss": 2.8572, "step": 15256 }, { "epoch": 0.9471103110062697, "grad_norm": 0.5238280191784097, "learning_rate": 8.648440729072492e-05, "loss": 2.8903, "step": 15257 }, { "epoch": 0.9471723881060277, "grad_norm": 0.5118121578780463, "learning_rate": 8.648193770722858e-05, "loss": 2.9191, "step": 15258 }, { "epoch": 0.9472344652057856, "grad_norm": 0.4701031386808447, "learning_rate": 8.647946793339715e-05, "loss": 2.9177, "step": 15259 }, { "epoch": 0.9472965423055435, "grad_norm": 0.462830422908894, "learning_rate": 8.647699796924345e-05, "loss": 2.9479, "step": 15260 }, { "epoch": 0.9473586194053014, "grad_norm": 0.49997511444151527, "learning_rate": 8.64745278147804e-05, "loss": 2.9835, "step": 15261 }, { "epoch": 0.9474206965050593, "grad_norm": 0.4788642547515127, "learning_rate": 8.64720574700209e-05, "loss": 2.9449, "step": 15262 }, { "epoch": 0.9474827736048171, "grad_norm": 0.5050034517226898, "learning_rate": 8.646958693497781e-05, "loss": 2.9134, "step": 15263 }, { "epoch": 0.9475448507045751, "grad_norm": 0.5779959560404616, "learning_rate": 8.646711620966404e-05, "loss": 2.923, "step": 15264 }, { "epoch": 0.947606927804333, "grad_norm": 0.4813447580310252, "learning_rate": 8.646464529409246e-05, "loss": 2.9244, "step": 15265 }, { "epoch": 0.9476690049040909, "grad_norm": 0.49335220827045506, "learning_rate": 8.646217418827599e-05, "loss": 2.8998, "step": 15266 }, { "epoch": 0.9477310820038488, "grad_norm": 0.5487695846017231, "learning_rate": 8.645970289222749e-05, "loss": 2.952, "step": 15267 }, { "epoch": 0.9477931591036067, "grad_norm": 0.5115867910519425, "learning_rate": 8.645723140595987e-05, "loss": 3.0059, "step": 15268 }, { "epoch": 0.9478552362033645, "grad_norm": 0.525105764316862, "learning_rate": 8.645475972948602e-05, "loss": 2.971, "step": 15269 }, { "epoch": 0.9479173133031225, "grad_norm": 0.4890937453943825, "learning_rate": 8.645228786281886e-05, "loss": 2.9881, "step": 15270 }, { "epoch": 0.9479793904028804, "grad_norm": 0.5507093612614357, "learning_rate": 8.644981580597125e-05, "loss": 2.969, "step": 15271 }, { "epoch": 0.9480414675026383, "grad_norm": 0.4519015929864071, "learning_rate": 8.64473435589561e-05, "loss": 2.9167, "step": 15272 }, { "epoch": 0.9481035446023962, "grad_norm": 0.5113012051763419, "learning_rate": 8.644487112178632e-05, "loss": 2.984, "step": 15273 }, { "epoch": 0.9481656217021541, "grad_norm": 0.4821616949962832, "learning_rate": 8.64423984944748e-05, "loss": 2.9931, "step": 15274 }, { "epoch": 0.9482276988019119, "grad_norm": 0.5070695283661985, "learning_rate": 8.643992567703443e-05, "loss": 2.9282, "step": 15275 }, { "epoch": 0.9482897759016699, "grad_norm": 0.5053323482090315, "learning_rate": 8.643745266947813e-05, "loss": 2.9593, "step": 15276 }, { "epoch": 0.9483518530014278, "grad_norm": 0.46990525619788365, "learning_rate": 8.643497947181878e-05, "loss": 2.8241, "step": 15277 }, { "epoch": 0.9484139301011857, "grad_norm": 0.48458799170794026, "learning_rate": 8.643250608406932e-05, "loss": 2.9705, "step": 15278 }, { "epoch": 0.9484760072009436, "grad_norm": 0.48267584845214, "learning_rate": 8.643003250624262e-05, "loss": 3.0323, "step": 15279 }, { "epoch": 0.9485380843007015, "grad_norm": 0.4785003348262656, "learning_rate": 8.64275587383516e-05, "loss": 2.933, "step": 15280 }, { "epoch": 0.9486001614004593, "grad_norm": 0.4777769498654642, "learning_rate": 8.642508478040916e-05, "loss": 2.903, "step": 15281 }, { "epoch": 0.9486622385002172, "grad_norm": 0.5468046688086317, "learning_rate": 8.64226106324282e-05, "loss": 2.8875, "step": 15282 }, { "epoch": 0.9487243155999752, "grad_norm": 0.5016195175704015, "learning_rate": 8.642013629442165e-05, "loss": 2.9031, "step": 15283 }, { "epoch": 0.9487863926997331, "grad_norm": 0.5084393080741935, "learning_rate": 8.64176617664024e-05, "loss": 2.8949, "step": 15284 }, { "epoch": 0.948848469799491, "grad_norm": 0.46779662773609715, "learning_rate": 8.641518704838338e-05, "loss": 2.9558, "step": 15285 }, { "epoch": 0.9489105468992489, "grad_norm": 0.5174188620773206, "learning_rate": 8.641271214037747e-05, "loss": 2.9569, "step": 15286 }, { "epoch": 0.9489726239990067, "grad_norm": 0.47018334272322276, "learning_rate": 8.641023704239764e-05, "loss": 2.9401, "step": 15287 }, { "epoch": 0.9490347010987646, "grad_norm": 0.43810814075438287, "learning_rate": 8.640776175445673e-05, "loss": 2.8898, "step": 15288 }, { "epoch": 0.9490967781985226, "grad_norm": 0.47576510835773816, "learning_rate": 8.64052862765677e-05, "loss": 2.8828, "step": 15289 }, { "epoch": 0.9491588552982805, "grad_norm": 0.5351731235043754, "learning_rate": 8.640281060874345e-05, "loss": 3.0123, "step": 15290 }, { "epoch": 0.9492209323980384, "grad_norm": 0.47889352006779884, "learning_rate": 8.640033475099689e-05, "loss": 2.9819, "step": 15291 }, { "epoch": 0.9492830094977963, "grad_norm": 0.592490843730307, "learning_rate": 8.639785870334096e-05, "loss": 2.9785, "step": 15292 }, { "epoch": 0.9493450865975541, "grad_norm": 0.5709625354324458, "learning_rate": 8.639538246578856e-05, "loss": 2.996, "step": 15293 }, { "epoch": 0.949407163697312, "grad_norm": 0.49053025150823226, "learning_rate": 8.639290603835262e-05, "loss": 3.0049, "step": 15294 }, { "epoch": 0.94946924079707, "grad_norm": 0.5757400446191342, "learning_rate": 8.639042942104605e-05, "loss": 2.9497, "step": 15295 }, { "epoch": 0.9495313178968279, "grad_norm": 0.54228801729975, "learning_rate": 8.638795261388178e-05, "loss": 2.9349, "step": 15296 }, { "epoch": 0.9495933949965858, "grad_norm": 0.5077394286181689, "learning_rate": 8.638547561687273e-05, "loss": 2.819, "step": 15297 }, { "epoch": 0.9496554720963437, "grad_norm": 0.636282320756762, "learning_rate": 8.638299843003181e-05, "loss": 2.9947, "step": 15298 }, { "epoch": 0.9497175491961015, "grad_norm": 0.5763310875182395, "learning_rate": 8.638052105337198e-05, "loss": 2.9667, "step": 15299 }, { "epoch": 0.9497796262958594, "grad_norm": 0.5639068970078002, "learning_rate": 8.63780434869061e-05, "loss": 2.8987, "step": 15300 }, { "epoch": 0.9498417033956174, "grad_norm": 0.49475454559281246, "learning_rate": 8.637556573064717e-05, "loss": 2.9336, "step": 15301 }, { "epoch": 0.9499037804953753, "grad_norm": 0.5476436662520664, "learning_rate": 8.637308778460807e-05, "loss": 2.8975, "step": 15302 }, { "epoch": 0.9499658575951332, "grad_norm": 0.48181122962743606, "learning_rate": 8.637060964880174e-05, "loss": 2.8488, "step": 15303 }, { "epoch": 0.9500279346948911, "grad_norm": 0.506951262008717, "learning_rate": 8.636813132324111e-05, "loss": 2.8658, "step": 15304 }, { "epoch": 0.9500900117946489, "grad_norm": 0.5590828041837588, "learning_rate": 8.636565280793911e-05, "loss": 2.9799, "step": 15305 }, { "epoch": 0.9501520888944068, "grad_norm": 0.4707905699994018, "learning_rate": 8.636317410290868e-05, "loss": 3.0094, "step": 15306 }, { "epoch": 0.9502141659941648, "grad_norm": 0.5686527556821264, "learning_rate": 8.636069520816273e-05, "loss": 2.8817, "step": 15307 }, { "epoch": 0.9502762430939227, "grad_norm": 0.496056908268249, "learning_rate": 8.635821612371422e-05, "loss": 2.9159, "step": 15308 }, { "epoch": 0.9503383201936806, "grad_norm": 0.4978879936959197, "learning_rate": 8.635573684957605e-05, "loss": 3.0154, "step": 15309 }, { "epoch": 0.9504003972934385, "grad_norm": 0.5273779116267852, "learning_rate": 8.635325738576119e-05, "loss": 2.9366, "step": 15310 }, { "epoch": 0.9504624743931963, "grad_norm": 0.4858620183283632, "learning_rate": 8.635077773228256e-05, "loss": 2.964, "step": 15311 }, { "epoch": 0.9505245514929542, "grad_norm": 0.5290234130282503, "learning_rate": 8.63482978891531e-05, "loss": 2.9504, "step": 15312 }, { "epoch": 0.9505866285927121, "grad_norm": 0.5555930737569031, "learning_rate": 8.634581785638574e-05, "loss": 2.8924, "step": 15313 }, { "epoch": 0.9506487056924701, "grad_norm": 0.5735713137377298, "learning_rate": 8.634333763399343e-05, "loss": 2.9534, "step": 15314 }, { "epoch": 0.950710782792228, "grad_norm": 0.5311124848049192, "learning_rate": 8.63408572219891e-05, "loss": 2.9695, "step": 15315 }, { "epoch": 0.9507728598919859, "grad_norm": 0.5220640297000394, "learning_rate": 8.63383766203857e-05, "loss": 2.8889, "step": 15316 }, { "epoch": 0.9508349369917437, "grad_norm": 0.6237789385471696, "learning_rate": 8.633589582919616e-05, "loss": 3.0306, "step": 15317 }, { "epoch": 0.9508970140915016, "grad_norm": 0.48638981500658834, "learning_rate": 8.633341484843345e-05, "loss": 2.9999, "step": 15318 }, { "epoch": 0.9509590911912595, "grad_norm": 0.5369906484990159, "learning_rate": 8.633093367811048e-05, "loss": 2.8868, "step": 15319 }, { "epoch": 0.9510211682910175, "grad_norm": 0.5784764896732542, "learning_rate": 8.63284523182402e-05, "loss": 2.8482, "step": 15320 }, { "epoch": 0.9510832453907754, "grad_norm": 0.5524061326945399, "learning_rate": 8.63259707688356e-05, "loss": 2.9971, "step": 15321 }, { "epoch": 0.9511453224905333, "grad_norm": 0.63512504378981, "learning_rate": 8.632348902990956e-05, "loss": 2.9545, "step": 15322 }, { "epoch": 0.9512073995902911, "grad_norm": 0.5446122906882764, "learning_rate": 8.632100710147506e-05, "loss": 2.8881, "step": 15323 }, { "epoch": 0.951269476690049, "grad_norm": 0.5769287310836728, "learning_rate": 8.631852498354506e-05, "loss": 3.0211, "step": 15324 }, { "epoch": 0.9513315537898069, "grad_norm": 0.6458058325747201, "learning_rate": 8.63160426761325e-05, "loss": 2.9587, "step": 15325 }, { "epoch": 0.9513936308895649, "grad_norm": 0.4994743832891842, "learning_rate": 8.631356017925032e-05, "loss": 2.9991, "step": 15326 }, { "epoch": 0.9514557079893228, "grad_norm": 0.5718858259249879, "learning_rate": 8.63110774929115e-05, "loss": 2.9721, "step": 15327 }, { "epoch": 0.9515177850890807, "grad_norm": 0.5779892272291794, "learning_rate": 8.630859461712895e-05, "loss": 2.9877, "step": 15328 }, { "epoch": 0.9515798621888385, "grad_norm": 0.5402424379355263, "learning_rate": 8.630611155191566e-05, "loss": 2.8154, "step": 15329 }, { "epoch": 0.9516419392885964, "grad_norm": 0.5143888568054624, "learning_rate": 8.630362829728456e-05, "loss": 2.92, "step": 15330 }, { "epoch": 0.9517040163883543, "grad_norm": 0.4963760323793525, "learning_rate": 8.630114485324864e-05, "loss": 2.9217, "step": 15331 }, { "epoch": 0.9517660934881123, "grad_norm": 0.5158606770909335, "learning_rate": 8.629866121982082e-05, "loss": 2.9179, "step": 15332 }, { "epoch": 0.9518281705878702, "grad_norm": 0.49415978858185944, "learning_rate": 8.629617739701407e-05, "loss": 2.9573, "step": 15333 }, { "epoch": 0.9518902476876281, "grad_norm": 0.5687745141415987, "learning_rate": 8.629369338484137e-05, "loss": 2.9562, "step": 15334 }, { "epoch": 0.9519523247873859, "grad_norm": 0.5301402224977357, "learning_rate": 8.629120918331564e-05, "loss": 2.9816, "step": 15335 }, { "epoch": 0.9520144018871438, "grad_norm": 0.5716872172744367, "learning_rate": 8.628872479244986e-05, "loss": 2.9983, "step": 15336 }, { "epoch": 0.9520764789869017, "grad_norm": 0.4857254583762732, "learning_rate": 8.628624021225702e-05, "loss": 3.0004, "step": 15337 }, { "epoch": 0.9521385560866596, "grad_norm": 0.5815954176715039, "learning_rate": 8.628375544275004e-05, "loss": 2.9732, "step": 15338 }, { "epoch": 0.9522006331864176, "grad_norm": 0.533861389731568, "learning_rate": 8.62812704839419e-05, "loss": 3.011, "step": 15339 }, { "epoch": 0.9522627102861755, "grad_norm": 0.5094348228757715, "learning_rate": 8.627878533584555e-05, "loss": 2.9557, "step": 15340 }, { "epoch": 0.9523247873859333, "grad_norm": 0.5338057302195998, "learning_rate": 8.627629999847399e-05, "loss": 3.0148, "step": 15341 }, { "epoch": 0.9523868644856912, "grad_norm": 0.5315327733253684, "learning_rate": 8.627381447184017e-05, "loss": 2.9497, "step": 15342 }, { "epoch": 0.9524489415854491, "grad_norm": 0.5315988180860666, "learning_rate": 8.627132875595703e-05, "loss": 2.9374, "step": 15343 }, { "epoch": 0.952511018685207, "grad_norm": 0.5788680791175483, "learning_rate": 8.626884285083758e-05, "loss": 3.0011, "step": 15344 }, { "epoch": 0.952573095784965, "grad_norm": 0.5331801804297355, "learning_rate": 8.626635675649477e-05, "loss": 2.9742, "step": 15345 }, { "epoch": 0.9526351728847229, "grad_norm": 0.570209023481385, "learning_rate": 8.626387047294158e-05, "loss": 3.0012, "step": 15346 }, { "epoch": 0.9526972499844807, "grad_norm": 0.48131534764401385, "learning_rate": 8.626138400019096e-05, "loss": 2.9448, "step": 15347 }, { "epoch": 0.9527593270842386, "grad_norm": 0.5594095528442605, "learning_rate": 8.625889733825589e-05, "loss": 3.009, "step": 15348 }, { "epoch": 0.9528214041839965, "grad_norm": 0.5918926501436966, "learning_rate": 8.625641048714938e-05, "loss": 2.9854, "step": 15349 }, { "epoch": 0.9528834812837544, "grad_norm": 0.4574397128411663, "learning_rate": 8.625392344688434e-05, "loss": 2.8696, "step": 15350 }, { "epoch": 0.9529455583835124, "grad_norm": 0.5853804928971742, "learning_rate": 8.62514362174738e-05, "loss": 2.885, "step": 15351 }, { "epoch": 0.9530076354832703, "grad_norm": 0.49567872055738366, "learning_rate": 8.624894879893073e-05, "loss": 3.0249, "step": 15352 }, { "epoch": 0.9530697125830281, "grad_norm": 0.4795112683759175, "learning_rate": 8.624646119126806e-05, "loss": 2.9325, "step": 15353 }, { "epoch": 0.953131789682786, "grad_norm": 0.5795136090656283, "learning_rate": 8.624397339449883e-05, "loss": 3.0712, "step": 15354 }, { "epoch": 0.9531938667825439, "grad_norm": 0.44150606225324124, "learning_rate": 8.624148540863599e-05, "loss": 2.9121, "step": 15355 }, { "epoch": 0.9532559438823018, "grad_norm": 0.4892769389116399, "learning_rate": 8.623899723369252e-05, "loss": 2.8836, "step": 15356 }, { "epoch": 0.9533180209820598, "grad_norm": 0.5078655508647214, "learning_rate": 8.62365088696814e-05, "loss": 2.9051, "step": 15357 }, { "epoch": 0.9533800980818177, "grad_norm": 0.4879748794703264, "learning_rate": 8.623402031661563e-05, "loss": 2.9007, "step": 15358 }, { "epoch": 0.9534421751815755, "grad_norm": 0.5235459798879841, "learning_rate": 8.623153157450816e-05, "loss": 2.9342, "step": 15359 }, { "epoch": 0.9535042522813334, "grad_norm": 0.4555653922671761, "learning_rate": 8.6229042643372e-05, "loss": 2.9039, "step": 15360 }, { "epoch": 0.9535663293810913, "grad_norm": 0.5316028545116427, "learning_rate": 8.622655352322014e-05, "loss": 2.9736, "step": 15361 }, { "epoch": 0.9536284064808492, "grad_norm": 0.5371003028797041, "learning_rate": 8.622406421406555e-05, "loss": 2.9502, "step": 15362 }, { "epoch": 0.9536904835806072, "grad_norm": 0.4923751412822659, "learning_rate": 8.622157471592122e-05, "loss": 2.9438, "step": 15363 }, { "epoch": 0.953752560680365, "grad_norm": 0.4919412087518903, "learning_rate": 8.621908502880014e-05, "loss": 2.9416, "step": 15364 }, { "epoch": 0.9538146377801229, "grad_norm": 0.47326749890596354, "learning_rate": 8.62165951527153e-05, "loss": 2.8996, "step": 15365 }, { "epoch": 0.9538767148798808, "grad_norm": 0.4540557510073936, "learning_rate": 8.62141050876797e-05, "loss": 2.8732, "step": 15366 }, { "epoch": 0.9539387919796387, "grad_norm": 0.4782829030160089, "learning_rate": 8.621161483370632e-05, "loss": 2.9869, "step": 15367 }, { "epoch": 0.9540008690793966, "grad_norm": 0.49662172952146877, "learning_rate": 8.620912439080815e-05, "loss": 2.9437, "step": 15368 }, { "epoch": 0.9540629461791545, "grad_norm": 0.44542831454140297, "learning_rate": 8.62066337589982e-05, "loss": 2.9074, "step": 15369 }, { "epoch": 0.9541250232789124, "grad_norm": 0.47740620830376146, "learning_rate": 8.620414293828944e-05, "loss": 2.8705, "step": 15370 }, { "epoch": 0.9541871003786703, "grad_norm": 0.537974905199863, "learning_rate": 8.620165192869488e-05, "loss": 2.9963, "step": 15371 }, { "epoch": 0.9542491774784282, "grad_norm": 0.5262148323450148, "learning_rate": 8.619916073022751e-05, "loss": 2.9733, "step": 15372 }, { "epoch": 0.9543112545781861, "grad_norm": 0.5303002731810975, "learning_rate": 8.619666934290033e-05, "loss": 2.8756, "step": 15373 }, { "epoch": 0.954373331677944, "grad_norm": 0.4730338172779928, "learning_rate": 8.619417776672634e-05, "loss": 2.9273, "step": 15374 }, { "epoch": 0.9544354087777019, "grad_norm": 0.5414541237102267, "learning_rate": 8.619168600171854e-05, "loss": 2.9412, "step": 15375 }, { "epoch": 0.9544974858774598, "grad_norm": 0.5250416273689159, "learning_rate": 8.618919404788994e-05, "loss": 2.9762, "step": 15376 }, { "epoch": 0.9545595629772177, "grad_norm": 0.4805276415949209, "learning_rate": 8.618670190525352e-05, "loss": 2.9333, "step": 15377 }, { "epoch": 0.9546216400769756, "grad_norm": 0.4574165350441869, "learning_rate": 8.618420957382229e-05, "loss": 2.9427, "step": 15378 }, { "epoch": 0.9546837171767335, "grad_norm": 0.49194299589006424, "learning_rate": 8.618171705360925e-05, "loss": 2.9197, "step": 15379 }, { "epoch": 0.9547457942764914, "grad_norm": 0.5316154961739957, "learning_rate": 8.617922434462742e-05, "loss": 2.9361, "step": 15380 }, { "epoch": 0.9548078713762493, "grad_norm": 0.48548296371981164, "learning_rate": 8.617673144688977e-05, "loss": 2.831, "step": 15381 }, { "epoch": 0.9548699484760071, "grad_norm": 0.5310818462154474, "learning_rate": 8.617423836040936e-05, "loss": 3.0152, "step": 15382 }, { "epoch": 0.9549320255757651, "grad_norm": 0.5216039456218992, "learning_rate": 8.617174508519916e-05, "loss": 3.0285, "step": 15383 }, { "epoch": 0.954994102675523, "grad_norm": 0.5149320222841299, "learning_rate": 8.616925162127218e-05, "loss": 3.0307, "step": 15384 }, { "epoch": 0.9550561797752809, "grad_norm": 0.5127428886500731, "learning_rate": 8.616675796864143e-05, "loss": 3.0093, "step": 15385 }, { "epoch": 0.9551182568750388, "grad_norm": 0.5461964621157276, "learning_rate": 8.616426412731992e-05, "loss": 3.0048, "step": 15386 }, { "epoch": 0.9551803339747967, "grad_norm": 0.4733904975749671, "learning_rate": 8.616177009732067e-05, "loss": 2.8776, "step": 15387 }, { "epoch": 0.9552424110745545, "grad_norm": 0.5628863703125494, "learning_rate": 8.61592758786567e-05, "loss": 2.9796, "step": 15388 }, { "epoch": 0.9553044881743125, "grad_norm": 0.5492149700181227, "learning_rate": 8.6156781471341e-05, "loss": 2.9882, "step": 15389 }, { "epoch": 0.9553665652740704, "grad_norm": 0.5820022934232665, "learning_rate": 8.61542868753866e-05, "loss": 2.9279, "step": 15390 }, { "epoch": 0.9554286423738283, "grad_norm": 0.5128924181660598, "learning_rate": 8.615179209080649e-05, "loss": 2.9976, "step": 15391 }, { "epoch": 0.9554907194735862, "grad_norm": 0.5657408916972425, "learning_rate": 8.614929711761372e-05, "loss": 2.9143, "step": 15392 }, { "epoch": 0.9555527965733441, "grad_norm": 0.5468645639786194, "learning_rate": 8.614680195582127e-05, "loss": 3.0485, "step": 15393 }, { "epoch": 0.9556148736731019, "grad_norm": 0.5004112221545793, "learning_rate": 8.61443066054422e-05, "loss": 2.7855, "step": 15394 }, { "epoch": 0.9556769507728599, "grad_norm": 0.5388850044272163, "learning_rate": 8.614181106648951e-05, "loss": 2.9041, "step": 15395 }, { "epoch": 0.9557390278726178, "grad_norm": 0.5196169158750844, "learning_rate": 8.613931533897621e-05, "loss": 2.9833, "step": 15396 }, { "epoch": 0.9558011049723757, "grad_norm": 0.4980320348585634, "learning_rate": 8.613681942291533e-05, "loss": 2.9873, "step": 15397 }, { "epoch": 0.9558631820721336, "grad_norm": 0.5511718339061654, "learning_rate": 8.61343233183199e-05, "loss": 2.9317, "step": 15398 }, { "epoch": 0.9559252591718915, "grad_norm": 0.4864862330548096, "learning_rate": 8.613182702520292e-05, "loss": 2.9716, "step": 15399 }, { "epoch": 0.9559873362716493, "grad_norm": 0.5413611510237725, "learning_rate": 8.612933054357744e-05, "loss": 3.0046, "step": 15400 }, { "epoch": 0.9560494133714073, "grad_norm": 0.6149010133554716, "learning_rate": 8.612683387345646e-05, "loss": 2.9637, "step": 15401 }, { "epoch": 0.9561114904711652, "grad_norm": 0.609467205994485, "learning_rate": 8.612433701485301e-05, "loss": 3.0143, "step": 15402 }, { "epoch": 0.9561735675709231, "grad_norm": 0.532070984105862, "learning_rate": 8.612183996778014e-05, "loss": 2.9519, "step": 15403 }, { "epoch": 0.956235644670681, "grad_norm": 0.5301392106101404, "learning_rate": 8.611934273225086e-05, "loss": 2.9525, "step": 15404 }, { "epoch": 0.9562977217704389, "grad_norm": 0.5654324282775759, "learning_rate": 8.611684530827821e-05, "loss": 2.9324, "step": 15405 }, { "epoch": 0.9563597988701967, "grad_norm": 0.5243633963263804, "learning_rate": 8.611434769587518e-05, "loss": 2.9631, "step": 15406 }, { "epoch": 0.9564218759699546, "grad_norm": 0.49451289270929777, "learning_rate": 8.611184989505486e-05, "loss": 2.9346, "step": 15407 }, { "epoch": 0.9564839530697126, "grad_norm": 0.4781817770263853, "learning_rate": 8.610935190583025e-05, "loss": 2.806, "step": 15408 }, { "epoch": 0.9565460301694705, "grad_norm": 0.4866372623471341, "learning_rate": 8.610685372821436e-05, "loss": 2.8414, "step": 15409 }, { "epoch": 0.9566081072692284, "grad_norm": 0.4515537280910356, "learning_rate": 8.610435536222028e-05, "loss": 2.8897, "step": 15410 }, { "epoch": 0.9566701843689863, "grad_norm": 0.557552166837365, "learning_rate": 8.6101856807861e-05, "loss": 3.0177, "step": 15411 }, { "epoch": 0.9567322614687441, "grad_norm": 0.5772993532694887, "learning_rate": 8.609935806514957e-05, "loss": 3.0852, "step": 15412 }, { "epoch": 0.956794338568502, "grad_norm": 0.5014272228121137, "learning_rate": 8.609685913409901e-05, "loss": 2.9389, "step": 15413 }, { "epoch": 0.95685641566826, "grad_norm": 0.5034536650588369, "learning_rate": 8.60943600147224e-05, "loss": 2.9674, "step": 15414 }, { "epoch": 0.9569184927680179, "grad_norm": 0.542624604642206, "learning_rate": 8.609186070703275e-05, "loss": 2.854, "step": 15415 }, { "epoch": 0.9569805698677758, "grad_norm": 0.5571958483264728, "learning_rate": 8.608936121104307e-05, "loss": 2.9574, "step": 15416 }, { "epoch": 0.9570426469675337, "grad_norm": 0.5561760017380168, "learning_rate": 8.608686152676646e-05, "loss": 2.9996, "step": 15417 }, { "epoch": 0.9571047240672915, "grad_norm": 0.5573731003442539, "learning_rate": 8.608436165421592e-05, "loss": 2.9453, "step": 15418 }, { "epoch": 0.9571668011670494, "grad_norm": 0.5875442599811118, "learning_rate": 8.608186159340452e-05, "loss": 2.9425, "step": 15419 }, { "epoch": 0.9572288782668074, "grad_norm": 0.5599123190885001, "learning_rate": 8.607936134434527e-05, "loss": 2.9839, "step": 15420 }, { "epoch": 0.9572909553665653, "grad_norm": 0.5189022472747554, "learning_rate": 8.607686090705125e-05, "loss": 2.9213, "step": 15421 }, { "epoch": 0.9573530324663232, "grad_norm": 0.5234783213625508, "learning_rate": 8.607436028153548e-05, "loss": 2.9597, "step": 15422 }, { "epoch": 0.9574151095660811, "grad_norm": 0.4982643699717516, "learning_rate": 8.607185946781102e-05, "loss": 3.0086, "step": 15423 }, { "epoch": 0.9574771866658389, "grad_norm": 0.4903996557522712, "learning_rate": 8.60693584658909e-05, "loss": 2.9682, "step": 15424 }, { "epoch": 0.9575392637655968, "grad_norm": 0.43372306957247336, "learning_rate": 8.60668572757882e-05, "loss": 2.973, "step": 15425 }, { "epoch": 0.9576013408653548, "grad_norm": 0.5414363175667334, "learning_rate": 8.606435589751593e-05, "loss": 2.9011, "step": 15426 }, { "epoch": 0.9576634179651127, "grad_norm": 0.47987535321062924, "learning_rate": 8.606185433108717e-05, "loss": 2.8713, "step": 15427 }, { "epoch": 0.9577254950648706, "grad_norm": 0.493899079041164, "learning_rate": 8.605935257651496e-05, "loss": 3.007, "step": 15428 }, { "epoch": 0.9577875721646285, "grad_norm": 0.4708767804070888, "learning_rate": 8.605685063381236e-05, "loss": 2.9122, "step": 15429 }, { "epoch": 0.9578496492643863, "grad_norm": 0.5337238773055215, "learning_rate": 8.605434850299241e-05, "loss": 2.9591, "step": 15430 }, { "epoch": 0.9579117263641442, "grad_norm": 0.48217889326897917, "learning_rate": 8.605184618406819e-05, "loss": 2.9737, "step": 15431 }, { "epoch": 0.9579738034639022, "grad_norm": 0.5104256810800085, "learning_rate": 8.604934367705271e-05, "loss": 2.9145, "step": 15432 }, { "epoch": 0.9580358805636601, "grad_norm": 0.5229284173726767, "learning_rate": 8.604684098195905e-05, "loss": 2.926, "step": 15433 }, { "epoch": 0.958097957663418, "grad_norm": 0.5205533005983242, "learning_rate": 8.604433809880029e-05, "loss": 3.0379, "step": 15434 }, { "epoch": 0.9581600347631759, "grad_norm": 0.47185352036554556, "learning_rate": 8.604183502758947e-05, "loss": 2.9, "step": 15435 }, { "epoch": 0.9582221118629337, "grad_norm": 0.5234688422085872, "learning_rate": 8.603933176833963e-05, "loss": 2.9126, "step": 15436 }, { "epoch": 0.9582841889626916, "grad_norm": 0.475914029289708, "learning_rate": 8.603682832106385e-05, "loss": 2.7841, "step": 15437 }, { "epoch": 0.9583462660624495, "grad_norm": 0.5995849952338936, "learning_rate": 8.60343246857752e-05, "loss": 2.9616, "step": 15438 }, { "epoch": 0.9584083431622075, "grad_norm": 0.4790808628492125, "learning_rate": 8.603182086248674e-05, "loss": 2.977, "step": 15439 }, { "epoch": 0.9584704202619654, "grad_norm": 0.5693511960845924, "learning_rate": 8.60293168512115e-05, "loss": 2.9084, "step": 15440 }, { "epoch": 0.9585324973617233, "grad_norm": 0.5422804881890548, "learning_rate": 8.602681265196257e-05, "loss": 2.9012, "step": 15441 }, { "epoch": 0.9585945744614811, "grad_norm": 0.5723500889232623, "learning_rate": 8.602430826475303e-05, "loss": 2.9464, "step": 15442 }, { "epoch": 0.958656651561239, "grad_norm": 0.5276032551051176, "learning_rate": 8.602180368959592e-05, "loss": 2.8772, "step": 15443 }, { "epoch": 0.9587187286609969, "grad_norm": 0.5228642128349892, "learning_rate": 8.60192989265043e-05, "loss": 2.9584, "step": 15444 }, { "epoch": 0.9587808057607549, "grad_norm": 0.49849828570672805, "learning_rate": 8.601679397549127e-05, "loss": 2.8754, "step": 15445 }, { "epoch": 0.9588428828605128, "grad_norm": 0.5163869141437232, "learning_rate": 8.601428883656988e-05, "loss": 2.9869, "step": 15446 }, { "epoch": 0.9589049599602707, "grad_norm": 0.5683758133546216, "learning_rate": 8.60117835097532e-05, "loss": 3.0466, "step": 15447 }, { "epoch": 0.9589670370600285, "grad_norm": 0.6047816161414434, "learning_rate": 8.60092779950543e-05, "loss": 2.9372, "step": 15448 }, { "epoch": 0.9590291141597864, "grad_norm": 0.5634418126436829, "learning_rate": 8.600677229248626e-05, "loss": 2.9222, "step": 15449 }, { "epoch": 0.9590911912595443, "grad_norm": 0.5345674881949363, "learning_rate": 8.600426640206214e-05, "loss": 2.9886, "step": 15450 }, { "epoch": 0.9591532683593023, "grad_norm": 0.574559402626774, "learning_rate": 8.600176032379503e-05, "loss": 2.9055, "step": 15451 }, { "epoch": 0.9592153454590602, "grad_norm": 0.49564147217113147, "learning_rate": 8.5999254057698e-05, "loss": 2.9968, "step": 15452 }, { "epoch": 0.9592774225588181, "grad_norm": 0.5347402027404334, "learning_rate": 8.599674760378412e-05, "loss": 2.9431, "step": 15453 }, { "epoch": 0.9593394996585759, "grad_norm": 0.48526796968117614, "learning_rate": 8.599424096206648e-05, "loss": 2.8906, "step": 15454 }, { "epoch": 0.9594015767583338, "grad_norm": 0.515378951119306, "learning_rate": 8.599173413255813e-05, "loss": 2.8483, "step": 15455 }, { "epoch": 0.9594636538580917, "grad_norm": 0.5383813396795348, "learning_rate": 8.598922711527217e-05, "loss": 2.9373, "step": 15456 }, { "epoch": 0.9595257309578497, "grad_norm": 0.5669178460951068, "learning_rate": 8.598671991022167e-05, "loss": 2.9143, "step": 15457 }, { "epoch": 0.9595878080576076, "grad_norm": 0.5182546261103874, "learning_rate": 8.598421251741973e-05, "loss": 3.0014, "step": 15458 }, { "epoch": 0.9596498851573655, "grad_norm": 0.5276356207260576, "learning_rate": 8.598170493687941e-05, "loss": 2.9233, "step": 15459 }, { "epoch": 0.9597119622571233, "grad_norm": 0.4846678433100092, "learning_rate": 8.597919716861379e-05, "loss": 2.9378, "step": 15460 }, { "epoch": 0.9597740393568812, "grad_norm": 0.5236935629743524, "learning_rate": 8.597668921263597e-05, "loss": 3.0204, "step": 15461 }, { "epoch": 0.9598361164566391, "grad_norm": 0.5282111204907891, "learning_rate": 8.597418106895903e-05, "loss": 2.9433, "step": 15462 }, { "epoch": 0.959898193556397, "grad_norm": 0.5231713144817574, "learning_rate": 8.597167273759607e-05, "loss": 2.979, "step": 15463 }, { "epoch": 0.959960270656155, "grad_norm": 0.5184821528910287, "learning_rate": 8.596916421856014e-05, "loss": 2.8635, "step": 15464 }, { "epoch": 0.9600223477559129, "grad_norm": 0.5106320380912969, "learning_rate": 8.596665551186436e-05, "loss": 2.9935, "step": 15465 }, { "epoch": 0.9600844248556707, "grad_norm": 0.5235750103143817, "learning_rate": 8.59641466175218e-05, "loss": 2.8874, "step": 15466 }, { "epoch": 0.9601465019554286, "grad_norm": 0.5374076431237724, "learning_rate": 8.596163753554555e-05, "loss": 2.9319, "step": 15467 }, { "epoch": 0.9602085790551865, "grad_norm": 0.4348534679526325, "learning_rate": 8.595912826594873e-05, "loss": 3.0316, "step": 15468 }, { "epoch": 0.9602706561549444, "grad_norm": 0.5512057623121172, "learning_rate": 8.595661880874439e-05, "loss": 2.9775, "step": 15469 }, { "epoch": 0.9603327332547024, "grad_norm": 0.45465082564694825, "learning_rate": 8.595410916394565e-05, "loss": 2.9022, "step": 15470 }, { "epoch": 0.9603948103544603, "grad_norm": 0.5873339225513146, "learning_rate": 8.59515993315656e-05, "loss": 2.9355, "step": 15471 }, { "epoch": 0.9604568874542181, "grad_norm": 0.43390518894416114, "learning_rate": 8.594908931161732e-05, "loss": 2.8604, "step": 15472 }, { "epoch": 0.960518964553976, "grad_norm": 0.5571304040432321, "learning_rate": 8.59465791041139e-05, "loss": 2.9217, "step": 15473 }, { "epoch": 0.9605810416537339, "grad_norm": 0.48264381770568243, "learning_rate": 8.594406870906847e-05, "loss": 2.9358, "step": 15474 }, { "epoch": 0.9606431187534918, "grad_norm": 0.5739244758003036, "learning_rate": 8.59415581264941e-05, "loss": 2.9089, "step": 15475 }, { "epoch": 0.9607051958532498, "grad_norm": 0.5592105900909105, "learning_rate": 8.59390473564039e-05, "loss": 2.9901, "step": 15476 }, { "epoch": 0.9607672729530077, "grad_norm": 0.5419157232010112, "learning_rate": 8.593653639881096e-05, "loss": 2.9799, "step": 15477 }, { "epoch": 0.9608293500527655, "grad_norm": 0.5637652155309383, "learning_rate": 8.593402525372838e-05, "loss": 3.0384, "step": 15478 }, { "epoch": 0.9608914271525234, "grad_norm": 0.5039731358339484, "learning_rate": 8.59315139211693e-05, "loss": 2.9771, "step": 15479 }, { "epoch": 0.9609535042522813, "grad_norm": 0.5378217654452472, "learning_rate": 8.592900240114676e-05, "loss": 2.9114, "step": 15480 }, { "epoch": 0.9610155813520392, "grad_norm": 0.5098506812683384, "learning_rate": 8.592649069367389e-05, "loss": 2.9177, "step": 15481 }, { "epoch": 0.9610776584517972, "grad_norm": 0.49364238242221486, "learning_rate": 8.59239787987638e-05, "loss": 2.955, "step": 15482 }, { "epoch": 0.9611397355515551, "grad_norm": 0.5549297341122249, "learning_rate": 8.59214667164296e-05, "loss": 2.9769, "step": 15483 }, { "epoch": 0.9612018126513129, "grad_norm": 0.5477040410106383, "learning_rate": 8.591895444668438e-05, "loss": 2.9478, "step": 15484 }, { "epoch": 0.9612638897510708, "grad_norm": 0.47615171144148455, "learning_rate": 8.591644198954126e-05, "loss": 2.9878, "step": 15485 }, { "epoch": 0.9613259668508287, "grad_norm": 0.523648034642499, "learning_rate": 8.591392934501334e-05, "loss": 2.907, "step": 15486 }, { "epoch": 0.9613880439505866, "grad_norm": 0.4855094207778775, "learning_rate": 8.591141651311374e-05, "loss": 2.9137, "step": 15487 }, { "epoch": 0.9614501210503446, "grad_norm": 0.4815825446658661, "learning_rate": 8.590890349385557e-05, "loss": 2.9844, "step": 15488 }, { "epoch": 0.9615121981501025, "grad_norm": 0.5244974535342968, "learning_rate": 8.590639028725191e-05, "loss": 2.9136, "step": 15489 }, { "epoch": 0.9615742752498603, "grad_norm": 0.48795941236724827, "learning_rate": 8.590387689331589e-05, "loss": 2.9453, "step": 15490 }, { "epoch": 0.9616363523496182, "grad_norm": 0.4731714105204795, "learning_rate": 8.590136331206064e-05, "loss": 2.9916, "step": 15491 }, { "epoch": 0.9616984294493761, "grad_norm": 0.4938815195227296, "learning_rate": 8.589884954349927e-05, "loss": 2.9293, "step": 15492 }, { "epoch": 0.961760506549134, "grad_norm": 0.4573352976110155, "learning_rate": 8.589633558764489e-05, "loss": 2.9284, "step": 15493 }, { "epoch": 0.961822583648892, "grad_norm": 0.45523199664428005, "learning_rate": 8.589382144451061e-05, "loss": 2.8861, "step": 15494 }, { "epoch": 0.9618846607486499, "grad_norm": 0.5045792572818664, "learning_rate": 8.589130711410956e-05, "loss": 2.9625, "step": 15495 }, { "epoch": 0.9619467378484077, "grad_norm": 0.4708999760340362, "learning_rate": 8.588879259645482e-05, "loss": 2.9579, "step": 15496 }, { "epoch": 0.9620088149481656, "grad_norm": 0.4917081587735399, "learning_rate": 8.588627789155955e-05, "loss": 2.9565, "step": 15497 }, { "epoch": 0.9620708920479235, "grad_norm": 0.5327250255027786, "learning_rate": 8.588376299943686e-05, "loss": 2.9181, "step": 15498 }, { "epoch": 0.9621329691476814, "grad_norm": 0.5026932243809501, "learning_rate": 8.588124792009985e-05, "loss": 2.9624, "step": 15499 }, { "epoch": 0.9621950462474393, "grad_norm": 0.5191305235604009, "learning_rate": 8.587873265356167e-05, "loss": 2.9065, "step": 15500 }, { "epoch": 0.9622571233471973, "grad_norm": 0.4583631816450417, "learning_rate": 8.587621719983546e-05, "loss": 2.8818, "step": 15501 }, { "epoch": 0.9623192004469551, "grad_norm": 0.4831554225852617, "learning_rate": 8.587370155893429e-05, "loss": 2.9243, "step": 15502 }, { "epoch": 0.962381277546713, "grad_norm": 0.4809143822054471, "learning_rate": 8.587118573087131e-05, "loss": 2.9556, "step": 15503 }, { "epoch": 0.9624433546464709, "grad_norm": 0.49906086641353614, "learning_rate": 8.586866971565965e-05, "loss": 2.9358, "step": 15504 }, { "epoch": 0.9625054317462288, "grad_norm": 0.5521693222689402, "learning_rate": 8.586615351331242e-05, "loss": 2.932, "step": 15505 }, { "epoch": 0.9625675088459867, "grad_norm": 0.504580084173114, "learning_rate": 8.586363712384277e-05, "loss": 3.0361, "step": 15506 }, { "epoch": 0.9626295859457447, "grad_norm": 0.47251104851576453, "learning_rate": 8.586112054726382e-05, "loss": 2.9467, "step": 15507 }, { "epoch": 0.9626916630455025, "grad_norm": 0.5812203307424364, "learning_rate": 8.58586037835887e-05, "loss": 2.9453, "step": 15508 }, { "epoch": 0.9627537401452604, "grad_norm": 0.5071066131775174, "learning_rate": 8.585608683283053e-05, "loss": 2.9311, "step": 15509 }, { "epoch": 0.9628158172450183, "grad_norm": 0.5192221388819601, "learning_rate": 8.585356969500247e-05, "loss": 2.9439, "step": 15510 }, { "epoch": 0.9628778943447762, "grad_norm": 0.5176907111020768, "learning_rate": 8.585105237011762e-05, "loss": 2.866, "step": 15511 }, { "epoch": 0.9629399714445341, "grad_norm": 0.4656204818659992, "learning_rate": 8.584853485818912e-05, "loss": 2.8731, "step": 15512 }, { "epoch": 0.963002048544292, "grad_norm": 0.6009004095238953, "learning_rate": 8.584601715923012e-05, "loss": 2.9459, "step": 15513 }, { "epoch": 0.9630641256440499, "grad_norm": 0.49854028233985404, "learning_rate": 8.584349927325375e-05, "loss": 3.0081, "step": 15514 }, { "epoch": 0.9631262027438078, "grad_norm": 0.5433045042128246, "learning_rate": 8.584098120027314e-05, "loss": 2.9174, "step": 15515 }, { "epoch": 0.9631882798435657, "grad_norm": 0.5312579939745664, "learning_rate": 8.583846294030142e-05, "loss": 2.8922, "step": 15516 }, { "epoch": 0.9632503569433236, "grad_norm": 0.5188180717938351, "learning_rate": 8.583594449335175e-05, "loss": 3.0045, "step": 15517 }, { "epoch": 0.9633124340430815, "grad_norm": 0.5066438873350688, "learning_rate": 8.583342585943725e-05, "loss": 2.9943, "step": 15518 }, { "epoch": 0.9633745111428395, "grad_norm": 0.583364840632994, "learning_rate": 8.583090703857108e-05, "loss": 2.9313, "step": 15519 }, { "epoch": 0.9634365882425973, "grad_norm": 0.5542898767740592, "learning_rate": 8.582838803076636e-05, "loss": 2.9241, "step": 15520 }, { "epoch": 0.9634986653423552, "grad_norm": 0.5391253282725986, "learning_rate": 8.582586883603625e-05, "loss": 3.0467, "step": 15521 }, { "epoch": 0.9635607424421131, "grad_norm": 0.5039476776836338, "learning_rate": 8.582334945439387e-05, "loss": 2.8078, "step": 15522 }, { "epoch": 0.963622819541871, "grad_norm": 0.48357528380108716, "learning_rate": 8.58208298858524e-05, "loss": 2.8807, "step": 15523 }, { "epoch": 0.9636848966416289, "grad_norm": 0.5775724038235848, "learning_rate": 8.581831013042495e-05, "loss": 2.9725, "step": 15524 }, { "epoch": 0.9637469737413868, "grad_norm": 0.467793967925352, "learning_rate": 8.58157901881247e-05, "loss": 2.9792, "step": 15525 }, { "epoch": 0.9638090508411447, "grad_norm": 0.5255050534568864, "learning_rate": 8.581327005896475e-05, "loss": 3.0281, "step": 15526 }, { "epoch": 0.9638711279409026, "grad_norm": 0.4834015354866401, "learning_rate": 8.581074974295829e-05, "loss": 3.0224, "step": 15527 }, { "epoch": 0.9639332050406605, "grad_norm": 0.5170199281220085, "learning_rate": 8.580822924011844e-05, "loss": 3.0669, "step": 15528 }, { "epoch": 0.9639952821404184, "grad_norm": 0.4801294477823345, "learning_rate": 8.58057085504584e-05, "loss": 2.9064, "step": 15529 }, { "epoch": 0.9640573592401763, "grad_norm": 0.5160600676301946, "learning_rate": 8.580318767399126e-05, "loss": 2.9563, "step": 15530 }, { "epoch": 0.9641194363399342, "grad_norm": 0.48466851970173197, "learning_rate": 8.580066661073021e-05, "loss": 2.9739, "step": 15531 }, { "epoch": 0.964181513439692, "grad_norm": 0.5312853969674428, "learning_rate": 8.579814536068838e-05, "loss": 2.885, "step": 15532 }, { "epoch": 0.96424359053945, "grad_norm": 0.43647864877888787, "learning_rate": 8.579562392387894e-05, "loss": 2.8414, "step": 15533 }, { "epoch": 0.9643056676392079, "grad_norm": 0.4883911924565444, "learning_rate": 8.579310230031504e-05, "loss": 3.0162, "step": 15534 }, { "epoch": 0.9643677447389658, "grad_norm": 0.49408825683531926, "learning_rate": 8.579058049000984e-05, "loss": 3.0244, "step": 15535 }, { "epoch": 0.9644298218387237, "grad_norm": 0.6149049391587392, "learning_rate": 8.578805849297648e-05, "loss": 2.9073, "step": 15536 }, { "epoch": 0.9644918989384816, "grad_norm": 0.5370508890387365, "learning_rate": 8.578553630922815e-05, "loss": 2.8776, "step": 15537 }, { "epoch": 0.9645539760382394, "grad_norm": 0.49176099250911276, "learning_rate": 8.578301393877799e-05, "loss": 2.9686, "step": 15538 }, { "epoch": 0.9646160531379974, "grad_norm": 0.5457399578402528, "learning_rate": 8.578049138163915e-05, "loss": 2.9375, "step": 15539 }, { "epoch": 0.9646781302377553, "grad_norm": 0.5659117019477489, "learning_rate": 8.57779686378248e-05, "loss": 2.936, "step": 15540 }, { "epoch": 0.9647402073375132, "grad_norm": 0.5148233192569422, "learning_rate": 8.57754457073481e-05, "loss": 2.9894, "step": 15541 }, { "epoch": 0.9648022844372711, "grad_norm": 0.45865890650932784, "learning_rate": 8.577292259022223e-05, "loss": 2.8759, "step": 15542 }, { "epoch": 0.964864361537029, "grad_norm": 0.49082607109556453, "learning_rate": 8.577039928646033e-05, "loss": 2.8749, "step": 15543 }, { "epoch": 0.9649264386367868, "grad_norm": 0.46532964555703205, "learning_rate": 8.576787579607557e-05, "loss": 2.8466, "step": 15544 }, { "epoch": 0.9649885157365448, "grad_norm": 0.530807198380657, "learning_rate": 8.576535211908112e-05, "loss": 2.9219, "step": 15545 }, { "epoch": 0.9650505928363027, "grad_norm": 0.46559793534443766, "learning_rate": 8.576282825549014e-05, "loss": 2.8811, "step": 15546 }, { "epoch": 0.9651126699360606, "grad_norm": 0.4947161432818885, "learning_rate": 8.576030420531581e-05, "loss": 2.8626, "step": 15547 }, { "epoch": 0.9651747470358185, "grad_norm": 0.4539944594219506, "learning_rate": 8.575777996857128e-05, "loss": 2.9159, "step": 15548 }, { "epoch": 0.9652368241355764, "grad_norm": 0.46248238504218103, "learning_rate": 8.575525554526975e-05, "loss": 2.8748, "step": 15549 }, { "epoch": 0.9652989012353342, "grad_norm": 0.47693541272866036, "learning_rate": 8.575273093542437e-05, "loss": 2.8902, "step": 15550 }, { "epoch": 0.9653609783350922, "grad_norm": 0.4503731696922512, "learning_rate": 8.57502061390483e-05, "loss": 2.97, "step": 15551 }, { "epoch": 0.9654230554348501, "grad_norm": 0.5040495321328582, "learning_rate": 8.574768115615475e-05, "loss": 2.9093, "step": 15552 }, { "epoch": 0.965485132534608, "grad_norm": 0.5405303762018313, "learning_rate": 8.574515598675684e-05, "loss": 2.9828, "step": 15553 }, { "epoch": 0.9655472096343659, "grad_norm": 0.43664917639048934, "learning_rate": 8.574263063086778e-05, "loss": 2.8864, "step": 15554 }, { "epoch": 0.9656092867341238, "grad_norm": 0.5419586996094509, "learning_rate": 8.574010508850076e-05, "loss": 2.9814, "step": 15555 }, { "epoch": 0.9656713638338816, "grad_norm": 0.46061313660457126, "learning_rate": 8.573757935966892e-05, "loss": 2.8447, "step": 15556 }, { "epoch": 0.9657334409336396, "grad_norm": 0.5451127951816476, "learning_rate": 8.573505344438546e-05, "loss": 2.854, "step": 15557 }, { "epoch": 0.9657955180333975, "grad_norm": 0.45457484694839534, "learning_rate": 8.573252734266354e-05, "loss": 2.8759, "step": 15558 }, { "epoch": 0.9658575951331554, "grad_norm": 0.5488551394510134, "learning_rate": 8.573000105451636e-05, "loss": 2.8827, "step": 15559 }, { "epoch": 0.9659196722329133, "grad_norm": 0.5090656956243937, "learning_rate": 8.572747457995708e-05, "loss": 2.909, "step": 15560 }, { "epoch": 0.9659817493326712, "grad_norm": 0.5395601435750929, "learning_rate": 8.57249479189989e-05, "loss": 2.97, "step": 15561 }, { "epoch": 0.966043826432429, "grad_norm": 0.5757998801210391, "learning_rate": 8.5722421071655e-05, "loss": 3.0092, "step": 15562 }, { "epoch": 0.966105903532187, "grad_norm": 0.5452970355008033, "learning_rate": 8.571989403793854e-05, "loss": 2.9851, "step": 15563 }, { "epoch": 0.9661679806319449, "grad_norm": 0.4986150604686098, "learning_rate": 8.571736681786273e-05, "loss": 2.9781, "step": 15564 }, { "epoch": 0.9662300577317028, "grad_norm": 0.5281363006375936, "learning_rate": 8.571483941144076e-05, "loss": 2.967, "step": 15565 }, { "epoch": 0.9662921348314607, "grad_norm": 0.5313835817575968, "learning_rate": 8.571231181868579e-05, "loss": 2.9358, "step": 15566 }, { "epoch": 0.9663542119312186, "grad_norm": 0.4773769922746797, "learning_rate": 8.570978403961101e-05, "loss": 2.8297, "step": 15567 }, { "epoch": 0.9664162890309764, "grad_norm": 0.5366894554350446, "learning_rate": 8.570725607422964e-05, "loss": 2.8557, "step": 15568 }, { "epoch": 0.9664783661307343, "grad_norm": 0.516278257681928, "learning_rate": 8.570472792255484e-05, "loss": 2.9448, "step": 15569 }, { "epoch": 0.9665404432304923, "grad_norm": 0.5319408244675763, "learning_rate": 8.57021995845998e-05, "loss": 2.9854, "step": 15570 }, { "epoch": 0.9666025203302502, "grad_norm": 0.5510552448505955, "learning_rate": 8.569967106037771e-05, "loss": 2.8905, "step": 15571 }, { "epoch": 0.9666645974300081, "grad_norm": 0.4865784364759773, "learning_rate": 8.569714234990178e-05, "loss": 2.9195, "step": 15572 }, { "epoch": 0.966726674529766, "grad_norm": 0.6142476722353346, "learning_rate": 8.569461345318519e-05, "loss": 2.9122, "step": 15573 }, { "epoch": 0.9667887516295238, "grad_norm": 0.49237480837994246, "learning_rate": 8.569208437024115e-05, "loss": 2.867, "step": 15574 }, { "epoch": 0.9668508287292817, "grad_norm": 0.5357875302742938, "learning_rate": 8.568955510108283e-05, "loss": 2.8542, "step": 15575 }, { "epoch": 0.9669129058290397, "grad_norm": 0.5071493954544052, "learning_rate": 8.568702564572344e-05, "loss": 2.8979, "step": 15576 }, { "epoch": 0.9669749829287976, "grad_norm": 0.519366159237605, "learning_rate": 8.568449600417617e-05, "loss": 2.9037, "step": 15577 }, { "epoch": 0.9670370600285555, "grad_norm": 0.46126454276132417, "learning_rate": 8.568196617645423e-05, "loss": 2.8133, "step": 15578 }, { "epoch": 0.9670991371283134, "grad_norm": 0.4465319850898761, "learning_rate": 8.567943616257079e-05, "loss": 2.9696, "step": 15579 }, { "epoch": 0.9671612142280712, "grad_norm": 0.5796693588256107, "learning_rate": 8.567690596253908e-05, "loss": 2.9492, "step": 15580 }, { "epoch": 0.9672232913278291, "grad_norm": 0.5020162935994276, "learning_rate": 8.56743755763723e-05, "loss": 2.9349, "step": 15581 }, { "epoch": 0.9672853684275871, "grad_norm": 0.48658335166356637, "learning_rate": 8.567184500408363e-05, "loss": 2.9968, "step": 15582 }, { "epoch": 0.967347445527345, "grad_norm": 0.6032481171649333, "learning_rate": 8.56693142456863e-05, "loss": 2.9565, "step": 15583 }, { "epoch": 0.9674095226271029, "grad_norm": 0.5050871857835708, "learning_rate": 8.566678330119348e-05, "loss": 2.9317, "step": 15584 }, { "epoch": 0.9674715997268608, "grad_norm": 0.5727683193152678, "learning_rate": 8.566425217061841e-05, "loss": 2.8455, "step": 15585 }, { "epoch": 0.9675336768266186, "grad_norm": 0.5376462992351789, "learning_rate": 8.566172085397427e-05, "loss": 2.9644, "step": 15586 }, { "epoch": 0.9675957539263765, "grad_norm": 0.5275896135025808, "learning_rate": 8.565918935127429e-05, "loss": 2.9051, "step": 15587 }, { "epoch": 0.9676578310261345, "grad_norm": 0.5450927029082245, "learning_rate": 8.565665766253163e-05, "loss": 2.997, "step": 15588 }, { "epoch": 0.9677199081258924, "grad_norm": 0.5760161884552935, "learning_rate": 8.565412578775955e-05, "loss": 2.8978, "step": 15589 }, { "epoch": 0.9677819852256503, "grad_norm": 0.48464069462655246, "learning_rate": 8.565159372697124e-05, "loss": 2.9535, "step": 15590 }, { "epoch": 0.9678440623254082, "grad_norm": 0.5273665811753261, "learning_rate": 8.564906148017992e-05, "loss": 2.9495, "step": 15591 }, { "epoch": 0.967906139425166, "grad_norm": 0.5754534622966577, "learning_rate": 8.564652904739878e-05, "loss": 2.9796, "step": 15592 }, { "epoch": 0.9679682165249239, "grad_norm": 0.5338813743791319, "learning_rate": 8.564399642864105e-05, "loss": 2.9953, "step": 15593 }, { "epoch": 0.9680302936246818, "grad_norm": 0.5163668872876531, "learning_rate": 8.564146362391994e-05, "loss": 3.0376, "step": 15594 }, { "epoch": 0.9680923707244398, "grad_norm": 0.5627605841296746, "learning_rate": 8.563893063324865e-05, "loss": 3.0277, "step": 15595 }, { "epoch": 0.9681544478241977, "grad_norm": 0.547885370146365, "learning_rate": 8.563639745664041e-05, "loss": 3.0055, "step": 15596 }, { "epoch": 0.9682165249239556, "grad_norm": 0.5638889542419362, "learning_rate": 8.563386409410845e-05, "loss": 2.9344, "step": 15597 }, { "epoch": 0.9682786020237134, "grad_norm": 0.48545328221819967, "learning_rate": 8.563133054566596e-05, "loss": 2.8614, "step": 15598 }, { "epoch": 0.9683406791234713, "grad_norm": 0.5280883900520332, "learning_rate": 8.562879681132616e-05, "loss": 2.8922, "step": 15599 }, { "epoch": 0.9684027562232292, "grad_norm": 0.49504760012401816, "learning_rate": 8.562626289110229e-05, "loss": 2.8961, "step": 15600 }, { "epoch": 0.9684648333229872, "grad_norm": 0.5426524221964637, "learning_rate": 8.562372878500755e-05, "loss": 2.9293, "step": 15601 }, { "epoch": 0.9685269104227451, "grad_norm": 0.5281741068654939, "learning_rate": 8.562119449305518e-05, "loss": 2.9722, "step": 15602 }, { "epoch": 0.968588987522503, "grad_norm": 0.4409614448615409, "learning_rate": 8.561866001525838e-05, "loss": 2.8526, "step": 15603 }, { "epoch": 0.9686510646222608, "grad_norm": 0.5504406626733833, "learning_rate": 8.561612535163039e-05, "loss": 2.8711, "step": 15604 }, { "epoch": 0.9687131417220187, "grad_norm": 0.5195745865953129, "learning_rate": 8.561359050218443e-05, "loss": 2.9617, "step": 15605 }, { "epoch": 0.9687752188217766, "grad_norm": 0.5049919438427705, "learning_rate": 8.56110554669337e-05, "loss": 2.9894, "step": 15606 }, { "epoch": 0.9688372959215346, "grad_norm": 0.5059073583755826, "learning_rate": 8.560852024589148e-05, "loss": 2.848, "step": 15607 }, { "epoch": 0.9688993730212925, "grad_norm": 0.5461920424644712, "learning_rate": 8.560598483907095e-05, "loss": 2.9948, "step": 15608 }, { "epoch": 0.9689614501210504, "grad_norm": 0.46865552903921337, "learning_rate": 8.560344924648536e-05, "loss": 2.887, "step": 15609 }, { "epoch": 0.9690235272208082, "grad_norm": 0.518501983090676, "learning_rate": 8.560091346814793e-05, "loss": 2.9508, "step": 15610 }, { "epoch": 0.9690856043205661, "grad_norm": 0.513988614787462, "learning_rate": 8.55983775040719e-05, "loss": 2.9797, "step": 15611 }, { "epoch": 0.969147681420324, "grad_norm": 0.5076562200835675, "learning_rate": 8.559584135427048e-05, "loss": 3.0646, "step": 15612 }, { "epoch": 0.969209758520082, "grad_norm": 0.51219350326027, "learning_rate": 8.559330501875691e-05, "loss": 2.9477, "step": 15613 }, { "epoch": 0.9692718356198399, "grad_norm": 0.5094492431080899, "learning_rate": 8.559076849754444e-05, "loss": 3.0225, "step": 15614 }, { "epoch": 0.9693339127195978, "grad_norm": 0.5502418961471993, "learning_rate": 8.558823179064628e-05, "loss": 2.9418, "step": 15615 }, { "epoch": 0.9693959898193556, "grad_norm": 0.49623060966871363, "learning_rate": 8.558569489807569e-05, "loss": 2.9774, "step": 15616 }, { "epoch": 0.9694580669191135, "grad_norm": 0.5780649153957662, "learning_rate": 8.558315781984587e-05, "loss": 2.9508, "step": 15617 }, { "epoch": 0.9695201440188714, "grad_norm": 0.5035918385998761, "learning_rate": 8.55806205559701e-05, "loss": 2.9314, "step": 15618 }, { "epoch": 0.9695822211186294, "grad_norm": 0.555909811427186, "learning_rate": 8.557808310646159e-05, "loss": 2.8468, "step": 15619 }, { "epoch": 0.9696442982183873, "grad_norm": 0.5273243088140919, "learning_rate": 8.557554547133357e-05, "loss": 2.9107, "step": 15620 }, { "epoch": 0.9697063753181452, "grad_norm": 0.5600617599065228, "learning_rate": 8.557300765059929e-05, "loss": 3.053, "step": 15621 }, { "epoch": 0.969768452417903, "grad_norm": 0.516606639123213, "learning_rate": 8.5570469644272e-05, "loss": 2.9238, "step": 15622 }, { "epoch": 0.9698305295176609, "grad_norm": 0.537452227665164, "learning_rate": 8.556793145236495e-05, "loss": 2.8388, "step": 15623 }, { "epoch": 0.9698926066174188, "grad_norm": 0.5464887617479122, "learning_rate": 8.556539307489136e-05, "loss": 2.8227, "step": 15624 }, { "epoch": 0.9699546837171767, "grad_norm": 0.5782017012569453, "learning_rate": 8.556285451186447e-05, "loss": 2.8791, "step": 15625 }, { "epoch": 0.9700167608169347, "grad_norm": 0.5937513050265777, "learning_rate": 8.556031576329755e-05, "loss": 2.8414, "step": 15626 }, { "epoch": 0.9700788379166926, "grad_norm": 0.49466588446021037, "learning_rate": 8.555777682920382e-05, "loss": 3.0239, "step": 15627 }, { "epoch": 0.9701409150164504, "grad_norm": 0.511801793006836, "learning_rate": 8.555523770959653e-05, "loss": 2.9228, "step": 15628 }, { "epoch": 0.9702029921162083, "grad_norm": 0.5045348394177044, "learning_rate": 8.555269840448894e-05, "loss": 2.9846, "step": 15629 }, { "epoch": 0.9702650692159662, "grad_norm": 0.45444300130893744, "learning_rate": 8.555015891389431e-05, "loss": 2.9547, "step": 15630 }, { "epoch": 0.9703271463157241, "grad_norm": 0.4962476806922735, "learning_rate": 8.554761923782584e-05, "loss": 2.8428, "step": 15631 }, { "epoch": 0.9703892234154821, "grad_norm": 0.47694539494229365, "learning_rate": 8.554507937629683e-05, "loss": 2.9045, "step": 15632 }, { "epoch": 0.97045130051524, "grad_norm": 0.5107190508005666, "learning_rate": 8.55425393293205e-05, "loss": 2.8798, "step": 15633 }, { "epoch": 0.9705133776149978, "grad_norm": 0.517921851841368, "learning_rate": 8.553999909691013e-05, "loss": 2.8666, "step": 15634 }, { "epoch": 0.9705754547147557, "grad_norm": 0.5361588546459318, "learning_rate": 8.553745867907894e-05, "loss": 2.9538, "step": 15635 }, { "epoch": 0.9706375318145136, "grad_norm": 0.48813333941380377, "learning_rate": 8.553491807584022e-05, "loss": 2.9248, "step": 15636 }, { "epoch": 0.9706996089142715, "grad_norm": 0.5007773555419698, "learning_rate": 8.55323772872072e-05, "loss": 2.9262, "step": 15637 }, { "epoch": 0.9707616860140295, "grad_norm": 0.5057295702675002, "learning_rate": 8.552983631319314e-05, "loss": 3.0152, "step": 15638 }, { "epoch": 0.9708237631137874, "grad_norm": 0.47609612830193676, "learning_rate": 8.552729515381129e-05, "loss": 2.8637, "step": 15639 }, { "epoch": 0.9708858402135452, "grad_norm": 0.4622876872074346, "learning_rate": 8.552475380907492e-05, "loss": 2.9216, "step": 15640 }, { "epoch": 0.9709479173133031, "grad_norm": 0.49668100096618534, "learning_rate": 8.552221227899728e-05, "loss": 2.9057, "step": 15641 }, { "epoch": 0.971009994413061, "grad_norm": 0.511740327336484, "learning_rate": 8.551967056359165e-05, "loss": 2.8624, "step": 15642 }, { "epoch": 0.9710720715128189, "grad_norm": 0.47450256715668976, "learning_rate": 8.551712866287127e-05, "loss": 2.9552, "step": 15643 }, { "epoch": 0.9711341486125769, "grad_norm": 0.5497824628820155, "learning_rate": 8.551458657684941e-05, "loss": 3.0035, "step": 15644 }, { "epoch": 0.9711962257123348, "grad_norm": 0.47168058786247047, "learning_rate": 8.551204430553932e-05, "loss": 2.998, "step": 15645 }, { "epoch": 0.9712583028120926, "grad_norm": 0.5615602166114029, "learning_rate": 8.550950184895429e-05, "loss": 3.0234, "step": 15646 }, { "epoch": 0.9713203799118505, "grad_norm": 0.6181266255063244, "learning_rate": 8.550695920710756e-05, "loss": 2.9988, "step": 15647 }, { "epoch": 0.9713824570116084, "grad_norm": 0.5299408995767546, "learning_rate": 8.55044163800124e-05, "loss": 2.9208, "step": 15648 }, { "epoch": 0.9714445341113663, "grad_norm": 0.4916521369947771, "learning_rate": 8.55018733676821e-05, "loss": 2.9091, "step": 15649 }, { "epoch": 0.9715066112111242, "grad_norm": 0.45947325622324836, "learning_rate": 8.549933017012987e-05, "loss": 2.7616, "step": 15650 }, { "epoch": 0.9715686883108822, "grad_norm": 0.4603107847038055, "learning_rate": 8.549678678736906e-05, "loss": 2.8202, "step": 15651 }, { "epoch": 0.97163076541064, "grad_norm": 0.5335885356240577, "learning_rate": 8.549424321941286e-05, "loss": 3.0424, "step": 15652 }, { "epoch": 0.9716928425103979, "grad_norm": 0.4964874087405315, "learning_rate": 8.54916994662746e-05, "loss": 2.9502, "step": 15653 }, { "epoch": 0.9717549196101558, "grad_norm": 0.5060521940275161, "learning_rate": 8.548915552796752e-05, "loss": 2.9788, "step": 15654 }, { "epoch": 0.9718169967099137, "grad_norm": 0.4892297458783602, "learning_rate": 8.54866114045049e-05, "loss": 2.9185, "step": 15655 }, { "epoch": 0.9718790738096716, "grad_norm": 0.51306630108987, "learning_rate": 8.548406709590003e-05, "loss": 3.0163, "step": 15656 }, { "epoch": 0.9719411509094296, "grad_norm": 0.45784509367026693, "learning_rate": 8.548152260216614e-05, "loss": 2.9646, "step": 15657 }, { "epoch": 0.9720032280091874, "grad_norm": 0.5406586608991718, "learning_rate": 8.547897792331655e-05, "loss": 2.9248, "step": 15658 }, { "epoch": 0.9720653051089453, "grad_norm": 0.49806621738348134, "learning_rate": 8.547643305936451e-05, "loss": 2.968, "step": 15659 }, { "epoch": 0.9721273822087032, "grad_norm": 0.49576389430240925, "learning_rate": 8.547388801032333e-05, "loss": 2.9552, "step": 15660 }, { "epoch": 0.9721894593084611, "grad_norm": 0.474907581222806, "learning_rate": 8.547134277620623e-05, "loss": 2.9346, "step": 15661 }, { "epoch": 0.972251536408219, "grad_norm": 0.5082032680877789, "learning_rate": 8.546879735702654e-05, "loss": 2.9171, "step": 15662 }, { "epoch": 0.972313613507977, "grad_norm": 0.4958255523450448, "learning_rate": 8.546625175279752e-05, "loss": 2.9029, "step": 15663 }, { "epoch": 0.9723756906077348, "grad_norm": 0.5664510248836755, "learning_rate": 8.546370596353246e-05, "loss": 3.0036, "step": 15664 }, { "epoch": 0.9724377677074927, "grad_norm": 0.48735672972903177, "learning_rate": 8.546115998924464e-05, "loss": 2.9287, "step": 15665 }, { "epoch": 0.9724998448072506, "grad_norm": 0.5207132646127176, "learning_rate": 8.545861382994733e-05, "loss": 2.8893, "step": 15666 }, { "epoch": 0.9725619219070085, "grad_norm": 0.47549172786232075, "learning_rate": 8.545606748565384e-05, "loss": 2.9802, "step": 15667 }, { "epoch": 0.9726239990067664, "grad_norm": 0.5033819738889669, "learning_rate": 8.545352095637743e-05, "loss": 2.9546, "step": 15668 }, { "epoch": 0.9726860761065244, "grad_norm": 0.4858760801057462, "learning_rate": 8.545097424213139e-05, "loss": 2.9255, "step": 15669 }, { "epoch": 0.9727481532062822, "grad_norm": 0.4979726698702893, "learning_rate": 8.544842734292901e-05, "loss": 2.9144, "step": 15670 }, { "epoch": 0.9728102303060401, "grad_norm": 0.481754242551223, "learning_rate": 8.544588025878358e-05, "loss": 2.9822, "step": 15671 }, { "epoch": 0.972872307405798, "grad_norm": 0.5401503325853539, "learning_rate": 8.544333298970839e-05, "loss": 2.9086, "step": 15672 }, { "epoch": 0.9729343845055559, "grad_norm": 0.4972582388761583, "learning_rate": 8.544078553571672e-05, "loss": 2.9918, "step": 15673 }, { "epoch": 0.9729964616053138, "grad_norm": 0.5462141130823354, "learning_rate": 8.543823789682189e-05, "loss": 2.9165, "step": 15674 }, { "epoch": 0.9730585387050718, "grad_norm": 0.5789489030247295, "learning_rate": 8.543569007303715e-05, "loss": 3.034, "step": 15675 }, { "epoch": 0.9731206158048296, "grad_norm": 0.49827607808579505, "learning_rate": 8.543314206437583e-05, "loss": 2.8849, "step": 15676 }, { "epoch": 0.9731826929045875, "grad_norm": 0.5096197102159645, "learning_rate": 8.543059387085119e-05, "loss": 2.9464, "step": 15677 }, { "epoch": 0.9732447700043454, "grad_norm": 0.5160398548337228, "learning_rate": 8.542804549247654e-05, "loss": 2.9166, "step": 15678 }, { "epoch": 0.9733068471041033, "grad_norm": 0.5417237863012213, "learning_rate": 8.542549692926518e-05, "loss": 2.828, "step": 15679 }, { "epoch": 0.9733689242038612, "grad_norm": 0.5360523434477716, "learning_rate": 8.542294818123043e-05, "loss": 3.0314, "step": 15680 }, { "epoch": 0.9734310013036191, "grad_norm": 0.5156512398256481, "learning_rate": 8.542039924838552e-05, "loss": 2.9818, "step": 15681 }, { "epoch": 0.973493078403377, "grad_norm": 0.562393496285298, "learning_rate": 8.541785013074381e-05, "loss": 2.9112, "step": 15682 }, { "epoch": 0.9735551555031349, "grad_norm": 0.49303242654533697, "learning_rate": 8.541530082831856e-05, "loss": 2.816, "step": 15683 }, { "epoch": 0.9736172326028928, "grad_norm": 0.5349173465056268, "learning_rate": 8.541275134112311e-05, "loss": 2.9161, "step": 15684 }, { "epoch": 0.9736793097026507, "grad_norm": 0.5117586717248238, "learning_rate": 8.541020166917073e-05, "loss": 3.012, "step": 15685 }, { "epoch": 0.9737413868024086, "grad_norm": 0.5060858790135073, "learning_rate": 8.540765181247473e-05, "loss": 2.8695, "step": 15686 }, { "epoch": 0.9738034639021665, "grad_norm": 0.507268261875459, "learning_rate": 8.540510177104842e-05, "loss": 2.897, "step": 15687 }, { "epoch": 0.9738655410019244, "grad_norm": 0.5290809275284589, "learning_rate": 8.540255154490509e-05, "loss": 2.9193, "step": 15688 }, { "epoch": 0.9739276181016823, "grad_norm": 0.4999834296341754, "learning_rate": 8.540000113405806e-05, "loss": 2.9565, "step": 15689 }, { "epoch": 0.9739896952014402, "grad_norm": 0.5599183336808342, "learning_rate": 8.539745053852064e-05, "loss": 2.9564, "step": 15690 }, { "epoch": 0.9740517723011981, "grad_norm": 0.4981471540437717, "learning_rate": 8.53948997583061e-05, "loss": 2.8848, "step": 15691 }, { "epoch": 0.974113849400956, "grad_norm": 0.5692926720853668, "learning_rate": 8.53923487934278e-05, "loss": 2.9363, "step": 15692 }, { "epoch": 0.9741759265007139, "grad_norm": 0.5153973105590988, "learning_rate": 8.538979764389903e-05, "loss": 2.8983, "step": 15693 }, { "epoch": 0.9742380036004717, "grad_norm": 0.4811149841261113, "learning_rate": 8.538724630973307e-05, "loss": 2.9399, "step": 15694 }, { "epoch": 0.9743000807002297, "grad_norm": 0.510291285263662, "learning_rate": 8.538469479094327e-05, "loss": 2.9495, "step": 15695 }, { "epoch": 0.9743621577999876, "grad_norm": 0.5204880300917258, "learning_rate": 8.538214308754292e-05, "loss": 3.0244, "step": 15696 }, { "epoch": 0.9744242348997455, "grad_norm": 0.5073856613944987, "learning_rate": 8.537959119954534e-05, "loss": 3.0086, "step": 15697 }, { "epoch": 0.9744863119995034, "grad_norm": 0.5400528311052081, "learning_rate": 8.537703912696387e-05, "loss": 3.0106, "step": 15698 }, { "epoch": 0.9745483890992613, "grad_norm": 0.5382865628670989, "learning_rate": 8.537448686981177e-05, "loss": 2.9059, "step": 15699 }, { "epoch": 0.9746104661990191, "grad_norm": 0.4723515277036705, "learning_rate": 8.537193442810238e-05, "loss": 2.9824, "step": 15700 }, { "epoch": 0.9746725432987771, "grad_norm": 0.4951921966736365, "learning_rate": 8.536938180184905e-05, "loss": 2.8566, "step": 15701 }, { "epoch": 0.974734620398535, "grad_norm": 0.47684612589223657, "learning_rate": 8.536682899106505e-05, "loss": 2.9046, "step": 15702 }, { "epoch": 0.9747966974982929, "grad_norm": 0.4930270013988225, "learning_rate": 8.536427599576371e-05, "loss": 3.0554, "step": 15703 }, { "epoch": 0.9748587745980508, "grad_norm": 0.5343117838270327, "learning_rate": 8.536172281595838e-05, "loss": 2.8793, "step": 15704 }, { "epoch": 0.9749208516978087, "grad_norm": 0.5114202429526362, "learning_rate": 8.535916945166233e-05, "loss": 2.9001, "step": 15705 }, { "epoch": 0.9749829287975665, "grad_norm": 0.5353252951389469, "learning_rate": 8.535661590288893e-05, "loss": 2.8945, "step": 15706 }, { "epoch": 0.9750450058973245, "grad_norm": 0.48028835538738274, "learning_rate": 8.535406216965149e-05, "loss": 2.9391, "step": 15707 }, { "epoch": 0.9751070829970824, "grad_norm": 0.5544629784308223, "learning_rate": 8.535150825196329e-05, "loss": 2.9273, "step": 15708 }, { "epoch": 0.9751691600968403, "grad_norm": 0.5148812623315745, "learning_rate": 8.534895414983772e-05, "loss": 3.0797, "step": 15709 }, { "epoch": 0.9752312371965982, "grad_norm": 0.5171434988798421, "learning_rate": 8.534639986328808e-05, "loss": 2.9383, "step": 15710 }, { "epoch": 0.9752933142963561, "grad_norm": 0.46067967720516273, "learning_rate": 8.534384539232767e-05, "loss": 2.9103, "step": 15711 }, { "epoch": 0.9753553913961139, "grad_norm": 0.5675495316404108, "learning_rate": 8.534129073696984e-05, "loss": 2.9464, "step": 15712 }, { "epoch": 0.9754174684958719, "grad_norm": 0.5104899321412382, "learning_rate": 8.533873589722792e-05, "loss": 2.9282, "step": 15713 }, { "epoch": 0.9754795455956298, "grad_norm": 0.5225690498083011, "learning_rate": 8.533618087311524e-05, "loss": 2.9887, "step": 15714 }, { "epoch": 0.9755416226953877, "grad_norm": 0.48660161857668965, "learning_rate": 8.533362566464514e-05, "loss": 2.8619, "step": 15715 }, { "epoch": 0.9756036997951456, "grad_norm": 0.5768550120452473, "learning_rate": 8.533107027183092e-05, "loss": 2.9547, "step": 15716 }, { "epoch": 0.9756657768949035, "grad_norm": 0.5272715766402398, "learning_rate": 8.532851469468593e-05, "loss": 2.9735, "step": 15717 }, { "epoch": 0.9757278539946613, "grad_norm": 0.48459220446237256, "learning_rate": 8.532595893322349e-05, "loss": 2.883, "step": 15718 }, { "epoch": 0.9757899310944192, "grad_norm": 0.5040741574192256, "learning_rate": 8.532340298745697e-05, "loss": 2.9369, "step": 15719 }, { "epoch": 0.9758520081941772, "grad_norm": 0.5407223233509055, "learning_rate": 8.532084685739967e-05, "loss": 2.9612, "step": 15720 }, { "epoch": 0.9759140852939351, "grad_norm": 0.4839986908329448, "learning_rate": 8.531829054306493e-05, "loss": 2.9399, "step": 15721 }, { "epoch": 0.975976162393693, "grad_norm": 0.5082809928098322, "learning_rate": 8.531573404446611e-05, "loss": 2.8535, "step": 15722 }, { "epoch": 0.9760382394934509, "grad_norm": 0.6016827747939646, "learning_rate": 8.531317736161652e-05, "loss": 2.9649, "step": 15723 }, { "epoch": 0.9761003165932087, "grad_norm": 0.551513671875, "learning_rate": 8.531062049452953e-05, "loss": 2.9797, "step": 15724 }, { "epoch": 0.9761623936929666, "grad_norm": 0.49895587321917984, "learning_rate": 8.530806344321844e-05, "loss": 2.9137, "step": 15725 }, { "epoch": 0.9762244707927246, "grad_norm": 0.5674216803731983, "learning_rate": 8.530550620769663e-05, "loss": 2.9961, "step": 15726 }, { "epoch": 0.9762865478924825, "grad_norm": 0.48969360557114183, "learning_rate": 8.530294878797743e-05, "loss": 2.9475, "step": 15727 }, { "epoch": 0.9763486249922404, "grad_norm": 0.47435029733241796, "learning_rate": 8.530039118407415e-05, "loss": 2.9083, "step": 15728 }, { "epoch": 0.9764107020919983, "grad_norm": 0.46730583567686734, "learning_rate": 8.529783339600017e-05, "loss": 2.911, "step": 15729 }, { "epoch": 0.9764727791917561, "grad_norm": 0.5070878947963668, "learning_rate": 8.529527542376882e-05, "loss": 2.9702, "step": 15730 }, { "epoch": 0.976534856291514, "grad_norm": 0.5573069549555062, "learning_rate": 8.529271726739345e-05, "loss": 2.9671, "step": 15731 }, { "epoch": 0.976596933391272, "grad_norm": 0.5707234770000446, "learning_rate": 8.52901589268874e-05, "loss": 2.9667, "step": 15732 }, { "epoch": 0.9766590104910299, "grad_norm": 0.5042156482053806, "learning_rate": 8.528760040226405e-05, "loss": 3.0108, "step": 15733 }, { "epoch": 0.9767210875907878, "grad_norm": 0.5524089380885016, "learning_rate": 8.52850416935367e-05, "loss": 2.9615, "step": 15734 }, { "epoch": 0.9767831646905456, "grad_norm": 0.49279773061097415, "learning_rate": 8.528248280071873e-05, "loss": 2.9336, "step": 15735 }, { "epoch": 0.9768452417903035, "grad_norm": 0.5593980454114441, "learning_rate": 8.527992372382347e-05, "loss": 2.9295, "step": 15736 }, { "epoch": 0.9769073188900614, "grad_norm": 0.5212708796977868, "learning_rate": 8.52773644628643e-05, "loss": 2.8703, "step": 15737 }, { "epoch": 0.9769693959898194, "grad_norm": 0.4769450668916447, "learning_rate": 8.527480501785454e-05, "loss": 2.9367, "step": 15738 }, { "epoch": 0.9770314730895773, "grad_norm": 0.5656722212411763, "learning_rate": 8.527224538880759e-05, "loss": 2.9381, "step": 15739 }, { "epoch": 0.9770935501893352, "grad_norm": 0.5675157401132912, "learning_rate": 8.526968557573674e-05, "loss": 3.01, "step": 15740 }, { "epoch": 0.977155627289093, "grad_norm": 0.5682247303883647, "learning_rate": 8.52671255786554e-05, "loss": 2.9166, "step": 15741 }, { "epoch": 0.9772177043888509, "grad_norm": 0.5119566328096476, "learning_rate": 8.526456539757688e-05, "loss": 2.8989, "step": 15742 }, { "epoch": 0.9772797814886088, "grad_norm": 0.5121320555063366, "learning_rate": 8.526200503251458e-05, "loss": 2.9546, "step": 15743 }, { "epoch": 0.9773418585883668, "grad_norm": 0.5787630684038338, "learning_rate": 8.525944448348185e-05, "loss": 2.9772, "step": 15744 }, { "epoch": 0.9774039356881247, "grad_norm": 0.4685106620124966, "learning_rate": 8.525688375049202e-05, "loss": 2.9313, "step": 15745 }, { "epoch": 0.9774660127878826, "grad_norm": 0.5683245041889492, "learning_rate": 8.525432283355846e-05, "loss": 2.9321, "step": 15746 }, { "epoch": 0.9775280898876404, "grad_norm": 0.5179135081664693, "learning_rate": 8.525176173269455e-05, "loss": 2.9205, "step": 15747 }, { "epoch": 0.9775901669873983, "grad_norm": 0.5711516125298677, "learning_rate": 8.524920044791365e-05, "loss": 3.0217, "step": 15748 }, { "epoch": 0.9776522440871562, "grad_norm": 0.4821675050833452, "learning_rate": 8.524663897922911e-05, "loss": 2.9739, "step": 15749 }, { "epoch": 0.9777143211869141, "grad_norm": 0.5170118292870057, "learning_rate": 8.524407732665429e-05, "loss": 2.948, "step": 15750 }, { "epoch": 0.9777763982866721, "grad_norm": 0.5297249614780554, "learning_rate": 8.524151549020257e-05, "loss": 2.9608, "step": 15751 }, { "epoch": 0.97783847538643, "grad_norm": 0.4821220114657109, "learning_rate": 8.523895346988731e-05, "loss": 2.8684, "step": 15752 }, { "epoch": 0.9779005524861878, "grad_norm": 0.5125830827219464, "learning_rate": 8.523639126572187e-05, "loss": 2.9304, "step": 15753 }, { "epoch": 0.9779626295859457, "grad_norm": 0.5142483105851956, "learning_rate": 8.523382887771962e-05, "loss": 2.9473, "step": 15754 }, { "epoch": 0.9780247066857036, "grad_norm": 0.5145415569701425, "learning_rate": 8.523126630589395e-05, "loss": 2.9027, "step": 15755 }, { "epoch": 0.9780867837854615, "grad_norm": 0.5505263909017579, "learning_rate": 8.522870355025818e-05, "loss": 2.9245, "step": 15756 }, { "epoch": 0.9781488608852195, "grad_norm": 0.5045316792211896, "learning_rate": 8.522614061082574e-05, "loss": 2.9202, "step": 15757 }, { "epoch": 0.9782109379849774, "grad_norm": 0.5297906409476318, "learning_rate": 8.522357748760996e-05, "loss": 2.8891, "step": 15758 }, { "epoch": 0.9782730150847352, "grad_norm": 0.45984798915515046, "learning_rate": 8.522101418062422e-05, "loss": 2.9215, "step": 15759 }, { "epoch": 0.9783350921844931, "grad_norm": 0.5499699075009199, "learning_rate": 8.521845068988191e-05, "loss": 3.0156, "step": 15760 }, { "epoch": 0.978397169284251, "grad_norm": 0.47883690149918773, "learning_rate": 8.52158870153964e-05, "loss": 2.8824, "step": 15761 }, { "epoch": 0.9784592463840089, "grad_norm": 0.471861067938868, "learning_rate": 8.521332315718104e-05, "loss": 2.9298, "step": 15762 }, { "epoch": 0.9785213234837669, "grad_norm": 0.4923292743003942, "learning_rate": 8.521075911524924e-05, "loss": 2.882, "step": 15763 }, { "epoch": 0.9785834005835248, "grad_norm": 0.5299794205682812, "learning_rate": 8.520819488961435e-05, "loss": 2.9355, "step": 15764 }, { "epoch": 0.9786454776832826, "grad_norm": 0.49606326152199476, "learning_rate": 8.520563048028976e-05, "loss": 2.8895, "step": 15765 }, { "epoch": 0.9787075547830405, "grad_norm": 0.5295840958570528, "learning_rate": 8.520306588728886e-05, "loss": 2.9109, "step": 15766 }, { "epoch": 0.9787696318827984, "grad_norm": 0.49164869698590685, "learning_rate": 8.520050111062501e-05, "loss": 2.8656, "step": 15767 }, { "epoch": 0.9788317089825563, "grad_norm": 0.5289489616153793, "learning_rate": 8.519793615031161e-05, "loss": 2.9979, "step": 15768 }, { "epoch": 0.9788937860823143, "grad_norm": 0.49056482006724333, "learning_rate": 8.519537100636203e-05, "loss": 2.8702, "step": 15769 }, { "epoch": 0.9789558631820722, "grad_norm": 0.5468587327990059, "learning_rate": 8.519280567878964e-05, "loss": 2.8815, "step": 15770 }, { "epoch": 0.97901794028183, "grad_norm": 0.5668789173903532, "learning_rate": 8.519024016760786e-05, "loss": 2.9138, "step": 15771 }, { "epoch": 0.9790800173815879, "grad_norm": 0.48900859828227217, "learning_rate": 8.518767447283005e-05, "loss": 2.884, "step": 15772 }, { "epoch": 0.9791420944813458, "grad_norm": 0.5061334876217689, "learning_rate": 8.518510859446959e-05, "loss": 2.9685, "step": 15773 }, { "epoch": 0.9792041715811037, "grad_norm": 0.46832793625790986, "learning_rate": 8.518254253253987e-05, "loss": 3.0017, "step": 15774 }, { "epoch": 0.9792662486808617, "grad_norm": 0.4913186169524876, "learning_rate": 8.517997628705431e-05, "loss": 2.9783, "step": 15775 }, { "epoch": 0.9793283257806196, "grad_norm": 0.4760827948947085, "learning_rate": 8.517740985802626e-05, "loss": 2.764, "step": 15776 }, { "epoch": 0.9793904028803774, "grad_norm": 0.4471401888267394, "learning_rate": 8.517484324546913e-05, "loss": 2.919, "step": 15777 }, { "epoch": 0.9794524799801353, "grad_norm": 0.489500248399463, "learning_rate": 8.51722764493963e-05, "loss": 2.9907, "step": 15778 }, { "epoch": 0.9795145570798932, "grad_norm": 0.5027737570204351, "learning_rate": 8.516970946982116e-05, "loss": 2.891, "step": 15779 }, { "epoch": 0.9795766341796511, "grad_norm": 0.5051532957897416, "learning_rate": 8.516714230675711e-05, "loss": 2.9632, "step": 15780 }, { "epoch": 0.979638711279409, "grad_norm": 0.482460792608813, "learning_rate": 8.516457496021754e-05, "loss": 2.8811, "step": 15781 }, { "epoch": 0.979700788379167, "grad_norm": 0.5195002439037555, "learning_rate": 8.516200743021585e-05, "loss": 2.9311, "step": 15782 }, { "epoch": 0.9797628654789248, "grad_norm": 0.5569337039835696, "learning_rate": 8.515943971676543e-05, "loss": 2.9536, "step": 15783 }, { "epoch": 0.9798249425786827, "grad_norm": 0.559817301247593, "learning_rate": 8.515687181987969e-05, "loss": 2.9755, "step": 15784 }, { "epoch": 0.9798870196784406, "grad_norm": 0.5248042241141655, "learning_rate": 8.515430373957198e-05, "loss": 3.0676, "step": 15785 }, { "epoch": 0.9799490967781985, "grad_norm": 0.6206410515621616, "learning_rate": 8.515173547585577e-05, "loss": 2.9588, "step": 15786 }, { "epoch": 0.9800111738779564, "grad_norm": 0.5334610367662539, "learning_rate": 8.51491670287444e-05, "loss": 2.9567, "step": 15787 }, { "epoch": 0.9800732509777144, "grad_norm": 0.5188551497461267, "learning_rate": 8.514659839825131e-05, "loss": 2.9315, "step": 15788 }, { "epoch": 0.9801353280774722, "grad_norm": 0.5143868869321108, "learning_rate": 8.514402958438987e-05, "loss": 2.9715, "step": 15789 }, { "epoch": 0.9801974051772301, "grad_norm": 0.5697452389098455, "learning_rate": 8.514146058717352e-05, "loss": 2.97, "step": 15790 }, { "epoch": 0.980259482276988, "grad_norm": 0.5265914071408833, "learning_rate": 8.513889140661562e-05, "loss": 2.9298, "step": 15791 }, { "epoch": 0.9803215593767459, "grad_norm": 0.4963965655694862, "learning_rate": 8.51363220427296e-05, "loss": 2.889, "step": 15792 }, { "epoch": 0.9803836364765038, "grad_norm": 0.5995463235614018, "learning_rate": 8.513375249552885e-05, "loss": 2.9518, "step": 15793 }, { "epoch": 0.9804457135762618, "grad_norm": 0.4673333217945769, "learning_rate": 8.513118276502677e-05, "loss": 2.9782, "step": 15794 }, { "epoch": 0.9805077906760196, "grad_norm": 0.5662444837356236, "learning_rate": 8.512861285123682e-05, "loss": 2.9655, "step": 15795 }, { "epoch": 0.9805698677757775, "grad_norm": 0.5713847805597309, "learning_rate": 8.512604275417232e-05, "loss": 2.8876, "step": 15796 }, { "epoch": 0.9806319448755354, "grad_norm": 0.5314359059144461, "learning_rate": 8.512347247384675e-05, "loss": 2.9059, "step": 15797 }, { "epoch": 0.9806940219752933, "grad_norm": 0.6311666254183415, "learning_rate": 8.512090201027352e-05, "loss": 2.9392, "step": 15798 }, { "epoch": 0.9807560990750512, "grad_norm": 0.5583547867975686, "learning_rate": 8.5118331363466e-05, "loss": 2.9363, "step": 15799 }, { "epoch": 0.9808181761748092, "grad_norm": 0.5439376145645121, "learning_rate": 8.511576053343762e-05, "loss": 2.9752, "step": 15800 }, { "epoch": 0.980880253274567, "grad_norm": 0.5028708710582749, "learning_rate": 8.51131895202018e-05, "loss": 2.9237, "step": 15801 }, { "epoch": 0.9809423303743249, "grad_norm": 0.5060616755177887, "learning_rate": 8.511061832377193e-05, "loss": 2.8251, "step": 15802 }, { "epoch": 0.9810044074740828, "grad_norm": 0.49867772558653795, "learning_rate": 8.510804694416146e-05, "loss": 2.9804, "step": 15803 }, { "epoch": 0.9810664845738407, "grad_norm": 0.49511183034191897, "learning_rate": 8.510547538138376e-05, "loss": 2.9038, "step": 15804 }, { "epoch": 0.9811285616735986, "grad_norm": 0.5353537703158568, "learning_rate": 8.51029036354523e-05, "loss": 2.9321, "step": 15805 }, { "epoch": 0.9811906387733565, "grad_norm": 0.48062959165326835, "learning_rate": 8.510033170638046e-05, "loss": 2.9421, "step": 15806 }, { "epoch": 0.9812527158731144, "grad_norm": 0.5219475050487904, "learning_rate": 8.509775959418166e-05, "loss": 2.9527, "step": 15807 }, { "epoch": 0.9813147929728723, "grad_norm": 0.5236633439858058, "learning_rate": 8.509518729886934e-05, "loss": 2.8788, "step": 15808 }, { "epoch": 0.9813768700726302, "grad_norm": 0.5614148374941705, "learning_rate": 8.509261482045692e-05, "loss": 2.9291, "step": 15809 }, { "epoch": 0.9814389471723881, "grad_norm": 0.5267393533982255, "learning_rate": 8.509004215895779e-05, "loss": 2.9994, "step": 15810 }, { "epoch": 0.981501024272146, "grad_norm": 0.5100040389817314, "learning_rate": 8.508746931438539e-05, "loss": 2.9435, "step": 15811 }, { "epoch": 0.981563101371904, "grad_norm": 0.5613329489652642, "learning_rate": 8.508489628675316e-05, "loss": 2.8833, "step": 15812 }, { "epoch": 0.9816251784716618, "grad_norm": 0.44735360425514237, "learning_rate": 8.508232307607449e-05, "loss": 2.8574, "step": 15813 }, { "epoch": 0.9816872555714197, "grad_norm": 0.5060008672246927, "learning_rate": 8.507974968236284e-05, "loss": 2.9753, "step": 15814 }, { "epoch": 0.9817493326711776, "grad_norm": 0.4993604951310797, "learning_rate": 8.507717610563161e-05, "loss": 2.8747, "step": 15815 }, { "epoch": 0.9818114097709355, "grad_norm": 0.4846058418505506, "learning_rate": 8.507460234589424e-05, "loss": 2.9193, "step": 15816 }, { "epoch": 0.9818734868706934, "grad_norm": 0.45465302156375437, "learning_rate": 8.507202840316415e-05, "loss": 2.9405, "step": 15817 }, { "epoch": 0.9819355639704513, "grad_norm": 0.5332467477022367, "learning_rate": 8.506945427745479e-05, "loss": 2.92, "step": 15818 }, { "epoch": 0.9819976410702091, "grad_norm": 0.4409009860497322, "learning_rate": 8.506687996877955e-05, "loss": 2.873, "step": 15819 }, { "epoch": 0.9820597181699671, "grad_norm": 0.5338924549297721, "learning_rate": 8.50643054771519e-05, "loss": 2.9472, "step": 15820 }, { "epoch": 0.982121795269725, "grad_norm": 0.4536183729594313, "learning_rate": 8.506173080258525e-05, "loss": 2.9685, "step": 15821 }, { "epoch": 0.9821838723694829, "grad_norm": 0.4729122855561895, "learning_rate": 8.505915594509303e-05, "loss": 2.8919, "step": 15822 }, { "epoch": 0.9822459494692408, "grad_norm": 0.4816863595947292, "learning_rate": 8.505658090468871e-05, "loss": 2.9334, "step": 15823 }, { "epoch": 0.9823080265689987, "grad_norm": 0.42922208028583836, "learning_rate": 8.505400568138569e-05, "loss": 2.9173, "step": 15824 }, { "epoch": 0.9823701036687565, "grad_norm": 0.46133262057136776, "learning_rate": 8.50514302751974e-05, "loss": 2.9517, "step": 15825 }, { "epoch": 0.9824321807685145, "grad_norm": 0.4095915039867063, "learning_rate": 8.50488546861373e-05, "loss": 2.8128, "step": 15826 }, { "epoch": 0.9824942578682724, "grad_norm": 0.45088986256858143, "learning_rate": 8.50462789142188e-05, "loss": 2.8471, "step": 15827 }, { "epoch": 0.9825563349680303, "grad_norm": 0.5203583236082123, "learning_rate": 8.504370295945537e-05, "loss": 2.9141, "step": 15828 }, { "epoch": 0.9826184120677882, "grad_norm": 0.508211800503406, "learning_rate": 8.504112682186043e-05, "loss": 3.0369, "step": 15829 }, { "epoch": 0.9826804891675461, "grad_norm": 0.5144940602577862, "learning_rate": 8.503855050144743e-05, "loss": 2.8402, "step": 15830 }, { "epoch": 0.9827425662673039, "grad_norm": 0.5320919881648014, "learning_rate": 8.50359739982298e-05, "loss": 2.9108, "step": 15831 }, { "epoch": 0.9828046433670619, "grad_norm": 0.47690523049263267, "learning_rate": 8.503339731222098e-05, "loss": 2.9478, "step": 15832 }, { "epoch": 0.9828667204668198, "grad_norm": 0.5107119024280193, "learning_rate": 8.503082044343443e-05, "loss": 2.8121, "step": 15833 }, { "epoch": 0.9829287975665777, "grad_norm": 0.4759526024048473, "learning_rate": 8.502824339188359e-05, "loss": 2.8479, "step": 15834 }, { "epoch": 0.9829908746663356, "grad_norm": 0.51465964259883, "learning_rate": 8.502566615758189e-05, "loss": 2.9436, "step": 15835 }, { "epoch": 0.9830529517660935, "grad_norm": 0.5087982294058582, "learning_rate": 8.50230887405428e-05, "loss": 2.9253, "step": 15836 }, { "epoch": 0.9831150288658513, "grad_norm": 0.4562113000839077, "learning_rate": 8.502051114077976e-05, "loss": 2.912, "step": 15837 }, { "epoch": 0.9831771059656093, "grad_norm": 0.5087124699145728, "learning_rate": 8.501793335830619e-05, "loss": 2.9493, "step": 15838 }, { "epoch": 0.9832391830653672, "grad_norm": 0.44501528944045243, "learning_rate": 8.501535539313556e-05, "loss": 2.9088, "step": 15839 }, { "epoch": 0.9833012601651251, "grad_norm": 0.49311908479050964, "learning_rate": 8.501277724528134e-05, "loss": 2.9734, "step": 15840 }, { "epoch": 0.983363337264883, "grad_norm": 0.46921244698024767, "learning_rate": 8.501019891475695e-05, "loss": 2.8769, "step": 15841 }, { "epoch": 0.9834254143646409, "grad_norm": 0.48129409365102205, "learning_rate": 8.500762040157586e-05, "loss": 2.8788, "step": 15842 }, { "epoch": 0.9834874914643987, "grad_norm": 0.4889218515093348, "learning_rate": 8.500504170575153e-05, "loss": 2.9495, "step": 15843 }, { "epoch": 0.9835495685641567, "grad_norm": 0.48809478251062677, "learning_rate": 8.500246282729739e-05, "loss": 2.9471, "step": 15844 }, { "epoch": 0.9836116456639146, "grad_norm": 0.5078377057202242, "learning_rate": 8.499988376622687e-05, "loss": 3.0348, "step": 15845 }, { "epoch": 0.9836737227636725, "grad_norm": 0.4663485524897106, "learning_rate": 8.49973045225535e-05, "loss": 2.904, "step": 15846 }, { "epoch": 0.9837357998634304, "grad_norm": 0.5114893799608741, "learning_rate": 8.499472509629067e-05, "loss": 2.9475, "step": 15847 }, { "epoch": 0.9837978769631883, "grad_norm": 0.5212553285651147, "learning_rate": 8.499214548745188e-05, "loss": 2.8991, "step": 15848 }, { "epoch": 0.9838599540629461, "grad_norm": 0.4820609653902993, "learning_rate": 8.498956569605057e-05, "loss": 2.9347, "step": 15849 }, { "epoch": 0.983922031162704, "grad_norm": 0.47719111960186816, "learning_rate": 8.498698572210019e-05, "loss": 2.8571, "step": 15850 }, { "epoch": 0.983984108262462, "grad_norm": 0.4636776667476625, "learning_rate": 8.498440556561424e-05, "loss": 2.8746, "step": 15851 }, { "epoch": 0.9840461853622199, "grad_norm": 0.47881480614998484, "learning_rate": 8.498182522660613e-05, "loss": 2.8748, "step": 15852 }, { "epoch": 0.9841082624619778, "grad_norm": 0.500305350763063, "learning_rate": 8.497924470508935e-05, "loss": 2.9096, "step": 15853 }, { "epoch": 0.9841703395617357, "grad_norm": 0.5394129788585859, "learning_rate": 8.497666400107736e-05, "loss": 2.8513, "step": 15854 }, { "epoch": 0.9842324166614935, "grad_norm": 0.474609375, "learning_rate": 8.497408311458362e-05, "loss": 2.9368, "step": 15855 }, { "epoch": 0.9842944937612514, "grad_norm": 0.5629315310458546, "learning_rate": 8.497150204562161e-05, "loss": 2.8997, "step": 15856 }, { "epoch": 0.9843565708610094, "grad_norm": 0.5838553244596011, "learning_rate": 8.496892079420478e-05, "loss": 2.9395, "step": 15857 }, { "epoch": 0.9844186479607673, "grad_norm": 0.495248073756108, "learning_rate": 8.496633936034659e-05, "loss": 2.9795, "step": 15858 }, { "epoch": 0.9844807250605252, "grad_norm": 0.5596994780212416, "learning_rate": 8.496375774406052e-05, "loss": 2.9186, "step": 15859 }, { "epoch": 0.9845428021602831, "grad_norm": 0.5188466200413436, "learning_rate": 8.496117594536006e-05, "loss": 2.8338, "step": 15860 }, { "epoch": 0.9846048792600409, "grad_norm": 0.4883493147140152, "learning_rate": 8.495859396425863e-05, "loss": 2.9919, "step": 15861 }, { "epoch": 0.9846669563597988, "grad_norm": 0.561536015908711, "learning_rate": 8.495601180076976e-05, "loss": 2.9867, "step": 15862 }, { "epoch": 0.9847290334595568, "grad_norm": 0.5835145827680037, "learning_rate": 8.495342945490686e-05, "loss": 2.9344, "step": 15863 }, { "epoch": 0.9847911105593147, "grad_norm": 0.553769170803843, "learning_rate": 8.495084692668344e-05, "loss": 2.9117, "step": 15864 }, { "epoch": 0.9848531876590726, "grad_norm": 0.47569424063808907, "learning_rate": 8.494826421611298e-05, "loss": 2.988, "step": 15865 }, { "epoch": 0.9849152647588305, "grad_norm": 0.5793378325457004, "learning_rate": 8.494568132320893e-05, "loss": 2.842, "step": 15866 }, { "epoch": 0.9849773418585883, "grad_norm": 0.4940260954201947, "learning_rate": 8.494309824798477e-05, "loss": 2.8549, "step": 15867 }, { "epoch": 0.9850394189583462, "grad_norm": 0.522537938798539, "learning_rate": 8.4940514990454e-05, "loss": 2.911, "step": 15868 }, { "epoch": 0.9851014960581042, "grad_norm": 0.46748808437604544, "learning_rate": 8.493793155063008e-05, "loss": 2.9313, "step": 15869 }, { "epoch": 0.9851635731578621, "grad_norm": 0.5333270533261509, "learning_rate": 8.493534792852648e-05, "loss": 2.9704, "step": 15870 }, { "epoch": 0.98522565025762, "grad_norm": 0.4545405245372706, "learning_rate": 8.49327641241567e-05, "loss": 2.9467, "step": 15871 }, { "epoch": 0.9852877273573779, "grad_norm": 0.4786269389322718, "learning_rate": 8.49301801375342e-05, "loss": 2.9988, "step": 15872 }, { "epoch": 0.9853498044571357, "grad_norm": 0.5220389399894082, "learning_rate": 8.492759596867248e-05, "loss": 2.9604, "step": 15873 }, { "epoch": 0.9854118815568936, "grad_norm": 0.4838703818695818, "learning_rate": 8.4925011617585e-05, "loss": 2.9407, "step": 15874 }, { "epoch": 0.9854739586566515, "grad_norm": 0.5014743822596508, "learning_rate": 8.492242708428527e-05, "loss": 3.0233, "step": 15875 }, { "epoch": 0.9855360357564095, "grad_norm": 0.4452157668405156, "learning_rate": 8.491984236878675e-05, "loss": 2.9448, "step": 15876 }, { "epoch": 0.9855981128561674, "grad_norm": 0.5565627729617633, "learning_rate": 8.491725747110294e-05, "loss": 2.9659, "step": 15877 }, { "epoch": 0.9856601899559253, "grad_norm": 0.48175254134085405, "learning_rate": 8.491467239124733e-05, "loss": 2.9739, "step": 15878 }, { "epoch": 0.9857222670556831, "grad_norm": 0.49539104765456043, "learning_rate": 8.491208712923337e-05, "loss": 2.942, "step": 15879 }, { "epoch": 0.985784344155441, "grad_norm": 0.5297020069037486, "learning_rate": 8.49095016850746e-05, "loss": 2.8271, "step": 15880 }, { "epoch": 0.985846421255199, "grad_norm": 0.5118753906919224, "learning_rate": 8.490691605878447e-05, "loss": 2.8998, "step": 15881 }, { "epoch": 0.9859084983549569, "grad_norm": 0.5009422959772171, "learning_rate": 8.49043302503765e-05, "loss": 2.9487, "step": 15882 }, { "epoch": 0.9859705754547148, "grad_norm": 0.468106750366835, "learning_rate": 8.490174425986417e-05, "loss": 2.9165, "step": 15883 }, { "epoch": 0.9860326525544727, "grad_norm": 0.5159565119992177, "learning_rate": 8.489915808726095e-05, "loss": 2.9355, "step": 15884 }, { "epoch": 0.9860947296542305, "grad_norm": 0.5026576519378284, "learning_rate": 8.489657173258037e-05, "loss": 2.8707, "step": 15885 }, { "epoch": 0.9861568067539884, "grad_norm": 0.5161108848783209, "learning_rate": 8.489398519583588e-05, "loss": 2.938, "step": 15886 }, { "epoch": 0.9862188838537463, "grad_norm": 0.4841588214719063, "learning_rate": 8.489139847704102e-05, "loss": 2.9538, "step": 15887 }, { "epoch": 0.9862809609535043, "grad_norm": 0.5315720479602203, "learning_rate": 8.488881157620925e-05, "loss": 2.957, "step": 15888 }, { "epoch": 0.9863430380532622, "grad_norm": 0.5198707045230674, "learning_rate": 8.488622449335408e-05, "loss": 2.8697, "step": 15889 }, { "epoch": 0.9864051151530201, "grad_norm": 0.43991241190180536, "learning_rate": 8.488363722848902e-05, "loss": 3.053, "step": 15890 }, { "epoch": 0.9864671922527779, "grad_norm": 0.5906476990309018, "learning_rate": 8.488104978162754e-05, "loss": 2.8439, "step": 15891 }, { "epoch": 0.9865292693525358, "grad_norm": 0.48480537737234125, "learning_rate": 8.487846215278316e-05, "loss": 2.971, "step": 15892 }, { "epoch": 0.9865913464522937, "grad_norm": 0.5562040256403257, "learning_rate": 8.48758743419694e-05, "loss": 2.8919, "step": 15893 }, { "epoch": 0.9866534235520517, "grad_norm": 0.4693640501672787, "learning_rate": 8.487328634919971e-05, "loss": 2.9194, "step": 15894 }, { "epoch": 0.9867155006518096, "grad_norm": 0.46620761916502246, "learning_rate": 8.487069817448763e-05, "loss": 2.9741, "step": 15895 }, { "epoch": 0.9867775777515675, "grad_norm": 0.4926092369424724, "learning_rate": 8.486810981784664e-05, "loss": 2.9562, "step": 15896 }, { "epoch": 0.9868396548513253, "grad_norm": 0.4672317235167432, "learning_rate": 8.486552127929026e-05, "loss": 2.9799, "step": 15897 }, { "epoch": 0.9869017319510832, "grad_norm": 0.43256407125336743, "learning_rate": 8.4862932558832e-05, "loss": 2.9287, "step": 15898 }, { "epoch": 0.9869638090508411, "grad_norm": 0.4889377910306431, "learning_rate": 8.486034365648536e-05, "loss": 2.9395, "step": 15899 }, { "epoch": 0.987025886150599, "grad_norm": 0.47612082228826974, "learning_rate": 8.485775457226383e-05, "loss": 2.8278, "step": 15900 }, { "epoch": 0.987087963250357, "grad_norm": 0.46587533046095553, "learning_rate": 8.485516530618095e-05, "loss": 2.9111, "step": 15901 }, { "epoch": 0.9871500403501149, "grad_norm": 0.5237240931947795, "learning_rate": 8.485257585825021e-05, "loss": 2.9686, "step": 15902 }, { "epoch": 0.9872121174498727, "grad_norm": 0.4895765138381113, "learning_rate": 8.484998622848511e-05, "loss": 2.9514, "step": 15903 }, { "epoch": 0.9872741945496306, "grad_norm": 0.5092901592160028, "learning_rate": 8.484739641689918e-05, "loss": 2.9594, "step": 15904 }, { "epoch": 0.9873362716493885, "grad_norm": 0.5365210067532874, "learning_rate": 8.484480642350592e-05, "loss": 2.9597, "step": 15905 }, { "epoch": 0.9873983487491464, "grad_norm": 0.5680762041075548, "learning_rate": 8.484221624831885e-05, "loss": 2.845, "step": 15906 }, { "epoch": 0.9874604258489044, "grad_norm": 0.5408505900715243, "learning_rate": 8.483962589135146e-05, "loss": 2.9115, "step": 15907 }, { "epoch": 0.9875225029486623, "grad_norm": 0.5546899446245508, "learning_rate": 8.48370353526173e-05, "loss": 2.8816, "step": 15908 }, { "epoch": 0.9875845800484201, "grad_norm": 0.5253836478865547, "learning_rate": 8.483444463212988e-05, "loss": 2.8975, "step": 15909 }, { "epoch": 0.987646657148178, "grad_norm": 0.5972831159666412, "learning_rate": 8.483185372990269e-05, "loss": 2.9465, "step": 15910 }, { "epoch": 0.9877087342479359, "grad_norm": 0.4842265270700462, "learning_rate": 8.482926264594927e-05, "loss": 2.8777, "step": 15911 }, { "epoch": 0.9877708113476938, "grad_norm": 0.5598273893409675, "learning_rate": 8.482667138028313e-05, "loss": 3.0062, "step": 15912 }, { "epoch": 0.9878328884474518, "grad_norm": 0.5235892403276626, "learning_rate": 8.482407993291779e-05, "loss": 2.9439, "step": 15913 }, { "epoch": 0.9878949655472097, "grad_norm": 0.5524961409714509, "learning_rate": 8.482148830386676e-05, "loss": 2.9689, "step": 15914 }, { "epoch": 0.9879570426469675, "grad_norm": 0.5331993801558341, "learning_rate": 8.481889649314358e-05, "loss": 2.7913, "step": 15915 }, { "epoch": 0.9880191197467254, "grad_norm": 0.49727200835896207, "learning_rate": 8.481630450076177e-05, "loss": 2.907, "step": 15916 }, { "epoch": 0.9880811968464833, "grad_norm": 0.5248014983068571, "learning_rate": 8.481371232673485e-05, "loss": 2.946, "step": 15917 }, { "epoch": 0.9881432739462412, "grad_norm": 0.5406506681137028, "learning_rate": 8.481111997107634e-05, "loss": 2.9978, "step": 15918 }, { "epoch": 0.9882053510459992, "grad_norm": 0.46491479731412594, "learning_rate": 8.480852743379977e-05, "loss": 2.9654, "step": 15919 }, { "epoch": 0.9882674281457571, "grad_norm": 0.48734408660808554, "learning_rate": 8.480593471491865e-05, "loss": 2.8516, "step": 15920 }, { "epoch": 0.9883295052455149, "grad_norm": 0.45889989197131853, "learning_rate": 8.480334181444653e-05, "loss": 2.9247, "step": 15921 }, { "epoch": 0.9883915823452728, "grad_norm": 0.4991711959516626, "learning_rate": 8.480074873239691e-05, "loss": 2.8782, "step": 15922 }, { "epoch": 0.9884536594450307, "grad_norm": 0.5160836001029553, "learning_rate": 8.479815546878335e-05, "loss": 3.0172, "step": 15923 }, { "epoch": 0.9885157365447886, "grad_norm": 0.5147905244072294, "learning_rate": 8.479556202361938e-05, "loss": 2.9122, "step": 15924 }, { "epoch": 0.9885778136445466, "grad_norm": 0.5339682821487304, "learning_rate": 8.479296839691849e-05, "loss": 2.8884, "step": 15925 }, { "epoch": 0.9886398907443045, "grad_norm": 0.5389130150117918, "learning_rate": 8.479037458869426e-05, "loss": 2.9439, "step": 15926 }, { "epoch": 0.9887019678440623, "grad_norm": 0.5098792347789441, "learning_rate": 8.478778059896017e-05, "loss": 2.9931, "step": 15927 }, { "epoch": 0.9887640449438202, "grad_norm": 0.5337916050671385, "learning_rate": 8.478518642772982e-05, "loss": 2.9036, "step": 15928 }, { "epoch": 0.9888261220435781, "grad_norm": 0.5083275457219492, "learning_rate": 8.478259207501668e-05, "loss": 3.0114, "step": 15929 }, { "epoch": 0.988888199143336, "grad_norm": 0.589258143267161, "learning_rate": 8.477999754083433e-05, "loss": 3.0167, "step": 15930 }, { "epoch": 0.988950276243094, "grad_norm": 0.5036346235740512, "learning_rate": 8.477740282519626e-05, "loss": 2.8751, "step": 15931 }, { "epoch": 0.9890123533428519, "grad_norm": 0.5379281756024066, "learning_rate": 8.477480792811608e-05, "loss": 2.9407, "step": 15932 }, { "epoch": 0.9890744304426097, "grad_norm": 0.5039249091427106, "learning_rate": 8.477221284960726e-05, "loss": 2.9196, "step": 15933 }, { "epoch": 0.9891365075423676, "grad_norm": 0.5017894374077062, "learning_rate": 8.476961758968336e-05, "loss": 3.012, "step": 15934 }, { "epoch": 0.9891985846421255, "grad_norm": 0.5644230818095486, "learning_rate": 8.476702214835793e-05, "loss": 2.9193, "step": 15935 }, { "epoch": 0.9892606617418834, "grad_norm": 0.552729306686403, "learning_rate": 8.476442652564452e-05, "loss": 2.8293, "step": 15936 }, { "epoch": 0.9893227388416413, "grad_norm": 0.47785103247725463, "learning_rate": 8.476183072155663e-05, "loss": 2.9108, "step": 15937 }, { "epoch": 0.9893848159413993, "grad_norm": 0.5540690937284594, "learning_rate": 8.475923473610785e-05, "loss": 2.924, "step": 15938 }, { "epoch": 0.9894468930411571, "grad_norm": 0.45998665966509805, "learning_rate": 8.475663856931171e-05, "loss": 2.9443, "step": 15939 }, { "epoch": 0.989508970140915, "grad_norm": 0.5190310544129538, "learning_rate": 8.475404222118173e-05, "loss": 2.8674, "step": 15940 }, { "epoch": 0.9895710472406729, "grad_norm": 0.4510452125738838, "learning_rate": 8.475144569173148e-05, "loss": 2.9158, "step": 15941 }, { "epoch": 0.9896331243404308, "grad_norm": 0.45287028092113846, "learning_rate": 8.474884898097452e-05, "loss": 2.9219, "step": 15942 }, { "epoch": 0.9896952014401887, "grad_norm": 0.49420776212660483, "learning_rate": 8.474625208892437e-05, "loss": 2.859, "step": 15943 }, { "epoch": 0.9897572785399467, "grad_norm": 0.47195404476991143, "learning_rate": 8.474365501559456e-05, "loss": 2.9711, "step": 15944 }, { "epoch": 0.9898193556397045, "grad_norm": 0.5268402238819804, "learning_rate": 8.47410577609987e-05, "loss": 2.9078, "step": 15945 }, { "epoch": 0.9898814327394624, "grad_norm": 0.5107005815353653, "learning_rate": 8.47384603251503e-05, "loss": 2.9417, "step": 15946 }, { "epoch": 0.9899435098392203, "grad_norm": 0.5368751153734858, "learning_rate": 8.473586270806291e-05, "loss": 2.8976, "step": 15947 }, { "epoch": 0.9900055869389782, "grad_norm": 0.5888125772284823, "learning_rate": 8.47332649097501e-05, "loss": 3.0279, "step": 15948 }, { "epoch": 0.9900676640387361, "grad_norm": 0.554422368904115, "learning_rate": 8.473066693022543e-05, "loss": 2.8696, "step": 15949 }, { "epoch": 0.9901297411384941, "grad_norm": 0.4913952066608865, "learning_rate": 8.472806876950241e-05, "loss": 2.8627, "step": 15950 }, { "epoch": 0.9901918182382519, "grad_norm": 0.48214924015886057, "learning_rate": 8.472547042759464e-05, "loss": 2.9518, "step": 15951 }, { "epoch": 0.9902538953380098, "grad_norm": 0.5033288057178371, "learning_rate": 8.472287190451566e-05, "loss": 2.9987, "step": 15952 }, { "epoch": 0.9903159724377677, "grad_norm": 0.5083453097522069, "learning_rate": 8.472027320027904e-05, "loss": 3.0157, "step": 15953 }, { "epoch": 0.9903780495375256, "grad_norm": 0.5593219358965661, "learning_rate": 8.47176743148983e-05, "loss": 2.8695, "step": 15954 }, { "epoch": 0.9904401266372835, "grad_norm": 0.4755843865527922, "learning_rate": 8.471507524838705e-05, "loss": 2.8833, "step": 15955 }, { "epoch": 0.9905022037370415, "grad_norm": 0.5379201145440824, "learning_rate": 8.47124760007588e-05, "loss": 2.9333, "step": 15956 }, { "epoch": 0.9905642808367993, "grad_norm": 0.5988277099369052, "learning_rate": 8.470987657202715e-05, "loss": 2.9901, "step": 15957 }, { "epoch": 0.9906263579365572, "grad_norm": 0.49540889957147527, "learning_rate": 8.470727696220567e-05, "loss": 2.8693, "step": 15958 }, { "epoch": 0.9906884350363151, "grad_norm": 0.5470187134601476, "learning_rate": 8.470467717130787e-05, "loss": 2.9422, "step": 15959 }, { "epoch": 0.990750512136073, "grad_norm": 0.5876670995439496, "learning_rate": 8.470207719934735e-05, "loss": 2.9703, "step": 15960 }, { "epoch": 0.9908125892358309, "grad_norm": 0.5419337885556534, "learning_rate": 8.469947704633768e-05, "loss": 2.8784, "step": 15961 }, { "epoch": 0.9908746663355888, "grad_norm": 0.5681454495147803, "learning_rate": 8.469687671229238e-05, "loss": 2.9498, "step": 15962 }, { "epoch": 0.9909367434353467, "grad_norm": 0.6151778179996946, "learning_rate": 8.469427619722508e-05, "loss": 3.0194, "step": 15963 }, { "epoch": 0.9909988205351046, "grad_norm": 0.4694677579735081, "learning_rate": 8.469167550114933e-05, "loss": 2.8355, "step": 15964 }, { "epoch": 0.9910608976348625, "grad_norm": 0.5966107817225816, "learning_rate": 8.468907462407867e-05, "loss": 2.9658, "step": 15965 }, { "epoch": 0.9911229747346204, "grad_norm": 0.5290465942082831, "learning_rate": 8.468647356602669e-05, "loss": 2.9571, "step": 15966 }, { "epoch": 0.9911850518343783, "grad_norm": 0.45615131081603205, "learning_rate": 8.468387232700696e-05, "loss": 2.8645, "step": 15967 }, { "epoch": 0.9912471289341362, "grad_norm": 0.5191830493593425, "learning_rate": 8.468127090703304e-05, "loss": 3.0019, "step": 15968 }, { "epoch": 0.991309206033894, "grad_norm": 0.47196340616966004, "learning_rate": 8.467866930611852e-05, "loss": 3.0258, "step": 15969 }, { "epoch": 0.991371283133652, "grad_norm": 0.5000888626287933, "learning_rate": 8.467606752427696e-05, "loss": 2.8385, "step": 15970 }, { "epoch": 0.9914333602334099, "grad_norm": 0.5095788313832371, "learning_rate": 8.467346556152193e-05, "loss": 2.9544, "step": 15971 }, { "epoch": 0.9914954373331678, "grad_norm": 0.4690011146578933, "learning_rate": 8.467086341786703e-05, "loss": 2.8433, "step": 15972 }, { "epoch": 0.9915575144329257, "grad_norm": 0.4739159907475984, "learning_rate": 8.46682610933258e-05, "loss": 2.934, "step": 15973 }, { "epoch": 0.9916195915326836, "grad_norm": 0.5269488233386465, "learning_rate": 8.466565858791186e-05, "loss": 2.9719, "step": 15974 }, { "epoch": 0.9916816686324414, "grad_norm": 0.5249695916679171, "learning_rate": 8.466305590163875e-05, "loss": 2.9896, "step": 15975 }, { "epoch": 0.9917437457321994, "grad_norm": 0.4916204789121352, "learning_rate": 8.466045303452006e-05, "loss": 2.9768, "step": 15976 }, { "epoch": 0.9918058228319573, "grad_norm": 0.4782032316694186, "learning_rate": 8.465784998656936e-05, "loss": 2.9785, "step": 15977 }, { "epoch": 0.9918678999317152, "grad_norm": 0.5242581837315059, "learning_rate": 8.465524675780025e-05, "loss": 2.8656, "step": 15978 }, { "epoch": 0.9919299770314731, "grad_norm": 0.49874216054429515, "learning_rate": 8.465264334822632e-05, "loss": 2.9765, "step": 15979 }, { "epoch": 0.991992054131231, "grad_norm": 0.5482969870256165, "learning_rate": 8.465003975786113e-05, "loss": 2.9675, "step": 15980 }, { "epoch": 0.9920541312309888, "grad_norm": 0.4801075981410923, "learning_rate": 8.464743598671828e-05, "loss": 2.9267, "step": 15981 }, { "epoch": 0.9921162083307468, "grad_norm": 0.5208544122880908, "learning_rate": 8.464483203481133e-05, "loss": 2.9839, "step": 15982 }, { "epoch": 0.9921782854305047, "grad_norm": 0.5312185839171346, "learning_rate": 8.464222790215388e-05, "loss": 2.9787, "step": 15983 }, { "epoch": 0.9922403625302626, "grad_norm": 0.5409864837840724, "learning_rate": 8.463962358875953e-05, "loss": 2.9644, "step": 15984 }, { "epoch": 0.9923024396300205, "grad_norm": 0.44257766440789276, "learning_rate": 8.463701909464185e-05, "loss": 2.9533, "step": 15985 }, { "epoch": 0.9923645167297784, "grad_norm": 0.5801113902613942, "learning_rate": 8.463441441981444e-05, "loss": 2.9191, "step": 15986 }, { "epoch": 0.9924265938295362, "grad_norm": 0.5349572361584429, "learning_rate": 8.463180956429086e-05, "loss": 2.9353, "step": 15987 }, { "epoch": 0.9924886709292942, "grad_norm": 0.5144199682165097, "learning_rate": 8.462920452808474e-05, "loss": 2.9923, "step": 15988 }, { "epoch": 0.9925507480290521, "grad_norm": 0.5034422401475008, "learning_rate": 8.462659931120965e-05, "loss": 2.9075, "step": 15989 }, { "epoch": 0.99261282512881, "grad_norm": 0.5889188070114852, "learning_rate": 8.462399391367919e-05, "loss": 2.9742, "step": 15990 }, { "epoch": 0.9926749022285679, "grad_norm": 0.580156314602249, "learning_rate": 8.462138833550694e-05, "loss": 3.0323, "step": 15991 }, { "epoch": 0.9927369793283258, "grad_norm": 0.580280358611246, "learning_rate": 8.46187825767065e-05, "loss": 2.9585, "step": 15992 }, { "epoch": 0.9927990564280836, "grad_norm": 0.5736026183180325, "learning_rate": 8.46161766372915e-05, "loss": 2.8474, "step": 15993 }, { "epoch": 0.9928611335278416, "grad_norm": 0.5423681106745253, "learning_rate": 8.461357051727546e-05, "loss": 2.9478, "step": 15994 }, { "epoch": 0.9929232106275995, "grad_norm": 0.5612535441739128, "learning_rate": 8.461096421667203e-05, "loss": 2.92, "step": 15995 }, { "epoch": 0.9929852877273574, "grad_norm": 0.5598990920807388, "learning_rate": 8.46083577354948e-05, "loss": 2.9336, "step": 15996 }, { "epoch": 0.9930473648271153, "grad_norm": 0.5069262473732041, "learning_rate": 8.460575107375738e-05, "loss": 2.9874, "step": 15997 }, { "epoch": 0.9931094419268732, "grad_norm": 0.5112232109363484, "learning_rate": 8.460314423147334e-05, "loss": 2.8614, "step": 15998 }, { "epoch": 0.993171519026631, "grad_norm": 0.4912425006410513, "learning_rate": 8.46005372086563e-05, "loss": 2.9895, "step": 15999 }, { "epoch": 0.993233596126389, "grad_norm": 0.5118475599002165, "learning_rate": 8.459793000531986e-05, "loss": 2.9344, "step": 16000 }, { "epoch": 0.9932956732261469, "grad_norm": 0.5049953962342817, "learning_rate": 8.459532262147763e-05, "loss": 2.9434, "step": 16001 }, { "epoch": 0.9933577503259048, "grad_norm": 0.4792529906038791, "learning_rate": 8.45927150571432e-05, "loss": 2.8575, "step": 16002 }, { "epoch": 0.9934198274256627, "grad_norm": 0.4400338807120506, "learning_rate": 8.459010731233017e-05, "loss": 2.8806, "step": 16003 }, { "epoch": 0.9934819045254206, "grad_norm": 0.5189198790302193, "learning_rate": 8.458749938705215e-05, "loss": 2.9214, "step": 16004 }, { "epoch": 0.9935439816251784, "grad_norm": 0.46079242168865625, "learning_rate": 8.458489128132275e-05, "loss": 2.9033, "step": 16005 }, { "epoch": 0.9936060587249363, "grad_norm": 0.5020000272576071, "learning_rate": 8.458228299515559e-05, "loss": 2.9519, "step": 16006 }, { "epoch": 0.9936681358246943, "grad_norm": 0.49328454689275036, "learning_rate": 8.457967452856426e-05, "loss": 2.8632, "step": 16007 }, { "epoch": 0.9937302129244522, "grad_norm": 0.5241975248448643, "learning_rate": 8.457706588156236e-05, "loss": 2.9711, "step": 16008 }, { "epoch": 0.9937922900242101, "grad_norm": 0.47943285790619006, "learning_rate": 8.457445705416351e-05, "loss": 2.9068, "step": 16009 }, { "epoch": 0.993854367123968, "grad_norm": 0.477428025518322, "learning_rate": 8.457184804638134e-05, "loss": 2.9805, "step": 16010 }, { "epoch": 0.9939164442237258, "grad_norm": 0.5039818285663537, "learning_rate": 8.456923885822944e-05, "loss": 2.9442, "step": 16011 }, { "epoch": 0.9939785213234837, "grad_norm": 0.5402766389788491, "learning_rate": 8.456662948972143e-05, "loss": 2.9428, "step": 16012 }, { "epoch": 0.9940405984232417, "grad_norm": 0.4406276858362897, "learning_rate": 8.456401994087093e-05, "loss": 2.8872, "step": 16013 }, { "epoch": 0.9941026755229996, "grad_norm": 0.46925074533983563, "learning_rate": 8.456141021169154e-05, "loss": 2.9628, "step": 16014 }, { "epoch": 0.9941647526227575, "grad_norm": 0.48282296190994584, "learning_rate": 8.455880030219688e-05, "loss": 2.9202, "step": 16015 }, { "epoch": 0.9942268297225154, "grad_norm": 0.5105802377717271, "learning_rate": 8.455619021240058e-05, "loss": 2.9012, "step": 16016 }, { "epoch": 0.9942889068222732, "grad_norm": 0.47341334970649945, "learning_rate": 8.455357994231624e-05, "loss": 3.0012, "step": 16017 }, { "epoch": 0.9943509839220311, "grad_norm": 0.5233306206290691, "learning_rate": 8.45509694919575e-05, "loss": 2.9242, "step": 16018 }, { "epoch": 0.9944130610217891, "grad_norm": 0.4780703750272933, "learning_rate": 8.454835886133793e-05, "loss": 2.8408, "step": 16019 }, { "epoch": 0.994475138121547, "grad_norm": 0.49253616396292915, "learning_rate": 8.454574805047121e-05, "loss": 2.8778, "step": 16020 }, { "epoch": 0.9945372152213049, "grad_norm": 0.5350712061225968, "learning_rate": 8.454313705937094e-05, "loss": 2.9187, "step": 16021 }, { "epoch": 0.9945992923210628, "grad_norm": 0.5160481421760328, "learning_rate": 8.454052588805073e-05, "loss": 2.9752, "step": 16022 }, { "epoch": 0.9946613694208206, "grad_norm": 0.46303427354538984, "learning_rate": 8.45379145365242e-05, "loss": 2.8488, "step": 16023 }, { "epoch": 0.9947234465205785, "grad_norm": 0.5101807172203593, "learning_rate": 8.4535303004805e-05, "loss": 2.9955, "step": 16024 }, { "epoch": 0.9947855236203365, "grad_norm": 0.4666392167851039, "learning_rate": 8.453269129290674e-05, "loss": 2.8787, "step": 16025 }, { "epoch": 0.9948476007200944, "grad_norm": 0.47853100810334137, "learning_rate": 8.453007940084303e-05, "loss": 2.9129, "step": 16026 }, { "epoch": 0.9949096778198523, "grad_norm": 0.48078498667795044, "learning_rate": 8.452746732862754e-05, "loss": 2.9849, "step": 16027 }, { "epoch": 0.9949717549196102, "grad_norm": 0.4468740556613574, "learning_rate": 8.452485507627386e-05, "loss": 2.8944, "step": 16028 }, { "epoch": 0.995033832019368, "grad_norm": 0.4644117030480785, "learning_rate": 8.452224264379563e-05, "loss": 2.9075, "step": 16029 }, { "epoch": 0.9950959091191259, "grad_norm": 0.5102891827668191, "learning_rate": 8.451963003120647e-05, "loss": 2.9524, "step": 16030 }, { "epoch": 0.9951579862188838, "grad_norm": 0.4667476644074451, "learning_rate": 8.451701723852002e-05, "loss": 2.9588, "step": 16031 }, { "epoch": 0.9952200633186418, "grad_norm": 0.5034651193081201, "learning_rate": 8.451440426574993e-05, "loss": 2.9086, "step": 16032 }, { "epoch": 0.9952821404183997, "grad_norm": 0.4553200074822523, "learning_rate": 8.45117911129098e-05, "loss": 2.9217, "step": 16033 }, { "epoch": 0.9953442175181576, "grad_norm": 0.47353586994163777, "learning_rate": 8.450917778001329e-05, "loss": 2.9353, "step": 16034 }, { "epoch": 0.9954062946179154, "grad_norm": 0.4417346644090087, "learning_rate": 8.450656426707399e-05, "loss": 2.8414, "step": 16035 }, { "epoch": 0.9954683717176733, "grad_norm": 0.46378781894493665, "learning_rate": 8.450395057410561e-05, "loss": 2.8679, "step": 16036 }, { "epoch": 0.9955304488174312, "grad_norm": 0.4652174721569152, "learning_rate": 8.450133670112172e-05, "loss": 2.8204, "step": 16037 }, { "epoch": 0.9955925259171892, "grad_norm": 0.4558239829111539, "learning_rate": 8.449872264813598e-05, "loss": 2.8398, "step": 16038 }, { "epoch": 0.9956546030169471, "grad_norm": 0.42349502552786017, "learning_rate": 8.449610841516205e-05, "loss": 2.9168, "step": 16039 }, { "epoch": 0.995716680116705, "grad_norm": 0.49518736693671644, "learning_rate": 8.449349400221353e-05, "loss": 2.8904, "step": 16040 }, { "epoch": 0.9957787572164628, "grad_norm": 0.45325479621579895, "learning_rate": 8.449087940930407e-05, "loss": 2.8791, "step": 16041 }, { "epoch": 0.9958408343162207, "grad_norm": 0.46477192235628095, "learning_rate": 8.448826463644734e-05, "loss": 2.987, "step": 16042 }, { "epoch": 0.9959029114159786, "grad_norm": 0.4664447046526245, "learning_rate": 8.448564968365694e-05, "loss": 2.9244, "step": 16043 }, { "epoch": 0.9959649885157366, "grad_norm": 0.48969048653125813, "learning_rate": 8.448303455094655e-05, "loss": 2.9685, "step": 16044 }, { "epoch": 0.9960270656154945, "grad_norm": 0.5106067660285005, "learning_rate": 8.448041923832979e-05, "loss": 2.8273, "step": 16045 }, { "epoch": 0.9960891427152524, "grad_norm": 0.5145085993512463, "learning_rate": 8.447780374582032e-05, "loss": 2.9297, "step": 16046 }, { "epoch": 0.9961512198150102, "grad_norm": 0.52647520524607, "learning_rate": 8.447518807343178e-05, "loss": 2.8286, "step": 16047 }, { "epoch": 0.9962132969147681, "grad_norm": 0.4727306307441229, "learning_rate": 8.447257222117779e-05, "loss": 2.8832, "step": 16048 }, { "epoch": 0.996275374014526, "grad_norm": 0.5480290633642672, "learning_rate": 8.446995618907205e-05, "loss": 2.9927, "step": 16049 }, { "epoch": 0.996337451114284, "grad_norm": 0.5813348175020404, "learning_rate": 8.446733997712817e-05, "loss": 3.0064, "step": 16050 }, { "epoch": 0.9963995282140419, "grad_norm": 0.5455006359596636, "learning_rate": 8.446472358535981e-05, "loss": 2.9131, "step": 16051 }, { "epoch": 0.9964616053137998, "grad_norm": 0.4773873398228017, "learning_rate": 8.446210701378063e-05, "loss": 2.973, "step": 16052 }, { "epoch": 0.9965236824135576, "grad_norm": 0.49734750150032064, "learning_rate": 8.445949026240425e-05, "loss": 2.8909, "step": 16053 }, { "epoch": 0.9965857595133155, "grad_norm": 0.5364090115981404, "learning_rate": 8.445687333124436e-05, "loss": 2.954, "step": 16054 }, { "epoch": 0.9966478366130734, "grad_norm": 0.4902969867496223, "learning_rate": 8.445425622031459e-05, "loss": 2.9116, "step": 16055 }, { "epoch": 0.9967099137128314, "grad_norm": 0.5486055513360452, "learning_rate": 8.445163892962861e-05, "loss": 2.9263, "step": 16056 }, { "epoch": 0.9967719908125893, "grad_norm": 0.5714877312979504, "learning_rate": 8.444902145920006e-05, "loss": 2.8023, "step": 16057 }, { "epoch": 0.9968340679123472, "grad_norm": 0.49037151774079935, "learning_rate": 8.444640380904261e-05, "loss": 2.8979, "step": 16058 }, { "epoch": 0.996896145012105, "grad_norm": 0.5618197248212652, "learning_rate": 8.44437859791699e-05, "loss": 2.9475, "step": 16059 }, { "epoch": 0.9969582221118629, "grad_norm": 0.47777172575796734, "learning_rate": 8.44411679695956e-05, "loss": 2.8464, "step": 16060 }, { "epoch": 0.9970202992116208, "grad_norm": 0.5253354862591798, "learning_rate": 8.443854978033337e-05, "loss": 2.8654, "step": 16061 }, { "epoch": 0.9970823763113787, "grad_norm": 0.5295726719175025, "learning_rate": 8.443593141139685e-05, "loss": 2.9935, "step": 16062 }, { "epoch": 0.9971444534111367, "grad_norm": 0.4845855778138959, "learning_rate": 8.443331286279974e-05, "loss": 2.8353, "step": 16063 }, { "epoch": 0.9972065305108946, "grad_norm": 0.5790976255646505, "learning_rate": 8.443069413455566e-05, "loss": 2.9849, "step": 16064 }, { "epoch": 0.9972686076106524, "grad_norm": 0.5715628204428231, "learning_rate": 8.44280752266783e-05, "loss": 2.9672, "step": 16065 }, { "epoch": 0.9973306847104103, "grad_norm": 0.5412387319012986, "learning_rate": 8.442545613918131e-05, "loss": 2.959, "step": 16066 }, { "epoch": 0.9973927618101682, "grad_norm": 0.5988932256280121, "learning_rate": 8.442283687207835e-05, "loss": 2.8172, "step": 16067 }, { "epoch": 0.9974548389099261, "grad_norm": 0.568689047513123, "learning_rate": 8.442021742538311e-05, "loss": 2.9556, "step": 16068 }, { "epoch": 0.9975169160096841, "grad_norm": 0.5158647788892894, "learning_rate": 8.441759779910922e-05, "loss": 2.899, "step": 16069 }, { "epoch": 0.997578993109442, "grad_norm": 0.6129940113518747, "learning_rate": 8.441497799327039e-05, "loss": 2.9714, "step": 16070 }, { "epoch": 0.9976410702091998, "grad_norm": 0.5541818758901903, "learning_rate": 8.441235800788025e-05, "loss": 2.9921, "step": 16071 }, { "epoch": 0.9977031473089577, "grad_norm": 0.5596398115688157, "learning_rate": 8.440973784295249e-05, "loss": 2.8764, "step": 16072 }, { "epoch": 0.9977652244087156, "grad_norm": 0.5327683796891826, "learning_rate": 8.440711749850076e-05, "loss": 2.9259, "step": 16073 }, { "epoch": 0.9978273015084735, "grad_norm": 0.47304273205052544, "learning_rate": 8.440449697453878e-05, "loss": 2.8811, "step": 16074 }, { "epoch": 0.9978893786082315, "grad_norm": 0.5244348049501453, "learning_rate": 8.440187627108016e-05, "loss": 3.0258, "step": 16075 }, { "epoch": 0.9979514557079894, "grad_norm": 0.5014795525909245, "learning_rate": 8.43992553881386e-05, "loss": 2.8913, "step": 16076 }, { "epoch": 0.9980135328077472, "grad_norm": 0.4895470500920285, "learning_rate": 8.439663432572778e-05, "loss": 2.7947, "step": 16077 }, { "epoch": 0.9980756099075051, "grad_norm": 0.5205752846105959, "learning_rate": 8.439401308386137e-05, "loss": 3.0084, "step": 16078 }, { "epoch": 0.998137687007263, "grad_norm": 0.5307592200647687, "learning_rate": 8.439139166255303e-05, "loss": 2.8814, "step": 16079 }, { "epoch": 0.9981997641070209, "grad_norm": 0.4808572888425915, "learning_rate": 8.438877006181647e-05, "loss": 2.8261, "step": 16080 }, { "epoch": 0.9982618412067789, "grad_norm": 0.46612741835323085, "learning_rate": 8.438614828166535e-05, "loss": 2.9662, "step": 16081 }, { "epoch": 0.9983239183065368, "grad_norm": 0.5138893682317245, "learning_rate": 8.438352632211334e-05, "loss": 2.9366, "step": 16082 }, { "epoch": 0.9983859954062946, "grad_norm": 0.5369858200840114, "learning_rate": 8.438090418317412e-05, "loss": 2.9207, "step": 16083 }, { "epoch": 0.9984480725060525, "grad_norm": 0.5021286594408513, "learning_rate": 8.437828186486139e-05, "loss": 3.0367, "step": 16084 }, { "epoch": 0.9985101496058104, "grad_norm": 0.598160450003415, "learning_rate": 8.437565936718881e-05, "loss": 2.9923, "step": 16085 }, { "epoch": 0.9985722267055683, "grad_norm": 0.5283952473013754, "learning_rate": 8.437303669017007e-05, "loss": 3.0169, "step": 16086 }, { "epoch": 0.9986343038053263, "grad_norm": 0.514179660467031, "learning_rate": 8.437041383381885e-05, "loss": 3.0028, "step": 16087 }, { "epoch": 0.9986963809050842, "grad_norm": 0.5918939089146363, "learning_rate": 8.436779079814883e-05, "loss": 2.9045, "step": 16088 }, { "epoch": 0.998758458004842, "grad_norm": 0.49064928711470374, "learning_rate": 8.436516758317372e-05, "loss": 2.9171, "step": 16089 }, { "epoch": 0.9988205351045999, "grad_norm": 0.47334127990618113, "learning_rate": 8.436254418890718e-05, "loss": 2.835, "step": 16090 }, { "epoch": 0.9988826122043578, "grad_norm": 0.48173481746459135, "learning_rate": 8.435992061536291e-05, "loss": 2.9363, "step": 16091 }, { "epoch": 0.9989446893041157, "grad_norm": 0.5363362522162384, "learning_rate": 8.435729686255458e-05, "loss": 2.8732, "step": 16092 }, { "epoch": 0.9990067664038736, "grad_norm": 0.4941777302075526, "learning_rate": 8.43546729304959e-05, "loss": 2.8446, "step": 16093 }, { "epoch": 0.9990688435036316, "grad_norm": 0.5158541199230016, "learning_rate": 8.435204881920056e-05, "loss": 2.9647, "step": 16094 }, { "epoch": 0.9991309206033894, "grad_norm": 0.4991176986504229, "learning_rate": 8.434942452868222e-05, "loss": 2.955, "step": 16095 }, { "epoch": 0.9991929977031473, "grad_norm": 0.5210239220635094, "learning_rate": 8.434680005895461e-05, "loss": 2.9738, "step": 16096 }, { "epoch": 0.9992550748029052, "grad_norm": 0.5522455540931824, "learning_rate": 8.434417541003141e-05, "loss": 2.8827, "step": 16097 }, { "epoch": 0.9993171519026631, "grad_norm": 0.5517529242645054, "learning_rate": 8.43415505819263e-05, "loss": 2.9162, "step": 16098 }, { "epoch": 0.999379229002421, "grad_norm": 0.5270558454359665, "learning_rate": 8.433892557465299e-05, "loss": 2.9347, "step": 16099 }, { "epoch": 0.999441306102179, "grad_norm": 0.47925188681977776, "learning_rate": 8.433630038822517e-05, "loss": 2.9682, "step": 16100 }, { "epoch": 0.9995033832019368, "grad_norm": 0.5258913171520349, "learning_rate": 8.433367502265654e-05, "loss": 2.9647, "step": 16101 }, { "epoch": 0.9995654603016947, "grad_norm": 0.5218425477953671, "learning_rate": 8.433104947796078e-05, "loss": 2.9855, "step": 16102 }, { "epoch": 0.9996275374014526, "grad_norm": 0.5346285034664258, "learning_rate": 8.43284237541516e-05, "loss": 3.0332, "step": 16103 }, { "epoch": 0.9996896145012105, "grad_norm": 0.48482546317908937, "learning_rate": 8.43257978512427e-05, "loss": 2.8952, "step": 16104 }, { "epoch": 0.9997516916009684, "grad_norm": 0.4644082858656619, "learning_rate": 8.43231717692478e-05, "loss": 2.957, "step": 16105 }, { "epoch": 0.9998137687007264, "grad_norm": 0.5061687863516656, "learning_rate": 8.432054550818056e-05, "loss": 2.8548, "step": 16106 }, { "epoch": 0.9998758458004842, "grad_norm": 0.4532459360730573, "learning_rate": 8.43179190680547e-05, "loss": 2.9222, "step": 16107 }, { "epoch": 0.9999379229002421, "grad_norm": 0.4937972734250098, "learning_rate": 8.431529244888394e-05, "loss": 2.9196, "step": 16108 }, { "epoch": 1.0, "grad_norm": 0.5463428633120343, "learning_rate": 8.431266565068196e-05, "loss": 2.9602, "step": 16109 }, { "epoch": 1.000062077099758, "grad_norm": 0.4463584061387532, "learning_rate": 8.431003867346247e-05, "loss": 2.9677, "step": 16110 }, { "epoch": 1.0001241541995158, "grad_norm": 0.5121985653885841, "learning_rate": 8.43074115172392e-05, "loss": 2.9161, "step": 16111 }, { "epoch": 1.0001862312992738, "grad_norm": 0.48674520356372414, "learning_rate": 8.43047841820258e-05, "loss": 2.9237, "step": 16112 }, { "epoch": 1.0002483083990317, "grad_norm": 0.44469549701333194, "learning_rate": 8.430215666783605e-05, "loss": 2.9533, "step": 16113 }, { "epoch": 1.0003103854987896, "grad_norm": 0.5109126607007491, "learning_rate": 8.429952897468361e-05, "loss": 2.9073, "step": 16114 }, { "epoch": 1.0003724625985473, "grad_norm": 0.4243800880396783, "learning_rate": 8.429690110258218e-05, "loss": 2.9249, "step": 16115 }, { "epoch": 1.0004345396983052, "grad_norm": 0.543486286757128, "learning_rate": 8.429427305154552e-05, "loss": 2.9614, "step": 16116 }, { "epoch": 1.0004966167980631, "grad_norm": 0.47500551433875937, "learning_rate": 8.429164482158729e-05, "loss": 2.9554, "step": 16117 }, { "epoch": 1.000558693897821, "grad_norm": 0.5739673401483009, "learning_rate": 8.428901641272124e-05, "loss": 2.9637, "step": 16118 }, { "epoch": 1.000620770997579, "grad_norm": 0.4877073806656706, "learning_rate": 8.428638782496106e-05, "loss": 3.002, "step": 16119 }, { "epoch": 1.0006828480973369, "grad_norm": 0.4978813195417716, "learning_rate": 8.428375905832046e-05, "loss": 2.9415, "step": 16120 }, { "epoch": 1.0007449251970948, "grad_norm": 0.47366479195457123, "learning_rate": 8.42811301128132e-05, "loss": 2.8808, "step": 16121 }, { "epoch": 1.0008070022968527, "grad_norm": 0.4713835961724656, "learning_rate": 8.427850098845293e-05, "loss": 3.0243, "step": 16122 }, { "epoch": 1.0008690793966106, "grad_norm": 0.499653368244024, "learning_rate": 8.427587168525341e-05, "loss": 2.8806, "step": 16123 }, { "epoch": 1.0009311564963685, "grad_norm": 0.4737057344596483, "learning_rate": 8.427324220322835e-05, "loss": 2.913, "step": 16124 }, { "epoch": 1.0009932335961265, "grad_norm": 0.49849093219560103, "learning_rate": 8.427061254239146e-05, "loss": 2.9062, "step": 16125 }, { "epoch": 1.0010553106958844, "grad_norm": 0.4793620350513135, "learning_rate": 8.426798270275646e-05, "loss": 2.9307, "step": 16126 }, { "epoch": 1.001117387795642, "grad_norm": 0.45492490554840453, "learning_rate": 8.426535268433708e-05, "loss": 2.9116, "step": 16127 }, { "epoch": 1.0011794648954, "grad_norm": 0.5056181520297729, "learning_rate": 8.426272248714705e-05, "loss": 2.9211, "step": 16128 }, { "epoch": 1.001241541995158, "grad_norm": 0.4713006085761172, "learning_rate": 8.426009211120008e-05, "loss": 2.7697, "step": 16129 }, { "epoch": 1.0013036190949158, "grad_norm": 0.5434601296116334, "learning_rate": 8.425746155650987e-05, "loss": 2.9068, "step": 16130 }, { "epoch": 1.0013656961946737, "grad_norm": 0.48711067266834374, "learning_rate": 8.425483082309019e-05, "loss": 2.89, "step": 16131 }, { "epoch": 1.0014277732944317, "grad_norm": 0.536353588680574, "learning_rate": 8.425219991095473e-05, "loss": 2.9598, "step": 16132 }, { "epoch": 1.0014898503941896, "grad_norm": 0.4864504710044702, "learning_rate": 8.424956882011721e-05, "loss": 2.793, "step": 16133 }, { "epoch": 1.0015519274939475, "grad_norm": 0.48943799140836336, "learning_rate": 8.42469375505914e-05, "loss": 2.9579, "step": 16134 }, { "epoch": 1.0016140045937054, "grad_norm": 0.5517195966871274, "learning_rate": 8.4244306102391e-05, "loss": 2.9218, "step": 16135 }, { "epoch": 1.0016760816934633, "grad_norm": 0.5301627646039503, "learning_rate": 8.424167447552973e-05, "loss": 2.8918, "step": 16136 }, { "epoch": 1.0017381587932213, "grad_norm": 0.4627376577276013, "learning_rate": 8.423904267002134e-05, "loss": 2.8305, "step": 16137 }, { "epoch": 1.0018002358929792, "grad_norm": 0.5207639425465079, "learning_rate": 8.423641068587955e-05, "loss": 2.935, "step": 16138 }, { "epoch": 1.0018623129927369, "grad_norm": 0.4888329555278289, "learning_rate": 8.42337785231181e-05, "loss": 2.8962, "step": 16139 }, { "epoch": 1.0019243900924948, "grad_norm": 0.4936347905480323, "learning_rate": 8.423114618175072e-05, "loss": 2.8099, "step": 16140 }, { "epoch": 1.0019864671922527, "grad_norm": 0.5311337792669073, "learning_rate": 8.422851366179111e-05, "loss": 2.8749, "step": 16141 }, { "epoch": 1.0020485442920106, "grad_norm": 0.5115703338414717, "learning_rate": 8.422588096325307e-05, "loss": 2.8835, "step": 16142 }, { "epoch": 1.0021106213917685, "grad_norm": 0.4811295253266311, "learning_rate": 8.422324808615029e-05, "loss": 2.933, "step": 16143 }, { "epoch": 1.0021726984915265, "grad_norm": 0.5395126950729823, "learning_rate": 8.422061503049652e-05, "loss": 2.9282, "step": 16144 }, { "epoch": 1.0022347755912844, "grad_norm": 0.5234361335394944, "learning_rate": 8.421798179630549e-05, "loss": 2.9327, "step": 16145 }, { "epoch": 1.0022968526910423, "grad_norm": 0.49592881952872436, "learning_rate": 8.421534838359094e-05, "loss": 3.0061, "step": 16146 }, { "epoch": 1.0023589297908002, "grad_norm": 0.4426064168265802, "learning_rate": 8.421271479236661e-05, "loss": 2.8379, "step": 16147 }, { "epoch": 1.0024210068905581, "grad_norm": 0.4854107826378368, "learning_rate": 8.421008102264624e-05, "loss": 2.8346, "step": 16148 }, { "epoch": 1.002483083990316, "grad_norm": 0.4795295250178908, "learning_rate": 8.420744707444359e-05, "loss": 2.9414, "step": 16149 }, { "epoch": 1.002545161090074, "grad_norm": 0.5162325372120835, "learning_rate": 8.420481294777237e-05, "loss": 2.9537, "step": 16150 }, { "epoch": 1.0026072381898317, "grad_norm": 0.48495313535967394, "learning_rate": 8.420217864264635e-05, "loss": 2.8329, "step": 16151 }, { "epoch": 1.0026693152895896, "grad_norm": 0.4794138671556322, "learning_rate": 8.419954415907924e-05, "loss": 2.8384, "step": 16152 }, { "epoch": 1.0027313923893475, "grad_norm": 0.5229024857299107, "learning_rate": 8.419690949708481e-05, "loss": 2.8683, "step": 16153 }, { "epoch": 1.0027934694891054, "grad_norm": 0.46385997565837056, "learning_rate": 8.419427465667682e-05, "loss": 2.876, "step": 16154 }, { "epoch": 1.0028555465888633, "grad_norm": 0.5041487945457483, "learning_rate": 8.419163963786899e-05, "loss": 2.801, "step": 16155 }, { "epoch": 1.0029176236886213, "grad_norm": 0.5283963189301497, "learning_rate": 8.418900444067507e-05, "loss": 2.9415, "step": 16156 }, { "epoch": 1.0029797007883792, "grad_norm": 0.49258381161897524, "learning_rate": 8.418636906510882e-05, "loss": 2.8574, "step": 16157 }, { "epoch": 1.003041777888137, "grad_norm": 0.5516324601368051, "learning_rate": 8.418373351118397e-05, "loss": 2.9785, "step": 16158 }, { "epoch": 1.003103854987895, "grad_norm": 0.569335283177623, "learning_rate": 8.418109777891429e-05, "loss": 2.8543, "step": 16159 }, { "epoch": 1.003165932087653, "grad_norm": 0.4843776456699105, "learning_rate": 8.417846186831353e-05, "loss": 2.9064, "step": 16160 }, { "epoch": 1.0032280091874108, "grad_norm": 0.5572469788457026, "learning_rate": 8.417582577939545e-05, "loss": 3.013, "step": 16161 }, { "epoch": 1.0032900862871688, "grad_norm": 0.5005472883470168, "learning_rate": 8.417318951217377e-05, "loss": 2.9215, "step": 16162 }, { "epoch": 1.0033521633869265, "grad_norm": 0.4515120144773723, "learning_rate": 8.417055306666227e-05, "loss": 2.8801, "step": 16163 }, { "epoch": 1.0034142404866844, "grad_norm": 0.5253677930444238, "learning_rate": 8.41679164428747e-05, "loss": 2.9188, "step": 16164 }, { "epoch": 1.0034763175864423, "grad_norm": 0.47860151804548956, "learning_rate": 8.416527964082481e-05, "loss": 2.91, "step": 16165 }, { "epoch": 1.0035383946862002, "grad_norm": 0.5070861610337934, "learning_rate": 8.416264266052638e-05, "loss": 2.9317, "step": 16166 }, { "epoch": 1.0036004717859581, "grad_norm": 0.5688695813320982, "learning_rate": 8.416000550199313e-05, "loss": 3.0155, "step": 16167 }, { "epoch": 1.003662548885716, "grad_norm": 0.5896554608477983, "learning_rate": 8.415736816523884e-05, "loss": 3.0227, "step": 16168 }, { "epoch": 1.003724625985474, "grad_norm": 0.4938936181581594, "learning_rate": 8.415473065027728e-05, "loss": 2.889, "step": 16169 }, { "epoch": 1.0037867030852319, "grad_norm": 0.5214580286349988, "learning_rate": 8.41520929571222e-05, "loss": 2.9922, "step": 16170 }, { "epoch": 1.0038487801849898, "grad_norm": 0.5029279097738605, "learning_rate": 8.414945508578736e-05, "loss": 2.9007, "step": 16171 }, { "epoch": 1.0039108572847477, "grad_norm": 0.4989545293064357, "learning_rate": 8.41468170362865e-05, "loss": 2.9459, "step": 16172 }, { "epoch": 1.0039729343845056, "grad_norm": 0.481003936184423, "learning_rate": 8.414417880863344e-05, "loss": 2.8447, "step": 16173 }, { "epoch": 1.0040350114842636, "grad_norm": 0.5229495036946294, "learning_rate": 8.414154040284188e-05, "loss": 2.8719, "step": 16174 }, { "epoch": 1.0040970885840212, "grad_norm": 0.5507954264229993, "learning_rate": 8.413890181892563e-05, "loss": 3.0156, "step": 16175 }, { "epoch": 1.0041591656837792, "grad_norm": 0.6092079373750556, "learning_rate": 8.413626305689843e-05, "loss": 2.9884, "step": 16176 }, { "epoch": 1.004221242783537, "grad_norm": 0.5332767310008331, "learning_rate": 8.413362411677408e-05, "loss": 2.8628, "step": 16177 }, { "epoch": 1.004283319883295, "grad_norm": 0.5353431096987508, "learning_rate": 8.413098499856631e-05, "loss": 2.8715, "step": 16178 }, { "epoch": 1.004345396983053, "grad_norm": 0.5271913944138983, "learning_rate": 8.412834570228893e-05, "loss": 2.9252, "step": 16179 }, { "epoch": 1.0044074740828108, "grad_norm": 0.5098147898383262, "learning_rate": 8.412570622795566e-05, "loss": 2.9591, "step": 16180 }, { "epoch": 1.0044695511825688, "grad_norm": 0.5708125809101702, "learning_rate": 8.412306657558032e-05, "loss": 2.9333, "step": 16181 }, { "epoch": 1.0045316282823267, "grad_norm": 0.5382831578989087, "learning_rate": 8.412042674517663e-05, "loss": 2.9121, "step": 16182 }, { "epoch": 1.0045937053820846, "grad_norm": 0.45657643894168704, "learning_rate": 8.41177867367584e-05, "loss": 2.9062, "step": 16183 }, { "epoch": 1.0046557824818425, "grad_norm": 0.5871750034141435, "learning_rate": 8.41151465503394e-05, "loss": 2.8517, "step": 16184 }, { "epoch": 1.0047178595816004, "grad_norm": 0.5201261742360181, "learning_rate": 8.411250618593338e-05, "loss": 2.9182, "step": 16185 }, { "epoch": 1.0047799366813583, "grad_norm": 0.5209662331458595, "learning_rate": 8.410986564355417e-05, "loss": 2.8546, "step": 16186 }, { "epoch": 1.004842013781116, "grad_norm": 0.5210443704934823, "learning_rate": 8.410722492321548e-05, "loss": 2.9881, "step": 16187 }, { "epoch": 1.004904090880874, "grad_norm": 0.508992072257263, "learning_rate": 8.410458402493114e-05, "loss": 2.8625, "step": 16188 }, { "epoch": 1.0049661679806319, "grad_norm": 0.5276570837428289, "learning_rate": 8.410194294871488e-05, "loss": 2.8573, "step": 16189 }, { "epoch": 1.0050282450803898, "grad_norm": 0.49609547712370833, "learning_rate": 8.409930169458051e-05, "loss": 2.8909, "step": 16190 }, { "epoch": 1.0050903221801477, "grad_norm": 0.4908286060948644, "learning_rate": 8.409666026254181e-05, "loss": 2.9102, "step": 16191 }, { "epoch": 1.0051523992799056, "grad_norm": 0.5171313390751306, "learning_rate": 8.409401865261257e-05, "loss": 2.9223, "step": 16192 }, { "epoch": 1.0052144763796635, "grad_norm": 0.48718124079766734, "learning_rate": 8.409137686480655e-05, "loss": 2.7979, "step": 16193 }, { "epoch": 1.0052765534794215, "grad_norm": 0.46648332607526377, "learning_rate": 8.408873489913753e-05, "loss": 2.8908, "step": 16194 }, { "epoch": 1.0053386305791794, "grad_norm": 0.49208063146133135, "learning_rate": 8.408609275561931e-05, "loss": 2.9907, "step": 16195 }, { "epoch": 1.0054007076789373, "grad_norm": 0.5783287411030141, "learning_rate": 8.408345043426569e-05, "loss": 2.8407, "step": 16196 }, { "epoch": 1.0054627847786952, "grad_norm": 0.5036429671081386, "learning_rate": 8.40808079350904e-05, "loss": 2.949, "step": 16197 }, { "epoch": 1.0055248618784531, "grad_norm": 0.5446456701398118, "learning_rate": 8.40781652581073e-05, "loss": 2.9098, "step": 16198 }, { "epoch": 1.0055869389782108, "grad_norm": 0.6027635993720769, "learning_rate": 8.407552240333012e-05, "loss": 2.8024, "step": 16199 }, { "epoch": 1.0056490160779687, "grad_norm": 0.5524961409714509, "learning_rate": 8.407287937077266e-05, "loss": 2.8624, "step": 16200 }, { "epoch": 1.0057110931777267, "grad_norm": 0.5501258955265411, "learning_rate": 8.407023616044873e-05, "loss": 2.8736, "step": 16201 }, { "epoch": 1.0057731702774846, "grad_norm": 0.633441871097277, "learning_rate": 8.406759277237209e-05, "loss": 2.9231, "step": 16202 }, { "epoch": 1.0058352473772425, "grad_norm": 0.5170971633061681, "learning_rate": 8.406494920655656e-05, "loss": 2.9197, "step": 16203 }, { "epoch": 1.0058973244770004, "grad_norm": 0.5886726374326956, "learning_rate": 8.406230546301592e-05, "loss": 2.7883, "step": 16204 }, { "epoch": 1.0059594015767583, "grad_norm": 0.5306615936817168, "learning_rate": 8.405966154176397e-05, "loss": 2.9295, "step": 16205 }, { "epoch": 1.0060214786765163, "grad_norm": 0.5508273761454069, "learning_rate": 8.405701744281448e-05, "loss": 2.908, "step": 16206 }, { "epoch": 1.0060835557762742, "grad_norm": 0.5987426009945445, "learning_rate": 8.405437316618128e-05, "loss": 2.9217, "step": 16207 }, { "epoch": 1.006145632876032, "grad_norm": 0.5355452912444244, "learning_rate": 8.405172871187814e-05, "loss": 2.9566, "step": 16208 }, { "epoch": 1.00620770997579, "grad_norm": 0.583140423643315, "learning_rate": 8.404908407991886e-05, "loss": 2.8925, "step": 16209 }, { "epoch": 1.006269787075548, "grad_norm": 0.5535495798791782, "learning_rate": 8.404643927031725e-05, "loss": 2.9689, "step": 16210 }, { "epoch": 1.0063318641753056, "grad_norm": 0.5705420933355535, "learning_rate": 8.404379428308709e-05, "loss": 2.8531, "step": 16211 }, { "epoch": 1.0063939412750635, "grad_norm": 0.5744789208843666, "learning_rate": 8.404114911824221e-05, "loss": 2.8822, "step": 16212 }, { "epoch": 1.0064560183748215, "grad_norm": 0.6054504391762434, "learning_rate": 8.403850377579639e-05, "loss": 2.943, "step": 16213 }, { "epoch": 1.0065180954745794, "grad_norm": 0.5707997631262974, "learning_rate": 8.40358582557634e-05, "loss": 2.8656, "step": 16214 }, { "epoch": 1.0065801725743373, "grad_norm": 0.5672227170745695, "learning_rate": 8.403321255815711e-05, "loss": 2.9991, "step": 16215 }, { "epoch": 1.0066422496740952, "grad_norm": 0.5231838180946521, "learning_rate": 8.403056668299128e-05, "loss": 2.9637, "step": 16216 }, { "epoch": 1.0067043267738531, "grad_norm": 0.553995910188748, "learning_rate": 8.402792063027973e-05, "loss": 2.8998, "step": 16217 }, { "epoch": 1.006766403873611, "grad_norm": 0.526847351407651, "learning_rate": 8.402527440003624e-05, "loss": 2.7878, "step": 16218 }, { "epoch": 1.006828480973369, "grad_norm": 0.5544947437040686, "learning_rate": 8.402262799227465e-05, "loss": 2.9474, "step": 16219 }, { "epoch": 1.0068905580731269, "grad_norm": 0.4774693787328012, "learning_rate": 8.401998140700873e-05, "loss": 2.8532, "step": 16220 }, { "epoch": 1.0069526351728848, "grad_norm": 0.49091280010015, "learning_rate": 8.401733464425232e-05, "loss": 2.8619, "step": 16221 }, { "epoch": 1.0070147122726427, "grad_norm": 0.5139764673193254, "learning_rate": 8.401468770401922e-05, "loss": 2.9624, "step": 16222 }, { "epoch": 1.0070767893724004, "grad_norm": 0.4486878450591348, "learning_rate": 8.401204058632324e-05, "loss": 2.8988, "step": 16223 }, { "epoch": 1.0071388664721583, "grad_norm": 0.5047637681051609, "learning_rate": 8.400939329117818e-05, "loss": 2.9785, "step": 16224 }, { "epoch": 1.0072009435719163, "grad_norm": 0.49539652211027635, "learning_rate": 8.400674581859785e-05, "loss": 2.8743, "step": 16225 }, { "epoch": 1.0072630206716742, "grad_norm": 0.4537192919877861, "learning_rate": 8.400409816859609e-05, "loss": 2.8176, "step": 16226 }, { "epoch": 1.007325097771432, "grad_norm": 0.5067850429141045, "learning_rate": 8.400145034118669e-05, "loss": 3.0198, "step": 16227 }, { "epoch": 1.00738717487119, "grad_norm": 0.47323609087233026, "learning_rate": 8.399880233638347e-05, "loss": 2.8296, "step": 16228 }, { "epoch": 1.007449251970948, "grad_norm": 0.5456118027357137, "learning_rate": 8.399615415420023e-05, "loss": 2.9033, "step": 16229 }, { "epoch": 1.0075113290707058, "grad_norm": 0.5071296796199632, "learning_rate": 8.399350579465081e-05, "loss": 2.9582, "step": 16230 }, { "epoch": 1.0075734061704638, "grad_norm": 0.548426443787336, "learning_rate": 8.399085725774903e-05, "loss": 2.9288, "step": 16231 }, { "epoch": 1.0076354832702217, "grad_norm": 0.5064561483016191, "learning_rate": 8.398820854350868e-05, "loss": 2.8972, "step": 16232 }, { "epoch": 1.0076975603699796, "grad_norm": 0.5999323816741792, "learning_rate": 8.39855596519436e-05, "loss": 2.8725, "step": 16233 }, { "epoch": 1.0077596374697375, "grad_norm": 0.5882980101124259, "learning_rate": 8.398291058306762e-05, "loss": 2.9183, "step": 16234 }, { "epoch": 1.0078217145694952, "grad_norm": 0.5331419745512271, "learning_rate": 8.398026133689451e-05, "loss": 2.86, "step": 16235 }, { "epoch": 1.0078837916692531, "grad_norm": 0.5254574417680618, "learning_rate": 8.397761191343816e-05, "loss": 2.9294, "step": 16236 }, { "epoch": 1.007945868769011, "grad_norm": 0.5917038797263272, "learning_rate": 8.397496231271235e-05, "loss": 2.8995, "step": 16237 }, { "epoch": 1.008007945868769, "grad_norm": 0.5510265264131793, "learning_rate": 8.397231253473093e-05, "loss": 2.861, "step": 16238 }, { "epoch": 1.0080700229685269, "grad_norm": 0.5596380276010374, "learning_rate": 8.396966257950767e-05, "loss": 2.8889, "step": 16239 }, { "epoch": 1.0081321000682848, "grad_norm": 0.5322875381677903, "learning_rate": 8.396701244705646e-05, "loss": 2.9288, "step": 16240 }, { "epoch": 1.0081941771680427, "grad_norm": 0.5449065613047603, "learning_rate": 8.396436213739109e-05, "loss": 2.8584, "step": 16241 }, { "epoch": 1.0082562542678006, "grad_norm": 0.49412140036115465, "learning_rate": 8.396171165052542e-05, "loss": 2.862, "step": 16242 }, { "epoch": 1.0083183313675586, "grad_norm": 0.54117022897793, "learning_rate": 8.395906098647324e-05, "loss": 2.8722, "step": 16243 }, { "epoch": 1.0083804084673165, "grad_norm": 0.49599322103955423, "learning_rate": 8.395641014524839e-05, "loss": 2.9732, "step": 16244 }, { "epoch": 1.0084424855670744, "grad_norm": 0.48728377119596283, "learning_rate": 8.395375912686472e-05, "loss": 2.9272, "step": 16245 }, { "epoch": 1.0085045626668323, "grad_norm": 0.4858350437253068, "learning_rate": 8.395110793133602e-05, "loss": 2.9248, "step": 16246 }, { "epoch": 1.00856663976659, "grad_norm": 0.5101058526141418, "learning_rate": 8.394845655867618e-05, "loss": 2.9746, "step": 16247 }, { "epoch": 1.008628716866348, "grad_norm": 0.5206802397475812, "learning_rate": 8.394580500889897e-05, "loss": 2.9279, "step": 16248 }, { "epoch": 1.0086907939661058, "grad_norm": 0.5068126224748655, "learning_rate": 8.394315328201827e-05, "loss": 2.8814, "step": 16249 }, { "epoch": 1.0087528710658638, "grad_norm": 0.5214057034360166, "learning_rate": 8.394050137804791e-05, "loss": 2.861, "step": 16250 }, { "epoch": 1.0088149481656217, "grad_norm": 0.5497511029598869, "learning_rate": 8.39378492970017e-05, "loss": 2.88, "step": 16251 }, { "epoch": 1.0088770252653796, "grad_norm": 0.49643317197920306, "learning_rate": 8.39351970388935e-05, "loss": 2.9077, "step": 16252 }, { "epoch": 1.0089391023651375, "grad_norm": 0.5884153236104068, "learning_rate": 8.393254460373714e-05, "loss": 2.9243, "step": 16253 }, { "epoch": 1.0090011794648954, "grad_norm": 0.5337720637261569, "learning_rate": 8.392989199154645e-05, "loss": 2.9546, "step": 16254 }, { "epoch": 1.0090632565646533, "grad_norm": 0.4880057511362069, "learning_rate": 8.392723920233528e-05, "loss": 2.8242, "step": 16255 }, { "epoch": 1.0091253336644113, "grad_norm": 0.5375926392587058, "learning_rate": 8.392458623611745e-05, "loss": 2.9539, "step": 16256 }, { "epoch": 1.0091874107641692, "grad_norm": 0.500946817388761, "learning_rate": 8.392193309290684e-05, "loss": 3.0316, "step": 16257 }, { "epoch": 1.009249487863927, "grad_norm": 0.5520893282534844, "learning_rate": 8.391927977271726e-05, "loss": 2.8777, "step": 16258 }, { "epoch": 1.0093115649636848, "grad_norm": 0.49419408818093324, "learning_rate": 8.391662627556258e-05, "loss": 2.9981, "step": 16259 }, { "epoch": 1.0093736420634427, "grad_norm": 0.5420385667357104, "learning_rate": 8.391397260145662e-05, "loss": 2.8388, "step": 16260 }, { "epoch": 1.0094357191632006, "grad_norm": 0.5429078212006729, "learning_rate": 8.391131875041322e-05, "loss": 2.8833, "step": 16261 }, { "epoch": 1.0094977962629585, "grad_norm": 0.5072203778641562, "learning_rate": 8.390866472244625e-05, "loss": 2.8752, "step": 16262 }, { "epoch": 1.0095598733627165, "grad_norm": 0.572818944283971, "learning_rate": 8.390601051756953e-05, "loss": 2.9572, "step": 16263 }, { "epoch": 1.0096219504624744, "grad_norm": 0.47329485919335856, "learning_rate": 8.390335613579693e-05, "loss": 2.8618, "step": 16264 }, { "epoch": 1.0096840275622323, "grad_norm": 0.5384496439285511, "learning_rate": 8.39007015771423e-05, "loss": 2.9308, "step": 16265 }, { "epoch": 1.0097461046619902, "grad_norm": 0.5160818099402836, "learning_rate": 8.389804684161947e-05, "loss": 2.9756, "step": 16266 }, { "epoch": 1.0098081817617481, "grad_norm": 0.5862145341178315, "learning_rate": 8.38953919292423e-05, "loss": 2.9298, "step": 16267 }, { "epoch": 1.009870258861506, "grad_norm": 0.5432888980795618, "learning_rate": 8.389273684002463e-05, "loss": 2.9167, "step": 16268 }, { "epoch": 1.009932335961264, "grad_norm": 0.5604208827894178, "learning_rate": 8.389008157398036e-05, "loss": 2.8916, "step": 16269 }, { "epoch": 1.009994413061022, "grad_norm": 0.5377321828553346, "learning_rate": 8.388742613112329e-05, "loss": 2.9789, "step": 16270 }, { "epoch": 1.0100564901607796, "grad_norm": 0.4652547541974498, "learning_rate": 8.388477051146728e-05, "loss": 2.7908, "step": 16271 }, { "epoch": 1.0101185672605375, "grad_norm": 0.5597450023613729, "learning_rate": 8.38821147150262e-05, "loss": 2.901, "step": 16272 }, { "epoch": 1.0101806443602954, "grad_norm": 0.4769534243021158, "learning_rate": 8.38794587418139e-05, "loss": 2.9126, "step": 16273 }, { "epoch": 1.0102427214600533, "grad_norm": 0.540315995080961, "learning_rate": 8.387680259184426e-05, "loss": 2.8934, "step": 16274 }, { "epoch": 1.0103047985598113, "grad_norm": 0.5142921792676054, "learning_rate": 8.387414626513109e-05, "loss": 3.0129, "step": 16275 }, { "epoch": 1.0103668756595692, "grad_norm": 0.5584919443258166, "learning_rate": 8.387148976168829e-05, "loss": 2.8778, "step": 16276 }, { "epoch": 1.010428952759327, "grad_norm": 0.5142370385691009, "learning_rate": 8.386883308152971e-05, "loss": 2.9039, "step": 16277 }, { "epoch": 1.010491029859085, "grad_norm": 0.5817901572829179, "learning_rate": 8.386617622466919e-05, "loss": 2.8854, "step": 16278 }, { "epoch": 1.010553106958843, "grad_norm": 0.49210229771974345, "learning_rate": 8.386351919112062e-05, "loss": 2.9334, "step": 16279 }, { "epoch": 1.0106151840586008, "grad_norm": 0.579345703125, "learning_rate": 8.386086198089785e-05, "loss": 2.9331, "step": 16280 }, { "epoch": 1.0106772611583588, "grad_norm": 0.48313451170348404, "learning_rate": 8.385820459401473e-05, "loss": 2.8761, "step": 16281 }, { "epoch": 1.0107393382581167, "grad_norm": 0.5216820731442317, "learning_rate": 8.385554703048515e-05, "loss": 2.9347, "step": 16282 }, { "epoch": 1.0108014153578744, "grad_norm": 0.5072002534796506, "learning_rate": 8.385288929032295e-05, "loss": 2.8676, "step": 16283 }, { "epoch": 1.0108634924576323, "grad_norm": 0.5056790652799242, "learning_rate": 8.385023137354202e-05, "loss": 2.9052, "step": 16284 }, { "epoch": 1.0109255695573902, "grad_norm": 0.45237719555480377, "learning_rate": 8.384757328015623e-05, "loss": 2.9218, "step": 16285 }, { "epoch": 1.0109876466571481, "grad_norm": 0.5500223057732007, "learning_rate": 8.384491501017942e-05, "loss": 2.8834, "step": 16286 }, { "epoch": 1.011049723756906, "grad_norm": 0.45397996812689323, "learning_rate": 8.384225656362545e-05, "loss": 2.9221, "step": 16287 }, { "epoch": 1.011111800856664, "grad_norm": 0.5750010604433978, "learning_rate": 8.383959794050823e-05, "loss": 2.9148, "step": 16288 }, { "epoch": 1.0111738779564219, "grad_norm": 0.47895672767052544, "learning_rate": 8.38369391408416e-05, "loss": 2.9145, "step": 16289 }, { "epoch": 1.0112359550561798, "grad_norm": 0.5100988709222457, "learning_rate": 8.383428016463948e-05, "loss": 2.9626, "step": 16290 }, { "epoch": 1.0112980321559377, "grad_norm": 0.5381336225645128, "learning_rate": 8.383162101191568e-05, "loss": 2.8885, "step": 16291 }, { "epoch": 1.0113601092556956, "grad_norm": 0.5734332154657025, "learning_rate": 8.38289616826841e-05, "loss": 3.0345, "step": 16292 }, { "epoch": 1.0114221863554536, "grad_norm": 0.532129653367281, "learning_rate": 8.382630217695862e-05, "loss": 2.9131, "step": 16293 }, { "epoch": 1.0114842634552115, "grad_norm": 0.5368018639856913, "learning_rate": 8.382364249475312e-05, "loss": 3.0084, "step": 16294 }, { "epoch": 1.0115463405549692, "grad_norm": 0.6292477267091264, "learning_rate": 8.382098263608145e-05, "loss": 2.9115, "step": 16295 }, { "epoch": 1.011608417654727, "grad_norm": 0.4957169581197339, "learning_rate": 8.38183226009575e-05, "loss": 2.9927, "step": 16296 }, { "epoch": 1.011670494754485, "grad_norm": 0.5408769836160191, "learning_rate": 8.381566238939517e-05, "loss": 2.8292, "step": 16297 }, { "epoch": 1.011732571854243, "grad_norm": 0.521318902326125, "learning_rate": 8.381300200140831e-05, "loss": 2.9051, "step": 16298 }, { "epoch": 1.0117946489540008, "grad_norm": 0.536198151711747, "learning_rate": 8.381034143701081e-05, "loss": 2.8806, "step": 16299 }, { "epoch": 1.0118567260537588, "grad_norm": 0.525668669801829, "learning_rate": 8.380768069621655e-05, "loss": 2.9334, "step": 16300 }, { "epoch": 1.0119188031535167, "grad_norm": 0.5710396768460357, "learning_rate": 8.380501977903942e-05, "loss": 2.9027, "step": 16301 }, { "epoch": 1.0119808802532746, "grad_norm": 0.5153354351892003, "learning_rate": 8.380235868549329e-05, "loss": 2.8959, "step": 16302 }, { "epoch": 1.0120429573530325, "grad_norm": 0.5397240546061183, "learning_rate": 8.379969741559203e-05, "loss": 2.9268, "step": 16303 }, { "epoch": 1.0121050344527904, "grad_norm": 0.48900413408657684, "learning_rate": 8.379703596934958e-05, "loss": 2.9415, "step": 16304 }, { "epoch": 1.0121671115525483, "grad_norm": 0.5032530994708317, "learning_rate": 8.379437434677978e-05, "loss": 2.9462, "step": 16305 }, { "epoch": 1.0122291886523063, "grad_norm": 0.4675123405867717, "learning_rate": 8.379171254789651e-05, "loss": 2.8705, "step": 16306 }, { "epoch": 1.012291265752064, "grad_norm": 0.5249924693248695, "learning_rate": 8.378905057271367e-05, "loss": 2.8941, "step": 16307 }, { "epoch": 1.0123533428518219, "grad_norm": 0.5381330133743079, "learning_rate": 8.378638842124516e-05, "loss": 2.922, "step": 16308 }, { "epoch": 1.0124154199515798, "grad_norm": 0.46284365516287135, "learning_rate": 8.378372609350485e-05, "loss": 2.9043, "step": 16309 }, { "epoch": 1.0124774970513377, "grad_norm": 0.5188483719450144, "learning_rate": 8.378106358950666e-05, "loss": 2.8836, "step": 16310 }, { "epoch": 1.0125395741510956, "grad_norm": 0.5017465841864166, "learning_rate": 8.377840090926444e-05, "loss": 2.9204, "step": 16311 }, { "epoch": 1.0126016512508536, "grad_norm": 0.45780172367888033, "learning_rate": 8.37757380527921e-05, "loss": 2.9067, "step": 16312 }, { "epoch": 1.0126637283506115, "grad_norm": 0.5446226057044198, "learning_rate": 8.377307502010356e-05, "loss": 2.9031, "step": 16313 }, { "epoch": 1.0127258054503694, "grad_norm": 0.5029266357329285, "learning_rate": 8.377041181121267e-05, "loss": 2.8848, "step": 16314 }, { "epoch": 1.0127878825501273, "grad_norm": 0.505037236162805, "learning_rate": 8.376774842613335e-05, "loss": 2.9221, "step": 16315 }, { "epoch": 1.0128499596498852, "grad_norm": 0.5009624337923776, "learning_rate": 8.37650848648795e-05, "loss": 2.7806, "step": 16316 }, { "epoch": 1.0129120367496431, "grad_norm": 0.4815904193585028, "learning_rate": 8.3762421127465e-05, "loss": 2.8258, "step": 16317 }, { "epoch": 1.012974113849401, "grad_norm": 0.568861775366288, "learning_rate": 8.375975721390374e-05, "loss": 2.8488, "step": 16318 }, { "epoch": 1.0130361909491588, "grad_norm": 0.5604683159729995, "learning_rate": 8.375709312420963e-05, "loss": 2.9105, "step": 16319 }, { "epoch": 1.0130982680489167, "grad_norm": 0.4987539918878414, "learning_rate": 8.375442885839659e-05, "loss": 2.8324, "step": 16320 }, { "epoch": 1.0131603451486746, "grad_norm": 0.5320827464810179, "learning_rate": 8.375176441647847e-05, "loss": 2.7738, "step": 16321 }, { "epoch": 1.0132224222484325, "grad_norm": 0.506829910389486, "learning_rate": 8.374909979846923e-05, "loss": 2.8776, "step": 16322 }, { "epoch": 1.0132844993481904, "grad_norm": 0.5595955302096763, "learning_rate": 8.374643500438273e-05, "loss": 2.9723, "step": 16323 }, { "epoch": 1.0133465764479483, "grad_norm": 0.5369357574065177, "learning_rate": 8.374377003423288e-05, "loss": 2.9266, "step": 16324 }, { "epoch": 1.0134086535477063, "grad_norm": 0.5063777609311756, "learning_rate": 8.374110488803361e-05, "loss": 2.8442, "step": 16325 }, { "epoch": 1.0134707306474642, "grad_norm": 0.49175488669212686, "learning_rate": 8.373843956579879e-05, "loss": 2.9751, "step": 16326 }, { "epoch": 1.013532807747222, "grad_norm": 0.4762717125916798, "learning_rate": 8.373577406754235e-05, "loss": 2.9356, "step": 16327 }, { "epoch": 1.01359488484698, "grad_norm": 0.5103086889326724, "learning_rate": 8.373310839327818e-05, "loss": 2.9615, "step": 16328 }, { "epoch": 1.013656961946738, "grad_norm": 0.4483578435084337, "learning_rate": 8.37304425430202e-05, "loss": 2.886, "step": 16329 }, { "epoch": 1.0137190390464959, "grad_norm": 0.5111069552827354, "learning_rate": 8.37277765167823e-05, "loss": 2.956, "step": 16330 }, { "epoch": 1.0137811161462535, "grad_norm": 0.481745952960378, "learning_rate": 8.372511031457841e-05, "loss": 2.88, "step": 16331 }, { "epoch": 1.0138431932460115, "grad_norm": 0.5316667604994417, "learning_rate": 8.372244393642246e-05, "loss": 2.9488, "step": 16332 }, { "epoch": 1.0139052703457694, "grad_norm": 0.49395302087358384, "learning_rate": 8.37197773823283e-05, "loss": 2.862, "step": 16333 }, { "epoch": 1.0139673474455273, "grad_norm": 0.5724576816576878, "learning_rate": 8.371711065230988e-05, "loss": 2.9009, "step": 16334 }, { "epoch": 1.0140294245452852, "grad_norm": 0.5806241532105019, "learning_rate": 8.371444374638112e-05, "loss": 2.9632, "step": 16335 }, { "epoch": 1.0140915016450431, "grad_norm": 0.4905482195650311, "learning_rate": 8.371177666455591e-05, "loss": 2.8841, "step": 16336 }, { "epoch": 1.014153578744801, "grad_norm": 0.5341771769072343, "learning_rate": 8.370910940684817e-05, "loss": 2.8013, "step": 16337 }, { "epoch": 1.014215655844559, "grad_norm": 0.46894306339612707, "learning_rate": 8.370644197327184e-05, "loss": 2.9332, "step": 16338 }, { "epoch": 1.014277732944317, "grad_norm": 0.4835214323532407, "learning_rate": 8.370377436384081e-05, "loss": 2.8931, "step": 16339 }, { "epoch": 1.0143398100440748, "grad_norm": 0.48456222238364177, "learning_rate": 8.370110657856903e-05, "loss": 2.933, "step": 16340 }, { "epoch": 1.0144018871438327, "grad_norm": 0.5088100319200266, "learning_rate": 8.369843861747037e-05, "loss": 2.924, "step": 16341 }, { "epoch": 1.0144639642435906, "grad_norm": 0.5229772850504955, "learning_rate": 8.369577048055878e-05, "loss": 3.0084, "step": 16342 }, { "epoch": 1.0145260413433483, "grad_norm": 0.53600216578288, "learning_rate": 8.369310216784818e-05, "loss": 2.8887, "step": 16343 }, { "epoch": 1.0145881184431063, "grad_norm": 0.5248072622368776, "learning_rate": 8.369043367935247e-05, "loss": 2.8513, "step": 16344 }, { "epoch": 1.0146501955428642, "grad_norm": 0.5387022500882868, "learning_rate": 8.368776501508562e-05, "loss": 2.9626, "step": 16345 }, { "epoch": 1.014712272642622, "grad_norm": 0.5021528744585008, "learning_rate": 8.368509617506151e-05, "loss": 2.8676, "step": 16346 }, { "epoch": 1.01477434974238, "grad_norm": 0.4970700726761455, "learning_rate": 8.368242715929406e-05, "loss": 2.9286, "step": 16347 }, { "epoch": 1.014836426842138, "grad_norm": 0.5028147742087817, "learning_rate": 8.367975796779722e-05, "loss": 2.8304, "step": 16348 }, { "epoch": 1.0148985039418958, "grad_norm": 0.47919630567132826, "learning_rate": 8.367708860058492e-05, "loss": 2.9065, "step": 16349 }, { "epoch": 1.0149605810416538, "grad_norm": 0.511135700991686, "learning_rate": 8.367441905767106e-05, "loss": 2.8616, "step": 16350 }, { "epoch": 1.0150226581414117, "grad_norm": 0.5423016463729372, "learning_rate": 8.367174933906959e-05, "loss": 2.8934, "step": 16351 }, { "epoch": 1.0150847352411696, "grad_norm": 0.5535861889018384, "learning_rate": 8.366907944479441e-05, "loss": 2.9666, "step": 16352 }, { "epoch": 1.0151468123409275, "grad_norm": 0.5188639090613297, "learning_rate": 8.366640937485949e-05, "loss": 2.8752, "step": 16353 }, { "epoch": 1.0152088894406854, "grad_norm": 0.5561522901633809, "learning_rate": 8.366373912927871e-05, "loss": 2.9791, "step": 16354 }, { "epoch": 1.0152709665404431, "grad_norm": 0.5511366056300511, "learning_rate": 8.366106870806608e-05, "loss": 2.8077, "step": 16355 }, { "epoch": 1.015333043640201, "grad_norm": 0.5729135744415988, "learning_rate": 8.365839811123544e-05, "loss": 2.8996, "step": 16356 }, { "epoch": 1.015395120739959, "grad_norm": 0.5936476468658012, "learning_rate": 8.365572733880078e-05, "loss": 2.9097, "step": 16357 }, { "epoch": 1.0154571978397169, "grad_norm": 0.5287548259758362, "learning_rate": 8.365305639077603e-05, "loss": 2.8617, "step": 16358 }, { "epoch": 1.0155192749394748, "grad_norm": 0.5795717951624572, "learning_rate": 8.36503852671751e-05, "loss": 2.9206, "step": 16359 }, { "epoch": 1.0155813520392327, "grad_norm": 0.49740147379394933, "learning_rate": 8.364771396801197e-05, "loss": 2.8438, "step": 16360 }, { "epoch": 1.0156434291389906, "grad_norm": 0.5914240503088989, "learning_rate": 8.364504249330052e-05, "loss": 2.9069, "step": 16361 }, { "epoch": 1.0157055062387486, "grad_norm": 0.5081176134630686, "learning_rate": 8.364237084305472e-05, "loss": 2.8854, "step": 16362 }, { "epoch": 1.0157675833385065, "grad_norm": 0.6147004247643769, "learning_rate": 8.36396990172885e-05, "loss": 2.9257, "step": 16363 }, { "epoch": 1.0158296604382644, "grad_norm": 0.5334513159975665, "learning_rate": 8.363702701601581e-05, "loss": 2.9493, "step": 16364 }, { "epoch": 1.0158917375380223, "grad_norm": 0.5908780660854365, "learning_rate": 8.36343548392506e-05, "loss": 2.9264, "step": 16365 }, { "epoch": 1.0159538146377802, "grad_norm": 0.5803390070418581, "learning_rate": 8.363168248700677e-05, "loss": 3.0078, "step": 16366 }, { "epoch": 1.016015891737538, "grad_norm": 0.5809823896805408, "learning_rate": 8.36290099592983e-05, "loss": 2.9099, "step": 16367 }, { "epoch": 1.0160779688372958, "grad_norm": 0.5141862100084472, "learning_rate": 8.362633725613911e-05, "loss": 2.9633, "step": 16368 }, { "epoch": 1.0161400459370538, "grad_norm": 0.5911158303420586, "learning_rate": 8.362366437754317e-05, "loss": 2.9076, "step": 16369 }, { "epoch": 1.0162021230368117, "grad_norm": 0.5128596451199636, "learning_rate": 8.362099132352441e-05, "loss": 2.9019, "step": 16370 }, { "epoch": 1.0162642001365696, "grad_norm": 0.5878544539670031, "learning_rate": 8.361831809409677e-05, "loss": 2.8869, "step": 16371 }, { "epoch": 1.0163262772363275, "grad_norm": 0.49479532407780824, "learning_rate": 8.36156446892742e-05, "loss": 2.8648, "step": 16372 }, { "epoch": 1.0163883543360854, "grad_norm": 0.5697764137174539, "learning_rate": 8.361297110907067e-05, "loss": 2.918, "step": 16373 }, { "epoch": 1.0164504314358433, "grad_norm": 0.5637038382431514, "learning_rate": 8.361029735350008e-05, "loss": 2.9175, "step": 16374 }, { "epoch": 1.0165125085356013, "grad_norm": 0.5103303842862162, "learning_rate": 8.360762342257645e-05, "loss": 2.8475, "step": 16375 }, { "epoch": 1.0165745856353592, "grad_norm": 0.535144750095632, "learning_rate": 8.360494931631365e-05, "loss": 2.9524, "step": 16376 }, { "epoch": 1.016636662735117, "grad_norm": 0.5149472435067495, "learning_rate": 8.360227503472568e-05, "loss": 2.9161, "step": 16377 }, { "epoch": 1.016698739834875, "grad_norm": 0.5178304956655324, "learning_rate": 8.35996005778265e-05, "loss": 2.927, "step": 16378 }, { "epoch": 1.0167608169346327, "grad_norm": 0.531178189080357, "learning_rate": 8.359692594563004e-05, "loss": 3.0274, "step": 16379 }, { "epoch": 1.0168228940343906, "grad_norm": 0.5043480880345576, "learning_rate": 8.359425113815027e-05, "loss": 2.941, "step": 16380 }, { "epoch": 1.0168849711341486, "grad_norm": 0.5620567906896848, "learning_rate": 8.359157615540112e-05, "loss": 2.9574, "step": 16381 }, { "epoch": 1.0169470482339065, "grad_norm": 0.5450562342233403, "learning_rate": 8.358890099739656e-05, "loss": 2.9454, "step": 16382 }, { "epoch": 1.0170091253336644, "grad_norm": 0.5118762931295044, "learning_rate": 8.358622566415057e-05, "loss": 2.9506, "step": 16383 }, { "epoch": 1.0170712024334223, "grad_norm": 0.5346094386842013, "learning_rate": 8.358355015567707e-05, "loss": 2.9485, "step": 16384 }, { "epoch": 1.0171332795331802, "grad_norm": 0.5381333179694965, "learning_rate": 8.358087447199004e-05, "loss": 2.8602, "step": 16385 }, { "epoch": 1.0171953566329381, "grad_norm": 0.5425752270307383, "learning_rate": 8.357819861310345e-05, "loss": 2.8755, "step": 16386 }, { "epoch": 1.017257433732696, "grad_norm": 0.4845309621692579, "learning_rate": 8.357552257903123e-05, "loss": 2.8082, "step": 16387 }, { "epoch": 1.017319510832454, "grad_norm": 0.5672935376210577, "learning_rate": 8.357284636978736e-05, "loss": 2.9228, "step": 16388 }, { "epoch": 1.017381587932212, "grad_norm": 0.5026008493844356, "learning_rate": 8.357016998538581e-05, "loss": 2.9075, "step": 16389 }, { "epoch": 1.0174436650319698, "grad_norm": 0.4796186385290837, "learning_rate": 8.356749342584052e-05, "loss": 2.8145, "step": 16390 }, { "epoch": 1.0175057421317275, "grad_norm": 0.5239601088456495, "learning_rate": 8.356481669116548e-05, "loss": 3.024, "step": 16391 }, { "epoch": 1.0175678192314854, "grad_norm": 0.5046952450751431, "learning_rate": 8.356213978137464e-05, "loss": 2.9169, "step": 16392 }, { "epoch": 1.0176298963312433, "grad_norm": 0.46871143818232525, "learning_rate": 8.355946269648196e-05, "loss": 2.9014, "step": 16393 }, { "epoch": 1.0176919734310013, "grad_norm": 0.5810557645899943, "learning_rate": 8.355678543650143e-05, "loss": 2.8795, "step": 16394 }, { "epoch": 1.0177540505307592, "grad_norm": 0.5051754190269246, "learning_rate": 8.3554108001447e-05, "loss": 2.7915, "step": 16395 }, { "epoch": 1.017816127630517, "grad_norm": 0.5043057181648943, "learning_rate": 8.355143039133264e-05, "loss": 2.883, "step": 16396 }, { "epoch": 1.017878204730275, "grad_norm": 0.5169361091293424, "learning_rate": 8.354875260617232e-05, "loss": 2.8997, "step": 16397 }, { "epoch": 1.017940281830033, "grad_norm": 0.4665556248326296, "learning_rate": 8.354607464598e-05, "loss": 2.8795, "step": 16398 }, { "epoch": 1.0180023589297909, "grad_norm": 0.5245074538854927, "learning_rate": 8.35433965107697e-05, "loss": 2.795, "step": 16399 }, { "epoch": 1.0180644360295488, "grad_norm": 0.5237224714104962, "learning_rate": 8.354071820055533e-05, "loss": 2.8998, "step": 16400 }, { "epoch": 1.0181265131293067, "grad_norm": 0.5665357770472219, "learning_rate": 8.353803971535091e-05, "loss": 2.9276, "step": 16401 }, { "epoch": 1.0181885902290646, "grad_norm": 0.5711723012691718, "learning_rate": 8.353536105517039e-05, "loss": 2.9642, "step": 16402 }, { "epoch": 1.0182506673288223, "grad_norm": 0.5487688786033385, "learning_rate": 8.353268222002773e-05, "loss": 2.8412, "step": 16403 }, { "epoch": 1.0183127444285802, "grad_norm": 0.551284370864607, "learning_rate": 8.353000320993695e-05, "loss": 3.0076, "step": 16404 }, { "epoch": 1.0183748215283381, "grad_norm": 0.5824214216367414, "learning_rate": 8.352732402491199e-05, "loss": 2.8943, "step": 16405 }, { "epoch": 1.018436898628096, "grad_norm": 0.5804330786607172, "learning_rate": 8.352464466496686e-05, "loss": 2.94, "step": 16406 }, { "epoch": 1.018498975727854, "grad_norm": 0.511501382568594, "learning_rate": 8.35219651301155e-05, "loss": 2.93, "step": 16407 }, { "epoch": 1.018561052827612, "grad_norm": 0.672019610256494, "learning_rate": 8.351928542037192e-05, "loss": 2.9827, "step": 16408 }, { "epoch": 1.0186231299273698, "grad_norm": 0.5958459901688054, "learning_rate": 8.351660553575009e-05, "loss": 2.8583, "step": 16409 }, { "epoch": 1.0186852070271277, "grad_norm": 0.48880553517844355, "learning_rate": 8.351392547626398e-05, "loss": 2.8179, "step": 16410 }, { "epoch": 1.0187472841268856, "grad_norm": 0.5811393868398262, "learning_rate": 8.351124524192762e-05, "loss": 2.9261, "step": 16411 }, { "epoch": 1.0188093612266436, "grad_norm": 0.6519554261566382, "learning_rate": 8.350856483275494e-05, "loss": 2.8818, "step": 16412 }, { "epoch": 1.0188714383264015, "grad_norm": 0.5354226834395748, "learning_rate": 8.350588424875993e-05, "loss": 2.937, "step": 16413 }, { "epoch": 1.0189335154261594, "grad_norm": 0.627332744763139, "learning_rate": 8.350320348995659e-05, "loss": 2.9888, "step": 16414 }, { "epoch": 1.018995592525917, "grad_norm": 0.5126329656566685, "learning_rate": 8.350052255635891e-05, "loss": 2.9018, "step": 16415 }, { "epoch": 1.019057669625675, "grad_norm": 0.51597643926525, "learning_rate": 8.349784144798088e-05, "loss": 2.9062, "step": 16416 }, { "epoch": 1.019119746725433, "grad_norm": 0.5325329939600615, "learning_rate": 8.349516016483646e-05, "loss": 2.8892, "step": 16417 }, { "epoch": 1.0191818238251908, "grad_norm": 0.48081546772019024, "learning_rate": 8.349247870693967e-05, "loss": 2.8547, "step": 16418 }, { "epoch": 1.0192439009249488, "grad_norm": 0.4715587863813812, "learning_rate": 8.348979707430448e-05, "loss": 2.8065, "step": 16419 }, { "epoch": 1.0193059780247067, "grad_norm": 0.5117844401121131, "learning_rate": 8.348711526694491e-05, "loss": 3.0118, "step": 16420 }, { "epoch": 1.0193680551244646, "grad_norm": 0.4563564307040498, "learning_rate": 8.34844332848749e-05, "loss": 2.9223, "step": 16421 }, { "epoch": 1.0194301322242225, "grad_norm": 0.5034421809503938, "learning_rate": 8.348175112810849e-05, "loss": 2.9345, "step": 16422 }, { "epoch": 1.0194922093239804, "grad_norm": 0.4750479724149773, "learning_rate": 8.347906879665966e-05, "loss": 2.8643, "step": 16423 }, { "epoch": 1.0195542864237384, "grad_norm": 0.48254839207293687, "learning_rate": 8.34763862905424e-05, "loss": 2.9569, "step": 16424 }, { "epoch": 1.0196163635234963, "grad_norm": 0.484981357299733, "learning_rate": 8.34737036097707e-05, "loss": 2.9202, "step": 16425 }, { "epoch": 1.0196784406232542, "grad_norm": 0.5023592778916162, "learning_rate": 8.347102075435856e-05, "loss": 2.9245, "step": 16426 }, { "epoch": 1.0197405177230119, "grad_norm": 0.4715695933875908, "learning_rate": 8.346833772432e-05, "loss": 2.7551, "step": 16427 }, { "epoch": 1.0198025948227698, "grad_norm": 0.489848908636013, "learning_rate": 8.346565451966897e-05, "loss": 2.8073, "step": 16428 }, { "epoch": 1.0198646719225277, "grad_norm": 0.44432171809322074, "learning_rate": 8.34629711404195e-05, "loss": 2.9011, "step": 16429 }, { "epoch": 1.0199267490222856, "grad_norm": 0.575467108611004, "learning_rate": 8.34602875865856e-05, "loss": 2.9363, "step": 16430 }, { "epoch": 1.0199888261220436, "grad_norm": 0.502500243305508, "learning_rate": 8.345760385818124e-05, "loss": 2.9321, "step": 16431 }, { "epoch": 1.0200509032218015, "grad_norm": 0.4829973811262225, "learning_rate": 8.345491995522046e-05, "loss": 2.8016, "step": 16432 }, { "epoch": 1.0201129803215594, "grad_norm": 0.49113472864967617, "learning_rate": 8.345223587771722e-05, "loss": 2.8892, "step": 16433 }, { "epoch": 1.0201750574213173, "grad_norm": 0.5508809641945196, "learning_rate": 8.344955162568555e-05, "loss": 2.8186, "step": 16434 }, { "epoch": 1.0202371345210752, "grad_norm": 0.5214438547686556, "learning_rate": 8.344686719913946e-05, "loss": 2.9536, "step": 16435 }, { "epoch": 1.0202992116208331, "grad_norm": 0.564357102618317, "learning_rate": 8.344418259809293e-05, "loss": 2.9691, "step": 16436 }, { "epoch": 1.020361288720591, "grad_norm": 0.5284891757704742, "learning_rate": 8.344149782255999e-05, "loss": 2.9862, "step": 16437 }, { "epoch": 1.020423365820349, "grad_norm": 0.5354300028504408, "learning_rate": 8.343881287255464e-05, "loss": 2.8493, "step": 16438 }, { "epoch": 1.0204854429201067, "grad_norm": 0.5319471833391232, "learning_rate": 8.343612774809089e-05, "loss": 2.9634, "step": 16439 }, { "epoch": 1.0205475200198646, "grad_norm": 0.550384351425208, "learning_rate": 8.343344244918273e-05, "loss": 2.9139, "step": 16440 }, { "epoch": 1.0206095971196225, "grad_norm": 0.531801106198076, "learning_rate": 8.34307569758442e-05, "loss": 2.9149, "step": 16441 }, { "epoch": 1.0206716742193804, "grad_norm": 0.5267808523715753, "learning_rate": 8.342807132808927e-05, "loss": 2.8877, "step": 16442 }, { "epoch": 1.0207337513191383, "grad_norm": 0.5323918357205935, "learning_rate": 8.3425385505932e-05, "loss": 2.8574, "step": 16443 }, { "epoch": 1.0207958284188963, "grad_norm": 0.5376978476811346, "learning_rate": 8.342269950938637e-05, "loss": 2.9077, "step": 16444 }, { "epoch": 1.0208579055186542, "grad_norm": 0.4937169516645535, "learning_rate": 8.342001333846641e-05, "loss": 2.9035, "step": 16445 }, { "epoch": 1.020919982618412, "grad_norm": 0.4915825441157659, "learning_rate": 8.341732699318611e-05, "loss": 2.8953, "step": 16446 }, { "epoch": 1.02098205971817, "grad_norm": 0.43892565118396226, "learning_rate": 8.341464047355952e-05, "loss": 2.8596, "step": 16447 }, { "epoch": 1.021044136817928, "grad_norm": 0.496339136545219, "learning_rate": 8.341195377960065e-05, "loss": 2.8469, "step": 16448 }, { "epoch": 1.0211062139176859, "grad_norm": 0.4959774031087051, "learning_rate": 8.340926691132349e-05, "loss": 2.9125, "step": 16449 }, { "epoch": 1.0211682910174438, "grad_norm": 0.4999265020710464, "learning_rate": 8.340657986874207e-05, "loss": 2.8773, "step": 16450 }, { "epoch": 1.0212303681172015, "grad_norm": 0.5060091128452231, "learning_rate": 8.340389265187042e-05, "loss": 2.8914, "step": 16451 }, { "epoch": 1.0212924452169594, "grad_norm": 0.5155024527372069, "learning_rate": 8.340120526072255e-05, "loss": 2.8651, "step": 16452 }, { "epoch": 1.0213545223167173, "grad_norm": 0.5384438599904399, "learning_rate": 8.339851769531247e-05, "loss": 2.8591, "step": 16453 }, { "epoch": 1.0214165994164752, "grad_norm": 0.47174030806870815, "learning_rate": 8.339582995565423e-05, "loss": 2.8337, "step": 16454 }, { "epoch": 1.0214786765162331, "grad_norm": 0.5277445365102528, "learning_rate": 8.339314204176185e-05, "loss": 2.7835, "step": 16455 }, { "epoch": 1.021540753615991, "grad_norm": 0.4958258528772906, "learning_rate": 8.339045395364933e-05, "loss": 2.9312, "step": 16456 }, { "epoch": 1.021602830715749, "grad_norm": 0.48550675058336973, "learning_rate": 8.338776569133071e-05, "loss": 2.8828, "step": 16457 }, { "epoch": 1.021664907815507, "grad_norm": 0.5098431115434339, "learning_rate": 8.338507725482e-05, "loss": 2.9193, "step": 16458 }, { "epoch": 1.0217269849152648, "grad_norm": 0.4860646096951785, "learning_rate": 8.338238864413124e-05, "loss": 2.7608, "step": 16459 }, { "epoch": 1.0217890620150227, "grad_norm": 0.48722008408376094, "learning_rate": 8.337969985927846e-05, "loss": 2.8935, "step": 16460 }, { "epoch": 1.0218511391147806, "grad_norm": 0.5243429636575091, "learning_rate": 8.337701090027568e-05, "loss": 2.9242, "step": 16461 }, { "epoch": 1.0219132162145383, "grad_norm": 0.5430171588202848, "learning_rate": 8.337432176713693e-05, "loss": 2.7523, "step": 16462 }, { "epoch": 1.0219752933142963, "grad_norm": 0.5212077003349315, "learning_rate": 8.337163245987624e-05, "loss": 2.9683, "step": 16463 }, { "epoch": 1.0220373704140542, "grad_norm": 0.5365565281104079, "learning_rate": 8.336894297850766e-05, "loss": 2.7929, "step": 16464 }, { "epoch": 1.022099447513812, "grad_norm": 0.4896673289406628, "learning_rate": 8.33662533230452e-05, "loss": 2.9014, "step": 16465 }, { "epoch": 1.02216152461357, "grad_norm": 0.528984061846558, "learning_rate": 8.336356349350287e-05, "loss": 2.8496, "step": 16466 }, { "epoch": 1.022223601713328, "grad_norm": 0.505927269109399, "learning_rate": 8.336087348989477e-05, "loss": 2.8585, "step": 16467 }, { "epoch": 1.0222856788130859, "grad_norm": 0.5198927460613646, "learning_rate": 8.335818331223485e-05, "loss": 2.8256, "step": 16468 }, { "epoch": 1.0223477559128438, "grad_norm": 0.49027627408742924, "learning_rate": 8.335549296053722e-05, "loss": 2.9672, "step": 16469 }, { "epoch": 1.0224098330126017, "grad_norm": 0.5480463562226779, "learning_rate": 8.335280243481587e-05, "loss": 2.9117, "step": 16470 }, { "epoch": 1.0224719101123596, "grad_norm": 0.5205093011602188, "learning_rate": 8.335011173508485e-05, "loss": 2.8781, "step": 16471 }, { "epoch": 1.0225339872121175, "grad_norm": 0.47291675986346476, "learning_rate": 8.334742086135821e-05, "loss": 2.8877, "step": 16472 }, { "epoch": 1.0225960643118754, "grad_norm": 0.5678744262702491, "learning_rate": 8.334472981364999e-05, "loss": 2.955, "step": 16473 }, { "epoch": 1.0226581414116334, "grad_norm": 0.5325461731623949, "learning_rate": 8.334203859197421e-05, "loss": 2.8627, "step": 16474 }, { "epoch": 1.022720218511391, "grad_norm": 0.5617787717709486, "learning_rate": 8.333934719634493e-05, "loss": 2.954, "step": 16475 }, { "epoch": 1.022782295611149, "grad_norm": 0.5516324061111159, "learning_rate": 8.333665562677617e-05, "loss": 2.969, "step": 16476 }, { "epoch": 1.022844372710907, "grad_norm": 0.571145298793092, "learning_rate": 8.333396388328198e-05, "loss": 2.9748, "step": 16477 }, { "epoch": 1.0229064498106648, "grad_norm": 0.4702151764960612, "learning_rate": 8.333127196587642e-05, "loss": 2.8625, "step": 16478 }, { "epoch": 1.0229685269104227, "grad_norm": 0.5095439443495499, "learning_rate": 8.332857987457352e-05, "loss": 2.8836, "step": 16479 }, { "epoch": 1.0230306040101806, "grad_norm": 0.5206343048015242, "learning_rate": 8.332588760938734e-05, "loss": 2.9192, "step": 16480 }, { "epoch": 1.0230926811099386, "grad_norm": 0.49975174462842575, "learning_rate": 8.33231951703319e-05, "loss": 2.8975, "step": 16481 }, { "epoch": 1.0231547582096965, "grad_norm": 0.4744863782460968, "learning_rate": 8.332050255742125e-05, "loss": 2.8647, "step": 16482 }, { "epoch": 1.0232168353094544, "grad_norm": 0.4783490260747782, "learning_rate": 8.331780977066948e-05, "loss": 2.8637, "step": 16483 }, { "epoch": 1.0232789124092123, "grad_norm": 0.5199367403560665, "learning_rate": 8.331511681009058e-05, "loss": 2.9219, "step": 16484 }, { "epoch": 1.0233409895089702, "grad_norm": 0.5320519677160109, "learning_rate": 8.331242367569863e-05, "loss": 2.9036, "step": 16485 }, { "epoch": 1.023403066608728, "grad_norm": 0.5320974490843164, "learning_rate": 8.330973036750768e-05, "loss": 2.9487, "step": 16486 }, { "epoch": 1.0234651437084858, "grad_norm": 0.5110300686009109, "learning_rate": 8.33070368855318e-05, "loss": 2.8675, "step": 16487 }, { "epoch": 1.0235272208082438, "grad_norm": 0.4551215417149596, "learning_rate": 8.330434322978501e-05, "loss": 2.873, "step": 16488 }, { "epoch": 1.0235892979080017, "grad_norm": 0.548754514054347, "learning_rate": 8.330164940028137e-05, "loss": 2.9341, "step": 16489 }, { "epoch": 1.0236513750077596, "grad_norm": 0.455350900521754, "learning_rate": 8.329895539703496e-05, "loss": 2.9375, "step": 16490 }, { "epoch": 1.0237134521075175, "grad_norm": 0.49092629226236306, "learning_rate": 8.329626122005981e-05, "loss": 2.9274, "step": 16491 }, { "epoch": 1.0237755292072754, "grad_norm": 0.468519631044742, "learning_rate": 8.329356686936997e-05, "loss": 2.9071, "step": 16492 }, { "epoch": 1.0238376063070334, "grad_norm": 0.5052491561546621, "learning_rate": 8.329087234497952e-05, "loss": 2.8905, "step": 16493 }, { "epoch": 1.0238996834067913, "grad_norm": 0.4827840271982917, "learning_rate": 8.32881776469025e-05, "loss": 2.9355, "step": 16494 }, { "epoch": 1.0239617605065492, "grad_norm": 0.5099230994421272, "learning_rate": 8.328548277515298e-05, "loss": 2.9562, "step": 16495 }, { "epoch": 1.024023837606307, "grad_norm": 0.4622918936631601, "learning_rate": 8.328278772974502e-05, "loss": 2.8656, "step": 16496 }, { "epoch": 1.024085914706065, "grad_norm": 0.47281697590708116, "learning_rate": 8.328009251069266e-05, "loss": 2.9049, "step": 16497 }, { "epoch": 1.024147991805823, "grad_norm": 0.5325225566957857, "learning_rate": 8.327739711801e-05, "loss": 2.9859, "step": 16498 }, { "epoch": 1.0242100689055806, "grad_norm": 0.4758984520035521, "learning_rate": 8.327470155171107e-05, "loss": 2.8769, "step": 16499 }, { "epoch": 1.0242721460053386, "grad_norm": 0.5318345332188441, "learning_rate": 8.327200581180993e-05, "loss": 2.8896, "step": 16500 }, { "epoch": 1.0243342231050965, "grad_norm": 0.5506911779031274, "learning_rate": 8.326930989832069e-05, "loss": 2.947, "step": 16501 }, { "epoch": 1.0243963002048544, "grad_norm": 0.4548532315009321, "learning_rate": 8.326661381125737e-05, "loss": 2.9639, "step": 16502 }, { "epoch": 1.0244583773046123, "grad_norm": 0.5205381001719483, "learning_rate": 8.326391755063404e-05, "loss": 2.981, "step": 16503 }, { "epoch": 1.0245204544043702, "grad_norm": 0.45150315313832945, "learning_rate": 8.326122111646479e-05, "loss": 2.9118, "step": 16504 }, { "epoch": 1.0245825315041281, "grad_norm": 0.5053586506809037, "learning_rate": 8.325852450876367e-05, "loss": 2.9767, "step": 16505 }, { "epoch": 1.024644608603886, "grad_norm": 0.47666740826180043, "learning_rate": 8.325582772754474e-05, "loss": 2.8728, "step": 16506 }, { "epoch": 1.024706685703644, "grad_norm": 0.5476206464526924, "learning_rate": 8.32531307728221e-05, "loss": 2.869, "step": 16507 }, { "epoch": 1.024768762803402, "grad_norm": 0.5182999957066807, "learning_rate": 8.325043364460981e-05, "loss": 2.8794, "step": 16508 }, { "epoch": 1.0248308399031598, "grad_norm": 0.4789891605265687, "learning_rate": 8.324773634292194e-05, "loss": 2.8393, "step": 16509 }, { "epoch": 1.0248929170029175, "grad_norm": 0.5370181196892313, "learning_rate": 8.324503886777255e-05, "loss": 2.879, "step": 16510 }, { "epoch": 1.0249549941026754, "grad_norm": 0.5393237434600676, "learning_rate": 8.324234121917572e-05, "loss": 2.8403, "step": 16511 }, { "epoch": 1.0250170712024333, "grad_norm": 0.508938143336599, "learning_rate": 8.323964339714552e-05, "loss": 2.9455, "step": 16512 }, { "epoch": 1.0250791483021913, "grad_norm": 0.5585352726985008, "learning_rate": 8.323694540169602e-05, "loss": 3.0055, "step": 16513 }, { "epoch": 1.0251412254019492, "grad_norm": 0.5265584679514291, "learning_rate": 8.323424723284132e-05, "loss": 2.9408, "step": 16514 }, { "epoch": 1.025203302501707, "grad_norm": 0.5045660859028627, "learning_rate": 8.32315488905955e-05, "loss": 2.8276, "step": 16515 }, { "epoch": 1.025265379601465, "grad_norm": 0.532306966045405, "learning_rate": 8.322885037497261e-05, "loss": 2.966, "step": 16516 }, { "epoch": 1.025327456701223, "grad_norm": 0.56151712165266, "learning_rate": 8.322615168598673e-05, "loss": 2.9012, "step": 16517 }, { "epoch": 1.0253895338009809, "grad_norm": 0.5481334919635974, "learning_rate": 8.322345282365196e-05, "loss": 2.833, "step": 16518 }, { "epoch": 1.0254516109007388, "grad_norm": 0.5073310623955665, "learning_rate": 8.322075378798237e-05, "loss": 2.9, "step": 16519 }, { "epoch": 1.0255136880004967, "grad_norm": 0.6039250537466613, "learning_rate": 8.321805457899206e-05, "loss": 2.9165, "step": 16520 }, { "epoch": 1.0255757651002546, "grad_norm": 0.5019795748499266, "learning_rate": 8.321535519669506e-05, "loss": 2.9051, "step": 16521 }, { "epoch": 1.0256378422000123, "grad_norm": 0.5441020691761311, "learning_rate": 8.321265564110552e-05, "loss": 2.8469, "step": 16522 }, { "epoch": 1.0256999192997702, "grad_norm": 0.5088646479307749, "learning_rate": 8.320995591223748e-05, "loss": 2.8532, "step": 16523 }, { "epoch": 1.0257619963995281, "grad_norm": 0.5112604024989975, "learning_rate": 8.320725601010503e-05, "loss": 2.9448, "step": 16524 }, { "epoch": 1.025824073499286, "grad_norm": 0.5904177468594416, "learning_rate": 8.320455593472229e-05, "loss": 2.9328, "step": 16525 }, { "epoch": 1.025886150599044, "grad_norm": 0.4713296954158682, "learning_rate": 8.32018556861033e-05, "loss": 2.8064, "step": 16526 }, { "epoch": 1.025948227698802, "grad_norm": 0.5759283294609977, "learning_rate": 8.319915526426217e-05, "loss": 2.9219, "step": 16527 }, { "epoch": 1.0260103047985598, "grad_norm": 0.5170528409432189, "learning_rate": 8.3196454669213e-05, "loss": 2.9299, "step": 16528 }, { "epoch": 1.0260723818983177, "grad_norm": 0.5531172757229021, "learning_rate": 8.319375390096985e-05, "loss": 2.8685, "step": 16529 }, { "epoch": 1.0261344589980756, "grad_norm": 0.4867610612973146, "learning_rate": 8.319105295954683e-05, "loss": 2.9342, "step": 16530 }, { "epoch": 1.0261965360978336, "grad_norm": 0.5559432630577956, "learning_rate": 8.318835184495804e-05, "loss": 2.976, "step": 16531 }, { "epoch": 1.0262586131975915, "grad_norm": 0.574148653903145, "learning_rate": 8.318565055721756e-05, "loss": 2.8587, "step": 16532 }, { "epoch": 1.0263206902973494, "grad_norm": 0.5606579819995843, "learning_rate": 8.318294909633947e-05, "loss": 2.9251, "step": 16533 }, { "epoch": 1.026382767397107, "grad_norm": 0.4932326316376105, "learning_rate": 8.31802474623379e-05, "loss": 2.9582, "step": 16534 }, { "epoch": 1.026444844496865, "grad_norm": 0.5790600303436663, "learning_rate": 8.317754565522691e-05, "loss": 2.9215, "step": 16535 }, { "epoch": 1.026506921596623, "grad_norm": 0.5100948103963988, "learning_rate": 8.317484367502062e-05, "loss": 2.865, "step": 16536 }, { "epoch": 1.0265689986963809, "grad_norm": 0.4898792819775063, "learning_rate": 8.317214152173312e-05, "loss": 2.8408, "step": 16537 }, { "epoch": 1.0266310757961388, "grad_norm": 0.5498905148772567, "learning_rate": 8.31694391953785e-05, "loss": 2.9599, "step": 16538 }, { "epoch": 1.0266931528958967, "grad_norm": 0.5320214112390159, "learning_rate": 8.316673669597087e-05, "loss": 3.041, "step": 16539 }, { "epoch": 1.0267552299956546, "grad_norm": 0.5349783218751275, "learning_rate": 8.31640340235243e-05, "loss": 2.9148, "step": 16540 }, { "epoch": 1.0268173070954125, "grad_norm": 0.5273870944824164, "learning_rate": 8.316133117805293e-05, "loss": 2.8426, "step": 16541 }, { "epoch": 1.0268793841951704, "grad_norm": 0.594179274325835, "learning_rate": 8.315862815957083e-05, "loss": 2.8862, "step": 16542 }, { "epoch": 1.0269414612949284, "grad_norm": 0.5493247612211837, "learning_rate": 8.315592496809215e-05, "loss": 2.914, "step": 16543 }, { "epoch": 1.0270035383946863, "grad_norm": 0.5784067549899353, "learning_rate": 8.315322160363092e-05, "loss": 2.8875, "step": 16544 }, { "epoch": 1.0270656154944442, "grad_norm": 0.55352955156573, "learning_rate": 8.315051806620131e-05, "loss": 2.8647, "step": 16545 }, { "epoch": 1.027127692594202, "grad_norm": 0.5495383123833962, "learning_rate": 8.31478143558174e-05, "loss": 2.8525, "step": 16546 }, { "epoch": 1.0271897696939598, "grad_norm": 0.530435442545984, "learning_rate": 8.31451104724933e-05, "loss": 3.0101, "step": 16547 }, { "epoch": 1.0272518467937177, "grad_norm": 0.5430358735453145, "learning_rate": 8.31424064162431e-05, "loss": 2.8554, "step": 16548 }, { "epoch": 1.0273139238934756, "grad_norm": 0.5105757724840733, "learning_rate": 8.313970218708092e-05, "loss": 2.8884, "step": 16549 }, { "epoch": 1.0273760009932336, "grad_norm": 0.5654068120305195, "learning_rate": 8.313699778502086e-05, "loss": 2.9122, "step": 16550 }, { "epoch": 1.0274380780929915, "grad_norm": 0.5530822791288321, "learning_rate": 8.313429321007706e-05, "loss": 2.8726, "step": 16551 }, { "epoch": 1.0275001551927494, "grad_norm": 0.570837249765628, "learning_rate": 8.31315884622636e-05, "loss": 2.9206, "step": 16552 }, { "epoch": 1.0275622322925073, "grad_norm": 0.5429211876984215, "learning_rate": 8.31288835415946e-05, "loss": 2.9161, "step": 16553 }, { "epoch": 1.0276243093922652, "grad_norm": 0.5446581732273701, "learning_rate": 8.312617844808417e-05, "loss": 2.8482, "step": 16554 }, { "epoch": 1.0276863864920232, "grad_norm": 0.45928344722506953, "learning_rate": 8.312347318174643e-05, "loss": 2.9567, "step": 16555 }, { "epoch": 1.027748463591781, "grad_norm": 0.5752869874600873, "learning_rate": 8.312076774259548e-05, "loss": 2.9145, "step": 16556 }, { "epoch": 1.027810540691539, "grad_norm": 0.5772962558398274, "learning_rate": 8.311806213064546e-05, "loss": 2.9983, "step": 16557 }, { "epoch": 1.0278726177912967, "grad_norm": 0.4914158569997934, "learning_rate": 8.311535634591045e-05, "loss": 2.8542, "step": 16558 }, { "epoch": 1.0279346948910546, "grad_norm": 0.5316474493683249, "learning_rate": 8.311265038840462e-05, "loss": 2.8724, "step": 16559 }, { "epoch": 1.0279967719908125, "grad_norm": 0.49651134366698185, "learning_rate": 8.310994425814201e-05, "loss": 2.9108, "step": 16560 }, { "epoch": 1.0280588490905704, "grad_norm": 0.505771644083889, "learning_rate": 8.310723795513681e-05, "loss": 2.8824, "step": 16561 }, { "epoch": 1.0281209261903284, "grad_norm": 0.48942666556893377, "learning_rate": 8.310453147940311e-05, "loss": 2.8613, "step": 16562 }, { "epoch": 1.0281830032900863, "grad_norm": 0.5070968867580059, "learning_rate": 8.310182483095503e-05, "loss": 2.8787, "step": 16563 }, { "epoch": 1.0282450803898442, "grad_norm": 0.506233930921429, "learning_rate": 8.309911800980669e-05, "loss": 2.9134, "step": 16564 }, { "epoch": 1.028307157489602, "grad_norm": 0.4749068595519349, "learning_rate": 8.309641101597223e-05, "loss": 2.8742, "step": 16565 }, { "epoch": 1.02836923458936, "grad_norm": 0.49837439744490497, "learning_rate": 8.309370384946576e-05, "loss": 2.9259, "step": 16566 }, { "epoch": 1.028431311689118, "grad_norm": 0.5396644438041412, "learning_rate": 8.309099651030138e-05, "loss": 2.9062, "step": 16567 }, { "epoch": 1.0284933887888759, "grad_norm": 0.4538519388558156, "learning_rate": 8.308828899849324e-05, "loss": 2.9521, "step": 16568 }, { "epoch": 1.0285554658886338, "grad_norm": 0.5412348499336208, "learning_rate": 8.308558131405548e-05, "loss": 2.9451, "step": 16569 }, { "epoch": 1.0286175429883915, "grad_norm": 0.4898535628605944, "learning_rate": 8.30828734570022e-05, "loss": 2.9283, "step": 16570 }, { "epoch": 1.0286796200881494, "grad_norm": 0.5851261115875476, "learning_rate": 8.308016542734756e-05, "loss": 2.946, "step": 16571 }, { "epoch": 1.0287416971879073, "grad_norm": 0.4780844166305492, "learning_rate": 8.307745722510565e-05, "loss": 2.9391, "step": 16572 }, { "epoch": 1.0288037742876652, "grad_norm": 0.5272307380804292, "learning_rate": 8.307474885029061e-05, "loss": 2.9553, "step": 16573 }, { "epoch": 1.0288658513874231, "grad_norm": 0.44443854520114506, "learning_rate": 8.307204030291658e-05, "loss": 2.8185, "step": 16574 }, { "epoch": 1.028927928487181, "grad_norm": 0.47604277678123047, "learning_rate": 8.30693315829977e-05, "loss": 2.857, "step": 16575 }, { "epoch": 1.028990005586939, "grad_norm": 0.5289145633827382, "learning_rate": 8.306662269054808e-05, "loss": 2.8561, "step": 16576 }, { "epoch": 1.029052082686697, "grad_norm": 0.41793170657848117, "learning_rate": 8.306391362558186e-05, "loss": 2.7923, "step": 16577 }, { "epoch": 1.0291141597864548, "grad_norm": 0.529097797638097, "learning_rate": 8.306120438811317e-05, "loss": 2.8364, "step": 16578 }, { "epoch": 1.0291762368862127, "grad_norm": 0.4897454395841297, "learning_rate": 8.305849497815616e-05, "loss": 2.8614, "step": 16579 }, { "epoch": 1.0292383139859707, "grad_norm": 0.4787266480051893, "learning_rate": 8.305578539572496e-05, "loss": 2.9528, "step": 16580 }, { "epoch": 1.0293003910857286, "grad_norm": 0.5351824231952195, "learning_rate": 8.305307564083369e-05, "loss": 2.8941, "step": 16581 }, { "epoch": 1.0293624681854863, "grad_norm": 0.5200209066883218, "learning_rate": 8.305036571349651e-05, "loss": 2.9186, "step": 16582 }, { "epoch": 1.0294245452852442, "grad_norm": 0.46152738229661494, "learning_rate": 8.304765561372754e-05, "loss": 2.868, "step": 16583 }, { "epoch": 1.029486622385002, "grad_norm": 0.47227707147728065, "learning_rate": 8.304494534154094e-05, "loss": 2.8792, "step": 16584 }, { "epoch": 1.02954869948476, "grad_norm": 0.5233208540574583, "learning_rate": 8.304223489695083e-05, "loss": 2.9486, "step": 16585 }, { "epoch": 1.029610776584518, "grad_norm": 0.46625613571609237, "learning_rate": 8.303952427997137e-05, "loss": 2.9129, "step": 16586 }, { "epoch": 1.0296728536842759, "grad_norm": 0.5470250332611745, "learning_rate": 8.303681349061668e-05, "loss": 2.8908, "step": 16587 }, { "epoch": 1.0297349307840338, "grad_norm": 0.5258993642470584, "learning_rate": 8.303410252890092e-05, "loss": 2.9138, "step": 16588 }, { "epoch": 1.0297970078837917, "grad_norm": 0.5510276081140305, "learning_rate": 8.303139139483822e-05, "loss": 2.8684, "step": 16589 }, { "epoch": 1.0298590849835496, "grad_norm": 0.4865673809504956, "learning_rate": 8.302868008844273e-05, "loss": 2.9856, "step": 16590 }, { "epoch": 1.0299211620833075, "grad_norm": 0.5444661357374387, "learning_rate": 8.302596860972862e-05, "loss": 2.9228, "step": 16591 }, { "epoch": 1.0299832391830654, "grad_norm": 0.555641884520407, "learning_rate": 8.302325695870999e-05, "loss": 2.9319, "step": 16592 }, { "epoch": 1.0300453162828234, "grad_norm": 0.5852030410699318, "learning_rate": 8.302054513540103e-05, "loss": 2.8426, "step": 16593 }, { "epoch": 1.030107393382581, "grad_norm": 0.507512928558437, "learning_rate": 8.301783313981585e-05, "loss": 2.8327, "step": 16594 }, { "epoch": 1.030169470482339, "grad_norm": 0.5412121908007468, "learning_rate": 8.301512097196864e-05, "loss": 2.9181, "step": 16595 }, { "epoch": 1.030231547582097, "grad_norm": 0.5692102414546827, "learning_rate": 8.301240863187352e-05, "loss": 2.8871, "step": 16596 }, { "epoch": 1.0302936246818548, "grad_norm": 0.4896667355320317, "learning_rate": 8.300969611954466e-05, "loss": 2.8147, "step": 16597 }, { "epoch": 1.0303557017816127, "grad_norm": 0.5698824528321654, "learning_rate": 8.300698343499618e-05, "loss": 2.9061, "step": 16598 }, { "epoch": 1.0304177788813706, "grad_norm": 0.5415627367621166, "learning_rate": 8.300427057824227e-05, "loss": 2.9732, "step": 16599 }, { "epoch": 1.0304798559811286, "grad_norm": 0.5348855329324715, "learning_rate": 8.300155754929706e-05, "loss": 2.8311, "step": 16600 }, { "epoch": 1.0305419330808865, "grad_norm": 0.5199089111729603, "learning_rate": 8.299884434817472e-05, "loss": 2.9462, "step": 16601 }, { "epoch": 1.0306040101806444, "grad_norm": 0.5986480213118883, "learning_rate": 8.29961309748894e-05, "loss": 2.8723, "step": 16602 }, { "epoch": 1.0306660872804023, "grad_norm": 0.5404755413315797, "learning_rate": 8.299341742945526e-05, "loss": 2.9288, "step": 16603 }, { "epoch": 1.0307281643801602, "grad_norm": 0.5443953838539475, "learning_rate": 8.299070371188643e-05, "loss": 2.8788, "step": 16604 }, { "epoch": 1.0307902414799182, "grad_norm": 0.49101292769511357, "learning_rate": 8.29879898221971e-05, "loss": 2.9444, "step": 16605 }, { "epoch": 1.0308523185796759, "grad_norm": 0.5991816035448397, "learning_rate": 8.298527576040143e-05, "loss": 2.93, "step": 16606 }, { "epoch": 1.0309143956794338, "grad_norm": 0.55520659321792, "learning_rate": 8.298256152651356e-05, "loss": 2.8623, "step": 16607 }, { "epoch": 1.0309764727791917, "grad_norm": 0.5245437035603506, "learning_rate": 8.297984712054767e-05, "loss": 2.9145, "step": 16608 }, { "epoch": 1.0310385498789496, "grad_norm": 0.5300378153014088, "learning_rate": 8.297713254251788e-05, "loss": 2.8659, "step": 16609 }, { "epoch": 1.0311006269787075, "grad_norm": 0.5038773875879052, "learning_rate": 8.297441779243842e-05, "loss": 2.9388, "step": 16610 }, { "epoch": 1.0311627040784654, "grad_norm": 0.5373484963542701, "learning_rate": 8.297170287032342e-05, "loss": 2.8934, "step": 16611 }, { "epoch": 1.0312247811782234, "grad_norm": 0.493407161207576, "learning_rate": 8.2968987776187e-05, "loss": 2.9391, "step": 16612 }, { "epoch": 1.0312868582779813, "grad_norm": 0.534151205527461, "learning_rate": 8.296627251004341e-05, "loss": 2.9781, "step": 16613 }, { "epoch": 1.0313489353777392, "grad_norm": 0.5252276029409808, "learning_rate": 8.296355707190675e-05, "loss": 2.8911, "step": 16614 }, { "epoch": 1.0314110124774971, "grad_norm": 0.470316763950422, "learning_rate": 8.296084146179121e-05, "loss": 2.8817, "step": 16615 }, { "epoch": 1.031473089577255, "grad_norm": 0.5775753191757624, "learning_rate": 8.295812567971097e-05, "loss": 2.8798, "step": 16616 }, { "epoch": 1.031535166677013, "grad_norm": 0.48273039616607144, "learning_rate": 8.295540972568018e-05, "loss": 2.879, "step": 16617 }, { "epoch": 1.0315972437767706, "grad_norm": 0.506622004699888, "learning_rate": 8.295269359971303e-05, "loss": 2.7874, "step": 16618 }, { "epoch": 1.0316593208765286, "grad_norm": 0.5980281794520487, "learning_rate": 8.294997730182365e-05, "loss": 2.8929, "step": 16619 }, { "epoch": 1.0317213979762865, "grad_norm": 0.5264537223500271, "learning_rate": 8.294726083202627e-05, "loss": 2.9884, "step": 16620 }, { "epoch": 1.0317834750760444, "grad_norm": 0.569318820179135, "learning_rate": 8.294454419033502e-05, "loss": 2.8434, "step": 16621 }, { "epoch": 1.0318455521758023, "grad_norm": 0.5669199751369046, "learning_rate": 8.294182737676408e-05, "loss": 2.8693, "step": 16622 }, { "epoch": 1.0319076292755602, "grad_norm": 0.5652854353297277, "learning_rate": 8.293911039132763e-05, "loss": 2.8933, "step": 16623 }, { "epoch": 1.0319697063753182, "grad_norm": 0.592865234506904, "learning_rate": 8.293639323403986e-05, "loss": 2.9146, "step": 16624 }, { "epoch": 1.032031783475076, "grad_norm": 0.5368218501985847, "learning_rate": 8.293367590491492e-05, "loss": 2.9303, "step": 16625 }, { "epoch": 1.032093860574834, "grad_norm": 0.5686201565683596, "learning_rate": 8.2930958403967e-05, "loss": 2.8138, "step": 16626 }, { "epoch": 1.032155937674592, "grad_norm": 0.4898250740543018, "learning_rate": 8.292824073121028e-05, "loss": 2.8699, "step": 16627 }, { "epoch": 1.0322180147743498, "grad_norm": 0.5344610780688239, "learning_rate": 8.292552288665891e-05, "loss": 2.9417, "step": 16628 }, { "epoch": 1.0322800918741077, "grad_norm": 0.4577798825254685, "learning_rate": 8.292280487032713e-05, "loss": 2.85, "step": 16629 }, { "epoch": 1.0323421689738654, "grad_norm": 0.4506018534521054, "learning_rate": 8.292008668222906e-05, "loss": 2.8344, "step": 16630 }, { "epoch": 1.0324042460736234, "grad_norm": 0.5125802628539867, "learning_rate": 8.291736832237891e-05, "loss": 2.8351, "step": 16631 }, { "epoch": 1.0324663231733813, "grad_norm": 0.4547102755318633, "learning_rate": 8.291464979079086e-05, "loss": 2.8178, "step": 16632 }, { "epoch": 1.0325284002731392, "grad_norm": 0.48590326709924697, "learning_rate": 8.29119310874791e-05, "loss": 2.9543, "step": 16633 }, { "epoch": 1.032590477372897, "grad_norm": 0.45241084218920774, "learning_rate": 8.29092122124578e-05, "loss": 2.9363, "step": 16634 }, { "epoch": 1.032652554472655, "grad_norm": 0.4761504596403948, "learning_rate": 8.290649316574115e-05, "loss": 2.9252, "step": 16635 }, { "epoch": 1.032714631572413, "grad_norm": 0.5063189918696143, "learning_rate": 8.290377394734334e-05, "loss": 2.9646, "step": 16636 }, { "epoch": 1.0327767086721709, "grad_norm": 0.5142068144376041, "learning_rate": 8.290105455727855e-05, "loss": 2.8964, "step": 16637 }, { "epoch": 1.0328387857719288, "grad_norm": 0.5464896615613192, "learning_rate": 8.289833499556097e-05, "loss": 2.9083, "step": 16638 }, { "epoch": 1.0329008628716867, "grad_norm": 0.49822396514599815, "learning_rate": 8.289561526220479e-05, "loss": 2.8591, "step": 16639 }, { "epoch": 1.0329629399714446, "grad_norm": 0.4764851992090705, "learning_rate": 8.28928953572242e-05, "loss": 2.8847, "step": 16640 }, { "epoch": 1.0330250170712025, "grad_norm": 0.5235725627131158, "learning_rate": 8.289017528063339e-05, "loss": 2.9833, "step": 16641 }, { "epoch": 1.0330870941709602, "grad_norm": 0.5037284951553886, "learning_rate": 8.288745503244654e-05, "loss": 2.9025, "step": 16642 }, { "epoch": 1.0331491712707181, "grad_norm": 0.5086705807549426, "learning_rate": 8.288473461267789e-05, "loss": 2.9451, "step": 16643 }, { "epoch": 1.033211248370476, "grad_norm": 0.491914007101614, "learning_rate": 8.288201402134157e-05, "loss": 2.9487, "step": 16644 }, { "epoch": 1.033273325470234, "grad_norm": 0.541227223575614, "learning_rate": 8.28792932584518e-05, "loss": 2.9322, "step": 16645 }, { "epoch": 1.033335402569992, "grad_norm": 0.4702300706024065, "learning_rate": 8.287657232402278e-05, "loss": 2.8716, "step": 16646 }, { "epoch": 1.0333974796697498, "grad_norm": 0.544070135344556, "learning_rate": 8.287385121806869e-05, "loss": 2.9022, "step": 16647 }, { "epoch": 1.0334595567695077, "grad_norm": 0.4973686686848885, "learning_rate": 8.287112994060375e-05, "loss": 2.8269, "step": 16648 }, { "epoch": 1.0335216338692657, "grad_norm": 0.5130472773467151, "learning_rate": 8.286840849164214e-05, "loss": 2.9136, "step": 16649 }, { "epoch": 1.0335837109690236, "grad_norm": 0.5020803742961297, "learning_rate": 8.286568687119806e-05, "loss": 2.9345, "step": 16650 }, { "epoch": 1.0336457880687815, "grad_norm": 0.5641980238229111, "learning_rate": 8.286296507928574e-05, "loss": 2.9574, "step": 16651 }, { "epoch": 1.0337078651685394, "grad_norm": 0.4540616253754501, "learning_rate": 8.286024311591934e-05, "loss": 2.937, "step": 16652 }, { "epoch": 1.0337699422682973, "grad_norm": 0.49979457509587394, "learning_rate": 8.285752098111305e-05, "loss": 2.9623, "step": 16653 }, { "epoch": 1.033832019368055, "grad_norm": 0.5180524559911207, "learning_rate": 8.285479867488112e-05, "loss": 2.9068, "step": 16654 }, { "epoch": 1.033894096467813, "grad_norm": 0.5420205598570401, "learning_rate": 8.285207619723771e-05, "loss": 2.9277, "step": 16655 }, { "epoch": 1.0339561735675709, "grad_norm": 0.5166558163282909, "learning_rate": 8.284935354819707e-05, "loss": 2.9043, "step": 16656 }, { "epoch": 1.0340182506673288, "grad_norm": 0.5513170490446965, "learning_rate": 8.284663072777337e-05, "loss": 2.9031, "step": 16657 }, { "epoch": 1.0340803277670867, "grad_norm": 0.5308772911433717, "learning_rate": 8.28439077359808e-05, "loss": 2.928, "step": 16658 }, { "epoch": 1.0341424048668446, "grad_norm": 0.4954799098427477, "learning_rate": 8.284118457283364e-05, "loss": 2.9643, "step": 16659 }, { "epoch": 1.0342044819666025, "grad_norm": 0.5742654846272608, "learning_rate": 8.2838461238346e-05, "loss": 2.8869, "step": 16660 }, { "epoch": 1.0342665590663604, "grad_norm": 0.4918851226337, "learning_rate": 8.283573773253214e-05, "loss": 2.9008, "step": 16661 }, { "epoch": 1.0343286361661184, "grad_norm": 0.6111288751443029, "learning_rate": 8.28330140554063e-05, "loss": 2.9607, "step": 16662 }, { "epoch": 1.0343907132658763, "grad_norm": 0.5600891104967193, "learning_rate": 8.283029020698264e-05, "loss": 2.8986, "step": 16663 }, { "epoch": 1.0344527903656342, "grad_norm": 0.580108461967463, "learning_rate": 8.282756618727537e-05, "loss": 2.9212, "step": 16664 }, { "epoch": 1.0345148674653921, "grad_norm": 0.5776798751329505, "learning_rate": 8.282484199629873e-05, "loss": 2.9881, "step": 16665 }, { "epoch": 1.0345769445651498, "grad_norm": 0.5827046537459281, "learning_rate": 8.28221176340669e-05, "loss": 2.9737, "step": 16666 }, { "epoch": 1.0346390216649077, "grad_norm": 0.5515295856370096, "learning_rate": 8.281939310059414e-05, "loss": 2.859, "step": 16667 }, { "epoch": 1.0347010987646656, "grad_norm": 0.5215338922772311, "learning_rate": 8.281666839589464e-05, "loss": 2.9618, "step": 16668 }, { "epoch": 1.0347631758644236, "grad_norm": 0.5508958683739779, "learning_rate": 8.28139435199826e-05, "loss": 2.887, "step": 16669 }, { "epoch": 1.0348252529641815, "grad_norm": 0.5545520079708024, "learning_rate": 8.281121847287226e-05, "loss": 2.8895, "step": 16670 }, { "epoch": 1.0348873300639394, "grad_norm": 0.6120445027205706, "learning_rate": 8.280849325457782e-05, "loss": 2.9168, "step": 16671 }, { "epoch": 1.0349494071636973, "grad_norm": 0.6404266980919575, "learning_rate": 8.280576786511351e-05, "loss": 2.9806, "step": 16672 }, { "epoch": 1.0350114842634552, "grad_norm": 0.581565030157539, "learning_rate": 8.280304230449355e-05, "loss": 2.8986, "step": 16673 }, { "epoch": 1.0350735613632132, "grad_norm": 0.5953882354460126, "learning_rate": 8.280031657273215e-05, "loss": 2.8743, "step": 16674 }, { "epoch": 1.035135638462971, "grad_norm": 0.5016658866544639, "learning_rate": 8.279759066984354e-05, "loss": 2.8093, "step": 16675 }, { "epoch": 1.035197715562729, "grad_norm": 0.5765497811007794, "learning_rate": 8.279486459584194e-05, "loss": 2.8922, "step": 16676 }, { "epoch": 1.035259792662487, "grad_norm": 0.5619621884736631, "learning_rate": 8.279213835074155e-05, "loss": 2.8465, "step": 16677 }, { "epoch": 1.0353218697622446, "grad_norm": 0.5711929892591251, "learning_rate": 8.278941193455664e-05, "loss": 2.7625, "step": 16678 }, { "epoch": 1.0353839468620025, "grad_norm": 0.5430868281129609, "learning_rate": 8.27866853473014e-05, "loss": 2.8535, "step": 16679 }, { "epoch": 1.0354460239617604, "grad_norm": 0.492530400553004, "learning_rate": 8.278395858899006e-05, "loss": 2.8777, "step": 16680 }, { "epoch": 1.0355081010615184, "grad_norm": 0.5390353265077743, "learning_rate": 8.278123165963683e-05, "loss": 2.9741, "step": 16681 }, { "epoch": 1.0355701781612763, "grad_norm": 0.502528858637192, "learning_rate": 8.277850455925599e-05, "loss": 2.8657, "step": 16682 }, { "epoch": 1.0356322552610342, "grad_norm": 0.4974476968645418, "learning_rate": 8.277577728786173e-05, "loss": 2.9196, "step": 16683 }, { "epoch": 1.0356943323607921, "grad_norm": 0.4783329361931662, "learning_rate": 8.277304984546825e-05, "loss": 2.8065, "step": 16684 }, { "epoch": 1.03575640946055, "grad_norm": 0.5569489010241195, "learning_rate": 8.277032223208984e-05, "loss": 2.917, "step": 16685 }, { "epoch": 1.035818486560308, "grad_norm": 0.5146681548364319, "learning_rate": 8.276759444774071e-05, "loss": 2.9101, "step": 16686 }, { "epoch": 1.0358805636600659, "grad_norm": 0.5442231323170887, "learning_rate": 8.276486649243506e-05, "loss": 2.954, "step": 16687 }, { "epoch": 1.0359426407598238, "grad_norm": 0.5452923079635282, "learning_rate": 8.276213836618718e-05, "loss": 2.8687, "step": 16688 }, { "epoch": 1.0360047178595817, "grad_norm": 0.5361917598816159, "learning_rate": 8.275941006901124e-05, "loss": 2.9698, "step": 16689 }, { "epoch": 1.0360667949593394, "grad_norm": 0.5053326430873804, "learning_rate": 8.275668160092154e-05, "loss": 2.8979, "step": 16690 }, { "epoch": 1.0361288720590973, "grad_norm": 0.4948872289897877, "learning_rate": 8.275395296193225e-05, "loss": 2.9362, "step": 16691 }, { "epoch": 1.0361909491588552, "grad_norm": 0.49783700747369186, "learning_rate": 8.275122415205765e-05, "loss": 2.9119, "step": 16692 }, { "epoch": 1.0362530262586132, "grad_norm": 0.5581048813597357, "learning_rate": 8.274849517131195e-05, "loss": 2.8999, "step": 16693 }, { "epoch": 1.036315103358371, "grad_norm": 0.47817311387664607, "learning_rate": 8.274576601970941e-05, "loss": 2.9119, "step": 16694 }, { "epoch": 1.036377180458129, "grad_norm": 0.4840789474687669, "learning_rate": 8.274303669726426e-05, "loss": 2.9178, "step": 16695 }, { "epoch": 1.036439257557887, "grad_norm": 0.49318110370464374, "learning_rate": 8.274030720399073e-05, "loss": 2.843, "step": 16696 }, { "epoch": 1.0365013346576448, "grad_norm": 0.4797861942068061, "learning_rate": 8.273757753990307e-05, "loss": 2.9153, "step": 16697 }, { "epoch": 1.0365634117574027, "grad_norm": 0.5437087361255356, "learning_rate": 8.273484770501554e-05, "loss": 2.9883, "step": 16698 }, { "epoch": 1.0366254888571607, "grad_norm": 0.47645333087175973, "learning_rate": 8.273211769934234e-05, "loss": 2.9202, "step": 16699 }, { "epoch": 1.0366875659569186, "grad_norm": 0.49936106210004244, "learning_rate": 8.272938752289775e-05, "loss": 2.8627, "step": 16700 }, { "epoch": 1.0367496430566765, "grad_norm": 0.5306497717243402, "learning_rate": 8.2726657175696e-05, "loss": 2.8655, "step": 16701 }, { "epoch": 1.0368117201564342, "grad_norm": 0.5334832709620303, "learning_rate": 8.272392665775133e-05, "loss": 2.9888, "step": 16702 }, { "epoch": 1.036873797256192, "grad_norm": 0.5093531492730384, "learning_rate": 8.2721195969078e-05, "loss": 2.9613, "step": 16703 }, { "epoch": 1.03693587435595, "grad_norm": 0.4898901867187922, "learning_rate": 8.271846510969024e-05, "loss": 2.8326, "step": 16704 }, { "epoch": 1.036997951455708, "grad_norm": 0.4845013607466636, "learning_rate": 8.271573407960229e-05, "loss": 2.9143, "step": 16705 }, { "epoch": 1.0370600285554659, "grad_norm": 0.47451368395486143, "learning_rate": 8.271300287882843e-05, "loss": 2.8978, "step": 16706 }, { "epoch": 1.0371221056552238, "grad_norm": 0.4838933703661142, "learning_rate": 8.271027150738289e-05, "loss": 2.8861, "step": 16707 }, { "epoch": 1.0371841827549817, "grad_norm": 0.5151908954497826, "learning_rate": 8.270753996527992e-05, "loss": 2.8994, "step": 16708 }, { "epoch": 1.0372462598547396, "grad_norm": 0.4309664332999928, "learning_rate": 8.270480825253379e-05, "loss": 2.8773, "step": 16709 }, { "epoch": 1.0373083369544975, "grad_norm": 0.4946095348906812, "learning_rate": 8.270207636915871e-05, "loss": 2.9058, "step": 16710 }, { "epoch": 1.0373704140542555, "grad_norm": 0.5083752961374794, "learning_rate": 8.269934431516898e-05, "loss": 2.9214, "step": 16711 }, { "epoch": 1.0374324911540134, "grad_norm": 0.4661541268518416, "learning_rate": 8.269661209057882e-05, "loss": 2.8685, "step": 16712 }, { "epoch": 1.0374945682537713, "grad_norm": 0.5068742741251497, "learning_rate": 8.26938796954025e-05, "loss": 2.8673, "step": 16713 }, { "epoch": 1.037556645353529, "grad_norm": 0.5008843469054653, "learning_rate": 8.269114712965427e-05, "loss": 2.9194, "step": 16714 }, { "epoch": 1.037618722453287, "grad_norm": 0.5501102932968838, "learning_rate": 8.268841439334839e-05, "loss": 2.9424, "step": 16715 }, { "epoch": 1.0376807995530448, "grad_norm": 0.5139662040795371, "learning_rate": 8.268568148649912e-05, "loss": 2.9651, "step": 16716 }, { "epoch": 1.0377428766528027, "grad_norm": 0.5054935325431448, "learning_rate": 8.268294840912072e-05, "loss": 2.918, "step": 16717 }, { "epoch": 1.0378049537525607, "grad_norm": 0.4901183853627309, "learning_rate": 8.268021516122743e-05, "loss": 2.8578, "step": 16718 }, { "epoch": 1.0378670308523186, "grad_norm": 0.5292755347873126, "learning_rate": 8.267748174283354e-05, "loss": 2.845, "step": 16719 }, { "epoch": 1.0379291079520765, "grad_norm": 0.570347954680062, "learning_rate": 8.267474815395329e-05, "loss": 2.8607, "step": 16720 }, { "epoch": 1.0379911850518344, "grad_norm": 0.5361141625445442, "learning_rate": 8.267201439460095e-05, "loss": 2.9011, "step": 16721 }, { "epoch": 1.0380532621515923, "grad_norm": 0.5594735053507613, "learning_rate": 8.266928046479076e-05, "loss": 2.9593, "step": 16722 }, { "epoch": 1.0381153392513502, "grad_norm": 0.49866103657605826, "learning_rate": 8.266654636453703e-05, "loss": 2.8964, "step": 16723 }, { "epoch": 1.0381774163511082, "grad_norm": 0.5055947219025376, "learning_rate": 8.266381209385399e-05, "loss": 2.9344, "step": 16724 }, { "epoch": 1.038239493450866, "grad_norm": 0.5592286563154674, "learning_rate": 8.26610776527559e-05, "loss": 2.943, "step": 16725 }, { "epoch": 1.0383015705506238, "grad_norm": 0.5637788804516534, "learning_rate": 8.265834304125704e-05, "loss": 2.9659, "step": 16726 }, { "epoch": 1.0383636476503817, "grad_norm": 0.5208819048803549, "learning_rate": 8.265560825937169e-05, "loss": 2.9351, "step": 16727 }, { "epoch": 1.0384257247501396, "grad_norm": 0.5034985630408866, "learning_rate": 8.26528733071141e-05, "loss": 2.8605, "step": 16728 }, { "epoch": 1.0384878018498975, "grad_norm": 0.5089061990280525, "learning_rate": 8.265013818449854e-05, "loss": 2.9731, "step": 16729 }, { "epoch": 1.0385498789496554, "grad_norm": 0.5118135262582794, "learning_rate": 8.264740289153927e-05, "loss": 2.9083, "step": 16730 }, { "epoch": 1.0386119560494134, "grad_norm": 0.49989372852590047, "learning_rate": 8.26446674282506e-05, "loss": 2.7003, "step": 16731 }, { "epoch": 1.0386740331491713, "grad_norm": 0.5247271578116679, "learning_rate": 8.264193179464677e-05, "loss": 2.8943, "step": 16732 }, { "epoch": 1.0387361102489292, "grad_norm": 0.5232297285703356, "learning_rate": 8.263919599074203e-05, "loss": 3.0108, "step": 16733 }, { "epoch": 1.0387981873486871, "grad_norm": 0.5533600356321419, "learning_rate": 8.26364600165507e-05, "loss": 2.8113, "step": 16734 }, { "epoch": 1.038860264448445, "grad_norm": 0.4369607906395528, "learning_rate": 8.263372387208705e-05, "loss": 2.8162, "step": 16735 }, { "epoch": 1.038922341548203, "grad_norm": 0.5478917887398047, "learning_rate": 8.263098755736532e-05, "loss": 2.9397, "step": 16736 }, { "epoch": 1.0389844186479609, "grad_norm": 0.5876708015764157, "learning_rate": 8.262825107239982e-05, "loss": 2.7349, "step": 16737 }, { "epoch": 1.0390464957477186, "grad_norm": 0.4579901946040572, "learning_rate": 8.262551441720481e-05, "loss": 2.8806, "step": 16738 }, { "epoch": 1.0391085728474765, "grad_norm": 0.5088555115051094, "learning_rate": 8.262277759179457e-05, "loss": 2.8739, "step": 16739 }, { "epoch": 1.0391706499472344, "grad_norm": 0.46958216868709995, "learning_rate": 8.262004059618338e-05, "loss": 2.8639, "step": 16740 }, { "epoch": 1.0392327270469923, "grad_norm": 0.581331305816196, "learning_rate": 8.261730343038551e-05, "loss": 2.905, "step": 16741 }, { "epoch": 1.0392948041467502, "grad_norm": 0.5412853408710863, "learning_rate": 8.261456609441527e-05, "loss": 2.9775, "step": 16742 }, { "epoch": 1.0393568812465082, "grad_norm": 0.541061647296175, "learning_rate": 8.26118285882869e-05, "loss": 2.8977, "step": 16743 }, { "epoch": 1.039418958346266, "grad_norm": 0.48812363178681734, "learning_rate": 8.260909091201471e-05, "loss": 2.8023, "step": 16744 }, { "epoch": 1.039481035446024, "grad_norm": 0.5170761265074446, "learning_rate": 8.260635306561299e-05, "loss": 2.8771, "step": 16745 }, { "epoch": 1.039543112545782, "grad_norm": 0.5544202456240049, "learning_rate": 8.2603615049096e-05, "loss": 2.9127, "step": 16746 }, { "epoch": 1.0396051896455398, "grad_norm": 0.5448664428518568, "learning_rate": 8.260087686247802e-05, "loss": 2.9443, "step": 16747 }, { "epoch": 1.0396672667452977, "grad_norm": 0.5324249737678074, "learning_rate": 8.259813850577336e-05, "loss": 2.8695, "step": 16748 }, { "epoch": 1.0397293438450557, "grad_norm": 0.5139943840094843, "learning_rate": 8.259539997899631e-05, "loss": 2.8597, "step": 16749 }, { "epoch": 1.0397914209448134, "grad_norm": 0.5460809938089703, "learning_rate": 8.259266128216115e-05, "loss": 2.897, "step": 16750 }, { "epoch": 1.0398534980445713, "grad_norm": 0.522309554084996, "learning_rate": 8.258992241528214e-05, "loss": 2.8907, "step": 16751 }, { "epoch": 1.0399155751443292, "grad_norm": 0.5055434076045571, "learning_rate": 8.258718337837362e-05, "loss": 2.8842, "step": 16752 }, { "epoch": 1.0399776522440871, "grad_norm": 0.44506460953062393, "learning_rate": 8.258444417144983e-05, "loss": 2.9342, "step": 16753 }, { "epoch": 1.040039729343845, "grad_norm": 0.5461188674847874, "learning_rate": 8.25817047945251e-05, "loss": 2.9427, "step": 16754 }, { "epoch": 1.040101806443603, "grad_norm": 0.48145207522209765, "learning_rate": 8.25789652476137e-05, "loss": 2.9268, "step": 16755 }, { "epoch": 1.0401638835433609, "grad_norm": 0.5243505798218782, "learning_rate": 8.257622553072992e-05, "loss": 2.9505, "step": 16756 }, { "epoch": 1.0402259606431188, "grad_norm": 0.46416324202725523, "learning_rate": 8.257348564388808e-05, "loss": 2.8412, "step": 16757 }, { "epoch": 1.0402880377428767, "grad_norm": 0.5013344956525727, "learning_rate": 8.257074558710245e-05, "loss": 2.8239, "step": 16758 }, { "epoch": 1.0403501148426346, "grad_norm": 0.54429697281218, "learning_rate": 8.256800536038733e-05, "loss": 2.8108, "step": 16759 }, { "epoch": 1.0404121919423925, "grad_norm": 0.5116792911138062, "learning_rate": 8.256526496375705e-05, "loss": 2.9302, "step": 16760 }, { "epoch": 1.0404742690421505, "grad_norm": 0.5841891079790018, "learning_rate": 8.256252439722585e-05, "loss": 2.9361, "step": 16761 }, { "epoch": 1.0405363461419082, "grad_norm": 0.4819618688326013, "learning_rate": 8.255978366080806e-05, "loss": 2.8655, "step": 16762 }, { "epoch": 1.040598423241666, "grad_norm": 0.5625895852700397, "learning_rate": 8.255704275451796e-05, "loss": 2.9613, "step": 16763 }, { "epoch": 1.040660500341424, "grad_norm": 0.5115678870605487, "learning_rate": 8.255430167836988e-05, "loss": 2.7851, "step": 16764 }, { "epoch": 1.040722577441182, "grad_norm": 0.5544658809171186, "learning_rate": 8.255156043237809e-05, "loss": 2.9477, "step": 16765 }, { "epoch": 1.0407846545409398, "grad_norm": 0.6444255626609859, "learning_rate": 8.254881901655694e-05, "loss": 2.9143, "step": 16766 }, { "epoch": 1.0408467316406977, "grad_norm": 0.553759645072346, "learning_rate": 8.254607743092066e-05, "loss": 2.8878, "step": 16767 }, { "epoch": 1.0409088087404557, "grad_norm": 0.5835992059618573, "learning_rate": 8.254333567548362e-05, "loss": 2.8864, "step": 16768 }, { "epoch": 1.0409708858402136, "grad_norm": 0.49120849548096857, "learning_rate": 8.254059375026009e-05, "loss": 2.8325, "step": 16769 }, { "epoch": 1.0410329629399715, "grad_norm": 0.6555989305533461, "learning_rate": 8.253785165526438e-05, "loss": 2.9155, "step": 16770 }, { "epoch": 1.0410950400397294, "grad_norm": 0.5471849925333954, "learning_rate": 8.253510939051082e-05, "loss": 2.9029, "step": 16771 }, { "epoch": 1.0411571171394873, "grad_norm": 0.5832301854717852, "learning_rate": 8.253236695601366e-05, "loss": 2.8744, "step": 16772 }, { "epoch": 1.0412191942392452, "grad_norm": 0.6303825347831322, "learning_rate": 8.252962435178727e-05, "loss": 2.8864, "step": 16773 }, { "epoch": 1.041281271339003, "grad_norm": 0.5612956771020133, "learning_rate": 8.252688157784591e-05, "loss": 2.8983, "step": 16774 }, { "epoch": 1.0413433484387609, "grad_norm": 0.5946570044118863, "learning_rate": 8.252413863420394e-05, "loss": 2.8624, "step": 16775 }, { "epoch": 1.0414054255385188, "grad_norm": 0.5628438534388395, "learning_rate": 8.252139552087563e-05, "loss": 2.8575, "step": 16776 }, { "epoch": 1.0414675026382767, "grad_norm": 0.607071336873359, "learning_rate": 8.251865223787531e-05, "loss": 2.8744, "step": 16777 }, { "epoch": 1.0415295797380346, "grad_norm": 0.5354885545356939, "learning_rate": 8.251590878521728e-05, "loss": 2.9718, "step": 16778 }, { "epoch": 1.0415916568377925, "grad_norm": 0.5584390065792448, "learning_rate": 8.251316516291586e-05, "loss": 2.9726, "step": 16779 }, { "epoch": 1.0416537339375505, "grad_norm": 0.5963266580114583, "learning_rate": 8.251042137098537e-05, "loss": 2.8513, "step": 16780 }, { "epoch": 1.0417158110373084, "grad_norm": 0.5186169258311132, "learning_rate": 8.250767740944011e-05, "loss": 2.9632, "step": 16781 }, { "epoch": 1.0417778881370663, "grad_norm": 0.5793257949822989, "learning_rate": 8.25049332782944e-05, "loss": 2.7766, "step": 16782 }, { "epoch": 1.0418399652368242, "grad_norm": 0.5785149213025478, "learning_rate": 8.250218897756259e-05, "loss": 2.9677, "step": 16783 }, { "epoch": 1.0419020423365821, "grad_norm": 0.6307333002393385, "learning_rate": 8.249944450725895e-05, "loss": 2.8758, "step": 16784 }, { "epoch": 1.04196411943634, "grad_norm": 0.553198279359711, "learning_rate": 8.249669986739783e-05, "loss": 2.9321, "step": 16785 }, { "epoch": 1.0420261965360977, "grad_norm": 0.5777301213002767, "learning_rate": 8.249395505799353e-05, "loss": 2.9399, "step": 16786 }, { "epoch": 1.0420882736358557, "grad_norm": 0.651583840000131, "learning_rate": 8.249121007906038e-05, "loss": 2.9851, "step": 16787 }, { "epoch": 1.0421503507356136, "grad_norm": 0.5519584898475342, "learning_rate": 8.24884649306127e-05, "loss": 2.9686, "step": 16788 }, { "epoch": 1.0422124278353715, "grad_norm": 0.6289342080293295, "learning_rate": 8.24857196126648e-05, "loss": 2.9037, "step": 16789 }, { "epoch": 1.0422745049351294, "grad_norm": 0.7292267547553876, "learning_rate": 8.248297412523103e-05, "loss": 2.9325, "step": 16790 }, { "epoch": 1.0423365820348873, "grad_norm": 0.4929723990376349, "learning_rate": 8.24802284683257e-05, "loss": 2.963, "step": 16791 }, { "epoch": 1.0423986591346452, "grad_norm": 0.5393687497834394, "learning_rate": 8.247748264196313e-05, "loss": 2.7902, "step": 16792 }, { "epoch": 1.0424607362344032, "grad_norm": 0.5737538953827577, "learning_rate": 8.247473664615763e-05, "loss": 2.927, "step": 16793 }, { "epoch": 1.042522813334161, "grad_norm": 0.5467015400042351, "learning_rate": 8.247199048092357e-05, "loss": 3.0009, "step": 16794 }, { "epoch": 1.042584890433919, "grad_norm": 0.5381713633219627, "learning_rate": 8.246924414627523e-05, "loss": 2.801, "step": 16795 }, { "epoch": 1.042646967533677, "grad_norm": 0.515416739049283, "learning_rate": 8.246649764222698e-05, "loss": 2.8687, "step": 16796 }, { "epoch": 1.0427090446334346, "grad_norm": 0.5434349309192582, "learning_rate": 8.246375096879313e-05, "loss": 2.9024, "step": 16797 }, { "epoch": 1.0427711217331925, "grad_norm": 0.48987680290203833, "learning_rate": 8.246100412598801e-05, "loss": 2.8996, "step": 16798 }, { "epoch": 1.0428331988329504, "grad_norm": 0.5160409809960667, "learning_rate": 8.245825711382595e-05, "loss": 2.8896, "step": 16799 }, { "epoch": 1.0428952759327084, "grad_norm": 0.4766992310285518, "learning_rate": 8.245550993232127e-05, "loss": 2.8635, "step": 16800 }, { "epoch": 1.0429573530324663, "grad_norm": 0.466987586866282, "learning_rate": 8.245276258148833e-05, "loss": 2.9358, "step": 16801 }, { "epoch": 1.0430194301322242, "grad_norm": 0.542551744999328, "learning_rate": 8.245001506134144e-05, "loss": 2.9142, "step": 16802 }, { "epoch": 1.0430815072319821, "grad_norm": 0.4785920996766712, "learning_rate": 8.244726737189495e-05, "loss": 2.9372, "step": 16803 }, { "epoch": 1.04314358433174, "grad_norm": 0.49328282503124915, "learning_rate": 8.244451951316319e-05, "loss": 2.959, "step": 16804 }, { "epoch": 1.043205661431498, "grad_norm": 0.4943179238427345, "learning_rate": 8.24417714851605e-05, "loss": 2.9338, "step": 16805 }, { "epoch": 1.0432677385312559, "grad_norm": 0.485229584316029, "learning_rate": 8.24390232879012e-05, "loss": 2.972, "step": 16806 }, { "epoch": 1.0433298156310138, "grad_norm": 0.5271776290901081, "learning_rate": 8.243627492139965e-05, "loss": 2.9241, "step": 16807 }, { "epoch": 1.0433918927307717, "grad_norm": 0.4651208580527742, "learning_rate": 8.243352638567017e-05, "loss": 2.8551, "step": 16808 }, { "epoch": 1.0434539698305296, "grad_norm": 0.5074766368843889, "learning_rate": 8.243077768072713e-05, "loss": 2.9051, "step": 16809 }, { "epoch": 1.0435160469302873, "grad_norm": 0.4887345155716777, "learning_rate": 8.242802880658483e-05, "loss": 2.8632, "step": 16810 }, { "epoch": 1.0435781240300452, "grad_norm": 0.5072662937260363, "learning_rate": 8.242527976325764e-05, "loss": 2.8606, "step": 16811 }, { "epoch": 1.0436402011298032, "grad_norm": 0.47997359215592356, "learning_rate": 8.242253055075988e-05, "loss": 2.9249, "step": 16812 }, { "epoch": 1.043702278229561, "grad_norm": 0.5616604049156231, "learning_rate": 8.241978116910592e-05, "loss": 3.0153, "step": 16813 }, { "epoch": 1.043764355329319, "grad_norm": 0.47325208640205424, "learning_rate": 8.24170316183101e-05, "loss": 2.9183, "step": 16814 }, { "epoch": 1.043826432429077, "grad_norm": 0.47023699461674207, "learning_rate": 8.241428189838674e-05, "loss": 2.989, "step": 16815 }, { "epoch": 1.0438885095288348, "grad_norm": 0.6086922268408305, "learning_rate": 8.241153200935021e-05, "loss": 2.8485, "step": 16816 }, { "epoch": 1.0439505866285927, "grad_norm": 0.4568406710271414, "learning_rate": 8.240878195121485e-05, "loss": 2.8722, "step": 16817 }, { "epoch": 1.0440126637283507, "grad_norm": 0.5595614980166499, "learning_rate": 8.240603172399501e-05, "loss": 2.908, "step": 16818 }, { "epoch": 1.0440747408281086, "grad_norm": 0.4635907122207437, "learning_rate": 8.240328132770503e-05, "loss": 2.9282, "step": 16819 }, { "epoch": 1.0441368179278665, "grad_norm": 0.5349798259763974, "learning_rate": 8.240053076235926e-05, "loss": 2.9293, "step": 16820 }, { "epoch": 1.0441988950276242, "grad_norm": 0.529406011949867, "learning_rate": 8.239778002797207e-05, "loss": 2.913, "step": 16821 }, { "epoch": 1.0442609721273821, "grad_norm": 0.5122754512215029, "learning_rate": 8.239502912455779e-05, "loss": 2.8436, "step": 16822 }, { "epoch": 1.04432304922714, "grad_norm": 0.5249048419271948, "learning_rate": 8.239227805213078e-05, "loss": 2.9334, "step": 16823 }, { "epoch": 1.044385126326898, "grad_norm": 0.5600442260921348, "learning_rate": 8.238952681070538e-05, "loss": 2.9513, "step": 16824 }, { "epoch": 1.0444472034266559, "grad_norm": 0.536581383330089, "learning_rate": 8.238677540029598e-05, "loss": 2.8613, "step": 16825 }, { "epoch": 1.0445092805264138, "grad_norm": 0.5038566564950971, "learning_rate": 8.23840238209169e-05, "loss": 2.9584, "step": 16826 }, { "epoch": 1.0445713576261717, "grad_norm": 0.5467762994253791, "learning_rate": 8.238127207258249e-05, "loss": 2.9111, "step": 16827 }, { "epoch": 1.0446334347259296, "grad_norm": 0.5041850893211969, "learning_rate": 8.237852015530714e-05, "loss": 2.9425, "step": 16828 }, { "epoch": 1.0446955118256875, "grad_norm": 0.5062605103237643, "learning_rate": 8.237576806910518e-05, "loss": 2.8681, "step": 16829 }, { "epoch": 1.0447575889254455, "grad_norm": 0.5703563673314597, "learning_rate": 8.237301581399098e-05, "loss": 2.839, "step": 16830 }, { "epoch": 1.0448196660252034, "grad_norm": 0.4713055724355838, "learning_rate": 8.237026338997891e-05, "loss": 2.93, "step": 16831 }, { "epoch": 1.0448817431249613, "grad_norm": 0.5359811203446354, "learning_rate": 8.236751079708331e-05, "loss": 2.8786, "step": 16832 }, { "epoch": 1.0449438202247192, "grad_norm": 0.5099161736935393, "learning_rate": 8.236475803531853e-05, "loss": 2.9068, "step": 16833 }, { "epoch": 1.045005897324477, "grad_norm": 0.5083699907572414, "learning_rate": 8.236200510469898e-05, "loss": 2.9519, "step": 16834 }, { "epoch": 1.0450679744242348, "grad_norm": 0.5577575453500496, "learning_rate": 8.235925200523897e-05, "loss": 2.9707, "step": 16835 }, { "epoch": 1.0451300515239927, "grad_norm": 0.6088823625765133, "learning_rate": 8.235649873695291e-05, "loss": 2.9961, "step": 16836 }, { "epoch": 1.0451921286237507, "grad_norm": 0.5169834967770139, "learning_rate": 8.235374529985513e-05, "loss": 2.973, "step": 16837 }, { "epoch": 1.0452542057235086, "grad_norm": 0.5360332181106363, "learning_rate": 8.235099169396e-05, "loss": 2.9193, "step": 16838 }, { "epoch": 1.0453162828232665, "grad_norm": 0.5011533607399149, "learning_rate": 8.23482379192819e-05, "loss": 2.8449, "step": 16839 }, { "epoch": 1.0453783599230244, "grad_norm": 0.4533647527775095, "learning_rate": 8.23454839758352e-05, "loss": 2.8766, "step": 16840 }, { "epoch": 1.0454404370227823, "grad_norm": 0.4686361174527709, "learning_rate": 8.234272986363424e-05, "loss": 2.9168, "step": 16841 }, { "epoch": 1.0455025141225402, "grad_norm": 0.4368500138642871, "learning_rate": 8.233997558269342e-05, "loss": 2.8711, "step": 16842 }, { "epoch": 1.0455645912222982, "grad_norm": 0.4864823583215289, "learning_rate": 8.233722113302709e-05, "loss": 2.7958, "step": 16843 }, { "epoch": 1.045626668322056, "grad_norm": 0.4391014047945973, "learning_rate": 8.233446651464963e-05, "loss": 2.8575, "step": 16844 }, { "epoch": 1.0456887454218138, "grad_norm": 0.47349807553167106, "learning_rate": 8.233171172757539e-05, "loss": 2.9161, "step": 16845 }, { "epoch": 1.0457508225215717, "grad_norm": 0.4918053520993069, "learning_rate": 8.232895677181878e-05, "loss": 2.8874, "step": 16846 }, { "epoch": 1.0458128996213296, "grad_norm": 0.4733990121688899, "learning_rate": 8.232620164739415e-05, "loss": 2.9067, "step": 16847 }, { "epoch": 1.0458749767210875, "grad_norm": 0.43478485983416104, "learning_rate": 8.232344635431589e-05, "loss": 2.8219, "step": 16848 }, { "epoch": 1.0459370538208455, "grad_norm": 0.4899877254495978, "learning_rate": 8.232069089259835e-05, "loss": 2.9242, "step": 16849 }, { "epoch": 1.0459991309206034, "grad_norm": 0.44975644978091167, "learning_rate": 8.231793526225592e-05, "loss": 2.9207, "step": 16850 }, { "epoch": 1.0460612080203613, "grad_norm": 0.5424969771076837, "learning_rate": 8.231517946330298e-05, "loss": 2.8758, "step": 16851 }, { "epoch": 1.0461232851201192, "grad_norm": 0.4790759328960781, "learning_rate": 8.231242349575391e-05, "loss": 2.9122, "step": 16852 }, { "epoch": 1.0461853622198771, "grad_norm": 0.49306323843180333, "learning_rate": 8.230966735962306e-05, "loss": 2.9863, "step": 16853 }, { "epoch": 1.046247439319635, "grad_norm": 0.5116152767950924, "learning_rate": 8.230691105492487e-05, "loss": 2.8789, "step": 16854 }, { "epoch": 1.046309516419393, "grad_norm": 0.5270494275502435, "learning_rate": 8.230415458167365e-05, "loss": 2.8914, "step": 16855 }, { "epoch": 1.0463715935191509, "grad_norm": 0.48357385092177074, "learning_rate": 8.230139793988383e-05, "loss": 3.0219, "step": 16856 }, { "epoch": 1.0464336706189088, "grad_norm": 0.4889988927945667, "learning_rate": 8.229864112956977e-05, "loss": 2.9327, "step": 16857 }, { "epoch": 1.0464957477186665, "grad_norm": 0.47371422765946086, "learning_rate": 8.229588415074585e-05, "loss": 2.8892, "step": 16858 }, { "epoch": 1.0465578248184244, "grad_norm": 0.4794085987208338, "learning_rate": 8.229312700342649e-05, "loss": 2.9612, "step": 16859 }, { "epoch": 1.0466199019181823, "grad_norm": 0.5229122031225258, "learning_rate": 8.229036968762602e-05, "loss": 2.8893, "step": 16860 }, { "epoch": 1.0466819790179402, "grad_norm": 0.4288565143035733, "learning_rate": 8.228761220335884e-05, "loss": 2.8553, "step": 16861 }, { "epoch": 1.0467440561176982, "grad_norm": 0.480756704368996, "learning_rate": 8.228485455063937e-05, "loss": 2.9225, "step": 16862 }, { "epoch": 1.046806133217456, "grad_norm": 0.5281963943438543, "learning_rate": 8.228209672948198e-05, "loss": 2.9492, "step": 16863 }, { "epoch": 1.046868210317214, "grad_norm": 0.5255605713483081, "learning_rate": 8.227933873990105e-05, "loss": 2.9998, "step": 16864 }, { "epoch": 1.046930287416972, "grad_norm": 0.46615075440890463, "learning_rate": 8.227658058191098e-05, "loss": 2.9479, "step": 16865 }, { "epoch": 1.0469923645167298, "grad_norm": 0.5081387278918263, "learning_rate": 8.227382225552615e-05, "loss": 2.9008, "step": 16866 }, { "epoch": 1.0470544416164878, "grad_norm": 0.5315777665087111, "learning_rate": 8.227106376076095e-05, "loss": 2.9589, "step": 16867 }, { "epoch": 1.0471165187162457, "grad_norm": 0.4553302508515612, "learning_rate": 8.226830509762977e-05, "loss": 2.9418, "step": 16868 }, { "epoch": 1.0471785958160034, "grad_norm": 0.4447335277442486, "learning_rate": 8.226554626614702e-05, "loss": 2.9729, "step": 16869 }, { "epoch": 1.0472406729157613, "grad_norm": 0.4833174049970116, "learning_rate": 8.226278726632707e-05, "loss": 2.9088, "step": 16870 }, { "epoch": 1.0473027500155192, "grad_norm": 0.4346356632086449, "learning_rate": 8.226002809818434e-05, "loss": 2.9219, "step": 16871 }, { "epoch": 1.0473648271152771, "grad_norm": 0.48699761082065135, "learning_rate": 8.225726876173321e-05, "loss": 2.9209, "step": 16872 }, { "epoch": 1.047426904215035, "grad_norm": 0.4811271715092881, "learning_rate": 8.225450925698808e-05, "loss": 2.8432, "step": 16873 }, { "epoch": 1.047488981314793, "grad_norm": 0.49814549385999335, "learning_rate": 8.225174958396335e-05, "loss": 2.8705, "step": 16874 }, { "epoch": 1.0475510584145509, "grad_norm": 0.4774020882028265, "learning_rate": 8.22489897426734e-05, "loss": 2.9085, "step": 16875 }, { "epoch": 1.0476131355143088, "grad_norm": 0.5461869134416365, "learning_rate": 8.224622973313263e-05, "loss": 2.8387, "step": 16876 }, { "epoch": 1.0476752126140667, "grad_norm": 0.4967666927790874, "learning_rate": 8.224346955535549e-05, "loss": 2.839, "step": 16877 }, { "epoch": 1.0477372897138246, "grad_norm": 0.5088320840230266, "learning_rate": 8.224070920935631e-05, "loss": 2.9834, "step": 16878 }, { "epoch": 1.0477993668135825, "grad_norm": 0.5356308718863578, "learning_rate": 8.223794869514953e-05, "loss": 2.909, "step": 16879 }, { "epoch": 1.0478614439133405, "grad_norm": 0.5332697173513751, "learning_rate": 8.223518801274955e-05, "loss": 2.9096, "step": 16880 }, { "epoch": 1.0479235210130984, "grad_norm": 0.4802569721218232, "learning_rate": 8.223242716217077e-05, "loss": 2.8405, "step": 16881 }, { "epoch": 1.047985598112856, "grad_norm": 0.5667942165362362, "learning_rate": 8.222966614342758e-05, "loss": 2.8524, "step": 16882 }, { "epoch": 1.048047675212614, "grad_norm": 0.552064415469626, "learning_rate": 8.222690495653443e-05, "loss": 2.8415, "step": 16883 }, { "epoch": 1.048109752312372, "grad_norm": 0.5302598364896504, "learning_rate": 8.222414360150565e-05, "loss": 2.8804, "step": 16884 }, { "epoch": 1.0481718294121298, "grad_norm": 0.5243602419607646, "learning_rate": 8.222138207835573e-05, "loss": 2.8794, "step": 16885 }, { "epoch": 1.0482339065118877, "grad_norm": 0.5305834403235882, "learning_rate": 8.221862038709901e-05, "loss": 2.9667, "step": 16886 }, { "epoch": 1.0482959836116457, "grad_norm": 0.5443546802481782, "learning_rate": 8.221585852774993e-05, "loss": 2.9048, "step": 16887 }, { "epoch": 1.0483580607114036, "grad_norm": 0.5713967767878902, "learning_rate": 8.22130965003229e-05, "loss": 2.875, "step": 16888 }, { "epoch": 1.0484201378111615, "grad_norm": 0.5524337274409847, "learning_rate": 8.221033430483234e-05, "loss": 2.8539, "step": 16889 }, { "epoch": 1.0484822149109194, "grad_norm": 0.5626937214737193, "learning_rate": 8.220757194129263e-05, "loss": 2.8929, "step": 16890 }, { "epoch": 1.0485442920106773, "grad_norm": 0.5179041573365901, "learning_rate": 8.220480940971819e-05, "loss": 2.9303, "step": 16891 }, { "epoch": 1.0486063691104353, "grad_norm": 0.5643459601128495, "learning_rate": 8.220204671012346e-05, "loss": 2.9358, "step": 16892 }, { "epoch": 1.048668446210193, "grad_norm": 0.4798214125953934, "learning_rate": 8.219928384252283e-05, "loss": 2.9857, "step": 16893 }, { "epoch": 1.0487305233099509, "grad_norm": 0.5340188741527353, "learning_rate": 8.219652080693071e-05, "loss": 2.8694, "step": 16894 }, { "epoch": 1.0487926004097088, "grad_norm": 0.49933499578246743, "learning_rate": 8.219375760336153e-05, "loss": 2.9598, "step": 16895 }, { "epoch": 1.0488546775094667, "grad_norm": 0.45632237291490524, "learning_rate": 8.21909942318297e-05, "loss": 2.9413, "step": 16896 }, { "epoch": 1.0489167546092246, "grad_norm": 0.5298578027938857, "learning_rate": 8.218823069234964e-05, "loss": 3.0152, "step": 16897 }, { "epoch": 1.0489788317089825, "grad_norm": 0.4879184137742546, "learning_rate": 8.218546698493576e-05, "loss": 2.9603, "step": 16898 }, { "epoch": 1.0490409088087405, "grad_norm": 0.5428261054952255, "learning_rate": 8.218270310960252e-05, "loss": 2.8708, "step": 16899 }, { "epoch": 1.0491029859084984, "grad_norm": 0.5675914334904821, "learning_rate": 8.217993906636427e-05, "loss": 2.9332, "step": 16900 }, { "epoch": 1.0491650630082563, "grad_norm": 0.5036588845309236, "learning_rate": 8.217717485523547e-05, "loss": 2.9191, "step": 16901 }, { "epoch": 1.0492271401080142, "grad_norm": 0.57589619393718, "learning_rate": 8.217441047623055e-05, "loss": 2.8616, "step": 16902 }, { "epoch": 1.0492892172077721, "grad_norm": 0.4502021640082452, "learning_rate": 8.217164592936391e-05, "loss": 2.8335, "step": 16903 }, { "epoch": 1.04935129430753, "grad_norm": 0.5284612894145736, "learning_rate": 8.216888121464999e-05, "loss": 2.9298, "step": 16904 }, { "epoch": 1.049413371407288, "grad_norm": 0.5105227697471015, "learning_rate": 8.216611633210321e-05, "loss": 2.8589, "step": 16905 }, { "epoch": 1.0494754485070457, "grad_norm": 0.47420412195007816, "learning_rate": 8.216335128173797e-05, "loss": 2.9309, "step": 16906 }, { "epoch": 1.0495375256068036, "grad_norm": 0.48339179448659125, "learning_rate": 8.216058606356873e-05, "loss": 2.8855, "step": 16907 }, { "epoch": 1.0495996027065615, "grad_norm": 0.47659508015314006, "learning_rate": 8.215782067760991e-05, "loss": 2.927, "step": 16908 }, { "epoch": 1.0496616798063194, "grad_norm": 0.49505678067962877, "learning_rate": 8.215505512387594e-05, "loss": 2.8263, "step": 16909 }, { "epoch": 1.0497237569060773, "grad_norm": 0.4644896657315626, "learning_rate": 8.215228940238123e-05, "loss": 2.8881, "step": 16910 }, { "epoch": 1.0497858340058352, "grad_norm": 0.43620213368972716, "learning_rate": 8.214952351314023e-05, "loss": 2.8935, "step": 16911 }, { "epoch": 1.0498479111055932, "grad_norm": 0.48748621493191946, "learning_rate": 8.214675745616735e-05, "loss": 2.8604, "step": 16912 }, { "epoch": 1.049909988205351, "grad_norm": 0.45354143277911196, "learning_rate": 8.214399123147704e-05, "loss": 2.9042, "step": 16913 }, { "epoch": 1.049972065305109, "grad_norm": 0.5426514061805799, "learning_rate": 8.214122483908372e-05, "loss": 2.9254, "step": 16914 }, { "epoch": 1.050034142404867, "grad_norm": 0.48631462783161666, "learning_rate": 8.213845827900184e-05, "loss": 2.8752, "step": 16915 }, { "epoch": 1.0500962195046248, "grad_norm": 0.5533267240627473, "learning_rate": 8.21356915512458e-05, "loss": 2.9407, "step": 16916 }, { "epoch": 1.0501582966043825, "grad_norm": 0.5146637539692634, "learning_rate": 8.213292465583007e-05, "loss": 2.8911, "step": 16917 }, { "epoch": 1.0502203737041405, "grad_norm": 0.5112818534553267, "learning_rate": 8.213015759276908e-05, "loss": 2.8159, "step": 16918 }, { "epoch": 1.0502824508038984, "grad_norm": 0.5466552292782587, "learning_rate": 8.212739036207724e-05, "loss": 2.9078, "step": 16919 }, { "epoch": 1.0503445279036563, "grad_norm": 0.48064076825630203, "learning_rate": 8.212462296376902e-05, "loss": 2.9073, "step": 16920 }, { "epoch": 1.0504066050034142, "grad_norm": 0.5853515747192504, "learning_rate": 8.212185539785884e-05, "loss": 2.9012, "step": 16921 }, { "epoch": 1.0504686821031721, "grad_norm": 0.5298061946787551, "learning_rate": 8.211908766436114e-05, "loss": 2.7645, "step": 16922 }, { "epoch": 1.05053075920293, "grad_norm": 0.4597572637039148, "learning_rate": 8.211631976329035e-05, "loss": 2.8816, "step": 16923 }, { "epoch": 1.050592836302688, "grad_norm": 0.5541144083474996, "learning_rate": 8.211355169466093e-05, "loss": 2.9803, "step": 16924 }, { "epoch": 1.0506549134024459, "grad_norm": 0.4701066095598773, "learning_rate": 8.211078345848733e-05, "loss": 2.7728, "step": 16925 }, { "epoch": 1.0507169905022038, "grad_norm": 0.5650424454789169, "learning_rate": 8.210801505478396e-05, "loss": 2.9168, "step": 16926 }, { "epoch": 1.0507790676019617, "grad_norm": 0.4507125068226342, "learning_rate": 8.210524648356528e-05, "loss": 2.9055, "step": 16927 }, { "epoch": 1.0508411447017196, "grad_norm": 0.5796181495908305, "learning_rate": 8.210247774484574e-05, "loss": 2.8953, "step": 16928 }, { "epoch": 1.0509032218014773, "grad_norm": 0.47207599679083795, "learning_rate": 8.20997088386398e-05, "loss": 2.8739, "step": 16929 }, { "epoch": 1.0509652989012352, "grad_norm": 0.5161539889967026, "learning_rate": 8.209693976496186e-05, "loss": 2.8652, "step": 16930 }, { "epoch": 1.0510273760009932, "grad_norm": 0.5111321151547892, "learning_rate": 8.20941705238264e-05, "loss": 2.9044, "step": 16931 }, { "epoch": 1.051089453100751, "grad_norm": 0.5103881367345836, "learning_rate": 8.209140111524787e-05, "loss": 2.9057, "step": 16932 }, { "epoch": 1.051151530200509, "grad_norm": 0.5481025542444861, "learning_rate": 8.208863153924071e-05, "loss": 2.9427, "step": 16933 }, { "epoch": 1.051213607300267, "grad_norm": 0.5497962314097721, "learning_rate": 8.208586179581937e-05, "loss": 2.8937, "step": 16934 }, { "epoch": 1.0512756844000248, "grad_norm": 0.5227775683820933, "learning_rate": 8.208309188499827e-05, "loss": 2.9719, "step": 16935 }, { "epoch": 1.0513377614997828, "grad_norm": 0.5509492063441672, "learning_rate": 8.208032180679192e-05, "loss": 2.8981, "step": 16936 }, { "epoch": 1.0513998385995407, "grad_norm": 0.4908350118431359, "learning_rate": 8.207755156121474e-05, "loss": 2.8563, "step": 16937 }, { "epoch": 1.0514619156992986, "grad_norm": 0.5168036081995651, "learning_rate": 8.20747811482812e-05, "loss": 2.9019, "step": 16938 }, { "epoch": 1.0515239927990565, "grad_norm": 0.530960789664197, "learning_rate": 8.207201056800572e-05, "loss": 2.8219, "step": 16939 }, { "epoch": 1.0515860698988144, "grad_norm": 0.548227300329678, "learning_rate": 8.206923982040277e-05, "loss": 2.9225, "step": 16940 }, { "epoch": 1.0516481469985721, "grad_norm": 0.5034897435818987, "learning_rate": 8.206646890548684e-05, "loss": 2.9092, "step": 16941 }, { "epoch": 1.05171022409833, "grad_norm": 0.5410203899077314, "learning_rate": 8.206369782327233e-05, "loss": 2.7775, "step": 16942 }, { "epoch": 1.051772301198088, "grad_norm": 0.4972447387192577, "learning_rate": 8.206092657377372e-05, "loss": 2.925, "step": 16943 }, { "epoch": 1.0518343782978459, "grad_norm": 0.5344743770081671, "learning_rate": 8.20581551570055e-05, "loss": 2.9251, "step": 16944 }, { "epoch": 1.0518964553976038, "grad_norm": 0.5007616738526522, "learning_rate": 8.205538357298208e-05, "loss": 2.8336, "step": 16945 }, { "epoch": 1.0519585324973617, "grad_norm": 0.47865665459533274, "learning_rate": 8.205261182171796e-05, "loss": 2.8632, "step": 16946 }, { "epoch": 1.0520206095971196, "grad_norm": 0.4357476904781066, "learning_rate": 8.204983990322756e-05, "loss": 2.9219, "step": 16947 }, { "epoch": 1.0520826866968775, "grad_norm": 0.5442936328207154, "learning_rate": 8.204706781752538e-05, "loss": 2.932, "step": 16948 }, { "epoch": 1.0521447637966355, "grad_norm": 0.47610789644503065, "learning_rate": 8.204429556462588e-05, "loss": 2.9708, "step": 16949 }, { "epoch": 1.0522068408963934, "grad_norm": 0.5067573148277578, "learning_rate": 8.204152314454348e-05, "loss": 2.8816, "step": 16950 }, { "epoch": 1.0522689179961513, "grad_norm": 0.46622118704899307, "learning_rate": 8.203875055729271e-05, "loss": 2.8656, "step": 16951 }, { "epoch": 1.0523309950959092, "grad_norm": 0.5150995756381195, "learning_rate": 8.203597780288797e-05, "loss": 2.9643, "step": 16952 }, { "epoch": 1.052393072195667, "grad_norm": 0.4639471848422692, "learning_rate": 8.203320488134378e-05, "loss": 2.9178, "step": 16953 }, { "epoch": 1.0524551492954248, "grad_norm": 0.4907965001964689, "learning_rate": 8.203043179267457e-05, "loss": 2.9003, "step": 16954 }, { "epoch": 1.0525172263951827, "grad_norm": 0.5249956766586311, "learning_rate": 8.202765853689482e-05, "loss": 2.8964, "step": 16955 }, { "epoch": 1.0525793034949407, "grad_norm": 0.5251445264793088, "learning_rate": 8.202488511401901e-05, "loss": 2.9372, "step": 16956 }, { "epoch": 1.0526413805946986, "grad_norm": 0.5242608839405094, "learning_rate": 8.20221115240616e-05, "loss": 2.7755, "step": 16957 }, { "epoch": 1.0527034576944565, "grad_norm": 0.5202782215820979, "learning_rate": 8.201933776703705e-05, "loss": 2.8512, "step": 16958 }, { "epoch": 1.0527655347942144, "grad_norm": 0.5565176577015899, "learning_rate": 8.201656384295985e-05, "loss": 2.8648, "step": 16959 }, { "epoch": 1.0528276118939723, "grad_norm": 0.513705553466485, "learning_rate": 8.201378975184446e-05, "loss": 2.9285, "step": 16960 }, { "epoch": 1.0528896889937303, "grad_norm": 0.5334257003650142, "learning_rate": 8.201101549370536e-05, "loss": 2.9232, "step": 16961 }, { "epoch": 1.0529517660934882, "grad_norm": 0.5381816633477474, "learning_rate": 8.200824106855702e-05, "loss": 2.9052, "step": 16962 }, { "epoch": 1.053013843193246, "grad_norm": 0.5487004467568335, "learning_rate": 8.200546647641392e-05, "loss": 2.9277, "step": 16963 }, { "epoch": 1.053075920293004, "grad_norm": 0.5193846933261355, "learning_rate": 8.200269171729052e-05, "loss": 2.8907, "step": 16964 }, { "epoch": 1.0531379973927617, "grad_norm": 0.6310734105262162, "learning_rate": 8.199991679120133e-05, "loss": 2.8692, "step": 16965 }, { "epoch": 1.0532000744925196, "grad_norm": 0.5187447122511176, "learning_rate": 8.19971416981608e-05, "loss": 2.7892, "step": 16966 }, { "epoch": 1.0532621515922775, "grad_norm": 0.5130356594467115, "learning_rate": 8.199436643818339e-05, "loss": 2.8557, "step": 16967 }, { "epoch": 1.0533242286920355, "grad_norm": 0.5817950492698081, "learning_rate": 8.199159101128362e-05, "loss": 2.8892, "step": 16968 }, { "epoch": 1.0533863057917934, "grad_norm": 0.5402546567826498, "learning_rate": 8.198881541747596e-05, "loss": 2.7271, "step": 16969 }, { "epoch": 1.0534483828915513, "grad_norm": 0.4893265331378966, "learning_rate": 8.198603965677488e-05, "loss": 2.902, "step": 16970 }, { "epoch": 1.0535104599913092, "grad_norm": 0.6441838802251861, "learning_rate": 8.198326372919487e-05, "loss": 2.8749, "step": 16971 }, { "epoch": 1.0535725370910671, "grad_norm": 0.5467263428410736, "learning_rate": 8.19804876347504e-05, "loss": 2.9362, "step": 16972 }, { "epoch": 1.053634614190825, "grad_norm": 0.5530970701073494, "learning_rate": 8.197771137345596e-05, "loss": 2.9009, "step": 16973 }, { "epoch": 1.053696691290583, "grad_norm": 0.6175007809699676, "learning_rate": 8.197493494532605e-05, "loss": 2.8712, "step": 16974 }, { "epoch": 1.0537587683903409, "grad_norm": 0.5404639891839083, "learning_rate": 8.197215835037512e-05, "loss": 2.8844, "step": 16975 }, { "epoch": 1.0538208454900988, "grad_norm": 0.6044954167342942, "learning_rate": 8.19693815886177e-05, "loss": 2.9127, "step": 16976 }, { "epoch": 1.0538829225898565, "grad_norm": 0.5970045288703294, "learning_rate": 8.196660466006825e-05, "loss": 2.9691, "step": 16977 }, { "epoch": 1.0539449996896144, "grad_norm": 0.5877020904261727, "learning_rate": 8.196382756474126e-05, "loss": 2.9093, "step": 16978 }, { "epoch": 1.0540070767893723, "grad_norm": 0.5710617265189837, "learning_rate": 8.196105030265121e-05, "loss": 2.9122, "step": 16979 }, { "epoch": 1.0540691538891303, "grad_norm": 0.5630021767667356, "learning_rate": 8.195827287381263e-05, "loss": 2.9672, "step": 16980 }, { "epoch": 1.0541312309888882, "grad_norm": 0.5583002613246032, "learning_rate": 8.195549527823998e-05, "loss": 2.7866, "step": 16981 }, { "epoch": 1.054193308088646, "grad_norm": 0.5715966334091653, "learning_rate": 8.195271751594773e-05, "loss": 2.9821, "step": 16982 }, { "epoch": 1.054255385188404, "grad_norm": 0.5695355492365487, "learning_rate": 8.194993958695041e-05, "loss": 2.8602, "step": 16983 }, { "epoch": 1.054317462288162, "grad_norm": 0.5908126200025431, "learning_rate": 8.19471614912625e-05, "loss": 2.9404, "step": 16984 }, { "epoch": 1.0543795393879198, "grad_norm": 0.6257049399721226, "learning_rate": 8.194438322889848e-05, "loss": 2.9732, "step": 16985 }, { "epoch": 1.0544416164876778, "grad_norm": 0.5219473908521505, "learning_rate": 8.194160479987287e-05, "loss": 2.852, "step": 16986 }, { "epoch": 1.0545036935874357, "grad_norm": 0.6477028235242752, "learning_rate": 8.193882620420016e-05, "loss": 2.9199, "step": 16987 }, { "epoch": 1.0545657706871936, "grad_norm": 0.5242633851743361, "learning_rate": 8.193604744189482e-05, "loss": 2.9779, "step": 16988 }, { "epoch": 1.0546278477869513, "grad_norm": 0.575799103748635, "learning_rate": 8.193326851297138e-05, "loss": 2.9474, "step": 16989 }, { "epoch": 1.0546899248867092, "grad_norm": 0.5968141045388511, "learning_rate": 8.193048941744434e-05, "loss": 2.9707, "step": 16990 }, { "epoch": 1.0547520019864671, "grad_norm": 0.5143706931284999, "learning_rate": 8.192771015532817e-05, "loss": 2.9188, "step": 16991 }, { "epoch": 1.054814079086225, "grad_norm": 0.5680251039780171, "learning_rate": 8.192493072663739e-05, "loss": 2.9333, "step": 16992 }, { "epoch": 1.054876156185983, "grad_norm": 0.514374980625231, "learning_rate": 8.192215113138648e-05, "loss": 2.8963, "step": 16993 }, { "epoch": 1.0549382332857409, "grad_norm": 0.48203206239999724, "learning_rate": 8.191937136958999e-05, "loss": 2.855, "step": 16994 }, { "epoch": 1.0550003103854988, "grad_norm": 0.5014839205878737, "learning_rate": 8.191659144126236e-05, "loss": 2.9273, "step": 16995 }, { "epoch": 1.0550623874852567, "grad_norm": 0.4711303671401837, "learning_rate": 8.191381134641814e-05, "loss": 2.8246, "step": 16996 }, { "epoch": 1.0551244645850146, "grad_norm": 0.5343079630870746, "learning_rate": 8.191103108507181e-05, "loss": 2.7886, "step": 16997 }, { "epoch": 1.0551865416847725, "grad_norm": 0.4730402277409347, "learning_rate": 8.190825065723789e-05, "loss": 2.7962, "step": 16998 }, { "epoch": 1.0552486187845305, "grad_norm": 0.4740124311768648, "learning_rate": 8.190547006293088e-05, "loss": 2.8382, "step": 16999 }, { "epoch": 1.0553106958842884, "grad_norm": 0.49121782362492805, "learning_rate": 8.190268930216529e-05, "loss": 2.8588, "step": 17000 }, { "epoch": 1.055372772984046, "grad_norm": 0.5266426229219747, "learning_rate": 8.189990837495563e-05, "loss": 2.9904, "step": 17001 }, { "epoch": 1.055434850083804, "grad_norm": 0.47664502476811016, "learning_rate": 8.189712728131639e-05, "loss": 2.881, "step": 17002 }, { "epoch": 1.055496927183562, "grad_norm": 0.48340839414264686, "learning_rate": 8.18943460212621e-05, "loss": 2.8958, "step": 17003 }, { "epoch": 1.0555590042833198, "grad_norm": 0.4830305298531295, "learning_rate": 8.189156459480727e-05, "loss": 3.0373, "step": 17004 }, { "epoch": 1.0556210813830778, "grad_norm": 0.5242616513657934, "learning_rate": 8.188878300196639e-05, "loss": 2.7834, "step": 17005 }, { "epoch": 1.0556831584828357, "grad_norm": 0.4848320558141455, "learning_rate": 8.1886001242754e-05, "loss": 2.8867, "step": 17006 }, { "epoch": 1.0557452355825936, "grad_norm": 0.5508717401718383, "learning_rate": 8.188321931718459e-05, "loss": 2.8075, "step": 17007 }, { "epoch": 1.0558073126823515, "grad_norm": 0.49558568454584734, "learning_rate": 8.18804372252727e-05, "loss": 2.9159, "step": 17008 }, { "epoch": 1.0558693897821094, "grad_norm": 0.5611661620276123, "learning_rate": 8.187765496703281e-05, "loss": 2.825, "step": 17009 }, { "epoch": 1.0559314668818673, "grad_norm": 0.5130599115143532, "learning_rate": 8.187487254247948e-05, "loss": 2.975, "step": 17010 }, { "epoch": 1.0559935439816253, "grad_norm": 0.5331160365935408, "learning_rate": 8.187208995162718e-05, "loss": 2.8838, "step": 17011 }, { "epoch": 1.0560556210813832, "grad_norm": 0.5194959126551246, "learning_rate": 8.186930719449047e-05, "loss": 2.8731, "step": 17012 }, { "epoch": 1.0561176981811409, "grad_norm": 0.47682707899503723, "learning_rate": 8.186652427108383e-05, "loss": 2.9169, "step": 17013 }, { "epoch": 1.0561797752808988, "grad_norm": 0.4912876198222636, "learning_rate": 8.186374118142182e-05, "loss": 2.8305, "step": 17014 }, { "epoch": 1.0562418523806567, "grad_norm": 0.4962731284894279, "learning_rate": 8.186095792551892e-05, "loss": 2.9658, "step": 17015 }, { "epoch": 1.0563039294804146, "grad_norm": 0.5223569392148949, "learning_rate": 8.185817450338966e-05, "loss": 2.9257, "step": 17016 }, { "epoch": 1.0563660065801725, "grad_norm": 0.502085746136747, "learning_rate": 8.185539091504859e-05, "loss": 2.9523, "step": 17017 }, { "epoch": 1.0564280836799305, "grad_norm": 0.5303564129678053, "learning_rate": 8.18526071605102e-05, "loss": 2.9247, "step": 17018 }, { "epoch": 1.0564901607796884, "grad_norm": 0.4839311535062059, "learning_rate": 8.184982323978902e-05, "loss": 2.9051, "step": 17019 }, { "epoch": 1.0565522378794463, "grad_norm": 0.5716850277905732, "learning_rate": 8.18470391528996e-05, "loss": 2.961, "step": 17020 }, { "epoch": 1.0566143149792042, "grad_norm": 0.5172299635589288, "learning_rate": 8.184425489985643e-05, "loss": 2.937, "step": 17021 }, { "epoch": 1.0566763920789621, "grad_norm": 0.46272092839285567, "learning_rate": 8.184147048067405e-05, "loss": 2.7957, "step": 17022 }, { "epoch": 1.05673846917872, "grad_norm": 0.527753120049922, "learning_rate": 8.1838685895367e-05, "loss": 2.9211, "step": 17023 }, { "epoch": 1.056800546278478, "grad_norm": 0.4617276450654206, "learning_rate": 8.183590114394979e-05, "loss": 2.8227, "step": 17024 }, { "epoch": 1.0568626233782357, "grad_norm": 0.4884919894351332, "learning_rate": 8.183311622643696e-05, "loss": 3.013, "step": 17025 }, { "epoch": 1.0569247004779936, "grad_norm": 0.49876395569304455, "learning_rate": 8.183033114284303e-05, "loss": 2.8173, "step": 17026 }, { "epoch": 1.0569867775777515, "grad_norm": 0.5327142563352484, "learning_rate": 8.182754589318253e-05, "loss": 2.9418, "step": 17027 }, { "epoch": 1.0570488546775094, "grad_norm": 0.47878724777501463, "learning_rate": 8.182476047747001e-05, "loss": 2.7807, "step": 17028 }, { "epoch": 1.0571109317772673, "grad_norm": 0.525317729445099, "learning_rate": 8.182197489571998e-05, "loss": 2.908, "step": 17029 }, { "epoch": 1.0571730088770253, "grad_norm": 0.49069002716750837, "learning_rate": 8.181918914794699e-05, "loss": 2.8553, "step": 17030 }, { "epoch": 1.0572350859767832, "grad_norm": 0.5542607344713618, "learning_rate": 8.181640323416554e-05, "loss": 2.8975, "step": 17031 }, { "epoch": 1.057297163076541, "grad_norm": 0.44543629314312105, "learning_rate": 8.181361715439023e-05, "loss": 2.8894, "step": 17032 }, { "epoch": 1.057359240176299, "grad_norm": 0.5074936085566301, "learning_rate": 8.181083090863553e-05, "loss": 2.8129, "step": 17033 }, { "epoch": 1.057421317276057, "grad_norm": 0.4852233502461125, "learning_rate": 8.180804449691601e-05, "loss": 2.8817, "step": 17034 }, { "epoch": 1.0574833943758148, "grad_norm": 0.489419784689039, "learning_rate": 8.18052579192462e-05, "loss": 2.9232, "step": 17035 }, { "epoch": 1.0575454714755728, "grad_norm": 0.458096835141053, "learning_rate": 8.180247117564063e-05, "loss": 2.9157, "step": 17036 }, { "epoch": 1.0576075485753305, "grad_norm": 0.5238676154422413, "learning_rate": 8.179968426611386e-05, "loss": 2.8507, "step": 17037 }, { "epoch": 1.0576696256750884, "grad_norm": 0.4922463820824832, "learning_rate": 8.179689719068042e-05, "loss": 2.8676, "step": 17038 }, { "epoch": 1.0577317027748463, "grad_norm": 0.45338264901245656, "learning_rate": 8.179410994935483e-05, "loss": 2.852, "step": 17039 }, { "epoch": 1.0577937798746042, "grad_norm": 0.5447355382486884, "learning_rate": 8.179132254215167e-05, "loss": 2.904, "step": 17040 }, { "epoch": 1.0578558569743621, "grad_norm": 0.43677351442741286, "learning_rate": 8.178853496908545e-05, "loss": 2.8873, "step": 17041 }, { "epoch": 1.05791793407412, "grad_norm": 0.5880239088475293, "learning_rate": 8.178574723017073e-05, "loss": 2.8911, "step": 17042 }, { "epoch": 1.057980011173878, "grad_norm": 0.5309165581338979, "learning_rate": 8.178295932542206e-05, "loss": 2.8429, "step": 17043 }, { "epoch": 1.0580420882736359, "grad_norm": 0.5422104679486701, "learning_rate": 8.178017125485396e-05, "loss": 2.8922, "step": 17044 }, { "epoch": 1.0581041653733938, "grad_norm": 0.5293662669734929, "learning_rate": 8.1777383018481e-05, "loss": 2.9702, "step": 17045 }, { "epoch": 1.0581662424731517, "grad_norm": 0.5106915363119112, "learning_rate": 8.177459461631771e-05, "loss": 2.7755, "step": 17046 }, { "epoch": 1.0582283195729096, "grad_norm": 0.5241648616466565, "learning_rate": 8.177180604837866e-05, "loss": 2.9152, "step": 17047 }, { "epoch": 1.0582903966726676, "grad_norm": 0.5570487951910229, "learning_rate": 8.176901731467839e-05, "loss": 2.8145, "step": 17048 }, { "epoch": 1.0583524737724253, "grad_norm": 0.49798607548622664, "learning_rate": 8.176622841523142e-05, "loss": 2.9091, "step": 17049 }, { "epoch": 1.0584145508721832, "grad_norm": 0.5079704039167445, "learning_rate": 8.176343935005233e-05, "loss": 2.9128, "step": 17050 }, { "epoch": 1.058476627971941, "grad_norm": 0.5220554953376602, "learning_rate": 8.176065011915568e-05, "loss": 2.7432, "step": 17051 }, { "epoch": 1.058538705071699, "grad_norm": 0.4852410541812369, "learning_rate": 8.1757860722556e-05, "loss": 2.9418, "step": 17052 }, { "epoch": 1.058600782171457, "grad_norm": 0.5360965681766361, "learning_rate": 8.175507116026786e-05, "loss": 2.8915, "step": 17053 }, { "epoch": 1.0586628592712148, "grad_norm": 0.5417485969887923, "learning_rate": 8.17522814323058e-05, "loss": 2.8975, "step": 17054 }, { "epoch": 1.0587249363709728, "grad_norm": 0.5267607114922139, "learning_rate": 8.174949153868438e-05, "loss": 2.8159, "step": 17055 }, { "epoch": 1.0587870134707307, "grad_norm": 0.5066254753959528, "learning_rate": 8.174670147941816e-05, "loss": 2.8958, "step": 17056 }, { "epoch": 1.0588490905704886, "grad_norm": 0.4985303290093255, "learning_rate": 8.174391125452168e-05, "loss": 2.8892, "step": 17057 }, { "epoch": 1.0589111676702465, "grad_norm": 0.5282255359189624, "learning_rate": 8.17411208640095e-05, "loss": 2.8494, "step": 17058 }, { "epoch": 1.0589732447700044, "grad_norm": 0.5168203024192761, "learning_rate": 8.17383303078962e-05, "loss": 2.8926, "step": 17059 }, { "epoch": 1.0590353218697623, "grad_norm": 0.45167564571729507, "learning_rate": 8.173553958619632e-05, "loss": 2.8828, "step": 17060 }, { "epoch": 1.05909739896952, "grad_norm": 0.5576190848243088, "learning_rate": 8.173274869892444e-05, "loss": 2.8637, "step": 17061 }, { "epoch": 1.059159476069278, "grad_norm": 0.5086469103897924, "learning_rate": 8.17299576460951e-05, "loss": 2.8519, "step": 17062 }, { "epoch": 1.0592215531690359, "grad_norm": 0.5226618015239662, "learning_rate": 8.172716642772287e-05, "loss": 2.8991, "step": 17063 }, { "epoch": 1.0592836302687938, "grad_norm": 0.54715239437827, "learning_rate": 8.172437504382231e-05, "loss": 2.8945, "step": 17064 }, { "epoch": 1.0593457073685517, "grad_norm": 0.5403567547211916, "learning_rate": 8.172158349440797e-05, "loss": 2.8753, "step": 17065 }, { "epoch": 1.0594077844683096, "grad_norm": 0.4952929333453239, "learning_rate": 8.171879177949445e-05, "loss": 2.8525, "step": 17066 }, { "epoch": 1.0594698615680676, "grad_norm": 0.5207372449612135, "learning_rate": 8.171599989909629e-05, "loss": 2.8992, "step": 17067 }, { "epoch": 1.0595319386678255, "grad_norm": 0.4724870962318587, "learning_rate": 8.171320785322805e-05, "loss": 2.8579, "step": 17068 }, { "epoch": 1.0595940157675834, "grad_norm": 0.4931813152052572, "learning_rate": 8.17104156419043e-05, "loss": 2.8828, "step": 17069 }, { "epoch": 1.0596560928673413, "grad_norm": 0.5210485458784218, "learning_rate": 8.170762326513964e-05, "loss": 2.9001, "step": 17070 }, { "epoch": 1.0597181699670992, "grad_norm": 0.5010330733463142, "learning_rate": 8.170483072294858e-05, "loss": 2.9432, "step": 17071 }, { "epoch": 1.0597802470668571, "grad_norm": 0.5166016201891804, "learning_rate": 8.170203801534573e-05, "loss": 2.9227, "step": 17072 }, { "epoch": 1.0598423241666148, "grad_norm": 0.5209302780209121, "learning_rate": 8.169924514234566e-05, "loss": 2.9253, "step": 17073 }, { "epoch": 1.0599044012663728, "grad_norm": 0.5175650828895569, "learning_rate": 8.169645210396293e-05, "loss": 2.8999, "step": 17074 }, { "epoch": 1.0599664783661307, "grad_norm": 0.5714339634947462, "learning_rate": 8.169365890021211e-05, "loss": 2.9048, "step": 17075 }, { "epoch": 1.0600285554658886, "grad_norm": 0.49769151763271646, "learning_rate": 8.169086553110779e-05, "loss": 2.9215, "step": 17076 }, { "epoch": 1.0600906325656465, "grad_norm": 0.5728161868212437, "learning_rate": 8.168807199666451e-05, "loss": 2.8737, "step": 17077 }, { "epoch": 1.0601527096654044, "grad_norm": 0.5749830492257774, "learning_rate": 8.16852782968969e-05, "loss": 2.9322, "step": 17078 }, { "epoch": 1.0602147867651623, "grad_norm": 0.5071784829908821, "learning_rate": 8.168248443181947e-05, "loss": 2.8327, "step": 17079 }, { "epoch": 1.0602768638649203, "grad_norm": 0.5495250254894869, "learning_rate": 8.167969040144686e-05, "loss": 2.9123, "step": 17080 }, { "epoch": 1.0603389409646782, "grad_norm": 0.6080652491040487, "learning_rate": 8.167689620579357e-05, "loss": 2.9622, "step": 17081 }, { "epoch": 1.060401018064436, "grad_norm": 0.5297388293846925, "learning_rate": 8.167410184487425e-05, "loss": 2.9034, "step": 17082 }, { "epoch": 1.060463095164194, "grad_norm": 0.6060090146448657, "learning_rate": 8.167130731870346e-05, "loss": 2.7582, "step": 17083 }, { "epoch": 1.060525172263952, "grad_norm": 0.4850166132136416, "learning_rate": 8.166851262729576e-05, "loss": 2.7828, "step": 17084 }, { "epoch": 1.0605872493637096, "grad_norm": 0.5180317168706859, "learning_rate": 8.166571777066572e-05, "loss": 2.9396, "step": 17085 }, { "epoch": 1.0606493264634675, "grad_norm": 0.49702387444816887, "learning_rate": 8.166292274882797e-05, "loss": 3.0007, "step": 17086 }, { "epoch": 1.0607114035632255, "grad_norm": 0.513441926376385, "learning_rate": 8.166012756179706e-05, "loss": 2.8676, "step": 17087 }, { "epoch": 1.0607734806629834, "grad_norm": 0.5313225303953283, "learning_rate": 8.165733220958758e-05, "loss": 2.9105, "step": 17088 }, { "epoch": 1.0608355577627413, "grad_norm": 0.5528932754131815, "learning_rate": 8.165453669221411e-05, "loss": 2.9876, "step": 17089 }, { "epoch": 1.0608976348624992, "grad_norm": 0.5262573920340128, "learning_rate": 8.165174100969124e-05, "loss": 2.8955, "step": 17090 }, { "epoch": 1.0609597119622571, "grad_norm": 0.5018840400860237, "learning_rate": 8.164894516203355e-05, "loss": 2.8478, "step": 17091 }, { "epoch": 1.061021789062015, "grad_norm": 0.5017910112955679, "learning_rate": 8.164614914925563e-05, "loss": 2.9312, "step": 17092 }, { "epoch": 1.061083866161773, "grad_norm": 0.4520831132996664, "learning_rate": 8.164335297137206e-05, "loss": 2.7968, "step": 17093 }, { "epoch": 1.0611459432615309, "grad_norm": 0.5204864841120818, "learning_rate": 8.164055662839745e-05, "loss": 2.8495, "step": 17094 }, { "epoch": 1.0612080203612888, "grad_norm": 0.5034501428987893, "learning_rate": 8.163776012034638e-05, "loss": 2.9113, "step": 17095 }, { "epoch": 1.0612700974610467, "grad_norm": 0.471554551989779, "learning_rate": 8.163496344723342e-05, "loss": 2.76, "step": 17096 }, { "epoch": 1.0613321745608044, "grad_norm": 0.5074877067072957, "learning_rate": 8.163216660907318e-05, "loss": 2.936, "step": 17097 }, { "epoch": 1.0613942516605623, "grad_norm": 0.5028201974738337, "learning_rate": 8.162936960588026e-05, "loss": 2.8115, "step": 17098 }, { "epoch": 1.0614563287603203, "grad_norm": 0.45185867483214787, "learning_rate": 8.162657243766923e-05, "loss": 2.8688, "step": 17099 }, { "epoch": 1.0615184058600782, "grad_norm": 0.4785618195326522, "learning_rate": 8.162377510445469e-05, "loss": 2.8403, "step": 17100 }, { "epoch": 1.061580482959836, "grad_norm": 0.4582167842094121, "learning_rate": 8.162097760625125e-05, "loss": 2.8903, "step": 17101 }, { "epoch": 1.061642560059594, "grad_norm": 0.47950870462159934, "learning_rate": 8.161817994307348e-05, "loss": 2.8547, "step": 17102 }, { "epoch": 1.061704637159352, "grad_norm": 0.4817681303893551, "learning_rate": 8.1615382114936e-05, "loss": 2.874, "step": 17103 }, { "epoch": 1.0617667142591098, "grad_norm": 0.45749669949325966, "learning_rate": 8.161258412185339e-05, "loss": 2.8637, "step": 17104 }, { "epoch": 1.0618287913588678, "grad_norm": 0.4960649587131245, "learning_rate": 8.160978596384026e-05, "loss": 2.8449, "step": 17105 }, { "epoch": 1.0618908684586257, "grad_norm": 0.4891389561569681, "learning_rate": 8.16069876409112e-05, "loss": 2.8282, "step": 17106 }, { "epoch": 1.0619529455583836, "grad_norm": 0.5136384264132724, "learning_rate": 8.16041891530808e-05, "loss": 3.0509, "step": 17107 }, { "epoch": 1.0620150226581415, "grad_norm": 0.5564442338327147, "learning_rate": 8.160139050036369e-05, "loss": 2.8829, "step": 17108 }, { "epoch": 1.0620770997578992, "grad_norm": 0.5257189267363481, "learning_rate": 8.159859168277444e-05, "loss": 2.8624, "step": 17109 }, { "epoch": 1.0621391768576571, "grad_norm": 0.5496250489256375, "learning_rate": 8.159579270032768e-05, "loss": 2.8271, "step": 17110 }, { "epoch": 1.062201253957415, "grad_norm": 0.5676463790358729, "learning_rate": 8.159299355303799e-05, "loss": 2.8919, "step": 17111 }, { "epoch": 1.062263331057173, "grad_norm": 0.5260402534642137, "learning_rate": 8.159019424091999e-05, "loss": 3.0115, "step": 17112 }, { "epoch": 1.0623254081569309, "grad_norm": 0.4664126295103906, "learning_rate": 8.158739476398828e-05, "loss": 3.0475, "step": 17113 }, { "epoch": 1.0623874852566888, "grad_norm": 0.5068611623848414, "learning_rate": 8.158459512225746e-05, "loss": 2.9227, "step": 17114 }, { "epoch": 1.0624495623564467, "grad_norm": 0.5134887078649171, "learning_rate": 8.158179531574214e-05, "loss": 2.8202, "step": 17115 }, { "epoch": 1.0625116394562046, "grad_norm": 0.5013352090038727, "learning_rate": 8.157899534445693e-05, "loss": 2.9431, "step": 17116 }, { "epoch": 1.0625737165559626, "grad_norm": 0.5136980985612601, "learning_rate": 8.157619520841644e-05, "loss": 2.9133, "step": 17117 }, { "epoch": 1.0626357936557205, "grad_norm": 0.5043548834285864, "learning_rate": 8.157339490763527e-05, "loss": 2.8345, "step": 17118 }, { "epoch": 1.0626978707554784, "grad_norm": 0.47554919905805154, "learning_rate": 8.157059444212803e-05, "loss": 2.9556, "step": 17119 }, { "epoch": 1.0627599478552363, "grad_norm": 0.5615797727769809, "learning_rate": 8.156779381190933e-05, "loss": 2.8547, "step": 17120 }, { "epoch": 1.062822024954994, "grad_norm": 0.5554933470096706, "learning_rate": 8.156499301699378e-05, "loss": 2.9365, "step": 17121 }, { "epoch": 1.062884102054752, "grad_norm": 0.50721132933459, "learning_rate": 8.156219205739602e-05, "loss": 2.8318, "step": 17122 }, { "epoch": 1.0629461791545098, "grad_norm": 0.5307725556082339, "learning_rate": 8.155939093313063e-05, "loss": 2.9712, "step": 17123 }, { "epoch": 1.0630082562542678, "grad_norm": 0.5750235283225742, "learning_rate": 8.155658964421222e-05, "loss": 2.8872, "step": 17124 }, { "epoch": 1.0630703333540257, "grad_norm": 0.4741856131374699, "learning_rate": 8.155378819065544e-05, "loss": 2.882, "step": 17125 }, { "epoch": 1.0631324104537836, "grad_norm": 0.4769014185215848, "learning_rate": 8.15509865724749e-05, "loss": 2.9039, "step": 17126 }, { "epoch": 1.0631944875535415, "grad_norm": 0.5010163289548044, "learning_rate": 8.154818478968518e-05, "loss": 2.9735, "step": 17127 }, { "epoch": 1.0632565646532994, "grad_norm": 0.5018979944018422, "learning_rate": 8.154538284230093e-05, "loss": 2.8488, "step": 17128 }, { "epoch": 1.0633186417530573, "grad_norm": 0.5251821792225874, "learning_rate": 8.154258073033676e-05, "loss": 2.9312, "step": 17129 }, { "epoch": 1.0633807188528153, "grad_norm": 0.5172989290178039, "learning_rate": 8.15397784538073e-05, "loss": 2.9494, "step": 17130 }, { "epoch": 1.0634427959525732, "grad_norm": 0.4840619552294016, "learning_rate": 8.153697601272714e-05, "loss": 3.0329, "step": 17131 }, { "epoch": 1.0635048730523309, "grad_norm": 0.5017405850373423, "learning_rate": 8.153417340711093e-05, "loss": 2.852, "step": 17132 }, { "epoch": 1.0635669501520888, "grad_norm": 0.5244538134066472, "learning_rate": 8.153137063697329e-05, "loss": 2.8974, "step": 17133 }, { "epoch": 1.0636290272518467, "grad_norm": 0.519373762310587, "learning_rate": 8.152856770232883e-05, "loss": 2.9672, "step": 17134 }, { "epoch": 1.0636911043516046, "grad_norm": 0.5314264845816266, "learning_rate": 8.152576460319216e-05, "loss": 2.9341, "step": 17135 }, { "epoch": 1.0637531814513626, "grad_norm": 0.5062438799641228, "learning_rate": 8.152296133957795e-05, "loss": 2.8237, "step": 17136 }, { "epoch": 1.0638152585511205, "grad_norm": 0.5030892424582618, "learning_rate": 8.152015791150079e-05, "loss": 2.8956, "step": 17137 }, { "epoch": 1.0638773356508784, "grad_norm": 0.5708760651607111, "learning_rate": 8.151735431897531e-05, "loss": 2.904, "step": 17138 }, { "epoch": 1.0639394127506363, "grad_norm": 0.48000088063298346, "learning_rate": 8.151455056201615e-05, "loss": 2.9565, "step": 17139 }, { "epoch": 1.0640014898503942, "grad_norm": 0.5230142960718845, "learning_rate": 8.15117466406379e-05, "loss": 2.9085, "step": 17140 }, { "epoch": 1.0640635669501521, "grad_norm": 0.5464312524177682, "learning_rate": 8.150894255485524e-05, "loss": 2.8772, "step": 17141 }, { "epoch": 1.06412564404991, "grad_norm": 0.4767355057663982, "learning_rate": 8.150613830468279e-05, "loss": 2.9568, "step": 17142 }, { "epoch": 1.064187721149668, "grad_norm": 0.5150936162914067, "learning_rate": 8.150333389013515e-05, "loss": 2.8936, "step": 17143 }, { "epoch": 1.064249798249426, "grad_norm": 0.4908680562829184, "learning_rate": 8.150052931122699e-05, "loss": 2.8575, "step": 17144 }, { "epoch": 1.0643118753491836, "grad_norm": 0.4822847889711226, "learning_rate": 8.149772456797289e-05, "loss": 2.9355, "step": 17145 }, { "epoch": 1.0643739524489415, "grad_norm": 0.47359543485728783, "learning_rate": 8.149491966038753e-05, "loss": 2.8772, "step": 17146 }, { "epoch": 1.0644360295486994, "grad_norm": 0.455655333578037, "learning_rate": 8.149211458848554e-05, "loss": 2.884, "step": 17147 }, { "epoch": 1.0644981066484573, "grad_norm": 0.47490224710785506, "learning_rate": 8.148930935228154e-05, "loss": 2.9613, "step": 17148 }, { "epoch": 1.0645601837482153, "grad_norm": 0.47921609796338105, "learning_rate": 8.148650395179015e-05, "loss": 2.8632, "step": 17149 }, { "epoch": 1.0646222608479732, "grad_norm": 0.5665382757547157, "learning_rate": 8.148369838702603e-05, "loss": 2.9826, "step": 17150 }, { "epoch": 1.064684337947731, "grad_norm": 0.4985906736302305, "learning_rate": 8.148089265800383e-05, "loss": 2.8449, "step": 17151 }, { "epoch": 1.064746415047489, "grad_norm": 0.47486460851740275, "learning_rate": 8.147808676473816e-05, "loss": 2.8512, "step": 17152 }, { "epoch": 1.064808492147247, "grad_norm": 0.5075027988619741, "learning_rate": 8.147528070724367e-05, "loss": 2.9369, "step": 17153 }, { "epoch": 1.0648705692470049, "grad_norm": 0.5008534062573912, "learning_rate": 8.1472474485535e-05, "loss": 2.8777, "step": 17154 }, { "epoch": 1.0649326463467628, "grad_norm": 0.5152771383243282, "learning_rate": 8.14696680996268e-05, "loss": 2.8268, "step": 17155 }, { "epoch": 1.0649947234465205, "grad_norm": 0.5214023025450149, "learning_rate": 8.146686154953367e-05, "loss": 2.8537, "step": 17156 }, { "epoch": 1.0650568005462784, "grad_norm": 0.5616458659843525, "learning_rate": 8.146405483527031e-05, "loss": 2.795, "step": 17157 }, { "epoch": 1.0651188776460363, "grad_norm": 0.5675131144247663, "learning_rate": 8.146124795685133e-05, "loss": 2.926, "step": 17158 }, { "epoch": 1.0651809547457942, "grad_norm": 0.5400838160811632, "learning_rate": 8.14584409142914e-05, "loss": 2.8873, "step": 17159 }, { "epoch": 1.0652430318455521, "grad_norm": 0.6024193049640572, "learning_rate": 8.145563370760513e-05, "loss": 2.9577, "step": 17160 }, { "epoch": 1.06530510894531, "grad_norm": 0.5863267749869951, "learning_rate": 8.145282633680717e-05, "loss": 2.8904, "step": 17161 }, { "epoch": 1.065367186045068, "grad_norm": 0.5744497651928344, "learning_rate": 8.14500188019122e-05, "loss": 2.9702, "step": 17162 }, { "epoch": 1.0654292631448259, "grad_norm": 0.5428442229036702, "learning_rate": 8.144721110293485e-05, "loss": 2.9329, "step": 17163 }, { "epoch": 1.0654913402445838, "grad_norm": 0.5853105115067513, "learning_rate": 8.144440323988975e-05, "loss": 2.9524, "step": 17164 }, { "epoch": 1.0655534173443417, "grad_norm": 0.49873319721838766, "learning_rate": 8.144159521279158e-05, "loss": 2.8737, "step": 17165 }, { "epoch": 1.0656154944440996, "grad_norm": 0.5111938580044031, "learning_rate": 8.143878702165497e-05, "loss": 2.8937, "step": 17166 }, { "epoch": 1.0656775715438576, "grad_norm": 0.5620764621326887, "learning_rate": 8.143597866649457e-05, "loss": 2.9485, "step": 17167 }, { "epoch": 1.0657396486436155, "grad_norm": 0.48475888637190223, "learning_rate": 8.143317014732504e-05, "loss": 2.8658, "step": 17168 }, { "epoch": 1.0658017257433732, "grad_norm": 0.5142883546700893, "learning_rate": 8.143036146416104e-05, "loss": 2.8637, "step": 17169 }, { "epoch": 1.065863802843131, "grad_norm": 0.5151366030435297, "learning_rate": 8.14275526170172e-05, "loss": 2.8374, "step": 17170 }, { "epoch": 1.065925879942889, "grad_norm": 0.47843625714460186, "learning_rate": 8.142474360590821e-05, "loss": 2.9162, "step": 17171 }, { "epoch": 1.065987957042647, "grad_norm": 0.5128192860713983, "learning_rate": 8.142193443084869e-05, "loss": 2.8311, "step": 17172 }, { "epoch": 1.0660500341424048, "grad_norm": 0.45526814893402656, "learning_rate": 8.14191250918533e-05, "loss": 2.8019, "step": 17173 }, { "epoch": 1.0661121112421628, "grad_norm": 0.5423867653746313, "learning_rate": 8.141631558893672e-05, "loss": 2.8918, "step": 17174 }, { "epoch": 1.0661741883419207, "grad_norm": 0.5195161630834854, "learning_rate": 8.141350592211358e-05, "loss": 2.9643, "step": 17175 }, { "epoch": 1.0662362654416786, "grad_norm": 0.49226000418312327, "learning_rate": 8.141069609139857e-05, "loss": 2.8565, "step": 17176 }, { "epoch": 1.0662983425414365, "grad_norm": 0.4925506705151282, "learning_rate": 8.140788609680634e-05, "loss": 2.7855, "step": 17177 }, { "epoch": 1.0663604196411944, "grad_norm": 0.5236365096556816, "learning_rate": 8.140507593835152e-05, "loss": 2.8401, "step": 17178 }, { "epoch": 1.0664224967409524, "grad_norm": 0.5805715395457459, "learning_rate": 8.140226561604881e-05, "loss": 2.8672, "step": 17179 }, { "epoch": 1.06648457384071, "grad_norm": 0.463222207847134, "learning_rate": 8.139945512991285e-05, "loss": 2.8902, "step": 17180 }, { "epoch": 1.066546650940468, "grad_norm": 0.5639517910590429, "learning_rate": 8.139664447995832e-05, "loss": 2.8745, "step": 17181 }, { "epoch": 1.0666087280402259, "grad_norm": 0.5614534550055597, "learning_rate": 8.139383366619987e-05, "loss": 2.7987, "step": 17182 }, { "epoch": 1.0666708051399838, "grad_norm": 0.4762386722669319, "learning_rate": 8.139102268865218e-05, "loss": 2.8698, "step": 17183 }, { "epoch": 1.0667328822397417, "grad_norm": 0.5477059454688712, "learning_rate": 8.138821154732988e-05, "loss": 2.9547, "step": 17184 }, { "epoch": 1.0667949593394996, "grad_norm": 0.5382488301871735, "learning_rate": 8.138540024224768e-05, "loss": 2.8278, "step": 17185 }, { "epoch": 1.0668570364392576, "grad_norm": 0.46980776649963857, "learning_rate": 8.138258877342022e-05, "loss": 2.9238, "step": 17186 }, { "epoch": 1.0669191135390155, "grad_norm": 0.5730235578061578, "learning_rate": 8.137977714086217e-05, "loss": 2.8324, "step": 17187 }, { "epoch": 1.0669811906387734, "grad_norm": 0.4714266017102711, "learning_rate": 8.13769653445882e-05, "loss": 2.9117, "step": 17188 }, { "epoch": 1.0670432677385313, "grad_norm": 0.52167219001641, "learning_rate": 8.1374153384613e-05, "loss": 2.8678, "step": 17189 }, { "epoch": 1.0671053448382892, "grad_norm": 0.4466957726561855, "learning_rate": 8.137134126095122e-05, "loss": 2.908, "step": 17190 }, { "epoch": 1.0671674219380471, "grad_norm": 0.5085160590161317, "learning_rate": 8.136852897361754e-05, "loss": 2.8886, "step": 17191 }, { "epoch": 1.067229499037805, "grad_norm": 0.43466711787712825, "learning_rate": 8.136571652262664e-05, "loss": 2.7974, "step": 17192 }, { "epoch": 1.0672915761375628, "grad_norm": 0.43687781301637857, "learning_rate": 8.136290390799316e-05, "loss": 2.8394, "step": 17193 }, { "epoch": 1.0673536532373207, "grad_norm": 0.44294670830573696, "learning_rate": 8.13600911297318e-05, "loss": 2.8782, "step": 17194 }, { "epoch": 1.0674157303370786, "grad_norm": 0.46809274371340964, "learning_rate": 8.135727818785724e-05, "loss": 2.8087, "step": 17195 }, { "epoch": 1.0674778074368365, "grad_norm": 0.46917397081212747, "learning_rate": 8.135446508238415e-05, "loss": 2.904, "step": 17196 }, { "epoch": 1.0675398845365944, "grad_norm": 0.4820893102762774, "learning_rate": 8.13516518133272e-05, "loss": 2.8818, "step": 17197 }, { "epoch": 1.0676019616363523, "grad_norm": 0.5171830594889237, "learning_rate": 8.134883838070108e-05, "loss": 2.9621, "step": 17198 }, { "epoch": 1.0676640387361103, "grad_norm": 0.5295921431148076, "learning_rate": 8.134602478452044e-05, "loss": 2.9488, "step": 17199 }, { "epoch": 1.0677261158358682, "grad_norm": 0.46926217708020196, "learning_rate": 8.134321102480001e-05, "loss": 2.8713, "step": 17200 }, { "epoch": 1.067788192935626, "grad_norm": 0.5452115509245339, "learning_rate": 8.134039710155442e-05, "loss": 2.9259, "step": 17201 }, { "epoch": 1.067850270035384, "grad_norm": 0.49030551166307423, "learning_rate": 8.133758301479836e-05, "loss": 2.845, "step": 17202 }, { "epoch": 1.067912347135142, "grad_norm": 0.4707100737126089, "learning_rate": 8.133476876454653e-05, "loss": 2.9185, "step": 17203 }, { "epoch": 1.0679744242348996, "grad_norm": 0.4583787696887263, "learning_rate": 8.133195435081361e-05, "loss": 2.8341, "step": 17204 }, { "epoch": 1.0680365013346576, "grad_norm": 0.44440835215958774, "learning_rate": 8.132913977361427e-05, "loss": 2.9949, "step": 17205 }, { "epoch": 1.0680985784344155, "grad_norm": 0.516750407995276, "learning_rate": 8.13263250329632e-05, "loss": 2.8944, "step": 17206 }, { "epoch": 1.0681606555341734, "grad_norm": 0.4933461523511618, "learning_rate": 8.13235101288751e-05, "loss": 2.8998, "step": 17207 }, { "epoch": 1.0682227326339313, "grad_norm": 0.5037980665839918, "learning_rate": 8.132069506136464e-05, "loss": 2.9303, "step": 17208 }, { "epoch": 1.0682848097336892, "grad_norm": 0.517483886200663, "learning_rate": 8.13178798304465e-05, "loss": 2.8825, "step": 17209 }, { "epoch": 1.0683468868334471, "grad_norm": 0.517594621503396, "learning_rate": 8.131506443613539e-05, "loss": 2.9232, "step": 17210 }, { "epoch": 1.068408963933205, "grad_norm": 0.530709159656157, "learning_rate": 8.131224887844599e-05, "loss": 2.8947, "step": 17211 }, { "epoch": 1.068471041032963, "grad_norm": 0.5544789419174336, "learning_rate": 8.130943315739299e-05, "loss": 2.841, "step": 17212 }, { "epoch": 1.068533118132721, "grad_norm": 0.5445844637949777, "learning_rate": 8.130661727299106e-05, "loss": 2.9187, "step": 17213 }, { "epoch": 1.0685951952324788, "grad_norm": 0.5755907910896031, "learning_rate": 8.130380122525492e-05, "loss": 2.8724, "step": 17214 }, { "epoch": 1.0686572723322367, "grad_norm": 0.48475050983333284, "learning_rate": 8.130098501419925e-05, "loss": 2.8245, "step": 17215 }, { "epoch": 1.0687193494319946, "grad_norm": 0.5547000722400597, "learning_rate": 8.129816863983873e-05, "loss": 2.919, "step": 17216 }, { "epoch": 1.0687814265317523, "grad_norm": 0.5925210732925548, "learning_rate": 8.129535210218808e-05, "loss": 2.8224, "step": 17217 }, { "epoch": 1.0688435036315103, "grad_norm": 0.489126480963064, "learning_rate": 8.129253540126196e-05, "loss": 2.8153, "step": 17218 }, { "epoch": 1.0689055807312682, "grad_norm": 0.518372067121456, "learning_rate": 8.12897185370751e-05, "loss": 2.9015, "step": 17219 }, { "epoch": 1.068967657831026, "grad_norm": 0.5680680200017101, "learning_rate": 8.128690150964219e-05, "loss": 2.8131, "step": 17220 }, { "epoch": 1.069029734930784, "grad_norm": 0.5432216412128754, "learning_rate": 8.128408431897791e-05, "loss": 2.9484, "step": 17221 }, { "epoch": 1.069091812030542, "grad_norm": 0.5661510813544994, "learning_rate": 8.128126696509697e-05, "loss": 2.8341, "step": 17222 }, { "epoch": 1.0691538891302999, "grad_norm": 0.5163881261187337, "learning_rate": 8.127844944801406e-05, "loss": 2.903, "step": 17223 }, { "epoch": 1.0692159662300578, "grad_norm": 0.5330892308479283, "learning_rate": 8.12756317677439e-05, "loss": 2.8986, "step": 17224 }, { "epoch": 1.0692780433298157, "grad_norm": 0.621522311737727, "learning_rate": 8.127281392430117e-05, "loss": 2.8646, "step": 17225 }, { "epoch": 1.0693401204295736, "grad_norm": 0.6432878928204047, "learning_rate": 8.126999591770057e-05, "loss": 2.9902, "step": 17226 }, { "epoch": 1.0694021975293315, "grad_norm": 0.6131547688092579, "learning_rate": 8.126717774795682e-05, "loss": 3.0077, "step": 17227 }, { "epoch": 1.0694642746290892, "grad_norm": 0.5881080605543313, "learning_rate": 8.126435941508461e-05, "loss": 2.9012, "step": 17228 }, { "epoch": 1.0695263517288471, "grad_norm": 0.5550298037559817, "learning_rate": 8.126154091909865e-05, "loss": 2.881, "step": 17229 }, { "epoch": 1.069588428828605, "grad_norm": 0.4748132526813028, "learning_rate": 8.125872226001364e-05, "loss": 2.9133, "step": 17230 }, { "epoch": 1.069650505928363, "grad_norm": 0.5211007639628805, "learning_rate": 8.125590343784428e-05, "loss": 2.8748, "step": 17231 }, { "epoch": 1.069712583028121, "grad_norm": 0.46623625668579016, "learning_rate": 8.125308445260531e-05, "loss": 2.8598, "step": 17232 }, { "epoch": 1.0697746601278788, "grad_norm": 0.5099710803153705, "learning_rate": 8.12502653043114e-05, "loss": 2.8554, "step": 17233 }, { "epoch": 1.0698367372276367, "grad_norm": 0.46085610721746867, "learning_rate": 8.124744599297725e-05, "loss": 2.757, "step": 17234 }, { "epoch": 1.0698988143273946, "grad_norm": 0.48417025514981016, "learning_rate": 8.124462651861761e-05, "loss": 2.8283, "step": 17235 }, { "epoch": 1.0699608914271526, "grad_norm": 0.4689818921294316, "learning_rate": 8.124180688124716e-05, "loss": 2.8259, "step": 17236 }, { "epoch": 1.0700229685269105, "grad_norm": 0.535229086270009, "learning_rate": 8.123898708088063e-05, "loss": 2.9768, "step": 17237 }, { "epoch": 1.0700850456266684, "grad_norm": 0.5258641714654599, "learning_rate": 8.12361671175327e-05, "loss": 2.8745, "step": 17238 }, { "epoch": 1.0701471227264263, "grad_norm": 0.5222130873852882, "learning_rate": 8.123334699121811e-05, "loss": 2.8791, "step": 17239 }, { "epoch": 1.0702091998261842, "grad_norm": 0.5234813102493772, "learning_rate": 8.123052670195156e-05, "loss": 2.8448, "step": 17240 }, { "epoch": 1.070271276925942, "grad_norm": 0.5439064381814712, "learning_rate": 8.122770624974778e-05, "loss": 2.8671, "step": 17241 }, { "epoch": 1.0703333540256998, "grad_norm": 0.5022315355327507, "learning_rate": 8.122488563462149e-05, "loss": 2.8546, "step": 17242 }, { "epoch": 1.0703954311254578, "grad_norm": 0.5009080271103021, "learning_rate": 8.122206485658737e-05, "loss": 2.9827, "step": 17243 }, { "epoch": 1.0704575082252157, "grad_norm": 0.5600475785786052, "learning_rate": 8.121924391566017e-05, "loss": 2.7974, "step": 17244 }, { "epoch": 1.0705195853249736, "grad_norm": 0.4973641596856965, "learning_rate": 8.12164228118546e-05, "loss": 2.9374, "step": 17245 }, { "epoch": 1.0705816624247315, "grad_norm": 0.5929151236870804, "learning_rate": 8.121360154518536e-05, "loss": 2.9364, "step": 17246 }, { "epoch": 1.0706437395244894, "grad_norm": 0.5693142397689913, "learning_rate": 8.121078011566719e-05, "loss": 2.7929, "step": 17247 }, { "epoch": 1.0707058166242474, "grad_norm": 0.49059352409327944, "learning_rate": 8.12079585233148e-05, "loss": 2.9454, "step": 17248 }, { "epoch": 1.0707678937240053, "grad_norm": 0.5464880528029723, "learning_rate": 8.120513676814292e-05, "loss": 2.8473, "step": 17249 }, { "epoch": 1.0708299708237632, "grad_norm": 0.4725766706073798, "learning_rate": 8.120231485016627e-05, "loss": 2.7697, "step": 17250 }, { "epoch": 1.070892047923521, "grad_norm": 0.5126895867890846, "learning_rate": 8.119949276939955e-05, "loss": 2.922, "step": 17251 }, { "epoch": 1.0709541250232788, "grad_norm": 0.5503676464517457, "learning_rate": 8.119667052585753e-05, "loss": 2.877, "step": 17252 }, { "epoch": 1.0710162021230367, "grad_norm": 0.5177391233697044, "learning_rate": 8.119384811955489e-05, "loss": 2.9376, "step": 17253 }, { "epoch": 1.0710782792227946, "grad_norm": 0.5118907028044016, "learning_rate": 8.119102555050637e-05, "loss": 2.8057, "step": 17254 }, { "epoch": 1.0711403563225526, "grad_norm": 0.5164578676035716, "learning_rate": 8.11882028187267e-05, "loss": 2.8415, "step": 17255 }, { "epoch": 1.0712024334223105, "grad_norm": 0.46789112880219147, "learning_rate": 8.118537992423061e-05, "loss": 2.8405, "step": 17256 }, { "epoch": 1.0712645105220684, "grad_norm": 0.49793117899741285, "learning_rate": 8.118255686703282e-05, "loss": 2.9391, "step": 17257 }, { "epoch": 1.0713265876218263, "grad_norm": 0.4631728587473911, "learning_rate": 8.117973364714805e-05, "loss": 2.7049, "step": 17258 }, { "epoch": 1.0713886647215842, "grad_norm": 0.461931579128674, "learning_rate": 8.117691026459105e-05, "loss": 2.9549, "step": 17259 }, { "epoch": 1.0714507418213421, "grad_norm": 0.5041543808010225, "learning_rate": 8.117408671937654e-05, "loss": 2.9589, "step": 17260 }, { "epoch": 1.0715128189211, "grad_norm": 0.4885936348149158, "learning_rate": 8.117126301151925e-05, "loss": 2.8683, "step": 17261 }, { "epoch": 1.071574896020858, "grad_norm": 0.4888902299840311, "learning_rate": 8.116843914103393e-05, "loss": 2.8694, "step": 17262 }, { "epoch": 1.071636973120616, "grad_norm": 0.47772735752743933, "learning_rate": 8.116561510793528e-05, "loss": 2.869, "step": 17263 }, { "epoch": 1.0716990502203738, "grad_norm": 0.5119259829626917, "learning_rate": 8.116279091223804e-05, "loss": 2.8752, "step": 17264 }, { "epoch": 1.0717611273201315, "grad_norm": 0.4901093859365923, "learning_rate": 8.115996655395698e-05, "loss": 2.8473, "step": 17265 }, { "epoch": 1.0718232044198894, "grad_norm": 0.5193431485045955, "learning_rate": 8.115714203310679e-05, "loss": 2.8753, "step": 17266 }, { "epoch": 1.0718852815196473, "grad_norm": 0.485393883070855, "learning_rate": 8.115431734970224e-05, "loss": 2.863, "step": 17267 }, { "epoch": 1.0719473586194053, "grad_norm": 0.522958878286357, "learning_rate": 8.115149250375804e-05, "loss": 2.8835, "step": 17268 }, { "epoch": 1.0720094357191632, "grad_norm": 0.5082088684179713, "learning_rate": 8.114866749528896e-05, "loss": 2.9041, "step": 17269 }, { "epoch": 1.072071512818921, "grad_norm": 0.49735683434736794, "learning_rate": 8.11458423243097e-05, "loss": 2.8782, "step": 17270 }, { "epoch": 1.072133589918679, "grad_norm": 0.4824216433381671, "learning_rate": 8.114301699083503e-05, "loss": 2.7656, "step": 17271 }, { "epoch": 1.072195667018437, "grad_norm": 0.5746056946222317, "learning_rate": 8.114019149487969e-05, "loss": 2.9496, "step": 17272 }, { "epoch": 1.0722577441181949, "grad_norm": 0.5326315082993063, "learning_rate": 8.11373658364584e-05, "loss": 2.8863, "step": 17273 }, { "epoch": 1.0723198212179528, "grad_norm": 0.5088059318220692, "learning_rate": 8.11345400155859e-05, "loss": 2.926, "step": 17274 }, { "epoch": 1.0723818983177107, "grad_norm": 0.49955403287621053, "learning_rate": 8.113171403227697e-05, "loss": 2.8793, "step": 17275 }, { "epoch": 1.0724439754174684, "grad_norm": 0.5158525022828528, "learning_rate": 8.112888788654631e-05, "loss": 2.9325, "step": 17276 }, { "epoch": 1.0725060525172263, "grad_norm": 0.5192339914376789, "learning_rate": 8.11260615784087e-05, "loss": 2.9951, "step": 17277 }, { "epoch": 1.0725681296169842, "grad_norm": 0.46501815152887865, "learning_rate": 8.112323510787887e-05, "loss": 2.8014, "step": 17278 }, { "epoch": 1.0726302067167421, "grad_norm": 0.5784568092177551, "learning_rate": 8.112040847497156e-05, "loss": 2.9431, "step": 17279 }, { "epoch": 1.0726922838165, "grad_norm": 0.5477919112589151, "learning_rate": 8.111758167970153e-05, "loss": 2.9401, "step": 17280 }, { "epoch": 1.072754360916258, "grad_norm": 0.4823518926231619, "learning_rate": 8.111475472208352e-05, "loss": 2.9021, "step": 17281 }, { "epoch": 1.072816438016016, "grad_norm": 0.6264578743812327, "learning_rate": 8.111192760213229e-05, "loss": 2.9052, "step": 17282 }, { "epoch": 1.0728785151157738, "grad_norm": 0.6108559439387492, "learning_rate": 8.110910031986256e-05, "loss": 2.9285, "step": 17283 }, { "epoch": 1.0729405922155317, "grad_norm": 0.5441350417661982, "learning_rate": 8.110627287528912e-05, "loss": 2.8741, "step": 17284 }, { "epoch": 1.0730026693152896, "grad_norm": 0.6160360048704935, "learning_rate": 8.11034452684267e-05, "loss": 2.9023, "step": 17285 }, { "epoch": 1.0730647464150476, "grad_norm": 0.5440185607134993, "learning_rate": 8.110061749929006e-05, "loss": 2.8837, "step": 17286 }, { "epoch": 1.0731268235148055, "grad_norm": 0.5574752707192773, "learning_rate": 8.109778956789394e-05, "loss": 2.8415, "step": 17287 }, { "epoch": 1.0731889006145634, "grad_norm": 0.5186147134238389, "learning_rate": 8.109496147425309e-05, "loss": 2.9099, "step": 17288 }, { "epoch": 1.073250977714321, "grad_norm": 0.5233744398250274, "learning_rate": 8.109213321838229e-05, "loss": 2.9826, "step": 17289 }, { "epoch": 1.073313054814079, "grad_norm": 0.4948721284862516, "learning_rate": 8.108930480029629e-05, "loss": 2.9008, "step": 17290 }, { "epoch": 1.073375131913837, "grad_norm": 0.5772837885173722, "learning_rate": 8.108647622000983e-05, "loss": 2.8946, "step": 17291 }, { "epoch": 1.0734372090135949, "grad_norm": 0.5299263058566736, "learning_rate": 8.108364747753767e-05, "loss": 2.7916, "step": 17292 }, { "epoch": 1.0734992861133528, "grad_norm": 0.6092019202115685, "learning_rate": 8.10808185728946e-05, "loss": 2.9046, "step": 17293 }, { "epoch": 1.0735613632131107, "grad_norm": 0.6299645895694491, "learning_rate": 8.107798950609534e-05, "loss": 2.8212, "step": 17294 }, { "epoch": 1.0736234403128686, "grad_norm": 0.48480135089014076, "learning_rate": 8.107516027715467e-05, "loss": 2.7475, "step": 17295 }, { "epoch": 1.0736855174126265, "grad_norm": 0.6463122873522648, "learning_rate": 8.107233088608734e-05, "loss": 2.9464, "step": 17296 }, { "epoch": 1.0737475945123844, "grad_norm": 0.5523950458539941, "learning_rate": 8.106950133290811e-05, "loss": 2.8722, "step": 17297 }, { "epoch": 1.0738096716121424, "grad_norm": 0.5854421938108034, "learning_rate": 8.106667161763175e-05, "loss": 2.8508, "step": 17298 }, { "epoch": 1.0738717487119003, "grad_norm": 0.5541713892465391, "learning_rate": 8.106384174027304e-05, "loss": 2.9374, "step": 17299 }, { "epoch": 1.073933825811658, "grad_norm": 0.5183371969622814, "learning_rate": 8.106101170084671e-05, "loss": 2.8651, "step": 17300 }, { "epoch": 1.073995902911416, "grad_norm": 0.5098518795542551, "learning_rate": 8.105818149936755e-05, "loss": 2.819, "step": 17301 }, { "epoch": 1.0740579800111738, "grad_norm": 0.5363001327822476, "learning_rate": 8.105535113585031e-05, "loss": 2.8562, "step": 17302 }, { "epoch": 1.0741200571109317, "grad_norm": 0.5517253764923248, "learning_rate": 8.105252061030977e-05, "loss": 2.9077, "step": 17303 }, { "epoch": 1.0741821342106896, "grad_norm": 0.46034166651541664, "learning_rate": 8.104968992276069e-05, "loss": 2.8071, "step": 17304 }, { "epoch": 1.0742442113104476, "grad_norm": 0.5483688659324893, "learning_rate": 8.104685907321783e-05, "loss": 2.8567, "step": 17305 }, { "epoch": 1.0743062884102055, "grad_norm": 0.50122798685838, "learning_rate": 8.104402806169598e-05, "loss": 2.8772, "step": 17306 }, { "epoch": 1.0743683655099634, "grad_norm": 0.5163706676326345, "learning_rate": 8.10411968882099e-05, "loss": 2.9649, "step": 17307 }, { "epoch": 1.0744304426097213, "grad_norm": 0.5249095543654446, "learning_rate": 8.103836555277436e-05, "loss": 2.7925, "step": 17308 }, { "epoch": 1.0744925197094792, "grad_norm": 0.4818977721408389, "learning_rate": 8.103553405540413e-05, "loss": 2.8358, "step": 17309 }, { "epoch": 1.0745545968092372, "grad_norm": 0.4839997068228898, "learning_rate": 8.103270239611399e-05, "loss": 2.9313, "step": 17310 }, { "epoch": 1.074616673908995, "grad_norm": 0.5181192124875258, "learning_rate": 8.102987057491871e-05, "loss": 2.8857, "step": 17311 }, { "epoch": 1.074678751008753, "grad_norm": 0.49921700144477954, "learning_rate": 8.102703859183305e-05, "loss": 2.8928, "step": 17312 }, { "epoch": 1.0747408281085107, "grad_norm": 0.5366609122972813, "learning_rate": 8.102420644687181e-05, "loss": 2.895, "step": 17313 }, { "epoch": 1.0748029052082686, "grad_norm": 0.5495108434094931, "learning_rate": 8.102137414004976e-05, "loss": 2.8743, "step": 17314 }, { "epoch": 1.0748649823080265, "grad_norm": 0.5255563751038006, "learning_rate": 8.101854167138165e-05, "loss": 2.9354, "step": 17315 }, { "epoch": 1.0749270594077844, "grad_norm": 0.4695688248513906, "learning_rate": 8.101570904088228e-05, "loss": 2.8752, "step": 17316 }, { "epoch": 1.0749891365075424, "grad_norm": 0.5254887202129364, "learning_rate": 8.101287624856644e-05, "loss": 2.8249, "step": 17317 }, { "epoch": 1.0750512136073003, "grad_norm": 0.5210028150101685, "learning_rate": 8.101004329444888e-05, "loss": 2.8812, "step": 17318 }, { "epoch": 1.0751132907070582, "grad_norm": 0.5304236998270305, "learning_rate": 8.10072101785444e-05, "loss": 2.8589, "step": 17319 }, { "epoch": 1.075175367806816, "grad_norm": 0.4982071861614817, "learning_rate": 8.100437690086779e-05, "loss": 2.8759, "step": 17320 }, { "epoch": 1.075237444906574, "grad_norm": 0.5311334145468131, "learning_rate": 8.10015434614338e-05, "loss": 2.854, "step": 17321 }, { "epoch": 1.075299522006332, "grad_norm": 0.5308941603064393, "learning_rate": 8.099870986025724e-05, "loss": 2.9946, "step": 17322 }, { "epoch": 1.0753615991060899, "grad_norm": 0.4926941398474181, "learning_rate": 8.09958760973529e-05, "loss": 2.7765, "step": 17323 }, { "epoch": 1.0754236762058476, "grad_norm": 0.47137887022361236, "learning_rate": 8.099304217273553e-05, "loss": 2.8587, "step": 17324 }, { "epoch": 1.0754857533056055, "grad_norm": 0.46857562000478076, "learning_rate": 8.099020808641994e-05, "loss": 2.8944, "step": 17325 }, { "epoch": 1.0755478304053634, "grad_norm": 0.48086464861186723, "learning_rate": 8.098737383842092e-05, "loss": 2.9067, "step": 17326 }, { "epoch": 1.0756099075051213, "grad_norm": 0.45711198729409364, "learning_rate": 8.098453942875325e-05, "loss": 2.8483, "step": 17327 }, { "epoch": 1.0756719846048792, "grad_norm": 0.5141841234379364, "learning_rate": 8.09817048574317e-05, "loss": 2.8918, "step": 17328 }, { "epoch": 1.0757340617046371, "grad_norm": 0.549802654797877, "learning_rate": 8.09788701244711e-05, "loss": 2.858, "step": 17329 }, { "epoch": 1.075796138804395, "grad_norm": 0.42342540399933365, "learning_rate": 8.097603522988619e-05, "loss": 2.9243, "step": 17330 }, { "epoch": 1.075858215904153, "grad_norm": 0.5527368282627618, "learning_rate": 8.09732001736918e-05, "loss": 2.8448, "step": 17331 }, { "epoch": 1.075920293003911, "grad_norm": 0.4899685051342805, "learning_rate": 8.097036495590269e-05, "loss": 2.9362, "step": 17332 }, { "epoch": 1.0759823701036688, "grad_norm": 0.452339972182131, "learning_rate": 8.096752957653368e-05, "loss": 2.7978, "step": 17333 }, { "epoch": 1.0760444472034267, "grad_norm": 0.4706283446619911, "learning_rate": 8.096469403559955e-05, "loss": 2.8586, "step": 17334 }, { "epoch": 1.0761065243031847, "grad_norm": 0.44805869541689364, "learning_rate": 8.096185833311509e-05, "loss": 2.8728, "step": 17335 }, { "epoch": 1.0761686014029426, "grad_norm": 0.5407220202141346, "learning_rate": 8.095902246909509e-05, "loss": 2.8945, "step": 17336 }, { "epoch": 1.0762306785027003, "grad_norm": 0.49959632850535624, "learning_rate": 8.095618644355438e-05, "loss": 2.9252, "step": 17337 }, { "epoch": 1.0762927556024582, "grad_norm": 0.4724141911896959, "learning_rate": 8.09533502565077e-05, "loss": 2.905, "step": 17338 }, { "epoch": 1.076354832702216, "grad_norm": 0.4545372134512816, "learning_rate": 8.095051390796989e-05, "loss": 2.8224, "step": 17339 }, { "epoch": 1.076416909801974, "grad_norm": 0.47821367039961904, "learning_rate": 8.094767739795574e-05, "loss": 2.8086, "step": 17340 }, { "epoch": 1.076478986901732, "grad_norm": 0.4899326322916908, "learning_rate": 8.094484072648004e-05, "loss": 2.9026, "step": 17341 }, { "epoch": 1.0765410640014899, "grad_norm": 0.5016412619667067, "learning_rate": 8.09420038935576e-05, "loss": 2.8567, "step": 17342 }, { "epoch": 1.0766031411012478, "grad_norm": 0.5021813612196564, "learning_rate": 8.093916689920322e-05, "loss": 2.8957, "step": 17343 }, { "epoch": 1.0766652182010057, "grad_norm": 0.47245797025861325, "learning_rate": 8.093632974343167e-05, "loss": 2.8931, "step": 17344 }, { "epoch": 1.0767272953007636, "grad_norm": 0.5073942075591203, "learning_rate": 8.093349242625779e-05, "loss": 2.8806, "step": 17345 }, { "epoch": 1.0767893724005215, "grad_norm": 0.5235707127743342, "learning_rate": 8.093065494769638e-05, "loss": 2.9078, "step": 17346 }, { "epoch": 1.0768514495002794, "grad_norm": 0.5116077623052088, "learning_rate": 8.092781730776222e-05, "loss": 2.98, "step": 17347 }, { "epoch": 1.0769135266000371, "grad_norm": 0.4633410230291222, "learning_rate": 8.092497950647014e-05, "loss": 2.9264, "step": 17348 }, { "epoch": 1.076975603699795, "grad_norm": 0.5025837421385002, "learning_rate": 8.092214154383494e-05, "loss": 2.8297, "step": 17349 }, { "epoch": 1.077037680799553, "grad_norm": 0.4518014717692753, "learning_rate": 8.091930341987141e-05, "loss": 2.9953, "step": 17350 }, { "epoch": 1.077099757899311, "grad_norm": 0.5306919195889441, "learning_rate": 8.091646513459438e-05, "loss": 2.8745, "step": 17351 }, { "epoch": 1.0771618349990688, "grad_norm": 0.5304998262317024, "learning_rate": 8.091362668801863e-05, "loss": 2.9341, "step": 17352 }, { "epoch": 1.0772239120988267, "grad_norm": 0.5653450594541009, "learning_rate": 8.091078808015899e-05, "loss": 2.8314, "step": 17353 }, { "epoch": 1.0772859891985846, "grad_norm": 0.5171200722710192, "learning_rate": 8.090794931103026e-05, "loss": 2.8792, "step": 17354 }, { "epoch": 1.0773480662983426, "grad_norm": 0.6168167111068179, "learning_rate": 8.090511038064728e-05, "loss": 2.9158, "step": 17355 }, { "epoch": 1.0774101433981005, "grad_norm": 0.5398363557300873, "learning_rate": 8.09022712890248e-05, "loss": 2.8519, "step": 17356 }, { "epoch": 1.0774722204978584, "grad_norm": 0.5479499062245063, "learning_rate": 8.089943203617769e-05, "loss": 2.8534, "step": 17357 }, { "epoch": 1.0775342975976163, "grad_norm": 0.5223540009483487, "learning_rate": 8.089659262212075e-05, "loss": 2.9486, "step": 17358 }, { "epoch": 1.0775963746973742, "grad_norm": 0.5858502386601355, "learning_rate": 8.089375304686877e-05, "loss": 2.854, "step": 17359 }, { "epoch": 1.0776584517971322, "grad_norm": 0.45807560986248885, "learning_rate": 8.089091331043659e-05, "loss": 2.8576, "step": 17360 }, { "epoch": 1.0777205288968899, "grad_norm": 0.4941600298181219, "learning_rate": 8.0888073412839e-05, "loss": 2.9161, "step": 17361 }, { "epoch": 1.0777826059966478, "grad_norm": 0.5055478289151142, "learning_rate": 8.088523335409085e-05, "loss": 2.853, "step": 17362 }, { "epoch": 1.0778446830964057, "grad_norm": 0.47937479543868056, "learning_rate": 8.088239313420694e-05, "loss": 2.8243, "step": 17363 }, { "epoch": 1.0779067601961636, "grad_norm": 0.5475255366688568, "learning_rate": 8.087955275320208e-05, "loss": 2.8735, "step": 17364 }, { "epoch": 1.0779688372959215, "grad_norm": 0.5225663123418898, "learning_rate": 8.08767122110911e-05, "loss": 2.9352, "step": 17365 }, { "epoch": 1.0780309143956794, "grad_norm": 0.530703852919887, "learning_rate": 8.087387150788881e-05, "loss": 2.9193, "step": 17366 }, { "epoch": 1.0780929914954374, "grad_norm": 0.5488671122648581, "learning_rate": 8.087103064361005e-05, "loss": 2.9197, "step": 17367 }, { "epoch": 1.0781550685951953, "grad_norm": 0.5613687584172644, "learning_rate": 8.086818961826963e-05, "loss": 2.9354, "step": 17368 }, { "epoch": 1.0782171456949532, "grad_norm": 0.5790032082931628, "learning_rate": 8.086534843188237e-05, "loss": 2.9041, "step": 17369 }, { "epoch": 1.0782792227947111, "grad_norm": 0.5084719557479099, "learning_rate": 8.086250708446308e-05, "loss": 2.9129, "step": 17370 }, { "epoch": 1.078341299894469, "grad_norm": 0.5361360365695979, "learning_rate": 8.08596655760266e-05, "loss": 2.9265, "step": 17371 }, { "epoch": 1.0784033769942267, "grad_norm": 0.5023797741634867, "learning_rate": 8.085682390658776e-05, "loss": 2.8399, "step": 17372 }, { "epoch": 1.0784654540939846, "grad_norm": 0.5722677081073434, "learning_rate": 8.085398207616138e-05, "loss": 2.9563, "step": 17373 }, { "epoch": 1.0785275311937426, "grad_norm": 0.5187813645957818, "learning_rate": 8.085114008476229e-05, "loss": 2.9043, "step": 17374 }, { "epoch": 1.0785896082935005, "grad_norm": 0.5636938723939803, "learning_rate": 8.08482979324053e-05, "loss": 2.791, "step": 17375 }, { "epoch": 1.0786516853932584, "grad_norm": 0.5628528018397428, "learning_rate": 8.084545561910526e-05, "loss": 2.8952, "step": 17376 }, { "epoch": 1.0787137624930163, "grad_norm": 0.5055215362862553, "learning_rate": 8.0842613144877e-05, "loss": 2.8833, "step": 17377 }, { "epoch": 1.0787758395927742, "grad_norm": 0.5467289593384681, "learning_rate": 8.083977050973532e-05, "loss": 2.791, "step": 17378 }, { "epoch": 1.0788379166925322, "grad_norm": 0.5094665292989686, "learning_rate": 8.083692771369509e-05, "loss": 2.8653, "step": 17379 }, { "epoch": 1.07889999379229, "grad_norm": 0.5918043280909598, "learning_rate": 8.08340847567711e-05, "loss": 2.9019, "step": 17380 }, { "epoch": 1.078962070892048, "grad_norm": 0.5729333672834795, "learning_rate": 8.083124163897822e-05, "loss": 2.897, "step": 17381 }, { "epoch": 1.079024147991806, "grad_norm": 0.49871772016309823, "learning_rate": 8.082839836033126e-05, "loss": 2.8429, "step": 17382 }, { "epoch": 1.0790862250915638, "grad_norm": 0.5432614148358708, "learning_rate": 8.082555492084507e-05, "loss": 2.809, "step": 17383 }, { "epoch": 1.0791483021913215, "grad_norm": 0.5146475978316057, "learning_rate": 8.082271132053448e-05, "loss": 2.7984, "step": 17384 }, { "epoch": 1.0792103792910794, "grad_norm": 0.5002797953241195, "learning_rate": 8.081986755941431e-05, "loss": 2.8728, "step": 17385 }, { "epoch": 1.0792724563908374, "grad_norm": 0.5732498298408532, "learning_rate": 8.08170236374994e-05, "loss": 2.9006, "step": 17386 }, { "epoch": 1.0793345334905953, "grad_norm": 0.567684940124472, "learning_rate": 8.081417955480462e-05, "loss": 2.8569, "step": 17387 }, { "epoch": 1.0793966105903532, "grad_norm": 0.5188705430672235, "learning_rate": 8.081133531134477e-05, "loss": 2.8307, "step": 17388 }, { "epoch": 1.079458687690111, "grad_norm": 0.5046709748860697, "learning_rate": 8.080849090713472e-05, "loss": 2.9573, "step": 17389 }, { "epoch": 1.079520764789869, "grad_norm": 0.5342242067186103, "learning_rate": 8.080564634218928e-05, "loss": 2.9637, "step": 17390 }, { "epoch": 1.079582841889627, "grad_norm": 0.45313501346968, "learning_rate": 8.08028016165233e-05, "loss": 2.9436, "step": 17391 }, { "epoch": 1.0796449189893849, "grad_norm": 0.5361551026538077, "learning_rate": 8.079995673015164e-05, "loss": 2.8719, "step": 17392 }, { "epoch": 1.0797069960891428, "grad_norm": 0.5022938977981353, "learning_rate": 8.079711168308912e-05, "loss": 2.8658, "step": 17393 }, { "epoch": 1.0797690731889007, "grad_norm": 0.526900295722793, "learning_rate": 8.07942664753506e-05, "loss": 2.8415, "step": 17394 }, { "epoch": 1.0798311502886586, "grad_norm": 0.5690535406070397, "learning_rate": 8.07914211069509e-05, "loss": 2.9314, "step": 17395 }, { "epoch": 1.0798932273884163, "grad_norm": 0.523057885371038, "learning_rate": 8.07885755779049e-05, "loss": 2.9335, "step": 17396 }, { "epoch": 1.0799553044881742, "grad_norm": 0.5509420660633542, "learning_rate": 8.078572988822742e-05, "loss": 2.8426, "step": 17397 }, { "epoch": 1.0800173815879321, "grad_norm": 0.546893364734195, "learning_rate": 8.078288403793331e-05, "loss": 2.9453, "step": 17398 }, { "epoch": 1.08007945868769, "grad_norm": 0.5139042723847587, "learning_rate": 8.078003802703743e-05, "loss": 2.7537, "step": 17399 }, { "epoch": 1.080141535787448, "grad_norm": 0.49539897356396706, "learning_rate": 8.077719185555461e-05, "loss": 2.8695, "step": 17400 }, { "epoch": 1.080203612887206, "grad_norm": 0.529997527278674, "learning_rate": 8.077434552349971e-05, "loss": 2.8601, "step": 17401 }, { "epoch": 1.0802656899869638, "grad_norm": 0.4942476511008482, "learning_rate": 8.077149903088759e-05, "loss": 2.8198, "step": 17402 }, { "epoch": 1.0803277670867217, "grad_norm": 0.5610874241318556, "learning_rate": 8.076865237773308e-05, "loss": 2.9249, "step": 17403 }, { "epoch": 1.0803898441864797, "grad_norm": 0.5253044256239942, "learning_rate": 8.076580556405104e-05, "loss": 2.8755, "step": 17404 }, { "epoch": 1.0804519212862376, "grad_norm": 0.49634472063115337, "learning_rate": 8.076295858985632e-05, "loss": 2.8251, "step": 17405 }, { "epoch": 1.0805139983859955, "grad_norm": 0.5785277484496678, "learning_rate": 8.076011145516379e-05, "loss": 2.8869, "step": 17406 }, { "epoch": 1.0805760754857534, "grad_norm": 0.5089232987056636, "learning_rate": 8.075726415998828e-05, "loss": 2.9175, "step": 17407 }, { "epoch": 1.080638152585511, "grad_norm": 0.5509661911857452, "learning_rate": 8.075441670434467e-05, "loss": 2.8623, "step": 17408 }, { "epoch": 1.080700229685269, "grad_norm": 0.5274261692772434, "learning_rate": 8.075156908824779e-05, "loss": 2.8166, "step": 17409 }, { "epoch": 1.080762306785027, "grad_norm": 0.5643507920794396, "learning_rate": 8.074872131171251e-05, "loss": 2.8554, "step": 17410 }, { "epoch": 1.0808243838847849, "grad_norm": 0.4811606039271559, "learning_rate": 8.074587337475369e-05, "loss": 2.807, "step": 17411 }, { "epoch": 1.0808864609845428, "grad_norm": 0.5020812053028559, "learning_rate": 8.074302527738618e-05, "loss": 2.8996, "step": 17412 }, { "epoch": 1.0809485380843007, "grad_norm": 0.5282921915802629, "learning_rate": 8.074017701962485e-05, "loss": 2.8937, "step": 17413 }, { "epoch": 1.0810106151840586, "grad_norm": 0.532105934401901, "learning_rate": 8.073732860148455e-05, "loss": 2.9738, "step": 17414 }, { "epoch": 1.0810726922838165, "grad_norm": 0.5298826066497279, "learning_rate": 8.073448002298014e-05, "loss": 2.9067, "step": 17415 }, { "epoch": 1.0811347693835744, "grad_norm": 0.5152481320202471, "learning_rate": 8.073163128412649e-05, "loss": 2.8914, "step": 17416 }, { "epoch": 1.0811968464833324, "grad_norm": 0.5214688873848626, "learning_rate": 8.072878238493846e-05, "loss": 2.81, "step": 17417 }, { "epoch": 1.0812589235830903, "grad_norm": 0.5272436258845079, "learning_rate": 8.072593332543092e-05, "loss": 2.8296, "step": 17418 }, { "epoch": 1.0813210006828482, "grad_norm": 0.5046519594699285, "learning_rate": 8.072308410561872e-05, "loss": 2.852, "step": 17419 }, { "epoch": 1.081383077782606, "grad_norm": 0.4985737575490467, "learning_rate": 8.072023472551673e-05, "loss": 2.833, "step": 17420 }, { "epoch": 1.0814451548823638, "grad_norm": 0.5747383258242923, "learning_rate": 8.071738518513983e-05, "loss": 2.8986, "step": 17421 }, { "epoch": 1.0815072319821217, "grad_norm": 0.4495607318299809, "learning_rate": 8.071453548450286e-05, "loss": 2.7587, "step": 17422 }, { "epoch": 1.0815693090818796, "grad_norm": 0.5523543922014181, "learning_rate": 8.07116856236207e-05, "loss": 2.8362, "step": 17423 }, { "epoch": 1.0816313861816376, "grad_norm": 0.448629888941033, "learning_rate": 8.070883560250823e-05, "loss": 2.8642, "step": 17424 }, { "epoch": 1.0816934632813955, "grad_norm": 0.49703958412625476, "learning_rate": 8.07059854211803e-05, "loss": 2.9896, "step": 17425 }, { "epoch": 1.0817555403811534, "grad_norm": 0.5450606357458161, "learning_rate": 8.070313507965181e-05, "loss": 2.9576, "step": 17426 }, { "epoch": 1.0818176174809113, "grad_norm": 0.5403699912907489, "learning_rate": 8.07002845779376e-05, "loss": 3.0153, "step": 17427 }, { "epoch": 1.0818796945806692, "grad_norm": 0.5247548734521305, "learning_rate": 8.069743391605256e-05, "loss": 2.8781, "step": 17428 }, { "epoch": 1.0819417716804272, "grad_norm": 0.5154896471795126, "learning_rate": 8.069458309401155e-05, "loss": 2.994, "step": 17429 }, { "epoch": 1.082003848780185, "grad_norm": 0.5622195498378268, "learning_rate": 8.069173211182943e-05, "loss": 2.8676, "step": 17430 }, { "epoch": 1.082065925879943, "grad_norm": 0.4914864588353591, "learning_rate": 8.068888096952111e-05, "loss": 2.8835, "step": 17431 }, { "epoch": 1.0821280029797007, "grad_norm": 0.5283802724719139, "learning_rate": 8.068602966710146e-05, "loss": 2.8805, "step": 17432 }, { "epoch": 1.0821900800794586, "grad_norm": 0.4801554084569357, "learning_rate": 8.068317820458532e-05, "loss": 2.85, "step": 17433 }, { "epoch": 1.0822521571792165, "grad_norm": 0.47808331014848005, "learning_rate": 8.06803265819876e-05, "loss": 2.9429, "step": 17434 }, { "epoch": 1.0823142342789744, "grad_norm": 0.49591806259475896, "learning_rate": 8.067747479932318e-05, "loss": 2.92, "step": 17435 }, { "epoch": 1.0823763113787324, "grad_norm": 0.44377392449157577, "learning_rate": 8.06746228566069e-05, "loss": 2.7953, "step": 17436 }, { "epoch": 1.0824383884784903, "grad_norm": 0.4914334742975377, "learning_rate": 8.067177075385369e-05, "loss": 2.831, "step": 17437 }, { "epoch": 1.0825004655782482, "grad_norm": 0.4615360511738242, "learning_rate": 8.06689184910784e-05, "loss": 2.883, "step": 17438 }, { "epoch": 1.0825625426780061, "grad_norm": 0.48803742996976246, "learning_rate": 8.066606606829591e-05, "loss": 2.8729, "step": 17439 }, { "epoch": 1.082624619777764, "grad_norm": 0.49743501066200185, "learning_rate": 8.066321348552111e-05, "loss": 2.7419, "step": 17440 }, { "epoch": 1.082686696877522, "grad_norm": 0.5081883726683698, "learning_rate": 8.06603607427689e-05, "loss": 2.8252, "step": 17441 }, { "epoch": 1.0827487739772799, "grad_norm": 0.5492117413308307, "learning_rate": 8.065750784005413e-05, "loss": 2.9369, "step": 17442 }, { "epoch": 1.0828108510770378, "grad_norm": 0.507627365403628, "learning_rate": 8.06546547773917e-05, "loss": 2.9849, "step": 17443 }, { "epoch": 1.0828729281767955, "grad_norm": 0.6173692990796933, "learning_rate": 8.065180155479648e-05, "loss": 2.9652, "step": 17444 }, { "epoch": 1.0829350052765534, "grad_norm": 0.569222152620209, "learning_rate": 8.06489481722834e-05, "loss": 2.7618, "step": 17445 }, { "epoch": 1.0829970823763113, "grad_norm": 0.5653108462029555, "learning_rate": 8.06460946298673e-05, "loss": 2.8358, "step": 17446 }, { "epoch": 1.0830591594760692, "grad_norm": 0.5378345934038716, "learning_rate": 8.064324092756308e-05, "loss": 2.8905, "step": 17447 }, { "epoch": 1.0831212365758272, "grad_norm": 0.5456032816616213, "learning_rate": 8.064038706538563e-05, "loss": 2.8282, "step": 17448 }, { "epoch": 1.083183313675585, "grad_norm": 0.5382980786974708, "learning_rate": 8.063753304334986e-05, "loss": 2.9137, "step": 17449 }, { "epoch": 1.083245390775343, "grad_norm": 0.523667526944136, "learning_rate": 8.063467886147063e-05, "loss": 2.8318, "step": 17450 }, { "epoch": 1.083307467875101, "grad_norm": 0.5521581766992087, "learning_rate": 8.063182451976284e-05, "loss": 2.8671, "step": 17451 }, { "epoch": 1.0833695449748588, "grad_norm": 0.5383299674036167, "learning_rate": 8.06289700182414e-05, "loss": 2.878, "step": 17452 }, { "epoch": 1.0834316220746167, "grad_norm": 0.5376021188382701, "learning_rate": 8.062611535692118e-05, "loss": 2.8371, "step": 17453 }, { "epoch": 1.0834936991743747, "grad_norm": 0.5477612534885653, "learning_rate": 8.062326053581709e-05, "loss": 2.9193, "step": 17454 }, { "epoch": 1.0835557762741326, "grad_norm": 0.49732634839610673, "learning_rate": 8.062040555494401e-05, "loss": 2.9109, "step": 17455 }, { "epoch": 1.0836178533738903, "grad_norm": 0.5193411113456905, "learning_rate": 8.061755041431683e-05, "loss": 2.9559, "step": 17456 }, { "epoch": 1.0836799304736482, "grad_norm": 0.5094482778720201, "learning_rate": 8.061469511395047e-05, "loss": 2.9898, "step": 17457 }, { "epoch": 1.083742007573406, "grad_norm": 0.43891546631503997, "learning_rate": 8.06118396538598e-05, "loss": 2.8209, "step": 17458 }, { "epoch": 1.083804084673164, "grad_norm": 0.47638711672577777, "learning_rate": 8.060898403405975e-05, "loss": 2.8667, "step": 17459 }, { "epoch": 1.083866161772922, "grad_norm": 0.4355795125322396, "learning_rate": 8.060612825456519e-05, "loss": 2.8226, "step": 17460 }, { "epoch": 1.0839282388726799, "grad_norm": 0.49282929795203834, "learning_rate": 8.060327231539104e-05, "loss": 2.8815, "step": 17461 }, { "epoch": 1.0839903159724378, "grad_norm": 0.48124032753971535, "learning_rate": 8.060041621655217e-05, "loss": 2.8285, "step": 17462 }, { "epoch": 1.0840523930721957, "grad_norm": 0.4696602883644715, "learning_rate": 8.05975599580635e-05, "loss": 2.8768, "step": 17463 }, { "epoch": 1.0841144701719536, "grad_norm": 0.4625689358499971, "learning_rate": 8.059470353993994e-05, "loss": 2.9923, "step": 17464 }, { "epoch": 1.0841765472717115, "grad_norm": 0.49904498148956383, "learning_rate": 8.05918469621964e-05, "loss": 2.9211, "step": 17465 }, { "epoch": 1.0842386243714695, "grad_norm": 0.47230770721871024, "learning_rate": 8.058899022484777e-05, "loss": 2.7798, "step": 17466 }, { "epoch": 1.0843007014712274, "grad_norm": 0.5129144690637228, "learning_rate": 8.058613332790891e-05, "loss": 2.9001, "step": 17467 }, { "epoch": 1.084362778570985, "grad_norm": 0.5445802225998418, "learning_rate": 8.05832762713948e-05, "loss": 2.8598, "step": 17468 }, { "epoch": 1.084424855670743, "grad_norm": 0.44157850229757345, "learning_rate": 8.05804190553203e-05, "loss": 2.9139, "step": 17469 }, { "epoch": 1.084486932770501, "grad_norm": 0.5049705502921619, "learning_rate": 8.057756167970035e-05, "loss": 2.8184, "step": 17470 }, { "epoch": 1.0845490098702588, "grad_norm": 0.4706949573615481, "learning_rate": 8.057470414454982e-05, "loss": 2.9334, "step": 17471 }, { "epoch": 1.0846110869700167, "grad_norm": 0.5180538366520477, "learning_rate": 8.057184644988364e-05, "loss": 2.8725, "step": 17472 }, { "epoch": 1.0846731640697747, "grad_norm": 0.45915058454272584, "learning_rate": 8.056898859571671e-05, "loss": 2.8754, "step": 17473 }, { "epoch": 1.0847352411695326, "grad_norm": 0.5165923898383735, "learning_rate": 8.056613058206395e-05, "loss": 2.8698, "step": 17474 }, { "epoch": 1.0847973182692905, "grad_norm": 0.49904871389496397, "learning_rate": 8.056327240894027e-05, "loss": 2.8685, "step": 17475 }, { "epoch": 1.0848593953690484, "grad_norm": 0.5121068283704439, "learning_rate": 8.056041407636056e-05, "loss": 2.8301, "step": 17476 }, { "epoch": 1.0849214724688063, "grad_norm": 0.513405560401464, "learning_rate": 8.055755558433979e-05, "loss": 2.844, "step": 17477 }, { "epoch": 1.0849835495685642, "grad_norm": 0.5083659164221238, "learning_rate": 8.05546969328928e-05, "loss": 2.9026, "step": 17478 }, { "epoch": 1.0850456266683222, "grad_norm": 0.5114527877486253, "learning_rate": 8.055183812203455e-05, "loss": 2.7248, "step": 17479 }, { "epoch": 1.0851077037680799, "grad_norm": 0.5612185504657444, "learning_rate": 8.054897915177995e-05, "loss": 2.8933, "step": 17480 }, { "epoch": 1.0851697808678378, "grad_norm": 0.5272521893169589, "learning_rate": 8.05461200221439e-05, "loss": 2.8245, "step": 17481 }, { "epoch": 1.0852318579675957, "grad_norm": 0.5768538754481393, "learning_rate": 8.054326073314134e-05, "loss": 2.9287, "step": 17482 }, { "epoch": 1.0852939350673536, "grad_norm": 0.5401457531135019, "learning_rate": 8.054040128478718e-05, "loss": 2.7609, "step": 17483 }, { "epoch": 1.0853560121671115, "grad_norm": 0.5321653558376634, "learning_rate": 8.053754167709629e-05, "loss": 2.8499, "step": 17484 }, { "epoch": 1.0854180892668694, "grad_norm": 0.6124506628835309, "learning_rate": 8.053468191008367e-05, "loss": 2.8653, "step": 17485 }, { "epoch": 1.0854801663666274, "grad_norm": 0.5100717027598466, "learning_rate": 8.05318219837642e-05, "loss": 2.8798, "step": 17486 }, { "epoch": 1.0855422434663853, "grad_norm": 0.5926249033620675, "learning_rate": 8.052896189815278e-05, "loss": 2.8607, "step": 17487 }, { "epoch": 1.0856043205661432, "grad_norm": 0.5300522935008583, "learning_rate": 8.052610165326436e-05, "loss": 2.8331, "step": 17488 }, { "epoch": 1.0856663976659011, "grad_norm": 0.5543019203779032, "learning_rate": 8.052324124911387e-05, "loss": 2.8928, "step": 17489 }, { "epoch": 1.085728474765659, "grad_norm": 0.5437959366610465, "learning_rate": 8.052038068571621e-05, "loss": 2.9168, "step": 17490 }, { "epoch": 1.0857905518654167, "grad_norm": 0.5803218547866131, "learning_rate": 8.051751996308633e-05, "loss": 2.9147, "step": 17491 }, { "epoch": 1.0858526289651746, "grad_norm": 0.5684789420398972, "learning_rate": 8.051465908123912e-05, "loss": 2.8926, "step": 17492 }, { "epoch": 1.0859147060649326, "grad_norm": 0.5074941370771027, "learning_rate": 8.051179804018956e-05, "loss": 2.8516, "step": 17493 }, { "epoch": 1.0859767831646905, "grad_norm": 0.58382482486151, "learning_rate": 8.05089368399525e-05, "loss": 2.9079, "step": 17494 }, { "epoch": 1.0860388602644484, "grad_norm": 0.5278024726940572, "learning_rate": 8.050607548054294e-05, "loss": 2.876, "step": 17495 }, { "epoch": 1.0861009373642063, "grad_norm": 0.5089612731609262, "learning_rate": 8.050321396197577e-05, "loss": 2.8134, "step": 17496 }, { "epoch": 1.0861630144639642, "grad_norm": 0.5078161973085054, "learning_rate": 8.050035228426592e-05, "loss": 2.8548, "step": 17497 }, { "epoch": 1.0862250915637222, "grad_norm": 0.48680316750174535, "learning_rate": 8.049749044742834e-05, "loss": 2.8698, "step": 17498 }, { "epoch": 1.08628716866348, "grad_norm": 0.48556404912011375, "learning_rate": 8.049462845147794e-05, "loss": 2.8545, "step": 17499 }, { "epoch": 1.086349245763238, "grad_norm": 0.5143248900041669, "learning_rate": 8.049176629642966e-05, "loss": 2.9721, "step": 17500 }, { "epoch": 1.086411322862996, "grad_norm": 0.5019907362090767, "learning_rate": 8.048890398229843e-05, "loss": 2.901, "step": 17501 }, { "epoch": 1.0864733999627538, "grad_norm": 0.4758207458238576, "learning_rate": 8.048604150909919e-05, "loss": 2.9429, "step": 17502 }, { "epoch": 1.0865354770625117, "grad_norm": 0.5468401216556296, "learning_rate": 8.048317887684688e-05, "loss": 2.9072, "step": 17503 }, { "epoch": 1.0865975541622694, "grad_norm": 0.45557453438559686, "learning_rate": 8.048031608555641e-05, "loss": 2.9677, "step": 17504 }, { "epoch": 1.0866596312620274, "grad_norm": 0.5357413330516336, "learning_rate": 8.047745313524276e-05, "loss": 2.8492, "step": 17505 }, { "epoch": 1.0867217083617853, "grad_norm": 0.561100490341557, "learning_rate": 8.047459002592081e-05, "loss": 2.9253, "step": 17506 }, { "epoch": 1.0867837854615432, "grad_norm": 0.4884831888328251, "learning_rate": 8.047172675760553e-05, "loss": 2.9367, "step": 17507 }, { "epoch": 1.0868458625613011, "grad_norm": 0.49856677875114996, "learning_rate": 8.046886333031188e-05, "loss": 2.9227, "step": 17508 }, { "epoch": 1.086907939661059, "grad_norm": 0.4942070837148463, "learning_rate": 8.046599974405475e-05, "loss": 2.9109, "step": 17509 }, { "epoch": 1.086970016760817, "grad_norm": 0.5282741110408748, "learning_rate": 8.04631359988491e-05, "loss": 2.9157, "step": 17510 }, { "epoch": 1.0870320938605749, "grad_norm": 0.5255135885091494, "learning_rate": 8.046027209470988e-05, "loss": 2.7908, "step": 17511 }, { "epoch": 1.0870941709603328, "grad_norm": 0.5508907560965131, "learning_rate": 8.045740803165202e-05, "loss": 2.8809, "step": 17512 }, { "epoch": 1.0871562480600907, "grad_norm": 0.4902359555891477, "learning_rate": 8.045454380969049e-05, "loss": 2.8987, "step": 17513 }, { "epoch": 1.0872183251598486, "grad_norm": 0.6354489214063845, "learning_rate": 8.04516794288402e-05, "loss": 2.865, "step": 17514 }, { "epoch": 1.0872804022596063, "grad_norm": 0.5788055744224333, "learning_rate": 8.044881488911609e-05, "loss": 2.8634, "step": 17515 }, { "epoch": 1.0873424793593642, "grad_norm": 0.5488169658788933, "learning_rate": 8.044595019053314e-05, "loss": 2.8074, "step": 17516 }, { "epoch": 1.0874045564591222, "grad_norm": 0.5874912261307735, "learning_rate": 8.044308533310627e-05, "loss": 2.9084, "step": 17517 }, { "epoch": 1.08746663355888, "grad_norm": 0.5948997207894875, "learning_rate": 8.044022031685044e-05, "loss": 2.7754, "step": 17518 }, { "epoch": 1.087528710658638, "grad_norm": 0.5371646921975901, "learning_rate": 8.043735514178057e-05, "loss": 2.8713, "step": 17519 }, { "epoch": 1.087590787758396, "grad_norm": 0.5374681418980085, "learning_rate": 8.043448980791165e-05, "loss": 2.8713, "step": 17520 }, { "epoch": 1.0876528648581538, "grad_norm": 0.5267618995991677, "learning_rate": 8.04316243152586e-05, "loss": 2.8268, "step": 17521 }, { "epoch": 1.0877149419579117, "grad_norm": 0.5566516540505511, "learning_rate": 8.04287586638364e-05, "loss": 2.9417, "step": 17522 }, { "epoch": 1.0877770190576697, "grad_norm": 0.5307240968511365, "learning_rate": 8.042589285365995e-05, "loss": 2.8839, "step": 17523 }, { "epoch": 1.0878390961574276, "grad_norm": 0.4895057430738959, "learning_rate": 8.042302688474424e-05, "loss": 2.9385, "step": 17524 }, { "epoch": 1.0879011732571855, "grad_norm": 0.4790499759338424, "learning_rate": 8.042016075710421e-05, "loss": 2.8748, "step": 17525 }, { "epoch": 1.0879632503569434, "grad_norm": 0.4670731752273923, "learning_rate": 8.041729447075483e-05, "loss": 2.9059, "step": 17526 }, { "epoch": 1.0880253274567013, "grad_norm": 0.49373534699921406, "learning_rate": 8.041442802571103e-05, "loss": 2.8924, "step": 17527 }, { "epoch": 1.088087404556459, "grad_norm": 0.47492921515125014, "learning_rate": 8.041156142198778e-05, "loss": 2.8124, "step": 17528 }, { "epoch": 1.088149481656217, "grad_norm": 0.46156391316625633, "learning_rate": 8.040869465960003e-05, "loss": 2.8492, "step": 17529 }, { "epoch": 1.0882115587559749, "grad_norm": 0.5038473109589291, "learning_rate": 8.040582773856274e-05, "loss": 2.8307, "step": 17530 }, { "epoch": 1.0882736358557328, "grad_norm": 0.5505257954249034, "learning_rate": 8.040296065889086e-05, "loss": 2.867, "step": 17531 }, { "epoch": 1.0883357129554907, "grad_norm": 0.48226214092608394, "learning_rate": 8.040009342059936e-05, "loss": 2.8933, "step": 17532 }, { "epoch": 1.0883977900552486, "grad_norm": 0.48669128952814283, "learning_rate": 8.039722602370318e-05, "loss": 2.8372, "step": 17533 }, { "epoch": 1.0884598671550065, "grad_norm": 0.533381533555682, "learning_rate": 8.03943584682173e-05, "loss": 2.9216, "step": 17534 }, { "epoch": 1.0885219442547645, "grad_norm": 0.4939453281871692, "learning_rate": 8.039149075415667e-05, "loss": 2.8631, "step": 17535 }, { "epoch": 1.0885840213545224, "grad_norm": 0.5278150924519575, "learning_rate": 8.038862288153627e-05, "loss": 2.9319, "step": 17536 }, { "epoch": 1.0886460984542803, "grad_norm": 0.5391361075746721, "learning_rate": 8.038575485037104e-05, "loss": 2.9833, "step": 17537 }, { "epoch": 1.0887081755540382, "grad_norm": 0.5769374351976664, "learning_rate": 8.038288666067596e-05, "loss": 2.8921, "step": 17538 }, { "epoch": 1.088770252653796, "grad_norm": 0.5540657050679598, "learning_rate": 8.038001831246595e-05, "loss": 2.9311, "step": 17539 }, { "epoch": 1.0888323297535538, "grad_norm": 0.5160265428310727, "learning_rate": 8.037714980575604e-05, "loss": 2.8005, "step": 17540 }, { "epoch": 1.0888944068533117, "grad_norm": 0.5433018712235343, "learning_rate": 8.037428114056116e-05, "loss": 2.9169, "step": 17541 }, { "epoch": 1.0889564839530697, "grad_norm": 0.4991288792244229, "learning_rate": 8.037141231689626e-05, "loss": 2.9287, "step": 17542 }, { "epoch": 1.0890185610528276, "grad_norm": 0.4736603719027715, "learning_rate": 8.036854333477635e-05, "loss": 2.8146, "step": 17543 }, { "epoch": 1.0890806381525855, "grad_norm": 0.5336622560014244, "learning_rate": 8.036567419421639e-05, "loss": 2.9115, "step": 17544 }, { "epoch": 1.0891427152523434, "grad_norm": 0.4773049277893406, "learning_rate": 8.036280489523132e-05, "loss": 2.8106, "step": 17545 }, { "epoch": 1.0892047923521013, "grad_norm": 0.5239136082069648, "learning_rate": 8.035993543783613e-05, "loss": 2.9048, "step": 17546 }, { "epoch": 1.0892668694518592, "grad_norm": 0.5422571857533793, "learning_rate": 8.035706582204579e-05, "loss": 2.9138, "step": 17547 }, { "epoch": 1.0893289465516172, "grad_norm": 0.4697794577146258, "learning_rate": 8.035419604787525e-05, "loss": 2.9451, "step": 17548 }, { "epoch": 1.089391023651375, "grad_norm": 0.4948955242846801, "learning_rate": 8.03513261153395e-05, "loss": 2.8635, "step": 17549 }, { "epoch": 1.089453100751133, "grad_norm": 0.4829419842019767, "learning_rate": 8.034845602445353e-05, "loss": 2.8187, "step": 17550 }, { "epoch": 1.089515177850891, "grad_norm": 0.4334273663459021, "learning_rate": 8.03455857752323e-05, "loss": 2.8663, "step": 17551 }, { "epoch": 1.0895772549506486, "grad_norm": 0.49576148973886563, "learning_rate": 8.034271536769078e-05, "loss": 2.8613, "step": 17552 }, { "epoch": 1.0896393320504065, "grad_norm": 0.5496701606154026, "learning_rate": 8.033984480184395e-05, "loss": 2.8244, "step": 17553 }, { "epoch": 1.0897014091501644, "grad_norm": 0.5102832840488759, "learning_rate": 8.033697407770677e-05, "loss": 2.9096, "step": 17554 }, { "epoch": 1.0897634862499224, "grad_norm": 0.465790753656308, "learning_rate": 8.033410319529424e-05, "loss": 2.8429, "step": 17555 }, { "epoch": 1.0898255633496803, "grad_norm": 0.5607530965511218, "learning_rate": 8.033123215462133e-05, "loss": 2.9072, "step": 17556 }, { "epoch": 1.0898876404494382, "grad_norm": 0.511712197997102, "learning_rate": 8.032836095570302e-05, "loss": 2.8744, "step": 17557 }, { "epoch": 1.0899497175491961, "grad_norm": 0.5020548795285339, "learning_rate": 8.032548959855429e-05, "loss": 2.776, "step": 17558 }, { "epoch": 1.090011794648954, "grad_norm": 0.5628786137376129, "learning_rate": 8.032261808319012e-05, "loss": 2.8205, "step": 17559 }, { "epoch": 1.090073871748712, "grad_norm": 0.4697146979278947, "learning_rate": 8.031974640962547e-05, "loss": 2.8989, "step": 17560 }, { "epoch": 1.0901359488484699, "grad_norm": 0.4828951595126203, "learning_rate": 8.031687457787537e-05, "loss": 2.8889, "step": 17561 }, { "epoch": 1.0901980259482278, "grad_norm": 0.4648636244684487, "learning_rate": 8.031400258795476e-05, "loss": 2.7865, "step": 17562 }, { "epoch": 1.0902601030479855, "grad_norm": 0.5164715724149418, "learning_rate": 8.031113043987866e-05, "loss": 2.9838, "step": 17563 }, { "epoch": 1.0903221801477434, "grad_norm": 0.5047870892033659, "learning_rate": 8.0308258133662e-05, "loss": 2.977, "step": 17564 }, { "epoch": 1.0903842572475013, "grad_norm": 0.4862270939685523, "learning_rate": 8.030538566931981e-05, "loss": 2.8477, "step": 17565 }, { "epoch": 1.0904463343472592, "grad_norm": 0.5533624322643323, "learning_rate": 8.030251304686709e-05, "loss": 2.9379, "step": 17566 }, { "epoch": 1.0905084114470172, "grad_norm": 0.539282132062861, "learning_rate": 8.029964026631878e-05, "loss": 2.7856, "step": 17567 }, { "epoch": 1.090570488546775, "grad_norm": 0.5109352345080997, "learning_rate": 8.02967673276899e-05, "loss": 2.8709, "step": 17568 }, { "epoch": 1.090632565646533, "grad_norm": 0.5364704007269838, "learning_rate": 8.029389423099543e-05, "loss": 2.873, "step": 17569 }, { "epoch": 1.090694642746291, "grad_norm": 0.5297566067786351, "learning_rate": 8.029102097625037e-05, "loss": 2.8829, "step": 17570 }, { "epoch": 1.0907567198460488, "grad_norm": 0.47633334419302975, "learning_rate": 8.028814756346969e-05, "loss": 2.8529, "step": 17571 }, { "epoch": 1.0908187969458067, "grad_norm": 0.5200358070369635, "learning_rate": 8.028527399266839e-05, "loss": 2.8654, "step": 17572 }, { "epoch": 1.0908808740455647, "grad_norm": 0.5175017678419499, "learning_rate": 8.028240026386146e-05, "loss": 2.8172, "step": 17573 }, { "epoch": 1.0909429511453226, "grad_norm": 0.5208503497840752, "learning_rate": 8.027952637706389e-05, "loss": 2.844, "step": 17574 }, { "epoch": 1.0910050282450805, "grad_norm": 0.47515177247014784, "learning_rate": 8.027665233229069e-05, "loss": 2.9083, "step": 17575 }, { "epoch": 1.0910671053448382, "grad_norm": 0.5522943099582329, "learning_rate": 8.027377812955685e-05, "loss": 2.8883, "step": 17576 }, { "epoch": 1.0911291824445961, "grad_norm": 0.45358525940047356, "learning_rate": 8.027090376887737e-05, "loss": 2.883, "step": 17577 }, { "epoch": 1.091191259544354, "grad_norm": 0.439111178133097, "learning_rate": 8.026802925026722e-05, "loss": 2.821, "step": 17578 }, { "epoch": 1.091253336644112, "grad_norm": 0.4096826089236789, "learning_rate": 8.026515457374141e-05, "loss": 2.9392, "step": 17579 }, { "epoch": 1.0913154137438699, "grad_norm": 0.47006717632997164, "learning_rate": 8.026227973931495e-05, "loss": 2.8896, "step": 17580 }, { "epoch": 1.0913774908436278, "grad_norm": 0.4460203668259107, "learning_rate": 8.025940474700282e-05, "loss": 2.8692, "step": 17581 }, { "epoch": 1.0914395679433857, "grad_norm": 0.4864813934621973, "learning_rate": 8.025652959682004e-05, "loss": 2.8487, "step": 17582 }, { "epoch": 1.0915016450431436, "grad_norm": 0.5134105815552641, "learning_rate": 8.02536542887816e-05, "loss": 2.9181, "step": 17583 }, { "epoch": 1.0915637221429015, "grad_norm": 0.48909741663982725, "learning_rate": 8.025077882290249e-05, "loss": 2.8913, "step": 17584 }, { "epoch": 1.0916257992426595, "grad_norm": 0.5004137830415596, "learning_rate": 8.024790319919773e-05, "loss": 2.9528, "step": 17585 }, { "epoch": 1.0916878763424174, "grad_norm": 0.5394871186547291, "learning_rate": 8.024502741768232e-05, "loss": 2.9912, "step": 17586 }, { "epoch": 1.091749953442175, "grad_norm": 0.5200397899424599, "learning_rate": 8.024215147837126e-05, "loss": 2.8473, "step": 17587 }, { "epoch": 1.091812030541933, "grad_norm": 0.48862551950313404, "learning_rate": 8.023927538127955e-05, "loss": 2.8798, "step": 17588 }, { "epoch": 1.091874107641691, "grad_norm": 0.53120876601021, "learning_rate": 8.02363991264222e-05, "loss": 2.8154, "step": 17589 }, { "epoch": 1.0919361847414488, "grad_norm": 0.49848898917657564, "learning_rate": 8.023352271381421e-05, "loss": 2.8324, "step": 17590 }, { "epoch": 1.0919982618412067, "grad_norm": 0.5634270287597004, "learning_rate": 8.02306461434706e-05, "loss": 2.8262, "step": 17591 }, { "epoch": 1.0920603389409647, "grad_norm": 0.5738601083586958, "learning_rate": 8.022776941540638e-05, "loss": 2.7563, "step": 17592 }, { "epoch": 1.0921224160407226, "grad_norm": 0.4620442922620019, "learning_rate": 8.022489252963654e-05, "loss": 2.8553, "step": 17593 }, { "epoch": 1.0921844931404805, "grad_norm": 0.5579546760572349, "learning_rate": 8.02220154861761e-05, "loss": 2.9317, "step": 17594 }, { "epoch": 1.0922465702402384, "grad_norm": 0.500202763690734, "learning_rate": 8.021913828504006e-05, "loss": 2.8704, "step": 17595 }, { "epoch": 1.0923086473399963, "grad_norm": 0.5227002602880343, "learning_rate": 8.021626092624345e-05, "loss": 2.9162, "step": 17596 }, { "epoch": 1.0923707244397542, "grad_norm": 0.5892146715879345, "learning_rate": 8.021338340980126e-05, "loss": 2.9227, "step": 17597 }, { "epoch": 1.0924328015395122, "grad_norm": 0.5177916752018228, "learning_rate": 8.021050573572853e-05, "loss": 2.8978, "step": 17598 }, { "epoch": 1.09249487863927, "grad_norm": 0.5433419406487984, "learning_rate": 8.020762790404024e-05, "loss": 2.9223, "step": 17599 }, { "epoch": 1.0925569557390278, "grad_norm": 0.5264240580960724, "learning_rate": 8.020474991475142e-05, "loss": 2.8923, "step": 17600 }, { "epoch": 1.0926190328387857, "grad_norm": 0.5381516764338863, "learning_rate": 8.02018717678771e-05, "loss": 2.9642, "step": 17601 }, { "epoch": 1.0926811099385436, "grad_norm": 0.5281891721753839, "learning_rate": 8.01989934634323e-05, "loss": 2.8171, "step": 17602 }, { "epoch": 1.0927431870383015, "grad_norm": 0.5502866594008727, "learning_rate": 8.019611500143198e-05, "loss": 2.895, "step": 17603 }, { "epoch": 1.0928052641380595, "grad_norm": 0.5219085339947968, "learning_rate": 8.019323638189123e-05, "loss": 2.9558, "step": 17604 }, { "epoch": 1.0928673412378174, "grad_norm": 0.5359236512466259, "learning_rate": 8.019035760482503e-05, "loss": 2.8156, "step": 17605 }, { "epoch": 1.0929294183375753, "grad_norm": 0.5491921788783976, "learning_rate": 8.01874786702484e-05, "loss": 2.7475, "step": 17606 }, { "epoch": 1.0929914954373332, "grad_norm": 0.48655091969571007, "learning_rate": 8.018459957817636e-05, "loss": 2.871, "step": 17607 }, { "epoch": 1.0930535725370911, "grad_norm": 0.5412368872818313, "learning_rate": 8.018172032862395e-05, "loss": 2.83, "step": 17608 }, { "epoch": 1.093115649636849, "grad_norm": 0.4878525797759791, "learning_rate": 8.017884092160617e-05, "loss": 2.957, "step": 17609 }, { "epoch": 1.093177726736607, "grad_norm": 0.5133188869664662, "learning_rate": 8.017596135713806e-05, "loss": 2.8731, "step": 17610 }, { "epoch": 1.0932398038363647, "grad_norm": 0.4537328556331842, "learning_rate": 8.01730816352346e-05, "loss": 2.7941, "step": 17611 }, { "epoch": 1.0933018809361226, "grad_norm": 0.5086788416874483, "learning_rate": 8.017020175591087e-05, "loss": 2.9914, "step": 17612 }, { "epoch": 1.0933639580358805, "grad_norm": 0.5348284197568033, "learning_rate": 8.016732171918189e-05, "loss": 2.8198, "step": 17613 }, { "epoch": 1.0934260351356384, "grad_norm": 0.4331779720341159, "learning_rate": 8.016444152506266e-05, "loss": 2.8496, "step": 17614 }, { "epoch": 1.0934881122353963, "grad_norm": 0.483453011562891, "learning_rate": 8.01615611735682e-05, "loss": 2.8758, "step": 17615 }, { "epoch": 1.0935501893351542, "grad_norm": 0.47886689974994573, "learning_rate": 8.015868066471357e-05, "loss": 2.8689, "step": 17616 }, { "epoch": 1.0936122664349122, "grad_norm": 0.5031477075984955, "learning_rate": 8.015579999851379e-05, "loss": 2.8537, "step": 17617 }, { "epoch": 1.09367434353467, "grad_norm": 0.479039306576857, "learning_rate": 8.015291917498386e-05, "loss": 2.8813, "step": 17618 }, { "epoch": 1.093736420634428, "grad_norm": 0.49804141471711955, "learning_rate": 8.015003819413885e-05, "loss": 2.8446, "step": 17619 }, { "epoch": 1.093798497734186, "grad_norm": 0.5737464415579091, "learning_rate": 8.014715705599374e-05, "loss": 2.9246, "step": 17620 }, { "epoch": 1.0938605748339438, "grad_norm": 0.5180149178621365, "learning_rate": 8.014427576056362e-05, "loss": 2.9477, "step": 17621 }, { "epoch": 1.0939226519337018, "grad_norm": 0.5557594954828767, "learning_rate": 8.014139430786349e-05, "loss": 2.7673, "step": 17622 }, { "epoch": 1.0939847290334597, "grad_norm": 0.5360692721376425, "learning_rate": 8.013851269790839e-05, "loss": 2.8944, "step": 17623 }, { "epoch": 1.0940468061332174, "grad_norm": 0.5513571575791645, "learning_rate": 8.013563093071335e-05, "loss": 2.8854, "step": 17624 }, { "epoch": 1.0941088832329753, "grad_norm": 0.5238261132482576, "learning_rate": 8.013274900629341e-05, "loss": 2.8959, "step": 17625 }, { "epoch": 1.0941709603327332, "grad_norm": 0.5146680390246201, "learning_rate": 8.01298669246636e-05, "loss": 2.8027, "step": 17626 }, { "epoch": 1.0942330374324911, "grad_norm": 0.5721648456249302, "learning_rate": 8.012698468583897e-05, "loss": 2.93, "step": 17627 }, { "epoch": 1.094295114532249, "grad_norm": 0.6048045052578184, "learning_rate": 8.012410228983453e-05, "loss": 2.8127, "step": 17628 }, { "epoch": 1.094357191632007, "grad_norm": 0.5568343511402944, "learning_rate": 8.012121973666536e-05, "loss": 2.9349, "step": 17629 }, { "epoch": 1.0944192687317649, "grad_norm": 0.5951570352923199, "learning_rate": 8.011833702634644e-05, "loss": 2.9162, "step": 17630 }, { "epoch": 1.0944813458315228, "grad_norm": 0.5720382863799577, "learning_rate": 8.011545415889287e-05, "loss": 2.9645, "step": 17631 }, { "epoch": 1.0945434229312807, "grad_norm": 0.4642324675822248, "learning_rate": 8.011257113431966e-05, "loss": 2.841, "step": 17632 }, { "epoch": 1.0946055000310386, "grad_norm": 0.5534385535964113, "learning_rate": 8.010968795264185e-05, "loss": 2.835, "step": 17633 }, { "epoch": 1.0946675771307965, "grad_norm": 0.6732553895209774, "learning_rate": 8.010680461387449e-05, "loss": 2.9314, "step": 17634 }, { "epoch": 1.0947296542305542, "grad_norm": 0.5131885880391494, "learning_rate": 8.010392111803263e-05, "loss": 2.9068, "step": 17635 }, { "epoch": 1.0947917313303122, "grad_norm": 0.4782639133385726, "learning_rate": 8.01010374651313e-05, "loss": 2.8253, "step": 17636 }, { "epoch": 1.09485380843007, "grad_norm": 0.5154146574603644, "learning_rate": 8.009815365518554e-05, "loss": 2.8345, "step": 17637 }, { "epoch": 1.094915885529828, "grad_norm": 0.5193983209104722, "learning_rate": 8.009526968821043e-05, "loss": 2.9309, "step": 17638 }, { "epoch": 1.094977962629586, "grad_norm": 0.526757005712834, "learning_rate": 8.009238556422098e-05, "loss": 2.9103, "step": 17639 }, { "epoch": 1.0950400397293438, "grad_norm": 0.4767148759222515, "learning_rate": 8.008950128323225e-05, "loss": 2.8658, "step": 17640 }, { "epoch": 1.0951021168291017, "grad_norm": 0.5025738392378387, "learning_rate": 8.008661684525927e-05, "loss": 2.9651, "step": 17641 }, { "epoch": 1.0951641939288597, "grad_norm": 0.502181123836657, "learning_rate": 8.008373225031713e-05, "loss": 2.8858, "step": 17642 }, { "epoch": 1.0952262710286176, "grad_norm": 0.5279724611827589, "learning_rate": 8.008084749842085e-05, "loss": 2.9421, "step": 17643 }, { "epoch": 1.0952883481283755, "grad_norm": 0.4951275856385858, "learning_rate": 8.007796258958549e-05, "loss": 2.8862, "step": 17644 }, { "epoch": 1.0953504252281334, "grad_norm": 0.44793353122929275, "learning_rate": 8.007507752382608e-05, "loss": 2.9778, "step": 17645 }, { "epoch": 1.0954125023278913, "grad_norm": 0.5502482060095986, "learning_rate": 8.00721923011577e-05, "loss": 2.8193, "step": 17646 }, { "epoch": 1.0954745794276493, "grad_norm": 0.5379717475481862, "learning_rate": 8.006930692159539e-05, "loss": 2.8366, "step": 17647 }, { "epoch": 1.095536656527407, "grad_norm": 0.5327016127750108, "learning_rate": 8.006642138515422e-05, "loss": 2.9308, "step": 17648 }, { "epoch": 1.0955987336271649, "grad_norm": 0.5122595397392304, "learning_rate": 8.00635356918492e-05, "loss": 2.7907, "step": 17649 }, { "epoch": 1.0956608107269228, "grad_norm": 0.5105137213731986, "learning_rate": 8.006064984169546e-05, "loss": 2.8253, "step": 17650 }, { "epoch": 1.0957228878266807, "grad_norm": 0.5256158849263771, "learning_rate": 8.005776383470798e-05, "loss": 2.7953, "step": 17651 }, { "epoch": 1.0957849649264386, "grad_norm": 0.5451756095127476, "learning_rate": 8.005487767090185e-05, "loss": 2.9221, "step": 17652 }, { "epoch": 1.0958470420261965, "grad_norm": 0.5322958244962603, "learning_rate": 8.005199135029215e-05, "loss": 2.9425, "step": 17653 }, { "epoch": 1.0959091191259545, "grad_norm": 0.5242770849024828, "learning_rate": 8.00491048728939e-05, "loss": 3.0058, "step": 17654 }, { "epoch": 1.0959711962257124, "grad_norm": 0.5367883451114692, "learning_rate": 8.004621823872216e-05, "loss": 2.8393, "step": 17655 }, { "epoch": 1.0960332733254703, "grad_norm": 0.535355022854844, "learning_rate": 8.004333144779203e-05, "loss": 2.86, "step": 17656 }, { "epoch": 1.0960953504252282, "grad_norm": 0.46329438832583697, "learning_rate": 8.004044450011854e-05, "loss": 2.873, "step": 17657 }, { "epoch": 1.0961574275249861, "grad_norm": 0.5158122040740698, "learning_rate": 8.003755739571676e-05, "loss": 2.9054, "step": 17658 }, { "epoch": 1.0962195046247438, "grad_norm": 0.5368765586501778, "learning_rate": 8.003467013460174e-05, "loss": 2.8459, "step": 17659 }, { "epoch": 1.0962815817245017, "grad_norm": 0.451975103320322, "learning_rate": 8.003178271678857e-05, "loss": 2.8127, "step": 17660 }, { "epoch": 1.0963436588242597, "grad_norm": 0.5552553037727623, "learning_rate": 8.002889514229228e-05, "loss": 2.9032, "step": 17661 }, { "epoch": 1.0964057359240176, "grad_norm": 0.49767320864540865, "learning_rate": 8.002600741112798e-05, "loss": 2.9172, "step": 17662 }, { "epoch": 1.0964678130237755, "grad_norm": 0.46898135198088764, "learning_rate": 8.00231195233107e-05, "loss": 2.7942, "step": 17663 }, { "epoch": 1.0965298901235334, "grad_norm": 0.4187946893559339, "learning_rate": 8.002023147885549e-05, "loss": 2.8952, "step": 17664 }, { "epoch": 1.0965919672232913, "grad_norm": 0.46758333297289395, "learning_rate": 8.001734327777748e-05, "loss": 2.7962, "step": 17665 }, { "epoch": 1.0966540443230492, "grad_norm": 0.4411871205825568, "learning_rate": 8.001445492009168e-05, "loss": 2.9548, "step": 17666 }, { "epoch": 1.0967161214228072, "grad_norm": 0.5164638689130973, "learning_rate": 8.001156640581319e-05, "loss": 2.9149, "step": 17667 }, { "epoch": 1.096778198522565, "grad_norm": 0.480229558565268, "learning_rate": 8.000867773495709e-05, "loss": 2.8301, "step": 17668 }, { "epoch": 1.096840275622323, "grad_norm": 0.5426785358978922, "learning_rate": 8.000578890753842e-05, "loss": 2.9503, "step": 17669 }, { "epoch": 1.096902352722081, "grad_norm": 0.4929296710791775, "learning_rate": 8.000289992357226e-05, "loss": 2.8588, "step": 17670 }, { "epoch": 1.0969644298218388, "grad_norm": 0.4784062474643439, "learning_rate": 8.000001078307369e-05, "loss": 2.8526, "step": 17671 }, { "epoch": 1.0970265069215965, "grad_norm": 0.5340724467807147, "learning_rate": 7.999712148605779e-05, "loss": 2.9237, "step": 17672 }, { "epoch": 1.0970885840213545, "grad_norm": 0.4633157604673491, "learning_rate": 7.999423203253962e-05, "loss": 2.7972, "step": 17673 }, { "epoch": 1.0971506611211124, "grad_norm": 0.4934525807246224, "learning_rate": 7.999134242253426e-05, "loss": 2.9621, "step": 17674 }, { "epoch": 1.0972127382208703, "grad_norm": 0.44697913110371806, "learning_rate": 7.998845265605677e-05, "loss": 2.8455, "step": 17675 }, { "epoch": 1.0972748153206282, "grad_norm": 0.4762638438207661, "learning_rate": 7.998556273312225e-05, "loss": 2.7747, "step": 17676 }, { "epoch": 1.0973368924203861, "grad_norm": 0.4216221652214144, "learning_rate": 7.99826726537458e-05, "loss": 2.8071, "step": 17677 }, { "epoch": 1.097398969520144, "grad_norm": 0.44717121377181146, "learning_rate": 7.997978241794243e-05, "loss": 2.9428, "step": 17678 }, { "epoch": 1.097461046619902, "grad_norm": 0.4639251512150519, "learning_rate": 7.997689202572729e-05, "loss": 2.959, "step": 17679 }, { "epoch": 1.0975231237196599, "grad_norm": 0.492615196049163, "learning_rate": 7.997400147711539e-05, "loss": 2.9095, "step": 17680 }, { "epoch": 1.0975852008194178, "grad_norm": 0.49529111316609653, "learning_rate": 7.997111077212187e-05, "loss": 2.8208, "step": 17681 }, { "epoch": 1.0976472779191757, "grad_norm": 0.5253509166134079, "learning_rate": 7.996821991076178e-05, "loss": 2.9027, "step": 17682 }, { "epoch": 1.0977093550189334, "grad_norm": 0.48308122804248893, "learning_rate": 7.996532889305022e-05, "loss": 2.9325, "step": 17683 }, { "epoch": 1.0977714321186913, "grad_norm": 0.4486697615610729, "learning_rate": 7.996243771900227e-05, "loss": 2.905, "step": 17684 }, { "epoch": 1.0978335092184492, "grad_norm": 0.492576294183921, "learning_rate": 7.995954638863299e-05, "loss": 2.9027, "step": 17685 }, { "epoch": 1.0978955863182072, "grad_norm": 0.452761257449796, "learning_rate": 7.995665490195751e-05, "loss": 2.8328, "step": 17686 }, { "epoch": 1.097957663417965, "grad_norm": 0.489191001244021, "learning_rate": 7.995376325899087e-05, "loss": 2.8423, "step": 17687 }, { "epoch": 1.098019740517723, "grad_norm": 0.49047420152330573, "learning_rate": 7.995087145974817e-05, "loss": 2.8923, "step": 17688 }, { "epoch": 1.098081817617481, "grad_norm": 0.46859991531920797, "learning_rate": 7.994797950424453e-05, "loss": 2.929, "step": 17689 }, { "epoch": 1.0981438947172388, "grad_norm": 0.5560240157913177, "learning_rate": 7.994508739249498e-05, "loss": 2.8983, "step": 17690 }, { "epoch": 1.0982059718169968, "grad_norm": 0.5048510543308996, "learning_rate": 7.994219512451466e-05, "loss": 2.85, "step": 17691 }, { "epoch": 1.0982680489167547, "grad_norm": 0.509824581317826, "learning_rate": 7.993930270031864e-05, "loss": 2.9096, "step": 17692 }, { "epoch": 1.0983301260165126, "grad_norm": 0.4879322330517478, "learning_rate": 7.993641011992199e-05, "loss": 2.8724, "step": 17693 }, { "epoch": 1.0983922031162705, "grad_norm": 0.525987194249881, "learning_rate": 7.993351738333983e-05, "loss": 2.9019, "step": 17694 }, { "epoch": 1.0984542802160284, "grad_norm": 0.5200353772251943, "learning_rate": 7.993062449058726e-05, "loss": 2.8459, "step": 17695 }, { "epoch": 1.0985163573157861, "grad_norm": 0.5339276767145453, "learning_rate": 7.992773144167933e-05, "loss": 2.8679, "step": 17696 }, { "epoch": 1.098578434415544, "grad_norm": 0.5198674655706785, "learning_rate": 7.992483823663118e-05, "loss": 2.8958, "step": 17697 }, { "epoch": 1.098640511515302, "grad_norm": 0.4931375628612871, "learning_rate": 7.992194487545787e-05, "loss": 2.9014, "step": 17698 }, { "epoch": 1.0987025886150599, "grad_norm": 0.49166816984636513, "learning_rate": 7.991905135817453e-05, "loss": 2.7888, "step": 17699 }, { "epoch": 1.0987646657148178, "grad_norm": 0.5196480870627448, "learning_rate": 7.991615768479621e-05, "loss": 2.8901, "step": 17700 }, { "epoch": 1.0988267428145757, "grad_norm": 0.5659864514232266, "learning_rate": 7.991326385533804e-05, "loss": 2.9278, "step": 17701 }, { "epoch": 1.0988888199143336, "grad_norm": 0.5247192347314561, "learning_rate": 7.991036986981513e-05, "loss": 2.905, "step": 17702 }, { "epoch": 1.0989508970140915, "grad_norm": 0.5555862881849043, "learning_rate": 7.990747572824253e-05, "loss": 2.8877, "step": 17703 }, { "epoch": 1.0990129741138495, "grad_norm": 0.5346108044587519, "learning_rate": 7.99045814306354e-05, "loss": 2.8904, "step": 17704 }, { "epoch": 1.0990750512136074, "grad_norm": 0.5865602617012773, "learning_rate": 7.99016869770088e-05, "loss": 2.8991, "step": 17705 }, { "epoch": 1.0991371283133653, "grad_norm": 0.5083567417206134, "learning_rate": 7.989879236737783e-05, "loss": 2.8138, "step": 17706 }, { "epoch": 1.099199205413123, "grad_norm": 0.5774755438460668, "learning_rate": 7.989589760175759e-05, "loss": 2.9104, "step": 17707 }, { "epoch": 1.099261282512881, "grad_norm": 0.5038016454665241, "learning_rate": 7.989300268016322e-05, "loss": 2.911, "step": 17708 }, { "epoch": 1.0993233596126388, "grad_norm": 0.5072716400245768, "learning_rate": 7.98901076026098e-05, "loss": 2.9221, "step": 17709 }, { "epoch": 1.0993854367123967, "grad_norm": 0.5957201595684938, "learning_rate": 7.988721236911241e-05, "loss": 2.9152, "step": 17710 }, { "epoch": 1.0994475138121547, "grad_norm": 0.5180204984151295, "learning_rate": 7.988431697968619e-05, "loss": 2.8736, "step": 17711 }, { "epoch": 1.0995095909119126, "grad_norm": 0.5073367310964534, "learning_rate": 7.988142143434623e-05, "loss": 2.9245, "step": 17712 }, { "epoch": 1.0995716680116705, "grad_norm": 0.5012232598761202, "learning_rate": 7.987852573310764e-05, "loss": 2.756, "step": 17713 }, { "epoch": 1.0996337451114284, "grad_norm": 0.5359296292096831, "learning_rate": 7.987562987598553e-05, "loss": 2.8734, "step": 17714 }, { "epoch": 1.0996958222111863, "grad_norm": 0.5283978699679474, "learning_rate": 7.987273386299501e-05, "loss": 2.9149, "step": 17715 }, { "epoch": 1.0997578993109443, "grad_norm": 0.5131108806865299, "learning_rate": 7.986983769415118e-05, "loss": 2.8259, "step": 17716 }, { "epoch": 1.0998199764107022, "grad_norm": 0.47938505323827607, "learning_rate": 7.986694136946918e-05, "loss": 2.8733, "step": 17717 }, { "epoch": 1.09988205351046, "grad_norm": 0.5663010598207957, "learning_rate": 7.986404488896407e-05, "loss": 2.8379, "step": 17718 }, { "epoch": 1.099944130610218, "grad_norm": 0.5101004191717906, "learning_rate": 7.986114825265099e-05, "loss": 2.8843, "step": 17719 }, { "epoch": 1.1000062077099757, "grad_norm": 0.5367612230578295, "learning_rate": 7.985825146054505e-05, "loss": 2.7504, "step": 17720 }, { "epoch": 1.1000682848097336, "grad_norm": 0.5686087045068361, "learning_rate": 7.985535451266136e-05, "loss": 2.939, "step": 17721 }, { "epoch": 1.1001303619094915, "grad_norm": 0.47366386390441145, "learning_rate": 7.985245740901506e-05, "loss": 2.8029, "step": 17722 }, { "epoch": 1.1001924390092495, "grad_norm": 0.5019789811536588, "learning_rate": 7.984956014962121e-05, "loss": 2.8689, "step": 17723 }, { "epoch": 1.1002545161090074, "grad_norm": 0.48151061445046117, "learning_rate": 7.984666273449499e-05, "loss": 2.8503, "step": 17724 }, { "epoch": 1.1003165932087653, "grad_norm": 0.5338800067385496, "learning_rate": 7.984376516365146e-05, "loss": 2.836, "step": 17725 }, { "epoch": 1.1003786703085232, "grad_norm": 0.5007675954590765, "learning_rate": 7.98408674371058e-05, "loss": 2.8577, "step": 17726 }, { "epoch": 1.1004407474082811, "grad_norm": 0.5248715981638722, "learning_rate": 7.983796955487305e-05, "loss": 2.8713, "step": 17727 }, { "epoch": 1.100502824508039, "grad_norm": 0.4712136377021479, "learning_rate": 7.983507151696838e-05, "loss": 2.7793, "step": 17728 }, { "epoch": 1.100564901607797, "grad_norm": 0.5373809959944931, "learning_rate": 7.98321733234069e-05, "loss": 2.9654, "step": 17729 }, { "epoch": 1.1006269787075549, "grad_norm": 0.4889861702387045, "learning_rate": 7.982927497420374e-05, "loss": 2.9633, "step": 17730 }, { "epoch": 1.1006890558073126, "grad_norm": 0.4757304510549313, "learning_rate": 7.9826376469374e-05, "loss": 2.7565, "step": 17731 }, { "epoch": 1.1007511329070705, "grad_norm": 0.5512297136981257, "learning_rate": 7.982347780893283e-05, "loss": 2.788, "step": 17732 }, { "epoch": 1.1008132100068284, "grad_norm": 0.4376874590215537, "learning_rate": 7.982057899289531e-05, "loss": 2.8488, "step": 17733 }, { "epoch": 1.1008752871065863, "grad_norm": 0.46990546231927777, "learning_rate": 7.98176800212766e-05, "loss": 2.8523, "step": 17734 }, { "epoch": 1.1009373642063442, "grad_norm": 0.4796601290768711, "learning_rate": 7.981478089409183e-05, "loss": 2.8902, "step": 17735 }, { "epoch": 1.1009994413061022, "grad_norm": 0.5230525864737291, "learning_rate": 7.981188161135609e-05, "loss": 2.8694, "step": 17736 }, { "epoch": 1.10106151840586, "grad_norm": 0.5111425519109355, "learning_rate": 7.980898217308454e-05, "loss": 2.9704, "step": 17737 }, { "epoch": 1.101123595505618, "grad_norm": 0.4503307676613996, "learning_rate": 7.980608257929228e-05, "loss": 2.8401, "step": 17738 }, { "epoch": 1.101185672605376, "grad_norm": 0.5030232166616424, "learning_rate": 7.980318282999447e-05, "loss": 2.875, "step": 17739 }, { "epoch": 1.1012477497051338, "grad_norm": 0.4763419000311524, "learning_rate": 7.98002829252062e-05, "loss": 2.9575, "step": 17740 }, { "epoch": 1.1013098268048918, "grad_norm": 0.5248715413836537, "learning_rate": 7.979738286494265e-05, "loss": 2.9411, "step": 17741 }, { "epoch": 1.1013719039046497, "grad_norm": 0.5035184506499064, "learning_rate": 7.979448264921889e-05, "loss": 2.9435, "step": 17742 }, { "epoch": 1.1014339810044076, "grad_norm": 0.49016088717610706, "learning_rate": 7.979158227805009e-05, "loss": 2.8731, "step": 17743 }, { "epoch": 1.1014960581041653, "grad_norm": 0.5438716434470748, "learning_rate": 7.978868175145138e-05, "loss": 2.8707, "step": 17744 }, { "epoch": 1.1015581352039232, "grad_norm": 0.5096964295877978, "learning_rate": 7.978578106943788e-05, "loss": 2.7852, "step": 17745 }, { "epoch": 1.1016202123036811, "grad_norm": 0.519106095818325, "learning_rate": 7.978288023202471e-05, "loss": 2.7319, "step": 17746 }, { "epoch": 1.101682289403439, "grad_norm": 0.47918546853978883, "learning_rate": 7.977997923922707e-05, "loss": 2.8867, "step": 17747 }, { "epoch": 1.101744366503197, "grad_norm": 0.49091800578147393, "learning_rate": 7.977707809106e-05, "loss": 2.7963, "step": 17748 }, { "epoch": 1.1018064436029549, "grad_norm": 0.48346102530922763, "learning_rate": 7.977417678753871e-05, "loss": 2.942, "step": 17749 }, { "epoch": 1.1018685207027128, "grad_norm": 0.4452824331473056, "learning_rate": 7.977127532867832e-05, "loss": 2.905, "step": 17750 }, { "epoch": 1.1019305978024707, "grad_norm": 0.5321446066934604, "learning_rate": 7.976837371449395e-05, "loss": 2.9677, "step": 17751 }, { "epoch": 1.1019926749022286, "grad_norm": 0.5510244982183596, "learning_rate": 7.976547194500075e-05, "loss": 2.9008, "step": 17752 }, { "epoch": 1.1020547520019865, "grad_norm": 0.46199680095448536, "learning_rate": 7.976257002021385e-05, "loss": 2.939, "step": 17753 }, { "epoch": 1.1021168291017445, "grad_norm": 0.5615186077407781, "learning_rate": 7.97596679401484e-05, "loss": 2.7792, "step": 17754 }, { "epoch": 1.1021789062015022, "grad_norm": 0.4512172362408324, "learning_rate": 7.975676570481954e-05, "loss": 2.9195, "step": 17755 }, { "epoch": 1.10224098330126, "grad_norm": 0.5254867068786233, "learning_rate": 7.975386331424241e-05, "loss": 2.7517, "step": 17756 }, { "epoch": 1.102303060401018, "grad_norm": 0.5285161020484006, "learning_rate": 7.975096076843215e-05, "loss": 2.9728, "step": 17757 }, { "epoch": 1.102365137500776, "grad_norm": 0.49550360774450364, "learning_rate": 7.974805806740391e-05, "loss": 2.8278, "step": 17758 }, { "epoch": 1.1024272146005338, "grad_norm": 0.6078045979356342, "learning_rate": 7.974515521117283e-05, "loss": 2.8768, "step": 17759 }, { "epoch": 1.1024892917002918, "grad_norm": 0.48851713967587973, "learning_rate": 7.974225219975404e-05, "loss": 2.8867, "step": 17760 }, { "epoch": 1.1025513688000497, "grad_norm": 0.5303126087087453, "learning_rate": 7.973934903316271e-05, "loss": 2.8689, "step": 17761 }, { "epoch": 1.1026134458998076, "grad_norm": 0.5227626606681903, "learning_rate": 7.973644571141397e-05, "loss": 2.8171, "step": 17762 }, { "epoch": 1.1026755229995655, "grad_norm": 0.5513088053523397, "learning_rate": 7.973354223452298e-05, "loss": 2.8791, "step": 17763 }, { "epoch": 1.1027376000993234, "grad_norm": 0.4990460862844708, "learning_rate": 7.973063860250489e-05, "loss": 2.8642, "step": 17764 }, { "epoch": 1.1027996771990813, "grad_norm": 0.5387434360697292, "learning_rate": 7.972773481537484e-05, "loss": 2.8175, "step": 17765 }, { "epoch": 1.1028617542988393, "grad_norm": 0.5274310004499126, "learning_rate": 7.972483087314798e-05, "loss": 2.8985, "step": 17766 }, { "epoch": 1.102923831398597, "grad_norm": 0.5733844639005624, "learning_rate": 7.972192677583944e-05, "loss": 2.9091, "step": 17767 }, { "epoch": 1.1029859084983549, "grad_norm": 0.5298416037306518, "learning_rate": 7.971902252346442e-05, "loss": 2.8641, "step": 17768 }, { "epoch": 1.1030479855981128, "grad_norm": 0.48607161471206256, "learning_rate": 7.971611811603804e-05, "loss": 2.8107, "step": 17769 }, { "epoch": 1.1031100626978707, "grad_norm": 0.6248338240006539, "learning_rate": 7.971321355357545e-05, "loss": 2.861, "step": 17770 }, { "epoch": 1.1031721397976286, "grad_norm": 0.5520396905119348, "learning_rate": 7.97103088360918e-05, "loss": 2.9054, "step": 17771 }, { "epoch": 1.1032342168973865, "grad_norm": 0.5163204531590997, "learning_rate": 7.970740396360226e-05, "loss": 2.8916, "step": 17772 }, { "epoch": 1.1032962939971445, "grad_norm": 0.5785002650373242, "learning_rate": 7.9704498936122e-05, "loss": 2.8959, "step": 17773 }, { "epoch": 1.1033583710969024, "grad_norm": 0.5363038281898928, "learning_rate": 7.970159375366615e-05, "loss": 2.875, "step": 17774 }, { "epoch": 1.1034204481966603, "grad_norm": 0.45724594746478514, "learning_rate": 7.969868841624988e-05, "loss": 2.8445, "step": 17775 }, { "epoch": 1.1034825252964182, "grad_norm": 0.6109632189086028, "learning_rate": 7.969578292388833e-05, "loss": 2.9085, "step": 17776 }, { "epoch": 1.1035446023961761, "grad_norm": 0.42267941677605964, "learning_rate": 7.969287727659669e-05, "loss": 2.832, "step": 17777 }, { "epoch": 1.103606679495934, "grad_norm": 0.5592037951208981, "learning_rate": 7.968997147439008e-05, "loss": 2.9249, "step": 17778 }, { "epoch": 1.1036687565956917, "grad_norm": 0.5619976396562011, "learning_rate": 7.96870655172837e-05, "loss": 2.8624, "step": 17779 }, { "epoch": 1.1037308336954497, "grad_norm": 0.4367142842564328, "learning_rate": 7.968415940529267e-05, "loss": 2.8018, "step": 17780 }, { "epoch": 1.1037929107952076, "grad_norm": 0.5629041862045471, "learning_rate": 7.968125313843218e-05, "loss": 2.9268, "step": 17781 }, { "epoch": 1.1038549878949655, "grad_norm": 0.670502613913087, "learning_rate": 7.96783467167174e-05, "loss": 2.8386, "step": 17782 }, { "epoch": 1.1039170649947234, "grad_norm": 0.49649125041147824, "learning_rate": 7.967544014016348e-05, "loss": 2.9081, "step": 17783 }, { "epoch": 1.1039791420944813, "grad_norm": 0.5562020431162109, "learning_rate": 7.967253340878557e-05, "loss": 2.8971, "step": 17784 }, { "epoch": 1.1040412191942393, "grad_norm": 0.5777933097115977, "learning_rate": 7.966962652259887e-05, "loss": 2.8922, "step": 17785 }, { "epoch": 1.1041032962939972, "grad_norm": 0.5168633473753743, "learning_rate": 7.966671948161852e-05, "loss": 2.8896, "step": 17786 }, { "epoch": 1.104165373393755, "grad_norm": 0.5056056266251274, "learning_rate": 7.966381228585968e-05, "loss": 2.8194, "step": 17787 }, { "epoch": 1.104227450493513, "grad_norm": 0.5541194102089041, "learning_rate": 7.966090493533755e-05, "loss": 2.8734, "step": 17788 }, { "epoch": 1.104289527593271, "grad_norm": 0.47343922227602886, "learning_rate": 7.965799743006726e-05, "loss": 2.8746, "step": 17789 }, { "epoch": 1.1043516046930288, "grad_norm": 0.6000822855235648, "learning_rate": 7.965508977006402e-05, "loss": 2.8143, "step": 17790 }, { "epoch": 1.1044136817927865, "grad_norm": 0.5177602195707394, "learning_rate": 7.965218195534295e-05, "loss": 2.798, "step": 17791 }, { "epoch": 1.1044757588925445, "grad_norm": 0.5228125414075909, "learning_rate": 7.964927398591926e-05, "loss": 2.8656, "step": 17792 }, { "epoch": 1.1045378359923024, "grad_norm": 0.4713703033326581, "learning_rate": 7.964636586180811e-05, "loss": 2.8719, "step": 17793 }, { "epoch": 1.1045999130920603, "grad_norm": 0.5044773916333661, "learning_rate": 7.964345758302468e-05, "loss": 2.9044, "step": 17794 }, { "epoch": 1.1046619901918182, "grad_norm": 0.47721129170122, "learning_rate": 7.964054914958413e-05, "loss": 2.8479, "step": 17795 }, { "epoch": 1.1047240672915761, "grad_norm": 0.45976037514281315, "learning_rate": 7.963764056150162e-05, "loss": 2.8063, "step": 17796 }, { "epoch": 1.104786144391334, "grad_norm": 0.5291729602539583, "learning_rate": 7.963473181879238e-05, "loss": 2.9205, "step": 17797 }, { "epoch": 1.104848221491092, "grad_norm": 0.503903440718989, "learning_rate": 7.963182292147154e-05, "loss": 2.8326, "step": 17798 }, { "epoch": 1.1049102985908499, "grad_norm": 0.49603412299873373, "learning_rate": 7.962891386955427e-05, "loss": 2.7958, "step": 17799 }, { "epoch": 1.1049723756906078, "grad_norm": 0.48756886203733923, "learning_rate": 7.962600466305576e-05, "loss": 2.8976, "step": 17800 }, { "epoch": 1.1050344527903657, "grad_norm": 0.5458335769691604, "learning_rate": 7.962309530199121e-05, "loss": 2.9382, "step": 17801 }, { "epoch": 1.1050965298901236, "grad_norm": 0.4799435078388572, "learning_rate": 7.962018578637577e-05, "loss": 2.9187, "step": 17802 }, { "epoch": 1.1051586069898813, "grad_norm": 0.5082966770442495, "learning_rate": 7.961727611622465e-05, "loss": 2.839, "step": 17803 }, { "epoch": 1.1052206840896392, "grad_norm": 0.5518582413009205, "learning_rate": 7.9614366291553e-05, "loss": 2.9462, "step": 17804 }, { "epoch": 1.1052827611893972, "grad_norm": 0.573468555160422, "learning_rate": 7.961145631237601e-05, "loss": 2.8651, "step": 17805 }, { "epoch": 1.105344838289155, "grad_norm": 0.525444737026651, "learning_rate": 7.960854617870886e-05, "loss": 2.8542, "step": 17806 }, { "epoch": 1.105406915388913, "grad_norm": 0.5178214311030964, "learning_rate": 7.960563589056673e-05, "loss": 2.8084, "step": 17807 }, { "epoch": 1.105468992488671, "grad_norm": 0.5846522922813154, "learning_rate": 7.960272544796483e-05, "loss": 2.9405, "step": 17808 }, { "epoch": 1.1055310695884288, "grad_norm": 0.4954776693098888, "learning_rate": 7.959981485091832e-05, "loss": 2.8866, "step": 17809 }, { "epoch": 1.1055931466881868, "grad_norm": 0.5105560430332541, "learning_rate": 7.959690409944238e-05, "loss": 2.8623, "step": 17810 }, { "epoch": 1.1056552237879447, "grad_norm": 0.4957493014280213, "learning_rate": 7.95939931935522e-05, "loss": 2.9311, "step": 17811 }, { "epoch": 1.1057173008877026, "grad_norm": 0.5381638319955462, "learning_rate": 7.959108213326299e-05, "loss": 2.9292, "step": 17812 }, { "epoch": 1.1057793779874605, "grad_norm": 0.5100027241821087, "learning_rate": 7.958817091858992e-05, "loss": 2.9165, "step": 17813 }, { "epoch": 1.1058414550872184, "grad_norm": 0.5130346428679438, "learning_rate": 7.958525954954816e-05, "loss": 2.8868, "step": 17814 }, { "epoch": 1.1059035321869761, "grad_norm": 0.4692057619129683, "learning_rate": 7.958234802615294e-05, "loss": 2.8194, "step": 17815 }, { "epoch": 1.105965609286734, "grad_norm": 0.5178164239434313, "learning_rate": 7.957943634841942e-05, "loss": 2.9039, "step": 17816 }, { "epoch": 1.106027686386492, "grad_norm": 0.44520627812845526, "learning_rate": 7.957652451636281e-05, "loss": 2.8822, "step": 17817 }, { "epoch": 1.1060897634862499, "grad_norm": 0.572611499198371, "learning_rate": 7.957361252999829e-05, "loss": 2.9514, "step": 17818 }, { "epoch": 1.1061518405860078, "grad_norm": 0.5383044455171605, "learning_rate": 7.957070038934105e-05, "loss": 2.9131, "step": 17819 }, { "epoch": 1.1062139176857657, "grad_norm": 0.4988642100737498, "learning_rate": 7.956778809440628e-05, "loss": 2.8843, "step": 17820 }, { "epoch": 1.1062759947855236, "grad_norm": 0.47023075191726404, "learning_rate": 7.95648756452092e-05, "loss": 2.8117, "step": 17821 }, { "epoch": 1.1063380718852815, "grad_norm": 0.47720673275644204, "learning_rate": 7.956196304176498e-05, "loss": 2.8671, "step": 17822 }, { "epoch": 1.1064001489850395, "grad_norm": 0.47676413601030293, "learning_rate": 7.955905028408881e-05, "loss": 2.944, "step": 17823 }, { "epoch": 1.1064622260847974, "grad_norm": 0.49852225859544286, "learning_rate": 7.95561373721959e-05, "loss": 2.8717, "step": 17824 }, { "epoch": 1.1065243031845553, "grad_norm": 0.4731891402403952, "learning_rate": 7.955322430610146e-05, "loss": 2.9234, "step": 17825 }, { "epoch": 1.1065863802843132, "grad_norm": 0.494415583650766, "learning_rate": 7.955031108582066e-05, "loss": 2.9056, "step": 17826 }, { "epoch": 1.106648457384071, "grad_norm": 0.4939695823683175, "learning_rate": 7.954739771136872e-05, "loss": 2.7814, "step": 17827 }, { "epoch": 1.1067105344838288, "grad_norm": 0.5238599638113393, "learning_rate": 7.954448418276084e-05, "loss": 2.8163, "step": 17828 }, { "epoch": 1.1067726115835868, "grad_norm": 0.4884197954079706, "learning_rate": 7.954157050001221e-05, "loss": 2.8035, "step": 17829 }, { "epoch": 1.1068346886833447, "grad_norm": 0.5269226370801059, "learning_rate": 7.953865666313802e-05, "loss": 2.8882, "step": 17830 }, { "epoch": 1.1068967657831026, "grad_norm": 0.49488883988370386, "learning_rate": 7.95357426721535e-05, "loss": 2.9351, "step": 17831 }, { "epoch": 1.1069588428828605, "grad_norm": 0.5057996029686718, "learning_rate": 7.953282852707385e-05, "loss": 2.9191, "step": 17832 }, { "epoch": 1.1070209199826184, "grad_norm": 0.5401638224906552, "learning_rate": 7.952991422791424e-05, "loss": 2.8071, "step": 17833 }, { "epoch": 1.1070829970823763, "grad_norm": 0.5002153052252335, "learning_rate": 7.952699977468993e-05, "loss": 2.8595, "step": 17834 }, { "epoch": 1.1071450741821343, "grad_norm": 0.536397871912647, "learning_rate": 7.952408516741607e-05, "loss": 2.9685, "step": 17835 }, { "epoch": 1.1072071512818922, "grad_norm": 0.5253608156277562, "learning_rate": 7.95211704061079e-05, "loss": 2.8741, "step": 17836 }, { "epoch": 1.10726922838165, "grad_norm": 0.47745865845091445, "learning_rate": 7.951825549078061e-05, "loss": 2.9194, "step": 17837 }, { "epoch": 1.107331305481408, "grad_norm": 0.5138364462971106, "learning_rate": 7.951534042144943e-05, "loss": 2.8849, "step": 17838 }, { "epoch": 1.1073933825811657, "grad_norm": 0.5308704422785862, "learning_rate": 7.951242519812955e-05, "loss": 2.8432, "step": 17839 }, { "epoch": 1.1074554596809236, "grad_norm": 0.5031022451716509, "learning_rate": 7.950950982083618e-05, "loss": 2.8273, "step": 17840 }, { "epoch": 1.1075175367806815, "grad_norm": 0.5117176725652869, "learning_rate": 7.950659428958454e-05, "loss": 2.8268, "step": 17841 }, { "epoch": 1.1075796138804395, "grad_norm": 0.5520282453967553, "learning_rate": 7.95036786043898e-05, "loss": 2.9382, "step": 17842 }, { "epoch": 1.1076416909801974, "grad_norm": 0.48270886491008935, "learning_rate": 7.950076276526724e-05, "loss": 2.8861, "step": 17843 }, { "epoch": 1.1077037680799553, "grad_norm": 0.5128466573494145, "learning_rate": 7.949784677223203e-05, "loss": 2.9187, "step": 17844 }, { "epoch": 1.1077658451797132, "grad_norm": 0.5066996190791094, "learning_rate": 7.94949306252994e-05, "loss": 2.7695, "step": 17845 }, { "epoch": 1.1078279222794711, "grad_norm": 0.48999955530049133, "learning_rate": 7.949201432448454e-05, "loss": 2.8316, "step": 17846 }, { "epoch": 1.107889999379229, "grad_norm": 0.5744680784601598, "learning_rate": 7.94890978698027e-05, "loss": 2.8943, "step": 17847 }, { "epoch": 1.107952076478987, "grad_norm": 0.5556154146000565, "learning_rate": 7.948618126126907e-05, "loss": 2.9288, "step": 17848 }, { "epoch": 1.1080141535787449, "grad_norm": 0.5480530176341879, "learning_rate": 7.948326449889888e-05, "loss": 2.9196, "step": 17849 }, { "epoch": 1.1080762306785026, "grad_norm": 0.5720577187810856, "learning_rate": 7.948034758270733e-05, "loss": 2.8776, "step": 17850 }, { "epoch": 1.1081383077782605, "grad_norm": 0.5715227741266025, "learning_rate": 7.947743051270967e-05, "loss": 2.9159, "step": 17851 }, { "epoch": 1.1082003848780184, "grad_norm": 0.6064331004315282, "learning_rate": 7.947451328892108e-05, "loss": 2.8271, "step": 17852 }, { "epoch": 1.1082624619777763, "grad_norm": 0.5676176599541681, "learning_rate": 7.94715959113568e-05, "loss": 2.9193, "step": 17853 }, { "epoch": 1.1083245390775343, "grad_norm": 0.6102749707802715, "learning_rate": 7.946867838003207e-05, "loss": 2.9837, "step": 17854 }, { "epoch": 1.1083866161772922, "grad_norm": 0.5802958686676293, "learning_rate": 7.946576069496207e-05, "loss": 2.8515, "step": 17855 }, { "epoch": 1.10844869327705, "grad_norm": 0.5651111133576231, "learning_rate": 7.946284285616204e-05, "loss": 2.9472, "step": 17856 }, { "epoch": 1.108510770376808, "grad_norm": 0.6034790324256196, "learning_rate": 7.945992486364723e-05, "loss": 2.9192, "step": 17857 }, { "epoch": 1.108572847476566, "grad_norm": 0.5578559057193392, "learning_rate": 7.945700671743283e-05, "loss": 2.9049, "step": 17858 }, { "epoch": 1.1086349245763238, "grad_norm": 0.520006177847184, "learning_rate": 7.945408841753407e-05, "loss": 2.9371, "step": 17859 }, { "epoch": 1.1086970016760818, "grad_norm": 0.5587589846492441, "learning_rate": 7.945116996396618e-05, "loss": 2.7977, "step": 17860 }, { "epoch": 1.1087590787758397, "grad_norm": 0.5386529002122271, "learning_rate": 7.944825135674438e-05, "loss": 2.9525, "step": 17861 }, { "epoch": 1.1088211558755976, "grad_norm": 0.6571155698105765, "learning_rate": 7.944533259588391e-05, "loss": 2.8814, "step": 17862 }, { "epoch": 1.1088832329753553, "grad_norm": 0.5200337152830109, "learning_rate": 7.94424136814e-05, "loss": 2.8822, "step": 17863 }, { "epoch": 1.1089453100751132, "grad_norm": 0.5903074702729653, "learning_rate": 7.943949461330787e-05, "loss": 2.8424, "step": 17864 }, { "epoch": 1.1090073871748711, "grad_norm": 0.5731734018468995, "learning_rate": 7.943657539162274e-05, "loss": 2.7996, "step": 17865 }, { "epoch": 1.109069464274629, "grad_norm": 0.5570075448704876, "learning_rate": 7.943365601635986e-05, "loss": 2.9608, "step": 17866 }, { "epoch": 1.109131541374387, "grad_norm": 0.5644425123860407, "learning_rate": 7.943073648753444e-05, "loss": 2.9021, "step": 17867 }, { "epoch": 1.1091936184741449, "grad_norm": 0.5297207138315876, "learning_rate": 7.942781680516174e-05, "loss": 2.8724, "step": 17868 }, { "epoch": 1.1092556955739028, "grad_norm": 0.5298799913319319, "learning_rate": 7.942489696925694e-05, "loss": 2.9141, "step": 17869 }, { "epoch": 1.1093177726736607, "grad_norm": 0.5443133165732477, "learning_rate": 7.942197697983535e-05, "loss": 2.8414, "step": 17870 }, { "epoch": 1.1093798497734186, "grad_norm": 0.5114895838910272, "learning_rate": 7.941905683691215e-05, "loss": 2.911, "step": 17871 }, { "epoch": 1.1094419268731766, "grad_norm": 0.5125964841504032, "learning_rate": 7.941613654050258e-05, "loss": 2.8962, "step": 17872 }, { "epoch": 1.1095040039729345, "grad_norm": 0.5315129246544312, "learning_rate": 7.941321609062188e-05, "loss": 2.9295, "step": 17873 }, { "epoch": 1.1095660810726922, "grad_norm": 0.6075057434862317, "learning_rate": 7.941029548728532e-05, "loss": 2.9188, "step": 17874 }, { "epoch": 1.10962815817245, "grad_norm": 0.49422535527965383, "learning_rate": 7.94073747305081e-05, "loss": 2.8389, "step": 17875 }, { "epoch": 1.109690235272208, "grad_norm": 0.537155565529825, "learning_rate": 7.940445382030546e-05, "loss": 2.8728, "step": 17876 }, { "epoch": 1.109752312371966, "grad_norm": 0.5842369324475977, "learning_rate": 7.940153275669264e-05, "loss": 2.8367, "step": 17877 }, { "epoch": 1.1098143894717238, "grad_norm": 0.47823700870650304, "learning_rate": 7.939861153968489e-05, "loss": 2.9049, "step": 17878 }, { "epoch": 1.1098764665714818, "grad_norm": 0.5896154555566616, "learning_rate": 7.939569016929744e-05, "loss": 2.9525, "step": 17879 }, { "epoch": 1.1099385436712397, "grad_norm": 0.5698349403346786, "learning_rate": 7.939276864554554e-05, "loss": 2.8729, "step": 17880 }, { "epoch": 1.1100006207709976, "grad_norm": 0.5565674850866043, "learning_rate": 7.938984696844444e-05, "loss": 2.8643, "step": 17881 }, { "epoch": 1.1100626978707555, "grad_norm": 0.5248894552378426, "learning_rate": 7.938692513800939e-05, "loss": 2.8464, "step": 17882 }, { "epoch": 1.1101247749705134, "grad_norm": 0.5693337389738319, "learning_rate": 7.938400315425558e-05, "loss": 2.8019, "step": 17883 }, { "epoch": 1.1101868520702713, "grad_norm": 0.5464563402097622, "learning_rate": 7.938108101719833e-05, "loss": 2.8911, "step": 17884 }, { "epoch": 1.1102489291700293, "grad_norm": 0.5820423867613089, "learning_rate": 7.937815872685281e-05, "loss": 2.8586, "step": 17885 }, { "epoch": 1.1103110062697872, "grad_norm": 0.5601116976717353, "learning_rate": 7.937523628323433e-05, "loss": 2.8664, "step": 17886 }, { "epoch": 1.1103730833695449, "grad_norm": 0.48466267810572594, "learning_rate": 7.93723136863581e-05, "loss": 2.9269, "step": 17887 }, { "epoch": 1.1104351604693028, "grad_norm": 0.49535853054709966, "learning_rate": 7.936939093623939e-05, "loss": 2.89, "step": 17888 }, { "epoch": 1.1104972375690607, "grad_norm": 0.5436560034350337, "learning_rate": 7.936646803289343e-05, "loss": 2.8872, "step": 17889 }, { "epoch": 1.1105593146688186, "grad_norm": 0.45750225282854406, "learning_rate": 7.936354497633547e-05, "loss": 2.839, "step": 17890 }, { "epoch": 1.1106213917685765, "grad_norm": 0.5666068675474323, "learning_rate": 7.936062176658078e-05, "loss": 2.8507, "step": 17891 }, { "epoch": 1.1106834688683345, "grad_norm": 0.4557652994999751, "learning_rate": 7.93576984036446e-05, "loss": 2.8127, "step": 17892 }, { "epoch": 1.1107455459680924, "grad_norm": 0.5359119731729385, "learning_rate": 7.935477488754217e-05, "loss": 2.8046, "step": 17893 }, { "epoch": 1.1108076230678503, "grad_norm": 0.5275998094334233, "learning_rate": 7.935185121828875e-05, "loss": 2.8342, "step": 17894 }, { "epoch": 1.1108697001676082, "grad_norm": 0.559534494475306, "learning_rate": 7.93489273958996e-05, "loss": 2.8992, "step": 17895 }, { "epoch": 1.1109317772673661, "grad_norm": 0.4965944841862089, "learning_rate": 7.934600342038998e-05, "loss": 2.9374, "step": 17896 }, { "epoch": 1.110993854367124, "grad_norm": 0.5507659909995677, "learning_rate": 7.934307929177513e-05, "loss": 2.8716, "step": 17897 }, { "epoch": 1.1110559314668818, "grad_norm": 0.4993114200360237, "learning_rate": 7.934015501007033e-05, "loss": 2.8571, "step": 17898 }, { "epoch": 1.1111180085666397, "grad_norm": 0.6064863697841301, "learning_rate": 7.933723057529078e-05, "loss": 2.9134, "step": 17899 }, { "epoch": 1.1111800856663976, "grad_norm": 0.5348475045939682, "learning_rate": 7.933430598745182e-05, "loss": 2.9091, "step": 17900 }, { "epoch": 1.1112421627661555, "grad_norm": 0.523209678833282, "learning_rate": 7.933138124656865e-05, "loss": 2.8761, "step": 17901 }, { "epoch": 1.1113042398659134, "grad_norm": 0.5323257493854547, "learning_rate": 7.932845635265653e-05, "loss": 2.9323, "step": 17902 }, { "epoch": 1.1113663169656713, "grad_norm": 0.5107130986932532, "learning_rate": 7.932553130573075e-05, "loss": 2.9299, "step": 17903 }, { "epoch": 1.1114283940654293, "grad_norm": 0.49041580551231745, "learning_rate": 7.932260610580655e-05, "loss": 2.8633, "step": 17904 }, { "epoch": 1.1114904711651872, "grad_norm": 0.5344622769385621, "learning_rate": 7.931968075289918e-05, "loss": 2.937, "step": 17905 }, { "epoch": 1.111552548264945, "grad_norm": 0.4470579673092536, "learning_rate": 7.931675524702393e-05, "loss": 2.7652, "step": 17906 }, { "epoch": 1.111614625364703, "grad_norm": 0.48088699063195306, "learning_rate": 7.931382958819605e-05, "loss": 2.901, "step": 17907 }, { "epoch": 1.111676702464461, "grad_norm": 0.48891487209072426, "learning_rate": 7.93109037764308e-05, "loss": 2.8142, "step": 17908 }, { "epoch": 1.1117387795642188, "grad_norm": 0.4922097670347127, "learning_rate": 7.930797781174346e-05, "loss": 2.9433, "step": 17909 }, { "epoch": 1.1118008566639768, "grad_norm": 0.4733852407991151, "learning_rate": 7.930505169414927e-05, "loss": 2.9244, "step": 17910 }, { "epoch": 1.1118629337637345, "grad_norm": 0.517165541398798, "learning_rate": 7.930212542366351e-05, "loss": 2.9402, "step": 17911 }, { "epoch": 1.1119250108634924, "grad_norm": 0.5147512430951382, "learning_rate": 7.929919900030147e-05, "loss": 2.9902, "step": 17912 }, { "epoch": 1.1119870879632503, "grad_norm": 0.5054150253587926, "learning_rate": 7.929627242407838e-05, "loss": 2.8535, "step": 17913 }, { "epoch": 1.1120491650630082, "grad_norm": 0.4961247772366232, "learning_rate": 7.929334569500954e-05, "loss": 2.8113, "step": 17914 }, { "epoch": 1.1121112421627661, "grad_norm": 0.569660964238997, "learning_rate": 7.929041881311019e-05, "loss": 2.9734, "step": 17915 }, { "epoch": 1.112173319262524, "grad_norm": 0.46347728203525945, "learning_rate": 7.928749177839562e-05, "loss": 2.8089, "step": 17916 }, { "epoch": 1.112235396362282, "grad_norm": 0.48730081927459407, "learning_rate": 7.92845645908811e-05, "loss": 2.966, "step": 17917 }, { "epoch": 1.1122974734620399, "grad_norm": 0.4880903710159739, "learning_rate": 7.928163725058188e-05, "loss": 2.9961, "step": 17918 }, { "epoch": 1.1123595505617978, "grad_norm": 0.4713400334580502, "learning_rate": 7.927870975751327e-05, "loss": 2.7851, "step": 17919 }, { "epoch": 1.1124216276615557, "grad_norm": 0.5233772869492905, "learning_rate": 7.92757821116905e-05, "loss": 2.8643, "step": 17920 }, { "epoch": 1.1124837047613136, "grad_norm": 0.46824243404387983, "learning_rate": 7.92728543131289e-05, "loss": 2.899, "step": 17921 }, { "epoch": 1.1125457818610713, "grad_norm": 0.5952281872174482, "learning_rate": 7.92699263618437e-05, "loss": 2.8953, "step": 17922 }, { "epoch": 1.1126078589608293, "grad_norm": 0.5107030908270339, "learning_rate": 7.926699825785017e-05, "loss": 2.8222, "step": 17923 }, { "epoch": 1.1126699360605872, "grad_norm": 0.51634775427748, "learning_rate": 7.926407000116363e-05, "loss": 2.8477, "step": 17924 }, { "epoch": 1.112732013160345, "grad_norm": 0.5394440282100238, "learning_rate": 7.926114159179932e-05, "loss": 2.9152, "step": 17925 }, { "epoch": 1.112794090260103, "grad_norm": 0.541585161116901, "learning_rate": 7.925821302977254e-05, "loss": 2.9597, "step": 17926 }, { "epoch": 1.112856167359861, "grad_norm": 0.541403951268456, "learning_rate": 7.925528431509855e-05, "loss": 2.9196, "step": 17927 }, { "epoch": 1.1129182444596188, "grad_norm": 0.5920614022746421, "learning_rate": 7.925235544779264e-05, "loss": 2.8522, "step": 17928 }, { "epoch": 1.1129803215593768, "grad_norm": 0.49061185430421894, "learning_rate": 7.924942642787009e-05, "loss": 2.8637, "step": 17929 }, { "epoch": 1.1130423986591347, "grad_norm": 0.521611087563263, "learning_rate": 7.924649725534619e-05, "loss": 2.8396, "step": 17930 }, { "epoch": 1.1131044757588926, "grad_norm": 0.5519118912528787, "learning_rate": 7.92435679302362e-05, "loss": 2.9494, "step": 17931 }, { "epoch": 1.1131665528586505, "grad_norm": 0.5103453339796296, "learning_rate": 7.924063845255542e-05, "loss": 2.8517, "step": 17932 }, { "epoch": 1.1132286299584084, "grad_norm": 0.5243288961973753, "learning_rate": 7.923770882231915e-05, "loss": 2.8018, "step": 17933 }, { "epoch": 1.1132907070581664, "grad_norm": 0.5579766018757545, "learning_rate": 7.923477903954264e-05, "loss": 2.9186, "step": 17934 }, { "epoch": 1.113352784157924, "grad_norm": 0.5325108040353862, "learning_rate": 7.92318491042412e-05, "loss": 2.8166, "step": 17935 }, { "epoch": 1.113414861257682, "grad_norm": 0.5555101124403574, "learning_rate": 7.922891901643009e-05, "loss": 2.8486, "step": 17936 }, { "epoch": 1.1134769383574399, "grad_norm": 0.5291951775066561, "learning_rate": 7.922598877612463e-05, "loss": 2.8829, "step": 17937 }, { "epoch": 1.1135390154571978, "grad_norm": 0.5428426856924903, "learning_rate": 7.922305838334008e-05, "loss": 2.8285, "step": 17938 }, { "epoch": 1.1136010925569557, "grad_norm": 0.5507066283843125, "learning_rate": 7.922012783809176e-05, "loss": 2.8319, "step": 17939 }, { "epoch": 1.1136631696567136, "grad_norm": 0.5395548962742541, "learning_rate": 7.921719714039492e-05, "loss": 2.9112, "step": 17940 }, { "epoch": 1.1137252467564716, "grad_norm": 0.5408773417659076, "learning_rate": 7.921426629026487e-05, "loss": 2.9348, "step": 17941 }, { "epoch": 1.1137873238562295, "grad_norm": 0.5265350357273666, "learning_rate": 7.92113352877169e-05, "loss": 2.8427, "step": 17942 }, { "epoch": 1.1138494009559874, "grad_norm": 0.48158105944417734, "learning_rate": 7.92084041327663e-05, "loss": 2.7957, "step": 17943 }, { "epoch": 1.1139114780557453, "grad_norm": 0.49163946794452235, "learning_rate": 7.920547282542839e-05, "loss": 2.884, "step": 17944 }, { "epoch": 1.1139735551555032, "grad_norm": 0.5211280719943784, "learning_rate": 7.920254136571843e-05, "loss": 2.8526, "step": 17945 }, { "epoch": 1.114035632255261, "grad_norm": 0.5196098612345874, "learning_rate": 7.91996097536517e-05, "loss": 2.8578, "step": 17946 }, { "epoch": 1.1140977093550188, "grad_norm": 0.5688275116722133, "learning_rate": 7.919667798924354e-05, "loss": 2.778, "step": 17947 }, { "epoch": 1.1141597864547768, "grad_norm": 0.5622028784692114, "learning_rate": 7.919374607250922e-05, "loss": 2.937, "step": 17948 }, { "epoch": 1.1142218635545347, "grad_norm": 0.5103307346746674, "learning_rate": 7.919081400346403e-05, "loss": 2.9945, "step": 17949 }, { "epoch": 1.1142839406542926, "grad_norm": 0.5817032472244081, "learning_rate": 7.918788178212328e-05, "loss": 2.9188, "step": 17950 }, { "epoch": 1.1143460177540505, "grad_norm": 0.5385913725098586, "learning_rate": 7.918494940850227e-05, "loss": 2.7828, "step": 17951 }, { "epoch": 1.1144080948538084, "grad_norm": 0.5043069591740234, "learning_rate": 7.918201688261628e-05, "loss": 2.8345, "step": 17952 }, { "epoch": 1.1144701719535663, "grad_norm": 0.5344285960396467, "learning_rate": 7.917908420448065e-05, "loss": 2.9017, "step": 17953 }, { "epoch": 1.1145322490533243, "grad_norm": 0.5143184581193139, "learning_rate": 7.917615137411064e-05, "loss": 2.9305, "step": 17954 }, { "epoch": 1.1145943261530822, "grad_norm": 0.5267843599740786, "learning_rate": 7.917321839152156e-05, "loss": 2.8788, "step": 17955 }, { "epoch": 1.11465640325284, "grad_norm": 0.5482026469476058, "learning_rate": 7.917028525672872e-05, "loss": 2.9165, "step": 17956 }, { "epoch": 1.114718480352598, "grad_norm": 0.5728731283348317, "learning_rate": 7.916735196974742e-05, "loss": 2.8415, "step": 17957 }, { "epoch": 1.114780557452356, "grad_norm": 0.517946076508204, "learning_rate": 7.916441853059297e-05, "loss": 2.841, "step": 17958 }, { "epoch": 1.1148426345521136, "grad_norm": 0.5540442968606706, "learning_rate": 7.916148493928066e-05, "loss": 2.9196, "step": 17959 }, { "epoch": 1.1149047116518715, "grad_norm": 0.5704286012963768, "learning_rate": 7.915855119582582e-05, "loss": 2.8633, "step": 17960 }, { "epoch": 1.1149667887516295, "grad_norm": 0.5193102373676911, "learning_rate": 7.915561730024373e-05, "loss": 2.8755, "step": 17961 }, { "epoch": 1.1150288658513874, "grad_norm": 0.515192197008899, "learning_rate": 7.91526832525497e-05, "loss": 2.8643, "step": 17962 }, { "epoch": 1.1150909429511453, "grad_norm": 0.6313771582464421, "learning_rate": 7.914974905275905e-05, "loss": 2.9113, "step": 17963 }, { "epoch": 1.1151530200509032, "grad_norm": 0.5547033495720561, "learning_rate": 7.914681470088708e-05, "loss": 2.7815, "step": 17964 }, { "epoch": 1.1152150971506611, "grad_norm": 0.5407108040341373, "learning_rate": 7.914388019694911e-05, "loss": 2.9389, "step": 17965 }, { "epoch": 1.115277174250419, "grad_norm": 0.5104070260311714, "learning_rate": 7.914094554096043e-05, "loss": 2.8033, "step": 17966 }, { "epoch": 1.115339251350177, "grad_norm": 0.5261341208865731, "learning_rate": 7.913801073293639e-05, "loss": 2.8275, "step": 17967 }, { "epoch": 1.115401328449935, "grad_norm": 0.48471734039316133, "learning_rate": 7.913507577289225e-05, "loss": 2.9382, "step": 17968 }, { "epoch": 1.1154634055496928, "grad_norm": 0.538239472737641, "learning_rate": 7.913214066084335e-05, "loss": 2.8095, "step": 17969 }, { "epoch": 1.1155254826494505, "grad_norm": 0.5153447458832304, "learning_rate": 7.9129205396805e-05, "loss": 2.9701, "step": 17970 }, { "epoch": 1.1155875597492084, "grad_norm": 0.5449573956469416, "learning_rate": 7.912626998079251e-05, "loss": 2.8917, "step": 17971 }, { "epoch": 1.1156496368489663, "grad_norm": 0.5411403250635755, "learning_rate": 7.91233344128212e-05, "loss": 2.8571, "step": 17972 }, { "epoch": 1.1157117139487243, "grad_norm": 0.4809896080486606, "learning_rate": 7.912039869290641e-05, "loss": 2.9215, "step": 17973 }, { "epoch": 1.1157737910484822, "grad_norm": 0.5275046774128006, "learning_rate": 7.911746282106341e-05, "loss": 2.9643, "step": 17974 }, { "epoch": 1.11583586814824, "grad_norm": 0.491920943968229, "learning_rate": 7.911452679730753e-05, "loss": 2.8783, "step": 17975 }, { "epoch": 1.115897945247998, "grad_norm": 0.5006557575164119, "learning_rate": 7.911159062165409e-05, "loss": 2.8117, "step": 17976 }, { "epoch": 1.115960022347756, "grad_norm": 0.5615823996727122, "learning_rate": 7.910865429411844e-05, "loss": 2.9423, "step": 17977 }, { "epoch": 1.1160220994475138, "grad_norm": 0.46239251485849653, "learning_rate": 7.910571781471585e-05, "loss": 2.8325, "step": 17978 }, { "epoch": 1.1160841765472718, "grad_norm": 0.5535470225408026, "learning_rate": 7.910278118346168e-05, "loss": 2.8744, "step": 17979 }, { "epoch": 1.1161462536470297, "grad_norm": 0.4496335975094358, "learning_rate": 7.909984440037124e-05, "loss": 2.9096, "step": 17980 }, { "epoch": 1.1162083307467876, "grad_norm": 0.5623935492772354, "learning_rate": 7.909690746545985e-05, "loss": 2.8233, "step": 17981 }, { "epoch": 1.1162704078465455, "grad_norm": 0.5423706657744647, "learning_rate": 7.909397037874282e-05, "loss": 2.8562, "step": 17982 }, { "epoch": 1.1163324849463032, "grad_norm": 0.49740553308852165, "learning_rate": 7.909103314023549e-05, "loss": 2.9174, "step": 17983 }, { "epoch": 1.1163945620460611, "grad_norm": 0.5718387925823115, "learning_rate": 7.908809574995316e-05, "loss": 2.884, "step": 17984 }, { "epoch": 1.116456639145819, "grad_norm": 0.5564740918992898, "learning_rate": 7.908515820791119e-05, "loss": 2.9173, "step": 17985 }, { "epoch": 1.116518716245577, "grad_norm": 0.4987104894050861, "learning_rate": 7.908222051412488e-05, "loss": 2.8368, "step": 17986 }, { "epoch": 1.1165807933453349, "grad_norm": 0.6111136111897085, "learning_rate": 7.907928266860958e-05, "loss": 2.9538, "step": 17987 }, { "epoch": 1.1166428704450928, "grad_norm": 0.5167972936650178, "learning_rate": 7.907634467138059e-05, "loss": 2.8269, "step": 17988 }, { "epoch": 1.1167049475448507, "grad_norm": 0.5559366425823357, "learning_rate": 7.907340652245325e-05, "loss": 2.9129, "step": 17989 }, { "epoch": 1.1167670246446086, "grad_norm": 0.5916667077462984, "learning_rate": 7.90704682218429e-05, "loss": 2.8837, "step": 17990 }, { "epoch": 1.1168291017443666, "grad_norm": 0.5645436145094643, "learning_rate": 7.906752976956484e-05, "loss": 2.8369, "step": 17991 }, { "epoch": 1.1168911788441245, "grad_norm": 0.6207930358720387, "learning_rate": 7.906459116563444e-05, "loss": 2.9054, "step": 17992 }, { "epoch": 1.1169532559438824, "grad_norm": 0.5673628522871369, "learning_rate": 7.9061652410067e-05, "loss": 2.8599, "step": 17993 }, { "epoch": 1.11701533304364, "grad_norm": 0.5730749402912896, "learning_rate": 7.905871350287788e-05, "loss": 2.9528, "step": 17994 }, { "epoch": 1.117077410143398, "grad_norm": 0.5731954474296983, "learning_rate": 7.905577444408238e-05, "loss": 2.8789, "step": 17995 }, { "epoch": 1.117139487243156, "grad_norm": 0.5804064300065256, "learning_rate": 7.905283523369586e-05, "loss": 2.9732, "step": 17996 }, { "epoch": 1.1172015643429138, "grad_norm": 0.5901837916129944, "learning_rate": 7.904989587173366e-05, "loss": 2.834, "step": 17997 }, { "epoch": 1.1172636414426718, "grad_norm": 0.4830112794966853, "learning_rate": 7.904695635821109e-05, "loss": 2.864, "step": 17998 }, { "epoch": 1.1173257185424297, "grad_norm": 0.570180302805429, "learning_rate": 7.90440166931435e-05, "loss": 2.9366, "step": 17999 }, { "epoch": 1.1173877956421876, "grad_norm": 0.5486289915425056, "learning_rate": 7.904107687654622e-05, "loss": 2.8957, "step": 18000 }, { "epoch": 1.1174498727419455, "grad_norm": 0.5223606762096341, "learning_rate": 7.903813690843459e-05, "loss": 2.8716, "step": 18001 }, { "epoch": 1.1175119498417034, "grad_norm": 0.5741902558399793, "learning_rate": 7.903519678882396e-05, "loss": 2.9052, "step": 18002 }, { "epoch": 1.1175740269414614, "grad_norm": 0.5583960443360695, "learning_rate": 7.903225651772964e-05, "loss": 2.8309, "step": 18003 }, { "epoch": 1.1176361040412193, "grad_norm": 0.5312958304887979, "learning_rate": 7.9029316095167e-05, "loss": 2.9638, "step": 18004 }, { "epoch": 1.1176981811409772, "grad_norm": 0.5853836493723708, "learning_rate": 7.902637552115138e-05, "loss": 2.8804, "step": 18005 }, { "epoch": 1.117760258240735, "grad_norm": 0.5051136485247147, "learning_rate": 7.902343479569811e-05, "loss": 2.8328, "step": 18006 }, { "epoch": 1.1178223353404928, "grad_norm": 0.5313763748650093, "learning_rate": 7.902049391882252e-05, "loss": 2.8961, "step": 18007 }, { "epoch": 1.1178844124402507, "grad_norm": 0.5221363235681384, "learning_rate": 7.901755289053998e-05, "loss": 2.8638, "step": 18008 }, { "epoch": 1.1179464895400086, "grad_norm": 0.4998068287346288, "learning_rate": 7.901461171086581e-05, "loss": 2.8547, "step": 18009 }, { "epoch": 1.1180085666397666, "grad_norm": 0.4833747164368075, "learning_rate": 7.901167037981539e-05, "loss": 2.8162, "step": 18010 }, { "epoch": 1.1180706437395245, "grad_norm": 0.5005123672272153, "learning_rate": 7.900872889740402e-05, "loss": 2.8273, "step": 18011 }, { "epoch": 1.1181327208392824, "grad_norm": 0.5102199415374187, "learning_rate": 7.900578726364709e-05, "loss": 2.8095, "step": 18012 }, { "epoch": 1.1181947979390403, "grad_norm": 0.5090057731574567, "learning_rate": 7.900284547855991e-05, "loss": 2.9352, "step": 18013 }, { "epoch": 1.1182568750387982, "grad_norm": 0.5236913435599503, "learning_rate": 7.899990354215785e-05, "loss": 2.9468, "step": 18014 }, { "epoch": 1.1183189521385561, "grad_norm": 0.5288574816084152, "learning_rate": 7.899696145445625e-05, "loss": 3.012, "step": 18015 }, { "epoch": 1.118381029238314, "grad_norm": 0.4907383095441465, "learning_rate": 7.899401921547047e-05, "loss": 2.8402, "step": 18016 }, { "epoch": 1.118443106338072, "grad_norm": 0.4998803144498213, "learning_rate": 7.899107682521586e-05, "loss": 2.8526, "step": 18017 }, { "epoch": 1.1185051834378297, "grad_norm": 0.49477121578583366, "learning_rate": 7.898813428370775e-05, "loss": 2.8505, "step": 18018 }, { "epoch": 1.1185672605375876, "grad_norm": 0.46527763761293695, "learning_rate": 7.898519159096151e-05, "loss": 2.9067, "step": 18019 }, { "epoch": 1.1186293376373455, "grad_norm": 0.45947643444684827, "learning_rate": 7.89822487469925e-05, "loss": 2.8694, "step": 18020 }, { "epoch": 1.1186914147371034, "grad_norm": 0.5389180750114962, "learning_rate": 7.897930575181604e-05, "loss": 2.8773, "step": 18021 }, { "epoch": 1.1187534918368613, "grad_norm": 0.4920284226094934, "learning_rate": 7.897636260544752e-05, "loss": 2.8193, "step": 18022 }, { "epoch": 1.1188155689366193, "grad_norm": 0.5500731549595835, "learning_rate": 7.897341930790229e-05, "loss": 2.9307, "step": 18023 }, { "epoch": 1.1188776460363772, "grad_norm": 0.4628796636514912, "learning_rate": 7.897047585919569e-05, "loss": 2.8471, "step": 18024 }, { "epoch": 1.118939723136135, "grad_norm": 0.4597591759448983, "learning_rate": 7.896753225934308e-05, "loss": 2.8399, "step": 18025 }, { "epoch": 1.119001800235893, "grad_norm": 0.4387161517875703, "learning_rate": 7.896458850835981e-05, "loss": 2.8604, "step": 18026 }, { "epoch": 1.119063877335651, "grad_norm": 0.4804651213718256, "learning_rate": 7.896164460626128e-05, "loss": 2.9259, "step": 18027 }, { "epoch": 1.1191259544354089, "grad_norm": 0.5206373958783447, "learning_rate": 7.89587005530628e-05, "loss": 2.9071, "step": 18028 }, { "epoch": 1.1191880315351668, "grad_norm": 0.5051126455029491, "learning_rate": 7.895575634877977e-05, "loss": 2.8132, "step": 18029 }, { "epoch": 1.1192501086349247, "grad_norm": 0.464703049723438, "learning_rate": 7.89528119934275e-05, "loss": 2.937, "step": 18030 }, { "epoch": 1.1193121857346824, "grad_norm": 0.5162274280444256, "learning_rate": 7.89498674870214e-05, "loss": 2.8902, "step": 18031 }, { "epoch": 1.1193742628344403, "grad_norm": 0.49546449657624686, "learning_rate": 7.89469228295768e-05, "loss": 2.854, "step": 18032 }, { "epoch": 1.1194363399341982, "grad_norm": 0.5011084607887856, "learning_rate": 7.894397802110909e-05, "loss": 2.8532, "step": 18033 }, { "epoch": 1.1194984170339561, "grad_norm": 0.48099639266776945, "learning_rate": 7.894103306163363e-05, "loss": 2.8446, "step": 18034 }, { "epoch": 1.119560494133714, "grad_norm": 0.4710385563846276, "learning_rate": 7.893808795116575e-05, "loss": 2.916, "step": 18035 }, { "epoch": 1.119622571233472, "grad_norm": 0.4994083271435755, "learning_rate": 7.893514268972087e-05, "loss": 2.9111, "step": 18036 }, { "epoch": 1.11968464833323, "grad_norm": 0.49545183478298727, "learning_rate": 7.893219727731429e-05, "loss": 2.9449, "step": 18037 }, { "epoch": 1.1197467254329878, "grad_norm": 0.5267307250484373, "learning_rate": 7.892925171396144e-05, "loss": 2.8937, "step": 18038 }, { "epoch": 1.1198088025327457, "grad_norm": 0.4810156771926393, "learning_rate": 7.892630599967765e-05, "loss": 2.9535, "step": 18039 }, { "epoch": 1.1198708796325036, "grad_norm": 0.5549510007569999, "learning_rate": 7.89233601344783e-05, "loss": 2.7812, "step": 18040 }, { "epoch": 1.1199329567322616, "grad_norm": 0.44335193039613774, "learning_rate": 7.892041411837876e-05, "loss": 2.7857, "step": 18041 }, { "epoch": 1.1199950338320193, "grad_norm": 0.4940116774389266, "learning_rate": 7.89174679513944e-05, "loss": 2.8602, "step": 18042 }, { "epoch": 1.1200571109317772, "grad_norm": 0.46361338849927636, "learning_rate": 7.891452163354059e-05, "loss": 2.8665, "step": 18043 }, { "epoch": 1.120119188031535, "grad_norm": 0.5244330148806294, "learning_rate": 7.891157516483269e-05, "loss": 2.9398, "step": 18044 }, { "epoch": 1.120181265131293, "grad_norm": 0.481380605314829, "learning_rate": 7.89086285452861e-05, "loss": 2.8875, "step": 18045 }, { "epoch": 1.120243342231051, "grad_norm": 0.5504939635428301, "learning_rate": 7.890568177491618e-05, "loss": 2.9227, "step": 18046 }, { "epoch": 1.1203054193308088, "grad_norm": 0.49738318408906346, "learning_rate": 7.890273485373828e-05, "loss": 2.9148, "step": 18047 }, { "epoch": 1.1203674964305668, "grad_norm": 0.5795856016111472, "learning_rate": 7.88997877817678e-05, "loss": 2.889, "step": 18048 }, { "epoch": 1.1204295735303247, "grad_norm": 0.5425562217691039, "learning_rate": 7.889684055902013e-05, "loss": 2.8313, "step": 18049 }, { "epoch": 1.1204916506300826, "grad_norm": 0.4758196184193818, "learning_rate": 7.889389318551061e-05, "loss": 2.8648, "step": 18050 }, { "epoch": 1.1205537277298405, "grad_norm": 0.5503778536027535, "learning_rate": 7.889094566125465e-05, "loss": 2.9198, "step": 18051 }, { "epoch": 1.1206158048295984, "grad_norm": 0.492343045436654, "learning_rate": 7.88879979862676e-05, "loss": 2.9183, "step": 18052 }, { "epoch": 1.1206778819293564, "grad_norm": 0.5117282138560845, "learning_rate": 7.888505016056487e-05, "loss": 2.9134, "step": 18053 }, { "epoch": 1.1207399590291143, "grad_norm": 0.45837694921461386, "learning_rate": 7.888210218416182e-05, "loss": 2.8985, "step": 18054 }, { "epoch": 1.120802036128872, "grad_norm": 0.48733853698466717, "learning_rate": 7.88791540570738e-05, "loss": 2.8663, "step": 18055 }, { "epoch": 1.1208641132286299, "grad_norm": 0.4984727273050204, "learning_rate": 7.887620577931624e-05, "loss": 2.8212, "step": 18056 }, { "epoch": 1.1209261903283878, "grad_norm": 0.4762363099210621, "learning_rate": 7.88732573509045e-05, "loss": 2.9039, "step": 18057 }, { "epoch": 1.1209882674281457, "grad_norm": 0.529279814160146, "learning_rate": 7.887030877185397e-05, "loss": 2.8563, "step": 18058 }, { "epoch": 1.1210503445279036, "grad_norm": 0.49304437979119836, "learning_rate": 7.886736004218005e-05, "loss": 2.8822, "step": 18059 }, { "epoch": 1.1211124216276616, "grad_norm": 0.5230724713057308, "learning_rate": 7.886441116189809e-05, "loss": 2.8595, "step": 18060 }, { "epoch": 1.1211744987274195, "grad_norm": 0.5276929758635618, "learning_rate": 7.886146213102349e-05, "loss": 2.9785, "step": 18061 }, { "epoch": 1.1212365758271774, "grad_norm": 0.5524101789312863, "learning_rate": 7.885851294957162e-05, "loss": 2.8101, "step": 18062 }, { "epoch": 1.1212986529269353, "grad_norm": 0.5267387593196561, "learning_rate": 7.88555636175579e-05, "loss": 2.9078, "step": 18063 }, { "epoch": 1.1213607300266932, "grad_norm": 0.559454861060119, "learning_rate": 7.885261413499772e-05, "loss": 2.9439, "step": 18064 }, { "epoch": 1.1214228071264511, "grad_norm": 0.5259538772399478, "learning_rate": 7.884966450190641e-05, "loss": 2.8518, "step": 18065 }, { "epoch": 1.1214848842262088, "grad_norm": 0.5725656185275505, "learning_rate": 7.88467147182994e-05, "loss": 2.9166, "step": 18066 }, { "epoch": 1.1215469613259668, "grad_norm": 0.5381333733504214, "learning_rate": 7.884376478419212e-05, "loss": 2.8971, "step": 18067 }, { "epoch": 1.1216090384257247, "grad_norm": 0.5852471671885258, "learning_rate": 7.884081469959988e-05, "loss": 2.8805, "step": 18068 }, { "epoch": 1.1216711155254826, "grad_norm": 0.5089102397571394, "learning_rate": 7.88378644645381e-05, "loss": 2.945, "step": 18069 }, { "epoch": 1.1217331926252405, "grad_norm": 0.5243458623599727, "learning_rate": 7.883491407902219e-05, "loss": 2.9502, "step": 18070 }, { "epoch": 1.1217952697249984, "grad_norm": 0.643073195622201, "learning_rate": 7.883196354306755e-05, "loss": 2.914, "step": 18071 }, { "epoch": 1.1218573468247564, "grad_norm": 0.5187234837292894, "learning_rate": 7.882901285668955e-05, "loss": 2.8403, "step": 18072 }, { "epoch": 1.1219194239245143, "grad_norm": 0.6410824607768142, "learning_rate": 7.882606201990357e-05, "loss": 2.8856, "step": 18073 }, { "epoch": 1.1219815010242722, "grad_norm": 0.510074624137065, "learning_rate": 7.882311103272503e-05, "loss": 2.822, "step": 18074 }, { "epoch": 1.12204357812403, "grad_norm": 0.6072675012019433, "learning_rate": 7.882015989516935e-05, "loss": 2.8849, "step": 18075 }, { "epoch": 1.122105655223788, "grad_norm": 0.5022218927215425, "learning_rate": 7.881720860725188e-05, "loss": 2.8657, "step": 18076 }, { "epoch": 1.122167732323546, "grad_norm": 0.5199199455792906, "learning_rate": 7.881425716898804e-05, "loss": 2.8594, "step": 18077 }, { "epoch": 1.1222298094233039, "grad_norm": 0.5277209875299164, "learning_rate": 7.881130558039325e-05, "loss": 2.8785, "step": 18078 }, { "epoch": 1.1222918865230616, "grad_norm": 0.48766463446383823, "learning_rate": 7.880835384148286e-05, "loss": 2.83, "step": 18079 }, { "epoch": 1.1223539636228195, "grad_norm": 0.5196915859536616, "learning_rate": 7.880540195227229e-05, "loss": 2.8468, "step": 18080 }, { "epoch": 1.1224160407225774, "grad_norm": 0.48461568144133027, "learning_rate": 7.880244991277694e-05, "loss": 2.8925, "step": 18081 }, { "epoch": 1.1224781178223353, "grad_norm": 0.5942366011508834, "learning_rate": 7.879949772301223e-05, "loss": 2.908, "step": 18082 }, { "epoch": 1.1225401949220932, "grad_norm": 0.5378590848120897, "learning_rate": 7.879654538299356e-05, "loss": 2.8618, "step": 18083 }, { "epoch": 1.1226022720218511, "grad_norm": 0.5176070870987352, "learning_rate": 7.87935928927363e-05, "loss": 2.9072, "step": 18084 }, { "epoch": 1.122664349121609, "grad_norm": 0.5335131851075701, "learning_rate": 7.879064025225589e-05, "loss": 2.8949, "step": 18085 }, { "epoch": 1.122726426221367, "grad_norm": 0.5312598451936285, "learning_rate": 7.878768746156771e-05, "loss": 2.9084, "step": 18086 }, { "epoch": 1.122788503321125, "grad_norm": 0.4690648928586957, "learning_rate": 7.87847345206872e-05, "loss": 2.9128, "step": 18087 }, { "epoch": 1.1228505804208828, "grad_norm": 0.5121855026680974, "learning_rate": 7.87817814296297e-05, "loss": 2.9211, "step": 18088 }, { "epoch": 1.1229126575206407, "grad_norm": 0.5438105417945643, "learning_rate": 7.87788281884107e-05, "loss": 2.8408, "step": 18089 }, { "epoch": 1.1229747346203984, "grad_norm": 0.4969431746394779, "learning_rate": 7.877587479704554e-05, "loss": 2.8943, "step": 18090 }, { "epoch": 1.1230368117201563, "grad_norm": 0.5713558321024136, "learning_rate": 7.877292125554967e-05, "loss": 2.8449, "step": 18091 }, { "epoch": 1.1230988888199143, "grad_norm": 0.5164130864001053, "learning_rate": 7.876996756393848e-05, "loss": 2.9453, "step": 18092 }, { "epoch": 1.1231609659196722, "grad_norm": 0.47663828763090127, "learning_rate": 7.876701372222739e-05, "loss": 2.9005, "step": 18093 }, { "epoch": 1.12322304301943, "grad_norm": 0.5644273059048175, "learning_rate": 7.87640597304318e-05, "loss": 2.8941, "step": 18094 }, { "epoch": 1.123285120119188, "grad_norm": 0.48822765819180064, "learning_rate": 7.876110558856713e-05, "loss": 2.9187, "step": 18095 }, { "epoch": 1.123347197218946, "grad_norm": 0.5556091656912044, "learning_rate": 7.87581512966488e-05, "loss": 2.9802, "step": 18096 }, { "epoch": 1.1234092743187039, "grad_norm": 0.5857303761928752, "learning_rate": 7.87551968546922e-05, "loss": 2.9013, "step": 18097 }, { "epoch": 1.1234713514184618, "grad_norm": 0.4824267553167557, "learning_rate": 7.875224226271276e-05, "loss": 2.8027, "step": 18098 }, { "epoch": 1.1235334285182197, "grad_norm": 0.5432273742802262, "learning_rate": 7.87492875207259e-05, "loss": 2.9616, "step": 18099 }, { "epoch": 1.1235955056179776, "grad_norm": 0.5192881711044183, "learning_rate": 7.874633262874702e-05, "loss": 2.8585, "step": 18100 }, { "epoch": 1.1236575827177355, "grad_norm": 0.5193566624127498, "learning_rate": 7.874337758679155e-05, "loss": 2.809, "step": 18101 }, { "epoch": 1.1237196598174934, "grad_norm": 0.5178406823177725, "learning_rate": 7.874042239487489e-05, "loss": 2.8917, "step": 18102 }, { "epoch": 1.1237817369172511, "grad_norm": 0.4769781207658917, "learning_rate": 7.873746705301248e-05, "loss": 2.8442, "step": 18103 }, { "epoch": 1.123843814017009, "grad_norm": 0.4935122780794318, "learning_rate": 7.873451156121972e-05, "loss": 2.7962, "step": 18104 }, { "epoch": 1.123905891116767, "grad_norm": 0.5116705835401653, "learning_rate": 7.873155591951205e-05, "loss": 2.7965, "step": 18105 }, { "epoch": 1.123967968216525, "grad_norm": 0.49016097837775247, "learning_rate": 7.872860012790486e-05, "loss": 2.8519, "step": 18106 }, { "epoch": 1.1240300453162828, "grad_norm": 0.5562356646972971, "learning_rate": 7.872564418641359e-05, "loss": 2.9431, "step": 18107 }, { "epoch": 1.1240921224160407, "grad_norm": 0.4535122564882337, "learning_rate": 7.872268809505367e-05, "loss": 2.8886, "step": 18108 }, { "epoch": 1.1241541995157986, "grad_norm": 0.5133752873180506, "learning_rate": 7.871973185384051e-05, "loss": 2.8645, "step": 18109 }, { "epoch": 1.1242162766155566, "grad_norm": 0.49647187668190423, "learning_rate": 7.871677546278955e-05, "loss": 2.9659, "step": 18110 }, { "epoch": 1.1242783537153145, "grad_norm": 0.5182836654096182, "learning_rate": 7.871381892191619e-05, "loss": 2.8297, "step": 18111 }, { "epoch": 1.1243404308150724, "grad_norm": 0.6400997753525518, "learning_rate": 7.871086223123587e-05, "loss": 3.0599, "step": 18112 }, { "epoch": 1.1244025079148303, "grad_norm": 0.6512216980288639, "learning_rate": 7.870790539076402e-05, "loss": 2.9488, "step": 18113 }, { "epoch": 1.124464585014588, "grad_norm": 0.6172913813989989, "learning_rate": 7.870494840051604e-05, "loss": 2.8593, "step": 18114 }, { "epoch": 1.124526662114346, "grad_norm": 0.6620096002081435, "learning_rate": 7.87019912605074e-05, "loss": 2.9617, "step": 18115 }, { "epoch": 1.1245887392141038, "grad_norm": 0.6701875387880314, "learning_rate": 7.869903397075348e-05, "loss": 2.8953, "step": 18116 }, { "epoch": 1.1246508163138618, "grad_norm": 0.5839053964953147, "learning_rate": 7.869607653126973e-05, "loss": 2.8803, "step": 18117 }, { "epoch": 1.1247128934136197, "grad_norm": 0.6450255608114401, "learning_rate": 7.86931189420716e-05, "loss": 2.8404, "step": 18118 }, { "epoch": 1.1247749705133776, "grad_norm": 0.6690764361323891, "learning_rate": 7.86901612031745e-05, "loss": 2.9043, "step": 18119 }, { "epoch": 1.1248370476131355, "grad_norm": 0.6648750270923107, "learning_rate": 7.868720331459387e-05, "loss": 2.899, "step": 18120 }, { "epoch": 1.1248991247128934, "grad_norm": 0.5931311192814878, "learning_rate": 7.868424527634511e-05, "loss": 2.7583, "step": 18121 }, { "epoch": 1.1249612018126514, "grad_norm": 0.6048965457469524, "learning_rate": 7.86812870884437e-05, "loss": 2.8768, "step": 18122 }, { "epoch": 1.1250232789124093, "grad_norm": 0.5959381390364631, "learning_rate": 7.867832875090503e-05, "loss": 2.8682, "step": 18123 }, { "epoch": 1.1250853560121672, "grad_norm": 0.5906853135411263, "learning_rate": 7.867537026374458e-05, "loss": 2.8609, "step": 18124 }, { "epoch": 1.125147433111925, "grad_norm": 0.6516342187650496, "learning_rate": 7.867241162697776e-05, "loss": 2.9422, "step": 18125 }, { "epoch": 1.125209510211683, "grad_norm": 0.6176469275931211, "learning_rate": 7.866945284062001e-05, "loss": 2.9629, "step": 18126 }, { "epoch": 1.1252715873114407, "grad_norm": 0.6970188548344117, "learning_rate": 7.866649390468677e-05, "loss": 2.9055, "step": 18127 }, { "epoch": 1.1253336644111986, "grad_norm": 0.5135378354479682, "learning_rate": 7.866353481919346e-05, "loss": 2.8813, "step": 18128 }, { "epoch": 1.1253957415109566, "grad_norm": 0.6024121563532792, "learning_rate": 7.866057558415554e-05, "loss": 2.993, "step": 18129 }, { "epoch": 1.1254578186107145, "grad_norm": 0.5899991468649692, "learning_rate": 7.865761619958843e-05, "loss": 2.888, "step": 18130 }, { "epoch": 1.1255198957104724, "grad_norm": 0.5631824697791206, "learning_rate": 7.865465666550757e-05, "loss": 2.8208, "step": 18131 }, { "epoch": 1.1255819728102303, "grad_norm": 0.5362870180473227, "learning_rate": 7.865169698192842e-05, "loss": 2.9226, "step": 18132 }, { "epoch": 1.1256440499099882, "grad_norm": 0.6128051270270446, "learning_rate": 7.864873714886643e-05, "loss": 2.8932, "step": 18133 }, { "epoch": 1.1257061270097461, "grad_norm": 0.5796039068609661, "learning_rate": 7.864577716633701e-05, "loss": 2.9223, "step": 18134 }, { "epoch": 1.125768204109504, "grad_norm": 0.5084503276311093, "learning_rate": 7.864281703435561e-05, "loss": 2.9311, "step": 18135 }, { "epoch": 1.125830281209262, "grad_norm": 0.5971457351700282, "learning_rate": 7.863985675293768e-05, "loss": 2.915, "step": 18136 }, { "epoch": 1.1258923583090197, "grad_norm": 0.5339412680181171, "learning_rate": 7.863689632209868e-05, "loss": 2.8318, "step": 18137 }, { "epoch": 1.1259544354087776, "grad_norm": 0.5867984296026194, "learning_rate": 7.863393574185403e-05, "loss": 3.0111, "step": 18138 }, { "epoch": 1.1260165125085355, "grad_norm": 0.5866335739705679, "learning_rate": 7.863097501221917e-05, "loss": 2.9098, "step": 18139 }, { "epoch": 1.1260785896082934, "grad_norm": 0.5465233898619516, "learning_rate": 7.862801413320958e-05, "loss": 2.9097, "step": 18140 }, { "epoch": 1.1261406667080514, "grad_norm": 0.6427944248311979, "learning_rate": 7.86250531048407e-05, "loss": 2.9238, "step": 18141 }, { "epoch": 1.1262027438078093, "grad_norm": 0.5522675986449656, "learning_rate": 7.862209192712795e-05, "loss": 2.8149, "step": 18142 }, { "epoch": 1.1262648209075672, "grad_norm": 0.5425442470138302, "learning_rate": 7.861913060008682e-05, "loss": 2.8329, "step": 18143 }, { "epoch": 1.126326898007325, "grad_norm": 0.5405927857960917, "learning_rate": 7.861616912373271e-05, "loss": 2.9415, "step": 18144 }, { "epoch": 1.126388975107083, "grad_norm": 0.5452157325470909, "learning_rate": 7.861320749808111e-05, "loss": 2.9289, "step": 18145 }, { "epoch": 1.126451052206841, "grad_norm": 0.5577411680800415, "learning_rate": 7.861024572314746e-05, "loss": 2.8581, "step": 18146 }, { "epoch": 1.1265131293065989, "grad_norm": 0.5235238644535455, "learning_rate": 7.860728379894722e-05, "loss": 2.8862, "step": 18147 }, { "epoch": 1.1265752064063568, "grad_norm": 0.5743792043704299, "learning_rate": 7.860432172549583e-05, "loss": 2.8719, "step": 18148 }, { "epoch": 1.1266372835061147, "grad_norm": 0.5594426886665567, "learning_rate": 7.860135950280876e-05, "loss": 2.8639, "step": 18149 }, { "epoch": 1.1266993606058726, "grad_norm": 0.5337161994612706, "learning_rate": 7.859839713090144e-05, "loss": 2.8642, "step": 18150 }, { "epoch": 1.1267614377056303, "grad_norm": 0.5563381243132425, "learning_rate": 7.859543460978935e-05, "loss": 2.8828, "step": 18151 }, { "epoch": 1.1268235148053882, "grad_norm": 0.5176280735314288, "learning_rate": 7.859247193948793e-05, "loss": 2.9599, "step": 18152 }, { "epoch": 1.1268855919051461, "grad_norm": 0.5115091316754162, "learning_rate": 7.858950912001264e-05, "loss": 2.8491, "step": 18153 }, { "epoch": 1.126947669004904, "grad_norm": 0.5269866582043029, "learning_rate": 7.858654615137895e-05, "loss": 2.8858, "step": 18154 }, { "epoch": 1.127009746104662, "grad_norm": 0.5183553078900796, "learning_rate": 7.858358303360229e-05, "loss": 2.8483, "step": 18155 }, { "epoch": 1.12707182320442, "grad_norm": 0.455655039253612, "learning_rate": 7.858061976669815e-05, "loss": 2.8844, "step": 18156 }, { "epoch": 1.1271339003041778, "grad_norm": 0.49449732876548913, "learning_rate": 7.857765635068198e-05, "loss": 2.7802, "step": 18157 }, { "epoch": 1.1271959774039357, "grad_norm": 0.45665407511773326, "learning_rate": 7.857469278556924e-05, "loss": 2.8867, "step": 18158 }, { "epoch": 1.1272580545036937, "grad_norm": 0.49045940568911667, "learning_rate": 7.85717290713754e-05, "loss": 2.9163, "step": 18159 }, { "epoch": 1.1273201316034516, "grad_norm": 0.5105433469036056, "learning_rate": 7.85687652081159e-05, "loss": 2.8642, "step": 18160 }, { "epoch": 1.1273822087032093, "grad_norm": 0.4716292328633208, "learning_rate": 7.856580119580622e-05, "loss": 2.8375, "step": 18161 }, { "epoch": 1.1274442858029672, "grad_norm": 0.4934534413599504, "learning_rate": 7.856283703446183e-05, "loss": 2.782, "step": 18162 }, { "epoch": 1.127506362902725, "grad_norm": 0.5494075987673291, "learning_rate": 7.855987272409817e-05, "loss": 2.8738, "step": 18163 }, { "epoch": 1.127568440002483, "grad_norm": 0.5516676297376281, "learning_rate": 7.855690826473072e-05, "loss": 2.888, "step": 18164 }, { "epoch": 1.127630517102241, "grad_norm": 0.5436220149986501, "learning_rate": 7.855394365637496e-05, "loss": 2.8737, "step": 18165 }, { "epoch": 1.1276925942019989, "grad_norm": 0.55106598007801, "learning_rate": 7.855097889904635e-05, "loss": 2.99, "step": 18166 }, { "epoch": 1.1277546713017568, "grad_norm": 0.5493955021180075, "learning_rate": 7.854801399276033e-05, "loss": 2.915, "step": 18167 }, { "epoch": 1.1278167484015147, "grad_norm": 0.5168522477129781, "learning_rate": 7.854504893753241e-05, "loss": 2.8696, "step": 18168 }, { "epoch": 1.1278788255012726, "grad_norm": 0.576370592924869, "learning_rate": 7.854208373337804e-05, "loss": 2.873, "step": 18169 }, { "epoch": 1.1279409026010305, "grad_norm": 0.509386586718625, "learning_rate": 7.853911838031267e-05, "loss": 2.9005, "step": 18170 }, { "epoch": 1.1280029797007884, "grad_norm": 0.49933678630000816, "learning_rate": 7.85361528783518e-05, "loss": 2.9039, "step": 18171 }, { "epoch": 1.1280650568005464, "grad_norm": 0.5060907666835968, "learning_rate": 7.853318722751088e-05, "loss": 2.8317, "step": 18172 }, { "epoch": 1.1281271339003043, "grad_norm": 0.491047538139038, "learning_rate": 7.853022142780542e-05, "loss": 2.9117, "step": 18173 }, { "epoch": 1.1281892110000622, "grad_norm": 0.49322698210649046, "learning_rate": 7.852725547925086e-05, "loss": 2.8446, "step": 18174 }, { "epoch": 1.12825128809982, "grad_norm": 0.46009311275260567, "learning_rate": 7.852428938186269e-05, "loss": 2.8189, "step": 18175 }, { "epoch": 1.1283133651995778, "grad_norm": 0.4295874912746472, "learning_rate": 7.852132313565636e-05, "loss": 2.8307, "step": 18176 }, { "epoch": 1.1283754422993357, "grad_norm": 0.5016353803787341, "learning_rate": 7.851835674064738e-05, "loss": 2.8334, "step": 18177 }, { "epoch": 1.1284375193990936, "grad_norm": 0.48157375703758276, "learning_rate": 7.851539019685122e-05, "loss": 2.8225, "step": 18178 }, { "epoch": 1.1284995964988516, "grad_norm": 0.5135066996102996, "learning_rate": 7.851242350428332e-05, "loss": 2.8614, "step": 18179 }, { "epoch": 1.1285616735986095, "grad_norm": 0.44737931851276924, "learning_rate": 7.85094566629592e-05, "loss": 2.8738, "step": 18180 }, { "epoch": 1.1286237506983674, "grad_norm": 0.5152698218333234, "learning_rate": 7.850648967289432e-05, "loss": 2.8356, "step": 18181 }, { "epoch": 1.1286858277981253, "grad_norm": 0.5530710172442462, "learning_rate": 7.850352253410415e-05, "loss": 2.9052, "step": 18182 }, { "epoch": 1.1287479048978832, "grad_norm": 0.5064326392492133, "learning_rate": 7.850055524660421e-05, "loss": 2.8371, "step": 18183 }, { "epoch": 1.1288099819976412, "grad_norm": 0.558820611729106, "learning_rate": 7.849758781040994e-05, "loss": 2.9264, "step": 18184 }, { "epoch": 1.1288720590973989, "grad_norm": 0.5112588869082597, "learning_rate": 7.849462022553685e-05, "loss": 2.878, "step": 18185 }, { "epoch": 1.1289341361971568, "grad_norm": 0.5337178187976127, "learning_rate": 7.849165249200039e-05, "loss": 2.8878, "step": 18186 }, { "epoch": 1.1289962132969147, "grad_norm": 0.5391046535209015, "learning_rate": 7.848868460981606e-05, "loss": 2.881, "step": 18187 }, { "epoch": 1.1290582903966726, "grad_norm": 0.5605067062606279, "learning_rate": 7.848571657899936e-05, "loss": 2.78, "step": 18188 }, { "epoch": 1.1291203674964305, "grad_norm": 0.5054607514540098, "learning_rate": 7.848274839956575e-05, "loss": 2.9079, "step": 18189 }, { "epoch": 1.1291824445961884, "grad_norm": 0.5751640686979439, "learning_rate": 7.847978007153075e-05, "loss": 2.9448, "step": 18190 }, { "epoch": 1.1292445216959464, "grad_norm": 0.5566295957309195, "learning_rate": 7.84768115949098e-05, "loss": 2.886, "step": 18191 }, { "epoch": 1.1293065987957043, "grad_norm": 0.5314629632779806, "learning_rate": 7.847384296971842e-05, "loss": 2.8821, "step": 18192 }, { "epoch": 1.1293686758954622, "grad_norm": 0.5297717395964586, "learning_rate": 7.847087419597208e-05, "loss": 2.846, "step": 18193 }, { "epoch": 1.12943075299522, "grad_norm": 0.523336543125992, "learning_rate": 7.846790527368629e-05, "loss": 2.9054, "step": 18194 }, { "epoch": 1.129492830094978, "grad_norm": 0.5439066299573037, "learning_rate": 7.846493620287652e-05, "loss": 2.8904, "step": 18195 }, { "epoch": 1.129554907194736, "grad_norm": 0.580291400581314, "learning_rate": 7.846196698355828e-05, "loss": 2.9218, "step": 18196 }, { "epoch": 1.1296169842944939, "grad_norm": 0.5511379304486413, "learning_rate": 7.845899761574704e-05, "loss": 2.8112, "step": 18197 }, { "epoch": 1.1296790613942518, "grad_norm": 0.545825851063975, "learning_rate": 7.84560280994583e-05, "loss": 2.8808, "step": 18198 }, { "epoch": 1.1297411384940095, "grad_norm": 0.5781073696438118, "learning_rate": 7.845305843470756e-05, "loss": 2.8919, "step": 18199 }, { "epoch": 1.1298032155937674, "grad_norm": 0.5119177744300395, "learning_rate": 7.845008862151029e-05, "loss": 2.9209, "step": 18200 }, { "epoch": 1.1298652926935253, "grad_norm": 0.5235919439864651, "learning_rate": 7.844711865988201e-05, "loss": 2.8962, "step": 18201 }, { "epoch": 1.1299273697932832, "grad_norm": 0.5226036946283978, "learning_rate": 7.844414854983821e-05, "loss": 2.9129, "step": 18202 }, { "epoch": 1.1299894468930411, "grad_norm": 0.5306523551674429, "learning_rate": 7.844117829139439e-05, "loss": 2.7838, "step": 18203 }, { "epoch": 1.130051523992799, "grad_norm": 0.505259743922604, "learning_rate": 7.843820788456602e-05, "loss": 2.8691, "step": 18204 }, { "epoch": 1.130113601092557, "grad_norm": 0.5507218890295209, "learning_rate": 7.843523732936862e-05, "loss": 2.9595, "step": 18205 }, { "epoch": 1.130175678192315, "grad_norm": 0.5120262503358043, "learning_rate": 7.843226662581768e-05, "loss": 2.8291, "step": 18206 }, { "epoch": 1.1302377552920728, "grad_norm": 0.47541793581798475, "learning_rate": 7.842929577392872e-05, "loss": 2.8473, "step": 18207 }, { "epoch": 1.1302998323918307, "grad_norm": 0.54326618747495, "learning_rate": 7.842632477371722e-05, "loss": 2.927, "step": 18208 }, { "epoch": 1.1303619094915884, "grad_norm": 0.47500505946627, "learning_rate": 7.842335362519868e-05, "loss": 2.7909, "step": 18209 }, { "epoch": 1.1304239865913464, "grad_norm": 0.5281207541571596, "learning_rate": 7.842038232838859e-05, "loss": 2.8664, "step": 18210 }, { "epoch": 1.1304860636911043, "grad_norm": 0.515003031518032, "learning_rate": 7.841741088330248e-05, "loss": 2.8415, "step": 18211 }, { "epoch": 1.1305481407908622, "grad_norm": 0.5207712676896776, "learning_rate": 7.841443928995584e-05, "loss": 2.9392, "step": 18212 }, { "epoch": 1.13061021789062, "grad_norm": 0.5092896910768009, "learning_rate": 7.841146754836417e-05, "loss": 2.8874, "step": 18213 }, { "epoch": 1.130672294990378, "grad_norm": 0.5412343268294646, "learning_rate": 7.840849565854298e-05, "loss": 2.9088, "step": 18214 }, { "epoch": 1.130734372090136, "grad_norm": 0.4659295423865626, "learning_rate": 7.840552362050776e-05, "loss": 2.9092, "step": 18215 }, { "epoch": 1.1307964491898939, "grad_norm": 0.5755999037566247, "learning_rate": 7.840255143427404e-05, "loss": 2.8835, "step": 18216 }, { "epoch": 1.1308585262896518, "grad_norm": 0.49248519855855094, "learning_rate": 7.839957909985731e-05, "loss": 2.8965, "step": 18217 }, { "epoch": 1.1309206033894097, "grad_norm": 0.49144708860964953, "learning_rate": 7.839660661727307e-05, "loss": 2.8665, "step": 18218 }, { "epoch": 1.1309826804891676, "grad_norm": 0.49981930567124533, "learning_rate": 7.839363398653685e-05, "loss": 2.8933, "step": 18219 }, { "epoch": 1.1310447575889255, "grad_norm": 0.45090562635089526, "learning_rate": 7.839066120766416e-05, "loss": 2.8589, "step": 18220 }, { "epoch": 1.1311068346886834, "grad_norm": 0.5232036124972356, "learning_rate": 7.838768828067047e-05, "loss": 2.8065, "step": 18221 }, { "epoch": 1.1311689117884414, "grad_norm": 0.4712466825171146, "learning_rate": 7.838471520557133e-05, "loss": 2.9125, "step": 18222 }, { "epoch": 1.131230988888199, "grad_norm": 0.5384833223723176, "learning_rate": 7.838174198238227e-05, "loss": 2.8624, "step": 18223 }, { "epoch": 1.131293065987957, "grad_norm": 0.4653087021702813, "learning_rate": 7.837876861111874e-05, "loss": 2.8232, "step": 18224 }, { "epoch": 1.131355143087715, "grad_norm": 0.5445796206210812, "learning_rate": 7.837579509179628e-05, "loss": 2.8116, "step": 18225 }, { "epoch": 1.1314172201874728, "grad_norm": 0.48254263289273364, "learning_rate": 7.837282142443043e-05, "loss": 2.9348, "step": 18226 }, { "epoch": 1.1314792972872307, "grad_norm": 0.5077760243153688, "learning_rate": 7.836984760903668e-05, "loss": 2.866, "step": 18227 }, { "epoch": 1.1315413743869887, "grad_norm": 0.5560447582112716, "learning_rate": 7.836687364563053e-05, "loss": 2.8433, "step": 18228 }, { "epoch": 1.1316034514867466, "grad_norm": 0.5031524757322828, "learning_rate": 7.836389953422754e-05, "loss": 2.9134, "step": 18229 }, { "epoch": 1.1316655285865045, "grad_norm": 0.50087869442976, "learning_rate": 7.836092527484318e-05, "loss": 2.8798, "step": 18230 }, { "epoch": 1.1317276056862624, "grad_norm": 0.5127710776886089, "learning_rate": 7.8357950867493e-05, "loss": 2.8906, "step": 18231 }, { "epoch": 1.1317896827860203, "grad_norm": 0.4954501655675735, "learning_rate": 7.835497631219251e-05, "loss": 2.899, "step": 18232 }, { "epoch": 1.131851759885778, "grad_norm": 0.5224748270915851, "learning_rate": 7.83520016089572e-05, "loss": 2.9035, "step": 18233 }, { "epoch": 1.131913836985536, "grad_norm": 0.5400418769528549, "learning_rate": 7.834902675780263e-05, "loss": 2.8842, "step": 18234 }, { "epoch": 1.1319759140852939, "grad_norm": 0.5616971750585995, "learning_rate": 7.834605175874432e-05, "loss": 2.8599, "step": 18235 }, { "epoch": 1.1320379911850518, "grad_norm": 0.5481728276490017, "learning_rate": 7.834307661179776e-05, "loss": 2.9, "step": 18236 }, { "epoch": 1.1321000682848097, "grad_norm": 0.4402024725496628, "learning_rate": 7.834010131697849e-05, "loss": 2.8425, "step": 18237 }, { "epoch": 1.1321621453845676, "grad_norm": 0.5863497745956856, "learning_rate": 7.833712587430204e-05, "loss": 2.9099, "step": 18238 }, { "epoch": 1.1322242224843255, "grad_norm": 0.5360230157985091, "learning_rate": 7.83341502837839e-05, "loss": 2.9169, "step": 18239 }, { "epoch": 1.1322862995840834, "grad_norm": 0.5185745436607861, "learning_rate": 7.833117454543964e-05, "loss": 2.9012, "step": 18240 }, { "epoch": 1.1323483766838414, "grad_norm": 0.5514194497538868, "learning_rate": 7.832819865928477e-05, "loss": 2.8484, "step": 18241 }, { "epoch": 1.1324104537835993, "grad_norm": 0.49138477502909467, "learning_rate": 7.83252226253348e-05, "loss": 2.8117, "step": 18242 }, { "epoch": 1.1324725308833572, "grad_norm": 0.5809776190929798, "learning_rate": 7.832224644360527e-05, "loss": 2.8831, "step": 18243 }, { "epoch": 1.1325346079831151, "grad_norm": 0.5306103164923762, "learning_rate": 7.831927011411171e-05, "loss": 2.7664, "step": 18244 }, { "epoch": 1.132596685082873, "grad_norm": 0.5416281087307664, "learning_rate": 7.831629363686965e-05, "loss": 2.8433, "step": 18245 }, { "epoch": 1.132658762182631, "grad_norm": 0.5451099249022148, "learning_rate": 7.83133170118946e-05, "loss": 2.8201, "step": 18246 }, { "epoch": 1.1327208392823886, "grad_norm": 0.571065536198702, "learning_rate": 7.83103402392021e-05, "loss": 2.8203, "step": 18247 }, { "epoch": 1.1327829163821466, "grad_norm": 0.5358366991009563, "learning_rate": 7.830736331880768e-05, "loss": 2.8819, "step": 18248 }, { "epoch": 1.1328449934819045, "grad_norm": 0.5087267055849, "learning_rate": 7.830438625072689e-05, "loss": 2.9025, "step": 18249 }, { "epoch": 1.1329070705816624, "grad_norm": 0.5616571416481125, "learning_rate": 7.830140903497522e-05, "loss": 2.805, "step": 18250 }, { "epoch": 1.1329691476814203, "grad_norm": 0.5380753864008502, "learning_rate": 7.829843167156826e-05, "loss": 2.8937, "step": 18251 }, { "epoch": 1.1330312247811782, "grad_norm": 0.5880354642666854, "learning_rate": 7.829545416052148e-05, "loss": 2.8728, "step": 18252 }, { "epoch": 1.1330933018809362, "grad_norm": 0.4854856493918032, "learning_rate": 7.829247650185047e-05, "loss": 2.8769, "step": 18253 }, { "epoch": 1.133155378980694, "grad_norm": 0.5289772166391884, "learning_rate": 7.828949869557073e-05, "loss": 2.8195, "step": 18254 }, { "epoch": 1.133217456080452, "grad_norm": 0.5218765270187972, "learning_rate": 7.828652074169781e-05, "loss": 2.9107, "step": 18255 }, { "epoch": 1.13327953318021, "grad_norm": 0.5486362705685311, "learning_rate": 7.828354264024725e-05, "loss": 2.8459, "step": 18256 }, { "epoch": 1.1333416102799676, "grad_norm": 0.5317497147100676, "learning_rate": 7.828056439123458e-05, "loss": 2.9013, "step": 18257 }, { "epoch": 1.1334036873797255, "grad_norm": 0.4909960085061748, "learning_rate": 7.827758599467533e-05, "loss": 2.8179, "step": 18258 }, { "epoch": 1.1334657644794834, "grad_norm": 0.5371912113614056, "learning_rate": 7.827460745058508e-05, "loss": 2.8789, "step": 18259 }, { "epoch": 1.1335278415792414, "grad_norm": 0.5296825678772985, "learning_rate": 7.827162875897931e-05, "loss": 2.8346, "step": 18260 }, { "epoch": 1.1335899186789993, "grad_norm": 0.5330750308036035, "learning_rate": 7.826864991987359e-05, "loss": 2.977, "step": 18261 }, { "epoch": 1.1336519957787572, "grad_norm": 0.5644717097935259, "learning_rate": 7.826567093328347e-05, "loss": 3.0295, "step": 18262 }, { "epoch": 1.1337140728785151, "grad_norm": 0.5587075924408162, "learning_rate": 7.826269179922448e-05, "loss": 2.8264, "step": 18263 }, { "epoch": 1.133776149978273, "grad_norm": 0.5231296431085861, "learning_rate": 7.825971251771217e-05, "loss": 2.8109, "step": 18264 }, { "epoch": 1.133838227078031, "grad_norm": 0.6593744061001819, "learning_rate": 7.825673308876206e-05, "loss": 2.8639, "step": 18265 }, { "epoch": 1.1339003041777889, "grad_norm": 0.5112047307367924, "learning_rate": 7.825375351238974e-05, "loss": 2.8126, "step": 18266 }, { "epoch": 1.1339623812775468, "grad_norm": 0.6394624629005934, "learning_rate": 7.825077378861073e-05, "loss": 2.8694, "step": 18267 }, { "epoch": 1.1340244583773047, "grad_norm": 0.5570284646755096, "learning_rate": 7.824779391744058e-05, "loss": 2.8578, "step": 18268 }, { "epoch": 1.1340865354770626, "grad_norm": 0.5910116343038884, "learning_rate": 7.824481389889479e-05, "loss": 2.8913, "step": 18269 }, { "epoch": 1.1341486125768203, "grad_norm": 0.5556656715088194, "learning_rate": 7.824183373298898e-05, "loss": 2.9361, "step": 18270 }, { "epoch": 1.1342106896765782, "grad_norm": 0.6580456508537681, "learning_rate": 7.823885341973866e-05, "loss": 2.919, "step": 18271 }, { "epoch": 1.1342727667763361, "grad_norm": 0.4954108847491117, "learning_rate": 7.82358729591594e-05, "loss": 2.7753, "step": 18272 }, { "epoch": 1.134334843876094, "grad_norm": 0.5886142877206281, "learning_rate": 7.823289235126672e-05, "loss": 2.8387, "step": 18273 }, { "epoch": 1.134396920975852, "grad_norm": 0.4970133660752719, "learning_rate": 7.822991159607619e-05, "loss": 2.8015, "step": 18274 }, { "epoch": 1.13445899807561, "grad_norm": 0.5900322821239227, "learning_rate": 7.822693069360337e-05, "loss": 2.9069, "step": 18275 }, { "epoch": 1.1345210751753678, "grad_norm": 0.6064041295711143, "learning_rate": 7.822394964386378e-05, "loss": 2.8036, "step": 18276 }, { "epoch": 1.1345831522751257, "grad_norm": 0.5376618474321412, "learning_rate": 7.822096844687301e-05, "loss": 2.8308, "step": 18277 }, { "epoch": 1.1346452293748837, "grad_norm": 0.5801321961347617, "learning_rate": 7.821798710264658e-05, "loss": 2.8406, "step": 18278 }, { "epoch": 1.1347073064746416, "grad_norm": 0.5121473017541697, "learning_rate": 7.821500561120007e-05, "loss": 2.9372, "step": 18279 }, { "epoch": 1.1347693835743995, "grad_norm": 0.561370961592866, "learning_rate": 7.821202397254903e-05, "loss": 2.9263, "step": 18280 }, { "epoch": 1.1348314606741572, "grad_norm": 0.5007854788805485, "learning_rate": 7.820904218670901e-05, "loss": 2.7932, "step": 18281 }, { "epoch": 1.134893537773915, "grad_norm": 0.5364649287685639, "learning_rate": 7.820606025369558e-05, "loss": 2.8384, "step": 18282 }, { "epoch": 1.134955614873673, "grad_norm": 0.5315715153471801, "learning_rate": 7.820307817352428e-05, "loss": 2.8419, "step": 18283 }, { "epoch": 1.135017691973431, "grad_norm": 0.5354654852681613, "learning_rate": 7.820009594621066e-05, "loss": 2.7637, "step": 18284 }, { "epoch": 1.1350797690731889, "grad_norm": 0.5205881081890295, "learning_rate": 7.819711357177031e-05, "loss": 2.7954, "step": 18285 }, { "epoch": 1.1351418461729468, "grad_norm": 0.5347343786792662, "learning_rate": 7.819413105021875e-05, "loss": 2.8103, "step": 18286 }, { "epoch": 1.1352039232727047, "grad_norm": 0.5173570556680804, "learning_rate": 7.81911483815716e-05, "loss": 2.9155, "step": 18287 }, { "epoch": 1.1352660003724626, "grad_norm": 0.46048172545750804, "learning_rate": 7.818816556584436e-05, "loss": 2.8316, "step": 18288 }, { "epoch": 1.1353280774722205, "grad_norm": 0.5178685075471234, "learning_rate": 7.818518260305262e-05, "loss": 2.9321, "step": 18289 }, { "epoch": 1.1353901545719784, "grad_norm": 0.5030501434366916, "learning_rate": 7.818219949321195e-05, "loss": 2.8337, "step": 18290 }, { "epoch": 1.1354522316717364, "grad_norm": 0.5755100134153233, "learning_rate": 7.817921623633789e-05, "loss": 2.7683, "step": 18291 }, { "epoch": 1.1355143087714943, "grad_norm": 0.5180269418654537, "learning_rate": 7.817623283244604e-05, "loss": 2.8947, "step": 18292 }, { "epoch": 1.1355763858712522, "grad_norm": 0.6183462251191562, "learning_rate": 7.817324928155191e-05, "loss": 2.9098, "step": 18293 }, { "epoch": 1.13563846297101, "grad_norm": 0.6587577995620321, "learning_rate": 7.817026558367114e-05, "loss": 2.9596, "step": 18294 }, { "epoch": 1.1357005400707678, "grad_norm": 0.5713383579993923, "learning_rate": 7.816728173881922e-05, "loss": 2.944, "step": 18295 }, { "epoch": 1.1357626171705257, "grad_norm": 0.5704847624346109, "learning_rate": 7.816429774701177e-05, "loss": 2.924, "step": 18296 }, { "epoch": 1.1358246942702837, "grad_norm": 0.5794742150729159, "learning_rate": 7.816131360826434e-05, "loss": 2.8292, "step": 18297 }, { "epoch": 1.1358867713700416, "grad_norm": 0.6294459757707215, "learning_rate": 7.81583293225925e-05, "loss": 2.885, "step": 18298 }, { "epoch": 1.1359488484697995, "grad_norm": 0.618098130408735, "learning_rate": 7.815534489001181e-05, "loss": 2.8462, "step": 18299 }, { "epoch": 1.1360109255695574, "grad_norm": 0.5716217637244697, "learning_rate": 7.815236031053788e-05, "loss": 2.937, "step": 18300 }, { "epoch": 1.1360730026693153, "grad_norm": 0.6416693293115767, "learning_rate": 7.814937558418623e-05, "loss": 2.771, "step": 18301 }, { "epoch": 1.1361350797690732, "grad_norm": 0.5502970846812578, "learning_rate": 7.814639071097245e-05, "loss": 2.897, "step": 18302 }, { "epoch": 1.1361971568688312, "grad_norm": 0.6094120210014691, "learning_rate": 7.814340569091213e-05, "loss": 2.845, "step": 18303 }, { "epoch": 1.136259233968589, "grad_norm": 0.5589659724481894, "learning_rate": 7.814042052402082e-05, "loss": 2.7997, "step": 18304 }, { "epoch": 1.1363213110683468, "grad_norm": 0.5791942400051833, "learning_rate": 7.81374352103141e-05, "loss": 2.8384, "step": 18305 }, { "epoch": 1.1363833881681047, "grad_norm": 0.5326122041699255, "learning_rate": 7.813444974980757e-05, "loss": 2.9066, "step": 18306 }, { "epoch": 1.1364454652678626, "grad_norm": 0.540611556883799, "learning_rate": 7.813146414251678e-05, "loss": 2.908, "step": 18307 }, { "epoch": 1.1365075423676205, "grad_norm": 0.5644226593982831, "learning_rate": 7.812847838845729e-05, "loss": 2.8736, "step": 18308 }, { "epoch": 1.1365696194673784, "grad_norm": 0.5580691027870699, "learning_rate": 7.812549248764472e-05, "loss": 2.8817, "step": 18309 }, { "epoch": 1.1366316965671364, "grad_norm": 0.5289689346649394, "learning_rate": 7.812250644009462e-05, "loss": 2.9462, "step": 18310 }, { "epoch": 1.1366937736668943, "grad_norm": 0.5741110979752075, "learning_rate": 7.811952024582259e-05, "loss": 2.8557, "step": 18311 }, { "epoch": 1.1367558507666522, "grad_norm": 0.47860333942765876, "learning_rate": 7.811653390484419e-05, "loss": 2.8861, "step": 18312 }, { "epoch": 1.1368179278664101, "grad_norm": 0.576121027367689, "learning_rate": 7.8113547417175e-05, "loss": 2.9472, "step": 18313 }, { "epoch": 1.136880004966168, "grad_norm": 0.5216068024101793, "learning_rate": 7.811056078283062e-05, "loss": 2.8968, "step": 18314 }, { "epoch": 1.136942082065926, "grad_norm": 0.5487780565114941, "learning_rate": 7.810757400182661e-05, "loss": 2.9241, "step": 18315 }, { "epoch": 1.1370041591656839, "grad_norm": 0.515219153005128, "learning_rate": 7.810458707417857e-05, "loss": 2.7778, "step": 18316 }, { "epoch": 1.1370662362654418, "grad_norm": 0.5084865791158858, "learning_rate": 7.810159999990207e-05, "loss": 2.9041, "step": 18317 }, { "epoch": 1.1371283133651995, "grad_norm": 0.49887158795238695, "learning_rate": 7.809861277901271e-05, "loss": 2.8906, "step": 18318 }, { "epoch": 1.1371903904649574, "grad_norm": 0.5063499811131988, "learning_rate": 7.809562541152606e-05, "loss": 2.8987, "step": 18319 }, { "epoch": 1.1372524675647153, "grad_norm": 0.45619648724102035, "learning_rate": 7.809263789745771e-05, "loss": 2.8922, "step": 18320 }, { "epoch": 1.1373145446644732, "grad_norm": 0.5986098118605925, "learning_rate": 7.808965023682324e-05, "loss": 2.8971, "step": 18321 }, { "epoch": 1.1373766217642312, "grad_norm": 0.5016059416974953, "learning_rate": 7.808666242963827e-05, "loss": 2.9459, "step": 18322 }, { "epoch": 1.137438698863989, "grad_norm": 0.5487600535468937, "learning_rate": 7.808367447591834e-05, "loss": 2.8783, "step": 18323 }, { "epoch": 1.137500775963747, "grad_norm": 0.4669119401792733, "learning_rate": 7.808068637567908e-05, "loss": 2.8215, "step": 18324 }, { "epoch": 1.137562853063505, "grad_norm": 0.5157996373220044, "learning_rate": 7.807769812893607e-05, "loss": 2.9097, "step": 18325 }, { "epoch": 1.1376249301632628, "grad_norm": 0.5151176557584326, "learning_rate": 7.807470973570487e-05, "loss": 2.91, "step": 18326 }, { "epoch": 1.1376870072630207, "grad_norm": 0.4843105919377372, "learning_rate": 7.80717211960011e-05, "loss": 2.8725, "step": 18327 }, { "epoch": 1.1377490843627787, "grad_norm": 0.5161988215444475, "learning_rate": 7.806873250984035e-05, "loss": 2.861, "step": 18328 }, { "epoch": 1.1378111614625364, "grad_norm": 0.4777841543617974, "learning_rate": 7.80657436772382e-05, "loss": 2.9478, "step": 18329 }, { "epoch": 1.1378732385622943, "grad_norm": 0.5201722972866211, "learning_rate": 7.806275469821026e-05, "loss": 2.848, "step": 18330 }, { "epoch": 1.1379353156620522, "grad_norm": 0.5097768789634931, "learning_rate": 7.805976557277212e-05, "loss": 2.8844, "step": 18331 }, { "epoch": 1.1379973927618101, "grad_norm": 0.568873038973998, "learning_rate": 7.805677630093937e-05, "loss": 2.8979, "step": 18332 }, { "epoch": 1.138059469861568, "grad_norm": 0.5158299125636022, "learning_rate": 7.80537868827276e-05, "loss": 2.8922, "step": 18333 }, { "epoch": 1.138121546961326, "grad_norm": 0.5715921494608206, "learning_rate": 7.805079731815242e-05, "loss": 2.7696, "step": 18334 }, { "epoch": 1.1381836240610839, "grad_norm": 0.5023817317954878, "learning_rate": 7.80478076072294e-05, "loss": 2.8662, "step": 18335 }, { "epoch": 1.1382457011608418, "grad_norm": 0.5769469656644557, "learning_rate": 7.804481774997418e-05, "loss": 2.8691, "step": 18336 }, { "epoch": 1.1383077782605997, "grad_norm": 0.5817080630935331, "learning_rate": 7.804182774640234e-05, "loss": 2.8668, "step": 18337 }, { "epoch": 1.1383698553603576, "grad_norm": 0.5446547260152573, "learning_rate": 7.803883759652946e-05, "loss": 2.9164, "step": 18338 }, { "epoch": 1.1384319324601155, "grad_norm": 0.5404581992318022, "learning_rate": 7.803584730037118e-05, "loss": 2.8399, "step": 18339 }, { "epoch": 1.1384940095598735, "grad_norm": 0.5171665498575826, "learning_rate": 7.803285685794307e-05, "loss": 2.855, "step": 18340 }, { "epoch": 1.1385560866596314, "grad_norm": 0.5273386354551601, "learning_rate": 7.802986626926073e-05, "loss": 2.8687, "step": 18341 }, { "epoch": 1.138618163759389, "grad_norm": 0.5585056048652454, "learning_rate": 7.802687553433979e-05, "loss": 2.902, "step": 18342 }, { "epoch": 1.138680240859147, "grad_norm": 0.49932522242772276, "learning_rate": 7.802388465319583e-05, "loss": 2.8641, "step": 18343 }, { "epoch": 1.138742317958905, "grad_norm": 0.5131156433590559, "learning_rate": 7.802089362584445e-05, "loss": 2.7909, "step": 18344 }, { "epoch": 1.1388043950586628, "grad_norm": 0.5022537577988372, "learning_rate": 7.801790245230128e-05, "loss": 2.9307, "step": 18345 }, { "epoch": 1.1388664721584207, "grad_norm": 0.5820235693508006, "learning_rate": 7.80149111325819e-05, "loss": 2.8106, "step": 18346 }, { "epoch": 1.1389285492581787, "grad_norm": 0.44039639433823496, "learning_rate": 7.801191966670193e-05, "loss": 2.8445, "step": 18347 }, { "epoch": 1.1389906263579366, "grad_norm": 0.5501844811989914, "learning_rate": 7.800892805467699e-05, "loss": 2.82, "step": 18348 }, { "epoch": 1.1390527034576945, "grad_norm": 0.44513493059388415, "learning_rate": 7.800593629652266e-05, "loss": 2.8795, "step": 18349 }, { "epoch": 1.1391147805574524, "grad_norm": 0.5393089062982909, "learning_rate": 7.800294439225456e-05, "loss": 2.8727, "step": 18350 }, { "epoch": 1.1391768576572103, "grad_norm": 0.452005598604292, "learning_rate": 7.79999523418883e-05, "loss": 2.9416, "step": 18351 }, { "epoch": 1.1392389347569682, "grad_norm": 0.5466611444041215, "learning_rate": 7.799696014543949e-05, "loss": 2.9222, "step": 18352 }, { "epoch": 1.139301011856726, "grad_norm": 0.4873310915228521, "learning_rate": 7.799396780292375e-05, "loss": 2.7359, "step": 18353 }, { "epoch": 1.1393630889564839, "grad_norm": 0.5592138143419144, "learning_rate": 7.799097531435668e-05, "loss": 2.8792, "step": 18354 }, { "epoch": 1.1394251660562418, "grad_norm": 0.4560067684639572, "learning_rate": 7.798798267975392e-05, "loss": 2.8323, "step": 18355 }, { "epoch": 1.1394872431559997, "grad_norm": 0.4918525252242596, "learning_rate": 7.798498989913103e-05, "loss": 2.8279, "step": 18356 }, { "epoch": 1.1395493202557576, "grad_norm": 0.5255270571634707, "learning_rate": 7.798199697250366e-05, "loss": 2.9786, "step": 18357 }, { "epoch": 1.1396113973555155, "grad_norm": 0.5117993473324596, "learning_rate": 7.797900389988742e-05, "loss": 2.9924, "step": 18358 }, { "epoch": 1.1396734744552734, "grad_norm": 0.5296685578326694, "learning_rate": 7.797601068129791e-05, "loss": 2.7402, "step": 18359 }, { "epoch": 1.1397355515550314, "grad_norm": 0.4927839268230613, "learning_rate": 7.79730173167508e-05, "loss": 2.8097, "step": 18360 }, { "epoch": 1.1397976286547893, "grad_norm": 0.48042765495899464, "learning_rate": 7.797002380626164e-05, "loss": 2.9211, "step": 18361 }, { "epoch": 1.1398597057545472, "grad_norm": 0.5020000272576071, "learning_rate": 7.796703014984607e-05, "loss": 2.8289, "step": 18362 }, { "epoch": 1.1399217828543051, "grad_norm": 0.45065540626467526, "learning_rate": 7.796403634751975e-05, "loss": 2.8785, "step": 18363 }, { "epoch": 1.139983859954063, "grad_norm": 0.48826542637934595, "learning_rate": 7.796104239929822e-05, "loss": 2.8338, "step": 18364 }, { "epoch": 1.140045937053821, "grad_norm": 0.4408760431070212, "learning_rate": 7.795804830519718e-05, "loss": 2.8593, "step": 18365 }, { "epoch": 1.1401080141535787, "grad_norm": 0.5251604448109519, "learning_rate": 7.795505406523219e-05, "loss": 2.846, "step": 18366 }, { "epoch": 1.1401700912533366, "grad_norm": 0.5104183534347793, "learning_rate": 7.795205967941889e-05, "loss": 2.7158, "step": 18367 }, { "epoch": 1.1402321683530945, "grad_norm": 0.5028647667917252, "learning_rate": 7.794906514777294e-05, "loss": 2.9194, "step": 18368 }, { "epoch": 1.1402942454528524, "grad_norm": 0.509276202631227, "learning_rate": 7.79460704703099e-05, "loss": 2.9145, "step": 18369 }, { "epoch": 1.1403563225526103, "grad_norm": 0.5243749621712963, "learning_rate": 7.794307564704545e-05, "loss": 2.8549, "step": 18370 }, { "epoch": 1.1404183996523682, "grad_norm": 0.4817567170283084, "learning_rate": 7.794008067799518e-05, "loss": 2.8981, "step": 18371 }, { "epoch": 1.1404804767521262, "grad_norm": 0.5349742830637575, "learning_rate": 7.793708556317473e-05, "loss": 2.8681, "step": 18372 }, { "epoch": 1.140542553851884, "grad_norm": 0.5401729810891589, "learning_rate": 7.793409030259972e-05, "loss": 2.8757, "step": 18373 }, { "epoch": 1.140604630951642, "grad_norm": 0.497073055477106, "learning_rate": 7.79310948962858e-05, "loss": 2.8608, "step": 18374 }, { "epoch": 1.1406667080514, "grad_norm": 0.6053007323421508, "learning_rate": 7.792809934424856e-05, "loss": 2.8416, "step": 18375 }, { "epoch": 1.1407287851511578, "grad_norm": 0.5115343882876775, "learning_rate": 7.792510364650364e-05, "loss": 2.8927, "step": 18376 }, { "epoch": 1.1407908622509155, "grad_norm": 0.5769596727086195, "learning_rate": 7.792210780306668e-05, "loss": 2.8257, "step": 18377 }, { "epoch": 1.1408529393506734, "grad_norm": 0.5090047192551502, "learning_rate": 7.79191118139533e-05, "loss": 2.8894, "step": 18378 }, { "epoch": 1.1409150164504314, "grad_norm": 0.5354347339753268, "learning_rate": 7.791611567917914e-05, "loss": 2.8341, "step": 18379 }, { "epoch": 1.1409770935501893, "grad_norm": 0.5805658159180793, "learning_rate": 7.791311939875984e-05, "loss": 2.866, "step": 18380 }, { "epoch": 1.1410391706499472, "grad_norm": 0.5211626410520469, "learning_rate": 7.791012297271101e-05, "loss": 2.9187, "step": 18381 }, { "epoch": 1.1411012477497051, "grad_norm": 0.6166629735303384, "learning_rate": 7.790712640104829e-05, "loss": 2.8801, "step": 18382 }, { "epoch": 1.141163324849463, "grad_norm": 0.623191768814875, "learning_rate": 7.790412968378731e-05, "loss": 2.872, "step": 18383 }, { "epoch": 1.141225401949221, "grad_norm": 0.5509648389086256, "learning_rate": 7.790113282094372e-05, "loss": 2.8555, "step": 18384 }, { "epoch": 1.1412874790489789, "grad_norm": 0.6442849584545732, "learning_rate": 7.789813581253315e-05, "loss": 2.8755, "step": 18385 }, { "epoch": 1.1413495561487368, "grad_norm": 0.4738680383235786, "learning_rate": 7.789513865857124e-05, "loss": 2.8619, "step": 18386 }, { "epoch": 1.1414116332484947, "grad_norm": 0.6508188298653097, "learning_rate": 7.789214135907361e-05, "loss": 2.7916, "step": 18387 }, { "epoch": 1.1414737103482526, "grad_norm": 0.5045533276473989, "learning_rate": 7.788914391405591e-05, "loss": 2.9027, "step": 18388 }, { "epoch": 1.1415357874480105, "grad_norm": 0.5809021565242145, "learning_rate": 7.788614632353377e-05, "loss": 2.9173, "step": 18389 }, { "epoch": 1.1415978645477682, "grad_norm": 0.5200486725714604, "learning_rate": 7.788314858752283e-05, "loss": 2.7777, "step": 18390 }, { "epoch": 1.1416599416475262, "grad_norm": 0.5457363263285698, "learning_rate": 7.788015070603876e-05, "loss": 2.9471, "step": 18391 }, { "epoch": 1.141722018747284, "grad_norm": 0.48146510518768604, "learning_rate": 7.787715267909715e-05, "loss": 2.8194, "step": 18392 }, { "epoch": 1.141784095847042, "grad_norm": 0.5207644575990048, "learning_rate": 7.787415450671368e-05, "loss": 2.8041, "step": 18393 }, { "epoch": 1.1418461729468, "grad_norm": 0.478945994011469, "learning_rate": 7.787115618890397e-05, "loss": 2.9143, "step": 18394 }, { "epoch": 1.1419082500465578, "grad_norm": 0.5493016762347978, "learning_rate": 7.786815772568369e-05, "loss": 2.8429, "step": 18395 }, { "epoch": 1.1419703271463157, "grad_norm": 0.5222184518696017, "learning_rate": 7.786515911706844e-05, "loss": 2.8609, "step": 18396 }, { "epoch": 1.1420324042460737, "grad_norm": 0.5030959363800183, "learning_rate": 7.786216036307392e-05, "loss": 2.9168, "step": 18397 }, { "epoch": 1.1420944813458316, "grad_norm": 0.5951759882723691, "learning_rate": 7.785916146371573e-05, "loss": 2.8398, "step": 18398 }, { "epoch": 1.1421565584455895, "grad_norm": 0.5134176633303309, "learning_rate": 7.785616241900952e-05, "loss": 2.8449, "step": 18399 }, { "epoch": 1.1422186355453474, "grad_norm": 0.5343309150045351, "learning_rate": 7.785316322897096e-05, "loss": 2.8482, "step": 18400 }, { "epoch": 1.1422807126451051, "grad_norm": 0.5110673908467531, "learning_rate": 7.785016389361569e-05, "loss": 2.8905, "step": 18401 }, { "epoch": 1.142342789744863, "grad_norm": 0.48718905558163433, "learning_rate": 7.784716441295935e-05, "loss": 2.8992, "step": 18402 }, { "epoch": 1.142404866844621, "grad_norm": 0.5461984809568717, "learning_rate": 7.784416478701759e-05, "loss": 2.8078, "step": 18403 }, { "epoch": 1.1424669439443789, "grad_norm": 0.5137909725771167, "learning_rate": 7.784116501580608e-05, "loss": 2.8081, "step": 18404 }, { "epoch": 1.1425290210441368, "grad_norm": 0.5175356865524504, "learning_rate": 7.783816509934043e-05, "loss": 2.8571, "step": 18405 }, { "epoch": 1.1425910981438947, "grad_norm": 0.5940216095731256, "learning_rate": 7.783516503763632e-05, "loss": 2.9187, "step": 18406 }, { "epoch": 1.1426531752436526, "grad_norm": 0.5369260440404015, "learning_rate": 7.783216483070941e-05, "loss": 2.8399, "step": 18407 }, { "epoch": 1.1427152523434105, "grad_norm": 0.5498317352742705, "learning_rate": 7.782916447857532e-05, "loss": 2.8704, "step": 18408 }, { "epoch": 1.1427773294431685, "grad_norm": 0.5080660553792655, "learning_rate": 7.782616398124973e-05, "loss": 2.8012, "step": 18409 }, { "epoch": 1.1428394065429264, "grad_norm": 0.515844356232156, "learning_rate": 7.78231633387483e-05, "loss": 2.7355, "step": 18410 }, { "epoch": 1.1429014836426843, "grad_norm": 0.5434636392388721, "learning_rate": 7.782016255108666e-05, "loss": 2.8836, "step": 18411 }, { "epoch": 1.1429635607424422, "grad_norm": 0.5077871756474649, "learning_rate": 7.781716161828049e-05, "loss": 2.8511, "step": 18412 }, { "epoch": 1.1430256378422001, "grad_norm": 0.5764037360790317, "learning_rate": 7.781416054034543e-05, "loss": 2.8549, "step": 18413 }, { "epoch": 1.1430877149419578, "grad_norm": 0.49189743698500876, "learning_rate": 7.781115931729714e-05, "loss": 2.9058, "step": 18414 }, { "epoch": 1.1431497920417157, "grad_norm": 0.5367176638929547, "learning_rate": 7.780815794915127e-05, "loss": 2.8026, "step": 18415 }, { "epoch": 1.1432118691414737, "grad_norm": 0.54146408608837, "learning_rate": 7.78051564359235e-05, "loss": 2.8672, "step": 18416 }, { "epoch": 1.1432739462412316, "grad_norm": 0.5613323914983614, "learning_rate": 7.780215477762947e-05, "loss": 3.0223, "step": 18417 }, { "epoch": 1.1433360233409895, "grad_norm": 0.5418968872965035, "learning_rate": 7.779915297428486e-05, "loss": 2.8541, "step": 18418 }, { "epoch": 1.1433981004407474, "grad_norm": 0.5619896586949822, "learning_rate": 7.779615102590532e-05, "loss": 2.901, "step": 18419 }, { "epoch": 1.1434601775405053, "grad_norm": 0.5407722008142828, "learning_rate": 7.779314893250651e-05, "loss": 2.8788, "step": 18420 }, { "epoch": 1.1435222546402632, "grad_norm": 0.5579277282864195, "learning_rate": 7.779014669410409e-05, "loss": 2.8545, "step": 18421 }, { "epoch": 1.1435843317400212, "grad_norm": 0.5369341477751367, "learning_rate": 7.778714431071375e-05, "loss": 2.8338, "step": 18422 }, { "epoch": 1.143646408839779, "grad_norm": 0.579769193542955, "learning_rate": 7.77841417823511e-05, "loss": 2.9854, "step": 18423 }, { "epoch": 1.143708485939537, "grad_norm": 0.5227695017369114, "learning_rate": 7.778113910903186e-05, "loss": 2.7966, "step": 18424 }, { "epoch": 1.1437705630392947, "grad_norm": 0.6207521566875108, "learning_rate": 7.777813629077166e-05, "loss": 2.8072, "step": 18425 }, { "epoch": 1.1438326401390526, "grad_norm": 0.47799969321404673, "learning_rate": 7.777513332758619e-05, "loss": 2.9485, "step": 18426 }, { "epoch": 1.1438947172388105, "grad_norm": 0.5507413700897755, "learning_rate": 7.777213021949111e-05, "loss": 2.8586, "step": 18427 }, { "epoch": 1.1439567943385685, "grad_norm": 0.4861081250038468, "learning_rate": 7.776912696650208e-05, "loss": 2.8734, "step": 18428 }, { "epoch": 1.1440188714383264, "grad_norm": 0.5308564916536624, "learning_rate": 7.776612356863478e-05, "loss": 2.8522, "step": 18429 }, { "epoch": 1.1440809485380843, "grad_norm": 0.4957528933288201, "learning_rate": 7.776312002590486e-05, "loss": 2.9135, "step": 18430 }, { "epoch": 1.1441430256378422, "grad_norm": 0.5166629689863643, "learning_rate": 7.776011633832801e-05, "loss": 2.8513, "step": 18431 }, { "epoch": 1.1442051027376001, "grad_norm": 0.5502049833564313, "learning_rate": 7.775711250591988e-05, "loss": 2.9005, "step": 18432 }, { "epoch": 1.144267179837358, "grad_norm": 0.5475637186326625, "learning_rate": 7.775410852869618e-05, "loss": 2.9133, "step": 18433 }, { "epoch": 1.144329256937116, "grad_norm": 0.45722093476845144, "learning_rate": 7.775110440667252e-05, "loss": 2.9342, "step": 18434 }, { "epoch": 1.1443913340368739, "grad_norm": 0.6169837663685043, "learning_rate": 7.774810013986465e-05, "loss": 2.8785, "step": 18435 }, { "epoch": 1.1444534111366318, "grad_norm": 0.5784981270976564, "learning_rate": 7.774509572828819e-05, "loss": 2.8248, "step": 18436 }, { "epoch": 1.1445154882363897, "grad_norm": 0.4912870890324548, "learning_rate": 7.774209117195883e-05, "loss": 2.8665, "step": 18437 }, { "epoch": 1.1445775653361474, "grad_norm": 0.5307301895336394, "learning_rate": 7.773908647089225e-05, "loss": 2.778, "step": 18438 }, { "epoch": 1.1446396424359053, "grad_norm": 0.4811539919565415, "learning_rate": 7.773608162510413e-05, "loss": 2.8864, "step": 18439 }, { "epoch": 1.1447017195356632, "grad_norm": 0.4634087154017495, "learning_rate": 7.773307663461013e-05, "loss": 2.7523, "step": 18440 }, { "epoch": 1.1447637966354212, "grad_norm": 0.5019883614725941, "learning_rate": 7.773007149942593e-05, "loss": 2.848, "step": 18441 }, { "epoch": 1.144825873735179, "grad_norm": 0.5025512456621728, "learning_rate": 7.772706621956722e-05, "loss": 2.8794, "step": 18442 }, { "epoch": 1.144887950834937, "grad_norm": 0.5180352837133457, "learning_rate": 7.772406079504968e-05, "loss": 2.843, "step": 18443 }, { "epoch": 1.144950027934695, "grad_norm": 0.5121489310973227, "learning_rate": 7.772105522588898e-05, "loss": 2.9256, "step": 18444 }, { "epoch": 1.1450121050344528, "grad_norm": 0.5261272102775377, "learning_rate": 7.771804951210082e-05, "loss": 2.9283, "step": 18445 }, { "epoch": 1.1450741821342107, "grad_norm": 0.4994734584441068, "learning_rate": 7.771504365370084e-05, "loss": 2.8897, "step": 18446 }, { "epoch": 1.1451362592339687, "grad_norm": 0.5394334760508517, "learning_rate": 7.771203765070476e-05, "loss": 2.9109, "step": 18447 }, { "epoch": 1.1451983363337266, "grad_norm": 0.5116467607252294, "learning_rate": 7.770903150312826e-05, "loss": 2.8449, "step": 18448 }, { "epoch": 1.1452604134334843, "grad_norm": 0.48585387549912246, "learning_rate": 7.7706025210987e-05, "loss": 2.8064, "step": 18449 }, { "epoch": 1.1453224905332422, "grad_norm": 0.5823177170314549, "learning_rate": 7.770301877429669e-05, "loss": 2.8133, "step": 18450 }, { "epoch": 1.1453845676330001, "grad_norm": 0.4880976675149777, "learning_rate": 7.770001219307301e-05, "loss": 2.8909, "step": 18451 }, { "epoch": 1.145446644732758, "grad_norm": 0.5656403197010578, "learning_rate": 7.769700546733164e-05, "loss": 2.8905, "step": 18452 }, { "epoch": 1.145508721832516, "grad_norm": 0.49794799720747496, "learning_rate": 7.769399859708827e-05, "loss": 2.7531, "step": 18453 }, { "epoch": 1.1455707989322739, "grad_norm": 0.5434570860884126, "learning_rate": 7.769099158235858e-05, "loss": 2.8874, "step": 18454 }, { "epoch": 1.1456328760320318, "grad_norm": 0.5055816653982683, "learning_rate": 7.768798442315828e-05, "loss": 2.8358, "step": 18455 }, { "epoch": 1.1456949531317897, "grad_norm": 0.5377897634803815, "learning_rate": 7.768497711950302e-05, "loss": 2.8407, "step": 18456 }, { "epoch": 1.1457570302315476, "grad_norm": 0.5729018960840481, "learning_rate": 7.768196967140853e-05, "loss": 2.8632, "step": 18457 }, { "epoch": 1.1458191073313055, "grad_norm": 0.5514683056765423, "learning_rate": 7.767896207889047e-05, "loss": 2.88, "step": 18458 }, { "epoch": 1.1458811844310635, "grad_norm": 0.5300276943834452, "learning_rate": 7.767595434196456e-05, "loss": 2.8289, "step": 18459 }, { "epoch": 1.1459432615308214, "grad_norm": 0.5313002338220795, "learning_rate": 7.767294646064648e-05, "loss": 2.9063, "step": 18460 }, { "epoch": 1.1460053386305793, "grad_norm": 0.5339368585571308, "learning_rate": 7.766993843495192e-05, "loss": 2.8963, "step": 18461 }, { "epoch": 1.146067415730337, "grad_norm": 0.5265912373565298, "learning_rate": 7.766693026489655e-05, "loss": 2.8558, "step": 18462 }, { "epoch": 1.146129492830095, "grad_norm": 0.5500704731007415, "learning_rate": 7.766392195049612e-05, "loss": 2.8842, "step": 18463 }, { "epoch": 1.1461915699298528, "grad_norm": 0.5054947116794031, "learning_rate": 7.766091349176628e-05, "loss": 2.8445, "step": 18464 }, { "epoch": 1.1462536470296107, "grad_norm": 0.4966601195026534, "learning_rate": 7.765790488872273e-05, "loss": 2.9268, "step": 18465 }, { "epoch": 1.1463157241293687, "grad_norm": 0.5052585642379948, "learning_rate": 7.765489614138118e-05, "loss": 2.8263, "step": 18466 }, { "epoch": 1.1463778012291266, "grad_norm": 0.5057013128475945, "learning_rate": 7.765188724975734e-05, "loss": 2.968, "step": 18467 }, { "epoch": 1.1464398783288845, "grad_norm": 0.5393868175639742, "learning_rate": 7.764887821386689e-05, "loss": 2.8766, "step": 18468 }, { "epoch": 1.1465019554286424, "grad_norm": 0.4702332395003388, "learning_rate": 7.764586903372551e-05, "loss": 2.9324, "step": 18469 }, { "epoch": 1.1465640325284003, "grad_norm": 0.5044871685562627, "learning_rate": 7.764285970934893e-05, "loss": 2.7485, "step": 18470 }, { "epoch": 1.1466261096281583, "grad_norm": 0.43757683215557314, "learning_rate": 7.763985024075284e-05, "loss": 2.8467, "step": 18471 }, { "epoch": 1.146688186727916, "grad_norm": 0.5009282852914988, "learning_rate": 7.763684062795295e-05, "loss": 2.951, "step": 18472 }, { "epoch": 1.1467502638276739, "grad_norm": 0.48354334345138816, "learning_rate": 7.763383087096495e-05, "loss": 2.9318, "step": 18473 }, { "epoch": 1.1468123409274318, "grad_norm": 0.5203006754496697, "learning_rate": 7.763082096980454e-05, "loss": 2.8539, "step": 18474 }, { "epoch": 1.1468744180271897, "grad_norm": 0.5262603651371547, "learning_rate": 7.762781092448743e-05, "loss": 2.86, "step": 18475 }, { "epoch": 1.1469364951269476, "grad_norm": 0.5074196983566762, "learning_rate": 7.762480073502931e-05, "loss": 2.9173, "step": 18476 }, { "epoch": 1.1469985722267055, "grad_norm": 0.532553839894377, "learning_rate": 7.762179040144592e-05, "loss": 2.9303, "step": 18477 }, { "epoch": 1.1470606493264635, "grad_norm": 0.5221554727737456, "learning_rate": 7.761877992375294e-05, "loss": 2.8541, "step": 18478 }, { "epoch": 1.1471227264262214, "grad_norm": 0.5211476013633458, "learning_rate": 7.761576930196608e-05, "loss": 2.9174, "step": 18479 }, { "epoch": 1.1471848035259793, "grad_norm": 0.6087183960577185, "learning_rate": 7.761275853610104e-05, "loss": 2.9051, "step": 18480 }, { "epoch": 1.1472468806257372, "grad_norm": 0.5510557856731001, "learning_rate": 7.760974762617354e-05, "loss": 2.8817, "step": 18481 }, { "epoch": 1.1473089577254951, "grad_norm": 0.4536201632565861, "learning_rate": 7.760673657219929e-05, "loss": 2.8949, "step": 18482 }, { "epoch": 1.147371034825253, "grad_norm": 0.5609359188641975, "learning_rate": 7.760372537419397e-05, "loss": 2.9111, "step": 18483 }, { "epoch": 1.147433111925011, "grad_norm": 0.5323784847588832, "learning_rate": 7.760071403217334e-05, "loss": 2.9392, "step": 18484 }, { "epoch": 1.1474951890247689, "grad_norm": 0.5724309740974233, "learning_rate": 7.759770254615308e-05, "loss": 2.8986, "step": 18485 }, { "epoch": 1.1475572661245266, "grad_norm": 0.5117224482045963, "learning_rate": 7.759469091614891e-05, "loss": 2.8651, "step": 18486 }, { "epoch": 1.1476193432242845, "grad_norm": 0.6333514379331281, "learning_rate": 7.759167914217652e-05, "loss": 2.9477, "step": 18487 }, { "epoch": 1.1476814203240424, "grad_norm": 0.5955894486824258, "learning_rate": 7.758866722425167e-05, "loss": 2.8531, "step": 18488 }, { "epoch": 1.1477434974238003, "grad_norm": 0.513198605507497, "learning_rate": 7.758565516239002e-05, "loss": 2.8002, "step": 18489 }, { "epoch": 1.1478055745235582, "grad_norm": 0.5573226230853165, "learning_rate": 7.758264295660731e-05, "loss": 2.9075, "step": 18490 }, { "epoch": 1.1478676516233162, "grad_norm": 0.6090581999656802, "learning_rate": 7.757963060691927e-05, "loss": 2.8939, "step": 18491 }, { "epoch": 1.147929728723074, "grad_norm": 0.493770807769394, "learning_rate": 7.75766181133416e-05, "loss": 2.8228, "step": 18492 }, { "epoch": 1.147991805822832, "grad_norm": 0.5339289325989924, "learning_rate": 7.757360547589e-05, "loss": 2.9179, "step": 18493 }, { "epoch": 1.14805388292259, "grad_norm": 0.5604276098248858, "learning_rate": 7.757059269458024e-05, "loss": 2.906, "step": 18494 }, { "epoch": 1.1481159600223478, "grad_norm": 0.5313384038634569, "learning_rate": 7.7567579769428e-05, "loss": 2.8778, "step": 18495 }, { "epoch": 1.1481780371221055, "grad_norm": 0.5752362688440877, "learning_rate": 7.756456670044898e-05, "loss": 2.9047, "step": 18496 }, { "epoch": 1.1482401142218635, "grad_norm": 0.5103613635354566, "learning_rate": 7.756155348765893e-05, "loss": 2.839, "step": 18497 }, { "epoch": 1.1483021913216214, "grad_norm": 0.49343520160934595, "learning_rate": 7.755854013107358e-05, "loss": 2.8216, "step": 18498 }, { "epoch": 1.1483642684213793, "grad_norm": 0.5623118827759621, "learning_rate": 7.755552663070862e-05, "loss": 2.9254, "step": 18499 }, { "epoch": 1.1484263455211372, "grad_norm": 0.5073001330829117, "learning_rate": 7.755251298657979e-05, "loss": 2.8466, "step": 18500 }, { "epoch": 1.1484884226208951, "grad_norm": 0.49133655617895716, "learning_rate": 7.754949919870281e-05, "loss": 2.9221, "step": 18501 }, { "epoch": 1.148550499720653, "grad_norm": 0.5552988041877751, "learning_rate": 7.75464852670934e-05, "loss": 2.8256, "step": 18502 }, { "epoch": 1.148612576820411, "grad_norm": 0.5025395037178033, "learning_rate": 7.754347119176731e-05, "loss": 2.8386, "step": 18503 }, { "epoch": 1.1486746539201689, "grad_norm": 0.5077805435689416, "learning_rate": 7.754045697274024e-05, "loss": 2.8836, "step": 18504 }, { "epoch": 1.1487367310199268, "grad_norm": 0.5576983123931605, "learning_rate": 7.75374426100279e-05, "loss": 2.8584, "step": 18505 }, { "epoch": 1.1487988081196847, "grad_norm": 0.5055357733839703, "learning_rate": 7.753442810364606e-05, "loss": 2.897, "step": 18506 }, { "epoch": 1.1488608852194426, "grad_norm": 0.516300135110826, "learning_rate": 7.753141345361042e-05, "loss": 2.8631, "step": 18507 }, { "epoch": 1.1489229623192005, "grad_norm": 0.5618666421016165, "learning_rate": 7.752839865993671e-05, "loss": 2.92, "step": 18508 }, { "epoch": 1.1489850394189585, "grad_norm": 0.5057757393174686, "learning_rate": 7.752538372264067e-05, "loss": 2.9223, "step": 18509 }, { "epoch": 1.1490471165187162, "grad_norm": 0.5677570677904827, "learning_rate": 7.752236864173799e-05, "loss": 2.7686, "step": 18510 }, { "epoch": 1.149109193618474, "grad_norm": 0.5241547126144781, "learning_rate": 7.751935341724446e-05, "loss": 2.8579, "step": 18511 }, { "epoch": 1.149171270718232, "grad_norm": 0.590401240757282, "learning_rate": 7.751633804917578e-05, "loss": 2.9038, "step": 18512 }, { "epoch": 1.14923334781799, "grad_norm": 0.5372493214787121, "learning_rate": 7.751332253754767e-05, "loss": 2.8997, "step": 18513 }, { "epoch": 1.1492954249177478, "grad_norm": 0.5686816583172608, "learning_rate": 7.75103068823759e-05, "loss": 2.8917, "step": 18514 }, { "epoch": 1.1493575020175057, "grad_norm": 0.5066070627876801, "learning_rate": 7.750729108367615e-05, "loss": 2.8989, "step": 18515 }, { "epoch": 1.1494195791172637, "grad_norm": 0.5581727742405541, "learning_rate": 7.750427514146421e-05, "loss": 2.9257, "step": 18516 }, { "epoch": 1.1494816562170216, "grad_norm": 0.5637438849056178, "learning_rate": 7.750125905575579e-05, "loss": 2.7841, "step": 18517 }, { "epoch": 1.1495437333167795, "grad_norm": 0.5784586124287143, "learning_rate": 7.74982428265666e-05, "loss": 2.884, "step": 18518 }, { "epoch": 1.1496058104165374, "grad_norm": 0.5107852779030067, "learning_rate": 7.749522645391243e-05, "loss": 2.8952, "step": 18519 }, { "epoch": 1.1496678875162951, "grad_norm": 0.5447760219284822, "learning_rate": 7.749220993780897e-05, "loss": 2.7834, "step": 18520 }, { "epoch": 1.149729964616053, "grad_norm": 0.6176409685177889, "learning_rate": 7.748919327827198e-05, "loss": 2.8417, "step": 18521 }, { "epoch": 1.149792041715811, "grad_norm": 0.6196865959757233, "learning_rate": 7.748617647531721e-05, "loss": 2.8902, "step": 18522 }, { "epoch": 1.1498541188155689, "grad_norm": 0.5622357171476259, "learning_rate": 7.748315952896037e-05, "loss": 2.905, "step": 18523 }, { "epoch": 1.1499161959153268, "grad_norm": 0.5721296077173508, "learning_rate": 7.748014243921722e-05, "loss": 2.8443, "step": 18524 }, { "epoch": 1.1499782730150847, "grad_norm": 0.5020310159653594, "learning_rate": 7.74771252061035e-05, "loss": 2.9424, "step": 18525 }, { "epoch": 1.1500403501148426, "grad_norm": 0.5435026823133801, "learning_rate": 7.747410782963495e-05, "loss": 2.8182, "step": 18526 }, { "epoch": 1.1501024272146005, "grad_norm": 0.5614834447799669, "learning_rate": 7.747109030982729e-05, "loss": 2.9379, "step": 18527 }, { "epoch": 1.1501645043143585, "grad_norm": 0.5289738644271367, "learning_rate": 7.74680726466963e-05, "loss": 2.9556, "step": 18528 }, { "epoch": 1.1502265814141164, "grad_norm": 0.5274335149023887, "learning_rate": 7.74650548402577e-05, "loss": 2.8187, "step": 18529 }, { "epoch": 1.1502886585138743, "grad_norm": 0.5662530099818248, "learning_rate": 7.746203689052725e-05, "loss": 2.8398, "step": 18530 }, { "epoch": 1.1503507356136322, "grad_norm": 0.5735241847821363, "learning_rate": 7.745901879752067e-05, "loss": 2.8724, "step": 18531 }, { "epoch": 1.1504128127133901, "grad_norm": 0.5054873715114635, "learning_rate": 7.745600056125373e-05, "loss": 2.8564, "step": 18532 }, { "epoch": 1.150474889813148, "grad_norm": 0.571662428699314, "learning_rate": 7.745298218174219e-05, "loss": 2.9194, "step": 18533 }, { "epoch": 1.1505369669129057, "grad_norm": 0.5486900726129429, "learning_rate": 7.744996365900175e-05, "loss": 2.857, "step": 18534 }, { "epoch": 1.1505990440126637, "grad_norm": 0.6112234007619785, "learning_rate": 7.744694499304821e-05, "loss": 2.8855, "step": 18535 }, { "epoch": 1.1506611211124216, "grad_norm": 0.5478380442972766, "learning_rate": 7.744392618389727e-05, "loss": 2.8294, "step": 18536 }, { "epoch": 1.1507231982121795, "grad_norm": 0.5665532414668236, "learning_rate": 7.744090723156472e-05, "loss": 2.765, "step": 18537 }, { "epoch": 1.1507852753119374, "grad_norm": 0.586614675218431, "learning_rate": 7.743788813606629e-05, "loss": 2.9549, "step": 18538 }, { "epoch": 1.1508473524116953, "grad_norm": 0.5790233076721182, "learning_rate": 7.743486889741774e-05, "loss": 2.9053, "step": 18539 }, { "epoch": 1.1509094295114533, "grad_norm": 0.6677343169069828, "learning_rate": 7.743184951563482e-05, "loss": 2.8452, "step": 18540 }, { "epoch": 1.1509715066112112, "grad_norm": 0.6021988958004039, "learning_rate": 7.742882999073329e-05, "loss": 2.768, "step": 18541 }, { "epoch": 1.151033583710969, "grad_norm": 0.6763051818925522, "learning_rate": 7.742581032272889e-05, "loss": 2.9121, "step": 18542 }, { "epoch": 1.151095660810727, "grad_norm": 0.604986429377542, "learning_rate": 7.742279051163737e-05, "loss": 2.8445, "step": 18543 }, { "epoch": 1.1511577379104847, "grad_norm": 0.5410073895980966, "learning_rate": 7.741977055747449e-05, "loss": 2.8705, "step": 18544 }, { "epoch": 1.1512198150102426, "grad_norm": 0.6820452126103914, "learning_rate": 7.741675046025602e-05, "loss": 2.756, "step": 18545 }, { "epoch": 1.1512818921100005, "grad_norm": 0.5702399638830146, "learning_rate": 7.741373021999772e-05, "loss": 2.9114, "step": 18546 }, { "epoch": 1.1513439692097585, "grad_norm": 0.6082750075828611, "learning_rate": 7.741070983671532e-05, "loss": 2.8704, "step": 18547 }, { "epoch": 1.1514060463095164, "grad_norm": 0.5315135133968629, "learning_rate": 7.740768931042459e-05, "loss": 2.8333, "step": 18548 }, { "epoch": 1.1514681234092743, "grad_norm": 0.5755522936569402, "learning_rate": 7.740466864114128e-05, "loss": 2.9701, "step": 18549 }, { "epoch": 1.1515302005090322, "grad_norm": 0.5726427521617589, "learning_rate": 7.74016478288812e-05, "loss": 2.7986, "step": 18550 }, { "epoch": 1.1515922776087901, "grad_norm": 0.5445584141264831, "learning_rate": 7.739862687366004e-05, "loss": 2.8955, "step": 18551 }, { "epoch": 1.151654354708548, "grad_norm": 0.5580838417007996, "learning_rate": 7.739560577549361e-05, "loss": 2.8536, "step": 18552 }, { "epoch": 1.151716431808306, "grad_norm": 0.5694398866900069, "learning_rate": 7.739258453439764e-05, "loss": 2.8626, "step": 18553 }, { "epoch": 1.1517785089080639, "grad_norm": 0.5013704294620757, "learning_rate": 7.738956315038791e-05, "loss": 2.9138, "step": 18554 }, { "epoch": 1.1518405860078218, "grad_norm": 0.5113427622705776, "learning_rate": 7.738654162348018e-05, "loss": 2.8649, "step": 18555 }, { "epoch": 1.1519026631075797, "grad_norm": 0.47088098924826355, "learning_rate": 7.738351995369023e-05, "loss": 2.814, "step": 18556 }, { "epoch": 1.1519647402073376, "grad_norm": 0.5519178580364809, "learning_rate": 7.738049814103379e-05, "loss": 2.8858, "step": 18557 }, { "epoch": 1.1520268173070953, "grad_norm": 0.5501371093557926, "learning_rate": 7.737747618552664e-05, "loss": 2.8854, "step": 18558 }, { "epoch": 1.1520888944068532, "grad_norm": 0.5166527014301142, "learning_rate": 7.737445408718457e-05, "loss": 2.9008, "step": 18559 }, { "epoch": 1.1521509715066112, "grad_norm": 0.605261564184465, "learning_rate": 7.737143184602331e-05, "loss": 2.8944, "step": 18560 }, { "epoch": 1.152213048606369, "grad_norm": 0.5464161719189599, "learning_rate": 7.736840946205866e-05, "loss": 2.9379, "step": 18561 }, { "epoch": 1.152275125706127, "grad_norm": 0.5371314304331751, "learning_rate": 7.736538693530638e-05, "loss": 2.9109, "step": 18562 }, { "epoch": 1.152337202805885, "grad_norm": 0.5866506686883164, "learning_rate": 7.73623642657822e-05, "loss": 2.9111, "step": 18563 }, { "epoch": 1.1523992799056428, "grad_norm": 0.4643717380886597, "learning_rate": 7.735934145350195e-05, "loss": 2.901, "step": 18564 }, { "epoch": 1.1524613570054008, "grad_norm": 0.5571992446936741, "learning_rate": 7.735631849848136e-05, "loss": 2.8066, "step": 18565 }, { "epoch": 1.1525234341051587, "grad_norm": 0.4676167299273012, "learning_rate": 7.735329540073621e-05, "loss": 2.8089, "step": 18566 }, { "epoch": 1.1525855112049166, "grad_norm": 0.48326870497150337, "learning_rate": 7.73502721602823e-05, "loss": 2.8942, "step": 18567 }, { "epoch": 1.1526475883046743, "grad_norm": 0.5658334400680659, "learning_rate": 7.734724877713537e-05, "loss": 2.8198, "step": 18568 }, { "epoch": 1.1527096654044322, "grad_norm": 0.5186181613271724, "learning_rate": 7.734422525131121e-05, "loss": 2.8563, "step": 18569 }, { "epoch": 1.1527717425041901, "grad_norm": 0.5115895290143537, "learning_rate": 7.734120158282559e-05, "loss": 2.8438, "step": 18570 }, { "epoch": 1.152833819603948, "grad_norm": 0.5485710546568181, "learning_rate": 7.733817777169428e-05, "loss": 2.9065, "step": 18571 }, { "epoch": 1.152895896703706, "grad_norm": 0.6845771691255119, "learning_rate": 7.733515381793304e-05, "loss": 2.8409, "step": 18572 }, { "epoch": 1.1529579738034639, "grad_norm": 0.5303526480228766, "learning_rate": 7.73321297215577e-05, "loss": 2.9417, "step": 18573 }, { "epoch": 1.1530200509032218, "grad_norm": 0.5290939674099865, "learning_rate": 7.732910548258399e-05, "loss": 2.8501, "step": 18574 }, { "epoch": 1.1530821280029797, "grad_norm": 0.5257643324240271, "learning_rate": 7.73260811010277e-05, "loss": 2.8245, "step": 18575 }, { "epoch": 1.1531442051027376, "grad_norm": 0.4994954155939714, "learning_rate": 7.732305657690461e-05, "loss": 2.9442, "step": 18576 }, { "epoch": 1.1532062822024955, "grad_norm": 0.5235823530495284, "learning_rate": 7.732003191023049e-05, "loss": 2.86, "step": 18577 }, { "epoch": 1.1532683593022535, "grad_norm": 0.5020382285798076, "learning_rate": 7.731700710102115e-05, "loss": 2.841, "step": 18578 }, { "epoch": 1.1533304364020114, "grad_norm": 0.5580314527076395, "learning_rate": 7.731398214929235e-05, "loss": 2.9334, "step": 18579 }, { "epoch": 1.1533925135017693, "grad_norm": 0.5453940639324589, "learning_rate": 7.731095705505988e-05, "loss": 2.8406, "step": 18580 }, { "epoch": 1.1534545906015272, "grad_norm": 0.5094524020497142, "learning_rate": 7.73079318183395e-05, "loss": 2.8879, "step": 18581 }, { "epoch": 1.153516667701285, "grad_norm": 0.5023365560108453, "learning_rate": 7.730490643914701e-05, "loss": 2.7838, "step": 18582 }, { "epoch": 1.1535787448010428, "grad_norm": 0.5387678031900588, "learning_rate": 7.730188091749822e-05, "loss": 2.864, "step": 18583 }, { "epoch": 1.1536408219008008, "grad_norm": 0.5431828522681416, "learning_rate": 7.729885525340887e-05, "loss": 2.9405, "step": 18584 }, { "epoch": 1.1537028990005587, "grad_norm": 0.6009529791002346, "learning_rate": 7.729582944689476e-05, "loss": 2.902, "step": 18585 }, { "epoch": 1.1537649761003166, "grad_norm": 0.563610437842545, "learning_rate": 7.72928034979717e-05, "loss": 2.8112, "step": 18586 }, { "epoch": 1.1538270532000745, "grad_norm": 0.5386899130431144, "learning_rate": 7.728977740665543e-05, "loss": 2.8964, "step": 18587 }, { "epoch": 1.1538891302998324, "grad_norm": 0.5756579159614731, "learning_rate": 7.72867511729618e-05, "loss": 2.9435, "step": 18588 }, { "epoch": 1.1539512073995903, "grad_norm": 0.4231569099508455, "learning_rate": 7.728372479690656e-05, "loss": 2.8281, "step": 18589 }, { "epoch": 1.1540132844993483, "grad_norm": 0.5103001039754951, "learning_rate": 7.728069827850548e-05, "loss": 2.8216, "step": 18590 }, { "epoch": 1.1540753615991062, "grad_norm": 0.48494004547556296, "learning_rate": 7.72776716177744e-05, "loss": 2.8486, "step": 18591 }, { "epoch": 1.1541374386988639, "grad_norm": 0.4835071325920039, "learning_rate": 7.727464481472907e-05, "loss": 2.7254, "step": 18592 }, { "epoch": 1.1541995157986218, "grad_norm": 0.4804725336685686, "learning_rate": 7.727161786938531e-05, "loss": 2.8449, "step": 18593 }, { "epoch": 1.1542615928983797, "grad_norm": 0.5332887182598551, "learning_rate": 7.726859078175887e-05, "loss": 2.915, "step": 18594 }, { "epoch": 1.1543236699981376, "grad_norm": 0.5722380751920956, "learning_rate": 7.72655635518656e-05, "loss": 2.8969, "step": 18595 }, { "epoch": 1.1543857470978955, "grad_norm": 0.49939604880763927, "learning_rate": 7.726253617972126e-05, "loss": 2.8893, "step": 18596 }, { "epoch": 1.1544478241976535, "grad_norm": 0.5667850937463487, "learning_rate": 7.725950866534165e-05, "loss": 2.8927, "step": 18597 }, { "epoch": 1.1545099012974114, "grad_norm": 0.5256885407151445, "learning_rate": 7.725648100874257e-05, "loss": 2.8918, "step": 18598 }, { "epoch": 1.1545719783971693, "grad_norm": 0.5214555996731612, "learning_rate": 7.725345320993979e-05, "loss": 2.8357, "step": 18599 }, { "epoch": 1.1546340554969272, "grad_norm": 0.545164457627294, "learning_rate": 7.725042526894915e-05, "loss": 2.9241, "step": 18600 }, { "epoch": 1.1546961325966851, "grad_norm": 0.505071165858546, "learning_rate": 7.724739718578642e-05, "loss": 2.8558, "step": 18601 }, { "epoch": 1.154758209696443, "grad_norm": 0.5417407028023644, "learning_rate": 7.724436896046741e-05, "loss": 2.8056, "step": 18602 }, { "epoch": 1.154820286796201, "grad_norm": 0.524778555534268, "learning_rate": 7.72413405930079e-05, "loss": 2.9389, "step": 18603 }, { "epoch": 1.1548823638959589, "grad_norm": 0.5632693803994263, "learning_rate": 7.723831208342371e-05, "loss": 2.8903, "step": 18604 }, { "epoch": 1.1549444409957168, "grad_norm": 0.47587781718517164, "learning_rate": 7.723528343173066e-05, "loss": 2.8641, "step": 18605 }, { "epoch": 1.1550065180954745, "grad_norm": 0.5581927427336749, "learning_rate": 7.723225463794449e-05, "loss": 2.8348, "step": 18606 }, { "epoch": 1.1550685951952324, "grad_norm": 0.49772933111956763, "learning_rate": 7.722922570208104e-05, "loss": 2.8868, "step": 18607 }, { "epoch": 1.1551306722949903, "grad_norm": 0.5517276451843591, "learning_rate": 7.722619662415612e-05, "loss": 2.8428, "step": 18608 }, { "epoch": 1.1551927493947483, "grad_norm": 0.5010444639713401, "learning_rate": 7.722316740418551e-05, "loss": 2.8724, "step": 18609 }, { "epoch": 1.1552548264945062, "grad_norm": 0.5460447002861735, "learning_rate": 7.722013804218505e-05, "loss": 2.8687, "step": 18610 }, { "epoch": 1.155316903594264, "grad_norm": 0.5672180671967304, "learning_rate": 7.72171085381705e-05, "loss": 2.8381, "step": 18611 }, { "epoch": 1.155378980694022, "grad_norm": 0.5395547029516696, "learning_rate": 7.721407889215771e-05, "loss": 2.8637, "step": 18612 }, { "epoch": 1.15544105779378, "grad_norm": 0.5365170906505257, "learning_rate": 7.721104910416243e-05, "loss": 2.9311, "step": 18613 }, { "epoch": 1.1555031348935378, "grad_norm": 0.5160018815109785, "learning_rate": 7.720801917420053e-05, "loss": 2.8423, "step": 18614 }, { "epoch": 1.1555652119932958, "grad_norm": 0.49121391037634804, "learning_rate": 7.720498910228777e-05, "loss": 2.8561, "step": 18615 }, { "epoch": 1.1556272890930535, "grad_norm": 0.5403730246256069, "learning_rate": 7.720195888844e-05, "loss": 2.9016, "step": 18616 }, { "epoch": 1.1556893661928114, "grad_norm": 0.5618459024416204, "learning_rate": 7.7198928532673e-05, "loss": 2.9323, "step": 18617 }, { "epoch": 1.1557514432925693, "grad_norm": 0.5251103046888128, "learning_rate": 7.719589803500257e-05, "loss": 2.8551, "step": 18618 }, { "epoch": 1.1558135203923272, "grad_norm": 0.5626479854893112, "learning_rate": 7.719286739544455e-05, "loss": 2.881, "step": 18619 }, { "epoch": 1.1558755974920851, "grad_norm": 0.5914490939989986, "learning_rate": 7.718983661401473e-05, "loss": 2.9866, "step": 18620 }, { "epoch": 1.155937674591843, "grad_norm": 0.5567103024424515, "learning_rate": 7.718680569072895e-05, "loss": 2.8911, "step": 18621 }, { "epoch": 1.155999751691601, "grad_norm": 0.5194228783492134, "learning_rate": 7.7183774625603e-05, "loss": 2.8668, "step": 18622 }, { "epoch": 1.1560618287913589, "grad_norm": 0.5669180038021807, "learning_rate": 7.718074341865269e-05, "loss": 2.8714, "step": 18623 }, { "epoch": 1.1561239058911168, "grad_norm": 0.47265883515967233, "learning_rate": 7.717771206989385e-05, "loss": 2.8469, "step": 18624 }, { "epoch": 1.1561859829908747, "grad_norm": 0.5493340383454925, "learning_rate": 7.717468057934229e-05, "loss": 2.8177, "step": 18625 }, { "epoch": 1.1562480600906326, "grad_norm": 0.5468336907179917, "learning_rate": 7.717164894701381e-05, "loss": 2.8873, "step": 18626 }, { "epoch": 1.1563101371903906, "grad_norm": 0.50423985165515, "learning_rate": 7.716861717292425e-05, "loss": 2.8788, "step": 18627 }, { "epoch": 1.1563722142901485, "grad_norm": 0.5435671082207331, "learning_rate": 7.716558525708944e-05, "loss": 2.9057, "step": 18628 }, { "epoch": 1.1564342913899064, "grad_norm": 0.4908026179362368, "learning_rate": 7.716255319952515e-05, "loss": 2.8592, "step": 18629 }, { "epoch": 1.156496368489664, "grad_norm": 0.5010218014285088, "learning_rate": 7.715952100024723e-05, "loss": 2.8212, "step": 18630 }, { "epoch": 1.156558445589422, "grad_norm": 0.5121334812243215, "learning_rate": 7.71564886592715e-05, "loss": 2.8122, "step": 18631 }, { "epoch": 1.15662052268918, "grad_norm": 0.49847203975096765, "learning_rate": 7.715345617661377e-05, "loss": 2.8923, "step": 18632 }, { "epoch": 1.1566825997889378, "grad_norm": 0.47654480197750404, "learning_rate": 7.715042355228987e-05, "loss": 2.8372, "step": 18633 }, { "epoch": 1.1567446768886958, "grad_norm": 0.5054998408901217, "learning_rate": 7.714739078631563e-05, "loss": 2.8442, "step": 18634 }, { "epoch": 1.1568067539884537, "grad_norm": 0.4951543248561367, "learning_rate": 7.714435787870685e-05, "loss": 2.905, "step": 18635 }, { "epoch": 1.1568688310882116, "grad_norm": 0.517720530370747, "learning_rate": 7.714132482947937e-05, "loss": 2.8337, "step": 18636 }, { "epoch": 1.1569309081879695, "grad_norm": 0.49258601993825646, "learning_rate": 7.713829163864901e-05, "loss": 2.9027, "step": 18637 }, { "epoch": 1.1569929852877274, "grad_norm": 0.5374087799646241, "learning_rate": 7.713525830623159e-05, "loss": 2.7659, "step": 18638 }, { "epoch": 1.1570550623874853, "grad_norm": 0.5197355112640756, "learning_rate": 7.713222483224294e-05, "loss": 2.8718, "step": 18639 }, { "epoch": 1.157117139487243, "grad_norm": 0.5178358479993201, "learning_rate": 7.712919121669889e-05, "loss": 2.876, "step": 18640 }, { "epoch": 1.157179216587001, "grad_norm": 0.5536350689401078, "learning_rate": 7.712615745961527e-05, "loss": 2.7612, "step": 18641 }, { "epoch": 1.1572412936867589, "grad_norm": 0.5280240227587966, "learning_rate": 7.71231235610079e-05, "loss": 2.8588, "step": 18642 }, { "epoch": 1.1573033707865168, "grad_norm": 0.5459598786473564, "learning_rate": 7.71200895208926e-05, "loss": 2.8386, "step": 18643 }, { "epoch": 1.1573654478862747, "grad_norm": 0.5135597715868228, "learning_rate": 7.711705533928521e-05, "loss": 2.9079, "step": 18644 }, { "epoch": 1.1574275249860326, "grad_norm": 0.4826135293288551, "learning_rate": 7.711402101620158e-05, "loss": 2.8378, "step": 18645 }, { "epoch": 1.1574896020857905, "grad_norm": 0.48965386295241936, "learning_rate": 7.711098655165749e-05, "loss": 2.8611, "step": 18646 }, { "epoch": 1.1575516791855485, "grad_norm": 0.48522109306311706, "learning_rate": 7.710795194566882e-05, "loss": 2.8983, "step": 18647 }, { "epoch": 1.1576137562853064, "grad_norm": 0.5266770846608056, "learning_rate": 7.710491719825138e-05, "loss": 2.884, "step": 18648 }, { "epoch": 1.1576758333850643, "grad_norm": 0.5399485783023876, "learning_rate": 7.710188230942102e-05, "loss": 2.8899, "step": 18649 }, { "epoch": 1.1577379104848222, "grad_norm": 0.5206862210093308, "learning_rate": 7.709884727919353e-05, "loss": 2.8207, "step": 18650 }, { "epoch": 1.1577999875845801, "grad_norm": 0.5298840408507809, "learning_rate": 7.70958121075848e-05, "loss": 2.8507, "step": 18651 }, { "epoch": 1.157862064684338, "grad_norm": 0.4545900895950849, "learning_rate": 7.709277679461064e-05, "loss": 2.9011, "step": 18652 }, { "epoch": 1.1579241417840958, "grad_norm": 0.49328992389523935, "learning_rate": 7.708974134028687e-05, "loss": 2.9186, "step": 18653 }, { "epoch": 1.1579862188838537, "grad_norm": 0.45901570314668405, "learning_rate": 7.708670574462935e-05, "loss": 2.8867, "step": 18654 }, { "epoch": 1.1580482959836116, "grad_norm": 0.48173537424549395, "learning_rate": 7.708367000765393e-05, "loss": 2.8118, "step": 18655 }, { "epoch": 1.1581103730833695, "grad_norm": 0.5221943397918689, "learning_rate": 7.708063412937641e-05, "loss": 2.8597, "step": 18656 }, { "epoch": 1.1581724501831274, "grad_norm": 0.4750505759267525, "learning_rate": 7.707759810981265e-05, "loss": 2.9383, "step": 18657 }, { "epoch": 1.1582345272828853, "grad_norm": 0.49883592221611833, "learning_rate": 7.707456194897848e-05, "loss": 2.8072, "step": 18658 }, { "epoch": 1.1582966043826433, "grad_norm": 0.4873092131318533, "learning_rate": 7.707152564688975e-05, "loss": 2.8605, "step": 18659 }, { "epoch": 1.1583586814824012, "grad_norm": 0.4848507421433759, "learning_rate": 7.70684892035623e-05, "loss": 2.8669, "step": 18660 }, { "epoch": 1.158420758582159, "grad_norm": 0.4654141300625208, "learning_rate": 7.7065452619012e-05, "loss": 2.9847, "step": 18661 }, { "epoch": 1.158482835681917, "grad_norm": 0.45104787204106417, "learning_rate": 7.706241589325462e-05, "loss": 2.8221, "step": 18662 }, { "epoch": 1.158544912781675, "grad_norm": 0.4742464946693696, "learning_rate": 7.705937902630606e-05, "loss": 2.8516, "step": 18663 }, { "epoch": 1.1586069898814326, "grad_norm": 0.5403378644718343, "learning_rate": 7.705634201818216e-05, "loss": 2.9553, "step": 18664 }, { "epoch": 1.1586690669811905, "grad_norm": 0.5267576563477417, "learning_rate": 7.705330486889874e-05, "loss": 3.0156, "step": 18665 }, { "epoch": 1.1587311440809485, "grad_norm": 0.5822255623088078, "learning_rate": 7.705026757847168e-05, "loss": 2.8243, "step": 18666 }, { "epoch": 1.1587932211807064, "grad_norm": 0.5230404785767954, "learning_rate": 7.70472301469168e-05, "loss": 2.8712, "step": 18667 }, { "epoch": 1.1588552982804643, "grad_norm": 0.5958487410876243, "learning_rate": 7.704419257424995e-05, "loss": 2.8899, "step": 18668 }, { "epoch": 1.1589173753802222, "grad_norm": 0.6058026039388087, "learning_rate": 7.704115486048699e-05, "loss": 2.9013, "step": 18669 }, { "epoch": 1.1589794524799801, "grad_norm": 0.5822783592036487, "learning_rate": 7.703811700564376e-05, "loss": 2.9322, "step": 18670 }, { "epoch": 1.159041529579738, "grad_norm": 0.5209744707248463, "learning_rate": 7.703507900973611e-05, "loss": 2.8613, "step": 18671 }, { "epoch": 1.159103606679496, "grad_norm": 0.540025459104422, "learning_rate": 7.703204087277988e-05, "loss": 2.9084, "step": 18672 }, { "epoch": 1.1591656837792539, "grad_norm": 0.5386883639771185, "learning_rate": 7.702900259479094e-05, "loss": 2.8174, "step": 18673 }, { "epoch": 1.1592277608790118, "grad_norm": 0.5200052322069445, "learning_rate": 7.702596417578514e-05, "loss": 2.8755, "step": 18674 }, { "epoch": 1.1592898379787697, "grad_norm": 0.5544670902819244, "learning_rate": 7.70229256157783e-05, "loss": 2.9073, "step": 18675 }, { "epoch": 1.1593519150785276, "grad_norm": 0.4686757982652263, "learning_rate": 7.701988691478633e-05, "loss": 2.7258, "step": 18676 }, { "epoch": 1.1594139921782853, "grad_norm": 0.561991514742589, "learning_rate": 7.701684807282502e-05, "loss": 2.7894, "step": 18677 }, { "epoch": 1.1594760692780433, "grad_norm": 0.5221501647185508, "learning_rate": 7.701380908991027e-05, "loss": 2.9224, "step": 18678 }, { "epoch": 1.1595381463778012, "grad_norm": 0.5388272919495614, "learning_rate": 7.701076996605791e-05, "loss": 2.8063, "step": 18679 }, { "epoch": 1.159600223477559, "grad_norm": 0.5114424155944347, "learning_rate": 7.70077307012838e-05, "loss": 2.7867, "step": 18680 }, { "epoch": 1.159662300577317, "grad_norm": 0.5291541213203012, "learning_rate": 7.700469129560382e-05, "loss": 2.8812, "step": 18681 }, { "epoch": 1.159724377677075, "grad_norm": 0.513848771070106, "learning_rate": 7.70016517490338e-05, "loss": 2.8906, "step": 18682 }, { "epoch": 1.1597864547768328, "grad_norm": 0.5738925396501651, "learning_rate": 7.699861206158963e-05, "loss": 2.8136, "step": 18683 }, { "epoch": 1.1598485318765908, "grad_norm": 0.5677065951352822, "learning_rate": 7.69955722332871e-05, "loss": 2.9213, "step": 18684 }, { "epoch": 1.1599106089763487, "grad_norm": 0.5529166416473881, "learning_rate": 7.699253226414216e-05, "loss": 2.7689, "step": 18685 }, { "epoch": 1.1599726860761066, "grad_norm": 0.5214409970876438, "learning_rate": 7.698949215417061e-05, "loss": 2.9503, "step": 18686 }, { "epoch": 1.1600347631758645, "grad_norm": 0.6521877627698939, "learning_rate": 7.698645190338834e-05, "loss": 2.8646, "step": 18687 }, { "epoch": 1.1600968402756222, "grad_norm": 0.4910137926065976, "learning_rate": 7.698341151181118e-05, "loss": 2.8426, "step": 18688 }, { "epoch": 1.1601589173753801, "grad_norm": 0.6610164117261765, "learning_rate": 7.698037097945503e-05, "loss": 2.8291, "step": 18689 }, { "epoch": 1.160220994475138, "grad_norm": 0.5459191279166826, "learning_rate": 7.697733030633571e-05, "loss": 2.9568, "step": 18690 }, { "epoch": 1.160283071574896, "grad_norm": 0.6467119490567026, "learning_rate": 7.697428949246913e-05, "loss": 2.923, "step": 18691 }, { "epoch": 1.1603451486746539, "grad_norm": 0.5905237131210613, "learning_rate": 7.697124853787113e-05, "loss": 2.8576, "step": 18692 }, { "epoch": 1.1604072257744118, "grad_norm": 0.6398948369573948, "learning_rate": 7.696820744255757e-05, "loss": 2.9149, "step": 18693 }, { "epoch": 1.1604693028741697, "grad_norm": 0.5890979725809458, "learning_rate": 7.696516620654434e-05, "loss": 2.8803, "step": 18694 }, { "epoch": 1.1605313799739276, "grad_norm": 0.6325558978097003, "learning_rate": 7.696212482984728e-05, "loss": 2.8785, "step": 18695 }, { "epoch": 1.1605934570736856, "grad_norm": 0.5459273438144951, "learning_rate": 7.695908331248227e-05, "loss": 2.984, "step": 18696 }, { "epoch": 1.1606555341734435, "grad_norm": 0.6009967918326238, "learning_rate": 7.695604165446518e-05, "loss": 2.7609, "step": 18697 }, { "epoch": 1.1607176112732014, "grad_norm": 0.5793410476645414, "learning_rate": 7.695299985581188e-05, "loss": 2.812, "step": 18698 }, { "epoch": 1.1607796883729593, "grad_norm": 0.6009523344068687, "learning_rate": 7.694995791653823e-05, "loss": 2.8391, "step": 18699 }, { "epoch": 1.1608417654727172, "grad_norm": 0.506817150319026, "learning_rate": 7.694691583666012e-05, "loss": 2.8647, "step": 18700 }, { "epoch": 1.160903842572475, "grad_norm": 0.6263547519630562, "learning_rate": 7.69438736161934e-05, "loss": 2.9496, "step": 18701 }, { "epoch": 1.1609659196722328, "grad_norm": 0.4955466700659768, "learning_rate": 7.694083125515395e-05, "loss": 2.8373, "step": 18702 }, { "epoch": 1.1610279967719908, "grad_norm": 0.638856936832901, "learning_rate": 7.693778875355764e-05, "loss": 2.8445, "step": 18703 }, { "epoch": 1.1610900738717487, "grad_norm": 0.5275586290907801, "learning_rate": 7.693474611142035e-05, "loss": 2.9347, "step": 18704 }, { "epoch": 1.1611521509715066, "grad_norm": 0.6650207113683327, "learning_rate": 7.693170332875794e-05, "loss": 2.9436, "step": 18705 }, { "epoch": 1.1612142280712645, "grad_norm": 0.6110464306279418, "learning_rate": 7.69286604055863e-05, "loss": 2.8248, "step": 18706 }, { "epoch": 1.1612763051710224, "grad_norm": 0.5154624451537586, "learning_rate": 7.692561734192132e-05, "loss": 2.8191, "step": 18707 }, { "epoch": 1.1613383822707803, "grad_norm": 0.6225887515673456, "learning_rate": 7.692257413777886e-05, "loss": 2.7567, "step": 18708 }, { "epoch": 1.1614004593705383, "grad_norm": 0.48963496429689735, "learning_rate": 7.691953079317478e-05, "loss": 2.7662, "step": 18709 }, { "epoch": 1.1614625364702962, "grad_norm": 0.590043318378615, "learning_rate": 7.691648730812497e-05, "loss": 2.9557, "step": 18710 }, { "epoch": 1.161524613570054, "grad_norm": 0.5046705615150765, "learning_rate": 7.691344368264531e-05, "loss": 2.7984, "step": 18711 }, { "epoch": 1.1615866906698118, "grad_norm": 0.5567037446223807, "learning_rate": 7.691039991675169e-05, "loss": 2.8586, "step": 18712 }, { "epoch": 1.1616487677695697, "grad_norm": 0.5317020456457326, "learning_rate": 7.690735601045998e-05, "loss": 2.9138, "step": 18713 }, { "epoch": 1.1617108448693276, "grad_norm": 0.5242177926294991, "learning_rate": 7.690431196378604e-05, "loss": 2.837, "step": 18714 }, { "epoch": 1.1617729219690855, "grad_norm": 0.5263595722051019, "learning_rate": 7.69012677767458e-05, "loss": 2.91, "step": 18715 }, { "epoch": 1.1618349990688435, "grad_norm": 0.5268835532239673, "learning_rate": 7.689822344935511e-05, "loss": 2.8173, "step": 18716 }, { "epoch": 1.1618970761686014, "grad_norm": 0.6054977163953741, "learning_rate": 7.689517898162985e-05, "loss": 2.838, "step": 18717 }, { "epoch": 1.1619591532683593, "grad_norm": 0.4970174135490572, "learning_rate": 7.689213437358593e-05, "loss": 2.9129, "step": 18718 }, { "epoch": 1.1620212303681172, "grad_norm": 0.5954009243232388, "learning_rate": 7.688908962523921e-05, "loss": 2.7936, "step": 18719 }, { "epoch": 1.1620833074678751, "grad_norm": 0.5645576037166412, "learning_rate": 7.688604473660557e-05, "loss": 2.845, "step": 18720 }, { "epoch": 1.162145384567633, "grad_norm": 0.5240017141758202, "learning_rate": 7.688299970770093e-05, "loss": 2.8474, "step": 18721 }, { "epoch": 1.162207461667391, "grad_norm": 0.5539257568023646, "learning_rate": 7.687995453854113e-05, "loss": 2.8509, "step": 18722 }, { "epoch": 1.162269538767149, "grad_norm": 0.5366505275495191, "learning_rate": 7.687690922914209e-05, "loss": 2.9486, "step": 18723 }, { "epoch": 1.1623316158669068, "grad_norm": 0.5083181358108462, "learning_rate": 7.687386377951968e-05, "loss": 2.8343, "step": 18724 }, { "epoch": 1.1623936929666645, "grad_norm": 0.5260338231877416, "learning_rate": 7.687081818968982e-05, "loss": 2.7972, "step": 18725 }, { "epoch": 1.1624557700664224, "grad_norm": 0.4998313500171137, "learning_rate": 7.686777245966837e-05, "loss": 2.8591, "step": 18726 }, { "epoch": 1.1625178471661803, "grad_norm": 0.5165055299280084, "learning_rate": 7.686472658947122e-05, "loss": 2.7472, "step": 18727 }, { "epoch": 1.1625799242659383, "grad_norm": 0.49928369472764916, "learning_rate": 7.686168057911429e-05, "loss": 2.8585, "step": 18728 }, { "epoch": 1.1626420013656962, "grad_norm": 0.521698896882052, "learning_rate": 7.685863442861343e-05, "loss": 2.8186, "step": 18729 }, { "epoch": 1.162704078465454, "grad_norm": 0.5321779841380361, "learning_rate": 7.685558813798456e-05, "loss": 2.8949, "step": 18730 }, { "epoch": 1.162766155565212, "grad_norm": 0.5245068004592766, "learning_rate": 7.685254170724356e-05, "loss": 2.9215, "step": 18731 }, { "epoch": 1.16282823266497, "grad_norm": 0.47764820185685936, "learning_rate": 7.684949513640636e-05, "loss": 2.8313, "step": 18732 }, { "epoch": 1.1628903097647278, "grad_norm": 0.5376033107034035, "learning_rate": 7.68464484254888e-05, "loss": 2.8847, "step": 18733 }, { "epoch": 1.1629523868644858, "grad_norm": 0.5178633857454107, "learning_rate": 7.68434015745068e-05, "loss": 2.9196, "step": 18734 }, { "epoch": 1.1630144639642437, "grad_norm": 0.5260108775076597, "learning_rate": 7.684035458347627e-05, "loss": 2.848, "step": 18735 }, { "epoch": 1.1630765410640014, "grad_norm": 0.4976618606275063, "learning_rate": 7.68373074524131e-05, "loss": 2.9337, "step": 18736 }, { "epoch": 1.1631386181637593, "grad_norm": 0.4784502412929558, "learning_rate": 7.683426018133317e-05, "loss": 2.7817, "step": 18737 }, { "epoch": 1.1632006952635172, "grad_norm": 0.5010221583268916, "learning_rate": 7.683121277025238e-05, "loss": 2.8784, "step": 18738 }, { "epoch": 1.1632627723632751, "grad_norm": 0.5437188764412274, "learning_rate": 7.682816521918666e-05, "loss": 2.8847, "step": 18739 }, { "epoch": 1.163324849463033, "grad_norm": 0.5439222457624547, "learning_rate": 7.682511752815188e-05, "loss": 2.8983, "step": 18740 }, { "epoch": 1.163386926562791, "grad_norm": 0.5248724214763494, "learning_rate": 7.682206969716395e-05, "loss": 2.9305, "step": 18741 }, { "epoch": 1.1634490036625489, "grad_norm": 0.588335394932248, "learning_rate": 7.681902172623877e-05, "loss": 2.863, "step": 18742 }, { "epoch": 1.1635110807623068, "grad_norm": 0.5157169202309686, "learning_rate": 7.681597361539226e-05, "loss": 2.8789, "step": 18743 }, { "epoch": 1.1635731578620647, "grad_norm": 0.5475915574026249, "learning_rate": 7.681292536464028e-05, "loss": 2.9047, "step": 18744 }, { "epoch": 1.1636352349618226, "grad_norm": 0.6195227469670117, "learning_rate": 7.680987697399877e-05, "loss": 2.8986, "step": 18745 }, { "epoch": 1.1636973120615806, "grad_norm": 0.4917331596714163, "learning_rate": 7.680682844348362e-05, "loss": 2.8669, "step": 18746 }, { "epoch": 1.1637593891613385, "grad_norm": 0.512153731810405, "learning_rate": 7.680377977311074e-05, "loss": 2.7833, "step": 18747 }, { "epoch": 1.1638214662610964, "grad_norm": 0.5121676390767241, "learning_rate": 7.680073096289604e-05, "loss": 2.8961, "step": 18748 }, { "epoch": 1.163883543360854, "grad_norm": 0.529423547205364, "learning_rate": 7.679768201285541e-05, "loss": 2.831, "step": 18749 }, { "epoch": 1.163945620460612, "grad_norm": 0.5324260652731264, "learning_rate": 7.679463292300479e-05, "loss": 2.8899, "step": 18750 }, { "epoch": 1.16400769756037, "grad_norm": 0.5268754363143038, "learning_rate": 7.679158369336003e-05, "loss": 2.7652, "step": 18751 }, { "epoch": 1.1640697746601278, "grad_norm": 0.5734287458785826, "learning_rate": 7.678853432393712e-05, "loss": 2.8586, "step": 18752 }, { "epoch": 1.1641318517598858, "grad_norm": 0.6092910708803843, "learning_rate": 7.678548481475189e-05, "loss": 2.8875, "step": 18753 }, { "epoch": 1.1641939288596437, "grad_norm": 0.5298780228046461, "learning_rate": 7.678243516582029e-05, "loss": 2.8078, "step": 18754 }, { "epoch": 1.1642560059594016, "grad_norm": 0.6029920065715352, "learning_rate": 7.677938537715824e-05, "loss": 2.7763, "step": 18755 }, { "epoch": 1.1643180830591595, "grad_norm": 0.6021248059580314, "learning_rate": 7.67763354487816e-05, "loss": 2.9186, "step": 18756 }, { "epoch": 1.1643801601589174, "grad_norm": 0.5675709292749195, "learning_rate": 7.677328538070634e-05, "loss": 2.9225, "step": 18757 }, { "epoch": 1.1644422372586754, "grad_norm": 0.5495978554153523, "learning_rate": 7.677023517294835e-05, "loss": 2.9057, "step": 18758 }, { "epoch": 1.1645043143584333, "grad_norm": 0.5299008010302448, "learning_rate": 7.676718482552354e-05, "loss": 2.9531, "step": 18759 }, { "epoch": 1.164566391458191, "grad_norm": 0.5587052187373237, "learning_rate": 7.676413433844782e-05, "loss": 2.911, "step": 18760 }, { "epoch": 1.1646284685579489, "grad_norm": 0.49525565594385407, "learning_rate": 7.676108371173713e-05, "loss": 2.8591, "step": 18761 }, { "epoch": 1.1646905456577068, "grad_norm": 0.5003614311427366, "learning_rate": 7.675803294540737e-05, "loss": 2.9094, "step": 18762 }, { "epoch": 1.1647526227574647, "grad_norm": 0.5129097045231734, "learning_rate": 7.675498203947444e-05, "loss": 2.8761, "step": 18763 }, { "epoch": 1.1648146998572226, "grad_norm": 0.536768607498885, "learning_rate": 7.67519309939543e-05, "loss": 2.8261, "step": 18764 }, { "epoch": 1.1648767769569806, "grad_norm": 0.45874736429127744, "learning_rate": 7.674887980886283e-05, "loss": 2.9092, "step": 18765 }, { "epoch": 1.1649388540567385, "grad_norm": 0.5408899870599662, "learning_rate": 7.674582848421596e-05, "loss": 2.8919, "step": 18766 }, { "epoch": 1.1650009311564964, "grad_norm": 0.4724065735958359, "learning_rate": 7.67427770200296e-05, "loss": 2.7975, "step": 18767 }, { "epoch": 1.1650630082562543, "grad_norm": 0.5196741810994103, "learning_rate": 7.673972541631969e-05, "loss": 2.8315, "step": 18768 }, { "epoch": 1.1651250853560122, "grad_norm": 0.49436983062414225, "learning_rate": 7.673667367310213e-05, "loss": 2.8923, "step": 18769 }, { "epoch": 1.1651871624557701, "grad_norm": 0.5381549714765872, "learning_rate": 7.673362179039287e-05, "loss": 2.9065, "step": 18770 }, { "epoch": 1.165249239555528, "grad_norm": 0.5339413238338456, "learning_rate": 7.673056976820781e-05, "loss": 2.925, "step": 18771 }, { "epoch": 1.165311316655286, "grad_norm": 0.48815214357601927, "learning_rate": 7.672751760656288e-05, "loss": 2.8652, "step": 18772 }, { "epoch": 1.1653733937550437, "grad_norm": 0.5015882360682794, "learning_rate": 7.672446530547401e-05, "loss": 2.902, "step": 18773 }, { "epoch": 1.1654354708548016, "grad_norm": 0.4414163605198811, "learning_rate": 7.672141286495709e-05, "loss": 2.8263, "step": 18774 }, { "epoch": 1.1654975479545595, "grad_norm": 0.5333447949350132, "learning_rate": 7.67183602850281e-05, "loss": 2.8731, "step": 18775 }, { "epoch": 1.1655596250543174, "grad_norm": 0.4415160186125383, "learning_rate": 7.671530756570291e-05, "loss": 2.8579, "step": 18776 }, { "epoch": 1.1656217021540753, "grad_norm": 0.5491480861910764, "learning_rate": 7.67122547069975e-05, "loss": 2.8509, "step": 18777 }, { "epoch": 1.1656837792538333, "grad_norm": 0.5006843294570187, "learning_rate": 7.670920170892776e-05, "loss": 2.823, "step": 18778 }, { "epoch": 1.1657458563535912, "grad_norm": 0.5182355052873449, "learning_rate": 7.670614857150964e-05, "loss": 2.8599, "step": 18779 }, { "epoch": 1.165807933453349, "grad_norm": 0.5974788776558706, "learning_rate": 7.670309529475905e-05, "loss": 2.9086, "step": 18780 }, { "epoch": 1.165870010553107, "grad_norm": 0.536794840875868, "learning_rate": 7.670004187869193e-05, "loss": 2.8708, "step": 18781 }, { "epoch": 1.165932087652865, "grad_norm": 0.5313663355287613, "learning_rate": 7.669698832332422e-05, "loss": 2.8318, "step": 18782 }, { "epoch": 1.1659941647526229, "grad_norm": 0.48696929153248747, "learning_rate": 7.669393462867183e-05, "loss": 2.8242, "step": 18783 }, { "epoch": 1.1660562418523805, "grad_norm": 0.5259793468108427, "learning_rate": 7.66908807947507e-05, "loss": 2.8207, "step": 18784 }, { "epoch": 1.1661183189521385, "grad_norm": 0.5348527423571873, "learning_rate": 7.668782682157677e-05, "loss": 2.8484, "step": 18785 }, { "epoch": 1.1661803960518964, "grad_norm": 0.5623266694786401, "learning_rate": 7.668477270916598e-05, "loss": 2.8461, "step": 18786 }, { "epoch": 1.1662424731516543, "grad_norm": 0.5874662927332364, "learning_rate": 7.668171845753425e-05, "loss": 2.8154, "step": 18787 }, { "epoch": 1.1663045502514122, "grad_norm": 0.5209739844824817, "learning_rate": 7.66786640666975e-05, "loss": 2.8523, "step": 18788 }, { "epoch": 1.1663666273511701, "grad_norm": 0.5860131278504186, "learning_rate": 7.66756095366717e-05, "loss": 2.8323, "step": 18789 }, { "epoch": 1.166428704450928, "grad_norm": 0.5854607231785106, "learning_rate": 7.667255486747277e-05, "loss": 2.7878, "step": 18790 }, { "epoch": 1.166490781550686, "grad_norm": 0.5363395306290617, "learning_rate": 7.666950005911664e-05, "loss": 2.8883, "step": 18791 }, { "epoch": 1.166552858650444, "grad_norm": 0.5444459374914435, "learning_rate": 7.666644511161925e-05, "loss": 2.8308, "step": 18792 }, { "epoch": 1.1666149357502018, "grad_norm": 0.4769416770217259, "learning_rate": 7.666339002499656e-05, "loss": 2.8871, "step": 18793 }, { "epoch": 1.1666770128499597, "grad_norm": 0.5777051535251881, "learning_rate": 7.666033479926447e-05, "loss": 2.8757, "step": 18794 }, { "epoch": 1.1667390899497176, "grad_norm": 0.5177820919411513, "learning_rate": 7.665727943443895e-05, "loss": 2.8747, "step": 18795 }, { "epoch": 1.1668011670494756, "grad_norm": 0.5348428518570255, "learning_rate": 7.665422393053593e-05, "loss": 2.7859, "step": 18796 }, { "epoch": 1.1668632441492333, "grad_norm": 0.5534194098111035, "learning_rate": 7.665116828757135e-05, "loss": 2.8867, "step": 18797 }, { "epoch": 1.1669253212489912, "grad_norm": 0.509404109052883, "learning_rate": 7.664811250556117e-05, "loss": 2.8359, "step": 18798 }, { "epoch": 1.166987398348749, "grad_norm": 0.5636380392323056, "learning_rate": 7.66450565845213e-05, "loss": 2.8606, "step": 18799 }, { "epoch": 1.167049475448507, "grad_norm": 0.48304917790436036, "learning_rate": 7.66420005244677e-05, "loss": 2.8696, "step": 18800 }, { "epoch": 1.167111552548265, "grad_norm": 0.5468227906509614, "learning_rate": 7.663894432541635e-05, "loss": 2.9122, "step": 18801 }, { "epoch": 1.1671736296480228, "grad_norm": 0.5813493766739987, "learning_rate": 7.663588798738313e-05, "loss": 2.9598, "step": 18802 }, { "epoch": 1.1672357067477808, "grad_norm": 0.48846354324214186, "learning_rate": 7.6632831510384e-05, "loss": 2.8908, "step": 18803 }, { "epoch": 1.1672977838475387, "grad_norm": 0.4914259544267035, "learning_rate": 7.662977489443496e-05, "loss": 2.8498, "step": 18804 }, { "epoch": 1.1673598609472966, "grad_norm": 0.4877712179991747, "learning_rate": 7.66267181395519e-05, "loss": 2.8941, "step": 18805 }, { "epoch": 1.1674219380470545, "grad_norm": 0.5230195383758509, "learning_rate": 7.66236612457508e-05, "loss": 2.8254, "step": 18806 }, { "epoch": 1.1674840151468124, "grad_norm": 0.48216137049467556, "learning_rate": 7.662060421304759e-05, "loss": 2.8365, "step": 18807 }, { "epoch": 1.1675460922465701, "grad_norm": 0.5578313573219686, "learning_rate": 7.661754704145821e-05, "loss": 2.9515, "step": 18808 }, { "epoch": 1.167608169346328, "grad_norm": 0.4817784764138944, "learning_rate": 7.661448973099866e-05, "loss": 2.8398, "step": 18809 }, { "epoch": 1.167670246446086, "grad_norm": 0.6200875583088296, "learning_rate": 7.661143228168482e-05, "loss": 2.8278, "step": 18810 }, { "epoch": 1.1677323235458439, "grad_norm": 0.6015057475084172, "learning_rate": 7.66083746935327e-05, "loss": 2.8413, "step": 18811 }, { "epoch": 1.1677944006456018, "grad_norm": 0.5545291405941594, "learning_rate": 7.660531696655821e-05, "loss": 2.7967, "step": 18812 }, { "epoch": 1.1678564777453597, "grad_norm": 0.5830140744484821, "learning_rate": 7.660225910077734e-05, "loss": 2.9102, "step": 18813 }, { "epoch": 1.1679185548451176, "grad_norm": 0.5396160379707848, "learning_rate": 7.659920109620602e-05, "loss": 2.8098, "step": 18814 }, { "epoch": 1.1679806319448756, "grad_norm": 0.5175218659540581, "learning_rate": 7.659614295286021e-05, "loss": 2.858, "step": 18815 }, { "epoch": 1.1680427090446335, "grad_norm": 0.5328393051251119, "learning_rate": 7.659308467075588e-05, "loss": 2.8675, "step": 18816 }, { "epoch": 1.1681047861443914, "grad_norm": 0.5371411955684179, "learning_rate": 7.659002624990895e-05, "loss": 2.8233, "step": 18817 }, { "epoch": 1.1681668632441493, "grad_norm": 0.4989238721803917, "learning_rate": 7.658696769033542e-05, "loss": 2.8473, "step": 18818 }, { "epoch": 1.1682289403439072, "grad_norm": 0.5045155825859614, "learning_rate": 7.658390899205121e-05, "loss": 2.8875, "step": 18819 }, { "epoch": 1.1682910174436651, "grad_norm": 0.5361020438948885, "learning_rate": 7.65808501550723e-05, "loss": 2.8455, "step": 18820 }, { "epoch": 1.1683530945434228, "grad_norm": 0.4998605801991507, "learning_rate": 7.657779117941464e-05, "loss": 2.9399, "step": 18821 }, { "epoch": 1.1684151716431808, "grad_norm": 0.5158735023238319, "learning_rate": 7.657473206509418e-05, "loss": 2.8494, "step": 18822 }, { "epoch": 1.1684772487429387, "grad_norm": 0.48583200726347203, "learning_rate": 7.657167281212689e-05, "loss": 2.8811, "step": 18823 }, { "epoch": 1.1685393258426966, "grad_norm": 0.5598163696203021, "learning_rate": 7.656861342052877e-05, "loss": 2.926, "step": 18824 }, { "epoch": 1.1686014029424545, "grad_norm": 0.5058752226832574, "learning_rate": 7.656555389031572e-05, "loss": 2.9541, "step": 18825 }, { "epoch": 1.1686634800422124, "grad_norm": 0.504753406106945, "learning_rate": 7.65624942215037e-05, "loss": 2.8446, "step": 18826 }, { "epoch": 1.1687255571419704, "grad_norm": 0.4735987228238196, "learning_rate": 7.655943441410873e-05, "loss": 2.8062, "step": 18827 }, { "epoch": 1.1687876342417283, "grad_norm": 0.47788241783629315, "learning_rate": 7.655637446814673e-05, "loss": 2.8136, "step": 18828 }, { "epoch": 1.1688497113414862, "grad_norm": 0.4741250223611856, "learning_rate": 7.655331438363369e-05, "loss": 2.8688, "step": 18829 }, { "epoch": 1.168911788441244, "grad_norm": 0.4859002003966282, "learning_rate": 7.655025416058555e-05, "loss": 2.8798, "step": 18830 }, { "epoch": 1.1689738655410018, "grad_norm": 0.4506570926037016, "learning_rate": 7.654719379901831e-05, "loss": 2.9067, "step": 18831 }, { "epoch": 1.1690359426407597, "grad_norm": 0.4659330923280044, "learning_rate": 7.654413329894791e-05, "loss": 2.823, "step": 18832 }, { "epoch": 1.1690980197405176, "grad_norm": 0.46330368347918566, "learning_rate": 7.65410726603903e-05, "loss": 2.834, "step": 18833 }, { "epoch": 1.1691600968402756, "grad_norm": 0.5308011908009732, "learning_rate": 7.653801188336149e-05, "loss": 2.9391, "step": 18834 }, { "epoch": 1.1692221739400335, "grad_norm": 0.521739911395418, "learning_rate": 7.653495096787741e-05, "loss": 2.9115, "step": 18835 }, { "epoch": 1.1692842510397914, "grad_norm": 0.4977220261250948, "learning_rate": 7.653188991395408e-05, "loss": 2.8139, "step": 18836 }, { "epoch": 1.1693463281395493, "grad_norm": 0.5605238002586804, "learning_rate": 7.652882872160742e-05, "loss": 2.8975, "step": 18837 }, { "epoch": 1.1694084052393072, "grad_norm": 0.6036942103576058, "learning_rate": 7.652576739085343e-05, "loss": 2.9846, "step": 18838 }, { "epoch": 1.1694704823390651, "grad_norm": 0.5032002434235667, "learning_rate": 7.652270592170806e-05, "loss": 2.8701, "step": 18839 }, { "epoch": 1.169532559438823, "grad_norm": 0.6043742223456507, "learning_rate": 7.65196443141873e-05, "loss": 2.9029, "step": 18840 }, { "epoch": 1.169594636538581, "grad_norm": 0.5832608439046102, "learning_rate": 7.651658256830711e-05, "loss": 2.9221, "step": 18841 }, { "epoch": 1.169656713638339, "grad_norm": 0.5294783727967758, "learning_rate": 7.65135206840835e-05, "loss": 2.8008, "step": 18842 }, { "epoch": 1.1697187907380968, "grad_norm": 0.5984181801887003, "learning_rate": 7.651045866153239e-05, "loss": 2.9265, "step": 18843 }, { "epoch": 1.1697808678378547, "grad_norm": 0.5514157475523456, "learning_rate": 7.650739650066979e-05, "loss": 2.8385, "step": 18844 }, { "epoch": 1.1698429449376124, "grad_norm": 0.5385097765995822, "learning_rate": 7.650433420151166e-05, "loss": 2.8876, "step": 18845 }, { "epoch": 1.1699050220373703, "grad_norm": 0.5203837806974307, "learning_rate": 7.650127176407399e-05, "loss": 2.7702, "step": 18846 }, { "epoch": 1.1699670991371283, "grad_norm": 0.5708340389590764, "learning_rate": 7.649820918837276e-05, "loss": 2.8943, "step": 18847 }, { "epoch": 1.1700291762368862, "grad_norm": 0.5097298737482044, "learning_rate": 7.649514647442393e-05, "loss": 2.8531, "step": 18848 }, { "epoch": 1.170091253336644, "grad_norm": 0.5370065486622317, "learning_rate": 7.649208362224349e-05, "loss": 2.8704, "step": 18849 }, { "epoch": 1.170153330436402, "grad_norm": 0.5289242266666669, "learning_rate": 7.648902063184742e-05, "loss": 2.8256, "step": 18850 }, { "epoch": 1.17021540753616, "grad_norm": 0.6155781103566438, "learning_rate": 7.648595750325169e-05, "loss": 2.914, "step": 18851 }, { "epoch": 1.1702774846359179, "grad_norm": 0.5494519418296621, "learning_rate": 7.64828942364723e-05, "loss": 2.8428, "step": 18852 }, { "epoch": 1.1703395617356758, "grad_norm": 0.512729810734803, "learning_rate": 7.647983083152523e-05, "loss": 2.849, "step": 18853 }, { "epoch": 1.1704016388354337, "grad_norm": 0.5981721832734556, "learning_rate": 7.647676728842643e-05, "loss": 2.8183, "step": 18854 }, { "epoch": 1.1704637159351914, "grad_norm": 0.5443592790595724, "learning_rate": 7.647370360719193e-05, "loss": 2.8757, "step": 18855 }, { "epoch": 1.1705257930349493, "grad_norm": 0.5516145233172514, "learning_rate": 7.647063978783769e-05, "loss": 2.861, "step": 18856 }, { "epoch": 1.1705878701347072, "grad_norm": 0.5291856599638111, "learning_rate": 7.646757583037967e-05, "loss": 2.836, "step": 18857 }, { "epoch": 1.1706499472344651, "grad_norm": 0.5305784412622035, "learning_rate": 7.646451173483392e-05, "loss": 2.8154, "step": 18858 }, { "epoch": 1.170712024334223, "grad_norm": 0.5313102463612844, "learning_rate": 7.646144750121635e-05, "loss": 2.8475, "step": 18859 }, { "epoch": 1.170774101433981, "grad_norm": 0.582786147311869, "learning_rate": 7.645838312954302e-05, "loss": 2.8552, "step": 18860 }, { "epoch": 1.170836178533739, "grad_norm": 0.5413307071183224, "learning_rate": 7.645531861982986e-05, "loss": 2.7919, "step": 18861 }, { "epoch": 1.1708982556334968, "grad_norm": 0.5617728831982775, "learning_rate": 7.645225397209287e-05, "loss": 2.9257, "step": 18862 }, { "epoch": 1.1709603327332547, "grad_norm": 0.6257031062146757, "learning_rate": 7.644918918634809e-05, "loss": 2.968, "step": 18863 }, { "epoch": 1.1710224098330126, "grad_norm": 0.5184217666988719, "learning_rate": 7.644612426261144e-05, "loss": 2.8289, "step": 18864 }, { "epoch": 1.1710844869327706, "grad_norm": 0.5499391543804282, "learning_rate": 7.644305920089893e-05, "loss": 2.966, "step": 18865 }, { "epoch": 1.1711465640325285, "grad_norm": 0.5543594464826418, "learning_rate": 7.643999400122659e-05, "loss": 2.8908, "step": 18866 }, { "epoch": 1.1712086411322864, "grad_norm": 0.47594785920739774, "learning_rate": 7.643692866361036e-05, "loss": 2.7444, "step": 18867 }, { "epoch": 1.1712707182320443, "grad_norm": 0.6054298388375547, "learning_rate": 7.643386318806628e-05, "loss": 2.8405, "step": 18868 }, { "epoch": 1.171332795331802, "grad_norm": 0.5148959641324921, "learning_rate": 7.643079757461029e-05, "loss": 2.8658, "step": 18869 }, { "epoch": 1.17139487243156, "grad_norm": 0.5694320885421336, "learning_rate": 7.642773182325844e-05, "loss": 2.86, "step": 18870 }, { "epoch": 1.1714569495313178, "grad_norm": 0.506291503146905, "learning_rate": 7.642466593402667e-05, "loss": 2.8317, "step": 18871 }, { "epoch": 1.1715190266310758, "grad_norm": 0.5900902643499267, "learning_rate": 7.642159990693104e-05, "loss": 2.7757, "step": 18872 }, { "epoch": 1.1715811037308337, "grad_norm": 0.499864425756244, "learning_rate": 7.641853374198747e-05, "loss": 2.861, "step": 18873 }, { "epoch": 1.1716431808305916, "grad_norm": 0.5280633608268903, "learning_rate": 7.641546743921202e-05, "loss": 2.8011, "step": 18874 }, { "epoch": 1.1717052579303495, "grad_norm": 0.5172129944571041, "learning_rate": 7.641240099862063e-05, "loss": 2.9526, "step": 18875 }, { "epoch": 1.1717673350301074, "grad_norm": 0.5630442582396867, "learning_rate": 7.640933442022936e-05, "loss": 2.8148, "step": 18876 }, { "epoch": 1.1718294121298654, "grad_norm": 0.4896726239396876, "learning_rate": 7.64062677040542e-05, "loss": 2.7476, "step": 18877 }, { "epoch": 1.1718914892296233, "grad_norm": 0.51498972164934, "learning_rate": 7.640320085011109e-05, "loss": 2.8742, "step": 18878 }, { "epoch": 1.171953566329381, "grad_norm": 0.5256644460729569, "learning_rate": 7.640013385841609e-05, "loss": 2.8399, "step": 18879 }, { "epoch": 1.1720156434291389, "grad_norm": 0.5592857023173712, "learning_rate": 7.639706672898519e-05, "loss": 2.8094, "step": 18880 }, { "epoch": 1.1720777205288968, "grad_norm": 0.5161080842852704, "learning_rate": 7.639399946183439e-05, "loss": 2.8217, "step": 18881 }, { "epoch": 1.1721397976286547, "grad_norm": 0.5668880912421536, "learning_rate": 7.639093205697967e-05, "loss": 2.8767, "step": 18882 }, { "epoch": 1.1722018747284126, "grad_norm": 0.5645970091858449, "learning_rate": 7.638786451443705e-05, "loss": 2.8208, "step": 18883 }, { "epoch": 1.1722639518281706, "grad_norm": 0.6160820102177794, "learning_rate": 7.638479683422254e-05, "loss": 2.8614, "step": 18884 }, { "epoch": 1.1723260289279285, "grad_norm": 0.6378287019592204, "learning_rate": 7.638172901635214e-05, "loss": 2.9315, "step": 18885 }, { "epoch": 1.1723881060276864, "grad_norm": 0.5369763851247896, "learning_rate": 7.637866106084186e-05, "loss": 2.9032, "step": 18886 }, { "epoch": 1.1724501831274443, "grad_norm": 0.5090155509248198, "learning_rate": 7.63755929677077e-05, "loss": 2.8222, "step": 18887 }, { "epoch": 1.1725122602272022, "grad_norm": 0.5682376062574723, "learning_rate": 7.637252473696568e-05, "loss": 2.8989, "step": 18888 }, { "epoch": 1.1725743373269601, "grad_norm": 0.47425609359888443, "learning_rate": 7.636945636863178e-05, "loss": 2.9081, "step": 18889 }, { "epoch": 1.172636414426718, "grad_norm": 0.5668017354976329, "learning_rate": 7.636638786272204e-05, "loss": 2.8451, "step": 18890 }, { "epoch": 1.172698491526476, "grad_norm": 0.4752577289522719, "learning_rate": 7.636331921925242e-05, "loss": 2.8225, "step": 18891 }, { "epoch": 1.172760568626234, "grad_norm": 0.5426679367972997, "learning_rate": 7.636025043823897e-05, "loss": 2.7514, "step": 18892 }, { "epoch": 1.1728226457259916, "grad_norm": 0.5603879111711009, "learning_rate": 7.635718151969772e-05, "loss": 2.8652, "step": 18893 }, { "epoch": 1.1728847228257495, "grad_norm": 0.46519586706420735, "learning_rate": 7.635411246364466e-05, "loss": 2.8396, "step": 18894 }, { "epoch": 1.1729467999255074, "grad_norm": 0.6045983487829412, "learning_rate": 7.635104327009577e-05, "loss": 2.941, "step": 18895 }, { "epoch": 1.1730088770252654, "grad_norm": 0.501590077958163, "learning_rate": 7.634797393906711e-05, "loss": 2.8961, "step": 18896 }, { "epoch": 1.1730709541250233, "grad_norm": 0.5620687474035405, "learning_rate": 7.634490447057468e-05, "loss": 2.9107, "step": 18897 }, { "epoch": 1.1731330312247812, "grad_norm": 0.5207219068226294, "learning_rate": 7.634183486463446e-05, "loss": 2.7527, "step": 18898 }, { "epoch": 1.173195108324539, "grad_norm": 0.5347621329662366, "learning_rate": 7.633876512126252e-05, "loss": 2.8671, "step": 18899 }, { "epoch": 1.173257185424297, "grad_norm": 0.5219591815233388, "learning_rate": 7.633569524047483e-05, "loss": 2.9106, "step": 18900 }, { "epoch": 1.173319262524055, "grad_norm": 0.508382213566676, "learning_rate": 7.633262522228744e-05, "loss": 2.7145, "step": 18901 }, { "epoch": 1.1733813396238129, "grad_norm": 0.5770485367136575, "learning_rate": 7.632955506671633e-05, "loss": 2.8296, "step": 18902 }, { "epoch": 1.1734434167235706, "grad_norm": 0.5661245500575842, "learning_rate": 7.632648477377754e-05, "loss": 2.8354, "step": 18903 }, { "epoch": 1.1735054938233285, "grad_norm": 0.5484588032032923, "learning_rate": 7.632341434348711e-05, "loss": 2.9027, "step": 18904 }, { "epoch": 1.1735675709230864, "grad_norm": 0.5901232934889534, "learning_rate": 7.632034377586104e-05, "loss": 2.7873, "step": 18905 }, { "epoch": 1.1736296480228443, "grad_norm": 0.5108366490495533, "learning_rate": 7.631727307091533e-05, "loss": 2.7863, "step": 18906 }, { "epoch": 1.1736917251226022, "grad_norm": 0.5617554293184663, "learning_rate": 7.631420222866602e-05, "loss": 2.9081, "step": 18907 }, { "epoch": 1.1737538022223601, "grad_norm": 0.5613378865050924, "learning_rate": 7.631113124912911e-05, "loss": 2.8411, "step": 18908 }, { "epoch": 1.173815879322118, "grad_norm": 0.5408654124920178, "learning_rate": 7.630806013232067e-05, "loss": 2.9668, "step": 18909 }, { "epoch": 1.173877956421876, "grad_norm": 0.5837078141319915, "learning_rate": 7.63049888782567e-05, "loss": 2.8646, "step": 18910 }, { "epoch": 1.173940033521634, "grad_norm": 0.5193539367058694, "learning_rate": 7.63019174869532e-05, "loss": 2.9364, "step": 18911 }, { "epoch": 1.1740021106213918, "grad_norm": 0.5270434619564269, "learning_rate": 7.629884595842621e-05, "loss": 2.8903, "step": 18912 }, { "epoch": 1.1740641877211497, "grad_norm": 0.5490967714125504, "learning_rate": 7.629577429269177e-05, "loss": 2.8418, "step": 18913 }, { "epoch": 1.1741262648209077, "grad_norm": 0.5970189055879045, "learning_rate": 7.629270248976588e-05, "loss": 2.805, "step": 18914 }, { "epoch": 1.1741883419206656, "grad_norm": 0.5699730476900453, "learning_rate": 7.628963054966458e-05, "loss": 2.845, "step": 18915 }, { "epoch": 1.1742504190204235, "grad_norm": 0.5907571301685799, "learning_rate": 7.62865584724039e-05, "loss": 2.8736, "step": 18916 }, { "epoch": 1.1743124961201812, "grad_norm": 0.6073051168698257, "learning_rate": 7.628348625799986e-05, "loss": 2.7848, "step": 18917 }, { "epoch": 1.174374573219939, "grad_norm": 0.6311491781866886, "learning_rate": 7.62804139064685e-05, "loss": 2.9856, "step": 18918 }, { "epoch": 1.174436650319697, "grad_norm": 0.5729733150103712, "learning_rate": 7.627734141782583e-05, "loss": 2.75, "step": 18919 }, { "epoch": 1.174498727419455, "grad_norm": 0.6528005085926202, "learning_rate": 7.627426879208788e-05, "loss": 2.8058, "step": 18920 }, { "epoch": 1.1745608045192129, "grad_norm": 0.49557194338001515, "learning_rate": 7.627119602927071e-05, "loss": 2.9126, "step": 18921 }, { "epoch": 1.1746228816189708, "grad_norm": 0.6772871713206531, "learning_rate": 7.626812312939033e-05, "loss": 2.9224, "step": 18922 }, { "epoch": 1.1746849587187287, "grad_norm": 0.4888100011878996, "learning_rate": 7.626505009246277e-05, "loss": 2.8436, "step": 18923 }, { "epoch": 1.1747470358184866, "grad_norm": 0.6096935906587476, "learning_rate": 7.626197691850408e-05, "loss": 2.919, "step": 18924 }, { "epoch": 1.1748091129182445, "grad_norm": 0.5297830748627028, "learning_rate": 7.625890360753025e-05, "loss": 2.8055, "step": 18925 }, { "epoch": 1.1748711900180024, "grad_norm": 0.5544576034160781, "learning_rate": 7.625583015955737e-05, "loss": 2.884, "step": 18926 }, { "epoch": 1.1749332671177601, "grad_norm": 0.530213185745948, "learning_rate": 7.625275657460144e-05, "loss": 2.8616, "step": 18927 }, { "epoch": 1.174995344217518, "grad_norm": 0.5932368018172807, "learning_rate": 7.624968285267853e-05, "loss": 2.8266, "step": 18928 }, { "epoch": 1.175057421317276, "grad_norm": 0.5826053732001215, "learning_rate": 7.624660899380463e-05, "loss": 2.8844, "step": 18929 }, { "epoch": 1.175119498417034, "grad_norm": 0.5487258924519837, "learning_rate": 7.62435349979958e-05, "loss": 2.8666, "step": 18930 }, { "epoch": 1.1751815755167918, "grad_norm": 0.643974595565429, "learning_rate": 7.624046086526808e-05, "loss": 2.8488, "step": 18931 }, { "epoch": 1.1752436526165497, "grad_norm": 0.5295833924196192, "learning_rate": 7.623738659563751e-05, "loss": 2.8177, "step": 18932 }, { "epoch": 1.1753057297163076, "grad_norm": 0.6202099347779965, "learning_rate": 7.623431218912012e-05, "loss": 2.8869, "step": 18933 }, { "epoch": 1.1753678068160656, "grad_norm": 0.4983859385305867, "learning_rate": 7.623123764573196e-05, "loss": 2.9032, "step": 18934 }, { "epoch": 1.1754298839158235, "grad_norm": 0.5590578831950612, "learning_rate": 7.622816296548907e-05, "loss": 2.7673, "step": 18935 }, { "epoch": 1.1754919610155814, "grad_norm": 0.5154163342965405, "learning_rate": 7.622508814840747e-05, "loss": 2.8855, "step": 18936 }, { "epoch": 1.1755540381153393, "grad_norm": 0.5254120946418517, "learning_rate": 7.622201319450323e-05, "loss": 2.8508, "step": 18937 }, { "epoch": 1.1756161152150972, "grad_norm": 0.5718941117308561, "learning_rate": 7.621893810379237e-05, "loss": 2.8569, "step": 18938 }, { "epoch": 1.1756781923148552, "grad_norm": 0.485262595912845, "learning_rate": 7.621586287629097e-05, "loss": 2.738, "step": 18939 }, { "epoch": 1.175740269414613, "grad_norm": 0.5969254257711237, "learning_rate": 7.621278751201501e-05, "loss": 2.8287, "step": 18940 }, { "epoch": 1.1758023465143708, "grad_norm": 0.46853658268286136, "learning_rate": 7.62097120109806e-05, "loss": 2.783, "step": 18941 }, { "epoch": 1.1758644236141287, "grad_norm": 0.5910856801711625, "learning_rate": 7.620663637320376e-05, "loss": 2.8396, "step": 18942 }, { "epoch": 1.1759265007138866, "grad_norm": 0.49568599554094994, "learning_rate": 7.620356059870053e-05, "loss": 2.8834, "step": 18943 }, { "epoch": 1.1759885778136445, "grad_norm": 0.5668852260722905, "learning_rate": 7.620048468748697e-05, "loss": 2.7987, "step": 18944 }, { "epoch": 1.1760506549134024, "grad_norm": 0.5565217811472536, "learning_rate": 7.619740863957911e-05, "loss": 2.8625, "step": 18945 }, { "epoch": 1.1761127320131604, "grad_norm": 0.5678408115280072, "learning_rate": 7.619433245499303e-05, "loss": 2.8561, "step": 18946 }, { "epoch": 1.1761748091129183, "grad_norm": 0.6455930575642158, "learning_rate": 7.619125613374474e-05, "loss": 2.8188, "step": 18947 }, { "epoch": 1.1762368862126762, "grad_norm": 0.6011334164657499, "learning_rate": 7.618817967585031e-05, "loss": 2.8277, "step": 18948 }, { "epoch": 1.176298963312434, "grad_norm": 0.6033443220158263, "learning_rate": 7.61851030813258e-05, "loss": 2.849, "step": 18949 }, { "epoch": 1.176361040412192, "grad_norm": 0.6210035102594529, "learning_rate": 7.618202635018725e-05, "loss": 2.8753, "step": 18950 }, { "epoch": 1.1764231175119497, "grad_norm": 0.5213386532066137, "learning_rate": 7.61789494824507e-05, "loss": 2.8409, "step": 18951 }, { "epoch": 1.1764851946117076, "grad_norm": 0.5364165397832741, "learning_rate": 7.617587247813223e-05, "loss": 2.8388, "step": 18952 }, { "epoch": 1.1765472717114656, "grad_norm": 0.5412816519435405, "learning_rate": 7.617279533724786e-05, "loss": 2.7895, "step": 18953 }, { "epoch": 1.1766093488112235, "grad_norm": 0.5351320525775929, "learning_rate": 7.616971805981367e-05, "loss": 2.8126, "step": 18954 }, { "epoch": 1.1766714259109814, "grad_norm": 0.5697982246072679, "learning_rate": 7.616664064584572e-05, "loss": 2.9, "step": 18955 }, { "epoch": 1.1767335030107393, "grad_norm": 0.5518881044743633, "learning_rate": 7.616356309536004e-05, "loss": 2.8925, "step": 18956 }, { "epoch": 1.1767955801104972, "grad_norm": 0.5418450783076847, "learning_rate": 7.61604854083727e-05, "loss": 2.8724, "step": 18957 }, { "epoch": 1.1768576572102551, "grad_norm": 0.5648444907789003, "learning_rate": 7.615740758489977e-05, "loss": 2.866, "step": 18958 }, { "epoch": 1.176919734310013, "grad_norm": 0.6011338130807008, "learning_rate": 7.615432962495728e-05, "loss": 2.8928, "step": 18959 }, { "epoch": 1.176981811409771, "grad_norm": 0.5238592526875497, "learning_rate": 7.615125152856131e-05, "loss": 2.7706, "step": 18960 }, { "epoch": 1.177043888509529, "grad_norm": 0.5142189274859752, "learning_rate": 7.614817329572792e-05, "loss": 2.8283, "step": 18961 }, { "epoch": 1.1771059656092868, "grad_norm": 0.5383471289462399, "learning_rate": 7.614509492647315e-05, "loss": 2.8143, "step": 18962 }, { "epoch": 1.1771680427090447, "grad_norm": 0.607793737087575, "learning_rate": 7.614201642081306e-05, "loss": 2.8726, "step": 18963 }, { "epoch": 1.1772301198088027, "grad_norm": 0.4702744807594503, "learning_rate": 7.613893777876375e-05, "loss": 2.8282, "step": 18964 }, { "epoch": 1.1772921969085604, "grad_norm": 0.6260154819143215, "learning_rate": 7.613585900034126e-05, "loss": 2.9306, "step": 18965 }, { "epoch": 1.1773542740083183, "grad_norm": 0.5196202997743413, "learning_rate": 7.613278008556164e-05, "loss": 2.8618, "step": 18966 }, { "epoch": 1.1774163511080762, "grad_norm": 0.5410504381567754, "learning_rate": 7.612970103444096e-05, "loss": 2.8696, "step": 18967 }, { "epoch": 1.177478428207834, "grad_norm": 0.567156012635778, "learning_rate": 7.612662184699528e-05, "loss": 2.7846, "step": 18968 }, { "epoch": 1.177540505307592, "grad_norm": 0.5714286438056355, "learning_rate": 7.612354252324068e-05, "loss": 2.7807, "step": 18969 }, { "epoch": 1.17760258240735, "grad_norm": 0.5525357593857007, "learning_rate": 7.61204630631932e-05, "loss": 2.7575, "step": 18970 }, { "epoch": 1.1776646595071079, "grad_norm": 0.5200192446998917, "learning_rate": 7.611738346686895e-05, "loss": 2.7528, "step": 18971 }, { "epoch": 1.1777267366068658, "grad_norm": 0.5651453650673196, "learning_rate": 7.611430373428395e-05, "loss": 2.8702, "step": 18972 }, { "epoch": 1.1777888137066237, "grad_norm": 0.5261246612594866, "learning_rate": 7.611122386545431e-05, "loss": 2.8368, "step": 18973 }, { "epoch": 1.1778508908063816, "grad_norm": 0.5464609758713364, "learning_rate": 7.610814386039605e-05, "loss": 2.8523, "step": 18974 }, { "epoch": 1.1779129679061393, "grad_norm": 0.5094186179085585, "learning_rate": 7.610506371912526e-05, "loss": 2.7789, "step": 18975 }, { "epoch": 1.1779750450058972, "grad_norm": 0.5631219815968221, "learning_rate": 7.610198344165805e-05, "loss": 2.8924, "step": 18976 }, { "epoch": 1.1780371221056551, "grad_norm": 0.5442018571674744, "learning_rate": 7.609890302801044e-05, "loss": 2.9084, "step": 18977 }, { "epoch": 1.178099199205413, "grad_norm": 0.6369155626416781, "learning_rate": 7.609582247819851e-05, "loss": 2.9599, "step": 18978 }, { "epoch": 1.178161276305171, "grad_norm": 0.5232127831752262, "learning_rate": 7.609274179223834e-05, "loss": 2.8932, "step": 18979 }, { "epoch": 1.178223353404929, "grad_norm": 0.5305036182208268, "learning_rate": 7.6089660970146e-05, "loss": 2.7561, "step": 18980 }, { "epoch": 1.1782854305046868, "grad_norm": 0.5983422275730448, "learning_rate": 7.608658001193758e-05, "loss": 2.8724, "step": 18981 }, { "epoch": 1.1783475076044447, "grad_norm": 0.572552033187858, "learning_rate": 7.608349891762913e-05, "loss": 2.9111, "step": 18982 }, { "epoch": 1.1784095847042027, "grad_norm": 0.5250127915685995, "learning_rate": 7.608041768723673e-05, "loss": 2.9456, "step": 18983 }, { "epoch": 1.1784716618039606, "grad_norm": 0.571308207394889, "learning_rate": 7.607733632077647e-05, "loss": 2.8558, "step": 18984 }, { "epoch": 1.1785337389037185, "grad_norm": 0.5494552233452652, "learning_rate": 7.607425481826442e-05, "loss": 2.9519, "step": 18985 }, { "epoch": 1.1785958160034764, "grad_norm": 0.6581013767824493, "learning_rate": 7.607117317971663e-05, "loss": 2.9161, "step": 18986 }, { "epoch": 1.1786578931032343, "grad_norm": 0.633852492627765, "learning_rate": 7.606809140514923e-05, "loss": 2.9505, "step": 18987 }, { "epoch": 1.1787199702029922, "grad_norm": 0.5476495162832867, "learning_rate": 7.606500949457824e-05, "loss": 2.784, "step": 18988 }, { "epoch": 1.17878204730275, "grad_norm": 0.5800476579981716, "learning_rate": 7.60619274480198e-05, "loss": 2.797, "step": 18989 }, { "epoch": 1.1788441244025079, "grad_norm": 0.5287727773553569, "learning_rate": 7.605884526548992e-05, "loss": 2.7711, "step": 18990 }, { "epoch": 1.1789062015022658, "grad_norm": 0.5598689375930499, "learning_rate": 7.605576294700474e-05, "loss": 2.9013, "step": 18991 }, { "epoch": 1.1789682786020237, "grad_norm": 0.5557725528940635, "learning_rate": 7.605268049258032e-05, "loss": 2.914, "step": 18992 }, { "epoch": 1.1790303557017816, "grad_norm": 0.5632585867503659, "learning_rate": 7.604959790223273e-05, "loss": 2.8744, "step": 18993 }, { "epoch": 1.1790924328015395, "grad_norm": 0.5364219011200808, "learning_rate": 7.604651517597807e-05, "loss": 2.8838, "step": 18994 }, { "epoch": 1.1791545099012974, "grad_norm": 0.5164327652164907, "learning_rate": 7.604343231383241e-05, "loss": 2.8846, "step": 18995 }, { "epoch": 1.1792165870010554, "grad_norm": 0.5953554483375361, "learning_rate": 7.604034931581187e-05, "loss": 2.9388, "step": 18996 }, { "epoch": 1.1792786641008133, "grad_norm": 0.5136264447330925, "learning_rate": 7.603726618193247e-05, "loss": 2.9272, "step": 18997 }, { "epoch": 1.1793407412005712, "grad_norm": 0.6098526769713324, "learning_rate": 7.603418291221036e-05, "loss": 2.9086, "step": 18998 }, { "epoch": 1.179402818300329, "grad_norm": 0.5119733104454963, "learning_rate": 7.603109950666158e-05, "loss": 2.8136, "step": 18999 }, { "epoch": 1.1794648954000868, "grad_norm": 0.5433691181595514, "learning_rate": 7.602801596530225e-05, "loss": 2.7517, "step": 19000 }, { "epoch": 1.1795269724998447, "grad_norm": 0.5173661571686148, "learning_rate": 7.602493228814842e-05, "loss": 2.8475, "step": 19001 }, { "epoch": 1.1795890495996026, "grad_norm": 0.5294962996460716, "learning_rate": 7.602184847521623e-05, "loss": 2.8936, "step": 19002 }, { "epoch": 1.1796511266993606, "grad_norm": 0.5882008136608096, "learning_rate": 7.601876452652172e-05, "loss": 2.8733, "step": 19003 }, { "epoch": 1.1797132037991185, "grad_norm": 0.5188662927182319, "learning_rate": 7.6015680442081e-05, "loss": 2.9097, "step": 19004 }, { "epoch": 1.1797752808988764, "grad_norm": 0.5165376677778231, "learning_rate": 7.601259622191017e-05, "loss": 2.7685, "step": 19005 }, { "epoch": 1.1798373579986343, "grad_norm": 0.5142032500173593, "learning_rate": 7.60095118660253e-05, "loss": 2.8976, "step": 19006 }, { "epoch": 1.1798994350983922, "grad_norm": 0.5332421089425482, "learning_rate": 7.600642737444248e-05, "loss": 2.8979, "step": 19007 }, { "epoch": 1.1799615121981502, "grad_norm": 0.5258466592038207, "learning_rate": 7.600334274717782e-05, "loss": 2.8701, "step": 19008 }, { "epoch": 1.180023589297908, "grad_norm": 0.5012480537395791, "learning_rate": 7.600025798424744e-05, "loss": 2.8444, "step": 19009 }, { "epoch": 1.180085666397666, "grad_norm": 0.4987824487049753, "learning_rate": 7.599717308566738e-05, "loss": 2.8869, "step": 19010 }, { "epoch": 1.180147743497424, "grad_norm": 0.5767604085035125, "learning_rate": 7.599408805145373e-05, "loss": 2.8355, "step": 19011 }, { "epoch": 1.1802098205971818, "grad_norm": 0.5050940597392566, "learning_rate": 7.599100288162266e-05, "loss": 2.8297, "step": 19012 }, { "epoch": 1.1802718976969395, "grad_norm": 0.5333097581818629, "learning_rate": 7.598791757619019e-05, "loss": 2.9527, "step": 19013 }, { "epoch": 1.1803339747966974, "grad_norm": 0.5662073509626433, "learning_rate": 7.598483213517246e-05, "loss": 2.7225, "step": 19014 }, { "epoch": 1.1803960518964554, "grad_norm": 0.5334343321605048, "learning_rate": 7.598174655858554e-05, "loss": 2.8039, "step": 19015 }, { "epoch": 1.1804581289962133, "grad_norm": 0.5400873476485608, "learning_rate": 7.597866084644556e-05, "loss": 2.8639, "step": 19016 }, { "epoch": 1.1805202060959712, "grad_norm": 0.6055299051225265, "learning_rate": 7.597557499876857e-05, "loss": 2.8813, "step": 19017 }, { "epoch": 1.180582283195729, "grad_norm": 0.531444373755588, "learning_rate": 7.597248901557072e-05, "loss": 2.8303, "step": 19018 }, { "epoch": 1.180644360295487, "grad_norm": 0.5291730728913056, "learning_rate": 7.59694028968681e-05, "loss": 2.9029, "step": 19019 }, { "epoch": 1.180706437395245, "grad_norm": 0.5599796735105889, "learning_rate": 7.596631664267678e-05, "loss": 2.8399, "step": 19020 }, { "epoch": 1.1807685144950029, "grad_norm": 0.5321961000456585, "learning_rate": 7.59632302530129e-05, "loss": 2.8371, "step": 19021 }, { "epoch": 1.1808305915947608, "grad_norm": 0.5091801346695329, "learning_rate": 7.596014372789253e-05, "loss": 2.9185, "step": 19022 }, { "epoch": 1.1808926686945185, "grad_norm": 0.5710042651162466, "learning_rate": 7.595705706733178e-05, "loss": 2.8295, "step": 19023 }, { "epoch": 1.1809547457942764, "grad_norm": 0.47388583631409364, "learning_rate": 7.595397027134677e-05, "loss": 2.8645, "step": 19024 }, { "epoch": 1.1810168228940343, "grad_norm": 0.6057303571047009, "learning_rate": 7.595088333995361e-05, "loss": 2.9019, "step": 19025 }, { "epoch": 1.1810788999937922, "grad_norm": 0.5613045174473146, "learning_rate": 7.594779627316837e-05, "loss": 2.9459, "step": 19026 }, { "epoch": 1.1811409770935501, "grad_norm": 0.5480213140289668, "learning_rate": 7.594470907100721e-05, "loss": 2.8716, "step": 19027 }, { "epoch": 1.181203054193308, "grad_norm": 0.6039377359766609, "learning_rate": 7.594162173348618e-05, "loss": 2.9351, "step": 19028 }, { "epoch": 1.181265131293066, "grad_norm": 0.5074251898700604, "learning_rate": 7.593853426062141e-05, "loss": 2.8202, "step": 19029 }, { "epoch": 1.181327208392824, "grad_norm": 0.5915599386459011, "learning_rate": 7.593544665242902e-05, "loss": 2.8038, "step": 19030 }, { "epoch": 1.1813892854925818, "grad_norm": 0.49080008280270726, "learning_rate": 7.59323589089251e-05, "loss": 2.8337, "step": 19031 }, { "epoch": 1.1814513625923397, "grad_norm": 0.5769041677273845, "learning_rate": 7.592927103012578e-05, "loss": 2.8454, "step": 19032 }, { "epoch": 1.1815134396920977, "grad_norm": 0.488456511514953, "learning_rate": 7.592618301604715e-05, "loss": 2.8385, "step": 19033 }, { "epoch": 1.1815755167918556, "grad_norm": 0.5479869709188894, "learning_rate": 7.592309486670534e-05, "loss": 2.8707, "step": 19034 }, { "epoch": 1.1816375938916135, "grad_norm": 0.5335696291742716, "learning_rate": 7.592000658211645e-05, "loss": 2.8408, "step": 19035 }, { "epoch": 1.1816996709913712, "grad_norm": 0.4843868284934656, "learning_rate": 7.591691816229658e-05, "loss": 2.8378, "step": 19036 }, { "epoch": 1.181761748091129, "grad_norm": 0.5899074090622015, "learning_rate": 7.591382960726185e-05, "loss": 2.8573, "step": 19037 }, { "epoch": 1.181823825190887, "grad_norm": 0.5661810065336207, "learning_rate": 7.59107409170284e-05, "loss": 2.8596, "step": 19038 }, { "epoch": 1.181885902290645, "grad_norm": 0.5972843134808538, "learning_rate": 7.590765209161232e-05, "loss": 2.826, "step": 19039 }, { "epoch": 1.1819479793904029, "grad_norm": 0.5035264409858998, "learning_rate": 7.590456313102972e-05, "loss": 2.8201, "step": 19040 }, { "epoch": 1.1820100564901608, "grad_norm": 0.5256011427682562, "learning_rate": 7.590147403529672e-05, "loss": 2.8407, "step": 19041 }, { "epoch": 1.1820721335899187, "grad_norm": 0.515645084567922, "learning_rate": 7.589838480442946e-05, "loss": 2.8393, "step": 19042 }, { "epoch": 1.1821342106896766, "grad_norm": 0.5058049058493757, "learning_rate": 7.589529543844404e-05, "loss": 2.8261, "step": 19043 }, { "epoch": 1.1821962877894345, "grad_norm": 0.521362290396899, "learning_rate": 7.589220593735658e-05, "loss": 2.9307, "step": 19044 }, { "epoch": 1.1822583648891924, "grad_norm": 0.5312545439581923, "learning_rate": 7.588911630118318e-05, "loss": 2.787, "step": 19045 }, { "epoch": 1.1823204419889504, "grad_norm": 0.49606549940996386, "learning_rate": 7.588602652993999e-05, "loss": 2.8978, "step": 19046 }, { "epoch": 1.182382519088708, "grad_norm": 0.5770059787351242, "learning_rate": 7.58829366236431e-05, "loss": 2.8372, "step": 19047 }, { "epoch": 1.182444596188466, "grad_norm": 0.4621924434498086, "learning_rate": 7.587984658230866e-05, "loss": 2.8297, "step": 19048 }, { "epoch": 1.182506673288224, "grad_norm": 0.5395707484904753, "learning_rate": 7.587675640595277e-05, "loss": 2.7863, "step": 19049 }, { "epoch": 1.1825687503879818, "grad_norm": 0.5640312386934985, "learning_rate": 7.587366609459156e-05, "loss": 2.8977, "step": 19050 }, { "epoch": 1.1826308274877397, "grad_norm": 0.4785394935104634, "learning_rate": 7.587057564824116e-05, "loss": 2.789, "step": 19051 }, { "epoch": 1.1826929045874977, "grad_norm": 0.5462093934825263, "learning_rate": 7.586748506691769e-05, "loss": 2.8567, "step": 19052 }, { "epoch": 1.1827549816872556, "grad_norm": 0.5070059312985125, "learning_rate": 7.586439435063726e-05, "loss": 2.9434, "step": 19053 }, { "epoch": 1.1828170587870135, "grad_norm": 0.5716557035377345, "learning_rate": 7.5861303499416e-05, "loss": 2.8483, "step": 19054 }, { "epoch": 1.1828791358867714, "grad_norm": 0.5762817795238396, "learning_rate": 7.585821251327006e-05, "loss": 2.9154, "step": 19055 }, { "epoch": 1.1829412129865293, "grad_norm": 0.4900380079456475, "learning_rate": 7.585512139221553e-05, "loss": 2.8666, "step": 19056 }, { "epoch": 1.1830032900862872, "grad_norm": 0.5533446592393586, "learning_rate": 7.585203013626857e-05, "loss": 2.8666, "step": 19057 }, { "epoch": 1.1830653671860452, "grad_norm": 0.47978091433136977, "learning_rate": 7.584893874544528e-05, "loss": 2.9355, "step": 19058 }, { "epoch": 1.183127444285803, "grad_norm": 0.5391105962112084, "learning_rate": 7.584584721976181e-05, "loss": 2.8434, "step": 19059 }, { "epoch": 1.1831895213855608, "grad_norm": 0.5521833281259586, "learning_rate": 7.584275555923427e-05, "loss": 2.9087, "step": 19060 }, { "epoch": 1.1832515984853187, "grad_norm": 0.5691651076560313, "learning_rate": 7.583966376387883e-05, "loss": 2.9213, "step": 19061 }, { "epoch": 1.1833136755850766, "grad_norm": 0.5201039993384675, "learning_rate": 7.583657183371155e-05, "loss": 2.7478, "step": 19062 }, { "epoch": 1.1833757526848345, "grad_norm": 0.527326597711867, "learning_rate": 7.583347976874865e-05, "loss": 2.943, "step": 19063 }, { "epoch": 1.1834378297845924, "grad_norm": 0.5528550841823676, "learning_rate": 7.583038756900618e-05, "loss": 2.9324, "step": 19064 }, { "epoch": 1.1834999068843504, "grad_norm": 0.5071361145294717, "learning_rate": 7.582729523450032e-05, "loss": 2.8572, "step": 19065 }, { "epoch": 1.1835619839841083, "grad_norm": 0.5392046478381747, "learning_rate": 7.582420276524719e-05, "loss": 2.8006, "step": 19066 }, { "epoch": 1.1836240610838662, "grad_norm": 0.5144801000244341, "learning_rate": 7.582111016126292e-05, "loss": 2.829, "step": 19067 }, { "epoch": 1.1836861381836241, "grad_norm": 0.5920321308531139, "learning_rate": 7.581801742256365e-05, "loss": 2.9082, "step": 19068 }, { "epoch": 1.183748215283382, "grad_norm": 0.4516198052604583, "learning_rate": 7.581492454916551e-05, "loss": 2.7753, "step": 19069 }, { "epoch": 1.18381029238314, "grad_norm": 0.5780889396762229, "learning_rate": 7.581183154108466e-05, "loss": 2.8355, "step": 19070 }, { "epoch": 1.1838723694828976, "grad_norm": 0.5138659962575423, "learning_rate": 7.58087383983372e-05, "loss": 2.8484, "step": 19071 }, { "epoch": 1.1839344465826556, "grad_norm": 0.4799643093737981, "learning_rate": 7.580564512093928e-05, "loss": 2.7694, "step": 19072 }, { "epoch": 1.1839965236824135, "grad_norm": 0.5205040049458386, "learning_rate": 7.580255170890708e-05, "loss": 2.8845, "step": 19073 }, { "epoch": 1.1840586007821714, "grad_norm": 0.5186042259102976, "learning_rate": 7.579945816225668e-05, "loss": 2.8892, "step": 19074 }, { "epoch": 1.1841206778819293, "grad_norm": 0.5436949230669198, "learning_rate": 7.579636448100425e-05, "loss": 2.859, "step": 19075 }, { "epoch": 1.1841827549816872, "grad_norm": 0.6277591598355949, "learning_rate": 7.579327066516591e-05, "loss": 2.9072, "step": 19076 }, { "epoch": 1.1842448320814452, "grad_norm": 0.5556511902617125, "learning_rate": 7.579017671475784e-05, "loss": 2.7721, "step": 19077 }, { "epoch": 1.184306909181203, "grad_norm": 0.5918783503178776, "learning_rate": 7.578708262979614e-05, "loss": 2.9179, "step": 19078 }, { "epoch": 1.184368986280961, "grad_norm": 0.6013107268344264, "learning_rate": 7.578398841029697e-05, "loss": 2.8097, "step": 19079 }, { "epoch": 1.184431063380719, "grad_norm": 0.5791119322081558, "learning_rate": 7.578089405627648e-05, "loss": 2.8907, "step": 19080 }, { "epoch": 1.1844931404804768, "grad_norm": 0.5969413520721452, "learning_rate": 7.577779956775081e-05, "loss": 2.9402, "step": 19081 }, { "epoch": 1.1845552175802347, "grad_norm": 0.6564257250255209, "learning_rate": 7.57747049447361e-05, "loss": 2.9091, "step": 19082 }, { "epoch": 1.1846172946799927, "grad_norm": 0.5016737579824584, "learning_rate": 7.577161018724848e-05, "loss": 2.7913, "step": 19083 }, { "epoch": 1.1846793717797504, "grad_norm": 0.6011499749173576, "learning_rate": 7.576851529530413e-05, "loss": 2.9112, "step": 19084 }, { "epoch": 1.1847414488795083, "grad_norm": 0.5312014725904562, "learning_rate": 7.576542026891917e-05, "loss": 2.8633, "step": 19085 }, { "epoch": 1.1848035259792662, "grad_norm": 0.6345441226362711, "learning_rate": 7.576232510810978e-05, "loss": 2.8482, "step": 19086 }, { "epoch": 1.184865603079024, "grad_norm": 0.511888257554814, "learning_rate": 7.575922981289207e-05, "loss": 2.8323, "step": 19087 }, { "epoch": 1.184927680178782, "grad_norm": 0.6023541727908348, "learning_rate": 7.575613438328221e-05, "loss": 2.8598, "step": 19088 }, { "epoch": 1.18498975727854, "grad_norm": 0.4907034647211151, "learning_rate": 7.575303881929633e-05, "loss": 2.8767, "step": 19089 }, { "epoch": 1.1850518343782979, "grad_norm": 0.6087565340243679, "learning_rate": 7.574994312095061e-05, "loss": 2.9125, "step": 19090 }, { "epoch": 1.1851139114780558, "grad_norm": 0.4942193100594728, "learning_rate": 7.574684728826117e-05, "loss": 2.8632, "step": 19091 }, { "epoch": 1.1851759885778137, "grad_norm": 0.5885142570347031, "learning_rate": 7.574375132124419e-05, "loss": 2.9061, "step": 19092 }, { "epoch": 1.1852380656775716, "grad_norm": 0.5099397851420783, "learning_rate": 7.574065521991582e-05, "loss": 2.8553, "step": 19093 }, { "epoch": 1.1853001427773295, "grad_norm": 0.5153513095322775, "learning_rate": 7.573755898429217e-05, "loss": 2.8035, "step": 19094 }, { "epoch": 1.1853622198770872, "grad_norm": 0.5665080011986222, "learning_rate": 7.573446261438945e-05, "loss": 2.8587, "step": 19095 }, { "epoch": 1.1854242969768451, "grad_norm": 0.46427151375317655, "learning_rate": 7.57313661102238e-05, "loss": 2.744, "step": 19096 }, { "epoch": 1.185486374076603, "grad_norm": 0.5908341335224854, "learning_rate": 7.572826947181135e-05, "loss": 2.9292, "step": 19097 }, { "epoch": 1.185548451176361, "grad_norm": 0.5123153877377027, "learning_rate": 7.572517269916828e-05, "loss": 2.8562, "step": 19098 }, { "epoch": 1.185610528276119, "grad_norm": 0.5728481309529796, "learning_rate": 7.572207579231075e-05, "loss": 2.766, "step": 19099 }, { "epoch": 1.1856726053758768, "grad_norm": 0.575511696399179, "learning_rate": 7.571897875125487e-05, "loss": 2.7748, "step": 19100 }, { "epoch": 1.1857346824756347, "grad_norm": 0.5568971009583654, "learning_rate": 7.571588157601687e-05, "loss": 2.8358, "step": 19101 }, { "epoch": 1.1857967595753927, "grad_norm": 0.5871357427730538, "learning_rate": 7.571278426661286e-05, "loss": 2.8441, "step": 19102 }, { "epoch": 1.1858588366751506, "grad_norm": 0.544474428296263, "learning_rate": 7.570968682305901e-05, "loss": 2.7765, "step": 19103 }, { "epoch": 1.1859209137749085, "grad_norm": 0.5272935066306593, "learning_rate": 7.57065892453715e-05, "loss": 2.7202, "step": 19104 }, { "epoch": 1.1859829908746664, "grad_norm": 0.5384682960138626, "learning_rate": 7.570349153356644e-05, "loss": 2.7895, "step": 19105 }, { "epoch": 1.1860450679744243, "grad_norm": 0.5124451526508242, "learning_rate": 7.570039368766004e-05, "loss": 2.9547, "step": 19106 }, { "epoch": 1.1861071450741822, "grad_norm": 0.5973385483786943, "learning_rate": 7.569729570766845e-05, "loss": 2.9104, "step": 19107 }, { "epoch": 1.18616922217394, "grad_norm": 0.5504841916519608, "learning_rate": 7.569419759360784e-05, "loss": 2.8441, "step": 19108 }, { "epoch": 1.1862312992736979, "grad_norm": 0.502016412329912, "learning_rate": 7.569109934549436e-05, "loss": 2.8986, "step": 19109 }, { "epoch": 1.1862933763734558, "grad_norm": 0.5553729429393641, "learning_rate": 7.568800096334416e-05, "loss": 2.8624, "step": 19110 }, { "epoch": 1.1863554534732137, "grad_norm": 0.5122710007104335, "learning_rate": 7.568490244717346e-05, "loss": 2.8976, "step": 19111 }, { "epoch": 1.1864175305729716, "grad_norm": 0.6064402999302203, "learning_rate": 7.568180379699834e-05, "loss": 2.8919, "step": 19112 }, { "epoch": 1.1864796076727295, "grad_norm": 0.5887719325619156, "learning_rate": 7.567870501283505e-05, "loss": 2.9653, "step": 19113 }, { "epoch": 1.1865416847724874, "grad_norm": 0.5467357730750031, "learning_rate": 7.567560609469971e-05, "loss": 2.8749, "step": 19114 }, { "epoch": 1.1866037618722454, "grad_norm": 0.5918537276250924, "learning_rate": 7.56725070426085e-05, "loss": 2.8152, "step": 19115 }, { "epoch": 1.1866658389720033, "grad_norm": 0.5746316268834192, "learning_rate": 7.56694078565776e-05, "loss": 2.8879, "step": 19116 }, { "epoch": 1.1867279160717612, "grad_norm": 0.5575201481227566, "learning_rate": 7.566630853662315e-05, "loss": 2.8746, "step": 19117 }, { "epoch": 1.1867899931715191, "grad_norm": 0.5782603543579485, "learning_rate": 7.566320908276134e-05, "loss": 2.8751, "step": 19118 }, { "epoch": 1.1868520702712768, "grad_norm": 0.5661152585455539, "learning_rate": 7.566010949500835e-05, "loss": 2.8795, "step": 19119 }, { "epoch": 1.1869141473710347, "grad_norm": 0.5317413918480384, "learning_rate": 7.565700977338034e-05, "loss": 2.8301, "step": 19120 }, { "epoch": 1.1869762244707927, "grad_norm": 0.5091347133497558, "learning_rate": 7.565390991789346e-05, "loss": 2.8736, "step": 19121 }, { "epoch": 1.1870383015705506, "grad_norm": 0.6121928038277291, "learning_rate": 7.565080992856392e-05, "loss": 2.8935, "step": 19122 }, { "epoch": 1.1871003786703085, "grad_norm": 0.540243210241043, "learning_rate": 7.564770980540788e-05, "loss": 2.8752, "step": 19123 }, { "epoch": 1.1871624557700664, "grad_norm": 0.5589277695972605, "learning_rate": 7.564460954844152e-05, "loss": 2.8653, "step": 19124 }, { "epoch": 1.1872245328698243, "grad_norm": 0.5398860114375226, "learning_rate": 7.5641509157681e-05, "loss": 2.9287, "step": 19125 }, { "epoch": 1.1872866099695822, "grad_norm": 0.531715441616997, "learning_rate": 7.563840863314248e-05, "loss": 2.8661, "step": 19126 }, { "epoch": 1.1873486870693402, "grad_norm": 0.5146630301388228, "learning_rate": 7.563530797484218e-05, "loss": 2.8273, "step": 19127 }, { "epoch": 1.187410764169098, "grad_norm": 0.4904560788435801, "learning_rate": 7.563220718279624e-05, "loss": 2.8444, "step": 19128 }, { "epoch": 1.187472841268856, "grad_norm": 0.5905324187201828, "learning_rate": 7.562910625702089e-05, "loss": 2.9128, "step": 19129 }, { "epoch": 1.187534918368614, "grad_norm": 0.5553776651593675, "learning_rate": 7.562600519753222e-05, "loss": 2.8895, "step": 19130 }, { "epoch": 1.1875969954683718, "grad_norm": 0.556247425073386, "learning_rate": 7.562290400434651e-05, "loss": 2.8984, "step": 19131 }, { "epoch": 1.1876590725681295, "grad_norm": 0.5374072826615367, "learning_rate": 7.561980267747986e-05, "loss": 2.8826, "step": 19132 }, { "epoch": 1.1877211496678874, "grad_norm": 0.5741388174462762, "learning_rate": 7.561670121694848e-05, "loss": 2.83, "step": 19133 }, { "epoch": 1.1877832267676454, "grad_norm": 0.5584643554207217, "learning_rate": 7.561359962276855e-05, "loss": 2.8113, "step": 19134 }, { "epoch": 1.1878453038674033, "grad_norm": 0.5519469350587868, "learning_rate": 7.561049789495626e-05, "loss": 2.8091, "step": 19135 }, { "epoch": 1.1879073809671612, "grad_norm": 0.5514038841259499, "learning_rate": 7.560739603352777e-05, "loss": 2.9136, "step": 19136 }, { "epoch": 1.1879694580669191, "grad_norm": 0.5573617914828713, "learning_rate": 7.560429403849931e-05, "loss": 2.7255, "step": 19137 }, { "epoch": 1.188031535166677, "grad_norm": 0.48132544025303325, "learning_rate": 7.5601191909887e-05, "loss": 2.8819, "step": 19138 }, { "epoch": 1.188093612266435, "grad_norm": 0.5319939341181362, "learning_rate": 7.559808964770707e-05, "loss": 2.8601, "step": 19139 }, { "epoch": 1.1881556893661929, "grad_norm": 0.4905016956766665, "learning_rate": 7.55949872519757e-05, "loss": 2.8761, "step": 19140 }, { "epoch": 1.1882177664659508, "grad_norm": 0.5307662949632383, "learning_rate": 7.559188472270906e-05, "loss": 2.7902, "step": 19141 }, { "epoch": 1.1882798435657087, "grad_norm": 0.5100309770189658, "learning_rate": 7.558878205992334e-05, "loss": 2.8809, "step": 19142 }, { "epoch": 1.1883419206654664, "grad_norm": 0.5710794961092445, "learning_rate": 7.558567926363474e-05, "loss": 2.815, "step": 19143 }, { "epoch": 1.1884039977652243, "grad_norm": 0.5260358344280376, "learning_rate": 7.558257633385943e-05, "loss": 2.8253, "step": 19144 }, { "epoch": 1.1884660748649822, "grad_norm": 0.5693038748191329, "learning_rate": 7.557947327061363e-05, "loss": 2.848, "step": 19145 }, { "epoch": 1.1885281519647402, "grad_norm": 0.5817903365711217, "learning_rate": 7.55763700739135e-05, "loss": 2.7701, "step": 19146 }, { "epoch": 1.188590229064498, "grad_norm": 0.5687883992908762, "learning_rate": 7.557326674377522e-05, "loss": 2.8572, "step": 19147 }, { "epoch": 1.188652306164256, "grad_norm": 0.5299949125279664, "learning_rate": 7.557016328021503e-05, "loss": 2.8616, "step": 19148 }, { "epoch": 1.188714383264014, "grad_norm": 0.5833815969074142, "learning_rate": 7.556705968324907e-05, "loss": 2.8322, "step": 19149 }, { "epoch": 1.1887764603637718, "grad_norm": 0.5682946394401547, "learning_rate": 7.556395595289355e-05, "loss": 2.9213, "step": 19150 }, { "epoch": 1.1888385374635297, "grad_norm": 0.48645165035168997, "learning_rate": 7.556085208916468e-05, "loss": 2.8649, "step": 19151 }, { "epoch": 1.1889006145632877, "grad_norm": 0.5598608730549822, "learning_rate": 7.555774809207862e-05, "loss": 2.8092, "step": 19152 }, { "epoch": 1.1889626916630456, "grad_norm": 0.5181050910851982, "learning_rate": 7.555464396165161e-05, "loss": 2.8842, "step": 19153 }, { "epoch": 1.1890247687628035, "grad_norm": 0.4969300557470015, "learning_rate": 7.55515396978998e-05, "loss": 2.8574, "step": 19154 }, { "epoch": 1.1890868458625614, "grad_norm": 0.5405984640499739, "learning_rate": 7.55484353008394e-05, "loss": 2.8809, "step": 19155 }, { "epoch": 1.189148922962319, "grad_norm": 0.4859599516119161, "learning_rate": 7.554533077048663e-05, "loss": 2.8445, "step": 19156 }, { "epoch": 1.189211000062077, "grad_norm": 0.5557944843645874, "learning_rate": 7.554222610685765e-05, "loss": 2.8437, "step": 19157 }, { "epoch": 1.189273077161835, "grad_norm": 0.5695412790595429, "learning_rate": 7.55391213099687e-05, "loss": 2.9016, "step": 19158 }, { "epoch": 1.1893351542615929, "grad_norm": 0.5230947197703962, "learning_rate": 7.553601637983593e-05, "loss": 2.9154, "step": 19159 }, { "epoch": 1.1893972313613508, "grad_norm": 0.5528771313827384, "learning_rate": 7.553291131647558e-05, "loss": 2.8392, "step": 19160 }, { "epoch": 1.1894593084611087, "grad_norm": 0.5250448627831261, "learning_rate": 7.552980611990381e-05, "loss": 2.8731, "step": 19161 }, { "epoch": 1.1895213855608666, "grad_norm": 0.6156801576642233, "learning_rate": 7.552670079013687e-05, "loss": 2.8812, "step": 19162 }, { "epoch": 1.1895834626606245, "grad_norm": 0.5396266694015472, "learning_rate": 7.552359532719093e-05, "loss": 2.7912, "step": 19163 }, { "epoch": 1.1896455397603825, "grad_norm": 0.5172534716154539, "learning_rate": 7.55204897310822e-05, "loss": 2.8947, "step": 19164 }, { "epoch": 1.1897076168601404, "grad_norm": 0.5994245511476951, "learning_rate": 7.551738400182687e-05, "loss": 2.9352, "step": 19165 }, { "epoch": 1.1897696939598983, "grad_norm": 0.583621444964208, "learning_rate": 7.551427813944113e-05, "loss": 2.8482, "step": 19166 }, { "epoch": 1.189831771059656, "grad_norm": 0.4607302474560424, "learning_rate": 7.551117214394125e-05, "loss": 2.8038, "step": 19167 }, { "epoch": 1.189893848159414, "grad_norm": 0.5743177419438403, "learning_rate": 7.550806601534337e-05, "loss": 2.7975, "step": 19168 }, { "epoch": 1.1899559252591718, "grad_norm": 0.5239366172956922, "learning_rate": 7.550495975366372e-05, "loss": 2.866, "step": 19169 }, { "epoch": 1.1900180023589297, "grad_norm": 0.5329623953989275, "learning_rate": 7.55018533589185e-05, "loss": 2.8488, "step": 19170 }, { "epoch": 1.1900800794586877, "grad_norm": 0.5092934361783675, "learning_rate": 7.549874683112392e-05, "loss": 2.8124, "step": 19171 }, { "epoch": 1.1901421565584456, "grad_norm": 0.5584527217794906, "learning_rate": 7.54956401702962e-05, "loss": 2.871, "step": 19172 }, { "epoch": 1.1902042336582035, "grad_norm": 0.5821829476539415, "learning_rate": 7.549253337645153e-05, "loss": 2.8988, "step": 19173 }, { "epoch": 1.1902663107579614, "grad_norm": 0.571822349542731, "learning_rate": 7.548942644960612e-05, "loss": 2.8472, "step": 19174 }, { "epoch": 1.1903283878577193, "grad_norm": 0.571118060219494, "learning_rate": 7.548631938977617e-05, "loss": 2.8583, "step": 19175 }, { "epoch": 1.1903904649574772, "grad_norm": 0.5522689747138785, "learning_rate": 7.548321219697793e-05, "loss": 2.8379, "step": 19176 }, { "epoch": 1.1904525420572352, "grad_norm": 0.5548496278039503, "learning_rate": 7.548010487122759e-05, "loss": 2.9317, "step": 19177 }, { "epoch": 1.190514619156993, "grad_norm": 0.5393062261715389, "learning_rate": 7.547699741254133e-05, "loss": 2.837, "step": 19178 }, { "epoch": 1.190576696256751, "grad_norm": 0.5386499125185351, "learning_rate": 7.547388982093541e-05, "loss": 2.8421, "step": 19179 }, { "epoch": 1.1906387733565087, "grad_norm": 0.525685309262553, "learning_rate": 7.547078209642601e-05, "loss": 2.8498, "step": 19180 }, { "epoch": 1.1907008504562666, "grad_norm": 0.52116632942654, "learning_rate": 7.546767423902938e-05, "loss": 2.8355, "step": 19181 }, { "epoch": 1.1907629275560245, "grad_norm": 0.554984268532572, "learning_rate": 7.546456624876169e-05, "loss": 2.8852, "step": 19182 }, { "epoch": 1.1908250046557824, "grad_norm": 0.5252689092955388, "learning_rate": 7.546145812563918e-05, "loss": 2.877, "step": 19183 }, { "epoch": 1.1908870817555404, "grad_norm": 0.48681095774518063, "learning_rate": 7.545834986967806e-05, "loss": 2.8399, "step": 19184 }, { "epoch": 1.1909491588552983, "grad_norm": 0.5186582702516468, "learning_rate": 7.545524148089455e-05, "loss": 2.8333, "step": 19185 }, { "epoch": 1.1910112359550562, "grad_norm": 0.48148819304931123, "learning_rate": 7.545213295930488e-05, "loss": 2.8159, "step": 19186 }, { "epoch": 1.1910733130548141, "grad_norm": 0.5112690879131149, "learning_rate": 7.544902430492523e-05, "loss": 2.904, "step": 19187 }, { "epoch": 1.191135390154572, "grad_norm": 0.5345038175080791, "learning_rate": 7.544591551777186e-05, "loss": 2.7994, "step": 19188 }, { "epoch": 1.19119746725433, "grad_norm": 0.468257963704476, "learning_rate": 7.544280659786096e-05, "loss": 2.8336, "step": 19189 }, { "epoch": 1.1912595443540879, "grad_norm": 0.5563784868812013, "learning_rate": 7.543969754520876e-05, "loss": 2.8435, "step": 19190 }, { "epoch": 1.1913216214538456, "grad_norm": 0.48018589843121456, "learning_rate": 7.543658835983149e-05, "loss": 2.8199, "step": 19191 }, { "epoch": 1.1913836985536035, "grad_norm": 0.6019125081196417, "learning_rate": 7.543347904174538e-05, "loss": 2.9077, "step": 19192 }, { "epoch": 1.1914457756533614, "grad_norm": 0.5235268530803602, "learning_rate": 7.543036959096661e-05, "loss": 2.8425, "step": 19193 }, { "epoch": 1.1915078527531193, "grad_norm": 0.5409908357855905, "learning_rate": 7.542726000751143e-05, "loss": 2.8519, "step": 19194 }, { "epoch": 1.1915699298528772, "grad_norm": 0.5950334130049468, "learning_rate": 7.542415029139607e-05, "loss": 2.9135, "step": 19195 }, { "epoch": 1.1916320069526352, "grad_norm": 0.560713527243312, "learning_rate": 7.542104044263676e-05, "loss": 2.8984, "step": 19196 }, { "epoch": 1.191694084052393, "grad_norm": 0.5199571169080286, "learning_rate": 7.541793046124967e-05, "loss": 2.9684, "step": 19197 }, { "epoch": 1.191756161152151, "grad_norm": 0.5942912646254859, "learning_rate": 7.541482034725111e-05, "loss": 2.9356, "step": 19198 }, { "epoch": 1.191818238251909, "grad_norm": 0.590885102053531, "learning_rate": 7.541171010065724e-05, "loss": 2.8525, "step": 19199 }, { "epoch": 1.1918803153516668, "grad_norm": 0.5299254622767519, "learning_rate": 7.540859972148431e-05, "loss": 2.8258, "step": 19200 }, { "epoch": 1.1919423924514247, "grad_norm": 0.6553824002992057, "learning_rate": 7.540548920974856e-05, "loss": 2.8049, "step": 19201 }, { "epoch": 1.1920044695511827, "grad_norm": 0.5418179618620877, "learning_rate": 7.540237856546619e-05, "loss": 2.7526, "step": 19202 }, { "epoch": 1.1920665466509406, "grad_norm": 0.5489631839612693, "learning_rate": 7.539926778865345e-05, "loss": 2.7613, "step": 19203 }, { "epoch": 1.1921286237506983, "grad_norm": 0.5766960475259657, "learning_rate": 7.539615687932655e-05, "loss": 2.7311, "step": 19204 }, { "epoch": 1.1921907008504562, "grad_norm": 0.5571215242102411, "learning_rate": 7.539304583750176e-05, "loss": 2.8621, "step": 19205 }, { "epoch": 1.1922527779502141, "grad_norm": 0.5986611140221277, "learning_rate": 7.538993466319529e-05, "loss": 2.8527, "step": 19206 }, { "epoch": 1.192314855049972, "grad_norm": 0.5453847744328322, "learning_rate": 7.538682335642334e-05, "loss": 2.8527, "step": 19207 }, { "epoch": 1.19237693214973, "grad_norm": 0.5156612961449782, "learning_rate": 7.538371191720218e-05, "loss": 2.8601, "step": 19208 }, { "epoch": 1.1924390092494879, "grad_norm": 0.5090980982408312, "learning_rate": 7.538060034554804e-05, "loss": 2.911, "step": 19209 }, { "epoch": 1.1925010863492458, "grad_norm": 0.5498557736122023, "learning_rate": 7.537748864147713e-05, "loss": 2.8963, "step": 19210 }, { "epoch": 1.1925631634490037, "grad_norm": 0.45946490517381194, "learning_rate": 7.537437680500572e-05, "loss": 2.8154, "step": 19211 }, { "epoch": 1.1926252405487616, "grad_norm": 0.5763545118623085, "learning_rate": 7.537126483615002e-05, "loss": 2.8378, "step": 19212 }, { "epoch": 1.1926873176485195, "grad_norm": 0.5907332427125761, "learning_rate": 7.536815273492627e-05, "loss": 2.8301, "step": 19213 }, { "epoch": 1.1927493947482772, "grad_norm": 0.5197550069486139, "learning_rate": 7.536504050135072e-05, "loss": 2.8531, "step": 19214 }, { "epoch": 1.1928114718480352, "grad_norm": 0.647156656839021, "learning_rate": 7.536192813543957e-05, "loss": 2.8688, "step": 19215 }, { "epoch": 1.192873548947793, "grad_norm": 0.5177439585913013, "learning_rate": 7.53588156372091e-05, "loss": 2.81, "step": 19216 }, { "epoch": 1.192935626047551, "grad_norm": 0.5644156895667588, "learning_rate": 7.535570300667554e-05, "loss": 2.7645, "step": 19217 }, { "epoch": 1.192997703147309, "grad_norm": 0.5209615708472525, "learning_rate": 7.53525902438551e-05, "loss": 2.8106, "step": 19218 }, { "epoch": 1.1930597802470668, "grad_norm": 0.5266515639509786, "learning_rate": 7.534947734876407e-05, "loss": 2.8061, "step": 19219 }, { "epoch": 1.1931218573468247, "grad_norm": 0.5087405893734814, "learning_rate": 7.534636432141863e-05, "loss": 2.7664, "step": 19220 }, { "epoch": 1.1931839344465827, "grad_norm": 0.4904816448440342, "learning_rate": 7.534325116183509e-05, "loss": 2.9344, "step": 19221 }, { "epoch": 1.1932460115463406, "grad_norm": 0.568016709266614, "learning_rate": 7.534013787002962e-05, "loss": 2.9047, "step": 19222 }, { "epoch": 1.1933080886460985, "grad_norm": 0.547550628751133, "learning_rate": 7.533702444601853e-05, "loss": 2.8367, "step": 19223 }, { "epoch": 1.1933701657458564, "grad_norm": 0.581668433452681, "learning_rate": 7.533391088981803e-05, "loss": 2.8325, "step": 19224 }, { "epoch": 1.1934322428456143, "grad_norm": 0.5389690318137557, "learning_rate": 7.533079720144436e-05, "loss": 2.7731, "step": 19225 }, { "epoch": 1.1934943199453723, "grad_norm": 0.5915098847831352, "learning_rate": 7.532768338091377e-05, "loss": 2.8359, "step": 19226 }, { "epoch": 1.1935563970451302, "grad_norm": 0.5530232729282868, "learning_rate": 7.532456942824251e-05, "loss": 2.8434, "step": 19227 }, { "epoch": 1.1936184741448879, "grad_norm": 0.6126158925788113, "learning_rate": 7.532145534344682e-05, "loss": 2.8134, "step": 19228 }, { "epoch": 1.1936805512446458, "grad_norm": 0.5275583748811518, "learning_rate": 7.531834112654295e-05, "loss": 2.8741, "step": 19229 }, { "epoch": 1.1937426283444037, "grad_norm": 0.5317663039993149, "learning_rate": 7.531522677754716e-05, "loss": 2.817, "step": 19230 }, { "epoch": 1.1938047054441616, "grad_norm": 0.504567857857274, "learning_rate": 7.531211229647567e-05, "loss": 2.8025, "step": 19231 }, { "epoch": 1.1938667825439195, "grad_norm": 0.501292406129116, "learning_rate": 7.530899768334476e-05, "loss": 2.9176, "step": 19232 }, { "epoch": 1.1939288596436775, "grad_norm": 0.5365295887508711, "learning_rate": 7.530588293817064e-05, "loss": 2.8352, "step": 19233 }, { "epoch": 1.1939909367434354, "grad_norm": 0.5019832260665358, "learning_rate": 7.530276806096961e-05, "loss": 2.8844, "step": 19234 }, { "epoch": 1.1940530138431933, "grad_norm": 0.5685671659179397, "learning_rate": 7.52996530517579e-05, "loss": 2.8145, "step": 19235 }, { "epoch": 1.1941150909429512, "grad_norm": 0.5183674677293987, "learning_rate": 7.529653791055173e-05, "loss": 2.9329, "step": 19236 }, { "epoch": 1.1941771680427091, "grad_norm": 0.5331631320584994, "learning_rate": 7.52934226373674e-05, "loss": 2.9413, "step": 19237 }, { "epoch": 1.1942392451424668, "grad_norm": 0.5232268521637874, "learning_rate": 7.529030723222114e-05, "loss": 2.8085, "step": 19238 }, { "epoch": 1.1943013222422247, "grad_norm": 0.5509731147925989, "learning_rate": 7.528719169512921e-05, "loss": 2.8134, "step": 19239 }, { "epoch": 1.1943633993419827, "grad_norm": 0.5245810869647047, "learning_rate": 7.528407602610786e-05, "loss": 2.8819, "step": 19240 }, { "epoch": 1.1944254764417406, "grad_norm": 0.515327627955845, "learning_rate": 7.528096022517335e-05, "loss": 2.8671, "step": 19241 }, { "epoch": 1.1944875535414985, "grad_norm": 0.5338299877614924, "learning_rate": 7.527784429234195e-05, "loss": 2.9076, "step": 19242 }, { "epoch": 1.1945496306412564, "grad_norm": 0.4811500897673145, "learning_rate": 7.527472822762986e-05, "loss": 2.8108, "step": 19243 }, { "epoch": 1.1946117077410143, "grad_norm": 0.5640422024865809, "learning_rate": 7.527161203105342e-05, "loss": 2.8941, "step": 19244 }, { "epoch": 1.1946737848407722, "grad_norm": 0.5451155834386249, "learning_rate": 7.526849570262882e-05, "loss": 2.9159, "step": 19245 }, { "epoch": 1.1947358619405302, "grad_norm": 0.5242907274295353, "learning_rate": 7.526537924237236e-05, "loss": 2.8196, "step": 19246 }, { "epoch": 1.194797939040288, "grad_norm": 0.5709506344759047, "learning_rate": 7.526226265030026e-05, "loss": 2.8449, "step": 19247 }, { "epoch": 1.194860016140046, "grad_norm": 0.5759486655085957, "learning_rate": 7.525914592642882e-05, "loss": 2.9578, "step": 19248 }, { "epoch": 1.194922093239804, "grad_norm": 0.579845368926066, "learning_rate": 7.525602907077427e-05, "loss": 2.9033, "step": 19249 }, { "epoch": 1.1949841703395618, "grad_norm": 0.6613243197553004, "learning_rate": 7.52529120833529e-05, "loss": 2.9026, "step": 19250 }, { "epoch": 1.1950462474393198, "grad_norm": 0.6158159927908701, "learning_rate": 7.524979496418095e-05, "loss": 2.9334, "step": 19251 }, { "epoch": 1.1951083245390774, "grad_norm": 0.6081476322208582, "learning_rate": 7.524667771327468e-05, "loss": 2.9638, "step": 19252 }, { "epoch": 1.1951704016388354, "grad_norm": 0.646642751972516, "learning_rate": 7.524356033065039e-05, "loss": 2.8304, "step": 19253 }, { "epoch": 1.1952324787385933, "grad_norm": 0.6286466074393331, "learning_rate": 7.52404428163243e-05, "loss": 2.9007, "step": 19254 }, { "epoch": 1.1952945558383512, "grad_norm": 0.653010319036762, "learning_rate": 7.52373251703127e-05, "loss": 2.8823, "step": 19255 }, { "epoch": 1.1953566329381091, "grad_norm": 0.6331304293114459, "learning_rate": 7.523420739263183e-05, "loss": 2.8644, "step": 19256 }, { "epoch": 1.195418710037867, "grad_norm": 0.527110040940446, "learning_rate": 7.5231089483298e-05, "loss": 2.8107, "step": 19257 }, { "epoch": 1.195480787137625, "grad_norm": 0.5564724316721819, "learning_rate": 7.522797144232742e-05, "loss": 2.7549, "step": 19258 }, { "epoch": 1.1955428642373829, "grad_norm": 0.5259724341682728, "learning_rate": 7.522485326973641e-05, "loss": 2.8526, "step": 19259 }, { "epoch": 1.1956049413371408, "grad_norm": 0.5340860623005103, "learning_rate": 7.522173496554123e-05, "loss": 2.892, "step": 19260 }, { "epoch": 1.1956670184368987, "grad_norm": 0.6089536481719048, "learning_rate": 7.521861652975811e-05, "loss": 2.7483, "step": 19261 }, { "epoch": 1.1957290955366564, "grad_norm": 0.5179982046092523, "learning_rate": 7.521549796240337e-05, "loss": 2.8753, "step": 19262 }, { "epoch": 1.1957911726364143, "grad_norm": 0.6812245757932125, "learning_rate": 7.521237926349322e-05, "loss": 2.9041, "step": 19263 }, { "epoch": 1.1958532497361722, "grad_norm": 0.5466317589544993, "learning_rate": 7.5209260433044e-05, "loss": 2.7938, "step": 19264 }, { "epoch": 1.1959153268359302, "grad_norm": 0.5666852592709212, "learning_rate": 7.520614147107193e-05, "loss": 2.813, "step": 19265 }, { "epoch": 1.195977403935688, "grad_norm": 0.6036487173341236, "learning_rate": 7.520302237759331e-05, "loss": 2.947, "step": 19266 }, { "epoch": 1.196039481035446, "grad_norm": 0.6114659260717605, "learning_rate": 7.51999031526244e-05, "loss": 2.8998, "step": 19267 }, { "epoch": 1.196101558135204, "grad_norm": 0.5159366706088243, "learning_rate": 7.519678379618149e-05, "loss": 2.9151, "step": 19268 }, { "epoch": 1.1961636352349618, "grad_norm": 0.6331860417033037, "learning_rate": 7.519366430828083e-05, "loss": 2.9156, "step": 19269 }, { "epoch": 1.1962257123347197, "grad_norm": 0.5465509817175842, "learning_rate": 7.51905446889387e-05, "loss": 2.8892, "step": 19270 }, { "epoch": 1.1962877894344777, "grad_norm": 0.5472916242262823, "learning_rate": 7.518742493817139e-05, "loss": 2.9297, "step": 19271 }, { "epoch": 1.1963498665342356, "grad_norm": 0.553417498088722, "learning_rate": 7.518430505599517e-05, "loss": 2.9265, "step": 19272 }, { "epoch": 1.1964119436339935, "grad_norm": 0.5019910924185802, "learning_rate": 7.518118504242633e-05, "loss": 2.853, "step": 19273 }, { "epoch": 1.1964740207337514, "grad_norm": 0.5189114939704075, "learning_rate": 7.517806489748111e-05, "loss": 2.906, "step": 19274 }, { "epoch": 1.1965360978335093, "grad_norm": 0.4866382115871168, "learning_rate": 7.517494462117584e-05, "loss": 2.7805, "step": 19275 }, { "epoch": 1.196598174933267, "grad_norm": 0.5690221690718099, "learning_rate": 7.517182421352675e-05, "loss": 2.7441, "step": 19276 }, { "epoch": 1.196660252033025, "grad_norm": 0.5296839463556414, "learning_rate": 7.516870367455016e-05, "loss": 2.8559, "step": 19277 }, { "epoch": 1.1967223291327829, "grad_norm": 0.5864043855218193, "learning_rate": 7.516558300426233e-05, "loss": 2.8859, "step": 19278 }, { "epoch": 1.1967844062325408, "grad_norm": 0.5855296941047271, "learning_rate": 7.516246220267953e-05, "loss": 2.9302, "step": 19279 }, { "epoch": 1.1968464833322987, "grad_norm": 0.5678595216183738, "learning_rate": 7.515934126981807e-05, "loss": 2.8582, "step": 19280 }, { "epoch": 1.1969085604320566, "grad_norm": 0.6348723101828372, "learning_rate": 7.515622020569421e-05, "loss": 2.8168, "step": 19281 }, { "epoch": 1.1969706375318145, "grad_norm": 0.6081953612942533, "learning_rate": 7.515309901032426e-05, "loss": 2.8779, "step": 19282 }, { "epoch": 1.1970327146315725, "grad_norm": 0.5829585068906271, "learning_rate": 7.514997768372447e-05, "loss": 2.8116, "step": 19283 }, { "epoch": 1.1970947917313304, "grad_norm": 0.5559811617520917, "learning_rate": 7.514685622591114e-05, "loss": 2.8886, "step": 19284 }, { "epoch": 1.1971568688310883, "grad_norm": 0.5937125043574144, "learning_rate": 7.514373463690057e-05, "loss": 2.8029, "step": 19285 }, { "epoch": 1.197218945930846, "grad_norm": 0.5342824443057005, "learning_rate": 7.514061291670899e-05, "loss": 2.8527, "step": 19286 }, { "epoch": 1.197281023030604, "grad_norm": 0.5916627536787564, "learning_rate": 7.513749106535278e-05, "loss": 2.8265, "step": 19287 }, { "epoch": 1.1973431001303618, "grad_norm": 0.5617676842269566, "learning_rate": 7.513436908284815e-05, "loss": 2.91, "step": 19288 }, { "epoch": 1.1974051772301197, "grad_norm": 0.5952975784394123, "learning_rate": 7.513124696921144e-05, "loss": 2.7804, "step": 19289 }, { "epoch": 1.1974672543298777, "grad_norm": 0.5964893093542069, "learning_rate": 7.512812472445889e-05, "loss": 2.8132, "step": 19290 }, { "epoch": 1.1975293314296356, "grad_norm": 0.5294536907254377, "learning_rate": 7.512500234860682e-05, "loss": 2.899, "step": 19291 }, { "epoch": 1.1975914085293935, "grad_norm": 0.5918448652191185, "learning_rate": 7.512187984167151e-05, "loss": 2.8701, "step": 19292 }, { "epoch": 1.1976534856291514, "grad_norm": 0.558754024320443, "learning_rate": 7.511875720366927e-05, "loss": 2.7971, "step": 19293 }, { "epoch": 1.1977155627289093, "grad_norm": 0.5351337790122164, "learning_rate": 7.511563443461636e-05, "loss": 2.8985, "step": 19294 }, { "epoch": 1.1977776398286673, "grad_norm": 0.5318044966287011, "learning_rate": 7.51125115345291e-05, "loss": 2.8585, "step": 19295 }, { "epoch": 1.1978397169284252, "grad_norm": 0.5480179695451434, "learning_rate": 7.510938850342376e-05, "loss": 2.8191, "step": 19296 }, { "epoch": 1.197901794028183, "grad_norm": 0.5837275472974984, "learning_rate": 7.510626534131664e-05, "loss": 2.8604, "step": 19297 }, { "epoch": 1.197963871127941, "grad_norm": 0.5248886887304532, "learning_rate": 7.510314204822406e-05, "loss": 2.9726, "step": 19298 }, { "epoch": 1.198025948227699, "grad_norm": 0.5890993385027602, "learning_rate": 7.510001862416227e-05, "loss": 2.7899, "step": 19299 }, { "epoch": 1.1980880253274566, "grad_norm": 0.5443141378548239, "learning_rate": 7.509689506914761e-05, "loss": 2.8315, "step": 19300 }, { "epoch": 1.1981501024272145, "grad_norm": 0.634856936416173, "learning_rate": 7.509377138319633e-05, "loss": 2.8712, "step": 19301 }, { "epoch": 1.1982121795269725, "grad_norm": 0.559484185585925, "learning_rate": 7.509064756632478e-05, "loss": 2.8448, "step": 19302 }, { "epoch": 1.1982742566267304, "grad_norm": 0.5645815165397813, "learning_rate": 7.508752361854923e-05, "loss": 2.7998, "step": 19303 }, { "epoch": 1.1983363337264883, "grad_norm": 0.48539825767460587, "learning_rate": 7.508439953988596e-05, "loss": 2.8678, "step": 19304 }, { "epoch": 1.1983984108262462, "grad_norm": 0.5699001546045251, "learning_rate": 7.50812753303513e-05, "loss": 2.8849, "step": 19305 }, { "epoch": 1.1984604879260041, "grad_norm": 0.5287032229336652, "learning_rate": 7.507815098996153e-05, "loss": 2.8832, "step": 19306 }, { "epoch": 1.198522565025762, "grad_norm": 0.6511209643072831, "learning_rate": 7.507502651873296e-05, "loss": 2.8239, "step": 19307 }, { "epoch": 1.19858464212552, "grad_norm": 0.5574315659042172, "learning_rate": 7.50719019166819e-05, "loss": 2.8684, "step": 19308 }, { "epoch": 1.1986467192252779, "grad_norm": 0.5511119472007451, "learning_rate": 7.506877718382464e-05, "loss": 2.8469, "step": 19309 }, { "epoch": 1.1987087963250356, "grad_norm": 0.6231562599138895, "learning_rate": 7.506565232017748e-05, "loss": 2.9552, "step": 19310 }, { "epoch": 1.1987708734247935, "grad_norm": 0.5780978325311412, "learning_rate": 7.506252732575672e-05, "loss": 2.7926, "step": 19311 }, { "epoch": 1.1988329505245514, "grad_norm": 0.5546646113441669, "learning_rate": 7.505940220057868e-05, "loss": 2.8317, "step": 19312 }, { "epoch": 1.1988950276243093, "grad_norm": 0.5781616508619672, "learning_rate": 7.505627694465964e-05, "loss": 2.8012, "step": 19313 }, { "epoch": 1.1989571047240672, "grad_norm": 0.6227726347740506, "learning_rate": 7.505315155801594e-05, "loss": 2.9514, "step": 19314 }, { "epoch": 1.1990191818238252, "grad_norm": 0.4896310536144507, "learning_rate": 7.505002604066386e-05, "loss": 2.8929, "step": 19315 }, { "epoch": 1.199081258923583, "grad_norm": 0.5673584399347862, "learning_rate": 7.504690039261971e-05, "loss": 2.8453, "step": 19316 }, { "epoch": 1.199143336023341, "grad_norm": 0.46566736361422545, "learning_rate": 7.50437746138998e-05, "loss": 2.7134, "step": 19317 }, { "epoch": 1.199205413123099, "grad_norm": 0.5862609733335391, "learning_rate": 7.504064870452044e-05, "loss": 2.9209, "step": 19318 }, { "epoch": 1.1992674902228568, "grad_norm": 0.5425262569190914, "learning_rate": 7.503752266449792e-05, "loss": 2.9467, "step": 19319 }, { "epoch": 1.1993295673226148, "grad_norm": 0.5797604291338916, "learning_rate": 7.50343964938486e-05, "loss": 2.8443, "step": 19320 }, { "epoch": 1.1993916444223727, "grad_norm": 0.5365022314190566, "learning_rate": 7.503127019258873e-05, "loss": 2.9317, "step": 19321 }, { "epoch": 1.1994537215221306, "grad_norm": 0.5617873127441421, "learning_rate": 7.502814376073465e-05, "loss": 2.8283, "step": 19322 }, { "epoch": 1.1995157986218885, "grad_norm": 0.5400655784788487, "learning_rate": 7.502501719830269e-05, "loss": 2.7996, "step": 19323 }, { "epoch": 1.1995778757216462, "grad_norm": 0.5826326629701088, "learning_rate": 7.502189050530913e-05, "loss": 2.806, "step": 19324 }, { "epoch": 1.1996399528214041, "grad_norm": 0.5758203242176236, "learning_rate": 7.501876368177028e-05, "loss": 2.8238, "step": 19325 }, { "epoch": 1.199702029921162, "grad_norm": 0.5555185620256542, "learning_rate": 7.501563672770246e-05, "loss": 2.8197, "step": 19326 }, { "epoch": 1.19976410702092, "grad_norm": 0.6486749558971171, "learning_rate": 7.501250964312202e-05, "loss": 2.8679, "step": 19327 }, { "epoch": 1.1998261841206779, "grad_norm": 0.6122536037131843, "learning_rate": 7.500938242804523e-05, "loss": 2.8285, "step": 19328 }, { "epoch": 1.1998882612204358, "grad_norm": 0.5762248645273719, "learning_rate": 7.500625508248842e-05, "loss": 2.8129, "step": 19329 }, { "epoch": 1.1999503383201937, "grad_norm": 0.6749786726796542, "learning_rate": 7.50031276064679e-05, "loss": 2.9103, "step": 19330 }, { "epoch": 1.2000124154199516, "grad_norm": 0.5706863484407044, "learning_rate": 7.500000000000001e-05, "loss": 2.8469, "step": 19331 }, { "epoch": 1.2000744925197095, "grad_norm": 0.5845664195330352, "learning_rate": 7.499687226310103e-05, "loss": 2.8618, "step": 19332 }, { "epoch": 1.2001365696194675, "grad_norm": 0.6764460694736775, "learning_rate": 7.499374439578733e-05, "loss": 2.7934, "step": 19333 }, { "epoch": 1.2001986467192252, "grad_norm": 0.5409443391734508, "learning_rate": 7.499061639807516e-05, "loss": 2.9, "step": 19334 }, { "epoch": 1.200260723818983, "grad_norm": 0.6378187495312809, "learning_rate": 7.49874882699809e-05, "loss": 2.8405, "step": 19335 }, { "epoch": 1.200322800918741, "grad_norm": 0.5664849325315936, "learning_rate": 7.498436001152085e-05, "loss": 2.7819, "step": 19336 }, { "epoch": 1.200384878018499, "grad_norm": 0.6051535001023011, "learning_rate": 7.49812316227113e-05, "loss": 2.8405, "step": 19337 }, { "epoch": 1.2004469551182568, "grad_norm": 0.5644477659188855, "learning_rate": 7.497810310356864e-05, "loss": 2.9166, "step": 19338 }, { "epoch": 1.2005090322180147, "grad_norm": 0.5161237038937696, "learning_rate": 7.497497445410912e-05, "loss": 2.782, "step": 19339 }, { "epoch": 1.2005711093177727, "grad_norm": 0.5213391105263866, "learning_rate": 7.497184567434912e-05, "loss": 2.8392, "step": 19340 }, { "epoch": 1.2006331864175306, "grad_norm": 0.5735839137120072, "learning_rate": 7.496871676430495e-05, "loss": 2.8475, "step": 19341 }, { "epoch": 1.2006952635172885, "grad_norm": 0.5434021900676221, "learning_rate": 7.49655877239929e-05, "loss": 2.8062, "step": 19342 }, { "epoch": 1.2007573406170464, "grad_norm": 0.5630295962470147, "learning_rate": 7.496245855342931e-05, "loss": 2.7008, "step": 19343 }, { "epoch": 1.2008194177168043, "grad_norm": 0.5813499662102525, "learning_rate": 7.495932925263053e-05, "loss": 2.8826, "step": 19344 }, { "epoch": 1.2008814948165623, "grad_norm": 0.5864704251666132, "learning_rate": 7.495619982161287e-05, "loss": 2.7976, "step": 19345 }, { "epoch": 1.2009435719163202, "grad_norm": 0.4611283569896232, "learning_rate": 7.495307026039265e-05, "loss": 2.8559, "step": 19346 }, { "epoch": 1.201005649016078, "grad_norm": 0.6062898544110713, "learning_rate": 7.494994056898621e-05, "loss": 2.9271, "step": 19347 }, { "epoch": 1.2010677261158358, "grad_norm": 0.48314700282577655, "learning_rate": 7.494681074740985e-05, "loss": 2.8241, "step": 19348 }, { "epoch": 1.2011298032155937, "grad_norm": 0.531947379426465, "learning_rate": 7.494368079567997e-05, "loss": 2.8107, "step": 19349 }, { "epoch": 1.2011918803153516, "grad_norm": 0.5421555006714776, "learning_rate": 7.49405507138128e-05, "loss": 2.8675, "step": 19350 }, { "epoch": 1.2012539574151095, "grad_norm": 0.5022259872301493, "learning_rate": 7.493742050182476e-05, "loss": 2.852, "step": 19351 }, { "epoch": 1.2013160345148675, "grad_norm": 0.544043129801277, "learning_rate": 7.493429015973212e-05, "loss": 2.9192, "step": 19352 }, { "epoch": 1.2013781116146254, "grad_norm": 0.5424983504925099, "learning_rate": 7.493115968755126e-05, "loss": 2.8029, "step": 19353 }, { "epoch": 1.2014401887143833, "grad_norm": 0.5201062054106762, "learning_rate": 7.492802908529847e-05, "loss": 2.8158, "step": 19354 }, { "epoch": 1.2015022658141412, "grad_norm": 0.5291394495785549, "learning_rate": 7.49248983529901e-05, "loss": 2.8821, "step": 19355 }, { "epoch": 1.2015643429138991, "grad_norm": 0.5193588716594093, "learning_rate": 7.492176749064249e-05, "loss": 2.8464, "step": 19356 }, { "epoch": 1.201626420013657, "grad_norm": 0.5028290879457367, "learning_rate": 7.491863649827197e-05, "loss": 2.8239, "step": 19357 }, { "epoch": 1.2016884971134147, "grad_norm": 0.5986921769495195, "learning_rate": 7.491550537589487e-05, "loss": 2.8398, "step": 19358 }, { "epoch": 1.2017505742131727, "grad_norm": 0.44356925130809527, "learning_rate": 7.491237412352754e-05, "loss": 2.7779, "step": 19359 }, { "epoch": 1.2018126513129306, "grad_norm": 0.5321092668823616, "learning_rate": 7.490924274118629e-05, "loss": 2.8448, "step": 19360 }, { "epoch": 1.2018747284126885, "grad_norm": 0.5480214227923257, "learning_rate": 7.490611122888749e-05, "loss": 2.8264, "step": 19361 }, { "epoch": 1.2019368055124464, "grad_norm": 0.509254286802786, "learning_rate": 7.490297958664744e-05, "loss": 2.8242, "step": 19362 }, { "epoch": 1.2019988826122043, "grad_norm": 0.5323432164571612, "learning_rate": 7.489984781448253e-05, "loss": 2.8846, "step": 19363 }, { "epoch": 1.2020609597119623, "grad_norm": 0.6166083117101471, "learning_rate": 7.489671591240905e-05, "loss": 2.8402, "step": 19364 }, { "epoch": 1.2021230368117202, "grad_norm": 0.5088310297609713, "learning_rate": 7.489358388044336e-05, "loss": 2.8303, "step": 19365 }, { "epoch": 1.202185113911478, "grad_norm": 0.6152636324268524, "learning_rate": 7.489045171860181e-05, "loss": 2.9026, "step": 19366 }, { "epoch": 1.202247191011236, "grad_norm": 0.4754265238012786, "learning_rate": 7.488731942690072e-05, "loss": 2.8527, "step": 19367 }, { "epoch": 1.202309268110994, "grad_norm": 0.5769692028080862, "learning_rate": 7.488418700535644e-05, "loss": 2.8589, "step": 19368 }, { "epoch": 1.2023713452107518, "grad_norm": 0.5325655356451483, "learning_rate": 7.488105445398532e-05, "loss": 2.9283, "step": 19369 }, { "epoch": 1.2024334223105098, "grad_norm": 0.5479565144204287, "learning_rate": 7.487792177280371e-05, "loss": 2.909, "step": 19370 }, { "epoch": 1.2024954994102677, "grad_norm": 0.4930744807397617, "learning_rate": 7.487478896182792e-05, "loss": 2.936, "step": 19371 }, { "epoch": 1.2025575765100254, "grad_norm": 0.5495335942170375, "learning_rate": 7.487165602107435e-05, "loss": 2.9632, "step": 19372 }, { "epoch": 1.2026196536097833, "grad_norm": 0.493833242714846, "learning_rate": 7.486852295055928e-05, "loss": 2.889, "step": 19373 }, { "epoch": 1.2026817307095412, "grad_norm": 0.5104213312156486, "learning_rate": 7.486538975029911e-05, "loss": 2.7493, "step": 19374 }, { "epoch": 1.2027438078092991, "grad_norm": 0.5492680642791097, "learning_rate": 7.486225642031015e-05, "loss": 2.8771, "step": 19375 }, { "epoch": 1.202805884909057, "grad_norm": 0.5048773228477845, "learning_rate": 7.485912296060877e-05, "loss": 2.8566, "step": 19376 }, { "epoch": 1.202867962008815, "grad_norm": 0.549375891961425, "learning_rate": 7.485598937121133e-05, "loss": 2.8302, "step": 19377 }, { "epoch": 1.2029300391085729, "grad_norm": 0.6127470812391462, "learning_rate": 7.485285565213413e-05, "loss": 2.7979, "step": 19378 }, { "epoch": 1.2029921162083308, "grad_norm": 0.5407437629569285, "learning_rate": 7.484972180339356e-05, "loss": 2.9151, "step": 19379 }, { "epoch": 1.2030541933080887, "grad_norm": 0.576206374347838, "learning_rate": 7.484658782500595e-05, "loss": 2.8795, "step": 19380 }, { "epoch": 1.2031162704078466, "grad_norm": 0.5663058224797876, "learning_rate": 7.484345371698767e-05, "loss": 2.7764, "step": 19381 }, { "epoch": 1.2031783475076043, "grad_norm": 0.5797682939762787, "learning_rate": 7.484031947935506e-05, "loss": 2.8124, "step": 19382 }, { "epoch": 1.2032404246073622, "grad_norm": 0.5976308680112431, "learning_rate": 7.483718511212447e-05, "loss": 2.8319, "step": 19383 }, { "epoch": 1.2033025017071202, "grad_norm": 0.6156229880827738, "learning_rate": 7.483405061531227e-05, "loss": 2.8471, "step": 19384 }, { "epoch": 1.203364578806878, "grad_norm": 0.5571488854104842, "learning_rate": 7.483091598893478e-05, "loss": 2.7947, "step": 19385 }, { "epoch": 1.203426655906636, "grad_norm": 0.619673250143214, "learning_rate": 7.482778123300837e-05, "loss": 2.9127, "step": 19386 }, { "epoch": 1.203488733006394, "grad_norm": 0.49995124102313593, "learning_rate": 7.482464634754942e-05, "loss": 2.9097, "step": 19387 }, { "epoch": 1.2035508101061518, "grad_norm": 0.5138312553042448, "learning_rate": 7.482151133257426e-05, "loss": 2.736, "step": 19388 }, { "epoch": 1.2036128872059098, "grad_norm": 0.49190935722578016, "learning_rate": 7.481837618809924e-05, "loss": 2.7862, "step": 19389 }, { "epoch": 1.2036749643056677, "grad_norm": 0.5477823631932094, "learning_rate": 7.481524091414072e-05, "loss": 2.8973, "step": 19390 }, { "epoch": 1.2037370414054256, "grad_norm": 0.488296859491707, "learning_rate": 7.481210551071508e-05, "loss": 2.7854, "step": 19391 }, { "epoch": 1.2037991185051835, "grad_norm": 0.5708441150881508, "learning_rate": 7.480896997783866e-05, "loss": 2.7778, "step": 19392 }, { "epoch": 1.2038611956049414, "grad_norm": 0.5106608397140594, "learning_rate": 7.48058343155278e-05, "loss": 2.924, "step": 19393 }, { "epoch": 1.2039232727046993, "grad_norm": 0.543015347683633, "learning_rate": 7.480269852379891e-05, "loss": 2.894, "step": 19394 }, { "epoch": 1.203985349804457, "grad_norm": 0.5753207887532126, "learning_rate": 7.479956260266831e-05, "loss": 2.8559, "step": 19395 }, { "epoch": 1.204047426904215, "grad_norm": 0.5570421343538526, "learning_rate": 7.479642655215236e-05, "loss": 2.8646, "step": 19396 }, { "epoch": 1.2041095040039729, "grad_norm": 0.611089470930536, "learning_rate": 7.479329037226745e-05, "loss": 2.9409, "step": 19397 }, { "epoch": 1.2041715811037308, "grad_norm": 0.5987232880417626, "learning_rate": 7.479015406302992e-05, "loss": 2.8568, "step": 19398 }, { "epoch": 1.2042336582034887, "grad_norm": 0.5239097969624551, "learning_rate": 7.478701762445612e-05, "loss": 2.8198, "step": 19399 }, { "epoch": 1.2042957353032466, "grad_norm": 0.5630953869670148, "learning_rate": 7.478388105656244e-05, "loss": 2.807, "step": 19400 }, { "epoch": 1.2043578124030045, "grad_norm": 0.5908534017077082, "learning_rate": 7.478074435936526e-05, "loss": 2.8275, "step": 19401 }, { "epoch": 1.2044198895027625, "grad_norm": 0.5472822852529243, "learning_rate": 7.47776075328809e-05, "loss": 2.8785, "step": 19402 }, { "epoch": 1.2044819666025204, "grad_norm": 0.5571822626506382, "learning_rate": 7.477447057712574e-05, "loss": 2.8325, "step": 19403 }, { "epoch": 1.2045440437022783, "grad_norm": 0.5553007631022195, "learning_rate": 7.477133349211619e-05, "loss": 2.9047, "step": 19404 }, { "epoch": 1.2046061208020362, "grad_norm": 0.5739496859287333, "learning_rate": 7.476819627786852e-05, "loss": 2.8467, "step": 19405 }, { "epoch": 1.204668197901794, "grad_norm": 0.5040491182522194, "learning_rate": 7.476505893439921e-05, "loss": 2.7704, "step": 19406 }, { "epoch": 1.2047302750015518, "grad_norm": 0.5375183213238592, "learning_rate": 7.476192146172455e-05, "loss": 2.8905, "step": 19407 }, { "epoch": 1.2047923521013097, "grad_norm": 0.5490208351944814, "learning_rate": 7.475878385986095e-05, "loss": 2.8803, "step": 19408 }, { "epoch": 1.2048544292010677, "grad_norm": 0.5632828986240339, "learning_rate": 7.475564612882476e-05, "loss": 2.7773, "step": 19409 }, { "epoch": 1.2049165063008256, "grad_norm": 0.5909194232630887, "learning_rate": 7.475250826863233e-05, "loss": 2.8757, "step": 19410 }, { "epoch": 1.2049785834005835, "grad_norm": 0.4932630685060137, "learning_rate": 7.474937027930009e-05, "loss": 2.7899, "step": 19411 }, { "epoch": 1.2050406605003414, "grad_norm": 0.6276937133305297, "learning_rate": 7.474623216084436e-05, "loss": 2.8336, "step": 19412 }, { "epoch": 1.2051027376000993, "grad_norm": 0.5391163729962054, "learning_rate": 7.474309391328154e-05, "loss": 2.8017, "step": 19413 }, { "epoch": 1.2051648146998573, "grad_norm": 0.6353623856878269, "learning_rate": 7.473995553662798e-05, "loss": 2.8927, "step": 19414 }, { "epoch": 1.2052268917996152, "grad_norm": 0.5760896268427533, "learning_rate": 7.473681703090007e-05, "loss": 2.8764, "step": 19415 }, { "epoch": 1.205288968899373, "grad_norm": 0.6086342294964296, "learning_rate": 7.473367839611416e-05, "loss": 2.9178, "step": 19416 }, { "epoch": 1.205351045999131, "grad_norm": 0.5619042473586507, "learning_rate": 7.473053963228668e-05, "loss": 2.8763, "step": 19417 }, { "epoch": 1.205413123098889, "grad_norm": 0.6111212432146617, "learning_rate": 7.472740073943395e-05, "loss": 2.8921, "step": 19418 }, { "epoch": 1.2054752001986466, "grad_norm": 0.5317318918564127, "learning_rate": 7.472426171757239e-05, "loss": 2.849, "step": 19419 }, { "epoch": 1.2055372772984045, "grad_norm": 0.5375017432251097, "learning_rate": 7.472112256671832e-05, "loss": 2.9095, "step": 19420 }, { "epoch": 1.2055993543981625, "grad_norm": 0.5592482673676159, "learning_rate": 7.471798328688819e-05, "loss": 2.7993, "step": 19421 }, { "epoch": 1.2056614314979204, "grad_norm": 0.5216315415422649, "learning_rate": 7.471484387809832e-05, "loss": 2.8663, "step": 19422 }, { "epoch": 1.2057235085976783, "grad_norm": 0.5782050515503442, "learning_rate": 7.471170434036511e-05, "loss": 2.8774, "step": 19423 }, { "epoch": 1.2057855856974362, "grad_norm": 0.45874416477535584, "learning_rate": 7.470856467370493e-05, "loss": 2.863, "step": 19424 }, { "epoch": 1.2058476627971941, "grad_norm": 0.6478851447660791, "learning_rate": 7.470542487813419e-05, "loss": 2.8734, "step": 19425 }, { "epoch": 1.205909739896952, "grad_norm": 0.49591499772728764, "learning_rate": 7.470228495366924e-05, "loss": 2.7865, "step": 19426 }, { "epoch": 1.20597181699671, "grad_norm": 0.5390358517459375, "learning_rate": 7.469914490032648e-05, "loss": 2.788, "step": 19427 }, { "epoch": 1.2060338940964679, "grad_norm": 0.5472569358075969, "learning_rate": 7.469600471812229e-05, "loss": 2.8238, "step": 19428 }, { "epoch": 1.2060959711962258, "grad_norm": 0.5867156901070238, "learning_rate": 7.469286440707303e-05, "loss": 2.8995, "step": 19429 }, { "epoch": 1.2061580482959835, "grad_norm": 0.6068337589319349, "learning_rate": 7.468972396719512e-05, "loss": 2.9442, "step": 19430 }, { "epoch": 1.2062201253957414, "grad_norm": 0.6192742092296148, "learning_rate": 7.468658339850492e-05, "loss": 2.8719, "step": 19431 }, { "epoch": 1.2062822024954993, "grad_norm": 0.5117576234917757, "learning_rate": 7.468344270101879e-05, "loss": 2.8586, "step": 19432 }, { "epoch": 1.2063442795952573, "grad_norm": 0.6053720212394845, "learning_rate": 7.468030187475319e-05, "loss": 2.8001, "step": 19433 }, { "epoch": 1.2064063566950152, "grad_norm": 0.5323015352691749, "learning_rate": 7.467716091972445e-05, "loss": 2.855, "step": 19434 }, { "epoch": 1.206468433794773, "grad_norm": 0.5413149064418719, "learning_rate": 7.467401983594897e-05, "loss": 2.9034, "step": 19435 }, { "epoch": 1.206530510894531, "grad_norm": 0.5140801029611919, "learning_rate": 7.467087862344313e-05, "loss": 2.7621, "step": 19436 }, { "epoch": 1.206592587994289, "grad_norm": 0.5345994879357298, "learning_rate": 7.466773728222333e-05, "loss": 2.7777, "step": 19437 }, { "epoch": 1.2066546650940468, "grad_norm": 0.5112884109729571, "learning_rate": 7.466459581230599e-05, "loss": 2.8408, "step": 19438 }, { "epoch": 1.2067167421938048, "grad_norm": 0.5386311283980201, "learning_rate": 7.466145421370743e-05, "loss": 2.8957, "step": 19439 }, { "epoch": 1.2067788192935627, "grad_norm": 0.48491289666703186, "learning_rate": 7.465831248644409e-05, "loss": 2.9456, "step": 19440 }, { "epoch": 1.2068408963933206, "grad_norm": 0.5283509984156891, "learning_rate": 7.465517063053234e-05, "loss": 2.8858, "step": 19441 }, { "epoch": 1.2069029734930785, "grad_norm": 0.5440832267989381, "learning_rate": 7.465202864598858e-05, "loss": 2.8388, "step": 19442 }, { "epoch": 1.2069650505928362, "grad_norm": 0.5406017166151368, "learning_rate": 7.46488865328292e-05, "loss": 2.8087, "step": 19443 }, { "epoch": 1.2070271276925941, "grad_norm": 0.5500242834799228, "learning_rate": 7.46457442910706e-05, "loss": 2.8385, "step": 19444 }, { "epoch": 1.207089204792352, "grad_norm": 0.6248679259942013, "learning_rate": 7.464260192072917e-05, "loss": 2.8315, "step": 19445 }, { "epoch": 1.20715128189211, "grad_norm": 0.549784170415014, "learning_rate": 7.463945942182131e-05, "loss": 2.9167, "step": 19446 }, { "epoch": 1.2072133589918679, "grad_norm": 0.5438135833396244, "learning_rate": 7.463631679436339e-05, "loss": 2.9482, "step": 19447 }, { "epoch": 1.2072754360916258, "grad_norm": 0.5554429940262857, "learning_rate": 7.463317403837185e-05, "loss": 2.8319, "step": 19448 }, { "epoch": 1.2073375131913837, "grad_norm": 0.48291984525242815, "learning_rate": 7.463003115386304e-05, "loss": 2.7512, "step": 19449 }, { "epoch": 1.2073995902911416, "grad_norm": 0.514113117033391, "learning_rate": 7.462688814085338e-05, "loss": 2.8237, "step": 19450 }, { "epoch": 1.2074616673908996, "grad_norm": 0.5473946963031355, "learning_rate": 7.462374499935928e-05, "loss": 2.8301, "step": 19451 }, { "epoch": 1.2075237444906575, "grad_norm": 0.5558504351864094, "learning_rate": 7.46206017293971e-05, "loss": 2.9929, "step": 19452 }, { "epoch": 1.2075858215904154, "grad_norm": 0.5346711180433078, "learning_rate": 7.46174583309833e-05, "loss": 2.9669, "step": 19453 }, { "epoch": 1.207647898690173, "grad_norm": 0.5413994924941008, "learning_rate": 7.46143148041342e-05, "loss": 2.8177, "step": 19454 }, { "epoch": 1.207709975789931, "grad_norm": 0.5718048896962752, "learning_rate": 7.461117114886629e-05, "loss": 2.8641, "step": 19455 }, { "epoch": 1.207772052889689, "grad_norm": 0.47005092976464685, "learning_rate": 7.46080273651959e-05, "loss": 2.8441, "step": 19456 }, { "epoch": 1.2078341299894468, "grad_norm": 0.5732373524970447, "learning_rate": 7.460488345313947e-05, "loss": 2.8811, "step": 19457 }, { "epoch": 1.2078962070892048, "grad_norm": 0.6402748825916178, "learning_rate": 7.460173941271338e-05, "loss": 2.8626, "step": 19458 }, { "epoch": 1.2079582841889627, "grad_norm": 0.61332457079623, "learning_rate": 7.459859524393404e-05, "loss": 2.9344, "step": 19459 }, { "epoch": 1.2080203612887206, "grad_norm": 0.5553037417123105, "learning_rate": 7.459545094681785e-05, "loss": 2.9411, "step": 19460 }, { "epoch": 1.2080824383884785, "grad_norm": 0.5575050735670428, "learning_rate": 7.459230652138124e-05, "loss": 2.8353, "step": 19461 }, { "epoch": 1.2081445154882364, "grad_norm": 0.5082723442866475, "learning_rate": 7.45891619676406e-05, "loss": 2.8478, "step": 19462 }, { "epoch": 1.2082065925879943, "grad_norm": 0.6050935875017156, "learning_rate": 7.458601728561232e-05, "loss": 2.8636, "step": 19463 }, { "epoch": 1.2082686696877523, "grad_norm": 0.5391657357254521, "learning_rate": 7.458287247531282e-05, "loss": 2.8446, "step": 19464 }, { "epoch": 1.2083307467875102, "grad_norm": 0.5842316783240763, "learning_rate": 7.45797275367585e-05, "loss": 2.879, "step": 19465 }, { "epoch": 1.208392823887268, "grad_norm": 0.6008225633879025, "learning_rate": 7.457658246996579e-05, "loss": 2.812, "step": 19466 }, { "epoch": 1.2084549009870258, "grad_norm": 0.532903733930429, "learning_rate": 7.457343727495108e-05, "loss": 2.851, "step": 19467 }, { "epoch": 1.2085169780867837, "grad_norm": 0.612429203106882, "learning_rate": 7.457029195173078e-05, "loss": 2.7997, "step": 19468 }, { "epoch": 1.2085790551865416, "grad_norm": 0.48681095774518063, "learning_rate": 7.45671465003213e-05, "loss": 2.8141, "step": 19469 }, { "epoch": 1.2086411322862995, "grad_norm": 0.5852141938595664, "learning_rate": 7.456400092073906e-05, "loss": 2.9425, "step": 19470 }, { "epoch": 1.2087032093860575, "grad_norm": 0.5468402579034902, "learning_rate": 7.456085521300045e-05, "loss": 2.9424, "step": 19471 }, { "epoch": 1.2087652864858154, "grad_norm": 0.5752540130972605, "learning_rate": 7.45577093771219e-05, "loss": 2.9003, "step": 19472 }, { "epoch": 1.2088273635855733, "grad_norm": 0.5230368319082104, "learning_rate": 7.455456341311983e-05, "loss": 2.8931, "step": 19473 }, { "epoch": 1.2088894406853312, "grad_norm": 0.5937919601872631, "learning_rate": 7.455141732101063e-05, "loss": 2.8148, "step": 19474 }, { "epoch": 1.2089515177850891, "grad_norm": 0.4828329921435247, "learning_rate": 7.454827110081072e-05, "loss": 2.8376, "step": 19475 }, { "epoch": 1.209013594884847, "grad_norm": 0.544773751639581, "learning_rate": 7.454512475253655e-05, "loss": 2.8748, "step": 19476 }, { "epoch": 1.209075671984605, "grad_norm": 0.516365646405023, "learning_rate": 7.454197827620449e-05, "loss": 2.8023, "step": 19477 }, { "epoch": 1.2091377490843627, "grad_norm": 0.5207290608613966, "learning_rate": 7.453883167183096e-05, "loss": 2.8729, "step": 19478 }, { "epoch": 1.2091998261841206, "grad_norm": 0.53422342571166, "learning_rate": 7.453568493943239e-05, "loss": 2.8366, "step": 19479 }, { "epoch": 1.2092619032838785, "grad_norm": 0.5841607940485664, "learning_rate": 7.453253807902521e-05, "loss": 2.9265, "step": 19480 }, { "epoch": 1.2093239803836364, "grad_norm": 0.5555882460857265, "learning_rate": 7.452939109062582e-05, "loss": 2.7823, "step": 19481 }, { "epoch": 1.2093860574833943, "grad_norm": 0.6098126282742158, "learning_rate": 7.452624397425062e-05, "loss": 2.8635, "step": 19482 }, { "epoch": 1.2094481345831523, "grad_norm": 0.5545054123427728, "learning_rate": 7.452309672991607e-05, "loss": 2.7818, "step": 19483 }, { "epoch": 1.2095102116829102, "grad_norm": 0.5358296633407343, "learning_rate": 7.451994935763856e-05, "loss": 2.8549, "step": 19484 }, { "epoch": 1.209572288782668, "grad_norm": 0.5775437913119889, "learning_rate": 7.451680185743455e-05, "loss": 2.8422, "step": 19485 }, { "epoch": 1.209634365882426, "grad_norm": 0.5308897536050817, "learning_rate": 7.45136542293204e-05, "loss": 2.8259, "step": 19486 }, { "epoch": 1.209696442982184, "grad_norm": 0.6425837832494516, "learning_rate": 7.451050647331259e-05, "loss": 2.8917, "step": 19487 }, { "epoch": 1.2097585200819418, "grad_norm": 0.5776840538822653, "learning_rate": 7.45073585894275e-05, "loss": 2.7718, "step": 19488 }, { "epoch": 1.2098205971816998, "grad_norm": 0.6414270962151407, "learning_rate": 7.450421057768157e-05, "loss": 2.8843, "step": 19489 }, { "epoch": 1.2098826742814577, "grad_norm": 0.601260072064648, "learning_rate": 7.450106243809123e-05, "loss": 2.9922, "step": 19490 }, { "epoch": 1.2099447513812154, "grad_norm": 0.7567921878106026, "learning_rate": 7.44979141706729e-05, "loss": 2.8667, "step": 19491 }, { "epoch": 1.2100068284809733, "grad_norm": 0.47539105818892513, "learning_rate": 7.449476577544301e-05, "loss": 2.81, "step": 19492 }, { "epoch": 1.2100689055807312, "grad_norm": 0.6171873309944018, "learning_rate": 7.449161725241799e-05, "loss": 2.9257, "step": 19493 }, { "epoch": 1.2101309826804891, "grad_norm": 0.6016064912638581, "learning_rate": 7.448846860161425e-05, "loss": 2.801, "step": 19494 }, { "epoch": 1.210193059780247, "grad_norm": 0.46645441621285993, "learning_rate": 7.44853198230482e-05, "loss": 2.7788, "step": 19495 }, { "epoch": 1.210255136880005, "grad_norm": 0.6407673840346881, "learning_rate": 7.448217091673633e-05, "loss": 2.7885, "step": 19496 }, { "epoch": 1.2103172139797629, "grad_norm": 0.5421419778671688, "learning_rate": 7.4479021882695e-05, "loss": 2.896, "step": 19497 }, { "epoch": 1.2103792910795208, "grad_norm": 0.5744637725602825, "learning_rate": 7.447587272094067e-05, "loss": 2.8858, "step": 19498 }, { "epoch": 1.2104413681792787, "grad_norm": 0.6288765610031198, "learning_rate": 7.44727234314898e-05, "loss": 2.859, "step": 19499 }, { "epoch": 1.2105034452790366, "grad_norm": 0.5729526392885155, "learning_rate": 7.446957401435877e-05, "loss": 2.9141, "step": 19500 }, { "epoch": 1.2105655223787946, "grad_norm": 0.6450136864489044, "learning_rate": 7.446642446956405e-05, "loss": 2.9146, "step": 19501 }, { "epoch": 1.2106275994785523, "grad_norm": 0.6806162703823878, "learning_rate": 7.446327479712202e-05, "loss": 2.8248, "step": 19502 }, { "epoch": 1.2106896765783102, "grad_norm": 0.5490771505868737, "learning_rate": 7.446012499704917e-05, "loss": 2.8286, "step": 19503 }, { "epoch": 1.210751753678068, "grad_norm": 0.6700347049108638, "learning_rate": 7.445697506936192e-05, "loss": 2.9176, "step": 19504 }, { "epoch": 1.210813830777826, "grad_norm": 0.6353958288017413, "learning_rate": 7.445382501407667e-05, "loss": 2.8515, "step": 19505 }, { "epoch": 1.210875907877584, "grad_norm": 0.5583514508349198, "learning_rate": 7.445067483120988e-05, "loss": 2.7877, "step": 19506 }, { "epoch": 1.2109379849773418, "grad_norm": 0.6604927153579084, "learning_rate": 7.444752452077801e-05, "loss": 2.8475, "step": 19507 }, { "epoch": 1.2110000620770998, "grad_norm": 0.5236982009510069, "learning_rate": 7.444437408279745e-05, "loss": 2.9372, "step": 19508 }, { "epoch": 1.2110621391768577, "grad_norm": 0.6801296418390389, "learning_rate": 7.444122351728467e-05, "loss": 2.7225, "step": 19509 }, { "epoch": 1.2111242162766156, "grad_norm": 0.5472019852878294, "learning_rate": 7.443807282425606e-05, "loss": 2.8676, "step": 19510 }, { "epoch": 1.2111862933763735, "grad_norm": 0.5780589607254979, "learning_rate": 7.443492200372813e-05, "loss": 2.7545, "step": 19511 }, { "epoch": 1.2112483704761314, "grad_norm": 0.5123872830718723, "learning_rate": 7.443177105571726e-05, "loss": 2.8363, "step": 19512 }, { "epoch": 1.2113104475758893, "grad_norm": 0.5021381259899592, "learning_rate": 7.442861998023992e-05, "loss": 2.8379, "step": 19513 }, { "epoch": 1.2113725246756473, "grad_norm": 0.5092933776613673, "learning_rate": 7.442546877731252e-05, "loss": 2.8269, "step": 19514 }, { "epoch": 1.211434601775405, "grad_norm": 0.49888989776278275, "learning_rate": 7.442231744695152e-05, "loss": 2.8347, "step": 19515 }, { "epoch": 1.2114966788751629, "grad_norm": 0.5688754226391303, "learning_rate": 7.44191659891734e-05, "loss": 2.881, "step": 19516 }, { "epoch": 1.2115587559749208, "grad_norm": 0.46559691120397995, "learning_rate": 7.441601440399453e-05, "loss": 2.8113, "step": 19517 }, { "epoch": 1.2116208330746787, "grad_norm": 0.5361124392670347, "learning_rate": 7.44128626914314e-05, "loss": 2.828, "step": 19518 }, { "epoch": 1.2116829101744366, "grad_norm": 0.5114363553620468, "learning_rate": 7.440971085150042e-05, "loss": 2.8802, "step": 19519 }, { "epoch": 1.2117449872741946, "grad_norm": 0.4953199945598837, "learning_rate": 7.440655888421806e-05, "loss": 2.821, "step": 19520 }, { "epoch": 1.2118070643739525, "grad_norm": 0.5104361906664242, "learning_rate": 7.440340678960077e-05, "loss": 2.8601, "step": 19521 }, { "epoch": 1.2118691414737104, "grad_norm": 0.5093483806745762, "learning_rate": 7.440025456766497e-05, "loss": 2.7742, "step": 19522 }, { "epoch": 1.2119312185734683, "grad_norm": 0.49733560673515376, "learning_rate": 7.439710221842712e-05, "loss": 2.7749, "step": 19523 }, { "epoch": 1.2119932956732262, "grad_norm": 0.6077686558774462, "learning_rate": 7.439394974190368e-05, "loss": 2.7898, "step": 19524 }, { "epoch": 1.2120553727729841, "grad_norm": 0.5373171040505337, "learning_rate": 7.439079713811106e-05, "loss": 2.9318, "step": 19525 }, { "epoch": 1.2121174498727418, "grad_norm": 0.540306811327577, "learning_rate": 7.438764440706574e-05, "loss": 2.9522, "step": 19526 }, { "epoch": 1.2121795269724998, "grad_norm": 0.6670008333464054, "learning_rate": 7.438449154878418e-05, "loss": 2.9215, "step": 19527 }, { "epoch": 1.2122416040722577, "grad_norm": 0.6244300627843509, "learning_rate": 7.438133856328277e-05, "loss": 2.8038, "step": 19528 }, { "epoch": 1.2123036811720156, "grad_norm": 0.5591078306683759, "learning_rate": 7.437818545057802e-05, "loss": 2.8162, "step": 19529 }, { "epoch": 1.2123657582717735, "grad_norm": 0.5483959845237996, "learning_rate": 7.437503221068635e-05, "loss": 2.8019, "step": 19530 }, { "epoch": 1.2124278353715314, "grad_norm": 0.5594839458821356, "learning_rate": 7.437187884362423e-05, "loss": 2.8958, "step": 19531 }, { "epoch": 1.2124899124712893, "grad_norm": 0.5414937794927751, "learning_rate": 7.436872534940809e-05, "loss": 2.9553, "step": 19532 }, { "epoch": 1.2125519895710473, "grad_norm": 0.5309085870855609, "learning_rate": 7.43655717280544e-05, "loss": 2.8413, "step": 19533 }, { "epoch": 1.2126140666708052, "grad_norm": 0.5866236674425527, "learning_rate": 7.436241797957961e-05, "loss": 2.8609, "step": 19534 }, { "epoch": 1.212676143770563, "grad_norm": 0.4848608687264927, "learning_rate": 7.435926410400015e-05, "loss": 2.9181, "step": 19535 }, { "epoch": 1.212738220870321, "grad_norm": 0.635466930610973, "learning_rate": 7.435611010133253e-05, "loss": 2.8889, "step": 19536 }, { "epoch": 1.212800297970079, "grad_norm": 0.5193403940343146, "learning_rate": 7.435295597159317e-05, "loss": 2.898, "step": 19537 }, { "epoch": 1.2128623750698369, "grad_norm": 0.5452224558729843, "learning_rate": 7.43498017147985e-05, "loss": 2.8875, "step": 19538 }, { "epoch": 1.2129244521695945, "grad_norm": 0.48416405360566694, "learning_rate": 7.434664733096502e-05, "loss": 2.7809, "step": 19539 }, { "epoch": 1.2129865292693525, "grad_norm": 0.5210112808144817, "learning_rate": 7.434349282010916e-05, "loss": 2.7988, "step": 19540 }, { "epoch": 1.2130486063691104, "grad_norm": 0.5146025432702475, "learning_rate": 7.434033818224739e-05, "loss": 2.811, "step": 19541 }, { "epoch": 1.2131106834688683, "grad_norm": 0.5188431162162558, "learning_rate": 7.433718341739617e-05, "loss": 2.9233, "step": 19542 }, { "epoch": 1.2131727605686262, "grad_norm": 0.5043923746304785, "learning_rate": 7.433402852557197e-05, "loss": 2.8401, "step": 19543 }, { "epoch": 1.2132348376683841, "grad_norm": 0.5386235758350283, "learning_rate": 7.43308735067912e-05, "loss": 2.8405, "step": 19544 }, { "epoch": 1.213296914768142, "grad_norm": 0.4750055300240098, "learning_rate": 7.43277183610704e-05, "loss": 2.777, "step": 19545 }, { "epoch": 1.2133589918679, "grad_norm": 0.5166111676597753, "learning_rate": 7.432456308842596e-05, "loss": 2.8529, "step": 19546 }, { "epoch": 1.213421068967658, "grad_norm": 0.5413934097859173, "learning_rate": 7.432140768887437e-05, "loss": 2.8758, "step": 19547 }, { "epoch": 1.2134831460674158, "grad_norm": 0.5369185785053008, "learning_rate": 7.43182521624321e-05, "loss": 2.8097, "step": 19548 }, { "epoch": 1.2135452231671737, "grad_norm": 0.5411747171760469, "learning_rate": 7.43150965091156e-05, "loss": 2.836, "step": 19549 }, { "epoch": 1.2136073002669314, "grad_norm": 0.5875285861994434, "learning_rate": 7.431194072894135e-05, "loss": 2.97, "step": 19550 }, { "epoch": 1.2136693773666893, "grad_norm": 0.5834900668164157, "learning_rate": 7.43087848219258e-05, "loss": 2.8885, "step": 19551 }, { "epoch": 1.2137314544664473, "grad_norm": 0.5088159477174244, "learning_rate": 7.43056287880854e-05, "loss": 2.8868, "step": 19552 }, { "epoch": 1.2137935315662052, "grad_norm": 0.5734643976623134, "learning_rate": 7.430247262743665e-05, "loss": 2.9416, "step": 19553 }, { "epoch": 1.213855608665963, "grad_norm": 0.5876279478553005, "learning_rate": 7.4299316339996e-05, "loss": 2.949, "step": 19554 }, { "epoch": 1.213917685765721, "grad_norm": 0.5203530831304376, "learning_rate": 7.429615992577992e-05, "loss": 2.7806, "step": 19555 }, { "epoch": 1.213979762865479, "grad_norm": 0.5320553285412765, "learning_rate": 7.429300338480488e-05, "loss": 2.9881, "step": 19556 }, { "epoch": 1.2140418399652368, "grad_norm": 0.622933380408801, "learning_rate": 7.428984671708733e-05, "loss": 2.8504, "step": 19557 }, { "epoch": 1.2141039170649948, "grad_norm": 0.6192758695260211, "learning_rate": 7.428668992264375e-05, "loss": 2.7942, "step": 19558 }, { "epoch": 1.2141659941647527, "grad_norm": 0.5806565916502717, "learning_rate": 7.428353300149063e-05, "loss": 2.9047, "step": 19559 }, { "epoch": 1.2142280712645106, "grad_norm": 0.5718398609726448, "learning_rate": 7.428037595364441e-05, "loss": 2.8135, "step": 19560 }, { "epoch": 1.2142901483642685, "grad_norm": 0.6393923181181533, "learning_rate": 7.427721877912158e-05, "loss": 2.8665, "step": 19561 }, { "epoch": 1.2143522254640264, "grad_norm": 0.5697533989236238, "learning_rate": 7.427406147793861e-05, "loss": 2.9056, "step": 19562 }, { "epoch": 1.2144143025637841, "grad_norm": 0.5746034384601694, "learning_rate": 7.427090405011196e-05, "loss": 2.8528, "step": 19563 }, { "epoch": 1.214476379663542, "grad_norm": 0.5773921652020194, "learning_rate": 7.426774649565812e-05, "loss": 2.8434, "step": 19564 }, { "epoch": 1.2145384567633, "grad_norm": 0.5073970562486733, "learning_rate": 7.426458881459355e-05, "loss": 2.957, "step": 19565 }, { "epoch": 1.2146005338630579, "grad_norm": 0.5979887841314724, "learning_rate": 7.426143100693472e-05, "loss": 2.7859, "step": 19566 }, { "epoch": 1.2146626109628158, "grad_norm": 0.48139673262625615, "learning_rate": 7.425827307269813e-05, "loss": 2.8784, "step": 19567 }, { "epoch": 1.2147246880625737, "grad_norm": 0.5665405903370883, "learning_rate": 7.425511501190023e-05, "loss": 2.9104, "step": 19568 }, { "epoch": 1.2147867651623316, "grad_norm": 0.46614479263729375, "learning_rate": 7.425195682455752e-05, "loss": 2.8547, "step": 19569 }, { "epoch": 1.2148488422620896, "grad_norm": 0.5169192456885182, "learning_rate": 7.424879851068646e-05, "loss": 2.8846, "step": 19570 }, { "epoch": 1.2149109193618475, "grad_norm": 0.4969970259385572, "learning_rate": 7.424564007030353e-05, "loss": 2.7944, "step": 19571 }, { "epoch": 1.2149729964616054, "grad_norm": 0.563723505078436, "learning_rate": 7.424248150342521e-05, "loss": 2.8773, "step": 19572 }, { "epoch": 1.2150350735613633, "grad_norm": 0.5193996406154395, "learning_rate": 7.423932281006797e-05, "loss": 2.8505, "step": 19573 }, { "epoch": 1.215097150661121, "grad_norm": 0.5106630282208096, "learning_rate": 7.423616399024831e-05, "loss": 2.9022, "step": 19574 }, { "epoch": 1.215159227760879, "grad_norm": 0.4802924352164115, "learning_rate": 7.42330050439827e-05, "loss": 2.9077, "step": 19575 }, { "epoch": 1.2152213048606368, "grad_norm": 0.5249782205787069, "learning_rate": 7.422984597128762e-05, "loss": 2.8768, "step": 19576 }, { "epoch": 1.2152833819603948, "grad_norm": 0.49857130676177086, "learning_rate": 7.422668677217955e-05, "loss": 2.8455, "step": 19577 }, { "epoch": 1.2153454590601527, "grad_norm": 0.6009262237443052, "learning_rate": 7.422352744667496e-05, "loss": 3.0096, "step": 19578 }, { "epoch": 1.2154075361599106, "grad_norm": 0.5004458525746583, "learning_rate": 7.422036799479035e-05, "loss": 2.8323, "step": 19579 }, { "epoch": 1.2154696132596685, "grad_norm": 0.4953104879536963, "learning_rate": 7.42172084165422e-05, "loss": 2.7551, "step": 19580 }, { "epoch": 1.2155316903594264, "grad_norm": 0.5648718472110211, "learning_rate": 7.421404871194701e-05, "loss": 2.8405, "step": 19581 }, { "epoch": 1.2155937674591843, "grad_norm": 0.5633483423494569, "learning_rate": 7.421088888102125e-05, "loss": 2.8156, "step": 19582 }, { "epoch": 1.2156558445589423, "grad_norm": 0.5518990664996591, "learning_rate": 7.420772892378138e-05, "loss": 2.7602, "step": 19583 }, { "epoch": 1.2157179216587002, "grad_norm": 0.5366240094342565, "learning_rate": 7.420456884024393e-05, "loss": 2.7589, "step": 19584 }, { "epoch": 1.215779998758458, "grad_norm": 0.5577298933993421, "learning_rate": 7.420140863042537e-05, "loss": 2.9721, "step": 19585 }, { "epoch": 1.215842075858216, "grad_norm": 0.5091935963956952, "learning_rate": 7.419824829434218e-05, "loss": 2.6806, "step": 19586 }, { "epoch": 1.2159041529579737, "grad_norm": 0.6506375597237144, "learning_rate": 7.419508783201086e-05, "loss": 2.865, "step": 19587 }, { "epoch": 1.2159662300577316, "grad_norm": 0.5645915195063989, "learning_rate": 7.41919272434479e-05, "loss": 2.8333, "step": 19588 }, { "epoch": 1.2160283071574896, "grad_norm": 0.5582613990002566, "learning_rate": 7.418876652866978e-05, "loss": 2.7593, "step": 19589 }, { "epoch": 1.2160903842572475, "grad_norm": 0.5936403173228431, "learning_rate": 7.418560568769298e-05, "loss": 2.8425, "step": 19590 }, { "epoch": 1.2161524613570054, "grad_norm": 0.5921844874886658, "learning_rate": 7.418244472053401e-05, "loss": 2.8426, "step": 19591 }, { "epoch": 1.2162145384567633, "grad_norm": 0.6216565347447551, "learning_rate": 7.417928362720935e-05, "loss": 2.8756, "step": 19592 }, { "epoch": 1.2162766155565212, "grad_norm": 0.5480405648072831, "learning_rate": 7.417612240773552e-05, "loss": 2.778, "step": 19593 }, { "epoch": 1.2163386926562791, "grad_norm": 0.5877379665952098, "learning_rate": 7.417296106212896e-05, "loss": 2.9564, "step": 19594 }, { "epoch": 1.216400769756037, "grad_norm": 0.5720382863799577, "learning_rate": 7.416979959040622e-05, "loss": 2.8293, "step": 19595 }, { "epoch": 1.216462846855795, "grad_norm": 0.5979209014751472, "learning_rate": 7.416663799258374e-05, "loss": 2.8798, "step": 19596 }, { "epoch": 1.2165249239555527, "grad_norm": 0.6612358745443558, "learning_rate": 7.416347626867806e-05, "loss": 2.8852, "step": 19597 }, { "epoch": 1.2165870010553106, "grad_norm": 0.597088463045363, "learning_rate": 7.416031441870565e-05, "loss": 2.8517, "step": 19598 }, { "epoch": 1.2166490781550685, "grad_norm": 0.6024305842604495, "learning_rate": 7.415715244268303e-05, "loss": 2.8016, "step": 19599 }, { "epoch": 1.2167111552548264, "grad_norm": 0.6296885109412926, "learning_rate": 7.415399034062668e-05, "loss": 2.8012, "step": 19600 }, { "epoch": 1.2167732323545843, "grad_norm": 0.5068960282823928, "learning_rate": 7.415082811255309e-05, "loss": 2.872, "step": 19601 }, { "epoch": 1.2168353094543423, "grad_norm": 0.5463640823652391, "learning_rate": 7.414766575847876e-05, "loss": 2.8298, "step": 19602 }, { "epoch": 1.2168973865541002, "grad_norm": 0.5709252919195292, "learning_rate": 7.414450327842021e-05, "loss": 2.8199, "step": 19603 }, { "epoch": 1.216959463653858, "grad_norm": 0.5892636054503092, "learning_rate": 7.414134067239392e-05, "loss": 2.8755, "step": 19604 }, { "epoch": 1.217021540753616, "grad_norm": 0.5912318790231582, "learning_rate": 7.41381779404164e-05, "loss": 2.9042, "step": 19605 }, { "epoch": 1.217083617853374, "grad_norm": 0.7336220635260946, "learning_rate": 7.413501508250413e-05, "loss": 2.8391, "step": 19606 }, { "epoch": 1.2171456949531319, "grad_norm": 0.5650106666199477, "learning_rate": 7.413185209867364e-05, "loss": 2.7415, "step": 19607 }, { "epoch": 1.2172077720528898, "grad_norm": 0.6365401128578619, "learning_rate": 7.412868898894141e-05, "loss": 2.8325, "step": 19608 }, { "epoch": 1.2172698491526477, "grad_norm": 0.6412578806104747, "learning_rate": 7.412552575332397e-05, "loss": 2.9053, "step": 19609 }, { "epoch": 1.2173319262524056, "grad_norm": 0.581964501771455, "learning_rate": 7.412236239183779e-05, "loss": 2.7872, "step": 19610 }, { "epoch": 1.2173940033521633, "grad_norm": 0.648679550219216, "learning_rate": 7.411919890449941e-05, "loss": 2.8069, "step": 19611 }, { "epoch": 1.2174560804519212, "grad_norm": 0.5795148690120616, "learning_rate": 7.411603529132529e-05, "loss": 2.8503, "step": 19612 }, { "epoch": 1.2175181575516791, "grad_norm": 0.6108996318583738, "learning_rate": 7.411287155233196e-05, "loss": 2.7639, "step": 19613 }, { "epoch": 1.217580234651437, "grad_norm": 0.6488336938369864, "learning_rate": 7.410970768753594e-05, "loss": 2.8369, "step": 19614 }, { "epoch": 1.217642311751195, "grad_norm": 0.5792435316038296, "learning_rate": 7.410654369695371e-05, "loss": 2.8086, "step": 19615 }, { "epoch": 1.217704388850953, "grad_norm": 0.6524751724986027, "learning_rate": 7.41033795806018e-05, "loss": 2.7835, "step": 19616 }, { "epoch": 1.2177664659507108, "grad_norm": 0.5634338256919245, "learning_rate": 7.410021533849671e-05, "loss": 2.8276, "step": 19617 }, { "epoch": 1.2178285430504687, "grad_norm": 0.5612375609754758, "learning_rate": 7.409705097065493e-05, "loss": 2.8199, "step": 19618 }, { "epoch": 1.2178906201502266, "grad_norm": 0.6069955832510802, "learning_rate": 7.4093886477093e-05, "loss": 2.8306, "step": 19619 }, { "epoch": 1.2179526972499846, "grad_norm": 0.5421909002367854, "learning_rate": 7.409072185782742e-05, "loss": 2.851, "step": 19620 }, { "epoch": 1.2180147743497423, "grad_norm": 0.5624175011217953, "learning_rate": 7.408755711287467e-05, "loss": 2.8828, "step": 19621 }, { "epoch": 1.2180768514495002, "grad_norm": 0.6547198394581336, "learning_rate": 7.408439224225131e-05, "loss": 2.876, "step": 19622 }, { "epoch": 1.218138928549258, "grad_norm": 0.49405360303730295, "learning_rate": 7.408122724597381e-05, "loss": 2.7565, "step": 19623 }, { "epoch": 1.218201005649016, "grad_norm": 0.5477303763599881, "learning_rate": 7.407806212405872e-05, "loss": 2.7641, "step": 19624 }, { "epoch": 1.218263082748774, "grad_norm": 0.5497504524326241, "learning_rate": 7.407489687652252e-05, "loss": 2.8731, "step": 19625 }, { "epoch": 1.2183251598485318, "grad_norm": 0.5715628204428231, "learning_rate": 7.407173150338176e-05, "loss": 2.8253, "step": 19626 }, { "epoch": 1.2183872369482898, "grad_norm": 0.7078808058901921, "learning_rate": 7.406856600465293e-05, "loss": 2.8156, "step": 19627 }, { "epoch": 1.2184493140480477, "grad_norm": 0.5156710055085909, "learning_rate": 7.406540038035252e-05, "loss": 2.725, "step": 19628 }, { "epoch": 1.2185113911478056, "grad_norm": 0.6260946462961345, "learning_rate": 7.406223463049709e-05, "loss": 2.8818, "step": 19629 }, { "epoch": 1.2185734682475635, "grad_norm": 0.6646448386938709, "learning_rate": 7.405906875510314e-05, "loss": 2.8298, "step": 19630 }, { "epoch": 1.2186355453473214, "grad_norm": 0.5185415837415682, "learning_rate": 7.40559027541872e-05, "loss": 2.7831, "step": 19631 }, { "epoch": 1.2186976224470794, "grad_norm": 0.6004557646817191, "learning_rate": 7.405273662776574e-05, "loss": 2.7619, "step": 19632 }, { "epoch": 1.2187596995468373, "grad_norm": 0.5551482422774822, "learning_rate": 7.404957037585536e-05, "loss": 2.8724, "step": 19633 }, { "epoch": 1.2188217766465952, "grad_norm": 0.5506712079874994, "learning_rate": 7.404640399847251e-05, "loss": 2.926, "step": 19634 }, { "epoch": 1.2188838537463529, "grad_norm": 0.5446907291546436, "learning_rate": 7.404323749563374e-05, "loss": 2.8128, "step": 19635 }, { "epoch": 1.2189459308461108, "grad_norm": 0.5737358450130416, "learning_rate": 7.404007086735557e-05, "loss": 2.7542, "step": 19636 }, { "epoch": 1.2190080079458687, "grad_norm": 0.5246148604094152, "learning_rate": 7.403690411365449e-05, "loss": 2.9108, "step": 19637 }, { "epoch": 1.2190700850456266, "grad_norm": 0.4789560743242408, "learning_rate": 7.403373723454707e-05, "loss": 2.824, "step": 19638 }, { "epoch": 1.2191321621453846, "grad_norm": 0.6825412604327636, "learning_rate": 7.403057023004979e-05, "loss": 2.8988, "step": 19639 }, { "epoch": 1.2191942392451425, "grad_norm": 0.5929366865774747, "learning_rate": 7.40274031001792e-05, "loss": 2.8682, "step": 19640 }, { "epoch": 1.2192563163449004, "grad_norm": 0.5921344864197908, "learning_rate": 7.402423584495181e-05, "loss": 2.8799, "step": 19641 }, { "epoch": 1.2193183934446583, "grad_norm": 0.5704628210666045, "learning_rate": 7.402106846438417e-05, "loss": 2.7651, "step": 19642 }, { "epoch": 1.2193804705444162, "grad_norm": 0.6714732721197498, "learning_rate": 7.401790095849275e-05, "loss": 2.8807, "step": 19643 }, { "epoch": 1.2194425476441741, "grad_norm": 0.5685062284271004, "learning_rate": 7.401473332729414e-05, "loss": 2.7743, "step": 19644 }, { "epoch": 1.2195046247439318, "grad_norm": 0.585847796885154, "learning_rate": 7.401156557080482e-05, "loss": 2.8866, "step": 19645 }, { "epoch": 1.2195667018436898, "grad_norm": 0.6219930555915776, "learning_rate": 7.400839768904133e-05, "loss": 2.8579, "step": 19646 }, { "epoch": 1.2196287789434477, "grad_norm": 0.5441027812299157, "learning_rate": 7.400522968202022e-05, "loss": 2.8798, "step": 19647 }, { "epoch": 1.2196908560432056, "grad_norm": 0.5737419744108954, "learning_rate": 7.400206154975798e-05, "loss": 2.8566, "step": 19648 }, { "epoch": 1.2197529331429635, "grad_norm": 0.5715529916263362, "learning_rate": 7.399889329227117e-05, "loss": 2.8589, "step": 19649 }, { "epoch": 1.2198150102427214, "grad_norm": 0.5677680383720141, "learning_rate": 7.399572490957631e-05, "loss": 2.868, "step": 19650 }, { "epoch": 1.2198770873424793, "grad_norm": 0.5059118943217366, "learning_rate": 7.399255640168992e-05, "loss": 2.8578, "step": 19651 }, { "epoch": 1.2199391644422373, "grad_norm": 0.543135528199026, "learning_rate": 7.398938776862855e-05, "loss": 2.8648, "step": 19652 }, { "epoch": 1.2200012415419952, "grad_norm": 0.556893461929647, "learning_rate": 7.39862190104087e-05, "loss": 2.8816, "step": 19653 }, { "epoch": 1.220063318641753, "grad_norm": 0.5880848763265318, "learning_rate": 7.398305012704695e-05, "loss": 2.8631, "step": 19654 }, { "epoch": 1.220125395741511, "grad_norm": 0.5692787207091659, "learning_rate": 7.397988111855978e-05, "loss": 2.8038, "step": 19655 }, { "epoch": 1.220187472841269, "grad_norm": 0.5367759363173823, "learning_rate": 7.397671198496377e-05, "loss": 2.8647, "step": 19656 }, { "epoch": 1.2202495499410269, "grad_norm": 0.5217247741022397, "learning_rate": 7.397354272627541e-05, "loss": 2.7898, "step": 19657 }, { "epoch": 1.2203116270407848, "grad_norm": 0.5892944304329444, "learning_rate": 7.397037334251128e-05, "loss": 2.9114, "step": 19658 }, { "epoch": 1.2203737041405425, "grad_norm": 0.6180288157828152, "learning_rate": 7.396720383368788e-05, "loss": 2.8773, "step": 19659 }, { "epoch": 1.2204357812403004, "grad_norm": 0.5551351970242002, "learning_rate": 7.396403419982177e-05, "loss": 2.7926, "step": 19660 }, { "epoch": 1.2204978583400583, "grad_norm": 0.5797500710250857, "learning_rate": 7.396086444092946e-05, "loss": 2.8665, "step": 19661 }, { "epoch": 1.2205599354398162, "grad_norm": 0.6229034067314002, "learning_rate": 7.39576945570275e-05, "loss": 2.7911, "step": 19662 }, { "epoch": 1.2206220125395741, "grad_norm": 0.5905024406389128, "learning_rate": 7.395452454813246e-05, "loss": 2.9184, "step": 19663 }, { "epoch": 1.220684089639332, "grad_norm": 0.6061282841913449, "learning_rate": 7.395135441426082e-05, "loss": 2.8671, "step": 19664 }, { "epoch": 1.22074616673909, "grad_norm": 0.5258253773371022, "learning_rate": 7.394818415542916e-05, "loss": 2.8839, "step": 19665 }, { "epoch": 1.220808243838848, "grad_norm": 0.5574974292216879, "learning_rate": 7.394501377165402e-05, "loss": 2.8882, "step": 19666 }, { "epoch": 1.2208703209386058, "grad_norm": 0.480693128963293, "learning_rate": 7.394184326295193e-05, "loss": 2.8122, "step": 19667 }, { "epoch": 1.2209323980383637, "grad_norm": 0.6009913619050636, "learning_rate": 7.393867262933943e-05, "loss": 2.8979, "step": 19668 }, { "epoch": 1.2209944751381214, "grad_norm": 0.5050615772625501, "learning_rate": 7.393550187083305e-05, "loss": 2.8521, "step": 19669 }, { "epoch": 1.2210565522378793, "grad_norm": 0.48670535796280656, "learning_rate": 7.393233098744936e-05, "loss": 2.8381, "step": 19670 }, { "epoch": 1.2211186293376373, "grad_norm": 0.516043839704825, "learning_rate": 7.392915997920488e-05, "loss": 2.9033, "step": 19671 }, { "epoch": 1.2211807064373952, "grad_norm": 0.5411701739077832, "learning_rate": 7.392598884611616e-05, "loss": 2.8272, "step": 19672 }, { "epoch": 1.221242783537153, "grad_norm": 0.5414467481298688, "learning_rate": 7.392281758819977e-05, "loss": 2.8538, "step": 19673 }, { "epoch": 1.221304860636911, "grad_norm": 0.5799571209710849, "learning_rate": 7.391964620547221e-05, "loss": 2.9137, "step": 19674 }, { "epoch": 1.221366937736669, "grad_norm": 0.5444055661227345, "learning_rate": 7.391647469795006e-05, "loss": 2.8136, "step": 19675 }, { "epoch": 1.2214290148364269, "grad_norm": 0.5744041091798527, "learning_rate": 7.391330306564986e-05, "loss": 2.7572, "step": 19676 }, { "epoch": 1.2214910919361848, "grad_norm": 0.5070100459527466, "learning_rate": 7.391013130858814e-05, "loss": 2.8621, "step": 19677 }, { "epoch": 1.2215531690359427, "grad_norm": 0.5197651272141645, "learning_rate": 7.390695942678146e-05, "loss": 2.8119, "step": 19678 }, { "epoch": 1.2216152461357006, "grad_norm": 0.5553639545119312, "learning_rate": 7.390378742024638e-05, "loss": 2.7859, "step": 19679 }, { "epoch": 1.2216773232354585, "grad_norm": 0.5285356685562219, "learning_rate": 7.390061528899943e-05, "loss": 2.9035, "step": 19680 }, { "epoch": 1.2217394003352164, "grad_norm": 0.5034836468185976, "learning_rate": 7.389744303305718e-05, "loss": 2.8173, "step": 19681 }, { "epoch": 1.2218014774349744, "grad_norm": 0.5254039266384393, "learning_rate": 7.389427065243615e-05, "loss": 2.8583, "step": 19682 }, { "epoch": 1.221863554534732, "grad_norm": 0.5317178237104349, "learning_rate": 7.389109814715292e-05, "loss": 2.7589, "step": 19683 }, { "epoch": 1.22192563163449, "grad_norm": 0.5279135839537367, "learning_rate": 7.388792551722404e-05, "loss": 2.8674, "step": 19684 }, { "epoch": 1.221987708734248, "grad_norm": 0.5233337242536037, "learning_rate": 7.388475276266604e-05, "loss": 2.7945, "step": 19685 }, { "epoch": 1.2220497858340058, "grad_norm": 0.5552005275824126, "learning_rate": 7.388157988349549e-05, "loss": 2.8788, "step": 19686 }, { "epoch": 1.2221118629337637, "grad_norm": 0.5370035795575541, "learning_rate": 7.387840687972895e-05, "loss": 2.8295, "step": 19687 }, { "epoch": 1.2221739400335216, "grad_norm": 0.5544237396249302, "learning_rate": 7.387523375138295e-05, "loss": 2.763, "step": 19688 }, { "epoch": 1.2222360171332796, "grad_norm": 0.5218840364142108, "learning_rate": 7.387206049847407e-05, "loss": 2.8146, "step": 19689 }, { "epoch": 1.2222980942330375, "grad_norm": 0.5783680070034268, "learning_rate": 7.386888712101885e-05, "loss": 2.9344, "step": 19690 }, { "epoch": 1.2223601713327954, "grad_norm": 0.5517266458806841, "learning_rate": 7.386571361903385e-05, "loss": 2.8698, "step": 19691 }, { "epoch": 1.2224222484325533, "grad_norm": 0.4982399061532842, "learning_rate": 7.386253999253564e-05, "loss": 2.8805, "step": 19692 }, { "epoch": 1.222484325532311, "grad_norm": 0.5849914542821112, "learning_rate": 7.385936624154076e-05, "loss": 2.8177, "step": 19693 }, { "epoch": 1.222546402632069, "grad_norm": 0.5167673922870624, "learning_rate": 7.385619236606576e-05, "loss": 2.8328, "step": 19694 }, { "epoch": 1.2226084797318268, "grad_norm": 0.5632571581636673, "learning_rate": 7.385301836612722e-05, "loss": 2.7646, "step": 19695 }, { "epoch": 1.2226705568315848, "grad_norm": 0.48023661766689074, "learning_rate": 7.384984424174169e-05, "loss": 2.8129, "step": 19696 }, { "epoch": 1.2227326339313427, "grad_norm": 0.5650088732368681, "learning_rate": 7.384666999292576e-05, "loss": 2.8187, "step": 19697 }, { "epoch": 1.2227947110311006, "grad_norm": 0.5282427718538645, "learning_rate": 7.384349561969594e-05, "loss": 2.8222, "step": 19698 }, { "epoch": 1.2228567881308585, "grad_norm": 0.5017268639415986, "learning_rate": 7.384032112206881e-05, "loss": 2.8796, "step": 19699 }, { "epoch": 1.2229188652306164, "grad_norm": 0.6030334225243971, "learning_rate": 7.383714650006095e-05, "loss": 2.6903, "step": 19700 }, { "epoch": 1.2229809423303744, "grad_norm": 0.5266629946657125, "learning_rate": 7.383397175368892e-05, "loss": 2.8543, "step": 19701 }, { "epoch": 1.2230430194301323, "grad_norm": 0.5910239885322109, "learning_rate": 7.383079688296924e-05, "loss": 2.8627, "step": 19702 }, { "epoch": 1.2231050965298902, "grad_norm": 0.5327505349738784, "learning_rate": 7.382762188791853e-05, "loss": 2.8177, "step": 19703 }, { "epoch": 1.223167173629648, "grad_norm": 0.6251981182804088, "learning_rate": 7.382444676855332e-05, "loss": 2.8752, "step": 19704 }, { "epoch": 1.223229250729406, "grad_norm": 0.5184457955550374, "learning_rate": 7.382127152489022e-05, "loss": 2.843, "step": 19705 }, { "epoch": 1.223291327829164, "grad_norm": 0.6157988849738825, "learning_rate": 7.381809615694573e-05, "loss": 2.9058, "step": 19706 }, { "epoch": 1.2233534049289216, "grad_norm": 0.5473745516757562, "learning_rate": 7.381492066473646e-05, "loss": 2.855, "step": 19707 }, { "epoch": 1.2234154820286796, "grad_norm": 0.53676361051981, "learning_rate": 7.381174504827897e-05, "loss": 2.8687, "step": 19708 }, { "epoch": 1.2234775591284375, "grad_norm": 0.5825123174944942, "learning_rate": 7.380856930758982e-05, "loss": 2.8825, "step": 19709 }, { "epoch": 1.2235396362281954, "grad_norm": 0.48562783067386395, "learning_rate": 7.38053934426856e-05, "loss": 2.8154, "step": 19710 }, { "epoch": 1.2236017133279533, "grad_norm": 0.595621647500113, "learning_rate": 7.380221745358283e-05, "loss": 2.8446, "step": 19711 }, { "epoch": 1.2236637904277112, "grad_norm": 0.5772489405358374, "learning_rate": 7.379904134029813e-05, "loss": 2.852, "step": 19712 }, { "epoch": 1.2237258675274691, "grad_norm": 0.5371285729878675, "learning_rate": 7.379586510284806e-05, "loss": 2.8334, "step": 19713 }, { "epoch": 1.223787944627227, "grad_norm": 0.5833585824497428, "learning_rate": 7.379268874124919e-05, "loss": 2.9001, "step": 19714 }, { "epoch": 1.223850021726985, "grad_norm": 0.5571722871305016, "learning_rate": 7.37895122555181e-05, "loss": 2.8441, "step": 19715 }, { "epoch": 1.223912098826743, "grad_norm": 0.6057891244206788, "learning_rate": 7.378633564567132e-05, "loss": 2.8279, "step": 19716 }, { "epoch": 1.2239741759265006, "grad_norm": 0.5145555155361269, "learning_rate": 7.378315891172548e-05, "loss": 2.9138, "step": 19717 }, { "epoch": 1.2240362530262585, "grad_norm": 0.6106593706263932, "learning_rate": 7.37799820536971e-05, "loss": 2.8864, "step": 19718 }, { "epoch": 1.2240983301260164, "grad_norm": 0.6195707301371777, "learning_rate": 7.377680507160279e-05, "loss": 2.8483, "step": 19719 }, { "epoch": 1.2241604072257743, "grad_norm": 0.4747275688030771, "learning_rate": 7.37736279654591e-05, "loss": 2.7837, "step": 19720 }, { "epoch": 1.2242224843255323, "grad_norm": 0.6713558453483892, "learning_rate": 7.377045073528264e-05, "loss": 2.7941, "step": 19721 }, { "epoch": 1.2242845614252902, "grad_norm": 0.5594171177868743, "learning_rate": 7.376727338108997e-05, "loss": 2.7991, "step": 19722 }, { "epoch": 1.224346638525048, "grad_norm": 0.6050495542377229, "learning_rate": 7.376409590289766e-05, "loss": 2.8074, "step": 19723 }, { "epoch": 1.224408715624806, "grad_norm": 0.6242147042115912, "learning_rate": 7.376091830072229e-05, "loss": 2.8362, "step": 19724 }, { "epoch": 1.224470792724564, "grad_norm": 0.639295687487321, "learning_rate": 7.375774057458045e-05, "loss": 2.772, "step": 19725 }, { "epoch": 1.2245328698243219, "grad_norm": 0.592245152467701, "learning_rate": 7.375456272448871e-05, "loss": 2.8738, "step": 19726 }, { "epoch": 1.2245949469240798, "grad_norm": 0.6241251783456668, "learning_rate": 7.375138475046363e-05, "loss": 2.7864, "step": 19727 }, { "epoch": 1.2246570240238377, "grad_norm": 0.5472453906656756, "learning_rate": 7.374820665252184e-05, "loss": 2.7861, "step": 19728 }, { "epoch": 1.2247191011235956, "grad_norm": 0.6168276063439675, "learning_rate": 7.374502843067987e-05, "loss": 2.8376, "step": 19729 }, { "epoch": 1.2247811782233535, "grad_norm": 0.5760937395294992, "learning_rate": 7.374185008495432e-05, "loss": 2.8569, "step": 19730 }, { "epoch": 1.2248432553231112, "grad_norm": 0.5778410446188657, "learning_rate": 7.373867161536179e-05, "loss": 2.7645, "step": 19731 }, { "epoch": 1.2249053324228691, "grad_norm": 0.5457326128781004, "learning_rate": 7.373549302191883e-05, "loss": 2.8933, "step": 19732 }, { "epoch": 1.224967409522627, "grad_norm": 0.5601108729499181, "learning_rate": 7.373231430464206e-05, "loss": 2.8223, "step": 19733 }, { "epoch": 1.225029486622385, "grad_norm": 0.5556432522327892, "learning_rate": 7.372913546354802e-05, "loss": 2.8042, "step": 19734 }, { "epoch": 1.225091563722143, "grad_norm": 0.6244863783860328, "learning_rate": 7.372595649865334e-05, "loss": 2.8799, "step": 19735 }, { "epoch": 1.2251536408219008, "grad_norm": 0.518203271445707, "learning_rate": 7.372277740997458e-05, "loss": 2.8567, "step": 19736 }, { "epoch": 1.2252157179216587, "grad_norm": 0.5789363940752534, "learning_rate": 7.371959819752834e-05, "loss": 2.7973, "step": 19737 }, { "epoch": 1.2252777950214166, "grad_norm": 0.5600908398179365, "learning_rate": 7.37164188613312e-05, "loss": 2.8576, "step": 19738 }, { "epoch": 1.2253398721211746, "grad_norm": 0.5191120378073295, "learning_rate": 7.371323940139974e-05, "loss": 2.8294, "step": 19739 }, { "epoch": 1.2254019492209325, "grad_norm": 0.5438895342460096, "learning_rate": 7.371005981775054e-05, "loss": 2.9229, "step": 19740 }, { "epoch": 1.2254640263206902, "grad_norm": 0.5772476756436095, "learning_rate": 7.370688011040021e-05, "loss": 2.8323, "step": 19741 }, { "epoch": 1.225526103420448, "grad_norm": 0.5506919084960649, "learning_rate": 7.370370027936533e-05, "loss": 2.771, "step": 19742 }, { "epoch": 1.225588180520206, "grad_norm": 0.5444995241487104, "learning_rate": 7.370052032466251e-05, "loss": 2.9195, "step": 19743 }, { "epoch": 1.225650257619964, "grad_norm": 0.593386162158872, "learning_rate": 7.369734024630828e-05, "loss": 2.7922, "step": 19744 }, { "epoch": 1.2257123347197219, "grad_norm": 0.5667292496071699, "learning_rate": 7.36941600443193e-05, "loss": 2.8028, "step": 19745 }, { "epoch": 1.2257744118194798, "grad_norm": 0.5169161612047849, "learning_rate": 7.369097971871214e-05, "loss": 2.8523, "step": 19746 }, { "epoch": 1.2258364889192377, "grad_norm": 0.5655700562067505, "learning_rate": 7.368779926950338e-05, "loss": 2.8215, "step": 19747 }, { "epoch": 1.2258985660189956, "grad_norm": 0.48491993368579756, "learning_rate": 7.368461869670962e-05, "loss": 2.8871, "step": 19748 }, { "epoch": 1.2259606431187535, "grad_norm": 0.5277726019081276, "learning_rate": 7.368143800034746e-05, "loss": 2.8178, "step": 19749 }, { "epoch": 1.2260227202185114, "grad_norm": 0.5235719650413011, "learning_rate": 7.367825718043348e-05, "loss": 2.8503, "step": 19750 }, { "epoch": 1.2260847973182694, "grad_norm": 0.47922956185198207, "learning_rate": 7.367507623698431e-05, "loss": 2.7957, "step": 19751 }, { "epoch": 1.2261468744180273, "grad_norm": 0.5332631507035287, "learning_rate": 7.367189517001649e-05, "loss": 2.8731, "step": 19752 }, { "epoch": 1.2262089515177852, "grad_norm": 0.5452125075076866, "learning_rate": 7.366871397954667e-05, "loss": 2.8345, "step": 19753 }, { "epoch": 1.2262710286175431, "grad_norm": 0.5676829189479824, "learning_rate": 7.36655326655914e-05, "loss": 2.8719, "step": 19754 }, { "epoch": 1.2263331057173008, "grad_norm": 0.5646246151301043, "learning_rate": 7.366235122816732e-05, "loss": 2.7951, "step": 19755 }, { "epoch": 1.2263951828170587, "grad_norm": 0.5284807733631222, "learning_rate": 7.365916966729098e-05, "loss": 2.8535, "step": 19756 }, { "epoch": 1.2264572599168166, "grad_norm": 0.5369209652675877, "learning_rate": 7.365598798297904e-05, "loss": 2.7842, "step": 19757 }, { "epoch": 1.2265193370165746, "grad_norm": 0.5725459431033559, "learning_rate": 7.365280617524805e-05, "loss": 2.8894, "step": 19758 }, { "epoch": 1.2265814141163325, "grad_norm": 0.5241385647296428, "learning_rate": 7.364962424411465e-05, "loss": 2.8978, "step": 19759 }, { "epoch": 1.2266434912160904, "grad_norm": 0.6153066683483674, "learning_rate": 7.364644218959538e-05, "loss": 2.8636, "step": 19760 }, { "epoch": 1.2267055683158483, "grad_norm": 0.5515176706224004, "learning_rate": 7.364326001170692e-05, "loss": 2.8534, "step": 19761 }, { "epoch": 1.2267676454156062, "grad_norm": 0.5915329347729867, "learning_rate": 7.364007771046582e-05, "loss": 2.8206, "step": 19762 }, { "epoch": 1.2268297225153642, "grad_norm": 0.5249950806069366, "learning_rate": 7.363689528588869e-05, "loss": 2.803, "step": 19763 }, { "epoch": 1.226891799615122, "grad_norm": 0.6085123411680872, "learning_rate": 7.363371273799214e-05, "loss": 2.8213, "step": 19764 }, { "epoch": 1.2269538767148798, "grad_norm": 0.5486668251279939, "learning_rate": 7.363053006679277e-05, "loss": 2.7978, "step": 19765 }, { "epoch": 1.2270159538146377, "grad_norm": 0.6055386164563092, "learning_rate": 7.36273472723072e-05, "loss": 2.8706, "step": 19766 }, { "epoch": 1.2270780309143956, "grad_norm": 0.545152020819308, "learning_rate": 7.362416435455202e-05, "loss": 2.8245, "step": 19767 }, { "epoch": 1.2271401080141535, "grad_norm": 0.5631150485904867, "learning_rate": 7.362098131354384e-05, "loss": 2.8179, "step": 19768 }, { "epoch": 1.2272021851139114, "grad_norm": 0.5522629577598213, "learning_rate": 7.361779814929927e-05, "loss": 2.8716, "step": 19769 }, { "epoch": 1.2272642622136694, "grad_norm": 0.5033823291139858, "learning_rate": 7.36146148618349e-05, "loss": 2.8973, "step": 19770 }, { "epoch": 1.2273263393134273, "grad_norm": 0.5505470156571323, "learning_rate": 7.361143145116737e-05, "loss": 2.9096, "step": 19771 }, { "epoch": 1.2273884164131852, "grad_norm": 0.555710373274395, "learning_rate": 7.360824791731324e-05, "loss": 2.8306, "step": 19772 }, { "epoch": 1.227450493512943, "grad_norm": 0.5401320972289744, "learning_rate": 7.360506426028919e-05, "loss": 2.9058, "step": 19773 }, { "epoch": 1.227512570612701, "grad_norm": 0.5940730069599351, "learning_rate": 7.360188048011177e-05, "loss": 2.8931, "step": 19774 }, { "epoch": 1.227574647712459, "grad_norm": 0.5411581960177095, "learning_rate": 7.35986965767976e-05, "loss": 2.934, "step": 19775 }, { "epoch": 1.2276367248122169, "grad_norm": 0.6293525060291483, "learning_rate": 7.359551255036329e-05, "loss": 2.8941, "step": 19776 }, { "epoch": 1.2276988019119748, "grad_norm": 0.5883896949380432, "learning_rate": 7.359232840082549e-05, "loss": 2.8957, "step": 19777 }, { "epoch": 1.2277608790117325, "grad_norm": 0.5198619908353607, "learning_rate": 7.358914412820078e-05, "loss": 2.8253, "step": 19778 }, { "epoch": 1.2278229561114904, "grad_norm": 0.601010849910299, "learning_rate": 7.358595973250577e-05, "loss": 2.8341, "step": 19779 }, { "epoch": 1.2278850332112483, "grad_norm": 0.6207994927607018, "learning_rate": 7.35827752137571e-05, "loss": 2.8619, "step": 19780 }, { "epoch": 1.2279471103110062, "grad_norm": 0.5489606866900714, "learning_rate": 7.357959057197135e-05, "loss": 2.9197, "step": 19781 }, { "epoch": 1.2280091874107641, "grad_norm": 0.5933562026902783, "learning_rate": 7.357640580716516e-05, "loss": 2.8502, "step": 19782 }, { "epoch": 1.228071264510522, "grad_norm": 0.4795612044687917, "learning_rate": 7.357322091935514e-05, "loss": 2.7526, "step": 19783 }, { "epoch": 1.22813334161028, "grad_norm": 0.5763528830479895, "learning_rate": 7.35700359085579e-05, "loss": 2.7932, "step": 19784 }, { "epoch": 1.228195418710038, "grad_norm": 0.5384372457458285, "learning_rate": 7.356685077479005e-05, "loss": 2.8629, "step": 19785 }, { "epoch": 1.2282574958097958, "grad_norm": 0.6052855183619615, "learning_rate": 7.356366551806824e-05, "loss": 2.7861, "step": 19786 }, { "epoch": 1.2283195729095537, "grad_norm": 0.5562764900843489, "learning_rate": 7.356048013840906e-05, "loss": 2.8355, "step": 19787 }, { "epoch": 1.2283816500093117, "grad_norm": 0.6500622719499415, "learning_rate": 7.355729463582912e-05, "loss": 2.8656, "step": 19788 }, { "epoch": 1.2284437271090693, "grad_norm": 0.6166230529964756, "learning_rate": 7.355410901034507e-05, "loss": 2.7792, "step": 19789 }, { "epoch": 1.2285058042088273, "grad_norm": 0.5799937331337918, "learning_rate": 7.35509232619735e-05, "loss": 2.8509, "step": 19790 }, { "epoch": 1.2285678813085852, "grad_norm": 0.5603582882402738, "learning_rate": 7.354773739073107e-05, "loss": 2.849, "step": 19791 }, { "epoch": 1.228629958408343, "grad_norm": 0.6403930755974154, "learning_rate": 7.354455139663436e-05, "loss": 2.7951, "step": 19792 }, { "epoch": 1.228692035508101, "grad_norm": 0.522325986742152, "learning_rate": 7.354136527970001e-05, "loss": 2.7763, "step": 19793 }, { "epoch": 1.228754112607859, "grad_norm": 0.6095505974932585, "learning_rate": 7.353817903994465e-05, "loss": 2.8532, "step": 19794 }, { "epoch": 1.2288161897076169, "grad_norm": 0.5457739235915647, "learning_rate": 7.353499267738489e-05, "loss": 2.8514, "step": 19795 }, { "epoch": 1.2288782668073748, "grad_norm": 0.5732320495436852, "learning_rate": 7.353180619203736e-05, "loss": 2.8303, "step": 19796 }, { "epoch": 1.2289403439071327, "grad_norm": 0.6463833410137636, "learning_rate": 7.35286195839187e-05, "loss": 2.835, "step": 19797 }, { "epoch": 1.2290024210068906, "grad_norm": 0.5836657161844699, "learning_rate": 7.35254328530455e-05, "loss": 2.8359, "step": 19798 }, { "epoch": 1.2290644981066485, "grad_norm": 0.5737420263546651, "learning_rate": 7.352224599943441e-05, "loss": 2.8977, "step": 19799 }, { "epoch": 1.2291265752064064, "grad_norm": 0.7109356555285998, "learning_rate": 7.351905902310205e-05, "loss": 2.8864, "step": 19800 }, { "epoch": 1.2291886523061644, "grad_norm": 0.610187477944071, "learning_rate": 7.351587192406506e-05, "loss": 2.8041, "step": 19801 }, { "epoch": 1.229250729405922, "grad_norm": 0.5683466067633269, "learning_rate": 7.351268470234006e-05, "loss": 2.8319, "step": 19802 }, { "epoch": 1.22931280650568, "grad_norm": 0.617429082404373, "learning_rate": 7.350949735794365e-05, "loss": 2.7847, "step": 19803 }, { "epoch": 1.229374883605438, "grad_norm": 0.53907750979549, "learning_rate": 7.350630989089253e-05, "loss": 2.9463, "step": 19804 }, { "epoch": 1.2294369607051958, "grad_norm": 0.6067812322508619, "learning_rate": 7.350312230120325e-05, "loss": 2.8619, "step": 19805 }, { "epoch": 1.2294990378049537, "grad_norm": 0.5581276289310887, "learning_rate": 7.349993458889248e-05, "loss": 2.79, "step": 19806 }, { "epoch": 1.2295611149047116, "grad_norm": 0.5416252750085255, "learning_rate": 7.349674675397687e-05, "loss": 2.833, "step": 19807 }, { "epoch": 1.2296231920044696, "grad_norm": 0.6059590477183663, "learning_rate": 7.3493558796473e-05, "loss": 2.8622, "step": 19808 }, { "epoch": 1.2296852691042275, "grad_norm": 0.5338923432880913, "learning_rate": 7.349037071639752e-05, "loss": 2.8349, "step": 19809 }, { "epoch": 1.2297473462039854, "grad_norm": 0.5134714119799066, "learning_rate": 7.348718251376709e-05, "loss": 2.7423, "step": 19810 }, { "epoch": 1.2298094233037433, "grad_norm": 0.5103033160512126, "learning_rate": 7.348399418859834e-05, "loss": 2.8479, "step": 19811 }, { "epoch": 1.2298715004035012, "grad_norm": 0.4953317572282811, "learning_rate": 7.348080574090788e-05, "loss": 2.8301, "step": 19812 }, { "epoch": 1.229933577503259, "grad_norm": 0.5269867713089469, "learning_rate": 7.347761717071235e-05, "loss": 2.8299, "step": 19813 }, { "epoch": 1.2299956546030169, "grad_norm": 0.5101803667288771, "learning_rate": 7.347442847802842e-05, "loss": 2.8813, "step": 19814 }, { "epoch": 1.2300577317027748, "grad_norm": 0.4990294842437668, "learning_rate": 7.347123966287267e-05, "loss": 2.728, "step": 19815 }, { "epoch": 1.2301198088025327, "grad_norm": 0.47979566677886887, "learning_rate": 7.346805072526177e-05, "loss": 2.8856, "step": 19816 }, { "epoch": 1.2301818859022906, "grad_norm": 0.4508369821246854, "learning_rate": 7.346486166521235e-05, "loss": 2.8312, "step": 19817 }, { "epoch": 1.2302439630020485, "grad_norm": 0.5002340722548573, "learning_rate": 7.346167248274106e-05, "loss": 2.8849, "step": 19818 }, { "epoch": 1.2303060401018064, "grad_norm": 0.5214160774454831, "learning_rate": 7.345848317786453e-05, "loss": 2.7582, "step": 19819 }, { "epoch": 1.2303681172015644, "grad_norm": 0.5351058769186366, "learning_rate": 7.345529375059939e-05, "loss": 2.7748, "step": 19820 }, { "epoch": 1.2304301943013223, "grad_norm": 0.5389597421644171, "learning_rate": 7.345210420096229e-05, "loss": 2.9179, "step": 19821 }, { "epoch": 1.2304922714010802, "grad_norm": 0.5780175341137721, "learning_rate": 7.344891452896987e-05, "loss": 2.8079, "step": 19822 }, { "epoch": 1.2305543485008381, "grad_norm": 0.6181579878811421, "learning_rate": 7.344572473463878e-05, "loss": 2.8866, "step": 19823 }, { "epoch": 1.230616425600596, "grad_norm": 0.5401673811211707, "learning_rate": 7.344253481798563e-05, "loss": 2.8357, "step": 19824 }, { "epoch": 1.230678502700354, "grad_norm": 0.6684824212144722, "learning_rate": 7.34393447790271e-05, "loss": 2.8266, "step": 19825 }, { "epoch": 1.2307405798001116, "grad_norm": 0.5700352138383352, "learning_rate": 7.343615461777981e-05, "loss": 2.8382, "step": 19826 }, { "epoch": 1.2308026568998696, "grad_norm": 0.5490646395698594, "learning_rate": 7.343296433426042e-05, "loss": 2.7974, "step": 19827 }, { "epoch": 1.2308647339996275, "grad_norm": 0.6835976300810642, "learning_rate": 7.342977392848555e-05, "loss": 2.9055, "step": 19828 }, { "epoch": 1.2309268110993854, "grad_norm": 0.6158057571951037, "learning_rate": 7.342658340047189e-05, "loss": 2.763, "step": 19829 }, { "epoch": 1.2309888881991433, "grad_norm": 0.542734383697971, "learning_rate": 7.342339275023605e-05, "loss": 2.7258, "step": 19830 }, { "epoch": 1.2310509652989012, "grad_norm": 0.5975292295225562, "learning_rate": 7.342020197779466e-05, "loss": 2.8265, "step": 19831 }, { "epoch": 1.2311130423986592, "grad_norm": 0.5870267788369243, "learning_rate": 7.341701108316442e-05, "loss": 2.8773, "step": 19832 }, { "epoch": 1.231175119498417, "grad_norm": 0.5711650746446426, "learning_rate": 7.341382006636194e-05, "loss": 2.8496, "step": 19833 }, { "epoch": 1.231237196598175, "grad_norm": 0.5634741295509109, "learning_rate": 7.341062892740388e-05, "loss": 2.8621, "step": 19834 }, { "epoch": 1.231299273697933, "grad_norm": 0.5366415309745637, "learning_rate": 7.340743766630687e-05, "loss": 2.8376, "step": 19835 }, { "epoch": 1.2313613507976908, "grad_norm": 0.6034789583493321, "learning_rate": 7.34042462830876e-05, "loss": 2.7181, "step": 19836 }, { "epoch": 1.2314234278974485, "grad_norm": 0.6212067651191758, "learning_rate": 7.340105477776267e-05, "loss": 2.8529, "step": 19837 }, { "epoch": 1.2314855049972064, "grad_norm": 0.55648713254319, "learning_rate": 7.339786315034876e-05, "loss": 2.7489, "step": 19838 }, { "epoch": 1.2315475820969644, "grad_norm": 0.6032717313808428, "learning_rate": 7.339467140086253e-05, "loss": 2.7919, "step": 19839 }, { "epoch": 1.2316096591967223, "grad_norm": 0.6065341804492959, "learning_rate": 7.339147952932062e-05, "loss": 2.9158, "step": 19840 }, { "epoch": 1.2316717362964802, "grad_norm": 0.5791683319351443, "learning_rate": 7.338828753573968e-05, "loss": 2.8924, "step": 19841 }, { "epoch": 1.231733813396238, "grad_norm": 0.5661215757374668, "learning_rate": 7.338509542013635e-05, "loss": 2.8393, "step": 19842 }, { "epoch": 1.231795890495996, "grad_norm": 0.5758833340504856, "learning_rate": 7.338190318252731e-05, "loss": 2.9301, "step": 19843 }, { "epoch": 1.231857967595754, "grad_norm": 0.5484439143053, "learning_rate": 7.33787108229292e-05, "loss": 2.8882, "step": 19844 }, { "epoch": 1.2319200446955119, "grad_norm": 0.5954847842449782, "learning_rate": 7.337551834135871e-05, "loss": 2.8078, "step": 19845 }, { "epoch": 1.2319821217952698, "grad_norm": 0.6049773159976365, "learning_rate": 7.337232573783243e-05, "loss": 2.8709, "step": 19846 }, { "epoch": 1.2320441988950277, "grad_norm": 0.5328440592528156, "learning_rate": 7.336913301236708e-05, "loss": 2.7863, "step": 19847 }, { "epoch": 1.2321062759947856, "grad_norm": 0.5181748315387305, "learning_rate": 7.336594016497929e-05, "loss": 2.7606, "step": 19848 }, { "epoch": 1.2321683530945435, "grad_norm": 0.5839703153110652, "learning_rate": 7.33627471956857e-05, "loss": 2.8591, "step": 19849 }, { "epoch": 1.2322304301943012, "grad_norm": 0.4941112977021155, "learning_rate": 7.335955410450299e-05, "loss": 2.9206, "step": 19850 }, { "epoch": 1.2322925072940591, "grad_norm": 0.5998380144326485, "learning_rate": 7.335636089144782e-05, "loss": 2.8734, "step": 19851 }, { "epoch": 1.232354584393817, "grad_norm": 0.5764815193516013, "learning_rate": 7.335316755653684e-05, "loss": 2.8287, "step": 19852 }, { "epoch": 1.232416661493575, "grad_norm": 0.5674069476616271, "learning_rate": 7.334997409978672e-05, "loss": 2.9622, "step": 19853 }, { "epoch": 1.232478738593333, "grad_norm": 0.5343654665180209, "learning_rate": 7.334678052121412e-05, "loss": 2.7921, "step": 19854 }, { "epoch": 1.2325408156930908, "grad_norm": 0.5280978709368849, "learning_rate": 7.334358682083569e-05, "loss": 2.9239, "step": 19855 }, { "epoch": 1.2326028927928487, "grad_norm": 0.5422060982540275, "learning_rate": 7.334039299866811e-05, "loss": 2.8885, "step": 19856 }, { "epoch": 1.2326649698926067, "grad_norm": 0.4800465131863261, "learning_rate": 7.333719905472803e-05, "loss": 2.776, "step": 19857 }, { "epoch": 1.2327270469923646, "grad_norm": 0.48531049712014046, "learning_rate": 7.33340049890321e-05, "loss": 2.8565, "step": 19858 }, { "epoch": 1.2327891240921225, "grad_norm": 0.4654266805547669, "learning_rate": 7.333081080159702e-05, "loss": 2.8819, "step": 19859 }, { "epoch": 1.2328512011918804, "grad_norm": 0.5855978171880485, "learning_rate": 7.332761649243944e-05, "loss": 2.8672, "step": 19860 }, { "epoch": 1.232913278291638, "grad_norm": 0.4636347298659219, "learning_rate": 7.332442206157601e-05, "loss": 2.8714, "step": 19861 }, { "epoch": 1.232975355391396, "grad_norm": 0.5649819981821387, "learning_rate": 7.332122750902342e-05, "loss": 2.8853, "step": 19862 }, { "epoch": 1.233037432491154, "grad_norm": 0.5441960796022922, "learning_rate": 7.331803283479831e-05, "loss": 2.865, "step": 19863 }, { "epoch": 1.2330995095909119, "grad_norm": 0.5392398266139932, "learning_rate": 7.331483803891736e-05, "loss": 2.8952, "step": 19864 }, { "epoch": 1.2331615866906698, "grad_norm": 0.6393663323002264, "learning_rate": 7.331164312139727e-05, "loss": 2.7857, "step": 19865 }, { "epoch": 1.2332236637904277, "grad_norm": 0.5585284961920047, "learning_rate": 7.330844808225466e-05, "loss": 2.7754, "step": 19866 }, { "epoch": 1.2332857408901856, "grad_norm": 0.5195292709427619, "learning_rate": 7.330525292150621e-05, "loss": 2.8523, "step": 19867 }, { "epoch": 1.2333478179899435, "grad_norm": 0.5505914022736738, "learning_rate": 7.330205763916861e-05, "loss": 2.8694, "step": 19868 }, { "epoch": 1.2334098950897014, "grad_norm": 0.5222601388830014, "learning_rate": 7.329886223525851e-05, "loss": 2.9007, "step": 19869 }, { "epoch": 1.2334719721894594, "grad_norm": 0.5510802843759527, "learning_rate": 7.32956667097926e-05, "loss": 2.8929, "step": 19870 }, { "epoch": 1.2335340492892173, "grad_norm": 0.49801218248598805, "learning_rate": 7.329247106278752e-05, "loss": 2.8633, "step": 19871 }, { "epoch": 1.2335961263889752, "grad_norm": 0.5684859144785877, "learning_rate": 7.328927529425999e-05, "loss": 2.8128, "step": 19872 }, { "epoch": 1.2336582034887331, "grad_norm": 0.5717406745703913, "learning_rate": 7.328607940422663e-05, "loss": 2.8918, "step": 19873 }, { "epoch": 1.2337202805884908, "grad_norm": 0.5089779025622891, "learning_rate": 7.328288339270417e-05, "loss": 2.9085, "step": 19874 }, { "epoch": 1.2337823576882487, "grad_norm": 0.5734142714532958, "learning_rate": 7.327968725970924e-05, "loss": 2.7697, "step": 19875 }, { "epoch": 1.2338444347880066, "grad_norm": 0.5368429459494148, "learning_rate": 7.327649100525853e-05, "loss": 2.8958, "step": 19876 }, { "epoch": 1.2339065118877646, "grad_norm": 0.5507076295388921, "learning_rate": 7.327329462936871e-05, "loss": 2.8753, "step": 19877 }, { "epoch": 1.2339685889875225, "grad_norm": 0.4966107325478778, "learning_rate": 7.327009813205647e-05, "loss": 2.8081, "step": 19878 }, { "epoch": 1.2340306660872804, "grad_norm": 0.5970764838671238, "learning_rate": 7.326690151333848e-05, "loss": 2.7946, "step": 19879 }, { "epoch": 1.2340927431870383, "grad_norm": 0.5385645626029061, "learning_rate": 7.326370477323141e-05, "loss": 2.8081, "step": 19880 }, { "epoch": 1.2341548202867962, "grad_norm": 0.5983412314091164, "learning_rate": 7.326050791175196e-05, "loss": 2.8395, "step": 19881 }, { "epoch": 1.2342168973865542, "grad_norm": 0.5465570070264425, "learning_rate": 7.325731092891678e-05, "loss": 2.8923, "step": 19882 }, { "epoch": 1.234278974486312, "grad_norm": 0.5576341028537275, "learning_rate": 7.325411382474256e-05, "loss": 2.7964, "step": 19883 }, { "epoch": 1.23434105158607, "grad_norm": 0.6052598408257879, "learning_rate": 7.325091659924599e-05, "loss": 2.7965, "step": 19884 }, { "epoch": 1.2344031286858277, "grad_norm": 0.5269126825670194, "learning_rate": 7.324771925244376e-05, "loss": 2.8352, "step": 19885 }, { "epoch": 1.2344652057855856, "grad_norm": 0.5313355433088234, "learning_rate": 7.324452178435252e-05, "loss": 2.7746, "step": 19886 }, { "epoch": 1.2345272828853435, "grad_norm": 0.5358996275040431, "learning_rate": 7.324132419498895e-05, "loss": 2.924, "step": 19887 }, { "epoch": 1.2345893599851014, "grad_norm": 0.5615382715007332, "learning_rate": 7.323812648436979e-05, "loss": 2.9315, "step": 19888 }, { "epoch": 1.2346514370848594, "grad_norm": 0.5187310674958274, "learning_rate": 7.323492865251164e-05, "loss": 2.8204, "step": 19889 }, { "epoch": 1.2347135141846173, "grad_norm": 0.5702266628518929, "learning_rate": 7.323173069943125e-05, "loss": 2.7681, "step": 19890 }, { "epoch": 1.2347755912843752, "grad_norm": 0.604638767496573, "learning_rate": 7.322853262514528e-05, "loss": 2.8175, "step": 19891 }, { "epoch": 1.2348376683841331, "grad_norm": 0.5588377839667875, "learning_rate": 7.32253344296704e-05, "loss": 2.757, "step": 19892 }, { "epoch": 1.234899745483891, "grad_norm": 0.6236096653523875, "learning_rate": 7.322213611302333e-05, "loss": 2.8745, "step": 19893 }, { "epoch": 1.234961822583649, "grad_norm": 0.5167471783214466, "learning_rate": 7.321893767522072e-05, "loss": 2.8494, "step": 19894 }, { "epoch": 1.2350238996834069, "grad_norm": 0.5168508350115734, "learning_rate": 7.32157391162793e-05, "loss": 2.9091, "step": 19895 }, { "epoch": 1.2350859767831648, "grad_norm": 0.5487267342851266, "learning_rate": 7.321254043621572e-05, "loss": 2.8566, "step": 19896 }, { "epoch": 1.2351480538829227, "grad_norm": 0.55343567264714, "learning_rate": 7.320934163504669e-05, "loss": 2.7907, "step": 19897 }, { "epoch": 1.2352101309826804, "grad_norm": 0.580091713897086, "learning_rate": 7.320614271278887e-05, "loss": 2.8627, "step": 19898 }, { "epoch": 1.2352722080824383, "grad_norm": 0.5670638115059506, "learning_rate": 7.320294366945899e-05, "loss": 2.8302, "step": 19899 }, { "epoch": 1.2353342851821962, "grad_norm": 0.599977659763233, "learning_rate": 7.319974450507372e-05, "loss": 2.9029, "step": 19900 }, { "epoch": 1.2353963622819542, "grad_norm": 0.6005532108477354, "learning_rate": 7.319654521964975e-05, "loss": 2.8626, "step": 19901 }, { "epoch": 1.235458439381712, "grad_norm": 0.5874098525990349, "learning_rate": 7.319334581320376e-05, "loss": 2.9162, "step": 19902 }, { "epoch": 1.23552051648147, "grad_norm": 0.6237269071611976, "learning_rate": 7.319014628575247e-05, "loss": 2.9089, "step": 19903 }, { "epoch": 1.235582593581228, "grad_norm": 0.5556324713497504, "learning_rate": 7.318694663731255e-05, "loss": 2.8722, "step": 19904 }, { "epoch": 1.2356446706809858, "grad_norm": 0.5962051777416799, "learning_rate": 7.318374686790069e-05, "loss": 2.7799, "step": 19905 }, { "epoch": 1.2357067477807437, "grad_norm": 0.6294161701847749, "learning_rate": 7.318054697753362e-05, "loss": 2.8044, "step": 19906 }, { "epoch": 1.2357688248805017, "grad_norm": 0.5622422634506721, "learning_rate": 7.317734696622797e-05, "loss": 2.7776, "step": 19907 }, { "epoch": 1.2358309019802596, "grad_norm": 0.5598371578491728, "learning_rate": 7.31741468340005e-05, "loss": 2.7581, "step": 19908 }, { "epoch": 1.2358929790800173, "grad_norm": 0.5577336071226597, "learning_rate": 7.317094658086787e-05, "loss": 2.735, "step": 19909 }, { "epoch": 1.2359550561797752, "grad_norm": 0.5172523769012458, "learning_rate": 7.316774620684681e-05, "loss": 2.84, "step": 19910 }, { "epoch": 1.236017133279533, "grad_norm": 0.5891286038815228, "learning_rate": 7.316454571195398e-05, "loss": 2.8752, "step": 19911 }, { "epoch": 1.236079210379291, "grad_norm": 0.5183961845155253, "learning_rate": 7.316134509620609e-05, "loss": 2.7864, "step": 19912 }, { "epoch": 1.236141287479049, "grad_norm": 0.5950259001888905, "learning_rate": 7.315814435961985e-05, "loss": 2.9078, "step": 19913 }, { "epoch": 1.2362033645788069, "grad_norm": 0.5572420317985904, "learning_rate": 7.315494350221193e-05, "loss": 2.7608, "step": 19914 }, { "epoch": 1.2362654416785648, "grad_norm": 0.5853738744261805, "learning_rate": 7.315174252399905e-05, "loss": 2.9796, "step": 19915 }, { "epoch": 1.2363275187783227, "grad_norm": 0.5001933498857954, "learning_rate": 7.314854142499792e-05, "loss": 2.9376, "step": 19916 }, { "epoch": 1.2363895958780806, "grad_norm": 0.6518617322527964, "learning_rate": 7.314534020522523e-05, "loss": 2.8579, "step": 19917 }, { "epoch": 1.2364516729778385, "grad_norm": 0.5227228381700094, "learning_rate": 7.314213886469768e-05, "loss": 2.873, "step": 19918 }, { "epoch": 1.2365137500775965, "grad_norm": 0.6812497961411478, "learning_rate": 7.313893740343197e-05, "loss": 2.8323, "step": 19919 }, { "epoch": 1.2365758271773544, "grad_norm": 0.6129881042801093, "learning_rate": 7.31357358214448e-05, "loss": 2.7676, "step": 19920 }, { "epoch": 1.2366379042771123, "grad_norm": 0.5751971000293876, "learning_rate": 7.313253411875289e-05, "loss": 2.776, "step": 19921 }, { "epoch": 1.23669998137687, "grad_norm": 0.6035041931452301, "learning_rate": 7.312933229537294e-05, "loss": 2.8341, "step": 19922 }, { "epoch": 1.236762058476628, "grad_norm": 0.5768778985013708, "learning_rate": 7.312613035132165e-05, "loss": 2.8314, "step": 19923 }, { "epoch": 1.2368241355763858, "grad_norm": 0.5769766666514067, "learning_rate": 7.31229282866157e-05, "loss": 2.8211, "step": 19924 }, { "epoch": 1.2368862126761437, "grad_norm": 0.6588016810765153, "learning_rate": 7.311972610127183e-05, "loss": 2.8906, "step": 19925 }, { "epoch": 1.2369482897759017, "grad_norm": 0.5699071880955565, "learning_rate": 7.311652379530675e-05, "loss": 2.8326, "step": 19926 }, { "epoch": 1.2370103668756596, "grad_norm": 0.5822675340550281, "learning_rate": 7.311332136873714e-05, "loss": 2.762, "step": 19927 }, { "epoch": 1.2370724439754175, "grad_norm": 0.651306744334735, "learning_rate": 7.311011882157971e-05, "loss": 2.8504, "step": 19928 }, { "epoch": 1.2371345210751754, "grad_norm": 0.5949334597445889, "learning_rate": 7.310691615385122e-05, "loss": 2.782, "step": 19929 }, { "epoch": 1.2371965981749333, "grad_norm": 0.5350636311596977, "learning_rate": 7.31037133655683e-05, "loss": 2.8315, "step": 19930 }, { "epoch": 1.2372586752746912, "grad_norm": 0.6028727839154526, "learning_rate": 7.310051045674773e-05, "loss": 2.8216, "step": 19931 }, { "epoch": 1.2373207523744492, "grad_norm": 0.503019276759597, "learning_rate": 7.309730742740617e-05, "loss": 2.9153, "step": 19932 }, { "epoch": 1.2373828294742069, "grad_norm": 0.54108816824184, "learning_rate": 7.309410427756034e-05, "loss": 2.7523, "step": 19933 }, { "epoch": 1.2374449065739648, "grad_norm": 0.5237709236062501, "learning_rate": 7.309090100722698e-05, "loss": 2.8996, "step": 19934 }, { "epoch": 1.2375069836737227, "grad_norm": 0.5890146453603982, "learning_rate": 7.308769761642277e-05, "loss": 2.8047, "step": 19935 }, { "epoch": 1.2375690607734806, "grad_norm": 0.5910337456687961, "learning_rate": 7.308449410516446e-05, "loss": 2.892, "step": 19936 }, { "epoch": 1.2376311378732385, "grad_norm": 0.5489758872954001, "learning_rate": 7.308129047346872e-05, "loss": 2.8843, "step": 19937 }, { "epoch": 1.2376932149729964, "grad_norm": 0.5641228524675833, "learning_rate": 7.307808672135229e-05, "loss": 2.8717, "step": 19938 }, { "epoch": 1.2377552920727544, "grad_norm": 0.5419874861229794, "learning_rate": 7.307488284883188e-05, "loss": 2.8085, "step": 19939 }, { "epoch": 1.2378173691725123, "grad_norm": 0.5897880174188649, "learning_rate": 7.307167885592421e-05, "loss": 2.8129, "step": 19940 }, { "epoch": 1.2378794462722702, "grad_norm": 0.6236906641868315, "learning_rate": 7.306847474264598e-05, "loss": 2.8356, "step": 19941 }, { "epoch": 1.2379415233720281, "grad_norm": 0.6488164690545065, "learning_rate": 7.306527050901392e-05, "loss": 2.8196, "step": 19942 }, { "epoch": 1.238003600471786, "grad_norm": 0.5508645177354154, "learning_rate": 7.306206615504474e-05, "loss": 2.8762, "step": 19943 }, { "epoch": 1.238065677571544, "grad_norm": 0.5516234647866503, "learning_rate": 7.305886168075518e-05, "loss": 2.841, "step": 19944 }, { "epoch": 1.2381277546713019, "grad_norm": 0.5286818869437437, "learning_rate": 7.305565708616191e-05, "loss": 2.8416, "step": 19945 }, { "epoch": 1.2381898317710596, "grad_norm": 0.6019555331311338, "learning_rate": 7.305245237128171e-05, "loss": 2.8019, "step": 19946 }, { "epoch": 1.2382519088708175, "grad_norm": 0.5848414787882295, "learning_rate": 7.304924753613127e-05, "loss": 2.8172, "step": 19947 }, { "epoch": 1.2383139859705754, "grad_norm": 0.5615292756150312, "learning_rate": 7.304604258072729e-05, "loss": 2.795, "step": 19948 }, { "epoch": 1.2383760630703333, "grad_norm": 0.5763410416880581, "learning_rate": 7.304283750508653e-05, "loss": 2.8328, "step": 19949 }, { "epoch": 1.2384381401700912, "grad_norm": 0.5646254332589106, "learning_rate": 7.303963230922567e-05, "loss": 2.84, "step": 19950 }, { "epoch": 1.2385002172698492, "grad_norm": 0.5456659849139697, "learning_rate": 7.303642699316147e-05, "loss": 2.8165, "step": 19951 }, { "epoch": 1.238562294369607, "grad_norm": 0.6548312837310208, "learning_rate": 7.303322155691064e-05, "loss": 2.79, "step": 19952 }, { "epoch": 1.238624371469365, "grad_norm": 0.5867925381643042, "learning_rate": 7.30300160004899e-05, "loss": 2.8026, "step": 19953 }, { "epoch": 1.238686448569123, "grad_norm": 0.5904970656200115, "learning_rate": 7.302681032391597e-05, "loss": 2.7572, "step": 19954 }, { "epoch": 1.2387485256688808, "grad_norm": 0.5709909296952139, "learning_rate": 7.302360452720559e-05, "loss": 2.8087, "step": 19955 }, { "epoch": 1.2388106027686385, "grad_norm": 0.5756909966093945, "learning_rate": 7.302039861037546e-05, "loss": 2.9158, "step": 19956 }, { "epoch": 1.2388726798683964, "grad_norm": 0.5753714481221752, "learning_rate": 7.301719257344232e-05, "loss": 2.8358, "step": 19957 }, { "epoch": 1.2389347569681544, "grad_norm": 0.5469001491758472, "learning_rate": 7.301398641642294e-05, "loss": 2.8117, "step": 19958 }, { "epoch": 1.2389968340679123, "grad_norm": 0.5374998580577575, "learning_rate": 7.301078013933396e-05, "loss": 2.7547, "step": 19959 }, { "epoch": 1.2390589111676702, "grad_norm": 0.5455573147262824, "learning_rate": 7.300757374219218e-05, "loss": 2.8508, "step": 19960 }, { "epoch": 1.2391209882674281, "grad_norm": 0.5660620334046715, "learning_rate": 7.30043672250143e-05, "loss": 2.7773, "step": 19961 }, { "epoch": 1.239183065367186, "grad_norm": 0.6006255842319151, "learning_rate": 7.300116058781703e-05, "loss": 2.8923, "step": 19962 }, { "epoch": 1.239245142466944, "grad_norm": 0.5791175415510604, "learning_rate": 7.299795383061714e-05, "loss": 2.8754, "step": 19963 }, { "epoch": 1.2393072195667019, "grad_norm": 0.6209761550229126, "learning_rate": 7.299474695343133e-05, "loss": 2.8302, "step": 19964 }, { "epoch": 1.2393692966664598, "grad_norm": 0.5651063933672889, "learning_rate": 7.299153995627637e-05, "loss": 2.8842, "step": 19965 }, { "epoch": 1.2394313737662177, "grad_norm": 0.6540511851418755, "learning_rate": 7.298833283916893e-05, "loss": 2.8843, "step": 19966 }, { "epoch": 1.2394934508659756, "grad_norm": 0.5654180654013229, "learning_rate": 7.298512560212581e-05, "loss": 2.8044, "step": 19967 }, { "epoch": 1.2395555279657335, "grad_norm": 0.515398813980316, "learning_rate": 7.298191824516369e-05, "loss": 2.8415, "step": 19968 }, { "epoch": 1.2396176050654915, "grad_norm": 0.5655154095699558, "learning_rate": 7.297871076829932e-05, "loss": 2.8105, "step": 19969 }, { "epoch": 1.2396796821652492, "grad_norm": 0.5172272842640735, "learning_rate": 7.297550317154946e-05, "loss": 2.8167, "step": 19970 }, { "epoch": 1.239741759265007, "grad_norm": 0.5694779076732979, "learning_rate": 7.29722954549308e-05, "loss": 2.9158, "step": 19971 }, { "epoch": 1.239803836364765, "grad_norm": 0.5455414179507567, "learning_rate": 7.296908761846011e-05, "loss": 2.8986, "step": 19972 }, { "epoch": 1.239865913464523, "grad_norm": 0.601088944354678, "learning_rate": 7.296587966215409e-05, "loss": 2.9397, "step": 19973 }, { "epoch": 1.2399279905642808, "grad_norm": 0.5986239259940817, "learning_rate": 7.296267158602953e-05, "loss": 2.8434, "step": 19974 }, { "epoch": 1.2399900676640387, "grad_norm": 0.5024179169527561, "learning_rate": 7.295946339010312e-05, "loss": 2.8559, "step": 19975 }, { "epoch": 1.2400521447637967, "grad_norm": 0.6001443867999406, "learning_rate": 7.295625507439163e-05, "loss": 2.809, "step": 19976 }, { "epoch": 1.2401142218635546, "grad_norm": 0.5793721176536655, "learning_rate": 7.295304663891177e-05, "loss": 2.8281, "step": 19977 }, { "epoch": 1.2401762989633125, "grad_norm": 0.6018854983276742, "learning_rate": 7.294983808368032e-05, "loss": 2.8415, "step": 19978 }, { "epoch": 1.2402383760630704, "grad_norm": 0.5448470799240694, "learning_rate": 7.294662940871397e-05, "loss": 2.8275, "step": 19979 }, { "epoch": 1.240300453162828, "grad_norm": 0.5646008360942686, "learning_rate": 7.294342061402948e-05, "loss": 2.8733, "step": 19980 }, { "epoch": 1.240362530262586, "grad_norm": 0.5427123912237345, "learning_rate": 7.29402116996436e-05, "loss": 2.7965, "step": 19981 }, { "epoch": 1.240424607362344, "grad_norm": 0.6503170074863942, "learning_rate": 7.293700266557307e-05, "loss": 2.9006, "step": 19982 }, { "epoch": 1.2404866844621019, "grad_norm": 0.5501064739336621, "learning_rate": 7.293379351183465e-05, "loss": 2.7736, "step": 19983 }, { "epoch": 1.2405487615618598, "grad_norm": 0.597977869598726, "learning_rate": 7.293058423844504e-05, "loss": 2.8988, "step": 19984 }, { "epoch": 1.2406108386616177, "grad_norm": 0.5931970380523286, "learning_rate": 7.292737484542101e-05, "loss": 2.7898, "step": 19985 }, { "epoch": 1.2406729157613756, "grad_norm": 0.5611393419278218, "learning_rate": 7.292416533277928e-05, "loss": 2.9318, "step": 19986 }, { "epoch": 1.2407349928611335, "grad_norm": 0.5937985099250211, "learning_rate": 7.292095570053663e-05, "loss": 2.7814, "step": 19987 }, { "epoch": 1.2407970699608915, "grad_norm": 0.5463703824653029, "learning_rate": 7.291774594870979e-05, "loss": 2.7576, "step": 19988 }, { "epoch": 1.2408591470606494, "grad_norm": 0.5783695013223074, "learning_rate": 7.29145360773155e-05, "loss": 2.8696, "step": 19989 }, { "epoch": 1.2409212241604073, "grad_norm": 0.627537202758123, "learning_rate": 7.291132608637052e-05, "loss": 2.8438, "step": 19990 }, { "epoch": 1.2409833012601652, "grad_norm": 0.5773571172877957, "learning_rate": 7.290811597589158e-05, "loss": 2.8225, "step": 19991 }, { "epoch": 1.2410453783599231, "grad_norm": 0.5268851652782505, "learning_rate": 7.290490574589546e-05, "loss": 2.7954, "step": 19992 }, { "epoch": 1.241107455459681, "grad_norm": 0.5961680364753884, "learning_rate": 7.290169539639887e-05, "loss": 2.7942, "step": 19993 }, { "epoch": 1.2411695325594387, "grad_norm": 0.5178071864700962, "learning_rate": 7.289848492741859e-05, "loss": 2.7793, "step": 19994 }, { "epoch": 1.2412316096591967, "grad_norm": 0.605076841447822, "learning_rate": 7.289527433897133e-05, "loss": 2.9121, "step": 19995 }, { "epoch": 1.2412936867589546, "grad_norm": 0.5250849064787911, "learning_rate": 7.289206363107389e-05, "loss": 2.9002, "step": 19996 }, { "epoch": 1.2413557638587125, "grad_norm": 0.5522149277045814, "learning_rate": 7.288885280374299e-05, "loss": 2.797, "step": 19997 }, { "epoch": 1.2414178409584704, "grad_norm": 0.5462313269996045, "learning_rate": 7.288564185699537e-05, "loss": 2.7944, "step": 19998 }, { "epoch": 1.2414799180582283, "grad_norm": 0.6134572596581293, "learning_rate": 7.288243079084782e-05, "loss": 2.8189, "step": 19999 }, { "epoch": 1.2415419951579862, "grad_norm": 0.5792210216081487, "learning_rate": 7.287921960531708e-05, "loss": 2.8036, "step": 20000 }, { "epoch": 1.2416040722577442, "grad_norm": 0.5884752883163012, "learning_rate": 7.287600830041988e-05, "loss": 2.8538, "step": 20001 }, { "epoch": 1.241666149357502, "grad_norm": 0.5941102541171671, "learning_rate": 7.2872796876173e-05, "loss": 2.9072, "step": 20002 }, { "epoch": 1.24172822645726, "grad_norm": 0.5827897780747279, "learning_rate": 7.286958533259318e-05, "loss": 2.8434, "step": 20003 }, { "epoch": 1.2417903035570177, "grad_norm": 0.5699036844342946, "learning_rate": 7.286637366969718e-05, "loss": 2.7943, "step": 20004 }, { "epoch": 1.2418523806567756, "grad_norm": 0.5766882183020453, "learning_rate": 7.286316188750177e-05, "loss": 2.8299, "step": 20005 }, { "epoch": 1.2419144577565335, "grad_norm": 0.5187677208045306, "learning_rate": 7.285994998602369e-05, "loss": 2.7707, "step": 20006 }, { "epoch": 1.2419765348562914, "grad_norm": 0.642384833393299, "learning_rate": 7.28567379652797e-05, "loss": 2.9261, "step": 20007 }, { "epoch": 1.2420386119560494, "grad_norm": 0.48230249263677705, "learning_rate": 7.285352582528655e-05, "loss": 2.838, "step": 20008 }, { "epoch": 1.2421006890558073, "grad_norm": 0.6268005898565379, "learning_rate": 7.285031356606101e-05, "loss": 2.8691, "step": 20009 }, { "epoch": 1.2421627661555652, "grad_norm": 0.5103878447770638, "learning_rate": 7.284710118761986e-05, "loss": 2.82, "step": 20010 }, { "epoch": 1.2422248432553231, "grad_norm": 0.5561363211208474, "learning_rate": 7.284388868997981e-05, "loss": 2.8119, "step": 20011 }, { "epoch": 1.242286920355081, "grad_norm": 0.5937916590481577, "learning_rate": 7.284067607315765e-05, "loss": 2.8918, "step": 20012 }, { "epoch": 1.242348997454839, "grad_norm": 0.5698008659173446, "learning_rate": 7.283746333717015e-05, "loss": 2.8068, "step": 20013 }, { "epoch": 1.2424110745545969, "grad_norm": 0.6043149722249255, "learning_rate": 7.283425048203405e-05, "loss": 2.8204, "step": 20014 }, { "epoch": 1.2424731516543548, "grad_norm": 0.5734489626869175, "learning_rate": 7.283103750776613e-05, "loss": 2.8694, "step": 20015 }, { "epoch": 1.2425352287541127, "grad_norm": 0.5130072235264378, "learning_rate": 7.282782441438315e-05, "loss": 2.8098, "step": 20016 }, { "epoch": 1.2425973058538706, "grad_norm": 0.527966082663006, "learning_rate": 7.282461120190185e-05, "loss": 2.7933, "step": 20017 }, { "epoch": 1.2426593829536283, "grad_norm": 0.6010630131660104, "learning_rate": 7.282139787033904e-05, "loss": 2.8217, "step": 20018 }, { "epoch": 1.2427214600533862, "grad_norm": 0.5641614695027423, "learning_rate": 7.281818441971143e-05, "loss": 2.8848, "step": 20019 }, { "epoch": 1.2427835371531442, "grad_norm": 0.5645383090779674, "learning_rate": 7.281497085003582e-05, "loss": 2.9302, "step": 20020 }, { "epoch": 1.242845614252902, "grad_norm": 0.5375407203508482, "learning_rate": 7.281175716132897e-05, "loss": 2.8055, "step": 20021 }, { "epoch": 1.24290769135266, "grad_norm": 0.5941100785468196, "learning_rate": 7.280854335360764e-05, "loss": 2.8403, "step": 20022 }, { "epoch": 1.242969768452418, "grad_norm": 0.5802416843802938, "learning_rate": 7.280532942688862e-05, "loss": 2.7963, "step": 20023 }, { "epoch": 1.2430318455521758, "grad_norm": 0.5248225377649205, "learning_rate": 7.280211538118864e-05, "loss": 2.8689, "step": 20024 }, { "epoch": 1.2430939226519337, "grad_norm": 0.6121792216030891, "learning_rate": 7.27989012165245e-05, "loss": 2.86, "step": 20025 }, { "epoch": 1.2431559997516917, "grad_norm": 0.5810176036124466, "learning_rate": 7.279568693291296e-05, "loss": 2.8433, "step": 20026 }, { "epoch": 1.2432180768514496, "grad_norm": 0.7271637428511107, "learning_rate": 7.279247253037078e-05, "loss": 2.7916, "step": 20027 }, { "epoch": 1.2432801539512073, "grad_norm": 0.5908727701402088, "learning_rate": 7.278925800891474e-05, "loss": 2.7856, "step": 20028 }, { "epoch": 1.2433422310509652, "grad_norm": 0.5994578614303221, "learning_rate": 7.278604336856161e-05, "loss": 2.7549, "step": 20029 }, { "epoch": 1.2434043081507231, "grad_norm": 0.5509165063451094, "learning_rate": 7.278282860932817e-05, "loss": 2.8778, "step": 20030 }, { "epoch": 1.243466385250481, "grad_norm": 0.6032142009886694, "learning_rate": 7.277961373123115e-05, "loss": 2.9016, "step": 20031 }, { "epoch": 1.243528462350239, "grad_norm": 0.5568983585583496, "learning_rate": 7.277639873428739e-05, "loss": 2.8786, "step": 20032 }, { "epoch": 1.2435905394499969, "grad_norm": 0.6113535720355189, "learning_rate": 7.277318361851361e-05, "loss": 2.9275, "step": 20033 }, { "epoch": 1.2436526165497548, "grad_norm": 0.5354239914785173, "learning_rate": 7.276996838392662e-05, "loss": 2.8851, "step": 20034 }, { "epoch": 1.2437146936495127, "grad_norm": 0.5294535218589571, "learning_rate": 7.276675303054316e-05, "loss": 2.8138, "step": 20035 }, { "epoch": 1.2437767707492706, "grad_norm": 0.5489192086114553, "learning_rate": 7.276353755838002e-05, "loss": 2.9061, "step": 20036 }, { "epoch": 1.2438388478490285, "grad_norm": 0.5445480431658583, "learning_rate": 7.276032196745399e-05, "loss": 2.7693, "step": 20037 }, { "epoch": 1.2439009249487865, "grad_norm": 0.5419112411400282, "learning_rate": 7.275710625778182e-05, "loss": 2.8245, "step": 20038 }, { "epoch": 1.2439630020485444, "grad_norm": 0.5609326248169609, "learning_rate": 7.275389042938032e-05, "loss": 2.8742, "step": 20039 }, { "epoch": 1.2440250791483023, "grad_norm": 0.5598896174122768, "learning_rate": 7.275067448226624e-05, "loss": 2.9329, "step": 20040 }, { "epoch": 1.2440871562480602, "grad_norm": 0.5095091426654244, "learning_rate": 7.274745841645638e-05, "loss": 2.8822, "step": 20041 }, { "epoch": 1.244149233347818, "grad_norm": 0.580912160623458, "learning_rate": 7.27442422319675e-05, "loss": 2.8458, "step": 20042 }, { "epoch": 1.2442113104475758, "grad_norm": 0.5460169191027201, "learning_rate": 7.274102592881639e-05, "loss": 2.8435, "step": 20043 }, { "epoch": 1.2442733875473337, "grad_norm": 0.5922411771014856, "learning_rate": 7.273780950701983e-05, "loss": 2.7818, "step": 20044 }, { "epoch": 1.2443354646470917, "grad_norm": 0.5499529189852831, "learning_rate": 7.273459296659458e-05, "loss": 2.8043, "step": 20045 }, { "epoch": 1.2443975417468496, "grad_norm": 0.5353830511274662, "learning_rate": 7.273137630755747e-05, "loss": 2.8651, "step": 20046 }, { "epoch": 1.2444596188466075, "grad_norm": 0.4734125470676791, "learning_rate": 7.272815952992523e-05, "loss": 2.751, "step": 20047 }, { "epoch": 1.2445216959463654, "grad_norm": 0.5639963644887885, "learning_rate": 7.272494263371466e-05, "loss": 2.8286, "step": 20048 }, { "epoch": 1.2445837730461233, "grad_norm": 0.5569922424183684, "learning_rate": 7.272172561894255e-05, "loss": 2.8512, "step": 20049 }, { "epoch": 1.2446458501458812, "grad_norm": 0.5370667041821326, "learning_rate": 7.27185084856257e-05, "loss": 2.833, "step": 20050 }, { "epoch": 1.2447079272456392, "grad_norm": 0.7119978239288723, "learning_rate": 7.271529123378085e-05, "loss": 2.8858, "step": 20051 }, { "epoch": 1.2447700043453969, "grad_norm": 0.5439660498126424, "learning_rate": 7.271207386342482e-05, "loss": 2.8139, "step": 20052 }, { "epoch": 1.2448320814451548, "grad_norm": 0.6018608889009979, "learning_rate": 7.270885637457439e-05, "loss": 2.7521, "step": 20053 }, { "epoch": 1.2448941585449127, "grad_norm": 0.5985023891637117, "learning_rate": 7.270563876724635e-05, "loss": 2.8449, "step": 20054 }, { "epoch": 1.2449562356446706, "grad_norm": 0.5326988714325417, "learning_rate": 7.270242104145747e-05, "loss": 2.891, "step": 20055 }, { "epoch": 1.2450183127444285, "grad_norm": 0.6187199989906219, "learning_rate": 7.269920319722454e-05, "loss": 2.8816, "step": 20056 }, { "epoch": 1.2450803898441865, "grad_norm": 0.5789712691763664, "learning_rate": 7.269598523456438e-05, "loss": 2.862, "step": 20057 }, { "epoch": 1.2451424669439444, "grad_norm": 0.6541123998409965, "learning_rate": 7.269276715349374e-05, "loss": 2.9159, "step": 20058 }, { "epoch": 1.2452045440437023, "grad_norm": 0.5824508178806056, "learning_rate": 7.268954895402943e-05, "loss": 2.8581, "step": 20059 }, { "epoch": 1.2452666211434602, "grad_norm": 0.5644105149391991, "learning_rate": 7.268633063618824e-05, "loss": 2.8237, "step": 20060 }, { "epoch": 1.2453286982432181, "grad_norm": 0.6080205977897282, "learning_rate": 7.268311219998694e-05, "loss": 2.7716, "step": 20061 }, { "epoch": 1.245390775342976, "grad_norm": 0.5434854642245545, "learning_rate": 7.267989364544235e-05, "loss": 2.7653, "step": 20062 }, { "epoch": 1.245452852442734, "grad_norm": 0.566054162397767, "learning_rate": 7.267667497257125e-05, "loss": 2.8203, "step": 20063 }, { "epoch": 1.2455149295424919, "grad_norm": 0.5919128153320901, "learning_rate": 7.267345618139043e-05, "loss": 2.7748, "step": 20064 }, { "epoch": 1.2455770066422498, "grad_norm": 0.545493014606467, "learning_rate": 7.267023727191668e-05, "loss": 2.7558, "step": 20065 }, { "epoch": 1.2456390837420075, "grad_norm": 0.646519247998784, "learning_rate": 7.266701824416681e-05, "loss": 2.8262, "step": 20066 }, { "epoch": 1.2457011608417654, "grad_norm": 0.6078855946186478, "learning_rate": 7.266379909815759e-05, "loss": 2.8505, "step": 20067 }, { "epoch": 1.2457632379415233, "grad_norm": 0.66976524279152, "learning_rate": 7.266057983390584e-05, "loss": 2.8691, "step": 20068 }, { "epoch": 1.2458253150412812, "grad_norm": 0.5777011038979813, "learning_rate": 7.265736045142834e-05, "loss": 2.7637, "step": 20069 }, { "epoch": 1.2458873921410392, "grad_norm": 0.628330773007702, "learning_rate": 7.265414095074188e-05, "loss": 2.9222, "step": 20070 }, { "epoch": 1.245949469240797, "grad_norm": 0.5080580191201716, "learning_rate": 7.265092133186328e-05, "loss": 2.7854, "step": 20071 }, { "epoch": 1.246011546340555, "grad_norm": 0.5462919396770324, "learning_rate": 7.264770159480932e-05, "loss": 2.9627, "step": 20072 }, { "epoch": 1.246073623440313, "grad_norm": 0.5444804492169281, "learning_rate": 7.264448173959679e-05, "loss": 2.8658, "step": 20073 }, { "epoch": 1.2461357005400708, "grad_norm": 0.5905984004656236, "learning_rate": 7.264126176624252e-05, "loss": 2.8454, "step": 20074 }, { "epoch": 1.2461977776398288, "grad_norm": 0.5665132881838325, "learning_rate": 7.26380416747633e-05, "loss": 2.8352, "step": 20075 }, { "epoch": 1.2462598547395864, "grad_norm": 0.514745366561512, "learning_rate": 7.263482146517588e-05, "loss": 2.852, "step": 20076 }, { "epoch": 1.2463219318393444, "grad_norm": 0.6792486242758445, "learning_rate": 7.263160113749714e-05, "loss": 2.9061, "step": 20077 }, { "epoch": 1.2463840089391023, "grad_norm": 0.6032789933074263, "learning_rate": 7.262838069174382e-05, "loss": 2.8522, "step": 20078 }, { "epoch": 1.2464460860388602, "grad_norm": 0.6451849887661931, "learning_rate": 7.262516012793277e-05, "loss": 2.8036, "step": 20079 }, { "epoch": 1.2465081631386181, "grad_norm": 0.5669907809354054, "learning_rate": 7.262193944608075e-05, "loss": 2.8686, "step": 20080 }, { "epoch": 1.246570240238376, "grad_norm": 0.6713156034779479, "learning_rate": 7.261871864620457e-05, "loss": 2.8353, "step": 20081 }, { "epoch": 1.246632317338134, "grad_norm": 0.5708800065898761, "learning_rate": 7.261549772832106e-05, "loss": 2.8133, "step": 20082 }, { "epoch": 1.2466943944378919, "grad_norm": 0.5957326912968834, "learning_rate": 7.2612276692447e-05, "loss": 2.8829, "step": 20083 }, { "epoch": 1.2467564715376498, "grad_norm": 0.5751792762994984, "learning_rate": 7.260905553859923e-05, "loss": 2.8726, "step": 20084 }, { "epoch": 1.2468185486374077, "grad_norm": 0.7108071595486665, "learning_rate": 7.260583426679451e-05, "loss": 2.8689, "step": 20085 }, { "epoch": 1.2468806257371656, "grad_norm": 0.5554088415361111, "learning_rate": 7.260261287704965e-05, "loss": 2.7934, "step": 20086 }, { "epoch": 1.2469427028369235, "grad_norm": 0.6361692432797449, "learning_rate": 7.259939136938147e-05, "loss": 2.714, "step": 20087 }, { "epoch": 1.2470047799366815, "grad_norm": 0.6110500641757466, "learning_rate": 7.25961697438068e-05, "loss": 2.8665, "step": 20088 }, { "epoch": 1.2470668570364394, "grad_norm": 0.58663773974264, "learning_rate": 7.259294800034242e-05, "loss": 2.8939, "step": 20089 }, { "epoch": 1.247128934136197, "grad_norm": 0.6780446246633186, "learning_rate": 7.258972613900515e-05, "loss": 2.8434, "step": 20090 }, { "epoch": 1.247191011235955, "grad_norm": 0.648244392411958, "learning_rate": 7.258650415981178e-05, "loss": 2.7171, "step": 20091 }, { "epoch": 1.247253088335713, "grad_norm": 0.5689152362284547, "learning_rate": 7.258328206277915e-05, "loss": 2.7749, "step": 20092 }, { "epoch": 1.2473151654354708, "grad_norm": 0.6971937938630087, "learning_rate": 7.258005984792405e-05, "loss": 2.883, "step": 20093 }, { "epoch": 1.2473772425352287, "grad_norm": 0.6368611416417959, "learning_rate": 7.25768375152633e-05, "loss": 2.9303, "step": 20094 }, { "epoch": 1.2474393196349867, "grad_norm": 0.6712673233680961, "learning_rate": 7.257361506481371e-05, "loss": 2.8777, "step": 20095 }, { "epoch": 1.2475013967347446, "grad_norm": 0.6830088483689667, "learning_rate": 7.257039249659208e-05, "loss": 2.8776, "step": 20096 }, { "epoch": 1.2475634738345025, "grad_norm": 0.6374499796896765, "learning_rate": 7.256716981061523e-05, "loss": 2.8553, "step": 20097 }, { "epoch": 1.2476255509342604, "grad_norm": 0.6606647406600352, "learning_rate": 7.256394700689998e-05, "loss": 2.8047, "step": 20098 }, { "epoch": 1.2476876280340183, "grad_norm": 0.5685457009404982, "learning_rate": 7.256072408546315e-05, "loss": 2.7613, "step": 20099 }, { "epoch": 1.247749705133776, "grad_norm": 0.6260106022427502, "learning_rate": 7.255750104632156e-05, "loss": 2.7448, "step": 20100 }, { "epoch": 1.247811782233534, "grad_norm": 0.6470192716107407, "learning_rate": 7.255427788949199e-05, "loss": 2.8594, "step": 20101 }, { "epoch": 1.2478738593332919, "grad_norm": 0.6588274883754373, "learning_rate": 7.255105461499127e-05, "loss": 2.8064, "step": 20102 }, { "epoch": 1.2479359364330498, "grad_norm": 0.5696852906101091, "learning_rate": 7.254783122283623e-05, "loss": 2.9371, "step": 20103 }, { "epoch": 1.2479980135328077, "grad_norm": 0.5615620740783809, "learning_rate": 7.254460771304368e-05, "loss": 2.8021, "step": 20104 }, { "epoch": 1.2480600906325656, "grad_norm": 0.5217871769087951, "learning_rate": 7.254138408563044e-05, "loss": 2.852, "step": 20105 }, { "epoch": 1.2481221677323235, "grad_norm": 0.5846524961790344, "learning_rate": 7.253816034061332e-05, "loss": 2.8509, "step": 20106 }, { "epoch": 1.2481842448320815, "grad_norm": 0.6014993807883813, "learning_rate": 7.253493647800917e-05, "loss": 2.8532, "step": 20107 }, { "epoch": 1.2482463219318394, "grad_norm": 0.6178703876055412, "learning_rate": 7.253171249783477e-05, "loss": 2.9083, "step": 20108 }, { "epoch": 1.2483083990315973, "grad_norm": 0.6081480977689246, "learning_rate": 7.252848840010696e-05, "loss": 2.8473, "step": 20109 }, { "epoch": 1.2483704761313552, "grad_norm": 0.7025524351296477, "learning_rate": 7.252526418484254e-05, "loss": 2.862, "step": 20110 }, { "epoch": 1.2484325532311131, "grad_norm": 0.6218681306315816, "learning_rate": 7.252203985205837e-05, "loss": 2.8376, "step": 20111 }, { "epoch": 1.248494630330871, "grad_norm": 0.6283736254053085, "learning_rate": 7.251881540177125e-05, "loss": 2.81, "step": 20112 }, { "epoch": 1.248556707430629, "grad_norm": 0.6809805704739691, "learning_rate": 7.2515590833998e-05, "loss": 2.8492, "step": 20113 }, { "epoch": 1.2486187845303867, "grad_norm": 0.5825647047230517, "learning_rate": 7.251236614875543e-05, "loss": 2.8252, "step": 20114 }, { "epoch": 1.2486808616301446, "grad_norm": 0.6618861044221483, "learning_rate": 7.25091413460604e-05, "loss": 2.8354, "step": 20115 }, { "epoch": 1.2487429387299025, "grad_norm": 0.5928280348041854, "learning_rate": 7.250591642592971e-05, "loss": 2.8659, "step": 20116 }, { "epoch": 1.2488050158296604, "grad_norm": 0.6371071221962585, "learning_rate": 7.25026913883802e-05, "loss": 2.8339, "step": 20117 }, { "epoch": 1.2488670929294183, "grad_norm": 0.646533722161012, "learning_rate": 7.249946623342867e-05, "loss": 2.7878, "step": 20118 }, { "epoch": 1.2489291700291762, "grad_norm": 0.5814737555812767, "learning_rate": 7.249624096109196e-05, "loss": 2.8811, "step": 20119 }, { "epoch": 1.2489912471289342, "grad_norm": 0.5457123522841927, "learning_rate": 7.249301557138691e-05, "loss": 2.7593, "step": 20120 }, { "epoch": 1.249053324228692, "grad_norm": 0.6268349176601696, "learning_rate": 7.248979006433033e-05, "loss": 2.8985, "step": 20121 }, { "epoch": 1.24911540132845, "grad_norm": 0.5681424070908233, "learning_rate": 7.248656443993907e-05, "loss": 2.825, "step": 20122 }, { "epoch": 1.249177478428208, "grad_norm": 0.5487438422272493, "learning_rate": 7.248333869822993e-05, "loss": 2.8459, "step": 20123 }, { "epoch": 1.2492395555279656, "grad_norm": 0.540741200169579, "learning_rate": 7.248011283921976e-05, "loss": 2.818, "step": 20124 }, { "epoch": 1.2493016326277235, "grad_norm": 0.5548391806377532, "learning_rate": 7.24768868629254e-05, "loss": 2.8524, "step": 20125 }, { "epoch": 1.2493637097274815, "grad_norm": 0.5921447285527763, "learning_rate": 7.247366076936363e-05, "loss": 2.8876, "step": 20126 }, { "epoch": 1.2494257868272394, "grad_norm": 0.5269326479632651, "learning_rate": 7.247043455855135e-05, "loss": 2.8943, "step": 20127 }, { "epoch": 1.2494878639269973, "grad_norm": 0.6598165642614162, "learning_rate": 7.246720823050534e-05, "loss": 2.7898, "step": 20128 }, { "epoch": 1.2495499410267552, "grad_norm": 0.5511326311551725, "learning_rate": 7.246398178524245e-05, "loss": 2.8048, "step": 20129 }, { "epoch": 1.2496120181265131, "grad_norm": 0.5949857801453193, "learning_rate": 7.246075522277952e-05, "loss": 2.837, "step": 20130 }, { "epoch": 1.249674095226271, "grad_norm": 0.5646066159964388, "learning_rate": 7.245752854313337e-05, "loss": 2.9177, "step": 20131 }, { "epoch": 1.249736172326029, "grad_norm": 0.5989706010529655, "learning_rate": 7.245430174632085e-05, "loss": 2.8952, "step": 20132 }, { "epoch": 1.2497982494257869, "grad_norm": 0.640010469209456, "learning_rate": 7.245107483235879e-05, "loss": 2.9265, "step": 20133 }, { "epoch": 1.2498603265255448, "grad_norm": 0.6221321829744665, "learning_rate": 7.244784780126401e-05, "loss": 2.9238, "step": 20134 }, { "epoch": 1.2499224036253027, "grad_norm": 0.6209406874046437, "learning_rate": 7.244462065305338e-05, "loss": 2.8684, "step": 20135 }, { "epoch": 1.2499844807250606, "grad_norm": 0.6648762149258217, "learning_rate": 7.244139338774369e-05, "loss": 2.8151, "step": 20136 }, { "epoch": 1.2500465578248185, "grad_norm": 0.6677226455352748, "learning_rate": 7.24381660053518e-05, "loss": 2.7842, "step": 20137 }, { "epoch": 1.2501086349245762, "grad_norm": 0.6567604032853473, "learning_rate": 7.243493850589457e-05, "loss": 2.8678, "step": 20138 }, { "epoch": 1.2501707120243342, "grad_norm": 0.7206122695182161, "learning_rate": 7.243171088938882e-05, "loss": 2.789, "step": 20139 }, { "epoch": 1.250232789124092, "grad_norm": 0.597610945636276, "learning_rate": 7.242848315585138e-05, "loss": 2.7991, "step": 20140 }, { "epoch": 1.25029486622385, "grad_norm": 0.7204512903054087, "learning_rate": 7.24252553052991e-05, "loss": 2.6779, "step": 20141 }, { "epoch": 1.250356943323608, "grad_norm": 0.5745385764959199, "learning_rate": 7.242202733774882e-05, "loss": 2.8387, "step": 20142 }, { "epoch": 1.2504190204233658, "grad_norm": 0.5701731420112284, "learning_rate": 7.24187992532174e-05, "loss": 2.8115, "step": 20143 }, { "epoch": 1.2504810975231238, "grad_norm": 0.5520007653542064, "learning_rate": 7.241557105172163e-05, "loss": 2.8673, "step": 20144 }, { "epoch": 1.2505431746228817, "grad_norm": 0.5782915854799497, "learning_rate": 7.241234273327839e-05, "loss": 2.8913, "step": 20145 }, { "epoch": 1.2506052517226396, "grad_norm": 0.5549139984186248, "learning_rate": 7.240911429790452e-05, "loss": 2.7534, "step": 20146 }, { "epoch": 1.2506673288223973, "grad_norm": 0.5655321413817176, "learning_rate": 7.240588574561688e-05, "loss": 2.8061, "step": 20147 }, { "epoch": 1.2507294059221552, "grad_norm": 0.628068092956755, "learning_rate": 7.240265707643226e-05, "loss": 2.8386, "step": 20148 }, { "epoch": 1.2507914830219131, "grad_norm": 0.5992684901953282, "learning_rate": 7.239942829036758e-05, "loss": 2.8543, "step": 20149 }, { "epoch": 1.250853560121671, "grad_norm": 0.5605345934010875, "learning_rate": 7.239619938743961e-05, "loss": 2.7522, "step": 20150 }, { "epoch": 1.250915637221429, "grad_norm": 0.51828527546177, "learning_rate": 7.239297036766526e-05, "loss": 2.8274, "step": 20151 }, { "epoch": 1.2509777143211869, "grad_norm": 0.5741208310880156, "learning_rate": 7.238974123106133e-05, "loss": 2.8013, "step": 20152 }, { "epoch": 1.2510397914209448, "grad_norm": 0.5662195095164746, "learning_rate": 7.238651197764468e-05, "loss": 2.7848, "step": 20153 }, { "epoch": 1.2511018685207027, "grad_norm": 0.5390937906665155, "learning_rate": 7.238328260743217e-05, "loss": 2.8645, "step": 20154 }, { "epoch": 1.2511639456204606, "grad_norm": 0.6171150889351267, "learning_rate": 7.238005312044063e-05, "loss": 2.8652, "step": 20155 }, { "epoch": 1.2512260227202185, "grad_norm": 0.5095647072474627, "learning_rate": 7.237682351668694e-05, "loss": 2.8785, "step": 20156 }, { "epoch": 1.2512880998199765, "grad_norm": 0.5873084750368912, "learning_rate": 7.237359379618791e-05, "loss": 2.8123, "step": 20157 }, { "epoch": 1.2513501769197344, "grad_norm": 0.5522945527824279, "learning_rate": 7.237036395896044e-05, "loss": 2.8399, "step": 20158 }, { "epoch": 1.2514122540194923, "grad_norm": 0.5695964027678291, "learning_rate": 7.23671340050213e-05, "loss": 2.8617, "step": 20159 }, { "epoch": 1.2514743311192502, "grad_norm": 0.5408681124430932, "learning_rate": 7.236390393438745e-05, "loss": 2.8574, "step": 20160 }, { "epoch": 1.2515364082190081, "grad_norm": 0.5631878409003116, "learning_rate": 7.236067374707565e-05, "loss": 2.8703, "step": 20161 }, { "epoch": 1.2515984853187658, "grad_norm": 0.5653551279826804, "learning_rate": 7.235744344310279e-05, "loss": 2.8507, "step": 20162 }, { "epoch": 1.2516605624185237, "grad_norm": 0.5156185554332849, "learning_rate": 7.235421302248572e-05, "loss": 2.7943, "step": 20163 }, { "epoch": 1.2517226395182817, "grad_norm": 0.5079592566070956, "learning_rate": 7.23509824852413e-05, "loss": 2.8412, "step": 20164 }, { "epoch": 1.2517847166180396, "grad_norm": 0.5448114973826809, "learning_rate": 7.234775183138638e-05, "loss": 2.9492, "step": 20165 }, { "epoch": 1.2518467937177975, "grad_norm": 0.5293618475531047, "learning_rate": 7.23445210609378e-05, "loss": 2.8322, "step": 20166 }, { "epoch": 1.2519088708175554, "grad_norm": 0.5296497927947366, "learning_rate": 7.234129017391245e-05, "loss": 2.8971, "step": 20167 }, { "epoch": 1.2519709479173133, "grad_norm": 0.5518861874475717, "learning_rate": 7.233805917032715e-05, "loss": 2.87, "step": 20168 }, { "epoch": 1.2520330250170713, "grad_norm": 0.4741279766553532, "learning_rate": 7.233482805019878e-05, "loss": 2.7673, "step": 20169 }, { "epoch": 1.2520951021168292, "grad_norm": 0.5036311618544438, "learning_rate": 7.233159681354419e-05, "loss": 2.7409, "step": 20170 }, { "epoch": 1.2521571792165869, "grad_norm": 0.5323027949902462, "learning_rate": 7.232836546038022e-05, "loss": 2.8575, "step": 20171 }, { "epoch": 1.2522192563163448, "grad_norm": 0.5433185191806765, "learning_rate": 7.232513399072377e-05, "loss": 2.8463, "step": 20172 }, { "epoch": 1.2522813334161027, "grad_norm": 0.574199676196885, "learning_rate": 7.232190240459167e-05, "loss": 2.8789, "step": 20173 }, { "epoch": 1.2523434105158606, "grad_norm": 0.5250915754014741, "learning_rate": 7.23186707020008e-05, "loss": 2.8187, "step": 20174 }, { "epoch": 1.2524054876156185, "grad_norm": 0.577635764268746, "learning_rate": 7.231543888296797e-05, "loss": 2.8884, "step": 20175 }, { "epoch": 1.2524675647153765, "grad_norm": 0.5469239349269771, "learning_rate": 7.231220694751011e-05, "loss": 2.7718, "step": 20176 }, { "epoch": 1.2525296418151344, "grad_norm": 0.5111318819287538, "learning_rate": 7.230897489564403e-05, "loss": 2.9025, "step": 20177 }, { "epoch": 1.2525917189148923, "grad_norm": 0.5499776022858827, "learning_rate": 7.230574272738664e-05, "loss": 2.8603, "step": 20178 }, { "epoch": 1.2526537960146502, "grad_norm": 0.4860897935628434, "learning_rate": 7.230251044275476e-05, "loss": 2.815, "step": 20179 }, { "epoch": 1.2527158731144081, "grad_norm": 0.5524797965364692, "learning_rate": 7.229927804176526e-05, "loss": 2.8175, "step": 20180 }, { "epoch": 1.252777950214166, "grad_norm": 0.6117981636268588, "learning_rate": 7.229604552443502e-05, "loss": 2.7968, "step": 20181 }, { "epoch": 1.252840027313924, "grad_norm": 0.51437897839404, "learning_rate": 7.22928128907809e-05, "loss": 2.8438, "step": 20182 }, { "epoch": 1.2529021044136819, "grad_norm": 0.5561014341341708, "learning_rate": 7.228958014081975e-05, "loss": 2.799, "step": 20183 }, { "epoch": 1.2529641815134398, "grad_norm": 0.5344567286583611, "learning_rate": 7.228634727456847e-05, "loss": 2.8804, "step": 20184 }, { "epoch": 1.2530262586131977, "grad_norm": 0.6023245356917735, "learning_rate": 7.22831142920439e-05, "loss": 2.8666, "step": 20185 }, { "epoch": 1.2530883357129554, "grad_norm": 0.5478619797204091, "learning_rate": 7.22798811932629e-05, "loss": 2.8544, "step": 20186 }, { "epoch": 1.2531504128127133, "grad_norm": 0.5795803053214831, "learning_rate": 7.227664797824236e-05, "loss": 2.8527, "step": 20187 }, { "epoch": 1.2532124899124713, "grad_norm": 0.5116358391372957, "learning_rate": 7.227341464699914e-05, "loss": 2.7998, "step": 20188 }, { "epoch": 1.2532745670122292, "grad_norm": 0.6392601173981706, "learning_rate": 7.22701811995501e-05, "loss": 2.6861, "step": 20189 }, { "epoch": 1.253336644111987, "grad_norm": 0.5100942261454154, "learning_rate": 7.226694763591214e-05, "loss": 2.7582, "step": 20190 }, { "epoch": 1.253398721211745, "grad_norm": 0.5688457439865257, "learning_rate": 7.226371395610208e-05, "loss": 2.7972, "step": 20191 }, { "epoch": 1.253460798311503, "grad_norm": 0.6333577667927373, "learning_rate": 7.226048016013685e-05, "loss": 2.9285, "step": 20192 }, { "epoch": 1.2535228754112608, "grad_norm": 0.5398617499921049, "learning_rate": 7.225724624803327e-05, "loss": 2.8392, "step": 20193 }, { "epoch": 1.2535849525110188, "grad_norm": 0.557201089955364, "learning_rate": 7.225401221980826e-05, "loss": 2.8568, "step": 20194 }, { "epoch": 1.2536470296107765, "grad_norm": 0.6247539035756071, "learning_rate": 7.225077807547863e-05, "loss": 2.9072, "step": 20195 }, { "epoch": 1.2537091067105344, "grad_norm": 0.5289081962245368, "learning_rate": 7.224754381506132e-05, "loss": 2.8561, "step": 20196 }, { "epoch": 1.2537711838102923, "grad_norm": 0.5981398228051289, "learning_rate": 7.224430943857315e-05, "loss": 2.9226, "step": 20197 }, { "epoch": 1.2538332609100502, "grad_norm": 0.5600397028642989, "learning_rate": 7.224107494603103e-05, "loss": 2.9306, "step": 20198 }, { "epoch": 1.2538953380098081, "grad_norm": 0.5687749857182669, "learning_rate": 7.223784033745183e-05, "loss": 2.7693, "step": 20199 }, { "epoch": 1.253957415109566, "grad_norm": 0.5413976208988617, "learning_rate": 7.223460561285239e-05, "loss": 2.8178, "step": 20200 }, { "epoch": 1.254019492209324, "grad_norm": 0.5150084421706596, "learning_rate": 7.223137077224963e-05, "loss": 2.8696, "step": 20201 }, { "epoch": 1.2540815693090819, "grad_norm": 0.5873713433638343, "learning_rate": 7.222813581566041e-05, "loss": 2.7668, "step": 20202 }, { "epoch": 1.2541436464088398, "grad_norm": 0.5137004771809054, "learning_rate": 7.222490074310163e-05, "loss": 2.807, "step": 20203 }, { "epoch": 1.2542057235085977, "grad_norm": 0.6324990991729232, "learning_rate": 7.222166555459012e-05, "loss": 2.8168, "step": 20204 }, { "epoch": 1.2542678006083556, "grad_norm": 0.6152784785856796, "learning_rate": 7.221843025014278e-05, "loss": 2.8961, "step": 20205 }, { "epoch": 1.2543298777081135, "grad_norm": 0.5218948006442768, "learning_rate": 7.221519482977654e-05, "loss": 2.9341, "step": 20206 }, { "epoch": 1.2543919548078715, "grad_norm": 0.6375964849049084, "learning_rate": 7.221195929350819e-05, "loss": 2.8708, "step": 20207 }, { "epoch": 1.2544540319076294, "grad_norm": 0.5108574760834584, "learning_rate": 7.22087236413547e-05, "loss": 2.8471, "step": 20208 }, { "epoch": 1.2545161090073873, "grad_norm": 0.5615060289421032, "learning_rate": 7.220548787333287e-05, "loss": 2.9593, "step": 20209 }, { "epoch": 1.254578186107145, "grad_norm": 0.5788745402970833, "learning_rate": 7.220225198945962e-05, "loss": 2.8022, "step": 20210 }, { "epoch": 1.254640263206903, "grad_norm": 0.5289500321223535, "learning_rate": 7.219901598975184e-05, "loss": 2.895, "step": 20211 }, { "epoch": 1.2547023403066608, "grad_norm": 0.5789616690813321, "learning_rate": 7.219577987422642e-05, "loss": 2.7661, "step": 20212 }, { "epoch": 1.2547644174064188, "grad_norm": 0.585466857084498, "learning_rate": 7.219254364290022e-05, "loss": 2.8441, "step": 20213 }, { "epoch": 1.2548264945061767, "grad_norm": 0.5750755612855555, "learning_rate": 7.218930729579011e-05, "loss": 2.8968, "step": 20214 }, { "epoch": 1.2548885716059346, "grad_norm": 0.607578637602488, "learning_rate": 7.218607083291304e-05, "loss": 2.85, "step": 20215 }, { "epoch": 1.2549506487056925, "grad_norm": 0.5605975935583218, "learning_rate": 7.218283425428582e-05, "loss": 2.764, "step": 20216 }, { "epoch": 1.2550127258054504, "grad_norm": 0.6091319846714305, "learning_rate": 7.217959755992538e-05, "loss": 2.816, "step": 20217 }, { "epoch": 1.2550748029052083, "grad_norm": 0.6605298491843599, "learning_rate": 7.21763607498486e-05, "loss": 2.883, "step": 20218 }, { "epoch": 1.255136880004966, "grad_norm": 0.5544643490512444, "learning_rate": 7.217312382407234e-05, "loss": 2.8701, "step": 20219 }, { "epoch": 1.255198957104724, "grad_norm": 0.6167220521446568, "learning_rate": 7.216988678261352e-05, "loss": 2.9032, "step": 20220 }, { "epoch": 1.2552610342044819, "grad_norm": 0.5921767875487548, "learning_rate": 7.216664962548902e-05, "loss": 2.856, "step": 20221 }, { "epoch": 1.2553231113042398, "grad_norm": 0.582087290069341, "learning_rate": 7.216341235271574e-05, "loss": 2.7794, "step": 20222 }, { "epoch": 1.2553851884039977, "grad_norm": 0.6735420836205396, "learning_rate": 7.216017496431055e-05, "loss": 2.8462, "step": 20223 }, { "epoch": 1.2554472655037556, "grad_norm": 0.6348011417722981, "learning_rate": 7.215693746029034e-05, "loss": 2.785, "step": 20224 }, { "epoch": 1.2555093426035135, "grad_norm": 0.592275017182191, "learning_rate": 7.215369984067201e-05, "loss": 2.8792, "step": 20225 }, { "epoch": 1.2555714197032715, "grad_norm": 0.6400103062306508, "learning_rate": 7.215046210547248e-05, "loss": 2.9178, "step": 20226 }, { "epoch": 1.2556334968030294, "grad_norm": 0.6521363756790058, "learning_rate": 7.214722425470857e-05, "loss": 2.9047, "step": 20227 }, { "epoch": 1.2556955739027873, "grad_norm": 0.6206117355066293, "learning_rate": 7.214398628839723e-05, "loss": 2.8381, "step": 20228 }, { "epoch": 1.2557576510025452, "grad_norm": 0.5939626062115282, "learning_rate": 7.214074820655534e-05, "loss": 2.8376, "step": 20229 }, { "epoch": 1.2558197281023031, "grad_norm": 0.5855574583353031, "learning_rate": 7.21375100091998e-05, "loss": 2.8262, "step": 20230 }, { "epoch": 1.255881805202061, "grad_norm": 0.6398701756648159, "learning_rate": 7.213427169634748e-05, "loss": 2.8272, "step": 20231 }, { "epoch": 1.255943882301819, "grad_norm": 0.5809665901653217, "learning_rate": 7.21310332680153e-05, "loss": 2.8858, "step": 20232 }, { "epoch": 1.2560059594015769, "grad_norm": 0.6044657120584184, "learning_rate": 7.212779472422013e-05, "loss": 2.845, "step": 20233 }, { "epoch": 1.2560680365013346, "grad_norm": 0.5595020032800805, "learning_rate": 7.212455606497891e-05, "loss": 2.8991, "step": 20234 }, { "epoch": 1.2561301136010925, "grad_norm": 0.5979039046845105, "learning_rate": 7.21213172903085e-05, "loss": 2.8455, "step": 20235 }, { "epoch": 1.2561921907008504, "grad_norm": 0.5357511513377683, "learning_rate": 7.21180784002258e-05, "loss": 2.8255, "step": 20236 }, { "epoch": 1.2562542678006083, "grad_norm": 0.5713477732168242, "learning_rate": 7.211483939474773e-05, "loss": 2.8713, "step": 20237 }, { "epoch": 1.2563163449003663, "grad_norm": 0.6827449215197597, "learning_rate": 7.211160027389116e-05, "loss": 2.9485, "step": 20238 }, { "epoch": 1.2563784220001242, "grad_norm": 0.49276375717591453, "learning_rate": 7.2108361037673e-05, "loss": 2.8174, "step": 20239 }, { "epoch": 1.256440499099882, "grad_norm": 0.6209596453146031, "learning_rate": 7.210512168611018e-05, "loss": 2.773, "step": 20240 }, { "epoch": 1.25650257619964, "grad_norm": 0.5491121310717494, "learning_rate": 7.210188221921955e-05, "loss": 2.8091, "step": 20241 }, { "epoch": 1.256564653299398, "grad_norm": 0.5908321158751233, "learning_rate": 7.209864263701805e-05, "loss": 2.8134, "step": 20242 }, { "epoch": 1.2566267303991556, "grad_norm": 0.5508617044761267, "learning_rate": 7.209540293952255e-05, "loss": 2.8401, "step": 20243 }, { "epoch": 1.2566888074989135, "grad_norm": 0.5369523807311232, "learning_rate": 7.209216312674998e-05, "loss": 2.8596, "step": 20244 }, { "epoch": 1.2567508845986715, "grad_norm": 0.6065653069503433, "learning_rate": 7.208892319871723e-05, "loss": 2.85, "step": 20245 }, { "epoch": 1.2568129616984294, "grad_norm": 0.5039485647488237, "learning_rate": 7.20856831554412e-05, "loss": 2.7987, "step": 20246 }, { "epoch": 1.2568750387981873, "grad_norm": 0.5426498684231915, "learning_rate": 7.20824429969388e-05, "loss": 2.7656, "step": 20247 }, { "epoch": 1.2569371158979452, "grad_norm": 0.579085402882409, "learning_rate": 7.207920272322696e-05, "loss": 2.8047, "step": 20248 }, { "epoch": 1.2569991929977031, "grad_norm": 0.5829509407048955, "learning_rate": 7.207596233432252e-05, "loss": 2.8171, "step": 20249 }, { "epoch": 1.257061270097461, "grad_norm": 0.5955174641438541, "learning_rate": 7.207272183024245e-05, "loss": 2.8081, "step": 20250 }, { "epoch": 1.257123347197219, "grad_norm": 0.5634059233658436, "learning_rate": 7.206948121100363e-05, "loss": 2.879, "step": 20251 }, { "epoch": 1.2571854242969769, "grad_norm": 0.5894801048338303, "learning_rate": 7.206624047662296e-05, "loss": 2.8375, "step": 20252 }, { "epoch": 1.2572475013967348, "grad_norm": 0.5734619551181189, "learning_rate": 7.206299962711737e-05, "loss": 2.8096, "step": 20253 }, { "epoch": 1.2573095784964927, "grad_norm": 0.5588963096018357, "learning_rate": 7.205975866250375e-05, "loss": 2.8285, "step": 20254 }, { "epoch": 1.2573716555962506, "grad_norm": 0.5968063394889904, "learning_rate": 7.2056517582799e-05, "loss": 2.8926, "step": 20255 }, { "epoch": 1.2574337326960086, "grad_norm": 0.5406290870026914, "learning_rate": 7.205327638802004e-05, "loss": 2.8025, "step": 20256 }, { "epoch": 1.2574958097957665, "grad_norm": 0.6079559921694698, "learning_rate": 7.205003507818381e-05, "loss": 2.7957, "step": 20257 }, { "epoch": 1.2575578868955242, "grad_norm": 0.6010187837772327, "learning_rate": 7.204679365330717e-05, "loss": 2.7444, "step": 20258 }, { "epoch": 1.257619963995282, "grad_norm": 0.5912202096494916, "learning_rate": 7.204355211340708e-05, "loss": 2.8394, "step": 20259 }, { "epoch": 1.25768204109504, "grad_norm": 0.5676185525252032, "learning_rate": 7.204031045850041e-05, "loss": 2.8144, "step": 20260 }, { "epoch": 1.257744118194798, "grad_norm": 0.5819030979063542, "learning_rate": 7.20370686886041e-05, "loss": 2.7885, "step": 20261 }, { "epoch": 1.2578061952945558, "grad_norm": 0.6002162633008116, "learning_rate": 7.203382680373505e-05, "loss": 2.9229, "step": 20262 }, { "epoch": 1.2578682723943138, "grad_norm": 0.5423188196165523, "learning_rate": 7.203058480391016e-05, "loss": 2.7866, "step": 20263 }, { "epoch": 1.2579303494940717, "grad_norm": 0.5765817768213565, "learning_rate": 7.202734268914638e-05, "loss": 2.7764, "step": 20264 }, { "epoch": 1.2579924265938296, "grad_norm": 0.5732931605068976, "learning_rate": 7.20241004594606e-05, "loss": 2.8119, "step": 20265 }, { "epoch": 1.2580545036935875, "grad_norm": 0.544942192327819, "learning_rate": 7.202085811486975e-05, "loss": 2.8661, "step": 20266 }, { "epoch": 1.2581165807933452, "grad_norm": 0.6054822858712513, "learning_rate": 7.201761565539072e-05, "loss": 2.8358, "step": 20267 }, { "epoch": 1.2581786578931031, "grad_norm": 0.5520844969347322, "learning_rate": 7.201437308104046e-05, "loss": 2.8517, "step": 20268 }, { "epoch": 1.258240734992861, "grad_norm": 0.5878973672169707, "learning_rate": 7.201113039183586e-05, "loss": 2.7819, "step": 20269 }, { "epoch": 1.258302812092619, "grad_norm": 0.6221727798843519, "learning_rate": 7.200788758779385e-05, "loss": 2.8101, "step": 20270 }, { "epoch": 1.2583648891923769, "grad_norm": 0.6544319355050203, "learning_rate": 7.200464466893135e-05, "loss": 2.8799, "step": 20271 }, { "epoch": 1.2584269662921348, "grad_norm": 0.5765632463982773, "learning_rate": 7.200140163526528e-05, "loss": 2.8134, "step": 20272 }, { "epoch": 1.2584890433918927, "grad_norm": 0.5242947348544023, "learning_rate": 7.199815848681257e-05, "loss": 2.8242, "step": 20273 }, { "epoch": 1.2585511204916506, "grad_norm": 0.6655186339806887, "learning_rate": 7.19949152235901e-05, "loss": 2.8038, "step": 20274 }, { "epoch": 1.2586131975914086, "grad_norm": 0.5330436104634395, "learning_rate": 7.199167184561484e-05, "loss": 2.7469, "step": 20275 }, { "epoch": 1.2586752746911665, "grad_norm": 0.6351860649365181, "learning_rate": 7.198842835290367e-05, "loss": 2.9029, "step": 20276 }, { "epoch": 1.2587373517909244, "grad_norm": 0.6490474268302223, "learning_rate": 7.198518474547354e-05, "loss": 2.7894, "step": 20277 }, { "epoch": 1.2587994288906823, "grad_norm": 0.5925888452836384, "learning_rate": 7.198194102334138e-05, "loss": 2.7887, "step": 20278 }, { "epoch": 1.2588615059904402, "grad_norm": 0.5827056254966246, "learning_rate": 7.197869718652407e-05, "loss": 2.7277, "step": 20279 }, { "epoch": 1.2589235830901981, "grad_norm": 0.528029017783535, "learning_rate": 7.197545323503859e-05, "loss": 2.8101, "step": 20280 }, { "epoch": 1.258985660189956, "grad_norm": 0.6008464217336632, "learning_rate": 7.197220916890181e-05, "loss": 2.8472, "step": 20281 }, { "epoch": 1.2590477372897138, "grad_norm": 0.4730368886408567, "learning_rate": 7.196896498813069e-05, "loss": 2.8223, "step": 20282 }, { "epoch": 1.2591098143894717, "grad_norm": 0.5671695958614048, "learning_rate": 7.196572069274214e-05, "loss": 2.8968, "step": 20283 }, { "epoch": 1.2591718914892296, "grad_norm": 0.518935442720452, "learning_rate": 7.19624762827531e-05, "loss": 2.84, "step": 20284 }, { "epoch": 1.2592339685889875, "grad_norm": 0.5685355578670154, "learning_rate": 7.195923175818048e-05, "loss": 2.9289, "step": 20285 }, { "epoch": 1.2592960456887454, "grad_norm": 0.5655368578187441, "learning_rate": 7.195598711904124e-05, "loss": 2.9093, "step": 20286 }, { "epoch": 1.2593581227885033, "grad_norm": 0.5281733733375624, "learning_rate": 7.195274236535226e-05, "loss": 2.8139, "step": 20287 }, { "epoch": 1.2594201998882613, "grad_norm": 0.5811948206561819, "learning_rate": 7.194949749713051e-05, "loss": 2.9186, "step": 20288 }, { "epoch": 1.2594822769880192, "grad_norm": 0.6176219087410243, "learning_rate": 7.19462525143929e-05, "loss": 2.8816, "step": 20289 }, { "epoch": 1.259544354087777, "grad_norm": 0.6043929355962082, "learning_rate": 7.194300741715637e-05, "loss": 2.8727, "step": 20290 }, { "epoch": 1.2596064311875348, "grad_norm": 0.5534473578998355, "learning_rate": 7.193976220543783e-05, "loss": 2.7264, "step": 20291 }, { "epoch": 1.2596685082872927, "grad_norm": 0.6514905043603109, "learning_rate": 7.193651687925422e-05, "loss": 2.9171, "step": 20292 }, { "epoch": 1.2597305853870506, "grad_norm": 0.5766736189702442, "learning_rate": 7.193327143862249e-05, "loss": 2.7833, "step": 20293 }, { "epoch": 1.2597926624868085, "grad_norm": 0.5813554514319554, "learning_rate": 7.193002588355955e-05, "loss": 2.7692, "step": 20294 }, { "epoch": 1.2598547395865665, "grad_norm": 0.6199956579594695, "learning_rate": 7.192678021408236e-05, "loss": 2.8319, "step": 20295 }, { "epoch": 1.2599168166863244, "grad_norm": 0.6000227357609872, "learning_rate": 7.192353443020782e-05, "loss": 2.863, "step": 20296 }, { "epoch": 1.2599788937860823, "grad_norm": 0.6230639751323879, "learning_rate": 7.192028853195288e-05, "loss": 2.8314, "step": 20297 }, { "epoch": 1.2600409708858402, "grad_norm": 0.6027005317764126, "learning_rate": 7.191704251933447e-05, "loss": 2.8555, "step": 20298 }, { "epoch": 1.2601030479855981, "grad_norm": 0.5996683197373314, "learning_rate": 7.191379639236953e-05, "loss": 2.7329, "step": 20299 }, { "epoch": 1.260165125085356, "grad_norm": 0.5844846897910139, "learning_rate": 7.191055015107499e-05, "loss": 2.8522, "step": 20300 }, { "epoch": 1.260227202185114, "grad_norm": 0.5970432154303279, "learning_rate": 7.19073037954678e-05, "loss": 2.7771, "step": 20301 }, { "epoch": 1.2602892792848719, "grad_norm": 0.5478875731466564, "learning_rate": 7.190405732556488e-05, "loss": 2.8565, "step": 20302 }, { "epoch": 1.2603513563846298, "grad_norm": 0.5379683405874325, "learning_rate": 7.190081074138318e-05, "loss": 2.8787, "step": 20303 }, { "epoch": 1.2604134334843877, "grad_norm": 0.5782044588075409, "learning_rate": 7.189756404293964e-05, "loss": 2.8575, "step": 20304 }, { "epoch": 1.2604755105841456, "grad_norm": 0.5678713299134961, "learning_rate": 7.189431723025118e-05, "loss": 2.7999, "step": 20305 }, { "epoch": 1.2605375876839033, "grad_norm": 0.5882455001850251, "learning_rate": 7.189107030333475e-05, "loss": 2.9746, "step": 20306 }, { "epoch": 1.2605996647836613, "grad_norm": 0.5679428042187935, "learning_rate": 7.18878232622073e-05, "loss": 2.8503, "step": 20307 }, { "epoch": 1.2606617418834192, "grad_norm": 0.5811868982005304, "learning_rate": 7.188457610688577e-05, "loss": 2.7564, "step": 20308 }, { "epoch": 1.260723818983177, "grad_norm": 0.5757920645910656, "learning_rate": 7.18813288373871e-05, "loss": 2.7874, "step": 20309 }, { "epoch": 1.260785896082935, "grad_norm": 0.5172114963093147, "learning_rate": 7.18780814537282e-05, "loss": 2.8077, "step": 20310 }, { "epoch": 1.260847973182693, "grad_norm": 0.6180473084798447, "learning_rate": 7.187483395592604e-05, "loss": 2.8205, "step": 20311 }, { "epoch": 1.2609100502824508, "grad_norm": 0.5738734030729395, "learning_rate": 7.187158634399758e-05, "loss": 2.8535, "step": 20312 }, { "epoch": 1.2609721273822088, "grad_norm": 0.5744150825187916, "learning_rate": 7.186833861795975e-05, "loss": 2.8495, "step": 20313 }, { "epoch": 1.2610342044819667, "grad_norm": 0.48670965954663276, "learning_rate": 7.186509077782947e-05, "loss": 2.9337, "step": 20314 }, { "epoch": 1.2610962815817244, "grad_norm": 0.627223613062363, "learning_rate": 7.18618428236237e-05, "loss": 2.8363, "step": 20315 }, { "epoch": 1.2611583586814823, "grad_norm": 0.5389425724803453, "learning_rate": 7.18585947553594e-05, "loss": 2.7334, "step": 20316 }, { "epoch": 1.2612204357812402, "grad_norm": 0.5498656108702382, "learning_rate": 7.185534657305348e-05, "loss": 2.8705, "step": 20317 }, { "epoch": 1.2612825128809981, "grad_norm": 0.5843947636735259, "learning_rate": 7.185209827672294e-05, "loss": 2.8046, "step": 20318 }, { "epoch": 1.261344589980756, "grad_norm": 0.5619525364554908, "learning_rate": 7.18488498663847e-05, "loss": 2.8258, "step": 20319 }, { "epoch": 1.261406667080514, "grad_norm": 0.5790188812362769, "learning_rate": 7.184560134205569e-05, "loss": 2.719, "step": 20320 }, { "epoch": 1.2614687441802719, "grad_norm": 0.5587430634377386, "learning_rate": 7.184235270375289e-05, "loss": 2.8642, "step": 20321 }, { "epoch": 1.2615308212800298, "grad_norm": 0.6320729349203481, "learning_rate": 7.183910395149322e-05, "loss": 2.8012, "step": 20322 }, { "epoch": 1.2615928983797877, "grad_norm": 0.55408036222028, "learning_rate": 7.183585508529366e-05, "loss": 2.8717, "step": 20323 }, { "epoch": 1.2616549754795456, "grad_norm": 0.6261023813047334, "learning_rate": 7.183260610517112e-05, "loss": 2.8539, "step": 20324 }, { "epoch": 1.2617170525793036, "grad_norm": 0.6155649901444381, "learning_rate": 7.182935701114259e-05, "loss": 2.8592, "step": 20325 }, { "epoch": 1.2617791296790615, "grad_norm": 0.6127551306482902, "learning_rate": 7.182610780322499e-05, "loss": 2.8853, "step": 20326 }, { "epoch": 1.2618412067788194, "grad_norm": 0.5598229175964147, "learning_rate": 7.18228584814353e-05, "loss": 2.9202, "step": 20327 }, { "epoch": 1.2619032838785773, "grad_norm": 0.7072375929045223, "learning_rate": 7.181960904579046e-05, "loss": 2.7851, "step": 20328 }, { "epoch": 1.2619653609783352, "grad_norm": 0.5511322526322607, "learning_rate": 7.181635949630743e-05, "loss": 2.9384, "step": 20329 }, { "epoch": 1.262027438078093, "grad_norm": 0.6724395154675848, "learning_rate": 7.181310983300315e-05, "loss": 2.9443, "step": 20330 }, { "epoch": 1.2620895151778508, "grad_norm": 0.5969028586232427, "learning_rate": 7.180986005589457e-05, "loss": 2.7782, "step": 20331 }, { "epoch": 1.2621515922776088, "grad_norm": 0.6654514596936119, "learning_rate": 7.180661016499868e-05, "loss": 2.8997, "step": 20332 }, { "epoch": 1.2622136693773667, "grad_norm": 0.6320397167793954, "learning_rate": 7.180336016033236e-05, "loss": 2.78, "step": 20333 }, { "epoch": 1.2622757464771246, "grad_norm": 0.5916886435264271, "learning_rate": 7.180011004191267e-05, "loss": 2.7808, "step": 20334 }, { "epoch": 1.2623378235768825, "grad_norm": 0.6712953373535537, "learning_rate": 7.179685980975649e-05, "loss": 2.8659, "step": 20335 }, { "epoch": 1.2623999006766404, "grad_norm": 0.6223060005724395, "learning_rate": 7.17936094638808e-05, "loss": 2.8765, "step": 20336 }, { "epoch": 1.2624619777763983, "grad_norm": 0.6286445215227862, "learning_rate": 7.179035900430257e-05, "loss": 2.7938, "step": 20337 }, { "epoch": 1.2625240548761563, "grad_norm": 0.6314284884343959, "learning_rate": 7.178710843103875e-05, "loss": 2.8303, "step": 20338 }, { "epoch": 1.262586131975914, "grad_norm": 0.6210040381556748, "learning_rate": 7.178385774410629e-05, "loss": 2.769, "step": 20339 }, { "epoch": 1.2626482090756719, "grad_norm": 0.5825222171999712, "learning_rate": 7.178060694352214e-05, "loss": 2.7993, "step": 20340 }, { "epoch": 1.2627102861754298, "grad_norm": 0.6031441641021072, "learning_rate": 7.17773560293033e-05, "loss": 2.7937, "step": 20341 }, { "epoch": 1.2627723632751877, "grad_norm": 0.5927143601668137, "learning_rate": 7.177410500146669e-05, "loss": 2.8553, "step": 20342 }, { "epoch": 1.2628344403749456, "grad_norm": 0.5960147480340052, "learning_rate": 7.177085386002931e-05, "loss": 2.8706, "step": 20343 }, { "epoch": 1.2628965174747036, "grad_norm": 0.6173430620989567, "learning_rate": 7.176760260500808e-05, "loss": 2.8849, "step": 20344 }, { "epoch": 1.2629585945744615, "grad_norm": 0.6388039641904039, "learning_rate": 7.176435123642e-05, "loss": 2.8249, "step": 20345 }, { "epoch": 1.2630206716742194, "grad_norm": 0.6309739236204385, "learning_rate": 7.176109975428199e-05, "loss": 2.8712, "step": 20346 }, { "epoch": 1.2630827487739773, "grad_norm": 0.5547568318194311, "learning_rate": 7.175784815861106e-05, "loss": 2.8262, "step": 20347 }, { "epoch": 1.2631448258737352, "grad_norm": 0.5607218186928536, "learning_rate": 7.175459644942414e-05, "loss": 2.8239, "step": 20348 }, { "epoch": 1.2632069029734931, "grad_norm": 0.5592719809007112, "learning_rate": 7.175134462673822e-05, "loss": 2.8366, "step": 20349 }, { "epoch": 1.263268980073251, "grad_norm": 0.570809552702066, "learning_rate": 7.174809269057026e-05, "loss": 2.8348, "step": 20350 }, { "epoch": 1.263331057173009, "grad_norm": 0.5365144521254553, "learning_rate": 7.17448406409372e-05, "loss": 2.7877, "step": 20351 }, { "epoch": 1.263393134272767, "grad_norm": 0.6256552837332545, "learning_rate": 7.174158847785605e-05, "loss": 2.82, "step": 20352 }, { "epoch": 1.2634552113725248, "grad_norm": 0.654091896858231, "learning_rate": 7.173833620134374e-05, "loss": 2.9167, "step": 20353 }, { "epoch": 1.2635172884722825, "grad_norm": 0.6256340625270429, "learning_rate": 7.173508381141727e-05, "loss": 2.8262, "step": 20354 }, { "epoch": 1.2635793655720404, "grad_norm": 0.5885514508899762, "learning_rate": 7.173183130809358e-05, "loss": 2.833, "step": 20355 }, { "epoch": 1.2636414426717983, "grad_norm": 0.5703799847952042, "learning_rate": 7.172857869138965e-05, "loss": 2.8233, "step": 20356 }, { "epoch": 1.2637035197715563, "grad_norm": 0.5480159030246113, "learning_rate": 7.172532596132245e-05, "loss": 2.8569, "step": 20357 }, { "epoch": 1.2637655968713142, "grad_norm": 0.5218619076577048, "learning_rate": 7.172207311790895e-05, "loss": 2.8603, "step": 20358 }, { "epoch": 1.263827673971072, "grad_norm": 0.5386790694875959, "learning_rate": 7.171882016116613e-05, "loss": 2.8283, "step": 20359 }, { "epoch": 1.26388975107083, "grad_norm": 0.489969873694321, "learning_rate": 7.171556709111094e-05, "loss": 2.9044, "step": 20360 }, { "epoch": 1.263951828170588, "grad_norm": 0.5225070255665061, "learning_rate": 7.171231390776037e-05, "loss": 2.8028, "step": 20361 }, { "epoch": 1.2640139052703458, "grad_norm": 0.5214482555667844, "learning_rate": 7.170906061113139e-05, "loss": 2.8744, "step": 20362 }, { "epoch": 1.2640759823701035, "grad_norm": 0.628055281114844, "learning_rate": 7.170580720124098e-05, "loss": 2.8511, "step": 20363 }, { "epoch": 1.2641380594698615, "grad_norm": 0.5710529589403422, "learning_rate": 7.170255367810609e-05, "loss": 2.7851, "step": 20364 }, { "epoch": 1.2642001365696194, "grad_norm": 0.5886020601281897, "learning_rate": 7.169930004174371e-05, "loss": 2.8182, "step": 20365 }, { "epoch": 1.2642622136693773, "grad_norm": 0.6122675006774932, "learning_rate": 7.169604629217082e-05, "loss": 2.8264, "step": 20366 }, { "epoch": 1.2643242907691352, "grad_norm": 0.5432759520541449, "learning_rate": 7.169279242940438e-05, "loss": 2.9907, "step": 20367 }, { "epoch": 1.2643863678688931, "grad_norm": 0.5497597494783097, "learning_rate": 7.168953845346138e-05, "loss": 2.7379, "step": 20368 }, { "epoch": 1.264448444968651, "grad_norm": 0.5313305232786604, "learning_rate": 7.168628436435879e-05, "loss": 2.7585, "step": 20369 }, { "epoch": 1.264510522068409, "grad_norm": 0.5458833695790553, "learning_rate": 7.16830301621136e-05, "loss": 2.7925, "step": 20370 }, { "epoch": 1.2645725991681669, "grad_norm": 0.6083055795800903, "learning_rate": 7.167977584674277e-05, "loss": 2.7486, "step": 20371 }, { "epoch": 1.2646346762679248, "grad_norm": 0.5676570630034913, "learning_rate": 7.167652141826329e-05, "loss": 2.8697, "step": 20372 }, { "epoch": 1.2646967533676827, "grad_norm": 0.6680310315102652, "learning_rate": 7.167326687669214e-05, "loss": 2.8826, "step": 20373 }, { "epoch": 1.2647588304674406, "grad_norm": 0.5622591986745442, "learning_rate": 7.167001222204629e-05, "loss": 2.8444, "step": 20374 }, { "epoch": 1.2648209075671986, "grad_norm": 0.6369444792174876, "learning_rate": 7.166675745434273e-05, "loss": 2.8925, "step": 20375 }, { "epoch": 1.2648829846669565, "grad_norm": 0.5727071263298378, "learning_rate": 7.166350257359843e-05, "loss": 2.8514, "step": 20376 }, { "epoch": 1.2649450617667144, "grad_norm": 0.6051272752179683, "learning_rate": 7.166024757983039e-05, "loss": 2.75, "step": 20377 }, { "epoch": 1.265007138866472, "grad_norm": 0.6066204315774422, "learning_rate": 7.165699247305555e-05, "loss": 2.9629, "step": 20378 }, { "epoch": 1.26506921596623, "grad_norm": 0.5435224221262248, "learning_rate": 7.165373725329096e-05, "loss": 2.8288, "step": 20379 }, { "epoch": 1.265131293065988, "grad_norm": 0.5903171887673498, "learning_rate": 7.165048192055353e-05, "loss": 2.8251, "step": 20380 }, { "epoch": 1.2651933701657458, "grad_norm": 0.5517039314900898, "learning_rate": 7.164722647486032e-05, "loss": 2.9575, "step": 20381 }, { "epoch": 1.2652554472655038, "grad_norm": 0.6244345252651834, "learning_rate": 7.164397091622824e-05, "loss": 2.8293, "step": 20382 }, { "epoch": 1.2653175243652617, "grad_norm": 0.5097121579715296, "learning_rate": 7.164071524467433e-05, "loss": 2.7665, "step": 20383 }, { "epoch": 1.2653796014650196, "grad_norm": 0.7265489638513526, "learning_rate": 7.163745946021554e-05, "loss": 2.8962, "step": 20384 }, { "epoch": 1.2654416785647775, "grad_norm": 0.5697646187517493, "learning_rate": 7.163420356286887e-05, "loss": 2.8025, "step": 20385 }, { "epoch": 1.2655037556645354, "grad_norm": 0.6301429861778791, "learning_rate": 7.163094755265132e-05, "loss": 2.8065, "step": 20386 }, { "epoch": 1.2655658327642931, "grad_norm": 0.6172356647095285, "learning_rate": 7.162769142957987e-05, "loss": 2.8569, "step": 20387 }, { "epoch": 1.265627909864051, "grad_norm": 0.5852455376622374, "learning_rate": 7.16244351936715e-05, "loss": 2.7905, "step": 20388 }, { "epoch": 1.265689986963809, "grad_norm": 0.6927192204320173, "learning_rate": 7.162117884494319e-05, "loss": 2.8861, "step": 20389 }, { "epoch": 1.2657520640635669, "grad_norm": 0.556655107276312, "learning_rate": 7.161792238341196e-05, "loss": 2.7487, "step": 20390 }, { "epoch": 1.2658141411633248, "grad_norm": 0.6316443127658368, "learning_rate": 7.161466580909477e-05, "loss": 2.8687, "step": 20391 }, { "epoch": 1.2658762182630827, "grad_norm": 0.625709179027208, "learning_rate": 7.161140912200863e-05, "loss": 2.8699, "step": 20392 }, { "epoch": 1.2659382953628406, "grad_norm": 0.5469722116445033, "learning_rate": 7.160815232217053e-05, "loss": 2.8345, "step": 20393 }, { "epoch": 1.2660003724625986, "grad_norm": 0.6124532905596102, "learning_rate": 7.160489540959742e-05, "loss": 2.9025, "step": 20394 }, { "epoch": 1.2660624495623565, "grad_norm": 0.5656916615947671, "learning_rate": 7.160163838430635e-05, "loss": 2.8811, "step": 20395 }, { "epoch": 1.2661245266621144, "grad_norm": 0.5307942005876257, "learning_rate": 7.159838124631428e-05, "loss": 2.827, "step": 20396 }, { "epoch": 1.2661866037618723, "grad_norm": 0.6021991432461936, "learning_rate": 7.159512399563821e-05, "loss": 2.8735, "step": 20397 }, { "epoch": 1.2662486808616302, "grad_norm": 0.5617108372328602, "learning_rate": 7.159186663229513e-05, "loss": 2.9047, "step": 20398 }, { "epoch": 1.2663107579613881, "grad_norm": 0.5789152878275882, "learning_rate": 7.158860915630205e-05, "loss": 2.8345, "step": 20399 }, { "epoch": 1.266372835061146, "grad_norm": 0.6184280579064462, "learning_rate": 7.158535156767596e-05, "loss": 2.7546, "step": 20400 }, { "epoch": 1.266434912160904, "grad_norm": 0.5304697140669454, "learning_rate": 7.158209386643385e-05, "loss": 2.846, "step": 20401 }, { "epoch": 1.2664969892606617, "grad_norm": 0.5412892224766578, "learning_rate": 7.157883605259273e-05, "loss": 2.9057, "step": 20402 }, { "epoch": 1.2665590663604196, "grad_norm": 0.4823895648221446, "learning_rate": 7.157557812616955e-05, "loss": 2.7714, "step": 20403 }, { "epoch": 1.2666211434601775, "grad_norm": 0.567885709461324, "learning_rate": 7.157232008718137e-05, "loss": 2.7582, "step": 20404 }, { "epoch": 1.2666832205599354, "grad_norm": 0.5451418251584854, "learning_rate": 7.156906193564515e-05, "loss": 2.8305, "step": 20405 }, { "epoch": 1.2667452976596933, "grad_norm": 0.5864445590428112, "learning_rate": 7.156580367157789e-05, "loss": 2.8381, "step": 20406 }, { "epoch": 1.2668073747594513, "grad_norm": 0.579647430927023, "learning_rate": 7.156254529499659e-05, "loss": 2.8749, "step": 20407 }, { "epoch": 1.2668694518592092, "grad_norm": 0.6140397841738834, "learning_rate": 7.155928680591829e-05, "loss": 2.8532, "step": 20408 }, { "epoch": 1.266931528958967, "grad_norm": 0.5924007243269845, "learning_rate": 7.155602820435993e-05, "loss": 2.8349, "step": 20409 }, { "epoch": 1.266993606058725, "grad_norm": 0.7496980615645483, "learning_rate": 7.155276949033855e-05, "loss": 2.9274, "step": 20410 }, { "epoch": 1.2670556831584827, "grad_norm": 0.6078203372401016, "learning_rate": 7.154951066387114e-05, "loss": 2.8187, "step": 20411 }, { "epoch": 1.2671177602582406, "grad_norm": 0.5984740303383861, "learning_rate": 7.15462517249747e-05, "loss": 2.8668, "step": 20412 }, { "epoch": 1.2671798373579986, "grad_norm": 0.6147519354508607, "learning_rate": 7.154299267366622e-05, "loss": 2.8138, "step": 20413 }, { "epoch": 1.2672419144577565, "grad_norm": 0.5721514851659614, "learning_rate": 7.153973350996272e-05, "loss": 2.8369, "step": 20414 }, { "epoch": 1.2673039915575144, "grad_norm": 0.5748723261522791, "learning_rate": 7.15364742338812e-05, "loss": 2.8681, "step": 20415 }, { "epoch": 1.2673660686572723, "grad_norm": 0.5353607566625728, "learning_rate": 7.153321484543866e-05, "loss": 2.8342, "step": 20416 }, { "epoch": 1.2674281457570302, "grad_norm": 0.5852261866907075, "learning_rate": 7.152995534465214e-05, "loss": 2.8808, "step": 20417 }, { "epoch": 1.2674902228567881, "grad_norm": 0.5566962231091476, "learning_rate": 7.15266957315386e-05, "loss": 2.9111, "step": 20418 }, { "epoch": 1.267552299956546, "grad_norm": 0.5509076074982405, "learning_rate": 7.152343600611507e-05, "loss": 2.8091, "step": 20419 }, { "epoch": 1.267614377056304, "grad_norm": 0.6010789289994921, "learning_rate": 7.152017616839855e-05, "loss": 2.8977, "step": 20420 }, { "epoch": 1.267676454156062, "grad_norm": 0.5184687311516106, "learning_rate": 7.151691621840603e-05, "loss": 2.7892, "step": 20421 }, { "epoch": 1.2677385312558198, "grad_norm": 0.5506792176883367, "learning_rate": 7.151365615615455e-05, "loss": 2.83, "step": 20422 }, { "epoch": 1.2678006083555777, "grad_norm": 0.5179791318563575, "learning_rate": 7.151039598166109e-05, "loss": 2.7917, "step": 20423 }, { "epoch": 1.2678626854553356, "grad_norm": 0.5844331885757982, "learning_rate": 7.15071356949427e-05, "loss": 2.8931, "step": 20424 }, { "epoch": 1.2679247625550933, "grad_norm": 0.5621038207309251, "learning_rate": 7.150387529601633e-05, "loss": 2.7924, "step": 20425 }, { "epoch": 1.2679868396548513, "grad_norm": 0.5774387720190844, "learning_rate": 7.150061478489904e-05, "loss": 2.7873, "step": 20426 }, { "epoch": 1.2680489167546092, "grad_norm": 0.567310216990079, "learning_rate": 7.14973541616078e-05, "loss": 2.9438, "step": 20427 }, { "epoch": 1.268110993854367, "grad_norm": 0.5616437169498114, "learning_rate": 7.149409342615968e-05, "loss": 2.7323, "step": 20428 }, { "epoch": 1.268173070954125, "grad_norm": 0.5927834423546905, "learning_rate": 7.149083257857163e-05, "loss": 2.8087, "step": 20429 }, { "epoch": 1.268235148053883, "grad_norm": 0.5373012408108195, "learning_rate": 7.14875716188607e-05, "loss": 2.8052, "step": 20430 }, { "epoch": 1.2682972251536409, "grad_norm": 0.617910926901083, "learning_rate": 7.14843105470439e-05, "loss": 2.848, "step": 20431 }, { "epoch": 1.2683593022533988, "grad_norm": 0.5183533530902418, "learning_rate": 7.148104936313821e-05, "loss": 2.8322, "step": 20432 }, { "epoch": 1.2684213793531567, "grad_norm": 0.5663143478024923, "learning_rate": 7.14777880671607e-05, "loss": 2.8805, "step": 20433 }, { "epoch": 1.2684834564529144, "grad_norm": 0.562480343369249, "learning_rate": 7.147452665912834e-05, "loss": 2.7821, "step": 20434 }, { "epoch": 1.2685455335526723, "grad_norm": 0.5779616924266469, "learning_rate": 7.147126513905818e-05, "loss": 2.8618, "step": 20435 }, { "epoch": 1.2686076106524302, "grad_norm": 0.6583162657621786, "learning_rate": 7.14680035069672e-05, "loss": 2.7247, "step": 20436 }, { "epoch": 1.2686696877521881, "grad_norm": 0.6263181614282548, "learning_rate": 7.146474176287243e-05, "loss": 2.8203, "step": 20437 }, { "epoch": 1.268731764851946, "grad_norm": 0.6095191833897603, "learning_rate": 7.14614799067909e-05, "loss": 2.7514, "step": 20438 }, { "epoch": 1.268793841951704, "grad_norm": 0.6204038182839607, "learning_rate": 7.145821793873961e-05, "loss": 2.8466, "step": 20439 }, { "epoch": 1.268855919051462, "grad_norm": 0.6010789289994921, "learning_rate": 7.145495585873561e-05, "loss": 2.8317, "step": 20440 }, { "epoch": 1.2689179961512198, "grad_norm": 0.5569344798988545, "learning_rate": 7.145169366679587e-05, "loss": 2.8651, "step": 20441 }, { "epoch": 1.2689800732509777, "grad_norm": 0.6321588837675606, "learning_rate": 7.144843136293746e-05, "loss": 2.896, "step": 20442 }, { "epoch": 1.2690421503507356, "grad_norm": 0.5181367270933687, "learning_rate": 7.144516894717737e-05, "loss": 2.8, "step": 20443 }, { "epoch": 1.2691042274504936, "grad_norm": 0.6318193106254514, "learning_rate": 7.144190641953262e-05, "loss": 2.7529, "step": 20444 }, { "epoch": 1.2691663045502515, "grad_norm": 0.47798466711112386, "learning_rate": 7.143864378002025e-05, "loss": 2.8231, "step": 20445 }, { "epoch": 1.2692283816500094, "grad_norm": 0.5960875724582371, "learning_rate": 7.143538102865726e-05, "loss": 2.8469, "step": 20446 }, { "epoch": 1.2692904587497673, "grad_norm": 0.5072593610791878, "learning_rate": 7.143211816546068e-05, "loss": 2.7545, "step": 20447 }, { "epoch": 1.2693525358495252, "grad_norm": 0.5194776120139323, "learning_rate": 7.142885519044755e-05, "loss": 2.8209, "step": 20448 }, { "epoch": 1.269414612949283, "grad_norm": 0.5298133385541264, "learning_rate": 7.142559210363487e-05, "loss": 2.9022, "step": 20449 }, { "epoch": 1.2694766900490408, "grad_norm": 0.5266822622148978, "learning_rate": 7.142232890503967e-05, "loss": 2.8759, "step": 20450 }, { "epoch": 1.2695387671487988, "grad_norm": 0.49158471146423954, "learning_rate": 7.1419065594679e-05, "loss": 2.8746, "step": 20451 }, { "epoch": 1.2696008442485567, "grad_norm": 0.5207714965987414, "learning_rate": 7.141580217256984e-05, "loss": 2.7896, "step": 20452 }, { "epoch": 1.2696629213483146, "grad_norm": 0.4568415517082846, "learning_rate": 7.141253863872927e-05, "loss": 2.7755, "step": 20453 }, { "epoch": 1.2697249984480725, "grad_norm": 0.4933336325217104, "learning_rate": 7.140927499317428e-05, "loss": 2.828, "step": 20454 }, { "epoch": 1.2697870755478304, "grad_norm": 0.5116184223627379, "learning_rate": 7.140601123592189e-05, "loss": 2.7979, "step": 20455 }, { "epoch": 1.2698491526475884, "grad_norm": 0.47899308032013144, "learning_rate": 7.140274736698916e-05, "loss": 2.9132, "step": 20456 }, { "epoch": 1.2699112297473463, "grad_norm": 0.5258983441995746, "learning_rate": 7.139948338639309e-05, "loss": 2.8106, "step": 20457 }, { "epoch": 1.269973306847104, "grad_norm": 0.47236215884280575, "learning_rate": 7.139621929415073e-05, "loss": 2.893, "step": 20458 }, { "epoch": 1.2700353839468619, "grad_norm": 0.5171155193751652, "learning_rate": 7.139295509027908e-05, "loss": 2.8428, "step": 20459 }, { "epoch": 1.2700974610466198, "grad_norm": 0.47434866381027063, "learning_rate": 7.13896907747952e-05, "loss": 2.8109, "step": 20460 }, { "epoch": 1.2701595381463777, "grad_norm": 0.5004811356197215, "learning_rate": 7.138642634771611e-05, "loss": 2.8154, "step": 20461 }, { "epoch": 1.2702216152461356, "grad_norm": 0.4815810285020101, "learning_rate": 7.138316180905885e-05, "loss": 2.8814, "step": 20462 }, { "epoch": 1.2702836923458936, "grad_norm": 0.5447288636233861, "learning_rate": 7.137989715884043e-05, "loss": 2.8229, "step": 20463 }, { "epoch": 1.2703457694456515, "grad_norm": 0.5023972442763219, "learning_rate": 7.13766323970779e-05, "loss": 2.8265, "step": 20464 }, { "epoch": 1.2704078465454094, "grad_norm": 0.5708081691188784, "learning_rate": 7.137336752378831e-05, "loss": 2.7602, "step": 20465 }, { "epoch": 1.2704699236451673, "grad_norm": 0.5122207334302108, "learning_rate": 7.137010253898864e-05, "loss": 2.8508, "step": 20466 }, { "epoch": 1.2705320007449252, "grad_norm": 0.5542704397810831, "learning_rate": 7.1366837442696e-05, "loss": 2.9005, "step": 20467 }, { "epoch": 1.2705940778446831, "grad_norm": 0.48743005947768897, "learning_rate": 7.136357223492734e-05, "loss": 2.7658, "step": 20468 }, { "epoch": 1.270656154944441, "grad_norm": 0.5309232941377857, "learning_rate": 7.136030691569976e-05, "loss": 2.8254, "step": 20469 }, { "epoch": 1.270718232044199, "grad_norm": 0.5084304863950685, "learning_rate": 7.135704148503026e-05, "loss": 2.7321, "step": 20470 }, { "epoch": 1.270780309143957, "grad_norm": 0.5451465266665242, "learning_rate": 7.135377594293591e-05, "loss": 2.8135, "step": 20471 }, { "epoch": 1.2708423862437148, "grad_norm": 0.5028873759128161, "learning_rate": 7.13505102894337e-05, "loss": 2.8044, "step": 20472 }, { "epoch": 1.2709044633434725, "grad_norm": 0.49189337767571206, "learning_rate": 7.134724452454072e-05, "loss": 2.752, "step": 20473 }, { "epoch": 1.2709665404432304, "grad_norm": 0.5039083495576869, "learning_rate": 7.134397864827398e-05, "loss": 2.8448, "step": 20474 }, { "epoch": 1.2710286175429883, "grad_norm": 0.5624972714251991, "learning_rate": 7.134071266065051e-05, "loss": 2.8805, "step": 20475 }, { "epoch": 1.2710906946427463, "grad_norm": 0.5123854799950786, "learning_rate": 7.133744656168737e-05, "loss": 2.8056, "step": 20476 }, { "epoch": 1.2711527717425042, "grad_norm": 0.5130844819012321, "learning_rate": 7.133418035140158e-05, "loss": 2.9205, "step": 20477 }, { "epoch": 1.271214848842262, "grad_norm": 0.48906958456057315, "learning_rate": 7.13309140298102e-05, "loss": 2.7598, "step": 20478 }, { "epoch": 1.27127692594202, "grad_norm": 0.5789584518556953, "learning_rate": 7.132764759693027e-05, "loss": 2.861, "step": 20479 }, { "epoch": 1.271339003041778, "grad_norm": 0.5259696294226328, "learning_rate": 7.132438105277882e-05, "loss": 2.8807, "step": 20480 }, { "epoch": 1.2714010801415359, "grad_norm": 0.5448550111500102, "learning_rate": 7.132111439737289e-05, "loss": 2.768, "step": 20481 }, { "epoch": 1.2714631572412936, "grad_norm": 0.5345601569741757, "learning_rate": 7.131784763072953e-05, "loss": 2.8335, "step": 20482 }, { "epoch": 1.2715252343410515, "grad_norm": 0.5150185689078456, "learning_rate": 7.13145807528658e-05, "loss": 2.8356, "step": 20483 }, { "epoch": 1.2715873114408094, "grad_norm": 0.4804355795970661, "learning_rate": 7.131131376379871e-05, "loss": 2.8107, "step": 20484 }, { "epoch": 1.2716493885405673, "grad_norm": 0.5211694745314573, "learning_rate": 7.130804666354533e-05, "loss": 2.818, "step": 20485 }, { "epoch": 1.2717114656403252, "grad_norm": 0.5074551131688199, "learning_rate": 7.13047794521227e-05, "loss": 2.8456, "step": 20486 }, { "epoch": 1.2717735427400831, "grad_norm": 0.5170924084881656, "learning_rate": 7.130151212954786e-05, "loss": 2.8259, "step": 20487 }, { "epoch": 1.271835619839841, "grad_norm": 0.5966612817700658, "learning_rate": 7.129824469583786e-05, "loss": 2.7302, "step": 20488 }, { "epoch": 1.271897696939599, "grad_norm": 0.5446639732493073, "learning_rate": 7.129497715100975e-05, "loss": 2.8206, "step": 20489 }, { "epoch": 1.271959774039357, "grad_norm": 0.5795016265810108, "learning_rate": 7.129170949508057e-05, "loss": 2.8785, "step": 20490 }, { "epoch": 1.2720218511391148, "grad_norm": 0.6134421265126921, "learning_rate": 7.128844172806739e-05, "loss": 2.7622, "step": 20491 }, { "epoch": 1.2720839282388727, "grad_norm": 0.49381503208053423, "learning_rate": 7.128517384998723e-05, "loss": 2.7591, "step": 20492 }, { "epoch": 1.2721460053386306, "grad_norm": 0.5782848085728399, "learning_rate": 7.128190586085715e-05, "loss": 2.8092, "step": 20493 }, { "epoch": 1.2722080824383886, "grad_norm": 0.5608515692216975, "learning_rate": 7.12786377606942e-05, "loss": 2.8715, "step": 20494 }, { "epoch": 1.2722701595381465, "grad_norm": 0.5899249898782567, "learning_rate": 7.127536954951543e-05, "loss": 2.8179, "step": 20495 }, { "epoch": 1.2723322366379044, "grad_norm": 0.6023469491915715, "learning_rate": 7.127210122733789e-05, "loss": 2.8103, "step": 20496 }, { "epoch": 1.272394313737662, "grad_norm": 0.583453903907519, "learning_rate": 7.126883279417863e-05, "loss": 2.8594, "step": 20497 }, { "epoch": 1.27245639083742, "grad_norm": 0.5651377449665735, "learning_rate": 7.126556425005471e-05, "loss": 2.8616, "step": 20498 }, { "epoch": 1.272518467937178, "grad_norm": 0.5994469487878699, "learning_rate": 7.126229559498318e-05, "loss": 2.8484, "step": 20499 }, { "epoch": 1.2725805450369359, "grad_norm": 0.5990471953664273, "learning_rate": 7.125902682898108e-05, "loss": 2.8104, "step": 20500 }, { "epoch": 1.2726426221366938, "grad_norm": 0.5716092508425029, "learning_rate": 7.125575795206551e-05, "loss": 2.7309, "step": 20501 }, { "epoch": 1.2727046992364517, "grad_norm": 0.5916882153961347, "learning_rate": 7.125248896425347e-05, "loss": 2.7743, "step": 20502 }, { "epoch": 1.2727667763362096, "grad_norm": 0.5677670935455831, "learning_rate": 7.124921986556204e-05, "loss": 2.8878, "step": 20503 }, { "epoch": 1.2728288534359675, "grad_norm": 0.5536649170173652, "learning_rate": 7.124595065600826e-05, "loss": 2.7766, "step": 20504 }, { "epoch": 1.2728909305357254, "grad_norm": 0.588965589611079, "learning_rate": 7.124268133560922e-05, "loss": 2.8316, "step": 20505 }, { "epoch": 1.2729530076354831, "grad_norm": 0.5263255143521977, "learning_rate": 7.123941190438193e-05, "loss": 2.8896, "step": 20506 }, { "epoch": 1.273015084735241, "grad_norm": 0.6015343597567953, "learning_rate": 7.123614236234349e-05, "loss": 2.8935, "step": 20507 }, { "epoch": 1.273077161834999, "grad_norm": 0.5101740286330277, "learning_rate": 7.123287270951094e-05, "loss": 2.8549, "step": 20508 }, { "epoch": 1.273139238934757, "grad_norm": 0.5158422474817598, "learning_rate": 7.122960294590133e-05, "loss": 2.7863, "step": 20509 }, { "epoch": 1.2732013160345148, "grad_norm": 0.527237832064196, "learning_rate": 7.122633307153173e-05, "loss": 2.7533, "step": 20510 }, { "epoch": 1.2732633931342727, "grad_norm": 0.5378150881361986, "learning_rate": 7.122306308641919e-05, "loss": 2.839, "step": 20511 }, { "epoch": 1.2733254702340306, "grad_norm": 0.5497468745189187, "learning_rate": 7.121979299058079e-05, "loss": 2.915, "step": 20512 }, { "epoch": 1.2733875473337886, "grad_norm": 0.5363888711000283, "learning_rate": 7.121652278403356e-05, "loss": 2.8303, "step": 20513 }, { "epoch": 1.2734496244335465, "grad_norm": 0.5777692215568186, "learning_rate": 7.12132524667946e-05, "loss": 2.8563, "step": 20514 }, { "epoch": 1.2735117015333044, "grad_norm": 0.5662470363598334, "learning_rate": 7.120998203888093e-05, "loss": 2.8143, "step": 20515 }, { "epoch": 1.2735737786330623, "grad_norm": 0.5874805984945879, "learning_rate": 7.120671150030966e-05, "loss": 2.8143, "step": 20516 }, { "epoch": 1.2736358557328202, "grad_norm": 0.6189334895941926, "learning_rate": 7.120344085109783e-05, "loss": 2.9059, "step": 20517 }, { "epoch": 1.2736979328325782, "grad_norm": 0.5605704272702058, "learning_rate": 7.120017009126248e-05, "loss": 2.8443, "step": 20518 }, { "epoch": 1.273760009932336, "grad_norm": 0.5838643336594803, "learning_rate": 7.119689922082072e-05, "loss": 2.7178, "step": 20519 }, { "epoch": 1.273822087032094, "grad_norm": 0.48972607283587904, "learning_rate": 7.119362823978957e-05, "loss": 2.7628, "step": 20520 }, { "epoch": 1.2738841641318517, "grad_norm": 0.6180021727887941, "learning_rate": 7.119035714818612e-05, "loss": 2.8338, "step": 20521 }, { "epoch": 1.2739462412316096, "grad_norm": 0.5735856283262866, "learning_rate": 7.118708594602744e-05, "loss": 2.8545, "step": 20522 }, { "epoch": 1.2740083183313675, "grad_norm": 0.5483656050897047, "learning_rate": 7.118381463333059e-05, "loss": 2.8175, "step": 20523 }, { "epoch": 1.2740703954311254, "grad_norm": 0.5955222183441858, "learning_rate": 7.118054321011263e-05, "loss": 2.8042, "step": 20524 }, { "epoch": 1.2741324725308834, "grad_norm": 0.5410847533643263, "learning_rate": 7.117727167639064e-05, "loss": 2.8253, "step": 20525 }, { "epoch": 1.2741945496306413, "grad_norm": 0.5301951145055676, "learning_rate": 7.117400003218168e-05, "loss": 2.7422, "step": 20526 }, { "epoch": 1.2742566267303992, "grad_norm": 0.5974538373073199, "learning_rate": 7.117072827750282e-05, "loss": 2.8866, "step": 20527 }, { "epoch": 1.274318703830157, "grad_norm": 0.527423626537011, "learning_rate": 7.116745641237113e-05, "loss": 2.9265, "step": 20528 }, { "epoch": 1.274380780929915, "grad_norm": 0.5668263423103089, "learning_rate": 7.116418443680368e-05, "loss": 2.8917, "step": 20529 }, { "epoch": 1.2744428580296727, "grad_norm": 0.5400906308819118, "learning_rate": 7.116091235081754e-05, "loss": 2.8702, "step": 20530 }, { "epoch": 1.2745049351294306, "grad_norm": 0.6325366513937712, "learning_rate": 7.11576401544298e-05, "loss": 2.8035, "step": 20531 }, { "epoch": 1.2745670122291886, "grad_norm": 0.5397945077964406, "learning_rate": 7.11543678476575e-05, "loss": 2.8828, "step": 20532 }, { "epoch": 1.2746290893289465, "grad_norm": 0.6825408892906455, "learning_rate": 7.115109543051772e-05, "loss": 2.9336, "step": 20533 }, { "epoch": 1.2746911664287044, "grad_norm": 0.5440829529223653, "learning_rate": 7.114782290302755e-05, "loss": 2.8089, "step": 20534 }, { "epoch": 1.2747532435284623, "grad_norm": 0.6206992954066392, "learning_rate": 7.114455026520408e-05, "loss": 2.8671, "step": 20535 }, { "epoch": 1.2748153206282202, "grad_norm": 0.5953444855274824, "learning_rate": 7.114127751706432e-05, "loss": 2.8848, "step": 20536 }, { "epoch": 1.2748773977279781, "grad_norm": 0.585637460818284, "learning_rate": 7.113800465862541e-05, "loss": 2.8356, "step": 20537 }, { "epoch": 1.274939474827736, "grad_norm": 0.5836066616116093, "learning_rate": 7.113473168990437e-05, "loss": 2.7132, "step": 20538 }, { "epoch": 1.275001551927494, "grad_norm": 0.5702717910242185, "learning_rate": 7.113145861091833e-05, "loss": 2.9455, "step": 20539 }, { "epoch": 1.275063629027252, "grad_norm": 0.6444702813182515, "learning_rate": 7.112818542168433e-05, "loss": 2.8951, "step": 20540 }, { "epoch": 1.2751257061270098, "grad_norm": 0.5505256059548601, "learning_rate": 7.112491212221946e-05, "loss": 2.934, "step": 20541 }, { "epoch": 1.2751877832267677, "grad_norm": 0.6314944681467464, "learning_rate": 7.11216387125408e-05, "loss": 2.7927, "step": 20542 }, { "epoch": 1.2752498603265257, "grad_norm": 0.562305522784255, "learning_rate": 7.111836519266542e-05, "loss": 2.8713, "step": 20543 }, { "epoch": 1.2753119374262836, "grad_norm": 0.5835624369808873, "learning_rate": 7.111509156261038e-05, "loss": 2.8283, "step": 20544 }, { "epoch": 1.2753740145260413, "grad_norm": 0.4860265785090403, "learning_rate": 7.111181782239282e-05, "loss": 2.8041, "step": 20545 }, { "epoch": 1.2754360916257992, "grad_norm": 0.6469879724381037, "learning_rate": 7.110854397202974e-05, "loss": 2.7523, "step": 20546 }, { "epoch": 1.275498168725557, "grad_norm": 0.5751762969884452, "learning_rate": 7.110527001153828e-05, "loss": 2.9219, "step": 20547 }, { "epoch": 1.275560245825315, "grad_norm": 0.5946027003499934, "learning_rate": 7.110199594093551e-05, "loss": 2.8676, "step": 20548 }, { "epoch": 1.275622322925073, "grad_norm": 0.6323366141853355, "learning_rate": 7.10987217602385e-05, "loss": 2.8167, "step": 20549 }, { "epoch": 1.2756844000248309, "grad_norm": 0.5512732344334902, "learning_rate": 7.109544746946435e-05, "loss": 2.8649, "step": 20550 }, { "epoch": 1.2757464771245888, "grad_norm": 0.5596670496541513, "learning_rate": 7.109217306863011e-05, "loss": 2.8019, "step": 20551 }, { "epoch": 1.2758085542243467, "grad_norm": 0.5967870387827154, "learning_rate": 7.10888985577529e-05, "loss": 2.8322, "step": 20552 }, { "epoch": 1.2758706313241046, "grad_norm": 0.5669127731608226, "learning_rate": 7.108562393684978e-05, "loss": 2.8444, "step": 20553 }, { "epoch": 1.2759327084238623, "grad_norm": 0.580701525134394, "learning_rate": 7.108234920593783e-05, "loss": 2.8206, "step": 20554 }, { "epoch": 1.2759947855236202, "grad_norm": 0.5393793032088853, "learning_rate": 7.107907436503417e-05, "loss": 2.8854, "step": 20555 }, { "epoch": 1.2760568626233781, "grad_norm": 0.6266905094955952, "learning_rate": 7.107579941415584e-05, "loss": 2.8545, "step": 20556 }, { "epoch": 1.276118939723136, "grad_norm": 0.6034561424200674, "learning_rate": 7.107252435331995e-05, "loss": 2.9165, "step": 20557 }, { "epoch": 1.276181016822894, "grad_norm": 0.5452460141899861, "learning_rate": 7.106924918254358e-05, "loss": 2.7784, "step": 20558 }, { "epoch": 1.276243093922652, "grad_norm": 0.6614300554093201, "learning_rate": 7.106597390184384e-05, "loss": 2.8442, "step": 20559 }, { "epoch": 1.2763051710224098, "grad_norm": 0.5768943524240109, "learning_rate": 7.106269851123777e-05, "loss": 2.7671, "step": 20560 }, { "epoch": 1.2763672481221677, "grad_norm": 0.5314155208560738, "learning_rate": 7.105942301074253e-05, "loss": 2.8075, "step": 20561 }, { "epoch": 1.2764293252219256, "grad_norm": 0.5792671468647507, "learning_rate": 7.105614740037514e-05, "loss": 2.8264, "step": 20562 }, { "epoch": 1.2764914023216836, "grad_norm": 0.5245114028352998, "learning_rate": 7.105287168015272e-05, "loss": 2.881, "step": 20563 }, { "epoch": 1.2765534794214415, "grad_norm": 0.6079540803659583, "learning_rate": 7.104959585009236e-05, "loss": 2.8519, "step": 20564 }, { "epoch": 1.2766155565211994, "grad_norm": 0.4928740300075381, "learning_rate": 7.104631991021115e-05, "loss": 2.8632, "step": 20565 }, { "epoch": 1.2766776336209573, "grad_norm": 0.5561549694880126, "learning_rate": 7.104304386052619e-05, "loss": 2.8183, "step": 20566 }, { "epoch": 1.2767397107207152, "grad_norm": 0.6029534298467163, "learning_rate": 7.103976770105455e-05, "loss": 2.7844, "step": 20567 }, { "epoch": 1.2768017878204732, "grad_norm": 0.5848031061936035, "learning_rate": 7.103649143181335e-05, "loss": 2.8587, "step": 20568 }, { "epoch": 1.2768638649202309, "grad_norm": 0.6287061479569138, "learning_rate": 7.103321505281965e-05, "loss": 2.7746, "step": 20569 }, { "epoch": 1.2769259420199888, "grad_norm": 0.5239596822530422, "learning_rate": 7.102993856409057e-05, "loss": 2.8958, "step": 20570 }, { "epoch": 1.2769880191197467, "grad_norm": 0.6310118972122241, "learning_rate": 7.10266619656432e-05, "loss": 2.9938, "step": 20571 }, { "epoch": 1.2770500962195046, "grad_norm": 0.545873296769518, "learning_rate": 7.102338525749461e-05, "loss": 2.777, "step": 20572 }, { "epoch": 1.2771121733192625, "grad_norm": 0.5427681257242644, "learning_rate": 7.102010843966192e-05, "loss": 2.7801, "step": 20573 }, { "epoch": 1.2771742504190204, "grad_norm": 0.5999729657043178, "learning_rate": 7.101683151216222e-05, "loss": 2.8472, "step": 20574 }, { "epoch": 1.2772363275187784, "grad_norm": 0.5243258553028475, "learning_rate": 7.101355447501264e-05, "loss": 2.814, "step": 20575 }, { "epoch": 1.2772984046185363, "grad_norm": 0.5576653401282464, "learning_rate": 7.101027732823022e-05, "loss": 2.794, "step": 20576 }, { "epoch": 1.2773604817182942, "grad_norm": 0.5201234812110411, "learning_rate": 7.100700007183207e-05, "loss": 2.8053, "step": 20577 }, { "epoch": 1.277422558818052, "grad_norm": 0.5233686601151396, "learning_rate": 7.100372270583532e-05, "loss": 2.7764, "step": 20578 }, { "epoch": 1.2774846359178098, "grad_norm": 0.5456922547794403, "learning_rate": 7.100044523025705e-05, "loss": 2.9117, "step": 20579 }, { "epoch": 1.2775467130175677, "grad_norm": 0.46092376041948113, "learning_rate": 7.099716764511435e-05, "loss": 2.662, "step": 20580 }, { "epoch": 1.2776087901173256, "grad_norm": 0.5580542032721131, "learning_rate": 7.099388995042432e-05, "loss": 2.9189, "step": 20581 }, { "epoch": 1.2776708672170836, "grad_norm": 0.5812502491858164, "learning_rate": 7.09906121462041e-05, "loss": 2.8585, "step": 20582 }, { "epoch": 1.2777329443168415, "grad_norm": 0.5172157314466674, "learning_rate": 7.098733423247074e-05, "loss": 2.7681, "step": 20583 }, { "epoch": 1.2777950214165994, "grad_norm": 0.5485039293101497, "learning_rate": 7.098405620924137e-05, "loss": 2.7783, "step": 20584 }, { "epoch": 1.2778570985163573, "grad_norm": 0.5043067818871919, "learning_rate": 7.098077807653307e-05, "loss": 2.9527, "step": 20585 }, { "epoch": 1.2779191756161152, "grad_norm": 0.5406350404986707, "learning_rate": 7.097749983436297e-05, "loss": 2.7224, "step": 20586 }, { "epoch": 1.2779812527158732, "grad_norm": 0.5845392455344699, "learning_rate": 7.097422148274816e-05, "loss": 2.7969, "step": 20587 }, { "epoch": 1.278043329815631, "grad_norm": 0.5276427939893226, "learning_rate": 7.097094302170575e-05, "loss": 2.8075, "step": 20588 }, { "epoch": 1.278105406915389, "grad_norm": 0.6388064134847313, "learning_rate": 7.096766445125283e-05, "loss": 2.8833, "step": 20589 }, { "epoch": 1.278167484015147, "grad_norm": 0.6150482865117651, "learning_rate": 7.096438577140651e-05, "loss": 2.7796, "step": 20590 }, { "epoch": 1.2782295611149048, "grad_norm": 0.6361244329943597, "learning_rate": 7.096110698218392e-05, "loss": 2.7885, "step": 20591 }, { "epoch": 1.2782916382146627, "grad_norm": 0.5290616628388993, "learning_rate": 7.095782808360214e-05, "loss": 2.7557, "step": 20592 }, { "epoch": 1.2783537153144204, "grad_norm": 0.5537264920964763, "learning_rate": 7.095454907567827e-05, "loss": 2.7306, "step": 20593 }, { "epoch": 1.2784157924141784, "grad_norm": 0.6538932352354629, "learning_rate": 7.095126995842945e-05, "loss": 2.9328, "step": 20594 }, { "epoch": 1.2784778695139363, "grad_norm": 0.5865691277399316, "learning_rate": 7.094799073187274e-05, "loss": 2.846, "step": 20595 }, { "epoch": 1.2785399466136942, "grad_norm": 0.5604306409528182, "learning_rate": 7.09447113960253e-05, "loss": 2.8217, "step": 20596 }, { "epoch": 1.278602023713452, "grad_norm": 0.5783021500816792, "learning_rate": 7.094143195090421e-05, "loss": 2.8038, "step": 20597 }, { "epoch": 1.27866410081321, "grad_norm": 0.49021175985872084, "learning_rate": 7.09381523965266e-05, "loss": 2.8581, "step": 20598 }, { "epoch": 1.278726177912968, "grad_norm": 0.585188093961467, "learning_rate": 7.093487273290954e-05, "loss": 2.7751, "step": 20599 }, { "epoch": 1.2787882550127259, "grad_norm": 0.5507268946455091, "learning_rate": 7.093159296007018e-05, "loss": 2.8734, "step": 20600 }, { "epoch": 1.2788503321124838, "grad_norm": 0.5958326355278712, "learning_rate": 7.09283130780256e-05, "loss": 2.8684, "step": 20601 }, { "epoch": 1.2789124092122415, "grad_norm": 0.5549856915648552, "learning_rate": 7.092503308679297e-05, "loss": 2.7291, "step": 20602 }, { "epoch": 1.2789744863119994, "grad_norm": 0.520424068506205, "learning_rate": 7.092175298638933e-05, "loss": 2.7781, "step": 20603 }, { "epoch": 1.2790365634117573, "grad_norm": 0.6223050906590868, "learning_rate": 7.091847277683183e-05, "loss": 2.8521, "step": 20604 }, { "epoch": 1.2790986405115152, "grad_norm": 0.4989434792199647, "learning_rate": 7.09151924581376e-05, "loss": 2.8716, "step": 20605 }, { "epoch": 1.2791607176112731, "grad_norm": 0.5403855438456795, "learning_rate": 7.091191203032371e-05, "loss": 2.7761, "step": 20606 }, { "epoch": 1.279222794711031, "grad_norm": 0.5213087836401672, "learning_rate": 7.090863149340731e-05, "loss": 2.8482, "step": 20607 }, { "epoch": 1.279284871810789, "grad_norm": 0.5430645480824705, "learning_rate": 7.09053508474055e-05, "loss": 2.9233, "step": 20608 }, { "epoch": 1.279346948910547, "grad_norm": 0.5061413189119521, "learning_rate": 7.09020700923354e-05, "loss": 2.8215, "step": 20609 }, { "epoch": 1.2794090260103048, "grad_norm": 0.5768043279702378, "learning_rate": 7.089878922821412e-05, "loss": 2.8713, "step": 20610 }, { "epoch": 1.2794711031100627, "grad_norm": 0.538219262306312, "learning_rate": 7.08955082550588e-05, "loss": 2.8056, "step": 20611 }, { "epoch": 1.2795331802098207, "grad_norm": 0.5442058548780937, "learning_rate": 7.089222717288651e-05, "loss": 2.8726, "step": 20612 }, { "epoch": 1.2795952573095786, "grad_norm": 0.5649819981821387, "learning_rate": 7.088894598171443e-05, "loss": 2.8381, "step": 20613 }, { "epoch": 1.2796573344093365, "grad_norm": 0.5225960530054277, "learning_rate": 7.088566468155965e-05, "loss": 2.8423, "step": 20614 }, { "epoch": 1.2797194115090944, "grad_norm": 0.6169557015407593, "learning_rate": 7.088238327243926e-05, "loss": 2.8264, "step": 20615 }, { "epoch": 1.2797814886088523, "grad_norm": 0.5392608001416214, "learning_rate": 7.087910175437044e-05, "loss": 2.7938, "step": 20616 }, { "epoch": 1.27984356570861, "grad_norm": 0.6027687166805489, "learning_rate": 7.087582012737025e-05, "loss": 2.7229, "step": 20617 }, { "epoch": 1.279905642808368, "grad_norm": 0.6182623569628034, "learning_rate": 7.087253839145584e-05, "loss": 2.8067, "step": 20618 }, { "epoch": 1.2799677199081259, "grad_norm": 0.5967178457392623, "learning_rate": 7.086925654664435e-05, "loss": 2.8257, "step": 20619 }, { "epoch": 1.2800297970078838, "grad_norm": 0.6717022074800706, "learning_rate": 7.086597459295287e-05, "loss": 2.8898, "step": 20620 }, { "epoch": 1.2800918741076417, "grad_norm": 0.5670643370607115, "learning_rate": 7.086269253039853e-05, "loss": 2.8161, "step": 20621 }, { "epoch": 1.2801539512073996, "grad_norm": 0.5940867021457751, "learning_rate": 7.085941035899847e-05, "loss": 2.9078, "step": 20622 }, { "epoch": 1.2802160283071575, "grad_norm": 0.5964857369977569, "learning_rate": 7.085612807876979e-05, "loss": 2.7688, "step": 20623 }, { "epoch": 1.2802781054069154, "grad_norm": 0.5399010811379622, "learning_rate": 7.085284568972964e-05, "loss": 2.7263, "step": 20624 }, { "epoch": 1.2803401825066734, "grad_norm": 0.5671903510324162, "learning_rate": 7.084956319189512e-05, "loss": 2.8944, "step": 20625 }, { "epoch": 1.280402259606431, "grad_norm": 0.6985827702406208, "learning_rate": 7.084628058528338e-05, "loss": 2.8071, "step": 20626 }, { "epoch": 1.280464336706189, "grad_norm": 0.5575123970723234, "learning_rate": 7.084299786991154e-05, "loss": 2.8713, "step": 20627 }, { "epoch": 1.280526413805947, "grad_norm": 0.6289589901247119, "learning_rate": 7.083971504579668e-05, "loss": 2.8396, "step": 20628 }, { "epoch": 1.2805884909057048, "grad_norm": 0.6645440991897481, "learning_rate": 7.0836432112956e-05, "loss": 2.8652, "step": 20629 }, { "epoch": 1.2806505680054627, "grad_norm": 0.5692475448223739, "learning_rate": 7.083314907140659e-05, "loss": 2.8401, "step": 20630 }, { "epoch": 1.2807126451052206, "grad_norm": 0.5903860971383942, "learning_rate": 7.082986592116558e-05, "loss": 2.8359, "step": 20631 }, { "epoch": 1.2807747222049786, "grad_norm": 0.5884387986973608, "learning_rate": 7.082658266225012e-05, "loss": 2.8849, "step": 20632 }, { "epoch": 1.2808367993047365, "grad_norm": 0.5637723255600328, "learning_rate": 7.082329929467731e-05, "loss": 2.7587, "step": 20633 }, { "epoch": 1.2808988764044944, "grad_norm": 0.5861890888495591, "learning_rate": 7.082001581846429e-05, "loss": 2.7748, "step": 20634 }, { "epoch": 1.2809609535042523, "grad_norm": 0.5284240959177433, "learning_rate": 7.081673223362819e-05, "loss": 2.851, "step": 20635 }, { "epoch": 1.2810230306040102, "grad_norm": 0.5396910886308364, "learning_rate": 7.081344854018614e-05, "loss": 2.8046, "step": 20636 }, { "epoch": 1.2810851077037682, "grad_norm": 0.535277192866009, "learning_rate": 7.08101647381553e-05, "loss": 2.8451, "step": 20637 }, { "epoch": 1.281147184803526, "grad_norm": 0.5293272790938932, "learning_rate": 7.080688082755275e-05, "loss": 2.8477, "step": 20638 }, { "epoch": 1.281209261903284, "grad_norm": 0.5490715600158038, "learning_rate": 7.080359680839566e-05, "loss": 2.8143, "step": 20639 }, { "epoch": 1.281271339003042, "grad_norm": 0.5863850727927918, "learning_rate": 7.080031268070117e-05, "loss": 2.8665, "step": 20640 }, { "epoch": 1.2813334161027996, "grad_norm": 0.5165852362030587, "learning_rate": 7.07970284444864e-05, "loss": 2.8101, "step": 20641 }, { "epoch": 1.2813954932025575, "grad_norm": 0.5624849794289677, "learning_rate": 7.079374409976847e-05, "loss": 2.6659, "step": 20642 }, { "epoch": 1.2814575703023154, "grad_norm": 0.6052151302627868, "learning_rate": 7.079045964656452e-05, "loss": 2.7889, "step": 20643 }, { "epoch": 1.2815196474020734, "grad_norm": 0.5237887613069617, "learning_rate": 7.07871750848917e-05, "loss": 2.7956, "step": 20644 }, { "epoch": 1.2815817245018313, "grad_norm": 0.5953155257180685, "learning_rate": 7.078389041476716e-05, "loss": 2.8535, "step": 20645 }, { "epoch": 1.2816438016015892, "grad_norm": 0.623800628954285, "learning_rate": 7.078060563620801e-05, "loss": 2.8153, "step": 20646 }, { "epoch": 1.2817058787013471, "grad_norm": 0.595120955362339, "learning_rate": 7.07773207492314e-05, "loss": 2.715, "step": 20647 }, { "epoch": 1.281767955801105, "grad_norm": 0.5575584741469942, "learning_rate": 7.077403575385445e-05, "loss": 2.8037, "step": 20648 }, { "epoch": 1.281830032900863, "grad_norm": 0.6218217625665715, "learning_rate": 7.077075065009433e-05, "loss": 2.8568, "step": 20649 }, { "epoch": 1.2818921100006206, "grad_norm": 0.5389207017685659, "learning_rate": 7.076746543796814e-05, "loss": 2.7562, "step": 20650 }, { "epoch": 1.2819541871003786, "grad_norm": 0.5749534266725292, "learning_rate": 7.076418011749307e-05, "loss": 2.8929, "step": 20651 }, { "epoch": 1.2820162642001365, "grad_norm": 0.545651921004685, "learning_rate": 7.076089468868621e-05, "loss": 2.7074, "step": 20652 }, { "epoch": 1.2820783412998944, "grad_norm": 0.6266598594918068, "learning_rate": 7.075760915156472e-05, "loss": 2.8955, "step": 20653 }, { "epoch": 1.2821404183996523, "grad_norm": 0.6449717779047347, "learning_rate": 7.075432350614576e-05, "loss": 2.805, "step": 20654 }, { "epoch": 1.2822024954994102, "grad_norm": 0.5472946191985674, "learning_rate": 7.075103775244643e-05, "loss": 2.8271, "step": 20655 }, { "epoch": 1.2822645725991682, "grad_norm": 0.6298941914321632, "learning_rate": 7.074775189048392e-05, "loss": 2.7959, "step": 20656 }, { "epoch": 1.282326649698926, "grad_norm": 0.5905342102913249, "learning_rate": 7.074446592027536e-05, "loss": 2.7188, "step": 20657 }, { "epoch": 1.282388726798684, "grad_norm": 0.598882501749864, "learning_rate": 7.074117984183787e-05, "loss": 2.8748, "step": 20658 }, { "epoch": 1.282450803898442, "grad_norm": 0.6286220026648324, "learning_rate": 7.073789365518861e-05, "loss": 2.7364, "step": 20659 }, { "epoch": 1.2825128809981998, "grad_norm": 0.55703509892529, "learning_rate": 7.073460736034473e-05, "loss": 2.9256, "step": 20660 }, { "epoch": 1.2825749580979577, "grad_norm": 0.5979474174974528, "learning_rate": 7.073132095732335e-05, "loss": 2.8014, "step": 20661 }, { "epoch": 1.2826370351977157, "grad_norm": 0.5364670675609866, "learning_rate": 7.072803444614165e-05, "loss": 2.8729, "step": 20662 }, { "epoch": 1.2826991122974736, "grad_norm": 0.5231102731926539, "learning_rate": 7.072474782681677e-05, "loss": 2.8033, "step": 20663 }, { "epoch": 1.2827611893972315, "grad_norm": 0.5209013290525264, "learning_rate": 7.072146109936585e-05, "loss": 2.8832, "step": 20664 }, { "epoch": 1.2828232664969892, "grad_norm": 0.5437291810007336, "learning_rate": 7.071817426380603e-05, "loss": 2.8267, "step": 20665 }, { "epoch": 1.282885343596747, "grad_norm": 0.5439944835744279, "learning_rate": 7.071488732015446e-05, "loss": 2.8537, "step": 20666 }, { "epoch": 1.282947420696505, "grad_norm": 0.5632884010569988, "learning_rate": 7.071160026842831e-05, "loss": 2.8459, "step": 20667 }, { "epoch": 1.283009497796263, "grad_norm": 0.5587971723728686, "learning_rate": 7.07083131086447e-05, "loss": 2.8748, "step": 20668 }, { "epoch": 1.2830715748960209, "grad_norm": 0.5736309858065112, "learning_rate": 7.070502584082081e-05, "loss": 2.8527, "step": 20669 }, { "epoch": 1.2831336519957788, "grad_norm": 0.626199120341097, "learning_rate": 7.070173846497377e-05, "loss": 2.8985, "step": 20670 }, { "epoch": 1.2831957290955367, "grad_norm": 0.5114620817193413, "learning_rate": 7.069845098112071e-05, "loss": 2.8508, "step": 20671 }, { "epoch": 1.2832578061952946, "grad_norm": 0.6321601802185737, "learning_rate": 7.069516338927883e-05, "loss": 2.8515, "step": 20672 }, { "epoch": 1.2833198832950525, "grad_norm": 0.5307035440605864, "learning_rate": 7.069187568946525e-05, "loss": 2.973, "step": 20673 }, { "epoch": 1.2833819603948102, "grad_norm": 0.5710964823812724, "learning_rate": 7.068858788169714e-05, "loss": 2.9074, "step": 20674 }, { "epoch": 1.2834440374945681, "grad_norm": 0.4847286686731245, "learning_rate": 7.068529996599163e-05, "loss": 2.7751, "step": 20675 }, { "epoch": 1.283506114594326, "grad_norm": 0.538687700090328, "learning_rate": 7.06820119423659e-05, "loss": 2.7632, "step": 20676 }, { "epoch": 1.283568191694084, "grad_norm": 0.5209006138896638, "learning_rate": 7.067872381083709e-05, "loss": 2.8176, "step": 20677 }, { "epoch": 1.283630268793842, "grad_norm": 0.6293283076993191, "learning_rate": 7.067543557142234e-05, "loss": 2.7708, "step": 20678 }, { "epoch": 1.2836923458935998, "grad_norm": 0.5690692780964044, "learning_rate": 7.067214722413885e-05, "loss": 2.8345, "step": 20679 }, { "epoch": 1.2837544229933577, "grad_norm": 0.5381559129136533, "learning_rate": 7.066885876900373e-05, "loss": 2.8066, "step": 20680 }, { "epoch": 1.2838165000931157, "grad_norm": 0.6821766579256144, "learning_rate": 7.066557020603418e-05, "loss": 2.837, "step": 20681 }, { "epoch": 1.2838785771928736, "grad_norm": 0.6053576952083365, "learning_rate": 7.066228153524733e-05, "loss": 2.7553, "step": 20682 }, { "epoch": 1.2839406542926315, "grad_norm": 0.59881429737149, "learning_rate": 7.065899275666033e-05, "loss": 2.8773, "step": 20683 }, { "epoch": 1.2840027313923894, "grad_norm": 0.7388103143562714, "learning_rate": 7.065570387029033e-05, "loss": 2.8662, "step": 20684 }, { "epoch": 1.2840648084921473, "grad_norm": 0.536666743217289, "learning_rate": 7.065241487615455e-05, "loss": 2.8188, "step": 20685 }, { "epoch": 1.2841268855919052, "grad_norm": 0.5778574453299538, "learning_rate": 7.064912577427008e-05, "loss": 2.8036, "step": 20686 }, { "epoch": 1.2841889626916632, "grad_norm": 0.5850426007107977, "learning_rate": 7.064583656465411e-05, "loss": 2.8957, "step": 20687 }, { "epoch": 1.284251039791421, "grad_norm": 0.5347375275814581, "learning_rate": 7.064254724732382e-05, "loss": 2.8456, "step": 20688 }, { "epoch": 1.2843131168911788, "grad_norm": 0.5876211011132829, "learning_rate": 7.063925782229632e-05, "loss": 2.8673, "step": 20689 }, { "epoch": 1.2843751939909367, "grad_norm": 0.53882682181775, "learning_rate": 7.063596828958883e-05, "loss": 2.8364, "step": 20690 }, { "epoch": 1.2844372710906946, "grad_norm": 0.5714871315879855, "learning_rate": 7.063267864921847e-05, "loss": 2.7791, "step": 20691 }, { "epoch": 1.2844993481904525, "grad_norm": 0.5601235628899202, "learning_rate": 7.062938890120241e-05, "loss": 2.8157, "step": 20692 }, { "epoch": 1.2845614252902104, "grad_norm": 0.49531734716889647, "learning_rate": 7.062609904555782e-05, "loss": 2.8734, "step": 20693 }, { "epoch": 1.2846235023899684, "grad_norm": 0.5947070189022524, "learning_rate": 7.062280908230188e-05, "loss": 2.7537, "step": 20694 }, { "epoch": 1.2846855794897263, "grad_norm": 0.518053951706959, "learning_rate": 7.061951901145173e-05, "loss": 2.8654, "step": 20695 }, { "epoch": 1.2847476565894842, "grad_norm": 0.558489969923341, "learning_rate": 7.061622883302454e-05, "loss": 2.8768, "step": 20696 }, { "epoch": 1.2848097336892421, "grad_norm": 0.512830008115496, "learning_rate": 7.061293854703747e-05, "loss": 2.7717, "step": 20697 }, { "epoch": 1.2848718107889998, "grad_norm": 0.5814607115254641, "learning_rate": 7.060964815350771e-05, "loss": 2.8316, "step": 20698 }, { "epoch": 1.2849338878887577, "grad_norm": 0.5665846445055636, "learning_rate": 7.060635765245242e-05, "loss": 2.7863, "step": 20699 }, { "epoch": 1.2849959649885156, "grad_norm": 0.5423921226441447, "learning_rate": 7.060306704388875e-05, "loss": 2.8484, "step": 20700 }, { "epoch": 1.2850580420882736, "grad_norm": 0.5844263554002973, "learning_rate": 7.059977632783388e-05, "loss": 2.9049, "step": 20701 }, { "epoch": 1.2851201191880315, "grad_norm": 0.5267613621225444, "learning_rate": 7.059648550430497e-05, "loss": 2.7796, "step": 20702 }, { "epoch": 1.2851821962877894, "grad_norm": 0.5106521147739176, "learning_rate": 7.05931945733192e-05, "loss": 2.8852, "step": 20703 }, { "epoch": 1.2852442733875473, "grad_norm": 0.7621856823313405, "learning_rate": 7.058990353489372e-05, "loss": 2.9261, "step": 20704 }, { "epoch": 1.2853063504873052, "grad_norm": 0.6148750103343475, "learning_rate": 7.058661238904574e-05, "loss": 2.7716, "step": 20705 }, { "epoch": 1.2853684275870632, "grad_norm": 0.5379831870227875, "learning_rate": 7.058332113579241e-05, "loss": 2.7613, "step": 20706 }, { "epoch": 1.285430504686821, "grad_norm": 0.6717725941883429, "learning_rate": 7.058002977515087e-05, "loss": 2.8793, "step": 20707 }, { "epoch": 1.285492581786579, "grad_norm": 0.5850501143761603, "learning_rate": 7.057673830713834e-05, "loss": 2.7899, "step": 20708 }, { "epoch": 1.285554658886337, "grad_norm": 0.6149700989858883, "learning_rate": 7.057344673177195e-05, "loss": 2.863, "step": 20709 }, { "epoch": 1.2856167359860948, "grad_norm": 0.5969209074171261, "learning_rate": 7.05701550490689e-05, "loss": 2.8115, "step": 20710 }, { "epoch": 1.2856788130858527, "grad_norm": 0.6030351522457502, "learning_rate": 7.056686325904636e-05, "loss": 2.8555, "step": 20711 }, { "epoch": 1.2857408901856107, "grad_norm": 0.5880693183103896, "learning_rate": 7.056357136172152e-05, "loss": 2.8176, "step": 20712 }, { "epoch": 1.2858029672853684, "grad_norm": 0.6053575967464683, "learning_rate": 7.056027935711152e-05, "loss": 2.7762, "step": 20713 }, { "epoch": 1.2858650443851263, "grad_norm": 0.5717669192047146, "learning_rate": 7.055698724523354e-05, "loss": 2.8768, "step": 20714 }, { "epoch": 1.2859271214848842, "grad_norm": 0.668059114288964, "learning_rate": 7.055369502610478e-05, "loss": 2.9032, "step": 20715 }, { "epoch": 1.2859891985846421, "grad_norm": 0.5304676915468604, "learning_rate": 7.05504026997424e-05, "loss": 2.8814, "step": 20716 }, { "epoch": 1.2860512756844, "grad_norm": 0.6749999540823461, "learning_rate": 7.054711026616359e-05, "loss": 2.8426, "step": 20717 }, { "epoch": 1.286113352784158, "grad_norm": 0.5555893189051688, "learning_rate": 7.054381772538551e-05, "loss": 2.8977, "step": 20718 }, { "epoch": 1.2861754298839159, "grad_norm": 0.6138270433864748, "learning_rate": 7.054052507742535e-05, "loss": 2.8445, "step": 20719 }, { "epoch": 1.2862375069836738, "grad_norm": 0.6066948811769878, "learning_rate": 7.053723232230028e-05, "loss": 2.7251, "step": 20720 }, { "epoch": 1.2862995840834317, "grad_norm": 0.5168918306131332, "learning_rate": 7.053393946002748e-05, "loss": 2.8686, "step": 20721 }, { "epoch": 1.2863616611831894, "grad_norm": 0.5880813035937597, "learning_rate": 7.053064649062414e-05, "loss": 2.8725, "step": 20722 }, { "epoch": 1.2864237382829473, "grad_norm": 0.5820624323946966, "learning_rate": 7.052735341410744e-05, "loss": 2.812, "step": 20723 }, { "epoch": 1.2864858153827052, "grad_norm": 0.527681708655046, "learning_rate": 7.052406023049453e-05, "loss": 2.8965, "step": 20724 }, { "epoch": 1.2865478924824632, "grad_norm": 0.5809564074578514, "learning_rate": 7.052076693980264e-05, "loss": 2.8343, "step": 20725 }, { "epoch": 1.286609969582221, "grad_norm": 0.5184243823341229, "learning_rate": 7.051747354204892e-05, "loss": 2.87, "step": 20726 }, { "epoch": 1.286672046681979, "grad_norm": 0.6650044212337746, "learning_rate": 7.051418003725054e-05, "loss": 2.7662, "step": 20727 }, { "epoch": 1.286734123781737, "grad_norm": 0.6507679758172092, "learning_rate": 7.051088642542472e-05, "loss": 2.8659, "step": 20728 }, { "epoch": 1.2867962008814948, "grad_norm": 0.5146358133754688, "learning_rate": 7.050759270658864e-05, "loss": 2.8472, "step": 20729 }, { "epoch": 1.2868582779812527, "grad_norm": 0.6617298670243613, "learning_rate": 7.050429888075944e-05, "loss": 2.7404, "step": 20730 }, { "epoch": 1.2869203550810107, "grad_norm": 0.5484182924999017, "learning_rate": 7.050100494795437e-05, "loss": 2.8931, "step": 20731 }, { "epoch": 1.2869824321807686, "grad_norm": 0.6645512072851202, "learning_rate": 7.049771090819055e-05, "loss": 2.8427, "step": 20732 }, { "epoch": 1.2870445092805265, "grad_norm": 0.5272966717085976, "learning_rate": 7.049441676148522e-05, "loss": 2.8018, "step": 20733 }, { "epoch": 1.2871065863802844, "grad_norm": 0.6598647788994632, "learning_rate": 7.049112250785552e-05, "loss": 2.82, "step": 20734 }, { "epoch": 1.2871686634800423, "grad_norm": 0.5761654352377805, "learning_rate": 7.048782814731868e-05, "loss": 2.8847, "step": 20735 }, { "epoch": 1.2872307405798002, "grad_norm": 0.5851106022345083, "learning_rate": 7.048453367989185e-05, "loss": 2.7957, "step": 20736 }, { "epoch": 1.287292817679558, "grad_norm": 0.6145360950052987, "learning_rate": 7.048123910559225e-05, "loss": 2.8818, "step": 20737 }, { "epoch": 1.2873548947793159, "grad_norm": 0.5794839866785615, "learning_rate": 7.047794442443703e-05, "loss": 2.8738, "step": 20738 }, { "epoch": 1.2874169718790738, "grad_norm": 0.5422389388075237, "learning_rate": 7.047464963644344e-05, "loss": 2.7727, "step": 20739 }, { "epoch": 1.2874790489788317, "grad_norm": 0.5253027803519185, "learning_rate": 7.04713547416286e-05, "loss": 2.904, "step": 20740 }, { "epoch": 1.2875411260785896, "grad_norm": 0.5823541807853088, "learning_rate": 7.046805974000975e-05, "loss": 2.7608, "step": 20741 }, { "epoch": 1.2876032031783475, "grad_norm": 0.546487616528706, "learning_rate": 7.046476463160407e-05, "loss": 2.8673, "step": 20742 }, { "epoch": 1.2876652802781055, "grad_norm": 0.6029036051326825, "learning_rate": 7.04614694164287e-05, "loss": 2.8536, "step": 20743 }, { "epoch": 1.2877273573778634, "grad_norm": 0.5727520590619856, "learning_rate": 7.045817409450092e-05, "loss": 2.8882, "step": 20744 }, { "epoch": 1.2877894344776213, "grad_norm": 0.46237316261064626, "learning_rate": 7.045487866583787e-05, "loss": 2.8007, "step": 20745 }, { "epoch": 1.287851511577379, "grad_norm": 0.6253196614096996, "learning_rate": 7.045158313045674e-05, "loss": 2.8077, "step": 20746 }, { "epoch": 1.287913588677137, "grad_norm": 0.5375691058956363, "learning_rate": 7.044828748837475e-05, "loss": 2.7583, "step": 20747 }, { "epoch": 1.2879756657768948, "grad_norm": 0.5220428219802563, "learning_rate": 7.044499173960907e-05, "loss": 2.8149, "step": 20748 }, { "epoch": 1.2880377428766527, "grad_norm": 0.5253459812185227, "learning_rate": 7.044169588417693e-05, "loss": 2.7873, "step": 20749 }, { "epoch": 1.2880998199764107, "grad_norm": 0.5340399690164399, "learning_rate": 7.043839992209547e-05, "loss": 2.7059, "step": 20750 }, { "epoch": 1.2881618970761686, "grad_norm": 0.533792442537194, "learning_rate": 7.043510385338195e-05, "loss": 2.7897, "step": 20751 }, { "epoch": 1.2882239741759265, "grad_norm": 0.52802723989879, "learning_rate": 7.04318076780535e-05, "loss": 2.7326, "step": 20752 }, { "epoch": 1.2882860512756844, "grad_norm": 0.6131550118336785, "learning_rate": 7.042851139612736e-05, "loss": 2.7983, "step": 20753 }, { "epoch": 1.2883481283754423, "grad_norm": 0.5222614513564272, "learning_rate": 7.042521500762072e-05, "loss": 2.9022, "step": 20754 }, { "epoch": 1.2884102054752002, "grad_norm": 0.5922009942120159, "learning_rate": 7.042191851255076e-05, "loss": 2.8307, "step": 20755 }, { "epoch": 1.2884722825749582, "grad_norm": 0.5481578766152948, "learning_rate": 7.041862191093471e-05, "loss": 2.8874, "step": 20756 }, { "epoch": 1.288534359674716, "grad_norm": 0.5629442368124715, "learning_rate": 7.041532520278976e-05, "loss": 2.823, "step": 20757 }, { "epoch": 1.288596436774474, "grad_norm": 0.6457561298083855, "learning_rate": 7.041202838813307e-05, "loss": 2.7946, "step": 20758 }, { "epoch": 1.288658513874232, "grad_norm": 0.5019120670848665, "learning_rate": 7.04087314669819e-05, "loss": 2.7924, "step": 20759 }, { "epoch": 1.2887205909739898, "grad_norm": 0.5231375617794428, "learning_rate": 7.040543443935342e-05, "loss": 2.8209, "step": 20760 }, { "epoch": 1.2887826680737475, "grad_norm": 0.627964665460603, "learning_rate": 7.040213730526482e-05, "loss": 2.86, "step": 20761 }, { "epoch": 1.2888447451735054, "grad_norm": 0.546276391611731, "learning_rate": 7.039884006473334e-05, "loss": 2.8441, "step": 20762 }, { "epoch": 1.2889068222732634, "grad_norm": 0.5776449478608017, "learning_rate": 7.039554271777615e-05, "loss": 2.8436, "step": 20763 }, { "epoch": 1.2889688993730213, "grad_norm": 0.48698407100989044, "learning_rate": 7.039224526441047e-05, "loss": 2.8589, "step": 20764 }, { "epoch": 1.2890309764727792, "grad_norm": 0.6066623858241421, "learning_rate": 7.038894770465346e-05, "loss": 2.8835, "step": 20765 }, { "epoch": 1.2890930535725371, "grad_norm": 0.45885067879617364, "learning_rate": 7.03856500385224e-05, "loss": 2.7695, "step": 20766 }, { "epoch": 1.289155130672295, "grad_norm": 0.5311171421646753, "learning_rate": 7.038235226603445e-05, "loss": 2.7858, "step": 20767 }, { "epoch": 1.289217207772053, "grad_norm": 0.541155690264645, "learning_rate": 7.03790543872068e-05, "loss": 2.8251, "step": 20768 }, { "epoch": 1.2892792848718109, "grad_norm": 0.5734118286955804, "learning_rate": 7.037575640205669e-05, "loss": 2.8709, "step": 20769 }, { "epoch": 1.2893413619715686, "grad_norm": 0.6095380075994923, "learning_rate": 7.037245831060131e-05, "loss": 2.8079, "step": 20770 }, { "epoch": 1.2894034390713265, "grad_norm": 0.5713512941031516, "learning_rate": 7.036916011285788e-05, "loss": 2.8807, "step": 20771 }, { "epoch": 1.2894655161710844, "grad_norm": 0.5805295735218092, "learning_rate": 7.036586180884358e-05, "loss": 2.8176, "step": 20772 }, { "epoch": 1.2895275932708423, "grad_norm": 0.5895874527651382, "learning_rate": 7.036256339857564e-05, "loss": 2.8349, "step": 20773 }, { "epoch": 1.2895896703706002, "grad_norm": 0.6029268621212297, "learning_rate": 7.035926488207127e-05, "loss": 2.8214, "step": 20774 }, { "epoch": 1.2896517474703582, "grad_norm": 0.512825969214633, "learning_rate": 7.035596625934767e-05, "loss": 2.8496, "step": 20775 }, { "epoch": 1.289713824570116, "grad_norm": 0.6361199588197336, "learning_rate": 7.035266753042204e-05, "loss": 2.89, "step": 20776 }, { "epoch": 1.289775901669874, "grad_norm": 0.5453470956928061, "learning_rate": 7.034936869531161e-05, "loss": 2.9219, "step": 20777 }, { "epoch": 1.289837978769632, "grad_norm": 0.5738351280039565, "learning_rate": 7.03460697540336e-05, "loss": 2.8025, "step": 20778 }, { "epoch": 1.2899000558693898, "grad_norm": 0.5175102333310188, "learning_rate": 7.034277070660518e-05, "loss": 2.8314, "step": 20779 }, { "epoch": 1.2899621329691477, "grad_norm": 0.5213281633241863, "learning_rate": 7.033947155304361e-05, "loss": 2.8058, "step": 20780 }, { "epoch": 1.2900242100689057, "grad_norm": 0.5475583575339885, "learning_rate": 7.033617229336606e-05, "loss": 2.791, "step": 20781 }, { "epoch": 1.2900862871686636, "grad_norm": 0.5304091474844669, "learning_rate": 7.033287292758977e-05, "loss": 2.8354, "step": 20782 }, { "epoch": 1.2901483642684215, "grad_norm": 0.6352640396001927, "learning_rate": 7.032957345573194e-05, "loss": 2.8172, "step": 20783 }, { "epoch": 1.2902104413681794, "grad_norm": 0.5510654392655104, "learning_rate": 7.03262738778098e-05, "loss": 2.8167, "step": 20784 }, { "epoch": 1.2902725184679371, "grad_norm": 0.5882620161136762, "learning_rate": 7.032297419384056e-05, "loss": 2.8135, "step": 20785 }, { "epoch": 1.290334595567695, "grad_norm": 0.535277192866009, "learning_rate": 7.031967440384142e-05, "loss": 2.8938, "step": 20786 }, { "epoch": 1.290396672667453, "grad_norm": 0.5973858938634704, "learning_rate": 7.03163745078296e-05, "loss": 2.8994, "step": 20787 }, { "epoch": 1.2904587497672109, "grad_norm": 0.5075725281852912, "learning_rate": 7.031307450582233e-05, "loss": 2.8669, "step": 20788 }, { "epoch": 1.2905208268669688, "grad_norm": 0.6385862893904767, "learning_rate": 7.030977439783681e-05, "loss": 2.8898, "step": 20789 }, { "epoch": 1.2905829039667267, "grad_norm": 0.5631486278915862, "learning_rate": 7.030647418389027e-05, "loss": 2.887, "step": 20790 }, { "epoch": 1.2906449810664846, "grad_norm": 0.5744305952975155, "learning_rate": 7.030317386399993e-05, "loss": 2.9078, "step": 20791 }, { "epoch": 1.2907070581662425, "grad_norm": 0.5061871265881479, "learning_rate": 7.029987343818301e-05, "loss": 2.7782, "step": 20792 }, { "epoch": 1.2907691352660002, "grad_norm": 0.6076718518232443, "learning_rate": 7.029657290645671e-05, "loss": 2.8336, "step": 20793 }, { "epoch": 1.2908312123657582, "grad_norm": 0.5430032732849096, "learning_rate": 7.029327226883828e-05, "loss": 2.7223, "step": 20794 }, { "epoch": 1.290893289465516, "grad_norm": 0.5955703337201508, "learning_rate": 7.02899715253449e-05, "loss": 2.8917, "step": 20795 }, { "epoch": 1.290955366565274, "grad_norm": 0.5928355503376854, "learning_rate": 7.028667067599382e-05, "loss": 2.7773, "step": 20796 }, { "epoch": 1.291017443665032, "grad_norm": 0.56974160348146, "learning_rate": 7.028336972080225e-05, "loss": 2.8812, "step": 20797 }, { "epoch": 1.2910795207647898, "grad_norm": 0.5944528685945341, "learning_rate": 7.028006865978743e-05, "loss": 2.7904, "step": 20798 }, { "epoch": 1.2911415978645477, "grad_norm": 0.6200798684252231, "learning_rate": 7.027676749296655e-05, "loss": 2.8143, "step": 20799 }, { "epoch": 1.2912036749643057, "grad_norm": 0.5525556079791519, "learning_rate": 7.027346622035687e-05, "loss": 2.7283, "step": 20800 }, { "epoch": 1.2912657520640636, "grad_norm": 0.5402209784335512, "learning_rate": 7.027016484197559e-05, "loss": 2.7829, "step": 20801 }, { "epoch": 1.2913278291638215, "grad_norm": 0.5586346431248986, "learning_rate": 7.026686335783994e-05, "loss": 2.7877, "step": 20802 }, { "epoch": 1.2913899062635794, "grad_norm": 0.46387635874465716, "learning_rate": 7.026356176796713e-05, "loss": 2.7795, "step": 20803 }, { "epoch": 1.2914519833633373, "grad_norm": 0.5689945931638658, "learning_rate": 7.026026007237442e-05, "loss": 2.7906, "step": 20804 }, { "epoch": 1.2915140604630952, "grad_norm": 0.4661143112890538, "learning_rate": 7.025695827107901e-05, "loss": 2.8383, "step": 20805 }, { "epoch": 1.2915761375628532, "grad_norm": 0.5402663237309066, "learning_rate": 7.025365636409813e-05, "loss": 2.8732, "step": 20806 }, { "epoch": 1.291638214662611, "grad_norm": 0.5240260843216531, "learning_rate": 7.0250354351449e-05, "loss": 2.7585, "step": 20807 }, { "epoch": 1.2917002917623688, "grad_norm": 0.5628448594791753, "learning_rate": 7.024705223314884e-05, "loss": 2.8034, "step": 20808 }, { "epoch": 1.2917623688621267, "grad_norm": 0.51652436860494, "learning_rate": 7.024375000921493e-05, "loss": 2.7792, "step": 20809 }, { "epoch": 1.2918244459618846, "grad_norm": 0.5487644796698661, "learning_rate": 7.024044767966445e-05, "loss": 2.8265, "step": 20810 }, { "epoch": 1.2918865230616425, "grad_norm": 0.5310263162815972, "learning_rate": 7.023714524451463e-05, "loss": 2.9476, "step": 20811 }, { "epoch": 1.2919486001614005, "grad_norm": 0.5833212618487046, "learning_rate": 7.023384270378272e-05, "loss": 2.8397, "step": 20812 }, { "epoch": 1.2920106772611584, "grad_norm": 0.5469004488881959, "learning_rate": 7.023054005748592e-05, "loss": 2.8271, "step": 20813 }, { "epoch": 1.2920727543609163, "grad_norm": 0.5950164839921509, "learning_rate": 7.022723730564152e-05, "loss": 2.8517, "step": 20814 }, { "epoch": 1.2921348314606742, "grad_norm": 0.5305218755668241, "learning_rate": 7.022393444826668e-05, "loss": 2.9087, "step": 20815 }, { "epoch": 1.2921969085604321, "grad_norm": 0.6354399869631339, "learning_rate": 7.022063148537868e-05, "loss": 2.8594, "step": 20816 }, { "epoch": 1.2922589856601898, "grad_norm": 0.5514396898177966, "learning_rate": 7.021732841699473e-05, "loss": 2.8214, "step": 20817 }, { "epoch": 1.2923210627599477, "grad_norm": 0.6169498565537014, "learning_rate": 7.021402524313208e-05, "loss": 2.8158, "step": 20818 }, { "epoch": 1.2923831398597057, "grad_norm": 0.5250188653826114, "learning_rate": 7.021072196380793e-05, "loss": 2.8673, "step": 20819 }, { "epoch": 1.2924452169594636, "grad_norm": 0.5775179382114409, "learning_rate": 7.020741857903957e-05, "loss": 2.7535, "step": 20820 }, { "epoch": 1.2925072940592215, "grad_norm": 0.4932480693311459, "learning_rate": 7.020411508884417e-05, "loss": 2.8136, "step": 20821 }, { "epoch": 1.2925693711589794, "grad_norm": 0.5540604606715678, "learning_rate": 7.020081149323902e-05, "loss": 2.935, "step": 20822 }, { "epoch": 1.2926314482587373, "grad_norm": 0.5015570417445866, "learning_rate": 7.019750779224133e-05, "loss": 2.8842, "step": 20823 }, { "epoch": 1.2926935253584952, "grad_norm": 0.5659553311433634, "learning_rate": 7.019420398586832e-05, "loss": 2.8661, "step": 20824 }, { "epoch": 1.2927556024582532, "grad_norm": 0.5349573475778883, "learning_rate": 7.019090007413726e-05, "loss": 2.9398, "step": 20825 }, { "epoch": 1.292817679558011, "grad_norm": 0.5683312949883311, "learning_rate": 7.018759605706537e-05, "loss": 2.8249, "step": 20826 }, { "epoch": 1.292879756657769, "grad_norm": 0.523527849285507, "learning_rate": 7.018429193466988e-05, "loss": 2.8787, "step": 20827 }, { "epoch": 1.292941833757527, "grad_norm": 0.5662584835536976, "learning_rate": 7.018098770696806e-05, "loss": 2.8521, "step": 20828 }, { "epoch": 1.2930039108572848, "grad_norm": 0.5222423631028456, "learning_rate": 7.017768337397712e-05, "loss": 2.8294, "step": 20829 }, { "epoch": 1.2930659879570428, "grad_norm": 0.49220115400478986, "learning_rate": 7.01743789357143e-05, "loss": 2.7953, "step": 20830 }, { "epoch": 1.2931280650568007, "grad_norm": 0.5466359842323407, "learning_rate": 7.017107439219686e-05, "loss": 2.7911, "step": 20831 }, { "epoch": 1.2931901421565584, "grad_norm": 0.5460118157257587, "learning_rate": 7.016776974344202e-05, "loss": 2.8273, "step": 20832 }, { "epoch": 1.2932522192563163, "grad_norm": 0.5732672196857627, "learning_rate": 7.016446498946703e-05, "loss": 2.8499, "step": 20833 }, { "epoch": 1.2933142963560742, "grad_norm": 0.5300002825934178, "learning_rate": 7.016116013028914e-05, "loss": 2.9098, "step": 20834 }, { "epoch": 1.2933763734558321, "grad_norm": 0.6054046104667689, "learning_rate": 7.015785516592557e-05, "loss": 2.791, "step": 20835 }, { "epoch": 1.29343845055559, "grad_norm": 0.5927752222980106, "learning_rate": 7.015455009639357e-05, "loss": 2.81, "step": 20836 }, { "epoch": 1.293500527655348, "grad_norm": 0.5958424889810309, "learning_rate": 7.015124492171041e-05, "loss": 2.8597, "step": 20837 }, { "epoch": 1.2935626047551059, "grad_norm": 0.541195643252594, "learning_rate": 7.01479396418933e-05, "loss": 2.8883, "step": 20838 }, { "epoch": 1.2936246818548638, "grad_norm": 0.5588376506439934, "learning_rate": 7.01446342569595e-05, "loss": 2.8897, "step": 20839 }, { "epoch": 1.2936867589546217, "grad_norm": 0.5271249387803019, "learning_rate": 7.014132876692626e-05, "loss": 2.7989, "step": 20840 }, { "epoch": 1.2937488360543794, "grad_norm": 0.5017645812032808, "learning_rate": 7.013802317181082e-05, "loss": 2.735, "step": 20841 }, { "epoch": 1.2938109131541373, "grad_norm": 0.5685431848461233, "learning_rate": 7.013471747163041e-05, "loss": 2.865, "step": 20842 }, { "epoch": 1.2938729902538952, "grad_norm": 0.5045071353388861, "learning_rate": 7.01314116664023e-05, "loss": 2.8248, "step": 20843 }, { "epoch": 1.2939350673536532, "grad_norm": 0.5734250558443579, "learning_rate": 7.012810575614373e-05, "loss": 2.8468, "step": 20844 }, { "epoch": 1.293997144453411, "grad_norm": 0.5908246000780634, "learning_rate": 7.012479974087194e-05, "loss": 2.9041, "step": 20845 }, { "epoch": 1.294059221553169, "grad_norm": 0.5385788669195385, "learning_rate": 7.012149362060418e-05, "loss": 2.8566, "step": 20846 }, { "epoch": 1.294121298652927, "grad_norm": 0.5245206926935055, "learning_rate": 7.011818739535772e-05, "loss": 2.8004, "step": 20847 }, { "epoch": 1.2941833757526848, "grad_norm": 0.5942207528180121, "learning_rate": 7.011488106514978e-05, "loss": 2.7474, "step": 20848 }, { "epoch": 1.2942454528524427, "grad_norm": 0.5003951120882657, "learning_rate": 7.011157462999762e-05, "loss": 2.8674, "step": 20849 }, { "epoch": 1.2943075299522007, "grad_norm": 0.6052930269274848, "learning_rate": 7.010826808991851e-05, "loss": 2.8538, "step": 20850 }, { "epoch": 1.2943696070519586, "grad_norm": 0.5364104561289406, "learning_rate": 7.010496144492966e-05, "loss": 2.8985, "step": 20851 }, { "epoch": 1.2944316841517165, "grad_norm": 0.5277199710035089, "learning_rate": 7.010165469504836e-05, "loss": 2.7617, "step": 20852 }, { "epoch": 1.2944937612514744, "grad_norm": 0.5224689233472624, "learning_rate": 7.009834784029184e-05, "loss": 2.8435, "step": 20853 }, { "epoch": 1.2945558383512323, "grad_norm": 0.5868186936672738, "learning_rate": 7.009504088067738e-05, "loss": 2.8337, "step": 20854 }, { "epoch": 1.2946179154509903, "grad_norm": 0.5078171949903154, "learning_rate": 7.00917338162222e-05, "loss": 2.8046, "step": 20855 }, { "epoch": 1.294679992550748, "grad_norm": 0.5896393630247801, "learning_rate": 7.008842664694356e-05, "loss": 2.8567, "step": 20856 }, { "epoch": 1.2947420696505059, "grad_norm": 0.5061850070442168, "learning_rate": 7.008511937285873e-05, "loss": 2.7821, "step": 20857 }, { "epoch": 1.2948041467502638, "grad_norm": 0.6110782539199575, "learning_rate": 7.008181199398494e-05, "loss": 2.8984, "step": 20858 }, { "epoch": 1.2948662238500217, "grad_norm": 0.5356382719107131, "learning_rate": 7.007850451033949e-05, "loss": 2.7803, "step": 20859 }, { "epoch": 1.2949283009497796, "grad_norm": 0.5708735593372908, "learning_rate": 7.007519692193958e-05, "loss": 2.9015, "step": 20860 }, { "epoch": 1.2949903780495375, "grad_norm": 0.5433852156000659, "learning_rate": 7.007188922880252e-05, "loss": 2.9408, "step": 20861 }, { "epoch": 1.2950524551492955, "grad_norm": 0.5945938287952363, "learning_rate": 7.006858143094553e-05, "loss": 2.805, "step": 20862 }, { "epoch": 1.2951145322490534, "grad_norm": 0.5329666731275869, "learning_rate": 7.006527352838588e-05, "loss": 2.8204, "step": 20863 }, { "epoch": 1.2951766093488113, "grad_norm": 0.561703621523977, "learning_rate": 7.006196552114081e-05, "loss": 2.8477, "step": 20864 }, { "epoch": 1.295238686448569, "grad_norm": 0.5751448189766113, "learning_rate": 7.005865740922763e-05, "loss": 2.8072, "step": 20865 }, { "epoch": 1.295300763548327, "grad_norm": 0.605401632218985, "learning_rate": 7.005534919266355e-05, "loss": 2.8525, "step": 20866 }, { "epoch": 1.2953628406480848, "grad_norm": 0.5491441515946205, "learning_rate": 7.005204087146582e-05, "loss": 2.803, "step": 20867 }, { "epoch": 1.2954249177478427, "grad_norm": 0.5274454371992764, "learning_rate": 7.004873244565176e-05, "loss": 2.8378, "step": 20868 }, { "epoch": 1.2954869948476007, "grad_norm": 0.5384159080263071, "learning_rate": 7.004542391523858e-05, "loss": 2.8542, "step": 20869 }, { "epoch": 1.2955490719473586, "grad_norm": 0.5756323923470883, "learning_rate": 7.004211528024355e-05, "loss": 2.7281, "step": 20870 }, { "epoch": 1.2956111490471165, "grad_norm": 0.5925454419250517, "learning_rate": 7.003880654068395e-05, "loss": 2.9045, "step": 20871 }, { "epoch": 1.2956732261468744, "grad_norm": 0.6203916167724584, "learning_rate": 7.003549769657703e-05, "loss": 2.7399, "step": 20872 }, { "epoch": 1.2957353032466323, "grad_norm": 0.4987428477259663, "learning_rate": 7.003218874794006e-05, "loss": 2.7711, "step": 20873 }, { "epoch": 1.2957973803463902, "grad_norm": 0.6074447627523867, "learning_rate": 7.002887969479028e-05, "loss": 2.8938, "step": 20874 }, { "epoch": 1.2958594574461482, "grad_norm": 0.5117567208449304, "learning_rate": 7.002557053714499e-05, "loss": 2.7587, "step": 20875 }, { "epoch": 1.295921534545906, "grad_norm": 0.5985460327375136, "learning_rate": 7.002226127502142e-05, "loss": 2.8491, "step": 20876 }, { "epoch": 1.295983611645664, "grad_norm": 0.6268831493270147, "learning_rate": 7.001895190843689e-05, "loss": 2.8707, "step": 20877 }, { "epoch": 1.296045688745422, "grad_norm": 0.54373104457043, "learning_rate": 7.001564243740858e-05, "loss": 2.8504, "step": 20878 }, { "epoch": 1.2961077658451798, "grad_norm": 0.5957585543499789, "learning_rate": 7.001233286195383e-05, "loss": 2.7352, "step": 20879 }, { "epoch": 1.2961698429449375, "grad_norm": 0.5667055851764272, "learning_rate": 7.000902318208989e-05, "loss": 2.8051, "step": 20880 }, { "epoch": 1.2962319200446955, "grad_norm": 0.4683503195401842, "learning_rate": 7.000571339783399e-05, "loss": 2.8676, "step": 20881 }, { "epoch": 1.2962939971444534, "grad_norm": 0.6229173531629507, "learning_rate": 7.000240350920344e-05, "loss": 2.853, "step": 20882 }, { "epoch": 1.2963560742442113, "grad_norm": 0.5531758139594259, "learning_rate": 6.99990935162155e-05, "loss": 2.8836, "step": 20883 }, { "epoch": 1.2964181513439692, "grad_norm": 0.5679439324121436, "learning_rate": 6.999578341888746e-05, "loss": 2.7764, "step": 20884 }, { "epoch": 1.2964802284437271, "grad_norm": 0.5515017024780862, "learning_rate": 6.999247321723652e-05, "loss": 2.868, "step": 20885 }, { "epoch": 1.296542305543485, "grad_norm": 0.5944379535312745, "learning_rate": 6.998916291128002e-05, "loss": 2.8149, "step": 20886 }, { "epoch": 1.296604382643243, "grad_norm": 0.6280860290963812, "learning_rate": 6.998585250103519e-05, "loss": 2.8258, "step": 20887 }, { "epoch": 1.2966664597430009, "grad_norm": 0.5903422289299619, "learning_rate": 6.998254198651933e-05, "loss": 2.8342, "step": 20888 }, { "epoch": 1.2967285368427586, "grad_norm": 0.5812892921046887, "learning_rate": 6.997923136774968e-05, "loss": 2.9305, "step": 20889 }, { "epoch": 1.2967906139425165, "grad_norm": 0.5753501075002607, "learning_rate": 6.997592064474353e-05, "loss": 2.8333, "step": 20890 }, { "epoch": 1.2968526910422744, "grad_norm": 0.5813938977949754, "learning_rate": 6.997260981751816e-05, "loss": 2.8454, "step": 20891 }, { "epoch": 1.2969147681420323, "grad_norm": 0.6564578453803864, "learning_rate": 6.996929888609082e-05, "loss": 2.8305, "step": 20892 }, { "epoch": 1.2969768452417902, "grad_norm": 0.6289850979299658, "learning_rate": 6.996598785047882e-05, "loss": 2.8661, "step": 20893 }, { "epoch": 1.2970389223415482, "grad_norm": 0.6656456644136951, "learning_rate": 6.996267671069943e-05, "loss": 2.8279, "step": 20894 }, { "epoch": 1.297100999441306, "grad_norm": 0.5515514696871594, "learning_rate": 6.995936546676988e-05, "loss": 2.8861, "step": 20895 }, { "epoch": 1.297163076541064, "grad_norm": 0.6056237299288727, "learning_rate": 6.995605411870749e-05, "loss": 2.8169, "step": 20896 }, { "epoch": 1.297225153640822, "grad_norm": 0.5630882948535977, "learning_rate": 6.995274266652952e-05, "loss": 2.7512, "step": 20897 }, { "epoch": 1.2972872307405798, "grad_norm": 0.5115385539066922, "learning_rate": 6.994943111025324e-05, "loss": 2.8631, "step": 20898 }, { "epoch": 1.2973493078403378, "grad_norm": 0.584127963475931, "learning_rate": 6.994611944989594e-05, "loss": 2.7948, "step": 20899 }, { "epoch": 1.2974113849400957, "grad_norm": 0.5144175929237248, "learning_rate": 6.99428076854749e-05, "loss": 2.7694, "step": 20900 }, { "epoch": 1.2974734620398536, "grad_norm": 0.5591604652525185, "learning_rate": 6.993949581700739e-05, "loss": 2.8581, "step": 20901 }, { "epoch": 1.2975355391396115, "grad_norm": 0.5266642112865835, "learning_rate": 6.993618384451069e-05, "loss": 2.7651, "step": 20902 }, { "epoch": 1.2975976162393694, "grad_norm": 0.5627569035801877, "learning_rate": 6.993287176800205e-05, "loss": 2.8349, "step": 20903 }, { "epoch": 1.2976596933391271, "grad_norm": 0.5237275359131912, "learning_rate": 6.992955958749882e-05, "loss": 2.8109, "step": 20904 }, { "epoch": 1.297721770438885, "grad_norm": 0.5318195151358269, "learning_rate": 6.992624730301821e-05, "loss": 2.7581, "step": 20905 }, { "epoch": 1.297783847538643, "grad_norm": 0.5435550186644192, "learning_rate": 6.992293491457753e-05, "loss": 2.8714, "step": 20906 }, { "epoch": 1.2978459246384009, "grad_norm": 0.4940412971873978, "learning_rate": 6.991962242219407e-05, "loss": 2.7812, "step": 20907 }, { "epoch": 1.2979080017381588, "grad_norm": 0.5986008005466328, "learning_rate": 6.991630982588512e-05, "loss": 2.8097, "step": 20908 }, { "epoch": 1.2979700788379167, "grad_norm": 0.6009216362837578, "learning_rate": 6.991299712566794e-05, "loss": 2.8612, "step": 20909 }, { "epoch": 1.2980321559376746, "grad_norm": 0.5902293378613755, "learning_rate": 6.990968432155981e-05, "loss": 2.8191, "step": 20910 }, { "epoch": 1.2980942330374325, "grad_norm": 0.5844610303903263, "learning_rate": 6.990637141357803e-05, "loss": 2.8553, "step": 20911 }, { "epoch": 1.2981563101371905, "grad_norm": 0.5836471298296875, "learning_rate": 6.990305840173988e-05, "loss": 2.7551, "step": 20912 }, { "epoch": 1.2982183872369482, "grad_norm": 0.5591171320262066, "learning_rate": 6.989974528606263e-05, "loss": 2.8188, "step": 20913 }, { "epoch": 1.298280464336706, "grad_norm": 0.565565234657073, "learning_rate": 6.98964320665636e-05, "loss": 2.7455, "step": 20914 }, { "epoch": 1.298342541436464, "grad_norm": 0.5157294889985969, "learning_rate": 6.989311874326003e-05, "loss": 2.8081, "step": 20915 }, { "epoch": 1.298404618536222, "grad_norm": 0.6106579309210696, "learning_rate": 6.988980531616924e-05, "loss": 2.8418, "step": 20916 }, { "epoch": 1.2984666956359798, "grad_norm": 0.5418725508884228, "learning_rate": 6.988649178530852e-05, "loss": 2.849, "step": 20917 }, { "epoch": 1.2985287727357377, "grad_norm": 0.5508061126075461, "learning_rate": 6.988317815069512e-05, "loss": 2.9098, "step": 20918 }, { "epoch": 1.2985908498354957, "grad_norm": 0.6294235092606645, "learning_rate": 6.987986441234638e-05, "loss": 2.9386, "step": 20919 }, { "epoch": 1.2986529269352536, "grad_norm": 0.6068775645240161, "learning_rate": 6.987655057027956e-05, "loss": 2.8284, "step": 20920 }, { "epoch": 1.2987150040350115, "grad_norm": 0.5311977136350168, "learning_rate": 6.987323662451192e-05, "loss": 2.774, "step": 20921 }, { "epoch": 1.2987770811347694, "grad_norm": 0.6007108728983506, "learning_rate": 6.98699225750608e-05, "loss": 2.8178, "step": 20922 }, { "epoch": 1.2988391582345273, "grad_norm": 0.553240729471925, "learning_rate": 6.986660842194345e-05, "loss": 2.9259, "step": 20923 }, { "epoch": 1.2989012353342853, "grad_norm": 0.6067926269241101, "learning_rate": 6.986329416517721e-05, "loss": 2.8452, "step": 20924 }, { "epoch": 1.2989633124340432, "grad_norm": 0.5433345358441314, "learning_rate": 6.985997980477933e-05, "loss": 2.7371, "step": 20925 }, { "epoch": 1.299025389533801, "grad_norm": 0.5805147371006817, "learning_rate": 6.985666534076712e-05, "loss": 2.896, "step": 20926 }, { "epoch": 1.299087466633559, "grad_norm": 0.5261085171916993, "learning_rate": 6.985335077315787e-05, "loss": 2.8529, "step": 20927 }, { "epoch": 1.2991495437333167, "grad_norm": 0.5862600837279518, "learning_rate": 6.985003610196886e-05, "loss": 2.7542, "step": 20928 }, { "epoch": 1.2992116208330746, "grad_norm": 0.48151966627310155, "learning_rate": 6.98467213272174e-05, "loss": 2.916, "step": 20929 }, { "epoch": 1.2992736979328325, "grad_norm": 0.5703057589197867, "learning_rate": 6.984340644892076e-05, "loss": 2.8979, "step": 20930 }, { "epoch": 1.2993357750325905, "grad_norm": 0.48155429372658587, "learning_rate": 6.984009146709628e-05, "loss": 2.7339, "step": 20931 }, { "epoch": 1.2993978521323484, "grad_norm": 0.539642160539988, "learning_rate": 6.98367763817612e-05, "loss": 2.8269, "step": 20932 }, { "epoch": 1.2994599292321063, "grad_norm": 0.46222182942712786, "learning_rate": 6.983346119293285e-05, "loss": 2.7478, "step": 20933 }, { "epoch": 1.2995220063318642, "grad_norm": 0.5648340701886989, "learning_rate": 6.983014590062851e-05, "loss": 2.872, "step": 20934 }, { "epoch": 1.2995840834316221, "grad_norm": 0.507025769502383, "learning_rate": 6.98268305048655e-05, "loss": 2.7824, "step": 20935 }, { "epoch": 1.29964616053138, "grad_norm": 0.5263059789282827, "learning_rate": 6.98235150056611e-05, "loss": 2.8575, "step": 20936 }, { "epoch": 1.2997082376311377, "grad_norm": 0.5314251667033032, "learning_rate": 6.98201994030326e-05, "loss": 2.8357, "step": 20937 }, { "epoch": 1.2997703147308957, "grad_norm": 0.5853855585224953, "learning_rate": 6.981688369699733e-05, "loss": 2.8073, "step": 20938 }, { "epoch": 1.2998323918306536, "grad_norm": 0.5137199698440138, "learning_rate": 6.981356788757253e-05, "loss": 2.6894, "step": 20939 }, { "epoch": 1.2998944689304115, "grad_norm": 0.5781236854744478, "learning_rate": 6.981025197477557e-05, "loss": 2.8633, "step": 20940 }, { "epoch": 1.2999565460301694, "grad_norm": 0.5953257631841887, "learning_rate": 6.98069359586237e-05, "loss": 2.8796, "step": 20941 }, { "epoch": 1.3000186231299273, "grad_norm": 0.5369117234429912, "learning_rate": 6.980361983913424e-05, "loss": 2.7116, "step": 20942 }, { "epoch": 1.3000807002296852, "grad_norm": 0.5965800849956487, "learning_rate": 6.980030361632448e-05, "loss": 2.8039, "step": 20943 }, { "epoch": 1.3001427773294432, "grad_norm": 0.6646668320900811, "learning_rate": 6.979698729021175e-05, "loss": 2.8792, "step": 20944 }, { "epoch": 1.300204854429201, "grad_norm": 0.5630641598282692, "learning_rate": 6.979367086081333e-05, "loss": 2.8493, "step": 20945 }, { "epoch": 1.300266931528959, "grad_norm": 0.5991941126168769, "learning_rate": 6.97903543281465e-05, "loss": 2.9067, "step": 20946 }, { "epoch": 1.300329008628717, "grad_norm": 0.562586910103456, "learning_rate": 6.978703769222861e-05, "loss": 2.7457, "step": 20947 }, { "epoch": 1.3003910857284748, "grad_norm": 0.5832760447799471, "learning_rate": 6.978372095307692e-05, "loss": 2.7629, "step": 20948 }, { "epoch": 1.3004531628282328, "grad_norm": 0.582892427256331, "learning_rate": 6.978040411070879e-05, "loss": 2.8731, "step": 20949 }, { "epoch": 1.3005152399279907, "grad_norm": 0.6734731872732445, "learning_rate": 6.977708716514146e-05, "loss": 2.8338, "step": 20950 }, { "epoch": 1.3005773170277486, "grad_norm": 0.4769364593868428, "learning_rate": 6.977377011639227e-05, "loss": 2.7801, "step": 20951 }, { "epoch": 1.3006393941275063, "grad_norm": 0.6944043089606595, "learning_rate": 6.977045296447851e-05, "loss": 2.8525, "step": 20952 }, { "epoch": 1.3007014712272642, "grad_norm": 0.5802131264543964, "learning_rate": 6.976713570941752e-05, "loss": 2.8678, "step": 20953 }, { "epoch": 1.3007635483270221, "grad_norm": 0.5677701117356156, "learning_rate": 6.976381835122658e-05, "loss": 2.757, "step": 20954 }, { "epoch": 1.30082562542678, "grad_norm": 0.620058288180021, "learning_rate": 6.976050088992299e-05, "loss": 2.8347, "step": 20955 }, { "epoch": 1.300887702526538, "grad_norm": 0.6452555661336441, "learning_rate": 6.975718332552408e-05, "loss": 2.7862, "step": 20956 }, { "epoch": 1.3009497796262959, "grad_norm": 0.5515571702045848, "learning_rate": 6.975386565804714e-05, "loss": 2.7073, "step": 20957 }, { "epoch": 1.3010118567260538, "grad_norm": 0.6145264442967864, "learning_rate": 6.97505478875095e-05, "loss": 2.7418, "step": 20958 }, { "epoch": 1.3010739338258117, "grad_norm": 0.578280505324438, "learning_rate": 6.974723001392844e-05, "loss": 2.7598, "step": 20959 }, { "epoch": 1.3011360109255696, "grad_norm": 0.5873196132172728, "learning_rate": 6.97439120373213e-05, "loss": 2.8106, "step": 20960 }, { "epoch": 1.3011980880253273, "grad_norm": 0.5783253915261604, "learning_rate": 6.974059395770537e-05, "loss": 2.8103, "step": 20961 }, { "epoch": 1.3012601651250852, "grad_norm": 0.5625674684116755, "learning_rate": 6.973727577509799e-05, "loss": 2.7796, "step": 20962 }, { "epoch": 1.3013222422248432, "grad_norm": 0.5746562355170532, "learning_rate": 6.973395748951643e-05, "loss": 2.7589, "step": 20963 }, { "epoch": 1.301384319324601, "grad_norm": 0.5497990772238877, "learning_rate": 6.973063910097803e-05, "loss": 2.8031, "step": 20964 }, { "epoch": 1.301446396424359, "grad_norm": 0.5437160262094382, "learning_rate": 6.97273206095001e-05, "loss": 2.8097, "step": 20965 }, { "epoch": 1.301508473524117, "grad_norm": 0.6260828889000419, "learning_rate": 6.972400201509993e-05, "loss": 2.8399, "step": 20966 }, { "epoch": 1.3015705506238748, "grad_norm": 0.5442199835378603, "learning_rate": 6.972068331779487e-05, "loss": 2.8483, "step": 20967 }, { "epoch": 1.3016326277236328, "grad_norm": 0.6036748336614828, "learning_rate": 6.971736451760222e-05, "loss": 2.8109, "step": 20968 }, { "epoch": 1.3016947048233907, "grad_norm": 0.5585395680005966, "learning_rate": 6.971404561453929e-05, "loss": 2.7234, "step": 20969 }, { "epoch": 1.3017567819231486, "grad_norm": 0.48523515805949147, "learning_rate": 6.971072660862339e-05, "loss": 2.8539, "step": 20970 }, { "epoch": 1.3018188590229065, "grad_norm": 0.5747833849439459, "learning_rate": 6.970740749987186e-05, "loss": 2.7401, "step": 20971 }, { "epoch": 1.3018809361226644, "grad_norm": 0.5273555047879711, "learning_rate": 6.970408828830197e-05, "loss": 2.8666, "step": 20972 }, { "epoch": 1.3019430132224223, "grad_norm": 0.5929585753762856, "learning_rate": 6.97007689739311e-05, "loss": 2.9413, "step": 20973 }, { "epoch": 1.3020050903221803, "grad_norm": 0.504121896894549, "learning_rate": 6.969744955677652e-05, "loss": 2.8072, "step": 20974 }, { "epoch": 1.3020671674219382, "grad_norm": 0.5552200125300265, "learning_rate": 6.969413003685557e-05, "loss": 2.8512, "step": 20975 }, { "epoch": 1.3021292445216959, "grad_norm": 0.4773598082942731, "learning_rate": 6.969081041418555e-05, "loss": 2.7936, "step": 20976 }, { "epoch": 1.3021913216214538, "grad_norm": 0.5036016031257164, "learning_rate": 6.968749068878381e-05, "loss": 2.8422, "step": 20977 }, { "epoch": 1.3022533987212117, "grad_norm": 0.464835591605263, "learning_rate": 6.968417086066765e-05, "loss": 2.73, "step": 20978 }, { "epoch": 1.3023154758209696, "grad_norm": 0.569991818503172, "learning_rate": 6.968085092985438e-05, "loss": 2.8335, "step": 20979 }, { "epoch": 1.3023775529207275, "grad_norm": 0.5342569243054218, "learning_rate": 6.967753089636134e-05, "loss": 2.7549, "step": 20980 }, { "epoch": 1.3024396300204855, "grad_norm": 0.5518403658254118, "learning_rate": 6.967421076020586e-05, "loss": 2.832, "step": 20981 }, { "epoch": 1.3025017071202434, "grad_norm": 0.5595832543599103, "learning_rate": 6.967089052140522e-05, "loss": 2.8698, "step": 20982 }, { "epoch": 1.3025637842200013, "grad_norm": 0.631513887900549, "learning_rate": 6.966757017997678e-05, "loss": 2.8323, "step": 20983 }, { "epoch": 1.3026258613197592, "grad_norm": 0.558370825827705, "learning_rate": 6.966424973593786e-05, "loss": 2.8022, "step": 20984 }, { "epoch": 1.302687938419517, "grad_norm": 0.5761405807013827, "learning_rate": 6.966092918930577e-05, "loss": 2.7254, "step": 20985 }, { "epoch": 1.3027500155192748, "grad_norm": 0.579942167128275, "learning_rate": 6.965760854009783e-05, "loss": 2.9314, "step": 20986 }, { "epoch": 1.3028120926190327, "grad_norm": 0.6455600273032525, "learning_rate": 6.965428778833138e-05, "loss": 2.8183, "step": 20987 }, { "epoch": 1.3028741697187907, "grad_norm": 0.5801911934239146, "learning_rate": 6.965096693402375e-05, "loss": 2.8284, "step": 20988 }, { "epoch": 1.3029362468185486, "grad_norm": 0.6050581986044197, "learning_rate": 6.964764597719223e-05, "loss": 2.8971, "step": 20989 }, { "epoch": 1.3029983239183065, "grad_norm": 0.6363191111039195, "learning_rate": 6.96443249178542e-05, "loss": 2.864, "step": 20990 }, { "epoch": 1.3030604010180644, "grad_norm": 0.48950895461404875, "learning_rate": 6.964100375602694e-05, "loss": 2.7902, "step": 20991 }, { "epoch": 1.3031224781178223, "grad_norm": 0.6140859148060378, "learning_rate": 6.963768249172779e-05, "loss": 2.8381, "step": 20992 }, { "epoch": 1.3031845552175803, "grad_norm": 0.523954762193203, "learning_rate": 6.96343611249741e-05, "loss": 2.8143, "step": 20993 }, { "epoch": 1.3032466323173382, "grad_norm": 0.5922562984471403, "learning_rate": 6.963103965578317e-05, "loss": 2.8351, "step": 20994 }, { "epoch": 1.303308709417096, "grad_norm": 0.5778230445605257, "learning_rate": 6.962771808417234e-05, "loss": 2.8364, "step": 20995 }, { "epoch": 1.303370786516854, "grad_norm": 0.5303047971778598, "learning_rate": 6.962439641015895e-05, "loss": 2.8129, "step": 20996 }, { "epoch": 1.303432863616612, "grad_norm": 0.5606071891634671, "learning_rate": 6.962107463376032e-05, "loss": 2.8666, "step": 20997 }, { "epoch": 1.3034949407163698, "grad_norm": 0.5198132316177831, "learning_rate": 6.961775275499378e-05, "loss": 2.81, "step": 20998 }, { "epoch": 1.3035570178161278, "grad_norm": 0.49409309731598927, "learning_rate": 6.961443077387665e-05, "loss": 2.8948, "step": 20999 }, { "epoch": 1.3036190949158855, "grad_norm": 0.5464163355432419, "learning_rate": 6.961110869042628e-05, "loss": 2.8099, "step": 21000 }, { "epoch": 1.3036811720156434, "grad_norm": 0.6160572180678778, "learning_rate": 6.960778650466e-05, "loss": 2.8585, "step": 21001 }, { "epoch": 1.3037432491154013, "grad_norm": 0.5724756682110294, "learning_rate": 6.960446421659513e-05, "loss": 2.8758, "step": 21002 }, { "epoch": 1.3038053262151592, "grad_norm": 0.5067592261443243, "learning_rate": 6.960114182624903e-05, "loss": 2.829, "step": 21003 }, { "epoch": 1.3038674033149171, "grad_norm": 0.5469172597633241, "learning_rate": 6.959781933363898e-05, "loss": 2.78, "step": 21004 }, { "epoch": 1.303929480414675, "grad_norm": 0.590101249016476, "learning_rate": 6.959449673878238e-05, "loss": 2.7808, "step": 21005 }, { "epoch": 1.303991557514433, "grad_norm": 0.5918328806181522, "learning_rate": 6.959117404169652e-05, "loss": 2.8684, "step": 21006 }, { "epoch": 1.3040536346141909, "grad_norm": 0.6166962950479364, "learning_rate": 6.958785124239875e-05, "loss": 2.8607, "step": 21007 }, { "epoch": 1.3041157117139488, "grad_norm": 0.5997561296926546, "learning_rate": 6.95845283409064e-05, "loss": 2.7884, "step": 21008 }, { "epoch": 1.3041777888137065, "grad_norm": 0.559830476953265, "learning_rate": 6.958120533723682e-05, "loss": 2.9067, "step": 21009 }, { "epoch": 1.3042398659134644, "grad_norm": 0.5565666551129643, "learning_rate": 6.957788223140734e-05, "loss": 2.8471, "step": 21010 }, { "epoch": 1.3043019430132223, "grad_norm": 0.5323934590851411, "learning_rate": 6.957455902343528e-05, "loss": 2.8014, "step": 21011 }, { "epoch": 1.3043640201129802, "grad_norm": 0.5597233854170777, "learning_rate": 6.957123571333801e-05, "loss": 3.0045, "step": 21012 }, { "epoch": 1.3044260972127382, "grad_norm": 0.5931850054086264, "learning_rate": 6.956791230113284e-05, "loss": 2.8916, "step": 21013 }, { "epoch": 1.304488174312496, "grad_norm": 0.49288293360575464, "learning_rate": 6.956458878683714e-05, "loss": 2.8071, "step": 21014 }, { "epoch": 1.304550251412254, "grad_norm": 0.5584192336840884, "learning_rate": 6.956126517046822e-05, "loss": 2.9003, "step": 21015 }, { "epoch": 1.304612328512012, "grad_norm": 0.5191894783817178, "learning_rate": 6.955794145204342e-05, "loss": 2.8141, "step": 21016 }, { "epoch": 1.3046744056117698, "grad_norm": 0.5103249532344593, "learning_rate": 6.955461763158011e-05, "loss": 2.814, "step": 21017 }, { "epoch": 1.3047364827115278, "grad_norm": 0.5689740871351555, "learning_rate": 6.955129370909562e-05, "loss": 2.8047, "step": 21018 }, { "epoch": 1.3047985598112857, "grad_norm": 0.5262796191128011, "learning_rate": 6.954796968460726e-05, "loss": 2.834, "step": 21019 }, { "epoch": 1.3048606369110436, "grad_norm": 0.5923297358983068, "learning_rate": 6.95446455581324e-05, "loss": 2.9642, "step": 21020 }, { "epoch": 1.3049227140108015, "grad_norm": 0.5930670524805959, "learning_rate": 6.954132132968838e-05, "loss": 2.8437, "step": 21021 }, { "epoch": 1.3049847911105594, "grad_norm": 0.5729467875336579, "learning_rate": 6.953799699929255e-05, "loss": 2.7635, "step": 21022 }, { "epoch": 1.3050468682103173, "grad_norm": 0.5440826516579761, "learning_rate": 6.953467256696225e-05, "loss": 2.8754, "step": 21023 }, { "epoch": 1.305108945310075, "grad_norm": 0.6479520704239716, "learning_rate": 6.95313480327148e-05, "loss": 2.8652, "step": 21024 }, { "epoch": 1.305171022409833, "grad_norm": 0.47129916999560884, "learning_rate": 6.95280233965676e-05, "loss": 2.7957, "step": 21025 }, { "epoch": 1.3052330995095909, "grad_norm": 0.6055299543394536, "learning_rate": 6.952469865853794e-05, "loss": 2.8489, "step": 21026 }, { "epoch": 1.3052951766093488, "grad_norm": 0.5262052325781287, "learning_rate": 6.952137381864318e-05, "loss": 2.8274, "step": 21027 }, { "epoch": 1.3053572537091067, "grad_norm": 0.5595620306177727, "learning_rate": 6.95180488769007e-05, "loss": 2.7975, "step": 21028 }, { "epoch": 1.3054193308088646, "grad_norm": 0.5513404009991425, "learning_rate": 6.95147238333278e-05, "loss": 2.8516, "step": 21029 }, { "epoch": 1.3054814079086225, "grad_norm": 0.5188079617762169, "learning_rate": 6.951139868794185e-05, "loss": 2.7314, "step": 21030 }, { "epoch": 1.3055434850083805, "grad_norm": 0.5793004586950719, "learning_rate": 6.95080734407602e-05, "loss": 2.7922, "step": 21031 }, { "epoch": 1.3056055621081384, "grad_norm": 0.5919175229613054, "learning_rate": 6.95047480918002e-05, "loss": 2.7593, "step": 21032 }, { "epoch": 1.305667639207896, "grad_norm": 0.6208462731935163, "learning_rate": 6.950142264107918e-05, "loss": 2.8025, "step": 21033 }, { "epoch": 1.305729716307654, "grad_norm": 0.5945951820918058, "learning_rate": 6.949809708861452e-05, "loss": 2.9434, "step": 21034 }, { "epoch": 1.305791793407412, "grad_norm": 0.6560172394822018, "learning_rate": 6.949477143442354e-05, "loss": 2.8494, "step": 21035 }, { "epoch": 1.3058538705071698, "grad_norm": 0.5604244989133818, "learning_rate": 6.949144567852363e-05, "loss": 2.8788, "step": 21036 }, { "epoch": 1.3059159476069278, "grad_norm": 0.6218983458924366, "learning_rate": 6.948811982093209e-05, "loss": 2.7761, "step": 21037 }, { "epoch": 1.3059780247066857, "grad_norm": 0.5742164923175882, "learning_rate": 6.948479386166629e-05, "loss": 2.803, "step": 21038 }, { "epoch": 1.3060401018064436, "grad_norm": 0.5810718437675327, "learning_rate": 6.948146780074363e-05, "loss": 2.7871, "step": 21039 }, { "epoch": 1.3061021789062015, "grad_norm": 0.5703879528560414, "learning_rate": 6.947814163818138e-05, "loss": 2.8792, "step": 21040 }, { "epoch": 1.3061642560059594, "grad_norm": 0.6114355852053616, "learning_rate": 6.947481537399696e-05, "loss": 2.851, "step": 21041 }, { "epoch": 1.3062263331057173, "grad_norm": 0.593988646682939, "learning_rate": 6.94714890082077e-05, "loss": 2.8483, "step": 21042 }, { "epoch": 1.3062884102054753, "grad_norm": 0.5844798203159127, "learning_rate": 6.946816254083095e-05, "loss": 2.8086, "step": 21043 }, { "epoch": 1.3063504873052332, "grad_norm": 0.5925700107287978, "learning_rate": 6.94648359718841e-05, "loss": 2.8295, "step": 21044 }, { "epoch": 1.306412564404991, "grad_norm": 0.55762095541942, "learning_rate": 6.946150930138442e-05, "loss": 2.7373, "step": 21045 }, { "epoch": 1.306474641504749, "grad_norm": 0.5672628305395508, "learning_rate": 6.945818252934935e-05, "loss": 2.6927, "step": 21046 }, { "epoch": 1.306536718604507, "grad_norm": 0.6105582657799093, "learning_rate": 6.945485565579622e-05, "loss": 2.8366, "step": 21047 }, { "epoch": 1.3065987957042646, "grad_norm": 0.5703854448882408, "learning_rate": 6.945152868074238e-05, "loss": 2.845, "step": 21048 }, { "epoch": 1.3066608728040225, "grad_norm": 0.5936802019955186, "learning_rate": 6.94482016042052e-05, "loss": 2.8278, "step": 21049 }, { "epoch": 1.3067229499037805, "grad_norm": 0.517007361903799, "learning_rate": 6.944487442620203e-05, "loss": 2.8062, "step": 21050 }, { "epoch": 1.3067850270035384, "grad_norm": 0.5518962045140572, "learning_rate": 6.944154714675022e-05, "loss": 2.8477, "step": 21051 }, { "epoch": 1.3068471041032963, "grad_norm": 0.5100052953426485, "learning_rate": 6.943821976586715e-05, "loss": 2.8314, "step": 21052 }, { "epoch": 1.3069091812030542, "grad_norm": 0.5698357248329543, "learning_rate": 6.943489228357016e-05, "loss": 2.8748, "step": 21053 }, { "epoch": 1.3069712583028121, "grad_norm": 0.5254757893680934, "learning_rate": 6.943156469987661e-05, "loss": 2.8263, "step": 21054 }, { "epoch": 1.30703333540257, "grad_norm": 0.5919786180472082, "learning_rate": 6.94282370148039e-05, "loss": 2.8056, "step": 21055 }, { "epoch": 1.307095412502328, "grad_norm": 0.5099506846127373, "learning_rate": 6.942490922836933e-05, "loss": 2.8056, "step": 21056 }, { "epoch": 1.3071574896020857, "grad_norm": 0.5564337362663239, "learning_rate": 6.942158134059032e-05, "loss": 2.8361, "step": 21057 }, { "epoch": 1.3072195667018436, "grad_norm": 0.5137938728069635, "learning_rate": 6.941825335148419e-05, "loss": 2.7725, "step": 21058 }, { "epoch": 1.3072816438016015, "grad_norm": 0.5260531421076968, "learning_rate": 6.941492526106832e-05, "loss": 2.7962, "step": 21059 }, { "epoch": 1.3073437209013594, "grad_norm": 0.4902463508754159, "learning_rate": 6.941159706936006e-05, "loss": 2.778, "step": 21060 }, { "epoch": 1.3074057980011173, "grad_norm": 0.5620631004728174, "learning_rate": 6.94082687763768e-05, "loss": 2.8133, "step": 21061 }, { "epoch": 1.3074678751008753, "grad_norm": 0.576057370979493, "learning_rate": 6.940494038213589e-05, "loss": 2.7464, "step": 21062 }, { "epoch": 1.3075299522006332, "grad_norm": 0.5834264737657008, "learning_rate": 6.94016118866547e-05, "loss": 2.7987, "step": 21063 }, { "epoch": 1.307592029300391, "grad_norm": 0.5290435522644292, "learning_rate": 6.93982832899506e-05, "loss": 2.7616, "step": 21064 }, { "epoch": 1.307654106400149, "grad_norm": 0.6124361131397206, "learning_rate": 6.939495459204092e-05, "loss": 2.7884, "step": 21065 }, { "epoch": 1.307716183499907, "grad_norm": 0.5301343197928684, "learning_rate": 6.939162579294307e-05, "loss": 2.88, "step": 21066 }, { "epoch": 1.3077782605996648, "grad_norm": 0.51424413793966, "learning_rate": 6.938829689267439e-05, "loss": 2.8253, "step": 21067 }, { "epoch": 1.3078403376994228, "grad_norm": 0.5392583132046893, "learning_rate": 6.938496789125227e-05, "loss": 2.7713, "step": 21068 }, { "epoch": 1.3079024147991807, "grad_norm": 0.5117720947415723, "learning_rate": 6.938163878869405e-05, "loss": 2.8971, "step": 21069 }, { "epoch": 1.3079644918989386, "grad_norm": 0.517766464802326, "learning_rate": 6.937830958501713e-05, "loss": 2.851, "step": 21070 }, { "epoch": 1.3080265689986965, "grad_norm": 0.5831835304642724, "learning_rate": 6.937498028023885e-05, "loss": 2.8808, "step": 21071 }, { "epoch": 1.3080886460984542, "grad_norm": 0.5295007179442699, "learning_rate": 6.937165087437661e-05, "loss": 2.8474, "step": 21072 }, { "epoch": 1.3081507231982121, "grad_norm": 0.5575858940790147, "learning_rate": 6.936832136744777e-05, "loss": 2.8088, "step": 21073 }, { "epoch": 1.30821280029797, "grad_norm": 0.5556587527378393, "learning_rate": 6.936499175946967e-05, "loss": 2.8572, "step": 21074 }, { "epoch": 1.308274877397728, "grad_norm": 0.6316090195485925, "learning_rate": 6.936166205045973e-05, "loss": 2.8728, "step": 21075 }, { "epoch": 1.3083369544974859, "grad_norm": 0.49821221094339146, "learning_rate": 6.935833224043529e-05, "loss": 2.8881, "step": 21076 }, { "epoch": 1.3083990315972438, "grad_norm": 0.5511344156168266, "learning_rate": 6.935500232941374e-05, "loss": 2.8373, "step": 21077 }, { "epoch": 1.3084611086970017, "grad_norm": 0.5481759265395136, "learning_rate": 6.935167231741242e-05, "loss": 2.7707, "step": 21078 }, { "epoch": 1.3085231857967596, "grad_norm": 0.5512495011913439, "learning_rate": 6.934834220444875e-05, "loss": 2.8336, "step": 21079 }, { "epoch": 1.3085852628965176, "grad_norm": 0.5711773624607901, "learning_rate": 6.934501199054009e-05, "loss": 2.9198, "step": 21080 }, { "epoch": 1.3086473399962752, "grad_norm": 0.5631828931205815, "learning_rate": 6.934168167570378e-05, "loss": 2.7885, "step": 21081 }, { "epoch": 1.3087094170960332, "grad_norm": 0.6178549525523719, "learning_rate": 6.933835125995724e-05, "loss": 2.7939, "step": 21082 }, { "epoch": 1.308771494195791, "grad_norm": 0.5118480256999766, "learning_rate": 6.933502074331782e-05, "loss": 2.8219, "step": 21083 }, { "epoch": 1.308833571295549, "grad_norm": 0.567809163043347, "learning_rate": 6.93316901258029e-05, "loss": 2.8074, "step": 21084 }, { "epoch": 1.308895648395307, "grad_norm": 0.5499538131307684, "learning_rate": 6.932835940742987e-05, "loss": 2.8406, "step": 21085 }, { "epoch": 1.3089577254950648, "grad_norm": 0.509199507736903, "learning_rate": 6.932502858821608e-05, "loss": 2.7656, "step": 21086 }, { "epoch": 1.3090198025948228, "grad_norm": 0.5570701414337655, "learning_rate": 6.932169766817895e-05, "loss": 2.802, "step": 21087 }, { "epoch": 1.3090818796945807, "grad_norm": 0.5267821535816806, "learning_rate": 6.93183666473358e-05, "loss": 2.8957, "step": 21088 }, { "epoch": 1.3091439567943386, "grad_norm": 0.57836061264456, "learning_rate": 6.931503552570407e-05, "loss": 2.8315, "step": 21089 }, { "epoch": 1.3092060338940965, "grad_norm": 0.5529904261100039, "learning_rate": 6.931170430330109e-05, "loss": 2.8619, "step": 21090 }, { "epoch": 1.3092681109938544, "grad_norm": 0.5920023797396564, "learning_rate": 6.930837298014428e-05, "loss": 2.8942, "step": 21091 }, { "epoch": 1.3093301880936123, "grad_norm": 0.5491612465330012, "learning_rate": 6.930504155625098e-05, "loss": 2.7407, "step": 21092 }, { "epoch": 1.3093922651933703, "grad_norm": 0.5964850125312521, "learning_rate": 6.93017100316386e-05, "loss": 2.8565, "step": 21093 }, { "epoch": 1.3094543422931282, "grad_norm": 0.570884652742953, "learning_rate": 6.929837840632453e-05, "loss": 2.8319, "step": 21094 }, { "epoch": 1.309516419392886, "grad_norm": 0.5776610445912709, "learning_rate": 6.929504668032613e-05, "loss": 2.7809, "step": 21095 }, { "epoch": 1.3095784964926438, "grad_norm": 0.5106641662406131, "learning_rate": 6.929171485366077e-05, "loss": 2.7716, "step": 21096 }, { "epoch": 1.3096405735924017, "grad_norm": 0.5160413852588727, "learning_rate": 6.928838292634586e-05, "loss": 2.815, "step": 21097 }, { "epoch": 1.3097026506921596, "grad_norm": 0.6038423166323653, "learning_rate": 6.928505089839879e-05, "loss": 2.9141, "step": 21098 }, { "epoch": 1.3097647277919175, "grad_norm": 0.5829868792125596, "learning_rate": 6.928171876983689e-05, "loss": 2.8315, "step": 21099 }, { "epoch": 1.3098268048916755, "grad_norm": 0.551967371752023, "learning_rate": 6.927838654067761e-05, "loss": 2.8033, "step": 21100 }, { "epoch": 1.3098888819914334, "grad_norm": 0.6304312986763052, "learning_rate": 6.92750542109383e-05, "loss": 2.8233, "step": 21101 }, { "epoch": 1.3099509590911913, "grad_norm": 0.5713956293334812, "learning_rate": 6.927172178063635e-05, "loss": 2.7798, "step": 21102 }, { "epoch": 1.3100130361909492, "grad_norm": 0.6258670038528059, "learning_rate": 6.926838924978914e-05, "loss": 2.7998, "step": 21103 }, { "epoch": 1.3100751132907071, "grad_norm": 0.6760777831655167, "learning_rate": 6.92650566184141e-05, "loss": 2.9229, "step": 21104 }, { "epoch": 1.3101371903904648, "grad_norm": 0.5441782536473044, "learning_rate": 6.926172388652855e-05, "loss": 2.845, "step": 21105 }, { "epoch": 1.3101992674902228, "grad_norm": 0.6308075732998094, "learning_rate": 6.925839105414992e-05, "loss": 2.8424, "step": 21106 }, { "epoch": 1.3102613445899807, "grad_norm": 0.5127812486107239, "learning_rate": 6.92550581212956e-05, "loss": 2.7752, "step": 21107 }, { "epoch": 1.3103234216897386, "grad_norm": 0.6527134881139209, "learning_rate": 6.925172508798295e-05, "loss": 2.7929, "step": 21108 }, { "epoch": 1.3103854987894965, "grad_norm": 0.551694748236749, "learning_rate": 6.924839195422938e-05, "loss": 2.8646, "step": 21109 }, { "epoch": 1.3104475758892544, "grad_norm": 0.6106240846671952, "learning_rate": 6.924505872005229e-05, "loss": 2.8635, "step": 21110 }, { "epoch": 1.3105096529890123, "grad_norm": 0.5732742638524876, "learning_rate": 6.924172538546904e-05, "loss": 2.8263, "step": 21111 }, { "epoch": 1.3105717300887703, "grad_norm": 0.6331018798764063, "learning_rate": 6.923839195049705e-05, "loss": 2.8376, "step": 21112 }, { "epoch": 1.3106338071885282, "grad_norm": 0.5711371325754814, "learning_rate": 6.923505841515369e-05, "loss": 2.8778, "step": 21113 }, { "epoch": 1.310695884288286, "grad_norm": 0.5425519921838987, "learning_rate": 6.923172477945637e-05, "loss": 2.8547, "step": 21114 }, { "epoch": 1.310757961388044, "grad_norm": 0.6075170019305921, "learning_rate": 6.922839104342246e-05, "loss": 2.8516, "step": 21115 }, { "epoch": 1.310820038487802, "grad_norm": 0.6260878394095184, "learning_rate": 6.922505720706939e-05, "loss": 2.8781, "step": 21116 }, { "epoch": 1.3108821155875598, "grad_norm": 0.6189976957429264, "learning_rate": 6.922172327041451e-05, "loss": 2.805, "step": 21117 }, { "epoch": 1.3109441926873178, "grad_norm": 0.5402839202117643, "learning_rate": 6.921838923347524e-05, "loss": 2.7953, "step": 21118 }, { "epoch": 1.3110062697870757, "grad_norm": 0.6231928448105932, "learning_rate": 6.921505509626896e-05, "loss": 2.8741, "step": 21119 }, { "epoch": 1.3110683468868334, "grad_norm": 0.5636202200989144, "learning_rate": 6.921172085881308e-05, "loss": 2.897, "step": 21120 }, { "epoch": 1.3111304239865913, "grad_norm": 0.6350181194784031, "learning_rate": 6.920838652112498e-05, "loss": 2.8352, "step": 21121 }, { "epoch": 1.3111925010863492, "grad_norm": 0.5720005137384548, "learning_rate": 6.920505208322208e-05, "loss": 2.8303, "step": 21122 }, { "epoch": 1.3112545781861071, "grad_norm": 0.5910449649161641, "learning_rate": 6.920171754512175e-05, "loss": 2.8203, "step": 21123 }, { "epoch": 1.311316655285865, "grad_norm": 0.6033111769162683, "learning_rate": 6.919838290684139e-05, "loss": 2.8459, "step": 21124 }, { "epoch": 1.311378732385623, "grad_norm": 0.5268134382309492, "learning_rate": 6.919504816839843e-05, "loss": 2.7227, "step": 21125 }, { "epoch": 1.3114408094853809, "grad_norm": 0.5911009823313569, "learning_rate": 6.919171332981023e-05, "loss": 2.8752, "step": 21126 }, { "epoch": 1.3115028865851388, "grad_norm": 0.6235110189355921, "learning_rate": 6.91883783910942e-05, "loss": 2.7632, "step": 21127 }, { "epoch": 1.3115649636848967, "grad_norm": 0.5017661551691089, "learning_rate": 6.918504335226775e-05, "loss": 2.8807, "step": 21128 }, { "epoch": 1.3116270407846544, "grad_norm": 0.6959479996052332, "learning_rate": 6.918170821334826e-05, "loss": 2.9382, "step": 21129 }, { "epoch": 1.3116891178844123, "grad_norm": 0.5669325652329845, "learning_rate": 6.917837297435315e-05, "loss": 2.845, "step": 21130 }, { "epoch": 1.3117511949841703, "grad_norm": 0.6442457549370147, "learning_rate": 6.917503763529981e-05, "loss": 2.8638, "step": 21131 }, { "epoch": 1.3118132720839282, "grad_norm": 0.6553857425694728, "learning_rate": 6.917170219620565e-05, "loss": 2.8629, "step": 21132 }, { "epoch": 1.311875349183686, "grad_norm": 0.5919968421423303, "learning_rate": 6.916836665708806e-05, "loss": 2.8498, "step": 21133 }, { "epoch": 1.311937426283444, "grad_norm": 0.6087771686546188, "learning_rate": 6.916503101796445e-05, "loss": 2.8966, "step": 21134 }, { "epoch": 1.311999503383202, "grad_norm": 0.6189006015256916, "learning_rate": 6.916169527885222e-05, "loss": 2.9105, "step": 21135 }, { "epoch": 1.3120615804829598, "grad_norm": 0.5769444087254042, "learning_rate": 6.915835943976877e-05, "loss": 2.8495, "step": 21136 }, { "epoch": 1.3121236575827178, "grad_norm": 0.6514167823955694, "learning_rate": 6.91550235007315e-05, "loss": 2.8075, "step": 21137 }, { "epoch": 1.3121857346824757, "grad_norm": 0.5904760950080367, "learning_rate": 6.915168746175782e-05, "loss": 2.806, "step": 21138 }, { "epoch": 1.3122478117822336, "grad_norm": 0.6065511073713609, "learning_rate": 6.914835132286515e-05, "loss": 2.887, "step": 21139 }, { "epoch": 1.3123098888819915, "grad_norm": 0.6278805631937535, "learning_rate": 6.914501508407089e-05, "loss": 2.9168, "step": 21140 }, { "epoch": 1.3123719659817494, "grad_norm": 0.5658335717424987, "learning_rate": 6.914167874539244e-05, "loss": 2.7792, "step": 21141 }, { "epoch": 1.3124340430815074, "grad_norm": 0.515411968728902, "learning_rate": 6.913834230684719e-05, "loss": 2.6751, "step": 21142 }, { "epoch": 1.3124961201812653, "grad_norm": 0.5971145668388029, "learning_rate": 6.913500576845258e-05, "loss": 2.7673, "step": 21143 }, { "epoch": 1.312558197281023, "grad_norm": 0.4781971241275795, "learning_rate": 6.913166913022599e-05, "loss": 2.7811, "step": 21144 }, { "epoch": 1.3126202743807809, "grad_norm": 0.5732138787579248, "learning_rate": 6.912833239218484e-05, "loss": 2.8957, "step": 21145 }, { "epoch": 1.3126823514805388, "grad_norm": 0.5036130540102745, "learning_rate": 6.912499555434653e-05, "loss": 2.8738, "step": 21146 }, { "epoch": 1.3127444285802967, "grad_norm": 0.6528069912909494, "learning_rate": 6.912165861672847e-05, "loss": 2.8873, "step": 21147 }, { "epoch": 1.3128065056800546, "grad_norm": 0.5798324424321603, "learning_rate": 6.911832157934808e-05, "loss": 2.9134, "step": 21148 }, { "epoch": 1.3128685827798126, "grad_norm": 0.5926773521313267, "learning_rate": 6.911498444222277e-05, "loss": 2.771, "step": 21149 }, { "epoch": 1.3129306598795705, "grad_norm": 0.5747100648500145, "learning_rate": 6.911164720536994e-05, "loss": 2.7591, "step": 21150 }, { "epoch": 1.3129927369793284, "grad_norm": 0.5904562593033824, "learning_rate": 6.910830986880702e-05, "loss": 2.8709, "step": 21151 }, { "epoch": 1.3130548140790863, "grad_norm": 0.6636431323513268, "learning_rate": 6.91049724325514e-05, "loss": 2.89, "step": 21152 }, { "epoch": 1.313116891178844, "grad_norm": 0.5203337530429295, "learning_rate": 6.91016348966205e-05, "loss": 2.9191, "step": 21153 }, { "epoch": 1.313178968278602, "grad_norm": 0.657558023155251, "learning_rate": 6.909829726103173e-05, "loss": 2.9175, "step": 21154 }, { "epoch": 1.3132410453783598, "grad_norm": 0.6750030446866916, "learning_rate": 6.909495952580251e-05, "loss": 2.813, "step": 21155 }, { "epoch": 1.3133031224781178, "grad_norm": 0.5749248393319686, "learning_rate": 6.909162169095025e-05, "loss": 2.7256, "step": 21156 }, { "epoch": 1.3133651995778757, "grad_norm": 0.6332185878665914, "learning_rate": 6.908828375649237e-05, "loss": 2.8511, "step": 21157 }, { "epoch": 1.3134272766776336, "grad_norm": 0.5866628861531273, "learning_rate": 6.908494572244628e-05, "loss": 2.7958, "step": 21158 }, { "epoch": 1.3134893537773915, "grad_norm": 0.6231082658806854, "learning_rate": 6.90816075888294e-05, "loss": 2.8422, "step": 21159 }, { "epoch": 1.3135514308771494, "grad_norm": 0.6055553005253939, "learning_rate": 6.907826935565912e-05, "loss": 2.9036, "step": 21160 }, { "epoch": 1.3136135079769073, "grad_norm": 0.6295881188204665, "learning_rate": 6.90749310229529e-05, "loss": 2.8649, "step": 21161 }, { "epoch": 1.3136755850766653, "grad_norm": 0.5865414113972339, "learning_rate": 6.907159259072812e-05, "loss": 2.8667, "step": 21162 }, { "epoch": 1.3137376621764232, "grad_norm": 0.606657080295124, "learning_rate": 6.906825405900221e-05, "loss": 2.8419, "step": 21163 }, { "epoch": 1.313799739276181, "grad_norm": 0.5729835615696097, "learning_rate": 6.90649154277926e-05, "loss": 2.7799, "step": 21164 }, { "epoch": 1.313861816375939, "grad_norm": 0.5780286193101449, "learning_rate": 6.90615766971167e-05, "loss": 2.7961, "step": 21165 }, { "epoch": 1.313923893475697, "grad_norm": 0.6174129846785484, "learning_rate": 6.905823786699192e-05, "loss": 2.8306, "step": 21166 }, { "epoch": 1.3139859705754546, "grad_norm": 0.6087372939881084, "learning_rate": 6.905489893743569e-05, "loss": 2.829, "step": 21167 }, { "epoch": 1.3140480476752125, "grad_norm": 0.581991258309773, "learning_rate": 6.905155990846542e-05, "loss": 2.8297, "step": 21168 }, { "epoch": 1.3141101247749705, "grad_norm": 0.6143506681915347, "learning_rate": 6.904822078009854e-05, "loss": 2.8356, "step": 21169 }, { "epoch": 1.3141722018747284, "grad_norm": 0.5827248044550732, "learning_rate": 6.904488155235246e-05, "loss": 2.7488, "step": 21170 }, { "epoch": 1.3142342789744863, "grad_norm": 0.5262966073298855, "learning_rate": 6.904154222524462e-05, "loss": 2.7995, "step": 21171 }, { "epoch": 1.3142963560742442, "grad_norm": 0.563120711434109, "learning_rate": 6.903820279879243e-05, "loss": 2.7196, "step": 21172 }, { "epoch": 1.3143584331740021, "grad_norm": 0.5356387448411248, "learning_rate": 6.903486327301331e-05, "loss": 2.8286, "step": 21173 }, { "epoch": 1.31442051027376, "grad_norm": 0.5349760657152935, "learning_rate": 6.90315236479247e-05, "loss": 2.7122, "step": 21174 }, { "epoch": 1.314482587373518, "grad_norm": 0.4855730867561169, "learning_rate": 6.9028183923544e-05, "loss": 2.7605, "step": 21175 }, { "epoch": 1.3145446644732757, "grad_norm": 0.5585808652445492, "learning_rate": 6.902484409988865e-05, "loss": 2.8189, "step": 21176 }, { "epoch": 1.3146067415730336, "grad_norm": 0.5377562632796739, "learning_rate": 6.902150417697608e-05, "loss": 2.7955, "step": 21177 }, { "epoch": 1.3146688186727915, "grad_norm": 0.5169900107977411, "learning_rate": 6.901816415482368e-05, "loss": 2.846, "step": 21178 }, { "epoch": 1.3147308957725494, "grad_norm": 0.49451049712932044, "learning_rate": 6.901482403344893e-05, "loss": 2.798, "step": 21179 }, { "epoch": 1.3147929728723073, "grad_norm": 0.5863234965162682, "learning_rate": 6.901148381286921e-05, "loss": 2.8496, "step": 21180 }, { "epoch": 1.3148550499720653, "grad_norm": 0.48098770276169983, "learning_rate": 6.900814349310196e-05, "loss": 2.7513, "step": 21181 }, { "epoch": 1.3149171270718232, "grad_norm": 0.6223770894270081, "learning_rate": 6.900480307416461e-05, "loss": 2.8212, "step": 21182 }, { "epoch": 1.314979204171581, "grad_norm": 0.4917607652468116, "learning_rate": 6.90014625560746e-05, "loss": 2.7353, "step": 21183 }, { "epoch": 1.315041281271339, "grad_norm": 0.6001245478664317, "learning_rate": 6.899812193884934e-05, "loss": 2.7754, "step": 21184 }, { "epoch": 1.315103358371097, "grad_norm": 0.5466171474193707, "learning_rate": 6.899478122250627e-05, "loss": 2.817, "step": 21185 }, { "epoch": 1.3151654354708548, "grad_norm": 0.557636213896614, "learning_rate": 6.899144040706282e-05, "loss": 2.8512, "step": 21186 }, { "epoch": 1.3152275125706128, "grad_norm": 0.543594356710869, "learning_rate": 6.89880994925364e-05, "loss": 2.7616, "step": 21187 }, { "epoch": 1.3152895896703707, "grad_norm": 0.534661279923914, "learning_rate": 6.898475847894446e-05, "loss": 2.8491, "step": 21188 }, { "epoch": 1.3153516667701286, "grad_norm": 0.54303543449757, "learning_rate": 6.898141736630442e-05, "loss": 2.9061, "step": 21189 }, { "epoch": 1.3154137438698865, "grad_norm": 0.5289140844398758, "learning_rate": 6.897807615463374e-05, "loss": 2.7686, "step": 21190 }, { "epoch": 1.3154758209696442, "grad_norm": 0.5127144946188361, "learning_rate": 6.89747348439498e-05, "loss": 2.7892, "step": 21191 }, { "epoch": 1.3155378980694021, "grad_norm": 0.5219185268501527, "learning_rate": 6.89713934342701e-05, "loss": 2.8214, "step": 21192 }, { "epoch": 1.31559997516916, "grad_norm": 0.5004574054405261, "learning_rate": 6.8968051925612e-05, "loss": 2.8236, "step": 21193 }, { "epoch": 1.315662052268918, "grad_norm": 0.531368242456384, "learning_rate": 6.896471031799298e-05, "loss": 2.796, "step": 21194 }, { "epoch": 1.3157241293686759, "grad_norm": 0.5555050962765437, "learning_rate": 6.896136861143047e-05, "loss": 2.8232, "step": 21195 }, { "epoch": 1.3157862064684338, "grad_norm": 0.5510200361634784, "learning_rate": 6.89580268059419e-05, "loss": 2.8523, "step": 21196 }, { "epoch": 1.3158482835681917, "grad_norm": 0.5453957852047204, "learning_rate": 6.89546849015447e-05, "loss": 2.8316, "step": 21197 }, { "epoch": 1.3159103606679496, "grad_norm": 0.5274324978219319, "learning_rate": 6.895134289825628e-05, "loss": 2.7464, "step": 21198 }, { "epoch": 1.3159724377677076, "grad_norm": 0.5035710957515432, "learning_rate": 6.894800079609414e-05, "loss": 2.8492, "step": 21199 }, { "epoch": 1.3160345148674653, "grad_norm": 0.4985522081408667, "learning_rate": 6.894465859507566e-05, "loss": 2.7358, "step": 21200 }, { "epoch": 1.3160965919672232, "grad_norm": 0.6135336241224667, "learning_rate": 6.894131629521829e-05, "loss": 2.8404, "step": 21201 }, { "epoch": 1.316158669066981, "grad_norm": 0.5453809492987604, "learning_rate": 6.89379738965395e-05, "loss": 2.7426, "step": 21202 }, { "epoch": 1.316220746166739, "grad_norm": 0.6077278813786208, "learning_rate": 6.893463139905668e-05, "loss": 2.9251, "step": 21203 }, { "epoch": 1.316282823266497, "grad_norm": 0.5756341785192024, "learning_rate": 6.89312888027873e-05, "loss": 2.8648, "step": 21204 }, { "epoch": 1.3163449003662548, "grad_norm": 0.5709752973667902, "learning_rate": 6.89279461077488e-05, "loss": 2.6902, "step": 21205 }, { "epoch": 1.3164069774660128, "grad_norm": 0.5810712539491049, "learning_rate": 6.89246033139586e-05, "loss": 2.7726, "step": 21206 }, { "epoch": 1.3164690545657707, "grad_norm": 0.5581301385828787, "learning_rate": 6.892126042143416e-05, "loss": 2.7986, "step": 21207 }, { "epoch": 1.3165311316655286, "grad_norm": 0.5803508695326242, "learning_rate": 6.891791743019291e-05, "loss": 2.7049, "step": 21208 }, { "epoch": 1.3165932087652865, "grad_norm": 0.6001756708945184, "learning_rate": 6.891457434025228e-05, "loss": 2.7979, "step": 21209 }, { "epoch": 1.3166552858650444, "grad_norm": 0.5830178827026286, "learning_rate": 6.891123115162972e-05, "loss": 2.768, "step": 21210 }, { "epoch": 1.3167173629648024, "grad_norm": 0.6121182439243115, "learning_rate": 6.890788786434268e-05, "loss": 2.7958, "step": 21211 }, { "epoch": 1.3167794400645603, "grad_norm": 0.575642876321595, "learning_rate": 6.890454447840861e-05, "loss": 2.9332, "step": 21212 }, { "epoch": 1.3168415171643182, "grad_norm": 0.5815618017111496, "learning_rate": 6.890120099384494e-05, "loss": 2.8699, "step": 21213 }, { "epoch": 1.316903594264076, "grad_norm": 0.5710155386539112, "learning_rate": 6.889785741066912e-05, "loss": 2.7904, "step": 21214 }, { "epoch": 1.3169656713638338, "grad_norm": 0.607594063937163, "learning_rate": 6.889451372889858e-05, "loss": 2.7992, "step": 21215 }, { "epoch": 1.3170277484635917, "grad_norm": 0.5822975522348381, "learning_rate": 6.889116994855079e-05, "loss": 2.8261, "step": 21216 }, { "epoch": 1.3170898255633496, "grad_norm": 0.5965172629292554, "learning_rate": 6.888782606964314e-05, "loss": 2.7993, "step": 21217 }, { "epoch": 1.3171519026631076, "grad_norm": 0.5923616842595327, "learning_rate": 6.888448209219315e-05, "loss": 2.8114, "step": 21218 }, { "epoch": 1.3172139797628655, "grad_norm": 0.6064961484282013, "learning_rate": 6.888113801621823e-05, "loss": 2.9512, "step": 21219 }, { "epoch": 1.3172760568626234, "grad_norm": 0.6072826900658459, "learning_rate": 6.887779384173582e-05, "loss": 2.8257, "step": 21220 }, { "epoch": 1.3173381339623813, "grad_norm": 0.5579481595699293, "learning_rate": 6.887444956876339e-05, "loss": 2.9137, "step": 21221 }, { "epoch": 1.3174002110621392, "grad_norm": 0.6391464476721751, "learning_rate": 6.887110519731837e-05, "loss": 2.8477, "step": 21222 }, { "epoch": 1.3174622881618971, "grad_norm": 0.6523967424443519, "learning_rate": 6.886776072741821e-05, "loss": 2.9166, "step": 21223 }, { "epoch": 1.3175243652616548, "grad_norm": 0.562457453920068, "learning_rate": 6.886441615908036e-05, "loss": 2.8031, "step": 21224 }, { "epoch": 1.3175864423614128, "grad_norm": 0.6227376523288778, "learning_rate": 6.886107149232228e-05, "loss": 2.8036, "step": 21225 }, { "epoch": 1.3176485194611707, "grad_norm": 0.5399627079714229, "learning_rate": 6.885772672716141e-05, "loss": 2.7593, "step": 21226 }, { "epoch": 1.3177105965609286, "grad_norm": 0.5708475607702367, "learning_rate": 6.88543818636152e-05, "loss": 2.7452, "step": 21227 }, { "epoch": 1.3177726736606865, "grad_norm": 0.5511717798353257, "learning_rate": 6.885103690170109e-05, "loss": 2.8077, "step": 21228 }, { "epoch": 1.3178347507604444, "grad_norm": 0.6047282212614541, "learning_rate": 6.884769184143656e-05, "loss": 2.8394, "step": 21229 }, { "epoch": 1.3178968278602023, "grad_norm": 0.5873811358066218, "learning_rate": 6.884434668283905e-05, "loss": 2.86, "step": 21230 }, { "epoch": 1.3179589049599603, "grad_norm": 0.5887051892488774, "learning_rate": 6.884100142592599e-05, "loss": 2.8103, "step": 21231 }, { "epoch": 1.3180209820597182, "grad_norm": 0.6447862294162471, "learning_rate": 6.883765607071486e-05, "loss": 2.761, "step": 21232 }, { "epoch": 1.318083059159476, "grad_norm": 0.5561407689203757, "learning_rate": 6.883431061722312e-05, "loss": 2.8818, "step": 21233 }, { "epoch": 1.318145136259234, "grad_norm": 0.5625948296191033, "learning_rate": 6.883096506546818e-05, "loss": 2.8843, "step": 21234 }, { "epoch": 1.318207213358992, "grad_norm": 0.5953337477849534, "learning_rate": 6.882761941546755e-05, "loss": 2.9343, "step": 21235 }, { "epoch": 1.3182692904587499, "grad_norm": 0.49398579640857426, "learning_rate": 6.882427366723864e-05, "loss": 2.7549, "step": 21236 }, { "epoch": 1.3183313675585078, "grad_norm": 0.5971067557850807, "learning_rate": 6.882092782079895e-05, "loss": 2.8734, "step": 21237 }, { "epoch": 1.3183934446582657, "grad_norm": 0.5933263923965901, "learning_rate": 6.881758187616586e-05, "loss": 2.8596, "step": 21238 }, { "epoch": 1.3184555217580234, "grad_norm": 0.5860316645933491, "learning_rate": 6.881423583335693e-05, "loss": 2.8083, "step": 21239 }, { "epoch": 1.3185175988577813, "grad_norm": 0.6701116487723892, "learning_rate": 6.881088969238956e-05, "loss": 2.8456, "step": 21240 }, { "epoch": 1.3185796759575392, "grad_norm": 0.5391238910160382, "learning_rate": 6.880754345328119e-05, "loss": 2.7527, "step": 21241 }, { "epoch": 1.3186417530572971, "grad_norm": 0.5650498295221965, "learning_rate": 6.880419711604931e-05, "loss": 2.7857, "step": 21242 }, { "epoch": 1.318703830157055, "grad_norm": 0.5614094759526985, "learning_rate": 6.880085068071136e-05, "loss": 2.8028, "step": 21243 }, { "epoch": 1.318765907256813, "grad_norm": 0.528923099762183, "learning_rate": 6.879750414728482e-05, "loss": 2.8237, "step": 21244 }, { "epoch": 1.318827984356571, "grad_norm": 0.5706145649318297, "learning_rate": 6.879415751578712e-05, "loss": 2.8824, "step": 21245 }, { "epoch": 1.3188900614563288, "grad_norm": 0.5478108436833622, "learning_rate": 6.879081078623575e-05, "loss": 2.7857, "step": 21246 }, { "epoch": 1.3189521385560867, "grad_norm": 0.5449488643402354, "learning_rate": 6.878746395864817e-05, "loss": 2.8664, "step": 21247 }, { "epoch": 1.3190142156558444, "grad_norm": 0.6225452376784238, "learning_rate": 6.878411703304181e-05, "loss": 2.8524, "step": 21248 }, { "epoch": 1.3190762927556023, "grad_norm": 0.5830047709578979, "learning_rate": 6.878077000943416e-05, "loss": 2.8155, "step": 21249 }, { "epoch": 1.3191383698553603, "grad_norm": 0.5214959475486916, "learning_rate": 6.877742288784267e-05, "loss": 2.8112, "step": 21250 }, { "epoch": 1.3192004469551182, "grad_norm": 0.5483413927255086, "learning_rate": 6.877407566828482e-05, "loss": 2.8845, "step": 21251 }, { "epoch": 1.319262524054876, "grad_norm": 0.5406388716489143, "learning_rate": 6.877072835077804e-05, "loss": 2.7852, "step": 21252 }, { "epoch": 1.319324601154634, "grad_norm": 0.5541247078300823, "learning_rate": 6.876738093533983e-05, "loss": 2.8252, "step": 21253 }, { "epoch": 1.319386678254392, "grad_norm": 0.5857015261730666, "learning_rate": 6.876403342198762e-05, "loss": 2.8403, "step": 21254 }, { "epoch": 1.3194487553541498, "grad_norm": 0.5611675428297369, "learning_rate": 6.876068581073891e-05, "loss": 2.8792, "step": 21255 }, { "epoch": 1.3195108324539078, "grad_norm": 0.531215442206661, "learning_rate": 6.875733810161114e-05, "loss": 2.8624, "step": 21256 }, { "epoch": 1.3195729095536657, "grad_norm": 0.5848622437839093, "learning_rate": 6.875399029462179e-05, "loss": 2.8965, "step": 21257 }, { "epoch": 1.3196349866534236, "grad_norm": 0.5604306409528182, "learning_rate": 6.875064238978831e-05, "loss": 2.917, "step": 21258 }, { "epoch": 1.3196970637531815, "grad_norm": 0.554116963067299, "learning_rate": 6.874729438712818e-05, "loss": 2.837, "step": 21259 }, { "epoch": 1.3197591408529394, "grad_norm": 0.657305051019443, "learning_rate": 6.874394628665888e-05, "loss": 2.8199, "step": 21260 }, { "epoch": 1.3198212179526974, "grad_norm": 0.529139280611789, "learning_rate": 6.874059808839784e-05, "loss": 2.8466, "step": 21261 }, { "epoch": 1.3198832950524553, "grad_norm": 0.6022648859904035, "learning_rate": 6.873724979236255e-05, "loss": 2.851, "step": 21262 }, { "epoch": 1.319945372152213, "grad_norm": 0.5986175535164638, "learning_rate": 6.873390139857048e-05, "loss": 2.8864, "step": 21263 }, { "epoch": 1.3200074492519709, "grad_norm": 0.5666035538710401, "learning_rate": 6.873055290703911e-05, "loss": 2.7948, "step": 21264 }, { "epoch": 1.3200695263517288, "grad_norm": 0.5070141605735884, "learning_rate": 6.87272043177859e-05, "loss": 2.8416, "step": 21265 }, { "epoch": 1.3201316034514867, "grad_norm": 0.631080990062826, "learning_rate": 6.872385563082831e-05, "loss": 2.7827, "step": 21266 }, { "epoch": 1.3201936805512446, "grad_norm": 0.4940814408755957, "learning_rate": 6.872050684618382e-05, "loss": 2.815, "step": 21267 }, { "epoch": 1.3202557576510026, "grad_norm": 0.5650847441914395, "learning_rate": 6.87171579638699e-05, "loss": 2.8107, "step": 21268 }, { "epoch": 1.3203178347507605, "grad_norm": 0.5783692694454927, "learning_rate": 6.871380898390403e-05, "loss": 2.9196, "step": 21269 }, { "epoch": 1.3203799118505184, "grad_norm": 0.5627232744233424, "learning_rate": 6.871045990630367e-05, "loss": 2.8494, "step": 21270 }, { "epoch": 1.3204419889502763, "grad_norm": 0.5897843792068855, "learning_rate": 6.87071107310863e-05, "loss": 2.8536, "step": 21271 }, { "epoch": 1.320504066050034, "grad_norm": 0.585381282017569, "learning_rate": 6.87037614582694e-05, "loss": 2.8258, "step": 21272 }, { "epoch": 1.320566143149792, "grad_norm": 0.5741980672212503, "learning_rate": 6.870041208787042e-05, "loss": 2.8375, "step": 21273 }, { "epoch": 1.3206282202495498, "grad_norm": 0.58320593861989, "learning_rate": 6.869706261990686e-05, "loss": 2.7902, "step": 21274 }, { "epoch": 1.3206902973493078, "grad_norm": 0.5975001511513747, "learning_rate": 6.86937130543962e-05, "loss": 2.8735, "step": 21275 }, { "epoch": 1.3207523744490657, "grad_norm": 0.5128080117019352, "learning_rate": 6.869036339135589e-05, "loss": 2.8702, "step": 21276 }, { "epoch": 1.3208144515488236, "grad_norm": 0.5893456332595699, "learning_rate": 6.86870136308034e-05, "loss": 2.8729, "step": 21277 }, { "epoch": 1.3208765286485815, "grad_norm": 0.5300999422580304, "learning_rate": 6.868366377275625e-05, "loss": 2.7497, "step": 21278 }, { "epoch": 1.3209386057483394, "grad_norm": 0.5938808397388565, "learning_rate": 6.868031381723187e-05, "loss": 2.8236, "step": 21279 }, { "epoch": 1.3210006828480974, "grad_norm": 0.6243837417815377, "learning_rate": 6.867696376424777e-05, "loss": 2.9399, "step": 21280 }, { "epoch": 1.3210627599478553, "grad_norm": 0.5343284051223263, "learning_rate": 6.86736136138214e-05, "loss": 2.8385, "step": 21281 }, { "epoch": 1.3211248370476132, "grad_norm": 0.5516997450259077, "learning_rate": 6.867026336597027e-05, "loss": 2.8518, "step": 21282 }, { "epoch": 1.321186914147371, "grad_norm": 0.5155888169048768, "learning_rate": 6.866691302071183e-05, "loss": 2.7648, "step": 21283 }, { "epoch": 1.321248991247129, "grad_norm": 0.5592005175176089, "learning_rate": 6.866356257806358e-05, "loss": 2.8008, "step": 21284 }, { "epoch": 1.321311068346887, "grad_norm": 0.588415855418861, "learning_rate": 6.8660212038043e-05, "loss": 2.8526, "step": 21285 }, { "epoch": 1.3213731454466449, "grad_norm": 0.6493985052324767, "learning_rate": 6.865686140066757e-05, "loss": 2.8253, "step": 21286 }, { "epoch": 1.3214352225464026, "grad_norm": 0.5963424503780221, "learning_rate": 6.865351066595475e-05, "loss": 2.8828, "step": 21287 }, { "epoch": 1.3214972996461605, "grad_norm": 0.5523245272603009, "learning_rate": 6.865015983392203e-05, "loss": 2.904, "step": 21288 }, { "epoch": 1.3215593767459184, "grad_norm": 0.6079893742310439, "learning_rate": 6.864680890458691e-05, "loss": 2.7596, "step": 21289 }, { "epoch": 1.3216214538456763, "grad_norm": 0.54231560482672, "learning_rate": 6.864345787796686e-05, "loss": 2.7773, "step": 21290 }, { "epoch": 1.3216835309454342, "grad_norm": 0.6448122279063164, "learning_rate": 6.864010675407937e-05, "loss": 2.806, "step": 21291 }, { "epoch": 1.3217456080451921, "grad_norm": 0.5398469552194753, "learning_rate": 6.863675553294191e-05, "loss": 2.8032, "step": 21292 }, { "epoch": 1.32180768514495, "grad_norm": 0.5824342395025203, "learning_rate": 6.863340421457199e-05, "loss": 2.8003, "step": 21293 }, { "epoch": 1.321869762244708, "grad_norm": 0.593478492096432, "learning_rate": 6.863005279898705e-05, "loss": 2.8415, "step": 21294 }, { "epoch": 1.321931839344466, "grad_norm": 0.5895870989307438, "learning_rate": 6.862670128620462e-05, "loss": 2.9152, "step": 21295 }, { "epoch": 1.3219939164442236, "grad_norm": 0.6201175960028797, "learning_rate": 6.862334967624217e-05, "loss": 2.9076, "step": 21296 }, { "epoch": 1.3220559935439815, "grad_norm": 0.5729789844490412, "learning_rate": 6.861999796911716e-05, "loss": 2.8665, "step": 21297 }, { "epoch": 1.3221180706437394, "grad_norm": 0.641553810235441, "learning_rate": 6.861664616484713e-05, "loss": 2.7567, "step": 21298 }, { "epoch": 1.3221801477434973, "grad_norm": 0.6199087438412542, "learning_rate": 6.86132942634495e-05, "loss": 2.8746, "step": 21299 }, { "epoch": 1.3222422248432553, "grad_norm": 0.645585348346317, "learning_rate": 6.860994226494183e-05, "loss": 2.7923, "step": 21300 }, { "epoch": 1.3223043019430132, "grad_norm": 0.6387937703598019, "learning_rate": 6.860659016934157e-05, "loss": 2.8945, "step": 21301 }, { "epoch": 1.322366379042771, "grad_norm": 0.6203266901271707, "learning_rate": 6.860323797666619e-05, "loss": 2.8284, "step": 21302 }, { "epoch": 1.322428456142529, "grad_norm": 0.5306155118329857, "learning_rate": 6.859988568693323e-05, "loss": 2.8493, "step": 21303 }, { "epoch": 1.322490533242287, "grad_norm": 0.6209778107688593, "learning_rate": 6.859653330016013e-05, "loss": 2.8098, "step": 21304 }, { "epoch": 1.3225526103420449, "grad_norm": 0.5419654358456376, "learning_rate": 6.859318081636442e-05, "loss": 2.8527, "step": 21305 }, { "epoch": 1.3226146874418028, "grad_norm": 0.5809810046752844, "learning_rate": 6.858982823556355e-05, "loss": 2.7311, "step": 21306 }, { "epoch": 1.3226767645415607, "grad_norm": 0.5936435553855139, "learning_rate": 6.858647555777504e-05, "loss": 2.8253, "step": 21307 }, { "epoch": 1.3227388416413186, "grad_norm": 0.5747403481159615, "learning_rate": 6.858312278301637e-05, "loss": 2.8617, "step": 21308 }, { "epoch": 1.3228009187410765, "grad_norm": 0.5839213716847204, "learning_rate": 6.857976991130506e-05, "loss": 2.8334, "step": 21309 }, { "epoch": 1.3228629958408344, "grad_norm": 0.5956094887105716, "learning_rate": 6.857641694265856e-05, "loss": 2.8245, "step": 21310 }, { "epoch": 1.3229250729405921, "grad_norm": 0.6035475737871128, "learning_rate": 6.857306387709439e-05, "loss": 2.8457, "step": 21311 }, { "epoch": 1.32298715004035, "grad_norm": 0.5926826570935111, "learning_rate": 6.856971071463004e-05, "loss": 2.8451, "step": 21312 }, { "epoch": 1.323049227140108, "grad_norm": 0.6352841885361907, "learning_rate": 6.856635745528299e-05, "loss": 2.8671, "step": 21313 }, { "epoch": 1.323111304239866, "grad_norm": 0.5674884323588045, "learning_rate": 6.856300409907077e-05, "loss": 2.7066, "step": 21314 }, { "epoch": 1.3231733813396238, "grad_norm": 0.6000530825419175, "learning_rate": 6.855965064601082e-05, "loss": 2.8237, "step": 21315 }, { "epoch": 1.3232354584393817, "grad_norm": 0.5859697460202511, "learning_rate": 6.855629709612068e-05, "loss": 2.8158, "step": 21316 }, { "epoch": 1.3232975355391396, "grad_norm": 0.5280297515120022, "learning_rate": 6.855294344941782e-05, "loss": 2.8439, "step": 21317 }, { "epoch": 1.3233596126388976, "grad_norm": 0.6474264834803752, "learning_rate": 6.854958970591978e-05, "loss": 2.8684, "step": 21318 }, { "epoch": 1.3234216897386555, "grad_norm": 0.5239628674694576, "learning_rate": 6.854623586564401e-05, "loss": 2.8618, "step": 21319 }, { "epoch": 1.3234837668384132, "grad_norm": 0.5630568820682845, "learning_rate": 6.854288192860803e-05, "loss": 2.7775, "step": 21320 }, { "epoch": 1.323545843938171, "grad_norm": 0.5350255319260705, "learning_rate": 6.853952789482933e-05, "loss": 2.8366, "step": 21321 }, { "epoch": 1.323607921037929, "grad_norm": 0.5577897908693155, "learning_rate": 6.853617376432542e-05, "loss": 2.7646, "step": 21322 }, { "epoch": 1.323669998137687, "grad_norm": 0.6120666333182369, "learning_rate": 6.853281953711379e-05, "loss": 2.8441, "step": 21323 }, { "epoch": 1.3237320752374448, "grad_norm": 0.6666936918582479, "learning_rate": 6.852946521321193e-05, "loss": 2.6942, "step": 21324 }, { "epoch": 1.3237941523372028, "grad_norm": 0.5894839724286224, "learning_rate": 6.852611079263737e-05, "loss": 2.7867, "step": 21325 }, { "epoch": 1.3238562294369607, "grad_norm": 0.6555401732459414, "learning_rate": 6.852275627540757e-05, "loss": 2.8473, "step": 21326 }, { "epoch": 1.3239183065367186, "grad_norm": 0.6185796946417227, "learning_rate": 6.851940166154007e-05, "loss": 2.7687, "step": 21327 }, { "epoch": 1.3239803836364765, "grad_norm": 0.5522305244389485, "learning_rate": 6.851604695105235e-05, "loss": 2.8211, "step": 21328 }, { "epoch": 1.3240424607362344, "grad_norm": 0.6355257149182535, "learning_rate": 6.851269214396192e-05, "loss": 2.9071, "step": 21329 }, { "epoch": 1.3241045378359924, "grad_norm": 0.48237422754147286, "learning_rate": 6.850933724028629e-05, "loss": 2.8328, "step": 21330 }, { "epoch": 1.3241666149357503, "grad_norm": 0.6494585294486025, "learning_rate": 6.850598224004293e-05, "loss": 2.7583, "step": 21331 }, { "epoch": 1.3242286920355082, "grad_norm": 0.589946333699328, "learning_rate": 6.850262714324938e-05, "loss": 2.8651, "step": 21332 }, { "epoch": 1.324290769135266, "grad_norm": 0.5453439260789301, "learning_rate": 6.849927194992312e-05, "loss": 2.8281, "step": 21333 }, { "epoch": 1.324352846235024, "grad_norm": 0.6594307428454335, "learning_rate": 6.84959166600817e-05, "loss": 2.8375, "step": 21334 }, { "epoch": 1.3244149233347817, "grad_norm": 0.48311876626082845, "learning_rate": 6.849256127374258e-05, "loss": 2.7847, "step": 21335 }, { "epoch": 1.3244770004345396, "grad_norm": 0.6087027778952533, "learning_rate": 6.848920579092327e-05, "loss": 2.8439, "step": 21336 }, { "epoch": 1.3245390775342976, "grad_norm": 0.49701104251061384, "learning_rate": 6.848585021164129e-05, "loss": 2.9113, "step": 21337 }, { "epoch": 1.3246011546340555, "grad_norm": 0.6296157862337673, "learning_rate": 6.848249453591413e-05, "loss": 2.797, "step": 21338 }, { "epoch": 1.3246632317338134, "grad_norm": 0.5913997362551651, "learning_rate": 6.847913876375933e-05, "loss": 2.8261, "step": 21339 }, { "epoch": 1.3247253088335713, "grad_norm": 0.6115434163916643, "learning_rate": 6.847578289519436e-05, "loss": 2.9455, "step": 21340 }, { "epoch": 1.3247873859333292, "grad_norm": 0.6786899041957832, "learning_rate": 6.847242693023676e-05, "loss": 2.9243, "step": 21341 }, { "epoch": 1.3248494630330871, "grad_norm": 0.671920952223539, "learning_rate": 6.846907086890402e-05, "loss": 2.7449, "step": 21342 }, { "epoch": 1.324911540132845, "grad_norm": 0.6346059510334597, "learning_rate": 6.846571471121365e-05, "loss": 2.8145, "step": 21343 }, { "epoch": 1.3249736172326028, "grad_norm": 0.7011506016301388, "learning_rate": 6.846235845718316e-05, "loss": 2.771, "step": 21344 }, { "epoch": 1.3250356943323607, "grad_norm": 0.5586968706887603, "learning_rate": 6.845900210683007e-05, "loss": 2.7134, "step": 21345 }, { "epoch": 1.3250977714321186, "grad_norm": 0.6782545352746197, "learning_rate": 6.845564566017189e-05, "loss": 2.868, "step": 21346 }, { "epoch": 1.3251598485318765, "grad_norm": 0.5666972498027143, "learning_rate": 6.845228911722612e-05, "loss": 2.7174, "step": 21347 }, { "epoch": 1.3252219256316344, "grad_norm": 0.6170237601649867, "learning_rate": 6.844893247801026e-05, "loss": 2.8328, "step": 21348 }, { "epoch": 1.3252840027313924, "grad_norm": 0.5577220918280934, "learning_rate": 6.844557574254186e-05, "loss": 2.8531, "step": 21349 }, { "epoch": 1.3253460798311503, "grad_norm": 0.6004508013776237, "learning_rate": 6.844221891083842e-05, "loss": 2.8358, "step": 21350 }, { "epoch": 1.3254081569309082, "grad_norm": 0.5384959685405829, "learning_rate": 6.843886198291743e-05, "loss": 2.8154, "step": 21351 }, { "epoch": 1.325470234030666, "grad_norm": 0.5750860554112435, "learning_rate": 6.843550495879643e-05, "loss": 2.7954, "step": 21352 }, { "epoch": 1.325532311130424, "grad_norm": 0.5929221606139666, "learning_rate": 6.843214783849292e-05, "loss": 2.8894, "step": 21353 }, { "epoch": 1.325594388230182, "grad_norm": 0.48612570475818795, "learning_rate": 6.842879062202444e-05, "loss": 2.8273, "step": 21354 }, { "epoch": 1.3256564653299399, "grad_norm": 0.5341154963056057, "learning_rate": 6.842543330940846e-05, "loss": 2.7816, "step": 21355 }, { "epoch": 1.3257185424296978, "grad_norm": 0.4934445631549668, "learning_rate": 6.842207590066254e-05, "loss": 2.7821, "step": 21356 }, { "epoch": 1.3257806195294557, "grad_norm": 0.5462622615825997, "learning_rate": 6.841871839580417e-05, "loss": 2.726, "step": 21357 }, { "epoch": 1.3258426966292136, "grad_norm": 0.5524967343246816, "learning_rate": 6.841536079485087e-05, "loss": 2.777, "step": 21358 }, { "epoch": 1.3259047737289713, "grad_norm": 0.5951501249407285, "learning_rate": 6.841200309782017e-05, "loss": 2.8712, "step": 21359 }, { "epoch": 1.3259668508287292, "grad_norm": 0.4783577016140472, "learning_rate": 6.840864530472957e-05, "loss": 2.7933, "step": 21360 }, { "epoch": 1.3260289279284871, "grad_norm": 0.5687021486882012, "learning_rate": 6.840528741559662e-05, "loss": 2.8519, "step": 21361 }, { "epoch": 1.326091005028245, "grad_norm": 0.5341447055108975, "learning_rate": 6.840192943043879e-05, "loss": 2.8277, "step": 21362 }, { "epoch": 1.326153082128003, "grad_norm": 0.5544567702841285, "learning_rate": 6.839857134927366e-05, "loss": 2.8193, "step": 21363 }, { "epoch": 1.326215159227761, "grad_norm": 0.5117227394006735, "learning_rate": 6.83952131721187e-05, "loss": 2.8296, "step": 21364 }, { "epoch": 1.3262772363275188, "grad_norm": 0.5507109847463746, "learning_rate": 6.839185489899143e-05, "loss": 2.8788, "step": 21365 }, { "epoch": 1.3263393134272767, "grad_norm": 0.5276604160802936, "learning_rate": 6.83884965299094e-05, "loss": 2.7534, "step": 21366 }, { "epoch": 1.3264013905270347, "grad_norm": 0.5763912753039164, "learning_rate": 6.838513806489012e-05, "loss": 2.857, "step": 21367 }, { "epoch": 1.3264634676267923, "grad_norm": 0.5983891450155591, "learning_rate": 6.838177950395111e-05, "loss": 2.7366, "step": 21368 }, { "epoch": 1.3265255447265503, "grad_norm": 0.5864648099262737, "learning_rate": 6.83784208471099e-05, "loss": 2.8522, "step": 21369 }, { "epoch": 1.3265876218263082, "grad_norm": 0.5843139789973469, "learning_rate": 6.837506209438402e-05, "loss": 2.7111, "step": 21370 }, { "epoch": 1.326649698926066, "grad_norm": 0.5692363933349338, "learning_rate": 6.837170324579095e-05, "loss": 2.7903, "step": 21371 }, { "epoch": 1.326711776025824, "grad_norm": 0.6023817057840887, "learning_rate": 6.836834430134826e-05, "loss": 2.8743, "step": 21372 }, { "epoch": 1.326773853125582, "grad_norm": 0.6350019513873657, "learning_rate": 6.836498526107346e-05, "loss": 2.843, "step": 21373 }, { "epoch": 1.3268359302253399, "grad_norm": 0.5313671207350819, "learning_rate": 6.836162612498409e-05, "loss": 2.7272, "step": 21374 }, { "epoch": 1.3268980073250978, "grad_norm": 0.5893821173002527, "learning_rate": 6.835826689309763e-05, "loss": 2.8676, "step": 21375 }, { "epoch": 1.3269600844248557, "grad_norm": 0.5757314000489856, "learning_rate": 6.835490756543165e-05, "loss": 2.7844, "step": 21376 }, { "epoch": 1.3270221615246136, "grad_norm": 0.4984605006588272, "learning_rate": 6.835154814200367e-05, "loss": 2.8015, "step": 21377 }, { "epoch": 1.3270842386243715, "grad_norm": 0.5739364708707718, "learning_rate": 6.834818862283117e-05, "loss": 2.8367, "step": 21378 }, { "epoch": 1.3271463157241294, "grad_norm": 0.5236123775036262, "learning_rate": 6.834482900793175e-05, "loss": 2.835, "step": 21379 }, { "epoch": 1.3272083928238874, "grad_norm": 0.5773167501906203, "learning_rate": 6.834146929732289e-05, "loss": 2.816, "step": 21380 }, { "epoch": 1.3272704699236453, "grad_norm": 0.5283286610506044, "learning_rate": 6.833810949102213e-05, "loss": 2.7191, "step": 21381 }, { "epoch": 1.3273325470234032, "grad_norm": 0.5486148134608291, "learning_rate": 6.833474958904702e-05, "loss": 2.8324, "step": 21382 }, { "epoch": 1.327394624123161, "grad_norm": 0.5721393224495763, "learning_rate": 6.833138959141506e-05, "loss": 2.8373, "step": 21383 }, { "epoch": 1.3274567012229188, "grad_norm": 0.5818954155656287, "learning_rate": 6.832802949814378e-05, "loss": 2.93, "step": 21384 }, { "epoch": 1.3275187783226767, "grad_norm": 0.5277137023807267, "learning_rate": 6.832466930925073e-05, "loss": 2.8508, "step": 21385 }, { "epoch": 1.3275808554224346, "grad_norm": 0.5543391786203287, "learning_rate": 6.832130902475344e-05, "loss": 2.8573, "step": 21386 }, { "epoch": 1.3276429325221926, "grad_norm": 0.573761712704784, "learning_rate": 6.831794864466941e-05, "loss": 2.7861, "step": 21387 }, { "epoch": 1.3277050096219505, "grad_norm": 0.5915978477867678, "learning_rate": 6.831458816901623e-05, "loss": 2.8273, "step": 21388 }, { "epoch": 1.3277670867217084, "grad_norm": 0.5360355254186081, "learning_rate": 6.831122759781137e-05, "loss": 2.7036, "step": 21389 }, { "epoch": 1.3278291638214663, "grad_norm": 0.5996992062553711, "learning_rate": 6.83078669310724e-05, "loss": 2.7562, "step": 21390 }, { "epoch": 1.3278912409212242, "grad_norm": 0.5911378873943631, "learning_rate": 6.830450616881684e-05, "loss": 2.8908, "step": 21391 }, { "epoch": 1.327953318020982, "grad_norm": 0.5775693078582583, "learning_rate": 6.830114531106224e-05, "loss": 2.785, "step": 21392 }, { "epoch": 1.3280153951207398, "grad_norm": 0.5645111212585199, "learning_rate": 6.829778435782611e-05, "loss": 2.8304, "step": 21393 }, { "epoch": 1.3280774722204978, "grad_norm": 0.6428793111452991, "learning_rate": 6.829442330912599e-05, "loss": 2.824, "step": 21394 }, { "epoch": 1.3281395493202557, "grad_norm": 0.5384442751076849, "learning_rate": 6.829106216497944e-05, "loss": 2.7592, "step": 21395 }, { "epoch": 1.3282016264200136, "grad_norm": 0.6071403562470227, "learning_rate": 6.828770092540396e-05, "loss": 2.792, "step": 21396 }, { "epoch": 1.3282637035197715, "grad_norm": 0.5359844009296654, "learning_rate": 6.828433959041712e-05, "loss": 2.8619, "step": 21397 }, { "epoch": 1.3283257806195294, "grad_norm": 0.6101888943388306, "learning_rate": 6.828097816003644e-05, "loss": 2.7896, "step": 21398 }, { "epoch": 1.3283878577192874, "grad_norm": 0.5690033924591622, "learning_rate": 6.827761663427944e-05, "loss": 2.8439, "step": 21399 }, { "epoch": 1.3284499348190453, "grad_norm": 0.6000047802734685, "learning_rate": 6.82742550131637e-05, "loss": 2.8407, "step": 21400 }, { "epoch": 1.3285120119188032, "grad_norm": 0.5323213825279739, "learning_rate": 6.827089329670673e-05, "loss": 2.8683, "step": 21401 }, { "epoch": 1.328574089018561, "grad_norm": 0.5635907141823435, "learning_rate": 6.826753148492607e-05, "loss": 2.8132, "step": 21402 }, { "epoch": 1.328636166118319, "grad_norm": 0.5155517930745581, "learning_rate": 6.826416957783926e-05, "loss": 2.7702, "step": 21403 }, { "epoch": 1.328698243218077, "grad_norm": 0.520294489276743, "learning_rate": 6.826080757546385e-05, "loss": 2.8454, "step": 21404 }, { "epoch": 1.3287603203178349, "grad_norm": 0.5322759762844653, "learning_rate": 6.825744547781738e-05, "loss": 2.7809, "step": 21405 }, { "epoch": 1.3288223974175928, "grad_norm": 0.5167601834028757, "learning_rate": 6.825408328491737e-05, "loss": 2.752, "step": 21406 }, { "epoch": 1.3288844745173505, "grad_norm": 0.5384836821142497, "learning_rate": 6.825072099678138e-05, "loss": 2.846, "step": 21407 }, { "epoch": 1.3289465516171084, "grad_norm": 0.5032477993109274, "learning_rate": 6.824735861342696e-05, "loss": 2.7844, "step": 21408 }, { "epoch": 1.3290086287168663, "grad_norm": 0.55310415562724, "learning_rate": 6.824399613487163e-05, "loss": 2.7828, "step": 21409 }, { "epoch": 1.3290707058166242, "grad_norm": 0.5398214498460133, "learning_rate": 6.824063356113294e-05, "loss": 2.7703, "step": 21410 }, { "epoch": 1.3291327829163821, "grad_norm": 0.5415796032652055, "learning_rate": 6.823727089222845e-05, "loss": 2.7588, "step": 21411 }, { "epoch": 1.32919486001614, "grad_norm": 0.530090272296206, "learning_rate": 6.823390812817568e-05, "loss": 2.7486, "step": 21412 }, { "epoch": 1.329256937115898, "grad_norm": 0.5530429962492086, "learning_rate": 6.823054526899219e-05, "loss": 2.8442, "step": 21413 }, { "epoch": 1.329319014215656, "grad_norm": 0.527500270644566, "learning_rate": 6.822718231469551e-05, "loss": 2.7466, "step": 21414 }, { "epoch": 1.3293810913154138, "grad_norm": 0.5780431586407879, "learning_rate": 6.82238192653032e-05, "loss": 2.8682, "step": 21415 }, { "epoch": 1.3294431684151715, "grad_norm": 0.57727175974816, "learning_rate": 6.82204561208328e-05, "loss": 2.8209, "step": 21416 }, { "epoch": 1.3295052455149294, "grad_norm": 0.5647474003631079, "learning_rate": 6.821709288130187e-05, "loss": 2.7992, "step": 21417 }, { "epoch": 1.3295673226146874, "grad_norm": 0.6192944693539699, "learning_rate": 6.821372954672794e-05, "loss": 2.8839, "step": 21418 }, { "epoch": 1.3296293997144453, "grad_norm": 0.5555158259835332, "learning_rate": 6.821036611712854e-05, "loss": 2.7039, "step": 21419 }, { "epoch": 1.3296914768142032, "grad_norm": 0.5716011433914954, "learning_rate": 6.820700259252127e-05, "loss": 2.8797, "step": 21420 }, { "epoch": 1.329753553913961, "grad_norm": 0.5869862389892987, "learning_rate": 6.820363897292363e-05, "loss": 2.8319, "step": 21421 }, { "epoch": 1.329815631013719, "grad_norm": 0.5497213676563363, "learning_rate": 6.82002752583532e-05, "loss": 2.8304, "step": 21422 }, { "epoch": 1.329877708113477, "grad_norm": 0.5142349522048332, "learning_rate": 6.81969114488275e-05, "loss": 2.8444, "step": 21423 }, { "epoch": 1.3299397852132349, "grad_norm": 0.5544943137294398, "learning_rate": 6.819354754436409e-05, "loss": 2.8288, "step": 21424 }, { "epoch": 1.3300018623129928, "grad_norm": 0.5087784017372198, "learning_rate": 6.819018354498053e-05, "loss": 2.8588, "step": 21425 }, { "epoch": 1.3300639394127507, "grad_norm": 0.48989940308910007, "learning_rate": 6.818681945069438e-05, "loss": 2.7535, "step": 21426 }, { "epoch": 1.3301260165125086, "grad_norm": 0.5807424780399157, "learning_rate": 6.818345526152316e-05, "loss": 2.7979, "step": 21427 }, { "epoch": 1.3301880936122665, "grad_norm": 0.5041778187529831, "learning_rate": 6.818009097748447e-05, "loss": 2.7918, "step": 21428 }, { "epoch": 1.3302501707120244, "grad_norm": 0.5912638363015634, "learning_rate": 6.817672659859582e-05, "loss": 2.8669, "step": 21429 }, { "epoch": 1.3303122478117824, "grad_norm": 0.5422236867476229, "learning_rate": 6.817336212487475e-05, "loss": 2.871, "step": 21430 }, { "epoch": 1.33037432491154, "grad_norm": 0.5247914751644878, "learning_rate": 6.816999755633886e-05, "loss": 2.8466, "step": 21431 }, { "epoch": 1.330436402011298, "grad_norm": 0.5870159143308411, "learning_rate": 6.816663289300566e-05, "loss": 2.8614, "step": 21432 }, { "epoch": 1.330498479111056, "grad_norm": 0.5404832885763029, "learning_rate": 6.816326813489275e-05, "loss": 2.7806, "step": 21433 }, { "epoch": 1.3305605562108138, "grad_norm": 0.6302156500483562, "learning_rate": 6.815990328201766e-05, "loss": 2.8727, "step": 21434 }, { "epoch": 1.3306226333105717, "grad_norm": 0.5278888287188365, "learning_rate": 6.815653833439795e-05, "loss": 2.8857, "step": 21435 }, { "epoch": 1.3306847104103297, "grad_norm": 0.5860409708851809, "learning_rate": 6.815317329205116e-05, "loss": 2.8175, "step": 21436 }, { "epoch": 1.3307467875100876, "grad_norm": 0.6013323108140339, "learning_rate": 6.814980815499486e-05, "loss": 2.8248, "step": 21437 }, { "epoch": 1.3308088646098455, "grad_norm": 0.5216827301085195, "learning_rate": 6.81464429232466e-05, "loss": 2.8503, "step": 21438 }, { "epoch": 1.3308709417096034, "grad_norm": 0.6066917127724518, "learning_rate": 6.814307759682395e-05, "loss": 2.8039, "step": 21439 }, { "epoch": 1.330933018809361, "grad_norm": 0.5562870441917406, "learning_rate": 6.813971217574446e-05, "loss": 2.7996, "step": 21440 }, { "epoch": 1.330995095909119, "grad_norm": 0.5427061310306706, "learning_rate": 6.813634666002568e-05, "loss": 2.8478, "step": 21441 }, { "epoch": 1.331057173008877, "grad_norm": 0.5814374160102259, "learning_rate": 6.813298104968518e-05, "loss": 2.914, "step": 21442 }, { "epoch": 1.3311192501086349, "grad_norm": 0.5821973064536553, "learning_rate": 6.812961534474052e-05, "loss": 2.8142, "step": 21443 }, { "epoch": 1.3311813272083928, "grad_norm": 0.5646011791951361, "learning_rate": 6.812624954520925e-05, "loss": 2.7859, "step": 21444 }, { "epoch": 1.3312434043081507, "grad_norm": 0.5646925685636095, "learning_rate": 6.812288365110893e-05, "loss": 2.744, "step": 21445 }, { "epoch": 1.3313054814079086, "grad_norm": 0.5558546171969909, "learning_rate": 6.811951766245715e-05, "loss": 2.8014, "step": 21446 }, { "epoch": 1.3313675585076665, "grad_norm": 0.612460370630712, "learning_rate": 6.811615157927143e-05, "loss": 2.7306, "step": 21447 }, { "epoch": 1.3314296356074244, "grad_norm": 0.5362312212681531, "learning_rate": 6.811278540156934e-05, "loss": 2.7563, "step": 21448 }, { "epoch": 1.3314917127071824, "grad_norm": 0.6324268625968688, "learning_rate": 6.810941912936848e-05, "loss": 2.8937, "step": 21449 }, { "epoch": 1.3315537898069403, "grad_norm": 0.5689715205558971, "learning_rate": 6.810605276268635e-05, "loss": 2.7837, "step": 21450 }, { "epoch": 1.3316158669066982, "grad_norm": 0.618470126732753, "learning_rate": 6.810268630154057e-05, "loss": 2.8594, "step": 21451 }, { "epoch": 1.3316779440064561, "grad_norm": 0.5751539387095593, "learning_rate": 6.809931974594867e-05, "loss": 2.849, "step": 21452 }, { "epoch": 1.331740021106214, "grad_norm": 0.6249316893440108, "learning_rate": 6.809595309592825e-05, "loss": 2.8116, "step": 21453 }, { "epoch": 1.331802098205972, "grad_norm": 0.5164078347416972, "learning_rate": 6.809258635149684e-05, "loss": 2.8282, "step": 21454 }, { "epoch": 1.3318641753057296, "grad_norm": 0.6232702638045645, "learning_rate": 6.8089219512672e-05, "loss": 2.8483, "step": 21455 }, { "epoch": 1.3319262524054876, "grad_norm": 0.5756520140498095, "learning_rate": 6.80858525794713e-05, "loss": 2.8232, "step": 21456 }, { "epoch": 1.3319883295052455, "grad_norm": 0.5362599540000016, "learning_rate": 6.808248555191235e-05, "loss": 2.8477, "step": 21457 }, { "epoch": 1.3320504066050034, "grad_norm": 0.5499315132502838, "learning_rate": 6.807911843001267e-05, "loss": 2.9087, "step": 21458 }, { "epoch": 1.3321124837047613, "grad_norm": 0.5324137227364172, "learning_rate": 6.807575121378982e-05, "loss": 2.9091, "step": 21459 }, { "epoch": 1.3321745608045192, "grad_norm": 0.5243344379691492, "learning_rate": 6.807238390326142e-05, "loss": 2.7984, "step": 21460 }, { "epoch": 1.3322366379042772, "grad_norm": 0.5485529362406705, "learning_rate": 6.806901649844499e-05, "loss": 2.8511, "step": 21461 }, { "epoch": 1.332298715004035, "grad_norm": 0.5542678857686855, "learning_rate": 6.806564899935811e-05, "loss": 2.796, "step": 21462 }, { "epoch": 1.332360792103793, "grad_norm": 0.5279967327215512, "learning_rate": 6.806228140601836e-05, "loss": 2.823, "step": 21463 }, { "epoch": 1.3324228692035507, "grad_norm": 0.639433077601059, "learning_rate": 6.805891371844329e-05, "loss": 2.8253, "step": 21464 }, { "epoch": 1.3324849463033086, "grad_norm": 0.5580521205125258, "learning_rate": 6.80555459366505e-05, "loss": 2.8692, "step": 21465 }, { "epoch": 1.3325470234030665, "grad_norm": 0.5781748982108637, "learning_rate": 6.805217806065753e-05, "loss": 2.7949, "step": 21466 }, { "epoch": 1.3326091005028244, "grad_norm": 0.5942918663972245, "learning_rate": 6.804881009048198e-05, "loss": 2.819, "step": 21467 }, { "epoch": 1.3326711776025824, "grad_norm": 0.5687622530109873, "learning_rate": 6.80454420261414e-05, "loss": 2.8431, "step": 21468 }, { "epoch": 1.3327332547023403, "grad_norm": 0.6697673118834375, "learning_rate": 6.804207386765338e-05, "loss": 2.7969, "step": 21469 }, { "epoch": 1.3327953318020982, "grad_norm": 0.552397069016677, "learning_rate": 6.803870561503546e-05, "loss": 2.8886, "step": 21470 }, { "epoch": 1.3328574089018561, "grad_norm": 0.6390240133023974, "learning_rate": 6.803533726830524e-05, "loss": 2.7765, "step": 21471 }, { "epoch": 1.332919486001614, "grad_norm": 0.6315010751773987, "learning_rate": 6.803196882748029e-05, "loss": 2.8889, "step": 21472 }, { "epoch": 1.332981563101372, "grad_norm": 0.6675124143310419, "learning_rate": 6.802860029257818e-05, "loss": 2.7891, "step": 21473 }, { "epoch": 1.3330436402011299, "grad_norm": 0.555959693289876, "learning_rate": 6.80252316636165e-05, "loss": 2.7533, "step": 21474 }, { "epoch": 1.3331057173008878, "grad_norm": 0.6723738969395602, "learning_rate": 6.802186294061278e-05, "loss": 2.8324, "step": 21475 }, { "epoch": 1.3331677944006457, "grad_norm": 0.5672878638971388, "learning_rate": 6.801849412358466e-05, "loss": 2.88, "step": 21476 }, { "epoch": 1.3332298715004036, "grad_norm": 0.6077635316319158, "learning_rate": 6.801512521254966e-05, "loss": 2.8889, "step": 21477 }, { "epoch": 1.3332919486001615, "grad_norm": 0.6010459068937714, "learning_rate": 6.801175620752539e-05, "loss": 2.8152, "step": 21478 }, { "epoch": 1.3333540256999192, "grad_norm": 0.5422865884254651, "learning_rate": 6.80083871085294e-05, "loss": 2.8436, "step": 21479 }, { "epoch": 1.3334161027996771, "grad_norm": 0.617935065932065, "learning_rate": 6.80050179155793e-05, "loss": 2.8815, "step": 21480 }, { "epoch": 1.333478179899435, "grad_norm": 0.575276859636166, "learning_rate": 6.800164862869264e-05, "loss": 2.8847, "step": 21481 }, { "epoch": 1.333540256999193, "grad_norm": 0.5744663405405998, "learning_rate": 6.7998279247887e-05, "loss": 2.787, "step": 21482 }, { "epoch": 1.333602334098951, "grad_norm": 0.580154182765328, "learning_rate": 6.799490977318e-05, "loss": 2.8, "step": 21483 }, { "epoch": 1.3336644111987088, "grad_norm": 0.5545703871763819, "learning_rate": 6.799154020458916e-05, "loss": 2.7844, "step": 21484 }, { "epoch": 1.3337264882984667, "grad_norm": 0.5735627404011154, "learning_rate": 6.798817054213209e-05, "loss": 2.7655, "step": 21485 }, { "epoch": 1.3337885653982247, "grad_norm": 0.5372104064247328, "learning_rate": 6.798480078582637e-05, "loss": 2.7449, "step": 21486 }, { "epoch": 1.3338506424979826, "grad_norm": 0.6579167906302235, "learning_rate": 6.798143093568958e-05, "loss": 2.8107, "step": 21487 }, { "epoch": 1.3339127195977403, "grad_norm": 0.5710291866435769, "learning_rate": 6.797806099173929e-05, "loss": 2.8247, "step": 21488 }, { "epoch": 1.3339747966974982, "grad_norm": 0.6022887861652915, "learning_rate": 6.797469095399311e-05, "loss": 2.8019, "step": 21489 }, { "epoch": 1.334036873797256, "grad_norm": 0.5267714892214774, "learning_rate": 6.79713208224686e-05, "loss": 2.7836, "step": 21490 }, { "epoch": 1.334098950897014, "grad_norm": 0.6535282231842067, "learning_rate": 6.796795059718333e-05, "loss": 2.8984, "step": 21491 }, { "epoch": 1.334161027996772, "grad_norm": 0.6071906677022106, "learning_rate": 6.796458027815492e-05, "loss": 2.6928, "step": 21492 }, { "epoch": 1.3342231050965299, "grad_norm": 0.5155961577700682, "learning_rate": 6.796120986540092e-05, "loss": 2.8556, "step": 21493 }, { "epoch": 1.3342851821962878, "grad_norm": 0.6017072676813918, "learning_rate": 6.795783935893895e-05, "loss": 2.8191, "step": 21494 }, { "epoch": 1.3343472592960457, "grad_norm": 0.5497902687038598, "learning_rate": 6.795446875878655e-05, "loss": 2.8125, "step": 21495 }, { "epoch": 1.3344093363958036, "grad_norm": 0.6120842593164193, "learning_rate": 6.795109806496132e-05, "loss": 2.7678, "step": 21496 }, { "epoch": 1.3344714134955615, "grad_norm": 0.6121762276313173, "learning_rate": 6.794772727748088e-05, "loss": 2.7924, "step": 21497 }, { "epoch": 1.3345334905953194, "grad_norm": 0.6112262531262523, "learning_rate": 6.794435639636276e-05, "loss": 2.7758, "step": 21498 }, { "epoch": 1.3345955676950774, "grad_norm": 0.5741329258729431, "learning_rate": 6.79409854216246e-05, "loss": 2.8007, "step": 21499 }, { "epoch": 1.3346576447948353, "grad_norm": 0.6141173864785991, "learning_rate": 6.793761435328395e-05, "loss": 2.7869, "step": 21500 }, { "epoch": 1.3347197218945932, "grad_norm": 0.5664768306559667, "learning_rate": 6.793424319135843e-05, "loss": 2.8506, "step": 21501 }, { "epoch": 1.3347817989943511, "grad_norm": 0.5712552574132814, "learning_rate": 6.793087193586559e-05, "loss": 2.8698, "step": 21502 }, { "epoch": 1.3348438760941088, "grad_norm": 0.5511383900788367, "learning_rate": 6.792750058682305e-05, "loss": 2.8473, "step": 21503 }, { "epoch": 1.3349059531938667, "grad_norm": 0.563552375334451, "learning_rate": 6.792412914424838e-05, "loss": 2.7219, "step": 21504 }, { "epoch": 1.3349680302936247, "grad_norm": 0.5380103028686468, "learning_rate": 6.792075760815919e-05, "loss": 2.7157, "step": 21505 }, { "epoch": 1.3350301073933826, "grad_norm": 0.6140773247246988, "learning_rate": 6.791738597857304e-05, "loss": 2.7524, "step": 21506 }, { "epoch": 1.3350921844931405, "grad_norm": 0.5287482878023747, "learning_rate": 6.791401425550756e-05, "loss": 2.7909, "step": 21507 }, { "epoch": 1.3351542615928984, "grad_norm": 0.534845470767829, "learning_rate": 6.79106424389803e-05, "loss": 2.7775, "step": 21508 }, { "epoch": 1.3352163386926563, "grad_norm": 0.5635832052616596, "learning_rate": 6.790727052900887e-05, "loss": 2.8714, "step": 21509 }, { "epoch": 1.3352784157924142, "grad_norm": 0.5136691191276626, "learning_rate": 6.790389852561088e-05, "loss": 2.6648, "step": 21510 }, { "epoch": 1.3353404928921722, "grad_norm": 0.5073595521167708, "learning_rate": 6.790052642880388e-05, "loss": 2.7514, "step": 21511 }, { "epoch": 1.3354025699919299, "grad_norm": 0.5673239803835739, "learning_rate": 6.78971542386055e-05, "loss": 2.7691, "step": 21512 }, { "epoch": 1.3354646470916878, "grad_norm": 0.5345031484238166, "learning_rate": 6.789378195503331e-05, "loss": 2.7601, "step": 21513 }, { "epoch": 1.3355267241914457, "grad_norm": 0.5762631877374839, "learning_rate": 6.789040957810494e-05, "loss": 2.7086, "step": 21514 }, { "epoch": 1.3355888012912036, "grad_norm": 0.5947442764625772, "learning_rate": 6.788703710783794e-05, "loss": 2.7728, "step": 21515 }, { "epoch": 1.3356508783909615, "grad_norm": 0.49396243293870445, "learning_rate": 6.788366454424994e-05, "loss": 2.7911, "step": 21516 }, { "epoch": 1.3357129554907194, "grad_norm": 0.6067970963280369, "learning_rate": 6.788029188735852e-05, "loss": 2.8125, "step": 21517 }, { "epoch": 1.3357750325904774, "grad_norm": 0.5348210363889275, "learning_rate": 6.787691913718125e-05, "loss": 2.9296, "step": 21518 }, { "epoch": 1.3358371096902353, "grad_norm": 0.5322524878110927, "learning_rate": 6.787354629373579e-05, "loss": 2.7247, "step": 21519 }, { "epoch": 1.3358991867899932, "grad_norm": 0.5465514724692068, "learning_rate": 6.787017335703967e-05, "loss": 2.6893, "step": 21520 }, { "epoch": 1.3359612638897511, "grad_norm": 0.5353558022109571, "learning_rate": 6.786680032711055e-05, "loss": 2.8132, "step": 21521 }, { "epoch": 1.336023340989509, "grad_norm": 0.5507686153613035, "learning_rate": 6.786342720396597e-05, "loss": 2.8065, "step": 21522 }, { "epoch": 1.336085418089267, "grad_norm": 0.5808454120653459, "learning_rate": 6.786005398762357e-05, "loss": 2.8172, "step": 21523 }, { "epoch": 1.3361474951890249, "grad_norm": 0.5666875206625575, "learning_rate": 6.785668067810092e-05, "loss": 2.8383, "step": 21524 }, { "epoch": 1.3362095722887828, "grad_norm": 0.5443310012292437, "learning_rate": 6.785330727541563e-05, "loss": 2.8311, "step": 21525 }, { "epoch": 1.3362716493885407, "grad_norm": 0.5034905130710491, "learning_rate": 6.784993377958532e-05, "loss": 2.8851, "step": 21526 }, { "epoch": 1.3363337264882984, "grad_norm": 0.5428025245079307, "learning_rate": 6.784656019062756e-05, "loss": 2.8307, "step": 21527 }, { "epoch": 1.3363958035880563, "grad_norm": 0.5203615308691222, "learning_rate": 6.784318650855995e-05, "loss": 2.8032, "step": 21528 }, { "epoch": 1.3364578806878142, "grad_norm": 0.4833730825845369, "learning_rate": 6.783981273340011e-05, "loss": 2.8128, "step": 21529 }, { "epoch": 1.3365199577875722, "grad_norm": 0.5948846916856892, "learning_rate": 6.783643886516564e-05, "loss": 2.8063, "step": 21530 }, { "epoch": 1.33658203488733, "grad_norm": 0.5342820817349978, "learning_rate": 6.783306490387414e-05, "loss": 2.7991, "step": 21531 }, { "epoch": 1.336644111987088, "grad_norm": 0.5508556450996293, "learning_rate": 6.782969084954321e-05, "loss": 2.8053, "step": 21532 }, { "epoch": 1.336706189086846, "grad_norm": 0.5270895451216118, "learning_rate": 6.782631670219047e-05, "loss": 2.7886, "step": 21533 }, { "epoch": 1.3367682661866038, "grad_norm": 0.5607774904751649, "learning_rate": 6.78229424618335e-05, "loss": 2.8293, "step": 21534 }, { "epoch": 1.3368303432863617, "grad_norm": 0.5187813071489921, "learning_rate": 6.781956812848991e-05, "loss": 2.8054, "step": 21535 }, { "epoch": 1.3368924203861194, "grad_norm": 0.5873755800106727, "learning_rate": 6.781619370217729e-05, "loss": 2.8346, "step": 21536 }, { "epoch": 1.3369544974858774, "grad_norm": 0.5819288586153243, "learning_rate": 6.78128191829133e-05, "loss": 2.8659, "step": 21537 }, { "epoch": 1.3370165745856353, "grad_norm": 0.523574953393552, "learning_rate": 6.780944457071547e-05, "loss": 2.7737, "step": 21538 }, { "epoch": 1.3370786516853932, "grad_norm": 0.6371605399407058, "learning_rate": 6.780606986560146e-05, "loss": 2.7775, "step": 21539 }, { "epoch": 1.3371407287851511, "grad_norm": 0.5543910832958908, "learning_rate": 6.780269506758887e-05, "loss": 2.8191, "step": 21540 }, { "epoch": 1.337202805884909, "grad_norm": 0.5752002087629176, "learning_rate": 6.779932017669528e-05, "loss": 2.8403, "step": 21541 }, { "epoch": 1.337264882984667, "grad_norm": 0.562863947430082, "learning_rate": 6.779594519293833e-05, "loss": 2.8175, "step": 21542 }, { "epoch": 1.3373269600844249, "grad_norm": 0.5337516003837088, "learning_rate": 6.77925701163356e-05, "loss": 2.8161, "step": 21543 }, { "epoch": 1.3373890371841828, "grad_norm": 0.48602369654050803, "learning_rate": 6.778919494690473e-05, "loss": 2.7403, "step": 21544 }, { "epoch": 1.3374511142839407, "grad_norm": 0.5425426265564606, "learning_rate": 6.778581968466329e-05, "loss": 2.8521, "step": 21545 }, { "epoch": 1.3375131913836986, "grad_norm": 0.5377168583874627, "learning_rate": 6.778244432962893e-05, "loss": 2.8127, "step": 21546 }, { "epoch": 1.3375752684834565, "grad_norm": 0.5865360508901695, "learning_rate": 6.777906888181922e-05, "loss": 2.8018, "step": 21547 }, { "epoch": 1.3376373455832145, "grad_norm": 0.5602337165161569, "learning_rate": 6.777569334125182e-05, "loss": 2.7819, "step": 21548 }, { "epoch": 1.3376994226829724, "grad_norm": 0.5566898524953204, "learning_rate": 6.77723177079443e-05, "loss": 2.8217, "step": 21549 }, { "epoch": 1.33776149978273, "grad_norm": 0.5889711304083832, "learning_rate": 6.776894198191428e-05, "loss": 2.8067, "step": 21550 }, { "epoch": 1.337823576882488, "grad_norm": 0.619538116417027, "learning_rate": 6.776556616317938e-05, "loss": 2.8022, "step": 21551 }, { "epoch": 1.337885653982246, "grad_norm": 0.5452437458580414, "learning_rate": 6.776219025175721e-05, "loss": 2.8049, "step": 21552 }, { "epoch": 1.3379477310820038, "grad_norm": 0.576177409517248, "learning_rate": 6.775881424766538e-05, "loss": 2.8104, "step": 21553 }, { "epoch": 1.3380098081817617, "grad_norm": 0.5862208634885497, "learning_rate": 6.77554381509215e-05, "loss": 2.8732, "step": 21554 }, { "epoch": 1.3380718852815197, "grad_norm": 0.5246713245254565, "learning_rate": 6.775206196154321e-05, "loss": 2.8364, "step": 21555 }, { "epoch": 1.3381339623812776, "grad_norm": 0.6448271332345185, "learning_rate": 6.774868567954807e-05, "loss": 2.848, "step": 21556 }, { "epoch": 1.3381960394810355, "grad_norm": 0.5255258095552613, "learning_rate": 6.774530930495376e-05, "loss": 2.842, "step": 21557 }, { "epoch": 1.3382581165807934, "grad_norm": 0.6124364051112223, "learning_rate": 6.774193283777784e-05, "loss": 2.7892, "step": 21558 }, { "epoch": 1.338320193680551, "grad_norm": 0.5766466156922654, "learning_rate": 6.773855627803795e-05, "loss": 2.888, "step": 21559 }, { "epoch": 1.338382270780309, "grad_norm": 0.5619720789686615, "learning_rate": 6.77351796257517e-05, "loss": 2.9137, "step": 21560 }, { "epoch": 1.338444347880067, "grad_norm": 0.5426483032013746, "learning_rate": 6.773180288093673e-05, "loss": 2.8266, "step": 21561 }, { "epoch": 1.3385064249798249, "grad_norm": 0.5153946217371405, "learning_rate": 6.772842604361065e-05, "loss": 2.8619, "step": 21562 }, { "epoch": 1.3385685020795828, "grad_norm": 0.5656834166483415, "learning_rate": 6.772504911379104e-05, "loss": 2.8665, "step": 21563 }, { "epoch": 1.3386305791793407, "grad_norm": 0.6151175494007343, "learning_rate": 6.772167209149557e-05, "loss": 2.863, "step": 21564 }, { "epoch": 1.3386926562790986, "grad_norm": 0.6533707398253283, "learning_rate": 6.771829497674181e-05, "loss": 2.7431, "step": 21565 }, { "epoch": 1.3387547333788565, "grad_norm": 0.624257504975315, "learning_rate": 6.771491776954742e-05, "loss": 2.8449, "step": 21566 }, { "epoch": 1.3388168104786144, "grad_norm": 0.6721305804624876, "learning_rate": 6.771154046993e-05, "loss": 2.784, "step": 21567 }, { "epoch": 1.3388788875783724, "grad_norm": 0.5822253831546258, "learning_rate": 6.770816307790719e-05, "loss": 2.8448, "step": 21568 }, { "epoch": 1.3389409646781303, "grad_norm": 0.6581264190803436, "learning_rate": 6.770478559349656e-05, "loss": 2.8581, "step": 21569 }, { "epoch": 1.3390030417778882, "grad_norm": 0.6405473289855695, "learning_rate": 6.77014080167158e-05, "loss": 2.9253, "step": 21570 }, { "epoch": 1.3390651188776461, "grad_norm": 0.5932465225596856, "learning_rate": 6.769803034758248e-05, "loss": 2.8571, "step": 21571 }, { "epoch": 1.339127195977404, "grad_norm": 0.6070383216318357, "learning_rate": 6.769465258611424e-05, "loss": 2.8169, "step": 21572 }, { "epoch": 1.339189273077162, "grad_norm": 0.63875049725418, "learning_rate": 6.769127473232871e-05, "loss": 2.8634, "step": 21573 }, { "epoch": 1.3392513501769197, "grad_norm": 0.579168023192466, "learning_rate": 6.76878967862435e-05, "loss": 2.8228, "step": 21574 }, { "epoch": 1.3393134272766776, "grad_norm": 0.5913847189930898, "learning_rate": 6.768451874787624e-05, "loss": 2.811, "step": 21575 }, { "epoch": 1.3393755043764355, "grad_norm": 0.5571457027006318, "learning_rate": 6.768114061724454e-05, "loss": 2.8111, "step": 21576 }, { "epoch": 1.3394375814761934, "grad_norm": 0.5870270072940058, "learning_rate": 6.767776239436605e-05, "loss": 2.7919, "step": 21577 }, { "epoch": 1.3394996585759513, "grad_norm": 0.5428110346447673, "learning_rate": 6.767438407925836e-05, "loss": 2.8183, "step": 21578 }, { "epoch": 1.3395617356757092, "grad_norm": 0.6370314315934118, "learning_rate": 6.767100567193915e-05, "loss": 2.8597, "step": 21579 }, { "epoch": 1.3396238127754672, "grad_norm": 0.5173482996430349, "learning_rate": 6.7667627172426e-05, "loss": 2.7588, "step": 21580 }, { "epoch": 1.339685889875225, "grad_norm": 0.5844749762951943, "learning_rate": 6.766424858073654e-05, "loss": 2.8167, "step": 21581 }, { "epoch": 1.339747966974983, "grad_norm": 0.5719014333650649, "learning_rate": 6.766086989688842e-05, "loss": 2.7661, "step": 21582 }, { "epoch": 1.3398100440747407, "grad_norm": 0.5332698291234045, "learning_rate": 6.765749112089926e-05, "loss": 2.8474, "step": 21583 }, { "epoch": 1.3398721211744986, "grad_norm": 0.6011806613807996, "learning_rate": 6.765411225278667e-05, "loss": 2.8725, "step": 21584 }, { "epoch": 1.3399341982742565, "grad_norm": 0.505553959735103, "learning_rate": 6.765073329256829e-05, "loss": 2.8243, "step": 21585 }, { "epoch": 1.3399962753740144, "grad_norm": 0.544804030497524, "learning_rate": 6.764735424026176e-05, "loss": 2.8103, "step": 21586 }, { "epoch": 1.3400583524737724, "grad_norm": 0.538380370958904, "learning_rate": 6.76439750958847e-05, "loss": 2.8522, "step": 21587 }, { "epoch": 1.3401204295735303, "grad_norm": 0.5928313024391533, "learning_rate": 6.764059585945472e-05, "loss": 2.7756, "step": 21588 }, { "epoch": 1.3401825066732882, "grad_norm": 0.49401458821590727, "learning_rate": 6.763721653098949e-05, "loss": 2.8415, "step": 21589 }, { "epoch": 1.3402445837730461, "grad_norm": 0.6344544337484265, "learning_rate": 6.76338371105066e-05, "loss": 2.8739, "step": 21590 }, { "epoch": 1.340306660872804, "grad_norm": 0.5341852386576709, "learning_rate": 6.763045759802372e-05, "loss": 2.7631, "step": 21591 }, { "epoch": 1.340368737972562, "grad_norm": 0.5766929985339073, "learning_rate": 6.762707799355845e-05, "loss": 2.7881, "step": 21592 }, { "epoch": 1.3404308150723199, "grad_norm": 0.5445018503141885, "learning_rate": 6.762369829712844e-05, "loss": 2.8423, "step": 21593 }, { "epoch": 1.3404928921720778, "grad_norm": 0.5470175693504321, "learning_rate": 6.762031850875132e-05, "loss": 2.8244, "step": 21594 }, { "epoch": 1.3405549692718357, "grad_norm": 0.5354793158168615, "learning_rate": 6.761693862844471e-05, "loss": 2.7789, "step": 21595 }, { "epoch": 1.3406170463715936, "grad_norm": 0.6148605421940804, "learning_rate": 6.761355865622628e-05, "loss": 2.9225, "step": 21596 }, { "epoch": 1.3406791234713515, "grad_norm": 0.5465064032738786, "learning_rate": 6.76101785921136e-05, "loss": 2.7692, "step": 21597 }, { "epoch": 1.3407412005711092, "grad_norm": 0.5413293307711153, "learning_rate": 6.760679843612438e-05, "loss": 2.8655, "step": 21598 }, { "epoch": 1.3408032776708672, "grad_norm": 0.5489925260469358, "learning_rate": 6.760341818827619e-05, "loss": 2.8586, "step": 21599 }, { "epoch": 1.340865354770625, "grad_norm": 0.6393809916941741, "learning_rate": 6.760003784858671e-05, "loss": 2.8516, "step": 21600 }, { "epoch": 1.340927431870383, "grad_norm": 0.6132116338980546, "learning_rate": 6.759665741707355e-05, "loss": 2.8417, "step": 21601 }, { "epoch": 1.340989508970141, "grad_norm": 0.6173427724480522, "learning_rate": 6.759327689375437e-05, "loss": 2.7388, "step": 21602 }, { "epoch": 1.3410515860698988, "grad_norm": 0.5505462848719171, "learning_rate": 6.758989627864678e-05, "loss": 2.7886, "step": 21603 }, { "epoch": 1.3411136631696567, "grad_norm": 0.563746263825751, "learning_rate": 6.758651557176845e-05, "loss": 2.7932, "step": 21604 }, { "epoch": 1.3411757402694147, "grad_norm": 0.5975710739575554, "learning_rate": 6.758313477313698e-05, "loss": 2.8725, "step": 21605 }, { "epoch": 1.3412378173691726, "grad_norm": 0.5307328006618723, "learning_rate": 6.757975388277004e-05, "loss": 2.836, "step": 21606 }, { "epoch": 1.3412998944689303, "grad_norm": 0.6598647337351744, "learning_rate": 6.757637290068526e-05, "loss": 2.8608, "step": 21607 }, { "epoch": 1.3413619715686882, "grad_norm": 0.5585517867538754, "learning_rate": 6.757299182690026e-05, "loss": 2.7247, "step": 21608 }, { "epoch": 1.3414240486684461, "grad_norm": 0.6003847110732519, "learning_rate": 6.75696106614327e-05, "loss": 2.7969, "step": 21609 }, { "epoch": 1.341486125768204, "grad_norm": 0.5419698349750971, "learning_rate": 6.75662294043002e-05, "loss": 2.8578, "step": 21610 }, { "epoch": 1.341548202867962, "grad_norm": 0.573076136387387, "learning_rate": 6.756284805552045e-05, "loss": 2.8164, "step": 21611 }, { "epoch": 1.3416102799677199, "grad_norm": 0.504207373255195, "learning_rate": 6.755946661511103e-05, "loss": 2.8686, "step": 21612 }, { "epoch": 1.3416723570674778, "grad_norm": 0.5620601046602459, "learning_rate": 6.755608508308962e-05, "loss": 2.7542, "step": 21613 }, { "epoch": 1.3417344341672357, "grad_norm": 0.5695893131166384, "learning_rate": 6.755270345947387e-05, "loss": 2.753, "step": 21614 }, { "epoch": 1.3417965112669936, "grad_norm": 0.5008869053734983, "learning_rate": 6.754932174428138e-05, "loss": 2.8471, "step": 21615 }, { "epoch": 1.3418585883667515, "grad_norm": 0.6181030966539364, "learning_rate": 6.754593993752983e-05, "loss": 2.8757, "step": 21616 }, { "epoch": 1.3419206654665095, "grad_norm": 0.5191981746558473, "learning_rate": 6.754255803923685e-05, "loss": 2.7661, "step": 21617 }, { "epoch": 1.3419827425662674, "grad_norm": 0.5567642342366983, "learning_rate": 6.75391760494201e-05, "loss": 2.8119, "step": 21618 }, { "epoch": 1.3420448196660253, "grad_norm": 0.5642619089603107, "learning_rate": 6.753579396809718e-05, "loss": 2.7692, "step": 21619 }, { "epoch": 1.3421068967657832, "grad_norm": 0.5200730560408701, "learning_rate": 6.75324117952858e-05, "loss": 2.8064, "step": 21620 }, { "epoch": 1.3421689738655411, "grad_norm": 0.5495920261228698, "learning_rate": 6.752902953100353e-05, "loss": 2.7648, "step": 21621 }, { "epoch": 1.3422310509652988, "grad_norm": 0.5295712368731905, "learning_rate": 6.75256471752681e-05, "loss": 2.7374, "step": 21622 }, { "epoch": 1.3422931280650567, "grad_norm": 0.5586463796724485, "learning_rate": 6.752226472809709e-05, "loss": 2.8192, "step": 21623 }, { "epoch": 1.3423552051648147, "grad_norm": 0.5963062172704571, "learning_rate": 6.751888218950818e-05, "loss": 2.8568, "step": 21624 }, { "epoch": 1.3424172822645726, "grad_norm": 0.5295950130862436, "learning_rate": 6.751549955951902e-05, "loss": 2.7815, "step": 21625 }, { "epoch": 1.3424793593643305, "grad_norm": 0.5704637091859832, "learning_rate": 6.751211683814722e-05, "loss": 2.8626, "step": 21626 }, { "epoch": 1.3425414364640884, "grad_norm": 0.5673400284989951, "learning_rate": 6.750873402541047e-05, "loss": 2.8404, "step": 21627 }, { "epoch": 1.3426035135638463, "grad_norm": 0.5437927854081832, "learning_rate": 6.750535112132639e-05, "loss": 2.8752, "step": 21628 }, { "epoch": 1.3426655906636042, "grad_norm": 0.6406451431457104, "learning_rate": 6.750196812591266e-05, "loss": 2.7701, "step": 21629 }, { "epoch": 1.3427276677633622, "grad_norm": 0.5109193396184427, "learning_rate": 6.74985850391869e-05, "loss": 2.839, "step": 21630 }, { "epoch": 1.3427897448631199, "grad_norm": 0.6156736712894607, "learning_rate": 6.74952018611668e-05, "loss": 2.8875, "step": 21631 }, { "epoch": 1.3428518219628778, "grad_norm": 0.4607600824326134, "learning_rate": 6.749181859186997e-05, "loss": 2.7359, "step": 21632 }, { "epoch": 1.3429138990626357, "grad_norm": 0.5752537281573731, "learning_rate": 6.748843523131406e-05, "loss": 2.8159, "step": 21633 }, { "epoch": 1.3429759761623936, "grad_norm": 0.5801516142767051, "learning_rate": 6.748505177951676e-05, "loss": 2.7619, "step": 21634 }, { "epoch": 1.3430380532621515, "grad_norm": 0.6375781852797863, "learning_rate": 6.748166823649569e-05, "loss": 2.8225, "step": 21635 }, { "epoch": 1.3431001303619095, "grad_norm": 0.6048408698227978, "learning_rate": 6.747828460226853e-05, "loss": 2.8745, "step": 21636 }, { "epoch": 1.3431622074616674, "grad_norm": 0.6605735453089067, "learning_rate": 6.747490087685289e-05, "loss": 2.7906, "step": 21637 }, { "epoch": 1.3432242845614253, "grad_norm": 0.5637606428626178, "learning_rate": 6.747151706026646e-05, "loss": 2.8412, "step": 21638 }, { "epoch": 1.3432863616611832, "grad_norm": 0.5934836894656743, "learning_rate": 6.746813315252689e-05, "loss": 2.8121, "step": 21639 }, { "epoch": 1.3433484387609411, "grad_norm": 0.5761537452043698, "learning_rate": 6.746474915365182e-05, "loss": 2.7684, "step": 21640 }, { "epoch": 1.343410515860699, "grad_norm": 0.648339873168963, "learning_rate": 6.746136506365892e-05, "loss": 2.774, "step": 21641 }, { "epoch": 1.343472592960457, "grad_norm": 0.5793917670184804, "learning_rate": 6.745798088256585e-05, "loss": 2.8631, "step": 21642 }, { "epoch": 1.3435346700602149, "grad_norm": 0.6183677928204598, "learning_rate": 6.745459661039024e-05, "loss": 2.8402, "step": 21643 }, { "epoch": 1.3435967471599728, "grad_norm": 0.5657821373773214, "learning_rate": 6.745121224714975e-05, "loss": 2.8103, "step": 21644 }, { "epoch": 1.3436588242597307, "grad_norm": 0.5589553622881166, "learning_rate": 6.744782779286207e-05, "loss": 2.8418, "step": 21645 }, { "epoch": 1.3437209013594884, "grad_norm": 0.6159118313330104, "learning_rate": 6.744444324754483e-05, "loss": 2.7974, "step": 21646 }, { "epoch": 1.3437829784592463, "grad_norm": 0.5558156911127162, "learning_rate": 6.74410586112157e-05, "loss": 2.9457, "step": 21647 }, { "epoch": 1.3438450555590042, "grad_norm": 0.5776488173081171, "learning_rate": 6.743767388389233e-05, "loss": 2.7878, "step": 21648 }, { "epoch": 1.3439071326587622, "grad_norm": 0.5773162855912314, "learning_rate": 6.743428906559238e-05, "loss": 2.9173, "step": 21649 }, { "epoch": 1.34396920975852, "grad_norm": 0.5459089192866684, "learning_rate": 6.743090415633351e-05, "loss": 2.7983, "step": 21650 }, { "epoch": 1.344031286858278, "grad_norm": 0.5315185877688845, "learning_rate": 6.742751915613338e-05, "loss": 2.7562, "step": 21651 }, { "epoch": 1.344093363958036, "grad_norm": 0.5768360512460103, "learning_rate": 6.742413406500967e-05, "loss": 2.8123, "step": 21652 }, { "epoch": 1.3441554410577938, "grad_norm": 0.5979393182732652, "learning_rate": 6.742074888298002e-05, "loss": 2.9096, "step": 21653 }, { "epoch": 1.3442175181575517, "grad_norm": 0.559882644379695, "learning_rate": 6.741736361006209e-05, "loss": 2.7792, "step": 21654 }, { "epoch": 1.3442795952573094, "grad_norm": 0.562625473651123, "learning_rate": 6.741397824627352e-05, "loss": 2.7355, "step": 21655 }, { "epoch": 1.3443416723570674, "grad_norm": 0.5610365373745511, "learning_rate": 6.741059279163202e-05, "loss": 2.7722, "step": 21656 }, { "epoch": 1.3444037494568253, "grad_norm": 0.5402569736434869, "learning_rate": 6.740720724615524e-05, "loss": 2.7755, "step": 21657 }, { "epoch": 1.3444658265565832, "grad_norm": 0.5872414131446181, "learning_rate": 6.740382160986081e-05, "loss": 2.8212, "step": 21658 }, { "epoch": 1.3445279036563411, "grad_norm": 0.578163686952842, "learning_rate": 6.740043588276643e-05, "loss": 2.9353, "step": 21659 }, { "epoch": 1.344589980756099, "grad_norm": 0.5687742783531181, "learning_rate": 6.739705006488975e-05, "loss": 2.8638, "step": 21660 }, { "epoch": 1.344652057855857, "grad_norm": 0.6175572941335499, "learning_rate": 6.739366415624845e-05, "loss": 2.7995, "step": 21661 }, { "epoch": 1.3447141349556149, "grad_norm": 0.5615730860778531, "learning_rate": 6.739027815686017e-05, "loss": 2.9035, "step": 21662 }, { "epoch": 1.3447762120553728, "grad_norm": 0.5962275963498922, "learning_rate": 6.738689206674259e-05, "loss": 2.8432, "step": 21663 }, { "epoch": 1.3448382891551307, "grad_norm": 0.5737915525993308, "learning_rate": 6.738350588591336e-05, "loss": 2.774, "step": 21664 }, { "epoch": 1.3449003662548886, "grad_norm": 0.6203920731325233, "learning_rate": 6.738011961439018e-05, "loss": 2.7277, "step": 21665 }, { "epoch": 1.3449624433546465, "grad_norm": 0.5640154398891349, "learning_rate": 6.737673325219068e-05, "loss": 2.7925, "step": 21666 }, { "epoch": 1.3450245204544045, "grad_norm": 0.588813184599483, "learning_rate": 6.737334679933256e-05, "loss": 2.8905, "step": 21667 }, { "epoch": 1.3450865975541624, "grad_norm": 0.5410410189952164, "learning_rate": 6.736996025583347e-05, "loss": 2.8185, "step": 21668 }, { "epoch": 1.3451486746539203, "grad_norm": 0.5555607007371217, "learning_rate": 6.736657362171106e-05, "loss": 2.7951, "step": 21669 }, { "epoch": 1.345210751753678, "grad_norm": 0.5299611164170601, "learning_rate": 6.736318689698305e-05, "loss": 2.7792, "step": 21670 }, { "epoch": 1.345272828853436, "grad_norm": 0.5505135068042006, "learning_rate": 6.735980008166706e-05, "loss": 2.836, "step": 21671 }, { "epoch": 1.3453349059531938, "grad_norm": 0.5256015113272486, "learning_rate": 6.735641317578079e-05, "loss": 2.8349, "step": 21672 }, { "epoch": 1.3453969830529517, "grad_norm": 0.5673590177447847, "learning_rate": 6.735302617934187e-05, "loss": 2.8797, "step": 21673 }, { "epoch": 1.3454590601527097, "grad_norm": 0.5337290144196386, "learning_rate": 6.734963909236804e-05, "loss": 2.8053, "step": 21674 }, { "epoch": 1.3455211372524676, "grad_norm": 0.5647186657866093, "learning_rate": 6.73462519148769e-05, "loss": 2.8461, "step": 21675 }, { "epoch": 1.3455832143522255, "grad_norm": 0.4952906919666446, "learning_rate": 6.734286464688616e-05, "loss": 2.7459, "step": 21676 }, { "epoch": 1.3456452914519834, "grad_norm": 0.4994708927358946, "learning_rate": 6.73394772884135e-05, "loss": 2.9142, "step": 21677 }, { "epoch": 1.3457073685517413, "grad_norm": 0.5346440558115674, "learning_rate": 6.733608983947656e-05, "loss": 2.8409, "step": 21678 }, { "epoch": 1.345769445651499, "grad_norm": 0.5304744332504895, "learning_rate": 6.733270230009304e-05, "loss": 2.8297, "step": 21679 }, { "epoch": 1.345831522751257, "grad_norm": 0.5219684311612794, "learning_rate": 6.73293146702806e-05, "loss": 2.7969, "step": 21680 }, { "epoch": 1.3458935998510149, "grad_norm": 0.5605624791542039, "learning_rate": 6.732592695005692e-05, "loss": 2.7764, "step": 21681 }, { "epoch": 1.3459556769507728, "grad_norm": 0.5464133630278128, "learning_rate": 6.732253913943968e-05, "loss": 2.7865, "step": 21682 }, { "epoch": 1.3460177540505307, "grad_norm": 0.5356471462411955, "learning_rate": 6.731915123844654e-05, "loss": 2.7888, "step": 21683 }, { "epoch": 1.3460798311502886, "grad_norm": 0.5723394144546463, "learning_rate": 6.73157632470952e-05, "loss": 2.8494, "step": 21684 }, { "epoch": 1.3461419082500465, "grad_norm": 0.5454469837401505, "learning_rate": 6.73123751654033e-05, "loss": 2.7575, "step": 21685 }, { "epoch": 1.3462039853498045, "grad_norm": 0.4881257992297088, "learning_rate": 6.730898699338855e-05, "loss": 2.8141, "step": 21686 }, { "epoch": 1.3462660624495624, "grad_norm": 0.5388198527569076, "learning_rate": 6.730559873106858e-05, "loss": 2.7924, "step": 21687 }, { "epoch": 1.3463281395493203, "grad_norm": 0.5199546809347007, "learning_rate": 6.730221037846114e-05, "loss": 2.7479, "step": 21688 }, { "epoch": 1.3463902166490782, "grad_norm": 0.5312064938401795, "learning_rate": 6.729882193558385e-05, "loss": 2.8283, "step": 21689 }, { "epoch": 1.3464522937488361, "grad_norm": 0.552691993840916, "learning_rate": 6.729543340245441e-05, "loss": 2.8187, "step": 21690 }, { "epoch": 1.346514370848594, "grad_norm": 0.5990998778127062, "learning_rate": 6.72920447790905e-05, "loss": 2.8647, "step": 21691 }, { "epoch": 1.346576447948352, "grad_norm": 0.5168606661800594, "learning_rate": 6.728865606550978e-05, "loss": 2.8239, "step": 21692 }, { "epoch": 1.3466385250481099, "grad_norm": 0.6086129044736998, "learning_rate": 6.728526726172996e-05, "loss": 2.8658, "step": 21693 }, { "epoch": 1.3467006021478676, "grad_norm": 0.5309824830971671, "learning_rate": 6.728187836776869e-05, "loss": 2.8395, "step": 21694 }, { "epoch": 1.3467626792476255, "grad_norm": 0.5799289858440041, "learning_rate": 6.727848938364368e-05, "loss": 2.8003, "step": 21695 }, { "epoch": 1.3468247563473834, "grad_norm": 0.5593478041702862, "learning_rate": 6.727510030937258e-05, "loss": 2.7326, "step": 21696 }, { "epoch": 1.3468868334471413, "grad_norm": 0.5678514131135279, "learning_rate": 6.727171114497311e-05, "loss": 2.81, "step": 21697 }, { "epoch": 1.3469489105468992, "grad_norm": 0.5285435344265585, "learning_rate": 6.726832189046291e-05, "loss": 2.7445, "step": 21698 }, { "epoch": 1.3470109876466572, "grad_norm": 0.5788167732475636, "learning_rate": 6.72649325458597e-05, "loss": 2.917, "step": 21699 }, { "epoch": 1.347073064746415, "grad_norm": 0.5824841267137973, "learning_rate": 6.726154311118114e-05, "loss": 2.8321, "step": 21700 }, { "epoch": 1.347135141846173, "grad_norm": 0.5874958677349456, "learning_rate": 6.725815358644491e-05, "loss": 2.8011, "step": 21701 }, { "epoch": 1.347197218945931, "grad_norm": 0.5416465413193983, "learning_rate": 6.72547639716687e-05, "loss": 2.8163, "step": 21702 }, { "epoch": 1.3472592960456886, "grad_norm": 0.5170589218043504, "learning_rate": 6.725137426687022e-05, "loss": 2.8, "step": 21703 }, { "epoch": 1.3473213731454465, "grad_norm": 0.5816509105011208, "learning_rate": 6.724798447206713e-05, "loss": 2.7642, "step": 21704 }, { "epoch": 1.3473834502452045, "grad_norm": 0.6171250372073985, "learning_rate": 6.72445945872771e-05, "loss": 2.866, "step": 21705 }, { "epoch": 1.3474455273449624, "grad_norm": 0.5809067994737497, "learning_rate": 6.724120461251784e-05, "loss": 2.7357, "step": 21706 }, { "epoch": 1.3475076044447203, "grad_norm": 0.5747332700639879, "learning_rate": 6.723781454780702e-05, "loss": 2.7863, "step": 21707 }, { "epoch": 1.3475696815444782, "grad_norm": 0.6058341861606911, "learning_rate": 6.723442439316236e-05, "loss": 2.8266, "step": 21708 }, { "epoch": 1.3476317586442361, "grad_norm": 0.58542155120097, "learning_rate": 6.72310341486015e-05, "loss": 2.8688, "step": 21709 }, { "epoch": 1.347693835743994, "grad_norm": 0.5963114399714134, "learning_rate": 6.722764381414217e-05, "loss": 2.7941, "step": 21710 }, { "epoch": 1.347755912843752, "grad_norm": 0.6844594433042962, "learning_rate": 6.722425338980204e-05, "loss": 2.8449, "step": 21711 }, { "epoch": 1.3478179899435099, "grad_norm": 0.5802417357422036, "learning_rate": 6.722086287559878e-05, "loss": 2.7749, "step": 21712 }, { "epoch": 1.3478800670432678, "grad_norm": 0.531622279381086, "learning_rate": 6.721747227155012e-05, "loss": 2.7268, "step": 21713 }, { "epoch": 1.3479421441430257, "grad_norm": 0.5996092010398388, "learning_rate": 6.72140815776737e-05, "loss": 2.6997, "step": 21714 }, { "epoch": 1.3480042212427836, "grad_norm": 0.5243241785403775, "learning_rate": 6.721069079398727e-05, "loss": 2.7944, "step": 21715 }, { "epoch": 1.3480662983425415, "grad_norm": 0.5872953575923142, "learning_rate": 6.720729992050847e-05, "loss": 2.8549, "step": 21716 }, { "epoch": 1.3481283754422995, "grad_norm": 0.5714967268719062, "learning_rate": 6.720390895725502e-05, "loss": 2.8466, "step": 21717 }, { "epoch": 1.3481904525420572, "grad_norm": 0.5742552868903206, "learning_rate": 6.720051790424459e-05, "loss": 2.8419, "step": 21718 }, { "epoch": 1.348252529641815, "grad_norm": 0.6171473236527862, "learning_rate": 6.719712676149488e-05, "loss": 2.7382, "step": 21719 }, { "epoch": 1.348314606741573, "grad_norm": 0.5776801072864832, "learning_rate": 6.71937355290236e-05, "loss": 2.833, "step": 21720 }, { "epoch": 1.348376683841331, "grad_norm": 0.6704129568376272, "learning_rate": 6.719034420684842e-05, "loss": 2.8207, "step": 21721 }, { "epoch": 1.3484387609410888, "grad_norm": 0.5686295119714754, "learning_rate": 6.718695279498705e-05, "loss": 2.8082, "step": 21722 }, { "epoch": 1.3485008380408467, "grad_norm": 0.6871689519595162, "learning_rate": 6.718356129345716e-05, "loss": 2.7812, "step": 21723 }, { "epoch": 1.3485629151406047, "grad_norm": 0.5673108736481174, "learning_rate": 6.718016970227646e-05, "loss": 2.8371, "step": 21724 }, { "epoch": 1.3486249922403626, "grad_norm": 0.6564611821751918, "learning_rate": 6.717677802146263e-05, "loss": 2.76, "step": 21725 }, { "epoch": 1.3486870693401205, "grad_norm": 0.537381994245843, "learning_rate": 6.717338625103341e-05, "loss": 2.8254, "step": 21726 }, { "epoch": 1.3487491464398782, "grad_norm": 0.6265102735713697, "learning_rate": 6.716999439100645e-05, "loss": 2.8366, "step": 21727 }, { "epoch": 1.3488112235396361, "grad_norm": 0.5389714371432914, "learning_rate": 6.716660244139946e-05, "loss": 2.8207, "step": 21728 }, { "epoch": 1.348873300639394, "grad_norm": 0.54901980382259, "learning_rate": 6.716321040223015e-05, "loss": 2.8683, "step": 21729 }, { "epoch": 1.348935377739152, "grad_norm": 0.5677380656118084, "learning_rate": 6.715981827351618e-05, "loss": 2.8786, "step": 21730 }, { "epoch": 1.3489974548389099, "grad_norm": 0.5559867096468583, "learning_rate": 6.715642605527531e-05, "loss": 2.8218, "step": 21731 }, { "epoch": 1.3490595319386678, "grad_norm": 0.5052366805890366, "learning_rate": 6.715303374752516e-05, "loss": 2.8159, "step": 21732 }, { "epoch": 1.3491216090384257, "grad_norm": 0.5274953271107455, "learning_rate": 6.71496413502835e-05, "loss": 2.845, "step": 21733 }, { "epoch": 1.3491836861381836, "grad_norm": 0.5413103092943398, "learning_rate": 6.714624886356797e-05, "loss": 2.8049, "step": 21734 }, { "epoch": 1.3492457632379415, "grad_norm": 0.5155638456203341, "learning_rate": 6.714285628739632e-05, "loss": 2.7335, "step": 21735 }, { "epoch": 1.3493078403376995, "grad_norm": 0.5203431174861075, "learning_rate": 6.713946362178622e-05, "loss": 2.6993, "step": 21736 }, { "epoch": 1.3493699174374574, "grad_norm": 0.5383809245137636, "learning_rate": 6.713607086675539e-05, "loss": 2.8045, "step": 21737 }, { "epoch": 1.3494319945372153, "grad_norm": 0.5612880578338384, "learning_rate": 6.71326780223215e-05, "loss": 2.8227, "step": 21738 }, { "epoch": 1.3494940716369732, "grad_norm": 0.5507022719877891, "learning_rate": 6.712928508850227e-05, "loss": 2.8658, "step": 21739 }, { "epoch": 1.3495561487367311, "grad_norm": 0.6106908968818412, "learning_rate": 6.712589206531541e-05, "loss": 2.7544, "step": 21740 }, { "epoch": 1.349618225836489, "grad_norm": 0.5059844933798797, "learning_rate": 6.71224989527786e-05, "loss": 2.7979, "step": 21741 }, { "epoch": 1.3496803029362467, "grad_norm": 0.5849382399293377, "learning_rate": 6.711910575090957e-05, "loss": 2.7826, "step": 21742 }, { "epoch": 1.3497423800360047, "grad_norm": 0.5449611417233933, "learning_rate": 6.7115712459726e-05, "loss": 2.845, "step": 21743 }, { "epoch": 1.3498044571357626, "grad_norm": 0.5334985215203232, "learning_rate": 6.711231907924563e-05, "loss": 2.8177, "step": 21744 }, { "epoch": 1.3498665342355205, "grad_norm": 0.528551935836196, "learning_rate": 6.710892560948611e-05, "loss": 2.7965, "step": 21745 }, { "epoch": 1.3499286113352784, "grad_norm": 0.5181973477919597, "learning_rate": 6.710553205046518e-05, "loss": 2.7908, "step": 21746 }, { "epoch": 1.3499906884350363, "grad_norm": 0.5633073417133669, "learning_rate": 6.710213840220054e-05, "loss": 2.9579, "step": 21747 }, { "epoch": 1.3500527655347943, "grad_norm": 0.49636369404516956, "learning_rate": 6.709874466470987e-05, "loss": 2.9035, "step": 21748 }, { "epoch": 1.3501148426345522, "grad_norm": 0.49947642689239546, "learning_rate": 6.709535083801093e-05, "loss": 2.7974, "step": 21749 }, { "epoch": 1.35017691973431, "grad_norm": 0.5043190736265376, "learning_rate": 6.709195692212137e-05, "loss": 2.8805, "step": 21750 }, { "epoch": 1.3502389968340678, "grad_norm": 0.5218749565992509, "learning_rate": 6.708856291705893e-05, "loss": 2.8029, "step": 21751 }, { "epoch": 1.3503010739338257, "grad_norm": 0.5756997453204968, "learning_rate": 6.708516882284131e-05, "loss": 2.758, "step": 21752 }, { "epoch": 1.3503631510335836, "grad_norm": 0.5328271958905798, "learning_rate": 6.708177463948623e-05, "loss": 2.851, "step": 21753 }, { "epoch": 1.3504252281333415, "grad_norm": 0.5337771724600796, "learning_rate": 6.707838036701138e-05, "loss": 2.774, "step": 21754 }, { "epoch": 1.3504873052330995, "grad_norm": 0.5324189004793991, "learning_rate": 6.707498600543446e-05, "loss": 2.8132, "step": 21755 }, { "epoch": 1.3505493823328574, "grad_norm": 0.5477554862628197, "learning_rate": 6.707159155477319e-05, "loss": 2.8965, "step": 21756 }, { "epoch": 1.3506114594326153, "grad_norm": 0.5305944774103034, "learning_rate": 6.70681970150453e-05, "loss": 2.8496, "step": 21757 }, { "epoch": 1.3506735365323732, "grad_norm": 0.4846602338380824, "learning_rate": 6.706480238626846e-05, "loss": 2.7677, "step": 21758 }, { "epoch": 1.3507356136321311, "grad_norm": 0.510508584160513, "learning_rate": 6.706140766846042e-05, "loss": 2.7677, "step": 21759 }, { "epoch": 1.350797690731889, "grad_norm": 0.48327962012375325, "learning_rate": 6.705801286163887e-05, "loss": 2.8832, "step": 21760 }, { "epoch": 1.350859767831647, "grad_norm": 0.5129804127320464, "learning_rate": 6.705461796582153e-05, "loss": 2.8908, "step": 21761 }, { "epoch": 1.3509218449314049, "grad_norm": 0.5506068555546451, "learning_rate": 6.70512229810261e-05, "loss": 2.8498, "step": 21762 }, { "epoch": 1.3509839220311628, "grad_norm": 0.5307183971816032, "learning_rate": 6.704782790727031e-05, "loss": 2.8357, "step": 21763 }, { "epoch": 1.3510459991309207, "grad_norm": 0.6054488394139929, "learning_rate": 6.704443274457187e-05, "loss": 2.8272, "step": 21764 }, { "epoch": 1.3511080762306786, "grad_norm": 0.573232699418058, "learning_rate": 6.704103749294847e-05, "loss": 2.8365, "step": 21765 }, { "epoch": 1.3511701533304363, "grad_norm": 0.5657408126796439, "learning_rate": 6.703764215241786e-05, "loss": 2.8207, "step": 21766 }, { "epoch": 1.3512322304301942, "grad_norm": 0.6228827137661783, "learning_rate": 6.703424672299771e-05, "loss": 2.8953, "step": 21767 }, { "epoch": 1.3512943075299522, "grad_norm": 0.5975130695046865, "learning_rate": 6.703085120470577e-05, "loss": 2.8151, "step": 21768 }, { "epoch": 1.35135638462971, "grad_norm": 0.5138284422862016, "learning_rate": 6.702745559755977e-05, "loss": 2.8303, "step": 21769 }, { "epoch": 1.351418461729468, "grad_norm": 0.6444163826975075, "learning_rate": 6.702405990157736e-05, "loss": 2.8001, "step": 21770 }, { "epoch": 1.351480538829226, "grad_norm": 0.5318960020712925, "learning_rate": 6.702066411677632e-05, "loss": 2.7557, "step": 21771 }, { "epoch": 1.3515426159289838, "grad_norm": 0.585902581763907, "learning_rate": 6.701726824317432e-05, "loss": 2.8239, "step": 21772 }, { "epoch": 1.3516046930287418, "grad_norm": 0.5871298293397716, "learning_rate": 6.701387228078912e-05, "loss": 2.8168, "step": 21773 }, { "epoch": 1.3516667701284997, "grad_norm": 0.5919417150534857, "learning_rate": 6.701047622963842e-05, "loss": 2.8998, "step": 21774 }, { "epoch": 1.3517288472282574, "grad_norm": 0.5647032820285856, "learning_rate": 6.700708008973992e-05, "loss": 2.8958, "step": 21775 }, { "epoch": 1.3517909243280153, "grad_norm": 0.6144582060684456, "learning_rate": 6.700368386111137e-05, "loss": 2.8468, "step": 21776 }, { "epoch": 1.3518530014277732, "grad_norm": 0.5677049152609676, "learning_rate": 6.700028754377048e-05, "loss": 2.8225, "step": 21777 }, { "epoch": 1.3519150785275311, "grad_norm": 0.5808949483356626, "learning_rate": 6.699689113773495e-05, "loss": 2.8419, "step": 21778 }, { "epoch": 1.351977155627289, "grad_norm": 0.6087594713809782, "learning_rate": 6.699349464302252e-05, "loss": 2.7962, "step": 21779 }, { "epoch": 1.352039232727047, "grad_norm": 0.6061264157914934, "learning_rate": 6.69900980596509e-05, "loss": 2.8531, "step": 21780 }, { "epoch": 1.3521013098268049, "grad_norm": 0.6637316833454447, "learning_rate": 6.69867013876378e-05, "loss": 2.86, "step": 21781 }, { "epoch": 1.3521633869265628, "grad_norm": 0.6606107422617312, "learning_rate": 6.698330462700097e-05, "loss": 2.7947, "step": 21782 }, { "epoch": 1.3522254640263207, "grad_norm": 0.6006061333462867, "learning_rate": 6.697990777775811e-05, "loss": 2.8, "step": 21783 }, { "epoch": 1.3522875411260786, "grad_norm": 0.7488317928313929, "learning_rate": 6.697651083992696e-05, "loss": 2.7463, "step": 21784 }, { "epoch": 1.3523496182258365, "grad_norm": 0.5638519041912308, "learning_rate": 6.697311381352524e-05, "loss": 2.8043, "step": 21785 }, { "epoch": 1.3524116953255945, "grad_norm": 0.6101992241529111, "learning_rate": 6.696971669857064e-05, "loss": 2.8608, "step": 21786 }, { "epoch": 1.3524737724253524, "grad_norm": 0.5975263616500489, "learning_rate": 6.696631949508093e-05, "loss": 2.7568, "step": 21787 }, { "epoch": 1.3525358495251103, "grad_norm": 0.5778735361412828, "learning_rate": 6.696292220307378e-05, "loss": 2.8243, "step": 21788 }, { "epoch": 1.3525979266248682, "grad_norm": 0.5091298256337958, "learning_rate": 6.695952482256698e-05, "loss": 2.8382, "step": 21789 }, { "epoch": 1.352660003724626, "grad_norm": 0.5774462814034939, "learning_rate": 6.695612735357822e-05, "loss": 2.7663, "step": 21790 }, { "epoch": 1.3527220808243838, "grad_norm": 0.5798850717625816, "learning_rate": 6.69527297961252e-05, "loss": 2.8276, "step": 21791 }, { "epoch": 1.3527841579241418, "grad_norm": 0.6068885890881072, "learning_rate": 6.69493321502257e-05, "loss": 2.9362, "step": 21792 }, { "epoch": 1.3528462350238997, "grad_norm": 0.6378539094093885, "learning_rate": 6.69459344158974e-05, "loss": 2.8657, "step": 21793 }, { "epoch": 1.3529083121236576, "grad_norm": 0.609037868457726, "learning_rate": 6.694253659315806e-05, "loss": 2.7677, "step": 21794 }, { "epoch": 1.3529703892234155, "grad_norm": 0.6498199112544639, "learning_rate": 6.69391386820254e-05, "loss": 2.8429, "step": 21795 }, { "epoch": 1.3530324663231734, "grad_norm": 0.671739276218318, "learning_rate": 6.693574068251712e-05, "loss": 2.8078, "step": 21796 }, { "epoch": 1.3530945434229313, "grad_norm": 0.5658390493717538, "learning_rate": 6.693234259465097e-05, "loss": 2.8124, "step": 21797 }, { "epoch": 1.3531566205226893, "grad_norm": 0.6991561137014144, "learning_rate": 6.69289444184447e-05, "loss": 2.7147, "step": 21798 }, { "epoch": 1.353218697622447, "grad_norm": 0.6341615825262135, "learning_rate": 6.692554615391601e-05, "loss": 2.8187, "step": 21799 }, { "epoch": 1.3532807747222049, "grad_norm": 0.6426024968589409, "learning_rate": 6.692214780108263e-05, "loss": 2.7767, "step": 21800 }, { "epoch": 1.3533428518219628, "grad_norm": 0.6546212374491839, "learning_rate": 6.691874935996229e-05, "loss": 2.7945, "step": 21801 }, { "epoch": 1.3534049289217207, "grad_norm": 0.643134829647175, "learning_rate": 6.691535083057276e-05, "loss": 2.8435, "step": 21802 }, { "epoch": 1.3534670060214786, "grad_norm": 0.7245999564409572, "learning_rate": 6.691195221293171e-05, "loss": 2.8748, "step": 21803 }, { "epoch": 1.3535290831212365, "grad_norm": 0.7241442233588969, "learning_rate": 6.69085535070569e-05, "loss": 2.8157, "step": 21804 }, { "epoch": 1.3535911602209945, "grad_norm": 0.6332350368094845, "learning_rate": 6.690515471296608e-05, "loss": 2.861, "step": 21805 }, { "epoch": 1.3536532373207524, "grad_norm": 0.6632674899881242, "learning_rate": 6.690175583067695e-05, "loss": 2.8269, "step": 21806 }, { "epoch": 1.3537153144205103, "grad_norm": 0.5899714906318585, "learning_rate": 6.689835686020725e-05, "loss": 2.7082, "step": 21807 }, { "epoch": 1.3537773915202682, "grad_norm": 0.5384728897517396, "learning_rate": 6.689495780157472e-05, "loss": 2.7194, "step": 21808 }, { "epoch": 1.3538394686200261, "grad_norm": 0.6001271053627681, "learning_rate": 6.68915586547971e-05, "loss": 2.8694, "step": 21809 }, { "epoch": 1.353901545719784, "grad_norm": 0.6068394066914333, "learning_rate": 6.688815941989214e-05, "loss": 2.7869, "step": 21810 }, { "epoch": 1.353963622819542, "grad_norm": 0.5134037028523544, "learning_rate": 6.688476009687752e-05, "loss": 2.8125, "step": 21811 }, { "epoch": 1.3540256999192999, "grad_norm": 0.5366064317833188, "learning_rate": 6.688136068577102e-05, "loss": 2.8195, "step": 21812 }, { "epoch": 1.3540877770190578, "grad_norm": 0.5432528843247953, "learning_rate": 6.687796118659037e-05, "loss": 2.7727, "step": 21813 }, { "epoch": 1.3541498541188155, "grad_norm": 0.5939852599810092, "learning_rate": 6.68745615993533e-05, "loss": 2.8818, "step": 21814 }, { "epoch": 1.3542119312185734, "grad_norm": 0.5395935041679237, "learning_rate": 6.687116192407753e-05, "loss": 2.7819, "step": 21815 }, { "epoch": 1.3542740083183313, "grad_norm": 0.5331085456553906, "learning_rate": 6.686776216078083e-05, "loss": 2.8119, "step": 21816 }, { "epoch": 1.3543360854180893, "grad_norm": 0.5407087922622572, "learning_rate": 6.68643623094809e-05, "loss": 2.8054, "step": 21817 }, { "epoch": 1.3543981625178472, "grad_norm": 0.5293891516621326, "learning_rate": 6.686096237019552e-05, "loss": 2.7637, "step": 21818 }, { "epoch": 1.354460239617605, "grad_norm": 0.5416137473836686, "learning_rate": 6.685756234294238e-05, "loss": 2.7076, "step": 21819 }, { "epoch": 1.354522316717363, "grad_norm": 0.5726616436338713, "learning_rate": 6.685416222773927e-05, "loss": 2.769, "step": 21820 }, { "epoch": 1.354584393817121, "grad_norm": 0.5421187244515593, "learning_rate": 6.68507620246039e-05, "loss": 2.883, "step": 21821 }, { "epoch": 1.3546464709168788, "grad_norm": 0.6164776788477685, "learning_rate": 6.6847361733554e-05, "loss": 2.8338, "step": 21822 }, { "epoch": 1.3547085480166365, "grad_norm": 0.5180913719611846, "learning_rate": 6.684396135460733e-05, "loss": 2.7847, "step": 21823 }, { "epoch": 1.3547706251163945, "grad_norm": 0.609002953442332, "learning_rate": 6.684056088778162e-05, "loss": 2.7564, "step": 21824 }, { "epoch": 1.3548327022161524, "grad_norm": 0.5704882625271745, "learning_rate": 6.683716033309463e-05, "loss": 2.9921, "step": 21825 }, { "epoch": 1.3548947793159103, "grad_norm": 0.5725917212545002, "learning_rate": 6.683375969056407e-05, "loss": 2.7263, "step": 21826 }, { "epoch": 1.3549568564156682, "grad_norm": 0.5985490202017736, "learning_rate": 6.683035896020772e-05, "loss": 2.8752, "step": 21827 }, { "epoch": 1.3550189335154261, "grad_norm": 0.5909719981790674, "learning_rate": 6.68269581420433e-05, "loss": 2.8051, "step": 21828 }, { "epoch": 1.355081010615184, "grad_norm": 0.5725970301380879, "learning_rate": 6.682355723608852e-05, "loss": 2.7942, "step": 21829 }, { "epoch": 1.355143087714942, "grad_norm": 0.5962945222526568, "learning_rate": 6.68201562423612e-05, "loss": 2.8147, "step": 21830 }, { "epoch": 1.3552051648146999, "grad_norm": 0.5743582938735746, "learning_rate": 6.681675516087902e-05, "loss": 2.8107, "step": 21831 }, { "epoch": 1.3552672419144578, "grad_norm": 0.6181867694752073, "learning_rate": 6.681335399165975e-05, "loss": 2.8812, "step": 21832 }, { "epoch": 1.3553293190142157, "grad_norm": 0.6199898175972371, "learning_rate": 6.680995273472113e-05, "loss": 2.7887, "step": 21833 }, { "epoch": 1.3553913961139736, "grad_norm": 0.5463693188171425, "learning_rate": 6.680655139008092e-05, "loss": 2.8197, "step": 21834 }, { "epoch": 1.3554534732137316, "grad_norm": 0.6655392551265785, "learning_rate": 6.680314995775683e-05, "loss": 2.8465, "step": 21835 }, { "epoch": 1.3555155503134895, "grad_norm": 0.5263399531125518, "learning_rate": 6.679974843776665e-05, "loss": 2.8376, "step": 21836 }, { "epoch": 1.3555776274132474, "grad_norm": 0.6541176393890451, "learning_rate": 6.679634683012808e-05, "loss": 2.7744, "step": 21837 }, { "epoch": 1.355639704513005, "grad_norm": 0.52795470838567, "learning_rate": 6.679294513485889e-05, "loss": 2.8865, "step": 21838 }, { "epoch": 1.355701781612763, "grad_norm": 0.5635983287603076, "learning_rate": 6.678954335197686e-05, "loss": 2.8429, "step": 21839 }, { "epoch": 1.355763858712521, "grad_norm": 0.6196779632988753, "learning_rate": 6.678614148149966e-05, "loss": 2.7889, "step": 21840 }, { "epoch": 1.3558259358122788, "grad_norm": 0.5418654560009802, "learning_rate": 6.678273952344512e-05, "loss": 2.7882, "step": 21841 }, { "epoch": 1.3558880129120368, "grad_norm": 0.6485561124485801, "learning_rate": 6.677933747783094e-05, "loss": 2.8438, "step": 21842 }, { "epoch": 1.3559500900117947, "grad_norm": 0.5988757090316521, "learning_rate": 6.67759353446749e-05, "loss": 2.8536, "step": 21843 }, { "epoch": 1.3560121671115526, "grad_norm": 0.6794635579173504, "learning_rate": 6.67725331239947e-05, "loss": 2.8891, "step": 21844 }, { "epoch": 1.3560742442113105, "grad_norm": 0.6058633810337617, "learning_rate": 6.676913081580816e-05, "loss": 2.8385, "step": 21845 }, { "epoch": 1.3561363213110684, "grad_norm": 0.616052162765584, "learning_rate": 6.676572842013298e-05, "loss": 2.8154, "step": 21846 }, { "epoch": 1.3561983984108261, "grad_norm": 0.5656095755565745, "learning_rate": 6.676232593698692e-05, "loss": 2.8145, "step": 21847 }, { "epoch": 1.356260475510584, "grad_norm": 0.5956827879389033, "learning_rate": 6.675892336638773e-05, "loss": 2.784, "step": 21848 }, { "epoch": 1.356322552610342, "grad_norm": 0.5566578377187246, "learning_rate": 6.675552070835318e-05, "loss": 2.8819, "step": 21849 }, { "epoch": 1.3563846297100999, "grad_norm": 0.5938273178755559, "learning_rate": 6.675211796290102e-05, "loss": 2.8247, "step": 21850 }, { "epoch": 1.3564467068098578, "grad_norm": 0.5552143496198734, "learning_rate": 6.674871513004897e-05, "loss": 2.7941, "step": 21851 }, { "epoch": 1.3565087839096157, "grad_norm": 0.574318701939437, "learning_rate": 6.674531220981483e-05, "loss": 2.7907, "step": 21852 }, { "epoch": 1.3565708610093736, "grad_norm": 0.5532448234729637, "learning_rate": 6.674190920221631e-05, "loss": 2.9091, "step": 21853 }, { "epoch": 1.3566329381091315, "grad_norm": 0.5275652949884277, "learning_rate": 6.67385061072712e-05, "loss": 2.8108, "step": 21854 }, { "epoch": 1.3566950152088895, "grad_norm": 0.5695286419760353, "learning_rate": 6.673510292499724e-05, "loss": 2.762, "step": 21855 }, { "epoch": 1.3567570923086474, "grad_norm": 0.47818062402630174, "learning_rate": 6.673169965541217e-05, "loss": 2.8319, "step": 21856 }, { "epoch": 1.3568191694084053, "grad_norm": 0.5499447632275396, "learning_rate": 6.672829629853378e-05, "loss": 2.8641, "step": 21857 }, { "epoch": 1.3568812465081632, "grad_norm": 0.5163295152125703, "learning_rate": 6.672489285437979e-05, "loss": 2.9207, "step": 21858 }, { "epoch": 1.3569433236079211, "grad_norm": 0.550300956877955, "learning_rate": 6.6721489322968e-05, "loss": 2.7908, "step": 21859 }, { "epoch": 1.357005400707679, "grad_norm": 0.4841381539907611, "learning_rate": 6.671808570431611e-05, "loss": 2.7129, "step": 21860 }, { "epoch": 1.357067477807437, "grad_norm": 0.5392614633228663, "learning_rate": 6.671468199844193e-05, "loss": 2.7549, "step": 21861 }, { "epoch": 1.3571295549071947, "grad_norm": 0.5079035163889857, "learning_rate": 6.671127820536318e-05, "loss": 2.8454, "step": 21862 }, { "epoch": 1.3571916320069526, "grad_norm": 0.5067806323973842, "learning_rate": 6.670787432509766e-05, "loss": 2.7984, "step": 21863 }, { "epoch": 1.3572537091067105, "grad_norm": 0.5374407768890459, "learning_rate": 6.670447035766309e-05, "loss": 2.7909, "step": 21864 }, { "epoch": 1.3573157862064684, "grad_norm": 0.5495185717393843, "learning_rate": 6.670106630307723e-05, "loss": 2.7339, "step": 21865 }, { "epoch": 1.3573778633062263, "grad_norm": 0.5424116830968531, "learning_rate": 6.669766216135787e-05, "loss": 2.7591, "step": 21866 }, { "epoch": 1.3574399404059843, "grad_norm": 0.5543163293458269, "learning_rate": 6.669425793252275e-05, "loss": 2.902, "step": 21867 }, { "epoch": 1.3575020175057422, "grad_norm": 0.5832766323688561, "learning_rate": 6.669085361658963e-05, "loss": 2.8298, "step": 21868 }, { "epoch": 1.3575640946055, "grad_norm": 0.5026589563052031, "learning_rate": 6.668744921357627e-05, "loss": 2.8321, "step": 21869 }, { "epoch": 1.357626171705258, "grad_norm": 0.668590189653952, "learning_rate": 6.668404472350044e-05, "loss": 2.873, "step": 21870 }, { "epoch": 1.3576882488050157, "grad_norm": 0.5594785924707436, "learning_rate": 6.66806401463799e-05, "loss": 2.8173, "step": 21871 }, { "epoch": 1.3577503259047736, "grad_norm": 0.5763963940781621, "learning_rate": 6.66772354822324e-05, "loss": 2.9221, "step": 21872 }, { "epoch": 1.3578124030045315, "grad_norm": 0.5339877046873345, "learning_rate": 6.667383073107574e-05, "loss": 2.8859, "step": 21873 }, { "epoch": 1.3578744801042895, "grad_norm": 0.5191929798693464, "learning_rate": 6.667042589292764e-05, "loss": 2.8768, "step": 21874 }, { "epoch": 1.3579365572040474, "grad_norm": 0.5633783370537675, "learning_rate": 6.666702096780588e-05, "loss": 2.7416, "step": 21875 }, { "epoch": 1.3579986343038053, "grad_norm": 0.485369031503484, "learning_rate": 6.666361595572823e-05, "loss": 2.7171, "step": 21876 }, { "epoch": 1.3580607114035632, "grad_norm": 0.5585131020114787, "learning_rate": 6.666021085671246e-05, "loss": 2.7835, "step": 21877 }, { "epoch": 1.3581227885033211, "grad_norm": 0.5097342002799831, "learning_rate": 6.66568056707763e-05, "loss": 2.8281, "step": 21878 }, { "epoch": 1.358184865603079, "grad_norm": 0.5820792261614539, "learning_rate": 6.665340039793756e-05, "loss": 2.767, "step": 21879 }, { "epoch": 1.358246942702837, "grad_norm": 0.5642837217409573, "learning_rate": 6.664999503821398e-05, "loss": 2.8586, "step": 21880 }, { "epoch": 1.3583090198025949, "grad_norm": 0.5791431174076038, "learning_rate": 6.664658959162334e-05, "loss": 2.8495, "step": 21881 }, { "epoch": 1.3583710969023528, "grad_norm": 0.5564925682860344, "learning_rate": 6.66431840581834e-05, "loss": 2.782, "step": 21882 }, { "epoch": 1.3584331740021107, "grad_norm": 0.5595069037123157, "learning_rate": 6.663977843791193e-05, "loss": 2.8292, "step": 21883 }, { "epoch": 1.3584952511018686, "grad_norm": 0.5696232432308401, "learning_rate": 6.66363727308267e-05, "loss": 2.7917, "step": 21884 }, { "epoch": 1.3585573282016266, "grad_norm": 0.6258610992350747, "learning_rate": 6.663296693694546e-05, "loss": 2.818, "step": 21885 }, { "epoch": 1.3586194053013843, "grad_norm": 0.5801039924377008, "learning_rate": 6.662956105628601e-05, "loss": 2.7084, "step": 21886 }, { "epoch": 1.3586814824011422, "grad_norm": 0.6097563504009188, "learning_rate": 6.662615508886609e-05, "loss": 2.8087, "step": 21887 }, { "epoch": 1.3587435595009, "grad_norm": 0.5385551553017088, "learning_rate": 6.662274903470348e-05, "loss": 2.817, "step": 21888 }, { "epoch": 1.358805636600658, "grad_norm": 0.6750698645074976, "learning_rate": 6.661934289381597e-05, "loss": 2.8558, "step": 21889 }, { "epoch": 1.358867713700416, "grad_norm": 0.5512632601088165, "learning_rate": 6.66159366662213e-05, "loss": 2.8109, "step": 21890 }, { "epoch": 1.3589297908001738, "grad_norm": 0.5990484391036353, "learning_rate": 6.661253035193726e-05, "loss": 2.8011, "step": 21891 }, { "epoch": 1.3589918678999318, "grad_norm": 0.6216982652655015, "learning_rate": 6.660912395098161e-05, "loss": 2.7391, "step": 21892 }, { "epoch": 1.3590539449996897, "grad_norm": 0.6233252259311994, "learning_rate": 6.660571746337214e-05, "loss": 2.7901, "step": 21893 }, { "epoch": 1.3591160220994476, "grad_norm": 0.5576725012072522, "learning_rate": 6.66023108891266e-05, "loss": 2.8145, "step": 21894 }, { "epoch": 1.3591780991992053, "grad_norm": 0.5641226147348298, "learning_rate": 6.659890422826278e-05, "loss": 2.8831, "step": 21895 }, { "epoch": 1.3592401762989632, "grad_norm": 0.5861356781556124, "learning_rate": 6.659549748079844e-05, "loss": 2.821, "step": 21896 }, { "epoch": 1.3593022533987211, "grad_norm": 0.5662475100414107, "learning_rate": 6.659209064675138e-05, "loss": 2.8035, "step": 21897 }, { "epoch": 1.359364330498479, "grad_norm": 0.584818063141802, "learning_rate": 6.658868372613934e-05, "loss": 2.8285, "step": 21898 }, { "epoch": 1.359426407598237, "grad_norm": 0.5579619936612781, "learning_rate": 6.658527671898011e-05, "loss": 2.7605, "step": 21899 }, { "epoch": 1.3594884846979949, "grad_norm": 0.5087553221778012, "learning_rate": 6.658186962529147e-05, "loss": 2.8508, "step": 21900 }, { "epoch": 1.3595505617977528, "grad_norm": 0.5892319190253918, "learning_rate": 6.657846244509118e-05, "loss": 2.7892, "step": 21901 }, { "epoch": 1.3596126388975107, "grad_norm": 0.5289292977071335, "learning_rate": 6.657505517839704e-05, "loss": 2.7937, "step": 21902 }, { "epoch": 1.3596747159972686, "grad_norm": 0.5690084729426959, "learning_rate": 6.65716478252268e-05, "loss": 2.7405, "step": 21903 }, { "epoch": 1.3597367930970266, "grad_norm": 0.5755995413237593, "learning_rate": 6.656824038559827e-05, "loss": 2.8144, "step": 21904 }, { "epoch": 1.3597988701967845, "grad_norm": 0.5875588681778598, "learning_rate": 6.656483285952918e-05, "loss": 2.7596, "step": 21905 }, { "epoch": 1.3598609472965424, "grad_norm": 0.5413544898453932, "learning_rate": 6.656142524703736e-05, "loss": 2.7984, "step": 21906 }, { "epoch": 1.3599230243963003, "grad_norm": 0.6356644121402446, "learning_rate": 6.655801754814056e-05, "loss": 2.7224, "step": 21907 }, { "epoch": 1.3599851014960582, "grad_norm": 0.5964910330709955, "learning_rate": 6.655460976285655e-05, "loss": 2.847, "step": 21908 }, { "epoch": 1.3600471785958161, "grad_norm": 0.6219904921746408, "learning_rate": 6.655120189120314e-05, "loss": 2.8401, "step": 21909 }, { "epoch": 1.3601092556955738, "grad_norm": 0.5297293778690607, "learning_rate": 6.654779393319807e-05, "loss": 2.7861, "step": 21910 }, { "epoch": 1.3601713327953318, "grad_norm": 0.6807374846211219, "learning_rate": 6.654438588885916e-05, "loss": 2.8235, "step": 21911 }, { "epoch": 1.3602334098950897, "grad_norm": 0.6309160614528062, "learning_rate": 6.654097775820416e-05, "loss": 2.7393, "step": 21912 }, { "epoch": 1.3602954869948476, "grad_norm": 0.6059564410628673, "learning_rate": 6.653756954125088e-05, "loss": 2.8031, "step": 21913 }, { "epoch": 1.3603575640946055, "grad_norm": 0.5665693903155297, "learning_rate": 6.653416123801708e-05, "loss": 2.8529, "step": 21914 }, { "epoch": 1.3604196411943634, "grad_norm": 0.600522219371492, "learning_rate": 6.653075284852053e-05, "loss": 2.8318, "step": 21915 }, { "epoch": 1.3604817182941213, "grad_norm": 0.5501577757901174, "learning_rate": 6.652734437277905e-05, "loss": 2.841, "step": 21916 }, { "epoch": 1.3605437953938793, "grad_norm": 0.6076416403087509, "learning_rate": 6.65239358108104e-05, "loss": 2.8839, "step": 21917 }, { "epoch": 1.360605872493637, "grad_norm": 0.5236602423148583, "learning_rate": 6.652052716263236e-05, "loss": 2.7458, "step": 21918 }, { "epoch": 1.3606679495933949, "grad_norm": 0.5478317066749134, "learning_rate": 6.651711842826272e-05, "loss": 2.8508, "step": 21919 }, { "epoch": 1.3607300266931528, "grad_norm": 0.5211346771776463, "learning_rate": 6.651370960771927e-05, "loss": 2.7808, "step": 21920 }, { "epoch": 1.3607921037929107, "grad_norm": 0.5860441238029963, "learning_rate": 6.651030070101977e-05, "loss": 2.9616, "step": 21921 }, { "epoch": 1.3608541808926686, "grad_norm": 0.5517433367155288, "learning_rate": 6.650689170818203e-05, "loss": 2.7906, "step": 21922 }, { "epoch": 1.3609162579924265, "grad_norm": 0.5278089096552515, "learning_rate": 6.650348262922384e-05, "loss": 2.7193, "step": 21923 }, { "epoch": 1.3609783350921845, "grad_norm": 0.6088091105640997, "learning_rate": 6.650007346416297e-05, "loss": 2.8469, "step": 21924 }, { "epoch": 1.3610404121919424, "grad_norm": 0.570450543980586, "learning_rate": 6.64966642130172e-05, "loss": 2.8096, "step": 21925 }, { "epoch": 1.3611024892917003, "grad_norm": 0.5646661006839457, "learning_rate": 6.649325487580433e-05, "loss": 2.8553, "step": 21926 }, { "epoch": 1.3611645663914582, "grad_norm": 0.5646612714155408, "learning_rate": 6.648984545254217e-05, "loss": 2.8337, "step": 21927 }, { "epoch": 1.3612266434912161, "grad_norm": 0.5518044241062464, "learning_rate": 6.648643594324846e-05, "loss": 2.8405, "step": 21928 }, { "epoch": 1.361288720590974, "grad_norm": 0.5955621020879907, "learning_rate": 6.648302634794101e-05, "loss": 2.9166, "step": 21929 }, { "epoch": 1.361350797690732, "grad_norm": 0.5467135327251859, "learning_rate": 6.647961666663762e-05, "loss": 2.8494, "step": 21930 }, { "epoch": 1.36141287479049, "grad_norm": 0.5973511958275602, "learning_rate": 6.647620689935608e-05, "loss": 2.8131, "step": 21931 }, { "epoch": 1.3614749518902478, "grad_norm": 0.5552820055418313, "learning_rate": 6.647279704611414e-05, "loss": 2.8161, "step": 21932 }, { "epoch": 1.3615370289900055, "grad_norm": 0.5656700611606474, "learning_rate": 6.646938710692963e-05, "loss": 2.8653, "step": 21933 }, { "epoch": 1.3615991060897634, "grad_norm": 0.6353110684267531, "learning_rate": 6.646597708182034e-05, "loss": 2.8722, "step": 21934 }, { "epoch": 1.3616611831895213, "grad_norm": 0.6381088481581352, "learning_rate": 6.646256697080405e-05, "loss": 2.8567, "step": 21935 }, { "epoch": 1.3617232602892793, "grad_norm": 0.6054892505743693, "learning_rate": 6.645915677389853e-05, "loss": 2.8079, "step": 21936 }, { "epoch": 1.3617853373890372, "grad_norm": 0.5710876631510227, "learning_rate": 6.645574649112163e-05, "loss": 2.7672, "step": 21937 }, { "epoch": 1.361847414488795, "grad_norm": 0.5570239972127922, "learning_rate": 6.645233612249107e-05, "loss": 2.8201, "step": 21938 }, { "epoch": 1.361909491588553, "grad_norm": 0.6126938941168183, "learning_rate": 6.644892566802469e-05, "loss": 2.7987, "step": 21939 }, { "epoch": 1.361971568688311, "grad_norm": 0.624324982427738, "learning_rate": 6.644551512774027e-05, "loss": 2.8966, "step": 21940 }, { "epoch": 1.3620336457880688, "grad_norm": 0.5799751318746894, "learning_rate": 6.644210450165562e-05, "loss": 2.8402, "step": 21941 }, { "epoch": 1.3620957228878265, "grad_norm": 0.5823947104471247, "learning_rate": 6.643869378978851e-05, "loss": 2.9003, "step": 21942 }, { "epoch": 1.3621577999875845, "grad_norm": 0.5703557141797382, "learning_rate": 6.643528299215675e-05, "loss": 2.8279, "step": 21943 }, { "epoch": 1.3622198770873424, "grad_norm": 0.5856980152271047, "learning_rate": 6.643187210877811e-05, "loss": 2.7633, "step": 21944 }, { "epoch": 1.3622819541871003, "grad_norm": 0.5633743431420959, "learning_rate": 6.642846113967043e-05, "loss": 2.8414, "step": 21945 }, { "epoch": 1.3623440312868582, "grad_norm": 0.5260301122875756, "learning_rate": 6.642505008485146e-05, "loss": 2.8118, "step": 21946 }, { "epoch": 1.3624061083866161, "grad_norm": 0.6071368465606911, "learning_rate": 6.642163894433903e-05, "loss": 2.7614, "step": 21947 }, { "epoch": 1.362468185486374, "grad_norm": 0.5051426172939023, "learning_rate": 6.641822771815092e-05, "loss": 2.7587, "step": 21948 }, { "epoch": 1.362530262586132, "grad_norm": 0.5578529140246887, "learning_rate": 6.641481640630492e-05, "loss": 2.8274, "step": 21949 }, { "epoch": 1.3625923396858899, "grad_norm": 0.5196976646122035, "learning_rate": 6.641140500881884e-05, "loss": 2.7332, "step": 21950 }, { "epoch": 1.3626544167856478, "grad_norm": 0.5671193599491944, "learning_rate": 6.640799352571049e-05, "loss": 2.8047, "step": 21951 }, { "epoch": 1.3627164938854057, "grad_norm": 0.5218755562145444, "learning_rate": 6.640458195699763e-05, "loss": 2.8424, "step": 21952 }, { "epoch": 1.3627785709851636, "grad_norm": 0.5840834466597834, "learning_rate": 6.640117030269811e-05, "loss": 2.8803, "step": 21953 }, { "epoch": 1.3628406480849216, "grad_norm": 0.5540833473909333, "learning_rate": 6.639775856282968e-05, "loss": 2.8554, "step": 21954 }, { "epoch": 1.3629027251846795, "grad_norm": 0.5760215433847057, "learning_rate": 6.639434673741018e-05, "loss": 2.871, "step": 21955 }, { "epoch": 1.3629648022844374, "grad_norm": 0.5248861336977381, "learning_rate": 6.63909348264574e-05, "loss": 2.806, "step": 21956 }, { "epoch": 1.363026879384195, "grad_norm": 0.5875757838130161, "learning_rate": 6.63875228299891e-05, "loss": 2.7907, "step": 21957 }, { "epoch": 1.363088956483953, "grad_norm": 0.5367595018577211, "learning_rate": 6.638411074802316e-05, "loss": 2.735, "step": 21958 }, { "epoch": 1.363151033583711, "grad_norm": 0.5843460086016083, "learning_rate": 6.63806985805773e-05, "loss": 2.8621, "step": 21959 }, { "epoch": 1.3632131106834688, "grad_norm": 0.5431317146597343, "learning_rate": 6.637728632766937e-05, "loss": 2.709, "step": 21960 }, { "epoch": 1.3632751877832268, "grad_norm": 0.5699490996271259, "learning_rate": 6.637387398931715e-05, "loss": 2.8352, "step": 21961 }, { "epoch": 1.3633372648829847, "grad_norm": 0.5610949398953412, "learning_rate": 6.637046156553849e-05, "loss": 2.8034, "step": 21962 }, { "epoch": 1.3633993419827426, "grad_norm": 0.5948036281721313, "learning_rate": 6.636704905635114e-05, "loss": 2.8537, "step": 21963 }, { "epoch": 1.3634614190825005, "grad_norm": 0.6697608153583103, "learning_rate": 6.636363646177292e-05, "loss": 2.8628, "step": 21964 }, { "epoch": 1.3635234961822584, "grad_norm": 0.575663093027149, "learning_rate": 6.636022378182163e-05, "loss": 2.8601, "step": 21965 }, { "epoch": 1.3635855732820161, "grad_norm": 0.7894985769137022, "learning_rate": 6.635681101651508e-05, "loss": 2.8258, "step": 21966 }, { "epoch": 1.363647650381774, "grad_norm": 0.6218212832920323, "learning_rate": 6.635339816587109e-05, "loss": 2.8125, "step": 21967 }, { "epoch": 1.363709727481532, "grad_norm": 0.623683090405387, "learning_rate": 6.634998522990743e-05, "loss": 2.8848, "step": 21968 }, { "epoch": 1.3637718045812899, "grad_norm": 0.6428518900580985, "learning_rate": 6.634657220864195e-05, "loss": 2.8765, "step": 21969 }, { "epoch": 1.3638338816810478, "grad_norm": 0.6762324023469087, "learning_rate": 6.634315910209243e-05, "loss": 2.9027, "step": 21970 }, { "epoch": 1.3638959587808057, "grad_norm": 0.5318657728351845, "learning_rate": 6.633974591027668e-05, "loss": 2.8755, "step": 21971 }, { "epoch": 1.3639580358805636, "grad_norm": 0.6262052121393553, "learning_rate": 6.63363326332125e-05, "loss": 2.8124, "step": 21972 }, { "epoch": 1.3640201129803216, "grad_norm": 0.6225996654650521, "learning_rate": 6.633291927091772e-05, "loss": 2.824, "step": 21973 }, { "epoch": 1.3640821900800795, "grad_norm": 0.619417074681755, "learning_rate": 6.632950582341013e-05, "loss": 2.9026, "step": 21974 }, { "epoch": 1.3641442671798374, "grad_norm": 0.5777216612659501, "learning_rate": 6.632609229070753e-05, "loss": 2.7822, "step": 21975 }, { "epoch": 1.3642063442795953, "grad_norm": 0.5837022744300294, "learning_rate": 6.632267867282777e-05, "loss": 2.8784, "step": 21976 }, { "epoch": 1.3642684213793532, "grad_norm": 0.5270065076976326, "learning_rate": 6.631926496978861e-05, "loss": 2.7668, "step": 21977 }, { "epoch": 1.3643304984791111, "grad_norm": 0.5551575294520684, "learning_rate": 6.631585118160791e-05, "loss": 2.7809, "step": 21978 }, { "epoch": 1.364392575578869, "grad_norm": 0.6095106022908561, "learning_rate": 6.63124373083034e-05, "loss": 2.8178, "step": 21979 }, { "epoch": 1.364454652678627, "grad_norm": 0.5178018338400443, "learning_rate": 6.6309023349893e-05, "loss": 2.7277, "step": 21980 }, { "epoch": 1.3645167297783847, "grad_norm": 0.6283403539855618, "learning_rate": 6.630560930639445e-05, "loss": 2.7639, "step": 21981 }, { "epoch": 1.3645788068781426, "grad_norm": 0.49451485135097073, "learning_rate": 6.630219517782556e-05, "loss": 2.7502, "step": 21982 }, { "epoch": 1.3646408839779005, "grad_norm": 0.6322964341823404, "learning_rate": 6.629878096420419e-05, "loss": 2.8217, "step": 21983 }, { "epoch": 1.3647029610776584, "grad_norm": 0.5102633972745446, "learning_rate": 6.62953666655481e-05, "loss": 2.7732, "step": 21984 }, { "epoch": 1.3647650381774163, "grad_norm": 0.564480236418741, "learning_rate": 6.629195228187515e-05, "loss": 2.8248, "step": 21985 }, { "epoch": 1.3648271152771743, "grad_norm": 0.5866100774460605, "learning_rate": 6.62885378132031e-05, "loss": 2.8568, "step": 21986 }, { "epoch": 1.3648891923769322, "grad_norm": 0.6336303418149355, "learning_rate": 6.62851232595498e-05, "loss": 2.8664, "step": 21987 }, { "epoch": 1.36495126947669, "grad_norm": 0.6785470229435133, "learning_rate": 6.628170862093308e-05, "loss": 2.8434, "step": 21988 }, { "epoch": 1.365013346576448, "grad_norm": 0.5646913547099308, "learning_rate": 6.62782938973707e-05, "loss": 2.877, "step": 21989 }, { "epoch": 1.3650754236762057, "grad_norm": 0.6230491470725985, "learning_rate": 6.627487908888054e-05, "loss": 2.7199, "step": 21990 }, { "epoch": 1.3651375007759636, "grad_norm": 0.6169820999040574, "learning_rate": 6.627146419548037e-05, "loss": 2.8438, "step": 21991 }, { "epoch": 1.3651995778757215, "grad_norm": 0.6154545465725227, "learning_rate": 6.626804921718803e-05, "loss": 2.7115, "step": 21992 }, { "epoch": 1.3652616549754795, "grad_norm": 0.5839848598108225, "learning_rate": 6.626463415402132e-05, "loss": 2.7386, "step": 21993 }, { "epoch": 1.3653237320752374, "grad_norm": 0.5839359174035507, "learning_rate": 6.626121900599808e-05, "loss": 2.842, "step": 21994 }, { "epoch": 1.3653858091749953, "grad_norm": 0.5507023261047275, "learning_rate": 6.625780377313609e-05, "loss": 2.8935, "step": 21995 }, { "epoch": 1.3654478862747532, "grad_norm": 0.5611887589582959, "learning_rate": 6.625438845545321e-05, "loss": 2.8309, "step": 21996 }, { "epoch": 1.3655099633745111, "grad_norm": 0.5573489584544405, "learning_rate": 6.625097305296722e-05, "loss": 2.8844, "step": 21997 }, { "epoch": 1.365572040474269, "grad_norm": 0.6161354127477272, "learning_rate": 6.624755756569598e-05, "loss": 2.8189, "step": 21998 }, { "epoch": 1.365634117574027, "grad_norm": 0.578003767550706, "learning_rate": 6.624414199365728e-05, "loss": 2.8075, "step": 21999 }, { "epoch": 1.365696194673785, "grad_norm": 0.5866167581752683, "learning_rate": 6.624072633686894e-05, "loss": 2.8075, "step": 22000 }, { "epoch": 1.3657582717735428, "grad_norm": 0.58947466993552, "learning_rate": 6.623731059534881e-05, "loss": 2.8196, "step": 22001 }, { "epoch": 1.3658203488733007, "grad_norm": 0.5026650927153887, "learning_rate": 6.623389476911465e-05, "loss": 2.7504, "step": 22002 }, { "epoch": 1.3658824259730586, "grad_norm": 0.6788362236474976, "learning_rate": 6.623047885818434e-05, "loss": 2.8252, "step": 22003 }, { "epoch": 1.3659445030728166, "grad_norm": 0.5257139097624821, "learning_rate": 6.622706286257569e-05, "loss": 2.8251, "step": 22004 }, { "epoch": 1.3660065801725743, "grad_norm": 0.604872354448449, "learning_rate": 6.622364678230651e-05, "loss": 2.6731, "step": 22005 }, { "epoch": 1.3660686572723322, "grad_norm": 0.5560361022308716, "learning_rate": 6.622023061739463e-05, "loss": 2.7816, "step": 22006 }, { "epoch": 1.36613073437209, "grad_norm": 0.5993552896641887, "learning_rate": 6.621681436785786e-05, "loss": 2.8332, "step": 22007 }, { "epoch": 1.366192811471848, "grad_norm": 0.538030576540777, "learning_rate": 6.621339803371405e-05, "loss": 2.8375, "step": 22008 }, { "epoch": 1.366254888571606, "grad_norm": 0.596320335950025, "learning_rate": 6.620998161498098e-05, "loss": 2.7812, "step": 22009 }, { "epoch": 1.3663169656713638, "grad_norm": 0.5838044313543578, "learning_rate": 6.620656511167653e-05, "loss": 2.7713, "step": 22010 }, { "epoch": 1.3663790427711218, "grad_norm": 0.5650307626704736, "learning_rate": 6.620314852381846e-05, "loss": 2.8767, "step": 22011 }, { "epoch": 1.3664411198708797, "grad_norm": 0.5789881268669129, "learning_rate": 6.619973185142466e-05, "loss": 2.7766, "step": 22012 }, { "epoch": 1.3665031969706376, "grad_norm": 0.5274647609185367, "learning_rate": 6.619631509451291e-05, "loss": 2.7474, "step": 22013 }, { "epoch": 1.3665652740703953, "grad_norm": 0.554164828271135, "learning_rate": 6.619289825310107e-05, "loss": 2.8625, "step": 22014 }, { "epoch": 1.3666273511701532, "grad_norm": 0.543790812440573, "learning_rate": 6.618948132720693e-05, "loss": 2.8036, "step": 22015 }, { "epoch": 1.3666894282699111, "grad_norm": 0.543206636199949, "learning_rate": 6.618606431684835e-05, "loss": 2.8589, "step": 22016 }, { "epoch": 1.366751505369669, "grad_norm": 0.580104403430381, "learning_rate": 6.618264722204314e-05, "loss": 2.8098, "step": 22017 }, { "epoch": 1.366813582469427, "grad_norm": 0.4885902647669389, "learning_rate": 6.617923004280912e-05, "loss": 2.8161, "step": 22018 }, { "epoch": 1.3668756595691849, "grad_norm": 0.6244547372790259, "learning_rate": 6.617581277916414e-05, "loss": 2.7673, "step": 22019 }, { "epoch": 1.3669377366689428, "grad_norm": 0.5679873792696837, "learning_rate": 6.617239543112601e-05, "loss": 2.8234, "step": 22020 }, { "epoch": 1.3669998137687007, "grad_norm": 0.5420726544557141, "learning_rate": 6.616897799871258e-05, "loss": 2.8175, "step": 22021 }, { "epoch": 1.3670618908684586, "grad_norm": 0.4965622709500629, "learning_rate": 6.616556048194165e-05, "loss": 2.704, "step": 22022 }, { "epoch": 1.3671239679682166, "grad_norm": 0.5584636883606869, "learning_rate": 6.616214288083108e-05, "loss": 2.7123, "step": 22023 }, { "epoch": 1.3671860450679745, "grad_norm": 0.5254321737771571, "learning_rate": 6.615872519539869e-05, "loss": 2.8405, "step": 22024 }, { "epoch": 1.3672481221677324, "grad_norm": 0.5403721973541524, "learning_rate": 6.61553074256623e-05, "loss": 2.8069, "step": 22025 }, { "epoch": 1.3673101992674903, "grad_norm": 0.5257444927182894, "learning_rate": 6.615188957163975e-05, "loss": 2.8109, "step": 22026 }, { "epoch": 1.3673722763672482, "grad_norm": 0.499018047624658, "learning_rate": 6.614847163334886e-05, "loss": 2.7909, "step": 22027 }, { "epoch": 1.3674343534670061, "grad_norm": 0.5495742667344217, "learning_rate": 6.61450536108075e-05, "loss": 2.7067, "step": 22028 }, { "epoch": 1.3674964305667638, "grad_norm": 0.5240505386379616, "learning_rate": 6.614163550403345e-05, "loss": 2.7825, "step": 22029 }, { "epoch": 1.3675585076665218, "grad_norm": 0.526003370370329, "learning_rate": 6.613821731304458e-05, "loss": 2.8726, "step": 22030 }, { "epoch": 1.3676205847662797, "grad_norm": 0.5912026169402181, "learning_rate": 6.61347990378587e-05, "loss": 2.8247, "step": 22031 }, { "epoch": 1.3676826618660376, "grad_norm": 0.5137822137836016, "learning_rate": 6.613138067849369e-05, "loss": 2.8066, "step": 22032 }, { "epoch": 1.3677447389657955, "grad_norm": 0.6053709874135352, "learning_rate": 6.612796223496731e-05, "loss": 2.7851, "step": 22033 }, { "epoch": 1.3678068160655534, "grad_norm": 0.6489081914970514, "learning_rate": 6.612454370729747e-05, "loss": 2.7857, "step": 22034 }, { "epoch": 1.3678688931653114, "grad_norm": 0.5256205059408239, "learning_rate": 6.612112509550196e-05, "loss": 2.8243, "step": 22035 }, { "epoch": 1.3679309702650693, "grad_norm": 0.631371564773824, "learning_rate": 6.61177063995986e-05, "loss": 2.824, "step": 22036 }, { "epoch": 1.3679930473648272, "grad_norm": 0.5312999533560797, "learning_rate": 6.611428761960528e-05, "loss": 2.8867, "step": 22037 }, { "epoch": 1.3680551244645849, "grad_norm": 0.6943640079538159, "learning_rate": 6.61108687555398e-05, "loss": 2.9206, "step": 22038 }, { "epoch": 1.3681172015643428, "grad_norm": 0.7476371659062502, "learning_rate": 6.610744980742001e-05, "loss": 2.7334, "step": 22039 }, { "epoch": 1.3681792786641007, "grad_norm": 0.5235881304005331, "learning_rate": 6.610403077526374e-05, "loss": 2.8223, "step": 22040 }, { "epoch": 1.3682413557638586, "grad_norm": 0.7422062118580761, "learning_rate": 6.610061165908885e-05, "loss": 2.8393, "step": 22041 }, { "epoch": 1.3683034328636166, "grad_norm": 0.579800497593885, "learning_rate": 6.609719245891316e-05, "loss": 2.7794, "step": 22042 }, { "epoch": 1.3683655099633745, "grad_norm": 0.5583707190802754, "learning_rate": 6.609377317475449e-05, "loss": 2.8261, "step": 22043 }, { "epoch": 1.3684275870631324, "grad_norm": 0.6245479617961986, "learning_rate": 6.609035380663071e-05, "loss": 2.8698, "step": 22044 }, { "epoch": 1.3684896641628903, "grad_norm": 0.542303679713364, "learning_rate": 6.608693435455964e-05, "loss": 2.8599, "step": 22045 }, { "epoch": 1.3685517412626482, "grad_norm": 0.5845943059818192, "learning_rate": 6.608351481855915e-05, "loss": 2.8462, "step": 22046 }, { "epoch": 1.3686138183624061, "grad_norm": 0.5225298399366946, "learning_rate": 6.608009519864702e-05, "loss": 2.8117, "step": 22047 }, { "epoch": 1.368675895462164, "grad_norm": 0.5633042467055087, "learning_rate": 6.607667549484117e-05, "loss": 2.8639, "step": 22048 }, { "epoch": 1.368737972561922, "grad_norm": 0.537666087771461, "learning_rate": 6.607325570715937e-05, "loss": 2.9131, "step": 22049 }, { "epoch": 1.36880004966168, "grad_norm": 0.5294106561752008, "learning_rate": 6.606983583561952e-05, "loss": 2.8683, "step": 22050 }, { "epoch": 1.3688621267614378, "grad_norm": 0.5570327715966628, "learning_rate": 6.606641588023941e-05, "loss": 2.8079, "step": 22051 }, { "epoch": 1.3689242038611957, "grad_norm": 0.5944411872490312, "learning_rate": 6.606299584103692e-05, "loss": 2.7943, "step": 22052 }, { "epoch": 1.3689862809609534, "grad_norm": 0.5266399915142745, "learning_rate": 6.605957571802988e-05, "loss": 2.7716, "step": 22053 }, { "epoch": 1.3690483580607113, "grad_norm": 0.5329518267456265, "learning_rate": 6.605615551123612e-05, "loss": 2.7694, "step": 22054 }, { "epoch": 1.3691104351604693, "grad_norm": 0.5505439571792109, "learning_rate": 6.605273522067352e-05, "loss": 2.8333, "step": 22055 }, { "epoch": 1.3691725122602272, "grad_norm": 0.481537196926044, "learning_rate": 6.60493148463599e-05, "loss": 2.731, "step": 22056 }, { "epoch": 1.369234589359985, "grad_norm": 0.5620772839704464, "learning_rate": 6.604589438831311e-05, "loss": 2.7431, "step": 22057 }, { "epoch": 1.369296666459743, "grad_norm": 0.496507637205287, "learning_rate": 6.604247384655099e-05, "loss": 2.8076, "step": 22058 }, { "epoch": 1.369358743559501, "grad_norm": 0.5276722767717017, "learning_rate": 6.603905322109139e-05, "loss": 2.7807, "step": 22059 }, { "epoch": 1.3694208206592589, "grad_norm": 0.5276840807189368, "learning_rate": 6.603563251195216e-05, "loss": 2.7748, "step": 22060 }, { "epoch": 1.3694828977590168, "grad_norm": 0.5442026238539788, "learning_rate": 6.603221171915113e-05, "loss": 2.8643, "step": 22061 }, { "epoch": 1.3695449748587745, "grad_norm": 0.5457011567546545, "learning_rate": 6.602879084270618e-05, "loss": 2.9234, "step": 22062 }, { "epoch": 1.3696070519585324, "grad_norm": 0.6007059861203216, "learning_rate": 6.602536988263512e-05, "loss": 2.8262, "step": 22063 }, { "epoch": 1.3696691290582903, "grad_norm": 0.5157210231726979, "learning_rate": 6.602194883895582e-05, "loss": 2.7842, "step": 22064 }, { "epoch": 1.3697312061580482, "grad_norm": 0.5144728301263658, "learning_rate": 6.601852771168612e-05, "loss": 2.7115, "step": 22065 }, { "epoch": 1.3697932832578061, "grad_norm": 0.5996636729553556, "learning_rate": 6.601510650084388e-05, "loss": 2.9197, "step": 22066 }, { "epoch": 1.369855360357564, "grad_norm": 0.5783322452549589, "learning_rate": 6.601168520644692e-05, "loss": 2.822, "step": 22067 }, { "epoch": 1.369917437457322, "grad_norm": 0.5750048181083166, "learning_rate": 6.600826382851314e-05, "loss": 2.9153, "step": 22068 }, { "epoch": 1.36997951455708, "grad_norm": 0.595625650339482, "learning_rate": 6.600484236706034e-05, "loss": 2.897, "step": 22069 }, { "epoch": 1.3700415916568378, "grad_norm": 0.5857244231305437, "learning_rate": 6.60014208221064e-05, "loss": 2.8331, "step": 22070 }, { "epoch": 1.3701036687565957, "grad_norm": 0.6594676427256604, "learning_rate": 6.599799919366917e-05, "loss": 2.8927, "step": 22071 }, { "epoch": 1.3701657458563536, "grad_norm": 0.6587869562137431, "learning_rate": 6.599457748176649e-05, "loss": 2.8492, "step": 22072 }, { "epoch": 1.3702278229561116, "grad_norm": 0.6048248312496142, "learning_rate": 6.59911556864162e-05, "loss": 2.8175, "step": 22073 }, { "epoch": 1.3702899000558695, "grad_norm": 0.6445420928967808, "learning_rate": 6.598773380763619e-05, "loss": 2.8391, "step": 22074 }, { "epoch": 1.3703519771556274, "grad_norm": 0.5784711832289303, "learning_rate": 6.59843118454443e-05, "loss": 2.7696, "step": 22075 }, { "epoch": 1.3704140542553853, "grad_norm": 0.6785796772663437, "learning_rate": 6.598088979985835e-05, "loss": 2.8789, "step": 22076 }, { "epoch": 1.370476131355143, "grad_norm": 0.5684724413433326, "learning_rate": 6.597746767089623e-05, "loss": 2.8247, "step": 22077 }, { "epoch": 1.370538208454901, "grad_norm": 0.622167774272859, "learning_rate": 6.597404545857579e-05, "loss": 2.7491, "step": 22078 }, { "epoch": 1.3706002855546588, "grad_norm": 0.5556318813451391, "learning_rate": 6.597062316291486e-05, "loss": 2.7673, "step": 22079 }, { "epoch": 1.3706623626544168, "grad_norm": 0.6377014580469929, "learning_rate": 6.596720078393132e-05, "loss": 2.9054, "step": 22080 }, { "epoch": 1.3707244397541747, "grad_norm": 0.6122619516647755, "learning_rate": 6.596377832164302e-05, "loss": 2.7833, "step": 22081 }, { "epoch": 1.3707865168539326, "grad_norm": 0.5806209451967039, "learning_rate": 6.596035577606781e-05, "loss": 2.7835, "step": 22082 }, { "epoch": 1.3708485939536905, "grad_norm": 0.6126055062178875, "learning_rate": 6.595693314722356e-05, "loss": 2.8648, "step": 22083 }, { "epoch": 1.3709106710534484, "grad_norm": 0.5710904028673102, "learning_rate": 6.59535104351281e-05, "loss": 2.7627, "step": 22084 }, { "epoch": 1.3709727481532064, "grad_norm": 0.6157344662700642, "learning_rate": 6.595008763979931e-05, "loss": 2.7556, "step": 22085 }, { "epoch": 1.371034825252964, "grad_norm": 0.5537295868170163, "learning_rate": 6.594666476125503e-05, "loss": 2.8047, "step": 22086 }, { "epoch": 1.371096902352722, "grad_norm": 0.5299438801329751, "learning_rate": 6.594324179951314e-05, "loss": 2.8629, "step": 22087 }, { "epoch": 1.3711589794524799, "grad_norm": 0.5583578825530766, "learning_rate": 6.593981875459149e-05, "loss": 2.887, "step": 22088 }, { "epoch": 1.3712210565522378, "grad_norm": 0.562513642675473, "learning_rate": 6.593639562650793e-05, "loss": 2.8977, "step": 22089 }, { "epoch": 1.3712831336519957, "grad_norm": 0.5350302944786426, "learning_rate": 6.593297241528033e-05, "loss": 2.7918, "step": 22090 }, { "epoch": 1.3713452107517536, "grad_norm": 0.5305474347957151, "learning_rate": 6.592954912092654e-05, "loss": 2.7489, "step": 22091 }, { "epoch": 1.3714072878515116, "grad_norm": 0.5562342984415849, "learning_rate": 6.592612574346442e-05, "loss": 2.7743, "step": 22092 }, { "epoch": 1.3714693649512695, "grad_norm": 0.5474924960993904, "learning_rate": 6.592270228291185e-05, "loss": 2.7163, "step": 22093 }, { "epoch": 1.3715314420510274, "grad_norm": 0.5478733758782647, "learning_rate": 6.591927873928667e-05, "loss": 2.7821, "step": 22094 }, { "epoch": 1.3715935191507853, "grad_norm": 0.5016911041711988, "learning_rate": 6.591585511260675e-05, "loss": 2.7624, "step": 22095 }, { "epoch": 1.3716555962505432, "grad_norm": 0.5458977823816508, "learning_rate": 6.591243140288995e-05, "loss": 2.7863, "step": 22096 }, { "epoch": 1.3717176733503011, "grad_norm": 0.5801937103711658, "learning_rate": 6.590900761015413e-05, "loss": 2.732, "step": 22097 }, { "epoch": 1.371779750450059, "grad_norm": 0.4898568633786786, "learning_rate": 6.590558373441717e-05, "loss": 2.7671, "step": 22098 }, { "epoch": 1.371841827549817, "grad_norm": 0.5685694198246009, "learning_rate": 6.59021597756969e-05, "loss": 2.8445, "step": 22099 }, { "epoch": 1.371903904649575, "grad_norm": 0.5016398955433272, "learning_rate": 6.58987357340112e-05, "loss": 2.7985, "step": 22100 }, { "epoch": 1.3719659817493326, "grad_norm": 0.6181039886452067, "learning_rate": 6.589531160937794e-05, "loss": 2.8415, "step": 22101 }, { "epoch": 1.3720280588490905, "grad_norm": 0.5428190779740358, "learning_rate": 6.589188740181498e-05, "loss": 2.7531, "step": 22102 }, { "epoch": 1.3720901359488484, "grad_norm": 0.557053636954398, "learning_rate": 6.588846311134019e-05, "loss": 2.8339, "step": 22103 }, { "epoch": 1.3721522130486064, "grad_norm": 0.6043273257256606, "learning_rate": 6.588503873797143e-05, "loss": 2.85, "step": 22104 }, { "epoch": 1.3722142901483643, "grad_norm": 0.524175181049318, "learning_rate": 6.588161428172656e-05, "loss": 2.8676, "step": 22105 }, { "epoch": 1.3722763672481222, "grad_norm": 0.5866235404346971, "learning_rate": 6.587818974262346e-05, "loss": 2.8456, "step": 22106 }, { "epoch": 1.37233844434788, "grad_norm": 0.5241043625491183, "learning_rate": 6.587476512068e-05, "loss": 2.8563, "step": 22107 }, { "epoch": 1.372400521447638, "grad_norm": 0.5473823646124895, "learning_rate": 6.587134041591402e-05, "loss": 2.694, "step": 22108 }, { "epoch": 1.372462598547396, "grad_norm": 0.541891992603927, "learning_rate": 6.586791562834342e-05, "loss": 2.8409, "step": 22109 }, { "epoch": 1.3725246756471536, "grad_norm": 0.5973470299321638, "learning_rate": 6.586449075798604e-05, "loss": 2.8027, "step": 22110 }, { "epoch": 1.3725867527469116, "grad_norm": 0.5407812664447508, "learning_rate": 6.586106580485976e-05, "loss": 2.8399, "step": 22111 }, { "epoch": 1.3726488298466695, "grad_norm": 0.6080188087280228, "learning_rate": 6.585764076898245e-05, "loss": 2.9049, "step": 22112 }, { "epoch": 1.3727109069464274, "grad_norm": 0.6089592028609585, "learning_rate": 6.585421565037199e-05, "loss": 2.8123, "step": 22113 }, { "epoch": 1.3727729840461853, "grad_norm": 0.570517907972383, "learning_rate": 6.585079044904623e-05, "loss": 2.8418, "step": 22114 }, { "epoch": 1.3728350611459432, "grad_norm": 0.5530123062420063, "learning_rate": 6.584736516502304e-05, "loss": 2.801, "step": 22115 }, { "epoch": 1.3728971382457011, "grad_norm": 0.5642482029197919, "learning_rate": 6.584393979832031e-05, "loss": 2.7998, "step": 22116 }, { "epoch": 1.372959215345459, "grad_norm": 0.5332853093231688, "learning_rate": 6.584051434895589e-05, "loss": 2.8167, "step": 22117 }, { "epoch": 1.373021292445217, "grad_norm": 0.6181152469108563, "learning_rate": 6.583708881694767e-05, "loss": 2.8247, "step": 22118 }, { "epoch": 1.373083369544975, "grad_norm": 0.49344474434442037, "learning_rate": 6.58336632023135e-05, "loss": 2.8375, "step": 22119 }, { "epoch": 1.3731454466447328, "grad_norm": 0.5748538443116389, "learning_rate": 6.583023750507129e-05, "loss": 2.8083, "step": 22120 }, { "epoch": 1.3732075237444907, "grad_norm": 0.5666025808036567, "learning_rate": 6.582681172523887e-05, "loss": 2.8239, "step": 22121 }, { "epoch": 1.3732696008442487, "grad_norm": 0.5432264142013948, "learning_rate": 6.582338586283413e-05, "loss": 2.8669, "step": 22122 }, { "epoch": 1.3733316779440066, "grad_norm": 0.600461596511608, "learning_rate": 6.581995991787495e-05, "loss": 2.7757, "step": 22123 }, { "epoch": 1.3733937550437645, "grad_norm": 0.542390694044115, "learning_rate": 6.58165338903792e-05, "loss": 2.7876, "step": 22124 }, { "epoch": 1.3734558321435222, "grad_norm": 0.5162305166433481, "learning_rate": 6.581310778036474e-05, "loss": 2.8519, "step": 22125 }, { "epoch": 1.37351790924328, "grad_norm": 0.6214237895712693, "learning_rate": 6.580968158784947e-05, "loss": 2.8346, "step": 22126 }, { "epoch": 1.373579986343038, "grad_norm": 0.5142705351476002, "learning_rate": 6.580625531285126e-05, "loss": 2.8477, "step": 22127 }, { "epoch": 1.373642063442796, "grad_norm": 0.5932078898306455, "learning_rate": 6.580282895538797e-05, "loss": 2.8255, "step": 22128 }, { "epoch": 1.3737041405425539, "grad_norm": 0.5329783039025148, "learning_rate": 6.579940251547749e-05, "loss": 2.893, "step": 22129 }, { "epoch": 1.3737662176423118, "grad_norm": 0.5113648217120063, "learning_rate": 6.579597599313768e-05, "loss": 2.7914, "step": 22130 }, { "epoch": 1.3738282947420697, "grad_norm": 0.5606663805809304, "learning_rate": 6.579254938838644e-05, "loss": 2.7197, "step": 22131 }, { "epoch": 1.3738903718418276, "grad_norm": 0.5137619403741673, "learning_rate": 6.578912270124163e-05, "loss": 2.8262, "step": 22132 }, { "epoch": 1.3739524489415855, "grad_norm": 0.4875908204601697, "learning_rate": 6.578569593172114e-05, "loss": 2.6945, "step": 22133 }, { "epoch": 1.3740145260413432, "grad_norm": 0.5378757627119631, "learning_rate": 6.578226907984284e-05, "loss": 2.8315, "step": 22134 }, { "epoch": 1.3740766031411011, "grad_norm": 0.5460659855018251, "learning_rate": 6.57788421456246e-05, "loss": 2.7886, "step": 22135 }, { "epoch": 1.374138680240859, "grad_norm": 0.5646686076649194, "learning_rate": 6.577541512908432e-05, "loss": 2.8822, "step": 22136 }, { "epoch": 1.374200757340617, "grad_norm": 0.5380774634052269, "learning_rate": 6.577198803023988e-05, "loss": 2.8255, "step": 22137 }, { "epoch": 1.374262834440375, "grad_norm": 0.5122724842184194, "learning_rate": 6.576856084910912e-05, "loss": 2.7494, "step": 22138 }, { "epoch": 1.3743249115401328, "grad_norm": 0.5972959392232023, "learning_rate": 6.576513358570998e-05, "loss": 2.8467, "step": 22139 }, { "epoch": 1.3743869886398907, "grad_norm": 0.46465074715795923, "learning_rate": 6.576170624006031e-05, "loss": 2.8091, "step": 22140 }, { "epoch": 1.3744490657396486, "grad_norm": 0.5692514189969262, "learning_rate": 6.575827881217798e-05, "loss": 2.8164, "step": 22141 }, { "epoch": 1.3745111428394066, "grad_norm": 0.5540164595751317, "learning_rate": 6.575485130208088e-05, "loss": 2.7948, "step": 22142 }, { "epoch": 1.3745732199391645, "grad_norm": 0.5086121645170607, "learning_rate": 6.57514237097869e-05, "loss": 2.8529, "step": 22143 }, { "epoch": 1.3746352970389224, "grad_norm": 0.6090117129685586, "learning_rate": 6.574799603531393e-05, "loss": 2.8301, "step": 22144 }, { "epoch": 1.3746973741386803, "grad_norm": 0.5294640759104895, "learning_rate": 6.574456827867984e-05, "loss": 2.7686, "step": 22145 }, { "epoch": 1.3747594512384382, "grad_norm": 0.581587013863573, "learning_rate": 6.57411404399025e-05, "loss": 2.8261, "step": 22146 }, { "epoch": 1.3748215283381962, "grad_norm": 0.5099779468798501, "learning_rate": 6.573771251899981e-05, "loss": 2.7999, "step": 22147 }, { "epoch": 1.374883605437954, "grad_norm": 0.51037628312502, "learning_rate": 6.573428451598966e-05, "loss": 2.7028, "step": 22148 }, { "epoch": 1.3749456825377118, "grad_norm": 0.5608510378451598, "learning_rate": 6.573085643088993e-05, "loss": 2.8506, "step": 22149 }, { "epoch": 1.3750077596374697, "grad_norm": 0.5612727393405675, "learning_rate": 6.572742826371849e-05, "loss": 2.8228, "step": 22150 }, { "epoch": 1.3750698367372276, "grad_norm": 0.5706000452340446, "learning_rate": 6.572400001449326e-05, "loss": 2.7394, "step": 22151 }, { "epoch": 1.3751319138369855, "grad_norm": 0.5657952531895255, "learning_rate": 6.572057168323209e-05, "loss": 2.7991, "step": 22152 }, { "epoch": 1.3751939909367434, "grad_norm": 0.5526404418927753, "learning_rate": 6.571714326995289e-05, "loss": 2.7592, "step": 22153 }, { "epoch": 1.3752560680365014, "grad_norm": 0.5521644916501182, "learning_rate": 6.571371477467354e-05, "loss": 2.8186, "step": 22154 }, { "epoch": 1.3753181451362593, "grad_norm": 0.6007330985547223, "learning_rate": 6.571028619741191e-05, "loss": 2.8245, "step": 22155 }, { "epoch": 1.3753802222360172, "grad_norm": 0.5528550841823676, "learning_rate": 6.570685753818591e-05, "loss": 2.8191, "step": 22156 }, { "epoch": 1.375442299335775, "grad_norm": 0.5445066668134699, "learning_rate": 6.570342879701342e-05, "loss": 2.8654, "step": 22157 }, { "epoch": 1.3755043764355328, "grad_norm": 0.5775601746638547, "learning_rate": 6.569999997391235e-05, "loss": 2.7679, "step": 22158 }, { "epoch": 1.3755664535352907, "grad_norm": 0.5494193968122358, "learning_rate": 6.569657106890054e-05, "loss": 2.7874, "step": 22159 }, { "epoch": 1.3756285306350486, "grad_norm": 0.6765282752210274, "learning_rate": 6.569314208199593e-05, "loss": 2.85, "step": 22160 }, { "epoch": 1.3756906077348066, "grad_norm": 0.5724996407270345, "learning_rate": 6.568971301321638e-05, "loss": 2.8529, "step": 22161 }, { "epoch": 1.3757526848345645, "grad_norm": 0.5838856693404497, "learning_rate": 6.568628386257978e-05, "loss": 2.8366, "step": 22162 }, { "epoch": 1.3758147619343224, "grad_norm": 0.577904762382905, "learning_rate": 6.568285463010404e-05, "loss": 2.8242, "step": 22163 }, { "epoch": 1.3758768390340803, "grad_norm": 0.5338653253148943, "learning_rate": 6.567942531580705e-05, "loss": 2.8368, "step": 22164 }, { "epoch": 1.3759389161338382, "grad_norm": 0.5983953456125192, "learning_rate": 6.567599591970667e-05, "loss": 2.767, "step": 22165 }, { "epoch": 1.3760009932335961, "grad_norm": 0.5784575820231402, "learning_rate": 6.567256644182082e-05, "loss": 2.7751, "step": 22166 }, { "epoch": 1.376063070333354, "grad_norm": 0.6341087816248457, "learning_rate": 6.566913688216738e-05, "loss": 2.8952, "step": 22167 }, { "epoch": 1.376125147433112, "grad_norm": 0.6901684473197914, "learning_rate": 6.566570724076426e-05, "loss": 2.6877, "step": 22168 }, { "epoch": 1.37618722453287, "grad_norm": 0.60036892575605, "learning_rate": 6.566227751762934e-05, "loss": 2.8093, "step": 22169 }, { "epoch": 1.3762493016326278, "grad_norm": 0.5808984626594041, "learning_rate": 6.565884771278052e-05, "loss": 2.8247, "step": 22170 }, { "epoch": 1.3763113787323857, "grad_norm": 0.6837910394713548, "learning_rate": 6.56554178262357e-05, "loss": 2.8757, "step": 22171 }, { "epoch": 1.3763734558321437, "grad_norm": 0.5694219350888846, "learning_rate": 6.565198785801274e-05, "loss": 2.8432, "step": 22172 }, { "epoch": 1.3764355329319014, "grad_norm": 0.6254351769803781, "learning_rate": 6.564855780812957e-05, "loss": 2.8536, "step": 22173 }, { "epoch": 1.3764976100316593, "grad_norm": 0.5575311063205347, "learning_rate": 6.564512767660408e-05, "loss": 2.7603, "step": 22174 }, { "epoch": 1.3765596871314172, "grad_norm": 0.654717859370001, "learning_rate": 6.564169746345414e-05, "loss": 2.8582, "step": 22175 }, { "epoch": 1.376621764231175, "grad_norm": 0.6086849070923765, "learning_rate": 6.56382671686977e-05, "loss": 2.8145, "step": 22176 }, { "epoch": 1.376683841330933, "grad_norm": 0.6412713581566096, "learning_rate": 6.56348367923526e-05, "loss": 2.788, "step": 22177 }, { "epoch": 1.376745918430691, "grad_norm": 0.5501679868348115, "learning_rate": 6.563140633443677e-05, "loss": 2.7948, "step": 22178 }, { "epoch": 1.3768079955304489, "grad_norm": 0.612193874814704, "learning_rate": 6.562797579496812e-05, "loss": 2.7976, "step": 22179 }, { "epoch": 1.3768700726302068, "grad_norm": 0.5428732094995395, "learning_rate": 6.562454517396448e-05, "loss": 2.759, "step": 22180 }, { "epoch": 1.3769321497299647, "grad_norm": 0.5909350575397111, "learning_rate": 6.562111447144384e-05, "loss": 2.8798, "step": 22181 }, { "epoch": 1.3769942268297224, "grad_norm": 0.5722379710315149, "learning_rate": 6.561768368742401e-05, "loss": 2.7116, "step": 22182 }, { "epoch": 1.3770563039294803, "grad_norm": 0.5518886444807082, "learning_rate": 6.561425282192297e-05, "loss": 2.8265, "step": 22183 }, { "epoch": 1.3771183810292382, "grad_norm": 0.5786090577047857, "learning_rate": 6.561082187495856e-05, "loss": 2.8215, "step": 22184 }, { "epoch": 1.3771804581289961, "grad_norm": 0.5918369846159744, "learning_rate": 6.56073908465487e-05, "loss": 2.6942, "step": 22185 }, { "epoch": 1.377242535228754, "grad_norm": 0.5975794774031574, "learning_rate": 6.560395973671132e-05, "loss": 2.7723, "step": 22186 }, { "epoch": 1.377304612328512, "grad_norm": 0.6388627913115903, "learning_rate": 6.560052854546427e-05, "loss": 2.7594, "step": 22187 }, { "epoch": 1.37736668942827, "grad_norm": 0.6111171224206945, "learning_rate": 6.559709727282548e-05, "loss": 2.7971, "step": 22188 }, { "epoch": 1.3774287665280278, "grad_norm": 0.6626011195475305, "learning_rate": 6.559366591881284e-05, "loss": 2.8427, "step": 22189 }, { "epoch": 1.3774908436277857, "grad_norm": 0.581230098629774, "learning_rate": 6.559023448344429e-05, "loss": 2.7876, "step": 22190 }, { "epoch": 1.3775529207275437, "grad_norm": 0.6756458257029193, "learning_rate": 6.558680296673768e-05, "loss": 2.8603, "step": 22191 }, { "epoch": 1.3776149978273016, "grad_norm": 0.6040912339924419, "learning_rate": 6.558337136871093e-05, "loss": 2.748, "step": 22192 }, { "epoch": 1.3776770749270595, "grad_norm": 0.5745651601536639, "learning_rate": 6.557993968938195e-05, "loss": 2.8501, "step": 22193 }, { "epoch": 1.3777391520268174, "grad_norm": 0.595718858855037, "learning_rate": 6.557650792876865e-05, "loss": 2.6799, "step": 22194 }, { "epoch": 1.3778012291265753, "grad_norm": 0.5166888100226907, "learning_rate": 6.557307608688892e-05, "loss": 2.7868, "step": 22195 }, { "epoch": 1.3778633062263332, "grad_norm": 0.5497786954491806, "learning_rate": 6.556964416376068e-05, "loss": 2.7466, "step": 22196 }, { "epoch": 1.377925383326091, "grad_norm": 0.5157026752573406, "learning_rate": 6.556621215940182e-05, "loss": 2.8765, "step": 22197 }, { "epoch": 1.3779874604258489, "grad_norm": 0.5512037888445143, "learning_rate": 6.556278007383026e-05, "loss": 2.8208, "step": 22198 }, { "epoch": 1.3780495375256068, "grad_norm": 0.541300536785145, "learning_rate": 6.55593479070639e-05, "loss": 2.7942, "step": 22199 }, { "epoch": 1.3781116146253647, "grad_norm": 0.52995985112979, "learning_rate": 6.555591565912062e-05, "loss": 2.7558, "step": 22200 }, { "epoch": 1.3781736917251226, "grad_norm": 0.5396091343321484, "learning_rate": 6.555248333001838e-05, "loss": 2.7943, "step": 22201 }, { "epoch": 1.3782357688248805, "grad_norm": 0.5307980746931418, "learning_rate": 6.554905091977506e-05, "loss": 2.8108, "step": 22202 }, { "epoch": 1.3782978459246384, "grad_norm": 0.4978472290808257, "learning_rate": 6.554561842840854e-05, "loss": 2.7714, "step": 22203 }, { "epoch": 1.3783599230243964, "grad_norm": 0.5377568728966748, "learning_rate": 6.554218585593678e-05, "loss": 2.7691, "step": 22204 }, { "epoch": 1.3784220001241543, "grad_norm": 0.5411060409185425, "learning_rate": 6.553875320237765e-05, "loss": 2.7758, "step": 22205 }, { "epoch": 1.378484077223912, "grad_norm": 0.5509925599262255, "learning_rate": 6.553532046774909e-05, "loss": 2.9089, "step": 22206 }, { "epoch": 1.37854615432367, "grad_norm": 0.5936230473569596, "learning_rate": 6.553188765206896e-05, "loss": 2.7734, "step": 22207 }, { "epoch": 1.3786082314234278, "grad_norm": 0.516478064040768, "learning_rate": 6.552845475535523e-05, "loss": 2.8039, "step": 22208 }, { "epoch": 1.3786703085231857, "grad_norm": 0.584869148300127, "learning_rate": 6.552502177762576e-05, "loss": 2.8717, "step": 22209 }, { "epoch": 1.3787323856229436, "grad_norm": 0.5724103568828868, "learning_rate": 6.55215887188985e-05, "loss": 2.8385, "step": 22210 }, { "epoch": 1.3787944627227016, "grad_norm": 0.5186626372252325, "learning_rate": 6.551815557919133e-05, "loss": 2.7392, "step": 22211 }, { "epoch": 1.3788565398224595, "grad_norm": 0.5469843618852958, "learning_rate": 6.551472235852218e-05, "loss": 2.7454, "step": 22212 }, { "epoch": 1.3789186169222174, "grad_norm": 0.5041009689380768, "learning_rate": 6.551128905690897e-05, "loss": 2.7944, "step": 22213 }, { "epoch": 1.3789806940219753, "grad_norm": 0.5878993949362737, "learning_rate": 6.550785567436958e-05, "loss": 2.7042, "step": 22214 }, { "epoch": 1.3790427711217332, "grad_norm": 0.5197742438852427, "learning_rate": 6.550442221092197e-05, "loss": 2.7724, "step": 22215 }, { "epoch": 1.3791048482214912, "grad_norm": 0.5614960771612164, "learning_rate": 6.550098866658398e-05, "loss": 2.6704, "step": 22216 }, { "epoch": 1.379166925321249, "grad_norm": 0.6274443273810285, "learning_rate": 6.549755504137362e-05, "loss": 2.7617, "step": 22217 }, { "epoch": 1.379229002421007, "grad_norm": 0.5438113090327604, "learning_rate": 6.549412133530871e-05, "loss": 2.9219, "step": 22218 }, { "epoch": 1.379291079520765, "grad_norm": 0.5978338688223299, "learning_rate": 6.549068754840723e-05, "loss": 2.8389, "step": 22219 }, { "epoch": 1.3793531566205228, "grad_norm": 0.510981137340005, "learning_rate": 6.548725368068708e-05, "loss": 2.8405, "step": 22220 }, { "epoch": 1.3794152337202805, "grad_norm": 0.6074548939258366, "learning_rate": 6.548381973216615e-05, "loss": 2.8815, "step": 22221 }, { "epoch": 1.3794773108200384, "grad_norm": 0.5065330232284078, "learning_rate": 6.548038570286239e-05, "loss": 2.8142, "step": 22222 }, { "epoch": 1.3795393879197964, "grad_norm": 0.6254955472978264, "learning_rate": 6.54769515927937e-05, "loss": 2.7879, "step": 22223 }, { "epoch": 1.3796014650195543, "grad_norm": 0.5010657872390406, "learning_rate": 6.547351740197798e-05, "loss": 2.7615, "step": 22224 }, { "epoch": 1.3796635421193122, "grad_norm": 0.6056133466766668, "learning_rate": 6.547008313043317e-05, "loss": 2.7883, "step": 22225 }, { "epoch": 1.37972561921907, "grad_norm": 0.5344926102362033, "learning_rate": 6.546664877817719e-05, "loss": 2.8324, "step": 22226 }, { "epoch": 1.379787696318828, "grad_norm": 0.559647746162576, "learning_rate": 6.546321434522793e-05, "loss": 2.8359, "step": 22227 }, { "epoch": 1.379849773418586, "grad_norm": 0.5486856730476656, "learning_rate": 6.545977983160334e-05, "loss": 2.7968, "step": 22228 }, { "epoch": 1.3799118505183439, "grad_norm": 0.5508012700331485, "learning_rate": 6.545634523732132e-05, "loss": 2.7606, "step": 22229 }, { "epoch": 1.3799739276181016, "grad_norm": 0.5799083782456502, "learning_rate": 6.545291056239981e-05, "loss": 2.8179, "step": 22230 }, { "epoch": 1.3800360047178595, "grad_norm": 0.5198967587246153, "learning_rate": 6.54494758068567e-05, "loss": 2.745, "step": 22231 }, { "epoch": 1.3800980818176174, "grad_norm": 0.5567238996354824, "learning_rate": 6.544604097070992e-05, "loss": 2.789, "step": 22232 }, { "epoch": 1.3801601589173753, "grad_norm": 0.5410094828893928, "learning_rate": 6.544260605397739e-05, "loss": 2.8232, "step": 22233 }, { "epoch": 1.3802222360171332, "grad_norm": 0.5587841856395998, "learning_rate": 6.543917105667705e-05, "loss": 2.9009, "step": 22234 }, { "epoch": 1.3802843131168911, "grad_norm": 0.5627098751600154, "learning_rate": 6.54357359788268e-05, "loss": 2.8575, "step": 22235 }, { "epoch": 1.380346390216649, "grad_norm": 0.5358340850348216, "learning_rate": 6.543230082044457e-05, "loss": 2.8503, "step": 22236 }, { "epoch": 1.380408467316407, "grad_norm": 0.5669735139668627, "learning_rate": 6.542886558154829e-05, "loss": 2.8686, "step": 22237 }, { "epoch": 1.380470544416165, "grad_norm": 0.5101590739196853, "learning_rate": 6.542543026215586e-05, "loss": 2.8636, "step": 22238 }, { "epoch": 1.3805326215159228, "grad_norm": 0.5809566896007647, "learning_rate": 6.542199486228521e-05, "loss": 2.8248, "step": 22239 }, { "epoch": 1.3805946986156807, "grad_norm": 0.5451939221631104, "learning_rate": 6.54185593819543e-05, "loss": 2.7822, "step": 22240 }, { "epoch": 1.3806567757154387, "grad_norm": 0.5116935898514066, "learning_rate": 6.541512382118099e-05, "loss": 2.6539, "step": 22241 }, { "epoch": 1.3807188528151966, "grad_norm": 0.5648188478670965, "learning_rate": 6.541168817998326e-05, "loss": 2.834, "step": 22242 }, { "epoch": 1.3807809299149545, "grad_norm": 0.540078932547576, "learning_rate": 6.540825245837898e-05, "loss": 2.7939, "step": 22243 }, { "epoch": 1.3808430070147124, "grad_norm": 0.5764294323456424, "learning_rate": 6.540481665638614e-05, "loss": 2.8116, "step": 22244 }, { "epoch": 1.38090508411447, "grad_norm": 0.49524694544443776, "learning_rate": 6.540138077402262e-05, "loss": 2.7865, "step": 22245 }, { "epoch": 1.380967161214228, "grad_norm": 0.5487444939478313, "learning_rate": 6.539794481130635e-05, "loss": 2.7504, "step": 22246 }, { "epoch": 1.381029238313986, "grad_norm": 0.5447666124766831, "learning_rate": 6.539450876825527e-05, "loss": 2.8305, "step": 22247 }, { "epoch": 1.3810913154137439, "grad_norm": 0.5449895235682084, "learning_rate": 6.539107264488731e-05, "loss": 2.7908, "step": 22248 }, { "epoch": 1.3811533925135018, "grad_norm": 0.594455901696063, "learning_rate": 6.538763644122037e-05, "loss": 2.7796, "step": 22249 }, { "epoch": 1.3812154696132597, "grad_norm": 0.5423077463713462, "learning_rate": 6.538420015727238e-05, "loss": 2.8426, "step": 22250 }, { "epoch": 1.3812775467130176, "grad_norm": 0.599323440617389, "learning_rate": 6.538076379306132e-05, "loss": 2.8507, "step": 22251 }, { "epoch": 1.3813396238127755, "grad_norm": 0.5728174354852552, "learning_rate": 6.537732734860507e-05, "loss": 2.7909, "step": 22252 }, { "epoch": 1.3814017009125334, "grad_norm": 0.5495496466634741, "learning_rate": 6.537389082392157e-05, "loss": 2.8281, "step": 22253 }, { "epoch": 1.3814637780122911, "grad_norm": 0.644156514672166, "learning_rate": 6.537045421902874e-05, "loss": 2.8294, "step": 22254 }, { "epoch": 1.381525855112049, "grad_norm": 0.5551647497203389, "learning_rate": 6.536701753394454e-05, "loss": 2.7412, "step": 22255 }, { "epoch": 1.381587932211807, "grad_norm": 0.6165288718595013, "learning_rate": 6.536358076868688e-05, "loss": 2.8399, "step": 22256 }, { "epoch": 1.381650009311565, "grad_norm": 0.5859695680108148, "learning_rate": 6.536014392327366e-05, "loss": 2.7529, "step": 22257 }, { "epoch": 1.3817120864113228, "grad_norm": 0.5541664416329446, "learning_rate": 6.535670699772286e-05, "loss": 2.7546, "step": 22258 }, { "epoch": 1.3817741635110807, "grad_norm": 0.5929570675630158, "learning_rate": 6.535326999205238e-05, "loss": 2.7649, "step": 22259 }, { "epoch": 1.3818362406108387, "grad_norm": 0.563285226583002, "learning_rate": 6.534983290628017e-05, "loss": 2.7883, "step": 22260 }, { "epoch": 1.3818983177105966, "grad_norm": 0.5761777198624355, "learning_rate": 6.534639574042416e-05, "loss": 2.8081, "step": 22261 }, { "epoch": 1.3819603948103545, "grad_norm": 0.5377238417434402, "learning_rate": 6.534295849450227e-05, "loss": 2.7788, "step": 22262 }, { "epoch": 1.3820224719101124, "grad_norm": 0.5551537179714688, "learning_rate": 6.533952116853245e-05, "loss": 2.824, "step": 22263 }, { "epoch": 1.3820845490098703, "grad_norm": 0.548811725629376, "learning_rate": 6.533608376253262e-05, "loss": 2.84, "step": 22264 }, { "epoch": 1.3821466261096282, "grad_norm": 0.5164304857439018, "learning_rate": 6.533264627652072e-05, "loss": 2.8365, "step": 22265 }, { "epoch": 1.3822087032093862, "grad_norm": 0.5463547820847509, "learning_rate": 6.532920871051466e-05, "loss": 2.8468, "step": 22266 }, { "epoch": 1.382270780309144, "grad_norm": 0.5643008861692299, "learning_rate": 6.532577106453243e-05, "loss": 2.7688, "step": 22267 }, { "epoch": 1.382332857408902, "grad_norm": 0.5181034804731076, "learning_rate": 6.53223333385919e-05, "loss": 2.8015, "step": 22268 }, { "epoch": 1.3823949345086597, "grad_norm": 0.589329779822797, "learning_rate": 6.531889553271107e-05, "loss": 2.7571, "step": 22269 }, { "epoch": 1.3824570116084176, "grad_norm": 0.5862209905836474, "learning_rate": 6.531545764690782e-05, "loss": 2.8018, "step": 22270 }, { "epoch": 1.3825190887081755, "grad_norm": 0.6176404859990516, "learning_rate": 6.531201968120012e-05, "loss": 2.8149, "step": 22271 }, { "epoch": 1.3825811658079334, "grad_norm": 0.5978583698452183, "learning_rate": 6.53085816356059e-05, "loss": 2.7963, "step": 22272 }, { "epoch": 1.3826432429076914, "grad_norm": 0.647962028187073, "learning_rate": 6.530514351014309e-05, "loss": 2.8665, "step": 22273 }, { "epoch": 1.3827053200074493, "grad_norm": 0.6105700048554947, "learning_rate": 6.530170530482963e-05, "loss": 2.8427, "step": 22274 }, { "epoch": 1.3827673971072072, "grad_norm": 0.5700581649750953, "learning_rate": 6.529826701968345e-05, "loss": 2.8178, "step": 22275 }, { "epoch": 1.3828294742069651, "grad_norm": 0.6632196799953282, "learning_rate": 6.529482865472252e-05, "loss": 2.7972, "step": 22276 }, { "epoch": 1.382891551306723, "grad_norm": 0.6345275667257899, "learning_rate": 6.529139020996473e-05, "loss": 2.8443, "step": 22277 }, { "epoch": 1.3829536284064807, "grad_norm": 0.6277224375189708, "learning_rate": 6.528795168542805e-05, "loss": 2.958, "step": 22278 }, { "epoch": 1.3830157055062386, "grad_norm": 0.6496797176039925, "learning_rate": 6.528451308113042e-05, "loss": 2.8842, "step": 22279 }, { "epoch": 1.3830777826059966, "grad_norm": 0.5856689855124014, "learning_rate": 6.528107439708978e-05, "loss": 2.7894, "step": 22280 }, { "epoch": 1.3831398597057545, "grad_norm": 0.6106499514757869, "learning_rate": 6.527763563332406e-05, "loss": 2.7842, "step": 22281 }, { "epoch": 1.3832019368055124, "grad_norm": 0.5720483673460509, "learning_rate": 6.527419678985119e-05, "loss": 2.8229, "step": 22282 }, { "epoch": 1.3832640139052703, "grad_norm": 0.5977675888970947, "learning_rate": 6.527075786668915e-05, "loss": 2.7906, "step": 22283 }, { "epoch": 1.3833260910050282, "grad_norm": 0.5884379630307297, "learning_rate": 6.526731886385585e-05, "loss": 2.8406, "step": 22284 }, { "epoch": 1.3833881681047862, "grad_norm": 0.5780173536553284, "learning_rate": 6.526387978136926e-05, "loss": 2.82, "step": 22285 }, { "epoch": 1.383450245204544, "grad_norm": 0.5639556751909616, "learning_rate": 6.526044061924727e-05, "loss": 2.8257, "step": 22286 }, { "epoch": 1.383512322304302, "grad_norm": 0.6035930003715282, "learning_rate": 6.525700137750789e-05, "loss": 2.8082, "step": 22287 }, { "epoch": 1.38357439940406, "grad_norm": 0.5358154803021531, "learning_rate": 6.5253562056169e-05, "loss": 2.8003, "step": 22288 }, { "epoch": 1.3836364765038178, "grad_norm": 0.5458071773998573, "learning_rate": 6.52501226552486e-05, "loss": 2.7585, "step": 22289 }, { "epoch": 1.3836985536035757, "grad_norm": 0.6164656896905358, "learning_rate": 6.52466831747646e-05, "loss": 2.8755, "step": 22290 }, { "epoch": 1.3837606307033337, "grad_norm": 0.6152455162371677, "learning_rate": 6.524324361473495e-05, "loss": 2.7666, "step": 22291 }, { "epoch": 1.3838227078030914, "grad_norm": 0.5872276344408798, "learning_rate": 6.523980397517759e-05, "loss": 2.7932, "step": 22292 }, { "epoch": 1.3838847849028493, "grad_norm": 0.5491228771503002, "learning_rate": 6.523636425611049e-05, "loss": 2.8531, "step": 22293 }, { "epoch": 1.3839468620026072, "grad_norm": 0.6259600418457869, "learning_rate": 6.523292445755157e-05, "loss": 2.9293, "step": 22294 }, { "epoch": 1.384008939102365, "grad_norm": 0.5677748358269352, "learning_rate": 6.522948457951877e-05, "loss": 2.8933, "step": 22295 }, { "epoch": 1.384071016202123, "grad_norm": 0.6038925080506284, "learning_rate": 6.522604462203009e-05, "loss": 2.7941, "step": 22296 }, { "epoch": 1.384133093301881, "grad_norm": 0.5632444594559847, "learning_rate": 6.52226045851034e-05, "loss": 2.7528, "step": 22297 }, { "epoch": 1.3841951704016389, "grad_norm": 0.6312253380669657, "learning_rate": 6.521916446875669e-05, "loss": 2.739, "step": 22298 }, { "epoch": 1.3842572475013968, "grad_norm": 0.5481074478488316, "learning_rate": 6.521572427300793e-05, "loss": 2.7049, "step": 22299 }, { "epoch": 1.3843193246011547, "grad_norm": 0.5973193646561691, "learning_rate": 6.5212283997875e-05, "loss": 2.9235, "step": 22300 }, { "epoch": 1.3843814017009124, "grad_norm": 0.593318054287344, "learning_rate": 6.520884364337593e-05, "loss": 2.7929, "step": 22301 }, { "epoch": 1.3844434788006703, "grad_norm": 0.5625516814755587, "learning_rate": 6.520540320952863e-05, "loss": 2.7971, "step": 22302 }, { "epoch": 1.3845055559004282, "grad_norm": 0.6282179958147514, "learning_rate": 6.520196269635104e-05, "loss": 2.7885, "step": 22303 }, { "epoch": 1.3845676330001861, "grad_norm": 0.5317109856723843, "learning_rate": 6.519852210386112e-05, "loss": 2.8794, "step": 22304 }, { "epoch": 1.384629710099944, "grad_norm": 0.6462449382384531, "learning_rate": 6.519508143207682e-05, "loss": 2.7579, "step": 22305 }, { "epoch": 1.384691787199702, "grad_norm": 0.5649814706903532, "learning_rate": 6.519164068101609e-05, "loss": 2.8566, "step": 22306 }, { "epoch": 1.38475386429946, "grad_norm": 0.6072389382383124, "learning_rate": 6.518819985069689e-05, "loss": 2.7952, "step": 22307 }, { "epoch": 1.3848159413992178, "grad_norm": 0.5685201201288059, "learning_rate": 6.518475894113715e-05, "loss": 2.7812, "step": 22308 }, { "epoch": 1.3848780184989757, "grad_norm": 0.5790205025510066, "learning_rate": 6.518131795235486e-05, "loss": 2.7668, "step": 22309 }, { "epoch": 1.3849400955987337, "grad_norm": 0.5719391083351343, "learning_rate": 6.517787688436792e-05, "loss": 2.8254, "step": 22310 }, { "epoch": 1.3850021726984916, "grad_norm": 0.575807695544931, "learning_rate": 6.517443573719434e-05, "loss": 2.8613, "step": 22311 }, { "epoch": 1.3850642497982495, "grad_norm": 0.6253034570241218, "learning_rate": 6.517099451085204e-05, "loss": 2.8603, "step": 22312 }, { "epoch": 1.3851263268980074, "grad_norm": 0.5043360629172794, "learning_rate": 6.516755320535897e-05, "loss": 2.8555, "step": 22313 }, { "epoch": 1.3851884039977653, "grad_norm": 0.5797272979565102, "learning_rate": 6.51641118207331e-05, "loss": 2.912, "step": 22314 }, { "epoch": 1.3852504810975232, "grad_norm": 0.5133006563974297, "learning_rate": 6.516067035699235e-05, "loss": 2.8557, "step": 22315 }, { "epoch": 1.385312558197281, "grad_norm": 0.5405981608437497, "learning_rate": 6.515722881415473e-05, "loss": 2.698, "step": 22316 }, { "epoch": 1.3853746352970389, "grad_norm": 0.5350195160095927, "learning_rate": 6.515378719223816e-05, "loss": 2.8559, "step": 22317 }, { "epoch": 1.3854367123967968, "grad_norm": 0.5367789899622237, "learning_rate": 6.51503454912606e-05, "loss": 2.8421, "step": 22318 }, { "epoch": 1.3854987894965547, "grad_norm": 0.5080035217174348, "learning_rate": 6.514690371124001e-05, "loss": 2.8228, "step": 22319 }, { "epoch": 1.3855608665963126, "grad_norm": 0.5332946698757711, "learning_rate": 6.514346185219436e-05, "loss": 2.7447, "step": 22320 }, { "epoch": 1.3856229436960705, "grad_norm": 0.5470048750201612, "learning_rate": 6.514001991414159e-05, "loss": 2.9101, "step": 22321 }, { "epoch": 1.3856850207958284, "grad_norm": 0.5849451435485199, "learning_rate": 6.513657789709963e-05, "loss": 2.7543, "step": 22322 }, { "epoch": 1.3857470978955864, "grad_norm": 0.5403435729817805, "learning_rate": 6.513313580108651e-05, "loss": 2.8129, "step": 22323 }, { "epoch": 1.3858091749953443, "grad_norm": 0.4925779428867424, "learning_rate": 6.512969362612013e-05, "loss": 2.8007, "step": 22324 }, { "epoch": 1.385871252095102, "grad_norm": 0.5084621968335005, "learning_rate": 6.512625137221846e-05, "loss": 2.7319, "step": 22325 }, { "epoch": 1.38593332919486, "grad_norm": 0.5355715288960821, "learning_rate": 6.512280903939947e-05, "loss": 2.8732, "step": 22326 }, { "epoch": 1.3859954062946178, "grad_norm": 0.49665272377378683, "learning_rate": 6.511936662768112e-05, "loss": 2.7863, "step": 22327 }, { "epoch": 1.3860574833943757, "grad_norm": 0.5181472815863692, "learning_rate": 6.511592413708136e-05, "loss": 2.8384, "step": 22328 }, { "epoch": 1.3861195604941337, "grad_norm": 0.5232085111402606, "learning_rate": 6.511248156761815e-05, "loss": 2.842, "step": 22329 }, { "epoch": 1.3861816375938916, "grad_norm": 0.5419913626999636, "learning_rate": 6.510903891930946e-05, "loss": 2.7715, "step": 22330 }, { "epoch": 1.3862437146936495, "grad_norm": 0.5253130206684752, "learning_rate": 6.510559619217323e-05, "loss": 2.7647, "step": 22331 }, { "epoch": 1.3863057917934074, "grad_norm": 0.5836328322309283, "learning_rate": 6.510215338622745e-05, "loss": 2.7622, "step": 22332 }, { "epoch": 1.3863678688931653, "grad_norm": 0.5862861867363048, "learning_rate": 6.509871050149007e-05, "loss": 2.7332, "step": 22333 }, { "epoch": 1.3864299459929232, "grad_norm": 0.5602979738785376, "learning_rate": 6.509526753797906e-05, "loss": 2.7752, "step": 22334 }, { "epoch": 1.3864920230926812, "grad_norm": 0.5428593202821544, "learning_rate": 6.509182449571238e-05, "loss": 2.8524, "step": 22335 }, { "epoch": 1.386554100192439, "grad_norm": 0.5384430851040596, "learning_rate": 6.508838137470796e-05, "loss": 2.8224, "step": 22336 }, { "epoch": 1.386616177292197, "grad_norm": 0.47296058693108806, "learning_rate": 6.508493817498383e-05, "loss": 2.8257, "step": 22337 }, { "epoch": 1.386678254391955, "grad_norm": 0.5722627607150215, "learning_rate": 6.508149489655787e-05, "loss": 2.7839, "step": 22338 }, { "epoch": 1.3867403314917128, "grad_norm": 0.4618867378137107, "learning_rate": 6.507805153944813e-05, "loss": 2.7828, "step": 22339 }, { "epoch": 1.3868024085914705, "grad_norm": 0.5805878631245714, "learning_rate": 6.50746081036725e-05, "loss": 2.8526, "step": 22340 }, { "epoch": 1.3868644856912284, "grad_norm": 0.5056846051631262, "learning_rate": 6.5071164589249e-05, "loss": 2.7609, "step": 22341 }, { "epoch": 1.3869265627909864, "grad_norm": 0.5651514558006071, "learning_rate": 6.506772099619557e-05, "loss": 2.8725, "step": 22342 }, { "epoch": 1.3869886398907443, "grad_norm": 0.5452987024261382, "learning_rate": 6.50642773245302e-05, "loss": 2.7959, "step": 22343 }, { "epoch": 1.3870507169905022, "grad_norm": 0.619956072204816, "learning_rate": 6.506083357427081e-05, "loss": 2.8996, "step": 22344 }, { "epoch": 1.3871127940902601, "grad_norm": 0.5717053323251835, "learning_rate": 6.505738974543541e-05, "loss": 2.8441, "step": 22345 }, { "epoch": 1.387174871190018, "grad_norm": 0.5722402365198663, "learning_rate": 6.505394583804195e-05, "loss": 2.7917, "step": 22346 }, { "epoch": 1.387236948289776, "grad_norm": 0.5801989496957293, "learning_rate": 6.50505018521084e-05, "loss": 2.8208, "step": 22347 }, { "epoch": 1.3872990253895339, "grad_norm": 0.5403921618184, "learning_rate": 6.504705778765272e-05, "loss": 2.8129, "step": 22348 }, { "epoch": 1.3873611024892916, "grad_norm": 0.6270453839737397, "learning_rate": 6.504361364469289e-05, "loss": 2.7965, "step": 22349 }, { "epoch": 1.3874231795890495, "grad_norm": 0.6014361310727435, "learning_rate": 6.504016942324689e-05, "loss": 2.8921, "step": 22350 }, { "epoch": 1.3874852566888074, "grad_norm": 0.6536627358205405, "learning_rate": 6.503672512333267e-05, "loss": 2.8121, "step": 22351 }, { "epoch": 1.3875473337885653, "grad_norm": 0.6242823056366136, "learning_rate": 6.503328074496819e-05, "loss": 2.8305, "step": 22352 }, { "epoch": 1.3876094108883232, "grad_norm": 0.6449355966736021, "learning_rate": 6.502983628817145e-05, "loss": 2.8671, "step": 22353 }, { "epoch": 1.3876714879880812, "grad_norm": 0.5541932496408994, "learning_rate": 6.502639175296042e-05, "loss": 2.9048, "step": 22354 }, { "epoch": 1.387733565087839, "grad_norm": 0.659264023648664, "learning_rate": 6.502294713935303e-05, "loss": 2.7202, "step": 22355 }, { "epoch": 1.387795642187597, "grad_norm": 0.5991277095861545, "learning_rate": 6.501950244736728e-05, "loss": 2.8053, "step": 22356 }, { "epoch": 1.387857719287355, "grad_norm": 0.598642346055752, "learning_rate": 6.501605767702115e-05, "loss": 2.8655, "step": 22357 }, { "epoch": 1.3879197963871128, "grad_norm": 0.6442925444645113, "learning_rate": 6.50126128283326e-05, "loss": 2.8092, "step": 22358 }, { "epoch": 1.3879818734868707, "grad_norm": 0.5541161294231854, "learning_rate": 6.50091679013196e-05, "loss": 2.7896, "step": 22359 }, { "epoch": 1.3880439505866287, "grad_norm": 0.592467478893781, "learning_rate": 6.500572289600012e-05, "loss": 2.7886, "step": 22360 }, { "epoch": 1.3881060276863866, "grad_norm": 0.5847032644979606, "learning_rate": 6.500227781239215e-05, "loss": 2.7873, "step": 22361 }, { "epoch": 1.3881681047861445, "grad_norm": 0.5929129120642388, "learning_rate": 6.499883265051366e-05, "loss": 2.8293, "step": 22362 }, { "epoch": 1.3882301818859024, "grad_norm": 0.5525891278300163, "learning_rate": 6.49953874103826e-05, "loss": 2.8367, "step": 22363 }, { "epoch": 1.38829225898566, "grad_norm": 0.5912019616142864, "learning_rate": 6.499194209201699e-05, "loss": 2.8177, "step": 22364 }, { "epoch": 1.388354336085418, "grad_norm": 0.5531368879344635, "learning_rate": 6.498849669543475e-05, "loss": 2.753, "step": 22365 }, { "epoch": 1.388416413185176, "grad_norm": 0.5994112016566704, "learning_rate": 6.498505122065391e-05, "loss": 2.836, "step": 22366 }, { "epoch": 1.3884784902849339, "grad_norm": 0.5492293225243446, "learning_rate": 6.49816056676924e-05, "loss": 2.7843, "step": 22367 }, { "epoch": 1.3885405673846918, "grad_norm": 0.620065449636415, "learning_rate": 6.497816003656824e-05, "loss": 2.866, "step": 22368 }, { "epoch": 1.3886026444844497, "grad_norm": 0.6499137848750515, "learning_rate": 6.497471432729936e-05, "loss": 2.7411, "step": 22369 }, { "epoch": 1.3886647215842076, "grad_norm": 0.6060269397674181, "learning_rate": 6.497126853990377e-05, "loss": 2.7822, "step": 22370 }, { "epoch": 1.3887267986839655, "grad_norm": 0.6449736954995549, "learning_rate": 6.496782267439943e-05, "loss": 2.8973, "step": 22371 }, { "epoch": 1.3887888757837235, "grad_norm": 0.6271560198035349, "learning_rate": 6.496437673080435e-05, "loss": 2.8144, "step": 22372 }, { "epoch": 1.3888509528834811, "grad_norm": 0.620747931791525, "learning_rate": 6.496093070913647e-05, "loss": 2.8171, "step": 22373 }, { "epoch": 1.388913029983239, "grad_norm": 0.671851157719054, "learning_rate": 6.495748460941376e-05, "loss": 2.8513, "step": 22374 }, { "epoch": 1.388975107082997, "grad_norm": 0.580733292113107, "learning_rate": 6.495403843165425e-05, "loss": 2.7589, "step": 22375 }, { "epoch": 1.389037184182755, "grad_norm": 0.6078389198817733, "learning_rate": 6.495059217587587e-05, "loss": 2.7299, "step": 22376 }, { "epoch": 1.3890992612825128, "grad_norm": 0.5654577007832202, "learning_rate": 6.494714584209665e-05, "loss": 2.7545, "step": 22377 }, { "epoch": 1.3891613383822707, "grad_norm": 0.6811284980565089, "learning_rate": 6.494369943033451e-05, "loss": 2.7414, "step": 22378 }, { "epoch": 1.3892234154820287, "grad_norm": 0.5446157928907289, "learning_rate": 6.494025294060749e-05, "loss": 2.7581, "step": 22379 }, { "epoch": 1.3892854925817866, "grad_norm": 0.6259424018763149, "learning_rate": 6.493680637293353e-05, "loss": 2.7958, "step": 22380 }, { "epoch": 1.3893475696815445, "grad_norm": 0.5393715401071061, "learning_rate": 6.493335972733062e-05, "loss": 2.7015, "step": 22381 }, { "epoch": 1.3894096467813024, "grad_norm": 0.652003399354463, "learning_rate": 6.492991300381676e-05, "loss": 2.8385, "step": 22382 }, { "epoch": 1.3894717238810603, "grad_norm": 0.5616630844950937, "learning_rate": 6.49264662024099e-05, "loss": 2.7653, "step": 22383 }, { "epoch": 1.3895338009808182, "grad_norm": 0.6660437032648362, "learning_rate": 6.492301932312806e-05, "loss": 2.8284, "step": 22384 }, { "epoch": 1.3895958780805762, "grad_norm": 0.6206210754977302, "learning_rate": 6.49195723659892e-05, "loss": 2.8057, "step": 22385 }, { "epoch": 1.389657955180334, "grad_norm": 0.5898298047166695, "learning_rate": 6.491612533101132e-05, "loss": 2.7601, "step": 22386 }, { "epoch": 1.389720032280092, "grad_norm": 0.610357226536219, "learning_rate": 6.491267821821236e-05, "loss": 2.8272, "step": 22387 }, { "epoch": 1.3897821093798497, "grad_norm": 0.6225003563064107, "learning_rate": 6.490923102761037e-05, "loss": 2.8135, "step": 22388 }, { "epoch": 1.3898441864796076, "grad_norm": 0.6418861675312434, "learning_rate": 6.49057837592233e-05, "loss": 2.7933, "step": 22389 }, { "epoch": 1.3899062635793655, "grad_norm": 0.5650578200031806, "learning_rate": 6.490233641306914e-05, "loss": 2.7898, "step": 22390 }, { "epoch": 1.3899683406791234, "grad_norm": 0.543440881096718, "learning_rate": 6.489888898916585e-05, "loss": 2.7345, "step": 22391 }, { "epoch": 1.3900304177788814, "grad_norm": 0.6377928865100047, "learning_rate": 6.489544148753145e-05, "loss": 2.8777, "step": 22392 }, { "epoch": 1.3900924948786393, "grad_norm": 0.5834415681653229, "learning_rate": 6.489199390818392e-05, "loss": 2.8142, "step": 22393 }, { "epoch": 1.3901545719783972, "grad_norm": 0.6054362626742891, "learning_rate": 6.488854625114125e-05, "loss": 2.8814, "step": 22394 }, { "epoch": 1.3902166490781551, "grad_norm": 0.5503867339408873, "learning_rate": 6.48850985164214e-05, "loss": 2.8095, "step": 22395 }, { "epoch": 1.390278726177913, "grad_norm": 0.5166699196600284, "learning_rate": 6.488165070404237e-05, "loss": 2.8131, "step": 22396 }, { "epoch": 1.3903408032776707, "grad_norm": 0.5400614673457621, "learning_rate": 6.48782028140222e-05, "loss": 2.8291, "step": 22397 }, { "epoch": 1.3904028803774287, "grad_norm": 0.5764992769721494, "learning_rate": 6.48747548463788e-05, "loss": 2.887, "step": 22398 }, { "epoch": 1.3904649574771866, "grad_norm": 0.6013256201110188, "learning_rate": 6.487130680113019e-05, "loss": 2.7304, "step": 22399 }, { "epoch": 1.3905270345769445, "grad_norm": 0.5536223648518465, "learning_rate": 6.486785867829439e-05, "loss": 2.765, "step": 22400 }, { "epoch": 1.3905891116767024, "grad_norm": 0.5829228733878443, "learning_rate": 6.486441047788933e-05, "loss": 2.7273, "step": 22401 }, { "epoch": 1.3906511887764603, "grad_norm": 0.5401188272407301, "learning_rate": 6.486096219993306e-05, "loss": 2.6623, "step": 22402 }, { "epoch": 1.3907132658762182, "grad_norm": 0.5540160561259283, "learning_rate": 6.485751384444352e-05, "loss": 2.8048, "step": 22403 }, { "epoch": 1.3907753429759762, "grad_norm": 0.5423719845310142, "learning_rate": 6.485406541143874e-05, "loss": 2.8265, "step": 22404 }, { "epoch": 1.390837420075734, "grad_norm": 0.5495316418604881, "learning_rate": 6.485061690093668e-05, "loss": 2.8641, "step": 22405 }, { "epoch": 1.390899497175492, "grad_norm": 0.5968031185865738, "learning_rate": 6.484716831295537e-05, "loss": 2.865, "step": 22406 }, { "epoch": 1.39096157427525, "grad_norm": 0.5812434555013423, "learning_rate": 6.484371964751275e-05, "loss": 2.8206, "step": 22407 }, { "epoch": 1.3910236513750078, "grad_norm": 0.5667241223976973, "learning_rate": 6.484027090462687e-05, "loss": 2.6867, "step": 22408 }, { "epoch": 1.3910857284747657, "grad_norm": 0.5409177560800691, "learning_rate": 6.483682208431567e-05, "loss": 2.7783, "step": 22409 }, { "epoch": 1.3911478055745237, "grad_norm": 0.6594389003145493, "learning_rate": 6.483337318659718e-05, "loss": 2.8575, "step": 22410 }, { "epoch": 1.3912098826742816, "grad_norm": 0.5968854334054168, "learning_rate": 6.482992421148937e-05, "loss": 2.7443, "step": 22411 }, { "epoch": 1.3912719597740393, "grad_norm": 0.5939405035283983, "learning_rate": 6.482647515901026e-05, "loss": 2.8535, "step": 22412 }, { "epoch": 1.3913340368737972, "grad_norm": 0.6508466021454727, "learning_rate": 6.482302602917781e-05, "loss": 2.7693, "step": 22413 }, { "epoch": 1.3913961139735551, "grad_norm": 0.6158213645957528, "learning_rate": 6.481957682201006e-05, "loss": 2.8261, "step": 22414 }, { "epoch": 1.391458191073313, "grad_norm": 0.5650741434398718, "learning_rate": 6.481612753752496e-05, "loss": 2.9133, "step": 22415 }, { "epoch": 1.391520268173071, "grad_norm": 0.6058218387902077, "learning_rate": 6.481267817574054e-05, "loss": 2.7951, "step": 22416 }, { "epoch": 1.3915823452728289, "grad_norm": 0.5938096518416618, "learning_rate": 6.480922873667477e-05, "loss": 2.8353, "step": 22417 }, { "epoch": 1.3916444223725868, "grad_norm": 0.5129166189028947, "learning_rate": 6.480577922034566e-05, "loss": 2.8444, "step": 22418 }, { "epoch": 1.3917064994723447, "grad_norm": 0.5887207304609738, "learning_rate": 6.480232962677121e-05, "loss": 2.6944, "step": 22419 }, { "epoch": 1.3917685765721026, "grad_norm": 0.5510487008562238, "learning_rate": 6.479887995596942e-05, "loss": 2.8182, "step": 22420 }, { "epoch": 1.3918306536718603, "grad_norm": 0.5836886164148758, "learning_rate": 6.479543020795827e-05, "loss": 2.7379, "step": 22421 }, { "epoch": 1.3918927307716182, "grad_norm": 0.5503299839463524, "learning_rate": 6.479198038275578e-05, "loss": 2.7626, "step": 22422 }, { "epoch": 1.3919548078713762, "grad_norm": 0.6008581521529324, "learning_rate": 6.478853048037991e-05, "loss": 2.8177, "step": 22423 }, { "epoch": 1.392016884971134, "grad_norm": 0.5555502669427069, "learning_rate": 6.478508050084871e-05, "loss": 2.8177, "step": 22424 }, { "epoch": 1.392078962070892, "grad_norm": 0.6267280531130699, "learning_rate": 6.478163044418015e-05, "loss": 2.8823, "step": 22425 }, { "epoch": 1.39214103917065, "grad_norm": 0.5804295615192177, "learning_rate": 6.477818031039225e-05, "loss": 2.8166, "step": 22426 }, { "epoch": 1.3922031162704078, "grad_norm": 0.5886534497491679, "learning_rate": 6.477473009950299e-05, "loss": 2.7641, "step": 22427 }, { "epoch": 1.3922651933701657, "grad_norm": 0.5492823068843322, "learning_rate": 6.477127981153036e-05, "loss": 2.8853, "step": 22428 }, { "epoch": 1.3923272704699237, "grad_norm": 0.5857276286331635, "learning_rate": 6.476782944649238e-05, "loss": 2.7654, "step": 22429 }, { "epoch": 1.3923893475696816, "grad_norm": 0.5674577358237195, "learning_rate": 6.476437900440705e-05, "loss": 2.7102, "step": 22430 }, { "epoch": 1.3924514246694395, "grad_norm": 0.5796076089760376, "learning_rate": 6.476092848529238e-05, "loss": 2.7285, "step": 22431 }, { "epoch": 1.3925135017691974, "grad_norm": 0.6712914749559434, "learning_rate": 6.475747788916637e-05, "loss": 2.8375, "step": 22432 }, { "epoch": 1.3925755788689553, "grad_norm": 0.5442038286448747, "learning_rate": 6.4754027216047e-05, "loss": 2.8655, "step": 22433 }, { "epoch": 1.3926376559687133, "grad_norm": 0.6534313569402707, "learning_rate": 6.47505764659523e-05, "loss": 2.8737, "step": 22434 }, { "epoch": 1.3926997330684712, "grad_norm": 0.5643142476550688, "learning_rate": 6.474712563890024e-05, "loss": 2.7852, "step": 22435 }, { "epoch": 1.3927618101682289, "grad_norm": 0.6794564084476693, "learning_rate": 6.474367473490888e-05, "loss": 2.8423, "step": 22436 }, { "epoch": 1.3928238872679868, "grad_norm": 0.6237666119444688, "learning_rate": 6.474022375399616e-05, "loss": 2.804, "step": 22437 }, { "epoch": 1.3928859643677447, "grad_norm": 0.6343000612015615, "learning_rate": 6.473677269618012e-05, "loss": 2.8136, "step": 22438 }, { "epoch": 1.3929480414675026, "grad_norm": 0.6728194939178517, "learning_rate": 6.473332156147877e-05, "loss": 2.9125, "step": 22439 }, { "epoch": 1.3930101185672605, "grad_norm": 0.6193648693360978, "learning_rate": 6.47298703499101e-05, "loss": 2.8574, "step": 22440 }, { "epoch": 1.3930721956670185, "grad_norm": 0.5900862239613668, "learning_rate": 6.472641906149211e-05, "loss": 2.8484, "step": 22441 }, { "epoch": 1.3931342727667764, "grad_norm": 0.5967733805966016, "learning_rate": 6.472296769624283e-05, "loss": 2.7688, "step": 22442 }, { "epoch": 1.3931963498665343, "grad_norm": 0.6284700622544893, "learning_rate": 6.471951625418025e-05, "loss": 2.84, "step": 22443 }, { "epoch": 1.3932584269662922, "grad_norm": 0.5832743586519651, "learning_rate": 6.471606473532238e-05, "loss": 2.7848, "step": 22444 }, { "epoch": 1.39332050406605, "grad_norm": 0.6710380397849816, "learning_rate": 6.471261313968723e-05, "loss": 2.8851, "step": 22445 }, { "epoch": 1.3933825811658078, "grad_norm": 0.6011306153652091, "learning_rate": 6.470916146729279e-05, "loss": 2.8002, "step": 22446 }, { "epoch": 1.3934446582655657, "grad_norm": 0.7545307320415072, "learning_rate": 6.47057097181571e-05, "loss": 2.8428, "step": 22447 }, { "epoch": 1.3935067353653237, "grad_norm": 0.6141361668008434, "learning_rate": 6.470225789229815e-05, "loss": 2.8249, "step": 22448 }, { "epoch": 1.3935688124650816, "grad_norm": 0.6089710706391659, "learning_rate": 6.469880598973395e-05, "loss": 2.7565, "step": 22449 }, { "epoch": 1.3936308895648395, "grad_norm": 0.6947801673848812, "learning_rate": 6.46953540104825e-05, "loss": 2.7926, "step": 22450 }, { "epoch": 1.3936929666645974, "grad_norm": 0.5904022503180637, "learning_rate": 6.469190195456184e-05, "loss": 2.8823, "step": 22451 }, { "epoch": 1.3937550437643553, "grad_norm": 0.6867824190762247, "learning_rate": 6.468844982198996e-05, "loss": 2.843, "step": 22452 }, { "epoch": 1.3938171208641132, "grad_norm": 0.6098600560050326, "learning_rate": 6.468499761278485e-05, "loss": 2.8782, "step": 22453 }, { "epoch": 1.3938791979638712, "grad_norm": 0.6348373842025844, "learning_rate": 6.468154532696457e-05, "loss": 2.8075, "step": 22454 }, { "epoch": 1.393941275063629, "grad_norm": 0.6564407298533349, "learning_rate": 6.467809296454709e-05, "loss": 2.8218, "step": 22455 }, { "epoch": 1.394003352163387, "grad_norm": 0.6011439514615442, "learning_rate": 6.467464052555044e-05, "loss": 2.7547, "step": 22456 }, { "epoch": 1.394065429263145, "grad_norm": 0.6599238511429963, "learning_rate": 6.467118800999262e-05, "loss": 2.8034, "step": 22457 }, { "epoch": 1.3941275063629028, "grad_norm": 0.708199436958755, "learning_rate": 6.466773541789166e-05, "loss": 2.8736, "step": 22458 }, { "epoch": 1.3941895834626608, "grad_norm": 0.5908197071923552, "learning_rate": 6.466428274926556e-05, "loss": 2.7772, "step": 22459 }, { "epoch": 1.3942516605624184, "grad_norm": 0.6863451057291816, "learning_rate": 6.466083000413234e-05, "loss": 2.8681, "step": 22460 }, { "epoch": 1.3943137376621764, "grad_norm": 0.5829089926032058, "learning_rate": 6.465737718251e-05, "loss": 2.7421, "step": 22461 }, { "epoch": 1.3943758147619343, "grad_norm": 0.645118954061595, "learning_rate": 6.465392428441656e-05, "loss": 2.8342, "step": 22462 }, { "epoch": 1.3944378918616922, "grad_norm": 0.6268603294839004, "learning_rate": 6.465047130987006e-05, "loss": 2.7258, "step": 22463 }, { "epoch": 1.3944999689614501, "grad_norm": 0.6088209078417405, "learning_rate": 6.464701825888848e-05, "loss": 2.7376, "step": 22464 }, { "epoch": 1.394562046061208, "grad_norm": 0.6434947841243204, "learning_rate": 6.464356513148987e-05, "loss": 2.9314, "step": 22465 }, { "epoch": 1.394624123160966, "grad_norm": 0.5707601854231753, "learning_rate": 6.46401119276922e-05, "loss": 2.7708, "step": 22466 }, { "epoch": 1.3946862002607239, "grad_norm": 0.6165931108999321, "learning_rate": 6.463665864751353e-05, "loss": 2.8336, "step": 22467 }, { "epoch": 1.3947482773604818, "grad_norm": 0.5761791939997932, "learning_rate": 6.463320529097186e-05, "loss": 2.8111, "step": 22468 }, { "epoch": 1.3948103544602395, "grad_norm": 0.5439842388257545, "learning_rate": 6.46297518580852e-05, "loss": 2.8167, "step": 22469 }, { "epoch": 1.3948724315599974, "grad_norm": 0.6344535412579756, "learning_rate": 6.462629834887157e-05, "loss": 2.8135, "step": 22470 }, { "epoch": 1.3949345086597553, "grad_norm": 0.4963815860306562, "learning_rate": 6.462284476334899e-05, "loss": 2.7738, "step": 22471 }, { "epoch": 1.3949965857595132, "grad_norm": 0.5401059983558704, "learning_rate": 6.461939110153549e-05, "loss": 2.7678, "step": 22472 }, { "epoch": 1.3950586628592712, "grad_norm": 0.588445256081313, "learning_rate": 6.461593736344906e-05, "loss": 2.8241, "step": 22473 }, { "epoch": 1.395120739959029, "grad_norm": 0.5091769740387578, "learning_rate": 6.461248354910775e-05, "loss": 2.7975, "step": 22474 }, { "epoch": 1.395182817058787, "grad_norm": 0.5895474428591344, "learning_rate": 6.460902965852956e-05, "loss": 2.8458, "step": 22475 }, { "epoch": 1.395244894158545, "grad_norm": 0.5561804224282625, "learning_rate": 6.460557569173252e-05, "loss": 2.7814, "step": 22476 }, { "epoch": 1.3953069712583028, "grad_norm": 0.5500364204832551, "learning_rate": 6.460212164873464e-05, "loss": 2.8613, "step": 22477 }, { "epoch": 1.3953690483580607, "grad_norm": 0.5655825709662923, "learning_rate": 6.459866752955395e-05, "loss": 2.7421, "step": 22478 }, { "epoch": 1.3954311254578187, "grad_norm": 0.5806089085709193, "learning_rate": 6.459521333420847e-05, "loss": 2.905, "step": 22479 }, { "epoch": 1.3954932025575766, "grad_norm": 0.6146557948396726, "learning_rate": 6.459175906271621e-05, "loss": 2.7849, "step": 22480 }, { "epoch": 1.3955552796573345, "grad_norm": 0.58186752781644, "learning_rate": 6.458830471509521e-05, "loss": 2.7937, "step": 22481 }, { "epoch": 1.3956173567570924, "grad_norm": 0.5884122846957338, "learning_rate": 6.458485029136347e-05, "loss": 2.8692, "step": 22482 }, { "epoch": 1.3956794338568503, "grad_norm": 0.612994375984069, "learning_rate": 6.458139579153904e-05, "loss": 2.8334, "step": 22483 }, { "epoch": 1.395741510956608, "grad_norm": 0.5650590066986405, "learning_rate": 6.457794121563992e-05, "loss": 2.7502, "step": 22484 }, { "epoch": 1.395803588056366, "grad_norm": 0.6879238426077596, "learning_rate": 6.457448656368415e-05, "loss": 2.8531, "step": 22485 }, { "epoch": 1.3958656651561239, "grad_norm": 0.5316584083111064, "learning_rate": 6.457103183568974e-05, "loss": 2.8231, "step": 22486 }, { "epoch": 1.3959277422558818, "grad_norm": 0.5648360224151685, "learning_rate": 6.456757703167473e-05, "loss": 2.8788, "step": 22487 }, { "epoch": 1.3959898193556397, "grad_norm": 0.5189178689254883, "learning_rate": 6.456412215165713e-05, "loss": 2.8326, "step": 22488 }, { "epoch": 1.3960518964553976, "grad_norm": 0.5575100182800624, "learning_rate": 6.456066719565497e-05, "loss": 2.7354, "step": 22489 }, { "epoch": 1.3961139735551555, "grad_norm": 0.5491784766070098, "learning_rate": 6.455721216368626e-05, "loss": 2.8283, "step": 22490 }, { "epoch": 1.3961760506549135, "grad_norm": 0.5417365493688968, "learning_rate": 6.455375705576904e-05, "loss": 2.7926, "step": 22491 }, { "epoch": 1.3962381277546714, "grad_norm": 0.5841208206050098, "learning_rate": 6.455030187192136e-05, "loss": 2.8387, "step": 22492 }, { "epoch": 1.396300204854429, "grad_norm": 0.5401853394337711, "learning_rate": 6.454684661216121e-05, "loss": 2.7564, "step": 22493 }, { "epoch": 1.396362281954187, "grad_norm": 0.6076182694948152, "learning_rate": 6.454339127650663e-05, "loss": 2.8676, "step": 22494 }, { "epoch": 1.396424359053945, "grad_norm": 0.5181565992707203, "learning_rate": 6.453993586497564e-05, "loss": 2.8128, "step": 22495 }, { "epoch": 1.3964864361537028, "grad_norm": 0.6094247602051126, "learning_rate": 6.453648037758627e-05, "loss": 2.8518, "step": 22496 }, { "epoch": 1.3965485132534607, "grad_norm": 0.6008927716745224, "learning_rate": 6.453302481435657e-05, "loss": 2.7775, "step": 22497 }, { "epoch": 1.3966105903532187, "grad_norm": 0.541124986963587, "learning_rate": 6.452956917530454e-05, "loss": 2.7382, "step": 22498 }, { "epoch": 1.3966726674529766, "grad_norm": 0.5642986944328509, "learning_rate": 6.452611346044823e-05, "loss": 2.7916, "step": 22499 }, { "epoch": 1.3967347445527345, "grad_norm": 0.5428862474731379, "learning_rate": 6.452265766980565e-05, "loss": 2.718, "step": 22500 }, { "epoch": 1.3967968216524924, "grad_norm": 0.6020190005488112, "learning_rate": 6.451920180339486e-05, "loss": 2.8362, "step": 22501 }, { "epoch": 1.3968588987522503, "grad_norm": 0.5614585772663028, "learning_rate": 6.451574586123383e-05, "loss": 2.7213, "step": 22502 }, { "epoch": 1.3969209758520083, "grad_norm": 0.5453659763730668, "learning_rate": 6.451228984334066e-05, "loss": 2.8153, "step": 22503 }, { "epoch": 1.3969830529517662, "grad_norm": 0.6136115818526511, "learning_rate": 6.450883374973336e-05, "loss": 2.6839, "step": 22504 }, { "epoch": 1.397045130051524, "grad_norm": 0.5184928727887845, "learning_rate": 6.450537758042994e-05, "loss": 2.8289, "step": 22505 }, { "epoch": 1.397107207151282, "grad_norm": 0.5903568183737151, "learning_rate": 6.450192133544843e-05, "loss": 2.8113, "step": 22506 }, { "epoch": 1.39716928425104, "grad_norm": 0.5634468110426487, "learning_rate": 6.44984650148069e-05, "loss": 2.8074, "step": 22507 }, { "epoch": 1.3972313613507976, "grad_norm": 0.5435888742270448, "learning_rate": 6.449500861852335e-05, "loss": 2.8994, "step": 22508 }, { "epoch": 1.3972934384505555, "grad_norm": 0.5779028543042379, "learning_rate": 6.44915521466158e-05, "loss": 2.8188, "step": 22509 }, { "epoch": 1.3973555155503135, "grad_norm": 0.5865618113623973, "learning_rate": 6.448809559910232e-05, "loss": 2.8134, "step": 22510 }, { "epoch": 1.3974175926500714, "grad_norm": 0.5809288080887742, "learning_rate": 6.448463897600092e-05, "loss": 2.8364, "step": 22511 }, { "epoch": 1.3974796697498293, "grad_norm": 0.5729176318998587, "learning_rate": 6.448118227732967e-05, "loss": 2.7841, "step": 22512 }, { "epoch": 1.3975417468495872, "grad_norm": 0.5854790228987725, "learning_rate": 6.447772550310655e-05, "loss": 2.7516, "step": 22513 }, { "epoch": 1.3976038239493451, "grad_norm": 0.6299217980910051, "learning_rate": 6.447426865334963e-05, "loss": 2.8602, "step": 22514 }, { "epoch": 1.397665901049103, "grad_norm": 0.5550592546792305, "learning_rate": 6.447081172807694e-05, "loss": 2.8022, "step": 22515 }, { "epoch": 1.397727978148861, "grad_norm": 0.6525871827720745, "learning_rate": 6.44673547273065e-05, "loss": 2.8624, "step": 22516 }, { "epoch": 1.3977900552486187, "grad_norm": 0.5816273151665624, "learning_rate": 6.446389765105639e-05, "loss": 2.7347, "step": 22517 }, { "epoch": 1.3978521323483766, "grad_norm": 0.6385152315535058, "learning_rate": 6.446044049934456e-05, "loss": 2.7384, "step": 22518 }, { "epoch": 1.3979142094481345, "grad_norm": 0.5348341592100968, "learning_rate": 6.445698327218916e-05, "loss": 2.8519, "step": 22519 }, { "epoch": 1.3979762865478924, "grad_norm": 0.642219234527068, "learning_rate": 6.445352596960813e-05, "loss": 2.8153, "step": 22520 }, { "epoch": 1.3980383636476503, "grad_norm": 0.5280156693915015, "learning_rate": 6.445006859161956e-05, "loss": 2.7958, "step": 22521 }, { "epoch": 1.3981004407474082, "grad_norm": 0.5524969770599106, "learning_rate": 6.444661113824147e-05, "loss": 2.8219, "step": 22522 }, { "epoch": 1.3981625178471662, "grad_norm": 0.5527964850262905, "learning_rate": 6.444315360949191e-05, "loss": 2.7306, "step": 22523 }, { "epoch": 1.398224594946924, "grad_norm": 0.521680130810194, "learning_rate": 6.44396960053889e-05, "loss": 2.7963, "step": 22524 }, { "epoch": 1.398286672046682, "grad_norm": 0.580158343692765, "learning_rate": 6.44362383259505e-05, "loss": 2.7956, "step": 22525 }, { "epoch": 1.39834874914644, "grad_norm": 0.552298491915568, "learning_rate": 6.443278057119475e-05, "loss": 2.8441, "step": 22526 }, { "epoch": 1.3984108262461978, "grad_norm": 0.5525938199075643, "learning_rate": 6.442932274113967e-05, "loss": 2.7919, "step": 22527 }, { "epoch": 1.3984729033459558, "grad_norm": 0.5578643197745276, "learning_rate": 6.442586483580328e-05, "loss": 2.7787, "step": 22528 }, { "epoch": 1.3985349804457137, "grad_norm": 0.5467120336494539, "learning_rate": 6.442240685520369e-05, "loss": 2.859, "step": 22529 }, { "epoch": 1.3985970575454716, "grad_norm": 0.5414755893798647, "learning_rate": 6.44189487993589e-05, "loss": 2.8487, "step": 22530 }, { "epoch": 1.3986591346452295, "grad_norm": 0.5791044958854089, "learning_rate": 6.441549066828694e-05, "loss": 2.7947, "step": 22531 }, { "epoch": 1.3987212117449872, "grad_norm": 0.5676931035043388, "learning_rate": 6.441203246200587e-05, "loss": 2.7577, "step": 22532 }, { "epoch": 1.3987832888447451, "grad_norm": 0.5485512520396549, "learning_rate": 6.440857418053374e-05, "loss": 2.7805, "step": 22533 }, { "epoch": 1.398845365944503, "grad_norm": 0.5412883690764015, "learning_rate": 6.440511582388856e-05, "loss": 2.8425, "step": 22534 }, { "epoch": 1.398907443044261, "grad_norm": 0.5944677080844037, "learning_rate": 6.440165739208839e-05, "loss": 2.8103, "step": 22535 }, { "epoch": 1.3989695201440189, "grad_norm": 0.5318815740589203, "learning_rate": 6.43981988851513e-05, "loss": 2.8101, "step": 22536 }, { "epoch": 1.3990315972437768, "grad_norm": 0.64316754425617, "learning_rate": 6.439474030309531e-05, "loss": 2.8577, "step": 22537 }, { "epoch": 1.3990936743435347, "grad_norm": 0.49309274894351157, "learning_rate": 6.439128164593845e-05, "loss": 2.8329, "step": 22538 }, { "epoch": 1.3991557514432926, "grad_norm": 0.565899484116932, "learning_rate": 6.438782291369879e-05, "loss": 2.8003, "step": 22539 }, { "epoch": 1.3992178285430505, "grad_norm": 0.6100890060277098, "learning_rate": 6.438436410639435e-05, "loss": 2.6928, "step": 22540 }, { "epoch": 1.3992799056428082, "grad_norm": 0.6263959316013752, "learning_rate": 6.43809052240432e-05, "loss": 2.8105, "step": 22541 }, { "epoch": 1.3993419827425662, "grad_norm": 0.5774090173812626, "learning_rate": 6.437744626666339e-05, "loss": 2.8772, "step": 22542 }, { "epoch": 1.399404059842324, "grad_norm": 0.6530870782529737, "learning_rate": 6.437398723427294e-05, "loss": 2.7884, "step": 22543 }, { "epoch": 1.399466136942082, "grad_norm": 0.5396430441565819, "learning_rate": 6.437052812688991e-05, "loss": 2.7759, "step": 22544 }, { "epoch": 1.39952821404184, "grad_norm": 0.6448461514708547, "learning_rate": 6.436706894453234e-05, "loss": 2.7903, "step": 22545 }, { "epoch": 1.3995902911415978, "grad_norm": 0.5832893803477183, "learning_rate": 6.43636096872183e-05, "loss": 2.7983, "step": 22546 }, { "epoch": 1.3996523682413557, "grad_norm": 0.6150658028375051, "learning_rate": 6.436015035496581e-05, "loss": 2.7571, "step": 22547 }, { "epoch": 1.3997144453411137, "grad_norm": 0.6596391898879486, "learning_rate": 6.435669094779294e-05, "loss": 2.8664, "step": 22548 }, { "epoch": 1.3997765224408716, "grad_norm": 0.5981832188142278, "learning_rate": 6.43532314657177e-05, "loss": 2.7983, "step": 22549 }, { "epoch": 1.3998385995406295, "grad_norm": 0.6207206133040682, "learning_rate": 6.434977190875819e-05, "loss": 2.88, "step": 22550 }, { "epoch": 1.3999006766403874, "grad_norm": 0.603259504470222, "learning_rate": 6.434631227693245e-05, "loss": 2.7231, "step": 22551 }, { "epoch": 1.3999627537401453, "grad_norm": 0.555839658321823, "learning_rate": 6.434285257025849e-05, "loss": 2.8434, "step": 22552 }, { "epoch": 1.4000248308399033, "grad_norm": 0.6242802767456003, "learning_rate": 6.43393927887544e-05, "loss": 2.7907, "step": 22553 }, { "epoch": 1.4000869079396612, "grad_norm": 0.4762410502456415, "learning_rate": 6.43359329324382e-05, "loss": 2.7669, "step": 22554 }, { "epoch": 1.400148985039419, "grad_norm": 0.5621974979378884, "learning_rate": 6.433247300132799e-05, "loss": 2.843, "step": 22555 }, { "epoch": 1.4002110621391768, "grad_norm": 0.5656388971297023, "learning_rate": 6.432901299544176e-05, "loss": 2.8117, "step": 22556 }, { "epoch": 1.4002731392389347, "grad_norm": 0.6181585182073889, "learning_rate": 6.432555291479761e-05, "loss": 2.9006, "step": 22557 }, { "epoch": 1.4003352163386926, "grad_norm": 0.5957949208057488, "learning_rate": 6.432209275941354e-05, "loss": 2.8216, "step": 22558 }, { "epoch": 1.4003972934384505, "grad_norm": 0.652319376598668, "learning_rate": 6.431863252930768e-05, "loss": 2.8834, "step": 22559 }, { "epoch": 1.4004593705382085, "grad_norm": 0.5770066501839131, "learning_rate": 6.4315172224498e-05, "loss": 2.8254, "step": 22560 }, { "epoch": 1.4005214476379664, "grad_norm": 0.6475005677121299, "learning_rate": 6.431171184500261e-05, "loss": 2.8088, "step": 22561 }, { "epoch": 1.4005835247377243, "grad_norm": 0.594503175972427, "learning_rate": 6.430825139083955e-05, "loss": 2.8891, "step": 22562 }, { "epoch": 1.4006456018374822, "grad_norm": 0.6422138747022378, "learning_rate": 6.430479086202685e-05, "loss": 2.697, "step": 22563 }, { "epoch": 1.4007076789372401, "grad_norm": 0.617182791969595, "learning_rate": 6.43013302585826e-05, "loss": 2.7408, "step": 22564 }, { "epoch": 1.4007697560369978, "grad_norm": 0.6127398585675194, "learning_rate": 6.429786958052483e-05, "loss": 2.8389, "step": 22565 }, { "epoch": 1.4008318331367557, "grad_norm": 0.6291502722021013, "learning_rate": 6.429440882787161e-05, "loss": 2.8685, "step": 22566 }, { "epoch": 1.4008939102365137, "grad_norm": 0.6227685192942578, "learning_rate": 6.429094800064098e-05, "loss": 2.7476, "step": 22567 }, { "epoch": 1.4009559873362716, "grad_norm": 0.6726400322459918, "learning_rate": 6.428748709885102e-05, "loss": 2.8323, "step": 22568 }, { "epoch": 1.4010180644360295, "grad_norm": 0.5950432296083131, "learning_rate": 6.428402612251975e-05, "loss": 2.8079, "step": 22569 }, { "epoch": 1.4010801415357874, "grad_norm": 0.6630989938610683, "learning_rate": 6.428056507166527e-05, "loss": 2.8647, "step": 22570 }, { "epoch": 1.4011422186355453, "grad_norm": 0.6033170799398468, "learning_rate": 6.427710394630562e-05, "loss": 2.7832, "step": 22571 }, { "epoch": 1.4012042957353033, "grad_norm": 0.6345190889930498, "learning_rate": 6.427364274645881e-05, "loss": 2.7341, "step": 22572 }, { "epoch": 1.4012663728350612, "grad_norm": 0.625872194156258, "learning_rate": 6.427018147214298e-05, "loss": 2.8167, "step": 22573 }, { "epoch": 1.401328449934819, "grad_norm": 0.641978137376381, "learning_rate": 6.426672012337614e-05, "loss": 2.881, "step": 22574 }, { "epoch": 1.401390527034577, "grad_norm": 0.5953284414164439, "learning_rate": 6.426325870017637e-05, "loss": 2.746, "step": 22575 }, { "epoch": 1.401452604134335, "grad_norm": 0.6137992712082425, "learning_rate": 6.425979720256169e-05, "loss": 2.8103, "step": 22576 }, { "epoch": 1.4015146812340928, "grad_norm": 0.567402955845034, "learning_rate": 6.42563356305502e-05, "loss": 2.823, "step": 22577 }, { "epoch": 1.4015767583338508, "grad_norm": 0.641181703912589, "learning_rate": 6.425287398415994e-05, "loss": 2.9402, "step": 22578 }, { "epoch": 1.4016388354336087, "grad_norm": 0.5352391088340689, "learning_rate": 6.424941226340899e-05, "loss": 2.7252, "step": 22579 }, { "epoch": 1.4017009125333664, "grad_norm": 0.5891708173250414, "learning_rate": 6.424595046831539e-05, "loss": 2.8667, "step": 22580 }, { "epoch": 1.4017629896331243, "grad_norm": 0.5852410309799705, "learning_rate": 6.42424885988972e-05, "loss": 2.7315, "step": 22581 }, { "epoch": 1.4018250667328822, "grad_norm": 0.581908168195674, "learning_rate": 6.42390266551725e-05, "loss": 2.8096, "step": 22582 }, { "epoch": 1.4018871438326401, "grad_norm": 0.665513394630103, "learning_rate": 6.423556463715933e-05, "loss": 2.8179, "step": 22583 }, { "epoch": 1.401949220932398, "grad_norm": 0.5943759078651659, "learning_rate": 6.423210254487577e-05, "loss": 2.7873, "step": 22584 }, { "epoch": 1.402011298032156, "grad_norm": 0.6705644156401698, "learning_rate": 6.422864037833988e-05, "loss": 2.8342, "step": 22585 }, { "epoch": 1.4020733751319139, "grad_norm": 0.6047669065319222, "learning_rate": 6.422517813756972e-05, "loss": 2.7935, "step": 22586 }, { "epoch": 1.4021354522316718, "grad_norm": 0.6137071819092172, "learning_rate": 6.422171582258334e-05, "loss": 2.7903, "step": 22587 }, { "epoch": 1.4021975293314297, "grad_norm": 0.6125991332347839, "learning_rate": 6.421825343339883e-05, "loss": 2.8248, "step": 22588 }, { "epoch": 1.4022596064311874, "grad_norm": 0.6578590331302631, "learning_rate": 6.421479097003424e-05, "loss": 2.8365, "step": 22589 }, { "epoch": 1.4023216835309453, "grad_norm": 0.5682542841389414, "learning_rate": 6.421132843250763e-05, "loss": 2.7813, "step": 22590 }, { "epoch": 1.4023837606307032, "grad_norm": 0.5840047622235014, "learning_rate": 6.420786582083708e-05, "loss": 2.8912, "step": 22591 }, { "epoch": 1.4024458377304612, "grad_norm": 0.5763447389073465, "learning_rate": 6.420440313504062e-05, "loss": 2.8375, "step": 22592 }, { "epoch": 1.402507914830219, "grad_norm": 0.5639008985325202, "learning_rate": 6.420094037513635e-05, "loss": 2.8466, "step": 22593 }, { "epoch": 1.402569991929977, "grad_norm": 0.6138808119958384, "learning_rate": 6.419747754114234e-05, "loss": 2.7548, "step": 22594 }, { "epoch": 1.402632069029735, "grad_norm": 0.6324955888451109, "learning_rate": 6.419401463307663e-05, "loss": 2.7958, "step": 22595 }, { "epoch": 1.4026941461294928, "grad_norm": 0.6009979819471781, "learning_rate": 6.41905516509573e-05, "loss": 2.7716, "step": 22596 }, { "epoch": 1.4027562232292508, "grad_norm": 0.5848010167768618, "learning_rate": 6.418708859480242e-05, "loss": 2.8304, "step": 22597 }, { "epoch": 1.4028183003290087, "grad_norm": 0.5926659123593276, "learning_rate": 6.418362546463006e-05, "loss": 2.7015, "step": 22598 }, { "epoch": 1.4028803774287666, "grad_norm": 0.6332644273475112, "learning_rate": 6.418016226045828e-05, "loss": 2.7196, "step": 22599 }, { "epoch": 1.4029424545285245, "grad_norm": 0.6125720838052559, "learning_rate": 6.417669898230516e-05, "loss": 2.8273, "step": 22600 }, { "epoch": 1.4030045316282824, "grad_norm": 0.6141731191690416, "learning_rate": 6.417323563018874e-05, "loss": 2.6865, "step": 22601 }, { "epoch": 1.4030666087280403, "grad_norm": 0.6691665617664424, "learning_rate": 6.416977220412713e-05, "loss": 2.8319, "step": 22602 }, { "epoch": 1.4031286858277983, "grad_norm": 0.593017276592657, "learning_rate": 6.416630870413835e-05, "loss": 2.8042, "step": 22603 }, { "epoch": 1.403190762927556, "grad_norm": 0.6500952339098582, "learning_rate": 6.416284513024053e-05, "loss": 2.8032, "step": 22604 }, { "epoch": 1.4032528400273139, "grad_norm": 0.5627468150441466, "learning_rate": 6.41593814824517e-05, "loss": 2.8044, "step": 22605 }, { "epoch": 1.4033149171270718, "grad_norm": 0.6939421877265921, "learning_rate": 6.415591776078993e-05, "loss": 2.7733, "step": 22606 }, { "epoch": 1.4033769942268297, "grad_norm": 0.589999197377451, "learning_rate": 6.415245396527331e-05, "loss": 2.7053, "step": 22607 }, { "epoch": 1.4034390713265876, "grad_norm": 0.5891516205738256, "learning_rate": 6.414899009591989e-05, "loss": 2.7666, "step": 22608 }, { "epoch": 1.4035011484263455, "grad_norm": 0.6220200546778843, "learning_rate": 6.414552615274776e-05, "loss": 2.7569, "step": 22609 }, { "epoch": 1.4035632255261035, "grad_norm": 0.559261616345335, "learning_rate": 6.414206213577498e-05, "loss": 2.8291, "step": 22610 }, { "epoch": 1.4036253026258614, "grad_norm": 0.7064316878772823, "learning_rate": 6.413859804501964e-05, "loss": 2.8674, "step": 22611 }, { "epoch": 1.4036873797256193, "grad_norm": 0.627573318567513, "learning_rate": 6.413513388049981e-05, "loss": 2.7529, "step": 22612 }, { "epoch": 1.403749456825377, "grad_norm": 0.5954714465274179, "learning_rate": 6.413166964223354e-05, "loss": 2.8186, "step": 22613 }, { "epoch": 1.403811533925135, "grad_norm": 0.6582170026544356, "learning_rate": 6.412820533023892e-05, "loss": 2.7624, "step": 22614 }, { "epoch": 1.4038736110248928, "grad_norm": 0.5609186780316661, "learning_rate": 6.412474094453402e-05, "loss": 2.7585, "step": 22615 }, { "epoch": 1.4039356881246507, "grad_norm": 0.6046043624699722, "learning_rate": 6.412127648513693e-05, "loss": 2.82, "step": 22616 }, { "epoch": 1.4039977652244087, "grad_norm": 0.5878480154418801, "learning_rate": 6.41178119520657e-05, "loss": 2.8342, "step": 22617 }, { "epoch": 1.4040598423241666, "grad_norm": 0.6113094046654459, "learning_rate": 6.411434734533842e-05, "loss": 2.85, "step": 22618 }, { "epoch": 1.4041219194239245, "grad_norm": 0.529034257327612, "learning_rate": 6.411088266497316e-05, "loss": 2.7781, "step": 22619 }, { "epoch": 1.4041839965236824, "grad_norm": 0.5957156320728536, "learning_rate": 6.410741791098802e-05, "loss": 2.8339, "step": 22620 }, { "epoch": 1.4042460736234403, "grad_norm": 0.5406877371047032, "learning_rate": 6.410395308340104e-05, "loss": 2.9152, "step": 22621 }, { "epoch": 1.4043081507231983, "grad_norm": 0.5282871144241996, "learning_rate": 6.410048818223032e-05, "loss": 2.8361, "step": 22622 }, { "epoch": 1.4043702278229562, "grad_norm": 0.5931066239688969, "learning_rate": 6.409702320749392e-05, "loss": 2.821, "step": 22623 }, { "epoch": 1.404432304922714, "grad_norm": 0.5592880735572033, "learning_rate": 6.409355815920993e-05, "loss": 2.8517, "step": 22624 }, { "epoch": 1.404494382022472, "grad_norm": 0.6098956549231846, "learning_rate": 6.409009303739644e-05, "loss": 2.8916, "step": 22625 }, { "epoch": 1.40455645912223, "grad_norm": 0.6083682130803802, "learning_rate": 6.408662784207149e-05, "loss": 2.8603, "step": 22626 }, { "epoch": 1.4046185362219878, "grad_norm": 0.5662602203490995, "learning_rate": 6.40831625732532e-05, "loss": 2.8505, "step": 22627 }, { "epoch": 1.4046806133217455, "grad_norm": 0.5671774776613626, "learning_rate": 6.407969723095961e-05, "loss": 2.8719, "step": 22628 }, { "epoch": 1.4047426904215035, "grad_norm": 0.5589759959319969, "learning_rate": 6.407623181520885e-05, "loss": 2.8271, "step": 22629 }, { "epoch": 1.4048047675212614, "grad_norm": 0.5737698155972586, "learning_rate": 6.407276632601896e-05, "loss": 2.8013, "step": 22630 }, { "epoch": 1.4048668446210193, "grad_norm": 0.5335522301802536, "learning_rate": 6.406930076340803e-05, "loss": 2.7578, "step": 22631 }, { "epoch": 1.4049289217207772, "grad_norm": 0.517470064218786, "learning_rate": 6.406583512739415e-05, "loss": 2.7948, "step": 22632 }, { "epoch": 1.4049909988205351, "grad_norm": 0.559782563809755, "learning_rate": 6.406236941799539e-05, "loss": 2.7968, "step": 22633 }, { "epoch": 1.405053075920293, "grad_norm": 0.47691351048870884, "learning_rate": 6.405890363522984e-05, "loss": 2.7881, "step": 22634 }, { "epoch": 1.405115153020051, "grad_norm": 0.5517091172597387, "learning_rate": 6.405543777911555e-05, "loss": 2.79, "step": 22635 }, { "epoch": 1.4051772301198089, "grad_norm": 0.5224204644899649, "learning_rate": 6.405197184967067e-05, "loss": 2.7859, "step": 22636 }, { "epoch": 1.4052393072195666, "grad_norm": 0.5199363104624025, "learning_rate": 6.40485058469132e-05, "loss": 2.879, "step": 22637 }, { "epoch": 1.4053013843193245, "grad_norm": 0.5894942353349337, "learning_rate": 6.40450397708613e-05, "loss": 2.7155, "step": 22638 }, { "epoch": 1.4053634614190824, "grad_norm": 0.5250218454993043, "learning_rate": 6.4041573621533e-05, "loss": 2.7845, "step": 22639 }, { "epoch": 1.4054255385188403, "grad_norm": 0.5566517075890984, "learning_rate": 6.403810739894641e-05, "loss": 2.7552, "step": 22640 }, { "epoch": 1.4054876156185983, "grad_norm": 0.5453610308456192, "learning_rate": 6.403464110311961e-05, "loss": 2.7392, "step": 22641 }, { "epoch": 1.4055496927183562, "grad_norm": 0.5359543466839607, "learning_rate": 6.403117473407065e-05, "loss": 2.7649, "step": 22642 }, { "epoch": 1.405611769818114, "grad_norm": 0.5557988276582855, "learning_rate": 6.402770829181768e-05, "loss": 2.8174, "step": 22643 }, { "epoch": 1.405673846917872, "grad_norm": 0.564559530504809, "learning_rate": 6.402424177637873e-05, "loss": 2.7929, "step": 22644 }, { "epoch": 1.40573592401763, "grad_norm": 0.5587780521698767, "learning_rate": 6.402077518777192e-05, "loss": 2.8698, "step": 22645 }, { "epoch": 1.4057980011173878, "grad_norm": 0.5140904218980934, "learning_rate": 6.40173085260153e-05, "loss": 2.7235, "step": 22646 }, { "epoch": 1.4058600782171458, "grad_norm": 0.5110619093126576, "learning_rate": 6.401384179112701e-05, "loss": 2.7489, "step": 22647 }, { "epoch": 1.4059221553169037, "grad_norm": 0.5665293329612462, "learning_rate": 6.401037498312508e-05, "loss": 2.8172, "step": 22648 }, { "epoch": 1.4059842324166616, "grad_norm": 0.550071475413152, "learning_rate": 6.400690810202764e-05, "loss": 2.817, "step": 22649 }, { "epoch": 1.4060463095164195, "grad_norm": 0.5589253968288397, "learning_rate": 6.400344114785275e-05, "loss": 2.8324, "step": 22650 }, { "epoch": 1.4061083866161774, "grad_norm": 0.540283506507978, "learning_rate": 6.399997412061851e-05, "loss": 2.7764, "step": 22651 }, { "epoch": 1.4061704637159351, "grad_norm": 0.5559055226110166, "learning_rate": 6.399650702034302e-05, "loss": 2.7408, "step": 22652 }, { "epoch": 1.406232540815693, "grad_norm": 0.6176069258959969, "learning_rate": 6.399303984704432e-05, "loss": 2.7921, "step": 22653 }, { "epoch": 1.406294617915451, "grad_norm": 0.5498748517705045, "learning_rate": 6.398957260074057e-05, "loss": 2.8015, "step": 22654 }, { "epoch": 1.4063566950152089, "grad_norm": 0.5249901986342537, "learning_rate": 6.39861052814498e-05, "loss": 2.7857, "step": 22655 }, { "epoch": 1.4064187721149668, "grad_norm": 0.6127262641499294, "learning_rate": 6.398263788919014e-05, "loss": 2.7642, "step": 22656 }, { "epoch": 1.4064808492147247, "grad_norm": 0.5994028984713171, "learning_rate": 6.397917042397965e-05, "loss": 2.8541, "step": 22657 }, { "epoch": 1.4065429263144826, "grad_norm": 0.6074283758962438, "learning_rate": 6.397570288583646e-05, "loss": 2.8137, "step": 22658 }, { "epoch": 1.4066050034142406, "grad_norm": 0.6499489553289223, "learning_rate": 6.39722352747786e-05, "loss": 2.8225, "step": 22659 }, { "epoch": 1.4066670805139985, "grad_norm": 0.6292618403754884, "learning_rate": 6.39687675908242e-05, "loss": 2.9038, "step": 22660 }, { "epoch": 1.4067291576137562, "grad_norm": 0.7396147860075951, "learning_rate": 6.396529983399138e-05, "loss": 2.8532, "step": 22661 }, { "epoch": 1.406791234713514, "grad_norm": 0.7241869412281176, "learning_rate": 6.396183200429817e-05, "loss": 2.8588, "step": 22662 }, { "epoch": 1.406853311813272, "grad_norm": 0.5935694520399996, "learning_rate": 6.395836410176271e-05, "loss": 2.7361, "step": 22663 }, { "epoch": 1.40691538891303, "grad_norm": 0.7214949891570017, "learning_rate": 6.395489612640307e-05, "loss": 2.8832, "step": 22664 }, { "epoch": 1.4069774660127878, "grad_norm": 0.6663652101068437, "learning_rate": 6.395142807823735e-05, "loss": 2.797, "step": 22665 }, { "epoch": 1.4070395431125458, "grad_norm": 0.6021808319831201, "learning_rate": 6.394795995728366e-05, "loss": 2.7416, "step": 22666 }, { "epoch": 1.4071016202123037, "grad_norm": 0.6942824116395064, "learning_rate": 6.394449176356006e-05, "loss": 2.7415, "step": 22667 }, { "epoch": 1.4071636973120616, "grad_norm": 0.5283986313848364, "learning_rate": 6.394102349708467e-05, "loss": 2.8049, "step": 22668 }, { "epoch": 1.4072257744118195, "grad_norm": 0.7385366492556081, "learning_rate": 6.393755515787556e-05, "loss": 2.8307, "step": 22669 }, { "epoch": 1.4072878515115774, "grad_norm": 0.5414721494259639, "learning_rate": 6.393408674595084e-05, "loss": 2.8595, "step": 22670 }, { "epoch": 1.4073499286113353, "grad_norm": 0.6351917890339902, "learning_rate": 6.393061826132861e-05, "loss": 2.7921, "step": 22671 }, { "epoch": 1.4074120057110933, "grad_norm": 0.6743877574785184, "learning_rate": 6.392714970402698e-05, "loss": 2.7972, "step": 22672 }, { "epoch": 1.4074740828108512, "grad_norm": 0.5427321323245755, "learning_rate": 6.392368107406401e-05, "loss": 2.8147, "step": 22673 }, { "epoch": 1.407536159910609, "grad_norm": 0.6440938216602622, "learning_rate": 6.392021237145782e-05, "loss": 2.7701, "step": 22674 }, { "epoch": 1.4075982370103668, "grad_norm": 0.5404836469870723, "learning_rate": 6.391674359622651e-05, "loss": 2.8858, "step": 22675 }, { "epoch": 1.4076603141101247, "grad_norm": 0.5857110666807951, "learning_rate": 6.391327474838817e-05, "loss": 2.8546, "step": 22676 }, { "epoch": 1.4077223912098826, "grad_norm": 0.5578181076836477, "learning_rate": 6.390980582796089e-05, "loss": 2.8131, "step": 22677 }, { "epoch": 1.4077844683096405, "grad_norm": 0.5934879829102837, "learning_rate": 6.390633683496278e-05, "loss": 2.7946, "step": 22678 }, { "epoch": 1.4078465454093985, "grad_norm": 0.5494284011266198, "learning_rate": 6.390286776941194e-05, "loss": 2.7628, "step": 22679 }, { "epoch": 1.4079086225091564, "grad_norm": 0.5759735800690788, "learning_rate": 6.389939863132645e-05, "loss": 2.8083, "step": 22680 }, { "epoch": 1.4079706996089143, "grad_norm": 0.5832195312890484, "learning_rate": 6.389592942072445e-05, "loss": 2.7659, "step": 22681 }, { "epoch": 1.4080327767086722, "grad_norm": 0.5270531878187599, "learning_rate": 6.389246013762397e-05, "loss": 2.7154, "step": 22682 }, { "epoch": 1.4080948538084301, "grad_norm": 0.5742833107706188, "learning_rate": 6.38889907820432e-05, "loss": 2.8328, "step": 22683 }, { "epoch": 1.4081569309081878, "grad_norm": 0.6097637306003114, "learning_rate": 6.388552135400016e-05, "loss": 2.8062, "step": 22684 }, { "epoch": 1.4082190080079457, "grad_norm": 0.5295088227521284, "learning_rate": 6.3882051853513e-05, "loss": 2.7951, "step": 22685 }, { "epoch": 1.4082810851077037, "grad_norm": 0.5266091776005, "learning_rate": 6.387858228059981e-05, "loss": 2.8707, "step": 22686 }, { "epoch": 1.4083431622074616, "grad_norm": 0.594694615901323, "learning_rate": 6.387511263527868e-05, "loss": 2.7542, "step": 22687 }, { "epoch": 1.4084052393072195, "grad_norm": 0.5667527290046321, "learning_rate": 6.387164291756771e-05, "loss": 2.8777, "step": 22688 }, { "epoch": 1.4084673164069774, "grad_norm": 0.5787101310443699, "learning_rate": 6.386817312748502e-05, "loss": 2.7394, "step": 22689 }, { "epoch": 1.4085293935067353, "grad_norm": 0.5961140450438679, "learning_rate": 6.38647032650487e-05, "loss": 2.8701, "step": 22690 }, { "epoch": 1.4085914706064933, "grad_norm": 0.6100034651892374, "learning_rate": 6.386123333027686e-05, "loss": 2.8847, "step": 22691 }, { "epoch": 1.4086535477062512, "grad_norm": 0.5817812184152811, "learning_rate": 6.38577633231876e-05, "loss": 2.7973, "step": 22692 }, { "epoch": 1.408715624806009, "grad_norm": 0.5984231852520052, "learning_rate": 6.385429324379904e-05, "loss": 2.833, "step": 22693 }, { "epoch": 1.408777701905767, "grad_norm": 0.5556170237497551, "learning_rate": 6.385082309212924e-05, "loss": 2.8199, "step": 22694 }, { "epoch": 1.408839779005525, "grad_norm": 0.5819223545280312, "learning_rate": 6.384735286819635e-05, "loss": 2.7451, "step": 22695 }, { "epoch": 1.4089018561052828, "grad_norm": 0.5776271738664855, "learning_rate": 6.384388257201846e-05, "loss": 2.8775, "step": 22696 }, { "epoch": 1.4089639332050408, "grad_norm": 0.5958261581691346, "learning_rate": 6.384041220361368e-05, "loss": 2.8723, "step": 22697 }, { "epoch": 1.4090260103047987, "grad_norm": 0.5128484297491155, "learning_rate": 6.383694176300008e-05, "loss": 2.7778, "step": 22698 }, { "epoch": 1.4090880874045564, "grad_norm": 0.578871322587206, "learning_rate": 6.383347125019583e-05, "loss": 2.8282, "step": 22699 }, { "epoch": 1.4091501645043143, "grad_norm": 0.5525761300278427, "learning_rate": 6.383000066521898e-05, "loss": 2.8354, "step": 22700 }, { "epoch": 1.4092122416040722, "grad_norm": 0.6313050293306208, "learning_rate": 6.382653000808767e-05, "loss": 2.8311, "step": 22701 }, { "epoch": 1.4092743187038301, "grad_norm": 0.5388190231008032, "learning_rate": 6.382305927882e-05, "loss": 2.7451, "step": 22702 }, { "epoch": 1.409336395803588, "grad_norm": 0.6631422735224168, "learning_rate": 6.381958847743407e-05, "loss": 2.8067, "step": 22703 }, { "epoch": 1.409398472903346, "grad_norm": 0.5467708760999955, "learning_rate": 6.381611760394798e-05, "loss": 2.7715, "step": 22704 }, { "epoch": 1.4094605500031039, "grad_norm": 0.5969960674111635, "learning_rate": 6.381264665837987e-05, "loss": 2.751, "step": 22705 }, { "epoch": 1.4095226271028618, "grad_norm": 0.5838632107073058, "learning_rate": 6.380917564074782e-05, "loss": 2.7641, "step": 22706 }, { "epoch": 1.4095847042026197, "grad_norm": 0.560269888785711, "learning_rate": 6.380570455106994e-05, "loss": 2.8449, "step": 22707 }, { "epoch": 1.4096467813023774, "grad_norm": 0.5746023752083103, "learning_rate": 6.380223338936436e-05, "loss": 2.8082, "step": 22708 }, { "epoch": 1.4097088584021353, "grad_norm": 0.542769910231938, "learning_rate": 6.379876215564918e-05, "loss": 2.8097, "step": 22709 }, { "epoch": 1.4097709355018933, "grad_norm": 0.5429992118359039, "learning_rate": 6.379529084994249e-05, "loss": 2.8048, "step": 22710 }, { "epoch": 1.4098330126016512, "grad_norm": 0.5897385964546368, "learning_rate": 6.379181947226245e-05, "loss": 2.8597, "step": 22711 }, { "epoch": 1.409895089701409, "grad_norm": 0.49090794343806227, "learning_rate": 6.378834802262711e-05, "loss": 2.7568, "step": 22712 }, { "epoch": 1.409957166801167, "grad_norm": 0.5961225690162191, "learning_rate": 6.378487650105462e-05, "loss": 2.8111, "step": 22713 }, { "epoch": 1.410019243900925, "grad_norm": 0.5685622649505253, "learning_rate": 6.378140490756308e-05, "loss": 2.8924, "step": 22714 }, { "epoch": 1.4100813210006828, "grad_norm": 0.6010290480621818, "learning_rate": 6.377793324217062e-05, "loss": 2.8351, "step": 22715 }, { "epoch": 1.4101433981004408, "grad_norm": 0.5740406512742265, "learning_rate": 6.377446150489532e-05, "loss": 2.7732, "step": 22716 }, { "epoch": 1.4102054752001987, "grad_norm": 0.6098129215018123, "learning_rate": 6.377098969575533e-05, "loss": 2.7916, "step": 22717 }, { "epoch": 1.4102675522999566, "grad_norm": 0.6273338849157827, "learning_rate": 6.376751781476873e-05, "loss": 2.8316, "step": 22718 }, { "epoch": 1.4103296293997145, "grad_norm": 0.6019031501437878, "learning_rate": 6.376404586195365e-05, "loss": 2.7431, "step": 22719 }, { "epoch": 1.4103917064994724, "grad_norm": 0.5914537297367864, "learning_rate": 6.376057383732819e-05, "loss": 2.88, "step": 22720 }, { "epoch": 1.4104537835992303, "grad_norm": 0.6449406104136205, "learning_rate": 6.37571017409105e-05, "loss": 2.7439, "step": 22721 }, { "epoch": 1.4105158606989883, "grad_norm": 0.6085079823145145, "learning_rate": 6.375362957271865e-05, "loss": 2.7456, "step": 22722 }, { "epoch": 1.410577937798746, "grad_norm": 0.6010866140641602, "learning_rate": 6.375015733277079e-05, "loss": 2.8052, "step": 22723 }, { "epoch": 1.4106400148985039, "grad_norm": 0.6400157776392802, "learning_rate": 6.3746685021085e-05, "loss": 2.883, "step": 22724 }, { "epoch": 1.4107020919982618, "grad_norm": 0.573430071166769, "learning_rate": 6.374321263767943e-05, "loss": 2.8038, "step": 22725 }, { "epoch": 1.4107641690980197, "grad_norm": 0.6364343159998757, "learning_rate": 6.373974018257218e-05, "loss": 2.7777, "step": 22726 }, { "epoch": 1.4108262461977776, "grad_norm": 0.55598092053802, "learning_rate": 6.373626765578138e-05, "loss": 2.8178, "step": 22727 }, { "epoch": 1.4108883232975356, "grad_norm": 0.5523987414922334, "learning_rate": 6.373279505732513e-05, "loss": 2.8367, "step": 22728 }, { "epoch": 1.4109504003972935, "grad_norm": 0.6098108933413845, "learning_rate": 6.372932238722156e-05, "loss": 2.8963, "step": 22729 }, { "epoch": 1.4110124774970514, "grad_norm": 0.49545320323467196, "learning_rate": 6.372584964548878e-05, "loss": 2.7303, "step": 22730 }, { "epoch": 1.4110745545968093, "grad_norm": 0.5581671679843729, "learning_rate": 6.37223768321449e-05, "loss": 2.7975, "step": 22731 }, { "epoch": 1.411136631696567, "grad_norm": 0.5288879391361617, "learning_rate": 6.371890394720806e-05, "loss": 2.8181, "step": 22732 }, { "epoch": 1.411198708796325, "grad_norm": 0.5968401454315561, "learning_rate": 6.371543099069636e-05, "loss": 2.811, "step": 22733 }, { "epoch": 1.4112607858960828, "grad_norm": 0.5600099018617957, "learning_rate": 6.371195796262794e-05, "loss": 2.878, "step": 22734 }, { "epoch": 1.4113228629958408, "grad_norm": 0.5585436231623239, "learning_rate": 6.37084848630209e-05, "loss": 2.893, "step": 22735 }, { "epoch": 1.4113849400955987, "grad_norm": 0.5527304389732594, "learning_rate": 6.370501169189336e-05, "loss": 2.8423, "step": 22736 }, { "epoch": 1.4114470171953566, "grad_norm": 0.5407708230455889, "learning_rate": 6.370153844926345e-05, "loss": 2.7419, "step": 22737 }, { "epoch": 1.4115090942951145, "grad_norm": 0.5287318856493461, "learning_rate": 6.36980651351493e-05, "loss": 2.7891, "step": 22738 }, { "epoch": 1.4115711713948724, "grad_norm": 0.5220060275394833, "learning_rate": 6.369459174956901e-05, "loss": 2.8069, "step": 22739 }, { "epoch": 1.4116332484946303, "grad_norm": 0.5408341416684513, "learning_rate": 6.36911182925407e-05, "loss": 2.8072, "step": 22740 }, { "epoch": 1.4116953255943883, "grad_norm": 0.5323457636905672, "learning_rate": 6.368764476408251e-05, "loss": 2.8424, "step": 22741 }, { "epoch": 1.4117574026941462, "grad_norm": 0.5366350333544064, "learning_rate": 6.368417116421256e-05, "loss": 2.8874, "step": 22742 }, { "epoch": 1.411819479793904, "grad_norm": 0.5279408218389473, "learning_rate": 6.368069749294896e-05, "loss": 2.7263, "step": 22743 }, { "epoch": 1.411881556893662, "grad_norm": 0.5621731391247253, "learning_rate": 6.367722375030985e-05, "loss": 2.9239, "step": 22744 }, { "epoch": 1.41194363399342, "grad_norm": 0.5872992903192157, "learning_rate": 6.367374993631332e-05, "loss": 2.8836, "step": 22745 }, { "epoch": 1.4120057110931779, "grad_norm": 0.5661447644918179, "learning_rate": 6.367027605097755e-05, "loss": 2.7897, "step": 22746 }, { "epoch": 1.4120677881929355, "grad_norm": 0.5891215722291694, "learning_rate": 6.36668020943206e-05, "loss": 2.8094, "step": 22747 }, { "epoch": 1.4121298652926935, "grad_norm": 0.5538833860943219, "learning_rate": 6.366332806636065e-05, "loss": 2.8156, "step": 22748 }, { "epoch": 1.4121919423924514, "grad_norm": 0.5773600337270526, "learning_rate": 6.365985396711581e-05, "loss": 2.8076, "step": 22749 }, { "epoch": 1.4122540194922093, "grad_norm": 0.5747638372901864, "learning_rate": 6.365637979660416e-05, "loss": 2.7253, "step": 22750 }, { "epoch": 1.4123160965919672, "grad_norm": 0.5720701176625783, "learning_rate": 6.365290555484387e-05, "loss": 2.7974, "step": 22751 }, { "epoch": 1.4123781736917251, "grad_norm": 0.5446159296950551, "learning_rate": 6.364943124185307e-05, "loss": 2.7879, "step": 22752 }, { "epoch": 1.412440250791483, "grad_norm": 0.49840528270434553, "learning_rate": 6.364595685764988e-05, "loss": 2.7825, "step": 22753 }, { "epoch": 1.412502327891241, "grad_norm": 0.5520480312600802, "learning_rate": 6.36424824022524e-05, "loss": 2.6984, "step": 22754 }, { "epoch": 1.412564404990999, "grad_norm": 0.5708129463824756, "learning_rate": 6.363900787567879e-05, "loss": 2.8737, "step": 22755 }, { "epoch": 1.4126264820907566, "grad_norm": 0.6230001402161319, "learning_rate": 6.363553327794716e-05, "loss": 2.8545, "step": 22756 }, { "epoch": 1.4126885591905145, "grad_norm": 0.5554212364768096, "learning_rate": 6.363205860907564e-05, "loss": 2.8438, "step": 22757 }, { "epoch": 1.4127506362902724, "grad_norm": 0.5802191617447208, "learning_rate": 6.362858386908238e-05, "loss": 2.8174, "step": 22758 }, { "epoch": 1.4128127133900303, "grad_norm": 0.5946477076677521, "learning_rate": 6.362510905798546e-05, "loss": 2.8285, "step": 22759 }, { "epoch": 1.4128747904897883, "grad_norm": 0.6103590087436407, "learning_rate": 6.362163417580306e-05, "loss": 2.7838, "step": 22760 }, { "epoch": 1.4129368675895462, "grad_norm": 0.5319030338344986, "learning_rate": 6.361815922255327e-05, "loss": 2.7181, "step": 22761 }, { "epoch": 1.412998944689304, "grad_norm": 0.5906252200010182, "learning_rate": 6.361468419825426e-05, "loss": 2.783, "step": 22762 }, { "epoch": 1.413061021789062, "grad_norm": 0.5833445843111128, "learning_rate": 6.361120910292413e-05, "loss": 2.8406, "step": 22763 }, { "epoch": 1.41312309888882, "grad_norm": 0.6008219185545981, "learning_rate": 6.360773393658101e-05, "loss": 2.81, "step": 22764 }, { "epoch": 1.4131851759885778, "grad_norm": 0.5581636440230947, "learning_rate": 6.360425869924304e-05, "loss": 2.8137, "step": 22765 }, { "epoch": 1.4132472530883358, "grad_norm": 0.594866330615263, "learning_rate": 6.360078339092837e-05, "loss": 2.6895, "step": 22766 }, { "epoch": 1.4133093301880937, "grad_norm": 0.5752157003678635, "learning_rate": 6.359730801165509e-05, "loss": 2.8448, "step": 22767 }, { "epoch": 1.4133714072878516, "grad_norm": 0.5367585302100614, "learning_rate": 6.359383256144135e-05, "loss": 2.8102, "step": 22768 }, { "epoch": 1.4134334843876095, "grad_norm": 0.5630587345979327, "learning_rate": 6.359035704030529e-05, "loss": 2.7386, "step": 22769 }, { "epoch": 1.4134955614873674, "grad_norm": 0.5044460214956888, "learning_rate": 6.358688144826504e-05, "loss": 2.7825, "step": 22770 }, { "epoch": 1.4135576385871251, "grad_norm": 0.5506900414232979, "learning_rate": 6.358340578533874e-05, "loss": 2.7381, "step": 22771 }, { "epoch": 1.413619715686883, "grad_norm": 0.49590430059043217, "learning_rate": 6.35799300515445e-05, "loss": 2.8027, "step": 22772 }, { "epoch": 1.413681792786641, "grad_norm": 0.5559510627880517, "learning_rate": 6.357645424690047e-05, "loss": 2.7669, "step": 22773 }, { "epoch": 1.4137438698863989, "grad_norm": 0.537273811831551, "learning_rate": 6.357297837142479e-05, "loss": 2.716, "step": 22774 }, { "epoch": 1.4138059469861568, "grad_norm": 0.5444684620455642, "learning_rate": 6.356950242513558e-05, "loss": 2.7758, "step": 22775 }, { "epoch": 1.4138680240859147, "grad_norm": 0.5150495265553013, "learning_rate": 6.356602640805099e-05, "loss": 2.7424, "step": 22776 }, { "epoch": 1.4139301011856726, "grad_norm": 0.5396438725458247, "learning_rate": 6.356255032018914e-05, "loss": 2.8055, "step": 22777 }, { "epoch": 1.4139921782854306, "grad_norm": 0.5602261360019053, "learning_rate": 6.355907416156818e-05, "loss": 2.7739, "step": 22778 }, { "epoch": 1.4140542553851885, "grad_norm": 0.5431290259612618, "learning_rate": 6.355559793220624e-05, "loss": 2.8139, "step": 22779 }, { "epoch": 1.4141163324849462, "grad_norm": 0.5261785278805958, "learning_rate": 6.355212163212145e-05, "loss": 2.8585, "step": 22780 }, { "epoch": 1.414178409584704, "grad_norm": 0.489600573567603, "learning_rate": 6.354864526133194e-05, "loss": 2.7531, "step": 22781 }, { "epoch": 1.414240486684462, "grad_norm": 0.5493529447483387, "learning_rate": 6.354516881985588e-05, "loss": 2.857, "step": 22782 }, { "epoch": 1.41430256378422, "grad_norm": 0.5888544337471501, "learning_rate": 6.354169230771137e-05, "loss": 2.8277, "step": 22783 }, { "epoch": 1.4143646408839778, "grad_norm": 0.5817729197447363, "learning_rate": 6.353821572491657e-05, "loss": 2.8059, "step": 22784 }, { "epoch": 1.4144267179837358, "grad_norm": 0.5175205990474051, "learning_rate": 6.353473907148961e-05, "loss": 2.8334, "step": 22785 }, { "epoch": 1.4144887950834937, "grad_norm": 0.5996262488957533, "learning_rate": 6.353126234744864e-05, "loss": 2.7624, "step": 22786 }, { "epoch": 1.4145508721832516, "grad_norm": 0.5319101215336176, "learning_rate": 6.352778555281178e-05, "loss": 2.8742, "step": 22787 }, { "epoch": 1.4146129492830095, "grad_norm": 0.6621577827514727, "learning_rate": 6.352430868759718e-05, "loss": 2.8613, "step": 22788 }, { "epoch": 1.4146750263827674, "grad_norm": 0.5970855930558023, "learning_rate": 6.352083175182298e-05, "loss": 2.7885, "step": 22789 }, { "epoch": 1.4147371034825253, "grad_norm": 0.5433849139488359, "learning_rate": 6.351735474550731e-05, "loss": 2.7579, "step": 22790 }, { "epoch": 1.4147991805822833, "grad_norm": 0.5938538912179367, "learning_rate": 6.351387766866832e-05, "loss": 2.745, "step": 22791 }, { "epoch": 1.4148612576820412, "grad_norm": 0.5049052722296007, "learning_rate": 6.351040052132415e-05, "loss": 2.799, "step": 22792 }, { "epoch": 1.414923334781799, "grad_norm": 0.6533061255495562, "learning_rate": 6.350692330349294e-05, "loss": 2.8247, "step": 22793 }, { "epoch": 1.414985411881557, "grad_norm": 0.5083578262786231, "learning_rate": 6.350344601519286e-05, "loss": 2.7297, "step": 22794 }, { "epoch": 1.4150474889813147, "grad_norm": 0.6078726025360918, "learning_rate": 6.349996865644199e-05, "loss": 2.7185, "step": 22795 }, { "epoch": 1.4151095660810726, "grad_norm": 0.4981434896667994, "learning_rate": 6.349649122725853e-05, "loss": 2.8125, "step": 22796 }, { "epoch": 1.4151716431808306, "grad_norm": 0.6573448131453846, "learning_rate": 6.349301372766058e-05, "loss": 2.8269, "step": 22797 }, { "epoch": 1.4152337202805885, "grad_norm": 0.5554701159925873, "learning_rate": 6.348953615766631e-05, "loss": 2.7653, "step": 22798 }, { "epoch": 1.4152957973803464, "grad_norm": 0.6359140190269685, "learning_rate": 6.348605851729386e-05, "loss": 2.8107, "step": 22799 }, { "epoch": 1.4153578744801043, "grad_norm": 0.6024767875948472, "learning_rate": 6.348258080656136e-05, "loss": 2.7368, "step": 22800 }, { "epoch": 1.4154199515798622, "grad_norm": 0.5182341826178585, "learning_rate": 6.347910302548698e-05, "loss": 2.724, "step": 22801 }, { "epoch": 1.4154820286796201, "grad_norm": 0.5643922713584604, "learning_rate": 6.347562517408883e-05, "loss": 2.854, "step": 22802 }, { "epoch": 1.415544105779378, "grad_norm": 0.5801474533010035, "learning_rate": 6.347214725238507e-05, "loss": 2.7612, "step": 22803 }, { "epoch": 1.4156061828791358, "grad_norm": 0.6708689853916749, "learning_rate": 6.346866926039386e-05, "loss": 2.722, "step": 22804 }, { "epoch": 1.4156682599788937, "grad_norm": 0.5721663821879271, "learning_rate": 6.346519119813332e-05, "loss": 2.7801, "step": 22805 }, { "epoch": 1.4157303370786516, "grad_norm": 0.7546437268564519, "learning_rate": 6.34617130656216e-05, "loss": 2.8739, "step": 22806 }, { "epoch": 1.4157924141784095, "grad_norm": 0.632006944865687, "learning_rate": 6.345823486287687e-05, "loss": 2.8175, "step": 22807 }, { "epoch": 1.4158544912781674, "grad_norm": 0.6749081566811008, "learning_rate": 6.345475658991727e-05, "loss": 2.8571, "step": 22808 }, { "epoch": 1.4159165683779253, "grad_norm": 0.6262087339331897, "learning_rate": 6.345127824676093e-05, "loss": 2.8111, "step": 22809 }, { "epoch": 1.4159786454776833, "grad_norm": 0.619744063934852, "learning_rate": 6.3447799833426e-05, "loss": 2.8161, "step": 22810 }, { "epoch": 1.4160407225774412, "grad_norm": 0.6518771621580466, "learning_rate": 6.344432134993063e-05, "loss": 2.8549, "step": 22811 }, { "epoch": 1.416102799677199, "grad_norm": 0.6155181472529482, "learning_rate": 6.3440842796293e-05, "loss": 2.8166, "step": 22812 }, { "epoch": 1.416164876776957, "grad_norm": 0.599190183353991, "learning_rate": 6.34373641725312e-05, "loss": 2.8246, "step": 22813 }, { "epoch": 1.416226953876715, "grad_norm": 0.5850179195512392, "learning_rate": 6.343388547866343e-05, "loss": 2.785, "step": 22814 }, { "epoch": 1.4162890309764729, "grad_norm": 0.5784586639489449, "learning_rate": 6.34304067147078e-05, "loss": 2.7912, "step": 22815 }, { "epoch": 1.4163511080762308, "grad_norm": 0.5972964132293553, "learning_rate": 6.34269278806825e-05, "loss": 2.7167, "step": 22816 }, { "epoch": 1.4164131851759887, "grad_norm": 0.610046090550086, "learning_rate": 6.342344897660564e-05, "loss": 2.8051, "step": 22817 }, { "epoch": 1.4164752622757466, "grad_norm": 0.6678695158835909, "learning_rate": 6.341997000249538e-05, "loss": 2.8859, "step": 22818 }, { "epoch": 1.4165373393755043, "grad_norm": 0.6008580281539263, "learning_rate": 6.34164909583699e-05, "loss": 2.7391, "step": 22819 }, { "epoch": 1.4165994164752622, "grad_norm": 0.6008055246754392, "learning_rate": 6.341301184424733e-05, "loss": 2.8144, "step": 22820 }, { "epoch": 1.4166614935750201, "grad_norm": 0.6404961944976248, "learning_rate": 6.340953266014581e-05, "loss": 2.9198, "step": 22821 }, { "epoch": 1.416723570674778, "grad_norm": 0.5249772555102024, "learning_rate": 6.340605340608349e-05, "loss": 2.7402, "step": 22822 }, { "epoch": 1.416785647774536, "grad_norm": 0.6405456075087009, "learning_rate": 6.340257408207855e-05, "loss": 2.8887, "step": 22823 }, { "epoch": 1.416847724874294, "grad_norm": 0.543051486949959, "learning_rate": 6.339909468814913e-05, "loss": 2.7902, "step": 22824 }, { "epoch": 1.4169098019740518, "grad_norm": 0.540115544178778, "learning_rate": 6.339561522431337e-05, "loss": 2.8081, "step": 22825 }, { "epoch": 1.4169718790738097, "grad_norm": 0.5599405817883163, "learning_rate": 6.339213569058947e-05, "loss": 2.7838, "step": 22826 }, { "epoch": 1.4170339561735676, "grad_norm": 0.5308769262470664, "learning_rate": 6.33886560869955e-05, "loss": 2.7639, "step": 22827 }, { "epoch": 1.4170960332733253, "grad_norm": 0.5062286914061441, "learning_rate": 6.338517641354968e-05, "loss": 2.7515, "step": 22828 }, { "epoch": 1.4171581103730833, "grad_norm": 0.5651796937973115, "learning_rate": 6.338169667027015e-05, "loss": 2.8089, "step": 22829 }, { "epoch": 1.4172201874728412, "grad_norm": 0.49503823879200204, "learning_rate": 6.337821685717504e-05, "loss": 2.8009, "step": 22830 }, { "epoch": 1.417282264572599, "grad_norm": 0.559830476953265, "learning_rate": 6.337473697428255e-05, "loss": 2.768, "step": 22831 }, { "epoch": 1.417344341672357, "grad_norm": 0.5872702128901125, "learning_rate": 6.33712570216108e-05, "loss": 2.7858, "step": 22832 }, { "epoch": 1.417406418772115, "grad_norm": 0.5867668386122805, "learning_rate": 6.336777699917795e-05, "loss": 2.8223, "step": 22833 }, { "epoch": 1.4174684958718728, "grad_norm": 0.5089524312509273, "learning_rate": 6.336429690700217e-05, "loss": 2.8606, "step": 22834 }, { "epoch": 1.4175305729716308, "grad_norm": 0.5384775941414787, "learning_rate": 6.336081674510161e-05, "loss": 2.8097, "step": 22835 }, { "epoch": 1.4175926500713887, "grad_norm": 0.5595856775967066, "learning_rate": 6.335733651349442e-05, "loss": 2.7793, "step": 22836 }, { "epoch": 1.4176547271711466, "grad_norm": 0.6187406144249131, "learning_rate": 6.335385621219877e-05, "loss": 2.7735, "step": 22837 }, { "epoch": 1.4177168042709045, "grad_norm": 0.6588892317978837, "learning_rate": 6.33503758412328e-05, "loss": 2.7351, "step": 22838 }, { "epoch": 1.4177788813706624, "grad_norm": 0.6775121626870716, "learning_rate": 6.334689540061468e-05, "loss": 2.9829, "step": 22839 }, { "epoch": 1.4178409584704204, "grad_norm": 0.6998945412214861, "learning_rate": 6.334341489036256e-05, "loss": 2.7915, "step": 22840 }, { "epoch": 1.4179030355701783, "grad_norm": 0.643472483917336, "learning_rate": 6.333993431049462e-05, "loss": 2.7194, "step": 22841 }, { "epoch": 1.4179651126699362, "grad_norm": 0.6037348623727685, "learning_rate": 6.333645366102901e-05, "loss": 2.8709, "step": 22842 }, { "epoch": 1.4180271897696939, "grad_norm": 0.6198546326866864, "learning_rate": 6.333297294198387e-05, "loss": 2.8008, "step": 22843 }, { "epoch": 1.4180892668694518, "grad_norm": 0.5612229579876323, "learning_rate": 6.332949215337738e-05, "loss": 2.6775, "step": 22844 }, { "epoch": 1.4181513439692097, "grad_norm": 0.6210762837191292, "learning_rate": 6.332601129522769e-05, "loss": 2.8507, "step": 22845 }, { "epoch": 1.4182134210689676, "grad_norm": 0.5533266702024826, "learning_rate": 6.332253036755296e-05, "loss": 2.8874, "step": 22846 }, { "epoch": 1.4182754981687256, "grad_norm": 0.6472363434857796, "learning_rate": 6.331904937037137e-05, "loss": 2.8356, "step": 22847 }, { "epoch": 1.4183375752684835, "grad_norm": 0.5835383060780291, "learning_rate": 6.331556830370105e-05, "loss": 2.8555, "step": 22848 }, { "epoch": 1.4183996523682414, "grad_norm": 0.6177204342107662, "learning_rate": 6.331208716756019e-05, "loss": 2.7681, "step": 22849 }, { "epoch": 1.4184617294679993, "grad_norm": 0.5970264682014214, "learning_rate": 6.330860596196692e-05, "loss": 2.7617, "step": 22850 }, { "epoch": 1.4185238065677572, "grad_norm": 0.5818234783271544, "learning_rate": 6.330512468693945e-05, "loss": 2.795, "step": 22851 }, { "epoch": 1.418585883667515, "grad_norm": 0.5967562262710092, "learning_rate": 6.33016433424959e-05, "loss": 2.7718, "step": 22852 }, { "epoch": 1.4186479607672728, "grad_norm": 0.57073360729863, "learning_rate": 6.329816192865443e-05, "loss": 2.7588, "step": 22853 }, { "epoch": 1.4187100378670308, "grad_norm": 0.5494194510555355, "learning_rate": 6.329468044543326e-05, "loss": 2.7412, "step": 22854 }, { "epoch": 1.4187721149667887, "grad_norm": 0.5398290684604071, "learning_rate": 6.329119889285047e-05, "loss": 2.7785, "step": 22855 }, { "epoch": 1.4188341920665466, "grad_norm": 0.60143053167834, "learning_rate": 6.32877172709243e-05, "loss": 2.8193, "step": 22856 }, { "epoch": 1.4188962691663045, "grad_norm": 0.6353759179312233, "learning_rate": 6.328423557967287e-05, "loss": 2.8091, "step": 22857 }, { "epoch": 1.4189583462660624, "grad_norm": 0.6178847611520816, "learning_rate": 6.328075381911435e-05, "loss": 2.8406, "step": 22858 }, { "epoch": 1.4190204233658203, "grad_norm": 0.6167389169028847, "learning_rate": 6.327727198926694e-05, "loss": 2.8403, "step": 22859 }, { "epoch": 1.4190825004655783, "grad_norm": 0.6462088744102852, "learning_rate": 6.327379009014877e-05, "loss": 2.8385, "step": 22860 }, { "epoch": 1.4191445775653362, "grad_norm": 0.6478607646298951, "learning_rate": 6.3270308121778e-05, "loss": 2.7808, "step": 22861 }, { "epoch": 1.419206654665094, "grad_norm": 0.5589122797699664, "learning_rate": 6.326682608417284e-05, "loss": 2.8457, "step": 22862 }, { "epoch": 1.419268731764852, "grad_norm": 0.6140282813091735, "learning_rate": 6.32633439773514e-05, "loss": 2.8539, "step": 22863 }, { "epoch": 1.41933080886461, "grad_norm": 0.6193531044824705, "learning_rate": 6.325986180133189e-05, "loss": 2.8547, "step": 22864 }, { "epoch": 1.4193928859643679, "grad_norm": 0.5525471670204299, "learning_rate": 6.325637955613245e-05, "loss": 2.8272, "step": 22865 }, { "epoch": 1.4194549630641258, "grad_norm": 0.6290090725643193, "learning_rate": 6.325289724177127e-05, "loss": 2.8224, "step": 22866 }, { "epoch": 1.4195170401638835, "grad_norm": 0.6190561664147545, "learning_rate": 6.32494148582665e-05, "loss": 2.8381, "step": 22867 }, { "epoch": 1.4195791172636414, "grad_norm": 0.5413223388732605, "learning_rate": 6.324593240563632e-05, "loss": 2.7216, "step": 22868 }, { "epoch": 1.4196411943633993, "grad_norm": 0.6797502642185922, "learning_rate": 6.32424498838989e-05, "loss": 2.7565, "step": 22869 }, { "epoch": 1.4197032714631572, "grad_norm": 0.6295053933392063, "learning_rate": 6.323896729307239e-05, "loss": 2.8049, "step": 22870 }, { "epoch": 1.4197653485629151, "grad_norm": 0.719850940904415, "learning_rate": 6.3235484633175e-05, "loss": 2.79, "step": 22871 }, { "epoch": 1.419827425662673, "grad_norm": 0.5756943873978195, "learning_rate": 6.323200190422483e-05, "loss": 2.7871, "step": 22872 }, { "epoch": 1.419889502762431, "grad_norm": 0.6226334590402336, "learning_rate": 6.322851910624013e-05, "loss": 2.7959, "step": 22873 }, { "epoch": 1.419951579862189, "grad_norm": 0.5889289279537779, "learning_rate": 6.322503623923902e-05, "loss": 2.7926, "step": 22874 }, { "epoch": 1.4200136569619468, "grad_norm": 0.6017113290951714, "learning_rate": 6.322155330323969e-05, "loss": 2.7542, "step": 22875 }, { "epoch": 1.4200757340617045, "grad_norm": 0.6370978367811417, "learning_rate": 6.32180702982603e-05, "loss": 2.8222, "step": 22876 }, { "epoch": 1.4201378111614624, "grad_norm": 0.6307407657387321, "learning_rate": 6.321458722431904e-05, "loss": 2.72, "step": 22877 }, { "epoch": 1.4201998882612203, "grad_norm": 0.5204279339102327, "learning_rate": 6.321110408143407e-05, "loss": 2.7107, "step": 22878 }, { "epoch": 1.4202619653609783, "grad_norm": 0.6139895970988528, "learning_rate": 6.320762086962357e-05, "loss": 2.6918, "step": 22879 }, { "epoch": 1.4203240424607362, "grad_norm": 0.5414165292179652, "learning_rate": 6.320413758890568e-05, "loss": 2.7644, "step": 22880 }, { "epoch": 1.420386119560494, "grad_norm": 0.636677746534885, "learning_rate": 6.320065423929862e-05, "loss": 2.8152, "step": 22881 }, { "epoch": 1.420448196660252, "grad_norm": 0.5839194577473732, "learning_rate": 6.319717082082054e-05, "loss": 2.7908, "step": 22882 }, { "epoch": 1.42051027376001, "grad_norm": 0.650718216974659, "learning_rate": 6.31936873334896e-05, "loss": 2.6862, "step": 22883 }, { "epoch": 1.4205723508597679, "grad_norm": 0.5995040453121326, "learning_rate": 6.3190203777324e-05, "loss": 2.7757, "step": 22884 }, { "epoch": 1.4206344279595258, "grad_norm": 0.7098556083614029, "learning_rate": 6.318672015234192e-05, "loss": 2.7802, "step": 22885 }, { "epoch": 1.4206965050592837, "grad_norm": 0.6634963820099377, "learning_rate": 6.31832364585615e-05, "loss": 2.7671, "step": 22886 }, { "epoch": 1.4207585821590416, "grad_norm": 0.5747568113795372, "learning_rate": 6.317975269600093e-05, "loss": 2.7977, "step": 22887 }, { "epoch": 1.4208206592587995, "grad_norm": 0.63873773637708, "learning_rate": 6.317626886467841e-05, "loss": 2.6983, "step": 22888 }, { "epoch": 1.4208827363585574, "grad_norm": 0.5343811380580381, "learning_rate": 6.31727849646121e-05, "loss": 2.8807, "step": 22889 }, { "epoch": 1.4209448134583154, "grad_norm": 0.59726156030024, "learning_rate": 6.316930099582015e-05, "loss": 2.7499, "step": 22890 }, { "epoch": 1.421006890558073, "grad_norm": 0.5746018306151327, "learning_rate": 6.316581695832077e-05, "loss": 2.8508, "step": 22891 }, { "epoch": 1.421068967657831, "grad_norm": 0.6140262428043421, "learning_rate": 6.316233285213213e-05, "loss": 2.7433, "step": 22892 }, { "epoch": 1.421131044757589, "grad_norm": 0.5273543462748344, "learning_rate": 6.315884867727241e-05, "loss": 2.7717, "step": 22893 }, { "epoch": 1.4211931218573468, "grad_norm": 0.5496248591450833, "learning_rate": 6.315536443375977e-05, "loss": 2.7447, "step": 22894 }, { "epoch": 1.4212551989571047, "grad_norm": 0.5635045671032534, "learning_rate": 6.315188012161241e-05, "loss": 2.7894, "step": 22895 }, { "epoch": 1.4213172760568626, "grad_norm": 0.5528422004480044, "learning_rate": 6.314839574084848e-05, "loss": 2.7688, "step": 22896 }, { "epoch": 1.4213793531566206, "grad_norm": 0.5581959461687528, "learning_rate": 6.31449112914862e-05, "loss": 2.7641, "step": 22897 }, { "epoch": 1.4214414302563785, "grad_norm": 0.6717497686734243, "learning_rate": 6.314142677354371e-05, "loss": 2.773, "step": 22898 }, { "epoch": 1.4215035073561364, "grad_norm": 0.47333750219535237, "learning_rate": 6.313794218703922e-05, "loss": 2.7087, "step": 22899 }, { "epoch": 1.421565584455894, "grad_norm": 0.6101254219736783, "learning_rate": 6.31344575319909e-05, "loss": 2.7892, "step": 22900 }, { "epoch": 1.421627661555652, "grad_norm": 0.5311023001951967, "learning_rate": 6.313097280841691e-05, "loss": 2.7632, "step": 22901 }, { "epoch": 1.42168973865541, "grad_norm": 0.6510699278737435, "learning_rate": 6.312748801633545e-05, "loss": 2.9101, "step": 22902 }, { "epoch": 1.4217518157551678, "grad_norm": 0.5908215735556015, "learning_rate": 6.312400315576471e-05, "loss": 2.7547, "step": 22903 }, { "epoch": 1.4218138928549258, "grad_norm": 0.5645465971430883, "learning_rate": 6.312051822672285e-05, "loss": 2.8008, "step": 22904 }, { "epoch": 1.4218759699546837, "grad_norm": 0.6213166178900791, "learning_rate": 6.311703322922806e-05, "loss": 2.8414, "step": 22905 }, { "epoch": 1.4219380470544416, "grad_norm": 0.5471225450742612, "learning_rate": 6.311354816329853e-05, "loss": 2.8133, "step": 22906 }, { "epoch": 1.4220001241541995, "grad_norm": 0.5699265101390077, "learning_rate": 6.311006302895244e-05, "loss": 2.8153, "step": 22907 }, { "epoch": 1.4220622012539574, "grad_norm": 0.5478262937977633, "learning_rate": 6.310657782620794e-05, "loss": 2.823, "step": 22908 }, { "epoch": 1.4221242783537154, "grad_norm": 0.5710660841642421, "learning_rate": 6.310309255508328e-05, "loss": 2.8485, "step": 22909 }, { "epoch": 1.4221863554534733, "grad_norm": 0.6209553498388836, "learning_rate": 6.309960721559658e-05, "loss": 2.7885, "step": 22910 }, { "epoch": 1.4222484325532312, "grad_norm": 0.5253379256468109, "learning_rate": 6.309612180776606e-05, "loss": 2.8411, "step": 22911 }, { "epoch": 1.422310509652989, "grad_norm": 0.607584229387498, "learning_rate": 6.30926363316099e-05, "loss": 2.8217, "step": 22912 }, { "epoch": 1.422372586752747, "grad_norm": 0.5294622184126748, "learning_rate": 6.308915078714626e-05, "loss": 2.813, "step": 22913 }, { "epoch": 1.422434663852505, "grad_norm": 0.5702496846761338, "learning_rate": 6.308566517439335e-05, "loss": 2.7228, "step": 22914 }, { "epoch": 1.4224967409522626, "grad_norm": 0.5744328002571415, "learning_rate": 6.308217949336935e-05, "loss": 2.8174, "step": 22915 }, { "epoch": 1.4225588180520206, "grad_norm": 0.5558648576285115, "learning_rate": 6.307869374409243e-05, "loss": 2.8474, "step": 22916 }, { "epoch": 1.4226208951517785, "grad_norm": 0.6175720610251513, "learning_rate": 6.307520792658082e-05, "loss": 2.9037, "step": 22917 }, { "epoch": 1.4226829722515364, "grad_norm": 0.5526814519664954, "learning_rate": 6.307172204085265e-05, "loss": 2.7822, "step": 22918 }, { "epoch": 1.4227450493512943, "grad_norm": 0.5958352614640559, "learning_rate": 6.306823608692615e-05, "loss": 2.9242, "step": 22919 }, { "epoch": 1.4228071264510522, "grad_norm": 0.5057494585370924, "learning_rate": 6.306475006481949e-05, "loss": 2.7604, "step": 22920 }, { "epoch": 1.4228692035508101, "grad_norm": 0.6003465625128558, "learning_rate": 6.306126397455083e-05, "loss": 2.8828, "step": 22921 }, { "epoch": 1.422931280650568, "grad_norm": 0.563894318633828, "learning_rate": 6.30577778161384e-05, "loss": 2.7764, "step": 22922 }, { "epoch": 1.422993357750326, "grad_norm": 0.5196063052006308, "learning_rate": 6.305429158960038e-05, "loss": 2.8459, "step": 22923 }, { "epoch": 1.4230554348500837, "grad_norm": 0.5704855199191508, "learning_rate": 6.305080529495495e-05, "loss": 2.7563, "step": 22924 }, { "epoch": 1.4231175119498416, "grad_norm": 0.48867943359188865, "learning_rate": 6.304731893222031e-05, "loss": 2.7439, "step": 22925 }, { "epoch": 1.4231795890495995, "grad_norm": 0.592990766274866, "learning_rate": 6.304383250141462e-05, "loss": 2.8195, "step": 22926 }, { "epoch": 1.4232416661493574, "grad_norm": 0.5295010837889649, "learning_rate": 6.304034600255611e-05, "loss": 2.7835, "step": 22927 }, { "epoch": 1.4233037432491153, "grad_norm": 0.5360857833497175, "learning_rate": 6.303685943566293e-05, "loss": 2.8417, "step": 22928 }, { "epoch": 1.4233658203488733, "grad_norm": 0.6030688808205199, "learning_rate": 6.30333728007533e-05, "loss": 2.7763, "step": 22929 }, { "epoch": 1.4234278974486312, "grad_norm": 0.5739274616096774, "learning_rate": 6.302988609784539e-05, "loss": 2.8347, "step": 22930 }, { "epoch": 1.423489974548389, "grad_norm": 0.6055982635784956, "learning_rate": 6.302639932695742e-05, "loss": 2.7777, "step": 22931 }, { "epoch": 1.423552051648147, "grad_norm": 0.5682081040891057, "learning_rate": 6.302291248810754e-05, "loss": 2.8446, "step": 22932 }, { "epoch": 1.423614128747905, "grad_norm": 0.5915937925063373, "learning_rate": 6.301942558131398e-05, "loss": 2.8252, "step": 22933 }, { "epoch": 1.4236762058476629, "grad_norm": 0.5544520670964164, "learning_rate": 6.301593860659492e-05, "loss": 2.7255, "step": 22934 }, { "epoch": 1.4237382829474208, "grad_norm": 0.5500306500217634, "learning_rate": 6.301245156396854e-05, "loss": 2.8056, "step": 22935 }, { "epoch": 1.4238003600471787, "grad_norm": 0.5594030000295924, "learning_rate": 6.300896445345304e-05, "loss": 2.8672, "step": 22936 }, { "epoch": 1.4238624371469366, "grad_norm": 0.6011306153652091, "learning_rate": 6.30054772750666e-05, "loss": 2.7916, "step": 22937 }, { "epoch": 1.4239245142466945, "grad_norm": 0.5009771573842107, "learning_rate": 6.300199002882745e-05, "loss": 2.8037, "step": 22938 }, { "epoch": 1.4239865913464522, "grad_norm": 0.6647270243613351, "learning_rate": 6.299850271475376e-05, "loss": 2.8448, "step": 22939 }, { "epoch": 1.4240486684462101, "grad_norm": 0.5368384215358446, "learning_rate": 6.299501533286371e-05, "loss": 2.8394, "step": 22940 }, { "epoch": 1.424110745545968, "grad_norm": 0.6273183739115324, "learning_rate": 6.299152788317553e-05, "loss": 2.7271, "step": 22941 }, { "epoch": 1.424172822645726, "grad_norm": 0.5736694822971922, "learning_rate": 6.298804036570739e-05, "loss": 2.8118, "step": 22942 }, { "epoch": 1.424234899745484, "grad_norm": 0.5607353451855133, "learning_rate": 6.298455278047748e-05, "loss": 2.8711, "step": 22943 }, { "epoch": 1.4242969768452418, "grad_norm": 0.6058024809518604, "learning_rate": 6.298106512750403e-05, "loss": 2.774, "step": 22944 }, { "epoch": 1.4243590539449997, "grad_norm": 0.5584639285023912, "learning_rate": 6.29775774068052e-05, "loss": 2.7535, "step": 22945 }, { "epoch": 1.4244211310447576, "grad_norm": 0.6270147513170778, "learning_rate": 6.297408961839918e-05, "loss": 2.7535, "step": 22946 }, { "epoch": 1.4244832081445156, "grad_norm": 0.5470275393689505, "learning_rate": 6.29706017623042e-05, "loss": 2.7924, "step": 22947 }, { "epoch": 1.4245452852442733, "grad_norm": 0.5885442604300359, "learning_rate": 6.296711383853844e-05, "loss": 2.7719, "step": 22948 }, { "epoch": 1.4246073623440312, "grad_norm": 0.595391864442455, "learning_rate": 6.29636258471201e-05, "loss": 2.8864, "step": 22949 }, { "epoch": 1.424669439443789, "grad_norm": 0.5435997843153879, "learning_rate": 6.296013778806738e-05, "loss": 2.8113, "step": 22950 }, { "epoch": 1.424731516543547, "grad_norm": 0.5981032750440414, "learning_rate": 6.295664966139847e-05, "loss": 2.8098, "step": 22951 }, { "epoch": 1.424793593643305, "grad_norm": 0.5653883897527134, "learning_rate": 6.295316146713157e-05, "loss": 2.7726, "step": 22952 }, { "epoch": 1.4248556707430629, "grad_norm": 0.5874503885789807, "learning_rate": 6.294967320528489e-05, "loss": 2.8022, "step": 22953 }, { "epoch": 1.4249177478428208, "grad_norm": 0.5092311995495271, "learning_rate": 6.294618487587663e-05, "loss": 2.7366, "step": 22954 }, { "epoch": 1.4249798249425787, "grad_norm": 0.5263354233480216, "learning_rate": 6.294269647892494e-05, "loss": 2.7436, "step": 22955 }, { "epoch": 1.4250419020423366, "grad_norm": 0.5167040948488115, "learning_rate": 6.29392080144481e-05, "loss": 2.7303, "step": 22956 }, { "epoch": 1.4251039791420945, "grad_norm": 0.5594613599976853, "learning_rate": 6.293571948246427e-05, "loss": 2.9141, "step": 22957 }, { "epoch": 1.4251660562418524, "grad_norm": 0.4749363373457167, "learning_rate": 6.293223088299164e-05, "loss": 2.8219, "step": 22958 }, { "epoch": 1.4252281333416104, "grad_norm": 0.539607477445733, "learning_rate": 6.292874221604842e-05, "loss": 2.8405, "step": 22959 }, { "epoch": 1.4252902104413683, "grad_norm": 0.5182378918346167, "learning_rate": 6.292525348165281e-05, "loss": 2.7748, "step": 22960 }, { "epoch": 1.4253522875411262, "grad_norm": 0.5269891747768936, "learning_rate": 6.292176467982303e-05, "loss": 2.8497, "step": 22961 }, { "epoch": 1.4254143646408841, "grad_norm": 0.5423811058426938, "learning_rate": 6.291827581057725e-05, "loss": 2.7244, "step": 22962 }, { "epoch": 1.4254764417406418, "grad_norm": 0.5302013818950913, "learning_rate": 6.29147868739337e-05, "loss": 2.7705, "step": 22963 }, { "epoch": 1.4255385188403997, "grad_norm": 0.5597379210405332, "learning_rate": 6.291129786991058e-05, "loss": 2.7969, "step": 22964 }, { "epoch": 1.4256005959401576, "grad_norm": 0.5890359209777989, "learning_rate": 6.290780879852606e-05, "loss": 2.7953, "step": 22965 }, { "epoch": 1.4256626730399156, "grad_norm": 0.5074753742632516, "learning_rate": 6.29043196597984e-05, "loss": 2.7828, "step": 22966 }, { "epoch": 1.4257247501396735, "grad_norm": 0.6127565411078871, "learning_rate": 6.290083045374575e-05, "loss": 2.8328, "step": 22967 }, { "epoch": 1.4257868272394314, "grad_norm": 0.5351865161217143, "learning_rate": 6.289734118038634e-05, "loss": 2.8064, "step": 22968 }, { "epoch": 1.4258489043391893, "grad_norm": 0.5628333693325372, "learning_rate": 6.289385183973838e-05, "loss": 2.8383, "step": 22969 }, { "epoch": 1.4259109814389472, "grad_norm": 0.5871642939198124, "learning_rate": 6.289036243182003e-05, "loss": 2.8451, "step": 22970 }, { "epoch": 1.4259730585387052, "grad_norm": 0.5278113940754174, "learning_rate": 6.288687295664956e-05, "loss": 2.7536, "step": 22971 }, { "epoch": 1.4260351356384628, "grad_norm": 0.555273230328221, "learning_rate": 6.288338341424515e-05, "loss": 2.8064, "step": 22972 }, { "epoch": 1.4260972127382208, "grad_norm": 0.5478811817016274, "learning_rate": 6.287989380462498e-05, "loss": 2.7786, "step": 22973 }, { "epoch": 1.4261592898379787, "grad_norm": 0.5637009833229344, "learning_rate": 6.287640412780728e-05, "loss": 2.7071, "step": 22974 }, { "epoch": 1.4262213669377366, "grad_norm": 0.5550894825341869, "learning_rate": 6.287291438381026e-05, "loss": 2.8407, "step": 22975 }, { "epoch": 1.4262834440374945, "grad_norm": 0.5301463499980258, "learning_rate": 6.286942457265214e-05, "loss": 2.7961, "step": 22976 }, { "epoch": 1.4263455211372524, "grad_norm": 0.6103657469054916, "learning_rate": 6.286593469435108e-05, "loss": 2.8448, "step": 22977 }, { "epoch": 1.4264075982370104, "grad_norm": 0.5056514238919715, "learning_rate": 6.286244474892533e-05, "loss": 2.8248, "step": 22978 }, { "epoch": 1.4264696753367683, "grad_norm": 0.5138728687721918, "learning_rate": 6.285895473639308e-05, "loss": 2.6839, "step": 22979 }, { "epoch": 1.4265317524365262, "grad_norm": 0.5783217069692139, "learning_rate": 6.285546465677254e-05, "loss": 2.7846, "step": 22980 }, { "epoch": 1.426593829536284, "grad_norm": 0.5180559076365393, "learning_rate": 6.285197451008194e-05, "loss": 2.8283, "step": 22981 }, { "epoch": 1.426655906636042, "grad_norm": 0.5590255775434688, "learning_rate": 6.284848429633943e-05, "loss": 2.9013, "step": 22982 }, { "epoch": 1.4267179837358, "grad_norm": 0.5916768320535658, "learning_rate": 6.284499401556328e-05, "loss": 2.771, "step": 22983 }, { "epoch": 1.4267800608355579, "grad_norm": 0.5480206070666078, "learning_rate": 6.284150366777168e-05, "loss": 2.8005, "step": 22984 }, { "epoch": 1.4268421379353158, "grad_norm": 0.5934711604643628, "learning_rate": 6.283801325298284e-05, "loss": 2.8869, "step": 22985 }, { "epoch": 1.4269042150350737, "grad_norm": 0.6300486265065959, "learning_rate": 6.283452277121495e-05, "loss": 2.8887, "step": 22986 }, { "epoch": 1.4269662921348314, "grad_norm": 0.5707028503208427, "learning_rate": 6.283103222248626e-05, "loss": 2.779, "step": 22987 }, { "epoch": 1.4270283692345893, "grad_norm": 0.5576009130023695, "learning_rate": 6.282754160681494e-05, "loss": 2.7859, "step": 22988 }, { "epoch": 1.4270904463343472, "grad_norm": 0.5738918126276846, "learning_rate": 6.282405092421923e-05, "loss": 2.8714, "step": 22989 }, { "epoch": 1.4271525234341051, "grad_norm": 0.5803934903343047, "learning_rate": 6.282056017471735e-05, "loss": 2.7237, "step": 22990 }, { "epoch": 1.427214600533863, "grad_norm": 0.5412706125400221, "learning_rate": 6.281706935832748e-05, "loss": 2.873, "step": 22991 }, { "epoch": 1.427276677633621, "grad_norm": 0.5712137808471693, "learning_rate": 6.281357847506785e-05, "loss": 2.6672, "step": 22992 }, { "epoch": 1.427338754733379, "grad_norm": 0.5439415046380546, "learning_rate": 6.281008752495666e-05, "loss": 2.709, "step": 22993 }, { "epoch": 1.4274008318331368, "grad_norm": 0.6209898567876321, "learning_rate": 6.280659650801216e-05, "loss": 2.7798, "step": 22994 }, { "epoch": 1.4274629089328947, "grad_norm": 0.49312055037005464, "learning_rate": 6.28031054242525e-05, "loss": 2.7092, "step": 22995 }, { "epoch": 1.4275249860326524, "grad_norm": 0.5886025411355904, "learning_rate": 6.279961427369596e-05, "loss": 2.7763, "step": 22996 }, { "epoch": 1.4275870631324103, "grad_norm": 0.5679665220445175, "learning_rate": 6.279612305636073e-05, "loss": 2.8171, "step": 22997 }, { "epoch": 1.4276491402321683, "grad_norm": 0.542441489452781, "learning_rate": 6.2792631772265e-05, "loss": 2.8167, "step": 22998 }, { "epoch": 1.4277112173319262, "grad_norm": 0.5616608294043122, "learning_rate": 6.278914042142701e-05, "loss": 2.798, "step": 22999 }, { "epoch": 1.427773294431684, "grad_norm": 0.6004170746397088, "learning_rate": 6.278564900386496e-05, "loss": 2.7572, "step": 23000 }, { "epoch": 1.427835371531442, "grad_norm": 0.5384387401847704, "learning_rate": 6.27821575195971e-05, "loss": 2.7946, "step": 23001 }, { "epoch": 1.4278974486312, "grad_norm": 0.6212254270686677, "learning_rate": 6.277866596864159e-05, "loss": 2.7729, "step": 23002 }, { "epoch": 1.4279595257309579, "grad_norm": 0.5892088295994944, "learning_rate": 6.277517435101669e-05, "loss": 2.8405, "step": 23003 }, { "epoch": 1.4280216028307158, "grad_norm": 0.573016640581294, "learning_rate": 6.277168266674063e-05, "loss": 2.8721, "step": 23004 }, { "epoch": 1.4280836799304737, "grad_norm": 0.6271021300463239, "learning_rate": 6.276819091583157e-05, "loss": 2.6985, "step": 23005 }, { "epoch": 1.4281457570302316, "grad_norm": 0.5911350893453083, "learning_rate": 6.276469909830776e-05, "loss": 2.6789, "step": 23006 }, { "epoch": 1.4282078341299895, "grad_norm": 0.5469508254753849, "learning_rate": 6.276120721418744e-05, "loss": 2.785, "step": 23007 }, { "epoch": 1.4282699112297474, "grad_norm": 0.6490786726088776, "learning_rate": 6.275771526348879e-05, "loss": 2.8648, "step": 23008 }, { "epoch": 1.4283319883295054, "grad_norm": 0.5955360553533771, "learning_rate": 6.275422324623003e-05, "loss": 2.9297, "step": 23009 }, { "epoch": 1.4283940654292633, "grad_norm": 0.6070660349184195, "learning_rate": 6.27507311624294e-05, "loss": 2.8377, "step": 23010 }, { "epoch": 1.428456142529021, "grad_norm": 0.559576144362812, "learning_rate": 6.274723901210511e-05, "loss": 2.7416, "step": 23011 }, { "epoch": 1.428518219628779, "grad_norm": 0.5609748881001396, "learning_rate": 6.274374679527539e-05, "loss": 2.796, "step": 23012 }, { "epoch": 1.4285802967285368, "grad_norm": 0.574188880403038, "learning_rate": 6.274025451195842e-05, "loss": 2.8405, "step": 23013 }, { "epoch": 1.4286423738282947, "grad_norm": 0.5815284144248217, "learning_rate": 6.273676216217249e-05, "loss": 2.7973, "step": 23014 }, { "epoch": 1.4287044509280526, "grad_norm": 0.5125171460795809, "learning_rate": 6.273326974593576e-05, "loss": 2.8441, "step": 23015 }, { "epoch": 1.4287665280278106, "grad_norm": 0.6525756058426812, "learning_rate": 6.272977726326646e-05, "loss": 2.9557, "step": 23016 }, { "epoch": 1.4288286051275685, "grad_norm": 0.6016286590658291, "learning_rate": 6.272628471418285e-05, "loss": 2.6874, "step": 23017 }, { "epoch": 1.4288906822273264, "grad_norm": 0.5636214626977054, "learning_rate": 6.27227920987031e-05, "loss": 2.8339, "step": 23018 }, { "epoch": 1.4289527593270843, "grad_norm": 0.6917257917557159, "learning_rate": 6.271929941684545e-05, "loss": 2.7887, "step": 23019 }, { "epoch": 1.429014836426842, "grad_norm": 0.557637763771114, "learning_rate": 6.271580666862813e-05, "loss": 2.7877, "step": 23020 }, { "epoch": 1.4290769135266, "grad_norm": 0.72858123892903, "learning_rate": 6.271231385406938e-05, "loss": 2.7759, "step": 23021 }, { "epoch": 1.4291389906263579, "grad_norm": 0.5484122604692, "learning_rate": 6.27088209731874e-05, "loss": 2.7598, "step": 23022 }, { "epoch": 1.4292010677261158, "grad_norm": 0.6785751536274249, "learning_rate": 6.27053280260004e-05, "loss": 2.8096, "step": 23023 }, { "epoch": 1.4292631448258737, "grad_norm": 0.5716364920783334, "learning_rate": 6.270183501252662e-05, "loss": 2.7748, "step": 23024 }, { "epoch": 1.4293252219256316, "grad_norm": 0.6033056443229071, "learning_rate": 6.26983419327843e-05, "loss": 2.8315, "step": 23025 }, { "epoch": 1.4293872990253895, "grad_norm": 0.6133869833217801, "learning_rate": 6.269484878679163e-05, "loss": 2.8005, "step": 23026 }, { "epoch": 1.4294493761251474, "grad_norm": 0.6116741285943018, "learning_rate": 6.269135557456686e-05, "loss": 2.7702, "step": 23027 }, { "epoch": 1.4295114532249054, "grad_norm": 0.6199128542690254, "learning_rate": 6.268786229612821e-05, "loss": 2.8366, "step": 23028 }, { "epoch": 1.4295735303246633, "grad_norm": 0.6013068609652261, "learning_rate": 6.26843689514939e-05, "loss": 2.904, "step": 23029 }, { "epoch": 1.4296356074244212, "grad_norm": 0.6564943903039108, "learning_rate": 6.268087554068215e-05, "loss": 2.7961, "step": 23030 }, { "epoch": 1.4296976845241791, "grad_norm": 0.6428915957838612, "learning_rate": 6.267738206371119e-05, "loss": 2.7675, "step": 23031 }, { "epoch": 1.429759761623937, "grad_norm": 0.5860936020495259, "learning_rate": 6.267388852059926e-05, "loss": 2.8101, "step": 23032 }, { "epoch": 1.429821838723695, "grad_norm": 0.6282075116066407, "learning_rate": 6.267039491136459e-05, "loss": 2.8246, "step": 23033 }, { "epoch": 1.4298839158234529, "grad_norm": 0.6230430004991941, "learning_rate": 6.266690123602538e-05, "loss": 2.8111, "step": 23034 }, { "epoch": 1.4299459929232106, "grad_norm": 0.5873818461329086, "learning_rate": 6.266340749459985e-05, "loss": 2.7826, "step": 23035 }, { "epoch": 1.4300080700229685, "grad_norm": 0.5825845021853778, "learning_rate": 6.265991368710628e-05, "loss": 2.673, "step": 23036 }, { "epoch": 1.4300701471227264, "grad_norm": 0.5614737845319486, "learning_rate": 6.265641981356286e-05, "loss": 2.9049, "step": 23037 }, { "epoch": 1.4301322242224843, "grad_norm": 0.5474372970261879, "learning_rate": 6.265292587398781e-05, "loss": 2.7008, "step": 23038 }, { "epoch": 1.4301943013222422, "grad_norm": 0.6550308437947631, "learning_rate": 6.26494318683994e-05, "loss": 2.9547, "step": 23039 }, { "epoch": 1.4302563784220002, "grad_norm": 0.5651016205995456, "learning_rate": 6.264593779681582e-05, "loss": 2.8249, "step": 23040 }, { "epoch": 1.430318455521758, "grad_norm": 0.6176360226828604, "learning_rate": 6.264244365925532e-05, "loss": 2.7786, "step": 23041 }, { "epoch": 1.430380532621516, "grad_norm": 0.5976194731922151, "learning_rate": 6.263894945573612e-05, "loss": 2.8992, "step": 23042 }, { "epoch": 1.4304426097212737, "grad_norm": 0.6443577388371814, "learning_rate": 6.263545518627643e-05, "loss": 2.8576, "step": 23043 }, { "epoch": 1.4305046868210316, "grad_norm": 0.5894592497187009, "learning_rate": 6.263196085089454e-05, "loss": 2.7487, "step": 23044 }, { "epoch": 1.4305667639207895, "grad_norm": 0.6351862526126629, "learning_rate": 6.262846644960862e-05, "loss": 2.8235, "step": 23045 }, { "epoch": 1.4306288410205474, "grad_norm": 0.5759307615452052, "learning_rate": 6.262497198243694e-05, "loss": 2.7625, "step": 23046 }, { "epoch": 1.4306909181203054, "grad_norm": 0.6505563885572779, "learning_rate": 6.26214774493977e-05, "loss": 2.8162, "step": 23047 }, { "epoch": 1.4307529952200633, "grad_norm": 0.5352460131579126, "learning_rate": 6.261798285050915e-05, "loss": 2.8378, "step": 23048 }, { "epoch": 1.4308150723198212, "grad_norm": 0.623281070138289, "learning_rate": 6.261448818578952e-05, "loss": 2.8624, "step": 23049 }, { "epoch": 1.430877149419579, "grad_norm": 0.609867263915326, "learning_rate": 6.261099345525705e-05, "loss": 2.7426, "step": 23050 }, { "epoch": 1.430939226519337, "grad_norm": 0.6186929281481981, "learning_rate": 6.260749865892995e-05, "loss": 2.829, "step": 23051 }, { "epoch": 1.431001303619095, "grad_norm": 0.624100920593348, "learning_rate": 6.260400379682649e-05, "loss": 2.8291, "step": 23052 }, { "epoch": 1.4310633807188529, "grad_norm": 0.649883977915682, "learning_rate": 6.260050886896487e-05, "loss": 2.8197, "step": 23053 }, { "epoch": 1.4311254578186108, "grad_norm": 0.5685254408227627, "learning_rate": 6.259701387536334e-05, "loss": 2.7725, "step": 23054 }, { "epoch": 1.4311875349183687, "grad_norm": 0.6173204689202857, "learning_rate": 6.259351881604013e-05, "loss": 2.8796, "step": 23055 }, { "epoch": 1.4312496120181266, "grad_norm": 0.5701087169534165, "learning_rate": 6.259002369101347e-05, "loss": 2.7778, "step": 23056 }, { "epoch": 1.4313116891178845, "grad_norm": 0.5803223169797779, "learning_rate": 6.258652850030163e-05, "loss": 2.7848, "step": 23057 }, { "epoch": 1.4313737662176422, "grad_norm": 0.5322208509226215, "learning_rate": 6.258303324392279e-05, "loss": 2.8505, "step": 23058 }, { "epoch": 1.4314358433174001, "grad_norm": 0.5651702812885072, "learning_rate": 6.25795379218952e-05, "loss": 2.8198, "step": 23059 }, { "epoch": 1.431497920417158, "grad_norm": 0.5635700908503766, "learning_rate": 6.257604253423712e-05, "loss": 2.8238, "step": 23060 }, { "epoch": 1.431559997516916, "grad_norm": 0.5902390576422379, "learning_rate": 6.257254708096677e-05, "loss": 2.7599, "step": 23061 }, { "epoch": 1.431622074616674, "grad_norm": 0.5361180815905155, "learning_rate": 6.25690515621024e-05, "loss": 2.7827, "step": 23062 }, { "epoch": 1.4316841517164318, "grad_norm": 0.6485137894575389, "learning_rate": 6.256555597766223e-05, "loss": 2.7452, "step": 23063 }, { "epoch": 1.4317462288161897, "grad_norm": 0.5589899911871601, "learning_rate": 6.256206032766451e-05, "loss": 2.8865, "step": 23064 }, { "epoch": 1.4318083059159477, "grad_norm": 0.6416336817953867, "learning_rate": 6.255856461212746e-05, "loss": 2.8111, "step": 23065 }, { "epoch": 1.4318703830157056, "grad_norm": 0.6208030692240337, "learning_rate": 6.255506883106934e-05, "loss": 2.8286, "step": 23066 }, { "epoch": 1.4319324601154633, "grad_norm": 0.6395852560041358, "learning_rate": 6.255157298450836e-05, "loss": 2.715, "step": 23067 }, { "epoch": 1.4319945372152212, "grad_norm": 0.5719748790216462, "learning_rate": 6.254807707246279e-05, "loss": 2.788, "step": 23068 }, { "epoch": 1.432056614314979, "grad_norm": 0.6401912569954161, "learning_rate": 6.254458109495086e-05, "loss": 2.8292, "step": 23069 }, { "epoch": 1.432118691414737, "grad_norm": 0.5995720718162135, "learning_rate": 6.254108505199078e-05, "loss": 2.7808, "step": 23070 }, { "epoch": 1.432180768514495, "grad_norm": 0.535057365024783, "learning_rate": 6.253758894360085e-05, "loss": 2.7878, "step": 23071 }, { "epoch": 1.4322428456142529, "grad_norm": 0.5465944933156442, "learning_rate": 6.253409276979926e-05, "loss": 2.8193, "step": 23072 }, { "epoch": 1.4323049227140108, "grad_norm": 0.5624210249243553, "learning_rate": 6.253059653060426e-05, "loss": 2.8433, "step": 23073 }, { "epoch": 1.4323669998137687, "grad_norm": 0.51523960044357, "learning_rate": 6.25271002260341e-05, "loss": 2.7073, "step": 23074 }, { "epoch": 1.4324290769135266, "grad_norm": 0.5988972812441832, "learning_rate": 6.252360385610703e-05, "loss": 2.853, "step": 23075 }, { "epoch": 1.4324911540132845, "grad_norm": 0.6294553267161211, "learning_rate": 6.252010742084127e-05, "loss": 2.8517, "step": 23076 }, { "epoch": 1.4325532311130424, "grad_norm": 0.5548542739097855, "learning_rate": 6.251661092025505e-05, "loss": 2.8323, "step": 23077 }, { "epoch": 1.4326153082128004, "grad_norm": 0.6256745512981576, "learning_rate": 6.251311435436666e-05, "loss": 2.8529, "step": 23078 }, { "epoch": 1.4326773853125583, "grad_norm": 0.5502934561645191, "learning_rate": 6.25096177231943e-05, "loss": 2.8152, "step": 23079 }, { "epoch": 1.4327394624123162, "grad_norm": 0.6204215196565296, "learning_rate": 6.250612102675623e-05, "loss": 2.7425, "step": 23080 }, { "epoch": 1.4328015395120741, "grad_norm": 0.5481755731581472, "learning_rate": 6.25026242650707e-05, "loss": 2.8109, "step": 23081 }, { "epoch": 1.4328636166118318, "grad_norm": 0.6769903314736806, "learning_rate": 6.249912743815595e-05, "loss": 2.8171, "step": 23082 }, { "epoch": 1.4329256937115897, "grad_norm": 0.5201657371767331, "learning_rate": 6.249563054603019e-05, "loss": 2.6833, "step": 23083 }, { "epoch": 1.4329877708113476, "grad_norm": 0.6544320038138525, "learning_rate": 6.24921335887117e-05, "loss": 2.8274, "step": 23084 }, { "epoch": 1.4330498479111056, "grad_norm": 0.54371934234308, "learning_rate": 6.248863656621872e-05, "loss": 2.7681, "step": 23085 }, { "epoch": 1.4331119250108635, "grad_norm": 0.5967513820199102, "learning_rate": 6.24851394785695e-05, "loss": 2.7408, "step": 23086 }, { "epoch": 1.4331740021106214, "grad_norm": 0.5460960289903493, "learning_rate": 6.248164232578227e-05, "loss": 2.8248, "step": 23087 }, { "epoch": 1.4332360792103793, "grad_norm": 0.6148139122359908, "learning_rate": 6.247814510787528e-05, "loss": 2.7981, "step": 23088 }, { "epoch": 1.4332981563101372, "grad_norm": 0.565058953956673, "learning_rate": 6.247464782486679e-05, "loss": 2.8745, "step": 23089 }, { "epoch": 1.4333602334098952, "grad_norm": 0.6141513071076241, "learning_rate": 6.247115047677502e-05, "loss": 2.8637, "step": 23090 }, { "epoch": 1.4334223105096529, "grad_norm": 0.6910223460800544, "learning_rate": 6.246765306361824e-05, "loss": 2.8406, "step": 23091 }, { "epoch": 1.4334843876094108, "grad_norm": 0.5770931830638634, "learning_rate": 6.246415558541467e-05, "loss": 2.8924, "step": 23092 }, { "epoch": 1.4335464647091687, "grad_norm": 0.6277958087971428, "learning_rate": 6.24606580421826e-05, "loss": 2.8566, "step": 23093 }, { "epoch": 1.4336085418089266, "grad_norm": 0.5910633186414531, "learning_rate": 6.245716043394024e-05, "loss": 2.8, "step": 23094 }, { "epoch": 1.4336706189086845, "grad_norm": 0.6480919020394801, "learning_rate": 6.245366276070585e-05, "loss": 2.8325, "step": 23095 }, { "epoch": 1.4337326960084424, "grad_norm": 0.6511371211964895, "learning_rate": 6.245016502249767e-05, "loss": 2.8016, "step": 23096 }, { "epoch": 1.4337947731082004, "grad_norm": 0.6353143521050637, "learning_rate": 6.244666721933396e-05, "loss": 2.7952, "step": 23097 }, { "epoch": 1.4338568502079583, "grad_norm": 0.6785480331209682, "learning_rate": 6.244316935123297e-05, "loss": 2.7788, "step": 23098 }, { "epoch": 1.4339189273077162, "grad_norm": 0.607901993662255, "learning_rate": 6.243967141821295e-05, "loss": 2.811, "step": 23099 }, { "epoch": 1.4339810044074741, "grad_norm": 0.638792347409411, "learning_rate": 6.243617342029215e-05, "loss": 2.7781, "step": 23100 }, { "epoch": 1.434043081507232, "grad_norm": 0.6561989764405266, "learning_rate": 6.243267535748879e-05, "loss": 2.7851, "step": 23101 }, { "epoch": 1.43410515860699, "grad_norm": 0.567449332718806, "learning_rate": 6.242917722982116e-05, "loss": 2.8306, "step": 23102 }, { "epoch": 1.4341672357067479, "grad_norm": 0.6360524910102491, "learning_rate": 6.242567903730749e-05, "loss": 2.812, "step": 23103 }, { "epoch": 1.4342293128065058, "grad_norm": 0.5189528436369132, "learning_rate": 6.242218077996604e-05, "loss": 2.752, "step": 23104 }, { "epoch": 1.4342913899062637, "grad_norm": 0.6411534432631767, "learning_rate": 6.241868245781504e-05, "loss": 2.8023, "step": 23105 }, { "epoch": 1.4343534670060214, "grad_norm": 0.49120432430237054, "learning_rate": 6.241518407087278e-05, "loss": 2.7403, "step": 23106 }, { "epoch": 1.4344155441057793, "grad_norm": 0.5846975558306682, "learning_rate": 6.241168561915749e-05, "loss": 2.7572, "step": 23107 }, { "epoch": 1.4344776212055372, "grad_norm": 0.5282063528751347, "learning_rate": 6.24081871026874e-05, "loss": 2.8658, "step": 23108 }, { "epoch": 1.4345396983052952, "grad_norm": 0.573641896005472, "learning_rate": 6.240468852148083e-05, "loss": 2.8502, "step": 23109 }, { "epoch": 1.434601775405053, "grad_norm": 0.5847695472193319, "learning_rate": 6.240118987555595e-05, "loss": 2.8078, "step": 23110 }, { "epoch": 1.434663852504811, "grad_norm": 0.5565993176576364, "learning_rate": 6.239769116493107e-05, "loss": 2.8059, "step": 23111 }, { "epoch": 1.434725929604569, "grad_norm": 0.5954447451707572, "learning_rate": 6.239419238962441e-05, "loss": 2.7599, "step": 23112 }, { "epoch": 1.4347880067043268, "grad_norm": 0.5579193151875033, "learning_rate": 6.239069354965426e-05, "loss": 2.8955, "step": 23113 }, { "epoch": 1.4348500838040847, "grad_norm": 0.6689114081708045, "learning_rate": 6.238719464503885e-05, "loss": 2.9255, "step": 23114 }, { "epoch": 1.4349121609038424, "grad_norm": 0.52134448400367, "learning_rate": 6.238369567579643e-05, "loss": 2.7953, "step": 23115 }, { "epoch": 1.4349742380036004, "grad_norm": 0.6013380350230787, "learning_rate": 6.238019664194526e-05, "loss": 2.754, "step": 23116 }, { "epoch": 1.4350363151033583, "grad_norm": 0.5887139976735286, "learning_rate": 6.237669754350359e-05, "loss": 2.6872, "step": 23117 }, { "epoch": 1.4350983922031162, "grad_norm": 0.54248917621591, "learning_rate": 6.237319838048971e-05, "loss": 2.7993, "step": 23118 }, { "epoch": 1.435160469302874, "grad_norm": 0.6121013735775866, "learning_rate": 6.236969915292183e-05, "loss": 2.7831, "step": 23119 }, { "epoch": 1.435222546402632, "grad_norm": 0.5482897850278959, "learning_rate": 6.236619986081824e-05, "loss": 2.7372, "step": 23120 }, { "epoch": 1.43528462350239, "grad_norm": 0.5796637548225816, "learning_rate": 6.23627005041972e-05, "loss": 2.7893, "step": 23121 }, { "epoch": 1.4353467006021479, "grad_norm": 0.5634468903819573, "learning_rate": 6.235920108307691e-05, "loss": 2.7745, "step": 23122 }, { "epoch": 1.4354087777019058, "grad_norm": 0.5978177668912261, "learning_rate": 6.23557015974757e-05, "loss": 2.819, "step": 23123 }, { "epoch": 1.4354708548016637, "grad_norm": 0.6093905520288598, "learning_rate": 6.235220204741176e-05, "loss": 2.8748, "step": 23124 }, { "epoch": 1.4355329319014216, "grad_norm": 0.5514528224925755, "learning_rate": 6.234870243290342e-05, "loss": 2.6787, "step": 23125 }, { "epoch": 1.4355950090011795, "grad_norm": 0.600147118016024, "learning_rate": 6.234520275396887e-05, "loss": 2.7585, "step": 23126 }, { "epoch": 1.4356570861009375, "grad_norm": 0.5509073099659226, "learning_rate": 6.234170301062644e-05, "loss": 2.7201, "step": 23127 }, { "epoch": 1.4357191632006954, "grad_norm": 0.589568497052118, "learning_rate": 6.233820320289432e-05, "loss": 2.8124, "step": 23128 }, { "epoch": 1.4357812403004533, "grad_norm": 0.6445616975008266, "learning_rate": 6.233470333079081e-05, "loss": 2.8257, "step": 23129 }, { "epoch": 1.435843317400211, "grad_norm": 0.5768130339614757, "learning_rate": 6.233120339433415e-05, "loss": 2.7621, "step": 23130 }, { "epoch": 1.435905394499969, "grad_norm": 0.5842150484566194, "learning_rate": 6.232770339354262e-05, "loss": 2.7836, "step": 23131 }, { "epoch": 1.4359674715997268, "grad_norm": 0.6147431849985792, "learning_rate": 6.232420332843446e-05, "loss": 2.7468, "step": 23132 }, { "epoch": 1.4360295486994847, "grad_norm": 0.561501384799639, "learning_rate": 6.232070319902792e-05, "loss": 2.8224, "step": 23133 }, { "epoch": 1.4360916257992427, "grad_norm": 0.6603161093472841, "learning_rate": 6.23172030053413e-05, "loss": 2.8399, "step": 23134 }, { "epoch": 1.4361537028990006, "grad_norm": 0.5999536426835135, "learning_rate": 6.231370274739284e-05, "loss": 2.8119, "step": 23135 }, { "epoch": 1.4362157799987585, "grad_norm": 0.5734587330226295, "learning_rate": 6.231020242520082e-05, "loss": 2.6831, "step": 23136 }, { "epoch": 1.4362778570985164, "grad_norm": 0.6133260771289346, "learning_rate": 6.230670203878347e-05, "loss": 2.7635, "step": 23137 }, { "epoch": 1.4363399341982743, "grad_norm": 0.5330934516543195, "learning_rate": 6.230320158815906e-05, "loss": 2.7377, "step": 23138 }, { "epoch": 1.436402011298032, "grad_norm": 0.5838339366238187, "learning_rate": 6.229970107334588e-05, "loss": 2.8113, "step": 23139 }, { "epoch": 1.43646408839779, "grad_norm": 0.5666401873273539, "learning_rate": 6.229620049436216e-05, "loss": 2.7216, "step": 23140 }, { "epoch": 1.4365261654975479, "grad_norm": 0.5713008520751988, "learning_rate": 6.229269985122619e-05, "loss": 2.774, "step": 23141 }, { "epoch": 1.4365882425973058, "grad_norm": 0.49293727381311575, "learning_rate": 6.228919914395621e-05, "loss": 2.7091, "step": 23142 }, { "epoch": 1.4366503196970637, "grad_norm": 0.5602658993117419, "learning_rate": 6.22856983725705e-05, "loss": 2.7486, "step": 23143 }, { "epoch": 1.4367123967968216, "grad_norm": 0.542442808037148, "learning_rate": 6.228219753708731e-05, "loss": 2.9322, "step": 23144 }, { "epoch": 1.4367744738965795, "grad_norm": 0.5715724926675188, "learning_rate": 6.227869663752493e-05, "loss": 2.7201, "step": 23145 }, { "epoch": 1.4368365509963374, "grad_norm": 0.5449365319909424, "learning_rate": 6.227519567390158e-05, "loss": 2.8256, "step": 23146 }, { "epoch": 1.4368986280960954, "grad_norm": 0.6160106545043976, "learning_rate": 6.227169464623558e-05, "loss": 2.845, "step": 23147 }, { "epoch": 1.4369607051958533, "grad_norm": 0.5774569388845103, "learning_rate": 6.226819355454516e-05, "loss": 2.7928, "step": 23148 }, { "epoch": 1.4370227822956112, "grad_norm": 0.6396888545422174, "learning_rate": 6.22646923988486e-05, "loss": 2.7754, "step": 23149 }, { "epoch": 1.4370848593953691, "grad_norm": 0.5978670184914455, "learning_rate": 6.226119117916415e-05, "loss": 2.8462, "step": 23150 }, { "epoch": 1.437146936495127, "grad_norm": 0.564831432034101, "learning_rate": 6.22576898955101e-05, "loss": 2.774, "step": 23151 }, { "epoch": 1.437209013594885, "grad_norm": 0.6522570798028363, "learning_rate": 6.225418854790471e-05, "loss": 2.7833, "step": 23152 }, { "epoch": 1.4372710906946429, "grad_norm": 0.5511091352045151, "learning_rate": 6.225068713636623e-05, "loss": 2.728, "step": 23153 }, { "epoch": 1.4373331677944006, "grad_norm": 0.6338909284786578, "learning_rate": 6.224718566091297e-05, "loss": 2.8291, "step": 23154 }, { "epoch": 1.4373952448941585, "grad_norm": 0.6746913743007252, "learning_rate": 6.224368412156314e-05, "loss": 2.8521, "step": 23155 }, { "epoch": 1.4374573219939164, "grad_norm": 0.5724601024568233, "learning_rate": 6.224018251833505e-05, "loss": 2.7668, "step": 23156 }, { "epoch": 1.4375193990936743, "grad_norm": 0.6020981769205981, "learning_rate": 6.223668085124697e-05, "loss": 2.8179, "step": 23157 }, { "epoch": 1.4375814761934322, "grad_norm": 0.5424180565751429, "learning_rate": 6.223317912031713e-05, "loss": 2.8348, "step": 23158 }, { "epoch": 1.4376435532931902, "grad_norm": 0.5895044222302667, "learning_rate": 6.222967732556384e-05, "loss": 2.8029, "step": 23159 }, { "epoch": 1.437705630392948, "grad_norm": 0.5224520103191025, "learning_rate": 6.222617546700534e-05, "loss": 2.8553, "step": 23160 }, { "epoch": 1.437767707492706, "grad_norm": 0.5674791894365581, "learning_rate": 6.222267354465992e-05, "loss": 2.7945, "step": 23161 }, { "epoch": 1.437829784592464, "grad_norm": 0.5273201548631237, "learning_rate": 6.221917155854585e-05, "loss": 2.8416, "step": 23162 }, { "epoch": 1.4378918616922216, "grad_norm": 0.5533370382068113, "learning_rate": 6.22156695086814e-05, "loss": 2.7809, "step": 23163 }, { "epoch": 1.4379539387919795, "grad_norm": 0.5427086296251578, "learning_rate": 6.221216739508483e-05, "loss": 2.7519, "step": 23164 }, { "epoch": 1.4380160158917374, "grad_norm": 0.562255567847908, "learning_rate": 6.220866521777442e-05, "loss": 2.8383, "step": 23165 }, { "epoch": 1.4380780929914954, "grad_norm": 0.5758823249124166, "learning_rate": 6.220516297676844e-05, "loss": 2.8989, "step": 23166 }, { "epoch": 1.4381401700912533, "grad_norm": 0.5753885407678753, "learning_rate": 6.220166067208515e-05, "loss": 2.7965, "step": 23167 }, { "epoch": 1.4382022471910112, "grad_norm": 0.6271300972223282, "learning_rate": 6.219815830374285e-05, "loss": 2.7542, "step": 23168 }, { "epoch": 1.4382643242907691, "grad_norm": 0.5827680187278971, "learning_rate": 6.219465587175978e-05, "loss": 2.8855, "step": 23169 }, { "epoch": 1.438326401390527, "grad_norm": 0.6504608895319177, "learning_rate": 6.219115337615425e-05, "loss": 2.8551, "step": 23170 }, { "epoch": 1.438388478490285, "grad_norm": 0.6358093601255398, "learning_rate": 6.21876508169445e-05, "loss": 2.8312, "step": 23171 }, { "epoch": 1.4384505555900429, "grad_norm": 0.6262788801329703, "learning_rate": 6.218414819414883e-05, "loss": 2.8921, "step": 23172 }, { "epoch": 1.4385126326898008, "grad_norm": 0.6742976225405852, "learning_rate": 6.21806455077855e-05, "loss": 2.7562, "step": 23173 }, { "epoch": 1.4385747097895587, "grad_norm": 0.5849209932626586, "learning_rate": 6.217714275787278e-05, "loss": 2.7863, "step": 23174 }, { "epoch": 1.4386367868893166, "grad_norm": 0.5995994840421232, "learning_rate": 6.217363994442895e-05, "loss": 2.7922, "step": 23175 }, { "epoch": 1.4386988639890745, "grad_norm": 0.6150166444154337, "learning_rate": 6.217013706747228e-05, "loss": 2.714, "step": 23176 }, { "epoch": 1.4387609410888325, "grad_norm": 0.5620686943809807, "learning_rate": 6.216663412702106e-05, "loss": 2.7559, "step": 23177 }, { "epoch": 1.4388230181885902, "grad_norm": 0.621782364966547, "learning_rate": 6.216313112309354e-05, "loss": 2.7916, "step": 23178 }, { "epoch": 1.438885095288348, "grad_norm": 0.5647747614980723, "learning_rate": 6.215962805570803e-05, "loss": 2.8363, "step": 23179 }, { "epoch": 1.438947172388106, "grad_norm": 0.6215530712396421, "learning_rate": 6.215612492488278e-05, "loss": 2.8195, "step": 23180 }, { "epoch": 1.439009249487864, "grad_norm": 0.5651263014388799, "learning_rate": 6.215262173063607e-05, "loss": 2.8087, "step": 23181 }, { "epoch": 1.4390713265876218, "grad_norm": 0.5665925081431008, "learning_rate": 6.214911847298619e-05, "loss": 2.7698, "step": 23182 }, { "epoch": 1.4391334036873797, "grad_norm": 0.661267400703724, "learning_rate": 6.214561515195141e-05, "loss": 2.9096, "step": 23183 }, { "epoch": 1.4391954807871377, "grad_norm": 0.5456244475459113, "learning_rate": 6.214211176755001e-05, "loss": 2.7606, "step": 23184 }, { "epoch": 1.4392575578868956, "grad_norm": 0.6010932826374281, "learning_rate": 6.213860831980025e-05, "loss": 2.8175, "step": 23185 }, { "epoch": 1.4393196349866535, "grad_norm": 0.5620371185581754, "learning_rate": 6.213510480872045e-05, "loss": 2.7392, "step": 23186 }, { "epoch": 1.4393817120864112, "grad_norm": 0.5764791412881051, "learning_rate": 6.213160123432884e-05, "loss": 2.9314, "step": 23187 }, { "epoch": 1.439443789186169, "grad_norm": 0.6168055499326255, "learning_rate": 6.212809759664372e-05, "loss": 2.8584, "step": 23188 }, { "epoch": 1.439505866285927, "grad_norm": 0.5544940181216892, "learning_rate": 6.212459389568338e-05, "loss": 2.8484, "step": 23189 }, { "epoch": 1.439567943385685, "grad_norm": 0.5418124063984364, "learning_rate": 6.21210901314661e-05, "loss": 2.8112, "step": 23190 }, { "epoch": 1.4396300204854429, "grad_norm": 0.5583371460011057, "learning_rate": 6.211758630401013e-05, "loss": 2.8336, "step": 23191 }, { "epoch": 1.4396920975852008, "grad_norm": 0.5621678908417974, "learning_rate": 6.211408241333379e-05, "loss": 2.8254, "step": 23192 }, { "epoch": 1.4397541746849587, "grad_norm": 0.612872309316867, "learning_rate": 6.211057845945534e-05, "loss": 2.862, "step": 23193 }, { "epoch": 1.4398162517847166, "grad_norm": 0.5295313635966972, "learning_rate": 6.210707444239304e-05, "loss": 2.8824, "step": 23194 }, { "epoch": 1.4398783288844745, "grad_norm": 0.55820256654299, "learning_rate": 6.210357036216522e-05, "loss": 2.6906, "step": 23195 }, { "epoch": 1.4399404059842325, "grad_norm": 0.5766798205014733, "learning_rate": 6.210006621879011e-05, "loss": 2.8368, "step": 23196 }, { "epoch": 1.4400024830839904, "grad_norm": 0.5560775318015571, "learning_rate": 6.209656201228604e-05, "loss": 2.7546, "step": 23197 }, { "epoch": 1.4400645601837483, "grad_norm": 0.5937851092348312, "learning_rate": 6.209305774267126e-05, "loss": 2.9002, "step": 23198 }, { "epoch": 1.4401266372835062, "grad_norm": 0.5447008511892566, "learning_rate": 6.208955340996405e-05, "loss": 2.7863, "step": 23199 }, { "epoch": 1.4401887143832641, "grad_norm": 0.5747331922827127, "learning_rate": 6.208604901418273e-05, "loss": 2.7898, "step": 23200 }, { "epoch": 1.440250791483022, "grad_norm": 0.6021045373088975, "learning_rate": 6.208254455534552e-05, "loss": 2.7624, "step": 23201 }, { "epoch": 1.4403128685827797, "grad_norm": 0.6148387787556135, "learning_rate": 6.207904003347079e-05, "loss": 2.7848, "step": 23202 }, { "epoch": 1.4403749456825377, "grad_norm": 0.5868256005569756, "learning_rate": 6.207553544857674e-05, "loss": 2.8542, "step": 23203 }, { "epoch": 1.4404370227822956, "grad_norm": 0.6260504240120642, "learning_rate": 6.20720308006817e-05, "loss": 2.8627, "step": 23204 }, { "epoch": 1.4404990998820535, "grad_norm": 0.566593586426027, "learning_rate": 6.206852608980395e-05, "loss": 2.7519, "step": 23205 }, { "epoch": 1.4405611769818114, "grad_norm": 0.5767932450791128, "learning_rate": 6.206502131596177e-05, "loss": 2.8373, "step": 23206 }, { "epoch": 1.4406232540815693, "grad_norm": 0.6007780930692871, "learning_rate": 6.206151647917344e-05, "loss": 2.7513, "step": 23207 }, { "epoch": 1.4406853311813272, "grad_norm": 0.5616871205472189, "learning_rate": 6.205801157945725e-05, "loss": 2.8103, "step": 23208 }, { "epoch": 1.4407474082810852, "grad_norm": 0.6138858851790363, "learning_rate": 6.205450661683149e-05, "loss": 2.8718, "step": 23209 }, { "epoch": 1.440809485380843, "grad_norm": 0.556022997409687, "learning_rate": 6.205100159131443e-05, "loss": 2.8428, "step": 23210 }, { "epoch": 1.4408715624806008, "grad_norm": 0.6704985469330156, "learning_rate": 6.20474965029244e-05, "loss": 2.8607, "step": 23211 }, { "epoch": 1.4409336395803587, "grad_norm": 0.5479450112131833, "learning_rate": 6.204399135167964e-05, "loss": 2.825, "step": 23212 }, { "epoch": 1.4409957166801166, "grad_norm": 0.6308457222515268, "learning_rate": 6.204048613759845e-05, "loss": 2.7239, "step": 23213 }, { "epoch": 1.4410577937798745, "grad_norm": 0.49839212755193874, "learning_rate": 6.203698086069911e-05, "loss": 2.755, "step": 23214 }, { "epoch": 1.4411198708796324, "grad_norm": 0.5907726930706342, "learning_rate": 6.203347552099993e-05, "loss": 2.7726, "step": 23215 }, { "epoch": 1.4411819479793904, "grad_norm": 0.5379593383515335, "learning_rate": 6.202997011851918e-05, "loss": 2.8283, "step": 23216 }, { "epoch": 1.4412440250791483, "grad_norm": 0.5070288847646224, "learning_rate": 6.202646465327516e-05, "loss": 2.6602, "step": 23217 }, { "epoch": 1.4413061021789062, "grad_norm": 0.5520276515399639, "learning_rate": 6.202295912528616e-05, "loss": 2.7857, "step": 23218 }, { "epoch": 1.4413681792786641, "grad_norm": 0.4918524343362689, "learning_rate": 6.201945353457046e-05, "loss": 2.7317, "step": 23219 }, { "epoch": 1.441430256378422, "grad_norm": 0.6205625601760877, "learning_rate": 6.201594788114636e-05, "loss": 2.721, "step": 23220 }, { "epoch": 1.44149233347818, "grad_norm": 0.5269763373033625, "learning_rate": 6.201244216503213e-05, "loss": 2.8311, "step": 23221 }, { "epoch": 1.4415544105779379, "grad_norm": 0.5804114876968153, "learning_rate": 6.200893638624608e-05, "loss": 2.7042, "step": 23222 }, { "epoch": 1.4416164876776958, "grad_norm": 0.5775635286785799, "learning_rate": 6.20054305448065e-05, "loss": 2.8331, "step": 23223 }, { "epoch": 1.4416785647774537, "grad_norm": 0.6593620217659637, "learning_rate": 6.200192464073166e-05, "loss": 2.7838, "step": 23224 }, { "epoch": 1.4417406418772116, "grad_norm": 0.5636351046660002, "learning_rate": 6.199841867403987e-05, "loss": 2.7711, "step": 23225 }, { "epoch": 1.4418027189769693, "grad_norm": 0.6236671779971301, "learning_rate": 6.199491264474942e-05, "loss": 2.7254, "step": 23226 }, { "epoch": 1.4418647960767272, "grad_norm": 0.6115709498956655, "learning_rate": 6.199140655287861e-05, "loss": 2.8216, "step": 23227 }, { "epoch": 1.4419268731764852, "grad_norm": 0.6156851676158661, "learning_rate": 6.19879003984457e-05, "loss": 2.6451, "step": 23228 }, { "epoch": 1.441988950276243, "grad_norm": 0.5484741807283723, "learning_rate": 6.198439418146902e-05, "loss": 2.7943, "step": 23229 }, { "epoch": 1.442051027376001, "grad_norm": 0.6278759828089336, "learning_rate": 6.198088790196683e-05, "loss": 2.851, "step": 23230 }, { "epoch": 1.442113104475759, "grad_norm": 0.5891977018422215, "learning_rate": 6.197738155995746e-05, "loss": 2.7798, "step": 23231 }, { "epoch": 1.4421751815755168, "grad_norm": 0.6666837903644895, "learning_rate": 6.197387515545917e-05, "loss": 2.8256, "step": 23232 }, { "epoch": 1.4422372586752747, "grad_norm": 0.6367267303902803, "learning_rate": 6.197036868849026e-05, "loss": 2.8601, "step": 23233 }, { "epoch": 1.4422993357750327, "grad_norm": 0.541791201605105, "learning_rate": 6.196686215906904e-05, "loss": 2.7709, "step": 23234 }, { "epoch": 1.4423614128747904, "grad_norm": 0.5285202184056008, "learning_rate": 6.19633555672138e-05, "loss": 2.7792, "step": 23235 }, { "epoch": 1.4424234899745483, "grad_norm": 0.5494637389224379, "learning_rate": 6.195984891294282e-05, "loss": 2.7795, "step": 23236 }, { "epoch": 1.4424855670743062, "grad_norm": 0.5599785558818171, "learning_rate": 6.195634219627441e-05, "loss": 2.7967, "step": 23237 }, { "epoch": 1.4425476441740641, "grad_norm": 0.5556095143446226, "learning_rate": 6.195283541722686e-05, "loss": 2.775, "step": 23238 }, { "epoch": 1.442609721273822, "grad_norm": 0.6105837937100754, "learning_rate": 6.194932857581845e-05, "loss": 2.7888, "step": 23239 }, { "epoch": 1.44267179837358, "grad_norm": 0.5154980878988744, "learning_rate": 6.194582167206752e-05, "loss": 2.7556, "step": 23240 }, { "epoch": 1.4427338754733379, "grad_norm": 0.5853497927425438, "learning_rate": 6.194231470599232e-05, "loss": 2.7883, "step": 23241 }, { "epoch": 1.4427959525730958, "grad_norm": 0.5159075280858278, "learning_rate": 6.193880767761118e-05, "loss": 2.8784, "step": 23242 }, { "epoch": 1.4428580296728537, "grad_norm": 0.5274356055616125, "learning_rate": 6.193530058694236e-05, "loss": 2.8106, "step": 23243 }, { "epoch": 1.4429201067726116, "grad_norm": 0.5692934311560659, "learning_rate": 6.193179343400421e-05, "loss": 2.9091, "step": 23244 }, { "epoch": 1.4429821838723695, "grad_norm": 0.5295189535875033, "learning_rate": 6.192828621881497e-05, "loss": 2.8023, "step": 23245 }, { "epoch": 1.4430442609721275, "grad_norm": 0.5455568503940459, "learning_rate": 6.192477894139297e-05, "loss": 2.7677, "step": 23246 }, { "epoch": 1.4431063380718854, "grad_norm": 0.5896574319795992, "learning_rate": 6.19212716017565e-05, "loss": 2.8671, "step": 23247 }, { "epoch": 1.4431684151716433, "grad_norm": 0.49896753524333287, "learning_rate": 6.191776419992387e-05, "loss": 2.8195, "step": 23248 }, { "epoch": 1.4432304922714012, "grad_norm": 0.5857650757452033, "learning_rate": 6.191425673591338e-05, "loss": 2.7108, "step": 23249 }, { "epoch": 1.443292569371159, "grad_norm": 0.48726329742360125, "learning_rate": 6.19107492097433e-05, "loss": 2.7933, "step": 23250 }, { "epoch": 1.4433546464709168, "grad_norm": 0.6032456717069098, "learning_rate": 6.190724162143197e-05, "loss": 2.7636, "step": 23251 }, { "epoch": 1.4434167235706747, "grad_norm": 0.49168386880501885, "learning_rate": 6.190373397099765e-05, "loss": 2.7099, "step": 23252 }, { "epoch": 1.4434788006704327, "grad_norm": 0.6115471200856002, "learning_rate": 6.190022625845868e-05, "loss": 2.8018, "step": 23253 }, { "epoch": 1.4435408777701906, "grad_norm": 0.5485048258177064, "learning_rate": 6.189671848383333e-05, "loss": 2.7944, "step": 23254 }, { "epoch": 1.4436029548699485, "grad_norm": 0.6504008203572265, "learning_rate": 6.189321064713991e-05, "loss": 2.7851, "step": 23255 }, { "epoch": 1.4436650319697064, "grad_norm": 0.5985349540940397, "learning_rate": 6.188970274839674e-05, "loss": 2.706, "step": 23256 }, { "epoch": 1.4437271090694643, "grad_norm": 0.63376270586177, "learning_rate": 6.188619478762208e-05, "loss": 2.8508, "step": 23257 }, { "epoch": 1.4437891861692222, "grad_norm": 0.6675230178721859, "learning_rate": 6.188268676483428e-05, "loss": 2.8052, "step": 23258 }, { "epoch": 1.44385126326898, "grad_norm": 0.5822367721322772, "learning_rate": 6.187917868005158e-05, "loss": 2.6994, "step": 23259 }, { "epoch": 1.4439133403687379, "grad_norm": 0.6388658934638115, "learning_rate": 6.187567053329236e-05, "loss": 2.8379, "step": 23260 }, { "epoch": 1.4439754174684958, "grad_norm": 0.5894401634818678, "learning_rate": 6.187216232457486e-05, "loss": 2.8222, "step": 23261 }, { "epoch": 1.4440374945682537, "grad_norm": 0.7302988028068189, "learning_rate": 6.186865405391743e-05, "loss": 2.828, "step": 23262 }, { "epoch": 1.4440995716680116, "grad_norm": 0.6676806893976621, "learning_rate": 6.186514572133833e-05, "loss": 2.7919, "step": 23263 }, { "epoch": 1.4441616487677695, "grad_norm": 0.575468299735041, "learning_rate": 6.186163732685589e-05, "loss": 2.7525, "step": 23264 }, { "epoch": 1.4442237258675275, "grad_norm": 0.6437376743358858, "learning_rate": 6.185812887048842e-05, "loss": 2.7891, "step": 23265 }, { "epoch": 1.4442858029672854, "grad_norm": 0.5752910281789515, "learning_rate": 6.185462035225419e-05, "loss": 2.7969, "step": 23266 }, { "epoch": 1.4443478800670433, "grad_norm": 0.6003164043230046, "learning_rate": 6.185111177217156e-05, "loss": 2.8304, "step": 23267 }, { "epoch": 1.4444099571668012, "grad_norm": 0.5930247897278542, "learning_rate": 6.184760313025877e-05, "loss": 2.744, "step": 23268 }, { "epoch": 1.4444720342665591, "grad_norm": 0.5855602066934861, "learning_rate": 6.184409442653418e-05, "loss": 2.7752, "step": 23269 }, { "epoch": 1.444534111366317, "grad_norm": 0.5179681711783418, "learning_rate": 6.184058566101605e-05, "loss": 2.8273, "step": 23270 }, { "epoch": 1.444596188466075, "grad_norm": 0.6664876300095978, "learning_rate": 6.183707683372274e-05, "loss": 2.7897, "step": 23271 }, { "epoch": 1.4446582655658329, "grad_norm": 0.5799086095070513, "learning_rate": 6.183356794467251e-05, "loss": 2.7734, "step": 23272 }, { "epoch": 1.4447203426655908, "grad_norm": 0.6163737811879331, "learning_rate": 6.18300589938837e-05, "loss": 2.7377, "step": 23273 }, { "epoch": 1.4447824197653485, "grad_norm": 0.5691509960646075, "learning_rate": 6.182654998137458e-05, "loss": 2.7686, "step": 23274 }, { "epoch": 1.4448444968651064, "grad_norm": 0.6315804013410153, "learning_rate": 6.182304090716349e-05, "loss": 2.7994, "step": 23275 }, { "epoch": 1.4449065739648643, "grad_norm": 0.6115037221054497, "learning_rate": 6.181953177126873e-05, "loss": 2.7873, "step": 23276 }, { "epoch": 1.4449686510646222, "grad_norm": 0.6198443195406752, "learning_rate": 6.181602257370857e-05, "loss": 2.7691, "step": 23277 }, { "epoch": 1.4450307281643802, "grad_norm": 0.6538047192140377, "learning_rate": 6.18125133145014e-05, "loss": 2.9135, "step": 23278 }, { "epoch": 1.445092805264138, "grad_norm": 0.5623340626841716, "learning_rate": 6.180900399366544e-05, "loss": 2.7556, "step": 23279 }, { "epoch": 1.445154882363896, "grad_norm": 0.6079711393232412, "learning_rate": 6.180549461121906e-05, "loss": 2.7316, "step": 23280 }, { "epoch": 1.445216959463654, "grad_norm": 0.6422646636268948, "learning_rate": 6.180198516718054e-05, "loss": 2.8731, "step": 23281 }, { "epoch": 1.4452790365634118, "grad_norm": 0.5929260308881557, "learning_rate": 6.17984756615682e-05, "loss": 2.8352, "step": 23282 }, { "epoch": 1.4453411136631695, "grad_norm": 0.6048607757860381, "learning_rate": 6.179496609440036e-05, "loss": 2.8088, "step": 23283 }, { "epoch": 1.4454031907629274, "grad_norm": 0.5394304374327075, "learning_rate": 6.179145646569531e-05, "loss": 2.7965, "step": 23284 }, { "epoch": 1.4454652678626854, "grad_norm": 0.5578394512002117, "learning_rate": 6.178794677547137e-05, "loss": 2.7905, "step": 23285 }, { "epoch": 1.4455273449624433, "grad_norm": 0.5661901127386189, "learning_rate": 6.178443702374684e-05, "loss": 2.744, "step": 23286 }, { "epoch": 1.4455894220622012, "grad_norm": 0.603506983229113, "learning_rate": 6.178092721054006e-05, "loss": 2.7777, "step": 23287 }, { "epoch": 1.4456514991619591, "grad_norm": 0.5941805282525634, "learning_rate": 6.17774173358693e-05, "loss": 2.7878, "step": 23288 }, { "epoch": 1.445713576261717, "grad_norm": 0.6406923584300306, "learning_rate": 6.177390739975292e-05, "loss": 2.9069, "step": 23289 }, { "epoch": 1.445775653361475, "grad_norm": 0.5701891883374346, "learning_rate": 6.177039740220919e-05, "loss": 2.8002, "step": 23290 }, { "epoch": 1.4458377304612329, "grad_norm": 0.6485226127193824, "learning_rate": 6.176688734325644e-05, "loss": 2.7913, "step": 23291 }, { "epoch": 1.4458998075609908, "grad_norm": 0.5482539638753461, "learning_rate": 6.176337722291299e-05, "loss": 2.8124, "step": 23292 }, { "epoch": 1.4459618846607487, "grad_norm": 0.6946934505247261, "learning_rate": 6.175986704119714e-05, "loss": 2.7834, "step": 23293 }, { "epoch": 1.4460239617605066, "grad_norm": 0.5111813526187369, "learning_rate": 6.175635679812723e-05, "loss": 2.7063, "step": 23294 }, { "epoch": 1.4460860388602645, "grad_norm": 0.6021678405466612, "learning_rate": 6.175284649372151e-05, "loss": 2.816, "step": 23295 }, { "epoch": 1.4461481159600225, "grad_norm": 0.5520962377755484, "learning_rate": 6.174933612799837e-05, "loss": 2.7818, "step": 23296 }, { "epoch": 1.4462101930597804, "grad_norm": 0.5767787775814539, "learning_rate": 6.174582570097608e-05, "loss": 2.8038, "step": 23297 }, { "epoch": 1.446272270159538, "grad_norm": 0.48003724733515923, "learning_rate": 6.174231521267296e-05, "loss": 2.8247, "step": 23298 }, { "epoch": 1.446334347259296, "grad_norm": 0.5197303791902323, "learning_rate": 6.173880466310734e-05, "loss": 2.7846, "step": 23299 }, { "epoch": 1.446396424359054, "grad_norm": 0.5210864088007469, "learning_rate": 6.173529405229752e-05, "loss": 2.848, "step": 23300 }, { "epoch": 1.4464585014588118, "grad_norm": 0.5315516681229087, "learning_rate": 6.173178338026183e-05, "loss": 2.7514, "step": 23301 }, { "epoch": 1.4465205785585697, "grad_norm": 0.5204926680031066, "learning_rate": 6.172827264701857e-05, "loss": 2.7319, "step": 23302 }, { "epoch": 1.4465826556583277, "grad_norm": 0.5745006310564896, "learning_rate": 6.172476185258607e-05, "loss": 2.8288, "step": 23303 }, { "epoch": 1.4466447327580856, "grad_norm": 0.5514456076667494, "learning_rate": 6.172125099698265e-05, "loss": 2.7485, "step": 23304 }, { "epoch": 1.4467068098578435, "grad_norm": 0.5712223633522969, "learning_rate": 6.17177400802266e-05, "loss": 2.7804, "step": 23305 }, { "epoch": 1.4467688869576014, "grad_norm": 0.5356720158385645, "learning_rate": 6.171422910233626e-05, "loss": 2.777, "step": 23306 }, { "epoch": 1.4468309640573591, "grad_norm": 0.6602116172189637, "learning_rate": 6.171071806332995e-05, "loss": 2.9569, "step": 23307 }, { "epoch": 1.446893041157117, "grad_norm": 0.5606529587373235, "learning_rate": 6.170720696322599e-05, "loss": 2.7499, "step": 23308 }, { "epoch": 1.446955118256875, "grad_norm": 0.5562729809332194, "learning_rate": 6.170369580204267e-05, "loss": 2.8281, "step": 23309 }, { "epoch": 1.4470171953566329, "grad_norm": 0.5530779414444577, "learning_rate": 6.170018457979834e-05, "loss": 2.7881, "step": 23310 }, { "epoch": 1.4470792724563908, "grad_norm": 0.6015572485756479, "learning_rate": 6.16966732965113e-05, "loss": 2.8023, "step": 23311 }, { "epoch": 1.4471413495561487, "grad_norm": 0.5503454174917363, "learning_rate": 6.169316195219988e-05, "loss": 2.7802, "step": 23312 }, { "epoch": 1.4472034266559066, "grad_norm": 0.5453046049404551, "learning_rate": 6.16896505468824e-05, "loss": 2.7241, "step": 23313 }, { "epoch": 1.4472655037556645, "grad_norm": 0.5822757489025365, "learning_rate": 6.168613908057715e-05, "loss": 2.8451, "step": 23314 }, { "epoch": 1.4473275808554225, "grad_norm": 0.5173835244611289, "learning_rate": 6.168262755330249e-05, "loss": 2.7317, "step": 23315 }, { "epoch": 1.4473896579551804, "grad_norm": 0.6031782571473583, "learning_rate": 6.167911596507674e-05, "loss": 2.7595, "step": 23316 }, { "epoch": 1.4474517350549383, "grad_norm": 0.5310068134869301, "learning_rate": 6.16756043159182e-05, "loss": 2.7733, "step": 23317 }, { "epoch": 1.4475138121546962, "grad_norm": 0.5937041716708078, "learning_rate": 6.16720926058452e-05, "loss": 2.8303, "step": 23318 }, { "epoch": 1.4475758892544541, "grad_norm": 0.5576832960928607, "learning_rate": 6.166858083487604e-05, "loss": 2.7572, "step": 23319 }, { "epoch": 1.447637966354212, "grad_norm": 0.5141996276137474, "learning_rate": 6.166506900302908e-05, "loss": 2.8216, "step": 23320 }, { "epoch": 1.44770004345397, "grad_norm": 0.5962621598433294, "learning_rate": 6.166155711032261e-05, "loss": 2.8416, "step": 23321 }, { "epoch": 1.4477621205537277, "grad_norm": 0.5504268290495778, "learning_rate": 6.165804515677498e-05, "loss": 2.7757, "step": 23322 }, { "epoch": 1.4478241976534856, "grad_norm": 0.5832931868044242, "learning_rate": 6.16545331424045e-05, "loss": 2.7721, "step": 23323 }, { "epoch": 1.4478862747532435, "grad_norm": 0.6097062262196695, "learning_rate": 6.165102106722947e-05, "loss": 2.7572, "step": 23324 }, { "epoch": 1.4479483518530014, "grad_norm": 0.60705255890786, "learning_rate": 6.164750893126826e-05, "loss": 2.8129, "step": 23325 }, { "epoch": 1.4480104289527593, "grad_norm": 0.6843513867907458, "learning_rate": 6.164399673453915e-05, "loss": 2.7111, "step": 23326 }, { "epoch": 1.4480725060525172, "grad_norm": 0.6512672313306945, "learning_rate": 6.164048447706049e-05, "loss": 2.8251, "step": 23327 }, { "epoch": 1.4481345831522752, "grad_norm": 0.6517137237625724, "learning_rate": 6.16369721588506e-05, "loss": 2.6874, "step": 23328 }, { "epoch": 1.448196660252033, "grad_norm": 0.6854082277081862, "learning_rate": 6.163345977992779e-05, "loss": 2.9313, "step": 23329 }, { "epoch": 1.448258737351791, "grad_norm": 0.7868850804926256, "learning_rate": 6.162994734031041e-05, "loss": 2.7705, "step": 23330 }, { "epoch": 1.4483208144515487, "grad_norm": 0.5943902729531064, "learning_rate": 6.162643484001676e-05, "loss": 2.7764, "step": 23331 }, { "epoch": 1.4483828915513066, "grad_norm": 0.7261923390542525, "learning_rate": 6.162292227906517e-05, "loss": 2.7359, "step": 23332 }, { "epoch": 1.4484449686510645, "grad_norm": 0.6067637714635381, "learning_rate": 6.1619409657474e-05, "loss": 2.8647, "step": 23333 }, { "epoch": 1.4485070457508225, "grad_norm": 0.6888915846906063, "learning_rate": 6.161589697526153e-05, "loss": 2.8166, "step": 23334 }, { "epoch": 1.4485691228505804, "grad_norm": 0.6156270545111905, "learning_rate": 6.16123842324461e-05, "loss": 2.8337, "step": 23335 }, { "epoch": 1.4486311999503383, "grad_norm": 0.7018681206709987, "learning_rate": 6.160887142904605e-05, "loss": 2.8872, "step": 23336 }, { "epoch": 1.4486932770500962, "grad_norm": 0.5892159360799714, "learning_rate": 6.160535856507972e-05, "loss": 2.8241, "step": 23337 }, { "epoch": 1.4487553541498541, "grad_norm": 0.6705465268371169, "learning_rate": 6.16018456405654e-05, "loss": 2.8472, "step": 23338 }, { "epoch": 1.448817431249612, "grad_norm": 0.5571366626356712, "learning_rate": 6.159833265552143e-05, "loss": 2.8262, "step": 23339 }, { "epoch": 1.44887950834937, "grad_norm": 0.6191424781940055, "learning_rate": 6.159481960996614e-05, "loss": 2.7409, "step": 23340 }, { "epoch": 1.4489415854491279, "grad_norm": 0.5998587571234548, "learning_rate": 6.159130650391787e-05, "loss": 2.8237, "step": 23341 }, { "epoch": 1.4490036625488858, "grad_norm": 0.5823053828098576, "learning_rate": 6.158779333739493e-05, "loss": 2.8379, "step": 23342 }, { "epoch": 1.4490657396486437, "grad_norm": 0.6580838058490593, "learning_rate": 6.158428011041568e-05, "loss": 2.8422, "step": 23343 }, { "epoch": 1.4491278167484016, "grad_norm": 0.6206139924839391, "learning_rate": 6.15807668229984e-05, "loss": 2.7814, "step": 23344 }, { "epoch": 1.4491898938481595, "grad_norm": 0.6437092944117925, "learning_rate": 6.157725347516148e-05, "loss": 2.8516, "step": 23345 }, { "epoch": 1.4492519709479172, "grad_norm": 0.6917021813141369, "learning_rate": 6.157374006692319e-05, "loss": 2.8187, "step": 23346 }, { "epoch": 1.4493140480476752, "grad_norm": 0.6111585973101676, "learning_rate": 6.15702265983019e-05, "loss": 2.7848, "step": 23347 }, { "epoch": 1.449376125147433, "grad_norm": 0.6658917085253109, "learning_rate": 6.156671306931593e-05, "loss": 2.7506, "step": 23348 }, { "epoch": 1.449438202247191, "grad_norm": 0.6687405015146104, "learning_rate": 6.15631994799836e-05, "loss": 2.844, "step": 23349 }, { "epoch": 1.449500279346949, "grad_norm": 0.6350502901738421, "learning_rate": 6.155968583032326e-05, "loss": 2.7376, "step": 23350 }, { "epoch": 1.4495623564467068, "grad_norm": 0.6925119728816109, "learning_rate": 6.155617212035324e-05, "loss": 2.8386, "step": 23351 }, { "epoch": 1.4496244335464648, "grad_norm": 0.65469503102489, "learning_rate": 6.155265835009186e-05, "loss": 2.781, "step": 23352 }, { "epoch": 1.4496865106462227, "grad_norm": 0.6545927602824475, "learning_rate": 6.154914451955746e-05, "loss": 2.8374, "step": 23353 }, { "epoch": 1.4497485877459806, "grad_norm": 0.6555763374575364, "learning_rate": 6.154563062876836e-05, "loss": 2.8447, "step": 23354 }, { "epoch": 1.4498106648457383, "grad_norm": 0.600247146448727, "learning_rate": 6.154211667774291e-05, "loss": 2.7289, "step": 23355 }, { "epoch": 1.4498727419454962, "grad_norm": 0.6998554506237203, "learning_rate": 6.153860266649942e-05, "loss": 2.7729, "step": 23356 }, { "epoch": 1.4499348190452541, "grad_norm": 0.6321942877110441, "learning_rate": 6.153508859505627e-05, "loss": 2.8325, "step": 23357 }, { "epoch": 1.449996896145012, "grad_norm": 0.6005820669654244, "learning_rate": 6.153157446343173e-05, "loss": 2.8728, "step": 23358 }, { "epoch": 1.45005897324477, "grad_norm": 0.6350628670412723, "learning_rate": 6.152806027164419e-05, "loss": 2.8099, "step": 23359 }, { "epoch": 1.4501210503445279, "grad_norm": 0.5333839082083505, "learning_rate": 6.152454601971194e-05, "loss": 2.8099, "step": 23360 }, { "epoch": 1.4501831274442858, "grad_norm": 0.6501909470690931, "learning_rate": 6.152103170765336e-05, "loss": 2.8187, "step": 23361 }, { "epoch": 1.4502452045440437, "grad_norm": 0.6124941154119449, "learning_rate": 6.151751733548673e-05, "loss": 2.8134, "step": 23362 }, { "epoch": 1.4503072816438016, "grad_norm": 0.6657371498154119, "learning_rate": 6.151400290323043e-05, "loss": 2.6749, "step": 23363 }, { "epoch": 1.4503693587435595, "grad_norm": 0.5841049529149234, "learning_rate": 6.151048841090278e-05, "loss": 2.7645, "step": 23364 }, { "epoch": 1.4504314358433175, "grad_norm": 0.6564223653499075, "learning_rate": 6.150697385852213e-05, "loss": 2.6775, "step": 23365 }, { "epoch": 1.4504935129430754, "grad_norm": 0.6038860184413035, "learning_rate": 6.15034592461068e-05, "loss": 2.8211, "step": 23366 }, { "epoch": 1.4505555900428333, "grad_norm": 0.6390767578549102, "learning_rate": 6.149994457367511e-05, "loss": 2.8542, "step": 23367 }, { "epoch": 1.4506176671425912, "grad_norm": 0.5937154910449254, "learning_rate": 6.149642984124545e-05, "loss": 2.8211, "step": 23368 }, { "epoch": 1.4506797442423491, "grad_norm": 0.7297019401608145, "learning_rate": 6.149291504883611e-05, "loss": 2.7921, "step": 23369 }, { "epoch": 1.4507418213421068, "grad_norm": 0.584021040872169, "learning_rate": 6.148940019646544e-05, "loss": 2.8162, "step": 23370 }, { "epoch": 1.4508038984418647, "grad_norm": 0.6789827310307187, "learning_rate": 6.148588528415178e-05, "loss": 2.7848, "step": 23371 }, { "epoch": 1.4508659755416227, "grad_norm": 0.6327708430766593, "learning_rate": 6.148237031191348e-05, "loss": 2.8051, "step": 23372 }, { "epoch": 1.4509280526413806, "grad_norm": 0.6159436209813364, "learning_rate": 6.147885527976885e-05, "loss": 2.7713, "step": 23373 }, { "epoch": 1.4509901297411385, "grad_norm": 0.6508941076267796, "learning_rate": 6.147534018773625e-05, "loss": 2.779, "step": 23374 }, { "epoch": 1.4510522068408964, "grad_norm": 0.6155278066215487, "learning_rate": 6.147182503583403e-05, "loss": 2.7599, "step": 23375 }, { "epoch": 1.4511142839406543, "grad_norm": 0.6839198397392413, "learning_rate": 6.14683098240805e-05, "loss": 2.7431, "step": 23376 }, { "epoch": 1.4511763610404123, "grad_norm": 0.5841692118469095, "learning_rate": 6.146479455249401e-05, "loss": 2.7563, "step": 23377 }, { "epoch": 1.4512384381401702, "grad_norm": 0.6063032244885203, "learning_rate": 6.14612792210929e-05, "loss": 2.7632, "step": 23378 }, { "epoch": 1.4513005152399279, "grad_norm": 0.5936790223111202, "learning_rate": 6.145776382989552e-05, "loss": 2.8666, "step": 23379 }, { "epoch": 1.4513625923396858, "grad_norm": 0.6466972715060411, "learning_rate": 6.145424837892021e-05, "loss": 2.8361, "step": 23380 }, { "epoch": 1.4514246694394437, "grad_norm": 0.5311583832672527, "learning_rate": 6.145073286818529e-05, "loss": 2.7565, "step": 23381 }, { "epoch": 1.4514867465392016, "grad_norm": 0.6452262368169921, "learning_rate": 6.144721729770913e-05, "loss": 2.7559, "step": 23382 }, { "epoch": 1.4515488236389595, "grad_norm": 0.5470783672945075, "learning_rate": 6.144370166751005e-05, "loss": 2.8555, "step": 23383 }, { "epoch": 1.4516109007387175, "grad_norm": 0.638128510296439, "learning_rate": 6.144018597760642e-05, "loss": 2.6841, "step": 23384 }, { "epoch": 1.4516729778384754, "grad_norm": 0.5581292575361954, "learning_rate": 6.143667022801655e-05, "loss": 2.7682, "step": 23385 }, { "epoch": 1.4517350549382333, "grad_norm": 0.5810674329369607, "learning_rate": 6.143315441875879e-05, "loss": 2.7292, "step": 23386 }, { "epoch": 1.4517971320379912, "grad_norm": 0.5071475443728958, "learning_rate": 6.14296385498515e-05, "loss": 2.7623, "step": 23387 }, { "epoch": 1.4518592091377491, "grad_norm": 0.5480227007601758, "learning_rate": 6.1426122621313e-05, "loss": 2.7993, "step": 23388 }, { "epoch": 1.451921286237507, "grad_norm": 0.5733180605116456, "learning_rate": 6.142260663316165e-05, "loss": 2.7051, "step": 23389 }, { "epoch": 1.451983363337265, "grad_norm": 0.5531376691757419, "learning_rate": 6.141909058541579e-05, "loss": 2.7917, "step": 23390 }, { "epoch": 1.4520454404370229, "grad_norm": 0.6061415103308702, "learning_rate": 6.141557447809377e-05, "loss": 2.8253, "step": 23391 }, { "epoch": 1.4521075175367808, "grad_norm": 0.48320931482320917, "learning_rate": 6.14120583112139e-05, "loss": 2.7909, "step": 23392 }, { "epoch": 1.4521695946365387, "grad_norm": 0.6579562440153393, "learning_rate": 6.140854208479458e-05, "loss": 2.8722, "step": 23393 }, { "epoch": 1.4522316717362964, "grad_norm": 0.5220957968528968, "learning_rate": 6.14050257988541e-05, "loss": 2.8337, "step": 23394 }, { "epoch": 1.4522937488360543, "grad_norm": 0.6269636064170221, "learning_rate": 6.140150945341084e-05, "loss": 2.8887, "step": 23395 }, { "epoch": 1.4523558259358123, "grad_norm": 0.5156977053159413, "learning_rate": 6.139799304848315e-05, "loss": 2.8175, "step": 23396 }, { "epoch": 1.4524179030355702, "grad_norm": 0.5706463972928699, "learning_rate": 6.139447658408936e-05, "loss": 2.7068, "step": 23397 }, { "epoch": 1.452479980135328, "grad_norm": 0.5678436193929464, "learning_rate": 6.139096006024782e-05, "loss": 2.8142, "step": 23398 }, { "epoch": 1.452542057235086, "grad_norm": 0.5678750297954467, "learning_rate": 6.138744347697686e-05, "loss": 2.7139, "step": 23399 }, { "epoch": 1.452604134334844, "grad_norm": 0.5745515183522096, "learning_rate": 6.138392683429486e-05, "loss": 2.8083, "step": 23400 }, { "epoch": 1.4526662114346018, "grad_norm": 0.6032601713990094, "learning_rate": 6.138041013222014e-05, "loss": 2.8421, "step": 23401 }, { "epoch": 1.4527282885343598, "grad_norm": 0.5578389169544428, "learning_rate": 6.137689337077105e-05, "loss": 2.7798, "step": 23402 }, { "epoch": 1.4527903656341175, "grad_norm": 0.5943412096063637, "learning_rate": 6.137337654996597e-05, "loss": 2.8674, "step": 23403 }, { "epoch": 1.4528524427338754, "grad_norm": 0.5846381212156364, "learning_rate": 6.136985966982322e-05, "loss": 2.7766, "step": 23404 }, { "epoch": 1.4529145198336333, "grad_norm": 0.5243970132965302, "learning_rate": 6.136634273036114e-05, "loss": 2.7589, "step": 23405 }, { "epoch": 1.4529765969333912, "grad_norm": 0.6174201285487056, "learning_rate": 6.13628257315981e-05, "loss": 2.8451, "step": 23406 }, { "epoch": 1.4530386740331491, "grad_norm": 0.47626288954746226, "learning_rate": 6.135930867355242e-05, "loss": 2.7387, "step": 23407 }, { "epoch": 1.453100751132907, "grad_norm": 0.5996669530404881, "learning_rate": 6.135579155624249e-05, "loss": 2.7562, "step": 23408 }, { "epoch": 1.453162828232665, "grad_norm": 0.5957084280309553, "learning_rate": 6.135227437968663e-05, "loss": 2.8339, "step": 23409 }, { "epoch": 1.4532249053324229, "grad_norm": 0.5779164427812065, "learning_rate": 6.134875714390321e-05, "loss": 2.8662, "step": 23410 }, { "epoch": 1.4532869824321808, "grad_norm": 0.708159163510328, "learning_rate": 6.134523984891055e-05, "loss": 2.772, "step": 23411 }, { "epoch": 1.4533490595319387, "grad_norm": 0.5115807616575213, "learning_rate": 6.134172249472702e-05, "loss": 2.7739, "step": 23412 }, { "epoch": 1.4534111366316966, "grad_norm": 0.6131802129432052, "learning_rate": 6.133820508137098e-05, "loss": 2.8276, "step": 23413 }, { "epoch": 1.4534732137314545, "grad_norm": 0.5841198001877479, "learning_rate": 6.133468760886078e-05, "loss": 2.7979, "step": 23414 }, { "epoch": 1.4535352908312125, "grad_norm": 0.6010966788712163, "learning_rate": 6.133117007721475e-05, "loss": 2.764, "step": 23415 }, { "epoch": 1.4535973679309704, "grad_norm": 0.5727919413888123, "learning_rate": 6.132765248645125e-05, "loss": 2.8167, "step": 23416 }, { "epoch": 1.453659445030728, "grad_norm": 0.6571297652224356, "learning_rate": 6.132413483658866e-05, "loss": 2.9229, "step": 23417 }, { "epoch": 1.453721522130486, "grad_norm": 0.6179438917559259, "learning_rate": 6.13206171276453e-05, "loss": 2.8163, "step": 23418 }, { "epoch": 1.453783599230244, "grad_norm": 0.6124209790929467, "learning_rate": 6.131709935963952e-05, "loss": 2.8611, "step": 23419 }, { "epoch": 1.4538456763300018, "grad_norm": 0.6022639952823244, "learning_rate": 6.131358153258969e-05, "loss": 2.8514, "step": 23420 }, { "epoch": 1.4539077534297598, "grad_norm": 0.5358471274293669, "learning_rate": 6.131006364651416e-05, "loss": 2.8158, "step": 23421 }, { "epoch": 1.4539698305295177, "grad_norm": 0.647534304413404, "learning_rate": 6.13065457014313e-05, "loss": 2.8571, "step": 23422 }, { "epoch": 1.4540319076292756, "grad_norm": 0.5773353596698376, "learning_rate": 6.130302769735941e-05, "loss": 2.9095, "step": 23423 }, { "epoch": 1.4540939847290335, "grad_norm": 0.687995190446976, "learning_rate": 6.129950963431691e-05, "loss": 2.8134, "step": 23424 }, { "epoch": 1.4541560618287914, "grad_norm": 0.5878317413750163, "learning_rate": 6.129599151232213e-05, "loss": 2.8173, "step": 23425 }, { "epoch": 1.4542181389285491, "grad_norm": 0.6098511864934221, "learning_rate": 6.129247333139339e-05, "loss": 2.8048, "step": 23426 }, { "epoch": 1.454280216028307, "grad_norm": 0.6766799067233913, "learning_rate": 6.12889550915491e-05, "loss": 2.7922, "step": 23427 }, { "epoch": 1.454342293128065, "grad_norm": 0.5954032518394143, "learning_rate": 6.128543679280756e-05, "loss": 2.7701, "step": 23428 }, { "epoch": 1.4544043702278229, "grad_norm": 0.6424288126921018, "learning_rate": 6.12819184351872e-05, "loss": 2.8103, "step": 23429 }, { "epoch": 1.4544664473275808, "grad_norm": 0.6078689745087242, "learning_rate": 6.12784000187063e-05, "loss": 2.7624, "step": 23430 }, { "epoch": 1.4545285244273387, "grad_norm": 0.6314424353645209, "learning_rate": 6.127488154338326e-05, "loss": 2.7904, "step": 23431 }, { "epoch": 1.4545906015270966, "grad_norm": 0.6093627732835394, "learning_rate": 6.127136300923644e-05, "loss": 2.7997, "step": 23432 }, { "epoch": 1.4546526786268545, "grad_norm": 0.6104371033807151, "learning_rate": 6.126784441628416e-05, "loss": 2.8208, "step": 23433 }, { "epoch": 1.4547147557266125, "grad_norm": 0.6380160402909423, "learning_rate": 6.126432576454482e-05, "loss": 2.7792, "step": 23434 }, { "epoch": 1.4547768328263704, "grad_norm": 0.5353821048144903, "learning_rate": 6.126080705403674e-05, "loss": 2.7578, "step": 23435 }, { "epoch": 1.4548389099261283, "grad_norm": 0.5831230982496972, "learning_rate": 6.125728828477832e-05, "loss": 2.7575, "step": 23436 }, { "epoch": 1.4549009870258862, "grad_norm": 0.5077517546487726, "learning_rate": 6.125376945678787e-05, "loss": 2.8478, "step": 23437 }, { "epoch": 1.4549630641256441, "grad_norm": 0.5552655552570632, "learning_rate": 6.125025057008378e-05, "loss": 2.7421, "step": 23438 }, { "epoch": 1.455025141225402, "grad_norm": 0.5515447964947505, "learning_rate": 6.12467316246844e-05, "loss": 2.7994, "step": 23439 }, { "epoch": 1.45508721832516, "grad_norm": 0.5798634860915877, "learning_rate": 6.12432126206081e-05, "loss": 2.8102, "step": 23440 }, { "epoch": 1.4551492954249177, "grad_norm": 0.5989894831200653, "learning_rate": 6.123969355787322e-05, "loss": 2.7333, "step": 23441 }, { "epoch": 1.4552113725246756, "grad_norm": 0.598236326063773, "learning_rate": 6.123617443649815e-05, "loss": 2.8869, "step": 23442 }, { "epoch": 1.4552734496244335, "grad_norm": 0.6562225245219117, "learning_rate": 6.12326552565012e-05, "loss": 2.7397, "step": 23443 }, { "epoch": 1.4553355267241914, "grad_norm": 0.5492653784893089, "learning_rate": 6.122913601790078e-05, "loss": 2.7737, "step": 23444 }, { "epoch": 1.4553976038239493, "grad_norm": 0.5772935713873445, "learning_rate": 6.122561672071521e-05, "loss": 2.7676, "step": 23445 }, { "epoch": 1.4554596809237073, "grad_norm": 0.5105141883899703, "learning_rate": 6.12220973649629e-05, "loss": 2.7608, "step": 23446 }, { "epoch": 1.4555217580234652, "grad_norm": 0.5712426843320828, "learning_rate": 6.121857795066216e-05, "loss": 2.9298, "step": 23447 }, { "epoch": 1.455583835123223, "grad_norm": 0.5823913842553646, "learning_rate": 6.121505847783139e-05, "loss": 2.795, "step": 23448 }, { "epoch": 1.455645912222981, "grad_norm": 0.584686749986405, "learning_rate": 6.121153894648894e-05, "loss": 2.7431, "step": 23449 }, { "epoch": 1.4557079893227387, "grad_norm": 0.6078905462449545, "learning_rate": 6.120801935665318e-05, "loss": 2.8004, "step": 23450 }, { "epoch": 1.4557700664224966, "grad_norm": 0.6077509537551659, "learning_rate": 6.120449970834243e-05, "loss": 2.8281, "step": 23451 }, { "epoch": 1.4558321435222545, "grad_norm": 0.5955054783861378, "learning_rate": 6.120098000157511e-05, "loss": 2.7873, "step": 23452 }, { "epoch": 1.4558942206220125, "grad_norm": 0.5716092508425029, "learning_rate": 6.119746023636955e-05, "loss": 2.7045, "step": 23453 }, { "epoch": 1.4559562977217704, "grad_norm": 0.6276705905908477, "learning_rate": 6.119394041274413e-05, "loss": 2.8895, "step": 23454 }, { "epoch": 1.4560183748215283, "grad_norm": 0.5424987075719951, "learning_rate": 6.11904205307172e-05, "loss": 2.8451, "step": 23455 }, { "epoch": 1.4560804519212862, "grad_norm": 0.5930268250424083, "learning_rate": 6.118690059030712e-05, "loss": 2.8359, "step": 23456 }, { "epoch": 1.4561425290210441, "grad_norm": 0.5664468947987548, "learning_rate": 6.118338059153228e-05, "loss": 2.7743, "step": 23457 }, { "epoch": 1.456204606120802, "grad_norm": 0.5934540111219458, "learning_rate": 6.117986053441101e-05, "loss": 2.8062, "step": 23458 }, { "epoch": 1.45626668322056, "grad_norm": 0.6139986009814822, "learning_rate": 6.11763404189617e-05, "loss": 2.8288, "step": 23459 }, { "epoch": 1.4563287603203179, "grad_norm": 0.5922932069990766, "learning_rate": 6.117282024520272e-05, "loss": 2.8358, "step": 23460 }, { "epoch": 1.4563908374200758, "grad_norm": 0.5805828583024912, "learning_rate": 6.11693000131524e-05, "loss": 2.8422, "step": 23461 }, { "epoch": 1.4564529145198337, "grad_norm": 0.6023799741863779, "learning_rate": 6.116577972282915e-05, "loss": 2.7881, "step": 23462 }, { "epoch": 1.4565149916195916, "grad_norm": 0.6487770110958736, "learning_rate": 6.116225937425132e-05, "loss": 2.8306, "step": 23463 }, { "epoch": 1.4565770687193496, "grad_norm": 0.6446890204407537, "learning_rate": 6.115873896743726e-05, "loss": 2.7976, "step": 23464 }, { "epoch": 1.4566391458191073, "grad_norm": 0.5522190832789317, "learning_rate": 6.115521850240537e-05, "loss": 2.7502, "step": 23465 }, { "epoch": 1.4567012229188652, "grad_norm": 0.6176651180990412, "learning_rate": 6.115169797917397e-05, "loss": 2.7034, "step": 23466 }, { "epoch": 1.456763300018623, "grad_norm": 0.5718447599076367, "learning_rate": 6.114817739776147e-05, "loss": 2.8297, "step": 23467 }, { "epoch": 1.456825377118381, "grad_norm": 0.6935015895222969, "learning_rate": 6.114465675818623e-05, "loss": 2.7873, "step": 23468 }, { "epoch": 1.456887454218139, "grad_norm": 0.593363711534014, "learning_rate": 6.11411360604666e-05, "loss": 2.8124, "step": 23469 }, { "epoch": 1.4569495313178968, "grad_norm": 0.6398578330179735, "learning_rate": 6.113761530462097e-05, "loss": 2.7446, "step": 23470 }, { "epoch": 1.4570116084176548, "grad_norm": 0.6889528615492203, "learning_rate": 6.113409449066767e-05, "loss": 2.8352, "step": 23471 }, { "epoch": 1.4570736855174127, "grad_norm": 0.6442092553585645, "learning_rate": 6.113057361862513e-05, "loss": 2.8007, "step": 23472 }, { "epoch": 1.4571357626171706, "grad_norm": 0.5833746236992426, "learning_rate": 6.112705268851166e-05, "loss": 2.7823, "step": 23473 }, { "epoch": 1.4571978397169283, "grad_norm": 0.6210750840938999, "learning_rate": 6.112353170034566e-05, "loss": 2.8252, "step": 23474 }, { "epoch": 1.4572599168166862, "grad_norm": 0.6177620929227537, "learning_rate": 6.11200106541455e-05, "loss": 2.7885, "step": 23475 }, { "epoch": 1.4573219939164441, "grad_norm": 0.6169797088819889, "learning_rate": 6.111648954992954e-05, "loss": 2.8083, "step": 23476 }, { "epoch": 1.457384071016202, "grad_norm": 0.6026974412698897, "learning_rate": 6.111296838771616e-05, "loss": 2.7183, "step": 23477 }, { "epoch": 1.45744614811596, "grad_norm": 0.5321984239891798, "learning_rate": 6.110944716752371e-05, "loss": 2.8882, "step": 23478 }, { "epoch": 1.4575082252157179, "grad_norm": 0.6765619520224804, "learning_rate": 6.11059258893706e-05, "loss": 2.8613, "step": 23479 }, { "epoch": 1.4575703023154758, "grad_norm": 0.6446540253196, "learning_rate": 6.110240455327517e-05, "loss": 2.8489, "step": 23480 }, { "epoch": 1.4576323794152337, "grad_norm": 0.5636670668219137, "learning_rate": 6.10988831592558e-05, "loss": 2.857, "step": 23481 }, { "epoch": 1.4576944565149916, "grad_norm": 0.6260855783643735, "learning_rate": 6.109536170733087e-05, "loss": 2.7433, "step": 23482 }, { "epoch": 1.4577565336147496, "grad_norm": 0.5829092738010183, "learning_rate": 6.109184019751874e-05, "loss": 2.8274, "step": 23483 }, { "epoch": 1.4578186107145075, "grad_norm": 0.6675013194981938, "learning_rate": 6.108831862983777e-05, "loss": 2.7471, "step": 23484 }, { "epoch": 1.4578806878142654, "grad_norm": 0.5883749047614192, "learning_rate": 6.108479700430637e-05, "loss": 2.8303, "step": 23485 }, { "epoch": 1.4579427649140233, "grad_norm": 0.577095042177624, "learning_rate": 6.108127532094289e-05, "loss": 2.8633, "step": 23486 }, { "epoch": 1.4580048420137812, "grad_norm": 0.6944773727133673, "learning_rate": 6.107775357976571e-05, "loss": 2.7862, "step": 23487 }, { "epoch": 1.4580669191135391, "grad_norm": 0.5863840563157496, "learning_rate": 6.107423178079319e-05, "loss": 2.7995, "step": 23488 }, { "epoch": 1.4581289962132968, "grad_norm": 0.6896201301583922, "learning_rate": 6.107070992404371e-05, "loss": 2.8557, "step": 23489 }, { "epoch": 1.4581910733130548, "grad_norm": 0.6368985068612932, "learning_rate": 6.106718800953566e-05, "loss": 2.8489, "step": 23490 }, { "epoch": 1.4582531504128127, "grad_norm": 0.6903660379339546, "learning_rate": 6.106366603728741e-05, "loss": 2.8061, "step": 23491 }, { "epoch": 1.4583152275125706, "grad_norm": 0.6575980872248817, "learning_rate": 6.106014400731732e-05, "loss": 2.7194, "step": 23492 }, { "epoch": 1.4583773046123285, "grad_norm": 0.6099768869523823, "learning_rate": 6.105662191964376e-05, "loss": 2.8367, "step": 23493 }, { "epoch": 1.4584393817120864, "grad_norm": 0.6209874811968494, "learning_rate": 6.105309977428513e-05, "loss": 2.8509, "step": 23494 }, { "epoch": 1.4585014588118443, "grad_norm": 0.6098802623721119, "learning_rate": 6.104957757125981e-05, "loss": 2.7605, "step": 23495 }, { "epoch": 1.4585635359116023, "grad_norm": 0.63620997503478, "learning_rate": 6.104605531058613e-05, "loss": 2.8318, "step": 23496 }, { "epoch": 1.4586256130113602, "grad_norm": 0.607620182354915, "learning_rate": 6.104253299228253e-05, "loss": 2.765, "step": 23497 }, { "epoch": 1.4586876901111179, "grad_norm": 0.5795336393182107, "learning_rate": 6.103901061636735e-05, "loss": 2.7861, "step": 23498 }, { "epoch": 1.4587497672108758, "grad_norm": 0.5583426971741609, "learning_rate": 6.103548818285895e-05, "loss": 2.7561, "step": 23499 }, { "epoch": 1.4588118443106337, "grad_norm": 0.6117756336018851, "learning_rate": 6.103196569177575e-05, "loss": 2.7634, "step": 23500 }, { "epoch": 1.4588739214103916, "grad_norm": 0.5716710566008294, "learning_rate": 6.1028443143136104e-05, "loss": 2.7473, "step": 23501 }, { "epoch": 1.4589359985101495, "grad_norm": 0.5293167786181482, "learning_rate": 6.1024920536958384e-05, "loss": 2.8183, "step": 23502 }, { "epoch": 1.4589980756099075, "grad_norm": 0.5947715604449112, "learning_rate": 6.102139787326099e-05, "loss": 2.8981, "step": 23503 }, { "epoch": 1.4590601527096654, "grad_norm": 0.5653002760510771, "learning_rate": 6.1017875152062264e-05, "loss": 2.7962, "step": 23504 }, { "epoch": 1.4591222298094233, "grad_norm": 0.5353402427080584, "learning_rate": 6.101435237338063e-05, "loss": 2.8197, "step": 23505 }, { "epoch": 1.4591843069091812, "grad_norm": 0.5745700099175834, "learning_rate": 6.101082953723445e-05, "loss": 2.8277, "step": 23506 }, { "epoch": 1.4592463840089391, "grad_norm": 0.5844152640684841, "learning_rate": 6.100730664364208e-05, "loss": 2.8056, "step": 23507 }, { "epoch": 1.459308461108697, "grad_norm": 0.6250677787268739, "learning_rate": 6.100378369262194e-05, "loss": 2.8593, "step": 23508 }, { "epoch": 1.459370538208455, "grad_norm": 0.5778537320021708, "learning_rate": 6.1000260684192364e-05, "loss": 2.8022, "step": 23509 }, { "epoch": 1.4594326153082129, "grad_norm": 0.5931710633137012, "learning_rate": 6.099673761837177e-05, "loss": 2.7868, "step": 23510 }, { "epoch": 1.4594946924079708, "grad_norm": 0.6460253650495225, "learning_rate": 6.099321449517852e-05, "loss": 2.7895, "step": 23511 }, { "epoch": 1.4595567695077287, "grad_norm": 0.5624728196252955, "learning_rate": 6.098969131463102e-05, "loss": 2.7792, "step": 23512 }, { "epoch": 1.4596188466074864, "grad_norm": 0.6261602599363079, "learning_rate": 6.098616807674762e-05, "loss": 2.8559, "step": 23513 }, { "epoch": 1.4596809237072443, "grad_norm": 0.6001654168431726, "learning_rate": 6.098264478154672e-05, "loss": 2.8767, "step": 23514 }, { "epoch": 1.4597430008070023, "grad_norm": 0.5454982594200918, "learning_rate": 6.0979121429046695e-05, "loss": 2.8517, "step": 23515 }, { "epoch": 1.4598050779067602, "grad_norm": 0.6570245622255966, "learning_rate": 6.0975598019265914e-05, "loss": 2.6983, "step": 23516 }, { "epoch": 1.459867155006518, "grad_norm": 0.5168995277369418, "learning_rate": 6.097207455222279e-05, "loss": 2.7647, "step": 23517 }, { "epoch": 1.459929232106276, "grad_norm": 0.6022335372217065, "learning_rate": 6.0968551027935686e-05, "loss": 2.8804, "step": 23518 }, { "epoch": 1.459991309206034, "grad_norm": 0.5469126824611605, "learning_rate": 6.0965027446422986e-05, "loss": 2.8657, "step": 23519 }, { "epoch": 1.4600533863057918, "grad_norm": 0.51306194456974, "learning_rate": 6.096150380770307e-05, "loss": 2.8197, "step": 23520 }, { "epoch": 1.4601154634055498, "grad_norm": 0.5961379418547064, "learning_rate": 6.095798011179434e-05, "loss": 2.7919, "step": 23521 }, { "epoch": 1.4601775405053075, "grad_norm": 0.5109803791310417, "learning_rate": 6.095445635871516e-05, "loss": 2.7836, "step": 23522 }, { "epoch": 1.4602396176050654, "grad_norm": 0.618022016503454, "learning_rate": 6.095093254848393e-05, "loss": 2.8046, "step": 23523 }, { "epoch": 1.4603016947048233, "grad_norm": 0.5986959352553142, "learning_rate": 6.094740868111902e-05, "loss": 2.7802, "step": 23524 }, { "epoch": 1.4603637718045812, "grad_norm": 0.5788890841226051, "learning_rate": 6.094388475663881e-05, "loss": 2.8147, "step": 23525 }, { "epoch": 1.4604258489043391, "grad_norm": 0.5867345856054658, "learning_rate": 6.094036077506171e-05, "loss": 2.8101, "step": 23526 }, { "epoch": 1.460487926004097, "grad_norm": 0.5689408778480379, "learning_rate": 6.093683673640609e-05, "loss": 2.7106, "step": 23527 }, { "epoch": 1.460550003103855, "grad_norm": 0.5871395496728387, "learning_rate": 6.093331264069034e-05, "loss": 2.8008, "step": 23528 }, { "epoch": 1.4606120802036129, "grad_norm": 0.5868359099543454, "learning_rate": 6.0929788487932836e-05, "loss": 2.8289, "step": 23529 }, { "epoch": 1.4606741573033708, "grad_norm": 0.575643419929881, "learning_rate": 6.092626427815198e-05, "loss": 2.8637, "step": 23530 }, { "epoch": 1.4607362344031287, "grad_norm": 0.5674352047181895, "learning_rate": 6.0922740011366154e-05, "loss": 2.7497, "step": 23531 }, { "epoch": 1.4607983115028866, "grad_norm": 0.6149656162884812, "learning_rate": 6.091921568759373e-05, "loss": 2.7866, "step": 23532 }, { "epoch": 1.4608603886026446, "grad_norm": 0.5823757764864738, "learning_rate": 6.091569130685312e-05, "loss": 2.7779, "step": 23533 }, { "epoch": 1.4609224657024025, "grad_norm": 0.594466805693571, "learning_rate": 6.091216686916269e-05, "loss": 2.7288, "step": 23534 }, { "epoch": 1.4609845428021604, "grad_norm": 0.5247961318323439, "learning_rate": 6.0908642374540835e-05, "loss": 2.8387, "step": 23535 }, { "epoch": 1.4610466199019183, "grad_norm": 0.6094230486193412, "learning_rate": 6.090511782300595e-05, "loss": 2.857, "step": 23536 }, { "epoch": 1.461108697001676, "grad_norm": 0.52232379004535, "learning_rate": 6.0901593214576425e-05, "loss": 2.8149, "step": 23537 }, { "epoch": 1.461170774101434, "grad_norm": 0.6043366215068129, "learning_rate": 6.089806854927063e-05, "loss": 2.8551, "step": 23538 }, { "epoch": 1.4612328512011918, "grad_norm": 0.6086720299968441, "learning_rate": 6.089454382710697e-05, "loss": 2.853, "step": 23539 }, { "epoch": 1.4612949283009498, "grad_norm": 0.6243894694487463, "learning_rate": 6.0891019048103817e-05, "loss": 2.8345, "step": 23540 }, { "epoch": 1.4613570054007077, "grad_norm": 0.5184028819956259, "learning_rate": 6.0887494212279595e-05, "loss": 2.8057, "step": 23541 }, { "epoch": 1.4614190825004656, "grad_norm": 0.5943774622195295, "learning_rate": 6.088396931965267e-05, "loss": 2.6538, "step": 23542 }, { "epoch": 1.4614811596002235, "grad_norm": 0.5561889957879826, "learning_rate": 6.088044437024141e-05, "loss": 2.758, "step": 23543 }, { "epoch": 1.4615432366999814, "grad_norm": 0.6101822274865376, "learning_rate": 6.087691936406425e-05, "loss": 2.8002, "step": 23544 }, { "epoch": 1.4616053137997393, "grad_norm": 0.6078996649759872, "learning_rate": 6.0873394301139555e-05, "loss": 2.8973, "step": 23545 }, { "epoch": 1.461667390899497, "grad_norm": 0.6187014059771255, "learning_rate": 6.0869869181485726e-05, "loss": 2.7824, "step": 23546 }, { "epoch": 1.461729467999255, "grad_norm": 0.5977597863765499, "learning_rate": 6.086634400512116e-05, "loss": 2.8267, "step": 23547 }, { "epoch": 1.4617915450990129, "grad_norm": 0.5817745846109734, "learning_rate": 6.0862818772064235e-05, "loss": 2.8433, "step": 23548 }, { "epoch": 1.4618536221987708, "grad_norm": 0.6211224675287158, "learning_rate": 6.0859293482333347e-05, "loss": 2.8651, "step": 23549 }, { "epoch": 1.4619156992985287, "grad_norm": 0.5679399443697957, "learning_rate": 6.085576813594688e-05, "loss": 2.8444, "step": 23550 }, { "epoch": 1.4619777763982866, "grad_norm": 0.6510348866427559, "learning_rate": 6.085224273292324e-05, "loss": 2.8024, "step": 23551 }, { "epoch": 1.4620398534980446, "grad_norm": 0.5357766279776035, "learning_rate": 6.084871727328081e-05, "loss": 2.7052, "step": 23552 }, { "epoch": 1.4621019305978025, "grad_norm": 0.6314350253587834, "learning_rate": 6.0845191757038e-05, "loss": 2.8104, "step": 23553 }, { "epoch": 1.4621640076975604, "grad_norm": 0.5922911691635606, "learning_rate": 6.0841666184213184e-05, "loss": 2.7286, "step": 23554 }, { "epoch": 1.4622260847973183, "grad_norm": 0.681767645863153, "learning_rate": 6.083814055482476e-05, "loss": 2.799, "step": 23555 }, { "epoch": 1.4622881618970762, "grad_norm": 0.5829942915736758, "learning_rate": 6.083461486889113e-05, "loss": 2.8215, "step": 23556 }, { "epoch": 1.4623502389968341, "grad_norm": 0.600606679170545, "learning_rate": 6.08310891264307e-05, "loss": 2.7569, "step": 23557 }, { "epoch": 1.462412316096592, "grad_norm": 0.5684706064558498, "learning_rate": 6.0827563327461825e-05, "loss": 2.8931, "step": 23558 }, { "epoch": 1.46247439319635, "grad_norm": 0.5171247692038482, "learning_rate": 6.0824037472002923e-05, "loss": 2.8204, "step": 23559 }, { "epoch": 1.462536470296108, "grad_norm": 0.5526875182870795, "learning_rate": 6.08205115600724e-05, "loss": 2.858, "step": 23560 }, { "epoch": 1.4625985473958656, "grad_norm": 0.4816977900901613, "learning_rate": 6.081698559168865e-05, "loss": 2.827, "step": 23561 }, { "epoch": 1.4626606244956235, "grad_norm": 0.4974726789007698, "learning_rate": 6.081345956687005e-05, "loss": 2.8201, "step": 23562 }, { "epoch": 1.4627227015953814, "grad_norm": 0.5275910539368956, "learning_rate": 6.0809933485634996e-05, "loss": 2.8567, "step": 23563 }, { "epoch": 1.4627847786951393, "grad_norm": 0.4871829383445681, "learning_rate": 6.0806407348001915e-05, "loss": 2.8529, "step": 23564 }, { "epoch": 1.4628468557948973, "grad_norm": 0.6004829380444732, "learning_rate": 6.080288115398917e-05, "loss": 2.737, "step": 23565 }, { "epoch": 1.4629089328946552, "grad_norm": 0.4810304692158632, "learning_rate": 6.079935490361518e-05, "loss": 2.7345, "step": 23566 }, { "epoch": 1.462971009994413, "grad_norm": 0.6013127093218114, "learning_rate": 6.079582859689834e-05, "loss": 2.7425, "step": 23567 }, { "epoch": 1.463033087094171, "grad_norm": 0.5688496732880682, "learning_rate": 6.0792302233857024e-05, "loss": 2.7857, "step": 23568 }, { "epoch": 1.463095164193929, "grad_norm": 0.5543082377886805, "learning_rate": 6.0788775814509655e-05, "loss": 2.7857, "step": 23569 }, { "epoch": 1.4631572412936866, "grad_norm": 0.5974860104714769, "learning_rate": 6.0785249338874615e-05, "loss": 2.8237, "step": 23570 }, { "epoch": 1.4632193183934445, "grad_norm": 0.5160744182351105, "learning_rate": 6.078172280697033e-05, "loss": 2.7986, "step": 23571 }, { "epoch": 1.4632813954932025, "grad_norm": 0.5494263127897407, "learning_rate": 6.077819621881516e-05, "loss": 2.7907, "step": 23572 }, { "epoch": 1.4633434725929604, "grad_norm": 0.5571792138511025, "learning_rate": 6.0774669574427544e-05, "loss": 2.8626, "step": 23573 }, { "epoch": 1.4634055496927183, "grad_norm": 0.5131437248038084, "learning_rate": 6.077114287382584e-05, "loss": 2.8528, "step": 23574 }, { "epoch": 1.4634676267924762, "grad_norm": 0.5698511530795627, "learning_rate": 6.076761611702848e-05, "loss": 2.7003, "step": 23575 }, { "epoch": 1.4635297038922341, "grad_norm": 0.5363563946902205, "learning_rate": 6.0764089304053854e-05, "loss": 2.8292, "step": 23576 }, { "epoch": 1.463591780991992, "grad_norm": 0.5402535810938838, "learning_rate": 6.0760562434920354e-05, "loss": 2.8021, "step": 23577 }, { "epoch": 1.46365385809175, "grad_norm": 0.5491253736839005, "learning_rate": 6.07570355096464e-05, "loss": 2.8043, "step": 23578 }, { "epoch": 1.4637159351915079, "grad_norm": 0.5362152981144858, "learning_rate": 6.075350852825037e-05, "loss": 2.7952, "step": 23579 }, { "epoch": 1.4637780122912658, "grad_norm": 0.591743114283105, "learning_rate": 6.074998149075068e-05, "loss": 2.7198, "step": 23580 }, { "epoch": 1.4638400893910237, "grad_norm": 0.5973341578801813, "learning_rate": 6.074645439716572e-05, "loss": 2.7768, "step": 23581 }, { "epoch": 1.4639021664907816, "grad_norm": 0.5770229713141494, "learning_rate": 6.07429272475139e-05, "loss": 2.7374, "step": 23582 }, { "epoch": 1.4639642435905396, "grad_norm": 0.553253765527776, "learning_rate": 6.073940004181362e-05, "loss": 2.7517, "step": 23583 }, { "epoch": 1.4640263206902975, "grad_norm": 0.5751647422969832, "learning_rate": 6.07358727800833e-05, "loss": 2.8642, "step": 23584 }, { "epoch": 1.4640883977900552, "grad_norm": 0.6979396351313357, "learning_rate": 6.073234546234131e-05, "loss": 2.7737, "step": 23585 }, { "epoch": 1.464150474889813, "grad_norm": 0.5461735723002845, "learning_rate": 6.0728818088606066e-05, "loss": 2.7299, "step": 23586 }, { "epoch": 1.464212551989571, "grad_norm": 0.6141974293636063, "learning_rate": 6.0725290658895985e-05, "loss": 2.8554, "step": 23587 }, { "epoch": 1.464274629089329, "grad_norm": 0.5558582630266982, "learning_rate": 6.072176317322945e-05, "loss": 2.8267, "step": 23588 }, { "epoch": 1.4643367061890868, "grad_norm": 0.5696531429592133, "learning_rate": 6.0718235631624886e-05, "loss": 2.6958, "step": 23589 }, { "epoch": 1.4643987832888448, "grad_norm": 0.5109744009055772, "learning_rate": 6.0714708034100664e-05, "loss": 2.7389, "step": 23590 }, { "epoch": 1.4644608603886027, "grad_norm": 0.5672544245473798, "learning_rate": 6.071118038067523e-05, "loss": 2.8695, "step": 23591 }, { "epoch": 1.4645229374883606, "grad_norm": 0.5743577490489646, "learning_rate": 6.070765267136696e-05, "loss": 2.7785, "step": 23592 }, { "epoch": 1.4645850145881185, "grad_norm": 0.5321081187191814, "learning_rate": 6.070412490619425e-05, "loss": 2.7863, "step": 23593 }, { "epoch": 1.4646470916878762, "grad_norm": 0.5615808341504729, "learning_rate": 6.070059708517554e-05, "loss": 2.7338, "step": 23594 }, { "epoch": 1.4647091687876341, "grad_norm": 0.5178651697495094, "learning_rate": 6.0697069208329204e-05, "loss": 2.8081, "step": 23595 }, { "epoch": 1.464771245887392, "grad_norm": 0.6038046579912896, "learning_rate": 6.069354127567368e-05, "loss": 2.8358, "step": 23596 }, { "epoch": 1.46483332298715, "grad_norm": 0.4938950512686589, "learning_rate": 6.069001328722734e-05, "loss": 2.8156, "step": 23597 }, { "epoch": 1.4648954000869079, "grad_norm": 0.5825053338801238, "learning_rate": 6.068648524300862e-05, "loss": 2.8162, "step": 23598 }, { "epoch": 1.4649574771866658, "grad_norm": 0.5670949760613369, "learning_rate": 6.06829571430359e-05, "loss": 2.7596, "step": 23599 }, { "epoch": 1.4650195542864237, "grad_norm": 0.5704923894744011, "learning_rate": 6.06794289873276e-05, "loss": 2.8085, "step": 23600 }, { "epoch": 1.4650816313861816, "grad_norm": 0.5526203536653921, "learning_rate": 6.067590077590213e-05, "loss": 2.8715, "step": 23601 }, { "epoch": 1.4651437084859396, "grad_norm": 0.5732640744763582, "learning_rate": 6.06723725087779e-05, "loss": 2.8055, "step": 23602 }, { "epoch": 1.4652057855856975, "grad_norm": 0.5136942405328062, "learning_rate": 6.0668844185973305e-05, "loss": 2.7392, "step": 23603 }, { "epoch": 1.4652678626854554, "grad_norm": 0.6198272748276489, "learning_rate": 6.066531580750676e-05, "loss": 2.8302, "step": 23604 }, { "epoch": 1.4653299397852133, "grad_norm": 0.5561738583604011, "learning_rate": 6.0661787373396674e-05, "loss": 2.8178, "step": 23605 }, { "epoch": 1.4653920168849712, "grad_norm": 0.6018654692024963, "learning_rate": 6.065825888366146e-05, "loss": 2.716, "step": 23606 }, { "epoch": 1.4654540939847291, "grad_norm": 0.569650579439463, "learning_rate": 6.0654730338319524e-05, "loss": 2.7699, "step": 23607 }, { "epoch": 1.465516171084487, "grad_norm": 0.5917121650323351, "learning_rate": 6.065120173738924e-05, "loss": 2.7994, "step": 23608 }, { "epoch": 1.4655782481842448, "grad_norm": 0.5853151704041949, "learning_rate": 6.0647673080889076e-05, "loss": 2.8843, "step": 23609 }, { "epoch": 1.4656403252840027, "grad_norm": 0.6056647937011241, "learning_rate": 6.064414436883743e-05, "loss": 2.7894, "step": 23610 }, { "epoch": 1.4657024023837606, "grad_norm": 0.5618628230912971, "learning_rate": 6.064061560125268e-05, "loss": 2.7976, "step": 23611 }, { "epoch": 1.4657644794835185, "grad_norm": 0.5730235578061578, "learning_rate": 6.0637086778153276e-05, "loss": 2.8241, "step": 23612 }, { "epoch": 1.4658265565832764, "grad_norm": 0.5923936308977525, "learning_rate": 6.063355789955758e-05, "loss": 2.7546, "step": 23613 }, { "epoch": 1.4658886336830343, "grad_norm": 0.5484502448532917, "learning_rate": 6.063002896548405e-05, "loss": 2.7244, "step": 23614 }, { "epoch": 1.4659507107827923, "grad_norm": 0.5782871276754662, "learning_rate": 6.062649997595107e-05, "loss": 2.856, "step": 23615 }, { "epoch": 1.4660127878825502, "grad_norm": 0.5454061673666524, "learning_rate": 6.062297093097706e-05, "loss": 2.7022, "step": 23616 }, { "epoch": 1.466074864982308, "grad_norm": 0.6526737178102412, "learning_rate": 6.061944183058044e-05, "loss": 2.7988, "step": 23617 }, { "epoch": 1.4661369420820658, "grad_norm": 0.6965016044672158, "learning_rate": 6.0615912674779605e-05, "loss": 2.7901, "step": 23618 }, { "epoch": 1.4661990191818237, "grad_norm": 0.6065097350764646, "learning_rate": 6.061238346359298e-05, "loss": 2.7667, "step": 23619 }, { "epoch": 1.4662610962815816, "grad_norm": 0.618395383922601, "learning_rate": 6.0608854197038975e-05, "loss": 2.7152, "step": 23620 }, { "epoch": 1.4663231733813396, "grad_norm": 0.6765779198349589, "learning_rate": 6.0605324875136016e-05, "loss": 2.8433, "step": 23621 }, { "epoch": 1.4663852504810975, "grad_norm": 0.5580658185195645, "learning_rate": 6.060179549790248e-05, "loss": 2.6849, "step": 23622 }, { "epoch": 1.4664473275808554, "grad_norm": 0.6697980805283247, "learning_rate": 6.059826606535681e-05, "loss": 2.7657, "step": 23623 }, { "epoch": 1.4665094046806133, "grad_norm": 0.5311435985599219, "learning_rate": 6.059473657751742e-05, "loss": 2.7961, "step": 23624 }, { "epoch": 1.4665714817803712, "grad_norm": 0.7442193177433479, "learning_rate": 6.059120703440271e-05, "loss": 2.8686, "step": 23625 }, { "epoch": 1.4666335588801291, "grad_norm": 0.5839964186022956, "learning_rate": 6.058767743603109e-05, "loss": 2.8207, "step": 23626 }, { "epoch": 1.466695635979887, "grad_norm": 0.6047828727246647, "learning_rate": 6.0584147782421006e-05, "loss": 2.8691, "step": 23627 }, { "epoch": 1.466757713079645, "grad_norm": 0.6594806803403395, "learning_rate": 6.0580618073590856e-05, "loss": 2.7717, "step": 23628 }, { "epoch": 1.466819790179403, "grad_norm": 0.6248255963179297, "learning_rate": 6.057708830955904e-05, "loss": 2.9741, "step": 23629 }, { "epoch": 1.4668818672791608, "grad_norm": 0.6624308118139745, "learning_rate": 6.0573558490343996e-05, "loss": 2.819, "step": 23630 }, { "epoch": 1.4669439443789187, "grad_norm": 0.6340813572590822, "learning_rate": 6.057002861596412e-05, "loss": 2.7447, "step": 23631 }, { "epoch": 1.4670060214786766, "grad_norm": 0.7043928583448676, "learning_rate": 6.056649868643786e-05, "loss": 2.7084, "step": 23632 }, { "epoch": 1.4670680985784343, "grad_norm": 0.6056053745802286, "learning_rate": 6.0562968701783583e-05, "loss": 2.7005, "step": 23633 }, { "epoch": 1.4671301756781923, "grad_norm": 0.6405580066914498, "learning_rate": 6.0559438662019754e-05, "loss": 2.8169, "step": 23634 }, { "epoch": 1.4671922527779502, "grad_norm": 0.5825563660556444, "learning_rate": 6.055590856716475e-05, "loss": 2.7986, "step": 23635 }, { "epoch": 1.467254329877708, "grad_norm": 0.6632157480991292, "learning_rate": 6.055237841723703e-05, "loss": 2.7897, "step": 23636 }, { "epoch": 1.467316406977466, "grad_norm": 0.5975361372980893, "learning_rate": 6.054884821225498e-05, "loss": 2.7909, "step": 23637 }, { "epoch": 1.467378484077224, "grad_norm": 0.6311439840597446, "learning_rate": 6.054531795223703e-05, "loss": 2.9033, "step": 23638 }, { "epoch": 1.4674405611769819, "grad_norm": 0.6886253899493371, "learning_rate": 6.05417876372016e-05, "loss": 2.7578, "step": 23639 }, { "epoch": 1.4675026382767398, "grad_norm": 0.6292853308766662, "learning_rate": 6.05382572671671e-05, "loss": 2.7412, "step": 23640 }, { "epoch": 1.4675647153764977, "grad_norm": 0.5876434921993566, "learning_rate": 6.0534726842151957e-05, "loss": 2.7522, "step": 23641 }, { "epoch": 1.4676267924762554, "grad_norm": 0.6332394372410768, "learning_rate": 6.053119636217457e-05, "loss": 2.781, "step": 23642 }, { "epoch": 1.4676888695760133, "grad_norm": 0.5945186658773113, "learning_rate": 6.05276658272534e-05, "loss": 2.8184, "step": 23643 }, { "epoch": 1.4677509466757712, "grad_norm": 0.5726789212390416, "learning_rate": 6.052413523740682e-05, "loss": 2.8179, "step": 23644 }, { "epoch": 1.4678130237755291, "grad_norm": 0.6500518878987108, "learning_rate": 6.0520604592653295e-05, "loss": 2.8495, "step": 23645 }, { "epoch": 1.467875100875287, "grad_norm": 0.6369942379216277, "learning_rate": 6.0517073893011213e-05, "loss": 2.8758, "step": 23646 }, { "epoch": 1.467937177975045, "grad_norm": 0.5340415315657892, "learning_rate": 6.0513543138498994e-05, "loss": 2.7923, "step": 23647 }, { "epoch": 1.467999255074803, "grad_norm": 0.6734475429343665, "learning_rate": 6.0510012329135077e-05, "loss": 2.7521, "step": 23648 }, { "epoch": 1.4680613321745608, "grad_norm": 0.5830826192176062, "learning_rate": 6.050648146493787e-05, "loss": 2.8159, "step": 23649 }, { "epoch": 1.4681234092743187, "grad_norm": 0.6419055514693273, "learning_rate": 6.050295054592581e-05, "loss": 2.7806, "step": 23650 }, { "epoch": 1.4681854863740766, "grad_norm": 0.6130489225172295, "learning_rate": 6.049941957211729e-05, "loss": 2.7961, "step": 23651 }, { "epoch": 1.4682475634738346, "grad_norm": 0.6850812368532166, "learning_rate": 6.0495888543530754e-05, "loss": 2.8396, "step": 23652 }, { "epoch": 1.4683096405735925, "grad_norm": 0.514382425720456, "learning_rate": 6.0492357460184624e-05, "loss": 2.7024, "step": 23653 }, { "epoch": 1.4683717176733504, "grad_norm": 0.708964534093072, "learning_rate": 6.0488826322097316e-05, "loss": 2.8264, "step": 23654 }, { "epoch": 1.4684337947731083, "grad_norm": 0.5220554953376602, "learning_rate": 6.0485295129287255e-05, "loss": 2.7858, "step": 23655 }, { "epoch": 1.4684958718728662, "grad_norm": 0.6664996807257881, "learning_rate": 6.048176388177287e-05, "loss": 2.7909, "step": 23656 }, { "epoch": 1.468557948972624, "grad_norm": 0.5657805834758112, "learning_rate": 6.047823257957257e-05, "loss": 2.7831, "step": 23657 }, { "epoch": 1.4686200260723818, "grad_norm": 0.7525966912308623, "learning_rate": 6.047470122270478e-05, "loss": 2.8032, "step": 23658 }, { "epoch": 1.4686821031721398, "grad_norm": 0.5189517812214447, "learning_rate": 6.047116981118794e-05, "loss": 2.7237, "step": 23659 }, { "epoch": 1.4687441802718977, "grad_norm": 0.7081521354115724, "learning_rate": 6.0467638345040464e-05, "loss": 2.7984, "step": 23660 }, { "epoch": 1.4688062573716556, "grad_norm": 0.5562929640475701, "learning_rate": 6.046410682428079e-05, "loss": 2.8396, "step": 23661 }, { "epoch": 1.4688683344714135, "grad_norm": 0.71488360517952, "learning_rate": 6.046057524892731e-05, "loss": 2.7722, "step": 23662 }, { "epoch": 1.4689304115711714, "grad_norm": 0.6117693737611702, "learning_rate": 6.0457043618998476e-05, "loss": 2.7337, "step": 23663 }, { "epoch": 1.4689924886709294, "grad_norm": 0.6750058482941069, "learning_rate": 6.045351193451271e-05, "loss": 2.7715, "step": 23664 }, { "epoch": 1.4690545657706873, "grad_norm": 0.6131790221715657, "learning_rate": 6.0449980195488446e-05, "loss": 2.7925, "step": 23665 }, { "epoch": 1.469116642870445, "grad_norm": 0.6779298088317605, "learning_rate": 6.044644840194409e-05, "loss": 2.7854, "step": 23666 }, { "epoch": 1.4691787199702029, "grad_norm": 0.6748343653025249, "learning_rate": 6.044291655389807e-05, "loss": 2.8556, "step": 23667 }, { "epoch": 1.4692407970699608, "grad_norm": 0.5828542587006226, "learning_rate": 6.043938465136882e-05, "loss": 2.8411, "step": 23668 }, { "epoch": 1.4693028741697187, "grad_norm": 0.6708703625175607, "learning_rate": 6.0435852694374765e-05, "loss": 2.881, "step": 23669 }, { "epoch": 1.4693649512694766, "grad_norm": 0.5857394328725163, "learning_rate": 6.043232068293434e-05, "loss": 2.8198, "step": 23670 }, { "epoch": 1.4694270283692346, "grad_norm": 0.5890961766418886, "learning_rate": 6.0428788617065955e-05, "loss": 2.7055, "step": 23671 }, { "epoch": 1.4694891054689925, "grad_norm": 0.6365299295923649, "learning_rate": 6.042525649678806e-05, "loss": 2.8567, "step": 23672 }, { "epoch": 1.4695511825687504, "grad_norm": 0.6125504336710476, "learning_rate": 6.0421724322119064e-05, "loss": 2.7903, "step": 23673 }, { "epoch": 1.4696132596685083, "grad_norm": 0.6324214433429416, "learning_rate": 6.0418192093077396e-05, "loss": 2.8124, "step": 23674 }, { "epoch": 1.4696753367682662, "grad_norm": 0.556257310026334, "learning_rate": 6.041465980968151e-05, "loss": 2.8168, "step": 23675 }, { "epoch": 1.4697374138680241, "grad_norm": 0.646138078305928, "learning_rate": 6.04111274719498e-05, "loss": 2.7191, "step": 23676 }, { "epoch": 1.469799490967782, "grad_norm": 0.5614078568632772, "learning_rate": 6.040759507990073e-05, "loss": 2.7831, "step": 23677 }, { "epoch": 1.46986156806754, "grad_norm": 0.586943259123912, "learning_rate": 6.040406263355268e-05, "loss": 2.8343, "step": 23678 }, { "epoch": 1.469923645167298, "grad_norm": 0.5618095398855784, "learning_rate": 6.0400530132924136e-05, "loss": 2.7496, "step": 23679 }, { "epoch": 1.4699857222670558, "grad_norm": 0.6167543557256399, "learning_rate": 6.0396997578033474e-05, "loss": 2.7922, "step": 23680 }, { "epoch": 1.4700477993668135, "grad_norm": 0.5622328017654631, "learning_rate": 6.039346496889918e-05, "loss": 2.7316, "step": 23681 }, { "epoch": 1.4701098764665714, "grad_norm": 0.5536168740116073, "learning_rate": 6.038993230553964e-05, "loss": 2.6835, "step": 23682 }, { "epoch": 1.4701719535663293, "grad_norm": 0.5798770286243344, "learning_rate": 6.0386399587973306e-05, "loss": 2.7281, "step": 23683 }, { "epoch": 1.4702340306660873, "grad_norm": 0.5390465222630695, "learning_rate": 6.038286681621862e-05, "loss": 2.7806, "step": 23684 }, { "epoch": 1.4702961077658452, "grad_norm": 0.6083238044628256, "learning_rate": 6.037933399029397e-05, "loss": 2.6445, "step": 23685 }, { "epoch": 1.470358184865603, "grad_norm": 0.5430982695960579, "learning_rate": 6.037580111021783e-05, "loss": 2.8092, "step": 23686 }, { "epoch": 1.470420261965361, "grad_norm": 0.561512716439772, "learning_rate": 6.0372268176008605e-05, "loss": 2.8072, "step": 23687 }, { "epoch": 1.470482339065119, "grad_norm": 0.6365085793206902, "learning_rate": 6.036873518768475e-05, "loss": 2.7743, "step": 23688 }, { "epoch": 1.4705444161648769, "grad_norm": 0.6166983247268732, "learning_rate": 6.036520214526469e-05, "loss": 2.7638, "step": 23689 }, { "epoch": 1.4706064932646346, "grad_norm": 0.6209939600593971, "learning_rate": 6.0361669048766835e-05, "loss": 2.7564, "step": 23690 }, { "epoch": 1.4706685703643925, "grad_norm": 0.7712971520500767, "learning_rate": 6.035813589820966e-05, "loss": 2.74, "step": 23691 }, { "epoch": 1.4707306474641504, "grad_norm": 0.6594929720668264, "learning_rate": 6.035460269361155e-05, "loss": 2.773, "step": 23692 }, { "epoch": 1.4707927245639083, "grad_norm": 0.6711812096693607, "learning_rate": 6.035106943499098e-05, "loss": 2.7428, "step": 23693 }, { "epoch": 1.4708548016636662, "grad_norm": 0.6940770692768983, "learning_rate": 6.034753612236637e-05, "loss": 2.7969, "step": 23694 }, { "epoch": 1.4709168787634241, "grad_norm": 0.6585196800708996, "learning_rate": 6.034400275575616e-05, "loss": 2.8326, "step": 23695 }, { "epoch": 1.470978955863182, "grad_norm": 0.8443139628635503, "learning_rate": 6.034046933517875e-05, "loss": 2.8295, "step": 23696 }, { "epoch": 1.47104103296294, "grad_norm": 0.6126564633739502, "learning_rate": 6.0336935860652624e-05, "loss": 2.7777, "step": 23697 }, { "epoch": 1.471103110062698, "grad_norm": 0.7059789382167666, "learning_rate": 6.033340233219618e-05, "loss": 2.873, "step": 23698 }, { "epoch": 1.4711651871624558, "grad_norm": 0.6588684251370744, "learning_rate": 6.032986874982788e-05, "loss": 2.8623, "step": 23699 }, { "epoch": 1.4712272642622137, "grad_norm": 0.5657404702699224, "learning_rate": 6.032633511356613e-05, "loss": 2.7469, "step": 23700 }, { "epoch": 1.4712893413619716, "grad_norm": 0.6648499476214532, "learning_rate": 6.0322801423429406e-05, "loss": 2.7671, "step": 23701 }, { "epoch": 1.4713514184617296, "grad_norm": 0.6301081764353008, "learning_rate": 6.031926767943612e-05, "loss": 2.7783, "step": 23702 }, { "epoch": 1.4714134955614875, "grad_norm": 0.6311306916220307, "learning_rate": 6.0315733881604685e-05, "loss": 2.7351, "step": 23703 }, { "epoch": 1.4714755726612454, "grad_norm": 0.64472084724692, "learning_rate": 6.031220002995358e-05, "loss": 2.84, "step": 23704 }, { "epoch": 1.471537649761003, "grad_norm": 0.5728740127175413, "learning_rate": 6.030866612450122e-05, "loss": 2.7671, "step": 23705 }, { "epoch": 1.471599726860761, "grad_norm": 0.6463400688345058, "learning_rate": 6.0305132165266056e-05, "loss": 2.7968, "step": 23706 }, { "epoch": 1.471661803960519, "grad_norm": 0.5605932342817828, "learning_rate": 6.030159815226648e-05, "loss": 2.8275, "step": 23707 }, { "epoch": 1.4717238810602769, "grad_norm": 0.5933958804545235, "learning_rate": 6.0298064085521e-05, "loss": 2.766, "step": 23708 }, { "epoch": 1.4717859581600348, "grad_norm": 0.6303792017763199, "learning_rate": 6.029452996504802e-05, "loss": 2.7608, "step": 23709 }, { "epoch": 1.4718480352597927, "grad_norm": 0.5503112194213178, "learning_rate": 6.0290995790865964e-05, "loss": 2.8016, "step": 23710 }, { "epoch": 1.4719101123595506, "grad_norm": 0.6044638138676847, "learning_rate": 6.02874615629933e-05, "loss": 2.8016, "step": 23711 }, { "epoch": 1.4719721894593085, "grad_norm": 0.589259635257399, "learning_rate": 6.028392728144843e-05, "loss": 2.7728, "step": 23712 }, { "epoch": 1.4720342665590664, "grad_norm": 0.6121789781912967, "learning_rate": 6.028039294624983e-05, "loss": 2.7967, "step": 23713 }, { "epoch": 1.4720963436588241, "grad_norm": 0.5460272894544134, "learning_rate": 6.0276858557415914e-05, "loss": 2.822, "step": 23714 }, { "epoch": 1.472158420758582, "grad_norm": 0.6225493067628387, "learning_rate": 6.027332411496515e-05, "loss": 2.8125, "step": 23715 }, { "epoch": 1.47222049785834, "grad_norm": 0.5832414526331491, "learning_rate": 6.0269789618915936e-05, "loss": 2.8636, "step": 23716 }, { "epoch": 1.472282574958098, "grad_norm": 0.6346706847569886, "learning_rate": 6.026625506928676e-05, "loss": 2.8448, "step": 23717 }, { "epoch": 1.4723446520578558, "grad_norm": 0.5262864710930092, "learning_rate": 6.026272046609602e-05, "loss": 2.734, "step": 23718 }, { "epoch": 1.4724067291576137, "grad_norm": 0.601748276795605, "learning_rate": 6.025918580936219e-05, "loss": 2.8065, "step": 23719 }, { "epoch": 1.4724688062573716, "grad_norm": 0.5448907002381245, "learning_rate": 6.025565109910371e-05, "loss": 2.8054, "step": 23720 }, { "epoch": 1.4725308833571296, "grad_norm": 0.6365023754342246, "learning_rate": 6.0252116335338984e-05, "loss": 2.8635, "step": 23721 }, { "epoch": 1.4725929604568875, "grad_norm": 0.5390428180142186, "learning_rate": 6.0248581518086475e-05, "loss": 2.7541, "step": 23722 }, { "epoch": 1.4726550375566454, "grad_norm": 0.6235150578186415, "learning_rate": 6.024504664736463e-05, "loss": 2.8439, "step": 23723 }, { "epoch": 1.4727171146564033, "grad_norm": 0.5369442772715656, "learning_rate": 6.02415117231919e-05, "loss": 2.8214, "step": 23724 }, { "epoch": 1.4727791917561612, "grad_norm": 0.5714759716529484, "learning_rate": 6.023797674558671e-05, "loss": 2.7351, "step": 23725 }, { "epoch": 1.4728412688559192, "grad_norm": 0.5310493819717763, "learning_rate": 6.0234441714567524e-05, "loss": 2.7176, "step": 23726 }, { "epoch": 1.472903345955677, "grad_norm": 0.5407567144921467, "learning_rate": 6.023090663015276e-05, "loss": 2.795, "step": 23727 }, { "epoch": 1.472965423055435, "grad_norm": 0.5825621468623048, "learning_rate": 6.022737149236087e-05, "loss": 2.7982, "step": 23728 }, { "epoch": 1.4730275001551927, "grad_norm": 0.5198347022894793, "learning_rate": 6.022383630121031e-05, "loss": 2.8128, "step": 23729 }, { "epoch": 1.4730895772549506, "grad_norm": 0.5319832621931465, "learning_rate": 6.022030105671951e-05, "loss": 2.7498, "step": 23730 }, { "epoch": 1.4731516543547085, "grad_norm": 0.5836431980249442, "learning_rate": 6.021676575890692e-05, "loss": 2.8229, "step": 23731 }, { "epoch": 1.4732137314544664, "grad_norm": 0.5288023943903122, "learning_rate": 6.021323040779098e-05, "loss": 2.8486, "step": 23732 }, { "epoch": 1.4732758085542244, "grad_norm": 0.5326774716926528, "learning_rate": 6.020969500339013e-05, "loss": 2.7167, "step": 23733 }, { "epoch": 1.4733378856539823, "grad_norm": 0.5472974235668489, "learning_rate": 6.020615954572283e-05, "loss": 2.8016, "step": 23734 }, { "epoch": 1.4733999627537402, "grad_norm": 0.5790078664699883, "learning_rate": 6.020262403480752e-05, "loss": 2.895, "step": 23735 }, { "epoch": 1.473462039853498, "grad_norm": 0.5104186453748483, "learning_rate": 6.0199088470662644e-05, "loss": 2.8575, "step": 23736 }, { "epoch": 1.473524116953256, "grad_norm": 0.5086724848824888, "learning_rate": 6.019555285330665e-05, "loss": 2.7274, "step": 23737 }, { "epoch": 1.4735861940530137, "grad_norm": 0.49188038160198705, "learning_rate": 6.019201718275798e-05, "loss": 2.883, "step": 23738 }, { "epoch": 1.4736482711527716, "grad_norm": 0.5187714836543738, "learning_rate": 6.018848145903506e-05, "loss": 2.7076, "step": 23739 }, { "epoch": 1.4737103482525296, "grad_norm": 0.5906631386770296, "learning_rate": 6.018494568215639e-05, "loss": 2.846, "step": 23740 }, { "epoch": 1.4737724253522875, "grad_norm": 0.5536202115876551, "learning_rate": 6.018140985214037e-05, "loss": 2.8267, "step": 23741 }, { "epoch": 1.4738345024520454, "grad_norm": 0.6369107430847383, "learning_rate": 6.017787396900547e-05, "loss": 2.8374, "step": 23742 }, { "epoch": 1.4738965795518033, "grad_norm": 0.530197447219741, "learning_rate": 6.0174338032770114e-05, "loss": 2.8626, "step": 23743 }, { "epoch": 1.4739586566515612, "grad_norm": 0.6497437292870071, "learning_rate": 6.017080204345279e-05, "loss": 2.7909, "step": 23744 }, { "epoch": 1.4740207337513191, "grad_norm": 0.5676996131250576, "learning_rate": 6.016726600107192e-05, "loss": 2.7432, "step": 23745 }, { "epoch": 1.474082810851077, "grad_norm": 0.5809274229557724, "learning_rate": 6.016372990564594e-05, "loss": 2.7179, "step": 23746 }, { "epoch": 1.474144887950835, "grad_norm": 0.6312825581680702, "learning_rate": 6.016019375719333e-05, "loss": 2.7946, "step": 23747 }, { "epoch": 1.474206965050593, "grad_norm": 0.5389541295903486, "learning_rate": 6.015665755573251e-05, "loss": 2.821, "step": 23748 }, { "epoch": 1.4742690421503508, "grad_norm": 0.5824238010223698, "learning_rate": 6.015312130128196e-05, "loss": 2.6861, "step": 23749 }, { "epoch": 1.4743311192501087, "grad_norm": 0.5969036325108039, "learning_rate": 6.0149584993860085e-05, "loss": 2.7757, "step": 23750 }, { "epoch": 1.4743931963498667, "grad_norm": 0.5648474190620509, "learning_rate": 6.014604863348537e-05, "loss": 2.7855, "step": 23751 }, { "epoch": 1.4744552734496246, "grad_norm": 0.5767588583435849, "learning_rate": 6.014251222017626e-05, "loss": 2.7198, "step": 23752 }, { "epoch": 1.4745173505493823, "grad_norm": 0.5394797990462191, "learning_rate": 6.013897575395121e-05, "loss": 2.807, "step": 23753 }, { "epoch": 1.4745794276491402, "grad_norm": 0.6053160444085959, "learning_rate": 6.013543923482865e-05, "loss": 2.8579, "step": 23754 }, { "epoch": 1.474641504748898, "grad_norm": 0.5202981838055317, "learning_rate": 6.013190266282706e-05, "loss": 2.7097, "step": 23755 }, { "epoch": 1.474703581848656, "grad_norm": 0.6573253857708938, "learning_rate": 6.012836603796484e-05, "loss": 2.7804, "step": 23756 }, { "epoch": 1.474765658948414, "grad_norm": 0.5216246855492813, "learning_rate": 6.0124829360260504e-05, "loss": 2.7011, "step": 23757 }, { "epoch": 1.4748277360481719, "grad_norm": 0.5869915446138736, "learning_rate": 6.0121292629732474e-05, "loss": 2.8253, "step": 23758 }, { "epoch": 1.4748898131479298, "grad_norm": 0.5535184064645815, "learning_rate": 6.011775584639918e-05, "loss": 2.7248, "step": 23759 }, { "epoch": 1.4749518902476877, "grad_norm": 0.54342706122965, "learning_rate": 6.0114219010279124e-05, "loss": 2.7494, "step": 23760 }, { "epoch": 1.4750139673474456, "grad_norm": 0.657066881300003, "learning_rate": 6.011068212139073e-05, "loss": 2.7593, "step": 23761 }, { "epoch": 1.4750760444472033, "grad_norm": 0.5513548873626632, "learning_rate": 6.010714517975244e-05, "loss": 2.7394, "step": 23762 }, { "epoch": 1.4751381215469612, "grad_norm": 0.5785533761348367, "learning_rate": 6.010360818538273e-05, "loss": 2.7733, "step": 23763 }, { "epoch": 1.4752001986467191, "grad_norm": 0.5631369851770753, "learning_rate": 6.010007113830004e-05, "loss": 2.6258, "step": 23764 }, { "epoch": 1.475262275746477, "grad_norm": 0.5754595992940049, "learning_rate": 6.009653403852283e-05, "loss": 2.865, "step": 23765 }, { "epoch": 1.475324352846235, "grad_norm": 0.6881217962358475, "learning_rate": 6.009299688606954e-05, "loss": 2.7, "step": 23766 }, { "epoch": 1.475386429945993, "grad_norm": 0.5529118714583398, "learning_rate": 6.008945968095865e-05, "loss": 2.8012, "step": 23767 }, { "epoch": 1.4754485070457508, "grad_norm": 0.6593436030284614, "learning_rate": 6.008592242320859e-05, "loss": 2.7556, "step": 23768 }, { "epoch": 1.4755105841455087, "grad_norm": 0.6064000750145954, "learning_rate": 6.008238511283784e-05, "loss": 2.8037, "step": 23769 }, { "epoch": 1.4755726612452666, "grad_norm": 0.5682552543779855, "learning_rate": 6.007884774986481e-05, "loss": 2.7469, "step": 23770 }, { "epoch": 1.4756347383450246, "grad_norm": 0.5960428738662387, "learning_rate": 6.007531033430801e-05, "loss": 2.7326, "step": 23771 }, { "epoch": 1.4756968154447825, "grad_norm": 0.5781955675925728, "learning_rate": 6.0071772866185846e-05, "loss": 2.8015, "step": 23772 }, { "epoch": 1.4757588925445404, "grad_norm": 0.5793979651746219, "learning_rate": 6.006823534551681e-05, "loss": 2.7543, "step": 23773 }, { "epoch": 1.4758209696442983, "grad_norm": 0.5520463577220146, "learning_rate": 6.0064697772319355e-05, "loss": 2.7398, "step": 23774 }, { "epoch": 1.4758830467440562, "grad_norm": 0.5825195824110675, "learning_rate": 6.0061160146611914e-05, "loss": 2.8085, "step": 23775 }, { "epoch": 1.4759451238438142, "grad_norm": 0.6472993536882431, "learning_rate": 6.005762246841298e-05, "loss": 2.8194, "step": 23776 }, { "epoch": 1.4760072009435719, "grad_norm": 0.61309855467278, "learning_rate": 6.0054084737740965e-05, "loss": 2.7806, "step": 23777 }, { "epoch": 1.4760692780433298, "grad_norm": 0.626443554820524, "learning_rate": 6.005054695461436e-05, "loss": 2.8321, "step": 23778 }, { "epoch": 1.4761313551430877, "grad_norm": 0.5904273878248973, "learning_rate": 6.00470091190516e-05, "loss": 2.7732, "step": 23779 }, { "epoch": 1.4761934322428456, "grad_norm": 0.5743871947811018, "learning_rate": 6.004347123107117e-05, "loss": 2.8463, "step": 23780 }, { "epoch": 1.4762555093426035, "grad_norm": 0.6142699660416431, "learning_rate": 6.003993329069152e-05, "loss": 2.8431, "step": 23781 }, { "epoch": 1.4763175864423614, "grad_norm": 0.6031663989154059, "learning_rate": 6.003639529793108e-05, "loss": 2.7407, "step": 23782 }, { "epoch": 1.4763796635421194, "grad_norm": 0.6217854564743215, "learning_rate": 6.003285725280834e-05, "loss": 2.8607, "step": 23783 }, { "epoch": 1.4764417406418773, "grad_norm": 0.6149551242257256, "learning_rate": 6.002931915534173e-05, "loss": 2.7274, "step": 23784 }, { "epoch": 1.4765038177416352, "grad_norm": 0.6483712219928364, "learning_rate": 6.0025781005549765e-05, "loss": 2.7571, "step": 23785 }, { "epoch": 1.476565894841393, "grad_norm": 0.678776206850673, "learning_rate": 6.0022242803450836e-05, "loss": 2.7817, "step": 23786 }, { "epoch": 1.4766279719411508, "grad_norm": 0.690852940602866, "learning_rate": 6.0018704549063446e-05, "loss": 2.6739, "step": 23787 }, { "epoch": 1.4766900490409087, "grad_norm": 0.6752883789612641, "learning_rate": 6.001516624240603e-05, "loss": 2.7409, "step": 23788 }, { "epoch": 1.4767521261406666, "grad_norm": 0.6560132417025653, "learning_rate": 6.001162788349706e-05, "loss": 2.8378, "step": 23789 }, { "epoch": 1.4768142032404246, "grad_norm": 0.6656136516266289, "learning_rate": 6.000808947235501e-05, "loss": 2.7114, "step": 23790 }, { "epoch": 1.4768762803401825, "grad_norm": 0.6500067050294573, "learning_rate": 6.000455100899831e-05, "loss": 2.793, "step": 23791 }, { "epoch": 1.4769383574399404, "grad_norm": 0.5763930591243444, "learning_rate": 6.0001012493445464e-05, "loss": 2.7311, "step": 23792 }, { "epoch": 1.4770004345396983, "grad_norm": 0.6673417746073202, "learning_rate": 5.9997473925714884e-05, "loss": 2.8541, "step": 23793 }, { "epoch": 1.4770625116394562, "grad_norm": 0.5063292335388639, "learning_rate": 5.999393530582508e-05, "loss": 2.8382, "step": 23794 }, { "epoch": 1.4771245887392142, "grad_norm": 0.694658100056895, "learning_rate": 5.9990396633794466e-05, "loss": 2.8412, "step": 23795 }, { "epoch": 1.477186665838972, "grad_norm": 0.5619818632280944, "learning_rate": 5.9986857909641546e-05, "loss": 2.7225, "step": 23796 }, { "epoch": 1.47724874293873, "grad_norm": 0.6953696109591022, "learning_rate": 5.998331913338475e-05, "loss": 2.8494, "step": 23797 }, { "epoch": 1.477310820038488, "grad_norm": 0.6607137955291874, "learning_rate": 5.997978030504257e-05, "loss": 2.8206, "step": 23798 }, { "epoch": 1.4773728971382458, "grad_norm": 0.677215641494974, "learning_rate": 5.997624142463344e-05, "loss": 2.8565, "step": 23799 }, { "epoch": 1.4774349742380035, "grad_norm": 0.6295662017686169, "learning_rate": 5.9972702492175834e-05, "loss": 2.8526, "step": 23800 }, { "epoch": 1.4774970513377614, "grad_norm": 0.647440131818333, "learning_rate": 5.996916350768823e-05, "loss": 2.7651, "step": 23801 }, { "epoch": 1.4775591284375194, "grad_norm": 0.6359877573535404, "learning_rate": 5.996562447118907e-05, "loss": 2.8096, "step": 23802 }, { "epoch": 1.4776212055372773, "grad_norm": 0.5982578964238321, "learning_rate": 5.996208538269684e-05, "loss": 2.7809, "step": 23803 }, { "epoch": 1.4776832826370352, "grad_norm": 0.6570105459827119, "learning_rate": 5.9958546242229974e-05, "loss": 2.7741, "step": 23804 }, { "epoch": 1.477745359736793, "grad_norm": 0.6002166853481081, "learning_rate": 5.9955007049806964e-05, "loss": 2.9011, "step": 23805 }, { "epoch": 1.477807436836551, "grad_norm": 0.633891751238978, "learning_rate": 5.995146780544627e-05, "loss": 2.8187, "step": 23806 }, { "epoch": 1.477869513936309, "grad_norm": 0.5486255964413693, "learning_rate": 5.994792850916635e-05, "loss": 2.8753, "step": 23807 }, { "epoch": 1.4779315910360669, "grad_norm": 0.5894479244500773, "learning_rate": 5.994438916098567e-05, "loss": 2.6988, "step": 23808 }, { "epoch": 1.4779936681358246, "grad_norm": 0.5624673357062577, "learning_rate": 5.994084976092269e-05, "loss": 2.8194, "step": 23809 }, { "epoch": 1.4780557452355825, "grad_norm": 0.6641756971033311, "learning_rate": 5.99373103089959e-05, "loss": 2.8904, "step": 23810 }, { "epoch": 1.4781178223353404, "grad_norm": 0.5324820928741583, "learning_rate": 5.993377080522374e-05, "loss": 2.7918, "step": 23811 }, { "epoch": 1.4781798994350983, "grad_norm": 0.5768082805582138, "learning_rate": 5.993023124962469e-05, "loss": 2.7563, "step": 23812 }, { "epoch": 1.4782419765348562, "grad_norm": 0.5084449644191448, "learning_rate": 5.99266916422172e-05, "loss": 2.8166, "step": 23813 }, { "epoch": 1.4783040536346141, "grad_norm": 0.5765813633177146, "learning_rate": 5.992315198301978e-05, "loss": 2.82, "step": 23814 }, { "epoch": 1.478366130734372, "grad_norm": 0.5696495069430835, "learning_rate": 5.991961227205084e-05, "loss": 2.7661, "step": 23815 }, { "epoch": 1.47842820783413, "grad_norm": 0.5602643567075246, "learning_rate": 5.991607250932889e-05, "loss": 2.6895, "step": 23816 }, { "epoch": 1.478490284933888, "grad_norm": 0.5804603165312955, "learning_rate": 5.991253269487238e-05, "loss": 2.7848, "step": 23817 }, { "epoch": 1.4785523620336458, "grad_norm": 0.610208088605716, "learning_rate": 5.990899282869977e-05, "loss": 2.7655, "step": 23818 }, { "epoch": 1.4786144391334037, "grad_norm": 0.5843929532807071, "learning_rate": 5.990545291082955e-05, "loss": 2.7206, "step": 23819 }, { "epoch": 1.4786765162331617, "grad_norm": 0.6197146814014722, "learning_rate": 5.990191294128017e-05, "loss": 2.8103, "step": 23820 }, { "epoch": 1.4787385933329196, "grad_norm": 0.6204862682719696, "learning_rate": 5.9898372920070104e-05, "loss": 2.7574, "step": 23821 }, { "epoch": 1.4788006704326775, "grad_norm": 0.6631556658256211, "learning_rate": 5.989483284721782e-05, "loss": 2.7939, "step": 23822 }, { "epoch": 1.4788627475324354, "grad_norm": 0.5205208095051524, "learning_rate": 5.989129272274181e-05, "loss": 2.8149, "step": 23823 }, { "epoch": 1.478924824632193, "grad_norm": 0.6627798595961563, "learning_rate": 5.988775254666052e-05, "loss": 2.8697, "step": 23824 }, { "epoch": 1.478986901731951, "grad_norm": 0.5710814530787772, "learning_rate": 5.98842123189924e-05, "loss": 2.7727, "step": 23825 }, { "epoch": 1.479048978831709, "grad_norm": 0.6193240883667245, "learning_rate": 5.988067203975598e-05, "loss": 2.6927, "step": 23826 }, { "epoch": 1.4791110559314669, "grad_norm": 0.5686776754484625, "learning_rate": 5.987713170896967e-05, "loss": 2.7392, "step": 23827 }, { "epoch": 1.4791731330312248, "grad_norm": 0.5785837157860482, "learning_rate": 5.987359132665198e-05, "loss": 2.7358, "step": 23828 }, { "epoch": 1.4792352101309827, "grad_norm": 0.6020897127977285, "learning_rate": 5.9870050892821364e-05, "loss": 2.7211, "step": 23829 }, { "epoch": 1.4792972872307406, "grad_norm": 0.6010034118149277, "learning_rate": 5.986651040749629e-05, "loss": 2.8031, "step": 23830 }, { "epoch": 1.4793593643304985, "grad_norm": 0.6090696499024656, "learning_rate": 5.986296987069524e-05, "loss": 2.8009, "step": 23831 }, { "epoch": 1.4794214414302564, "grad_norm": 0.5066403285299947, "learning_rate": 5.985942928243668e-05, "loss": 2.7501, "step": 23832 }, { "epoch": 1.4794835185300141, "grad_norm": 0.6334842366166972, "learning_rate": 5.985588864273909e-05, "loss": 2.8069, "step": 23833 }, { "epoch": 1.479545595629772, "grad_norm": 0.5112166525823791, "learning_rate": 5.9852347951620935e-05, "loss": 2.7872, "step": 23834 }, { "epoch": 1.47960767272953, "grad_norm": 0.6098805555671903, "learning_rate": 5.984880720910069e-05, "loss": 2.8624, "step": 23835 }, { "epoch": 1.479669749829288, "grad_norm": 0.5708241975939633, "learning_rate": 5.984526641519682e-05, "loss": 2.7975, "step": 23836 }, { "epoch": 1.4797318269290458, "grad_norm": 0.4979607451519373, "learning_rate": 5.984172556992782e-05, "loss": 2.8154, "step": 23837 }, { "epoch": 1.4797939040288037, "grad_norm": 0.5687739377695845, "learning_rate": 5.9838184673312116e-05, "loss": 2.875, "step": 23838 }, { "epoch": 1.4798559811285616, "grad_norm": 0.5480574494671965, "learning_rate": 5.983464372536825e-05, "loss": 2.8077, "step": 23839 }, { "epoch": 1.4799180582283196, "grad_norm": 0.5278454972065972, "learning_rate": 5.983110272611463e-05, "loss": 2.8549, "step": 23840 }, { "epoch": 1.4799801353280775, "grad_norm": 0.5041060532107887, "learning_rate": 5.9827561675569785e-05, "loss": 2.8911, "step": 23841 }, { "epoch": 1.4800422124278354, "grad_norm": 0.5485333774586028, "learning_rate": 5.982402057375216e-05, "loss": 2.7487, "step": 23842 }, { "epoch": 1.4801042895275933, "grad_norm": 0.5064928661401377, "learning_rate": 5.982047942068023e-05, "loss": 2.7624, "step": 23843 }, { "epoch": 1.4801663666273512, "grad_norm": 0.5527531381394614, "learning_rate": 5.981693821637249e-05, "loss": 2.7789, "step": 23844 }, { "epoch": 1.4802284437271092, "grad_norm": 0.46896961153970196, "learning_rate": 5.981339696084737e-05, "loss": 2.7026, "step": 23845 }, { "epoch": 1.480290520826867, "grad_norm": 0.5758839809329332, "learning_rate": 5.9809855654123404e-05, "loss": 2.7488, "step": 23846 }, { "epoch": 1.480352597926625, "grad_norm": 0.5138085477189985, "learning_rate": 5.980631429621901e-05, "loss": 2.7927, "step": 23847 }, { "epoch": 1.4804146750263827, "grad_norm": 0.5190461267211962, "learning_rate": 5.980277288715272e-05, "loss": 2.8242, "step": 23848 }, { "epoch": 1.4804767521261406, "grad_norm": 0.5471814795423763, "learning_rate": 5.979923142694297e-05, "loss": 2.8629, "step": 23849 }, { "epoch": 1.4805388292258985, "grad_norm": 0.4942281291215219, "learning_rate": 5.9795689915608264e-05, "loss": 2.7942, "step": 23850 }, { "epoch": 1.4806009063256564, "grad_norm": 0.5701363696549773, "learning_rate": 5.979214835316706e-05, "loss": 2.8479, "step": 23851 }, { "epoch": 1.4806629834254144, "grad_norm": 0.5535085264408972, "learning_rate": 5.9788606739637845e-05, "loss": 2.7753, "step": 23852 }, { "epoch": 1.4807250605251723, "grad_norm": 0.6247688581300256, "learning_rate": 5.978506507503907e-05, "loss": 2.9141, "step": 23853 }, { "epoch": 1.4807871376249302, "grad_norm": 0.5832618402766387, "learning_rate": 5.978152335938925e-05, "loss": 2.7963, "step": 23854 }, { "epoch": 1.4808492147246881, "grad_norm": 0.6171264135334098, "learning_rate": 5.977798159270686e-05, "loss": 2.8415, "step": 23855 }, { "epoch": 1.480911291824446, "grad_norm": 0.5806085492648215, "learning_rate": 5.977443977501035e-05, "loss": 2.8105, "step": 23856 }, { "epoch": 1.4809733689242037, "grad_norm": 0.6769161286895704, "learning_rate": 5.9770897906318226e-05, "loss": 2.8607, "step": 23857 }, { "epoch": 1.4810354460239616, "grad_norm": 0.6099156157958009, "learning_rate": 5.9767355986648944e-05, "loss": 2.8628, "step": 23858 }, { "epoch": 1.4810975231237196, "grad_norm": 0.6030517326087185, "learning_rate": 5.9763814016021e-05, "loss": 2.8012, "step": 23859 }, { "epoch": 1.4811596002234775, "grad_norm": 0.6163817107098921, "learning_rate": 5.9760271994452864e-05, "loss": 2.8361, "step": 23860 }, { "epoch": 1.4812216773232354, "grad_norm": 0.6812438903324644, "learning_rate": 5.9756729921963036e-05, "loss": 2.826, "step": 23861 }, { "epoch": 1.4812837544229933, "grad_norm": 0.5477787180188156, "learning_rate": 5.975318779856998e-05, "loss": 2.8107, "step": 23862 }, { "epoch": 1.4813458315227512, "grad_norm": 0.600471448453011, "learning_rate": 5.974964562429215e-05, "loss": 2.7768, "step": 23863 }, { "epoch": 1.4814079086225092, "grad_norm": 0.5416484945868087, "learning_rate": 5.9746103399148076e-05, "loss": 2.8485, "step": 23864 }, { "epoch": 1.481469985722267, "grad_norm": 0.5392607448764808, "learning_rate": 5.974256112315619e-05, "loss": 2.8224, "step": 23865 }, { "epoch": 1.481532062822025, "grad_norm": 0.5175756778299312, "learning_rate": 5.9739018796335025e-05, "loss": 2.7597, "step": 23866 }, { "epoch": 1.481594139921783, "grad_norm": 0.5401811464556825, "learning_rate": 5.973547641870302e-05, "loss": 2.7818, "step": 23867 }, { "epoch": 1.4816562170215408, "grad_norm": 0.5337388418045282, "learning_rate": 5.9731933990278673e-05, "loss": 2.7218, "step": 23868 }, { "epoch": 1.4817182941212987, "grad_norm": 0.5798426191791269, "learning_rate": 5.972839151108046e-05, "loss": 2.8723, "step": 23869 }, { "epoch": 1.4817803712210567, "grad_norm": 0.553667904427944, "learning_rate": 5.972484898112687e-05, "loss": 2.8874, "step": 23870 }, { "epoch": 1.4818424483208146, "grad_norm": 0.5612577921235128, "learning_rate": 5.972130640043638e-05, "loss": 2.7646, "step": 23871 }, { "epoch": 1.4819045254205723, "grad_norm": 0.5428702175835314, "learning_rate": 5.971776376902748e-05, "loss": 2.8423, "step": 23872 }, { "epoch": 1.4819666025203302, "grad_norm": 0.5740107983693271, "learning_rate": 5.971422108691864e-05, "loss": 2.8174, "step": 23873 }, { "epoch": 1.482028679620088, "grad_norm": 0.5446754089606765, "learning_rate": 5.971067835412836e-05, "loss": 2.7515, "step": 23874 }, { "epoch": 1.482090756719846, "grad_norm": 0.6117764130328207, "learning_rate": 5.970713557067511e-05, "loss": 2.7245, "step": 23875 }, { "epoch": 1.482152833819604, "grad_norm": 0.5742197101610941, "learning_rate": 5.970359273657737e-05, "loss": 2.7758, "step": 23876 }, { "epoch": 1.4822149109193619, "grad_norm": 0.5804947663711155, "learning_rate": 5.9700049851853644e-05, "loss": 2.784, "step": 23877 }, { "epoch": 1.4822769880191198, "grad_norm": 0.5622214581346576, "learning_rate": 5.969650691652239e-05, "loss": 2.7241, "step": 23878 }, { "epoch": 1.4823390651188777, "grad_norm": 0.5909994815252906, "learning_rate": 5.969296393060212e-05, "loss": 2.6977, "step": 23879 }, { "epoch": 1.4824011422186356, "grad_norm": 0.6175439988032564, "learning_rate": 5.968942089411129e-05, "loss": 2.8878, "step": 23880 }, { "epoch": 1.4824632193183933, "grad_norm": 0.5593219358965661, "learning_rate": 5.96858778070684e-05, "loss": 2.8167, "step": 23881 }, { "epoch": 1.4825252964181512, "grad_norm": 0.6220164852119616, "learning_rate": 5.9682334669491944e-05, "loss": 2.8041, "step": 23882 }, { "epoch": 1.4825873735179091, "grad_norm": 0.5656681644977722, "learning_rate": 5.96787914814004e-05, "loss": 2.6722, "step": 23883 }, { "epoch": 1.482649450617667, "grad_norm": 0.5850415819012853, "learning_rate": 5.9675248242812246e-05, "loss": 2.9218, "step": 23884 }, { "epoch": 1.482711527717425, "grad_norm": 0.5859404754563123, "learning_rate": 5.9671704953745964e-05, "loss": 2.8438, "step": 23885 }, { "epoch": 1.482773604817183, "grad_norm": 0.5758428636398032, "learning_rate": 5.9668161614220066e-05, "loss": 2.7546, "step": 23886 }, { "epoch": 1.4828356819169408, "grad_norm": 0.5826566267843504, "learning_rate": 5.9664618224252997e-05, "loss": 2.7284, "step": 23887 }, { "epoch": 1.4828977590166987, "grad_norm": 0.5209667766012641, "learning_rate": 5.966107478386328e-05, "loss": 2.6612, "step": 23888 }, { "epoch": 1.4829598361164567, "grad_norm": 0.6047008690354196, "learning_rate": 5.96575312930694e-05, "loss": 2.8423, "step": 23889 }, { "epoch": 1.4830219132162146, "grad_norm": 0.6058571338919553, "learning_rate": 5.965398775188983e-05, "loss": 2.7101, "step": 23890 }, { "epoch": 1.4830839903159725, "grad_norm": 0.6191216113904475, "learning_rate": 5.965044416034307e-05, "loss": 2.8083, "step": 23891 }, { "epoch": 1.4831460674157304, "grad_norm": 0.6578596220558507, "learning_rate": 5.964690051844758e-05, "loss": 2.8076, "step": 23892 }, { "epoch": 1.4832081445154883, "grad_norm": 0.6130715515840538, "learning_rate": 5.964335682622189e-05, "loss": 2.7453, "step": 23893 }, { "epoch": 1.4832702216152462, "grad_norm": 0.7113639893726201, "learning_rate": 5.963981308368445e-05, "loss": 2.8819, "step": 23894 }, { "epoch": 1.4833322987150042, "grad_norm": 0.5707457738628815, "learning_rate": 5.963626929085377e-05, "loss": 2.7242, "step": 23895 }, { "epoch": 1.4833943758147619, "grad_norm": 0.7875863421503101, "learning_rate": 5.963272544774833e-05, "loss": 2.7317, "step": 23896 }, { "epoch": 1.4834564529145198, "grad_norm": 0.641226230500551, "learning_rate": 5.962918155438664e-05, "loss": 2.7702, "step": 23897 }, { "epoch": 1.4835185300142777, "grad_norm": 0.6893693910925663, "learning_rate": 5.962563761078715e-05, "loss": 2.8311, "step": 23898 }, { "epoch": 1.4835806071140356, "grad_norm": 0.6618582550322489, "learning_rate": 5.962209361696838e-05, "loss": 2.8544, "step": 23899 }, { "epoch": 1.4836426842137935, "grad_norm": 0.6312038319766919, "learning_rate": 5.9618549572948826e-05, "loss": 2.8096, "step": 23900 }, { "epoch": 1.4837047613135514, "grad_norm": 0.6374762072517594, "learning_rate": 5.961500547874694e-05, "loss": 2.8173, "step": 23901 }, { "epoch": 1.4837668384133094, "grad_norm": 0.6435092800022548, "learning_rate": 5.9611461334381255e-05, "loss": 2.8338, "step": 23902 }, { "epoch": 1.4838289155130673, "grad_norm": 0.6178234783194774, "learning_rate": 5.960791713987022e-05, "loss": 2.8443, "step": 23903 }, { "epoch": 1.4838909926128252, "grad_norm": 0.7375455955376445, "learning_rate": 5.960437289523237e-05, "loss": 2.8864, "step": 23904 }, { "epoch": 1.483953069712583, "grad_norm": 0.5710948646628464, "learning_rate": 5.960082860048617e-05, "loss": 2.8104, "step": 23905 }, { "epoch": 1.4840151468123408, "grad_norm": 0.7473939680725287, "learning_rate": 5.959728425565012e-05, "loss": 2.7942, "step": 23906 }, { "epoch": 1.4840772239120987, "grad_norm": 0.6143851822856269, "learning_rate": 5.95937398607427e-05, "loss": 2.782, "step": 23907 }, { "epoch": 1.4841393010118566, "grad_norm": 0.6605854106650813, "learning_rate": 5.959019541578241e-05, "loss": 2.823, "step": 23908 }, { "epoch": 1.4842013781116146, "grad_norm": 0.64390737221057, "learning_rate": 5.958665092078776e-05, "loss": 2.8225, "step": 23909 }, { "epoch": 1.4842634552113725, "grad_norm": 0.6080560103250593, "learning_rate": 5.95831063757772e-05, "loss": 2.8408, "step": 23910 }, { "epoch": 1.4843255323111304, "grad_norm": 0.6513110455477634, "learning_rate": 5.957956178076927e-05, "loss": 2.8399, "step": 23911 }, { "epoch": 1.4843876094108883, "grad_norm": 0.5754937529485842, "learning_rate": 5.9576017135782424e-05, "loss": 2.6995, "step": 23912 }, { "epoch": 1.4844496865106462, "grad_norm": 0.5916137663828359, "learning_rate": 5.957247244083518e-05, "loss": 2.6492, "step": 23913 }, { "epoch": 1.4845117636104042, "grad_norm": 0.5499460096302404, "learning_rate": 5.956892769594602e-05, "loss": 2.7316, "step": 23914 }, { "epoch": 1.484573840710162, "grad_norm": 0.672681634758474, "learning_rate": 5.956538290113345e-05, "loss": 2.7174, "step": 23915 }, { "epoch": 1.48463591780992, "grad_norm": 0.5512606381019138, "learning_rate": 5.956183805641595e-05, "loss": 2.8082, "step": 23916 }, { "epoch": 1.484697994909678, "grad_norm": 0.7027361960437838, "learning_rate": 5.955829316181202e-05, "loss": 2.8064, "step": 23917 }, { "epoch": 1.4847600720094358, "grad_norm": 0.5588815921146383, "learning_rate": 5.955474821734015e-05, "loss": 2.7463, "step": 23918 }, { "epoch": 1.4848221491091937, "grad_norm": 0.6147813855040508, "learning_rate": 5.955120322301882e-05, "loss": 2.7422, "step": 23919 }, { "epoch": 1.4848842262089514, "grad_norm": 0.5986462789144119, "learning_rate": 5.954765817886658e-05, "loss": 2.734, "step": 23920 }, { "epoch": 1.4849463033087094, "grad_norm": 0.5432862650146378, "learning_rate": 5.9544113084901867e-05, "loss": 2.7511, "step": 23921 }, { "epoch": 1.4850083804084673, "grad_norm": 0.6175962133008616, "learning_rate": 5.954056794114321e-05, "loss": 2.8208, "step": 23922 }, { "epoch": 1.4850704575082252, "grad_norm": 0.5395222237740874, "learning_rate": 5.95370227476091e-05, "loss": 2.8344, "step": 23923 }, { "epoch": 1.4851325346079831, "grad_norm": 0.6095690051611437, "learning_rate": 5.9533477504318015e-05, "loss": 2.7638, "step": 23924 }, { "epoch": 1.485194611707741, "grad_norm": 0.6231656574158795, "learning_rate": 5.952993221128847e-05, "loss": 2.7804, "step": 23925 }, { "epoch": 1.485256688807499, "grad_norm": 0.6239378486883761, "learning_rate": 5.952638686853894e-05, "loss": 2.8211, "step": 23926 }, { "epoch": 1.4853187659072569, "grad_norm": 0.648825058536529, "learning_rate": 5.952284147608795e-05, "loss": 2.8024, "step": 23927 }, { "epoch": 1.4853808430070148, "grad_norm": 0.6304871491461939, "learning_rate": 5.951929603395399e-05, "loss": 2.8011, "step": 23928 }, { "epoch": 1.4854429201067725, "grad_norm": 0.6719985005625958, "learning_rate": 5.951575054215555e-05, "loss": 2.7819, "step": 23929 }, { "epoch": 1.4855049972065304, "grad_norm": 0.6394825028537889, "learning_rate": 5.951220500071112e-05, "loss": 2.7462, "step": 23930 }, { "epoch": 1.4855670743062883, "grad_norm": 0.5999578153165255, "learning_rate": 5.950865940963921e-05, "loss": 2.8137, "step": 23931 }, { "epoch": 1.4856291514060462, "grad_norm": 0.6728167476422521, "learning_rate": 5.950511376895831e-05, "loss": 2.8206, "step": 23932 }, { "epoch": 1.4856912285058042, "grad_norm": 0.6309132036327281, "learning_rate": 5.950156807868694e-05, "loss": 2.8174, "step": 23933 }, { "epoch": 1.485753305605562, "grad_norm": 0.6468355083850023, "learning_rate": 5.949802233884358e-05, "loss": 2.7861, "step": 23934 }, { "epoch": 1.48581538270532, "grad_norm": 0.6762135616787274, "learning_rate": 5.949447654944672e-05, "loss": 2.8376, "step": 23935 }, { "epoch": 1.485877459805078, "grad_norm": 0.57196521360151, "learning_rate": 5.949093071051487e-05, "loss": 2.8319, "step": 23936 }, { "epoch": 1.4859395369048358, "grad_norm": 0.6445562184439315, "learning_rate": 5.948738482206653e-05, "loss": 2.7314, "step": 23937 }, { "epoch": 1.4860016140045937, "grad_norm": 0.5658956133267149, "learning_rate": 5.9483838884120216e-05, "loss": 2.7084, "step": 23938 }, { "epoch": 1.4860636911043517, "grad_norm": 0.6383142432436758, "learning_rate": 5.94802928966944e-05, "loss": 2.7805, "step": 23939 }, { "epoch": 1.4861257682041096, "grad_norm": 0.6129814921816422, "learning_rate": 5.94767468598076e-05, "loss": 2.7903, "step": 23940 }, { "epoch": 1.4861878453038675, "grad_norm": 0.6080532656209661, "learning_rate": 5.947320077347831e-05, "loss": 2.8009, "step": 23941 }, { "epoch": 1.4862499224036254, "grad_norm": 0.6081696106013667, "learning_rate": 5.946965463772504e-05, "loss": 2.8014, "step": 23942 }, { "epoch": 1.4863119995033833, "grad_norm": 0.5507643947321894, "learning_rate": 5.946610845256627e-05, "loss": 2.829, "step": 23943 }, { "epoch": 1.486374076603141, "grad_norm": 0.6282365206506922, "learning_rate": 5.946256221802051e-05, "loss": 2.8147, "step": 23944 }, { "epoch": 1.486436153702899, "grad_norm": 0.6091231045582897, "learning_rate": 5.945901593410629e-05, "loss": 2.8206, "step": 23945 }, { "epoch": 1.4864982308026569, "grad_norm": 0.6577012500335196, "learning_rate": 5.9455469600842064e-05, "loss": 2.8544, "step": 23946 }, { "epoch": 1.4865603079024148, "grad_norm": 0.5936424760332528, "learning_rate": 5.945192321824636e-05, "loss": 2.8161, "step": 23947 }, { "epoch": 1.4866223850021727, "grad_norm": 0.6187449734416202, "learning_rate": 5.9448376786337676e-05, "loss": 2.7737, "step": 23948 }, { "epoch": 1.4866844621019306, "grad_norm": 0.6206159613297829, "learning_rate": 5.944483030513454e-05, "loss": 2.7409, "step": 23949 }, { "epoch": 1.4867465392016885, "grad_norm": 0.6408916709209285, "learning_rate": 5.944128377465541e-05, "loss": 2.8656, "step": 23950 }, { "epoch": 1.4868086163014465, "grad_norm": 0.6777876027777, "learning_rate": 5.943773719491882e-05, "loss": 2.8879, "step": 23951 }, { "epoch": 1.4868706934012044, "grad_norm": 0.6517402461367936, "learning_rate": 5.943419056594326e-05, "loss": 2.7697, "step": 23952 }, { "epoch": 1.486932770500962, "grad_norm": 0.6539131975431541, "learning_rate": 5.943064388774723e-05, "loss": 2.7961, "step": 23953 }, { "epoch": 1.48699484760072, "grad_norm": 0.6082839980652508, "learning_rate": 5.942709716034926e-05, "loss": 2.876, "step": 23954 }, { "epoch": 1.487056924700478, "grad_norm": 0.7292137584920046, "learning_rate": 5.9423550383767814e-05, "loss": 2.8277, "step": 23955 }, { "epoch": 1.4871190018002358, "grad_norm": 0.6342046753307725, "learning_rate": 5.942000355802143e-05, "loss": 2.709, "step": 23956 }, { "epoch": 1.4871810788999937, "grad_norm": 0.7145204073352098, "learning_rate": 5.94164566831286e-05, "loss": 2.7618, "step": 23957 }, { "epoch": 1.4872431559997517, "grad_norm": 0.6830725072252795, "learning_rate": 5.9412909759107823e-05, "loss": 2.758, "step": 23958 }, { "epoch": 1.4873052330995096, "grad_norm": 0.5974050754496263, "learning_rate": 5.940936278597763e-05, "loss": 2.732, "step": 23959 }, { "epoch": 1.4873673101992675, "grad_norm": 0.7100117703254657, "learning_rate": 5.940581576375649e-05, "loss": 2.7836, "step": 23960 }, { "epoch": 1.4874293872990254, "grad_norm": 0.6465384930320135, "learning_rate": 5.9402268692462924e-05, "loss": 2.7919, "step": 23961 }, { "epoch": 1.4874914643987833, "grad_norm": 0.7135155016331893, "learning_rate": 5.939872157211544e-05, "loss": 2.7452, "step": 23962 }, { "epoch": 1.4875535414985412, "grad_norm": 0.6424359103347463, "learning_rate": 5.939517440273257e-05, "loss": 2.6641, "step": 23963 }, { "epoch": 1.4876156185982992, "grad_norm": 0.6187853109514403, "learning_rate": 5.9391627184332756e-05, "loss": 2.7875, "step": 23964 }, { "epoch": 1.487677695698057, "grad_norm": 0.607736953512151, "learning_rate": 5.938807991693457e-05, "loss": 2.7724, "step": 23965 }, { "epoch": 1.487739772797815, "grad_norm": 0.6554182094508123, "learning_rate": 5.938453260055648e-05, "loss": 2.8389, "step": 23966 }, { "epoch": 1.487801849897573, "grad_norm": 0.6374491381457528, "learning_rate": 5.938098523521701e-05, "loss": 2.7337, "step": 23967 }, { "epoch": 1.4878639269973306, "grad_norm": 0.6363329976775195, "learning_rate": 5.9377437820934656e-05, "loss": 2.8051, "step": 23968 }, { "epoch": 1.4879260040970885, "grad_norm": 0.6342533803029513, "learning_rate": 5.9373890357727936e-05, "loss": 2.786, "step": 23969 }, { "epoch": 1.4879880811968464, "grad_norm": 0.590088547188169, "learning_rate": 5.937034284561536e-05, "loss": 2.7005, "step": 23970 }, { "epoch": 1.4880501582966044, "grad_norm": 0.6582393692859068, "learning_rate": 5.936679528461542e-05, "loss": 2.7391, "step": 23971 }, { "epoch": 1.4881122353963623, "grad_norm": 0.5830520280847721, "learning_rate": 5.9363247674746645e-05, "loss": 2.86, "step": 23972 }, { "epoch": 1.4881743124961202, "grad_norm": 0.6256489722263647, "learning_rate": 5.935970001602752e-05, "loss": 2.8409, "step": 23973 }, { "epoch": 1.4882363895958781, "grad_norm": 0.6145610212597087, "learning_rate": 5.935615230847659e-05, "loss": 2.6805, "step": 23974 }, { "epoch": 1.488298466695636, "grad_norm": 0.598222850421036, "learning_rate": 5.935260455211232e-05, "loss": 2.7553, "step": 23975 }, { "epoch": 1.488360543795394, "grad_norm": 0.6853917916450064, "learning_rate": 5.934905674695326e-05, "loss": 2.8127, "step": 23976 }, { "epoch": 1.4884226208951516, "grad_norm": 0.5787091783335248, "learning_rate": 5.9345508893017896e-05, "loss": 2.7442, "step": 23977 }, { "epoch": 1.4884846979949096, "grad_norm": 0.6448721244379666, "learning_rate": 5.934196099032474e-05, "loss": 2.7876, "step": 23978 }, { "epoch": 1.4885467750946675, "grad_norm": 0.710856254881868, "learning_rate": 5.933841303889231e-05, "loss": 2.8327, "step": 23979 }, { "epoch": 1.4886088521944254, "grad_norm": 0.6349479530811547, "learning_rate": 5.9334865038739104e-05, "loss": 2.7886, "step": 23980 }, { "epoch": 1.4886709292941833, "grad_norm": 0.7480093047664864, "learning_rate": 5.933131698988365e-05, "loss": 2.7519, "step": 23981 }, { "epoch": 1.4887330063939412, "grad_norm": 0.6340649067704729, "learning_rate": 5.9327768892344437e-05, "loss": 2.8334, "step": 23982 }, { "epoch": 1.4887950834936992, "grad_norm": 0.6878335533750253, "learning_rate": 5.9324220746140005e-05, "loss": 2.8318, "step": 23983 }, { "epoch": 1.488857160593457, "grad_norm": 0.6388901503753187, "learning_rate": 5.932067255128884e-05, "loss": 2.7767, "step": 23984 }, { "epoch": 1.488919237693215, "grad_norm": 0.6026301384694702, "learning_rate": 5.9317124307809456e-05, "loss": 2.8067, "step": 23985 }, { "epoch": 1.488981314792973, "grad_norm": 0.690564154624424, "learning_rate": 5.93135760157204e-05, "loss": 2.8239, "step": 23986 }, { "epoch": 1.4890433918927308, "grad_norm": 0.5438880273867132, "learning_rate": 5.931002767504012e-05, "loss": 2.8457, "step": 23987 }, { "epoch": 1.4891054689924887, "grad_norm": 0.6855273773811545, "learning_rate": 5.9306479285787197e-05, "loss": 2.8165, "step": 23988 }, { "epoch": 1.4891675460922467, "grad_norm": 0.5934446453128346, "learning_rate": 5.93029308479801e-05, "loss": 2.7508, "step": 23989 }, { "epoch": 1.4892296231920046, "grad_norm": 0.6954022038919383, "learning_rate": 5.929938236163736e-05, "loss": 2.8252, "step": 23990 }, { "epoch": 1.4892917002917625, "grad_norm": 0.5962417168911607, "learning_rate": 5.929583382677747e-05, "loss": 2.8824, "step": 23991 }, { "epoch": 1.4893537773915202, "grad_norm": 0.5744042907734559, "learning_rate": 5.9292285243418974e-05, "loss": 2.7444, "step": 23992 }, { "epoch": 1.4894158544912781, "grad_norm": 0.6123562537082577, "learning_rate": 5.928873661158035e-05, "loss": 2.7608, "step": 23993 }, { "epoch": 1.489477931591036, "grad_norm": 0.5357930091508639, "learning_rate": 5.928518793128016e-05, "loss": 2.8366, "step": 23994 }, { "epoch": 1.489540008690794, "grad_norm": 0.6355271920765213, "learning_rate": 5.9281639202536886e-05, "loss": 2.8238, "step": 23995 }, { "epoch": 1.4896020857905519, "grad_norm": 0.5569917608658495, "learning_rate": 5.9278090425369026e-05, "loss": 2.7983, "step": 23996 }, { "epoch": 1.4896641628903098, "grad_norm": 0.5607824329056571, "learning_rate": 5.927454159979513e-05, "loss": 2.8257, "step": 23997 }, { "epoch": 1.4897262399900677, "grad_norm": 0.594542376179795, "learning_rate": 5.927099272583369e-05, "loss": 2.8663, "step": 23998 }, { "epoch": 1.4897883170898256, "grad_norm": 0.6070674095037761, "learning_rate": 5.926744380350325e-05, "loss": 2.7662, "step": 23999 }, { "epoch": 1.4898503941895835, "grad_norm": 0.5908587482655364, "learning_rate": 5.9263894832822286e-05, "loss": 2.8234, "step": 24000 }, { "epoch": 1.4899124712893412, "grad_norm": 0.6366732060344585, "learning_rate": 5.9260345813809336e-05, "loss": 2.8584, "step": 24001 }, { "epoch": 1.4899745483890992, "grad_norm": 0.6396857563809409, "learning_rate": 5.925679674648292e-05, "loss": 2.7651, "step": 24002 }, { "epoch": 1.490036625488857, "grad_norm": 0.5643804166751447, "learning_rate": 5.925324763086155e-05, "loss": 2.7658, "step": 24003 }, { "epoch": 1.490098702588615, "grad_norm": 0.6888740636548895, "learning_rate": 5.9249698466963745e-05, "loss": 2.6755, "step": 24004 }, { "epoch": 1.490160779688373, "grad_norm": 0.5734068651877422, "learning_rate": 5.9246149254808005e-05, "loss": 2.7353, "step": 24005 }, { "epoch": 1.4902228567881308, "grad_norm": 0.6620408869517538, "learning_rate": 5.924259999441286e-05, "loss": 2.7469, "step": 24006 }, { "epoch": 1.4902849338878887, "grad_norm": 0.6755832536417445, "learning_rate": 5.9239050685796834e-05, "loss": 2.7334, "step": 24007 }, { "epoch": 1.4903470109876467, "grad_norm": 0.5888036437409713, "learning_rate": 5.923550132897844e-05, "loss": 2.7911, "step": 24008 }, { "epoch": 1.4904090880874046, "grad_norm": 0.6813268731867617, "learning_rate": 5.9231951923976184e-05, "loss": 2.8058, "step": 24009 }, { "epoch": 1.4904711651871625, "grad_norm": 0.6407635236630914, "learning_rate": 5.922840247080861e-05, "loss": 2.7791, "step": 24010 }, { "epoch": 1.4905332422869204, "grad_norm": 0.5988833228424504, "learning_rate": 5.9224852969494204e-05, "loss": 2.8061, "step": 24011 }, { "epoch": 1.4905953193866783, "grad_norm": 0.7232304946320942, "learning_rate": 5.922130342005151e-05, "loss": 2.7652, "step": 24012 }, { "epoch": 1.4906573964864362, "grad_norm": 0.6191651252207131, "learning_rate": 5.921775382249902e-05, "loss": 2.8112, "step": 24013 }, { "epoch": 1.4907194735861942, "grad_norm": 0.6755094477729061, "learning_rate": 5.921420417685529e-05, "loss": 2.7585, "step": 24014 }, { "epoch": 1.490781550685952, "grad_norm": 0.6158731686514741, "learning_rate": 5.921065448313883e-05, "loss": 2.747, "step": 24015 }, { "epoch": 1.4908436277857098, "grad_norm": 0.6303006464496573, "learning_rate": 5.9207104741368123e-05, "loss": 2.7718, "step": 24016 }, { "epoch": 1.4909057048854677, "grad_norm": 0.5971118716624829, "learning_rate": 5.920355495156174e-05, "loss": 2.8098, "step": 24017 }, { "epoch": 1.4909677819852256, "grad_norm": 0.6462752127755768, "learning_rate": 5.920000511373813e-05, "loss": 2.7785, "step": 24018 }, { "epoch": 1.4910298590849835, "grad_norm": 0.5635662569426557, "learning_rate": 5.9196455227915904e-05, "loss": 2.711, "step": 24019 }, { "epoch": 1.4910919361847415, "grad_norm": 0.6706031027113724, "learning_rate": 5.919290529411353e-05, "loss": 2.8583, "step": 24020 }, { "epoch": 1.4911540132844994, "grad_norm": 0.580598283282379, "learning_rate": 5.9189355312349535e-05, "loss": 2.7709, "step": 24021 }, { "epoch": 1.4912160903842573, "grad_norm": 0.5954749498967944, "learning_rate": 5.9185805282642445e-05, "loss": 2.8633, "step": 24022 }, { "epoch": 1.4912781674840152, "grad_norm": 0.5968353268285483, "learning_rate": 5.9182255205010773e-05, "loss": 2.8403, "step": 24023 }, { "epoch": 1.4913402445837731, "grad_norm": 0.6215103959175434, "learning_rate": 5.917870507947305e-05, "loss": 2.8384, "step": 24024 }, { "epoch": 1.4914023216835308, "grad_norm": 0.6011293015701932, "learning_rate": 5.917515490604779e-05, "loss": 2.7875, "step": 24025 }, { "epoch": 1.4914643987832887, "grad_norm": 0.6266029544622473, "learning_rate": 5.9171604684753536e-05, "loss": 2.7263, "step": 24026 }, { "epoch": 1.4915264758830467, "grad_norm": 0.5994133893031283, "learning_rate": 5.916805441560878e-05, "loss": 2.7854, "step": 24027 }, { "epoch": 1.4915885529828046, "grad_norm": 0.628429658746954, "learning_rate": 5.916450409863206e-05, "loss": 2.8637, "step": 24028 }, { "epoch": 1.4916506300825625, "grad_norm": 0.6215385188045406, "learning_rate": 5.91609537338419e-05, "loss": 2.8779, "step": 24029 }, { "epoch": 1.4917127071823204, "grad_norm": 0.6102857630547812, "learning_rate": 5.915740332125681e-05, "loss": 2.8168, "step": 24030 }, { "epoch": 1.4917747842820783, "grad_norm": 0.6109583409649825, "learning_rate": 5.915385286089533e-05, "loss": 2.8114, "step": 24031 }, { "epoch": 1.4918368613818362, "grad_norm": 0.6270736388368543, "learning_rate": 5.915030235277599e-05, "loss": 2.8063, "step": 24032 }, { "epoch": 1.4918989384815942, "grad_norm": 0.6309000952838627, "learning_rate": 5.914675179691729e-05, "loss": 2.8637, "step": 24033 }, { "epoch": 1.491961015581352, "grad_norm": 0.6354064992706151, "learning_rate": 5.9143201193337746e-05, "loss": 2.8007, "step": 24034 }, { "epoch": 1.49202309268111, "grad_norm": 0.5778322251761019, "learning_rate": 5.913965054205594e-05, "loss": 2.7358, "step": 24035 }, { "epoch": 1.492085169780868, "grad_norm": 0.6858495495139451, "learning_rate": 5.913609984309032e-05, "loss": 2.8179, "step": 24036 }, { "epoch": 1.4921472468806258, "grad_norm": 0.5909285012805194, "learning_rate": 5.913254909645948e-05, "loss": 2.7728, "step": 24037 }, { "epoch": 1.4922093239803838, "grad_norm": 0.6884605243519247, "learning_rate": 5.91289983021819e-05, "loss": 2.7659, "step": 24038 }, { "epoch": 1.4922714010801417, "grad_norm": 0.6759175069517865, "learning_rate": 5.912544746027613e-05, "loss": 2.8373, "step": 24039 }, { "epoch": 1.4923334781798994, "grad_norm": 0.5745918722488725, "learning_rate": 5.912189657076068e-05, "loss": 2.7274, "step": 24040 }, { "epoch": 1.4923955552796573, "grad_norm": 0.6575571846842777, "learning_rate": 5.911834563365407e-05, "loss": 2.7712, "step": 24041 }, { "epoch": 1.4924576323794152, "grad_norm": 0.608210869986651, "learning_rate": 5.911479464897486e-05, "loss": 2.7954, "step": 24042 }, { "epoch": 1.4925197094791731, "grad_norm": 0.6829852420349056, "learning_rate": 5.911124361674153e-05, "loss": 2.8455, "step": 24043 }, { "epoch": 1.492581786578931, "grad_norm": 0.6027598417072745, "learning_rate": 5.910769253697265e-05, "loss": 2.8159, "step": 24044 }, { "epoch": 1.492643863678689, "grad_norm": 0.6626872913562101, "learning_rate": 5.910414140968671e-05, "loss": 2.8095, "step": 24045 }, { "epoch": 1.4927059407784469, "grad_norm": 0.5197278848120187, "learning_rate": 5.910059023490227e-05, "loss": 2.8037, "step": 24046 }, { "epoch": 1.4927680178782048, "grad_norm": 0.7606726872988497, "learning_rate": 5.9097039012637833e-05, "loss": 2.8234, "step": 24047 }, { "epoch": 1.4928300949779627, "grad_norm": 0.5637272057503094, "learning_rate": 5.9093487742911934e-05, "loss": 2.7671, "step": 24048 }, { "epoch": 1.4928921720777204, "grad_norm": 0.6666998830110005, "learning_rate": 5.9089936425743096e-05, "loss": 2.7829, "step": 24049 }, { "epoch": 1.4929542491774783, "grad_norm": 0.5697216474434189, "learning_rate": 5.908638506114986e-05, "loss": 2.7604, "step": 24050 }, { "epoch": 1.4930163262772362, "grad_norm": 0.6098548760304497, "learning_rate": 5.908283364915075e-05, "loss": 2.7115, "step": 24051 }, { "epoch": 1.4930784033769942, "grad_norm": 0.571490051908842, "learning_rate": 5.907928218976429e-05, "loss": 2.7967, "step": 24052 }, { "epoch": 1.493140480476752, "grad_norm": 0.5886097055677133, "learning_rate": 5.907573068300901e-05, "loss": 2.7044, "step": 24053 }, { "epoch": 1.49320255757651, "grad_norm": 0.5250329995002159, "learning_rate": 5.9072179128903435e-05, "loss": 2.7166, "step": 24054 }, { "epoch": 1.493264634676268, "grad_norm": 0.6411650869875432, "learning_rate": 5.906862752746611e-05, "loss": 2.8358, "step": 24055 }, { "epoch": 1.4933267117760258, "grad_norm": 0.5788425168516125, "learning_rate": 5.906507587871555e-05, "loss": 2.812, "step": 24056 }, { "epoch": 1.4933887888757837, "grad_norm": 0.6357510708189724, "learning_rate": 5.9061524182670295e-05, "loss": 2.6904, "step": 24057 }, { "epoch": 1.4934508659755417, "grad_norm": 0.5601929401208885, "learning_rate": 5.905797243934886e-05, "loss": 2.8005, "step": 24058 }, { "epoch": 1.4935129430752996, "grad_norm": 0.5437730554099488, "learning_rate": 5.9054420648769785e-05, "loss": 2.7859, "step": 24059 }, { "epoch": 1.4935750201750575, "grad_norm": 0.5334263428659302, "learning_rate": 5.90508688109516e-05, "loss": 2.7187, "step": 24060 }, { "epoch": 1.4936370972748154, "grad_norm": 0.6262365030596946, "learning_rate": 5.904731692591283e-05, "loss": 2.7637, "step": 24061 }, { "epoch": 1.4936991743745733, "grad_norm": 0.5730830528941246, "learning_rate": 5.904376499367201e-05, "loss": 2.8024, "step": 24062 }, { "epoch": 1.4937612514743313, "grad_norm": 0.6488556949607482, "learning_rate": 5.9040213014247684e-05, "loss": 2.8153, "step": 24063 }, { "epoch": 1.493823328574089, "grad_norm": 0.5658607486896202, "learning_rate": 5.903666098765837e-05, "loss": 2.7431, "step": 24064 }, { "epoch": 1.4938854056738469, "grad_norm": 0.5946197413832874, "learning_rate": 5.903310891392259e-05, "loss": 2.8532, "step": 24065 }, { "epoch": 1.4939474827736048, "grad_norm": 0.6312889549775047, "learning_rate": 5.9029556793058904e-05, "loss": 2.8346, "step": 24066 }, { "epoch": 1.4940095598733627, "grad_norm": 0.530352844699766, "learning_rate": 5.902600462508581e-05, "loss": 2.7754, "step": 24067 }, { "epoch": 1.4940716369731206, "grad_norm": 0.5743258629370428, "learning_rate": 5.9022452410021865e-05, "loss": 2.7659, "step": 24068 }, { "epoch": 1.4941337140728785, "grad_norm": 0.6165515182143555, "learning_rate": 5.90189001478856e-05, "loss": 2.6903, "step": 24069 }, { "epoch": 1.4941957911726365, "grad_norm": 0.5894168041205178, "learning_rate": 5.901534783869555e-05, "loss": 2.7915, "step": 24070 }, { "epoch": 1.4942578682723944, "grad_norm": 0.6471934967278941, "learning_rate": 5.9011795482470234e-05, "loss": 2.8057, "step": 24071 }, { "epoch": 1.4943199453721523, "grad_norm": 0.6397637649119656, "learning_rate": 5.900824307922819e-05, "loss": 2.7573, "step": 24072 }, { "epoch": 1.49438202247191, "grad_norm": 0.6041749728936291, "learning_rate": 5.900469062898796e-05, "loss": 2.7451, "step": 24073 }, { "epoch": 1.494444099571668, "grad_norm": 0.5662199305872544, "learning_rate": 5.900113813176806e-05, "loss": 2.8552, "step": 24074 }, { "epoch": 1.4945061766714258, "grad_norm": 0.6400466494769725, "learning_rate": 5.899758558758706e-05, "loss": 2.7447, "step": 24075 }, { "epoch": 1.4945682537711837, "grad_norm": 0.549170906294616, "learning_rate": 5.899403299646347e-05, "loss": 2.7895, "step": 24076 }, { "epoch": 1.4946303308709417, "grad_norm": 0.6325396903404824, "learning_rate": 5.8990480358415814e-05, "loss": 2.8905, "step": 24077 }, { "epoch": 1.4946924079706996, "grad_norm": 0.5543008450667705, "learning_rate": 5.8986927673462645e-05, "loss": 2.8351, "step": 24078 }, { "epoch": 1.4947544850704575, "grad_norm": 0.6639713673608475, "learning_rate": 5.898337494162248e-05, "loss": 2.8924, "step": 24079 }, { "epoch": 1.4948165621702154, "grad_norm": 0.5616939650626506, "learning_rate": 5.8979822162913886e-05, "loss": 2.7663, "step": 24080 }, { "epoch": 1.4948786392699733, "grad_norm": 0.6383293236472918, "learning_rate": 5.897626933735536e-05, "loss": 2.7853, "step": 24081 }, { "epoch": 1.4949407163697312, "grad_norm": 0.5955827935426617, "learning_rate": 5.8972716464965485e-05, "loss": 2.7446, "step": 24082 }, { "epoch": 1.4950027934694892, "grad_norm": 0.5682011806916857, "learning_rate": 5.896916354576274e-05, "loss": 2.7191, "step": 24083 }, { "epoch": 1.495064870569247, "grad_norm": 0.6427981802792361, "learning_rate": 5.8965610579765705e-05, "loss": 2.7786, "step": 24084 }, { "epoch": 1.495126947669005, "grad_norm": 0.5429382864550188, "learning_rate": 5.896205756699291e-05, "loss": 2.8801, "step": 24085 }, { "epoch": 1.495189024768763, "grad_norm": 0.6633453311226513, "learning_rate": 5.895850450746287e-05, "loss": 2.8132, "step": 24086 }, { "epoch": 1.4952511018685208, "grad_norm": 0.4986478433461582, "learning_rate": 5.895495140119416e-05, "loss": 2.7428, "step": 24087 }, { "epoch": 1.4953131789682785, "grad_norm": 0.7143152503310207, "learning_rate": 5.895139824820527e-05, "loss": 2.8417, "step": 24088 }, { "epoch": 1.4953752560680365, "grad_norm": 0.5626560365633572, "learning_rate": 5.894784504851478e-05, "loss": 2.7846, "step": 24089 }, { "epoch": 1.4954373331677944, "grad_norm": 0.7049343564876339, "learning_rate": 5.89442918021412e-05, "loss": 2.8615, "step": 24090 }, { "epoch": 1.4954994102675523, "grad_norm": 0.7023117448518051, "learning_rate": 5.894073850910309e-05, "loss": 2.7677, "step": 24091 }, { "epoch": 1.4955614873673102, "grad_norm": 0.6326870264060804, "learning_rate": 5.893718516941895e-05, "loss": 2.7486, "step": 24092 }, { "epoch": 1.4956235644670681, "grad_norm": 0.6364794322344366, "learning_rate": 5.893363178310738e-05, "loss": 2.7951, "step": 24093 }, { "epoch": 1.495685641566826, "grad_norm": 0.65170645278527, "learning_rate": 5.8930078350186866e-05, "loss": 2.7601, "step": 24094 }, { "epoch": 1.495747718666584, "grad_norm": 0.6433181212122291, "learning_rate": 5.892652487067597e-05, "loss": 2.7873, "step": 24095 }, { "epoch": 1.4958097957663419, "grad_norm": 0.6633447470673434, "learning_rate": 5.892297134459322e-05, "loss": 2.7929, "step": 24096 }, { "epoch": 1.4958718728660996, "grad_norm": 0.6611066443985176, "learning_rate": 5.891941777195716e-05, "loss": 2.7826, "step": 24097 }, { "epoch": 1.4959339499658575, "grad_norm": 0.7283553287173168, "learning_rate": 5.891586415278635e-05, "loss": 2.8701, "step": 24098 }, { "epoch": 1.4959960270656154, "grad_norm": 0.6571664315046792, "learning_rate": 5.8912310487099296e-05, "loss": 2.7091, "step": 24099 }, { "epoch": 1.4960581041653733, "grad_norm": 0.7322953178419648, "learning_rate": 5.890875677491455e-05, "loss": 2.8128, "step": 24100 }, { "epoch": 1.4961201812651312, "grad_norm": 0.7419590246988864, "learning_rate": 5.8905203016250674e-05, "loss": 2.7028, "step": 24101 }, { "epoch": 1.4961822583648892, "grad_norm": 0.5948865953931716, "learning_rate": 5.890164921112619e-05, "loss": 2.7507, "step": 24102 }, { "epoch": 1.496244335464647, "grad_norm": 0.7641096104489169, "learning_rate": 5.8898095359559636e-05, "loss": 2.8227, "step": 24103 }, { "epoch": 1.496306412564405, "grad_norm": 0.6121321926477247, "learning_rate": 5.889454146156955e-05, "loss": 2.7704, "step": 24104 }, { "epoch": 1.496368489664163, "grad_norm": 0.7366106619482489, "learning_rate": 5.88909875171745e-05, "loss": 2.7389, "step": 24105 }, { "epoch": 1.4964305667639208, "grad_norm": 0.6364845594090037, "learning_rate": 5.888743352639299e-05, "loss": 2.8699, "step": 24106 }, { "epoch": 1.4964926438636788, "grad_norm": 0.6583411188107887, "learning_rate": 5.88838794892436e-05, "loss": 2.7361, "step": 24107 }, { "epoch": 1.4965547209634367, "grad_norm": 0.634661903710418, "learning_rate": 5.888032540574484e-05, "loss": 2.77, "step": 24108 }, { "epoch": 1.4966167980631946, "grad_norm": 0.5737816061402423, "learning_rate": 5.8876771275915276e-05, "loss": 2.6631, "step": 24109 }, { "epoch": 1.4966788751629525, "grad_norm": 0.6359962857980658, "learning_rate": 5.887321709977344e-05, "loss": 2.8309, "step": 24110 }, { "epoch": 1.4967409522627104, "grad_norm": 0.5724796767146518, "learning_rate": 5.8869662877337874e-05, "loss": 2.7479, "step": 24111 }, { "epoch": 1.4968030293624681, "grad_norm": 0.6094132680370002, "learning_rate": 5.886610860862713e-05, "loss": 2.781, "step": 24112 }, { "epoch": 1.496865106462226, "grad_norm": 0.5568204622717394, "learning_rate": 5.886255429365973e-05, "loss": 2.7368, "step": 24113 }, { "epoch": 1.496927183561984, "grad_norm": 0.5794608688505659, "learning_rate": 5.885899993245425e-05, "loss": 2.8005, "step": 24114 }, { "epoch": 1.4969892606617419, "grad_norm": 0.5736492992365921, "learning_rate": 5.885544552502921e-05, "loss": 2.8517, "step": 24115 }, { "epoch": 1.4970513377614998, "grad_norm": 0.6214935887083533, "learning_rate": 5.8851891071403156e-05, "loss": 2.7776, "step": 24116 }, { "epoch": 1.4971134148612577, "grad_norm": 0.6184879557270251, "learning_rate": 5.884833657159463e-05, "loss": 2.8826, "step": 24117 }, { "epoch": 1.4971754919610156, "grad_norm": 0.560882149093291, "learning_rate": 5.8844782025622205e-05, "loss": 2.756, "step": 24118 }, { "epoch": 1.4972375690607735, "grad_norm": 0.5818321092165277, "learning_rate": 5.88412274335044e-05, "loss": 2.7208, "step": 24119 }, { "epoch": 1.4972996461605315, "grad_norm": 0.608846973630143, "learning_rate": 5.883767279525974e-05, "loss": 2.8675, "step": 24120 }, { "epoch": 1.4973617232602892, "grad_norm": 0.583622696042492, "learning_rate": 5.883411811090683e-05, "loss": 2.7834, "step": 24121 }, { "epoch": 1.497423800360047, "grad_norm": 0.5718578148432868, "learning_rate": 5.8830563380464165e-05, "loss": 2.8722, "step": 24122 }, { "epoch": 1.497485877459805, "grad_norm": 0.5199428161485048, "learning_rate": 5.882700860395032e-05, "loss": 2.8305, "step": 24123 }, { "epoch": 1.497547954559563, "grad_norm": 0.6113441635806681, "learning_rate": 5.882345378138381e-05, "loss": 2.8223, "step": 24124 }, { "epoch": 1.4976100316593208, "grad_norm": 0.5180324935240018, "learning_rate": 5.881989891278321e-05, "loss": 2.7718, "step": 24125 }, { "epoch": 1.4976721087590787, "grad_norm": 0.590744669461991, "learning_rate": 5.881634399816704e-05, "loss": 2.6819, "step": 24126 }, { "epoch": 1.4977341858588367, "grad_norm": 0.6170358592267189, "learning_rate": 5.8812789037553894e-05, "loss": 2.8833, "step": 24127 }, { "epoch": 1.4977962629585946, "grad_norm": 0.5805014918092128, "learning_rate": 5.880923403096226e-05, "loss": 2.7419, "step": 24128 }, { "epoch": 1.4978583400583525, "grad_norm": 0.544581590730914, "learning_rate": 5.880567897841073e-05, "loss": 2.6944, "step": 24129 }, { "epoch": 1.4979204171581104, "grad_norm": 0.5917536653663823, "learning_rate": 5.8802123879917836e-05, "loss": 2.7523, "step": 24130 }, { "epoch": 1.4979824942578683, "grad_norm": 0.5641653521910793, "learning_rate": 5.8798568735502116e-05, "loss": 2.6949, "step": 24131 }, { "epoch": 1.4980445713576263, "grad_norm": 0.5307309756811256, "learning_rate": 5.879501354518212e-05, "loss": 2.7981, "step": 24132 }, { "epoch": 1.4981066484573842, "grad_norm": 0.5717918856731862, "learning_rate": 5.879145830897641e-05, "loss": 2.7137, "step": 24133 }, { "epoch": 1.498168725557142, "grad_norm": 0.5262063653030986, "learning_rate": 5.8787903026903537e-05, "loss": 2.8026, "step": 24134 }, { "epoch": 1.4982308026569, "grad_norm": 0.590100163184982, "learning_rate": 5.8784347698982024e-05, "loss": 2.7877, "step": 24135 }, { "epoch": 1.4982928797566577, "grad_norm": 0.49393749962446737, "learning_rate": 5.8780792325230446e-05, "loss": 2.7328, "step": 24136 }, { "epoch": 1.4983549568564156, "grad_norm": 0.6197150901696753, "learning_rate": 5.8777236905667355e-05, "loss": 2.7544, "step": 24137 }, { "epoch": 1.4984170339561735, "grad_norm": 0.5491195936486101, "learning_rate": 5.877368144031128e-05, "loss": 2.8242, "step": 24138 }, { "epoch": 1.4984791110559315, "grad_norm": 0.6174833819821214, "learning_rate": 5.8770125929180774e-05, "loss": 2.7985, "step": 24139 }, { "epoch": 1.4985411881556894, "grad_norm": 0.5294351713804594, "learning_rate": 5.8766570372294384e-05, "loss": 2.8074, "step": 24140 }, { "epoch": 1.4986032652554473, "grad_norm": 0.5429096326960311, "learning_rate": 5.876301476967069e-05, "loss": 2.8021, "step": 24141 }, { "epoch": 1.4986653423552052, "grad_norm": 0.5505149955283598, "learning_rate": 5.87594591213282e-05, "loss": 2.694, "step": 24142 }, { "epoch": 1.4987274194549631, "grad_norm": 0.5216354836974275, "learning_rate": 5.87559034272855e-05, "loss": 2.7832, "step": 24143 }, { "epoch": 1.498789496554721, "grad_norm": 0.5960295736380994, "learning_rate": 5.875234768756112e-05, "loss": 2.866, "step": 24144 }, { "epoch": 1.4988515736544787, "grad_norm": 0.5594956379806482, "learning_rate": 5.874879190217362e-05, "loss": 2.8032, "step": 24145 }, { "epoch": 1.4989136507542367, "grad_norm": 0.5538908112907188, "learning_rate": 5.874523607114154e-05, "loss": 2.659, "step": 24146 }, { "epoch": 1.4989757278539946, "grad_norm": 0.5914386634561253, "learning_rate": 5.874168019448345e-05, "loss": 2.8526, "step": 24147 }, { "epoch": 1.4990378049537525, "grad_norm": 0.629244150880978, "learning_rate": 5.8738124272217895e-05, "loss": 2.8342, "step": 24148 }, { "epoch": 1.4990998820535104, "grad_norm": 0.560781662314522, "learning_rate": 5.8734568304363405e-05, "loss": 2.785, "step": 24149 }, { "epoch": 1.4991619591532683, "grad_norm": 0.6161166933660247, "learning_rate": 5.873101229093857e-05, "loss": 2.8576, "step": 24150 }, { "epoch": 1.4992240362530262, "grad_norm": 0.5894368264869237, "learning_rate": 5.872745623196192e-05, "loss": 2.727, "step": 24151 }, { "epoch": 1.4992861133527842, "grad_norm": 0.5783895969597628, "learning_rate": 5.872390012745201e-05, "loss": 2.7729, "step": 24152 }, { "epoch": 1.499348190452542, "grad_norm": 0.547484031514051, "learning_rate": 5.872034397742739e-05, "loss": 2.7407, "step": 24153 }, { "epoch": 1.4994102675523, "grad_norm": 0.5896664283436405, "learning_rate": 5.8716787781906635e-05, "loss": 2.7888, "step": 24154 }, { "epoch": 1.499472344652058, "grad_norm": 0.5680464574659899, "learning_rate": 5.871323154090828e-05, "loss": 2.8441, "step": 24155 }, { "epoch": 1.4995344217518158, "grad_norm": 0.5559239910901064, "learning_rate": 5.870967525445087e-05, "loss": 2.7541, "step": 24156 }, { "epoch": 1.4995964988515738, "grad_norm": 0.5478210984734133, "learning_rate": 5.8706118922552975e-05, "loss": 2.7902, "step": 24157 }, { "epoch": 1.4996585759513317, "grad_norm": 0.5753880746115458, "learning_rate": 5.870256254523314e-05, "loss": 2.7659, "step": 24158 }, { "epoch": 1.4997206530510896, "grad_norm": 0.5594648225155626, "learning_rate": 5.8699006122509924e-05, "loss": 2.8143, "step": 24159 }, { "epoch": 1.4997827301508473, "grad_norm": 0.497262179474382, "learning_rate": 5.869544965440188e-05, "loss": 2.7239, "step": 24160 }, { "epoch": 1.4998448072506052, "grad_norm": 0.551326049394283, "learning_rate": 5.869189314092758e-05, "loss": 2.7334, "step": 24161 }, { "epoch": 1.4999068843503631, "grad_norm": 0.5691712600956437, "learning_rate": 5.868833658210554e-05, "loss": 2.8619, "step": 24162 }, { "epoch": 1.499968961450121, "grad_norm": 0.634964474562876, "learning_rate": 5.8684779977954365e-05, "loss": 2.7645, "step": 24163 }, { "epoch": 1.500031038549879, "grad_norm": 0.5945557347392669, "learning_rate": 5.868122332849255e-05, "loss": 2.7567, "step": 24164 }, { "epoch": 1.5000931156496369, "grad_norm": 0.5688806876223248, "learning_rate": 5.8677666633738704e-05, "loss": 2.7602, "step": 24165 }, { "epoch": 1.5001551927493948, "grad_norm": 0.5580625609343277, "learning_rate": 5.8674109893711374e-05, "loss": 2.759, "step": 24166 }, { "epoch": 1.5002172698491525, "grad_norm": 0.5121987981289128, "learning_rate": 5.867055310842909e-05, "loss": 2.7403, "step": 24167 }, { "epoch": 1.5002793469489104, "grad_norm": 0.5862169743652392, "learning_rate": 5.866699627791045e-05, "loss": 2.8443, "step": 24168 }, { "epoch": 1.5003414240486683, "grad_norm": 0.6250096558778648, "learning_rate": 5.866343940217396e-05, "loss": 2.8045, "step": 24169 }, { "epoch": 1.5004035011484262, "grad_norm": 0.5851719496369089, "learning_rate": 5.865988248123821e-05, "loss": 2.726, "step": 24170 }, { "epoch": 1.5004655782481842, "grad_norm": 0.6253809959723491, "learning_rate": 5.8656325515121757e-05, "loss": 2.7434, "step": 24171 }, { "epoch": 1.500527655347942, "grad_norm": 0.5439826774433718, "learning_rate": 5.865276850384316e-05, "loss": 2.7599, "step": 24172 }, { "epoch": 1.5005897324477, "grad_norm": 0.5942733867121234, "learning_rate": 5.864921144742096e-05, "loss": 2.7343, "step": 24173 }, { "epoch": 1.500651809547458, "grad_norm": 0.5842735059593946, "learning_rate": 5.864565434587372e-05, "loss": 2.8389, "step": 24174 }, { "epoch": 1.5007138866472158, "grad_norm": 0.5631583917097722, "learning_rate": 5.864209719922001e-05, "loss": 2.8537, "step": 24175 }, { "epoch": 1.5007759637469738, "grad_norm": 0.5517821990168778, "learning_rate": 5.863854000747837e-05, "loss": 2.6963, "step": 24176 }, { "epoch": 1.5008380408467317, "grad_norm": 0.5235528961062194, "learning_rate": 5.863498277066739e-05, "loss": 2.8558, "step": 24177 }, { "epoch": 1.5009001179464896, "grad_norm": 0.5488550037033273, "learning_rate": 5.863142548880558e-05, "loss": 2.7981, "step": 24178 }, { "epoch": 1.5009621950462475, "grad_norm": 0.5368607656614695, "learning_rate": 5.862786816191156e-05, "loss": 2.7567, "step": 24179 }, { "epoch": 1.5010242721460054, "grad_norm": 0.5478897217433463, "learning_rate": 5.862431079000383e-05, "loss": 2.8192, "step": 24180 }, { "epoch": 1.5010863492457633, "grad_norm": 0.5761816508870082, "learning_rate": 5.8620753373100975e-05, "loss": 2.7887, "step": 24181 }, { "epoch": 1.5011484263455213, "grad_norm": 0.5284979445833425, "learning_rate": 5.8617195911221576e-05, "loss": 2.7437, "step": 24182 }, { "epoch": 1.5012105034452792, "grad_norm": 0.5875115931158439, "learning_rate": 5.861363840438416e-05, "loss": 2.7528, "step": 24183 }, { "epoch": 1.501272580545037, "grad_norm": 0.6371672051505886, "learning_rate": 5.861008085260733e-05, "loss": 2.7287, "step": 24184 }, { "epoch": 1.5013346576447948, "grad_norm": 0.5398821749447789, "learning_rate": 5.8606523255909584e-05, "loss": 2.7212, "step": 24185 }, { "epoch": 1.5013967347445527, "grad_norm": 0.5660170961180552, "learning_rate": 5.860296561430953e-05, "loss": 2.7527, "step": 24186 }, { "epoch": 1.5014588118443106, "grad_norm": 0.552464908115457, "learning_rate": 5.859940792782572e-05, "loss": 2.7314, "step": 24187 }, { "epoch": 1.5015208889440685, "grad_norm": 0.6083937839189251, "learning_rate": 5.859585019647671e-05, "loss": 2.7964, "step": 24188 }, { "epoch": 1.5015829660438265, "grad_norm": 0.5088217170179464, "learning_rate": 5.859229242028106e-05, "loss": 2.7714, "step": 24189 }, { "epoch": 1.5016450431435844, "grad_norm": 0.5714622821826721, "learning_rate": 5.8588734599257335e-05, "loss": 2.9018, "step": 24190 }, { "epoch": 1.501707120243342, "grad_norm": 0.5791307670343991, "learning_rate": 5.8585176733424095e-05, "loss": 2.898, "step": 24191 }, { "epoch": 1.5017691973431, "grad_norm": 0.5305891695270284, "learning_rate": 5.858161882279991e-05, "loss": 2.7407, "step": 24192 }, { "epoch": 1.501831274442858, "grad_norm": 0.6030759228123727, "learning_rate": 5.857806086740334e-05, "loss": 2.8354, "step": 24193 }, { "epoch": 1.5018933515426158, "grad_norm": 0.5821964618281811, "learning_rate": 5.857450286725292e-05, "loss": 2.7326, "step": 24194 }, { "epoch": 1.5019554286423737, "grad_norm": 0.6000106909912035, "learning_rate": 5.857094482236726e-05, "loss": 2.7415, "step": 24195 }, { "epoch": 1.5020175057421317, "grad_norm": 0.5870341655708551, "learning_rate": 5.856738673276488e-05, "loss": 2.9183, "step": 24196 }, { "epoch": 1.5020795828418896, "grad_norm": 0.6441156839385127, "learning_rate": 5.8563828598464386e-05, "loss": 2.8333, "step": 24197 }, { "epoch": 1.5021416599416475, "grad_norm": 0.5621175790576409, "learning_rate": 5.8560270419484306e-05, "loss": 2.7592, "step": 24198 }, { "epoch": 1.5022037370414054, "grad_norm": 0.5634782285351643, "learning_rate": 5.855671219584322e-05, "loss": 2.8435, "step": 24199 }, { "epoch": 1.5022658141411633, "grad_norm": 0.5277125728912271, "learning_rate": 5.855315392755969e-05, "loss": 2.8141, "step": 24200 }, { "epoch": 1.5023278912409213, "grad_norm": 0.5759980020190396, "learning_rate": 5.8549595614652276e-05, "loss": 2.7819, "step": 24201 }, { "epoch": 1.5023899683406792, "grad_norm": 0.5236217402276815, "learning_rate": 5.854603725713955e-05, "loss": 2.8369, "step": 24202 }, { "epoch": 1.502452045440437, "grad_norm": 0.5712257024075941, "learning_rate": 5.8542478855040076e-05, "loss": 2.8487, "step": 24203 }, { "epoch": 1.502514122540195, "grad_norm": 0.539248945639282, "learning_rate": 5.853892040837241e-05, "loss": 2.7591, "step": 24204 }, { "epoch": 1.502576199639953, "grad_norm": 0.5800913542706401, "learning_rate": 5.853536191715513e-05, "loss": 2.8134, "step": 24205 }, { "epoch": 1.5026382767397108, "grad_norm": 0.5651185492400738, "learning_rate": 5.853180338140679e-05, "loss": 2.8102, "step": 24206 }, { "epoch": 1.5027003538394688, "grad_norm": 0.6311029253427486, "learning_rate": 5.852824480114596e-05, "loss": 2.8042, "step": 24207 }, { "epoch": 1.5027624309392267, "grad_norm": 0.6140508014836178, "learning_rate": 5.852468617639122e-05, "loss": 2.766, "step": 24208 }, { "epoch": 1.5028245080389844, "grad_norm": 0.5890923065010829, "learning_rate": 5.85211275071611e-05, "loss": 2.7704, "step": 24209 }, { "epoch": 1.5028865851387423, "grad_norm": 0.6330447300659585, "learning_rate": 5.851756879347421e-05, "loss": 2.802, "step": 24210 }, { "epoch": 1.5029486622385002, "grad_norm": 0.6384308853217534, "learning_rate": 5.851401003534909e-05, "loss": 2.8211, "step": 24211 }, { "epoch": 1.5030107393382581, "grad_norm": 0.6037738827706617, "learning_rate": 5.8510451232804296e-05, "loss": 2.7546, "step": 24212 }, { "epoch": 1.503072816438016, "grad_norm": 0.5895474428591344, "learning_rate": 5.850689238585844e-05, "loss": 2.7455, "step": 24213 }, { "epoch": 1.503134893537774, "grad_norm": 0.6032694589232742, "learning_rate": 5.8503333494530025e-05, "loss": 2.7959, "step": 24214 }, { "epoch": 1.5031969706375317, "grad_norm": 0.6329192554235934, "learning_rate": 5.8499774558837686e-05, "loss": 2.8385, "step": 24215 }, { "epoch": 1.5032590477372896, "grad_norm": 0.6597530781629654, "learning_rate": 5.849621557879995e-05, "loss": 2.7214, "step": 24216 }, { "epoch": 1.5033211248370475, "grad_norm": 0.6072580785170794, "learning_rate": 5.8492656554435386e-05, "loss": 2.8489, "step": 24217 }, { "epoch": 1.5033832019368054, "grad_norm": 0.6719756161957612, "learning_rate": 5.848909748576259e-05, "loss": 2.805, "step": 24218 }, { "epoch": 1.5034452790365633, "grad_norm": 0.5323087576295328, "learning_rate": 5.8485538372800096e-05, "loss": 2.6819, "step": 24219 }, { "epoch": 1.5035073561363212, "grad_norm": 0.6953799182910603, "learning_rate": 5.848197921556649e-05, "loss": 2.7843, "step": 24220 }, { "epoch": 1.5035694332360792, "grad_norm": 0.5275586290907801, "learning_rate": 5.847842001408035e-05, "loss": 2.8427, "step": 24221 }, { "epoch": 1.503631510335837, "grad_norm": 0.6765489572486681, "learning_rate": 5.847486076836023e-05, "loss": 2.8647, "step": 24222 }, { "epoch": 1.503693587435595, "grad_norm": 0.5867619626809083, "learning_rate": 5.8471301478424686e-05, "loss": 2.7972, "step": 24223 }, { "epoch": 1.503755664535353, "grad_norm": 0.640263083037168, "learning_rate": 5.846774214429234e-05, "loss": 2.8615, "step": 24224 }, { "epoch": 1.5038177416351108, "grad_norm": 0.5633702963020527, "learning_rate": 5.84641827659817e-05, "loss": 2.7924, "step": 24225 }, { "epoch": 1.5038798187348688, "grad_norm": 0.6828166795685601, "learning_rate": 5.8460623343511376e-05, "loss": 2.8256, "step": 24226 }, { "epoch": 1.5039418958346267, "grad_norm": 0.548028465173866, "learning_rate": 5.845706387689992e-05, "loss": 2.8225, "step": 24227 }, { "epoch": 1.5040039729343846, "grad_norm": 0.5985668948847551, "learning_rate": 5.8453504366165915e-05, "loss": 2.8614, "step": 24228 }, { "epoch": 1.5040660500341425, "grad_norm": 0.6221550085837215, "learning_rate": 5.8449944811327925e-05, "loss": 2.7276, "step": 24229 }, { "epoch": 1.5041281271339004, "grad_norm": 0.5577240689489391, "learning_rate": 5.844638521240451e-05, "loss": 2.6949, "step": 24230 }, { "epoch": 1.5041902042336583, "grad_norm": 0.5574724106312671, "learning_rate": 5.8442825569414264e-05, "loss": 2.8086, "step": 24231 }, { "epoch": 1.5042522813334163, "grad_norm": 0.5895217622973714, "learning_rate": 5.8439265882375735e-05, "loss": 2.7765, "step": 24232 }, { "epoch": 1.504314358433174, "grad_norm": 0.5835282703928896, "learning_rate": 5.8435706151307525e-05, "loss": 2.7678, "step": 24233 }, { "epoch": 1.5043764355329319, "grad_norm": 0.5439149036502606, "learning_rate": 5.843214637622817e-05, "loss": 2.8392, "step": 24234 }, { "epoch": 1.5044385126326898, "grad_norm": 0.6196075026931536, "learning_rate": 5.8428586557156284e-05, "loss": 2.7964, "step": 24235 }, { "epoch": 1.5045005897324477, "grad_norm": 0.5235365588840565, "learning_rate": 5.842502669411041e-05, "loss": 2.7906, "step": 24236 }, { "epoch": 1.5045626668322056, "grad_norm": 0.594337348547698, "learning_rate": 5.842146678710912e-05, "loss": 2.7348, "step": 24237 }, { "epoch": 1.5046247439319635, "grad_norm": 0.5716520541921353, "learning_rate": 5.841790683617099e-05, "loss": 2.838, "step": 24238 }, { "epoch": 1.5046868210317212, "grad_norm": 0.5748436570122103, "learning_rate": 5.8414346841314594e-05, "loss": 2.7242, "step": 24239 }, { "epoch": 1.5047488981314792, "grad_norm": 0.46642980148686924, "learning_rate": 5.8410786802558515e-05, "loss": 2.7001, "step": 24240 }, { "epoch": 1.504810975231237, "grad_norm": 0.6191584587537183, "learning_rate": 5.840722671992132e-05, "loss": 2.7546, "step": 24241 }, { "epoch": 1.504873052330995, "grad_norm": 0.49102175882494414, "learning_rate": 5.840366659342158e-05, "loss": 2.6872, "step": 24242 }, { "epoch": 1.504935129430753, "grad_norm": 0.5593282498904273, "learning_rate": 5.840010642307786e-05, "loss": 2.75, "step": 24243 }, { "epoch": 1.5049972065305108, "grad_norm": 0.4893803547728221, "learning_rate": 5.8396546208908765e-05, "loss": 2.7764, "step": 24244 }, { "epoch": 1.5050592836302688, "grad_norm": 0.5211700749584204, "learning_rate": 5.8392985950932835e-05, "loss": 2.7446, "step": 24245 }, { "epoch": 1.5051213607300267, "grad_norm": 0.5177149753668969, "learning_rate": 5.8389425649168675e-05, "loss": 2.8499, "step": 24246 }, { "epoch": 1.5051834378297846, "grad_norm": 0.5486214679700798, "learning_rate": 5.838586530363482e-05, "loss": 2.7207, "step": 24247 }, { "epoch": 1.5052455149295425, "grad_norm": 0.5202481192775243, "learning_rate": 5.8382304914349884e-05, "loss": 2.6576, "step": 24248 }, { "epoch": 1.5053075920293004, "grad_norm": 0.5620975379148161, "learning_rate": 5.837874448133244e-05, "loss": 2.737, "step": 24249 }, { "epoch": 1.5053696691290583, "grad_norm": 0.5736684173143145, "learning_rate": 5.8375184004601035e-05, "loss": 2.8106, "step": 24250 }, { "epoch": 1.5054317462288163, "grad_norm": 0.5387253190514202, "learning_rate": 5.8371623484174277e-05, "loss": 2.7742, "step": 24251 }, { "epoch": 1.5054938233285742, "grad_norm": 0.5692380163349524, "learning_rate": 5.836806292007071e-05, "loss": 2.8251, "step": 24252 }, { "epoch": 1.505555900428332, "grad_norm": 0.5817814489322176, "learning_rate": 5.836450231230894e-05, "loss": 2.7599, "step": 24253 }, { "epoch": 1.50561797752809, "grad_norm": 0.566297954861321, "learning_rate": 5.8360941660907545e-05, "loss": 2.8411, "step": 24254 }, { "epoch": 1.505680054627848, "grad_norm": 0.5697312463065016, "learning_rate": 5.8357380965885056e-05, "loss": 2.7819, "step": 24255 }, { "epoch": 1.5057421317276058, "grad_norm": 0.5676763041382473, "learning_rate": 5.8353820227260104e-05, "loss": 2.8069, "step": 24256 }, { "epoch": 1.5058042088273635, "grad_norm": 0.6055970086871691, "learning_rate": 5.835025944505123e-05, "loss": 2.7671, "step": 24257 }, { "epoch": 1.5058662859271215, "grad_norm": 0.562528556522719, "learning_rate": 5.834669861927704e-05, "loss": 2.778, "step": 24258 }, { "epoch": 1.5059283630268794, "grad_norm": 0.571174623161616, "learning_rate": 5.8343137749956075e-05, "loss": 2.7708, "step": 24259 }, { "epoch": 1.5059904401266373, "grad_norm": 0.5776672355220186, "learning_rate": 5.8339576837106955e-05, "loss": 2.8046, "step": 24260 }, { "epoch": 1.5060525172263952, "grad_norm": 0.5580692896960214, "learning_rate": 5.833601588074823e-05, "loss": 2.7837, "step": 24261 }, { "epoch": 1.5061145943261531, "grad_norm": 0.5783246185441314, "learning_rate": 5.833245488089849e-05, "loss": 2.7675, "step": 24262 }, { "epoch": 1.5061766714259108, "grad_norm": 0.5138927028560296, "learning_rate": 5.83288938375763e-05, "loss": 2.7751, "step": 24263 }, { "epoch": 1.5062387485256687, "grad_norm": 0.4844446747413043, "learning_rate": 5.8325332750800256e-05, "loss": 2.7957, "step": 24264 }, { "epoch": 1.5063008256254267, "grad_norm": 0.5913546581265149, "learning_rate": 5.8321771620588936e-05, "loss": 2.8992, "step": 24265 }, { "epoch": 1.5063629027251846, "grad_norm": 0.503320753032949, "learning_rate": 5.831821044696091e-05, "loss": 2.8257, "step": 24266 }, { "epoch": 1.5064249798249425, "grad_norm": 0.5320423332325894, "learning_rate": 5.831464922993476e-05, "loss": 2.7316, "step": 24267 }, { "epoch": 1.5064870569247004, "grad_norm": 0.5498875882406845, "learning_rate": 5.8311087969529066e-05, "loss": 2.7388, "step": 24268 }, { "epoch": 1.5065491340244583, "grad_norm": 0.6211732537589294, "learning_rate": 5.8307526665762416e-05, "loss": 2.7897, "step": 24269 }, { "epoch": 1.5066112111242163, "grad_norm": 0.49695220023640235, "learning_rate": 5.830396531865336e-05, "loss": 2.6466, "step": 24270 }, { "epoch": 1.5066732882239742, "grad_norm": 0.6272832173826074, "learning_rate": 5.830040392822052e-05, "loss": 2.7621, "step": 24271 }, { "epoch": 1.506735365323732, "grad_norm": 0.5219949515909761, "learning_rate": 5.829684249448246e-05, "loss": 2.8292, "step": 24272 }, { "epoch": 1.50679744242349, "grad_norm": 0.5887128080376672, "learning_rate": 5.829328101745774e-05, "loss": 2.7615, "step": 24273 }, { "epoch": 1.506859519523248, "grad_norm": 0.5732822956565372, "learning_rate": 5.828971949716498e-05, "loss": 2.822, "step": 24274 }, { "epoch": 1.5069215966230058, "grad_norm": 0.556400715844073, "learning_rate": 5.8286157933622734e-05, "loss": 2.7979, "step": 24275 }, { "epoch": 1.5069836737227638, "grad_norm": 0.5390990977372262, "learning_rate": 5.828259632684959e-05, "loss": 2.7903, "step": 24276 }, { "epoch": 1.5070457508225217, "grad_norm": 0.6258580754778138, "learning_rate": 5.827903467686413e-05, "loss": 2.8092, "step": 24277 }, { "epoch": 1.5071078279222796, "grad_norm": 0.5913173633936293, "learning_rate": 5.827547298368494e-05, "loss": 2.8223, "step": 24278 }, { "epoch": 1.5071699050220375, "grad_norm": 0.6179639061353351, "learning_rate": 5.827191124733059e-05, "loss": 2.7751, "step": 24279 }, { "epoch": 1.5072319821217954, "grad_norm": 0.5734299932087252, "learning_rate": 5.826834946781967e-05, "loss": 2.7482, "step": 24280 }, { "epoch": 1.5072940592215531, "grad_norm": 0.6050217732144967, "learning_rate": 5.826478764517077e-05, "loss": 2.7655, "step": 24281 }, { "epoch": 1.507356136321311, "grad_norm": 0.5735957600333114, "learning_rate": 5.826122577940246e-05, "loss": 2.8009, "step": 24282 }, { "epoch": 1.507418213421069, "grad_norm": 0.635103763602927, "learning_rate": 5.825766387053335e-05, "loss": 2.7424, "step": 24283 }, { "epoch": 1.5074802905208269, "grad_norm": 0.5296956210793864, "learning_rate": 5.825410191858198e-05, "loss": 2.8158, "step": 24284 }, { "epoch": 1.5075423676205848, "grad_norm": 0.587446330033658, "learning_rate": 5.825053992356697e-05, "loss": 2.8205, "step": 24285 }, { "epoch": 1.5076044447203427, "grad_norm": 0.60997605636382, "learning_rate": 5.824697788550688e-05, "loss": 2.8224, "step": 24286 }, { "epoch": 1.5076665218201004, "grad_norm": 0.5804793899813719, "learning_rate": 5.824341580442031e-05, "loss": 2.7423, "step": 24287 }, { "epoch": 1.5077285989198583, "grad_norm": 0.6017900011961842, "learning_rate": 5.823985368032584e-05, "loss": 2.8314, "step": 24288 }, { "epoch": 1.5077906760196162, "grad_norm": 0.568061960539686, "learning_rate": 5.8236291513242056e-05, "loss": 2.7549, "step": 24289 }, { "epoch": 1.5078527531193742, "grad_norm": 0.5768366970600515, "learning_rate": 5.823272930318754e-05, "loss": 2.831, "step": 24290 }, { "epoch": 1.507914830219132, "grad_norm": 0.6284442650095583, "learning_rate": 5.8229167050180866e-05, "loss": 2.8448, "step": 24291 }, { "epoch": 1.50797690731889, "grad_norm": 0.5697485081601666, "learning_rate": 5.8225604754240635e-05, "loss": 2.84, "step": 24292 }, { "epoch": 1.508038984418648, "grad_norm": 0.6195678680873883, "learning_rate": 5.8222042415385426e-05, "loss": 2.7216, "step": 24293 }, { "epoch": 1.5081010615184058, "grad_norm": 0.5578268695764922, "learning_rate": 5.8218480033633825e-05, "loss": 2.835, "step": 24294 }, { "epoch": 1.5081631386181638, "grad_norm": 0.5879692455961768, "learning_rate": 5.8214917609004414e-05, "loss": 2.8381, "step": 24295 }, { "epoch": 1.5082252157179217, "grad_norm": 0.6161382181874527, "learning_rate": 5.821135514151579e-05, "loss": 2.8082, "step": 24296 }, { "epoch": 1.5082872928176796, "grad_norm": 0.6174454693133696, "learning_rate": 5.820779263118654e-05, "loss": 2.801, "step": 24297 }, { "epoch": 1.5083493699174375, "grad_norm": 0.6174560155917915, "learning_rate": 5.8204230078035224e-05, "loss": 2.8295, "step": 24298 }, { "epoch": 1.5084114470171954, "grad_norm": 0.6207678317605667, "learning_rate": 5.820066748208047e-05, "loss": 2.8416, "step": 24299 }, { "epoch": 1.5084735241169533, "grad_norm": 0.5985390619295854, "learning_rate": 5.8197104843340835e-05, "loss": 2.6704, "step": 24300 }, { "epoch": 1.5085356012167113, "grad_norm": 0.6217177513327371, "learning_rate": 5.8193542161834914e-05, "loss": 2.7519, "step": 24301 }, { "epoch": 1.5085976783164692, "grad_norm": 0.5925318117092285, "learning_rate": 5.8189979437581286e-05, "loss": 2.8049, "step": 24302 }, { "epoch": 1.508659755416227, "grad_norm": 0.6021165896892098, "learning_rate": 5.8186416670598566e-05, "loss": 2.7767, "step": 24303 }, { "epoch": 1.508721832515985, "grad_norm": 0.6553186210597489, "learning_rate": 5.818285386090529e-05, "loss": 2.8678, "step": 24304 }, { "epoch": 1.5087839096157427, "grad_norm": 0.5750864699900545, "learning_rate": 5.817929100852012e-05, "loss": 2.8195, "step": 24305 }, { "epoch": 1.5088459867155006, "grad_norm": 0.6881600592583882, "learning_rate": 5.817572811346157e-05, "loss": 2.7872, "step": 24306 }, { "epoch": 1.5089080638152585, "grad_norm": 0.5642950503221967, "learning_rate": 5.817216517574828e-05, "loss": 2.7911, "step": 24307 }, { "epoch": 1.5089701409150165, "grad_norm": 0.6231428927394199, "learning_rate": 5.816860219539882e-05, "loss": 2.7493, "step": 24308 }, { "epoch": 1.5090322180147744, "grad_norm": 0.5550300453834054, "learning_rate": 5.816503917243178e-05, "loss": 2.7591, "step": 24309 }, { "epoch": 1.5090942951145323, "grad_norm": 0.5465494276678725, "learning_rate": 5.816147610686574e-05, "loss": 2.7855, "step": 24310 }, { "epoch": 1.50915637221429, "grad_norm": 0.5690678902820614, "learning_rate": 5.8157912998719296e-05, "loss": 2.8265, "step": 24311 }, { "epoch": 1.509218449314048, "grad_norm": 0.5830922792419108, "learning_rate": 5.815434984801105e-05, "loss": 2.8154, "step": 24312 }, { "epoch": 1.5092805264138058, "grad_norm": 0.6702959445429469, "learning_rate": 5.815078665475956e-05, "loss": 2.8418, "step": 24313 }, { "epoch": 1.5093426035135638, "grad_norm": 0.5478316522744041, "learning_rate": 5.8147223418983466e-05, "loss": 2.819, "step": 24314 }, { "epoch": 1.5094046806133217, "grad_norm": 0.6566395057791227, "learning_rate": 5.8143660140701315e-05, "loss": 2.8071, "step": 24315 }, { "epoch": 1.5094667577130796, "grad_norm": 0.7048164579553164, "learning_rate": 5.814009681993171e-05, "loss": 2.7537, "step": 24316 }, { "epoch": 1.5095288348128375, "grad_norm": 0.5603206856039794, "learning_rate": 5.8136533456693255e-05, "loss": 2.6947, "step": 24317 }, { "epoch": 1.5095909119125954, "grad_norm": 0.7087192933092987, "learning_rate": 5.813297005100452e-05, "loss": 2.82, "step": 24318 }, { "epoch": 1.5096529890123533, "grad_norm": 0.5912561999796353, "learning_rate": 5.8129406602884105e-05, "loss": 2.8849, "step": 24319 }, { "epoch": 1.5097150661121113, "grad_norm": 0.613801359020567, "learning_rate": 5.812584311235061e-05, "loss": 2.8702, "step": 24320 }, { "epoch": 1.5097771432118692, "grad_norm": 0.6089814211105147, "learning_rate": 5.812227957942261e-05, "loss": 2.7797, "step": 24321 }, { "epoch": 1.509839220311627, "grad_norm": 0.57624715539557, "learning_rate": 5.8118716004118715e-05, "loss": 2.8258, "step": 24322 }, { "epoch": 1.509901297411385, "grad_norm": 0.632068691394213, "learning_rate": 5.8115152386457506e-05, "loss": 2.7625, "step": 24323 }, { "epoch": 1.509963374511143, "grad_norm": 0.5762979918994169, "learning_rate": 5.8111588726457565e-05, "loss": 2.7577, "step": 24324 }, { "epoch": 1.5100254516109008, "grad_norm": 0.5824366700041352, "learning_rate": 5.810802502413751e-05, "loss": 2.8665, "step": 24325 }, { "epoch": 1.5100875287106588, "grad_norm": 0.5893466951970323, "learning_rate": 5.810446127951592e-05, "loss": 2.8271, "step": 24326 }, { "epoch": 1.5101496058104167, "grad_norm": 0.6030805680111105, "learning_rate": 5.810089749261138e-05, "loss": 2.7796, "step": 24327 }, { "epoch": 1.5102116829101746, "grad_norm": 0.5780274334620829, "learning_rate": 5.80973336634425e-05, "loss": 2.7904, "step": 24328 }, { "epoch": 1.5102737600099323, "grad_norm": 0.5828764494309963, "learning_rate": 5.809376979202784e-05, "loss": 2.8029, "step": 24329 }, { "epoch": 1.5103358371096902, "grad_norm": 0.6192051225154631, "learning_rate": 5.809020587838605e-05, "loss": 2.7983, "step": 24330 }, { "epoch": 1.5103979142094481, "grad_norm": 0.5969692345572155, "learning_rate": 5.808664192253567e-05, "loss": 2.7698, "step": 24331 }, { "epoch": 1.510459991309206, "grad_norm": 0.5748986611006137, "learning_rate": 5.8083077924495333e-05, "loss": 2.7528, "step": 24332 }, { "epoch": 1.510522068408964, "grad_norm": 0.5969676869511781, "learning_rate": 5.807951388428361e-05, "loss": 2.7789, "step": 24333 }, { "epoch": 1.5105841455087219, "grad_norm": 0.5891881672102854, "learning_rate": 5.8075949801919096e-05, "loss": 2.7913, "step": 24334 }, { "epoch": 1.5106462226084796, "grad_norm": 0.5777934902400156, "learning_rate": 5.80723856774204e-05, "loss": 2.7533, "step": 24335 }, { "epoch": 1.5107082997082375, "grad_norm": 0.5704022167131974, "learning_rate": 5.8068821510806105e-05, "loss": 2.756, "step": 24336 }, { "epoch": 1.5107703768079954, "grad_norm": 0.5949492390000769, "learning_rate": 5.806525730209481e-05, "loss": 2.7933, "step": 24337 }, { "epoch": 1.5108324539077533, "grad_norm": 0.5205717637960681, "learning_rate": 5.806169305130511e-05, "loss": 2.7972, "step": 24338 }, { "epoch": 1.5108945310075113, "grad_norm": 0.734411279309781, "learning_rate": 5.80581287584556e-05, "loss": 2.8123, "step": 24339 }, { "epoch": 1.5109566081072692, "grad_norm": 0.5491067850907032, "learning_rate": 5.805456442356487e-05, "loss": 2.7737, "step": 24340 }, { "epoch": 1.511018685207027, "grad_norm": 0.6655032068858847, "learning_rate": 5.805100004665153e-05, "loss": 2.8198, "step": 24341 }, { "epoch": 1.511080762306785, "grad_norm": 0.6133009306064166, "learning_rate": 5.804743562773416e-05, "loss": 2.7107, "step": 24342 }, { "epoch": 1.511142839406543, "grad_norm": 0.6312755239633099, "learning_rate": 5.8043871166831375e-05, "loss": 2.8318, "step": 24343 }, { "epoch": 1.5112049165063008, "grad_norm": 0.5380248988817445, "learning_rate": 5.804030666396176e-05, "loss": 2.7664, "step": 24344 }, { "epoch": 1.5112669936060588, "grad_norm": 0.6522571483394343, "learning_rate": 5.803674211914389e-05, "loss": 2.738, "step": 24345 }, { "epoch": 1.5113290707058167, "grad_norm": 0.5930427806613915, "learning_rate": 5.803317753239641e-05, "loss": 2.8131, "step": 24346 }, { "epoch": 1.5113911478055746, "grad_norm": 0.6026380757446539, "learning_rate": 5.802961290373788e-05, "loss": 2.7332, "step": 24347 }, { "epoch": 1.5114532249053325, "grad_norm": 0.5478860772832262, "learning_rate": 5.802604823318694e-05, "loss": 2.7812, "step": 24348 }, { "epoch": 1.5115153020050904, "grad_norm": 0.579593237438727, "learning_rate": 5.802248352076212e-05, "loss": 2.7978, "step": 24349 }, { "epoch": 1.5115773791048484, "grad_norm": 0.5514322586389837, "learning_rate": 5.8018918766482076e-05, "loss": 2.8573, "step": 24350 }, { "epoch": 1.5116394562046063, "grad_norm": 0.5700923808425951, "learning_rate": 5.8015353970365395e-05, "loss": 2.742, "step": 24351 }, { "epoch": 1.511701533304364, "grad_norm": 0.5732083416228662, "learning_rate": 5.8011789132430636e-05, "loss": 2.7641, "step": 24352 }, { "epoch": 1.5117636104041219, "grad_norm": 0.5556851669621234, "learning_rate": 5.8008224252696454e-05, "loss": 2.7371, "step": 24353 }, { "epoch": 1.5118256875038798, "grad_norm": 0.5002381830340096, "learning_rate": 5.800465933118141e-05, "loss": 2.7065, "step": 24354 }, { "epoch": 1.5118877646036377, "grad_norm": 0.5406049967245926, "learning_rate": 5.800109436790413e-05, "loss": 2.7573, "step": 24355 }, { "epoch": 1.5119498417033956, "grad_norm": 0.5807037832645672, "learning_rate": 5.799752936288317e-05, "loss": 2.8595, "step": 24356 }, { "epoch": 1.5120119188031536, "grad_norm": 0.5313050017214239, "learning_rate": 5.799396431613718e-05, "loss": 2.8021, "step": 24357 }, { "epoch": 1.5120739959029112, "grad_norm": 0.5408892707766981, "learning_rate": 5.799039922768473e-05, "loss": 2.8208, "step": 24358 }, { "epoch": 1.5121360730026692, "grad_norm": 0.6091046344868232, "learning_rate": 5.798683409754444e-05, "loss": 2.7973, "step": 24359 }, { "epoch": 1.512198150102427, "grad_norm": 0.5351385962922771, "learning_rate": 5.798326892573488e-05, "loss": 2.7944, "step": 24360 }, { "epoch": 1.512260227202185, "grad_norm": 0.5612724473028392, "learning_rate": 5.797970371227468e-05, "loss": 2.7464, "step": 24361 }, { "epoch": 1.512322304301943, "grad_norm": 0.5523230434135294, "learning_rate": 5.7976138457182414e-05, "loss": 2.7442, "step": 24362 }, { "epoch": 1.5123843814017008, "grad_norm": 0.5913549101099534, "learning_rate": 5.797257316047671e-05, "loss": 2.758, "step": 24363 }, { "epoch": 1.5124464585014588, "grad_norm": 0.6473485236299202, "learning_rate": 5.7969007822176155e-05, "loss": 2.8136, "step": 24364 }, { "epoch": 1.5125085356012167, "grad_norm": 0.5422600986126013, "learning_rate": 5.796544244229934e-05, "loss": 2.7779, "step": 24365 }, { "epoch": 1.5125706127009746, "grad_norm": 0.5660901733335988, "learning_rate": 5.796187702086491e-05, "loss": 2.7203, "step": 24366 }, { "epoch": 1.5126326898007325, "grad_norm": 0.5151922837793899, "learning_rate": 5.79583115578914e-05, "loss": 2.7358, "step": 24367 }, { "epoch": 1.5126947669004904, "grad_norm": 0.5851781374984896, "learning_rate": 5.795474605339747e-05, "loss": 2.781, "step": 24368 }, { "epoch": 1.5127568440002483, "grad_norm": 0.5730690377805131, "learning_rate": 5.795118050740169e-05, "loss": 2.708, "step": 24369 }, { "epoch": 1.5128189211000063, "grad_norm": 0.48206353102655, "learning_rate": 5.7947614919922666e-05, "loss": 2.8427, "step": 24370 }, { "epoch": 1.5128809981997642, "grad_norm": 0.6609488700447839, "learning_rate": 5.7944049290979006e-05, "loss": 2.8236, "step": 24371 }, { "epoch": 1.512943075299522, "grad_norm": 0.5267426915415367, "learning_rate": 5.794048362058931e-05, "loss": 2.8048, "step": 24372 }, { "epoch": 1.51300515239928, "grad_norm": 0.6233242218818714, "learning_rate": 5.79369179087722e-05, "loss": 2.767, "step": 24373 }, { "epoch": 1.513067229499038, "grad_norm": 0.5360042508209399, "learning_rate": 5.793335215554625e-05, "loss": 2.7974, "step": 24374 }, { "epoch": 1.5131293065987959, "grad_norm": 0.6034669331781088, "learning_rate": 5.792978636093008e-05, "loss": 2.8125, "step": 24375 }, { "epoch": 1.5131913836985535, "grad_norm": 0.5733996667256899, "learning_rate": 5.792622052494229e-05, "loss": 2.851, "step": 24376 }, { "epoch": 1.5132534607983115, "grad_norm": 0.5700103011371646, "learning_rate": 5.792265464760148e-05, "loss": 2.7673, "step": 24377 }, { "epoch": 1.5133155378980694, "grad_norm": 0.593767165889554, "learning_rate": 5.791908872892625e-05, "loss": 2.8961, "step": 24378 }, { "epoch": 1.5133776149978273, "grad_norm": 0.5409902298127935, "learning_rate": 5.7915522768935213e-05, "loss": 2.803, "step": 24379 }, { "epoch": 1.5134396920975852, "grad_norm": 0.5682682344437097, "learning_rate": 5.791195676764699e-05, "loss": 2.7792, "step": 24380 }, { "epoch": 1.5135017691973431, "grad_norm": 0.571951092936701, "learning_rate": 5.7908390725080154e-05, "loss": 2.8001, "step": 24381 }, { "epoch": 1.5135638462971008, "grad_norm": 0.571664383673303, "learning_rate": 5.790482464125333e-05, "loss": 2.7353, "step": 24382 }, { "epoch": 1.5136259233968588, "grad_norm": 0.5432333267311082, "learning_rate": 5.790125851618511e-05, "loss": 2.7132, "step": 24383 }, { "epoch": 1.5136880004966167, "grad_norm": 0.5949793436568035, "learning_rate": 5.789769234989411e-05, "loss": 2.8574, "step": 24384 }, { "epoch": 1.5137500775963746, "grad_norm": 0.5734488847314418, "learning_rate": 5.7894126142398933e-05, "loss": 2.7809, "step": 24385 }, { "epoch": 1.5138121546961325, "grad_norm": 0.5909224240676143, "learning_rate": 5.789055989371818e-05, "loss": 2.7787, "step": 24386 }, { "epoch": 1.5138742317958904, "grad_norm": 0.6265820745135955, "learning_rate": 5.788699360387047e-05, "loss": 2.768, "step": 24387 }, { "epoch": 1.5139363088956483, "grad_norm": 0.5604013925430237, "learning_rate": 5.788342727287439e-05, "loss": 2.8226, "step": 24388 }, { "epoch": 1.5139983859954063, "grad_norm": 0.6212422895048977, "learning_rate": 5.787986090074856e-05, "loss": 2.7894, "step": 24389 }, { "epoch": 1.5140604630951642, "grad_norm": 0.5516874825544691, "learning_rate": 5.787629448751157e-05, "loss": 2.8595, "step": 24390 }, { "epoch": 1.514122540194922, "grad_norm": 0.6085685381974214, "learning_rate": 5.7872728033182064e-05, "loss": 2.8246, "step": 24391 }, { "epoch": 1.51418461729468, "grad_norm": 0.5484134560120882, "learning_rate": 5.786916153777861e-05, "loss": 2.7811, "step": 24392 }, { "epoch": 1.514246694394438, "grad_norm": 0.5589654925960519, "learning_rate": 5.786559500131983e-05, "loss": 2.8376, "step": 24393 }, { "epoch": 1.5143087714941958, "grad_norm": 0.5727590315151252, "learning_rate": 5.786202842382432e-05, "loss": 2.7249, "step": 24394 }, { "epoch": 1.5143708485939538, "grad_norm": 0.5800136440415196, "learning_rate": 5.7858461805310715e-05, "loss": 2.7327, "step": 24395 }, { "epoch": 1.5144329256937117, "grad_norm": 0.5549624661478064, "learning_rate": 5.7854895145797606e-05, "loss": 2.8268, "step": 24396 }, { "epoch": 1.5144950027934696, "grad_norm": 0.596702163198062, "learning_rate": 5.78513284453036e-05, "loss": 2.8309, "step": 24397 }, { "epoch": 1.5145570798932275, "grad_norm": 0.5408298986156393, "learning_rate": 5.784776170384731e-05, "loss": 2.7315, "step": 24398 }, { "epoch": 1.5146191569929854, "grad_norm": 0.5740706324118103, "learning_rate": 5.7844194921447334e-05, "loss": 2.7763, "step": 24399 }, { "epoch": 1.5146812340927431, "grad_norm": 0.5433119642798533, "learning_rate": 5.7840628098122295e-05, "loss": 2.8157, "step": 24400 }, { "epoch": 1.514743311192501, "grad_norm": 0.618917888444289, "learning_rate": 5.783706123389079e-05, "loss": 2.842, "step": 24401 }, { "epoch": 1.514805388292259, "grad_norm": 0.5846126073339841, "learning_rate": 5.783349432877145e-05, "loss": 2.7787, "step": 24402 }, { "epoch": 1.5148674653920169, "grad_norm": 0.6112249610312646, "learning_rate": 5.7829927382782856e-05, "loss": 2.7864, "step": 24403 }, { "epoch": 1.5149295424917748, "grad_norm": 0.6056820155450355, "learning_rate": 5.782636039594364e-05, "loss": 2.7662, "step": 24404 }, { "epoch": 1.5149916195915327, "grad_norm": 0.5279008818687506, "learning_rate": 5.7822793368272374e-05, "loss": 2.7757, "step": 24405 }, { "epoch": 1.5150536966912904, "grad_norm": 0.5423495652335699, "learning_rate": 5.781922629978773e-05, "loss": 2.7904, "step": 24406 }, { "epoch": 1.5151157737910483, "grad_norm": 0.5717047328434789, "learning_rate": 5.781565919050828e-05, "loss": 2.7511, "step": 24407 }, { "epoch": 1.5151778508908063, "grad_norm": 0.520290193280047, "learning_rate": 5.781209204045261e-05, "loss": 2.7585, "step": 24408 }, { "epoch": 1.5152399279905642, "grad_norm": 0.5896194486261684, "learning_rate": 5.780852484963939e-05, "loss": 2.778, "step": 24409 }, { "epoch": 1.515302005090322, "grad_norm": 0.5291117946329597, "learning_rate": 5.780495761808719e-05, "loss": 2.7373, "step": 24410 }, { "epoch": 1.51536408219008, "grad_norm": 0.5822128678188959, "learning_rate": 5.780139034581462e-05, "loss": 2.7728, "step": 24411 }, { "epoch": 1.515426159289838, "grad_norm": 0.555012164669924, "learning_rate": 5.7797823032840315e-05, "loss": 2.6907, "step": 24412 }, { "epoch": 1.5154882363895958, "grad_norm": 0.6007810198283654, "learning_rate": 5.7794255679182866e-05, "loss": 2.8073, "step": 24413 }, { "epoch": 1.5155503134893538, "grad_norm": 0.6232742086162442, "learning_rate": 5.77906882848609e-05, "loss": 2.8091, "step": 24414 }, { "epoch": 1.5156123905891117, "grad_norm": 0.5889744194414892, "learning_rate": 5.7787120849893014e-05, "loss": 2.8278, "step": 24415 }, { "epoch": 1.5156744676888696, "grad_norm": 0.5597526958753333, "learning_rate": 5.778355337429785e-05, "loss": 2.8183, "step": 24416 }, { "epoch": 1.5157365447886275, "grad_norm": 0.5821862238461999, "learning_rate": 5.7779985858093976e-05, "loss": 2.7484, "step": 24417 }, { "epoch": 1.5157986218883854, "grad_norm": 0.511492788497041, "learning_rate": 5.777641830130004e-05, "loss": 2.8445, "step": 24418 }, { "epoch": 1.5158606989881434, "grad_norm": 0.5370324374494613, "learning_rate": 5.777285070393463e-05, "loss": 2.8224, "step": 24419 }, { "epoch": 1.5159227760879013, "grad_norm": 0.5420761730660457, "learning_rate": 5.7769283066016375e-05, "loss": 2.7757, "step": 24420 }, { "epoch": 1.5159848531876592, "grad_norm": 0.5359177010218306, "learning_rate": 5.776571538756389e-05, "loss": 2.7949, "step": 24421 }, { "epoch": 1.516046930287417, "grad_norm": 0.5287635340209786, "learning_rate": 5.7762147668595776e-05, "loss": 2.6849, "step": 24422 }, { "epoch": 1.516109007387175, "grad_norm": 0.5319732903322234, "learning_rate": 5.775857990913066e-05, "loss": 2.7439, "step": 24423 }, { "epoch": 1.5161710844869327, "grad_norm": 0.5577042172688197, "learning_rate": 5.7755012109187146e-05, "loss": 2.7952, "step": 24424 }, { "epoch": 1.5162331615866906, "grad_norm": 0.528902364291141, "learning_rate": 5.775144426878386e-05, "loss": 2.7473, "step": 24425 }, { "epoch": 1.5162952386864486, "grad_norm": 0.5101401754420113, "learning_rate": 5.774787638793939e-05, "loss": 2.7629, "step": 24426 }, { "epoch": 1.5163573157862065, "grad_norm": 0.5465366134053206, "learning_rate": 5.774430846667237e-05, "loss": 2.8197, "step": 24427 }, { "epoch": 1.5164193928859644, "grad_norm": 0.5897671479258428, "learning_rate": 5.774074050500141e-05, "loss": 2.8131, "step": 24428 }, { "epoch": 1.5164814699857223, "grad_norm": 0.5386000875172648, "learning_rate": 5.773717250294515e-05, "loss": 2.725, "step": 24429 }, { "epoch": 1.51654354708548, "grad_norm": 0.5690441659888914, "learning_rate": 5.773360446052216e-05, "loss": 2.8234, "step": 24430 }, { "epoch": 1.516605624185238, "grad_norm": 0.6025981410414101, "learning_rate": 5.77300363777511e-05, "loss": 2.7717, "step": 24431 }, { "epoch": 1.5166677012849958, "grad_norm": 0.6309444263865944, "learning_rate": 5.772646825465056e-05, "loss": 2.7596, "step": 24432 }, { "epoch": 1.5167297783847538, "grad_norm": 0.5351604823481134, "learning_rate": 5.772290009123914e-05, "loss": 2.7436, "step": 24433 }, { "epoch": 1.5167918554845117, "grad_norm": 0.5552014132766847, "learning_rate": 5.771933188753549e-05, "loss": 2.6991, "step": 24434 }, { "epoch": 1.5168539325842696, "grad_norm": 0.5948320617810047, "learning_rate": 5.77157636435582e-05, "loss": 2.8334, "step": 24435 }, { "epoch": 1.5169160096840275, "grad_norm": 0.5829088647859736, "learning_rate": 5.771219535932592e-05, "loss": 2.8682, "step": 24436 }, { "epoch": 1.5169780867837854, "grad_norm": 0.5795437698883757, "learning_rate": 5.770862703485722e-05, "loss": 2.7977, "step": 24437 }, { "epoch": 1.5170401638835433, "grad_norm": 0.6167005235381833, "learning_rate": 5.7705058670170756e-05, "loss": 2.7935, "step": 24438 }, { "epoch": 1.5171022409833013, "grad_norm": 0.5374860795014906, "learning_rate": 5.770149026528512e-05, "loss": 2.8084, "step": 24439 }, { "epoch": 1.5171643180830592, "grad_norm": 0.6135786270986243, "learning_rate": 5.7697921820218946e-05, "loss": 2.779, "step": 24440 }, { "epoch": 1.517226395182817, "grad_norm": 0.5815353328994654, "learning_rate": 5.769435333499085e-05, "loss": 2.7914, "step": 24441 }, { "epoch": 1.517288472282575, "grad_norm": 0.6575495704124451, "learning_rate": 5.769078480961944e-05, "loss": 2.7295, "step": 24442 }, { "epoch": 1.517350549382333, "grad_norm": 0.6245561215599784, "learning_rate": 5.7687216244123335e-05, "loss": 2.7868, "step": 24443 }, { "epoch": 1.5174126264820909, "grad_norm": 0.6794122379699362, "learning_rate": 5.768364763852116e-05, "loss": 2.8178, "step": 24444 }, { "epoch": 1.5174747035818488, "grad_norm": 0.6466530064200208, "learning_rate": 5.768007899283153e-05, "loss": 2.7901, "step": 24445 }, { "epoch": 1.5175367806816067, "grad_norm": 0.5775164932934296, "learning_rate": 5.767651030707306e-05, "loss": 2.8261, "step": 24446 }, { "epoch": 1.5175988577813646, "grad_norm": 0.615351081556884, "learning_rate": 5.767294158126437e-05, "loss": 2.692, "step": 24447 }, { "epoch": 1.5176609348811223, "grad_norm": 0.6237945614494207, "learning_rate": 5.7669372815424094e-05, "loss": 2.8209, "step": 24448 }, { "epoch": 1.5177230119808802, "grad_norm": 0.643130242054802, "learning_rate": 5.766580400957083e-05, "loss": 2.8131, "step": 24449 }, { "epoch": 1.5177850890806381, "grad_norm": 0.588650817089442, "learning_rate": 5.766223516372321e-05, "loss": 2.7955, "step": 24450 }, { "epoch": 1.517847166180396, "grad_norm": 0.5916297349238614, "learning_rate": 5.765866627789984e-05, "loss": 2.7726, "step": 24451 }, { "epoch": 1.517909243280154, "grad_norm": 0.6124215143871548, "learning_rate": 5.765509735211937e-05, "loss": 2.8572, "step": 24452 }, { "epoch": 1.517971320379912, "grad_norm": 0.6463973571667253, "learning_rate": 5.7651528386400376e-05, "loss": 2.8106, "step": 24453 }, { "epoch": 1.5180333974796696, "grad_norm": 0.5722183103823769, "learning_rate": 5.7647959380761505e-05, "loss": 2.762, "step": 24454 }, { "epoch": 1.5180954745794275, "grad_norm": 0.5782505363052857, "learning_rate": 5.764439033522138e-05, "loss": 2.7692, "step": 24455 }, { "epoch": 1.5181575516791854, "grad_norm": 0.5931060963664193, "learning_rate": 5.764082124979862e-05, "loss": 2.8357, "step": 24456 }, { "epoch": 1.5182196287789433, "grad_norm": 0.5655053965954414, "learning_rate": 5.763725212451182e-05, "loss": 2.8036, "step": 24457 }, { "epoch": 1.5182817058787013, "grad_norm": 0.6150515087778713, "learning_rate": 5.7633682959379644e-05, "loss": 2.8697, "step": 24458 }, { "epoch": 1.5183437829784592, "grad_norm": 0.5853900895472, "learning_rate": 5.763011375442067e-05, "loss": 2.8197, "step": 24459 }, { "epoch": 1.518405860078217, "grad_norm": 0.5825179196740726, "learning_rate": 5.7626544509653546e-05, "loss": 2.8289, "step": 24460 }, { "epoch": 1.518467937177975, "grad_norm": 0.5296339531223456, "learning_rate": 5.76229752250969e-05, "loss": 2.7319, "step": 24461 }, { "epoch": 1.518530014277733, "grad_norm": 0.5992072182631187, "learning_rate": 5.761940590076933e-05, "loss": 2.8822, "step": 24462 }, { "epoch": 1.5185920913774908, "grad_norm": 0.5128234703104007, "learning_rate": 5.7615836536689484e-05, "loss": 2.7566, "step": 24463 }, { "epoch": 1.5186541684772488, "grad_norm": 0.5971908003527732, "learning_rate": 5.761226713287595e-05, "loss": 2.727, "step": 24464 }, { "epoch": 1.5187162455770067, "grad_norm": 0.5321285052480883, "learning_rate": 5.760869768934739e-05, "loss": 2.7951, "step": 24465 }, { "epoch": 1.5187783226767646, "grad_norm": 0.5768868875156974, "learning_rate": 5.760512820612238e-05, "loss": 2.8052, "step": 24466 }, { "epoch": 1.5188403997765225, "grad_norm": 0.5669642626501707, "learning_rate": 5.76015586832196e-05, "loss": 2.8164, "step": 24467 }, { "epoch": 1.5189024768762804, "grad_norm": 0.5567016568103912, "learning_rate": 5.759798912065764e-05, "loss": 2.8475, "step": 24468 }, { "epoch": 1.5189645539760384, "grad_norm": 0.5406576411365674, "learning_rate": 5.759441951845511e-05, "loss": 2.6571, "step": 24469 }, { "epoch": 1.5190266310757963, "grad_norm": 0.5801458351357265, "learning_rate": 5.759084987663066e-05, "loss": 2.7898, "step": 24470 }, { "epoch": 1.5190887081755542, "grad_norm": 0.5672777771367232, "learning_rate": 5.75872801952029e-05, "loss": 2.8178, "step": 24471 }, { "epoch": 1.5191507852753119, "grad_norm": 0.6048437276498746, "learning_rate": 5.758371047419045e-05, "loss": 2.8387, "step": 24472 }, { "epoch": 1.5192128623750698, "grad_norm": 0.5847647310809462, "learning_rate": 5.7580140713611953e-05, "loss": 2.7161, "step": 24473 }, { "epoch": 1.5192749394748277, "grad_norm": 0.5624882643852956, "learning_rate": 5.757657091348603e-05, "loss": 2.8684, "step": 24474 }, { "epoch": 1.5193370165745856, "grad_norm": 0.6104001933751231, "learning_rate": 5.7573001073831276e-05, "loss": 2.7444, "step": 24475 }, { "epoch": 1.5193990936743436, "grad_norm": 0.574814130969283, "learning_rate": 5.7569431194666335e-05, "loss": 2.7775, "step": 24476 }, { "epoch": 1.5194611707741015, "grad_norm": 0.6266556982042637, "learning_rate": 5.756586127600986e-05, "loss": 2.7707, "step": 24477 }, { "epoch": 1.5195232478738592, "grad_norm": 0.5239345979970812, "learning_rate": 5.7562291317880425e-05, "loss": 2.7673, "step": 24478 }, { "epoch": 1.519585324973617, "grad_norm": 0.5935963381645329, "learning_rate": 5.755872132029669e-05, "loss": 2.816, "step": 24479 }, { "epoch": 1.519647402073375, "grad_norm": 0.5403269436760644, "learning_rate": 5.755515128327727e-05, "loss": 2.8341, "step": 24480 }, { "epoch": 1.519709479173133, "grad_norm": 0.5518531919428015, "learning_rate": 5.75515812068408e-05, "loss": 2.8305, "step": 24481 }, { "epoch": 1.5197715562728908, "grad_norm": 0.5600293791251008, "learning_rate": 5.754801109100588e-05, "loss": 2.7406, "step": 24482 }, { "epoch": 1.5198336333726488, "grad_norm": 0.563811125179907, "learning_rate": 5.754444093579118e-05, "loss": 2.8328, "step": 24483 }, { "epoch": 1.5198957104724067, "grad_norm": 0.588086548662074, "learning_rate": 5.754087074121527e-05, "loss": 2.8071, "step": 24484 }, { "epoch": 1.5199577875721646, "grad_norm": 0.4877712179991747, "learning_rate": 5.753730050729683e-05, "loss": 2.7257, "step": 24485 }, { "epoch": 1.5200198646719225, "grad_norm": 0.6072368524022485, "learning_rate": 5.7533730234054465e-05, "loss": 2.853, "step": 24486 }, { "epoch": 1.5200819417716804, "grad_norm": 0.5681010442452552, "learning_rate": 5.7530159921506776e-05, "loss": 2.764, "step": 24487 }, { "epoch": 1.5201440188714384, "grad_norm": 0.558758077925763, "learning_rate": 5.752658956967244e-05, "loss": 2.7321, "step": 24488 }, { "epoch": 1.5202060959711963, "grad_norm": 0.5675665710491407, "learning_rate": 5.752301917857004e-05, "loss": 2.8288, "step": 24489 }, { "epoch": 1.5202681730709542, "grad_norm": 0.5622250361737574, "learning_rate": 5.751944874821824e-05, "loss": 2.7932, "step": 24490 }, { "epoch": 1.520330250170712, "grad_norm": 0.6565740103350418, "learning_rate": 5.751587827863564e-05, "loss": 2.8462, "step": 24491 }, { "epoch": 1.52039232727047, "grad_norm": 0.5460440453433035, "learning_rate": 5.751230776984088e-05, "loss": 2.8636, "step": 24492 }, { "epoch": 1.520454404370228, "grad_norm": 0.6646295706789277, "learning_rate": 5.750873722185258e-05, "loss": 2.7295, "step": 24493 }, { "epoch": 1.5205164814699859, "grad_norm": 0.6053715289416336, "learning_rate": 5.750516663468938e-05, "loss": 2.8332, "step": 24494 }, { "epoch": 1.5205785585697438, "grad_norm": 0.6176570120511709, "learning_rate": 5.7501596008369906e-05, "loss": 2.8306, "step": 24495 }, { "epoch": 1.5206406356695015, "grad_norm": 0.6778196781353466, "learning_rate": 5.749802534291278e-05, "loss": 2.8447, "step": 24496 }, { "epoch": 1.5207027127692594, "grad_norm": 0.6078222984955899, "learning_rate": 5.749445463833665e-05, "loss": 2.8037, "step": 24497 }, { "epoch": 1.5207647898690173, "grad_norm": 0.6194648735980816, "learning_rate": 5.749088389466011e-05, "loss": 2.764, "step": 24498 }, { "epoch": 1.5208268669687752, "grad_norm": 0.6322676585466533, "learning_rate": 5.748731311190183e-05, "loss": 2.7889, "step": 24499 }, { "epoch": 1.5208889440685331, "grad_norm": 0.6174264517702216, "learning_rate": 5.74837422900804e-05, "loss": 2.8253, "step": 24500 }, { "epoch": 1.520951021168291, "grad_norm": 0.605390432891638, "learning_rate": 5.748017142921448e-05, "loss": 2.8268, "step": 24501 }, { "epoch": 1.5210130982680488, "grad_norm": 0.6268869050132516, "learning_rate": 5.747660052932269e-05, "loss": 2.8784, "step": 24502 }, { "epoch": 1.5210751753678067, "grad_norm": 0.6373723659651725, "learning_rate": 5.747302959042367e-05, "loss": 2.7823, "step": 24503 }, { "epoch": 1.5211372524675646, "grad_norm": 0.6337249911563552, "learning_rate": 5.746945861253603e-05, "loss": 2.7637, "step": 24504 }, { "epoch": 1.5211993295673225, "grad_norm": 0.5797073772138025, "learning_rate": 5.74658875956784e-05, "loss": 2.8998, "step": 24505 }, { "epoch": 1.5212614066670804, "grad_norm": 0.6260520901396724, "learning_rate": 5.746231653986944e-05, "loss": 2.8734, "step": 24506 }, { "epoch": 1.5213234837668383, "grad_norm": 0.5713677506989457, "learning_rate": 5.745874544512776e-05, "loss": 2.846, "step": 24507 }, { "epoch": 1.5213855608665963, "grad_norm": 0.6366310059170313, "learning_rate": 5.745517431147199e-05, "loss": 2.8661, "step": 24508 }, { "epoch": 1.5214476379663542, "grad_norm": 0.614694703781822, "learning_rate": 5.745160313892075e-05, "loss": 2.8006, "step": 24509 }, { "epoch": 1.521509715066112, "grad_norm": 0.5917135501025684, "learning_rate": 5.744803192749271e-05, "loss": 2.7561, "step": 24510 }, { "epoch": 1.52157179216587, "grad_norm": 0.6011461080143887, "learning_rate": 5.744446067720648e-05, "loss": 2.7845, "step": 24511 }, { "epoch": 1.521633869265628, "grad_norm": 0.6119587482265624, "learning_rate": 5.744088938808068e-05, "loss": 2.7868, "step": 24512 }, { "epoch": 1.5216959463653859, "grad_norm": 0.579694036359515, "learning_rate": 5.743731806013397e-05, "loss": 2.824, "step": 24513 }, { "epoch": 1.5217580234651438, "grad_norm": 0.6364177390189396, "learning_rate": 5.743374669338495e-05, "loss": 2.8405, "step": 24514 }, { "epoch": 1.5218201005649017, "grad_norm": 0.6601988648874901, "learning_rate": 5.743017528785227e-05, "loss": 2.8247, "step": 24515 }, { "epoch": 1.5218821776646596, "grad_norm": 0.5630007475294093, "learning_rate": 5.742660384355456e-05, "loss": 2.7224, "step": 24516 }, { "epoch": 1.5219442547644175, "grad_norm": 0.5964180830195638, "learning_rate": 5.742303236051047e-05, "loss": 2.7171, "step": 24517 }, { "epoch": 1.5220063318641754, "grad_norm": 0.6003719537893195, "learning_rate": 5.74194608387386e-05, "loss": 2.6465, "step": 24518 }, { "epoch": 1.5220684089639334, "grad_norm": 0.6077717451146429, "learning_rate": 5.741588927825762e-05, "loss": 2.8599, "step": 24519 }, { "epoch": 1.522130486063691, "grad_norm": 0.6049535959354608, "learning_rate": 5.741231767908613e-05, "loss": 2.7973, "step": 24520 }, { "epoch": 1.522192563163449, "grad_norm": 0.5891261756945136, "learning_rate": 5.740874604124279e-05, "loss": 2.7366, "step": 24521 }, { "epoch": 1.522254640263207, "grad_norm": 0.5705241502971062, "learning_rate": 5.740517436474623e-05, "loss": 2.8221, "step": 24522 }, { "epoch": 1.5223167173629648, "grad_norm": 0.5936875310460689, "learning_rate": 5.740160264961505e-05, "loss": 2.8133, "step": 24523 }, { "epoch": 1.5223787944627227, "grad_norm": 0.5734600322568599, "learning_rate": 5.739803089586794e-05, "loss": 2.8266, "step": 24524 }, { "epoch": 1.5224408715624806, "grad_norm": 0.6362178212780237, "learning_rate": 5.739445910352348e-05, "loss": 2.8167, "step": 24525 }, { "epoch": 1.5225029486622383, "grad_norm": 0.578504515134885, "learning_rate": 5.7390887272600347e-05, "loss": 2.7647, "step": 24526 }, { "epoch": 1.5225650257619963, "grad_norm": 0.5839017726694891, "learning_rate": 5.738731540311716e-05, "loss": 2.836, "step": 24527 }, { "epoch": 1.5226271028617542, "grad_norm": 0.803083766071704, "learning_rate": 5.7383743495092555e-05, "loss": 2.7952, "step": 24528 }, { "epoch": 1.522689179961512, "grad_norm": 0.6814236008521941, "learning_rate": 5.738017154854518e-05, "loss": 2.715, "step": 24529 }, { "epoch": 1.52275125706127, "grad_norm": 0.7232896245841194, "learning_rate": 5.737659956349364e-05, "loss": 2.7632, "step": 24530 }, { "epoch": 1.522813334161028, "grad_norm": 0.6446878416399806, "learning_rate": 5.7373027539956604e-05, "loss": 2.7299, "step": 24531 }, { "epoch": 1.5228754112607858, "grad_norm": 0.6093429410277514, "learning_rate": 5.736945547795268e-05, "loss": 2.7731, "step": 24532 }, { "epoch": 1.5229374883605438, "grad_norm": 0.6881995759929425, "learning_rate": 5.736588337750053e-05, "loss": 2.8291, "step": 24533 }, { "epoch": 1.5229995654603017, "grad_norm": 0.6208798982024972, "learning_rate": 5.736231123861876e-05, "loss": 2.8348, "step": 24534 }, { "epoch": 1.5230616425600596, "grad_norm": 0.6954122750349515, "learning_rate": 5.735873906132605e-05, "loss": 2.8246, "step": 24535 }, { "epoch": 1.5231237196598175, "grad_norm": 0.6095126314505059, "learning_rate": 5.7355166845641e-05, "loss": 2.6912, "step": 24536 }, { "epoch": 1.5231857967595754, "grad_norm": 0.6368200771469923, "learning_rate": 5.7351594591582256e-05, "loss": 2.6753, "step": 24537 }, { "epoch": 1.5232478738593334, "grad_norm": 0.6256523780662933, "learning_rate": 5.7348022299168446e-05, "loss": 2.8223, "step": 24538 }, { "epoch": 1.5233099509590913, "grad_norm": 0.6469700996780265, "learning_rate": 5.7344449968418244e-05, "loss": 2.7885, "step": 24539 }, { "epoch": 1.5233720280588492, "grad_norm": 0.6208388327241926, "learning_rate": 5.734087759935026e-05, "loss": 2.7859, "step": 24540 }, { "epoch": 1.523434105158607, "grad_norm": 0.638178293429417, "learning_rate": 5.7337305191983126e-05, "loss": 2.7732, "step": 24541 }, { "epoch": 1.523496182258365, "grad_norm": 0.5717811486616835, "learning_rate": 5.7333732746335486e-05, "loss": 2.9052, "step": 24542 }, { "epoch": 1.523558259358123, "grad_norm": 0.6181018912582827, "learning_rate": 5.733016026242598e-05, "loss": 2.7269, "step": 24543 }, { "epoch": 1.5236203364578806, "grad_norm": 0.5639866680199579, "learning_rate": 5.7326587740273266e-05, "loss": 2.7628, "step": 24544 }, { "epoch": 1.5236824135576386, "grad_norm": 0.6277180221507059, "learning_rate": 5.7323015179895955e-05, "loss": 2.7937, "step": 24545 }, { "epoch": 1.5237444906573965, "grad_norm": 0.6647405640691691, "learning_rate": 5.73194425813127e-05, "loss": 2.7806, "step": 24546 }, { "epoch": 1.5238065677571544, "grad_norm": 0.582463737431432, "learning_rate": 5.731586994454213e-05, "loss": 2.8084, "step": 24547 }, { "epoch": 1.5238686448569123, "grad_norm": 0.6066505220076276, "learning_rate": 5.73122972696029e-05, "loss": 2.6921, "step": 24548 }, { "epoch": 1.5239307219566702, "grad_norm": 0.5381482152328518, "learning_rate": 5.730872455651364e-05, "loss": 2.8294, "step": 24549 }, { "epoch": 1.523992799056428, "grad_norm": 0.6316799578572634, "learning_rate": 5.730515180529298e-05, "loss": 2.7516, "step": 24550 }, { "epoch": 1.5240548761561858, "grad_norm": 0.5639716870441817, "learning_rate": 5.7301579015959574e-05, "loss": 2.6797, "step": 24551 }, { "epoch": 1.5241169532559438, "grad_norm": 0.6475154111567312, "learning_rate": 5.729800618853205e-05, "loss": 2.8107, "step": 24552 }, { "epoch": 1.5241790303557017, "grad_norm": 0.5832471244551455, "learning_rate": 5.7294433323029074e-05, "loss": 2.7523, "step": 24553 }, { "epoch": 1.5242411074554596, "grad_norm": 0.6036105775672289, "learning_rate": 5.729086041946925e-05, "loss": 2.8816, "step": 24554 }, { "epoch": 1.5243031845552175, "grad_norm": 0.5735647668376986, "learning_rate": 5.7287287477871256e-05, "loss": 2.7531, "step": 24555 }, { "epoch": 1.5243652616549754, "grad_norm": 0.626355774944691, "learning_rate": 5.7283714498253694e-05, "loss": 2.6754, "step": 24556 }, { "epoch": 1.5244273387547334, "grad_norm": 0.5679485763472184, "learning_rate": 5.728014148063524e-05, "loss": 2.7732, "step": 24557 }, { "epoch": 1.5244894158544913, "grad_norm": 0.5659491437433517, "learning_rate": 5.7276568425034505e-05, "loss": 2.8282, "step": 24558 }, { "epoch": 1.5245514929542492, "grad_norm": 0.5883596583340258, "learning_rate": 5.727299533147016e-05, "loss": 2.7853, "step": 24559 }, { "epoch": 1.524613570054007, "grad_norm": 0.5557996587785193, "learning_rate": 5.7269422199960834e-05, "loss": 2.7123, "step": 24560 }, { "epoch": 1.524675647153765, "grad_norm": 0.5751554931950688, "learning_rate": 5.7265849030525154e-05, "loss": 2.7902, "step": 24561 }, { "epoch": 1.524737724253523, "grad_norm": 0.5870071058103171, "learning_rate": 5.726227582318179e-05, "loss": 2.7159, "step": 24562 }, { "epoch": 1.5247998013532809, "grad_norm": 0.5944665048963224, "learning_rate": 5.7258702577949364e-05, "loss": 2.7791, "step": 24563 }, { "epoch": 1.5248618784530388, "grad_norm": 0.590557575918665, "learning_rate": 5.7255129294846524e-05, "loss": 2.7505, "step": 24564 }, { "epoch": 1.5249239555527967, "grad_norm": 0.5358370606197371, "learning_rate": 5.725155597389192e-05, "loss": 2.8249, "step": 24565 }, { "epoch": 1.5249860326525546, "grad_norm": 0.575345212513001, "learning_rate": 5.724798261510418e-05, "loss": 2.7779, "step": 24566 }, { "epoch": 1.5250481097523125, "grad_norm": 0.577026793284266, "learning_rate": 5.724440921850196e-05, "loss": 2.7314, "step": 24567 }, { "epoch": 1.5251101868520702, "grad_norm": 0.5599115472998508, "learning_rate": 5.7240835784103886e-05, "loss": 2.7739, "step": 24568 }, { "epoch": 1.5251722639518281, "grad_norm": 0.5347377783781806, "learning_rate": 5.723726231192863e-05, "loss": 2.8568, "step": 24569 }, { "epoch": 1.525234341051586, "grad_norm": 0.5012361920802242, "learning_rate": 5.723368880199481e-05, "loss": 2.8452, "step": 24570 }, { "epoch": 1.525296418151344, "grad_norm": 0.5485423148219356, "learning_rate": 5.7230115254321094e-05, "loss": 2.748, "step": 24571 }, { "epoch": 1.525358495251102, "grad_norm": 0.5730081889714107, "learning_rate": 5.722654166892609e-05, "loss": 2.7807, "step": 24572 }, { "epoch": 1.5254205723508598, "grad_norm": 0.533584067395049, "learning_rate": 5.7222968045828484e-05, "loss": 2.7769, "step": 24573 }, { "epoch": 1.5254826494506175, "grad_norm": 0.646973853919882, "learning_rate": 5.721939438504689e-05, "loss": 2.8294, "step": 24574 }, { "epoch": 1.5255447265503754, "grad_norm": 0.5408436195363239, "learning_rate": 5.721582068659995e-05, "loss": 2.8623, "step": 24575 }, { "epoch": 1.5256068036501333, "grad_norm": 0.5993830598561747, "learning_rate": 5.721224695050634e-05, "loss": 2.8555, "step": 24576 }, { "epoch": 1.5256688807498913, "grad_norm": 0.5927033987884617, "learning_rate": 5.720867317678468e-05, "loss": 2.7922, "step": 24577 }, { "epoch": 1.5257309578496492, "grad_norm": 0.5089158323075096, "learning_rate": 5.720509936545363e-05, "loss": 2.6955, "step": 24578 }, { "epoch": 1.525793034949407, "grad_norm": 0.5724548703942318, "learning_rate": 5.720152551653182e-05, "loss": 2.803, "step": 24579 }, { "epoch": 1.525855112049165, "grad_norm": 0.5782798611228197, "learning_rate": 5.71979516300379e-05, "loss": 2.8049, "step": 24580 }, { "epoch": 1.525917189148923, "grad_norm": 0.5523571978581671, "learning_rate": 5.719437770599052e-05, "loss": 2.8544, "step": 24581 }, { "epoch": 1.5259792662486809, "grad_norm": 0.611095786504588, "learning_rate": 5.7190803744408336e-05, "loss": 2.8327, "step": 24582 }, { "epoch": 1.5260413433484388, "grad_norm": 0.5598905755321049, "learning_rate": 5.718722974530999e-05, "loss": 2.7745, "step": 24583 }, { "epoch": 1.5261034204481967, "grad_norm": 0.5866733762057584, "learning_rate": 5.71836557087141e-05, "loss": 2.7709, "step": 24584 }, { "epoch": 1.5261654975479546, "grad_norm": 0.6280842497403363, "learning_rate": 5.718008163463935e-05, "loss": 2.7373, "step": 24585 }, { "epoch": 1.5262275746477125, "grad_norm": 0.607821513994154, "learning_rate": 5.717650752310435e-05, "loss": 2.7328, "step": 24586 }, { "epoch": 1.5262896517474704, "grad_norm": 0.6203460030912834, "learning_rate": 5.717293337412779e-05, "loss": 2.767, "step": 24587 }, { "epoch": 1.5263517288472284, "grad_norm": 0.5788838329427948, "learning_rate": 5.7169359187728274e-05, "loss": 2.811, "step": 24588 }, { "epoch": 1.5264138059469863, "grad_norm": 0.6378935524755931, "learning_rate": 5.7165784963924494e-05, "loss": 2.7998, "step": 24589 }, { "epoch": 1.5264758830467442, "grad_norm": 0.5415540969469661, "learning_rate": 5.7162210702735064e-05, "loss": 2.8116, "step": 24590 }, { "epoch": 1.5265379601465021, "grad_norm": 0.652983734148801, "learning_rate": 5.7158636404178634e-05, "loss": 2.8552, "step": 24591 }, { "epoch": 1.5266000372462598, "grad_norm": 0.6234131456938671, "learning_rate": 5.715506206827388e-05, "loss": 2.8131, "step": 24592 }, { "epoch": 1.5266621143460177, "grad_norm": 0.6114162101295032, "learning_rate": 5.7151487695039405e-05, "loss": 2.8389, "step": 24593 }, { "epoch": 1.5267241914457756, "grad_norm": 0.6370896739240851, "learning_rate": 5.71479132844939e-05, "loss": 2.756, "step": 24594 }, { "epoch": 1.5267862685455336, "grad_norm": 0.5893617896726332, "learning_rate": 5.7144338836655986e-05, "loss": 2.8764, "step": 24595 }, { "epoch": 1.5268483456452915, "grad_norm": 0.6120308685106769, "learning_rate": 5.714076435154434e-05, "loss": 2.7473, "step": 24596 }, { "epoch": 1.5269104227450494, "grad_norm": 0.6052608994609852, "learning_rate": 5.7137189829177576e-05, "loss": 2.7583, "step": 24597 }, { "epoch": 1.526972499844807, "grad_norm": 0.5710713289506673, "learning_rate": 5.713361526957437e-05, "loss": 2.8202, "step": 24598 }, { "epoch": 1.527034576944565, "grad_norm": 0.6776556359539632, "learning_rate": 5.713004067275335e-05, "loss": 2.751, "step": 24599 }, { "epoch": 1.527096654044323, "grad_norm": 0.5651713095533986, "learning_rate": 5.712646603873319e-05, "loss": 2.7407, "step": 24600 }, { "epoch": 1.5271587311440808, "grad_norm": 0.6394351050203685, "learning_rate": 5.7122891367532505e-05, "loss": 2.8174, "step": 24601 }, { "epoch": 1.5272208082438388, "grad_norm": 0.5294279942564877, "learning_rate": 5.711931665916999e-05, "loss": 2.7821, "step": 24602 }, { "epoch": 1.5272828853435967, "grad_norm": 0.6553501590209637, "learning_rate": 5.7115741913664264e-05, "loss": 2.7745, "step": 24603 }, { "epoch": 1.5273449624433546, "grad_norm": 0.5700514993073557, "learning_rate": 5.711216713103398e-05, "loss": 2.7241, "step": 24604 }, { "epoch": 1.5274070395431125, "grad_norm": 0.7160265366575563, "learning_rate": 5.7108592311297804e-05, "loss": 2.7015, "step": 24605 }, { "epoch": 1.5274691166428704, "grad_norm": 0.5919052377326308, "learning_rate": 5.7105017454474365e-05, "loss": 2.7774, "step": 24606 }, { "epoch": 1.5275311937426284, "grad_norm": 0.6361524954243409, "learning_rate": 5.710144256058233e-05, "loss": 2.8258, "step": 24607 }, { "epoch": 1.5275932708423863, "grad_norm": 0.5690652193845559, "learning_rate": 5.7097867629640346e-05, "loss": 2.8157, "step": 24608 }, { "epoch": 1.5276553479421442, "grad_norm": 0.59147144094385, "learning_rate": 5.709429266166706e-05, "loss": 2.8316, "step": 24609 }, { "epoch": 1.527717425041902, "grad_norm": 0.6449916928887477, "learning_rate": 5.7090717656681136e-05, "loss": 2.7376, "step": 24610 }, { "epoch": 1.52777950214166, "grad_norm": 0.6131914887069637, "learning_rate": 5.708714261470121e-05, "loss": 2.881, "step": 24611 }, { "epoch": 1.527841579241418, "grad_norm": 0.5967829188874698, "learning_rate": 5.708356753574594e-05, "loss": 2.7249, "step": 24612 }, { "epoch": 1.5279036563411759, "grad_norm": 0.6316938992352904, "learning_rate": 5.7079992419833975e-05, "loss": 2.7707, "step": 24613 }, { "epoch": 1.5279657334409338, "grad_norm": 0.6378086801327635, "learning_rate": 5.7076417266983986e-05, "loss": 2.8311, "step": 24614 }, { "epoch": 1.5280278105406917, "grad_norm": 0.5700519175476233, "learning_rate": 5.707284207721458e-05, "loss": 2.7159, "step": 24615 }, { "epoch": 1.5280898876404494, "grad_norm": 0.7117867061527963, "learning_rate": 5.706926685054448e-05, "loss": 2.7213, "step": 24616 }, { "epoch": 1.5281519647402073, "grad_norm": 0.5118543430673632, "learning_rate": 5.706569158699227e-05, "loss": 2.8121, "step": 24617 }, { "epoch": 1.5282140418399652, "grad_norm": 0.6846878451507966, "learning_rate": 5.706211628657664e-05, "loss": 2.7511, "step": 24618 }, { "epoch": 1.5282761189397231, "grad_norm": 0.55352473281765, "learning_rate": 5.705854094931623e-05, "loss": 2.7366, "step": 24619 }, { "epoch": 1.528338196039481, "grad_norm": 0.567889094374931, "learning_rate": 5.705496557522971e-05, "loss": 2.7361, "step": 24620 }, { "epoch": 1.528400273139239, "grad_norm": 0.6165509140001068, "learning_rate": 5.7051390164335715e-05, "loss": 2.7321, "step": 24621 }, { "epoch": 1.5284623502389967, "grad_norm": 0.5844309958490815, "learning_rate": 5.7047814716652904e-05, "loss": 2.6864, "step": 24622 }, { "epoch": 1.5285244273387546, "grad_norm": 0.5986923511761375, "learning_rate": 5.704423923219994e-05, "loss": 2.7847, "step": 24623 }, { "epoch": 1.5285865044385125, "grad_norm": 0.6426807541909152, "learning_rate": 5.704066371099545e-05, "loss": 2.7339, "step": 24624 }, { "epoch": 1.5286485815382704, "grad_norm": 0.5281124587516802, "learning_rate": 5.7037088153058125e-05, "loss": 2.8086, "step": 24625 }, { "epoch": 1.5287106586380284, "grad_norm": 0.6384035299173083, "learning_rate": 5.70335125584066e-05, "loss": 2.785, "step": 24626 }, { "epoch": 1.5287727357377863, "grad_norm": 0.5059480626191784, "learning_rate": 5.7029936927059535e-05, "loss": 2.7096, "step": 24627 }, { "epoch": 1.5288348128375442, "grad_norm": 0.6361908860090731, "learning_rate": 5.702636125903559e-05, "loss": 2.7702, "step": 24628 }, { "epoch": 1.528896889937302, "grad_norm": 0.5444709525526584, "learning_rate": 5.7022785554353395e-05, "loss": 2.7613, "step": 24629 }, { "epoch": 1.52895896703706, "grad_norm": 0.5947128569929528, "learning_rate": 5.701920981303164e-05, "loss": 2.7411, "step": 24630 }, { "epoch": 1.529021044136818, "grad_norm": 0.600570802575431, "learning_rate": 5.701563403508894e-05, "loss": 2.7798, "step": 24631 }, { "epoch": 1.5290831212365759, "grad_norm": 0.5922210230938452, "learning_rate": 5.7012058220543996e-05, "loss": 2.8147, "step": 24632 }, { "epoch": 1.5291451983363338, "grad_norm": 0.6836738539492623, "learning_rate": 5.700848236941543e-05, "loss": 2.7534, "step": 24633 }, { "epoch": 1.5292072754360917, "grad_norm": 0.540352645803444, "learning_rate": 5.700490648172192e-05, "loss": 2.6903, "step": 24634 }, { "epoch": 1.5292693525358496, "grad_norm": 0.5925787114119724, "learning_rate": 5.70013305574821e-05, "loss": 2.7544, "step": 24635 }, { "epoch": 1.5293314296356075, "grad_norm": 0.5843282854308625, "learning_rate": 5.699775459671466e-05, "loss": 2.8102, "step": 24636 }, { "epoch": 1.5293935067353654, "grad_norm": 0.5954907648505885, "learning_rate": 5.699417859943821e-05, "loss": 2.8365, "step": 24637 }, { "epoch": 1.5294555838351234, "grad_norm": 0.6295398576933356, "learning_rate": 5.6990602565671447e-05, "loss": 2.8203, "step": 24638 }, { "epoch": 1.5295176609348813, "grad_norm": 0.5893978429192408, "learning_rate": 5.698702649543301e-05, "loss": 2.8353, "step": 24639 }, { "epoch": 1.529579738034639, "grad_norm": 0.6339037399110657, "learning_rate": 5.6983450388741554e-05, "loss": 2.7429, "step": 24640 }, { "epoch": 1.529641815134397, "grad_norm": 0.5958893281683622, "learning_rate": 5.697987424561575e-05, "loss": 2.7574, "step": 24641 }, { "epoch": 1.5297038922341548, "grad_norm": 0.6097338182461316, "learning_rate": 5.697629806607424e-05, "loss": 2.7075, "step": 24642 }, { "epoch": 1.5297659693339127, "grad_norm": 0.6089401894640875, "learning_rate": 5.6972721850135705e-05, "loss": 2.8566, "step": 24643 }, { "epoch": 1.5298280464336707, "grad_norm": 0.7071839478555659, "learning_rate": 5.6969145597818764e-05, "loss": 2.8257, "step": 24644 }, { "epoch": 1.5298901235334286, "grad_norm": 0.538480693481672, "learning_rate": 5.696556930914212e-05, "loss": 2.7281, "step": 24645 }, { "epoch": 1.5299522006331863, "grad_norm": 0.7390848458845566, "learning_rate": 5.6961992984124416e-05, "loss": 2.8602, "step": 24646 }, { "epoch": 1.5300142777329442, "grad_norm": 0.5500189734567478, "learning_rate": 5.695841662278427e-05, "loss": 2.7747, "step": 24647 }, { "epoch": 1.530076354832702, "grad_norm": 0.6320688092703236, "learning_rate": 5.6954840225140404e-05, "loss": 2.7749, "step": 24648 }, { "epoch": 1.53013843193246, "grad_norm": 0.5653444532280704, "learning_rate": 5.695126379121144e-05, "loss": 2.7454, "step": 24649 }, { "epoch": 1.530200509032218, "grad_norm": 0.594597863614799, "learning_rate": 5.694768732101605e-05, "loss": 2.7927, "step": 24650 }, { "epoch": 1.5302625861319759, "grad_norm": 0.602342966279585, "learning_rate": 5.694411081457288e-05, "loss": 2.808, "step": 24651 }, { "epoch": 1.5303246632317338, "grad_norm": 0.6030623329297574, "learning_rate": 5.6940534271900606e-05, "loss": 2.7934, "step": 24652 }, { "epoch": 1.5303867403314917, "grad_norm": 0.58130413439114, "learning_rate": 5.6936957693017864e-05, "loss": 2.719, "step": 24653 }, { "epoch": 1.5304488174312496, "grad_norm": 0.6765746382049539, "learning_rate": 5.693338107794335e-05, "loss": 2.8442, "step": 24654 }, { "epoch": 1.5305108945310075, "grad_norm": 0.5573443866166783, "learning_rate": 5.692980442669569e-05, "loss": 2.752, "step": 24655 }, { "epoch": 1.5305729716307654, "grad_norm": 0.6650715958624198, "learning_rate": 5.692622773929356e-05, "loss": 2.7356, "step": 24656 }, { "epoch": 1.5306350487305234, "grad_norm": 0.5506801106545226, "learning_rate": 5.692265101575562e-05, "loss": 2.8513, "step": 24657 }, { "epoch": 1.5306971258302813, "grad_norm": 0.6729058629951042, "learning_rate": 5.691907425610054e-05, "loss": 2.7957, "step": 24658 }, { "epoch": 1.5307592029300392, "grad_norm": 0.6134386042984804, "learning_rate": 5.6915497460346956e-05, "loss": 2.8237, "step": 24659 }, { "epoch": 1.5308212800297971, "grad_norm": 0.5954807304112043, "learning_rate": 5.6911920628513545e-05, "loss": 2.7473, "step": 24660 }, { "epoch": 1.530883357129555, "grad_norm": 0.6075404256941388, "learning_rate": 5.6908343760618974e-05, "loss": 2.8396, "step": 24661 }, { "epoch": 1.530945434229313, "grad_norm": 0.5253232040540139, "learning_rate": 5.690476685668189e-05, "loss": 2.798, "step": 24662 }, { "epoch": 1.5310075113290709, "grad_norm": 0.6124730707593563, "learning_rate": 5.690118991672096e-05, "loss": 2.787, "step": 24663 }, { "epoch": 1.5310695884288286, "grad_norm": 0.5950352661516563, "learning_rate": 5.689761294075486e-05, "loss": 2.8158, "step": 24664 }, { "epoch": 1.5311316655285865, "grad_norm": 0.5658043129304519, "learning_rate": 5.6894035928802225e-05, "loss": 2.7909, "step": 24665 }, { "epoch": 1.5311937426283444, "grad_norm": 0.6438182007670813, "learning_rate": 5.689045888088174e-05, "loss": 2.8787, "step": 24666 }, { "epoch": 1.5312558197281023, "grad_norm": 0.5401198756092507, "learning_rate": 5.688688179701205e-05, "loss": 2.7081, "step": 24667 }, { "epoch": 1.5313178968278602, "grad_norm": 0.6263363380079218, "learning_rate": 5.6883304677211836e-05, "loss": 2.7863, "step": 24668 }, { "epoch": 1.5313799739276182, "grad_norm": 0.5471617628021673, "learning_rate": 5.6879727521499724e-05, "loss": 2.7264, "step": 24669 }, { "epoch": 1.5314420510273759, "grad_norm": 0.664590782514417, "learning_rate": 5.687615032989444e-05, "loss": 2.772, "step": 24670 }, { "epoch": 1.5315041281271338, "grad_norm": 0.5572393577007841, "learning_rate": 5.6872573102414585e-05, "loss": 2.7892, "step": 24671 }, { "epoch": 1.5315662052268917, "grad_norm": 0.5854470807339506, "learning_rate": 5.6868995839078854e-05, "loss": 2.7775, "step": 24672 }, { "epoch": 1.5316282823266496, "grad_norm": 0.5496874560626664, "learning_rate": 5.6865418539905904e-05, "loss": 2.8359, "step": 24673 }, { "epoch": 1.5316903594264075, "grad_norm": 0.5717397884346034, "learning_rate": 5.6861841204914404e-05, "loss": 2.7341, "step": 24674 }, { "epoch": 1.5317524365261654, "grad_norm": 0.5192183218996417, "learning_rate": 5.6858263834123e-05, "loss": 2.7446, "step": 24675 }, { "epoch": 1.5318145136259234, "grad_norm": 0.6159610150791986, "learning_rate": 5.685468642755039e-05, "loss": 2.7561, "step": 24676 }, { "epoch": 1.5318765907256813, "grad_norm": 0.5188736446512708, "learning_rate": 5.6851108985215196e-05, "loss": 2.7166, "step": 24677 }, { "epoch": 1.5319386678254392, "grad_norm": 0.617729842040409, "learning_rate": 5.684753150713611e-05, "loss": 2.6501, "step": 24678 }, { "epoch": 1.532000744925197, "grad_norm": 0.5797232881588111, "learning_rate": 5.68439539933318e-05, "loss": 2.8162, "step": 24679 }, { "epoch": 1.532062822024955, "grad_norm": 0.6269611821592745, "learning_rate": 5.684037644382091e-05, "loss": 2.8392, "step": 24680 }, { "epoch": 1.532124899124713, "grad_norm": 0.6278850486159361, "learning_rate": 5.683679885862212e-05, "loss": 2.7588, "step": 24681 }, { "epoch": 1.5321869762244709, "grad_norm": 0.6896174507882625, "learning_rate": 5.68332212377541e-05, "loss": 2.8141, "step": 24682 }, { "epoch": 1.5322490533242288, "grad_norm": 0.6539960482795854, "learning_rate": 5.682964358123548e-05, "loss": 2.748, "step": 24683 }, { "epoch": 1.5323111304239867, "grad_norm": 0.6111222429297198, "learning_rate": 5.682606588908497e-05, "loss": 2.8139, "step": 24684 }, { "epoch": 1.5323732075237446, "grad_norm": 0.5787757613213929, "learning_rate": 5.682248816132121e-05, "loss": 2.8329, "step": 24685 }, { "epoch": 1.5324352846235025, "grad_norm": 0.6799948205820746, "learning_rate": 5.6818910397962885e-05, "loss": 2.7739, "step": 24686 }, { "epoch": 1.5324973617232605, "grad_norm": 0.5442966716662493, "learning_rate": 5.681533259902864e-05, "loss": 2.743, "step": 24687 }, { "epoch": 1.5325594388230181, "grad_norm": 0.6450449890149066, "learning_rate": 5.6811754764537154e-05, "loss": 2.7617, "step": 24688 }, { "epoch": 1.532621515922776, "grad_norm": 0.5649543305734522, "learning_rate": 5.680817689450708e-05, "loss": 2.7597, "step": 24689 }, { "epoch": 1.532683593022534, "grad_norm": 0.582494078061695, "learning_rate": 5.6804598988957095e-05, "loss": 2.8547, "step": 24690 }, { "epoch": 1.532745670122292, "grad_norm": 0.6127813938461104, "learning_rate": 5.680102104790588e-05, "loss": 2.7466, "step": 24691 }, { "epoch": 1.5328077472220498, "grad_norm": 0.5660403944536369, "learning_rate": 5.679744307137207e-05, "loss": 2.8067, "step": 24692 }, { "epoch": 1.5328698243218077, "grad_norm": 0.6202018860172281, "learning_rate": 5.679386505937436e-05, "loss": 2.7608, "step": 24693 }, { "epoch": 1.5329319014215654, "grad_norm": 0.6611737642888884, "learning_rate": 5.6790287011931396e-05, "loss": 2.7588, "step": 24694 }, { "epoch": 1.5329939785213234, "grad_norm": 0.5960589737453355, "learning_rate": 5.678670892906187e-05, "loss": 2.8042, "step": 24695 }, { "epoch": 1.5330560556210813, "grad_norm": 0.7257847058036335, "learning_rate": 5.6783130810784434e-05, "loss": 2.7242, "step": 24696 }, { "epoch": 1.5331181327208392, "grad_norm": 0.5513079674619532, "learning_rate": 5.677955265711775e-05, "loss": 2.8317, "step": 24697 }, { "epoch": 1.533180209820597, "grad_norm": 0.6431690965369157, "learning_rate": 5.677597446808049e-05, "loss": 2.7758, "step": 24698 }, { "epoch": 1.533242286920355, "grad_norm": 0.5476677462123525, "learning_rate": 5.677239624369135e-05, "loss": 2.7524, "step": 24699 }, { "epoch": 1.533304364020113, "grad_norm": 0.6205159504674685, "learning_rate": 5.676881798396896e-05, "loss": 2.8321, "step": 24700 }, { "epoch": 1.5333664411198709, "grad_norm": 0.5346937756864035, "learning_rate": 5.6765239688932e-05, "loss": 2.7867, "step": 24701 }, { "epoch": 1.5334285182196288, "grad_norm": 0.6206416517945323, "learning_rate": 5.676166135859915e-05, "loss": 2.7402, "step": 24702 }, { "epoch": 1.5334905953193867, "grad_norm": 0.5274688572300693, "learning_rate": 5.675808299298906e-05, "loss": 2.7681, "step": 24703 }, { "epoch": 1.5335526724191446, "grad_norm": 0.6237393062187058, "learning_rate": 5.6754504592120426e-05, "loss": 2.7426, "step": 24704 }, { "epoch": 1.5336147495189025, "grad_norm": 0.5287199078106465, "learning_rate": 5.675092615601187e-05, "loss": 2.7621, "step": 24705 }, { "epoch": 1.5336768266186604, "grad_norm": 0.6478052158930692, "learning_rate": 5.674734768468213e-05, "loss": 2.8189, "step": 24706 }, { "epoch": 1.5337389037184184, "grad_norm": 0.5566448546131796, "learning_rate": 5.674376917814983e-05, "loss": 2.7572, "step": 24707 }, { "epoch": 1.5338009808181763, "grad_norm": 0.6435676770209037, "learning_rate": 5.674019063643363e-05, "loss": 2.7511, "step": 24708 }, { "epoch": 1.5338630579179342, "grad_norm": 0.5382670739775137, "learning_rate": 5.673661205955225e-05, "loss": 2.7748, "step": 24709 }, { "epoch": 1.5339251350176921, "grad_norm": 0.5792851534973946, "learning_rate": 5.6733033447524305e-05, "loss": 2.7733, "step": 24710 }, { "epoch": 1.53398721211745, "grad_norm": 0.6036248710049199, "learning_rate": 5.672945480036851e-05, "loss": 2.8812, "step": 24711 }, { "epoch": 1.5340492892172077, "grad_norm": 0.49786794103511867, "learning_rate": 5.672587611810349e-05, "loss": 2.756, "step": 24712 }, { "epoch": 1.5341113663169657, "grad_norm": 0.5409265713438943, "learning_rate": 5.672229740074796e-05, "loss": 2.7608, "step": 24713 }, { "epoch": 1.5341734434167236, "grad_norm": 0.5536226070935439, "learning_rate": 5.671871864832057e-05, "loss": 2.8167, "step": 24714 }, { "epoch": 1.5342355205164815, "grad_norm": 0.5747554891505875, "learning_rate": 5.671513986083999e-05, "loss": 2.7822, "step": 24715 }, { "epoch": 1.5342975976162394, "grad_norm": 0.527558685581792, "learning_rate": 5.671156103832489e-05, "loss": 2.7385, "step": 24716 }, { "epoch": 1.5343596747159973, "grad_norm": 0.6374495589178536, "learning_rate": 5.670798218079396e-05, "loss": 2.8641, "step": 24717 }, { "epoch": 1.534421751815755, "grad_norm": 0.5762948373740382, "learning_rate": 5.6704403288265865e-05, "loss": 2.7194, "step": 24718 }, { "epoch": 1.534483828915513, "grad_norm": 0.6017571171623751, "learning_rate": 5.670082436075925e-05, "loss": 2.785, "step": 24719 }, { "epoch": 1.5345459060152709, "grad_norm": 0.579780091039819, "learning_rate": 5.669724539829282e-05, "loss": 2.7886, "step": 24720 }, { "epoch": 1.5346079831150288, "grad_norm": 0.6123070724416522, "learning_rate": 5.6693666400885226e-05, "loss": 2.8088, "step": 24721 }, { "epoch": 1.5346700602147867, "grad_norm": 0.5465493186115864, "learning_rate": 5.669008736855515e-05, "loss": 2.8027, "step": 24722 }, { "epoch": 1.5347321373145446, "grad_norm": 0.5617010482524308, "learning_rate": 5.668650830132126e-05, "loss": 2.6995, "step": 24723 }, { "epoch": 1.5347942144143025, "grad_norm": 0.7379061033403602, "learning_rate": 5.668292919920224e-05, "loss": 2.7413, "step": 24724 }, { "epoch": 1.5348562915140604, "grad_norm": 0.6305009750960673, "learning_rate": 5.667935006221676e-05, "loss": 2.8378, "step": 24725 }, { "epoch": 1.5349183686138184, "grad_norm": 0.6748972496665118, "learning_rate": 5.6675770890383485e-05, "loss": 2.6513, "step": 24726 }, { "epoch": 1.5349804457135763, "grad_norm": 0.6219819154318211, "learning_rate": 5.66721916837211e-05, "loss": 2.8009, "step": 24727 }, { "epoch": 1.5350425228133342, "grad_norm": 0.6954703848242587, "learning_rate": 5.6668612442248256e-05, "loss": 2.769, "step": 24728 }, { "epoch": 1.5351045999130921, "grad_norm": 0.638973596446755, "learning_rate": 5.666503316598366e-05, "loss": 2.745, "step": 24729 }, { "epoch": 1.53516667701285, "grad_norm": 0.695213267928335, "learning_rate": 5.6661453854945954e-05, "loss": 2.8288, "step": 24730 }, { "epoch": 1.535228754112608, "grad_norm": 0.649392653951341, "learning_rate": 5.665787450915384e-05, "loss": 2.8262, "step": 24731 }, { "epoch": 1.5352908312123659, "grad_norm": 0.6378135629931969, "learning_rate": 5.665429512862597e-05, "loss": 2.8301, "step": 24732 }, { "epoch": 1.5353529083121238, "grad_norm": 0.6334398480199158, "learning_rate": 5.665071571338102e-05, "loss": 2.8545, "step": 24733 }, { "epoch": 1.5354149854118817, "grad_norm": 0.6341078886478985, "learning_rate": 5.6647136263437676e-05, "loss": 2.707, "step": 24734 }, { "epoch": 1.5354770625116394, "grad_norm": 0.6777141029338251, "learning_rate": 5.664355677881461e-05, "loss": 2.8421, "step": 24735 }, { "epoch": 1.5355391396113973, "grad_norm": 0.6685228336289285, "learning_rate": 5.6639977259530506e-05, "loss": 2.7602, "step": 24736 }, { "epoch": 1.5356012167111552, "grad_norm": 0.636828594445044, "learning_rate": 5.6636397705604016e-05, "loss": 2.7792, "step": 24737 }, { "epoch": 1.5356632938109132, "grad_norm": 0.6873594487178002, "learning_rate": 5.663281811705382e-05, "loss": 2.7563, "step": 24738 }, { "epoch": 1.535725370910671, "grad_norm": 0.5731047899433106, "learning_rate": 5.66292384938986e-05, "loss": 2.795, "step": 24739 }, { "epoch": 1.535787448010429, "grad_norm": 0.6162692857011052, "learning_rate": 5.662565883615707e-05, "loss": 2.8585, "step": 24740 }, { "epoch": 1.5358495251101867, "grad_norm": 0.5910107014231001, "learning_rate": 5.662207914384783e-05, "loss": 2.7429, "step": 24741 }, { "epoch": 1.5359116022099446, "grad_norm": 0.5933402303855588, "learning_rate": 5.661849941698962e-05, "loss": 2.787, "step": 24742 }, { "epoch": 1.5359736793097025, "grad_norm": 0.5784312545381497, "learning_rate": 5.6614919655601086e-05, "loss": 2.7474, "step": 24743 }, { "epoch": 1.5360357564094604, "grad_norm": 0.557017790857817, "learning_rate": 5.6611339859700906e-05, "loss": 2.783, "step": 24744 }, { "epoch": 1.5360978335092184, "grad_norm": 0.6085927539665484, "learning_rate": 5.660776002930777e-05, "loss": 2.7214, "step": 24745 }, { "epoch": 1.5361599106089763, "grad_norm": 0.5674064749479701, "learning_rate": 5.6604180164440335e-05, "loss": 2.7281, "step": 24746 }, { "epoch": 1.5362219877087342, "grad_norm": 0.5597796356513514, "learning_rate": 5.660060026511729e-05, "loss": 2.7614, "step": 24747 }, { "epoch": 1.5362840648084921, "grad_norm": 0.6024113400705275, "learning_rate": 5.659702033135731e-05, "loss": 2.7699, "step": 24748 }, { "epoch": 1.53634614190825, "grad_norm": 0.5983674548717135, "learning_rate": 5.6593440363179087e-05, "loss": 2.7461, "step": 24749 }, { "epoch": 1.536408219008008, "grad_norm": 0.6115722168943344, "learning_rate": 5.658986036060127e-05, "loss": 2.7888, "step": 24750 }, { "epoch": 1.5364702961077659, "grad_norm": 0.6213715130115895, "learning_rate": 5.658628032364255e-05, "loss": 2.7115, "step": 24751 }, { "epoch": 1.5365323732075238, "grad_norm": 0.6276771429115788, "learning_rate": 5.658270025232161e-05, "loss": 2.792, "step": 24752 }, { "epoch": 1.5365944503072817, "grad_norm": 0.5963301563565682, "learning_rate": 5.657912014665714e-05, "loss": 2.7754, "step": 24753 }, { "epoch": 1.5366565274070396, "grad_norm": 0.5713742445342169, "learning_rate": 5.657554000666778e-05, "loss": 2.6934, "step": 24754 }, { "epoch": 1.5367186045067975, "grad_norm": 0.6008471905419389, "learning_rate": 5.657195983237224e-05, "loss": 2.756, "step": 24755 }, { "epoch": 1.5367806816065555, "grad_norm": 0.5839969544347622, "learning_rate": 5.656837962378919e-05, "loss": 2.7169, "step": 24756 }, { "epoch": 1.5368427587063134, "grad_norm": 0.5910174332595463, "learning_rate": 5.656479938093731e-05, "loss": 2.8612, "step": 24757 }, { "epoch": 1.5369048358060713, "grad_norm": 0.5918892010976052, "learning_rate": 5.6561219103835284e-05, "loss": 2.7701, "step": 24758 }, { "epoch": 1.536966912905829, "grad_norm": 0.5475083634421424, "learning_rate": 5.655763879250178e-05, "loss": 2.6656, "step": 24759 }, { "epoch": 1.537028990005587, "grad_norm": 0.6135306367600518, "learning_rate": 5.655405844695548e-05, "loss": 2.8082, "step": 24760 }, { "epoch": 1.5370910671053448, "grad_norm": 0.6518785336875048, "learning_rate": 5.6550478067215086e-05, "loss": 2.8284, "step": 24761 }, { "epoch": 1.5371531442051027, "grad_norm": 0.5216724185301505, "learning_rate": 5.654689765329924e-05, "loss": 2.7503, "step": 24762 }, { "epoch": 1.5372152213048607, "grad_norm": 0.6326655228967581, "learning_rate": 5.6543317205226635e-05, "loss": 2.8806, "step": 24763 }, { "epoch": 1.5372772984046186, "grad_norm": 0.5879659762837891, "learning_rate": 5.6539736723015955e-05, "loss": 2.8001, "step": 24764 }, { "epoch": 1.5373393755043763, "grad_norm": 0.58066200643484, "learning_rate": 5.6536156206685894e-05, "loss": 2.8351, "step": 24765 }, { "epoch": 1.5374014526041342, "grad_norm": 0.6330058661892638, "learning_rate": 5.6532575656255106e-05, "loss": 2.8613, "step": 24766 }, { "epoch": 1.537463529703892, "grad_norm": 0.6374525978192233, "learning_rate": 5.6528995071742295e-05, "loss": 2.8125, "step": 24767 }, { "epoch": 1.53752560680365, "grad_norm": 0.5837032189912248, "learning_rate": 5.652541445316613e-05, "loss": 2.7444, "step": 24768 }, { "epoch": 1.537587683903408, "grad_norm": 0.6023729982708952, "learning_rate": 5.65218338005453e-05, "loss": 2.8323, "step": 24769 }, { "epoch": 1.5376497610031659, "grad_norm": 0.5655757735102035, "learning_rate": 5.6518253113898456e-05, "loss": 2.7364, "step": 24770 }, { "epoch": 1.5377118381029238, "grad_norm": 0.6443076699616387, "learning_rate": 5.6514672393244315e-05, "loss": 2.8398, "step": 24771 }, { "epoch": 1.5377739152026817, "grad_norm": 0.5556759958712149, "learning_rate": 5.6511091638601555e-05, "loss": 2.8096, "step": 24772 }, { "epoch": 1.5378359923024396, "grad_norm": 0.6008729822495136, "learning_rate": 5.650751084998884e-05, "loss": 2.6722, "step": 24773 }, { "epoch": 1.5378980694021975, "grad_norm": 0.6722325105729261, "learning_rate": 5.650393002742487e-05, "loss": 2.8289, "step": 24774 }, { "epoch": 1.5379601465019554, "grad_norm": 0.5452031602518117, "learning_rate": 5.6500349170928314e-05, "loss": 2.7221, "step": 24775 }, { "epoch": 1.5380222236017134, "grad_norm": 0.5734686331132237, "learning_rate": 5.649676828051787e-05, "loss": 2.7179, "step": 24776 }, { "epoch": 1.5380843007014713, "grad_norm": 0.5872269746789988, "learning_rate": 5.6493187356212185e-05, "loss": 2.8293, "step": 24777 }, { "epoch": 1.5381463778012292, "grad_norm": 0.5368665944108654, "learning_rate": 5.648960639802998e-05, "loss": 2.7106, "step": 24778 }, { "epoch": 1.5382084549009871, "grad_norm": 0.6220638688021716, "learning_rate": 5.648602540598993e-05, "loss": 2.7655, "step": 24779 }, { "epoch": 1.538270532000745, "grad_norm": 0.5888999056860236, "learning_rate": 5.648244438011069e-05, "loss": 2.7657, "step": 24780 }, { "epoch": 1.538332609100503, "grad_norm": 0.5891651519457362, "learning_rate": 5.647886332041098e-05, "loss": 2.8207, "step": 24781 }, { "epoch": 1.5383946862002609, "grad_norm": 0.6367135076950603, "learning_rate": 5.647528222690946e-05, "loss": 2.7841, "step": 24782 }, { "epoch": 1.5384567633000186, "grad_norm": 0.6102737255055467, "learning_rate": 5.647170109962483e-05, "loss": 2.8622, "step": 24783 }, { "epoch": 1.5385188403997765, "grad_norm": 0.6141455810098446, "learning_rate": 5.646811993857575e-05, "loss": 2.8278, "step": 24784 }, { "epoch": 1.5385809174995344, "grad_norm": 0.6419995378584699, "learning_rate": 5.646453874378094e-05, "loss": 2.8013, "step": 24785 }, { "epoch": 1.5386429945992923, "grad_norm": 0.5326701424337686, "learning_rate": 5.646095751525904e-05, "loss": 2.8144, "step": 24786 }, { "epoch": 1.5387050716990502, "grad_norm": 0.6803144873544931, "learning_rate": 5.645737625302875e-05, "loss": 2.7587, "step": 24787 }, { "epoch": 1.5387671487988082, "grad_norm": 0.6135550695573435, "learning_rate": 5.6453794957108784e-05, "loss": 2.6963, "step": 24788 }, { "epoch": 1.5388292258985659, "grad_norm": 0.6450121386073332, "learning_rate": 5.645021362751779e-05, "loss": 2.7609, "step": 24789 }, { "epoch": 1.5388913029983238, "grad_norm": 0.6105998274379028, "learning_rate": 5.6446632264274466e-05, "loss": 2.7948, "step": 24790 }, { "epoch": 1.5389533800980817, "grad_norm": 0.601509760737003, "learning_rate": 5.6443050867397495e-05, "loss": 2.8039, "step": 24791 }, { "epoch": 1.5390154571978396, "grad_norm": 0.6651181077404694, "learning_rate": 5.643946943690558e-05, "loss": 2.7708, "step": 24792 }, { "epoch": 1.5390775342975975, "grad_norm": 0.5372223891223243, "learning_rate": 5.643588797281737e-05, "loss": 2.8447, "step": 24793 }, { "epoch": 1.5391396113973554, "grad_norm": 0.6303471945583936, "learning_rate": 5.643230647515158e-05, "loss": 2.804, "step": 24794 }, { "epoch": 1.5392016884971134, "grad_norm": 0.5521599038724448, "learning_rate": 5.642872494392688e-05, "loss": 2.847, "step": 24795 }, { "epoch": 1.5392637655968713, "grad_norm": 0.5891701091556075, "learning_rate": 5.642514337916197e-05, "loss": 2.7802, "step": 24796 }, { "epoch": 1.5393258426966292, "grad_norm": 0.5478268922089219, "learning_rate": 5.642156178087552e-05, "loss": 2.7925, "step": 24797 }, { "epoch": 1.5393879197963871, "grad_norm": 0.557533137572558, "learning_rate": 5.6417980149086235e-05, "loss": 2.8107, "step": 24798 }, { "epoch": 1.539449996896145, "grad_norm": 0.5320387482639889, "learning_rate": 5.641439848381278e-05, "loss": 2.734, "step": 24799 }, { "epoch": 1.539512073995903, "grad_norm": 0.5874479534551519, "learning_rate": 5.641081678507385e-05, "loss": 2.8101, "step": 24800 }, { "epoch": 1.5395741510956609, "grad_norm": 0.5976482714949475, "learning_rate": 5.640723505288813e-05, "loss": 2.7484, "step": 24801 }, { "epoch": 1.5396362281954188, "grad_norm": 0.5475465193890657, "learning_rate": 5.6403653287274316e-05, "loss": 2.6935, "step": 24802 }, { "epoch": 1.5396983052951767, "grad_norm": 0.6437417946449958, "learning_rate": 5.6400071488251085e-05, "loss": 2.7363, "step": 24803 }, { "epoch": 1.5397603823949346, "grad_norm": 0.649543577200662, "learning_rate": 5.6396489655837124e-05, "loss": 2.7874, "step": 24804 }, { "epoch": 1.5398224594946925, "grad_norm": 0.5802153864853599, "learning_rate": 5.639290779005112e-05, "loss": 2.7342, "step": 24805 }, { "epoch": 1.5398845365944505, "grad_norm": 0.5721130428069011, "learning_rate": 5.638932589091178e-05, "loss": 2.7303, "step": 24806 }, { "epoch": 1.5399466136942082, "grad_norm": 0.641826038777614, "learning_rate": 5.6385743958437765e-05, "loss": 2.79, "step": 24807 }, { "epoch": 1.540008690793966, "grad_norm": 0.524844002253752, "learning_rate": 5.6382161992647774e-05, "loss": 2.7507, "step": 24808 }, { "epoch": 1.540070767893724, "grad_norm": 0.6654094947144985, "learning_rate": 5.637857999356049e-05, "loss": 2.8136, "step": 24809 }, { "epoch": 1.540132844993482, "grad_norm": 0.555934015813141, "learning_rate": 5.637499796119462e-05, "loss": 2.7671, "step": 24810 }, { "epoch": 1.5401949220932398, "grad_norm": 0.6538705832030121, "learning_rate": 5.6371415895568824e-05, "loss": 2.7973, "step": 24811 }, { "epoch": 1.5402569991929977, "grad_norm": 0.654490883375398, "learning_rate": 5.636783379670182e-05, "loss": 2.758, "step": 24812 }, { "epoch": 1.5403190762927554, "grad_norm": 0.5886595757002632, "learning_rate": 5.6364251664612263e-05, "loss": 2.7625, "step": 24813 }, { "epoch": 1.5403811533925134, "grad_norm": 0.6233701674352644, "learning_rate": 5.6360669499318874e-05, "loss": 2.8612, "step": 24814 }, { "epoch": 1.5404432304922713, "grad_norm": 0.5813828767641919, "learning_rate": 5.635708730084032e-05, "loss": 2.7843, "step": 24815 }, { "epoch": 1.5405053075920292, "grad_norm": 0.618628256794806, "learning_rate": 5.63535050691953e-05, "loss": 2.7503, "step": 24816 }, { "epoch": 1.5405673846917871, "grad_norm": 0.6467880043096972, "learning_rate": 5.634992280440251e-05, "loss": 2.8528, "step": 24817 }, { "epoch": 1.540629461791545, "grad_norm": 0.5717379379701468, "learning_rate": 5.634634050648062e-05, "loss": 2.7828, "step": 24818 }, { "epoch": 1.540691538891303, "grad_norm": 0.5984292609939945, "learning_rate": 5.634275817544834e-05, "loss": 2.7225, "step": 24819 }, { "epoch": 1.5407536159910609, "grad_norm": 0.5987089273624719, "learning_rate": 5.633917581132433e-05, "loss": 2.7048, "step": 24820 }, { "epoch": 1.5408156930908188, "grad_norm": 0.6737636369446878, "learning_rate": 5.633559341412732e-05, "loss": 2.7137, "step": 24821 }, { "epoch": 1.5408777701905767, "grad_norm": 0.7179432363200434, "learning_rate": 5.6332010983875974e-05, "loss": 2.8282, "step": 24822 }, { "epoch": 1.5409398472903346, "grad_norm": 0.6649454866021635, "learning_rate": 5.6328428520588995e-05, "loss": 2.8257, "step": 24823 }, { "epoch": 1.5410019243900925, "grad_norm": 0.6701716855335628, "learning_rate": 5.632484602428507e-05, "loss": 2.7406, "step": 24824 }, { "epoch": 1.5410640014898505, "grad_norm": 0.6830345482375506, "learning_rate": 5.632126349498289e-05, "loss": 2.7842, "step": 24825 }, { "epoch": 1.5411260785896084, "grad_norm": 0.5929741307593714, "learning_rate": 5.631768093270113e-05, "loss": 2.844, "step": 24826 }, { "epoch": 1.5411881556893663, "grad_norm": 0.6904419030561653, "learning_rate": 5.631409833745851e-05, "loss": 2.7785, "step": 24827 }, { "epoch": 1.5412502327891242, "grad_norm": 0.5821685885000775, "learning_rate": 5.63105157092737e-05, "loss": 2.7265, "step": 24828 }, { "epoch": 1.5413123098888821, "grad_norm": 0.6067571160944795, "learning_rate": 5.630693304816539e-05, "loss": 2.8077, "step": 24829 }, { "epoch": 1.54137438698864, "grad_norm": 0.5884494849963255, "learning_rate": 5.63033503541523e-05, "loss": 2.7826, "step": 24830 }, { "epoch": 1.5414364640883977, "grad_norm": 0.5943832283373054, "learning_rate": 5.6299767627253074e-05, "loss": 2.7689, "step": 24831 }, { "epoch": 1.5414985411881557, "grad_norm": 0.5879247661828398, "learning_rate": 5.6296184867486456e-05, "loss": 2.7435, "step": 24832 }, { "epoch": 1.5415606182879136, "grad_norm": 0.5530594587558204, "learning_rate": 5.6292602074871095e-05, "loss": 2.7731, "step": 24833 }, { "epoch": 1.5416226953876715, "grad_norm": 0.6065368091895119, "learning_rate": 5.628901924942571e-05, "loss": 2.8378, "step": 24834 }, { "epoch": 1.5416847724874294, "grad_norm": 0.6494436845261923, "learning_rate": 5.628543639116899e-05, "loss": 2.7776, "step": 24835 }, { "epoch": 1.5417468495871873, "grad_norm": 0.5827393544704544, "learning_rate": 5.62818535001196e-05, "loss": 2.758, "step": 24836 }, { "epoch": 1.541808926686945, "grad_norm": 0.6172717797402579, "learning_rate": 5.627827057629628e-05, "loss": 2.7734, "step": 24837 }, { "epoch": 1.541871003786703, "grad_norm": 0.5773128785176207, "learning_rate": 5.627468761971768e-05, "loss": 2.8357, "step": 24838 }, { "epoch": 1.5419330808864609, "grad_norm": 0.6599165803035977, "learning_rate": 5.6271104630402524e-05, "loss": 2.7971, "step": 24839 }, { "epoch": 1.5419951579862188, "grad_norm": 0.5483405503007748, "learning_rate": 5.6267521608369475e-05, "loss": 2.7798, "step": 24840 }, { "epoch": 1.5420572350859767, "grad_norm": 0.5914100918966299, "learning_rate": 5.6263938553637274e-05, "loss": 2.6759, "step": 24841 }, { "epoch": 1.5421193121857346, "grad_norm": 0.5995082956413029, "learning_rate": 5.626035546622457e-05, "loss": 2.7414, "step": 24842 }, { "epoch": 1.5421813892854925, "grad_norm": 0.5914258391790745, "learning_rate": 5.6256772346150066e-05, "loss": 2.8206, "step": 24843 }, { "epoch": 1.5422434663852504, "grad_norm": 0.5808991809127347, "learning_rate": 5.625318919343248e-05, "loss": 2.8208, "step": 24844 }, { "epoch": 1.5423055434850084, "grad_norm": 0.59412944113534, "learning_rate": 5.624960600809046e-05, "loss": 2.7483, "step": 24845 }, { "epoch": 1.5423676205847663, "grad_norm": 0.5448701348574339, "learning_rate": 5.624602279014275e-05, "loss": 2.7888, "step": 24846 }, { "epoch": 1.5424296976845242, "grad_norm": 0.5930736102240303, "learning_rate": 5.624243953960801e-05, "loss": 2.7087, "step": 24847 }, { "epoch": 1.5424917747842821, "grad_norm": 0.5626404375027622, "learning_rate": 5.6238856256504955e-05, "loss": 2.8222, "step": 24848 }, { "epoch": 1.54255385188404, "grad_norm": 0.6136989021784371, "learning_rate": 5.623527294085227e-05, "loss": 2.817, "step": 24849 }, { "epoch": 1.542615928983798, "grad_norm": 0.5869544803964742, "learning_rate": 5.623168959266866e-05, "loss": 2.7773, "step": 24850 }, { "epoch": 1.5426780060835559, "grad_norm": 0.5894861716378736, "learning_rate": 5.6228106211972795e-05, "loss": 2.8186, "step": 24851 }, { "epoch": 1.5427400831833138, "grad_norm": 0.5856136698731124, "learning_rate": 5.6224522798783406e-05, "loss": 2.8097, "step": 24852 }, { "epoch": 1.5428021602830717, "grad_norm": 0.5565129451548487, "learning_rate": 5.622093935311916e-05, "loss": 2.7215, "step": 24853 }, { "epoch": 1.5428642373828296, "grad_norm": 0.6102559007064794, "learning_rate": 5.6217355874998755e-05, "loss": 2.8215, "step": 24854 }, { "epoch": 1.5429263144825873, "grad_norm": 0.5993342311943989, "learning_rate": 5.621377236444092e-05, "loss": 2.7563, "step": 24855 }, { "epoch": 1.5429883915823452, "grad_norm": 0.5810214249522868, "learning_rate": 5.62101888214643e-05, "loss": 2.834, "step": 24856 }, { "epoch": 1.5430504686821032, "grad_norm": 0.5670690407541321, "learning_rate": 5.620660524608764e-05, "loss": 2.7174, "step": 24857 }, { "epoch": 1.543112545781861, "grad_norm": 0.575611941575699, "learning_rate": 5.62030216383296e-05, "loss": 2.8696, "step": 24858 }, { "epoch": 1.543174622881619, "grad_norm": 0.6069750844533285, "learning_rate": 5.619943799820889e-05, "loss": 2.8278, "step": 24859 }, { "epoch": 1.543236699981377, "grad_norm": 0.5641360860995146, "learning_rate": 5.619585432574422e-05, "loss": 2.7576, "step": 24860 }, { "epoch": 1.5432987770811346, "grad_norm": 0.5369620936208223, "learning_rate": 5.619227062095427e-05, "loss": 2.8293, "step": 24861 }, { "epoch": 1.5433608541808925, "grad_norm": 0.5367595018577211, "learning_rate": 5.618868688385773e-05, "loss": 2.697, "step": 24862 }, { "epoch": 1.5434229312806504, "grad_norm": 0.5440096312048979, "learning_rate": 5.618510311447331e-05, "loss": 2.7875, "step": 24863 }, { "epoch": 1.5434850083804084, "grad_norm": 0.5223619313774418, "learning_rate": 5.618151931281972e-05, "loss": 2.8008, "step": 24864 }, { "epoch": 1.5435470854801663, "grad_norm": 0.48992601704660044, "learning_rate": 5.617793547891562e-05, "loss": 2.723, "step": 24865 }, { "epoch": 1.5436091625799242, "grad_norm": 0.5254073867109423, "learning_rate": 5.617435161277975e-05, "loss": 2.81, "step": 24866 }, { "epoch": 1.5436712396796821, "grad_norm": 0.5429486058292533, "learning_rate": 5.6170767714430784e-05, "loss": 2.7295, "step": 24867 }, { "epoch": 1.54373331677944, "grad_norm": 0.5279925558385457, "learning_rate": 5.616718378388742e-05, "loss": 2.7921, "step": 24868 }, { "epoch": 1.543795393879198, "grad_norm": 0.5726440532473143, "learning_rate": 5.6163599821168365e-05, "loss": 2.7788, "step": 24869 }, { "epoch": 1.5438574709789559, "grad_norm": 0.5251013090387234, "learning_rate": 5.61600158262923e-05, "loss": 2.8057, "step": 24870 }, { "epoch": 1.5439195480787138, "grad_norm": 0.6732074698717132, "learning_rate": 5.615643179927796e-05, "loss": 2.81, "step": 24871 }, { "epoch": 1.5439816251784717, "grad_norm": 0.587261509679818, "learning_rate": 5.615284774014401e-05, "loss": 2.8618, "step": 24872 }, { "epoch": 1.5440437022782296, "grad_norm": 0.6038366655181805, "learning_rate": 5.6149263648909156e-05, "loss": 2.7103, "step": 24873 }, { "epoch": 1.5441057793779875, "grad_norm": 0.5808020803981754, "learning_rate": 5.614567952559211e-05, "loss": 2.8154, "step": 24874 }, { "epoch": 1.5441678564777455, "grad_norm": 0.5781416246240536, "learning_rate": 5.614209537021156e-05, "loss": 2.7705, "step": 24875 }, { "epoch": 1.5442299335775034, "grad_norm": 0.5922310372807482, "learning_rate": 5.613851118278619e-05, "loss": 2.7815, "step": 24876 }, { "epoch": 1.5442920106772613, "grad_norm": 0.5568811265156743, "learning_rate": 5.6134926963334746e-05, "loss": 2.745, "step": 24877 }, { "epoch": 1.5443540877770192, "grad_norm": 0.5546401097644833, "learning_rate": 5.613134271187589e-05, "loss": 2.7956, "step": 24878 }, { "epoch": 1.544416164876777, "grad_norm": 0.6108149852009293, "learning_rate": 5.612775842842831e-05, "loss": 2.8694, "step": 24879 }, { "epoch": 1.5444782419765348, "grad_norm": 0.5606663805809304, "learning_rate": 5.612417411301075e-05, "loss": 2.9242, "step": 24880 }, { "epoch": 1.5445403190762927, "grad_norm": 0.554384578677991, "learning_rate": 5.612058976564187e-05, "loss": 2.7735, "step": 24881 }, { "epoch": 1.5446023961760507, "grad_norm": 0.558774532061194, "learning_rate": 5.6117005386340406e-05, "loss": 2.7152, "step": 24882 }, { "epoch": 1.5446644732758086, "grad_norm": 0.564632664410364, "learning_rate": 5.611342097512502e-05, "loss": 2.7474, "step": 24883 }, { "epoch": 1.5447265503755665, "grad_norm": 0.6113188867815477, "learning_rate": 5.610983653201445e-05, "loss": 2.8012, "step": 24884 }, { "epoch": 1.5447886274753242, "grad_norm": 0.5801649189246655, "learning_rate": 5.6106252057027364e-05, "loss": 2.8082, "step": 24885 }, { "epoch": 1.5448507045750821, "grad_norm": 0.6259518289663795, "learning_rate": 5.610266755018249e-05, "loss": 2.8032, "step": 24886 }, { "epoch": 1.54491278167484, "grad_norm": 0.6694215513296747, "learning_rate": 5.609908301149851e-05, "loss": 2.7944, "step": 24887 }, { "epoch": 1.544974858774598, "grad_norm": 0.5624964237099447, "learning_rate": 5.609549844099413e-05, "loss": 2.763, "step": 24888 }, { "epoch": 1.5450369358743559, "grad_norm": 0.6577088172370291, "learning_rate": 5.609191383868807e-05, "loss": 2.8001, "step": 24889 }, { "epoch": 1.5450990129741138, "grad_norm": 0.6194797874521554, "learning_rate": 5.608832920459901e-05, "loss": 2.7264, "step": 24890 }, { "epoch": 1.5451610900738717, "grad_norm": 0.5820999105009433, "learning_rate": 5.608474453874566e-05, "loss": 2.6968, "step": 24891 }, { "epoch": 1.5452231671736296, "grad_norm": 0.6158095562422281, "learning_rate": 5.6081159841146714e-05, "loss": 2.7705, "step": 24892 }, { "epoch": 1.5452852442733875, "grad_norm": 0.6161180719445423, "learning_rate": 5.607757511182089e-05, "loss": 2.8111, "step": 24893 }, { "epoch": 1.5453473213731455, "grad_norm": 0.6085290906295111, "learning_rate": 5.6073990350786875e-05, "loss": 2.7495, "step": 24894 }, { "epoch": 1.5454093984729034, "grad_norm": 0.5635713335596958, "learning_rate": 5.6070405558063386e-05, "loss": 2.757, "step": 24895 }, { "epoch": 1.5454714755726613, "grad_norm": 0.6225720692092457, "learning_rate": 5.606682073366911e-05, "loss": 2.815, "step": 24896 }, { "epoch": 1.5455335526724192, "grad_norm": 0.5698166612190958, "learning_rate": 5.6063235877622754e-05, "loss": 2.783, "step": 24897 }, { "epoch": 1.5455956297721771, "grad_norm": 0.5813521192941007, "learning_rate": 5.6059650989943036e-05, "loss": 2.8274, "step": 24898 }, { "epoch": 1.545657706871935, "grad_norm": 0.6172881708251071, "learning_rate": 5.605606607064864e-05, "loss": 2.8628, "step": 24899 }, { "epoch": 1.545719783971693, "grad_norm": 0.5725978628996748, "learning_rate": 5.605248111975828e-05, "loss": 2.7717, "step": 24900 }, { "epoch": 1.5457818610714509, "grad_norm": 0.5712388758337925, "learning_rate": 5.604889613729065e-05, "loss": 2.7533, "step": 24901 }, { "epoch": 1.5458439381712088, "grad_norm": 0.6346616689214089, "learning_rate": 5.6045311123264456e-05, "loss": 2.8286, "step": 24902 }, { "epoch": 1.5459060152709665, "grad_norm": 0.6553059781631808, "learning_rate": 5.604172607769842e-05, "loss": 2.758, "step": 24903 }, { "epoch": 1.5459680923707244, "grad_norm": 0.6222242706791143, "learning_rate": 5.603814100061122e-05, "loss": 2.7464, "step": 24904 }, { "epoch": 1.5460301694704823, "grad_norm": 0.6158870565272968, "learning_rate": 5.603455589202159e-05, "loss": 2.8738, "step": 24905 }, { "epoch": 1.5460922465702402, "grad_norm": 0.6135401816962708, "learning_rate": 5.603097075194821e-05, "loss": 2.8019, "step": 24906 }, { "epoch": 1.5461543236699982, "grad_norm": 0.6260275262091504, "learning_rate": 5.602738558040979e-05, "loss": 2.8322, "step": 24907 }, { "epoch": 1.546216400769756, "grad_norm": 0.6018107508892345, "learning_rate": 5.602380037742503e-05, "loss": 2.7839, "step": 24908 }, { "epoch": 1.5462784778695138, "grad_norm": 0.6018970104403767, "learning_rate": 5.602021514301265e-05, "loss": 2.825, "step": 24909 }, { "epoch": 1.5463405549692717, "grad_norm": 0.6046906177923372, "learning_rate": 5.601662987719134e-05, "loss": 2.782, "step": 24910 }, { "epoch": 1.5464026320690296, "grad_norm": 0.6327552769537745, "learning_rate": 5.601304457997981e-05, "loss": 2.7939, "step": 24911 }, { "epoch": 1.5464647091687875, "grad_norm": 0.5741926433830726, "learning_rate": 5.600945925139677e-05, "loss": 2.8246, "step": 24912 }, { "epoch": 1.5465267862685455, "grad_norm": 0.6247006653185105, "learning_rate": 5.600587389146093e-05, "loss": 2.8422, "step": 24913 }, { "epoch": 1.5465888633683034, "grad_norm": 0.5317577852397517, "learning_rate": 5.600228850019098e-05, "loss": 2.7005, "step": 24914 }, { "epoch": 1.5466509404680613, "grad_norm": 0.6369384433416314, "learning_rate": 5.599870307760563e-05, "loss": 2.826, "step": 24915 }, { "epoch": 1.5467130175678192, "grad_norm": 0.6263532055922971, "learning_rate": 5.59951176237236e-05, "loss": 2.8337, "step": 24916 }, { "epoch": 1.5467750946675771, "grad_norm": 0.6371220440660341, "learning_rate": 5.5991532138563574e-05, "loss": 2.7974, "step": 24917 }, { "epoch": 1.546837171767335, "grad_norm": 0.6799990499012986, "learning_rate": 5.598794662214427e-05, "loss": 2.8088, "step": 24918 }, { "epoch": 1.546899248867093, "grad_norm": 0.6297171678014455, "learning_rate": 5.5984361074484403e-05, "loss": 2.7563, "step": 24919 }, { "epoch": 1.5469613259668509, "grad_norm": 0.7138772486279074, "learning_rate": 5.5980775495602666e-05, "loss": 2.8648, "step": 24920 }, { "epoch": 1.5470234030666088, "grad_norm": 0.5719042994596898, "learning_rate": 5.597718988551779e-05, "loss": 2.7591, "step": 24921 }, { "epoch": 1.5470854801663667, "grad_norm": 0.6695903035386428, "learning_rate": 5.5973604244248435e-05, "loss": 2.7576, "step": 24922 }, { "epoch": 1.5471475572661246, "grad_norm": 0.621902347327607, "learning_rate": 5.5970018571813366e-05, "loss": 2.7372, "step": 24923 }, { "epoch": 1.5472096343658825, "grad_norm": 0.6590585781033176, "learning_rate": 5.5966432868231235e-05, "loss": 2.7575, "step": 24924 }, { "epoch": 1.5472717114656405, "grad_norm": 0.6738462362757924, "learning_rate": 5.596284713352079e-05, "loss": 2.7296, "step": 24925 }, { "epoch": 1.5473337885653984, "grad_norm": 0.6344673747188835, "learning_rate": 5.5959261367700714e-05, "loss": 2.7195, "step": 24926 }, { "epoch": 1.547395865665156, "grad_norm": 0.6988086670010543, "learning_rate": 5.5955675570789735e-05, "loss": 2.8091, "step": 24927 }, { "epoch": 1.547457942764914, "grad_norm": 0.5929405064949412, "learning_rate": 5.595208974280654e-05, "loss": 2.759, "step": 24928 }, { "epoch": 1.547520019864672, "grad_norm": 0.642267354933415, "learning_rate": 5.594850388376985e-05, "loss": 2.8356, "step": 24929 }, { "epoch": 1.5475820969644298, "grad_norm": 0.5388222310973267, "learning_rate": 5.5944917993698376e-05, "loss": 2.7734, "step": 24930 }, { "epoch": 1.5476441740641877, "grad_norm": 0.6658430798891051, "learning_rate": 5.594133207261083e-05, "loss": 2.7565, "step": 24931 }, { "epoch": 1.5477062511639457, "grad_norm": 0.5860018631235, "learning_rate": 5.59377461205259e-05, "loss": 2.8806, "step": 24932 }, { "epoch": 1.5477683282637034, "grad_norm": 0.6030481497074932, "learning_rate": 5.593416013746231e-05, "loss": 2.8065, "step": 24933 }, { "epoch": 1.5478304053634613, "grad_norm": 0.6185519190377909, "learning_rate": 5.5930574123438774e-05, "loss": 2.7761, "step": 24934 }, { "epoch": 1.5478924824632192, "grad_norm": 0.5375431043507434, "learning_rate": 5.5926988078473966e-05, "loss": 2.6861, "step": 24935 }, { "epoch": 1.5479545595629771, "grad_norm": 0.6344323794718507, "learning_rate": 5.592340200258664e-05, "loss": 2.7046, "step": 24936 }, { "epoch": 1.548016636662735, "grad_norm": 0.5415780624645385, "learning_rate": 5.591981589579549e-05, "loss": 2.7577, "step": 24937 }, { "epoch": 1.548078713762493, "grad_norm": 0.6888930123109193, "learning_rate": 5.591622975811923e-05, "loss": 2.7925, "step": 24938 }, { "epoch": 1.5481407908622509, "grad_norm": 0.5639754653524937, "learning_rate": 5.5912643589576566e-05, "loss": 2.7616, "step": 24939 }, { "epoch": 1.5482028679620088, "grad_norm": 0.6818747569222721, "learning_rate": 5.590905739018618e-05, "loss": 2.8402, "step": 24940 }, { "epoch": 1.5482649450617667, "grad_norm": 0.5673896145700136, "learning_rate": 5.590547115996683e-05, "loss": 2.7446, "step": 24941 }, { "epoch": 1.5483270221615246, "grad_norm": 0.6439073027852336, "learning_rate": 5.5901884898937185e-05, "loss": 2.8606, "step": 24942 }, { "epoch": 1.5483890992612825, "grad_norm": 0.5994409330777046, "learning_rate": 5.5898298607115996e-05, "loss": 2.8882, "step": 24943 }, { "epoch": 1.5484511763610405, "grad_norm": 0.5854747216251251, "learning_rate": 5.5894712284521936e-05, "loss": 2.7363, "step": 24944 }, { "epoch": 1.5485132534607984, "grad_norm": 0.5735903304997506, "learning_rate": 5.589112593117374e-05, "loss": 2.6786, "step": 24945 }, { "epoch": 1.5485753305605563, "grad_norm": 0.5136354672833828, "learning_rate": 5.588753954709009e-05, "loss": 2.6894, "step": 24946 }, { "epoch": 1.5486374076603142, "grad_norm": 0.5820220588111977, "learning_rate": 5.5883953132289735e-05, "loss": 2.7417, "step": 24947 }, { "epoch": 1.5486994847600721, "grad_norm": 0.5668239500281779, "learning_rate": 5.5880366686791364e-05, "loss": 2.8116, "step": 24948 }, { "epoch": 1.54876156185983, "grad_norm": 0.5767117056549467, "learning_rate": 5.58767802106137e-05, "loss": 2.8192, "step": 24949 }, { "epoch": 1.548823638959588, "grad_norm": 0.6241669109337213, "learning_rate": 5.587319370377543e-05, "loss": 2.8418, "step": 24950 }, { "epoch": 1.5488857160593457, "grad_norm": 0.5236733602656114, "learning_rate": 5.586960716629529e-05, "loss": 2.7185, "step": 24951 }, { "epoch": 1.5489477931591036, "grad_norm": 0.5552308014217535, "learning_rate": 5.586602059819199e-05, "loss": 2.7906, "step": 24952 }, { "epoch": 1.5490098702588615, "grad_norm": 0.5791618225754986, "learning_rate": 5.5862433999484234e-05, "loss": 2.7973, "step": 24953 }, { "epoch": 1.5490719473586194, "grad_norm": 0.5898116147211627, "learning_rate": 5.585884737019074e-05, "loss": 2.7699, "step": 24954 }, { "epoch": 1.5491340244583773, "grad_norm": 0.6227649541182666, "learning_rate": 5.58552607103302e-05, "loss": 2.9545, "step": 24955 }, { "epoch": 1.5491961015581353, "grad_norm": 0.5614968467718977, "learning_rate": 5.585167401992136e-05, "loss": 2.7549, "step": 24956 }, { "epoch": 1.549258178657893, "grad_norm": 0.6276052535642549, "learning_rate": 5.584808729898292e-05, "loss": 2.8396, "step": 24957 }, { "epoch": 1.5493202557576509, "grad_norm": 0.5796134963186641, "learning_rate": 5.584450054753357e-05, "loss": 2.7439, "step": 24958 }, { "epoch": 1.5493823328574088, "grad_norm": 0.5459872533512032, "learning_rate": 5.584091376559206e-05, "loss": 2.7549, "step": 24959 }, { "epoch": 1.5494444099571667, "grad_norm": 0.535476032137436, "learning_rate": 5.5837326953177085e-05, "loss": 2.791, "step": 24960 }, { "epoch": 1.5495064870569246, "grad_norm": 0.614660740411942, "learning_rate": 5.5833740110307356e-05, "loss": 2.8158, "step": 24961 }, { "epoch": 1.5495685641566825, "grad_norm": 0.5954016501088288, "learning_rate": 5.583015323700157e-05, "loss": 2.789, "step": 24962 }, { "epoch": 1.5496306412564405, "grad_norm": 0.6208828022042101, "learning_rate": 5.582656633327849e-05, "loss": 2.7692, "step": 24963 }, { "epoch": 1.5496927183561984, "grad_norm": 0.5363162201229037, "learning_rate": 5.5822979399156775e-05, "loss": 2.6915, "step": 24964 }, { "epoch": 1.5497547954559563, "grad_norm": 0.5750595217456527, "learning_rate": 5.581939243465517e-05, "loss": 2.7001, "step": 24965 }, { "epoch": 1.5498168725557142, "grad_norm": 0.5056503924676562, "learning_rate": 5.5815805439792377e-05, "loss": 2.6779, "step": 24966 }, { "epoch": 1.5498789496554721, "grad_norm": 0.5339697611887331, "learning_rate": 5.581221841458712e-05, "loss": 2.7452, "step": 24967 }, { "epoch": 1.54994102675523, "grad_norm": 0.5554882233931069, "learning_rate": 5.5808631359058104e-05, "loss": 2.8068, "step": 24968 }, { "epoch": 1.550003103854988, "grad_norm": 0.562930286924125, "learning_rate": 5.580504427322405e-05, "loss": 2.8039, "step": 24969 }, { "epoch": 1.5500651809547459, "grad_norm": 0.5046343311848421, "learning_rate": 5.580145715710367e-05, "loss": 2.77, "step": 24970 }, { "epoch": 1.5501272580545038, "grad_norm": 0.6002119186790373, "learning_rate": 5.579787001071569e-05, "loss": 2.8177, "step": 24971 }, { "epoch": 1.5501893351542617, "grad_norm": 0.49161359842727614, "learning_rate": 5.579428283407882e-05, "loss": 2.7826, "step": 24972 }, { "epoch": 1.5502514122540196, "grad_norm": 0.5425661913752015, "learning_rate": 5.5790695627211743e-05, "loss": 2.7159, "step": 24973 }, { "epoch": 1.5503134893537776, "grad_norm": 0.5540490035098454, "learning_rate": 5.5787108390133226e-05, "loss": 2.8574, "step": 24974 }, { "epoch": 1.5503755664535352, "grad_norm": 0.5187942325164103, "learning_rate": 5.5783521122861935e-05, "loss": 2.7534, "step": 24975 }, { "epoch": 1.5504376435532932, "grad_norm": 0.5672485928171255, "learning_rate": 5.577993382541664e-05, "loss": 2.7732, "step": 24976 }, { "epoch": 1.550499720653051, "grad_norm": 0.5209044185448096, "learning_rate": 5.577634649781601e-05, "loss": 2.7378, "step": 24977 }, { "epoch": 1.550561797752809, "grad_norm": 0.5546629994313956, "learning_rate": 5.5772759140078775e-05, "loss": 2.8029, "step": 24978 }, { "epoch": 1.550623874852567, "grad_norm": 0.522879174650135, "learning_rate": 5.576917175222367e-05, "loss": 2.7548, "step": 24979 }, { "epoch": 1.5506859519523248, "grad_norm": 0.4854775923312569, "learning_rate": 5.5765584334269386e-05, "loss": 2.7865, "step": 24980 }, { "epoch": 1.5507480290520825, "grad_norm": 0.5697781397895155, "learning_rate": 5.576199688623464e-05, "loss": 2.8493, "step": 24981 }, { "epoch": 1.5508101061518405, "grad_norm": 0.5582760527630705, "learning_rate": 5.5758409408138166e-05, "loss": 2.6808, "step": 24982 }, { "epoch": 1.5508721832515984, "grad_norm": 0.6004642270205963, "learning_rate": 5.575482189999868e-05, "loss": 2.8492, "step": 24983 }, { "epoch": 1.5509342603513563, "grad_norm": 0.5872545825427055, "learning_rate": 5.575123436183487e-05, "loss": 2.7337, "step": 24984 }, { "epoch": 1.5509963374511142, "grad_norm": 0.5969194096126729, "learning_rate": 5.574764679366549e-05, "loss": 2.7854, "step": 24985 }, { "epoch": 1.5510584145508721, "grad_norm": 0.6326233145407381, "learning_rate": 5.574405919550923e-05, "loss": 2.7469, "step": 24986 }, { "epoch": 1.55112049165063, "grad_norm": 0.6133980852349679, "learning_rate": 5.574047156738483e-05, "loss": 2.816, "step": 24987 }, { "epoch": 1.551182568750388, "grad_norm": 0.6646913355979855, "learning_rate": 5.5736883909311e-05, "loss": 2.8118, "step": 24988 }, { "epoch": 1.5512446458501459, "grad_norm": 0.5006575730736508, "learning_rate": 5.573329622130644e-05, "loss": 2.7393, "step": 24989 }, { "epoch": 1.5513067229499038, "grad_norm": 0.6086603522431047, "learning_rate": 5.5729708503389896e-05, "loss": 2.8121, "step": 24990 }, { "epoch": 1.5513688000496617, "grad_norm": 0.5297521343575256, "learning_rate": 5.5726120755580057e-05, "loss": 2.7885, "step": 24991 }, { "epoch": 1.5514308771494196, "grad_norm": 0.5670290712560158, "learning_rate": 5.572253297789567e-05, "loss": 2.7464, "step": 24992 }, { "epoch": 1.5514929542491775, "grad_norm": 0.5189767617662543, "learning_rate": 5.571894517035543e-05, "loss": 2.7682, "step": 24993 }, { "epoch": 1.5515550313489355, "grad_norm": 0.5961717357092949, "learning_rate": 5.571535733297807e-05, "loss": 2.7677, "step": 24994 }, { "epoch": 1.5516171084486934, "grad_norm": 0.5932232126054945, "learning_rate": 5.5711769465782306e-05, "loss": 2.8604, "step": 24995 }, { "epoch": 1.5516791855484513, "grad_norm": 0.5316333789878199, "learning_rate": 5.570818156878684e-05, "loss": 2.7591, "step": 24996 }, { "epoch": 1.5517412626482092, "grad_norm": 0.5807784249209049, "learning_rate": 5.570459364201043e-05, "loss": 2.7978, "step": 24997 }, { "epoch": 1.5518033397479671, "grad_norm": 0.5240920799246503, "learning_rate": 5.570100568547174e-05, "loss": 2.8186, "step": 24998 }, { "epoch": 1.5518654168477248, "grad_norm": 0.5621233579846187, "learning_rate": 5.5697417699189546e-05, "loss": 2.7046, "step": 24999 }, { "epoch": 1.5519274939474828, "grad_norm": 0.5971134188948206, "learning_rate": 5.569382968318252e-05, "loss": 2.7847, "step": 25000 }, { "epoch": 1.5519895710472407, "grad_norm": 0.5589818339953192, "learning_rate": 5.569024163746941e-05, "loss": 2.6646, "step": 25001 }, { "epoch": 1.5520516481469986, "grad_norm": 0.6003570616752769, "learning_rate": 5.5686653562068934e-05, "loss": 2.8066, "step": 25002 }, { "epoch": 1.5521137252467565, "grad_norm": 0.5221004490229925, "learning_rate": 5.5683065456999794e-05, "loss": 2.7184, "step": 25003 }, { "epoch": 1.5521758023465144, "grad_norm": 0.5597724216684783, "learning_rate": 5.567947732228074e-05, "loss": 2.8213, "step": 25004 }, { "epoch": 1.5522378794462721, "grad_norm": 0.5418583060204295, "learning_rate": 5.567588915793045e-05, "loss": 2.7612, "step": 25005 }, { "epoch": 1.55229995654603, "grad_norm": 0.5265618072414652, "learning_rate": 5.567230096396768e-05, "loss": 2.7453, "step": 25006 }, { "epoch": 1.552362033645788, "grad_norm": 0.5907877006672958, "learning_rate": 5.566871274041114e-05, "loss": 2.8047, "step": 25007 }, { "epoch": 1.5524241107455459, "grad_norm": 0.5129304473827166, "learning_rate": 5.566512448727955e-05, "loss": 2.7846, "step": 25008 }, { "epoch": 1.5524861878453038, "grad_norm": 0.5962952969297328, "learning_rate": 5.566153620459163e-05, "loss": 2.8756, "step": 25009 }, { "epoch": 1.5525482649450617, "grad_norm": 0.5224460492731211, "learning_rate": 5.56579478923661e-05, "loss": 2.7934, "step": 25010 }, { "epoch": 1.5526103420448196, "grad_norm": 0.5712362411728564, "learning_rate": 5.5654359550621684e-05, "loss": 2.7803, "step": 25011 }, { "epoch": 1.5526724191445775, "grad_norm": 0.6144073013320728, "learning_rate": 5.56507711793771e-05, "loss": 2.8198, "step": 25012 }, { "epoch": 1.5527344962443355, "grad_norm": 0.5195219569701639, "learning_rate": 5.5647182778651085e-05, "loss": 2.7569, "step": 25013 }, { "epoch": 1.5527965733440934, "grad_norm": 0.6180419319118144, "learning_rate": 5.564359434846232e-05, "loss": 2.7725, "step": 25014 }, { "epoch": 1.5528586504438513, "grad_norm": 0.5086979409184598, "learning_rate": 5.564000588882958e-05, "loss": 2.8635, "step": 25015 }, { "epoch": 1.5529207275436092, "grad_norm": 0.6068940644653281, "learning_rate": 5.5636417399771546e-05, "loss": 2.8089, "step": 25016 }, { "epoch": 1.5529828046433671, "grad_norm": 0.54914366315999, "learning_rate": 5.563282888130694e-05, "loss": 2.8489, "step": 25017 }, { "epoch": 1.553044881743125, "grad_norm": 0.5382866182322573, "learning_rate": 5.562924033345451e-05, "loss": 2.7108, "step": 25018 }, { "epoch": 1.553106958842883, "grad_norm": 0.5686710460063029, "learning_rate": 5.562565175623298e-05, "loss": 2.7316, "step": 25019 }, { "epoch": 1.5531690359426409, "grad_norm": 0.5492025706701273, "learning_rate": 5.5622063149661055e-05, "loss": 2.758, "step": 25020 }, { "epoch": 1.5532311130423988, "grad_norm": 0.5408486063710473, "learning_rate": 5.5618474513757445e-05, "loss": 2.7975, "step": 25021 }, { "epoch": 1.5532931901421567, "grad_norm": 0.618295544084808, "learning_rate": 5.561488584854091e-05, "loss": 2.7937, "step": 25022 }, { "epoch": 1.5533552672419144, "grad_norm": 0.5168609544812983, "learning_rate": 5.561129715403013e-05, "loss": 2.7897, "step": 25023 }, { "epoch": 1.5534173443416723, "grad_norm": 0.6809073715232238, "learning_rate": 5.5607708430243875e-05, "loss": 2.8046, "step": 25024 }, { "epoch": 1.5534794214414303, "grad_norm": 0.559416878054347, "learning_rate": 5.5604119677200826e-05, "loss": 2.8082, "step": 25025 }, { "epoch": 1.5535414985411882, "grad_norm": 0.59239071300859, "learning_rate": 5.560053089491973e-05, "loss": 2.7711, "step": 25026 }, { "epoch": 1.553603575640946, "grad_norm": 0.5717573284551865, "learning_rate": 5.55969420834193e-05, "loss": 2.9127, "step": 25027 }, { "epoch": 1.553665652740704, "grad_norm": 0.5518553791035944, "learning_rate": 5.559335324271828e-05, "loss": 2.7436, "step": 25028 }, { "epoch": 1.5537277298404617, "grad_norm": 0.5786803902769074, "learning_rate": 5.5589764372835364e-05, "loss": 2.8209, "step": 25029 }, { "epoch": 1.5537898069402196, "grad_norm": 0.5535809937982347, "learning_rate": 5.558617547378929e-05, "loss": 2.803, "step": 25030 }, { "epoch": 1.5538518840399775, "grad_norm": 0.5346417982411485, "learning_rate": 5.5582586545598804e-05, "loss": 2.721, "step": 25031 }, { "epoch": 1.5539139611397355, "grad_norm": 0.5397917748699819, "learning_rate": 5.557899758828258e-05, "loss": 2.8355, "step": 25032 }, { "epoch": 1.5539760382394934, "grad_norm": 0.5786616695467725, "learning_rate": 5.557540860185938e-05, "loss": 2.7526, "step": 25033 }, { "epoch": 1.5540381153392513, "grad_norm": 0.5486238038194939, "learning_rate": 5.557181958634792e-05, "loss": 2.817, "step": 25034 }, { "epoch": 1.5541001924390092, "grad_norm": 0.5996557460088797, "learning_rate": 5.556823054176693e-05, "loss": 2.809, "step": 25035 }, { "epoch": 1.5541622695387671, "grad_norm": 0.5224843527992578, "learning_rate": 5.556464146813511e-05, "loss": 2.7495, "step": 25036 }, { "epoch": 1.554224346638525, "grad_norm": 0.6380619321288687, "learning_rate": 5.556105236547123e-05, "loss": 2.8004, "step": 25037 }, { "epoch": 1.554286423738283, "grad_norm": 0.5588783126217541, "learning_rate": 5.555746323379399e-05, "loss": 2.7914, "step": 25038 }, { "epoch": 1.5543485008380409, "grad_norm": 0.6335088642421506, "learning_rate": 5.5553874073122104e-05, "loss": 2.813, "step": 25039 }, { "epoch": 1.5544105779377988, "grad_norm": 0.5017916646060607, "learning_rate": 5.5550284883474316e-05, "loss": 2.7917, "step": 25040 }, { "epoch": 1.5544726550375567, "grad_norm": 0.5902074488336857, "learning_rate": 5.554669566486933e-05, "loss": 2.8144, "step": 25041 }, { "epoch": 1.5545347321373146, "grad_norm": 0.6027420666464072, "learning_rate": 5.5543106417325906e-05, "loss": 2.7681, "step": 25042 }, { "epoch": 1.5545968092370726, "grad_norm": 0.5905557339535326, "learning_rate": 5.553951714086273e-05, "loss": 2.7669, "step": 25043 }, { "epoch": 1.5546588863368305, "grad_norm": 0.5668687707796376, "learning_rate": 5.5535927835498566e-05, "loss": 2.7518, "step": 25044 }, { "epoch": 1.5547209634365884, "grad_norm": 0.5326998784579451, "learning_rate": 5.553233850125211e-05, "loss": 2.8552, "step": 25045 }, { "epoch": 1.5547830405363463, "grad_norm": 0.5620992875668637, "learning_rate": 5.552874913814211e-05, "loss": 2.754, "step": 25046 }, { "epoch": 1.554845117636104, "grad_norm": 0.5237449483240906, "learning_rate": 5.5525159746187274e-05, "loss": 2.8753, "step": 25047 }, { "epoch": 1.554907194735862, "grad_norm": 0.618394323675918, "learning_rate": 5.5521570325406347e-05, "loss": 2.7468, "step": 25048 }, { "epoch": 1.5549692718356198, "grad_norm": 0.5568252257377521, "learning_rate": 5.5517980875818045e-05, "loss": 2.8127, "step": 25049 }, { "epoch": 1.5550313489353778, "grad_norm": 0.6117575846265404, "learning_rate": 5.551439139744108e-05, "loss": 2.7029, "step": 25050 }, { "epoch": 1.5550934260351357, "grad_norm": 0.6072682128050187, "learning_rate": 5.551080189029422e-05, "loss": 2.7658, "step": 25051 }, { "epoch": 1.5551555031348936, "grad_norm": 0.6379921472030284, "learning_rate": 5.5507212354396155e-05, "loss": 2.7591, "step": 25052 }, { "epoch": 1.5552175802346513, "grad_norm": 0.6117477439352385, "learning_rate": 5.550362278976564e-05, "loss": 2.785, "step": 25053 }, { "epoch": 1.5552796573344092, "grad_norm": 0.645580224204448, "learning_rate": 5.5500033196421375e-05, "loss": 2.7103, "step": 25054 }, { "epoch": 1.5553417344341671, "grad_norm": 0.6956466986055876, "learning_rate": 5.549644357438212e-05, "loss": 2.8517, "step": 25055 }, { "epoch": 1.555403811533925, "grad_norm": 0.5165981011373879, "learning_rate": 5.5492853923666576e-05, "loss": 2.7377, "step": 25056 }, { "epoch": 1.555465888633683, "grad_norm": 0.6963372768920736, "learning_rate": 5.548926424429347e-05, "loss": 2.7712, "step": 25057 }, { "epoch": 1.5555279657334409, "grad_norm": 0.5520147214996209, "learning_rate": 5.548567453628155e-05, "loss": 2.8184, "step": 25058 }, { "epoch": 1.5555900428331988, "grad_norm": 0.606201884729809, "learning_rate": 5.548208479964952e-05, "loss": 2.712, "step": 25059 }, { "epoch": 1.5556521199329567, "grad_norm": 0.5932013335716501, "learning_rate": 5.547849503441614e-05, "loss": 2.8007, "step": 25060 }, { "epoch": 1.5557141970327146, "grad_norm": 0.6208934820936799, "learning_rate": 5.547490524060011e-05, "loss": 2.8124, "step": 25061 }, { "epoch": 1.5557762741324725, "grad_norm": 0.6862340021266388, "learning_rate": 5.547131541822018e-05, "loss": 2.7323, "step": 25062 }, { "epoch": 1.5558383512322305, "grad_norm": 0.5712027460082348, "learning_rate": 5.546772556729506e-05, "loss": 2.812, "step": 25063 }, { "epoch": 1.5559004283319884, "grad_norm": 0.6627984751108271, "learning_rate": 5.54641356878435e-05, "loss": 2.7748, "step": 25064 }, { "epoch": 1.5559625054317463, "grad_norm": 0.5937660365699656, "learning_rate": 5.54605457798842e-05, "loss": 2.7858, "step": 25065 }, { "epoch": 1.5560245825315042, "grad_norm": 0.5666387146712841, "learning_rate": 5.545695584343591e-05, "loss": 2.6936, "step": 25066 }, { "epoch": 1.5560866596312621, "grad_norm": 0.5731623787375223, "learning_rate": 5.545336587851737e-05, "loss": 2.899, "step": 25067 }, { "epoch": 1.55614873673102, "grad_norm": 0.5679737893658885, "learning_rate": 5.544977588514728e-05, "loss": 2.7889, "step": 25068 }, { "epoch": 1.556210813830778, "grad_norm": 0.6077687049130799, "learning_rate": 5.54461858633444e-05, "loss": 2.7786, "step": 25069 }, { "epoch": 1.556272890930536, "grad_norm": 0.5419056316372425, "learning_rate": 5.544259581312744e-05, "loss": 2.8085, "step": 25070 }, { "epoch": 1.5563349680302936, "grad_norm": 0.5943540963348893, "learning_rate": 5.5439005734515146e-05, "loss": 2.7175, "step": 25071 }, { "epoch": 1.5563970451300515, "grad_norm": 0.6170917629882794, "learning_rate": 5.543541562752623e-05, "loss": 2.7753, "step": 25072 }, { "epoch": 1.5564591222298094, "grad_norm": 0.5879111809162563, "learning_rate": 5.543182549217943e-05, "loss": 2.792, "step": 25073 }, { "epoch": 1.5565211993295673, "grad_norm": 0.6336746230058199, "learning_rate": 5.542823532849349e-05, "loss": 2.7317, "step": 25074 }, { "epoch": 1.5565832764293253, "grad_norm": 0.6235454561462709, "learning_rate": 5.5424645136487116e-05, "loss": 2.7499, "step": 25075 }, { "epoch": 1.5566453535290832, "grad_norm": 0.6752200801015488, "learning_rate": 5.5421054916179063e-05, "loss": 2.8427, "step": 25076 }, { "epoch": 1.5567074306288409, "grad_norm": 0.5813061082072477, "learning_rate": 5.541746466758804e-05, "loss": 2.7293, "step": 25077 }, { "epoch": 1.5567695077285988, "grad_norm": 0.6431541527831195, "learning_rate": 5.541387439073279e-05, "loss": 2.7508, "step": 25078 }, { "epoch": 1.5568315848283567, "grad_norm": 0.6772902074856989, "learning_rate": 5.541028408563204e-05, "loss": 2.7953, "step": 25079 }, { "epoch": 1.5568936619281146, "grad_norm": 0.676199260049137, "learning_rate": 5.540669375230453e-05, "loss": 2.742, "step": 25080 }, { "epoch": 1.5569557390278725, "grad_norm": 0.643059848528933, "learning_rate": 5.540310339076899e-05, "loss": 2.8145, "step": 25081 }, { "epoch": 1.5570178161276305, "grad_norm": 0.6057279708688685, "learning_rate": 5.539951300104412e-05, "loss": 2.7776, "step": 25082 }, { "epoch": 1.5570798932273884, "grad_norm": 0.6205120121348775, "learning_rate": 5.539592258314872e-05, "loss": 2.8426, "step": 25083 }, { "epoch": 1.5571419703271463, "grad_norm": 0.6276619490197844, "learning_rate": 5.539233213710145e-05, "loss": 2.7446, "step": 25084 }, { "epoch": 1.5572040474269042, "grad_norm": 0.6621146187813761, "learning_rate": 5.538874166292109e-05, "loss": 2.8963, "step": 25085 }, { "epoch": 1.5572661245266621, "grad_norm": 0.6255553400927855, "learning_rate": 5.538515116062636e-05, "loss": 2.7533, "step": 25086 }, { "epoch": 1.55732820162642, "grad_norm": 0.679965565244344, "learning_rate": 5.5381560630235985e-05, "loss": 2.7306, "step": 25087 }, { "epoch": 1.557390278726178, "grad_norm": 0.5842191039434005, "learning_rate": 5.5377970071768704e-05, "loss": 2.6878, "step": 25088 }, { "epoch": 1.5574523558259359, "grad_norm": 0.6164637801056599, "learning_rate": 5.5374379485243246e-05, "loss": 2.8229, "step": 25089 }, { "epoch": 1.5575144329256938, "grad_norm": 0.6149308197530213, "learning_rate": 5.5370788870678335e-05, "loss": 2.7593, "step": 25090 }, { "epoch": 1.5575765100254517, "grad_norm": 0.5838205114006266, "learning_rate": 5.5367198228092734e-05, "loss": 2.8402, "step": 25091 }, { "epoch": 1.5576385871252096, "grad_norm": 0.5857348536699751, "learning_rate": 5.536360755750515e-05, "loss": 2.81, "step": 25092 }, { "epoch": 1.5577006642249676, "grad_norm": 0.5475775701623697, "learning_rate": 5.536001685893432e-05, "loss": 2.8486, "step": 25093 }, { "epoch": 1.5577627413247255, "grad_norm": 0.617072348173845, "learning_rate": 5.5356426132398977e-05, "loss": 2.7295, "step": 25094 }, { "epoch": 1.5578248184244832, "grad_norm": 0.549987735394801, "learning_rate": 5.5352835377917856e-05, "loss": 2.6835, "step": 25095 }, { "epoch": 1.557886895524241, "grad_norm": 0.622033350159301, "learning_rate": 5.534924459550971e-05, "loss": 2.8581, "step": 25096 }, { "epoch": 1.557948972623999, "grad_norm": 0.5648186895739021, "learning_rate": 5.534565378519324e-05, "loss": 2.7146, "step": 25097 }, { "epoch": 1.558011049723757, "grad_norm": 0.6390244330375058, "learning_rate": 5.5342062946987195e-05, "loss": 2.8338, "step": 25098 }, { "epoch": 1.5580731268235148, "grad_norm": 0.5752782842769057, "learning_rate": 5.533847208091032e-05, "loss": 2.8617, "step": 25099 }, { "epoch": 1.5581352039232728, "grad_norm": 0.5225642877481315, "learning_rate": 5.5334881186981336e-05, "loss": 2.6943, "step": 25100 }, { "epoch": 1.5581972810230305, "grad_norm": 0.5775068948177258, "learning_rate": 5.5331290265218995e-05, "loss": 2.8146, "step": 25101 }, { "epoch": 1.5582593581227884, "grad_norm": 0.5396484838893676, "learning_rate": 5.5327699315642e-05, "loss": 2.7649, "step": 25102 }, { "epoch": 1.5583214352225463, "grad_norm": 0.5690970336725485, "learning_rate": 5.532410833826912e-05, "loss": 2.8701, "step": 25103 }, { "epoch": 1.5583835123223042, "grad_norm": 0.6024693923314385, "learning_rate": 5.532051733311906e-05, "loss": 2.8139, "step": 25104 }, { "epoch": 1.5584455894220621, "grad_norm": 0.5213896704699353, "learning_rate": 5.5316926300210584e-05, "loss": 2.7257, "step": 25105 }, { "epoch": 1.55850766652182, "grad_norm": 0.6039425719319702, "learning_rate": 5.531333523956239e-05, "loss": 2.8648, "step": 25106 }, { "epoch": 1.558569743621578, "grad_norm": 0.6268165178108371, "learning_rate": 5.5309744151193264e-05, "loss": 2.7952, "step": 25107 }, { "epoch": 1.5586318207213359, "grad_norm": 0.5207209624821689, "learning_rate": 5.530615303512189e-05, "loss": 2.7687, "step": 25108 }, { "epoch": 1.5586938978210938, "grad_norm": 0.6186136114675854, "learning_rate": 5.5302561891367046e-05, "loss": 2.8621, "step": 25109 }, { "epoch": 1.5587559749208517, "grad_norm": 0.501911413931095, "learning_rate": 5.529897071994745e-05, "loss": 2.7483, "step": 25110 }, { "epoch": 1.5588180520206096, "grad_norm": 0.5681919493638969, "learning_rate": 5.5295379520881815e-05, "loss": 2.6661, "step": 25111 }, { "epoch": 1.5588801291203676, "grad_norm": 0.5444229740766398, "learning_rate": 5.529178829418893e-05, "loss": 2.7874, "step": 25112 }, { "epoch": 1.5589422062201255, "grad_norm": 0.6467564174671384, "learning_rate": 5.528819703988748e-05, "loss": 2.7419, "step": 25113 }, { "epoch": 1.5590042833198834, "grad_norm": 0.5799947351183846, "learning_rate": 5.528460575799622e-05, "loss": 2.8294, "step": 25114 }, { "epoch": 1.5590663604196413, "grad_norm": 0.6207011439470063, "learning_rate": 5.528101444853388e-05, "loss": 2.7532, "step": 25115 }, { "epoch": 1.5591284375193992, "grad_norm": 0.6011505698232988, "learning_rate": 5.5277423111519236e-05, "loss": 2.8363, "step": 25116 }, { "epoch": 1.5591905146191571, "grad_norm": 0.5807083251588907, "learning_rate": 5.5273831746970984e-05, "loss": 2.7817, "step": 25117 }, { "epoch": 1.5592525917189148, "grad_norm": 0.6321372915743392, "learning_rate": 5.5270240354907856e-05, "loss": 2.8041, "step": 25118 }, { "epoch": 1.5593146688186728, "grad_norm": 0.5378915536023382, "learning_rate": 5.526664893534862e-05, "loss": 2.7854, "step": 25119 }, { "epoch": 1.5593767459184307, "grad_norm": 0.64644051025569, "learning_rate": 5.5263057488312e-05, "loss": 2.7666, "step": 25120 }, { "epoch": 1.5594388230181886, "grad_norm": 0.6452536031895107, "learning_rate": 5.5259466013816715e-05, "loss": 2.7486, "step": 25121 }, { "epoch": 1.5595009001179465, "grad_norm": 0.562293359099766, "learning_rate": 5.5255874511881535e-05, "loss": 2.7534, "step": 25122 }, { "epoch": 1.5595629772177044, "grad_norm": 0.5671394338286557, "learning_rate": 5.525228298252518e-05, "loss": 2.754, "step": 25123 }, { "epoch": 1.5596250543174621, "grad_norm": 0.6365379591693177, "learning_rate": 5.5248691425766377e-05, "loss": 2.7007, "step": 25124 }, { "epoch": 1.55968713141722, "grad_norm": 0.6001380801262243, "learning_rate": 5.524509984162389e-05, "loss": 2.742, "step": 25125 }, { "epoch": 1.559749208516978, "grad_norm": 0.5668210319560851, "learning_rate": 5.5241508230116434e-05, "loss": 2.7656, "step": 25126 }, { "epoch": 1.5598112856167359, "grad_norm": 0.6655354488828037, "learning_rate": 5.5237916591262776e-05, "loss": 2.7418, "step": 25127 }, { "epoch": 1.5598733627164938, "grad_norm": 0.5577997553462042, "learning_rate": 5.5234324925081626e-05, "loss": 2.7308, "step": 25128 }, { "epoch": 1.5599354398162517, "grad_norm": 0.6365323174071718, "learning_rate": 5.523073323159171e-05, "loss": 2.7808, "step": 25129 }, { "epoch": 1.5599975169160096, "grad_norm": 0.5572400797084565, "learning_rate": 5.522714151081182e-05, "loss": 2.7404, "step": 25130 }, { "epoch": 1.5600595940157675, "grad_norm": 0.5172479115956083, "learning_rate": 5.522354976276063e-05, "loss": 2.7318, "step": 25131 }, { "epoch": 1.5601216711155255, "grad_norm": 0.6239283195201563, "learning_rate": 5.521995798745694e-05, "loss": 2.7711, "step": 25132 }, { "epoch": 1.5601837482152834, "grad_norm": 0.510118676477081, "learning_rate": 5.521636618491945e-05, "loss": 2.7526, "step": 25133 }, { "epoch": 1.5602458253150413, "grad_norm": 0.5921848397711033, "learning_rate": 5.521277435516693e-05, "loss": 2.7776, "step": 25134 }, { "epoch": 1.5603079024147992, "grad_norm": 0.5207212486461257, "learning_rate": 5.5209182498218095e-05, "loss": 2.7814, "step": 25135 }, { "epoch": 1.5603699795145571, "grad_norm": 0.538592755853783, "learning_rate": 5.5205590614091675e-05, "loss": 2.6854, "step": 25136 }, { "epoch": 1.560432056614315, "grad_norm": 0.52655654360518, "learning_rate": 5.520199870280644e-05, "loss": 2.7943, "step": 25137 }, { "epoch": 1.560494133714073, "grad_norm": 0.6083374727665709, "learning_rate": 5.519840676438109e-05, "loss": 2.8586, "step": 25138 }, { "epoch": 1.560556210813831, "grad_norm": 0.5407276144380826, "learning_rate": 5.5194814798834415e-05, "loss": 2.8605, "step": 25139 }, { "epoch": 1.5606182879135888, "grad_norm": 0.5883969632618856, "learning_rate": 5.519122280618513e-05, "loss": 2.7639, "step": 25140 }, { "epoch": 1.5606803650133467, "grad_norm": 0.5710452089149958, "learning_rate": 5.5187630786451974e-05, "loss": 2.8349, "step": 25141 }, { "epoch": 1.5607424421131044, "grad_norm": 0.5853530766668328, "learning_rate": 5.5184038739653675e-05, "loss": 2.8009, "step": 25142 }, { "epoch": 1.5608045192128623, "grad_norm": 0.5498964222999227, "learning_rate": 5.518044666580899e-05, "loss": 2.7819, "step": 25143 }, { "epoch": 1.5608665963126203, "grad_norm": 0.6072843586093195, "learning_rate": 5.5176854564936665e-05, "loss": 2.869, "step": 25144 }, { "epoch": 1.5609286734123782, "grad_norm": 0.5303790020761352, "learning_rate": 5.517326243705544e-05, "loss": 2.8377, "step": 25145 }, { "epoch": 1.560990750512136, "grad_norm": 0.6270981380282319, "learning_rate": 5.5169670282184025e-05, "loss": 2.6175, "step": 25146 }, { "epoch": 1.561052827611894, "grad_norm": 0.5192618280954413, "learning_rate": 5.516607810034121e-05, "loss": 2.719, "step": 25147 }, { "epoch": 1.5611149047116517, "grad_norm": 0.5512511230855215, "learning_rate": 5.51624858915457e-05, "loss": 2.7726, "step": 25148 }, { "epoch": 1.5611769818114096, "grad_norm": 0.521124783668464, "learning_rate": 5.5158893655816234e-05, "loss": 2.6603, "step": 25149 }, { "epoch": 1.5612390589111675, "grad_norm": 0.5422309693162288, "learning_rate": 5.51553013931716e-05, "loss": 2.7702, "step": 25150 }, { "epoch": 1.5613011360109255, "grad_norm": 0.513768147196788, "learning_rate": 5.515170910363048e-05, "loss": 2.7791, "step": 25151 }, { "epoch": 1.5613632131106834, "grad_norm": 0.5146807492154492, "learning_rate": 5.514811678721166e-05, "loss": 2.8383, "step": 25152 }, { "epoch": 1.5614252902104413, "grad_norm": 0.5321754921024936, "learning_rate": 5.514452444393386e-05, "loss": 2.7251, "step": 25153 }, { "epoch": 1.5614873673101992, "grad_norm": 0.5551060454120317, "learning_rate": 5.514093207381582e-05, "loss": 2.8362, "step": 25154 }, { "epoch": 1.5615494444099571, "grad_norm": 0.5536797731701999, "learning_rate": 5.51373396768763e-05, "loss": 2.7401, "step": 25155 }, { "epoch": 1.561611521509715, "grad_norm": 0.5485102591654873, "learning_rate": 5.5133747253134027e-05, "loss": 2.7609, "step": 25156 }, { "epoch": 1.561673598609473, "grad_norm": 0.5673257664483291, "learning_rate": 5.5130154802607747e-05, "loss": 2.7464, "step": 25157 }, { "epoch": 1.5617356757092309, "grad_norm": 0.5654435757263903, "learning_rate": 5.51265623253162e-05, "loss": 2.7424, "step": 25158 }, { "epoch": 1.5617977528089888, "grad_norm": 0.5326094064119669, "learning_rate": 5.5122969821278134e-05, "loss": 2.818, "step": 25159 }, { "epoch": 1.5618598299087467, "grad_norm": 0.5412976187622786, "learning_rate": 5.51193772905123e-05, "loss": 2.7365, "step": 25160 }, { "epoch": 1.5619219070085046, "grad_norm": 0.5658659890554765, "learning_rate": 5.5115784733037425e-05, "loss": 2.8003, "step": 25161 }, { "epoch": 1.5619839841082626, "grad_norm": 0.5091996247924717, "learning_rate": 5.5112192148872256e-05, "loss": 2.8071, "step": 25162 }, { "epoch": 1.5620460612080205, "grad_norm": 0.5962061274880013, "learning_rate": 5.510859953803555e-05, "loss": 2.7617, "step": 25163 }, { "epoch": 1.5621081383077784, "grad_norm": 0.5796594618111355, "learning_rate": 5.510500690054603e-05, "loss": 2.8066, "step": 25164 }, { "epoch": 1.5621702154075363, "grad_norm": 0.595747173744442, "learning_rate": 5.5101414236422446e-05, "loss": 2.803, "step": 25165 }, { "epoch": 1.562232292507294, "grad_norm": 0.6094909948821289, "learning_rate": 5.509782154568356e-05, "loss": 2.8632, "step": 25166 }, { "epoch": 1.562294369607052, "grad_norm": 0.584218262240912, "learning_rate": 5.509422882834809e-05, "loss": 2.6157, "step": 25167 }, { "epoch": 1.5623564467068098, "grad_norm": 0.5739016013888005, "learning_rate": 5.50906360844348e-05, "loss": 2.7192, "step": 25168 }, { "epoch": 1.5624185238065678, "grad_norm": 0.5520967235980647, "learning_rate": 5.508704331396242e-05, "loss": 2.7866, "step": 25169 }, { "epoch": 1.5624806009063257, "grad_norm": 0.58823638076191, "learning_rate": 5.508345051694971e-05, "loss": 2.7459, "step": 25170 }, { "epoch": 1.5625426780060836, "grad_norm": 0.5511456089263249, "learning_rate": 5.507985769341538e-05, "loss": 2.731, "step": 25171 }, { "epoch": 1.5626047551058413, "grad_norm": 0.649694144291593, "learning_rate": 5.5076264843378225e-05, "loss": 2.7254, "step": 25172 }, { "epoch": 1.5626668322055992, "grad_norm": 0.5648666503822657, "learning_rate": 5.507267196685696e-05, "loss": 2.7313, "step": 25173 }, { "epoch": 1.5627289093053571, "grad_norm": 0.6229210609952969, "learning_rate": 5.506907906387032e-05, "loss": 2.8196, "step": 25174 }, { "epoch": 1.562790986405115, "grad_norm": 0.6001183403057201, "learning_rate": 5.506548613443708e-05, "loss": 2.8154, "step": 25175 }, { "epoch": 1.562853063504873, "grad_norm": 0.648063367894736, "learning_rate": 5.506189317857596e-05, "loss": 2.7915, "step": 25176 }, { "epoch": 1.5629151406046309, "grad_norm": 0.6491213490195968, "learning_rate": 5.505830019630572e-05, "loss": 2.7565, "step": 25177 }, { "epoch": 1.5629772177043888, "grad_norm": 0.6209133294003515, "learning_rate": 5.505470718764509e-05, "loss": 2.7138, "step": 25178 }, { "epoch": 1.5630392948041467, "grad_norm": 0.6328880830060829, "learning_rate": 5.505111415261284e-05, "loss": 2.7916, "step": 25179 }, { "epoch": 1.5631013719039046, "grad_norm": 0.6268172309933536, "learning_rate": 5.504752109122768e-05, "loss": 2.7974, "step": 25180 }, { "epoch": 1.5631634490036626, "grad_norm": 0.6893356482513352, "learning_rate": 5.50439280035084e-05, "loss": 2.8255, "step": 25181 }, { "epoch": 1.5632255261034205, "grad_norm": 0.5872666352012017, "learning_rate": 5.504033488947371e-05, "loss": 2.7315, "step": 25182 }, { "epoch": 1.5632876032031784, "grad_norm": 0.6858784452490522, "learning_rate": 5.503674174914236e-05, "loss": 2.794, "step": 25183 }, { "epoch": 1.5633496803029363, "grad_norm": 0.5706847817817133, "learning_rate": 5.5033148582533124e-05, "loss": 2.8212, "step": 25184 }, { "epoch": 1.5634117574026942, "grad_norm": 0.7148293248790023, "learning_rate": 5.502955538966472e-05, "loss": 2.7559, "step": 25185 }, { "epoch": 1.5634738345024521, "grad_norm": 0.6493363871765324, "learning_rate": 5.502596217055591e-05, "loss": 2.7599, "step": 25186 }, { "epoch": 1.56353591160221, "grad_norm": 0.6989680005772752, "learning_rate": 5.5022368925225435e-05, "loss": 2.7591, "step": 25187 }, { "epoch": 1.563597988701968, "grad_norm": 0.6535698794630216, "learning_rate": 5.501877565369205e-05, "loss": 2.7705, "step": 25188 }, { "epoch": 1.563660065801726, "grad_norm": 0.7101448586938335, "learning_rate": 5.501518235597448e-05, "loss": 2.7799, "step": 25189 }, { "epoch": 1.5637221429014836, "grad_norm": 0.6140345423888411, "learning_rate": 5.5011589032091493e-05, "loss": 2.8279, "step": 25190 }, { "epoch": 1.5637842200012415, "grad_norm": 0.6714428021675155, "learning_rate": 5.500799568206183e-05, "loss": 2.7069, "step": 25191 }, { "epoch": 1.5638462971009994, "grad_norm": 0.5807755513048907, "learning_rate": 5.500440230590423e-05, "loss": 2.8875, "step": 25192 }, { "epoch": 1.5639083742007573, "grad_norm": 0.6383439835123916, "learning_rate": 5.5000808903637454e-05, "loss": 2.807, "step": 25193 }, { "epoch": 1.5639704513005153, "grad_norm": 0.5637824486065898, "learning_rate": 5.499721547528025e-05, "loss": 2.7311, "step": 25194 }, { "epoch": 1.5640325284002732, "grad_norm": 0.6286255109189078, "learning_rate": 5.499362202085136e-05, "loss": 2.7922, "step": 25195 }, { "epoch": 1.5640946055000309, "grad_norm": 0.595070174350539, "learning_rate": 5.499002854036952e-05, "loss": 2.795, "step": 25196 }, { "epoch": 1.5641566825997888, "grad_norm": 0.5708845222335512, "learning_rate": 5.498643503385349e-05, "loss": 2.8088, "step": 25197 }, { "epoch": 1.5642187596995467, "grad_norm": 0.6172065731916149, "learning_rate": 5.4982841501322024e-05, "loss": 2.8867, "step": 25198 }, { "epoch": 1.5642808367993046, "grad_norm": 0.6456565281646884, "learning_rate": 5.4979247942793856e-05, "loss": 2.8738, "step": 25199 }, { "epoch": 1.5643429138990625, "grad_norm": 0.5450732113284705, "learning_rate": 5.497565435828775e-05, "loss": 2.8566, "step": 25200 }, { "epoch": 1.5644049909988205, "grad_norm": 0.6139298671918605, "learning_rate": 5.497206074782245e-05, "loss": 2.8003, "step": 25201 }, { "epoch": 1.5644670680985784, "grad_norm": 0.5168582732735703, "learning_rate": 5.49684671114167e-05, "loss": 2.7806, "step": 25202 }, { "epoch": 1.5645291451983363, "grad_norm": 0.6199492700264977, "learning_rate": 5.496487344908925e-05, "loss": 2.6814, "step": 25203 }, { "epoch": 1.5645912222980942, "grad_norm": 0.503000021155264, "learning_rate": 5.496127976085885e-05, "loss": 2.7157, "step": 25204 }, { "epoch": 1.5646532993978521, "grad_norm": 0.6277081468060463, "learning_rate": 5.495768604674425e-05, "loss": 2.7534, "step": 25205 }, { "epoch": 1.56471537649761, "grad_norm": 0.5719819651558222, "learning_rate": 5.4954092306764204e-05, "loss": 2.8063, "step": 25206 }, { "epoch": 1.564777453597368, "grad_norm": 0.5758729579567231, "learning_rate": 5.4950498540937435e-05, "loss": 2.7466, "step": 25207 }, { "epoch": 1.564839530697126, "grad_norm": 0.5894042897961151, "learning_rate": 5.4946904749282735e-05, "loss": 2.709, "step": 25208 }, { "epoch": 1.5649016077968838, "grad_norm": 0.5562647035348478, "learning_rate": 5.494331093181884e-05, "loss": 2.7759, "step": 25209 }, { "epoch": 1.5649636848966417, "grad_norm": 0.6344134248643762, "learning_rate": 5.493971708856447e-05, "loss": 2.8303, "step": 25210 }, { "epoch": 1.5650257619963996, "grad_norm": 0.5379210840935706, "learning_rate": 5.4936123219538404e-05, "loss": 2.682, "step": 25211 }, { "epoch": 1.5650878390961576, "grad_norm": 0.6753653199997004, "learning_rate": 5.493252932475937e-05, "loss": 2.8238, "step": 25212 }, { "epoch": 1.5651499161959155, "grad_norm": 0.5992439970942023, "learning_rate": 5.492893540424615e-05, "loss": 2.7597, "step": 25213 }, { "epoch": 1.5652119932956732, "grad_norm": 0.6249393195259227, "learning_rate": 5.492534145801747e-05, "loss": 2.7271, "step": 25214 }, { "epoch": 1.565274070395431, "grad_norm": 0.6030602573570912, "learning_rate": 5.492174748609209e-05, "loss": 2.7626, "step": 25215 }, { "epoch": 1.565336147495189, "grad_norm": 0.560916659039728, "learning_rate": 5.4918153488488776e-05, "loss": 2.787, "step": 25216 }, { "epoch": 1.565398224594947, "grad_norm": 0.5732908212024439, "learning_rate": 5.491455946522623e-05, "loss": 2.7907, "step": 25217 }, { "epoch": 1.5654603016947048, "grad_norm": 0.5682302636387414, "learning_rate": 5.491096541632327e-05, "loss": 2.8071, "step": 25218 }, { "epoch": 1.5655223787944628, "grad_norm": 0.5496619464427923, "learning_rate": 5.490737134179857e-05, "loss": 2.7047, "step": 25219 }, { "epoch": 1.5655844558942205, "grad_norm": 0.6258257656003887, "learning_rate": 5.490377724167096e-05, "loss": 2.8046, "step": 25220 }, { "epoch": 1.5656465329939784, "grad_norm": 0.5691318571505811, "learning_rate": 5.4900183115959136e-05, "loss": 2.7211, "step": 25221 }, { "epoch": 1.5657086100937363, "grad_norm": 0.5599549521055488, "learning_rate": 5.489658896468187e-05, "loss": 2.7342, "step": 25222 }, { "epoch": 1.5657706871934942, "grad_norm": 0.5397663221024851, "learning_rate": 5.489299478785791e-05, "loss": 2.7121, "step": 25223 }, { "epoch": 1.5658327642932521, "grad_norm": 0.5900839512306444, "learning_rate": 5.488940058550601e-05, "loss": 2.7148, "step": 25224 }, { "epoch": 1.56589484139301, "grad_norm": 0.6094746630427325, "learning_rate": 5.488580635764491e-05, "loss": 2.8128, "step": 25225 }, { "epoch": 1.565956918492768, "grad_norm": 0.5403022883406124, "learning_rate": 5.4882212104293385e-05, "loss": 2.7918, "step": 25226 }, { "epoch": 1.5660189955925259, "grad_norm": 0.6103482910068574, "learning_rate": 5.487861782547018e-05, "loss": 2.777, "step": 25227 }, { "epoch": 1.5660810726922838, "grad_norm": 0.5068683650644391, "learning_rate": 5.4875023521194025e-05, "loss": 2.7269, "step": 25228 }, { "epoch": 1.5661431497920417, "grad_norm": 0.574409297545903, "learning_rate": 5.48714291914837e-05, "loss": 2.7619, "step": 25229 }, { "epoch": 1.5662052268917996, "grad_norm": 0.5822319862294812, "learning_rate": 5.486783483635792e-05, "loss": 2.907, "step": 25230 }, { "epoch": 1.5662673039915576, "grad_norm": 0.6468050527303322, "learning_rate": 5.486424045583549e-05, "loss": 2.7803, "step": 25231 }, { "epoch": 1.5663293810913155, "grad_norm": 0.6178132277442455, "learning_rate": 5.486064604993514e-05, "loss": 2.8491, "step": 25232 }, { "epoch": 1.5663914581910734, "grad_norm": 0.613962997256573, "learning_rate": 5.4857051618675615e-05, "loss": 2.6875, "step": 25233 }, { "epoch": 1.5664535352908313, "grad_norm": 0.5709559847173227, "learning_rate": 5.485345716207567e-05, "loss": 2.7561, "step": 25234 }, { "epoch": 1.5665156123905892, "grad_norm": 0.5953140238744901, "learning_rate": 5.4849862680154054e-05, "loss": 2.7687, "step": 25235 }, { "epoch": 1.5665776894903471, "grad_norm": 0.6267713954805306, "learning_rate": 5.484626817292954e-05, "loss": 2.8492, "step": 25236 }, { "epoch": 1.566639766590105, "grad_norm": 0.6100956250500785, "learning_rate": 5.484267364042086e-05, "loss": 2.8834, "step": 25237 }, { "epoch": 1.5667018436898628, "grad_norm": 0.5835191538575447, "learning_rate": 5.4839079082646784e-05, "loss": 2.7857, "step": 25238 }, { "epoch": 1.5667639207896207, "grad_norm": 0.5627739557075988, "learning_rate": 5.4835484499626046e-05, "loss": 2.7971, "step": 25239 }, { "epoch": 1.5668259978893786, "grad_norm": 0.585348672640123, "learning_rate": 5.483188989137742e-05, "loss": 2.7399, "step": 25240 }, { "epoch": 1.5668880749891365, "grad_norm": 0.5803229332367584, "learning_rate": 5.482829525791964e-05, "loss": 2.7393, "step": 25241 }, { "epoch": 1.5669501520888944, "grad_norm": 0.5874831095763312, "learning_rate": 5.4824700599271484e-05, "loss": 2.7969, "step": 25242 }, { "epoch": 1.5670122291886524, "grad_norm": 0.5695313891919546, "learning_rate": 5.4821105915451684e-05, "loss": 2.8236, "step": 25243 }, { "epoch": 1.56707430628841, "grad_norm": 0.5842434872830269, "learning_rate": 5.4817511206479024e-05, "loss": 2.8033, "step": 25244 }, { "epoch": 1.567136383388168, "grad_norm": 0.5407494396254431, "learning_rate": 5.481391647237223e-05, "loss": 2.7518, "step": 25245 }, { "epoch": 1.5671984604879259, "grad_norm": 0.5701216548210151, "learning_rate": 5.4810321713150035e-05, "loss": 2.6999, "step": 25246 }, { "epoch": 1.5672605375876838, "grad_norm": 0.5059984229613596, "learning_rate": 5.4806726928831254e-05, "loss": 2.8098, "step": 25247 }, { "epoch": 1.5673226146874417, "grad_norm": 0.6025925277161849, "learning_rate": 5.48031321194346e-05, "loss": 2.7347, "step": 25248 }, { "epoch": 1.5673846917871996, "grad_norm": 0.5514856528113212, "learning_rate": 5.479953728497885e-05, "loss": 2.8169, "step": 25249 }, { "epoch": 1.5674467688869576, "grad_norm": 0.6114892960420202, "learning_rate": 5.479594242548274e-05, "loss": 2.7591, "step": 25250 }, { "epoch": 1.5675088459867155, "grad_norm": 0.544861547487367, "learning_rate": 5.479234754096505e-05, "loss": 2.8735, "step": 25251 }, { "epoch": 1.5675709230864734, "grad_norm": 0.5333175536394271, "learning_rate": 5.4788752631444504e-05, "loss": 2.7669, "step": 25252 }, { "epoch": 1.5676330001862313, "grad_norm": 0.568799742944273, "learning_rate": 5.478515769693988e-05, "loss": 2.8417, "step": 25253 }, { "epoch": 1.5676950772859892, "grad_norm": 0.5542503568607707, "learning_rate": 5.478156273746993e-05, "loss": 2.7621, "step": 25254 }, { "epoch": 1.5677571543857471, "grad_norm": 0.5754672898691687, "learning_rate": 5.4777967753053394e-05, "loss": 2.8083, "step": 25255 }, { "epoch": 1.567819231485505, "grad_norm": 0.5428205054466407, "learning_rate": 5.477437274370905e-05, "loss": 2.7509, "step": 25256 }, { "epoch": 1.567881308585263, "grad_norm": 0.5465808622352158, "learning_rate": 5.477077770945563e-05, "loss": 2.7915, "step": 25257 }, { "epoch": 1.567943385685021, "grad_norm": 0.5876565511591656, "learning_rate": 5.4767182650311934e-05, "loss": 2.8773, "step": 25258 }, { "epoch": 1.5680054627847788, "grad_norm": 0.6102379532937308, "learning_rate": 5.476358756629666e-05, "loss": 2.7632, "step": 25259 }, { "epoch": 1.5680675398845367, "grad_norm": 0.5414716540708021, "learning_rate": 5.475999245742861e-05, "loss": 2.7295, "step": 25260 }, { "epoch": 1.5681296169842946, "grad_norm": 0.6201597424421056, "learning_rate": 5.475639732372651e-05, "loss": 2.7707, "step": 25261 }, { "epoch": 1.5681916940840523, "grad_norm": 0.5337916608985164, "learning_rate": 5.475280216520913e-05, "loss": 2.8384, "step": 25262 }, { "epoch": 1.5682537711838103, "grad_norm": 0.6435279203877191, "learning_rate": 5.474920698189524e-05, "loss": 2.7777, "step": 25263 }, { "epoch": 1.5683158482835682, "grad_norm": 0.5543725368698776, "learning_rate": 5.474561177380357e-05, "loss": 2.7794, "step": 25264 }, { "epoch": 1.568377925383326, "grad_norm": 0.5321876441655213, "learning_rate": 5.474201654095291e-05, "loss": 2.7966, "step": 25265 }, { "epoch": 1.568440002483084, "grad_norm": 0.5902226475295955, "learning_rate": 5.4738421283361986e-05, "loss": 2.7174, "step": 25266 }, { "epoch": 1.568502079582842, "grad_norm": 0.5242529538134603, "learning_rate": 5.4734826001049576e-05, "loss": 2.7539, "step": 25267 }, { "epoch": 1.5685641566825996, "grad_norm": 0.5345265379973718, "learning_rate": 5.473123069403442e-05, "loss": 2.7529, "step": 25268 }, { "epoch": 1.5686262337823575, "grad_norm": 0.6044807494626667, "learning_rate": 5.4727635362335294e-05, "loss": 2.7814, "step": 25269 }, { "epoch": 1.5686883108821155, "grad_norm": 0.5461842397829763, "learning_rate": 5.4724040005970955e-05, "loss": 2.7545, "step": 25270 }, { "epoch": 1.5687503879818734, "grad_norm": 0.5234764710907251, "learning_rate": 5.472044462496013e-05, "loss": 2.7372, "step": 25271 }, { "epoch": 1.5688124650816313, "grad_norm": 0.5470359292983817, "learning_rate": 5.471684921932162e-05, "loss": 2.7736, "step": 25272 }, { "epoch": 1.5688745421813892, "grad_norm": 0.5488955364707425, "learning_rate": 5.471325378907414e-05, "loss": 2.7863, "step": 25273 }, { "epoch": 1.5689366192811471, "grad_norm": 0.5606617826390057, "learning_rate": 5.470965833423649e-05, "loss": 2.7318, "step": 25274 }, { "epoch": 1.568998696380905, "grad_norm": 0.5755795812757444, "learning_rate": 5.47060628548274e-05, "loss": 2.7567, "step": 25275 }, { "epoch": 1.569060773480663, "grad_norm": 0.583722365172441, "learning_rate": 5.470246735086564e-05, "loss": 2.8269, "step": 25276 }, { "epoch": 1.569122850580421, "grad_norm": 0.6048576716814188, "learning_rate": 5.4698871822369956e-05, "loss": 2.7701, "step": 25277 }, { "epoch": 1.5691849276801788, "grad_norm": 0.57800776349698, "learning_rate": 5.469527626935914e-05, "loss": 2.8781, "step": 25278 }, { "epoch": 1.5692470047799367, "grad_norm": 0.5495942222821802, "learning_rate": 5.469168069185191e-05, "loss": 2.7639, "step": 25279 }, { "epoch": 1.5693090818796946, "grad_norm": 0.6022969506039758, "learning_rate": 5.468808508986704e-05, "loss": 2.829, "step": 25280 }, { "epoch": 1.5693711589794526, "grad_norm": 0.568401924978818, "learning_rate": 5.4684489463423304e-05, "loss": 2.7962, "step": 25281 }, { "epoch": 1.5694332360792105, "grad_norm": 0.5885458555044359, "learning_rate": 5.4680893812539436e-05, "loss": 2.842, "step": 25282 }, { "epoch": 1.5694953131789684, "grad_norm": 0.6206512554335168, "learning_rate": 5.4677298137234225e-05, "loss": 2.8284, "step": 25283 }, { "epoch": 1.5695573902787263, "grad_norm": 0.558213057539753, "learning_rate": 5.4673702437526395e-05, "loss": 2.8108, "step": 25284 }, { "epoch": 1.5696194673784842, "grad_norm": 0.5797023647901053, "learning_rate": 5.4670106713434743e-05, "loss": 2.7383, "step": 25285 }, { "epoch": 1.569681544478242, "grad_norm": 0.5899960403390331, "learning_rate": 5.4666510964978e-05, "loss": 2.8309, "step": 25286 }, { "epoch": 1.5697436215779998, "grad_norm": 0.561780761139659, "learning_rate": 5.4662915192174946e-05, "loss": 2.8758, "step": 25287 }, { "epoch": 1.5698056986777578, "grad_norm": 0.6282669988655746, "learning_rate": 5.465931939504434e-05, "loss": 2.8352, "step": 25288 }, { "epoch": 1.5698677757775157, "grad_norm": 0.5201851594453585, "learning_rate": 5.4655723573604914e-05, "loss": 2.7428, "step": 25289 }, { "epoch": 1.5699298528772736, "grad_norm": 0.6212704724981646, "learning_rate": 5.4652127727875454e-05, "loss": 2.7455, "step": 25290 }, { "epoch": 1.5699919299770315, "grad_norm": 0.5710910551787747, "learning_rate": 5.4648531857874707e-05, "loss": 2.8019, "step": 25291 }, { "epoch": 1.5700540070767892, "grad_norm": 0.5841597992098382, "learning_rate": 5.464493596362146e-05, "loss": 2.8106, "step": 25292 }, { "epoch": 1.5701160841765471, "grad_norm": 0.5702399638830146, "learning_rate": 5.4641340045134426e-05, "loss": 2.7907, "step": 25293 }, { "epoch": 1.570178161276305, "grad_norm": 0.5815351535326435, "learning_rate": 5.463774410243242e-05, "loss": 2.7211, "step": 25294 }, { "epoch": 1.570240238376063, "grad_norm": 0.5742766941444292, "learning_rate": 5.463414813553416e-05, "loss": 2.8283, "step": 25295 }, { "epoch": 1.5703023154758209, "grad_norm": 0.5989482353575509, "learning_rate": 5.463055214445843e-05, "loss": 2.7687, "step": 25296 }, { "epoch": 1.5703643925755788, "grad_norm": 0.5842282860735941, "learning_rate": 5.4626956129223994e-05, "loss": 2.7372, "step": 25297 }, { "epoch": 1.5704264696753367, "grad_norm": 0.6654160561157687, "learning_rate": 5.462336008984958e-05, "loss": 2.7826, "step": 25298 }, { "epoch": 1.5704885467750946, "grad_norm": 0.605241843744129, "learning_rate": 5.4619764026354004e-05, "loss": 2.8087, "step": 25299 }, { "epoch": 1.5705506238748526, "grad_norm": 0.6443946462424218, "learning_rate": 5.461616793875598e-05, "loss": 2.8429, "step": 25300 }, { "epoch": 1.5706127009746105, "grad_norm": 0.5806744782109978, "learning_rate": 5.4612571827074297e-05, "loss": 2.6884, "step": 25301 }, { "epoch": 1.5706747780743684, "grad_norm": 0.6290912000814693, "learning_rate": 5.46089756913277e-05, "loss": 2.629, "step": 25302 }, { "epoch": 1.5707368551741263, "grad_norm": 0.5730549183120468, "learning_rate": 5.460537953153496e-05, "loss": 2.8408, "step": 25303 }, { "epoch": 1.5707989322738842, "grad_norm": 0.5546525756157146, "learning_rate": 5.460178334771484e-05, "loss": 2.7554, "step": 25304 }, { "epoch": 1.5708610093736421, "grad_norm": 0.5811874879015966, "learning_rate": 5.45981871398861e-05, "loss": 2.9026, "step": 25305 }, { "epoch": 1.5709230864734, "grad_norm": 0.5269058387301803, "learning_rate": 5.45945909080675e-05, "loss": 2.7533, "step": 25306 }, { "epoch": 1.570985163573158, "grad_norm": 0.5555166575271718, "learning_rate": 5.45909946522778e-05, "loss": 2.8262, "step": 25307 }, { "epoch": 1.571047240672916, "grad_norm": 0.5282257898075183, "learning_rate": 5.458739837253577e-05, "loss": 2.7186, "step": 25308 }, { "epoch": 1.5711093177726738, "grad_norm": 0.5539503706777864, "learning_rate": 5.458380206886017e-05, "loss": 2.8061, "step": 25309 }, { "epoch": 1.5711713948724315, "grad_norm": 0.5923580618547928, "learning_rate": 5.458020574126976e-05, "loss": 2.8498, "step": 25310 }, { "epoch": 1.5712334719721894, "grad_norm": 0.6064984825033527, "learning_rate": 5.4576609389783285e-05, "loss": 2.6633, "step": 25311 }, { "epoch": 1.5712955490719474, "grad_norm": 0.5788136839381348, "learning_rate": 5.4573013014419546e-05, "loss": 2.8756, "step": 25312 }, { "epoch": 1.5713576261717053, "grad_norm": 0.5997418185904807, "learning_rate": 5.45694166151973e-05, "loss": 2.8482, "step": 25313 }, { "epoch": 1.5714197032714632, "grad_norm": 0.5875811601947653, "learning_rate": 5.456582019213529e-05, "loss": 2.7897, "step": 25314 }, { "epoch": 1.571481780371221, "grad_norm": 0.5883327608492969, "learning_rate": 5.456222374525228e-05, "loss": 2.8123, "step": 25315 }, { "epoch": 1.5715438574709788, "grad_norm": 0.5145715587313897, "learning_rate": 5.455862727456704e-05, "loss": 2.8121, "step": 25316 }, { "epoch": 1.5716059345707367, "grad_norm": 0.6042467891795623, "learning_rate": 5.455503078009835e-05, "loss": 2.8055, "step": 25317 }, { "epoch": 1.5716680116704946, "grad_norm": 0.5046654829295263, "learning_rate": 5.455143426186494e-05, "loss": 2.8216, "step": 25318 }, { "epoch": 1.5717300887702526, "grad_norm": 0.6398785126387173, "learning_rate": 5.4547837719885605e-05, "loss": 2.8734, "step": 25319 }, { "epoch": 1.5717921658700105, "grad_norm": 0.5619652378405479, "learning_rate": 5.4544241154179096e-05, "loss": 2.7571, "step": 25320 }, { "epoch": 1.5718542429697684, "grad_norm": 0.5640426251827705, "learning_rate": 5.4540644564764176e-05, "loss": 2.8295, "step": 25321 }, { "epoch": 1.5719163200695263, "grad_norm": 0.5879597163868623, "learning_rate": 5.4537047951659614e-05, "loss": 2.8228, "step": 25322 }, { "epoch": 1.5719783971692842, "grad_norm": 0.5452520265904491, "learning_rate": 5.453345131488417e-05, "loss": 2.6928, "step": 25323 }, { "epoch": 1.5720404742690421, "grad_norm": 0.6227769894963106, "learning_rate": 5.4529854654456614e-05, "loss": 2.8489, "step": 25324 }, { "epoch": 1.5721025513688, "grad_norm": 0.5737413251133762, "learning_rate": 5.452625797039569e-05, "loss": 2.7656, "step": 25325 }, { "epoch": 1.572164628468558, "grad_norm": 0.6477830410712249, "learning_rate": 5.4522661262720197e-05, "loss": 2.8111, "step": 25326 }, { "epoch": 1.572226705568316, "grad_norm": 0.572083714446376, "learning_rate": 5.4519064531448874e-05, "loss": 2.7616, "step": 25327 }, { "epoch": 1.5722887826680738, "grad_norm": 0.6755093374771696, "learning_rate": 5.4515467776600484e-05, "loss": 2.7863, "step": 25328 }, { "epoch": 1.5723508597678317, "grad_norm": 0.5808483879522766, "learning_rate": 5.451187099819382e-05, "loss": 2.7187, "step": 25329 }, { "epoch": 1.5724129368675897, "grad_norm": 0.6498920947154503, "learning_rate": 5.4508274196247624e-05, "loss": 2.7443, "step": 25330 }, { "epoch": 1.5724750139673476, "grad_norm": 0.6229921514221849, "learning_rate": 5.450467737078067e-05, "loss": 2.7918, "step": 25331 }, { "epoch": 1.5725370910671055, "grad_norm": 0.6126721510006934, "learning_rate": 5.450108052181171e-05, "loss": 2.7884, "step": 25332 }, { "epoch": 1.5725991681668634, "grad_norm": 0.5383133312570143, "learning_rate": 5.449748364935954e-05, "loss": 2.7194, "step": 25333 }, { "epoch": 1.572661245266621, "grad_norm": 0.5851904875615795, "learning_rate": 5.449388675344288e-05, "loss": 2.855, "step": 25334 }, { "epoch": 1.572723322366379, "grad_norm": 0.550943905235442, "learning_rate": 5.4490289834080545e-05, "loss": 2.7387, "step": 25335 }, { "epoch": 1.572785399466137, "grad_norm": 0.5738527078472084, "learning_rate": 5.448669289129126e-05, "loss": 2.836, "step": 25336 }, { "epoch": 1.5728474765658949, "grad_norm": 0.5169355326105333, "learning_rate": 5.4483095925093816e-05, "loss": 2.7945, "step": 25337 }, { "epoch": 1.5729095536656528, "grad_norm": 0.5345950002826813, "learning_rate": 5.4479498935506975e-05, "loss": 2.8341, "step": 25338 }, { "epoch": 1.5729716307654107, "grad_norm": 0.5397832999494446, "learning_rate": 5.44759019225495e-05, "loss": 2.7841, "step": 25339 }, { "epoch": 1.5730337078651684, "grad_norm": 0.6202471259041696, "learning_rate": 5.447230488624015e-05, "loss": 2.8143, "step": 25340 }, { "epoch": 1.5730957849649263, "grad_norm": 0.5286816332742191, "learning_rate": 5.446870782659772e-05, "loss": 2.7172, "step": 25341 }, { "epoch": 1.5731578620646842, "grad_norm": 0.6429704436677339, "learning_rate": 5.4465110743640936e-05, "loss": 2.7905, "step": 25342 }, { "epoch": 1.5732199391644421, "grad_norm": 0.5497264908087331, "learning_rate": 5.4461513637388595e-05, "loss": 2.6706, "step": 25343 }, { "epoch": 1.5732820162642, "grad_norm": 0.6360614871316608, "learning_rate": 5.445791650785945e-05, "loss": 2.7732, "step": 25344 }, { "epoch": 1.573344093363958, "grad_norm": 0.6047627917194907, "learning_rate": 5.445431935507227e-05, "loss": 2.7337, "step": 25345 }, { "epoch": 1.573406170463716, "grad_norm": 0.5813773149088718, "learning_rate": 5.4450722179045835e-05, "loss": 2.7604, "step": 25346 }, { "epoch": 1.5734682475634738, "grad_norm": 0.6167583422133365, "learning_rate": 5.444712497979889e-05, "loss": 2.7734, "step": 25347 }, { "epoch": 1.5735303246632317, "grad_norm": 0.5730455051397945, "learning_rate": 5.444352775735022e-05, "loss": 2.7535, "step": 25348 }, { "epoch": 1.5735924017629896, "grad_norm": 0.6273916974396617, "learning_rate": 5.443993051171859e-05, "loss": 2.6788, "step": 25349 }, { "epoch": 1.5736544788627476, "grad_norm": 0.6357765247078435, "learning_rate": 5.4436333242922764e-05, "loss": 2.823, "step": 25350 }, { "epoch": 1.5737165559625055, "grad_norm": 0.5832156477016149, "learning_rate": 5.443273595098151e-05, "loss": 2.8326, "step": 25351 }, { "epoch": 1.5737786330622634, "grad_norm": 0.6380195669563856, "learning_rate": 5.442913863591359e-05, "loss": 2.9138, "step": 25352 }, { "epoch": 1.5738407101620213, "grad_norm": 0.5920246302903344, "learning_rate": 5.4425541297737795e-05, "loss": 2.7238, "step": 25353 }, { "epoch": 1.5739027872617792, "grad_norm": 0.6142750360108902, "learning_rate": 5.442194393647285e-05, "loss": 2.8702, "step": 25354 }, { "epoch": 1.5739648643615372, "grad_norm": 0.5338796997167129, "learning_rate": 5.441834655213758e-05, "loss": 2.77, "step": 25355 }, { "epoch": 1.574026941461295, "grad_norm": 0.529787490769266, "learning_rate": 5.44147491447507e-05, "loss": 2.798, "step": 25356 }, { "epoch": 1.574089018561053, "grad_norm": 0.5771726290155906, "learning_rate": 5.441115171433101e-05, "loss": 2.7999, "step": 25357 }, { "epoch": 1.5741510956608107, "grad_norm": 0.5805377872943136, "learning_rate": 5.440755426089727e-05, "loss": 2.8305, "step": 25358 }, { "epoch": 1.5742131727605686, "grad_norm": 0.5843918313443145, "learning_rate": 5.440395678446826e-05, "loss": 2.7603, "step": 25359 }, { "epoch": 1.5742752498603265, "grad_norm": 0.6170597668547343, "learning_rate": 5.440035928506272e-05, "loss": 2.8479, "step": 25360 }, { "epoch": 1.5743373269600844, "grad_norm": 0.6038348147050946, "learning_rate": 5.439676176269945e-05, "loss": 2.6323, "step": 25361 }, { "epoch": 1.5743994040598424, "grad_norm": 0.6715999750089612, "learning_rate": 5.4393164217397194e-05, "loss": 2.9243, "step": 25362 }, { "epoch": 1.5744614811596003, "grad_norm": 0.5943409839607682, "learning_rate": 5.438956664917474e-05, "loss": 2.7252, "step": 25363 }, { "epoch": 1.574523558259358, "grad_norm": 0.620827791863444, "learning_rate": 5.4385969058050853e-05, "loss": 2.7201, "step": 25364 }, { "epoch": 1.5745856353591159, "grad_norm": 0.6054549923225903, "learning_rate": 5.43823714440443e-05, "loss": 2.7958, "step": 25365 }, { "epoch": 1.5746477124588738, "grad_norm": 0.6198047962454286, "learning_rate": 5.437877380717387e-05, "loss": 2.7495, "step": 25366 }, { "epoch": 1.5747097895586317, "grad_norm": 0.5455605377273852, "learning_rate": 5.437517614745829e-05, "loss": 2.7748, "step": 25367 }, { "epoch": 1.5747718666583896, "grad_norm": 0.6182420871180422, "learning_rate": 5.437157846491636e-05, "loss": 2.7763, "step": 25368 }, { "epoch": 1.5748339437581476, "grad_norm": 0.605068271229191, "learning_rate": 5.436798075956685e-05, "loss": 2.8554, "step": 25369 }, { "epoch": 1.5748960208579055, "grad_norm": 0.6120688244229858, "learning_rate": 5.436438303142851e-05, "loss": 2.8196, "step": 25370 }, { "epoch": 1.5749580979576634, "grad_norm": 0.5667231758308762, "learning_rate": 5.436078528052014e-05, "loss": 2.7729, "step": 25371 }, { "epoch": 1.5750201750574213, "grad_norm": 0.6648797335899947, "learning_rate": 5.4357187506860483e-05, "loss": 2.8118, "step": 25372 }, { "epoch": 1.5750822521571792, "grad_norm": 0.5979307205152394, "learning_rate": 5.4353589710468337e-05, "loss": 2.8465, "step": 25373 }, { "epoch": 1.5751443292569371, "grad_norm": 0.5823861390682219, "learning_rate": 5.434999189136244e-05, "loss": 2.7918, "step": 25374 }, { "epoch": 1.575206406356695, "grad_norm": 0.5863488851248271, "learning_rate": 5.434639404956159e-05, "loss": 2.8441, "step": 25375 }, { "epoch": 1.575268483456453, "grad_norm": 0.6163034989229974, "learning_rate": 5.434279618508453e-05, "loss": 2.821, "step": 25376 }, { "epoch": 1.575330560556211, "grad_norm": 0.5978313762917066, "learning_rate": 5.433919829795006e-05, "loss": 2.7391, "step": 25377 }, { "epoch": 1.5753926376559688, "grad_norm": 0.5882483119786366, "learning_rate": 5.4335600388176946e-05, "loss": 2.8028, "step": 25378 }, { "epoch": 1.5754547147557267, "grad_norm": 0.5508411997100195, "learning_rate": 5.4332002455783936e-05, "loss": 2.7395, "step": 25379 }, { "epoch": 1.5755167918554847, "grad_norm": 0.5924096035708223, "learning_rate": 5.432840450078983e-05, "loss": 2.8432, "step": 25380 }, { "epoch": 1.5755788689552426, "grad_norm": 0.5443059249832982, "learning_rate": 5.432480652321338e-05, "loss": 2.8017, "step": 25381 }, { "epoch": 1.5756409460550003, "grad_norm": 0.5612676684820609, "learning_rate": 5.432120852307337e-05, "loss": 2.8139, "step": 25382 }, { "epoch": 1.5757030231547582, "grad_norm": 0.5395957134058889, "learning_rate": 5.431761050038856e-05, "loss": 2.779, "step": 25383 }, { "epoch": 1.575765100254516, "grad_norm": 0.5363123303092935, "learning_rate": 5.431401245517774e-05, "loss": 2.7519, "step": 25384 }, { "epoch": 1.575827177354274, "grad_norm": 0.5209892293947983, "learning_rate": 5.431041438745965e-05, "loss": 2.7881, "step": 25385 }, { "epoch": 1.575889254454032, "grad_norm": 0.5179197227784776, "learning_rate": 5.4306816297253104e-05, "loss": 2.7586, "step": 25386 }, { "epoch": 1.5759513315537899, "grad_norm": 0.4888577834111441, "learning_rate": 5.430321818457684e-05, "loss": 2.782, "step": 25387 }, { "epoch": 1.5760134086535476, "grad_norm": 0.5125893910301811, "learning_rate": 5.4299620049449654e-05, "loss": 2.7139, "step": 25388 }, { "epoch": 1.5760754857533055, "grad_norm": 0.4945462337592305, "learning_rate": 5.4296021891890295e-05, "loss": 2.7752, "step": 25389 }, { "epoch": 1.5761375628530634, "grad_norm": 0.4969110739706317, "learning_rate": 5.4292423711917536e-05, "loss": 2.7927, "step": 25390 }, { "epoch": 1.5761996399528213, "grad_norm": 0.4922266899128702, "learning_rate": 5.4288825509550175e-05, "loss": 2.6898, "step": 25391 }, { "epoch": 1.5762617170525792, "grad_norm": 0.5433568870767201, "learning_rate": 5.4285227284806975e-05, "loss": 2.7593, "step": 25392 }, { "epoch": 1.5763237941523371, "grad_norm": 0.5397389631799409, "learning_rate": 5.428162903770668e-05, "loss": 2.844, "step": 25393 }, { "epoch": 1.576385871252095, "grad_norm": 0.49522452911973885, "learning_rate": 5.427803076826812e-05, "loss": 2.7691, "step": 25394 }, { "epoch": 1.576447948351853, "grad_norm": 0.5707018842429081, "learning_rate": 5.427443247651e-05, "loss": 2.8269, "step": 25395 }, { "epoch": 1.576510025451611, "grad_norm": 0.5045119496963361, "learning_rate": 5.4270834162451157e-05, "loss": 2.7513, "step": 25396 }, { "epoch": 1.5765721025513688, "grad_norm": 0.6112800311100836, "learning_rate": 5.4267235826110315e-05, "loss": 2.7666, "step": 25397 }, { "epoch": 1.5766341796511267, "grad_norm": 0.5516502883252756, "learning_rate": 5.426363746750629e-05, "loss": 2.7891, "step": 25398 }, { "epoch": 1.5766962567508847, "grad_norm": 0.5196271535366748, "learning_rate": 5.426003908665782e-05, "loss": 2.7311, "step": 25399 }, { "epoch": 1.5767583338506426, "grad_norm": 0.5979685248062272, "learning_rate": 5.42564406835837e-05, "loss": 2.821, "step": 25400 }, { "epoch": 1.5768204109504005, "grad_norm": 0.4860120458552365, "learning_rate": 5.425284225830268e-05, "loss": 2.776, "step": 25401 }, { "epoch": 1.5768824880501584, "grad_norm": 0.6297602098007599, "learning_rate": 5.4249243810833564e-05, "loss": 2.8769, "step": 25402 }, { "epoch": 1.5769445651499163, "grad_norm": 0.5528422004480044, "learning_rate": 5.4245645341195094e-05, "loss": 2.7722, "step": 25403 }, { "epoch": 1.5770066422496742, "grad_norm": 0.5353802956820868, "learning_rate": 5.424204684940608e-05, "loss": 2.7264, "step": 25404 }, { "epoch": 1.5770687193494322, "grad_norm": 0.5798491722928635, "learning_rate": 5.423844833548528e-05, "loss": 2.7588, "step": 25405 }, { "epoch": 1.5771307964491899, "grad_norm": 0.567691922315159, "learning_rate": 5.4234849799451446e-05, "loss": 2.7955, "step": 25406 }, { "epoch": 1.5771928735489478, "grad_norm": 0.5750358632555006, "learning_rate": 5.42312512413234e-05, "loss": 2.725, "step": 25407 }, { "epoch": 1.5772549506487057, "grad_norm": 0.5836695967772634, "learning_rate": 5.4227652661119865e-05, "loss": 2.8565, "step": 25408 }, { "epoch": 1.5773170277484636, "grad_norm": 0.564718771334102, "learning_rate": 5.422405405885965e-05, "loss": 2.7849, "step": 25409 }, { "epoch": 1.5773791048482215, "grad_norm": 0.6047216666965779, "learning_rate": 5.42204554345615e-05, "loss": 2.7883, "step": 25410 }, { "epoch": 1.5774411819479794, "grad_norm": 0.5930383834864127, "learning_rate": 5.421685678824423e-05, "loss": 2.8841, "step": 25411 }, { "epoch": 1.5775032590477371, "grad_norm": 0.6362874026818143, "learning_rate": 5.42132581199266e-05, "loss": 2.8332, "step": 25412 }, { "epoch": 1.577565336147495, "grad_norm": 0.50338475647824, "learning_rate": 5.420965942962737e-05, "loss": 2.7608, "step": 25413 }, { "epoch": 1.577627413247253, "grad_norm": 0.6154207704140792, "learning_rate": 5.420606071736533e-05, "loss": 2.8595, "step": 25414 }, { "epoch": 1.577689490347011, "grad_norm": 0.5450062840213188, "learning_rate": 5.4202461983159234e-05, "loss": 2.8881, "step": 25415 }, { "epoch": 1.5777515674467688, "grad_norm": 0.5969398043938204, "learning_rate": 5.4198863227027895e-05, "loss": 2.797, "step": 25416 }, { "epoch": 1.5778136445465267, "grad_norm": 0.5311337792669073, "learning_rate": 5.419526444899005e-05, "loss": 2.7012, "step": 25417 }, { "epoch": 1.5778757216462846, "grad_norm": 0.5618976971175998, "learning_rate": 5.4191665649064504e-05, "loss": 2.7371, "step": 25418 }, { "epoch": 1.5779377987460426, "grad_norm": 0.5288139476839715, "learning_rate": 5.418806682727001e-05, "loss": 2.7739, "step": 25419 }, { "epoch": 1.5779998758458005, "grad_norm": 0.6436006937673444, "learning_rate": 5.4184467983625364e-05, "loss": 2.8029, "step": 25420 }, { "epoch": 1.5780619529455584, "grad_norm": 0.6054345890375676, "learning_rate": 5.418086911814932e-05, "loss": 2.8469, "step": 25421 }, { "epoch": 1.5781240300453163, "grad_norm": 0.6290599800965216, "learning_rate": 5.417727023086069e-05, "loss": 2.7785, "step": 25422 }, { "epoch": 1.5781861071450742, "grad_norm": 0.6190456714682456, "learning_rate": 5.417367132177822e-05, "loss": 2.7435, "step": 25423 }, { "epoch": 1.5782481842448322, "grad_norm": 0.6535705406523904, "learning_rate": 5.4170072390920676e-05, "loss": 2.7086, "step": 25424 }, { "epoch": 1.57831026134459, "grad_norm": 0.6101947308273772, "learning_rate": 5.416647343830687e-05, "loss": 2.8001, "step": 25425 }, { "epoch": 1.578372338444348, "grad_norm": 0.6911491088291294, "learning_rate": 5.416287446395554e-05, "loss": 2.8178, "step": 25426 }, { "epoch": 1.578434415544106, "grad_norm": 0.6671619338915098, "learning_rate": 5.415927546788551e-05, "loss": 2.8488, "step": 25427 }, { "epoch": 1.5784964926438638, "grad_norm": 0.6750715862390577, "learning_rate": 5.41556764501155e-05, "loss": 2.8784, "step": 25428 }, { "epoch": 1.5785585697436217, "grad_norm": 0.6161060999755467, "learning_rate": 5.415207741066435e-05, "loss": 2.7742, "step": 25429 }, { "epoch": 1.5786206468433794, "grad_norm": 0.6085671670027004, "learning_rate": 5.414847834955079e-05, "loss": 2.7312, "step": 25430 }, { "epoch": 1.5786827239431374, "grad_norm": 0.585104413658698, "learning_rate": 5.41448792667936e-05, "loss": 2.7536, "step": 25431 }, { "epoch": 1.5787448010428953, "grad_norm": 0.5810579187639934, "learning_rate": 5.414128016241159e-05, "loss": 2.7993, "step": 25432 }, { "epoch": 1.5788068781426532, "grad_norm": 0.6001717480577939, "learning_rate": 5.413768103642349e-05, "loss": 2.7148, "step": 25433 }, { "epoch": 1.578868955242411, "grad_norm": 0.592001977007052, "learning_rate": 5.4134081888848133e-05, "loss": 2.711, "step": 25434 }, { "epoch": 1.578931032342169, "grad_norm": 0.5859653974885457, "learning_rate": 5.413048271970425e-05, "loss": 2.7881, "step": 25435 }, { "epoch": 1.5789931094419267, "grad_norm": 0.5892309074594191, "learning_rate": 5.4126883529010645e-05, "loss": 2.8283, "step": 25436 }, { "epoch": 1.5790551865416846, "grad_norm": 0.5972272043488553, "learning_rate": 5.4123284316786074e-05, "loss": 2.8483, "step": 25437 }, { "epoch": 1.5791172636414426, "grad_norm": 0.6149522649258525, "learning_rate": 5.411968508304933e-05, "loss": 2.7858, "step": 25438 }, { "epoch": 1.5791793407412005, "grad_norm": 0.589206148840822, "learning_rate": 5.4116085827819185e-05, "loss": 2.7172, "step": 25439 }, { "epoch": 1.5792414178409584, "grad_norm": 0.5814594045410713, "learning_rate": 5.411248655111444e-05, "loss": 2.8156, "step": 25440 }, { "epoch": 1.5793034949407163, "grad_norm": 0.6166624419179315, "learning_rate": 5.4108887252953844e-05, "loss": 2.6926, "step": 25441 }, { "epoch": 1.5793655720404742, "grad_norm": 0.5372105173768624, "learning_rate": 5.410528793335617e-05, "loss": 2.7542, "step": 25442 }, { "epoch": 1.5794276491402321, "grad_norm": 0.6231176880203736, "learning_rate": 5.410168859234024e-05, "loss": 2.6914, "step": 25443 }, { "epoch": 1.57948972623999, "grad_norm": 0.5579208642755629, "learning_rate": 5.409808922992479e-05, "loss": 2.7875, "step": 25444 }, { "epoch": 1.579551803339748, "grad_norm": 0.6496120065164397, "learning_rate": 5.409448984612861e-05, "loss": 2.6922, "step": 25445 }, { "epoch": 1.579613880439506, "grad_norm": 0.5268188690115537, "learning_rate": 5.409089044097049e-05, "loss": 2.8464, "step": 25446 }, { "epoch": 1.5796759575392638, "grad_norm": 0.5936032415014251, "learning_rate": 5.40872910144692e-05, "loss": 2.7422, "step": 25447 }, { "epoch": 1.5797380346390217, "grad_norm": 0.607114855336946, "learning_rate": 5.408369156664352e-05, "loss": 2.7744, "step": 25448 }, { "epoch": 1.5798001117387797, "grad_norm": 0.5972301984101573, "learning_rate": 5.4080092097512224e-05, "loss": 2.7333, "step": 25449 }, { "epoch": 1.5798621888385376, "grad_norm": 0.5865439264997768, "learning_rate": 5.407649260709411e-05, "loss": 2.8281, "step": 25450 }, { "epoch": 1.5799242659382955, "grad_norm": 0.5835639179985089, "learning_rate": 5.407289309540794e-05, "loss": 2.8538, "step": 25451 }, { "epoch": 1.5799863430380534, "grad_norm": 0.5747961398184882, "learning_rate": 5.40692935624725e-05, "loss": 2.7975, "step": 25452 }, { "epoch": 1.5800484201378113, "grad_norm": 0.5859650923272142, "learning_rate": 5.406569400830656e-05, "loss": 2.7522, "step": 25453 }, { "epoch": 1.580110497237569, "grad_norm": 0.6483212791987746, "learning_rate": 5.406209443292891e-05, "loss": 2.8335, "step": 25454 }, { "epoch": 1.580172574337327, "grad_norm": 0.5198854945270761, "learning_rate": 5.405849483635832e-05, "loss": 2.7877, "step": 25455 }, { "epoch": 1.5802346514370849, "grad_norm": 0.636431998058169, "learning_rate": 5.40548952186136e-05, "loss": 2.7523, "step": 25456 }, { "epoch": 1.5802967285368428, "grad_norm": 0.5814368009344707, "learning_rate": 5.405129557971348e-05, "loss": 2.807, "step": 25457 }, { "epoch": 1.5803588056366007, "grad_norm": 0.5965172629292554, "learning_rate": 5.404769591967679e-05, "loss": 2.7797, "step": 25458 }, { "epoch": 1.5804208827363586, "grad_norm": 0.5961925309754056, "learning_rate": 5.404409623852227e-05, "loss": 2.7423, "step": 25459 }, { "epoch": 1.5804829598361163, "grad_norm": 0.551348995566705, "learning_rate": 5.404049653626872e-05, "loss": 2.7364, "step": 25460 }, { "epoch": 1.5805450369358742, "grad_norm": 0.5783543003121925, "learning_rate": 5.4036896812934924e-05, "loss": 2.5824, "step": 25461 }, { "epoch": 1.5806071140356321, "grad_norm": 0.5972077924872814, "learning_rate": 5.403329706853966e-05, "loss": 2.6656, "step": 25462 }, { "epoch": 1.58066919113539, "grad_norm": 0.5927596114178607, "learning_rate": 5.40296973031017e-05, "loss": 2.835, "step": 25463 }, { "epoch": 1.580731268235148, "grad_norm": 0.5980185364350905, "learning_rate": 5.402609751663983e-05, "loss": 2.7848, "step": 25464 }, { "epoch": 1.580793345334906, "grad_norm": 0.598778363155336, "learning_rate": 5.4022497709172845e-05, "loss": 2.7813, "step": 25465 }, { "epoch": 1.5808554224346638, "grad_norm": 0.5614039816466831, "learning_rate": 5.401889788071951e-05, "loss": 2.8187, "step": 25466 }, { "epoch": 1.5809174995344217, "grad_norm": 0.5887532796181209, "learning_rate": 5.40152980312986e-05, "loss": 2.7862, "step": 25467 }, { "epoch": 1.5809795766341797, "grad_norm": 0.6041075633725602, "learning_rate": 5.401169816092891e-05, "loss": 2.8065, "step": 25468 }, { "epoch": 1.5810416537339376, "grad_norm": 0.5606269912088983, "learning_rate": 5.400809826962921e-05, "loss": 2.8194, "step": 25469 }, { "epoch": 1.5811037308336955, "grad_norm": 0.6196879185200489, "learning_rate": 5.40044983574183e-05, "loss": 2.8459, "step": 25470 }, { "epoch": 1.5811658079334534, "grad_norm": 0.6109228039703257, "learning_rate": 5.4000898424314935e-05, "loss": 2.7391, "step": 25471 }, { "epoch": 1.5812278850332113, "grad_norm": 0.6110349201862207, "learning_rate": 5.3997298470337934e-05, "loss": 2.7749, "step": 25472 }, { "epoch": 1.5812899621329692, "grad_norm": 0.5338501968856467, "learning_rate": 5.399369849550604e-05, "loss": 2.7708, "step": 25473 }, { "epoch": 1.5813520392327272, "grad_norm": 0.6370176772055762, "learning_rate": 5.399009849983806e-05, "loss": 2.833, "step": 25474 }, { "epoch": 1.581414116332485, "grad_norm": 0.5938897219442804, "learning_rate": 5.3986498483352755e-05, "loss": 2.7857, "step": 25475 }, { "epoch": 1.581476193432243, "grad_norm": 0.5715314823533739, "learning_rate": 5.398289844606892e-05, "loss": 2.7918, "step": 25476 }, { "epoch": 1.5815382705320007, "grad_norm": 0.5394710430067142, "learning_rate": 5.397929838800535e-05, "loss": 2.7659, "step": 25477 }, { "epoch": 1.5816003476317586, "grad_norm": 0.5933082593707831, "learning_rate": 5.39756983091808e-05, "loss": 2.7534, "step": 25478 }, { "epoch": 1.5816624247315165, "grad_norm": 0.5788409722676566, "learning_rate": 5.3972098209614084e-05, "loss": 2.8035, "step": 25479 }, { "epoch": 1.5817245018312744, "grad_norm": 0.6185879090436387, "learning_rate": 5.396849808932395e-05, "loss": 2.6636, "step": 25480 }, { "epoch": 1.5817865789310324, "grad_norm": 0.49895119937370686, "learning_rate": 5.396489794832922e-05, "loss": 2.7518, "step": 25481 }, { "epoch": 1.5818486560307903, "grad_norm": 0.6687045144272292, "learning_rate": 5.396129778664863e-05, "loss": 2.7222, "step": 25482 }, { "epoch": 1.581910733130548, "grad_norm": 0.5283752807677808, "learning_rate": 5.3957697604301004e-05, "loss": 2.8317, "step": 25483 }, { "epoch": 1.581972810230306, "grad_norm": 0.7259407664770302, "learning_rate": 5.395409740130511e-05, "loss": 2.7338, "step": 25484 }, { "epoch": 1.5820348873300638, "grad_norm": 0.5063332359739959, "learning_rate": 5.395049717767971e-05, "loss": 2.7469, "step": 25485 }, { "epoch": 1.5820969644298217, "grad_norm": 0.6651872422191748, "learning_rate": 5.3946896933443636e-05, "loss": 2.6902, "step": 25486 }, { "epoch": 1.5821590415295796, "grad_norm": 0.5637071160966811, "learning_rate": 5.394329666861562e-05, "loss": 2.8175, "step": 25487 }, { "epoch": 1.5822211186293376, "grad_norm": 0.6274809234880616, "learning_rate": 5.393969638321448e-05, "loss": 2.8323, "step": 25488 }, { "epoch": 1.5822831957290955, "grad_norm": 0.5713941428550252, "learning_rate": 5.393609607725898e-05, "loss": 2.6667, "step": 25489 }, { "epoch": 1.5823452728288534, "grad_norm": 0.5739263971055085, "learning_rate": 5.393249575076792e-05, "loss": 2.7427, "step": 25490 }, { "epoch": 1.5824073499286113, "grad_norm": 0.5740535783982949, "learning_rate": 5.392889540376006e-05, "loss": 2.7602, "step": 25491 }, { "epoch": 1.5824694270283692, "grad_norm": 0.570286763276773, "learning_rate": 5.3925295036254206e-05, "loss": 2.8452, "step": 25492 }, { "epoch": 1.5825315041281272, "grad_norm": 0.6704537862985013, "learning_rate": 5.392169464826914e-05, "loss": 2.7073, "step": 25493 }, { "epoch": 1.582593581227885, "grad_norm": 0.5827307114471271, "learning_rate": 5.3918094239823635e-05, "loss": 2.6894, "step": 25494 }, { "epoch": 1.582655658327643, "grad_norm": 0.5890486202071967, "learning_rate": 5.39144938109365e-05, "loss": 2.7239, "step": 25495 }, { "epoch": 1.582717735427401, "grad_norm": 0.6232195528200533, "learning_rate": 5.3910893361626466e-05, "loss": 2.7081, "step": 25496 }, { "epoch": 1.5827798125271588, "grad_norm": 0.5816326184358972, "learning_rate": 5.390729289191239e-05, "loss": 2.7913, "step": 25497 }, { "epoch": 1.5828418896269167, "grad_norm": 0.6012487955933877, "learning_rate": 5.390369240181299e-05, "loss": 2.8316, "step": 25498 }, { "epoch": 1.5829039667266747, "grad_norm": 0.5693406485989154, "learning_rate": 5.390009189134709e-05, "loss": 2.8484, "step": 25499 }, { "epoch": 1.5829660438264326, "grad_norm": 0.5419192428302769, "learning_rate": 5.389649136053345e-05, "loss": 2.6213, "step": 25500 }, { "epoch": 1.5830281209261903, "grad_norm": 0.5738215467518699, "learning_rate": 5.38928908093909e-05, "loss": 2.8022, "step": 25501 }, { "epoch": 1.5830901980259482, "grad_norm": 0.5814065845368593, "learning_rate": 5.3889290237938174e-05, "loss": 2.7643, "step": 25502 }, { "epoch": 1.583152275125706, "grad_norm": 0.5829182976224313, "learning_rate": 5.388568964619407e-05, "loss": 2.7972, "step": 25503 }, { "epoch": 1.583214352225464, "grad_norm": 0.5810749467154894, "learning_rate": 5.3882089034177393e-05, "loss": 2.8283, "step": 25504 }, { "epoch": 1.583276429325222, "grad_norm": 0.6259810377606699, "learning_rate": 5.3878488401906904e-05, "loss": 2.8577, "step": 25505 }, { "epoch": 1.5833385064249799, "grad_norm": 0.590076021857133, "learning_rate": 5.3874887749401415e-05, "loss": 2.7709, "step": 25506 }, { "epoch": 1.5834005835247376, "grad_norm": 0.55880647890162, "learning_rate": 5.3871287076679675e-05, "loss": 2.768, "step": 25507 }, { "epoch": 1.5834626606244955, "grad_norm": 0.6159906492328612, "learning_rate": 5.3867686383760495e-05, "loss": 2.7762, "step": 25508 }, { "epoch": 1.5835247377242534, "grad_norm": 0.6030672747406869, "learning_rate": 5.386408567066267e-05, "loss": 2.7446, "step": 25509 }, { "epoch": 1.5835868148240113, "grad_norm": 0.558076632498153, "learning_rate": 5.386048493740495e-05, "loss": 2.7712, "step": 25510 }, { "epoch": 1.5836488919237692, "grad_norm": 0.6410147016612765, "learning_rate": 5.385688418400617e-05, "loss": 2.7761, "step": 25511 }, { "epoch": 1.5837109690235271, "grad_norm": 0.5483408492258285, "learning_rate": 5.385328341048507e-05, "loss": 2.8203, "step": 25512 }, { "epoch": 1.583773046123285, "grad_norm": 0.6677749084831341, "learning_rate": 5.384968261686046e-05, "loss": 2.7562, "step": 25513 }, { "epoch": 1.583835123223043, "grad_norm": 0.5405279502417273, "learning_rate": 5.3846081803151125e-05, "loss": 2.7442, "step": 25514 }, { "epoch": 1.583897200322801, "grad_norm": 0.6864479859131329, "learning_rate": 5.3842480969375855e-05, "loss": 2.7772, "step": 25515 }, { "epoch": 1.5839592774225588, "grad_norm": 0.5370193683463055, "learning_rate": 5.3838880115553416e-05, "loss": 2.9137, "step": 25516 }, { "epoch": 1.5840213545223167, "grad_norm": 0.564665731233177, "learning_rate": 5.3835279241702605e-05, "loss": 2.7656, "step": 25517 }, { "epoch": 1.5840834316220747, "grad_norm": 0.5176829106045581, "learning_rate": 5.383167834784222e-05, "loss": 2.8833, "step": 25518 }, { "epoch": 1.5841455087218326, "grad_norm": 0.5401449254939645, "learning_rate": 5.382807743399103e-05, "loss": 2.6969, "step": 25519 }, { "epoch": 1.5842075858215905, "grad_norm": 0.5429640296346636, "learning_rate": 5.382447650016785e-05, "loss": 2.7654, "step": 25520 }, { "epoch": 1.5842696629213484, "grad_norm": 0.5436708864120001, "learning_rate": 5.382087554639143e-05, "loss": 2.8254, "step": 25521 }, { "epoch": 1.5843317400211063, "grad_norm": 0.563452946582875, "learning_rate": 5.3817274572680574e-05, "loss": 2.7883, "step": 25522 }, { "epoch": 1.5843938171208642, "grad_norm": 0.6090765491194233, "learning_rate": 5.381367357905408e-05, "loss": 2.7104, "step": 25523 }, { "epoch": 1.5844558942206222, "grad_norm": 0.5700625825707705, "learning_rate": 5.3810072565530714e-05, "loss": 2.6745, "step": 25524 }, { "epoch": 1.5845179713203799, "grad_norm": 0.6806502265563825, "learning_rate": 5.3806471532129274e-05, "loss": 2.7647, "step": 25525 }, { "epoch": 1.5845800484201378, "grad_norm": 0.6440297343041761, "learning_rate": 5.380287047886856e-05, "loss": 2.7296, "step": 25526 }, { "epoch": 1.5846421255198957, "grad_norm": 0.664509297574961, "learning_rate": 5.379926940576734e-05, "loss": 2.7638, "step": 25527 }, { "epoch": 1.5847042026196536, "grad_norm": 0.6720749091889423, "learning_rate": 5.3795668312844405e-05, "loss": 2.8006, "step": 25528 }, { "epoch": 1.5847662797194115, "grad_norm": 0.7234576344005434, "learning_rate": 5.379206720011857e-05, "loss": 2.7836, "step": 25529 }, { "epoch": 1.5848283568191694, "grad_norm": 0.6759634709406247, "learning_rate": 5.3788466067608566e-05, "loss": 2.8001, "step": 25530 }, { "epoch": 1.5848904339189271, "grad_norm": 0.7556142089047969, "learning_rate": 5.378486491533324e-05, "loss": 2.8043, "step": 25531 }, { "epoch": 1.584952511018685, "grad_norm": 0.6745049436103855, "learning_rate": 5.3781263743311335e-05, "loss": 2.7174, "step": 25532 }, { "epoch": 1.585014588118443, "grad_norm": 0.7369404884855602, "learning_rate": 5.377766255156168e-05, "loss": 2.8814, "step": 25533 }, { "epoch": 1.585076665218201, "grad_norm": 0.6306185661343098, "learning_rate": 5.377406134010303e-05, "loss": 2.8111, "step": 25534 }, { "epoch": 1.5851387423179588, "grad_norm": 0.7447916222063242, "learning_rate": 5.3770460108954194e-05, "loss": 2.8282, "step": 25535 }, { "epoch": 1.5852008194177167, "grad_norm": 0.641893503317974, "learning_rate": 5.376685885813394e-05, "loss": 2.7178, "step": 25536 }, { "epoch": 1.5852628965174747, "grad_norm": 0.7330069684481926, "learning_rate": 5.376325758766109e-05, "loss": 2.7421, "step": 25537 }, { "epoch": 1.5853249736172326, "grad_norm": 0.6259173098891685, "learning_rate": 5.3759656297554386e-05, "loss": 2.7561, "step": 25538 }, { "epoch": 1.5853870507169905, "grad_norm": 0.6553742378124204, "learning_rate": 5.375605498783266e-05, "loss": 2.7334, "step": 25539 }, { "epoch": 1.5854491278167484, "grad_norm": 0.5916824985678389, "learning_rate": 5.375245365851468e-05, "loss": 2.7086, "step": 25540 }, { "epoch": 1.5855112049165063, "grad_norm": 0.5677630255249356, "learning_rate": 5.3748852309619224e-05, "loss": 2.7874, "step": 25541 }, { "epoch": 1.5855732820162642, "grad_norm": 0.6507999861677364, "learning_rate": 5.374525094116511e-05, "loss": 2.7564, "step": 25542 }, { "epoch": 1.5856353591160222, "grad_norm": 0.5349183214976885, "learning_rate": 5.3741649553171103e-05, "loss": 2.7687, "step": 25543 }, { "epoch": 1.58569743621578, "grad_norm": 0.6659812359028073, "learning_rate": 5.373804814565602e-05, "loss": 2.7425, "step": 25544 }, { "epoch": 1.585759513315538, "grad_norm": 0.5408770387160173, "learning_rate": 5.373444671863862e-05, "loss": 2.7438, "step": 25545 }, { "epoch": 1.585821590415296, "grad_norm": 0.6023801473463729, "learning_rate": 5.373084527213771e-05, "loss": 2.8107, "step": 25546 }, { "epoch": 1.5858836675150538, "grad_norm": 0.5550462073344727, "learning_rate": 5.3727243806172065e-05, "loss": 2.8043, "step": 25547 }, { "epoch": 1.5859457446148117, "grad_norm": 0.6499631926312136, "learning_rate": 5.372364232076049e-05, "loss": 2.7666, "step": 25548 }, { "epoch": 1.5860078217145694, "grad_norm": 0.6017773727511576, "learning_rate": 5.372004081592177e-05, "loss": 2.8329, "step": 25549 }, { "epoch": 1.5860698988143274, "grad_norm": 0.6274899950027751, "learning_rate": 5.371643929167469e-05, "loss": 2.7516, "step": 25550 }, { "epoch": 1.5861319759140853, "grad_norm": 0.6219882881067715, "learning_rate": 5.371283774803806e-05, "loss": 2.8204, "step": 25551 }, { "epoch": 1.5861940530138432, "grad_norm": 0.5765286133274923, "learning_rate": 5.370923618503063e-05, "loss": 2.7479, "step": 25552 }, { "epoch": 1.5862561301136011, "grad_norm": 0.5186134491868731, "learning_rate": 5.370563460267123e-05, "loss": 2.6879, "step": 25553 }, { "epoch": 1.586318207213359, "grad_norm": 0.6146877464134913, "learning_rate": 5.370203300097863e-05, "loss": 2.8119, "step": 25554 }, { "epoch": 1.5863802843131167, "grad_norm": 0.6081504745087626, "learning_rate": 5.369843137997163e-05, "loss": 2.763, "step": 25555 }, { "epoch": 1.5864423614128746, "grad_norm": 0.6386411700319307, "learning_rate": 5.3694829739669006e-05, "loss": 2.8356, "step": 25556 }, { "epoch": 1.5865044385126326, "grad_norm": 0.5540765433298703, "learning_rate": 5.369122808008955e-05, "loss": 2.7287, "step": 25557 }, { "epoch": 1.5865665156123905, "grad_norm": 0.6507941474872965, "learning_rate": 5.368762640125208e-05, "loss": 2.7392, "step": 25558 }, { "epoch": 1.5866285927121484, "grad_norm": 0.5993223714943422, "learning_rate": 5.368402470317536e-05, "loss": 2.8048, "step": 25559 }, { "epoch": 1.5866906698119063, "grad_norm": 0.6413705255965333, "learning_rate": 5.36804229858782e-05, "loss": 2.7947, "step": 25560 }, { "epoch": 1.5867527469116642, "grad_norm": 0.5758914329663279, "learning_rate": 5.367682124937936e-05, "loss": 2.7874, "step": 25561 }, { "epoch": 1.5868148240114222, "grad_norm": 0.654681442888231, "learning_rate": 5.367321949369767e-05, "loss": 2.8138, "step": 25562 }, { "epoch": 1.58687690111118, "grad_norm": 0.5950541478850918, "learning_rate": 5.3669617718851883e-05, "loss": 2.8269, "step": 25563 }, { "epoch": 1.586938978210938, "grad_norm": 0.6034761187514551, "learning_rate": 5.366601592486083e-05, "loss": 2.7876, "step": 25564 }, { "epoch": 1.587001055310696, "grad_norm": 0.5416823042544131, "learning_rate": 5.366241411174328e-05, "loss": 2.8477, "step": 25565 }, { "epoch": 1.5870631324104538, "grad_norm": 0.5911260145004837, "learning_rate": 5.3658812279518014e-05, "loss": 2.841, "step": 25566 }, { "epoch": 1.5871252095102117, "grad_norm": 0.5066049744144553, "learning_rate": 5.365521042820384e-05, "loss": 2.778, "step": 25567 }, { "epoch": 1.5871872866099697, "grad_norm": 0.6098648205654488, "learning_rate": 5.3651608557819546e-05, "loss": 2.8276, "step": 25568 }, { "epoch": 1.5872493637097276, "grad_norm": 0.5268115996727473, "learning_rate": 5.364800666838393e-05, "loss": 2.7284, "step": 25569 }, { "epoch": 1.5873114408094855, "grad_norm": 0.6192811631949859, "learning_rate": 5.3644404759915766e-05, "loss": 2.7784, "step": 25570 }, { "epoch": 1.5873735179092434, "grad_norm": 0.5313165847339497, "learning_rate": 5.364080283243387e-05, "loss": 2.7865, "step": 25571 }, { "epoch": 1.5874355950090013, "grad_norm": 0.6084498694464939, "learning_rate": 5.363720088595701e-05, "loss": 2.8021, "step": 25572 }, { "epoch": 1.587497672108759, "grad_norm": 0.560941922481179, "learning_rate": 5.363359892050399e-05, "loss": 2.8489, "step": 25573 }, { "epoch": 1.587559749208517, "grad_norm": 0.5708448198884513, "learning_rate": 5.3629996936093605e-05, "loss": 2.7369, "step": 25574 }, { "epoch": 1.5876218263082749, "grad_norm": 0.5916423784786863, "learning_rate": 5.362639493274465e-05, "loss": 2.7722, "step": 25575 }, { "epoch": 1.5876839034080328, "grad_norm": 0.6512087696518282, "learning_rate": 5.3622792910475914e-05, "loss": 2.7247, "step": 25576 }, { "epoch": 1.5877459805077907, "grad_norm": 0.5943598125374623, "learning_rate": 5.361919086930618e-05, "loss": 2.7523, "step": 25577 }, { "epoch": 1.5878080576075486, "grad_norm": 0.5982486556373218, "learning_rate": 5.361558880925426e-05, "loss": 2.7046, "step": 25578 }, { "epoch": 1.5878701347073063, "grad_norm": 0.5712445364008422, "learning_rate": 5.361198673033893e-05, "loss": 2.8783, "step": 25579 }, { "epoch": 1.5879322118070642, "grad_norm": 0.5793041627493665, "learning_rate": 5.360838463257898e-05, "loss": 2.7146, "step": 25580 }, { "epoch": 1.5879942889068221, "grad_norm": 0.5869218822599387, "learning_rate": 5.360478251599321e-05, "loss": 2.6855, "step": 25581 }, { "epoch": 1.58805636600658, "grad_norm": 0.6677010652341939, "learning_rate": 5.3601180380600434e-05, "loss": 2.7644, "step": 25582 }, { "epoch": 1.588118443106338, "grad_norm": 0.594308916343277, "learning_rate": 5.359757822641942e-05, "loss": 2.7256, "step": 25583 }, { "epoch": 1.588180520206096, "grad_norm": 0.6296376068946378, "learning_rate": 5.359397605346895e-05, "loss": 2.7172, "step": 25584 }, { "epoch": 1.5882425973058538, "grad_norm": 0.6061809166156387, "learning_rate": 5.359037386176785e-05, "loss": 2.7643, "step": 25585 }, { "epoch": 1.5883046744056117, "grad_norm": 0.567667536771641, "learning_rate": 5.3586771651334896e-05, "loss": 2.7599, "step": 25586 }, { "epoch": 1.5883667515053697, "grad_norm": 0.561370908504399, "learning_rate": 5.358316942218888e-05, "loss": 2.7883, "step": 25587 }, { "epoch": 1.5884288286051276, "grad_norm": 0.6001825233832324, "learning_rate": 5.35795671743486e-05, "loss": 2.7056, "step": 25588 }, { "epoch": 1.5884909057048855, "grad_norm": 0.5944428417024763, "learning_rate": 5.3575964907832855e-05, "loss": 2.7609, "step": 25589 }, { "epoch": 1.5885529828046434, "grad_norm": 0.6276235590638815, "learning_rate": 5.357236262266042e-05, "loss": 2.8327, "step": 25590 }, { "epoch": 1.5886150599044013, "grad_norm": 0.645428605060562, "learning_rate": 5.356876031885011e-05, "loss": 2.7398, "step": 25591 }, { "epoch": 1.5886771370041592, "grad_norm": 0.5922310876028679, "learning_rate": 5.3565157996420715e-05, "loss": 2.8198, "step": 25592 }, { "epoch": 1.5887392141039172, "grad_norm": 0.6406883115420003, "learning_rate": 5.3561555655391014e-05, "loss": 2.8358, "step": 25593 }, { "epoch": 1.588801291203675, "grad_norm": 0.6538650454137359, "learning_rate": 5.355795329577983e-05, "loss": 2.8295, "step": 25594 }, { "epoch": 1.588863368303433, "grad_norm": 0.5964606299729744, "learning_rate": 5.355435091760592e-05, "loss": 2.7378, "step": 25595 }, { "epoch": 1.588925445403191, "grad_norm": 0.6227186289021583, "learning_rate": 5.355074852088811e-05, "loss": 2.8269, "step": 25596 }, { "epoch": 1.5889875225029486, "grad_norm": 0.5795840590075025, "learning_rate": 5.3547146105645175e-05, "loss": 2.8352, "step": 25597 }, { "epoch": 1.5890495996027065, "grad_norm": 0.6226962307642702, "learning_rate": 5.3543543671895933e-05, "loss": 2.8008, "step": 25598 }, { "epoch": 1.5891116767024644, "grad_norm": 0.6277635037038144, "learning_rate": 5.353994121965914e-05, "loss": 2.8507, "step": 25599 }, { "epoch": 1.5891737538022224, "grad_norm": 0.6059688348719425, "learning_rate": 5.353633874895364e-05, "loss": 2.8289, "step": 25600 }, { "epoch": 1.5892358309019803, "grad_norm": 0.6406120671153116, "learning_rate": 5.35327362597982e-05, "loss": 2.7301, "step": 25601 }, { "epoch": 1.5892979080017382, "grad_norm": 0.6276570584076169, "learning_rate": 5.35291337522116e-05, "loss": 2.8387, "step": 25602 }, { "epoch": 1.589359985101496, "grad_norm": 0.6239399981047006, "learning_rate": 5.3525531226212664e-05, "loss": 2.7555, "step": 25603 }, { "epoch": 1.5894220622012538, "grad_norm": 0.6001643492200086, "learning_rate": 5.3521928681820164e-05, "loss": 2.7367, "step": 25604 }, { "epoch": 1.5894841393010117, "grad_norm": 0.6417853848339509, "learning_rate": 5.351832611905292e-05, "loss": 2.7849, "step": 25605 }, { "epoch": 1.5895462164007697, "grad_norm": 0.5882367607406958, "learning_rate": 5.35147235379297e-05, "loss": 2.7647, "step": 25606 }, { "epoch": 1.5896082935005276, "grad_norm": 0.654632618863966, "learning_rate": 5.3511120938469325e-05, "loss": 2.827, "step": 25607 }, { "epoch": 1.5896703706002855, "grad_norm": 0.5896780021157303, "learning_rate": 5.350751832069059e-05, "loss": 2.7722, "step": 25608 }, { "epoch": 1.5897324477000434, "grad_norm": 0.5504302942560837, "learning_rate": 5.350391568461226e-05, "loss": 2.7605, "step": 25609 }, { "epoch": 1.5897945247998013, "grad_norm": 0.580709813455518, "learning_rate": 5.350031303025316e-05, "loss": 2.77, "step": 25610 }, { "epoch": 1.5898566018995592, "grad_norm": 0.5120102729027542, "learning_rate": 5.3496710357632075e-05, "loss": 2.7638, "step": 25611 }, { "epoch": 1.5899186789993172, "grad_norm": 0.564362172121449, "learning_rate": 5.3493107666767806e-05, "loss": 2.8211, "step": 25612 }, { "epoch": 1.589980756099075, "grad_norm": 0.5414493075777193, "learning_rate": 5.348950495767915e-05, "loss": 2.6094, "step": 25613 }, { "epoch": 1.590042833198833, "grad_norm": 0.5458184526565087, "learning_rate": 5.34859022303849e-05, "loss": 2.8475, "step": 25614 }, { "epoch": 1.590104910298591, "grad_norm": 0.5730602229045335, "learning_rate": 5.348229948490384e-05, "loss": 2.6564, "step": 25615 }, { "epoch": 1.5901669873983488, "grad_norm": 0.529505727180436, "learning_rate": 5.3478696721254785e-05, "loss": 2.7365, "step": 25616 }, { "epoch": 1.5902290644981067, "grad_norm": 0.5619248257568485, "learning_rate": 5.347509393945652e-05, "loss": 2.7651, "step": 25617 }, { "epoch": 1.5902911415978647, "grad_norm": 0.5174231531149578, "learning_rate": 5.347149113952786e-05, "loss": 2.7726, "step": 25618 }, { "epoch": 1.5903532186976226, "grad_norm": 0.5518192763401253, "learning_rate": 5.346788832148758e-05, "loss": 2.8536, "step": 25619 }, { "epoch": 1.5904152957973805, "grad_norm": 0.5538550026326319, "learning_rate": 5.346428548535448e-05, "loss": 2.7736, "step": 25620 }, { "epoch": 1.5904773728971382, "grad_norm": 0.5870202296961147, "learning_rate": 5.3460682631147375e-05, "loss": 2.7632, "step": 25621 }, { "epoch": 1.5905394499968961, "grad_norm": 0.5432647337444986, "learning_rate": 5.345707975888503e-05, "loss": 2.738, "step": 25622 }, { "epoch": 1.590601527096654, "grad_norm": 0.5817643392047295, "learning_rate": 5.345347686858627e-05, "loss": 2.792, "step": 25623 }, { "epoch": 1.590663604196412, "grad_norm": 0.6054431786828568, "learning_rate": 5.344987396026988e-05, "loss": 2.7678, "step": 25624 }, { "epoch": 1.5907256812961699, "grad_norm": 0.5381827985535826, "learning_rate": 5.344627103395466e-05, "loss": 2.761, "step": 25625 }, { "epoch": 1.5907877583959278, "grad_norm": 0.5970671748814174, "learning_rate": 5.344266808965942e-05, "loss": 2.7415, "step": 25626 }, { "epoch": 1.5908498354956855, "grad_norm": 0.5789417734714227, "learning_rate": 5.343906512740293e-05, "loss": 2.7897, "step": 25627 }, { "epoch": 1.5909119125954434, "grad_norm": 0.5773277197897863, "learning_rate": 5.343546214720401e-05, "loss": 2.744, "step": 25628 }, { "epoch": 1.5909739896952013, "grad_norm": 0.6217831318536301, "learning_rate": 5.3431859149081445e-05, "loss": 2.7867, "step": 25629 }, { "epoch": 1.5910360667949592, "grad_norm": 0.5830768435710719, "learning_rate": 5.342825613305405e-05, "loss": 2.743, "step": 25630 }, { "epoch": 1.5910981438947172, "grad_norm": 0.6052526272854984, "learning_rate": 5.342465309914059e-05, "loss": 2.7963, "step": 25631 }, { "epoch": 1.591160220994475, "grad_norm": 0.5306735557743166, "learning_rate": 5.34210500473599e-05, "loss": 2.7745, "step": 25632 }, { "epoch": 1.591222298094233, "grad_norm": 0.6173613581056772, "learning_rate": 5.341744697773075e-05, "loss": 2.7687, "step": 25633 }, { "epoch": 1.591284375193991, "grad_norm": 0.5574920834633559, "learning_rate": 5.3413843890271964e-05, "loss": 2.7579, "step": 25634 }, { "epoch": 1.5913464522937488, "grad_norm": 0.6032562192176557, "learning_rate": 5.341024078500231e-05, "loss": 2.8337, "step": 25635 }, { "epoch": 1.5914085293935067, "grad_norm": 0.5505847445093108, "learning_rate": 5.3406637661940615e-05, "loss": 2.7581, "step": 25636 }, { "epoch": 1.5914706064932647, "grad_norm": 0.5860416828358478, "learning_rate": 5.3403034521105663e-05, "loss": 2.7951, "step": 25637 }, { "epoch": 1.5915326835930226, "grad_norm": 0.5410165063733795, "learning_rate": 5.339943136251624e-05, "loss": 2.7619, "step": 25638 }, { "epoch": 1.5915947606927805, "grad_norm": 0.6426423803048255, "learning_rate": 5.339582818619117e-05, "loss": 2.8075, "step": 25639 }, { "epoch": 1.5916568377925384, "grad_norm": 0.5329393845439064, "learning_rate": 5.339222499214922e-05, "loss": 2.8026, "step": 25640 }, { "epoch": 1.5917189148922963, "grad_norm": 0.6180907532912322, "learning_rate": 5.338862178040924e-05, "loss": 2.8198, "step": 25641 }, { "epoch": 1.5917809919920543, "grad_norm": 0.6394430981166257, "learning_rate": 5.338501855098997e-05, "loss": 2.6798, "step": 25642 }, { "epoch": 1.5918430690918122, "grad_norm": 0.6063018973248955, "learning_rate": 5.338141530391026e-05, "loss": 2.7535, "step": 25643 }, { "epoch": 1.59190514619157, "grad_norm": 0.6358036415880186, "learning_rate": 5.337781203918888e-05, "loss": 2.7685, "step": 25644 }, { "epoch": 1.5919672232913278, "grad_norm": 0.5459355049992113, "learning_rate": 5.337420875684462e-05, "loss": 2.7662, "step": 25645 }, { "epoch": 1.5920293003910857, "grad_norm": 0.6083771531988085, "learning_rate": 5.3370605456896315e-05, "loss": 2.7307, "step": 25646 }, { "epoch": 1.5920913774908436, "grad_norm": 0.5631906455048129, "learning_rate": 5.336700213936273e-05, "loss": 2.7945, "step": 25647 }, { "epoch": 1.5921534545906015, "grad_norm": 0.6476280721647887, "learning_rate": 5.336339880426269e-05, "loss": 2.76, "step": 25648 }, { "epoch": 1.5922155316903595, "grad_norm": 0.6078792211788836, "learning_rate": 5.335979545161496e-05, "loss": 2.7826, "step": 25649 }, { "epoch": 1.5922776087901174, "grad_norm": 0.6636171306014905, "learning_rate": 5.335619208143837e-05, "loss": 2.7717, "step": 25650 }, { "epoch": 1.592339685889875, "grad_norm": 0.5895108679415922, "learning_rate": 5.335258869375171e-05, "loss": 2.7573, "step": 25651 }, { "epoch": 1.592401762989633, "grad_norm": 0.6725929327925567, "learning_rate": 5.334898528857379e-05, "loss": 2.8402, "step": 25652 }, { "epoch": 1.592463840089391, "grad_norm": 0.649045039140894, "learning_rate": 5.3345381865923395e-05, "loss": 2.7981, "step": 25653 }, { "epoch": 1.5925259171891488, "grad_norm": 0.5578907097021919, "learning_rate": 5.334177842581933e-05, "loss": 2.7388, "step": 25654 }, { "epoch": 1.5925879942889067, "grad_norm": 0.6324114765068701, "learning_rate": 5.3338174968280393e-05, "loss": 2.803, "step": 25655 }, { "epoch": 1.5926500713886647, "grad_norm": 0.5457381557397313, "learning_rate": 5.333457149332539e-05, "loss": 2.7552, "step": 25656 }, { "epoch": 1.5927121484884226, "grad_norm": 0.6151434210499815, "learning_rate": 5.3330968000973126e-05, "loss": 2.7292, "step": 25657 }, { "epoch": 1.5927742255881805, "grad_norm": 0.5382222523905608, "learning_rate": 5.3327364491242384e-05, "loss": 2.6836, "step": 25658 }, { "epoch": 1.5928363026879384, "grad_norm": 0.5474786696350535, "learning_rate": 5.332376096415198e-05, "loss": 2.795, "step": 25659 }, { "epoch": 1.5928983797876963, "grad_norm": 0.6072023491803312, "learning_rate": 5.3320157419720696e-05, "loss": 2.6895, "step": 25660 }, { "epoch": 1.5929604568874542, "grad_norm": 0.5431187100150493, "learning_rate": 5.331655385796736e-05, "loss": 2.7519, "step": 25661 }, { "epoch": 1.5930225339872122, "grad_norm": 0.6514324515845309, "learning_rate": 5.331295027891076e-05, "loss": 2.7852, "step": 25662 }, { "epoch": 1.59308461108697, "grad_norm": 0.5648131492841492, "learning_rate": 5.330934668256967e-05, "loss": 2.7337, "step": 25663 }, { "epoch": 1.593146688186728, "grad_norm": 0.6700305906116129, "learning_rate": 5.330574306896294e-05, "loss": 2.8623, "step": 25664 }, { "epoch": 1.593208765286486, "grad_norm": 0.6080026089022528, "learning_rate": 5.330213943810933e-05, "loss": 2.7436, "step": 25665 }, { "epoch": 1.5932708423862438, "grad_norm": 0.5914910410640315, "learning_rate": 5.329853579002767e-05, "loss": 2.7062, "step": 25666 }, { "epoch": 1.5933329194860018, "grad_norm": 0.588244664243792, "learning_rate": 5.329493212473673e-05, "loss": 2.7875, "step": 25667 }, { "epoch": 1.5933949965857597, "grad_norm": 0.6614826353119401, "learning_rate": 5.329132844225535e-05, "loss": 2.8123, "step": 25668 }, { "epoch": 1.5934570736855174, "grad_norm": 0.5394864557508784, "learning_rate": 5.3287724742602296e-05, "loss": 2.7807, "step": 25669 }, { "epoch": 1.5935191507852753, "grad_norm": 0.630902598881444, "learning_rate": 5.3284121025796386e-05, "loss": 2.7442, "step": 25670 }, { "epoch": 1.5935812278850332, "grad_norm": 0.6018600471120659, "learning_rate": 5.328051729185642e-05, "loss": 2.792, "step": 25671 }, { "epoch": 1.5936433049847911, "grad_norm": 0.6164665840510387, "learning_rate": 5.3276913540801196e-05, "loss": 2.7786, "step": 25672 }, { "epoch": 1.593705382084549, "grad_norm": 0.613169423009607, "learning_rate": 5.3273309772649524e-05, "loss": 2.7848, "step": 25673 }, { "epoch": 1.593767459184307, "grad_norm": 0.5665933497299507, "learning_rate": 5.326970598742019e-05, "loss": 2.7371, "step": 25674 }, { "epoch": 1.5938295362840647, "grad_norm": 0.5490535395351923, "learning_rate": 5.326610218513203e-05, "loss": 2.7338, "step": 25675 }, { "epoch": 1.5938916133838226, "grad_norm": 0.5724287614268888, "learning_rate": 5.3262498365803804e-05, "loss": 2.7013, "step": 25676 }, { "epoch": 1.5939536904835805, "grad_norm": 0.5800504581519647, "learning_rate": 5.325889452945434e-05, "loss": 2.7404, "step": 25677 }, { "epoch": 1.5940157675833384, "grad_norm": 0.5701413877708366, "learning_rate": 5.325529067610242e-05, "loss": 2.783, "step": 25678 }, { "epoch": 1.5940778446830963, "grad_norm": 0.6415006421755686, "learning_rate": 5.325168680576686e-05, "loss": 2.759, "step": 25679 }, { "epoch": 1.5941399217828542, "grad_norm": 0.5658364422375194, "learning_rate": 5.324808291846648e-05, "loss": 2.7166, "step": 25680 }, { "epoch": 1.5942019988826122, "grad_norm": 0.6309441902142255, "learning_rate": 5.324447901422004e-05, "loss": 2.7591, "step": 25681 }, { "epoch": 1.59426407598237, "grad_norm": 0.60143548689728, "learning_rate": 5.3240875093046385e-05, "loss": 2.8143, "step": 25682 }, { "epoch": 1.594326153082128, "grad_norm": 0.5527073075100087, "learning_rate": 5.3237271154964285e-05, "loss": 2.8061, "step": 25683 }, { "epoch": 1.594388230181886, "grad_norm": 0.6083807292093949, "learning_rate": 5.3233667199992564e-05, "loss": 2.7317, "step": 25684 }, { "epoch": 1.5944503072816438, "grad_norm": 0.5620194607755734, "learning_rate": 5.3230063228149994e-05, "loss": 2.7822, "step": 25685 }, { "epoch": 1.5945123843814017, "grad_norm": 0.5863174224019031, "learning_rate": 5.3226459239455425e-05, "loss": 2.7916, "step": 25686 }, { "epoch": 1.5945744614811597, "grad_norm": 0.6205786962609536, "learning_rate": 5.322285523392763e-05, "loss": 2.7952, "step": 25687 }, { "epoch": 1.5946365385809176, "grad_norm": 0.6087567298485829, "learning_rate": 5.3219251211585394e-05, "loss": 2.7574, "step": 25688 }, { "epoch": 1.5946986156806755, "grad_norm": 0.5974058985728317, "learning_rate": 5.321564717244757e-05, "loss": 2.8215, "step": 25689 }, { "epoch": 1.5947606927804334, "grad_norm": 0.5297987413202774, "learning_rate": 5.3212043116532925e-05, "loss": 2.7936, "step": 25690 }, { "epoch": 1.5948227698801913, "grad_norm": 0.6360759181439986, "learning_rate": 5.320843904386028e-05, "loss": 2.7939, "step": 25691 }, { "epoch": 1.5948848469799493, "grad_norm": 0.529131705213021, "learning_rate": 5.320483495444841e-05, "loss": 2.7991, "step": 25692 }, { "epoch": 1.594946924079707, "grad_norm": 0.576184702584961, "learning_rate": 5.3201230848316165e-05, "loss": 2.7717, "step": 25693 }, { "epoch": 1.5950090011794649, "grad_norm": 0.53661751160194, "learning_rate": 5.3197626725482294e-05, "loss": 2.7867, "step": 25694 }, { "epoch": 1.5950710782792228, "grad_norm": 0.533068489711054, "learning_rate": 5.319402258596564e-05, "loss": 2.7916, "step": 25695 }, { "epoch": 1.5951331553789807, "grad_norm": 0.5585929763895293, "learning_rate": 5.319041842978499e-05, "loss": 2.7517, "step": 25696 }, { "epoch": 1.5951952324787386, "grad_norm": 0.5647321229328344, "learning_rate": 5.318681425695917e-05, "loss": 2.7894, "step": 25697 }, { "epoch": 1.5952573095784965, "grad_norm": 0.5877525699775096, "learning_rate": 5.3183210067506964e-05, "loss": 2.7551, "step": 25698 }, { "epoch": 1.5953193866782542, "grad_norm": 0.5652733357677008, "learning_rate": 5.317960586144716e-05, "loss": 2.8947, "step": 25699 }, { "epoch": 1.5953814637780122, "grad_norm": 0.621975590607161, "learning_rate": 5.317600163879859e-05, "loss": 2.7994, "step": 25700 }, { "epoch": 1.59544354087777, "grad_norm": 0.6400732362463418, "learning_rate": 5.3172397399580045e-05, "loss": 2.7649, "step": 25701 }, { "epoch": 1.595505617977528, "grad_norm": 0.5895257054484486, "learning_rate": 5.316879314381034e-05, "loss": 2.7317, "step": 25702 }, { "epoch": 1.595567695077286, "grad_norm": 0.6216578530981397, "learning_rate": 5.316518887150827e-05, "loss": 2.7577, "step": 25703 }, { "epoch": 1.5956297721770438, "grad_norm": 0.5868217408345163, "learning_rate": 5.3161584582692625e-05, "loss": 2.6878, "step": 25704 }, { "epoch": 1.5956918492768017, "grad_norm": 0.611362346629761, "learning_rate": 5.3157980277382243e-05, "loss": 2.7448, "step": 25705 }, { "epoch": 1.5957539263765597, "grad_norm": 0.5343093854095213, "learning_rate": 5.3154375955595905e-05, "loss": 2.7417, "step": 25706 }, { "epoch": 1.5958160034763176, "grad_norm": 0.6019377095639286, "learning_rate": 5.315077161735242e-05, "loss": 2.6924, "step": 25707 }, { "epoch": 1.5958780805760755, "grad_norm": 0.6027953655364727, "learning_rate": 5.3147167262670604e-05, "loss": 2.769, "step": 25708 }, { "epoch": 1.5959401576758334, "grad_norm": 0.6073709693503314, "learning_rate": 5.3143562891569255e-05, "loss": 2.8501, "step": 25709 }, { "epoch": 1.5960022347755913, "grad_norm": 0.5501245140969788, "learning_rate": 5.313995850406715e-05, "loss": 2.786, "step": 25710 }, { "epoch": 1.5960643118753493, "grad_norm": 0.5891335613989244, "learning_rate": 5.313635410018315e-05, "loss": 2.7826, "step": 25711 }, { "epoch": 1.5961263889751072, "grad_norm": 0.5578727069921292, "learning_rate": 5.3132749679936014e-05, "loss": 2.7344, "step": 25712 }, { "epoch": 1.596188466074865, "grad_norm": 0.5714640553134994, "learning_rate": 5.312914524334457e-05, "loss": 2.8146, "step": 25713 }, { "epoch": 1.596250543174623, "grad_norm": 0.569764671058123, "learning_rate": 5.31255407904276e-05, "loss": 2.7612, "step": 25714 }, { "epoch": 1.596312620274381, "grad_norm": 0.5213178447325466, "learning_rate": 5.3121936321203935e-05, "loss": 2.7606, "step": 25715 }, { "epoch": 1.5963746973741388, "grad_norm": 0.5887741344308024, "learning_rate": 5.311833183569237e-05, "loss": 2.8521, "step": 25716 }, { "epoch": 1.5964367744738965, "grad_norm": 0.5121409880004903, "learning_rate": 5.311472733391172e-05, "loss": 2.7491, "step": 25717 }, { "epoch": 1.5964988515736545, "grad_norm": 0.5703203644928776, "learning_rate": 5.311112281588078e-05, "loss": 2.8154, "step": 25718 }, { "epoch": 1.5965609286734124, "grad_norm": 0.5700172809859001, "learning_rate": 5.310751828161834e-05, "loss": 2.7429, "step": 25719 }, { "epoch": 1.5966230057731703, "grad_norm": 0.5720538375633747, "learning_rate": 5.310391373114324e-05, "loss": 2.8069, "step": 25720 }, { "epoch": 1.5966850828729282, "grad_norm": 0.5619758442051923, "learning_rate": 5.310030916447424e-05, "loss": 2.8039, "step": 25721 }, { "epoch": 1.5967471599726861, "grad_norm": 0.6112162576032958, "learning_rate": 5.3096704581630197e-05, "loss": 2.807, "step": 25722 }, { "epoch": 1.5968092370724438, "grad_norm": 0.5475576499738974, "learning_rate": 5.3093099982629904e-05, "loss": 2.7885, "step": 25723 }, { "epoch": 1.5968713141722017, "grad_norm": 0.6087267435319527, "learning_rate": 5.3089495367492136e-05, "loss": 2.8035, "step": 25724 }, { "epoch": 1.5969333912719597, "grad_norm": 0.5973457078155755, "learning_rate": 5.3085890736235744e-05, "loss": 2.8046, "step": 25725 }, { "epoch": 1.5969954683717176, "grad_norm": 0.5671145252961386, "learning_rate": 5.3082286088879486e-05, "loss": 2.9011, "step": 25726 }, { "epoch": 1.5970575454714755, "grad_norm": 0.528067283185204, "learning_rate": 5.307868142544221e-05, "loss": 2.6984, "step": 25727 }, { "epoch": 1.5971196225712334, "grad_norm": 0.5638814492927625, "learning_rate": 5.30750767459427e-05, "loss": 2.7644, "step": 25728 }, { "epoch": 1.5971816996709913, "grad_norm": 0.56345307881361, "learning_rate": 5.307147205039977e-05, "loss": 2.8241, "step": 25729 }, { "epoch": 1.5972437767707492, "grad_norm": 0.5494473585230006, "learning_rate": 5.3067867338832224e-05, "loss": 2.867, "step": 25730 }, { "epoch": 1.5973058538705072, "grad_norm": 0.5780192613560258, "learning_rate": 5.3064262611258876e-05, "loss": 2.7766, "step": 25731 }, { "epoch": 1.597367930970265, "grad_norm": 0.5473874279892849, "learning_rate": 5.3060657867698505e-05, "loss": 2.7341, "step": 25732 }, { "epoch": 1.597430008070023, "grad_norm": 0.510353918175967, "learning_rate": 5.3057053108169965e-05, "loss": 2.739, "step": 25733 }, { "epoch": 1.597492085169781, "grad_norm": 0.5798316714592028, "learning_rate": 5.305344833269201e-05, "loss": 2.7846, "step": 25734 }, { "epoch": 1.5975541622695388, "grad_norm": 0.5004081848065199, "learning_rate": 5.3049843541283495e-05, "loss": 2.6543, "step": 25735 }, { "epoch": 1.5976162393692968, "grad_norm": 0.6319389907778583, "learning_rate": 5.304623873396321e-05, "loss": 2.7712, "step": 25736 }, { "epoch": 1.5976783164690547, "grad_norm": 0.5613607154257033, "learning_rate": 5.3042633910749926e-05, "loss": 2.7728, "step": 25737 }, { "epoch": 1.5977403935688126, "grad_norm": 0.601899387107191, "learning_rate": 5.3039029071662515e-05, "loss": 2.7945, "step": 25738 }, { "epoch": 1.5978024706685705, "grad_norm": 0.5812696812486813, "learning_rate": 5.303542421671973e-05, "loss": 2.7726, "step": 25739 }, { "epoch": 1.5978645477683284, "grad_norm": 0.5519425344564771, "learning_rate": 5.3031819345940414e-05, "loss": 2.7782, "step": 25740 }, { "epoch": 1.5979266248680861, "grad_norm": 0.5832019527395416, "learning_rate": 5.302821445934336e-05, "loss": 2.8301, "step": 25741 }, { "epoch": 1.597988701967844, "grad_norm": 0.5801566228189871, "learning_rate": 5.302460955694737e-05, "loss": 2.7456, "step": 25742 }, { "epoch": 1.598050779067602, "grad_norm": 0.5480182142641641, "learning_rate": 5.3021004638771266e-05, "loss": 2.7797, "step": 25743 }, { "epoch": 1.5981128561673599, "grad_norm": 0.602146434997139, "learning_rate": 5.301739970483384e-05, "loss": 2.759, "step": 25744 }, { "epoch": 1.5981749332671178, "grad_norm": 0.573356863895567, "learning_rate": 5.301379475515391e-05, "loss": 2.8049, "step": 25745 }, { "epoch": 1.5982370103668757, "grad_norm": 0.6404577595199819, "learning_rate": 5.301018978975028e-05, "loss": 2.7822, "step": 25746 }, { "epoch": 1.5982990874666334, "grad_norm": 0.5655740609599329, "learning_rate": 5.300658480864176e-05, "loss": 2.6929, "step": 25747 }, { "epoch": 1.5983611645663913, "grad_norm": 0.6377470454076907, "learning_rate": 5.3002979811847154e-05, "loss": 2.7087, "step": 25748 }, { "epoch": 1.5984232416661492, "grad_norm": 0.6117638933032884, "learning_rate": 5.299937479938528e-05, "loss": 2.7126, "step": 25749 }, { "epoch": 1.5984853187659072, "grad_norm": 0.5719676104325144, "learning_rate": 5.299576977127493e-05, "loss": 2.705, "step": 25750 }, { "epoch": 1.598547395865665, "grad_norm": 0.6073196423228266, "learning_rate": 5.2992164727534934e-05, "loss": 2.8295, "step": 25751 }, { "epoch": 1.598609472965423, "grad_norm": 0.6161704556427158, "learning_rate": 5.298855966818408e-05, "loss": 2.7188, "step": 25752 }, { "epoch": 1.598671550065181, "grad_norm": 0.5576289187398178, "learning_rate": 5.2984954593241184e-05, "loss": 2.6894, "step": 25753 }, { "epoch": 1.5987336271649388, "grad_norm": 0.5901190008154656, "learning_rate": 5.298134950272505e-05, "loss": 2.8608, "step": 25754 }, { "epoch": 1.5987957042646967, "grad_norm": 0.6184682233367848, "learning_rate": 5.29777443966545e-05, "loss": 2.808, "step": 25755 }, { "epoch": 1.5988577813644547, "grad_norm": 0.519617432064531, "learning_rate": 5.297413927504834e-05, "loss": 2.8365, "step": 25756 }, { "epoch": 1.5989198584642126, "grad_norm": 0.5979014124458526, "learning_rate": 5.297053413792535e-05, "loss": 2.69, "step": 25757 }, { "epoch": 1.5989819355639705, "grad_norm": 0.5798113431023614, "learning_rate": 5.29669289853044e-05, "loss": 2.7906, "step": 25758 }, { "epoch": 1.5990440126637284, "grad_norm": 0.5882511744213904, "learning_rate": 5.2963323817204226e-05, "loss": 2.8409, "step": 25759 }, { "epoch": 1.5991060897634863, "grad_norm": 0.5916710395606619, "learning_rate": 5.2959718633643694e-05, "loss": 2.7814, "step": 25760 }, { "epoch": 1.5991681668632443, "grad_norm": 0.5642366885320855, "learning_rate": 5.2956113434641596e-05, "loss": 2.8085, "step": 25761 }, { "epoch": 1.5992302439630022, "grad_norm": 0.5551736340107992, "learning_rate": 5.2952508220216715e-05, "loss": 2.8012, "step": 25762 }, { "epoch": 1.59929232106276, "grad_norm": 0.5763928264524277, "learning_rate": 5.2948902990387896e-05, "loss": 2.7699, "step": 25763 }, { "epoch": 1.599354398162518, "grad_norm": 0.5464579490614702, "learning_rate": 5.294529774517393e-05, "loss": 2.7839, "step": 25764 }, { "epoch": 1.5994164752622757, "grad_norm": 0.5855611482576747, "learning_rate": 5.294169248459363e-05, "loss": 2.7962, "step": 25765 }, { "epoch": 1.5994785523620336, "grad_norm": 0.5510686030111037, "learning_rate": 5.29380872086658e-05, "loss": 2.7999, "step": 25766 }, { "epoch": 1.5995406294617915, "grad_norm": 0.6239474493572986, "learning_rate": 5.293448191740927e-05, "loss": 2.7398, "step": 25767 }, { "epoch": 1.5996027065615495, "grad_norm": 0.5474283144019414, "learning_rate": 5.293087661084283e-05, "loss": 2.7742, "step": 25768 }, { "epoch": 1.5996647836613074, "grad_norm": 0.6352987310265673, "learning_rate": 5.292727128898529e-05, "loss": 2.797, "step": 25769 }, { "epoch": 1.5997268607610653, "grad_norm": 0.5942033492566902, "learning_rate": 5.292366595185547e-05, "loss": 2.7427, "step": 25770 }, { "epoch": 1.599788937860823, "grad_norm": 0.6444416099404704, "learning_rate": 5.2920060599472165e-05, "loss": 2.702, "step": 25771 }, { "epoch": 1.599851014960581, "grad_norm": 0.6287554447552619, "learning_rate": 5.291645523185421e-05, "loss": 2.8045, "step": 25772 }, { "epoch": 1.5999130920603388, "grad_norm": 0.6436077321814996, "learning_rate": 5.2912849849020385e-05, "loss": 2.7497, "step": 25773 }, { "epoch": 1.5999751691600967, "grad_norm": 0.5716484830242453, "learning_rate": 5.2909244450989525e-05, "loss": 2.8528, "step": 25774 }, { "epoch": 1.6000372462598547, "grad_norm": 0.6021327993890785, "learning_rate": 5.290563903778043e-05, "loss": 2.7577, "step": 25775 }, { "epoch": 1.6000993233596126, "grad_norm": 0.5985377424460897, "learning_rate": 5.2902033609411914e-05, "loss": 2.7457, "step": 25776 }, { "epoch": 1.6001614004593705, "grad_norm": 0.5611183098241028, "learning_rate": 5.289842816590278e-05, "loss": 2.7437, "step": 25777 }, { "epoch": 1.6002234775591284, "grad_norm": 0.6749692017981882, "learning_rate": 5.289482270727184e-05, "loss": 2.879, "step": 25778 }, { "epoch": 1.6002855546588863, "grad_norm": 0.6312861932633016, "learning_rate": 5.289121723353791e-05, "loss": 2.7542, "step": 25779 }, { "epoch": 1.6003476317586443, "grad_norm": 0.6430504405154893, "learning_rate": 5.28876117447198e-05, "loss": 2.6296, "step": 25780 }, { "epoch": 1.6004097088584022, "grad_norm": 0.5653886269532379, "learning_rate": 5.288400624083632e-05, "loss": 2.7235, "step": 25781 }, { "epoch": 1.60047178595816, "grad_norm": 0.7200005474353405, "learning_rate": 5.2880400721906275e-05, "loss": 2.7251, "step": 25782 }, { "epoch": 1.600533863057918, "grad_norm": 0.5536663434406434, "learning_rate": 5.287679518794848e-05, "loss": 2.701, "step": 25783 }, { "epoch": 1.600595940157676, "grad_norm": 0.7386879179375848, "learning_rate": 5.2873189638981744e-05, "loss": 2.7638, "step": 25784 }, { "epoch": 1.6006580172574338, "grad_norm": 0.5176275841455685, "learning_rate": 5.286958407502489e-05, "loss": 2.7708, "step": 25785 }, { "epoch": 1.6007200943571918, "grad_norm": 0.7024245906213472, "learning_rate": 5.286597849609671e-05, "loss": 2.8348, "step": 25786 }, { "epoch": 1.6007821714569497, "grad_norm": 0.5543729669389571, "learning_rate": 5.286237290221602e-05, "loss": 2.7924, "step": 25787 }, { "epoch": 1.6008442485567076, "grad_norm": 0.687573667394117, "learning_rate": 5.285876729340165e-05, "loss": 2.7386, "step": 25788 }, { "epoch": 1.6009063256564653, "grad_norm": 0.5715517401999054, "learning_rate": 5.285516166967238e-05, "loss": 2.8021, "step": 25789 }, { "epoch": 1.6009684027562232, "grad_norm": 0.5891942876018343, "learning_rate": 5.285155603104706e-05, "loss": 2.8231, "step": 25790 }, { "epoch": 1.6010304798559811, "grad_norm": 0.5761150008757671, "learning_rate": 5.284795037754446e-05, "loss": 2.7879, "step": 25791 }, { "epoch": 1.601092556955739, "grad_norm": 0.5512262535170581, "learning_rate": 5.284434470918342e-05, "loss": 2.7725, "step": 25792 }, { "epoch": 1.601154634055497, "grad_norm": 0.5765250206734389, "learning_rate": 5.2840739025982746e-05, "loss": 2.7115, "step": 25793 }, { "epoch": 1.6012167111552549, "grad_norm": 0.5743871947811018, "learning_rate": 5.283713332796124e-05, "loss": 2.7446, "step": 25794 }, { "epoch": 1.6012787882550126, "grad_norm": 0.5591664879320581, "learning_rate": 5.2833527615137724e-05, "loss": 2.7517, "step": 25795 }, { "epoch": 1.6013408653547705, "grad_norm": 0.5491354953831831, "learning_rate": 5.2829921887531007e-05, "loss": 2.7258, "step": 25796 }, { "epoch": 1.6014029424545284, "grad_norm": 0.6088993465824164, "learning_rate": 5.2826316145159915e-05, "loss": 2.7928, "step": 25797 }, { "epoch": 1.6014650195542863, "grad_norm": 0.5921175500225124, "learning_rate": 5.2822710388043214e-05, "loss": 2.8224, "step": 25798 }, { "epoch": 1.6015270966540442, "grad_norm": 0.5235779986492516, "learning_rate": 5.2819104616199774e-05, "loss": 2.7339, "step": 25799 }, { "epoch": 1.6015891737538022, "grad_norm": 0.5813602189190595, "learning_rate": 5.2815498829648366e-05, "loss": 2.7945, "step": 25800 }, { "epoch": 1.60165125085356, "grad_norm": 0.5649996424986122, "learning_rate": 5.2811893028407834e-05, "loss": 2.7108, "step": 25801 }, { "epoch": 1.601713327953318, "grad_norm": 0.5603178134459903, "learning_rate": 5.2808287212496956e-05, "loss": 2.756, "step": 25802 }, { "epoch": 1.601775405053076, "grad_norm": 0.5671999401903591, "learning_rate": 5.280468138193456e-05, "loss": 2.731, "step": 25803 }, { "epoch": 1.6018374821528338, "grad_norm": 0.5254768102358341, "learning_rate": 5.280107553673947e-05, "loss": 2.9133, "step": 25804 }, { "epoch": 1.6018995592525918, "grad_norm": 0.5526427607552344, "learning_rate": 5.279746967693049e-05, "loss": 2.7846, "step": 25805 }, { "epoch": 1.6019616363523497, "grad_norm": 0.48583636258665674, "learning_rate": 5.2793863802526436e-05, "loss": 2.8094, "step": 25806 }, { "epoch": 1.6020237134521076, "grad_norm": 0.5340644112291061, "learning_rate": 5.279025791354609e-05, "loss": 2.7357, "step": 25807 }, { "epoch": 1.6020857905518655, "grad_norm": 0.4876634427719345, "learning_rate": 5.278665201000832e-05, "loss": 2.7585, "step": 25808 }, { "epoch": 1.6021478676516234, "grad_norm": 0.5575587681300247, "learning_rate": 5.27830460919319e-05, "loss": 2.747, "step": 25809 }, { "epoch": 1.6022099447513813, "grad_norm": 0.5120196440623755, "learning_rate": 5.277944015933566e-05, "loss": 2.699, "step": 25810 }, { "epoch": 1.6022720218511393, "grad_norm": 0.5088476048504915, "learning_rate": 5.2775834212238386e-05, "loss": 2.7834, "step": 25811 }, { "epoch": 1.6023340989508972, "grad_norm": 0.49056986237754474, "learning_rate": 5.277222825065893e-05, "loss": 2.7282, "step": 25812 }, { "epoch": 1.6023961760506549, "grad_norm": 0.5724762148268124, "learning_rate": 5.276862227461607e-05, "loss": 2.7032, "step": 25813 }, { "epoch": 1.6024582531504128, "grad_norm": 0.5212656769807792, "learning_rate": 5.2765016284128655e-05, "loss": 2.802, "step": 25814 }, { "epoch": 1.6025203302501707, "grad_norm": 0.5768866550429532, "learning_rate": 5.276141027921549e-05, "loss": 2.7809, "step": 25815 }, { "epoch": 1.6025824073499286, "grad_norm": 0.5323454277923542, "learning_rate": 5.275780425989534e-05, "loss": 2.7793, "step": 25816 }, { "epoch": 1.6026444844496865, "grad_norm": 0.5943608905870489, "learning_rate": 5.275419822618708e-05, "loss": 2.8431, "step": 25817 }, { "epoch": 1.6027065615494445, "grad_norm": 0.5446322638846093, "learning_rate": 5.275059217810948e-05, "loss": 2.7261, "step": 25818 }, { "epoch": 1.6027686386492022, "grad_norm": 0.5741503149240852, "learning_rate": 5.274698611568139e-05, "loss": 2.7374, "step": 25819 }, { "epoch": 1.60283071574896, "grad_norm": 0.5586492337501493, "learning_rate": 5.274338003892158e-05, "loss": 2.7494, "step": 25820 }, { "epoch": 1.602892792848718, "grad_norm": 0.5651773209107102, "learning_rate": 5.273977394784892e-05, "loss": 2.6858, "step": 25821 }, { "epoch": 1.602954869948476, "grad_norm": 0.556201935952544, "learning_rate": 5.27361678424822e-05, "loss": 2.8449, "step": 25822 }, { "epoch": 1.6030169470482338, "grad_norm": 0.5525461961670688, "learning_rate": 5.2732561722840215e-05, "loss": 2.8532, "step": 25823 }, { "epoch": 1.6030790241479917, "grad_norm": 0.6639634451197074, "learning_rate": 5.2728955588941784e-05, "loss": 2.8175, "step": 25824 }, { "epoch": 1.6031411012477497, "grad_norm": 0.5599668738666218, "learning_rate": 5.272534944080574e-05, "loss": 2.8314, "step": 25825 }, { "epoch": 1.6032031783475076, "grad_norm": 0.6019741978010916, "learning_rate": 5.2721743278450885e-05, "loss": 2.7249, "step": 25826 }, { "epoch": 1.6032652554472655, "grad_norm": 0.6284839562877329, "learning_rate": 5.2718137101896024e-05, "loss": 2.7818, "step": 25827 }, { "epoch": 1.6033273325470234, "grad_norm": 0.5923606780382158, "learning_rate": 5.2714530911159996e-05, "loss": 2.6604, "step": 25828 }, { "epoch": 1.6033894096467813, "grad_norm": 0.6152784785856796, "learning_rate": 5.271092470626159e-05, "loss": 2.7654, "step": 25829 }, { "epoch": 1.6034514867465393, "grad_norm": 0.5612337642351888, "learning_rate": 5.2707318487219645e-05, "loss": 2.7674, "step": 25830 }, { "epoch": 1.6035135638462972, "grad_norm": 0.6372505962343595, "learning_rate": 5.270371225405295e-05, "loss": 2.7974, "step": 25831 }, { "epoch": 1.603575640946055, "grad_norm": 0.6244599154582906, "learning_rate": 5.270010600678034e-05, "loss": 2.7739, "step": 25832 }, { "epoch": 1.603637718045813, "grad_norm": 0.6690454561217629, "learning_rate": 5.269649974542062e-05, "loss": 2.7834, "step": 25833 }, { "epoch": 1.603699795145571, "grad_norm": 0.5911862839850525, "learning_rate": 5.26928934699926e-05, "loss": 2.7615, "step": 25834 }, { "epoch": 1.6037618722453288, "grad_norm": 0.6205995140720424, "learning_rate": 5.26892871805151e-05, "loss": 2.8639, "step": 25835 }, { "epoch": 1.6038239493450868, "grad_norm": 0.5571705220064649, "learning_rate": 5.268568087700694e-05, "loss": 2.7302, "step": 25836 }, { "epoch": 1.6038860264448445, "grad_norm": 0.6563207497379312, "learning_rate": 5.2682074559486936e-05, "loss": 2.7909, "step": 25837 }, { "epoch": 1.6039481035446024, "grad_norm": 0.5788621327093082, "learning_rate": 5.267846822797389e-05, "loss": 2.8553, "step": 25838 }, { "epoch": 1.6040101806443603, "grad_norm": 0.5666034749737469, "learning_rate": 5.267486188248663e-05, "loss": 2.7707, "step": 25839 }, { "epoch": 1.6040722577441182, "grad_norm": 0.5665674703625126, "learning_rate": 5.267125552304397e-05, "loss": 2.809, "step": 25840 }, { "epoch": 1.6041343348438761, "grad_norm": 0.548704574633884, "learning_rate": 5.266764914966471e-05, "loss": 2.7385, "step": 25841 }, { "epoch": 1.604196411943634, "grad_norm": 0.5344327783858939, "learning_rate": 5.266404276236769e-05, "loss": 2.7208, "step": 25842 }, { "epoch": 1.6042584890433917, "grad_norm": 0.5794121100959193, "learning_rate": 5.2660436361171704e-05, "loss": 2.723, "step": 25843 }, { "epoch": 1.6043205661431497, "grad_norm": 0.5841264838884127, "learning_rate": 5.265682994609558e-05, "loss": 2.8433, "step": 25844 }, { "epoch": 1.6043826432429076, "grad_norm": 0.5591777069984083, "learning_rate": 5.265322351715812e-05, "loss": 2.6673, "step": 25845 }, { "epoch": 1.6044447203426655, "grad_norm": 0.5353828562984608, "learning_rate": 5.2649617074378164e-05, "loss": 2.7392, "step": 25846 }, { "epoch": 1.6045067974424234, "grad_norm": 0.5421754819214128, "learning_rate": 5.26460106177745e-05, "loss": 2.7627, "step": 25847 }, { "epoch": 1.6045688745421813, "grad_norm": 0.5865823630382593, "learning_rate": 5.2642404147365966e-05, "loss": 2.8479, "step": 25848 }, { "epoch": 1.6046309516419393, "grad_norm": 0.5235835199088097, "learning_rate": 5.263879766317135e-05, "loss": 2.8294, "step": 25849 }, { "epoch": 1.6046930287416972, "grad_norm": 0.5673555771402041, "learning_rate": 5.263519116520951e-05, "loss": 2.8039, "step": 25850 }, { "epoch": 1.604755105841455, "grad_norm": 0.5163701770550775, "learning_rate": 5.263158465349922e-05, "loss": 2.8059, "step": 25851 }, { "epoch": 1.604817182941213, "grad_norm": 0.525093022688403, "learning_rate": 5.2627978128059304e-05, "loss": 2.7757, "step": 25852 }, { "epoch": 1.604879260040971, "grad_norm": 0.5849301134244926, "learning_rate": 5.262437158890862e-05, "loss": 2.7737, "step": 25853 }, { "epoch": 1.6049413371407288, "grad_norm": 0.5581443953267814, "learning_rate": 5.2620765036065925e-05, "loss": 2.8686, "step": 25854 }, { "epoch": 1.6050034142404868, "grad_norm": 0.6210435571122427, "learning_rate": 5.261715846955008e-05, "loss": 2.6659, "step": 25855 }, { "epoch": 1.6050654913402447, "grad_norm": 0.5591009278414756, "learning_rate": 5.261355188937988e-05, "loss": 2.7242, "step": 25856 }, { "epoch": 1.6051275684400026, "grad_norm": 0.5909274169691048, "learning_rate": 5.2609945295574135e-05, "loss": 2.73, "step": 25857 }, { "epoch": 1.6051896455397605, "grad_norm": 0.6704150461572851, "learning_rate": 5.260633868815169e-05, "loss": 2.7637, "step": 25858 }, { "epoch": 1.6052517226395184, "grad_norm": 0.5643039228983374, "learning_rate": 5.260273206713133e-05, "loss": 2.7983, "step": 25859 }, { "epoch": 1.6053137997392761, "grad_norm": 0.6261402457679721, "learning_rate": 5.259912543253189e-05, "loss": 2.858, "step": 25860 }, { "epoch": 1.605375876839034, "grad_norm": 0.5362453377349848, "learning_rate": 5.259551878437218e-05, "loss": 2.8173, "step": 25861 }, { "epoch": 1.605437953938792, "grad_norm": 0.5935642805262573, "learning_rate": 5.2591912122671025e-05, "loss": 2.8023, "step": 25862 }, { "epoch": 1.6055000310385499, "grad_norm": 0.5666504432188418, "learning_rate": 5.2588305447447214e-05, "loss": 2.7354, "step": 25863 }, { "epoch": 1.6055621081383078, "grad_norm": 0.5449612784410249, "learning_rate": 5.258469875871961e-05, "loss": 2.8178, "step": 25864 }, { "epoch": 1.6056241852380657, "grad_norm": 0.6047307346471863, "learning_rate": 5.2581092056507e-05, "loss": 2.7477, "step": 25865 }, { "epoch": 1.6056862623378234, "grad_norm": 0.5515424730160955, "learning_rate": 5.2577485340828204e-05, "loss": 2.7779, "step": 25866 }, { "epoch": 1.6057483394375813, "grad_norm": 0.614564900741458, "learning_rate": 5.257387861170203e-05, "loss": 2.7718, "step": 25867 }, { "epoch": 1.6058104165373392, "grad_norm": 0.5435255749466006, "learning_rate": 5.2570271869147305e-05, "loss": 2.69, "step": 25868 }, { "epoch": 1.6058724936370972, "grad_norm": 0.6254312934495105, "learning_rate": 5.2566665113182864e-05, "loss": 2.8489, "step": 25869 }, { "epoch": 1.605934570736855, "grad_norm": 0.5333363293275527, "learning_rate": 5.2563058343827495e-05, "loss": 2.6836, "step": 25870 }, { "epoch": 1.605996647836613, "grad_norm": 0.6087472323015967, "learning_rate": 5.2559451561100045e-05, "loss": 2.7641, "step": 25871 }, { "epoch": 1.606058724936371, "grad_norm": 0.5482172977680733, "learning_rate": 5.2555844765019304e-05, "loss": 2.7999, "step": 25872 }, { "epoch": 1.6061208020361288, "grad_norm": 0.6269468027907237, "learning_rate": 5.25522379556041e-05, "loss": 2.7884, "step": 25873 }, { "epoch": 1.6061828791358868, "grad_norm": 0.595712605385307, "learning_rate": 5.254863113287325e-05, "loss": 2.739, "step": 25874 }, { "epoch": 1.6062449562356447, "grad_norm": 0.6006425536227371, "learning_rate": 5.254502429684558e-05, "loss": 2.8743, "step": 25875 }, { "epoch": 1.6063070333354026, "grad_norm": 0.5912373733785229, "learning_rate": 5.25414174475399e-05, "loss": 2.7653, "step": 25876 }, { "epoch": 1.6063691104351605, "grad_norm": 0.58369379883956, "learning_rate": 5.2537810584975025e-05, "loss": 2.7793, "step": 25877 }, { "epoch": 1.6064311875349184, "grad_norm": 0.6216449810554784, "learning_rate": 5.2534203709169774e-05, "loss": 2.793, "step": 25878 }, { "epoch": 1.6064932646346763, "grad_norm": 0.7144891662016558, "learning_rate": 5.253059682014296e-05, "loss": 2.8463, "step": 25879 }, { "epoch": 1.6065553417344343, "grad_norm": 0.6188610901953885, "learning_rate": 5.252698991791343e-05, "loss": 2.7714, "step": 25880 }, { "epoch": 1.6066174188341922, "grad_norm": 0.5849321514300447, "learning_rate": 5.2523383002499946e-05, "loss": 2.699, "step": 25881 }, { "epoch": 1.60667949593395, "grad_norm": 0.6186675421361579, "learning_rate": 5.251977607392139e-05, "loss": 2.7798, "step": 25882 }, { "epoch": 1.606741573033708, "grad_norm": 0.5655050803933585, "learning_rate": 5.251616913219653e-05, "loss": 2.7057, "step": 25883 }, { "epoch": 1.6068036501334657, "grad_norm": 0.5845403162037223, "learning_rate": 5.251256217734423e-05, "loss": 2.7269, "step": 25884 }, { "epoch": 1.6068657272332236, "grad_norm": 0.4921577535450438, "learning_rate": 5.2508955209383257e-05, "loss": 2.7088, "step": 25885 }, { "epoch": 1.6069278043329815, "grad_norm": 0.6148881937369125, "learning_rate": 5.250534822833246e-05, "loss": 2.8341, "step": 25886 }, { "epoch": 1.6069898814327395, "grad_norm": 0.5237282472158272, "learning_rate": 5.250174123421068e-05, "loss": 2.7279, "step": 25887 }, { "epoch": 1.6070519585324974, "grad_norm": 0.5783587575989307, "learning_rate": 5.2498134227036676e-05, "loss": 2.7596, "step": 25888 }, { "epoch": 1.6071140356322553, "grad_norm": 0.5583274313154521, "learning_rate": 5.249452720682932e-05, "loss": 2.8023, "step": 25889 }, { "epoch": 1.607176112732013, "grad_norm": 0.5553627739315494, "learning_rate": 5.2490920173607396e-05, "loss": 2.7242, "step": 25890 }, { "epoch": 1.607238189831771, "grad_norm": 0.5622849583187349, "learning_rate": 5.248731312738975e-05, "loss": 2.816, "step": 25891 }, { "epoch": 1.6073002669315288, "grad_norm": 0.596302218999577, "learning_rate": 5.248370606819517e-05, "loss": 2.7999, "step": 25892 }, { "epoch": 1.6073623440312867, "grad_norm": 0.5548666812387444, "learning_rate": 5.2480098996042516e-05, "loss": 2.7835, "step": 25893 }, { "epoch": 1.6074244211310447, "grad_norm": 0.5550631205002605, "learning_rate": 5.2476491910950575e-05, "loss": 2.6981, "step": 25894 }, { "epoch": 1.6074864982308026, "grad_norm": 0.5486054426884922, "learning_rate": 5.2472884812938163e-05, "loss": 2.824, "step": 25895 }, { "epoch": 1.6075485753305605, "grad_norm": 0.6655739581099472, "learning_rate": 5.2469277702024125e-05, "loss": 2.7766, "step": 25896 }, { "epoch": 1.6076106524303184, "grad_norm": 0.6638313564362406, "learning_rate": 5.246567057822726e-05, "loss": 2.7837, "step": 25897 }, { "epoch": 1.6076727295300763, "grad_norm": 0.5835800812740629, "learning_rate": 5.246206344156639e-05, "loss": 2.719, "step": 25898 }, { "epoch": 1.6077348066298343, "grad_norm": 0.6400642965101776, "learning_rate": 5.2458456292060334e-05, "loss": 2.7807, "step": 25899 }, { "epoch": 1.6077968837295922, "grad_norm": 0.5613568398836776, "learning_rate": 5.2454849129727926e-05, "loss": 2.8718, "step": 25900 }, { "epoch": 1.60785896082935, "grad_norm": 0.6206767523740326, "learning_rate": 5.245124195458796e-05, "loss": 2.8054, "step": 25901 }, { "epoch": 1.607921037929108, "grad_norm": 0.587576214939782, "learning_rate": 5.2447634766659274e-05, "loss": 2.8746, "step": 25902 }, { "epoch": 1.607983115028866, "grad_norm": 0.5732333752866242, "learning_rate": 5.24440275659607e-05, "loss": 2.8091, "step": 25903 }, { "epoch": 1.6080451921286238, "grad_norm": 0.5945413736491526, "learning_rate": 5.244042035251101e-05, "loss": 2.7804, "step": 25904 }, { "epoch": 1.6081072692283818, "grad_norm": 0.5545363959156708, "learning_rate": 5.243681312632909e-05, "loss": 2.8041, "step": 25905 }, { "epoch": 1.6081693463281397, "grad_norm": 0.6326395905294574, "learning_rate": 5.2433205887433686e-05, "loss": 2.8749, "step": 25906 }, { "epoch": 1.6082314234278976, "grad_norm": 0.5496899500321117, "learning_rate": 5.242959863584369e-05, "loss": 2.812, "step": 25907 }, { "epoch": 1.6082935005276553, "grad_norm": 0.6260404033510618, "learning_rate": 5.242599137157787e-05, "loss": 2.7572, "step": 25908 }, { "epoch": 1.6083555776274132, "grad_norm": 0.5511119472007451, "learning_rate": 5.2422384094655076e-05, "loss": 2.8057, "step": 25909 }, { "epoch": 1.6084176547271711, "grad_norm": 0.632829854279845, "learning_rate": 5.2418776805094104e-05, "loss": 2.7637, "step": 25910 }, { "epoch": 1.608479731826929, "grad_norm": 0.5540706535810886, "learning_rate": 5.24151695029138e-05, "loss": 2.7719, "step": 25911 }, { "epoch": 1.608541808926687, "grad_norm": 0.6025198217748203, "learning_rate": 5.241156218813296e-05, "loss": 2.8191, "step": 25912 }, { "epoch": 1.6086038860264449, "grad_norm": 0.5933086863320552, "learning_rate": 5.240795486077042e-05, "loss": 2.7858, "step": 25913 }, { "epoch": 1.6086659631262026, "grad_norm": 0.6157586664231925, "learning_rate": 5.240434752084499e-05, "loss": 2.7164, "step": 25914 }, { "epoch": 1.6087280402259605, "grad_norm": 0.5566129711041273, "learning_rate": 5.24007401683755e-05, "loss": 2.7747, "step": 25915 }, { "epoch": 1.6087901173257184, "grad_norm": 0.5660710362282176, "learning_rate": 5.239713280338075e-05, "loss": 2.7782, "step": 25916 }, { "epoch": 1.6088521944254763, "grad_norm": 0.5601943765218411, "learning_rate": 5.239352542587959e-05, "loss": 2.6845, "step": 25917 }, { "epoch": 1.6089142715252343, "grad_norm": 0.6032752882537653, "learning_rate": 5.2389918035890826e-05, "loss": 2.7731, "step": 25918 }, { "epoch": 1.6089763486249922, "grad_norm": 0.6068825734821922, "learning_rate": 5.238631063343328e-05, "loss": 2.7328, "step": 25919 }, { "epoch": 1.60903842572475, "grad_norm": 0.6022136186479399, "learning_rate": 5.238270321852576e-05, "loss": 2.7109, "step": 25920 }, { "epoch": 1.609100502824508, "grad_norm": 0.6466654498008539, "learning_rate": 5.2379095791187124e-05, "loss": 2.7105, "step": 25921 }, { "epoch": 1.609162579924266, "grad_norm": 0.504620127713218, "learning_rate": 5.237548835143614e-05, "loss": 2.7471, "step": 25922 }, { "epoch": 1.6092246570240238, "grad_norm": 0.6649143589416545, "learning_rate": 5.237188089929168e-05, "loss": 2.7656, "step": 25923 }, { "epoch": 1.6092867341237818, "grad_norm": 0.5417093174599781, "learning_rate": 5.236827343477253e-05, "loss": 2.8894, "step": 25924 }, { "epoch": 1.6093488112235397, "grad_norm": 0.6568897852517643, "learning_rate": 5.236466595789753e-05, "loss": 2.7615, "step": 25925 }, { "epoch": 1.6094108883232976, "grad_norm": 0.5180595606028923, "learning_rate": 5.236105846868549e-05, "loss": 2.6449, "step": 25926 }, { "epoch": 1.6094729654230555, "grad_norm": 0.6830425765211957, "learning_rate": 5.235745096715523e-05, "loss": 2.8441, "step": 25927 }, { "epoch": 1.6095350425228134, "grad_norm": 0.5952959013299421, "learning_rate": 5.2353843453325566e-05, "loss": 2.7632, "step": 25928 }, { "epoch": 1.6095971196225713, "grad_norm": 0.6311977412054671, "learning_rate": 5.235023592721535e-05, "loss": 2.8683, "step": 25929 }, { "epoch": 1.6096591967223293, "grad_norm": 0.7021143537586572, "learning_rate": 5.234662838884338e-05, "loss": 2.8393, "step": 25930 }, { "epoch": 1.6097212738220872, "grad_norm": 0.5114986441343503, "learning_rate": 5.234302083822847e-05, "loss": 2.7994, "step": 25931 }, { "epoch": 1.6097833509218449, "grad_norm": 0.789042368717032, "learning_rate": 5.233941327538947e-05, "loss": 2.7668, "step": 25932 }, { "epoch": 1.6098454280216028, "grad_norm": 0.5624047039689226, "learning_rate": 5.2335805700345165e-05, "loss": 2.8184, "step": 25933 }, { "epoch": 1.6099075051213607, "grad_norm": 0.6613377939363504, "learning_rate": 5.2332198113114394e-05, "loss": 2.742, "step": 25934 }, { "epoch": 1.6099695822211186, "grad_norm": 0.6281131458587292, "learning_rate": 5.232859051371597e-05, "loss": 2.8757, "step": 25935 }, { "epoch": 1.6100316593208766, "grad_norm": 0.5904136834734398, "learning_rate": 5.232498290216875e-05, "loss": 2.6431, "step": 25936 }, { "epoch": 1.6100937364206345, "grad_norm": 0.6460083191191305, "learning_rate": 5.232137527849151e-05, "loss": 2.7885, "step": 25937 }, { "epoch": 1.6101558135203922, "grad_norm": 0.6370644596248979, "learning_rate": 5.231776764270311e-05, "loss": 2.7061, "step": 25938 }, { "epoch": 1.61021789062015, "grad_norm": 0.6542050880844196, "learning_rate": 5.2314159994822344e-05, "loss": 2.7236, "step": 25939 }, { "epoch": 1.610279967719908, "grad_norm": 0.6001558826527841, "learning_rate": 5.2310552334868034e-05, "loss": 2.7657, "step": 25940 }, { "epoch": 1.610342044819666, "grad_norm": 0.6761353507664541, "learning_rate": 5.230694466285902e-05, "loss": 2.83, "step": 25941 }, { "epoch": 1.6104041219194238, "grad_norm": 0.6087387871933497, "learning_rate": 5.230333697881412e-05, "loss": 2.8039, "step": 25942 }, { "epoch": 1.6104661990191818, "grad_norm": 0.6106120293839539, "learning_rate": 5.229972928275215e-05, "loss": 2.7285, "step": 25943 }, { "epoch": 1.6105282761189397, "grad_norm": 0.6508684665446526, "learning_rate": 5.2296121574691925e-05, "loss": 2.7793, "step": 25944 }, { "epoch": 1.6105903532186976, "grad_norm": 0.6344076467673133, "learning_rate": 5.229251385465229e-05, "loss": 2.7895, "step": 25945 }, { "epoch": 1.6106524303184555, "grad_norm": 0.5582728497876537, "learning_rate": 5.228890612265205e-05, "loss": 2.7629, "step": 25946 }, { "epoch": 1.6107145074182134, "grad_norm": 0.6537774828573506, "learning_rate": 5.228529837871004e-05, "loss": 2.774, "step": 25947 }, { "epoch": 1.6107765845179713, "grad_norm": 0.5575073989240336, "learning_rate": 5.228169062284505e-05, "loss": 2.7222, "step": 25948 }, { "epoch": 1.6108386616177293, "grad_norm": 0.6315924574793608, "learning_rate": 5.227808285507594e-05, "loss": 2.7506, "step": 25949 }, { "epoch": 1.6109007387174872, "grad_norm": 0.5575219923223345, "learning_rate": 5.2274475075421515e-05, "loss": 2.8281, "step": 25950 }, { "epoch": 1.610962815817245, "grad_norm": 0.5451352648469475, "learning_rate": 5.227086728390059e-05, "loss": 2.777, "step": 25951 }, { "epoch": 1.611024892917003, "grad_norm": 0.6034143603652706, "learning_rate": 5.226725948053202e-05, "loss": 2.7431, "step": 25952 }, { "epoch": 1.611086970016761, "grad_norm": 0.5809842876453445, "learning_rate": 5.226365166533459e-05, "loss": 2.7936, "step": 25953 }, { "epoch": 1.6111490471165189, "grad_norm": 0.6412297395104397, "learning_rate": 5.226004383832716e-05, "loss": 2.7482, "step": 25954 }, { "epoch": 1.6112111242162768, "grad_norm": 0.5366065428603299, "learning_rate": 5.225643599952851e-05, "loss": 2.7394, "step": 25955 }, { "epoch": 1.6112732013160345, "grad_norm": 0.591146709984845, "learning_rate": 5.22528281489575e-05, "loss": 2.6655, "step": 25956 }, { "epoch": 1.6113352784157924, "grad_norm": 0.5425387539183966, "learning_rate": 5.224922028663295e-05, "loss": 2.8025, "step": 25957 }, { "epoch": 1.6113973555155503, "grad_norm": 0.6336454161099585, "learning_rate": 5.2245612412573654e-05, "loss": 2.801, "step": 25958 }, { "epoch": 1.6114594326153082, "grad_norm": 0.5619739350743295, "learning_rate": 5.2242004526798464e-05, "loss": 2.8741, "step": 25959 }, { "epoch": 1.6115215097150661, "grad_norm": 0.5446675298359839, "learning_rate": 5.223839662932617e-05, "loss": 2.7175, "step": 25960 }, { "epoch": 1.611583586814824, "grad_norm": 0.5596269242707537, "learning_rate": 5.2234788720175646e-05, "loss": 2.7584, "step": 25961 }, { "epoch": 1.6116456639145817, "grad_norm": 0.5164059879919821, "learning_rate": 5.223118079936567e-05, "loss": 2.7261, "step": 25962 }, { "epoch": 1.6117077410143397, "grad_norm": 0.5597584193551864, "learning_rate": 5.222757286691508e-05, "loss": 2.8407, "step": 25963 }, { "epoch": 1.6117698181140976, "grad_norm": 0.5220921721272193, "learning_rate": 5.2223964922842696e-05, "loss": 2.8488, "step": 25964 }, { "epoch": 1.6118318952138555, "grad_norm": 0.562182310248992, "learning_rate": 5.222035696716736e-05, "loss": 2.801, "step": 25965 }, { "epoch": 1.6118939723136134, "grad_norm": 0.5757956618183692, "learning_rate": 5.221674899990788e-05, "loss": 2.7462, "step": 25966 }, { "epoch": 1.6119560494133713, "grad_norm": 0.5361940387167506, "learning_rate": 5.221314102108308e-05, "loss": 2.7204, "step": 25967 }, { "epoch": 1.6120181265131293, "grad_norm": 0.5453310015143943, "learning_rate": 5.220953303071179e-05, "loss": 2.7906, "step": 25968 }, { "epoch": 1.6120802036128872, "grad_norm": 0.5370442298916583, "learning_rate": 5.220592502881282e-05, "loss": 2.8139, "step": 25969 }, { "epoch": 1.612142280712645, "grad_norm": 0.5548573623401151, "learning_rate": 5.220231701540502e-05, "loss": 2.7624, "step": 25970 }, { "epoch": 1.612204357812403, "grad_norm": 0.5710077359281767, "learning_rate": 5.219870899050718e-05, "loss": 2.7089, "step": 25971 }, { "epoch": 1.612266434912161, "grad_norm": 0.521080088972522, "learning_rate": 5.219510095413817e-05, "loss": 2.6651, "step": 25972 }, { "epoch": 1.6123285120119188, "grad_norm": 0.6202431378152171, "learning_rate": 5.219149290631675e-05, "loss": 2.732, "step": 25973 }, { "epoch": 1.6123905891116768, "grad_norm": 0.5333626197681598, "learning_rate": 5.2187884847061806e-05, "loss": 2.6865, "step": 25974 }, { "epoch": 1.6124526662114347, "grad_norm": 0.7110811025609178, "learning_rate": 5.218427677639213e-05, "loss": 2.834, "step": 25975 }, { "epoch": 1.6125147433111926, "grad_norm": 0.5488034443040287, "learning_rate": 5.218066869432654e-05, "loss": 2.7446, "step": 25976 }, { "epoch": 1.6125768204109505, "grad_norm": 0.7017239917872781, "learning_rate": 5.217706060088389e-05, "loss": 2.7878, "step": 25977 }, { "epoch": 1.6126388975107084, "grad_norm": 0.5531215053383297, "learning_rate": 5.2173452496082965e-05, "loss": 2.6857, "step": 25978 }, { "epoch": 1.6127009746104664, "grad_norm": 0.6248365665374864, "learning_rate": 5.216984437994262e-05, "loss": 2.7351, "step": 25979 }, { "epoch": 1.612763051710224, "grad_norm": 0.5531685946682743, "learning_rate": 5.216623625248167e-05, "loss": 2.7778, "step": 25980 }, { "epoch": 1.612825128809982, "grad_norm": 0.6374266731875506, "learning_rate": 5.216262811371895e-05, "loss": 2.7995, "step": 25981 }, { "epoch": 1.6128872059097399, "grad_norm": 0.5888997791690836, "learning_rate": 5.215901996367326e-05, "loss": 2.7988, "step": 25982 }, { "epoch": 1.6129492830094978, "grad_norm": 0.6100850736593828, "learning_rate": 5.215541180236343e-05, "loss": 2.7807, "step": 25983 }, { "epoch": 1.6130113601092557, "grad_norm": 0.5482506751696895, "learning_rate": 5.215180362980832e-05, "loss": 2.6951, "step": 25984 }, { "epoch": 1.6130734372090136, "grad_norm": 0.5605798904319269, "learning_rate": 5.21481954460267e-05, "loss": 2.6714, "step": 25985 }, { "epoch": 1.6131355143087713, "grad_norm": 0.5599540473186763, "learning_rate": 5.2144587251037445e-05, "loss": 2.7268, "step": 25986 }, { "epoch": 1.6131975914085293, "grad_norm": 0.6197032357823151, "learning_rate": 5.214097904485935e-05, "loss": 2.6886, "step": 25987 }, { "epoch": 1.6132596685082872, "grad_norm": 0.5637401314778817, "learning_rate": 5.2137370827511255e-05, "loss": 2.7407, "step": 25988 }, { "epoch": 1.613321745608045, "grad_norm": 0.5903980606295359, "learning_rate": 5.213376259901197e-05, "loss": 2.6493, "step": 25989 }, { "epoch": 1.613383822707803, "grad_norm": 0.604135509761839, "learning_rate": 5.213015435938034e-05, "loss": 2.6908, "step": 25990 }, { "epoch": 1.613445899807561, "grad_norm": 0.5959068575313895, "learning_rate": 5.2126546108635156e-05, "loss": 2.7917, "step": 25991 }, { "epoch": 1.6135079769073188, "grad_norm": 0.557158647390325, "learning_rate": 5.212293784679527e-05, "loss": 2.7544, "step": 25992 }, { "epoch": 1.6135700540070768, "grad_norm": 0.5572185527815499, "learning_rate": 5.211932957387951e-05, "loss": 2.7368, "step": 25993 }, { "epoch": 1.6136321311068347, "grad_norm": 0.5930333832306764, "learning_rate": 5.211572128990669e-05, "loss": 2.7437, "step": 25994 }, { "epoch": 1.6136942082065926, "grad_norm": 0.5669446556675092, "learning_rate": 5.2112112994895645e-05, "loss": 2.7928, "step": 25995 }, { "epoch": 1.6137562853063505, "grad_norm": 0.6331540351679444, "learning_rate": 5.210850468886518e-05, "loss": 2.8198, "step": 25996 }, { "epoch": 1.6138183624061084, "grad_norm": 0.6177358002584101, "learning_rate": 5.210489637183415e-05, "loss": 2.841, "step": 25997 }, { "epoch": 1.6138804395058663, "grad_norm": 0.5950172603331567, "learning_rate": 5.210128804382135e-05, "loss": 2.7438, "step": 25998 }, { "epoch": 1.6139425166056243, "grad_norm": 0.5646846256911269, "learning_rate": 5.2097679704845616e-05, "loss": 2.7756, "step": 25999 }, { "epoch": 1.6140045937053822, "grad_norm": 0.5871050328787657, "learning_rate": 5.20940713549258e-05, "loss": 2.7898, "step": 26000 }, { "epoch": 1.61406667080514, "grad_norm": 0.6270735913108196, "learning_rate": 5.209046299408068e-05, "loss": 2.8332, "step": 26001 }, { "epoch": 1.614128747904898, "grad_norm": 0.5354387693135184, "learning_rate": 5.208685462232913e-05, "loss": 2.6867, "step": 26002 }, { "epoch": 1.614190825004656, "grad_norm": 0.5909396468778654, "learning_rate": 5.208324623968993e-05, "loss": 2.7097, "step": 26003 }, { "epoch": 1.6142529021044136, "grad_norm": 0.5231163975818445, "learning_rate": 5.207963784618195e-05, "loss": 2.6775, "step": 26004 }, { "epoch": 1.6143149792041716, "grad_norm": 0.5776281283619346, "learning_rate": 5.207602944182398e-05, "loss": 2.7491, "step": 26005 }, { "epoch": 1.6143770563039295, "grad_norm": 0.5481412940827687, "learning_rate": 5.2072421026634874e-05, "loss": 2.7244, "step": 26006 }, { "epoch": 1.6144391334036874, "grad_norm": 0.5413457090759326, "learning_rate": 5.206881260063343e-05, "loss": 2.7866, "step": 26007 }, { "epoch": 1.6145012105034453, "grad_norm": 0.5700972686560294, "learning_rate": 5.206520416383851e-05, "loss": 2.7261, "step": 26008 }, { "epoch": 1.6145632876032032, "grad_norm": 0.5691021656873938, "learning_rate": 5.20615957162689e-05, "loss": 2.747, "step": 26009 }, { "epoch": 1.614625364702961, "grad_norm": 0.5476261429939105, "learning_rate": 5.2057987257943453e-05, "loss": 2.8492, "step": 26010 }, { "epoch": 1.6146874418027188, "grad_norm": 0.5910375778851037, "learning_rate": 5.2054378788880985e-05, "loss": 2.7054, "step": 26011 }, { "epoch": 1.6147495189024768, "grad_norm": 0.5138429422143095, "learning_rate": 5.205077030910032e-05, "loss": 2.803, "step": 26012 }, { "epoch": 1.6148115960022347, "grad_norm": 0.6401968665130019, "learning_rate": 5.2047161818620296e-05, "loss": 2.7659, "step": 26013 }, { "epoch": 1.6148736731019926, "grad_norm": 0.5620835405023682, "learning_rate": 5.204355331745972e-05, "loss": 2.7844, "step": 26014 }, { "epoch": 1.6149357502017505, "grad_norm": 0.6167249032393279, "learning_rate": 5.203994480563743e-05, "loss": 2.8492, "step": 26015 }, { "epoch": 1.6149978273015084, "grad_norm": 0.5664790928796526, "learning_rate": 5.203633628317225e-05, "loss": 2.8233, "step": 26016 }, { "epoch": 1.6150599044012663, "grad_norm": 0.5758002941858932, "learning_rate": 5.203272775008302e-05, "loss": 2.7748, "step": 26017 }, { "epoch": 1.6151219815010243, "grad_norm": 0.5309209926794023, "learning_rate": 5.202911920638857e-05, "loss": 2.8169, "step": 26018 }, { "epoch": 1.6151840586007822, "grad_norm": 0.6573464226220678, "learning_rate": 5.202551065210768e-05, "loss": 2.7939, "step": 26019 }, { "epoch": 1.61524613570054, "grad_norm": 0.5425971152138576, "learning_rate": 5.2021902087259237e-05, "loss": 2.7907, "step": 26020 }, { "epoch": 1.615308212800298, "grad_norm": 0.6151275542031188, "learning_rate": 5.2018293511862014e-05, "loss": 2.7359, "step": 26021 }, { "epoch": 1.615370289900056, "grad_norm": 0.5350220783526588, "learning_rate": 5.2014684925934884e-05, "loss": 2.7247, "step": 26022 }, { "epoch": 1.6154323669998139, "grad_norm": 0.5668685341984646, "learning_rate": 5.2011076329496645e-05, "loss": 2.7812, "step": 26023 }, { "epoch": 1.6154944440995718, "grad_norm": 0.5245702926382353, "learning_rate": 5.200746772256615e-05, "loss": 2.6342, "step": 26024 }, { "epoch": 1.6155565211993297, "grad_norm": 0.5429318641922515, "learning_rate": 5.200385910516219e-05, "loss": 2.7516, "step": 26025 }, { "epoch": 1.6156185982990876, "grad_norm": 0.5955231191357613, "learning_rate": 5.2000250477303615e-05, "loss": 2.754, "step": 26026 }, { "epoch": 1.6156806753988455, "grad_norm": 0.5137733678306518, "learning_rate": 5.199664183900924e-05, "loss": 2.8465, "step": 26027 }, { "epoch": 1.6157427524986032, "grad_norm": 0.5606177414814405, "learning_rate": 5.199303319029791e-05, "loss": 2.745, "step": 26028 }, { "epoch": 1.6158048295983611, "grad_norm": 0.4757437630152844, "learning_rate": 5.198942453118845e-05, "loss": 2.697, "step": 26029 }, { "epoch": 1.615866906698119, "grad_norm": 0.545139201043344, "learning_rate": 5.198581586169966e-05, "loss": 2.779, "step": 26030 }, { "epoch": 1.615928983797877, "grad_norm": 0.5759630762424811, "learning_rate": 5.198220718185041e-05, "loss": 2.855, "step": 26031 }, { "epoch": 1.615991060897635, "grad_norm": 0.495458752192327, "learning_rate": 5.1978598491659466e-05, "loss": 2.7842, "step": 26032 }, { "epoch": 1.6160531379973928, "grad_norm": 0.5715476209018833, "learning_rate": 5.197498979114572e-05, "loss": 2.7396, "step": 26033 }, { "epoch": 1.6161152150971505, "grad_norm": 0.4995542715075762, "learning_rate": 5.1971381080327966e-05, "loss": 2.7179, "step": 26034 }, { "epoch": 1.6161772921969084, "grad_norm": 0.5644904259629205, "learning_rate": 5.196777235922505e-05, "loss": 2.7199, "step": 26035 }, { "epoch": 1.6162393692966663, "grad_norm": 0.5075528581462313, "learning_rate": 5.196416362785578e-05, "loss": 2.7366, "step": 26036 }, { "epoch": 1.6163014463964243, "grad_norm": 0.5327651632094222, "learning_rate": 5.1960554886238985e-05, "loss": 2.7246, "step": 26037 }, { "epoch": 1.6163635234961822, "grad_norm": 0.5152282633502103, "learning_rate": 5.195694613439352e-05, "loss": 2.7766, "step": 26038 }, { "epoch": 1.61642560059594, "grad_norm": 0.577520182987609, "learning_rate": 5.1953337372338164e-05, "loss": 2.781, "step": 26039 }, { "epoch": 1.616487677695698, "grad_norm": 0.5634695280749983, "learning_rate": 5.19497286000918e-05, "loss": 2.8044, "step": 26040 }, { "epoch": 1.616549754795456, "grad_norm": 0.5313048895360485, "learning_rate": 5.19461198176732e-05, "loss": 2.6649, "step": 26041 }, { "epoch": 1.6166118318952138, "grad_norm": 0.5610877959387767, "learning_rate": 5.194251102510125e-05, "loss": 2.7639, "step": 26042 }, { "epoch": 1.6166739089949718, "grad_norm": 0.5287053085720831, "learning_rate": 5.1938902222394725e-05, "loss": 2.7628, "step": 26043 }, { "epoch": 1.6167359860947297, "grad_norm": 0.5662304835714099, "learning_rate": 5.193529340957248e-05, "loss": 2.7575, "step": 26044 }, { "epoch": 1.6167980631944876, "grad_norm": 0.5044807293957169, "learning_rate": 5.193168458665335e-05, "loss": 2.7781, "step": 26045 }, { "epoch": 1.6168601402942455, "grad_norm": 0.6344969195388074, "learning_rate": 5.1928075753656144e-05, "loss": 2.8435, "step": 26046 }, { "epoch": 1.6169222173940034, "grad_norm": 0.5348356358555941, "learning_rate": 5.1924466910599715e-05, "loss": 2.7582, "step": 26047 }, { "epoch": 1.6169842944937614, "grad_norm": 0.6418849835831226, "learning_rate": 5.192085805750284e-05, "loss": 2.83, "step": 26048 }, { "epoch": 1.6170463715935193, "grad_norm": 0.6063656716259291, "learning_rate": 5.191724919438441e-05, "loss": 2.7723, "step": 26049 }, { "epoch": 1.6171084486932772, "grad_norm": 0.5594090733726192, "learning_rate": 5.191364032126321e-05, "loss": 2.744, "step": 26050 }, { "epoch": 1.617170525793035, "grad_norm": 0.5904767006679241, "learning_rate": 5.19100314381581e-05, "loss": 2.7483, "step": 26051 }, { "epoch": 1.6172326028927928, "grad_norm": 0.547841716276688, "learning_rate": 5.190642254508788e-05, "loss": 2.7775, "step": 26052 }, { "epoch": 1.6172946799925507, "grad_norm": 0.5624136328619854, "learning_rate": 5.1902813642071414e-05, "loss": 2.7841, "step": 26053 }, { "epoch": 1.6173567570923086, "grad_norm": 0.5629834640574752, "learning_rate": 5.1899204729127496e-05, "loss": 2.7742, "step": 26054 }, { "epoch": 1.6174188341920666, "grad_norm": 0.5798689596770613, "learning_rate": 5.1895595806274946e-05, "loss": 2.8058, "step": 26055 }, { "epoch": 1.6174809112918245, "grad_norm": 0.5630534945697366, "learning_rate": 5.189198687353263e-05, "loss": 2.7836, "step": 26056 }, { "epoch": 1.6175429883915824, "grad_norm": 0.5993151113525047, "learning_rate": 5.188837793091935e-05, "loss": 2.8072, "step": 26057 }, { "epoch": 1.61760506549134, "grad_norm": 0.5751029233453244, "learning_rate": 5.188476897845396e-05, "loss": 2.8054, "step": 26058 }, { "epoch": 1.617667142591098, "grad_norm": 0.625448233091393, "learning_rate": 5.188116001615525e-05, "loss": 2.8314, "step": 26059 }, { "epoch": 1.617729219690856, "grad_norm": 0.5984091161864153, "learning_rate": 5.1877551044042093e-05, "loss": 2.8353, "step": 26060 }, { "epoch": 1.6177912967906138, "grad_norm": 0.5918263846615863, "learning_rate": 5.187394206213328e-05, "loss": 2.8217, "step": 26061 }, { "epoch": 1.6178533738903718, "grad_norm": 0.5828023577260179, "learning_rate": 5.187033307044768e-05, "loss": 2.7768, "step": 26062 }, { "epoch": 1.6179154509901297, "grad_norm": 0.5742863985034186, "learning_rate": 5.186672406900407e-05, "loss": 2.768, "step": 26063 }, { "epoch": 1.6179775280898876, "grad_norm": 0.62835014825004, "learning_rate": 5.186311505782132e-05, "loss": 2.7992, "step": 26064 }, { "epoch": 1.6180396051896455, "grad_norm": 0.5567212230489331, "learning_rate": 5.185950603691824e-05, "loss": 2.7176, "step": 26065 }, { "epoch": 1.6181016822894034, "grad_norm": 0.58544392460075, "learning_rate": 5.185589700631367e-05, "loss": 2.728, "step": 26066 }, { "epoch": 1.6181637593891613, "grad_norm": 0.5448063553628482, "learning_rate": 5.1852287966026436e-05, "loss": 2.6929, "step": 26067 }, { "epoch": 1.6182258364889193, "grad_norm": 0.5621902354641958, "learning_rate": 5.1848678916075365e-05, "loss": 2.7503, "step": 26068 }, { "epoch": 1.6182879135886772, "grad_norm": 0.5569376370602451, "learning_rate": 5.184506985647929e-05, "loss": 2.7662, "step": 26069 }, { "epoch": 1.618349990688435, "grad_norm": 0.5255579912289631, "learning_rate": 5.184146078725703e-05, "loss": 2.8084, "step": 26070 }, { "epoch": 1.618412067788193, "grad_norm": 0.6103580566060766, "learning_rate": 5.183785170842743e-05, "loss": 2.678, "step": 26071 }, { "epoch": 1.618474144887951, "grad_norm": 0.5345304129292062, "learning_rate": 5.183424262000932e-05, "loss": 2.8756, "step": 26072 }, { "epoch": 1.6185362219877089, "grad_norm": 0.6096649213840302, "learning_rate": 5.18306335220215e-05, "loss": 2.6629, "step": 26073 }, { "epoch": 1.6185982990874668, "grad_norm": 0.544430966221238, "learning_rate": 5.1827024414482836e-05, "loss": 2.7455, "step": 26074 }, { "epoch": 1.6186603761872247, "grad_norm": 0.6016286590658291, "learning_rate": 5.1823415297412137e-05, "loss": 2.7337, "step": 26075 }, { "epoch": 1.6187224532869824, "grad_norm": 0.6104204062736801, "learning_rate": 5.181980617082825e-05, "loss": 2.7009, "step": 26076 }, { "epoch": 1.6187845303867403, "grad_norm": 0.6032030598621014, "learning_rate": 5.1816197034749966e-05, "loss": 2.8548, "step": 26077 }, { "epoch": 1.6188466074864982, "grad_norm": 0.6366836444561623, "learning_rate": 5.1812587889196164e-05, "loss": 2.7718, "step": 26078 }, { "epoch": 1.6189086845862561, "grad_norm": 0.5968074880236321, "learning_rate": 5.180897873418563e-05, "loss": 2.7755, "step": 26079 }, { "epoch": 1.618970761686014, "grad_norm": 0.6051589173076697, "learning_rate": 5.1805369569737236e-05, "loss": 2.7772, "step": 26080 }, { "epoch": 1.619032838785772, "grad_norm": 0.5755184282853865, "learning_rate": 5.180176039586978e-05, "loss": 2.7329, "step": 26081 }, { "epoch": 1.6190949158855297, "grad_norm": 0.5947204739864992, "learning_rate": 5.17981512126021e-05, "loss": 2.8263, "step": 26082 }, { "epoch": 1.6191569929852876, "grad_norm": 0.5696834857859929, "learning_rate": 5.179454201995303e-05, "loss": 2.8175, "step": 26083 }, { "epoch": 1.6192190700850455, "grad_norm": 0.5858981055685976, "learning_rate": 5.17909328179414e-05, "loss": 2.8068, "step": 26084 }, { "epoch": 1.6192811471848034, "grad_norm": 0.5474222986832954, "learning_rate": 5.1787323606586044e-05, "loss": 2.7375, "step": 26085 }, { "epoch": 1.6193432242845613, "grad_norm": 0.5613873126018167, "learning_rate": 5.178371438590578e-05, "loss": 2.7581, "step": 26086 }, { "epoch": 1.6194053013843193, "grad_norm": 0.5712500143155487, "learning_rate": 5.1780105155919456e-05, "loss": 2.7369, "step": 26087 }, { "epoch": 1.6194673784840772, "grad_norm": 0.5613540526622554, "learning_rate": 5.177649591664587e-05, "loss": 2.8451, "step": 26088 }, { "epoch": 1.619529455583835, "grad_norm": 0.5608784562172957, "learning_rate": 5.1772886668103893e-05, "loss": 2.8266, "step": 26089 }, { "epoch": 1.619591532683593, "grad_norm": 0.5588669275661959, "learning_rate": 5.176927741031233e-05, "loss": 2.6684, "step": 26090 }, { "epoch": 1.619653609783351, "grad_norm": 0.514901752127924, "learning_rate": 5.1765668143290027e-05, "loss": 2.7443, "step": 26091 }, { "epoch": 1.6197156868831089, "grad_norm": 0.5885942626902284, "learning_rate": 5.17620588670558e-05, "loss": 2.6741, "step": 26092 }, { "epoch": 1.6197777639828668, "grad_norm": 0.48925466022503866, "learning_rate": 5.1758449581628465e-05, "loss": 2.8006, "step": 26093 }, { "epoch": 1.6198398410826247, "grad_norm": 0.5921828770519972, "learning_rate": 5.1754840287026896e-05, "loss": 2.7618, "step": 26094 }, { "epoch": 1.6199019181823826, "grad_norm": 0.5116148107834261, "learning_rate": 5.1751230983269885e-05, "loss": 2.8234, "step": 26095 }, { "epoch": 1.6199639952821405, "grad_norm": 0.5683058618476474, "learning_rate": 5.174762167037628e-05, "loss": 2.7166, "step": 26096 }, { "epoch": 1.6200260723818984, "grad_norm": 0.5907883060077338, "learning_rate": 5.17440123483649e-05, "loss": 2.8292, "step": 26097 }, { "epoch": 1.6200881494816564, "grad_norm": 0.5874430578110745, "learning_rate": 5.174040301725459e-05, "loss": 2.7692, "step": 26098 }, { "epoch": 1.6201502265814143, "grad_norm": 0.5736751448558748, "learning_rate": 5.173679367706419e-05, "loss": 2.7855, "step": 26099 }, { "epoch": 1.620212303681172, "grad_norm": 0.5757496466691477, "learning_rate": 5.173318432781249e-05, "loss": 2.7947, "step": 26100 }, { "epoch": 1.62027438078093, "grad_norm": 0.5761360545257592, "learning_rate": 5.1729574969518366e-05, "loss": 2.8517, "step": 26101 }, { "epoch": 1.6203364578806878, "grad_norm": 0.5920039906673342, "learning_rate": 5.172596560220062e-05, "loss": 2.7658, "step": 26102 }, { "epoch": 1.6203985349804457, "grad_norm": 0.5466631070123863, "learning_rate": 5.172235622587811e-05, "loss": 2.7702, "step": 26103 }, { "epoch": 1.6204606120802036, "grad_norm": 0.49788338464835363, "learning_rate": 5.1718746840569634e-05, "loss": 2.7198, "step": 26104 }, { "epoch": 1.6205226891799616, "grad_norm": 0.5741637845882874, "learning_rate": 5.171513744629404e-05, "loss": 2.8005, "step": 26105 }, { "epoch": 1.6205847662797193, "grad_norm": 0.5181115910045452, "learning_rate": 5.171152804307016e-05, "loss": 2.7245, "step": 26106 }, { "epoch": 1.6206468433794772, "grad_norm": 0.5682858816130645, "learning_rate": 5.170791863091682e-05, "loss": 2.8038, "step": 26107 }, { "epoch": 1.620708920479235, "grad_norm": 0.4945759420102374, "learning_rate": 5.170430920985285e-05, "loss": 2.6904, "step": 26108 }, { "epoch": 1.620770997578993, "grad_norm": 0.5888961607730976, "learning_rate": 5.170069977989709e-05, "loss": 2.8333, "step": 26109 }, { "epoch": 1.620833074678751, "grad_norm": 0.5831236348846582, "learning_rate": 5.1697090341068376e-05, "loss": 2.7865, "step": 26110 }, { "epoch": 1.6208951517785088, "grad_norm": 0.5954469473898056, "learning_rate": 5.169348089338552e-05, "loss": 2.8107, "step": 26111 }, { "epoch": 1.6209572288782668, "grad_norm": 0.5947522438154511, "learning_rate": 5.168987143686737e-05, "loss": 2.7645, "step": 26112 }, { "epoch": 1.6210193059780247, "grad_norm": 0.5888189039790116, "learning_rate": 5.168626197153274e-05, "loss": 2.8176, "step": 26113 }, { "epoch": 1.6210813830777826, "grad_norm": 0.6635368510679869, "learning_rate": 5.168265249740046e-05, "loss": 2.7707, "step": 26114 }, { "epoch": 1.6211434601775405, "grad_norm": 0.580968795967603, "learning_rate": 5.1679043014489395e-05, "loss": 2.7316, "step": 26115 }, { "epoch": 1.6212055372772984, "grad_norm": 0.6740257747303431, "learning_rate": 5.167543352281834e-05, "loss": 2.8058, "step": 26116 }, { "epoch": 1.6212676143770564, "grad_norm": 0.5656538602937509, "learning_rate": 5.167182402240616e-05, "loss": 2.7769, "step": 26117 }, { "epoch": 1.6213296914768143, "grad_norm": 0.5478677186266678, "learning_rate": 5.166821451327165e-05, "loss": 2.7808, "step": 26118 }, { "epoch": 1.6213917685765722, "grad_norm": 0.5498518982804115, "learning_rate": 5.166460499543367e-05, "loss": 2.7582, "step": 26119 }, { "epoch": 1.62145384567633, "grad_norm": 0.557837901886072, "learning_rate": 5.166099546891104e-05, "loss": 2.6552, "step": 26120 }, { "epoch": 1.621515922776088, "grad_norm": 0.5338137138360938, "learning_rate": 5.16573859337226e-05, "loss": 2.7584, "step": 26121 }, { "epoch": 1.621577999875846, "grad_norm": 0.5802511605756955, "learning_rate": 5.165377638988716e-05, "loss": 2.7609, "step": 26122 }, { "epoch": 1.6216400769756039, "grad_norm": 0.5418108662598146, "learning_rate": 5.1650166837423585e-05, "loss": 2.8332, "step": 26123 }, { "epoch": 1.6217021540753616, "grad_norm": 0.5572893077289414, "learning_rate": 5.164655727635067e-05, "loss": 2.8053, "step": 26124 }, { "epoch": 1.6217642311751195, "grad_norm": 0.5481302025351259, "learning_rate": 5.164294770668728e-05, "loss": 2.7009, "step": 26125 }, { "epoch": 1.6218263082748774, "grad_norm": 0.5727845530916547, "learning_rate": 5.163933812845223e-05, "loss": 2.8215, "step": 26126 }, { "epoch": 1.6218883853746353, "grad_norm": 0.6028779744472919, "learning_rate": 5.1635728541664364e-05, "loss": 2.7132, "step": 26127 }, { "epoch": 1.6219504624743932, "grad_norm": 0.5009962825351696, "learning_rate": 5.16321189463425e-05, "loss": 2.6537, "step": 26128 }, { "epoch": 1.6220125395741511, "grad_norm": 0.5301349100663127, "learning_rate": 5.162850934250546e-05, "loss": 2.7492, "step": 26129 }, { "epoch": 1.6220746166739088, "grad_norm": 0.5941210390533158, "learning_rate": 5.162489973017209e-05, "loss": 2.6956, "step": 26130 }, { "epoch": 1.6221366937736668, "grad_norm": 0.6148612450094312, "learning_rate": 5.162129010936123e-05, "loss": 2.7778, "step": 26131 }, { "epoch": 1.6221987708734247, "grad_norm": 0.5108669267263393, "learning_rate": 5.1617680480091726e-05, "loss": 2.6962, "step": 26132 }, { "epoch": 1.6222608479731826, "grad_norm": 0.5774467458982915, "learning_rate": 5.1614070842382366e-05, "loss": 2.7449, "step": 26133 }, { "epoch": 1.6223229250729405, "grad_norm": 0.5476894852897082, "learning_rate": 5.161046119625202e-05, "loss": 2.7303, "step": 26134 }, { "epoch": 1.6223850021726984, "grad_norm": 0.5841554372046451, "learning_rate": 5.1606851541719516e-05, "loss": 2.7311, "step": 26135 }, { "epoch": 1.6224470792724563, "grad_norm": 0.5590963436776779, "learning_rate": 5.1603241878803655e-05, "loss": 2.8235, "step": 26136 }, { "epoch": 1.6225091563722143, "grad_norm": 0.5484926820908418, "learning_rate": 5.159963220752331e-05, "loss": 2.7636, "step": 26137 }, { "epoch": 1.6225712334719722, "grad_norm": 0.5823715290540111, "learning_rate": 5.159602252789728e-05, "loss": 2.6162, "step": 26138 }, { "epoch": 1.62263331057173, "grad_norm": 0.55113428043054, "learning_rate": 5.1592412839944424e-05, "loss": 2.8002, "step": 26139 }, { "epoch": 1.622695387671488, "grad_norm": 0.5302987558070806, "learning_rate": 5.158880314368355e-05, "loss": 2.7543, "step": 26140 }, { "epoch": 1.622757464771246, "grad_norm": 0.604311692710256, "learning_rate": 5.158519343913353e-05, "loss": 2.7869, "step": 26141 }, { "epoch": 1.6228195418710039, "grad_norm": 0.5490983725278263, "learning_rate": 5.158158372631314e-05, "loss": 2.7181, "step": 26142 }, { "epoch": 1.6228816189707618, "grad_norm": 0.6493040981265035, "learning_rate": 5.157797400524127e-05, "loss": 2.8237, "step": 26143 }, { "epoch": 1.6229436960705197, "grad_norm": 0.5670881179039683, "learning_rate": 5.1574364275936704e-05, "loss": 2.7618, "step": 26144 }, { "epoch": 1.6230057731702776, "grad_norm": 0.5684582077037785, "learning_rate": 5.157075453841832e-05, "loss": 2.8059, "step": 26145 }, { "epoch": 1.6230678502700355, "grad_norm": 0.6511608294469543, "learning_rate": 5.156714479270492e-05, "loss": 2.7128, "step": 26146 }, { "epoch": 1.6231299273697934, "grad_norm": 0.514303565969089, "learning_rate": 5.156353503881532e-05, "loss": 2.7419, "step": 26147 }, { "epoch": 1.6231920044695511, "grad_norm": 0.6512546470681304, "learning_rate": 5.1559925276768404e-05, "loss": 2.7834, "step": 26148 }, { "epoch": 1.623254081569309, "grad_norm": 0.5340166697471965, "learning_rate": 5.1556315506582976e-05, "loss": 2.7662, "step": 26149 }, { "epoch": 1.623316158669067, "grad_norm": 0.623558336751605, "learning_rate": 5.155270572827787e-05, "loss": 2.7182, "step": 26150 }, { "epoch": 1.623378235768825, "grad_norm": 0.5741916312736257, "learning_rate": 5.154909594187193e-05, "loss": 2.7494, "step": 26151 }, { "epoch": 1.6234403128685828, "grad_norm": 0.5752971669090688, "learning_rate": 5.154548614738397e-05, "loss": 2.7577, "step": 26152 }, { "epoch": 1.6235023899683407, "grad_norm": 0.6049540146777483, "learning_rate": 5.154187634483284e-05, "loss": 2.7536, "step": 26153 }, { "epoch": 1.6235644670680984, "grad_norm": 0.5334771259252095, "learning_rate": 5.153826653423737e-05, "loss": 2.7305, "step": 26154 }, { "epoch": 1.6236265441678563, "grad_norm": 0.6134634536950594, "learning_rate": 5.153465671561638e-05, "loss": 2.7157, "step": 26155 }, { "epoch": 1.6236886212676143, "grad_norm": 0.5140368829159144, "learning_rate": 5.153104688898871e-05, "loss": 2.7301, "step": 26156 }, { "epoch": 1.6237506983673722, "grad_norm": 0.587590289786928, "learning_rate": 5.152743705437322e-05, "loss": 2.7667, "step": 26157 }, { "epoch": 1.62381277546713, "grad_norm": 0.5528129818290478, "learning_rate": 5.1523827211788686e-05, "loss": 2.7842, "step": 26158 }, { "epoch": 1.623874852566888, "grad_norm": 0.617470205723365, "learning_rate": 5.152021736125401e-05, "loss": 2.8348, "step": 26159 }, { "epoch": 1.623936929666646, "grad_norm": 0.5559672783718559, "learning_rate": 5.151660750278796e-05, "loss": 2.772, "step": 26160 }, { "epoch": 1.6239990067664039, "grad_norm": 0.587883603887359, "learning_rate": 5.151299763640942e-05, "loss": 2.7901, "step": 26161 }, { "epoch": 1.6240610838661618, "grad_norm": 0.5368058057699187, "learning_rate": 5.1509387762137196e-05, "loss": 2.7295, "step": 26162 }, { "epoch": 1.6241231609659197, "grad_norm": 0.5205620599644771, "learning_rate": 5.1505777879990126e-05, "loss": 2.6218, "step": 26163 }, { "epoch": 1.6241852380656776, "grad_norm": 0.6151844307139748, "learning_rate": 5.150216798998706e-05, "loss": 2.7434, "step": 26164 }, { "epoch": 1.6242473151654355, "grad_norm": 0.517262863015834, "learning_rate": 5.149855809214681e-05, "loss": 2.8278, "step": 26165 }, { "epoch": 1.6243093922651934, "grad_norm": 0.5849481750013096, "learning_rate": 5.149494818648822e-05, "loss": 2.7522, "step": 26166 }, { "epoch": 1.6243714693649514, "grad_norm": 0.559536092354264, "learning_rate": 5.1491338273030124e-05, "loss": 2.8343, "step": 26167 }, { "epoch": 1.6244335464647093, "grad_norm": 0.5609139227627394, "learning_rate": 5.1487728351791365e-05, "loss": 2.8666, "step": 26168 }, { "epoch": 1.6244956235644672, "grad_norm": 0.4847528614266643, "learning_rate": 5.148411842279074e-05, "loss": 2.6688, "step": 26169 }, { "epoch": 1.6245577006642251, "grad_norm": 0.6068274972240899, "learning_rate": 5.148050848604713e-05, "loss": 2.7589, "step": 26170 }, { "epoch": 1.624619777763983, "grad_norm": 0.5625508338422955, "learning_rate": 5.147689854157936e-05, "loss": 2.7482, "step": 26171 }, { "epoch": 1.6246818548637407, "grad_norm": 0.6427238554190456, "learning_rate": 5.1473288589406224e-05, "loss": 2.7769, "step": 26172 }, { "epoch": 1.6247439319634986, "grad_norm": 0.5319749709960379, "learning_rate": 5.146967862954659e-05, "loss": 2.6304, "step": 26173 }, { "epoch": 1.6248060090632566, "grad_norm": 0.6418336074171106, "learning_rate": 5.1466068662019285e-05, "loss": 2.7402, "step": 26174 }, { "epoch": 1.6248680861630145, "grad_norm": 0.6497614340207264, "learning_rate": 5.146245868684315e-05, "loss": 2.6936, "step": 26175 }, { "epoch": 1.6249301632627724, "grad_norm": 0.5584205945940447, "learning_rate": 5.145884870403701e-05, "loss": 2.7307, "step": 26176 }, { "epoch": 1.6249922403625303, "grad_norm": 0.6149089376363029, "learning_rate": 5.1455238713619714e-05, "loss": 2.8456, "step": 26177 }, { "epoch": 1.625054317462288, "grad_norm": 0.5662628518471343, "learning_rate": 5.145162871561007e-05, "loss": 2.731, "step": 26178 }, { "epoch": 1.625116394562046, "grad_norm": 0.6046923427732863, "learning_rate": 5.1448018710026915e-05, "loss": 2.7331, "step": 26179 }, { "epoch": 1.6251784716618038, "grad_norm": 0.5329891516080677, "learning_rate": 5.144440869688912e-05, "loss": 2.716, "step": 26180 }, { "epoch": 1.6252405487615618, "grad_norm": 0.6352987310265673, "learning_rate": 5.144079867621547e-05, "loss": 2.6837, "step": 26181 }, { "epoch": 1.6253026258613197, "grad_norm": 0.5640945351504153, "learning_rate": 5.143718864802485e-05, "loss": 2.7891, "step": 26182 }, { "epoch": 1.6253647029610776, "grad_norm": 0.5838088470302227, "learning_rate": 5.143357861233605e-05, "loss": 2.7325, "step": 26183 }, { "epoch": 1.6254267800608355, "grad_norm": 0.5686910127086255, "learning_rate": 5.142996856916793e-05, "loss": 2.7675, "step": 26184 }, { "epoch": 1.6254888571605934, "grad_norm": 0.5930091100333922, "learning_rate": 5.142635851853931e-05, "loss": 2.666, "step": 26185 }, { "epoch": 1.6255509342603514, "grad_norm": 0.6313609441198216, "learning_rate": 5.142274846046905e-05, "loss": 2.7796, "step": 26186 }, { "epoch": 1.6256130113601093, "grad_norm": 0.5775429656819566, "learning_rate": 5.1419138394975943e-05, "loss": 2.7927, "step": 26187 }, { "epoch": 1.6256750884598672, "grad_norm": 0.6219539563456531, "learning_rate": 5.1415528322078856e-05, "loss": 2.7116, "step": 26188 }, { "epoch": 1.625737165559625, "grad_norm": 0.57496159052065, "learning_rate": 5.141191824179662e-05, "loss": 2.7929, "step": 26189 }, { "epoch": 1.625799242659383, "grad_norm": 0.6111230231939948, "learning_rate": 5.140830815414805e-05, "loss": 2.6775, "step": 26190 }, { "epoch": 1.625861319759141, "grad_norm": 0.6136939488550069, "learning_rate": 5.140469805915201e-05, "loss": 2.7306, "step": 26191 }, { "epoch": 1.6259233968588989, "grad_norm": 0.6195962955920682, "learning_rate": 5.140108795682731e-05, "loss": 2.8006, "step": 26192 }, { "epoch": 1.6259854739586568, "grad_norm": 0.6596151086507674, "learning_rate": 5.13974778471928e-05, "loss": 2.7372, "step": 26193 }, { "epoch": 1.6260475510584147, "grad_norm": 0.6338627894328841, "learning_rate": 5.13938677302673e-05, "loss": 2.723, "step": 26194 }, { "epoch": 1.6261096281581726, "grad_norm": 0.6080114318563471, "learning_rate": 5.1390257606069655e-05, "loss": 2.802, "step": 26195 }, { "epoch": 1.6261717052579303, "grad_norm": 0.5897580267653854, "learning_rate": 5.13866474746187e-05, "loss": 2.7214, "step": 26196 }, { "epoch": 1.6262337823576882, "grad_norm": 0.5376931918954931, "learning_rate": 5.138303733593326e-05, "loss": 2.7861, "step": 26197 }, { "epoch": 1.6262958594574461, "grad_norm": 0.5957928199154869, "learning_rate": 5.1379427190032206e-05, "loss": 2.6473, "step": 26198 }, { "epoch": 1.626357936557204, "grad_norm": 0.5267222662994784, "learning_rate": 5.137581703693431e-05, "loss": 2.829, "step": 26199 }, { "epoch": 1.626420013656962, "grad_norm": 0.597123326077829, "learning_rate": 5.1372206876658466e-05, "loss": 2.7545, "step": 26200 }, { "epoch": 1.62648209075672, "grad_norm": 0.5370235305155847, "learning_rate": 5.136859670922348e-05, "loss": 2.7354, "step": 26201 }, { "epoch": 1.6265441678564776, "grad_norm": 0.5893818644733693, "learning_rate": 5.1364986534648194e-05, "loss": 2.7999, "step": 26202 }, { "epoch": 1.6266062449562355, "grad_norm": 0.531499439470672, "learning_rate": 5.136137635295144e-05, "loss": 2.6635, "step": 26203 }, { "epoch": 1.6266683220559934, "grad_norm": 0.6392075511591979, "learning_rate": 5.1357766164152056e-05, "loss": 2.8372, "step": 26204 }, { "epoch": 1.6267303991557513, "grad_norm": 0.5616099946025223, "learning_rate": 5.135415596826887e-05, "loss": 2.7912, "step": 26205 }, { "epoch": 1.6267924762555093, "grad_norm": 0.5950767350603168, "learning_rate": 5.1350545765320735e-05, "loss": 2.6847, "step": 26206 }, { "epoch": 1.6268545533552672, "grad_norm": 0.554306866782246, "learning_rate": 5.134693555532648e-05, "loss": 2.7385, "step": 26207 }, { "epoch": 1.626916630455025, "grad_norm": 0.6208905541422607, "learning_rate": 5.1343325338304914e-05, "loss": 2.7234, "step": 26208 }, { "epoch": 1.626978707554783, "grad_norm": 0.6217127420576612, "learning_rate": 5.133971511427491e-05, "loss": 2.8642, "step": 26209 }, { "epoch": 1.627040784654541, "grad_norm": 0.6067770083090028, "learning_rate": 5.133610488325528e-05, "loss": 2.7506, "step": 26210 }, { "epoch": 1.6271028617542989, "grad_norm": 0.6058110653809011, "learning_rate": 5.1332494645264875e-05, "loss": 2.8067, "step": 26211 }, { "epoch": 1.6271649388540568, "grad_norm": 0.5840749766005554, "learning_rate": 5.1328884400322494e-05, "loss": 2.7451, "step": 26212 }, { "epoch": 1.6272270159538147, "grad_norm": 0.6473442881639708, "learning_rate": 5.132527414844702e-05, "loss": 2.7674, "step": 26213 }, { "epoch": 1.6272890930535726, "grad_norm": 0.5427892099939208, "learning_rate": 5.132166388965728e-05, "loss": 2.7586, "step": 26214 }, { "epoch": 1.6273511701533305, "grad_norm": 0.6234604231365497, "learning_rate": 5.131805362397208e-05, "loss": 2.7497, "step": 26215 }, { "epoch": 1.6274132472530884, "grad_norm": 0.5595452800697216, "learning_rate": 5.131444335141029e-05, "loss": 2.7708, "step": 26216 }, { "epoch": 1.6274753243528464, "grad_norm": 0.6551642973696963, "learning_rate": 5.131083307199071e-05, "loss": 2.8343, "step": 26217 }, { "epoch": 1.6275374014526043, "grad_norm": 0.5598731162023993, "learning_rate": 5.130722278573222e-05, "loss": 2.7557, "step": 26218 }, { "epoch": 1.6275994785523622, "grad_norm": 0.6143803557650487, "learning_rate": 5.130361249265362e-05, "loss": 2.8074, "step": 26219 }, { "epoch": 1.62766155565212, "grad_norm": 0.6007332225795154, "learning_rate": 5.130000219277377e-05, "loss": 2.7816, "step": 26220 }, { "epoch": 1.6277236327518778, "grad_norm": 0.6091879043929994, "learning_rate": 5.129639188611147e-05, "loss": 2.7676, "step": 26221 }, { "epoch": 1.6277857098516357, "grad_norm": 0.6131196507678451, "learning_rate": 5.129278157268559e-05, "loss": 2.7747, "step": 26222 }, { "epoch": 1.6278477869513936, "grad_norm": 0.601748821583867, "learning_rate": 5.1289171252514946e-05, "loss": 2.7795, "step": 26223 }, { "epoch": 1.6279098640511516, "grad_norm": 0.6078999101012775, "learning_rate": 5.12855609256184e-05, "loss": 2.7486, "step": 26224 }, { "epoch": 1.6279719411509093, "grad_norm": 0.6297736258093783, "learning_rate": 5.128195059201477e-05, "loss": 2.7523, "step": 26225 }, { "epoch": 1.6280340182506672, "grad_norm": 0.597578105929007, "learning_rate": 5.127834025172288e-05, "loss": 2.7753, "step": 26226 }, { "epoch": 1.628096095350425, "grad_norm": 0.6486271730181834, "learning_rate": 5.127472990476159e-05, "loss": 2.6823, "step": 26227 }, { "epoch": 1.628158172450183, "grad_norm": 0.6383099011419002, "learning_rate": 5.127111955114969e-05, "loss": 2.8234, "step": 26228 }, { "epoch": 1.628220249549941, "grad_norm": 0.6528500402530162, "learning_rate": 5.126750919090608e-05, "loss": 2.7276, "step": 26229 }, { "epoch": 1.6282823266496989, "grad_norm": 0.5867882973078686, "learning_rate": 5.126389882404956e-05, "loss": 2.7103, "step": 26230 }, { "epoch": 1.6283444037494568, "grad_norm": 0.6375435477697873, "learning_rate": 5.126028845059898e-05, "loss": 2.6528, "step": 26231 }, { "epoch": 1.6284064808492147, "grad_norm": 0.5733520298596627, "learning_rate": 5.1256678070573186e-05, "loss": 2.7611, "step": 26232 }, { "epoch": 1.6284685579489726, "grad_norm": 0.6353756834060623, "learning_rate": 5.1253067683990965e-05, "loss": 2.7875, "step": 26233 }, { "epoch": 1.6285306350487305, "grad_norm": 0.6127410015548951, "learning_rate": 5.12494572908712e-05, "loss": 2.7994, "step": 26234 }, { "epoch": 1.6285927121484884, "grad_norm": 0.5559848603547538, "learning_rate": 5.124584689123271e-05, "loss": 2.8594, "step": 26235 }, { "epoch": 1.6286547892482464, "grad_norm": 0.6188683618179118, "learning_rate": 5.124223648509434e-05, "loss": 2.8139, "step": 26236 }, { "epoch": 1.6287168663480043, "grad_norm": 0.659275709152173, "learning_rate": 5.1238626072474916e-05, "loss": 2.7898, "step": 26237 }, { "epoch": 1.6287789434477622, "grad_norm": 0.5171763173994773, "learning_rate": 5.123501565339328e-05, "loss": 2.698, "step": 26238 }, { "epoch": 1.6288410205475201, "grad_norm": 0.6741811739638127, "learning_rate": 5.123140522786827e-05, "loss": 2.7694, "step": 26239 }, { "epoch": 1.628903097647278, "grad_norm": 0.5382866182322573, "learning_rate": 5.122779479591872e-05, "loss": 2.759, "step": 26240 }, { "epoch": 1.628965174747036, "grad_norm": 0.642541994511048, "learning_rate": 5.122418435756345e-05, "loss": 2.8472, "step": 26241 }, { "epoch": 1.6290272518467939, "grad_norm": 0.6322484504934127, "learning_rate": 5.1220573912821334e-05, "loss": 2.7532, "step": 26242 }, { "epoch": 1.6290893289465516, "grad_norm": 0.6194582103569101, "learning_rate": 5.121696346171119e-05, "loss": 2.8059, "step": 26243 }, { "epoch": 1.6291514060463095, "grad_norm": 0.6627917978667875, "learning_rate": 5.121335300425183e-05, "loss": 2.777, "step": 26244 }, { "epoch": 1.6292134831460674, "grad_norm": 0.63273931009403, "learning_rate": 5.1209742540462125e-05, "loss": 2.7521, "step": 26245 }, { "epoch": 1.6292755602458253, "grad_norm": 0.6454009459616576, "learning_rate": 5.120613207036089e-05, "loss": 2.8072, "step": 26246 }, { "epoch": 1.6293376373455832, "grad_norm": 0.6336142323825092, "learning_rate": 5.120252159396699e-05, "loss": 2.687, "step": 26247 }, { "epoch": 1.6293997144453412, "grad_norm": 0.6006710581431954, "learning_rate": 5.1198911111299234e-05, "loss": 2.7788, "step": 26248 }, { "epoch": 1.6294617915450988, "grad_norm": 0.6563751691702874, "learning_rate": 5.1195300622376466e-05, "loss": 2.8293, "step": 26249 }, { "epoch": 1.6295238686448568, "grad_norm": 0.5589137194621994, "learning_rate": 5.119169012721753e-05, "loss": 2.7444, "step": 26250 }, { "epoch": 1.6295859457446147, "grad_norm": 0.6434831593984158, "learning_rate": 5.118807962584126e-05, "loss": 2.7657, "step": 26251 }, { "epoch": 1.6296480228443726, "grad_norm": 0.5810299395251664, "learning_rate": 5.118446911826647e-05, "loss": 2.8376, "step": 26252 }, { "epoch": 1.6297100999441305, "grad_norm": 0.6255814946660284, "learning_rate": 5.118085860451204e-05, "loss": 2.7469, "step": 26253 }, { "epoch": 1.6297721770438884, "grad_norm": 0.5734093859274586, "learning_rate": 5.117724808459677e-05, "loss": 2.8061, "step": 26254 }, { "epoch": 1.6298342541436464, "grad_norm": 0.5761910645473663, "learning_rate": 5.117363755853951e-05, "loss": 2.7822, "step": 26255 }, { "epoch": 1.6298963312434043, "grad_norm": 0.6601315104346448, "learning_rate": 5.117002702635911e-05, "loss": 2.7392, "step": 26256 }, { "epoch": 1.6299584083431622, "grad_norm": 0.6226014604956962, "learning_rate": 5.1166416488074386e-05, "loss": 2.7422, "step": 26257 }, { "epoch": 1.63002048544292, "grad_norm": 0.5949490135850879, "learning_rate": 5.116280594370419e-05, "loss": 2.8345, "step": 26258 }, { "epoch": 1.630082562542678, "grad_norm": 0.7107718976995272, "learning_rate": 5.1159195393267344e-05, "loss": 2.7546, "step": 26259 }, { "epoch": 1.630144639642436, "grad_norm": 0.6935917209802671, "learning_rate": 5.115558483678271e-05, "loss": 2.7965, "step": 26260 }, { "epoch": 1.6302067167421939, "grad_norm": 0.6170616262980894, "learning_rate": 5.11519742742691e-05, "loss": 2.8198, "step": 26261 }, { "epoch": 1.6302687938419518, "grad_norm": 0.592452413241471, "learning_rate": 5.114836370574535e-05, "loss": 2.6956, "step": 26262 }, { "epoch": 1.6303308709417097, "grad_norm": 0.6204036501545273, "learning_rate": 5.1144753131230317e-05, "loss": 2.7341, "step": 26263 }, { "epoch": 1.6303929480414676, "grad_norm": 0.6030302595925203, "learning_rate": 5.114114255074283e-05, "loss": 2.7819, "step": 26264 }, { "epoch": 1.6304550251412255, "grad_norm": 0.6086095012149424, "learning_rate": 5.113753196430173e-05, "loss": 2.7386, "step": 26265 }, { "epoch": 1.6305171022409835, "grad_norm": 0.5954829575208294, "learning_rate": 5.1133921371925844e-05, "loss": 2.7197, "step": 26266 }, { "epoch": 1.6305791793407411, "grad_norm": 0.5136587046976344, "learning_rate": 5.113031077363401e-05, "loss": 2.8233, "step": 26267 }, { "epoch": 1.630641256440499, "grad_norm": 0.6630108977654903, "learning_rate": 5.1126700169445085e-05, "loss": 2.727, "step": 26268 }, { "epoch": 1.630703333540257, "grad_norm": 0.5361468203856511, "learning_rate": 5.112308955937788e-05, "loss": 2.7642, "step": 26269 }, { "epoch": 1.630765410640015, "grad_norm": 0.661580552551727, "learning_rate": 5.111947894345125e-05, "loss": 2.7392, "step": 26270 }, { "epoch": 1.6308274877397728, "grad_norm": 0.5544889658921333, "learning_rate": 5.111586832168401e-05, "loss": 2.7977, "step": 26271 }, { "epoch": 1.6308895648395307, "grad_norm": 0.5926964346881909, "learning_rate": 5.111225769409505e-05, "loss": 2.7683, "step": 26272 }, { "epoch": 1.6309516419392884, "grad_norm": 0.5524286563690658, "learning_rate": 5.110864706070313e-05, "loss": 2.7746, "step": 26273 }, { "epoch": 1.6310137190390464, "grad_norm": 0.6277465790640572, "learning_rate": 5.110503642152715e-05, "loss": 2.882, "step": 26274 }, { "epoch": 1.6310757961388043, "grad_norm": 0.5933415865413421, "learning_rate": 5.110142577658592e-05, "loss": 2.6961, "step": 26275 }, { "epoch": 1.6311378732385622, "grad_norm": 0.592118254666827, "learning_rate": 5.109781512589829e-05, "loss": 2.8081, "step": 26276 }, { "epoch": 1.63119995033832, "grad_norm": 0.5995718729922274, "learning_rate": 5.1094204469483095e-05, "loss": 2.8047, "step": 26277 }, { "epoch": 1.631262027438078, "grad_norm": 0.6193352282137194, "learning_rate": 5.109059380735917e-05, "loss": 2.6968, "step": 26278 }, { "epoch": 1.631324104537836, "grad_norm": 0.581708370488081, "learning_rate": 5.108698313954534e-05, "loss": 2.7634, "step": 26279 }, { "epoch": 1.6313861816375939, "grad_norm": 0.6426459511437846, "learning_rate": 5.108337246606046e-05, "loss": 2.8342, "step": 26280 }, { "epoch": 1.6314482587373518, "grad_norm": 0.5532707065546172, "learning_rate": 5.1079761786923375e-05, "loss": 2.6759, "step": 26281 }, { "epoch": 1.6315103358371097, "grad_norm": 0.6145919594455681, "learning_rate": 5.107615110215289e-05, "loss": 2.8156, "step": 26282 }, { "epoch": 1.6315724129368676, "grad_norm": 0.5935770837061276, "learning_rate": 5.107254041176789e-05, "loss": 2.8049, "step": 26283 }, { "epoch": 1.6316344900366255, "grad_norm": 0.5893127881060632, "learning_rate": 5.1068929715787165e-05, "loss": 2.7308, "step": 26284 }, { "epoch": 1.6316965671363834, "grad_norm": 0.6127909990864137, "learning_rate": 5.1065319014229583e-05, "loss": 2.8409, "step": 26285 }, { "epoch": 1.6317586442361414, "grad_norm": 0.5589519232734168, "learning_rate": 5.1061708307113975e-05, "loss": 2.746, "step": 26286 }, { "epoch": 1.6318207213358993, "grad_norm": 0.5642570234156525, "learning_rate": 5.1058097594459164e-05, "loss": 2.742, "step": 26287 }, { "epoch": 1.6318827984356572, "grad_norm": 0.5298793445309161, "learning_rate": 5.105448687628402e-05, "loss": 2.8519, "step": 26288 }, { "epoch": 1.6319448755354151, "grad_norm": 0.5985946767586245, "learning_rate": 5.105087615260734e-05, "loss": 2.7824, "step": 26289 }, { "epoch": 1.632006952635173, "grad_norm": 0.5165123384724669, "learning_rate": 5.104726542344801e-05, "loss": 2.7092, "step": 26290 }, { "epoch": 1.6320690297349307, "grad_norm": 0.5419131384587114, "learning_rate": 5.104365468882481e-05, "loss": 2.8336, "step": 26291 }, { "epoch": 1.6321311068346886, "grad_norm": 0.6450222803673792, "learning_rate": 5.104004394875664e-05, "loss": 2.796, "step": 26292 }, { "epoch": 1.6321931839344466, "grad_norm": 0.549396885380632, "learning_rate": 5.103643320326229e-05, "loss": 2.6685, "step": 26293 }, { "epoch": 1.6322552610342045, "grad_norm": 0.7025243100613044, "learning_rate": 5.103282245236062e-05, "loss": 2.7896, "step": 26294 }, { "epoch": 1.6323173381339624, "grad_norm": 0.5891342190256498, "learning_rate": 5.1029211696070466e-05, "loss": 2.7513, "step": 26295 }, { "epoch": 1.6323794152337203, "grad_norm": 0.6341458625572465, "learning_rate": 5.1025600934410665e-05, "loss": 2.7858, "step": 26296 }, { "epoch": 1.632441492333478, "grad_norm": 0.6097995061949335, "learning_rate": 5.1021990167400056e-05, "loss": 2.7037, "step": 26297 }, { "epoch": 1.632503569433236, "grad_norm": 0.574011914636321, "learning_rate": 5.101837939505748e-05, "loss": 2.8015, "step": 26298 }, { "epoch": 1.6325656465329939, "grad_norm": 0.6252306035910099, "learning_rate": 5.1014768617401764e-05, "loss": 2.7404, "step": 26299 }, { "epoch": 1.6326277236327518, "grad_norm": 0.5692691142157179, "learning_rate": 5.101115783445176e-05, "loss": 2.8051, "step": 26300 }, { "epoch": 1.6326898007325097, "grad_norm": 0.5966437495970118, "learning_rate": 5.10075470462263e-05, "loss": 2.7323, "step": 26301 }, { "epoch": 1.6327518778322676, "grad_norm": 0.6135656827059349, "learning_rate": 5.100393625274421e-05, "loss": 2.8261, "step": 26302 }, { "epoch": 1.6328139549320255, "grad_norm": 0.6473987716358518, "learning_rate": 5.100032545402435e-05, "loss": 2.7282, "step": 26303 }, { "epoch": 1.6328760320317834, "grad_norm": 0.6029090919844845, "learning_rate": 5.099671465008555e-05, "loss": 2.7749, "step": 26304 }, { "epoch": 1.6329381091315414, "grad_norm": 0.5635665742325605, "learning_rate": 5.0993103840946656e-05, "loss": 2.7837, "step": 26305 }, { "epoch": 1.6330001862312993, "grad_norm": 0.64017987487715, "learning_rate": 5.09894930266265e-05, "loss": 2.798, "step": 26306 }, { "epoch": 1.6330622633310572, "grad_norm": 0.5681357976309125, "learning_rate": 5.09858822071439e-05, "loss": 2.7727, "step": 26307 }, { "epoch": 1.6331243404308151, "grad_norm": 0.5544947974508737, "learning_rate": 5.098227138251773e-05, "loss": 2.7465, "step": 26308 }, { "epoch": 1.633186417530573, "grad_norm": 0.49953844701828354, "learning_rate": 5.09786605527668e-05, "loss": 2.6932, "step": 26309 }, { "epoch": 1.633248494630331, "grad_norm": 0.5871978429154422, "learning_rate": 5.097504971790995e-05, "loss": 2.731, "step": 26310 }, { "epoch": 1.6333105717300889, "grad_norm": 0.5846556310975116, "learning_rate": 5.097143887796605e-05, "loss": 2.8147, "step": 26311 }, { "epoch": 1.6333726488298468, "grad_norm": 0.5166770721228613, "learning_rate": 5.09678280329539e-05, "loss": 2.7512, "step": 26312 }, { "epoch": 1.6334347259296047, "grad_norm": 0.6202810956947172, "learning_rate": 5.096421718289238e-05, "loss": 2.7083, "step": 26313 }, { "epoch": 1.6334968030293626, "grad_norm": 0.5241174693841433, "learning_rate": 5.0960606327800285e-05, "loss": 2.8275, "step": 26314 }, { "epoch": 1.6335588801291203, "grad_norm": 0.621417171314197, "learning_rate": 5.095699546769648e-05, "loss": 2.7357, "step": 26315 }, { "epoch": 1.6336209572288782, "grad_norm": 0.5626720059515896, "learning_rate": 5.095338460259979e-05, "loss": 2.7968, "step": 26316 }, { "epoch": 1.6336830343286362, "grad_norm": 0.5880733218783042, "learning_rate": 5.094977373252908e-05, "loss": 2.6852, "step": 26317 }, { "epoch": 1.633745111428394, "grad_norm": 0.5893856315827022, "learning_rate": 5.0946162857503155e-05, "loss": 2.7589, "step": 26318 }, { "epoch": 1.633807188528152, "grad_norm": 0.5888093125986201, "learning_rate": 5.094255197754088e-05, "loss": 2.8138, "step": 26319 }, { "epoch": 1.63386926562791, "grad_norm": 0.5741479272049725, "learning_rate": 5.093894109266106e-05, "loss": 2.7593, "step": 26320 }, { "epoch": 1.6339313427276676, "grad_norm": 0.6500710283498811, "learning_rate": 5.093533020288258e-05, "loss": 2.7761, "step": 26321 }, { "epoch": 1.6339934198274255, "grad_norm": 0.5528788563099865, "learning_rate": 5.0931719308224235e-05, "loss": 2.7315, "step": 26322 }, { "epoch": 1.6340554969271834, "grad_norm": 0.6943072649780455, "learning_rate": 5.09281084087049e-05, "loss": 2.6437, "step": 26323 }, { "epoch": 1.6341175740269414, "grad_norm": 0.575901756978257, "learning_rate": 5.09244975043434e-05, "loss": 2.7635, "step": 26324 }, { "epoch": 1.6341796511266993, "grad_norm": 0.6217622098763034, "learning_rate": 5.092088659515857e-05, "loss": 2.729, "step": 26325 }, { "epoch": 1.6342417282264572, "grad_norm": 0.6067179437160907, "learning_rate": 5.091727568116924e-05, "loss": 2.7731, "step": 26326 }, { "epoch": 1.634303805326215, "grad_norm": 0.6356318505517217, "learning_rate": 5.091366476239425e-05, "loss": 2.7202, "step": 26327 }, { "epoch": 1.634365882425973, "grad_norm": 0.5809592545300557, "learning_rate": 5.0910053838852476e-05, "loss": 2.8272, "step": 26328 }, { "epoch": 1.634427959525731, "grad_norm": 0.5976658988622544, "learning_rate": 5.0906442910562715e-05, "loss": 2.7917, "step": 26329 }, { "epoch": 1.6344900366254889, "grad_norm": 0.5483148149602155, "learning_rate": 5.090283197754383e-05, "loss": 2.7633, "step": 26330 }, { "epoch": 1.6345521137252468, "grad_norm": 0.5855574074396596, "learning_rate": 5.089922103981466e-05, "loss": 2.7454, "step": 26331 }, { "epoch": 1.6346141908250047, "grad_norm": 0.5304231098749617, "learning_rate": 5.089561009739401e-05, "loss": 2.7168, "step": 26332 }, { "epoch": 1.6346762679247626, "grad_norm": 0.6175618786638353, "learning_rate": 5.089199915030076e-05, "loss": 2.81, "step": 26333 }, { "epoch": 1.6347383450245205, "grad_norm": 0.5638494464341477, "learning_rate": 5.0888388198553724e-05, "loss": 2.8437, "step": 26334 }, { "epoch": 1.6348004221242785, "grad_norm": 0.5839938414964206, "learning_rate": 5.0884777242171776e-05, "loss": 2.7715, "step": 26335 }, { "epoch": 1.6348624992240364, "grad_norm": 0.5613706165177405, "learning_rate": 5.0881166281173695e-05, "loss": 2.6724, "step": 26336 }, { "epoch": 1.6349245763237943, "grad_norm": 0.6009060138581114, "learning_rate": 5.087755531557839e-05, "loss": 2.738, "step": 26337 }, { "epoch": 1.6349866534235522, "grad_norm": 0.5173044308785137, "learning_rate": 5.087394434540465e-05, "loss": 2.6817, "step": 26338 }, { "epoch": 1.63504873052331, "grad_norm": 0.5975493790560817, "learning_rate": 5.087033337067133e-05, "loss": 2.768, "step": 26339 }, { "epoch": 1.6351108076230678, "grad_norm": 0.5458757808671856, "learning_rate": 5.086672239139726e-05, "loss": 2.7919, "step": 26340 }, { "epoch": 1.6351728847228257, "grad_norm": 0.6402295217743215, "learning_rate": 5.08631114076013e-05, "loss": 2.817, "step": 26341 }, { "epoch": 1.6352349618225837, "grad_norm": 0.6012993521867209, "learning_rate": 5.0859500419302285e-05, "loss": 2.7456, "step": 26342 }, { "epoch": 1.6352970389223416, "grad_norm": 0.6042260491576661, "learning_rate": 5.085588942651902e-05, "loss": 2.7026, "step": 26343 }, { "epoch": 1.6353591160220995, "grad_norm": 0.5952640103193761, "learning_rate": 5.08522784292704e-05, "loss": 2.8368, "step": 26344 }, { "epoch": 1.6354211931218572, "grad_norm": 0.5955890984137658, "learning_rate": 5.084866742757523e-05, "loss": 2.7764, "step": 26345 }, { "epoch": 1.635483270221615, "grad_norm": 0.6357788450392947, "learning_rate": 5.0845056421452354e-05, "loss": 2.694, "step": 26346 }, { "epoch": 1.635545347321373, "grad_norm": 0.5957171829315884, "learning_rate": 5.0841445410920606e-05, "loss": 2.7863, "step": 26347 }, { "epoch": 1.635607424421131, "grad_norm": 0.5801688486175287, "learning_rate": 5.083783439599885e-05, "loss": 2.768, "step": 26348 }, { "epoch": 1.6356695015208889, "grad_norm": 0.5809447881806767, "learning_rate": 5.08342233767059e-05, "loss": 2.8115, "step": 26349 }, { "epoch": 1.6357315786206468, "grad_norm": 0.5720228650234752, "learning_rate": 5.0830612353060605e-05, "loss": 2.8944, "step": 26350 }, { "epoch": 1.6357936557204047, "grad_norm": 0.5227369773570779, "learning_rate": 5.08270013250818e-05, "loss": 2.7534, "step": 26351 }, { "epoch": 1.6358557328201626, "grad_norm": 0.5060845540345674, "learning_rate": 5.082339029278832e-05, "loss": 2.727, "step": 26352 }, { "epoch": 1.6359178099199205, "grad_norm": 0.6297319571654009, "learning_rate": 5.081977925619903e-05, "loss": 2.7573, "step": 26353 }, { "epoch": 1.6359798870196784, "grad_norm": 0.5627978384087594, "learning_rate": 5.081616821533274e-05, "loss": 2.7382, "step": 26354 }, { "epoch": 1.6360419641194364, "grad_norm": 0.5996537828930936, "learning_rate": 5.081255717020832e-05, "loss": 2.7535, "step": 26355 }, { "epoch": 1.6361040412191943, "grad_norm": 0.5647145494190102, "learning_rate": 5.080894612084457e-05, "loss": 2.7821, "step": 26356 }, { "epoch": 1.6361661183189522, "grad_norm": 0.6558453356732369, "learning_rate": 5.0805335067260374e-05, "loss": 2.7892, "step": 26357 }, { "epoch": 1.6362281954187101, "grad_norm": 0.5409312268471332, "learning_rate": 5.080172400947453e-05, "loss": 2.783, "step": 26358 }, { "epoch": 1.636290272518468, "grad_norm": 0.5858839137687424, "learning_rate": 5.079811294750589e-05, "loss": 2.7625, "step": 26359 }, { "epoch": 1.636352349618226, "grad_norm": 0.5587316756262609, "learning_rate": 5.079450188137332e-05, "loss": 2.7196, "step": 26360 }, { "epoch": 1.6364144267179839, "grad_norm": 0.5665141298883678, "learning_rate": 5.079089081109564e-05, "loss": 2.7879, "step": 26361 }, { "epoch": 1.6364765038177418, "grad_norm": 0.5922147075270939, "learning_rate": 5.078727973669168e-05, "loss": 2.8502, "step": 26362 }, { "epoch": 1.6365385809174995, "grad_norm": 0.5470320067499893, "learning_rate": 5.0783668658180296e-05, "loss": 2.7571, "step": 26363 }, { "epoch": 1.6366006580172574, "grad_norm": 0.6344631941801566, "learning_rate": 5.078005757558033e-05, "loss": 2.7542, "step": 26364 }, { "epoch": 1.6366627351170153, "grad_norm": 0.5997948623692226, "learning_rate": 5.07764464889106e-05, "loss": 2.7862, "step": 26365 }, { "epoch": 1.6367248122167732, "grad_norm": 0.6348153901766265, "learning_rate": 5.077283539818998e-05, "loss": 2.7811, "step": 26366 }, { "epoch": 1.6367868893165312, "grad_norm": 0.6138588681751136, "learning_rate": 5.076922430343728e-05, "loss": 2.7768, "step": 26367 }, { "epoch": 1.636848966416289, "grad_norm": 0.5328441151834773, "learning_rate": 5.0765613204671345e-05, "loss": 2.7014, "step": 26368 }, { "epoch": 1.6369110435160468, "grad_norm": 0.5934945862176885, "learning_rate": 5.076200210191103e-05, "loss": 2.7622, "step": 26369 }, { "epoch": 1.6369731206158047, "grad_norm": 0.6001061733004603, "learning_rate": 5.075839099517515e-05, "loss": 2.6847, "step": 26370 }, { "epoch": 1.6370351977155626, "grad_norm": 0.5617793818447687, "learning_rate": 5.075477988448257e-05, "loss": 2.7933, "step": 26371 }, { "epoch": 1.6370972748153205, "grad_norm": 0.6034646614556087, "learning_rate": 5.075116876985212e-05, "loss": 2.8452, "step": 26372 }, { "epoch": 1.6371593519150784, "grad_norm": 0.5586519277727323, "learning_rate": 5.0747557651302645e-05, "loss": 2.7612, "step": 26373 }, { "epoch": 1.6372214290148364, "grad_norm": 0.5745020576221082, "learning_rate": 5.074394652885297e-05, "loss": 2.6919, "step": 26374 }, { "epoch": 1.6372835061145943, "grad_norm": 0.5227563325999174, "learning_rate": 5.074033540252196e-05, "loss": 2.711, "step": 26375 }, { "epoch": 1.6373455832143522, "grad_norm": 0.5429178941361006, "learning_rate": 5.0736724272328415e-05, "loss": 2.8007, "step": 26376 }, { "epoch": 1.6374076603141101, "grad_norm": 0.5693324303232725, "learning_rate": 5.073311313829121e-05, "loss": 2.7792, "step": 26377 }, { "epoch": 1.637469737413868, "grad_norm": 0.511049575645973, "learning_rate": 5.072950200042918e-05, "loss": 2.8227, "step": 26378 }, { "epoch": 1.637531814513626, "grad_norm": 0.5214556568253345, "learning_rate": 5.072589085876116e-05, "loss": 2.7198, "step": 26379 }, { "epoch": 1.6375938916133839, "grad_norm": 0.521256500633559, "learning_rate": 5.0722279713306e-05, "loss": 2.8032, "step": 26380 }, { "epoch": 1.6376559687131418, "grad_norm": 0.5917986123843232, "learning_rate": 5.071866856408252e-05, "loss": 2.7823, "step": 26381 }, { "epoch": 1.6377180458128997, "grad_norm": 0.5584175525554459, "learning_rate": 5.0715057411109576e-05, "loss": 2.8452, "step": 26382 }, { "epoch": 1.6377801229126576, "grad_norm": 0.5946285624322196, "learning_rate": 5.0711446254406e-05, "loss": 2.8338, "step": 26383 }, { "epoch": 1.6378422000124155, "grad_norm": 0.5833855815603752, "learning_rate": 5.070783509399065e-05, "loss": 2.7008, "step": 26384 }, { "epoch": 1.6379042771121735, "grad_norm": 0.6347034599968382, "learning_rate": 5.0704223929882356e-05, "loss": 2.7241, "step": 26385 }, { "epoch": 1.6379663542119314, "grad_norm": 0.6102068920342053, "learning_rate": 5.070061276209993e-05, "loss": 2.7781, "step": 26386 }, { "epoch": 1.638028431311689, "grad_norm": 0.6593763270070974, "learning_rate": 5.069700159066225e-05, "loss": 2.7249, "step": 26387 }, { "epoch": 1.638090508411447, "grad_norm": 0.6602519716466652, "learning_rate": 5.069339041558814e-05, "loss": 2.8, "step": 26388 }, { "epoch": 1.638152585511205, "grad_norm": 0.6299214432573075, "learning_rate": 5.068977923689645e-05, "loss": 2.7918, "step": 26389 }, { "epoch": 1.6382146626109628, "grad_norm": 0.6166962950479364, "learning_rate": 5.068616805460601e-05, "loss": 2.7416, "step": 26390 }, { "epoch": 1.6382767397107207, "grad_norm": 0.5524772881892941, "learning_rate": 5.068255686873567e-05, "loss": 2.777, "step": 26391 }, { "epoch": 1.6383388168104787, "grad_norm": 0.6191167736573646, "learning_rate": 5.067894567930426e-05, "loss": 2.7264, "step": 26392 }, { "epoch": 1.6384008939102364, "grad_norm": 0.5241995431306399, "learning_rate": 5.067533448633062e-05, "loss": 2.7319, "step": 26393 }, { "epoch": 1.6384629710099943, "grad_norm": 0.5993291094240036, "learning_rate": 5.067172328983361e-05, "loss": 2.7551, "step": 26394 }, { "epoch": 1.6385250481097522, "grad_norm": 0.5758979792911021, "learning_rate": 5.0668112089832035e-05, "loss": 2.7777, "step": 26395 }, { "epoch": 1.63858712520951, "grad_norm": 0.6186994310387202, "learning_rate": 5.066450088634478e-05, "loss": 2.7861, "step": 26396 }, { "epoch": 1.638649202309268, "grad_norm": 0.6108899237017962, "learning_rate": 5.0660889679390644e-05, "loss": 2.7748, "step": 26397 }, { "epoch": 1.638711279409026, "grad_norm": 0.573943377009746, "learning_rate": 5.06572784689885e-05, "loss": 2.8243, "step": 26398 }, { "epoch": 1.6387733565087839, "grad_norm": 0.55287009685808, "learning_rate": 5.0653667255157165e-05, "loss": 2.736, "step": 26399 }, { "epoch": 1.6388354336085418, "grad_norm": 0.5433573807131974, "learning_rate": 5.0650056037915496e-05, "loss": 2.8337, "step": 26400 }, { "epoch": 1.6388975107082997, "grad_norm": 0.5688166139179558, "learning_rate": 5.064644481728232e-05, "loss": 2.7681, "step": 26401 }, { "epoch": 1.6389595878080576, "grad_norm": 0.5176948847517426, "learning_rate": 5.064283359327649e-05, "loss": 2.7164, "step": 26402 }, { "epoch": 1.6390216649078155, "grad_norm": 0.5663880707026699, "learning_rate": 5.063922236591685e-05, "loss": 2.768, "step": 26403 }, { "epoch": 1.6390837420075735, "grad_norm": 0.5171617380508313, "learning_rate": 5.0635611135222214e-05, "loss": 2.751, "step": 26404 }, { "epoch": 1.6391458191073314, "grad_norm": 0.5935459037148396, "learning_rate": 5.0631999901211454e-05, "loss": 2.7947, "step": 26405 }, { "epoch": 1.6392078962070893, "grad_norm": 0.5133176387156679, "learning_rate": 5.062838866390339e-05, "loss": 2.742, "step": 26406 }, { "epoch": 1.6392699733068472, "grad_norm": 0.6350705632104388, "learning_rate": 5.062477742331688e-05, "loss": 2.6362, "step": 26407 }, { "epoch": 1.6393320504066051, "grad_norm": 0.5383052206031114, "learning_rate": 5.062116617947073e-05, "loss": 2.7077, "step": 26408 }, { "epoch": 1.639394127506363, "grad_norm": 0.6720621602396256, "learning_rate": 5.0617554932383825e-05, "loss": 2.7347, "step": 26409 }, { "epoch": 1.639456204606121, "grad_norm": 0.6083401916935979, "learning_rate": 5.061394368207498e-05, "loss": 2.7948, "step": 26410 }, { "epoch": 1.6395182817058787, "grad_norm": 0.6445871039956129, "learning_rate": 5.061033242856305e-05, "loss": 2.7699, "step": 26411 }, { "epoch": 1.6395803588056366, "grad_norm": 0.5947183442448394, "learning_rate": 5.060672117186687e-05, "loss": 2.7497, "step": 26412 }, { "epoch": 1.6396424359053945, "grad_norm": 0.6170931393885034, "learning_rate": 5.0603109912005255e-05, "loss": 2.7187, "step": 26413 }, { "epoch": 1.6397045130051524, "grad_norm": 0.6509398927612573, "learning_rate": 5.059949864899709e-05, "loss": 2.7918, "step": 26414 }, { "epoch": 1.6397665901049103, "grad_norm": 0.6024655586307343, "learning_rate": 5.059588738286118e-05, "loss": 2.6495, "step": 26415 }, { "epoch": 1.6398286672046682, "grad_norm": 0.5940002114973109, "learning_rate": 5.05922761136164e-05, "loss": 2.7724, "step": 26416 }, { "epoch": 1.639890744304426, "grad_norm": 0.661871200529005, "learning_rate": 5.058866484128156e-05, "loss": 2.7845, "step": 26417 }, { "epoch": 1.6399528214041839, "grad_norm": 0.6183677928204598, "learning_rate": 5.0585053565875515e-05, "loss": 2.7689, "step": 26418 }, { "epoch": 1.6400148985039418, "grad_norm": 0.7043380234430227, "learning_rate": 5.05814422874171e-05, "loss": 2.6939, "step": 26419 }, { "epoch": 1.6400769756036997, "grad_norm": 0.5779087074444953, "learning_rate": 5.057783100592517e-05, "loss": 2.7826, "step": 26420 }, { "epoch": 1.6401390527034576, "grad_norm": 0.6549224823297385, "learning_rate": 5.057421972141855e-05, "loss": 2.8182, "step": 26421 }, { "epoch": 1.6402011298032155, "grad_norm": 0.5610176262967912, "learning_rate": 5.0570608433916075e-05, "loss": 2.7537, "step": 26422 }, { "epoch": 1.6402632069029734, "grad_norm": 0.6074886224018788, "learning_rate": 5.056699714343661e-05, "loss": 2.7418, "step": 26423 }, { "epoch": 1.6403252840027314, "grad_norm": 0.6440407013292867, "learning_rate": 5.0563385849998975e-05, "loss": 2.7875, "step": 26424 }, { "epoch": 1.6403873611024893, "grad_norm": 0.5566316838131954, "learning_rate": 5.055977455362202e-05, "loss": 2.6873, "step": 26425 }, { "epoch": 1.6404494382022472, "grad_norm": 0.611018361352652, "learning_rate": 5.0556163254324584e-05, "loss": 2.7211, "step": 26426 }, { "epoch": 1.6405115153020051, "grad_norm": 0.5782768204814939, "learning_rate": 5.0552551952125525e-05, "loss": 2.6887, "step": 26427 }, { "epoch": 1.640573592401763, "grad_norm": 0.5789213881270109, "learning_rate": 5.054894064704367e-05, "loss": 2.7433, "step": 26428 }, { "epoch": 1.640635669501521, "grad_norm": 0.6783543809660569, "learning_rate": 5.0545329339097834e-05, "loss": 2.7973, "step": 26429 }, { "epoch": 1.6406977466012789, "grad_norm": 0.5655455001264191, "learning_rate": 5.054171802830689e-05, "loss": 2.7391, "step": 26430 }, { "epoch": 1.6407598237010368, "grad_norm": 0.6098285356676755, "learning_rate": 5.053810671468968e-05, "loss": 2.7476, "step": 26431 }, { "epoch": 1.6408219008007947, "grad_norm": 0.563851296359831, "learning_rate": 5.0534495398265034e-05, "loss": 2.7644, "step": 26432 }, { "epoch": 1.6408839779005526, "grad_norm": 0.618388179028664, "learning_rate": 5.0530884079051786e-05, "loss": 2.6871, "step": 26433 }, { "epoch": 1.6409460550003105, "grad_norm": 0.58295845576808, "learning_rate": 5.052727275706881e-05, "loss": 2.7884, "step": 26434 }, { "epoch": 1.6410081321000682, "grad_norm": 0.6491367292866896, "learning_rate": 5.05236614323349e-05, "loss": 2.7831, "step": 26435 }, { "epoch": 1.6410702091998262, "grad_norm": 0.5969883796097702, "learning_rate": 5.0520050104868934e-05, "loss": 2.7926, "step": 26436 }, { "epoch": 1.641132286299584, "grad_norm": 0.6671829286012223, "learning_rate": 5.0516438774689734e-05, "loss": 2.7755, "step": 26437 }, { "epoch": 1.641194363399342, "grad_norm": 0.6035775458412026, "learning_rate": 5.0512827441816155e-05, "loss": 2.7301, "step": 26438 }, { "epoch": 1.6412564404991, "grad_norm": 0.6562608536322285, "learning_rate": 5.050921610626703e-05, "loss": 2.8765, "step": 26439 }, { "epoch": 1.6413185175988578, "grad_norm": 0.6128365915281185, "learning_rate": 5.0505604768061197e-05, "loss": 2.7717, "step": 26440 }, { "epoch": 1.6413805946986155, "grad_norm": 0.5873141836978705, "learning_rate": 5.05019934272175e-05, "loss": 2.8456, "step": 26441 }, { "epoch": 1.6414426717983734, "grad_norm": 0.5708692002224227, "learning_rate": 5.049838208375477e-05, "loss": 2.7539, "step": 26442 }, { "epoch": 1.6415047488981314, "grad_norm": 0.5608831055176424, "learning_rate": 5.049477073769188e-05, "loss": 2.7108, "step": 26443 }, { "epoch": 1.6415668259978893, "grad_norm": 0.6230170024538684, "learning_rate": 5.0491159389047637e-05, "loss": 2.7782, "step": 26444 }, { "epoch": 1.6416289030976472, "grad_norm": 0.5602569095929365, "learning_rate": 5.0487548037840904e-05, "loss": 2.7161, "step": 26445 }, { "epoch": 1.6416909801974051, "grad_norm": 0.5622968571797761, "learning_rate": 5.0483936684090525e-05, "loss": 2.6643, "step": 26446 }, { "epoch": 1.641753057297163, "grad_norm": 0.5494289977928416, "learning_rate": 5.048032532781532e-05, "loss": 2.7305, "step": 26447 }, { "epoch": 1.641815134396921, "grad_norm": 0.5482367319257581, "learning_rate": 5.047671396903414e-05, "loss": 2.7283, "step": 26448 }, { "epoch": 1.6418772114966789, "grad_norm": 0.5805352205028899, "learning_rate": 5.047310260776582e-05, "loss": 2.6965, "step": 26449 }, { "epoch": 1.6419392885964368, "grad_norm": 0.6143309969584221, "learning_rate": 5.046949124402922e-05, "loss": 2.8432, "step": 26450 }, { "epoch": 1.6420013656961947, "grad_norm": 0.5731307901154447, "learning_rate": 5.046587987784316e-05, "loss": 2.7368, "step": 26451 }, { "epoch": 1.6420634427959526, "grad_norm": 0.5552617981905476, "learning_rate": 5.046226850922651e-05, "loss": 2.7627, "step": 26452 }, { "epoch": 1.6421255198957105, "grad_norm": 0.5949633147446592, "learning_rate": 5.0458657138198075e-05, "loss": 2.6865, "step": 26453 }, { "epoch": 1.6421875969954685, "grad_norm": 0.565369070753659, "learning_rate": 5.045504576477672e-05, "loss": 2.8205, "step": 26454 }, { "epoch": 1.6422496740952264, "grad_norm": 0.614279572263694, "learning_rate": 5.0451434388981276e-05, "loss": 2.7993, "step": 26455 }, { "epoch": 1.6423117511949843, "grad_norm": 0.57104463483486, "learning_rate": 5.0447823010830595e-05, "loss": 2.7763, "step": 26456 }, { "epoch": 1.6423738282947422, "grad_norm": 0.6177943903613711, "learning_rate": 5.044421163034352e-05, "loss": 2.7385, "step": 26457 }, { "epoch": 1.6424359053945001, "grad_norm": 0.5446283787448977, "learning_rate": 5.044060024753886e-05, "loss": 2.6984, "step": 26458 }, { "epoch": 1.6424979824942578, "grad_norm": 0.623923781769851, "learning_rate": 5.043698886243551e-05, "loss": 2.9062, "step": 26459 }, { "epoch": 1.6425600595940157, "grad_norm": 0.5642393294674631, "learning_rate": 5.043337747505227e-05, "loss": 2.7843, "step": 26460 }, { "epoch": 1.6426221366937737, "grad_norm": 0.6251536657255647, "learning_rate": 5.0429766085408e-05, "loss": 2.7673, "step": 26461 }, { "epoch": 1.6426842137935316, "grad_norm": 0.6294568181198844, "learning_rate": 5.042615469352152e-05, "loss": 2.7301, "step": 26462 }, { "epoch": 1.6427462908932895, "grad_norm": 0.5800141064802868, "learning_rate": 5.0422543299411705e-05, "loss": 2.8513, "step": 26463 }, { "epoch": 1.6428083679930474, "grad_norm": 0.570751308780528, "learning_rate": 5.041893190309739e-05, "loss": 2.7782, "step": 26464 }, { "epoch": 1.642870445092805, "grad_norm": 0.5531005185846419, "learning_rate": 5.0415320504597374e-05, "loss": 2.684, "step": 26465 }, { "epoch": 1.642932522192563, "grad_norm": 0.5700371743929475, "learning_rate": 5.041170910393056e-05, "loss": 2.8948, "step": 26466 }, { "epoch": 1.642994599292321, "grad_norm": 0.5742453225144176, "learning_rate": 5.040809770111574e-05, "loss": 2.8024, "step": 26467 }, { "epoch": 1.6430566763920789, "grad_norm": 0.547138695492937, "learning_rate": 5.040448629617178e-05, "loss": 2.7477, "step": 26468 }, { "epoch": 1.6431187534918368, "grad_norm": 0.527430887440531, "learning_rate": 5.040087488911751e-05, "loss": 2.6632, "step": 26469 }, { "epoch": 1.6431808305915947, "grad_norm": 0.5677895589927334, "learning_rate": 5.039726347997179e-05, "loss": 2.8319, "step": 26470 }, { "epoch": 1.6432429076913526, "grad_norm": 0.5663174263597065, "learning_rate": 5.039365206875345e-05, "loss": 2.6485, "step": 26471 }, { "epoch": 1.6433049847911105, "grad_norm": 0.5478296394517614, "learning_rate": 5.039004065548133e-05, "loss": 2.6507, "step": 26472 }, { "epoch": 1.6433670618908685, "grad_norm": 0.6196971522054201, "learning_rate": 5.038642924017426e-05, "loss": 2.8018, "step": 26473 }, { "epoch": 1.6434291389906264, "grad_norm": 0.5675619764977754, "learning_rate": 5.03828178228511e-05, "loss": 2.7958, "step": 26474 }, { "epoch": 1.6434912160903843, "grad_norm": 0.5399689171816924, "learning_rate": 5.03792064035307e-05, "loss": 2.763, "step": 26475 }, { "epoch": 1.6435532931901422, "grad_norm": 0.5427339169505954, "learning_rate": 5.037559498223188e-05, "loss": 2.7332, "step": 26476 }, { "epoch": 1.6436153702899001, "grad_norm": 0.5276196641017307, "learning_rate": 5.037198355897349e-05, "loss": 2.7298, "step": 26477 }, { "epoch": 1.643677447389658, "grad_norm": 0.5507085224590086, "learning_rate": 5.036837213377437e-05, "loss": 2.7368, "step": 26478 }, { "epoch": 1.643739524489416, "grad_norm": 0.5192105156529386, "learning_rate": 5.036476070665337e-05, "loss": 2.7168, "step": 26479 }, { "epoch": 1.6438016015891739, "grad_norm": 0.5515966399436256, "learning_rate": 5.036114927762931e-05, "loss": 2.6854, "step": 26480 }, { "epoch": 1.6438636786889318, "grad_norm": 0.5266128278276774, "learning_rate": 5.035753784672106e-05, "loss": 2.901, "step": 26481 }, { "epoch": 1.6439257557886897, "grad_norm": 0.5360818640676287, "learning_rate": 5.035392641394745e-05, "loss": 2.6757, "step": 26482 }, { "epoch": 1.6439878328884474, "grad_norm": 0.5831450232188685, "learning_rate": 5.0350314979327306e-05, "loss": 2.7502, "step": 26483 }, { "epoch": 1.6440499099882053, "grad_norm": 0.5342917316098286, "learning_rate": 5.03467035428795e-05, "loss": 2.7563, "step": 26484 }, { "epoch": 1.6441119870879632, "grad_norm": 0.5711445160935879, "learning_rate": 5.034309210462285e-05, "loss": 2.689, "step": 26485 }, { "epoch": 1.6441740641877212, "grad_norm": 0.6156729451984806, "learning_rate": 5.03394806645762e-05, "loss": 2.7992, "step": 26486 }, { "epoch": 1.644236141287479, "grad_norm": 0.6651578060577307, "learning_rate": 5.0335869222758405e-05, "loss": 2.7837, "step": 26487 }, { "epoch": 1.644298218387237, "grad_norm": 0.5905838674522509, "learning_rate": 5.033225777918829e-05, "loss": 2.81, "step": 26488 }, { "epoch": 1.6443602954869947, "grad_norm": 0.6380758274796545, "learning_rate": 5.032864633388471e-05, "loss": 2.8202, "step": 26489 }, { "epoch": 1.6444223725867526, "grad_norm": 0.5404136145227851, "learning_rate": 5.03250348868665e-05, "loss": 2.8277, "step": 26490 }, { "epoch": 1.6444844496865105, "grad_norm": 0.6689987718146074, "learning_rate": 5.032142343815252e-05, "loss": 2.7419, "step": 26491 }, { "epoch": 1.6445465267862684, "grad_norm": 0.48300716093137414, "learning_rate": 5.031781198776157e-05, "loss": 2.7374, "step": 26492 }, { "epoch": 1.6446086038860264, "grad_norm": 0.6264748576219457, "learning_rate": 5.031420053571253e-05, "loss": 2.8106, "step": 26493 }, { "epoch": 1.6446706809857843, "grad_norm": 0.5402347424023883, "learning_rate": 5.031058908202423e-05, "loss": 2.769, "step": 26494 }, { "epoch": 1.6447327580855422, "grad_norm": 0.6023027893467966, "learning_rate": 5.030697762671551e-05, "loss": 2.7561, "step": 26495 }, { "epoch": 1.6447948351853001, "grad_norm": 0.6093064538599171, "learning_rate": 5.0303366169805214e-05, "loss": 2.7741, "step": 26496 }, { "epoch": 1.644856912285058, "grad_norm": 0.5845160982054124, "learning_rate": 5.0299754711312184e-05, "loss": 2.8473, "step": 26497 }, { "epoch": 1.644918989384816, "grad_norm": 0.5795285739667195, "learning_rate": 5.029614325125526e-05, "loss": 2.766, "step": 26498 }, { "epoch": 1.6449810664845739, "grad_norm": 0.5407145519727464, "learning_rate": 5.0292531789653294e-05, "loss": 2.7949, "step": 26499 }, { "epoch": 1.6450431435843318, "grad_norm": 0.5716830468217569, "learning_rate": 5.02889203265251e-05, "loss": 2.7936, "step": 26500 }, { "epoch": 1.6451052206840897, "grad_norm": 0.570744024608878, "learning_rate": 5.028530886188956e-05, "loss": 2.7303, "step": 26501 }, { "epoch": 1.6451672977838476, "grad_norm": 0.6081592463135358, "learning_rate": 5.028169739576548e-05, "loss": 2.7972, "step": 26502 }, { "epoch": 1.6452293748836055, "grad_norm": 0.6242703231494571, "learning_rate": 5.027808592817171e-05, "loss": 2.8928, "step": 26503 }, { "epoch": 1.6452914519833635, "grad_norm": 0.5780616931792106, "learning_rate": 5.027447445912712e-05, "loss": 2.7303, "step": 26504 }, { "epoch": 1.6453535290831214, "grad_norm": 0.6153044645560689, "learning_rate": 5.0270862988650516e-05, "loss": 2.7604, "step": 26505 }, { "epoch": 1.6454156061828793, "grad_norm": 0.5258859901340719, "learning_rate": 5.026725151676076e-05, "loss": 2.7308, "step": 26506 }, { "epoch": 1.645477683282637, "grad_norm": 0.601694934683729, "learning_rate": 5.026364004347668e-05, "loss": 2.8058, "step": 26507 }, { "epoch": 1.645539760382395, "grad_norm": 0.5599118400473948, "learning_rate": 5.026002856881712e-05, "loss": 2.7322, "step": 26508 }, { "epoch": 1.6456018374821528, "grad_norm": 0.557725378119696, "learning_rate": 5.0256417092800945e-05, "loss": 2.7041, "step": 26509 }, { "epoch": 1.6456639145819107, "grad_norm": 0.5519881856248346, "learning_rate": 5.0252805615446965e-05, "loss": 2.6575, "step": 26510 }, { "epoch": 1.6457259916816687, "grad_norm": 0.5790476781983169, "learning_rate": 5.0249194136774046e-05, "loss": 2.7664, "step": 26511 }, { "epoch": 1.6457880687814266, "grad_norm": 0.5425509210499455, "learning_rate": 5.024558265680102e-05, "loss": 2.7273, "step": 26512 }, { "epoch": 1.6458501458811843, "grad_norm": 0.5403913897257588, "learning_rate": 5.024197117554673e-05, "loss": 2.6877, "step": 26513 }, { "epoch": 1.6459122229809422, "grad_norm": 0.5837733675345589, "learning_rate": 5.0238359693030016e-05, "loss": 2.7512, "step": 26514 }, { "epoch": 1.6459743000807001, "grad_norm": 0.5396699937677034, "learning_rate": 5.023474820926972e-05, "loss": 2.8157, "step": 26515 }, { "epoch": 1.646036377180458, "grad_norm": 0.48919522005001076, "learning_rate": 5.023113672428468e-05, "loss": 2.8137, "step": 26516 }, { "epoch": 1.646098454280216, "grad_norm": 0.5840216277115314, "learning_rate": 5.022752523809375e-05, "loss": 2.8278, "step": 26517 }, { "epoch": 1.6461605313799739, "grad_norm": 0.49879250156514454, "learning_rate": 5.022391375071577e-05, "loss": 2.7357, "step": 26518 }, { "epoch": 1.6462226084797318, "grad_norm": 0.5264014125023098, "learning_rate": 5.022030226216958e-05, "loss": 2.8323, "step": 26519 }, { "epoch": 1.6462846855794897, "grad_norm": 0.5095925457491558, "learning_rate": 5.021669077247402e-05, "loss": 2.7835, "step": 26520 }, { "epoch": 1.6463467626792476, "grad_norm": 0.5093384337610909, "learning_rate": 5.021307928164791e-05, "loss": 2.7955, "step": 26521 }, { "epoch": 1.6464088397790055, "grad_norm": 0.5230574580426, "learning_rate": 5.020946778971014e-05, "loss": 2.712, "step": 26522 }, { "epoch": 1.6464709168787635, "grad_norm": 0.48037776240562824, "learning_rate": 5.0205856296679496e-05, "loss": 2.737, "step": 26523 }, { "epoch": 1.6465329939785214, "grad_norm": 0.5401413391279825, "learning_rate": 5.020224480257487e-05, "loss": 2.7897, "step": 26524 }, { "epoch": 1.6465950710782793, "grad_norm": 0.5010504119754868, "learning_rate": 5.019863330741509e-05, "loss": 2.7674, "step": 26525 }, { "epoch": 1.6466571481780372, "grad_norm": 0.5434353970645449, "learning_rate": 5.019502181121898e-05, "loss": 2.7863, "step": 26526 }, { "epoch": 1.6467192252777951, "grad_norm": 0.5775266850486505, "learning_rate": 5.01914103140054e-05, "loss": 2.6844, "step": 26527 }, { "epoch": 1.646781302377553, "grad_norm": 0.5535205601248075, "learning_rate": 5.018779881579319e-05, "loss": 2.7793, "step": 26528 }, { "epoch": 1.646843379477311, "grad_norm": 0.5458514034638616, "learning_rate": 5.018418731660118e-05, "loss": 2.7834, "step": 26529 }, { "epoch": 1.6469054565770689, "grad_norm": 0.578902597969263, "learning_rate": 5.018057581644822e-05, "loss": 2.7373, "step": 26530 }, { "epoch": 1.6469675336768266, "grad_norm": 0.5737206251233345, "learning_rate": 5.017696431535316e-05, "loss": 2.7678, "step": 26531 }, { "epoch": 1.6470296107765845, "grad_norm": 0.5805377359585964, "learning_rate": 5.017335281333482e-05, "loss": 2.767, "step": 26532 }, { "epoch": 1.6470916878763424, "grad_norm": 0.5613615383113199, "learning_rate": 5.016974131041206e-05, "loss": 2.7604, "step": 26533 }, { "epoch": 1.6471537649761003, "grad_norm": 0.5607550098384174, "learning_rate": 5.0166129806603726e-05, "loss": 2.7997, "step": 26534 }, { "epoch": 1.6472158420758582, "grad_norm": 0.5647070290545667, "learning_rate": 5.016251830192865e-05, "loss": 2.7913, "step": 26535 }, { "epoch": 1.6472779191756162, "grad_norm": 0.5073309449088678, "learning_rate": 5.015890679640567e-05, "loss": 2.7104, "step": 26536 }, { "epoch": 1.6473399962753739, "grad_norm": 0.6183856247651736, "learning_rate": 5.015529529005365e-05, "loss": 2.7769, "step": 26537 }, { "epoch": 1.6474020733751318, "grad_norm": 0.5612747570516299, "learning_rate": 5.0151683782891414e-05, "loss": 2.6976, "step": 26538 }, { "epoch": 1.6474641504748897, "grad_norm": 0.5547783468292226, "learning_rate": 5.01480722749378e-05, "loss": 2.791, "step": 26539 }, { "epoch": 1.6475262275746476, "grad_norm": 0.5804570819405952, "learning_rate": 5.014446076621164e-05, "loss": 2.855, "step": 26540 }, { "epoch": 1.6475883046744055, "grad_norm": 0.543926656461307, "learning_rate": 5.0140849256731804e-05, "loss": 2.7129, "step": 26541 }, { "epoch": 1.6476503817741635, "grad_norm": 0.5458570816125078, "learning_rate": 5.013723774651714e-05, "loss": 2.6771, "step": 26542 }, { "epoch": 1.6477124588739214, "grad_norm": 0.5475685626259187, "learning_rate": 5.013362623558645e-05, "loss": 2.7178, "step": 26543 }, { "epoch": 1.6477745359736793, "grad_norm": 0.5453073375643358, "learning_rate": 5.013001472395862e-05, "loss": 2.7135, "step": 26544 }, { "epoch": 1.6478366130734372, "grad_norm": 0.5377284418410846, "learning_rate": 5.0126403211652476e-05, "loss": 2.7229, "step": 26545 }, { "epoch": 1.6478986901731951, "grad_norm": 0.5286728111361398, "learning_rate": 5.012279169868683e-05, "loss": 2.719, "step": 26546 }, { "epoch": 1.647960767272953, "grad_norm": 0.5919989313328609, "learning_rate": 5.011918018508057e-05, "loss": 2.7067, "step": 26547 }, { "epoch": 1.648022844372711, "grad_norm": 0.5192143900915236, "learning_rate": 5.011556867085251e-05, "loss": 2.6592, "step": 26548 }, { "epoch": 1.6480849214724689, "grad_norm": 0.6288000929427139, "learning_rate": 5.0111957156021503e-05, "loss": 2.7614, "step": 26549 }, { "epoch": 1.6481469985722268, "grad_norm": 0.5701670526519751, "learning_rate": 5.010834564060638e-05, "loss": 2.7517, "step": 26550 }, { "epoch": 1.6482090756719847, "grad_norm": 0.5898057786454061, "learning_rate": 5.0104734124626005e-05, "loss": 2.8448, "step": 26551 }, { "epoch": 1.6482711527717426, "grad_norm": 0.5339973598818575, "learning_rate": 5.010112260809919e-05, "loss": 2.7995, "step": 26552 }, { "epoch": 1.6483332298715005, "grad_norm": 0.6152063514745586, "learning_rate": 5.0097511091044816e-05, "loss": 2.7511, "step": 26553 }, { "epoch": 1.6483953069712585, "grad_norm": 0.5377761586050392, "learning_rate": 5.009389957348169e-05, "loss": 2.756, "step": 26554 }, { "epoch": 1.6484573840710162, "grad_norm": 0.5889621993258362, "learning_rate": 5.009028805542867e-05, "loss": 2.6279, "step": 26555 }, { "epoch": 1.648519461170774, "grad_norm": 0.5348582308093558, "learning_rate": 5.0086676536904584e-05, "loss": 2.7026, "step": 26556 }, { "epoch": 1.648581538270532, "grad_norm": 0.6334071250223293, "learning_rate": 5.00830650179283e-05, "loss": 2.7645, "step": 26557 }, { "epoch": 1.64864361537029, "grad_norm": 0.5946547992673968, "learning_rate": 5.007945349851864e-05, "loss": 2.8963, "step": 26558 }, { "epoch": 1.6487056924700478, "grad_norm": 0.5122238461886351, "learning_rate": 5.0075841978694446e-05, "loss": 2.7394, "step": 26559 }, { "epoch": 1.6487677695698058, "grad_norm": 0.61319010354888, "learning_rate": 5.007223045847458e-05, "loss": 2.8414, "step": 26560 }, { "epoch": 1.6488298466695634, "grad_norm": 0.5508088990975388, "learning_rate": 5.0068618937877854e-05, "loss": 2.8391, "step": 26561 }, { "epoch": 1.6488919237693214, "grad_norm": 0.5808116757292519, "learning_rate": 5.006500741692314e-05, "loss": 2.7152, "step": 26562 }, { "epoch": 1.6489540008690793, "grad_norm": 0.5686567911337416, "learning_rate": 5.006139589562927e-05, "loss": 2.7178, "step": 26563 }, { "epoch": 1.6490160779688372, "grad_norm": 0.5721786745433528, "learning_rate": 5.005778437401507e-05, "loss": 2.791, "step": 26564 }, { "epoch": 1.6490781550685951, "grad_norm": 0.58495790611335, "learning_rate": 5.00541728520994e-05, "loss": 2.7855, "step": 26565 }, { "epoch": 1.649140232168353, "grad_norm": 0.5339031724224306, "learning_rate": 5.005056132990109e-05, "loss": 2.731, "step": 26566 }, { "epoch": 1.649202309268111, "grad_norm": 0.6139474396979143, "learning_rate": 5.0046949807439006e-05, "loss": 2.7897, "step": 26567 }, { "epoch": 1.6492643863678689, "grad_norm": 0.5293777516641452, "learning_rate": 5.0043338284731955e-05, "loss": 2.7046, "step": 26568 }, { "epoch": 1.6493264634676268, "grad_norm": 0.5687199920032027, "learning_rate": 5.003972676179881e-05, "loss": 2.788, "step": 26569 }, { "epoch": 1.6493885405673847, "grad_norm": 0.5733123424433749, "learning_rate": 5.0036115238658385e-05, "loss": 2.8044, "step": 26570 }, { "epoch": 1.6494506176671426, "grad_norm": 0.5950182119755885, "learning_rate": 5.003250371532956e-05, "loss": 2.674, "step": 26571 }, { "epoch": 1.6495126947669005, "grad_norm": 0.6093077011115066, "learning_rate": 5.002889219183113e-05, "loss": 2.7485, "step": 26572 }, { "epoch": 1.6495747718666585, "grad_norm": 0.5422770258621868, "learning_rate": 5.002528066818198e-05, "loss": 2.7374, "step": 26573 }, { "epoch": 1.6496368489664164, "grad_norm": 0.5850079092485518, "learning_rate": 5.002166914440094e-05, "loss": 2.8115, "step": 26574 }, { "epoch": 1.6496989260661743, "grad_norm": 0.5229725267076163, "learning_rate": 5.001805762050683e-05, "loss": 2.6884, "step": 26575 }, { "epoch": 1.6497610031659322, "grad_norm": 0.6275568636931848, "learning_rate": 5.0014446096518516e-05, "loss": 2.7933, "step": 26576 }, { "epoch": 1.6498230802656901, "grad_norm": 0.5224408012146191, "learning_rate": 5.0010834572454836e-05, "loss": 2.879, "step": 26577 }, { "epoch": 1.649885157365448, "grad_norm": 0.6620599733900621, "learning_rate": 5.000722304833463e-05, "loss": 2.8542, "step": 26578 }, { "epoch": 1.6499472344652057, "grad_norm": 0.5112073541496555, "learning_rate": 5.000361152417673e-05, "loss": 2.7508, "step": 26579 }, { "epoch": 1.6500093115649637, "grad_norm": 0.5631480193011081, "learning_rate": 5e-05, "loss": 2.8252, "step": 26580 }, { "epoch": 1.6500713886647216, "grad_norm": 0.5383983335236808, "learning_rate": 4.9996388475823274e-05, "loss": 2.8347, "step": 26581 }, { "epoch": 1.6501334657644795, "grad_norm": 0.5358892558101175, "learning_rate": 4.999277695166538e-05, "loss": 2.8154, "step": 26582 }, { "epoch": 1.6501955428642374, "grad_norm": 0.5096698539481206, "learning_rate": 4.9989165427545176e-05, "loss": 2.8019, "step": 26583 }, { "epoch": 1.6502576199639953, "grad_norm": 0.5613345417247928, "learning_rate": 4.998555390348149e-05, "loss": 2.763, "step": 26584 }, { "epoch": 1.650319697063753, "grad_norm": 0.5694724388788084, "learning_rate": 4.998194237949318e-05, "loss": 2.688, "step": 26585 }, { "epoch": 1.650381774163511, "grad_norm": 0.5904045723013104, "learning_rate": 4.997833085559908e-05, "loss": 2.7932, "step": 26586 }, { "epoch": 1.6504438512632689, "grad_norm": 0.6457816969830517, "learning_rate": 4.9974719331818035e-05, "loss": 2.7732, "step": 26587 }, { "epoch": 1.6505059283630268, "grad_norm": 0.5784617036344238, "learning_rate": 4.997110780816887e-05, "loss": 2.7592, "step": 26588 }, { "epoch": 1.6505680054627847, "grad_norm": 0.602316668504028, "learning_rate": 4.996749628467046e-05, "loss": 2.7969, "step": 26589 }, { "epoch": 1.6506300825625426, "grad_norm": 0.6218333129712403, "learning_rate": 4.996388476134162e-05, "loss": 2.671, "step": 26590 }, { "epoch": 1.6506921596623005, "grad_norm": 0.635055593137121, "learning_rate": 4.996027323820121e-05, "loss": 2.7827, "step": 26591 }, { "epoch": 1.6507542367620585, "grad_norm": 0.6178665047612771, "learning_rate": 4.9956661715268064e-05, "loss": 2.7059, "step": 26592 }, { "epoch": 1.6508163138618164, "grad_norm": 0.6551230609110315, "learning_rate": 4.995305019256102e-05, "loss": 2.6588, "step": 26593 }, { "epoch": 1.6508783909615743, "grad_norm": 0.6400380120313892, "learning_rate": 4.994943867009893e-05, "loss": 2.7483, "step": 26594 }, { "epoch": 1.6509404680613322, "grad_norm": 0.6313784091006275, "learning_rate": 4.994582714790062e-05, "loss": 2.8049, "step": 26595 }, { "epoch": 1.6510025451610901, "grad_norm": 0.7308590911913616, "learning_rate": 4.994221562598495e-05, "loss": 2.7315, "step": 26596 }, { "epoch": 1.651064622260848, "grad_norm": 0.6644001046476907, "learning_rate": 4.9938604104370737e-05, "loss": 2.8252, "step": 26597 }, { "epoch": 1.651126699360606, "grad_norm": 0.6271227550683112, "learning_rate": 4.9934992583076854e-05, "loss": 2.8886, "step": 26598 }, { "epoch": 1.6511887764603639, "grad_norm": 0.6855040751516247, "learning_rate": 4.9931381062122144e-05, "loss": 2.8272, "step": 26599 }, { "epoch": 1.6512508535601218, "grad_norm": 0.5839966482448414, "learning_rate": 4.9927769541525424e-05, "loss": 2.7466, "step": 26600 }, { "epoch": 1.6513129306598797, "grad_norm": 0.6684834911830192, "learning_rate": 4.992415802130555e-05, "loss": 2.8448, "step": 26601 }, { "epoch": 1.6513750077596374, "grad_norm": 0.5452576289948462, "learning_rate": 4.992054650148136e-05, "loss": 2.7691, "step": 26602 }, { "epoch": 1.6514370848593953, "grad_norm": 0.6647676874887093, "learning_rate": 4.9916934982071706e-05, "loss": 2.8391, "step": 26603 }, { "epoch": 1.6514991619591533, "grad_norm": 0.5616427618207092, "learning_rate": 4.9913323463095414e-05, "loss": 2.7589, "step": 26604 }, { "epoch": 1.6515612390589112, "grad_norm": 0.6096869183903691, "learning_rate": 4.9909711944571343e-05, "loss": 2.8601, "step": 26605 }, { "epoch": 1.651623316158669, "grad_norm": 0.5397264289621118, "learning_rate": 4.990610042651832e-05, "loss": 2.8199, "step": 26606 }, { "epoch": 1.651685393258427, "grad_norm": 0.5944017798150038, "learning_rate": 4.990248890895519e-05, "loss": 2.7412, "step": 26607 }, { "epoch": 1.6517474703581847, "grad_norm": 0.5836278790525276, "learning_rate": 4.9898877391900814e-05, "loss": 2.7179, "step": 26608 }, { "epoch": 1.6518095474579426, "grad_norm": 0.6039603363333739, "learning_rate": 4.9895265875374006e-05, "loss": 2.8022, "step": 26609 }, { "epoch": 1.6518716245577005, "grad_norm": 0.5680440178574884, "learning_rate": 4.989165435939362e-05, "loss": 2.7602, "step": 26610 }, { "epoch": 1.6519337016574585, "grad_norm": 0.5254823682587407, "learning_rate": 4.988804284397851e-05, "loss": 2.7441, "step": 26611 }, { "epoch": 1.6519957787572164, "grad_norm": 0.5788859179757743, "learning_rate": 4.98844313291475e-05, "loss": 2.7742, "step": 26612 }, { "epoch": 1.6520578558569743, "grad_norm": 0.5191556677206823, "learning_rate": 4.988081981491944e-05, "loss": 2.6419, "step": 26613 }, { "epoch": 1.6521199329567322, "grad_norm": 0.5302181882510748, "learning_rate": 4.987720830131318e-05, "loss": 2.6966, "step": 26614 }, { "epoch": 1.6521820100564901, "grad_norm": 0.5572412563115478, "learning_rate": 4.9873596788347535e-05, "loss": 2.7855, "step": 26615 }, { "epoch": 1.652244087156248, "grad_norm": 0.5737428055106477, "learning_rate": 4.9869985276041384e-05, "loss": 2.693, "step": 26616 }, { "epoch": 1.652306164256006, "grad_norm": 0.5181242742406826, "learning_rate": 4.986637376441356e-05, "loss": 2.7857, "step": 26617 }, { "epoch": 1.6523682413557639, "grad_norm": 0.5679785642340046, "learning_rate": 4.986276225348287e-05, "loss": 2.8101, "step": 26618 }, { "epoch": 1.6524303184555218, "grad_norm": 0.5613735363774914, "learning_rate": 4.98591507432682e-05, "loss": 2.7626, "step": 26619 }, { "epoch": 1.6524923955552797, "grad_norm": 0.5397881861576583, "learning_rate": 4.985553923378837e-05, "loss": 2.7249, "step": 26620 }, { "epoch": 1.6525544726550376, "grad_norm": 0.5506845213450496, "learning_rate": 4.985192772506223e-05, "loss": 2.7534, "step": 26621 }, { "epoch": 1.6526165497547955, "grad_norm": 0.5644589064146907, "learning_rate": 4.984831621710861e-05, "loss": 2.7725, "step": 26622 }, { "epoch": 1.6526786268545535, "grad_norm": 0.5376813305451302, "learning_rate": 4.984470470994637e-05, "loss": 2.8016, "step": 26623 }, { "epoch": 1.6527407039543114, "grad_norm": 0.5589255034703332, "learning_rate": 4.984109320359434e-05, "loss": 2.6848, "step": 26624 }, { "epoch": 1.6528027810540693, "grad_norm": 0.5836689585236425, "learning_rate": 4.983748169807137e-05, "loss": 2.8185, "step": 26625 }, { "epoch": 1.652864858153827, "grad_norm": 0.5374680032743213, "learning_rate": 4.9833870193396286e-05, "loss": 2.6921, "step": 26626 }, { "epoch": 1.652926935253585, "grad_norm": 0.5845847217582298, "learning_rate": 4.9830258689587955e-05, "loss": 2.7532, "step": 26627 }, { "epoch": 1.6529890123533428, "grad_norm": 0.5654390956906323, "learning_rate": 4.9826647186665194e-05, "loss": 2.7499, "step": 26628 }, { "epoch": 1.6530510894531008, "grad_norm": 0.5624170772042102, "learning_rate": 4.982303568464687e-05, "loss": 2.7071, "step": 26629 }, { "epoch": 1.6531131665528587, "grad_norm": 0.6281148302384197, "learning_rate": 4.981942418355178e-05, "loss": 2.7294, "step": 26630 }, { "epoch": 1.6531752436526166, "grad_norm": 0.5485916171118013, "learning_rate": 4.981581268339883e-05, "loss": 2.6423, "step": 26631 }, { "epoch": 1.6532373207523743, "grad_norm": 0.6208291599807597, "learning_rate": 4.9812201184206816e-05, "loss": 2.7679, "step": 26632 }, { "epoch": 1.6532993978521322, "grad_norm": 0.5494393579848448, "learning_rate": 4.98085896859946e-05, "loss": 2.7814, "step": 26633 }, { "epoch": 1.6533614749518901, "grad_norm": 0.6035385621289155, "learning_rate": 4.980497818878101e-05, "loss": 2.8335, "step": 26634 }, { "epoch": 1.653423552051648, "grad_norm": 0.5501230243161122, "learning_rate": 4.980136669258491e-05, "loss": 2.7323, "step": 26635 }, { "epoch": 1.653485629151406, "grad_norm": 0.5615657093888301, "learning_rate": 4.979775519742512e-05, "loss": 2.6768, "step": 26636 }, { "epoch": 1.6535477062511639, "grad_norm": 0.5699421189434011, "learning_rate": 4.97941437033205e-05, "loss": 2.8053, "step": 26637 }, { "epoch": 1.6536097833509218, "grad_norm": 0.5819756398100666, "learning_rate": 4.979053221028987e-05, "loss": 2.735, "step": 26638 }, { "epoch": 1.6536718604506797, "grad_norm": 0.5514253948430418, "learning_rate": 4.978692071835209e-05, "loss": 2.694, "step": 26639 }, { "epoch": 1.6537339375504376, "grad_norm": 0.5701525215576904, "learning_rate": 4.978330922752599e-05, "loss": 2.776, "step": 26640 }, { "epoch": 1.6537960146501955, "grad_norm": 0.5487326270809656, "learning_rate": 4.9779697737830436e-05, "loss": 2.8158, "step": 26641 }, { "epoch": 1.6538580917499535, "grad_norm": 0.5422555094782434, "learning_rate": 4.9776086249284244e-05, "loss": 2.5978, "step": 26642 }, { "epoch": 1.6539201688497114, "grad_norm": 0.656287737170068, "learning_rate": 4.977247476190625e-05, "loss": 2.8026, "step": 26643 }, { "epoch": 1.6539822459494693, "grad_norm": 0.6260092930561011, "learning_rate": 4.9768863275715326e-05, "loss": 2.7711, "step": 26644 }, { "epoch": 1.6540443230492272, "grad_norm": 0.6342872342813474, "learning_rate": 4.976525179073029e-05, "loss": 2.7361, "step": 26645 }, { "epoch": 1.6541064001489851, "grad_norm": 0.6044554814794054, "learning_rate": 4.976164030697e-05, "loss": 2.7504, "step": 26646 }, { "epoch": 1.654168477248743, "grad_norm": 0.6284803761140043, "learning_rate": 4.975802882445328e-05, "loss": 2.7227, "step": 26647 }, { "epoch": 1.654230554348501, "grad_norm": 0.6073347317413968, "learning_rate": 4.975441734319899e-05, "loss": 2.679, "step": 26648 }, { "epoch": 1.6542926314482589, "grad_norm": 0.7015356963544924, "learning_rate": 4.9750805863225966e-05, "loss": 2.8222, "step": 26649 }, { "epoch": 1.6543547085480166, "grad_norm": 0.5406161048466084, "learning_rate": 4.974719438455305e-05, "loss": 2.7149, "step": 26650 }, { "epoch": 1.6544167856477745, "grad_norm": 0.6044132754288014, "learning_rate": 4.9743582907199066e-05, "loss": 2.6946, "step": 26651 }, { "epoch": 1.6544788627475324, "grad_norm": 0.6267138110555637, "learning_rate": 4.973997143118289e-05, "loss": 2.8325, "step": 26652 }, { "epoch": 1.6545409398472903, "grad_norm": 0.6152626394409684, "learning_rate": 4.9736359956523335e-05, "loss": 2.7355, "step": 26653 }, { "epoch": 1.6546030169470483, "grad_norm": 0.6686481790481152, "learning_rate": 4.9732748483239266e-05, "loss": 2.8005, "step": 26654 }, { "epoch": 1.6546650940468062, "grad_norm": 0.597413007316959, "learning_rate": 4.972913701134951e-05, "loss": 2.8785, "step": 26655 }, { "epoch": 1.6547271711465639, "grad_norm": 0.6229689020138017, "learning_rate": 4.97255255408729e-05, "loss": 2.7366, "step": 26656 }, { "epoch": 1.6547892482463218, "grad_norm": 0.6087797877095048, "learning_rate": 4.97219140718283e-05, "loss": 2.7647, "step": 26657 }, { "epoch": 1.6548513253460797, "grad_norm": 0.703051923557149, "learning_rate": 4.971830260423454e-05, "loss": 2.7516, "step": 26658 }, { "epoch": 1.6549134024458376, "grad_norm": 0.589641334210395, "learning_rate": 4.9714691138110466e-05, "loss": 2.8516, "step": 26659 }, { "epoch": 1.6549754795455955, "grad_norm": 0.6162676173046135, "learning_rate": 4.971107967347491e-05, "loss": 2.7085, "step": 26660 }, { "epoch": 1.6550375566453535, "grad_norm": 0.5952395276779014, "learning_rate": 4.970746821034673e-05, "loss": 2.6869, "step": 26661 }, { "epoch": 1.6550996337451114, "grad_norm": 0.620943423138778, "learning_rate": 4.970385674874474e-05, "loss": 2.7757, "step": 26662 }, { "epoch": 1.6551617108448693, "grad_norm": 0.5436076515045216, "learning_rate": 4.970024528868781e-05, "loss": 2.7376, "step": 26663 }, { "epoch": 1.6552237879446272, "grad_norm": 0.6838978554182346, "learning_rate": 4.9696633830194784e-05, "loss": 2.8043, "step": 26664 }, { "epoch": 1.6552858650443851, "grad_norm": 0.6227734722230039, "learning_rate": 4.9693022373284486e-05, "loss": 2.7917, "step": 26665 }, { "epoch": 1.655347942144143, "grad_norm": 0.5854528584498386, "learning_rate": 4.968941091797577e-05, "loss": 2.6774, "step": 26666 }, { "epoch": 1.655410019243901, "grad_norm": 0.6243919036914, "learning_rate": 4.968579946428748e-05, "loss": 2.7219, "step": 26667 }, { "epoch": 1.6554720963436589, "grad_norm": 0.5859509784419815, "learning_rate": 4.968218801223844e-05, "loss": 2.7646, "step": 26668 }, { "epoch": 1.6555341734434168, "grad_norm": 0.5716288803038956, "learning_rate": 4.96785765618475e-05, "loss": 2.8025, "step": 26669 }, { "epoch": 1.6555962505431747, "grad_norm": 0.5677659387555877, "learning_rate": 4.96749651131335e-05, "loss": 2.7313, "step": 26670 }, { "epoch": 1.6556583276429326, "grad_norm": 0.5369579864774742, "learning_rate": 4.96713536661153e-05, "loss": 2.8581, "step": 26671 }, { "epoch": 1.6557204047426906, "grad_norm": 0.6002302155306505, "learning_rate": 4.966774222081171e-05, "loss": 2.7765, "step": 26672 }, { "epoch": 1.6557824818424485, "grad_norm": 0.5854657881096552, "learning_rate": 4.966413077724161e-05, "loss": 2.847, "step": 26673 }, { "epoch": 1.6558445589422062, "grad_norm": 0.5373252573516659, "learning_rate": 4.96605193354238e-05, "loss": 2.7518, "step": 26674 }, { "epoch": 1.655906636041964, "grad_norm": 0.5907872214390093, "learning_rate": 4.965690789537717e-05, "loss": 2.7847, "step": 26675 }, { "epoch": 1.655968713141722, "grad_norm": 0.614748299543985, "learning_rate": 4.965329645712051e-05, "loss": 2.7085, "step": 26676 }, { "epoch": 1.65603079024148, "grad_norm": 0.5516786501649913, "learning_rate": 4.9649685020672705e-05, "loss": 2.7227, "step": 26677 }, { "epoch": 1.6560928673412378, "grad_norm": 0.5616666660912548, "learning_rate": 4.964607358605256e-05, "loss": 2.644, "step": 26678 }, { "epoch": 1.6561549444409958, "grad_norm": 0.5911531377892831, "learning_rate": 4.9642462153278954e-05, "loss": 2.8008, "step": 26679 }, { "epoch": 1.6562170215407535, "grad_norm": 0.5573664433827096, "learning_rate": 4.96388507223707e-05, "loss": 2.7384, "step": 26680 }, { "epoch": 1.6562790986405114, "grad_norm": 0.5458255507615655, "learning_rate": 4.963523929334665e-05, "loss": 2.7481, "step": 26681 }, { "epoch": 1.6563411757402693, "grad_norm": 0.5889704219988487, "learning_rate": 4.9631627866225645e-05, "loss": 2.6926, "step": 26682 }, { "epoch": 1.6564032528400272, "grad_norm": 0.56155490952904, "learning_rate": 4.962801644102652e-05, "loss": 2.7862, "step": 26683 }, { "epoch": 1.6564653299397851, "grad_norm": 0.5391737228867316, "learning_rate": 4.962440501776814e-05, "loss": 2.7208, "step": 26684 }, { "epoch": 1.656527407039543, "grad_norm": 0.5314469252971612, "learning_rate": 4.962079359646931e-05, "loss": 2.6232, "step": 26685 }, { "epoch": 1.656589484139301, "grad_norm": 0.5441573875019713, "learning_rate": 4.9617182177148905e-05, "loss": 2.7808, "step": 26686 }, { "epoch": 1.6566515612390589, "grad_norm": 0.6069621465324069, "learning_rate": 4.961357075982574e-05, "loss": 2.7625, "step": 26687 }, { "epoch": 1.6567136383388168, "grad_norm": 0.5360894246428602, "learning_rate": 4.9609959344518693e-05, "loss": 2.7589, "step": 26688 }, { "epoch": 1.6567757154385747, "grad_norm": 0.5358824710009267, "learning_rate": 4.9606347931246564e-05, "loss": 2.8097, "step": 26689 }, { "epoch": 1.6568377925383326, "grad_norm": 0.548411309467314, "learning_rate": 4.9602736520028225e-05, "loss": 2.8183, "step": 26690 }, { "epoch": 1.6568998696380905, "grad_norm": 0.5362450598553651, "learning_rate": 4.959912511088251e-05, "loss": 2.7963, "step": 26691 }, { "epoch": 1.6569619467378485, "grad_norm": 0.5053677028966104, "learning_rate": 4.9595513703828236e-05, "loss": 2.8033, "step": 26692 }, { "epoch": 1.6570240238376064, "grad_norm": 0.5689484993936716, "learning_rate": 4.9591902298884284e-05, "loss": 2.7719, "step": 26693 }, { "epoch": 1.6570861009373643, "grad_norm": 0.48835674465205214, "learning_rate": 4.958829089606947e-05, "loss": 2.687, "step": 26694 }, { "epoch": 1.6571481780371222, "grad_norm": 0.5656931367162785, "learning_rate": 4.958467949540262e-05, "loss": 2.6979, "step": 26695 }, { "epoch": 1.6572102551368801, "grad_norm": 0.5493537313707995, "learning_rate": 4.958106809690263e-05, "loss": 2.7816, "step": 26696 }, { "epoch": 1.657272332236638, "grad_norm": 0.549105889565946, "learning_rate": 4.9577456700588287e-05, "loss": 2.8423, "step": 26697 }, { "epoch": 1.6573344093363958, "grad_norm": 0.5642170395847659, "learning_rate": 4.9573845306478475e-05, "loss": 2.7202, "step": 26698 }, { "epoch": 1.6573964864361537, "grad_norm": 0.549281248874928, "learning_rate": 4.9570233914592e-05, "loss": 2.8172, "step": 26699 }, { "epoch": 1.6574585635359116, "grad_norm": 0.49436696715164563, "learning_rate": 4.956662252494773e-05, "loss": 2.6879, "step": 26700 }, { "epoch": 1.6575206406356695, "grad_norm": 0.5574328222984182, "learning_rate": 4.956301113756449e-05, "loss": 2.7846, "step": 26701 }, { "epoch": 1.6575827177354274, "grad_norm": 0.5411311277631649, "learning_rate": 4.955939975246113e-05, "loss": 2.7523, "step": 26702 }, { "epoch": 1.6576447948351853, "grad_norm": 0.6107900768470731, "learning_rate": 4.95557883696565e-05, "loss": 2.7838, "step": 26703 }, { "epoch": 1.657706871934943, "grad_norm": 0.5102168457590104, "learning_rate": 4.955217698916941e-05, "loss": 2.6956, "step": 26704 }, { "epoch": 1.657768949034701, "grad_norm": 0.5805479260094906, "learning_rate": 4.9548565611018736e-05, "loss": 2.6752, "step": 26705 }, { "epoch": 1.6578310261344589, "grad_norm": 0.5661098362329412, "learning_rate": 4.954495423522329e-05, "loss": 2.7796, "step": 26706 }, { "epoch": 1.6578931032342168, "grad_norm": 0.5235227828510651, "learning_rate": 4.954134286180194e-05, "loss": 2.7943, "step": 26707 }, { "epoch": 1.6579551803339747, "grad_norm": 0.5676983532042924, "learning_rate": 4.95377314907735e-05, "loss": 2.7185, "step": 26708 }, { "epoch": 1.6580172574337326, "grad_norm": 0.5135630503207366, "learning_rate": 4.9534120122156854e-05, "loss": 2.8709, "step": 26709 }, { "epoch": 1.6580793345334905, "grad_norm": 0.6500689424119065, "learning_rate": 4.953050875597079e-05, "loss": 2.7666, "step": 26710 }, { "epoch": 1.6581414116332485, "grad_norm": 0.5338725544313702, "learning_rate": 4.952689739223419e-05, "loss": 2.8194, "step": 26711 }, { "epoch": 1.6582034887330064, "grad_norm": 0.5831034724032539, "learning_rate": 4.9523286030965874e-05, "loss": 2.7425, "step": 26712 }, { "epoch": 1.6582655658327643, "grad_norm": 0.5570400478101636, "learning_rate": 4.95196746721847e-05, "loss": 2.7982, "step": 26713 }, { "epoch": 1.6583276429325222, "grad_norm": 0.5847710251797715, "learning_rate": 4.9516063315909487e-05, "loss": 2.7187, "step": 26714 }, { "epoch": 1.6583897200322801, "grad_norm": 0.5532047171179901, "learning_rate": 4.95124519621591e-05, "loss": 2.6769, "step": 26715 }, { "epoch": 1.658451797132038, "grad_norm": 0.5841733696843675, "learning_rate": 4.9508840610952375e-05, "loss": 2.8106, "step": 26716 }, { "epoch": 1.658513874231796, "grad_norm": 0.5619595633532652, "learning_rate": 4.9505229262308134e-05, "loss": 2.8344, "step": 26717 }, { "epoch": 1.6585759513315539, "grad_norm": 0.5607330332147992, "learning_rate": 4.950161791624524e-05, "loss": 2.8238, "step": 26718 }, { "epoch": 1.6586380284313118, "grad_norm": 0.5647181380488504, "learning_rate": 4.949800657278252e-05, "loss": 2.6977, "step": 26719 }, { "epoch": 1.6587001055310697, "grad_norm": 0.6086040657550119, "learning_rate": 4.949439523193883e-05, "loss": 2.744, "step": 26720 }, { "epoch": 1.6587621826308276, "grad_norm": 0.5666058681867517, "learning_rate": 4.949078389373299e-05, "loss": 2.8075, "step": 26721 }, { "epoch": 1.6588242597305853, "grad_norm": 0.5683435916347727, "learning_rate": 4.948717255818386e-05, "loss": 2.6778, "step": 26722 }, { "epoch": 1.6588863368303433, "grad_norm": 0.663498829986472, "learning_rate": 4.948356122531028e-05, "loss": 2.7757, "step": 26723 }, { "epoch": 1.6589484139301012, "grad_norm": 0.490596774072923, "learning_rate": 4.9479949895131084e-05, "loss": 2.7013, "step": 26724 }, { "epoch": 1.659010491029859, "grad_norm": 0.6332229413479915, "learning_rate": 4.9476338567665106e-05, "loss": 2.7897, "step": 26725 }, { "epoch": 1.659072568129617, "grad_norm": 0.5366665210881638, "learning_rate": 4.9472727242931216e-05, "loss": 2.8595, "step": 26726 }, { "epoch": 1.659134645229375, "grad_norm": 0.5803035978621511, "learning_rate": 4.9469115920948225e-05, "loss": 2.7353, "step": 26727 }, { "epoch": 1.6591967223291326, "grad_norm": 0.4899380460809817, "learning_rate": 4.946550460173498e-05, "loss": 2.7686, "step": 26728 }, { "epoch": 1.6592587994288905, "grad_norm": 0.5397240546061183, "learning_rate": 4.9461893285310325e-05, "loss": 2.7637, "step": 26729 }, { "epoch": 1.6593208765286485, "grad_norm": 0.5096979205892356, "learning_rate": 4.945828197169311e-05, "loss": 2.8832, "step": 26730 }, { "epoch": 1.6593829536284064, "grad_norm": 0.6296390268680537, "learning_rate": 4.945467066090217e-05, "loss": 2.9042, "step": 26731 }, { "epoch": 1.6594450307281643, "grad_norm": 0.4762053480266967, "learning_rate": 4.9451059352956345e-05, "loss": 2.7196, "step": 26732 }, { "epoch": 1.6595071078279222, "grad_norm": 0.5387229402831563, "learning_rate": 4.944744804787448e-05, "loss": 2.7132, "step": 26733 }, { "epoch": 1.6595691849276801, "grad_norm": 0.5747570447137419, "learning_rate": 4.9443836745675414e-05, "loss": 2.7397, "step": 26734 }, { "epoch": 1.659631262027438, "grad_norm": 0.529409586599644, "learning_rate": 4.944022544637798e-05, "loss": 2.8453, "step": 26735 }, { "epoch": 1.659693339127196, "grad_norm": 0.5877967582107688, "learning_rate": 4.943661415000103e-05, "loss": 2.8699, "step": 26736 }, { "epoch": 1.6597554162269539, "grad_norm": 0.5285405459734812, "learning_rate": 4.9433002856563394e-05, "loss": 2.6853, "step": 26737 }, { "epoch": 1.6598174933267118, "grad_norm": 0.5788652217601739, "learning_rate": 4.942939156608393e-05, "loss": 2.7635, "step": 26738 }, { "epoch": 1.6598795704264697, "grad_norm": 0.558675586601158, "learning_rate": 4.9425780278581454e-05, "loss": 2.8034, "step": 26739 }, { "epoch": 1.6599416475262276, "grad_norm": 0.5387557165723242, "learning_rate": 4.942216899407484e-05, "loss": 2.7777, "step": 26740 }, { "epoch": 1.6600037246259856, "grad_norm": 0.5873246113753426, "learning_rate": 4.941855771258291e-05, "loss": 2.6794, "step": 26741 }, { "epoch": 1.6600658017257435, "grad_norm": 0.571614151753905, "learning_rate": 4.941494643412449e-05, "loss": 2.7735, "step": 26742 }, { "epoch": 1.6601278788255014, "grad_norm": 0.6200022192884533, "learning_rate": 4.941133515871845e-05, "loss": 2.6473, "step": 26743 }, { "epoch": 1.6601899559252593, "grad_norm": 0.5632126849767117, "learning_rate": 4.940772388638361e-05, "loss": 2.6898, "step": 26744 }, { "epoch": 1.6602520330250172, "grad_norm": 0.6173994448870309, "learning_rate": 4.940411261713882e-05, "loss": 2.8485, "step": 26745 }, { "epoch": 1.660314110124775, "grad_norm": 0.6077807430414481, "learning_rate": 4.940050135100292e-05, "loss": 2.7821, "step": 26746 }, { "epoch": 1.6603761872245328, "grad_norm": 0.6362987139176712, "learning_rate": 4.939689008799475e-05, "loss": 2.814, "step": 26747 }, { "epoch": 1.6604382643242908, "grad_norm": 0.5916366612074805, "learning_rate": 4.939327882813315e-05, "loss": 2.7583, "step": 26748 }, { "epoch": 1.6605003414240487, "grad_norm": 0.6536526597313732, "learning_rate": 4.938966757143696e-05, "loss": 2.6951, "step": 26749 }, { "epoch": 1.6605624185238066, "grad_norm": 0.5560708057357376, "learning_rate": 4.9386056317925026e-05, "loss": 2.7241, "step": 26750 }, { "epoch": 1.6606244956235645, "grad_norm": 0.6496454959445197, "learning_rate": 4.938244506761619e-05, "loss": 2.7316, "step": 26751 }, { "epoch": 1.6606865727233222, "grad_norm": 0.5667373478743052, "learning_rate": 4.937883382052928e-05, "loss": 2.6729, "step": 26752 }, { "epoch": 1.6607486498230801, "grad_norm": 0.6295917636989397, "learning_rate": 4.9375222576683146e-05, "loss": 2.713, "step": 26753 }, { "epoch": 1.660810726922838, "grad_norm": 0.5504689785569491, "learning_rate": 4.937161133609663e-05, "loss": 2.7453, "step": 26754 }, { "epoch": 1.660872804022596, "grad_norm": 0.5805544968290679, "learning_rate": 4.9368000098788564e-05, "loss": 2.6979, "step": 26755 }, { "epoch": 1.6609348811223539, "grad_norm": 0.5458443056949909, "learning_rate": 4.9364388864777805e-05, "loss": 2.7326, "step": 26756 }, { "epoch": 1.6609969582221118, "grad_norm": 0.5900622588380301, "learning_rate": 4.936077763408317e-05, "loss": 2.7884, "step": 26757 }, { "epoch": 1.6610590353218697, "grad_norm": 0.5568588364521277, "learning_rate": 4.9357166406723526e-05, "loss": 2.7736, "step": 26758 }, { "epoch": 1.6611211124216276, "grad_norm": 0.5586208790413502, "learning_rate": 4.935355518271769e-05, "loss": 2.8465, "step": 26759 }, { "epoch": 1.6611831895213856, "grad_norm": 0.5824082452930532, "learning_rate": 4.934994396208452e-05, "loss": 2.7145, "step": 26760 }, { "epoch": 1.6612452666211435, "grad_norm": 0.5792470044966992, "learning_rate": 4.934633274484284e-05, "loss": 2.8439, "step": 26761 }, { "epoch": 1.6613073437209014, "grad_norm": 0.6233289791488886, "learning_rate": 4.9342721531011496e-05, "loss": 2.7386, "step": 26762 }, { "epoch": 1.6613694208206593, "grad_norm": 0.5687408477969061, "learning_rate": 4.9339110320609354e-05, "loss": 2.7223, "step": 26763 }, { "epoch": 1.6614314979204172, "grad_norm": 0.5802186224234589, "learning_rate": 4.9335499113655226e-05, "loss": 2.7762, "step": 26764 }, { "epoch": 1.6614935750201751, "grad_norm": 0.6024408986953178, "learning_rate": 4.933188791016797e-05, "loss": 2.7642, "step": 26765 }, { "epoch": 1.661555652119933, "grad_norm": 0.5813426866772331, "learning_rate": 4.932827671016641e-05, "loss": 2.8087, "step": 26766 }, { "epoch": 1.661617729219691, "grad_norm": 0.558450160211251, "learning_rate": 4.9324665513669385e-05, "loss": 2.8077, "step": 26767 }, { "epoch": 1.661679806319449, "grad_norm": 0.5386465651753631, "learning_rate": 4.932105432069576e-05, "loss": 2.7836, "step": 26768 }, { "epoch": 1.6617418834192068, "grad_norm": 0.6084444570511001, "learning_rate": 4.931744313126434e-05, "loss": 2.8174, "step": 26769 }, { "epoch": 1.6618039605189645, "grad_norm": 0.5359789240095377, "learning_rate": 4.9313831945394e-05, "loss": 2.6899, "step": 26770 }, { "epoch": 1.6618660376187224, "grad_norm": 0.6355342261164099, "learning_rate": 4.931022076310356e-05, "loss": 2.7394, "step": 26771 }, { "epoch": 1.6619281147184803, "grad_norm": 0.5827363115271588, "learning_rate": 4.9306609584411864e-05, "loss": 2.7331, "step": 26772 }, { "epoch": 1.6619901918182383, "grad_norm": 0.5965295530945849, "learning_rate": 4.930299840933776e-05, "loss": 2.8529, "step": 26773 }, { "epoch": 1.6620522689179962, "grad_norm": 0.6011844289161675, "learning_rate": 4.929938723790008e-05, "loss": 2.6996, "step": 26774 }, { "epoch": 1.662114346017754, "grad_norm": 0.5833935252424386, "learning_rate": 4.929577607011766e-05, "loss": 2.8075, "step": 26775 }, { "epoch": 1.6621764231175118, "grad_norm": 0.6616943994430649, "learning_rate": 4.929216490600936e-05, "loss": 2.8074, "step": 26776 }, { "epoch": 1.6622385002172697, "grad_norm": 0.5545308066393471, "learning_rate": 4.928855374559401e-05, "loss": 2.698, "step": 26777 }, { "epoch": 1.6623005773170276, "grad_norm": 0.5643714132860342, "learning_rate": 4.928494258889043e-05, "loss": 2.6972, "step": 26778 }, { "epoch": 1.6623626544167855, "grad_norm": 0.5546392231738748, "learning_rate": 4.928133143591749e-05, "loss": 2.6828, "step": 26779 }, { "epoch": 1.6624247315165435, "grad_norm": 0.5889417053950178, "learning_rate": 4.927772028669401e-05, "loss": 2.7942, "step": 26780 }, { "epoch": 1.6624868086163014, "grad_norm": 0.5972076677303464, "learning_rate": 4.927410914123885e-05, "loss": 2.7761, "step": 26781 }, { "epoch": 1.6625488857160593, "grad_norm": 0.5514995409340214, "learning_rate": 4.927049799957082e-05, "loss": 2.8044, "step": 26782 }, { "epoch": 1.6626109628158172, "grad_norm": 0.6155797806218956, "learning_rate": 4.92668868617088e-05, "loss": 2.7505, "step": 26783 }, { "epoch": 1.6626730399155751, "grad_norm": 0.5680885325059559, "learning_rate": 4.92632757276716e-05, "loss": 2.7576, "step": 26784 }, { "epoch": 1.662735117015333, "grad_norm": 0.5866391621946212, "learning_rate": 4.9259664597478074e-05, "loss": 2.7807, "step": 26785 }, { "epoch": 1.662797194115091, "grad_norm": 0.5783152653796961, "learning_rate": 4.925605347114704e-05, "loss": 2.7488, "step": 26786 }, { "epoch": 1.6628592712148489, "grad_norm": 0.6512991713853178, "learning_rate": 4.925244234869738e-05, "loss": 2.7425, "step": 26787 }, { "epoch": 1.6629213483146068, "grad_norm": 0.5391695496788163, "learning_rate": 4.9248831230147905e-05, "loss": 2.7568, "step": 26788 }, { "epoch": 1.6629834254143647, "grad_norm": 0.6532883115951086, "learning_rate": 4.924522011551745e-05, "loss": 2.7133, "step": 26789 }, { "epoch": 1.6630455025141226, "grad_norm": 0.5737738670005846, "learning_rate": 4.9241609004824873e-05, "loss": 2.8064, "step": 26790 }, { "epoch": 1.6631075796138806, "grad_norm": 0.6111812476046936, "learning_rate": 4.9237997898088994e-05, "loss": 2.7697, "step": 26791 }, { "epoch": 1.6631696567136385, "grad_norm": 0.6382901745549314, "learning_rate": 4.923438679532868e-05, "loss": 2.692, "step": 26792 }, { "epoch": 1.6632317338133964, "grad_norm": 0.6220511728453374, "learning_rate": 4.923077569656273e-05, "loss": 2.7605, "step": 26793 }, { "epoch": 1.663293810913154, "grad_norm": 0.6216027195702224, "learning_rate": 4.922716460181002e-05, "loss": 2.761, "step": 26794 }, { "epoch": 1.663355888012912, "grad_norm": 0.636737472165313, "learning_rate": 4.922355351108939e-05, "loss": 2.7977, "step": 26795 }, { "epoch": 1.66341796511267, "grad_norm": 0.649806496354232, "learning_rate": 4.921994242441967e-05, "loss": 2.7482, "step": 26796 }, { "epoch": 1.6634800422124278, "grad_norm": 0.5678852109064524, "learning_rate": 4.92163313418197e-05, "loss": 2.7882, "step": 26797 }, { "epoch": 1.6635421193121858, "grad_norm": 0.582016119024234, "learning_rate": 4.921272026330831e-05, "loss": 2.839, "step": 26798 }, { "epoch": 1.6636041964119437, "grad_norm": 0.6172280358631962, "learning_rate": 4.920910918890436e-05, "loss": 2.7786, "step": 26799 }, { "epoch": 1.6636662735117014, "grad_norm": 0.589933754830769, "learning_rate": 4.9205498118626676e-05, "loss": 2.7174, "step": 26800 }, { "epoch": 1.6637283506114593, "grad_norm": 0.5938502528229982, "learning_rate": 4.9201887052494106e-05, "loss": 2.7295, "step": 26801 }, { "epoch": 1.6637904277112172, "grad_norm": 0.5940577311878746, "learning_rate": 4.919827599052548e-05, "loss": 2.7916, "step": 26802 }, { "epoch": 1.6638525048109751, "grad_norm": 0.5804352864904784, "learning_rate": 4.9194664932739644e-05, "loss": 2.7528, "step": 26803 }, { "epoch": 1.663914581910733, "grad_norm": 0.6279653298806185, "learning_rate": 4.919105387915544e-05, "loss": 2.8305, "step": 26804 }, { "epoch": 1.663976659010491, "grad_norm": 0.5570818306962801, "learning_rate": 4.91874428297917e-05, "loss": 2.8031, "step": 26805 }, { "epoch": 1.6640387361102489, "grad_norm": 0.6348989493398164, "learning_rate": 4.9183831784667266e-05, "loss": 2.6883, "step": 26806 }, { "epoch": 1.6641008132100068, "grad_norm": 0.565483630939283, "learning_rate": 4.918022074380098e-05, "loss": 2.7338, "step": 26807 }, { "epoch": 1.6641628903097647, "grad_norm": 0.6278134678956124, "learning_rate": 4.9176609707211685e-05, "loss": 2.761, "step": 26808 }, { "epoch": 1.6642249674095226, "grad_norm": 0.5700705550680413, "learning_rate": 4.917299867491822e-05, "loss": 2.7537, "step": 26809 }, { "epoch": 1.6642870445092806, "grad_norm": 0.6603127694659715, "learning_rate": 4.916938764693941e-05, "loss": 2.7289, "step": 26810 }, { "epoch": 1.6643491216090385, "grad_norm": 0.566812882305811, "learning_rate": 4.916577662329411e-05, "loss": 2.7592, "step": 26811 }, { "epoch": 1.6644111987087964, "grad_norm": 0.6775898408267248, "learning_rate": 4.916216560400116e-05, "loss": 2.7258, "step": 26812 }, { "epoch": 1.6644732758085543, "grad_norm": 0.5512534477921875, "learning_rate": 4.9158554589079405e-05, "loss": 2.8222, "step": 26813 }, { "epoch": 1.6645353529083122, "grad_norm": 0.674286484658711, "learning_rate": 4.915494357854766e-05, "loss": 2.7212, "step": 26814 }, { "epoch": 1.6645974300080701, "grad_norm": 0.5514073972400936, "learning_rate": 4.915133257242479e-05, "loss": 2.7607, "step": 26815 }, { "epoch": 1.664659507107828, "grad_norm": 0.6516013117878298, "learning_rate": 4.914772157072961e-05, "loss": 2.8311, "step": 26816 }, { "epoch": 1.664721584207586, "grad_norm": 0.5994405850595669, "learning_rate": 4.914411057348099e-05, "loss": 2.7635, "step": 26817 }, { "epoch": 1.6647836613073437, "grad_norm": 0.5780662300549618, "learning_rate": 4.9140499580697734e-05, "loss": 2.678, "step": 26818 }, { "epoch": 1.6648457384071016, "grad_norm": 0.5435618173595614, "learning_rate": 4.913688859239872e-05, "loss": 2.7455, "step": 26819 }, { "epoch": 1.6649078155068595, "grad_norm": 0.5707830813023731, "learning_rate": 4.913327760860275e-05, "loss": 2.6913, "step": 26820 }, { "epoch": 1.6649698926066174, "grad_norm": 0.6099992879879033, "learning_rate": 4.9129666629328694e-05, "loss": 2.6805, "step": 26821 }, { "epoch": 1.6650319697063753, "grad_norm": 0.572966345164857, "learning_rate": 4.912605565459537e-05, "loss": 2.6834, "step": 26822 }, { "epoch": 1.6650940468061333, "grad_norm": 0.6271175038309976, "learning_rate": 4.912244468442164e-05, "loss": 2.8133, "step": 26823 }, { "epoch": 1.665156123905891, "grad_norm": 0.5553470773402418, "learning_rate": 4.911883371882631e-05, "loss": 2.721, "step": 26824 }, { "epoch": 1.6652182010056489, "grad_norm": 0.6113145235638716, "learning_rate": 4.911522275782825e-05, "loss": 2.7898, "step": 26825 }, { "epoch": 1.6652802781054068, "grad_norm": 0.5613024732957613, "learning_rate": 4.911161180144627e-05, "loss": 2.7496, "step": 26826 }, { "epoch": 1.6653423552051647, "grad_norm": 0.572153386380916, "learning_rate": 4.910800084969925e-05, "loss": 2.6743, "step": 26827 }, { "epoch": 1.6654044323049226, "grad_norm": 0.5837736738415906, "learning_rate": 4.910438990260599e-05, "loss": 2.6674, "step": 26828 }, { "epoch": 1.6654665094046806, "grad_norm": 0.546274482169693, "learning_rate": 4.910077896018536e-05, "loss": 2.7616, "step": 26829 }, { "epoch": 1.6655285865044385, "grad_norm": 0.5977497650985224, "learning_rate": 4.909716802245617e-05, "loss": 2.8579, "step": 26830 }, { "epoch": 1.6655906636041964, "grad_norm": 0.586309137106794, "learning_rate": 4.909355708943728e-05, "loss": 2.774, "step": 26831 }, { "epoch": 1.6656527407039543, "grad_norm": 0.5289742306362095, "learning_rate": 4.908994616114753e-05, "loss": 2.7629, "step": 26832 }, { "epoch": 1.6657148178037122, "grad_norm": 0.6099172282735164, "learning_rate": 4.9086335237605746e-05, "loss": 2.6916, "step": 26833 }, { "epoch": 1.6657768949034701, "grad_norm": 0.547570059356742, "learning_rate": 4.9082724318830766e-05, "loss": 2.6552, "step": 26834 }, { "epoch": 1.665838972003228, "grad_norm": 0.573835439615911, "learning_rate": 4.907911340484145e-05, "loss": 2.7579, "step": 26835 }, { "epoch": 1.665901049102986, "grad_norm": 0.5785558229398825, "learning_rate": 4.90755024956566e-05, "loss": 2.8287, "step": 26836 }, { "epoch": 1.665963126202744, "grad_norm": 0.5470974333583511, "learning_rate": 4.90718915912951e-05, "loss": 2.7777, "step": 26837 }, { "epoch": 1.6660252033025018, "grad_norm": 0.5727695940996114, "learning_rate": 4.906828069177577e-05, "loss": 2.7433, "step": 26838 }, { "epoch": 1.6660872804022597, "grad_norm": 0.5538939319957276, "learning_rate": 4.906466979711744e-05, "loss": 2.7732, "step": 26839 }, { "epoch": 1.6661493575020176, "grad_norm": 0.6054440154899237, "learning_rate": 4.9061058907338944e-05, "loss": 2.7478, "step": 26840 }, { "epoch": 1.6662114346017756, "grad_norm": 0.545674013468586, "learning_rate": 4.905744802245913e-05, "loss": 2.7408, "step": 26841 }, { "epoch": 1.6662735117015333, "grad_norm": 0.6733468147342991, "learning_rate": 4.905383714249686e-05, "loss": 2.8449, "step": 26842 }, { "epoch": 1.6663355888012912, "grad_norm": 0.6156855790593586, "learning_rate": 4.905022626747093e-05, "loss": 2.7475, "step": 26843 }, { "epoch": 1.666397665901049, "grad_norm": 0.5456458857013908, "learning_rate": 4.904661539740021e-05, "loss": 2.8254, "step": 26844 }, { "epoch": 1.666459743000807, "grad_norm": 0.6475999317600339, "learning_rate": 4.9043004532303526e-05, "loss": 2.8205, "step": 26845 }, { "epoch": 1.666521820100565, "grad_norm": 0.6027205084282928, "learning_rate": 4.903939367219973e-05, "loss": 2.7992, "step": 26846 }, { "epoch": 1.6665838972003229, "grad_norm": 0.6064009105011866, "learning_rate": 4.903578281710763e-05, "loss": 2.8252, "step": 26847 }, { "epoch": 1.6666459743000805, "grad_norm": 0.6143951990138669, "learning_rate": 4.9032171967046107e-05, "loss": 2.7236, "step": 26848 }, { "epoch": 1.6667080513998385, "grad_norm": 0.57745951935881, "learning_rate": 4.902856112203396e-05, "loss": 2.7603, "step": 26849 }, { "epoch": 1.6667701284995964, "grad_norm": 0.6295296558730146, "learning_rate": 4.902495028209006e-05, "loss": 2.8145, "step": 26850 }, { "epoch": 1.6668322055993543, "grad_norm": 0.6118383989696798, "learning_rate": 4.9021339447233224e-05, "loss": 2.8043, "step": 26851 }, { "epoch": 1.6668942826991122, "grad_norm": 0.6320553948278317, "learning_rate": 4.9017728617482295e-05, "loss": 2.7483, "step": 26852 }, { "epoch": 1.6669563597988701, "grad_norm": 0.5264262093768114, "learning_rate": 4.9014117792856126e-05, "loss": 2.6973, "step": 26853 }, { "epoch": 1.667018436898628, "grad_norm": 0.5883701941134042, "learning_rate": 4.9010506973373526e-05, "loss": 2.7479, "step": 26854 }, { "epoch": 1.667080513998386, "grad_norm": 0.5259950132423701, "learning_rate": 4.900689615905336e-05, "loss": 2.6678, "step": 26855 }, { "epoch": 1.6671425910981439, "grad_norm": 0.5607453104674279, "learning_rate": 4.900328534991446e-05, "loss": 2.8375, "step": 26856 }, { "epoch": 1.6672046681979018, "grad_norm": 0.5063395632885224, "learning_rate": 4.899967454597566e-05, "loss": 2.7625, "step": 26857 }, { "epoch": 1.6672667452976597, "grad_norm": 0.6641468443319635, "learning_rate": 4.89960637472558e-05, "loss": 2.7438, "step": 26858 }, { "epoch": 1.6673288223974176, "grad_norm": 0.5341610251449653, "learning_rate": 4.89924529537737e-05, "loss": 2.7538, "step": 26859 }, { "epoch": 1.6673908994971756, "grad_norm": 0.6621898950998828, "learning_rate": 4.898884216554824e-05, "loss": 2.8136, "step": 26860 }, { "epoch": 1.6674529765969335, "grad_norm": 0.5639701545762522, "learning_rate": 4.898523138259823e-05, "loss": 2.6615, "step": 26861 }, { "epoch": 1.6675150536966914, "grad_norm": 0.5950077187814843, "learning_rate": 4.898162060494252e-05, "loss": 2.9013, "step": 26862 }, { "epoch": 1.6675771307964493, "grad_norm": 0.5932628490295946, "learning_rate": 4.897800983259995e-05, "loss": 2.7495, "step": 26863 }, { "epoch": 1.6676392078962072, "grad_norm": 0.6423753459029075, "learning_rate": 4.897439906558933e-05, "loss": 2.8713, "step": 26864 }, { "epoch": 1.6677012849959651, "grad_norm": 0.6499533113804491, "learning_rate": 4.897078830392954e-05, "loss": 2.7401, "step": 26865 }, { "epoch": 1.6677633620957228, "grad_norm": 0.6085908931327237, "learning_rate": 4.896717754763938e-05, "loss": 2.8029, "step": 26866 }, { "epoch": 1.6678254391954808, "grad_norm": 0.5691818107202523, "learning_rate": 4.896356679673772e-05, "loss": 2.8352, "step": 26867 }, { "epoch": 1.6678875162952387, "grad_norm": 0.5822529210647551, "learning_rate": 4.8959956051243364e-05, "loss": 2.8019, "step": 26868 }, { "epoch": 1.6679495933949966, "grad_norm": 0.5951364041497514, "learning_rate": 4.895634531117519e-05, "loss": 2.7098, "step": 26869 }, { "epoch": 1.6680116704947545, "grad_norm": 0.598308332163529, "learning_rate": 4.8952734576552006e-05, "loss": 2.6804, "step": 26870 }, { "epoch": 1.6680737475945124, "grad_norm": 0.604562413291645, "learning_rate": 4.8949123847392663e-05, "loss": 2.8726, "step": 26871 }, { "epoch": 1.6681358246942701, "grad_norm": 0.6062781307733432, "learning_rate": 4.894551312371599e-05, "loss": 2.6733, "step": 26872 }, { "epoch": 1.668197901794028, "grad_norm": 0.637307813135795, "learning_rate": 4.894190240554084e-05, "loss": 2.7326, "step": 26873 }, { "epoch": 1.668259978893786, "grad_norm": 0.6376776700113231, "learning_rate": 4.8938291692886037e-05, "loss": 2.7724, "step": 26874 }, { "epoch": 1.6683220559935439, "grad_norm": 0.6442533645379196, "learning_rate": 4.893468098577043e-05, "loss": 2.7816, "step": 26875 }, { "epoch": 1.6683841330933018, "grad_norm": 0.5388880461208837, "learning_rate": 4.893107028421285e-05, "loss": 2.8252, "step": 26876 }, { "epoch": 1.6684462101930597, "grad_norm": 0.5472546213529104, "learning_rate": 4.892745958823213e-05, "loss": 2.7722, "step": 26877 }, { "epoch": 1.6685082872928176, "grad_norm": 0.5845475814075601, "learning_rate": 4.8923848897847117e-05, "loss": 2.7822, "step": 26878 }, { "epoch": 1.6685703643925756, "grad_norm": 0.5521521045629161, "learning_rate": 4.892023821307664e-05, "loss": 2.8464, "step": 26879 }, { "epoch": 1.6686324414923335, "grad_norm": 0.5867643244652265, "learning_rate": 4.891662753393954e-05, "loss": 2.6909, "step": 26880 }, { "epoch": 1.6686945185920914, "grad_norm": 0.6358929292773138, "learning_rate": 4.891301686045467e-05, "loss": 2.7045, "step": 26881 }, { "epoch": 1.6687565956918493, "grad_norm": 0.6206156491961023, "learning_rate": 4.890940619264085e-05, "loss": 2.7878, "step": 26882 }, { "epoch": 1.6688186727916072, "grad_norm": 0.5870580766288453, "learning_rate": 4.890579553051691e-05, "loss": 2.7059, "step": 26883 }, { "epoch": 1.6688807498913651, "grad_norm": 0.6038071505427369, "learning_rate": 4.890218487410172e-05, "loss": 2.787, "step": 26884 }, { "epoch": 1.668942826991123, "grad_norm": 0.5725027901360765, "learning_rate": 4.889857422341408e-05, "loss": 2.7978, "step": 26885 }, { "epoch": 1.669004904090881, "grad_norm": 0.661280515503889, "learning_rate": 4.889496357847286e-05, "loss": 2.7593, "step": 26886 }, { "epoch": 1.669066981190639, "grad_norm": 0.5551486717456291, "learning_rate": 4.8891352939296885e-05, "loss": 2.7929, "step": 26887 }, { "epoch": 1.6691290582903968, "grad_norm": 0.5871567819304526, "learning_rate": 4.8887742305904984e-05, "loss": 2.7356, "step": 26888 }, { "epoch": 1.6691911353901547, "grad_norm": 0.553917175314352, "learning_rate": 4.8884131678316005e-05, "loss": 2.7677, "step": 26889 }, { "epoch": 1.6692532124899124, "grad_norm": 0.6462625313065921, "learning_rate": 4.888052105654877e-05, "loss": 2.7291, "step": 26890 }, { "epoch": 1.6693152895896703, "grad_norm": 0.5793612124830246, "learning_rate": 4.887691044062214e-05, "loss": 2.7205, "step": 26891 }, { "epoch": 1.6693773666894283, "grad_norm": 0.5934569237831198, "learning_rate": 4.8873299830554927e-05, "loss": 2.7418, "step": 26892 }, { "epoch": 1.6694394437891862, "grad_norm": 0.5831676117188997, "learning_rate": 4.8869689226365985e-05, "loss": 2.7845, "step": 26893 }, { "epoch": 1.669501520888944, "grad_norm": 0.5954242241016987, "learning_rate": 4.886607862807416e-05, "loss": 2.786, "step": 26894 }, { "epoch": 1.669563597988702, "grad_norm": 0.5769580714298478, "learning_rate": 4.886246803569827e-05, "loss": 2.7828, "step": 26895 }, { "epoch": 1.6696256750884597, "grad_norm": 0.5528150304172496, "learning_rate": 4.885885744925717e-05, "loss": 2.8241, "step": 26896 }, { "epoch": 1.6696877521882176, "grad_norm": 0.5324952733362882, "learning_rate": 4.8855246868769675e-05, "loss": 2.749, "step": 26897 }, { "epoch": 1.6697498292879756, "grad_norm": 0.5462574878415112, "learning_rate": 4.8851636294254655e-05, "loss": 2.7165, "step": 26898 }, { "epoch": 1.6698119063877335, "grad_norm": 0.6526910006107712, "learning_rate": 4.8848025725730916e-05, "loss": 2.8066, "step": 26899 }, { "epoch": 1.6698739834874914, "grad_norm": 0.5129243756801336, "learning_rate": 4.8844415163217296e-05, "loss": 2.8113, "step": 26900 }, { "epoch": 1.6699360605872493, "grad_norm": 0.6386292469602479, "learning_rate": 4.884080460673267e-05, "loss": 2.6784, "step": 26901 }, { "epoch": 1.6699981376870072, "grad_norm": 0.5400050947602436, "learning_rate": 4.883719405629581e-05, "loss": 2.6149, "step": 26902 }, { "epoch": 1.6700602147867651, "grad_norm": 0.5753356814130265, "learning_rate": 4.8833583511925625e-05, "loss": 2.7976, "step": 26903 }, { "epoch": 1.670122291886523, "grad_norm": 0.5511964626221587, "learning_rate": 4.8829972973640895e-05, "loss": 2.6465, "step": 26904 }, { "epoch": 1.670184368986281, "grad_norm": 0.5774398558537923, "learning_rate": 4.8826362441460494e-05, "loss": 2.6539, "step": 26905 }, { "epoch": 1.670246446086039, "grad_norm": 0.5843102811990953, "learning_rate": 4.882275191540323e-05, "loss": 2.8303, "step": 26906 }, { "epoch": 1.6703085231857968, "grad_norm": 0.5124192720891765, "learning_rate": 4.881914139548798e-05, "loss": 2.7501, "step": 26907 }, { "epoch": 1.6703706002855547, "grad_norm": 0.5674340229921592, "learning_rate": 4.881553088173353e-05, "loss": 2.686, "step": 26908 }, { "epoch": 1.6704326773853126, "grad_norm": 0.5394485031471342, "learning_rate": 4.881192037415876e-05, "loss": 2.7768, "step": 26909 }, { "epoch": 1.6704947544850706, "grad_norm": 0.5345514318541744, "learning_rate": 4.8808309872782474e-05, "loss": 2.8506, "step": 26910 }, { "epoch": 1.6705568315848285, "grad_norm": 0.5379346022015385, "learning_rate": 4.880469937762354e-05, "loss": 2.7338, "step": 26911 }, { "epoch": 1.6706189086845864, "grad_norm": 0.5863258854812456, "learning_rate": 4.880108888870078e-05, "loss": 2.746, "step": 26912 }, { "epoch": 1.6706809857843443, "grad_norm": 0.5164008805404118, "learning_rate": 4.879747840603302e-05, "loss": 2.8551, "step": 26913 }, { "epoch": 1.670743062884102, "grad_norm": 0.6630559585828248, "learning_rate": 4.8793867929639114e-05, "loss": 2.76, "step": 26914 }, { "epoch": 1.67080513998386, "grad_norm": 0.5394845498977702, "learning_rate": 4.8790257459537886e-05, "loss": 2.7745, "step": 26915 }, { "epoch": 1.6708672170836179, "grad_norm": 0.5933961817944802, "learning_rate": 4.8786646995748186e-05, "loss": 2.7299, "step": 26916 }, { "epoch": 1.6709292941833758, "grad_norm": 0.5622004400081941, "learning_rate": 4.878303653828884e-05, "loss": 2.7335, "step": 26917 }, { "epoch": 1.6709913712831337, "grad_norm": 0.5398579133269477, "learning_rate": 4.8779426087178684e-05, "loss": 2.7319, "step": 26918 }, { "epoch": 1.6710534483828916, "grad_norm": 0.6283532311588423, "learning_rate": 4.877581564243656e-05, "loss": 2.789, "step": 26919 }, { "epoch": 1.6711155254826493, "grad_norm": 0.49212366023134424, "learning_rate": 4.877220520408131e-05, "loss": 2.8158, "step": 26920 }, { "epoch": 1.6711776025824072, "grad_norm": 0.6510741619901402, "learning_rate": 4.876859477213175e-05, "loss": 2.8091, "step": 26921 }, { "epoch": 1.6712396796821651, "grad_norm": 0.5433205487155257, "learning_rate": 4.876498434660674e-05, "loss": 2.802, "step": 26922 }, { "epoch": 1.671301756781923, "grad_norm": 0.6302863905515145, "learning_rate": 4.876137392752511e-05, "loss": 2.679, "step": 26923 }, { "epoch": 1.671363833881681, "grad_norm": 0.5926715694170351, "learning_rate": 4.875776351490567e-05, "loss": 2.6988, "step": 26924 }, { "epoch": 1.671425910981439, "grad_norm": 0.6373363845897135, "learning_rate": 4.875415310876729e-05, "loss": 2.7663, "step": 26925 }, { "epoch": 1.6714879880811968, "grad_norm": 0.5958688974326075, "learning_rate": 4.87505427091288e-05, "loss": 2.7155, "step": 26926 }, { "epoch": 1.6715500651809547, "grad_norm": 0.6030056474620484, "learning_rate": 4.874693231600904e-05, "loss": 2.8522, "step": 26927 }, { "epoch": 1.6716121422807126, "grad_norm": 0.6002741554878048, "learning_rate": 4.874332192942683e-05, "loss": 2.7202, "step": 26928 }, { "epoch": 1.6716742193804706, "grad_norm": 0.6226338658922167, "learning_rate": 4.873971154940101e-05, "loss": 2.7384, "step": 26929 }, { "epoch": 1.6717362964802285, "grad_norm": 0.6146846434503069, "learning_rate": 4.873610117595044e-05, "loss": 2.7538, "step": 26930 }, { "epoch": 1.6717983735799864, "grad_norm": 0.5477205007523704, "learning_rate": 4.8732490809093915e-05, "loss": 2.7548, "step": 26931 }, { "epoch": 1.6718604506797443, "grad_norm": 0.550403763212879, "learning_rate": 4.8728880448850306e-05, "loss": 2.7423, "step": 26932 }, { "epoch": 1.6719225277795022, "grad_norm": 0.5476008638898306, "learning_rate": 4.872527009523843e-05, "loss": 2.7727, "step": 26933 }, { "epoch": 1.6719846048792602, "grad_norm": 0.5488023853714487, "learning_rate": 4.872165974827714e-05, "loss": 2.6975, "step": 26934 }, { "epoch": 1.672046681979018, "grad_norm": 0.549225686953324, "learning_rate": 4.871804940798525e-05, "loss": 2.7347, "step": 26935 }, { "epoch": 1.672108759078776, "grad_norm": 0.5363199154197021, "learning_rate": 4.871443907438161e-05, "loss": 2.7282, "step": 26936 }, { "epoch": 1.672170836178534, "grad_norm": 0.5165636593158911, "learning_rate": 4.871082874748506e-05, "loss": 2.7414, "step": 26937 }, { "epoch": 1.6722329132782916, "grad_norm": 0.5498881844086569, "learning_rate": 4.8707218427314416e-05, "loss": 2.7316, "step": 26938 }, { "epoch": 1.6722949903780495, "grad_norm": 0.5528742744601459, "learning_rate": 4.870360811388854e-05, "loss": 2.8946, "step": 26939 }, { "epoch": 1.6723570674778074, "grad_norm": 0.617976541385652, "learning_rate": 4.869999780722625e-05, "loss": 2.7924, "step": 26940 }, { "epoch": 1.6724191445775654, "grad_norm": 0.6023559539389834, "learning_rate": 4.869638750734639e-05, "loss": 2.7461, "step": 26941 }, { "epoch": 1.6724812216773233, "grad_norm": 0.5469611781202687, "learning_rate": 4.8692777214267785e-05, "loss": 2.6612, "step": 26942 }, { "epoch": 1.6725432987770812, "grad_norm": 0.6086471563565653, "learning_rate": 4.868916692800929e-05, "loss": 2.8079, "step": 26943 }, { "epoch": 1.6726053758768389, "grad_norm": 0.6084005193536938, "learning_rate": 4.8685556648589716e-05, "loss": 2.8025, "step": 26944 }, { "epoch": 1.6726674529765968, "grad_norm": 0.5091168305200282, "learning_rate": 4.868194637602793e-05, "loss": 2.6889, "step": 26945 }, { "epoch": 1.6727295300763547, "grad_norm": 0.6339899341388882, "learning_rate": 4.867833611034273e-05, "loss": 2.7343, "step": 26946 }, { "epoch": 1.6727916071761126, "grad_norm": 0.5657901175161455, "learning_rate": 4.8674725851552984e-05, "loss": 2.7637, "step": 26947 }, { "epoch": 1.6728536842758706, "grad_norm": 0.5856850143671338, "learning_rate": 4.867111559967752e-05, "loss": 2.772, "step": 26948 }, { "epoch": 1.6729157613756285, "grad_norm": 0.6418206756692182, "learning_rate": 4.866750535473515e-05, "loss": 2.7954, "step": 26949 }, { "epoch": 1.6729778384753864, "grad_norm": 0.5887899522123597, "learning_rate": 4.866389511674475e-05, "loss": 2.7289, "step": 26950 }, { "epoch": 1.6730399155751443, "grad_norm": 0.6260761056864746, "learning_rate": 4.866028488572511e-05, "loss": 2.717, "step": 26951 }, { "epoch": 1.6731019926749022, "grad_norm": 0.584449302315114, "learning_rate": 4.8656674661695105e-05, "loss": 2.8571, "step": 26952 }, { "epoch": 1.6731640697746601, "grad_norm": 0.6105093301955204, "learning_rate": 4.8653064444673546e-05, "loss": 2.7492, "step": 26953 }, { "epoch": 1.673226146874418, "grad_norm": 0.5745744706268654, "learning_rate": 4.8649454234679284e-05, "loss": 2.7535, "step": 26954 }, { "epoch": 1.673288223974176, "grad_norm": 0.6867116613098141, "learning_rate": 4.8645844031731136e-05, "loss": 2.7778, "step": 26955 }, { "epoch": 1.673350301073934, "grad_norm": 0.5316943666313595, "learning_rate": 4.864223383584796e-05, "loss": 2.7719, "step": 26956 }, { "epoch": 1.6734123781736918, "grad_norm": 0.6874536368602705, "learning_rate": 4.8638623647048565e-05, "loss": 2.7754, "step": 26957 }, { "epoch": 1.6734744552734497, "grad_norm": 0.5156037875349321, "learning_rate": 4.8635013465351804e-05, "loss": 2.7662, "step": 26958 }, { "epoch": 1.6735365323732077, "grad_norm": 0.6499298800647119, "learning_rate": 4.863140329077652e-05, "loss": 2.7126, "step": 26959 }, { "epoch": 1.6735986094729656, "grad_norm": 0.5457185506859089, "learning_rate": 4.862779312334153e-05, "loss": 2.7529, "step": 26960 }, { "epoch": 1.6736606865727235, "grad_norm": 0.5905715292487526, "learning_rate": 4.8624182963065685e-05, "loss": 2.7516, "step": 26961 }, { "epoch": 1.6737227636724812, "grad_norm": 0.5744695051066153, "learning_rate": 4.862057280996781e-05, "loss": 2.7077, "step": 26962 }, { "epoch": 1.673784840772239, "grad_norm": 0.6061251128250303, "learning_rate": 4.861696266406674e-05, "loss": 2.7782, "step": 26963 }, { "epoch": 1.673846917871997, "grad_norm": 0.5528481032840705, "learning_rate": 4.861335252538131e-05, "loss": 2.716, "step": 26964 }, { "epoch": 1.673908994971755, "grad_norm": 0.5829831985568502, "learning_rate": 4.860974239393035e-05, "loss": 2.7365, "step": 26965 }, { "epoch": 1.6739710720715129, "grad_norm": 0.6009077497032401, "learning_rate": 4.860613226973271e-05, "loss": 2.8312, "step": 26966 }, { "epoch": 1.6740331491712708, "grad_norm": 0.5718115609841241, "learning_rate": 4.860252215280721e-05, "loss": 2.7526, "step": 26967 }, { "epoch": 1.6740952262710285, "grad_norm": 0.5584834863634145, "learning_rate": 4.85989120431727e-05, "loss": 2.7305, "step": 26968 }, { "epoch": 1.6741573033707864, "grad_norm": 0.5133170000745509, "learning_rate": 4.8595301940848e-05, "loss": 2.7991, "step": 26969 }, { "epoch": 1.6742193804705443, "grad_norm": 0.5424828035731849, "learning_rate": 4.859169184585195e-05, "loss": 2.7567, "step": 26970 }, { "epoch": 1.6742814575703022, "grad_norm": 0.5456024350091707, "learning_rate": 4.8588081758203394e-05, "loss": 2.7699, "step": 26971 }, { "epoch": 1.6743435346700601, "grad_norm": 0.5184223128216078, "learning_rate": 4.858447167792115e-05, "loss": 2.7688, "step": 26972 }, { "epoch": 1.674405611769818, "grad_norm": 0.6057339979412596, "learning_rate": 4.858086160502407e-05, "loss": 2.7636, "step": 26973 }, { "epoch": 1.674467688869576, "grad_norm": 0.5159477033247367, "learning_rate": 4.8577251539530964e-05, "loss": 2.7283, "step": 26974 }, { "epoch": 1.674529765969334, "grad_norm": 0.6119587969264482, "learning_rate": 4.8573641481460694e-05, "loss": 2.7927, "step": 26975 }, { "epoch": 1.6745918430690918, "grad_norm": 0.5612647480715548, "learning_rate": 4.8570031430832076e-05, "loss": 2.673, "step": 26976 }, { "epoch": 1.6746539201688497, "grad_norm": 0.5784297088559979, "learning_rate": 4.856642138766396e-05, "loss": 2.7047, "step": 26977 }, { "epoch": 1.6747159972686076, "grad_norm": 0.5920798251616105, "learning_rate": 4.856281135197516e-05, "loss": 2.7998, "step": 26978 }, { "epoch": 1.6747780743683656, "grad_norm": 0.6369452980334498, "learning_rate": 4.8559201323784534e-05, "loss": 2.7412, "step": 26979 }, { "epoch": 1.6748401514681235, "grad_norm": 0.5811550021731753, "learning_rate": 4.855559130311089e-05, "loss": 2.7534, "step": 26980 }, { "epoch": 1.6749022285678814, "grad_norm": 0.6282896964665909, "learning_rate": 4.855198128997309e-05, "loss": 2.7529, "step": 26981 }, { "epoch": 1.6749643056676393, "grad_norm": 0.633725296833189, "learning_rate": 4.854837128438995e-05, "loss": 2.7336, "step": 26982 }, { "epoch": 1.6750263827673972, "grad_norm": 0.6045108477989313, "learning_rate": 4.854476128638031e-05, "loss": 2.8112, "step": 26983 }, { "epoch": 1.6750884598671552, "grad_norm": 0.6261994296916557, "learning_rate": 4.854115129596301e-05, "loss": 2.7268, "step": 26984 }, { "epoch": 1.6751505369669129, "grad_norm": 0.5995941160167806, "learning_rate": 4.853754131315686e-05, "loss": 2.8216, "step": 26985 }, { "epoch": 1.6752126140666708, "grad_norm": 0.6328853989043848, "learning_rate": 4.853393133798074e-05, "loss": 2.7727, "step": 26986 }, { "epoch": 1.6752746911664287, "grad_norm": 0.5706849123367935, "learning_rate": 4.853032137045343e-05, "loss": 2.8233, "step": 26987 }, { "epoch": 1.6753367682661866, "grad_norm": 0.5980402143171017, "learning_rate": 4.852671141059381e-05, "loss": 2.8339, "step": 26988 }, { "epoch": 1.6753988453659445, "grad_norm": 0.5623732265029825, "learning_rate": 4.8523101458420674e-05, "loss": 2.7717, "step": 26989 }, { "epoch": 1.6754609224657024, "grad_norm": 0.6504457009396792, "learning_rate": 4.8519491513952864e-05, "loss": 2.8371, "step": 26990 }, { "epoch": 1.6755229995654601, "grad_norm": 0.5888584825875048, "learning_rate": 4.8515881577209256e-05, "loss": 2.7892, "step": 26991 }, { "epoch": 1.675585076665218, "grad_norm": 0.622049879280473, "learning_rate": 4.851227164820864e-05, "loss": 2.7063, "step": 26992 }, { "epoch": 1.675647153764976, "grad_norm": 0.4970621734132657, "learning_rate": 4.850866172696988e-05, "loss": 2.7945, "step": 26993 }, { "epoch": 1.675709230864734, "grad_norm": 0.66367382567929, "learning_rate": 4.850505181351178e-05, "loss": 2.7873, "step": 26994 }, { "epoch": 1.6757713079644918, "grad_norm": 0.5152793650618398, "learning_rate": 4.850144190785319e-05, "loss": 2.7241, "step": 26995 }, { "epoch": 1.6758333850642497, "grad_norm": 0.6196912609376726, "learning_rate": 4.849783201001294e-05, "loss": 2.79, "step": 26996 }, { "epoch": 1.6758954621640076, "grad_norm": 0.6243768446458788, "learning_rate": 4.849422212000987e-05, "loss": 2.7764, "step": 26997 }, { "epoch": 1.6759575392637656, "grad_norm": 0.5896782295456572, "learning_rate": 4.8490612237862815e-05, "loss": 2.7665, "step": 26998 }, { "epoch": 1.6760196163635235, "grad_norm": 0.6098230377675511, "learning_rate": 4.848700236359059e-05, "loss": 2.815, "step": 26999 }, { "epoch": 1.6760816934632814, "grad_norm": 0.5291487708282108, "learning_rate": 4.848339249721205e-05, "loss": 2.6071, "step": 27000 }, { "epoch": 1.6761437705630393, "grad_norm": 0.653972807383058, "learning_rate": 4.847978263874601e-05, "loss": 2.7959, "step": 27001 }, { "epoch": 1.6762058476627972, "grad_norm": 0.577306735409845, "learning_rate": 4.847617278821132e-05, "loss": 2.7851, "step": 27002 }, { "epoch": 1.6762679247625552, "grad_norm": 0.6305670754906333, "learning_rate": 4.84725629456268e-05, "loss": 2.7766, "step": 27003 }, { "epoch": 1.676330001862313, "grad_norm": 0.6273231483962705, "learning_rate": 4.8468953111011295e-05, "loss": 2.7065, "step": 27004 }, { "epoch": 1.676392078962071, "grad_norm": 0.6171732067915753, "learning_rate": 4.846534328438363e-05, "loss": 2.7525, "step": 27005 }, { "epoch": 1.676454156061829, "grad_norm": 0.5382460340432612, "learning_rate": 4.846173346576265e-05, "loss": 2.7833, "step": 27006 }, { "epoch": 1.6765162331615868, "grad_norm": 0.6221168057540337, "learning_rate": 4.845812365516716e-05, "loss": 2.8066, "step": 27007 }, { "epoch": 1.6765783102613447, "grad_norm": 0.5962042779806115, "learning_rate": 4.8454513852616044e-05, "loss": 2.7611, "step": 27008 }, { "epoch": 1.6766403873611024, "grad_norm": 0.5676983007075331, "learning_rate": 4.845090405812809e-05, "loss": 2.8172, "step": 27009 }, { "epoch": 1.6767024644608604, "grad_norm": 0.5962128006629147, "learning_rate": 4.8447294271722135e-05, "loss": 2.7608, "step": 27010 }, { "epoch": 1.6767645415606183, "grad_norm": 0.5318512319326232, "learning_rate": 4.8443684493417035e-05, "loss": 2.8249, "step": 27011 }, { "epoch": 1.6768266186603762, "grad_norm": 0.60805833841257, "learning_rate": 4.84400747232316e-05, "loss": 2.8682, "step": 27012 }, { "epoch": 1.676888695760134, "grad_norm": 0.551393506796541, "learning_rate": 4.843646496118469e-05, "loss": 2.7683, "step": 27013 }, { "epoch": 1.676950772859892, "grad_norm": 0.5514235032307178, "learning_rate": 4.8432855207295104e-05, "loss": 2.8614, "step": 27014 }, { "epoch": 1.6770128499596497, "grad_norm": 0.5771301318265392, "learning_rate": 4.8429245461581704e-05, "loss": 2.7098, "step": 27015 }, { "epoch": 1.6770749270594076, "grad_norm": 0.5011519335186053, "learning_rate": 4.842563572406331e-05, "loss": 2.7303, "step": 27016 }, { "epoch": 1.6771370041591656, "grad_norm": 0.6124312955898448, "learning_rate": 4.842202599475876e-05, "loss": 2.886, "step": 27017 }, { "epoch": 1.6771990812589235, "grad_norm": 0.4859288886432634, "learning_rate": 4.841841627368687e-05, "loss": 2.7686, "step": 27018 }, { "epoch": 1.6772611583586814, "grad_norm": 0.5649239975102418, "learning_rate": 4.84148065608665e-05, "loss": 2.7631, "step": 27019 }, { "epoch": 1.6773232354584393, "grad_norm": 0.5393044302028603, "learning_rate": 4.841119685631646e-05, "loss": 2.7885, "step": 27020 }, { "epoch": 1.6773853125581972, "grad_norm": 0.5230472590336092, "learning_rate": 4.84075871600556e-05, "loss": 2.7994, "step": 27021 }, { "epoch": 1.6774473896579551, "grad_norm": 0.5489492316700946, "learning_rate": 4.840397747210275e-05, "loss": 2.7886, "step": 27022 }, { "epoch": 1.677509466757713, "grad_norm": 0.5548594033802899, "learning_rate": 4.84003677924767e-05, "loss": 2.8436, "step": 27023 }, { "epoch": 1.677571543857471, "grad_norm": 0.49389809850032057, "learning_rate": 4.839675812119634e-05, "loss": 2.7551, "step": 27024 }, { "epoch": 1.677633620957229, "grad_norm": 0.6050671876296853, "learning_rate": 4.83931484582805e-05, "loss": 2.7682, "step": 27025 }, { "epoch": 1.6776956980569868, "grad_norm": 0.4960596568490049, "learning_rate": 4.838953880374798e-05, "loss": 2.7872, "step": 27026 }, { "epoch": 1.6777577751567447, "grad_norm": 0.5787802925997897, "learning_rate": 4.838592915761763e-05, "loss": 2.7887, "step": 27027 }, { "epoch": 1.6778198522565027, "grad_norm": 0.5748738554785331, "learning_rate": 4.838231951990828e-05, "loss": 2.8273, "step": 27028 }, { "epoch": 1.6778819293562606, "grad_norm": 0.5519157521202224, "learning_rate": 4.837870989063876e-05, "loss": 2.7744, "step": 27029 }, { "epoch": 1.6779440064560185, "grad_norm": 0.5325607790310867, "learning_rate": 4.8375100269827906e-05, "loss": 2.8186, "step": 27030 }, { "epoch": 1.6780060835557764, "grad_norm": 0.5355749789264459, "learning_rate": 4.837149065749455e-05, "loss": 2.8127, "step": 27031 }, { "epoch": 1.6780681606555343, "grad_norm": 0.5427197770604146, "learning_rate": 4.8367881053657514e-05, "loss": 2.8244, "step": 27032 }, { "epoch": 1.678130237755292, "grad_norm": 0.5416366373198983, "learning_rate": 4.8364271458335655e-05, "loss": 2.7179, "step": 27033 }, { "epoch": 1.67819231485505, "grad_norm": 0.5484678232879486, "learning_rate": 4.836066187154778e-05, "loss": 2.7977, "step": 27034 }, { "epoch": 1.6782543919548079, "grad_norm": 0.5194208128149755, "learning_rate": 4.8357052293312724e-05, "loss": 2.7655, "step": 27035 }, { "epoch": 1.6783164690545658, "grad_norm": 0.5307420377758486, "learning_rate": 4.835344272364934e-05, "loss": 2.8436, "step": 27036 }, { "epoch": 1.6783785461543237, "grad_norm": 0.539981197409532, "learning_rate": 4.834983316257643e-05, "loss": 2.7636, "step": 27037 }, { "epoch": 1.6784406232540816, "grad_norm": 0.5300968220279972, "learning_rate": 4.834622361011285e-05, "loss": 2.8752, "step": 27038 }, { "epoch": 1.6785027003538393, "grad_norm": 0.6063711762992854, "learning_rate": 4.834261406627741e-05, "loss": 2.8374, "step": 27039 }, { "epoch": 1.6785647774535972, "grad_norm": 0.5618299360948662, "learning_rate": 4.833900453108897e-05, "loss": 2.7772, "step": 27040 }, { "epoch": 1.6786268545533551, "grad_norm": 0.582610309495012, "learning_rate": 4.833539500456634e-05, "loss": 2.7532, "step": 27041 }, { "epoch": 1.678688931653113, "grad_norm": 0.580926653436011, "learning_rate": 4.8331785486728356e-05, "loss": 2.7279, "step": 27042 }, { "epoch": 1.678751008752871, "grad_norm": 0.5494879560686633, "learning_rate": 4.8328175977593854e-05, "loss": 2.7301, "step": 27043 }, { "epoch": 1.678813085852629, "grad_norm": 0.6195538703088874, "learning_rate": 4.832456647718167e-05, "loss": 2.8034, "step": 27044 }, { "epoch": 1.6788751629523868, "grad_norm": 0.5859319050839651, "learning_rate": 4.8320956985510617e-05, "loss": 2.8017, "step": 27045 }, { "epoch": 1.6789372400521447, "grad_norm": 0.5870758189182282, "learning_rate": 4.831734750259955e-05, "loss": 2.8526, "step": 27046 }, { "epoch": 1.6789993171519026, "grad_norm": 0.6083453355763525, "learning_rate": 4.8313738028467295e-05, "loss": 2.7416, "step": 27047 }, { "epoch": 1.6790613942516606, "grad_norm": 0.5668025504839606, "learning_rate": 4.8310128563132664e-05, "loss": 2.8532, "step": 27048 }, { "epoch": 1.6791234713514185, "grad_norm": 0.6198506420724517, "learning_rate": 4.8306519106614513e-05, "loss": 2.7072, "step": 27049 }, { "epoch": 1.6791855484511764, "grad_norm": 0.5140771173874438, "learning_rate": 4.830290965893165e-05, "loss": 2.7407, "step": 27050 }, { "epoch": 1.6792476255509343, "grad_norm": 0.5910100458843679, "learning_rate": 4.829930022010293e-05, "loss": 2.744, "step": 27051 }, { "epoch": 1.6793097026506922, "grad_norm": 0.5165124538707521, "learning_rate": 4.829569079014716e-05, "loss": 2.8516, "step": 27052 }, { "epoch": 1.6793717797504502, "grad_norm": 0.5778197694179839, "learning_rate": 4.82920813690832e-05, "loss": 2.7576, "step": 27053 }, { "epoch": 1.679433856850208, "grad_norm": 0.500265468219099, "learning_rate": 4.828847195692986e-05, "loss": 2.7516, "step": 27054 }, { "epoch": 1.679495933949966, "grad_norm": 0.5872207322799, "learning_rate": 4.8284862553705965e-05, "loss": 2.8142, "step": 27055 }, { "epoch": 1.679558011049724, "grad_norm": 0.5358151187688118, "learning_rate": 4.828125315943037e-05, "loss": 2.8431, "step": 27056 }, { "epoch": 1.6796200881494816, "grad_norm": 0.621389690405853, "learning_rate": 4.8277643774121894e-05, "loss": 2.7814, "step": 27057 }, { "epoch": 1.6796821652492395, "grad_norm": 0.5445148219253418, "learning_rate": 4.827403439779938e-05, "loss": 2.6993, "step": 27058 }, { "epoch": 1.6797442423489974, "grad_norm": 0.5635710955942936, "learning_rate": 4.827042503048164e-05, "loss": 2.7869, "step": 27059 }, { "epoch": 1.6798063194487554, "grad_norm": 0.5767149870901407, "learning_rate": 4.82668156721875e-05, "loss": 2.6961, "step": 27060 }, { "epoch": 1.6798683965485133, "grad_norm": 0.5914596251334442, "learning_rate": 4.826320632293582e-05, "loss": 2.7555, "step": 27061 }, { "epoch": 1.6799304736482712, "grad_norm": 0.5184838772863235, "learning_rate": 4.825959698274541e-05, "loss": 2.7015, "step": 27062 }, { "epoch": 1.679992550748029, "grad_norm": 0.6064475239146727, "learning_rate": 4.82559876516351e-05, "loss": 2.7737, "step": 27063 }, { "epoch": 1.6800546278477868, "grad_norm": 0.5192213640081204, "learning_rate": 4.825237832962373e-05, "loss": 2.7517, "step": 27064 }, { "epoch": 1.6801167049475447, "grad_norm": 0.6074402981138384, "learning_rate": 4.8248769016730134e-05, "loss": 2.7446, "step": 27065 }, { "epoch": 1.6801787820473026, "grad_norm": 0.5257328152843302, "learning_rate": 4.824515971297312e-05, "loss": 2.8037, "step": 27066 }, { "epoch": 1.6802408591470606, "grad_norm": 0.620992784270844, "learning_rate": 4.824155041837155e-05, "loss": 2.8406, "step": 27067 }, { "epoch": 1.6803029362468185, "grad_norm": 0.5734620070872527, "learning_rate": 4.823794113294422e-05, "loss": 2.7795, "step": 27068 }, { "epoch": 1.6803650133465764, "grad_norm": 0.6341051392109799, "learning_rate": 4.823433185670999e-05, "loss": 2.7512, "step": 27069 }, { "epoch": 1.6804270904463343, "grad_norm": 0.6204307664413788, "learning_rate": 4.8230722589687674e-05, "loss": 2.7078, "step": 27070 }, { "epoch": 1.6804891675460922, "grad_norm": 0.5741299930407717, "learning_rate": 4.822711333189611e-05, "loss": 2.7621, "step": 27071 }, { "epoch": 1.6805512446458502, "grad_norm": 0.5894674654581962, "learning_rate": 4.8223504083354135e-05, "loss": 2.7899, "step": 27072 }, { "epoch": 1.680613321745608, "grad_norm": 0.599835753808628, "learning_rate": 4.821989484408056e-05, "loss": 2.6698, "step": 27073 }, { "epoch": 1.680675398845366, "grad_norm": 0.6464068778341341, "learning_rate": 4.821628561409423e-05, "loss": 2.7661, "step": 27074 }, { "epoch": 1.680737475945124, "grad_norm": 0.5779643737759295, "learning_rate": 4.821267639341397e-05, "loss": 2.7009, "step": 27075 }, { "epoch": 1.6807995530448818, "grad_norm": 0.6675385768687042, "learning_rate": 4.820906718205861e-05, "loss": 2.7543, "step": 27076 }, { "epoch": 1.6808616301446397, "grad_norm": 0.6147369796162144, "learning_rate": 4.820545798004698e-05, "loss": 2.848, "step": 27077 }, { "epoch": 1.6809237072443977, "grad_norm": 0.621744091306495, "learning_rate": 4.820184878739792e-05, "loss": 2.7759, "step": 27078 }, { "epoch": 1.6809857843441556, "grad_norm": 0.5867710796238076, "learning_rate": 4.819823960413023e-05, "loss": 2.7245, "step": 27079 }, { "epoch": 1.6810478614439135, "grad_norm": 0.6117112053299928, "learning_rate": 4.819463043026279e-05, "loss": 2.7657, "step": 27080 }, { "epoch": 1.6811099385436712, "grad_norm": 0.5799035474308561, "learning_rate": 4.819102126581438e-05, "loss": 2.7915, "step": 27081 }, { "epoch": 1.681172015643429, "grad_norm": 0.5945438799725888, "learning_rate": 4.818741211080387e-05, "loss": 2.6614, "step": 27082 }, { "epoch": 1.681234092743187, "grad_norm": 0.612586727619519, "learning_rate": 4.818380296525006e-05, "loss": 2.7116, "step": 27083 }, { "epoch": 1.681296169842945, "grad_norm": 0.5438185977416362, "learning_rate": 4.8180193829171785e-05, "loss": 2.7919, "step": 27084 }, { "epoch": 1.6813582469427029, "grad_norm": 0.6198524931436951, "learning_rate": 4.817658470258789e-05, "loss": 2.7938, "step": 27085 }, { "epoch": 1.6814203240424608, "grad_norm": 0.5061359312328393, "learning_rate": 4.817297558551718e-05, "loss": 2.8042, "step": 27086 }, { "epoch": 1.6814824011422185, "grad_norm": 0.6089349282519811, "learning_rate": 4.816936647797852e-05, "loss": 2.7803, "step": 27087 }, { "epoch": 1.6815444782419764, "grad_norm": 0.5376754274858699, "learning_rate": 4.81657573799907e-05, "loss": 2.792, "step": 27088 }, { "epoch": 1.6816065553417343, "grad_norm": 0.5083898343542819, "learning_rate": 4.816214829157257e-05, "loss": 2.7875, "step": 27089 }, { "epoch": 1.6816686324414922, "grad_norm": 0.5535958522032107, "learning_rate": 4.8158539212742973e-05, "loss": 2.7785, "step": 27090 }, { "epoch": 1.6817307095412501, "grad_norm": 0.4980892238189915, "learning_rate": 4.815493014352071e-05, "loss": 2.7653, "step": 27091 }, { "epoch": 1.681792786641008, "grad_norm": 0.5051833536586011, "learning_rate": 4.815132108392464e-05, "loss": 2.6914, "step": 27092 }, { "epoch": 1.681854863740766, "grad_norm": 0.5379729385944167, "learning_rate": 4.814771203397356e-05, "loss": 2.7847, "step": 27093 }, { "epoch": 1.681916940840524, "grad_norm": 0.5011392370167966, "learning_rate": 4.814410299368633e-05, "loss": 2.7954, "step": 27094 }, { "epoch": 1.6819790179402818, "grad_norm": 0.47177732732571837, "learning_rate": 4.8140493963081765e-05, "loss": 2.7599, "step": 27095 }, { "epoch": 1.6820410950400397, "grad_norm": 0.5287615613372321, "learning_rate": 4.813688494217869e-05, "loss": 2.7671, "step": 27096 }, { "epoch": 1.6821031721397977, "grad_norm": 0.5102303677472945, "learning_rate": 4.813327593099595e-05, "loss": 2.7693, "step": 27097 }, { "epoch": 1.6821652492395556, "grad_norm": 0.504142232833514, "learning_rate": 4.812966692955234e-05, "loss": 2.8521, "step": 27098 }, { "epoch": 1.6822273263393135, "grad_norm": 0.5654910882741752, "learning_rate": 4.812605793786673e-05, "loss": 2.7498, "step": 27099 }, { "epoch": 1.6822894034390714, "grad_norm": 0.5327251094175148, "learning_rate": 4.812244895595792e-05, "loss": 2.7169, "step": 27100 }, { "epoch": 1.6823514805388293, "grad_norm": 0.5760984988284571, "learning_rate": 4.811883998384476e-05, "loss": 2.8163, "step": 27101 }, { "epoch": 1.6824135576385872, "grad_norm": 0.6106994857962695, "learning_rate": 4.811523102154605e-05, "loss": 2.6902, "step": 27102 }, { "epoch": 1.6824756347383452, "grad_norm": 0.5220890040392854, "learning_rate": 4.811162206908066e-05, "loss": 2.7615, "step": 27103 }, { "epoch": 1.682537711838103, "grad_norm": 0.6321675110097339, "learning_rate": 4.8108013126467375e-05, "loss": 2.6711, "step": 27104 }, { "epoch": 1.6825997889378608, "grad_norm": 0.5670821005251693, "learning_rate": 4.8104404193725065e-05, "loss": 2.8168, "step": 27105 }, { "epoch": 1.6826618660376187, "grad_norm": 0.6081624070725155, "learning_rate": 4.810079527087252e-05, "loss": 2.8823, "step": 27106 }, { "epoch": 1.6827239431373766, "grad_norm": 0.5796353226697397, "learning_rate": 4.8097186357928604e-05, "loss": 2.805, "step": 27107 }, { "epoch": 1.6827860202371345, "grad_norm": 0.5692600572641664, "learning_rate": 4.809357745491212e-05, "loss": 2.7886, "step": 27108 }, { "epoch": 1.6828480973368924, "grad_norm": 0.5657342541804848, "learning_rate": 4.808996856184191e-05, "loss": 2.763, "step": 27109 }, { "epoch": 1.6829101744366504, "grad_norm": 0.5874543202680235, "learning_rate": 4.808635967873679e-05, "loss": 2.7873, "step": 27110 }, { "epoch": 1.682972251536408, "grad_norm": 0.5259231931267039, "learning_rate": 4.80827508056156e-05, "loss": 2.8012, "step": 27111 }, { "epoch": 1.683034328636166, "grad_norm": 0.5888506885450363, "learning_rate": 4.807914194249717e-05, "loss": 2.8415, "step": 27112 }, { "epoch": 1.683096405735924, "grad_norm": 0.5400187262625984, "learning_rate": 4.807553308940031e-05, "loss": 2.7343, "step": 27113 }, { "epoch": 1.6831584828356818, "grad_norm": 0.5901276366327035, "learning_rate": 4.8071924246343875e-05, "loss": 2.7151, "step": 27114 }, { "epoch": 1.6832205599354397, "grad_norm": 0.5494365645491684, "learning_rate": 4.806831541334667e-05, "loss": 2.7141, "step": 27115 }, { "epoch": 1.6832826370351976, "grad_norm": 0.5297781807576495, "learning_rate": 4.806470659042753e-05, "loss": 2.7617, "step": 27116 }, { "epoch": 1.6833447141349556, "grad_norm": 0.5505438489141944, "learning_rate": 4.8061097777605294e-05, "loss": 2.8357, "step": 27117 }, { "epoch": 1.6834067912347135, "grad_norm": 0.4935711983567214, "learning_rate": 4.805748897489878e-05, "loss": 2.7979, "step": 27118 }, { "epoch": 1.6834688683344714, "grad_norm": 0.5466093508061193, "learning_rate": 4.805388018232681e-05, "loss": 2.6591, "step": 27119 }, { "epoch": 1.6835309454342293, "grad_norm": 0.531865884902254, "learning_rate": 4.8050271399908226e-05, "loss": 2.734, "step": 27120 }, { "epoch": 1.6835930225339872, "grad_norm": 0.5821214900370705, "learning_rate": 4.8046662627661834e-05, "loss": 2.7572, "step": 27121 }, { "epoch": 1.6836550996337452, "grad_norm": 0.52829427884138, "learning_rate": 4.8043053865606495e-05, "loss": 2.7107, "step": 27122 }, { "epoch": 1.683717176733503, "grad_norm": 0.5897967844069207, "learning_rate": 4.803944511376101e-05, "loss": 2.7829, "step": 27123 }, { "epoch": 1.683779253833261, "grad_norm": 0.520683044366909, "learning_rate": 4.803583637214423e-05, "loss": 2.7545, "step": 27124 }, { "epoch": 1.683841330933019, "grad_norm": 0.5710198966516448, "learning_rate": 4.8032227640774954e-05, "loss": 2.7412, "step": 27125 }, { "epoch": 1.6839034080327768, "grad_norm": 0.5336888375181206, "learning_rate": 4.802861891967204e-05, "loss": 2.8352, "step": 27126 }, { "epoch": 1.6839654851325347, "grad_norm": 0.5657294340302195, "learning_rate": 4.802501020885428e-05, "loss": 2.7998, "step": 27127 }, { "epoch": 1.6840275622322927, "grad_norm": 0.5466729472059301, "learning_rate": 4.802140150834054e-05, "loss": 2.6951, "step": 27128 }, { "epoch": 1.6840896393320504, "grad_norm": 0.5624924235363505, "learning_rate": 4.801779281814961e-05, "loss": 2.7719, "step": 27129 }, { "epoch": 1.6841517164318083, "grad_norm": 0.5800304457467419, "learning_rate": 4.801418413830035e-05, "loss": 2.7278, "step": 27130 }, { "epoch": 1.6842137935315662, "grad_norm": 0.5202949761541307, "learning_rate": 4.801057546881156e-05, "loss": 2.6782, "step": 27131 }, { "epoch": 1.6842758706313241, "grad_norm": 0.5616375085816128, "learning_rate": 4.8006966809702094e-05, "loss": 2.7513, "step": 27132 }, { "epoch": 1.684337947731082, "grad_norm": 0.5714582143911622, "learning_rate": 4.800335816099077e-05, "loss": 2.7012, "step": 27133 }, { "epoch": 1.68440002483084, "grad_norm": 0.5553965536378772, "learning_rate": 4.7999749522696396e-05, "loss": 2.8226, "step": 27134 }, { "epoch": 1.6844621019305976, "grad_norm": 0.5772001756433867, "learning_rate": 4.799614089483783e-05, "loss": 2.8388, "step": 27135 }, { "epoch": 1.6845241790303556, "grad_norm": 0.6344017276864862, "learning_rate": 4.7992532277433864e-05, "loss": 2.8348, "step": 27136 }, { "epoch": 1.6845862561301135, "grad_norm": 0.5467311124883903, "learning_rate": 4.798892367050336e-05, "loss": 2.8146, "step": 27137 }, { "epoch": 1.6846483332298714, "grad_norm": 0.5579672547990415, "learning_rate": 4.798531507406512e-05, "loss": 2.7509, "step": 27138 }, { "epoch": 1.6847104103296293, "grad_norm": 0.5178313301728278, "learning_rate": 4.798170648813799e-05, "loss": 2.7579, "step": 27139 }, { "epoch": 1.6847724874293872, "grad_norm": 0.5207557016372792, "learning_rate": 4.7978097912740775e-05, "loss": 2.7329, "step": 27140 }, { "epoch": 1.6848345645291452, "grad_norm": 0.5597302273234921, "learning_rate": 4.7974489347892325e-05, "loss": 2.778, "step": 27141 }, { "epoch": 1.684896641628903, "grad_norm": 0.5441398067427136, "learning_rate": 4.797088079361145e-05, "loss": 2.7567, "step": 27142 }, { "epoch": 1.684958718728661, "grad_norm": 0.5791245660185346, "learning_rate": 4.796727224991699e-05, "loss": 2.7959, "step": 27143 }, { "epoch": 1.685020795828419, "grad_norm": 0.53258842275881, "learning_rate": 4.796366371682776e-05, "loss": 2.7445, "step": 27144 }, { "epoch": 1.6850828729281768, "grad_norm": 0.5521321874871409, "learning_rate": 4.7960055194362584e-05, "loss": 2.7875, "step": 27145 }, { "epoch": 1.6851449500279347, "grad_norm": 0.5066952666358757, "learning_rate": 4.795644668254031e-05, "loss": 2.7688, "step": 27146 }, { "epoch": 1.6852070271276927, "grad_norm": 0.5327068436648661, "learning_rate": 4.7952838181379736e-05, "loss": 2.6889, "step": 27147 }, { "epoch": 1.6852691042274506, "grad_norm": 0.5150575115905205, "learning_rate": 4.7949229690899714e-05, "loss": 2.7752, "step": 27148 }, { "epoch": 1.6853311813272085, "grad_norm": 0.5258948873572769, "learning_rate": 4.794562121111904e-05, "loss": 2.8086, "step": 27149 }, { "epoch": 1.6853932584269664, "grad_norm": 0.5336770546982466, "learning_rate": 4.794201274205657e-05, "loss": 2.7417, "step": 27150 }, { "epoch": 1.6854553355267243, "grad_norm": 0.5598471657353079, "learning_rate": 4.793840428373112e-05, "loss": 2.7677, "step": 27151 }, { "epoch": 1.6855174126264822, "grad_norm": 0.5186458873831752, "learning_rate": 4.793479583616152e-05, "loss": 2.71, "step": 27152 }, { "epoch": 1.68557948972624, "grad_norm": 0.5602051494115895, "learning_rate": 4.793118739936658e-05, "loss": 2.6609, "step": 27153 }, { "epoch": 1.6856415668259979, "grad_norm": 0.5338609710507835, "learning_rate": 4.7927578973365124e-05, "loss": 2.7396, "step": 27154 }, { "epoch": 1.6857036439257558, "grad_norm": 0.6276942593397554, "learning_rate": 4.792397055817602e-05, "loss": 2.7459, "step": 27155 }, { "epoch": 1.6857657210255137, "grad_norm": 0.541454976849233, "learning_rate": 4.7920362153818045e-05, "loss": 2.7029, "step": 27156 }, { "epoch": 1.6858277981252716, "grad_norm": 0.6049478566737206, "learning_rate": 4.791675376031006e-05, "loss": 2.7099, "step": 27157 }, { "epoch": 1.6858898752250295, "grad_norm": 0.4854725278247577, "learning_rate": 4.791314537767088e-05, "loss": 2.7663, "step": 27158 }, { "epoch": 1.6859519523247872, "grad_norm": 0.5788060378263913, "learning_rate": 4.790953700591931e-05, "loss": 2.7925, "step": 27159 }, { "epoch": 1.6860140294245451, "grad_norm": 0.49159821549717136, "learning_rate": 4.790592864507422e-05, "loss": 2.7895, "step": 27160 }, { "epoch": 1.686076106524303, "grad_norm": 0.6066587014339134, "learning_rate": 4.790232029515438e-05, "loss": 2.8593, "step": 27161 }, { "epoch": 1.686138183624061, "grad_norm": 0.5393933096614687, "learning_rate": 4.789871195617866e-05, "loss": 2.7358, "step": 27162 }, { "epoch": 1.686200260723819, "grad_norm": 0.6014145012653968, "learning_rate": 4.789510362816586e-05, "loss": 2.7811, "step": 27163 }, { "epoch": 1.6862623378235768, "grad_norm": 0.6311230654612233, "learning_rate": 4.789149531113483e-05, "loss": 2.6695, "step": 27164 }, { "epoch": 1.6863244149233347, "grad_norm": 0.614337861344362, "learning_rate": 4.788788700510437e-05, "loss": 2.7465, "step": 27165 }, { "epoch": 1.6863864920230927, "grad_norm": 0.5705659903926102, "learning_rate": 4.7884278710093325e-05, "loss": 2.6959, "step": 27166 }, { "epoch": 1.6864485691228506, "grad_norm": 0.6029815038640501, "learning_rate": 4.788067042612049e-05, "loss": 2.7242, "step": 27167 }, { "epoch": 1.6865106462226085, "grad_norm": 0.5871624920662613, "learning_rate": 4.787706215320474e-05, "loss": 2.7251, "step": 27168 }, { "epoch": 1.6865727233223664, "grad_norm": 0.5642323573713068, "learning_rate": 4.787345389136486e-05, "loss": 2.8064, "step": 27169 }, { "epoch": 1.6866348004221243, "grad_norm": 0.6534593146325567, "learning_rate": 4.786984564061968e-05, "loss": 2.7855, "step": 27170 }, { "epoch": 1.6866968775218822, "grad_norm": 0.6196151984391686, "learning_rate": 4.786623740098805e-05, "loss": 2.7554, "step": 27171 }, { "epoch": 1.6867589546216402, "grad_norm": 0.5551685074436237, "learning_rate": 4.786262917248876e-05, "loss": 2.7656, "step": 27172 }, { "epoch": 1.686821031721398, "grad_norm": 0.6133988383115173, "learning_rate": 4.785902095514066e-05, "loss": 2.7483, "step": 27173 }, { "epoch": 1.686883108821156, "grad_norm": 0.5359314642919429, "learning_rate": 4.785541274896256e-05, "loss": 2.6848, "step": 27174 }, { "epoch": 1.686945185920914, "grad_norm": 0.5997208979273331, "learning_rate": 4.78518045539733e-05, "loss": 2.748, "step": 27175 }, { "epoch": 1.6870072630206718, "grad_norm": 0.5812498390033458, "learning_rate": 4.7848196370191693e-05, "loss": 2.6492, "step": 27176 }, { "epoch": 1.6870693401204295, "grad_norm": 0.5589590145518781, "learning_rate": 4.784458819763658e-05, "loss": 2.8296, "step": 27177 }, { "epoch": 1.6871314172201874, "grad_norm": 0.566165320357297, "learning_rate": 4.784098003632675e-05, "loss": 2.8236, "step": 27178 }, { "epoch": 1.6871934943199454, "grad_norm": 0.6044541009542794, "learning_rate": 4.783737188628108e-05, "loss": 2.7562, "step": 27179 }, { "epoch": 1.6872555714197033, "grad_norm": 0.5486599810545608, "learning_rate": 4.783376374751835e-05, "loss": 2.8368, "step": 27180 }, { "epoch": 1.6873176485194612, "grad_norm": 0.5771262589016033, "learning_rate": 4.7830155620057396e-05, "loss": 2.8258, "step": 27181 }, { "epoch": 1.6873797256192191, "grad_norm": 0.5126258730412351, "learning_rate": 4.782654750391706e-05, "loss": 2.7501, "step": 27182 }, { "epoch": 1.6874418027189768, "grad_norm": 0.5691199179124384, "learning_rate": 4.782293939911614e-05, "loss": 2.7669, "step": 27183 }, { "epoch": 1.6875038798187347, "grad_norm": 0.5200837144956744, "learning_rate": 4.781933130567348e-05, "loss": 2.7082, "step": 27184 }, { "epoch": 1.6875659569184926, "grad_norm": 0.55517578125, "learning_rate": 4.7815723223607896e-05, "loss": 2.7592, "step": 27185 }, { "epoch": 1.6876280340182506, "grad_norm": 0.5478129925809954, "learning_rate": 4.78121151529382e-05, "loss": 2.7933, "step": 27186 }, { "epoch": 1.6876901111180085, "grad_norm": 0.5302342071915138, "learning_rate": 4.780850709368325e-05, "loss": 2.7042, "step": 27187 }, { "epoch": 1.6877521882177664, "grad_norm": 0.5878931343303875, "learning_rate": 4.780489904586184e-05, "loss": 2.8614, "step": 27188 }, { "epoch": 1.6878142653175243, "grad_norm": 0.582805144644542, "learning_rate": 4.7801291009492815e-05, "loss": 2.8129, "step": 27189 }, { "epoch": 1.6878763424172822, "grad_norm": 0.6602614505169233, "learning_rate": 4.7797682984594976e-05, "loss": 2.7885, "step": 27190 }, { "epoch": 1.6879384195170402, "grad_norm": 0.533604202056436, "learning_rate": 4.7794074971187173e-05, "loss": 2.7691, "step": 27191 }, { "epoch": 1.688000496616798, "grad_norm": 0.6330246275469567, "learning_rate": 4.7790466969288204e-05, "loss": 2.7059, "step": 27192 }, { "epoch": 1.688062573716556, "grad_norm": 0.5551822765980248, "learning_rate": 4.778685897891692e-05, "loss": 2.6608, "step": 27193 }, { "epoch": 1.688124650816314, "grad_norm": 0.6018458107973879, "learning_rate": 4.778325100009213e-05, "loss": 2.7796, "step": 27194 }, { "epoch": 1.6881867279160718, "grad_norm": 0.5708805286314333, "learning_rate": 4.777964303283265e-05, "loss": 2.7476, "step": 27195 }, { "epoch": 1.6882488050158297, "grad_norm": 0.586816789179712, "learning_rate": 4.777603507715731e-05, "loss": 2.7774, "step": 27196 }, { "epoch": 1.6883108821155877, "grad_norm": 0.5747371331874082, "learning_rate": 4.7772427133084926e-05, "loss": 2.737, "step": 27197 }, { "epoch": 1.6883729592153456, "grad_norm": 0.5539031594947501, "learning_rate": 4.776881920063435e-05, "loss": 2.8318, "step": 27198 }, { "epoch": 1.6884350363151035, "grad_norm": 0.6477612795640887, "learning_rate": 4.776521127982437e-05, "loss": 2.8098, "step": 27199 }, { "epoch": 1.6884971134148614, "grad_norm": 0.5786899950381763, "learning_rate": 4.7761603370673844e-05, "loss": 2.7, "step": 27200 }, { "epoch": 1.6885591905146191, "grad_norm": 0.62485593088978, "learning_rate": 4.7757995473201555e-05, "loss": 2.6938, "step": 27201 }, { "epoch": 1.688621267614377, "grad_norm": 0.5731595189339459, "learning_rate": 4.7754387587426364e-05, "loss": 2.7933, "step": 27202 }, { "epoch": 1.688683344714135, "grad_norm": 0.6076521850826639, "learning_rate": 4.775077971336707e-05, "loss": 2.8396, "step": 27203 }, { "epoch": 1.6887454218138929, "grad_norm": 0.5365034812768081, "learning_rate": 4.7747171851042506e-05, "loss": 2.7321, "step": 27204 }, { "epoch": 1.6888074989136508, "grad_norm": 0.6036437062271817, "learning_rate": 4.7743564000471495e-05, "loss": 2.7352, "step": 27205 }, { "epoch": 1.6888695760134087, "grad_norm": 0.6239609664666134, "learning_rate": 4.773995616167285e-05, "loss": 2.6689, "step": 27206 }, { "epoch": 1.6889316531131664, "grad_norm": 0.5753030206582099, "learning_rate": 4.773634833466542e-05, "loss": 2.7646, "step": 27207 }, { "epoch": 1.6889937302129243, "grad_norm": 0.6217509457481686, "learning_rate": 4.773274051946799e-05, "loss": 2.7019, "step": 27208 }, { "epoch": 1.6890558073126822, "grad_norm": 0.5718609677870555, "learning_rate": 4.772913271609942e-05, "loss": 2.7401, "step": 27209 }, { "epoch": 1.6891178844124402, "grad_norm": 0.6459332793844865, "learning_rate": 4.77255249245785e-05, "loss": 2.7769, "step": 27210 }, { "epoch": 1.689179961512198, "grad_norm": 0.614496594035469, "learning_rate": 4.7721917144924086e-05, "loss": 2.8167, "step": 27211 }, { "epoch": 1.689242038611956, "grad_norm": 0.6256317522063759, "learning_rate": 4.771830937715497e-05, "loss": 2.7474, "step": 27212 }, { "epoch": 1.689304115711714, "grad_norm": 0.6410843202742105, "learning_rate": 4.7714701621289995e-05, "loss": 2.8673, "step": 27213 }, { "epoch": 1.6893661928114718, "grad_norm": 0.6767116822653374, "learning_rate": 4.771109387734797e-05, "loss": 2.7209, "step": 27214 }, { "epoch": 1.6894282699112297, "grad_norm": 0.6355618222477135, "learning_rate": 4.770748614534773e-05, "loss": 2.8019, "step": 27215 }, { "epoch": 1.6894903470109877, "grad_norm": 0.6285902852629761, "learning_rate": 4.770387842530808e-05, "loss": 2.7652, "step": 27216 }, { "epoch": 1.6895524241107456, "grad_norm": 0.6756491118441698, "learning_rate": 4.770027071724787e-05, "loss": 2.7234, "step": 27217 }, { "epoch": 1.6896145012105035, "grad_norm": 0.6198705468084184, "learning_rate": 4.76966630211859e-05, "loss": 2.7185, "step": 27218 }, { "epoch": 1.6896765783102614, "grad_norm": 0.6525076015073479, "learning_rate": 4.769305533714098e-05, "loss": 2.8128, "step": 27219 }, { "epoch": 1.6897386554100193, "grad_norm": 0.6220570178085948, "learning_rate": 4.7689447665131964e-05, "loss": 2.7079, "step": 27220 }, { "epoch": 1.6898007325097772, "grad_norm": 0.5934547141794002, "learning_rate": 4.768584000517767e-05, "loss": 2.7799, "step": 27221 }, { "epoch": 1.6898628096095352, "grad_norm": 0.5613057651725261, "learning_rate": 4.76822323572969e-05, "loss": 2.7179, "step": 27222 }, { "epoch": 1.689924886709293, "grad_norm": 0.5989786863324327, "learning_rate": 4.767862472150849e-05, "loss": 2.7229, "step": 27223 }, { "epoch": 1.689986963809051, "grad_norm": 0.4995420862477653, "learning_rate": 4.7675017097831255e-05, "loss": 2.7501, "step": 27224 }, { "epoch": 1.6900490409088087, "grad_norm": 0.5598042583244328, "learning_rate": 4.7671409486284024e-05, "loss": 2.7772, "step": 27225 }, { "epoch": 1.6901111180085666, "grad_norm": 0.5312985229771778, "learning_rate": 4.766780188688561e-05, "loss": 2.6875, "step": 27226 }, { "epoch": 1.6901731951083245, "grad_norm": 0.5369398092158743, "learning_rate": 4.766419429965485e-05, "loss": 2.754, "step": 27227 }, { "epoch": 1.6902352722080825, "grad_norm": 0.5077332948390578, "learning_rate": 4.766058672461055e-05, "loss": 2.7603, "step": 27228 }, { "epoch": 1.6902973493078404, "grad_norm": 0.615190026032078, "learning_rate": 4.765697916177154e-05, "loss": 2.7976, "step": 27229 }, { "epoch": 1.6903594264075983, "grad_norm": 0.512287493587012, "learning_rate": 4.765337161115664e-05, "loss": 2.7932, "step": 27230 }, { "epoch": 1.690421503507356, "grad_norm": 0.5353270766216047, "learning_rate": 4.7649764072784654e-05, "loss": 2.7861, "step": 27231 }, { "epoch": 1.690483580607114, "grad_norm": 0.5144629533273491, "learning_rate": 4.764615654667444e-05, "loss": 2.6331, "step": 27232 }, { "epoch": 1.6905456577068718, "grad_norm": 0.545315453376533, "learning_rate": 4.764254903284478e-05, "loss": 2.7876, "step": 27233 }, { "epoch": 1.6906077348066297, "grad_norm": 0.5148155621466154, "learning_rate": 4.763894153131453e-05, "loss": 2.7321, "step": 27234 }, { "epoch": 1.6906698119063877, "grad_norm": 0.5481615736361256, "learning_rate": 4.763533404210248e-05, "loss": 2.7334, "step": 27235 }, { "epoch": 1.6907318890061456, "grad_norm": 0.5776198990115124, "learning_rate": 4.763172656522748e-05, "loss": 2.8078, "step": 27236 }, { "epoch": 1.6907939661059035, "grad_norm": 0.5558155838745589, "learning_rate": 4.762811910070833e-05, "loss": 2.6904, "step": 27237 }, { "epoch": 1.6908560432056614, "grad_norm": 0.6399971280927594, "learning_rate": 4.7624511648563866e-05, "loss": 2.7403, "step": 27238 }, { "epoch": 1.6909181203054193, "grad_norm": 0.6216955807901697, "learning_rate": 4.762090420881289e-05, "loss": 2.7435, "step": 27239 }, { "epoch": 1.6909801974051772, "grad_norm": 0.597359228190563, "learning_rate": 4.761729678147424e-05, "loss": 2.818, "step": 27240 }, { "epoch": 1.6910422745049352, "grad_norm": 0.6320934213207551, "learning_rate": 4.7613689366566726e-05, "loss": 2.7972, "step": 27241 }, { "epoch": 1.691104351604693, "grad_norm": 0.6125832248120797, "learning_rate": 4.7610081964109185e-05, "loss": 2.7066, "step": 27242 }, { "epoch": 1.691166428704451, "grad_norm": 0.6092448217776909, "learning_rate": 4.760647457412043e-05, "loss": 2.715, "step": 27243 }, { "epoch": 1.691228505804209, "grad_norm": 0.5425176599150784, "learning_rate": 4.760286719661926e-05, "loss": 2.6728, "step": 27244 }, { "epoch": 1.6912905829039668, "grad_norm": 0.6024846526917726, "learning_rate": 4.7599259831624535e-05, "loss": 2.8481, "step": 27245 }, { "epoch": 1.6913526600037248, "grad_norm": 0.5914604565308091, "learning_rate": 4.759565247915504e-05, "loss": 2.7288, "step": 27246 }, { "epoch": 1.6914147371034827, "grad_norm": 0.6183115224341786, "learning_rate": 4.759204513922961e-05, "loss": 2.7668, "step": 27247 }, { "epoch": 1.6914768142032406, "grad_norm": 0.6178618019048044, "learning_rate": 4.758843781186706e-05, "loss": 2.7359, "step": 27248 }, { "epoch": 1.6915388913029983, "grad_norm": 0.5928773993086276, "learning_rate": 4.7584830497086226e-05, "loss": 2.6575, "step": 27249 }, { "epoch": 1.6916009684027562, "grad_norm": 0.6178773091369735, "learning_rate": 4.7581223194905914e-05, "loss": 2.803, "step": 27250 }, { "epoch": 1.6916630455025141, "grad_norm": 0.5996422774144031, "learning_rate": 4.757761590534495e-05, "loss": 2.7765, "step": 27251 }, { "epoch": 1.691725122602272, "grad_norm": 0.6215799455989779, "learning_rate": 4.757400862842213e-05, "loss": 2.7351, "step": 27252 }, { "epoch": 1.69178719970203, "grad_norm": 0.6571515112730573, "learning_rate": 4.757040136415631e-05, "loss": 2.8028, "step": 27253 }, { "epoch": 1.6918492768017879, "grad_norm": 0.5783873813211622, "learning_rate": 4.7566794112566305e-05, "loss": 2.7842, "step": 27254 }, { "epoch": 1.6919113539015456, "grad_norm": 0.6635557372808616, "learning_rate": 4.756318687367093e-05, "loss": 2.788, "step": 27255 }, { "epoch": 1.6919734310013035, "grad_norm": 0.6001743053529982, "learning_rate": 4.755957964748898e-05, "loss": 2.8132, "step": 27256 }, { "epoch": 1.6920355081010614, "grad_norm": 0.626754586700954, "learning_rate": 4.755597243403932e-05, "loss": 2.7875, "step": 27257 }, { "epoch": 1.6920975852008193, "grad_norm": 0.5683491237279574, "learning_rate": 4.755236523334073e-05, "loss": 2.7855, "step": 27258 }, { "epoch": 1.6921596623005772, "grad_norm": 0.5685545858096757, "learning_rate": 4.754875804541205e-05, "loss": 2.8072, "step": 27259 }, { "epoch": 1.6922217394003352, "grad_norm": 0.6057481182760309, "learning_rate": 4.7545150870272086e-05, "loss": 2.8598, "step": 27260 }, { "epoch": 1.692283816500093, "grad_norm": 0.5527688275171205, "learning_rate": 4.754154370793968e-05, "loss": 2.6503, "step": 27261 }, { "epoch": 1.692345893599851, "grad_norm": 0.6233623507149344, "learning_rate": 4.753793655843362e-05, "loss": 2.7098, "step": 27262 }, { "epoch": 1.692407970699609, "grad_norm": 0.5716607083166734, "learning_rate": 4.753432942177276e-05, "loss": 2.7471, "step": 27263 }, { "epoch": 1.6924700477993668, "grad_norm": 0.5194271815191615, "learning_rate": 4.7530722297975886e-05, "loss": 2.7472, "step": 27264 }, { "epoch": 1.6925321248991247, "grad_norm": 0.6180049697543439, "learning_rate": 4.752711518706185e-05, "loss": 2.7586, "step": 27265 }, { "epoch": 1.6925942019988827, "grad_norm": 0.5459142965834918, "learning_rate": 4.7523508089049436e-05, "loss": 2.7116, "step": 27266 }, { "epoch": 1.6926562790986406, "grad_norm": 0.6143616556715001, "learning_rate": 4.7519901003957495e-05, "loss": 2.7555, "step": 27267 }, { "epoch": 1.6927183561983985, "grad_norm": 0.5690037590935132, "learning_rate": 4.751629393180484e-05, "loss": 2.7819, "step": 27268 }, { "epoch": 1.6927804332981564, "grad_norm": 0.5734800141085606, "learning_rate": 4.7512686872610264e-05, "loss": 2.8469, "step": 27269 }, { "epoch": 1.6928425103979143, "grad_norm": 0.7049880036219442, "learning_rate": 4.7509079826392616e-05, "loss": 2.7862, "step": 27270 }, { "epoch": 1.6929045874976723, "grad_norm": 0.5420950027410003, "learning_rate": 4.750547279317069e-05, "loss": 2.7408, "step": 27271 }, { "epoch": 1.6929666645974302, "grad_norm": 0.6003400097510508, "learning_rate": 4.7501865772963336e-05, "loss": 2.7018, "step": 27272 }, { "epoch": 1.6930287416971879, "grad_norm": 0.6032217847420398, "learning_rate": 4.749825876578934e-05, "loss": 2.778, "step": 27273 }, { "epoch": 1.6930908187969458, "grad_norm": 0.6422811825028881, "learning_rate": 4.7494651771667544e-05, "loss": 2.771, "step": 27274 }, { "epoch": 1.6931528958967037, "grad_norm": 0.6238787129528038, "learning_rate": 4.749104479061675e-05, "loss": 2.7059, "step": 27275 }, { "epoch": 1.6932149729964616, "grad_norm": 0.5946758480403872, "learning_rate": 4.7487437822655796e-05, "loss": 2.6401, "step": 27276 }, { "epoch": 1.6932770500962195, "grad_norm": 0.6892447440383193, "learning_rate": 4.748383086780347e-05, "loss": 2.8111, "step": 27277 }, { "epoch": 1.6933391271959775, "grad_norm": 0.6261044042912178, "learning_rate": 4.748022392607863e-05, "loss": 2.7059, "step": 27278 }, { "epoch": 1.6934012042957352, "grad_norm": 0.5789254292182371, "learning_rate": 4.7476616997500066e-05, "loss": 2.7443, "step": 27279 }, { "epoch": 1.693463281395493, "grad_norm": 0.5555763376507982, "learning_rate": 4.74730100820866e-05, "loss": 2.8087, "step": 27280 }, { "epoch": 1.693525358495251, "grad_norm": 0.6059709742538064, "learning_rate": 4.746940317985706e-05, "loss": 2.83, "step": 27281 }, { "epoch": 1.693587435595009, "grad_norm": 0.6111427489192518, "learning_rate": 4.746579629083025e-05, "loss": 2.7953, "step": 27282 }, { "epoch": 1.6936495126947668, "grad_norm": 0.5664150105818326, "learning_rate": 4.7462189415025007e-05, "loss": 2.7615, "step": 27283 }, { "epoch": 1.6937115897945247, "grad_norm": 0.5509445002603912, "learning_rate": 4.745858255246012e-05, "loss": 2.696, "step": 27284 }, { "epoch": 1.6937736668942827, "grad_norm": 0.621812392682346, "learning_rate": 4.745497570315442e-05, "loss": 2.6727, "step": 27285 }, { "epoch": 1.6938357439940406, "grad_norm": 0.5431641974752479, "learning_rate": 4.745136886712675e-05, "loss": 2.7045, "step": 27286 }, { "epoch": 1.6938978210937985, "grad_norm": 0.595152979226112, "learning_rate": 4.7447762044395896e-05, "loss": 2.7895, "step": 27287 }, { "epoch": 1.6939598981935564, "grad_norm": 0.564630632307815, "learning_rate": 4.74441552349807e-05, "loss": 2.7585, "step": 27288 }, { "epoch": 1.6940219752933143, "grad_norm": 0.6146560857567899, "learning_rate": 4.744054843889996e-05, "loss": 2.7032, "step": 27289 }, { "epoch": 1.6940840523930722, "grad_norm": 0.5711921544493225, "learning_rate": 4.74369416561725e-05, "loss": 2.6727, "step": 27290 }, { "epoch": 1.6941461294928302, "grad_norm": 0.6152444505623386, "learning_rate": 4.7433334886817134e-05, "loss": 2.7974, "step": 27291 }, { "epoch": 1.694208206592588, "grad_norm": 0.6005486205378235, "learning_rate": 4.742972813085269e-05, "loss": 2.7523, "step": 27292 }, { "epoch": 1.694270283692346, "grad_norm": 0.6099759342183478, "learning_rate": 4.742612138829798e-05, "loss": 2.7375, "step": 27293 }, { "epoch": 1.694332360792104, "grad_norm": 0.5742496819501501, "learning_rate": 4.7422514659171814e-05, "loss": 2.6935, "step": 27294 }, { "epoch": 1.6943944378918618, "grad_norm": 0.6147167147281338, "learning_rate": 4.741890794349302e-05, "loss": 2.797, "step": 27295 }, { "epoch": 1.6944565149916198, "grad_norm": 0.5693996648962993, "learning_rate": 4.7415301241280394e-05, "loss": 2.713, "step": 27296 }, { "epoch": 1.6945185920913775, "grad_norm": 0.6246494740776988, "learning_rate": 4.741169455255279e-05, "loss": 2.8245, "step": 27297 }, { "epoch": 1.6945806691911354, "grad_norm": 0.6003293365389244, "learning_rate": 4.740808787732899e-05, "loss": 2.8207, "step": 27298 }, { "epoch": 1.6946427462908933, "grad_norm": 0.6007781674785928, "learning_rate": 4.7404481215627834e-05, "loss": 2.733, "step": 27299 }, { "epoch": 1.6947048233906512, "grad_norm": 0.5906781490571837, "learning_rate": 4.7400874567468115e-05, "loss": 2.7312, "step": 27300 }, { "epoch": 1.6947669004904091, "grad_norm": 0.6069251970614117, "learning_rate": 4.739726793286868e-05, "loss": 2.7511, "step": 27301 }, { "epoch": 1.694828977590167, "grad_norm": 0.5463553548338717, "learning_rate": 4.739366131184832e-05, "loss": 2.7407, "step": 27302 }, { "epoch": 1.6948910546899247, "grad_norm": 0.6381092684952202, "learning_rate": 4.739005470442587e-05, "loss": 2.7586, "step": 27303 }, { "epoch": 1.6949531317896827, "grad_norm": 0.6127692350981633, "learning_rate": 4.7386448110620133e-05, "loss": 2.7824, "step": 27304 }, { "epoch": 1.6950152088894406, "grad_norm": 0.6126906351306759, "learning_rate": 4.7382841530449926e-05, "loss": 2.7956, "step": 27305 }, { "epoch": 1.6950772859891985, "grad_norm": 0.5892019253566592, "learning_rate": 4.737923496393408e-05, "loss": 2.7351, "step": 27306 }, { "epoch": 1.6951393630889564, "grad_norm": 0.5696948116166002, "learning_rate": 4.737562841109139e-05, "loss": 2.7256, "step": 27307 }, { "epoch": 1.6952014401887143, "grad_norm": 0.6213652539176686, "learning_rate": 4.73720218719407e-05, "loss": 2.7699, "step": 27308 }, { "epoch": 1.6952635172884722, "grad_norm": 0.5851214511847657, "learning_rate": 4.7368415346500795e-05, "loss": 2.7913, "step": 27309 }, { "epoch": 1.6953255943882302, "grad_norm": 0.630058583419489, "learning_rate": 4.736480883479052e-05, "loss": 2.795, "step": 27310 }, { "epoch": 1.695387671487988, "grad_norm": 0.5703182220171936, "learning_rate": 4.736120233682866e-05, "loss": 2.7778, "step": 27311 }, { "epoch": 1.695449748587746, "grad_norm": 0.6325689482897804, "learning_rate": 4.735759585263406e-05, "loss": 2.6524, "step": 27312 }, { "epoch": 1.695511825687504, "grad_norm": 0.5655482403427402, "learning_rate": 4.735398938222552e-05, "loss": 2.8381, "step": 27313 }, { "epoch": 1.6955739027872618, "grad_norm": 0.5824725123292697, "learning_rate": 4.735038292562186e-05, "loss": 2.742, "step": 27314 }, { "epoch": 1.6956359798870198, "grad_norm": 0.5868938017282075, "learning_rate": 4.73467764828419e-05, "loss": 2.7982, "step": 27315 }, { "epoch": 1.6956980569867777, "grad_norm": 0.530611046651426, "learning_rate": 4.734317005390444e-05, "loss": 2.7237, "step": 27316 }, { "epoch": 1.6957601340865356, "grad_norm": 0.6053910728587819, "learning_rate": 4.73395636388283e-05, "loss": 2.6847, "step": 27317 }, { "epoch": 1.6958222111862935, "grad_norm": 0.5113693966685599, "learning_rate": 4.733595723763232e-05, "loss": 2.76, "step": 27318 }, { "epoch": 1.6958842882860514, "grad_norm": 0.5593637348084953, "learning_rate": 4.733235085033529e-05, "loss": 2.7388, "step": 27319 }, { "epoch": 1.6959463653858093, "grad_norm": 0.5426698314703728, "learning_rate": 4.732874447695604e-05, "loss": 2.7909, "step": 27320 }, { "epoch": 1.696008442485567, "grad_norm": 0.633816240900554, "learning_rate": 4.732513811751337e-05, "loss": 2.7697, "step": 27321 }, { "epoch": 1.696070519585325, "grad_norm": 0.5258577390358277, "learning_rate": 4.7321531772026115e-05, "loss": 2.7534, "step": 27322 }, { "epoch": 1.6961325966850829, "grad_norm": 0.6020069709672049, "learning_rate": 4.731792544051307e-05, "loss": 2.7852, "step": 27323 }, { "epoch": 1.6961946737848408, "grad_norm": 0.5586230663796176, "learning_rate": 4.731431912299306e-05, "loss": 2.7968, "step": 27324 }, { "epoch": 1.6962567508845987, "grad_norm": 0.6225185486125075, "learning_rate": 4.7310712819484895e-05, "loss": 2.7691, "step": 27325 }, { "epoch": 1.6963188279843566, "grad_norm": 0.5612589072049553, "learning_rate": 4.730710653000741e-05, "loss": 2.7004, "step": 27326 }, { "epoch": 1.6963809050841143, "grad_norm": 0.6131178036733095, "learning_rate": 4.730350025457938e-05, "loss": 2.6326, "step": 27327 }, { "epoch": 1.6964429821838722, "grad_norm": 0.540136290587823, "learning_rate": 4.7299893993219665e-05, "loss": 2.8024, "step": 27328 }, { "epoch": 1.6965050592836302, "grad_norm": 0.6424222484626042, "learning_rate": 4.729628774594706e-05, "loss": 2.7777, "step": 27329 }, { "epoch": 1.696567136383388, "grad_norm": 0.5375829377615191, "learning_rate": 4.7292681512780367e-05, "loss": 2.8107, "step": 27330 }, { "epoch": 1.696629213483146, "grad_norm": 0.567944955655337, "learning_rate": 4.7289075293738415e-05, "loss": 2.7578, "step": 27331 }, { "epoch": 1.696691290582904, "grad_norm": 0.5654994150760773, "learning_rate": 4.728546908884001e-05, "loss": 2.8063, "step": 27332 }, { "epoch": 1.6967533676826618, "grad_norm": 0.5368730892285061, "learning_rate": 4.728186289810399e-05, "loss": 2.7801, "step": 27333 }, { "epoch": 1.6968154447824197, "grad_norm": 0.6022248275661574, "learning_rate": 4.7278256721549127e-05, "loss": 2.6962, "step": 27334 }, { "epoch": 1.6968775218821777, "grad_norm": 0.5401186892973523, "learning_rate": 4.727465055919427e-05, "loss": 2.7775, "step": 27335 }, { "epoch": 1.6969395989819356, "grad_norm": 0.5561832891601256, "learning_rate": 4.727104441105823e-05, "loss": 2.7375, "step": 27336 }, { "epoch": 1.6970016760816935, "grad_norm": 0.5733932218165646, "learning_rate": 4.7267438277159804e-05, "loss": 2.7666, "step": 27337 }, { "epoch": 1.6970637531814514, "grad_norm": 0.5962898241894752, "learning_rate": 4.726383215751781e-05, "loss": 2.7417, "step": 27338 }, { "epoch": 1.6971258302812093, "grad_norm": 0.6110494789077613, "learning_rate": 4.726022605215108e-05, "loss": 2.7183, "step": 27339 }, { "epoch": 1.6971879073809673, "grad_norm": 0.5688431244370851, "learning_rate": 4.725661996107842e-05, "loss": 2.6982, "step": 27340 }, { "epoch": 1.6972499844807252, "grad_norm": 0.5670006888366902, "learning_rate": 4.7253013884318625e-05, "loss": 2.7585, "step": 27341 }, { "epoch": 1.697312061580483, "grad_norm": 0.5998707304220525, "learning_rate": 4.724940782189054e-05, "loss": 2.6224, "step": 27342 }, { "epoch": 1.697374138680241, "grad_norm": 0.5516462365148534, "learning_rate": 4.724580177381294e-05, "loss": 2.8066, "step": 27343 }, { "epoch": 1.697436215779999, "grad_norm": 0.5658471077107142, "learning_rate": 4.7242195740104686e-05, "loss": 2.7473, "step": 27344 }, { "epoch": 1.6974982928797566, "grad_norm": 0.5991219891305962, "learning_rate": 4.7238589720784545e-05, "loss": 2.8054, "step": 27345 }, { "epoch": 1.6975603699795145, "grad_norm": 0.5230565179188071, "learning_rate": 4.7234983715871364e-05, "loss": 2.7863, "step": 27346 }, { "epoch": 1.6976224470792725, "grad_norm": 0.6217727548325954, "learning_rate": 4.723137772538393e-05, "loss": 2.8701, "step": 27347 }, { "epoch": 1.6976845241790304, "grad_norm": 0.5701154342263717, "learning_rate": 4.722777174934109e-05, "loss": 2.7685, "step": 27348 }, { "epoch": 1.6977466012787883, "grad_norm": 0.5093427928735588, "learning_rate": 4.722416578776162e-05, "loss": 2.751, "step": 27349 }, { "epoch": 1.697808678378546, "grad_norm": 0.5271144509710687, "learning_rate": 4.722055984066434e-05, "loss": 2.6444, "step": 27350 }, { "epoch": 1.697870755478304, "grad_norm": 0.5177738035726674, "learning_rate": 4.72169539080681e-05, "loss": 2.8242, "step": 27351 }, { "epoch": 1.6979328325780618, "grad_norm": 0.5615419865737249, "learning_rate": 4.721334798999168e-05, "loss": 2.7235, "step": 27352 }, { "epoch": 1.6979949096778197, "grad_norm": 0.5694456174755679, "learning_rate": 4.72097420864539e-05, "loss": 2.7243, "step": 27353 }, { "epoch": 1.6980569867775777, "grad_norm": 0.5656209039276352, "learning_rate": 4.720613619747358e-05, "loss": 2.8069, "step": 27354 }, { "epoch": 1.6981190638773356, "grad_norm": 0.5990846804719853, "learning_rate": 4.7202530323069516e-05, "loss": 2.7808, "step": 27355 }, { "epoch": 1.6981811409770935, "grad_norm": 0.5242916937617385, "learning_rate": 4.719892446326053e-05, "loss": 2.7673, "step": 27356 }, { "epoch": 1.6982432180768514, "grad_norm": 0.6046698190632498, "learning_rate": 4.719531861806544e-05, "loss": 2.7405, "step": 27357 }, { "epoch": 1.6983052951766093, "grad_norm": 0.5502828954199469, "learning_rate": 4.7191712787503055e-05, "loss": 2.6712, "step": 27358 }, { "epoch": 1.6983673722763672, "grad_norm": 0.6010316512949906, "learning_rate": 4.7188106971592184e-05, "loss": 2.7503, "step": 27359 }, { "epoch": 1.6984294493761252, "grad_norm": 0.6088196595928874, "learning_rate": 4.718450117035164e-05, "loss": 2.7222, "step": 27360 }, { "epoch": 1.698491526475883, "grad_norm": 0.6546186196957937, "learning_rate": 4.718089538380023e-05, "loss": 2.7255, "step": 27361 }, { "epoch": 1.698553603575641, "grad_norm": 0.5812557866208382, "learning_rate": 4.717728961195679e-05, "loss": 2.7552, "step": 27362 }, { "epoch": 1.698615680675399, "grad_norm": 0.6403917027383289, "learning_rate": 4.71736838548401e-05, "loss": 2.782, "step": 27363 }, { "epoch": 1.6986777577751568, "grad_norm": 0.6039873273543198, "learning_rate": 4.7170078112469005e-05, "loss": 2.8009, "step": 27364 }, { "epoch": 1.6987398348749148, "grad_norm": 0.6564996789348947, "learning_rate": 4.716647238486229e-05, "loss": 2.6979, "step": 27365 }, { "epoch": 1.6988019119746727, "grad_norm": 0.6159102345494066, "learning_rate": 4.716286667203877e-05, "loss": 2.7838, "step": 27366 }, { "epoch": 1.6988639890744306, "grad_norm": 0.5332297573494948, "learning_rate": 4.715926097401727e-05, "loss": 2.7386, "step": 27367 }, { "epoch": 1.6989260661741883, "grad_norm": 0.5858580471847347, "learning_rate": 4.715565529081659e-05, "loss": 2.8113, "step": 27368 }, { "epoch": 1.6989881432739462, "grad_norm": 0.589469184429205, "learning_rate": 4.715204962245555e-05, "loss": 2.7599, "step": 27369 }, { "epoch": 1.6990502203737041, "grad_norm": 0.5979446762338557, "learning_rate": 4.7148443968952954e-05, "loss": 2.8009, "step": 27370 }, { "epoch": 1.699112297473462, "grad_norm": 0.5996678973040942, "learning_rate": 4.714483833032763e-05, "loss": 2.7243, "step": 27371 }, { "epoch": 1.69917437457322, "grad_norm": 0.649302537563153, "learning_rate": 4.7141232706598364e-05, "loss": 2.7848, "step": 27372 }, { "epoch": 1.6992364516729779, "grad_norm": 0.6092640702442984, "learning_rate": 4.713762709778399e-05, "loss": 2.7611, "step": 27373 }, { "epoch": 1.6992985287727356, "grad_norm": 0.6353597120389163, "learning_rate": 4.7134021503903306e-05, "loss": 2.8681, "step": 27374 }, { "epoch": 1.6993606058724935, "grad_norm": 0.6166451642652191, "learning_rate": 4.713041592497513e-05, "loss": 2.7399, "step": 27375 }, { "epoch": 1.6994226829722514, "grad_norm": 0.5884513841997654, "learning_rate": 4.7126810361018275e-05, "loss": 2.7672, "step": 27376 }, { "epoch": 1.6994847600720093, "grad_norm": 0.6151108148416357, "learning_rate": 4.712320481205154e-05, "loss": 2.7335, "step": 27377 }, { "epoch": 1.6995468371717672, "grad_norm": 0.5848065715512539, "learning_rate": 4.711959927809374e-05, "loss": 2.8196, "step": 27378 }, { "epoch": 1.6996089142715252, "grad_norm": 0.6716507604015518, "learning_rate": 4.7115993759163696e-05, "loss": 2.8272, "step": 27379 }, { "epoch": 1.699670991371283, "grad_norm": 0.5348814098440333, "learning_rate": 4.711238825528022e-05, "loss": 2.7733, "step": 27380 }, { "epoch": 1.699733068471041, "grad_norm": 0.6833832334891506, "learning_rate": 4.71087827664621e-05, "loss": 2.7435, "step": 27381 }, { "epoch": 1.699795145570799, "grad_norm": 0.5384611562711805, "learning_rate": 4.710517729272818e-05, "loss": 2.7506, "step": 27382 }, { "epoch": 1.6998572226705568, "grad_norm": 0.6997805847565475, "learning_rate": 4.7101571834097227e-05, "loss": 2.7556, "step": 27383 }, { "epoch": 1.6999192997703148, "grad_norm": 0.556677994393926, "learning_rate": 4.7097966390588084e-05, "loss": 2.7408, "step": 27384 }, { "epoch": 1.6999813768700727, "grad_norm": 0.6041549950053243, "learning_rate": 4.709436096221957e-05, "loss": 2.6988, "step": 27385 }, { "epoch": 1.7000434539698306, "grad_norm": 0.6106931661238151, "learning_rate": 4.7090755549010466e-05, "loss": 2.8224, "step": 27386 }, { "epoch": 1.7001055310695885, "grad_norm": 0.5574963066166901, "learning_rate": 4.708715015097961e-05, "loss": 2.7705, "step": 27387 }, { "epoch": 1.7001676081693464, "grad_norm": 0.5920939439533753, "learning_rate": 4.708354476814579e-05, "loss": 2.7774, "step": 27388 }, { "epoch": 1.7002296852691043, "grad_norm": 0.5520708664500151, "learning_rate": 4.707993940052783e-05, "loss": 2.774, "step": 27389 }, { "epoch": 1.7002917623688623, "grad_norm": 0.5715508537711924, "learning_rate": 4.707633404814454e-05, "loss": 2.7271, "step": 27390 }, { "epoch": 1.7003538394686202, "grad_norm": 0.5773993138894239, "learning_rate": 4.707272871101472e-05, "loss": 2.8161, "step": 27391 }, { "epoch": 1.7004159165683779, "grad_norm": 0.5751834473090469, "learning_rate": 4.706912338915719e-05, "loss": 2.816, "step": 27392 }, { "epoch": 1.7004779936681358, "grad_norm": 0.5239219700940426, "learning_rate": 4.7065518082590735e-05, "loss": 2.8033, "step": 27393 }, { "epoch": 1.7005400707678937, "grad_norm": 0.5558588527910215, "learning_rate": 4.7061912791334206e-05, "loss": 2.7493, "step": 27394 }, { "epoch": 1.7006021478676516, "grad_norm": 0.5613639273272432, "learning_rate": 4.705830751540638e-05, "loss": 2.7398, "step": 27395 }, { "epoch": 1.7006642249674095, "grad_norm": 0.5020737558720324, "learning_rate": 4.705470225482608e-05, "loss": 2.7321, "step": 27396 }, { "epoch": 1.7007263020671675, "grad_norm": 0.5695332206618722, "learning_rate": 4.705109700961211e-05, "loss": 2.6581, "step": 27397 }, { "epoch": 1.7007883791669252, "grad_norm": 0.5744915528286462, "learning_rate": 4.70474917797833e-05, "loss": 2.7824, "step": 27398 }, { "epoch": 1.700850456266683, "grad_norm": 0.5907831101487887, "learning_rate": 4.704388656535842e-05, "loss": 2.7372, "step": 27399 }, { "epoch": 1.700912533366441, "grad_norm": 0.5656993269163959, "learning_rate": 4.704028136635632e-05, "loss": 2.7069, "step": 27400 }, { "epoch": 1.700974610466199, "grad_norm": 0.5648410348576335, "learning_rate": 4.703667618279578e-05, "loss": 2.759, "step": 27401 }, { "epoch": 1.7010366875659568, "grad_norm": 0.6389755553627862, "learning_rate": 4.703307101469562e-05, "loss": 2.7428, "step": 27402 }, { "epoch": 1.7010987646657147, "grad_norm": 0.5616180870892366, "learning_rate": 4.702946586207465e-05, "loss": 2.7813, "step": 27403 }, { "epoch": 1.7011608417654727, "grad_norm": 0.6406712631640575, "learning_rate": 4.702586072495168e-05, "loss": 2.8116, "step": 27404 }, { "epoch": 1.7012229188652306, "grad_norm": 0.6098337647355813, "learning_rate": 4.702225560334551e-05, "loss": 2.6678, "step": 27405 }, { "epoch": 1.7012849959649885, "grad_norm": 0.6708940841777884, "learning_rate": 4.701865049727496e-05, "loss": 2.7519, "step": 27406 }, { "epoch": 1.7013470730647464, "grad_norm": 0.6350401534128448, "learning_rate": 4.7015045406758834e-05, "loss": 2.7763, "step": 27407 }, { "epoch": 1.7014091501645043, "grad_norm": 0.6407297793043153, "learning_rate": 4.701144033181594e-05, "loss": 2.8812, "step": 27408 }, { "epoch": 1.7014712272642623, "grad_norm": 0.6031581721300172, "learning_rate": 4.700783527246509e-05, "loss": 2.7456, "step": 27409 }, { "epoch": 1.7015333043640202, "grad_norm": 0.612303081309571, "learning_rate": 4.700423022872508e-05, "loss": 2.7629, "step": 27410 }, { "epoch": 1.701595381463778, "grad_norm": 0.5605456787521943, "learning_rate": 4.700062520061474e-05, "loss": 2.7273, "step": 27411 }, { "epoch": 1.701657458563536, "grad_norm": 0.567001634940224, "learning_rate": 4.699702018815286e-05, "loss": 2.6834, "step": 27412 }, { "epoch": 1.701719535663294, "grad_norm": 0.5503054789193286, "learning_rate": 4.699341519135826e-05, "loss": 2.7985, "step": 27413 }, { "epoch": 1.7017816127630518, "grad_norm": 0.572353421399516, "learning_rate": 4.698981021024975e-05, "loss": 2.723, "step": 27414 }, { "epoch": 1.7018436898628098, "grad_norm": 0.627302933786745, "learning_rate": 4.698620524484611e-05, "loss": 2.7037, "step": 27415 }, { "epoch": 1.7019057669625675, "grad_norm": 0.5657889850280501, "learning_rate": 4.698260029516616e-05, "loss": 2.7091, "step": 27416 }, { "epoch": 1.7019678440623254, "grad_norm": 0.5974341584483412, "learning_rate": 4.6978995361228745e-05, "loss": 2.7971, "step": 27417 }, { "epoch": 1.7020299211620833, "grad_norm": 0.5513330765921162, "learning_rate": 4.697539044305263e-05, "loss": 2.7566, "step": 27418 }, { "epoch": 1.7020919982618412, "grad_norm": 0.5865404714066463, "learning_rate": 4.6971785540656646e-05, "loss": 2.6667, "step": 27419 }, { "epoch": 1.7021540753615991, "grad_norm": 0.5883315704425556, "learning_rate": 4.6968180654059584e-05, "loss": 2.8366, "step": 27420 }, { "epoch": 1.702216152461357, "grad_norm": 0.5604035729316829, "learning_rate": 4.6964575783280266e-05, "loss": 2.7816, "step": 27421 }, { "epoch": 1.7022782295611147, "grad_norm": 0.5144998527745369, "learning_rate": 4.696097092833749e-05, "loss": 2.7261, "step": 27422 }, { "epoch": 1.7023403066608727, "grad_norm": 0.5372176182654915, "learning_rate": 4.695736608925007e-05, "loss": 2.6057, "step": 27423 }, { "epoch": 1.7024023837606306, "grad_norm": 0.5416453583512038, "learning_rate": 4.69537612660368e-05, "loss": 2.7299, "step": 27424 }, { "epoch": 1.7024644608603885, "grad_norm": 0.5402055314777573, "learning_rate": 4.695015645871651e-05, "loss": 2.7334, "step": 27425 }, { "epoch": 1.7025265379601464, "grad_norm": 0.5768928542840692, "learning_rate": 4.6946551667308e-05, "loss": 2.7469, "step": 27426 }, { "epoch": 1.7025886150599043, "grad_norm": 0.5111015324829539, "learning_rate": 4.694294689183005e-05, "loss": 2.6816, "step": 27427 }, { "epoch": 1.7026506921596622, "grad_norm": 0.6637120163460701, "learning_rate": 4.6939342132301506e-05, "loss": 2.7835, "step": 27428 }, { "epoch": 1.7027127692594202, "grad_norm": 0.5406637872442076, "learning_rate": 4.693573738874114e-05, "loss": 2.7459, "step": 27429 }, { "epoch": 1.702774846359178, "grad_norm": 0.6443967736725221, "learning_rate": 4.6932132661167794e-05, "loss": 2.7241, "step": 27430 }, { "epoch": 1.702836923458936, "grad_norm": 0.5236813560787397, "learning_rate": 4.692852794960024e-05, "loss": 2.8011, "step": 27431 }, { "epoch": 1.702899000558694, "grad_norm": 0.5928567642392905, "learning_rate": 4.692492325405731e-05, "loss": 2.8303, "step": 27432 }, { "epoch": 1.7029610776584518, "grad_norm": 0.5376512880104662, "learning_rate": 4.69213185745578e-05, "loss": 2.6594, "step": 27433 }, { "epoch": 1.7030231547582098, "grad_norm": 0.6176498467952786, "learning_rate": 4.691771391112052e-05, "loss": 2.7412, "step": 27434 }, { "epoch": 1.7030852318579677, "grad_norm": 0.5289412708042134, "learning_rate": 4.6914109263764274e-05, "loss": 2.6682, "step": 27435 }, { "epoch": 1.7031473089577256, "grad_norm": 0.605276163297549, "learning_rate": 4.6910504632507876e-05, "loss": 2.8009, "step": 27436 }, { "epoch": 1.7032093860574835, "grad_norm": 0.6301523977209681, "learning_rate": 4.690690001737011e-05, "loss": 2.8538, "step": 27437 }, { "epoch": 1.7032714631572414, "grad_norm": 0.5378893650680114, "learning_rate": 4.690329541836981e-05, "loss": 2.8324, "step": 27438 }, { "epoch": 1.7033335402569993, "grad_norm": 0.5912976819141174, "learning_rate": 4.6899690835525774e-05, "loss": 2.7839, "step": 27439 }, { "epoch": 1.703395617356757, "grad_norm": 0.5408663216607218, "learning_rate": 4.689608626885679e-05, "loss": 2.77, "step": 27440 }, { "epoch": 1.703457694456515, "grad_norm": 0.5432695063544212, "learning_rate": 4.689248171838168e-05, "loss": 2.715, "step": 27441 }, { "epoch": 1.7035197715562729, "grad_norm": 0.5714415778655348, "learning_rate": 4.688887718411925e-05, "loss": 2.7463, "step": 27442 }, { "epoch": 1.7035818486560308, "grad_norm": 0.4858900034712481, "learning_rate": 4.6885272666088305e-05, "loss": 2.6586, "step": 27443 }, { "epoch": 1.7036439257557887, "grad_norm": 0.5757125057761227, "learning_rate": 4.688166816430764e-05, "loss": 2.7842, "step": 27444 }, { "epoch": 1.7037060028555466, "grad_norm": 0.5207168989370109, "learning_rate": 4.687806367879608e-05, "loss": 2.835, "step": 27445 }, { "epoch": 1.7037680799553043, "grad_norm": 0.5075260528210039, "learning_rate": 4.6874459209572405e-05, "loss": 2.7601, "step": 27446 }, { "epoch": 1.7038301570550622, "grad_norm": 0.5399872132394548, "learning_rate": 4.687085475665546e-05, "loss": 2.7483, "step": 27447 }, { "epoch": 1.7038922341548202, "grad_norm": 0.5397543406663141, "learning_rate": 4.686725032006399e-05, "loss": 2.7328, "step": 27448 }, { "epoch": 1.703954311254578, "grad_norm": 0.5479306522610273, "learning_rate": 4.686364589981685e-05, "loss": 2.7301, "step": 27449 }, { "epoch": 1.704016388354336, "grad_norm": 0.5820499647575179, "learning_rate": 4.6860041495932845e-05, "loss": 2.7448, "step": 27450 }, { "epoch": 1.704078465454094, "grad_norm": 0.5286760806996904, "learning_rate": 4.685643710843076e-05, "loss": 2.88, "step": 27451 }, { "epoch": 1.7041405425538518, "grad_norm": 0.5671383040349377, "learning_rate": 4.68528327373294e-05, "loss": 2.7813, "step": 27452 }, { "epoch": 1.7042026196536098, "grad_norm": 0.5264540337026069, "learning_rate": 4.684922838264758e-05, "loss": 2.7864, "step": 27453 }, { "epoch": 1.7042646967533677, "grad_norm": 0.5712896102511397, "learning_rate": 4.684562404440409e-05, "loss": 2.7633, "step": 27454 }, { "epoch": 1.7043267738531256, "grad_norm": 0.554271999067077, "learning_rate": 4.684201972261776e-05, "loss": 2.7815, "step": 27455 }, { "epoch": 1.7043888509528835, "grad_norm": 0.5214959475486916, "learning_rate": 4.683841541730737e-05, "loss": 2.8295, "step": 27456 }, { "epoch": 1.7044509280526414, "grad_norm": 0.6017098432152613, "learning_rate": 4.683481112849175e-05, "loss": 2.7516, "step": 27457 }, { "epoch": 1.7045130051523993, "grad_norm": 0.5409079213818233, "learning_rate": 4.683120685618967e-05, "loss": 2.7835, "step": 27458 }, { "epoch": 1.7045750822521573, "grad_norm": 0.529356836955424, "learning_rate": 4.6827602600419966e-05, "loss": 2.6812, "step": 27459 }, { "epoch": 1.7046371593519152, "grad_norm": 0.5280575195630266, "learning_rate": 4.682399836120142e-05, "loss": 2.789, "step": 27460 }, { "epoch": 1.704699236451673, "grad_norm": 0.5475380556388917, "learning_rate": 4.682039413855285e-05, "loss": 2.8025, "step": 27461 }, { "epoch": 1.704761313551431, "grad_norm": 0.5258237620337539, "learning_rate": 4.681678993249305e-05, "loss": 2.6769, "step": 27462 }, { "epoch": 1.704823390651189, "grad_norm": 0.5595762242509216, "learning_rate": 4.681318574304084e-05, "loss": 2.7693, "step": 27463 }, { "epoch": 1.7048854677509466, "grad_norm": 0.5647866079042212, "learning_rate": 4.680958157021502e-05, "loss": 2.7325, "step": 27464 }, { "epoch": 1.7049475448507045, "grad_norm": 0.5208531534874177, "learning_rate": 4.680597741403436e-05, "loss": 2.7467, "step": 27465 }, { "epoch": 1.7050096219504625, "grad_norm": 0.5322540276116553, "learning_rate": 4.680237327451772e-05, "loss": 2.7245, "step": 27466 }, { "epoch": 1.7050716990502204, "grad_norm": 0.5249542636476917, "learning_rate": 4.679876915168385e-05, "loss": 2.797, "step": 27467 }, { "epoch": 1.7051337761499783, "grad_norm": 0.5013139863685951, "learning_rate": 4.6795165045551595e-05, "loss": 2.7196, "step": 27468 }, { "epoch": 1.7051958532497362, "grad_norm": 0.5662425626809492, "learning_rate": 4.6791560956139734e-05, "loss": 2.7213, "step": 27469 }, { "epoch": 1.705257930349494, "grad_norm": 0.4645773180185868, "learning_rate": 4.6787956883467086e-05, "loss": 2.7908, "step": 27470 }, { "epoch": 1.7053200074492518, "grad_norm": 0.6141449259020343, "learning_rate": 4.678435282755244e-05, "loss": 2.7759, "step": 27471 }, { "epoch": 1.7053820845490097, "grad_norm": 0.5232646998524543, "learning_rate": 4.678074878841461e-05, "loss": 2.736, "step": 27472 }, { "epoch": 1.7054441616487677, "grad_norm": 0.570035030852894, "learning_rate": 4.677714476607239e-05, "loss": 2.8158, "step": 27473 }, { "epoch": 1.7055062387485256, "grad_norm": 0.5216773029874164, "learning_rate": 4.677354076054459e-05, "loss": 2.7727, "step": 27474 }, { "epoch": 1.7055683158482835, "grad_norm": 0.5299998046316831, "learning_rate": 4.676993677185002e-05, "loss": 2.7581, "step": 27475 }, { "epoch": 1.7056303929480414, "grad_norm": 0.5497319662944902, "learning_rate": 4.676633280000746e-05, "loss": 2.7024, "step": 27476 }, { "epoch": 1.7056924700477993, "grad_norm": 0.510401362235096, "learning_rate": 4.676272884503574e-05, "loss": 2.6542, "step": 27477 }, { "epoch": 1.7057545471475573, "grad_norm": 0.5706190565620746, "learning_rate": 4.675912490695364e-05, "loss": 2.7324, "step": 27478 }, { "epoch": 1.7058166242473152, "grad_norm": 0.5157718160437345, "learning_rate": 4.675552098577998e-05, "loss": 2.7045, "step": 27479 }, { "epoch": 1.705878701347073, "grad_norm": 0.5561414923542849, "learning_rate": 4.6751917081533544e-05, "loss": 2.6435, "step": 27480 }, { "epoch": 1.705940778446831, "grad_norm": 0.5236986277566107, "learning_rate": 4.674831319423313e-05, "loss": 2.7058, "step": 27481 }, { "epoch": 1.706002855546589, "grad_norm": 0.6223076527801259, "learning_rate": 4.6744709323897584e-05, "loss": 2.7952, "step": 27482 }, { "epoch": 1.7060649326463468, "grad_norm": 0.4997330638492708, "learning_rate": 4.6741105470545666e-05, "loss": 2.7847, "step": 27483 }, { "epoch": 1.7061270097461048, "grad_norm": 0.6651332301215106, "learning_rate": 4.67375016341962e-05, "loss": 2.7996, "step": 27484 }, { "epoch": 1.7061890868458627, "grad_norm": 0.5293860553906665, "learning_rate": 4.673389781486797e-05, "loss": 2.8301, "step": 27485 }, { "epoch": 1.7062511639456206, "grad_norm": 0.6128301723174958, "learning_rate": 4.673029401257981e-05, "loss": 2.788, "step": 27486 }, { "epoch": 1.7063132410453785, "grad_norm": 0.5699725771046145, "learning_rate": 4.6726690227350474e-05, "loss": 2.7863, "step": 27487 }, { "epoch": 1.7063753181451362, "grad_norm": 0.5721640643201289, "learning_rate": 4.672308645919881e-05, "loss": 2.7355, "step": 27488 }, { "epoch": 1.7064373952448941, "grad_norm": 0.5412276365580415, "learning_rate": 4.671948270814359e-05, "loss": 2.7778, "step": 27489 }, { "epoch": 1.706499472344652, "grad_norm": 0.5427849273179314, "learning_rate": 4.671587897420362e-05, "loss": 2.7933, "step": 27490 }, { "epoch": 1.70656154944441, "grad_norm": 0.5429757206970206, "learning_rate": 4.671227525739772e-05, "loss": 2.7614, "step": 27491 }, { "epoch": 1.7066236265441679, "grad_norm": 0.5484331820934764, "learning_rate": 4.670867155774466e-05, "loss": 2.7688, "step": 27492 }, { "epoch": 1.7066857036439258, "grad_norm": 0.5639904726488543, "learning_rate": 4.670506787526328e-05, "loss": 2.7438, "step": 27493 }, { "epoch": 1.7067477807436835, "grad_norm": 0.6284907372231382, "learning_rate": 4.670146420997235e-05, "loss": 2.7433, "step": 27494 }, { "epoch": 1.7068098578434414, "grad_norm": 0.5897652529595178, "learning_rate": 4.6697860561890686e-05, "loss": 2.8617, "step": 27495 }, { "epoch": 1.7068719349431993, "grad_norm": 0.6685556209467424, "learning_rate": 4.669425693103707e-05, "loss": 2.6796, "step": 27496 }, { "epoch": 1.7069340120429572, "grad_norm": 0.5600623452444239, "learning_rate": 4.6690653317430335e-05, "loss": 2.7869, "step": 27497 }, { "epoch": 1.7069960891427152, "grad_norm": 0.6371382519079327, "learning_rate": 4.6687049721089254e-05, "loss": 2.653, "step": 27498 }, { "epoch": 1.707058166242473, "grad_norm": 0.5813857217479568, "learning_rate": 4.6683446142032646e-05, "loss": 2.7305, "step": 27499 }, { "epoch": 1.707120243342231, "grad_norm": 0.5679684897399894, "learning_rate": 4.6679842580279316e-05, "loss": 2.7563, "step": 27500 }, { "epoch": 1.707182320441989, "grad_norm": 0.5818038598649546, "learning_rate": 4.667623903584803e-05, "loss": 2.7747, "step": 27501 }, { "epoch": 1.7072443975417468, "grad_norm": 0.5615167501300159, "learning_rate": 4.667263550875763e-05, "loss": 2.7981, "step": 27502 }, { "epoch": 1.7073064746415048, "grad_norm": 0.5938717817597898, "learning_rate": 4.6669031999026885e-05, "loss": 2.81, "step": 27503 }, { "epoch": 1.7073685517412627, "grad_norm": 0.6066879057528682, "learning_rate": 4.6665428506674616e-05, "loss": 2.7469, "step": 27504 }, { "epoch": 1.7074306288410206, "grad_norm": 0.6051202324732828, "learning_rate": 4.666182503171961e-05, "loss": 2.6806, "step": 27505 }, { "epoch": 1.7074927059407785, "grad_norm": 0.6085049457913267, "learning_rate": 4.665822157418068e-05, "loss": 2.7517, "step": 27506 }, { "epoch": 1.7075547830405364, "grad_norm": 0.6100111599589219, "learning_rate": 4.6654618134076616e-05, "loss": 2.6555, "step": 27507 }, { "epoch": 1.7076168601402943, "grad_norm": 0.5965532584096505, "learning_rate": 4.665101471142623e-05, "loss": 2.7612, "step": 27508 }, { "epoch": 1.7076789372400523, "grad_norm": 0.6352586680064496, "learning_rate": 4.664741130624829e-05, "loss": 2.7821, "step": 27509 }, { "epoch": 1.7077410143398102, "grad_norm": 0.5792307974851603, "learning_rate": 4.664380791856164e-05, "loss": 2.7214, "step": 27510 }, { "epoch": 1.707803091439568, "grad_norm": 0.6226180702670867, "learning_rate": 4.6640204548385067e-05, "loss": 2.8137, "step": 27511 }, { "epoch": 1.7078651685393258, "grad_norm": 0.634450370558063, "learning_rate": 4.6636601195737336e-05, "loss": 2.7457, "step": 27512 }, { "epoch": 1.7079272456390837, "grad_norm": 0.5390453888761985, "learning_rate": 4.6632997860637297e-05, "loss": 2.8032, "step": 27513 }, { "epoch": 1.7079893227388416, "grad_norm": 0.6125411895604091, "learning_rate": 4.66293945431037e-05, "loss": 2.7378, "step": 27514 }, { "epoch": 1.7080513998385995, "grad_norm": 0.5600538843916404, "learning_rate": 4.6625791243155375e-05, "loss": 2.6999, "step": 27515 }, { "epoch": 1.7081134769383575, "grad_norm": 0.6047107011942359, "learning_rate": 4.662218796081113e-05, "loss": 2.8323, "step": 27516 }, { "epoch": 1.7081755540381154, "grad_norm": 0.5656725636922101, "learning_rate": 4.661858469608974e-05, "loss": 2.7888, "step": 27517 }, { "epoch": 1.708237631137873, "grad_norm": 0.5900898098075948, "learning_rate": 4.6614981449010025e-05, "loss": 2.7692, "step": 27518 }, { "epoch": 1.708299708237631, "grad_norm": 0.6120907593837757, "learning_rate": 4.661137821959076e-05, "loss": 2.7386, "step": 27519 }, { "epoch": 1.708361785337389, "grad_norm": 0.6044280429510881, "learning_rate": 4.660777500785077e-05, "loss": 2.7941, "step": 27520 }, { "epoch": 1.7084238624371468, "grad_norm": 0.6328309138880284, "learning_rate": 4.6604171813808836e-05, "loss": 2.6983, "step": 27521 }, { "epoch": 1.7084859395369048, "grad_norm": 0.5583380534081671, "learning_rate": 4.660056863748377e-05, "loss": 2.8156, "step": 27522 }, { "epoch": 1.7085480166366627, "grad_norm": 0.665246536224362, "learning_rate": 4.659696547889435e-05, "loss": 2.7435, "step": 27523 }, { "epoch": 1.7086100937364206, "grad_norm": 0.5724098102042134, "learning_rate": 4.65933623380594e-05, "loss": 2.6581, "step": 27524 }, { "epoch": 1.7086721708361785, "grad_norm": 0.6778656890035535, "learning_rate": 4.6589759214997704e-05, "loss": 2.6897, "step": 27525 }, { "epoch": 1.7087342479359364, "grad_norm": 0.6085838170153034, "learning_rate": 4.6586156109728054e-05, "loss": 2.7503, "step": 27526 }, { "epoch": 1.7087963250356943, "grad_norm": 0.6382200643659165, "learning_rate": 4.658255302226926e-05, "loss": 2.7894, "step": 27527 }, { "epoch": 1.7088584021354523, "grad_norm": 0.6062398614723891, "learning_rate": 4.657894995264011e-05, "loss": 2.6902, "step": 27528 }, { "epoch": 1.7089204792352102, "grad_norm": 0.6578933484916092, "learning_rate": 4.657534690085942e-05, "loss": 2.7886, "step": 27529 }, { "epoch": 1.708982556334968, "grad_norm": 0.5675255600806978, "learning_rate": 4.6571743866945964e-05, "loss": 2.7712, "step": 27530 }, { "epoch": 1.709044633434726, "grad_norm": 0.5624694286092041, "learning_rate": 4.656814085091857e-05, "loss": 2.7313, "step": 27531 }, { "epoch": 1.709106710534484, "grad_norm": 0.5887489263358724, "learning_rate": 4.6564537852795996e-05, "loss": 2.7224, "step": 27532 }, { "epoch": 1.7091687876342418, "grad_norm": 0.5556588063720541, "learning_rate": 4.656093487259708e-05, "loss": 2.7787, "step": 27533 }, { "epoch": 1.7092308647339998, "grad_norm": 0.629031648632611, "learning_rate": 4.655733191034059e-05, "loss": 2.7784, "step": 27534 }, { "epoch": 1.7092929418337577, "grad_norm": 0.6187040793593145, "learning_rate": 4.655372896604534e-05, "loss": 2.7702, "step": 27535 }, { "epoch": 1.7093550189335154, "grad_norm": 0.6089686726325239, "learning_rate": 4.655012603973013e-05, "loss": 2.7198, "step": 27536 }, { "epoch": 1.7094170960332733, "grad_norm": 0.618317185852563, "learning_rate": 4.654652313141374e-05, "loss": 2.6592, "step": 27537 }, { "epoch": 1.7094791731330312, "grad_norm": 0.5442095513652063, "learning_rate": 4.654292024111499e-05, "loss": 2.6394, "step": 27538 }, { "epoch": 1.7095412502327891, "grad_norm": 0.663066138981799, "learning_rate": 4.653931736885265e-05, "loss": 2.7449, "step": 27539 }, { "epoch": 1.709603327332547, "grad_norm": 0.5679095346259995, "learning_rate": 4.653571451464554e-05, "loss": 2.7945, "step": 27540 }, { "epoch": 1.709665404432305, "grad_norm": 0.6069565981174991, "learning_rate": 4.6532111678512434e-05, "loss": 2.8164, "step": 27541 }, { "epoch": 1.7097274815320627, "grad_norm": 0.5741059847805002, "learning_rate": 4.6528508860472164e-05, "loss": 2.8101, "step": 27542 }, { "epoch": 1.7097895586318206, "grad_norm": 0.6531739348920285, "learning_rate": 4.652490606054349e-05, "loss": 2.8006, "step": 27543 }, { "epoch": 1.7098516357315785, "grad_norm": 0.561382003884888, "learning_rate": 4.6521303278745234e-05, "loss": 2.7622, "step": 27544 }, { "epoch": 1.7099137128313364, "grad_norm": 0.6195325844210364, "learning_rate": 4.6517700515096175e-05, "loss": 2.7904, "step": 27545 }, { "epoch": 1.7099757899310943, "grad_norm": 0.5465967015186785, "learning_rate": 4.6514097769615126e-05, "loss": 2.7289, "step": 27546 }, { "epoch": 1.7100378670308523, "grad_norm": 0.5716593267935319, "learning_rate": 4.651049504232086e-05, "loss": 2.684, "step": 27547 }, { "epoch": 1.7100999441306102, "grad_norm": 0.49466395654036627, "learning_rate": 4.6506892333232186e-05, "loss": 2.6609, "step": 27548 }, { "epoch": 1.710162021230368, "grad_norm": 0.5414729750168932, "learning_rate": 4.650328964236792e-05, "loss": 2.7695, "step": 27549 }, { "epoch": 1.710224098330126, "grad_norm": 0.4939591296862754, "learning_rate": 4.6499686969746845e-05, "loss": 2.7211, "step": 27550 }, { "epoch": 1.710286175429884, "grad_norm": 0.5585046977304062, "learning_rate": 4.6496084315387744e-05, "loss": 2.7933, "step": 27551 }, { "epoch": 1.7103482525296418, "grad_norm": 0.5088708852040762, "learning_rate": 4.649248167930943e-05, "loss": 2.7934, "step": 27552 }, { "epoch": 1.7104103296293998, "grad_norm": 0.5304485614963141, "learning_rate": 4.648887906153068e-05, "loss": 2.803, "step": 27553 }, { "epoch": 1.7104724067291577, "grad_norm": 0.546666378010503, "learning_rate": 4.6485276462070304e-05, "loss": 2.7248, "step": 27554 }, { "epoch": 1.7105344838289156, "grad_norm": 0.5540275677611266, "learning_rate": 4.648167388094709e-05, "loss": 2.7242, "step": 27555 }, { "epoch": 1.7105965609286735, "grad_norm": 0.5710961692748379, "learning_rate": 4.647807131817984e-05, "loss": 2.6882, "step": 27556 }, { "epoch": 1.7106586380284314, "grad_norm": 0.5291559235791402, "learning_rate": 4.647446877378735e-05, "loss": 2.7494, "step": 27557 }, { "epoch": 1.7107207151281894, "grad_norm": 0.5656315734649916, "learning_rate": 4.647086624778841e-05, "loss": 2.6698, "step": 27558 }, { "epoch": 1.7107827922279473, "grad_norm": 0.5409497658205534, "learning_rate": 4.646726374020182e-05, "loss": 2.7095, "step": 27559 }, { "epoch": 1.710844869327705, "grad_norm": 0.520014717187224, "learning_rate": 4.646366125104637e-05, "loss": 2.6884, "step": 27560 }, { "epoch": 1.7109069464274629, "grad_norm": 0.6061173195520475, "learning_rate": 4.6460058780340864e-05, "loss": 2.8376, "step": 27561 }, { "epoch": 1.7109690235272208, "grad_norm": 0.5524252576458741, "learning_rate": 4.645645632810408e-05, "loss": 2.7592, "step": 27562 }, { "epoch": 1.7110311006269787, "grad_norm": 0.5486446087371311, "learning_rate": 4.645285389435484e-05, "loss": 2.7386, "step": 27563 }, { "epoch": 1.7110931777267366, "grad_norm": 0.5895173641362085, "learning_rate": 4.6449251479111896e-05, "loss": 2.7948, "step": 27564 }, { "epoch": 1.7111552548264946, "grad_norm": 0.5826794388060107, "learning_rate": 4.6445649082394095e-05, "loss": 2.767, "step": 27565 }, { "epoch": 1.7112173319262522, "grad_norm": 0.5753330655128648, "learning_rate": 4.644204670422019e-05, "loss": 2.7478, "step": 27566 }, { "epoch": 1.7112794090260102, "grad_norm": 0.6119290649250373, "learning_rate": 4.6438444344609e-05, "loss": 2.7646, "step": 27567 }, { "epoch": 1.711341486125768, "grad_norm": 0.584365006223687, "learning_rate": 4.6434842003579297e-05, "loss": 2.7527, "step": 27568 }, { "epoch": 1.711403563225526, "grad_norm": 0.631764945945972, "learning_rate": 4.643123968114991e-05, "loss": 2.7751, "step": 27569 }, { "epoch": 1.711465640325284, "grad_norm": 0.5728884227610997, "learning_rate": 4.642763737733959e-05, "loss": 2.8029, "step": 27570 }, { "epoch": 1.7115277174250418, "grad_norm": 0.6425376577882024, "learning_rate": 4.642403509216717e-05, "loss": 2.7964, "step": 27571 }, { "epoch": 1.7115897945247998, "grad_norm": 0.5888016697509428, "learning_rate": 4.6420432825651424e-05, "loss": 2.7472, "step": 27572 }, { "epoch": 1.7116518716245577, "grad_norm": 0.5889086605922463, "learning_rate": 4.6416830577811136e-05, "loss": 2.7086, "step": 27573 }, { "epoch": 1.7117139487243156, "grad_norm": 0.6273535522227203, "learning_rate": 4.641322834866513e-05, "loss": 2.6998, "step": 27574 }, { "epoch": 1.7117760258240735, "grad_norm": 0.6088655004651121, "learning_rate": 4.640962613823217e-05, "loss": 2.7786, "step": 27575 }, { "epoch": 1.7118381029238314, "grad_norm": 0.5984883966540113, "learning_rate": 4.640602394653107e-05, "loss": 2.6579, "step": 27576 }, { "epoch": 1.7119001800235893, "grad_norm": 0.5817535557197655, "learning_rate": 4.64024217735806e-05, "loss": 2.7563, "step": 27577 }, { "epoch": 1.7119622571233473, "grad_norm": 0.6200140198922116, "learning_rate": 4.639881961939959e-05, "loss": 2.8338, "step": 27578 }, { "epoch": 1.7120243342231052, "grad_norm": 0.6095846255747861, "learning_rate": 4.639521748400679e-05, "loss": 2.7239, "step": 27579 }, { "epoch": 1.712086411322863, "grad_norm": 0.6352298858112383, "learning_rate": 4.639161536742102e-05, "loss": 2.829, "step": 27580 }, { "epoch": 1.712148488422621, "grad_norm": 0.6006378647600493, "learning_rate": 4.6388013269661076e-05, "loss": 2.7805, "step": 27581 }, { "epoch": 1.712210565522379, "grad_norm": 0.615052817061547, "learning_rate": 4.638441119074574e-05, "loss": 2.7491, "step": 27582 }, { "epoch": 1.7122726426221369, "grad_norm": 0.6132256306496142, "learning_rate": 4.638080913069382e-05, "loss": 2.7498, "step": 27583 }, { "epoch": 1.7123347197218945, "grad_norm": 0.6273747152664928, "learning_rate": 4.6377207089524084e-05, "loss": 2.7435, "step": 27584 }, { "epoch": 1.7123967968216525, "grad_norm": 0.5914592724190576, "learning_rate": 4.637360506725535e-05, "loss": 2.8174, "step": 27585 }, { "epoch": 1.7124588739214104, "grad_norm": 0.6307501683622386, "learning_rate": 4.63700030639064e-05, "loss": 2.7004, "step": 27586 }, { "epoch": 1.7125209510211683, "grad_norm": 0.603079678507739, "learning_rate": 4.6366401079496014e-05, "loss": 2.7432, "step": 27587 }, { "epoch": 1.7125830281209262, "grad_norm": 0.6696491854232406, "learning_rate": 4.6362799114043005e-05, "loss": 2.812, "step": 27588 }, { "epoch": 1.7126451052206841, "grad_norm": 0.5845497736969483, "learning_rate": 4.635919716756614e-05, "loss": 2.6247, "step": 27589 }, { "epoch": 1.7127071823204418, "grad_norm": 0.6118899070630526, "learning_rate": 4.6355595240084245e-05, "loss": 2.7283, "step": 27590 }, { "epoch": 1.7127692594201998, "grad_norm": 0.5696086197825715, "learning_rate": 4.635199333161608e-05, "loss": 2.7763, "step": 27591 }, { "epoch": 1.7128313365199577, "grad_norm": 0.595681262005948, "learning_rate": 4.6348391442180465e-05, "loss": 2.7461, "step": 27592 }, { "epoch": 1.7128934136197156, "grad_norm": 0.5621550614994758, "learning_rate": 4.6344789571796164e-05, "loss": 2.7631, "step": 27593 }, { "epoch": 1.7129554907194735, "grad_norm": 0.6017327253590423, "learning_rate": 4.6341187720482e-05, "loss": 2.7357, "step": 27594 }, { "epoch": 1.7130175678192314, "grad_norm": 0.5526257735141051, "learning_rate": 4.6337585888256734e-05, "loss": 2.7314, "step": 27595 }, { "epoch": 1.7130796449189893, "grad_norm": 0.5488424876224132, "learning_rate": 4.633398407513918e-05, "loss": 2.781, "step": 27596 }, { "epoch": 1.7131417220187473, "grad_norm": 0.5787289017030447, "learning_rate": 4.633038228114812e-05, "loss": 2.7756, "step": 27597 }, { "epoch": 1.7132037991185052, "grad_norm": 0.5512597460761037, "learning_rate": 4.632678050630234e-05, "loss": 2.7517, "step": 27598 }, { "epoch": 1.713265876218263, "grad_norm": 0.5314987385681916, "learning_rate": 4.632317875062065e-05, "loss": 2.8318, "step": 27599 }, { "epoch": 1.713327953318021, "grad_norm": 0.5649271627755548, "learning_rate": 4.6319577014121815e-05, "loss": 2.6913, "step": 27600 }, { "epoch": 1.713390030417779, "grad_norm": 0.582497480411565, "learning_rate": 4.631597529682465e-05, "loss": 2.791, "step": 27601 }, { "epoch": 1.7134521075175368, "grad_norm": 0.5718724068337936, "learning_rate": 4.6312373598747926e-05, "loss": 2.7364, "step": 27602 }, { "epoch": 1.7135141846172948, "grad_norm": 0.5381813034039339, "learning_rate": 4.630877191991046e-05, "loss": 2.7786, "step": 27603 }, { "epoch": 1.7135762617170527, "grad_norm": 0.5336924672443422, "learning_rate": 4.630517026033101e-05, "loss": 2.7537, "step": 27604 }, { "epoch": 1.7136383388168106, "grad_norm": 0.6130865237405483, "learning_rate": 4.63015686200284e-05, "loss": 2.7123, "step": 27605 }, { "epoch": 1.7137004159165685, "grad_norm": 0.5520545904006874, "learning_rate": 4.6297966999021385e-05, "loss": 2.675, "step": 27606 }, { "epoch": 1.7137624930163264, "grad_norm": 0.5918986166942226, "learning_rate": 4.6294365397328796e-05, "loss": 2.7975, "step": 27607 }, { "epoch": 1.7138245701160841, "grad_norm": 0.5962798031969873, "learning_rate": 4.629076381496938e-05, "loss": 2.8048, "step": 27608 }, { "epoch": 1.713886647215842, "grad_norm": 0.4601977765139016, "learning_rate": 4.628716225196197e-05, "loss": 2.6734, "step": 27609 }, { "epoch": 1.7139487243156, "grad_norm": 0.5355510230150692, "learning_rate": 4.628356070832533e-05, "loss": 2.7477, "step": 27610 }, { "epoch": 1.7140108014153579, "grad_norm": 0.5201391521289648, "learning_rate": 4.627995918407825e-05, "loss": 2.7115, "step": 27611 }, { "epoch": 1.7140728785151158, "grad_norm": 0.49272653036675873, "learning_rate": 4.6276357679239514e-05, "loss": 2.7451, "step": 27612 }, { "epoch": 1.7141349556148737, "grad_norm": 0.5625692695775728, "learning_rate": 4.627275619382794e-05, "loss": 2.7679, "step": 27613 }, { "epoch": 1.7141970327146314, "grad_norm": 0.5306656372361446, "learning_rate": 4.62691547278623e-05, "loss": 2.6812, "step": 27614 }, { "epoch": 1.7142591098143893, "grad_norm": 0.5204402743783599, "learning_rate": 4.6265553281361385e-05, "loss": 2.7694, "step": 27615 }, { "epoch": 1.7143211869141473, "grad_norm": 0.5399010259383619, "learning_rate": 4.6261951854343985e-05, "loss": 2.7131, "step": 27616 }, { "epoch": 1.7143832640139052, "grad_norm": 0.49042829348519085, "learning_rate": 4.6258350446828895e-05, "loss": 2.6981, "step": 27617 }, { "epoch": 1.714445341113663, "grad_norm": 0.5087276721900473, "learning_rate": 4.625474905883489e-05, "loss": 2.851, "step": 27618 }, { "epoch": 1.714507418213421, "grad_norm": 0.555791079413078, "learning_rate": 4.6251147690380774e-05, "loss": 2.6952, "step": 27619 }, { "epoch": 1.714569495313179, "grad_norm": 0.49416756839198184, "learning_rate": 4.6247546341485324e-05, "loss": 2.8226, "step": 27620 }, { "epoch": 1.7146315724129368, "grad_norm": 0.5744171059484705, "learning_rate": 4.624394501216735e-05, "loss": 2.7572, "step": 27621 }, { "epoch": 1.7146936495126948, "grad_norm": 0.5611239131509916, "learning_rate": 4.624034370244562e-05, "loss": 2.7401, "step": 27622 }, { "epoch": 1.7147557266124527, "grad_norm": 0.5364579012477085, "learning_rate": 4.6236742412338924e-05, "loss": 2.8142, "step": 27623 }, { "epoch": 1.7148178037122106, "grad_norm": 0.5045983935606881, "learning_rate": 4.623314114186607e-05, "loss": 2.6553, "step": 27624 }, { "epoch": 1.7148798808119685, "grad_norm": 0.5339338723824909, "learning_rate": 4.622953989104581e-05, "loss": 2.7701, "step": 27625 }, { "epoch": 1.7149419579117264, "grad_norm": 0.5048941457867253, "learning_rate": 4.622593865989698e-05, "loss": 2.7664, "step": 27626 }, { "epoch": 1.7150040350114844, "grad_norm": 0.543671544213735, "learning_rate": 4.6222337448438325e-05, "loss": 2.7939, "step": 27627 }, { "epoch": 1.7150661121112423, "grad_norm": 0.5098883528610421, "learning_rate": 4.621873625668867e-05, "loss": 2.6899, "step": 27628 }, { "epoch": 1.7151281892110002, "grad_norm": 0.6189858276165258, "learning_rate": 4.621513508466677e-05, "loss": 2.8292, "step": 27629 }, { "epoch": 1.715190266310758, "grad_norm": 0.5166550952887168, "learning_rate": 4.6211533932391446e-05, "loss": 2.8107, "step": 27630 }, { "epoch": 1.715252343410516, "grad_norm": 0.6053996631214955, "learning_rate": 4.620793279988145e-05, "loss": 2.6931, "step": 27631 }, { "epoch": 1.7153144205102737, "grad_norm": 0.5129230102634338, "learning_rate": 4.62043316871556e-05, "loss": 2.7991, "step": 27632 }, { "epoch": 1.7153764976100316, "grad_norm": 0.5936890872026915, "learning_rate": 4.6200730594232666e-05, "loss": 2.7389, "step": 27633 }, { "epoch": 1.7154385747097896, "grad_norm": 0.5442693214456346, "learning_rate": 4.6197129521131455e-05, "loss": 2.753, "step": 27634 }, { "epoch": 1.7155006518095475, "grad_norm": 0.6629506845415614, "learning_rate": 4.619352846787074e-05, "loss": 2.8106, "step": 27635 }, { "epoch": 1.7155627289093054, "grad_norm": 0.5589325417639196, "learning_rate": 4.61899274344693e-05, "loss": 2.7043, "step": 27636 }, { "epoch": 1.7156248060090633, "grad_norm": 0.6098451023736766, "learning_rate": 4.6186326420945945e-05, "loss": 2.5954, "step": 27637 }, { "epoch": 1.715686883108821, "grad_norm": 0.5697684894104268, "learning_rate": 4.618272542731944e-05, "loss": 2.739, "step": 27638 }, { "epoch": 1.715748960208579, "grad_norm": 0.562965994771779, "learning_rate": 4.61791244536086e-05, "loss": 2.7552, "step": 27639 }, { "epoch": 1.7158110373083368, "grad_norm": 0.5928618916560995, "learning_rate": 4.617552349983217e-05, "loss": 2.8361, "step": 27640 }, { "epoch": 1.7158731144080948, "grad_norm": 0.607074503296585, "learning_rate": 4.617192256600898e-05, "loss": 2.6684, "step": 27641 }, { "epoch": 1.7159351915078527, "grad_norm": 0.5459643001780088, "learning_rate": 4.61683216521578e-05, "loss": 2.6991, "step": 27642 }, { "epoch": 1.7159972686076106, "grad_norm": 0.5857002795352435, "learning_rate": 4.616472075829741e-05, "loss": 2.7895, "step": 27643 }, { "epoch": 1.7160593457073685, "grad_norm": 0.559324973016434, "learning_rate": 4.61611198844466e-05, "loss": 2.8445, "step": 27644 }, { "epoch": 1.7161214228071264, "grad_norm": 0.5934273696524903, "learning_rate": 4.615751903062415e-05, "loss": 2.726, "step": 27645 }, { "epoch": 1.7161834999068843, "grad_norm": 0.5883157150081825, "learning_rate": 4.615391819684888e-05, "loss": 2.7911, "step": 27646 }, { "epoch": 1.7162455770066423, "grad_norm": 0.5719987162195485, "learning_rate": 4.615031738313954e-05, "loss": 2.7322, "step": 27647 }, { "epoch": 1.7163076541064002, "grad_norm": 0.647762843840574, "learning_rate": 4.614671658951493e-05, "loss": 2.8088, "step": 27648 }, { "epoch": 1.716369731206158, "grad_norm": 0.5737237938025938, "learning_rate": 4.614311581599384e-05, "loss": 2.8358, "step": 27649 }, { "epoch": 1.716431808305916, "grad_norm": 0.609391994728401, "learning_rate": 4.613951506259504e-05, "loss": 2.7619, "step": 27650 }, { "epoch": 1.716493885405674, "grad_norm": 0.5639238879372472, "learning_rate": 4.6135914329337345e-05, "loss": 2.6955, "step": 27651 }, { "epoch": 1.7165559625054319, "grad_norm": 0.6329317804641083, "learning_rate": 4.61323136162395e-05, "loss": 2.7344, "step": 27652 }, { "epoch": 1.7166180396051898, "grad_norm": 0.5999530465907141, "learning_rate": 4.6128712923320337e-05, "loss": 2.819, "step": 27653 }, { "epoch": 1.7166801167049477, "grad_norm": 0.6177610798316455, "learning_rate": 4.61251122505986e-05, "loss": 2.7503, "step": 27654 }, { "epoch": 1.7167421938047056, "grad_norm": 0.6093302979334568, "learning_rate": 4.612151159809311e-05, "loss": 2.7043, "step": 27655 }, { "epoch": 1.7168042709044633, "grad_norm": 0.5730817268034362, "learning_rate": 4.611791096582262e-05, "loss": 2.6688, "step": 27656 }, { "epoch": 1.7168663480042212, "grad_norm": 0.6051883171128817, "learning_rate": 4.611431035380594e-05, "loss": 2.8153, "step": 27657 }, { "epoch": 1.7169284251039791, "grad_norm": 0.6691320005663878, "learning_rate": 4.611070976206184e-05, "loss": 2.7011, "step": 27658 }, { "epoch": 1.716990502203737, "grad_norm": 0.6606479371202281, "learning_rate": 4.610710919060912e-05, "loss": 2.7815, "step": 27659 }, { "epoch": 1.717052579303495, "grad_norm": 0.6143940105970208, "learning_rate": 4.610350863946655e-05, "loss": 2.7573, "step": 27660 }, { "epoch": 1.717114656403253, "grad_norm": 0.5972494598456487, "learning_rate": 4.6099908108652925e-05, "loss": 2.7868, "step": 27661 }, { "epoch": 1.7171767335030106, "grad_norm": 0.6269013332947879, "learning_rate": 4.609630759818703e-05, "loss": 2.8017, "step": 27662 }, { "epoch": 1.7172388106027685, "grad_norm": 0.5878418049866578, "learning_rate": 4.609270710808763e-05, "loss": 2.7495, "step": 27663 }, { "epoch": 1.7173008877025264, "grad_norm": 0.6387826199423322, "learning_rate": 4.608910663837354e-05, "loss": 2.7882, "step": 27664 }, { "epoch": 1.7173629648022843, "grad_norm": 0.6263443793107276, "learning_rate": 4.608550618906352e-05, "loss": 2.7473, "step": 27665 }, { "epoch": 1.7174250419020423, "grad_norm": 0.5817132631627668, "learning_rate": 4.6081905760176377e-05, "loss": 2.7142, "step": 27666 }, { "epoch": 1.7174871190018002, "grad_norm": 0.6044490225668615, "learning_rate": 4.607830535173087e-05, "loss": 2.6975, "step": 27667 }, { "epoch": 1.717549196101558, "grad_norm": 0.616717751315893, "learning_rate": 4.60747049637458e-05, "loss": 2.8061, "step": 27668 }, { "epoch": 1.717611273201316, "grad_norm": 0.5892539707320333, "learning_rate": 4.607110459623995e-05, "loss": 2.7453, "step": 27669 }, { "epoch": 1.717673350301074, "grad_norm": 0.5956385092789402, "learning_rate": 4.60675042492321e-05, "loss": 2.7328, "step": 27670 }, { "epoch": 1.7177354274008318, "grad_norm": 0.5393077734628352, "learning_rate": 4.606390392274104e-05, "loss": 2.7007, "step": 27671 }, { "epoch": 1.7177975045005898, "grad_norm": 0.6074342143777189, "learning_rate": 4.606030361678554e-05, "loss": 2.7284, "step": 27672 }, { "epoch": 1.7178595816003477, "grad_norm": 0.5676465102897663, "learning_rate": 4.60567033313844e-05, "loss": 2.7818, "step": 27673 }, { "epoch": 1.7179216587001056, "grad_norm": 0.6016858705122325, "learning_rate": 4.605310306655638e-05, "loss": 2.7916, "step": 27674 }, { "epoch": 1.7179837357998635, "grad_norm": 0.5823768255463477, "learning_rate": 4.60495028223203e-05, "loss": 2.7879, "step": 27675 }, { "epoch": 1.7180458128996214, "grad_norm": 0.5785821705111965, "learning_rate": 4.604590259869491e-05, "loss": 2.7191, "step": 27676 }, { "epoch": 1.7181078899993794, "grad_norm": 0.5851894690095368, "learning_rate": 4.6042302395699014e-05, "loss": 2.7794, "step": 27677 }, { "epoch": 1.7181699670991373, "grad_norm": 0.5602712717967228, "learning_rate": 4.603870221335137e-05, "loss": 2.7081, "step": 27678 }, { "epoch": 1.7182320441988952, "grad_norm": 0.6122011526083853, "learning_rate": 4.6035102051670786e-05, "loss": 2.641, "step": 27679 }, { "epoch": 1.7182941212986529, "grad_norm": 0.5547361486888105, "learning_rate": 4.6031501910676046e-05, "loss": 2.7392, "step": 27680 }, { "epoch": 1.7183561983984108, "grad_norm": 0.5731110560926753, "learning_rate": 4.602790179038592e-05, "loss": 2.7034, "step": 27681 }, { "epoch": 1.7184182754981687, "grad_norm": 0.5949764885384367, "learning_rate": 4.6024301690819196e-05, "loss": 2.7417, "step": 27682 }, { "epoch": 1.7184803525979266, "grad_norm": 0.571965943072879, "learning_rate": 4.602070161199465e-05, "loss": 2.8244, "step": 27683 }, { "epoch": 1.7185424296976846, "grad_norm": 0.5921469933739899, "learning_rate": 4.601710155393108e-05, "loss": 2.7652, "step": 27684 }, { "epoch": 1.7186045067974425, "grad_norm": 0.5594394923705148, "learning_rate": 4.601350151664726e-05, "loss": 2.8225, "step": 27685 }, { "epoch": 1.7186665838972002, "grad_norm": 0.49773022926722443, "learning_rate": 4.6009901500161955e-05, "loss": 2.828, "step": 27686 }, { "epoch": 1.718728660996958, "grad_norm": 0.5877626602970604, "learning_rate": 4.600630150449398e-05, "loss": 2.7821, "step": 27687 }, { "epoch": 1.718790738096716, "grad_norm": 0.5522112308361801, "learning_rate": 4.600270152966208e-05, "loss": 2.7938, "step": 27688 }, { "epoch": 1.718852815196474, "grad_norm": 0.5922754952066563, "learning_rate": 4.599910157568507e-05, "loss": 2.6811, "step": 27689 }, { "epoch": 1.7189148922962318, "grad_norm": 0.5658616440319477, "learning_rate": 4.5995501642581704e-05, "loss": 2.8057, "step": 27690 }, { "epoch": 1.7189769693959898, "grad_norm": 0.5492473372376201, "learning_rate": 4.599190173037079e-05, "loss": 2.8376, "step": 27691 }, { "epoch": 1.7190390464957477, "grad_norm": 0.6453087942834448, "learning_rate": 4.59883018390711e-05, "loss": 2.8404, "step": 27692 }, { "epoch": 1.7191011235955056, "grad_norm": 0.5295203324917765, "learning_rate": 4.5984701968701413e-05, "loss": 2.6833, "step": 27693 }, { "epoch": 1.7191632006952635, "grad_norm": 0.5350298767125693, "learning_rate": 4.59811021192805e-05, "loss": 2.7681, "step": 27694 }, { "epoch": 1.7192252777950214, "grad_norm": 0.6517624920890519, "learning_rate": 4.597750229082717e-05, "loss": 2.718, "step": 27695 }, { "epoch": 1.7192873548947794, "grad_norm": 0.5333908924196316, "learning_rate": 4.5973902483360174e-05, "loss": 2.6808, "step": 27696 }, { "epoch": 1.7193494319945373, "grad_norm": 0.6042955414572996, "learning_rate": 4.5970302696898304e-05, "loss": 2.7068, "step": 27697 }, { "epoch": 1.7194115090942952, "grad_norm": 0.5538383754363334, "learning_rate": 4.5966702931460354e-05, "loss": 2.7544, "step": 27698 }, { "epoch": 1.719473586194053, "grad_norm": 0.5504968869556962, "learning_rate": 4.596310318706508e-05, "loss": 2.7409, "step": 27699 }, { "epoch": 1.719535663293811, "grad_norm": 0.5849519451895823, "learning_rate": 4.595950346373129e-05, "loss": 2.7699, "step": 27700 }, { "epoch": 1.719597740393569, "grad_norm": 0.5307744085203933, "learning_rate": 4.595590376147774e-05, "loss": 2.7483, "step": 27701 }, { "epoch": 1.7196598174933269, "grad_norm": 0.5258199079565979, "learning_rate": 4.595230408032324e-05, "loss": 2.7541, "step": 27702 }, { "epoch": 1.7197218945930848, "grad_norm": 0.5656218523391057, "learning_rate": 4.5948704420286534e-05, "loss": 2.6635, "step": 27703 }, { "epoch": 1.7197839716928425, "grad_norm": 0.5312135908328596, "learning_rate": 4.594510478138643e-05, "loss": 2.8037, "step": 27704 }, { "epoch": 1.7198460487926004, "grad_norm": 0.5241079449270335, "learning_rate": 4.5941505163641684e-05, "loss": 2.7741, "step": 27705 }, { "epoch": 1.7199081258923583, "grad_norm": 0.6204769502672829, "learning_rate": 4.593790556707111e-05, "loss": 2.8389, "step": 27706 }, { "epoch": 1.7199702029921162, "grad_norm": 0.572513747862283, "learning_rate": 4.593430599169347e-05, "loss": 2.7174, "step": 27707 }, { "epoch": 1.7200322800918741, "grad_norm": 0.558906147705956, "learning_rate": 4.5930706437527524e-05, "loss": 2.8261, "step": 27708 }, { "epoch": 1.720094357191632, "grad_norm": 0.5629367192685318, "learning_rate": 4.592710690459209e-05, "loss": 2.6995, "step": 27709 }, { "epoch": 1.7201564342913898, "grad_norm": 0.6044906345033186, "learning_rate": 4.5923507392905896e-05, "loss": 2.6937, "step": 27710 }, { "epoch": 1.7202185113911477, "grad_norm": 0.6065536623374328, "learning_rate": 4.591990790248777e-05, "loss": 2.8405, "step": 27711 }, { "epoch": 1.7202805884909056, "grad_norm": 0.5330525000358016, "learning_rate": 4.5916308433356484e-05, "loss": 2.7759, "step": 27712 }, { "epoch": 1.7203426655906635, "grad_norm": 0.6713843883127169, "learning_rate": 4.59127089855308e-05, "loss": 2.7932, "step": 27713 }, { "epoch": 1.7204047426904214, "grad_norm": 0.5414758920947618, "learning_rate": 4.5909109559029516e-05, "loss": 2.7561, "step": 27714 }, { "epoch": 1.7204668197901793, "grad_norm": 0.6257696657858763, "learning_rate": 4.590551015387139e-05, "loss": 2.6847, "step": 27715 }, { "epoch": 1.7205288968899373, "grad_norm": 0.5751256981579913, "learning_rate": 4.5901910770075216e-05, "loss": 2.7357, "step": 27716 }, { "epoch": 1.7205909739896952, "grad_norm": 0.5741210646806953, "learning_rate": 4.5898311407659763e-05, "loss": 2.8011, "step": 27717 }, { "epoch": 1.720653051089453, "grad_norm": 0.58452629537963, "learning_rate": 4.589471206664383e-05, "loss": 2.7748, "step": 27718 }, { "epoch": 1.720715128189211, "grad_norm": 0.6168271956622438, "learning_rate": 4.589111274704616e-05, "loss": 2.7575, "step": 27719 }, { "epoch": 1.720777205288969, "grad_norm": 0.5948016740944705, "learning_rate": 4.588751344888557e-05, "loss": 2.7702, "step": 27720 }, { "epoch": 1.7208392823887269, "grad_norm": 0.5563073214452879, "learning_rate": 4.588391417218082e-05, "loss": 2.8108, "step": 27721 }, { "epoch": 1.7209013594884848, "grad_norm": 0.5707149392118622, "learning_rate": 4.588031491695068e-05, "loss": 2.6962, "step": 27722 }, { "epoch": 1.7209634365882427, "grad_norm": 0.507027033243146, "learning_rate": 4.5876715683213945e-05, "loss": 2.6809, "step": 27723 }, { "epoch": 1.7210255136880006, "grad_norm": 0.6304013742137226, "learning_rate": 4.5873116470989374e-05, "loss": 2.7694, "step": 27724 }, { "epoch": 1.7210875907877585, "grad_norm": 0.5485085748334314, "learning_rate": 4.586951728029577e-05, "loss": 2.865, "step": 27725 }, { "epoch": 1.7211496678875164, "grad_norm": 0.5731344300454198, "learning_rate": 4.5865918111151885e-05, "loss": 2.8116, "step": 27726 }, { "epoch": 1.7212117449872741, "grad_norm": 0.5172149535668964, "learning_rate": 4.5862318963576514e-05, "loss": 2.7496, "step": 27727 }, { "epoch": 1.721273822087032, "grad_norm": 0.6330230268505568, "learning_rate": 4.585871983758843e-05, "loss": 2.815, "step": 27728 }, { "epoch": 1.72133589918679, "grad_norm": 0.5840441057667223, "learning_rate": 4.5855120733206406e-05, "loss": 2.6922, "step": 27729 }, { "epoch": 1.721397976286548, "grad_norm": 0.6079797176410046, "learning_rate": 4.585152165044923e-05, "loss": 2.8057, "step": 27730 }, { "epoch": 1.7214600533863058, "grad_norm": 0.5936248296005252, "learning_rate": 4.584792258933567e-05, "loss": 2.6929, "step": 27731 }, { "epoch": 1.7215221304860637, "grad_norm": 0.533332473659816, "learning_rate": 4.5844323549884506e-05, "loss": 2.716, "step": 27732 }, { "epoch": 1.7215842075858214, "grad_norm": 0.6618432379387893, "learning_rate": 4.584072453211451e-05, "loss": 2.7448, "step": 27733 }, { "epoch": 1.7216462846855793, "grad_norm": 0.5585411687258481, "learning_rate": 4.583712553604447e-05, "loss": 2.6904, "step": 27734 }, { "epoch": 1.7217083617853373, "grad_norm": 0.5994155520821188, "learning_rate": 4.5833526561693153e-05, "loss": 2.773, "step": 27735 }, { "epoch": 1.7217704388850952, "grad_norm": 0.5986334099074303, "learning_rate": 4.582992760907934e-05, "loss": 2.7262, "step": 27736 }, { "epoch": 1.721832515984853, "grad_norm": 0.584149238509655, "learning_rate": 4.5826328678221804e-05, "loss": 2.7871, "step": 27737 }, { "epoch": 1.721894593084611, "grad_norm": 0.6548311699523816, "learning_rate": 4.5822729769139333e-05, "loss": 2.8745, "step": 27738 }, { "epoch": 1.721956670184369, "grad_norm": 0.5559463990449698, "learning_rate": 4.581913088185069e-05, "loss": 2.7474, "step": 27739 }, { "epoch": 1.7220187472841268, "grad_norm": 0.6167178238020196, "learning_rate": 4.5815532016374654e-05, "loss": 2.7616, "step": 27740 }, { "epoch": 1.7220808243838848, "grad_norm": 0.5858387927519509, "learning_rate": 4.581193317273e-05, "loss": 2.8003, "step": 27741 }, { "epoch": 1.7221429014836427, "grad_norm": 0.6767280649139944, "learning_rate": 4.5808334350935514e-05, "loss": 2.7964, "step": 27742 }, { "epoch": 1.7222049785834006, "grad_norm": 0.5887823091133698, "learning_rate": 4.5804735551009956e-05, "loss": 2.7807, "step": 27743 }, { "epoch": 1.7222670556831585, "grad_norm": 0.5938162264781766, "learning_rate": 4.580113677297211e-05, "loss": 2.813, "step": 27744 }, { "epoch": 1.7223291327829164, "grad_norm": 0.6266892492844937, "learning_rate": 4.5797538016840764e-05, "loss": 2.7739, "step": 27745 }, { "epoch": 1.7223912098826744, "grad_norm": 0.5745608809162102, "learning_rate": 4.579393928263468e-05, "loss": 2.6675, "step": 27746 }, { "epoch": 1.7224532869824323, "grad_norm": 0.6189398214301876, "learning_rate": 4.579034057037264e-05, "loss": 2.7516, "step": 27747 }, { "epoch": 1.7225153640821902, "grad_norm": 0.6417187448728685, "learning_rate": 4.578674188007341e-05, "loss": 2.7539, "step": 27748 }, { "epoch": 1.722577441181948, "grad_norm": 0.6302332886307842, "learning_rate": 4.578314321175577e-05, "loss": 2.8279, "step": 27749 }, { "epoch": 1.722639518281706, "grad_norm": 0.5718995313125214, "learning_rate": 4.5779544565438495e-05, "loss": 2.8042, "step": 27750 }, { "epoch": 1.7227015953814637, "grad_norm": 0.6317310983270419, "learning_rate": 4.577594594114036e-05, "loss": 2.6899, "step": 27751 }, { "epoch": 1.7227636724812216, "grad_norm": 0.5319760354137074, "learning_rate": 4.577234733888015e-05, "loss": 2.7788, "step": 27752 }, { "epoch": 1.7228257495809796, "grad_norm": 0.5542438505913024, "learning_rate": 4.5768748758676615e-05, "loss": 2.6492, "step": 27753 }, { "epoch": 1.7228878266807375, "grad_norm": 0.5893451528586606, "learning_rate": 4.576515020054856e-05, "loss": 2.693, "step": 27754 }, { "epoch": 1.7229499037804954, "grad_norm": 0.4936002256687752, "learning_rate": 4.5761551664514727e-05, "loss": 2.77, "step": 27755 }, { "epoch": 1.7230119808802533, "grad_norm": 0.5571873172025867, "learning_rate": 4.575795315059393e-05, "loss": 2.7322, "step": 27756 }, { "epoch": 1.723074057980011, "grad_norm": 0.5471786201144314, "learning_rate": 4.575435465880491e-05, "loss": 2.7803, "step": 27757 }, { "epoch": 1.723136135079769, "grad_norm": 0.5550774291730113, "learning_rate": 4.575075618916645e-05, "loss": 2.7489, "step": 27758 }, { "epoch": 1.7231982121795268, "grad_norm": 0.5632141136762161, "learning_rate": 4.574715774169733e-05, "loss": 2.7989, "step": 27759 }, { "epoch": 1.7232602892792848, "grad_norm": 0.5639303353721306, "learning_rate": 4.5743559316416315e-05, "loss": 2.7494, "step": 27760 }, { "epoch": 1.7233223663790427, "grad_norm": 0.5929222108774297, "learning_rate": 4.573996091334219e-05, "loss": 2.7376, "step": 27761 }, { "epoch": 1.7233844434788006, "grad_norm": 0.610426972880497, "learning_rate": 4.573636253249372e-05, "loss": 2.7784, "step": 27762 }, { "epoch": 1.7234465205785585, "grad_norm": 0.5812192539574088, "learning_rate": 4.573276417388969e-05, "loss": 2.7641, "step": 27763 }, { "epoch": 1.7235085976783164, "grad_norm": 0.6257679512811082, "learning_rate": 4.572916583754885e-05, "loss": 2.7239, "step": 27764 }, { "epoch": 1.7235706747780744, "grad_norm": 0.5318781000712651, "learning_rate": 4.572556752349e-05, "loss": 2.709, "step": 27765 }, { "epoch": 1.7236327518778323, "grad_norm": 0.5998628807345995, "learning_rate": 4.57219692317319e-05, "loss": 2.6901, "step": 27766 }, { "epoch": 1.7236948289775902, "grad_norm": 0.5814331873512679, "learning_rate": 4.5718370962293325e-05, "loss": 2.7467, "step": 27767 }, { "epoch": 1.723756906077348, "grad_norm": 0.6000970603002533, "learning_rate": 4.571477271519305e-05, "loss": 2.693, "step": 27768 }, { "epoch": 1.723818983177106, "grad_norm": 0.5867975915999157, "learning_rate": 4.5711174490449836e-05, "loss": 2.8061, "step": 27769 }, { "epoch": 1.723881060276864, "grad_norm": 0.6459154466500947, "learning_rate": 4.570757628808248e-05, "loss": 2.8115, "step": 27770 }, { "epoch": 1.7239431373766219, "grad_norm": 0.55111486733549, "learning_rate": 4.570397810810973e-05, "loss": 2.7028, "step": 27771 }, { "epoch": 1.7240052144763798, "grad_norm": 0.5757334706165311, "learning_rate": 4.570037995055038e-05, "loss": 2.6846, "step": 27772 }, { "epoch": 1.7240672915761377, "grad_norm": 0.5609853272516694, "learning_rate": 4.5696781815423176e-05, "loss": 2.7357, "step": 27773 }, { "epoch": 1.7241293686758956, "grad_norm": 0.6388496828028128, "learning_rate": 4.5693183702746914e-05, "loss": 2.7382, "step": 27774 }, { "epoch": 1.7241914457756533, "grad_norm": 0.5781710838302355, "learning_rate": 4.5689585612540356e-05, "loss": 2.799, "step": 27775 }, { "epoch": 1.7242535228754112, "grad_norm": 0.5971240497698939, "learning_rate": 4.5685987544822265e-05, "loss": 2.6655, "step": 27776 }, { "epoch": 1.7243155999751691, "grad_norm": 0.5950069424280291, "learning_rate": 4.568238949961144e-05, "loss": 2.709, "step": 27777 }, { "epoch": 1.724377677074927, "grad_norm": 0.6053453135027876, "learning_rate": 4.5678791476926627e-05, "loss": 2.7509, "step": 27778 }, { "epoch": 1.724439754174685, "grad_norm": 0.6166781242912837, "learning_rate": 4.567519347678662e-05, "loss": 2.83, "step": 27779 }, { "epoch": 1.724501831274443, "grad_norm": 0.5925307303318184, "learning_rate": 4.567159549921017e-05, "loss": 2.7624, "step": 27780 }, { "epoch": 1.7245639083742006, "grad_norm": 0.5912581153699795, "learning_rate": 4.566799754421607e-05, "loss": 2.7637, "step": 27781 }, { "epoch": 1.7246259854739585, "grad_norm": 0.6128417219852331, "learning_rate": 4.566439961182307e-05, "loss": 2.7878, "step": 27782 }, { "epoch": 1.7246880625737164, "grad_norm": 0.6035667076837165, "learning_rate": 4.566080170204994e-05, "loss": 2.6848, "step": 27783 }, { "epoch": 1.7247501396734743, "grad_norm": 0.5955038769305117, "learning_rate": 4.565720381491548e-05, "loss": 2.7213, "step": 27784 }, { "epoch": 1.7248122167732323, "grad_norm": 0.5659950342112153, "learning_rate": 4.5653605950438424e-05, "loss": 2.6977, "step": 27785 }, { "epoch": 1.7248742938729902, "grad_norm": 0.5735359284122776, "learning_rate": 4.565000810863758e-05, "loss": 2.7885, "step": 27786 }, { "epoch": 1.724936370972748, "grad_norm": 0.5188327482017815, "learning_rate": 4.5646410289531675e-05, "loss": 2.7363, "step": 27787 }, { "epoch": 1.724998448072506, "grad_norm": 0.5792117086455376, "learning_rate": 4.564281249313953e-05, "loss": 2.7154, "step": 27788 }, { "epoch": 1.725060525172264, "grad_norm": 0.5461952071566608, "learning_rate": 4.563921471947987e-05, "loss": 2.7604, "step": 27789 }, { "epoch": 1.7251226022720219, "grad_norm": 0.5519793040552621, "learning_rate": 4.56356169685715e-05, "loss": 2.7664, "step": 27790 }, { "epoch": 1.7251846793717798, "grad_norm": 0.5722311745126455, "learning_rate": 4.563201924043316e-05, "loss": 2.7929, "step": 27791 }, { "epoch": 1.7252467564715377, "grad_norm": 0.5801817162490603, "learning_rate": 4.562842153508365e-05, "loss": 2.7139, "step": 27792 }, { "epoch": 1.7253088335712956, "grad_norm": 0.5766563318498819, "learning_rate": 4.5624823852541725e-05, "loss": 2.7523, "step": 27793 }, { "epoch": 1.7253709106710535, "grad_norm": 0.573186946462244, "learning_rate": 4.562122619282615e-05, "loss": 2.8133, "step": 27794 }, { "epoch": 1.7254329877708114, "grad_norm": 0.5129728310940039, "learning_rate": 4.5617628555955704e-05, "loss": 2.6993, "step": 27795 }, { "epoch": 1.7254950648705694, "grad_norm": 0.6045966728267173, "learning_rate": 4.561403094194915e-05, "loss": 2.5904, "step": 27796 }, { "epoch": 1.7255571419703273, "grad_norm": 0.5402279018364589, "learning_rate": 4.561043335082527e-05, "loss": 2.7896, "step": 27797 }, { "epoch": 1.7256192190700852, "grad_norm": 0.5707146520057579, "learning_rate": 4.560683578260281e-05, "loss": 2.7329, "step": 27798 }, { "epoch": 1.725681296169843, "grad_norm": 0.575484638600689, "learning_rate": 4.560323823730057e-05, "loss": 2.7806, "step": 27799 }, { "epoch": 1.7257433732696008, "grad_norm": 0.5292494354947791, "learning_rate": 4.559964071493729e-05, "loss": 2.7814, "step": 27800 }, { "epoch": 1.7258054503693587, "grad_norm": 0.5821858655135697, "learning_rate": 4.559604321553176e-05, "loss": 2.8151, "step": 27801 }, { "epoch": 1.7258675274691166, "grad_norm": 0.5904128506020481, "learning_rate": 4.559244573910274e-05, "loss": 2.7838, "step": 27802 }, { "epoch": 1.7259296045688746, "grad_norm": 0.5919990571874766, "learning_rate": 4.558884828566901e-05, "loss": 2.7144, "step": 27803 }, { "epoch": 1.7259916816686325, "grad_norm": 0.5649570736571282, "learning_rate": 4.558525085524931e-05, "loss": 2.7272, "step": 27804 }, { "epoch": 1.7260537587683902, "grad_norm": 0.5937895761651643, "learning_rate": 4.5581653447862446e-05, "loss": 2.7616, "step": 27805 }, { "epoch": 1.726115835868148, "grad_norm": 0.523785774174548, "learning_rate": 4.557805606352717e-05, "loss": 2.7495, "step": 27806 }, { "epoch": 1.726177912967906, "grad_norm": 0.5927200921830301, "learning_rate": 4.557445870226224e-05, "loss": 2.7918, "step": 27807 }, { "epoch": 1.726239990067664, "grad_norm": 0.5691926491235227, "learning_rate": 4.5570861364086434e-05, "loss": 2.6764, "step": 27808 }, { "epoch": 1.7263020671674218, "grad_norm": 0.5990536876462123, "learning_rate": 4.5567264049018496e-05, "loss": 2.7599, "step": 27809 }, { "epoch": 1.7263641442671798, "grad_norm": 0.558886284688874, "learning_rate": 4.556366675707724e-05, "loss": 2.8013, "step": 27810 }, { "epoch": 1.7264262213669377, "grad_norm": 0.5901681879324358, "learning_rate": 4.556006948828141e-05, "loss": 2.8125, "step": 27811 }, { "epoch": 1.7264882984666956, "grad_norm": 0.5642081392315716, "learning_rate": 4.555647224264978e-05, "loss": 2.7593, "step": 27812 }, { "epoch": 1.7265503755664535, "grad_norm": 0.5743869612966522, "learning_rate": 4.555287502020111e-05, "loss": 2.8049, "step": 27813 }, { "epoch": 1.7266124526662114, "grad_norm": 0.5492075087372266, "learning_rate": 4.554927782095417e-05, "loss": 2.6593, "step": 27814 }, { "epoch": 1.7266745297659694, "grad_norm": 0.5619516348848881, "learning_rate": 4.5545680644927734e-05, "loss": 2.7162, "step": 27815 }, { "epoch": 1.7267366068657273, "grad_norm": 0.6003218899994809, "learning_rate": 4.5542083492140555e-05, "loss": 2.735, "step": 27816 }, { "epoch": 1.7267986839654852, "grad_norm": 0.6124326094708419, "learning_rate": 4.5538486362611416e-05, "loss": 2.6979, "step": 27817 }, { "epoch": 1.726860761065243, "grad_norm": 0.6024794587713548, "learning_rate": 4.5534889256359075e-05, "loss": 2.876, "step": 27818 }, { "epoch": 1.726922838165001, "grad_norm": 0.6337225457363768, "learning_rate": 4.5531292173402295e-05, "loss": 2.7079, "step": 27819 }, { "epoch": 1.726984915264759, "grad_norm": 0.6422731319239813, "learning_rate": 4.552769511375985e-05, "loss": 2.8073, "step": 27820 }, { "epoch": 1.7270469923645169, "grad_norm": 0.5638679982712363, "learning_rate": 4.5524098077450505e-05, "loss": 2.7244, "step": 27821 }, { "epoch": 1.7271090694642748, "grad_norm": 0.5751685507074827, "learning_rate": 4.5520501064493036e-05, "loss": 2.7591, "step": 27822 }, { "epoch": 1.7271711465640325, "grad_norm": 0.6166723974923882, "learning_rate": 4.551690407490619e-05, "loss": 2.8049, "step": 27823 }, { "epoch": 1.7272332236637904, "grad_norm": 0.611145552895261, "learning_rate": 4.551330710870875e-05, "loss": 2.8642, "step": 27824 }, { "epoch": 1.7272953007635483, "grad_norm": 0.6725158075155269, "learning_rate": 4.550971016591947e-05, "loss": 2.7747, "step": 27825 }, { "epoch": 1.7273573778633062, "grad_norm": 0.5721073647869696, "learning_rate": 4.550611324655713e-05, "loss": 2.721, "step": 27826 }, { "epoch": 1.7274194549630641, "grad_norm": 0.7055679479667953, "learning_rate": 4.550251635064048e-05, "loss": 2.7391, "step": 27827 }, { "epoch": 1.727481532062822, "grad_norm": 0.5407483649209285, "learning_rate": 4.54989194781883e-05, "loss": 2.7914, "step": 27828 }, { "epoch": 1.7275436091625798, "grad_norm": 0.678102201087326, "learning_rate": 4.5495322629219345e-05, "loss": 2.7352, "step": 27829 }, { "epoch": 1.7276056862623377, "grad_norm": 0.5713232828960338, "learning_rate": 4.549172580375239e-05, "loss": 2.8234, "step": 27830 }, { "epoch": 1.7276677633620956, "grad_norm": 0.6054924252711172, "learning_rate": 4.54881290018062e-05, "loss": 2.7837, "step": 27831 }, { "epoch": 1.7277298404618535, "grad_norm": 0.6321899271459747, "learning_rate": 4.548453222339952e-05, "loss": 2.8244, "step": 27832 }, { "epoch": 1.7277919175616114, "grad_norm": 0.5464190353368201, "learning_rate": 4.548093546855115e-05, "loss": 2.7107, "step": 27833 }, { "epoch": 1.7278539946613694, "grad_norm": 0.6598989899601716, "learning_rate": 4.547733873727982e-05, "loss": 2.7774, "step": 27834 }, { "epoch": 1.7279160717611273, "grad_norm": 0.5593354696019691, "learning_rate": 4.547374202960433e-05, "loss": 2.7755, "step": 27835 }, { "epoch": 1.7279781488608852, "grad_norm": 0.6647094492299046, "learning_rate": 4.5470145345543405e-05, "loss": 2.7214, "step": 27836 }, { "epoch": 1.728040225960643, "grad_norm": 0.6522216225705629, "learning_rate": 4.546654868511585e-05, "loss": 2.8325, "step": 27837 }, { "epoch": 1.728102303060401, "grad_norm": 0.580769547454118, "learning_rate": 4.5462952048340404e-05, "loss": 2.7778, "step": 27838 }, { "epoch": 1.728164380160159, "grad_norm": 0.6732188469470985, "learning_rate": 4.545935543523584e-05, "loss": 2.822, "step": 27839 }, { "epoch": 1.7282264572599169, "grad_norm": 0.6606230806306586, "learning_rate": 4.5455758845820915e-05, "loss": 2.6703, "step": 27840 }, { "epoch": 1.7282885343596748, "grad_norm": 0.6046947330958963, "learning_rate": 4.5452162280114386e-05, "loss": 2.7862, "step": 27841 }, { "epoch": 1.7283506114594327, "grad_norm": 0.594270578351469, "learning_rate": 4.544856573813505e-05, "loss": 2.7348, "step": 27842 }, { "epoch": 1.7284126885591906, "grad_norm": 0.6015930663430857, "learning_rate": 4.544496921990166e-05, "loss": 2.8116, "step": 27843 }, { "epoch": 1.7284747656589485, "grad_norm": 0.6089098451041812, "learning_rate": 4.544137272543295e-05, "loss": 2.808, "step": 27844 }, { "epoch": 1.7285368427587064, "grad_norm": 0.5679932821188828, "learning_rate": 4.543777625474773e-05, "loss": 2.6896, "step": 27845 }, { "epoch": 1.7285989198584644, "grad_norm": 0.6021259443478804, "learning_rate": 4.5434179807864716e-05, "loss": 2.7118, "step": 27846 }, { "epoch": 1.728660996958222, "grad_norm": 0.5887299183135924, "learning_rate": 4.54305833848027e-05, "loss": 2.781, "step": 27847 }, { "epoch": 1.72872307405798, "grad_norm": 0.6478260098496389, "learning_rate": 4.5426986985580445e-05, "loss": 2.7274, "step": 27848 }, { "epoch": 1.728785151157738, "grad_norm": 0.6270382309426882, "learning_rate": 4.5423390610216707e-05, "loss": 2.761, "step": 27849 }, { "epoch": 1.7288472282574958, "grad_norm": 0.5704651980888984, "learning_rate": 4.541979425873025e-05, "loss": 2.7342, "step": 27850 }, { "epoch": 1.7289093053572537, "grad_norm": 0.6884335335345815, "learning_rate": 4.541619793113985e-05, "loss": 2.7015, "step": 27851 }, { "epoch": 1.7289713824570117, "grad_norm": 0.5790742607178185, "learning_rate": 4.5412601627464235e-05, "loss": 2.7455, "step": 27852 }, { "epoch": 1.7290334595567693, "grad_norm": 0.6720024919421109, "learning_rate": 4.540900534772221e-05, "loss": 2.7857, "step": 27853 }, { "epoch": 1.7290955366565273, "grad_norm": 0.6433883938385558, "learning_rate": 4.5405409091932506e-05, "loss": 2.807, "step": 27854 }, { "epoch": 1.7291576137562852, "grad_norm": 0.5966188240538836, "learning_rate": 4.540181286011391e-05, "loss": 2.6752, "step": 27855 }, { "epoch": 1.729219690856043, "grad_norm": 0.6205539156723135, "learning_rate": 4.539821665228518e-05, "loss": 2.7508, "step": 27856 }, { "epoch": 1.729281767955801, "grad_norm": 0.5988115103072116, "learning_rate": 4.5394620468465045e-05, "loss": 2.7683, "step": 27857 }, { "epoch": 1.729343845055559, "grad_norm": 0.5661066775802958, "learning_rate": 4.5391024308672316e-05, "loss": 2.7442, "step": 27858 }, { "epoch": 1.7294059221553169, "grad_norm": 0.511617490344711, "learning_rate": 4.538742817292571e-05, "loss": 2.7391, "step": 27859 }, { "epoch": 1.7294679992550748, "grad_norm": 0.5443471797224287, "learning_rate": 4.538383206124403e-05, "loss": 2.7791, "step": 27860 }, { "epoch": 1.7295300763548327, "grad_norm": 0.5527338358199115, "learning_rate": 4.538023597364601e-05, "loss": 2.7473, "step": 27861 }, { "epoch": 1.7295921534545906, "grad_norm": 0.5301727143060353, "learning_rate": 4.537663991015042e-05, "loss": 2.8287, "step": 27862 }, { "epoch": 1.7296542305543485, "grad_norm": 0.5993683917861717, "learning_rate": 4.5373043870776025e-05, "loss": 2.8454, "step": 27863 }, { "epoch": 1.7297163076541064, "grad_norm": 0.5577451488962721, "learning_rate": 4.536944785554158e-05, "loss": 2.7475, "step": 27864 }, { "epoch": 1.7297783847538644, "grad_norm": 0.5692613137284814, "learning_rate": 4.5365851864465846e-05, "loss": 2.7119, "step": 27865 }, { "epoch": 1.7298404618536223, "grad_norm": 0.5340674524764002, "learning_rate": 4.53622558975676e-05, "loss": 2.7532, "step": 27866 }, { "epoch": 1.7299025389533802, "grad_norm": 0.5332764515743321, "learning_rate": 4.535865995486559e-05, "loss": 2.7676, "step": 27867 }, { "epoch": 1.7299646160531381, "grad_norm": 0.5331812983366735, "learning_rate": 4.535506403637857e-05, "loss": 2.7389, "step": 27868 }, { "epoch": 1.730026693152896, "grad_norm": 0.5543353077506762, "learning_rate": 4.535146814212531e-05, "loss": 2.6822, "step": 27869 }, { "epoch": 1.730088770252654, "grad_norm": 0.5567388615172768, "learning_rate": 4.534787227212457e-05, "loss": 2.6819, "step": 27870 }, { "epoch": 1.7301508473524116, "grad_norm": 0.5602218536309708, "learning_rate": 4.534427642639512e-05, "loss": 2.8936, "step": 27871 }, { "epoch": 1.7302129244521696, "grad_norm": 0.5392436400433286, "learning_rate": 4.534068060495569e-05, "loss": 2.7543, "step": 27872 }, { "epoch": 1.7302750015519275, "grad_norm": 0.5876423764689459, "learning_rate": 4.533708480782507e-05, "loss": 2.8697, "step": 27873 }, { "epoch": 1.7303370786516854, "grad_norm": 0.5381586541474958, "learning_rate": 4.5333489035022e-05, "loss": 2.7105, "step": 27874 }, { "epoch": 1.7303991557514433, "grad_norm": 0.5595312720721938, "learning_rate": 4.532989328656525e-05, "loss": 2.7983, "step": 27875 }, { "epoch": 1.7304612328512012, "grad_norm": 0.5582253902151197, "learning_rate": 4.5326297562473596e-05, "loss": 2.8098, "step": 27876 }, { "epoch": 1.730523309950959, "grad_norm": 0.5357141579900316, "learning_rate": 4.532270186276578e-05, "loss": 2.7875, "step": 27877 }, { "epoch": 1.7305853870507169, "grad_norm": 0.5476203471345074, "learning_rate": 4.531910618746056e-05, "loss": 2.7958, "step": 27878 }, { "epoch": 1.7306474641504748, "grad_norm": 0.5192497465953108, "learning_rate": 4.53155105365767e-05, "loss": 2.8122, "step": 27879 }, { "epoch": 1.7307095412502327, "grad_norm": 0.5622190727626071, "learning_rate": 4.5311914910132965e-05, "loss": 2.7955, "step": 27880 }, { "epoch": 1.7307716183499906, "grad_norm": 0.5574897045844173, "learning_rate": 4.5308319308148106e-05, "loss": 2.6648, "step": 27881 }, { "epoch": 1.7308336954497485, "grad_norm": 0.6357106612631832, "learning_rate": 4.530472373064087e-05, "loss": 2.7844, "step": 27882 }, { "epoch": 1.7308957725495064, "grad_norm": 0.5573454025838663, "learning_rate": 4.530112817763005e-05, "loss": 2.7581, "step": 27883 }, { "epoch": 1.7309578496492644, "grad_norm": 0.6086442429404628, "learning_rate": 4.529753264913437e-05, "loss": 2.7421, "step": 27884 }, { "epoch": 1.7310199267490223, "grad_norm": 0.5981431610714536, "learning_rate": 4.5293937145172616e-05, "loss": 2.6948, "step": 27885 }, { "epoch": 1.7310820038487802, "grad_norm": 0.523215289419505, "learning_rate": 4.529034166576352e-05, "loss": 2.7912, "step": 27886 }, { "epoch": 1.731144080948538, "grad_norm": 0.5861108396352749, "learning_rate": 4.528674621092587e-05, "loss": 2.7493, "step": 27887 }, { "epoch": 1.731206158048296, "grad_norm": 0.594220577280313, "learning_rate": 4.5283150780678396e-05, "loss": 2.6059, "step": 27888 }, { "epoch": 1.731268235148054, "grad_norm": 0.5660866986940328, "learning_rate": 4.5279555375039886e-05, "loss": 2.7661, "step": 27889 }, { "epoch": 1.7313303122478119, "grad_norm": 0.6439498128320099, "learning_rate": 4.527595999402906e-05, "loss": 2.7156, "step": 27890 }, { "epoch": 1.7313923893475698, "grad_norm": 0.5736574556683308, "learning_rate": 4.527236463766472e-05, "loss": 2.802, "step": 27891 }, { "epoch": 1.7314544664473277, "grad_norm": 0.6265917535740257, "learning_rate": 4.5268769305965585e-05, "loss": 2.7492, "step": 27892 }, { "epoch": 1.7315165435470856, "grad_norm": 0.5998902299922555, "learning_rate": 4.5265173998950435e-05, "loss": 2.7581, "step": 27893 }, { "epoch": 1.7315786206468435, "grad_norm": 0.6420800734092083, "learning_rate": 4.526157871663802e-05, "loss": 2.7228, "step": 27894 }, { "epoch": 1.7316406977466012, "grad_norm": 0.5678536173787317, "learning_rate": 4.52579834590471e-05, "loss": 2.8154, "step": 27895 }, { "epoch": 1.7317027748463591, "grad_norm": 0.6639967493155776, "learning_rate": 4.5254388226196434e-05, "loss": 2.7693, "step": 27896 }, { "epoch": 1.731764851946117, "grad_norm": 0.53165714706321, "learning_rate": 4.5250793018104765e-05, "loss": 2.7704, "step": 27897 }, { "epoch": 1.731826929045875, "grad_norm": 0.712946432359168, "learning_rate": 4.5247197834790876e-05, "loss": 2.8287, "step": 27898 }, { "epoch": 1.731889006145633, "grad_norm": 0.5725354544726371, "learning_rate": 4.52436026762735e-05, "loss": 2.7873, "step": 27899 }, { "epoch": 1.7319510832453908, "grad_norm": 0.6742800537740882, "learning_rate": 4.524000754257142e-05, "loss": 2.6969, "step": 27900 }, { "epoch": 1.7320131603451485, "grad_norm": 0.6300039248874051, "learning_rate": 4.523641243370335e-05, "loss": 2.6721, "step": 27901 }, { "epoch": 1.7320752374449064, "grad_norm": 0.5627499925553242, "learning_rate": 4.52328173496881e-05, "loss": 2.6922, "step": 27902 }, { "epoch": 1.7321373145446644, "grad_norm": 0.5991885668776017, "learning_rate": 4.522922229054439e-05, "loss": 2.7664, "step": 27903 }, { "epoch": 1.7321993916444223, "grad_norm": 0.5913977709304733, "learning_rate": 4.522562725629097e-05, "loss": 2.6744, "step": 27904 }, { "epoch": 1.7322614687441802, "grad_norm": 0.5408276393141969, "learning_rate": 4.5222032246946624e-05, "loss": 2.739, "step": 27905 }, { "epoch": 1.732323545843938, "grad_norm": 0.5670560595166513, "learning_rate": 4.521843726253009e-05, "loss": 2.7469, "step": 27906 }, { "epoch": 1.732385622943696, "grad_norm": 0.565329086499, "learning_rate": 4.521484230306012e-05, "loss": 2.7825, "step": 27907 }, { "epoch": 1.732447700043454, "grad_norm": 0.6231201272312901, "learning_rate": 4.52112473685555e-05, "loss": 2.8295, "step": 27908 }, { "epoch": 1.7325097771432119, "grad_norm": 0.5070207145078325, "learning_rate": 4.520765245903496e-05, "loss": 2.7828, "step": 27909 }, { "epoch": 1.7325718542429698, "grad_norm": 0.5829379552653521, "learning_rate": 4.520405757451725e-05, "loss": 2.7924, "step": 27910 }, { "epoch": 1.7326339313427277, "grad_norm": 0.5259540188983691, "learning_rate": 4.5200462715021146e-05, "loss": 2.6831, "step": 27911 }, { "epoch": 1.7326960084424856, "grad_norm": 0.5485457104385248, "learning_rate": 4.51968678805654e-05, "loss": 2.8503, "step": 27912 }, { "epoch": 1.7327580855422435, "grad_norm": 0.6268613991826036, "learning_rate": 4.5193273071168744e-05, "loss": 2.7342, "step": 27913 }, { "epoch": 1.7328201626420014, "grad_norm": 0.5580906236064261, "learning_rate": 4.518967828684996e-05, "loss": 2.8159, "step": 27914 }, { "epoch": 1.7328822397417594, "grad_norm": 0.592576674558289, "learning_rate": 4.5186083527627785e-05, "loss": 2.8099, "step": 27915 }, { "epoch": 1.7329443168415173, "grad_norm": 0.5588723401625136, "learning_rate": 4.5182488793520994e-05, "loss": 2.8136, "step": 27916 }, { "epoch": 1.7330063939412752, "grad_norm": 0.5816753502619315, "learning_rate": 4.517889408454832e-05, "loss": 2.7545, "step": 27917 }, { "epoch": 1.7330684710410331, "grad_norm": 0.5978387043021036, "learning_rate": 4.517529940072852e-05, "loss": 2.7756, "step": 27918 }, { "epoch": 1.7331305481407908, "grad_norm": 0.5952555491522438, "learning_rate": 4.5171704742080365e-05, "loss": 2.7102, "step": 27919 }, { "epoch": 1.7331926252405487, "grad_norm": 0.5792540273942088, "learning_rate": 4.516811010862259e-05, "loss": 2.7937, "step": 27920 }, { "epoch": 1.7332547023403067, "grad_norm": 0.6377262499374743, "learning_rate": 4.516451550037397e-05, "loss": 2.7019, "step": 27921 }, { "epoch": 1.7333167794400646, "grad_norm": 0.6017314128786481, "learning_rate": 4.5160920917353234e-05, "loss": 2.7574, "step": 27922 }, { "epoch": 1.7333788565398225, "grad_norm": 0.6198458581094782, "learning_rate": 4.5157326359579154e-05, "loss": 2.7988, "step": 27923 }, { "epoch": 1.7334409336395804, "grad_norm": 0.6027281231163477, "learning_rate": 4.5153731827070474e-05, "loss": 2.713, "step": 27924 }, { "epoch": 1.733503010739338, "grad_norm": 0.5905118531128425, "learning_rate": 4.515013731984595e-05, "loss": 2.8379, "step": 27925 }, { "epoch": 1.733565087839096, "grad_norm": 0.5591258469116156, "learning_rate": 4.514654283792434e-05, "loss": 2.6866, "step": 27926 }, { "epoch": 1.733627164938854, "grad_norm": 0.6801418671159655, "learning_rate": 4.51429483813244e-05, "loss": 2.8971, "step": 27927 }, { "epoch": 1.7336892420386119, "grad_norm": 0.5914462722277729, "learning_rate": 4.513935395006488e-05, "loss": 2.6809, "step": 27928 }, { "epoch": 1.7337513191383698, "grad_norm": 0.6414735338277335, "learning_rate": 4.5135759544164516e-05, "loss": 2.7466, "step": 27929 }, { "epoch": 1.7338133962381277, "grad_norm": 0.5506314012932603, "learning_rate": 4.513216516364209e-05, "loss": 2.7695, "step": 27930 }, { "epoch": 1.7338754733378856, "grad_norm": 0.6455354208531944, "learning_rate": 4.5128570808516324e-05, "loss": 2.8133, "step": 27931 }, { "epoch": 1.7339375504376435, "grad_norm": 0.5522811972931544, "learning_rate": 4.5124976478805994e-05, "loss": 2.7095, "step": 27932 }, { "epoch": 1.7339996275374014, "grad_norm": 0.5914365722865045, "learning_rate": 4.512138217452984e-05, "loss": 2.7286, "step": 27933 }, { "epoch": 1.7340617046371594, "grad_norm": 0.56794679224106, "learning_rate": 4.511778789570663e-05, "loss": 2.7477, "step": 27934 }, { "epoch": 1.7341237817369173, "grad_norm": 0.6166455750681745, "learning_rate": 4.5114193642355096e-05, "loss": 2.7496, "step": 27935 }, { "epoch": 1.7341858588366752, "grad_norm": 0.5304882534750819, "learning_rate": 4.511059941449401e-05, "loss": 2.8261, "step": 27936 }, { "epoch": 1.7342479359364331, "grad_norm": 0.6073750910267248, "learning_rate": 4.5107005212142106e-05, "loss": 2.6971, "step": 27937 }, { "epoch": 1.734310013036191, "grad_norm": 0.5660766958246924, "learning_rate": 4.510341103531815e-05, "loss": 2.7709, "step": 27938 }, { "epoch": 1.734372090135949, "grad_norm": 0.6028501427872941, "learning_rate": 4.509981688404088e-05, "loss": 2.7838, "step": 27939 }, { "epoch": 1.7344341672357069, "grad_norm": 0.643999747560582, "learning_rate": 4.509622275832904e-05, "loss": 2.7887, "step": 27940 }, { "epoch": 1.7344962443354648, "grad_norm": 0.5233196011896005, "learning_rate": 4.509262865820142e-05, "loss": 2.7035, "step": 27941 }, { "epoch": 1.7345583214352227, "grad_norm": 0.6127185791578826, "learning_rate": 4.508903458367675e-05, "loss": 2.7146, "step": 27942 }, { "epoch": 1.7346203985349804, "grad_norm": 0.5338864820670661, "learning_rate": 4.508544053477376e-05, "loss": 2.761, "step": 27943 }, { "epoch": 1.7346824756347383, "grad_norm": 0.5588626881068551, "learning_rate": 4.508184651151124e-05, "loss": 2.7408, "step": 27944 }, { "epoch": 1.7347445527344962, "grad_norm": 0.541173588246282, "learning_rate": 4.507825251390791e-05, "loss": 2.7129, "step": 27945 }, { "epoch": 1.7348066298342542, "grad_norm": 0.516135454378243, "learning_rate": 4.507465854198253e-05, "loss": 2.7348, "step": 27946 }, { "epoch": 1.734868706934012, "grad_norm": 0.5386692769055026, "learning_rate": 4.507106459575385e-05, "loss": 2.7015, "step": 27947 }, { "epoch": 1.73493078403377, "grad_norm": 0.5296519872412956, "learning_rate": 4.506747067524064e-05, "loss": 2.7438, "step": 27948 }, { "epoch": 1.7349928611335277, "grad_norm": 0.5336545772706097, "learning_rate": 4.5063876780461614e-05, "loss": 2.7077, "step": 27949 }, { "epoch": 1.7350549382332856, "grad_norm": 0.48845928761023133, "learning_rate": 4.506028291143555e-05, "loss": 2.7664, "step": 27950 }, { "epoch": 1.7351170153330435, "grad_norm": 0.5912443294616143, "learning_rate": 4.505668906818118e-05, "loss": 2.7651, "step": 27951 }, { "epoch": 1.7351790924328014, "grad_norm": 0.4836733412069564, "learning_rate": 4.505309525071728e-05, "loss": 2.7354, "step": 27952 }, { "epoch": 1.7352411695325594, "grad_norm": 0.5686823133916945, "learning_rate": 4.504950145906257e-05, "loss": 2.8313, "step": 27953 }, { "epoch": 1.7353032466323173, "grad_norm": 0.5165643804830949, "learning_rate": 4.5045907693235814e-05, "loss": 2.7028, "step": 27954 }, { "epoch": 1.7353653237320752, "grad_norm": 0.5402870643501881, "learning_rate": 4.504231395325577e-05, "loss": 2.7307, "step": 27955 }, { "epoch": 1.7354274008318331, "grad_norm": 0.516431813032962, "learning_rate": 4.503872023914116e-05, "loss": 2.671, "step": 27956 }, { "epoch": 1.735489477931591, "grad_norm": 0.5525839233525685, "learning_rate": 4.503512655091077e-05, "loss": 2.7531, "step": 27957 }, { "epoch": 1.735551555031349, "grad_norm": 0.596879017401951, "learning_rate": 4.503153288858331e-05, "loss": 2.7436, "step": 27958 }, { "epoch": 1.7356136321311069, "grad_norm": 0.530485500694834, "learning_rate": 4.502793925217756e-05, "loss": 2.6183, "step": 27959 }, { "epoch": 1.7356757092308648, "grad_norm": 0.5360219594172221, "learning_rate": 4.5024345641712254e-05, "loss": 2.6316, "step": 27960 }, { "epoch": 1.7357377863306227, "grad_norm": 0.544551956242806, "learning_rate": 4.5020752057206155e-05, "loss": 2.7447, "step": 27961 }, { "epoch": 1.7357998634303806, "grad_norm": 0.553784562317497, "learning_rate": 4.501715849867799e-05, "loss": 2.8366, "step": 27962 }, { "epoch": 1.7358619405301385, "grad_norm": 0.5058274130131739, "learning_rate": 4.501356496614653e-05, "loss": 2.7638, "step": 27963 }, { "epoch": 1.7359240176298965, "grad_norm": 0.5041652871210229, "learning_rate": 4.5009971459630494e-05, "loss": 2.7638, "step": 27964 }, { "epoch": 1.7359860947296544, "grad_norm": 0.5715339852922082, "learning_rate": 4.500637797914867e-05, "loss": 2.6956, "step": 27965 }, { "epoch": 1.7360481718294123, "grad_norm": 0.48991099174430275, "learning_rate": 4.500278452471978e-05, "loss": 2.7794, "step": 27966 }, { "epoch": 1.73611024892917, "grad_norm": 0.5999171060679839, "learning_rate": 4.499919109636256e-05, "loss": 2.7568, "step": 27967 }, { "epoch": 1.736172326028928, "grad_norm": 0.5034803616388321, "learning_rate": 4.499559769409579e-05, "loss": 2.6182, "step": 27968 }, { "epoch": 1.7362344031286858, "grad_norm": 0.5836107723914561, "learning_rate": 4.4992004317938184e-05, "loss": 2.7961, "step": 27969 }, { "epoch": 1.7362964802284437, "grad_norm": 0.5764403929800651, "learning_rate": 4.498841096790853e-05, "loss": 2.797, "step": 27970 }, { "epoch": 1.7363585573282017, "grad_norm": 0.5910464523958929, "learning_rate": 4.498481764402553e-05, "loss": 2.6887, "step": 27971 }, { "epoch": 1.7364206344279596, "grad_norm": 0.5353514044008854, "learning_rate": 4.498122434630795e-05, "loss": 2.6882, "step": 27972 }, { "epoch": 1.7364827115277173, "grad_norm": 0.5404813862382407, "learning_rate": 4.4977631074774564e-05, "loss": 2.7716, "step": 27973 }, { "epoch": 1.7365447886274752, "grad_norm": 0.5951632194882119, "learning_rate": 4.497403782944408e-05, "loss": 2.8208, "step": 27974 }, { "epoch": 1.736606865727233, "grad_norm": 0.5995364812722638, "learning_rate": 4.4970444610335275e-05, "loss": 2.8993, "step": 27975 }, { "epoch": 1.736668942826991, "grad_norm": 0.5800474267922083, "learning_rate": 4.4966851417466874e-05, "loss": 2.8004, "step": 27976 }, { "epoch": 1.736731019926749, "grad_norm": 0.58942936874054, "learning_rate": 4.496325825085764e-05, "loss": 2.8601, "step": 27977 }, { "epoch": 1.7367930970265069, "grad_norm": 0.5742496819501501, "learning_rate": 4.495966511052631e-05, "loss": 2.7887, "step": 27978 }, { "epoch": 1.7368551741262648, "grad_norm": 0.6236066784685828, "learning_rate": 4.495607199649161e-05, "loss": 2.6973, "step": 27979 }, { "epoch": 1.7369172512260227, "grad_norm": 0.5796992287901624, "learning_rate": 4.4952478908772325e-05, "loss": 2.7386, "step": 27980 }, { "epoch": 1.7369793283257806, "grad_norm": 0.5656863405887809, "learning_rate": 4.4948885847387175e-05, "loss": 2.7156, "step": 27981 }, { "epoch": 1.7370414054255385, "grad_norm": 0.606384274272026, "learning_rate": 4.494529281235492e-05, "loss": 2.7913, "step": 27982 }, { "epoch": 1.7371034825252964, "grad_norm": 0.5745672608585689, "learning_rate": 4.494169980369429e-05, "loss": 2.7878, "step": 27983 }, { "epoch": 1.7371655596250544, "grad_norm": 0.6086763631943841, "learning_rate": 4.493810682142405e-05, "loss": 2.7415, "step": 27984 }, { "epoch": 1.7372276367248123, "grad_norm": 0.5622393481024538, "learning_rate": 4.493451386556293e-05, "loss": 2.7567, "step": 27985 }, { "epoch": 1.7372897138245702, "grad_norm": 0.5830024450620543, "learning_rate": 4.493092093612969e-05, "loss": 2.701, "step": 27986 }, { "epoch": 1.7373517909243281, "grad_norm": 0.5255825443789116, "learning_rate": 4.492732803314305e-05, "loss": 2.7502, "step": 27987 }, { "epoch": 1.737413868024086, "grad_norm": 0.5867594993043402, "learning_rate": 4.492373515662178e-05, "loss": 2.7574, "step": 27988 }, { "epoch": 1.737475945123844, "grad_norm": 0.6038762222069208, "learning_rate": 4.492014230658463e-05, "loss": 2.7797, "step": 27989 }, { "epoch": 1.7375380222236019, "grad_norm": 0.5859350077258194, "learning_rate": 4.491654948305031e-05, "loss": 2.8075, "step": 27990 }, { "epoch": 1.7376000993233596, "grad_norm": 0.6334681469918803, "learning_rate": 4.491295668603759e-05, "loss": 2.7285, "step": 27991 }, { "epoch": 1.7376621764231175, "grad_norm": 0.543172784246263, "learning_rate": 4.490936391556521e-05, "loss": 2.7031, "step": 27992 }, { "epoch": 1.7377242535228754, "grad_norm": 0.5921033562941826, "learning_rate": 4.4905771171651925e-05, "loss": 2.8018, "step": 27993 }, { "epoch": 1.7377863306226333, "grad_norm": 0.6006630700676189, "learning_rate": 4.490217845431645e-05, "loss": 2.7875, "step": 27994 }, { "epoch": 1.7378484077223912, "grad_norm": 0.6224635870751162, "learning_rate": 4.489858576357756e-05, "loss": 2.8039, "step": 27995 }, { "epoch": 1.7379104848221492, "grad_norm": 0.5994154775036629, "learning_rate": 4.489499309945398e-05, "loss": 2.7287, "step": 27996 }, { "epoch": 1.7379725619219069, "grad_norm": 0.5562759007628051, "learning_rate": 4.489140046196447e-05, "loss": 2.6684, "step": 27997 }, { "epoch": 1.7380346390216648, "grad_norm": 0.5935408575236129, "learning_rate": 4.488780785112775e-05, "loss": 2.7352, "step": 27998 }, { "epoch": 1.7380967161214227, "grad_norm": 0.5737395330484104, "learning_rate": 4.4884215266962593e-05, "loss": 2.684, "step": 27999 }, { "epoch": 1.7381587932211806, "grad_norm": 0.5413289729202683, "learning_rate": 4.488062270948771e-05, "loss": 2.722, "step": 28000 }, { "epoch": 1.7382208703209385, "grad_norm": 0.6425327644454603, "learning_rate": 4.487703017872188e-05, "loss": 2.7075, "step": 28001 }, { "epoch": 1.7382829474206964, "grad_norm": 0.5566844454465557, "learning_rate": 4.4873437674683824e-05, "loss": 2.7836, "step": 28002 }, { "epoch": 1.7383450245204544, "grad_norm": 0.6614292218467883, "learning_rate": 4.486984519739227e-05, "loss": 2.765, "step": 28003 }, { "epoch": 1.7384071016202123, "grad_norm": 0.6174472551946959, "learning_rate": 4.4866252746866e-05, "loss": 2.7637, "step": 28004 }, { "epoch": 1.7384691787199702, "grad_norm": 0.5921003866413145, "learning_rate": 4.4862660323123714e-05, "loss": 2.7001, "step": 28005 }, { "epoch": 1.7385312558197281, "grad_norm": 0.6068042423944906, "learning_rate": 4.485906792618418e-05, "loss": 2.7777, "step": 28006 }, { "epoch": 1.738593332919486, "grad_norm": 0.5444650136323188, "learning_rate": 4.485547555606615e-05, "loss": 2.7524, "step": 28007 }, { "epoch": 1.738655410019244, "grad_norm": 0.598124501418461, "learning_rate": 4.485188321278834e-05, "loss": 2.8277, "step": 28008 }, { "epoch": 1.7387174871190019, "grad_norm": 0.512886229799124, "learning_rate": 4.484829089636952e-05, "loss": 2.7379, "step": 28009 }, { "epoch": 1.7387795642187598, "grad_norm": 0.5888540035562263, "learning_rate": 4.4844698606828406e-05, "loss": 2.7491, "step": 28010 }, { "epoch": 1.7388416413185177, "grad_norm": 0.5627082598122534, "learning_rate": 4.484110634418376e-05, "loss": 2.7623, "step": 28011 }, { "epoch": 1.7389037184182756, "grad_norm": 0.5513710489384467, "learning_rate": 4.4837514108454306e-05, "loss": 2.7775, "step": 28012 }, { "epoch": 1.7389657955180335, "grad_norm": 0.5780081759801836, "learning_rate": 4.48339218996588e-05, "loss": 2.6786, "step": 28013 }, { "epoch": 1.7390278726177915, "grad_norm": 0.49287369744217974, "learning_rate": 4.483032971781598e-05, "loss": 2.7267, "step": 28014 }, { "epoch": 1.7390899497175492, "grad_norm": 0.5812316625020032, "learning_rate": 4.482673756294458e-05, "loss": 2.8004, "step": 28015 }, { "epoch": 1.739152026817307, "grad_norm": 0.553702056652078, "learning_rate": 4.4823145435063346e-05, "loss": 2.7528, "step": 28016 }, { "epoch": 1.739214103917065, "grad_norm": 0.5666635652298295, "learning_rate": 4.481955333419101e-05, "loss": 2.8119, "step": 28017 }, { "epoch": 1.739276181016823, "grad_norm": 0.5722927568793515, "learning_rate": 4.481596126034634e-05, "loss": 2.723, "step": 28018 }, { "epoch": 1.7393382581165808, "grad_norm": 0.5986807276606259, "learning_rate": 4.4812369213548044e-05, "loss": 2.7814, "step": 28019 }, { "epoch": 1.7394003352163387, "grad_norm": 0.5444297071925457, "learning_rate": 4.480877719381489e-05, "loss": 2.8671, "step": 28020 }, { "epoch": 1.7394624123160964, "grad_norm": 0.56219296552918, "learning_rate": 4.480518520116559e-05, "loss": 2.7556, "step": 28021 }, { "epoch": 1.7395244894158544, "grad_norm": 0.5202450831729485, "learning_rate": 4.480159323561891e-05, "loss": 2.704, "step": 28022 }, { "epoch": 1.7395865665156123, "grad_norm": 0.581771434167764, "learning_rate": 4.4798001297193574e-05, "loss": 2.7782, "step": 28023 }, { "epoch": 1.7396486436153702, "grad_norm": 0.5918491957295264, "learning_rate": 4.4794409385908343e-05, "loss": 2.6579, "step": 28024 }, { "epoch": 1.7397107207151281, "grad_norm": 0.5764290187327155, "learning_rate": 4.479081750178192e-05, "loss": 2.7825, "step": 28025 }, { "epoch": 1.739772797814886, "grad_norm": 0.5799537294116786, "learning_rate": 4.478722564483308e-05, "loss": 2.8104, "step": 28026 }, { "epoch": 1.739834874914644, "grad_norm": 0.5966894520461639, "learning_rate": 4.478363381508055e-05, "loss": 2.7241, "step": 28027 }, { "epoch": 1.7398969520144019, "grad_norm": 0.5642329383826105, "learning_rate": 4.4780042012543064e-05, "loss": 2.72, "step": 28028 }, { "epoch": 1.7399590291141598, "grad_norm": 0.5961024212487753, "learning_rate": 4.477645023723938e-05, "loss": 2.7238, "step": 28029 }, { "epoch": 1.7400211062139177, "grad_norm": 0.5651882888364913, "learning_rate": 4.47728584891882e-05, "loss": 2.8049, "step": 28030 }, { "epoch": 1.7400831833136756, "grad_norm": 0.6106504151161336, "learning_rate": 4.4769266768408305e-05, "loss": 2.735, "step": 28031 }, { "epoch": 1.7401452604134335, "grad_norm": 0.5348579243490442, "learning_rate": 4.47656750749184e-05, "loss": 2.7646, "step": 28032 }, { "epoch": 1.7402073375131915, "grad_norm": 0.5917049122479936, "learning_rate": 4.4762083408737256e-05, "loss": 2.7253, "step": 28033 }, { "epoch": 1.7402694146129494, "grad_norm": 0.5906398781411212, "learning_rate": 4.475849176988358e-05, "loss": 2.7897, "step": 28034 }, { "epoch": 1.7403314917127073, "grad_norm": 0.5229380772932698, "learning_rate": 4.475490015837613e-05, "loss": 2.7254, "step": 28035 }, { "epoch": 1.7403935688124652, "grad_norm": 0.574507893535479, "learning_rate": 4.475130857423363e-05, "loss": 2.7736, "step": 28036 }, { "epoch": 1.7404556459122231, "grad_norm": 0.5805091925968424, "learning_rate": 4.4747717017474844e-05, "loss": 2.7647, "step": 28037 }, { "epoch": 1.740517723011981, "grad_norm": 0.5290511289254454, "learning_rate": 4.474412548811847e-05, "loss": 2.7292, "step": 28038 }, { "epoch": 1.7405798001117387, "grad_norm": 0.5445189541669484, "learning_rate": 4.474053398618329e-05, "loss": 2.6659, "step": 28039 }, { "epoch": 1.7406418772114967, "grad_norm": 0.5545340312287168, "learning_rate": 4.4736942511688006e-05, "loss": 2.6908, "step": 28040 }, { "epoch": 1.7407039543112546, "grad_norm": 0.5669455492980889, "learning_rate": 4.473335106465138e-05, "loss": 2.7584, "step": 28041 }, { "epoch": 1.7407660314110125, "grad_norm": 0.5704479840442532, "learning_rate": 4.472975964509214e-05, "loss": 2.7156, "step": 28042 }, { "epoch": 1.7408281085107704, "grad_norm": 0.6042810172902046, "learning_rate": 4.4726168253029034e-05, "loss": 2.7773, "step": 28043 }, { "epoch": 1.7408901856105283, "grad_norm": 0.5614098210039389, "learning_rate": 4.472257688848077e-05, "loss": 2.7425, "step": 28044 }, { "epoch": 1.740952262710286, "grad_norm": 0.6134903423774113, "learning_rate": 4.4718985551466115e-05, "loss": 2.7065, "step": 28045 }, { "epoch": 1.741014339810044, "grad_norm": 0.5856379951492707, "learning_rate": 4.4715394242003783e-05, "loss": 2.6671, "step": 28046 }, { "epoch": 1.7410764169098019, "grad_norm": 0.693544325540323, "learning_rate": 4.471180296011254e-05, "loss": 2.76, "step": 28047 }, { "epoch": 1.7411384940095598, "grad_norm": 0.5646460180451774, "learning_rate": 4.4708211705811084e-05, "loss": 2.7528, "step": 28048 }, { "epoch": 1.7412005711093177, "grad_norm": 0.6590589850790062, "learning_rate": 4.470462047911819e-05, "loss": 2.7381, "step": 28049 }, { "epoch": 1.7412626482090756, "grad_norm": 0.5616961139049438, "learning_rate": 4.470102928005256e-05, "loss": 2.7971, "step": 28050 }, { "epoch": 1.7413247253088335, "grad_norm": 0.609934501062841, "learning_rate": 4.4697438108632965e-05, "loss": 2.6636, "step": 28051 }, { "epoch": 1.7413868024085914, "grad_norm": 0.6033574363028312, "learning_rate": 4.4693846964878115e-05, "loss": 2.6596, "step": 28052 }, { "epoch": 1.7414488795083494, "grad_norm": 0.5667397405219953, "learning_rate": 4.469025584880675e-05, "loss": 2.7331, "step": 28053 }, { "epoch": 1.7415109566081073, "grad_norm": 0.6231898559290111, "learning_rate": 4.468666476043761e-05, "loss": 2.7782, "step": 28054 }, { "epoch": 1.7415730337078652, "grad_norm": 0.5542567286291594, "learning_rate": 4.468307369978943e-05, "loss": 2.6841, "step": 28055 }, { "epoch": 1.7416351108076231, "grad_norm": 0.6487132256068352, "learning_rate": 4.467948266688095e-05, "loss": 2.8142, "step": 28056 }, { "epoch": 1.741697187907381, "grad_norm": 0.5382890819760442, "learning_rate": 4.4675891661730896e-05, "loss": 2.7642, "step": 28057 }, { "epoch": 1.741759265007139, "grad_norm": 0.687015297561141, "learning_rate": 4.467230068435801e-05, "loss": 2.7904, "step": 28058 }, { "epoch": 1.7418213421068969, "grad_norm": 0.546147216449766, "learning_rate": 4.4668709734781017e-05, "loss": 2.7424, "step": 28059 }, { "epoch": 1.7418834192066548, "grad_norm": 0.6251284467315696, "learning_rate": 4.466511881301867e-05, "loss": 2.7176, "step": 28060 }, { "epoch": 1.7419454963064127, "grad_norm": 0.5600522879899891, "learning_rate": 4.466152791908969e-05, "loss": 2.6681, "step": 28061 }, { "epoch": 1.7420075734061706, "grad_norm": 0.614240467208713, "learning_rate": 4.465793705301282e-05, "loss": 2.8397, "step": 28062 }, { "epoch": 1.7420696505059283, "grad_norm": 0.6033876647600012, "learning_rate": 4.465434621480678e-05, "loss": 2.69, "step": 28063 }, { "epoch": 1.7421317276056862, "grad_norm": 0.6504054712199135, "learning_rate": 4.465075540449031e-05, "loss": 2.795, "step": 28064 }, { "epoch": 1.7421938047054442, "grad_norm": 0.6276449504174194, "learning_rate": 4.464716462208216e-05, "loss": 2.7301, "step": 28065 }, { "epoch": 1.742255881805202, "grad_norm": 0.6135813713719335, "learning_rate": 4.464357386760104e-05, "loss": 2.7628, "step": 28066 }, { "epoch": 1.74231795890496, "grad_norm": 0.6059691791407984, "learning_rate": 4.4639983141065714e-05, "loss": 2.7774, "step": 28067 }, { "epoch": 1.742380036004718, "grad_norm": 0.5610927887507049, "learning_rate": 4.4636392442494873e-05, "loss": 2.6999, "step": 28068 }, { "epoch": 1.7424421131044756, "grad_norm": 0.6267983313824925, "learning_rate": 4.463280177190729e-05, "loss": 2.7859, "step": 28069 }, { "epoch": 1.7425041902042335, "grad_norm": 0.5601097289789356, "learning_rate": 4.4629211129321676e-05, "loss": 2.7965, "step": 28070 }, { "epoch": 1.7425662673039914, "grad_norm": 0.5914078998416681, "learning_rate": 4.462562051475676e-05, "loss": 2.6784, "step": 28071 }, { "epoch": 1.7426283444037494, "grad_norm": 0.5284494746299151, "learning_rate": 4.46220299282313e-05, "loss": 2.6725, "step": 28072 }, { "epoch": 1.7426904215035073, "grad_norm": 0.6063017252849536, "learning_rate": 4.461843936976401e-05, "loss": 2.7513, "step": 28073 }, { "epoch": 1.7427524986032652, "grad_norm": 0.5480755026988107, "learning_rate": 4.461484883937364e-05, "loss": 2.6377, "step": 28074 }, { "epoch": 1.7428145757030231, "grad_norm": 0.6070883960901138, "learning_rate": 4.4611258337078905e-05, "loss": 2.7585, "step": 28075 }, { "epoch": 1.742876652802781, "grad_norm": 0.540984693393537, "learning_rate": 4.4607667862898545e-05, "loss": 2.7371, "step": 28076 }, { "epoch": 1.742938729902539, "grad_norm": 0.6230292243011908, "learning_rate": 4.46040774168513e-05, "loss": 2.7019, "step": 28077 }, { "epoch": 1.7430008070022969, "grad_norm": 0.5587506907414, "learning_rate": 4.460048699895587e-05, "loss": 2.7354, "step": 28078 }, { "epoch": 1.7430628841020548, "grad_norm": 0.6309741125494374, "learning_rate": 4.459689660923103e-05, "loss": 2.7497, "step": 28079 }, { "epoch": 1.7431249612018127, "grad_norm": 0.5269053014003838, "learning_rate": 4.4593306247695475e-05, "loss": 2.7132, "step": 28080 }, { "epoch": 1.7431870383015706, "grad_norm": 0.6500051690776405, "learning_rate": 4.458971591436797e-05, "loss": 2.7317, "step": 28081 }, { "epoch": 1.7432491154013285, "grad_norm": 0.5832601285595401, "learning_rate": 4.458612560926722e-05, "loss": 2.7051, "step": 28082 }, { "epoch": 1.7433111925010865, "grad_norm": 0.6231912905939595, "learning_rate": 4.458253533241198e-05, "loss": 2.7545, "step": 28083 }, { "epoch": 1.7433732696008444, "grad_norm": 0.6088775413818827, "learning_rate": 4.457894508382095e-05, "loss": 2.742, "step": 28084 }, { "epoch": 1.7434353467006023, "grad_norm": 0.6713314741120578, "learning_rate": 4.4575354863512896e-05, "loss": 2.8352, "step": 28085 }, { "epoch": 1.7434974238003602, "grad_norm": 0.5107635726973835, "learning_rate": 4.457176467150652e-05, "loss": 2.6706, "step": 28086 }, { "epoch": 1.743559500900118, "grad_norm": 0.595028930369447, "learning_rate": 4.456817450782058e-05, "loss": 2.7895, "step": 28087 }, { "epoch": 1.7436215779998758, "grad_norm": 0.5370242519549794, "learning_rate": 4.456458437247378e-05, "loss": 2.7356, "step": 28088 }, { "epoch": 1.7436836550996337, "grad_norm": 0.5559372858708588, "learning_rate": 4.4560994265484865e-05, "loss": 2.7326, "step": 28089 }, { "epoch": 1.7437457321993917, "grad_norm": 0.5963762326440684, "learning_rate": 4.4557404186872564e-05, "loss": 2.7806, "step": 28090 }, { "epoch": 1.7438078092991496, "grad_norm": 0.49306255844527846, "learning_rate": 4.4553814136655605e-05, "loss": 2.6773, "step": 28091 }, { "epoch": 1.7438698863989075, "grad_norm": 0.5491642855774466, "learning_rate": 4.4550224114852726e-05, "loss": 2.6385, "step": 28092 }, { "epoch": 1.7439319634986652, "grad_norm": 0.5174387042283505, "learning_rate": 4.454663412148264e-05, "loss": 2.7874, "step": 28093 }, { "epoch": 1.7439940405984231, "grad_norm": 0.5510219832464155, "learning_rate": 4.4543044156564104e-05, "loss": 2.7988, "step": 28094 }, { "epoch": 1.744056117698181, "grad_norm": 0.5197731258115436, "learning_rate": 4.453945422011581e-05, "loss": 2.7198, "step": 28095 }, { "epoch": 1.744118194797939, "grad_norm": 0.5101783221904322, "learning_rate": 4.453586431215653e-05, "loss": 2.681, "step": 28096 }, { "epoch": 1.7441802718976969, "grad_norm": 0.5960867975112014, "learning_rate": 4.4532274432704954e-05, "loss": 2.6851, "step": 28097 }, { "epoch": 1.7442423489974548, "grad_norm": 0.5240348709457244, "learning_rate": 4.4528684581779846e-05, "loss": 2.7263, "step": 28098 }, { "epoch": 1.7443044260972127, "grad_norm": 0.5772504893796886, "learning_rate": 4.452509475939991e-05, "loss": 2.7934, "step": 28099 }, { "epoch": 1.7443665031969706, "grad_norm": 0.5123271964826014, "learning_rate": 4.452150496558388e-05, "loss": 2.7423, "step": 28100 }, { "epoch": 1.7444285802967285, "grad_norm": 0.6231970292207256, "learning_rate": 4.45179152003505e-05, "loss": 2.7467, "step": 28101 }, { "epoch": 1.7444906573964865, "grad_norm": 0.5343289349873299, "learning_rate": 4.451432546371848e-05, "loss": 2.6874, "step": 28102 }, { "epoch": 1.7445527344962444, "grad_norm": 0.5543547962296059, "learning_rate": 4.4510735755706536e-05, "loss": 2.6846, "step": 28103 }, { "epoch": 1.7446148115960023, "grad_norm": 0.5884818465853066, "learning_rate": 4.450714607633344e-05, "loss": 2.735, "step": 28104 }, { "epoch": 1.7446768886957602, "grad_norm": 0.6089272443535563, "learning_rate": 4.4503556425617886e-05, "loss": 2.7327, "step": 28105 }, { "epoch": 1.7447389657955181, "grad_norm": 0.573023193743773, "learning_rate": 4.4499966803578623e-05, "loss": 2.8167, "step": 28106 }, { "epoch": 1.744801042895276, "grad_norm": 0.6071670586506809, "learning_rate": 4.449637721023436e-05, "loss": 2.6708, "step": 28107 }, { "epoch": 1.744863119995034, "grad_norm": 0.5453362478861924, "learning_rate": 4.449278764560384e-05, "loss": 2.6998, "step": 28108 }, { "epoch": 1.7449251970947919, "grad_norm": 0.5842866402392857, "learning_rate": 4.448919810970578e-05, "loss": 2.7017, "step": 28109 }, { "epoch": 1.7449872741945496, "grad_norm": 0.5823248564303878, "learning_rate": 4.448560860255891e-05, "loss": 2.8179, "step": 28110 }, { "epoch": 1.7450493512943075, "grad_norm": 0.565494408469378, "learning_rate": 4.448201912418196e-05, "loss": 2.6485, "step": 28111 }, { "epoch": 1.7451114283940654, "grad_norm": 0.6310518992582901, "learning_rate": 4.4478429674593665e-05, "loss": 2.6913, "step": 28112 }, { "epoch": 1.7451735054938233, "grad_norm": 0.4855340349827446, "learning_rate": 4.447484025381274e-05, "loss": 2.6708, "step": 28113 }, { "epoch": 1.7452355825935812, "grad_norm": 0.5890481395640571, "learning_rate": 4.4471250861857906e-05, "loss": 2.8216, "step": 28114 }, { "epoch": 1.7452976596933392, "grad_norm": 0.5378696955623854, "learning_rate": 4.44676614987479e-05, "loss": 2.6753, "step": 28115 }, { "epoch": 1.7453597367930969, "grad_norm": 0.5329219090509841, "learning_rate": 4.4464072164501445e-05, "loss": 2.7654, "step": 28116 }, { "epoch": 1.7454218138928548, "grad_norm": 0.524658572346383, "learning_rate": 4.446048285913728e-05, "loss": 2.6386, "step": 28117 }, { "epoch": 1.7454838909926127, "grad_norm": 0.5105750428581958, "learning_rate": 4.4456893582674105e-05, "loss": 2.7617, "step": 28118 }, { "epoch": 1.7455459680923706, "grad_norm": 0.5665335676545733, "learning_rate": 4.445330433513068e-05, "loss": 2.7662, "step": 28119 }, { "epoch": 1.7456080451921285, "grad_norm": 0.5599224586957484, "learning_rate": 4.4449715116525696e-05, "loss": 2.7943, "step": 28120 }, { "epoch": 1.7456701222918865, "grad_norm": 0.5328928844860916, "learning_rate": 4.444612592687791e-05, "loss": 2.7828, "step": 28121 }, { "epoch": 1.7457321993916444, "grad_norm": 0.5507376092165408, "learning_rate": 4.444253676620602e-05, "loss": 2.7528, "step": 28122 }, { "epoch": 1.7457942764914023, "grad_norm": 0.5393137415448234, "learning_rate": 4.443894763452878e-05, "loss": 2.7738, "step": 28123 }, { "epoch": 1.7458563535911602, "grad_norm": 0.5335163970776333, "learning_rate": 4.443535853186489e-05, "loss": 2.661, "step": 28124 }, { "epoch": 1.7459184306909181, "grad_norm": 0.58790165076578, "learning_rate": 4.443176945823309e-05, "loss": 2.6812, "step": 28125 }, { "epoch": 1.745980507790676, "grad_norm": 0.5135821710156934, "learning_rate": 4.44281804136521e-05, "loss": 2.8114, "step": 28126 }, { "epoch": 1.746042584890434, "grad_norm": 0.5406198534417357, "learning_rate": 4.442459139814064e-05, "loss": 2.7661, "step": 28127 }, { "epoch": 1.7461046619901919, "grad_norm": 0.5334149732788808, "learning_rate": 4.442100241171744e-05, "loss": 2.7023, "step": 28128 }, { "epoch": 1.7461667390899498, "grad_norm": 0.472149963982397, "learning_rate": 4.441741345440123e-05, "loss": 2.7292, "step": 28129 }, { "epoch": 1.7462288161897077, "grad_norm": 0.5503517261672483, "learning_rate": 4.4413824526210725e-05, "loss": 2.7284, "step": 28130 }, { "epoch": 1.7462908932894656, "grad_norm": 0.5043934086272901, "learning_rate": 4.441023562716465e-05, "loss": 2.8022, "step": 28131 }, { "epoch": 1.7463529703892235, "grad_norm": 0.519236746472409, "learning_rate": 4.4406646757281745e-05, "loss": 2.7596, "step": 28132 }, { "epoch": 1.7464150474889815, "grad_norm": 0.530938281439823, "learning_rate": 4.440305791658071e-05, "loss": 2.7554, "step": 28133 }, { "epoch": 1.7464771245887392, "grad_norm": 0.5575181970064171, "learning_rate": 4.439946910508029e-05, "loss": 2.738, "step": 28134 }, { "epoch": 1.746539201688497, "grad_norm": 0.5949488883545015, "learning_rate": 4.4395880322799186e-05, "loss": 2.7741, "step": 28135 }, { "epoch": 1.746601278788255, "grad_norm": 0.5184786178627451, "learning_rate": 4.439229156975613e-05, "loss": 2.74, "step": 28136 }, { "epoch": 1.746663355888013, "grad_norm": 0.6156702102481015, "learning_rate": 4.438870284596986e-05, "loss": 2.7441, "step": 28137 }, { "epoch": 1.7467254329877708, "grad_norm": 0.5724336032595274, "learning_rate": 4.438511415145911e-05, "loss": 2.7178, "step": 28138 }, { "epoch": 1.7467875100875287, "grad_norm": 0.6045117351964217, "learning_rate": 4.4381525486242554e-05, "loss": 2.7332, "step": 28139 }, { "epoch": 1.7468495871872864, "grad_norm": 0.5605222583642282, "learning_rate": 4.437793685033896e-05, "loss": 2.7895, "step": 28140 }, { "epoch": 1.7469116642870444, "grad_norm": 0.6574012558186103, "learning_rate": 4.437434824376702e-05, "loss": 2.7699, "step": 28141 }, { "epoch": 1.7469737413868023, "grad_norm": 0.5550871470481222, "learning_rate": 4.437075966654549e-05, "loss": 2.7741, "step": 28142 }, { "epoch": 1.7470358184865602, "grad_norm": 0.6234305465042765, "learning_rate": 4.436717111869306e-05, "loss": 2.8466, "step": 28143 }, { "epoch": 1.7470978955863181, "grad_norm": 0.6270843796283024, "learning_rate": 4.436358260022847e-05, "loss": 2.8021, "step": 28144 }, { "epoch": 1.747159972686076, "grad_norm": 0.5723047861494011, "learning_rate": 4.435999411117043e-05, "loss": 2.7546, "step": 28145 }, { "epoch": 1.747222049785834, "grad_norm": 0.5436919356727702, "learning_rate": 4.4356405651537683e-05, "loss": 2.7746, "step": 28146 }, { "epoch": 1.7472841268855919, "grad_norm": 0.5213069256680132, "learning_rate": 4.435281722134893e-05, "loss": 2.7563, "step": 28147 }, { "epoch": 1.7473462039853498, "grad_norm": 0.5540462602106155, "learning_rate": 4.434922882062291e-05, "loss": 2.7577, "step": 28148 }, { "epoch": 1.7474082810851077, "grad_norm": 0.5508057609136325, "learning_rate": 4.434564044937833e-05, "loss": 2.7004, "step": 28149 }, { "epoch": 1.7474703581848656, "grad_norm": 0.5267773730331152, "learning_rate": 4.4342052107633905e-05, "loss": 2.7055, "step": 28150 }, { "epoch": 1.7475324352846235, "grad_norm": 0.5509704373143282, "learning_rate": 4.433846379540838e-05, "loss": 2.816, "step": 28151 }, { "epoch": 1.7475945123843815, "grad_norm": 0.47996094082819685, "learning_rate": 4.4334875512720454e-05, "loss": 2.6328, "step": 28152 }, { "epoch": 1.7476565894841394, "grad_norm": 0.5364223178020444, "learning_rate": 4.433128725958887e-05, "loss": 2.667, "step": 28153 }, { "epoch": 1.7477186665838973, "grad_norm": 0.5359669969187633, "learning_rate": 4.4327699036032324e-05, "loss": 2.7679, "step": 28154 }, { "epoch": 1.7477807436836552, "grad_norm": 0.5989308447680544, "learning_rate": 4.432411084206956e-05, "loss": 2.7367, "step": 28155 }, { "epoch": 1.7478428207834131, "grad_norm": 0.5522324402704043, "learning_rate": 4.432052267771928e-05, "loss": 2.8502, "step": 28156 }, { "epoch": 1.747904897883171, "grad_norm": 0.571281133091338, "learning_rate": 4.4316934543000224e-05, "loss": 2.664, "step": 28157 }, { "epoch": 1.7479669749829287, "grad_norm": 0.5060133828459045, "learning_rate": 4.431334643793108e-05, "loss": 2.7293, "step": 28158 }, { "epoch": 1.7480290520826867, "grad_norm": 0.5715405815264198, "learning_rate": 4.430975836253061e-05, "loss": 2.729, "step": 28159 }, { "epoch": 1.7480911291824446, "grad_norm": 0.552802388350507, "learning_rate": 4.430617031681749e-05, "loss": 2.7056, "step": 28160 }, { "epoch": 1.7481532062822025, "grad_norm": 0.5997476076712048, "learning_rate": 4.430258230081048e-05, "loss": 2.7608, "step": 28161 }, { "epoch": 1.7482152833819604, "grad_norm": 0.5795578341173353, "learning_rate": 4.429899431452828e-05, "loss": 2.7168, "step": 28162 }, { "epoch": 1.7482773604817183, "grad_norm": 0.6223492678111059, "learning_rate": 4.429540635798959e-05, "loss": 2.7511, "step": 28163 }, { "epoch": 1.748339437581476, "grad_norm": 0.6087806933613763, "learning_rate": 4.4291818431213175e-05, "loss": 2.8266, "step": 28164 }, { "epoch": 1.748401514681234, "grad_norm": 0.5129175776120527, "learning_rate": 4.428823053421771e-05, "loss": 2.6929, "step": 28165 }, { "epoch": 1.7484635917809919, "grad_norm": 0.6045681808600717, "learning_rate": 4.428464266702195e-05, "loss": 2.7684, "step": 28166 }, { "epoch": 1.7485256688807498, "grad_norm": 0.5093438168207409, "learning_rate": 4.428105482964458e-05, "loss": 2.7468, "step": 28167 }, { "epoch": 1.7485877459805077, "grad_norm": 0.5606583275133236, "learning_rate": 4.427746702210435e-05, "loss": 2.7925, "step": 28168 }, { "epoch": 1.7486498230802656, "grad_norm": 0.52762819318928, "learning_rate": 4.427387924441994e-05, "loss": 2.7995, "step": 28169 }, { "epoch": 1.7487119001800235, "grad_norm": 0.488841384042117, "learning_rate": 4.427029149661011e-05, "loss": 2.674, "step": 28170 }, { "epoch": 1.7487739772797815, "grad_norm": 0.574301707185549, "learning_rate": 4.426670377869356e-05, "loss": 2.8323, "step": 28171 }, { "epoch": 1.7488360543795394, "grad_norm": 0.5051543577272635, "learning_rate": 4.4263116090689e-05, "loss": 2.7607, "step": 28172 }, { "epoch": 1.7488981314792973, "grad_norm": 0.5264778091739017, "learning_rate": 4.4259528432615175e-05, "loss": 2.6975, "step": 28173 }, { "epoch": 1.7489602085790552, "grad_norm": 0.5427275197255244, "learning_rate": 4.425594080449077e-05, "loss": 2.7303, "step": 28174 }, { "epoch": 1.7490222856788131, "grad_norm": 0.5753602340348306, "learning_rate": 4.425235320633451e-05, "loss": 2.712, "step": 28175 }, { "epoch": 1.749084362778571, "grad_norm": 0.5650600087950881, "learning_rate": 4.424876563816513e-05, "loss": 2.6622, "step": 28176 }, { "epoch": 1.749146439878329, "grad_norm": 0.5995625281905048, "learning_rate": 4.4245178100001334e-05, "loss": 2.7524, "step": 28177 }, { "epoch": 1.7492085169780869, "grad_norm": 0.49224665452819033, "learning_rate": 4.4241590591861845e-05, "loss": 2.7607, "step": 28178 }, { "epoch": 1.7492705940778448, "grad_norm": 0.6055122605963845, "learning_rate": 4.423800311376536e-05, "loss": 2.833, "step": 28179 }, { "epoch": 1.7493326711776027, "grad_norm": 0.49968222116111793, "learning_rate": 4.423441566573063e-05, "loss": 2.7102, "step": 28180 }, { "epoch": 1.7493947482773606, "grad_norm": 0.6085631268574351, "learning_rate": 4.423082824777634e-05, "loss": 2.8103, "step": 28181 }, { "epoch": 1.7494568253771183, "grad_norm": 0.5898818452166698, "learning_rate": 4.422724085992123e-05, "loss": 2.7762, "step": 28182 }, { "epoch": 1.7495189024768762, "grad_norm": 0.646379583342146, "learning_rate": 4.4223653502183994e-05, "loss": 2.7097, "step": 28183 }, { "epoch": 1.7495809795766342, "grad_norm": 0.5799385185376644, "learning_rate": 4.422006617458338e-05, "loss": 2.757, "step": 28184 }, { "epoch": 1.749643056676392, "grad_norm": 0.658830292964025, "learning_rate": 4.421647887713807e-05, "loss": 2.8632, "step": 28185 }, { "epoch": 1.74970513377615, "grad_norm": 0.48501214299780554, "learning_rate": 4.4212891609866785e-05, "loss": 2.7277, "step": 28186 }, { "epoch": 1.749767210875908, "grad_norm": 0.5351604266595398, "learning_rate": 4.420930437278826e-05, "loss": 2.7167, "step": 28187 }, { "epoch": 1.7498292879756656, "grad_norm": 0.48511224445773793, "learning_rate": 4.42057171659212e-05, "loss": 2.7251, "step": 28188 }, { "epoch": 1.7498913650754235, "grad_norm": 0.5468842369389425, "learning_rate": 4.4202129989284325e-05, "loss": 2.7314, "step": 28189 }, { "epoch": 1.7499534421751815, "grad_norm": 0.4759559053956139, "learning_rate": 4.419854284289633e-05, "loss": 2.6902, "step": 28190 }, { "epoch": 1.7500155192749394, "grad_norm": 0.5259408445019609, "learning_rate": 4.419495572677596e-05, "loss": 2.7425, "step": 28191 }, { "epoch": 1.7500775963746973, "grad_norm": 0.5445127421097556, "learning_rate": 4.419136864094191e-05, "loss": 2.694, "step": 28192 }, { "epoch": 1.7501396734744552, "grad_norm": 0.5469944959604186, "learning_rate": 4.4187781585412895e-05, "loss": 2.7833, "step": 28193 }, { "epoch": 1.7502017505742131, "grad_norm": 0.5650039414040479, "learning_rate": 4.4184194560207635e-05, "loss": 2.6678, "step": 28194 }, { "epoch": 1.750263827673971, "grad_norm": 0.5706463450673167, "learning_rate": 4.418060756534485e-05, "loss": 2.8453, "step": 28195 }, { "epoch": 1.750325904773729, "grad_norm": 0.5080891955759449, "learning_rate": 4.417702060084324e-05, "loss": 2.7438, "step": 28196 }, { "epoch": 1.7503879818734869, "grad_norm": 0.5889090401372604, "learning_rate": 4.4173433666721544e-05, "loss": 2.8029, "step": 28197 }, { "epoch": 1.7504500589732448, "grad_norm": 0.534938573073844, "learning_rate": 4.416984676299845e-05, "loss": 2.7102, "step": 28198 }, { "epoch": 1.7505121360730027, "grad_norm": 0.5849277951845445, "learning_rate": 4.4166259889692676e-05, "loss": 2.7834, "step": 28199 }, { "epoch": 1.7505742131727606, "grad_norm": 0.5707253829720164, "learning_rate": 4.416267304682295e-05, "loss": 2.6857, "step": 28200 }, { "epoch": 1.7506362902725185, "grad_norm": 0.5771853827255401, "learning_rate": 4.4159086234407956e-05, "loss": 2.741, "step": 28201 }, { "epoch": 1.7506983673722765, "grad_norm": 0.5696341517125325, "learning_rate": 4.415549945246642e-05, "loss": 2.7152, "step": 28202 }, { "epoch": 1.7507604444720344, "grad_norm": 0.5410838170231785, "learning_rate": 4.415191270101709e-05, "loss": 2.6832, "step": 28203 }, { "epoch": 1.7508225215717923, "grad_norm": 0.5879817651204813, "learning_rate": 4.4148325980078635e-05, "loss": 2.8153, "step": 28204 }, { "epoch": 1.7508845986715502, "grad_norm": 0.5787906680638696, "learning_rate": 4.41447392896698e-05, "loss": 2.7932, "step": 28205 }, { "epoch": 1.750946675771308, "grad_norm": 0.6216498710248665, "learning_rate": 4.4141152629809266e-05, "loss": 2.7036, "step": 28206 }, { "epoch": 1.7510087528710658, "grad_norm": 0.546429288976752, "learning_rate": 4.413756600051577e-05, "loss": 2.6847, "step": 28207 }, { "epoch": 1.7510708299708238, "grad_norm": 0.6016606585192956, "learning_rate": 4.413397940180801e-05, "loss": 2.7673, "step": 28208 }, { "epoch": 1.7511329070705817, "grad_norm": 0.6236013259370152, "learning_rate": 4.413039283370471e-05, "loss": 2.7726, "step": 28209 }, { "epoch": 1.7511949841703396, "grad_norm": 0.5894364725621094, "learning_rate": 4.412680629622458e-05, "loss": 2.7395, "step": 28210 }, { "epoch": 1.7512570612700975, "grad_norm": 0.7021162426269807, "learning_rate": 4.4123219789386314e-05, "loss": 2.821, "step": 28211 }, { "epoch": 1.7513191383698552, "grad_norm": 0.5906116968503514, "learning_rate": 4.411963331320865e-05, "loss": 2.7182, "step": 28212 }, { "epoch": 1.7513812154696131, "grad_norm": 0.6321423596771968, "learning_rate": 4.411604686771027e-05, "loss": 2.709, "step": 28213 }, { "epoch": 1.751443292569371, "grad_norm": 0.578115102322369, "learning_rate": 4.411246045290992e-05, "loss": 2.7142, "step": 28214 }, { "epoch": 1.751505369669129, "grad_norm": 0.5649880643022748, "learning_rate": 4.4108874068826275e-05, "loss": 2.8046, "step": 28215 }, { "epoch": 1.7515674467688869, "grad_norm": 0.617763926131012, "learning_rate": 4.410528771547808e-05, "loss": 2.7293, "step": 28216 }, { "epoch": 1.7516295238686448, "grad_norm": 0.561134880638361, "learning_rate": 4.410170139288402e-05, "loss": 2.7578, "step": 28217 }, { "epoch": 1.7516916009684027, "grad_norm": 0.579291584301979, "learning_rate": 4.409811510106282e-05, "loss": 2.7229, "step": 28218 }, { "epoch": 1.7517536780681606, "grad_norm": 0.5841244940923929, "learning_rate": 4.409452884003318e-05, "loss": 2.7293, "step": 28219 }, { "epoch": 1.7518157551679185, "grad_norm": 0.5725445116638356, "learning_rate": 4.409094260981383e-05, "loss": 2.7703, "step": 28220 }, { "epoch": 1.7518778322676765, "grad_norm": 0.6070585237363206, "learning_rate": 4.408735641042345e-05, "loss": 2.8073, "step": 28221 }, { "epoch": 1.7519399093674344, "grad_norm": 0.5748104498286309, "learning_rate": 4.4083770241880786e-05, "loss": 2.736, "step": 28222 }, { "epoch": 1.7520019864671923, "grad_norm": 0.5755695103792003, "learning_rate": 4.408018410420452e-05, "loss": 2.6835, "step": 28223 }, { "epoch": 1.7520640635669502, "grad_norm": 0.6059540557173494, "learning_rate": 4.407659799741336e-05, "loss": 2.7936, "step": 28224 }, { "epoch": 1.7521261406667081, "grad_norm": 0.5276575638266426, "learning_rate": 4.4073011921526046e-05, "loss": 2.6497, "step": 28225 }, { "epoch": 1.752188217766466, "grad_norm": 0.5933686587804675, "learning_rate": 4.406942587656125e-05, "loss": 2.7909, "step": 28226 }, { "epoch": 1.752250294866224, "grad_norm": 0.5470374274865253, "learning_rate": 4.406583986253771e-05, "loss": 2.7167, "step": 28227 }, { "epoch": 1.7523123719659819, "grad_norm": 0.5284253366831206, "learning_rate": 4.406225387947411e-05, "loss": 2.7082, "step": 28228 }, { "epoch": 1.7523744490657398, "grad_norm": 0.5793650704677997, "learning_rate": 4.4058667927389196e-05, "loss": 2.6925, "step": 28229 }, { "epoch": 1.7524365261654975, "grad_norm": 0.5128592964592787, "learning_rate": 4.4055082006301636e-05, "loss": 2.7486, "step": 28230 }, { "epoch": 1.7524986032652554, "grad_norm": 0.607983957736229, "learning_rate": 4.4051496116230165e-05, "loss": 2.7677, "step": 28231 }, { "epoch": 1.7525606803650133, "grad_norm": 0.5044869322582458, "learning_rate": 4.4047910257193467e-05, "loss": 2.7489, "step": 28232 }, { "epoch": 1.7526227574647713, "grad_norm": 0.6311051920235835, "learning_rate": 4.404432442921029e-05, "loss": 2.8814, "step": 28233 }, { "epoch": 1.7526848345645292, "grad_norm": 0.6004565836229517, "learning_rate": 4.404073863229929e-05, "loss": 2.7384, "step": 28234 }, { "epoch": 1.752746911664287, "grad_norm": 0.5344464683703676, "learning_rate": 4.403715286647922e-05, "loss": 2.7226, "step": 28235 }, { "epoch": 1.7528089887640448, "grad_norm": 0.5762464313437953, "learning_rate": 4.403356713176876e-05, "loss": 2.7803, "step": 28236 }, { "epoch": 1.7528710658638027, "grad_norm": 0.5571323030252908, "learning_rate": 4.402998142818665e-05, "loss": 2.7146, "step": 28237 }, { "epoch": 1.7529331429635606, "grad_norm": 0.5659273951537783, "learning_rate": 4.4026395755751556e-05, "loss": 2.7364, "step": 28238 }, { "epoch": 1.7529952200633185, "grad_norm": 0.5676257717981659, "learning_rate": 4.4022810114482224e-05, "loss": 2.7414, "step": 28239 }, { "epoch": 1.7530572971630765, "grad_norm": 0.5815464279110361, "learning_rate": 4.4019224504397325e-05, "loss": 2.7961, "step": 28240 }, { "epoch": 1.7531193742628344, "grad_norm": 0.5999791499329865, "learning_rate": 4.40156389255156e-05, "loss": 2.7357, "step": 28241 }, { "epoch": 1.7531814513625923, "grad_norm": 0.6292838390664778, "learning_rate": 4.401205337785573e-05, "loss": 2.7349, "step": 28242 }, { "epoch": 1.7532435284623502, "grad_norm": 0.6470242692045264, "learning_rate": 4.400846786143644e-05, "loss": 2.7242, "step": 28243 }, { "epoch": 1.7533056055621081, "grad_norm": 0.6579991599080482, "learning_rate": 4.400488237627641e-05, "loss": 2.7148, "step": 28244 }, { "epoch": 1.753367682661866, "grad_norm": 0.599470364737721, "learning_rate": 4.400129692239438e-05, "loss": 2.7622, "step": 28245 }, { "epoch": 1.753429759761624, "grad_norm": 0.7359409275501558, "learning_rate": 4.399771149980903e-05, "loss": 2.7748, "step": 28246 }, { "epoch": 1.7534918368613819, "grad_norm": 0.7139548523143382, "learning_rate": 4.3994126108539084e-05, "loss": 2.7003, "step": 28247 }, { "epoch": 1.7535539139611398, "grad_norm": 0.6241818556785825, "learning_rate": 4.399054074860324e-05, "loss": 2.7707, "step": 28248 }, { "epoch": 1.7536159910608977, "grad_norm": 0.6997830335677553, "learning_rate": 4.39869554200202e-05, "loss": 2.7127, "step": 28249 }, { "epoch": 1.7536780681606556, "grad_norm": 0.6402929422380688, "learning_rate": 4.398337012280868e-05, "loss": 2.8191, "step": 28250 }, { "epoch": 1.7537401452604136, "grad_norm": 0.7467116747926247, "learning_rate": 4.397978485698736e-05, "loss": 2.7425, "step": 28251 }, { "epoch": 1.7538022223601715, "grad_norm": 0.6040850918606059, "learning_rate": 4.397619962257498e-05, "loss": 2.7855, "step": 28252 }, { "epoch": 1.7538642994599294, "grad_norm": 0.6739201579614555, "learning_rate": 4.397261441959022e-05, "loss": 2.7123, "step": 28253 }, { "epoch": 1.753926376559687, "grad_norm": 0.5792413964075069, "learning_rate": 4.396902924805181e-05, "loss": 2.7828, "step": 28254 }, { "epoch": 1.753988453659445, "grad_norm": 0.6107569942503285, "learning_rate": 4.396544410797842e-05, "loss": 2.6622, "step": 28255 }, { "epoch": 1.754050530759203, "grad_norm": 0.6113439198363904, "learning_rate": 4.3961858999388786e-05, "loss": 2.7524, "step": 28256 }, { "epoch": 1.7541126078589608, "grad_norm": 0.579501986573429, "learning_rate": 4.3958273922301584e-05, "loss": 2.7424, "step": 28257 }, { "epoch": 1.7541746849587188, "grad_norm": 0.6005954648634559, "learning_rate": 4.395468887673555e-05, "loss": 2.7463, "step": 28258 }, { "epoch": 1.7542367620584767, "grad_norm": 0.5970340307191427, "learning_rate": 4.3951103862709374e-05, "loss": 2.7011, "step": 28259 }, { "epoch": 1.7542988391582344, "grad_norm": 0.5797445192082079, "learning_rate": 4.394751888024174e-05, "loss": 2.7573, "step": 28260 }, { "epoch": 1.7543609162579923, "grad_norm": 0.5955419603805798, "learning_rate": 4.3943933929351385e-05, "loss": 2.756, "step": 28261 }, { "epoch": 1.7544229933577502, "grad_norm": 0.5567028613182637, "learning_rate": 4.394034901005699e-05, "loss": 2.6616, "step": 28262 }, { "epoch": 1.7544850704575081, "grad_norm": 0.6124640201199311, "learning_rate": 4.3936764122377264e-05, "loss": 2.6947, "step": 28263 }, { "epoch": 1.754547147557266, "grad_norm": 0.5851741395871713, "learning_rate": 4.393317926633091e-05, "loss": 2.7616, "step": 28264 }, { "epoch": 1.754609224657024, "grad_norm": 0.6143087053669796, "learning_rate": 4.392959444193664e-05, "loss": 2.765, "step": 28265 }, { "epoch": 1.7546713017567819, "grad_norm": 0.608338403571786, "learning_rate": 4.392600964921314e-05, "loss": 2.7495, "step": 28266 }, { "epoch": 1.7547333788565398, "grad_norm": 0.5579063347288792, "learning_rate": 4.392242488817911e-05, "loss": 2.7931, "step": 28267 }, { "epoch": 1.7547954559562977, "grad_norm": 0.5908270969474233, "learning_rate": 4.3918840158853284e-05, "loss": 2.8249, "step": 28268 }, { "epoch": 1.7548575330560556, "grad_norm": 0.5903077731894917, "learning_rate": 4.391525546125433e-05, "loss": 2.6368, "step": 28269 }, { "epoch": 1.7549196101558135, "grad_norm": 0.575316023020652, "learning_rate": 4.391167079540099e-05, "loss": 2.8308, "step": 28270 }, { "epoch": 1.7549816872555715, "grad_norm": 0.5561179400780668, "learning_rate": 4.3908086161311925e-05, "loss": 2.7816, "step": 28271 }, { "epoch": 1.7550437643553294, "grad_norm": 0.6154562898078152, "learning_rate": 4.390450155900586e-05, "loss": 2.7613, "step": 28272 }, { "epoch": 1.7551058414550873, "grad_norm": 0.5218755562145444, "learning_rate": 4.390091698850149e-05, "loss": 2.6449, "step": 28273 }, { "epoch": 1.7551679185548452, "grad_norm": 0.59224522794908, "learning_rate": 4.3897332449817516e-05, "loss": 2.7112, "step": 28274 }, { "epoch": 1.7552299956546031, "grad_norm": 0.6121376211225736, "learning_rate": 4.389374794297264e-05, "loss": 2.8478, "step": 28275 }, { "epoch": 1.755292072754361, "grad_norm": 0.5569992783882494, "learning_rate": 4.3890163467985555e-05, "loss": 2.7653, "step": 28276 }, { "epoch": 1.755354149854119, "grad_norm": 0.6155260151711527, "learning_rate": 4.388657902487498e-05, "loss": 2.8325, "step": 28277 }, { "epoch": 1.7554162269538767, "grad_norm": 0.5276246629545341, "learning_rate": 4.3882994613659605e-05, "loss": 2.7495, "step": 28278 }, { "epoch": 1.7554783040536346, "grad_norm": 0.6156806417193013, "learning_rate": 4.387941023435814e-05, "loss": 2.8065, "step": 28279 }, { "epoch": 1.7555403811533925, "grad_norm": 0.593940026843992, "learning_rate": 4.387582588698926e-05, "loss": 2.7857, "step": 28280 }, { "epoch": 1.7556024582531504, "grad_norm": 0.5680125905659646, "learning_rate": 4.3872241571571696e-05, "loss": 2.6611, "step": 28281 }, { "epoch": 1.7556645353529083, "grad_norm": 0.5874200502740946, "learning_rate": 4.386865728812412e-05, "loss": 2.7955, "step": 28282 }, { "epoch": 1.7557266124526663, "grad_norm": 0.636474609375, "learning_rate": 4.3865073036665265e-05, "loss": 2.6878, "step": 28283 }, { "epoch": 1.755788689552424, "grad_norm": 0.5993732148860884, "learning_rate": 4.386148881721382e-05, "loss": 2.7623, "step": 28284 }, { "epoch": 1.7558507666521819, "grad_norm": 0.571691335566603, "learning_rate": 4.3857904629788454e-05, "loss": 2.6823, "step": 28285 }, { "epoch": 1.7559128437519398, "grad_norm": 0.6354191863421735, "learning_rate": 4.3854320474407906e-05, "loss": 2.8079, "step": 28286 }, { "epoch": 1.7559749208516977, "grad_norm": 0.5857994679945109, "learning_rate": 4.385073635109085e-05, "loss": 2.7771, "step": 28287 }, { "epoch": 1.7560369979514556, "grad_norm": 0.6272158681234925, "learning_rate": 4.3847152259856005e-05, "loss": 2.7995, "step": 28288 }, { "epoch": 1.7560990750512135, "grad_norm": 0.5482253976825588, "learning_rate": 4.384356820072205e-05, "loss": 2.7901, "step": 28289 }, { "epoch": 1.7561611521509715, "grad_norm": 0.6407324072885547, "learning_rate": 4.3839984173707704e-05, "loss": 2.6333, "step": 28290 }, { "epoch": 1.7562232292507294, "grad_norm": 0.5677220287479244, "learning_rate": 4.3836400178831646e-05, "loss": 2.7869, "step": 28291 }, { "epoch": 1.7562853063504873, "grad_norm": 0.6520106213119921, "learning_rate": 4.38328162161126e-05, "loss": 2.7839, "step": 28292 }, { "epoch": 1.7563473834502452, "grad_norm": 0.6079735657739704, "learning_rate": 4.3829232285569235e-05, "loss": 2.7101, "step": 28293 }, { "epoch": 1.7564094605500031, "grad_norm": 0.6157083532379795, "learning_rate": 4.382564838722027e-05, "loss": 2.7635, "step": 28294 }, { "epoch": 1.756471537649761, "grad_norm": 0.571942495313208, "learning_rate": 4.38220645210844e-05, "loss": 2.7354, "step": 28295 }, { "epoch": 1.756533614749519, "grad_norm": 0.5445724789131907, "learning_rate": 4.38184806871803e-05, "loss": 2.671, "step": 28296 }, { "epoch": 1.7565956918492769, "grad_norm": 0.6055639868867723, "learning_rate": 4.381489688552671e-05, "loss": 2.6912, "step": 28297 }, { "epoch": 1.7566577689490348, "grad_norm": 0.5547050150940108, "learning_rate": 4.3811313116142285e-05, "loss": 2.7236, "step": 28298 }, { "epoch": 1.7567198460487927, "grad_norm": 0.7265504405343692, "learning_rate": 4.3807729379045757e-05, "loss": 2.7322, "step": 28299 }, { "epoch": 1.7567819231485506, "grad_norm": 0.5161281789255394, "learning_rate": 4.380414567425579e-05, "loss": 2.7104, "step": 28300 }, { "epoch": 1.7568440002483086, "grad_norm": 0.6706579407550262, "learning_rate": 4.38005620017911e-05, "loss": 2.7921, "step": 28301 }, { "epoch": 1.7569060773480663, "grad_norm": 0.508346746090564, "learning_rate": 4.3796978361670404e-05, "loss": 2.7949, "step": 28302 }, { "epoch": 1.7569681544478242, "grad_norm": 0.6091077414121985, "learning_rate": 4.379339475391236e-05, "loss": 2.7893, "step": 28303 }, { "epoch": 1.757030231547582, "grad_norm": 0.5364922324522204, "learning_rate": 4.3789811178535696e-05, "loss": 2.7545, "step": 28304 }, { "epoch": 1.75709230864734, "grad_norm": 0.6174898734787484, "learning_rate": 4.378622763555909e-05, "loss": 2.6758, "step": 28305 }, { "epoch": 1.757154385747098, "grad_norm": 0.5568805913505325, "learning_rate": 4.3782644125001236e-05, "loss": 2.7161, "step": 28306 }, { "epoch": 1.7572164628468558, "grad_norm": 0.6398693373038876, "learning_rate": 4.3779060646880844e-05, "loss": 2.7224, "step": 28307 }, { "epoch": 1.7572785399466135, "grad_norm": 0.6232162054151776, "learning_rate": 4.3775477201216605e-05, "loss": 2.6324, "step": 28308 }, { "epoch": 1.7573406170463715, "grad_norm": 0.672994789415115, "learning_rate": 4.377189378802721e-05, "loss": 2.6787, "step": 28309 }, { "epoch": 1.7574026941461294, "grad_norm": 0.5980062519531301, "learning_rate": 4.3768310407331354e-05, "loss": 2.7507, "step": 28310 }, { "epoch": 1.7574647712458873, "grad_norm": 0.6324514135916265, "learning_rate": 4.376472705914774e-05, "loss": 2.7056, "step": 28311 }, { "epoch": 1.7575268483456452, "grad_norm": 0.5525871053705099, "learning_rate": 4.3761143743495056e-05, "loss": 2.7748, "step": 28312 }, { "epoch": 1.7575889254454031, "grad_norm": 0.5785528352607459, "learning_rate": 4.3757560460392e-05, "loss": 2.7417, "step": 28313 }, { "epoch": 1.757651002545161, "grad_norm": 0.5279966480553319, "learning_rate": 4.375397720985726e-05, "loss": 2.7673, "step": 28314 }, { "epoch": 1.757713079644919, "grad_norm": 0.5685657768779891, "learning_rate": 4.375039399190955e-05, "loss": 2.8266, "step": 28315 }, { "epoch": 1.7577751567446769, "grad_norm": 0.5880485145218598, "learning_rate": 4.374681080656754e-05, "loss": 2.8067, "step": 28316 }, { "epoch": 1.7578372338444348, "grad_norm": 0.6619261541581921, "learning_rate": 4.3743227653849945e-05, "loss": 2.8621, "step": 28317 }, { "epoch": 1.7578993109441927, "grad_norm": 0.6250258678805147, "learning_rate": 4.3739644533775434e-05, "loss": 2.7254, "step": 28318 }, { "epoch": 1.7579613880439506, "grad_norm": 0.611213746507204, "learning_rate": 4.3736061446362744e-05, "loss": 2.7117, "step": 28319 }, { "epoch": 1.7580234651437086, "grad_norm": 0.6085903054998075, "learning_rate": 4.373247839163053e-05, "loss": 2.7548, "step": 28320 }, { "epoch": 1.7580855422434665, "grad_norm": 0.5329834202447562, "learning_rate": 4.372889536959749e-05, "loss": 2.6961, "step": 28321 }, { "epoch": 1.7581476193432244, "grad_norm": 0.5968498574199125, "learning_rate": 4.372531238028233e-05, "loss": 2.8685, "step": 28322 }, { "epoch": 1.7582096964429823, "grad_norm": 0.5405926755381444, "learning_rate": 4.372172942370374e-05, "loss": 2.6438, "step": 28323 }, { "epoch": 1.7582717735427402, "grad_norm": 0.5800712148338283, "learning_rate": 4.3718146499880414e-05, "loss": 2.7445, "step": 28324 }, { "epoch": 1.7583338506424981, "grad_norm": 0.5239262078702102, "learning_rate": 4.3714563608831035e-05, "loss": 2.7989, "step": 28325 }, { "epoch": 1.7583959277422558, "grad_norm": 0.5976204954933846, "learning_rate": 4.3710980750574306e-05, "loss": 2.6902, "step": 28326 }, { "epoch": 1.7584580048420138, "grad_norm": 0.5204379551943816, "learning_rate": 4.370739792512892e-05, "loss": 2.6585, "step": 28327 }, { "epoch": 1.7585200819417717, "grad_norm": 0.5332597416033983, "learning_rate": 4.370381513251357e-05, "loss": 2.7352, "step": 28328 }, { "epoch": 1.7585821590415296, "grad_norm": 0.5434823111714906, "learning_rate": 4.370023237274694e-05, "loss": 2.7826, "step": 28329 }, { "epoch": 1.7586442361412875, "grad_norm": 0.5178489696158368, "learning_rate": 4.3696649645847726e-05, "loss": 2.6909, "step": 28330 }, { "epoch": 1.7587063132410454, "grad_norm": 0.6090914726901714, "learning_rate": 4.3693066951834617e-05, "loss": 2.7603, "step": 28331 }, { "epoch": 1.7587683903408031, "grad_norm": 0.5884888858815229, "learning_rate": 4.368948429072632e-05, "loss": 2.8295, "step": 28332 }, { "epoch": 1.758830467440561, "grad_norm": 0.6555913617256834, "learning_rate": 4.36859016625415e-05, "loss": 2.6944, "step": 28333 }, { "epoch": 1.758892544540319, "grad_norm": 0.59017848943772, "learning_rate": 4.3682319067298874e-05, "loss": 2.8093, "step": 28334 }, { "epoch": 1.7589546216400769, "grad_norm": 0.5812674765882039, "learning_rate": 4.3678736505017114e-05, "loss": 2.7422, "step": 28335 }, { "epoch": 1.7590166987398348, "grad_norm": 0.6387965695972215, "learning_rate": 4.3675153975714934e-05, "loss": 2.8088, "step": 28336 }, { "epoch": 1.7590787758395927, "grad_norm": 0.592157310783237, "learning_rate": 4.3671571479410996e-05, "loss": 2.7511, "step": 28337 }, { "epoch": 1.7591408529393506, "grad_norm": 0.5775518927440684, "learning_rate": 4.3667989016124024e-05, "loss": 2.83, "step": 28338 }, { "epoch": 1.7592029300391085, "grad_norm": 0.5585567222006986, "learning_rate": 4.366440658587268e-05, "loss": 2.7665, "step": 28339 }, { "epoch": 1.7592650071388665, "grad_norm": 0.5678888057396738, "learning_rate": 4.366082418867567e-05, "loss": 2.7227, "step": 28340 }, { "epoch": 1.7593270842386244, "grad_norm": 0.5480101112873846, "learning_rate": 4.365724182455167e-05, "loss": 2.7678, "step": 28341 }, { "epoch": 1.7593891613383823, "grad_norm": 0.6026492025899426, "learning_rate": 4.3653659493519396e-05, "loss": 2.7021, "step": 28342 }, { "epoch": 1.7594512384381402, "grad_norm": 0.5262868391720659, "learning_rate": 4.36500771955975e-05, "loss": 2.6795, "step": 28343 }, { "epoch": 1.7595133155378981, "grad_norm": 0.6372726465109406, "learning_rate": 4.36464949308047e-05, "loss": 2.6512, "step": 28344 }, { "epoch": 1.759575392637656, "grad_norm": 0.5017980788820908, "learning_rate": 4.364291269915969e-05, "loss": 2.7251, "step": 28345 }, { "epoch": 1.759637469737414, "grad_norm": 0.5756741200218412, "learning_rate": 4.363933050068114e-05, "loss": 2.7604, "step": 28346 }, { "epoch": 1.759699546837172, "grad_norm": 0.5962709066136531, "learning_rate": 4.363574833538775e-05, "loss": 2.7477, "step": 28347 }, { "epoch": 1.7597616239369298, "grad_norm": 0.5486535442885475, "learning_rate": 4.36321662032982e-05, "loss": 2.7212, "step": 28348 }, { "epoch": 1.7598237010366877, "grad_norm": 0.61431324137116, "learning_rate": 4.362858410443119e-05, "loss": 2.7641, "step": 28349 }, { "epoch": 1.7598857781364454, "grad_norm": 0.5509777665420893, "learning_rate": 4.3625002038805394e-05, "loss": 2.8169, "step": 28350 }, { "epoch": 1.7599478552362033, "grad_norm": 0.6330231445489594, "learning_rate": 4.3621420006439514e-05, "loss": 2.7564, "step": 28351 }, { "epoch": 1.7600099323359613, "grad_norm": 0.5901913660664706, "learning_rate": 4.361783800735224e-05, "loss": 2.8085, "step": 28352 }, { "epoch": 1.7600720094357192, "grad_norm": 0.5506475570422926, "learning_rate": 4.361425604156225e-05, "loss": 2.6705, "step": 28353 }, { "epoch": 1.760134086535477, "grad_norm": 0.5836197853664274, "learning_rate": 4.3610674109088236e-05, "loss": 2.8084, "step": 28354 }, { "epoch": 1.760196163635235, "grad_norm": 0.5977165092058889, "learning_rate": 4.360709220994889e-05, "loss": 2.7391, "step": 28355 }, { "epoch": 1.7602582407349927, "grad_norm": 0.5412851481665384, "learning_rate": 4.360351034416289e-05, "loss": 2.7655, "step": 28356 }, { "epoch": 1.7603203178347506, "grad_norm": 0.5329294864950826, "learning_rate": 4.3599928511748934e-05, "loss": 2.7215, "step": 28357 }, { "epoch": 1.7603823949345085, "grad_norm": 0.5618433828655562, "learning_rate": 4.359634671272571e-05, "loss": 2.7236, "step": 28358 }, { "epoch": 1.7604444720342665, "grad_norm": 0.5440896080840554, "learning_rate": 4.3592764947111884e-05, "loss": 2.7167, "step": 28359 }, { "epoch": 1.7605065491340244, "grad_norm": 0.5491784223399088, "learning_rate": 4.358918321492618e-05, "loss": 2.7633, "step": 28360 }, { "epoch": 1.7605686262337823, "grad_norm": 0.5503188282019512, "learning_rate": 4.358560151618724e-05, "loss": 2.7438, "step": 28361 }, { "epoch": 1.7606307033335402, "grad_norm": 0.5772927970237316, "learning_rate": 4.358201985091379e-05, "loss": 2.7618, "step": 28362 }, { "epoch": 1.7606927804332981, "grad_norm": 0.5229003484492437, "learning_rate": 4.357843821912449e-05, "loss": 2.7705, "step": 28363 }, { "epoch": 1.760754857533056, "grad_norm": 0.57485853611147, "learning_rate": 4.357485662083805e-05, "loss": 2.7181, "step": 28364 }, { "epoch": 1.760816934632814, "grad_norm": 0.5373647186900083, "learning_rate": 4.357127505607312e-05, "loss": 2.7729, "step": 28365 }, { "epoch": 1.7608790117325719, "grad_norm": 0.5378698063785164, "learning_rate": 4.356769352484842e-05, "loss": 2.8152, "step": 28366 }, { "epoch": 1.7609410888323298, "grad_norm": 0.5410809529107861, "learning_rate": 4.356411202718263e-05, "loss": 2.749, "step": 28367 }, { "epoch": 1.7610031659320877, "grad_norm": 0.552569819781273, "learning_rate": 4.356053056309442e-05, "loss": 2.6428, "step": 28368 }, { "epoch": 1.7610652430318456, "grad_norm": 0.5442947826561559, "learning_rate": 4.35569491326025e-05, "loss": 2.7842, "step": 28369 }, { "epoch": 1.7611273201316036, "grad_norm": 0.5393021092498631, "learning_rate": 4.355336773572554e-05, "loss": 2.7463, "step": 28370 }, { "epoch": 1.7611893972313615, "grad_norm": 0.5314554490395336, "learning_rate": 4.354978637248221e-05, "loss": 2.7311, "step": 28371 }, { "epoch": 1.7612514743311194, "grad_norm": 0.5426555251521161, "learning_rate": 4.354620504289123e-05, "loss": 2.7378, "step": 28372 }, { "epoch": 1.7613135514308773, "grad_norm": 0.5662894292918252, "learning_rate": 4.3542623746971246e-05, "loss": 2.8241, "step": 28373 }, { "epoch": 1.761375628530635, "grad_norm": 0.5347948176438884, "learning_rate": 4.3539042484740977e-05, "loss": 2.7294, "step": 28374 }, { "epoch": 1.761437705630393, "grad_norm": 0.5377716974610024, "learning_rate": 4.353546125621908e-05, "loss": 2.7223, "step": 28375 }, { "epoch": 1.7614997827301508, "grad_norm": 0.5355507725991866, "learning_rate": 4.3531880061424255e-05, "loss": 2.8177, "step": 28376 }, { "epoch": 1.7615618598299088, "grad_norm": 0.4758909371505241, "learning_rate": 4.352829890037518e-05, "loss": 2.7379, "step": 28377 }, { "epoch": 1.7616239369296667, "grad_norm": 0.5581368932279137, "learning_rate": 4.352471777309055e-05, "loss": 2.6531, "step": 28378 }, { "epoch": 1.7616860140294246, "grad_norm": 0.4952474870343603, "learning_rate": 4.3521136679589026e-05, "loss": 2.8696, "step": 28379 }, { "epoch": 1.7617480911291823, "grad_norm": 0.5318441994845197, "learning_rate": 4.3517555619889314e-05, "loss": 2.8629, "step": 28380 }, { "epoch": 1.7618101682289402, "grad_norm": 0.519322145299696, "learning_rate": 4.351397459401009e-05, "loss": 2.781, "step": 28381 }, { "epoch": 1.7618722453286981, "grad_norm": 0.5562804813820079, "learning_rate": 4.351039360197003e-05, "loss": 2.7621, "step": 28382 }, { "epoch": 1.761934322428456, "grad_norm": 0.5074225762694977, "learning_rate": 4.350681264378783e-05, "loss": 2.7301, "step": 28383 }, { "epoch": 1.761996399528214, "grad_norm": 0.6335478383844237, "learning_rate": 4.350323171948215e-05, "loss": 2.6978, "step": 28384 }, { "epoch": 1.7620584766279719, "grad_norm": 0.5204650689587249, "learning_rate": 4.3499650829071705e-05, "loss": 2.7795, "step": 28385 }, { "epoch": 1.7621205537277298, "grad_norm": 0.6219729792024622, "learning_rate": 4.3496069972575136e-05, "loss": 2.7891, "step": 28386 }, { "epoch": 1.7621826308274877, "grad_norm": 0.5812050761018236, "learning_rate": 4.3492489150011175e-05, "loss": 2.7584, "step": 28387 }, { "epoch": 1.7622447079272456, "grad_norm": 0.5582970851784523, "learning_rate": 4.348890836139845e-05, "loss": 2.7365, "step": 28388 }, { "epoch": 1.7623067850270036, "grad_norm": 0.6139884564358351, "learning_rate": 4.3485327606755696e-05, "loss": 2.7799, "step": 28389 }, { "epoch": 1.7623688621267615, "grad_norm": 0.5368843578282466, "learning_rate": 4.3481746886101556e-05, "loss": 2.8192, "step": 28390 }, { "epoch": 1.7624309392265194, "grad_norm": 0.6372362385921729, "learning_rate": 4.347816619945473e-05, "loss": 2.7544, "step": 28391 }, { "epoch": 1.7624930163262773, "grad_norm": 0.5340970270057332, "learning_rate": 4.3474585546833884e-05, "loss": 2.7959, "step": 28392 }, { "epoch": 1.7625550934260352, "grad_norm": 0.6566632422932381, "learning_rate": 4.3471004928257724e-05, "loss": 2.6722, "step": 28393 }, { "epoch": 1.7626171705257931, "grad_norm": 0.5412727598730578, "learning_rate": 4.346742434374492e-05, "loss": 2.7708, "step": 28394 }, { "epoch": 1.762679247625551, "grad_norm": 0.6307870687613705, "learning_rate": 4.346384379331413e-05, "loss": 2.8053, "step": 28395 }, { "epoch": 1.762741324725309, "grad_norm": 0.6083025910391459, "learning_rate": 4.346026327698406e-05, "loss": 2.8252, "step": 28396 }, { "epoch": 1.762803401825067, "grad_norm": 0.6160762295032405, "learning_rate": 4.3456682794773384e-05, "loss": 2.679, "step": 28397 }, { "epoch": 1.7628654789248246, "grad_norm": 0.5984526171654846, "learning_rate": 4.345310234670077e-05, "loss": 2.8032, "step": 28398 }, { "epoch": 1.7629275560245825, "grad_norm": 0.6291132521113239, "learning_rate": 4.344952193278493e-05, "loss": 2.6689, "step": 28399 }, { "epoch": 1.7629896331243404, "grad_norm": 0.5556517802458212, "learning_rate": 4.344594155304451e-05, "loss": 2.8107, "step": 28400 }, { "epoch": 1.7630517102240983, "grad_norm": 0.6012749665959334, "learning_rate": 4.344236120749822e-05, "loss": 2.7704, "step": 28401 }, { "epoch": 1.7631137873238563, "grad_norm": 0.5673520051927634, "learning_rate": 4.343878089616471e-05, "loss": 2.7447, "step": 28402 }, { "epoch": 1.7631758644236142, "grad_norm": 0.5890166186366746, "learning_rate": 4.343520061906269e-05, "loss": 2.7652, "step": 28403 }, { "epoch": 1.7632379415233719, "grad_norm": 0.5667885641044749, "learning_rate": 4.34316203762108e-05, "loss": 2.7218, "step": 28404 }, { "epoch": 1.7633000186231298, "grad_norm": 0.6035994190697374, "learning_rate": 4.342804016762776e-05, "loss": 2.7841, "step": 28405 }, { "epoch": 1.7633620957228877, "grad_norm": 0.6415783835859018, "learning_rate": 4.3424459993332226e-05, "loss": 2.7214, "step": 28406 }, { "epoch": 1.7634241728226456, "grad_norm": 0.5603163773614749, "learning_rate": 4.3420879853342874e-05, "loss": 2.6617, "step": 28407 }, { "epoch": 1.7634862499224035, "grad_norm": 0.6983344597302854, "learning_rate": 4.3417299747678394e-05, "loss": 2.7692, "step": 28408 }, { "epoch": 1.7635483270221615, "grad_norm": 0.5762106155043938, "learning_rate": 4.3413719676357454e-05, "loss": 2.7694, "step": 28409 }, { "epoch": 1.7636104041219194, "grad_norm": 0.6351445402260122, "learning_rate": 4.341013963939875e-05, "loss": 2.768, "step": 28410 }, { "epoch": 1.7636724812216773, "grad_norm": 0.5512721532146044, "learning_rate": 4.340655963682093e-05, "loss": 2.7489, "step": 28411 }, { "epoch": 1.7637345583214352, "grad_norm": 0.5666746621643047, "learning_rate": 4.34029796686427e-05, "loss": 2.7435, "step": 28412 }, { "epoch": 1.7637966354211931, "grad_norm": 0.5881428225144748, "learning_rate": 4.3399399734882716e-05, "loss": 2.6963, "step": 28413 }, { "epoch": 1.763858712520951, "grad_norm": 0.5504754753037632, "learning_rate": 4.3395819835559683e-05, "loss": 2.8067, "step": 28414 }, { "epoch": 1.763920789620709, "grad_norm": 0.6441421490264315, "learning_rate": 4.3392239970692244e-05, "loss": 2.6976, "step": 28415 }, { "epoch": 1.763982866720467, "grad_norm": 0.5086035216252688, "learning_rate": 4.3388660140299106e-05, "loss": 2.7441, "step": 28416 }, { "epoch": 1.7640449438202248, "grad_norm": 0.5503391087439062, "learning_rate": 4.338508034439892e-05, "loss": 2.7455, "step": 28417 }, { "epoch": 1.7641070209199827, "grad_norm": 0.55426715988933, "learning_rate": 4.338150058301039e-05, "loss": 2.6126, "step": 28418 }, { "epoch": 1.7641690980197406, "grad_norm": 0.5653833030952942, "learning_rate": 4.337792085615218e-05, "loss": 2.7614, "step": 28419 }, { "epoch": 1.7642311751194986, "grad_norm": 0.5920469806803236, "learning_rate": 4.3374341163842946e-05, "loss": 2.7719, "step": 28420 }, { "epoch": 1.7642932522192565, "grad_norm": 0.6133948056972789, "learning_rate": 4.33707615061014e-05, "loss": 2.7568, "step": 28421 }, { "epoch": 1.7643553293190142, "grad_norm": 0.5860294778475963, "learning_rate": 4.3367181882946184e-05, "loss": 2.6834, "step": 28422 }, { "epoch": 1.764417406418772, "grad_norm": 0.6591204355242064, "learning_rate": 4.3363602294396e-05, "loss": 2.807, "step": 28423 }, { "epoch": 1.76447948351853, "grad_norm": 0.5999029975203228, "learning_rate": 4.336002274046951e-05, "loss": 2.807, "step": 28424 }, { "epoch": 1.764541560618288, "grad_norm": 0.6335961234804376, "learning_rate": 4.33564432211854e-05, "loss": 2.784, "step": 28425 }, { "epoch": 1.7646036377180458, "grad_norm": 0.6121653226347281, "learning_rate": 4.335286373656233e-05, "loss": 2.7079, "step": 28426 }, { "epoch": 1.7646657148178038, "grad_norm": 0.6085076884580941, "learning_rate": 4.334928428661899e-05, "loss": 2.7996, "step": 28427 }, { "epoch": 1.7647277919175615, "grad_norm": 0.6536155911525489, "learning_rate": 4.334570487137405e-05, "loss": 2.7747, "step": 28428 }, { "epoch": 1.7647898690173194, "grad_norm": 0.5907309724711414, "learning_rate": 4.334212549084619e-05, "loss": 2.7921, "step": 28429 }, { "epoch": 1.7648519461170773, "grad_norm": 0.5893031289058202, "learning_rate": 4.3338546145054065e-05, "loss": 2.841, "step": 28430 }, { "epoch": 1.7649140232168352, "grad_norm": 0.5834057342920972, "learning_rate": 4.333496683401635e-05, "loss": 2.7882, "step": 28431 }, { "epoch": 1.7649761003165931, "grad_norm": 0.5487387099006165, "learning_rate": 4.333138755775174e-05, "loss": 2.8874, "step": 28432 }, { "epoch": 1.765038177416351, "grad_norm": 0.5978138535080946, "learning_rate": 4.332780831627891e-05, "loss": 2.8579, "step": 28433 }, { "epoch": 1.765100254516109, "grad_norm": 0.5563304371485094, "learning_rate": 4.332422910961651e-05, "loss": 2.7362, "step": 28434 }, { "epoch": 1.7651623316158669, "grad_norm": 0.6094722670175061, "learning_rate": 4.3320649937783245e-05, "loss": 2.8181, "step": 28435 }, { "epoch": 1.7652244087156248, "grad_norm": 0.5365555838670713, "learning_rate": 4.331707080079775e-05, "loss": 2.8054, "step": 28436 }, { "epoch": 1.7652864858153827, "grad_norm": 0.6109211453641481, "learning_rate": 4.3313491698678745e-05, "loss": 2.8512, "step": 28437 }, { "epoch": 1.7653485629151406, "grad_norm": 0.507201604923182, "learning_rate": 4.330991263144486e-05, "loss": 2.6913, "step": 28438 }, { "epoch": 1.7654106400148986, "grad_norm": 0.5667343767702568, "learning_rate": 4.330633359911479e-05, "loss": 2.7042, "step": 28439 }, { "epoch": 1.7654727171146565, "grad_norm": 0.6007527685669132, "learning_rate": 4.33027546017072e-05, "loss": 2.7612, "step": 28440 }, { "epoch": 1.7655347942144144, "grad_norm": 0.5394209071102254, "learning_rate": 4.329917563924076e-05, "loss": 2.7855, "step": 28441 }, { "epoch": 1.7655968713141723, "grad_norm": 0.6542727337838821, "learning_rate": 4.329559671173415e-05, "loss": 2.8166, "step": 28442 }, { "epoch": 1.7656589484139302, "grad_norm": 0.5596253799052053, "learning_rate": 4.329201781920605e-05, "loss": 2.7089, "step": 28443 }, { "epoch": 1.7657210255136881, "grad_norm": 0.6388511056254973, "learning_rate": 4.328843896167512e-05, "loss": 2.7664, "step": 28444 }, { "epoch": 1.765783102613446, "grad_norm": 0.6275504288414274, "learning_rate": 4.328486013916002e-05, "loss": 2.7112, "step": 28445 }, { "epoch": 1.7658451797132038, "grad_norm": 0.526713185016449, "learning_rate": 4.328128135167945e-05, "loss": 2.6662, "step": 28446 }, { "epoch": 1.7659072568129617, "grad_norm": 0.5655073465043794, "learning_rate": 4.3277702599252045e-05, "loss": 2.7714, "step": 28447 }, { "epoch": 1.7659693339127196, "grad_norm": 0.5611726677002234, "learning_rate": 4.327412388189652e-05, "loss": 2.7009, "step": 28448 }, { "epoch": 1.7660314110124775, "grad_norm": 0.6046224771083825, "learning_rate": 4.3270545199631505e-05, "loss": 2.7479, "step": 28449 }, { "epoch": 1.7660934881122354, "grad_norm": 0.5382911027911819, "learning_rate": 4.3266966552475706e-05, "loss": 2.7872, "step": 28450 }, { "epoch": 1.7661555652119934, "grad_norm": 0.5792833528593178, "learning_rate": 4.326338794044777e-05, "loss": 2.7239, "step": 28451 }, { "epoch": 1.766217642311751, "grad_norm": 0.5299429241073343, "learning_rate": 4.3259809363566375e-05, "loss": 2.6459, "step": 28452 }, { "epoch": 1.766279719411509, "grad_norm": 0.5310666104346531, "learning_rate": 4.3256230821850186e-05, "loss": 2.7537, "step": 28453 }, { "epoch": 1.7663417965112669, "grad_norm": 0.5817435404751279, "learning_rate": 4.325265231531788e-05, "loss": 2.7869, "step": 28454 }, { "epoch": 1.7664038736110248, "grad_norm": 0.5587052720790859, "learning_rate": 4.324907384398813e-05, "loss": 2.8316, "step": 28455 }, { "epoch": 1.7664659507107827, "grad_norm": 0.5828597553312166, "learning_rate": 4.32454954078796e-05, "loss": 2.7656, "step": 28456 }, { "epoch": 1.7665280278105406, "grad_norm": 0.5715520269853711, "learning_rate": 4.3241917007010956e-05, "loss": 2.7309, "step": 28457 }, { "epoch": 1.7665901049102986, "grad_norm": 0.59419702968176, "learning_rate": 4.323833864140087e-05, "loss": 2.8159, "step": 28458 }, { "epoch": 1.7666521820100565, "grad_norm": 0.5236833763538397, "learning_rate": 4.3234760311068024e-05, "loss": 2.7527, "step": 28459 }, { "epoch": 1.7667142591098144, "grad_norm": 0.6113931542077863, "learning_rate": 4.323118201603106e-05, "loss": 2.7027, "step": 28460 }, { "epoch": 1.7667763362095723, "grad_norm": 0.5931932449070667, "learning_rate": 4.322760375630868e-05, "loss": 2.7939, "step": 28461 }, { "epoch": 1.7668384133093302, "grad_norm": 0.5793158920983899, "learning_rate": 4.322402553191951e-05, "loss": 2.7509, "step": 28462 }, { "epoch": 1.7669004904090881, "grad_norm": 0.6293643207269066, "learning_rate": 4.322044734288226e-05, "loss": 2.689, "step": 28463 }, { "epoch": 1.766962567508846, "grad_norm": 0.5798379163106816, "learning_rate": 4.321686918921557e-05, "loss": 2.7885, "step": 28464 }, { "epoch": 1.767024644608604, "grad_norm": 0.6274451110967955, "learning_rate": 4.3213291070938126e-05, "loss": 2.775, "step": 28465 }, { "epoch": 1.767086721708362, "grad_norm": 0.5876488679616438, "learning_rate": 4.32097129880686e-05, "loss": 2.899, "step": 28466 }, { "epoch": 1.7671487988081198, "grad_norm": 0.572274504192328, "learning_rate": 4.320613494062564e-05, "loss": 2.7812, "step": 28467 }, { "epoch": 1.7672108759078777, "grad_norm": 0.6294984813047698, "learning_rate": 4.3202556928627934e-05, "loss": 2.7391, "step": 28468 }, { "epoch": 1.7672729530076356, "grad_norm": 0.5376240709071057, "learning_rate": 4.319897895209414e-05, "loss": 2.7239, "step": 28469 }, { "epoch": 1.7673350301073933, "grad_norm": 0.6206168016889119, "learning_rate": 4.319540101104291e-05, "loss": 2.7405, "step": 28470 }, { "epoch": 1.7673971072071513, "grad_norm": 0.5601096757709261, "learning_rate": 4.319182310549294e-05, "loss": 2.7255, "step": 28471 }, { "epoch": 1.7674591843069092, "grad_norm": 0.6078857171841422, "learning_rate": 4.318824523546286e-05, "loss": 2.717, "step": 28472 }, { "epoch": 1.767521261406667, "grad_norm": 0.6054591516533806, "learning_rate": 4.318466740097138e-05, "loss": 2.7921, "step": 28473 }, { "epoch": 1.767583338506425, "grad_norm": 0.5514001277714506, "learning_rate": 4.3181089602037126e-05, "loss": 2.8589, "step": 28474 }, { "epoch": 1.7676454156061827, "grad_norm": 0.6219237437865511, "learning_rate": 4.31775118386788e-05, "loss": 2.7547, "step": 28475 }, { "epoch": 1.7677074927059406, "grad_norm": 0.6175478112865865, "learning_rate": 4.317393411091504e-05, "loss": 2.7595, "step": 28476 }, { "epoch": 1.7677695698056985, "grad_norm": 0.6422109511425672, "learning_rate": 4.3170356418764526e-05, "loss": 2.7498, "step": 28477 }, { "epoch": 1.7678316469054565, "grad_norm": 0.6040530234914255, "learning_rate": 4.316677876224592e-05, "loss": 2.7969, "step": 28478 }, { "epoch": 1.7678937240052144, "grad_norm": 0.6008408664446288, "learning_rate": 4.316320114137789e-05, "loss": 2.7811, "step": 28479 }, { "epoch": 1.7679558011049723, "grad_norm": 0.6173506412493196, "learning_rate": 4.315962355617911e-05, "loss": 2.7562, "step": 28480 }, { "epoch": 1.7680178782047302, "grad_norm": 0.5508925413416536, "learning_rate": 4.315604600666821e-05, "loss": 2.7663, "step": 28481 }, { "epoch": 1.7680799553044881, "grad_norm": 0.632304729617914, "learning_rate": 4.3152468492863904e-05, "loss": 2.7638, "step": 28482 }, { "epoch": 1.768142032404246, "grad_norm": 0.5175167407216746, "learning_rate": 4.314889101478481e-05, "loss": 2.7869, "step": 28483 }, { "epoch": 1.768204109504004, "grad_norm": 0.613170200668828, "learning_rate": 4.314531357244963e-05, "loss": 2.6876, "step": 28484 }, { "epoch": 1.768266186603762, "grad_norm": 0.5742964140602559, "learning_rate": 4.314173616587701e-05, "loss": 2.7978, "step": 28485 }, { "epoch": 1.7683282637035198, "grad_norm": 0.5982618816246366, "learning_rate": 4.3138158795085614e-05, "loss": 2.7544, "step": 28486 }, { "epoch": 1.7683903408032777, "grad_norm": 0.5292321478761436, "learning_rate": 4.31345814600941e-05, "loss": 2.8371, "step": 28487 }, { "epoch": 1.7684524179030356, "grad_norm": 0.5938655088401334, "learning_rate": 4.3131004160921164e-05, "loss": 2.7184, "step": 28488 }, { "epoch": 1.7685144950027936, "grad_norm": 0.6097370930361762, "learning_rate": 4.3127426897585426e-05, "loss": 2.7084, "step": 28489 }, { "epoch": 1.7685765721025515, "grad_norm": 0.5104711626764734, "learning_rate": 4.312384967010559e-05, "loss": 2.7773, "step": 28490 }, { "epoch": 1.7686386492023094, "grad_norm": 0.6109252674800135, "learning_rate": 4.312027247850029e-05, "loss": 2.7518, "step": 28491 }, { "epoch": 1.7687007263020673, "grad_norm": 0.5291779162745528, "learning_rate": 4.311669532278819e-05, "loss": 2.7172, "step": 28492 }, { "epoch": 1.768762803401825, "grad_norm": 0.5386488336250739, "learning_rate": 4.311311820298798e-05, "loss": 2.7367, "step": 28493 }, { "epoch": 1.768824880501583, "grad_norm": 0.4992778898260847, "learning_rate": 4.310954111911829e-05, "loss": 2.745, "step": 28494 }, { "epoch": 1.7688869576013408, "grad_norm": 0.5910529569242899, "learning_rate": 4.310596407119779e-05, "loss": 2.7979, "step": 28495 }, { "epoch": 1.7689490347010988, "grad_norm": 0.5443326711128927, "learning_rate": 4.310238705924515e-05, "loss": 2.7621, "step": 28496 }, { "epoch": 1.7690111118008567, "grad_norm": 0.5574309243401288, "learning_rate": 4.3098810083279035e-05, "loss": 2.714, "step": 28497 }, { "epoch": 1.7690731889006146, "grad_norm": 0.5336529577423652, "learning_rate": 4.3095233143318114e-05, "loss": 2.6384, "step": 28498 }, { "epoch": 1.7691352660003723, "grad_norm": 0.6193624153359953, "learning_rate": 4.3091656239381024e-05, "loss": 2.6857, "step": 28499 }, { "epoch": 1.7691973431001302, "grad_norm": 0.4921023582809737, "learning_rate": 4.308807937148646e-05, "loss": 2.5797, "step": 28500 }, { "epoch": 1.7692594201998881, "grad_norm": 0.6154096807764768, "learning_rate": 4.3084502539653035e-05, "loss": 2.7006, "step": 28501 }, { "epoch": 1.769321497299646, "grad_norm": 0.5542032250078901, "learning_rate": 4.308092574389947e-05, "loss": 2.656, "step": 28502 }, { "epoch": 1.769383574399404, "grad_norm": 0.5420753483937985, "learning_rate": 4.307734898424437e-05, "loss": 2.6683, "step": 28503 }, { "epoch": 1.7694456514991619, "grad_norm": 0.5996485893026168, "learning_rate": 4.307377226070644e-05, "loss": 2.8143, "step": 28504 }, { "epoch": 1.7695077285989198, "grad_norm": 0.606904450359046, "learning_rate": 4.3070195573304316e-05, "loss": 2.7459, "step": 28505 }, { "epoch": 1.7695698056986777, "grad_norm": 0.5507956158002262, "learning_rate": 4.3066618922056656e-05, "loss": 2.6304, "step": 28506 }, { "epoch": 1.7696318827984356, "grad_norm": 0.6475735849742998, "learning_rate": 4.306304230698214e-05, "loss": 2.7471, "step": 28507 }, { "epoch": 1.7696939598981936, "grad_norm": 0.5630629689285268, "learning_rate": 4.30594657280994e-05, "loss": 2.8302, "step": 28508 }, { "epoch": 1.7697560369979515, "grad_norm": 0.6691000432292245, "learning_rate": 4.305588918542713e-05, "loss": 2.795, "step": 28509 }, { "epoch": 1.7698181140977094, "grad_norm": 0.5799830451904252, "learning_rate": 4.305231267898396e-05, "loss": 2.6855, "step": 28510 }, { "epoch": 1.7698801911974673, "grad_norm": 0.5971225524749969, "learning_rate": 4.304873620878857e-05, "loss": 2.742, "step": 28511 }, { "epoch": 1.7699422682972252, "grad_norm": 0.5582557936340852, "learning_rate": 4.30451597748596e-05, "loss": 2.8462, "step": 28512 }, { "epoch": 1.7700043453969831, "grad_norm": 0.5623133137641828, "learning_rate": 4.304158337721573e-05, "loss": 2.7831, "step": 28513 }, { "epoch": 1.770066422496741, "grad_norm": 0.5508773936234164, "learning_rate": 4.30380070158756e-05, "loss": 2.7764, "step": 28514 }, { "epoch": 1.770128499596499, "grad_norm": 0.563549995595854, "learning_rate": 4.303443069085789e-05, "loss": 2.7798, "step": 28515 }, { "epoch": 1.770190576696257, "grad_norm": 0.6269981867202749, "learning_rate": 4.303085440218124e-05, "loss": 2.662, "step": 28516 }, { "epoch": 1.7702526537960146, "grad_norm": 0.5456432640103843, "learning_rate": 4.3027278149864306e-05, "loss": 2.7344, "step": 28517 }, { "epoch": 1.7703147308957725, "grad_norm": 0.619872157429391, "learning_rate": 4.302370193392577e-05, "loss": 2.7778, "step": 28518 }, { "epoch": 1.7703768079955304, "grad_norm": 0.5281917958651945, "learning_rate": 4.302012575438426e-05, "loss": 2.6814, "step": 28519 }, { "epoch": 1.7704388850952884, "grad_norm": 0.5715288490413093, "learning_rate": 4.301654961125846e-05, "loss": 2.6833, "step": 28520 }, { "epoch": 1.7705009621950463, "grad_norm": 0.5715317691490047, "learning_rate": 4.3012973504567e-05, "loss": 2.7451, "step": 28521 }, { "epoch": 1.7705630392948042, "grad_norm": 0.5796653743336102, "learning_rate": 4.300939743432857e-05, "loss": 2.7617, "step": 28522 }, { "epoch": 1.7706251163945619, "grad_norm": 0.6215199861328167, "learning_rate": 4.3005821400561804e-05, "loss": 2.8084, "step": 28523 }, { "epoch": 1.7706871934943198, "grad_norm": 0.5578435114513355, "learning_rate": 4.3002245403285366e-05, "loss": 2.7222, "step": 28524 }, { "epoch": 1.7707492705940777, "grad_norm": 0.5197682521288663, "learning_rate": 4.299866944251791e-05, "loss": 2.7237, "step": 28525 }, { "epoch": 1.7708113476938356, "grad_norm": 0.5672883367096327, "learning_rate": 4.29950935182781e-05, "loss": 2.6693, "step": 28526 }, { "epoch": 1.7708734247935936, "grad_norm": 0.527868165535825, "learning_rate": 4.299151763058458e-05, "loss": 2.7818, "step": 28527 }, { "epoch": 1.7709355018933515, "grad_norm": 0.5625361324997149, "learning_rate": 4.298794177945602e-05, "loss": 2.6057, "step": 28528 }, { "epoch": 1.7709975789931094, "grad_norm": 0.5513079134044652, "learning_rate": 4.298436596491105e-05, "loss": 2.7257, "step": 28529 }, { "epoch": 1.7710596560928673, "grad_norm": 0.550727192275297, "learning_rate": 4.298079018696838e-05, "loss": 2.8655, "step": 28530 }, { "epoch": 1.7711217331926252, "grad_norm": 0.5639802212290708, "learning_rate": 4.297721444564661e-05, "loss": 2.7585, "step": 28531 }, { "epoch": 1.7711838102923831, "grad_norm": 0.5576507504812749, "learning_rate": 4.297363874096443e-05, "loss": 2.7751, "step": 28532 }, { "epoch": 1.771245887392141, "grad_norm": 0.5109496707224978, "learning_rate": 4.297006307294046e-05, "loss": 2.739, "step": 28533 }, { "epoch": 1.771307964491899, "grad_norm": 0.5280059612580656, "learning_rate": 4.2966487441593404e-05, "loss": 2.6992, "step": 28534 }, { "epoch": 1.771370041591657, "grad_norm": 0.5721360929054242, "learning_rate": 4.296291184694187e-05, "loss": 2.8171, "step": 28535 }, { "epoch": 1.7714321186914148, "grad_norm": 0.585141289472642, "learning_rate": 4.295933628900455e-05, "loss": 2.8226, "step": 28536 }, { "epoch": 1.7714941957911727, "grad_norm": 0.5578212064175784, "learning_rate": 4.2955760767800066e-05, "loss": 2.6833, "step": 28537 }, { "epoch": 1.7715562728909307, "grad_norm": 0.6160219268379731, "learning_rate": 4.295218528334711e-05, "loss": 2.699, "step": 28538 }, { "epoch": 1.7716183499906886, "grad_norm": 0.5516453991369887, "learning_rate": 4.294860983566429e-05, "loss": 2.6757, "step": 28539 }, { "epoch": 1.7716804270904465, "grad_norm": 0.5785950476755233, "learning_rate": 4.29450344247703e-05, "loss": 2.7068, "step": 28540 }, { "epoch": 1.7717425041902042, "grad_norm": 0.5356553805981635, "learning_rate": 4.294145905068378e-05, "loss": 2.741, "step": 28541 }, { "epoch": 1.771804581289962, "grad_norm": 0.5503518073943584, "learning_rate": 4.2937883713423374e-05, "loss": 2.7684, "step": 28542 }, { "epoch": 1.77186665838972, "grad_norm": 0.5367365427413475, "learning_rate": 4.2934308413007744e-05, "loss": 2.6531, "step": 28543 }, { "epoch": 1.771928735489478, "grad_norm": 0.5880022671665548, "learning_rate": 4.293073314945554e-05, "loss": 2.7327, "step": 28544 }, { "epoch": 1.7719908125892359, "grad_norm": 0.567048964382713, "learning_rate": 4.292715792278542e-05, "loss": 2.767, "step": 28545 }, { "epoch": 1.7720528896889938, "grad_norm": 0.6310280022370155, "learning_rate": 4.292358273301603e-05, "loss": 2.7627, "step": 28546 }, { "epoch": 1.7721149667887515, "grad_norm": 0.6130561901603556, "learning_rate": 4.292000758016604e-05, "loss": 2.7541, "step": 28547 }, { "epoch": 1.7721770438885094, "grad_norm": 0.5488052363391249, "learning_rate": 4.291643246425407e-05, "loss": 2.7136, "step": 28548 }, { "epoch": 1.7722391209882673, "grad_norm": 0.6094598222043541, "learning_rate": 4.291285738529881e-05, "loss": 2.6988, "step": 28549 }, { "epoch": 1.7723011980880252, "grad_norm": 0.6290324303707236, "learning_rate": 4.290928234331888e-05, "loss": 2.7846, "step": 28550 }, { "epoch": 1.7723632751877831, "grad_norm": 0.5824419147362211, "learning_rate": 4.290570733833295e-05, "loss": 2.7306, "step": 28551 }, { "epoch": 1.772425352287541, "grad_norm": 0.6460574258651375, "learning_rate": 4.290213237035966e-05, "loss": 2.6929, "step": 28552 }, { "epoch": 1.772487429387299, "grad_norm": 0.5395418330352446, "learning_rate": 4.2898557439417686e-05, "loss": 2.8175, "step": 28553 }, { "epoch": 1.772549506487057, "grad_norm": 0.6051160708128949, "learning_rate": 4.289498254552565e-05, "loss": 2.7815, "step": 28554 }, { "epoch": 1.7726115835868148, "grad_norm": 0.6334553267497441, "learning_rate": 4.2891407688702215e-05, "loss": 2.7253, "step": 28555 }, { "epoch": 1.7726736606865727, "grad_norm": 0.5721803412830792, "learning_rate": 4.288783286896604e-05, "loss": 2.6736, "step": 28556 }, { "epoch": 1.7727357377863306, "grad_norm": 0.5505058195099308, "learning_rate": 4.288425808633575e-05, "loss": 2.72, "step": 28557 }, { "epoch": 1.7727978148860886, "grad_norm": 0.6067463838511163, "learning_rate": 4.288068334083003e-05, "loss": 2.6862, "step": 28558 }, { "epoch": 1.7728598919858465, "grad_norm": 0.5577760326528964, "learning_rate": 4.28771086324675e-05, "loss": 2.7335, "step": 28559 }, { "epoch": 1.7729219690856044, "grad_norm": 0.6311737316626981, "learning_rate": 4.287353396126684e-05, "loss": 2.7577, "step": 28560 }, { "epoch": 1.7729840461853623, "grad_norm": 0.5917894218151748, "learning_rate": 4.286995932724667e-05, "loss": 2.7148, "step": 28561 }, { "epoch": 1.7730461232851202, "grad_norm": 0.6376712905541994, "learning_rate": 4.286638473042564e-05, "loss": 2.7296, "step": 28562 }, { "epoch": 1.7731082003848782, "grad_norm": 0.6274292465991006, "learning_rate": 4.286281017082243e-05, "loss": 2.6932, "step": 28563 }, { "epoch": 1.773170277484636, "grad_norm": 0.5467429137797644, "learning_rate": 4.285923564845566e-05, "loss": 2.8315, "step": 28564 }, { "epoch": 1.7732323545843938, "grad_norm": 0.6261842237410242, "learning_rate": 4.285566116334401e-05, "loss": 2.787, "step": 28565 }, { "epoch": 1.7732944316841517, "grad_norm": 0.5958489161456648, "learning_rate": 4.285208671550611e-05, "loss": 2.8173, "step": 28566 }, { "epoch": 1.7733565087839096, "grad_norm": 0.5815618529564703, "learning_rate": 4.284851230496059e-05, "loss": 2.7338, "step": 28567 }, { "epoch": 1.7734185858836675, "grad_norm": 0.5241136027597391, "learning_rate": 4.2844937931726133e-05, "loss": 2.8188, "step": 28568 }, { "epoch": 1.7734806629834254, "grad_norm": 0.5820128930800507, "learning_rate": 4.2841363595821364e-05, "loss": 2.6713, "step": 28569 }, { "epoch": 1.7735427400831834, "grad_norm": 0.6128098930058329, "learning_rate": 4.2837789297264954e-05, "loss": 2.6893, "step": 28570 }, { "epoch": 1.773604817182941, "grad_norm": 0.6418328877036978, "learning_rate": 4.283421503607552e-05, "loss": 2.8137, "step": 28571 }, { "epoch": 1.773666894282699, "grad_norm": 0.566823003628064, "learning_rate": 4.283064081227173e-05, "loss": 2.7541, "step": 28572 }, { "epoch": 1.7737289713824569, "grad_norm": 0.6341580109082818, "learning_rate": 4.282706662587222e-05, "loss": 2.705, "step": 28573 }, { "epoch": 1.7737910484822148, "grad_norm": 0.5343522485273271, "learning_rate": 4.2823492476895655e-05, "loss": 2.6961, "step": 28574 }, { "epoch": 1.7738531255819727, "grad_norm": 0.6599015641860587, "learning_rate": 4.2819918365360664e-05, "loss": 2.7756, "step": 28575 }, { "epoch": 1.7739152026817306, "grad_norm": 0.5347532718142823, "learning_rate": 4.2816344291285915e-05, "loss": 2.7223, "step": 28576 }, { "epoch": 1.7739772797814886, "grad_norm": 0.5820708805391476, "learning_rate": 4.281277025469003e-05, "loss": 2.7496, "step": 28577 }, { "epoch": 1.7740393568812465, "grad_norm": 0.5288249935238332, "learning_rate": 4.280919625559167e-05, "loss": 2.7481, "step": 28578 }, { "epoch": 1.7741014339810044, "grad_norm": 0.6020333812764013, "learning_rate": 4.280562229400948e-05, "loss": 2.7765, "step": 28579 }, { "epoch": 1.7741635110807623, "grad_norm": 0.5029892969452647, "learning_rate": 4.28020483699621e-05, "loss": 2.6838, "step": 28580 }, { "epoch": 1.7742255881805202, "grad_norm": 0.6043070569749504, "learning_rate": 4.2798474483468194e-05, "loss": 2.7117, "step": 28581 }, { "epoch": 1.7742876652802781, "grad_norm": 0.5823468626290736, "learning_rate": 4.279490063454638e-05, "loss": 2.752, "step": 28582 }, { "epoch": 1.774349742380036, "grad_norm": 0.5918372363940555, "learning_rate": 4.2791326823215325e-05, "loss": 2.8014, "step": 28583 }, { "epoch": 1.774411819479794, "grad_norm": 0.5730476894255859, "learning_rate": 4.278775304949367e-05, "loss": 2.6534, "step": 28584 }, { "epoch": 1.774473896579552, "grad_norm": 0.674835204388088, "learning_rate": 4.2784179313400055e-05, "loss": 2.7774, "step": 28585 }, { "epoch": 1.7745359736793098, "grad_norm": 0.5594161854931354, "learning_rate": 4.278060561495313e-05, "loss": 2.7194, "step": 28586 }, { "epoch": 1.7745980507790677, "grad_norm": 0.6498945251498961, "learning_rate": 4.2777031954171535e-05, "loss": 2.6809, "step": 28587 }, { "epoch": 1.7746601278788257, "grad_norm": 0.5568252792596106, "learning_rate": 4.277345833107392e-05, "loss": 2.6646, "step": 28588 }, { "epoch": 1.7747222049785834, "grad_norm": 0.5962076770708566, "learning_rate": 4.276988474567893e-05, "loss": 2.6794, "step": 28589 }, { "epoch": 1.7747842820783413, "grad_norm": 0.5596303325106757, "learning_rate": 4.276631119800521e-05, "loss": 2.6836, "step": 28590 }, { "epoch": 1.7748463591780992, "grad_norm": 0.5663288721307791, "learning_rate": 4.276273768807138e-05, "loss": 2.7212, "step": 28591 }, { "epoch": 1.774908436277857, "grad_norm": 0.6019758563047756, "learning_rate": 4.275916421589613e-05, "loss": 2.8116, "step": 28592 }, { "epoch": 1.774970513377615, "grad_norm": 0.5011989404828502, "learning_rate": 4.275559078149806e-05, "loss": 2.7667, "step": 28593 }, { "epoch": 1.775032590477373, "grad_norm": 0.617481716866041, "learning_rate": 4.2752017384895844e-05, "loss": 2.7986, "step": 28594 }, { "epoch": 1.7750946675771306, "grad_norm": 0.5359161439427539, "learning_rate": 4.274844402610809e-05, "loss": 2.7714, "step": 28595 }, { "epoch": 1.7751567446768886, "grad_norm": 0.5736523903820129, "learning_rate": 4.2744870705153474e-05, "loss": 2.7053, "step": 28596 }, { "epoch": 1.7752188217766465, "grad_norm": 0.5807986937728885, "learning_rate": 4.274129742205064e-05, "loss": 2.7038, "step": 28597 }, { "epoch": 1.7752808988764044, "grad_norm": 0.5267495374980792, "learning_rate": 4.273772417681821e-05, "loss": 2.6401, "step": 28598 }, { "epoch": 1.7753429759761623, "grad_norm": 0.5322092035944944, "learning_rate": 4.273415096947485e-05, "loss": 2.7778, "step": 28599 }, { "epoch": 1.7754050530759202, "grad_norm": 0.5558241628617641, "learning_rate": 4.273057780003917e-05, "loss": 2.7865, "step": 28600 }, { "epoch": 1.7754671301756781, "grad_norm": 0.5615820281932538, "learning_rate": 4.2727004668529844e-05, "loss": 2.7819, "step": 28601 }, { "epoch": 1.775529207275436, "grad_norm": 0.5813558102764315, "learning_rate": 4.2723431574965487e-05, "loss": 2.7496, "step": 28602 }, { "epoch": 1.775591284375194, "grad_norm": 0.5119177162130258, "learning_rate": 4.271985851936477e-05, "loss": 2.7293, "step": 28603 }, { "epoch": 1.775653361474952, "grad_norm": 0.5533493179814131, "learning_rate": 4.271628550174632e-05, "loss": 2.9253, "step": 28604 }, { "epoch": 1.7757154385747098, "grad_norm": 0.6143379098556505, "learning_rate": 4.271271252212876e-05, "loss": 2.7413, "step": 28605 }, { "epoch": 1.7757775156744677, "grad_norm": 0.5706711516670134, "learning_rate": 4.270913958053076e-05, "loss": 2.752, "step": 28606 }, { "epoch": 1.7758395927742257, "grad_norm": 0.5704456591941603, "learning_rate": 4.2705566676970944e-05, "loss": 2.6769, "step": 28607 }, { "epoch": 1.7759016698739836, "grad_norm": 0.5622448872511436, "learning_rate": 4.270199381146796e-05, "loss": 2.8238, "step": 28608 }, { "epoch": 1.7759637469737415, "grad_norm": 0.601594081891847, "learning_rate": 4.269842098404044e-05, "loss": 2.7383, "step": 28609 }, { "epoch": 1.7760258240734994, "grad_norm": 0.5476113675128835, "learning_rate": 4.269484819470704e-05, "loss": 2.713, "step": 28610 }, { "epoch": 1.7760879011732573, "grad_norm": 0.5900814512167393, "learning_rate": 4.2691275443486374e-05, "loss": 2.7086, "step": 28611 }, { "epoch": 1.7761499782730152, "grad_norm": 0.5225634607991289, "learning_rate": 4.2687702730397114e-05, "loss": 2.8174, "step": 28612 }, { "epoch": 1.776212055372773, "grad_norm": 0.5853157559460591, "learning_rate": 4.268413005545787e-05, "loss": 2.6965, "step": 28613 }, { "epoch": 1.7762741324725309, "grad_norm": 0.5736495330212817, "learning_rate": 4.268055741868731e-05, "loss": 2.796, "step": 28614 }, { "epoch": 1.7763362095722888, "grad_norm": 0.6554114797532021, "learning_rate": 4.267698482010406e-05, "loss": 2.8369, "step": 28615 }, { "epoch": 1.7763982866720467, "grad_norm": 0.5561864505856087, "learning_rate": 4.267341225972674e-05, "loss": 2.6904, "step": 28616 }, { "epoch": 1.7764603637718046, "grad_norm": 0.5277134764830201, "learning_rate": 4.2669839737574026e-05, "loss": 2.7216, "step": 28617 }, { "epoch": 1.7765224408715625, "grad_norm": 0.604901497207103, "learning_rate": 4.2666267253664526e-05, "loss": 2.8229, "step": 28618 }, { "epoch": 1.7765845179713202, "grad_norm": 0.552821283955329, "learning_rate": 4.26626948080169e-05, "loss": 2.6787, "step": 28619 }, { "epoch": 1.7766465950710781, "grad_norm": 0.5897908218586148, "learning_rate": 4.265912240064976e-05, "loss": 2.847, "step": 28620 }, { "epoch": 1.776708672170836, "grad_norm": 0.5213227325119401, "learning_rate": 4.2655550031581774e-05, "loss": 2.8153, "step": 28621 }, { "epoch": 1.776770749270594, "grad_norm": 0.5699563678137084, "learning_rate": 4.265197770083156e-05, "loss": 2.8083, "step": 28622 }, { "epoch": 1.776832826370352, "grad_norm": 0.49690542127941867, "learning_rate": 4.264840540841777e-05, "loss": 2.6857, "step": 28623 }, { "epoch": 1.7768949034701098, "grad_norm": 0.5906725485902071, "learning_rate": 4.264483315435902e-05, "loss": 2.7784, "step": 28624 }, { "epoch": 1.7769569805698677, "grad_norm": 0.6142104817457178, "learning_rate": 4.2641260938673976e-05, "loss": 2.7855, "step": 28625 }, { "epoch": 1.7770190576696256, "grad_norm": 0.5932312003869175, "learning_rate": 4.2637688761381254e-05, "loss": 2.7239, "step": 28626 }, { "epoch": 1.7770811347693836, "grad_norm": 0.5904107053218407, "learning_rate": 4.2634116622499484e-05, "loss": 2.7005, "step": 28627 }, { "epoch": 1.7771432118691415, "grad_norm": 0.5576254715446115, "learning_rate": 4.263054452204732e-05, "loss": 2.7795, "step": 28628 }, { "epoch": 1.7772052889688994, "grad_norm": 0.5410045250810924, "learning_rate": 4.2626972460043415e-05, "loss": 2.7314, "step": 28629 }, { "epoch": 1.7772673660686573, "grad_norm": 0.6108609934559863, "learning_rate": 4.2623400436506363e-05, "loss": 2.7637, "step": 28630 }, { "epoch": 1.7773294431684152, "grad_norm": 0.5065439371688467, "learning_rate": 4.261982845145484e-05, "loss": 2.7077, "step": 28631 }, { "epoch": 1.7773915202681732, "grad_norm": 0.5527896381694717, "learning_rate": 4.261625650490744e-05, "loss": 2.6834, "step": 28632 }, { "epoch": 1.777453597367931, "grad_norm": 0.5692994513361096, "learning_rate": 4.261268459688284e-05, "loss": 2.5793, "step": 28633 }, { "epoch": 1.777515674467689, "grad_norm": 0.551812120313705, "learning_rate": 4.260911272739965e-05, "loss": 2.7253, "step": 28634 }, { "epoch": 1.777577751567447, "grad_norm": 0.623219361540259, "learning_rate": 4.2605540896476524e-05, "loss": 2.6893, "step": 28635 }, { "epoch": 1.7776398286672048, "grad_norm": 0.5490975312644498, "learning_rate": 4.2601969104132074e-05, "loss": 2.7857, "step": 28636 }, { "epoch": 1.7777019057669625, "grad_norm": 0.5897677543137809, "learning_rate": 4.259839735038496e-05, "loss": 2.7051, "step": 28637 }, { "epoch": 1.7777639828667204, "grad_norm": 0.6003218403555747, "learning_rate": 4.259482563525379e-05, "loss": 2.7627, "step": 28638 }, { "epoch": 1.7778260599664784, "grad_norm": 0.5900618547812426, "learning_rate": 4.259125395875722e-05, "loss": 2.7639, "step": 28639 }, { "epoch": 1.7778881370662363, "grad_norm": 0.5610302160458898, "learning_rate": 4.258768232091388e-05, "loss": 2.7053, "step": 28640 }, { "epoch": 1.7779502141659942, "grad_norm": 0.5822481608871563, "learning_rate": 4.2584110721742394e-05, "loss": 2.7433, "step": 28641 }, { "epoch": 1.778012291265752, "grad_norm": 0.5847407517274681, "learning_rate": 4.258053916126141e-05, "loss": 2.6224, "step": 28642 }, { "epoch": 1.7780743683655098, "grad_norm": 0.6396889710141531, "learning_rate": 4.257696763948954e-05, "loss": 2.7157, "step": 28643 }, { "epoch": 1.7781364454652677, "grad_norm": 0.5634790747733612, "learning_rate": 4.2573396156445444e-05, "loss": 2.7585, "step": 28644 }, { "epoch": 1.7781985225650256, "grad_norm": 0.6378821760787249, "learning_rate": 4.256982471214774e-05, "loss": 2.7198, "step": 28645 }, { "epoch": 1.7782605996647836, "grad_norm": 0.5593268379064263, "learning_rate": 4.256625330661507e-05, "loss": 2.7889, "step": 28646 }, { "epoch": 1.7783226767645415, "grad_norm": 0.6528554497044279, "learning_rate": 4.2562681939866043e-05, "loss": 2.7712, "step": 28647 }, { "epoch": 1.7783847538642994, "grad_norm": 0.5424545378028335, "learning_rate": 4.2559110611919327e-05, "loss": 2.6493, "step": 28648 }, { "epoch": 1.7784468309640573, "grad_norm": 0.6106125662639798, "learning_rate": 4.255553932279353e-05, "loss": 2.7089, "step": 28649 }, { "epoch": 1.7785089080638152, "grad_norm": 0.5754312960564858, "learning_rate": 4.2551968072507295e-05, "loss": 2.7778, "step": 28650 }, { "epoch": 1.7785709851635731, "grad_norm": 0.523842811236365, "learning_rate": 4.254839686107926e-05, "loss": 2.7043, "step": 28651 }, { "epoch": 1.778633062263331, "grad_norm": 0.5551194671547545, "learning_rate": 4.2544825688528026e-05, "loss": 2.8063, "step": 28652 }, { "epoch": 1.778695139363089, "grad_norm": 0.5589965488433078, "learning_rate": 4.254125455487227e-05, "loss": 2.7712, "step": 28653 }, { "epoch": 1.778757216462847, "grad_norm": 0.48200151917507045, "learning_rate": 4.2537683460130575e-05, "loss": 2.6998, "step": 28654 }, { "epoch": 1.7788192935626048, "grad_norm": 0.5079702865779169, "learning_rate": 4.253411240432161e-05, "loss": 2.6344, "step": 28655 }, { "epoch": 1.7788813706623627, "grad_norm": 0.5130320287990056, "learning_rate": 4.253054138746399e-05, "loss": 2.7663, "step": 28656 }, { "epoch": 1.7789434477621207, "grad_norm": 0.5138177990785723, "learning_rate": 4.252697040957636e-05, "loss": 2.7748, "step": 28657 }, { "epoch": 1.7790055248618786, "grad_norm": 0.520668505972293, "learning_rate": 4.252339947067732e-05, "loss": 2.728, "step": 28658 }, { "epoch": 1.7790676019616365, "grad_norm": 0.5170853770408029, "learning_rate": 4.2519828570785534e-05, "loss": 2.7949, "step": 28659 }, { "epoch": 1.7791296790613944, "grad_norm": 0.5647715426098178, "learning_rate": 4.25162577099196e-05, "loss": 2.7802, "step": 28660 }, { "epoch": 1.779191756161152, "grad_norm": 0.4831985214371828, "learning_rate": 4.2512686888098174e-05, "loss": 2.7726, "step": 28661 }, { "epoch": 1.77925383326091, "grad_norm": 0.6351495374111367, "learning_rate": 4.250911610533989e-05, "loss": 2.7525, "step": 28662 }, { "epoch": 1.779315910360668, "grad_norm": 0.5111217364932141, "learning_rate": 4.2505545361663354e-05, "loss": 2.8093, "step": 28663 }, { "epoch": 1.7793779874604259, "grad_norm": 0.5513928582069683, "learning_rate": 4.250197465708722e-05, "loss": 2.7586, "step": 28664 }, { "epoch": 1.7794400645601838, "grad_norm": 0.5636550118295862, "learning_rate": 4.2498403991630105e-05, "loss": 2.7459, "step": 28665 }, { "epoch": 1.7795021416599417, "grad_norm": 0.6036346466326893, "learning_rate": 4.249483336531063e-05, "loss": 2.7714, "step": 28666 }, { "epoch": 1.7795642187596994, "grad_norm": 0.5445450057231086, "learning_rate": 4.249126277814743e-05, "loss": 2.727, "step": 28667 }, { "epoch": 1.7796262958594573, "grad_norm": 0.5812468651717775, "learning_rate": 4.2487692230159134e-05, "loss": 2.7307, "step": 28668 }, { "epoch": 1.7796883729592152, "grad_norm": 0.49446567203046154, "learning_rate": 4.248412172136438e-05, "loss": 2.8454, "step": 28669 }, { "epoch": 1.7797504500589731, "grad_norm": 0.604456122436428, "learning_rate": 4.248055125178178e-05, "loss": 2.7895, "step": 28670 }, { "epoch": 1.779812527158731, "grad_norm": 0.5129149048426614, "learning_rate": 4.2476980821429966e-05, "loss": 2.7905, "step": 28671 }, { "epoch": 1.779874604258489, "grad_norm": 0.5496389840615535, "learning_rate": 4.2473410430327574e-05, "loss": 2.7417, "step": 28672 }, { "epoch": 1.779936681358247, "grad_norm": 0.49385309712341696, "learning_rate": 4.246984007849323e-05, "loss": 2.6476, "step": 28673 }, { "epoch": 1.7799987584580048, "grad_norm": 0.5067809852401344, "learning_rate": 4.246626976594555e-05, "loss": 2.773, "step": 28674 }, { "epoch": 1.7800608355577627, "grad_norm": 0.5314846081770341, "learning_rate": 4.246269949270318e-05, "loss": 2.7848, "step": 28675 }, { "epoch": 1.7801229126575207, "grad_norm": 0.527585150955621, "learning_rate": 4.2459129258784734e-05, "loss": 2.6874, "step": 28676 }, { "epoch": 1.7801849897572786, "grad_norm": 0.5353721962605303, "learning_rate": 4.245555906420884e-05, "loss": 2.6602, "step": 28677 }, { "epoch": 1.7802470668570365, "grad_norm": 0.5364187343265796, "learning_rate": 4.2451988908994126e-05, "loss": 2.6076, "step": 28678 }, { "epoch": 1.7803091439567944, "grad_norm": 0.5386025774934801, "learning_rate": 4.244841879315921e-05, "loss": 2.7412, "step": 28679 }, { "epoch": 1.7803712210565523, "grad_norm": 0.5609084236025969, "learning_rate": 4.244484871672274e-05, "loss": 2.7143, "step": 28680 }, { "epoch": 1.7804332981563102, "grad_norm": 0.5330586499676594, "learning_rate": 4.244127867970332e-05, "loss": 2.7863, "step": 28681 }, { "epoch": 1.7804953752560682, "grad_norm": 0.6328894015080109, "learning_rate": 4.243770868211959e-05, "loss": 2.7502, "step": 28682 }, { "epoch": 1.780557452355826, "grad_norm": 0.5798486583258352, "learning_rate": 4.243413872399016e-05, "loss": 2.8363, "step": 28683 }, { "epoch": 1.780619529455584, "grad_norm": 0.5777207585110697, "learning_rate": 4.243056880533367e-05, "loss": 2.7715, "step": 28684 }, { "epoch": 1.7806816065553417, "grad_norm": 0.6009004095238953, "learning_rate": 4.2426998926168735e-05, "loss": 2.7692, "step": 28685 }, { "epoch": 1.7807436836550996, "grad_norm": 0.6052691222852842, "learning_rate": 4.2423429086514e-05, "loss": 2.7317, "step": 28686 }, { "epoch": 1.7808057607548575, "grad_norm": 0.5846911589988667, "learning_rate": 4.241985928638807e-05, "loss": 2.7119, "step": 28687 }, { "epoch": 1.7808678378546154, "grad_norm": 0.5511887037491384, "learning_rate": 4.241628952580956e-05, "loss": 2.7374, "step": 28688 }, { "epoch": 1.7809299149543734, "grad_norm": 0.5766883991764046, "learning_rate": 4.2412719804797127e-05, "loss": 2.7689, "step": 28689 }, { "epoch": 1.7809919920541313, "grad_norm": 0.6128599092312792, "learning_rate": 4.2409150123369356e-05, "loss": 2.7502, "step": 28690 }, { "epoch": 1.781054069153889, "grad_norm": 0.6085392772110055, "learning_rate": 4.240558048154492e-05, "loss": 2.7259, "step": 28691 }, { "epoch": 1.781116146253647, "grad_norm": 0.6509228381750469, "learning_rate": 4.240201087934239e-05, "loss": 2.8111, "step": 28692 }, { "epoch": 1.7811782233534048, "grad_norm": 0.5742357472090417, "learning_rate": 4.23984413167804e-05, "loss": 2.7004, "step": 28693 }, { "epoch": 1.7812403004531627, "grad_norm": 0.6010894153697489, "learning_rate": 4.239487179387761e-05, "loss": 2.7042, "step": 28694 }, { "epoch": 1.7813023775529206, "grad_norm": 0.6152848480350354, "learning_rate": 4.2391302310652614e-05, "loss": 2.7488, "step": 28695 }, { "epoch": 1.7813644546526786, "grad_norm": 0.5825953470216501, "learning_rate": 4.2387732867124055e-05, "loss": 2.8229, "step": 28696 }, { "epoch": 1.7814265317524365, "grad_norm": 0.5524851098794327, "learning_rate": 4.238416346331052e-05, "loss": 2.7489, "step": 28697 }, { "epoch": 1.7814886088521944, "grad_norm": 0.6014519379313976, "learning_rate": 4.238059409923067e-05, "loss": 2.7413, "step": 28698 }, { "epoch": 1.7815506859519523, "grad_norm": 0.5383484298795824, "learning_rate": 4.23770247749031e-05, "loss": 2.7541, "step": 28699 }, { "epoch": 1.7816127630517102, "grad_norm": 0.5885596286654626, "learning_rate": 4.237345549034645e-05, "loss": 2.6143, "step": 28700 }, { "epoch": 1.7816748401514682, "grad_norm": 0.5647577170122973, "learning_rate": 4.236988624557934e-05, "loss": 2.8243, "step": 28701 }, { "epoch": 1.781736917251226, "grad_norm": 0.579847347707851, "learning_rate": 4.236631704062037e-05, "loss": 2.7044, "step": 28702 }, { "epoch": 1.781798994350984, "grad_norm": 0.5845205085050929, "learning_rate": 4.236274787548819e-05, "loss": 2.7534, "step": 28703 }, { "epoch": 1.781861071450742, "grad_norm": 0.5782974862203197, "learning_rate": 4.2359178750201393e-05, "loss": 2.7514, "step": 28704 }, { "epoch": 1.7819231485504998, "grad_norm": 0.588487518542804, "learning_rate": 4.235560966477863e-05, "loss": 2.7474, "step": 28705 }, { "epoch": 1.7819852256502577, "grad_norm": 0.5106336430876867, "learning_rate": 4.23520406192385e-05, "loss": 2.6746, "step": 28706 }, { "epoch": 1.7820473027500157, "grad_norm": 0.6199576826035975, "learning_rate": 4.2348471613599636e-05, "loss": 2.7796, "step": 28707 }, { "epoch": 1.7821093798497736, "grad_norm": 0.5361166640665564, "learning_rate": 4.234490264788065e-05, "loss": 2.8129, "step": 28708 }, { "epoch": 1.7821714569495313, "grad_norm": 0.5671450564895009, "learning_rate": 4.2341333722100166e-05, "loss": 2.7421, "step": 28709 }, { "epoch": 1.7822335340492892, "grad_norm": 0.573560064454658, "learning_rate": 4.2337764836276796e-05, "loss": 2.7574, "step": 28710 }, { "epoch": 1.782295611149047, "grad_norm": 0.6153026240201882, "learning_rate": 4.233419599042918e-05, "loss": 2.7445, "step": 28711 }, { "epoch": 1.782357688248805, "grad_norm": 0.5523065319768382, "learning_rate": 4.2330627184575924e-05, "loss": 2.6944, "step": 28712 }, { "epoch": 1.782419765348563, "grad_norm": 0.6444158508568248, "learning_rate": 4.232705841873563e-05, "loss": 2.7221, "step": 28713 }, { "epoch": 1.7824818424483209, "grad_norm": 0.4923689674207385, "learning_rate": 4.2323489692926956e-05, "loss": 2.8188, "step": 28714 }, { "epoch": 1.7825439195480786, "grad_norm": 0.6991114613998707, "learning_rate": 4.2319921007168486e-05, "loss": 2.6631, "step": 28715 }, { "epoch": 1.7826059966478365, "grad_norm": 0.5225217409493211, "learning_rate": 4.2316352361478855e-05, "loss": 2.7832, "step": 28716 }, { "epoch": 1.7826680737475944, "grad_norm": 0.585772681897319, "learning_rate": 4.231278375587668e-05, "loss": 2.7995, "step": 28717 }, { "epoch": 1.7827301508473523, "grad_norm": 0.611194071834148, "learning_rate": 4.230921519038058e-05, "loss": 2.7138, "step": 28718 }, { "epoch": 1.7827922279471102, "grad_norm": 0.568409317811152, "learning_rate": 4.230564666500917e-05, "loss": 2.795, "step": 28719 }, { "epoch": 1.7828543050468681, "grad_norm": 0.557968189513759, "learning_rate": 4.2302078179781065e-05, "loss": 2.7141, "step": 28720 }, { "epoch": 1.782916382146626, "grad_norm": 0.63457007110206, "learning_rate": 4.2298509734714894e-05, "loss": 2.6726, "step": 28721 }, { "epoch": 1.782978459246384, "grad_norm": 0.6120164305251596, "learning_rate": 4.229494132982927e-05, "loss": 2.7356, "step": 28722 }, { "epoch": 1.783040536346142, "grad_norm": 0.555600771123627, "learning_rate": 4.22913729651428e-05, "loss": 2.7808, "step": 28723 }, { "epoch": 1.7831026134458998, "grad_norm": 0.611088032238516, "learning_rate": 4.228780464067411e-05, "loss": 2.7743, "step": 28724 }, { "epoch": 1.7831646905456577, "grad_norm": 0.4778983669829421, "learning_rate": 4.228423635644182e-05, "loss": 2.6196, "step": 28725 }, { "epoch": 1.7832267676454157, "grad_norm": 0.5513950471937186, "learning_rate": 4.2280668112464524e-05, "loss": 2.6587, "step": 28726 }, { "epoch": 1.7832888447451736, "grad_norm": 0.5109151689508192, "learning_rate": 4.2277099908760866e-05, "loss": 2.7337, "step": 28727 }, { "epoch": 1.7833509218449315, "grad_norm": 0.5693086908799112, "learning_rate": 4.227353174534946e-05, "loss": 2.7963, "step": 28728 }, { "epoch": 1.7834129989446894, "grad_norm": 0.5676470878065369, "learning_rate": 4.22699636222489e-05, "loss": 2.7491, "step": 28729 }, { "epoch": 1.7834750760444473, "grad_norm": 0.5476565362382649, "learning_rate": 4.226639553947784e-05, "loss": 2.7984, "step": 28730 }, { "epoch": 1.7835371531442052, "grad_norm": 0.5739412999842061, "learning_rate": 4.226282749705485e-05, "loss": 2.7138, "step": 28731 }, { "epoch": 1.7835992302439632, "grad_norm": 0.5497511571704574, "learning_rate": 4.2259259494998585e-05, "loss": 2.7751, "step": 28732 }, { "epoch": 1.7836613073437209, "grad_norm": 0.5778030065867605, "learning_rate": 4.225569153332763e-05, "loss": 2.7786, "step": 28733 }, { "epoch": 1.7837233844434788, "grad_norm": 0.5546209268512804, "learning_rate": 4.225212361206062e-05, "loss": 2.7386, "step": 28734 }, { "epoch": 1.7837854615432367, "grad_norm": 0.5629389957184909, "learning_rate": 4.224855573121616e-05, "loss": 2.7579, "step": 28735 }, { "epoch": 1.7838475386429946, "grad_norm": 0.5740293852299043, "learning_rate": 4.224498789081286e-05, "loss": 2.7806, "step": 28736 }, { "epoch": 1.7839096157427525, "grad_norm": 0.5254688984589675, "learning_rate": 4.2241420090869356e-05, "loss": 2.7047, "step": 28737 }, { "epoch": 1.7839716928425104, "grad_norm": 0.5098425854579895, "learning_rate": 4.223785233140423e-05, "loss": 2.7429, "step": 28738 }, { "epoch": 1.7840337699422681, "grad_norm": 0.5458464623345243, "learning_rate": 4.2234284612436127e-05, "loss": 2.6488, "step": 28739 }, { "epoch": 1.784095847042026, "grad_norm": 0.5599896256303313, "learning_rate": 4.223071693398363e-05, "loss": 2.7659, "step": 28740 }, { "epoch": 1.784157924141784, "grad_norm": 0.5034646457525904, "learning_rate": 4.222714929606538e-05, "loss": 2.6877, "step": 28741 }, { "epoch": 1.784220001241542, "grad_norm": 0.5518472244600047, "learning_rate": 4.222358169869997e-05, "loss": 2.705, "step": 28742 }, { "epoch": 1.7842820783412998, "grad_norm": 0.5586342696852471, "learning_rate": 4.2220014141906036e-05, "loss": 2.7328, "step": 28743 }, { "epoch": 1.7843441554410577, "grad_norm": 0.5049077217848875, "learning_rate": 4.2216446625702164e-05, "loss": 2.641, "step": 28744 }, { "epoch": 1.7844062325408157, "grad_norm": 0.5660417107151628, "learning_rate": 4.221287915010699e-05, "loss": 2.7205, "step": 28745 }, { "epoch": 1.7844683096405736, "grad_norm": 0.5143643487265525, "learning_rate": 4.22093117151391e-05, "loss": 2.7195, "step": 28746 }, { "epoch": 1.7845303867403315, "grad_norm": 0.6001719715112074, "learning_rate": 4.220574432081714e-05, "loss": 2.7768, "step": 28747 }, { "epoch": 1.7845924638400894, "grad_norm": 0.5772198989441204, "learning_rate": 4.22021769671597e-05, "loss": 2.7691, "step": 28748 }, { "epoch": 1.7846545409398473, "grad_norm": 0.5327350392346564, "learning_rate": 4.2198609654185396e-05, "loss": 2.6483, "step": 28749 }, { "epoch": 1.7847166180396052, "grad_norm": 0.5367019772858692, "learning_rate": 4.2195042381912844e-05, "loss": 2.7299, "step": 28750 }, { "epoch": 1.7847786951393632, "grad_norm": 0.5736039951607633, "learning_rate": 4.219147515036063e-05, "loss": 2.8241, "step": 28751 }, { "epoch": 1.784840772239121, "grad_norm": 0.5454378590596279, "learning_rate": 4.21879079595474e-05, "loss": 2.8404, "step": 28752 }, { "epoch": 1.784902849338879, "grad_norm": 0.5276116714872666, "learning_rate": 4.2184340809491754e-05, "loss": 2.8147, "step": 28753 }, { "epoch": 1.784964926438637, "grad_norm": 0.5501877583454691, "learning_rate": 4.2180773700212295e-05, "loss": 2.6586, "step": 28754 }, { "epoch": 1.7850270035383948, "grad_norm": 0.5969146415768053, "learning_rate": 4.217720663172763e-05, "loss": 2.7737, "step": 28755 }, { "epoch": 1.7850890806381527, "grad_norm": 0.5417099501360209, "learning_rate": 4.2173639604056395e-05, "loss": 2.6566, "step": 28756 }, { "epoch": 1.7851511577379104, "grad_norm": 0.6113407511519353, "learning_rate": 4.217007261721716e-05, "loss": 2.7347, "step": 28757 }, { "epoch": 1.7852132348376684, "grad_norm": 0.6047710705950429, "learning_rate": 4.216650567122855e-05, "loss": 2.7286, "step": 28758 }, { "epoch": 1.7852753119374263, "grad_norm": 0.5554872040292013, "learning_rate": 4.216293876610921e-05, "loss": 2.7773, "step": 28759 }, { "epoch": 1.7853373890371842, "grad_norm": 0.5773871068715407, "learning_rate": 4.2159371901877696e-05, "loss": 2.7735, "step": 28760 }, { "epoch": 1.7853994661369421, "grad_norm": 0.5705267098916834, "learning_rate": 4.215580507855267e-05, "loss": 2.761, "step": 28761 }, { "epoch": 1.7854615432367, "grad_norm": 0.5871506910577148, "learning_rate": 4.21522382961527e-05, "loss": 2.7177, "step": 28762 }, { "epoch": 1.7855236203364577, "grad_norm": 0.5933223489305354, "learning_rate": 4.2148671554696404e-05, "loss": 2.7555, "step": 28763 }, { "epoch": 1.7855856974362156, "grad_norm": 0.559642021550864, "learning_rate": 4.2145104854202405e-05, "loss": 2.6818, "step": 28764 }, { "epoch": 1.7856477745359736, "grad_norm": 0.5980755700042538, "learning_rate": 4.214153819468929e-05, "loss": 2.79, "step": 28765 }, { "epoch": 1.7857098516357315, "grad_norm": 0.5617798062435573, "learning_rate": 4.2137971576175686e-05, "loss": 2.6242, "step": 28766 }, { "epoch": 1.7857719287354894, "grad_norm": 0.6284303700987761, "learning_rate": 4.213440499868018e-05, "loss": 2.7354, "step": 28767 }, { "epoch": 1.7858340058352473, "grad_norm": 0.5877576405114993, "learning_rate": 4.213083846222141e-05, "loss": 2.7465, "step": 28768 }, { "epoch": 1.7858960829350052, "grad_norm": 0.581036992164247, "learning_rate": 4.212727196681795e-05, "loss": 2.7621, "step": 28769 }, { "epoch": 1.7859581600347632, "grad_norm": 0.6058509850674321, "learning_rate": 4.212370551248844e-05, "loss": 2.6611, "step": 28770 }, { "epoch": 1.786020237134521, "grad_norm": 0.53565888572522, "learning_rate": 4.2120139099251455e-05, "loss": 2.7373, "step": 28771 }, { "epoch": 1.786082314234279, "grad_norm": 0.5764388936603023, "learning_rate": 4.211657272712563e-05, "loss": 2.7908, "step": 28772 }, { "epoch": 1.786144391334037, "grad_norm": 0.5737053269092274, "learning_rate": 4.211300639612954e-05, "loss": 2.8099, "step": 28773 }, { "epoch": 1.7862064684337948, "grad_norm": 0.5904731171712208, "learning_rate": 4.210944010628183e-05, "loss": 2.7444, "step": 28774 }, { "epoch": 1.7862685455335527, "grad_norm": 0.5766304131184949, "learning_rate": 4.2105873857601085e-05, "loss": 2.6967, "step": 28775 }, { "epoch": 1.7863306226333107, "grad_norm": 0.6338182627765111, "learning_rate": 4.2102307650105896e-05, "loss": 2.7908, "step": 28776 }, { "epoch": 1.7863926997330686, "grad_norm": 0.6414679819339308, "learning_rate": 4.209874148381491e-05, "loss": 2.7008, "step": 28777 }, { "epoch": 1.7864547768328265, "grad_norm": 0.4938812479788829, "learning_rate": 4.2095175358746684e-05, "loss": 2.8112, "step": 28778 }, { "epoch": 1.7865168539325844, "grad_norm": 0.6303887988971882, "learning_rate": 4.2091609274919864e-05, "loss": 2.9437, "step": 28779 }, { "epoch": 1.7865789310323423, "grad_norm": 0.5760312700856752, "learning_rate": 4.208804323235303e-05, "loss": 2.743, "step": 28780 }, { "epoch": 1.7866410081321, "grad_norm": 0.5156943245726174, "learning_rate": 4.20844772310648e-05, "loss": 2.6835, "step": 28781 }, { "epoch": 1.786703085231858, "grad_norm": 0.592069556737947, "learning_rate": 4.208091127107376e-05, "loss": 2.7662, "step": 28782 }, { "epoch": 1.7867651623316159, "grad_norm": 0.5695971875946393, "learning_rate": 4.207734535239854e-05, "loss": 2.7386, "step": 28783 }, { "epoch": 1.7868272394313738, "grad_norm": 0.5278570714440215, "learning_rate": 4.207377947505773e-05, "loss": 2.6601, "step": 28784 }, { "epoch": 1.7868893165311317, "grad_norm": 0.5922853323613068, "learning_rate": 4.207021363906994e-05, "loss": 2.6496, "step": 28785 }, { "epoch": 1.7869513936308896, "grad_norm": 0.5320506793940304, "learning_rate": 4.206664784445377e-05, "loss": 2.7674, "step": 28786 }, { "epoch": 1.7870134707306473, "grad_norm": 0.5972164755059279, "learning_rate": 4.206308209122782e-05, "loss": 2.6632, "step": 28787 }, { "epoch": 1.7870755478304052, "grad_norm": 0.5498187535927561, "learning_rate": 4.20595163794107e-05, "loss": 2.7192, "step": 28788 }, { "epoch": 1.7871376249301631, "grad_norm": 0.6116592923983616, "learning_rate": 4.2055950709021006e-05, "loss": 2.8365, "step": 28789 }, { "epoch": 1.787199702029921, "grad_norm": 0.5319737665208433, "learning_rate": 4.205238508007735e-05, "loss": 2.7056, "step": 28790 }, { "epoch": 1.787261779129679, "grad_norm": 0.5845941785330682, "learning_rate": 4.204881949259832e-05, "loss": 2.7759, "step": 28791 }, { "epoch": 1.787323856229437, "grad_norm": 0.5365880204509625, "learning_rate": 4.204525394660254e-05, "loss": 2.7354, "step": 28792 }, { "epoch": 1.7873859333291948, "grad_norm": 0.6304052507644757, "learning_rate": 4.2041688442108604e-05, "loss": 2.8247, "step": 28793 }, { "epoch": 1.7874480104289527, "grad_norm": 0.5414165292179652, "learning_rate": 4.2038122979135094e-05, "loss": 2.7163, "step": 28794 }, { "epoch": 1.7875100875287107, "grad_norm": 0.5694279015730387, "learning_rate": 4.203455755770065e-05, "loss": 2.8008, "step": 28795 }, { "epoch": 1.7875721646284686, "grad_norm": 0.5511609925967259, "learning_rate": 4.203099217782384e-05, "loss": 2.8005, "step": 28796 }, { "epoch": 1.7876342417282265, "grad_norm": 0.5348847528911506, "learning_rate": 4.202742683952329e-05, "loss": 2.7425, "step": 28797 }, { "epoch": 1.7876963188279844, "grad_norm": 0.5597679494848125, "learning_rate": 4.202386154281758e-05, "loss": 2.7328, "step": 28798 }, { "epoch": 1.7877583959277423, "grad_norm": 0.5982440725527863, "learning_rate": 4.202029628772533e-05, "loss": 2.6803, "step": 28799 }, { "epoch": 1.7878204730275002, "grad_norm": 0.596208801765568, "learning_rate": 4.201673107426513e-05, "loss": 2.6729, "step": 28800 }, { "epoch": 1.7878825501272582, "grad_norm": 0.5846585111297529, "learning_rate": 4.201316590245557e-05, "loss": 2.7193, "step": 28801 }, { "epoch": 1.787944627227016, "grad_norm": 0.6181269388968798, "learning_rate": 4.2009600772315275e-05, "loss": 2.698, "step": 28802 }, { "epoch": 1.788006704326774, "grad_norm": 0.5873087541282737, "learning_rate": 4.2006035683862824e-05, "loss": 2.775, "step": 28803 }, { "epoch": 1.788068781426532, "grad_norm": 0.6249649514861534, "learning_rate": 4.200247063711683e-05, "loss": 2.7869, "step": 28804 }, { "epoch": 1.7881308585262896, "grad_norm": 0.5901398578656892, "learning_rate": 4.1998905632095886e-05, "loss": 2.7418, "step": 28805 }, { "epoch": 1.7881929356260475, "grad_norm": 0.5585637651411867, "learning_rate": 4.19953406688186e-05, "loss": 2.6401, "step": 28806 }, { "epoch": 1.7882550127258054, "grad_norm": 0.5806308514861707, "learning_rate": 4.199177574730355e-05, "loss": 2.6861, "step": 28807 }, { "epoch": 1.7883170898255634, "grad_norm": 0.6246797933480004, "learning_rate": 4.198821086756937e-05, "loss": 2.7102, "step": 28808 }, { "epoch": 1.7883791669253213, "grad_norm": 0.5835980824995275, "learning_rate": 4.1984646029634624e-05, "loss": 2.7753, "step": 28809 }, { "epoch": 1.7884412440250792, "grad_norm": 0.5935352338569724, "learning_rate": 4.1981081233517936e-05, "loss": 2.8064, "step": 28810 }, { "epoch": 1.788503321124837, "grad_norm": 0.5740169248312073, "learning_rate": 4.1977516479237885e-05, "loss": 2.7696, "step": 28811 }, { "epoch": 1.7885653982245948, "grad_norm": 0.5601599816872382, "learning_rate": 4.197395176681308e-05, "loss": 2.6726, "step": 28812 }, { "epoch": 1.7886274753243527, "grad_norm": 0.5229628104358992, "learning_rate": 4.197038709626213e-05, "loss": 2.7335, "step": 28813 }, { "epoch": 1.7886895524241107, "grad_norm": 0.5937828506617596, "learning_rate": 4.1966822467603596e-05, "loss": 2.8138, "step": 28814 }, { "epoch": 1.7887516295238686, "grad_norm": 0.5119769486010273, "learning_rate": 4.196325788085612e-05, "loss": 2.7601, "step": 28815 }, { "epoch": 1.7888137066236265, "grad_norm": 0.5140300995607853, "learning_rate": 4.195969333603826e-05, "loss": 2.8331, "step": 28816 }, { "epoch": 1.7888757837233844, "grad_norm": 0.5216404827643911, "learning_rate": 4.1956128833168644e-05, "loss": 2.6652, "step": 28817 }, { "epoch": 1.7889378608231423, "grad_norm": 0.5430831788602177, "learning_rate": 4.1952564372265854e-05, "loss": 2.7566, "step": 28818 }, { "epoch": 1.7889999379229002, "grad_norm": 0.4836267259783399, "learning_rate": 4.194899995334849e-05, "loss": 2.6826, "step": 28819 }, { "epoch": 1.7890620150226582, "grad_norm": 0.5829617276020529, "learning_rate": 4.194543557643514e-05, "loss": 2.6955, "step": 28820 }, { "epoch": 1.789124092122416, "grad_norm": 0.5335345792566525, "learning_rate": 4.194187124154442e-05, "loss": 2.8186, "step": 28821 }, { "epoch": 1.789186169222174, "grad_norm": 0.5774122432400387, "learning_rate": 4.193830694869491e-05, "loss": 2.7548, "step": 28822 }, { "epoch": 1.789248246321932, "grad_norm": 0.5528714714274054, "learning_rate": 4.193474269790521e-05, "loss": 2.847, "step": 28823 }, { "epoch": 1.7893103234216898, "grad_norm": 0.5866300940041473, "learning_rate": 4.193117848919389e-05, "loss": 2.7616, "step": 28824 }, { "epoch": 1.7893724005214477, "grad_norm": 0.5619119908649655, "learning_rate": 4.19276143225796e-05, "loss": 2.7461, "step": 28825 }, { "epoch": 1.7894344776212057, "grad_norm": 0.5436527691430548, "learning_rate": 4.19240501980809e-05, "loss": 2.7313, "step": 28826 }, { "epoch": 1.7894965547209636, "grad_norm": 0.5971308374596711, "learning_rate": 4.19204861157164e-05, "loss": 2.765, "step": 28827 }, { "epoch": 1.7895586318207215, "grad_norm": 0.5124213076863301, "learning_rate": 4.1916922075504665e-05, "loss": 2.718, "step": 28828 }, { "epoch": 1.7896207089204792, "grad_norm": 0.6783195629604514, "learning_rate": 4.1913358077464323e-05, "loss": 2.8321, "step": 28829 }, { "epoch": 1.7896827860202371, "grad_norm": 0.5319501246416599, "learning_rate": 4.1909794121613955e-05, "loss": 2.7213, "step": 28830 }, { "epoch": 1.789744863119995, "grad_norm": 0.6037617400708213, "learning_rate": 4.190623020797215e-05, "loss": 2.7441, "step": 28831 }, { "epoch": 1.789806940219753, "grad_norm": 0.5273241392657133, "learning_rate": 4.190266633655751e-05, "loss": 2.824, "step": 28832 }, { "epoch": 1.7898690173195109, "grad_norm": 0.571576898559279, "learning_rate": 4.189910250738863e-05, "loss": 2.7661, "step": 28833 }, { "epoch": 1.7899310944192688, "grad_norm": 0.5411608945080342, "learning_rate": 4.1895538720484087e-05, "loss": 2.688, "step": 28834 }, { "epoch": 1.7899931715190265, "grad_norm": 0.5918312188686082, "learning_rate": 4.1891974975862505e-05, "loss": 2.7561, "step": 28835 }, { "epoch": 1.7900552486187844, "grad_norm": 0.5529512714470366, "learning_rate": 4.188841127354245e-05, "loss": 2.7698, "step": 28836 }, { "epoch": 1.7901173257185423, "grad_norm": 0.5939146365743023, "learning_rate": 4.188484761354251e-05, "loss": 2.7516, "step": 28837 }, { "epoch": 1.7901794028183002, "grad_norm": 0.5470725383937977, "learning_rate": 4.1881283995881304e-05, "loss": 2.7549, "step": 28838 }, { "epoch": 1.7902414799180582, "grad_norm": 0.5467715846782262, "learning_rate": 4.187772042057739e-05, "loss": 2.7339, "step": 28839 }, { "epoch": 1.790303557017816, "grad_norm": 0.5494093888318256, "learning_rate": 4.187415688764941e-05, "loss": 2.7507, "step": 28840 }, { "epoch": 1.790365634117574, "grad_norm": 0.5342057411772245, "learning_rate": 4.18705933971159e-05, "loss": 2.7711, "step": 28841 }, { "epoch": 1.790427711217332, "grad_norm": 0.5758766581811832, "learning_rate": 4.18670299489955e-05, "loss": 2.8038, "step": 28842 }, { "epoch": 1.7904897883170898, "grad_norm": 0.5044178398780828, "learning_rate": 4.1863466543306757e-05, "loss": 2.8165, "step": 28843 }, { "epoch": 1.7905518654168477, "grad_norm": 0.5352404173216373, "learning_rate": 4.18599031800683e-05, "loss": 2.6762, "step": 28844 }, { "epoch": 1.7906139425166057, "grad_norm": 0.5214462837852024, "learning_rate": 4.185633985929869e-05, "loss": 2.7785, "step": 28845 }, { "epoch": 1.7906760196163636, "grad_norm": 0.5534250641625943, "learning_rate": 4.185277658101655e-05, "loss": 2.7156, "step": 28846 }, { "epoch": 1.7907380967161215, "grad_norm": 0.5362631773052117, "learning_rate": 4.1849213345240446e-05, "loss": 2.7585, "step": 28847 }, { "epoch": 1.7908001738158794, "grad_norm": 0.5670645210047626, "learning_rate": 4.184565015198897e-05, "loss": 2.8292, "step": 28848 }, { "epoch": 1.7908622509156373, "grad_norm": 0.5319420570301147, "learning_rate": 4.184208700128072e-05, "loss": 2.7679, "step": 28849 }, { "epoch": 1.7909243280153953, "grad_norm": 0.5100634059573096, "learning_rate": 4.183852389313428e-05, "loss": 2.6618, "step": 28850 }, { "epoch": 1.7909864051151532, "grad_norm": 0.5634772765156736, "learning_rate": 4.183496082756825e-05, "loss": 2.735, "step": 28851 }, { "epoch": 1.7910484822149109, "grad_norm": 0.5632712322301936, "learning_rate": 4.18313978046012e-05, "loss": 2.7547, "step": 28852 }, { "epoch": 1.7911105593146688, "grad_norm": 0.5622112274675941, "learning_rate": 4.182783482425174e-05, "loss": 2.8053, "step": 28853 }, { "epoch": 1.7911726364144267, "grad_norm": 0.6023317595598356, "learning_rate": 4.1824271886538435e-05, "loss": 2.6769, "step": 28854 }, { "epoch": 1.7912347135141846, "grad_norm": 0.5365236452461232, "learning_rate": 4.182070899147991e-05, "loss": 2.769, "step": 28855 }, { "epoch": 1.7912967906139425, "grad_norm": 0.6746203865075879, "learning_rate": 4.181714613909471e-05, "loss": 2.7336, "step": 28856 }, { "epoch": 1.7913588677137005, "grad_norm": 0.5863661660313897, "learning_rate": 4.181358332940144e-05, "loss": 2.7483, "step": 28857 }, { "epoch": 1.7914209448134581, "grad_norm": 0.584597874535564, "learning_rate": 4.181002056241871e-05, "loss": 2.6862, "step": 28858 }, { "epoch": 1.791483021913216, "grad_norm": 0.5922048691992896, "learning_rate": 4.1806457838165084e-05, "loss": 2.7469, "step": 28859 }, { "epoch": 1.791545099012974, "grad_norm": 0.5526624976558071, "learning_rate": 4.180289515665917e-05, "loss": 2.6965, "step": 28860 }, { "epoch": 1.791607176112732, "grad_norm": 0.5794625917905769, "learning_rate": 4.1799332517919535e-05, "loss": 2.7632, "step": 28861 }, { "epoch": 1.7916692532124898, "grad_norm": 0.5345307474543783, "learning_rate": 4.179576992196477e-05, "loss": 2.7782, "step": 28862 }, { "epoch": 1.7917313303122477, "grad_norm": 0.5293740642048425, "learning_rate": 4.179220736881347e-05, "loss": 2.8308, "step": 28863 }, { "epoch": 1.7917934074120057, "grad_norm": 0.5604838691418396, "learning_rate": 4.1788644858484205e-05, "loss": 2.6913, "step": 28864 }, { "epoch": 1.7918554845117636, "grad_norm": 0.4958807871120007, "learning_rate": 4.178508239099559e-05, "loss": 2.7271, "step": 28865 }, { "epoch": 1.7919175616115215, "grad_norm": 0.5699579626192277, "learning_rate": 4.178151996636618e-05, "loss": 2.7778, "step": 28866 }, { "epoch": 1.7919796387112794, "grad_norm": 0.5071382594811633, "learning_rate": 4.1777957584614586e-05, "loss": 2.7124, "step": 28867 }, { "epoch": 1.7920417158110373, "grad_norm": 0.5533037252528602, "learning_rate": 4.177439524575938e-05, "loss": 2.9169, "step": 28868 }, { "epoch": 1.7921037929107952, "grad_norm": 0.5051097249282294, "learning_rate": 4.177083294981915e-05, "loss": 2.7123, "step": 28869 }, { "epoch": 1.7921658700105532, "grad_norm": 0.5303565253538103, "learning_rate": 4.1767270696812474e-05, "loss": 2.7431, "step": 28870 }, { "epoch": 1.792227947110311, "grad_norm": 0.5672319641759777, "learning_rate": 4.1763708486757956e-05, "loss": 2.7329, "step": 28871 }, { "epoch": 1.792290024210069, "grad_norm": 0.530534374464467, "learning_rate": 4.176014631967418e-05, "loss": 2.7316, "step": 28872 }, { "epoch": 1.792352101309827, "grad_norm": 0.5620093855158995, "learning_rate": 4.175658419557969e-05, "loss": 2.8094, "step": 28873 }, { "epoch": 1.7924141784095848, "grad_norm": 0.5399342826750055, "learning_rate": 4.175302211449314e-05, "loss": 2.6822, "step": 28874 }, { "epoch": 1.7924762555093428, "grad_norm": 0.582248570366329, "learning_rate": 4.174946007643304e-05, "loss": 2.7666, "step": 28875 }, { "epoch": 1.7925383326091004, "grad_norm": 0.5808689622017287, "learning_rate": 4.174589808141803e-05, "loss": 2.7094, "step": 28876 }, { "epoch": 1.7926004097088584, "grad_norm": 0.5822185240673392, "learning_rate": 4.174233612946666e-05, "loss": 2.7774, "step": 28877 }, { "epoch": 1.7926624868086163, "grad_norm": 0.5981833433677022, "learning_rate": 4.1738774220597544e-05, "loss": 2.7233, "step": 28878 }, { "epoch": 1.7927245639083742, "grad_norm": 0.5974830925117482, "learning_rate": 4.1735212354829236e-05, "loss": 2.8249, "step": 28879 }, { "epoch": 1.7927866410081321, "grad_norm": 0.5408185744610677, "learning_rate": 4.173165053218034e-05, "loss": 2.7138, "step": 28880 }, { "epoch": 1.79284871810789, "grad_norm": 0.5444723483263783, "learning_rate": 4.172808875266942e-05, "loss": 2.7617, "step": 28881 }, { "epoch": 1.7929107952076477, "grad_norm": 0.5857416970205463, "learning_rate": 4.172452701631508e-05, "loss": 2.7317, "step": 28882 }, { "epoch": 1.7929728723074057, "grad_norm": 0.5328035359453598, "learning_rate": 4.1720965323135893e-05, "loss": 2.7237, "step": 28883 }, { "epoch": 1.7930349494071636, "grad_norm": 0.6005035095715777, "learning_rate": 4.171740367315042e-05, "loss": 2.7931, "step": 28884 }, { "epoch": 1.7930970265069215, "grad_norm": 0.5561535226543146, "learning_rate": 4.1713842066377285e-05, "loss": 2.7317, "step": 28885 }, { "epoch": 1.7931591036066794, "grad_norm": 0.5721963573626534, "learning_rate": 4.171028050283503e-05, "loss": 2.7096, "step": 28886 }, { "epoch": 1.7932211807064373, "grad_norm": 0.5690236093721583, "learning_rate": 4.170671898254227e-05, "loss": 2.7741, "step": 28887 }, { "epoch": 1.7932832578061952, "grad_norm": 0.6856045168909244, "learning_rate": 4.170315750551756e-05, "loss": 2.8128, "step": 28888 }, { "epoch": 1.7933453349059532, "grad_norm": 0.5843484311515329, "learning_rate": 4.169959607177948e-05, "loss": 2.7618, "step": 28889 }, { "epoch": 1.793407412005711, "grad_norm": 0.5999453718430219, "learning_rate": 4.1696034681346637e-05, "loss": 2.7174, "step": 28890 }, { "epoch": 1.793469489105469, "grad_norm": 0.639100144077357, "learning_rate": 4.169247333423759e-05, "loss": 2.7801, "step": 28891 }, { "epoch": 1.793531566205227, "grad_norm": 0.5713273777298935, "learning_rate": 4.168891203047094e-05, "loss": 2.7534, "step": 28892 }, { "epoch": 1.7935936433049848, "grad_norm": 0.6385870594320999, "learning_rate": 4.1685350770065234e-05, "loss": 2.8285, "step": 28893 }, { "epoch": 1.7936557204047427, "grad_norm": 0.5376294202021961, "learning_rate": 4.168178955303909e-05, "loss": 2.68, "step": 28894 }, { "epoch": 1.7937177975045007, "grad_norm": 0.6368183455944438, "learning_rate": 4.167822837941106e-05, "loss": 2.7598, "step": 28895 }, { "epoch": 1.7937798746042586, "grad_norm": 0.5567722365739611, "learning_rate": 4.167466724919974e-05, "loss": 2.7508, "step": 28896 }, { "epoch": 1.7938419517040165, "grad_norm": 0.5924447167837621, "learning_rate": 4.16711061624237e-05, "loss": 2.7295, "step": 28897 }, { "epoch": 1.7939040288037744, "grad_norm": 0.604889204701591, "learning_rate": 4.1667545119101516e-05, "loss": 2.7427, "step": 28898 }, { "epoch": 1.7939661059035323, "grad_norm": 0.5961746100981725, "learning_rate": 4.166398411925178e-05, "loss": 2.7173, "step": 28899 }, { "epoch": 1.79402818300329, "grad_norm": 0.5322201229720812, "learning_rate": 4.166042316289305e-05, "loss": 2.7415, "step": 28900 }, { "epoch": 1.794090260103048, "grad_norm": 0.6533291392735071, "learning_rate": 4.165686225004393e-05, "loss": 2.6887, "step": 28901 }, { "epoch": 1.7941523372028059, "grad_norm": 0.5515701650027472, "learning_rate": 4.165330138072297e-05, "loss": 2.7179, "step": 28902 }, { "epoch": 1.7942144143025638, "grad_norm": 0.6207053691614585, "learning_rate": 4.164974055494878e-05, "loss": 2.7531, "step": 28903 }, { "epoch": 1.7942764914023217, "grad_norm": 0.5749978988028572, "learning_rate": 4.164617977273991e-05, "loss": 2.7774, "step": 28904 }, { "epoch": 1.7943385685020796, "grad_norm": 0.5978025868097915, "learning_rate": 4.164261903411495e-05, "loss": 2.6595, "step": 28905 }, { "epoch": 1.7944006456018373, "grad_norm": 0.5920348239850204, "learning_rate": 4.1639058339092474e-05, "loss": 2.8293, "step": 28906 }, { "epoch": 1.7944627227015952, "grad_norm": 0.5348582308093558, "learning_rate": 4.1635497687691064e-05, "loss": 2.734, "step": 28907 }, { "epoch": 1.7945247998013532, "grad_norm": 0.619521616491628, "learning_rate": 4.163193707992929e-05, "loss": 2.7659, "step": 28908 }, { "epoch": 1.794586876901111, "grad_norm": 0.58029556052485, "learning_rate": 4.1628376515825735e-05, "loss": 2.7641, "step": 28909 }, { "epoch": 1.794648954000869, "grad_norm": 0.6097540287965478, "learning_rate": 4.162481599539898e-05, "loss": 2.7379, "step": 28910 }, { "epoch": 1.794711031100627, "grad_norm": 0.5189040277072239, "learning_rate": 4.1621255518667565e-05, "loss": 2.6984, "step": 28911 }, { "epoch": 1.7947731082003848, "grad_norm": 0.5777601688964151, "learning_rate": 4.161769508565012e-05, "loss": 2.6956, "step": 28912 }, { "epoch": 1.7948351853001427, "grad_norm": 0.6132762933032675, "learning_rate": 4.1614134696365185e-05, "loss": 2.7816, "step": 28913 }, { "epoch": 1.7948972623999007, "grad_norm": 0.558560190326431, "learning_rate": 4.161057435083135e-05, "loss": 2.707, "step": 28914 }, { "epoch": 1.7949593394996586, "grad_norm": 0.6141859294513758, "learning_rate": 4.160701404906718e-05, "loss": 2.7002, "step": 28915 }, { "epoch": 1.7950214165994165, "grad_norm": 0.5893334208412163, "learning_rate": 4.160345379109126e-05, "loss": 2.6819, "step": 28916 }, { "epoch": 1.7950834936991744, "grad_norm": 0.580073706613276, "learning_rate": 4.159989357692215e-05, "loss": 2.7351, "step": 28917 }, { "epoch": 1.7951455707989323, "grad_norm": 0.5785862912349629, "learning_rate": 4.159633340657845e-05, "loss": 2.8511, "step": 28918 }, { "epoch": 1.7952076478986903, "grad_norm": 0.5210579546711511, "learning_rate": 4.15927732800787e-05, "loss": 2.7355, "step": 28919 }, { "epoch": 1.7952697249984482, "grad_norm": 0.5936467683294666, "learning_rate": 4.1589213197441503e-05, "loss": 2.8228, "step": 28920 }, { "epoch": 1.795331802098206, "grad_norm": 0.5822748020257115, "learning_rate": 4.158565315868543e-05, "loss": 2.8695, "step": 28921 }, { "epoch": 1.795393879197964, "grad_norm": 0.6159672807048239, "learning_rate": 4.158209316382903e-05, "loss": 2.6525, "step": 28922 }, { "epoch": 1.795455956297722, "grad_norm": 0.5453567410423946, "learning_rate": 4.157853321289089e-05, "loss": 2.798, "step": 28923 }, { "epoch": 1.7955180333974796, "grad_norm": 0.5791179017822343, "learning_rate": 4.157497330588961e-05, "loss": 2.7161, "step": 28924 }, { "epoch": 1.7955801104972375, "grad_norm": 0.5873311063801478, "learning_rate": 4.157141344284372e-05, "loss": 2.8004, "step": 28925 }, { "epoch": 1.7956421875969955, "grad_norm": 0.5855684262432951, "learning_rate": 4.1567853623771824e-05, "loss": 2.7876, "step": 28926 }, { "epoch": 1.7957042646967534, "grad_norm": 0.5118262199941456, "learning_rate": 4.156429384869247e-05, "loss": 2.7377, "step": 28927 }, { "epoch": 1.7957663417965113, "grad_norm": 0.5533340490099097, "learning_rate": 4.156073411762427e-05, "loss": 2.6531, "step": 28928 }, { "epoch": 1.7958284188962692, "grad_norm": 0.5699343015219386, "learning_rate": 4.155717443058574e-05, "loss": 2.7048, "step": 28929 }, { "epoch": 1.795890495996027, "grad_norm": 0.5217481367592472, "learning_rate": 4.155361478759549e-05, "loss": 2.737, "step": 28930 }, { "epoch": 1.7959525730957848, "grad_norm": 0.5587138600364003, "learning_rate": 4.155005518867209e-05, "loss": 2.6903, "step": 28931 }, { "epoch": 1.7960146501955427, "grad_norm": 0.5637937872601527, "learning_rate": 4.1546495633834096e-05, "loss": 2.776, "step": 28932 }, { "epoch": 1.7960767272953007, "grad_norm": 0.5315568542629175, "learning_rate": 4.1542936123100095e-05, "loss": 2.7475, "step": 28933 }, { "epoch": 1.7961388043950586, "grad_norm": 0.556369970007095, "learning_rate": 4.1539376656488635e-05, "loss": 2.789, "step": 28934 }, { "epoch": 1.7962008814948165, "grad_norm": 0.6012032662388597, "learning_rate": 4.153581723401831e-05, "loss": 2.7525, "step": 28935 }, { "epoch": 1.7962629585945744, "grad_norm": 0.5283280687591322, "learning_rate": 4.153225785570768e-05, "loss": 2.8387, "step": 28936 }, { "epoch": 1.7963250356943323, "grad_norm": 0.6208582497616905, "learning_rate": 4.152869852157532e-05, "loss": 2.7522, "step": 28937 }, { "epoch": 1.7963871127940902, "grad_norm": 0.47402726884259416, "learning_rate": 4.152513923163979e-05, "loss": 2.7441, "step": 28938 }, { "epoch": 1.7964491898938482, "grad_norm": 0.6254147345656866, "learning_rate": 4.152157998591967e-05, "loss": 2.7625, "step": 28939 }, { "epoch": 1.796511266993606, "grad_norm": 0.5200126253484374, "learning_rate": 4.151802078443351e-05, "loss": 2.6588, "step": 28940 }, { "epoch": 1.796573344093364, "grad_norm": 0.558263908050689, "learning_rate": 4.151446162719992e-05, "loss": 2.7113, "step": 28941 }, { "epoch": 1.796635421193122, "grad_norm": 0.5347170175846785, "learning_rate": 4.1510902514237424e-05, "loss": 2.7745, "step": 28942 }, { "epoch": 1.7966974982928798, "grad_norm": 0.5460479204105247, "learning_rate": 4.1507343445564625e-05, "loss": 2.7531, "step": 28943 }, { "epoch": 1.7967595753926378, "grad_norm": 0.5439683782594105, "learning_rate": 4.1503784421200066e-05, "loss": 2.6869, "step": 28944 }, { "epoch": 1.7968216524923957, "grad_norm": 0.5029032876205923, "learning_rate": 4.150022544116233e-05, "loss": 2.7648, "step": 28945 }, { "epoch": 1.7968837295921536, "grad_norm": 0.5904170906623533, "learning_rate": 4.149666650546999e-05, "loss": 2.7871, "step": 28946 }, { "epoch": 1.7969458066919115, "grad_norm": 0.47661701265831136, "learning_rate": 4.1493107614141594e-05, "loss": 2.6874, "step": 28947 }, { "epoch": 1.7970078837916692, "grad_norm": 0.5699423542487393, "learning_rate": 4.148954876719572e-05, "loss": 2.7826, "step": 28948 }, { "epoch": 1.7970699608914271, "grad_norm": 0.543734250991576, "learning_rate": 4.148598996465093e-05, "loss": 2.6796, "step": 28949 }, { "epoch": 1.797132037991185, "grad_norm": 0.5379937953254387, "learning_rate": 4.1482431206525815e-05, "loss": 2.7872, "step": 28950 }, { "epoch": 1.797194115090943, "grad_norm": 0.5513803997341845, "learning_rate": 4.147887249283891e-05, "loss": 2.7477, "step": 28951 }, { "epoch": 1.7972561921907009, "grad_norm": 0.5060637661330449, "learning_rate": 4.147531382360881e-05, "loss": 2.7549, "step": 28952 }, { "epoch": 1.7973182692904588, "grad_norm": 0.5084591489659922, "learning_rate": 4.147175519885405e-05, "loss": 2.7274, "step": 28953 }, { "epoch": 1.7973803463902165, "grad_norm": 0.5148596719181396, "learning_rate": 4.146819661859323e-05, "loss": 2.6114, "step": 28954 }, { "epoch": 1.7974424234899744, "grad_norm": 0.5485168606106199, "learning_rate": 4.146463808284488e-05, "loss": 2.8264, "step": 28955 }, { "epoch": 1.7975045005897323, "grad_norm": 0.566892612398171, "learning_rate": 4.146107959162758e-05, "loss": 2.7569, "step": 28956 }, { "epoch": 1.7975665776894902, "grad_norm": 0.568253340120984, "learning_rate": 4.145752114495993e-05, "loss": 2.7068, "step": 28957 }, { "epoch": 1.7976286547892482, "grad_norm": 0.5395866554727489, "learning_rate": 4.1453962742860454e-05, "loss": 2.7567, "step": 28958 }, { "epoch": 1.797690731889006, "grad_norm": 0.5753610627970331, "learning_rate": 4.145040438534772e-05, "loss": 2.7248, "step": 28959 }, { "epoch": 1.797752808988764, "grad_norm": 0.5468103915605653, "learning_rate": 4.1446846072440314e-05, "loss": 2.7532, "step": 28960 }, { "epoch": 1.797814886088522, "grad_norm": 0.5124818193886076, "learning_rate": 4.1443287804156784e-05, "loss": 2.759, "step": 28961 }, { "epoch": 1.7978769631882798, "grad_norm": 0.5626072781307314, "learning_rate": 4.14397295805157e-05, "loss": 2.8167, "step": 28962 }, { "epoch": 1.7979390402880377, "grad_norm": 0.48206240276650736, "learning_rate": 4.143617140153562e-05, "loss": 2.7285, "step": 28963 }, { "epoch": 1.7980011173877957, "grad_norm": 0.5684579717840983, "learning_rate": 4.143261326723512e-05, "loss": 2.7756, "step": 28964 }, { "epoch": 1.7980631944875536, "grad_norm": 0.531494140625, "learning_rate": 4.1429055177632746e-05, "loss": 2.6752, "step": 28965 }, { "epoch": 1.7981252715873115, "grad_norm": 0.5943781140443433, "learning_rate": 4.142549713274709e-05, "loss": 2.7297, "step": 28966 }, { "epoch": 1.7981873486870694, "grad_norm": 0.5852344618555223, "learning_rate": 4.1421939132596676e-05, "loss": 2.768, "step": 28967 }, { "epoch": 1.7982494257868273, "grad_norm": 0.5338878496910608, "learning_rate": 4.1418381177200103e-05, "loss": 2.7344, "step": 28968 }, { "epoch": 1.7983115028865853, "grad_norm": 0.5239799592375375, "learning_rate": 4.141482326657591e-05, "loss": 2.694, "step": 28969 }, { "epoch": 1.7983735799863432, "grad_norm": 0.5239832865218993, "learning_rate": 4.141126540074268e-05, "loss": 2.8196, "step": 28970 }, { "epoch": 1.798435657086101, "grad_norm": 0.5489046036727891, "learning_rate": 4.140770757971896e-05, "loss": 2.7114, "step": 28971 }, { "epoch": 1.7984977341858588, "grad_norm": 0.5874693365414224, "learning_rate": 4.1404149803523304e-05, "loss": 2.7399, "step": 28972 }, { "epoch": 1.7985598112856167, "grad_norm": 0.53387869491674, "learning_rate": 4.140059207217429e-05, "loss": 2.8165, "step": 28973 }, { "epoch": 1.7986218883853746, "grad_norm": 0.6293938209878895, "learning_rate": 4.139703438569047e-05, "loss": 2.7943, "step": 28974 }, { "epoch": 1.7986839654851325, "grad_norm": 0.5559881837158692, "learning_rate": 4.139347674409043e-05, "loss": 2.6348, "step": 28975 }, { "epoch": 1.7987460425848905, "grad_norm": 0.6337107417495426, "learning_rate": 4.1389919147392684e-05, "loss": 2.8712, "step": 28976 }, { "epoch": 1.7988081196846484, "grad_norm": 0.596731305426046, "learning_rate": 4.138636159561584e-05, "loss": 2.7161, "step": 28977 }, { "epoch": 1.798870196784406, "grad_norm": 0.5593616835644759, "learning_rate": 4.138280408877843e-05, "loss": 2.729, "step": 28978 }, { "epoch": 1.798932273884164, "grad_norm": 0.6637682318832567, "learning_rate": 4.137924662689903e-05, "loss": 2.7816, "step": 28979 }, { "epoch": 1.798994350983922, "grad_norm": 0.5583532389133781, "learning_rate": 4.1375689209996186e-05, "loss": 2.7715, "step": 28980 }, { "epoch": 1.7990564280836798, "grad_norm": 0.5423406357210073, "learning_rate": 4.137213183808847e-05, "loss": 2.7511, "step": 28981 }, { "epoch": 1.7991185051834377, "grad_norm": 0.6192037989399438, "learning_rate": 4.136857451119444e-05, "loss": 2.8138, "step": 28982 }, { "epoch": 1.7991805822831957, "grad_norm": 0.5137068007746933, "learning_rate": 4.136501722933263e-05, "loss": 2.7353, "step": 28983 }, { "epoch": 1.7992426593829536, "grad_norm": 0.6338759070382223, "learning_rate": 4.1361459992521645e-05, "loss": 2.8013, "step": 28984 }, { "epoch": 1.7993047364827115, "grad_norm": 0.584072527403909, "learning_rate": 4.135790280078001e-05, "loss": 2.6706, "step": 28985 }, { "epoch": 1.7993668135824694, "grad_norm": 0.6430342194794096, "learning_rate": 4.1354345654126304e-05, "loss": 2.8013, "step": 28986 }, { "epoch": 1.7994288906822273, "grad_norm": 0.5442111394778847, "learning_rate": 4.1350788552579064e-05, "loss": 2.7336, "step": 28987 }, { "epoch": 1.7994909677819853, "grad_norm": 0.6227824687464432, "learning_rate": 4.134723149615685e-05, "loss": 2.7442, "step": 28988 }, { "epoch": 1.7995530448817432, "grad_norm": 0.49795839608341297, "learning_rate": 4.134367448487824e-05, "loss": 2.7208, "step": 28989 }, { "epoch": 1.799615121981501, "grad_norm": 0.6002396988896379, "learning_rate": 4.134011751876178e-05, "loss": 2.7586, "step": 28990 }, { "epoch": 1.799677199081259, "grad_norm": 0.5378292738555698, "learning_rate": 4.133656059782604e-05, "loss": 2.5993, "step": 28991 }, { "epoch": 1.799739276181017, "grad_norm": 0.5436846726735737, "learning_rate": 4.133300372208956e-05, "loss": 2.6489, "step": 28992 }, { "epoch": 1.7998013532807748, "grad_norm": 0.5292432976101462, "learning_rate": 4.1329446891570905e-05, "loss": 2.7396, "step": 28993 }, { "epoch": 1.7998634303805328, "grad_norm": 0.6091538786122049, "learning_rate": 4.1325890106288624e-05, "loss": 2.7413, "step": 28994 }, { "epoch": 1.7999255074802907, "grad_norm": 0.5239145752347438, "learning_rate": 4.132233336626129e-05, "loss": 2.6636, "step": 28995 }, { "epoch": 1.7999875845800484, "grad_norm": 0.6186758276284084, "learning_rate": 4.1318776671507455e-05, "loss": 2.7562, "step": 28996 }, { "epoch": 1.8000496616798063, "grad_norm": 0.5608044873065564, "learning_rate": 4.1315220022045654e-05, "loss": 2.79, "step": 28997 }, { "epoch": 1.8001117387795642, "grad_norm": 0.5001683249858673, "learning_rate": 4.131166341789447e-05, "loss": 2.7132, "step": 28998 }, { "epoch": 1.8001738158793221, "grad_norm": 0.6258012642201802, "learning_rate": 4.130810685907243e-05, "loss": 2.7511, "step": 28999 }, { "epoch": 1.80023589297908, "grad_norm": 0.504571962861087, "learning_rate": 4.1304550345598124e-05, "loss": 2.8662, "step": 29000 }, { "epoch": 1.800297970078838, "grad_norm": 0.5772500505410193, "learning_rate": 4.130099387749008e-05, "loss": 2.8428, "step": 29001 }, { "epoch": 1.8003600471785957, "grad_norm": 0.544788467320031, "learning_rate": 4.1297437454766877e-05, "loss": 2.7093, "step": 29002 }, { "epoch": 1.8004221242783536, "grad_norm": 0.5277608564101374, "learning_rate": 4.129388107744703e-05, "loss": 2.7088, "step": 29003 }, { "epoch": 1.8004842013781115, "grad_norm": 0.5273260608108117, "learning_rate": 4.129032474554915e-05, "loss": 2.69, "step": 29004 }, { "epoch": 1.8005462784778694, "grad_norm": 0.6195610856981238, "learning_rate": 4.128676845909173e-05, "loss": 2.7979, "step": 29005 }, { "epoch": 1.8006083555776273, "grad_norm": 0.5865679337539959, "learning_rate": 4.128321221809338e-05, "loss": 2.7379, "step": 29006 }, { "epoch": 1.8006704326773852, "grad_norm": 0.5626190377482887, "learning_rate": 4.127965602257262e-05, "loss": 2.7812, "step": 29007 }, { "epoch": 1.8007325097771432, "grad_norm": 0.5468623023602627, "learning_rate": 4.1276099872548e-05, "loss": 2.8238, "step": 29008 }, { "epoch": 1.800794586876901, "grad_norm": 0.5430960746059958, "learning_rate": 4.12725437680381e-05, "loss": 2.7777, "step": 29009 }, { "epoch": 1.800856663976659, "grad_norm": 0.5054814757170542, "learning_rate": 4.126898770906144e-05, "loss": 2.6941, "step": 29010 }, { "epoch": 1.800918741076417, "grad_norm": 0.5835732125750116, "learning_rate": 4.12654316956366e-05, "loss": 2.8369, "step": 29011 }, { "epoch": 1.8009808181761748, "grad_norm": 0.48629290288511406, "learning_rate": 4.126187572778212e-05, "loss": 2.7887, "step": 29012 }, { "epoch": 1.8010428952759328, "grad_norm": 0.5882243479734043, "learning_rate": 4.125831980551657e-05, "loss": 2.7623, "step": 29013 }, { "epoch": 1.8011049723756907, "grad_norm": 0.5168601184072625, "learning_rate": 4.125476392885847e-05, "loss": 2.7457, "step": 29014 }, { "epoch": 1.8011670494754486, "grad_norm": 0.5706798990002673, "learning_rate": 4.12512080978264e-05, "loss": 2.7282, "step": 29015 }, { "epoch": 1.8012291265752065, "grad_norm": 0.5556961881606031, "learning_rate": 4.1247652312438894e-05, "loss": 2.6876, "step": 29016 }, { "epoch": 1.8012912036749644, "grad_norm": 0.5810680996925673, "learning_rate": 4.124409657271452e-05, "loss": 2.7733, "step": 29017 }, { "epoch": 1.8013532807747223, "grad_norm": 0.6036080595172335, "learning_rate": 4.124054087867182e-05, "loss": 2.7911, "step": 29018 }, { "epoch": 1.8014153578744803, "grad_norm": 0.5600353392451772, "learning_rate": 4.123698523032934e-05, "loss": 2.7291, "step": 29019 }, { "epoch": 1.801477434974238, "grad_norm": 0.5647588251833541, "learning_rate": 4.1233429627705614e-05, "loss": 2.695, "step": 29020 }, { "epoch": 1.8015395120739959, "grad_norm": 0.569588475956429, "learning_rate": 4.122987407081924e-05, "loss": 2.7174, "step": 29021 }, { "epoch": 1.8016015891737538, "grad_norm": 0.5591942020814914, "learning_rate": 4.122631855968873e-05, "loss": 2.7475, "step": 29022 }, { "epoch": 1.8016636662735117, "grad_norm": 0.6006915488078373, "learning_rate": 4.1222763094332657e-05, "loss": 2.7139, "step": 29023 }, { "epoch": 1.8017257433732696, "grad_norm": 0.5648778617551723, "learning_rate": 4.1219207674769545e-05, "loss": 2.7776, "step": 29024 }, { "epoch": 1.8017878204730275, "grad_norm": 0.5027847229259528, "learning_rate": 4.1215652301017974e-05, "loss": 2.7714, "step": 29025 }, { "epoch": 1.8018498975727852, "grad_norm": 0.5504174891265261, "learning_rate": 4.121209697309646e-05, "loss": 2.742, "step": 29026 }, { "epoch": 1.8019119746725432, "grad_norm": 0.6065636855619421, "learning_rate": 4.1208541691023596e-05, "loss": 2.7405, "step": 29027 }, { "epoch": 1.801974051772301, "grad_norm": 0.5227080714332935, "learning_rate": 4.1204986454817876e-05, "loss": 2.7268, "step": 29028 }, { "epoch": 1.802036128872059, "grad_norm": 0.5155136103678717, "learning_rate": 4.1201431264497896e-05, "loss": 2.7294, "step": 29029 }, { "epoch": 1.802098205971817, "grad_norm": 0.5245835866717918, "learning_rate": 4.1197876120082175e-05, "loss": 2.7848, "step": 29030 }, { "epoch": 1.8021602830715748, "grad_norm": 0.4954725867212797, "learning_rate": 4.1194321021589274e-05, "loss": 2.7709, "step": 29031 }, { "epoch": 1.8022223601713327, "grad_norm": 0.5178307258745758, "learning_rate": 4.1190765969037746e-05, "loss": 2.735, "step": 29032 }, { "epoch": 1.8022844372710907, "grad_norm": 0.49931133050575105, "learning_rate": 4.118721096244612e-05, "loss": 2.6779, "step": 29033 }, { "epoch": 1.8023465143708486, "grad_norm": 0.4902850121110562, "learning_rate": 4.118365600183296e-05, "loss": 2.7276, "step": 29034 }, { "epoch": 1.8024085914706065, "grad_norm": 0.5353938222778809, "learning_rate": 4.1180101087216806e-05, "loss": 2.744, "step": 29035 }, { "epoch": 1.8024706685703644, "grad_norm": 0.5506363536084194, "learning_rate": 4.117654621861621e-05, "loss": 2.7232, "step": 29036 }, { "epoch": 1.8025327456701223, "grad_norm": 0.5279324107041654, "learning_rate": 4.11729913960497e-05, "loss": 2.7662, "step": 29037 }, { "epoch": 1.8025948227698803, "grad_norm": 0.5676807140199682, "learning_rate": 4.1169436619535853e-05, "loss": 2.7636, "step": 29038 }, { "epoch": 1.8026568998696382, "grad_norm": 0.5552359274137605, "learning_rate": 4.1165881889093174e-05, "loss": 2.7697, "step": 29039 }, { "epoch": 1.802718976969396, "grad_norm": 0.5328283425042593, "learning_rate": 4.116232720474026e-05, "loss": 2.7807, "step": 29040 }, { "epoch": 1.802781054069154, "grad_norm": 0.5644320843744822, "learning_rate": 4.115877256649562e-05, "loss": 2.7089, "step": 29041 }, { "epoch": 1.802843131168912, "grad_norm": 0.4994005842184057, "learning_rate": 4.1155217974377814e-05, "loss": 2.7634, "step": 29042 }, { "epoch": 1.8029052082686698, "grad_norm": 0.5255342592075081, "learning_rate": 4.1151663428405385e-05, "loss": 2.6134, "step": 29043 }, { "epoch": 1.8029672853684275, "grad_norm": 0.49847807823675916, "learning_rate": 4.114810892859686e-05, "loss": 2.6412, "step": 29044 }, { "epoch": 1.8030293624681855, "grad_norm": 0.5070257107236662, "learning_rate": 4.114455447497081e-05, "loss": 2.7608, "step": 29045 }, { "epoch": 1.8030914395679434, "grad_norm": 0.575136968636129, "learning_rate": 4.114100006754577e-05, "loss": 2.6755, "step": 29046 }, { "epoch": 1.8031535166677013, "grad_norm": 0.5336952872453666, "learning_rate": 4.113744570634029e-05, "loss": 2.717, "step": 29047 }, { "epoch": 1.8032155937674592, "grad_norm": 0.5593615770061394, "learning_rate": 4.113389139137289e-05, "loss": 2.8024, "step": 29048 }, { "epoch": 1.8032776708672171, "grad_norm": 0.5490653994662261, "learning_rate": 4.1130337122662144e-05, "loss": 2.7667, "step": 29049 }, { "epoch": 1.8033397479669748, "grad_norm": 0.5430055235341184, "learning_rate": 4.1126782900226574e-05, "loss": 2.7043, "step": 29050 }, { "epoch": 1.8034018250667327, "grad_norm": 0.5601115912560851, "learning_rate": 4.112322872408474e-05, "loss": 2.7667, "step": 29051 }, { "epoch": 1.8034639021664907, "grad_norm": 0.5557921518410006, "learning_rate": 4.111967459425517e-05, "loss": 2.7159, "step": 29052 }, { "epoch": 1.8035259792662486, "grad_norm": 0.6189281929515889, "learning_rate": 4.11161205107564e-05, "loss": 2.7345, "step": 29053 }, { "epoch": 1.8035880563660065, "grad_norm": 0.534115301014235, "learning_rate": 4.1112566473607004e-05, "loss": 2.7161, "step": 29054 }, { "epoch": 1.8036501334657644, "grad_norm": 0.5829285483095771, "learning_rate": 4.1109012482825506e-05, "loss": 2.6104, "step": 29055 }, { "epoch": 1.8037122105655223, "grad_norm": 0.5965196610301791, "learning_rate": 4.1105458538430445e-05, "loss": 2.6454, "step": 29056 }, { "epoch": 1.8037742876652803, "grad_norm": 0.5634884890875884, "learning_rate": 4.1101904640440376e-05, "loss": 2.6805, "step": 29057 }, { "epoch": 1.8038363647650382, "grad_norm": 0.5696971395307915, "learning_rate": 4.109835078887382e-05, "loss": 2.7598, "step": 29058 }, { "epoch": 1.803898441864796, "grad_norm": 0.6182259382524103, "learning_rate": 4.109479698374933e-05, "loss": 2.8266, "step": 29059 }, { "epoch": 1.803960518964554, "grad_norm": 0.5313414326692244, "learning_rate": 4.109124322508545e-05, "loss": 2.7563, "step": 29060 }, { "epoch": 1.804022596064312, "grad_norm": 0.6599854242637142, "learning_rate": 4.1087689512900716e-05, "loss": 2.7628, "step": 29061 }, { "epoch": 1.8040846731640698, "grad_norm": 0.574007501475067, "learning_rate": 4.108413584721366e-05, "loss": 2.7475, "step": 29062 }, { "epoch": 1.8041467502638278, "grad_norm": 0.6543763297503786, "learning_rate": 4.1080582228042843e-05, "loss": 2.7825, "step": 29063 }, { "epoch": 1.8042088273635857, "grad_norm": 0.5885161053893452, "learning_rate": 4.107702865540678e-05, "loss": 2.7592, "step": 29064 }, { "epoch": 1.8042709044633436, "grad_norm": 0.5837134814224506, "learning_rate": 4.1073475129324046e-05, "loss": 2.6882, "step": 29065 }, { "epoch": 1.8043329815631015, "grad_norm": 0.6234577223478556, "learning_rate": 4.1069921649813145e-05, "loss": 2.7605, "step": 29066 }, { "epoch": 1.8043950586628594, "grad_norm": 0.5555558727846299, "learning_rate": 4.1066368216892635e-05, "loss": 2.7219, "step": 29067 }, { "epoch": 1.8044571357626171, "grad_norm": 0.6468074947646022, "learning_rate": 4.1062814830581054e-05, "loss": 2.8114, "step": 29068 }, { "epoch": 1.804519212862375, "grad_norm": 0.5564944158909696, "learning_rate": 4.105926149089693e-05, "loss": 2.6926, "step": 29069 }, { "epoch": 1.804581289962133, "grad_norm": 0.7272726300765102, "learning_rate": 4.105570819785881e-05, "loss": 2.7822, "step": 29070 }, { "epoch": 1.8046433670618909, "grad_norm": 0.5318307227006175, "learning_rate": 4.1052154951485236e-05, "loss": 2.6557, "step": 29071 }, { "epoch": 1.8047054441616488, "grad_norm": 0.6488902568063333, "learning_rate": 4.104860175179474e-05, "loss": 2.6749, "step": 29072 }, { "epoch": 1.8047675212614067, "grad_norm": 0.5713243000868025, "learning_rate": 4.104504859880586e-05, "loss": 2.6889, "step": 29073 }, { "epoch": 1.8048295983611644, "grad_norm": 0.5565973633133848, "learning_rate": 4.104149549253714e-05, "loss": 2.6894, "step": 29074 }, { "epoch": 1.8048916754609223, "grad_norm": 0.49946061485399323, "learning_rate": 4.10379424330071e-05, "loss": 2.6203, "step": 29075 }, { "epoch": 1.8049537525606802, "grad_norm": 0.6204012723190891, "learning_rate": 4.103438942023431e-05, "loss": 2.7451, "step": 29076 }, { "epoch": 1.8050158296604382, "grad_norm": 0.526311811319174, "learning_rate": 4.1030836454237266e-05, "loss": 2.6666, "step": 29077 }, { "epoch": 1.805077906760196, "grad_norm": 0.5687170050555643, "learning_rate": 4.102728353503454e-05, "loss": 2.7222, "step": 29078 }, { "epoch": 1.805139983859954, "grad_norm": 0.6154813240589436, "learning_rate": 4.1023730662644656e-05, "loss": 2.827, "step": 29079 }, { "epoch": 1.805202060959712, "grad_norm": 0.5873387429931446, "learning_rate": 4.102017783708614e-05, "loss": 2.7689, "step": 29080 }, { "epoch": 1.8052641380594698, "grad_norm": 0.6294033621086385, "learning_rate": 4.1016625058377536e-05, "loss": 2.6908, "step": 29081 }, { "epoch": 1.8053262151592278, "grad_norm": 0.521430252066888, "learning_rate": 4.101307232653738e-05, "loss": 2.7477, "step": 29082 }, { "epoch": 1.8053882922589857, "grad_norm": 0.5527544590836146, "learning_rate": 4.100951964158422e-05, "loss": 2.7202, "step": 29083 }, { "epoch": 1.8054503693587436, "grad_norm": 0.5195970995480678, "learning_rate": 4.1005967003536555e-05, "loss": 2.8028, "step": 29084 }, { "epoch": 1.8055124464585015, "grad_norm": 0.5444333474197173, "learning_rate": 4.100241441241296e-05, "loss": 2.6761, "step": 29085 }, { "epoch": 1.8055745235582594, "grad_norm": 0.5875718022155806, "learning_rate": 4.099886186823193e-05, "loss": 2.7743, "step": 29086 }, { "epoch": 1.8056366006580173, "grad_norm": 0.49422401357881607, "learning_rate": 4.099530937101204e-05, "loss": 2.762, "step": 29087 }, { "epoch": 1.8056986777577753, "grad_norm": 0.6031247944411486, "learning_rate": 4.099175692077181e-05, "loss": 2.7285, "step": 29088 }, { "epoch": 1.8057607548575332, "grad_norm": 0.5401853946043185, "learning_rate": 4.098820451752977e-05, "loss": 2.6919, "step": 29089 }, { "epoch": 1.805822831957291, "grad_norm": 0.5408838435226531, "learning_rate": 4.098465216130446e-05, "loss": 2.7778, "step": 29090 }, { "epoch": 1.805884909057049, "grad_norm": 0.5273535833501594, "learning_rate": 4.0981099852114396e-05, "loss": 2.7085, "step": 29091 }, { "epoch": 1.8059469861568067, "grad_norm": 0.5523054527807839, "learning_rate": 4.097754758997814e-05, "loss": 2.7605, "step": 29092 }, { "epoch": 1.8060090632565646, "grad_norm": 0.4911130652177828, "learning_rate": 4.0973995374914195e-05, "loss": 2.695, "step": 29093 }, { "epoch": 1.8060711403563225, "grad_norm": 0.6010839862789089, "learning_rate": 4.0970443206941114e-05, "loss": 2.7997, "step": 29094 }, { "epoch": 1.8061332174560805, "grad_norm": 0.47577830973000096, "learning_rate": 4.096689108607742e-05, "loss": 2.7149, "step": 29095 }, { "epoch": 1.8061952945558384, "grad_norm": 0.6543080798655786, "learning_rate": 4.0963339012341644e-05, "loss": 2.7304, "step": 29096 }, { "epoch": 1.8062573716555963, "grad_norm": 0.5168088558362406, "learning_rate": 4.095978698575233e-05, "loss": 2.6977, "step": 29097 }, { "epoch": 1.806319448755354, "grad_norm": 0.5701242162226111, "learning_rate": 4.0956235006327995e-05, "loss": 2.7545, "step": 29098 }, { "epoch": 1.806381525855112, "grad_norm": 0.5490479487237078, "learning_rate": 4.0952683074087185e-05, "loss": 2.8108, "step": 29099 }, { "epoch": 1.8064436029548698, "grad_norm": 0.5960458488766522, "learning_rate": 4.094913118904841e-05, "loss": 2.7178, "step": 29100 }, { "epoch": 1.8065056800546277, "grad_norm": 0.5592675313593367, "learning_rate": 4.0945579351230234e-05, "loss": 2.7604, "step": 29101 }, { "epoch": 1.8065677571543857, "grad_norm": 0.5810901534853132, "learning_rate": 4.094202756065115e-05, "loss": 2.7296, "step": 29102 }, { "epoch": 1.8066298342541436, "grad_norm": 0.5606659553393291, "learning_rate": 4.0938475817329724e-05, "loss": 2.8034, "step": 29103 }, { "epoch": 1.8066919113539015, "grad_norm": 0.5866650451381267, "learning_rate": 4.0934924121284465e-05, "loss": 2.6632, "step": 29104 }, { "epoch": 1.8067539884536594, "grad_norm": 0.5470980870401877, "learning_rate": 4.0931372472533893e-05, "loss": 2.6242, "step": 29105 }, { "epoch": 1.8068160655534173, "grad_norm": 0.5167438332808433, "learning_rate": 4.092782087109657e-05, "loss": 2.763, "step": 29106 }, { "epoch": 1.8068781426531753, "grad_norm": 0.5736718200575511, "learning_rate": 4.0924269316991e-05, "loss": 2.6506, "step": 29107 }, { "epoch": 1.8069402197529332, "grad_norm": 0.5246990431151436, "learning_rate": 4.092071781023572e-05, "loss": 2.7391, "step": 29108 }, { "epoch": 1.807002296852691, "grad_norm": 0.6173818016828392, "learning_rate": 4.0917166350849255e-05, "loss": 2.8237, "step": 29109 }, { "epoch": 1.807064373952449, "grad_norm": 0.5236201750423051, "learning_rate": 4.091361493885015e-05, "loss": 2.8228, "step": 29110 }, { "epoch": 1.807126451052207, "grad_norm": 0.6126673839504013, "learning_rate": 4.091006357425691e-05, "loss": 2.6577, "step": 29111 }, { "epoch": 1.8071885281519648, "grad_norm": 0.5728224041005036, "learning_rate": 4.0906512257088084e-05, "loss": 2.7487, "step": 29112 }, { "epoch": 1.8072506052517228, "grad_norm": 0.5736151916270508, "learning_rate": 4.0902960987362185e-05, "loss": 2.7189, "step": 29113 }, { "epoch": 1.8073126823514807, "grad_norm": 0.5370307448659191, "learning_rate": 4.089940976509775e-05, "loss": 2.7393, "step": 29114 }, { "epoch": 1.8073747594512386, "grad_norm": 0.6065570034306728, "learning_rate": 4.0895858590313295e-05, "loss": 2.7759, "step": 29115 }, { "epoch": 1.8074368365509963, "grad_norm": 0.5392543064488411, "learning_rate": 4.0892307463027376e-05, "loss": 2.6475, "step": 29116 }, { "epoch": 1.8074989136507542, "grad_norm": 0.6445692802537911, "learning_rate": 4.088875638325849e-05, "loss": 2.7353, "step": 29117 }, { "epoch": 1.8075609907505121, "grad_norm": 0.5284792225686383, "learning_rate": 4.0885205351025164e-05, "loss": 2.7678, "step": 29118 }, { "epoch": 1.80762306785027, "grad_norm": 0.5589991878783412, "learning_rate": 4.088165436634592e-05, "loss": 2.781, "step": 29119 }, { "epoch": 1.807685144950028, "grad_norm": 0.5343027757613604, "learning_rate": 4.0878103429239325e-05, "loss": 2.7447, "step": 29120 }, { "epoch": 1.8077472220497859, "grad_norm": 0.5362106016664752, "learning_rate": 4.087455253972388e-05, "loss": 2.7192, "step": 29121 }, { "epoch": 1.8078092991495436, "grad_norm": 0.5439099175324879, "learning_rate": 4.0871001697818104e-05, "loss": 2.7498, "step": 29122 }, { "epoch": 1.8078713762493015, "grad_norm": 0.5248570906183484, "learning_rate": 4.086745090354052e-05, "loss": 2.7416, "step": 29123 }, { "epoch": 1.8079334533490594, "grad_norm": 0.510986328125, "learning_rate": 4.0863900156909676e-05, "loss": 2.7098, "step": 29124 }, { "epoch": 1.8079955304488173, "grad_norm": 0.5316257550404396, "learning_rate": 4.0860349457944067e-05, "loss": 2.5878, "step": 29125 }, { "epoch": 1.8080576075485753, "grad_norm": 0.5494081954561426, "learning_rate": 4.085679880666225e-05, "loss": 2.7216, "step": 29126 }, { "epoch": 1.8081196846483332, "grad_norm": 0.5800544656880047, "learning_rate": 4.0853248203082724e-05, "loss": 2.6582, "step": 29127 }, { "epoch": 1.808181761748091, "grad_norm": 0.582420833185033, "learning_rate": 4.084969764722403e-05, "loss": 2.8031, "step": 29128 }, { "epoch": 1.808243838847849, "grad_norm": 0.561160983989385, "learning_rate": 4.084614713910468e-05, "loss": 2.8041, "step": 29129 }, { "epoch": 1.808305915947607, "grad_norm": 0.5711404199486881, "learning_rate": 4.08425966787432e-05, "loss": 2.7136, "step": 29130 }, { "epoch": 1.8083679930473648, "grad_norm": 0.5601282716624478, "learning_rate": 4.083904626615812e-05, "loss": 2.8059, "step": 29131 }, { "epoch": 1.8084300701471228, "grad_norm": 0.5674844411153935, "learning_rate": 4.0835495901367955e-05, "loss": 2.6544, "step": 29132 }, { "epoch": 1.8084921472468807, "grad_norm": 0.5289800054380688, "learning_rate": 4.083194558439124e-05, "loss": 2.7642, "step": 29133 }, { "epoch": 1.8085542243466386, "grad_norm": 0.5188046587448093, "learning_rate": 4.0828395315246476e-05, "loss": 2.8024, "step": 29134 }, { "epoch": 1.8086163014463965, "grad_norm": 0.5283031919176794, "learning_rate": 4.0824845093952216e-05, "loss": 2.7525, "step": 29135 }, { "epoch": 1.8086783785461544, "grad_norm": 0.5273371943313581, "learning_rate": 4.082129492052695e-05, "loss": 2.6952, "step": 29136 }, { "epoch": 1.8087404556459123, "grad_norm": 0.5267322527009918, "learning_rate": 4.081774479498924e-05, "loss": 2.7422, "step": 29137 }, { "epoch": 1.8088025327456703, "grad_norm": 0.5118485206117572, "learning_rate": 4.081419471735756e-05, "loss": 2.783, "step": 29138 }, { "epoch": 1.8088646098454282, "grad_norm": 0.5227157399258562, "learning_rate": 4.081064468765048e-05, "loss": 2.8088, "step": 29139 }, { "epoch": 1.8089266869451859, "grad_norm": 0.6019016399806113, "learning_rate": 4.0807094705886475e-05, "loss": 2.647, "step": 29140 }, { "epoch": 1.8089887640449438, "grad_norm": 0.46902931156961336, "learning_rate": 4.08035447720841e-05, "loss": 2.6554, "step": 29141 }, { "epoch": 1.8090508411447017, "grad_norm": 0.5971563405162573, "learning_rate": 4.0799994886261875e-05, "loss": 2.7436, "step": 29142 }, { "epoch": 1.8091129182444596, "grad_norm": 0.5378018718070513, "learning_rate": 4.0796445048438295e-05, "loss": 2.7762, "step": 29143 }, { "epoch": 1.8091749953442176, "grad_norm": 0.5645917570417105, "learning_rate": 4.0792895258631895e-05, "loss": 2.6709, "step": 29144 }, { "epoch": 1.8092370724439755, "grad_norm": 0.5311316189980821, "learning_rate": 4.07893455168612e-05, "loss": 2.7339, "step": 29145 }, { "epoch": 1.8092991495437332, "grad_norm": 0.5534386612951587, "learning_rate": 4.0785795823144726e-05, "loss": 2.7412, "step": 29146 }, { "epoch": 1.809361226643491, "grad_norm": 0.6088238448877702, "learning_rate": 4.078224617750098e-05, "loss": 2.7975, "step": 29147 }, { "epoch": 1.809423303743249, "grad_norm": 0.5621802692891934, "learning_rate": 4.0778696579948516e-05, "loss": 2.767, "step": 29148 }, { "epoch": 1.809485380843007, "grad_norm": 0.5981870799598737, "learning_rate": 4.0775147030505814e-05, "loss": 2.6902, "step": 29149 }, { "epoch": 1.8095474579427648, "grad_norm": 0.5568684964811705, "learning_rate": 4.0771597529191416e-05, "loss": 2.7226, "step": 29150 }, { "epoch": 1.8096095350425228, "grad_norm": 0.5902554420533763, "learning_rate": 4.0768048076023814e-05, "loss": 2.8243, "step": 29151 }, { "epoch": 1.8096716121422807, "grad_norm": 0.5990800291629944, "learning_rate": 4.0764498671021554e-05, "loss": 2.7344, "step": 29152 }, { "epoch": 1.8097336892420386, "grad_norm": 0.6058524361956454, "learning_rate": 4.076094931420317e-05, "loss": 2.7478, "step": 29153 }, { "epoch": 1.8097957663417965, "grad_norm": 0.5678129683027084, "learning_rate": 4.075740000558714e-05, "loss": 2.7583, "step": 29154 }, { "epoch": 1.8098578434415544, "grad_norm": 0.6210044700704318, "learning_rate": 4.0753850745192e-05, "loss": 2.7246, "step": 29155 }, { "epoch": 1.8099199205413123, "grad_norm": 0.5775922434074666, "learning_rate": 4.075030153303627e-05, "loss": 2.7804, "step": 29156 }, { "epoch": 1.8099819976410703, "grad_norm": 0.6096081654420415, "learning_rate": 4.074675236913845e-05, "loss": 2.7194, "step": 29157 }, { "epoch": 1.8100440747408282, "grad_norm": 0.5714065561353715, "learning_rate": 4.0743203253517084e-05, "loss": 2.8465, "step": 29158 }, { "epoch": 1.810106151840586, "grad_norm": 0.600562813166357, "learning_rate": 4.073965418619066e-05, "loss": 2.6638, "step": 29159 }, { "epoch": 1.810168228940344, "grad_norm": 0.5278789488825154, "learning_rate": 4.0736105167177725e-05, "loss": 2.7141, "step": 29160 }, { "epoch": 1.810230306040102, "grad_norm": 0.6560160583225721, "learning_rate": 4.073255619649676e-05, "loss": 2.91, "step": 29161 }, { "epoch": 1.8102923831398599, "grad_norm": 0.5773054448326755, "learning_rate": 4.072900727416631e-05, "loss": 2.76, "step": 29162 }, { "epoch": 1.8103544602396178, "grad_norm": 0.6455718223703047, "learning_rate": 4.0725458400204873e-05, "loss": 2.7304, "step": 29163 }, { "epoch": 1.8104165373393755, "grad_norm": 0.6018378135546436, "learning_rate": 4.072190957463098e-05, "loss": 2.6828, "step": 29164 }, { "epoch": 1.8104786144391334, "grad_norm": 0.6196766407333029, "learning_rate": 4.0718360797463126e-05, "loss": 2.7746, "step": 29165 }, { "epoch": 1.8105406915388913, "grad_norm": 0.5707533974147475, "learning_rate": 4.0714812068719856e-05, "loss": 2.6805, "step": 29166 }, { "epoch": 1.8106027686386492, "grad_norm": 0.6322675407076149, "learning_rate": 4.071126338841965e-05, "loss": 2.7234, "step": 29167 }, { "epoch": 1.8106648457384071, "grad_norm": 0.6252694741106373, "learning_rate": 4.070771475658104e-05, "loss": 2.6958, "step": 29168 }, { "epoch": 1.810726922838165, "grad_norm": 0.6305376301214028, "learning_rate": 4.070416617322254e-05, "loss": 2.8779, "step": 29169 }, { "epoch": 1.8107889999379227, "grad_norm": 0.6494295505598475, "learning_rate": 4.0700617638362655e-05, "loss": 2.7707, "step": 29170 }, { "epoch": 1.8108510770376807, "grad_norm": 0.577557930043121, "learning_rate": 4.069706915201992e-05, "loss": 2.7161, "step": 29171 }, { "epoch": 1.8109131541374386, "grad_norm": 0.5624513075204389, "learning_rate": 4.0693520714212815e-05, "loss": 2.714, "step": 29172 }, { "epoch": 1.8109752312371965, "grad_norm": 0.6300271039006775, "learning_rate": 4.068997232495988e-05, "loss": 2.6819, "step": 29173 }, { "epoch": 1.8110373083369544, "grad_norm": 0.5849328647303105, "learning_rate": 4.068642398427962e-05, "loss": 2.7381, "step": 29174 }, { "epoch": 1.8110993854367123, "grad_norm": 0.6097224296602298, "learning_rate": 4.0682875692190555e-05, "loss": 2.7533, "step": 29175 }, { "epoch": 1.8111614625364703, "grad_norm": 0.5470613162231538, "learning_rate": 4.067932744871117e-05, "loss": 2.6491, "step": 29176 }, { "epoch": 1.8112235396362282, "grad_norm": 0.6576328920370472, "learning_rate": 4.067577925386001e-05, "loss": 2.7579, "step": 29177 }, { "epoch": 1.811285616735986, "grad_norm": 0.5394743576125177, "learning_rate": 4.067223110765558e-05, "loss": 2.7991, "step": 29178 }, { "epoch": 1.811347693835744, "grad_norm": 0.6473499507915524, "learning_rate": 4.066868301011637e-05, "loss": 2.7948, "step": 29179 }, { "epoch": 1.811409770935502, "grad_norm": 0.5615109118847553, "learning_rate": 4.066513496126092e-05, "loss": 2.6853, "step": 29180 }, { "epoch": 1.8114718480352598, "grad_norm": 0.5657945947724665, "learning_rate": 4.066158696110771e-05, "loss": 2.737, "step": 29181 }, { "epoch": 1.8115339251350178, "grad_norm": 0.5830773802486034, "learning_rate": 4.065803900967529e-05, "loss": 2.7485, "step": 29182 }, { "epoch": 1.8115960022347757, "grad_norm": 0.5321252849005487, "learning_rate": 4.065449110698212e-05, "loss": 2.7331, "step": 29183 }, { "epoch": 1.8116580793345336, "grad_norm": 0.5774063592600875, "learning_rate": 4.065094325304674e-05, "loss": 2.721, "step": 29184 }, { "epoch": 1.8117201564342915, "grad_norm": 0.5845854354824227, "learning_rate": 4.064739544788768e-05, "loss": 2.7667, "step": 29185 }, { "epoch": 1.8117822335340494, "grad_norm": 0.5623805396067274, "learning_rate": 4.064384769152341e-05, "loss": 2.7007, "step": 29186 }, { "epoch": 1.8118443106338074, "grad_norm": 0.5346657112805511, "learning_rate": 4.0640299983972474e-05, "loss": 2.7229, "step": 29187 }, { "epoch": 1.811906387733565, "grad_norm": 0.5470287106979956, "learning_rate": 4.063675232525336e-05, "loss": 2.709, "step": 29188 }, { "epoch": 1.811968464833323, "grad_norm": 0.5823590680342362, "learning_rate": 4.063320471538458e-05, "loss": 2.7219, "step": 29189 }, { "epoch": 1.8120305419330809, "grad_norm": 0.5596717622675319, "learning_rate": 4.062965715438464e-05, "loss": 2.6272, "step": 29190 }, { "epoch": 1.8120926190328388, "grad_norm": 0.6673438512131729, "learning_rate": 4.062610964227207e-05, "loss": 2.7669, "step": 29191 }, { "epoch": 1.8121546961325967, "grad_norm": 0.5502695452656118, "learning_rate": 4.0622562179065356e-05, "loss": 2.713, "step": 29192 }, { "epoch": 1.8122167732323546, "grad_norm": 0.5807878923587924, "learning_rate": 4.0619014764783e-05, "loss": 2.7029, "step": 29193 }, { "epoch": 1.8122788503321123, "grad_norm": 0.5318206639220818, "learning_rate": 4.0615467399443534e-05, "loss": 2.6612, "step": 29194 }, { "epoch": 1.8123409274318703, "grad_norm": 0.5786455233962507, "learning_rate": 4.061192008306544e-05, "loss": 2.8359, "step": 29195 }, { "epoch": 1.8124030045316282, "grad_norm": 0.510374560533232, "learning_rate": 4.060837281566725e-05, "loss": 2.6961, "step": 29196 }, { "epoch": 1.812465081631386, "grad_norm": 0.5946102686338886, "learning_rate": 4.060482559726745e-05, "loss": 2.6822, "step": 29197 }, { "epoch": 1.812527158731144, "grad_norm": 0.5363962606671656, "learning_rate": 4.060127842788456e-05, "loss": 2.7632, "step": 29198 }, { "epoch": 1.812589235830902, "grad_norm": 0.4892756597800511, "learning_rate": 4.059773130753708e-05, "loss": 2.6662, "step": 29199 }, { "epoch": 1.8126513129306598, "grad_norm": 0.5750832051738262, "learning_rate": 4.059418423624353e-05, "loss": 2.7838, "step": 29200 }, { "epoch": 1.8127133900304178, "grad_norm": 0.5106880640706771, "learning_rate": 4.059063721402239e-05, "loss": 2.6703, "step": 29201 }, { "epoch": 1.8127754671301757, "grad_norm": 0.5513563197622583, "learning_rate": 4.058709024089218e-05, "loss": 2.7381, "step": 29202 }, { "epoch": 1.8128375442299336, "grad_norm": 0.5717620977882399, "learning_rate": 4.058354331687142e-05, "loss": 2.6627, "step": 29203 }, { "epoch": 1.8128996213296915, "grad_norm": 0.545775780179419, "learning_rate": 4.057999644197858e-05, "loss": 2.7471, "step": 29204 }, { "epoch": 1.8129616984294494, "grad_norm": 0.577341270183353, "learning_rate": 4.0576449616232204e-05, "loss": 2.7037, "step": 29205 }, { "epoch": 1.8130237755292073, "grad_norm": 0.6254056091229065, "learning_rate": 4.057290283965076e-05, "loss": 2.705, "step": 29206 }, { "epoch": 1.8130858526289653, "grad_norm": 0.5507579555049448, "learning_rate": 4.056935611225279e-05, "loss": 2.7826, "step": 29207 }, { "epoch": 1.8131479297287232, "grad_norm": 0.6289226458763866, "learning_rate": 4.056580943405676e-05, "loss": 2.7217, "step": 29208 }, { "epoch": 1.813210006828481, "grad_norm": 0.49069985104230723, "learning_rate": 4.05622628050812e-05, "loss": 2.783, "step": 29209 }, { "epoch": 1.813272083928239, "grad_norm": 0.5540835894310923, "learning_rate": 4.0558716225344604e-05, "loss": 2.8029, "step": 29210 }, { "epoch": 1.813334161027997, "grad_norm": 0.5247955355540105, "learning_rate": 4.055516969486549e-05, "loss": 2.767, "step": 29211 }, { "epoch": 1.8133962381277546, "grad_norm": 0.5426863340676987, "learning_rate": 4.055162321366233e-05, "loss": 2.7619, "step": 29212 }, { "epoch": 1.8134583152275126, "grad_norm": 0.5965567304473088, "learning_rate": 4.054807678175365e-05, "loss": 2.6411, "step": 29213 }, { "epoch": 1.8135203923272705, "grad_norm": 0.5705890768720436, "learning_rate": 4.054453039915797e-05, "loss": 2.6147, "step": 29214 }, { "epoch": 1.8135824694270284, "grad_norm": 0.5810906919976907, "learning_rate": 4.054098406589374e-05, "loss": 2.7864, "step": 29215 }, { "epoch": 1.8136445465267863, "grad_norm": 0.570981273716504, "learning_rate": 4.0537437781979506e-05, "loss": 2.6941, "step": 29216 }, { "epoch": 1.8137066236265442, "grad_norm": 0.5303882453517824, "learning_rate": 4.053389154743374e-05, "loss": 2.6376, "step": 29217 }, { "epoch": 1.813768700726302, "grad_norm": 0.6699898910471539, "learning_rate": 4.0530345362274965e-05, "loss": 2.7383, "step": 29218 }, { "epoch": 1.8138307778260598, "grad_norm": 0.5560358878395542, "learning_rate": 4.052679922652169e-05, "loss": 2.7475, "step": 29219 }, { "epoch": 1.8138928549258178, "grad_norm": 0.641259367801883, "learning_rate": 4.05232531401924e-05, "loss": 2.7527, "step": 29220 }, { "epoch": 1.8139549320255757, "grad_norm": 0.5535108955089715, "learning_rate": 4.0519707103305605e-05, "loss": 2.748, "step": 29221 }, { "epoch": 1.8140170091253336, "grad_norm": 0.5697192673188314, "learning_rate": 4.051616111587979e-05, "loss": 2.7064, "step": 29222 }, { "epoch": 1.8140790862250915, "grad_norm": 0.5798407174773546, "learning_rate": 4.0512615177933464e-05, "loss": 2.7256, "step": 29223 }, { "epoch": 1.8141411633248494, "grad_norm": 0.5563720590637546, "learning_rate": 4.050906928948513e-05, "loss": 2.682, "step": 29224 }, { "epoch": 1.8142032404246073, "grad_norm": 0.544582958858549, "learning_rate": 4.050552345055329e-05, "loss": 2.7109, "step": 29225 }, { "epoch": 1.8142653175243653, "grad_norm": 0.5517741513153998, "learning_rate": 4.050197766115643e-05, "loss": 2.7109, "step": 29226 }, { "epoch": 1.8143273946241232, "grad_norm": 0.5533560501983257, "learning_rate": 4.049843192131307e-05, "loss": 2.8257, "step": 29227 }, { "epoch": 1.814389471723881, "grad_norm": 0.5505400866916444, "learning_rate": 4.049488623104169e-05, "loss": 2.768, "step": 29228 }, { "epoch": 1.814451548823639, "grad_norm": 0.5578898282773908, "learning_rate": 4.04913405903608e-05, "loss": 2.6894, "step": 29229 }, { "epoch": 1.814513625923397, "grad_norm": 0.5967846417562144, "learning_rate": 4.0487794999288894e-05, "loss": 2.7941, "step": 29230 }, { "epoch": 1.8145757030231549, "grad_norm": 0.5366214547403386, "learning_rate": 4.0484249457844464e-05, "loss": 2.8545, "step": 29231 }, { "epoch": 1.8146377801229128, "grad_norm": 0.5888540035562263, "learning_rate": 4.0480703966046025e-05, "loss": 2.7292, "step": 29232 }, { "epoch": 1.8146998572226707, "grad_norm": 0.6159741994297311, "learning_rate": 4.0477158523912054e-05, "loss": 2.7308, "step": 29233 }, { "epoch": 1.8147619343224286, "grad_norm": 0.6684131593985149, "learning_rate": 4.0473613131461066e-05, "loss": 2.8029, "step": 29234 }, { "epoch": 1.8148240114221863, "grad_norm": 0.574718284010614, "learning_rate": 4.047006778871154e-05, "loss": 2.7196, "step": 29235 }, { "epoch": 1.8148860885219442, "grad_norm": 0.6072459809494981, "learning_rate": 4.0466522495682e-05, "loss": 2.7378, "step": 29236 }, { "epoch": 1.8149481656217021, "grad_norm": 0.6579220904762295, "learning_rate": 4.046297725239091e-05, "loss": 2.7684, "step": 29237 }, { "epoch": 1.81501024272146, "grad_norm": 0.5892009643201065, "learning_rate": 4.04594320588568e-05, "loss": 2.6765, "step": 29238 }, { "epoch": 1.815072319821218, "grad_norm": 0.5804005763810159, "learning_rate": 4.0455886915098145e-05, "loss": 2.7329, "step": 29239 }, { "epoch": 1.815134396920976, "grad_norm": 0.5565062243515381, "learning_rate": 4.0452341821133435e-05, "loss": 2.7337, "step": 29240 }, { "epoch": 1.8151964740207336, "grad_norm": 0.6018314998668857, "learning_rate": 4.044879677698118e-05, "loss": 2.654, "step": 29241 }, { "epoch": 1.8152585511204915, "grad_norm": 0.6450510645398773, "learning_rate": 4.044525178265987e-05, "loss": 2.7416, "step": 29242 }, { "epoch": 1.8153206282202494, "grad_norm": 0.5614418302197247, "learning_rate": 4.044170683818801e-05, "loss": 2.6956, "step": 29243 }, { "epoch": 1.8153827053200073, "grad_norm": 0.6232588356637314, "learning_rate": 4.0438161943584066e-05, "loss": 2.7101, "step": 29244 }, { "epoch": 1.8154447824197653, "grad_norm": 0.5553464333680475, "learning_rate": 4.0434617098866574e-05, "loss": 2.7716, "step": 29245 }, { "epoch": 1.8155068595195232, "grad_norm": 0.5927140836207699, "learning_rate": 4.043107230405399e-05, "loss": 2.737, "step": 29246 }, { "epoch": 1.815568936619281, "grad_norm": 0.5678998787325813, "learning_rate": 4.0427527559164835e-05, "loss": 2.6759, "step": 29247 }, { "epoch": 1.815631013719039, "grad_norm": 0.5083926188443347, "learning_rate": 4.042398286421759e-05, "loss": 2.6907, "step": 29248 }, { "epoch": 1.815693090818797, "grad_norm": 0.579396422071408, "learning_rate": 4.042043821923075e-05, "loss": 2.7309, "step": 29249 }, { "epoch": 1.8157551679185548, "grad_norm": 0.5423694294372874, "learning_rate": 4.04168936242228e-05, "loss": 2.6799, "step": 29250 }, { "epoch": 1.8158172450183128, "grad_norm": 0.6082414942051417, "learning_rate": 4.041334907921224e-05, "loss": 2.7782, "step": 29251 }, { "epoch": 1.8158793221180707, "grad_norm": 0.5514141801894594, "learning_rate": 4.040980458421759e-05, "loss": 2.65, "step": 29252 }, { "epoch": 1.8159413992178286, "grad_norm": 0.5173542042156697, "learning_rate": 4.04062601392573e-05, "loss": 2.6964, "step": 29253 }, { "epoch": 1.8160034763175865, "grad_norm": 0.566225588695485, "learning_rate": 4.040271574434989e-05, "loss": 2.7187, "step": 29254 }, { "epoch": 1.8160655534173444, "grad_norm": 0.5025306377749938, "learning_rate": 4.0399171399513835e-05, "loss": 2.7112, "step": 29255 }, { "epoch": 1.8161276305171024, "grad_norm": 0.5235737580546984, "learning_rate": 4.039562710476763e-05, "loss": 2.6893, "step": 29256 }, { "epoch": 1.8161897076168603, "grad_norm": 0.5460800660346719, "learning_rate": 4.039208286012978e-05, "loss": 2.71, "step": 29257 }, { "epoch": 1.8162517847166182, "grad_norm": 0.4933225773538772, "learning_rate": 4.0388538665618756e-05, "loss": 2.7994, "step": 29258 }, { "epoch": 1.8163138618163759, "grad_norm": 0.4857718722081521, "learning_rate": 4.038499452125307e-05, "loss": 2.6935, "step": 29259 }, { "epoch": 1.8163759389161338, "grad_norm": 0.4817263728753402, "learning_rate": 4.0381450427051186e-05, "loss": 2.7452, "step": 29260 }, { "epoch": 1.8164380160158917, "grad_norm": 0.5256344537871376, "learning_rate": 4.037790638303163e-05, "loss": 2.7542, "step": 29261 }, { "epoch": 1.8165000931156496, "grad_norm": 0.4834775301259493, "learning_rate": 4.037436238921285e-05, "loss": 2.7131, "step": 29262 }, { "epoch": 1.8165621702154076, "grad_norm": 0.49901134379265305, "learning_rate": 4.037081844561338e-05, "loss": 2.6963, "step": 29263 }, { "epoch": 1.8166242473151655, "grad_norm": 0.5570213220682652, "learning_rate": 4.036727455225168e-05, "loss": 2.7766, "step": 29264 }, { "epoch": 1.8166863244149232, "grad_norm": 0.4792591313281714, "learning_rate": 4.036373070914624e-05, "loss": 2.7521, "step": 29265 }, { "epoch": 1.816748401514681, "grad_norm": 0.5172112658246543, "learning_rate": 4.036018691631557e-05, "loss": 2.698, "step": 29266 }, { "epoch": 1.816810478614439, "grad_norm": 0.5060256036831998, "learning_rate": 4.035664317377813e-05, "loss": 2.7558, "step": 29267 }, { "epoch": 1.816872555714197, "grad_norm": 0.5793933872897649, "learning_rate": 4.0353099481552436e-05, "loss": 2.7402, "step": 29268 }, { "epoch": 1.8169346328139548, "grad_norm": 0.6140609935324635, "learning_rate": 4.034955583965695e-05, "loss": 2.813, "step": 29269 }, { "epoch": 1.8169967099137128, "grad_norm": 0.6501430234913634, "learning_rate": 4.034601224811019e-05, "loss": 2.861, "step": 29270 }, { "epoch": 1.8170587870134707, "grad_norm": 0.543772342924398, "learning_rate": 4.034246870693061e-05, "loss": 2.7945, "step": 29271 }, { "epoch": 1.8171208641132286, "grad_norm": 0.6219878808325059, "learning_rate": 4.033892521613673e-05, "loss": 2.76, "step": 29272 }, { "epoch": 1.8171829412129865, "grad_norm": 0.5813296396804556, "learning_rate": 4.033538177574701e-05, "loss": 2.7676, "step": 29273 }, { "epoch": 1.8172450183127444, "grad_norm": 0.5953205318359663, "learning_rate": 4.0331838385779966e-05, "loss": 2.7288, "step": 29274 }, { "epoch": 1.8173070954125023, "grad_norm": 0.5786508797449267, "learning_rate": 4.032829504625406e-05, "loss": 2.8075, "step": 29275 }, { "epoch": 1.8173691725122603, "grad_norm": 0.6465882047961417, "learning_rate": 4.032475175718778e-05, "loss": 2.7357, "step": 29276 }, { "epoch": 1.8174312496120182, "grad_norm": 0.5578048844445374, "learning_rate": 4.0321208518599627e-05, "loss": 2.7382, "step": 29277 }, { "epoch": 1.817493326711776, "grad_norm": 0.5463934549388576, "learning_rate": 4.031766533050807e-05, "loss": 2.7856, "step": 29278 }, { "epoch": 1.817555403811534, "grad_norm": 0.6043287558551448, "learning_rate": 4.0314122192931616e-05, "loss": 2.7883, "step": 29279 }, { "epoch": 1.817617480911292, "grad_norm": 0.5673640341614041, "learning_rate": 4.031057910588872e-05, "loss": 2.723, "step": 29280 }, { "epoch": 1.8176795580110499, "grad_norm": 0.5710614916748817, "learning_rate": 4.030703606939791e-05, "loss": 2.7144, "step": 29281 }, { "epoch": 1.8177416351108078, "grad_norm": 0.6276683827282378, "learning_rate": 4.030349308347761e-05, "loss": 2.7173, "step": 29282 }, { "epoch": 1.8178037122105655, "grad_norm": 0.5883719922684411, "learning_rate": 4.029995014814636e-05, "loss": 2.7691, "step": 29283 }, { "epoch": 1.8178657893103234, "grad_norm": 0.6090724389570293, "learning_rate": 4.0296407263422634e-05, "loss": 2.6941, "step": 29284 }, { "epoch": 1.8179278664100813, "grad_norm": 0.587162441309743, "learning_rate": 4.0292864429324894e-05, "loss": 2.7015, "step": 29285 }, { "epoch": 1.8179899435098392, "grad_norm": 0.5841836238563938, "learning_rate": 4.0289321645871646e-05, "loss": 2.7147, "step": 29286 }, { "epoch": 1.8180520206095971, "grad_norm": 0.5102902632176676, "learning_rate": 4.028577891308136e-05, "loss": 2.7469, "step": 29287 }, { "epoch": 1.818114097709355, "grad_norm": 0.576995493702671, "learning_rate": 4.0282236230972526e-05, "loss": 2.739, "step": 29288 }, { "epoch": 1.8181761748091128, "grad_norm": 0.5691524098576266, "learning_rate": 4.027869359956363e-05, "loss": 2.732, "step": 29289 }, { "epoch": 1.8182382519088707, "grad_norm": 0.607904910635626, "learning_rate": 4.027515101887314e-05, "loss": 2.8296, "step": 29290 }, { "epoch": 1.8183003290086286, "grad_norm": 0.5721257269873239, "learning_rate": 4.027160848891955e-05, "loss": 2.7942, "step": 29291 }, { "epoch": 1.8183624061083865, "grad_norm": 0.5432719475007604, "learning_rate": 4.026806600972134e-05, "loss": 2.7689, "step": 29292 }, { "epoch": 1.8184244832081444, "grad_norm": 0.5438014170573138, "learning_rate": 4.0264523581297e-05, "loss": 2.6543, "step": 29293 }, { "epoch": 1.8184865603079023, "grad_norm": 0.6059129869559796, "learning_rate": 4.026098120366499e-05, "loss": 2.841, "step": 29294 }, { "epoch": 1.8185486374076603, "grad_norm": 0.5497510487493111, "learning_rate": 4.0257438876843814e-05, "loss": 2.7755, "step": 29295 }, { "epoch": 1.8186107145074182, "grad_norm": 0.6571743449854898, "learning_rate": 4.0253896600851935e-05, "loss": 2.8074, "step": 29296 }, { "epoch": 1.818672791607176, "grad_norm": 0.6036315609086997, "learning_rate": 4.025035437570786e-05, "loss": 2.772, "step": 29297 }, { "epoch": 1.818734868706934, "grad_norm": 0.6246510246661738, "learning_rate": 4.024681220143004e-05, "loss": 2.7386, "step": 29298 }, { "epoch": 1.818796945806692, "grad_norm": 0.6193411950471985, "learning_rate": 4.0243270078036975e-05, "loss": 2.7547, "step": 29299 }, { "epoch": 1.8188590229064499, "grad_norm": 0.5520094036289542, "learning_rate": 4.023972800554714e-05, "loss": 2.6686, "step": 29300 }, { "epoch": 1.8189211000062078, "grad_norm": 0.6567164080639545, "learning_rate": 4.0236185983979e-05, "loss": 2.7736, "step": 29301 }, { "epoch": 1.8189831771059657, "grad_norm": 0.5854353215061094, "learning_rate": 4.023264401335107e-05, "loss": 2.786, "step": 29302 }, { "epoch": 1.8190452542057236, "grad_norm": 0.5980524482691608, "learning_rate": 4.0229102093681786e-05, "loss": 2.7787, "step": 29303 }, { "epoch": 1.8191073313054815, "grad_norm": 0.5118300921032027, "learning_rate": 4.0225560224989664e-05, "loss": 2.6904, "step": 29304 }, { "epoch": 1.8191694084052394, "grad_norm": 0.5884946084128008, "learning_rate": 4.0222018407293155e-05, "loss": 2.8766, "step": 29305 }, { "epoch": 1.8192314855049974, "grad_norm": 0.6121949457998053, "learning_rate": 4.0218476640610756e-05, "loss": 2.7283, "step": 29306 }, { "epoch": 1.819293562604755, "grad_norm": 0.5560279017040869, "learning_rate": 4.0214934924960934e-05, "loss": 2.7866, "step": 29307 }, { "epoch": 1.819355639704513, "grad_norm": 0.7127937146512315, "learning_rate": 4.021139326036219e-05, "loss": 2.7824, "step": 29308 }, { "epoch": 1.819417716804271, "grad_norm": 0.5389102776150992, "learning_rate": 4.0207851646832954e-05, "loss": 2.7738, "step": 29309 }, { "epoch": 1.8194797939040288, "grad_norm": 0.6680269271768539, "learning_rate": 4.0204310084391754e-05, "loss": 2.7571, "step": 29310 }, { "epoch": 1.8195418710037867, "grad_norm": 0.5482494792718323, "learning_rate": 4.020076857305703e-05, "loss": 2.731, "step": 29311 }, { "epoch": 1.8196039481035446, "grad_norm": 0.595944115092425, "learning_rate": 4.0197227112847295e-05, "loss": 2.5801, "step": 29312 }, { "epoch": 1.8196660252033023, "grad_norm": 0.5692436182672077, "learning_rate": 4.0193685703781e-05, "loss": 2.7559, "step": 29313 }, { "epoch": 1.8197281023030603, "grad_norm": 0.626203665399317, "learning_rate": 4.019014434587662e-05, "loss": 2.6923, "step": 29314 }, { "epoch": 1.8197901794028182, "grad_norm": 0.6017625897052925, "learning_rate": 4.018660303915263e-05, "loss": 2.6489, "step": 29315 }, { "epoch": 1.819852256502576, "grad_norm": 0.6025755390520766, "learning_rate": 4.018306178362753e-05, "loss": 2.7945, "step": 29316 }, { "epoch": 1.819914333602334, "grad_norm": 0.6013261652821984, "learning_rate": 4.0179520579319773e-05, "loss": 2.7844, "step": 29317 }, { "epoch": 1.819976410702092, "grad_norm": 0.5825926614088321, "learning_rate": 4.017597942624784e-05, "loss": 2.7866, "step": 29318 }, { "epoch": 1.8200384878018498, "grad_norm": 0.5974791021164007, "learning_rate": 4.017243832443021e-05, "loss": 2.8079, "step": 29319 }, { "epoch": 1.8201005649016078, "grad_norm": 0.5577201948542071, "learning_rate": 4.0168897273885365e-05, "loss": 2.7401, "step": 29320 }, { "epoch": 1.8201626420013657, "grad_norm": 0.620330149209721, "learning_rate": 4.016535627463176e-05, "loss": 2.7626, "step": 29321 }, { "epoch": 1.8202247191011236, "grad_norm": 0.5207243678230552, "learning_rate": 4.016181532668788e-05, "loss": 2.7825, "step": 29322 }, { "epoch": 1.8202867962008815, "grad_norm": 0.6312372121365439, "learning_rate": 4.01582744300722e-05, "loss": 2.7118, "step": 29323 }, { "epoch": 1.8203488733006394, "grad_norm": 0.5158175195736078, "learning_rate": 4.015473358480319e-05, "loss": 2.747, "step": 29324 }, { "epoch": 1.8204109504003974, "grad_norm": 0.6186550173442007, "learning_rate": 4.015119279089933e-05, "loss": 2.7446, "step": 29325 }, { "epoch": 1.8204730275001553, "grad_norm": 0.5760569570994235, "learning_rate": 4.014765204837908e-05, "loss": 2.7573, "step": 29326 }, { "epoch": 1.8205351045999132, "grad_norm": 0.6043023472287278, "learning_rate": 4.0144111357260924e-05, "loss": 2.7605, "step": 29327 }, { "epoch": 1.820597181699671, "grad_norm": 0.5551512217059088, "learning_rate": 4.014057071756332e-05, "loss": 2.6615, "step": 29328 }, { "epoch": 1.820659258799429, "grad_norm": 0.553008129682677, "learning_rate": 4.013703012930477e-05, "loss": 2.6881, "step": 29329 }, { "epoch": 1.820721335899187, "grad_norm": 0.6161515196674616, "learning_rate": 4.0133489592503715e-05, "loss": 2.731, "step": 29330 }, { "epoch": 1.8207834129989446, "grad_norm": 0.5594503330671912, "learning_rate": 4.0129949107178654e-05, "loss": 2.771, "step": 29331 }, { "epoch": 1.8208454900987026, "grad_norm": 0.5426775199308992, "learning_rate": 4.012640867334803e-05, "loss": 2.6464, "step": 29332 }, { "epoch": 1.8209075671984605, "grad_norm": 0.5908034393171051, "learning_rate": 4.0122868291030344e-05, "loss": 2.6535, "step": 29333 }, { "epoch": 1.8209696442982184, "grad_norm": 0.5742671453102777, "learning_rate": 4.011932796024404e-05, "loss": 2.7421, "step": 29334 }, { "epoch": 1.8210317213979763, "grad_norm": 0.5450660487495675, "learning_rate": 4.01157876810076e-05, "loss": 2.7342, "step": 29335 }, { "epoch": 1.8210937984977342, "grad_norm": 0.5982444959914166, "learning_rate": 4.0112247453339495e-05, "loss": 2.7017, "step": 29336 }, { "epoch": 1.821155875597492, "grad_norm": 0.5736722356583954, "learning_rate": 4.010870727725821e-05, "loss": 2.7658, "step": 29337 }, { "epoch": 1.8212179526972498, "grad_norm": 0.5754516755636099, "learning_rate": 4.010516715278219e-05, "loss": 2.7948, "step": 29338 }, { "epoch": 1.8212800297970078, "grad_norm": 0.5651450750307638, "learning_rate": 4.010162707992991e-05, "loss": 2.7814, "step": 29339 }, { "epoch": 1.8213421068967657, "grad_norm": 0.6243409496717901, "learning_rate": 4.009808705871986e-05, "loss": 2.7763, "step": 29340 }, { "epoch": 1.8214041839965236, "grad_norm": 0.5297693487525067, "learning_rate": 4.0094547089170474e-05, "loss": 2.8298, "step": 29341 }, { "epoch": 1.8214662610962815, "grad_norm": 0.6785348128535906, "learning_rate": 4.0091007171300256e-05, "loss": 2.7547, "step": 29342 }, { "epoch": 1.8215283381960394, "grad_norm": 0.5762836929698948, "learning_rate": 4.008746730512765e-05, "loss": 2.6444, "step": 29343 }, { "epoch": 1.8215904152957973, "grad_norm": 0.6276729646279998, "learning_rate": 4.0083927490671134e-05, "loss": 2.6312, "step": 29344 }, { "epoch": 1.8216524923955553, "grad_norm": 0.6091654979818104, "learning_rate": 4.008038772794917e-05, "loss": 2.7289, "step": 29345 }, { "epoch": 1.8217145694953132, "grad_norm": 0.6237264293505811, "learning_rate": 4.0076848016980246e-05, "loss": 2.7382, "step": 29346 }, { "epoch": 1.821776646595071, "grad_norm": 0.6092380956771118, "learning_rate": 4.0073308357782804e-05, "loss": 2.7081, "step": 29347 }, { "epoch": 1.821838723694829, "grad_norm": 0.5875633317210902, "learning_rate": 4.006976875037531e-05, "loss": 2.7553, "step": 29348 }, { "epoch": 1.821900800794587, "grad_norm": 0.5691265944909528, "learning_rate": 4.006622919477626e-05, "loss": 2.7371, "step": 29349 }, { "epoch": 1.8219628778943449, "grad_norm": 0.6011310863476287, "learning_rate": 4.0062689691004107e-05, "loss": 2.6626, "step": 29350 }, { "epoch": 1.8220249549941028, "grad_norm": 0.5427555516423368, "learning_rate": 4.00591502390773e-05, "loss": 2.8322, "step": 29351 }, { "epoch": 1.8220870320938607, "grad_norm": 0.6393037988687986, "learning_rate": 4.005561083901434e-05, "loss": 2.7678, "step": 29352 }, { "epoch": 1.8221491091936186, "grad_norm": 0.5871311490811175, "learning_rate": 4.0052071490833654e-05, "loss": 2.7503, "step": 29353 }, { "epoch": 1.8222111862933765, "grad_norm": 0.5797897032844872, "learning_rate": 4.004853219455374e-05, "loss": 2.7785, "step": 29354 }, { "epoch": 1.8222732633931342, "grad_norm": 0.5823635458441075, "learning_rate": 4.004499295019304e-05, "loss": 2.6375, "step": 29355 }, { "epoch": 1.8223353404928921, "grad_norm": 0.5498373994192234, "learning_rate": 4.004145375777003e-05, "loss": 2.773, "step": 29356 }, { "epoch": 1.82239741759265, "grad_norm": 0.588994380948412, "learning_rate": 4.0037914617303175e-05, "loss": 2.7154, "step": 29357 }, { "epoch": 1.822459494692408, "grad_norm": 0.5581826517930131, "learning_rate": 4.003437552881094e-05, "loss": 2.732, "step": 29358 }, { "epoch": 1.822521571792166, "grad_norm": 0.5550708252160367, "learning_rate": 4.003083649231178e-05, "loss": 2.7832, "step": 29359 }, { "epoch": 1.8225836488919238, "grad_norm": 0.5379455715640692, "learning_rate": 4.002729750782418e-05, "loss": 2.6939, "step": 29360 }, { "epoch": 1.8226457259916815, "grad_norm": 0.6091384183915375, "learning_rate": 4.0023758575366563e-05, "loss": 2.7729, "step": 29361 }, { "epoch": 1.8227078030914394, "grad_norm": 0.5301174545596614, "learning_rate": 4.0020219694957443e-05, "loss": 2.6931, "step": 29362 }, { "epoch": 1.8227698801911973, "grad_norm": 0.5972226882779966, "learning_rate": 4.001668086661526e-05, "loss": 2.7993, "step": 29363 }, { "epoch": 1.8228319572909553, "grad_norm": 0.5331546076684517, "learning_rate": 4.0013142090358466e-05, "loss": 2.7187, "step": 29364 }, { "epoch": 1.8228940343907132, "grad_norm": 0.6209603652238276, "learning_rate": 4.0009603366205546e-05, "loss": 2.737, "step": 29365 }, { "epoch": 1.822956111490471, "grad_norm": 0.5274666537009226, "learning_rate": 4.000606469417493e-05, "loss": 2.7438, "step": 29366 }, { "epoch": 1.823018188590229, "grad_norm": 0.541783390551192, "learning_rate": 4.000252607428512e-05, "loss": 2.7341, "step": 29367 }, { "epoch": 1.823080265689987, "grad_norm": 0.5943853342098524, "learning_rate": 3.999898750655455e-05, "loss": 2.8618, "step": 29368 }, { "epoch": 1.8231423427897449, "grad_norm": 0.505096183861149, "learning_rate": 3.999544899100169e-05, "loss": 2.7416, "step": 29369 }, { "epoch": 1.8232044198895028, "grad_norm": 0.5654577007832202, "learning_rate": 3.9991910527645e-05, "loss": 2.7519, "step": 29370 }, { "epoch": 1.8232664969892607, "grad_norm": 0.5158995850909096, "learning_rate": 3.9988372116502946e-05, "loss": 2.7232, "step": 29371 }, { "epoch": 1.8233285740890186, "grad_norm": 0.5948892255053286, "learning_rate": 3.998483375759398e-05, "loss": 2.8122, "step": 29372 }, { "epoch": 1.8233906511887765, "grad_norm": 0.6246301510369249, "learning_rate": 3.998129545093658e-05, "loss": 2.7196, "step": 29373 }, { "epoch": 1.8234527282885344, "grad_norm": 0.6077916531554849, "learning_rate": 3.9977757196549195e-05, "loss": 2.8098, "step": 29374 }, { "epoch": 1.8235148053882924, "grad_norm": 0.5887435099979035, "learning_rate": 3.997421899445026e-05, "loss": 2.7344, "step": 29375 }, { "epoch": 1.8235768824880503, "grad_norm": 0.5754546534489409, "learning_rate": 3.997068084465828e-05, "loss": 2.7465, "step": 29376 }, { "epoch": 1.8236389595878082, "grad_norm": 0.6165009071764185, "learning_rate": 3.9967142747191683e-05, "loss": 2.8171, "step": 29377 }, { "epoch": 1.8237010366875661, "grad_norm": 0.5662388258162007, "learning_rate": 3.9963604702068946e-05, "loss": 2.7674, "step": 29378 }, { "epoch": 1.8237631137873238, "grad_norm": 0.5756179215599679, "learning_rate": 3.996006670930851e-05, "loss": 2.6851, "step": 29379 }, { "epoch": 1.8238251908870817, "grad_norm": 0.522975433006018, "learning_rate": 3.9956528768928845e-05, "loss": 2.7058, "step": 29380 }, { "epoch": 1.8238872679868396, "grad_norm": 0.655045994315811, "learning_rate": 3.995299088094839e-05, "loss": 2.7043, "step": 29381 }, { "epoch": 1.8239493450865976, "grad_norm": 0.5054828612349216, "learning_rate": 3.994945304538564e-05, "loss": 2.7123, "step": 29382 }, { "epoch": 1.8240114221863555, "grad_norm": 0.5806623657078592, "learning_rate": 3.994591526225904e-05, "loss": 2.7263, "step": 29383 }, { "epoch": 1.8240734992861134, "grad_norm": 0.47498306821460584, "learning_rate": 3.9942377531587025e-05, "loss": 2.8083, "step": 29384 }, { "epoch": 1.824135576385871, "grad_norm": 0.6049758874013126, "learning_rate": 3.9938839853388084e-05, "loss": 2.739, "step": 29385 }, { "epoch": 1.824197653485629, "grad_norm": 0.4991530755771458, "learning_rate": 3.993530222768064e-05, "loss": 2.7092, "step": 29386 }, { "epoch": 1.824259730585387, "grad_norm": 0.6252817472552854, "learning_rate": 3.9931764654483186e-05, "loss": 2.8275, "step": 29387 }, { "epoch": 1.8243218076851448, "grad_norm": 0.5525444162648097, "learning_rate": 3.992822713381416e-05, "loss": 2.7331, "step": 29388 }, { "epoch": 1.8243838847849028, "grad_norm": 0.5923919707211983, "learning_rate": 3.9924689665692006e-05, "loss": 2.8694, "step": 29389 }, { "epoch": 1.8244459618846607, "grad_norm": 0.6132045382005783, "learning_rate": 3.99211522501352e-05, "loss": 2.7886, "step": 29390 }, { "epoch": 1.8245080389844186, "grad_norm": 0.5701512147864575, "learning_rate": 3.991761488716218e-05, "loss": 2.7277, "step": 29391 }, { "epoch": 1.8245701160841765, "grad_norm": 0.5401993525717625, "learning_rate": 3.991407757679142e-05, "loss": 2.8344, "step": 29392 }, { "epoch": 1.8246321931839344, "grad_norm": 0.5760563621463026, "learning_rate": 3.9910540319041356e-05, "loss": 2.6841, "step": 29393 }, { "epoch": 1.8246942702836924, "grad_norm": 0.5041085362090579, "learning_rate": 3.990700311393047e-05, "loss": 2.753, "step": 29394 }, { "epoch": 1.8247563473834503, "grad_norm": 0.5904338991732332, "learning_rate": 3.990346596147718e-05, "loss": 2.7101, "step": 29395 }, { "epoch": 1.8248184244832082, "grad_norm": 0.5602145123474952, "learning_rate": 3.989992886169997e-05, "loss": 2.7287, "step": 29396 }, { "epoch": 1.824880501582966, "grad_norm": 0.5717477114823031, "learning_rate": 3.9896391814617276e-05, "loss": 2.772, "step": 29397 }, { "epoch": 1.824942578682724, "grad_norm": 0.5802428657030695, "learning_rate": 3.989285482024756e-05, "loss": 2.681, "step": 29398 }, { "epoch": 1.825004655782482, "grad_norm": 0.5346267196559832, "learning_rate": 3.988931787860929e-05, "loss": 2.748, "step": 29399 }, { "epoch": 1.8250667328822399, "grad_norm": 0.5904351358165474, "learning_rate": 3.988578098972088e-05, "loss": 2.734, "step": 29400 }, { "epoch": 1.8251288099819978, "grad_norm": 0.6293454265749256, "learning_rate": 3.988224415360082e-05, "loss": 2.706, "step": 29401 }, { "epoch": 1.8251908870817557, "grad_norm": 0.6009991224714147, "learning_rate": 3.9878707370267545e-05, "loss": 2.6578, "step": 29402 }, { "epoch": 1.8252529641815134, "grad_norm": 0.6402778615374544, "learning_rate": 3.9875170639739515e-05, "loss": 2.8042, "step": 29403 }, { "epoch": 1.8253150412812713, "grad_norm": 0.5533010859840225, "learning_rate": 3.987163396203516e-05, "loss": 2.6849, "step": 29404 }, { "epoch": 1.8253771183810292, "grad_norm": 0.6664726725096335, "learning_rate": 3.9868097337172975e-05, "loss": 2.759, "step": 29405 }, { "epoch": 1.8254391954807871, "grad_norm": 0.5188877451376026, "learning_rate": 3.986456076517136e-05, "loss": 2.6957, "step": 29406 }, { "epoch": 1.825501272580545, "grad_norm": 0.6527633687947408, "learning_rate": 3.986102424604881e-05, "loss": 2.6812, "step": 29407 }, { "epoch": 1.825563349680303, "grad_norm": 0.5245434762974656, "learning_rate": 3.985748777982375e-05, "loss": 2.7662, "step": 29408 }, { "epoch": 1.8256254267800607, "grad_norm": 0.5769261224067218, "learning_rate": 3.985395136651464e-05, "loss": 2.6235, "step": 29409 }, { "epoch": 1.8256875038798186, "grad_norm": 0.5699212548172147, "learning_rate": 3.985041500613994e-05, "loss": 2.7631, "step": 29410 }, { "epoch": 1.8257495809795765, "grad_norm": 0.6314141164023908, "learning_rate": 3.9846878698718074e-05, "loss": 2.7279, "step": 29411 }, { "epoch": 1.8258116580793344, "grad_norm": 0.5436406266717876, "learning_rate": 3.9843342444267514e-05, "loss": 2.7632, "step": 29412 }, { "epoch": 1.8258737351790923, "grad_norm": 0.6281760816537382, "learning_rate": 3.9839806242806684e-05, "loss": 2.6988, "step": 29413 }, { "epoch": 1.8259358122788503, "grad_norm": 0.5561644541934853, "learning_rate": 3.983627009435406e-05, "loss": 2.6918, "step": 29414 }, { "epoch": 1.8259978893786082, "grad_norm": 0.6603591426582416, "learning_rate": 3.983273399892809e-05, "loss": 2.6287, "step": 29415 }, { "epoch": 1.826059966478366, "grad_norm": 0.5677249421889479, "learning_rate": 3.982919795654721e-05, "loss": 2.7013, "step": 29416 }, { "epoch": 1.826122043578124, "grad_norm": 0.6172786355636031, "learning_rate": 3.982566196722988e-05, "loss": 2.8436, "step": 29417 }, { "epoch": 1.826184120677882, "grad_norm": 0.5915345469791053, "learning_rate": 3.982212603099453e-05, "loss": 2.7667, "step": 29418 }, { "epoch": 1.8262461977776399, "grad_norm": 0.608235785967734, "learning_rate": 3.9818590147859634e-05, "loss": 2.6897, "step": 29419 }, { "epoch": 1.8263082748773978, "grad_norm": 0.5664569700485872, "learning_rate": 3.9815054317843616e-05, "loss": 2.7565, "step": 29420 }, { "epoch": 1.8263703519771557, "grad_norm": 0.5971928214689951, "learning_rate": 3.9811518540964936e-05, "loss": 2.7287, "step": 29421 }, { "epoch": 1.8264324290769136, "grad_norm": 0.6061564078301535, "learning_rate": 3.9807982817242026e-05, "loss": 2.8319, "step": 29422 }, { "epoch": 1.8264945061766715, "grad_norm": 0.582365874291634, "learning_rate": 3.9804447146693364e-05, "loss": 2.7513, "step": 29423 }, { "epoch": 1.8265565832764294, "grad_norm": 0.49005470174902854, "learning_rate": 3.980091152933737e-05, "loss": 2.7078, "step": 29424 }, { "epoch": 1.8266186603761874, "grad_norm": 0.5573790086652111, "learning_rate": 3.979737596519248e-05, "loss": 2.7328, "step": 29425 }, { "epoch": 1.8266807374759453, "grad_norm": 0.5805132996418083, "learning_rate": 3.9793840454277174e-05, "loss": 2.7582, "step": 29426 }, { "epoch": 1.826742814575703, "grad_norm": 0.5876988196276264, "learning_rate": 3.9790304996609874e-05, "loss": 2.8408, "step": 29427 }, { "epoch": 1.826804891675461, "grad_norm": 0.5281335638130672, "learning_rate": 3.978676959220904e-05, "loss": 2.7462, "step": 29428 }, { "epoch": 1.8268669687752188, "grad_norm": 0.6180993358117228, "learning_rate": 3.978323424109309e-05, "loss": 2.8598, "step": 29429 }, { "epoch": 1.8269290458749767, "grad_norm": 0.5656622110424375, "learning_rate": 3.9779698943280504e-05, "loss": 2.7632, "step": 29430 }, { "epoch": 1.8269911229747346, "grad_norm": 0.6101502109282455, "learning_rate": 3.977616369878969e-05, "loss": 2.7241, "step": 29431 }, { "epoch": 1.8270532000744926, "grad_norm": 0.6051207742257841, "learning_rate": 3.977262850763914e-05, "loss": 2.7312, "step": 29432 }, { "epoch": 1.8271152771742503, "grad_norm": 0.5850803972898744, "learning_rate": 3.9769093369847246e-05, "loss": 2.7107, "step": 29433 }, { "epoch": 1.8271773542740082, "grad_norm": 0.5248869002088585, "learning_rate": 3.976555828543249e-05, "loss": 2.7045, "step": 29434 }, { "epoch": 1.827239431373766, "grad_norm": 0.6302303567826587, "learning_rate": 3.97620232544133e-05, "loss": 2.8225, "step": 29435 }, { "epoch": 1.827301508473524, "grad_norm": 0.48605216289533865, "learning_rate": 3.975848827680811e-05, "loss": 2.7206, "step": 29436 }, { "epoch": 1.827363585573282, "grad_norm": 0.6194624921591063, "learning_rate": 3.975495335263538e-05, "loss": 2.7477, "step": 29437 }, { "epoch": 1.8274256626730399, "grad_norm": 0.5909531120929065, "learning_rate": 3.975141848191354e-05, "loss": 2.7862, "step": 29438 }, { "epoch": 1.8274877397727978, "grad_norm": 0.5531247381436003, "learning_rate": 3.974788366466105e-05, "loss": 2.7885, "step": 29439 }, { "epoch": 1.8275498168725557, "grad_norm": 0.6168900750974383, "learning_rate": 3.9744348900896324e-05, "loss": 2.7909, "step": 29440 }, { "epoch": 1.8276118939723136, "grad_norm": 0.5947904505497336, "learning_rate": 3.9740814190637825e-05, "loss": 2.7568, "step": 29441 }, { "epoch": 1.8276739710720715, "grad_norm": 0.6360197618596197, "learning_rate": 3.973727953390399e-05, "loss": 2.7732, "step": 29442 }, { "epoch": 1.8277360481718294, "grad_norm": 0.5284000696137445, "learning_rate": 3.9733744930713266e-05, "loss": 2.6391, "step": 29443 }, { "epoch": 1.8277981252715874, "grad_norm": 0.6401057814186366, "learning_rate": 3.973021038108407e-05, "loss": 2.6738, "step": 29444 }, { "epoch": 1.8278602023713453, "grad_norm": 0.4947129203641346, "learning_rate": 3.9726675885034876e-05, "loss": 2.7999, "step": 29445 }, { "epoch": 1.8279222794711032, "grad_norm": 0.6158298819560931, "learning_rate": 3.9723141442584085e-05, "loss": 2.8006, "step": 29446 }, { "epoch": 1.8279843565708611, "grad_norm": 0.5818148216982287, "learning_rate": 3.971960705375017e-05, "loss": 2.6957, "step": 29447 }, { "epoch": 1.828046433670619, "grad_norm": 0.5760434282309262, "learning_rate": 3.971607271855157e-05, "loss": 2.7666, "step": 29448 }, { "epoch": 1.828108510770377, "grad_norm": 0.5966474958311146, "learning_rate": 3.971253843700672e-05, "loss": 2.6907, "step": 29449 }, { "epoch": 1.8281705878701349, "grad_norm": 0.5592195966297956, "learning_rate": 3.970900420913404e-05, "loss": 2.7356, "step": 29450 }, { "epoch": 1.8282326649698926, "grad_norm": 0.5855765947837531, "learning_rate": 3.9705470034951994e-05, "loss": 2.8091, "step": 29451 }, { "epoch": 1.8282947420696505, "grad_norm": 0.5090792774599412, "learning_rate": 3.9701935914479e-05, "loss": 2.6575, "step": 29452 }, { "epoch": 1.8283568191694084, "grad_norm": 0.5889195913907306, "learning_rate": 3.969840184773351e-05, "loss": 2.7435, "step": 29453 }, { "epoch": 1.8284188962691663, "grad_norm": 0.5548598062162794, "learning_rate": 3.969486783473396e-05, "loss": 2.7558, "step": 29454 }, { "epoch": 1.8284809733689242, "grad_norm": 0.5200042865649854, "learning_rate": 3.969133387549879e-05, "loss": 2.6717, "step": 29455 }, { "epoch": 1.8285430504686822, "grad_norm": 0.5740691528599319, "learning_rate": 3.9687799970046424e-05, "loss": 2.7478, "step": 29456 }, { "epoch": 1.8286051275684398, "grad_norm": 0.5516915880818652, "learning_rate": 3.968426611839533e-05, "loss": 2.7326, "step": 29457 }, { "epoch": 1.8286672046681978, "grad_norm": 0.574231621215226, "learning_rate": 3.96807323205639e-05, "loss": 2.7696, "step": 29458 }, { "epoch": 1.8287292817679557, "grad_norm": 0.6083146430980871, "learning_rate": 3.9677198576570606e-05, "loss": 2.7831, "step": 29459 }, { "epoch": 1.8287913588677136, "grad_norm": 0.570314119937634, "learning_rate": 3.967366488643388e-05, "loss": 2.8004, "step": 29460 }, { "epoch": 1.8288534359674715, "grad_norm": 0.5720393022993189, "learning_rate": 3.967013125017213e-05, "loss": 2.7439, "step": 29461 }, { "epoch": 1.8289155130672294, "grad_norm": 0.5662266676771471, "learning_rate": 3.966659766780383e-05, "loss": 2.6835, "step": 29462 }, { "epoch": 1.8289775901669874, "grad_norm": 0.5562076691797307, "learning_rate": 3.966306413934739e-05, "loss": 2.6886, "step": 29463 }, { "epoch": 1.8290396672667453, "grad_norm": 0.5973473791700193, "learning_rate": 3.965953066482126e-05, "loss": 2.7555, "step": 29464 }, { "epoch": 1.8291017443665032, "grad_norm": 0.5884364942804414, "learning_rate": 3.965599724424386e-05, "loss": 2.7746, "step": 29465 }, { "epoch": 1.829163821466261, "grad_norm": 0.5692557119703636, "learning_rate": 3.9652463877633645e-05, "loss": 2.6281, "step": 29466 }, { "epoch": 1.829225898566019, "grad_norm": 0.6134648868162635, "learning_rate": 3.9648930565009026e-05, "loss": 2.7651, "step": 29467 }, { "epoch": 1.829287975665777, "grad_norm": 0.5962730558004388, "learning_rate": 3.964539730638846e-05, "loss": 2.719, "step": 29468 }, { "epoch": 1.8293500527655349, "grad_norm": 0.5831064623209484, "learning_rate": 3.964186410179036e-05, "loss": 2.8533, "step": 29469 }, { "epoch": 1.8294121298652928, "grad_norm": 0.6086725685880477, "learning_rate": 3.963833095123318e-05, "loss": 2.7089, "step": 29470 }, { "epoch": 1.8294742069650507, "grad_norm": 0.5770229196657312, "learning_rate": 3.963479785473534e-05, "loss": 2.7779, "step": 29471 }, { "epoch": 1.8295362840648086, "grad_norm": 0.5972653525572031, "learning_rate": 3.9631264812315265e-05, "loss": 2.7942, "step": 29472 }, { "epoch": 1.8295983611645665, "grad_norm": 0.5612907922869288, "learning_rate": 3.962773182399141e-05, "loss": 2.7206, "step": 29473 }, { "epoch": 1.8296604382643245, "grad_norm": 0.6946712708904847, "learning_rate": 3.962419888978219e-05, "loss": 2.7452, "step": 29474 }, { "epoch": 1.8297225153640821, "grad_norm": 0.5604857301758662, "learning_rate": 3.9620666009706054e-05, "loss": 2.7104, "step": 29475 }, { "epoch": 1.82978459246384, "grad_norm": 0.6111438217463747, "learning_rate": 3.961713318378141e-05, "loss": 2.7572, "step": 29476 }, { "epoch": 1.829846669563598, "grad_norm": 0.6448129442943135, "learning_rate": 3.9613600412026705e-05, "loss": 2.7624, "step": 29477 }, { "epoch": 1.829908746663356, "grad_norm": 0.5681273258694585, "learning_rate": 3.961006769446037e-05, "loss": 2.7879, "step": 29478 }, { "epoch": 1.8299708237631138, "grad_norm": 0.651669204902909, "learning_rate": 3.960653503110082e-05, "loss": 2.6647, "step": 29479 }, { "epoch": 1.8300329008628717, "grad_norm": 0.6002025093617873, "learning_rate": 3.960300242196652e-05, "loss": 2.6404, "step": 29480 }, { "epoch": 1.8300949779626294, "grad_norm": 0.6828813601062712, "learning_rate": 3.959946986707587e-05, "loss": 2.7057, "step": 29481 }, { "epoch": 1.8301570550623874, "grad_norm": 0.581243763141601, "learning_rate": 3.9595937366447316e-05, "loss": 2.7669, "step": 29482 }, { "epoch": 1.8302191321621453, "grad_norm": 0.6606617634197066, "learning_rate": 3.959240492009928e-05, "loss": 2.7291, "step": 29483 }, { "epoch": 1.8302812092619032, "grad_norm": 0.5560874465709464, "learning_rate": 3.95888725280502e-05, "loss": 2.7089, "step": 29484 }, { "epoch": 1.830343286361661, "grad_norm": 0.5964038916989184, "learning_rate": 3.9585340190318496e-05, "loss": 2.7572, "step": 29485 }, { "epoch": 1.830405363461419, "grad_norm": 0.5883451713293192, "learning_rate": 3.9581807906922595e-05, "loss": 2.6539, "step": 29486 }, { "epoch": 1.830467440561177, "grad_norm": 0.5514833831237788, "learning_rate": 3.957827567788094e-05, "loss": 2.6941, "step": 29487 }, { "epoch": 1.8305295176609349, "grad_norm": 0.5403052668972608, "learning_rate": 3.9574743503211944e-05, "loss": 2.7341, "step": 29488 }, { "epoch": 1.8305915947606928, "grad_norm": 0.5395251513991813, "learning_rate": 3.957121138293405e-05, "loss": 2.7549, "step": 29489 }, { "epoch": 1.8306536718604507, "grad_norm": 0.5407957325616041, "learning_rate": 3.956767931706566e-05, "loss": 2.6999, "step": 29490 }, { "epoch": 1.8307157489602086, "grad_norm": 0.5461883593944384, "learning_rate": 3.956414730562524e-05, "loss": 2.6831, "step": 29491 }, { "epoch": 1.8307778260599665, "grad_norm": 0.5498447437499623, "learning_rate": 3.956061534863119e-05, "loss": 2.719, "step": 29492 }, { "epoch": 1.8308399031597244, "grad_norm": 0.5118231630452502, "learning_rate": 3.9557083446101944e-05, "loss": 2.7978, "step": 29493 }, { "epoch": 1.8309019802594824, "grad_norm": 0.5312434925353967, "learning_rate": 3.9553551598055926e-05, "loss": 2.7404, "step": 29494 }, { "epoch": 1.8309640573592403, "grad_norm": 0.532282471126351, "learning_rate": 3.955001980451157e-05, "loss": 2.8054, "step": 29495 }, { "epoch": 1.8310261344589982, "grad_norm": 0.5314202877210636, "learning_rate": 3.95464880654873e-05, "loss": 2.7059, "step": 29496 }, { "epoch": 1.8310882115587561, "grad_norm": 0.5756209503594677, "learning_rate": 3.954295638100153e-05, "loss": 2.7527, "step": 29497 }, { "epoch": 1.831150288658514, "grad_norm": 0.5242958717069433, "learning_rate": 3.953942475107271e-05, "loss": 2.6585, "step": 29498 }, { "epoch": 1.8312123657582717, "grad_norm": 0.5456027900571325, "learning_rate": 3.953589317571923e-05, "loss": 2.8059, "step": 29499 }, { "epoch": 1.8312744428580296, "grad_norm": 0.5514470398306959, "learning_rate": 3.953236165495955e-05, "loss": 2.7733, "step": 29500 }, { "epoch": 1.8313365199577876, "grad_norm": 0.5791537179345577, "learning_rate": 3.952883018881207e-05, "loss": 2.7653, "step": 29501 }, { "epoch": 1.8313985970575455, "grad_norm": 0.488064939269317, "learning_rate": 3.9525298777295234e-05, "loss": 2.8375, "step": 29502 }, { "epoch": 1.8314606741573034, "grad_norm": 0.556395359548693, "learning_rate": 3.952176742042745e-05, "loss": 2.739, "step": 29503 }, { "epoch": 1.8315227512570613, "grad_norm": 0.5450816040015518, "learning_rate": 3.9518236118227154e-05, "loss": 2.7442, "step": 29504 }, { "epoch": 1.831584828356819, "grad_norm": 0.5616663211978035, "learning_rate": 3.951470487071276e-05, "loss": 2.8133, "step": 29505 }, { "epoch": 1.831646905456577, "grad_norm": 0.5518307797956514, "learning_rate": 3.95111736779027e-05, "loss": 2.7963, "step": 29506 }, { "epoch": 1.8317089825563349, "grad_norm": 0.5332294220081375, "learning_rate": 3.9507642539815394e-05, "loss": 2.6999, "step": 29507 }, { "epoch": 1.8317710596560928, "grad_norm": 0.553461896809887, "learning_rate": 3.9504111456469264e-05, "loss": 2.6961, "step": 29508 }, { "epoch": 1.8318331367558507, "grad_norm": 0.5570766949406701, "learning_rate": 3.9500580427882737e-05, "loss": 2.617, "step": 29509 }, { "epoch": 1.8318952138556086, "grad_norm": 0.5094124165933921, "learning_rate": 3.949704945407422e-05, "loss": 2.6843, "step": 29510 }, { "epoch": 1.8319572909553665, "grad_norm": 0.5908291902742095, "learning_rate": 3.9493518535062154e-05, "loss": 2.7, "step": 29511 }, { "epoch": 1.8320193680551244, "grad_norm": 0.4916194028955248, "learning_rate": 3.948998767086493e-05, "loss": 2.7533, "step": 29512 }, { "epoch": 1.8320814451548824, "grad_norm": 0.5693256776385797, "learning_rate": 3.948645686150101e-05, "loss": 2.7125, "step": 29513 }, { "epoch": 1.8321435222546403, "grad_norm": 0.5286526014023166, "learning_rate": 3.94829261069888e-05, "loss": 2.7137, "step": 29514 }, { "epoch": 1.8322055993543982, "grad_norm": 0.5548221532152258, "learning_rate": 3.947939540734671e-05, "loss": 2.671, "step": 29515 }, { "epoch": 1.8322676764541561, "grad_norm": 0.5541499851220953, "learning_rate": 3.9475864762593175e-05, "loss": 2.8312, "step": 29516 }, { "epoch": 1.832329753553914, "grad_norm": 0.56972347830172, "learning_rate": 3.94723341727466e-05, "loss": 2.7595, "step": 29517 }, { "epoch": 1.832391830653672, "grad_norm": 0.5282948147584445, "learning_rate": 3.9468803637825424e-05, "loss": 2.7595, "step": 29518 }, { "epoch": 1.8324539077534299, "grad_norm": 0.6007067551083324, "learning_rate": 3.946527315784805e-05, "loss": 2.7292, "step": 29519 }, { "epoch": 1.8325159848531878, "grad_norm": 0.5316615193764583, "learning_rate": 3.9461742732832906e-05, "loss": 2.7163, "step": 29520 }, { "epoch": 1.8325780619529457, "grad_norm": 0.5697512543159393, "learning_rate": 3.9458212362798415e-05, "loss": 2.735, "step": 29521 }, { "epoch": 1.8326401390527036, "grad_norm": 0.5312627622529988, "learning_rate": 3.945468204776297e-05, "loss": 2.683, "step": 29522 }, { "epoch": 1.8327022161524613, "grad_norm": 0.5719231892695116, "learning_rate": 3.945115178774503e-05, "loss": 2.652, "step": 29523 }, { "epoch": 1.8327642932522192, "grad_norm": 0.5450398854000879, "learning_rate": 3.944762158276298e-05, "loss": 2.7936, "step": 29524 }, { "epoch": 1.8328263703519772, "grad_norm": 0.6017795765577112, "learning_rate": 3.944409143283526e-05, "loss": 2.7064, "step": 29525 }, { "epoch": 1.832888447451735, "grad_norm": 0.5343656617179652, "learning_rate": 3.944056133798026e-05, "loss": 2.7006, "step": 29526 }, { "epoch": 1.832950524551493, "grad_norm": 0.5499482043759557, "learning_rate": 3.943703129821643e-05, "loss": 2.6994, "step": 29527 }, { "epoch": 1.833012601651251, "grad_norm": 0.4872850402663092, "learning_rate": 3.943350131356216e-05, "loss": 2.7703, "step": 29528 }, { "epoch": 1.8330746787510086, "grad_norm": 0.583098412507614, "learning_rate": 3.942997138403589e-05, "loss": 2.7049, "step": 29529 }, { "epoch": 1.8331367558507665, "grad_norm": 0.5024976040892157, "learning_rate": 3.9426441509656016e-05, "loss": 2.8287, "step": 29530 }, { "epoch": 1.8331988329505244, "grad_norm": 0.6494943440083127, "learning_rate": 3.942291169044097e-05, "loss": 2.762, "step": 29531 }, { "epoch": 1.8332609100502824, "grad_norm": 0.561077358693733, "learning_rate": 3.9419381926409156e-05, "loss": 2.7602, "step": 29532 }, { "epoch": 1.8333229871500403, "grad_norm": 0.5225304958351067, "learning_rate": 3.9415852217579005e-05, "loss": 2.7054, "step": 29533 }, { "epoch": 1.8333850642497982, "grad_norm": 0.5939020413934836, "learning_rate": 3.941232256396892e-05, "loss": 2.7134, "step": 29534 }, { "epoch": 1.833447141349556, "grad_norm": 0.5982116412447978, "learning_rate": 3.9408792965597305e-05, "loss": 2.678, "step": 29535 }, { "epoch": 1.833509218449314, "grad_norm": 0.530891606108198, "learning_rate": 3.9405263422482605e-05, "loss": 2.7568, "step": 29536 }, { "epoch": 1.833571295549072, "grad_norm": 0.6271685411432162, "learning_rate": 3.94017339346432e-05, "loss": 2.6222, "step": 29537 }, { "epoch": 1.8336333726488299, "grad_norm": 0.4848761273513831, "learning_rate": 3.939820450209754e-05, "loss": 2.6718, "step": 29538 }, { "epoch": 1.8336954497485878, "grad_norm": 0.5955731859916831, "learning_rate": 3.939467512486401e-05, "loss": 2.7638, "step": 29539 }, { "epoch": 1.8337575268483457, "grad_norm": 0.5124490200831728, "learning_rate": 3.9391145802961036e-05, "loss": 2.6774, "step": 29540 }, { "epoch": 1.8338196039481036, "grad_norm": 0.5515218314504026, "learning_rate": 3.938761653640703e-05, "loss": 2.7657, "step": 29541 }, { "epoch": 1.8338816810478615, "grad_norm": 0.5862284891456377, "learning_rate": 3.9384087325220407e-05, "loss": 2.7674, "step": 29542 }, { "epoch": 1.8339437581476195, "grad_norm": 0.5183721821057345, "learning_rate": 3.938055816941957e-05, "loss": 2.6758, "step": 29543 }, { "epoch": 1.8340058352473774, "grad_norm": 0.6636459165888514, "learning_rate": 3.937702906902295e-05, "loss": 2.7892, "step": 29544 }, { "epoch": 1.8340679123471353, "grad_norm": 0.5528189658417059, "learning_rate": 3.937350002404893e-05, "loss": 2.759, "step": 29545 }, { "epoch": 1.8341299894468932, "grad_norm": 0.5679900027657901, "learning_rate": 3.936997103451596e-05, "loss": 2.702, "step": 29546 }, { "epoch": 1.834192066546651, "grad_norm": 0.5743070262083909, "learning_rate": 3.936644210044242e-05, "loss": 2.7499, "step": 29547 }, { "epoch": 1.8342541436464088, "grad_norm": 0.5549674603657582, "learning_rate": 3.936291322184674e-05, "loss": 2.6837, "step": 29548 }, { "epoch": 1.8343162207461667, "grad_norm": 0.5470175148689574, "learning_rate": 3.935938439874732e-05, "loss": 2.7555, "step": 29549 }, { "epoch": 1.8343782978459247, "grad_norm": 0.6405630779578261, "learning_rate": 3.935585563116258e-05, "loss": 2.7211, "step": 29550 }, { "epoch": 1.8344403749456826, "grad_norm": 0.6021183715401617, "learning_rate": 3.9352326919110915e-05, "loss": 2.7938, "step": 29551 }, { "epoch": 1.8345024520454405, "grad_norm": 0.6513247497551671, "learning_rate": 3.934879826261076e-05, "loss": 2.7771, "step": 29552 }, { "epoch": 1.8345645291451982, "grad_norm": 0.6282176400187017, "learning_rate": 3.9345269661680494e-05, "loss": 2.6784, "step": 29553 }, { "epoch": 1.834626606244956, "grad_norm": 0.6250735001261222, "learning_rate": 3.934174111633856e-05, "loss": 2.821, "step": 29554 }, { "epoch": 1.834688683344714, "grad_norm": 0.6126699377319562, "learning_rate": 3.933821262660333e-05, "loss": 2.7028, "step": 29555 }, { "epoch": 1.834750760444472, "grad_norm": 0.6044371646126385, "learning_rate": 3.9334684192493256e-05, "loss": 2.7509, "step": 29556 }, { "epoch": 1.8348128375442299, "grad_norm": 0.5392716596318409, "learning_rate": 3.93311558140267e-05, "loss": 2.6518, "step": 29557 }, { "epoch": 1.8348749146439878, "grad_norm": 0.6142802757437489, "learning_rate": 3.9327627491222114e-05, "loss": 2.7, "step": 29558 }, { "epoch": 1.8349369917437457, "grad_norm": 0.6053641936562317, "learning_rate": 3.932409922409788e-05, "loss": 2.7485, "step": 29559 }, { "epoch": 1.8349990688435036, "grad_norm": 0.5971593848396755, "learning_rate": 3.9320571012672404e-05, "loss": 2.8409, "step": 29560 }, { "epoch": 1.8350611459432615, "grad_norm": 0.5678638775958237, "learning_rate": 3.9317042856964115e-05, "loss": 2.6861, "step": 29561 }, { "epoch": 1.8351232230430194, "grad_norm": 0.6518191209581733, "learning_rate": 3.9313514756991394e-05, "loss": 2.7674, "step": 29562 }, { "epoch": 1.8351853001427774, "grad_norm": 0.5612122311872249, "learning_rate": 3.930998671277267e-05, "loss": 2.7447, "step": 29563 }, { "epoch": 1.8352473772425353, "grad_norm": 0.6720157298478008, "learning_rate": 3.9306458724326326e-05, "loss": 2.7072, "step": 29564 }, { "epoch": 1.8353094543422932, "grad_norm": 0.5900561727033647, "learning_rate": 3.93029307916708e-05, "loss": 2.66, "step": 29565 }, { "epoch": 1.8353715314420511, "grad_norm": 0.6668687429733404, "learning_rate": 3.929940291482447e-05, "loss": 2.8442, "step": 29566 }, { "epoch": 1.835433608541809, "grad_norm": 0.5436208911511539, "learning_rate": 3.9295875093805764e-05, "loss": 2.7945, "step": 29567 }, { "epoch": 1.835495685641567, "grad_norm": 0.6551637060212858, "learning_rate": 3.929234732863306e-05, "loss": 2.7325, "step": 29568 }, { "epoch": 1.8355577627413249, "grad_norm": 0.5459500528954792, "learning_rate": 3.9288819619324796e-05, "loss": 2.7685, "step": 29569 }, { "epoch": 1.8356198398410828, "grad_norm": 0.573658105060778, "learning_rate": 3.9285291965899354e-05, "loss": 2.8045, "step": 29570 }, { "epoch": 1.8356819169408405, "grad_norm": 0.5571576845725189, "learning_rate": 3.928176436837514e-05, "loss": 2.7137, "step": 29571 }, { "epoch": 1.8357439940405984, "grad_norm": 0.5280844678504912, "learning_rate": 3.927823682677057e-05, "loss": 2.7038, "step": 29572 }, { "epoch": 1.8358060711403563, "grad_norm": 0.5625776661130806, "learning_rate": 3.9274709341104034e-05, "loss": 2.686, "step": 29573 }, { "epoch": 1.8358681482401142, "grad_norm": 0.5376564153146087, "learning_rate": 3.927118191139395e-05, "loss": 2.8145, "step": 29574 }, { "epoch": 1.8359302253398722, "grad_norm": 0.6362197886790665, "learning_rate": 3.926765453765871e-05, "loss": 2.7499, "step": 29575 }, { "epoch": 1.83599230243963, "grad_norm": 0.5554540200299084, "learning_rate": 3.9264127219916726e-05, "loss": 2.768, "step": 29576 }, { "epoch": 1.8360543795393878, "grad_norm": 0.5876166887261556, "learning_rate": 3.9260599958186376e-05, "loss": 2.8483, "step": 29577 }, { "epoch": 1.8361164566391457, "grad_norm": 0.5804875017836476, "learning_rate": 3.92570727524861e-05, "loss": 2.691, "step": 29578 }, { "epoch": 1.8361785337389036, "grad_norm": 0.5771722159352531, "learning_rate": 3.9253545602834286e-05, "loss": 2.7996, "step": 29579 }, { "epoch": 1.8362406108386615, "grad_norm": 0.649912546767029, "learning_rate": 3.925001850924932e-05, "loss": 2.6818, "step": 29580 }, { "epoch": 1.8363026879384194, "grad_norm": 0.5747509002147526, "learning_rate": 3.924649147174963e-05, "loss": 2.6795, "step": 29581 }, { "epoch": 1.8363647650381774, "grad_norm": 0.6376113953222436, "learning_rate": 3.9242964490353606e-05, "loss": 2.7897, "step": 29582 }, { "epoch": 1.8364268421379353, "grad_norm": 0.5597007293245226, "learning_rate": 3.923943756507965e-05, "loss": 2.6609, "step": 29583 }, { "epoch": 1.8364889192376932, "grad_norm": 0.5595223238141666, "learning_rate": 3.923591069594616e-05, "loss": 2.6862, "step": 29584 }, { "epoch": 1.8365509963374511, "grad_norm": 0.5710471920964787, "learning_rate": 3.9232383882971526e-05, "loss": 2.6556, "step": 29585 }, { "epoch": 1.836613073437209, "grad_norm": 0.5406183650320758, "learning_rate": 3.9228857126174166e-05, "loss": 2.77, "step": 29586 }, { "epoch": 1.836675150536967, "grad_norm": 0.6131061376845258, "learning_rate": 3.9225330425572474e-05, "loss": 2.6921, "step": 29587 }, { "epoch": 1.8367372276367249, "grad_norm": 0.5159828793620373, "learning_rate": 3.922180378118485e-05, "loss": 2.8335, "step": 29588 }, { "epoch": 1.8367993047364828, "grad_norm": 0.6081776715920139, "learning_rate": 3.921827719302968e-05, "loss": 2.7676, "step": 29589 }, { "epoch": 1.8368613818362407, "grad_norm": 0.5397994491131951, "learning_rate": 3.921475066112539e-05, "loss": 2.7374, "step": 29590 }, { "epoch": 1.8369234589359986, "grad_norm": 0.5076621786788732, "learning_rate": 3.921122418549035e-05, "loss": 2.7062, "step": 29591 }, { "epoch": 1.8369855360357565, "grad_norm": 0.5758793233630924, "learning_rate": 3.9207697766142994e-05, "loss": 2.7598, "step": 29592 }, { "epoch": 1.8370476131355145, "grad_norm": 0.5538480881472274, "learning_rate": 3.920417140310167e-05, "loss": 2.6507, "step": 29593 }, { "epoch": 1.8371096902352724, "grad_norm": 0.5617272844581443, "learning_rate": 3.920064509638483e-05, "loss": 2.7915, "step": 29594 }, { "epoch": 1.83717176733503, "grad_norm": 0.5925020102374263, "learning_rate": 3.919711884601084e-05, "loss": 2.7033, "step": 29595 }, { "epoch": 1.837233844434788, "grad_norm": 0.5190752652761443, "learning_rate": 3.91935926519981e-05, "loss": 2.7618, "step": 29596 }, { "epoch": 1.837295921534546, "grad_norm": 0.5285018637163257, "learning_rate": 3.919006651436501e-05, "loss": 2.7619, "step": 29597 }, { "epoch": 1.8373579986343038, "grad_norm": 0.6029638094772537, "learning_rate": 3.918654043312996e-05, "loss": 2.7369, "step": 29598 }, { "epoch": 1.8374200757340617, "grad_norm": 0.5426916609220941, "learning_rate": 3.918301440831137e-05, "loss": 2.7071, "step": 29599 }, { "epoch": 1.8374821528338194, "grad_norm": 0.5534978385872753, "learning_rate": 3.91794884399276e-05, "loss": 2.7736, "step": 29600 }, { "epoch": 1.8375442299335774, "grad_norm": 0.550214814370967, "learning_rate": 3.917596252799709e-05, "loss": 2.7287, "step": 29601 }, { "epoch": 1.8376063070333353, "grad_norm": 0.5449990111963244, "learning_rate": 3.9172436672538187e-05, "loss": 2.8496, "step": 29602 }, { "epoch": 1.8376683841330932, "grad_norm": 0.5192852154738912, "learning_rate": 3.916891087356933e-05, "loss": 2.6761, "step": 29603 }, { "epoch": 1.837730461232851, "grad_norm": 0.5370086298005842, "learning_rate": 3.916538513110888e-05, "loss": 2.824, "step": 29604 }, { "epoch": 1.837792538332609, "grad_norm": 0.5338360729097609, "learning_rate": 3.916185944517525e-05, "loss": 2.738, "step": 29605 }, { "epoch": 1.837854615432367, "grad_norm": 0.48353100124607407, "learning_rate": 3.915833381578684e-05, "loss": 2.7598, "step": 29606 }, { "epoch": 1.8379166925321249, "grad_norm": 0.5000790891085298, "learning_rate": 3.9154808242962024e-05, "loss": 2.6975, "step": 29607 }, { "epoch": 1.8379787696318828, "grad_norm": 0.5445690037954, "learning_rate": 3.9151282726719205e-05, "loss": 2.7668, "step": 29608 }, { "epoch": 1.8380408467316407, "grad_norm": 0.4772302764275949, "learning_rate": 3.914775726707678e-05, "loss": 2.7003, "step": 29609 }, { "epoch": 1.8381029238313986, "grad_norm": 0.49374390308623683, "learning_rate": 3.914423186405313e-05, "loss": 2.707, "step": 29610 }, { "epoch": 1.8381650009311565, "grad_norm": 0.5449890587522122, "learning_rate": 3.914070651766667e-05, "loss": 2.7075, "step": 29611 }, { "epoch": 1.8382270780309145, "grad_norm": 0.5362834336677774, "learning_rate": 3.913718122793577e-05, "loss": 2.7627, "step": 29612 }, { "epoch": 1.8382891551306724, "grad_norm": 0.5238214764041536, "learning_rate": 3.9133655994878846e-05, "loss": 2.6992, "step": 29613 }, { "epoch": 1.8383512322304303, "grad_norm": 0.5888780431553416, "learning_rate": 3.9130130818514265e-05, "loss": 2.7929, "step": 29614 }, { "epoch": 1.8384133093301882, "grad_norm": 0.4968386336879602, "learning_rate": 3.912660569886044e-05, "loss": 2.7292, "step": 29615 }, { "epoch": 1.8384753864299461, "grad_norm": 0.5359055501239381, "learning_rate": 3.912308063593575e-05, "loss": 2.6469, "step": 29616 }, { "epoch": 1.838537463529704, "grad_norm": 0.5602444886494402, "learning_rate": 3.911955562975859e-05, "loss": 2.6889, "step": 29617 }, { "epoch": 1.8385995406294617, "grad_norm": 0.5602611385022948, "learning_rate": 3.9116030680347344e-05, "loss": 2.8116, "step": 29618 }, { "epoch": 1.8386616177292197, "grad_norm": 0.5517790933741751, "learning_rate": 3.9112505787720416e-05, "loss": 2.7441, "step": 29619 }, { "epoch": 1.8387236948289776, "grad_norm": 0.5570516574540153, "learning_rate": 3.910898095189619e-05, "loss": 2.7399, "step": 29620 }, { "epoch": 1.8387857719287355, "grad_norm": 0.6082220663797475, "learning_rate": 3.9105456172893045e-05, "loss": 2.6624, "step": 29621 }, { "epoch": 1.8388478490284934, "grad_norm": 0.5428530068840492, "learning_rate": 3.910193145072939e-05, "loss": 2.7491, "step": 29622 }, { "epoch": 1.8389099261282513, "grad_norm": 0.6066724563766014, "learning_rate": 3.9098406785423594e-05, "loss": 2.7705, "step": 29623 }, { "epoch": 1.838972003228009, "grad_norm": 0.5844943521371978, "learning_rate": 3.909488217699406e-05, "loss": 2.8212, "step": 29624 }, { "epoch": 1.839034080327767, "grad_norm": 0.557561574323878, "learning_rate": 3.909135762545917e-05, "loss": 2.7665, "step": 29625 }, { "epoch": 1.8390961574275249, "grad_norm": 0.5932708865112762, "learning_rate": 3.908783313083732e-05, "loss": 2.6835, "step": 29626 }, { "epoch": 1.8391582345272828, "grad_norm": 0.5879949432791394, "learning_rate": 3.9084308693146884e-05, "loss": 2.7402, "step": 29627 }, { "epoch": 1.8392203116270407, "grad_norm": 0.6269947881907881, "learning_rate": 3.908078431240628e-05, "loss": 2.7211, "step": 29628 }, { "epoch": 1.8392823887267986, "grad_norm": 0.5563155982095493, "learning_rate": 3.907725998863385e-05, "loss": 2.7998, "step": 29629 }, { "epoch": 1.8393444658265565, "grad_norm": 0.6463793528094336, "learning_rate": 3.9073735721848024e-05, "loss": 2.7515, "step": 29630 }, { "epoch": 1.8394065429263144, "grad_norm": 0.5576344769631761, "learning_rate": 3.9070211512067176e-05, "loss": 2.7538, "step": 29631 }, { "epoch": 1.8394686200260724, "grad_norm": 0.5623163612269617, "learning_rate": 3.906668735930967e-05, "loss": 2.7238, "step": 29632 }, { "epoch": 1.8395306971258303, "grad_norm": 0.5998571424487174, "learning_rate": 3.906316326359393e-05, "loss": 2.7681, "step": 29633 }, { "epoch": 1.8395927742255882, "grad_norm": 0.5559633384305279, "learning_rate": 3.90596392249383e-05, "loss": 2.755, "step": 29634 }, { "epoch": 1.8396548513253461, "grad_norm": 0.6041547730250767, "learning_rate": 3.90561152433612e-05, "loss": 2.874, "step": 29635 }, { "epoch": 1.839716928425104, "grad_norm": 0.5534771084087824, "learning_rate": 3.9052591318881e-05, "loss": 2.7018, "step": 29636 }, { "epoch": 1.839779005524862, "grad_norm": 0.5568874414255092, "learning_rate": 3.90490674515161e-05, "loss": 2.8294, "step": 29637 }, { "epoch": 1.8398410826246199, "grad_norm": 0.5663948057925884, "learning_rate": 3.904554364128485e-05, "loss": 2.7261, "step": 29638 }, { "epoch": 1.8399031597243778, "grad_norm": 0.575320244840079, "learning_rate": 3.904201988820568e-05, "loss": 2.6889, "step": 29639 }, { "epoch": 1.8399652368241357, "grad_norm": 0.5934119265943387, "learning_rate": 3.9038496192296935e-05, "loss": 2.7842, "step": 29640 }, { "epoch": 1.8400273139238936, "grad_norm": 0.5339538543171981, "learning_rate": 3.903497255357703e-05, "loss": 2.7893, "step": 29641 }, { "epoch": 1.8400893910236513, "grad_norm": 0.5603321209018275, "learning_rate": 3.903144897206433e-05, "loss": 2.6711, "step": 29642 }, { "epoch": 1.8401514681234092, "grad_norm": 0.5242384007234453, "learning_rate": 3.9027925447777206e-05, "loss": 2.721, "step": 29643 }, { "epoch": 1.8402135452231672, "grad_norm": 0.5826155526639871, "learning_rate": 3.9024401980734084e-05, "loss": 2.8292, "step": 29644 }, { "epoch": 1.840275622322925, "grad_norm": 0.5729474897473965, "learning_rate": 3.9020878570953316e-05, "loss": 2.7865, "step": 29645 }, { "epoch": 1.840337699422683, "grad_norm": 0.5279570792229975, "learning_rate": 3.901735521845329e-05, "loss": 2.7207, "step": 29646 }, { "epoch": 1.840399776522441, "grad_norm": 0.5859913356208817, "learning_rate": 3.901383192325239e-05, "loss": 2.6608, "step": 29647 }, { "epoch": 1.8404618536221986, "grad_norm": 0.5637071160966811, "learning_rate": 3.901030868536898e-05, "loss": 2.7137, "step": 29648 }, { "epoch": 1.8405239307219565, "grad_norm": 0.5430952789199075, "learning_rate": 3.900678550482147e-05, "loss": 2.7728, "step": 29649 }, { "epoch": 1.8405860078217144, "grad_norm": 0.551829942697823, "learning_rate": 3.900326238162823e-05, "loss": 2.8123, "step": 29650 }, { "epoch": 1.8406480849214724, "grad_norm": 0.6572436122184847, "learning_rate": 3.899973931580764e-05, "loss": 2.6297, "step": 29651 }, { "epoch": 1.8407101620212303, "grad_norm": 0.5335945955998659, "learning_rate": 3.8996216307378073e-05, "loss": 2.6834, "step": 29652 }, { "epoch": 1.8407722391209882, "grad_norm": 0.5892990325479611, "learning_rate": 3.8992693356357926e-05, "loss": 2.8298, "step": 29653 }, { "epoch": 1.8408343162207461, "grad_norm": 0.611233103621972, "learning_rate": 3.898917046276556e-05, "loss": 2.6743, "step": 29654 }, { "epoch": 1.840896393320504, "grad_norm": 0.5531892286607021, "learning_rate": 3.898564762661938e-05, "loss": 2.7869, "step": 29655 }, { "epoch": 1.840958470420262, "grad_norm": 0.628453796835538, "learning_rate": 3.898212484793774e-05, "loss": 2.8135, "step": 29656 }, { "epoch": 1.8410205475200199, "grad_norm": 0.5873552590491053, "learning_rate": 3.8978602126739024e-05, "loss": 2.7103, "step": 29657 }, { "epoch": 1.8410826246197778, "grad_norm": 0.5859434508975151, "learning_rate": 3.897507946304163e-05, "loss": 2.6422, "step": 29658 }, { "epoch": 1.8411447017195357, "grad_norm": 0.6004516451422144, "learning_rate": 3.8971556856863914e-05, "loss": 2.6733, "step": 29659 }, { "epoch": 1.8412067788192936, "grad_norm": 0.5882944386740225, "learning_rate": 3.896803430822427e-05, "loss": 2.7564, "step": 29660 }, { "epoch": 1.8412688559190515, "grad_norm": 0.5544679771477723, "learning_rate": 3.896451181714105e-05, "loss": 2.747, "step": 29661 }, { "epoch": 1.8413309330188095, "grad_norm": 0.6709271109726778, "learning_rate": 3.8960989383632665e-05, "loss": 2.6765, "step": 29662 }, { "epoch": 1.8413930101185674, "grad_norm": 0.4991590610454372, "learning_rate": 3.895746700771748e-05, "loss": 2.7084, "step": 29663 }, { "epoch": 1.8414550872183253, "grad_norm": 0.6105903341552238, "learning_rate": 3.8953944689413865e-05, "loss": 2.7592, "step": 29664 }, { "epoch": 1.8415171643180832, "grad_norm": 0.532804430903305, "learning_rate": 3.89504224287402e-05, "loss": 2.8008, "step": 29665 }, { "epoch": 1.841579241417841, "grad_norm": 0.5727636884421459, "learning_rate": 3.894690022571488e-05, "loss": 2.8694, "step": 29666 }, { "epoch": 1.8416413185175988, "grad_norm": 0.6371797870058786, "learning_rate": 3.8943378080356244e-05, "loss": 2.685, "step": 29667 }, { "epoch": 1.8417033956173567, "grad_norm": 0.5968672587107405, "learning_rate": 3.89398559926827e-05, "loss": 2.6753, "step": 29668 }, { "epoch": 1.8417654727171147, "grad_norm": 0.6269067527282565, "learning_rate": 3.8936333962712615e-05, "loss": 2.7472, "step": 29669 }, { "epoch": 1.8418275498168726, "grad_norm": 0.5277917442509439, "learning_rate": 3.8932811990464354e-05, "loss": 2.7193, "step": 29670 }, { "epoch": 1.8418896269166305, "grad_norm": 0.6048462898281877, "learning_rate": 3.89292900759563e-05, "loss": 2.7593, "step": 29671 }, { "epoch": 1.8419517040163882, "grad_norm": 0.5239918179237221, "learning_rate": 3.8925768219206826e-05, "loss": 2.7545, "step": 29672 }, { "epoch": 1.842013781116146, "grad_norm": 0.5334512042635633, "learning_rate": 3.892224642023431e-05, "loss": 2.7669, "step": 29673 }, { "epoch": 1.842075858215904, "grad_norm": 0.5432548592460277, "learning_rate": 3.891872467905713e-05, "loss": 2.6896, "step": 29674 }, { "epoch": 1.842137935315662, "grad_norm": 0.5052203409662218, "learning_rate": 3.891520299569364e-05, "loss": 2.7567, "step": 29675 }, { "epoch": 1.8422000124154199, "grad_norm": 0.5732355588565458, "learning_rate": 3.8911681370162224e-05, "loss": 2.7225, "step": 29676 }, { "epoch": 1.8422620895151778, "grad_norm": 0.4975828514886858, "learning_rate": 3.8908159802481266e-05, "loss": 2.743, "step": 29677 }, { "epoch": 1.8423241666149357, "grad_norm": 0.5325414443566134, "learning_rate": 3.890463829266914e-05, "loss": 2.7954, "step": 29678 }, { "epoch": 1.8423862437146936, "grad_norm": 0.5241555086241039, "learning_rate": 3.89011168407442e-05, "loss": 2.6494, "step": 29679 }, { "epoch": 1.8424483208144515, "grad_norm": 0.5715610997603645, "learning_rate": 3.8897595446724836e-05, "loss": 2.7657, "step": 29680 }, { "epoch": 1.8425103979142095, "grad_norm": 0.5023636085773477, "learning_rate": 3.889407411062941e-05, "loss": 2.8161, "step": 29681 }, { "epoch": 1.8425724750139674, "grad_norm": 0.5877173539120405, "learning_rate": 3.889055283247628e-05, "loss": 2.7597, "step": 29682 }, { "epoch": 1.8426345521137253, "grad_norm": 0.5419436046373978, "learning_rate": 3.888703161228385e-05, "loss": 2.7214, "step": 29683 }, { "epoch": 1.8426966292134832, "grad_norm": 0.5593950086883077, "learning_rate": 3.8883510450070464e-05, "loss": 2.7821, "step": 29684 }, { "epoch": 1.8427587063132411, "grad_norm": 0.6193503136050525, "learning_rate": 3.887998934585451e-05, "loss": 2.6288, "step": 29685 }, { "epoch": 1.842820783412999, "grad_norm": 0.5725914349897708, "learning_rate": 3.887646829965434e-05, "loss": 2.7799, "step": 29686 }, { "epoch": 1.842882860512757, "grad_norm": 0.5866639021470579, "learning_rate": 3.887294731148835e-05, "loss": 2.7231, "step": 29687 }, { "epoch": 1.8429449376125149, "grad_norm": 0.6130463460076758, "learning_rate": 3.8869426381374886e-05, "loss": 2.7441, "step": 29688 }, { "epoch": 1.8430070147122728, "grad_norm": 0.5786774804920303, "learning_rate": 3.886590550933233e-05, "loss": 2.7437, "step": 29689 }, { "epoch": 1.8430690918120305, "grad_norm": 0.5142167251347677, "learning_rate": 3.886238469537904e-05, "loss": 2.7669, "step": 29690 }, { "epoch": 1.8431311689117884, "grad_norm": 0.641277910930149, "learning_rate": 3.8858863939533404e-05, "loss": 2.7604, "step": 29691 }, { "epoch": 1.8431932460115463, "grad_norm": 0.5390957531831264, "learning_rate": 3.885534324181378e-05, "loss": 2.7753, "step": 29692 }, { "epoch": 1.8432553231113042, "grad_norm": 0.5656496980448985, "learning_rate": 3.885182260223853e-05, "loss": 2.733, "step": 29693 }, { "epoch": 1.8433174002110622, "grad_norm": 0.5203080643931618, "learning_rate": 3.884830202082603e-05, "loss": 2.7646, "step": 29694 }, { "epoch": 1.84337947731082, "grad_norm": 0.5477249080703326, "learning_rate": 3.884478149759465e-05, "loss": 2.6596, "step": 29695 }, { "epoch": 1.8434415544105778, "grad_norm": 0.5082969702029021, "learning_rate": 3.884126103256275e-05, "loss": 2.719, "step": 29696 }, { "epoch": 1.8435036315103357, "grad_norm": 0.5750617502060662, "learning_rate": 3.8837740625748685e-05, "loss": 2.7036, "step": 29697 }, { "epoch": 1.8435657086100936, "grad_norm": 0.5549637012811605, "learning_rate": 3.883422027717086e-05, "loss": 2.7212, "step": 29698 }, { "epoch": 1.8436277857098515, "grad_norm": 0.5827254181713207, "learning_rate": 3.88306999868476e-05, "loss": 2.7625, "step": 29699 }, { "epoch": 1.8436898628096094, "grad_norm": 0.6292958208071725, "learning_rate": 3.88271797547973e-05, "loss": 2.7395, "step": 29700 }, { "epoch": 1.8437519399093674, "grad_norm": 0.5790126275512215, "learning_rate": 3.882365958103831e-05, "loss": 2.7607, "step": 29701 }, { "epoch": 1.8438140170091253, "grad_norm": 0.5515933171381285, "learning_rate": 3.8820139465589005e-05, "loss": 2.6677, "step": 29702 }, { "epoch": 1.8438760941088832, "grad_norm": 0.5715492894818747, "learning_rate": 3.881661940846774e-05, "loss": 2.7729, "step": 29703 }, { "epoch": 1.8439381712086411, "grad_norm": 0.616176960971176, "learning_rate": 3.8813099409692896e-05, "loss": 2.7387, "step": 29704 }, { "epoch": 1.844000248308399, "grad_norm": 0.5509687605031287, "learning_rate": 3.8809579469282834e-05, "loss": 2.7461, "step": 29705 }, { "epoch": 1.844062325408157, "grad_norm": 0.6002837374467502, "learning_rate": 3.880605958725589e-05, "loss": 2.7445, "step": 29706 }, { "epoch": 1.8441244025079149, "grad_norm": 0.5281240835689742, "learning_rate": 3.880253976363047e-05, "loss": 2.7446, "step": 29707 }, { "epoch": 1.8441864796076728, "grad_norm": 0.6134275759935346, "learning_rate": 3.87990199984249e-05, "loss": 2.7254, "step": 29708 }, { "epoch": 1.8442485567074307, "grad_norm": 0.5645510050772022, "learning_rate": 3.879550029165756e-05, "loss": 2.6297, "step": 29709 }, { "epoch": 1.8443106338071886, "grad_norm": 0.5829379041410027, "learning_rate": 3.8791980643346835e-05, "loss": 2.7058, "step": 29710 }, { "epoch": 1.8443727109069465, "grad_norm": 0.6596512301768729, "learning_rate": 3.878846105351105e-05, "loss": 2.7488, "step": 29711 }, { "epoch": 1.8444347880067045, "grad_norm": 0.5478574647092522, "learning_rate": 3.8784941522168606e-05, "loss": 2.7186, "step": 29712 }, { "epoch": 1.8444968651064624, "grad_norm": 0.6405605190669663, "learning_rate": 3.878142204933783e-05, "loss": 2.7206, "step": 29713 }, { "epoch": 1.84455894220622, "grad_norm": 0.5203130189360359, "learning_rate": 3.877790263503711e-05, "loss": 2.7115, "step": 29714 }, { "epoch": 1.844621019305978, "grad_norm": 0.5528820635822823, "learning_rate": 3.877438327928478e-05, "loss": 2.7149, "step": 29715 }, { "epoch": 1.844683096405736, "grad_norm": 0.6207860028276755, "learning_rate": 3.877086398209924e-05, "loss": 2.682, "step": 29716 }, { "epoch": 1.8447451735054938, "grad_norm": 0.5541887862110022, "learning_rate": 3.8767344743498813e-05, "loss": 2.6819, "step": 29717 }, { "epoch": 1.8448072506052517, "grad_norm": 0.5934085366009513, "learning_rate": 3.876382556350187e-05, "loss": 2.786, "step": 29718 }, { "epoch": 1.8448693277050097, "grad_norm": 0.5709413954043907, "learning_rate": 3.8760306442126796e-05, "loss": 2.7682, "step": 29719 }, { "epoch": 1.8449314048047674, "grad_norm": 0.5576928349639161, "learning_rate": 3.875678737939191e-05, "loss": 2.7008, "step": 29720 }, { "epoch": 1.8449934819045253, "grad_norm": 0.5848897084985193, "learning_rate": 3.875326837531561e-05, "loss": 2.6688, "step": 29721 }, { "epoch": 1.8450555590042832, "grad_norm": 0.5224235164752887, "learning_rate": 3.874974942991623e-05, "loss": 2.7534, "step": 29722 }, { "epoch": 1.8451176361040411, "grad_norm": 0.5820303539294828, "learning_rate": 3.874623054321215e-05, "loss": 2.8108, "step": 29723 }, { "epoch": 1.845179713203799, "grad_norm": 0.602883461532359, "learning_rate": 3.87427117152217e-05, "loss": 2.823, "step": 29724 }, { "epoch": 1.845241790303557, "grad_norm": 0.49685460414651544, "learning_rate": 3.873919294596327e-05, "loss": 2.6284, "step": 29725 }, { "epoch": 1.8453038674033149, "grad_norm": 0.6317592851464424, "learning_rate": 3.8735674235455194e-05, "loss": 2.7267, "step": 29726 }, { "epoch": 1.8453659445030728, "grad_norm": 0.4933449894853335, "learning_rate": 3.873215558371585e-05, "loss": 2.7199, "step": 29727 }, { "epoch": 1.8454280216028307, "grad_norm": 0.624746271129164, "learning_rate": 3.8728636990763575e-05, "loss": 2.7275, "step": 29728 }, { "epoch": 1.8454900987025886, "grad_norm": 0.5469086772903418, "learning_rate": 3.872511845661674e-05, "loss": 2.7417, "step": 29729 }, { "epoch": 1.8455521758023465, "grad_norm": 0.5836340832848026, "learning_rate": 3.872159998129371e-05, "loss": 2.7544, "step": 29730 }, { "epoch": 1.8456142529021045, "grad_norm": 0.5584487193239528, "learning_rate": 3.871808156481282e-05, "loss": 2.74, "step": 29731 }, { "epoch": 1.8456763300018624, "grad_norm": 0.5677642853021871, "learning_rate": 3.871456320719245e-05, "loss": 2.7259, "step": 29732 }, { "epoch": 1.8457384071016203, "grad_norm": 0.5740239597978841, "learning_rate": 3.871104490845092e-05, "loss": 2.7515, "step": 29733 }, { "epoch": 1.8458004842013782, "grad_norm": 0.5651186019764793, "learning_rate": 3.870752666860662e-05, "loss": 2.744, "step": 29734 }, { "epoch": 1.8458625613011361, "grad_norm": 0.5742227463356099, "learning_rate": 3.8704008487677894e-05, "loss": 2.7364, "step": 29735 }, { "epoch": 1.845924638400894, "grad_norm": 0.6400561947664264, "learning_rate": 3.870049036568311e-05, "loss": 2.725, "step": 29736 }, { "epoch": 1.845986715500652, "grad_norm": 0.585034552067294, "learning_rate": 3.869697230264059e-05, "loss": 2.7464, "step": 29737 }, { "epoch": 1.8460487926004097, "grad_norm": 0.5865206804442542, "learning_rate": 3.869345429856873e-05, "loss": 2.8068, "step": 29738 }, { "epoch": 1.8461108697001676, "grad_norm": 0.5448608364251796, "learning_rate": 3.868993635348585e-05, "loss": 2.7296, "step": 29739 }, { "epoch": 1.8461729467999255, "grad_norm": 0.5786399352296407, "learning_rate": 3.868641846741032e-05, "loss": 2.6634, "step": 29740 }, { "epoch": 1.8462350238996834, "grad_norm": 0.5433341244631353, "learning_rate": 3.868290064036049e-05, "loss": 2.7001, "step": 29741 }, { "epoch": 1.8462971009994413, "grad_norm": 0.5377230103962938, "learning_rate": 3.867938287235472e-05, "loss": 2.7087, "step": 29742 }, { "epoch": 1.8463591780991992, "grad_norm": 0.5995408307886506, "learning_rate": 3.867586516341135e-05, "loss": 2.7759, "step": 29743 }, { "epoch": 1.846421255198957, "grad_norm": 0.5138035304543298, "learning_rate": 3.8672347513548745e-05, "loss": 2.7334, "step": 29744 }, { "epoch": 1.8464833322987149, "grad_norm": 0.6197368506759701, "learning_rate": 3.8668829922785256e-05, "loss": 2.8102, "step": 29745 }, { "epoch": 1.8465454093984728, "grad_norm": 0.5419111311503356, "learning_rate": 3.866531239113923e-05, "loss": 2.6604, "step": 29746 }, { "epoch": 1.8466074864982307, "grad_norm": 0.5331610359096487, "learning_rate": 3.866179491862902e-05, "loss": 2.7131, "step": 29747 }, { "epoch": 1.8466695635979886, "grad_norm": 0.6010529477952123, "learning_rate": 3.865827750527298e-05, "loss": 2.7981, "step": 29748 }, { "epoch": 1.8467316406977465, "grad_norm": 0.5925684516333447, "learning_rate": 3.865476015108945e-05, "loss": 2.7302, "step": 29749 }, { "epoch": 1.8467937177975045, "grad_norm": 0.5810150389407415, "learning_rate": 3.865124285609681e-05, "loss": 2.6764, "step": 29750 }, { "epoch": 1.8468557948972624, "grad_norm": 0.5657644911261668, "learning_rate": 3.864772562031337e-05, "loss": 2.6458, "step": 29751 }, { "epoch": 1.8469178719970203, "grad_norm": 0.5852957962756564, "learning_rate": 3.8644208443757526e-05, "loss": 2.6424, "step": 29752 }, { "epoch": 1.8469799490967782, "grad_norm": 0.5056391350711915, "learning_rate": 3.864069132644758e-05, "loss": 2.7196, "step": 29753 }, { "epoch": 1.8470420261965361, "grad_norm": 0.5458096344974265, "learning_rate": 3.8637174268401924e-05, "loss": 2.7179, "step": 29754 }, { "epoch": 1.847104103296294, "grad_norm": 0.5859630070538625, "learning_rate": 3.8633657269638876e-05, "loss": 2.6574, "step": 29755 }, { "epoch": 1.847166180396052, "grad_norm": 0.5424154467547251, "learning_rate": 3.8630140330176794e-05, "loss": 2.8345, "step": 29756 }, { "epoch": 1.8472282574958099, "grad_norm": 0.5635306135728358, "learning_rate": 3.862662345003404e-05, "loss": 2.7948, "step": 29757 }, { "epoch": 1.8472903345955678, "grad_norm": 0.5660619807561594, "learning_rate": 3.8623106629228944e-05, "loss": 2.7913, "step": 29758 }, { "epoch": 1.8473524116953257, "grad_norm": 0.5741559986937601, "learning_rate": 3.861958986777987e-05, "loss": 2.7155, "step": 29759 }, { "epoch": 1.8474144887950836, "grad_norm": 0.5978520390565675, "learning_rate": 3.8616073165705154e-05, "loss": 2.7615, "step": 29760 }, { "epoch": 1.8474765658948415, "grad_norm": 0.542287000601126, "learning_rate": 3.861255652302315e-05, "loss": 2.7119, "step": 29761 }, { "epoch": 1.8475386429945992, "grad_norm": 0.6026690573168421, "learning_rate": 3.86090399397522e-05, "loss": 2.7877, "step": 29762 }, { "epoch": 1.8476007200943572, "grad_norm": 0.5200798465332951, "learning_rate": 3.860552341591066e-05, "loss": 2.669, "step": 29763 }, { "epoch": 1.847662797194115, "grad_norm": 0.5900534957943813, "learning_rate": 3.8602006951516864e-05, "loss": 2.7996, "step": 29764 }, { "epoch": 1.847724874293873, "grad_norm": 0.6461757372645127, "learning_rate": 3.859849054658917e-05, "loss": 2.7197, "step": 29765 }, { "epoch": 1.847786951393631, "grad_norm": 0.5769765891725602, "learning_rate": 3.859497420114592e-05, "loss": 2.8515, "step": 29766 }, { "epoch": 1.8478490284933888, "grad_norm": 0.6529269324223722, "learning_rate": 3.8591457915205446e-05, "loss": 2.7902, "step": 29767 }, { "epoch": 1.8479111055931465, "grad_norm": 0.5727357983470668, "learning_rate": 3.858794168878612e-05, "loss": 2.7188, "step": 29768 }, { "epoch": 1.8479731826929044, "grad_norm": 0.5774970639554646, "learning_rate": 3.858442552190626e-05, "loss": 2.7886, "step": 29769 }, { "epoch": 1.8480352597926624, "grad_norm": 0.5944927238182708, "learning_rate": 3.858090941458423e-05, "loss": 2.7711, "step": 29770 }, { "epoch": 1.8480973368924203, "grad_norm": 0.5551138032189027, "learning_rate": 3.8577393366838364e-05, "loss": 2.7712, "step": 29771 }, { "epoch": 1.8481594139921782, "grad_norm": 0.538732289335154, "learning_rate": 3.8573877378687015e-05, "loss": 2.7611, "step": 29772 }, { "epoch": 1.8482214910919361, "grad_norm": 0.5707908348868092, "learning_rate": 3.857036145014852e-05, "loss": 2.6915, "step": 29773 }, { "epoch": 1.848283568191694, "grad_norm": 0.6182298670451164, "learning_rate": 3.8566845581241206e-05, "loss": 2.7614, "step": 29774 }, { "epoch": 1.848345645291452, "grad_norm": 0.5971853108667865, "learning_rate": 3.856332977198345e-05, "loss": 2.7125, "step": 29775 }, { "epoch": 1.8484077223912099, "grad_norm": 0.5606707924435133, "learning_rate": 3.855981402239358e-05, "loss": 2.6846, "step": 29776 }, { "epoch": 1.8484697994909678, "grad_norm": 0.6266255934381633, "learning_rate": 3.855629833248994e-05, "loss": 2.7512, "step": 29777 }, { "epoch": 1.8485318765907257, "grad_norm": 0.5554564076271677, "learning_rate": 3.8552782702290865e-05, "loss": 2.7594, "step": 29778 }, { "epoch": 1.8485939536904836, "grad_norm": 0.5852832192970299, "learning_rate": 3.85492671318147e-05, "loss": 2.739, "step": 29779 }, { "epoch": 1.8486560307902415, "grad_norm": 0.6611925377062929, "learning_rate": 3.8545751621079806e-05, "loss": 2.7206, "step": 29780 }, { "epoch": 1.8487181078899995, "grad_norm": 0.6081133522023799, "learning_rate": 3.854223617010449e-05, "loss": 2.8395, "step": 29781 }, { "epoch": 1.8487801849897574, "grad_norm": 0.6483916530717269, "learning_rate": 3.853872077890711e-05, "loss": 2.7201, "step": 29782 }, { "epoch": 1.8488422620895153, "grad_norm": 0.6014810729961182, "learning_rate": 3.8535205447506e-05, "loss": 2.7366, "step": 29783 }, { "epoch": 1.8489043391892732, "grad_norm": 0.6642884038505954, "learning_rate": 3.853169017591952e-05, "loss": 2.7324, "step": 29784 }, { "epoch": 1.8489664162890311, "grad_norm": 0.5810797934357554, "learning_rate": 3.8528174964165986e-05, "loss": 2.7522, "step": 29785 }, { "epoch": 1.8490284933887888, "grad_norm": 0.6491676034989982, "learning_rate": 3.852465981226376e-05, "loss": 2.6921, "step": 29786 }, { "epoch": 1.8490905704885467, "grad_norm": 0.6097632662857642, "learning_rate": 3.852114472023115e-05, "loss": 2.6711, "step": 29787 }, { "epoch": 1.8491526475883047, "grad_norm": 0.5947091486844699, "learning_rate": 3.8517629688086535e-05, "loss": 2.722, "step": 29788 }, { "epoch": 1.8492147246880626, "grad_norm": 0.6175737982833024, "learning_rate": 3.851411471584822e-05, "loss": 2.7542, "step": 29789 }, { "epoch": 1.8492768017878205, "grad_norm": 0.6247203916713494, "learning_rate": 3.851059980353457e-05, "loss": 2.8059, "step": 29790 }, { "epoch": 1.8493388788875784, "grad_norm": 0.5909819578879738, "learning_rate": 3.8507084951163904e-05, "loss": 2.7151, "step": 29791 }, { "epoch": 1.8494009559873361, "grad_norm": 0.5234774389260343, "learning_rate": 3.850357015875456e-05, "loss": 2.7179, "step": 29792 }, { "epoch": 1.849463033087094, "grad_norm": 0.5805714112139054, "learning_rate": 3.850005542632489e-05, "loss": 2.811, "step": 29793 }, { "epoch": 1.849525110186852, "grad_norm": 0.5453454835639127, "learning_rate": 3.849654075389321e-05, "loss": 2.7532, "step": 29794 }, { "epoch": 1.8495871872866099, "grad_norm": 0.6373993681973646, "learning_rate": 3.8493026141477884e-05, "loss": 2.7534, "step": 29795 }, { "epoch": 1.8496492643863678, "grad_norm": 0.559051566147647, "learning_rate": 3.848951158909721e-05, "loss": 2.6661, "step": 29796 }, { "epoch": 1.8497113414861257, "grad_norm": 0.6067988398777855, "learning_rate": 3.848599709676958e-05, "loss": 2.8468, "step": 29797 }, { "epoch": 1.8497734185858836, "grad_norm": 0.5915141673628483, "learning_rate": 3.848248266451328e-05, "loss": 2.6427, "step": 29798 }, { "epoch": 1.8498354956856415, "grad_norm": 0.545917380998882, "learning_rate": 3.847896829234667e-05, "loss": 2.6509, "step": 29799 }, { "epoch": 1.8498975727853995, "grad_norm": 0.5937802909352262, "learning_rate": 3.847545398028807e-05, "loss": 2.7358, "step": 29800 }, { "epoch": 1.8499596498851574, "grad_norm": 0.5441668622550594, "learning_rate": 3.847193972835584e-05, "loss": 2.7182, "step": 29801 }, { "epoch": 1.8500217269849153, "grad_norm": 0.5818801017971897, "learning_rate": 3.8468425536568295e-05, "loss": 2.6775, "step": 29802 }, { "epoch": 1.8500838040846732, "grad_norm": 0.513341674267311, "learning_rate": 3.846491140494376e-05, "loss": 2.704, "step": 29803 }, { "epoch": 1.8501458811844311, "grad_norm": 0.5102994907587416, "learning_rate": 3.846139733350059e-05, "loss": 2.7053, "step": 29804 }, { "epoch": 1.850207958284189, "grad_norm": 0.545692609768962, "learning_rate": 3.845788332225711e-05, "loss": 2.6961, "step": 29805 }, { "epoch": 1.850270035383947, "grad_norm": 0.45926230923286926, "learning_rate": 3.845436937123166e-05, "loss": 2.5758, "step": 29806 }, { "epoch": 1.8503321124837049, "grad_norm": 0.5161850517557608, "learning_rate": 3.845085548044255e-05, "loss": 2.7402, "step": 29807 }, { "epoch": 1.8503941895834628, "grad_norm": 0.5147099611978238, "learning_rate": 3.844734164990814e-05, "loss": 2.7325, "step": 29808 }, { "epoch": 1.8504562666832207, "grad_norm": 0.4925168161861588, "learning_rate": 3.8443827879646765e-05, "loss": 2.7694, "step": 29809 }, { "epoch": 1.8505183437829784, "grad_norm": 0.5270186093010728, "learning_rate": 3.844031416967673e-05, "loss": 2.7368, "step": 29810 }, { "epoch": 1.8505804208827363, "grad_norm": 0.5384919007683119, "learning_rate": 3.8436800520016394e-05, "loss": 2.7154, "step": 29811 }, { "epoch": 1.8506424979824943, "grad_norm": 0.5253957017844721, "learning_rate": 3.843328693068407e-05, "loss": 2.7174, "step": 29812 }, { "epoch": 1.8507045750822522, "grad_norm": 0.5976939220647892, "learning_rate": 3.8429773401698106e-05, "loss": 2.7841, "step": 29813 }, { "epoch": 1.85076665218201, "grad_norm": 0.5133514855579295, "learning_rate": 3.842625993307681e-05, "loss": 2.6827, "step": 29814 }, { "epoch": 1.850828729281768, "grad_norm": 0.5338491082921003, "learning_rate": 3.842274652483854e-05, "loss": 2.7355, "step": 29815 }, { "epoch": 1.8508908063815257, "grad_norm": 0.5767547503996268, "learning_rate": 3.8419233177001605e-05, "loss": 2.721, "step": 29816 }, { "epoch": 1.8509528834812836, "grad_norm": 0.5653984575096737, "learning_rate": 3.8415719889584336e-05, "loss": 2.6989, "step": 29817 }, { "epoch": 1.8510149605810415, "grad_norm": 0.6106944105432531, "learning_rate": 3.8412206662605077e-05, "loss": 2.7196, "step": 29818 }, { "epoch": 1.8510770376807995, "grad_norm": 0.5815205477629094, "learning_rate": 3.8408693496082136e-05, "loss": 2.6689, "step": 29819 }, { "epoch": 1.8511391147805574, "grad_norm": 0.591246244890414, "learning_rate": 3.840518039003387e-05, "loss": 2.7623, "step": 29820 }, { "epoch": 1.8512011918803153, "grad_norm": 0.5425263667840758, "learning_rate": 3.840166734447858e-05, "loss": 2.7997, "step": 29821 }, { "epoch": 1.8512632689800732, "grad_norm": 0.5982408593910612, "learning_rate": 3.8398154359434625e-05, "loss": 2.6778, "step": 29822 }, { "epoch": 1.8513253460798311, "grad_norm": 0.4873062623023386, "learning_rate": 3.83946414349203e-05, "loss": 2.6918, "step": 29823 }, { "epoch": 1.851387423179589, "grad_norm": 0.5586770269033414, "learning_rate": 3.839112857095395e-05, "loss": 2.6918, "step": 29824 }, { "epoch": 1.851449500279347, "grad_norm": 0.4809300913745129, "learning_rate": 3.83876157675539e-05, "loss": 2.7054, "step": 29825 }, { "epoch": 1.8515115773791049, "grad_norm": 0.572391274885007, "learning_rate": 3.8384103024738484e-05, "loss": 2.8053, "step": 29826 }, { "epoch": 1.8515736544788628, "grad_norm": 0.4977748950995884, "learning_rate": 3.8380590342526025e-05, "loss": 2.7176, "step": 29827 }, { "epoch": 1.8516357315786207, "grad_norm": 0.5333384247874579, "learning_rate": 3.8377077720934826e-05, "loss": 2.8083, "step": 29828 }, { "epoch": 1.8516978086783786, "grad_norm": 0.5573290666764055, "learning_rate": 3.837356515998326e-05, "loss": 2.8349, "step": 29829 }, { "epoch": 1.8517598857781365, "grad_norm": 0.5404810278259722, "learning_rate": 3.8370052659689604e-05, "loss": 2.6219, "step": 29830 }, { "epoch": 1.8518219628778945, "grad_norm": 0.5484175045356945, "learning_rate": 3.836654022007222e-05, "loss": 2.7789, "step": 29831 }, { "epoch": 1.8518840399776524, "grad_norm": 0.5115864997784831, "learning_rate": 3.836302784114942e-05, "loss": 2.8143, "step": 29832 }, { "epoch": 1.8519461170774103, "grad_norm": 0.5711901195953168, "learning_rate": 3.8359515522939526e-05, "loss": 2.7119, "step": 29833 }, { "epoch": 1.852008194177168, "grad_norm": 0.5333471976902069, "learning_rate": 3.835600326546086e-05, "loss": 2.7518, "step": 29834 }, { "epoch": 1.852070271276926, "grad_norm": 0.5229355127299798, "learning_rate": 3.835249106873177e-05, "loss": 2.7009, "step": 29835 }, { "epoch": 1.8521323483766838, "grad_norm": 0.5341116741615151, "learning_rate": 3.834897893277054e-05, "loss": 2.6622, "step": 29836 }, { "epoch": 1.8521944254764418, "grad_norm": 0.5402422448589747, "learning_rate": 3.8345466857595524e-05, "loss": 2.8125, "step": 29837 }, { "epoch": 1.8522565025761997, "grad_norm": 0.5064322273161636, "learning_rate": 3.8341954843225033e-05, "loss": 2.7026, "step": 29838 }, { "epoch": 1.8523185796759576, "grad_norm": 0.5415076486993083, "learning_rate": 3.833844288967738e-05, "loss": 2.7701, "step": 29839 }, { "epoch": 1.8523806567757153, "grad_norm": 0.5046097627653932, "learning_rate": 3.8334930996970925e-05, "loss": 2.7083, "step": 29840 }, { "epoch": 1.8524427338754732, "grad_norm": 0.45964771791014236, "learning_rate": 3.8331419165123966e-05, "loss": 2.7649, "step": 29841 }, { "epoch": 1.8525048109752311, "grad_norm": 0.5236599577569706, "learning_rate": 3.832790739415482e-05, "loss": 2.7531, "step": 29842 }, { "epoch": 1.852566888074989, "grad_norm": 0.5219305751242029, "learning_rate": 3.8324395684081814e-05, "loss": 2.7414, "step": 29843 }, { "epoch": 1.852628965174747, "grad_norm": 0.5249854301519918, "learning_rate": 3.832088403492327e-05, "loss": 2.7534, "step": 29844 }, { "epoch": 1.8526910422745049, "grad_norm": 0.5609341390184961, "learning_rate": 3.8317372446697506e-05, "loss": 2.69, "step": 29845 }, { "epoch": 1.8527531193742628, "grad_norm": 0.5477297234327501, "learning_rate": 3.831386091942285e-05, "loss": 2.6909, "step": 29846 }, { "epoch": 1.8528151964740207, "grad_norm": 0.6615291742752409, "learning_rate": 3.8310349453117624e-05, "loss": 2.7938, "step": 29847 }, { "epoch": 1.8528772735737786, "grad_norm": 0.5447341157944159, "learning_rate": 3.830683804780013e-05, "loss": 2.7608, "step": 29848 }, { "epoch": 1.8529393506735365, "grad_norm": 0.5703760921610772, "learning_rate": 3.8303326703488714e-05, "loss": 2.7439, "step": 29849 }, { "epoch": 1.8530014277732945, "grad_norm": 0.5626136082256816, "learning_rate": 3.829981542020167e-05, "loss": 2.6722, "step": 29850 }, { "epoch": 1.8530635048730524, "grad_norm": 0.5321992079678376, "learning_rate": 3.829630419795734e-05, "loss": 2.6735, "step": 29851 }, { "epoch": 1.8531255819728103, "grad_norm": 0.508990754843578, "learning_rate": 3.829279303677403e-05, "loss": 2.7272, "step": 29852 }, { "epoch": 1.8531876590725682, "grad_norm": 0.5673431802850256, "learning_rate": 3.8289281936670054e-05, "loss": 2.7376, "step": 29853 }, { "epoch": 1.8532497361723261, "grad_norm": 0.5526138012400006, "learning_rate": 3.828577089766374e-05, "loss": 2.7401, "step": 29854 }, { "epoch": 1.853311813272084, "grad_norm": 0.541523223614948, "learning_rate": 3.82822599197734e-05, "loss": 2.7174, "step": 29855 }, { "epoch": 1.853373890371842, "grad_norm": 0.5257850216054332, "learning_rate": 3.8278749003017366e-05, "loss": 2.82, "step": 29856 }, { "epoch": 1.8534359674715999, "grad_norm": 0.531304048144978, "learning_rate": 3.8275238147413934e-05, "loss": 2.712, "step": 29857 }, { "epoch": 1.8534980445713576, "grad_norm": 0.5930935845129867, "learning_rate": 3.827172735298143e-05, "loss": 2.8054, "step": 29858 }, { "epoch": 1.8535601216711155, "grad_norm": 0.543366128974146, "learning_rate": 3.826821661973817e-05, "loss": 2.719, "step": 29859 }, { "epoch": 1.8536221987708734, "grad_norm": 0.5903368019687425, "learning_rate": 3.826470594770248e-05, "loss": 2.7899, "step": 29860 }, { "epoch": 1.8536842758706313, "grad_norm": 0.5322071876849035, "learning_rate": 3.826119533689266e-05, "loss": 2.731, "step": 29861 }, { "epoch": 1.8537463529703893, "grad_norm": 0.5371698241463061, "learning_rate": 3.825768478732705e-05, "loss": 2.797, "step": 29862 }, { "epoch": 1.8538084300701472, "grad_norm": 0.5637099709858386, "learning_rate": 3.8254174299023934e-05, "loss": 2.7773, "step": 29863 }, { "epoch": 1.8538705071699049, "grad_norm": 0.45773867119229605, "learning_rate": 3.8250663872001644e-05, "loss": 2.7662, "step": 29864 }, { "epoch": 1.8539325842696628, "grad_norm": 0.5800504067731177, "learning_rate": 3.8247153506278504e-05, "loss": 2.7286, "step": 29865 }, { "epoch": 1.8539946613694207, "grad_norm": 0.5200951749560586, "learning_rate": 3.8243643201872805e-05, "loss": 2.7481, "step": 29866 }, { "epoch": 1.8540567384691786, "grad_norm": 0.5453799383659877, "learning_rate": 3.8240132958802876e-05, "loss": 2.7505, "step": 29867 }, { "epoch": 1.8541188155689365, "grad_norm": 0.6480368330018229, "learning_rate": 3.823662277708702e-05, "loss": 2.6674, "step": 29868 }, { "epoch": 1.8541808926686945, "grad_norm": 0.5328525886109087, "learning_rate": 3.823311265674358e-05, "loss": 2.7436, "step": 29869 }, { "epoch": 1.8542429697684524, "grad_norm": 0.5969144169040185, "learning_rate": 3.822960259779083e-05, "loss": 2.796, "step": 29870 }, { "epoch": 1.8543050468682103, "grad_norm": 0.5416110236369494, "learning_rate": 3.822609260024711e-05, "loss": 2.6563, "step": 29871 }, { "epoch": 1.8543671239679682, "grad_norm": 0.528055798212074, "learning_rate": 3.82225826641307e-05, "loss": 2.6915, "step": 29872 }, { "epoch": 1.8544292010677261, "grad_norm": 0.5851773735685015, "learning_rate": 3.821907278945994e-05, "loss": 2.734, "step": 29873 }, { "epoch": 1.854491278167484, "grad_norm": 0.5832735411336132, "learning_rate": 3.821556297625316e-05, "loss": 2.8338, "step": 29874 }, { "epoch": 1.854553355267242, "grad_norm": 0.48879567323145584, "learning_rate": 3.821205322452863e-05, "loss": 2.5878, "step": 29875 }, { "epoch": 1.8546154323669999, "grad_norm": 0.603340518572298, "learning_rate": 3.82085435343047e-05, "loss": 2.7625, "step": 29876 }, { "epoch": 1.8546775094667578, "grad_norm": 0.5467127695598724, "learning_rate": 3.820503390559965e-05, "loss": 2.7274, "step": 29877 }, { "epoch": 1.8547395865665157, "grad_norm": 0.5892276704366359, "learning_rate": 3.82015243384318e-05, "loss": 2.7641, "step": 29878 }, { "epoch": 1.8548016636662736, "grad_norm": 0.5731343520471627, "learning_rate": 3.819801483281947e-05, "loss": 2.7943, "step": 29879 }, { "epoch": 1.8548637407660316, "grad_norm": 0.6349615410925379, "learning_rate": 3.819450538878094e-05, "loss": 2.7397, "step": 29880 }, { "epoch": 1.8549258178657895, "grad_norm": 0.5645735984985865, "learning_rate": 3.819099600633457e-05, "loss": 2.7283, "step": 29881 }, { "epoch": 1.8549878949655472, "grad_norm": 0.6026678705022331, "learning_rate": 3.8187486685498616e-05, "loss": 2.7638, "step": 29882 }, { "epoch": 1.855049972065305, "grad_norm": 0.5713286296477633, "learning_rate": 3.818397742629143e-05, "loss": 2.6733, "step": 29883 }, { "epoch": 1.855112049165063, "grad_norm": 0.6076412234185801, "learning_rate": 3.8180468228731283e-05, "loss": 2.8574, "step": 29884 }, { "epoch": 1.855174126264821, "grad_norm": 0.5354821820629516, "learning_rate": 3.8176959092836524e-05, "loss": 2.6822, "step": 29885 }, { "epoch": 1.8552362033645788, "grad_norm": 0.5865400903289519, "learning_rate": 3.817345001862542e-05, "loss": 2.7177, "step": 29886 }, { "epoch": 1.8552982804643368, "grad_norm": 0.48473340280123817, "learning_rate": 3.816994100611632e-05, "loss": 2.7531, "step": 29887 }, { "epoch": 1.8553603575640945, "grad_norm": 0.6162996303853727, "learning_rate": 3.81664320553275e-05, "loss": 2.8312, "step": 29888 }, { "epoch": 1.8554224346638524, "grad_norm": 0.550798726988203, "learning_rate": 3.816292316627727e-05, "loss": 2.836, "step": 29889 }, { "epoch": 1.8554845117636103, "grad_norm": 0.5685553458657838, "learning_rate": 3.815941433898395e-05, "loss": 2.7643, "step": 29890 }, { "epoch": 1.8555465888633682, "grad_norm": 0.5845209928712379, "learning_rate": 3.8155905573465834e-05, "loss": 2.7104, "step": 29891 }, { "epoch": 1.8556086659631261, "grad_norm": 0.5766896136170621, "learning_rate": 3.815239686974124e-05, "loss": 2.7014, "step": 29892 }, { "epoch": 1.855670743062884, "grad_norm": 0.5973791589602142, "learning_rate": 3.814888822782846e-05, "loss": 2.7119, "step": 29893 }, { "epoch": 1.855732820162642, "grad_norm": 0.5642769878611189, "learning_rate": 3.814537964774582e-05, "loss": 2.7494, "step": 29894 }, { "epoch": 1.8557948972623999, "grad_norm": 0.5813596806563444, "learning_rate": 3.81418711295116e-05, "loss": 2.7776, "step": 29895 }, { "epoch": 1.8558569743621578, "grad_norm": 0.5657480295747906, "learning_rate": 3.813836267314412e-05, "loss": 2.7225, "step": 29896 }, { "epoch": 1.8559190514619157, "grad_norm": 0.5608791735404349, "learning_rate": 3.8134854278661684e-05, "loss": 2.6977, "step": 29897 }, { "epoch": 1.8559811285616736, "grad_norm": 0.5094997253304921, "learning_rate": 3.813134594608259e-05, "loss": 2.6415, "step": 29898 }, { "epoch": 1.8560432056614315, "grad_norm": 0.5786625708337646, "learning_rate": 3.812783767542514e-05, "loss": 2.746, "step": 29899 }, { "epoch": 1.8561052827611895, "grad_norm": 0.5248920386433923, "learning_rate": 3.812432946670766e-05, "loss": 2.8324, "step": 29900 }, { "epoch": 1.8561673598609474, "grad_norm": 0.5677013980080219, "learning_rate": 3.812082131994843e-05, "loss": 2.8208, "step": 29901 }, { "epoch": 1.8562294369607053, "grad_norm": 0.5168536892410434, "learning_rate": 3.811731323516575e-05, "loss": 2.7083, "step": 29902 }, { "epoch": 1.8562915140604632, "grad_norm": 0.5145697633084012, "learning_rate": 3.811380521237795e-05, "loss": 2.6798, "step": 29903 }, { "epoch": 1.8563535911602211, "grad_norm": 0.5058091775739563, "learning_rate": 3.811029725160329e-05, "loss": 2.7183, "step": 29904 }, { "epoch": 1.856415668259979, "grad_norm": 0.49895862076285835, "learning_rate": 3.810678935286009e-05, "loss": 2.6162, "step": 29905 }, { "epoch": 1.8564777453597368, "grad_norm": 0.5106278067129966, "learning_rate": 3.810328151616668e-05, "loss": 2.7265, "step": 29906 }, { "epoch": 1.8565398224594947, "grad_norm": 0.5558553410033205, "learning_rate": 3.809977374154132e-05, "loss": 2.685, "step": 29907 }, { "epoch": 1.8566018995592526, "grad_norm": 0.5240472686501315, "learning_rate": 3.809626602900235e-05, "loss": 2.7338, "step": 29908 }, { "epoch": 1.8566639766590105, "grad_norm": 0.5688504067546835, "learning_rate": 3.8092758378568034e-05, "loss": 2.7622, "step": 29909 }, { "epoch": 1.8567260537587684, "grad_norm": 0.5980942810126395, "learning_rate": 3.8089250790256706e-05, "loss": 2.7265, "step": 29910 }, { "epoch": 1.8567881308585263, "grad_norm": 0.5317964548283308, "learning_rate": 3.808574326408663e-05, "loss": 2.7373, "step": 29911 }, { "epoch": 1.856850207958284, "grad_norm": 0.5513395091023828, "learning_rate": 3.808223580007613e-05, "loss": 2.681, "step": 29912 }, { "epoch": 1.856912285058042, "grad_norm": 0.5777939544556883, "learning_rate": 3.8078728398243506e-05, "loss": 2.7236, "step": 29913 }, { "epoch": 1.8569743621577999, "grad_norm": 0.5845922922883059, "learning_rate": 3.807522105860704e-05, "loss": 2.7139, "step": 29914 }, { "epoch": 1.8570364392575578, "grad_norm": 0.580256707545098, "learning_rate": 3.8071713781185045e-05, "loss": 2.8649, "step": 29915 }, { "epoch": 1.8570985163573157, "grad_norm": 0.6030159520606068, "learning_rate": 3.806820656599581e-05, "loss": 2.8058, "step": 29916 }, { "epoch": 1.8571605934570736, "grad_norm": 0.5891413769921426, "learning_rate": 3.806469941305765e-05, "loss": 2.7444, "step": 29917 }, { "epoch": 1.8572226705568315, "grad_norm": 0.5691063288695193, "learning_rate": 3.8061192322388836e-05, "loss": 2.6096, "step": 29918 }, { "epoch": 1.8572847476565895, "grad_norm": 0.6332926165099289, "learning_rate": 3.805768529400769e-05, "loss": 2.7409, "step": 29919 }, { "epoch": 1.8573468247563474, "grad_norm": 0.5763137901232968, "learning_rate": 3.805417832793249e-05, "loss": 2.6399, "step": 29920 }, { "epoch": 1.8574089018561053, "grad_norm": 0.576484155888793, "learning_rate": 3.805067142418155e-05, "loss": 2.6611, "step": 29921 }, { "epoch": 1.8574709789558632, "grad_norm": 0.5783114519248905, "learning_rate": 3.8047164582773154e-05, "loss": 2.7543, "step": 29922 }, { "epoch": 1.8575330560556211, "grad_norm": 0.5473180338714944, "learning_rate": 3.804365780372561e-05, "loss": 2.6758, "step": 29923 }, { "epoch": 1.857595133155379, "grad_norm": 0.5484185913825806, "learning_rate": 3.8040151087057194e-05, "loss": 2.7931, "step": 29924 }, { "epoch": 1.857657210255137, "grad_norm": 0.5697938834175724, "learning_rate": 3.8036644432786214e-05, "loss": 2.678, "step": 29925 }, { "epoch": 1.8577192873548949, "grad_norm": 0.5867651625154415, "learning_rate": 3.803313784093098e-05, "loss": 2.6947, "step": 29926 }, { "epoch": 1.8577813644546528, "grad_norm": 0.5979671542243806, "learning_rate": 3.8029631311509754e-05, "loss": 2.7362, "step": 29927 }, { "epoch": 1.8578434415544107, "grad_norm": 0.5502302511467709, "learning_rate": 3.802612484454086e-05, "loss": 2.7256, "step": 29928 }, { "epoch": 1.8579055186541686, "grad_norm": 0.4957500979605697, "learning_rate": 3.8022618440042554e-05, "loss": 2.756, "step": 29929 }, { "epoch": 1.8579675957539263, "grad_norm": 0.547264042483757, "learning_rate": 3.8019112098033184e-05, "loss": 2.7768, "step": 29930 }, { "epoch": 1.8580296728536843, "grad_norm": 0.5199152738981836, "learning_rate": 3.8015605818530996e-05, "loss": 2.7522, "step": 29931 }, { "epoch": 1.8580917499534422, "grad_norm": 0.5390749114445988, "learning_rate": 3.801209960155432e-05, "loss": 2.761, "step": 29932 }, { "epoch": 1.8581538270532, "grad_norm": 0.5260154100670057, "learning_rate": 3.800859344712141e-05, "loss": 2.6905, "step": 29933 }, { "epoch": 1.858215904152958, "grad_norm": 0.4854300913230031, "learning_rate": 3.8005087355250596e-05, "loss": 2.7383, "step": 29934 }, { "epoch": 1.858277981252716, "grad_norm": 0.5242813766523905, "learning_rate": 3.800158132596014e-05, "loss": 2.7101, "step": 29935 }, { "epoch": 1.8583400583524736, "grad_norm": 0.5229648334868262, "learning_rate": 3.799807535926836e-05, "loss": 2.6901, "step": 29936 }, { "epoch": 1.8584021354522315, "grad_norm": 0.5209031026521884, "learning_rate": 3.7994569455193526e-05, "loss": 2.6956, "step": 29937 }, { "epoch": 1.8584642125519895, "grad_norm": 0.5404627209139904, "learning_rate": 3.799106361375393e-05, "loss": 2.631, "step": 29938 }, { "epoch": 1.8585262896517474, "grad_norm": 0.49795639113718104, "learning_rate": 3.798755783496787e-05, "loss": 2.6558, "step": 29939 }, { "epoch": 1.8585883667515053, "grad_norm": 0.5322062917226343, "learning_rate": 3.7984052118853654e-05, "loss": 2.7638, "step": 29940 }, { "epoch": 1.8586504438512632, "grad_norm": 0.5376811088351556, "learning_rate": 3.798054646542954e-05, "loss": 2.6754, "step": 29941 }, { "epoch": 1.8587125209510211, "grad_norm": 0.5242301860079268, "learning_rate": 3.7977040874713844e-05, "loss": 2.7527, "step": 29942 }, { "epoch": 1.858774598050779, "grad_norm": 0.608059171620671, "learning_rate": 3.797353534672484e-05, "loss": 2.7025, "step": 29943 }, { "epoch": 1.858836675150537, "grad_norm": 0.5772859567667056, "learning_rate": 3.7970029881480825e-05, "loss": 2.7606, "step": 29944 }, { "epoch": 1.8588987522502949, "grad_norm": 0.6205330002775366, "learning_rate": 3.7966524479000075e-05, "loss": 2.7304, "step": 29945 }, { "epoch": 1.8589608293500528, "grad_norm": 0.5932396150717709, "learning_rate": 3.79630191393009e-05, "loss": 2.7378, "step": 29946 }, { "epoch": 1.8590229064498107, "grad_norm": 0.5802422750419822, "learning_rate": 3.7959513862401564e-05, "loss": 2.7842, "step": 29947 }, { "epoch": 1.8590849835495686, "grad_norm": 0.5888708313786513, "learning_rate": 3.795600864832038e-05, "loss": 2.7775, "step": 29948 }, { "epoch": 1.8591470606493266, "grad_norm": 0.5761106555549237, "learning_rate": 3.795250349707561e-05, "loss": 2.7519, "step": 29949 }, { "epoch": 1.8592091377490845, "grad_norm": 0.5583184370739214, "learning_rate": 3.7948998408685564e-05, "loss": 2.6704, "step": 29950 }, { "epoch": 1.8592712148488424, "grad_norm": 0.6596525855427271, "learning_rate": 3.7945493383168525e-05, "loss": 2.7264, "step": 29951 }, { "epoch": 1.8593332919486003, "grad_norm": 0.6036669593868602, "learning_rate": 3.7941988420542755e-05, "loss": 2.7342, "step": 29952 }, { "epoch": 1.8593953690483582, "grad_norm": 0.5981562150086565, "learning_rate": 3.7938483520826576e-05, "loss": 2.6671, "step": 29953 }, { "epoch": 1.859457446148116, "grad_norm": 0.5310588941726662, "learning_rate": 3.793497868403824e-05, "loss": 2.7463, "step": 29954 }, { "epoch": 1.8595195232478738, "grad_norm": 0.5977445798936885, "learning_rate": 3.7931473910196066e-05, "loss": 2.7489, "step": 29955 }, { "epoch": 1.8595816003476318, "grad_norm": 0.5619616581472985, "learning_rate": 3.79279691993183e-05, "loss": 2.6638, "step": 29956 }, { "epoch": 1.8596436774473897, "grad_norm": 0.5535511950341702, "learning_rate": 3.7924464551423274e-05, "loss": 2.7224, "step": 29957 }, { "epoch": 1.8597057545471476, "grad_norm": 0.5691922826108416, "learning_rate": 3.792095996652923e-05, "loss": 2.7573, "step": 29958 }, { "epoch": 1.8597678316469055, "grad_norm": 0.5660822500713347, "learning_rate": 3.791745544465448e-05, "loss": 2.7149, "step": 29959 }, { "epoch": 1.8598299087466632, "grad_norm": 0.5489707571381562, "learning_rate": 3.791395098581729e-05, "loss": 2.6851, "step": 29960 }, { "epoch": 1.8598919858464211, "grad_norm": 0.5527573166254722, "learning_rate": 3.791044659003596e-05, "loss": 2.7477, "step": 29961 }, { "epoch": 1.859954062946179, "grad_norm": 0.5544591621757182, "learning_rate": 3.790694225732877e-05, "loss": 2.7089, "step": 29962 }, { "epoch": 1.860016140045937, "grad_norm": 0.5248298062615562, "learning_rate": 3.790343798771399e-05, "loss": 2.7476, "step": 29963 }, { "epoch": 1.8600782171456949, "grad_norm": 0.5515229121798131, "learning_rate": 3.7899933781209905e-05, "loss": 2.6491, "step": 29964 }, { "epoch": 1.8601402942454528, "grad_norm": 0.5650804722707005, "learning_rate": 3.789642963783481e-05, "loss": 2.7762, "step": 29965 }, { "epoch": 1.8602023713452107, "grad_norm": 0.5751858307294946, "learning_rate": 3.789292555760698e-05, "loss": 2.8013, "step": 29966 }, { "epoch": 1.8602644484449686, "grad_norm": 0.5190734854310242, "learning_rate": 3.788942154054469e-05, "loss": 2.7798, "step": 29967 }, { "epoch": 1.8603265255447266, "grad_norm": 0.5849415006895109, "learning_rate": 3.788591758666623e-05, "loss": 2.7981, "step": 29968 }, { "epoch": 1.8603886026444845, "grad_norm": 0.5471359447807754, "learning_rate": 3.7882413695989873e-05, "loss": 2.7185, "step": 29969 }, { "epoch": 1.8604506797442424, "grad_norm": 0.5571578717872225, "learning_rate": 3.78789098685339e-05, "loss": 2.6946, "step": 29970 }, { "epoch": 1.8605127568440003, "grad_norm": 0.600248338049606, "learning_rate": 3.787540610431662e-05, "loss": 2.8036, "step": 29971 }, { "epoch": 1.8605748339437582, "grad_norm": 0.5560713416802202, "learning_rate": 3.787190240335627e-05, "loss": 2.6838, "step": 29972 }, { "epoch": 1.8606369110435161, "grad_norm": 0.5919440309979667, "learning_rate": 3.786839876567116e-05, "loss": 2.7425, "step": 29973 }, { "epoch": 1.860698988143274, "grad_norm": 0.5740773033264331, "learning_rate": 3.786489519127956e-05, "loss": 2.6849, "step": 29974 }, { "epoch": 1.860761065243032, "grad_norm": 0.5919892152757579, "learning_rate": 3.786139168019975e-05, "loss": 2.7894, "step": 29975 }, { "epoch": 1.86082314234279, "grad_norm": 0.621224515572589, "learning_rate": 3.785788823245e-05, "loss": 2.7, "step": 29976 }, { "epoch": 1.8608852194425476, "grad_norm": 0.6404227658636783, "learning_rate": 3.78543848480486e-05, "loss": 2.7631, "step": 29977 }, { "epoch": 1.8609472965423055, "grad_norm": 0.5715277800695832, "learning_rate": 3.785088152701381e-05, "loss": 2.7843, "step": 29978 }, { "epoch": 1.8610093736420634, "grad_norm": 0.6406777522999962, "learning_rate": 3.784737826936393e-05, "loss": 2.7732, "step": 29979 }, { "epoch": 1.8610714507418213, "grad_norm": 0.5920596656614693, "learning_rate": 3.784387507511723e-05, "loss": 2.8422, "step": 29980 }, { "epoch": 1.8611335278415793, "grad_norm": 0.6390555393054431, "learning_rate": 3.784037194429198e-05, "loss": 2.7166, "step": 29981 }, { "epoch": 1.8611956049413372, "grad_norm": 0.6126872302012641, "learning_rate": 3.783686887690647e-05, "loss": 2.7008, "step": 29982 }, { "epoch": 1.8612576820410949, "grad_norm": 0.6593118944750226, "learning_rate": 3.7833365872978956e-05, "loss": 2.8315, "step": 29983 }, { "epoch": 1.8613197591408528, "grad_norm": 0.6012458711113596, "learning_rate": 3.7829862932527734e-05, "loss": 2.7655, "step": 29984 }, { "epoch": 1.8613818362406107, "grad_norm": 0.5935524812728283, "learning_rate": 3.7826360055571065e-05, "loss": 2.7682, "step": 29985 }, { "epoch": 1.8614439133403686, "grad_norm": 0.6093043751015943, "learning_rate": 3.782285724212724e-05, "loss": 2.7262, "step": 29986 }, { "epoch": 1.8615059904401265, "grad_norm": 0.6900381356564302, "learning_rate": 3.781935449221452e-05, "loss": 2.7913, "step": 29987 }, { "epoch": 1.8615680675398845, "grad_norm": 0.574173101569973, "learning_rate": 3.781585180585118e-05, "loss": 2.6873, "step": 29988 }, { "epoch": 1.8616301446396424, "grad_norm": 0.6151756621910305, "learning_rate": 3.781234918305551e-05, "loss": 2.5831, "step": 29989 }, { "epoch": 1.8616922217394003, "grad_norm": 0.5901702078496049, "learning_rate": 3.780884662384576e-05, "loss": 2.7541, "step": 29990 }, { "epoch": 1.8617542988391582, "grad_norm": 0.5427975007193618, "learning_rate": 3.780534412824022e-05, "loss": 2.7207, "step": 29991 }, { "epoch": 1.8618163759389161, "grad_norm": 0.5784117528789766, "learning_rate": 3.780184169625716e-05, "loss": 2.6729, "step": 29992 }, { "epoch": 1.861878453038674, "grad_norm": 0.48302358870727813, "learning_rate": 3.7798339327914853e-05, "loss": 2.716, "step": 29993 }, { "epoch": 1.861940530138432, "grad_norm": 0.5701894758082166, "learning_rate": 3.779483702323158e-05, "loss": 2.851, "step": 29994 }, { "epoch": 1.8620026072381899, "grad_norm": 0.48478982451574826, "learning_rate": 3.7791334782225594e-05, "loss": 2.8032, "step": 29995 }, { "epoch": 1.8620646843379478, "grad_norm": 0.5106666757112175, "learning_rate": 3.7787832604915185e-05, "loss": 2.7164, "step": 29996 }, { "epoch": 1.8621267614377057, "grad_norm": 0.530573301728642, "learning_rate": 3.778433049131862e-05, "loss": 2.7127, "step": 29997 }, { "epoch": 1.8621888385374636, "grad_norm": 0.5512221985897283, "learning_rate": 3.778082844145417e-05, "loss": 2.794, "step": 29998 }, { "epoch": 1.8622509156372216, "grad_norm": 0.4904515670580995, "learning_rate": 3.777732645534009e-05, "loss": 2.6756, "step": 29999 }, { "epoch": 1.8623129927369795, "grad_norm": 0.6354672354498871, "learning_rate": 3.7773824532994675e-05, "loss": 2.6941, "step": 30000 }, { "epoch": 1.8623750698367372, "grad_norm": 0.49294322895422454, "learning_rate": 3.777032267443619e-05, "loss": 2.752, "step": 30001 }, { "epoch": 1.862437146936495, "grad_norm": 0.5990070212657997, "learning_rate": 3.7766820879682896e-05, "loss": 2.732, "step": 30002 }, { "epoch": 1.862499224036253, "grad_norm": 0.5015211807296467, "learning_rate": 3.776331914875304e-05, "loss": 2.7257, "step": 30003 }, { "epoch": 1.862561301136011, "grad_norm": 0.5720804324936676, "learning_rate": 3.7759817481664944e-05, "loss": 2.7458, "step": 30004 }, { "epoch": 1.8626233782357688, "grad_norm": 0.5342643154691523, "learning_rate": 3.775631587843686e-05, "loss": 2.6839, "step": 30005 }, { "epoch": 1.8626854553355268, "grad_norm": 0.556087607349534, "learning_rate": 3.7752814339087035e-05, "loss": 2.734, "step": 30006 }, { "epoch": 1.8627475324352845, "grad_norm": 0.5250906389195709, "learning_rate": 3.7749312863633764e-05, "loss": 2.7476, "step": 30007 }, { "epoch": 1.8628096095350424, "grad_norm": 0.5640624122936572, "learning_rate": 3.774581145209528e-05, "loss": 2.7209, "step": 30008 }, { "epoch": 1.8628716866348003, "grad_norm": 0.4952535949241495, "learning_rate": 3.7742310104489896e-05, "loss": 2.7552, "step": 30009 }, { "epoch": 1.8629337637345582, "grad_norm": 0.5134181277056111, "learning_rate": 3.773880882083584e-05, "loss": 2.732, "step": 30010 }, { "epoch": 1.8629958408343161, "grad_norm": 0.5447505011484013, "learning_rate": 3.77353076011514e-05, "loss": 2.7526, "step": 30011 }, { "epoch": 1.863057917934074, "grad_norm": 0.45417368326395746, "learning_rate": 3.773180644545485e-05, "loss": 2.7123, "step": 30012 }, { "epoch": 1.863119995033832, "grad_norm": 0.5440050842208853, "learning_rate": 3.7728305353764425e-05, "loss": 2.765, "step": 30013 }, { "epoch": 1.8631820721335899, "grad_norm": 0.5726448338972286, "learning_rate": 3.772480432609843e-05, "loss": 2.8484, "step": 30014 }, { "epoch": 1.8632441492333478, "grad_norm": 0.5245044140261803, "learning_rate": 3.7721303362475094e-05, "loss": 2.7592, "step": 30015 }, { "epoch": 1.8633062263331057, "grad_norm": 0.5331029273827014, "learning_rate": 3.7717802462912707e-05, "loss": 2.7938, "step": 30016 }, { "epoch": 1.8633683034328636, "grad_norm": 0.5238682696657194, "learning_rate": 3.7714301627429513e-05, "loss": 2.8364, "step": 30017 }, { "epoch": 1.8634303805326216, "grad_norm": 0.4968060313648413, "learning_rate": 3.7710800856043804e-05, "loss": 2.7292, "step": 30018 }, { "epoch": 1.8634924576323795, "grad_norm": 0.5115461859318428, "learning_rate": 3.770730014877382e-05, "loss": 2.7368, "step": 30019 }, { "epoch": 1.8635545347321374, "grad_norm": 0.517620502392506, "learning_rate": 3.770379950563785e-05, "loss": 2.795, "step": 30020 }, { "epoch": 1.8636166118318953, "grad_norm": 0.5478602661964596, "learning_rate": 3.770029892665413e-05, "loss": 2.7449, "step": 30021 }, { "epoch": 1.8636786889316532, "grad_norm": 0.4962063759143128, "learning_rate": 3.769679841184094e-05, "loss": 2.7846, "step": 30022 }, { "epoch": 1.8637407660314111, "grad_norm": 0.5258963040986715, "learning_rate": 3.7693297961216545e-05, "loss": 2.718, "step": 30023 }, { "epoch": 1.863802843131169, "grad_norm": 0.5675612151113852, "learning_rate": 3.768979757479919e-05, "loss": 2.8099, "step": 30024 }, { "epoch": 1.8638649202309268, "grad_norm": 0.489914869776363, "learning_rate": 3.7686297252607163e-05, "loss": 2.7425, "step": 30025 }, { "epoch": 1.8639269973306847, "grad_norm": 0.548886659154769, "learning_rate": 3.7682796994658695e-05, "loss": 2.713, "step": 30026 }, { "epoch": 1.8639890744304426, "grad_norm": 0.5107130986932532, "learning_rate": 3.767929680097209e-05, "loss": 2.7148, "step": 30027 }, { "epoch": 1.8640511515302005, "grad_norm": 0.5003538072011406, "learning_rate": 3.767579667156556e-05, "loss": 2.8172, "step": 30028 }, { "epoch": 1.8641132286299584, "grad_norm": 0.4892882682168793, "learning_rate": 3.767229660645741e-05, "loss": 2.5951, "step": 30029 }, { "epoch": 1.8641753057297163, "grad_norm": 0.564383320963574, "learning_rate": 3.766879660566587e-05, "loss": 2.7981, "step": 30030 }, { "epoch": 1.864237382829474, "grad_norm": 0.5507564674375194, "learning_rate": 3.766529666920922e-05, "loss": 2.7575, "step": 30031 }, { "epoch": 1.864299459929232, "grad_norm": 0.5912634078636707, "learning_rate": 3.7661796797105696e-05, "loss": 2.7022, "step": 30032 }, { "epoch": 1.8643615370289899, "grad_norm": 0.542506810466366, "learning_rate": 3.765829698937359e-05, "loss": 2.6781, "step": 30033 }, { "epoch": 1.8644236141287478, "grad_norm": 0.5892825203970026, "learning_rate": 3.7654797246031124e-05, "loss": 2.7447, "step": 30034 }, { "epoch": 1.8644856912285057, "grad_norm": 0.5816172465076714, "learning_rate": 3.7651297567096596e-05, "loss": 2.6884, "step": 30035 }, { "epoch": 1.8645477683282636, "grad_norm": 0.5527780737974608, "learning_rate": 3.764779795258823e-05, "loss": 2.8027, "step": 30036 }, { "epoch": 1.8646098454280216, "grad_norm": 0.5933741333502931, "learning_rate": 3.7644298402524316e-05, "loss": 2.7381, "step": 30037 }, { "epoch": 1.8646719225277795, "grad_norm": 0.5313195575729558, "learning_rate": 3.764079891692309e-05, "loss": 2.7965, "step": 30038 }, { "epoch": 1.8647339996275374, "grad_norm": 0.5736323625811539, "learning_rate": 3.7637299495802825e-05, "loss": 2.7954, "step": 30039 }, { "epoch": 1.8647960767272953, "grad_norm": 0.525887605246282, "learning_rate": 3.7633800139181756e-05, "loss": 2.8114, "step": 30040 }, { "epoch": 1.8648581538270532, "grad_norm": 0.5294166795340393, "learning_rate": 3.763030084707817e-05, "loss": 2.6276, "step": 30041 }, { "epoch": 1.8649202309268111, "grad_norm": 0.52916265383523, "learning_rate": 3.762680161951029e-05, "loss": 2.7117, "step": 30042 }, { "epoch": 1.864982308026569, "grad_norm": 0.555096623154768, "learning_rate": 3.76233024564964e-05, "loss": 2.7826, "step": 30043 }, { "epoch": 1.865044385126327, "grad_norm": 0.5195670151410499, "learning_rate": 3.761980335805474e-05, "loss": 2.6885, "step": 30044 }, { "epoch": 1.865106462226085, "grad_norm": 0.5496276516237685, "learning_rate": 3.761630432420358e-05, "loss": 2.7739, "step": 30045 }, { "epoch": 1.8651685393258428, "grad_norm": 0.5660924897481276, "learning_rate": 3.7612805354961165e-05, "loss": 2.7264, "step": 30046 }, { "epoch": 1.8652306164256007, "grad_norm": 0.4900403037574346, "learning_rate": 3.760930645034575e-05, "loss": 2.6306, "step": 30047 }, { "epoch": 1.8652926935253586, "grad_norm": 0.5951328236723251, "learning_rate": 3.760580761037559e-05, "loss": 2.6618, "step": 30048 }, { "epoch": 1.8653547706251163, "grad_norm": 0.5488790848175373, "learning_rate": 3.760230883506894e-05, "loss": 2.6934, "step": 30049 }, { "epoch": 1.8654168477248743, "grad_norm": 0.5767949243187293, "learning_rate": 3.759881012444406e-05, "loss": 2.709, "step": 30050 }, { "epoch": 1.8654789248246322, "grad_norm": 0.4996207706443243, "learning_rate": 3.759531147851919e-05, "loss": 2.7268, "step": 30051 }, { "epoch": 1.86554100192439, "grad_norm": 0.58825350489547, "learning_rate": 3.7591812897312594e-05, "loss": 2.7739, "step": 30052 }, { "epoch": 1.865603079024148, "grad_norm": 0.5499771145914397, "learning_rate": 3.758831438084251e-05, "loss": 2.7509, "step": 30053 }, { "epoch": 1.865665156123906, "grad_norm": 0.5005575290318316, "learning_rate": 3.7584815929127225e-05, "loss": 2.7906, "step": 30054 }, { "epoch": 1.8657272332236636, "grad_norm": 0.5793111849539557, "learning_rate": 3.7581317542184955e-05, "loss": 2.7491, "step": 30055 }, { "epoch": 1.8657893103234215, "grad_norm": 0.5142420516041954, "learning_rate": 3.7577819220033976e-05, "loss": 2.6685, "step": 30056 }, { "epoch": 1.8658513874231795, "grad_norm": 0.6236925755366999, "learning_rate": 3.757432096269252e-05, "loss": 2.7011, "step": 30057 }, { "epoch": 1.8659134645229374, "grad_norm": 0.5293334723063238, "learning_rate": 3.757082277017886e-05, "loss": 2.7485, "step": 30058 }, { "epoch": 1.8659755416226953, "grad_norm": 0.5589024950967251, "learning_rate": 3.756732464251122e-05, "loss": 2.7449, "step": 30059 }, { "epoch": 1.8660376187224532, "grad_norm": 0.5283318199272422, "learning_rate": 3.756382657970788e-05, "loss": 2.6651, "step": 30060 }, { "epoch": 1.8660996958222111, "grad_norm": 0.5862432826375129, "learning_rate": 3.756032858178708e-05, "loss": 2.8116, "step": 30061 }, { "epoch": 1.866161772921969, "grad_norm": 0.5812755261220344, "learning_rate": 3.7556830648767045e-05, "loss": 2.7322, "step": 30062 }, { "epoch": 1.866223850021727, "grad_norm": 0.4989787341136915, "learning_rate": 3.755333278066606e-05, "loss": 2.6839, "step": 30063 }, { "epoch": 1.8662859271214849, "grad_norm": 0.5467408969489388, "learning_rate": 3.7549834977502344e-05, "loss": 2.7101, "step": 30064 }, { "epoch": 1.8663480042212428, "grad_norm": 0.4756717017074804, "learning_rate": 3.754633723929418e-05, "loss": 2.7995, "step": 30065 }, { "epoch": 1.8664100813210007, "grad_norm": 0.5377225947222386, "learning_rate": 3.754283956605978e-05, "loss": 2.7075, "step": 30066 }, { "epoch": 1.8664721584207586, "grad_norm": 0.5037961736110502, "learning_rate": 3.7539341957817424e-05, "loss": 2.7472, "step": 30067 }, { "epoch": 1.8665342355205166, "grad_norm": 0.5989524149947851, "learning_rate": 3.753584441458534e-05, "loss": 2.8017, "step": 30068 }, { "epoch": 1.8665963126202745, "grad_norm": 0.5321862161725026, "learning_rate": 3.753234693638176e-05, "loss": 2.6899, "step": 30069 }, { "epoch": 1.8666583897200324, "grad_norm": 0.5546597218609813, "learning_rate": 3.7528849523224984e-05, "loss": 2.7441, "step": 30070 }, { "epoch": 1.8667204668197903, "grad_norm": 0.58339393391769, "learning_rate": 3.752535217513321e-05, "loss": 2.6532, "step": 30071 }, { "epoch": 1.8667825439195482, "grad_norm": 0.556453633271304, "learning_rate": 3.752185489212472e-05, "loss": 2.6155, "step": 30072 }, { "epoch": 1.866844621019306, "grad_norm": 0.5730006214241666, "learning_rate": 3.7518357674217733e-05, "loss": 2.7772, "step": 30073 }, { "epoch": 1.8669066981190638, "grad_norm": 0.5716112842044615, "learning_rate": 3.75148605214305e-05, "loss": 2.7955, "step": 30074 }, { "epoch": 1.8669687752188218, "grad_norm": 0.5717029604591104, "learning_rate": 3.751136343378128e-05, "loss": 2.6119, "step": 30075 }, { "epoch": 1.8670308523185797, "grad_norm": 0.5687716060768387, "learning_rate": 3.7507866411288293e-05, "loss": 2.6834, "step": 30076 }, { "epoch": 1.8670929294183376, "grad_norm": 0.5690730749296693, "learning_rate": 3.7504369453969824e-05, "loss": 2.6724, "step": 30077 }, { "epoch": 1.8671550065180955, "grad_norm": 0.5718762371696056, "learning_rate": 3.750087256184407e-05, "loss": 2.6571, "step": 30078 }, { "epoch": 1.8672170836178532, "grad_norm": 0.5679510950757832, "learning_rate": 3.7497375734929305e-05, "loss": 2.7038, "step": 30079 }, { "epoch": 1.8672791607176111, "grad_norm": 0.5283221740125765, "learning_rate": 3.749387897324377e-05, "loss": 2.7087, "step": 30080 }, { "epoch": 1.867341237817369, "grad_norm": 0.591487640061472, "learning_rate": 3.7490382276805705e-05, "loss": 2.7146, "step": 30081 }, { "epoch": 1.867403314917127, "grad_norm": 0.5240732289480193, "learning_rate": 3.748688564563334e-05, "loss": 2.8003, "step": 30082 }, { "epoch": 1.8674653920168849, "grad_norm": 0.5432035363924226, "learning_rate": 3.7483389079744944e-05, "loss": 2.7252, "step": 30083 }, { "epoch": 1.8675274691166428, "grad_norm": 0.5797486830758505, "learning_rate": 3.7479892579158735e-05, "loss": 2.7109, "step": 30084 }, { "epoch": 1.8675895462164007, "grad_norm": 0.5173648034745576, "learning_rate": 3.747639614389298e-05, "loss": 2.7141, "step": 30085 }, { "epoch": 1.8676516233161586, "grad_norm": 0.5552985626772966, "learning_rate": 3.747289977396591e-05, "loss": 2.7111, "step": 30086 }, { "epoch": 1.8677137004159166, "grad_norm": 0.5574182800305358, "learning_rate": 3.746940346939574e-05, "loss": 2.8121, "step": 30087 }, { "epoch": 1.8677757775156745, "grad_norm": 0.5603460556935075, "learning_rate": 3.7465907230200755e-05, "loss": 2.6856, "step": 30088 }, { "epoch": 1.8678378546154324, "grad_norm": 0.5616033082632289, "learning_rate": 3.746241105639916e-05, "loss": 2.7328, "step": 30089 }, { "epoch": 1.8678999317151903, "grad_norm": 0.5513932635755406, "learning_rate": 3.7458914948009215e-05, "loss": 2.6883, "step": 30090 }, { "epoch": 1.8679620088149482, "grad_norm": 0.5451362215639639, "learning_rate": 3.745541890504916e-05, "loss": 2.6936, "step": 30091 }, { "epoch": 1.8680240859147061, "grad_norm": 0.5569171954608199, "learning_rate": 3.745192292753723e-05, "loss": 2.6538, "step": 30092 }, { "epoch": 1.868086163014464, "grad_norm": 0.5535780866723531, "learning_rate": 3.744842701549165e-05, "loss": 2.7252, "step": 30093 }, { "epoch": 1.868148240114222, "grad_norm": 0.6109191208828583, "learning_rate": 3.744493116893069e-05, "loss": 2.7328, "step": 30094 }, { "epoch": 1.86821031721398, "grad_norm": 0.6031705740278301, "learning_rate": 3.744143538787256e-05, "loss": 2.8085, "step": 30095 }, { "epoch": 1.8682723943137378, "grad_norm": 0.5986454574966998, "learning_rate": 3.743793967233552e-05, "loss": 2.7916, "step": 30096 }, { "epoch": 1.8683344714134955, "grad_norm": 0.5872358052765859, "learning_rate": 3.743444402233779e-05, "loss": 2.7884, "step": 30097 }, { "epoch": 1.8683965485132534, "grad_norm": 0.47344490335045536, "learning_rate": 3.7430948437897615e-05, "loss": 2.7516, "step": 30098 }, { "epoch": 1.8684586256130113, "grad_norm": 0.6319359489424147, "learning_rate": 3.7427452919033244e-05, "loss": 2.7299, "step": 30099 }, { "epoch": 1.8685207027127693, "grad_norm": 0.5239082041953411, "learning_rate": 3.7423957465762894e-05, "loss": 2.703, "step": 30100 }, { "epoch": 1.8685827798125272, "grad_norm": 0.5501834249244794, "learning_rate": 3.74204620781048e-05, "loss": 2.8233, "step": 30101 }, { "epoch": 1.868644856912285, "grad_norm": 0.5571763522433782, "learning_rate": 3.7416966756077226e-05, "loss": 2.7831, "step": 30102 }, { "epoch": 1.8687069340120428, "grad_norm": 0.5225787163718401, "learning_rate": 3.7413471499698384e-05, "loss": 2.7052, "step": 30103 }, { "epoch": 1.8687690111118007, "grad_norm": 0.5325930112549249, "learning_rate": 3.740997630898653e-05, "loss": 2.6697, "step": 30104 }, { "epoch": 1.8688310882115586, "grad_norm": 0.615530784291077, "learning_rate": 3.7406481183959865e-05, "loss": 2.7379, "step": 30105 }, { "epoch": 1.8688931653113166, "grad_norm": 0.4939210728150102, "learning_rate": 3.7402986124636666e-05, "loss": 2.7008, "step": 30106 }, { "epoch": 1.8689552424110745, "grad_norm": 0.5624306159142358, "learning_rate": 3.7399491131035125e-05, "loss": 2.7592, "step": 30107 }, { "epoch": 1.8690173195108324, "grad_norm": 0.5448667983794083, "learning_rate": 3.739599620317352e-05, "loss": 2.7318, "step": 30108 }, { "epoch": 1.8690793966105903, "grad_norm": 0.5133988848030274, "learning_rate": 3.7392501341070055e-05, "loss": 2.6471, "step": 30109 }, { "epoch": 1.8691414737103482, "grad_norm": 0.5458660081979321, "learning_rate": 3.738900654474297e-05, "loss": 2.689, "step": 30110 }, { "epoch": 1.8692035508101061, "grad_norm": 0.5220284356341388, "learning_rate": 3.7385511814210495e-05, "loss": 2.7597, "step": 30111 }, { "epoch": 1.869265627909864, "grad_norm": 0.45945682964573015, "learning_rate": 3.738201714949087e-05, "loss": 2.736, "step": 30112 }, { "epoch": 1.869327705009622, "grad_norm": 0.5537677446454329, "learning_rate": 3.7378522550602324e-05, "loss": 2.6928, "step": 30113 }, { "epoch": 1.86938978210938, "grad_norm": 0.5524216161335364, "learning_rate": 3.737502801756308e-05, "loss": 2.7362, "step": 30114 }, { "epoch": 1.8694518592091378, "grad_norm": 0.5350271472995302, "learning_rate": 3.73715335503914e-05, "loss": 2.6868, "step": 30115 }, { "epoch": 1.8695139363088957, "grad_norm": 0.587940099921915, "learning_rate": 3.736803914910548e-05, "loss": 2.6649, "step": 30116 }, { "epoch": 1.8695760134086536, "grad_norm": 0.5303822892169349, "learning_rate": 3.736454481372357e-05, "loss": 2.7495, "step": 30117 }, { "epoch": 1.8696380905084116, "grad_norm": 0.5450644357894258, "learning_rate": 3.73610505442639e-05, "loss": 2.7047, "step": 30118 }, { "epoch": 1.8697001676081695, "grad_norm": 0.5186969109451508, "learning_rate": 3.73575563407447e-05, "loss": 2.692, "step": 30119 }, { "epoch": 1.8697622447079274, "grad_norm": 0.5091769740387578, "learning_rate": 3.735406220318418e-05, "loss": 2.7261, "step": 30120 }, { "epoch": 1.869824321807685, "grad_norm": 0.5581638842938886, "learning_rate": 3.735056813160061e-05, "loss": 2.7689, "step": 30121 }, { "epoch": 1.869886398907443, "grad_norm": 0.5579665871447131, "learning_rate": 3.734707412601219e-05, "loss": 2.735, "step": 30122 }, { "epoch": 1.869948476007201, "grad_norm": 0.5796442689224522, "learning_rate": 3.7343580186437154e-05, "loss": 2.7353, "step": 30123 }, { "epoch": 1.8700105531069589, "grad_norm": 0.5800396171144316, "learning_rate": 3.7340086312893734e-05, "loss": 2.7248, "step": 30124 }, { "epoch": 1.8700726302067168, "grad_norm": 0.6096822013214128, "learning_rate": 3.733659250540015e-05, "loss": 2.8346, "step": 30125 }, { "epoch": 1.8701347073064747, "grad_norm": 0.5319441019571174, "learning_rate": 3.733309876397465e-05, "loss": 2.7173, "step": 30126 }, { "epoch": 1.8701967844062324, "grad_norm": 0.5793969621580004, "learning_rate": 3.7329605088635435e-05, "loss": 2.6341, "step": 30127 }, { "epoch": 1.8702588615059903, "grad_norm": 0.5106124275462036, "learning_rate": 3.732611147940076e-05, "loss": 2.748, "step": 30128 }, { "epoch": 1.8703209386057482, "grad_norm": 0.5551947302759218, "learning_rate": 3.7322617936288814e-05, "loss": 2.7617, "step": 30129 }, { "epoch": 1.8703830157055061, "grad_norm": 0.5404679869717817, "learning_rate": 3.731912445931787e-05, "loss": 2.7936, "step": 30130 }, { "epoch": 1.870445092805264, "grad_norm": 0.5414963937553019, "learning_rate": 3.7315631048506125e-05, "loss": 2.719, "step": 30131 }, { "epoch": 1.870507169905022, "grad_norm": 0.6126217302687742, "learning_rate": 3.731213770387182e-05, "loss": 2.6748, "step": 30132 }, { "epoch": 1.87056924700478, "grad_norm": 0.5697212812710526, "learning_rate": 3.730864442543317e-05, "loss": 2.7914, "step": 30133 }, { "epoch": 1.8706313241045378, "grad_norm": 0.6634683531578106, "learning_rate": 3.730515121320838e-05, "loss": 2.747, "step": 30134 }, { "epoch": 1.8706934012042957, "grad_norm": 0.5714968572415023, "learning_rate": 3.730165806721571e-05, "loss": 2.7719, "step": 30135 }, { "epoch": 1.8707554783040536, "grad_norm": 0.6876032925084782, "learning_rate": 3.7298164987473385e-05, "loss": 2.6441, "step": 30136 }, { "epoch": 1.8708175554038116, "grad_norm": 0.58185797550688, "learning_rate": 3.729467197399961e-05, "loss": 2.6481, "step": 30137 }, { "epoch": 1.8708796325035695, "grad_norm": 0.6965584681836209, "learning_rate": 3.729117902681262e-05, "loss": 2.7383, "step": 30138 }, { "epoch": 1.8709417096033274, "grad_norm": 0.6411005209170796, "learning_rate": 3.7287686145930625e-05, "loss": 2.7623, "step": 30139 }, { "epoch": 1.8710037867030853, "grad_norm": 0.6171303976177183, "learning_rate": 3.7284193331371864e-05, "loss": 2.8207, "step": 30140 }, { "epoch": 1.8710658638028432, "grad_norm": 0.6307757531834417, "learning_rate": 3.728070058315455e-05, "loss": 2.6899, "step": 30141 }, { "epoch": 1.8711279409026011, "grad_norm": 0.5884527516224466, "learning_rate": 3.727720790129691e-05, "loss": 2.7348, "step": 30142 }, { "epoch": 1.871190018002359, "grad_norm": 0.5939891985880904, "learning_rate": 3.727371528581717e-05, "loss": 2.6741, "step": 30143 }, { "epoch": 1.871252095102117, "grad_norm": 0.5617697266857136, "learning_rate": 3.727022273673354e-05, "loss": 2.767, "step": 30144 }, { "epoch": 1.8713141722018747, "grad_norm": 0.5254470624710372, "learning_rate": 3.726673025406425e-05, "loss": 2.676, "step": 30145 }, { "epoch": 1.8713762493016326, "grad_norm": 0.5484122061262795, "learning_rate": 3.726323783782752e-05, "loss": 2.679, "step": 30146 }, { "epoch": 1.8714383264013905, "grad_norm": 0.523973219289083, "learning_rate": 3.7259745488041575e-05, "loss": 2.7123, "step": 30147 }, { "epoch": 1.8715004035011484, "grad_norm": 0.5169882237736759, "learning_rate": 3.725625320472462e-05, "loss": 2.6825, "step": 30148 }, { "epoch": 1.8715624806009064, "grad_norm": 0.5273481015626302, "learning_rate": 3.72527609878949e-05, "loss": 2.7331, "step": 30149 }, { "epoch": 1.8716245577006643, "grad_norm": 0.5338330303442972, "learning_rate": 3.724926883757061e-05, "loss": 2.739, "step": 30150 }, { "epoch": 1.871686634800422, "grad_norm": 0.5535989476540519, "learning_rate": 3.724577675376999e-05, "loss": 2.7595, "step": 30151 }, { "epoch": 1.8717487119001799, "grad_norm": 0.5606589919622238, "learning_rate": 3.724228473651123e-05, "loss": 2.7026, "step": 30152 }, { "epoch": 1.8718107889999378, "grad_norm": 0.5521314857884426, "learning_rate": 3.723879278581258e-05, "loss": 2.7787, "step": 30153 }, { "epoch": 1.8718728660996957, "grad_norm": 0.5514111535450713, "learning_rate": 3.723530090169224e-05, "loss": 2.7389, "step": 30154 }, { "epoch": 1.8719349431994536, "grad_norm": 0.5265195834728235, "learning_rate": 3.723180908416845e-05, "loss": 2.6656, "step": 30155 }, { "epoch": 1.8719970202992116, "grad_norm": 0.5266890523726266, "learning_rate": 3.722831733325939e-05, "loss": 2.7054, "step": 30156 }, { "epoch": 1.8720590973989695, "grad_norm": 0.5430756333350472, "learning_rate": 3.7224825648983313e-05, "loss": 2.7012, "step": 30157 }, { "epoch": 1.8721211744987274, "grad_norm": 0.593735217864086, "learning_rate": 3.722133403135842e-05, "loss": 2.7586, "step": 30158 }, { "epoch": 1.8721832515984853, "grad_norm": 0.5878885972504482, "learning_rate": 3.7217842480402925e-05, "loss": 2.7942, "step": 30159 }, { "epoch": 1.8722453286982432, "grad_norm": 0.5454394709154916, "learning_rate": 3.7214350996135056e-05, "loss": 2.7423, "step": 30160 }, { "epoch": 1.8723074057980011, "grad_norm": 0.5785699369396177, "learning_rate": 3.721085957857301e-05, "loss": 2.6882, "step": 30161 }, { "epoch": 1.872369482897759, "grad_norm": 0.5121660097930864, "learning_rate": 3.720736822773503e-05, "loss": 2.723, "step": 30162 }, { "epoch": 1.872431559997517, "grad_norm": 0.471841488210319, "learning_rate": 3.72038769436393e-05, "loss": 2.6535, "step": 30163 }, { "epoch": 1.872493637097275, "grad_norm": 0.5813358940730143, "learning_rate": 3.720038572630406e-05, "loss": 2.6138, "step": 30164 }, { "epoch": 1.8725557141970328, "grad_norm": 0.5653175677678082, "learning_rate": 3.719689457574751e-05, "loss": 2.8066, "step": 30165 }, { "epoch": 1.8726177912967907, "grad_norm": 0.5894922636574395, "learning_rate": 3.719340349198787e-05, "loss": 2.7107, "step": 30166 }, { "epoch": 1.8726798683965487, "grad_norm": 0.5671852805354303, "learning_rate": 3.7189912475043345e-05, "loss": 2.7087, "step": 30167 }, { "epoch": 1.8727419454963066, "grad_norm": 0.5728607208216999, "learning_rate": 3.718642152493215e-05, "loss": 2.707, "step": 30168 }, { "epoch": 1.8728040225960643, "grad_norm": 0.5282925300555988, "learning_rate": 3.718293064167253e-05, "loss": 2.8178, "step": 30169 }, { "epoch": 1.8728660996958222, "grad_norm": 0.5464355883250197, "learning_rate": 3.717943982528265e-05, "loss": 2.6712, "step": 30170 }, { "epoch": 1.87292817679558, "grad_norm": 0.5337858823199265, "learning_rate": 3.717594907578076e-05, "loss": 2.6514, "step": 30171 }, { "epoch": 1.872990253895338, "grad_norm": 0.5221545595637919, "learning_rate": 3.717245839318506e-05, "loss": 2.6628, "step": 30172 }, { "epoch": 1.873052330995096, "grad_norm": 0.5822702723504084, "learning_rate": 3.7168967777513754e-05, "loss": 2.7256, "step": 30173 }, { "epoch": 1.8731144080948539, "grad_norm": 0.5482642647348214, "learning_rate": 3.7165477228785055e-05, "loss": 2.7422, "step": 30174 }, { "epoch": 1.8731764851946116, "grad_norm": 0.5475108129051954, "learning_rate": 3.7161986747017177e-05, "loss": 2.7122, "step": 30175 }, { "epoch": 1.8732385622943695, "grad_norm": 0.5759242673400041, "learning_rate": 3.7158496332228334e-05, "loss": 2.7384, "step": 30176 }, { "epoch": 1.8733006393941274, "grad_norm": 0.543573331084806, "learning_rate": 3.715500598443673e-05, "loss": 2.7522, "step": 30177 }, { "epoch": 1.8733627164938853, "grad_norm": 0.5558713717315414, "learning_rate": 3.715151570366058e-05, "loss": 2.7573, "step": 30178 }, { "epoch": 1.8734247935936432, "grad_norm": 0.4848903099125168, "learning_rate": 3.714802548991808e-05, "loss": 2.7352, "step": 30179 }, { "epoch": 1.8734868706934011, "grad_norm": 0.6160620556388863, "learning_rate": 3.714453534322747e-05, "loss": 2.7793, "step": 30180 }, { "epoch": 1.873548947793159, "grad_norm": 0.5553671742638531, "learning_rate": 3.7141045263606925e-05, "loss": 2.6987, "step": 30181 }, { "epoch": 1.873611024892917, "grad_norm": 0.5701276662555518, "learning_rate": 3.713755525107468e-05, "loss": 2.8404, "step": 30182 }, { "epoch": 1.873673101992675, "grad_norm": 0.5224761390257086, "learning_rate": 3.713406530564893e-05, "loss": 2.7125, "step": 30183 }, { "epoch": 1.8737351790924328, "grad_norm": 0.5594627183726377, "learning_rate": 3.713057542734787e-05, "loss": 2.7802, "step": 30184 }, { "epoch": 1.8737972561921907, "grad_norm": 0.5674312131003816, "learning_rate": 3.712708561618974e-05, "loss": 2.7402, "step": 30185 }, { "epoch": 1.8738593332919486, "grad_norm": 0.5695788223638643, "learning_rate": 3.712359587219272e-05, "loss": 2.7074, "step": 30186 }, { "epoch": 1.8739214103917066, "grad_norm": 0.530573020878379, "learning_rate": 3.7120106195375034e-05, "loss": 2.7252, "step": 30187 }, { "epoch": 1.8739834874914645, "grad_norm": 0.5198332116973937, "learning_rate": 3.7116616585754874e-05, "loss": 2.6915, "step": 30188 }, { "epoch": 1.8740455645912224, "grad_norm": 0.5486557442044132, "learning_rate": 3.711312704335045e-05, "loss": 2.7386, "step": 30189 }, { "epoch": 1.8741076416909803, "grad_norm": 0.5316969169732507, "learning_rate": 3.710963756817997e-05, "loss": 2.7841, "step": 30190 }, { "epoch": 1.8741697187907382, "grad_norm": 0.5389309045245824, "learning_rate": 3.7106148160261654e-05, "loss": 2.7328, "step": 30191 }, { "epoch": 1.8742317958904962, "grad_norm": 0.5342033980719584, "learning_rate": 3.7102658819613676e-05, "loss": 2.7653, "step": 30192 }, { "epoch": 1.8742938729902539, "grad_norm": 0.5835149913584703, "learning_rate": 3.709916954625427e-05, "loss": 2.7876, "step": 30193 }, { "epoch": 1.8743559500900118, "grad_norm": 0.5293463933662396, "learning_rate": 3.709568034020163e-05, "loss": 2.772, "step": 30194 }, { "epoch": 1.8744180271897697, "grad_norm": 0.5392619883407733, "learning_rate": 3.709219120147395e-05, "loss": 2.6996, "step": 30195 }, { "epoch": 1.8744801042895276, "grad_norm": 0.5552502316333162, "learning_rate": 3.708870213008945e-05, "loss": 2.7572, "step": 30196 }, { "epoch": 1.8745421813892855, "grad_norm": 0.4839868528535113, "learning_rate": 3.708521312606631e-05, "loss": 2.7312, "step": 30197 }, { "epoch": 1.8746042584890434, "grad_norm": 0.5699766816422321, "learning_rate": 3.7081724189422764e-05, "loss": 2.8285, "step": 30198 }, { "epoch": 1.8746663355888011, "grad_norm": 0.5235711966051083, "learning_rate": 3.7078235320176986e-05, "loss": 2.8101, "step": 30199 }, { "epoch": 1.874728412688559, "grad_norm": 0.5781403874591927, "learning_rate": 3.7074746518347184e-05, "loss": 2.746, "step": 30200 }, { "epoch": 1.874790489788317, "grad_norm": 0.48831304828100847, "learning_rate": 3.7071257783951586e-05, "loss": 2.7179, "step": 30201 }, { "epoch": 1.874852566888075, "grad_norm": 0.5791309471459009, "learning_rate": 3.706776911700836e-05, "loss": 2.7312, "step": 30202 }, { "epoch": 1.8749146439878328, "grad_norm": 0.5529340511407203, "learning_rate": 3.706428051753574e-05, "loss": 2.8302, "step": 30203 }, { "epoch": 1.8749767210875907, "grad_norm": 0.5892015712907431, "learning_rate": 3.706079198555189e-05, "loss": 2.7257, "step": 30204 }, { "epoch": 1.8750387981873486, "grad_norm": 0.5511791063857626, "learning_rate": 3.705730352107505e-05, "loss": 2.7353, "step": 30205 }, { "epoch": 1.8751008752871066, "grad_norm": 0.6136665348862329, "learning_rate": 3.705381512412338e-05, "loss": 2.7574, "step": 30206 }, { "epoch": 1.8751629523868645, "grad_norm": 0.5401053913903275, "learning_rate": 3.7050326794715115e-05, "loss": 2.7242, "step": 30207 }, { "epoch": 1.8752250294866224, "grad_norm": 0.5603911286483026, "learning_rate": 3.704683853286844e-05, "loss": 2.7105, "step": 30208 }, { "epoch": 1.8752871065863803, "grad_norm": 0.5574009303947844, "learning_rate": 3.704335033860154e-05, "loss": 2.723, "step": 30209 }, { "epoch": 1.8753491836861382, "grad_norm": 0.5486466185698917, "learning_rate": 3.703986221193264e-05, "loss": 2.7757, "step": 30210 }, { "epoch": 1.8754112607858962, "grad_norm": 0.6462217644623418, "learning_rate": 3.7036374152879905e-05, "loss": 2.8274, "step": 30211 }, { "epoch": 1.875473337885654, "grad_norm": 0.5325709637292921, "learning_rate": 3.7032886161461574e-05, "loss": 2.8016, "step": 30212 }, { "epoch": 1.875535414985412, "grad_norm": 0.5161529208202545, "learning_rate": 3.7029398237695805e-05, "loss": 2.7039, "step": 30213 }, { "epoch": 1.87559749208517, "grad_norm": 0.6065172039180167, "learning_rate": 3.702591038160083e-05, "loss": 2.7061, "step": 30214 }, { "epoch": 1.8756595691849278, "grad_norm": 0.5145903234225837, "learning_rate": 3.702242259319482e-05, "loss": 2.7389, "step": 30215 }, { "epoch": 1.8757216462846857, "grad_norm": 0.6146119859988374, "learning_rate": 3.701893487249599e-05, "loss": 2.7371, "step": 30216 }, { "epoch": 1.8757837233844434, "grad_norm": 0.5625327948441015, "learning_rate": 3.7015447219522513e-05, "loss": 2.7458, "step": 30217 }, { "epoch": 1.8758458004842014, "grad_norm": 0.5255123975799462, "learning_rate": 3.701195963429262e-05, "loss": 2.7374, "step": 30218 }, { "epoch": 1.8759078775839593, "grad_norm": 0.6042261231423626, "learning_rate": 3.7008472116824485e-05, "loss": 2.7286, "step": 30219 }, { "epoch": 1.8759699546837172, "grad_norm": 0.5054581866568238, "learning_rate": 3.700498466713629e-05, "loss": 2.7279, "step": 30220 }, { "epoch": 1.876032031783475, "grad_norm": 0.5596005363357488, "learning_rate": 3.700149728524625e-05, "loss": 2.7692, "step": 30221 }, { "epoch": 1.876094108883233, "grad_norm": 0.5666471823914188, "learning_rate": 3.6998009971172556e-05, "loss": 2.6633, "step": 30222 }, { "epoch": 1.8761561859829907, "grad_norm": 0.5300861119095203, "learning_rate": 3.6994522724933405e-05, "loss": 2.768, "step": 30223 }, { "epoch": 1.8762182630827486, "grad_norm": 0.5524467825396179, "learning_rate": 3.6991035546546974e-05, "loss": 2.7105, "step": 30224 }, { "epoch": 1.8762803401825066, "grad_norm": 0.5411542033287697, "learning_rate": 3.698754843603148e-05, "loss": 2.798, "step": 30225 }, { "epoch": 1.8763424172822645, "grad_norm": 0.5204796702952269, "learning_rate": 3.69840613934051e-05, "loss": 2.7907, "step": 30226 }, { "epoch": 1.8764044943820224, "grad_norm": 0.5433042573706207, "learning_rate": 3.698057441868603e-05, "loss": 2.6083, "step": 30227 }, { "epoch": 1.8764665714817803, "grad_norm": 0.5639402970425714, "learning_rate": 3.697708751189246e-05, "loss": 2.7566, "step": 30228 }, { "epoch": 1.8765286485815382, "grad_norm": 0.5480394228308266, "learning_rate": 3.69736006730426e-05, "loss": 2.6777, "step": 30229 }, { "epoch": 1.8765907256812961, "grad_norm": 0.5294671717256993, "learning_rate": 3.6970113902154615e-05, "loss": 2.6226, "step": 30230 }, { "epoch": 1.876652802781054, "grad_norm": 0.5720793645526151, "learning_rate": 3.696662719924672e-05, "loss": 2.8024, "step": 30231 }, { "epoch": 1.876714879880812, "grad_norm": 0.5299677802134981, "learning_rate": 3.696314056433707e-05, "loss": 2.7605, "step": 30232 }, { "epoch": 1.87677695698057, "grad_norm": 0.5270842302136336, "learning_rate": 3.69596539974439e-05, "loss": 2.7849, "step": 30233 }, { "epoch": 1.8768390340803278, "grad_norm": 0.5233868247027239, "learning_rate": 3.695616749858538e-05, "loss": 2.6869, "step": 30234 }, { "epoch": 1.8769011111800857, "grad_norm": 0.515063702800741, "learning_rate": 3.69526810677797e-05, "loss": 2.688, "step": 30235 }, { "epoch": 1.8769631882798437, "grad_norm": 0.5227894828458701, "learning_rate": 3.6949194705045043e-05, "loss": 2.7419, "step": 30236 }, { "epoch": 1.8770252653796016, "grad_norm": 0.4962264656905088, "learning_rate": 3.694570841039962e-05, "loss": 2.7131, "step": 30237 }, { "epoch": 1.8770873424793595, "grad_norm": 0.5412092723015478, "learning_rate": 3.69422221838616e-05, "loss": 2.6888, "step": 30238 }, { "epoch": 1.8771494195791174, "grad_norm": 0.4885702117383514, "learning_rate": 3.693873602544917e-05, "loss": 2.7013, "step": 30239 }, { "epoch": 1.8772114966788753, "grad_norm": 0.529130691396779, "learning_rate": 3.693524993518053e-05, "loss": 2.7575, "step": 30240 }, { "epoch": 1.877273573778633, "grad_norm": 0.5088117598085556, "learning_rate": 3.6931763913073866e-05, "loss": 2.7061, "step": 30241 }, { "epoch": 1.877335650878391, "grad_norm": 0.4786141274722405, "learning_rate": 3.6928277959147354e-05, "loss": 2.718, "step": 30242 }, { "epoch": 1.8773977279781489, "grad_norm": 0.4982147233153422, "learning_rate": 3.69247920734192e-05, "loss": 2.6806, "step": 30243 }, { "epoch": 1.8774598050779068, "grad_norm": 0.5428489443107909, "learning_rate": 3.692130625590757e-05, "loss": 2.691, "step": 30244 }, { "epoch": 1.8775218821776647, "grad_norm": 0.5125973271786576, "learning_rate": 3.691782050663066e-05, "loss": 2.6619, "step": 30245 }, { "epoch": 1.8775839592774226, "grad_norm": 0.5421118251966714, "learning_rate": 3.6914334825606666e-05, "loss": 2.6571, "step": 30246 }, { "epoch": 1.8776460363771803, "grad_norm": 0.537962496076048, "learning_rate": 3.691084921285375e-05, "loss": 2.7115, "step": 30247 }, { "epoch": 1.8777081134769382, "grad_norm": 0.5390013508045489, "learning_rate": 3.6907363668390125e-05, "loss": 2.836, "step": 30248 }, { "epoch": 1.8777701905766961, "grad_norm": 0.5048876527889959, "learning_rate": 3.6903878192233955e-05, "loss": 2.7359, "step": 30249 }, { "epoch": 1.877832267676454, "grad_norm": 0.5818468352158722, "learning_rate": 3.6900392784403434e-05, "loss": 2.7501, "step": 30250 }, { "epoch": 1.877894344776212, "grad_norm": 0.5216689622491332, "learning_rate": 3.6896907444916736e-05, "loss": 2.778, "step": 30251 }, { "epoch": 1.87795642187597, "grad_norm": 0.5813699844560661, "learning_rate": 3.689342217379207e-05, "loss": 2.6672, "step": 30252 }, { "epoch": 1.8780184989757278, "grad_norm": 0.5374010161255719, "learning_rate": 3.6889936971047575e-05, "loss": 2.6639, "step": 30253 }, { "epoch": 1.8780805760754857, "grad_norm": 0.5238938975164974, "learning_rate": 3.688645183670148e-05, "loss": 2.7678, "step": 30254 }, { "epoch": 1.8781426531752436, "grad_norm": 0.5851079791013135, "learning_rate": 3.688296677077195e-05, "loss": 2.7413, "step": 30255 }, { "epoch": 1.8782047302750016, "grad_norm": 0.5292750561710944, "learning_rate": 3.687948177327717e-05, "loss": 2.8313, "step": 30256 }, { "epoch": 1.8782668073747595, "grad_norm": 0.5613774118042507, "learning_rate": 3.687599684423532e-05, "loss": 2.6807, "step": 30257 }, { "epoch": 1.8783288844745174, "grad_norm": 0.5225056281539563, "learning_rate": 3.687251198366456e-05, "loss": 2.845, "step": 30258 }, { "epoch": 1.8783909615742753, "grad_norm": 0.6000896605416013, "learning_rate": 3.6869027191583114e-05, "loss": 2.8128, "step": 30259 }, { "epoch": 1.8784530386740332, "grad_norm": 0.5561312570166319, "learning_rate": 3.6865542468009125e-05, "loss": 2.8034, "step": 30260 }, { "epoch": 1.8785151157737912, "grad_norm": 0.5213573458254207, "learning_rate": 3.68620578129608e-05, "loss": 2.8055, "step": 30261 }, { "epoch": 1.878577192873549, "grad_norm": 0.5497031245090263, "learning_rate": 3.68585732264563e-05, "loss": 2.6648, "step": 30262 }, { "epoch": 1.878639269973307, "grad_norm": 0.5552550085701491, "learning_rate": 3.685508870851382e-05, "loss": 2.6753, "step": 30263 }, { "epoch": 1.878701347073065, "grad_norm": 0.5613344886328814, "learning_rate": 3.685160425915153e-05, "loss": 2.7093, "step": 30264 }, { "epoch": 1.8787634241728226, "grad_norm": 0.5478457146262787, "learning_rate": 3.6848119878387595e-05, "loss": 2.7293, "step": 30265 }, { "epoch": 1.8788255012725805, "grad_norm": 0.6280126679974198, "learning_rate": 3.684463556624023e-05, "loss": 2.6476, "step": 30266 }, { "epoch": 1.8788875783723384, "grad_norm": 0.525598591198913, "learning_rate": 3.68411513227276e-05, "loss": 2.7036, "step": 30267 }, { "epoch": 1.8789496554720964, "grad_norm": 0.5892162395576565, "learning_rate": 3.6837667147867874e-05, "loss": 2.7148, "step": 30268 }, { "epoch": 1.8790117325718543, "grad_norm": 0.5798254008410724, "learning_rate": 3.683418304167924e-05, "loss": 2.6384, "step": 30269 }, { "epoch": 1.8790738096716122, "grad_norm": 0.5369378943234819, "learning_rate": 3.683069900417985e-05, "loss": 2.7391, "step": 30270 }, { "epoch": 1.87913588677137, "grad_norm": 0.5458859901169223, "learning_rate": 3.682721503538792e-05, "loss": 2.7363, "step": 30271 }, { "epoch": 1.8791979638711278, "grad_norm": 0.581275474851471, "learning_rate": 3.682373113532159e-05, "loss": 2.7488, "step": 30272 }, { "epoch": 1.8792600409708857, "grad_norm": 0.5244640134751969, "learning_rate": 3.6820247303999066e-05, "loss": 2.8187, "step": 30273 }, { "epoch": 1.8793221180706436, "grad_norm": 0.5761550642244357, "learning_rate": 3.6816763541438505e-05, "loss": 2.7682, "step": 30274 }, { "epoch": 1.8793841951704016, "grad_norm": 0.5974587257446572, "learning_rate": 3.68132798476581e-05, "loss": 2.7373, "step": 30275 }, { "epoch": 1.8794462722701595, "grad_norm": 0.5307967552545833, "learning_rate": 3.6809796222675994e-05, "loss": 2.71, "step": 30276 }, { "epoch": 1.8795083493699174, "grad_norm": 0.5496820343509221, "learning_rate": 3.6806312666510405e-05, "loss": 2.762, "step": 30277 }, { "epoch": 1.8795704264696753, "grad_norm": 0.5450374521720946, "learning_rate": 3.680282917917947e-05, "loss": 2.7387, "step": 30278 }, { "epoch": 1.8796325035694332, "grad_norm": 0.5550352269239547, "learning_rate": 3.679934576070139e-05, "loss": 2.7528, "step": 30279 }, { "epoch": 1.8796945806691912, "grad_norm": 0.5989449513364054, "learning_rate": 3.6795862411094316e-05, "loss": 2.8162, "step": 30280 }, { "epoch": 1.879756657768949, "grad_norm": 0.5562152240897961, "learning_rate": 3.6792379130376456e-05, "loss": 2.8076, "step": 30281 }, { "epoch": 1.879818734868707, "grad_norm": 0.5436090220833176, "learning_rate": 3.6788895918565946e-05, "loss": 2.7506, "step": 30282 }, { "epoch": 1.879880811968465, "grad_norm": 0.5533806086312651, "learning_rate": 3.678541277568096e-05, "loss": 2.7519, "step": 30283 }, { "epoch": 1.8799428890682228, "grad_norm": 0.5447596920493996, "learning_rate": 3.6781929701739706e-05, "loss": 2.7713, "step": 30284 }, { "epoch": 1.8800049661679807, "grad_norm": 0.5204095228767227, "learning_rate": 3.6778446696760316e-05, "loss": 2.7104, "step": 30285 }, { "epoch": 1.8800670432677387, "grad_norm": 0.5805255692655761, "learning_rate": 3.677496376076099e-05, "loss": 2.7822, "step": 30286 }, { "epoch": 1.8801291203674966, "grad_norm": 0.5051418503214238, "learning_rate": 3.677148089375988e-05, "loss": 2.7268, "step": 30287 }, { "epoch": 1.8801911974672545, "grad_norm": 0.5500984288052254, "learning_rate": 3.676799809577517e-05, "loss": 2.6664, "step": 30288 }, { "epoch": 1.8802532745670122, "grad_norm": 0.5421141341204332, "learning_rate": 3.6764515366825024e-05, "loss": 2.7736, "step": 30289 }, { "epoch": 1.88031535166677, "grad_norm": 0.5690547713415292, "learning_rate": 3.676103270692762e-05, "loss": 2.7825, "step": 30290 }, { "epoch": 1.880377428766528, "grad_norm": 0.5606960138106771, "learning_rate": 3.675755011610112e-05, "loss": 2.717, "step": 30291 }, { "epoch": 1.880439505866286, "grad_norm": 0.5370876793380979, "learning_rate": 3.6754067594363694e-05, "loss": 2.7221, "step": 30292 }, { "epoch": 1.8805015829660439, "grad_norm": 0.5462846294137284, "learning_rate": 3.675058514173352e-05, "loss": 2.7295, "step": 30293 }, { "epoch": 1.8805636600658018, "grad_norm": 0.5830621230804733, "learning_rate": 3.674710275822875e-05, "loss": 2.7628, "step": 30294 }, { "epoch": 1.8806257371655595, "grad_norm": 0.5468108003262688, "learning_rate": 3.674362044386757e-05, "loss": 2.7228, "step": 30295 }, { "epoch": 1.8806878142653174, "grad_norm": 0.510984111841228, "learning_rate": 3.674013819866813e-05, "loss": 2.6207, "step": 30296 }, { "epoch": 1.8807498913650753, "grad_norm": 0.533205220983389, "learning_rate": 3.673665602264862e-05, "loss": 2.7944, "step": 30297 }, { "epoch": 1.8808119684648332, "grad_norm": 0.5416764723061399, "learning_rate": 3.6733173915827176e-05, "loss": 2.7222, "step": 30298 }, { "epoch": 1.8808740455645911, "grad_norm": 0.5892698767833376, "learning_rate": 3.672969187822199e-05, "loss": 2.6867, "step": 30299 }, { "epoch": 1.880936122664349, "grad_norm": 0.5216682481386306, "learning_rate": 3.6726209909851234e-05, "loss": 2.7819, "step": 30300 }, { "epoch": 1.880998199764107, "grad_norm": 0.5701175251901273, "learning_rate": 3.6722728010733056e-05, "loss": 2.6846, "step": 30301 }, { "epoch": 1.881060276863865, "grad_norm": 0.6194146930590091, "learning_rate": 3.671924618088564e-05, "loss": 2.8237, "step": 30302 }, { "epoch": 1.8811223539636228, "grad_norm": 0.5629709973953454, "learning_rate": 3.6715764420327123e-05, "loss": 2.7006, "step": 30303 }, { "epoch": 1.8811844310633807, "grad_norm": 0.6300063137843115, "learning_rate": 3.671228272907571e-05, "loss": 2.8084, "step": 30304 }, { "epoch": 1.8812465081631387, "grad_norm": 0.5768852860348934, "learning_rate": 3.670880110714952e-05, "loss": 2.7384, "step": 30305 }, { "epoch": 1.8813085852628966, "grad_norm": 0.5666634337482607, "learning_rate": 3.670531955456676e-05, "loss": 2.7649, "step": 30306 }, { "epoch": 1.8813706623626545, "grad_norm": 0.5797606090497791, "learning_rate": 3.670183807134557e-05, "loss": 2.6844, "step": 30307 }, { "epoch": 1.8814327394624124, "grad_norm": 0.6041069713785752, "learning_rate": 3.6698356657504115e-05, "loss": 2.784, "step": 30308 }, { "epoch": 1.8814948165621703, "grad_norm": 0.6057649933663304, "learning_rate": 3.669487531306057e-05, "loss": 2.7676, "step": 30309 }, { "epoch": 1.8815568936619282, "grad_norm": 0.5436266748293341, "learning_rate": 3.6691394038033076e-05, "loss": 2.7144, "step": 30310 }, { "epoch": 1.8816189707616862, "grad_norm": 0.5543457375322041, "learning_rate": 3.668791283243983e-05, "loss": 2.7622, "step": 30311 }, { "epoch": 1.881681047861444, "grad_norm": 0.5790161018290354, "learning_rate": 3.668443169629896e-05, "loss": 2.7394, "step": 30312 }, { "epoch": 1.8817431249612018, "grad_norm": 0.613005047456892, "learning_rate": 3.668095062962865e-05, "loss": 2.7175, "step": 30313 }, { "epoch": 1.8818052020609597, "grad_norm": 0.6217199563596899, "learning_rate": 3.667746963244704e-05, "loss": 2.819, "step": 30314 }, { "epoch": 1.8818672791607176, "grad_norm": 0.6604170199572784, "learning_rate": 3.667398870477232e-05, "loss": 2.6696, "step": 30315 }, { "epoch": 1.8819293562604755, "grad_norm": 0.525223971457761, "learning_rate": 3.6670507846622624e-05, "loss": 2.6214, "step": 30316 }, { "epoch": 1.8819914333602334, "grad_norm": 0.7050548771541753, "learning_rate": 3.666702705801614e-05, "loss": 2.7646, "step": 30317 }, { "epoch": 1.8820535104599914, "grad_norm": 0.5649107032022568, "learning_rate": 3.666354633897101e-05, "loss": 2.7655, "step": 30318 }, { "epoch": 1.882115587559749, "grad_norm": 0.6095406478317723, "learning_rate": 3.666006568950539e-05, "loss": 2.8123, "step": 30319 }, { "epoch": 1.882177664659507, "grad_norm": 0.5915472429487427, "learning_rate": 3.665658510963744e-05, "loss": 2.7774, "step": 30320 }, { "epoch": 1.882239741759265, "grad_norm": 0.6018142173553218, "learning_rate": 3.665310459938533e-05, "loss": 2.6773, "step": 30321 }, { "epoch": 1.8823018188590228, "grad_norm": 0.5583677835179615, "learning_rate": 3.664962415876722e-05, "loss": 2.7066, "step": 30322 }, { "epoch": 1.8823638959587807, "grad_norm": 0.5348741943627887, "learning_rate": 3.664614378780125e-05, "loss": 2.7878, "step": 30323 }, { "epoch": 1.8824259730585386, "grad_norm": 0.5832317439809999, "learning_rate": 3.6642663486505604e-05, "loss": 2.7509, "step": 30324 }, { "epoch": 1.8824880501582966, "grad_norm": 0.5257878273358392, "learning_rate": 3.6639183254898404e-05, "loss": 2.7952, "step": 30325 }, { "epoch": 1.8825501272580545, "grad_norm": 0.5564004212491671, "learning_rate": 3.6635703092997855e-05, "loss": 2.651, "step": 30326 }, { "epoch": 1.8826122043578124, "grad_norm": 0.5493387582257987, "learning_rate": 3.6632223000822063e-05, "loss": 2.7148, "step": 30327 }, { "epoch": 1.8826742814575703, "grad_norm": 0.5387689371608759, "learning_rate": 3.6628742978389216e-05, "loss": 2.8316, "step": 30328 }, { "epoch": 1.8827363585573282, "grad_norm": 0.5697076281048408, "learning_rate": 3.6625263025717474e-05, "loss": 2.8167, "step": 30329 }, { "epoch": 1.8827984356570862, "grad_norm": 0.4997883438115441, "learning_rate": 3.662178314282497e-05, "loss": 2.7149, "step": 30330 }, { "epoch": 1.882860512756844, "grad_norm": 0.5626278573001566, "learning_rate": 3.661830332972986e-05, "loss": 2.687, "step": 30331 }, { "epoch": 1.882922589856602, "grad_norm": 0.50183026780698, "learning_rate": 3.661482358645033e-05, "loss": 2.7416, "step": 30332 }, { "epoch": 1.88298466695636, "grad_norm": 0.558280109838885, "learning_rate": 3.6611343913004505e-05, "loss": 2.7815, "step": 30333 }, { "epoch": 1.8830467440561178, "grad_norm": 0.5434360551513278, "learning_rate": 3.660786430941056e-05, "loss": 2.6698, "step": 30334 }, { "epoch": 1.8831088211558757, "grad_norm": 0.5148234929222537, "learning_rate": 3.660438477568662e-05, "loss": 2.8053, "step": 30335 }, { "epoch": 1.8831708982556337, "grad_norm": 0.5498991050076109, "learning_rate": 3.660090531185088e-05, "loss": 2.6668, "step": 30336 }, { "epoch": 1.8832329753553914, "grad_norm": 0.5248322479995565, "learning_rate": 3.6597425917921444e-05, "loss": 2.6687, "step": 30337 }, { "epoch": 1.8832950524551493, "grad_norm": 0.5541211850520978, "learning_rate": 3.659394659391651e-05, "loss": 2.7475, "step": 30338 }, { "epoch": 1.8833571295549072, "grad_norm": 0.5271099844013523, "learning_rate": 3.65904673398542e-05, "loss": 2.7436, "step": 30339 }, { "epoch": 1.8834192066546651, "grad_norm": 0.622368925036991, "learning_rate": 3.658698815575269e-05, "loss": 2.723, "step": 30340 }, { "epoch": 1.883481283754423, "grad_norm": 0.505820814157924, "learning_rate": 3.65835090416301e-05, "loss": 2.8403, "step": 30341 }, { "epoch": 1.883543360854181, "grad_norm": 0.53248685019111, "learning_rate": 3.6580029997504615e-05, "loss": 2.6758, "step": 30342 }, { "epoch": 1.8836054379539386, "grad_norm": 0.6009581118265909, "learning_rate": 3.6576551023394377e-05, "loss": 2.718, "step": 30343 }, { "epoch": 1.8836675150536966, "grad_norm": 0.5503699207405092, "learning_rate": 3.657307211931752e-05, "loss": 2.7779, "step": 30344 }, { "epoch": 1.8837295921534545, "grad_norm": 0.5274535453372753, "learning_rate": 3.656959328529221e-05, "loss": 2.6709, "step": 30345 }, { "epoch": 1.8837916692532124, "grad_norm": 0.6040598073268816, "learning_rate": 3.656611452133658e-05, "loss": 2.7294, "step": 30346 }, { "epoch": 1.8838537463529703, "grad_norm": 0.5685516504110285, "learning_rate": 3.656263582746881e-05, "loss": 2.7052, "step": 30347 }, { "epoch": 1.8839158234527282, "grad_norm": 0.609284223006223, "learning_rate": 3.6559157203707015e-05, "loss": 2.6915, "step": 30348 }, { "epoch": 1.8839779005524862, "grad_norm": 0.5791733747089255, "learning_rate": 3.655567865006937e-05, "loss": 2.7854, "step": 30349 }, { "epoch": 1.884039977652244, "grad_norm": 0.520138980238459, "learning_rate": 3.6552200166574e-05, "loss": 2.6678, "step": 30350 }, { "epoch": 1.884102054752002, "grad_norm": 0.6497250837544186, "learning_rate": 3.654872175323909e-05, "loss": 2.7896, "step": 30351 }, { "epoch": 1.88416413185176, "grad_norm": 0.5549058887231099, "learning_rate": 3.654524341008274e-05, "loss": 2.7426, "step": 30352 }, { "epoch": 1.8842262089515178, "grad_norm": 0.6309333970426305, "learning_rate": 3.654176513712314e-05, "loss": 2.8521, "step": 30353 }, { "epoch": 1.8842882860512757, "grad_norm": 0.6089126348907126, "learning_rate": 3.653828693437841e-05, "loss": 2.7813, "step": 30354 }, { "epoch": 1.8843503631510337, "grad_norm": 0.5914249321469857, "learning_rate": 3.653480880186671e-05, "loss": 2.7328, "step": 30355 }, { "epoch": 1.8844124402507916, "grad_norm": 0.5804584681959596, "learning_rate": 3.653133073960617e-05, "loss": 2.7889, "step": 30356 }, { "epoch": 1.8844745173505495, "grad_norm": 0.61502534251993, "learning_rate": 3.652785274761496e-05, "loss": 2.7688, "step": 30357 }, { "epoch": 1.8845365944503074, "grad_norm": 0.6066262041430153, "learning_rate": 3.65243748259112e-05, "loss": 2.7574, "step": 30358 }, { "epoch": 1.8845986715500653, "grad_norm": 0.5554040122635551, "learning_rate": 3.652089697451305e-05, "loss": 2.7841, "step": 30359 }, { "epoch": 1.884660748649823, "grad_norm": 0.6003582530579944, "learning_rate": 3.6517419193438665e-05, "loss": 2.7033, "step": 30360 }, { "epoch": 1.884722825749581, "grad_norm": 0.5687051881176871, "learning_rate": 3.651394148270616e-05, "loss": 2.7487, "step": 30361 }, { "epoch": 1.8847849028493389, "grad_norm": 0.610113430108096, "learning_rate": 3.651046384233371e-05, "loss": 2.8044, "step": 30362 }, { "epoch": 1.8848469799490968, "grad_norm": 0.5587163670549474, "learning_rate": 3.650698627233942e-05, "loss": 2.7487, "step": 30363 }, { "epoch": 1.8849090570488547, "grad_norm": 0.5427926141481668, "learning_rate": 3.650350877274147e-05, "loss": 2.718, "step": 30364 }, { "epoch": 1.8849711341486126, "grad_norm": 0.49297070631313117, "learning_rate": 3.6500031343558004e-05, "loss": 2.6344, "step": 30365 }, { "epoch": 1.8850332112483703, "grad_norm": 0.5589580548352451, "learning_rate": 3.649655398480715e-05, "loss": 2.7976, "step": 30366 }, { "epoch": 1.8850952883481282, "grad_norm": 0.5169082337088804, "learning_rate": 3.6493076696507046e-05, "loss": 2.7022, "step": 30367 }, { "epoch": 1.8851573654478861, "grad_norm": 0.5314955704774491, "learning_rate": 3.648959947867585e-05, "loss": 2.7406, "step": 30368 }, { "epoch": 1.885219442547644, "grad_norm": 0.5749394312346251, "learning_rate": 3.648612233133168e-05, "loss": 2.6955, "step": 30369 }, { "epoch": 1.885281519647402, "grad_norm": 0.5321822121716436, "learning_rate": 3.64826452544927e-05, "loss": 2.7795, "step": 30370 }, { "epoch": 1.88534359674716, "grad_norm": 0.5922363965624703, "learning_rate": 3.647916824817703e-05, "loss": 2.7158, "step": 30371 }, { "epoch": 1.8854056738469178, "grad_norm": 0.5346053134664146, "learning_rate": 3.6475691312402836e-05, "loss": 2.7775, "step": 30372 }, { "epoch": 1.8854677509466757, "grad_norm": 0.5962440911086212, "learning_rate": 3.647221444718823e-05, "loss": 2.7439, "step": 30373 }, { "epoch": 1.8855298280464337, "grad_norm": 0.5548061727455443, "learning_rate": 3.646873765255138e-05, "loss": 2.7543, "step": 30374 }, { "epoch": 1.8855919051461916, "grad_norm": 0.5211769941144595, "learning_rate": 3.646526092851039e-05, "loss": 2.7622, "step": 30375 }, { "epoch": 1.8856539822459495, "grad_norm": 0.5416004036451805, "learning_rate": 3.646178427508344e-05, "loss": 2.6978, "step": 30376 }, { "epoch": 1.8857160593457074, "grad_norm": 0.48839515883251317, "learning_rate": 3.6458307692288636e-05, "loss": 2.6838, "step": 30377 }, { "epoch": 1.8857781364454653, "grad_norm": 0.5478954875561101, "learning_rate": 3.6454831180144135e-05, "loss": 2.7641, "step": 30378 }, { "epoch": 1.8858402135452232, "grad_norm": 0.5043386925128732, "learning_rate": 3.645135473866807e-05, "loss": 2.7262, "step": 30379 }, { "epoch": 1.8859022906449812, "grad_norm": 0.569837163076988, "learning_rate": 3.6447878367878566e-05, "loss": 2.698, "step": 30380 }, { "epoch": 1.885964367744739, "grad_norm": 0.5447510482301821, "learning_rate": 3.6444402067793785e-05, "loss": 2.8177, "step": 30381 }, { "epoch": 1.886026444844497, "grad_norm": 0.5148089048318671, "learning_rate": 3.644092583843183e-05, "loss": 2.7557, "step": 30382 }, { "epoch": 1.886088521944255, "grad_norm": 0.5143076801859955, "learning_rate": 3.643744967981087e-05, "loss": 2.7154, "step": 30383 }, { "epoch": 1.8861505990440126, "grad_norm": 0.5675142959861058, "learning_rate": 3.643397359194902e-05, "loss": 2.6799, "step": 30384 }, { "epoch": 1.8862126761437705, "grad_norm": 0.49279122942277787, "learning_rate": 3.643049757486443e-05, "loss": 2.786, "step": 30385 }, { "epoch": 1.8862747532435284, "grad_norm": 0.5844278342286352, "learning_rate": 3.642702162857522e-05, "loss": 2.7566, "step": 30386 }, { "epoch": 1.8863368303432864, "grad_norm": 0.5164717455260227, "learning_rate": 3.642354575309954e-05, "loss": 2.7824, "step": 30387 }, { "epoch": 1.8863989074430443, "grad_norm": 0.5545227450838862, "learning_rate": 3.64200699484555e-05, "loss": 2.7808, "step": 30388 }, { "epoch": 1.8864609845428022, "grad_norm": 0.48664837752502593, "learning_rate": 3.641659421466128e-05, "loss": 2.7409, "step": 30389 }, { "epoch": 1.88652306164256, "grad_norm": 0.5427637056107625, "learning_rate": 3.641311855173498e-05, "loss": 2.6738, "step": 30390 }, { "epoch": 1.8865851387423178, "grad_norm": 0.5964287512724306, "learning_rate": 3.6409642959694725e-05, "loss": 2.7424, "step": 30391 }, { "epoch": 1.8866472158420757, "grad_norm": 0.5308738947911409, "learning_rate": 3.640616743855867e-05, "loss": 2.7991, "step": 30392 }, { "epoch": 1.8867092929418336, "grad_norm": 0.5752022294307022, "learning_rate": 3.6402691988344936e-05, "loss": 2.7634, "step": 30393 }, { "epoch": 1.8867713700415916, "grad_norm": 0.5432505528113216, "learning_rate": 3.6399216609071664e-05, "loss": 2.8023, "step": 30394 }, { "epoch": 1.8868334471413495, "grad_norm": 0.5436322665734307, "learning_rate": 3.639574130075697e-05, "loss": 2.806, "step": 30395 }, { "epoch": 1.8868955242411074, "grad_norm": 0.5631226960620893, "learning_rate": 3.639226606341899e-05, "loss": 2.6883, "step": 30396 }, { "epoch": 1.8869576013408653, "grad_norm": 0.5268555818828076, "learning_rate": 3.6388790897075876e-05, "loss": 2.6815, "step": 30397 }, { "epoch": 1.8870196784406232, "grad_norm": 0.5573823504479773, "learning_rate": 3.638531580174574e-05, "loss": 2.6734, "step": 30398 }, { "epoch": 1.8870817555403812, "grad_norm": 0.550326518108229, "learning_rate": 3.6381840777446725e-05, "loss": 2.7102, "step": 30399 }, { "epoch": 1.887143832640139, "grad_norm": 0.5673832852333712, "learning_rate": 3.6378365824196934e-05, "loss": 2.7574, "step": 30400 }, { "epoch": 1.887205909739897, "grad_norm": 0.54253441433368, "learning_rate": 3.637489094201454e-05, "loss": 2.8413, "step": 30401 }, { "epoch": 1.887267986839655, "grad_norm": 0.5721816434201121, "learning_rate": 3.637141613091763e-05, "loss": 2.7578, "step": 30402 }, { "epoch": 1.8873300639394128, "grad_norm": 0.5161076800747092, "learning_rate": 3.636794139092436e-05, "loss": 2.793, "step": 30403 }, { "epoch": 1.8873921410391707, "grad_norm": 0.5831444354973473, "learning_rate": 3.636446672205285e-05, "loss": 2.7728, "step": 30404 }, { "epoch": 1.8874542181389287, "grad_norm": 0.4842124020017891, "learning_rate": 3.636099212432121e-05, "loss": 2.7486, "step": 30405 }, { "epoch": 1.8875162952386866, "grad_norm": 0.6151033049945357, "learning_rate": 3.635751759774761e-05, "loss": 2.7442, "step": 30406 }, { "epoch": 1.8875783723384445, "grad_norm": 0.5234716603533712, "learning_rate": 3.635404314235014e-05, "loss": 2.7533, "step": 30407 }, { "epoch": 1.8876404494382022, "grad_norm": 0.5537955675256332, "learning_rate": 3.635056875814694e-05, "loss": 2.6822, "step": 30408 }, { "epoch": 1.8877025265379601, "grad_norm": 0.5768428451735167, "learning_rate": 3.6347094445156126e-05, "loss": 2.8817, "step": 30409 }, { "epoch": 1.887764603637718, "grad_norm": 0.5941324758879962, "learning_rate": 3.634362020339586e-05, "loss": 2.7848, "step": 30410 }, { "epoch": 1.887826680737476, "grad_norm": 0.5873162134306246, "learning_rate": 3.634014603288421e-05, "loss": 2.7725, "step": 30411 }, { "epoch": 1.8878887578372339, "grad_norm": 0.5411175518215035, "learning_rate": 3.633667193363936e-05, "loss": 2.786, "step": 30412 }, { "epoch": 1.8879508349369918, "grad_norm": 0.5560508414357572, "learning_rate": 3.633319790567939e-05, "loss": 2.7397, "step": 30413 }, { "epoch": 1.8880129120367495, "grad_norm": 0.5048782082795985, "learning_rate": 3.6329723949022465e-05, "loss": 2.6348, "step": 30414 }, { "epoch": 1.8880749891365074, "grad_norm": 0.5222062390273613, "learning_rate": 3.632625006368669e-05, "loss": 2.6195, "step": 30415 }, { "epoch": 1.8881370662362653, "grad_norm": 0.586188021191781, "learning_rate": 3.6322776249690166e-05, "loss": 2.7861, "step": 30416 }, { "epoch": 1.8881991433360232, "grad_norm": 0.47549918634864513, "learning_rate": 3.6319302507051054e-05, "loss": 2.6982, "step": 30417 }, { "epoch": 1.8882612204357812, "grad_norm": 0.5707045735909356, "learning_rate": 3.631582883578745e-05, "loss": 2.7011, "step": 30418 }, { "epoch": 1.888323297535539, "grad_norm": 0.5311964793453401, "learning_rate": 3.6312355235917505e-05, "loss": 2.6888, "step": 30419 }, { "epoch": 1.888385374635297, "grad_norm": 0.5621797921806395, "learning_rate": 3.630888170745931e-05, "loss": 2.6774, "step": 30420 }, { "epoch": 1.888447451735055, "grad_norm": 0.5283180561123259, "learning_rate": 3.630540825043102e-05, "loss": 2.7273, "step": 30421 }, { "epoch": 1.8885095288348128, "grad_norm": 0.6225917912028147, "learning_rate": 3.630193486485072e-05, "loss": 2.7758, "step": 30422 }, { "epoch": 1.8885716059345707, "grad_norm": 0.5604007277886963, "learning_rate": 3.6298461550736565e-05, "loss": 2.6541, "step": 30423 }, { "epoch": 1.8886336830343287, "grad_norm": 0.6506499039998369, "learning_rate": 3.629498830810665e-05, "loss": 2.7365, "step": 30424 }, { "epoch": 1.8886957601340866, "grad_norm": 0.5356296199924276, "learning_rate": 3.6291515136979116e-05, "loss": 2.7766, "step": 30425 }, { "epoch": 1.8887578372338445, "grad_norm": 0.6651806337575131, "learning_rate": 3.628804203737207e-05, "loss": 2.7697, "step": 30426 }, { "epoch": 1.8888199143336024, "grad_norm": 0.5737190148042174, "learning_rate": 3.628456900930365e-05, "loss": 2.7829, "step": 30427 }, { "epoch": 1.8888819914333603, "grad_norm": 0.5644760127195626, "learning_rate": 3.628109605279196e-05, "loss": 2.7011, "step": 30428 }, { "epoch": 1.8889440685331182, "grad_norm": 0.5755201630279182, "learning_rate": 3.62776231678551e-05, "loss": 2.7441, "step": 30429 }, { "epoch": 1.8890061456328762, "grad_norm": 0.5646654937289838, "learning_rate": 3.627415035451123e-05, "loss": 2.7218, "step": 30430 }, { "epoch": 1.889068222732634, "grad_norm": 0.6095563667557512, "learning_rate": 3.6270677612778446e-05, "loss": 2.7863, "step": 30431 }, { "epoch": 1.8891302998323918, "grad_norm": 0.532742367600716, "learning_rate": 3.626720494267487e-05, "loss": 2.7895, "step": 30432 }, { "epoch": 1.8891923769321497, "grad_norm": 0.6084699512168317, "learning_rate": 3.626373234421863e-05, "loss": 2.7522, "step": 30433 }, { "epoch": 1.8892544540319076, "grad_norm": 0.506790041453319, "learning_rate": 3.626025981742781e-05, "loss": 2.7134, "step": 30434 }, { "epoch": 1.8893165311316655, "grad_norm": 0.5990213250470298, "learning_rate": 3.625678736232058e-05, "loss": 2.6414, "step": 30435 }, { "epoch": 1.8893786082314235, "grad_norm": 0.48201623457362136, "learning_rate": 3.6253314978915e-05, "loss": 2.6991, "step": 30436 }, { "epoch": 1.8894406853311814, "grad_norm": 0.5193987512494169, "learning_rate": 3.6249842667229226e-05, "loss": 2.6917, "step": 30437 }, { "epoch": 1.889502762430939, "grad_norm": 0.575234066964279, "learning_rate": 3.624637042728135e-05, "loss": 2.6785, "step": 30438 }, { "epoch": 1.889564839530697, "grad_norm": 0.5024843781448507, "learning_rate": 3.624289825908952e-05, "loss": 2.7215, "step": 30439 }, { "epoch": 1.889626916630455, "grad_norm": 0.5779535193910295, "learning_rate": 3.623942616267182e-05, "loss": 2.7211, "step": 30440 }, { "epoch": 1.8896889937302128, "grad_norm": 0.5491248309601702, "learning_rate": 3.6235954138046364e-05, "loss": 2.7269, "step": 30441 }, { "epoch": 1.8897510708299707, "grad_norm": 0.579409358292831, "learning_rate": 3.6232482185231286e-05, "loss": 2.7584, "step": 30442 }, { "epoch": 1.8898131479297287, "grad_norm": 0.596887405626087, "learning_rate": 3.622901030424468e-05, "loss": 2.6547, "step": 30443 }, { "epoch": 1.8898752250294866, "grad_norm": 0.5255254976527461, "learning_rate": 3.622553849510469e-05, "loss": 2.7238, "step": 30444 }, { "epoch": 1.8899373021292445, "grad_norm": 0.50965453358777, "learning_rate": 3.622206675782939e-05, "loss": 2.6227, "step": 30445 }, { "epoch": 1.8899993792290024, "grad_norm": 0.5167838858358264, "learning_rate": 3.621859509243692e-05, "loss": 2.7624, "step": 30446 }, { "epoch": 1.8900614563287603, "grad_norm": 0.49107671462347485, "learning_rate": 3.621512349894539e-05, "loss": 2.6933, "step": 30447 }, { "epoch": 1.8901235334285182, "grad_norm": 0.5343866313538489, "learning_rate": 3.6211651977372904e-05, "loss": 2.7651, "step": 30448 }, { "epoch": 1.8901856105282762, "grad_norm": 0.5070096932694064, "learning_rate": 3.6208180527737564e-05, "loss": 2.7562, "step": 30449 }, { "epoch": 1.890247687628034, "grad_norm": 0.5479584451951398, "learning_rate": 3.620470915005751e-05, "loss": 2.7307, "step": 30450 }, { "epoch": 1.890309764727792, "grad_norm": 0.5282977763955758, "learning_rate": 3.620123784435083e-05, "loss": 2.9139, "step": 30451 }, { "epoch": 1.89037184182755, "grad_norm": 0.5605798372685432, "learning_rate": 3.6197766610635655e-05, "loss": 2.8122, "step": 30452 }, { "epoch": 1.8904339189273078, "grad_norm": 0.579169875646067, "learning_rate": 3.619429544893007e-05, "loss": 2.803, "step": 30453 }, { "epoch": 1.8904959960270658, "grad_norm": 0.5823603218244707, "learning_rate": 3.61908243592522e-05, "loss": 2.7051, "step": 30454 }, { "epoch": 1.8905580731268237, "grad_norm": 0.5205106753043341, "learning_rate": 3.618735334162015e-05, "loss": 2.6446, "step": 30455 }, { "epoch": 1.8906201502265814, "grad_norm": 0.5674135656114737, "learning_rate": 3.618388239605203e-05, "loss": 2.7161, "step": 30456 }, { "epoch": 1.8906822273263393, "grad_norm": 0.564779167651431, "learning_rate": 3.618041152256595e-05, "loss": 2.7883, "step": 30457 }, { "epoch": 1.8907443044260972, "grad_norm": 0.5009999409178739, "learning_rate": 3.617694072118002e-05, "loss": 2.7517, "step": 30458 }, { "epoch": 1.8908063815258551, "grad_norm": 0.649702974459025, "learning_rate": 3.6173469991912346e-05, "loss": 2.6991, "step": 30459 }, { "epoch": 1.890868458625613, "grad_norm": 0.5290465378761485, "learning_rate": 3.616999933478103e-05, "loss": 2.7451, "step": 30460 }, { "epoch": 1.890930535725371, "grad_norm": 0.5603235311533944, "learning_rate": 3.6166528749804193e-05, "loss": 2.759, "step": 30461 }, { "epoch": 1.8909926128251286, "grad_norm": 0.585501139599793, "learning_rate": 3.616305823699991e-05, "loss": 2.8646, "step": 30462 }, { "epoch": 1.8910546899248866, "grad_norm": 0.5265914071408833, "learning_rate": 3.615958779638633e-05, "loss": 2.7898, "step": 30463 }, { "epoch": 1.8911167670246445, "grad_norm": 0.5198506685558473, "learning_rate": 3.615611742798155e-05, "loss": 2.7251, "step": 30464 }, { "epoch": 1.8911788441244024, "grad_norm": 0.5334622378834224, "learning_rate": 3.615264713180366e-05, "loss": 2.7095, "step": 30465 }, { "epoch": 1.8912409212241603, "grad_norm": 0.5328360331419909, "learning_rate": 3.614917690787076e-05, "loss": 2.7269, "step": 30466 }, { "epoch": 1.8913029983239182, "grad_norm": 0.5036025499787585, "learning_rate": 3.614570675620098e-05, "loss": 2.7558, "step": 30467 }, { "epoch": 1.8913650754236762, "grad_norm": 0.5511655886891108, "learning_rate": 3.61422366768124e-05, "loss": 2.8099, "step": 30468 }, { "epoch": 1.891427152523434, "grad_norm": 0.5107712746501174, "learning_rate": 3.613876666972315e-05, "loss": 2.6485, "step": 30469 }, { "epoch": 1.891489229623192, "grad_norm": 0.5657475027972376, "learning_rate": 3.6135296734951307e-05, "loss": 2.7422, "step": 30470 }, { "epoch": 1.89155130672295, "grad_norm": 0.5508261858418211, "learning_rate": 3.6131826872514995e-05, "loss": 2.706, "step": 30471 }, { "epoch": 1.8916133838227078, "grad_norm": 0.5380327645007036, "learning_rate": 3.61283570824323e-05, "loss": 2.8427, "step": 30472 }, { "epoch": 1.8916754609224657, "grad_norm": 0.5910673775614352, "learning_rate": 3.6124887364721336e-05, "loss": 2.7148, "step": 30473 }, { "epoch": 1.8917375380222237, "grad_norm": 0.5616419393471251, "learning_rate": 3.61214177194002e-05, "loss": 2.781, "step": 30474 }, { "epoch": 1.8917996151219816, "grad_norm": 0.5357645851318072, "learning_rate": 3.6117948146487016e-05, "loss": 2.8259, "step": 30475 }, { "epoch": 1.8918616922217395, "grad_norm": 0.5814472315061128, "learning_rate": 3.6114478645999836e-05, "loss": 2.7868, "step": 30476 }, { "epoch": 1.8919237693214974, "grad_norm": 0.5285063467176996, "learning_rate": 3.6111009217956815e-05, "loss": 2.675, "step": 30477 }, { "epoch": 1.8919858464212553, "grad_norm": 0.552160848418023, "learning_rate": 3.6107539862376034e-05, "loss": 2.6598, "step": 30478 }, { "epoch": 1.8920479235210133, "grad_norm": 0.5329300736725998, "learning_rate": 3.610407057927557e-05, "loss": 2.7778, "step": 30479 }, { "epoch": 1.892110000620771, "grad_norm": 0.5583097362750966, "learning_rate": 3.610060136867356e-05, "loss": 2.7229, "step": 30480 }, { "epoch": 1.8921720777205289, "grad_norm": 0.5201808911918595, "learning_rate": 3.609713223058807e-05, "loss": 2.7582, "step": 30481 }, { "epoch": 1.8922341548202868, "grad_norm": 0.5712117460702045, "learning_rate": 3.6093663165037236e-05, "loss": 2.7283, "step": 30482 }, { "epoch": 1.8922962319200447, "grad_norm": 0.5833030221647924, "learning_rate": 3.609019417203912e-05, "loss": 2.7309, "step": 30483 }, { "epoch": 1.8923583090198026, "grad_norm": 0.5494607557734786, "learning_rate": 3.608672525161185e-05, "loss": 2.7194, "step": 30484 }, { "epoch": 1.8924203861195605, "grad_norm": 0.5546373156559427, "learning_rate": 3.6083256403773505e-05, "loss": 2.6695, "step": 30485 }, { "epoch": 1.8924824632193182, "grad_norm": 0.5701038815133861, "learning_rate": 3.607978762854219e-05, "loss": 2.6711, "step": 30486 }, { "epoch": 1.8925445403190762, "grad_norm": 0.5562561849185974, "learning_rate": 3.6076318925936006e-05, "loss": 2.7582, "step": 30487 }, { "epoch": 1.892606617418834, "grad_norm": 0.5844285481444246, "learning_rate": 3.6072850295973046e-05, "loss": 2.6304, "step": 30488 }, { "epoch": 1.892668694518592, "grad_norm": 0.5810081912118226, "learning_rate": 3.606938173867141e-05, "loss": 2.6423, "step": 30489 }, { "epoch": 1.89273077161835, "grad_norm": 0.585558959754798, "learning_rate": 3.6065913254049174e-05, "loss": 2.7484, "step": 30490 }, { "epoch": 1.8927928487181078, "grad_norm": 0.6309562821259072, "learning_rate": 3.6062444842124475e-05, "loss": 2.7444, "step": 30491 }, { "epoch": 1.8928549258178657, "grad_norm": 0.5934448210800501, "learning_rate": 3.6058976502915366e-05, "loss": 2.819, "step": 30492 }, { "epoch": 1.8929170029176237, "grad_norm": 0.5424274243534236, "learning_rate": 3.605550823643997e-05, "loss": 2.7415, "step": 30493 }, { "epoch": 1.8929790800173816, "grad_norm": 0.6370991465708176, "learning_rate": 3.6052040042716355e-05, "loss": 2.7838, "step": 30494 }, { "epoch": 1.8930411571171395, "grad_norm": 0.5195106846460528, "learning_rate": 3.604857192176264e-05, "loss": 2.715, "step": 30495 }, { "epoch": 1.8931032342168974, "grad_norm": 0.6231831846936143, "learning_rate": 3.604510387359693e-05, "loss": 2.7534, "step": 30496 }, { "epoch": 1.8931653113166553, "grad_norm": 0.531007823721223, "learning_rate": 3.604163589823729e-05, "loss": 2.737, "step": 30497 }, { "epoch": 1.8932273884164132, "grad_norm": 0.5694497258114588, "learning_rate": 3.6038167995701825e-05, "loss": 2.6679, "step": 30498 }, { "epoch": 1.8932894655161712, "grad_norm": 0.5611822003659084, "learning_rate": 3.603470016600863e-05, "loss": 2.7015, "step": 30499 }, { "epoch": 1.893351542615929, "grad_norm": 0.6261611642474766, "learning_rate": 3.603123240917579e-05, "loss": 2.7527, "step": 30500 }, { "epoch": 1.893413619715687, "grad_norm": 0.5312004066207203, "learning_rate": 3.60277647252214e-05, "loss": 2.7157, "step": 30501 }, { "epoch": 1.893475696815445, "grad_norm": 0.5600541238514957, "learning_rate": 3.602429711416357e-05, "loss": 2.7398, "step": 30502 }, { "epoch": 1.8935377739152028, "grad_norm": 0.6010439483184965, "learning_rate": 3.6020829576020356e-05, "loss": 2.7578, "step": 30503 }, { "epoch": 1.8935998510149605, "grad_norm": 0.5464534224496649, "learning_rate": 3.6017362110809875e-05, "loss": 2.6947, "step": 30504 }, { "epoch": 1.8936619281147185, "grad_norm": 0.5666274329469237, "learning_rate": 3.601389471855021e-05, "loss": 2.7614, "step": 30505 }, { "epoch": 1.8937240052144764, "grad_norm": 0.5582647355010185, "learning_rate": 3.601042739925944e-05, "loss": 2.637, "step": 30506 }, { "epoch": 1.8937860823142343, "grad_norm": 0.5945166356700413, "learning_rate": 3.6006960152955685e-05, "loss": 2.7419, "step": 30507 }, { "epoch": 1.8938481594139922, "grad_norm": 0.5180789180183157, "learning_rate": 3.6003492979657e-05, "loss": 2.6982, "step": 30508 }, { "epoch": 1.8939102365137501, "grad_norm": 0.6421795568415817, "learning_rate": 3.6000025879381504e-05, "loss": 2.7044, "step": 30509 }, { "epoch": 1.8939723136135078, "grad_norm": 0.5928047838045708, "learning_rate": 3.5996558852147256e-05, "loss": 2.6993, "step": 30510 }, { "epoch": 1.8940343907132657, "grad_norm": 0.6571636878400885, "learning_rate": 3.599309189797238e-05, "loss": 2.6853, "step": 30511 }, { "epoch": 1.8940964678130237, "grad_norm": 0.5664730164211553, "learning_rate": 3.5989625016874916e-05, "loss": 2.7465, "step": 30512 }, { "epoch": 1.8941585449127816, "grad_norm": 0.595871098084075, "learning_rate": 3.598615820887301e-05, "loss": 2.7304, "step": 30513 }, { "epoch": 1.8942206220125395, "grad_norm": 0.6182550300088104, "learning_rate": 3.598269147398471e-05, "loss": 2.7585, "step": 30514 }, { "epoch": 1.8942826991122974, "grad_norm": 0.6044138178149675, "learning_rate": 3.5979224812228094e-05, "loss": 2.8882, "step": 30515 }, { "epoch": 1.8943447762120553, "grad_norm": 0.6483191186781637, "learning_rate": 3.597575822362128e-05, "loss": 2.8265, "step": 30516 }, { "epoch": 1.8944068533118132, "grad_norm": 0.54159850523093, "learning_rate": 3.5972291708182336e-05, "loss": 2.7538, "step": 30517 }, { "epoch": 1.8944689304115712, "grad_norm": 0.5610970910317304, "learning_rate": 3.596882526592936e-05, "loss": 2.7593, "step": 30518 }, { "epoch": 1.894531007511329, "grad_norm": 0.5451525948320758, "learning_rate": 3.5965358896880414e-05, "loss": 2.6162, "step": 30519 }, { "epoch": 1.894593084611087, "grad_norm": 0.515733534054058, "learning_rate": 3.596189260105361e-05, "loss": 2.8074, "step": 30520 }, { "epoch": 1.894655161710845, "grad_norm": 0.5701710512514341, "learning_rate": 3.5958426378467015e-05, "loss": 2.7862, "step": 30521 }, { "epoch": 1.8947172388106028, "grad_norm": 0.5524414148814522, "learning_rate": 3.595496022913872e-05, "loss": 2.6946, "step": 30522 }, { "epoch": 1.8947793159103608, "grad_norm": 0.5566882999817431, "learning_rate": 3.59514941530868e-05, "loss": 2.6885, "step": 30523 }, { "epoch": 1.8948413930101187, "grad_norm": 0.5042897325123704, "learning_rate": 3.594802815032935e-05, "loss": 2.8108, "step": 30524 }, { "epoch": 1.8949034701098766, "grad_norm": 0.560229487399803, "learning_rate": 3.594456222088446e-05, "loss": 2.7331, "step": 30525 }, { "epoch": 1.8949655472096345, "grad_norm": 0.5098924734792653, "learning_rate": 3.594109636477019e-05, "loss": 2.8172, "step": 30526 }, { "epoch": 1.8950276243093924, "grad_norm": 0.5967062836508767, "learning_rate": 3.5937630582004614e-05, "loss": 2.6944, "step": 30527 }, { "epoch": 1.8950897014091501, "grad_norm": 0.5258258590932805, "learning_rate": 3.5934164872605855e-05, "loss": 2.7794, "step": 30528 }, { "epoch": 1.895151778508908, "grad_norm": 0.5884089418714659, "learning_rate": 3.593069923659197e-05, "loss": 2.7349, "step": 30529 }, { "epoch": 1.895213855608666, "grad_norm": 0.5466411635606246, "learning_rate": 3.592723367398104e-05, "loss": 2.6807, "step": 30530 }, { "epoch": 1.8952759327084239, "grad_norm": 0.5457899501051414, "learning_rate": 3.592376818479115e-05, "loss": 2.7304, "step": 30531 }, { "epoch": 1.8953380098081818, "grad_norm": 0.5120848881899903, "learning_rate": 3.5920302769040384e-05, "loss": 2.7569, "step": 30532 }, { "epoch": 1.8954000869079397, "grad_norm": 0.5875335318553166, "learning_rate": 3.591683742674681e-05, "loss": 2.7183, "step": 30533 }, { "epoch": 1.8954621640076974, "grad_norm": 0.5733358381414405, "learning_rate": 3.591337215792852e-05, "loss": 2.7235, "step": 30534 }, { "epoch": 1.8955242411074553, "grad_norm": 0.5699510604779594, "learning_rate": 3.5909906962603576e-05, "loss": 2.6732, "step": 30535 }, { "epoch": 1.8955863182072132, "grad_norm": 0.598406725657439, "learning_rate": 3.5906441840790075e-05, "loss": 2.6664, "step": 30536 }, { "epoch": 1.8956483953069712, "grad_norm": 0.5398891578932373, "learning_rate": 3.590297679250608e-05, "loss": 2.7359, "step": 30537 }, { "epoch": 1.895710472406729, "grad_norm": 0.5844029741170429, "learning_rate": 3.5899511817769695e-05, "loss": 2.7992, "step": 30538 }, { "epoch": 1.895772549506487, "grad_norm": 0.5706368660502583, "learning_rate": 3.589604691659898e-05, "loss": 2.5719, "step": 30539 }, { "epoch": 1.895834626606245, "grad_norm": 0.5025569089766627, "learning_rate": 3.5892582089011995e-05, "loss": 2.7142, "step": 30540 }, { "epoch": 1.8958967037060028, "grad_norm": 0.5955717598576244, "learning_rate": 3.588911733502685e-05, "loss": 2.7515, "step": 30541 }, { "epoch": 1.8959587808057607, "grad_norm": 0.4997466756911439, "learning_rate": 3.588565265466158e-05, "loss": 2.7529, "step": 30542 }, { "epoch": 1.8960208579055187, "grad_norm": 0.5237623317058827, "learning_rate": 3.588218804793432e-05, "loss": 2.6692, "step": 30543 }, { "epoch": 1.8960829350052766, "grad_norm": 0.5436134353235627, "learning_rate": 3.5878723514863084e-05, "loss": 2.7355, "step": 30544 }, { "epoch": 1.8961450121050345, "grad_norm": 0.5258957657373912, "learning_rate": 3.5875259055465994e-05, "loss": 2.7721, "step": 30545 }, { "epoch": 1.8962070892047924, "grad_norm": 0.6189364749545955, "learning_rate": 3.5871794669761096e-05, "loss": 2.7247, "step": 30546 }, { "epoch": 1.8962691663045503, "grad_norm": 0.4800828919511587, "learning_rate": 3.5868330357766475e-05, "loss": 2.6726, "step": 30547 }, { "epoch": 1.8963312434043083, "grad_norm": 0.6308507298683237, "learning_rate": 3.5864866119500204e-05, "loss": 2.7855, "step": 30548 }, { "epoch": 1.8963933205040662, "grad_norm": 0.49237364322003785, "learning_rate": 3.586140195498037e-05, "loss": 2.716, "step": 30549 }, { "epoch": 1.896455397603824, "grad_norm": 0.5666241719870805, "learning_rate": 3.5857937864225035e-05, "loss": 2.7471, "step": 30550 }, { "epoch": 1.896517474703582, "grad_norm": 0.5433670888061144, "learning_rate": 3.5854473847252254e-05, "loss": 2.7522, "step": 30551 }, { "epoch": 1.8965795518033397, "grad_norm": 0.5930142361329641, "learning_rate": 3.585100990408013e-05, "loss": 2.8149, "step": 30552 }, { "epoch": 1.8966416289030976, "grad_norm": 0.5818526230718445, "learning_rate": 3.584754603472671e-05, "loss": 2.6786, "step": 30553 }, { "epoch": 1.8967037060028555, "grad_norm": 0.5728907637086808, "learning_rate": 3.584408223921009e-05, "loss": 2.7435, "step": 30554 }, { "epoch": 1.8967657831026135, "grad_norm": 0.5582435150162731, "learning_rate": 3.584061851754832e-05, "loss": 2.6677, "step": 30555 }, { "epoch": 1.8968278602023714, "grad_norm": 0.5592327597718599, "learning_rate": 3.5837154869759496e-05, "loss": 2.7127, "step": 30556 }, { "epoch": 1.8968899373021293, "grad_norm": 0.6086913455379853, "learning_rate": 3.583369129586165e-05, "loss": 2.7646, "step": 30557 }, { "epoch": 1.896952014401887, "grad_norm": 0.5179302816936382, "learning_rate": 3.58302277958729e-05, "loss": 2.7682, "step": 30558 }, { "epoch": 1.897014091501645, "grad_norm": 0.5885748952301882, "learning_rate": 3.582676436981127e-05, "loss": 2.7207, "step": 30559 }, { "epoch": 1.8970761686014028, "grad_norm": 0.5642138703356296, "learning_rate": 3.5823301017694846e-05, "loss": 2.7426, "step": 30560 }, { "epoch": 1.8971382457011607, "grad_norm": 0.5770835259043605, "learning_rate": 3.5819837739541715e-05, "loss": 2.6812, "step": 30561 }, { "epoch": 1.8972003228009187, "grad_norm": 0.5918914668965723, "learning_rate": 3.581637453536993e-05, "loss": 2.6754, "step": 30562 }, { "epoch": 1.8972623999006766, "grad_norm": 0.6203500625806482, "learning_rate": 3.581291140519758e-05, "loss": 2.767, "step": 30563 }, { "epoch": 1.8973244770004345, "grad_norm": 0.5762106672256225, "learning_rate": 3.580944834904271e-05, "loss": 2.7533, "step": 30564 }, { "epoch": 1.8973865541001924, "grad_norm": 0.5593765482534794, "learning_rate": 3.580598536692337e-05, "loss": 2.8278, "step": 30565 }, { "epoch": 1.8974486311999503, "grad_norm": 0.5672246348096946, "learning_rate": 3.580252245885767e-05, "loss": 2.7268, "step": 30566 }, { "epoch": 1.8975107082997082, "grad_norm": 0.5004337336890519, "learning_rate": 3.5799059624863654e-05, "loss": 2.6481, "step": 30567 }, { "epoch": 1.8975727853994662, "grad_norm": 0.576116138930572, "learning_rate": 3.579559686495939e-05, "loss": 2.6745, "step": 30568 }, { "epoch": 1.897634862499224, "grad_norm": 0.5422927710274839, "learning_rate": 3.579213417916294e-05, "loss": 2.6077, "step": 30569 }, { "epoch": 1.897696939598982, "grad_norm": 0.5653546535534221, "learning_rate": 3.578867156749238e-05, "loss": 2.6723, "step": 30570 }, { "epoch": 1.89775901669874, "grad_norm": 0.588971383411582, "learning_rate": 3.5785209029965765e-05, "loss": 2.7222, "step": 30571 }, { "epoch": 1.8978210937984978, "grad_norm": 0.533847991783593, "learning_rate": 3.578174656660118e-05, "loss": 2.7223, "step": 30572 }, { "epoch": 1.8978831708982558, "grad_norm": 0.4919804637158729, "learning_rate": 3.5778284177416654e-05, "loss": 2.7301, "step": 30573 }, { "epoch": 1.8979452479980137, "grad_norm": 0.5238591389076538, "learning_rate": 3.577482186243029e-05, "loss": 2.6707, "step": 30574 }, { "epoch": 1.8980073250977716, "grad_norm": 0.5489037621124379, "learning_rate": 3.577135962166013e-05, "loss": 2.7243, "step": 30575 }, { "epoch": 1.8980694021975293, "grad_norm": 0.4980412052804197, "learning_rate": 3.576789745512423e-05, "loss": 2.6148, "step": 30576 }, { "epoch": 1.8981314792972872, "grad_norm": 0.5816113282069856, "learning_rate": 3.576443536284068e-05, "loss": 2.7293, "step": 30577 }, { "epoch": 1.8981935563970451, "grad_norm": 0.5096861678713587, "learning_rate": 3.576097334482751e-05, "loss": 2.8121, "step": 30578 }, { "epoch": 1.898255633496803, "grad_norm": 0.5718423104453868, "learning_rate": 3.575751140110282e-05, "loss": 2.7811, "step": 30579 }, { "epoch": 1.898317710596561, "grad_norm": 0.5212285989717881, "learning_rate": 3.5754049531684625e-05, "loss": 2.7869, "step": 30580 }, { "epoch": 1.8983797876963189, "grad_norm": 0.5476226872541382, "learning_rate": 3.575058773659103e-05, "loss": 2.779, "step": 30581 }, { "epoch": 1.8984418647960766, "grad_norm": 0.5106709651266036, "learning_rate": 3.5747126015840065e-05, "loss": 2.749, "step": 30582 }, { "epoch": 1.8985039418958345, "grad_norm": 0.4759315316019519, "learning_rate": 3.5743664369449815e-05, "loss": 2.8035, "step": 30583 }, { "epoch": 1.8985660189955924, "grad_norm": 0.551865396729291, "learning_rate": 3.574020279743832e-05, "loss": 2.7475, "step": 30584 }, { "epoch": 1.8986280960953503, "grad_norm": 0.49827814155630595, "learning_rate": 3.573674129982366e-05, "loss": 2.622, "step": 30585 }, { "epoch": 1.8986901731951082, "grad_norm": 0.5590059053658438, "learning_rate": 3.573327987662388e-05, "loss": 2.7073, "step": 30586 }, { "epoch": 1.8987522502948662, "grad_norm": 0.5408442532250544, "learning_rate": 3.5729818527857035e-05, "loss": 2.7325, "step": 30587 }, { "epoch": 1.898814327394624, "grad_norm": 0.5555961849005451, "learning_rate": 3.57263572535412e-05, "loss": 2.6632, "step": 30588 }, { "epoch": 1.898876404494382, "grad_norm": 0.4907069872731625, "learning_rate": 3.5722896053694413e-05, "loss": 2.787, "step": 30589 }, { "epoch": 1.89893848159414, "grad_norm": 0.5958874276643539, "learning_rate": 3.571943492833475e-05, "loss": 2.6752, "step": 30590 }, { "epoch": 1.8990005586938978, "grad_norm": 0.48804840639673147, "learning_rate": 3.571597387748026e-05, "loss": 2.6262, "step": 30591 }, { "epoch": 1.8990626357936558, "grad_norm": 0.5100797365069367, "learning_rate": 3.5712512901149e-05, "loss": 2.7156, "step": 30592 }, { "epoch": 1.8991247128934137, "grad_norm": 0.5547541994638153, "learning_rate": 3.570905199935902e-05, "loss": 2.8354, "step": 30593 }, { "epoch": 1.8991867899931716, "grad_norm": 0.5205823834162612, "learning_rate": 3.5705591172128385e-05, "loss": 2.6836, "step": 30594 }, { "epoch": 1.8992488670929295, "grad_norm": 0.5353541321607541, "learning_rate": 3.570213041947517e-05, "loss": 2.7957, "step": 30595 }, { "epoch": 1.8993109441926874, "grad_norm": 0.564672170197683, "learning_rate": 3.56986697414174e-05, "loss": 2.7227, "step": 30596 }, { "epoch": 1.8993730212924453, "grad_norm": 0.5323017312259818, "learning_rate": 3.569520913797315e-05, "loss": 2.7725, "step": 30597 }, { "epoch": 1.8994350983922033, "grad_norm": 0.5610422477572785, "learning_rate": 3.5691748609160455e-05, "loss": 2.7418, "step": 30598 }, { "epoch": 1.8994971754919612, "grad_norm": 0.5200024239318547, "learning_rate": 3.5688288154997396e-05, "loss": 2.6772, "step": 30599 }, { "epoch": 1.8995592525917189, "grad_norm": 0.5172114963093147, "learning_rate": 3.5684827775502006e-05, "loss": 2.8182, "step": 30600 }, { "epoch": 1.8996213296914768, "grad_norm": 0.5234482892191309, "learning_rate": 3.568136747069234e-05, "loss": 2.7546, "step": 30601 }, { "epoch": 1.8996834067912347, "grad_norm": 0.5671286087358531, "learning_rate": 3.567790724058646e-05, "loss": 2.6725, "step": 30602 }, { "epoch": 1.8997454838909926, "grad_norm": 0.5060865856675236, "learning_rate": 3.5674447085202414e-05, "loss": 2.6934, "step": 30603 }, { "epoch": 1.8998075609907505, "grad_norm": 0.5280215111183346, "learning_rate": 3.567098700455825e-05, "loss": 2.6371, "step": 30604 }, { "epoch": 1.8998696380905085, "grad_norm": 0.557599363025441, "learning_rate": 3.566752699867203e-05, "loss": 2.7615, "step": 30605 }, { "epoch": 1.8999317151902662, "grad_norm": 0.5711412548341086, "learning_rate": 3.56640670675618e-05, "loss": 2.7819, "step": 30606 }, { "epoch": 1.899993792290024, "grad_norm": 0.5738365042888119, "learning_rate": 3.5660607211245614e-05, "loss": 2.7204, "step": 30607 }, { "epoch": 1.900055869389782, "grad_norm": 0.5552278760962046, "learning_rate": 3.5657147429741525e-05, "loss": 2.7793, "step": 30608 }, { "epoch": 1.90011794648954, "grad_norm": 0.5742009218523444, "learning_rate": 3.5653687723067566e-05, "loss": 2.6818, "step": 30609 }, { "epoch": 1.9001800235892978, "grad_norm": 0.6105279286203397, "learning_rate": 3.565022809124181e-05, "loss": 2.7314, "step": 30610 }, { "epoch": 1.9002421006890557, "grad_norm": 0.5894829612951789, "learning_rate": 3.5646768534282306e-05, "loss": 2.8131, "step": 30611 }, { "epoch": 1.9003041777888137, "grad_norm": 0.6142331407986847, "learning_rate": 3.564330905220708e-05, "loss": 2.7377, "step": 30612 }, { "epoch": 1.9003662548885716, "grad_norm": 0.5316146552745249, "learning_rate": 3.563984964503421e-05, "loss": 2.6482, "step": 30613 }, { "epoch": 1.9004283319883295, "grad_norm": 0.6138950604947978, "learning_rate": 3.563639031278171e-05, "loss": 2.6682, "step": 30614 }, { "epoch": 1.9004904090880874, "grad_norm": 0.5413726565024705, "learning_rate": 3.563293105546767e-05, "loss": 2.7437, "step": 30615 }, { "epoch": 1.9005524861878453, "grad_norm": 0.6202388373773114, "learning_rate": 3.56294718731101e-05, "loss": 2.7558, "step": 30616 }, { "epoch": 1.9006145632876033, "grad_norm": 0.5252565687949903, "learning_rate": 3.562601276572708e-05, "loss": 2.6828, "step": 30617 }, { "epoch": 1.9006766403873612, "grad_norm": 0.5553112015845855, "learning_rate": 3.562255373333662e-05, "loss": 2.7127, "step": 30618 }, { "epoch": 1.900738717487119, "grad_norm": 0.570580954925266, "learning_rate": 3.561909477595681e-05, "loss": 2.7658, "step": 30619 }, { "epoch": 1.900800794586877, "grad_norm": 0.5828987415721593, "learning_rate": 3.561563589360566e-05, "loss": 2.7002, "step": 30620 }, { "epoch": 1.900862871686635, "grad_norm": 0.5981038729803925, "learning_rate": 3.5612177086301234e-05, "loss": 2.768, "step": 30621 }, { "epoch": 1.9009249487863928, "grad_norm": 0.5864543923696721, "learning_rate": 3.560871835406157e-05, "loss": 2.6974, "step": 30622 }, { "epoch": 1.9009870258861508, "grad_norm": 0.5249100369621543, "learning_rate": 3.5605259696904716e-05, "loss": 2.6817, "step": 30623 }, { "epoch": 1.9010491029859085, "grad_norm": 0.6254015586226084, "learning_rate": 3.560180111484872e-05, "loss": 2.728, "step": 30624 }, { "epoch": 1.9011111800856664, "grad_norm": 0.5867043118789323, "learning_rate": 3.5598342607911605e-05, "loss": 2.7107, "step": 30625 }, { "epoch": 1.9011732571854243, "grad_norm": 0.6308655635187708, "learning_rate": 3.559488417611144e-05, "loss": 2.7473, "step": 30626 }, { "epoch": 1.9012353342851822, "grad_norm": 0.515880348094655, "learning_rate": 3.559142581946628e-05, "loss": 2.6839, "step": 30627 }, { "epoch": 1.9012974113849401, "grad_norm": 0.6035351055667803, "learning_rate": 3.558796753799414e-05, "loss": 2.7367, "step": 30628 }, { "epoch": 1.901359488484698, "grad_norm": 0.506417691749756, "learning_rate": 3.558450933171307e-05, "loss": 2.5801, "step": 30629 }, { "epoch": 1.9014215655844557, "grad_norm": 0.5313245777067243, "learning_rate": 3.5581051200641116e-05, "loss": 2.7435, "step": 30630 }, { "epoch": 1.9014836426842137, "grad_norm": 0.5330926689907576, "learning_rate": 3.5577593144796315e-05, "loss": 2.6799, "step": 30631 }, { "epoch": 1.9015457197839716, "grad_norm": 0.517014279125924, "learning_rate": 3.557413516419671e-05, "loss": 2.7659, "step": 30632 }, { "epoch": 1.9016077968837295, "grad_norm": 0.5520724049572181, "learning_rate": 3.5570677258860355e-05, "loss": 2.7915, "step": 30633 }, { "epoch": 1.9016698739834874, "grad_norm": 0.5249975215762538, "learning_rate": 3.5567219428805265e-05, "loss": 2.7086, "step": 30634 }, { "epoch": 1.9017319510832453, "grad_norm": 0.5160019392672034, "learning_rate": 3.5563761674049514e-05, "loss": 2.6532, "step": 30635 }, { "epoch": 1.9017940281830032, "grad_norm": 0.6378020683530762, "learning_rate": 3.556030399461111e-05, "loss": 2.6587, "step": 30636 }, { "epoch": 1.9018561052827612, "grad_norm": 0.5163631934887248, "learning_rate": 3.5556846390508105e-05, "loss": 2.6789, "step": 30637 }, { "epoch": 1.901918182382519, "grad_norm": 0.5962100014375883, "learning_rate": 3.555338886175854e-05, "loss": 2.8239, "step": 30638 }, { "epoch": 1.901980259482277, "grad_norm": 0.6010830442398895, "learning_rate": 3.554993140838045e-05, "loss": 2.7731, "step": 30639 }, { "epoch": 1.902042336582035, "grad_norm": 0.5432611679740883, "learning_rate": 3.5546474030391886e-05, "loss": 2.7936, "step": 30640 }, { "epoch": 1.9021044136817928, "grad_norm": 0.5600246163049786, "learning_rate": 3.554301672781086e-05, "loss": 2.7477, "step": 30641 }, { "epoch": 1.9021664907815508, "grad_norm": 0.5158013129066363, "learning_rate": 3.553955950065544e-05, "loss": 2.6512, "step": 30642 }, { "epoch": 1.9022285678813087, "grad_norm": 0.5368008091371155, "learning_rate": 3.553610234894363e-05, "loss": 2.6958, "step": 30643 }, { "epoch": 1.9022906449810666, "grad_norm": 0.5450407876054596, "learning_rate": 3.553264527269351e-05, "loss": 2.6966, "step": 30644 }, { "epoch": 1.9023527220808245, "grad_norm": 0.4937999289655113, "learning_rate": 3.552918827192307e-05, "loss": 2.7056, "step": 30645 }, { "epoch": 1.9024147991805824, "grad_norm": 0.55146300955782, "learning_rate": 3.5525731346650384e-05, "loss": 2.7104, "step": 30646 }, { "epoch": 1.9024768762803403, "grad_norm": 0.4946787320466694, "learning_rate": 3.552227449689346e-05, "loss": 2.738, "step": 30647 }, { "epoch": 1.902538953380098, "grad_norm": 0.4889747425111239, "learning_rate": 3.551881772267035e-05, "loss": 2.656, "step": 30648 }, { "epoch": 1.902601030479856, "grad_norm": 0.6458370736741894, "learning_rate": 3.551536102399908e-05, "loss": 2.7315, "step": 30649 }, { "epoch": 1.9026631075796139, "grad_norm": 0.4820887384493571, "learning_rate": 3.551190440089769e-05, "loss": 2.7944, "step": 30650 }, { "epoch": 1.9027251846793718, "grad_norm": 0.5733686109527438, "learning_rate": 3.550844785338422e-05, "loss": 2.7141, "step": 30651 }, { "epoch": 1.9027872617791297, "grad_norm": 0.5846722739197991, "learning_rate": 3.5504991381476685e-05, "loss": 2.7428, "step": 30652 }, { "epoch": 1.9028493388788876, "grad_norm": 0.5223463271451472, "learning_rate": 3.550153498519313e-05, "loss": 2.7404, "step": 30653 }, { "epoch": 1.9029114159786453, "grad_norm": 0.5459681758112298, "learning_rate": 3.549807866455158e-05, "loss": 2.6981, "step": 30654 }, { "epoch": 1.9029734930784032, "grad_norm": 0.5751938876537557, "learning_rate": 3.549462241957009e-05, "loss": 2.723, "step": 30655 }, { "epoch": 1.9030355701781612, "grad_norm": 0.5370276649051321, "learning_rate": 3.549116625026666e-05, "loss": 2.7509, "step": 30656 }, { "epoch": 1.903097647277919, "grad_norm": 0.562677805666963, "learning_rate": 3.548771015665935e-05, "loss": 2.7782, "step": 30657 }, { "epoch": 1.903159724377677, "grad_norm": 0.5453914410309478, "learning_rate": 3.548425413876616e-05, "loss": 2.7809, "step": 30658 }, { "epoch": 1.903221801477435, "grad_norm": 0.6008888535193114, "learning_rate": 3.5480798196605145e-05, "loss": 2.6273, "step": 30659 }, { "epoch": 1.9032838785771928, "grad_norm": 0.5477280094950472, "learning_rate": 3.547734233019435e-05, "loss": 2.7441, "step": 30660 }, { "epoch": 1.9033459556769508, "grad_norm": 0.6489240590367595, "learning_rate": 3.547388653955178e-05, "loss": 2.7854, "step": 30661 }, { "epoch": 1.9034080327767087, "grad_norm": 0.5568423257098047, "learning_rate": 3.5470430824695456e-05, "loss": 2.7334, "step": 30662 }, { "epoch": 1.9034701098764666, "grad_norm": 0.5797906542198159, "learning_rate": 3.5466975185643435e-05, "loss": 2.7074, "step": 30663 }, { "epoch": 1.9035321869762245, "grad_norm": 0.6202890233130254, "learning_rate": 3.5463519622413724e-05, "loss": 2.7266, "step": 30664 }, { "epoch": 1.9035942640759824, "grad_norm": 0.5839363256992021, "learning_rate": 3.5460064135024375e-05, "loss": 2.8664, "step": 30665 }, { "epoch": 1.9036563411757403, "grad_norm": 0.5685844893197985, "learning_rate": 3.5456608723493386e-05, "loss": 2.749, "step": 30666 }, { "epoch": 1.9037184182754983, "grad_norm": 0.6012674574191234, "learning_rate": 3.545315338783881e-05, "loss": 2.6767, "step": 30667 }, { "epoch": 1.9037804953752562, "grad_norm": 0.65651404654561, "learning_rate": 3.5449698128078654e-05, "loss": 2.7211, "step": 30668 }, { "epoch": 1.903842572475014, "grad_norm": 0.5967609206623913, "learning_rate": 3.544624294423096e-05, "loss": 2.7364, "step": 30669 }, { "epoch": 1.903904649574772, "grad_norm": 0.6016231853046389, "learning_rate": 3.544278783631375e-05, "loss": 2.7359, "step": 30670 }, { "epoch": 1.90396672667453, "grad_norm": 0.6148872486126784, "learning_rate": 3.543933280434505e-05, "loss": 2.7082, "step": 30671 }, { "epoch": 1.9040288037742876, "grad_norm": 0.6253816393100071, "learning_rate": 3.543587784834288e-05, "loss": 2.7653, "step": 30672 }, { "epoch": 1.9040908808740455, "grad_norm": 0.5838489439223378, "learning_rate": 3.543242296832528e-05, "loss": 2.6234, "step": 30673 }, { "epoch": 1.9041529579738035, "grad_norm": 0.6289295642393447, "learning_rate": 3.5428968164310265e-05, "loss": 2.7735, "step": 30674 }, { "epoch": 1.9042150350735614, "grad_norm": 0.5754125732047924, "learning_rate": 3.542551343631585e-05, "loss": 2.7555, "step": 30675 }, { "epoch": 1.9042771121733193, "grad_norm": 0.5404986724554903, "learning_rate": 3.5422058784360084e-05, "loss": 2.6742, "step": 30676 }, { "epoch": 1.9043391892730772, "grad_norm": 0.6107396471283637, "learning_rate": 3.541860420846096e-05, "loss": 2.7543, "step": 30677 }, { "epoch": 1.904401266372835, "grad_norm": 0.5006946863828449, "learning_rate": 3.541514970863653e-05, "loss": 2.7011, "step": 30678 }, { "epoch": 1.9044633434725928, "grad_norm": 0.6043190407710205, "learning_rate": 3.54116952849048e-05, "loss": 2.6379, "step": 30679 }, { "epoch": 1.9045254205723507, "grad_norm": 0.5447317359106925, "learning_rate": 3.54082409372838e-05, "loss": 2.7451, "step": 30680 }, { "epoch": 1.9045874976721087, "grad_norm": 0.5470137556340978, "learning_rate": 3.5404786665791544e-05, "loss": 2.7336, "step": 30681 }, { "epoch": 1.9046495747718666, "grad_norm": 0.5758376105582582, "learning_rate": 3.540133247044607e-05, "loss": 2.6921, "step": 30682 }, { "epoch": 1.9047116518716245, "grad_norm": 0.5896351931873423, "learning_rate": 3.5397878351265365e-05, "loss": 2.7214, "step": 30683 }, { "epoch": 1.9047737289713824, "grad_norm": 0.6007089628426352, "learning_rate": 3.53944243082675e-05, "loss": 2.7833, "step": 30684 }, { "epoch": 1.9048358060711403, "grad_norm": 0.527510524798432, "learning_rate": 3.539097034147046e-05, "loss": 2.8011, "step": 30685 }, { "epoch": 1.9048978831708983, "grad_norm": 0.6027529937973575, "learning_rate": 3.5387516450892263e-05, "loss": 2.7162, "step": 30686 }, { "epoch": 1.9049599602706562, "grad_norm": 0.5364459292511661, "learning_rate": 3.538406263655096e-05, "loss": 2.7742, "step": 30687 }, { "epoch": 1.905022037370414, "grad_norm": 0.5474139418965598, "learning_rate": 3.538060889846454e-05, "loss": 2.7536, "step": 30688 }, { "epoch": 1.905084114470172, "grad_norm": 0.5497974239431945, "learning_rate": 3.537715523665103e-05, "loss": 2.6455, "step": 30689 }, { "epoch": 1.90514619156993, "grad_norm": 0.5477491748895877, "learning_rate": 3.537370165112845e-05, "loss": 2.7964, "step": 30690 }, { "epoch": 1.9052082686696878, "grad_norm": 0.5044577190632477, "learning_rate": 3.537024814191481e-05, "loss": 2.7384, "step": 30691 }, { "epoch": 1.9052703457694458, "grad_norm": 0.5487200538962583, "learning_rate": 3.536679470902815e-05, "loss": 2.7456, "step": 30692 }, { "epoch": 1.9053324228692037, "grad_norm": 0.4859105964400991, "learning_rate": 3.5363341352486465e-05, "loss": 2.7634, "step": 30693 }, { "epoch": 1.9053944999689616, "grad_norm": 0.5136856541399125, "learning_rate": 3.5359888072307795e-05, "loss": 2.7021, "step": 30694 }, { "epoch": 1.9054565770687195, "grad_norm": 0.5023044885269651, "learning_rate": 3.535643486851014e-05, "loss": 2.7951, "step": 30695 }, { "epoch": 1.9055186541684772, "grad_norm": 0.5105888180187805, "learning_rate": 3.535298174111151e-05, "loss": 2.8694, "step": 30696 }, { "epoch": 1.9055807312682351, "grad_norm": 0.5571909543121287, "learning_rate": 3.5349528690129934e-05, "loss": 2.6818, "step": 30697 }, { "epoch": 1.905642808367993, "grad_norm": 0.5494761594958225, "learning_rate": 3.534607571558343e-05, "loss": 2.764, "step": 30698 }, { "epoch": 1.905704885467751, "grad_norm": 0.5121559430328088, "learning_rate": 3.534262281749001e-05, "loss": 2.704, "step": 30699 }, { "epoch": 1.9057669625675089, "grad_norm": 0.5702213057674336, "learning_rate": 3.533916999586767e-05, "loss": 2.7023, "step": 30700 }, { "epoch": 1.9058290396672668, "grad_norm": 0.4766592646682113, "learning_rate": 3.533571725073446e-05, "loss": 2.6908, "step": 30701 }, { "epoch": 1.9058911167670245, "grad_norm": 0.5199096276999619, "learning_rate": 3.533226458210835e-05, "loss": 2.7969, "step": 30702 }, { "epoch": 1.9059531938667824, "grad_norm": 0.5005958106666929, "learning_rate": 3.532881199000739e-05, "loss": 2.7033, "step": 30703 }, { "epoch": 1.9060152709665403, "grad_norm": 0.5210843784570495, "learning_rate": 3.532535947444957e-05, "loss": 2.7634, "step": 30704 }, { "epoch": 1.9060773480662982, "grad_norm": 0.5545120767786584, "learning_rate": 3.5321907035452914e-05, "loss": 2.7347, "step": 30705 }, { "epoch": 1.9061394251660562, "grad_norm": 0.46905993705522026, "learning_rate": 3.531845467303544e-05, "loss": 2.6852, "step": 30706 }, { "epoch": 1.906201502265814, "grad_norm": 0.5605076633256175, "learning_rate": 3.5315002387215144e-05, "loss": 2.7661, "step": 30707 }, { "epoch": 1.906263579365572, "grad_norm": 0.5587676784526397, "learning_rate": 3.5311550178010044e-05, "loss": 2.7192, "step": 30708 }, { "epoch": 1.90632565646533, "grad_norm": 0.5132622191270179, "learning_rate": 3.5308098045438164e-05, "loss": 2.6497, "step": 30709 }, { "epoch": 1.9063877335650878, "grad_norm": 0.5763899568243998, "learning_rate": 3.5304645989517506e-05, "loss": 2.7288, "step": 30710 }, { "epoch": 1.9064498106648458, "grad_norm": 0.6002182990556227, "learning_rate": 3.5301194010266057e-05, "loss": 2.7271, "step": 30711 }, { "epoch": 1.9065118877646037, "grad_norm": 0.5944521416505463, "learning_rate": 3.529774210770187e-05, "loss": 2.7361, "step": 30712 }, { "epoch": 1.9065739648643616, "grad_norm": 0.6027201375806387, "learning_rate": 3.529429028184291e-05, "loss": 2.7016, "step": 30713 }, { "epoch": 1.9066360419641195, "grad_norm": 0.5002416384454259, "learning_rate": 3.529083853270723e-05, "loss": 2.7089, "step": 30714 }, { "epoch": 1.9066981190638774, "grad_norm": 0.5910508643900187, "learning_rate": 3.528738686031279e-05, "loss": 2.7039, "step": 30715 }, { "epoch": 1.9067601961636353, "grad_norm": 0.5516888600703564, "learning_rate": 3.528393526467764e-05, "loss": 2.6629, "step": 30716 }, { "epoch": 1.9068222732633933, "grad_norm": 0.5580115852588924, "learning_rate": 3.528048374581977e-05, "loss": 2.7691, "step": 30717 }, { "epoch": 1.9068843503631512, "grad_norm": 0.5665935338269075, "learning_rate": 3.527703230375719e-05, "loss": 2.72, "step": 30718 }, { "epoch": 1.906946427462909, "grad_norm": 0.5388877419522895, "learning_rate": 3.52735809385079e-05, "loss": 2.6983, "step": 30719 }, { "epoch": 1.9070085045626668, "grad_norm": 0.5798424392886858, "learning_rate": 3.5270129650089926e-05, "loss": 2.8012, "step": 30720 }, { "epoch": 1.9070705816624247, "grad_norm": 0.5867328078291264, "learning_rate": 3.526667843852126e-05, "loss": 2.7597, "step": 30721 }, { "epoch": 1.9071326587621826, "grad_norm": 0.5744463151776892, "learning_rate": 3.5263227303819895e-05, "loss": 2.7813, "step": 30722 }, { "epoch": 1.9071947358619405, "grad_norm": 0.5651635052397462, "learning_rate": 3.5259776246003865e-05, "loss": 2.7045, "step": 30723 }, { "epoch": 1.9072568129616985, "grad_norm": 0.4978206494724539, "learning_rate": 3.5256325265091136e-05, "loss": 2.7054, "step": 30724 }, { "epoch": 1.9073188900614562, "grad_norm": 0.6735019723782619, "learning_rate": 3.525287436109975e-05, "loss": 2.8622, "step": 30725 }, { "epoch": 1.907380967161214, "grad_norm": 0.5645777686807906, "learning_rate": 3.524942353404772e-05, "loss": 2.6667, "step": 30726 }, { "epoch": 1.907443044260972, "grad_norm": 0.5713856150881016, "learning_rate": 3.5245972783953e-05, "loss": 2.6881, "step": 30727 }, { "epoch": 1.90750512136073, "grad_norm": 0.5110004421277009, "learning_rate": 3.5242522110833645e-05, "loss": 2.6485, "step": 30728 }, { "epoch": 1.9075671984604878, "grad_norm": 0.6583486786595067, "learning_rate": 3.5239071514707614e-05, "loss": 2.7513, "step": 30729 }, { "epoch": 1.9076292755602458, "grad_norm": 0.4838460679979634, "learning_rate": 3.5235620995592945e-05, "loss": 2.7541, "step": 30730 }, { "epoch": 1.9076913526600037, "grad_norm": 0.7019091159820979, "learning_rate": 3.5232170553507614e-05, "loss": 2.7036, "step": 30731 }, { "epoch": 1.9077534297597616, "grad_norm": 0.4869281942716874, "learning_rate": 3.522872018846965e-05, "loss": 2.6606, "step": 30732 }, { "epoch": 1.9078155068595195, "grad_norm": 0.5673020481005405, "learning_rate": 3.522526990049703e-05, "loss": 2.6128, "step": 30733 }, { "epoch": 1.9078775839592774, "grad_norm": 0.5252411924093086, "learning_rate": 3.522181968960776e-05, "loss": 2.7167, "step": 30734 }, { "epoch": 1.9079396610590353, "grad_norm": 0.5391984021910954, "learning_rate": 3.521836955581985e-05, "loss": 2.7301, "step": 30735 }, { "epoch": 1.9080017381587933, "grad_norm": 0.5100983451005766, "learning_rate": 3.521491949915129e-05, "loss": 2.7498, "step": 30736 }, { "epoch": 1.9080638152585512, "grad_norm": 0.5915850521490233, "learning_rate": 3.521146951962009e-05, "loss": 2.783, "step": 30737 }, { "epoch": 1.908125892358309, "grad_norm": 0.4919902618237253, "learning_rate": 3.5208019617244236e-05, "loss": 2.7053, "step": 30738 }, { "epoch": 1.908187969458067, "grad_norm": 0.545887218489717, "learning_rate": 3.520456979204174e-05, "loss": 2.7232, "step": 30739 }, { "epoch": 1.908250046557825, "grad_norm": 0.5102015126173386, "learning_rate": 3.520112004403059e-05, "loss": 2.6405, "step": 30740 }, { "epoch": 1.9083121236575828, "grad_norm": 0.5222501810967408, "learning_rate": 3.51976703732288e-05, "loss": 2.7044, "step": 30741 }, { "epoch": 1.9083742007573408, "grad_norm": 0.5496797029984307, "learning_rate": 3.519422077965434e-05, "loss": 2.7079, "step": 30742 }, { "epoch": 1.9084362778570985, "grad_norm": 0.5016216564052773, "learning_rate": 3.519077126332524e-05, "loss": 2.6943, "step": 30743 }, { "epoch": 1.9084983549568564, "grad_norm": 0.5425481745429667, "learning_rate": 3.5187321824259476e-05, "loss": 2.7077, "step": 30744 }, { "epoch": 1.9085604320566143, "grad_norm": 0.5692922532873066, "learning_rate": 3.518387246247505e-05, "loss": 2.7094, "step": 30745 }, { "epoch": 1.9086225091563722, "grad_norm": 0.50710476193984, "learning_rate": 3.518042317798996e-05, "loss": 2.6988, "step": 30746 }, { "epoch": 1.9086845862561301, "grad_norm": 0.538611292319584, "learning_rate": 3.517697397082219e-05, "loss": 2.7688, "step": 30747 }, { "epoch": 1.908746663355888, "grad_norm": 0.5306815584322082, "learning_rate": 3.5173524840989764e-05, "loss": 2.7216, "step": 30748 }, { "epoch": 1.9088087404556457, "grad_norm": 0.5548301298488832, "learning_rate": 3.517007578851064e-05, "loss": 2.6658, "step": 30749 }, { "epoch": 1.9088708175554037, "grad_norm": 0.5240449938639544, "learning_rate": 3.516662681340284e-05, "loss": 2.7506, "step": 30750 }, { "epoch": 1.9089328946551616, "grad_norm": 0.5421903230884941, "learning_rate": 3.5163177915684344e-05, "loss": 2.7545, "step": 30751 }, { "epoch": 1.9089949717549195, "grad_norm": 0.5502620982842327, "learning_rate": 3.515972909537316e-05, "loss": 2.8175, "step": 30752 }, { "epoch": 1.9090570488546774, "grad_norm": 0.5613335329776161, "learning_rate": 3.515628035248726e-05, "loss": 2.7619, "step": 30753 }, { "epoch": 1.9091191259544353, "grad_norm": 0.5763596826709146, "learning_rate": 3.515283168704465e-05, "loss": 2.7472, "step": 30754 }, { "epoch": 1.9091812030541933, "grad_norm": 0.553172716140808, "learning_rate": 3.514938309906333e-05, "loss": 2.7652, "step": 30755 }, { "epoch": 1.9092432801539512, "grad_norm": 0.550243927226684, "learning_rate": 3.5145934588561255e-05, "loss": 2.7166, "step": 30756 }, { "epoch": 1.909305357253709, "grad_norm": 0.5778276348865466, "learning_rate": 3.5142486155556475e-05, "loss": 2.6597, "step": 30757 }, { "epoch": 1.909367434353467, "grad_norm": 0.5633092463251352, "learning_rate": 3.513903780006694e-05, "loss": 2.7136, "step": 30758 }, { "epoch": 1.909429511453225, "grad_norm": 0.5887109603006053, "learning_rate": 3.513558952211066e-05, "loss": 2.772, "step": 30759 }, { "epoch": 1.9094915885529828, "grad_norm": 0.5757399152103584, "learning_rate": 3.513214132170563e-05, "loss": 2.6957, "step": 30760 }, { "epoch": 1.9095536656527408, "grad_norm": 0.5464297525676282, "learning_rate": 3.51286931988698e-05, "loss": 2.7029, "step": 30761 }, { "epoch": 1.9096157427524987, "grad_norm": 0.5481994666621284, "learning_rate": 3.512524515362121e-05, "loss": 2.7447, "step": 30762 }, { "epoch": 1.9096778198522566, "grad_norm": 0.5738996280710836, "learning_rate": 3.512179718597781e-05, "loss": 2.7038, "step": 30763 }, { "epoch": 1.9097398969520145, "grad_norm": 0.5130262487548454, "learning_rate": 3.511834929595762e-05, "loss": 2.7301, "step": 30764 }, { "epoch": 1.9098019740517724, "grad_norm": 0.5851056615660848, "learning_rate": 3.51149014835786e-05, "loss": 2.7848, "step": 30765 }, { "epoch": 1.9098640511515304, "grad_norm": 0.5477987661778154, "learning_rate": 3.511145374885877e-05, "loss": 2.6905, "step": 30766 }, { "epoch": 1.909926128251288, "grad_norm": 0.5019554405307426, "learning_rate": 3.510800609181608e-05, "loss": 2.7595, "step": 30767 }, { "epoch": 1.909988205351046, "grad_norm": 0.6289079085618415, "learning_rate": 3.5104558512468554e-05, "loss": 2.7626, "step": 30768 }, { "epoch": 1.9100502824508039, "grad_norm": 0.5851766351018983, "learning_rate": 3.510111101083415e-05, "loss": 2.6847, "step": 30769 }, { "epoch": 1.9101123595505618, "grad_norm": 0.5975739914876697, "learning_rate": 3.509766358693089e-05, "loss": 2.7737, "step": 30770 }, { "epoch": 1.9101744366503197, "grad_norm": 0.5875292963474618, "learning_rate": 3.509421624077672e-05, "loss": 2.7207, "step": 30771 }, { "epoch": 1.9102365137500776, "grad_norm": 0.5400107515995484, "learning_rate": 3.5090768972389634e-05, "loss": 2.7278, "step": 30772 }, { "epoch": 1.9102985908498353, "grad_norm": 0.5788151513621694, "learning_rate": 3.508732178178764e-05, "loss": 2.7275, "step": 30773 }, { "epoch": 1.9103606679495932, "grad_norm": 0.6607105027646497, "learning_rate": 3.5083874668988695e-05, "loss": 2.7054, "step": 30774 }, { "epoch": 1.9104227450493512, "grad_norm": 0.5619229164526508, "learning_rate": 3.508042763401081e-05, "loss": 2.7937, "step": 30775 }, { "epoch": 1.910484822149109, "grad_norm": 0.5935370414699177, "learning_rate": 3.507698067687195e-05, "loss": 2.7494, "step": 30776 }, { "epoch": 1.910546899248867, "grad_norm": 0.5773954943851433, "learning_rate": 3.507353379759011e-05, "loss": 2.7218, "step": 30777 }, { "epoch": 1.910608976348625, "grad_norm": 0.5908899943977346, "learning_rate": 3.5070086996183255e-05, "loss": 2.7885, "step": 30778 }, { "epoch": 1.9106710534483828, "grad_norm": 0.5167237625824829, "learning_rate": 3.5066640272669396e-05, "loss": 2.7103, "step": 30779 }, { "epoch": 1.9107331305481408, "grad_norm": 0.6006295041045409, "learning_rate": 3.506319362706648e-05, "loss": 2.7815, "step": 30780 }, { "epoch": 1.9107952076478987, "grad_norm": 0.514484154903031, "learning_rate": 3.505974705939253e-05, "loss": 2.7293, "step": 30781 }, { "epoch": 1.9108572847476566, "grad_norm": 0.5533444976637859, "learning_rate": 3.50563005696655e-05, "loss": 2.6995, "step": 30782 }, { "epoch": 1.9109193618474145, "grad_norm": 0.5116134418717012, "learning_rate": 3.505285415790337e-05, "loss": 2.7055, "step": 30783 }, { "epoch": 1.9109814389471724, "grad_norm": 0.5883807550298366, "learning_rate": 3.504940782412414e-05, "loss": 2.7357, "step": 30784 }, { "epoch": 1.9110435160469303, "grad_norm": 0.5316254467167381, "learning_rate": 3.504596156834576e-05, "loss": 2.6698, "step": 30785 }, { "epoch": 1.9111055931466883, "grad_norm": 0.5822761583623083, "learning_rate": 3.504251539058625e-05, "loss": 2.8223, "step": 30786 }, { "epoch": 1.9111676702464462, "grad_norm": 0.6516953175149844, "learning_rate": 3.5039069290863555e-05, "loss": 2.7477, "step": 30787 }, { "epoch": 1.911229747346204, "grad_norm": 0.5193276544192279, "learning_rate": 3.503562326919568e-05, "loss": 2.7522, "step": 30788 }, { "epoch": 1.911291824445962, "grad_norm": 0.5464814813849495, "learning_rate": 3.503217732560056e-05, "loss": 2.7369, "step": 30789 }, { "epoch": 1.91135390154572, "grad_norm": 0.5826221768929343, "learning_rate": 3.5028731460096225e-05, "loss": 2.7817, "step": 30790 }, { "epoch": 1.9114159786454776, "grad_norm": 0.5202706317266024, "learning_rate": 3.5025285672700635e-05, "loss": 2.6517, "step": 30791 }, { "epoch": 1.9114780557452355, "grad_norm": 0.5612781021881995, "learning_rate": 3.5021839963431765e-05, "loss": 2.7376, "step": 30792 }, { "epoch": 1.9115401328449935, "grad_norm": 0.5878631483988619, "learning_rate": 3.5018394332307593e-05, "loss": 2.7418, "step": 30793 }, { "epoch": 1.9116022099447514, "grad_norm": 0.5305943369907947, "learning_rate": 3.501494877934609e-05, "loss": 2.7038, "step": 30794 }, { "epoch": 1.9116642870445093, "grad_norm": 0.5242945074835983, "learning_rate": 3.501150330456524e-05, "loss": 2.6507, "step": 30795 }, { "epoch": 1.9117263641442672, "grad_norm": 0.5710322136869531, "learning_rate": 3.500805790798303e-05, "loss": 2.7463, "step": 30796 }, { "epoch": 1.911788441244025, "grad_norm": 0.5298098228794658, "learning_rate": 3.50046125896174e-05, "loss": 2.74, "step": 30797 }, { "epoch": 1.9118505183437828, "grad_norm": 0.5814447456129013, "learning_rate": 3.500116734948635e-05, "loss": 2.7706, "step": 30798 }, { "epoch": 1.9119125954435408, "grad_norm": 0.5362550078560562, "learning_rate": 3.499772218760785e-05, "loss": 2.7862, "step": 30799 }, { "epoch": 1.9119746725432987, "grad_norm": 0.5090755307828791, "learning_rate": 3.499427710399989e-05, "loss": 2.6666, "step": 30800 }, { "epoch": 1.9120367496430566, "grad_norm": 0.5433729300326806, "learning_rate": 3.4990832098680414e-05, "loss": 2.723, "step": 30801 }, { "epoch": 1.9120988267428145, "grad_norm": 0.5665087113935342, "learning_rate": 3.4987387171667416e-05, "loss": 2.6601, "step": 30802 }, { "epoch": 1.9121609038425724, "grad_norm": 0.5210418824057251, "learning_rate": 3.498394232297886e-05, "loss": 2.6837, "step": 30803 }, { "epoch": 1.9122229809423303, "grad_norm": 0.5298590683248441, "learning_rate": 3.498049755263273e-05, "loss": 2.6756, "step": 30804 }, { "epoch": 1.9122850580420883, "grad_norm": 0.49785930617046775, "learning_rate": 3.497705286064698e-05, "loss": 2.6818, "step": 30805 }, { "epoch": 1.9123471351418462, "grad_norm": 0.4998026547794962, "learning_rate": 3.49736082470396e-05, "loss": 2.6371, "step": 30806 }, { "epoch": 1.912409212241604, "grad_norm": 0.5162471428023145, "learning_rate": 3.4970163711828564e-05, "loss": 2.7088, "step": 30807 }, { "epoch": 1.912471289341362, "grad_norm": 0.5162194899717618, "learning_rate": 3.496671925503181e-05, "loss": 2.6744, "step": 30808 }, { "epoch": 1.91253336644112, "grad_norm": 0.5618003626760125, "learning_rate": 3.4963274876667344e-05, "loss": 2.7256, "step": 30809 }, { "epoch": 1.9125954435408778, "grad_norm": 0.5585350859454865, "learning_rate": 3.4959830576753115e-05, "loss": 2.7753, "step": 30810 }, { "epoch": 1.9126575206406358, "grad_norm": 0.5828040196515124, "learning_rate": 3.495638635530711e-05, "loss": 2.6501, "step": 30811 }, { "epoch": 1.9127195977403937, "grad_norm": 0.6296553563321742, "learning_rate": 3.495294221234728e-05, "loss": 2.7131, "step": 30812 }, { "epoch": 1.9127816748401516, "grad_norm": 0.557619458943833, "learning_rate": 3.494949814789161e-05, "loss": 2.7437, "step": 30813 }, { "epoch": 1.9128437519399095, "grad_norm": 0.5734620070872527, "learning_rate": 3.494605416195806e-05, "loss": 2.7359, "step": 30814 }, { "epoch": 1.9129058290396672, "grad_norm": 0.5545687212501026, "learning_rate": 3.4942610254564606e-05, "loss": 2.7228, "step": 30815 }, { "epoch": 1.9129679061394251, "grad_norm": 0.5141632572671057, "learning_rate": 3.4939166425729194e-05, "loss": 2.7213, "step": 30816 }, { "epoch": 1.913029983239183, "grad_norm": 0.6106023166546948, "learning_rate": 3.493572267546983e-05, "loss": 2.6954, "step": 30817 }, { "epoch": 1.913092060338941, "grad_norm": 0.48362070233448595, "learning_rate": 3.493227900380444e-05, "loss": 2.6719, "step": 30818 }, { "epoch": 1.9131541374386989, "grad_norm": 0.5426450903635005, "learning_rate": 3.4928835410751015e-05, "loss": 2.6999, "step": 30819 }, { "epoch": 1.9132162145384568, "grad_norm": 0.5408254902138219, "learning_rate": 3.492539189632752e-05, "loss": 2.7412, "step": 30820 }, { "epoch": 1.9132782916382145, "grad_norm": 0.48077815258477635, "learning_rate": 3.49219484605519e-05, "loss": 2.7464, "step": 30821 }, { "epoch": 1.9133403687379724, "grad_norm": 0.5163055033092532, "learning_rate": 3.4918505103442125e-05, "loss": 2.7861, "step": 30822 }, { "epoch": 1.9134024458377303, "grad_norm": 0.5420123397242403, "learning_rate": 3.4915061825016197e-05, "loss": 2.797, "step": 30823 }, { "epoch": 1.9134645229374883, "grad_norm": 0.5331528189266824, "learning_rate": 3.4911618625292036e-05, "loss": 2.7131, "step": 30824 }, { "epoch": 1.9135266000372462, "grad_norm": 0.5117550611352857, "learning_rate": 3.4908175504287635e-05, "loss": 2.6746, "step": 30825 }, { "epoch": 1.913588677137004, "grad_norm": 0.5547205149536697, "learning_rate": 3.490473246202094e-05, "loss": 2.7036, "step": 30826 }, { "epoch": 1.913650754236762, "grad_norm": 0.5091106255288854, "learning_rate": 3.4901289498509925e-05, "loss": 2.8182, "step": 30827 }, { "epoch": 1.91371283133652, "grad_norm": 0.5862094756559563, "learning_rate": 3.4897846613772544e-05, "loss": 2.6004, "step": 30828 }, { "epoch": 1.9137749084362778, "grad_norm": 0.5518250551012799, "learning_rate": 3.489440380782677e-05, "loss": 2.8274, "step": 30829 }, { "epoch": 1.9138369855360358, "grad_norm": 0.547461685468671, "learning_rate": 3.489096108069054e-05, "loss": 2.7368, "step": 30830 }, { "epoch": 1.9138990626357937, "grad_norm": 0.548292394064413, "learning_rate": 3.488751843238186e-05, "loss": 2.7178, "step": 30831 }, { "epoch": 1.9139611397355516, "grad_norm": 0.5405973063525665, "learning_rate": 3.4884075862918656e-05, "loss": 2.6827, "step": 30832 }, { "epoch": 1.9140232168353095, "grad_norm": 0.5413403414507908, "learning_rate": 3.488063337231889e-05, "loss": 2.7536, "step": 30833 }, { "epoch": 1.9140852939350674, "grad_norm": 0.5711954154181886, "learning_rate": 3.4877190960600534e-05, "loss": 2.7072, "step": 30834 }, { "epoch": 1.9141473710348254, "grad_norm": 0.5419462992169705, "learning_rate": 3.487374862778154e-05, "loss": 2.6313, "step": 30835 }, { "epoch": 1.9142094481345833, "grad_norm": 0.5570831413763263, "learning_rate": 3.487030637387988e-05, "loss": 2.7003, "step": 30836 }, { "epoch": 1.9142715252343412, "grad_norm": 0.5757469032426273, "learning_rate": 3.4866864198913495e-05, "loss": 2.659, "step": 30837 }, { "epoch": 1.914333602334099, "grad_norm": 0.5243399796823519, "learning_rate": 3.486342210290037e-05, "loss": 2.6717, "step": 30838 }, { "epoch": 1.9143956794338568, "grad_norm": 0.6279402475373833, "learning_rate": 3.4859980085858424e-05, "loss": 2.7287, "step": 30839 }, { "epoch": 1.9144577565336147, "grad_norm": 0.5715603176302617, "learning_rate": 3.4856538147805655e-05, "loss": 2.8455, "step": 30840 }, { "epoch": 1.9145198336333726, "grad_norm": 0.5339411563866427, "learning_rate": 3.485309628875999e-05, "loss": 2.7276, "step": 30841 }, { "epoch": 1.9145819107331306, "grad_norm": 0.5857468358409048, "learning_rate": 3.484965450873941e-05, "loss": 2.6748, "step": 30842 }, { "epoch": 1.9146439878328885, "grad_norm": 0.5306387077061695, "learning_rate": 3.484621280776185e-05, "loss": 2.7004, "step": 30843 }, { "epoch": 1.9147060649326464, "grad_norm": 0.5914179026011306, "learning_rate": 3.484277118584529e-05, "loss": 2.7719, "step": 30844 }, { "epoch": 1.914768142032404, "grad_norm": 0.508452525652456, "learning_rate": 3.483932964300767e-05, "loss": 2.787, "step": 30845 }, { "epoch": 1.914830219132162, "grad_norm": 0.5945037023350423, "learning_rate": 3.483588817926693e-05, "loss": 2.7328, "step": 30846 }, { "epoch": 1.91489229623192, "grad_norm": 0.5449534034495322, "learning_rate": 3.483244679464106e-05, "loss": 2.6986, "step": 30847 }, { "epoch": 1.9149543733316778, "grad_norm": 0.5908555958198399, "learning_rate": 3.482900548914798e-05, "loss": 2.6971, "step": 30848 }, { "epoch": 1.9150164504314358, "grad_norm": 0.5764046926004285, "learning_rate": 3.482556426280568e-05, "loss": 2.7969, "step": 30849 }, { "epoch": 1.9150785275311937, "grad_norm": 0.5689203174992273, "learning_rate": 3.482212311563208e-05, "loss": 2.8102, "step": 30850 }, { "epoch": 1.9151406046309516, "grad_norm": 0.5620908838853187, "learning_rate": 3.481868204764516e-05, "loss": 2.7027, "step": 30851 }, { "epoch": 1.9152026817307095, "grad_norm": 0.5565149265714087, "learning_rate": 3.481524105886286e-05, "loss": 2.6791, "step": 30852 }, { "epoch": 1.9152647588304674, "grad_norm": 0.5660121993967546, "learning_rate": 3.4811800149303134e-05, "loss": 2.7183, "step": 30853 }, { "epoch": 1.9153268359302253, "grad_norm": 0.5344398046539083, "learning_rate": 3.480835931898392e-05, "loss": 2.7466, "step": 30854 }, { "epoch": 1.9153889130299833, "grad_norm": 0.523363592139663, "learning_rate": 3.4804918567923176e-05, "loss": 2.6756, "step": 30855 }, { "epoch": 1.9154509901297412, "grad_norm": 0.6176970587522203, "learning_rate": 3.480147789613888e-05, "loss": 2.8225, "step": 30856 }, { "epoch": 1.915513067229499, "grad_norm": 0.5162645477388053, "learning_rate": 3.479803730364897e-05, "loss": 2.7181, "step": 30857 }, { "epoch": 1.915575144329257, "grad_norm": 0.5275771578182945, "learning_rate": 3.4794596790471377e-05, "loss": 2.7381, "step": 30858 }, { "epoch": 1.915637221429015, "grad_norm": 0.575905871006628, "learning_rate": 3.479115635662407e-05, "loss": 2.7425, "step": 30859 }, { "epoch": 1.9156992985287729, "grad_norm": 0.5287301664938814, "learning_rate": 3.4787716002124985e-05, "loss": 2.822, "step": 30860 }, { "epoch": 1.9157613756285308, "grad_norm": 0.5897257352076384, "learning_rate": 3.4784275726992085e-05, "loss": 2.685, "step": 30861 }, { "epoch": 1.9158234527282887, "grad_norm": 0.5748524704624512, "learning_rate": 3.47808355312433e-05, "loss": 2.7295, "step": 30862 }, { "epoch": 1.9158855298280464, "grad_norm": 0.6715843104126478, "learning_rate": 3.4777395414896616e-05, "loss": 2.6893, "step": 30863 }, { "epoch": 1.9159476069278043, "grad_norm": 0.583346423501779, "learning_rate": 3.477395537796993e-05, "loss": 2.7155, "step": 30864 }, { "epoch": 1.9160096840275622, "grad_norm": 0.6146727162897474, "learning_rate": 3.477051542048124e-05, "loss": 2.7574, "step": 30865 }, { "epoch": 1.9160717611273201, "grad_norm": 0.5828284621929193, "learning_rate": 3.4767075542448444e-05, "loss": 2.6796, "step": 30866 }, { "epoch": 1.916133838227078, "grad_norm": 0.5744385071721657, "learning_rate": 3.4763635743889524e-05, "loss": 2.7884, "step": 30867 }, { "epoch": 1.916195915326836, "grad_norm": 0.6144430733238697, "learning_rate": 3.476019602482241e-05, "loss": 2.6803, "step": 30868 }, { "epoch": 1.9162579924265937, "grad_norm": 0.5883301520824425, "learning_rate": 3.475675638526507e-05, "loss": 2.6811, "step": 30869 }, { "epoch": 1.9163200695263516, "grad_norm": 0.5756942062111569, "learning_rate": 3.475331682523542e-05, "loss": 2.6985, "step": 30870 }, { "epoch": 1.9163821466261095, "grad_norm": 0.612307121113834, "learning_rate": 3.4749877344751416e-05, "loss": 2.7326, "step": 30871 }, { "epoch": 1.9164442237258674, "grad_norm": 0.574181380320996, "learning_rate": 3.4746437943831e-05, "loss": 2.7675, "step": 30872 }, { "epoch": 1.9165063008256253, "grad_norm": 0.5190901066502559, "learning_rate": 3.4742998622492127e-05, "loss": 2.6332, "step": 30873 }, { "epoch": 1.9165683779253833, "grad_norm": 0.5151060266978666, "learning_rate": 3.473955938075273e-05, "loss": 2.7231, "step": 30874 }, { "epoch": 1.9166304550251412, "grad_norm": 0.5443573628928797, "learning_rate": 3.473612021863075e-05, "loss": 2.7615, "step": 30875 }, { "epoch": 1.916692532124899, "grad_norm": 0.4879297898930482, "learning_rate": 3.473268113614415e-05, "loss": 2.6839, "step": 30876 }, { "epoch": 1.916754609224657, "grad_norm": 0.5284869764955509, "learning_rate": 3.472924213331086e-05, "loss": 2.7931, "step": 30877 }, { "epoch": 1.916816686324415, "grad_norm": 0.49686682532441584, "learning_rate": 3.4725803210148814e-05, "loss": 2.6913, "step": 30878 }, { "epoch": 1.9168787634241728, "grad_norm": 0.5153422881076696, "learning_rate": 3.4722364366675955e-05, "loss": 2.6243, "step": 30879 }, { "epoch": 1.9169408405239308, "grad_norm": 0.5384642557059884, "learning_rate": 3.471892560291024e-05, "loss": 2.7442, "step": 30880 }, { "epoch": 1.9170029176236887, "grad_norm": 0.4992312243220918, "learning_rate": 3.47154869188696e-05, "loss": 2.6968, "step": 30881 }, { "epoch": 1.9170649947234466, "grad_norm": 0.5519056814095866, "learning_rate": 3.4712048314571964e-05, "loss": 2.7133, "step": 30882 }, { "epoch": 1.9171270718232045, "grad_norm": 0.49970940251446616, "learning_rate": 3.4708609790035294e-05, "loss": 2.63, "step": 30883 }, { "epoch": 1.9171891489229624, "grad_norm": 0.5327463673915739, "learning_rate": 3.470517134527751e-05, "loss": 2.7429, "step": 30884 }, { "epoch": 1.9172512260227204, "grad_norm": 0.5048509362670643, "learning_rate": 3.470173298031657e-05, "loss": 2.6399, "step": 30885 }, { "epoch": 1.9173133031224783, "grad_norm": 0.524086478724872, "learning_rate": 3.469829469517039e-05, "loss": 2.6814, "step": 30886 }, { "epoch": 1.917375380222236, "grad_norm": 0.4605072082054787, "learning_rate": 3.469485648985691e-05, "loss": 2.6653, "step": 30887 }, { "epoch": 1.9174374573219939, "grad_norm": 0.5713113633718223, "learning_rate": 3.469141836439411e-05, "loss": 2.7277, "step": 30888 }, { "epoch": 1.9174995344217518, "grad_norm": 0.5350074282691996, "learning_rate": 3.468798031879987e-05, "loss": 2.695, "step": 30889 }, { "epoch": 1.9175616115215097, "grad_norm": 0.5264604022377019, "learning_rate": 3.4684542353092174e-05, "loss": 2.6809, "step": 30890 }, { "epoch": 1.9176236886212676, "grad_norm": 0.577807545493688, "learning_rate": 3.4681104467288934e-05, "loss": 2.7524, "step": 30891 }, { "epoch": 1.9176857657210256, "grad_norm": 0.5017235375588472, "learning_rate": 3.467766666140809e-05, "loss": 2.7334, "step": 30892 }, { "epoch": 1.9177478428207833, "grad_norm": 0.6313743025134487, "learning_rate": 3.4674228935467575e-05, "loss": 2.7815, "step": 30893 }, { "epoch": 1.9178099199205412, "grad_norm": 0.49606862342460767, "learning_rate": 3.4670791289485337e-05, "loss": 2.6786, "step": 30894 }, { "epoch": 1.917871997020299, "grad_norm": 0.6054560506165666, "learning_rate": 3.46673537234793e-05, "loss": 2.7294, "step": 30895 }, { "epoch": 1.917934074120057, "grad_norm": 0.5640624122936572, "learning_rate": 3.46639162374674e-05, "loss": 2.7378, "step": 30896 }, { "epoch": 1.917996151219815, "grad_norm": 0.5282795550125402, "learning_rate": 3.466047883146757e-05, "loss": 2.6801, "step": 30897 }, { "epoch": 1.9180582283195728, "grad_norm": 0.5197549496094375, "learning_rate": 3.465704150549773e-05, "loss": 2.6859, "step": 30898 }, { "epoch": 1.9181203054193308, "grad_norm": 0.5224264258278766, "learning_rate": 3.4653604259575856e-05, "loss": 2.7299, "step": 30899 }, { "epoch": 1.9181823825190887, "grad_norm": 0.5532917407166223, "learning_rate": 3.465016709371984e-05, "loss": 2.6089, "step": 30900 }, { "epoch": 1.9182444596188466, "grad_norm": 0.5445214991743204, "learning_rate": 3.464673000794763e-05, "loss": 2.7652, "step": 30901 }, { "epoch": 1.9183065367186045, "grad_norm": 0.5064713300383977, "learning_rate": 3.464329300227716e-05, "loss": 2.692, "step": 30902 }, { "epoch": 1.9183686138183624, "grad_norm": 0.5452812951021417, "learning_rate": 3.463985607672635e-05, "loss": 2.5996, "step": 30903 }, { "epoch": 1.9184306909181204, "grad_norm": 0.5531309881802142, "learning_rate": 3.463641923131314e-05, "loss": 2.6913, "step": 30904 }, { "epoch": 1.9184927680178783, "grad_norm": 0.5635268058392192, "learning_rate": 3.4632982466055474e-05, "loss": 2.6972, "step": 30905 }, { "epoch": 1.9185548451176362, "grad_norm": 0.5595768100967107, "learning_rate": 3.462954578097127e-05, "loss": 2.7266, "step": 30906 }, { "epoch": 1.918616922217394, "grad_norm": 0.5620363231744946, "learning_rate": 3.462610917607844e-05, "loss": 2.7432, "step": 30907 }, { "epoch": 1.918678999317152, "grad_norm": 0.570163419913123, "learning_rate": 3.462267265139494e-05, "loss": 2.7165, "step": 30908 }, { "epoch": 1.91874107641691, "grad_norm": 0.5713190315406553, "learning_rate": 3.461923620693869e-05, "loss": 2.7856, "step": 30909 }, { "epoch": 1.9188031535166679, "grad_norm": 0.4784175383018275, "learning_rate": 3.461579984272761e-05, "loss": 2.6308, "step": 30910 }, { "epoch": 1.9188652306164256, "grad_norm": 0.5239448081736283, "learning_rate": 3.4612363558779646e-05, "loss": 2.7396, "step": 30911 }, { "epoch": 1.9189273077161835, "grad_norm": 0.5493010251752869, "learning_rate": 3.460892735511272e-05, "loss": 2.7577, "step": 30912 }, { "epoch": 1.9189893848159414, "grad_norm": 0.5068101233227761, "learning_rate": 3.4605491231744746e-05, "loss": 2.7412, "step": 30913 }, { "epoch": 1.9190514619156993, "grad_norm": 0.5757923233851471, "learning_rate": 3.4602055188693674e-05, "loss": 2.6735, "step": 30914 }, { "epoch": 1.9191135390154572, "grad_norm": 0.5620542720589902, "learning_rate": 3.45986192259774e-05, "loss": 2.7093, "step": 30915 }, { "epoch": 1.9191756161152151, "grad_norm": 0.46492123959702547, "learning_rate": 3.459518334361388e-05, "loss": 2.6941, "step": 30916 }, { "epoch": 1.9192376932149728, "grad_norm": 0.6062731168213317, "learning_rate": 3.4591747541621033e-05, "loss": 2.7441, "step": 30917 }, { "epoch": 1.9192997703147308, "grad_norm": 0.5217215466602882, "learning_rate": 3.458831182001677e-05, "loss": 2.625, "step": 30918 }, { "epoch": 1.9193618474144887, "grad_norm": 0.5398697820589264, "learning_rate": 3.458487617881903e-05, "loss": 2.7173, "step": 30919 }, { "epoch": 1.9194239245142466, "grad_norm": 0.6076601793707026, "learning_rate": 3.458144061804571e-05, "loss": 2.7829, "step": 30920 }, { "epoch": 1.9194860016140045, "grad_norm": 0.5208483471295905, "learning_rate": 3.457800513771478e-05, "loss": 2.6667, "step": 30921 }, { "epoch": 1.9195480787137624, "grad_norm": 0.6179288684886084, "learning_rate": 3.457456973784414e-05, "loss": 2.6354, "step": 30922 }, { "epoch": 1.9196101558135203, "grad_norm": 0.5353507920447188, "learning_rate": 3.457113441845171e-05, "loss": 2.7476, "step": 30923 }, { "epoch": 1.9196722329132783, "grad_norm": 0.6158447870532366, "learning_rate": 3.456769917955543e-05, "loss": 2.751, "step": 30924 }, { "epoch": 1.9197343100130362, "grad_norm": 0.6158440127713591, "learning_rate": 3.45642640211732e-05, "loss": 2.7856, "step": 30925 }, { "epoch": 1.919796387112794, "grad_norm": 0.544885449576418, "learning_rate": 3.4560828943322954e-05, "loss": 2.7156, "step": 30926 }, { "epoch": 1.919858464212552, "grad_norm": 0.5704167414466171, "learning_rate": 3.455739394602261e-05, "loss": 2.7085, "step": 30927 }, { "epoch": 1.91992054131231, "grad_norm": 0.4863872262363615, "learning_rate": 3.45539590292901e-05, "loss": 2.6363, "step": 30928 }, { "epoch": 1.9199826184120679, "grad_norm": 0.5938298021194228, "learning_rate": 3.455052419314331e-05, "loss": 2.7962, "step": 30929 }, { "epoch": 1.9200446955118258, "grad_norm": 0.5510665749711465, "learning_rate": 3.454708943760021e-05, "loss": 2.7241, "step": 30930 }, { "epoch": 1.9201067726115837, "grad_norm": 0.5078925436465116, "learning_rate": 3.454365476267869e-05, "loss": 2.8477, "step": 30931 }, { "epoch": 1.9201688497113416, "grad_norm": 0.5830258825111997, "learning_rate": 3.454022016839667e-05, "loss": 2.7659, "step": 30932 }, { "epoch": 1.9202309268110995, "grad_norm": 0.47681321912564517, "learning_rate": 3.4536785654772084e-05, "loss": 2.7011, "step": 30933 }, { "epoch": 1.9202930039108574, "grad_norm": 0.5651009086362615, "learning_rate": 3.453335122182283e-05, "loss": 2.7646, "step": 30934 }, { "epoch": 1.9203550810106151, "grad_norm": 0.534034611669676, "learning_rate": 3.452991686956685e-05, "loss": 2.7237, "step": 30935 }, { "epoch": 1.920417158110373, "grad_norm": 0.5118601363449562, "learning_rate": 3.452648259802203e-05, "loss": 2.6014, "step": 30936 }, { "epoch": 1.920479235210131, "grad_norm": 0.5088923490392544, "learning_rate": 3.4523048407206325e-05, "loss": 2.7606, "step": 30937 }, { "epoch": 1.920541312309889, "grad_norm": 0.4733093415591535, "learning_rate": 3.451961429713761e-05, "loss": 2.6508, "step": 30938 }, { "epoch": 1.9206033894096468, "grad_norm": 0.5494217292692805, "learning_rate": 3.4516180267833856e-05, "loss": 2.6823, "step": 30939 }, { "epoch": 1.9206654665094047, "grad_norm": 0.47724395245656537, "learning_rate": 3.451274631931293e-05, "loss": 2.7196, "step": 30940 }, { "epoch": 1.9207275436091624, "grad_norm": 0.5260206507900635, "learning_rate": 3.4509312451592774e-05, "loss": 2.6197, "step": 30941 }, { "epoch": 1.9207896207089203, "grad_norm": 0.5809433517859015, "learning_rate": 3.4505878664691295e-05, "loss": 2.7794, "step": 30942 }, { "epoch": 1.9208516978086783, "grad_norm": 0.4871005163786768, "learning_rate": 3.45024449586264e-05, "loss": 2.7958, "step": 30943 }, { "epoch": 1.9209137749084362, "grad_norm": 0.558072440437497, "learning_rate": 3.449901133341602e-05, "loss": 2.6891, "step": 30944 }, { "epoch": 1.920975852008194, "grad_norm": 0.5167615963521, "learning_rate": 3.449557778907805e-05, "loss": 2.7204, "step": 30945 }, { "epoch": 1.921037929107952, "grad_norm": 0.5437578463810394, "learning_rate": 3.449214432563043e-05, "loss": 2.6969, "step": 30946 }, { "epoch": 1.92110000620771, "grad_norm": 0.5002917690139852, "learning_rate": 3.4488710943091045e-05, "loss": 2.8043, "step": 30947 }, { "epoch": 1.9211620833074678, "grad_norm": 0.560317866634235, "learning_rate": 3.448527764147783e-05, "loss": 2.7022, "step": 30948 }, { "epoch": 1.9212241604072258, "grad_norm": 0.5140792623645116, "learning_rate": 3.448184442080868e-05, "loss": 2.7649, "step": 30949 }, { "epoch": 1.9212862375069837, "grad_norm": 0.525151251394856, "learning_rate": 3.447841128110152e-05, "loss": 2.8257, "step": 30950 }, { "epoch": 1.9213483146067416, "grad_norm": 0.5472627355157564, "learning_rate": 3.447497822237425e-05, "loss": 2.6324, "step": 30951 }, { "epoch": 1.9214103917064995, "grad_norm": 0.5020338357240852, "learning_rate": 3.44715452446448e-05, "loss": 2.5621, "step": 30952 }, { "epoch": 1.9214724688062574, "grad_norm": 0.5530884757617831, "learning_rate": 3.446811234793104e-05, "loss": 2.7003, "step": 30953 }, { "epoch": 1.9215345459060154, "grad_norm": 0.5373104204699148, "learning_rate": 3.446467953225092e-05, "loss": 2.7767, "step": 30954 }, { "epoch": 1.9215966230057733, "grad_norm": 0.5547879893655795, "learning_rate": 3.446124679762235e-05, "loss": 2.7022, "step": 30955 }, { "epoch": 1.9216587001055312, "grad_norm": 0.5333140610650021, "learning_rate": 3.4457814144063234e-05, "loss": 2.8149, "step": 30956 }, { "epoch": 1.921720777205289, "grad_norm": 0.5396755988968537, "learning_rate": 3.4454381571591454e-05, "loss": 2.7593, "step": 30957 }, { "epoch": 1.921782854305047, "grad_norm": 0.5414831021632011, "learning_rate": 3.445094908022496e-05, "loss": 2.753, "step": 30958 }, { "epoch": 1.9218449314048047, "grad_norm": 0.5759175919451884, "learning_rate": 3.4447516669981623e-05, "loss": 2.7433, "step": 30959 }, { "epoch": 1.9219070085045626, "grad_norm": 0.5460880339272861, "learning_rate": 3.444408434087938e-05, "loss": 2.76, "step": 30960 }, { "epoch": 1.9219690856043206, "grad_norm": 0.5334817905732683, "learning_rate": 3.444065209293611e-05, "loss": 2.6647, "step": 30961 }, { "epoch": 1.9220311627040785, "grad_norm": 0.5337080468657425, "learning_rate": 3.443721992616975e-05, "loss": 2.7887, "step": 30962 }, { "epoch": 1.9220932398038364, "grad_norm": 0.48173600835629365, "learning_rate": 3.4433787840598173e-05, "loss": 2.7155, "step": 30963 }, { "epoch": 1.9221553169035943, "grad_norm": 0.5470238346838316, "learning_rate": 3.443035583623933e-05, "loss": 2.7548, "step": 30964 }, { "epoch": 1.922217394003352, "grad_norm": 0.5059237347168126, "learning_rate": 3.4426923913111075e-05, "loss": 2.7137, "step": 30965 }, { "epoch": 1.92227947110311, "grad_norm": 0.5460810756713328, "learning_rate": 3.4423492071231356e-05, "loss": 2.6758, "step": 30966 }, { "epoch": 1.9223415482028678, "grad_norm": 0.5070698810079214, "learning_rate": 3.4420060310618064e-05, "loss": 2.6737, "step": 30967 }, { "epoch": 1.9224036253026258, "grad_norm": 0.5508132546507456, "learning_rate": 3.441662863128908e-05, "loss": 2.7028, "step": 30968 }, { "epoch": 1.9224657024023837, "grad_norm": 0.508686897432144, "learning_rate": 3.441319703326234e-05, "loss": 2.7317, "step": 30969 }, { "epoch": 1.9225277795021416, "grad_norm": 0.5340302587858567, "learning_rate": 3.4409765516555726e-05, "loss": 2.7544, "step": 30970 }, { "epoch": 1.9225898566018995, "grad_norm": 0.588764314538215, "learning_rate": 3.440633408118716e-05, "loss": 2.6993, "step": 30971 }, { "epoch": 1.9226519337016574, "grad_norm": 0.5468553267073807, "learning_rate": 3.440290272717453e-05, "loss": 2.6145, "step": 30972 }, { "epoch": 1.9227140108014154, "grad_norm": 0.5920640701052451, "learning_rate": 3.439947145453575e-05, "loss": 2.7451, "step": 30973 }, { "epoch": 1.9227760879011733, "grad_norm": 0.6077277097423567, "learning_rate": 3.4396040263288696e-05, "loss": 2.7159, "step": 30974 }, { "epoch": 1.9228381650009312, "grad_norm": 0.5688558815291865, "learning_rate": 3.4392609153451306e-05, "loss": 2.7172, "step": 30975 }, { "epoch": 1.922900242100689, "grad_norm": 0.6069238221539667, "learning_rate": 3.4389178125041456e-05, "loss": 2.7536, "step": 30976 }, { "epoch": 1.922962319200447, "grad_norm": 0.5322130953983545, "learning_rate": 3.4385747178077056e-05, "loss": 2.7967, "step": 30977 }, { "epoch": 1.923024396300205, "grad_norm": 0.6095112623796763, "learning_rate": 3.4382316312576e-05, "loss": 2.7868, "step": 30978 }, { "epoch": 1.9230864733999629, "grad_norm": 0.5268395450649828, "learning_rate": 3.437888552855619e-05, "loss": 2.7254, "step": 30979 }, { "epoch": 1.9231485504997208, "grad_norm": 0.5785668463121842, "learning_rate": 3.437545482603553e-05, "loss": 2.6783, "step": 30980 }, { "epoch": 1.9232106275994787, "grad_norm": 0.5508065725146327, "learning_rate": 3.4372024205031916e-05, "loss": 2.7488, "step": 30981 }, { "epoch": 1.9232727046992366, "grad_norm": 0.507182948808352, "learning_rate": 3.436859366556325e-05, "loss": 2.7431, "step": 30982 }, { "epoch": 1.9233347817989943, "grad_norm": 0.5526708290047275, "learning_rate": 3.436516320764741e-05, "loss": 2.7696, "step": 30983 }, { "epoch": 1.9233968588987522, "grad_norm": 0.5503534048584217, "learning_rate": 3.4361732831302327e-05, "loss": 2.6666, "step": 30984 }, { "epoch": 1.9234589359985101, "grad_norm": 0.5423802266868023, "learning_rate": 3.435830253654586e-05, "loss": 2.7599, "step": 30985 }, { "epoch": 1.923521013098268, "grad_norm": 0.6073272729537088, "learning_rate": 3.435487232339593e-05, "loss": 2.6516, "step": 30986 }, { "epoch": 1.923583090198026, "grad_norm": 0.537877009378058, "learning_rate": 3.435144219187043e-05, "loss": 2.7177, "step": 30987 }, { "epoch": 1.923645167297784, "grad_norm": 0.5643228294702265, "learning_rate": 3.434801214198726e-05, "loss": 2.7496, "step": 30988 }, { "epoch": 1.9237072443975416, "grad_norm": 0.5585404750788023, "learning_rate": 3.434458217376432e-05, "loss": 2.807, "step": 30989 }, { "epoch": 1.9237693214972995, "grad_norm": 0.5990464739976634, "learning_rate": 3.434115228721948e-05, "loss": 2.7754, "step": 30990 }, { "epoch": 1.9238313985970574, "grad_norm": 0.5257012962551002, "learning_rate": 3.433772248237066e-05, "loss": 2.6847, "step": 30991 }, { "epoch": 1.9238934756968153, "grad_norm": 0.6101457416640598, "learning_rate": 3.4334292759235745e-05, "loss": 2.6662, "step": 30992 }, { "epoch": 1.9239555527965733, "grad_norm": 0.5416394434714763, "learning_rate": 3.4330863117832616e-05, "loss": 2.7199, "step": 30993 }, { "epoch": 1.9240176298963312, "grad_norm": 0.6168964520621216, "learning_rate": 3.432743355817919e-05, "loss": 2.6744, "step": 30994 }, { "epoch": 1.924079706996089, "grad_norm": 0.5987258515236713, "learning_rate": 3.432400408029334e-05, "loss": 2.7833, "step": 30995 }, { "epoch": 1.924141784095847, "grad_norm": 0.6432436711044099, "learning_rate": 3.432057468419298e-05, "loss": 2.6905, "step": 30996 }, { "epoch": 1.924203861195605, "grad_norm": 0.5911398031680595, "learning_rate": 3.431714536989596e-05, "loss": 2.7286, "step": 30997 }, { "epoch": 1.9242659382953629, "grad_norm": 0.5988855372986621, "learning_rate": 3.431371613742023e-05, "loss": 2.7059, "step": 30998 }, { "epoch": 1.9243280153951208, "grad_norm": 0.6456807838262717, "learning_rate": 3.431028698678362e-05, "loss": 2.6697, "step": 30999 }, { "epoch": 1.9243900924948787, "grad_norm": 0.5795096235027506, "learning_rate": 3.430685791800408e-05, "loss": 2.7636, "step": 31000 }, { "epoch": 1.9244521695946366, "grad_norm": 0.6226200567086393, "learning_rate": 3.430342893109946e-05, "loss": 2.7084, "step": 31001 }, { "epoch": 1.9245142466943945, "grad_norm": 0.5680607538858515, "learning_rate": 3.4300000026087666e-05, "loss": 2.7135, "step": 31002 }, { "epoch": 1.9245763237941524, "grad_norm": 0.6129925285120502, "learning_rate": 3.429657120298658e-05, "loss": 2.7759, "step": 31003 }, { "epoch": 1.9246384008939104, "grad_norm": 0.6196932808072096, "learning_rate": 3.429314246181409e-05, "loss": 2.7534, "step": 31004 }, { "epoch": 1.9247004779936683, "grad_norm": 0.5650119852803452, "learning_rate": 3.4289713802588106e-05, "loss": 2.6673, "step": 31005 }, { "epoch": 1.9247625550934262, "grad_norm": 0.6091680419826333, "learning_rate": 3.4286285225326474e-05, "loss": 2.6443, "step": 31006 }, { "epoch": 1.924824632193184, "grad_norm": 0.5692733023526838, "learning_rate": 3.428285673004713e-05, "loss": 2.6587, "step": 31007 }, { "epoch": 1.9248867092929418, "grad_norm": 0.583023556699578, "learning_rate": 3.427942831676791e-05, "loss": 2.7079, "step": 31008 }, { "epoch": 1.9249487863926997, "grad_norm": 0.6181223344504313, "learning_rate": 3.427599998550676e-05, "loss": 2.7948, "step": 31009 }, { "epoch": 1.9250108634924576, "grad_norm": 0.520579034395005, "learning_rate": 3.4272571736281514e-05, "loss": 2.7218, "step": 31010 }, { "epoch": 1.9250729405922156, "grad_norm": 0.550341897598432, "learning_rate": 3.426914356911009e-05, "loss": 2.7181, "step": 31011 }, { "epoch": 1.9251350176919735, "grad_norm": 0.569907475708584, "learning_rate": 3.4265715484010354e-05, "loss": 2.6946, "step": 31012 }, { "epoch": 1.9251970947917312, "grad_norm": 0.5040003666479048, "learning_rate": 3.426228748100021e-05, "loss": 2.7286, "step": 31013 }, { "epoch": 1.925259171891489, "grad_norm": 0.5880091854954481, "learning_rate": 3.425885956009752e-05, "loss": 2.6526, "step": 31014 }, { "epoch": 1.925321248991247, "grad_norm": 0.4962220814440094, "learning_rate": 3.425543172132019e-05, "loss": 2.6782, "step": 31015 }, { "epoch": 1.925383326091005, "grad_norm": 0.5504094485440819, "learning_rate": 3.42520039646861e-05, "loss": 2.7651, "step": 31016 }, { "epoch": 1.9254454031907628, "grad_norm": 0.5201069216658746, "learning_rate": 3.424857629021312e-05, "loss": 2.6474, "step": 31017 }, { "epoch": 1.9255074802905208, "grad_norm": 0.4975567519104235, "learning_rate": 3.424514869791912e-05, "loss": 2.6558, "step": 31018 }, { "epoch": 1.9255695573902787, "grad_norm": 0.4923785005567529, "learning_rate": 3.424172118782203e-05, "loss": 2.7592, "step": 31019 }, { "epoch": 1.9256316344900366, "grad_norm": 0.5219629499141379, "learning_rate": 3.42382937599397e-05, "loss": 2.6749, "step": 31020 }, { "epoch": 1.9256937115897945, "grad_norm": 0.5355196088197999, "learning_rate": 3.423486641429002e-05, "loss": 2.7474, "step": 31021 }, { "epoch": 1.9257557886895524, "grad_norm": 0.55260754535396, "learning_rate": 3.423143915089087e-05, "loss": 2.7337, "step": 31022 }, { "epoch": 1.9258178657893104, "grad_norm": 0.5585835062434852, "learning_rate": 3.422801196976013e-05, "loss": 2.6603, "step": 31023 }, { "epoch": 1.9258799428890683, "grad_norm": 0.550636894842349, "learning_rate": 3.422458487091568e-05, "loss": 2.6702, "step": 31024 }, { "epoch": 1.9259420199888262, "grad_norm": 0.537839663549375, "learning_rate": 3.42211578543754e-05, "loss": 2.8017, "step": 31025 }, { "epoch": 1.926004097088584, "grad_norm": 0.6005848706243699, "learning_rate": 3.4217730920157164e-05, "loss": 2.7512, "step": 31026 }, { "epoch": 1.926066174188342, "grad_norm": 0.5239793620304155, "learning_rate": 3.421430406827887e-05, "loss": 2.7744, "step": 31027 }, { "epoch": 1.9261282512881, "grad_norm": 0.5974304171470679, "learning_rate": 3.421087729875838e-05, "loss": 2.7453, "step": 31028 }, { "epoch": 1.9261903283878579, "grad_norm": 0.540834554951037, "learning_rate": 3.420745061161357e-05, "loss": 2.69, "step": 31029 }, { "epoch": 1.9262524054876158, "grad_norm": 0.5532236797865366, "learning_rate": 3.4204024006862334e-05, "loss": 2.7136, "step": 31030 }, { "epoch": 1.9263144825873735, "grad_norm": 0.5695208973722037, "learning_rate": 3.420059748452252e-05, "loss": 2.7405, "step": 31031 }, { "epoch": 1.9263765596871314, "grad_norm": 0.5577795590630408, "learning_rate": 3.419717104461204e-05, "loss": 2.7753, "step": 31032 }, { "epoch": 1.9264386367868893, "grad_norm": 0.5325660113042179, "learning_rate": 3.4193744687148754e-05, "loss": 2.8137, "step": 31033 }, { "epoch": 1.9265007138866472, "grad_norm": 0.5691975970216187, "learning_rate": 3.419031841215054e-05, "loss": 2.692, "step": 31034 }, { "epoch": 1.9265627909864051, "grad_norm": 0.5961328676120119, "learning_rate": 3.418689221963526e-05, "loss": 2.7876, "step": 31035 }, { "epoch": 1.926624868086163, "grad_norm": 0.5174897316360831, "learning_rate": 3.418346610962082e-05, "loss": 2.7586, "step": 31036 }, { "epoch": 1.9266869451859208, "grad_norm": 0.5720256523602419, "learning_rate": 3.418004008212506e-05, "loss": 2.7966, "step": 31037 }, { "epoch": 1.9267490222856787, "grad_norm": 0.5165916687102711, "learning_rate": 3.4176614137165885e-05, "loss": 2.6797, "step": 31038 }, { "epoch": 1.9268110993854366, "grad_norm": 0.5928211224400964, "learning_rate": 3.4173188274761144e-05, "loss": 2.7348, "step": 31039 }, { "epoch": 1.9268731764851945, "grad_norm": 0.562716362985461, "learning_rate": 3.416976249492873e-05, "loss": 2.7031, "step": 31040 }, { "epoch": 1.9269352535849524, "grad_norm": 0.4979614034876249, "learning_rate": 3.416633679768651e-05, "loss": 2.7135, "step": 31041 }, { "epoch": 1.9269973306847104, "grad_norm": 0.6076845049008927, "learning_rate": 3.4162911183052346e-05, "loss": 2.7308, "step": 31042 }, { "epoch": 1.9270594077844683, "grad_norm": 0.5370210332178882, "learning_rate": 3.415948565104413e-05, "loss": 2.7987, "step": 31043 }, { "epoch": 1.9271214848842262, "grad_norm": 0.5550360054932248, "learning_rate": 3.4156060201679704e-05, "loss": 2.7071, "step": 31044 }, { "epoch": 1.927183561983984, "grad_norm": 0.5062044358647301, "learning_rate": 3.415263483497698e-05, "loss": 2.7096, "step": 31045 }, { "epoch": 1.927245639083742, "grad_norm": 0.5570835426042954, "learning_rate": 3.414920955095379e-05, "loss": 2.7257, "step": 31046 }, { "epoch": 1.9273077161835, "grad_norm": 0.523785176746021, "learning_rate": 3.414578434962803e-05, "loss": 2.809, "step": 31047 }, { "epoch": 1.9273697932832579, "grad_norm": 0.4856606464485224, "learning_rate": 3.4142359231017565e-05, "loss": 2.6982, "step": 31048 }, { "epoch": 1.9274318703830158, "grad_norm": 0.5349225278716395, "learning_rate": 3.413893419514026e-05, "loss": 2.5832, "step": 31049 }, { "epoch": 1.9274939474827737, "grad_norm": 0.4969431446538329, "learning_rate": 3.413550924201397e-05, "loss": 2.7226, "step": 31050 }, { "epoch": 1.9275560245825316, "grad_norm": 0.4987184372573523, "learning_rate": 3.413208437165658e-05, "loss": 2.6683, "step": 31051 }, { "epoch": 1.9276181016822895, "grad_norm": 0.538034786278549, "learning_rate": 3.4128659584085974e-05, "loss": 2.6888, "step": 31052 }, { "epoch": 1.9276801787820474, "grad_norm": 0.5055014032255746, "learning_rate": 3.412523487932001e-05, "loss": 2.7144, "step": 31053 }, { "epoch": 1.9277422558818054, "grad_norm": 0.5208537542790276, "learning_rate": 3.412181025737653e-05, "loss": 2.7438, "step": 31054 }, { "epoch": 1.927804332981563, "grad_norm": 0.6022200520481422, "learning_rate": 3.411838571827344e-05, "loss": 2.7919, "step": 31055 }, { "epoch": 1.927866410081321, "grad_norm": 0.4863663930276, "learning_rate": 3.4114961262028566e-05, "loss": 2.7071, "step": 31056 }, { "epoch": 1.927928487181079, "grad_norm": 0.5850684014703175, "learning_rate": 3.411153688865981e-05, "loss": 2.7218, "step": 31057 }, { "epoch": 1.9279905642808368, "grad_norm": 0.5668003684211615, "learning_rate": 3.4108112598185014e-05, "loss": 2.7423, "step": 31058 }, { "epoch": 1.9280526413805947, "grad_norm": 0.6654017015668154, "learning_rate": 3.4104688390622064e-05, "loss": 2.7119, "step": 31059 }, { "epoch": 1.9281147184803527, "grad_norm": 0.5880497055015997, "learning_rate": 3.410126426598881e-05, "loss": 2.7954, "step": 31060 }, { "epoch": 1.9281767955801103, "grad_norm": 0.5807865838603079, "learning_rate": 3.409784022430312e-05, "loss": 2.8295, "step": 31061 }, { "epoch": 1.9282388726798683, "grad_norm": 0.5549315869205045, "learning_rate": 3.409441626558285e-05, "loss": 2.6932, "step": 31062 }, { "epoch": 1.9283009497796262, "grad_norm": 0.5871877682671536, "learning_rate": 3.409099238984588e-05, "loss": 2.6993, "step": 31063 }, { "epoch": 1.928363026879384, "grad_norm": 0.5511798904019968, "learning_rate": 3.4087568597110057e-05, "loss": 2.6932, "step": 31064 }, { "epoch": 1.928425103979142, "grad_norm": 0.6066066017607276, "learning_rate": 3.408414488739326e-05, "loss": 2.6931, "step": 31065 }, { "epoch": 1.9284871810789, "grad_norm": 0.5340230597078305, "learning_rate": 3.4080721260713344e-05, "loss": 2.7326, "step": 31066 }, { "epoch": 1.9285492581786579, "grad_norm": 0.6558824145488866, "learning_rate": 3.407729771708816e-05, "loss": 2.7635, "step": 31067 }, { "epoch": 1.9286113352784158, "grad_norm": 0.5931549101705318, "learning_rate": 3.4073874256535585e-05, "loss": 2.7581, "step": 31068 }, { "epoch": 1.9286734123781737, "grad_norm": 0.6225512694881659, "learning_rate": 3.407045087907347e-05, "loss": 2.6806, "step": 31069 }, { "epoch": 1.9287354894779316, "grad_norm": 0.598724532451827, "learning_rate": 3.406702758471969e-05, "loss": 2.6701, "step": 31070 }, { "epoch": 1.9287975665776895, "grad_norm": 0.5424737938471125, "learning_rate": 3.406360437349208e-05, "loss": 2.6789, "step": 31071 }, { "epoch": 1.9288596436774474, "grad_norm": 0.6162374644909933, "learning_rate": 3.406018124540853e-05, "loss": 2.7349, "step": 31072 }, { "epoch": 1.9289217207772054, "grad_norm": 0.5524601340259383, "learning_rate": 3.405675820048687e-05, "loss": 2.7456, "step": 31073 }, { "epoch": 1.9289837978769633, "grad_norm": 0.6627246170018087, "learning_rate": 3.405333523874498e-05, "loss": 2.7189, "step": 31074 }, { "epoch": 1.9290458749767212, "grad_norm": 0.5317060812876608, "learning_rate": 3.404991236020071e-05, "loss": 2.7247, "step": 31075 }, { "epoch": 1.9291079520764791, "grad_norm": 0.56193219773162, "learning_rate": 3.404648956487192e-05, "loss": 2.7615, "step": 31076 }, { "epoch": 1.929170029176237, "grad_norm": 0.5545480848451109, "learning_rate": 3.4043066852776474e-05, "loss": 2.8163, "step": 31077 }, { "epoch": 1.929232106275995, "grad_norm": 0.5574745490162103, "learning_rate": 3.403964422393221e-05, "loss": 2.79, "step": 31078 }, { "epoch": 1.9292941833757526, "grad_norm": 0.5388332653533314, "learning_rate": 3.4036221678357004e-05, "loss": 2.6895, "step": 31079 }, { "epoch": 1.9293562604755106, "grad_norm": 0.50038754821832, "learning_rate": 3.4032799216068694e-05, "loss": 2.6523, "step": 31080 }, { "epoch": 1.9294183375752685, "grad_norm": 0.5015531794552504, "learning_rate": 3.402937683708516e-05, "loss": 2.6022, "step": 31081 }, { "epoch": 1.9294804146750264, "grad_norm": 0.5603607879017537, "learning_rate": 3.402595454142423e-05, "loss": 2.6866, "step": 31082 }, { "epoch": 1.9295424917747843, "grad_norm": 0.4679584177242612, "learning_rate": 3.402253232910379e-05, "loss": 2.6886, "step": 31083 }, { "epoch": 1.9296045688745422, "grad_norm": 0.5812139981923331, "learning_rate": 3.4019110200141656e-05, "loss": 2.7388, "step": 31084 }, { "epoch": 1.9296666459743, "grad_norm": 0.4728925602884489, "learning_rate": 3.401568815455571e-05, "loss": 2.7026, "step": 31085 }, { "epoch": 1.9297287230740579, "grad_norm": 0.5452135187509087, "learning_rate": 3.4012266192363805e-05, "loss": 2.633, "step": 31086 }, { "epoch": 1.9297908001738158, "grad_norm": 0.5179116667856273, "learning_rate": 3.400884431358379e-05, "loss": 2.7223, "step": 31087 }, { "epoch": 1.9298528772735737, "grad_norm": 0.4948875300918527, "learning_rate": 3.400542251823352e-05, "loss": 2.7157, "step": 31088 }, { "epoch": 1.9299149543733316, "grad_norm": 0.5402772457520068, "learning_rate": 3.4002000806330836e-05, "loss": 2.7413, "step": 31089 }, { "epoch": 1.9299770314730895, "grad_norm": 0.5292733387799252, "learning_rate": 3.39985791778936e-05, "loss": 2.7231, "step": 31090 }, { "epoch": 1.9300391085728474, "grad_norm": 0.523566728268604, "learning_rate": 3.399515763293967e-05, "loss": 2.7116, "step": 31091 }, { "epoch": 1.9301011856726054, "grad_norm": 0.5264348993292632, "learning_rate": 3.399173617148687e-05, "loss": 2.7452, "step": 31092 }, { "epoch": 1.9301632627723633, "grad_norm": 0.498947735025487, "learning_rate": 3.398831479355308e-05, "loss": 2.6773, "step": 31093 }, { "epoch": 1.9302253398721212, "grad_norm": 0.5573961718487007, "learning_rate": 3.398489349915613e-05, "loss": 2.6929, "step": 31094 }, { "epoch": 1.930287416971879, "grad_norm": 0.5284485159007187, "learning_rate": 3.39814722883139e-05, "loss": 2.7224, "step": 31095 }, { "epoch": 1.930349494071637, "grad_norm": 0.5207128639769912, "learning_rate": 3.397805116104419e-05, "loss": 2.7233, "step": 31096 }, { "epoch": 1.930411571171395, "grad_norm": 0.5467498908668552, "learning_rate": 3.39746301173649e-05, "loss": 2.7657, "step": 31097 }, { "epoch": 1.9304736482711529, "grad_norm": 0.5446673656863408, "learning_rate": 3.397120915729384e-05, "loss": 2.7794, "step": 31098 }, { "epoch": 1.9305357253709108, "grad_norm": 0.5250191775861058, "learning_rate": 3.3967788280848875e-05, "loss": 2.6335, "step": 31099 }, { "epoch": 1.9305978024706687, "grad_norm": 0.5812278425443905, "learning_rate": 3.396436748804785e-05, "loss": 2.7631, "step": 31100 }, { "epoch": 1.9306598795704266, "grad_norm": 0.5792794943303057, "learning_rate": 3.396094677890862e-05, "loss": 2.7918, "step": 31101 }, { "epoch": 1.9307219566701843, "grad_norm": 0.635608852485193, "learning_rate": 3.395752615344902e-05, "loss": 2.6799, "step": 31102 }, { "epoch": 1.9307840337699422, "grad_norm": 0.5088676348035853, "learning_rate": 3.39541056116869e-05, "loss": 2.7666, "step": 31103 }, { "epoch": 1.9308461108697001, "grad_norm": 0.5601273139492083, "learning_rate": 3.3950685153640114e-05, "loss": 2.5894, "step": 31104 }, { "epoch": 1.930908187969458, "grad_norm": 0.6080470654848661, "learning_rate": 3.394726477932648e-05, "loss": 2.7324, "step": 31105 }, { "epoch": 1.930970265069216, "grad_norm": 0.5683235602876314, "learning_rate": 3.394384448876388e-05, "loss": 2.7514, "step": 31106 }, { "epoch": 1.931032342168974, "grad_norm": 0.6848562299255787, "learning_rate": 3.394042428197013e-05, "loss": 2.7914, "step": 31107 }, { "epoch": 1.9310944192687316, "grad_norm": 0.5424366546160126, "learning_rate": 3.39370041589631e-05, "loss": 2.7226, "step": 31108 }, { "epoch": 1.9311564963684895, "grad_norm": 0.6426464380748327, "learning_rate": 3.39335841197606e-05, "loss": 2.683, "step": 31109 }, { "epoch": 1.9312185734682474, "grad_norm": 0.5558183184411074, "learning_rate": 3.393016416438051e-05, "loss": 2.6768, "step": 31110 }, { "epoch": 1.9312806505680054, "grad_norm": 0.5957682839700227, "learning_rate": 3.392674429284064e-05, "loss": 2.7448, "step": 31111 }, { "epoch": 1.9313427276677633, "grad_norm": 0.5558625522075039, "learning_rate": 3.392332450515886e-05, "loss": 2.6721, "step": 31112 }, { "epoch": 1.9314048047675212, "grad_norm": 0.5367130551320644, "learning_rate": 3.3919904801352994e-05, "loss": 2.7669, "step": 31113 }, { "epoch": 1.931466881867279, "grad_norm": 0.5494017132111296, "learning_rate": 3.391648518144088e-05, "loss": 2.6888, "step": 31114 }, { "epoch": 1.931528958967037, "grad_norm": 0.5188881758999924, "learning_rate": 3.3913065645440376e-05, "loss": 2.738, "step": 31115 }, { "epoch": 1.931591036066795, "grad_norm": 0.568520854020588, "learning_rate": 3.3909646193369305e-05, "loss": 2.7343, "step": 31116 }, { "epoch": 1.9316531131665529, "grad_norm": 0.5153503553506047, "learning_rate": 3.390622682524551e-05, "loss": 2.6858, "step": 31117 }, { "epoch": 1.9317151902663108, "grad_norm": 0.5665146033466193, "learning_rate": 3.390280754108685e-05, "loss": 2.6885, "step": 31118 }, { "epoch": 1.9317772673660687, "grad_norm": 0.5575693781411051, "learning_rate": 3.389938834091114e-05, "loss": 2.7569, "step": 31119 }, { "epoch": 1.9318393444658266, "grad_norm": 0.5267803149142896, "learning_rate": 3.389596922473625e-05, "loss": 2.6388, "step": 31120 }, { "epoch": 1.9319014215655845, "grad_norm": 0.5919991075293154, "learning_rate": 3.389255019257998e-05, "loss": 2.761, "step": 31121 }, { "epoch": 1.9319634986653424, "grad_norm": 0.5782539893926807, "learning_rate": 3.38891312444602e-05, "loss": 2.7487, "step": 31122 }, { "epoch": 1.9320255757651004, "grad_norm": 0.5628115532934348, "learning_rate": 3.388571238039472e-05, "loss": 2.7643, "step": 31123 }, { "epoch": 1.9320876528648583, "grad_norm": 0.6385347178227827, "learning_rate": 3.38822936004014e-05, "loss": 2.7449, "step": 31124 }, { "epoch": 1.9321497299646162, "grad_norm": 0.535578011596073, "learning_rate": 3.387887490449806e-05, "loss": 2.6326, "step": 31125 }, { "epoch": 1.932211807064374, "grad_norm": 0.5550376163227303, "learning_rate": 3.3875456292702546e-05, "loss": 2.7116, "step": 31126 }, { "epoch": 1.9322738841641318, "grad_norm": 0.5615000048184755, "learning_rate": 3.38720377650327e-05, "loss": 2.739, "step": 31127 }, { "epoch": 1.9323359612638897, "grad_norm": 0.5485263415594948, "learning_rate": 3.386861932150633e-05, "loss": 2.6815, "step": 31128 }, { "epoch": 1.9323980383636477, "grad_norm": 0.5418709559218208, "learning_rate": 3.3865200962141305e-05, "loss": 2.6838, "step": 31129 }, { "epoch": 1.9324601154634056, "grad_norm": 0.6108812155004392, "learning_rate": 3.3861782686955435e-05, "loss": 2.723, "step": 31130 }, { "epoch": 1.9325221925631635, "grad_norm": 0.6050422889337264, "learning_rate": 3.3858364495966565e-05, "loss": 2.8042, "step": 31131 }, { "epoch": 1.9325842696629212, "grad_norm": 0.573355200575984, "learning_rate": 3.385494638919252e-05, "loss": 2.6977, "step": 31132 }, { "epoch": 1.932646346762679, "grad_norm": 0.6107434044933484, "learning_rate": 3.385152836665115e-05, "loss": 2.6485, "step": 31133 }, { "epoch": 1.932708423862437, "grad_norm": 0.5767497381516463, "learning_rate": 3.384811042836027e-05, "loss": 2.6695, "step": 31134 }, { "epoch": 1.932770500962195, "grad_norm": 0.5918460485594111, "learning_rate": 3.384469257433772e-05, "loss": 2.7519, "step": 31135 }, { "epoch": 1.9328325780619529, "grad_norm": 0.6194159440135123, "learning_rate": 3.384127480460133e-05, "loss": 2.6218, "step": 31136 }, { "epoch": 1.9328946551617108, "grad_norm": 0.5599266901310846, "learning_rate": 3.3837857119168936e-05, "loss": 2.7683, "step": 31137 }, { "epoch": 1.9329567322614687, "grad_norm": 0.6300274823264722, "learning_rate": 3.3834439518058364e-05, "loss": 2.7675, "step": 31138 }, { "epoch": 1.9330188093612266, "grad_norm": 0.4969789612172052, "learning_rate": 3.383102200128744e-05, "loss": 2.6597, "step": 31139 }, { "epoch": 1.9330808864609845, "grad_norm": 0.5744852498603675, "learning_rate": 3.382760456887401e-05, "loss": 2.7692, "step": 31140 }, { "epoch": 1.9331429635607424, "grad_norm": 0.5706341763831312, "learning_rate": 3.382418722083588e-05, "loss": 2.7152, "step": 31141 }, { "epoch": 1.9332050406605004, "grad_norm": 0.5568376159107465, "learning_rate": 3.38207699571909e-05, "loss": 2.7194, "step": 31142 }, { "epoch": 1.9332671177602583, "grad_norm": 0.5839776896728804, "learning_rate": 3.3817352777956876e-05, "loss": 2.7345, "step": 31143 }, { "epoch": 1.9333291948600162, "grad_norm": 0.5532455506936637, "learning_rate": 3.3813935683151676e-05, "loss": 2.7653, "step": 31144 }, { "epoch": 1.9333912719597741, "grad_norm": 0.5794885124190293, "learning_rate": 3.381051867279308e-05, "loss": 2.7464, "step": 31145 }, { "epoch": 1.933453349059532, "grad_norm": 0.5796552201817744, "learning_rate": 3.380710174689895e-05, "loss": 2.6579, "step": 31146 }, { "epoch": 1.93351542615929, "grad_norm": 0.5547147394933284, "learning_rate": 3.38036849054871e-05, "loss": 2.7433, "step": 31147 }, { "epoch": 1.9335775032590479, "grad_norm": 0.6141029490187166, "learning_rate": 3.380026814857536e-05, "loss": 2.7333, "step": 31148 }, { "epoch": 1.9336395803588058, "grad_norm": 0.5847530345797235, "learning_rate": 3.3796851476181534e-05, "loss": 2.7572, "step": 31149 }, { "epoch": 1.9337016574585635, "grad_norm": 0.602687897685718, "learning_rate": 3.3793434888323486e-05, "loss": 2.7915, "step": 31150 }, { "epoch": 1.9337637345583214, "grad_norm": 0.5991867265760962, "learning_rate": 3.379001838501902e-05, "loss": 2.7552, "step": 31151 }, { "epoch": 1.9338258116580793, "grad_norm": 0.5843953501376026, "learning_rate": 3.3786601966285966e-05, "loss": 2.7137, "step": 31152 }, { "epoch": 1.9338878887578372, "grad_norm": 0.5825504061373932, "learning_rate": 3.3783185632142136e-05, "loss": 2.7501, "step": 31153 }, { "epoch": 1.9339499658575952, "grad_norm": 0.5054424438234778, "learning_rate": 3.377976938260538e-05, "loss": 2.7084, "step": 31154 }, { "epoch": 1.934012042957353, "grad_norm": 0.6006445383154946, "learning_rate": 3.3776353217693494e-05, "loss": 2.7056, "step": 31155 }, { "epoch": 1.9340741200571108, "grad_norm": 0.5055507469589001, "learning_rate": 3.377293713742431e-05, "loss": 2.7632, "step": 31156 }, { "epoch": 1.9341361971568687, "grad_norm": 0.550759741180548, "learning_rate": 3.376952114181565e-05, "loss": 2.7155, "step": 31157 }, { "epoch": 1.9341982742566266, "grad_norm": 0.5396979912211404, "learning_rate": 3.376610523088535e-05, "loss": 2.6457, "step": 31158 }, { "epoch": 1.9342603513563845, "grad_norm": 0.5031602941892731, "learning_rate": 3.3762689404651204e-05, "loss": 2.6581, "step": 31159 }, { "epoch": 1.9343224284561424, "grad_norm": 0.566754806082548, "learning_rate": 3.375927366313106e-05, "loss": 2.7679, "step": 31160 }, { "epoch": 1.9343845055559004, "grad_norm": 0.4673832040124346, "learning_rate": 3.375585800634273e-05, "loss": 2.7046, "step": 31161 }, { "epoch": 1.9344465826556583, "grad_norm": 0.557715652777837, "learning_rate": 3.3752442434304035e-05, "loss": 2.6277, "step": 31162 }, { "epoch": 1.9345086597554162, "grad_norm": 0.5342677181592139, "learning_rate": 3.3749026947032786e-05, "loss": 2.6772, "step": 31163 }, { "epoch": 1.9345707368551741, "grad_norm": 0.5526420597048194, "learning_rate": 3.374561154454681e-05, "loss": 2.7217, "step": 31164 }, { "epoch": 1.934632813954932, "grad_norm": 0.5726008296029853, "learning_rate": 3.3742196226863924e-05, "loss": 2.7602, "step": 31165 }, { "epoch": 1.93469489105469, "grad_norm": 0.5606914692645633, "learning_rate": 3.373878099400194e-05, "loss": 2.7682, "step": 31166 }, { "epoch": 1.9347569681544479, "grad_norm": 0.5206671322447242, "learning_rate": 3.3735365845978696e-05, "loss": 2.8678, "step": 31167 }, { "epoch": 1.9348190452542058, "grad_norm": 0.5689821272625851, "learning_rate": 3.373195078281198e-05, "loss": 2.7393, "step": 31168 }, { "epoch": 1.9348811223539637, "grad_norm": 0.5335922777370623, "learning_rate": 3.372853580451964e-05, "loss": 2.7393, "step": 31169 }, { "epoch": 1.9349431994537216, "grad_norm": 0.5429576076763941, "learning_rate": 3.372512091111947e-05, "loss": 2.7228, "step": 31170 }, { "epoch": 1.9350052765534795, "grad_norm": 0.5584738275872289, "learning_rate": 3.37217061026293e-05, "loss": 2.7145, "step": 31171 }, { "epoch": 1.9350673536532375, "grad_norm": 0.6271313327867225, "learning_rate": 3.3718291379066936e-05, "loss": 2.7236, "step": 31172 }, { "epoch": 1.9351294307529954, "grad_norm": 0.6010466754468825, "learning_rate": 3.371487674045021e-05, "loss": 2.6556, "step": 31173 }, { "epoch": 1.935191507852753, "grad_norm": 0.6127993397172553, "learning_rate": 3.371146218679692e-05, "loss": 2.7332, "step": 31174 }, { "epoch": 1.935253584952511, "grad_norm": 0.5824401494414261, "learning_rate": 3.370804771812488e-05, "loss": 2.8558, "step": 31175 }, { "epoch": 1.935315662052269, "grad_norm": 0.6179763725958426, "learning_rate": 3.370463333445192e-05, "loss": 2.8237, "step": 31176 }, { "epoch": 1.9353777391520268, "grad_norm": 0.5718668827668987, "learning_rate": 3.370121903579583e-05, "loss": 2.8178, "step": 31177 }, { "epoch": 1.9354398162517847, "grad_norm": 0.6525835978277251, "learning_rate": 3.3697804822174453e-05, "loss": 2.7653, "step": 31178 }, { "epoch": 1.9355018933515427, "grad_norm": 0.6140216561437273, "learning_rate": 3.369439069360557e-05, "loss": 2.7354, "step": 31179 }, { "epoch": 1.9355639704513004, "grad_norm": 0.623482698221908, "learning_rate": 3.369097665010702e-05, "loss": 2.7611, "step": 31180 }, { "epoch": 1.9356260475510583, "grad_norm": 0.5625009536735079, "learning_rate": 3.3687562691696593e-05, "loss": 2.6718, "step": 31181 }, { "epoch": 1.9356881246508162, "grad_norm": 0.5429072997084644, "learning_rate": 3.36841488183921e-05, "loss": 2.7343, "step": 31182 }, { "epoch": 1.935750201750574, "grad_norm": 0.5964558083039654, "learning_rate": 3.368073503021139e-05, "loss": 2.7932, "step": 31183 }, { "epoch": 1.935812278850332, "grad_norm": 0.5767600467999021, "learning_rate": 3.367732132717224e-05, "loss": 2.7222, "step": 31184 }, { "epoch": 1.93587435595009, "grad_norm": 0.5654528255702717, "learning_rate": 3.367390770929246e-05, "loss": 2.6797, "step": 31185 }, { "epoch": 1.9359364330498479, "grad_norm": 0.5216618211001172, "learning_rate": 3.3670494176589875e-05, "loss": 2.7162, "step": 31186 }, { "epoch": 1.9359985101496058, "grad_norm": 0.5650255145582559, "learning_rate": 3.366708072908229e-05, "loss": 2.6947, "step": 31187 }, { "epoch": 1.9360605872493637, "grad_norm": 0.5997299669504919, "learning_rate": 3.366366736678751e-05, "loss": 2.7726, "step": 31188 }, { "epoch": 1.9361226643491216, "grad_norm": 0.5268027745040585, "learning_rate": 3.366025408972333e-05, "loss": 2.7728, "step": 31189 }, { "epoch": 1.9361847414488795, "grad_norm": 0.6297985878161446, "learning_rate": 3.3656840897907585e-05, "loss": 2.7191, "step": 31190 }, { "epoch": 1.9362468185486374, "grad_norm": 0.5015309558586833, "learning_rate": 3.365342779135805e-05, "loss": 2.7693, "step": 31191 }, { "epoch": 1.9363088956483954, "grad_norm": 0.5551749491957952, "learning_rate": 3.365001477009257e-05, "loss": 2.7419, "step": 31192 }, { "epoch": 1.9363709727481533, "grad_norm": 0.4801780938028225, "learning_rate": 3.364660183412892e-05, "loss": 2.7505, "step": 31193 }, { "epoch": 1.9364330498479112, "grad_norm": 0.582702096499509, "learning_rate": 3.3643188983484924e-05, "loss": 2.8146, "step": 31194 }, { "epoch": 1.9364951269476691, "grad_norm": 0.5250459980114573, "learning_rate": 3.363977621817838e-05, "loss": 2.7452, "step": 31195 }, { "epoch": 1.936557204047427, "grad_norm": 0.5797232367509683, "learning_rate": 3.36363635382271e-05, "loss": 2.7247, "step": 31196 }, { "epoch": 1.936619281147185, "grad_norm": 0.5405531466148222, "learning_rate": 3.3632950943648864e-05, "loss": 2.7039, "step": 31197 }, { "epoch": 1.9366813582469427, "grad_norm": 0.5193321592287463, "learning_rate": 3.362953843446152e-05, "loss": 2.7224, "step": 31198 }, { "epoch": 1.9367434353467006, "grad_norm": 0.5643093097515048, "learning_rate": 3.3626126010682845e-05, "loss": 2.7398, "step": 31199 }, { "epoch": 1.9368055124464585, "grad_norm": 0.5374566360102989, "learning_rate": 3.3622713672330635e-05, "loss": 2.7259, "step": 31200 }, { "epoch": 1.9368675895462164, "grad_norm": 0.525838072881886, "learning_rate": 3.361930141942271e-05, "loss": 2.795, "step": 31201 }, { "epoch": 1.9369296666459743, "grad_norm": 0.5158764197309446, "learning_rate": 3.361588925197686e-05, "loss": 2.6992, "step": 31202 }, { "epoch": 1.9369917437457322, "grad_norm": 0.5598109395332389, "learning_rate": 3.3612477170010904e-05, "loss": 2.7755, "step": 31203 }, { "epoch": 1.93705382084549, "grad_norm": 0.5272703896817496, "learning_rate": 3.360906517354262e-05, "loss": 2.7073, "step": 31204 }, { "epoch": 1.9371158979452479, "grad_norm": 0.5247461840748447, "learning_rate": 3.360565326258983e-05, "loss": 2.7301, "step": 31205 }, { "epoch": 1.9371779750450058, "grad_norm": 0.5480482594913715, "learning_rate": 3.360224143717032e-05, "loss": 2.6904, "step": 31206 }, { "epoch": 1.9372400521447637, "grad_norm": 0.5027852563963366, "learning_rate": 3.359882969730191e-05, "loss": 2.7416, "step": 31207 }, { "epoch": 1.9373021292445216, "grad_norm": 0.4905015285897813, "learning_rate": 3.3595418043002375e-05, "loss": 2.7709, "step": 31208 }, { "epoch": 1.9373642063442795, "grad_norm": 0.6024663501064264, "learning_rate": 3.3592006474289535e-05, "loss": 2.7154, "step": 31209 }, { "epoch": 1.9374262834440374, "grad_norm": 0.4978802869852685, "learning_rate": 3.3588594991181166e-05, "loss": 2.7718, "step": 31210 }, { "epoch": 1.9374883605437954, "grad_norm": 0.594189757071875, "learning_rate": 3.35851835936951e-05, "loss": 2.6157, "step": 31211 }, { "epoch": 1.9375504376435533, "grad_norm": 0.5328762463781683, "learning_rate": 3.358177228184911e-05, "loss": 2.7645, "step": 31212 }, { "epoch": 1.9376125147433112, "grad_norm": 0.5325064107123957, "learning_rate": 3.357836105566099e-05, "loss": 2.7384, "step": 31213 }, { "epoch": 1.9376745918430691, "grad_norm": 0.507396116476568, "learning_rate": 3.3574949915148555e-05, "loss": 2.7533, "step": 31214 }, { "epoch": 1.937736668942827, "grad_norm": 0.49385309712341696, "learning_rate": 3.357153886032958e-05, "loss": 2.7378, "step": 31215 }, { "epoch": 1.937798746042585, "grad_norm": 0.5419918850734657, "learning_rate": 3.3568127891221886e-05, "loss": 2.7235, "step": 31216 }, { "epoch": 1.9378608231423429, "grad_norm": 0.46853613743191713, "learning_rate": 3.356471700784326e-05, "loss": 2.7375, "step": 31217 }, { "epoch": 1.9379229002421008, "grad_norm": 0.5549773681167371, "learning_rate": 3.356130621021149e-05, "loss": 2.7038, "step": 31218 }, { "epoch": 1.9379849773418587, "grad_norm": 0.5144355811485456, "learning_rate": 3.355789549834438e-05, "loss": 2.7116, "step": 31219 }, { "epoch": 1.9380470544416166, "grad_norm": 0.49557944544228294, "learning_rate": 3.355448487225972e-05, "loss": 2.7109, "step": 31220 }, { "epoch": 1.9381091315413745, "grad_norm": 0.5364468736875146, "learning_rate": 3.3551074331975306e-05, "loss": 2.7688, "step": 31221 }, { "epoch": 1.9381712086411322, "grad_norm": 0.5300730964349039, "learning_rate": 3.354766387750893e-05, "loss": 2.7257, "step": 31222 }, { "epoch": 1.9382332857408902, "grad_norm": 0.5101229412894478, "learning_rate": 3.354425350887839e-05, "loss": 2.7095, "step": 31223 }, { "epoch": 1.938295362840648, "grad_norm": 0.5463414177681439, "learning_rate": 3.354084322610147e-05, "loss": 2.6395, "step": 31224 }, { "epoch": 1.938357439940406, "grad_norm": 0.5269140682917562, "learning_rate": 3.3537433029195964e-05, "loss": 2.5923, "step": 31225 }, { "epoch": 1.938419517040164, "grad_norm": 0.560061467237343, "learning_rate": 3.353402291817968e-05, "loss": 2.6968, "step": 31226 }, { "epoch": 1.9384815941399218, "grad_norm": 0.5439144379159205, "learning_rate": 3.353061289307037e-05, "loss": 2.7511, "step": 31227 }, { "epoch": 1.9385436712396795, "grad_norm": 0.5347456923476712, "learning_rate": 3.3527202953885874e-05, "loss": 2.7601, "step": 31228 }, { "epoch": 1.9386057483394374, "grad_norm": 0.48138236974950305, "learning_rate": 3.352379310064394e-05, "loss": 2.6877, "step": 31229 }, { "epoch": 1.9386678254391954, "grad_norm": 0.5455157144522561, "learning_rate": 3.352038333336239e-05, "loss": 2.7792, "step": 31230 }, { "epoch": 1.9387299025389533, "grad_norm": 0.5534643468478663, "learning_rate": 3.351697365205899e-05, "loss": 2.8114, "step": 31231 }, { "epoch": 1.9387919796387112, "grad_norm": 0.5233634213081739, "learning_rate": 3.351356405675156e-05, "loss": 2.782, "step": 31232 }, { "epoch": 1.9388540567384691, "grad_norm": 0.5894054780363134, "learning_rate": 3.351015454745785e-05, "loss": 2.7231, "step": 31233 }, { "epoch": 1.938916133838227, "grad_norm": 0.5428288505959796, "learning_rate": 3.350674512419567e-05, "loss": 2.7378, "step": 31234 }, { "epoch": 1.938978210937985, "grad_norm": 0.5301457597373189, "learning_rate": 3.35033357869828e-05, "loss": 2.7558, "step": 31235 }, { "epoch": 1.9390402880377429, "grad_norm": 0.5240291838272221, "learning_rate": 3.349992653583705e-05, "loss": 2.6717, "step": 31236 }, { "epoch": 1.9391023651375008, "grad_norm": 0.5170854923112027, "learning_rate": 3.3496517370776184e-05, "loss": 2.6984, "step": 31237 }, { "epoch": 1.9391644422372587, "grad_norm": 0.5145843581782752, "learning_rate": 3.3493108291817976e-05, "loss": 2.7189, "step": 31238 }, { "epoch": 1.9392265193370166, "grad_norm": 0.5094844583493794, "learning_rate": 3.348969929898025e-05, "loss": 2.6541, "step": 31239 }, { "epoch": 1.9392885964367745, "grad_norm": 0.5250094991914496, "learning_rate": 3.3486290392280756e-05, "loss": 2.7353, "step": 31240 }, { "epoch": 1.9393506735365325, "grad_norm": 0.5905039294852605, "learning_rate": 3.3482881571737304e-05, "loss": 2.7708, "step": 31241 }, { "epoch": 1.9394127506362904, "grad_norm": 0.5180346508882142, "learning_rate": 3.347947283736765e-05, "loss": 2.713, "step": 31242 }, { "epoch": 1.9394748277360483, "grad_norm": 0.5311761412047356, "learning_rate": 3.347606418918962e-05, "loss": 2.7715, "step": 31243 }, { "epoch": 1.9395369048358062, "grad_norm": 0.566092015936836, "learning_rate": 3.3472655627220964e-05, "loss": 2.7374, "step": 31244 }, { "epoch": 1.9395989819355641, "grad_norm": 0.5388030381437765, "learning_rate": 3.346924715147948e-05, "loss": 2.6159, "step": 31245 }, { "epoch": 1.9396610590353218, "grad_norm": 0.5745121212934698, "learning_rate": 3.3465838761982936e-05, "loss": 2.7814, "step": 31246 }, { "epoch": 1.9397231361350797, "grad_norm": 0.5488614381067277, "learning_rate": 3.3462430458749136e-05, "loss": 2.7099, "step": 31247 }, { "epoch": 1.9397852132348377, "grad_norm": 0.5505165383836947, "learning_rate": 3.3459022241795834e-05, "loss": 2.6813, "step": 31248 }, { "epoch": 1.9398472903345956, "grad_norm": 0.5429486058292533, "learning_rate": 3.345561411114084e-05, "loss": 2.8088, "step": 31249 }, { "epoch": 1.9399093674343535, "grad_norm": 0.5340476143036446, "learning_rate": 3.3452206066801925e-05, "loss": 2.7869, "step": 31250 }, { "epoch": 1.9399714445341114, "grad_norm": 0.5742818058192029, "learning_rate": 3.3448798108796874e-05, "loss": 2.7231, "step": 31251 }, { "epoch": 1.940033521633869, "grad_norm": 0.500231808333073, "learning_rate": 3.344539023714345e-05, "loss": 2.7403, "step": 31252 }, { "epoch": 1.940095598733627, "grad_norm": 0.5257648142361406, "learning_rate": 3.3441982451859457e-05, "loss": 2.7175, "step": 31253 }, { "epoch": 1.940157675833385, "grad_norm": 0.572579515840369, "learning_rate": 3.3438574752962645e-05, "loss": 2.7383, "step": 31254 }, { "epoch": 1.9402197529331429, "grad_norm": 0.5088686304239595, "learning_rate": 3.343516714047082e-05, "loss": 2.7599, "step": 31255 }, { "epoch": 1.9402818300329008, "grad_norm": 0.5770788780159771, "learning_rate": 3.3431759614401733e-05, "loss": 2.805, "step": 31256 }, { "epoch": 1.9403439071326587, "grad_norm": 0.4938736899485745, "learning_rate": 3.34283521747732e-05, "loss": 2.7128, "step": 31257 }, { "epoch": 1.9404059842324166, "grad_norm": 0.5180195491500494, "learning_rate": 3.3424944821602964e-05, "loss": 2.7156, "step": 31258 }, { "epoch": 1.9404680613321745, "grad_norm": 0.5556297895055595, "learning_rate": 3.342153755490882e-05, "loss": 2.7236, "step": 31259 }, { "epoch": 1.9405301384319324, "grad_norm": 0.49758502264837584, "learning_rate": 3.341813037470854e-05, "loss": 2.7587, "step": 31260 }, { "epoch": 1.9405922155316904, "grad_norm": 0.5699272160740642, "learning_rate": 3.34147232810199e-05, "loss": 2.6846, "step": 31261 }, { "epoch": 1.9406542926314483, "grad_norm": 0.5600718968082771, "learning_rate": 3.341131627386067e-05, "loss": 2.7955, "step": 31262 }, { "epoch": 1.9407163697312062, "grad_norm": 0.4821251949223863, "learning_rate": 3.340790935324863e-05, "loss": 2.7428, "step": 31263 }, { "epoch": 1.9407784468309641, "grad_norm": 0.5621952449907587, "learning_rate": 3.3404502519201565e-05, "loss": 2.6993, "step": 31264 }, { "epoch": 1.940840523930722, "grad_norm": 0.5233963622823526, "learning_rate": 3.340109577173722e-05, "loss": 2.6957, "step": 31265 }, { "epoch": 1.94090260103048, "grad_norm": 0.5440544963951314, "learning_rate": 3.339768911087341e-05, "loss": 2.7534, "step": 31266 }, { "epoch": 1.9409646781302379, "grad_norm": 0.5564845887044089, "learning_rate": 3.339428253662787e-05, "loss": 2.777, "step": 31267 }, { "epoch": 1.9410267552299958, "grad_norm": 0.5285835949821291, "learning_rate": 3.33908760490184e-05, "loss": 2.7311, "step": 31268 }, { "epoch": 1.9410888323297537, "grad_norm": 0.5598325264796108, "learning_rate": 3.338746964806275e-05, "loss": 2.8009, "step": 31269 }, { "epoch": 1.9411509094295114, "grad_norm": 0.5840216277115314, "learning_rate": 3.3384063333778716e-05, "loss": 2.6552, "step": 31270 }, { "epoch": 1.9412129865292693, "grad_norm": 0.5539186548907357, "learning_rate": 3.338065710618404e-05, "loss": 2.7275, "step": 31271 }, { "epoch": 1.9412750636290272, "grad_norm": 0.6165990801022052, "learning_rate": 3.337725096529653e-05, "loss": 2.6526, "step": 31272 }, { "epoch": 1.9413371407287852, "grad_norm": 0.5149727944715797, "learning_rate": 3.337384491113394e-05, "loss": 2.738, "step": 31273 }, { "epoch": 1.941399217828543, "grad_norm": 0.6500040916167372, "learning_rate": 3.337043894371401e-05, "loss": 2.7802, "step": 31274 }, { "epoch": 1.941461294928301, "grad_norm": 0.5321169399092821, "learning_rate": 3.336703306305456e-05, "loss": 2.8261, "step": 31275 }, { "epoch": 1.9415233720280587, "grad_norm": 0.577329216801274, "learning_rate": 3.336362726917332e-05, "loss": 2.6731, "step": 31276 }, { "epoch": 1.9415854491278166, "grad_norm": 0.6156461276384071, "learning_rate": 3.336022156208809e-05, "loss": 2.6452, "step": 31277 }, { "epoch": 1.9416475262275745, "grad_norm": 0.5741109162888662, "learning_rate": 3.335681594181661e-05, "loss": 2.7912, "step": 31278 }, { "epoch": 1.9417096033273324, "grad_norm": 0.5700710517117231, "learning_rate": 3.335341040837667e-05, "loss": 2.7478, "step": 31279 }, { "epoch": 1.9417716804270904, "grad_norm": 0.5212574725907579, "learning_rate": 3.335000496178602e-05, "loss": 2.5939, "step": 31280 }, { "epoch": 1.9418337575268483, "grad_norm": 0.5523327557927766, "learning_rate": 3.3346599602062434e-05, "loss": 2.7329, "step": 31281 }, { "epoch": 1.9418958346266062, "grad_norm": 0.5824132600143315, "learning_rate": 3.3343194329223695e-05, "loss": 2.7464, "step": 31282 }, { "epoch": 1.9419579117263641, "grad_norm": 0.46862187223899143, "learning_rate": 3.333978914328755e-05, "loss": 2.6973, "step": 31283 }, { "epoch": 1.942019988826122, "grad_norm": 0.5771056286834956, "learning_rate": 3.3336384044271774e-05, "loss": 2.7678, "step": 31284 }, { "epoch": 1.94208206592588, "grad_norm": 0.499037352366082, "learning_rate": 3.333297903219412e-05, "loss": 2.685, "step": 31285 }, { "epoch": 1.9421441430256379, "grad_norm": 0.547578930804541, "learning_rate": 3.332957410707237e-05, "loss": 2.7019, "step": 31286 }, { "epoch": 1.9422062201253958, "grad_norm": 0.5853432502958289, "learning_rate": 3.3326169268924275e-05, "loss": 2.6377, "step": 31287 }, { "epoch": 1.9422682972251537, "grad_norm": 0.537685626157843, "learning_rate": 3.332276451776759e-05, "loss": 2.658, "step": 31288 }, { "epoch": 1.9423303743249116, "grad_norm": 0.5543484793494132, "learning_rate": 3.331935985362011e-05, "loss": 2.6508, "step": 31289 }, { "epoch": 1.9423924514246695, "grad_norm": 0.4779651823531078, "learning_rate": 3.331595527649957e-05, "loss": 2.6437, "step": 31290 }, { "epoch": 1.9424545285244275, "grad_norm": 0.5297557629284856, "learning_rate": 3.331255078642374e-05, "loss": 2.6773, "step": 31291 }, { "epoch": 1.9425166056241854, "grad_norm": 0.5207665464177974, "learning_rate": 3.330914638341038e-05, "loss": 2.6677, "step": 31292 }, { "epoch": 1.9425786827239433, "grad_norm": 0.56138264093357, "learning_rate": 3.3305742067477264e-05, "loss": 2.661, "step": 31293 }, { "epoch": 1.942640759823701, "grad_norm": 0.5813768535544277, "learning_rate": 3.330233783864214e-05, "loss": 2.7342, "step": 31294 }, { "epoch": 1.942702836923459, "grad_norm": 0.5311796197832982, "learning_rate": 3.3298933696922776e-05, "loss": 2.7433, "step": 31295 }, { "epoch": 1.9427649140232168, "grad_norm": 0.533347784407854, "learning_rate": 3.329552964233692e-05, "loss": 2.6785, "step": 31296 }, { "epoch": 1.9428269911229747, "grad_norm": 0.5102208761083412, "learning_rate": 3.3292125674902355e-05, "loss": 2.7482, "step": 31297 }, { "epoch": 1.9428890682227327, "grad_norm": 0.5487506852547575, "learning_rate": 3.328872179463683e-05, "loss": 2.7692, "step": 31298 }, { "epoch": 1.9429511453224906, "grad_norm": 0.5262177490536574, "learning_rate": 3.328531800155808e-05, "loss": 2.7497, "step": 31299 }, { "epoch": 1.9430132224222483, "grad_norm": 0.5367539495616883, "learning_rate": 3.3281914295683897e-05, "loss": 2.6369, "step": 31300 }, { "epoch": 1.9430752995220062, "grad_norm": 0.5292819256800461, "learning_rate": 3.3278510677032016e-05, "loss": 2.7598, "step": 31301 }, { "epoch": 1.9431373766217641, "grad_norm": 0.5901218036822811, "learning_rate": 3.327510714562022e-05, "loss": 2.7019, "step": 31302 }, { "epoch": 1.943199453721522, "grad_norm": 0.5113705913956968, "learning_rate": 3.3271703701466226e-05, "loss": 2.7514, "step": 31303 }, { "epoch": 1.94326153082128, "grad_norm": 0.6176742372757383, "learning_rate": 3.326830034458784e-05, "loss": 2.7506, "step": 31304 }, { "epoch": 1.9433236079210379, "grad_norm": 0.5954686438170763, "learning_rate": 3.3264897075002774e-05, "loss": 2.7285, "step": 31305 }, { "epoch": 1.9433856850207958, "grad_norm": 0.5908396315765782, "learning_rate": 3.326149389272881e-05, "loss": 2.8015, "step": 31306 }, { "epoch": 1.9434477621205537, "grad_norm": 0.5868790754150327, "learning_rate": 3.32580907977837e-05, "loss": 2.7535, "step": 31307 }, { "epoch": 1.9435098392203116, "grad_norm": 0.5725833674703665, "learning_rate": 3.325468779018519e-05, "loss": 2.796, "step": 31308 }, { "epoch": 1.9435719163200695, "grad_norm": 0.6142347419421132, "learning_rate": 3.325128486995105e-05, "loss": 2.792, "step": 31309 }, { "epoch": 1.9436339934198275, "grad_norm": 0.5911079148106136, "learning_rate": 3.3247882037099e-05, "loss": 2.7745, "step": 31310 }, { "epoch": 1.9436960705195854, "grad_norm": 0.5247747789668966, "learning_rate": 3.324447929164683e-05, "loss": 2.6942, "step": 31311 }, { "epoch": 1.9437581476193433, "grad_norm": 0.5618804592762682, "learning_rate": 3.324107663361228e-05, "loss": 2.757, "step": 31312 }, { "epoch": 1.9438202247191012, "grad_norm": 0.5679081702162376, "learning_rate": 3.323767406301308e-05, "loss": 2.7545, "step": 31313 }, { "epoch": 1.9438823018188591, "grad_norm": 0.5063707278374414, "learning_rate": 3.323427157986703e-05, "loss": 2.7373, "step": 31314 }, { "epoch": 1.943944378918617, "grad_norm": 0.6097255823158169, "learning_rate": 3.323086918419184e-05, "loss": 2.7339, "step": 31315 }, { "epoch": 1.944006456018375, "grad_norm": 0.5090964298637881, "learning_rate": 3.322746687600529e-05, "loss": 2.783, "step": 31316 }, { "epoch": 1.9440685331181329, "grad_norm": 0.5592935087206848, "learning_rate": 3.3224064655325106e-05, "loss": 2.7618, "step": 31317 }, { "epoch": 1.9441306102178906, "grad_norm": 0.5533848900885975, "learning_rate": 3.3220662522169054e-05, "loss": 2.7259, "step": 31318 }, { "epoch": 1.9441926873176485, "grad_norm": 0.5811222838504273, "learning_rate": 3.321726047655488e-05, "loss": 2.7287, "step": 31319 }, { "epoch": 1.9442547644174064, "grad_norm": 0.5711101806197438, "learning_rate": 3.321385851850033e-05, "loss": 2.767, "step": 31320 }, { "epoch": 1.9443168415171643, "grad_norm": 0.4974963717352289, "learning_rate": 3.3210456648023156e-05, "loss": 2.6167, "step": 31321 }, { "epoch": 1.9443789186169222, "grad_norm": 0.531551836322838, "learning_rate": 3.3207054865141106e-05, "loss": 2.7008, "step": 31322 }, { "epoch": 1.9444409957166802, "grad_norm": 0.541925484616587, "learning_rate": 3.320365316987194e-05, "loss": 2.7231, "step": 31323 }, { "epoch": 1.9445030728164379, "grad_norm": 0.46602397441179194, "learning_rate": 3.320025156223337e-05, "loss": 2.6572, "step": 31324 }, { "epoch": 1.9445651499161958, "grad_norm": 0.542019240245459, "learning_rate": 3.3196850042243184e-05, "loss": 2.7326, "step": 31325 }, { "epoch": 1.9446272270159537, "grad_norm": 0.5720044995210978, "learning_rate": 3.3193448609919095e-05, "loss": 2.6335, "step": 31326 }, { "epoch": 1.9446893041157116, "grad_norm": 0.5611056689380364, "learning_rate": 3.319004726527888e-05, "loss": 2.8264, "step": 31327 }, { "epoch": 1.9447513812154695, "grad_norm": 0.5378418522859425, "learning_rate": 3.318664600834025e-05, "loss": 2.7266, "step": 31328 }, { "epoch": 1.9448134583152275, "grad_norm": 0.5803871230881026, "learning_rate": 3.318324483912099e-05, "loss": 2.7225, "step": 31329 }, { "epoch": 1.9448755354149854, "grad_norm": 0.5490302531587075, "learning_rate": 3.3179843757638806e-05, "loss": 2.7571, "step": 31330 }, { "epoch": 1.9449376125147433, "grad_norm": 0.5464364336865766, "learning_rate": 3.3176442763911474e-05, "loss": 2.7148, "step": 31331 }, { "epoch": 1.9449996896145012, "grad_norm": 0.5270942945685039, "learning_rate": 3.317304185795672e-05, "loss": 2.7521, "step": 31332 }, { "epoch": 1.9450617667142591, "grad_norm": 0.5833726568818675, "learning_rate": 3.3169641039792294e-05, "loss": 2.8022, "step": 31333 }, { "epoch": 1.945123843814017, "grad_norm": 0.5638888485391202, "learning_rate": 3.316624030943594e-05, "loss": 2.7424, "step": 31334 }, { "epoch": 1.945185920913775, "grad_norm": 0.5821595786540766, "learning_rate": 3.316283966690538e-05, "loss": 2.7184, "step": 31335 }, { "epoch": 1.9452479980135329, "grad_norm": 0.5275767906394657, "learning_rate": 3.315943911221839e-05, "loss": 2.6774, "step": 31336 }, { "epoch": 1.9453100751132908, "grad_norm": 0.5398561467956934, "learning_rate": 3.3156038645392685e-05, "loss": 2.7326, "step": 31337 }, { "epoch": 1.9453721522130487, "grad_norm": 0.5599279941518809, "learning_rate": 3.315263826644602e-05, "loss": 2.7775, "step": 31338 }, { "epoch": 1.9454342293128066, "grad_norm": 0.5080838285519563, "learning_rate": 3.314923797539612e-05, "loss": 2.7321, "step": 31339 }, { "epoch": 1.9454963064125645, "grad_norm": 0.48799491115949756, "learning_rate": 3.314583777226075e-05, "loss": 2.7143, "step": 31340 }, { "epoch": 1.9455583835123225, "grad_norm": 0.5226388503572574, "learning_rate": 3.3142437657057624e-05, "loss": 2.7611, "step": 31341 }, { "epoch": 1.9456204606120802, "grad_norm": 0.4960966185239026, "learning_rate": 3.31390376298045e-05, "loss": 2.6751, "step": 31342 }, { "epoch": 1.945682537711838, "grad_norm": 0.5459098746478414, "learning_rate": 3.3135637690519103e-05, "loss": 2.7367, "step": 31343 }, { "epoch": 1.945744614811596, "grad_norm": 0.5213778097588705, "learning_rate": 3.313223783921919e-05, "loss": 2.6954, "step": 31344 }, { "epoch": 1.945806691911354, "grad_norm": 0.5277738442051292, "learning_rate": 3.312883807592249e-05, "loss": 2.7525, "step": 31345 }, { "epoch": 1.9458687690111118, "grad_norm": 0.5956553205476156, "learning_rate": 3.31254384006467e-05, "loss": 2.7384, "step": 31346 }, { "epoch": 1.9459308461108697, "grad_norm": 0.5313025055912213, "learning_rate": 3.312203881340963e-05, "loss": 2.684, "step": 31347 }, { "epoch": 1.9459929232106274, "grad_norm": 0.5344917738632077, "learning_rate": 3.311863931422898e-05, "loss": 2.7829, "step": 31348 }, { "epoch": 1.9460550003103854, "grad_norm": 0.5435261780927624, "learning_rate": 3.311523990312248e-05, "loss": 2.6442, "step": 31349 }, { "epoch": 1.9461170774101433, "grad_norm": 0.5640083065046911, "learning_rate": 3.311184058010788e-05, "loss": 2.6766, "step": 31350 }, { "epoch": 1.9461791545099012, "grad_norm": 0.496494026600496, "learning_rate": 3.310844134520289e-05, "loss": 2.6007, "step": 31351 }, { "epoch": 1.9462412316096591, "grad_norm": 0.5709373239016875, "learning_rate": 3.310504219842528e-05, "loss": 2.7086, "step": 31352 }, { "epoch": 1.946303308709417, "grad_norm": 0.5327852449130295, "learning_rate": 3.310164313979275e-05, "loss": 2.6997, "step": 31353 }, { "epoch": 1.946365385809175, "grad_norm": 0.5966204474927037, "learning_rate": 3.3098244169323067e-05, "loss": 2.6854, "step": 31354 }, { "epoch": 1.9464274629089329, "grad_norm": 0.484980343366257, "learning_rate": 3.309484528703393e-05, "loss": 2.6839, "step": 31355 }, { "epoch": 1.9464895400086908, "grad_norm": 0.5728101255593459, "learning_rate": 3.309144649294311e-05, "loss": 2.7567, "step": 31356 }, { "epoch": 1.9465516171084487, "grad_norm": 0.5213680351869496, "learning_rate": 3.30880477870683e-05, "loss": 2.715, "step": 31357 }, { "epoch": 1.9466136942082066, "grad_norm": 0.5354098811876157, "learning_rate": 3.308464916942726e-05, "loss": 2.7072, "step": 31358 }, { "epoch": 1.9466757713079645, "grad_norm": 0.5458000790558559, "learning_rate": 3.3081250640037707e-05, "loss": 2.8034, "step": 31359 }, { "epoch": 1.9467378484077225, "grad_norm": 0.5085822799959651, "learning_rate": 3.307785219891737e-05, "loss": 2.7011, "step": 31360 }, { "epoch": 1.9467999255074804, "grad_norm": 0.5504863301138198, "learning_rate": 3.307445384608401e-05, "loss": 2.6781, "step": 31361 }, { "epoch": 1.9468620026072383, "grad_norm": 0.5772511347300667, "learning_rate": 3.307105558155531e-05, "loss": 2.7513, "step": 31362 }, { "epoch": 1.9469240797069962, "grad_norm": 0.5288772608871659, "learning_rate": 3.306765740534903e-05, "loss": 2.7948, "step": 31363 }, { "epoch": 1.9469861568067541, "grad_norm": 0.5884917977959851, "learning_rate": 3.3064259317482886e-05, "loss": 2.7592, "step": 31364 }, { "epoch": 1.947048233906512, "grad_norm": 0.5539687160462452, "learning_rate": 3.3060861317974624e-05, "loss": 2.7281, "step": 31365 }, { "epoch": 1.9471103110062697, "grad_norm": 0.5519385927829026, "learning_rate": 3.305746340684195e-05, "loss": 2.7099, "step": 31366 }, { "epoch": 1.9471723881060277, "grad_norm": 0.5504285075117034, "learning_rate": 3.3054065584102614e-05, "loss": 2.779, "step": 31367 }, { "epoch": 1.9472344652057856, "grad_norm": 0.6032326290966861, "learning_rate": 3.305066784977432e-05, "loss": 2.698, "step": 31368 }, { "epoch": 1.9472965423055435, "grad_norm": 0.5084290502932822, "learning_rate": 3.3047270203874816e-05, "loss": 2.6826, "step": 31369 }, { "epoch": 1.9473586194053014, "grad_norm": 0.574627529663997, "learning_rate": 3.30438726464218e-05, "loss": 2.7929, "step": 31370 }, { "epoch": 1.9474206965050593, "grad_norm": 0.5156475986961514, "learning_rate": 3.304047517743304e-05, "loss": 2.6546, "step": 31371 }, { "epoch": 1.947482773604817, "grad_norm": 0.5642083769282888, "learning_rate": 3.303707779692623e-05, "loss": 2.7563, "step": 31372 }, { "epoch": 1.947544850704575, "grad_norm": 0.48812714242892397, "learning_rate": 3.30336805049191e-05, "loss": 2.7093, "step": 31373 }, { "epoch": 1.9476069278043329, "grad_norm": 0.6155698799979126, "learning_rate": 3.303028330142938e-05, "loss": 2.7369, "step": 31374 }, { "epoch": 1.9476690049040908, "grad_norm": 0.5157584682311176, "learning_rate": 3.302688618647479e-05, "loss": 2.698, "step": 31375 }, { "epoch": 1.9477310820038487, "grad_norm": 0.6105168965521948, "learning_rate": 3.302348916007306e-05, "loss": 2.7741, "step": 31376 }, { "epoch": 1.9477931591036066, "grad_norm": 0.5614003452889768, "learning_rate": 3.302009222224189e-05, "loss": 2.6904, "step": 31377 }, { "epoch": 1.9478552362033645, "grad_norm": 0.5922411016195904, "learning_rate": 3.301669537299905e-05, "loss": 2.6671, "step": 31378 }, { "epoch": 1.9479173133031225, "grad_norm": 0.6348611143621208, "learning_rate": 3.30132986123622e-05, "loss": 2.7128, "step": 31379 }, { "epoch": 1.9479793904028804, "grad_norm": 0.5684312073714223, "learning_rate": 3.300990194034911e-05, "loss": 2.6024, "step": 31380 }, { "epoch": 1.9480414675026383, "grad_norm": 0.6246359957234265, "learning_rate": 3.300650535697749e-05, "loss": 2.8116, "step": 31381 }, { "epoch": 1.9481035446023962, "grad_norm": 0.5419787705968842, "learning_rate": 3.300310886226505e-05, "loss": 2.6575, "step": 31382 }, { "epoch": 1.9481656217021541, "grad_norm": 0.5511285755390164, "learning_rate": 3.2999712456229524e-05, "loss": 2.7823, "step": 31383 }, { "epoch": 1.948227698801912, "grad_norm": 0.5674517748839987, "learning_rate": 3.2996316138888634e-05, "loss": 2.6924, "step": 31384 }, { "epoch": 1.94828977590167, "grad_norm": 0.5434036982741463, "learning_rate": 3.299291991026007e-05, "loss": 2.6884, "step": 31385 }, { "epoch": 1.9483518530014279, "grad_norm": 0.5652149431728208, "learning_rate": 3.298952377036159e-05, "loss": 2.6217, "step": 31386 }, { "epoch": 1.9484139301011858, "grad_norm": 0.5714802739901662, "learning_rate": 3.298612771921088e-05, "loss": 2.7679, "step": 31387 }, { "epoch": 1.9484760072009437, "grad_norm": 0.5385154491464673, "learning_rate": 3.298273175682568e-05, "loss": 2.7043, "step": 31388 }, { "epoch": 1.9485380843007016, "grad_norm": 0.5984754495564939, "learning_rate": 3.29793358832237e-05, "loss": 2.7285, "step": 31389 }, { "epoch": 1.9486001614004593, "grad_norm": 0.5454924682688145, "learning_rate": 3.2975940098422653e-05, "loss": 2.7258, "step": 31390 }, { "epoch": 1.9486622385002172, "grad_norm": 0.579836605668417, "learning_rate": 3.297254440244025e-05, "loss": 2.7548, "step": 31391 }, { "epoch": 1.9487243155999752, "grad_norm": 0.5730522920005341, "learning_rate": 3.296914879529424e-05, "loss": 2.7385, "step": 31392 }, { "epoch": 1.948786392699733, "grad_norm": 0.6031760584511228, "learning_rate": 3.296575327700229e-05, "loss": 2.7273, "step": 31393 }, { "epoch": 1.948848469799491, "grad_norm": 0.5674937889834747, "learning_rate": 3.296235784758216e-05, "loss": 2.7547, "step": 31394 }, { "epoch": 1.948910546899249, "grad_norm": 0.5069012021066872, "learning_rate": 3.295896250705154e-05, "loss": 2.7282, "step": 31395 }, { "epoch": 1.9489726239990066, "grad_norm": 0.5511240332134901, "learning_rate": 3.295556725542814e-05, "loss": 2.73, "step": 31396 }, { "epoch": 1.9490347010987645, "grad_norm": 0.5369652572098207, "learning_rate": 3.2952172092729694e-05, "loss": 2.708, "step": 31397 }, { "epoch": 1.9490967781985225, "grad_norm": 0.5568135846112772, "learning_rate": 3.29487770189739e-05, "loss": 2.7098, "step": 31398 }, { "epoch": 1.9491588552982804, "grad_norm": 0.5365357266002204, "learning_rate": 3.294538203417848e-05, "loss": 2.7759, "step": 31399 }, { "epoch": 1.9492209323980383, "grad_norm": 0.5359084140903618, "learning_rate": 3.294198713836113e-05, "loss": 2.8041, "step": 31400 }, { "epoch": 1.9492830094977962, "grad_norm": 0.6056454800507892, "learning_rate": 3.2938592331539587e-05, "loss": 2.6741, "step": 31401 }, { "epoch": 1.9493450865975541, "grad_norm": 0.5338762387310673, "learning_rate": 3.293519761373154e-05, "loss": 2.7603, "step": 31402 }, { "epoch": 1.949407163697312, "grad_norm": 0.6119132121372737, "learning_rate": 3.293180298495472e-05, "loss": 2.6406, "step": 31403 }, { "epoch": 1.94946924079707, "grad_norm": 0.5290825328875925, "learning_rate": 3.292840844522682e-05, "loss": 2.6834, "step": 31404 }, { "epoch": 1.9495313178968279, "grad_norm": 0.6167307503583523, "learning_rate": 3.2925013994565564e-05, "loss": 2.804, "step": 31405 }, { "epoch": 1.9495933949965858, "grad_norm": 0.5411668421534818, "learning_rate": 3.292161963298864e-05, "loss": 2.7471, "step": 31406 }, { "epoch": 1.9496554720963437, "grad_norm": 0.5061449401066695, "learning_rate": 3.2918225360513785e-05, "loss": 2.6778, "step": 31407 }, { "epoch": 1.9497175491961016, "grad_norm": 0.562464289061467, "learning_rate": 3.29148311771587e-05, "loss": 2.6917, "step": 31408 }, { "epoch": 1.9497796262958595, "grad_norm": 0.5595138015306155, "learning_rate": 3.291143708294108e-05, "loss": 2.6991, "step": 31409 }, { "epoch": 1.9498417033956175, "grad_norm": 0.5196789983167531, "learning_rate": 3.290804307787865e-05, "loss": 2.7049, "step": 31410 }, { "epoch": 1.9499037804953754, "grad_norm": 0.5838459067621053, "learning_rate": 3.290464916198908e-05, "loss": 2.7062, "step": 31411 }, { "epoch": 1.9499658575951333, "grad_norm": 0.5517722338926998, "learning_rate": 3.290125533529012e-05, "loss": 2.6661, "step": 31412 }, { "epoch": 1.9500279346948912, "grad_norm": 0.5096354409674118, "learning_rate": 3.2897861597799475e-05, "loss": 2.7225, "step": 31413 }, { "epoch": 1.950090011794649, "grad_norm": 0.5551888255496757, "learning_rate": 3.289446794953482e-05, "loss": 2.7902, "step": 31414 }, { "epoch": 1.9501520888944068, "grad_norm": 0.5300813330466811, "learning_rate": 3.28910743905139e-05, "loss": 2.7401, "step": 31415 }, { "epoch": 1.9502141659941648, "grad_norm": 0.5021826668241468, "learning_rate": 3.2887680920754384e-05, "loss": 2.6658, "step": 31416 }, { "epoch": 1.9502762430939227, "grad_norm": 0.6110200196949379, "learning_rate": 3.288428754027399e-05, "loss": 2.7373, "step": 31417 }, { "epoch": 1.9503383201936806, "grad_norm": 0.5081152380349275, "learning_rate": 3.2880894249090424e-05, "loss": 2.7315, "step": 31418 }, { "epoch": 1.9504003972934385, "grad_norm": 0.5399563331079288, "learning_rate": 3.287750104722139e-05, "loss": 2.6791, "step": 31419 }, { "epoch": 1.9504624743931962, "grad_norm": 0.5558536521204186, "learning_rate": 3.2874107934684604e-05, "loss": 2.6381, "step": 31420 }, { "epoch": 1.9505245514929541, "grad_norm": 0.541775579384663, "learning_rate": 3.287071491149773e-05, "loss": 2.7179, "step": 31421 }, { "epoch": 1.950586628592712, "grad_norm": 0.5768015378918298, "learning_rate": 3.286732197767851e-05, "loss": 2.7443, "step": 31422 }, { "epoch": 1.95064870569247, "grad_norm": 0.517584084530475, "learning_rate": 3.286392913324463e-05, "loss": 2.7371, "step": 31423 }, { "epoch": 1.9507107827922279, "grad_norm": 0.5872201993890979, "learning_rate": 3.286053637821379e-05, "loss": 2.7245, "step": 31424 }, { "epoch": 1.9507728598919858, "grad_norm": 0.6012916450591241, "learning_rate": 3.285714371260368e-05, "loss": 2.7142, "step": 31425 }, { "epoch": 1.9508349369917437, "grad_norm": 0.5856098530539735, "learning_rate": 3.285375113643203e-05, "loss": 2.6203, "step": 31426 }, { "epoch": 1.9508970140915016, "grad_norm": 0.5909706365855674, "learning_rate": 3.285035864971651e-05, "loss": 2.6957, "step": 31427 }, { "epoch": 1.9509590911912595, "grad_norm": 0.6277790037099108, "learning_rate": 3.284696625247484e-05, "loss": 2.693, "step": 31428 }, { "epoch": 1.9510211682910175, "grad_norm": 0.4976964578000837, "learning_rate": 3.2843573944724704e-05, "loss": 2.7406, "step": 31429 }, { "epoch": 1.9510832453907754, "grad_norm": 0.5770465225111638, "learning_rate": 3.284018172648382e-05, "loss": 2.7525, "step": 31430 }, { "epoch": 1.9511453224905333, "grad_norm": 0.4964261030550375, "learning_rate": 3.283678959776986e-05, "loss": 2.7636, "step": 31431 }, { "epoch": 1.9512073995902912, "grad_norm": 0.540697934076586, "learning_rate": 3.283339755860055e-05, "loss": 2.8042, "step": 31432 }, { "epoch": 1.9512694766900491, "grad_norm": 0.52618014209486, "learning_rate": 3.2830005608993565e-05, "loss": 2.6866, "step": 31433 }, { "epoch": 1.951331553789807, "grad_norm": 0.5242828830198563, "learning_rate": 3.282661374896661e-05, "loss": 2.6779, "step": 31434 }, { "epoch": 1.951393630889565, "grad_norm": 0.5757775201767259, "learning_rate": 3.282322197853738e-05, "loss": 2.6344, "step": 31435 }, { "epoch": 1.9514557079893229, "grad_norm": 0.5466647152555728, "learning_rate": 3.2819830297723566e-05, "loss": 2.6974, "step": 31436 }, { "epoch": 1.9515177850890808, "grad_norm": 0.5749664369407332, "learning_rate": 3.281643870654287e-05, "loss": 2.737, "step": 31437 }, { "epoch": 1.9515798621888385, "grad_norm": 0.5974947143863576, "learning_rate": 3.281304720501298e-05, "loss": 2.7626, "step": 31438 }, { "epoch": 1.9516419392885964, "grad_norm": 0.5659502759109635, "learning_rate": 3.2809655793151605e-05, "loss": 2.791, "step": 31439 }, { "epoch": 1.9517040163883543, "grad_norm": 0.5327531641701042, "learning_rate": 3.2806264470976414e-05, "loss": 2.7894, "step": 31440 }, { "epoch": 1.9517660934881123, "grad_norm": 0.5019712333529902, "learning_rate": 3.2802873238505135e-05, "loss": 2.6867, "step": 31441 }, { "epoch": 1.9518281705878702, "grad_norm": 0.5103827646894835, "learning_rate": 3.279948209575542e-05, "loss": 2.8164, "step": 31442 }, { "epoch": 1.951890247687628, "grad_norm": 0.4754798191122184, "learning_rate": 3.2796091042745e-05, "loss": 2.7299, "step": 31443 }, { "epoch": 1.9519523247873858, "grad_norm": 0.5616109497873515, "learning_rate": 3.279270007949153e-05, "loss": 2.6581, "step": 31444 }, { "epoch": 1.9520144018871437, "grad_norm": 0.5239784804377058, "learning_rate": 3.2789309206012744e-05, "loss": 2.7879, "step": 31445 }, { "epoch": 1.9520764789869016, "grad_norm": 0.5007528657055128, "learning_rate": 3.278591842232629e-05, "loss": 2.7882, "step": 31446 }, { "epoch": 1.9521385560866595, "grad_norm": 0.5745890714272652, "learning_rate": 3.2782527728449894e-05, "loss": 2.8397, "step": 31447 }, { "epoch": 1.9522006331864175, "grad_norm": 0.5162249744713632, "learning_rate": 3.2779137124401215e-05, "loss": 2.6565, "step": 31448 }, { "epoch": 1.9522627102861754, "grad_norm": 0.5575873371959796, "learning_rate": 3.277574661019798e-05, "loss": 2.7496, "step": 31449 }, { "epoch": 1.9523247873859333, "grad_norm": 0.569622039884335, "learning_rate": 3.277235618585783e-05, "loss": 2.7726, "step": 31450 }, { "epoch": 1.9523868644856912, "grad_norm": 0.587607432748957, "learning_rate": 3.27689658513985e-05, "loss": 2.7422, "step": 31451 }, { "epoch": 1.9524489415854491, "grad_norm": 0.610997021987825, "learning_rate": 3.276557560683765e-05, "loss": 2.7493, "step": 31452 }, { "epoch": 1.952511018685207, "grad_norm": 0.6461280693708958, "learning_rate": 3.2762185452192986e-05, "loss": 2.8111, "step": 31453 }, { "epoch": 1.952573095784965, "grad_norm": 0.5968856331242622, "learning_rate": 3.275879538748217e-05, "loss": 2.7969, "step": 31454 }, { "epoch": 1.9526351728847229, "grad_norm": 0.6683286400064962, "learning_rate": 3.275540541272292e-05, "loss": 2.7985, "step": 31455 }, { "epoch": 1.9526972499844808, "grad_norm": 0.5664528137009196, "learning_rate": 3.275201552793289e-05, "loss": 2.7486, "step": 31456 }, { "epoch": 1.9527593270842387, "grad_norm": 0.5881838906667569, "learning_rate": 3.274862573312979e-05, "loss": 2.7465, "step": 31457 }, { "epoch": 1.9528214041839966, "grad_norm": 0.5566434358224673, "learning_rate": 3.2745236028331305e-05, "loss": 2.673, "step": 31458 }, { "epoch": 1.9528834812837546, "grad_norm": 0.5352274715060226, "learning_rate": 3.27418464135551e-05, "loss": 2.6805, "step": 31459 }, { "epoch": 1.9529455583835125, "grad_norm": 0.5807523822563139, "learning_rate": 3.273845688881888e-05, "loss": 2.7212, "step": 31460 }, { "epoch": 1.9530076354832704, "grad_norm": 0.526345133982941, "learning_rate": 3.2735067454140314e-05, "loss": 2.6736, "step": 31461 }, { "epoch": 1.953069712583028, "grad_norm": 0.5542058868679243, "learning_rate": 3.27316781095371e-05, "loss": 2.7016, "step": 31462 }, { "epoch": 1.953131789682786, "grad_norm": 0.5498055818865533, "learning_rate": 3.27282888550269e-05, "loss": 2.6693, "step": 31463 }, { "epoch": 1.953193866782544, "grad_norm": 0.55025494908729, "learning_rate": 3.2724899690627425e-05, "loss": 2.6848, "step": 31464 }, { "epoch": 1.9532559438823018, "grad_norm": 0.5787912602063332, "learning_rate": 3.2721510616356334e-05, "loss": 2.5962, "step": 31465 }, { "epoch": 1.9533180209820598, "grad_norm": 0.5782461812626755, "learning_rate": 3.271812163223132e-05, "loss": 2.7431, "step": 31466 }, { "epoch": 1.9533800980818177, "grad_norm": 0.5635473249883508, "learning_rate": 3.271473273827005e-05, "loss": 2.7949, "step": 31467 }, { "epoch": 1.9534421751815754, "grad_norm": 0.6631036455489409, "learning_rate": 3.2711343934490235e-05, "loss": 2.7697, "step": 31468 }, { "epoch": 1.9535042522813333, "grad_norm": 0.53371000126643, "learning_rate": 3.270795522090952e-05, "loss": 2.7082, "step": 31469 }, { "epoch": 1.9535663293810912, "grad_norm": 0.5851833321959609, "learning_rate": 3.2704566597545605e-05, "loss": 2.7125, "step": 31470 }, { "epoch": 1.9536284064808491, "grad_norm": 0.575448386924512, "learning_rate": 3.270117806441617e-05, "loss": 2.7265, "step": 31471 }, { "epoch": 1.953690483580607, "grad_norm": 0.5492081599075965, "learning_rate": 3.269778962153888e-05, "loss": 2.7846, "step": 31472 }, { "epoch": 1.953752560680365, "grad_norm": 0.6334278035186264, "learning_rate": 3.269440126893143e-05, "loss": 2.6849, "step": 31473 }, { "epoch": 1.9538146377801229, "grad_norm": 0.5678043867506765, "learning_rate": 3.269101300661148e-05, "loss": 2.7112, "step": 31474 }, { "epoch": 1.9538767148798808, "grad_norm": 0.5622469279763787, "learning_rate": 3.268762483459672e-05, "loss": 2.7253, "step": 31475 }, { "epoch": 1.9539387919796387, "grad_norm": 0.6213644625334876, "learning_rate": 3.268423675290483e-05, "loss": 2.6318, "step": 31476 }, { "epoch": 1.9540008690793966, "grad_norm": 0.5593868574025861, "learning_rate": 3.268084876155346e-05, "loss": 2.7205, "step": 31477 }, { "epoch": 1.9540629461791545, "grad_norm": 0.6463778082381405, "learning_rate": 3.2677460860560325e-05, "loss": 2.8046, "step": 31478 }, { "epoch": 1.9541250232789125, "grad_norm": 0.5911817469729806, "learning_rate": 3.2674073049943074e-05, "loss": 2.752, "step": 31479 }, { "epoch": 1.9541871003786704, "grad_norm": 0.5790350427103435, "learning_rate": 3.26706853297194e-05, "loss": 2.6737, "step": 31480 }, { "epoch": 1.9542491774784283, "grad_norm": 0.6213736473242203, "learning_rate": 3.266729769990696e-05, "loss": 2.741, "step": 31481 }, { "epoch": 1.9543112545781862, "grad_norm": 0.5117213125383119, "learning_rate": 3.266391016052344e-05, "loss": 2.6669, "step": 31482 }, { "epoch": 1.9543733316779441, "grad_norm": 0.6007008512396613, "learning_rate": 3.2660522711586514e-05, "loss": 2.7123, "step": 31483 }, { "epoch": 1.954435408777702, "grad_norm": 0.6103391600302396, "learning_rate": 3.2657135353113845e-05, "loss": 2.7498, "step": 31484 }, { "epoch": 1.9544974858774598, "grad_norm": 0.5568415496661102, "learning_rate": 3.265374808512311e-05, "loss": 2.6933, "step": 31485 }, { "epoch": 1.9545595629772177, "grad_norm": 0.5698968077827741, "learning_rate": 3.265036090763198e-05, "loss": 2.7109, "step": 31486 }, { "epoch": 1.9546216400769756, "grad_norm": 0.5968557494489378, "learning_rate": 3.264697382065813e-05, "loss": 2.834, "step": 31487 }, { "epoch": 1.9546837171767335, "grad_norm": 0.5942404879401304, "learning_rate": 3.264358682421923e-05, "loss": 2.7051, "step": 31488 }, { "epoch": 1.9547457942764914, "grad_norm": 0.5639181803104287, "learning_rate": 3.264019991833296e-05, "loss": 2.6954, "step": 31489 }, { "epoch": 1.9548078713762493, "grad_norm": 0.5112172355505834, "learning_rate": 3.263681310301696e-05, "loss": 2.7576, "step": 31490 }, { "epoch": 1.954869948476007, "grad_norm": 0.5874235509277009, "learning_rate": 3.263342637828894e-05, "loss": 2.6576, "step": 31491 }, { "epoch": 1.954932025575765, "grad_norm": 0.6353639335846853, "learning_rate": 3.263003974416654e-05, "loss": 2.7618, "step": 31492 }, { "epoch": 1.9549941026755229, "grad_norm": 0.49813138955300823, "learning_rate": 3.2626653200667454e-05, "loss": 2.6651, "step": 31493 }, { "epoch": 1.9550561797752808, "grad_norm": 0.7140069451622079, "learning_rate": 3.2623266747809336e-05, "loss": 2.6397, "step": 31494 }, { "epoch": 1.9551182568750387, "grad_norm": 0.55662619588789, "learning_rate": 3.261988038560983e-05, "loss": 2.7777, "step": 31495 }, { "epoch": 1.9551803339747966, "grad_norm": 0.6236962070853125, "learning_rate": 3.261649411408665e-05, "loss": 2.6294, "step": 31496 }, { "epoch": 1.9552424110745545, "grad_norm": 0.6170228183119049, "learning_rate": 3.2613107933257425e-05, "loss": 2.7818, "step": 31497 }, { "epoch": 1.9553044881743125, "grad_norm": 0.6497033872950853, "learning_rate": 3.260972184313985e-05, "loss": 2.7295, "step": 31498 }, { "epoch": 1.9553665652740704, "grad_norm": 0.6003606358163367, "learning_rate": 3.2606335843751564e-05, "loss": 2.6723, "step": 31499 }, { "epoch": 1.9554286423738283, "grad_norm": 0.6015660669790758, "learning_rate": 3.260294993511026e-05, "loss": 2.7384, "step": 31500 }, { "epoch": 1.9554907194735862, "grad_norm": 0.6308433365333183, "learning_rate": 3.2599564117233575e-05, "loss": 2.7823, "step": 31501 }, { "epoch": 1.9555527965733441, "grad_norm": 0.5555288891220419, "learning_rate": 3.25961783901392e-05, "loss": 2.7644, "step": 31502 }, { "epoch": 1.955614873673102, "grad_norm": 0.6225063167389411, "learning_rate": 3.2592792753844783e-05, "loss": 2.707, "step": 31503 }, { "epoch": 1.95567695077286, "grad_norm": 0.5663162949260943, "learning_rate": 3.2589407208367993e-05, "loss": 2.7458, "step": 31504 }, { "epoch": 1.9557390278726179, "grad_norm": 0.5963802304187438, "learning_rate": 3.2586021753726495e-05, "loss": 2.7412, "step": 31505 }, { "epoch": 1.9558011049723758, "grad_norm": 0.557239411182866, "learning_rate": 3.258263638993794e-05, "loss": 2.7343, "step": 31506 }, { "epoch": 1.9558631820721337, "grad_norm": 0.5566909231918167, "learning_rate": 3.257925111702001e-05, "loss": 2.7336, "step": 31507 }, { "epoch": 1.9559252591718916, "grad_norm": 0.5617552966881242, "learning_rate": 3.2575865934990345e-05, "loss": 2.7934, "step": 31508 }, { "epoch": 1.9559873362716493, "grad_norm": 0.5510648984524802, "learning_rate": 3.257248084386663e-05, "loss": 2.7324, "step": 31509 }, { "epoch": 1.9560494133714073, "grad_norm": 0.5262581565478536, "learning_rate": 3.2569095843666485e-05, "loss": 2.6818, "step": 31510 }, { "epoch": 1.9561114904711652, "grad_norm": 0.4922855219038453, "learning_rate": 3.2565710934407614e-05, "loss": 2.7259, "step": 31511 }, { "epoch": 1.956173567570923, "grad_norm": 0.48769894783030887, "learning_rate": 3.2562326116107674e-05, "loss": 2.7061, "step": 31512 }, { "epoch": 1.956235644670681, "grad_norm": 0.5189648746224729, "learning_rate": 3.2558941388784304e-05, "loss": 2.6956, "step": 31513 }, { "epoch": 1.956297721770439, "grad_norm": 0.4814317867708147, "learning_rate": 3.2555556752455176e-05, "loss": 2.8003, "step": 31514 }, { "epoch": 1.9563597988701966, "grad_norm": 0.5119837009491737, "learning_rate": 3.255217220713793e-05, "loss": 2.7536, "step": 31515 }, { "epoch": 1.9564218759699545, "grad_norm": 0.558282325203365, "learning_rate": 3.254878775285025e-05, "loss": 2.6767, "step": 31516 }, { "epoch": 1.9564839530697125, "grad_norm": 0.5386188727203184, "learning_rate": 3.254540338960977e-05, "loss": 2.6967, "step": 31517 }, { "epoch": 1.9565460301694704, "grad_norm": 0.5296913450658544, "learning_rate": 3.254201911743417e-05, "loss": 2.7671, "step": 31518 }, { "epoch": 1.9566081072692283, "grad_norm": 0.5358155359226454, "learning_rate": 3.253863493634109e-05, "loss": 2.6245, "step": 31519 }, { "epoch": 1.9566701843689862, "grad_norm": 0.565236692812688, "learning_rate": 3.253525084634819e-05, "loss": 2.7065, "step": 31520 }, { "epoch": 1.9567322614687441, "grad_norm": 0.47460678476639334, "learning_rate": 3.2531866847473126e-05, "loss": 2.6326, "step": 31521 }, { "epoch": 1.956794338568502, "grad_norm": 0.6300380308540257, "learning_rate": 3.2528482939733544e-05, "loss": 2.7166, "step": 31522 }, { "epoch": 1.95685641566826, "grad_norm": 0.5316032749708777, "learning_rate": 3.252509912314712e-05, "loss": 2.7256, "step": 31523 }, { "epoch": 1.9569184927680179, "grad_norm": 0.5625152585821075, "learning_rate": 3.252171539773149e-05, "loss": 2.7252, "step": 31524 }, { "epoch": 1.9569805698677758, "grad_norm": 0.5671741935913579, "learning_rate": 3.251833176350432e-05, "loss": 2.7076, "step": 31525 }, { "epoch": 1.9570426469675337, "grad_norm": 0.5526665959402268, "learning_rate": 3.2514948220483246e-05, "loss": 2.7211, "step": 31526 }, { "epoch": 1.9571047240672916, "grad_norm": 0.6139947179216083, "learning_rate": 3.251156476868594e-05, "loss": 2.6685, "step": 31527 }, { "epoch": 1.9571668011670496, "grad_norm": 0.5808135742519095, "learning_rate": 3.250818140813004e-05, "loss": 2.7476, "step": 31528 }, { "epoch": 1.9572288782668075, "grad_norm": 0.5917408479160898, "learning_rate": 3.2504798138833215e-05, "loss": 2.7328, "step": 31529 }, { "epoch": 1.9572909553665654, "grad_norm": 0.5850889801350918, "learning_rate": 3.2501414960813106e-05, "loss": 2.809, "step": 31530 }, { "epoch": 1.9573530324663233, "grad_norm": 0.569122562370551, "learning_rate": 3.2498031874087353e-05, "loss": 2.6672, "step": 31531 }, { "epoch": 1.9574151095660812, "grad_norm": 0.5447772801568104, "learning_rate": 3.2494648878673616e-05, "loss": 2.7056, "step": 31532 }, { "epoch": 1.957477186665839, "grad_norm": 0.576656771140374, "learning_rate": 3.2491265974589546e-05, "loss": 2.649, "step": 31533 }, { "epoch": 1.9575392637655968, "grad_norm": 0.6274763401859944, "learning_rate": 3.24878831618528e-05, "loss": 2.7565, "step": 31534 }, { "epoch": 1.9576013408653548, "grad_norm": 0.4920902004645664, "learning_rate": 3.248450044048101e-05, "loss": 2.7341, "step": 31535 }, { "epoch": 1.9576634179651127, "grad_norm": 0.6258124792567022, "learning_rate": 3.248111781049184e-05, "loss": 2.7385, "step": 31536 }, { "epoch": 1.9577254950648706, "grad_norm": 0.5255476139177391, "learning_rate": 3.247773527190292e-05, "loss": 2.7893, "step": 31537 }, { "epoch": 1.9577875721646285, "grad_norm": 0.5445046416996447, "learning_rate": 3.247435282473192e-05, "loss": 2.6986, "step": 31538 }, { "epoch": 1.9578496492643862, "grad_norm": 0.534736747324236, "learning_rate": 3.247097046899647e-05, "loss": 2.7363, "step": 31539 }, { "epoch": 1.9579117263641441, "grad_norm": 0.519972019084634, "learning_rate": 3.246758820471423e-05, "loss": 2.6638, "step": 31540 }, { "epoch": 1.957973803463902, "grad_norm": 0.5385951351970216, "learning_rate": 3.246420603190283e-05, "loss": 2.7526, "step": 31541 }, { "epoch": 1.95803588056366, "grad_norm": 0.5464957966132424, "learning_rate": 3.246082395057991e-05, "loss": 2.795, "step": 31542 }, { "epoch": 1.9580979576634179, "grad_norm": 0.5415114186404758, "learning_rate": 3.2457441960763144e-05, "loss": 2.7762, "step": 31543 }, { "epoch": 1.9581600347631758, "grad_norm": 0.537736422640022, "learning_rate": 3.245406006247018e-05, "loss": 2.7194, "step": 31544 }, { "epoch": 1.9582221118629337, "grad_norm": 0.5339842444017162, "learning_rate": 3.245067825571862e-05, "loss": 2.7361, "step": 31545 }, { "epoch": 1.9582841889626916, "grad_norm": 0.5094021491556171, "learning_rate": 3.244729654052614e-05, "loss": 2.6739, "step": 31546 }, { "epoch": 1.9583462660624495, "grad_norm": 0.5706751467343577, "learning_rate": 3.2443914916910375e-05, "loss": 2.7336, "step": 31547 }, { "epoch": 1.9584083431622075, "grad_norm": 0.5316365462573437, "learning_rate": 3.244053338488897e-05, "loss": 2.6837, "step": 31548 }, { "epoch": 1.9584704202619654, "grad_norm": 0.5376099906428039, "learning_rate": 3.243715194447956e-05, "loss": 2.7527, "step": 31549 }, { "epoch": 1.9585324973617233, "grad_norm": 0.5355849393732779, "learning_rate": 3.24337705956998e-05, "loss": 2.7745, "step": 31550 }, { "epoch": 1.9585945744614812, "grad_norm": 0.540773991908337, "learning_rate": 3.243038933856731e-05, "loss": 2.7139, "step": 31551 }, { "epoch": 1.9586566515612391, "grad_norm": 0.5596959104603592, "learning_rate": 3.242700817309976e-05, "loss": 2.7609, "step": 31552 }, { "epoch": 1.958718728660997, "grad_norm": 0.5655407310644295, "learning_rate": 3.2423627099314755e-05, "loss": 2.7063, "step": 31553 }, { "epoch": 1.958780805760755, "grad_norm": 0.5787579963082242, "learning_rate": 3.2420246117229974e-05, "loss": 2.7017, "step": 31554 }, { "epoch": 1.958842882860513, "grad_norm": 0.5716139692739144, "learning_rate": 3.241686522686304e-05, "loss": 2.6661, "step": 31555 }, { "epoch": 1.9589049599602708, "grad_norm": 0.5514215845887318, "learning_rate": 3.241348442823157e-05, "loss": 2.7407, "step": 31556 }, { "epoch": 1.9589670370600285, "grad_norm": 0.5657737883987322, "learning_rate": 3.2410103721353226e-05, "loss": 2.8178, "step": 31557 }, { "epoch": 1.9590291141597864, "grad_norm": 0.5196641737611218, "learning_rate": 3.240672310624564e-05, "loss": 2.7164, "step": 31558 }, { "epoch": 1.9590911912595443, "grad_norm": 0.5225331194205226, "learning_rate": 3.240334258292645e-05, "loss": 2.6933, "step": 31559 }, { "epoch": 1.9591532683593023, "grad_norm": 0.5411989197623671, "learning_rate": 3.2399962151413296e-05, "loss": 2.7428, "step": 31560 }, { "epoch": 1.9592153454590602, "grad_norm": 0.5781116484053951, "learning_rate": 3.239658181172382e-05, "loss": 2.6903, "step": 31561 }, { "epoch": 1.959277422558818, "grad_norm": 0.5501352674864837, "learning_rate": 3.2393201563875636e-05, "loss": 2.7482, "step": 31562 }, { "epoch": 1.9593394996585758, "grad_norm": 0.5922993456217165, "learning_rate": 3.238982140788641e-05, "loss": 2.6648, "step": 31563 }, { "epoch": 1.9594015767583337, "grad_norm": 0.6130526657287968, "learning_rate": 3.238644134377374e-05, "loss": 2.744, "step": 31564 }, { "epoch": 1.9594636538580916, "grad_norm": 0.5110299519646203, "learning_rate": 3.2383061371555304e-05, "loss": 2.7424, "step": 31565 }, { "epoch": 1.9595257309578495, "grad_norm": 0.6556126587745338, "learning_rate": 3.237968149124869e-05, "loss": 2.6411, "step": 31566 }, { "epoch": 1.9595878080576075, "grad_norm": 0.5715807048052908, "learning_rate": 3.237630170287158e-05, "loss": 2.7152, "step": 31567 }, { "epoch": 1.9596498851573654, "grad_norm": 0.5569484194341263, "learning_rate": 3.237292200644157e-05, "loss": 2.5906, "step": 31568 }, { "epoch": 1.9597119622571233, "grad_norm": 0.5945869118979998, "learning_rate": 3.2369542401976294e-05, "loss": 2.7257, "step": 31569 }, { "epoch": 1.9597740393568812, "grad_norm": 0.5481094596512783, "learning_rate": 3.236616288949341e-05, "loss": 2.7685, "step": 31570 }, { "epoch": 1.9598361164566391, "grad_norm": 0.6529411047902344, "learning_rate": 3.2362783469010527e-05, "loss": 2.7412, "step": 31571 }, { "epoch": 1.959898193556397, "grad_norm": 0.5012026568479362, "learning_rate": 3.2359404140545295e-05, "loss": 2.7049, "step": 31572 }, { "epoch": 1.959960270656155, "grad_norm": 0.5937601891697434, "learning_rate": 3.235602490411532e-05, "loss": 2.7335, "step": 31573 }, { "epoch": 1.9600223477559129, "grad_norm": 0.5984463673624385, "learning_rate": 3.235264575973826e-05, "loss": 2.7198, "step": 31574 }, { "epoch": 1.9600844248556708, "grad_norm": 0.5729563843802594, "learning_rate": 3.2349266707431705e-05, "loss": 2.6945, "step": 31575 }, { "epoch": 1.9601465019554287, "grad_norm": 0.625191754489202, "learning_rate": 3.234588774721333e-05, "loss": 2.7948, "step": 31576 }, { "epoch": 1.9602085790551866, "grad_norm": 0.5618421363342729, "learning_rate": 3.2342508879100747e-05, "loss": 2.8061, "step": 31577 }, { "epoch": 1.9602706561549446, "grad_norm": 0.5910739827008273, "learning_rate": 3.233913010311157e-05, "loss": 2.7432, "step": 31578 }, { "epoch": 1.9603327332547025, "grad_norm": 0.591536234752678, "learning_rate": 3.2335751419263456e-05, "loss": 2.7655, "step": 31579 }, { "epoch": 1.9603948103544604, "grad_norm": 0.5520288392529079, "learning_rate": 3.233237282757401e-05, "loss": 2.727, "step": 31580 }, { "epoch": 1.960456887454218, "grad_norm": 0.5236346030284713, "learning_rate": 3.232899432806086e-05, "loss": 2.5975, "step": 31581 }, { "epoch": 1.960518964553976, "grad_norm": 0.5961736353069758, "learning_rate": 3.2325615920741634e-05, "loss": 2.839, "step": 31582 }, { "epoch": 1.960581041653734, "grad_norm": 0.4708123139806795, "learning_rate": 3.232223760563396e-05, "loss": 2.7059, "step": 31583 }, { "epoch": 1.9606431187534918, "grad_norm": 0.5708920917295338, "learning_rate": 3.2318859382755474e-05, "loss": 2.7275, "step": 31584 }, { "epoch": 1.9607051958532498, "grad_norm": 0.5647960268070669, "learning_rate": 3.231548125212378e-05, "loss": 2.7492, "step": 31585 }, { "epoch": 1.9607672729530077, "grad_norm": 0.4663097920826798, "learning_rate": 3.2312103213756516e-05, "loss": 2.7328, "step": 31586 }, { "epoch": 1.9608293500527654, "grad_norm": 0.5601290431524777, "learning_rate": 3.23087252676713e-05, "loss": 2.7342, "step": 31587 }, { "epoch": 1.9608914271525233, "grad_norm": 0.5643460129214511, "learning_rate": 3.230534741388577e-05, "loss": 2.6704, "step": 31588 }, { "epoch": 1.9609535042522812, "grad_norm": 0.4945743903556984, "learning_rate": 3.230196965241753e-05, "loss": 2.6902, "step": 31589 }, { "epoch": 1.9610155813520391, "grad_norm": 0.5616160706116518, "learning_rate": 3.229859198328421e-05, "loss": 2.7451, "step": 31590 }, { "epoch": 1.961077658451797, "grad_norm": 0.5416831295250409, "learning_rate": 3.2295214406503445e-05, "loss": 2.7664, "step": 31591 }, { "epoch": 1.961139735551555, "grad_norm": 0.5069996710819479, "learning_rate": 3.2291836922092826e-05, "loss": 2.7023, "step": 31592 }, { "epoch": 1.9612018126513129, "grad_norm": 0.5284324992263022, "learning_rate": 3.228845953007e-05, "loss": 2.6934, "step": 31593 }, { "epoch": 1.9612638897510708, "grad_norm": 0.5598717322078524, "learning_rate": 3.228508223045258e-05, "loss": 2.7148, "step": 31594 }, { "epoch": 1.9613259668508287, "grad_norm": 0.5516434812667671, "learning_rate": 3.2281705023258197e-05, "loss": 2.7541, "step": 31595 }, { "epoch": 1.9613880439505866, "grad_norm": 0.48156387076408663, "learning_rate": 3.2278327908504445e-05, "loss": 2.5984, "step": 31596 }, { "epoch": 1.9614501210503446, "grad_norm": 0.5832933145374399, "learning_rate": 3.227495088620897e-05, "loss": 2.6788, "step": 31597 }, { "epoch": 1.9615121981501025, "grad_norm": 0.4827359061611209, "learning_rate": 3.227157395638937e-05, "loss": 2.7523, "step": 31598 }, { "epoch": 1.9615742752498604, "grad_norm": 0.5579431653252729, "learning_rate": 3.226819711906328e-05, "loss": 2.6893, "step": 31599 }, { "epoch": 1.9616363523496183, "grad_norm": 0.49906406125258596, "learning_rate": 3.226482037424829e-05, "loss": 2.7388, "step": 31600 }, { "epoch": 1.9616984294493762, "grad_norm": 0.585872417650174, "learning_rate": 3.226144372196206e-05, "loss": 2.6547, "step": 31601 }, { "epoch": 1.9617605065491341, "grad_norm": 0.5070357911739741, "learning_rate": 3.2258067162222176e-05, "loss": 2.6834, "step": 31602 }, { "epoch": 1.961822583648892, "grad_norm": 0.6155574858383107, "learning_rate": 3.2254690695046275e-05, "loss": 2.7855, "step": 31603 }, { "epoch": 1.96188466074865, "grad_norm": 0.5239400301770605, "learning_rate": 3.225131432045195e-05, "loss": 2.7966, "step": 31604 }, { "epoch": 1.9619467378484077, "grad_norm": 0.5467121699292358, "learning_rate": 3.224793803845682e-05, "loss": 2.7176, "step": 31605 }, { "epoch": 1.9620088149481656, "grad_norm": 0.5566259014123196, "learning_rate": 3.224456184907852e-05, "loss": 2.7147, "step": 31606 }, { "epoch": 1.9620708920479235, "grad_norm": 0.5328320340242236, "learning_rate": 3.2241185752334634e-05, "loss": 2.828, "step": 31607 }, { "epoch": 1.9621329691476814, "grad_norm": 0.5219535859837785, "learning_rate": 3.223780974824279e-05, "loss": 2.8204, "step": 31608 }, { "epoch": 1.9621950462474393, "grad_norm": 0.5764560063541138, "learning_rate": 3.223443383682062e-05, "loss": 2.7072, "step": 31609 }, { "epoch": 1.9622571233471973, "grad_norm": 0.5219330875254522, "learning_rate": 3.223105801808572e-05, "loss": 2.7556, "step": 31610 }, { "epoch": 1.962319200446955, "grad_norm": 0.5729375546433406, "learning_rate": 3.222768229205571e-05, "loss": 2.7446, "step": 31611 }, { "epoch": 1.9623812775467129, "grad_norm": 0.5163894246602313, "learning_rate": 3.222430665874818e-05, "loss": 2.742, "step": 31612 }, { "epoch": 1.9624433546464708, "grad_norm": 0.5674943929134767, "learning_rate": 3.2220931118180774e-05, "loss": 2.7018, "step": 31613 }, { "epoch": 1.9625054317462287, "grad_norm": 0.5680056910196148, "learning_rate": 3.221755567037107e-05, "loss": 2.8317, "step": 31614 }, { "epoch": 1.9625675088459866, "grad_norm": 0.5534683584250965, "learning_rate": 3.221418031533671e-05, "loss": 2.6311, "step": 31615 }, { "epoch": 1.9626295859457445, "grad_norm": 0.6169052685903282, "learning_rate": 3.221080505309529e-05, "loss": 2.7594, "step": 31616 }, { "epoch": 1.9626916630455025, "grad_norm": 0.5538652531311405, "learning_rate": 3.2207429883664404e-05, "loss": 2.6944, "step": 31617 }, { "epoch": 1.9627537401452604, "grad_norm": 0.5874479534551519, "learning_rate": 3.220405480706169e-05, "loss": 2.7435, "step": 31618 }, { "epoch": 1.9628158172450183, "grad_norm": 0.5535608051089332, "learning_rate": 3.220067982330472e-05, "loss": 2.6982, "step": 31619 }, { "epoch": 1.9628778943447762, "grad_norm": 0.5503036376135535, "learning_rate": 3.219730493241115e-05, "loss": 2.7916, "step": 31620 }, { "epoch": 1.9629399714445341, "grad_norm": 0.5654159834128374, "learning_rate": 3.219393013439855e-05, "loss": 2.7839, "step": 31621 }, { "epoch": 1.963002048544292, "grad_norm": 0.5276881470893706, "learning_rate": 3.2190555429284544e-05, "loss": 2.7484, "step": 31622 }, { "epoch": 1.96306412564405, "grad_norm": 0.6011972928885972, "learning_rate": 3.218718081708673e-05, "loss": 2.7838, "step": 31623 }, { "epoch": 1.963126202743808, "grad_norm": 0.5474893389060138, "learning_rate": 3.218380629782272e-05, "loss": 2.7501, "step": 31624 }, { "epoch": 1.9631882798435658, "grad_norm": 0.581043711326214, "learning_rate": 3.218043187151011e-05, "loss": 2.7546, "step": 31625 }, { "epoch": 1.9632503569433237, "grad_norm": 0.5364028167392499, "learning_rate": 3.217705753816652e-05, "loss": 2.6583, "step": 31626 }, { "epoch": 1.9633124340430816, "grad_norm": 0.5763423344247717, "learning_rate": 3.217368329780954e-05, "loss": 2.7169, "step": 31627 }, { "epoch": 1.9633745111428396, "grad_norm": 0.5620297214465, "learning_rate": 3.217030915045679e-05, "loss": 2.7456, "step": 31628 }, { "epoch": 1.9634365882425973, "grad_norm": 0.5123359510660566, "learning_rate": 3.2166935096125874e-05, "loss": 2.6041, "step": 31629 }, { "epoch": 1.9634986653423552, "grad_norm": 0.5804219366940244, "learning_rate": 3.216356113483436e-05, "loss": 2.8125, "step": 31630 }, { "epoch": 1.963560742442113, "grad_norm": 0.5055960481651451, "learning_rate": 3.216018726659991e-05, "loss": 2.6538, "step": 31631 }, { "epoch": 1.963622819541871, "grad_norm": 0.5473069528662563, "learning_rate": 3.2156813491440065e-05, "loss": 2.6104, "step": 31632 }, { "epoch": 1.963684896641629, "grad_norm": 0.5850627473044772, "learning_rate": 3.2153439809372476e-05, "loss": 2.7122, "step": 31633 }, { "epoch": 1.9637469737413868, "grad_norm": 0.5327760711924792, "learning_rate": 3.215006622041471e-05, "loss": 2.7318, "step": 31634 }, { "epoch": 1.9638090508411445, "grad_norm": 0.5039262102299048, "learning_rate": 3.2146692724584394e-05, "loss": 2.8334, "step": 31635 }, { "epoch": 1.9638711279409025, "grad_norm": 0.5343434921255995, "learning_rate": 3.21433193218991e-05, "loss": 2.6361, "step": 31636 }, { "epoch": 1.9639332050406604, "grad_norm": 0.4887960847849995, "learning_rate": 3.213994601237645e-05, "loss": 2.7277, "step": 31637 }, { "epoch": 1.9639952821404183, "grad_norm": 0.5091232110452633, "learning_rate": 3.2136572796034046e-05, "loss": 2.7054, "step": 31638 }, { "epoch": 1.9640573592401762, "grad_norm": 0.5538537381222957, "learning_rate": 3.2133199672889475e-05, "loss": 2.6858, "step": 31639 }, { "epoch": 1.9641194363399341, "grad_norm": 0.5154252098730292, "learning_rate": 3.212982664296034e-05, "loss": 2.7026, "step": 31640 }, { "epoch": 1.964181513439692, "grad_norm": 0.519767994109098, "learning_rate": 3.212645370626421e-05, "loss": 2.7131, "step": 31641 }, { "epoch": 1.96424359053945, "grad_norm": 0.5203616740499519, "learning_rate": 3.212308086281874e-05, "loss": 2.6329, "step": 31642 }, { "epoch": 1.9643056676392079, "grad_norm": 0.515368166450534, "learning_rate": 3.211970811264149e-05, "loss": 2.7713, "step": 31643 }, { "epoch": 1.9643677447389658, "grad_norm": 0.5729979168469574, "learning_rate": 3.211633545575006e-05, "loss": 2.7794, "step": 31644 }, { "epoch": 1.9644298218387237, "grad_norm": 0.5836365343213541, "learning_rate": 3.2112962892162055e-05, "loss": 2.742, "step": 31645 }, { "epoch": 1.9644918989384816, "grad_norm": 0.5513060754467185, "learning_rate": 3.2109590421895064e-05, "loss": 2.707, "step": 31646 }, { "epoch": 1.9645539760382396, "grad_norm": 0.5876232312193482, "learning_rate": 3.2106218044966686e-05, "loss": 2.6551, "step": 31647 }, { "epoch": 1.9646160531379975, "grad_norm": 0.5020180448712253, "learning_rate": 3.2102845761394505e-05, "loss": 2.6878, "step": 31648 }, { "epoch": 1.9646781302377554, "grad_norm": 0.6201046198975633, "learning_rate": 3.2099473571196125e-05, "loss": 2.7738, "step": 31649 }, { "epoch": 1.9647402073375133, "grad_norm": 0.47206220260806714, "learning_rate": 3.209610147438914e-05, "loss": 2.7143, "step": 31650 }, { "epoch": 1.9648022844372712, "grad_norm": 0.6160217575125998, "learning_rate": 3.209272947099114e-05, "loss": 2.7069, "step": 31651 }, { "epoch": 1.9648643615370291, "grad_norm": 0.492540928935955, "learning_rate": 3.20893575610197e-05, "loss": 2.6877, "step": 31652 }, { "epoch": 1.9649264386367868, "grad_norm": 0.5576180159099993, "learning_rate": 3.2085985744492456e-05, "loss": 2.7567, "step": 31653 }, { "epoch": 1.9649885157365448, "grad_norm": 0.5465834249042835, "learning_rate": 3.2082614021426974e-05, "loss": 2.7003, "step": 31654 }, { "epoch": 1.9650505928363027, "grad_norm": 0.5710179916627153, "learning_rate": 3.207924239184083e-05, "loss": 2.7199, "step": 31655 }, { "epoch": 1.9651126699360606, "grad_norm": 0.534274690817046, "learning_rate": 3.207587085575163e-05, "loss": 2.6059, "step": 31656 }, { "epoch": 1.9651747470358185, "grad_norm": 0.6105097451271877, "learning_rate": 3.207249941317696e-05, "loss": 2.7573, "step": 31657 }, { "epoch": 1.9652368241355764, "grad_norm": 0.5062770224058541, "learning_rate": 3.206912806413442e-05, "loss": 2.7947, "step": 31658 }, { "epoch": 1.9652989012353341, "grad_norm": 0.5622685539081334, "learning_rate": 3.206575680864158e-05, "loss": 2.6916, "step": 31659 }, { "epoch": 1.965360978335092, "grad_norm": 0.5358747128556383, "learning_rate": 3.206238564671605e-05, "loss": 2.6921, "step": 31660 }, { "epoch": 1.96542305543485, "grad_norm": 0.5152653393468016, "learning_rate": 3.2059014578375405e-05, "loss": 2.6789, "step": 31661 }, { "epoch": 1.9654851325346079, "grad_norm": 0.5189096274146829, "learning_rate": 3.205564360363724e-05, "loss": 2.7183, "step": 31662 }, { "epoch": 1.9655472096343658, "grad_norm": 0.5467833305912753, "learning_rate": 3.205227272251914e-05, "loss": 2.7078, "step": 31663 }, { "epoch": 1.9656092867341237, "grad_norm": 0.530258656220236, "learning_rate": 3.204890193503869e-05, "loss": 2.8039, "step": 31664 }, { "epoch": 1.9656713638338816, "grad_norm": 0.5417084647215339, "learning_rate": 3.204553124121348e-05, "loss": 2.7772, "step": 31665 }, { "epoch": 1.9657334409336396, "grad_norm": 0.5892957958993915, "learning_rate": 3.2042160641061084e-05, "loss": 2.774, "step": 31666 }, { "epoch": 1.9657955180333975, "grad_norm": 0.49041765897850864, "learning_rate": 3.2038790134599093e-05, "loss": 2.7123, "step": 31667 }, { "epoch": 1.9658575951331554, "grad_norm": 0.5343615903900749, "learning_rate": 3.2035419721845096e-05, "loss": 2.7754, "step": 31668 }, { "epoch": 1.9659196722329133, "grad_norm": 0.5577803872318874, "learning_rate": 3.2032049402816684e-05, "loss": 2.7991, "step": 31669 }, { "epoch": 1.9659817493326712, "grad_norm": 0.5080711879599794, "learning_rate": 3.202867917753142e-05, "loss": 2.6918, "step": 31670 }, { "epoch": 1.9660438264324291, "grad_norm": 0.5914674351838269, "learning_rate": 3.202530904600691e-05, "loss": 2.7686, "step": 31671 }, { "epoch": 1.966105903532187, "grad_norm": 0.5372253570176985, "learning_rate": 3.202193900826072e-05, "loss": 2.7446, "step": 31672 }, { "epoch": 1.966167980631945, "grad_norm": 0.5445218549270914, "learning_rate": 3.201856906431042e-05, "loss": 2.6833, "step": 31673 }, { "epoch": 1.966230057731703, "grad_norm": 0.6106048302652757, "learning_rate": 3.201519921417364e-05, "loss": 2.7172, "step": 31674 }, { "epoch": 1.9662921348314608, "grad_norm": 0.5207361003396053, "learning_rate": 3.201182945786791e-05, "loss": 2.7451, "step": 31675 }, { "epoch": 1.9663542119312187, "grad_norm": 0.5950574784204717, "learning_rate": 3.200845979541085e-05, "loss": 2.7561, "step": 31676 }, { "epoch": 1.9664162890309764, "grad_norm": 0.5735402672590408, "learning_rate": 3.200509022682001e-05, "loss": 2.6512, "step": 31677 }, { "epoch": 1.9664783661307343, "grad_norm": 0.5193179273405527, "learning_rate": 3.2001720752112996e-05, "loss": 2.7063, "step": 31678 }, { "epoch": 1.9665404432304923, "grad_norm": 0.5413756291701352, "learning_rate": 3.1998351371307376e-05, "loss": 2.6626, "step": 31679 }, { "epoch": 1.9666025203302502, "grad_norm": 0.5229926710125837, "learning_rate": 3.199498208442071e-05, "loss": 2.7121, "step": 31680 }, { "epoch": 1.966664597430008, "grad_norm": 0.5371571190161403, "learning_rate": 3.199161289147061e-05, "loss": 2.7253, "step": 31681 }, { "epoch": 1.966726674529766, "grad_norm": 0.5293984403678769, "learning_rate": 3.1988243792474624e-05, "loss": 2.7672, "step": 31682 }, { "epoch": 1.9667887516295237, "grad_norm": 0.5262572221418974, "learning_rate": 3.1984874787450354e-05, "loss": 2.7423, "step": 31683 }, { "epoch": 1.9668508287292816, "grad_norm": 0.5569816749201892, "learning_rate": 3.1981505876415354e-05, "loss": 2.8129, "step": 31684 }, { "epoch": 1.9669129058290395, "grad_norm": 0.5098322682281738, "learning_rate": 3.197813705938722e-05, "loss": 2.7744, "step": 31685 }, { "epoch": 1.9669749829287975, "grad_norm": 0.5517466046030499, "learning_rate": 3.197476833638351e-05, "loss": 2.6653, "step": 31686 }, { "epoch": 1.9670370600285554, "grad_norm": 0.5155129455417224, "learning_rate": 3.197139970742182e-05, "loss": 2.7641, "step": 31687 }, { "epoch": 1.9670991371283133, "grad_norm": 0.5314014443345949, "learning_rate": 3.196803117251971e-05, "loss": 2.7157, "step": 31688 }, { "epoch": 1.9671612142280712, "grad_norm": 0.5731508614415904, "learning_rate": 3.1964662731694766e-05, "loss": 2.7287, "step": 31689 }, { "epoch": 1.9672232913278291, "grad_norm": 0.4871397943489316, "learning_rate": 3.196129438496456e-05, "loss": 2.7539, "step": 31690 }, { "epoch": 1.967285368427587, "grad_norm": 0.5830688444616956, "learning_rate": 3.195792613234664e-05, "loss": 2.6609, "step": 31691 }, { "epoch": 1.967347445527345, "grad_norm": 0.5881910348643515, "learning_rate": 3.195455797385861e-05, "loss": 2.7483, "step": 31692 }, { "epoch": 1.967409522627103, "grad_norm": 0.5200775257420593, "learning_rate": 3.195118990951803e-05, "loss": 2.7563, "step": 31693 }, { "epoch": 1.9674715997268608, "grad_norm": 0.6086615273731163, "learning_rate": 3.194782193934247e-05, "loss": 2.7378, "step": 31694 }, { "epoch": 1.9675336768266187, "grad_norm": 0.48812604344802285, "learning_rate": 3.1944454063349506e-05, "loss": 2.7434, "step": 31695 }, { "epoch": 1.9675957539263766, "grad_norm": 0.5839382906180314, "learning_rate": 3.194108628155672e-05, "loss": 2.8044, "step": 31696 }, { "epoch": 1.9676578310261346, "grad_norm": 0.577575499772316, "learning_rate": 3.193771859398166e-05, "loss": 2.7278, "step": 31697 }, { "epoch": 1.9677199081258925, "grad_norm": 0.5090613341391311, "learning_rate": 3.193435100064191e-05, "loss": 2.6785, "step": 31698 }, { "epoch": 1.9677819852256504, "grad_norm": 0.6584973908476237, "learning_rate": 3.193098350155503e-05, "loss": 2.7787, "step": 31699 }, { "epoch": 1.9678440623254083, "grad_norm": 0.5678231767690901, "learning_rate": 3.19276160967386e-05, "loss": 2.7899, "step": 31700 }, { "epoch": 1.967906139425166, "grad_norm": 0.6226468132194194, "learning_rate": 3.19242487862102e-05, "loss": 2.6218, "step": 31701 }, { "epoch": 1.967968216524924, "grad_norm": 0.5565642722785401, "learning_rate": 3.192088156998736e-05, "loss": 2.7793, "step": 31702 }, { "epoch": 1.9680302936246818, "grad_norm": 0.5838271474810931, "learning_rate": 3.1917514448087685e-05, "loss": 2.7592, "step": 31703 }, { "epoch": 1.9680923707244398, "grad_norm": 0.5249748712157027, "learning_rate": 3.1914147420528706e-05, "loss": 2.7518, "step": 31704 }, { "epoch": 1.9681544478241977, "grad_norm": 0.6064722420676478, "learning_rate": 3.191078048732803e-05, "loss": 2.7683, "step": 31705 }, { "epoch": 1.9682165249239556, "grad_norm": 0.5626819104522187, "learning_rate": 3.1907413648503184e-05, "loss": 2.7749, "step": 31706 }, { "epoch": 1.9682786020237133, "grad_norm": 0.5483729963055142, "learning_rate": 3.190404690407176e-05, "loss": 2.7418, "step": 31707 }, { "epoch": 1.9683406791234712, "grad_norm": 0.5922363965624703, "learning_rate": 3.190068025405132e-05, "loss": 2.7602, "step": 31708 }, { "epoch": 1.9684027562232291, "grad_norm": 0.5015417409623563, "learning_rate": 3.1897313698459424e-05, "loss": 2.7325, "step": 31709 }, { "epoch": 1.968464833322987, "grad_norm": 0.5219591815233388, "learning_rate": 3.1893947237313646e-05, "loss": 2.6442, "step": 31710 }, { "epoch": 1.968526910422745, "grad_norm": 0.5442982321479033, "learning_rate": 3.1890580870631524e-05, "loss": 2.6226, "step": 31711 }, { "epoch": 1.9685889875225029, "grad_norm": 0.5286109956684523, "learning_rate": 3.188721459843066e-05, "loss": 2.717, "step": 31712 }, { "epoch": 1.9686510646222608, "grad_norm": 0.5300009854775389, "learning_rate": 3.1883848420728576e-05, "loss": 2.7675, "step": 31713 }, { "epoch": 1.9687131417220187, "grad_norm": 0.5470293099816534, "learning_rate": 3.188048233754286e-05, "loss": 2.6935, "step": 31714 }, { "epoch": 1.9687752188217766, "grad_norm": 0.4773845929842322, "learning_rate": 3.187711634889107e-05, "loss": 2.6823, "step": 31715 }, { "epoch": 1.9688372959215346, "grad_norm": 0.5728391305879652, "learning_rate": 3.1873750454790756e-05, "loss": 2.7811, "step": 31716 }, { "epoch": 1.9688993730212925, "grad_norm": 0.5005036142373087, "learning_rate": 3.187038465525949e-05, "loss": 2.7388, "step": 31717 }, { "epoch": 1.9689614501210504, "grad_norm": 0.4798008843780308, "learning_rate": 3.186701895031483e-05, "loss": 2.7387, "step": 31718 }, { "epoch": 1.9690235272208083, "grad_norm": 0.5764454596179935, "learning_rate": 3.186365333997433e-05, "loss": 2.6692, "step": 31719 }, { "epoch": 1.9690856043205662, "grad_norm": 0.497880930462761, "learning_rate": 3.1860287824255555e-05, "loss": 2.7254, "step": 31720 }, { "epoch": 1.9691476814203241, "grad_norm": 0.5375017986711081, "learning_rate": 3.1856922403176065e-05, "loss": 2.6621, "step": 31721 }, { "epoch": 1.969209758520082, "grad_norm": 0.6019493444523306, "learning_rate": 3.18535570767534e-05, "loss": 2.7628, "step": 31722 }, { "epoch": 1.96927183561984, "grad_norm": 0.5629903722163442, "learning_rate": 3.185019184500516e-05, "loss": 2.8176, "step": 31723 }, { "epoch": 1.969333912719598, "grad_norm": 0.5370614047736835, "learning_rate": 3.184682670794885e-05, "loss": 2.7017, "step": 31724 }, { "epoch": 1.9693959898193556, "grad_norm": 0.5669817401529256, "learning_rate": 3.184346166560207e-05, "loss": 2.7221, "step": 31725 }, { "epoch": 1.9694580669191135, "grad_norm": 0.602739693302368, "learning_rate": 3.184009671798236e-05, "loss": 2.7135, "step": 31726 }, { "epoch": 1.9695201440188714, "grad_norm": 0.584400347807596, "learning_rate": 3.183673186510725e-05, "loss": 2.7199, "step": 31727 }, { "epoch": 1.9695822211186294, "grad_norm": 0.5838910286593265, "learning_rate": 3.1833367106994345e-05, "loss": 2.7208, "step": 31728 }, { "epoch": 1.9696442982183873, "grad_norm": 0.5885549194959321, "learning_rate": 3.183000244366116e-05, "loss": 2.6651, "step": 31729 }, { "epoch": 1.9697063753181452, "grad_norm": 0.5721870602783647, "learning_rate": 3.182663787512526e-05, "loss": 2.7475, "step": 31730 }, { "epoch": 1.9697684524179029, "grad_norm": 0.5701544294383117, "learning_rate": 3.182327340140421e-05, "loss": 2.6315, "step": 31731 }, { "epoch": 1.9698305295176608, "grad_norm": 0.5784714150648934, "learning_rate": 3.181990902251555e-05, "loss": 2.6865, "step": 31732 }, { "epoch": 1.9698926066174187, "grad_norm": 0.5733882061718389, "learning_rate": 3.181654473847684e-05, "loss": 2.6869, "step": 31733 }, { "epoch": 1.9699546837171766, "grad_norm": 0.5865451205345824, "learning_rate": 3.181318054930563e-05, "loss": 2.7648, "step": 31734 }, { "epoch": 1.9700167608169346, "grad_norm": 0.582332661049507, "learning_rate": 3.180981645501947e-05, "loss": 2.6848, "step": 31735 }, { "epoch": 1.9700788379166925, "grad_norm": 0.5768982269058183, "learning_rate": 3.1806452455635926e-05, "loss": 2.7599, "step": 31736 }, { "epoch": 1.9701409150164504, "grad_norm": 0.6099126840072147, "learning_rate": 3.180308855117252e-05, "loss": 2.6673, "step": 31737 }, { "epoch": 1.9702029921162083, "grad_norm": 0.5509514241398269, "learning_rate": 3.179972474164683e-05, "loss": 2.6731, "step": 31738 }, { "epoch": 1.9702650692159662, "grad_norm": 0.5328585450980541, "learning_rate": 3.179636102707637e-05, "loss": 2.6692, "step": 31739 }, { "epoch": 1.9703271463157241, "grad_norm": 0.5568250384112067, "learning_rate": 3.179299740747874e-05, "loss": 2.7073, "step": 31740 }, { "epoch": 1.970389223415482, "grad_norm": 0.5364240956414527, "learning_rate": 3.178963388287145e-05, "loss": 2.7465, "step": 31741 }, { "epoch": 1.97045130051524, "grad_norm": 0.588658031561575, "learning_rate": 3.178627045327207e-05, "loss": 2.762, "step": 31742 }, { "epoch": 1.970513377614998, "grad_norm": 0.5772998953179708, "learning_rate": 3.178290711869814e-05, "loss": 2.6681, "step": 31743 }, { "epoch": 1.9705754547147558, "grad_norm": 0.5122279480189803, "learning_rate": 3.1779543879167196e-05, "loss": 2.6435, "step": 31744 }, { "epoch": 1.9706375318145137, "grad_norm": 0.5384414799787244, "learning_rate": 3.1776180734696796e-05, "loss": 2.7212, "step": 31745 }, { "epoch": 1.9706996089142716, "grad_norm": 0.5690830251311458, "learning_rate": 3.177281768530449e-05, "loss": 2.7426, "step": 31746 }, { "epoch": 1.9707616860140296, "grad_norm": 0.5234792891944391, "learning_rate": 3.176945473100782e-05, "loss": 2.6897, "step": 31747 }, { "epoch": 1.9708237631137875, "grad_norm": 0.5322391333539075, "learning_rate": 3.176609187182433e-05, "loss": 2.7187, "step": 31748 }, { "epoch": 1.9708858402135452, "grad_norm": 0.5707282549725546, "learning_rate": 3.176272910777155e-05, "loss": 2.8272, "step": 31749 }, { "epoch": 1.970947917313303, "grad_norm": 0.5253987932102036, "learning_rate": 3.175936643886707e-05, "loss": 2.7237, "step": 31750 }, { "epoch": 1.971009994413061, "grad_norm": 0.54533023641429, "learning_rate": 3.1756003865128384e-05, "loss": 2.693, "step": 31751 }, { "epoch": 1.971072071512819, "grad_norm": 0.5358573330902736, "learning_rate": 3.1752641386573054e-05, "loss": 2.7264, "step": 31752 }, { "epoch": 1.9711341486125769, "grad_norm": 0.5255285032510982, "learning_rate": 3.174927900321863e-05, "loss": 2.7275, "step": 31753 }, { "epoch": 1.9711962257123348, "grad_norm": 0.5239136082069648, "learning_rate": 3.174591671508264e-05, "loss": 2.7635, "step": 31754 }, { "epoch": 1.9712583028120925, "grad_norm": 0.49483924603497825, "learning_rate": 3.174255452218264e-05, "loss": 2.8091, "step": 31755 }, { "epoch": 1.9713203799118504, "grad_norm": 0.5589330749643677, "learning_rate": 3.173919242453616e-05, "loss": 2.697, "step": 31756 }, { "epoch": 1.9713824570116083, "grad_norm": 0.5231288170532832, "learning_rate": 3.173583042216075e-05, "loss": 2.8048, "step": 31757 }, { "epoch": 1.9714445341113662, "grad_norm": 0.5557535967571028, "learning_rate": 3.1732468515073946e-05, "loss": 2.7298, "step": 31758 }, { "epoch": 1.9715066112111241, "grad_norm": 0.5146755667381818, "learning_rate": 3.172910670329329e-05, "loss": 2.7873, "step": 31759 }, { "epoch": 1.971568688310882, "grad_norm": 0.5115355826294651, "learning_rate": 3.172574498683631e-05, "loss": 2.7546, "step": 31760 }, { "epoch": 1.97163076541064, "grad_norm": 0.5221735939534764, "learning_rate": 3.172238336572056e-05, "loss": 2.7676, "step": 31761 }, { "epoch": 1.971692842510398, "grad_norm": 0.5119723208627194, "learning_rate": 3.171902183996358e-05, "loss": 2.7236, "step": 31762 }, { "epoch": 1.9717549196101558, "grad_norm": 0.5285609291434923, "learning_rate": 3.17156604095829e-05, "loss": 2.7243, "step": 31763 }, { "epoch": 1.9718169967099137, "grad_norm": 0.49140227214616305, "learning_rate": 3.171229907459606e-05, "loss": 2.7542, "step": 31764 }, { "epoch": 1.9718790738096716, "grad_norm": 0.4907023866936678, "learning_rate": 3.170893783502058e-05, "loss": 2.7619, "step": 31765 }, { "epoch": 1.9719411509094296, "grad_norm": 0.48159373009456224, "learning_rate": 3.1705576690874025e-05, "loss": 2.751, "step": 31766 }, { "epoch": 1.9720032280091875, "grad_norm": 0.5081670843622916, "learning_rate": 3.170221564217391e-05, "loss": 2.7539, "step": 31767 }, { "epoch": 1.9720653051089454, "grad_norm": 0.5510636004990419, "learning_rate": 3.169885468893779e-05, "loss": 2.7771, "step": 31768 }, { "epoch": 1.9721273822087033, "grad_norm": 0.508256072937372, "learning_rate": 3.1695493831183174e-05, "loss": 2.7265, "step": 31769 }, { "epoch": 1.9721894593084612, "grad_norm": 0.5465423389640167, "learning_rate": 3.169213306892762e-05, "loss": 2.6814, "step": 31770 }, { "epoch": 1.9722515364082192, "grad_norm": 0.5257074754942839, "learning_rate": 3.168877240218864e-05, "loss": 2.7552, "step": 31771 }, { "epoch": 1.972313613507977, "grad_norm": 0.5990184643181082, "learning_rate": 3.168541183098378e-05, "loss": 2.7775, "step": 31772 }, { "epoch": 1.9723756906077348, "grad_norm": 0.5808705783525482, "learning_rate": 3.168205135533058e-05, "loss": 2.7855, "step": 31773 }, { "epoch": 1.9724377677074927, "grad_norm": 0.5390003831982356, "learning_rate": 3.167869097524656e-05, "loss": 2.6982, "step": 31774 }, { "epoch": 1.9724998448072506, "grad_norm": 0.5340105308483729, "learning_rate": 3.167533069074927e-05, "loss": 2.6738, "step": 31775 }, { "epoch": 1.9725619219070085, "grad_norm": 0.5502807561675146, "learning_rate": 3.167197050185622e-05, "loss": 2.6984, "step": 31776 }, { "epoch": 1.9726239990067664, "grad_norm": 0.5532914444663715, "learning_rate": 3.166861040858495e-05, "loss": 2.6977, "step": 31777 }, { "epoch": 1.9726860761065244, "grad_norm": 0.5565013778254752, "learning_rate": 3.166525041095299e-05, "loss": 2.7424, "step": 31778 }, { "epoch": 1.972748153206282, "grad_norm": 0.5563137500107327, "learning_rate": 3.1661890508977865e-05, "loss": 2.7237, "step": 31779 }, { "epoch": 1.97281023030604, "grad_norm": 0.5228187547892142, "learning_rate": 3.165853070267711e-05, "loss": 2.727, "step": 31780 }, { "epoch": 1.9728723074057979, "grad_norm": 0.5385875821293006, "learning_rate": 3.165517099206826e-05, "loss": 2.7815, "step": 31781 }, { "epoch": 1.9729343845055558, "grad_norm": 0.5769572191344957, "learning_rate": 3.1651811377168834e-05, "loss": 2.7535, "step": 31782 }, { "epoch": 1.9729964616053137, "grad_norm": 0.5215473779916888, "learning_rate": 3.164845185799635e-05, "loss": 2.7426, "step": 31783 }, { "epoch": 1.9730585387050716, "grad_norm": 0.5297982631767205, "learning_rate": 3.164509243456837e-05, "loss": 2.6898, "step": 31784 }, { "epoch": 1.9731206158048296, "grad_norm": 0.523781478363783, "learning_rate": 3.164173310690238e-05, "loss": 2.7381, "step": 31785 }, { "epoch": 1.9731826929045875, "grad_norm": 0.5576840175257866, "learning_rate": 3.163837387501594e-05, "loss": 2.7318, "step": 31786 }, { "epoch": 1.9732447700043454, "grad_norm": 0.48708622991398404, "learning_rate": 3.163501473892654e-05, "loss": 2.7603, "step": 31787 }, { "epoch": 1.9733068471041033, "grad_norm": 0.5156106946868132, "learning_rate": 3.163165569865174e-05, "loss": 2.73, "step": 31788 }, { "epoch": 1.9733689242038612, "grad_norm": 0.47880310453800295, "learning_rate": 3.162829675420906e-05, "loss": 2.692, "step": 31789 }, { "epoch": 1.9734310013036191, "grad_norm": 0.514337000271764, "learning_rate": 3.1624937905616e-05, "loss": 2.7205, "step": 31790 }, { "epoch": 1.973493078403377, "grad_norm": 0.5090872683274917, "learning_rate": 3.162157915289011e-05, "loss": 2.6981, "step": 31791 }, { "epoch": 1.973555155503135, "grad_norm": 0.4572834395042927, "learning_rate": 3.161822049604889e-05, "loss": 2.7046, "step": 31792 }, { "epoch": 1.973617232602893, "grad_norm": 0.5662862190132565, "learning_rate": 3.1614861935109894e-05, "loss": 2.6789, "step": 31793 }, { "epoch": 1.9736793097026508, "grad_norm": 0.5280999589660346, "learning_rate": 3.161150347009061e-05, "loss": 2.7152, "step": 31794 }, { "epoch": 1.9737413868024087, "grad_norm": 0.5308783016241388, "learning_rate": 3.160814510100859e-05, "loss": 2.6412, "step": 31795 }, { "epoch": 1.9738034639021667, "grad_norm": 0.5330761768842055, "learning_rate": 3.160478682788133e-05, "loss": 2.625, "step": 31796 }, { "epoch": 1.9738655410019244, "grad_norm": 0.4680292310144574, "learning_rate": 3.160142865072637e-05, "loss": 2.6645, "step": 31797 }, { "epoch": 1.9739276181016823, "grad_norm": 0.5154832009206177, "learning_rate": 3.159807056956121e-05, "loss": 2.7033, "step": 31798 }, { "epoch": 1.9739896952014402, "grad_norm": 0.5735999425576127, "learning_rate": 3.15947125844034e-05, "loss": 2.8035, "step": 31799 }, { "epoch": 1.974051772301198, "grad_norm": 0.5232863421461463, "learning_rate": 3.1591354695270445e-05, "loss": 2.7423, "step": 31800 }, { "epoch": 1.974113849400956, "grad_norm": 0.5542531529200534, "learning_rate": 3.1587996902179846e-05, "loss": 2.6774, "step": 31801 }, { "epoch": 1.974175926500714, "grad_norm": 0.5454594411413527, "learning_rate": 3.1584639205149146e-05, "loss": 2.7172, "step": 31802 }, { "epoch": 1.9742380036004716, "grad_norm": 0.5241801274678207, "learning_rate": 3.158128160419584e-05, "loss": 2.6991, "step": 31803 }, { "epoch": 1.9743000807002296, "grad_norm": 0.5598071331172784, "learning_rate": 3.157792409933746e-05, "loss": 2.8044, "step": 31804 }, { "epoch": 1.9743621577999875, "grad_norm": 0.5446720165737395, "learning_rate": 3.157456669059154e-05, "loss": 2.8234, "step": 31805 }, { "epoch": 1.9744242348997454, "grad_norm": 0.5222663303048073, "learning_rate": 3.157120937797556e-05, "loss": 2.8153, "step": 31806 }, { "epoch": 1.9744863119995033, "grad_norm": 0.5816778607871669, "learning_rate": 3.1567852161507074e-05, "loss": 2.6957, "step": 31807 }, { "epoch": 1.9745483890992612, "grad_norm": 0.5899707076511368, "learning_rate": 3.156449504120357e-05, "loss": 2.7848, "step": 31808 }, { "epoch": 1.9746104661990191, "grad_norm": 0.514756801134175, "learning_rate": 3.156113801708257e-05, "loss": 2.7083, "step": 31809 }, { "epoch": 1.974672543298777, "grad_norm": 0.5593412506322177, "learning_rate": 3.155778108916159e-05, "loss": 2.8099, "step": 31810 }, { "epoch": 1.974734620398535, "grad_norm": 0.5275030672517523, "learning_rate": 3.1554424257458134e-05, "loss": 2.6888, "step": 31811 }, { "epoch": 1.974796697498293, "grad_norm": 0.573212682949248, "learning_rate": 3.155106752198974e-05, "loss": 2.7582, "step": 31812 }, { "epoch": 1.9748587745980508, "grad_norm": 0.525439036808464, "learning_rate": 3.154771088277389e-05, "loss": 2.7096, "step": 31813 }, { "epoch": 1.9749208516978087, "grad_norm": 0.48452534957092913, "learning_rate": 3.154435433982813e-05, "loss": 2.6682, "step": 31814 }, { "epoch": 1.9749829287975667, "grad_norm": 0.5578306360795949, "learning_rate": 3.154099789316994e-05, "loss": 2.7759, "step": 31815 }, { "epoch": 1.9750450058973246, "grad_norm": 0.5276179413222768, "learning_rate": 3.1537641542816856e-05, "loss": 2.7268, "step": 31816 }, { "epoch": 1.9751070829970825, "grad_norm": 0.5757800822977263, "learning_rate": 3.1534285288786356e-05, "loss": 2.8257, "step": 31817 }, { "epoch": 1.9751691600968404, "grad_norm": 0.4695207140814729, "learning_rate": 3.1530929131096e-05, "loss": 2.6586, "step": 31818 }, { "epoch": 1.9752312371965983, "grad_norm": 0.5146982071208096, "learning_rate": 3.152757306976325e-05, "loss": 2.7656, "step": 31819 }, { "epoch": 1.9752933142963562, "grad_norm": 0.5803202371076374, "learning_rate": 3.152421710480564e-05, "loss": 2.7392, "step": 31820 }, { "epoch": 1.975355391396114, "grad_norm": 0.5454573102902651, "learning_rate": 3.152086123624068e-05, "loss": 2.8338, "step": 31821 }, { "epoch": 1.9754174684958719, "grad_norm": 0.5139921806961539, "learning_rate": 3.1517505464085874e-05, "loss": 2.7061, "step": 31822 }, { "epoch": 1.9754795455956298, "grad_norm": 0.508658482054929, "learning_rate": 3.1514149788358715e-05, "loss": 2.6978, "step": 31823 }, { "epoch": 1.9755416226953877, "grad_norm": 0.4857316246464097, "learning_rate": 3.151079420907674e-05, "loss": 2.6677, "step": 31824 }, { "epoch": 1.9756036997951456, "grad_norm": 0.48327386965760305, "learning_rate": 3.1507438726257447e-05, "loss": 2.7704, "step": 31825 }, { "epoch": 1.9756657768949035, "grad_norm": 0.5490413808044831, "learning_rate": 3.150408333991831e-05, "loss": 2.7039, "step": 31826 }, { "epoch": 1.9757278539946612, "grad_norm": 0.4745098684729792, "learning_rate": 3.150072805007688e-05, "loss": 2.7, "step": 31827 }, { "epoch": 1.9757899310944191, "grad_norm": 0.5288665824225676, "learning_rate": 3.149737285675063e-05, "loss": 2.7012, "step": 31828 }, { "epoch": 1.975852008194177, "grad_norm": 0.5367996710086957, "learning_rate": 3.1494017759957086e-05, "loss": 2.6424, "step": 31829 }, { "epoch": 1.975914085293935, "grad_norm": 0.5243520007358217, "learning_rate": 3.149066275971373e-05, "loss": 2.6783, "step": 31830 }, { "epoch": 1.975976162393693, "grad_norm": 0.5468102008031325, "learning_rate": 3.14873078560381e-05, "loss": 2.7084, "step": 31831 }, { "epoch": 1.9760382394934508, "grad_norm": 0.5679221814989358, "learning_rate": 3.1483953048947666e-05, "loss": 2.6817, "step": 31832 }, { "epoch": 1.9761003165932087, "grad_norm": 0.5498094575448051, "learning_rate": 3.148059833845995e-05, "loss": 2.812, "step": 31833 }, { "epoch": 1.9761623936929666, "grad_norm": 0.5123480501776835, "learning_rate": 3.147724372459244e-05, "loss": 2.7267, "step": 31834 }, { "epoch": 1.9762244707927246, "grad_norm": 0.5577540455170898, "learning_rate": 3.1473889207362665e-05, "loss": 2.6112, "step": 31835 }, { "epoch": 1.9762865478924825, "grad_norm": 0.50062614216954, "learning_rate": 3.147053478678809e-05, "loss": 2.7772, "step": 31836 }, { "epoch": 1.9763486249922404, "grad_norm": 0.5613726869653121, "learning_rate": 3.146718046288623e-05, "loss": 2.7536, "step": 31837 }, { "epoch": 1.9764107020919983, "grad_norm": 0.49636620076319266, "learning_rate": 3.1463826235674584e-05, "loss": 2.7605, "step": 31838 }, { "epoch": 1.9764727791917562, "grad_norm": 0.5234352225638421, "learning_rate": 3.146047210517067e-05, "loss": 2.718, "step": 31839 }, { "epoch": 1.9765348562915142, "grad_norm": 0.5788095648332472, "learning_rate": 3.1457118071391965e-05, "loss": 2.7557, "step": 31840 }, { "epoch": 1.976596933391272, "grad_norm": 0.5110968385076545, "learning_rate": 3.145376413435599e-05, "loss": 2.7366, "step": 31841 }, { "epoch": 1.97665901049103, "grad_norm": 0.6113948115334872, "learning_rate": 3.145041029408022e-05, "loss": 2.6744, "step": 31842 }, { "epoch": 1.976721087590788, "grad_norm": 0.5309526228252506, "learning_rate": 3.144705655058217e-05, "loss": 2.8209, "step": 31843 }, { "epoch": 1.9767831646905456, "grad_norm": 0.5856643294295193, "learning_rate": 3.144370290387932e-05, "loss": 2.7455, "step": 31844 }, { "epoch": 1.9768452417903035, "grad_norm": 0.5178918139826301, "learning_rate": 3.144034935398918e-05, "loss": 2.758, "step": 31845 }, { "epoch": 1.9769073188900614, "grad_norm": 0.5625152585821075, "learning_rate": 3.1436995900929244e-05, "loss": 2.6911, "step": 31846 }, { "epoch": 1.9769693959898194, "grad_norm": 0.5542864894633287, "learning_rate": 3.1433642544717014e-05, "loss": 2.768, "step": 31847 }, { "epoch": 1.9770314730895773, "grad_norm": 0.5346118078827289, "learning_rate": 3.143028928536996e-05, "loss": 2.679, "step": 31848 }, { "epoch": 1.9770935501893352, "grad_norm": 0.5310505043642282, "learning_rate": 3.142693612290561e-05, "loss": 2.7354, "step": 31849 }, { "epoch": 1.9771556272890929, "grad_norm": 0.4922589901063957, "learning_rate": 3.1423583057341456e-05, "loss": 2.6641, "step": 31850 }, { "epoch": 1.9772177043888508, "grad_norm": 0.5615088950222854, "learning_rate": 3.142023008869495e-05, "loss": 2.6762, "step": 31851 }, { "epoch": 1.9772797814886087, "grad_norm": 0.5208689454937254, "learning_rate": 3.141687721698363e-05, "loss": 2.7182, "step": 31852 }, { "epoch": 1.9773418585883666, "grad_norm": 0.5330951846909746, "learning_rate": 3.1413524442224964e-05, "loss": 2.6635, "step": 31853 }, { "epoch": 1.9774039356881246, "grad_norm": 0.5051919075828943, "learning_rate": 3.141017176443646e-05, "loss": 2.8232, "step": 31854 }, { "epoch": 1.9774660127878825, "grad_norm": 0.5366824863848694, "learning_rate": 3.14068191836356e-05, "loss": 2.7078, "step": 31855 }, { "epoch": 1.9775280898876404, "grad_norm": 0.5302344601182714, "learning_rate": 3.1403466699839884e-05, "loss": 2.6902, "step": 31856 }, { "epoch": 1.9775901669873983, "grad_norm": 0.5216358264921207, "learning_rate": 3.140011431306679e-05, "loss": 2.7728, "step": 31857 }, { "epoch": 1.9776522440871562, "grad_norm": 0.5169823150210953, "learning_rate": 3.139676202333381e-05, "loss": 2.7331, "step": 31858 }, { "epoch": 1.9777143211869141, "grad_norm": 0.540418136578014, "learning_rate": 3.1393409830658446e-05, "loss": 2.7581, "step": 31859 }, { "epoch": 1.977776398286672, "grad_norm": 0.5277540800423892, "learning_rate": 3.1390057735058185e-05, "loss": 2.7317, "step": 31860 }, { "epoch": 1.97783847538643, "grad_norm": 0.5530824138389824, "learning_rate": 3.1386705736550506e-05, "loss": 2.7597, "step": 31861 }, { "epoch": 1.977900552486188, "grad_norm": 0.4731506410249546, "learning_rate": 3.13833538351529e-05, "loss": 2.7419, "step": 31862 }, { "epoch": 1.9779626295859458, "grad_norm": 0.5412547550477089, "learning_rate": 3.138000203088286e-05, "loss": 2.748, "step": 31863 }, { "epoch": 1.9780247066857037, "grad_norm": 0.5249102640663348, "learning_rate": 3.1376650323757857e-05, "loss": 2.7958, "step": 31864 }, { "epoch": 1.9780867837854617, "grad_norm": 0.48457401555594126, "learning_rate": 3.1373298713795404e-05, "loss": 2.7407, "step": 31865 }, { "epoch": 1.9781488608852196, "grad_norm": 0.4884220683173332, "learning_rate": 3.136994720101296e-05, "loss": 2.6373, "step": 31866 }, { "epoch": 1.9782109379849775, "grad_norm": 0.4953471746251113, "learning_rate": 3.136659578542804e-05, "loss": 2.6962, "step": 31867 }, { "epoch": 1.9782730150847352, "grad_norm": 0.538151621054845, "learning_rate": 3.13632444670581e-05, "loss": 2.7876, "step": 31868 }, { "epoch": 1.978335092184493, "grad_norm": 0.48470550459397566, "learning_rate": 3.1359893245920646e-05, "loss": 2.7674, "step": 31869 }, { "epoch": 1.978397169284251, "grad_norm": 0.4896979874082482, "learning_rate": 3.135654212203314e-05, "loss": 2.784, "step": 31870 }, { "epoch": 1.978459246384009, "grad_norm": 0.5572175900672545, "learning_rate": 3.135319109541308e-05, "loss": 2.84, "step": 31871 }, { "epoch": 1.9785213234837669, "grad_norm": 0.5136725422089744, "learning_rate": 3.134984016607797e-05, "loss": 2.7698, "step": 31872 }, { "epoch": 1.9785834005835248, "grad_norm": 0.5321553313858766, "learning_rate": 3.1346489334045256e-05, "loss": 2.7517, "step": 31873 }, { "epoch": 1.9786454776832825, "grad_norm": 0.5326078396610906, "learning_rate": 3.1343138599332445e-05, "loss": 2.7698, "step": 31874 }, { "epoch": 1.9787075547830404, "grad_norm": 0.5012725491101259, "learning_rate": 3.1339787961957014e-05, "loss": 2.6867, "step": 31875 }, { "epoch": 1.9787696318827983, "grad_norm": 0.5523238258059603, "learning_rate": 3.133643742193641e-05, "loss": 2.6702, "step": 31876 }, { "epoch": 1.9788317089825562, "grad_norm": 0.544820906059285, "learning_rate": 3.133308697928817e-05, "loss": 2.7016, "step": 31877 }, { "epoch": 1.9788937860823141, "grad_norm": 0.5282683849376015, "learning_rate": 3.132973663402974e-05, "loss": 2.7417, "step": 31878 }, { "epoch": 1.978955863182072, "grad_norm": 0.5543829121933553, "learning_rate": 3.1326386386178605e-05, "loss": 2.5718, "step": 31879 }, { "epoch": 1.97901794028183, "grad_norm": 0.5274844794227368, "learning_rate": 3.1323036235752244e-05, "loss": 2.7833, "step": 31880 }, { "epoch": 1.979080017381588, "grad_norm": 0.5129514218022968, "learning_rate": 3.131968618276815e-05, "loss": 2.6787, "step": 31881 }, { "epoch": 1.9791420944813458, "grad_norm": 0.4621627976662217, "learning_rate": 3.131633622724377e-05, "loss": 2.7517, "step": 31882 }, { "epoch": 1.9792041715811037, "grad_norm": 0.5331311858557326, "learning_rate": 3.131298636919661e-05, "loss": 2.6496, "step": 31883 }, { "epoch": 1.9792662486808617, "grad_norm": 0.49030380973141063, "learning_rate": 3.1309636608644124e-05, "loss": 2.6534, "step": 31884 }, { "epoch": 1.9793283257806196, "grad_norm": 0.5133292792622554, "learning_rate": 3.130628694560382e-05, "loss": 2.7179, "step": 31885 }, { "epoch": 1.9793904028803775, "grad_norm": 0.5103481369983914, "learning_rate": 3.130293738009314e-05, "loss": 2.713, "step": 31886 }, { "epoch": 1.9794524799801354, "grad_norm": 0.5658572199737117, "learning_rate": 3.129958791212958e-05, "loss": 2.749, "step": 31887 }, { "epoch": 1.9795145570798933, "grad_norm": 0.4936617917327363, "learning_rate": 3.129623854173062e-05, "loss": 2.7208, "step": 31888 }, { "epoch": 1.9795766341796512, "grad_norm": 0.5211374221642491, "learning_rate": 3.1292889268913706e-05, "loss": 2.6709, "step": 31889 }, { "epoch": 1.9796387112794092, "grad_norm": 0.4964329918804739, "learning_rate": 3.128954009369634e-05, "loss": 2.562, "step": 31890 }, { "epoch": 1.979700788379167, "grad_norm": 0.5247460136934596, "learning_rate": 3.128619101609598e-05, "loss": 2.6845, "step": 31891 }, { "epoch": 1.9797628654789248, "grad_norm": 0.5123926922641844, "learning_rate": 3.128284203613011e-05, "loss": 2.6996, "step": 31892 }, { "epoch": 1.9798249425786827, "grad_norm": 0.5020923644027923, "learning_rate": 3.127949315381619e-05, "loss": 2.78, "step": 31893 }, { "epoch": 1.9798870196784406, "grad_norm": 0.5534514772964599, "learning_rate": 3.1276144369171704e-05, "loss": 2.7301, "step": 31894 }, { "epoch": 1.9799490967781985, "grad_norm": 0.5184766060472693, "learning_rate": 3.1272795682214115e-05, "loss": 2.6195, "step": 31895 }, { "epoch": 1.9800111738779564, "grad_norm": 0.5596256728024472, "learning_rate": 3.12694470929609e-05, "loss": 2.7246, "step": 31896 }, { "epoch": 1.9800732509777144, "grad_norm": 0.5961846078645968, "learning_rate": 3.1266098601429534e-05, "loss": 2.7519, "step": 31897 }, { "epoch": 1.980135328077472, "grad_norm": 0.5777464993683203, "learning_rate": 3.126275020763747e-05, "loss": 2.7323, "step": 31898 }, { "epoch": 1.98019740517723, "grad_norm": 0.5317246896847523, "learning_rate": 3.1259401911602184e-05, "loss": 2.721, "step": 31899 }, { "epoch": 1.980259482276988, "grad_norm": 0.5351976253358479, "learning_rate": 3.125605371334115e-05, "loss": 2.7017, "step": 31900 }, { "epoch": 1.9803215593767458, "grad_norm": 0.5469950407978488, "learning_rate": 3.125270561287184e-05, "loss": 2.6773, "step": 31901 }, { "epoch": 1.9803836364765037, "grad_norm": 0.5240172122410668, "learning_rate": 3.1249357610211705e-05, "loss": 2.771, "step": 31902 }, { "epoch": 1.9804457135762616, "grad_norm": 0.5584624342856647, "learning_rate": 3.1246009705378217e-05, "loss": 2.6647, "step": 31903 }, { "epoch": 1.9805077906760196, "grad_norm": 0.5382362612677603, "learning_rate": 3.1242661898388865e-05, "loss": 2.8125, "step": 31904 }, { "epoch": 1.9805698677757775, "grad_norm": 0.5409312819416062, "learning_rate": 3.123931418926109e-05, "loss": 2.6277, "step": 31905 }, { "epoch": 1.9806319448755354, "grad_norm": 0.49303037134365335, "learning_rate": 3.1235966578012374e-05, "loss": 2.7137, "step": 31906 }, { "epoch": 1.9806940219752933, "grad_norm": 0.5277291479070682, "learning_rate": 3.123261906466017e-05, "loss": 2.7954, "step": 31907 }, { "epoch": 1.9807560990750512, "grad_norm": 0.5198545382234185, "learning_rate": 3.122927164922196e-05, "loss": 2.7805, "step": 31908 }, { "epoch": 1.9808181761748092, "grad_norm": 0.5138411442454469, "learning_rate": 3.1225924331715186e-05, "loss": 2.7718, "step": 31909 }, { "epoch": 1.980880253274567, "grad_norm": 0.5110738928029883, "learning_rate": 3.122257711215733e-05, "loss": 2.7798, "step": 31910 }, { "epoch": 1.980942330374325, "grad_norm": 0.5175442666549285, "learning_rate": 3.1219229990565845e-05, "loss": 2.6934, "step": 31911 }, { "epoch": 1.981004407474083, "grad_norm": 0.5435673001162844, "learning_rate": 3.1215882966958194e-05, "loss": 2.8433, "step": 31912 }, { "epoch": 1.9810664845738408, "grad_norm": 0.5053040390863324, "learning_rate": 3.121253604135185e-05, "loss": 2.7389, "step": 31913 }, { "epoch": 1.9811285616735987, "grad_norm": 0.5701792835737874, "learning_rate": 3.1209189213764244e-05, "loss": 2.6521, "step": 31914 }, { "epoch": 1.9811906387733567, "grad_norm": 0.5107802017681934, "learning_rate": 3.120584248421289e-05, "loss": 2.6883, "step": 31915 }, { "epoch": 1.9812527158731144, "grad_norm": 0.5514351770775138, "learning_rate": 3.1202495852715194e-05, "loss": 2.7386, "step": 31916 }, { "epoch": 1.9813147929728723, "grad_norm": 0.5774892455912045, "learning_rate": 3.119914931928865e-05, "loss": 2.7769, "step": 31917 }, { "epoch": 1.9813768700726302, "grad_norm": 0.549174840699386, "learning_rate": 3.1195802883950705e-05, "loss": 2.7303, "step": 31918 }, { "epoch": 1.981438947172388, "grad_norm": 0.5734600322568599, "learning_rate": 3.119245654671883e-05, "loss": 2.737, "step": 31919 }, { "epoch": 1.981501024272146, "grad_norm": 0.5170224355647776, "learning_rate": 3.118911030761046e-05, "loss": 2.6311, "step": 31920 }, { "epoch": 1.981563101371904, "grad_norm": 0.5313272420042621, "learning_rate": 3.118576416664308e-05, "loss": 2.6214, "step": 31921 }, { "epoch": 1.9816251784716616, "grad_norm": 0.5215541778595334, "learning_rate": 3.1182418123834134e-05, "loss": 2.7095, "step": 31922 }, { "epoch": 1.9816872555714196, "grad_norm": 0.5403649724629545, "learning_rate": 3.117907217920107e-05, "loss": 2.8022, "step": 31923 }, { "epoch": 1.9817493326711775, "grad_norm": 0.564091391629228, "learning_rate": 3.117572633276137e-05, "loss": 2.7215, "step": 31924 }, { "epoch": 1.9818114097709354, "grad_norm": 0.5677886666905227, "learning_rate": 3.117238058453247e-05, "loss": 2.7225, "step": 31925 }, { "epoch": 1.9818734868706933, "grad_norm": 0.5389310704214584, "learning_rate": 3.116903493453183e-05, "loss": 2.6523, "step": 31926 }, { "epoch": 1.9819355639704512, "grad_norm": 0.6028969812991152, "learning_rate": 3.11656893827769e-05, "loss": 2.7152, "step": 31927 }, { "epoch": 1.9819976410702091, "grad_norm": 0.4963616076055447, "learning_rate": 3.1162343929285155e-05, "loss": 2.699, "step": 31928 }, { "epoch": 1.982059718169967, "grad_norm": 0.5591334156927034, "learning_rate": 3.1158998574074024e-05, "loss": 2.6365, "step": 31929 }, { "epoch": 1.982121795269725, "grad_norm": 0.544261628092869, "learning_rate": 3.115565331716098e-05, "loss": 2.7167, "step": 31930 }, { "epoch": 1.982183872369483, "grad_norm": 0.5260241634616969, "learning_rate": 3.115230815856345e-05, "loss": 2.8111, "step": 31931 }, { "epoch": 1.9822459494692408, "grad_norm": 0.5062009034073756, "learning_rate": 3.114896309829892e-05, "loss": 2.7744, "step": 31932 }, { "epoch": 1.9823080265689987, "grad_norm": 0.543392482601814, "learning_rate": 3.114561813638481e-05, "loss": 2.648, "step": 31933 }, { "epoch": 1.9823701036687567, "grad_norm": 0.4888255328279926, "learning_rate": 3.1142273272838604e-05, "loss": 2.6841, "step": 31934 }, { "epoch": 1.9824321807685146, "grad_norm": 0.5577163207325643, "learning_rate": 3.1138928507677726e-05, "loss": 2.7347, "step": 31935 }, { "epoch": 1.9824942578682725, "grad_norm": 0.4886235829984427, "learning_rate": 3.113558384091964e-05, "loss": 2.6981, "step": 31936 }, { "epoch": 1.9825563349680304, "grad_norm": 0.5010720026372043, "learning_rate": 3.113223927258179e-05, "loss": 2.6604, "step": 31937 }, { "epoch": 1.9826184120677883, "grad_norm": 0.5326074479726514, "learning_rate": 3.1128894802681634e-05, "loss": 2.7885, "step": 31938 }, { "epoch": 1.9826804891675462, "grad_norm": 0.5400088475968374, "learning_rate": 3.1125550431236605e-05, "loss": 2.6616, "step": 31939 }, { "epoch": 1.982742566267304, "grad_norm": 0.5585755031779941, "learning_rate": 3.112220615826418e-05, "loss": 2.7241, "step": 31940 }, { "epoch": 1.9828046433670619, "grad_norm": 0.5712237981050114, "learning_rate": 3.1118861983781766e-05, "loss": 2.7342, "step": 31941 }, { "epoch": 1.9828667204668198, "grad_norm": 0.5442753993912104, "learning_rate": 3.1115517907806853e-05, "loss": 2.7283, "step": 31942 }, { "epoch": 1.9829287975665777, "grad_norm": 0.5567465430427734, "learning_rate": 3.111217393035685e-05, "loss": 2.7122, "step": 31943 }, { "epoch": 1.9829908746663356, "grad_norm": 0.49226599778798585, "learning_rate": 3.110883005144923e-05, "loss": 2.6251, "step": 31944 }, { "epoch": 1.9830529517660935, "grad_norm": 0.5238232116696365, "learning_rate": 3.110548627110142e-05, "loss": 2.6661, "step": 31945 }, { "epoch": 1.9831150288658512, "grad_norm": 0.5233893301136068, "learning_rate": 3.1102142589330894e-05, "loss": 2.6809, "step": 31946 }, { "epoch": 1.9831771059656091, "grad_norm": 0.5217448523413256, "learning_rate": 3.1098799006155075e-05, "loss": 2.6358, "step": 31947 }, { "epoch": 1.983239183065367, "grad_norm": 0.5175613112687628, "learning_rate": 3.10954555215914e-05, "loss": 2.6802, "step": 31948 }, { "epoch": 1.983301260165125, "grad_norm": 0.5042847978370549, "learning_rate": 3.109211213565733e-05, "loss": 2.6792, "step": 31949 }, { "epoch": 1.983363337264883, "grad_norm": 0.5083086964103654, "learning_rate": 3.108876884837029e-05, "loss": 2.674, "step": 31950 }, { "epoch": 1.9834254143646408, "grad_norm": 0.5142636389715599, "learning_rate": 3.108542565974774e-05, "loss": 2.7567, "step": 31951 }, { "epoch": 1.9834874914643987, "grad_norm": 0.5065444666801914, "learning_rate": 3.1082082569807116e-05, "loss": 2.7031, "step": 31952 }, { "epoch": 1.9835495685641567, "grad_norm": 0.5521982241389572, "learning_rate": 3.1078739578565866e-05, "loss": 2.8007, "step": 31953 }, { "epoch": 1.9836116456639146, "grad_norm": 0.5977783826314229, "learning_rate": 3.1075396686041406e-05, "loss": 2.7546, "step": 31954 }, { "epoch": 1.9836737227636725, "grad_norm": 0.5341285527667109, "learning_rate": 3.107205389225121e-05, "loss": 2.7591, "step": 31955 }, { "epoch": 1.9837357998634304, "grad_norm": 0.5650908355781186, "learning_rate": 3.1068711197212695e-05, "loss": 2.6189, "step": 31956 }, { "epoch": 1.9837978769631883, "grad_norm": 0.5992133606590015, "learning_rate": 3.106536860094332e-05, "loss": 2.7285, "step": 31957 }, { "epoch": 1.9838599540629462, "grad_norm": 0.6088405857795953, "learning_rate": 3.106202610346051e-05, "loss": 2.6956, "step": 31958 }, { "epoch": 1.9839220311627042, "grad_norm": 0.6206787930443405, "learning_rate": 3.105868370478171e-05, "loss": 2.7335, "step": 31959 }, { "epoch": 1.983984108262462, "grad_norm": 0.5459658558932409, "learning_rate": 3.105534140492436e-05, "loss": 2.7278, "step": 31960 }, { "epoch": 1.98404618536222, "grad_norm": 0.6072724578681092, "learning_rate": 3.105199920390588e-05, "loss": 2.7078, "step": 31961 }, { "epoch": 1.984108262461978, "grad_norm": 0.4939831569499151, "learning_rate": 3.104865710174373e-05, "loss": 2.6386, "step": 31962 }, { "epoch": 1.9841703395617358, "grad_norm": 0.5542729400133993, "learning_rate": 3.104531509845532e-05, "loss": 2.7467, "step": 31963 }, { "epoch": 1.9842324166614935, "grad_norm": 0.586792055673201, "learning_rate": 3.1041973194058126e-05, "loss": 2.7492, "step": 31964 }, { "epoch": 1.9842944937612514, "grad_norm": 0.6245576723802196, "learning_rate": 3.103863138856954e-05, "loss": 2.7116, "step": 31965 }, { "epoch": 1.9843565708610094, "grad_norm": 0.5660764589124864, "learning_rate": 3.103528968200703e-05, "loss": 2.7842, "step": 31966 }, { "epoch": 1.9844186479607673, "grad_norm": 0.6005412263432139, "learning_rate": 3.103194807438801e-05, "loss": 2.7397, "step": 31967 }, { "epoch": 1.9844807250605252, "grad_norm": 0.5910732768115904, "learning_rate": 3.10286065657299e-05, "loss": 2.73, "step": 31968 }, { "epoch": 1.9845428021602831, "grad_norm": 0.4942223100657627, "learning_rate": 3.102526515605019e-05, "loss": 2.7167, "step": 31969 }, { "epoch": 1.9846048792600408, "grad_norm": 0.5702399638830146, "learning_rate": 3.102192384536626e-05, "loss": 2.6614, "step": 31970 }, { "epoch": 1.9846669563597987, "grad_norm": 0.5788751838369125, "learning_rate": 3.1018582633695576e-05, "loss": 2.704, "step": 31971 }, { "epoch": 1.9847290334595566, "grad_norm": 0.5585622178361156, "learning_rate": 3.101524152105555e-05, "loss": 2.6679, "step": 31972 }, { "epoch": 1.9847911105593146, "grad_norm": 0.5779028543042379, "learning_rate": 3.101190050746361e-05, "loss": 2.6972, "step": 31973 }, { "epoch": 1.9848531876590725, "grad_norm": 0.6438422710992537, "learning_rate": 3.1008559592937206e-05, "loss": 2.7337, "step": 31974 }, { "epoch": 1.9849152647588304, "grad_norm": 0.5057174601380192, "learning_rate": 3.100521877749374e-05, "loss": 2.6796, "step": 31975 }, { "epoch": 1.9849773418585883, "grad_norm": 0.5894852868995061, "learning_rate": 3.100187806115067e-05, "loss": 2.7552, "step": 31976 }, { "epoch": 1.9850394189583462, "grad_norm": 0.5510795002180099, "learning_rate": 3.0998537443925407e-05, "loss": 2.7013, "step": 31977 }, { "epoch": 1.9851014960581042, "grad_norm": 0.5516197099274254, "learning_rate": 3.0995196925835404e-05, "loss": 2.8377, "step": 31978 }, { "epoch": 1.985163573157862, "grad_norm": 0.5333685984972734, "learning_rate": 3.099185650689805e-05, "loss": 2.6561, "step": 31979 }, { "epoch": 1.98522565025762, "grad_norm": 0.5746163788555988, "learning_rate": 3.098851618713081e-05, "loss": 2.7622, "step": 31980 }, { "epoch": 1.985287727357378, "grad_norm": 0.5755259368347998, "learning_rate": 3.0985175966551086e-05, "loss": 2.7334, "step": 31981 }, { "epoch": 1.9853498044571358, "grad_norm": 0.5545042836803773, "learning_rate": 3.098183584517632e-05, "loss": 2.7811, "step": 31982 }, { "epoch": 1.9854118815568937, "grad_norm": 0.5010750954381856, "learning_rate": 3.0978495823023935e-05, "loss": 2.8065, "step": 31983 }, { "epoch": 1.9854739586566517, "grad_norm": 0.5416705027468879, "learning_rate": 3.097515590011135e-05, "loss": 2.5811, "step": 31984 }, { "epoch": 1.9855360357564096, "grad_norm": 0.5376632885969687, "learning_rate": 3.0971816076456004e-05, "loss": 2.6963, "step": 31985 }, { "epoch": 1.9855981128561675, "grad_norm": 0.5235095757352827, "learning_rate": 3.0968476352075304e-05, "loss": 2.7352, "step": 31986 }, { "epoch": 1.9856601899559254, "grad_norm": 0.5447897802238796, "learning_rate": 3.09651367269867e-05, "loss": 2.7003, "step": 31987 }, { "epoch": 1.985722267055683, "grad_norm": 0.5505965443942997, "learning_rate": 3.096179720120758e-05, "loss": 2.7395, "step": 31988 }, { "epoch": 1.985784344155441, "grad_norm": 0.5168091153333749, "learning_rate": 3.095845777475539e-05, "loss": 2.6932, "step": 31989 }, { "epoch": 1.985846421255199, "grad_norm": 0.5448575545876736, "learning_rate": 3.095511844764755e-05, "loss": 2.6987, "step": 31990 }, { "epoch": 1.9859084983549569, "grad_norm": 0.5325238158940159, "learning_rate": 3.0951779219901476e-05, "loss": 2.6705, "step": 31991 }, { "epoch": 1.9859705754547148, "grad_norm": 0.4920720767360287, "learning_rate": 3.0948440091534595e-05, "loss": 2.7479, "step": 31992 }, { "epoch": 1.9860326525544727, "grad_norm": 0.5029135395902027, "learning_rate": 3.094510106256433e-05, "loss": 2.6583, "step": 31993 }, { "epoch": 1.9860947296542304, "grad_norm": 0.4966564741571581, "learning_rate": 3.09417621330081e-05, "loss": 2.7073, "step": 31994 }, { "epoch": 1.9861568067539883, "grad_norm": 0.48810926841852637, "learning_rate": 3.093842330288333e-05, "loss": 2.4994, "step": 31995 }, { "epoch": 1.9862188838537462, "grad_norm": 0.5065926204812582, "learning_rate": 3.093508457220742e-05, "loss": 2.6347, "step": 31996 }, { "epoch": 1.9862809609535041, "grad_norm": 0.5253015038441718, "learning_rate": 3.0931745940997806e-05, "loss": 2.6693, "step": 31997 }, { "epoch": 1.986343038053262, "grad_norm": 0.5091403034344404, "learning_rate": 3.0928407409271906e-05, "loss": 2.6761, "step": 31998 }, { "epoch": 1.98640511515302, "grad_norm": 0.48792180373269833, "learning_rate": 3.092506897704712e-05, "loss": 2.7553, "step": 31999 }, { "epoch": 1.986467192252778, "grad_norm": 0.5246690524449731, "learning_rate": 3.09217306443409e-05, "loss": 2.6947, "step": 32000 }, { "epoch": 1.9865292693525358, "grad_norm": 0.5187589598651758, "learning_rate": 3.091839241117061e-05, "loss": 2.8006, "step": 32001 }, { "epoch": 1.9865913464522937, "grad_norm": 0.5588582086431754, "learning_rate": 3.091505427755372e-05, "loss": 2.7343, "step": 32002 }, { "epoch": 1.9866534235520517, "grad_norm": 0.5056648911535281, "learning_rate": 3.0911716243507635e-05, "loss": 2.6978, "step": 32003 }, { "epoch": 1.9867155006518096, "grad_norm": 0.5553786042324227, "learning_rate": 3.0908378309049746e-05, "loss": 2.6947, "step": 32004 }, { "epoch": 1.9867775777515675, "grad_norm": 0.541259930808077, "learning_rate": 3.0905040474197497e-05, "loss": 2.6832, "step": 32005 }, { "epoch": 1.9868396548513254, "grad_norm": 0.5163801905164068, "learning_rate": 3.090170273896827e-05, "loss": 2.7289, "step": 32006 }, { "epoch": 1.9869017319510833, "grad_norm": 0.543228498941842, "learning_rate": 3.089836510337951e-05, "loss": 2.7488, "step": 32007 }, { "epoch": 1.9869638090508412, "grad_norm": 0.582337804364271, "learning_rate": 3.0895027567448603e-05, "loss": 2.7746, "step": 32008 }, { "epoch": 1.9870258861505992, "grad_norm": 0.5233738419269643, "learning_rate": 3.089169013119299e-05, "loss": 2.688, "step": 32009 }, { "epoch": 1.987087963250357, "grad_norm": 0.5459648460435074, "learning_rate": 3.0888352794630064e-05, "loss": 2.7244, "step": 32010 }, { "epoch": 1.987150040350115, "grad_norm": 0.5132858218101218, "learning_rate": 3.0885015557777234e-05, "loss": 2.7275, "step": 32011 }, { "epoch": 1.9872121174498727, "grad_norm": 0.5148310762503905, "learning_rate": 3.088167842065193e-05, "loss": 2.731, "step": 32012 }, { "epoch": 1.9872741945496306, "grad_norm": 0.5525310128772797, "learning_rate": 3.087834138327154e-05, "loss": 2.6638, "step": 32013 }, { "epoch": 1.9873362716493885, "grad_norm": 0.5173528792930403, "learning_rate": 3.0875004445653486e-05, "loss": 2.7366, "step": 32014 }, { "epoch": 1.9873983487491464, "grad_norm": 0.4800757840303561, "learning_rate": 3.087166760781518e-05, "loss": 2.6656, "step": 32015 }, { "epoch": 1.9874604258489044, "grad_norm": 0.4916837172730647, "learning_rate": 3.086833086977403e-05, "loss": 2.6989, "step": 32016 }, { "epoch": 1.9875225029486623, "grad_norm": 0.4900974371188297, "learning_rate": 3.086499423154743e-05, "loss": 2.6995, "step": 32017 }, { "epoch": 1.98758458004842, "grad_norm": 0.5153404375395119, "learning_rate": 3.086165769315281e-05, "loss": 2.6608, "step": 32018 }, { "epoch": 1.987646657148178, "grad_norm": 0.5324376798706253, "learning_rate": 3.0858321254607567e-05, "loss": 2.7252, "step": 32019 }, { "epoch": 1.9877087342479358, "grad_norm": 0.5555644021389521, "learning_rate": 3.0854984915929114e-05, "loss": 2.6103, "step": 32020 }, { "epoch": 1.9877708113476937, "grad_norm": 0.5254193549827313, "learning_rate": 3.085164867713485e-05, "loss": 2.6141, "step": 32021 }, { "epoch": 1.9878328884474517, "grad_norm": 0.5538718176552089, "learning_rate": 3.0848312538242174e-05, "loss": 2.7486, "step": 32022 }, { "epoch": 1.9878949655472096, "grad_norm": 0.552518714915779, "learning_rate": 3.084497649926852e-05, "loss": 2.7636, "step": 32023 }, { "epoch": 1.9879570426469675, "grad_norm": 0.5099015328774208, "learning_rate": 3.084164056023125e-05, "loss": 2.7372, "step": 32024 }, { "epoch": 1.9880191197467254, "grad_norm": 0.6013411325160752, "learning_rate": 3.08383047211478e-05, "loss": 2.6964, "step": 32025 }, { "epoch": 1.9880811968464833, "grad_norm": 0.4987814478890843, "learning_rate": 3.083496898203557e-05, "loss": 2.7098, "step": 32026 }, { "epoch": 1.9881432739462412, "grad_norm": 0.5992958665783846, "learning_rate": 3.0831633342911966e-05, "loss": 2.7291, "step": 32027 }, { "epoch": 1.9882053510459992, "grad_norm": 0.5904920690796883, "learning_rate": 3.082829780379437e-05, "loss": 2.7985, "step": 32028 }, { "epoch": 1.988267428145757, "grad_norm": 0.5009988999173847, "learning_rate": 3.0824962364700206e-05, "loss": 2.7803, "step": 32029 }, { "epoch": 1.988329505245515, "grad_norm": 0.5825206567924478, "learning_rate": 3.082162702564686e-05, "loss": 2.7404, "step": 32030 }, { "epoch": 1.988391582345273, "grad_norm": 0.47568249358721704, "learning_rate": 3.0818291786651754e-05, "loss": 2.7549, "step": 32031 }, { "epoch": 1.9884536594450308, "grad_norm": 0.5241236104348611, "learning_rate": 3.081495664773228e-05, "loss": 2.7526, "step": 32032 }, { "epoch": 1.9885157365447887, "grad_norm": 0.5103936838957294, "learning_rate": 3.0811621608905815e-05, "loss": 2.7684, "step": 32033 }, { "epoch": 1.9885778136445467, "grad_norm": 0.5322663738514783, "learning_rate": 3.080828667018978e-05, "loss": 2.7453, "step": 32034 }, { "epoch": 1.9886398907443046, "grad_norm": 0.5043294445485879, "learning_rate": 3.080495183160158e-05, "loss": 2.7316, "step": 32035 }, { "epoch": 1.9887019678440623, "grad_norm": 0.5957929449686861, "learning_rate": 3.08016170931586e-05, "loss": 2.788, "step": 32036 }, { "epoch": 1.9887640449438202, "grad_norm": 0.5303062302371082, "learning_rate": 3.079828245487826e-05, "loss": 2.6679, "step": 32037 }, { "epoch": 1.9888261220435781, "grad_norm": 0.5438520259068569, "learning_rate": 3.0794947916777926e-05, "loss": 2.6553, "step": 32038 }, { "epoch": 1.988888199143336, "grad_norm": 0.5887913694648441, "learning_rate": 3.079161347887502e-05, "loss": 2.7518, "step": 32039 }, { "epoch": 1.988950276243094, "grad_norm": 0.5448335965749984, "learning_rate": 3.078827914118693e-05, "loss": 2.7487, "step": 32040 }, { "epoch": 1.9890123533428519, "grad_norm": 0.5321690519565959, "learning_rate": 3.078494490373105e-05, "loss": 2.6924, "step": 32041 }, { "epoch": 1.9890744304426096, "grad_norm": 0.5760595179775815, "learning_rate": 3.078161076652477e-05, "loss": 2.7269, "step": 32042 }, { "epoch": 1.9891365075423675, "grad_norm": 0.5423571483349522, "learning_rate": 3.07782767295855e-05, "loss": 2.6987, "step": 32043 }, { "epoch": 1.9891985846421254, "grad_norm": 0.5866399496219135, "learning_rate": 3.077494279293062e-05, "loss": 2.7115, "step": 32044 }, { "epoch": 1.9892606617418833, "grad_norm": 0.5320758851271106, "learning_rate": 3.0771608956577533e-05, "loss": 2.8261, "step": 32045 }, { "epoch": 1.9893227388416412, "grad_norm": 0.5561383306732323, "learning_rate": 3.076827522054364e-05, "loss": 2.7748, "step": 32046 }, { "epoch": 1.9893848159413992, "grad_norm": 0.5054810040505306, "learning_rate": 3.076494158484631e-05, "loss": 2.7871, "step": 32047 }, { "epoch": 1.989446893041157, "grad_norm": 0.5312323567321725, "learning_rate": 3.076160804950296e-05, "loss": 2.7374, "step": 32048 }, { "epoch": 1.989508970140915, "grad_norm": 0.4834825076585024, "learning_rate": 3.0758274614530964e-05, "loss": 2.6571, "step": 32049 }, { "epoch": 1.989571047240673, "grad_norm": 0.47880676132246747, "learning_rate": 3.075494127994772e-05, "loss": 2.6732, "step": 32050 }, { "epoch": 1.9896331243404308, "grad_norm": 0.5041899067537672, "learning_rate": 3.075160804577062e-05, "loss": 2.6228, "step": 32051 }, { "epoch": 1.9896952014401887, "grad_norm": 0.5125490979192749, "learning_rate": 3.074827491201706e-05, "loss": 2.6901, "step": 32052 }, { "epoch": 1.9897572785399467, "grad_norm": 0.5474639446109272, "learning_rate": 3.0744941878704414e-05, "loss": 2.6575, "step": 32053 }, { "epoch": 1.9898193556397046, "grad_norm": 0.4724811986432129, "learning_rate": 3.074160894585009e-05, "loss": 2.6429, "step": 32054 }, { "epoch": 1.9898814327394625, "grad_norm": 0.5219834471997586, "learning_rate": 3.073827611347145e-05, "loss": 2.7264, "step": 32055 }, { "epoch": 1.9899435098392204, "grad_norm": 0.5437726991672901, "learning_rate": 3.0734943381585924e-05, "loss": 2.7867, "step": 32056 }, { "epoch": 1.9900055869389783, "grad_norm": 0.4932908150220126, "learning_rate": 3.073161075021087e-05, "loss": 2.6694, "step": 32057 }, { "epoch": 1.9900676640387363, "grad_norm": 0.5462901121203773, "learning_rate": 3.0728278219363666e-05, "loss": 2.8015, "step": 32058 }, { "epoch": 1.9901297411384942, "grad_norm": 0.5067214983625334, "learning_rate": 3.072494578906172e-05, "loss": 2.7285, "step": 32059 }, { "epoch": 1.9901918182382519, "grad_norm": 0.5003748621010096, "learning_rate": 3.072161345932241e-05, "loss": 2.6157, "step": 32060 }, { "epoch": 1.9902538953380098, "grad_norm": 0.5343229670037619, "learning_rate": 3.0718281230163126e-05, "loss": 2.8014, "step": 32061 }, { "epoch": 1.9903159724377677, "grad_norm": 0.5185368134315032, "learning_rate": 3.071494910160124e-05, "loss": 2.8019, "step": 32062 }, { "epoch": 1.9903780495375256, "grad_norm": 0.5370140684707327, "learning_rate": 3.071161707365416e-05, "loss": 2.7031, "step": 32063 }, { "epoch": 1.9904401266372835, "grad_norm": 0.542153714141465, "learning_rate": 3.0708285146339244e-05, "loss": 2.7345, "step": 32064 }, { "epoch": 1.9905022037370415, "grad_norm": 0.5470150087152549, "learning_rate": 3.07049533196739e-05, "loss": 2.806, "step": 32065 }, { "epoch": 1.9905642808367991, "grad_norm": 0.5010452966961714, "learning_rate": 3.070162159367548e-05, "loss": 2.7563, "step": 32066 }, { "epoch": 1.990626357936557, "grad_norm": 0.5380527880750696, "learning_rate": 3.0698289968361385e-05, "loss": 2.6988, "step": 32067 }, { "epoch": 1.990688435036315, "grad_norm": 0.5397432424167324, "learning_rate": 3.069495844374901e-05, "loss": 2.7127, "step": 32068 }, { "epoch": 1.990750512136073, "grad_norm": 0.49645569881338353, "learning_rate": 3.0691627019855716e-05, "loss": 2.7186, "step": 32069 }, { "epoch": 1.9908125892358308, "grad_norm": 0.5793575087934708, "learning_rate": 3.068829569669891e-05, "loss": 2.7404, "step": 32070 }, { "epoch": 1.9908746663355887, "grad_norm": 0.5838880427591638, "learning_rate": 3.068496447429594e-05, "loss": 2.7917, "step": 32071 }, { "epoch": 1.9909367434353467, "grad_norm": 0.5632233736777278, "learning_rate": 3.06816333526642e-05, "loss": 2.7345, "step": 32072 }, { "epoch": 1.9909988205351046, "grad_norm": 0.583254150284293, "learning_rate": 3.067830233182107e-05, "loss": 2.6823, "step": 32073 }, { "epoch": 1.9910608976348625, "grad_norm": 0.6095689562703317, "learning_rate": 3.0674971411783916e-05, "loss": 2.7228, "step": 32074 }, { "epoch": 1.9911229747346204, "grad_norm": 0.5821289390150552, "learning_rate": 3.0671640592570144e-05, "loss": 2.705, "step": 32075 }, { "epoch": 1.9911850518343783, "grad_norm": 0.5466982147032045, "learning_rate": 3.06683098741971e-05, "loss": 2.727, "step": 32076 }, { "epoch": 1.9912471289341362, "grad_norm": 0.5201249136728712, "learning_rate": 3.066497925668219e-05, "loss": 2.6933, "step": 32077 }, { "epoch": 1.9913092060338942, "grad_norm": 0.563601184200521, "learning_rate": 3.066164874004277e-05, "loss": 2.7048, "step": 32078 }, { "epoch": 1.991371283133652, "grad_norm": 0.5079870950888213, "learning_rate": 3.065831832429623e-05, "loss": 2.7531, "step": 32079 }, { "epoch": 1.99143336023341, "grad_norm": 0.5563419544636897, "learning_rate": 3.0654988009459926e-05, "loss": 2.7859, "step": 32080 }, { "epoch": 1.991495437333168, "grad_norm": 0.5138885853167946, "learning_rate": 3.065165779555126e-05, "loss": 2.6938, "step": 32081 }, { "epoch": 1.9915575144329258, "grad_norm": 0.5378122897372147, "learning_rate": 3.064832768258758e-05, "loss": 2.6599, "step": 32082 }, { "epoch": 1.9916195915326838, "grad_norm": 0.5378757627119631, "learning_rate": 3.064499767058628e-05, "loss": 2.7506, "step": 32083 }, { "epoch": 1.9916816686324414, "grad_norm": 0.5305167354852661, "learning_rate": 3.0641667759564725e-05, "loss": 2.7216, "step": 32084 }, { "epoch": 1.9917437457321994, "grad_norm": 0.5216210004158178, "learning_rate": 3.063833794954028e-05, "loss": 2.7047, "step": 32085 }, { "epoch": 1.9918058228319573, "grad_norm": 0.5433471239520217, "learning_rate": 3.0635008240530336e-05, "loss": 2.7713, "step": 32086 }, { "epoch": 1.9918678999317152, "grad_norm": 0.5003691443091849, "learning_rate": 3.063167863255225e-05, "loss": 2.7375, "step": 32087 }, { "epoch": 1.9919299770314731, "grad_norm": 0.5399896968194183, "learning_rate": 3.06283491256234e-05, "loss": 2.741, "step": 32088 }, { "epoch": 1.991992054131231, "grad_norm": 0.45431577490658615, "learning_rate": 3.0625019719761156e-05, "loss": 2.6276, "step": 32089 }, { "epoch": 1.9920541312309887, "grad_norm": 0.5225262752570143, "learning_rate": 3.062169041498289e-05, "loss": 2.7297, "step": 32090 }, { "epoch": 1.9921162083307467, "grad_norm": 0.44790964527689, "learning_rate": 3.061836121130596e-05, "loss": 2.6987, "step": 32091 }, { "epoch": 1.9921782854305046, "grad_norm": 0.5244759180397055, "learning_rate": 3.061503210874775e-05, "loss": 2.7039, "step": 32092 }, { "epoch": 1.9922403625302625, "grad_norm": 0.5038457434944323, "learning_rate": 3.061170310732563e-05, "loss": 2.6948, "step": 32093 }, { "epoch": 1.9923024396300204, "grad_norm": 0.5046297837551061, "learning_rate": 3.060837420705696e-05, "loss": 2.7206, "step": 32094 }, { "epoch": 1.9923645167297783, "grad_norm": 0.508768267947927, "learning_rate": 3.0605045407959107e-05, "loss": 2.8334, "step": 32095 }, { "epoch": 1.9924265938295362, "grad_norm": 0.5760281658372528, "learning_rate": 3.060171671004943e-05, "loss": 2.8208, "step": 32096 }, { "epoch": 1.9924886709292942, "grad_norm": 0.5133423128777312, "learning_rate": 3.059838811334532e-05, "loss": 2.6899, "step": 32097 }, { "epoch": 1.992550748029052, "grad_norm": 0.5030894201741497, "learning_rate": 3.059505961786412e-05, "loss": 2.7142, "step": 32098 }, { "epoch": 1.99261282512881, "grad_norm": 0.5649467606484457, "learning_rate": 3.0591731223623195e-05, "loss": 2.7773, "step": 32099 }, { "epoch": 1.992674902228568, "grad_norm": 0.5260897384685482, "learning_rate": 3.0588402930639936e-05, "loss": 2.8287, "step": 32100 }, { "epoch": 1.9927369793283258, "grad_norm": 0.5317566363176561, "learning_rate": 3.0585074738931684e-05, "loss": 2.7028, "step": 32101 }, { "epoch": 1.9927990564280837, "grad_norm": 0.588673169008692, "learning_rate": 3.058174664851582e-05, "loss": 2.6631, "step": 32102 }, { "epoch": 1.9928611335278417, "grad_norm": 0.527406392085644, "learning_rate": 3.057841865940969e-05, "loss": 2.7538, "step": 32103 }, { "epoch": 1.9929232106275996, "grad_norm": 0.5920475595643497, "learning_rate": 3.057509077163067e-05, "loss": 2.7243, "step": 32104 }, { "epoch": 1.9929852877273575, "grad_norm": 0.5127449520240164, "learning_rate": 3.05717629851961e-05, "loss": 2.714, "step": 32105 }, { "epoch": 1.9930473648271154, "grad_norm": 0.5466682315674147, "learning_rate": 3.056843530012338e-05, "loss": 2.5913, "step": 32106 }, { "epoch": 1.9931094419268733, "grad_norm": 0.5645083232144397, "learning_rate": 3.0565107716429854e-05, "loss": 2.7273, "step": 32107 }, { "epoch": 1.993171519026631, "grad_norm": 0.531400911550562, "learning_rate": 3.056178023413286e-05, "loss": 2.743, "step": 32108 }, { "epoch": 1.993233596126389, "grad_norm": 0.5354844360903624, "learning_rate": 3.055845285324979e-05, "loss": 2.6842, "step": 32109 }, { "epoch": 1.9932956732261469, "grad_norm": 0.5775626772766865, "learning_rate": 3.0555125573797984e-05, "loss": 2.7571, "step": 32110 }, { "epoch": 1.9933577503259048, "grad_norm": 0.5339950995774031, "learning_rate": 3.055179839579481e-05, "loss": 2.6828, "step": 32111 }, { "epoch": 1.9934198274256627, "grad_norm": 0.5284795891204724, "learning_rate": 3.054847131925762e-05, "loss": 2.6526, "step": 32112 }, { "epoch": 1.9934819045254206, "grad_norm": 0.5381775378234513, "learning_rate": 3.054514434420379e-05, "loss": 2.7066, "step": 32113 }, { "epoch": 1.9935439816251783, "grad_norm": 0.5057168708299155, "learning_rate": 3.0541817470650645e-05, "loss": 2.7535, "step": 32114 }, { "epoch": 1.9936060587249362, "grad_norm": 0.5272543937424065, "learning_rate": 3.053849069861558e-05, "loss": 2.7595, "step": 32115 }, { "epoch": 1.9936681358246942, "grad_norm": 0.5506739681002683, "learning_rate": 3.0535164028115924e-05, "loss": 2.6296, "step": 32116 }, { "epoch": 1.993730212924452, "grad_norm": 0.5085182860564598, "learning_rate": 3.053183745916905e-05, "loss": 2.7376, "step": 32117 }, { "epoch": 1.99379229002421, "grad_norm": 0.5681499868926557, "learning_rate": 3.052851099179231e-05, "loss": 2.6952, "step": 32118 }, { "epoch": 1.993854367123968, "grad_norm": 0.5104981052539123, "learning_rate": 3.052518462600305e-05, "loss": 2.805, "step": 32119 }, { "epoch": 1.9939164442237258, "grad_norm": 0.5465995094676771, "learning_rate": 3.052185836181862e-05, "loss": 2.712, "step": 32120 }, { "epoch": 1.9939785213234837, "grad_norm": 0.5794808752114865, "learning_rate": 3.0518532199256386e-05, "loss": 2.6473, "step": 32121 }, { "epoch": 1.9940405984232417, "grad_norm": 0.5055769201743511, "learning_rate": 3.0515206138333713e-05, "loss": 2.6988, "step": 32122 }, { "epoch": 1.9941026755229996, "grad_norm": 0.6242864111370813, "learning_rate": 3.051188017906792e-05, "loss": 2.7265, "step": 32123 }, { "epoch": 1.9941647526227575, "grad_norm": 0.4960898752136295, "learning_rate": 3.0508554321476395e-05, "loss": 2.7736, "step": 32124 }, { "epoch": 1.9942268297225154, "grad_norm": 0.5125721229419267, "learning_rate": 3.050522856557646e-05, "loss": 2.655, "step": 32125 }, { "epoch": 1.9942889068222733, "grad_norm": 0.5716187919398044, "learning_rate": 3.0501902911385495e-05, "loss": 2.6905, "step": 32126 }, { "epoch": 1.9943509839220313, "grad_norm": 0.5273926888822389, "learning_rate": 3.049857735892082e-05, "loss": 2.6234, "step": 32127 }, { "epoch": 1.9944130610217892, "grad_norm": 0.5001480658164584, "learning_rate": 3.0495251908199817e-05, "loss": 2.7889, "step": 32128 }, { "epoch": 1.994475138121547, "grad_norm": 0.5261914131371561, "learning_rate": 3.0491926559239808e-05, "loss": 2.722, "step": 32129 }, { "epoch": 1.994537215221305, "grad_norm": 0.550984554780466, "learning_rate": 3.0488601312058164e-05, "loss": 2.7633, "step": 32130 }, { "epoch": 1.994599292321063, "grad_norm": 0.46906089009841656, "learning_rate": 3.0485276166672223e-05, "loss": 2.7545, "step": 32131 }, { "epoch": 1.9946613694208206, "grad_norm": 0.6137555954594582, "learning_rate": 3.048195112309931e-05, "loss": 2.7528, "step": 32132 }, { "epoch": 1.9947234465205785, "grad_norm": 0.5391135813518544, "learning_rate": 3.0478626181356812e-05, "loss": 2.7657, "step": 32133 }, { "epoch": 1.9947855236203365, "grad_norm": 0.5289425385280367, "learning_rate": 3.047530134146207e-05, "loss": 2.7523, "step": 32134 }, { "epoch": 1.9948476007200944, "grad_norm": 0.5065909732674028, "learning_rate": 3.0471976603432408e-05, "loss": 2.7176, "step": 32135 }, { "epoch": 1.9949096778198523, "grad_norm": 0.5466077969223555, "learning_rate": 3.046865196728519e-05, "loss": 2.8387, "step": 32136 }, { "epoch": 1.9949717549196102, "grad_norm": 0.5432981959879803, "learning_rate": 3.046532743303775e-05, "loss": 2.7497, "step": 32137 }, { "epoch": 1.995033832019368, "grad_norm": 0.5308723509874893, "learning_rate": 3.046200300070745e-05, "loss": 2.6451, "step": 32138 }, { "epoch": 1.9950959091191258, "grad_norm": 0.5569065730245578, "learning_rate": 3.0458678670311612e-05, "loss": 2.6925, "step": 32139 }, { "epoch": 1.9951579862188837, "grad_norm": 0.5388166170908765, "learning_rate": 3.0455354441867606e-05, "loss": 2.6653, "step": 32140 }, { "epoch": 1.9952200633186417, "grad_norm": 0.5026162365137794, "learning_rate": 3.0452030315392743e-05, "loss": 2.6747, "step": 32141 }, { "epoch": 1.9952821404183996, "grad_norm": 0.5405213339315573, "learning_rate": 3.0448706290904404e-05, "loss": 2.7575, "step": 32142 }, { "epoch": 1.9953442175181575, "grad_norm": 0.48873132943202335, "learning_rate": 3.04453823684199e-05, "loss": 2.6456, "step": 32143 }, { "epoch": 1.9954062946179154, "grad_norm": 0.5123215248287509, "learning_rate": 3.0442058547956577e-05, "loss": 2.7045, "step": 32144 }, { "epoch": 1.9954683717176733, "grad_norm": 0.5292004993716772, "learning_rate": 3.043873482953179e-05, "loss": 2.7068, "step": 32145 }, { "epoch": 1.9955304488174312, "grad_norm": 0.48982253386138014, "learning_rate": 3.0435411213162868e-05, "loss": 2.6566, "step": 32146 }, { "epoch": 1.9955925259171892, "grad_norm": 0.48770531830092667, "learning_rate": 3.0432087698867163e-05, "loss": 2.6532, "step": 32147 }, { "epoch": 1.995654603016947, "grad_norm": 0.5516483164479209, "learning_rate": 3.042876428666199e-05, "loss": 2.771, "step": 32148 }, { "epoch": 1.995716680116705, "grad_norm": 0.4899128319123625, "learning_rate": 3.0425440976564723e-05, "loss": 2.6993, "step": 32149 }, { "epoch": 1.995778757216463, "grad_norm": 0.5395899417526526, "learning_rate": 3.0422117768592673e-05, "loss": 2.7401, "step": 32150 }, { "epoch": 1.9958408343162208, "grad_norm": 0.49226658806330525, "learning_rate": 3.0418794662763194e-05, "loss": 2.7947, "step": 32151 }, { "epoch": 1.9959029114159788, "grad_norm": 0.49848940767362104, "learning_rate": 3.0415471659093603e-05, "loss": 2.6889, "step": 32152 }, { "epoch": 1.9959649885157367, "grad_norm": 0.494154993986713, "learning_rate": 3.0412148757601265e-05, "loss": 2.6385, "step": 32153 }, { "epoch": 1.9960270656154946, "grad_norm": 0.5998716992054717, "learning_rate": 3.040882595830349e-05, "loss": 2.75, "step": 32154 }, { "epoch": 1.9960891427152525, "grad_norm": 0.49035734177028617, "learning_rate": 3.040550326121764e-05, "loss": 2.7722, "step": 32155 }, { "epoch": 1.9961512198150102, "grad_norm": 0.5494081412117318, "learning_rate": 3.0402180666361037e-05, "loss": 2.706, "step": 32156 }, { "epoch": 1.9962132969147681, "grad_norm": 0.5767155813638088, "learning_rate": 3.0398858173751e-05, "loss": 2.7093, "step": 32157 }, { "epoch": 1.996275374014526, "grad_norm": 0.5961586883090333, "learning_rate": 3.0395535783404882e-05, "loss": 2.7248, "step": 32158 }, { "epoch": 1.996337451114284, "grad_norm": 0.5238172377806894, "learning_rate": 3.0392213495340015e-05, "loss": 2.6638, "step": 32159 }, { "epoch": 1.9963995282140419, "grad_norm": 0.5666117591295451, "learning_rate": 3.0388891309573735e-05, "loss": 2.7436, "step": 32160 }, { "epoch": 1.9964616053137998, "grad_norm": 0.5266631361334001, "learning_rate": 3.038556922612336e-05, "loss": 2.6173, "step": 32161 }, { "epoch": 1.9965236824135575, "grad_norm": 0.5456693711214774, "learning_rate": 3.0382247245006246e-05, "loss": 2.7135, "step": 32162 }, { "epoch": 1.9965857595133154, "grad_norm": 0.4896864241422071, "learning_rate": 3.0378925366239696e-05, "loss": 2.77, "step": 32163 }, { "epoch": 1.9966478366130733, "grad_norm": 0.5289792448580141, "learning_rate": 3.0375603589841062e-05, "loss": 2.6564, "step": 32164 }, { "epoch": 1.9967099137128312, "grad_norm": 0.49338925196414946, "learning_rate": 3.0372281915827654e-05, "loss": 2.5934, "step": 32165 }, { "epoch": 1.9967719908125892, "grad_norm": 0.5020320548256799, "learning_rate": 3.0368960344216823e-05, "loss": 2.7005, "step": 32166 }, { "epoch": 1.996834067912347, "grad_norm": 0.4581530407836829, "learning_rate": 3.0365638875025902e-05, "loss": 2.7349, "step": 32167 }, { "epoch": 1.996896145012105, "grad_norm": 0.525331430043042, "learning_rate": 3.0362317508272215e-05, "loss": 2.6461, "step": 32168 }, { "epoch": 1.996958222111863, "grad_norm": 0.4660809025926441, "learning_rate": 3.035899624397307e-05, "loss": 2.8267, "step": 32169 }, { "epoch": 1.9970202992116208, "grad_norm": 0.5405381226606355, "learning_rate": 3.0355675082145817e-05, "loss": 2.7049, "step": 32170 }, { "epoch": 1.9970823763113787, "grad_norm": 0.4771714775278407, "learning_rate": 3.0352354022807773e-05, "loss": 2.6397, "step": 32171 }, { "epoch": 1.9971444534111367, "grad_norm": 0.5196867115062942, "learning_rate": 3.0349033065976274e-05, "loss": 2.7448, "step": 32172 }, { "epoch": 1.9972065305108946, "grad_norm": 0.5210212336787605, "learning_rate": 3.0345712211668627e-05, "loss": 2.7112, "step": 32173 }, { "epoch": 1.9972686076106525, "grad_norm": 0.5187765964905753, "learning_rate": 3.0342391459902185e-05, "loss": 2.697, "step": 32174 }, { "epoch": 1.9973306847104104, "grad_norm": 0.49641678272728584, "learning_rate": 3.0339070810694248e-05, "loss": 2.7365, "step": 32175 }, { "epoch": 1.9973927618101683, "grad_norm": 0.5481805204765468, "learning_rate": 3.0335750264062157e-05, "loss": 2.7294, "step": 32176 }, { "epoch": 1.9974548389099263, "grad_norm": 0.5061257445352758, "learning_rate": 3.0332429820023222e-05, "loss": 2.7385, "step": 32177 }, { "epoch": 1.9975169160096842, "grad_norm": 0.53263198389946, "learning_rate": 3.0329109478594786e-05, "loss": 2.7895, "step": 32178 }, { "epoch": 1.997578993109442, "grad_norm": 0.5908150664798089, "learning_rate": 3.0325789239794157e-05, "loss": 2.7, "step": 32179 }, { "epoch": 1.9976410702091998, "grad_norm": 0.4807658478727899, "learning_rate": 3.0322469103638664e-05, "loss": 2.6086, "step": 32180 }, { "epoch": 1.9977031473089577, "grad_norm": 0.5870706409664354, "learning_rate": 3.031914907014563e-05, "loss": 2.7491, "step": 32181 }, { "epoch": 1.9977652244087156, "grad_norm": 0.5803132527910895, "learning_rate": 3.0315829139332364e-05, "loss": 2.6977, "step": 32182 }, { "epoch": 1.9978273015084735, "grad_norm": 0.6210826656863959, "learning_rate": 3.0312509311216198e-05, "loss": 2.7408, "step": 32183 }, { "epoch": 1.9978893786082315, "grad_norm": 0.5354504577076931, "learning_rate": 3.0309189585814445e-05, "loss": 2.6451, "step": 32184 }, { "epoch": 1.9979514557079894, "grad_norm": 0.5881298756980068, "learning_rate": 3.0305869963144446e-05, "loss": 2.7669, "step": 32185 }, { "epoch": 1.998013532807747, "grad_norm": 0.5853226805706372, "learning_rate": 3.0302550443223488e-05, "loss": 2.7704, "step": 32186 }, { "epoch": 1.998075609907505, "grad_norm": 0.5535625817431377, "learning_rate": 3.029923102606892e-05, "loss": 2.7199, "step": 32187 }, { "epoch": 1.998137687007263, "grad_norm": 0.5612646418745229, "learning_rate": 3.0295911711698034e-05, "loss": 2.7384, "step": 32188 }, { "epoch": 1.9981997641070208, "grad_norm": 0.5568123535813749, "learning_rate": 3.029259250012817e-05, "loss": 2.7027, "step": 32189 }, { "epoch": 1.9982618412067787, "grad_norm": 0.5613784204725811, "learning_rate": 3.0289273391376626e-05, "loss": 2.8254, "step": 32190 }, { "epoch": 1.9983239183065367, "grad_norm": 0.5524497495644636, "learning_rate": 3.0285954385460735e-05, "loss": 2.8246, "step": 32191 }, { "epoch": 1.9983859954062946, "grad_norm": 0.540681784188971, "learning_rate": 3.028263548239781e-05, "loss": 2.7411, "step": 32192 }, { "epoch": 1.9984480725060525, "grad_norm": 0.5493209363962508, "learning_rate": 3.027931668220515e-05, "loss": 2.7144, "step": 32193 }, { "epoch": 1.9985101496058104, "grad_norm": 0.5371031882189847, "learning_rate": 3.027599798490009e-05, "loss": 2.7068, "step": 32194 }, { "epoch": 1.9985722267055683, "grad_norm": 0.5229693639527687, "learning_rate": 3.0272679390499924e-05, "loss": 2.7515, "step": 32195 }, { "epoch": 1.9986343038053263, "grad_norm": 0.5322835349291065, "learning_rate": 3.0269360899022e-05, "loss": 2.7294, "step": 32196 }, { "epoch": 1.9986963809050842, "grad_norm": 0.47872713046816284, "learning_rate": 3.0266042510483582e-05, "loss": 2.6591, "step": 32197 }, { "epoch": 1.998758458004842, "grad_norm": 0.5577791316205744, "learning_rate": 3.0262724224902016e-05, "loss": 2.7466, "step": 32198 }, { "epoch": 1.9988205351046, "grad_norm": 0.5319417208769783, "learning_rate": 3.0259406042294625e-05, "loss": 2.6445, "step": 32199 }, { "epoch": 1.998882612204358, "grad_norm": 0.47768139424527684, "learning_rate": 3.02560879626787e-05, "loss": 2.74, "step": 32200 }, { "epoch": 1.9989446893041158, "grad_norm": 0.5406111434307374, "learning_rate": 3.025276998607156e-05, "loss": 2.7858, "step": 32201 }, { "epoch": 1.9990067664038738, "grad_norm": 0.4922643329050663, "learning_rate": 3.0249452112490506e-05, "loss": 2.6703, "step": 32202 }, { "epoch": 1.9990688435036317, "grad_norm": 0.5158563730561838, "learning_rate": 3.024613434195286e-05, "loss": 2.6869, "step": 32203 }, { "epoch": 1.9991309206033894, "grad_norm": 0.562910248244274, "learning_rate": 3.0242816674475925e-05, "loss": 2.68, "step": 32204 }, { "epoch": 1.9991929977031473, "grad_norm": 0.5573896755657087, "learning_rate": 3.023949911007702e-05, "loss": 2.7082, "step": 32205 }, { "epoch": 1.9992550748029052, "grad_norm": 0.5421128971982123, "learning_rate": 3.0236181648773436e-05, "loss": 2.708, "step": 32206 }, { "epoch": 1.9993171519026631, "grad_norm": 0.5452816503592066, "learning_rate": 3.0232864290582484e-05, "loss": 2.7256, "step": 32207 }, { "epoch": 1.999379229002421, "grad_norm": 0.5406443013621078, "learning_rate": 3.0229547035521493e-05, "loss": 2.715, "step": 32208 }, { "epoch": 1.999441306102179, "grad_norm": 0.5335003091030066, "learning_rate": 3.0226229883607743e-05, "loss": 2.6487, "step": 32209 }, { "epoch": 1.9995033832019367, "grad_norm": 0.5354336486033129, "learning_rate": 3.0222912834858557e-05, "loss": 2.7354, "step": 32210 }, { "epoch": 1.9995654603016946, "grad_norm": 0.5494971219665161, "learning_rate": 3.021959588929123e-05, "loss": 2.7301, "step": 32211 }, { "epoch": 1.9996275374014525, "grad_norm": 0.5723169453493988, "learning_rate": 3.0216279046923084e-05, "loss": 2.7013, "step": 32212 }, { "epoch": 1.9996896145012104, "grad_norm": 0.5586534214826253, "learning_rate": 3.0212962307771396e-05, "loss": 2.6925, "step": 32213 }, { "epoch": 1.9997516916009683, "grad_norm": 0.5117057041514516, "learning_rate": 3.0209645671853502e-05, "loss": 2.6354, "step": 32214 }, { "epoch": 1.9998137687007262, "grad_norm": 0.5479416118740363, "learning_rate": 3.0206329139186685e-05, "loss": 2.7678, "step": 32215 }, { "epoch": 1.9998758458004842, "grad_norm": 0.5018280704737118, "learning_rate": 3.0203012709788258e-05, "loss": 2.705, "step": 32216 }, { "epoch": 1.999937922900242, "grad_norm": 0.5843892049847907, "learning_rate": 3.0199696383675524e-05, "loss": 2.6912, "step": 32217 }, { "epoch": 2.0, "grad_norm": 0.5811136937033379, "learning_rate": 3.0196380160865766e-05, "loss": 2.7499, "step": 32218 }, { "epoch": 2.000062077099758, "grad_norm": 0.5880361231214613, "learning_rate": 3.0193064041376318e-05, "loss": 2.6841, "step": 32219 }, { "epoch": 2.000124154199516, "grad_norm": 0.6012148162000882, "learning_rate": 3.0189748025224446e-05, "loss": 2.721, "step": 32220 }, { "epoch": 2.0001862312992738, "grad_norm": 0.5416543543467778, "learning_rate": 3.0186432112427476e-05, "loss": 2.8076, "step": 32221 }, { "epoch": 2.0002483083990317, "grad_norm": 0.6071260473985691, "learning_rate": 3.0183116303002696e-05, "loss": 2.711, "step": 32222 }, { "epoch": 2.0003103854987896, "grad_norm": 0.47728944272850127, "learning_rate": 3.0179800596967417e-05, "loss": 2.6809, "step": 32223 }, { "epoch": 2.0003724625985475, "grad_norm": 0.5756839561299485, "learning_rate": 3.0176484994338917e-05, "loss": 2.6181, "step": 32224 }, { "epoch": 2.0004345396983054, "grad_norm": 0.5259863160224494, "learning_rate": 3.017316949513452e-05, "loss": 2.6867, "step": 32225 }, { "epoch": 2.0004966167980633, "grad_norm": 0.48329674778240084, "learning_rate": 3.0169854099371498e-05, "loss": 2.6468, "step": 32226 }, { "epoch": 2.0005586938978213, "grad_norm": 0.5874479534551519, "learning_rate": 3.0166538807067175e-05, "loss": 2.687, "step": 32227 }, { "epoch": 2.000620770997579, "grad_norm": 0.5493615975334578, "learning_rate": 3.0163223618238834e-05, "loss": 2.7863, "step": 32228 }, { "epoch": 2.000682848097337, "grad_norm": 0.5735039186601533, "learning_rate": 3.015990853290376e-05, "loss": 2.7787, "step": 32229 }, { "epoch": 2.0007449251970946, "grad_norm": 0.5942785771294731, "learning_rate": 3.0156593551079242e-05, "loss": 2.7724, "step": 32230 }, { "epoch": 2.0008070022968525, "grad_norm": 0.5686009473730373, "learning_rate": 3.015327867278262e-05, "loss": 2.7581, "step": 32231 }, { "epoch": 2.0008690793966104, "grad_norm": 0.6010258249964157, "learning_rate": 3.0149963898031154e-05, "loss": 2.6739, "step": 32232 }, { "epoch": 2.0009311564963683, "grad_norm": 0.6442224398555185, "learning_rate": 3.014664922684215e-05, "loss": 2.7891, "step": 32233 }, { "epoch": 2.0009932335961262, "grad_norm": 0.5630856220563362, "learning_rate": 3.014333465923288e-05, "loss": 2.777, "step": 32234 }, { "epoch": 2.001055310695884, "grad_norm": 0.5782312089428931, "learning_rate": 3.0140020195220675e-05, "loss": 2.7102, "step": 32235 }, { "epoch": 2.001117387795642, "grad_norm": 0.6351711679656299, "learning_rate": 3.0136705834822787e-05, "loss": 2.7288, "step": 32236 }, { "epoch": 2.0011794648954, "grad_norm": 0.546555098564761, "learning_rate": 3.0133391578056548e-05, "loss": 2.7222, "step": 32237 }, { "epoch": 2.001241541995158, "grad_norm": 0.6501156795708615, "learning_rate": 3.0130077424939206e-05, "loss": 2.681, "step": 32238 }, { "epoch": 2.001303619094916, "grad_norm": 0.5429054607582581, "learning_rate": 3.0126763375488086e-05, "loss": 2.6415, "step": 32239 }, { "epoch": 2.0013656961946737, "grad_norm": 0.5845038868552164, "learning_rate": 3.0123449429720462e-05, "loss": 2.7108, "step": 32240 }, { "epoch": 2.0014277732944317, "grad_norm": 0.5372132079589889, "learning_rate": 3.0120135587653636e-05, "loss": 2.7445, "step": 32241 }, { "epoch": 2.0014898503941896, "grad_norm": 0.5714961793192777, "learning_rate": 3.011682184930489e-05, "loss": 2.7352, "step": 32242 }, { "epoch": 2.0015519274939475, "grad_norm": 0.5169135379377812, "learning_rate": 3.01135082146915e-05, "loss": 2.7505, "step": 32243 }, { "epoch": 2.0016140045937054, "grad_norm": 0.49615445103577177, "learning_rate": 3.011019468383077e-05, "loss": 2.6819, "step": 32244 }, { "epoch": 2.0016760816934633, "grad_norm": 0.6184468277993499, "learning_rate": 3.010688125673997e-05, "loss": 2.7359, "step": 32245 }, { "epoch": 2.0017381587932213, "grad_norm": 0.4517682745118343, "learning_rate": 3.010356793343642e-05, "loss": 2.6713, "step": 32246 }, { "epoch": 2.001800235892979, "grad_norm": 0.5284363624474786, "learning_rate": 3.010025471393737e-05, "loss": 2.6724, "step": 32247 }, { "epoch": 2.001862312992737, "grad_norm": 0.5508444188464995, "learning_rate": 3.0096941598260135e-05, "loss": 2.6492, "step": 32248 }, { "epoch": 2.001924390092495, "grad_norm": 0.5580158311843865, "learning_rate": 3.009362858642198e-05, "loss": 2.665, "step": 32249 }, { "epoch": 2.001986467192253, "grad_norm": 0.589763459052453, "learning_rate": 3.00903156784402e-05, "loss": 2.6791, "step": 32250 }, { "epoch": 2.002048544292011, "grad_norm": 0.5383425341352317, "learning_rate": 3.008700287433207e-05, "loss": 2.6617, "step": 32251 }, { "epoch": 2.0021106213917688, "grad_norm": 0.5550682480407817, "learning_rate": 3.0083690174114894e-05, "loss": 2.7271, "step": 32252 }, { "epoch": 2.0021726984915267, "grad_norm": 0.555261905535658, "learning_rate": 3.0080377577805935e-05, "loss": 2.7353, "step": 32253 }, { "epoch": 2.002234775591284, "grad_norm": 0.5565360255426615, "learning_rate": 3.0077065085422474e-05, "loss": 2.7506, "step": 32254 }, { "epoch": 2.002296852691042, "grad_norm": 0.524891981865385, "learning_rate": 3.007375269698181e-05, "loss": 2.6678, "step": 32255 }, { "epoch": 2.0023589297908, "grad_norm": 0.5198326097263014, "learning_rate": 3.0070440412501205e-05, "loss": 2.6967, "step": 32256 }, { "epoch": 2.002421006890558, "grad_norm": 0.5647130717412712, "learning_rate": 3.0067128231997966e-05, "loss": 2.6823, "step": 32257 }, { "epoch": 2.002483083990316, "grad_norm": 0.5199812754139154, "learning_rate": 3.0063816155489342e-05, "loss": 2.6733, "step": 32258 }, { "epoch": 2.0025451610900737, "grad_norm": 0.538034620105314, "learning_rate": 3.0060504182992638e-05, "loss": 2.6777, "step": 32259 }, { "epoch": 2.0026072381898317, "grad_norm": 0.5495169176147603, "learning_rate": 3.0057192314525118e-05, "loss": 2.6916, "step": 32260 }, { "epoch": 2.0026693152895896, "grad_norm": 0.6009702119934951, "learning_rate": 3.0053880550104073e-05, "loss": 2.7279, "step": 32261 }, { "epoch": 2.0027313923893475, "grad_norm": 0.547059600191331, "learning_rate": 3.0050568889746768e-05, "loss": 2.7568, "step": 32262 }, { "epoch": 2.0027934694891054, "grad_norm": 0.6119049811884105, "learning_rate": 3.004725733347048e-05, "loss": 2.6186, "step": 32263 }, { "epoch": 2.0028555465888633, "grad_norm": 0.5705921062502424, "learning_rate": 3.004394588129251e-05, "loss": 2.6707, "step": 32264 }, { "epoch": 2.0029176236886213, "grad_norm": 0.5746279964373044, "learning_rate": 3.004063453323011e-05, "loss": 2.7835, "step": 32265 }, { "epoch": 2.002979700788379, "grad_norm": 0.5550971600398207, "learning_rate": 3.0037323289300577e-05, "loss": 2.7404, "step": 32266 }, { "epoch": 2.003041777888137, "grad_norm": 0.5650633578940049, "learning_rate": 3.0034012149521173e-05, "loss": 2.7149, "step": 32267 }, { "epoch": 2.003103854987895, "grad_norm": 0.5125636921824459, "learning_rate": 3.0030701113909167e-05, "loss": 2.7261, "step": 32268 }, { "epoch": 2.003165932087653, "grad_norm": 0.5650906509916082, "learning_rate": 3.002739018248185e-05, "loss": 2.7768, "step": 32269 }, { "epoch": 2.003228009187411, "grad_norm": 0.5580927062223228, "learning_rate": 3.0024079355256472e-05, "loss": 2.6618, "step": 32270 }, { "epoch": 2.0032900862871688, "grad_norm": 0.48679246910573487, "learning_rate": 3.002076863225034e-05, "loss": 2.5991, "step": 32271 }, { "epoch": 2.0033521633869267, "grad_norm": 0.5617243133821196, "learning_rate": 3.0017458013480692e-05, "loss": 2.5892, "step": 32272 }, { "epoch": 2.0034142404866846, "grad_norm": 0.5317247457331573, "learning_rate": 3.001414749896483e-05, "loss": 2.6698, "step": 32273 }, { "epoch": 2.0034763175864425, "grad_norm": 0.4841423091025037, "learning_rate": 3.0010837088719995e-05, "loss": 2.7248, "step": 32274 }, { "epoch": 2.0035383946862004, "grad_norm": 0.5768686510358689, "learning_rate": 3.000752678276349e-05, "loss": 2.7282, "step": 32275 }, { "epoch": 2.0036004717859583, "grad_norm": 0.49664345270455706, "learning_rate": 3.0004216581112564e-05, "loss": 2.5798, "step": 32276 }, { "epoch": 2.0036625488857163, "grad_norm": 0.5573008587040524, "learning_rate": 3.0000906483784497e-05, "loss": 2.7303, "step": 32277 }, { "epoch": 2.0037246259854737, "grad_norm": 0.5777774487804226, "learning_rate": 2.9997596490796566e-05, "loss": 2.7472, "step": 32278 }, { "epoch": 2.0037867030852317, "grad_norm": 0.5417934018816531, "learning_rate": 2.9994286602166012e-05, "loss": 2.6803, "step": 32279 }, { "epoch": 2.0038487801849896, "grad_norm": 0.5986375917427493, "learning_rate": 2.9990976817910134e-05, "loss": 2.7048, "step": 32280 }, { "epoch": 2.0039108572847475, "grad_norm": 0.6604704927196302, "learning_rate": 2.9987667138046178e-05, "loss": 2.643, "step": 32281 }, { "epoch": 2.0039729343845054, "grad_norm": 0.5006272137101243, "learning_rate": 2.9984357562591426e-05, "loss": 2.7186, "step": 32282 }, { "epoch": 2.0040350114842633, "grad_norm": 0.6510776179241948, "learning_rate": 2.9981048091563134e-05, "loss": 2.7194, "step": 32283 }, { "epoch": 2.0040970885840212, "grad_norm": 0.515439202328162, "learning_rate": 2.997773872497859e-05, "loss": 2.6488, "step": 32284 }, { "epoch": 2.004159165683779, "grad_norm": 0.4868632516924844, "learning_rate": 2.9974429462855026e-05, "loss": 2.7536, "step": 32285 }, { "epoch": 2.004221242783537, "grad_norm": 0.569934902865858, "learning_rate": 2.997112030520973e-05, "loss": 2.6922, "step": 32286 }, { "epoch": 2.004283319883295, "grad_norm": 0.5303569468011168, "learning_rate": 2.9967811252059957e-05, "loss": 2.6003, "step": 32287 }, { "epoch": 2.004345396983053, "grad_norm": 0.5416523735899371, "learning_rate": 2.9964502303422987e-05, "loss": 2.6801, "step": 32288 }, { "epoch": 2.004407474082811, "grad_norm": 0.5591086835213827, "learning_rate": 2.996119345931607e-05, "loss": 2.7135, "step": 32289 }, { "epoch": 2.0044695511825688, "grad_norm": 0.5473771650746716, "learning_rate": 2.9957884719756468e-05, "loss": 2.6983, "step": 32290 }, { "epoch": 2.0045316282823267, "grad_norm": 0.6103446532812061, "learning_rate": 2.995457608476145e-05, "loss": 2.7161, "step": 32291 }, { "epoch": 2.0045937053820846, "grad_norm": 0.5733395287498736, "learning_rate": 2.995126755434826e-05, "loss": 2.7361, "step": 32292 }, { "epoch": 2.0046557824818425, "grad_norm": 0.6320668996746241, "learning_rate": 2.9947959128534197e-05, "loss": 2.6137, "step": 32293 }, { "epoch": 2.0047178595816004, "grad_norm": 0.5230976253899007, "learning_rate": 2.9944650807336482e-05, "loss": 2.767, "step": 32294 }, { "epoch": 2.0047799366813583, "grad_norm": 0.615023743434739, "learning_rate": 2.9941342590772405e-05, "loss": 2.6676, "step": 32295 }, { "epoch": 2.0048420137811163, "grad_norm": 0.548155729069687, "learning_rate": 2.993803447885919e-05, "loss": 2.659, "step": 32296 }, { "epoch": 2.004904090880874, "grad_norm": 0.5360796959418229, "learning_rate": 2.9934726471614126e-05, "loss": 2.6786, "step": 32297 }, { "epoch": 2.004966167980632, "grad_norm": 0.6303194646871351, "learning_rate": 2.993141856905448e-05, "loss": 2.6902, "step": 32298 }, { "epoch": 2.00502824508039, "grad_norm": 0.5441473101268919, "learning_rate": 2.992811077119748e-05, "loss": 2.7468, "step": 32299 }, { "epoch": 2.005090322180148, "grad_norm": 0.6133499107072814, "learning_rate": 2.9924803078060415e-05, "loss": 2.6523, "step": 32300 }, { "epoch": 2.005152399279906, "grad_norm": 0.630114679760182, "learning_rate": 2.9921495489660512e-05, "loss": 2.6649, "step": 32301 }, { "epoch": 2.0052144763796633, "grad_norm": 0.5414177126861389, "learning_rate": 2.9918188006015057e-05, "loss": 2.7158, "step": 32302 }, { "epoch": 2.0052765534794212, "grad_norm": 0.5813606290236509, "learning_rate": 2.9914880627141282e-05, "loss": 2.6668, "step": 32303 }, { "epoch": 2.005338630579179, "grad_norm": 0.5980178636606629, "learning_rate": 2.9911573353056444e-05, "loss": 2.6981, "step": 32304 }, { "epoch": 2.005400707678937, "grad_norm": 0.5064515583846289, "learning_rate": 2.990826618377782e-05, "loss": 2.801, "step": 32305 }, { "epoch": 2.005462784778695, "grad_norm": 0.6182385199422303, "learning_rate": 2.990495911932263e-05, "loss": 2.7481, "step": 32306 }, { "epoch": 2.005524861878453, "grad_norm": 0.5360652137483958, "learning_rate": 2.990165215970816e-05, "loss": 2.7221, "step": 32307 }, { "epoch": 2.005586938978211, "grad_norm": 0.5947782998217042, "learning_rate": 2.9898345304951645e-05, "loss": 2.7444, "step": 32308 }, { "epoch": 2.0056490160779687, "grad_norm": 0.5267402020807339, "learning_rate": 2.989503855507034e-05, "loss": 2.8026, "step": 32309 }, { "epoch": 2.0057110931777267, "grad_norm": 0.5901532150808626, "learning_rate": 2.98917319100815e-05, "loss": 2.6722, "step": 32310 }, { "epoch": 2.0057731702774846, "grad_norm": 0.4840986478692099, "learning_rate": 2.9888425370002387e-05, "loss": 2.6746, "step": 32311 }, { "epoch": 2.0058352473772425, "grad_norm": 0.5472683445647123, "learning_rate": 2.9885118934850225e-05, "loss": 2.6641, "step": 32312 }, { "epoch": 2.0058973244770004, "grad_norm": 0.540125614011416, "learning_rate": 2.9881812604642294e-05, "loss": 2.7372, "step": 32313 }, { "epoch": 2.0059594015767583, "grad_norm": 0.5709141471024695, "learning_rate": 2.9878506379395834e-05, "loss": 2.698, "step": 32314 }, { "epoch": 2.0060214786765163, "grad_norm": 0.509840364202351, "learning_rate": 2.987520025912807e-05, "loss": 2.8321, "step": 32315 }, { "epoch": 2.006083555776274, "grad_norm": 0.5524816036182366, "learning_rate": 2.987189424385629e-05, "loss": 2.6451, "step": 32316 }, { "epoch": 2.006145632876032, "grad_norm": 0.5251683044866077, "learning_rate": 2.9868588333597714e-05, "loss": 2.6724, "step": 32317 }, { "epoch": 2.00620770997579, "grad_norm": 0.5492879766948779, "learning_rate": 2.986528252836961e-05, "loss": 2.7605, "step": 32318 }, { "epoch": 2.006269787075548, "grad_norm": 0.5406477465833185, "learning_rate": 2.98619768281892e-05, "loss": 2.6485, "step": 32319 }, { "epoch": 2.006331864175306, "grad_norm": 0.5078271130146549, "learning_rate": 2.9858671233073753e-05, "loss": 2.6766, "step": 32320 }, { "epoch": 2.0063939412750638, "grad_norm": 0.5875270390882883, "learning_rate": 2.985536574304051e-05, "loss": 2.6903, "step": 32321 }, { "epoch": 2.0064560183748217, "grad_norm": 0.5578981082736495, "learning_rate": 2.9852060358106715e-05, "loss": 2.6947, "step": 32322 }, { "epoch": 2.0065180954745796, "grad_norm": 0.5461393585587656, "learning_rate": 2.9848755078289603e-05, "loss": 2.7186, "step": 32323 }, { "epoch": 2.0065801725743375, "grad_norm": 0.5481053544789868, "learning_rate": 2.984544990360644e-05, "loss": 2.7265, "step": 32324 }, { "epoch": 2.0066422496740954, "grad_norm": 0.451202639236873, "learning_rate": 2.9842144834074448e-05, "loss": 2.6889, "step": 32325 }, { "epoch": 2.006704326773853, "grad_norm": 0.5731020598568073, "learning_rate": 2.983883986971089e-05, "loss": 2.7194, "step": 32326 }, { "epoch": 2.006766403873611, "grad_norm": 0.5068594866447075, "learning_rate": 2.983553501053299e-05, "loss": 2.6891, "step": 32327 }, { "epoch": 2.0068284809733687, "grad_norm": 0.5591622507439258, "learning_rate": 2.983223025655799e-05, "loss": 2.727, "step": 32328 }, { "epoch": 2.0068905580731267, "grad_norm": 0.5107774594674788, "learning_rate": 2.982892560780315e-05, "loss": 2.6516, "step": 32329 }, { "epoch": 2.0069526351728846, "grad_norm": 0.5562263151548393, "learning_rate": 2.9825621064285703e-05, "loss": 2.7233, "step": 32330 }, { "epoch": 2.0070147122726425, "grad_norm": 0.5559745953316757, "learning_rate": 2.9822316626022882e-05, "loss": 2.7248, "step": 32331 }, { "epoch": 2.0070767893724004, "grad_norm": 0.5470519188400547, "learning_rate": 2.9819012293031946e-05, "loss": 2.687, "step": 32332 }, { "epoch": 2.0071388664721583, "grad_norm": 0.5409774215926103, "learning_rate": 2.9815708065330118e-05, "loss": 2.664, "step": 32333 }, { "epoch": 2.0072009435719163, "grad_norm": 0.5506355146947773, "learning_rate": 2.9812403942934642e-05, "loss": 2.6722, "step": 32334 }, { "epoch": 2.007263020671674, "grad_norm": 0.4698719109080772, "learning_rate": 2.9809099925862742e-05, "loss": 2.6361, "step": 32335 }, { "epoch": 2.007325097771432, "grad_norm": 0.5556556687617761, "learning_rate": 2.980579601413169e-05, "loss": 2.6747, "step": 32336 }, { "epoch": 2.00738717487119, "grad_norm": 0.5116171699631221, "learning_rate": 2.9802492207758682e-05, "loss": 2.6416, "step": 32337 }, { "epoch": 2.007449251970948, "grad_norm": 0.55328500771721, "learning_rate": 2.9799188506760993e-05, "loss": 2.6609, "step": 32338 }, { "epoch": 2.007511329070706, "grad_norm": 0.5503942333831437, "learning_rate": 2.9795884911155835e-05, "loss": 2.7145, "step": 32339 }, { "epoch": 2.0075734061704638, "grad_norm": 0.5040703439782133, "learning_rate": 2.9792581420960448e-05, "loss": 2.728, "step": 32340 }, { "epoch": 2.0076354832702217, "grad_norm": 0.5612709605629572, "learning_rate": 2.9789278036192074e-05, "loss": 2.6946, "step": 32341 }, { "epoch": 2.0076975603699796, "grad_norm": 0.4943036198837902, "learning_rate": 2.9785974756867935e-05, "loss": 2.6917, "step": 32342 }, { "epoch": 2.0077596374697375, "grad_norm": 0.562513642675473, "learning_rate": 2.9782671583005282e-05, "loss": 2.6881, "step": 32343 }, { "epoch": 2.0078217145694954, "grad_norm": 0.5095989788000274, "learning_rate": 2.9779368514621332e-05, "loss": 2.6752, "step": 32344 }, { "epoch": 2.0078837916692533, "grad_norm": 0.5199721337151064, "learning_rate": 2.9776065551733333e-05, "loss": 2.7048, "step": 32345 }, { "epoch": 2.0079458687690113, "grad_norm": 0.5006738235059045, "learning_rate": 2.9772762694358498e-05, "loss": 2.6551, "step": 32346 }, { "epoch": 2.008007945868769, "grad_norm": 0.5832999821620227, "learning_rate": 2.9769459942514077e-05, "loss": 2.7041, "step": 32347 }, { "epoch": 2.008070022968527, "grad_norm": 0.5177389506823836, "learning_rate": 2.976615729621729e-05, "loss": 2.784, "step": 32348 }, { "epoch": 2.008132100068285, "grad_norm": 0.5723443091137503, "learning_rate": 2.9762854755485382e-05, "loss": 2.6743, "step": 32349 }, { "epoch": 2.0081941771680425, "grad_norm": 0.5434665730505, "learning_rate": 2.9759552320335566e-05, "loss": 2.6726, "step": 32350 }, { "epoch": 2.0082562542678004, "grad_norm": 0.5024954689827464, "learning_rate": 2.9756249990785086e-05, "loss": 2.6959, "step": 32351 }, { "epoch": 2.0083183313675583, "grad_norm": 0.5288622997060043, "learning_rate": 2.9752947766851158e-05, "loss": 2.7852, "step": 32352 }, { "epoch": 2.0083804084673162, "grad_norm": 0.5311034785897839, "learning_rate": 2.9749645648551017e-05, "loss": 2.7224, "step": 32353 }, { "epoch": 2.008442485567074, "grad_norm": 0.5323406412197892, "learning_rate": 2.9746343635901895e-05, "loss": 2.7085, "step": 32354 }, { "epoch": 2.008504562666832, "grad_norm": 0.5691551588895549, "learning_rate": 2.9743041728921005e-05, "loss": 2.7203, "step": 32355 }, { "epoch": 2.00856663976659, "grad_norm": 0.5349361774665786, "learning_rate": 2.9739739927625597e-05, "loss": 2.7368, "step": 32356 }, { "epoch": 2.008628716866348, "grad_norm": 0.5435742631376078, "learning_rate": 2.973643823203287e-05, "loss": 2.6682, "step": 32357 }, { "epoch": 2.008690793966106, "grad_norm": 0.4970808646350492, "learning_rate": 2.9733136642160075e-05, "loss": 2.7076, "step": 32358 }, { "epoch": 2.0087528710658638, "grad_norm": 0.503662050203735, "learning_rate": 2.9729835158024422e-05, "loss": 2.6089, "step": 32359 }, { "epoch": 2.0088149481656217, "grad_norm": 0.5673186746874629, "learning_rate": 2.972653377964314e-05, "loss": 2.7126, "step": 32360 }, { "epoch": 2.0088770252653796, "grad_norm": 0.5067533157420823, "learning_rate": 2.972323250703344e-05, "loss": 2.7184, "step": 32361 }, { "epoch": 2.0089391023651375, "grad_norm": 0.5187936293397608, "learning_rate": 2.971993134021257e-05, "loss": 2.6931, "step": 32362 }, { "epoch": 2.0090011794648954, "grad_norm": 0.47444193833905474, "learning_rate": 2.9716630279197745e-05, "loss": 2.7258, "step": 32363 }, { "epoch": 2.0090632565646533, "grad_norm": 0.5139476775430769, "learning_rate": 2.971332932400619e-05, "loss": 2.722, "step": 32364 }, { "epoch": 2.0091253336644113, "grad_norm": 0.5327318784960414, "learning_rate": 2.9710028474655104e-05, "loss": 2.6718, "step": 32365 }, { "epoch": 2.009187410764169, "grad_norm": 0.48074311277164566, "learning_rate": 2.970672773116174e-05, "loss": 2.7006, "step": 32366 }, { "epoch": 2.009249487863927, "grad_norm": 0.5720185667672606, "learning_rate": 2.970342709354329e-05, "loss": 2.7767, "step": 32367 }, { "epoch": 2.009311564963685, "grad_norm": 0.48592022561527787, "learning_rate": 2.9700126561817e-05, "loss": 2.7082, "step": 32368 }, { "epoch": 2.009373642063443, "grad_norm": 0.5029149025530822, "learning_rate": 2.9696826136000065e-05, "loss": 2.7376, "step": 32369 }, { "epoch": 2.009435719163201, "grad_norm": 0.5921850914012874, "learning_rate": 2.969352581610973e-05, "loss": 2.6652, "step": 32370 }, { "epoch": 2.0094977962629588, "grad_norm": 0.4773911791275, "learning_rate": 2.969022560216319e-05, "loss": 2.7316, "step": 32371 }, { "epoch": 2.0095598733627167, "grad_norm": 0.5521434415328783, "learning_rate": 2.9686925494177687e-05, "loss": 2.6924, "step": 32372 }, { "epoch": 2.0096219504624746, "grad_norm": 0.5710334662519323, "learning_rate": 2.9683625492170408e-05, "loss": 2.7592, "step": 32373 }, { "epoch": 2.009684027562232, "grad_norm": 0.512748671888457, "learning_rate": 2.96803255961586e-05, "loss": 2.6721, "step": 32374 }, { "epoch": 2.00974610466199, "grad_norm": 0.5633169440654228, "learning_rate": 2.9677025806159453e-05, "loss": 2.7203, "step": 32375 }, { "epoch": 2.009808181761748, "grad_norm": 0.5275052141320579, "learning_rate": 2.9673726122190216e-05, "loss": 2.773, "step": 32376 }, { "epoch": 2.009870258861506, "grad_norm": 0.5310558076364784, "learning_rate": 2.9670426544268072e-05, "loss": 2.6969, "step": 32377 }, { "epoch": 2.0099323359612637, "grad_norm": 0.5367069193106668, "learning_rate": 2.9667127072410238e-05, "loss": 2.7376, "step": 32378 }, { "epoch": 2.0099944130610217, "grad_norm": 0.547196266631307, "learning_rate": 2.9663827706633963e-05, "loss": 2.7063, "step": 32379 }, { "epoch": 2.0100564901607796, "grad_norm": 0.5087939534894923, "learning_rate": 2.9660528446956408e-05, "loss": 2.6916, "step": 32380 }, { "epoch": 2.0101185672605375, "grad_norm": 0.5270484097285879, "learning_rate": 2.9657229293394832e-05, "loss": 2.7341, "step": 32381 }, { "epoch": 2.0101806443602954, "grad_norm": 0.5337726499767956, "learning_rate": 2.965393024596642e-05, "loss": 2.69, "step": 32382 }, { "epoch": 2.0102427214600533, "grad_norm": 0.5474002495565462, "learning_rate": 2.9650631304688402e-05, "loss": 2.7278, "step": 32383 }, { "epoch": 2.0103047985598113, "grad_norm": 0.4811555714050887, "learning_rate": 2.964733246957797e-05, "loss": 2.6889, "step": 32384 }, { "epoch": 2.010366875659569, "grad_norm": 0.5067958926218837, "learning_rate": 2.9644033740652355e-05, "loss": 2.7077, "step": 32385 }, { "epoch": 2.010428952759327, "grad_norm": 0.4943743367920103, "learning_rate": 2.9640735117928746e-05, "loss": 2.5849, "step": 32386 }, { "epoch": 2.010491029859085, "grad_norm": 0.5452102936984093, "learning_rate": 2.9637436601424377e-05, "loss": 2.7674, "step": 32387 }, { "epoch": 2.010553106958843, "grad_norm": 0.5176202720899667, "learning_rate": 2.9634138191156446e-05, "loss": 2.6818, "step": 32388 }, { "epoch": 2.010615184058601, "grad_norm": 0.5003877864521882, "learning_rate": 2.9630839887142147e-05, "loss": 2.7142, "step": 32389 }, { "epoch": 2.0106772611583588, "grad_norm": 0.5818228636595261, "learning_rate": 2.962754168939871e-05, "loss": 2.7959, "step": 32390 }, { "epoch": 2.0107393382581167, "grad_norm": 0.5361995690202813, "learning_rate": 2.9624243597943323e-05, "loss": 2.7024, "step": 32391 }, { "epoch": 2.0108014153578746, "grad_norm": 0.5789588636615746, "learning_rate": 2.9620945612793215e-05, "loss": 2.7967, "step": 32392 }, { "epoch": 2.0108634924576325, "grad_norm": 0.5788776292817387, "learning_rate": 2.9617647733965576e-05, "loss": 2.678, "step": 32393 }, { "epoch": 2.0109255695573904, "grad_norm": 0.5555422201807857, "learning_rate": 2.9614349961477604e-05, "loss": 2.6476, "step": 32394 }, { "epoch": 2.0109876466571484, "grad_norm": 0.587490135475292, "learning_rate": 2.9611052295346532e-05, "loss": 2.6141, "step": 32395 }, { "epoch": 2.0110497237569063, "grad_norm": 0.5749417638312676, "learning_rate": 2.9607754735589543e-05, "loss": 2.6884, "step": 32396 }, { "epoch": 2.011111800856664, "grad_norm": 0.5573391998079609, "learning_rate": 2.9604457282223863e-05, "loss": 2.6861, "step": 32397 }, { "epoch": 2.0111738779564217, "grad_norm": 0.5798151980920493, "learning_rate": 2.9601159935266663e-05, "loss": 2.7299, "step": 32398 }, { "epoch": 2.0112359550561796, "grad_norm": 0.5197009332954199, "learning_rate": 2.959786269473518e-05, "loss": 2.7224, "step": 32399 }, { "epoch": 2.0112980321559375, "grad_norm": 0.5328401440919066, "learning_rate": 2.9594565560646582e-05, "loss": 2.7583, "step": 32400 }, { "epoch": 2.0113601092556954, "grad_norm": 0.5466663234940281, "learning_rate": 2.9591268533018103e-05, "loss": 2.6163, "step": 32401 }, { "epoch": 2.0114221863554533, "grad_norm": 0.6304456694773186, "learning_rate": 2.9587971611866934e-05, "loss": 2.7617, "step": 32402 }, { "epoch": 2.0114842634552113, "grad_norm": 0.4946982814266997, "learning_rate": 2.9584674797210253e-05, "loss": 2.6823, "step": 32403 }, { "epoch": 2.011546340554969, "grad_norm": 0.5872691979451183, "learning_rate": 2.9581378089065305e-05, "loss": 2.6734, "step": 32404 }, { "epoch": 2.011608417654727, "grad_norm": 0.555830007220968, "learning_rate": 2.957808148744924e-05, "loss": 2.5949, "step": 32405 }, { "epoch": 2.011670494754485, "grad_norm": 0.5218378362157788, "learning_rate": 2.95747849923793e-05, "loss": 2.6898, "step": 32406 }, { "epoch": 2.011732571854243, "grad_norm": 0.5823801774079145, "learning_rate": 2.9571488603872648e-05, "loss": 2.6828, "step": 32407 }, { "epoch": 2.011794648954001, "grad_norm": 0.5785895105299851, "learning_rate": 2.9568192321946514e-05, "loss": 2.7338, "step": 32408 }, { "epoch": 2.0118567260537588, "grad_norm": 0.5520901109771238, "learning_rate": 2.956489614661807e-05, "loss": 2.6226, "step": 32409 }, { "epoch": 2.0119188031535167, "grad_norm": 0.6756708132079148, "learning_rate": 2.9561600077904532e-05, "loss": 2.7602, "step": 32410 }, { "epoch": 2.0119808802532746, "grad_norm": 0.5762348722511959, "learning_rate": 2.9558304115823077e-05, "loss": 2.6234, "step": 32411 }, { "epoch": 2.0120429573530325, "grad_norm": 0.6091755026930118, "learning_rate": 2.9555008260390928e-05, "loss": 2.7461, "step": 32412 }, { "epoch": 2.0121050344527904, "grad_norm": 0.6128474359508954, "learning_rate": 2.9551712511625264e-05, "loss": 2.6709, "step": 32413 }, { "epoch": 2.0121671115525483, "grad_norm": 0.6002259951394684, "learning_rate": 2.954841686954326e-05, "loss": 2.6761, "step": 32414 }, { "epoch": 2.0122291886523063, "grad_norm": 0.5758767616836235, "learning_rate": 2.9545121334162145e-05, "loss": 2.7286, "step": 32415 }, { "epoch": 2.012291265752064, "grad_norm": 0.6095967989529646, "learning_rate": 2.954182590549909e-05, "loss": 2.7212, "step": 32416 }, { "epoch": 2.012353342851822, "grad_norm": 0.6022415044667524, "learning_rate": 2.95385305835713e-05, "loss": 2.6245, "step": 32417 }, { "epoch": 2.01241541995158, "grad_norm": 0.5803570060939418, "learning_rate": 2.9535235368395964e-05, "loss": 2.6724, "step": 32418 }, { "epoch": 2.012477497051338, "grad_norm": 0.6238568341761913, "learning_rate": 2.9531940259990275e-05, "loss": 2.6653, "step": 32419 }, { "epoch": 2.012539574151096, "grad_norm": 0.6572428867075593, "learning_rate": 2.9528645258371412e-05, "loss": 2.7514, "step": 32420 }, { "epoch": 2.0126016512508538, "grad_norm": 0.6371121507650843, "learning_rate": 2.9525350363556594e-05, "loss": 2.6771, "step": 32421 }, { "epoch": 2.0126637283506112, "grad_norm": 0.5307815674384164, "learning_rate": 2.9522055575562967e-05, "loss": 2.6724, "step": 32422 }, { "epoch": 2.012725805450369, "grad_norm": 0.6458338434976183, "learning_rate": 2.951876089440777e-05, "loss": 2.6965, "step": 32423 }, { "epoch": 2.012787882550127, "grad_norm": 0.505911364148311, "learning_rate": 2.951546632010817e-05, "loss": 2.5814, "step": 32424 }, { "epoch": 2.012849959649885, "grad_norm": 0.5324252816285086, "learning_rate": 2.951217185268134e-05, "loss": 2.6073, "step": 32425 }, { "epoch": 2.012912036749643, "grad_norm": 0.5984681791404706, "learning_rate": 2.9508877492144498e-05, "loss": 2.6268, "step": 32426 }, { "epoch": 2.012974113849401, "grad_norm": 0.5037034092222602, "learning_rate": 2.9505583238514796e-05, "loss": 2.6894, "step": 32427 }, { "epoch": 2.0130361909491588, "grad_norm": 0.5697891760660677, "learning_rate": 2.9502289091809444e-05, "loss": 2.7244, "step": 32428 }, { "epoch": 2.0130982680489167, "grad_norm": 0.528002320662008, "learning_rate": 2.949899505204564e-05, "loss": 2.6665, "step": 32429 }, { "epoch": 2.0131603451486746, "grad_norm": 0.5435459718817619, "learning_rate": 2.9495701119240547e-05, "loss": 2.6307, "step": 32430 }, { "epoch": 2.0132224222484325, "grad_norm": 0.5518106890972848, "learning_rate": 2.9492407293411374e-05, "loss": 2.7013, "step": 32431 }, { "epoch": 2.0132844993481904, "grad_norm": 0.5255047131861487, "learning_rate": 2.9489113574575273e-05, "loss": 2.758, "step": 32432 }, { "epoch": 2.0133465764479483, "grad_norm": 0.5758292779857893, "learning_rate": 2.9485819962749462e-05, "loss": 2.7053, "step": 32433 }, { "epoch": 2.0134086535477063, "grad_norm": 0.5742162587637286, "learning_rate": 2.9482526457951086e-05, "loss": 2.7115, "step": 32434 }, { "epoch": 2.013470730647464, "grad_norm": 0.6077294996610124, "learning_rate": 2.9479233060197375e-05, "loss": 2.7578, "step": 32435 }, { "epoch": 2.013532807747222, "grad_norm": 0.5558884742395627, "learning_rate": 2.947593976950547e-05, "loss": 2.6775, "step": 32436 }, { "epoch": 2.01359488484698, "grad_norm": 0.6224176943037558, "learning_rate": 2.9472646585892577e-05, "loss": 2.7534, "step": 32437 }, { "epoch": 2.013656961946738, "grad_norm": 0.5945814735563509, "learning_rate": 2.9469353509375874e-05, "loss": 2.6962, "step": 32438 }, { "epoch": 2.013719039046496, "grad_norm": 0.593369964645747, "learning_rate": 2.9466060539972523e-05, "loss": 2.7077, "step": 32439 }, { "epoch": 2.0137811161462538, "grad_norm": 0.5182403933844173, "learning_rate": 2.9462767677699733e-05, "loss": 2.6971, "step": 32440 }, { "epoch": 2.0138431932460117, "grad_norm": 0.5371807259070934, "learning_rate": 2.945947492257466e-05, "loss": 2.6291, "step": 32441 }, { "epoch": 2.0139052703457696, "grad_norm": 0.5636123414300623, "learning_rate": 2.9456182274614497e-05, "loss": 2.678, "step": 32442 }, { "epoch": 2.0139673474455275, "grad_norm": 0.5223614464266009, "learning_rate": 2.945288973383642e-05, "loss": 2.6594, "step": 32443 }, { "epoch": 2.0140294245452854, "grad_norm": 0.5497155667806415, "learning_rate": 2.9449597300257603e-05, "loss": 2.6806, "step": 32444 }, { "epoch": 2.0140915016450434, "grad_norm": 0.5408141659669423, "learning_rate": 2.9446304973895224e-05, "loss": 2.6943, "step": 32445 }, { "epoch": 2.014153578744801, "grad_norm": 0.4853443782282357, "learning_rate": 2.9443012754766463e-05, "loss": 2.6174, "step": 32446 }, { "epoch": 2.0142156558445587, "grad_norm": 0.5641387803338488, "learning_rate": 2.943972064288849e-05, "loss": 2.5919, "step": 32447 }, { "epoch": 2.0142777329443167, "grad_norm": 0.5061759400060994, "learning_rate": 2.94364286382785e-05, "loss": 2.6867, "step": 32448 }, { "epoch": 2.0143398100440746, "grad_norm": 0.5676984844461692, "learning_rate": 2.9433136740953647e-05, "loss": 2.7615, "step": 32449 }, { "epoch": 2.0144018871438325, "grad_norm": 0.5509941014443179, "learning_rate": 2.9429844950931107e-05, "loss": 2.6504, "step": 32450 }, { "epoch": 2.0144639642435904, "grad_norm": 0.5115336891594835, "learning_rate": 2.9426553268228062e-05, "loss": 2.7035, "step": 32451 }, { "epoch": 2.0145260413433483, "grad_norm": 0.5226301543103187, "learning_rate": 2.9423261692861682e-05, "loss": 2.6487, "step": 32452 }, { "epoch": 2.0145881184431063, "grad_norm": 0.5677513462069167, "learning_rate": 2.941997022484915e-05, "loss": 2.7786, "step": 32453 }, { "epoch": 2.014650195542864, "grad_norm": 0.5189008114376696, "learning_rate": 2.9416678864207613e-05, "loss": 2.6728, "step": 32454 }, { "epoch": 2.014712272642622, "grad_norm": 0.5554985510535103, "learning_rate": 2.9413387610954273e-05, "loss": 2.7296, "step": 32455 }, { "epoch": 2.01477434974238, "grad_norm": 0.5494820442546204, "learning_rate": 2.941009646510628e-05, "loss": 2.7151, "step": 32456 }, { "epoch": 2.014836426842138, "grad_norm": 0.5703705536023012, "learning_rate": 2.9406805426680817e-05, "loss": 2.6311, "step": 32457 }, { "epoch": 2.014898503941896, "grad_norm": 0.566081223461132, "learning_rate": 2.940351449569504e-05, "loss": 2.673, "step": 32458 }, { "epoch": 2.0149605810416538, "grad_norm": 0.5820117665556475, "learning_rate": 2.9400223672166122e-05, "loss": 2.6582, "step": 32459 }, { "epoch": 2.0150226581414117, "grad_norm": 0.5897412242584142, "learning_rate": 2.9396932956111255e-05, "loss": 2.6859, "step": 32460 }, { "epoch": 2.0150847352411696, "grad_norm": 0.5493449699604879, "learning_rate": 2.939364234754758e-05, "loss": 2.6789, "step": 32461 }, { "epoch": 2.0151468123409275, "grad_norm": 0.6802321696133414, "learning_rate": 2.9390351846492288e-05, "loss": 2.7196, "step": 32462 }, { "epoch": 2.0152088894406854, "grad_norm": 0.5224872903327555, "learning_rate": 2.938706145296253e-05, "loss": 2.7628, "step": 32463 }, { "epoch": 2.0152709665404434, "grad_norm": 0.637991002741366, "learning_rate": 2.9383771166975465e-05, "loss": 2.6769, "step": 32464 }, { "epoch": 2.0153330436402013, "grad_norm": 0.5656344976734244, "learning_rate": 2.9380480988548286e-05, "loss": 2.659, "step": 32465 }, { "epoch": 2.015395120739959, "grad_norm": 0.58674250934307, "learning_rate": 2.9377190917698128e-05, "loss": 2.7162, "step": 32466 }, { "epoch": 2.015457197839717, "grad_norm": 0.5986851331737904, "learning_rate": 2.9373900954442187e-05, "loss": 2.7488, "step": 32467 }, { "epoch": 2.015519274939475, "grad_norm": 0.5700508980864333, "learning_rate": 2.9370611098797597e-05, "loss": 2.6657, "step": 32468 }, { "epoch": 2.015581352039233, "grad_norm": 0.5106225539187391, "learning_rate": 2.936732135078155e-05, "loss": 2.713, "step": 32469 }, { "epoch": 2.0156434291389904, "grad_norm": 0.5810381205785556, "learning_rate": 2.936403171041118e-05, "loss": 2.7117, "step": 32470 }, { "epoch": 2.0157055062387483, "grad_norm": 0.5937228196598595, "learning_rate": 2.936074217770368e-05, "loss": 2.7816, "step": 32471 }, { "epoch": 2.0157675833385063, "grad_norm": 0.5767262006726392, "learning_rate": 2.9357452752676194e-05, "loss": 2.7058, "step": 32472 }, { "epoch": 2.015829660438264, "grad_norm": 0.5631040402829239, "learning_rate": 2.9354163435345895e-05, "loss": 2.7451, "step": 32473 }, { "epoch": 2.015891737538022, "grad_norm": 0.5312803484158091, "learning_rate": 2.9350874225729936e-05, "loss": 2.7549, "step": 32474 }, { "epoch": 2.01595381463778, "grad_norm": 0.5528652723622233, "learning_rate": 2.9347585123845465e-05, "loss": 2.7656, "step": 32475 }, { "epoch": 2.016015891737538, "grad_norm": 0.5445019050473742, "learning_rate": 2.934429612970967e-05, "loss": 2.6959, "step": 32476 }, { "epoch": 2.016077968837296, "grad_norm": 0.5274635461441732, "learning_rate": 2.9341007243339692e-05, "loss": 2.6714, "step": 32477 }, { "epoch": 2.0161400459370538, "grad_norm": 0.5270064511474315, "learning_rate": 2.9337718464752695e-05, "loss": 2.7035, "step": 32478 }, { "epoch": 2.0162021230368117, "grad_norm": 0.5249991394217568, "learning_rate": 2.933442979396583e-05, "loss": 2.7157, "step": 32479 }, { "epoch": 2.0162642001365696, "grad_norm": 0.532879769785556, "learning_rate": 2.9331141230996273e-05, "loss": 2.781, "step": 32480 }, { "epoch": 2.0163262772363275, "grad_norm": 0.5337927216935873, "learning_rate": 2.9327852775861152e-05, "loss": 2.7282, "step": 32481 }, { "epoch": 2.0163883543360854, "grad_norm": 0.5197587626508922, "learning_rate": 2.932456442857766e-05, "loss": 2.6929, "step": 32482 }, { "epoch": 2.0164504314358433, "grad_norm": 0.5519343541025109, "learning_rate": 2.9321276189162926e-05, "loss": 2.7385, "step": 32483 }, { "epoch": 2.0165125085356013, "grad_norm": 0.5436789992444407, "learning_rate": 2.9317988057634117e-05, "loss": 2.7336, "step": 32484 }, { "epoch": 2.016574585635359, "grad_norm": 0.524002681041003, "learning_rate": 2.9314700034008392e-05, "loss": 2.6701, "step": 32485 }, { "epoch": 2.016636662735117, "grad_norm": 0.4919809331814418, "learning_rate": 2.9311412118302884e-05, "loss": 2.6415, "step": 32486 }, { "epoch": 2.016698739834875, "grad_norm": 0.5420097829351084, "learning_rate": 2.9308124310534774e-05, "loss": 2.6328, "step": 32487 }, { "epoch": 2.016760816934633, "grad_norm": 0.4919285168550309, "learning_rate": 2.9304836610721187e-05, "loss": 2.6308, "step": 32488 }, { "epoch": 2.016822894034391, "grad_norm": 0.5108351613717822, "learning_rate": 2.9301549018879305e-05, "loss": 2.6494, "step": 32489 }, { "epoch": 2.0168849711341488, "grad_norm": 0.5340761018006442, "learning_rate": 2.929826153502626e-05, "loss": 2.5724, "step": 32490 }, { "epoch": 2.0169470482339067, "grad_norm": 0.5430078012158943, "learning_rate": 2.9294974159179218e-05, "loss": 2.6808, "step": 32491 }, { "epoch": 2.0170091253336646, "grad_norm": 0.5593291556896077, "learning_rate": 2.9291686891355296e-05, "loss": 2.7386, "step": 32492 }, { "epoch": 2.0170712024334225, "grad_norm": 0.5005390122916772, "learning_rate": 2.928839973157169e-05, "loss": 2.7861, "step": 32493 }, { "epoch": 2.01713327953318, "grad_norm": 0.5459143511750661, "learning_rate": 2.9285112679845535e-05, "loss": 2.6135, "step": 32494 }, { "epoch": 2.017195356632938, "grad_norm": 0.5268815735014332, "learning_rate": 2.928182573619397e-05, "loss": 2.7683, "step": 32495 }, { "epoch": 2.017257433732696, "grad_norm": 0.5107499481253427, "learning_rate": 2.9278538900634156e-05, "loss": 2.7087, "step": 32496 }, { "epoch": 2.0173195108324538, "grad_norm": 0.5409311717526546, "learning_rate": 2.927525217318322e-05, "loss": 2.7579, "step": 32497 }, { "epoch": 2.0173815879322117, "grad_norm": 0.5307329410447468, "learning_rate": 2.927196555385834e-05, "loss": 2.671, "step": 32498 }, { "epoch": 2.0174436650319696, "grad_norm": 0.5457449545423922, "learning_rate": 2.926867904267665e-05, "loss": 2.7386, "step": 32499 }, { "epoch": 2.0175057421317275, "grad_norm": 0.5521784976296638, "learning_rate": 2.926539263965528e-05, "loss": 2.7028, "step": 32500 }, { "epoch": 2.0175678192314854, "grad_norm": 0.5243433046821601, "learning_rate": 2.92621063448114e-05, "loss": 2.6563, "step": 32501 }, { "epoch": 2.0176298963312433, "grad_norm": 0.5188272051062784, "learning_rate": 2.9258820158162135e-05, "loss": 2.7355, "step": 32502 }, { "epoch": 2.0176919734310013, "grad_norm": 0.5376831873625784, "learning_rate": 2.9255534079724656e-05, "loss": 2.7337, "step": 32503 }, { "epoch": 2.017754050530759, "grad_norm": 0.5012814967564654, "learning_rate": 2.9252248109516077e-05, "loss": 2.651, "step": 32504 }, { "epoch": 2.017816127630517, "grad_norm": 0.4948451180615318, "learning_rate": 2.9248962247553568e-05, "loss": 2.6785, "step": 32505 }, { "epoch": 2.017878204730275, "grad_norm": 0.5339204762533611, "learning_rate": 2.924567649385425e-05, "loss": 2.6377, "step": 32506 }, { "epoch": 2.017940281830033, "grad_norm": 0.5650603779880682, "learning_rate": 2.9242390848435287e-05, "loss": 2.6504, "step": 32507 }, { "epoch": 2.018002358929791, "grad_norm": 0.506811093583286, "learning_rate": 2.9239105311313797e-05, "loss": 2.7089, "step": 32508 }, { "epoch": 2.0180644360295488, "grad_norm": 0.6032853165464521, "learning_rate": 2.9235819882506956e-05, "loss": 2.7369, "step": 32509 }, { "epoch": 2.0181265131293067, "grad_norm": 0.5464378517094771, "learning_rate": 2.9232534562031864e-05, "loss": 2.7219, "step": 32510 }, { "epoch": 2.0181885902290646, "grad_norm": 0.5181897849705581, "learning_rate": 2.9229249349905684e-05, "loss": 2.7547, "step": 32511 }, { "epoch": 2.0182506673288225, "grad_norm": 0.545659540138399, "learning_rate": 2.922596424614556e-05, "loss": 2.7015, "step": 32512 }, { "epoch": 2.0183127444285804, "grad_norm": 0.5716094854615595, "learning_rate": 2.9222679250768613e-05, "loss": 2.6901, "step": 32513 }, { "epoch": 2.0183748215283384, "grad_norm": 0.4901566462808551, "learning_rate": 2.9219394363792e-05, "loss": 2.6325, "step": 32514 }, { "epoch": 2.0184368986280963, "grad_norm": 0.621683476538681, "learning_rate": 2.9216109585232842e-05, "loss": 2.7041, "step": 32515 }, { "epoch": 2.018498975727854, "grad_norm": 0.5556799646659756, "learning_rate": 2.9212824915108296e-05, "loss": 2.6937, "step": 32516 }, { "epoch": 2.018561052827612, "grad_norm": 0.5712589092937184, "learning_rate": 2.9209540353435483e-05, "loss": 2.7517, "step": 32517 }, { "epoch": 2.0186231299273696, "grad_norm": 0.5512125477427029, "learning_rate": 2.9206255900231548e-05, "loss": 2.7514, "step": 32518 }, { "epoch": 2.0186852070271275, "grad_norm": 0.5643258132707666, "learning_rate": 2.920297155551362e-05, "loss": 2.683, "step": 32519 }, { "epoch": 2.0187472841268854, "grad_norm": 0.5259431677528206, "learning_rate": 2.9199687319298846e-05, "loss": 2.7062, "step": 32520 }, { "epoch": 2.0188093612266433, "grad_norm": 0.5710517847008385, "learning_rate": 2.9196403191604337e-05, "loss": 2.6696, "step": 32521 }, { "epoch": 2.0188714383264013, "grad_norm": 0.5106984515455535, "learning_rate": 2.919311917244726e-05, "loss": 2.709, "step": 32522 }, { "epoch": 2.018933515426159, "grad_norm": 0.5613790044376282, "learning_rate": 2.9189835261844734e-05, "loss": 2.6754, "step": 32523 }, { "epoch": 2.018995592525917, "grad_norm": 0.5963255085507284, "learning_rate": 2.9186551459813884e-05, "loss": 2.7341, "step": 32524 }, { "epoch": 2.019057669625675, "grad_norm": 0.5216397686149008, "learning_rate": 2.9183267766371825e-05, "loss": 2.6687, "step": 32525 }, { "epoch": 2.019119746725433, "grad_norm": 0.6410948030898733, "learning_rate": 2.9179984181535713e-05, "loss": 2.772, "step": 32526 }, { "epoch": 2.019181823825191, "grad_norm": 0.5156418768692806, "learning_rate": 2.91767007053227e-05, "loss": 2.7091, "step": 32527 }, { "epoch": 2.0192439009249488, "grad_norm": 0.5985998545997308, "learning_rate": 2.91734173377499e-05, "loss": 2.6229, "step": 32528 }, { "epoch": 2.0193059780247067, "grad_norm": 0.5209264735546343, "learning_rate": 2.9170134078834427e-05, "loss": 2.6921, "step": 32529 }, { "epoch": 2.0193680551244646, "grad_norm": 0.579230128614623, "learning_rate": 2.9166850928593404e-05, "loss": 2.6001, "step": 32530 }, { "epoch": 2.0194301322242225, "grad_norm": 0.525590170932194, "learning_rate": 2.9163567887043998e-05, "loss": 2.7049, "step": 32531 }, { "epoch": 2.0194922093239804, "grad_norm": 0.564788771363589, "learning_rate": 2.9160284954203322e-05, "loss": 2.6246, "step": 32532 }, { "epoch": 2.0195542864237384, "grad_norm": 0.5620937469833628, "learning_rate": 2.9157002130088497e-05, "loss": 2.627, "step": 32533 }, { "epoch": 2.0196163635234963, "grad_norm": 0.5385071755093218, "learning_rate": 2.9153719414716625e-05, "loss": 2.794, "step": 32534 }, { "epoch": 2.019678440623254, "grad_norm": 0.564345009557175, "learning_rate": 2.915043680810488e-05, "loss": 2.7079, "step": 32535 }, { "epoch": 2.019740517723012, "grad_norm": 0.5845599193012327, "learning_rate": 2.9147154310270376e-05, "loss": 2.6991, "step": 32536 }, { "epoch": 2.01980259482277, "grad_norm": 0.5392100919936498, "learning_rate": 2.914387192123022e-05, "loss": 2.6826, "step": 32537 }, { "epoch": 2.019864671922528, "grad_norm": 0.5730791006147817, "learning_rate": 2.9140589641001548e-05, "loss": 2.6997, "step": 32538 }, { "epoch": 2.019926749022286, "grad_norm": 0.5573438518963617, "learning_rate": 2.9137307469601472e-05, "loss": 2.6817, "step": 32539 }, { "epoch": 2.019988826122044, "grad_norm": 0.6006602171577394, "learning_rate": 2.913402540704714e-05, "loss": 2.7512, "step": 32540 }, { "epoch": 2.0200509032218017, "grad_norm": 0.5525833300929596, "learning_rate": 2.913074345335567e-05, "loss": 2.7238, "step": 32541 }, { "epoch": 2.020112980321559, "grad_norm": 0.566111573484384, "learning_rate": 2.9127461608544172e-05, "loss": 2.7116, "step": 32542 }, { "epoch": 2.020175057421317, "grad_norm": 0.5388923874175608, "learning_rate": 2.912417987262975e-05, "loss": 2.6098, "step": 32543 }, { "epoch": 2.020237134521075, "grad_norm": 0.5674844936319364, "learning_rate": 2.912089824562958e-05, "loss": 2.7589, "step": 32544 }, { "epoch": 2.020299211620833, "grad_norm": 0.5671223290363497, "learning_rate": 2.9117616727560747e-05, "loss": 2.772, "step": 32545 }, { "epoch": 2.020361288720591, "grad_norm": 0.5681304995162563, "learning_rate": 2.911433531844038e-05, "loss": 2.6893, "step": 32546 }, { "epoch": 2.0204233658203488, "grad_norm": 0.5959245614146025, "learning_rate": 2.911105401828559e-05, "loss": 2.7549, "step": 32547 }, { "epoch": 2.0204854429201067, "grad_norm": 0.5338158074220565, "learning_rate": 2.9107772827113483e-05, "loss": 2.6539, "step": 32548 }, { "epoch": 2.0205475200198646, "grad_norm": 0.5662745881794754, "learning_rate": 2.910449174494122e-05, "loss": 2.7622, "step": 32549 }, { "epoch": 2.0206095971196225, "grad_norm": 0.549277369489677, "learning_rate": 2.910121077178589e-05, "loss": 2.7392, "step": 32550 }, { "epoch": 2.0206716742193804, "grad_norm": 0.5065977091616594, "learning_rate": 2.9097929907664623e-05, "loss": 2.692, "step": 32551 }, { "epoch": 2.0207337513191383, "grad_norm": 0.6359920450198908, "learning_rate": 2.9094649152594512e-05, "loss": 2.7237, "step": 32552 }, { "epoch": 2.0207958284188963, "grad_norm": 0.516210974449015, "learning_rate": 2.909136850659271e-05, "loss": 2.6801, "step": 32553 }, { "epoch": 2.020857905518654, "grad_norm": 0.592433146757247, "learning_rate": 2.908808796967631e-05, "loss": 2.7304, "step": 32554 }, { "epoch": 2.020919982618412, "grad_norm": 0.5071318833652895, "learning_rate": 2.9084807541862435e-05, "loss": 2.7044, "step": 32555 }, { "epoch": 2.02098205971817, "grad_norm": 0.5849674332756378, "learning_rate": 2.908152722316817e-05, "loss": 2.6381, "step": 32556 }, { "epoch": 2.021044136817928, "grad_norm": 0.5154005197798477, "learning_rate": 2.9078247013610682e-05, "loss": 2.7569, "step": 32557 }, { "epoch": 2.021106213917686, "grad_norm": 0.5681839505068909, "learning_rate": 2.9074966913207036e-05, "loss": 2.6696, "step": 32558 }, { "epoch": 2.0211682910174438, "grad_norm": 0.5478536568395936, "learning_rate": 2.907168692197439e-05, "loss": 2.7582, "step": 32559 }, { "epoch": 2.0212303681172017, "grad_norm": 0.6264671034263002, "learning_rate": 2.9068407039929835e-05, "loss": 2.8295, "step": 32560 }, { "epoch": 2.0212924452169596, "grad_norm": 0.5344842185678369, "learning_rate": 2.9065127267090476e-05, "loss": 2.6682, "step": 32561 }, { "epoch": 2.0213545223167175, "grad_norm": 0.4957228948981009, "learning_rate": 2.9061847603473413e-05, "loss": 2.5866, "step": 32562 }, { "epoch": 2.0214165994164754, "grad_norm": 0.5442131109216587, "learning_rate": 2.905856804909579e-05, "loss": 2.7323, "step": 32563 }, { "epoch": 2.0214786765162334, "grad_norm": 0.5456946577809239, "learning_rate": 2.9055288603974705e-05, "loss": 2.6153, "step": 32564 }, { "epoch": 2.021540753615991, "grad_norm": 0.5914968353196499, "learning_rate": 2.9052009268127267e-05, "loss": 2.6572, "step": 32565 }, { "epoch": 2.0216028307157488, "grad_norm": 0.5543377270473766, "learning_rate": 2.9048730041570558e-05, "loss": 2.742, "step": 32566 }, { "epoch": 2.0216649078155067, "grad_norm": 0.5523887874905159, "learning_rate": 2.904545092432173e-05, "loss": 2.6738, "step": 32567 }, { "epoch": 2.0217269849152646, "grad_norm": 0.5861643542783824, "learning_rate": 2.9042171916397877e-05, "loss": 2.6771, "step": 32568 }, { "epoch": 2.0217890620150225, "grad_norm": 0.542600987435387, "learning_rate": 2.9038893017816093e-05, "loss": 2.6925, "step": 32569 }, { "epoch": 2.0218511391147804, "grad_norm": 0.560142823381597, "learning_rate": 2.9035614228593476e-05, "loss": 2.6841, "step": 32570 }, { "epoch": 2.0219132162145383, "grad_norm": 0.6225490674056682, "learning_rate": 2.903233554874717e-05, "loss": 2.7232, "step": 32571 }, { "epoch": 2.0219752933142963, "grad_norm": 0.5268683940161085, "learning_rate": 2.902905697829426e-05, "loss": 2.7005, "step": 32572 }, { "epoch": 2.022037370414054, "grad_norm": 0.5975893768595779, "learning_rate": 2.902577851725185e-05, "loss": 2.6461, "step": 32573 }, { "epoch": 2.022099447513812, "grad_norm": 0.4909642018876054, "learning_rate": 2.9022500165637044e-05, "loss": 2.716, "step": 32574 }, { "epoch": 2.02216152461357, "grad_norm": 0.5711069713476301, "learning_rate": 2.9019221923466926e-05, "loss": 2.7339, "step": 32575 }, { "epoch": 2.022223601713328, "grad_norm": 0.5542592289233258, "learning_rate": 2.9015943790758642e-05, "loss": 2.6948, "step": 32576 }, { "epoch": 2.022285678813086, "grad_norm": 0.5257468168368186, "learning_rate": 2.9012665767529278e-05, "loss": 2.7817, "step": 32577 }, { "epoch": 2.0223477559128438, "grad_norm": 0.5630493660283261, "learning_rate": 2.900938785379592e-05, "loss": 2.6244, "step": 32578 }, { "epoch": 2.0224098330126017, "grad_norm": 0.5211593529442508, "learning_rate": 2.9006110049575674e-05, "loss": 2.698, "step": 32579 }, { "epoch": 2.0224719101123596, "grad_norm": 0.5646123958394405, "learning_rate": 2.9002832354885663e-05, "loss": 2.8109, "step": 32580 }, { "epoch": 2.0225339872121175, "grad_norm": 0.5922477188291856, "learning_rate": 2.899955476974297e-05, "loss": 2.6093, "step": 32581 }, { "epoch": 2.0225960643118754, "grad_norm": 0.5718225840743532, "learning_rate": 2.89962772941647e-05, "loss": 2.7151, "step": 32582 }, { "epoch": 2.0226581414116334, "grad_norm": 0.6119534399157673, "learning_rate": 2.8992999928167952e-05, "loss": 2.7345, "step": 32583 }, { "epoch": 2.0227202185113913, "grad_norm": 0.5293130907340767, "learning_rate": 2.8989722671769804e-05, "loss": 2.702, "step": 32584 }, { "epoch": 2.022782295611149, "grad_norm": 0.608558474535722, "learning_rate": 2.8986445524987382e-05, "loss": 2.6905, "step": 32585 }, { "epoch": 2.022844372710907, "grad_norm": 0.5581486135444894, "learning_rate": 2.8983168487837787e-05, "loss": 2.7489, "step": 32586 }, { "epoch": 2.022906449810665, "grad_norm": 0.5639729552903534, "learning_rate": 2.89798915603381e-05, "loss": 2.585, "step": 32587 }, { "epoch": 2.022968526910423, "grad_norm": 0.5773473354945778, "learning_rate": 2.8976614742505403e-05, "loss": 2.7394, "step": 32588 }, { "epoch": 2.023030604010181, "grad_norm": 0.5858738419602924, "learning_rate": 2.897333803435683e-05, "loss": 2.7531, "step": 32589 }, { "epoch": 2.0230926811099383, "grad_norm": 0.6167563852135052, "learning_rate": 2.8970061435909435e-05, "loss": 2.8108, "step": 32590 }, { "epoch": 2.0231547582096963, "grad_norm": 0.5200901037334883, "learning_rate": 2.8966784947180358e-05, "loss": 2.6686, "step": 32591 }, { "epoch": 2.023216835309454, "grad_norm": 0.6706091688905925, "learning_rate": 2.896350856818667e-05, "loss": 2.7264, "step": 32592 }, { "epoch": 2.023278912409212, "grad_norm": 0.4830354194448824, "learning_rate": 2.8960232298945437e-05, "loss": 2.695, "step": 32593 }, { "epoch": 2.02334098950897, "grad_norm": 0.5853142793611037, "learning_rate": 2.8956956139473808e-05, "loss": 2.7295, "step": 32594 }, { "epoch": 2.023403066608728, "grad_norm": 0.5046900486365057, "learning_rate": 2.8953680089788842e-05, "loss": 2.7265, "step": 32595 }, { "epoch": 2.023465143708486, "grad_norm": 0.5410904815337176, "learning_rate": 2.895040414990764e-05, "loss": 2.737, "step": 32596 }, { "epoch": 2.0235272208082438, "grad_norm": 0.5790613170103206, "learning_rate": 2.8947128319847287e-05, "loss": 2.7776, "step": 32597 }, { "epoch": 2.0235892979080017, "grad_norm": 0.5142158847614258, "learning_rate": 2.894385259962486e-05, "loss": 2.7426, "step": 32598 }, { "epoch": 2.0236513750077596, "grad_norm": 0.5385290906853208, "learning_rate": 2.8940576989257483e-05, "loss": 2.7223, "step": 32599 }, { "epoch": 2.0237134521075175, "grad_norm": 0.5239998088774441, "learning_rate": 2.8937301488762227e-05, "loss": 2.7056, "step": 32600 }, { "epoch": 2.0237755292072754, "grad_norm": 0.5026731559037739, "learning_rate": 2.8934026098156186e-05, "loss": 2.7674, "step": 32601 }, { "epoch": 2.0238376063070334, "grad_norm": 0.546091444804684, "learning_rate": 2.893075081745642e-05, "loss": 2.6051, "step": 32602 }, { "epoch": 2.0238996834067913, "grad_norm": 0.45879379536350234, "learning_rate": 2.8927475646680062e-05, "loss": 2.7203, "step": 32603 }, { "epoch": 2.023961760506549, "grad_norm": 0.5532195586936809, "learning_rate": 2.8924200585844185e-05, "loss": 2.7323, "step": 32604 }, { "epoch": 2.024023837606307, "grad_norm": 0.5034036717572318, "learning_rate": 2.8920925634965867e-05, "loss": 2.6842, "step": 32605 }, { "epoch": 2.024085914706065, "grad_norm": 0.5202876443052402, "learning_rate": 2.8917650794062172e-05, "loss": 2.6863, "step": 32606 }, { "epoch": 2.024147991805823, "grad_norm": 0.5434581554363513, "learning_rate": 2.8914376063150233e-05, "loss": 2.6979, "step": 32607 }, { "epoch": 2.024210068905581, "grad_norm": 0.5073168169349959, "learning_rate": 2.8911101442247114e-05, "loss": 2.6248, "step": 32608 }, { "epoch": 2.024272146005339, "grad_norm": 0.5590617746739722, "learning_rate": 2.8907826931369903e-05, "loss": 2.7219, "step": 32609 }, { "epoch": 2.0243342231050967, "grad_norm": 0.5362288314358056, "learning_rate": 2.890455253053567e-05, "loss": 2.7092, "step": 32610 }, { "epoch": 2.0243963002048546, "grad_norm": 0.49818777447435586, "learning_rate": 2.89012782397615e-05, "loss": 2.7325, "step": 32611 }, { "epoch": 2.0244583773046125, "grad_norm": 0.5491223886967463, "learning_rate": 2.8898004059064493e-05, "loss": 2.7301, "step": 32612 }, { "epoch": 2.02452045440437, "grad_norm": 0.5244049412450911, "learning_rate": 2.8894729988461722e-05, "loss": 2.6503, "step": 32613 }, { "epoch": 2.024582531504128, "grad_norm": 0.5364155119561071, "learning_rate": 2.889145602797027e-05, "loss": 2.6734, "step": 32614 }, { "epoch": 2.024644608603886, "grad_norm": 0.4988887328845849, "learning_rate": 2.88881821776072e-05, "loss": 2.7441, "step": 32615 }, { "epoch": 2.0247066857036438, "grad_norm": 0.534202393880841, "learning_rate": 2.8884908437389624e-05, "loss": 2.6636, "step": 32616 }, { "epoch": 2.0247687628034017, "grad_norm": 0.514736595058931, "learning_rate": 2.8881634807334608e-05, "loss": 2.6259, "step": 32617 }, { "epoch": 2.0248308399031596, "grad_norm": 0.5179815195848438, "learning_rate": 2.8878361287459233e-05, "loss": 2.6713, "step": 32618 }, { "epoch": 2.0248929170029175, "grad_norm": 0.5370059659420476, "learning_rate": 2.887508787778055e-05, "loss": 2.6969, "step": 32619 }, { "epoch": 2.0249549941026754, "grad_norm": 0.5481443387820764, "learning_rate": 2.887181457831568e-05, "loss": 2.7166, "step": 32620 }, { "epoch": 2.0250170712024333, "grad_norm": 0.5415060526579135, "learning_rate": 2.8868541389081683e-05, "loss": 2.7212, "step": 32621 }, { "epoch": 2.0250791483021913, "grad_norm": 0.5710504538931486, "learning_rate": 2.886526831009564e-05, "loss": 2.7062, "step": 32622 }, { "epoch": 2.025141225401949, "grad_norm": 0.5158807524833388, "learning_rate": 2.8861995341374602e-05, "loss": 2.7872, "step": 32623 }, { "epoch": 2.025203302501707, "grad_norm": 0.5626188523508198, "learning_rate": 2.8858722482935685e-05, "loss": 2.8007, "step": 32624 }, { "epoch": 2.025265379601465, "grad_norm": 0.5249552003728786, "learning_rate": 2.8855449734795924e-05, "loss": 2.7662, "step": 32625 }, { "epoch": 2.025327456701223, "grad_norm": 0.49886550942466473, "learning_rate": 2.8852177096972433e-05, "loss": 2.6297, "step": 32626 }, { "epoch": 2.025389533800981, "grad_norm": 0.528148658493391, "learning_rate": 2.8848904569482272e-05, "loss": 2.7614, "step": 32627 }, { "epoch": 2.0254516109007388, "grad_norm": 0.4897761996477436, "learning_rate": 2.8845632152342506e-05, "loss": 2.6698, "step": 32628 }, { "epoch": 2.0255136880004967, "grad_norm": 0.5019432393447595, "learning_rate": 2.88423598455702e-05, "loss": 2.6929, "step": 32629 }, { "epoch": 2.0255757651002546, "grad_norm": 0.5036168117373554, "learning_rate": 2.883908764918245e-05, "loss": 2.6395, "step": 32630 }, { "epoch": 2.0256378422000125, "grad_norm": 0.5026325718523532, "learning_rate": 2.8835815563196324e-05, "loss": 2.6082, "step": 32631 }, { "epoch": 2.0256999192997704, "grad_norm": 0.5300893165345861, "learning_rate": 2.883254358762888e-05, "loss": 2.7225, "step": 32632 }, { "epoch": 2.0257619963995284, "grad_norm": 0.5301688918437069, "learning_rate": 2.88292717224972e-05, "loss": 2.7022, "step": 32633 }, { "epoch": 2.0258240734992863, "grad_norm": 0.5472324565499508, "learning_rate": 2.882599996781833e-05, "loss": 2.7021, "step": 32634 }, { "epoch": 2.025886150599044, "grad_norm": 0.5303325023037202, "learning_rate": 2.8822728323609372e-05, "loss": 2.7128, "step": 32635 }, { "epoch": 2.025948227698802, "grad_norm": 0.5235531807222964, "learning_rate": 2.881945678988738e-05, "loss": 2.675, "step": 32636 }, { "epoch": 2.0260103047985596, "grad_norm": 0.5557614527735043, "learning_rate": 2.881618536666943e-05, "loss": 2.6291, "step": 32637 }, { "epoch": 2.0260723818983175, "grad_norm": 0.5244966012228203, "learning_rate": 2.881291405397256e-05, "loss": 2.6916, "step": 32638 }, { "epoch": 2.0261344589980754, "grad_norm": 0.513104985366018, "learning_rate": 2.880964285181388e-05, "loss": 2.6992, "step": 32639 }, { "epoch": 2.0261965360978333, "grad_norm": 0.5559201848644943, "learning_rate": 2.8806371760210437e-05, "loss": 2.727, "step": 32640 }, { "epoch": 2.0262586131975913, "grad_norm": 0.5206944916360723, "learning_rate": 2.8803100779179305e-05, "loss": 2.7092, "step": 32641 }, { "epoch": 2.026320690297349, "grad_norm": 0.5660629284286267, "learning_rate": 2.879982990873752e-05, "loss": 2.7991, "step": 32642 }, { "epoch": 2.026382767397107, "grad_norm": 0.567487066936586, "learning_rate": 2.879655914890218e-05, "loss": 2.8002, "step": 32643 }, { "epoch": 2.026444844496865, "grad_norm": 0.5098602674768176, "learning_rate": 2.879328849969034e-05, "loss": 2.6713, "step": 32644 }, { "epoch": 2.026506921596623, "grad_norm": 0.5126827856077514, "learning_rate": 2.879001796111907e-05, "loss": 2.6826, "step": 32645 }, { "epoch": 2.026568998696381, "grad_norm": 0.5277322538998188, "learning_rate": 2.8786747533205423e-05, "loss": 2.7384, "step": 32646 }, { "epoch": 2.0266310757961388, "grad_norm": 0.5444865795404719, "learning_rate": 2.8783477215966436e-05, "loss": 2.6836, "step": 32647 }, { "epoch": 2.0266931528958967, "grad_norm": 0.5271796642324887, "learning_rate": 2.878020700941923e-05, "loss": 2.7155, "step": 32648 }, { "epoch": 2.0267552299956546, "grad_norm": 0.5258512498530795, "learning_rate": 2.8776936913580833e-05, "loss": 2.7636, "step": 32649 }, { "epoch": 2.0268173070954125, "grad_norm": 0.5410508512741427, "learning_rate": 2.87736669284683e-05, "loss": 2.7316, "step": 32650 }, { "epoch": 2.0268793841951704, "grad_norm": 0.545094479805794, "learning_rate": 2.8770397054098687e-05, "loss": 2.6729, "step": 32651 }, { "epoch": 2.0269414612949284, "grad_norm": 0.5493491201196327, "learning_rate": 2.8767127290489084e-05, "loss": 2.6036, "step": 32652 }, { "epoch": 2.0270035383946863, "grad_norm": 0.5448719945248429, "learning_rate": 2.8763857637656533e-05, "loss": 2.7253, "step": 32653 }, { "epoch": 2.027065615494444, "grad_norm": 0.5204507248940039, "learning_rate": 2.876058809561809e-05, "loss": 2.7462, "step": 32654 }, { "epoch": 2.027127692594202, "grad_norm": 0.5879912939724656, "learning_rate": 2.8757318664390796e-05, "loss": 2.7152, "step": 32655 }, { "epoch": 2.02718976969396, "grad_norm": 0.538792860623009, "learning_rate": 2.875404934399173e-05, "loss": 2.671, "step": 32656 }, { "epoch": 2.027251846793718, "grad_norm": 0.552207048218002, "learning_rate": 2.875078013443796e-05, "loss": 2.6737, "step": 32657 }, { "epoch": 2.027313923893476, "grad_norm": 0.5239484769628482, "learning_rate": 2.8747511035746534e-05, "loss": 2.6459, "step": 32658 }, { "epoch": 2.027376000993234, "grad_norm": 0.5774097399751947, "learning_rate": 2.87442420479345e-05, "loss": 2.7127, "step": 32659 }, { "epoch": 2.0274380780929917, "grad_norm": 0.5077181508374243, "learning_rate": 2.8740973171018914e-05, "loss": 2.6358, "step": 32660 }, { "epoch": 2.027500155192749, "grad_norm": 0.5369052846003, "learning_rate": 2.8737704405016812e-05, "loss": 2.6707, "step": 32661 }, { "epoch": 2.027562232292507, "grad_norm": 0.5265310453534174, "learning_rate": 2.873443574994529e-05, "loss": 2.7666, "step": 32662 }, { "epoch": 2.027624309392265, "grad_norm": 0.5439389843118906, "learning_rate": 2.8731167205821373e-05, "loss": 2.7198, "step": 32663 }, { "epoch": 2.027686386492023, "grad_norm": 0.54201775567859, "learning_rate": 2.872789877266213e-05, "loss": 2.5463, "step": 32664 }, { "epoch": 2.027748463591781, "grad_norm": 0.512400078918442, "learning_rate": 2.872463045048458e-05, "loss": 2.7495, "step": 32665 }, { "epoch": 2.0278105406915388, "grad_norm": 0.5507933973771968, "learning_rate": 2.8721362239305817e-05, "loss": 2.7139, "step": 32666 }, { "epoch": 2.0278726177912967, "grad_norm": 0.5024949945134102, "learning_rate": 2.871809413914287e-05, "loss": 2.6183, "step": 32667 }, { "epoch": 2.0279346948910546, "grad_norm": 0.518639681472451, "learning_rate": 2.871482615001279e-05, "loss": 2.6765, "step": 32668 }, { "epoch": 2.0279967719908125, "grad_norm": 0.5162530599692778, "learning_rate": 2.8711558271932616e-05, "loss": 2.6006, "step": 32669 }, { "epoch": 2.0280588490905704, "grad_norm": 0.5570514702036149, "learning_rate": 2.8708290504919427e-05, "loss": 2.7337, "step": 32670 }, { "epoch": 2.0281209261903284, "grad_norm": 0.48967995772307665, "learning_rate": 2.8705022848990258e-05, "loss": 2.6932, "step": 32671 }, { "epoch": 2.0281830032900863, "grad_norm": 0.5078951254895973, "learning_rate": 2.870175530416215e-05, "loss": 2.6789, "step": 32672 }, { "epoch": 2.028245080389844, "grad_norm": 0.5179925662818153, "learning_rate": 2.8698487870452156e-05, "loss": 2.6431, "step": 32673 }, { "epoch": 2.028307157489602, "grad_norm": 0.5012511157263613, "learning_rate": 2.86952205478773e-05, "loss": 2.7061, "step": 32674 }, { "epoch": 2.02836923458936, "grad_norm": 0.5126228789978872, "learning_rate": 2.869195333645468e-05, "loss": 2.6388, "step": 32675 }, { "epoch": 2.028431311689118, "grad_norm": 0.5344812912106658, "learning_rate": 2.86886862362013e-05, "loss": 2.7174, "step": 32676 }, { "epoch": 2.028493388788876, "grad_norm": 0.5341815843844863, "learning_rate": 2.8685419247134226e-05, "loss": 2.6917, "step": 32677 }, { "epoch": 2.028555465888634, "grad_norm": 0.5403080523844535, "learning_rate": 2.8682152369270464e-05, "loss": 2.7572, "step": 32678 }, { "epoch": 2.0286175429883917, "grad_norm": 0.5663036648160357, "learning_rate": 2.867888560262712e-05, "loss": 2.7631, "step": 32679 }, { "epoch": 2.0286796200881496, "grad_norm": 0.5820899524143992, "learning_rate": 2.86756189472212e-05, "loss": 2.746, "step": 32680 }, { "epoch": 2.0287416971879075, "grad_norm": 0.5551735266486212, "learning_rate": 2.867235240306975e-05, "loss": 2.7189, "step": 32681 }, { "epoch": 2.0288037742876655, "grad_norm": 0.5106849711300332, "learning_rate": 2.8669085970189825e-05, "loss": 2.7389, "step": 32682 }, { "epoch": 2.0288658513874234, "grad_norm": 0.5553054322676871, "learning_rate": 2.8665819648598425e-05, "loss": 2.6741, "step": 32683 }, { "epoch": 2.0289279284871813, "grad_norm": 0.5031341433434701, "learning_rate": 2.866255343831265e-05, "loss": 2.6615, "step": 32684 }, { "epoch": 2.0289900055869388, "grad_norm": 0.524540294606737, "learning_rate": 2.865928733934951e-05, "loss": 2.7538, "step": 32685 }, { "epoch": 2.0290520826866967, "grad_norm": 0.552742300886609, "learning_rate": 2.865602135172605e-05, "loss": 2.7162, "step": 32686 }, { "epoch": 2.0291141597864546, "grad_norm": 0.5269013138305623, "learning_rate": 2.8652755475459293e-05, "loss": 2.7112, "step": 32687 }, { "epoch": 2.0291762368862125, "grad_norm": 0.5123321991200365, "learning_rate": 2.86494897105663e-05, "loss": 2.6558, "step": 32688 }, { "epoch": 2.0292383139859704, "grad_norm": 0.49398300612328006, "learning_rate": 2.8646224057064097e-05, "loss": 2.5808, "step": 32689 }, { "epoch": 2.0293003910857284, "grad_norm": 0.5134466279345108, "learning_rate": 2.8642958514969735e-05, "loss": 2.7075, "step": 32690 }, { "epoch": 2.0293624681854863, "grad_norm": 0.517262863015834, "learning_rate": 2.863969308430025e-05, "loss": 2.7012, "step": 32691 }, { "epoch": 2.029424545285244, "grad_norm": 0.517833833686645, "learning_rate": 2.863642776507265e-05, "loss": 2.7431, "step": 32692 }, { "epoch": 2.029486622385002, "grad_norm": 0.54769503555416, "learning_rate": 2.863316255730401e-05, "loss": 2.7437, "step": 32693 }, { "epoch": 2.02954869948476, "grad_norm": 0.530222881569714, "learning_rate": 2.8629897461011353e-05, "loss": 2.6856, "step": 32694 }, { "epoch": 2.029610776584518, "grad_norm": 0.6126260112060595, "learning_rate": 2.86266324762117e-05, "loss": 2.6874, "step": 32695 }, { "epoch": 2.029672853684276, "grad_norm": 0.48712085933653204, "learning_rate": 2.86233676029221e-05, "loss": 2.6676, "step": 32696 }, { "epoch": 2.0297349307840338, "grad_norm": 0.5378931049630501, "learning_rate": 2.862010284115957e-05, "loss": 2.6766, "step": 32697 }, { "epoch": 2.0297970078837917, "grad_norm": 0.5223807014816898, "learning_rate": 2.861683819094116e-05, "loss": 2.6937, "step": 32698 }, { "epoch": 2.0298590849835496, "grad_norm": 0.5225549345934538, "learning_rate": 2.8613573652283897e-05, "loss": 2.7103, "step": 32699 }, { "epoch": 2.0299211620833075, "grad_norm": 0.5398277158881655, "learning_rate": 2.8610309225204818e-05, "loss": 2.7226, "step": 32700 }, { "epoch": 2.0299832391830654, "grad_norm": 0.5304057481345673, "learning_rate": 2.8607044909720927e-05, "loss": 2.7254, "step": 32701 }, { "epoch": 2.0300453162828234, "grad_norm": 0.5597281508004237, "learning_rate": 2.8603780705849293e-05, "loss": 2.653, "step": 32702 }, { "epoch": 2.0301073933825813, "grad_norm": 0.580504597805849, "learning_rate": 2.8600516613606932e-05, "loss": 2.6804, "step": 32703 }, { "epoch": 2.030169470482339, "grad_norm": 0.5334081011259594, "learning_rate": 2.8597252633010863e-05, "loss": 2.6551, "step": 32704 }, { "epoch": 2.030231547582097, "grad_norm": 0.5187438792131871, "learning_rate": 2.859398876407811e-05, "loss": 2.7142, "step": 32705 }, { "epoch": 2.030293624681855, "grad_norm": 0.5621092286682999, "learning_rate": 2.8590725006825736e-05, "loss": 2.6607, "step": 32706 }, { "epoch": 2.030355701781613, "grad_norm": 0.5521786865325474, "learning_rate": 2.858746136127074e-05, "loss": 2.6731, "step": 32707 }, { "epoch": 2.030417778881371, "grad_norm": 0.5614575687419755, "learning_rate": 2.8584197827430164e-05, "loss": 2.6688, "step": 32708 }, { "epoch": 2.0304798559811283, "grad_norm": 0.5426123842391065, "learning_rate": 2.8580934405321026e-05, "loss": 2.762, "step": 32709 }, { "epoch": 2.0305419330808863, "grad_norm": 0.5282589917372223, "learning_rate": 2.8577671094960325e-05, "loss": 2.6543, "step": 32710 }, { "epoch": 2.030604010180644, "grad_norm": 0.5594608273007252, "learning_rate": 2.857440789636514e-05, "loss": 2.6719, "step": 32711 }, { "epoch": 2.030666087280402, "grad_norm": 0.5272939870454979, "learning_rate": 2.857114480955247e-05, "loss": 2.7243, "step": 32712 }, { "epoch": 2.03072816438016, "grad_norm": 0.5089160079887246, "learning_rate": 2.856788183453934e-05, "loss": 2.6627, "step": 32713 }, { "epoch": 2.030790241479918, "grad_norm": 0.518997204719763, "learning_rate": 2.856461897134275e-05, "loss": 2.7456, "step": 32714 }, { "epoch": 2.030852318579676, "grad_norm": 0.5175412434806882, "learning_rate": 2.8561356219979773e-05, "loss": 2.7339, "step": 32715 }, { "epoch": 2.0309143956794338, "grad_norm": 0.5570542789530115, "learning_rate": 2.8558093580467397e-05, "loss": 2.6866, "step": 32716 }, { "epoch": 2.0309764727791917, "grad_norm": 0.5747483335058503, "learning_rate": 2.8554831052822656e-05, "loss": 2.7089, "step": 32717 }, { "epoch": 2.0310385498789496, "grad_norm": 0.5288761902328943, "learning_rate": 2.8551568637062565e-05, "loss": 2.6692, "step": 32718 }, { "epoch": 2.0311006269787075, "grad_norm": 0.556047438054048, "learning_rate": 2.8548306333204132e-05, "loss": 2.6289, "step": 32719 }, { "epoch": 2.0311627040784654, "grad_norm": 0.5443374617342761, "learning_rate": 2.854504414126441e-05, "loss": 2.704, "step": 32720 }, { "epoch": 2.0312247811782234, "grad_norm": 0.5693467206244607, "learning_rate": 2.854178206126038e-05, "loss": 2.6706, "step": 32721 }, { "epoch": 2.0312868582779813, "grad_norm": 0.5047250940357461, "learning_rate": 2.8538520093209104e-05, "loss": 2.704, "step": 32722 }, { "epoch": 2.031348935377739, "grad_norm": 0.5718994010347184, "learning_rate": 2.8535258237127578e-05, "loss": 2.7304, "step": 32723 }, { "epoch": 2.031411012477497, "grad_norm": 0.5857722748816467, "learning_rate": 2.85319964930328e-05, "loss": 2.6357, "step": 32724 }, { "epoch": 2.031473089577255, "grad_norm": 0.5305094044631072, "learning_rate": 2.852873486094183e-05, "loss": 2.6333, "step": 32725 }, { "epoch": 2.031535166677013, "grad_norm": 0.6054578472509418, "learning_rate": 2.8525473340871655e-05, "loss": 2.7193, "step": 32726 }, { "epoch": 2.031597243776771, "grad_norm": 0.4941282911608422, "learning_rate": 2.8522211932839305e-05, "loss": 2.66, "step": 32727 }, { "epoch": 2.031659320876529, "grad_norm": 0.6034761187514551, "learning_rate": 2.8518950636861775e-05, "loss": 2.7382, "step": 32728 }, { "epoch": 2.0317213979762867, "grad_norm": 0.5713283427501604, "learning_rate": 2.851568945295611e-05, "loss": 2.7056, "step": 32729 }, { "epoch": 2.0317834750760446, "grad_norm": 0.5820933571263934, "learning_rate": 2.85124283811393e-05, "loss": 2.6803, "step": 32730 }, { "epoch": 2.0318455521758025, "grad_norm": 0.6132252904545797, "learning_rate": 2.8509167421428373e-05, "loss": 2.7905, "step": 32731 }, { "epoch": 2.0319076292755605, "grad_norm": 0.5207773337458727, "learning_rate": 2.8505906573840336e-05, "loss": 2.7003, "step": 32732 }, { "epoch": 2.031969706375318, "grad_norm": 0.6156321133042877, "learning_rate": 2.8502645838392185e-05, "loss": 2.7964, "step": 32733 }, { "epoch": 2.032031783475076, "grad_norm": 0.5163330938140842, "learning_rate": 2.849938521510097e-05, "loss": 2.7774, "step": 32734 }, { "epoch": 2.0320938605748338, "grad_norm": 0.6036291170041062, "learning_rate": 2.8496124703983677e-05, "loss": 2.6367, "step": 32735 }, { "epoch": 2.0321559376745917, "grad_norm": 0.49094775147300473, "learning_rate": 2.8492864305057325e-05, "loss": 2.6661, "step": 32736 }, { "epoch": 2.0322180147743496, "grad_norm": 0.5350041973983248, "learning_rate": 2.84896040183389e-05, "loss": 2.6217, "step": 32737 }, { "epoch": 2.0322800918741075, "grad_norm": 0.5419839119494199, "learning_rate": 2.8486343843845453e-05, "loss": 2.6756, "step": 32738 }, { "epoch": 2.0323421689738654, "grad_norm": 0.5372365072818427, "learning_rate": 2.848308378159398e-05, "loss": 2.7126, "step": 32739 }, { "epoch": 2.0324042460736234, "grad_norm": 0.5768419410432932, "learning_rate": 2.847982383160147e-05, "loss": 2.7092, "step": 32740 }, { "epoch": 2.0324663231733813, "grad_norm": 0.6047391864461114, "learning_rate": 2.847656399388493e-05, "loss": 2.6737, "step": 32741 }, { "epoch": 2.032528400273139, "grad_norm": 0.5254928602846077, "learning_rate": 2.84733042684614e-05, "loss": 2.711, "step": 32742 }, { "epoch": 2.032590477372897, "grad_norm": 0.5873448826565062, "learning_rate": 2.8470044655347867e-05, "loss": 2.6627, "step": 32743 }, { "epoch": 2.032652554472655, "grad_norm": 0.5680346528114162, "learning_rate": 2.8466785154561336e-05, "loss": 2.6991, "step": 32744 }, { "epoch": 2.032714631572413, "grad_norm": 0.5129339044444207, "learning_rate": 2.8463525766118815e-05, "loss": 2.602, "step": 32745 }, { "epoch": 2.032776708672171, "grad_norm": 0.6358793377346821, "learning_rate": 2.846026649003729e-05, "loss": 2.7212, "step": 32746 }, { "epoch": 2.032838785771929, "grad_norm": 0.4962287478856855, "learning_rate": 2.84570073263338e-05, "loss": 2.6821, "step": 32747 }, { "epoch": 2.0329008628716867, "grad_norm": 0.5538404471368815, "learning_rate": 2.845374827502534e-05, "loss": 2.7027, "step": 32748 }, { "epoch": 2.0329629399714446, "grad_norm": 0.6330855451606067, "learning_rate": 2.8450489336128892e-05, "loss": 2.6621, "step": 32749 }, { "epoch": 2.0330250170712025, "grad_norm": 0.49183533194711204, "learning_rate": 2.8447230509661466e-05, "loss": 2.6607, "step": 32750 }, { "epoch": 2.0330870941709605, "grad_norm": 0.5497968276768066, "learning_rate": 2.844397179564009e-05, "loss": 2.7586, "step": 32751 }, { "epoch": 2.0331491712707184, "grad_norm": 0.5239464577099445, "learning_rate": 2.8440713194081736e-05, "loss": 2.6823, "step": 32752 }, { "epoch": 2.0332112483704763, "grad_norm": 0.5420231165953361, "learning_rate": 2.843745470500342e-05, "loss": 2.6088, "step": 32753 }, { "epoch": 2.033273325470234, "grad_norm": 0.47796225177678536, "learning_rate": 2.843419632842212e-05, "loss": 2.6174, "step": 32754 }, { "epoch": 2.033335402569992, "grad_norm": 0.5254750237159863, "learning_rate": 2.843093806435485e-05, "loss": 2.7235, "step": 32755 }, { "epoch": 2.03339747966975, "grad_norm": 0.5412741363641129, "learning_rate": 2.842767991281864e-05, "loss": 2.7166, "step": 32756 }, { "epoch": 2.0334595567695075, "grad_norm": 0.5356265319748895, "learning_rate": 2.8424421873830447e-05, "loss": 2.7354, "step": 32757 }, { "epoch": 2.0335216338692654, "grad_norm": 0.5225935437914265, "learning_rate": 2.8421163947407294e-05, "loss": 2.6193, "step": 32758 }, { "epoch": 2.0335837109690234, "grad_norm": 0.5179311735812447, "learning_rate": 2.8417906133566163e-05, "loss": 2.7142, "step": 32759 }, { "epoch": 2.0336457880687813, "grad_norm": 0.539841600295119, "learning_rate": 2.8414648432324033e-05, "loss": 2.6663, "step": 32760 }, { "epoch": 2.033707865168539, "grad_norm": 0.4664709637345896, "learning_rate": 2.8411390843697944e-05, "loss": 2.6594, "step": 32761 }, { "epoch": 2.033769942268297, "grad_norm": 0.5373223454726067, "learning_rate": 2.840813336770487e-05, "loss": 2.7516, "step": 32762 }, { "epoch": 2.033832019368055, "grad_norm": 0.546151691031621, "learning_rate": 2.8404876004361804e-05, "loss": 2.6414, "step": 32763 }, { "epoch": 2.033894096467813, "grad_norm": 0.5335291609762582, "learning_rate": 2.840161875368572e-05, "loss": 2.7542, "step": 32764 }, { "epoch": 2.033956173567571, "grad_norm": 0.542865084624996, "learning_rate": 2.839836161569366e-05, "loss": 2.7387, "step": 32765 }, { "epoch": 2.0340182506673288, "grad_norm": 0.5906734567776484, "learning_rate": 2.8395104590402587e-05, "loss": 2.6854, "step": 32766 }, { "epoch": 2.0340803277670867, "grad_norm": 0.5387626311465518, "learning_rate": 2.8391847677829507e-05, "loss": 2.6398, "step": 32767 }, { "epoch": 2.0341424048668446, "grad_norm": 0.5311224447715452, "learning_rate": 2.8388590877991396e-05, "loss": 2.618, "step": 32768 }, { "epoch": 2.0342044819666025, "grad_norm": 0.5431248008367702, "learning_rate": 2.838533419090523e-05, "loss": 2.6445, "step": 32769 }, { "epoch": 2.0342665590663604, "grad_norm": 0.5204809299991817, "learning_rate": 2.838207761658805e-05, "loss": 2.6417, "step": 32770 }, { "epoch": 2.0343286361661184, "grad_norm": 0.5214624293135095, "learning_rate": 2.8378821155056812e-05, "loss": 2.7018, "step": 32771 }, { "epoch": 2.0343907132658763, "grad_norm": 0.5322595148647102, "learning_rate": 2.8375564806328513e-05, "loss": 2.6286, "step": 32772 }, { "epoch": 2.034452790365634, "grad_norm": 0.5471848836037801, "learning_rate": 2.837230857042013e-05, "loss": 2.6418, "step": 32773 }, { "epoch": 2.034514867465392, "grad_norm": 0.5135446253053606, "learning_rate": 2.8369052447348675e-05, "loss": 2.6586, "step": 32774 }, { "epoch": 2.03457694456515, "grad_norm": 0.5239480219206195, "learning_rate": 2.836579643713113e-05, "loss": 2.5962, "step": 32775 }, { "epoch": 2.034639021664908, "grad_norm": 0.5204701365278059, "learning_rate": 2.8362540539784476e-05, "loss": 2.6869, "step": 32776 }, { "epoch": 2.034701098764666, "grad_norm": 0.5606673905284411, "learning_rate": 2.8359284755325677e-05, "loss": 2.7111, "step": 32777 }, { "epoch": 2.034763175864424, "grad_norm": 0.5179544771969632, "learning_rate": 2.835602908377177e-05, "loss": 2.7085, "step": 32778 }, { "epoch": 2.0348252529641817, "grad_norm": 0.6148710843399916, "learning_rate": 2.835277352513971e-05, "loss": 2.6692, "step": 32779 }, { "epoch": 2.0348873300639396, "grad_norm": 0.5591397851157079, "learning_rate": 2.834951807944648e-05, "loss": 2.6988, "step": 32780 }, { "epoch": 2.034949407163697, "grad_norm": 0.5719552614349035, "learning_rate": 2.8346262746709073e-05, "loss": 2.7086, "step": 32781 }, { "epoch": 2.035011484263455, "grad_norm": 0.5640080687236559, "learning_rate": 2.8343007526944454e-05, "loss": 2.7103, "step": 32782 }, { "epoch": 2.035073561363213, "grad_norm": 0.6272760670636469, "learning_rate": 2.833975242016964e-05, "loss": 2.7132, "step": 32783 }, { "epoch": 2.035135638462971, "grad_norm": 0.6187201194098727, "learning_rate": 2.8336497426401594e-05, "loss": 2.6845, "step": 32784 }, { "epoch": 2.0351977155627288, "grad_norm": 0.5197386650273315, "learning_rate": 2.83332425456573e-05, "loss": 2.7624, "step": 32785 }, { "epoch": 2.0352597926624867, "grad_norm": 0.6263204216337642, "learning_rate": 2.832998777795372e-05, "loss": 2.7335, "step": 32786 }, { "epoch": 2.0353218697622446, "grad_norm": 0.564356204889058, "learning_rate": 2.832673312330786e-05, "loss": 2.7127, "step": 32787 }, { "epoch": 2.0353839468620025, "grad_norm": 0.5616221996197652, "learning_rate": 2.8323478581736706e-05, "loss": 2.6769, "step": 32788 }, { "epoch": 2.0354460239617604, "grad_norm": 0.5685257291348061, "learning_rate": 2.832022415325723e-05, "loss": 2.65, "step": 32789 }, { "epoch": 2.0355081010615184, "grad_norm": 0.5228127124190826, "learning_rate": 2.831696983788641e-05, "loss": 2.705, "step": 32790 }, { "epoch": 2.0355701781612763, "grad_norm": 0.5326426708026937, "learning_rate": 2.83137156356412e-05, "loss": 2.6706, "step": 32791 }, { "epoch": 2.035632255261034, "grad_norm": 0.5614262239643872, "learning_rate": 2.8310461546538613e-05, "loss": 2.6936, "step": 32792 }, { "epoch": 2.035694332360792, "grad_norm": 0.5293607778789299, "learning_rate": 2.830720757059563e-05, "loss": 2.6096, "step": 32793 }, { "epoch": 2.03575640946055, "grad_norm": 0.5369564879173493, "learning_rate": 2.8303953707829195e-05, "loss": 2.7029, "step": 32794 }, { "epoch": 2.035818486560308, "grad_norm": 0.5038337063875793, "learning_rate": 2.8300699958256304e-05, "loss": 2.7001, "step": 32795 }, { "epoch": 2.035880563660066, "grad_norm": 0.570605111501498, "learning_rate": 2.8297446321893916e-05, "loss": 2.6978, "step": 32796 }, { "epoch": 2.035942640759824, "grad_norm": 0.5379466518679767, "learning_rate": 2.8294192798759034e-05, "loss": 2.6879, "step": 32797 }, { "epoch": 2.0360047178595817, "grad_norm": 0.5155577760281831, "learning_rate": 2.8290939388868622e-05, "loss": 2.6307, "step": 32798 }, { "epoch": 2.0360667949593396, "grad_norm": 0.5511496644171415, "learning_rate": 2.828768609223964e-05, "loss": 2.8011, "step": 32799 }, { "epoch": 2.0361288720590975, "grad_norm": 0.5099325089582682, "learning_rate": 2.8284432908889057e-05, "loss": 2.7588, "step": 32800 }, { "epoch": 2.0361909491588555, "grad_norm": 0.5719999666680813, "learning_rate": 2.8281179838833884e-05, "loss": 2.6525, "step": 32801 }, { "epoch": 2.0362530262586134, "grad_norm": 0.5122104350105119, "learning_rate": 2.8277926882091055e-05, "loss": 2.6353, "step": 32802 }, { "epoch": 2.0363151033583713, "grad_norm": 0.5115951214026837, "learning_rate": 2.8274674038677562e-05, "loss": 2.585, "step": 32803 }, { "epoch": 2.036377180458129, "grad_norm": 0.5931613413346682, "learning_rate": 2.827142130861037e-05, "loss": 2.7689, "step": 32804 }, { "epoch": 2.0364392575578867, "grad_norm": 0.4970701626101045, "learning_rate": 2.8268168691906426e-05, "loss": 2.5467, "step": 32805 }, { "epoch": 2.0365013346576446, "grad_norm": 0.5220077688399067, "learning_rate": 2.8264916188582746e-05, "loss": 2.7443, "step": 32806 }, { "epoch": 2.0365634117574025, "grad_norm": 0.5519860259858088, "learning_rate": 2.826166379865627e-05, "loss": 2.7511, "step": 32807 }, { "epoch": 2.0366254888571604, "grad_norm": 0.4804375801210422, "learning_rate": 2.8258411522143967e-05, "loss": 2.6399, "step": 32808 }, { "epoch": 2.0366875659569184, "grad_norm": 0.5151550868226313, "learning_rate": 2.82551593590628e-05, "loss": 2.6336, "step": 32809 }, { "epoch": 2.0367496430566763, "grad_norm": 0.486281457829742, "learning_rate": 2.825190730942976e-05, "loss": 2.6448, "step": 32810 }, { "epoch": 2.036811720156434, "grad_norm": 0.5039191724909322, "learning_rate": 2.8248655373261788e-05, "loss": 2.7346, "step": 32811 }, { "epoch": 2.036873797256192, "grad_norm": 0.4889131500791059, "learning_rate": 2.824540355057588e-05, "loss": 2.5775, "step": 32812 }, { "epoch": 2.03693587435595, "grad_norm": 0.4988291263196868, "learning_rate": 2.8242151841388955e-05, "loss": 2.7601, "step": 32813 }, { "epoch": 2.036997951455708, "grad_norm": 0.5150584663147068, "learning_rate": 2.8238900245718025e-05, "loss": 2.6564, "step": 32814 }, { "epoch": 2.037060028555466, "grad_norm": 0.5035423029068429, "learning_rate": 2.823564876358003e-05, "loss": 2.7545, "step": 32815 }, { "epoch": 2.037122105655224, "grad_norm": 0.5332846107678789, "learning_rate": 2.8232397394991944e-05, "loss": 2.6917, "step": 32816 }, { "epoch": 2.0371841827549817, "grad_norm": 0.46628748661224595, "learning_rate": 2.822914613997073e-05, "loss": 2.7193, "step": 32817 }, { "epoch": 2.0372462598547396, "grad_norm": 0.49137079508226095, "learning_rate": 2.8225894998533318e-05, "loss": 2.6801, "step": 32818 }, { "epoch": 2.0373083369544975, "grad_norm": 0.48587275254632917, "learning_rate": 2.8222643970696716e-05, "loss": 2.7051, "step": 32819 }, { "epoch": 2.0373704140542555, "grad_norm": 0.5442580688537287, "learning_rate": 2.8219393056477857e-05, "loss": 2.7853, "step": 32820 }, { "epoch": 2.0374324911540134, "grad_norm": 0.5161994566201797, "learning_rate": 2.8216142255893723e-05, "loss": 2.6737, "step": 32821 }, { "epoch": 2.0374945682537713, "grad_norm": 0.508420521523585, "learning_rate": 2.8212891568961264e-05, "loss": 2.7582, "step": 32822 }, { "epoch": 2.037556645353529, "grad_norm": 0.5145668964103187, "learning_rate": 2.8209640995697422e-05, "loss": 2.686, "step": 32823 }, { "epoch": 2.037618722453287, "grad_norm": 0.5290419467870103, "learning_rate": 2.820639053611919e-05, "loss": 2.728, "step": 32824 }, { "epoch": 2.037680799553045, "grad_norm": 0.5263058656773344, "learning_rate": 2.8203140190243516e-05, "loss": 2.6341, "step": 32825 }, { "epoch": 2.037742876652803, "grad_norm": 0.5969105475260881, "learning_rate": 2.8199889958087343e-05, "loss": 2.7843, "step": 32826 }, { "epoch": 2.037804953752561, "grad_norm": 0.5429352400065105, "learning_rate": 2.819663983966762e-05, "loss": 2.6406, "step": 32827 }, { "epoch": 2.037867030852319, "grad_norm": 0.5678476605947489, "learning_rate": 2.819338983500134e-05, "loss": 2.699, "step": 32828 }, { "epoch": 2.0379291079520763, "grad_norm": 0.5386293025133768, "learning_rate": 2.8190139944105437e-05, "loss": 2.6958, "step": 32829 }, { "epoch": 2.037991185051834, "grad_norm": 0.5253236862707179, "learning_rate": 2.818689016699687e-05, "loss": 2.7805, "step": 32830 }, { "epoch": 2.038053262151592, "grad_norm": 0.5260858580153418, "learning_rate": 2.818364050369259e-05, "loss": 2.5914, "step": 32831 }, { "epoch": 2.03811533925135, "grad_norm": 0.5641688650765496, "learning_rate": 2.8180390954209535e-05, "loss": 2.6965, "step": 32832 }, { "epoch": 2.038177416351108, "grad_norm": 0.5218985694999408, "learning_rate": 2.8177141518564697e-05, "loss": 2.701, "step": 32833 }, { "epoch": 2.038239493450866, "grad_norm": 0.5457529546314316, "learning_rate": 2.817389219677501e-05, "loss": 2.7603, "step": 32834 }, { "epoch": 2.0383015705506238, "grad_norm": 0.5542170988046996, "learning_rate": 2.8170642988857433e-05, "loss": 2.7006, "step": 32835 }, { "epoch": 2.0383636476503817, "grad_norm": 0.5807679310221516, "learning_rate": 2.8167393894828877e-05, "loss": 2.683, "step": 32836 }, { "epoch": 2.0384257247501396, "grad_norm": 0.5248678790465887, "learning_rate": 2.8164144914706357e-05, "loss": 2.6811, "step": 32837 }, { "epoch": 2.0384878018498975, "grad_norm": 0.4900197931106745, "learning_rate": 2.816089604850679e-05, "loss": 2.6271, "step": 32838 }, { "epoch": 2.0385498789496554, "grad_norm": 0.5783213462420973, "learning_rate": 2.8157647296247132e-05, "loss": 2.5959, "step": 32839 }, { "epoch": 2.0386119560494134, "grad_norm": 0.5386282789086723, "learning_rate": 2.8154398657944304e-05, "loss": 2.7123, "step": 32840 }, { "epoch": 2.0386740331491713, "grad_norm": 0.565779688005206, "learning_rate": 2.8151150133615313e-05, "loss": 2.5871, "step": 32841 }, { "epoch": 2.038736110248929, "grad_norm": 0.6162095107652981, "learning_rate": 2.8147901723277064e-05, "loss": 2.7628, "step": 32842 }, { "epoch": 2.038798187348687, "grad_norm": 0.5564865434447175, "learning_rate": 2.814465342694652e-05, "loss": 2.7558, "step": 32843 }, { "epoch": 2.038860264448445, "grad_norm": 0.7227374469401909, "learning_rate": 2.814140524464063e-05, "loss": 2.7552, "step": 32844 }, { "epoch": 2.038922341548203, "grad_norm": 0.5661029134962119, "learning_rate": 2.8138157176376312e-05, "loss": 2.6772, "step": 32845 }, { "epoch": 2.038984418647961, "grad_norm": 0.6216994636882465, "learning_rate": 2.8134909222170554e-05, "loss": 2.7427, "step": 32846 }, { "epoch": 2.039046495747719, "grad_norm": 0.6413496617434745, "learning_rate": 2.813166138204028e-05, "loss": 2.7618, "step": 32847 }, { "epoch": 2.0391085728474767, "grad_norm": 0.5088435343396533, "learning_rate": 2.812841365600244e-05, "loss": 2.5892, "step": 32848 }, { "epoch": 2.0391706499472346, "grad_norm": 0.6588038750755145, "learning_rate": 2.8125166044073957e-05, "loss": 2.6763, "step": 32849 }, { "epoch": 2.0392327270469925, "grad_norm": 0.5289138027085777, "learning_rate": 2.8121918546271814e-05, "loss": 2.6878, "step": 32850 }, { "epoch": 2.0392948041467505, "grad_norm": 0.6158837660484427, "learning_rate": 2.811867116261293e-05, "loss": 2.6682, "step": 32851 }, { "epoch": 2.0393568812465084, "grad_norm": 0.6024860624624457, "learning_rate": 2.8115423893114235e-05, "loss": 2.7373, "step": 32852 }, { "epoch": 2.039418958346266, "grad_norm": 0.5391931513698747, "learning_rate": 2.81121767377927e-05, "loss": 2.704, "step": 32853 }, { "epoch": 2.0394810354460238, "grad_norm": 0.5801560320700954, "learning_rate": 2.8108929696665253e-05, "loss": 2.755, "step": 32854 }, { "epoch": 2.0395431125457817, "grad_norm": 0.5495338924931218, "learning_rate": 2.810568276974882e-05, "loss": 2.6759, "step": 32855 }, { "epoch": 2.0396051896455396, "grad_norm": 0.5585054181023107, "learning_rate": 2.810243595706037e-05, "loss": 2.6405, "step": 32856 }, { "epoch": 2.0396672667452975, "grad_norm": 0.5863988967056356, "learning_rate": 2.8099189258616825e-05, "loss": 2.7191, "step": 32857 }, { "epoch": 2.0397293438450554, "grad_norm": 0.5355622081384076, "learning_rate": 2.8095942674435132e-05, "loss": 2.6995, "step": 32858 }, { "epoch": 2.0397914209448134, "grad_norm": 0.5374550833897223, "learning_rate": 2.8092696204532197e-05, "loss": 2.617, "step": 32859 }, { "epoch": 2.0398534980445713, "grad_norm": 0.5543463289056904, "learning_rate": 2.808944984892501e-05, "loss": 2.7201, "step": 32860 }, { "epoch": 2.039915575144329, "grad_norm": 0.5355334379707192, "learning_rate": 2.808620360763048e-05, "loss": 2.6619, "step": 32861 }, { "epoch": 2.039977652244087, "grad_norm": 0.5885258281455791, "learning_rate": 2.808295748066555e-05, "loss": 2.6884, "step": 32862 }, { "epoch": 2.040039729343845, "grad_norm": 0.5308293191627265, "learning_rate": 2.8079711468047125e-05, "loss": 2.7075, "step": 32863 }, { "epoch": 2.040101806443603, "grad_norm": 0.5369655624674223, "learning_rate": 2.8076465569792188e-05, "loss": 2.7045, "step": 32864 }, { "epoch": 2.040163883543361, "grad_norm": 0.567676776627213, "learning_rate": 2.8073219785917653e-05, "loss": 2.6638, "step": 32865 }, { "epoch": 2.040225960643119, "grad_norm": 0.5720151020887415, "learning_rate": 2.8069974116440457e-05, "loss": 2.6956, "step": 32866 }, { "epoch": 2.0402880377428767, "grad_norm": 0.576862813174054, "learning_rate": 2.8066728561377527e-05, "loss": 2.5345, "step": 32867 }, { "epoch": 2.0403501148426346, "grad_norm": 0.5165594765262542, "learning_rate": 2.8063483120745777e-05, "loss": 2.7051, "step": 32868 }, { "epoch": 2.0404121919423925, "grad_norm": 0.5976133892173753, "learning_rate": 2.8060237794562184e-05, "loss": 2.6388, "step": 32869 }, { "epoch": 2.0404742690421505, "grad_norm": 0.5857154171007308, "learning_rate": 2.805699258284365e-05, "loss": 2.6708, "step": 32870 }, { "epoch": 2.0405363461419084, "grad_norm": 0.5761434515714915, "learning_rate": 2.8053747485607117e-05, "loss": 2.6793, "step": 32871 }, { "epoch": 2.0405984232416663, "grad_norm": 0.61952084680508, "learning_rate": 2.8050502502869492e-05, "loss": 2.6783, "step": 32872 }, { "epoch": 2.040660500341424, "grad_norm": 0.5207040212956904, "learning_rate": 2.8047257634647737e-05, "loss": 2.7184, "step": 32873 }, { "epoch": 2.040722577441182, "grad_norm": 0.60857475750612, "learning_rate": 2.8044012880958782e-05, "loss": 2.6939, "step": 32874 }, { "epoch": 2.04078465454094, "grad_norm": 0.5475663583421658, "learning_rate": 2.8040768241819528e-05, "loss": 2.7055, "step": 32875 }, { "epoch": 2.040846731640698, "grad_norm": 0.6015482566435357, "learning_rate": 2.8037523717246907e-05, "loss": 2.742, "step": 32876 }, { "epoch": 2.0409088087404554, "grad_norm": 0.5641203430612443, "learning_rate": 2.803427930725787e-05, "loss": 2.7564, "step": 32877 }, { "epoch": 2.0409708858402134, "grad_norm": 0.6471354269005962, "learning_rate": 2.8031035011869333e-05, "loss": 2.7935, "step": 32878 }, { "epoch": 2.0410329629399713, "grad_norm": 0.549833957577332, "learning_rate": 2.8027790831098217e-05, "loss": 2.6659, "step": 32879 }, { "epoch": 2.041095040039729, "grad_norm": 0.6322477434373727, "learning_rate": 2.8024546764961447e-05, "loss": 2.6762, "step": 32880 }, { "epoch": 2.041157117139487, "grad_norm": 0.5572671944304528, "learning_rate": 2.8021302813475936e-05, "loss": 2.6647, "step": 32881 }, { "epoch": 2.041219194239245, "grad_norm": 0.6886512047862002, "learning_rate": 2.8018058976658645e-05, "loss": 2.6534, "step": 32882 }, { "epoch": 2.041281271339003, "grad_norm": 0.5295490072587377, "learning_rate": 2.8014815254526478e-05, "loss": 2.7473, "step": 32883 }, { "epoch": 2.041343348438761, "grad_norm": 0.5482899480930419, "learning_rate": 2.801157164709635e-05, "loss": 2.7063, "step": 32884 }, { "epoch": 2.041405425538519, "grad_norm": 0.5327813013512956, "learning_rate": 2.8008328154385176e-05, "loss": 2.6123, "step": 32885 }, { "epoch": 2.0414675026382767, "grad_norm": 0.49318575669718834, "learning_rate": 2.8005084776409895e-05, "loss": 2.6838, "step": 32886 }, { "epoch": 2.0415295797380346, "grad_norm": 0.6542391167953735, "learning_rate": 2.8001841513187443e-05, "loss": 2.6031, "step": 32887 }, { "epoch": 2.0415916568377925, "grad_norm": 0.5001042972504405, "learning_rate": 2.799859836473472e-05, "loss": 2.6524, "step": 32888 }, { "epoch": 2.0416537339375505, "grad_norm": 0.6020985234026249, "learning_rate": 2.799535533106865e-05, "loss": 2.7033, "step": 32889 }, { "epoch": 2.0417158110373084, "grad_norm": 0.5456568092783323, "learning_rate": 2.799211241220614e-05, "loss": 2.7955, "step": 32890 }, { "epoch": 2.0417778881370663, "grad_norm": 0.524525692671403, "learning_rate": 2.7988869608164138e-05, "loss": 2.6081, "step": 32891 }, { "epoch": 2.041839965236824, "grad_norm": 0.5586435255801663, "learning_rate": 2.7985626918959553e-05, "loss": 2.7256, "step": 32892 }, { "epoch": 2.041902042336582, "grad_norm": 0.5067059418185541, "learning_rate": 2.798238434460929e-05, "loss": 2.7161, "step": 32893 }, { "epoch": 2.04196411943634, "grad_norm": 0.5446698552839466, "learning_rate": 2.7979141885130273e-05, "loss": 2.6959, "step": 32894 }, { "epoch": 2.042026196536098, "grad_norm": 0.5209314222158615, "learning_rate": 2.7975899540539396e-05, "loss": 2.7208, "step": 32895 }, { "epoch": 2.042088273635856, "grad_norm": 0.5285250677715814, "learning_rate": 2.7972657310853624e-05, "loss": 2.7092, "step": 32896 }, { "epoch": 2.042150350735614, "grad_norm": 0.5445230590116689, "learning_rate": 2.7969415196089843e-05, "loss": 2.7044, "step": 32897 }, { "epoch": 2.0422124278353717, "grad_norm": 0.5325037802978436, "learning_rate": 2.7966173196264974e-05, "loss": 2.7024, "step": 32898 }, { "epoch": 2.0422745049351296, "grad_norm": 0.5428263525548618, "learning_rate": 2.7962931311395903e-05, "loss": 2.7662, "step": 32899 }, { "epoch": 2.0423365820348875, "grad_norm": 0.49653748319013286, "learning_rate": 2.795968954149959e-05, "loss": 2.6381, "step": 32900 }, { "epoch": 2.042398659134645, "grad_norm": 0.5688189978195402, "learning_rate": 2.795644788659293e-05, "loss": 2.6652, "step": 32901 }, { "epoch": 2.042460736234403, "grad_norm": 0.568647750634169, "learning_rate": 2.7953206346692833e-05, "loss": 2.7718, "step": 32902 }, { "epoch": 2.042522813334161, "grad_norm": 0.5585871876151074, "learning_rate": 2.794996492181621e-05, "loss": 2.6589, "step": 32903 }, { "epoch": 2.0425848904339188, "grad_norm": 0.545337969340941, "learning_rate": 2.7946723611979953e-05, "loss": 2.7142, "step": 32904 }, { "epoch": 2.0426469675336767, "grad_norm": 0.5931392339158165, "learning_rate": 2.794348241720101e-05, "loss": 2.7302, "step": 32905 }, { "epoch": 2.0427090446334346, "grad_norm": 0.469343667794143, "learning_rate": 2.7940241337496276e-05, "loss": 2.6237, "step": 32906 }, { "epoch": 2.0427711217331925, "grad_norm": 0.5734024733570218, "learning_rate": 2.793700037288266e-05, "loss": 2.6811, "step": 32907 }, { "epoch": 2.0428331988329504, "grad_norm": 0.5143882774317303, "learning_rate": 2.7933759523377045e-05, "loss": 2.6852, "step": 32908 }, { "epoch": 2.0428952759327084, "grad_norm": 0.5324438089175395, "learning_rate": 2.7930518788996386e-05, "loss": 2.6729, "step": 32909 }, { "epoch": 2.0429573530324663, "grad_norm": 0.4930394232338991, "learning_rate": 2.7927278169757565e-05, "loss": 2.7101, "step": 32910 }, { "epoch": 2.043019430132224, "grad_norm": 0.5340275242637734, "learning_rate": 2.7924037665677495e-05, "loss": 2.6993, "step": 32911 }, { "epoch": 2.043081507231982, "grad_norm": 0.49828925125430223, "learning_rate": 2.7920797276773058e-05, "loss": 2.6388, "step": 32912 }, { "epoch": 2.04314358433174, "grad_norm": 0.5257199471319532, "learning_rate": 2.79175570030612e-05, "loss": 2.6396, "step": 32913 }, { "epoch": 2.043205661431498, "grad_norm": 0.5405812912840615, "learning_rate": 2.7914316844558807e-05, "loss": 2.7456, "step": 32914 }, { "epoch": 2.043267738531256, "grad_norm": 0.502541371772759, "learning_rate": 2.7911076801282786e-05, "loss": 2.7404, "step": 32915 }, { "epoch": 2.043329815631014, "grad_norm": 0.5744997232401607, "learning_rate": 2.790783687325005e-05, "loss": 2.688, "step": 32916 }, { "epoch": 2.0433918927307717, "grad_norm": 0.5023737529142135, "learning_rate": 2.7904597060477457e-05, "loss": 2.7067, "step": 32917 }, { "epoch": 2.0434539698305296, "grad_norm": 0.49847591096685856, "learning_rate": 2.790135736298195e-05, "loss": 2.6564, "step": 32918 }, { "epoch": 2.0435160469302875, "grad_norm": 0.5264550526733992, "learning_rate": 2.7898117780780452e-05, "loss": 2.686, "step": 32919 }, { "epoch": 2.0435781240300455, "grad_norm": 0.5151300945127258, "learning_rate": 2.7894878313889833e-05, "loss": 2.7804, "step": 32920 }, { "epoch": 2.0436402011298034, "grad_norm": 0.5202038362232537, "learning_rate": 2.7891638962327e-05, "loss": 2.7208, "step": 32921 }, { "epoch": 2.0437022782295613, "grad_norm": 0.5113097441817819, "learning_rate": 2.788839972610884e-05, "loss": 2.7317, "step": 32922 }, { "epoch": 2.043764355329319, "grad_norm": 0.48462758092953084, "learning_rate": 2.7885160605252274e-05, "loss": 2.6554, "step": 32923 }, { "epoch": 2.0438264324290767, "grad_norm": 0.5230319031672968, "learning_rate": 2.7881921599774203e-05, "loss": 2.6613, "step": 32924 }, { "epoch": 2.0438885095288346, "grad_norm": 0.5607948419343235, "learning_rate": 2.7878682709691517e-05, "loss": 2.7457, "step": 32925 }, { "epoch": 2.0439505866285925, "grad_norm": 0.492351474392658, "learning_rate": 2.7875443935021085e-05, "loss": 2.7298, "step": 32926 }, { "epoch": 2.0440126637283504, "grad_norm": 0.6019131022719267, "learning_rate": 2.787220527577986e-05, "loss": 2.6814, "step": 32927 }, { "epoch": 2.0440747408281084, "grad_norm": 0.5032422622127488, "learning_rate": 2.7868966731984715e-05, "loss": 2.7296, "step": 32928 }, { "epoch": 2.0441368179278663, "grad_norm": 0.5422871105145826, "learning_rate": 2.786572830365254e-05, "loss": 2.583, "step": 32929 }, { "epoch": 2.044198895027624, "grad_norm": 0.5352656120435922, "learning_rate": 2.786248999080023e-05, "loss": 2.7634, "step": 32930 }, { "epoch": 2.044260972127382, "grad_norm": 0.47700109779483496, "learning_rate": 2.7859251793444665e-05, "loss": 2.6691, "step": 32931 }, { "epoch": 2.04432304922714, "grad_norm": 0.5272858764539522, "learning_rate": 2.785601371160278e-05, "loss": 2.6309, "step": 32932 }, { "epoch": 2.044385126326898, "grad_norm": 0.4544037008600318, "learning_rate": 2.7852775745291443e-05, "loss": 2.5958, "step": 32933 }, { "epoch": 2.044447203426656, "grad_norm": 0.5457039420054169, "learning_rate": 2.7849537894527555e-05, "loss": 2.7558, "step": 32934 }, { "epoch": 2.044509280526414, "grad_norm": 0.49145585129701197, "learning_rate": 2.7846300159327986e-05, "loss": 2.7108, "step": 32935 }, { "epoch": 2.0445713576261717, "grad_norm": 0.5286717400678211, "learning_rate": 2.7843062539709662e-05, "loss": 2.7448, "step": 32936 }, { "epoch": 2.0446334347259296, "grad_norm": 0.5461643233423112, "learning_rate": 2.7839825035689465e-05, "loss": 2.7125, "step": 32937 }, { "epoch": 2.0446955118256875, "grad_norm": 0.5145187359263187, "learning_rate": 2.783658764728428e-05, "loss": 2.6917, "step": 32938 }, { "epoch": 2.0447575889254455, "grad_norm": 0.4728643573982155, "learning_rate": 2.7833350374510996e-05, "loss": 2.6858, "step": 32939 }, { "epoch": 2.0448196660252034, "grad_norm": 0.5204915514726621, "learning_rate": 2.7830113217386485e-05, "loss": 2.7851, "step": 32940 }, { "epoch": 2.0448817431249613, "grad_norm": 0.49662828558222155, "learning_rate": 2.7826876175927673e-05, "loss": 2.7356, "step": 32941 }, { "epoch": 2.044943820224719, "grad_norm": 0.5153047549728114, "learning_rate": 2.7823639250151434e-05, "loss": 2.6679, "step": 32942 }, { "epoch": 2.045005897324477, "grad_norm": 0.5176233523779493, "learning_rate": 2.782040244007465e-05, "loss": 2.7417, "step": 32943 }, { "epoch": 2.045067974424235, "grad_norm": 0.5011204444143424, "learning_rate": 2.781716574571419e-05, "loss": 2.7401, "step": 32944 }, { "epoch": 2.045130051523993, "grad_norm": 0.4731365789762818, "learning_rate": 2.7813929167086984e-05, "loss": 2.627, "step": 32945 }, { "epoch": 2.045192128623751, "grad_norm": 0.4995714377523992, "learning_rate": 2.7810692704209897e-05, "loss": 2.6784, "step": 32946 }, { "epoch": 2.045254205723509, "grad_norm": 0.49983206551367604, "learning_rate": 2.7807456357099808e-05, "loss": 2.5592, "step": 32947 }, { "epoch": 2.0453162828232667, "grad_norm": 0.49364626132677947, "learning_rate": 2.780422012577359e-05, "loss": 2.7201, "step": 32948 }, { "epoch": 2.045378359923024, "grad_norm": 0.49349197212654305, "learning_rate": 2.7800984010248167e-05, "loss": 2.7044, "step": 32949 }, { "epoch": 2.045440437022782, "grad_norm": 0.5289282271581945, "learning_rate": 2.779774801054039e-05, "loss": 2.6971, "step": 32950 }, { "epoch": 2.04550251412254, "grad_norm": 0.46411925845366636, "learning_rate": 2.779451212666714e-05, "loss": 2.8281, "step": 32951 }, { "epoch": 2.045564591222298, "grad_norm": 0.5335691544098939, "learning_rate": 2.7791276358645324e-05, "loss": 2.6495, "step": 32952 }, { "epoch": 2.045626668322056, "grad_norm": 0.46975419246757777, "learning_rate": 2.7788040706491813e-05, "loss": 2.7421, "step": 32953 }, { "epoch": 2.045688745421814, "grad_norm": 0.4765332009736286, "learning_rate": 2.778480517022347e-05, "loss": 2.7962, "step": 32954 }, { "epoch": 2.0457508225215717, "grad_norm": 0.4963479329531988, "learning_rate": 2.7781569749857205e-05, "loss": 2.6526, "step": 32955 }, { "epoch": 2.0458128996213296, "grad_norm": 0.4553766213052324, "learning_rate": 2.777833444540989e-05, "loss": 2.6905, "step": 32956 }, { "epoch": 2.0458749767210875, "grad_norm": 0.5267920539880884, "learning_rate": 2.7775099256898396e-05, "loss": 2.7148, "step": 32957 }, { "epoch": 2.0459370538208455, "grad_norm": 0.5121867536780254, "learning_rate": 2.777186418433958e-05, "loss": 2.6369, "step": 32958 }, { "epoch": 2.0459991309206034, "grad_norm": 0.47228399703242174, "learning_rate": 2.7768629227750374e-05, "loss": 2.6691, "step": 32959 }, { "epoch": 2.0460612080203613, "grad_norm": 0.5217822078054739, "learning_rate": 2.776539438714762e-05, "loss": 2.7149, "step": 32960 }, { "epoch": 2.046123285120119, "grad_norm": 0.4931583366439337, "learning_rate": 2.7762159662548203e-05, "loss": 2.7136, "step": 32961 }, { "epoch": 2.046185362219877, "grad_norm": 0.48062568520644694, "learning_rate": 2.7758925053968976e-05, "loss": 2.6567, "step": 32962 }, { "epoch": 2.046247439319635, "grad_norm": 0.5449486455870252, "learning_rate": 2.775569056142686e-05, "loss": 2.7487, "step": 32963 }, { "epoch": 2.046309516419393, "grad_norm": 0.5313811420811665, "learning_rate": 2.7752456184938702e-05, "loss": 2.7308, "step": 32964 }, { "epoch": 2.046371593519151, "grad_norm": 0.5006067647482224, "learning_rate": 2.7749221924521384e-05, "loss": 2.6071, "step": 32965 }, { "epoch": 2.046433670618909, "grad_norm": 0.4986798770387415, "learning_rate": 2.7745987780191774e-05, "loss": 2.6851, "step": 32966 }, { "epoch": 2.0464957477186667, "grad_norm": 0.5008524542071396, "learning_rate": 2.7742753751966726e-05, "loss": 2.7529, "step": 32967 }, { "epoch": 2.0465578248184246, "grad_norm": 0.49255531433676447, "learning_rate": 2.773951983986316e-05, "loss": 2.7617, "step": 32968 }, { "epoch": 2.0466199019181825, "grad_norm": 0.49815300203445123, "learning_rate": 2.773628604389792e-05, "loss": 2.6923, "step": 32969 }, { "epoch": 2.0466819790179405, "grad_norm": 0.5064829514092323, "learning_rate": 2.7733052364087886e-05, "loss": 2.6509, "step": 32970 }, { "epoch": 2.0467440561176984, "grad_norm": 0.4934510255376789, "learning_rate": 2.7729818800449892e-05, "loss": 2.6542, "step": 32971 }, { "epoch": 2.046806133217456, "grad_norm": 0.4789382469602522, "learning_rate": 2.7726585353000868e-05, "loss": 2.7257, "step": 32972 }, { "epoch": 2.0468682103172138, "grad_norm": 0.4707110867282033, "learning_rate": 2.7723352021757652e-05, "loss": 2.5995, "step": 32973 }, { "epoch": 2.0469302874169717, "grad_norm": 0.569777067533244, "learning_rate": 2.7720118806737115e-05, "loss": 2.679, "step": 32974 }, { "epoch": 2.0469923645167296, "grad_norm": 0.4734384354182404, "learning_rate": 2.7716885707956115e-05, "loss": 2.6272, "step": 32975 }, { "epoch": 2.0470544416164875, "grad_norm": 0.5606226853202744, "learning_rate": 2.771365272543155e-05, "loss": 2.7536, "step": 32976 }, { "epoch": 2.0471165187162454, "grad_norm": 0.5034019253058553, "learning_rate": 2.771041985918026e-05, "loss": 2.714, "step": 32977 }, { "epoch": 2.0471785958160034, "grad_norm": 0.5488395825476359, "learning_rate": 2.7707187109219124e-05, "loss": 2.6417, "step": 32978 }, { "epoch": 2.0472406729157613, "grad_norm": 0.5309230696056997, "learning_rate": 2.7703954475565007e-05, "loss": 2.6937, "step": 32979 }, { "epoch": 2.047302750015519, "grad_norm": 0.5141038999728264, "learning_rate": 2.7700721958234754e-05, "loss": 2.6268, "step": 32980 }, { "epoch": 2.047364827115277, "grad_norm": 0.5493438849463911, "learning_rate": 2.7697489557245267e-05, "loss": 2.6693, "step": 32981 }, { "epoch": 2.047426904215035, "grad_norm": 0.5215482636937827, "learning_rate": 2.769425727261339e-05, "loss": 2.6734, "step": 32982 }, { "epoch": 2.047488981314793, "grad_norm": 0.5054333339824667, "learning_rate": 2.7691025104355965e-05, "loss": 2.671, "step": 32983 }, { "epoch": 2.047551058414551, "grad_norm": 0.5020596877211289, "learning_rate": 2.7687793052489897e-05, "loss": 2.7759, "step": 32984 }, { "epoch": 2.047613135514309, "grad_norm": 0.4889240611298335, "learning_rate": 2.7684561117032014e-05, "loss": 2.6651, "step": 32985 }, { "epoch": 2.0476752126140667, "grad_norm": 0.5660665611583872, "learning_rate": 2.7681329297999213e-05, "loss": 2.7173, "step": 32986 }, { "epoch": 2.0477372897138246, "grad_norm": 0.4844707886518188, "learning_rate": 2.7678097595408336e-05, "loss": 2.6278, "step": 32987 }, { "epoch": 2.0477993668135825, "grad_norm": 0.5473747966824346, "learning_rate": 2.7674866009276236e-05, "loss": 2.7909, "step": 32988 }, { "epoch": 2.0478614439133405, "grad_norm": 0.5406242083946121, "learning_rate": 2.7671634539619785e-05, "loss": 2.6378, "step": 32989 }, { "epoch": 2.0479235210130984, "grad_norm": 0.5131790640038548, "learning_rate": 2.7668403186455817e-05, "loss": 2.5965, "step": 32990 }, { "epoch": 2.0479855981128563, "grad_norm": 0.6065706624145654, "learning_rate": 2.7665171949801227e-05, "loss": 2.7691, "step": 32991 }, { "epoch": 2.048047675212614, "grad_norm": 0.5473308026140664, "learning_rate": 2.766194082967286e-05, "loss": 2.6284, "step": 32992 }, { "epoch": 2.048109752312372, "grad_norm": 0.5888037702785679, "learning_rate": 2.765870982608757e-05, "loss": 2.7264, "step": 32993 }, { "epoch": 2.04817182941213, "grad_norm": 0.5314217738525818, "learning_rate": 2.7655478939062196e-05, "loss": 2.7099, "step": 32994 }, { "epoch": 2.048233906511888, "grad_norm": 0.5647032292533941, "learning_rate": 2.7652248168613625e-05, "loss": 2.6375, "step": 32995 }, { "epoch": 2.048295983611646, "grad_norm": 0.511722855879058, "learning_rate": 2.7649017514758706e-05, "loss": 2.6447, "step": 32996 }, { "epoch": 2.0483580607114034, "grad_norm": 0.5064736543338975, "learning_rate": 2.7645786977514287e-05, "loss": 2.7679, "step": 32997 }, { "epoch": 2.0484201378111613, "grad_norm": 0.5300834694848048, "learning_rate": 2.764255655689721e-05, "loss": 2.7566, "step": 32998 }, { "epoch": 2.048482214910919, "grad_norm": 0.5199422716228336, "learning_rate": 2.763932625292436e-05, "loss": 2.6966, "step": 32999 }, { "epoch": 2.048544292010677, "grad_norm": 0.4960430599958367, "learning_rate": 2.763609606561257e-05, "loss": 2.6101, "step": 33000 }, { "epoch": 2.048606369110435, "grad_norm": 0.5097084159465369, "learning_rate": 2.7632865994978695e-05, "loss": 2.6047, "step": 33001 }, { "epoch": 2.048668446210193, "grad_norm": 0.5328894730172249, "learning_rate": 2.7629636041039592e-05, "loss": 2.6665, "step": 33002 }, { "epoch": 2.048730523309951, "grad_norm": 0.497354182817943, "learning_rate": 2.7626406203812082e-05, "loss": 2.6414, "step": 33003 }, { "epoch": 2.048792600409709, "grad_norm": 0.5202429922604376, "learning_rate": 2.7623176483313074e-05, "loss": 2.7427, "step": 33004 }, { "epoch": 2.0488546775094667, "grad_norm": 0.5000833203511151, "learning_rate": 2.7619946879559377e-05, "loss": 2.6976, "step": 33005 }, { "epoch": 2.0489167546092246, "grad_norm": 0.522089061122121, "learning_rate": 2.7616717392567854e-05, "loss": 2.7292, "step": 33006 }, { "epoch": 2.0489788317089825, "grad_norm": 0.5054976300298145, "learning_rate": 2.7613488022355327e-05, "loss": 2.6523, "step": 33007 }, { "epoch": 2.0490409088087405, "grad_norm": 0.48350152351845493, "learning_rate": 2.761025876893869e-05, "loss": 2.6848, "step": 33008 }, { "epoch": 2.0491029859084984, "grad_norm": 0.5022149201119452, "learning_rate": 2.760702963233477e-05, "loss": 2.6136, "step": 33009 }, { "epoch": 2.0491650630082563, "grad_norm": 0.5523720352290108, "learning_rate": 2.7603800612560405e-05, "loss": 2.7098, "step": 33010 }, { "epoch": 2.049227140108014, "grad_norm": 0.4730065206487544, "learning_rate": 2.7600571709632438e-05, "loss": 2.6853, "step": 33011 }, { "epoch": 2.049289217207772, "grad_norm": 0.5616917631539953, "learning_rate": 2.7597342923567737e-05, "loss": 2.6929, "step": 33012 }, { "epoch": 2.04935129430753, "grad_norm": 0.5263531741398338, "learning_rate": 2.7594114254383143e-05, "loss": 2.6989, "step": 33013 }, { "epoch": 2.049413371407288, "grad_norm": 0.5392965555003586, "learning_rate": 2.7590885702095492e-05, "loss": 2.7775, "step": 33014 }, { "epoch": 2.049475448507046, "grad_norm": 0.4885587437809703, "learning_rate": 2.7587657266721635e-05, "loss": 2.6865, "step": 33015 }, { "epoch": 2.049537525606804, "grad_norm": 0.526841524945448, "learning_rate": 2.7584428948278385e-05, "loss": 2.7626, "step": 33016 }, { "epoch": 2.0495996027065617, "grad_norm": 0.512511127630244, "learning_rate": 2.758120074678261e-05, "loss": 2.6931, "step": 33017 }, { "epoch": 2.0496616798063196, "grad_norm": 0.49858450201693394, "learning_rate": 2.7577972662251178e-05, "loss": 2.7739, "step": 33018 }, { "epoch": 2.0497237569060776, "grad_norm": 0.49091997876580096, "learning_rate": 2.7574744694700905e-05, "loss": 2.7216, "step": 33019 }, { "epoch": 2.049785834005835, "grad_norm": 0.4868645983749121, "learning_rate": 2.7571516844148626e-05, "loss": 2.7268, "step": 33020 }, { "epoch": 2.049847911105593, "grad_norm": 0.5011540148816567, "learning_rate": 2.7568289110611178e-05, "loss": 2.6853, "step": 33021 }, { "epoch": 2.049909988205351, "grad_norm": 0.5603901713842888, "learning_rate": 2.7565061494105426e-05, "loss": 2.715, "step": 33022 }, { "epoch": 2.049972065305109, "grad_norm": 0.5136052367640843, "learning_rate": 2.756183399464819e-05, "loss": 2.7492, "step": 33023 }, { "epoch": 2.0500341424048667, "grad_norm": 0.4997531758484681, "learning_rate": 2.7558606612256323e-05, "loss": 2.6926, "step": 33024 }, { "epoch": 2.0500962195046246, "grad_norm": 0.5313599697061138, "learning_rate": 2.7555379346946646e-05, "loss": 2.8176, "step": 33025 }, { "epoch": 2.0501582966043825, "grad_norm": 0.47916063705741774, "learning_rate": 2.7552152198735985e-05, "loss": 2.7364, "step": 33026 }, { "epoch": 2.0502203737041405, "grad_norm": 0.48134627493909543, "learning_rate": 2.7548925167641215e-05, "loss": 2.6771, "step": 33027 }, { "epoch": 2.0502824508038984, "grad_norm": 0.5037041192186789, "learning_rate": 2.7545698253679152e-05, "loss": 2.7632, "step": 33028 }, { "epoch": 2.0503445279036563, "grad_norm": 0.48505899372856187, "learning_rate": 2.7542471456866637e-05, "loss": 2.6803, "step": 33029 }, { "epoch": 2.050406605003414, "grad_norm": 0.4872644595128724, "learning_rate": 2.753924477722048e-05, "loss": 2.6598, "step": 33030 }, { "epoch": 2.050468682103172, "grad_norm": 0.4977747753574067, "learning_rate": 2.7536018214757555e-05, "loss": 2.5669, "step": 33031 }, { "epoch": 2.05053075920293, "grad_norm": 0.5081297836923115, "learning_rate": 2.753279176949467e-05, "loss": 2.8002, "step": 33032 }, { "epoch": 2.050592836302688, "grad_norm": 0.4885499138803614, "learning_rate": 2.752956544144867e-05, "loss": 2.6687, "step": 33033 }, { "epoch": 2.050654913402446, "grad_norm": 0.530512690795913, "learning_rate": 2.752633923063636e-05, "loss": 2.6041, "step": 33034 }, { "epoch": 2.050716990502204, "grad_norm": 0.541304886261729, "learning_rate": 2.7523113137074613e-05, "loss": 2.671, "step": 33035 }, { "epoch": 2.0507790676019617, "grad_norm": 0.49649435674085995, "learning_rate": 2.7519887160780245e-05, "loss": 2.6089, "step": 33036 }, { "epoch": 2.0508411447017196, "grad_norm": 0.5388212908244623, "learning_rate": 2.751666130177008e-05, "loss": 2.7773, "step": 33037 }, { "epoch": 2.0509032218014775, "grad_norm": 0.56563813315472, "learning_rate": 2.7513435560060952e-05, "loss": 2.761, "step": 33038 }, { "epoch": 2.0509652989012355, "grad_norm": 0.5420197351001784, "learning_rate": 2.7510209935669667e-05, "loss": 2.8084, "step": 33039 }, { "epoch": 2.0510273760009934, "grad_norm": 0.5657246665187936, "learning_rate": 2.7506984428613104e-05, "loss": 2.6836, "step": 33040 }, { "epoch": 2.0510894531007513, "grad_norm": 0.5251585720910964, "learning_rate": 2.750375903890805e-05, "loss": 2.7695, "step": 33041 }, { "epoch": 2.051151530200509, "grad_norm": 0.543973391234507, "learning_rate": 2.7500533766571355e-05, "loss": 2.5764, "step": 33042 }, { "epoch": 2.051213607300267, "grad_norm": 0.545016646262214, "learning_rate": 2.749730861161982e-05, "loss": 2.7176, "step": 33043 }, { "epoch": 2.0512756844000246, "grad_norm": 0.5258980891873946, "learning_rate": 2.7494083574070307e-05, "loss": 2.6944, "step": 33044 }, { "epoch": 2.0513377614997825, "grad_norm": 0.48417079374170646, "learning_rate": 2.7490858653939623e-05, "loss": 2.6849, "step": 33045 }, { "epoch": 2.0513998385995404, "grad_norm": 0.5660402891525825, "learning_rate": 2.7487633851244586e-05, "loss": 2.6272, "step": 33046 }, { "epoch": 2.0514619156992984, "grad_norm": 0.5087861337412707, "learning_rate": 2.7484409166002017e-05, "loss": 2.6866, "step": 33047 }, { "epoch": 2.0515239927990563, "grad_norm": 0.5239452063380422, "learning_rate": 2.7481184598228772e-05, "loss": 2.6977, "step": 33048 }, { "epoch": 2.051586069898814, "grad_norm": 0.5092418215717136, "learning_rate": 2.7477960147941627e-05, "loss": 2.635, "step": 33049 }, { "epoch": 2.051648146998572, "grad_norm": 0.5427458875211278, "learning_rate": 2.7474735815157453e-05, "loss": 2.7667, "step": 33050 }, { "epoch": 2.05171022409833, "grad_norm": 0.5326715691294713, "learning_rate": 2.7471511599893053e-05, "loss": 2.7247, "step": 33051 }, { "epoch": 2.051772301198088, "grad_norm": 0.4968034968719946, "learning_rate": 2.7468287502165245e-05, "loss": 2.815, "step": 33052 }, { "epoch": 2.051834378297846, "grad_norm": 0.5633755862832007, "learning_rate": 2.746506352199083e-05, "loss": 2.7237, "step": 33053 }, { "epoch": 2.051896455397604, "grad_norm": 0.5397972959192763, "learning_rate": 2.7461839659386674e-05, "loss": 2.6101, "step": 33054 }, { "epoch": 2.0519585324973617, "grad_norm": 0.5558410255475595, "learning_rate": 2.745861591436957e-05, "loss": 2.6621, "step": 33055 }, { "epoch": 2.0520206095971196, "grad_norm": 0.5936453375675104, "learning_rate": 2.7455392286956334e-05, "loss": 2.6322, "step": 33056 }, { "epoch": 2.0520826866968775, "grad_norm": 0.5722516679851976, "learning_rate": 2.745216877716377e-05, "loss": 2.7292, "step": 33057 }, { "epoch": 2.0521447637966355, "grad_norm": 0.5998682463548541, "learning_rate": 2.744894538500874e-05, "loss": 2.6994, "step": 33058 }, { "epoch": 2.0522068408963934, "grad_norm": 0.48922044074498305, "learning_rate": 2.7445722110508032e-05, "loss": 2.7078, "step": 33059 }, { "epoch": 2.0522689179961513, "grad_norm": 0.5334499472544153, "learning_rate": 2.744249895367847e-05, "loss": 2.7949, "step": 33060 }, { "epoch": 2.052330995095909, "grad_norm": 0.5010321216374621, "learning_rate": 2.7439275914536843e-05, "loss": 2.6745, "step": 33061 }, { "epoch": 2.052393072195667, "grad_norm": 0.484156036107048, "learning_rate": 2.7436052993100013e-05, "loss": 2.6136, "step": 33062 }, { "epoch": 2.052455149295425, "grad_norm": 0.5600223546164345, "learning_rate": 2.7432830189384774e-05, "loss": 2.6959, "step": 33063 }, { "epoch": 2.052517226395183, "grad_norm": 0.5212746244784388, "learning_rate": 2.742960750340794e-05, "loss": 2.7409, "step": 33064 }, { "epoch": 2.052579303494941, "grad_norm": 0.48095932393568397, "learning_rate": 2.7426384935186317e-05, "loss": 2.7644, "step": 33065 }, { "epoch": 2.052641380594699, "grad_norm": 0.5278037713855319, "learning_rate": 2.742316248473671e-05, "loss": 2.7194, "step": 33066 }, { "epoch": 2.0527034576944567, "grad_norm": 0.4886667484590862, "learning_rate": 2.7419940152075957e-05, "loss": 2.6975, "step": 33067 }, { "epoch": 2.052765534794214, "grad_norm": 0.5011412589636773, "learning_rate": 2.741671793722086e-05, "loss": 2.6918, "step": 33068 }, { "epoch": 2.052827611893972, "grad_norm": 0.5315526212551378, "learning_rate": 2.741349584018823e-05, "loss": 2.7437, "step": 33069 }, { "epoch": 2.05288968899373, "grad_norm": 0.47704634571648935, "learning_rate": 2.7410273860994856e-05, "loss": 2.7154, "step": 33070 }, { "epoch": 2.052951766093488, "grad_norm": 0.5401836015586442, "learning_rate": 2.7407051999657584e-05, "loss": 2.763, "step": 33071 }, { "epoch": 2.053013843193246, "grad_norm": 0.468864744765289, "learning_rate": 2.7403830256193208e-05, "loss": 2.6361, "step": 33072 }, { "epoch": 2.053075920293004, "grad_norm": 0.49340426194751263, "learning_rate": 2.740060863061854e-05, "loss": 2.6873, "step": 33073 }, { "epoch": 2.0531379973927617, "grad_norm": 0.5591622507439258, "learning_rate": 2.739738712295038e-05, "loss": 2.727, "step": 33074 }, { "epoch": 2.0532000744925196, "grad_norm": 0.4888826862336688, "learning_rate": 2.7394165733205513e-05, "loss": 2.6655, "step": 33075 }, { "epoch": 2.0532621515922775, "grad_norm": 0.57182896850934, "learning_rate": 2.739094446140079e-05, "loss": 2.7822, "step": 33076 }, { "epoch": 2.0533242286920355, "grad_norm": 0.5843344312243349, "learning_rate": 2.7387723307553003e-05, "loss": 2.6504, "step": 33077 }, { "epoch": 2.0533863057917934, "grad_norm": 0.45261010150361886, "learning_rate": 2.7384502271678957e-05, "loss": 2.6606, "step": 33078 }, { "epoch": 2.0534483828915513, "grad_norm": 0.57100408244135, "learning_rate": 2.738128135379543e-05, "loss": 2.7539, "step": 33079 }, { "epoch": 2.053510459991309, "grad_norm": 0.5125844490441414, "learning_rate": 2.7378060553919265e-05, "loss": 2.6859, "step": 33080 }, { "epoch": 2.053572537091067, "grad_norm": 0.5163043200014923, "learning_rate": 2.7374839872067254e-05, "loss": 2.703, "step": 33081 }, { "epoch": 2.053634614190825, "grad_norm": 0.567343469197869, "learning_rate": 2.7371619308256173e-05, "loss": 2.7291, "step": 33082 }, { "epoch": 2.053696691290583, "grad_norm": 0.5559849675602612, "learning_rate": 2.736839886250287e-05, "loss": 2.6741, "step": 33083 }, { "epoch": 2.053758768390341, "grad_norm": 0.5270463740793798, "learning_rate": 2.7365178534824104e-05, "loss": 2.6404, "step": 33084 }, { "epoch": 2.053820845490099, "grad_norm": 0.5641652465400496, "learning_rate": 2.7361958325236713e-05, "loss": 2.6899, "step": 33085 }, { "epoch": 2.0538829225898567, "grad_norm": 0.5637642639960778, "learning_rate": 2.7358738233757485e-05, "loss": 2.7019, "step": 33086 }, { "epoch": 2.0539449996896146, "grad_norm": 0.5108455458555988, "learning_rate": 2.7355518260403212e-05, "loss": 2.7218, "step": 33087 }, { "epoch": 2.0540070767893726, "grad_norm": 0.6033912209495191, "learning_rate": 2.7352298405190703e-05, "loss": 2.7202, "step": 33088 }, { "epoch": 2.0540691538891305, "grad_norm": 0.5810777419188841, "learning_rate": 2.734907866813673e-05, "loss": 2.6745, "step": 33089 }, { "epoch": 2.0541312309888884, "grad_norm": 0.5571447666060386, "learning_rate": 2.7345859049258127e-05, "loss": 2.6564, "step": 33090 }, { "epoch": 2.0541933080886463, "grad_norm": 0.5888944148258237, "learning_rate": 2.734263954857168e-05, "loss": 2.7122, "step": 33091 }, { "epoch": 2.054255385188404, "grad_norm": 0.5575739214090492, "learning_rate": 2.733942016609418e-05, "loss": 2.6295, "step": 33092 }, { "epoch": 2.0543174622881617, "grad_norm": 0.49924507370200727, "learning_rate": 2.7336200901842414e-05, "loss": 2.6922, "step": 33093 }, { "epoch": 2.0543795393879196, "grad_norm": 0.5758409487289283, "learning_rate": 2.7332981755833197e-05, "loss": 2.6964, "step": 33094 }, { "epoch": 2.0544416164876775, "grad_norm": 0.547787014835753, "learning_rate": 2.732976272808333e-05, "loss": 2.6964, "step": 33095 }, { "epoch": 2.0545036935874355, "grad_norm": 0.5024713297868514, "learning_rate": 2.7326543818609585e-05, "loss": 2.6506, "step": 33096 }, { "epoch": 2.0545657706871934, "grad_norm": 0.546233291152197, "learning_rate": 2.7323325027428748e-05, "loss": 2.7342, "step": 33097 }, { "epoch": 2.0546278477869513, "grad_norm": 0.5290140050663887, "learning_rate": 2.732010635455765e-05, "loss": 2.7361, "step": 33098 }, { "epoch": 2.054689924886709, "grad_norm": 0.5545524379006976, "learning_rate": 2.7316887800013065e-05, "loss": 2.6359, "step": 33099 }, { "epoch": 2.054752001986467, "grad_norm": 0.5332457417095864, "learning_rate": 2.731366936381178e-05, "loss": 2.6282, "step": 33100 }, { "epoch": 2.054814079086225, "grad_norm": 0.5414488947643681, "learning_rate": 2.7310451045970587e-05, "loss": 2.6705, "step": 33101 }, { "epoch": 2.054876156185983, "grad_norm": 0.523077200251933, "learning_rate": 2.7307232846506258e-05, "loss": 2.7566, "step": 33102 }, { "epoch": 2.054938233285741, "grad_norm": 0.5051401983767457, "learning_rate": 2.730401476543563e-05, "loss": 2.7855, "step": 33103 }, { "epoch": 2.055000310385499, "grad_norm": 0.5243007033027415, "learning_rate": 2.7300796802775463e-05, "loss": 2.7397, "step": 33104 }, { "epoch": 2.0550623874852567, "grad_norm": 0.5056405790967924, "learning_rate": 2.7297578958542552e-05, "loss": 2.7084, "step": 33105 }, { "epoch": 2.0551244645850146, "grad_norm": 0.49002491705172563, "learning_rate": 2.7294361232753664e-05, "loss": 2.6783, "step": 33106 }, { "epoch": 2.0551865416847725, "grad_norm": 0.5459477874887106, "learning_rate": 2.729114362542562e-05, "loss": 2.8141, "step": 33107 }, { "epoch": 2.0552486187845305, "grad_norm": 0.48019838867592396, "learning_rate": 2.7287926136575197e-05, "loss": 2.6638, "step": 33108 }, { "epoch": 2.0553106958842884, "grad_norm": 0.4978861830037501, "learning_rate": 2.7284708766219175e-05, "loss": 2.6717, "step": 33109 }, { "epoch": 2.0553727729840463, "grad_norm": 0.5807576165407999, "learning_rate": 2.7281491514374345e-05, "loss": 2.6713, "step": 33110 }, { "epoch": 2.055434850083804, "grad_norm": 0.5314698886200816, "learning_rate": 2.7278274381057468e-05, "loss": 2.749, "step": 33111 }, { "epoch": 2.055496927183562, "grad_norm": 0.5043430357474815, "learning_rate": 2.7275057366285363e-05, "loss": 2.6743, "step": 33112 }, { "epoch": 2.05555900428332, "grad_norm": 0.5697997152491133, "learning_rate": 2.7271840470074807e-05, "loss": 2.6517, "step": 33113 }, { "epoch": 2.055621081383078, "grad_norm": 0.524591312963845, "learning_rate": 2.726862369244255e-05, "loss": 2.6223, "step": 33114 }, { "epoch": 2.055683158482836, "grad_norm": 0.587470706249961, "learning_rate": 2.7265407033405432e-05, "loss": 2.7419, "step": 33115 }, { "epoch": 2.0557452355825934, "grad_norm": 0.5521177754964789, "learning_rate": 2.7262190492980176e-05, "loss": 2.7833, "step": 33116 }, { "epoch": 2.0558073126823513, "grad_norm": 0.5727085573629156, "learning_rate": 2.7258974071183612e-05, "loss": 2.7746, "step": 33117 }, { "epoch": 2.055869389782109, "grad_norm": 0.47552588551717967, "learning_rate": 2.725575776803251e-05, "loss": 2.7365, "step": 33118 }, { "epoch": 2.055931466881867, "grad_norm": 0.6146495400883446, "learning_rate": 2.725254158354363e-05, "loss": 2.7537, "step": 33119 }, { "epoch": 2.055993543981625, "grad_norm": 0.5193319870711389, "learning_rate": 2.724932551773375e-05, "loss": 2.6852, "step": 33120 }, { "epoch": 2.056055621081383, "grad_norm": 0.5527040453026868, "learning_rate": 2.7246109570619677e-05, "loss": 2.8133, "step": 33121 }, { "epoch": 2.056117698181141, "grad_norm": 0.5362472551004366, "learning_rate": 2.7242893742218174e-05, "loss": 2.7338, "step": 33122 }, { "epoch": 2.056179775280899, "grad_norm": 0.6141722214687843, "learning_rate": 2.7239678032546024e-05, "loss": 2.6872, "step": 33123 }, { "epoch": 2.0562418523806567, "grad_norm": 0.47107803482995664, "learning_rate": 2.7236462441619992e-05, "loss": 2.6634, "step": 33124 }, { "epoch": 2.0563039294804146, "grad_norm": 0.533913191967077, "learning_rate": 2.723324696945685e-05, "loss": 2.6814, "step": 33125 }, { "epoch": 2.0563660065801725, "grad_norm": 0.5807884311514292, "learning_rate": 2.723003161607339e-05, "loss": 2.7239, "step": 33126 }, { "epoch": 2.0564280836799305, "grad_norm": 0.5801003448649044, "learning_rate": 2.7226816381486396e-05, "loss": 2.7212, "step": 33127 }, { "epoch": 2.0564901607796884, "grad_norm": 0.5909994815252906, "learning_rate": 2.7223601265712627e-05, "loss": 2.7047, "step": 33128 }, { "epoch": 2.0565522378794463, "grad_norm": 0.5187805316167062, "learning_rate": 2.722038626876884e-05, "loss": 2.7064, "step": 33129 }, { "epoch": 2.056614314979204, "grad_norm": 0.4793612890011034, "learning_rate": 2.7217171390671848e-05, "loss": 2.6178, "step": 33130 }, { "epoch": 2.056676392078962, "grad_norm": 0.5377049975591641, "learning_rate": 2.7213956631438404e-05, "loss": 2.6999, "step": 33131 }, { "epoch": 2.05673846917872, "grad_norm": 0.5309244448632362, "learning_rate": 2.7210741991085275e-05, "loss": 2.6069, "step": 33132 }, { "epoch": 2.056800546278478, "grad_norm": 0.5012615798506829, "learning_rate": 2.7207527469629222e-05, "loss": 2.6168, "step": 33133 }, { "epoch": 2.056862623378236, "grad_norm": 0.5170476534222197, "learning_rate": 2.720431306708705e-05, "loss": 2.755, "step": 33134 }, { "epoch": 2.056924700477994, "grad_norm": 0.5298273447703681, "learning_rate": 2.720109878347551e-05, "loss": 2.738, "step": 33135 }, { "epoch": 2.0569867775777517, "grad_norm": 0.5299640968595796, "learning_rate": 2.7197884618811377e-05, "loss": 2.6969, "step": 33136 }, { "epoch": 2.0570488546775096, "grad_norm": 0.5130678403848049, "learning_rate": 2.7194670573111403e-05, "loss": 2.6607, "step": 33137 }, { "epoch": 2.0571109317772676, "grad_norm": 0.5152389931056067, "learning_rate": 2.7191456646392362e-05, "loss": 2.735, "step": 33138 }, { "epoch": 2.0571730088770255, "grad_norm": 0.5090790432934326, "learning_rate": 2.718824283867104e-05, "loss": 2.7206, "step": 33139 }, { "epoch": 2.057235085976783, "grad_norm": 0.5424155016984422, "learning_rate": 2.7185029149964196e-05, "loss": 2.6716, "step": 33140 }, { "epoch": 2.057297163076541, "grad_norm": 0.5794304722604693, "learning_rate": 2.7181815580288593e-05, "loss": 2.74, "step": 33141 }, { "epoch": 2.057359240176299, "grad_norm": 0.5415323867394083, "learning_rate": 2.717860212966098e-05, "loss": 2.6662, "step": 33142 }, { "epoch": 2.0574213172760567, "grad_norm": 0.5558199270033808, "learning_rate": 2.7175388798098154e-05, "loss": 2.7104, "step": 33143 }, { "epoch": 2.0574833943758146, "grad_norm": 0.5494624914257525, "learning_rate": 2.717217558561687e-05, "loss": 2.7162, "step": 33144 }, { "epoch": 2.0575454714755725, "grad_norm": 0.5598354277443366, "learning_rate": 2.7168962492233885e-05, "loss": 2.6304, "step": 33145 }, { "epoch": 2.0576075485753305, "grad_norm": 0.6198666043752447, "learning_rate": 2.716574951796595e-05, "loss": 2.7023, "step": 33146 }, { "epoch": 2.0576696256750884, "grad_norm": 0.47851822521515314, "learning_rate": 2.7162536662829842e-05, "loss": 2.657, "step": 33147 }, { "epoch": 2.0577317027748463, "grad_norm": 0.5562054991334007, "learning_rate": 2.715932392684234e-05, "loss": 2.7051, "step": 33148 }, { "epoch": 2.057793779874604, "grad_norm": 0.576066010658052, "learning_rate": 2.7156111310020194e-05, "loss": 2.709, "step": 33149 }, { "epoch": 2.057855856974362, "grad_norm": 0.5126741532092041, "learning_rate": 2.7152898812380155e-05, "loss": 2.7012, "step": 33150 }, { "epoch": 2.05791793407412, "grad_norm": 0.6021177775904303, "learning_rate": 2.714968643393899e-05, "loss": 2.8085, "step": 33151 }, { "epoch": 2.057980011173878, "grad_norm": 0.5180465018487218, "learning_rate": 2.7146474174713445e-05, "loss": 2.6899, "step": 33152 }, { "epoch": 2.058042088273636, "grad_norm": 0.5190742031112424, "learning_rate": 2.7143262034720308e-05, "loss": 2.6543, "step": 33153 }, { "epoch": 2.058104165373394, "grad_norm": 0.5377550717535403, "learning_rate": 2.7140050013976326e-05, "loss": 2.752, "step": 33154 }, { "epoch": 2.0581662424731517, "grad_norm": 0.5120434204275576, "learning_rate": 2.7136838112498242e-05, "loss": 2.7087, "step": 33155 }, { "epoch": 2.0582283195729096, "grad_norm": 0.5051296966020984, "learning_rate": 2.7133626330302814e-05, "loss": 2.7433, "step": 33156 }, { "epoch": 2.0582903966726676, "grad_norm": 0.5067439648156405, "learning_rate": 2.7130414667406823e-05, "loss": 2.6928, "step": 33157 }, { "epoch": 2.0583524737724255, "grad_norm": 0.558223788584363, "learning_rate": 2.7127203123827016e-05, "loss": 2.7833, "step": 33158 }, { "epoch": 2.0584145508721834, "grad_norm": 0.5046073117637002, "learning_rate": 2.7123991699580136e-05, "loss": 2.6777, "step": 33159 }, { "epoch": 2.0584766279719413, "grad_norm": 0.5007857764362363, "learning_rate": 2.7120780394682942e-05, "loss": 2.7034, "step": 33160 }, { "epoch": 2.0585387050716992, "grad_norm": 0.5416253850562679, "learning_rate": 2.7117569209152182e-05, "loss": 2.6032, "step": 33161 }, { "epoch": 2.058600782171457, "grad_norm": 0.520835701619168, "learning_rate": 2.711435814300463e-05, "loss": 2.729, "step": 33162 }, { "epoch": 2.058662859271215, "grad_norm": 0.5163017513484831, "learning_rate": 2.711114719625703e-05, "loss": 2.7287, "step": 33163 }, { "epoch": 2.0587249363709725, "grad_norm": 0.4935176676976182, "learning_rate": 2.7107936368926136e-05, "loss": 2.6755, "step": 33164 }, { "epoch": 2.0587870134707305, "grad_norm": 0.5354037860948817, "learning_rate": 2.710472566102867e-05, "loss": 2.6936, "step": 33165 }, { "epoch": 2.0588490905704884, "grad_norm": 0.5070752587638097, "learning_rate": 2.710151507258143e-05, "loss": 2.7625, "step": 33166 }, { "epoch": 2.0589111676702463, "grad_norm": 0.5436645824383456, "learning_rate": 2.7098304603601142e-05, "loss": 2.6931, "step": 33167 }, { "epoch": 2.058973244770004, "grad_norm": 0.5347542749709306, "learning_rate": 2.709509425410456e-05, "loss": 2.6798, "step": 33168 }, { "epoch": 2.059035321869762, "grad_norm": 0.5026137757906335, "learning_rate": 2.709188402410841e-05, "loss": 2.6637, "step": 33169 }, { "epoch": 2.05909739896952, "grad_norm": 0.532504367944775, "learning_rate": 2.708867391362948e-05, "loss": 2.7494, "step": 33170 }, { "epoch": 2.059159476069278, "grad_norm": 0.5834793918405922, "learning_rate": 2.7085463922684508e-05, "loss": 2.6453, "step": 33171 }, { "epoch": 2.059221553169036, "grad_norm": 0.5597282572889737, "learning_rate": 2.7082254051290228e-05, "loss": 2.6458, "step": 33172 }, { "epoch": 2.059283630268794, "grad_norm": 0.5680581307164991, "learning_rate": 2.7079044299463396e-05, "loss": 2.6878, "step": 33173 }, { "epoch": 2.0593457073685517, "grad_norm": 0.5309485253260849, "learning_rate": 2.7075834667220733e-05, "loss": 2.702, "step": 33174 }, { "epoch": 2.0594077844683096, "grad_norm": 0.5482367862860721, "learning_rate": 2.7072625154579024e-05, "loss": 2.667, "step": 33175 }, { "epoch": 2.0594698615680676, "grad_norm": 0.559521498223626, "learning_rate": 2.7069415761554993e-05, "loss": 2.6517, "step": 33176 }, { "epoch": 2.0595319386678255, "grad_norm": 0.5017561767361395, "learning_rate": 2.706620648816539e-05, "loss": 2.6581, "step": 33177 }, { "epoch": 2.0595940157675834, "grad_norm": 0.5625349934607394, "learning_rate": 2.706299733442693e-05, "loss": 2.6774, "step": 33178 }, { "epoch": 2.0596560928673413, "grad_norm": 0.5134086079284141, "learning_rate": 2.7059788300356408e-05, "loss": 2.7052, "step": 33179 }, { "epoch": 2.059718169967099, "grad_norm": 0.48419687626078056, "learning_rate": 2.7056579385970515e-05, "loss": 2.7565, "step": 33180 }, { "epoch": 2.059780247066857, "grad_norm": 0.5252782708611817, "learning_rate": 2.7053370591286042e-05, "loss": 2.6998, "step": 33181 }, { "epoch": 2.059842324166615, "grad_norm": 0.5019371534841996, "learning_rate": 2.70501619163197e-05, "loss": 2.676, "step": 33182 }, { "epoch": 2.059904401266373, "grad_norm": 0.5144390570615031, "learning_rate": 2.7046953361088216e-05, "loss": 2.7599, "step": 33183 }, { "epoch": 2.059966478366131, "grad_norm": 0.5364454848099561, "learning_rate": 2.7043744925608372e-05, "loss": 2.6953, "step": 33184 }, { "epoch": 2.060028555465889, "grad_norm": 0.4783651465653873, "learning_rate": 2.7040536609896876e-05, "loss": 2.7601, "step": 33185 }, { "epoch": 2.0600906325656467, "grad_norm": 0.4869424700722663, "learning_rate": 2.703732841397048e-05, "loss": 2.6886, "step": 33186 }, { "epoch": 2.0601527096654046, "grad_norm": 0.5376488490594425, "learning_rate": 2.7034120337845915e-05, "loss": 2.7192, "step": 33187 }, { "epoch": 2.060214786765162, "grad_norm": 0.4999178580521999, "learning_rate": 2.7030912381539898e-05, "loss": 2.7119, "step": 33188 }, { "epoch": 2.06027686386492, "grad_norm": 0.5037986581365774, "learning_rate": 2.702770454506921e-05, "loss": 2.7143, "step": 33189 }, { "epoch": 2.060338940964678, "grad_norm": 0.4810968960556635, "learning_rate": 2.7024496828450563e-05, "loss": 2.7513, "step": 33190 }, { "epoch": 2.060401018064436, "grad_norm": 0.5190773608924145, "learning_rate": 2.7021289231700687e-05, "loss": 2.7024, "step": 33191 }, { "epoch": 2.060463095164194, "grad_norm": 0.5025624536258794, "learning_rate": 2.701808175483631e-05, "loss": 2.716, "step": 33192 }, { "epoch": 2.0605251722639517, "grad_norm": 0.48283755968853787, "learning_rate": 2.70148743978742e-05, "loss": 2.6227, "step": 33193 }, { "epoch": 2.0605872493637096, "grad_norm": 0.5316103666665339, "learning_rate": 2.7011667160831066e-05, "loss": 2.6821, "step": 33194 }, { "epoch": 2.0606493264634675, "grad_norm": 0.5021579191068689, "learning_rate": 2.700846004372365e-05, "loss": 2.6924, "step": 33195 }, { "epoch": 2.0607114035632255, "grad_norm": 0.5295290842290573, "learning_rate": 2.700525304656868e-05, "loss": 2.7337, "step": 33196 }, { "epoch": 2.0607734806629834, "grad_norm": 0.5191581935507497, "learning_rate": 2.700204616938286e-05, "loss": 2.6915, "step": 33197 }, { "epoch": 2.0608355577627413, "grad_norm": 0.5000515553597756, "learning_rate": 2.6998839412182976e-05, "loss": 2.7232, "step": 33198 }, { "epoch": 2.060897634862499, "grad_norm": 0.5578677120719926, "learning_rate": 2.6995632774985724e-05, "loss": 2.6968, "step": 33199 }, { "epoch": 2.060959711962257, "grad_norm": 0.5279524786598973, "learning_rate": 2.6992426257807844e-05, "loss": 2.6327, "step": 33200 }, { "epoch": 2.061021789062015, "grad_norm": 0.5480917609737825, "learning_rate": 2.698921986066604e-05, "loss": 2.7719, "step": 33201 }, { "epoch": 2.061083866161773, "grad_norm": 0.5437375944191293, "learning_rate": 2.698601358357708e-05, "loss": 2.7428, "step": 33202 }, { "epoch": 2.061145943261531, "grad_norm": 0.5075656290792818, "learning_rate": 2.6982807426557675e-05, "loss": 2.7126, "step": 33203 }, { "epoch": 2.061208020361289, "grad_norm": 0.5135857977743264, "learning_rate": 2.6979601389624552e-05, "loss": 2.6842, "step": 33204 }, { "epoch": 2.0612700974610467, "grad_norm": 0.528910816347591, "learning_rate": 2.6976395472794426e-05, "loss": 2.6188, "step": 33205 }, { "epoch": 2.0613321745608046, "grad_norm": 0.505284251247394, "learning_rate": 2.6973189676084044e-05, "loss": 2.7526, "step": 33206 }, { "epoch": 2.0613942516605626, "grad_norm": 0.5303166549241111, "learning_rate": 2.6969983999510117e-05, "loss": 2.6595, "step": 33207 }, { "epoch": 2.0614563287603205, "grad_norm": 0.5701722011702697, "learning_rate": 2.696677844308938e-05, "loss": 2.7133, "step": 33208 }, { "epoch": 2.0615184058600784, "grad_norm": 0.5260309621143522, "learning_rate": 2.6963573006838554e-05, "loss": 2.6638, "step": 33209 }, { "epoch": 2.0615804829598363, "grad_norm": 0.54912453246189, "learning_rate": 2.6960367690774335e-05, "loss": 2.6456, "step": 33210 }, { "epoch": 2.0616425600595942, "grad_norm": 0.5116691565323596, "learning_rate": 2.6957162494913497e-05, "loss": 2.6943, "step": 33211 }, { "epoch": 2.0617046371593517, "grad_norm": 0.5356677040783597, "learning_rate": 2.695395741927273e-05, "loss": 2.56, "step": 33212 }, { "epoch": 2.0617667142591096, "grad_norm": 0.5800632770320627, "learning_rate": 2.6950752463868745e-05, "loss": 2.6709, "step": 33213 }, { "epoch": 2.0618287913588675, "grad_norm": 0.5068711579126636, "learning_rate": 2.69475476287183e-05, "loss": 2.7036, "step": 33214 }, { "epoch": 2.0618908684586255, "grad_norm": 0.5828949836677408, "learning_rate": 2.6944342913838074e-05, "loss": 2.6195, "step": 33215 }, { "epoch": 2.0619529455583834, "grad_norm": 0.5640471691468028, "learning_rate": 2.694113831924483e-05, "loss": 2.7573, "step": 33216 }, { "epoch": 2.0620150226581413, "grad_norm": 0.501146194858562, "learning_rate": 2.693793384495526e-05, "loss": 2.6872, "step": 33217 }, { "epoch": 2.062077099757899, "grad_norm": 0.5774374043200021, "learning_rate": 2.6934729490986088e-05, "loss": 2.7064, "step": 33218 }, { "epoch": 2.062139176857657, "grad_norm": 0.5491802674183347, "learning_rate": 2.6931525257354017e-05, "loss": 2.7781, "step": 33219 }, { "epoch": 2.062201253957415, "grad_norm": 0.5886714477132501, "learning_rate": 2.6928321144075797e-05, "loss": 2.6132, "step": 33220 }, { "epoch": 2.062263331057173, "grad_norm": 0.5315775142211042, "learning_rate": 2.692511715116812e-05, "loss": 2.5721, "step": 33221 }, { "epoch": 2.062325408156931, "grad_norm": 0.6150957950557463, "learning_rate": 2.692191327864772e-05, "loss": 2.814, "step": 33222 }, { "epoch": 2.062387485256689, "grad_norm": 0.5601236693032958, "learning_rate": 2.6918709526531294e-05, "loss": 2.6712, "step": 33223 }, { "epoch": 2.0624495623564467, "grad_norm": 0.6111421637400273, "learning_rate": 2.6915505894835547e-05, "loss": 2.7961, "step": 33224 }, { "epoch": 2.0625116394562046, "grad_norm": 0.5571205078347875, "learning_rate": 2.6912302383577226e-05, "loss": 2.6162, "step": 33225 }, { "epoch": 2.0625737165559626, "grad_norm": 0.6088922006318988, "learning_rate": 2.690909899277303e-05, "loss": 2.7923, "step": 33226 }, { "epoch": 2.0626357936557205, "grad_norm": 0.5247168492645193, "learning_rate": 2.6905895722439668e-05, "loss": 2.6221, "step": 33227 }, { "epoch": 2.0626978707554784, "grad_norm": 0.595256149949143, "learning_rate": 2.6902692572593836e-05, "loss": 2.7043, "step": 33228 }, { "epoch": 2.0627599478552363, "grad_norm": 0.5507889875585434, "learning_rate": 2.6899489543252287e-05, "loss": 2.6319, "step": 33229 }, { "epoch": 2.0628220249549942, "grad_norm": 0.6245550001952513, "learning_rate": 2.68962866344317e-05, "loss": 2.6477, "step": 33230 }, { "epoch": 2.062884102054752, "grad_norm": 0.545651565988692, "learning_rate": 2.6893083846148803e-05, "loss": 2.7789, "step": 33231 }, { "epoch": 2.06294617915451, "grad_norm": 0.5941519129857707, "learning_rate": 2.6889881178420273e-05, "loss": 2.777, "step": 33232 }, { "epoch": 2.063008256254268, "grad_norm": 0.5973756418138892, "learning_rate": 2.6886678631262862e-05, "loss": 2.7611, "step": 33233 }, { "epoch": 2.063070333354026, "grad_norm": 0.5604901966323185, "learning_rate": 2.6883476204693266e-05, "loss": 2.6608, "step": 33234 }, { "epoch": 2.0631324104537834, "grad_norm": 0.5828680385277779, "learning_rate": 2.688027389872818e-05, "loss": 2.751, "step": 33235 }, { "epoch": 2.0631944875535413, "grad_norm": 0.607431712178244, "learning_rate": 2.6877071713384316e-05, "loss": 2.7808, "step": 33236 }, { "epoch": 2.063256564653299, "grad_norm": 0.5735586355460967, "learning_rate": 2.687386964867837e-05, "loss": 2.6883, "step": 33237 }, { "epoch": 2.063318641753057, "grad_norm": 0.562903630293706, "learning_rate": 2.6870667704627074e-05, "loss": 2.7052, "step": 33238 }, { "epoch": 2.063380718852815, "grad_norm": 0.6130180279957363, "learning_rate": 2.6867465881247123e-05, "loss": 2.798, "step": 33239 }, { "epoch": 2.063442795952573, "grad_norm": 0.5158177506810407, "learning_rate": 2.6864264178555214e-05, "loss": 2.6548, "step": 33240 }, { "epoch": 2.063504873052331, "grad_norm": 0.5214770312988741, "learning_rate": 2.6861062596568038e-05, "loss": 2.7316, "step": 33241 }, { "epoch": 2.063566950152089, "grad_norm": 0.6323986584064047, "learning_rate": 2.685786113530233e-05, "loss": 2.7256, "step": 33242 }, { "epoch": 2.0636290272518467, "grad_norm": 0.5559598273028231, "learning_rate": 2.6854659794774782e-05, "loss": 2.6449, "step": 33243 }, { "epoch": 2.0636911043516046, "grad_norm": 0.6272817208109647, "learning_rate": 2.6851458575002096e-05, "loss": 2.7375, "step": 33244 }, { "epoch": 2.0637531814513626, "grad_norm": 0.618149815978273, "learning_rate": 2.684825747600095e-05, "loss": 2.7355, "step": 33245 }, { "epoch": 2.0638152585511205, "grad_norm": 0.538322216850197, "learning_rate": 2.6845056497788084e-05, "loss": 2.7468, "step": 33246 }, { "epoch": 2.0638773356508784, "grad_norm": 0.5270271763897193, "learning_rate": 2.684185564038016e-05, "loss": 2.649, "step": 33247 }, { "epoch": 2.0639394127506363, "grad_norm": 0.5144854292868468, "learning_rate": 2.6838654903793914e-05, "loss": 2.7569, "step": 33248 }, { "epoch": 2.064001489850394, "grad_norm": 0.502236965095742, "learning_rate": 2.683545428804603e-05, "loss": 2.6541, "step": 33249 }, { "epoch": 2.064063566950152, "grad_norm": 0.4792154449707132, "learning_rate": 2.6832253793153207e-05, "loss": 2.674, "step": 33250 }, { "epoch": 2.06412564404991, "grad_norm": 0.48953813148042663, "learning_rate": 2.682905341913211e-05, "loss": 2.6781, "step": 33251 }, { "epoch": 2.064187721149668, "grad_norm": 0.5338448376344435, "learning_rate": 2.6825853165999492e-05, "loss": 2.6587, "step": 33252 }, { "epoch": 2.064249798249426, "grad_norm": 0.5012445458094026, "learning_rate": 2.6822653033772028e-05, "loss": 2.6256, "step": 33253 }, { "epoch": 2.064311875349184, "grad_norm": 0.5017804988188596, "learning_rate": 2.6819453022466405e-05, "loss": 2.6491, "step": 33254 }, { "epoch": 2.0643739524489417, "grad_norm": 0.4717166167467168, "learning_rate": 2.6816253132099305e-05, "loss": 2.7274, "step": 33255 }, { "epoch": 2.0644360295486996, "grad_norm": 0.5302373828187174, "learning_rate": 2.6813053362687464e-05, "loss": 2.6984, "step": 33256 }, { "epoch": 2.0644981066484576, "grad_norm": 0.4957449430197195, "learning_rate": 2.6809853714247546e-05, "loss": 2.7528, "step": 33257 }, { "epoch": 2.0645601837482155, "grad_norm": 0.4956821175671306, "learning_rate": 2.6806654186796253e-05, "loss": 2.7437, "step": 33258 }, { "epoch": 2.0646222608479734, "grad_norm": 0.5621589315385356, "learning_rate": 2.6803454780350278e-05, "loss": 2.7554, "step": 33259 }, { "epoch": 2.064684337947731, "grad_norm": 0.49349579182311853, "learning_rate": 2.6800255494926286e-05, "loss": 2.709, "step": 33260 }, { "epoch": 2.064746415047489, "grad_norm": 0.5488354556860009, "learning_rate": 2.6797056330541015e-05, "loss": 2.654, "step": 33261 }, { "epoch": 2.0648084921472467, "grad_norm": 0.49650203994550873, "learning_rate": 2.6793857287211132e-05, "loss": 2.7578, "step": 33262 }, { "epoch": 2.0648705692470046, "grad_norm": 0.5193000508477157, "learning_rate": 2.6790658364953337e-05, "loss": 2.7119, "step": 33263 }, { "epoch": 2.0649326463467625, "grad_norm": 0.5110303018734123, "learning_rate": 2.6787459563784278e-05, "loss": 2.74, "step": 33264 }, { "epoch": 2.0649947234465205, "grad_norm": 0.5109646314597652, "learning_rate": 2.6784260883720706e-05, "loss": 2.7624, "step": 33265 }, { "epoch": 2.0650568005462784, "grad_norm": 0.518903309791641, "learning_rate": 2.6781062324779282e-05, "loss": 2.6707, "step": 33266 }, { "epoch": 2.0651188776460363, "grad_norm": 0.553135918116169, "learning_rate": 2.677786388697669e-05, "loss": 2.7275, "step": 33267 }, { "epoch": 2.065180954745794, "grad_norm": 0.47751771844075663, "learning_rate": 2.6774665570329598e-05, "loss": 2.7914, "step": 33268 }, { "epoch": 2.065243031845552, "grad_norm": 0.5469551027761053, "learning_rate": 2.6771467374854735e-05, "loss": 2.6236, "step": 33269 }, { "epoch": 2.06530510894531, "grad_norm": 0.5386104623422207, "learning_rate": 2.676826930056877e-05, "loss": 2.6457, "step": 33270 }, { "epoch": 2.065367186045068, "grad_norm": 0.501030337178492, "learning_rate": 2.6765071347488378e-05, "loss": 2.6712, "step": 33271 }, { "epoch": 2.065429263144826, "grad_norm": 0.5502994134182015, "learning_rate": 2.6761873515630254e-05, "loss": 2.7034, "step": 33272 }, { "epoch": 2.065491340244584, "grad_norm": 0.5378529897694847, "learning_rate": 2.6758675805011053e-05, "loss": 2.664, "step": 33273 }, { "epoch": 2.0655534173443417, "grad_norm": 0.5198145502693753, "learning_rate": 2.6755478215647502e-05, "loss": 2.6629, "step": 33274 }, { "epoch": 2.0656154944440996, "grad_norm": 0.5114603045154301, "learning_rate": 2.675228074755627e-05, "loss": 2.5566, "step": 33275 }, { "epoch": 2.0656775715438576, "grad_norm": 0.5095292928377925, "learning_rate": 2.6749083400754026e-05, "loss": 2.6129, "step": 33276 }, { "epoch": 2.0657396486436155, "grad_norm": 0.5731939656187564, "learning_rate": 2.6745886175257438e-05, "loss": 2.7245, "step": 33277 }, { "epoch": 2.0658017257433734, "grad_norm": 0.4871865781098771, "learning_rate": 2.6742689071083216e-05, "loss": 2.6494, "step": 33278 }, { "epoch": 2.0658638028431313, "grad_norm": 0.5602471750010293, "learning_rate": 2.673949208824804e-05, "loss": 2.5921, "step": 33279 }, { "epoch": 2.0659258799428892, "grad_norm": 0.5013545285489189, "learning_rate": 2.6736295226768582e-05, "loss": 2.7495, "step": 33280 }, { "epoch": 2.065987957042647, "grad_norm": 0.567520019959548, "learning_rate": 2.6733098486661524e-05, "loss": 2.796, "step": 33281 }, { "epoch": 2.066050034142405, "grad_norm": 0.5059958019911731, "learning_rate": 2.672990186794352e-05, "loss": 2.8235, "step": 33282 }, { "epoch": 2.0661121112421625, "grad_norm": 0.5279502207001511, "learning_rate": 2.672670537063129e-05, "loss": 2.7264, "step": 33283 }, { "epoch": 2.0661741883419205, "grad_norm": 0.5707001087440613, "learning_rate": 2.6723508994741474e-05, "loss": 2.7343, "step": 33284 }, { "epoch": 2.0662362654416784, "grad_norm": 0.5309261007808492, "learning_rate": 2.6720312740290775e-05, "loss": 2.7032, "step": 33285 }, { "epoch": 2.0662983425414363, "grad_norm": 0.5515904535661137, "learning_rate": 2.671711660729585e-05, "loss": 2.6629, "step": 33286 }, { "epoch": 2.066360419641194, "grad_norm": 0.5317715440895444, "learning_rate": 2.6713920595773363e-05, "loss": 2.6686, "step": 33287 }, { "epoch": 2.066422496740952, "grad_norm": 0.5960795979550989, "learning_rate": 2.671072470574002e-05, "loss": 2.7544, "step": 33288 }, { "epoch": 2.06648457384071, "grad_norm": 0.5088414551452942, "learning_rate": 2.670752893721249e-05, "loss": 2.6472, "step": 33289 }, { "epoch": 2.066546650940468, "grad_norm": 0.6620419673308781, "learning_rate": 2.6704333290207428e-05, "loss": 2.7143, "step": 33290 }, { "epoch": 2.066608728040226, "grad_norm": 0.5382127837334524, "learning_rate": 2.670113776474149e-05, "loss": 2.6669, "step": 33291 }, { "epoch": 2.066670805139984, "grad_norm": 0.5627471857547085, "learning_rate": 2.6697942360831403e-05, "loss": 2.6589, "step": 33292 }, { "epoch": 2.0667328822397417, "grad_norm": 0.5980396910671354, "learning_rate": 2.6694747078493797e-05, "loss": 2.6645, "step": 33293 }, { "epoch": 2.0667949593394996, "grad_norm": 0.5002621916925281, "learning_rate": 2.669155191774536e-05, "loss": 2.6691, "step": 33294 }, { "epoch": 2.0668570364392576, "grad_norm": 0.5459769914096868, "learning_rate": 2.668835687860276e-05, "loss": 2.6901, "step": 33295 }, { "epoch": 2.0669191135390155, "grad_norm": 0.5954655908497289, "learning_rate": 2.6685161961082633e-05, "loss": 2.7614, "step": 33296 }, { "epoch": 2.0669811906387734, "grad_norm": 0.5015873448288108, "learning_rate": 2.6681967165201694e-05, "loss": 2.6879, "step": 33297 }, { "epoch": 2.0670432677385313, "grad_norm": 0.5409160756540995, "learning_rate": 2.66787724909766e-05, "loss": 2.6898, "step": 33298 }, { "epoch": 2.0671053448382892, "grad_norm": 0.4652043554708537, "learning_rate": 2.6675577938424007e-05, "loss": 2.7415, "step": 33299 }, { "epoch": 2.067167421938047, "grad_norm": 0.5612220286937284, "learning_rate": 2.6672383507560573e-05, "loss": 2.7346, "step": 33300 }, { "epoch": 2.067229499037805, "grad_norm": 0.482286102092546, "learning_rate": 2.666918919840299e-05, "loss": 2.7273, "step": 33301 }, { "epoch": 2.067291576137563, "grad_norm": 0.5598242218589323, "learning_rate": 2.666599501096791e-05, "loss": 2.7039, "step": 33302 }, { "epoch": 2.067353653237321, "grad_norm": 0.5742703109739757, "learning_rate": 2.6662800945272e-05, "loss": 2.6586, "step": 33303 }, { "epoch": 2.067415730337079, "grad_norm": 0.5342071637718363, "learning_rate": 2.6659607001331906e-05, "loss": 2.7289, "step": 33304 }, { "epoch": 2.0674778074368367, "grad_norm": 0.5746663224185317, "learning_rate": 2.6656413179164318e-05, "loss": 2.6747, "step": 33305 }, { "epoch": 2.0675398845365947, "grad_norm": 0.5564500716793532, "learning_rate": 2.66532194787859e-05, "loss": 2.7201, "step": 33306 }, { "epoch": 2.0676019616363526, "grad_norm": 0.49973878891884893, "learning_rate": 2.6650025900213303e-05, "loss": 2.6729, "step": 33307 }, { "epoch": 2.06766403873611, "grad_norm": 0.5495949272203453, "learning_rate": 2.664683244346319e-05, "loss": 2.6314, "step": 33308 }, { "epoch": 2.067726115835868, "grad_norm": 0.5448433057139417, "learning_rate": 2.66436391085522e-05, "loss": 2.5587, "step": 33309 }, { "epoch": 2.067788192935626, "grad_norm": 0.5360158712844939, "learning_rate": 2.6640445895497028e-05, "loss": 2.7285, "step": 33310 }, { "epoch": 2.067850270035384, "grad_norm": 0.5319816375768657, "learning_rate": 2.6637252804314306e-05, "loss": 2.619, "step": 33311 }, { "epoch": 2.0679123471351417, "grad_norm": 0.5413273488249893, "learning_rate": 2.663405983502073e-05, "loss": 2.7878, "step": 33312 }, { "epoch": 2.0679744242348996, "grad_norm": 0.5537931189530084, "learning_rate": 2.663086698763293e-05, "loss": 2.72, "step": 33313 }, { "epoch": 2.0680365013346576, "grad_norm": 0.5428956071726859, "learning_rate": 2.6627674262167558e-05, "loss": 2.7134, "step": 33314 }, { "epoch": 2.0680985784344155, "grad_norm": 0.5547298898752236, "learning_rate": 2.6624481658641297e-05, "loss": 2.7961, "step": 33315 }, { "epoch": 2.0681606555341734, "grad_norm": 0.5633716452519637, "learning_rate": 2.6621289177070786e-05, "loss": 2.6327, "step": 33316 }, { "epoch": 2.0682227326339313, "grad_norm": 0.5138983861658485, "learning_rate": 2.6618096817472694e-05, "loss": 2.7191, "step": 33317 }, { "epoch": 2.0682848097336892, "grad_norm": 0.5824117760706962, "learning_rate": 2.661490457986364e-05, "loss": 2.6385, "step": 33318 }, { "epoch": 2.068346886833447, "grad_norm": 0.5592744321299384, "learning_rate": 2.661171246426033e-05, "loss": 2.7263, "step": 33319 }, { "epoch": 2.068408963933205, "grad_norm": 0.5360893134587029, "learning_rate": 2.6608520470679393e-05, "loss": 2.667, "step": 33320 }, { "epoch": 2.068471041032963, "grad_norm": 0.6174698437342659, "learning_rate": 2.660532859913748e-05, "loss": 2.649, "step": 33321 }, { "epoch": 2.068533118132721, "grad_norm": 0.4956913014017132, "learning_rate": 2.660213684965125e-05, "loss": 2.7065, "step": 33322 }, { "epoch": 2.068595195232479, "grad_norm": 0.5674307404069396, "learning_rate": 2.6598945222237336e-05, "loss": 2.6714, "step": 33323 }, { "epoch": 2.0686572723322367, "grad_norm": 0.5678532237605729, "learning_rate": 2.659575371691242e-05, "loss": 2.6839, "step": 33324 }, { "epoch": 2.0687193494319946, "grad_norm": 0.507285475513609, "learning_rate": 2.659256233369314e-05, "loss": 2.7584, "step": 33325 }, { "epoch": 2.0687814265317526, "grad_norm": 0.5247654936067486, "learning_rate": 2.658937107259615e-05, "loss": 2.7348, "step": 33326 }, { "epoch": 2.0688435036315105, "grad_norm": 0.5759234911352801, "learning_rate": 2.6586179933638067e-05, "loss": 2.738, "step": 33327 }, { "epoch": 2.0689055807312684, "grad_norm": 0.517503812241449, "learning_rate": 2.6582988916835584e-05, "loss": 2.6171, "step": 33328 }, { "epoch": 2.0689676578310263, "grad_norm": 0.5041024173698286, "learning_rate": 2.657979802220534e-05, "loss": 2.691, "step": 33329 }, { "epoch": 2.0690297349307842, "grad_norm": 0.5448382460434051, "learning_rate": 2.6576607249763973e-05, "loss": 2.7262, "step": 33330 }, { "epoch": 2.0690918120305417, "grad_norm": 0.4557015238053136, "learning_rate": 2.6573416599528134e-05, "loss": 2.6358, "step": 33331 }, { "epoch": 2.0691538891302996, "grad_norm": 0.5132410541745178, "learning_rate": 2.6570226071514442e-05, "loss": 2.6847, "step": 33332 }, { "epoch": 2.0692159662300575, "grad_norm": 0.4731915650349954, "learning_rate": 2.6567035665739593e-05, "loss": 2.6849, "step": 33333 }, { "epoch": 2.0692780433298155, "grad_norm": 0.5010374154950124, "learning_rate": 2.6563845382220203e-05, "loss": 2.7066, "step": 33334 }, { "epoch": 2.0693401204295734, "grad_norm": 0.4741931235916532, "learning_rate": 2.6560655220972923e-05, "loss": 2.692, "step": 33335 }, { "epoch": 2.0694021975293313, "grad_norm": 0.5075206798417888, "learning_rate": 2.655746518201438e-05, "loss": 2.6642, "step": 33336 }, { "epoch": 2.069464274629089, "grad_norm": 0.4876200815396221, "learning_rate": 2.655427526536125e-05, "loss": 2.6439, "step": 33337 }, { "epoch": 2.069526351728847, "grad_norm": 0.45403880023999593, "learning_rate": 2.6551085471030153e-05, "loss": 2.6509, "step": 33338 }, { "epoch": 2.069588428828605, "grad_norm": 0.5549408777209114, "learning_rate": 2.6547895799037736e-05, "loss": 2.6935, "step": 33339 }, { "epoch": 2.069650505928363, "grad_norm": 0.5144465591602561, "learning_rate": 2.6544706249400618e-05, "loss": 2.6814, "step": 33340 }, { "epoch": 2.069712583028121, "grad_norm": 0.4916793076727468, "learning_rate": 2.654151682213549e-05, "loss": 2.653, "step": 33341 }, { "epoch": 2.069774660127879, "grad_norm": 0.4934315022399394, "learning_rate": 2.653832751725896e-05, "loss": 2.7577, "step": 33342 }, { "epoch": 2.0698367372276367, "grad_norm": 0.5288837692914984, "learning_rate": 2.653513833478767e-05, "loss": 2.6782, "step": 33343 }, { "epoch": 2.0698988143273946, "grad_norm": 0.5319860072231751, "learning_rate": 2.6531949274738236e-05, "loss": 2.7407, "step": 33344 }, { "epoch": 2.0699608914271526, "grad_norm": 0.4804732469797967, "learning_rate": 2.6528760337127346e-05, "loss": 2.6606, "step": 33345 }, { "epoch": 2.0700229685269105, "grad_norm": 0.5399861922088232, "learning_rate": 2.652557152197159e-05, "loss": 2.6124, "step": 33346 }, { "epoch": 2.0700850456266684, "grad_norm": 0.5305591186045405, "learning_rate": 2.652238282928764e-05, "loss": 2.6965, "step": 33347 }, { "epoch": 2.0701471227264263, "grad_norm": 0.5432348354064551, "learning_rate": 2.6519194259092127e-05, "loss": 2.7515, "step": 33348 }, { "epoch": 2.0702091998261842, "grad_norm": 0.46696394161252325, "learning_rate": 2.6516005811401668e-05, "loss": 2.7954, "step": 33349 }, { "epoch": 2.070271276925942, "grad_norm": 0.5130149208412337, "learning_rate": 2.6512817486232898e-05, "loss": 2.7441, "step": 33350 }, { "epoch": 2.0703333540257, "grad_norm": 0.5301394354742174, "learning_rate": 2.6509629283602477e-05, "loss": 2.6742, "step": 33351 }, { "epoch": 2.070395431125458, "grad_norm": 0.5351612619875361, "learning_rate": 2.650644120352702e-05, "loss": 2.7214, "step": 33352 }, { "epoch": 2.070457508225216, "grad_norm": 0.526421963419537, "learning_rate": 2.6503253246023165e-05, "loss": 2.6879, "step": 33353 }, { "epoch": 2.070519585324974, "grad_norm": 0.5614726433378268, "learning_rate": 2.6500065411107517e-05, "loss": 2.7961, "step": 33354 }, { "epoch": 2.0705816624247317, "grad_norm": 0.48843457675676555, "learning_rate": 2.6496877698796758e-05, "loss": 2.6364, "step": 33355 }, { "epoch": 2.070643739524489, "grad_norm": 0.4809729713857962, "learning_rate": 2.649369010910749e-05, "loss": 2.6773, "step": 33356 }, { "epoch": 2.070705816624247, "grad_norm": 0.516721369042001, "learning_rate": 2.6490502642056346e-05, "loss": 2.6931, "step": 33357 }, { "epoch": 2.070767893724005, "grad_norm": 0.5278377620869368, "learning_rate": 2.6487315297659965e-05, "loss": 2.7222, "step": 33358 }, { "epoch": 2.070829970823763, "grad_norm": 0.4679911192577123, "learning_rate": 2.6484128075934938e-05, "loss": 2.687, "step": 33359 }, { "epoch": 2.070892047923521, "grad_norm": 0.5449345084746045, "learning_rate": 2.648094097689795e-05, "loss": 2.7637, "step": 33360 }, { "epoch": 2.070954125023279, "grad_norm": 0.49551380230535286, "learning_rate": 2.64777540005656e-05, "loss": 2.7067, "step": 33361 }, { "epoch": 2.0710162021230367, "grad_norm": 0.5142537871310809, "learning_rate": 2.647456714695452e-05, "loss": 2.6664, "step": 33362 }, { "epoch": 2.0710782792227946, "grad_norm": 0.557213899647047, "learning_rate": 2.647138041608131e-05, "loss": 2.7369, "step": 33363 }, { "epoch": 2.0711403563225526, "grad_norm": 0.5210014707650715, "learning_rate": 2.6468193807962644e-05, "loss": 2.7249, "step": 33364 }, { "epoch": 2.0712024334223105, "grad_norm": 0.5725292601041984, "learning_rate": 2.646500732261512e-05, "loss": 2.7358, "step": 33365 }, { "epoch": 2.0712645105220684, "grad_norm": 0.5096438616905886, "learning_rate": 2.646182096005537e-05, "loss": 2.7346, "step": 33366 }, { "epoch": 2.0713265876218263, "grad_norm": 0.5197233547612015, "learning_rate": 2.645863472029999e-05, "loss": 2.6904, "step": 33367 }, { "epoch": 2.0713886647215842, "grad_norm": 0.5444778219142802, "learning_rate": 2.645544860336565e-05, "loss": 2.6357, "step": 33368 }, { "epoch": 2.071450741821342, "grad_norm": 0.5342863767875151, "learning_rate": 2.6452262609268953e-05, "loss": 2.7697, "step": 33369 }, { "epoch": 2.0715128189211, "grad_norm": 0.5260275061435664, "learning_rate": 2.6449076738026514e-05, "loss": 2.7482, "step": 33370 }, { "epoch": 2.071574896020858, "grad_norm": 0.5409126321419505, "learning_rate": 2.644589098965496e-05, "loss": 2.6958, "step": 33371 }, { "epoch": 2.071636973120616, "grad_norm": 0.5326259689250621, "learning_rate": 2.644270536417089e-05, "loss": 2.6841, "step": 33372 }, { "epoch": 2.071699050220374, "grad_norm": 0.5101705236614195, "learning_rate": 2.6439519861590965e-05, "loss": 2.7017, "step": 33373 }, { "epoch": 2.0717611273201317, "grad_norm": 0.5215340922797811, "learning_rate": 2.6436334481931785e-05, "loss": 2.7217, "step": 33374 }, { "epoch": 2.0718232044198897, "grad_norm": 0.539660108718273, "learning_rate": 2.6433149225209967e-05, "loss": 2.6652, "step": 33375 }, { "epoch": 2.0718852815196476, "grad_norm": 0.48774937462583307, "learning_rate": 2.6429964091442117e-05, "loss": 2.7141, "step": 33376 }, { "epoch": 2.0719473586194055, "grad_norm": 0.5516816483314012, "learning_rate": 2.642677908064486e-05, "loss": 2.6809, "step": 33377 }, { "epoch": 2.0720094357191634, "grad_norm": 0.5219487041120012, "learning_rate": 2.642359419283484e-05, "loss": 2.657, "step": 33378 }, { "epoch": 2.072071512818921, "grad_norm": 0.5014574445941923, "learning_rate": 2.642040942802865e-05, "loss": 2.6645, "step": 33379 }, { "epoch": 2.072133589918679, "grad_norm": 0.5277587670393594, "learning_rate": 2.641722478624291e-05, "loss": 2.7132, "step": 33380 }, { "epoch": 2.0721956670184367, "grad_norm": 0.511528358275584, "learning_rate": 2.6414040267494232e-05, "loss": 2.7165, "step": 33381 }, { "epoch": 2.0722577441181946, "grad_norm": 0.590659606767439, "learning_rate": 2.6410855871799217e-05, "loss": 2.6965, "step": 33382 }, { "epoch": 2.0723198212179526, "grad_norm": 0.540719925809983, "learning_rate": 2.640767159917451e-05, "loss": 2.7258, "step": 33383 }, { "epoch": 2.0723818983177105, "grad_norm": 0.5646908797229985, "learning_rate": 2.6404487449636705e-05, "loss": 2.6827, "step": 33384 }, { "epoch": 2.0724439754174684, "grad_norm": 0.572632603592969, "learning_rate": 2.6401303423202423e-05, "loss": 2.6421, "step": 33385 }, { "epoch": 2.0725060525172263, "grad_norm": 0.5967617696441235, "learning_rate": 2.6398119519888243e-05, "loss": 2.7807, "step": 33386 }, { "epoch": 2.0725681296169842, "grad_norm": 0.5420102503060655, "learning_rate": 2.6394935739710823e-05, "loss": 2.678, "step": 33387 }, { "epoch": 2.072630206716742, "grad_norm": 0.5956957957332804, "learning_rate": 2.639175208268676e-05, "loss": 2.7693, "step": 33388 }, { "epoch": 2.0726922838165, "grad_norm": 0.5825248263966558, "learning_rate": 2.638856854883266e-05, "loss": 2.7008, "step": 33389 }, { "epoch": 2.072754360916258, "grad_norm": 0.4817711151397156, "learning_rate": 2.6385385138165096e-05, "loss": 2.6484, "step": 33390 }, { "epoch": 2.072816438016016, "grad_norm": 0.5429393568247611, "learning_rate": 2.6382201850700745e-05, "loss": 2.6175, "step": 33391 }, { "epoch": 2.072878515115774, "grad_norm": 0.528693922013538, "learning_rate": 2.637901868645617e-05, "loss": 2.719, "step": 33392 }, { "epoch": 2.0729405922155317, "grad_norm": 0.5269629339686573, "learning_rate": 2.6375835645447994e-05, "loss": 2.8015, "step": 33393 }, { "epoch": 2.0730026693152896, "grad_norm": 0.5500050750411578, "learning_rate": 2.6372652727692816e-05, "loss": 2.7271, "step": 33394 }, { "epoch": 2.0730647464150476, "grad_norm": 0.5532761738977109, "learning_rate": 2.6369469933207224e-05, "loss": 2.6499, "step": 33395 }, { "epoch": 2.0731268235148055, "grad_norm": 0.5769260449210875, "learning_rate": 2.6366287262007873e-05, "loss": 2.7525, "step": 33396 }, { "epoch": 2.0731889006145634, "grad_norm": 0.5371662456575109, "learning_rate": 2.6363104714111326e-05, "loss": 2.5945, "step": 33397 }, { "epoch": 2.0732509777143213, "grad_norm": 0.579579354041028, "learning_rate": 2.6359922289534206e-05, "loss": 2.6686, "step": 33398 }, { "epoch": 2.0733130548140792, "grad_norm": 0.5434445279474757, "learning_rate": 2.6356739988293087e-05, "loss": 2.6594, "step": 33399 }, { "epoch": 2.073375131913837, "grad_norm": 0.6019222125335247, "learning_rate": 2.6353557810404617e-05, "loss": 2.6734, "step": 33400 }, { "epoch": 2.073437209013595, "grad_norm": 0.5544797750160692, "learning_rate": 2.635037575588538e-05, "loss": 2.7732, "step": 33401 }, { "epoch": 2.073499286113353, "grad_norm": 0.49972563905197503, "learning_rate": 2.634719382475197e-05, "loss": 2.6129, "step": 33402 }, { "epoch": 2.073561363213111, "grad_norm": 0.6480889360184781, "learning_rate": 2.6344012017020968e-05, "loss": 2.7276, "step": 33403 }, { "epoch": 2.0736234403128684, "grad_norm": 0.5448598792246157, "learning_rate": 2.634083033270902e-05, "loss": 2.6348, "step": 33404 }, { "epoch": 2.0736855174126263, "grad_norm": 0.5397873027786577, "learning_rate": 2.6337648771832713e-05, "loss": 2.739, "step": 33405 }, { "epoch": 2.073747594512384, "grad_norm": 0.5672659302177285, "learning_rate": 2.6334467334408624e-05, "loss": 2.731, "step": 33406 }, { "epoch": 2.073809671612142, "grad_norm": 0.5405137526432547, "learning_rate": 2.633128602045337e-05, "loss": 2.7059, "step": 33407 }, { "epoch": 2.0738717487119, "grad_norm": 0.5623014682519947, "learning_rate": 2.6328104829983523e-05, "loss": 2.6868, "step": 33408 }, { "epoch": 2.073933825811658, "grad_norm": 0.5444178557580263, "learning_rate": 2.6324923763015697e-05, "loss": 2.7703, "step": 33409 }, { "epoch": 2.073995902911416, "grad_norm": 0.5209984104494678, "learning_rate": 2.6321742819566515e-05, "loss": 2.7566, "step": 33410 }, { "epoch": 2.074057980011174, "grad_norm": 0.5659731030863621, "learning_rate": 2.6318561999652546e-05, "loss": 2.6777, "step": 33411 }, { "epoch": 2.0741200571109317, "grad_norm": 0.5081933867264058, "learning_rate": 2.6315381303290386e-05, "loss": 2.6758, "step": 33412 }, { "epoch": 2.0741821342106896, "grad_norm": 0.5944839007537431, "learning_rate": 2.6312200730496618e-05, "loss": 2.7458, "step": 33413 }, { "epoch": 2.0742442113104476, "grad_norm": 0.5543884760821781, "learning_rate": 2.630902028128786e-05, "loss": 2.7356, "step": 33414 }, { "epoch": 2.0743062884102055, "grad_norm": 0.5256360129745998, "learning_rate": 2.63058399556807e-05, "loss": 2.6763, "step": 33415 }, { "epoch": 2.0743683655099634, "grad_norm": 0.545248200525098, "learning_rate": 2.630265975369172e-05, "loss": 2.6939, "step": 33416 }, { "epoch": 2.0744304426097213, "grad_norm": 0.5680461426783849, "learning_rate": 2.6299479675337523e-05, "loss": 2.6652, "step": 33417 }, { "epoch": 2.0744925197094792, "grad_norm": 0.4823923449416595, "learning_rate": 2.6296299720634678e-05, "loss": 2.6747, "step": 33418 }, { "epoch": 2.074554596809237, "grad_norm": 0.5507354717303785, "learning_rate": 2.6293119889599798e-05, "loss": 2.6662, "step": 33419 }, { "epoch": 2.074616673908995, "grad_norm": 0.5001300106298137, "learning_rate": 2.6289940182249472e-05, "loss": 2.6942, "step": 33420 }, { "epoch": 2.074678751008753, "grad_norm": 0.5137523399541059, "learning_rate": 2.6286760598600292e-05, "loss": 2.7168, "step": 33421 }, { "epoch": 2.074740828108511, "grad_norm": 0.5192931927681985, "learning_rate": 2.6283581138668812e-05, "loss": 2.6702, "step": 33422 }, { "epoch": 2.074802905208269, "grad_norm": 0.5726089489229207, "learning_rate": 2.6280401802471665e-05, "loss": 2.7304, "step": 33423 }, { "epoch": 2.0748649823080267, "grad_norm": 0.5496700521782799, "learning_rate": 2.6277222590025418e-05, "loss": 2.6345, "step": 33424 }, { "epoch": 2.0749270594077847, "grad_norm": 0.6362113334957606, "learning_rate": 2.627404350134667e-05, "loss": 2.6998, "step": 33425 }, { "epoch": 2.0749891365075426, "grad_norm": 0.5767340035437988, "learning_rate": 2.6270864536451967e-05, "loss": 2.644, "step": 33426 }, { "epoch": 2.0750512136073, "grad_norm": 0.5790041090485103, "learning_rate": 2.6267685695357947e-05, "loss": 2.755, "step": 33427 }, { "epoch": 2.075113290707058, "grad_norm": 0.6073930249001775, "learning_rate": 2.626450697808117e-05, "loss": 2.6777, "step": 33428 }, { "epoch": 2.075175367806816, "grad_norm": 0.6166859774100526, "learning_rate": 2.626132838463823e-05, "loss": 2.6351, "step": 33429 }, { "epoch": 2.075237444906574, "grad_norm": 0.5703163146845078, "learning_rate": 2.6258149915045694e-05, "loss": 2.6936, "step": 33430 }, { "epoch": 2.0752995220063317, "grad_norm": 0.5875238180405181, "learning_rate": 2.625497156932014e-05, "loss": 2.666, "step": 33431 }, { "epoch": 2.0753615991060896, "grad_norm": 0.5608704593305465, "learning_rate": 2.6251793347478183e-05, "loss": 2.6836, "step": 33432 }, { "epoch": 2.0754236762058476, "grad_norm": 0.5260745847115484, "learning_rate": 2.624861524953638e-05, "loss": 2.7555, "step": 33433 }, { "epoch": 2.0754857533056055, "grad_norm": 0.60648995694323, "learning_rate": 2.624543727551132e-05, "loss": 2.7942, "step": 33434 }, { "epoch": 2.0755478304053634, "grad_norm": 0.5240936152684509, "learning_rate": 2.624225942541956e-05, "loss": 2.6567, "step": 33435 }, { "epoch": 2.0756099075051213, "grad_norm": 0.5696228246758223, "learning_rate": 2.623908169927771e-05, "loss": 2.6595, "step": 33436 }, { "epoch": 2.0756719846048792, "grad_norm": 0.5200203049345037, "learning_rate": 2.6235904097102354e-05, "loss": 2.6726, "step": 33437 }, { "epoch": 2.075734061704637, "grad_norm": 0.5893447988787927, "learning_rate": 2.6232726618910052e-05, "loss": 2.7552, "step": 33438 }, { "epoch": 2.075796138804395, "grad_norm": 0.5676210464662438, "learning_rate": 2.6229549264717363e-05, "loss": 2.6821, "step": 33439 }, { "epoch": 2.075858215904153, "grad_norm": 0.5717114834924448, "learning_rate": 2.6226372034540903e-05, "loss": 2.7426, "step": 33440 }, { "epoch": 2.075920293003911, "grad_norm": 0.5461431238123112, "learning_rate": 2.622319492839723e-05, "loss": 2.7343, "step": 33441 }, { "epoch": 2.075982370103669, "grad_norm": 0.6491042925659074, "learning_rate": 2.622001794630291e-05, "loss": 2.6966, "step": 33442 }, { "epoch": 2.0760444472034267, "grad_norm": 0.576067924820797, "learning_rate": 2.6216841088274547e-05, "loss": 2.7374, "step": 33443 }, { "epoch": 2.0761065243031847, "grad_norm": 0.6721535925690489, "learning_rate": 2.6213664354328692e-05, "loss": 2.7502, "step": 33444 }, { "epoch": 2.0761686014029426, "grad_norm": 0.562453321003577, "learning_rate": 2.621048774448191e-05, "loss": 2.6759, "step": 33445 }, { "epoch": 2.0762306785027005, "grad_norm": 0.6383817287812751, "learning_rate": 2.6207311258750804e-05, "loss": 2.7043, "step": 33446 }, { "epoch": 2.0762927556024584, "grad_norm": 0.5191367236034566, "learning_rate": 2.6204134897151933e-05, "loss": 2.6375, "step": 33447 }, { "epoch": 2.0763548327022163, "grad_norm": 0.5583055726546847, "learning_rate": 2.6200958659701867e-05, "loss": 2.6549, "step": 33448 }, { "epoch": 2.0764169098019742, "grad_norm": 0.5667192317854371, "learning_rate": 2.619778254641716e-05, "loss": 2.7248, "step": 33449 }, { "epoch": 2.076478986901732, "grad_norm": 0.4952822979885829, "learning_rate": 2.6194606557314416e-05, "loss": 2.687, "step": 33450 }, { "epoch": 2.07654106400149, "grad_norm": 0.610881508215106, "learning_rate": 2.619143069241019e-05, "loss": 2.6801, "step": 33451 }, { "epoch": 2.0766031411012476, "grad_norm": 0.550316797398907, "learning_rate": 2.6188254951721048e-05, "loss": 2.7066, "step": 33452 }, { "epoch": 2.0766652182010055, "grad_norm": 0.5000725335915698, "learning_rate": 2.6185079335263542e-05, "loss": 2.7445, "step": 33453 }, { "epoch": 2.0767272953007634, "grad_norm": 0.582877037422731, "learning_rate": 2.618190384305428e-05, "loss": 2.7251, "step": 33454 }, { "epoch": 2.0767893724005213, "grad_norm": 0.5103788232073881, "learning_rate": 2.6178728475109803e-05, "loss": 2.644, "step": 33455 }, { "epoch": 2.0768514495002792, "grad_norm": 0.5519850271499016, "learning_rate": 2.617555323144668e-05, "loss": 2.7304, "step": 33456 }, { "epoch": 2.076913526600037, "grad_norm": 0.6118640196129636, "learning_rate": 2.6172378112081487e-05, "loss": 2.7514, "step": 33457 }, { "epoch": 2.076975603699795, "grad_norm": 0.5159616527299135, "learning_rate": 2.616920311703076e-05, "loss": 2.7225, "step": 33458 }, { "epoch": 2.077037680799553, "grad_norm": 0.5583871045718819, "learning_rate": 2.6166028246311103e-05, "loss": 2.6471, "step": 33459 }, { "epoch": 2.077099757899311, "grad_norm": 0.5616149297066534, "learning_rate": 2.6162853499939067e-05, "loss": 2.6848, "step": 33460 }, { "epoch": 2.077161834999069, "grad_norm": 0.5377179391508672, "learning_rate": 2.6159678877931204e-05, "loss": 2.7026, "step": 33461 }, { "epoch": 2.0772239120988267, "grad_norm": 0.487071545316571, "learning_rate": 2.615650438030407e-05, "loss": 2.6463, "step": 33462 }, { "epoch": 2.0772859891985846, "grad_norm": 0.5417349815082492, "learning_rate": 2.615333000707426e-05, "loss": 2.7983, "step": 33463 }, { "epoch": 2.0773480662983426, "grad_norm": 0.5105144510867167, "learning_rate": 2.615015575825831e-05, "loss": 2.6571, "step": 33464 }, { "epoch": 2.0774101433981005, "grad_norm": 0.528571782931279, "learning_rate": 2.614698163387279e-05, "loss": 2.6885, "step": 33465 }, { "epoch": 2.0774722204978584, "grad_norm": 0.5395357017653806, "learning_rate": 2.6143807633934265e-05, "loss": 2.7131, "step": 33466 }, { "epoch": 2.0775342975976163, "grad_norm": 0.5040537004832024, "learning_rate": 2.614063375845926e-05, "loss": 2.6485, "step": 33467 }, { "epoch": 2.0775963746973742, "grad_norm": 0.5400134834177991, "learning_rate": 2.6137460007464382e-05, "loss": 2.7259, "step": 33468 }, { "epoch": 2.077658451797132, "grad_norm": 0.5138988501067323, "learning_rate": 2.6134286380966166e-05, "loss": 2.6908, "step": 33469 }, { "epoch": 2.07772052889689, "grad_norm": 0.5126069201611423, "learning_rate": 2.6131112878981177e-05, "loss": 2.6821, "step": 33470 }, { "epoch": 2.077782605996648, "grad_norm": 0.5566274540999355, "learning_rate": 2.6127939501525946e-05, "loss": 2.6739, "step": 33471 }, { "epoch": 2.077844683096406, "grad_norm": 0.5349826948985755, "learning_rate": 2.612476624861706e-05, "loss": 2.718, "step": 33472 }, { "epoch": 2.077906760196164, "grad_norm": 0.5570297219789153, "learning_rate": 2.612159312027107e-05, "loss": 2.7622, "step": 33473 }, { "epoch": 2.0779688372959217, "grad_norm": 0.5793646589506477, "learning_rate": 2.611842011650453e-05, "loss": 2.8063, "step": 33474 }, { "epoch": 2.078030914395679, "grad_norm": 0.5451523761606165, "learning_rate": 2.611524723733397e-05, "loss": 2.7377, "step": 33475 }, { "epoch": 2.078092991495437, "grad_norm": 0.5663968315656857, "learning_rate": 2.6112074482775955e-05, "loss": 2.7274, "step": 33476 }, { "epoch": 2.078155068595195, "grad_norm": 0.5328328730024668, "learning_rate": 2.6108901852847072e-05, "loss": 2.7123, "step": 33477 }, { "epoch": 2.078217145694953, "grad_norm": 0.5251361840593308, "learning_rate": 2.6105729347563845e-05, "loss": 2.7089, "step": 33478 }, { "epoch": 2.078279222794711, "grad_norm": 0.5421101484720616, "learning_rate": 2.610255696694283e-05, "loss": 2.6921, "step": 33479 }, { "epoch": 2.078341299894469, "grad_norm": 0.5382952828093771, "learning_rate": 2.609938471100058e-05, "loss": 2.668, "step": 33480 }, { "epoch": 2.0784033769942267, "grad_norm": 0.4816511847582937, "learning_rate": 2.609621257975362e-05, "loss": 2.599, "step": 33481 }, { "epoch": 2.0784654540939846, "grad_norm": 0.5503270596468746, "learning_rate": 2.609304057321854e-05, "loss": 2.6533, "step": 33482 }, { "epoch": 2.0785275311937426, "grad_norm": 0.5224866914194992, "learning_rate": 2.6089868691411868e-05, "loss": 2.555, "step": 33483 }, { "epoch": 2.0785896082935005, "grad_norm": 0.5531897673969307, "learning_rate": 2.6086696934350163e-05, "loss": 2.6094, "step": 33484 }, { "epoch": 2.0786516853932584, "grad_norm": 0.5738215986884365, "learning_rate": 2.6083525302049937e-05, "loss": 2.6587, "step": 33485 }, { "epoch": 2.0787137624930163, "grad_norm": 0.5440784065511187, "learning_rate": 2.608035379452779e-05, "loss": 2.636, "step": 33486 }, { "epoch": 2.0787758395927742, "grad_norm": 0.5500088951085121, "learning_rate": 2.607718241180025e-05, "loss": 2.6974, "step": 33487 }, { "epoch": 2.078837916692532, "grad_norm": 0.5282699363510224, "learning_rate": 2.6074011153883848e-05, "loss": 2.6893, "step": 33488 }, { "epoch": 2.07889999379229, "grad_norm": 0.5076015621186066, "learning_rate": 2.6070840020795117e-05, "loss": 2.6379, "step": 33489 }, { "epoch": 2.078962070892048, "grad_norm": 0.5617093516533146, "learning_rate": 2.606766901255065e-05, "loss": 2.6726, "step": 33490 }, { "epoch": 2.079024147991806, "grad_norm": 0.5671333644452745, "learning_rate": 2.606449812916696e-05, "loss": 2.641, "step": 33491 }, { "epoch": 2.079086225091564, "grad_norm": 0.4661222075401457, "learning_rate": 2.6061327370660592e-05, "loss": 2.6682, "step": 33492 }, { "epoch": 2.0791483021913217, "grad_norm": 0.5788891356045244, "learning_rate": 2.605815673704809e-05, "loss": 2.7438, "step": 33493 }, { "epoch": 2.0792103792910797, "grad_norm": 0.5864964679992161, "learning_rate": 2.605498622834598e-05, "loss": 2.736, "step": 33494 }, { "epoch": 2.0792724563908376, "grad_norm": 0.5461961620170827, "learning_rate": 2.6051815844570837e-05, "loss": 2.7386, "step": 33495 }, { "epoch": 2.0793345334905955, "grad_norm": 0.5566477724921461, "learning_rate": 2.604864558573919e-05, "loss": 2.726, "step": 33496 }, { "epoch": 2.0793966105903534, "grad_norm": 0.5543094743798109, "learning_rate": 2.604547545186757e-05, "loss": 2.621, "step": 33497 }, { "epoch": 2.0794586876901113, "grad_norm": 0.4949171275307832, "learning_rate": 2.60423054429725e-05, "loss": 2.7631, "step": 33498 }, { "epoch": 2.079520764789869, "grad_norm": 0.5519308983389952, "learning_rate": 2.6039135559070553e-05, "loss": 2.6864, "step": 33499 }, { "epoch": 2.0795828418896267, "grad_norm": 0.5313618205699011, "learning_rate": 2.603596580017826e-05, "loss": 2.6831, "step": 33500 }, { "epoch": 2.0796449189893846, "grad_norm": 0.546208738737135, "learning_rate": 2.6032796166312146e-05, "loss": 2.6187, "step": 33501 }, { "epoch": 2.0797069960891426, "grad_norm": 0.5285514001798357, "learning_rate": 2.6029626657488753e-05, "loss": 2.7635, "step": 33502 }, { "epoch": 2.0797690731889005, "grad_norm": 0.49814621177798185, "learning_rate": 2.6026457273724597e-05, "loss": 2.6874, "step": 33503 }, { "epoch": 2.0798311502886584, "grad_norm": 0.5912960186604769, "learning_rate": 2.602328801503625e-05, "loss": 2.6878, "step": 33504 }, { "epoch": 2.0798932273884163, "grad_norm": 0.5786257971846502, "learning_rate": 2.6020118881440236e-05, "loss": 2.7803, "step": 33505 }, { "epoch": 2.0799553044881742, "grad_norm": 0.5630279288843191, "learning_rate": 2.6016949872953083e-05, "loss": 2.7087, "step": 33506 }, { "epoch": 2.080017381587932, "grad_norm": 0.5699796619886035, "learning_rate": 2.6013780989591306e-05, "loss": 2.7185, "step": 33507 }, { "epoch": 2.08007945868769, "grad_norm": 0.5661978765838621, "learning_rate": 2.601061223137145e-05, "loss": 2.6841, "step": 33508 }, { "epoch": 2.080141535787448, "grad_norm": 0.5257655511244014, "learning_rate": 2.6007443598310076e-05, "loss": 2.6707, "step": 33509 }, { "epoch": 2.080203612887206, "grad_norm": 0.5682620460049109, "learning_rate": 2.6004275090423697e-05, "loss": 2.6566, "step": 33510 }, { "epoch": 2.080265689986964, "grad_norm": 0.612002966127189, "learning_rate": 2.6001106707728835e-05, "loss": 2.7553, "step": 33511 }, { "epoch": 2.0803277670867217, "grad_norm": 0.5347788238568438, "learning_rate": 2.599793845024201e-05, "loss": 2.6401, "step": 33512 }, { "epoch": 2.0803898441864797, "grad_norm": 0.5524434378743219, "learning_rate": 2.5994770317979782e-05, "loss": 2.6777, "step": 33513 }, { "epoch": 2.0804519212862376, "grad_norm": 0.5654383841522749, "learning_rate": 2.5991602310958662e-05, "loss": 2.6828, "step": 33514 }, { "epoch": 2.0805139983859955, "grad_norm": 0.5450981155935385, "learning_rate": 2.5988434429195192e-05, "loss": 2.6967, "step": 33515 }, { "epoch": 2.0805760754857534, "grad_norm": 0.6332660980256727, "learning_rate": 2.5985266672705876e-05, "loss": 2.6532, "step": 33516 }, { "epoch": 2.0806381525855113, "grad_norm": 0.5312637720005876, "learning_rate": 2.5982099041507247e-05, "loss": 2.7093, "step": 33517 }, { "epoch": 2.0807002296852692, "grad_norm": 0.6014730213548948, "learning_rate": 2.5978931535615847e-05, "loss": 2.6236, "step": 33518 }, { "epoch": 2.080762306785027, "grad_norm": 0.5611134765777624, "learning_rate": 2.59757641550482e-05, "loss": 2.7007, "step": 33519 }, { "epoch": 2.080824383884785, "grad_norm": 0.48146846320955294, "learning_rate": 2.5972596899820818e-05, "loss": 2.7196, "step": 33520 }, { "epoch": 2.080886460984543, "grad_norm": 0.5902592540776529, "learning_rate": 2.5969429769950217e-05, "loss": 2.6544, "step": 33521 }, { "epoch": 2.080948538084301, "grad_norm": 0.5448305060240463, "learning_rate": 2.5966262765452943e-05, "loss": 2.6816, "step": 33522 }, { "epoch": 2.0810106151840584, "grad_norm": 0.5354064022619064, "learning_rate": 2.5963095886345517e-05, "loss": 2.7025, "step": 33523 }, { "epoch": 2.0810726922838163, "grad_norm": 0.5316135340731615, "learning_rate": 2.5959929132644456e-05, "loss": 2.5953, "step": 33524 }, { "epoch": 2.0811347693835742, "grad_norm": 0.5268497555116258, "learning_rate": 2.595676250436626e-05, "loss": 2.6724, "step": 33525 }, { "epoch": 2.081196846483332, "grad_norm": 0.5401589948536433, "learning_rate": 2.5953596001527493e-05, "loss": 2.6763, "step": 33526 }, { "epoch": 2.08125892358309, "grad_norm": 0.4933126244703711, "learning_rate": 2.595042962414465e-05, "loss": 2.6859, "step": 33527 }, { "epoch": 2.081321000682848, "grad_norm": 0.5326355089226548, "learning_rate": 2.5947263372234253e-05, "loss": 2.7982, "step": 33528 }, { "epoch": 2.081383077782606, "grad_norm": 0.4837384502007952, "learning_rate": 2.5944097245812827e-05, "loss": 2.6251, "step": 33529 }, { "epoch": 2.081445154882364, "grad_norm": 0.5200227979100742, "learning_rate": 2.594093124489686e-05, "loss": 2.6775, "step": 33530 }, { "epoch": 2.0815072319821217, "grad_norm": 0.5559407435339174, "learning_rate": 2.593776536950292e-05, "loss": 2.7933, "step": 33531 }, { "epoch": 2.0815693090818796, "grad_norm": 0.5331112848912501, "learning_rate": 2.5934599619647493e-05, "loss": 2.6415, "step": 33532 }, { "epoch": 2.0816313861816376, "grad_norm": 0.47289109504004534, "learning_rate": 2.593143399534711e-05, "loss": 2.6993, "step": 33533 }, { "epoch": 2.0816934632813955, "grad_norm": 0.5698018596744016, "learning_rate": 2.5928268496618252e-05, "loss": 2.7513, "step": 33534 }, { "epoch": 2.0817555403811534, "grad_norm": 0.4875056400950937, "learning_rate": 2.5925103123477483e-05, "loss": 2.7356, "step": 33535 }, { "epoch": 2.0818176174809113, "grad_norm": 0.49577101775345467, "learning_rate": 2.5921937875941292e-05, "loss": 2.6203, "step": 33536 }, { "epoch": 2.0818796945806692, "grad_norm": 0.5322427729666499, "learning_rate": 2.5918772754026205e-05, "loss": 2.7687, "step": 33537 }, { "epoch": 2.081941771680427, "grad_norm": 0.5754672898691687, "learning_rate": 2.59156077577487e-05, "loss": 2.5933, "step": 33538 }, { "epoch": 2.082003848780185, "grad_norm": 0.5059247655838676, "learning_rate": 2.591244288712534e-05, "loss": 2.7446, "step": 33539 }, { "epoch": 2.082065925879943, "grad_norm": 0.5921000846418834, "learning_rate": 2.590927814217259e-05, "loss": 2.7304, "step": 33540 }, { "epoch": 2.082128002979701, "grad_norm": 0.48066761584404244, "learning_rate": 2.5906113522907006e-05, "loss": 2.6634, "step": 33541 }, { "epoch": 2.082190080079459, "grad_norm": 0.5669518046727077, "learning_rate": 2.590294902934507e-05, "loss": 2.86, "step": 33542 }, { "epoch": 2.0822521571792167, "grad_norm": 0.5202838637802839, "learning_rate": 2.5899784661503308e-05, "loss": 2.6174, "step": 33543 }, { "epoch": 2.0823142342789747, "grad_norm": 0.49637360076111486, "learning_rate": 2.58966204193982e-05, "loss": 2.6638, "step": 33544 }, { "epoch": 2.0823763113787326, "grad_norm": 0.5452325953662108, "learning_rate": 2.5893456303046283e-05, "loss": 2.7835, "step": 33545 }, { "epoch": 2.0824383884784905, "grad_norm": 0.5233386786252687, "learning_rate": 2.5890292312464064e-05, "loss": 2.8095, "step": 33546 }, { "epoch": 2.082500465578248, "grad_norm": 0.5542847420330327, "learning_rate": 2.588712844766804e-05, "loss": 2.7099, "step": 33547 }, { "epoch": 2.082562542678006, "grad_norm": 0.48411682385573757, "learning_rate": 2.5883964708674713e-05, "loss": 2.6898, "step": 33548 }, { "epoch": 2.082624619777764, "grad_norm": 0.5415112260163809, "learning_rate": 2.5880801095500607e-05, "loss": 2.6071, "step": 33549 }, { "epoch": 2.0826866968775217, "grad_norm": 0.5675736072042807, "learning_rate": 2.5877637608162208e-05, "loss": 2.6727, "step": 33550 }, { "epoch": 2.0827487739772796, "grad_norm": 0.47546648409200415, "learning_rate": 2.587447424667604e-05, "loss": 2.6666, "step": 33551 }, { "epoch": 2.0828108510770376, "grad_norm": 0.48950773697097305, "learning_rate": 2.5871311011058595e-05, "loss": 2.6856, "step": 33552 }, { "epoch": 2.0828729281767955, "grad_norm": 0.5149353790943293, "learning_rate": 2.5868147901326356e-05, "loss": 2.7319, "step": 33553 }, { "epoch": 2.0829350052765534, "grad_norm": 0.5341008213551143, "learning_rate": 2.586498491749587e-05, "loss": 2.6568, "step": 33554 }, { "epoch": 2.0829970823763113, "grad_norm": 0.5078354463528094, "learning_rate": 2.586182205958362e-05, "loss": 2.7788, "step": 33555 }, { "epoch": 2.0830591594760692, "grad_norm": 0.4779357043013443, "learning_rate": 2.58586593276061e-05, "loss": 2.6681, "step": 33556 }, { "epoch": 2.083121236575827, "grad_norm": 0.5260583824548587, "learning_rate": 2.5855496721579798e-05, "loss": 2.6705, "step": 33557 }, { "epoch": 2.083183313675585, "grad_norm": 0.5363777865584387, "learning_rate": 2.5852334241521243e-05, "loss": 2.6203, "step": 33558 }, { "epoch": 2.083245390775343, "grad_norm": 0.5157001036075963, "learning_rate": 2.5849171887446926e-05, "loss": 2.7287, "step": 33559 }, { "epoch": 2.083307467875101, "grad_norm": 0.5422443799791218, "learning_rate": 2.5846009659373344e-05, "loss": 2.6733, "step": 33560 }, { "epoch": 2.083369544974859, "grad_norm": 0.5426373465222406, "learning_rate": 2.5842847557316974e-05, "loss": 2.6973, "step": 33561 }, { "epoch": 2.0834316220746167, "grad_norm": 0.5718883794251284, "learning_rate": 2.583968558129435e-05, "loss": 2.6023, "step": 33562 }, { "epoch": 2.0834936991743747, "grad_norm": 0.5727936063440281, "learning_rate": 2.583652373132195e-05, "loss": 2.7494, "step": 33563 }, { "epoch": 2.0835557762741326, "grad_norm": 0.603127783931968, "learning_rate": 2.5833362007416273e-05, "loss": 2.7087, "step": 33564 }, { "epoch": 2.0836178533738905, "grad_norm": 0.559148659546453, "learning_rate": 2.583020040959382e-05, "loss": 2.708, "step": 33565 }, { "epoch": 2.0836799304736484, "grad_norm": 0.5342618889474283, "learning_rate": 2.582703893787105e-05, "loss": 2.6937, "step": 33566 }, { "epoch": 2.0837420075734063, "grad_norm": 0.5589550157217025, "learning_rate": 2.582387759226451e-05, "loss": 2.6514, "step": 33567 }, { "epoch": 2.0838040846731642, "grad_norm": 0.4679254909893724, "learning_rate": 2.5820716372790664e-05, "loss": 2.6944, "step": 33568 }, { "epoch": 2.083866161772922, "grad_norm": 0.568647750634169, "learning_rate": 2.5817555279466017e-05, "loss": 2.6652, "step": 33569 }, { "epoch": 2.08392823887268, "grad_norm": 0.5105395234096011, "learning_rate": 2.5814394312307033e-05, "loss": 2.6211, "step": 33570 }, { "epoch": 2.0839903159724376, "grad_norm": 0.5196036094812532, "learning_rate": 2.581123347133024e-05, "loss": 2.7463, "step": 33571 }, { "epoch": 2.0840523930721955, "grad_norm": 0.5243187503403494, "learning_rate": 2.5808072756552127e-05, "loss": 2.7294, "step": 33572 }, { "epoch": 2.0841144701719534, "grad_norm": 0.5287478650728598, "learning_rate": 2.5804912167989138e-05, "loss": 2.7324, "step": 33573 }, { "epoch": 2.0841765472717113, "grad_norm": 0.5373525727821408, "learning_rate": 2.580175170565782e-05, "loss": 2.72, "step": 33574 }, { "epoch": 2.0842386243714692, "grad_norm": 0.5471363533032579, "learning_rate": 2.579859136957462e-05, "loss": 2.5974, "step": 33575 }, { "epoch": 2.084300701471227, "grad_norm": 0.5009714464666282, "learning_rate": 2.579543115975606e-05, "loss": 2.6247, "step": 33576 }, { "epoch": 2.084362778570985, "grad_norm": 0.6073280580935686, "learning_rate": 2.5792271076218618e-05, "loss": 2.7627, "step": 33577 }, { "epoch": 2.084424855670743, "grad_norm": 0.5414857990343585, "learning_rate": 2.578911111897877e-05, "loss": 2.7143, "step": 33578 }, { "epoch": 2.084486932770501, "grad_norm": 0.5530716908083843, "learning_rate": 2.5785951288053e-05, "loss": 2.6479, "step": 33579 }, { "epoch": 2.084549009870259, "grad_norm": 0.5349160372277771, "learning_rate": 2.5782791583457787e-05, "loss": 2.6873, "step": 33580 }, { "epoch": 2.0846110869700167, "grad_norm": 0.5405391426487194, "learning_rate": 2.5779632005209653e-05, "loss": 2.7063, "step": 33581 }, { "epoch": 2.0846731640697747, "grad_norm": 0.5820237485671035, "learning_rate": 2.577647255332506e-05, "loss": 2.7065, "step": 33582 }, { "epoch": 2.0847352411695326, "grad_norm": 0.6002746767900237, "learning_rate": 2.5773313227820484e-05, "loss": 2.747, "step": 33583 }, { "epoch": 2.0847973182692905, "grad_norm": 0.5977190770056444, "learning_rate": 2.577015402871239e-05, "loss": 2.7308, "step": 33584 }, { "epoch": 2.0848593953690484, "grad_norm": 0.5653802721636746, "learning_rate": 2.5766994956017308e-05, "loss": 2.7296, "step": 33585 }, { "epoch": 2.0849214724688063, "grad_norm": 0.5769391656735237, "learning_rate": 2.57638360097517e-05, "loss": 2.7975, "step": 33586 }, { "epoch": 2.0849835495685642, "grad_norm": 0.5441296195007853, "learning_rate": 2.5760677189932044e-05, "loss": 2.7442, "step": 33587 }, { "epoch": 2.085045626668322, "grad_norm": 0.5528855942554803, "learning_rate": 2.5757518496574794e-05, "loss": 2.7718, "step": 33588 }, { "epoch": 2.08510770376808, "grad_norm": 0.549482993403363, "learning_rate": 2.575435992969648e-05, "loss": 2.7799, "step": 33589 }, { "epoch": 2.085169780867838, "grad_norm": 0.52470148546166, "learning_rate": 2.5751201489313547e-05, "loss": 2.626, "step": 33590 }, { "epoch": 2.085231857967596, "grad_norm": 0.5663844400350376, "learning_rate": 2.5748043175442494e-05, "loss": 2.602, "step": 33591 }, { "epoch": 2.085293935067354, "grad_norm": 0.5241269936681741, "learning_rate": 2.5744884988099784e-05, "loss": 2.6854, "step": 33592 }, { "epoch": 2.0853560121671117, "grad_norm": 0.48845326254824334, "learning_rate": 2.5741726927301868e-05, "loss": 2.7473, "step": 33593 }, { "epoch": 2.0854180892668692, "grad_norm": 0.5423730010703327, "learning_rate": 2.5738568993065283e-05, "loss": 2.7463, "step": 33594 }, { "epoch": 2.085480166366627, "grad_norm": 0.5291909537661648, "learning_rate": 2.5735411185406467e-05, "loss": 2.6903, "step": 33595 }, { "epoch": 2.085542243466385, "grad_norm": 0.5355966802207845, "learning_rate": 2.57322535043419e-05, "loss": 2.7041, "step": 33596 }, { "epoch": 2.085604320566143, "grad_norm": 0.5156757734327212, "learning_rate": 2.5729095949888043e-05, "loss": 2.7058, "step": 33597 }, { "epoch": 2.085666397665901, "grad_norm": 0.5095865220006252, "learning_rate": 2.57259385220614e-05, "loss": 2.7286, "step": 33598 }, { "epoch": 2.085728474765659, "grad_norm": 0.583438579964928, "learning_rate": 2.5722781220878432e-05, "loss": 2.7134, "step": 33599 }, { "epoch": 2.0857905518654167, "grad_norm": 0.5190156658524122, "learning_rate": 2.5719624046355605e-05, "loss": 2.635, "step": 33600 }, { "epoch": 2.0858526289651746, "grad_norm": 0.5454524202287337, "learning_rate": 2.5716466998509398e-05, "loss": 2.714, "step": 33601 }, { "epoch": 2.0859147060649326, "grad_norm": 0.5409833161660162, "learning_rate": 2.5713310077356257e-05, "loss": 2.6346, "step": 33602 }, { "epoch": 2.0859767831646905, "grad_norm": 0.5677036553519695, "learning_rate": 2.5710153282912685e-05, "loss": 2.7392, "step": 33603 }, { "epoch": 2.0860388602644484, "grad_norm": 0.5568514508321214, "learning_rate": 2.570699661519515e-05, "loss": 2.7034, "step": 33604 }, { "epoch": 2.0861009373642063, "grad_norm": 0.584926368570779, "learning_rate": 2.570384007422011e-05, "loss": 2.7113, "step": 33605 }, { "epoch": 2.0861630144639642, "grad_norm": 0.5631976040386854, "learning_rate": 2.570068366000401e-05, "loss": 2.7644, "step": 33606 }, { "epoch": 2.086225091563722, "grad_norm": 0.5244603767047096, "learning_rate": 2.5697527372563342e-05, "loss": 2.6472, "step": 33607 }, { "epoch": 2.08628716866348, "grad_norm": 0.5606951102196467, "learning_rate": 2.5694371211914592e-05, "loss": 2.6289, "step": 33608 }, { "epoch": 2.086349245763238, "grad_norm": 0.5246659567194785, "learning_rate": 2.5691215178074213e-05, "loss": 2.6617, "step": 33609 }, { "epoch": 2.086411322862996, "grad_norm": 0.5496118183529027, "learning_rate": 2.568805927105866e-05, "loss": 2.7152, "step": 33610 }, { "epoch": 2.086473399962754, "grad_norm": 0.5236900915784457, "learning_rate": 2.5684903490884392e-05, "loss": 2.6059, "step": 33611 }, { "epoch": 2.0865354770625117, "grad_norm": 0.549435913649495, "learning_rate": 2.56817478375679e-05, "loss": 2.6945, "step": 33612 }, { "epoch": 2.0865975541622697, "grad_norm": 0.5443924003036289, "learning_rate": 2.5678592311125627e-05, "loss": 2.7425, "step": 33613 }, { "epoch": 2.0866596312620276, "grad_norm": 0.549903332278014, "learning_rate": 2.5675436911574052e-05, "loss": 2.6721, "step": 33614 }, { "epoch": 2.0867217083617855, "grad_norm": 0.5371245780873922, "learning_rate": 2.5672281638929623e-05, "loss": 2.6663, "step": 33615 }, { "epoch": 2.0867837854615434, "grad_norm": 0.5060460692437065, "learning_rate": 2.5669126493208785e-05, "loss": 2.6106, "step": 33616 }, { "epoch": 2.0868458625613013, "grad_norm": 0.5080381039010867, "learning_rate": 2.566597147442804e-05, "loss": 2.6954, "step": 33617 }, { "epoch": 2.0869079396610593, "grad_norm": 0.5491204891510182, "learning_rate": 2.566281658260384e-05, "loss": 2.7668, "step": 33618 }, { "epoch": 2.0869700167608167, "grad_norm": 0.4670216964658342, "learning_rate": 2.5659661817752622e-05, "loss": 2.75, "step": 33619 }, { "epoch": 2.0870320938605746, "grad_norm": 0.5620186388532611, "learning_rate": 2.565650717989084e-05, "loss": 2.6695, "step": 33620 }, { "epoch": 2.0870941709603326, "grad_norm": 0.5017069647161295, "learning_rate": 2.565335266903499e-05, "loss": 2.6384, "step": 33621 }, { "epoch": 2.0871562480600905, "grad_norm": 0.5484675244320981, "learning_rate": 2.565019828520151e-05, "loss": 2.7127, "step": 33622 }, { "epoch": 2.0872183251598484, "grad_norm": 0.5134863863034648, "learning_rate": 2.5647044028406856e-05, "loss": 2.7307, "step": 33623 }, { "epoch": 2.0872804022596063, "grad_norm": 0.5239247573600783, "learning_rate": 2.5643889898667473e-05, "loss": 2.6961, "step": 33624 }, { "epoch": 2.0873424793593642, "grad_norm": 0.5129378553435521, "learning_rate": 2.5640735895999833e-05, "loss": 2.7294, "step": 33625 }, { "epoch": 2.087404556459122, "grad_norm": 0.5062817316353808, "learning_rate": 2.5637582020420403e-05, "loss": 2.731, "step": 33626 }, { "epoch": 2.08746663355888, "grad_norm": 0.49527817616233033, "learning_rate": 2.563442827194561e-05, "loss": 2.6382, "step": 33627 }, { "epoch": 2.087528710658638, "grad_norm": 0.5087279651003355, "learning_rate": 2.563127465059193e-05, "loss": 2.7177, "step": 33628 }, { "epoch": 2.087590787758396, "grad_norm": 0.5352530009111203, "learning_rate": 2.5628121156375777e-05, "loss": 2.7198, "step": 33629 }, { "epoch": 2.087652864858154, "grad_norm": 0.5451350188337291, "learning_rate": 2.5624967789313658e-05, "loss": 2.6857, "step": 33630 }, { "epoch": 2.0877149419579117, "grad_norm": 0.5758031667526206, "learning_rate": 2.5621814549421996e-05, "loss": 2.7789, "step": 33631 }, { "epoch": 2.0877770190576697, "grad_norm": 0.5172383758783835, "learning_rate": 2.561866143671725e-05, "loss": 2.72, "step": 33632 }, { "epoch": 2.0878390961574276, "grad_norm": 0.5199977243355327, "learning_rate": 2.561550845121584e-05, "loss": 2.7402, "step": 33633 }, { "epoch": 2.0879011732571855, "grad_norm": 0.5692765219620284, "learning_rate": 2.5612355592934267e-05, "loss": 2.7793, "step": 33634 }, { "epoch": 2.0879632503569434, "grad_norm": 0.5171341341266397, "learning_rate": 2.5609202861888947e-05, "loss": 2.6219, "step": 33635 }, { "epoch": 2.0880253274567013, "grad_norm": 0.5381637212399596, "learning_rate": 2.560605025809635e-05, "loss": 2.6654, "step": 33636 }, { "epoch": 2.0880874045564592, "grad_norm": 0.5401267451315728, "learning_rate": 2.56028977815729e-05, "loss": 2.6623, "step": 33637 }, { "epoch": 2.088149481656217, "grad_norm": 0.526508093040844, "learning_rate": 2.5599745432335037e-05, "loss": 2.7332, "step": 33638 }, { "epoch": 2.088211558755975, "grad_norm": 0.5509281638874982, "learning_rate": 2.559659321039923e-05, "loss": 2.6962, "step": 33639 }, { "epoch": 2.088273635855733, "grad_norm": 0.5039455782897972, "learning_rate": 2.5593441115781937e-05, "loss": 2.5938, "step": 33640 }, { "epoch": 2.088335712955491, "grad_norm": 0.5426028274174923, "learning_rate": 2.5590289148499585e-05, "loss": 2.6886, "step": 33641 }, { "epoch": 2.0883977900552484, "grad_norm": 0.5368472205137083, "learning_rate": 2.558713730856862e-05, "loss": 2.6531, "step": 33642 }, { "epoch": 2.0884598671550063, "grad_norm": 0.5364533735866958, "learning_rate": 2.5583985596005467e-05, "loss": 2.7265, "step": 33643 }, { "epoch": 2.0885219442547642, "grad_norm": 0.5857470647971984, "learning_rate": 2.5580834010826616e-05, "loss": 2.7046, "step": 33644 }, { "epoch": 2.088584021354522, "grad_norm": 0.48958397080670785, "learning_rate": 2.557768255304847e-05, "loss": 2.6815, "step": 33645 }, { "epoch": 2.08864609845428, "grad_norm": 0.5384836821142497, "learning_rate": 2.557453122268749e-05, "loss": 2.6916, "step": 33646 }, { "epoch": 2.088708175554038, "grad_norm": 0.5261485648992162, "learning_rate": 2.557138001976009e-05, "loss": 2.7207, "step": 33647 }, { "epoch": 2.088770252653796, "grad_norm": 0.5127993232945011, "learning_rate": 2.5568228944282745e-05, "loss": 2.6808, "step": 33648 }, { "epoch": 2.088832329753554, "grad_norm": 0.5219284909639892, "learning_rate": 2.5565077996271886e-05, "loss": 2.711, "step": 33649 }, { "epoch": 2.0888944068533117, "grad_norm": 0.49240951980131714, "learning_rate": 2.5561927175743945e-05, "loss": 2.6789, "step": 33650 }, { "epoch": 2.0889564839530697, "grad_norm": 0.5073477746012618, "learning_rate": 2.5558776482715363e-05, "loss": 2.7246, "step": 33651 }, { "epoch": 2.0890185610528276, "grad_norm": 0.5042676295185445, "learning_rate": 2.5555625917202554e-05, "loss": 2.774, "step": 33652 }, { "epoch": 2.0890806381525855, "grad_norm": 0.49854702239528004, "learning_rate": 2.5552475479222003e-05, "loss": 2.7003, "step": 33653 }, { "epoch": 2.0891427152523434, "grad_norm": 0.5490974227142429, "learning_rate": 2.5549325168790123e-05, "loss": 2.6903, "step": 33654 }, { "epoch": 2.0892047923521013, "grad_norm": 0.531097782991724, "learning_rate": 2.5546174985923344e-05, "loss": 2.7325, "step": 33655 }, { "epoch": 2.0892668694518592, "grad_norm": 0.5503349389738847, "learning_rate": 2.5543024930638092e-05, "loss": 2.642, "step": 33656 }, { "epoch": 2.089328946551617, "grad_norm": 0.49488205000559465, "learning_rate": 2.553987500295083e-05, "loss": 2.6326, "step": 33657 }, { "epoch": 2.089391023651375, "grad_norm": 0.6106758171887208, "learning_rate": 2.5536725202877988e-05, "loss": 2.6897, "step": 33658 }, { "epoch": 2.089453100751133, "grad_norm": 0.48624828556513644, "learning_rate": 2.5533575530435983e-05, "loss": 2.6969, "step": 33659 }, { "epoch": 2.089515177850891, "grad_norm": 0.5870906164404779, "learning_rate": 2.5530425985641233e-05, "loss": 2.7212, "step": 33660 }, { "epoch": 2.089577254950649, "grad_norm": 0.6120246112805268, "learning_rate": 2.552727656851021e-05, "loss": 2.6966, "step": 33661 }, { "epoch": 2.0896393320504068, "grad_norm": 0.5568059040106657, "learning_rate": 2.552412727905933e-05, "loss": 2.7515, "step": 33662 }, { "epoch": 2.0897014091501647, "grad_norm": 0.6530909570464223, "learning_rate": 2.5520978117305015e-05, "loss": 2.6512, "step": 33663 }, { "epoch": 2.0897634862499226, "grad_norm": 0.5331458315966382, "learning_rate": 2.55178290832637e-05, "loss": 2.7426, "step": 33664 }, { "epoch": 2.0898255633496805, "grad_norm": 0.621744690473991, "learning_rate": 2.55146801769518e-05, "loss": 2.7845, "step": 33665 }, { "epoch": 2.0898876404494384, "grad_norm": 0.5937402875005502, "learning_rate": 2.5511531398385775e-05, "loss": 2.7281, "step": 33666 }, { "epoch": 2.089949717549196, "grad_norm": 0.5572509097112843, "learning_rate": 2.5508382747582032e-05, "loss": 2.7469, "step": 33667 }, { "epoch": 2.090011794648954, "grad_norm": 0.6188787153051981, "learning_rate": 2.5505234224557007e-05, "loss": 2.657, "step": 33668 }, { "epoch": 2.0900738717487117, "grad_norm": 0.5579639699320421, "learning_rate": 2.5502085829327092e-05, "loss": 2.6457, "step": 33669 }, { "epoch": 2.0901359488484696, "grad_norm": 0.62047195499066, "learning_rate": 2.5498937561908774e-05, "loss": 2.6419, "step": 33670 }, { "epoch": 2.0901980259482276, "grad_norm": 0.5560348158817268, "learning_rate": 2.5495789422318417e-05, "loss": 2.8031, "step": 33671 }, { "epoch": 2.0902601030479855, "grad_norm": 0.5685763387380744, "learning_rate": 2.5492641410572505e-05, "loss": 2.6585, "step": 33672 }, { "epoch": 2.0903221801477434, "grad_norm": 0.5220722499573133, "learning_rate": 2.5489493526687426e-05, "loss": 2.6693, "step": 33673 }, { "epoch": 2.0903842572475013, "grad_norm": 0.5375520581163887, "learning_rate": 2.5486345770679588e-05, "loss": 2.5845, "step": 33674 }, { "epoch": 2.0904463343472592, "grad_norm": 0.5096041544229142, "learning_rate": 2.5483198142565455e-05, "loss": 2.704, "step": 33675 }, { "epoch": 2.090508411447017, "grad_norm": 0.532332299604509, "learning_rate": 2.5480050642361432e-05, "loss": 2.6639, "step": 33676 }, { "epoch": 2.090570488546775, "grad_norm": 0.5716021600877992, "learning_rate": 2.547690327008394e-05, "loss": 2.7005, "step": 33677 }, { "epoch": 2.090632565646533, "grad_norm": 0.5095021235503172, "learning_rate": 2.547375602574939e-05, "loss": 2.6841, "step": 33678 }, { "epoch": 2.090694642746291, "grad_norm": 0.5548999809232823, "learning_rate": 2.5470608909374194e-05, "loss": 2.7176, "step": 33679 }, { "epoch": 2.090756719846049, "grad_norm": 0.533518128740427, "learning_rate": 2.54674619209748e-05, "loss": 2.7364, "step": 33680 }, { "epoch": 2.0908187969458067, "grad_norm": 0.5727472199214765, "learning_rate": 2.5464315060567617e-05, "loss": 2.6105, "step": 33681 }, { "epoch": 2.0908808740455647, "grad_norm": 0.5398082826393765, "learning_rate": 2.546116832816906e-05, "loss": 2.7154, "step": 33682 }, { "epoch": 2.0909429511453226, "grad_norm": 0.6180967803545977, "learning_rate": 2.545802172379552e-05, "loss": 2.7289, "step": 33683 }, { "epoch": 2.0910050282450805, "grad_norm": 0.5281232653256431, "learning_rate": 2.545487524746346e-05, "loss": 2.7671, "step": 33684 }, { "epoch": 2.0910671053448384, "grad_norm": 0.5618443376536049, "learning_rate": 2.5451728899189274e-05, "loss": 2.6786, "step": 33685 }, { "epoch": 2.0911291824445963, "grad_norm": 0.5663758631480553, "learning_rate": 2.5448582678989386e-05, "loss": 2.7598, "step": 33686 }, { "epoch": 2.0911912595443543, "grad_norm": 0.5150149522388517, "learning_rate": 2.5445436586880195e-05, "loss": 2.7161, "step": 33687 }, { "epoch": 2.091253336644112, "grad_norm": 0.6332886164547614, "learning_rate": 2.54422906228781e-05, "loss": 2.7047, "step": 33688 }, { "epoch": 2.09131541374387, "grad_norm": 0.4841055582103694, "learning_rate": 2.5439144786999557e-05, "loss": 2.7232, "step": 33689 }, { "epoch": 2.0913774908436276, "grad_norm": 0.5418135614995297, "learning_rate": 2.5435999079260965e-05, "loss": 2.739, "step": 33690 }, { "epoch": 2.0914395679433855, "grad_norm": 0.49917188254270495, "learning_rate": 2.543285349967872e-05, "loss": 2.6232, "step": 33691 }, { "epoch": 2.0915016450431434, "grad_norm": 0.5058153936052815, "learning_rate": 2.542970804826923e-05, "loss": 2.7257, "step": 33692 }, { "epoch": 2.0915637221429013, "grad_norm": 0.506553762455866, "learning_rate": 2.542656272504893e-05, "loss": 2.6614, "step": 33693 }, { "epoch": 2.0916257992426592, "grad_norm": 0.5300448998289976, "learning_rate": 2.5423417530034222e-05, "loss": 2.6226, "step": 33694 }, { "epoch": 2.091687876342417, "grad_norm": 0.5114151148460447, "learning_rate": 2.5420272463241514e-05, "loss": 2.7552, "step": 33695 }, { "epoch": 2.091749953442175, "grad_norm": 0.5024847933143158, "learning_rate": 2.541712752468719e-05, "loss": 2.5522, "step": 33696 }, { "epoch": 2.091812030541933, "grad_norm": 0.5579910761827185, "learning_rate": 2.5413982714387698e-05, "loss": 2.8315, "step": 33697 }, { "epoch": 2.091874107641691, "grad_norm": 0.49483850826261216, "learning_rate": 2.5410838032359418e-05, "loss": 2.6336, "step": 33698 }, { "epoch": 2.091936184741449, "grad_norm": 0.5049611958617928, "learning_rate": 2.5407693478618777e-05, "loss": 2.6174, "step": 33699 }, { "epoch": 2.0919982618412067, "grad_norm": 0.5329398598693086, "learning_rate": 2.540454905318217e-05, "loss": 2.6519, "step": 33700 }, { "epoch": 2.0920603389409647, "grad_norm": 0.5101873180983039, "learning_rate": 2.540140475606597e-05, "loss": 2.6808, "step": 33701 }, { "epoch": 2.0921224160407226, "grad_norm": 0.5034578975378944, "learning_rate": 2.5398260587286643e-05, "loss": 2.7361, "step": 33702 }, { "epoch": 2.0921844931404805, "grad_norm": 0.5751192466814533, "learning_rate": 2.5395116546860558e-05, "loss": 2.6461, "step": 33703 }, { "epoch": 2.0922465702402384, "grad_norm": 0.5692473615837353, "learning_rate": 2.5391972634804106e-05, "loss": 2.6418, "step": 33704 }, { "epoch": 2.0923086473399963, "grad_norm": 0.5282418691674653, "learning_rate": 2.538882885113373e-05, "loss": 2.7252, "step": 33705 }, { "epoch": 2.0923707244397542, "grad_norm": 0.5403871431965136, "learning_rate": 2.5385685195865777e-05, "loss": 2.6152, "step": 33706 }, { "epoch": 2.092432801539512, "grad_norm": 0.5422963431654112, "learning_rate": 2.5382541669016707e-05, "loss": 2.7814, "step": 33707 }, { "epoch": 2.09249487863927, "grad_norm": 0.49463654311706146, "learning_rate": 2.5379398270602894e-05, "loss": 2.6941, "step": 33708 }, { "epoch": 2.092556955739028, "grad_norm": 0.5251950038059245, "learning_rate": 2.5376255000640736e-05, "loss": 2.6805, "step": 33709 }, { "epoch": 2.092619032838786, "grad_norm": 0.5620700729659109, "learning_rate": 2.5373111859146616e-05, "loss": 2.7838, "step": 33710 }, { "epoch": 2.092681109938544, "grad_norm": 0.535438129227421, "learning_rate": 2.536996884613696e-05, "loss": 2.7473, "step": 33711 }, { "epoch": 2.0927431870383018, "grad_norm": 0.5472971785255332, "learning_rate": 2.536682596162816e-05, "loss": 2.6398, "step": 33712 }, { "epoch": 2.0928052641380597, "grad_norm": 0.5450781594673223, "learning_rate": 2.5363683205636614e-05, "loss": 2.6882, "step": 33713 }, { "epoch": 2.0928673412378176, "grad_norm": 0.5598983735019519, "learning_rate": 2.536054057817871e-05, "loss": 2.8334, "step": 33714 }, { "epoch": 2.092929418337575, "grad_norm": 0.6088579870081844, "learning_rate": 2.535739807927083e-05, "loss": 2.6588, "step": 33715 }, { "epoch": 2.092991495437333, "grad_norm": 0.5610169888336259, "learning_rate": 2.5354255708929405e-05, "loss": 2.6746, "step": 33716 }, { "epoch": 2.093053572537091, "grad_norm": 0.5775086751922056, "learning_rate": 2.5351113467170808e-05, "loss": 2.6677, "step": 33717 }, { "epoch": 2.093115649636849, "grad_norm": 0.5471791647685254, "learning_rate": 2.5347971354011436e-05, "loss": 2.7719, "step": 33718 }, { "epoch": 2.0931777267366067, "grad_norm": 0.6507673346782251, "learning_rate": 2.5344829369467665e-05, "loss": 2.6546, "step": 33719 }, { "epoch": 2.0932398038363647, "grad_norm": 0.6166847450811072, "learning_rate": 2.5341687513555923e-05, "loss": 2.7079, "step": 33720 }, { "epoch": 2.0933018809361226, "grad_norm": 0.6027093581757819, "learning_rate": 2.5338545786292585e-05, "loss": 2.6975, "step": 33721 }, { "epoch": 2.0933639580358805, "grad_norm": 0.63236343085464, "learning_rate": 2.5335404187694033e-05, "loss": 2.7335, "step": 33722 }, { "epoch": 2.0934260351356384, "grad_norm": 0.5241112145362976, "learning_rate": 2.533226271777668e-05, "loss": 2.6631, "step": 33723 }, { "epoch": 2.0934881122353963, "grad_norm": 0.6384270108220684, "learning_rate": 2.5329121376556864e-05, "loss": 2.6934, "step": 33724 }, { "epoch": 2.0935501893351542, "grad_norm": 0.6037037873993998, "learning_rate": 2.5325980164051044e-05, "loss": 2.7587, "step": 33725 }, { "epoch": 2.093612266434912, "grad_norm": 0.5304505840893389, "learning_rate": 2.5322839080275566e-05, "loss": 2.7006, "step": 33726 }, { "epoch": 2.09367434353467, "grad_norm": 0.595037269547012, "learning_rate": 2.5319698125246834e-05, "loss": 2.6445, "step": 33727 }, { "epoch": 2.093736420634428, "grad_norm": 0.5403199663854306, "learning_rate": 2.5316557298981204e-05, "loss": 2.6058, "step": 33728 }, { "epoch": 2.093798497734186, "grad_norm": 0.5387584547543314, "learning_rate": 2.531341660149511e-05, "loss": 2.7068, "step": 33729 }, { "epoch": 2.093860574833944, "grad_norm": 0.609571400806124, "learning_rate": 2.5310276032804907e-05, "loss": 2.764, "step": 33730 }, { "epoch": 2.0939226519337018, "grad_norm": 0.5021672960833026, "learning_rate": 2.5307135592926996e-05, "loss": 2.6048, "step": 33731 }, { "epoch": 2.0939847290334597, "grad_norm": 0.5784367416764311, "learning_rate": 2.5303995281877728e-05, "loss": 2.6079, "step": 33732 }, { "epoch": 2.0940468061332176, "grad_norm": 0.5543175928008991, "learning_rate": 2.530085509967354e-05, "loss": 2.699, "step": 33733 }, { "epoch": 2.0941088832329755, "grad_norm": 0.5961592881957677, "learning_rate": 2.5297715046330772e-05, "loss": 2.6646, "step": 33734 }, { "epoch": 2.0941709603327334, "grad_norm": 0.4778781303405013, "learning_rate": 2.5294575121865834e-05, "loss": 2.7542, "step": 33735 }, { "epoch": 2.0942330374324913, "grad_norm": 0.5221048728272707, "learning_rate": 2.5291435326295088e-05, "loss": 2.6733, "step": 33736 }, { "epoch": 2.0942951145322493, "grad_norm": 0.536935812910962, "learning_rate": 2.528829565963491e-05, "loss": 2.7642, "step": 33737 }, { "epoch": 2.0943571916320067, "grad_norm": 0.5098870962119789, "learning_rate": 2.528515612190168e-05, "loss": 2.6989, "step": 33738 }, { "epoch": 2.0944192687317646, "grad_norm": 0.47572323111828163, "learning_rate": 2.5282016713111824e-05, "loss": 2.6698, "step": 33739 }, { "epoch": 2.0944813458315226, "grad_norm": 0.4793632007524422, "learning_rate": 2.527887743328168e-05, "loss": 2.6437, "step": 33740 }, { "epoch": 2.0945434229312805, "grad_norm": 0.5044837717616301, "learning_rate": 2.527573828242763e-05, "loss": 2.6883, "step": 33741 }, { "epoch": 2.0946055000310384, "grad_norm": 0.4776129480162708, "learning_rate": 2.5272599260566044e-05, "loss": 2.6573, "step": 33742 }, { "epoch": 2.0946675771307963, "grad_norm": 0.4997398921799281, "learning_rate": 2.5269460367713326e-05, "loss": 2.6316, "step": 33743 }, { "epoch": 2.0947296542305542, "grad_norm": 0.4728581336288461, "learning_rate": 2.5266321603885834e-05, "loss": 2.7095, "step": 33744 }, { "epoch": 2.094791731330312, "grad_norm": 0.5065743243410022, "learning_rate": 2.5263182969099953e-05, "loss": 2.6832, "step": 33745 }, { "epoch": 2.09485380843007, "grad_norm": 0.5513365901574451, "learning_rate": 2.5260044463372022e-05, "loss": 2.7312, "step": 33746 }, { "epoch": 2.094915885529828, "grad_norm": 0.5076455942375854, "learning_rate": 2.525690608671848e-05, "loss": 2.7882, "step": 33747 }, { "epoch": 2.094977962629586, "grad_norm": 0.5740111358456239, "learning_rate": 2.525376783915565e-05, "loss": 2.7075, "step": 33748 }, { "epoch": 2.095040039729344, "grad_norm": 0.49945324567640886, "learning_rate": 2.5250629720699926e-05, "loss": 2.5874, "step": 33749 }, { "epoch": 2.0951021168291017, "grad_norm": 0.5681222637349825, "learning_rate": 2.5247491731367678e-05, "loss": 2.7324, "step": 33750 }, { "epoch": 2.0951641939288597, "grad_norm": 0.5350413233849965, "learning_rate": 2.5244353871175253e-05, "loss": 2.6862, "step": 33751 }, { "epoch": 2.0952262710286176, "grad_norm": 0.49183934629027215, "learning_rate": 2.5241216140139062e-05, "loss": 2.6748, "step": 33752 }, { "epoch": 2.0952883481283755, "grad_norm": 0.5669644991913985, "learning_rate": 2.5238078538275456e-05, "loss": 2.6782, "step": 33753 }, { "epoch": 2.0953504252281334, "grad_norm": 0.5748972355191379, "learning_rate": 2.5234941065600815e-05, "loss": 2.6765, "step": 33754 }, { "epoch": 2.0954125023278913, "grad_norm": 0.5228768662837293, "learning_rate": 2.5231803722131464e-05, "loss": 2.7011, "step": 33755 }, { "epoch": 2.0954745794276493, "grad_norm": 0.5722907780141727, "learning_rate": 2.5228666507883837e-05, "loss": 2.7526, "step": 33756 }, { "epoch": 2.095536656527407, "grad_norm": 0.5374340948464339, "learning_rate": 2.522552942287426e-05, "loss": 2.7411, "step": 33757 }, { "epoch": 2.095598733627165, "grad_norm": 0.517875125530503, "learning_rate": 2.5222392467119117e-05, "loss": 2.7629, "step": 33758 }, { "epoch": 2.095660810726923, "grad_norm": 0.6002613213808823, "learning_rate": 2.521925564063475e-05, "loss": 2.6483, "step": 33759 }, { "epoch": 2.095722887826681, "grad_norm": 0.5383307978134362, "learning_rate": 2.521611894343756e-05, "loss": 2.6317, "step": 33760 }, { "epoch": 2.095784964926439, "grad_norm": 0.5265105269956446, "learning_rate": 2.5212982375543893e-05, "loss": 2.631, "step": 33761 }, { "epoch": 2.0958470420261968, "grad_norm": 0.5832163120012989, "learning_rate": 2.520984593697011e-05, "loss": 2.6993, "step": 33762 }, { "epoch": 2.0959091191259542, "grad_norm": 0.519166775540646, "learning_rate": 2.5206709627732582e-05, "loss": 2.6099, "step": 33763 }, { "epoch": 2.095971196225712, "grad_norm": 0.4661422992195831, "learning_rate": 2.520357344784765e-05, "loss": 2.6971, "step": 33764 }, { "epoch": 2.09603327332547, "grad_norm": 0.5706800295564646, "learning_rate": 2.5200437397331712e-05, "loss": 2.686, "step": 33765 }, { "epoch": 2.096095350425228, "grad_norm": 0.5939228659505617, "learning_rate": 2.519730147620112e-05, "loss": 2.6855, "step": 33766 }, { "epoch": 2.096157427524986, "grad_norm": 0.5539368937077046, "learning_rate": 2.5194165684472215e-05, "loss": 2.7245, "step": 33767 }, { "epoch": 2.096219504624744, "grad_norm": 0.5824513551349645, "learning_rate": 2.5191030022161355e-05, "loss": 2.6647, "step": 33768 }, { "epoch": 2.0962815817245017, "grad_norm": 0.5212407775447258, "learning_rate": 2.518789448928492e-05, "loss": 2.6569, "step": 33769 }, { "epoch": 2.0963436588242597, "grad_norm": 0.5719116209634749, "learning_rate": 2.518475908585928e-05, "loss": 2.7486, "step": 33770 }, { "epoch": 2.0964057359240176, "grad_norm": 0.46841343877071867, "learning_rate": 2.5181623811900768e-05, "loss": 2.7195, "step": 33771 }, { "epoch": 2.0964678130237755, "grad_norm": 0.5320134847534586, "learning_rate": 2.517848866742576e-05, "loss": 2.7256, "step": 33772 }, { "epoch": 2.0965298901235334, "grad_norm": 0.5435046563269269, "learning_rate": 2.5175353652450595e-05, "loss": 2.6589, "step": 33773 }, { "epoch": 2.0965919672232913, "grad_norm": 0.5397892627738605, "learning_rate": 2.517221876699162e-05, "loss": 2.5829, "step": 33774 }, { "epoch": 2.0966540443230492, "grad_norm": 0.547383970742459, "learning_rate": 2.5169084011065224e-05, "loss": 2.688, "step": 33775 }, { "epoch": 2.096716121422807, "grad_norm": 0.5049119715875523, "learning_rate": 2.5165949384687748e-05, "loss": 2.6975, "step": 33776 }, { "epoch": 2.096778198522565, "grad_norm": 0.5466406456300046, "learning_rate": 2.5162814887875542e-05, "loss": 2.6236, "step": 33777 }, { "epoch": 2.096840275622323, "grad_norm": 0.5417326709684976, "learning_rate": 2.5159680520644936e-05, "loss": 2.715, "step": 33778 }, { "epoch": 2.096902352722081, "grad_norm": 0.5133263473788827, "learning_rate": 2.5156546283012327e-05, "loss": 2.7113, "step": 33779 }, { "epoch": 2.096964429821839, "grad_norm": 0.5659554627894372, "learning_rate": 2.515341217499405e-05, "loss": 2.714, "step": 33780 }, { "epoch": 2.0970265069215968, "grad_norm": 0.523343917679915, "learning_rate": 2.515027819660646e-05, "loss": 2.6835, "step": 33781 }, { "epoch": 2.0970885840213547, "grad_norm": 0.5402659651759656, "learning_rate": 2.514714434786587e-05, "loss": 2.6751, "step": 33782 }, { "epoch": 2.0971506611211126, "grad_norm": 0.5937452316092705, "learning_rate": 2.5144010628788684e-05, "loss": 2.7299, "step": 33783 }, { "epoch": 2.0972127382208705, "grad_norm": 0.540370294625001, "learning_rate": 2.5140877039391232e-05, "loss": 2.7878, "step": 33784 }, { "epoch": 2.0972748153206284, "grad_norm": 0.5749020565622558, "learning_rate": 2.513774357968986e-05, "loss": 2.743, "step": 33785 }, { "epoch": 2.097336892420386, "grad_norm": 0.5550835766883633, "learning_rate": 2.5134610249700907e-05, "loss": 2.7227, "step": 33786 }, { "epoch": 2.097398969520144, "grad_norm": 0.5665725463885428, "learning_rate": 2.5131477049440722e-05, "loss": 2.748, "step": 33787 }, { "epoch": 2.0974610466199017, "grad_norm": 0.5739022505049874, "learning_rate": 2.5128343978925666e-05, "loss": 2.6685, "step": 33788 }, { "epoch": 2.0975231237196597, "grad_norm": 0.6125342320588117, "learning_rate": 2.512521103817208e-05, "loss": 2.6647, "step": 33789 }, { "epoch": 2.0975852008194176, "grad_norm": 0.6267670922856907, "learning_rate": 2.512207822719631e-05, "loss": 2.6528, "step": 33790 }, { "epoch": 2.0976472779191755, "grad_norm": 0.5393713190918965, "learning_rate": 2.5118945546014677e-05, "loss": 2.7215, "step": 33791 }, { "epoch": 2.0977093550189334, "grad_norm": 0.5540066960219315, "learning_rate": 2.511581299464356e-05, "loss": 2.8117, "step": 33792 }, { "epoch": 2.0977714321186913, "grad_norm": 0.5122726587484883, "learning_rate": 2.5112680573099297e-05, "loss": 2.657, "step": 33793 }, { "epoch": 2.0978335092184492, "grad_norm": 0.5713360889348702, "learning_rate": 2.5109548281398215e-05, "loss": 2.7002, "step": 33794 }, { "epoch": 2.097895586318207, "grad_norm": 0.5166605174816457, "learning_rate": 2.510641611955664e-05, "loss": 2.6875, "step": 33795 }, { "epoch": 2.097957663417965, "grad_norm": 0.5987232880417626, "learning_rate": 2.5103284087590962e-05, "loss": 2.6937, "step": 33796 }, { "epoch": 2.098019740517723, "grad_norm": 0.552681559812776, "learning_rate": 2.510015218551749e-05, "loss": 2.748, "step": 33797 }, { "epoch": 2.098081817617481, "grad_norm": 0.5474296209746286, "learning_rate": 2.5097020413352568e-05, "loss": 2.6821, "step": 33798 }, { "epoch": 2.098143894717239, "grad_norm": 0.5017765194623036, "learning_rate": 2.5093888771112538e-05, "loss": 2.7257, "step": 33799 }, { "epoch": 2.0982059718169968, "grad_norm": 0.5558857400181312, "learning_rate": 2.509075725881371e-05, "loss": 2.8571, "step": 33800 }, { "epoch": 2.0982680489167547, "grad_norm": 0.4898365427852871, "learning_rate": 2.5087625876472477e-05, "loss": 2.6668, "step": 33801 }, { "epoch": 2.0983301260165126, "grad_norm": 0.5463719915701227, "learning_rate": 2.5084494624105125e-05, "loss": 2.7189, "step": 33802 }, { "epoch": 2.0983922031162705, "grad_norm": 0.5021720438526992, "learning_rate": 2.5081363501728032e-05, "loss": 2.7413, "step": 33803 }, { "epoch": 2.0984542802160284, "grad_norm": 0.5071629404642943, "learning_rate": 2.5078232509357514e-05, "loss": 2.7953, "step": 33804 }, { "epoch": 2.0985163573157863, "grad_norm": 0.4839830042932788, "learning_rate": 2.507510164700989e-05, "loss": 2.5966, "step": 33805 }, { "epoch": 2.0985784344155443, "grad_norm": 0.5320248562824935, "learning_rate": 2.5071970914701533e-05, "loss": 2.644, "step": 33806 }, { "epoch": 2.098640511515302, "grad_norm": 0.5674703139888517, "learning_rate": 2.506884031244875e-05, "loss": 2.7942, "step": 33807 }, { "epoch": 2.09870258861506, "grad_norm": 0.5139937462092283, "learning_rate": 2.5065709840267882e-05, "loss": 2.7229, "step": 33808 }, { "epoch": 2.098764665714818, "grad_norm": 0.5547504926561572, "learning_rate": 2.5062579498175237e-05, "loss": 2.7636, "step": 33809 }, { "epoch": 2.098826742814576, "grad_norm": 0.5136018422472526, "learning_rate": 2.5059449286187188e-05, "loss": 2.7766, "step": 33810 }, { "epoch": 2.0988888199143334, "grad_norm": 0.5369941726595708, "learning_rate": 2.5056319204320045e-05, "loss": 2.7176, "step": 33811 }, { "epoch": 2.0989508970140913, "grad_norm": 0.5273750295898979, "learning_rate": 2.5053189252590147e-05, "loss": 2.7161, "step": 33812 }, { "epoch": 2.0990129741138492, "grad_norm": 0.5911929130011988, "learning_rate": 2.505005943101381e-05, "loss": 2.6698, "step": 33813 }, { "epoch": 2.099075051213607, "grad_norm": 0.5380984545461668, "learning_rate": 2.5046929739607355e-05, "loss": 2.7927, "step": 33814 }, { "epoch": 2.099137128313365, "grad_norm": 0.5300457994447729, "learning_rate": 2.5043800178387143e-05, "loss": 2.7265, "step": 33815 }, { "epoch": 2.099199205413123, "grad_norm": 0.53158337286908, "learning_rate": 2.5040670747369482e-05, "loss": 2.6344, "step": 33816 }, { "epoch": 2.099261282512881, "grad_norm": 0.5874500841890542, "learning_rate": 2.50375414465707e-05, "loss": 2.7347, "step": 33817 }, { "epoch": 2.099323359612639, "grad_norm": 0.5020623885992405, "learning_rate": 2.5034412276007114e-05, "loss": 2.586, "step": 33818 }, { "epoch": 2.0993854367123967, "grad_norm": 0.5821324458892161, "learning_rate": 2.5031283235695068e-05, "loss": 2.7035, "step": 33819 }, { "epoch": 2.0994475138121547, "grad_norm": 0.5350521292646679, "learning_rate": 2.5028154325650883e-05, "loss": 2.7168, "step": 33820 }, { "epoch": 2.0995095909119126, "grad_norm": 0.5134643889878062, "learning_rate": 2.5025025545890878e-05, "loss": 2.6757, "step": 33821 }, { "epoch": 2.0995716680116705, "grad_norm": 0.5703467789890837, "learning_rate": 2.502189689643138e-05, "loss": 2.7175, "step": 33822 }, { "epoch": 2.0996337451114284, "grad_norm": 0.48354178721064695, "learning_rate": 2.5018768377288686e-05, "loss": 2.7006, "step": 33823 }, { "epoch": 2.0996958222111863, "grad_norm": 0.5400877615026035, "learning_rate": 2.5015639988479167e-05, "loss": 2.7776, "step": 33824 }, { "epoch": 2.0997578993109443, "grad_norm": 0.516694722131699, "learning_rate": 2.5012511730019118e-05, "loss": 2.6843, "step": 33825 }, { "epoch": 2.099819976410702, "grad_norm": 0.48128852070803535, "learning_rate": 2.5009383601924853e-05, "loss": 2.7305, "step": 33826 }, { "epoch": 2.09988205351046, "grad_norm": 0.5241732479535397, "learning_rate": 2.5006255604212693e-05, "loss": 2.6984, "step": 33827 }, { "epoch": 2.099944130610218, "grad_norm": 0.4427163927429219, "learning_rate": 2.5003127736898974e-05, "loss": 2.7201, "step": 33828 }, { "epoch": 2.100006207709976, "grad_norm": 0.48151847484634186, "learning_rate": 2.500000000000001e-05, "loss": 2.7443, "step": 33829 }, { "epoch": 2.100068284809734, "grad_norm": 0.5232216688962253, "learning_rate": 2.499687239353212e-05, "loss": 2.7014, "step": 33830 }, { "epoch": 2.1001303619094918, "grad_norm": 0.454431541078522, "learning_rate": 2.4993744917511592e-05, "loss": 2.6963, "step": 33831 }, { "epoch": 2.1001924390092497, "grad_norm": 0.5569045662443235, "learning_rate": 2.4990617571954784e-05, "loss": 2.7384, "step": 33832 }, { "epoch": 2.1002545161090076, "grad_norm": 0.4885100782168882, "learning_rate": 2.4987490356877996e-05, "loss": 2.6633, "step": 33833 }, { "epoch": 2.100316593208765, "grad_norm": 0.4782869220635339, "learning_rate": 2.4984363272297545e-05, "loss": 2.7348, "step": 33834 }, { "epoch": 2.100378670308523, "grad_norm": 0.56757641639111, "learning_rate": 2.498123631822973e-05, "loss": 2.6661, "step": 33835 }, { "epoch": 2.100440747408281, "grad_norm": 0.5078983234361323, "learning_rate": 2.497810949469089e-05, "loss": 2.72, "step": 33836 }, { "epoch": 2.100502824508039, "grad_norm": 0.5291126958352329, "learning_rate": 2.497498280169731e-05, "loss": 2.6174, "step": 33837 }, { "epoch": 2.1005649016077967, "grad_norm": 0.6165139107909825, "learning_rate": 2.497185623926534e-05, "loss": 2.7471, "step": 33838 }, { "epoch": 2.1006269787075547, "grad_norm": 0.5168653654702006, "learning_rate": 2.4968729807411266e-05, "loss": 2.745, "step": 33839 }, { "epoch": 2.1006890558073126, "grad_norm": 0.556163060970048, "learning_rate": 2.4965603506151414e-05, "loss": 2.7098, "step": 33840 }, { "epoch": 2.1007511329070705, "grad_norm": 0.5175300719788664, "learning_rate": 2.496247733550206e-05, "loss": 2.7251, "step": 33841 }, { "epoch": 2.1008132100068284, "grad_norm": 0.5444436658261153, "learning_rate": 2.4959351295479566e-05, "loss": 2.7364, "step": 33842 }, { "epoch": 2.1008752871065863, "grad_norm": 0.5334597239141623, "learning_rate": 2.4956225386100208e-05, "loss": 2.7017, "step": 33843 }, { "epoch": 2.1009373642063442, "grad_norm": 0.527558205407998, "learning_rate": 2.4953099607380303e-05, "loss": 2.7313, "step": 33844 }, { "epoch": 2.100999441306102, "grad_norm": 0.5685991129002476, "learning_rate": 2.4949973959336143e-05, "loss": 2.6092, "step": 33845 }, { "epoch": 2.10106151840586, "grad_norm": 0.5208070907339332, "learning_rate": 2.4946848441984066e-05, "loss": 2.6693, "step": 33846 }, { "epoch": 2.101123595505618, "grad_norm": 0.5359084140903618, "learning_rate": 2.494372305534036e-05, "loss": 2.7273, "step": 33847 }, { "epoch": 2.101185672605376, "grad_norm": 0.5350908671067898, "learning_rate": 2.494059779942134e-05, "loss": 2.6833, "step": 33848 }, { "epoch": 2.101247749705134, "grad_norm": 0.5235239213799299, "learning_rate": 2.49374726742433e-05, "loss": 2.7205, "step": 33849 }, { "epoch": 2.1013098268048918, "grad_norm": 0.5124122637571294, "learning_rate": 2.4934347679822533e-05, "loss": 2.7553, "step": 33850 }, { "epoch": 2.1013719039046497, "grad_norm": 0.5569831196026576, "learning_rate": 2.4931222816175377e-05, "loss": 2.6694, "step": 33851 }, { "epoch": 2.1014339810044076, "grad_norm": 0.5815150640985006, "learning_rate": 2.4928098083318112e-05, "loss": 2.7246, "step": 33852 }, { "epoch": 2.1014960581041655, "grad_norm": 0.5069644595974977, "learning_rate": 2.492497348126705e-05, "loss": 2.6867, "step": 33853 }, { "epoch": 2.1015581352039234, "grad_norm": 0.6317745927749655, "learning_rate": 2.4921849010038472e-05, "loss": 2.6686, "step": 33854 }, { "epoch": 2.1016202123036813, "grad_norm": 0.5285791126359156, "learning_rate": 2.491872466964871e-05, "loss": 2.6392, "step": 33855 }, { "epoch": 2.1016822894034393, "grad_norm": 0.5628491748423211, "learning_rate": 2.4915600460114052e-05, "loss": 2.7606, "step": 33856 }, { "epoch": 2.101744366503197, "grad_norm": 0.5600079860288121, "learning_rate": 2.49124763814508e-05, "loss": 2.7576, "step": 33857 }, { "epoch": 2.1018064436029547, "grad_norm": 0.5516780559318761, "learning_rate": 2.4909352433675242e-05, "loss": 2.6598, "step": 33858 }, { "epoch": 2.1018685207027126, "grad_norm": 0.5247963021974574, "learning_rate": 2.490622861680367e-05, "loss": 2.6255, "step": 33859 }, { "epoch": 2.1019305978024705, "grad_norm": 0.577239569941909, "learning_rate": 2.490310493085241e-05, "loss": 2.7157, "step": 33860 }, { "epoch": 2.1019926749022284, "grad_norm": 0.540967808342119, "learning_rate": 2.4899981375837745e-05, "loss": 2.6789, "step": 33861 }, { "epoch": 2.1020547520019863, "grad_norm": 0.5625112585424533, "learning_rate": 2.4896857951775972e-05, "loss": 2.7328, "step": 33862 }, { "epoch": 2.1021168291017442, "grad_norm": 0.5457706472446504, "learning_rate": 2.4893734658683365e-05, "loss": 2.6653, "step": 33863 }, { "epoch": 2.102178906201502, "grad_norm": 0.5555592791776907, "learning_rate": 2.489061149657626e-05, "loss": 2.6851, "step": 33864 }, { "epoch": 2.10224098330126, "grad_norm": 0.5438871506666558, "learning_rate": 2.4887488465470926e-05, "loss": 2.6973, "step": 33865 }, { "epoch": 2.102303060401018, "grad_norm": 0.5798757694647968, "learning_rate": 2.4884365565383667e-05, "loss": 2.7638, "step": 33866 }, { "epoch": 2.102365137500776, "grad_norm": 0.49460545264897, "learning_rate": 2.4881242796330744e-05, "loss": 2.6873, "step": 33867 }, { "epoch": 2.102427214600534, "grad_norm": 0.5891788094639282, "learning_rate": 2.487812015832848e-05, "loss": 2.7696, "step": 33868 }, { "epoch": 2.1024892917002918, "grad_norm": 0.5514305832321081, "learning_rate": 2.4874997651393178e-05, "loss": 2.6645, "step": 33869 }, { "epoch": 2.1025513688000497, "grad_norm": 0.49570234883325054, "learning_rate": 2.4871875275541112e-05, "loss": 2.7534, "step": 33870 }, { "epoch": 2.1026134458998076, "grad_norm": 0.5556597717870368, "learning_rate": 2.4868753030788577e-05, "loss": 2.6697, "step": 33871 }, { "epoch": 2.1026755229995655, "grad_norm": 0.5510397500603942, "learning_rate": 2.4865630917151856e-05, "loss": 2.7145, "step": 33872 }, { "epoch": 2.1027376000993234, "grad_norm": 0.515855795330665, "learning_rate": 2.4862508934647216e-05, "loss": 2.6767, "step": 33873 }, { "epoch": 2.1027996771990813, "grad_norm": 0.5413105570458009, "learning_rate": 2.4859387083291e-05, "loss": 2.6219, "step": 33874 }, { "epoch": 2.1028617542988393, "grad_norm": 0.5354180913626669, "learning_rate": 2.4856265363099452e-05, "loss": 2.6688, "step": 33875 }, { "epoch": 2.102923831398597, "grad_norm": 0.5119660631626338, "learning_rate": 2.4853143774088877e-05, "loss": 2.6979, "step": 33876 }, { "epoch": 2.102985908498355, "grad_norm": 0.5478660595177658, "learning_rate": 2.485002231627554e-05, "loss": 2.6794, "step": 33877 }, { "epoch": 2.103047985598113, "grad_norm": 0.5393748276976557, "learning_rate": 2.4846900989675753e-05, "loss": 2.6845, "step": 33878 }, { "epoch": 2.103110062697871, "grad_norm": 0.5482058543963235, "learning_rate": 2.48437797943058e-05, "loss": 2.7237, "step": 33879 }, { "epoch": 2.103172139797629, "grad_norm": 0.5873208056750919, "learning_rate": 2.484065873018195e-05, "loss": 2.7299, "step": 33880 }, { "epoch": 2.1032342168973868, "grad_norm": 0.5025083980997247, "learning_rate": 2.4837537797320463e-05, "loss": 2.7091, "step": 33881 }, { "epoch": 2.1032962939971442, "grad_norm": 0.5327563247824355, "learning_rate": 2.483441699573768e-05, "loss": 2.6879, "step": 33882 }, { "epoch": 2.103358371096902, "grad_norm": 0.5363887044168888, "learning_rate": 2.4831296325449848e-05, "loss": 2.6874, "step": 33883 }, { "epoch": 2.10342044819666, "grad_norm": 0.5841546719372177, "learning_rate": 2.4828175786473258e-05, "loss": 2.7663, "step": 33884 }, { "epoch": 2.103482525296418, "grad_norm": 0.4971450869334261, "learning_rate": 2.482505537882418e-05, "loss": 2.6809, "step": 33885 }, { "epoch": 2.103544602396176, "grad_norm": 0.569556819933076, "learning_rate": 2.4821935102518884e-05, "loss": 2.6844, "step": 33886 }, { "epoch": 2.103606679495934, "grad_norm": 0.4978194222265318, "learning_rate": 2.4818814957573683e-05, "loss": 2.7287, "step": 33887 }, { "epoch": 2.1036687565956917, "grad_norm": 0.5238378616327026, "learning_rate": 2.4815694944004837e-05, "loss": 2.7291, "step": 33888 }, { "epoch": 2.1037308336954497, "grad_norm": 0.5617537581738675, "learning_rate": 2.481257506182862e-05, "loss": 2.6832, "step": 33889 }, { "epoch": 2.1037929107952076, "grad_norm": 0.5123757083717618, "learning_rate": 2.4809455311061297e-05, "loss": 2.766, "step": 33890 }, { "epoch": 2.1038549878949655, "grad_norm": 0.47669085353173357, "learning_rate": 2.4806335691719186e-05, "loss": 2.6426, "step": 33891 }, { "epoch": 2.1039170649947234, "grad_norm": 0.5609969614970017, "learning_rate": 2.480321620381853e-05, "loss": 2.762, "step": 33892 }, { "epoch": 2.1039791420944813, "grad_norm": 0.5131761312622708, "learning_rate": 2.4800096847375615e-05, "loss": 2.6339, "step": 33893 }, { "epoch": 2.1040412191942393, "grad_norm": 0.5441976677542847, "learning_rate": 2.4796977622406708e-05, "loss": 2.7511, "step": 33894 }, { "epoch": 2.104103296293997, "grad_norm": 0.5353206465547184, "learning_rate": 2.4793858528928076e-05, "loss": 2.7349, "step": 33895 }, { "epoch": 2.104165373393755, "grad_norm": 0.4991413283272226, "learning_rate": 2.4790739566956018e-05, "loss": 2.6491, "step": 33896 }, { "epoch": 2.104227450493513, "grad_norm": 0.519142521724867, "learning_rate": 2.478762073650679e-05, "loss": 2.7469, "step": 33897 }, { "epoch": 2.104289527593271, "grad_norm": 0.531353463588305, "learning_rate": 2.4784502037596664e-05, "loss": 2.677, "step": 33898 }, { "epoch": 2.104351604693029, "grad_norm": 0.5172368201857821, "learning_rate": 2.4781383470241897e-05, "loss": 2.632, "step": 33899 }, { "epoch": 2.1044136817927868, "grad_norm": 0.5289029559394889, "learning_rate": 2.477826503445877e-05, "loss": 2.6669, "step": 33900 }, { "epoch": 2.1044757588925447, "grad_norm": 0.5258087423088973, "learning_rate": 2.4775146730263582e-05, "loss": 2.6646, "step": 33901 }, { "epoch": 2.1045378359923026, "grad_norm": 0.5332949772344139, "learning_rate": 2.4772028557672573e-05, "loss": 2.5834, "step": 33902 }, { "epoch": 2.1045999130920605, "grad_norm": 0.526197133523549, "learning_rate": 2.4768910516702014e-05, "loss": 2.6973, "step": 33903 }, { "epoch": 2.1046619901918184, "grad_norm": 0.4863103074329673, "learning_rate": 2.476579260736816e-05, "loss": 2.6958, "step": 33904 }, { "epoch": 2.1047240672915764, "grad_norm": 0.5068879146788621, "learning_rate": 2.476267482968731e-05, "loss": 2.65, "step": 33905 }, { "epoch": 2.104786144391334, "grad_norm": 0.5827556940370876, "learning_rate": 2.4759557183675708e-05, "loss": 2.746, "step": 33906 }, { "epoch": 2.1048482214910917, "grad_norm": 0.4571615506132883, "learning_rate": 2.4756439669349625e-05, "loss": 2.6565, "step": 33907 }, { "epoch": 2.1049102985908497, "grad_norm": 0.5473587621252173, "learning_rate": 2.4753322286725327e-05, "loss": 2.8204, "step": 33908 }, { "epoch": 2.1049723756906076, "grad_norm": 0.47450595874984197, "learning_rate": 2.475020503581905e-05, "loss": 2.6225, "step": 33909 }, { "epoch": 2.1050344527903655, "grad_norm": 0.544702766147619, "learning_rate": 2.474708791664711e-05, "loss": 2.6483, "step": 33910 }, { "epoch": 2.1050965298901234, "grad_norm": 0.519968981367906, "learning_rate": 2.474397092922574e-05, "loss": 2.7194, "step": 33911 }, { "epoch": 2.1051586069898813, "grad_norm": 0.5365914084057011, "learning_rate": 2.47408540735712e-05, "loss": 2.6515, "step": 33912 }, { "epoch": 2.1052206840896392, "grad_norm": 0.5381670715863707, "learning_rate": 2.473773734969974e-05, "loss": 2.6851, "step": 33913 }, { "epoch": 2.105282761189397, "grad_norm": 0.5303558229408886, "learning_rate": 2.4734620757627657e-05, "loss": 2.6591, "step": 33914 }, { "epoch": 2.105344838289155, "grad_norm": 0.5414546741226427, "learning_rate": 2.4731504297371194e-05, "loss": 2.6493, "step": 33915 }, { "epoch": 2.105406915388913, "grad_norm": 0.5466907735874634, "learning_rate": 2.4728387968946602e-05, "loss": 2.6715, "step": 33916 }, { "epoch": 2.105468992488671, "grad_norm": 0.5180826858709539, "learning_rate": 2.472527177237013e-05, "loss": 2.7335, "step": 33917 }, { "epoch": 2.105531069588429, "grad_norm": 0.5075235278214628, "learning_rate": 2.472215570765807e-05, "loss": 2.7364, "step": 33918 }, { "epoch": 2.1055931466881868, "grad_norm": 0.5590572434973053, "learning_rate": 2.4719039774826653e-05, "loss": 2.6407, "step": 33919 }, { "epoch": 2.1056552237879447, "grad_norm": 0.48095405693877274, "learning_rate": 2.4715923973892148e-05, "loss": 2.5588, "step": 33920 }, { "epoch": 2.1057173008877026, "grad_norm": 0.6095766320797832, "learning_rate": 2.4712808304870804e-05, "loss": 2.735, "step": 33921 }, { "epoch": 2.1057793779874605, "grad_norm": 0.5167519363160343, "learning_rate": 2.470969276777886e-05, "loss": 2.64, "step": 33922 }, { "epoch": 2.1058414550872184, "grad_norm": 0.5133300920586414, "learning_rate": 2.4706577362632605e-05, "loss": 2.6875, "step": 33923 }, { "epoch": 2.1059035321869763, "grad_norm": 0.5211138605586607, "learning_rate": 2.4703462089448283e-05, "loss": 2.7359, "step": 33924 }, { "epoch": 2.1059656092867343, "grad_norm": 0.5298711610237893, "learning_rate": 2.470034694824213e-05, "loss": 2.774, "step": 33925 }, { "epoch": 2.106027686386492, "grad_norm": 0.6013085213095302, "learning_rate": 2.4697231939030398e-05, "loss": 2.6634, "step": 33926 }, { "epoch": 2.10608976348625, "grad_norm": 0.5197010766582483, "learning_rate": 2.4694117061829364e-05, "loss": 2.6846, "step": 33927 }, { "epoch": 2.106151840586008, "grad_norm": 0.5929645312012178, "learning_rate": 2.4691002316655265e-05, "loss": 2.702, "step": 33928 }, { "epoch": 2.106213917685766, "grad_norm": 0.5557141809370592, "learning_rate": 2.4687887703524355e-05, "loss": 2.6192, "step": 33929 }, { "epoch": 2.1062759947855234, "grad_norm": 0.5551493159472266, "learning_rate": 2.4684773222452855e-05, "loss": 2.6323, "step": 33930 }, { "epoch": 2.1063380718852813, "grad_norm": 0.52022397316351, "learning_rate": 2.4681658873457058e-05, "loss": 2.6902, "step": 33931 }, { "epoch": 2.1064001489850392, "grad_norm": 0.6172347714634077, "learning_rate": 2.46785446565532e-05, "loss": 2.6858, "step": 33932 }, { "epoch": 2.106462226084797, "grad_norm": 0.5320543763096066, "learning_rate": 2.4675430571757495e-05, "loss": 2.6885, "step": 33933 }, { "epoch": 2.106524303184555, "grad_norm": 0.504183050026058, "learning_rate": 2.467231661908624e-05, "loss": 2.6709, "step": 33934 }, { "epoch": 2.106586380284313, "grad_norm": 0.6119268489686437, "learning_rate": 2.4669202798555652e-05, "loss": 2.631, "step": 33935 }, { "epoch": 2.106648457384071, "grad_norm": 0.49592247957632474, "learning_rate": 2.4666089110181965e-05, "loss": 2.7669, "step": 33936 }, { "epoch": 2.106710534483829, "grad_norm": 0.5598846138709803, "learning_rate": 2.4662975553981464e-05, "loss": 2.6933, "step": 33937 }, { "epoch": 2.1067726115835868, "grad_norm": 0.5352278334363052, "learning_rate": 2.465986212997037e-05, "loss": 2.7087, "step": 33938 }, { "epoch": 2.1068346886833447, "grad_norm": 0.5347761488910664, "learning_rate": 2.4656748838164924e-05, "loss": 2.6413, "step": 33939 }, { "epoch": 2.1068967657831026, "grad_norm": 0.5216528802437338, "learning_rate": 2.4653635678581357e-05, "loss": 2.747, "step": 33940 }, { "epoch": 2.1069588428828605, "grad_norm": 0.5717841456633341, "learning_rate": 2.4650522651235936e-05, "loss": 2.7457, "step": 33941 }, { "epoch": 2.1070209199826184, "grad_norm": 0.4839500747607694, "learning_rate": 2.46474097561449e-05, "loss": 2.6001, "step": 33942 }, { "epoch": 2.1070829970823763, "grad_norm": 0.494169377632596, "learning_rate": 2.4644296993324478e-05, "loss": 2.6356, "step": 33943 }, { "epoch": 2.1071450741821343, "grad_norm": 0.5689233033794395, "learning_rate": 2.4641184362790913e-05, "loss": 2.7202, "step": 33944 }, { "epoch": 2.107207151281892, "grad_norm": 0.489412431783469, "learning_rate": 2.4638071864560432e-05, "loss": 2.6811, "step": 33945 }, { "epoch": 2.10726922838165, "grad_norm": 0.506262305781246, "learning_rate": 2.46349594986493e-05, "loss": 2.6955, "step": 33946 }, { "epoch": 2.107331305481408, "grad_norm": 0.5322977280940394, "learning_rate": 2.4631847265073742e-05, "loss": 2.7007, "step": 33947 }, { "epoch": 2.107393382581166, "grad_norm": 0.5564497235526012, "learning_rate": 2.4628735163850002e-05, "loss": 2.7015, "step": 33948 }, { "epoch": 2.107455459680924, "grad_norm": 0.5036140600186981, "learning_rate": 2.4625623194994286e-05, "loss": 2.599, "step": 33949 }, { "epoch": 2.1075175367806818, "grad_norm": 0.49699928960423195, "learning_rate": 2.462251135852287e-05, "loss": 2.7698, "step": 33950 }, { "epoch": 2.1075796138804397, "grad_norm": 0.554049729675015, "learning_rate": 2.4619399654451974e-05, "loss": 2.7398, "step": 33951 }, { "epoch": 2.1076416909801976, "grad_norm": 0.5539973895548522, "learning_rate": 2.4616288082797834e-05, "loss": 2.7412, "step": 33952 }, { "epoch": 2.107703768079955, "grad_norm": 0.5268089125376791, "learning_rate": 2.4613176643576664e-05, "loss": 2.6949, "step": 33953 }, { "epoch": 2.107765845179713, "grad_norm": 0.5111616756932289, "learning_rate": 2.4610065336804726e-05, "loss": 2.6902, "step": 33954 }, { "epoch": 2.107827922279471, "grad_norm": 0.5331861891519685, "learning_rate": 2.4606954162498247e-05, "loss": 2.7133, "step": 33955 }, { "epoch": 2.107889999379229, "grad_norm": 0.5111347972465466, "learning_rate": 2.4603843120673454e-05, "loss": 2.7024, "step": 33956 }, { "epoch": 2.1079520764789867, "grad_norm": 0.5276932582466498, "learning_rate": 2.4600732211346573e-05, "loss": 2.7286, "step": 33957 }, { "epoch": 2.1080141535787447, "grad_norm": 0.5305231676013505, "learning_rate": 2.4597621434533818e-05, "loss": 2.7346, "step": 33958 }, { "epoch": 2.1080762306785026, "grad_norm": 0.4955617800844592, "learning_rate": 2.4594510790251457e-05, "loss": 2.7103, "step": 33959 }, { "epoch": 2.1081383077782605, "grad_norm": 0.5179408116235342, "learning_rate": 2.4591400278515702e-05, "loss": 2.7421, "step": 33960 }, { "epoch": 2.1082003848780184, "grad_norm": 0.578384753459726, "learning_rate": 2.458828989934278e-05, "loss": 2.6131, "step": 33961 }, { "epoch": 2.1082624619777763, "grad_norm": 0.48406036987230033, "learning_rate": 2.45851796527489e-05, "loss": 2.7073, "step": 33962 }, { "epoch": 2.1083245390775343, "grad_norm": 0.6004998618371123, "learning_rate": 2.458206953875033e-05, "loss": 2.714, "step": 33963 }, { "epoch": 2.108386616177292, "grad_norm": 0.5823256496915328, "learning_rate": 2.457895955736327e-05, "loss": 2.622, "step": 33964 }, { "epoch": 2.10844869327705, "grad_norm": 0.5748346360467337, "learning_rate": 2.457584970860395e-05, "loss": 2.7845, "step": 33965 }, { "epoch": 2.108510770376808, "grad_norm": 0.534163535813645, "learning_rate": 2.4572739992488568e-05, "loss": 2.666, "step": 33966 }, { "epoch": 2.108572847476566, "grad_norm": 0.5640107371495239, "learning_rate": 2.4569630409033384e-05, "loss": 2.6887, "step": 33967 }, { "epoch": 2.108634924576324, "grad_norm": 0.5324067536947442, "learning_rate": 2.4566520958254625e-05, "loss": 2.6677, "step": 33968 }, { "epoch": 2.1086970016760818, "grad_norm": 0.5200213078571471, "learning_rate": 2.4563411640168504e-05, "loss": 2.7558, "step": 33969 }, { "epoch": 2.1087590787758397, "grad_norm": 0.49256133460062007, "learning_rate": 2.4560302454791235e-05, "loss": 2.7093, "step": 33970 }, { "epoch": 2.1088211558755976, "grad_norm": 0.5331672963831002, "learning_rate": 2.455719340213905e-05, "loss": 2.6769, "step": 33971 }, { "epoch": 2.1088832329753555, "grad_norm": 0.505019916387189, "learning_rate": 2.455408448222814e-05, "loss": 2.6392, "step": 33972 }, { "epoch": 2.1089453100751134, "grad_norm": 0.5421735030684406, "learning_rate": 2.455097569507477e-05, "loss": 2.6758, "step": 33973 }, { "epoch": 2.1090073871748714, "grad_norm": 0.5345114561607114, "learning_rate": 2.4547867040695134e-05, "loss": 2.7783, "step": 33974 }, { "epoch": 2.1090694642746293, "grad_norm": 0.509167667623099, "learning_rate": 2.4544758519105454e-05, "loss": 2.6682, "step": 33975 }, { "epoch": 2.109131541374387, "grad_norm": 0.5297058889972325, "learning_rate": 2.4541650130321937e-05, "loss": 2.6859, "step": 33976 }, { "epoch": 2.109193618474145, "grad_norm": 0.4865402310664844, "learning_rate": 2.453854187436082e-05, "loss": 2.7473, "step": 33977 }, { "epoch": 2.1092556955739026, "grad_norm": 0.5453976977230841, "learning_rate": 2.4535433751238318e-05, "loss": 2.7354, "step": 33978 }, { "epoch": 2.1093177726736605, "grad_norm": 0.5554435842315375, "learning_rate": 2.4532325760970636e-05, "loss": 2.7466, "step": 33979 }, { "epoch": 2.1093798497734184, "grad_norm": 0.4969551237750086, "learning_rate": 2.4529217903573974e-05, "loss": 2.6807, "step": 33980 }, { "epoch": 2.1094419268731763, "grad_norm": 0.5633489507235243, "learning_rate": 2.4526110179064594e-05, "loss": 2.6964, "step": 33981 }, { "epoch": 2.1095040039729342, "grad_norm": 0.5632417080315839, "learning_rate": 2.452300258745867e-05, "loss": 2.7603, "step": 33982 }, { "epoch": 2.109566081072692, "grad_norm": 0.4610761983356066, "learning_rate": 2.451989512877243e-05, "loss": 2.6411, "step": 33983 }, { "epoch": 2.10962815817245, "grad_norm": 0.5394921180283674, "learning_rate": 2.4516787803022084e-05, "loss": 2.6036, "step": 33984 }, { "epoch": 2.109690235272208, "grad_norm": 0.5312846396877561, "learning_rate": 2.451368061022382e-05, "loss": 2.6742, "step": 33985 }, { "epoch": 2.109752312371966, "grad_norm": 0.5327493602224346, "learning_rate": 2.4510573550393896e-05, "loss": 2.6832, "step": 33986 }, { "epoch": 2.109814389471724, "grad_norm": 0.5379265412395198, "learning_rate": 2.450746662354849e-05, "loss": 2.6977, "step": 33987 }, { "epoch": 2.1098764665714818, "grad_norm": 0.5387980323763364, "learning_rate": 2.4504359829703822e-05, "loss": 2.6841, "step": 33988 }, { "epoch": 2.1099385436712397, "grad_norm": 0.5393312585892555, "learning_rate": 2.4501253168876083e-05, "loss": 2.7419, "step": 33989 }, { "epoch": 2.1100006207709976, "grad_norm": 0.5106760131679476, "learning_rate": 2.449814664108151e-05, "loss": 2.7072, "step": 33990 }, { "epoch": 2.1100626978707555, "grad_norm": 0.5241167017477442, "learning_rate": 2.4495040246336298e-05, "loss": 2.8152, "step": 33991 }, { "epoch": 2.1101247749705134, "grad_norm": 0.5462489768396218, "learning_rate": 2.4491933984656657e-05, "loss": 2.6902, "step": 33992 }, { "epoch": 2.1101868520702713, "grad_norm": 0.5469318632808141, "learning_rate": 2.4488827856058784e-05, "loss": 2.7287, "step": 33993 }, { "epoch": 2.1102489291700293, "grad_norm": 0.560808180669631, "learning_rate": 2.448572186055887e-05, "loss": 2.7185, "step": 33994 }, { "epoch": 2.110311006269787, "grad_norm": 0.49757823961459957, "learning_rate": 2.448261599817316e-05, "loss": 2.706, "step": 33995 }, { "epoch": 2.110373083369545, "grad_norm": 0.5439407101886763, "learning_rate": 2.447951026891784e-05, "loss": 2.7024, "step": 33996 }, { "epoch": 2.110435160469303, "grad_norm": 0.5679765441023928, "learning_rate": 2.44764046728091e-05, "loss": 2.6754, "step": 33997 }, { "epoch": 2.110497237569061, "grad_norm": 0.5477716452231863, "learning_rate": 2.4473299209863142e-05, "loss": 2.672, "step": 33998 }, { "epoch": 2.110559314668819, "grad_norm": 0.5603364822097627, "learning_rate": 2.4470193880096176e-05, "loss": 2.7052, "step": 33999 }, { "epoch": 2.1106213917685768, "grad_norm": 0.5474094231900047, "learning_rate": 2.446708868352443e-05, "loss": 2.6772, "step": 34000 }, { "epoch": 2.1106834688683342, "grad_norm": 0.5792878287207736, "learning_rate": 2.4463983620164078e-05, "loss": 2.7152, "step": 34001 }, { "epoch": 2.110745545968092, "grad_norm": 0.5668223989827196, "learning_rate": 2.446087869003132e-05, "loss": 2.6273, "step": 34002 }, { "epoch": 2.11080762306785, "grad_norm": 0.5616419924099734, "learning_rate": 2.445777389314234e-05, "loss": 2.7013, "step": 34003 }, { "epoch": 2.110869700167608, "grad_norm": 0.5897408705162801, "learning_rate": 2.4454669229513372e-05, "loss": 2.7151, "step": 34004 }, { "epoch": 2.110931777267366, "grad_norm": 0.5740830397318817, "learning_rate": 2.4451564699160595e-05, "loss": 2.6674, "step": 34005 }, { "epoch": 2.110993854367124, "grad_norm": 0.5789669195551754, "learning_rate": 2.444846030210021e-05, "loss": 2.6723, "step": 34006 }, { "epoch": 2.1110559314668818, "grad_norm": 0.5727745371141137, "learning_rate": 2.444535603834841e-05, "loss": 2.6571, "step": 34007 }, { "epoch": 2.1111180085666397, "grad_norm": 0.5745350232699403, "learning_rate": 2.444225190792137e-05, "loss": 2.768, "step": 34008 }, { "epoch": 2.1111800856663976, "grad_norm": 0.5948947612272085, "learning_rate": 2.443914791083533e-05, "loss": 2.7447, "step": 34009 }, { "epoch": 2.1112421627661555, "grad_norm": 0.578687368550931, "learning_rate": 2.443604404710646e-05, "loss": 2.6743, "step": 34010 }, { "epoch": 2.1113042398659134, "grad_norm": 0.5565471102146613, "learning_rate": 2.4432940316750953e-05, "loss": 2.7042, "step": 34011 }, { "epoch": 2.1113663169656713, "grad_norm": 0.5417717562726719, "learning_rate": 2.4429836719784983e-05, "loss": 2.6937, "step": 34012 }, { "epoch": 2.1114283940654293, "grad_norm": 0.5598432530989959, "learning_rate": 2.4426733256224778e-05, "loss": 2.7186, "step": 34013 }, { "epoch": 2.111490471165187, "grad_norm": 0.565693558178861, "learning_rate": 2.4423629926086516e-05, "loss": 2.7568, "step": 34014 }, { "epoch": 2.111552548264945, "grad_norm": 0.5627267963111459, "learning_rate": 2.4420526729386385e-05, "loss": 2.7275, "step": 34015 }, { "epoch": 2.111614625364703, "grad_norm": 0.5383016496621355, "learning_rate": 2.441742366614056e-05, "loss": 2.6409, "step": 34016 }, { "epoch": 2.111676702464461, "grad_norm": 0.6074531031980289, "learning_rate": 2.441432073636526e-05, "loss": 2.7848, "step": 34017 }, { "epoch": 2.111738779564219, "grad_norm": 0.5661950605573832, "learning_rate": 2.4411217940076665e-05, "loss": 2.7021, "step": 34018 }, { "epoch": 2.1118008566639768, "grad_norm": 0.5291988380544799, "learning_rate": 2.4408115277290955e-05, "loss": 2.6233, "step": 34019 }, { "epoch": 2.1118629337637347, "grad_norm": 0.5452967895605685, "learning_rate": 2.440501274802432e-05, "loss": 2.7391, "step": 34020 }, { "epoch": 2.1119250108634926, "grad_norm": 0.5385668313980126, "learning_rate": 2.4401910352292927e-05, "loss": 2.7291, "step": 34021 }, { "epoch": 2.1119870879632505, "grad_norm": 0.5226683017954907, "learning_rate": 2.4398808090113006e-05, "loss": 2.6371, "step": 34022 }, { "epoch": 2.1120491650630084, "grad_norm": 0.5843793114079838, "learning_rate": 2.4395705961500713e-05, "loss": 2.7109, "step": 34023 }, { "epoch": 2.1121112421627664, "grad_norm": 0.5585976180362826, "learning_rate": 2.4392603966472242e-05, "loss": 2.5932, "step": 34024 }, { "epoch": 2.1121733192625243, "grad_norm": 0.554792824000394, "learning_rate": 2.438950210504375e-05, "loss": 2.8078, "step": 34025 }, { "epoch": 2.1122353963622817, "grad_norm": 0.515545549918488, "learning_rate": 2.438640037723146e-05, "loss": 2.7287, "step": 34026 }, { "epoch": 2.1122974734620397, "grad_norm": 0.5456543514925868, "learning_rate": 2.4383298783051543e-05, "loss": 2.6621, "step": 34027 }, { "epoch": 2.1123595505617976, "grad_norm": 0.5450058465609046, "learning_rate": 2.438019732252017e-05, "loss": 2.7271, "step": 34028 }, { "epoch": 2.1124216276615555, "grad_norm": 0.5945659351632595, "learning_rate": 2.437709599565353e-05, "loss": 2.7105, "step": 34029 }, { "epoch": 2.1124837047613134, "grad_norm": 0.5525438768993568, "learning_rate": 2.437399480246778e-05, "loss": 2.6939, "step": 34030 }, { "epoch": 2.1125457818610713, "grad_norm": 0.49185463079134145, "learning_rate": 2.4370893742979122e-05, "loss": 2.7125, "step": 34031 }, { "epoch": 2.1126078589608293, "grad_norm": 0.5301997799236513, "learning_rate": 2.436779281720375e-05, "loss": 2.5751, "step": 34032 }, { "epoch": 2.112669936060587, "grad_norm": 0.48521412183301516, "learning_rate": 2.4364692025157826e-05, "loss": 2.7296, "step": 34033 }, { "epoch": 2.112732013160345, "grad_norm": 0.44984371001126483, "learning_rate": 2.436159136685752e-05, "loss": 2.671, "step": 34034 }, { "epoch": 2.112794090260103, "grad_norm": 0.55673225050281, "learning_rate": 2.435849084231901e-05, "loss": 2.6813, "step": 34035 }, { "epoch": 2.112856167359861, "grad_norm": 0.46884906039974233, "learning_rate": 2.4355390451558485e-05, "loss": 2.6098, "step": 34036 }, { "epoch": 2.112918244459619, "grad_norm": 0.47960585355980523, "learning_rate": 2.435229019459212e-05, "loss": 2.611, "step": 34037 }, { "epoch": 2.1129803215593768, "grad_norm": 0.5167666137324142, "learning_rate": 2.4349190071436085e-05, "loss": 2.6953, "step": 34038 }, { "epoch": 2.1130423986591347, "grad_norm": 0.5561962830398401, "learning_rate": 2.4346090082106525e-05, "loss": 2.6836, "step": 34039 }, { "epoch": 2.1131044757588926, "grad_norm": 0.472277103029015, "learning_rate": 2.4342990226619667e-05, "loss": 2.7874, "step": 34040 }, { "epoch": 2.1131665528586505, "grad_norm": 0.5100191735422074, "learning_rate": 2.4339890504991658e-05, "loss": 2.6829, "step": 34041 }, { "epoch": 2.1132286299584084, "grad_norm": 0.4830403552601744, "learning_rate": 2.4336790917238666e-05, "loss": 2.6998, "step": 34042 }, { "epoch": 2.1132907070581664, "grad_norm": 0.4978443107810145, "learning_rate": 2.4333691463376868e-05, "loss": 2.7419, "step": 34043 }, { "epoch": 2.1133527841579243, "grad_norm": 0.5013275107007976, "learning_rate": 2.433059214342241e-05, "loss": 2.6588, "step": 34044 }, { "epoch": 2.113414861257682, "grad_norm": 0.5254082091838411, "learning_rate": 2.4327492957391506e-05, "loss": 2.7529, "step": 34045 }, { "epoch": 2.11347693835744, "grad_norm": 0.5286323345195978, "learning_rate": 2.43243939053003e-05, "loss": 2.6847, "step": 34046 }, { "epoch": 2.113539015457198, "grad_norm": 0.5011270754288981, "learning_rate": 2.4321294987164967e-05, "loss": 2.7506, "step": 34047 }, { "epoch": 2.113601092556956, "grad_norm": 0.5245736729906754, "learning_rate": 2.4318196203001654e-05, "loss": 2.6671, "step": 34048 }, { "epoch": 2.1136631696567134, "grad_norm": 0.5246147467933975, "learning_rate": 2.431509755282656e-05, "loss": 2.7473, "step": 34049 }, { "epoch": 2.1137252467564713, "grad_norm": 0.5219634066869315, "learning_rate": 2.4311999036655843e-05, "loss": 2.6915, "step": 34050 }, { "epoch": 2.1137873238562292, "grad_norm": 0.4972248399284949, "learning_rate": 2.4308900654505662e-05, "loss": 2.6737, "step": 34051 }, { "epoch": 2.113849400955987, "grad_norm": 0.5377483381957726, "learning_rate": 2.430580240639216e-05, "loss": 2.726, "step": 34052 }, { "epoch": 2.113911478055745, "grad_norm": 0.516511905728668, "learning_rate": 2.4302704292331546e-05, "loss": 2.7424, "step": 34053 }, { "epoch": 2.113973555155503, "grad_norm": 0.48479884584808036, "learning_rate": 2.429960631233996e-05, "loss": 2.5737, "step": 34054 }, { "epoch": 2.114035632255261, "grad_norm": 0.5381519810185117, "learning_rate": 2.4296508466433572e-05, "loss": 2.6478, "step": 34055 }, { "epoch": 2.114097709355019, "grad_norm": 0.5086102015680866, "learning_rate": 2.4293410754628538e-05, "loss": 2.6363, "step": 34056 }, { "epoch": 2.1141597864547768, "grad_norm": 0.49981864978269197, "learning_rate": 2.4290313176940997e-05, "loss": 2.7537, "step": 34057 }, { "epoch": 2.1142218635545347, "grad_norm": 0.4981482907429874, "learning_rate": 2.4287215733387155e-05, "loss": 2.721, "step": 34058 }, { "epoch": 2.1142839406542926, "grad_norm": 0.5123828626142487, "learning_rate": 2.4284118423983147e-05, "loss": 2.7049, "step": 34059 }, { "epoch": 2.1143460177540505, "grad_norm": 0.5521557208654615, "learning_rate": 2.428102124874514e-05, "loss": 2.7003, "step": 34060 }, { "epoch": 2.1144080948538084, "grad_norm": 0.4704185353634223, "learning_rate": 2.427792420768927e-05, "loss": 2.7002, "step": 34061 }, { "epoch": 2.1144701719535663, "grad_norm": 0.55837576287403, "learning_rate": 2.427482730083173e-05, "loss": 2.7499, "step": 34062 }, { "epoch": 2.1145322490533243, "grad_norm": 0.5579654654836431, "learning_rate": 2.4271730528188664e-05, "loss": 2.7165, "step": 34063 }, { "epoch": 2.114594326153082, "grad_norm": 0.4455228108143737, "learning_rate": 2.4268633889776205e-05, "loss": 2.6748, "step": 34064 }, { "epoch": 2.11465640325284, "grad_norm": 0.5330946535996952, "learning_rate": 2.4265537385610548e-05, "loss": 2.6765, "step": 34065 }, { "epoch": 2.114718480352598, "grad_norm": 0.5675758650564737, "learning_rate": 2.4262441015707808e-05, "loss": 2.7271, "step": 34066 }, { "epoch": 2.114780557452356, "grad_norm": 0.5347349638748816, "learning_rate": 2.4259344780084185e-05, "loss": 2.6475, "step": 34067 }, { "epoch": 2.114842634552114, "grad_norm": 0.5771372321214211, "learning_rate": 2.425624867875581e-05, "loss": 2.7289, "step": 34068 }, { "epoch": 2.1149047116518718, "grad_norm": 0.534506800498558, "learning_rate": 2.425315271173883e-05, "loss": 2.7279, "step": 34069 }, { "epoch": 2.1149667887516297, "grad_norm": 0.6119438945805119, "learning_rate": 2.425005687904941e-05, "loss": 2.8155, "step": 34070 }, { "epoch": 2.1150288658513876, "grad_norm": 0.5428524029899482, "learning_rate": 2.424696118070367e-05, "loss": 2.5386, "step": 34071 }, { "epoch": 2.1150909429511455, "grad_norm": 0.5424421487453652, "learning_rate": 2.4243865616717802e-05, "loss": 2.6864, "step": 34072 }, { "epoch": 2.1151530200509034, "grad_norm": 0.5522282847965666, "learning_rate": 2.4240770187107943e-05, "loss": 2.7403, "step": 34073 }, { "epoch": 2.115215097150661, "grad_norm": 0.47664847927253073, "learning_rate": 2.423767489189024e-05, "loss": 2.7595, "step": 34074 }, { "epoch": 2.115277174250419, "grad_norm": 0.6011144530080424, "learning_rate": 2.423457973108082e-05, "loss": 2.7307, "step": 34075 }, { "epoch": 2.1153392513501768, "grad_norm": 0.5118567302550175, "learning_rate": 2.4231484704695874e-05, "loss": 2.7283, "step": 34076 }, { "epoch": 2.1154013284499347, "grad_norm": 0.5959601426286358, "learning_rate": 2.422838981275152e-05, "loss": 2.721, "step": 34077 }, { "epoch": 2.1154634055496926, "grad_norm": 0.5356236665012493, "learning_rate": 2.4225295055263926e-05, "loss": 2.6328, "step": 34078 }, { "epoch": 2.1155254826494505, "grad_norm": 0.5681561240975614, "learning_rate": 2.4222200432249214e-05, "loss": 2.7624, "step": 34079 }, { "epoch": 2.1155875597492084, "grad_norm": 0.4843392974247176, "learning_rate": 2.4219105943723524e-05, "loss": 2.5663, "step": 34080 }, { "epoch": 2.1156496368489663, "grad_norm": 0.5642135269992379, "learning_rate": 2.4216011589703037e-05, "loss": 2.7622, "step": 34081 }, { "epoch": 2.1157117139487243, "grad_norm": 0.5274357750741594, "learning_rate": 2.4212917370203874e-05, "loss": 2.6353, "step": 34082 }, { "epoch": 2.115773791048482, "grad_norm": 0.5658170595296118, "learning_rate": 2.4209823285242183e-05, "loss": 2.6568, "step": 34083 }, { "epoch": 2.11583586814824, "grad_norm": 0.5302072276445513, "learning_rate": 2.420672933483409e-05, "loss": 2.6491, "step": 34084 }, { "epoch": 2.115897945247998, "grad_norm": 0.5175141780864487, "learning_rate": 2.4203635518995764e-05, "loss": 2.7372, "step": 34085 }, { "epoch": 2.115960022347756, "grad_norm": 0.5361977348555828, "learning_rate": 2.4200541837743336e-05, "loss": 2.7408, "step": 34086 }, { "epoch": 2.116022099447514, "grad_norm": 0.5218501718976793, "learning_rate": 2.4197448291092943e-05, "loss": 2.6819, "step": 34087 }, { "epoch": 2.1160841765472718, "grad_norm": 0.48679054061449, "learning_rate": 2.419435487906071e-05, "loss": 2.6977, "step": 34088 }, { "epoch": 2.1161462536470297, "grad_norm": 0.4887202921177804, "learning_rate": 2.4191261601662813e-05, "loss": 2.7031, "step": 34089 }, { "epoch": 2.1162083307467876, "grad_norm": 0.5434424988756382, "learning_rate": 2.4188168458915362e-05, "loss": 2.7204, "step": 34090 }, { "epoch": 2.1162704078465455, "grad_norm": 0.5032935150556364, "learning_rate": 2.4185075450834504e-05, "loss": 2.7127, "step": 34091 }, { "epoch": 2.1163324849463034, "grad_norm": 0.45058467353871706, "learning_rate": 2.418198257743638e-05, "loss": 2.675, "step": 34092 }, { "epoch": 2.1163945620460614, "grad_norm": 0.5732065478915882, "learning_rate": 2.4178889838737096e-05, "loss": 2.7653, "step": 34093 }, { "epoch": 2.1164566391458193, "grad_norm": 0.4607183290961042, "learning_rate": 2.4175797234752835e-05, "loss": 2.7001, "step": 34094 }, { "epoch": 2.116518716245577, "grad_norm": 0.5644100925185283, "learning_rate": 2.4172704765499705e-05, "loss": 2.7419, "step": 34095 }, { "epoch": 2.116580793345335, "grad_norm": 0.4902300587498298, "learning_rate": 2.416961243099385e-05, "loss": 2.631, "step": 34096 }, { "epoch": 2.1166428704450926, "grad_norm": 0.5735655981942257, "learning_rate": 2.4166520231251373e-05, "loss": 2.7599, "step": 34097 }, { "epoch": 2.1167049475448505, "grad_norm": 0.5124344807124226, "learning_rate": 2.416342816628843e-05, "loss": 2.7841, "step": 34098 }, { "epoch": 2.1167670246446084, "grad_norm": 0.4796141024672057, "learning_rate": 2.416033623612118e-05, "loss": 2.6628, "step": 34099 }, { "epoch": 2.1168291017443663, "grad_norm": 0.5220792998665251, "learning_rate": 2.415724444076572e-05, "loss": 2.5753, "step": 34100 }, { "epoch": 2.1168911788441243, "grad_norm": 0.4813755441373822, "learning_rate": 2.4154152780238194e-05, "loss": 2.6639, "step": 34101 }, { "epoch": 2.116953255943882, "grad_norm": 0.554260546278081, "learning_rate": 2.415106125455471e-05, "loss": 2.6754, "step": 34102 }, { "epoch": 2.11701533304364, "grad_norm": 0.4926565296959922, "learning_rate": 2.4147969863731428e-05, "loss": 2.7051, "step": 34103 }, { "epoch": 2.117077410143398, "grad_norm": 0.5571260443838771, "learning_rate": 2.4144878607784473e-05, "loss": 2.6556, "step": 34104 }, { "epoch": 2.117139487243156, "grad_norm": 0.5745904977733083, "learning_rate": 2.4141787486729954e-05, "loss": 2.7281, "step": 34105 }, { "epoch": 2.117201564342914, "grad_norm": 0.5260000841934804, "learning_rate": 2.4138696500584006e-05, "loss": 2.6325, "step": 34106 }, { "epoch": 2.1172636414426718, "grad_norm": 0.5668998409157296, "learning_rate": 2.4135605649362742e-05, "loss": 2.6418, "step": 34107 }, { "epoch": 2.1173257185424297, "grad_norm": 0.550150625237506, "learning_rate": 2.4132514933082324e-05, "loss": 2.7219, "step": 34108 }, { "epoch": 2.1173877956421876, "grad_norm": 0.5158268215659961, "learning_rate": 2.412942435175885e-05, "loss": 2.6727, "step": 34109 }, { "epoch": 2.1174498727419455, "grad_norm": 0.5452662101270452, "learning_rate": 2.412633390540845e-05, "loss": 2.7351, "step": 34110 }, { "epoch": 2.1175119498417034, "grad_norm": 0.5067194692751861, "learning_rate": 2.412324359404723e-05, "loss": 2.7143, "step": 34111 }, { "epoch": 2.1175740269414614, "grad_norm": 0.5641182562832613, "learning_rate": 2.4120153417691345e-05, "loss": 2.6744, "step": 34112 }, { "epoch": 2.1176361040412193, "grad_norm": 0.5338331978254159, "learning_rate": 2.4117063376356907e-05, "loss": 2.6985, "step": 34113 }, { "epoch": 2.117698181140977, "grad_norm": 0.5431842787847582, "learning_rate": 2.4113973470060032e-05, "loss": 2.7343, "step": 34114 }, { "epoch": 2.117760258240735, "grad_norm": 0.5635005211984276, "learning_rate": 2.411088369881684e-05, "loss": 2.6673, "step": 34115 }, { "epoch": 2.117822335340493, "grad_norm": 0.49217152948051607, "learning_rate": 2.410779406264343e-05, "loss": 2.7382, "step": 34116 }, { "epoch": 2.117884412440251, "grad_norm": 0.5343825323009008, "learning_rate": 2.410470456155597e-05, "loss": 2.6728, "step": 34117 }, { "epoch": 2.117946489540009, "grad_norm": 0.5359874868853911, "learning_rate": 2.4101615195570544e-05, "loss": 2.6882, "step": 34118 }, { "epoch": 2.1180085666397668, "grad_norm": 0.499957231838258, "learning_rate": 2.4098525964703284e-05, "loss": 2.6436, "step": 34119 }, { "epoch": 2.1180706437395247, "grad_norm": 0.5389148122848869, "learning_rate": 2.4095436868970282e-05, "loss": 2.7441, "step": 34120 }, { "epoch": 2.1181327208392826, "grad_norm": 0.49275970500328425, "learning_rate": 2.4092347908387696e-05, "loss": 2.6935, "step": 34121 }, { "epoch": 2.11819479793904, "grad_norm": 0.5217968865143429, "learning_rate": 2.4089259082971615e-05, "loss": 2.7458, "step": 34122 }, { "epoch": 2.118256875038798, "grad_norm": 0.47082864504423905, "learning_rate": 2.4086170392738167e-05, "loss": 2.697, "step": 34123 }, { "epoch": 2.118318952138556, "grad_norm": 0.510572241085137, "learning_rate": 2.408308183770343e-05, "loss": 2.707, "step": 34124 }, { "epoch": 2.118381029238314, "grad_norm": 0.5385582541955467, "learning_rate": 2.4079993417883573e-05, "loss": 2.6263, "step": 34125 }, { "epoch": 2.1184431063380718, "grad_norm": 0.5210555238443216, "learning_rate": 2.407690513329468e-05, "loss": 2.6561, "step": 34126 }, { "epoch": 2.1185051834378297, "grad_norm": 0.5344839676521362, "learning_rate": 2.407381698395287e-05, "loss": 2.7661, "step": 34127 }, { "epoch": 2.1185672605375876, "grad_norm": 0.5181890660654406, "learning_rate": 2.4070728969874246e-05, "loss": 2.7233, "step": 34128 }, { "epoch": 2.1186293376373455, "grad_norm": 0.5691859209602782, "learning_rate": 2.4067641091074906e-05, "loss": 2.7059, "step": 34129 }, { "epoch": 2.1186914147371034, "grad_norm": 0.4860002109258296, "learning_rate": 2.406455334757098e-05, "loss": 2.6562, "step": 34130 }, { "epoch": 2.1187534918368613, "grad_norm": 0.5683544984588638, "learning_rate": 2.4061465739378584e-05, "loss": 2.6463, "step": 34131 }, { "epoch": 2.1188155689366193, "grad_norm": 0.5349082929230439, "learning_rate": 2.405837826651383e-05, "loss": 2.6285, "step": 34132 }, { "epoch": 2.118877646036377, "grad_norm": 0.4997387292830455, "learning_rate": 2.4055290928992806e-05, "loss": 2.757, "step": 34133 }, { "epoch": 2.118939723136135, "grad_norm": 0.5354067084082759, "learning_rate": 2.4052203726831618e-05, "loss": 2.7176, "step": 34134 }, { "epoch": 2.119001800235893, "grad_norm": 0.5220921150447239, "learning_rate": 2.4049116660046396e-05, "loss": 2.7243, "step": 34135 }, { "epoch": 2.119063877335651, "grad_norm": 0.5129373614828249, "learning_rate": 2.404602972865323e-05, "loss": 2.8408, "step": 34136 }, { "epoch": 2.119125954435409, "grad_norm": 0.5725991120397842, "learning_rate": 2.4042942932668227e-05, "loss": 2.7643, "step": 34137 }, { "epoch": 2.1191880315351668, "grad_norm": 0.5461307092910845, "learning_rate": 2.4039856272107474e-05, "loss": 2.7472, "step": 34138 }, { "epoch": 2.1192501086349247, "grad_norm": 0.5212422069360363, "learning_rate": 2.4036769746987113e-05, "loss": 2.7964, "step": 34139 }, { "epoch": 2.1193121857346826, "grad_norm": 0.5491942681055818, "learning_rate": 2.4033683357323227e-05, "loss": 2.7637, "step": 34140 }, { "epoch": 2.1193742628344405, "grad_norm": 0.5470592460888565, "learning_rate": 2.4030597103131918e-05, "loss": 2.7832, "step": 34141 }, { "epoch": 2.1194363399341984, "grad_norm": 0.5074132083096897, "learning_rate": 2.402751098442929e-05, "loss": 2.6725, "step": 34142 }, { "epoch": 2.1194984170339564, "grad_norm": 0.5519250666889977, "learning_rate": 2.4024425001231425e-05, "loss": 2.6466, "step": 34143 }, { "epoch": 2.1195604941337143, "grad_norm": 0.5073099143514692, "learning_rate": 2.4021339153554456e-05, "loss": 2.7209, "step": 34144 }, { "epoch": 2.1196225712334718, "grad_norm": 0.5127405637117375, "learning_rate": 2.4018253441414468e-05, "loss": 2.6596, "step": 34145 }, { "epoch": 2.1196846483332297, "grad_norm": 0.5753099362939408, "learning_rate": 2.4015167864827557e-05, "loss": 2.6852, "step": 34146 }, { "epoch": 2.1197467254329876, "grad_norm": 0.5086346939411878, "learning_rate": 2.4012082423809805e-05, "loss": 2.6479, "step": 34147 }, { "epoch": 2.1198088025327455, "grad_norm": 0.5487328171701004, "learning_rate": 2.4008997118377352e-05, "loss": 2.7356, "step": 34148 }, { "epoch": 2.1198708796325034, "grad_norm": 0.5097475012118846, "learning_rate": 2.4005911948546268e-05, "loss": 2.6176, "step": 34149 }, { "epoch": 2.1199329567322613, "grad_norm": 0.5019467423920114, "learning_rate": 2.400282691433265e-05, "loss": 2.739, "step": 34150 }, { "epoch": 2.1199950338320193, "grad_norm": 0.58023893651149, "learning_rate": 2.3999742015752575e-05, "loss": 2.838, "step": 34151 }, { "epoch": 2.120057110931777, "grad_norm": 0.5325148055655494, "learning_rate": 2.3996657252822173e-05, "loss": 2.6893, "step": 34152 }, { "epoch": 2.120119188031535, "grad_norm": 0.5221170309607968, "learning_rate": 2.3993572625557532e-05, "loss": 2.7445, "step": 34153 }, { "epoch": 2.120181265131293, "grad_norm": 0.5488373562182401, "learning_rate": 2.399048813397473e-05, "loss": 2.7782, "step": 34154 }, { "epoch": 2.120243342231051, "grad_norm": 0.5850055149014045, "learning_rate": 2.3987403778089868e-05, "loss": 2.6369, "step": 34155 }, { "epoch": 2.120305419330809, "grad_norm": 0.5468274232059975, "learning_rate": 2.3984319557919012e-05, "loss": 2.7406, "step": 34156 }, { "epoch": 2.1203674964305668, "grad_norm": 0.5285696403830338, "learning_rate": 2.3981235473478298e-05, "loss": 2.6922, "step": 34157 }, { "epoch": 2.1204295735303247, "grad_norm": 0.5817807317681153, "learning_rate": 2.39781515247838e-05, "loss": 2.689, "step": 34158 }, { "epoch": 2.1204916506300826, "grad_norm": 0.5227839817037545, "learning_rate": 2.3975067711851595e-05, "loss": 2.6737, "step": 34159 }, { "epoch": 2.1205537277298405, "grad_norm": 0.5691222481781513, "learning_rate": 2.3971984034697763e-05, "loss": 2.7296, "step": 34160 }, { "epoch": 2.1206158048295984, "grad_norm": 0.5431550070108258, "learning_rate": 2.3968900493338426e-05, "loss": 2.653, "step": 34161 }, { "epoch": 2.1206778819293564, "grad_norm": 0.5036812509143648, "learning_rate": 2.396581708778964e-05, "loss": 2.7768, "step": 34162 }, { "epoch": 2.1207399590291143, "grad_norm": 0.5363820926105757, "learning_rate": 2.3962733818067524e-05, "loss": 2.692, "step": 34163 }, { "epoch": 2.120802036128872, "grad_norm": 0.5240140557915544, "learning_rate": 2.3959650684188146e-05, "loss": 2.7516, "step": 34164 }, { "epoch": 2.12086411322863, "grad_norm": 0.49936276300306826, "learning_rate": 2.3956567686167587e-05, "loss": 2.7084, "step": 34165 }, { "epoch": 2.120926190328388, "grad_norm": 0.5403246822715775, "learning_rate": 2.395348482402192e-05, "loss": 2.7083, "step": 34166 }, { "epoch": 2.120988267428146, "grad_norm": 0.5321853201749268, "learning_rate": 2.3950402097767273e-05, "loss": 2.7111, "step": 34167 }, { "epoch": 2.121050344527904, "grad_norm": 0.4979586055549416, "learning_rate": 2.394731950741969e-05, "loss": 2.6741, "step": 34168 }, { "epoch": 2.121112421627662, "grad_norm": 0.5869013678626589, "learning_rate": 2.3944237052995272e-05, "loss": 2.7628, "step": 34169 }, { "epoch": 2.1211744987274193, "grad_norm": 0.4830968050606886, "learning_rate": 2.3941154734510073e-05, "loss": 2.6554, "step": 34170 }, { "epoch": 2.121236575827177, "grad_norm": 0.49238655061676034, "learning_rate": 2.393807255198022e-05, "loss": 2.7533, "step": 34171 }, { "epoch": 2.121298652926935, "grad_norm": 0.5817559890651898, "learning_rate": 2.3934990505421767e-05, "loss": 2.7621, "step": 34172 }, { "epoch": 2.121360730026693, "grad_norm": 0.5080252861822214, "learning_rate": 2.3931908594850794e-05, "loss": 2.5794, "step": 34173 }, { "epoch": 2.121422807126451, "grad_norm": 0.553486396701645, "learning_rate": 2.3928826820283364e-05, "loss": 2.7569, "step": 34174 }, { "epoch": 2.121484884226209, "grad_norm": 0.528276565066135, "learning_rate": 2.392574518173559e-05, "loss": 2.6144, "step": 34175 }, { "epoch": 2.1215469613259668, "grad_norm": 0.54473258391643, "learning_rate": 2.3922663679223533e-05, "loss": 2.7197, "step": 34176 }, { "epoch": 2.1216090384257247, "grad_norm": 0.5675081518402768, "learning_rate": 2.3919582312763278e-05, "loss": 2.7468, "step": 34177 }, { "epoch": 2.1216711155254826, "grad_norm": 0.5344564219678258, "learning_rate": 2.3916501082370884e-05, "loss": 2.7993, "step": 34178 }, { "epoch": 2.1217331926252405, "grad_norm": 0.5417262344129401, "learning_rate": 2.3913419988062424e-05, "loss": 2.6987, "step": 34179 }, { "epoch": 2.1217952697249984, "grad_norm": 0.5622774849632414, "learning_rate": 2.3910339029854002e-05, "loss": 2.6728, "step": 34180 }, { "epoch": 2.1218573468247564, "grad_norm": 0.5057547030171318, "learning_rate": 2.3907258207761667e-05, "loss": 2.6154, "step": 34181 }, { "epoch": 2.1219194239245143, "grad_norm": 0.4921400259222431, "learning_rate": 2.390417752180151e-05, "loss": 2.7214, "step": 34182 }, { "epoch": 2.121981501024272, "grad_norm": 0.49248751321928763, "learning_rate": 2.3901096971989568e-05, "loss": 2.6545, "step": 34183 }, { "epoch": 2.12204357812403, "grad_norm": 0.5260128888356898, "learning_rate": 2.389801655834196e-05, "loss": 2.6985, "step": 34184 }, { "epoch": 2.122105655223788, "grad_norm": 0.48627430261812277, "learning_rate": 2.3894936280874742e-05, "loss": 2.7287, "step": 34185 }, { "epoch": 2.122167732323546, "grad_norm": 0.4488223441526073, "learning_rate": 2.389185613960397e-05, "loss": 2.6573, "step": 34186 }, { "epoch": 2.122229809423304, "grad_norm": 0.5471164715219644, "learning_rate": 2.388877613454571e-05, "loss": 2.6899, "step": 34187 }, { "epoch": 2.1222918865230618, "grad_norm": 0.48076519698504844, "learning_rate": 2.3885696265716058e-05, "loss": 2.7393, "step": 34188 }, { "epoch": 2.1223539636228197, "grad_norm": 0.4941285324124819, "learning_rate": 2.388261653313107e-05, "loss": 2.703, "step": 34189 }, { "epoch": 2.1224160407225776, "grad_norm": 0.529048678493048, "learning_rate": 2.387953693680681e-05, "loss": 2.7616, "step": 34190 }, { "epoch": 2.1224781178223355, "grad_norm": 0.44258099762943026, "learning_rate": 2.3876457476759352e-05, "loss": 2.6737, "step": 34191 }, { "epoch": 2.1225401949220934, "grad_norm": 0.5219096474938656, "learning_rate": 2.3873378153004732e-05, "loss": 2.7656, "step": 34192 }, { "epoch": 2.122602272021851, "grad_norm": 0.4964637127770392, "learning_rate": 2.3870298965559062e-05, "loss": 2.7209, "step": 34193 }, { "epoch": 2.122664349121609, "grad_norm": 0.474182753477202, "learning_rate": 2.386721991443839e-05, "loss": 2.707, "step": 34194 }, { "epoch": 2.1227264262213668, "grad_norm": 0.5194546062180434, "learning_rate": 2.3864140999658747e-05, "loss": 2.7417, "step": 34195 }, { "epoch": 2.1227885033211247, "grad_norm": 0.4758286532385108, "learning_rate": 2.386106222123625e-05, "loss": 2.7077, "step": 34196 }, { "epoch": 2.1228505804208826, "grad_norm": 0.5655581471678996, "learning_rate": 2.3857983579186914e-05, "loss": 2.7071, "step": 34197 }, { "epoch": 2.1229126575206405, "grad_norm": 0.48896915044577377, "learning_rate": 2.3854905073526852e-05, "loss": 2.7165, "step": 34198 }, { "epoch": 2.1229747346203984, "grad_norm": 0.5081277602286167, "learning_rate": 2.385182670427209e-05, "loss": 2.6527, "step": 34199 }, { "epoch": 2.1230368117201563, "grad_norm": 0.5025663378089472, "learning_rate": 2.3848748471438696e-05, "loss": 2.6326, "step": 34200 }, { "epoch": 2.1230988888199143, "grad_norm": 0.5198267619708417, "learning_rate": 2.384567037504271e-05, "loss": 2.6427, "step": 34201 }, { "epoch": 2.123160965919672, "grad_norm": 0.5084692303041772, "learning_rate": 2.3842592415100235e-05, "loss": 2.6709, "step": 34202 }, { "epoch": 2.12322304301943, "grad_norm": 0.5382589626265917, "learning_rate": 2.38395145916273e-05, "loss": 2.653, "step": 34203 }, { "epoch": 2.123285120119188, "grad_norm": 0.532153959309348, "learning_rate": 2.383643690463997e-05, "loss": 2.6452, "step": 34204 }, { "epoch": 2.123347197218946, "grad_norm": 0.5180937016493599, "learning_rate": 2.38333593541543e-05, "loss": 2.8097, "step": 34205 }, { "epoch": 2.123409274318704, "grad_norm": 0.5228745294085926, "learning_rate": 2.3830281940186327e-05, "loss": 2.6863, "step": 34206 }, { "epoch": 2.1234713514184618, "grad_norm": 0.5426963561988503, "learning_rate": 2.3827204662752146e-05, "loss": 2.6249, "step": 34207 }, { "epoch": 2.1235334285182197, "grad_norm": 0.543018338800005, "learning_rate": 2.3824127521867788e-05, "loss": 2.8743, "step": 34208 }, { "epoch": 2.1235955056179776, "grad_norm": 0.6045020723536129, "learning_rate": 2.3821050517549316e-05, "loss": 2.7239, "step": 34209 }, { "epoch": 2.1236575827177355, "grad_norm": 0.5139480254654076, "learning_rate": 2.3817973649812758e-05, "loss": 2.6576, "step": 34210 }, { "epoch": 2.1237196598174934, "grad_norm": 0.5602100436930605, "learning_rate": 2.3814896918674208e-05, "loss": 2.682, "step": 34211 }, { "epoch": 2.1237817369172514, "grad_norm": 0.6533393343906129, "learning_rate": 2.3811820324149696e-05, "loss": 2.8379, "step": 34212 }, { "epoch": 2.1238438140170093, "grad_norm": 0.4934135636798659, "learning_rate": 2.3808743866255275e-05, "loss": 2.6635, "step": 34213 }, { "epoch": 2.123905891116767, "grad_norm": 0.6009870724755484, "learning_rate": 2.3805667545006993e-05, "loss": 2.6962, "step": 34214 }, { "epoch": 2.123967968216525, "grad_norm": 0.5343812774824881, "learning_rate": 2.3802591360420884e-05, "loss": 2.679, "step": 34215 }, { "epoch": 2.124030045316283, "grad_norm": 0.5798606850349044, "learning_rate": 2.379951531251304e-05, "loss": 2.6871, "step": 34216 }, { "epoch": 2.124092122416041, "grad_norm": 0.5525607587883764, "learning_rate": 2.3796439401299482e-05, "loss": 2.703, "step": 34217 }, { "epoch": 2.1241541995157984, "grad_norm": 0.4822944441922034, "learning_rate": 2.3793363626796257e-05, "loss": 2.6345, "step": 34218 }, { "epoch": 2.1242162766155563, "grad_norm": 0.6152689363974813, "learning_rate": 2.37902879890194e-05, "loss": 2.6887, "step": 34219 }, { "epoch": 2.1242783537153143, "grad_norm": 0.5499423246053896, "learning_rate": 2.378721248798499e-05, "loss": 2.7082, "step": 34220 }, { "epoch": 2.124340430815072, "grad_norm": 0.546726233820077, "learning_rate": 2.3784137123709056e-05, "loss": 2.7353, "step": 34221 }, { "epoch": 2.12440250791483, "grad_norm": 0.4900233813954902, "learning_rate": 2.3781061896207647e-05, "loss": 2.6287, "step": 34222 }, { "epoch": 2.124464585014588, "grad_norm": 0.5771853827255401, "learning_rate": 2.3777986805496783e-05, "loss": 2.7245, "step": 34223 }, { "epoch": 2.124526662114346, "grad_norm": 0.5378810818004984, "learning_rate": 2.377491185159254e-05, "loss": 2.6851, "step": 34224 }, { "epoch": 2.124588739214104, "grad_norm": 0.5381486305781514, "learning_rate": 2.377183703451095e-05, "loss": 2.6784, "step": 34225 }, { "epoch": 2.1246508163138618, "grad_norm": 0.5046312306690126, "learning_rate": 2.376876235426806e-05, "loss": 2.6536, "step": 34226 }, { "epoch": 2.1247128934136197, "grad_norm": 0.5430281627176733, "learning_rate": 2.37656878108799e-05, "loss": 2.6706, "step": 34227 }, { "epoch": 2.1247749705133776, "grad_norm": 0.5649135783866489, "learning_rate": 2.3762613404362498e-05, "loss": 2.621, "step": 34228 }, { "epoch": 2.1248370476131355, "grad_norm": 0.4742982103369509, "learning_rate": 2.3759539134731912e-05, "loss": 2.7128, "step": 34229 }, { "epoch": 2.1248991247128934, "grad_norm": 0.537722539299007, "learning_rate": 2.3756465002004198e-05, "loss": 2.7058, "step": 34230 }, { "epoch": 2.1249612018126514, "grad_norm": 0.6049771928429138, "learning_rate": 2.3753391006195376e-05, "loss": 2.7692, "step": 34231 }, { "epoch": 2.1250232789124093, "grad_norm": 0.5269330155909471, "learning_rate": 2.375031714732148e-05, "loss": 2.6548, "step": 34232 }, { "epoch": 2.125085356012167, "grad_norm": 0.55713444271563, "learning_rate": 2.374724342539854e-05, "loss": 2.684, "step": 34233 }, { "epoch": 2.125147433111925, "grad_norm": 0.5030712632098344, "learning_rate": 2.3744169840442625e-05, "loss": 2.7553, "step": 34234 }, { "epoch": 2.125209510211683, "grad_norm": 0.5194896595304861, "learning_rate": 2.3741096392469746e-05, "loss": 2.7298, "step": 34235 }, { "epoch": 2.125271587311441, "grad_norm": 0.5181869956131296, "learning_rate": 2.3738023081495942e-05, "loss": 2.7163, "step": 34236 }, { "epoch": 2.125333664411199, "grad_norm": 0.5537659417606768, "learning_rate": 2.373494990753723e-05, "loss": 2.6696, "step": 34237 }, { "epoch": 2.125395741510957, "grad_norm": 0.5067570795882975, "learning_rate": 2.3731876870609676e-05, "loss": 2.6506, "step": 34238 }, { "epoch": 2.1254578186107147, "grad_norm": 0.5004133661538744, "learning_rate": 2.3728803970729296e-05, "loss": 2.65, "step": 34239 }, { "epoch": 2.1255198957104726, "grad_norm": 0.5425829168318201, "learning_rate": 2.3725731207912128e-05, "loss": 2.7368, "step": 34240 }, { "epoch": 2.12558197281023, "grad_norm": 0.4706686806640185, "learning_rate": 2.3722658582174195e-05, "loss": 2.7072, "step": 34241 }, { "epoch": 2.125644049909988, "grad_norm": 0.5329751446065488, "learning_rate": 2.371958609353151e-05, "loss": 2.6521, "step": 34242 }, { "epoch": 2.125706127009746, "grad_norm": 0.5555427566352066, "learning_rate": 2.3716513742000145e-05, "loss": 2.6841, "step": 34243 }, { "epoch": 2.125768204109504, "grad_norm": 0.5496843385849476, "learning_rate": 2.3713441527596116e-05, "loss": 2.7406, "step": 34244 }, { "epoch": 2.1258302812092618, "grad_norm": 0.5769089978285247, "learning_rate": 2.3710369450335435e-05, "loss": 2.7248, "step": 34245 }, { "epoch": 2.1258923583090197, "grad_norm": 0.5556600667746137, "learning_rate": 2.3707297510234122e-05, "loss": 2.6468, "step": 34246 }, { "epoch": 2.1259544354087776, "grad_norm": 0.49680331690751384, "learning_rate": 2.3704225707308243e-05, "loss": 2.6718, "step": 34247 }, { "epoch": 2.1260165125085355, "grad_norm": 0.5266650883836075, "learning_rate": 2.3701154041573797e-05, "loss": 2.6773, "step": 34248 }, { "epoch": 2.1260785896082934, "grad_norm": 0.4778672477239049, "learning_rate": 2.3698082513046816e-05, "loss": 2.6145, "step": 34249 }, { "epoch": 2.1261406667080514, "grad_norm": 0.4995899157876572, "learning_rate": 2.369501112174332e-05, "loss": 2.7817, "step": 34250 }, { "epoch": 2.1262027438078093, "grad_norm": 0.508545214890548, "learning_rate": 2.3691939867679324e-05, "loss": 2.7903, "step": 34251 }, { "epoch": 2.126264820907567, "grad_norm": 0.5471949050376119, "learning_rate": 2.368886875087088e-05, "loss": 2.7953, "step": 34252 }, { "epoch": 2.126326898007325, "grad_norm": 0.496186015042179, "learning_rate": 2.3685797771334e-05, "loss": 2.7198, "step": 34253 }, { "epoch": 2.126388975107083, "grad_norm": 0.48647641598286473, "learning_rate": 2.3682726929084697e-05, "loss": 2.6781, "step": 34254 }, { "epoch": 2.126451052206841, "grad_norm": 0.5337813599104372, "learning_rate": 2.3679656224138974e-05, "loss": 2.6313, "step": 34255 }, { "epoch": 2.126513129306599, "grad_norm": 0.504236748713694, "learning_rate": 2.3676585656512895e-05, "loss": 2.685, "step": 34256 }, { "epoch": 2.1265752064063568, "grad_norm": 0.536985875583281, "learning_rate": 2.3673515226222463e-05, "loss": 2.7271, "step": 34257 }, { "epoch": 2.1266372835061147, "grad_norm": 0.46699932927342835, "learning_rate": 2.3670444933283693e-05, "loss": 2.6514, "step": 34258 }, { "epoch": 2.1266993606058726, "grad_norm": 0.4876128695723937, "learning_rate": 2.3667374777712583e-05, "loss": 2.6494, "step": 34259 }, { "epoch": 2.1267614377056305, "grad_norm": 0.48811656463504666, "learning_rate": 2.3664304759525185e-05, "loss": 2.6441, "step": 34260 }, { "epoch": 2.1268235148053884, "grad_norm": 0.5081226282193431, "learning_rate": 2.3661234878737488e-05, "loss": 2.8296, "step": 34261 }, { "epoch": 2.1268855919051464, "grad_norm": 0.5003991619873805, "learning_rate": 2.365816513536554e-05, "loss": 2.6916, "step": 34262 }, { "epoch": 2.1269476690049043, "grad_norm": 0.475669540169481, "learning_rate": 2.3655095529425343e-05, "loss": 2.753, "step": 34263 }, { "epoch": 2.1270097461046618, "grad_norm": 0.4842653919613643, "learning_rate": 2.3652026060932902e-05, "loss": 2.7793, "step": 34264 }, { "epoch": 2.12707182320442, "grad_norm": 0.4754536814939144, "learning_rate": 2.364895672990422e-05, "loss": 2.7268, "step": 34265 }, { "epoch": 2.1271339003041776, "grad_norm": 0.49878241882990315, "learning_rate": 2.3645887536355348e-05, "loss": 2.7769, "step": 34266 }, { "epoch": 2.1271959774039355, "grad_norm": 0.43562812218443425, "learning_rate": 2.364281848030228e-05, "loss": 2.6835, "step": 34267 }, { "epoch": 2.1272580545036934, "grad_norm": 0.5403437660221254, "learning_rate": 2.3639749561761027e-05, "loss": 2.7732, "step": 34268 }, { "epoch": 2.1273201316034513, "grad_norm": 0.5236572544493253, "learning_rate": 2.3636680780747576e-05, "loss": 2.7123, "step": 34269 }, { "epoch": 2.1273822087032093, "grad_norm": 0.4640777752349883, "learning_rate": 2.3633612137277982e-05, "loss": 2.6478, "step": 34270 }, { "epoch": 2.127444285802967, "grad_norm": 0.5491905780366055, "learning_rate": 2.3630543631368236e-05, "loss": 2.7032, "step": 34271 }, { "epoch": 2.127506362902725, "grad_norm": 0.507763376060713, "learning_rate": 2.3627475263034344e-05, "loss": 2.6196, "step": 34272 }, { "epoch": 2.127568440002483, "grad_norm": 0.4886067488075436, "learning_rate": 2.36244070322923e-05, "loss": 2.6259, "step": 34273 }, { "epoch": 2.127630517102241, "grad_norm": 0.5328743448470418, "learning_rate": 2.3621338939158146e-05, "loss": 2.7043, "step": 34274 }, { "epoch": 2.127692594201999, "grad_norm": 0.5016807381133974, "learning_rate": 2.361827098364786e-05, "loss": 2.626, "step": 34275 }, { "epoch": 2.1277546713017568, "grad_norm": 0.5239458604646368, "learning_rate": 2.3615203165777472e-05, "loss": 2.6266, "step": 34276 }, { "epoch": 2.1278167484015147, "grad_norm": 0.5713986805139861, "learning_rate": 2.3612135485562965e-05, "loss": 2.7152, "step": 34277 }, { "epoch": 2.1278788255012726, "grad_norm": 0.5087204665438544, "learning_rate": 2.360906794302034e-05, "loss": 2.6591, "step": 34278 }, { "epoch": 2.1279409026010305, "grad_norm": 0.591693101099891, "learning_rate": 2.360600053816563e-05, "loss": 2.7174, "step": 34279 }, { "epoch": 2.1280029797007884, "grad_norm": 0.5356937688350112, "learning_rate": 2.360293327101482e-05, "loss": 2.6259, "step": 34280 }, { "epoch": 2.1280650568005464, "grad_norm": 0.5811123346535075, "learning_rate": 2.359986614158392e-05, "loss": 2.7872, "step": 34281 }, { "epoch": 2.1281271339003043, "grad_norm": 0.5477253705646059, "learning_rate": 2.359679914988891e-05, "loss": 2.688, "step": 34282 }, { "epoch": 2.128189211000062, "grad_norm": 0.5136932252573446, "learning_rate": 2.3593732295945826e-05, "loss": 2.6529, "step": 34283 }, { "epoch": 2.12825128809982, "grad_norm": 0.5809342203360615, "learning_rate": 2.359066557977065e-05, "loss": 2.7529, "step": 34284 }, { "epoch": 2.128313365199578, "grad_norm": 0.5029908374535821, "learning_rate": 2.3587599001379378e-05, "loss": 2.6157, "step": 34285 }, { "epoch": 2.128375442299336, "grad_norm": 0.5208314100865776, "learning_rate": 2.358453256078802e-05, "loss": 2.6549, "step": 34286 }, { "epoch": 2.128437519399094, "grad_norm": 0.5693877836146943, "learning_rate": 2.3581466258012546e-05, "loss": 2.7278, "step": 34287 }, { "epoch": 2.128499596498852, "grad_norm": 0.4785352274713782, "learning_rate": 2.3578400093068996e-05, "loss": 2.6408, "step": 34288 }, { "epoch": 2.1285616735986093, "grad_norm": 0.5124239248707885, "learning_rate": 2.3575334065973347e-05, "loss": 2.7157, "step": 34289 }, { "epoch": 2.128623750698367, "grad_norm": 0.5100427802225668, "learning_rate": 2.3572268176741597e-05, "loss": 2.5435, "step": 34290 }, { "epoch": 2.128685827798125, "grad_norm": 0.5077502872798676, "learning_rate": 2.3569202425389714e-05, "loss": 2.7331, "step": 34291 }, { "epoch": 2.128747904897883, "grad_norm": 0.48119649589662716, "learning_rate": 2.356613681193374e-05, "loss": 2.6674, "step": 34292 }, { "epoch": 2.128809981997641, "grad_norm": 0.5036935283348896, "learning_rate": 2.356307133638963e-05, "loss": 2.6703, "step": 34293 }, { "epoch": 2.128872059097399, "grad_norm": 0.5029763505639289, "learning_rate": 2.3560005998773416e-05, "loss": 2.7629, "step": 34294 }, { "epoch": 2.1289341361971568, "grad_norm": 0.47856625658872254, "learning_rate": 2.3556940799101068e-05, "loss": 2.7749, "step": 34295 }, { "epoch": 2.1289962132969147, "grad_norm": 0.5254769803802648, "learning_rate": 2.3553875737388557e-05, "loss": 2.7351, "step": 34296 }, { "epoch": 2.1290582903966726, "grad_norm": 0.47921270861080945, "learning_rate": 2.355081081365192e-05, "loss": 2.6806, "step": 34297 }, { "epoch": 2.1291203674964305, "grad_norm": 0.4653373950116742, "learning_rate": 2.354774602790712e-05, "loss": 2.6187, "step": 34298 }, { "epoch": 2.1291824445961884, "grad_norm": 0.4793988232120335, "learning_rate": 2.354468138017015e-05, "loss": 2.6002, "step": 34299 }, { "epoch": 2.1292445216959464, "grad_norm": 0.44607814420804054, "learning_rate": 2.3541616870457002e-05, "loss": 2.7165, "step": 34300 }, { "epoch": 2.1293065987957043, "grad_norm": 0.5544347858983901, "learning_rate": 2.353855249878364e-05, "loss": 2.6912, "step": 34301 }, { "epoch": 2.129368675895462, "grad_norm": 0.4970114772428921, "learning_rate": 2.3535488265166096e-05, "loss": 2.6911, "step": 34302 }, { "epoch": 2.12943075299522, "grad_norm": 0.5083544846600397, "learning_rate": 2.353242416962033e-05, "loss": 2.6892, "step": 34303 }, { "epoch": 2.129492830094978, "grad_norm": 0.4933180464758146, "learning_rate": 2.3529360212162332e-05, "loss": 2.6519, "step": 34304 }, { "epoch": 2.129554907194736, "grad_norm": 0.5476083198502585, "learning_rate": 2.352629639280807e-05, "loss": 2.7864, "step": 34305 }, { "epoch": 2.129616984294494, "grad_norm": 0.4867939996489041, "learning_rate": 2.3523232711573568e-05, "loss": 2.6649, "step": 34306 }, { "epoch": 2.129679061394252, "grad_norm": 0.5376843790480101, "learning_rate": 2.3520169168474786e-05, "loss": 2.6851, "step": 34307 }, { "epoch": 2.1297411384940097, "grad_norm": 0.5882862318823463, "learning_rate": 2.3517105763527712e-05, "loss": 2.6467, "step": 34308 }, { "epoch": 2.1298032155937676, "grad_norm": 0.5079849243874492, "learning_rate": 2.3514042496748307e-05, "loss": 2.664, "step": 34309 }, { "epoch": 2.1298652926935255, "grad_norm": 0.540869765467712, "learning_rate": 2.3510979368152593e-05, "loss": 2.7737, "step": 34310 }, { "epoch": 2.1299273697932835, "grad_norm": 0.4717381285167809, "learning_rate": 2.350791637775652e-05, "loss": 2.6475, "step": 34311 }, { "epoch": 2.129989446893041, "grad_norm": 0.4964256828185839, "learning_rate": 2.3504853525576087e-05, "loss": 2.5669, "step": 34312 }, { "epoch": 2.1300515239927993, "grad_norm": 0.48898580455555823, "learning_rate": 2.3501790811627267e-05, "loss": 2.7436, "step": 34313 }, { "epoch": 2.1301136010925568, "grad_norm": 0.5146270400056209, "learning_rate": 2.3498728235926014e-05, "loss": 2.5891, "step": 34314 }, { "epoch": 2.1301756781923147, "grad_norm": 0.5221372938875491, "learning_rate": 2.3495665798488348e-05, "loss": 2.665, "step": 34315 }, { "epoch": 2.1302377552920726, "grad_norm": 0.5370418436771933, "learning_rate": 2.349260349933023e-05, "loss": 2.7343, "step": 34316 }, { "epoch": 2.1302998323918305, "grad_norm": 0.4767313329835839, "learning_rate": 2.348954133846763e-05, "loss": 2.7456, "step": 34317 }, { "epoch": 2.1303619094915884, "grad_norm": 0.5492263923645082, "learning_rate": 2.3486479315916515e-05, "loss": 2.7407, "step": 34318 }, { "epoch": 2.1304239865913464, "grad_norm": 0.5408942020922906, "learning_rate": 2.348341743169289e-05, "loss": 2.6249, "step": 34319 }, { "epoch": 2.1304860636911043, "grad_norm": 0.5593624827513519, "learning_rate": 2.348035568581271e-05, "loss": 2.7333, "step": 34320 }, { "epoch": 2.130548140790862, "grad_norm": 0.5280911835447215, "learning_rate": 2.3477294078291957e-05, "loss": 2.6616, "step": 34321 }, { "epoch": 2.13061021789062, "grad_norm": 0.549361461910918, "learning_rate": 2.347423260914658e-05, "loss": 2.7666, "step": 34322 }, { "epoch": 2.130672294990378, "grad_norm": 0.5024172051387265, "learning_rate": 2.347117127839259e-05, "loss": 2.6994, "step": 34323 }, { "epoch": 2.130734372090136, "grad_norm": 0.5520095116065331, "learning_rate": 2.3468110086045937e-05, "loss": 2.6736, "step": 34324 }, { "epoch": 2.130796449189894, "grad_norm": 0.5260126622075638, "learning_rate": 2.3465049032122593e-05, "loss": 2.5915, "step": 34325 }, { "epoch": 2.1308585262896518, "grad_norm": 0.5363120524643846, "learning_rate": 2.3461988116638518e-05, "loss": 2.6836, "step": 34326 }, { "epoch": 2.1309206033894097, "grad_norm": 0.4854268067519177, "learning_rate": 2.3458927339609705e-05, "loss": 2.7639, "step": 34327 }, { "epoch": 2.1309826804891676, "grad_norm": 0.5381073436875393, "learning_rate": 2.3455866701052098e-05, "loss": 2.731, "step": 34328 }, { "epoch": 2.1310447575889255, "grad_norm": 0.49616216953679737, "learning_rate": 2.345280620098169e-05, "loss": 2.7059, "step": 34329 }, { "epoch": 2.1311068346886834, "grad_norm": 0.5020533658287423, "learning_rate": 2.3449745839414445e-05, "loss": 2.6971, "step": 34330 }, { "epoch": 2.1311689117884414, "grad_norm": 0.48665580281233595, "learning_rate": 2.3446685616366315e-05, "loss": 2.6458, "step": 34331 }, { "epoch": 2.1312309888881993, "grad_norm": 0.4711618048506242, "learning_rate": 2.3443625531853257e-05, "loss": 2.6491, "step": 34332 }, { "epoch": 2.131293065987957, "grad_norm": 0.4556095638459871, "learning_rate": 2.3440565585891268e-05, "loss": 2.6118, "step": 34333 }, { "epoch": 2.131355143087715, "grad_norm": 0.5437485563324804, "learning_rate": 2.34375057784963e-05, "loss": 2.7789, "step": 34334 }, { "epoch": 2.131417220187473, "grad_norm": 0.5054216884773433, "learning_rate": 2.3434446109684305e-05, "loss": 2.6782, "step": 34335 }, { "epoch": 2.131479297287231, "grad_norm": 0.5032662163863343, "learning_rate": 2.343138657947126e-05, "loss": 2.6956, "step": 34336 }, { "epoch": 2.1315413743869884, "grad_norm": 0.5154217984317451, "learning_rate": 2.34283271878731e-05, "loss": 2.6735, "step": 34337 }, { "epoch": 2.1316034514867463, "grad_norm": 0.49820494293890966, "learning_rate": 2.3425267934905826e-05, "loss": 2.6924, "step": 34338 }, { "epoch": 2.1316655285865043, "grad_norm": 0.5751111628014687, "learning_rate": 2.342220882058538e-05, "loss": 2.7378, "step": 34339 }, { "epoch": 2.131727605686262, "grad_norm": 0.49557409328121227, "learning_rate": 2.3419149844927723e-05, "loss": 2.6668, "step": 34340 }, { "epoch": 2.13178968278602, "grad_norm": 0.5263378014792608, "learning_rate": 2.3416091007948797e-05, "loss": 2.7152, "step": 34341 }, { "epoch": 2.131851759885778, "grad_norm": 0.5239661379840218, "learning_rate": 2.3413032309664595e-05, "loss": 2.7659, "step": 34342 }, { "epoch": 2.131913836985536, "grad_norm": 0.5271340695062957, "learning_rate": 2.3409973750091057e-05, "loss": 2.7607, "step": 34343 }, { "epoch": 2.131975914085294, "grad_norm": 0.4776595886033929, "learning_rate": 2.3406915329244145e-05, "loss": 2.6803, "step": 34344 }, { "epoch": 2.1320379911850518, "grad_norm": 0.5396149886234043, "learning_rate": 2.340385704713979e-05, "loss": 2.7244, "step": 34345 }, { "epoch": 2.1321000682848097, "grad_norm": 0.49357220973743043, "learning_rate": 2.3400798903793986e-05, "loss": 2.6282, "step": 34346 }, { "epoch": 2.1321621453845676, "grad_norm": 0.5334292760124547, "learning_rate": 2.339774089922267e-05, "loss": 2.7618, "step": 34347 }, { "epoch": 2.1322242224843255, "grad_norm": 0.5140070238876607, "learning_rate": 2.33946830334418e-05, "loss": 2.6382, "step": 34348 }, { "epoch": 2.1322862995840834, "grad_norm": 0.5287250371771447, "learning_rate": 2.3391625306467328e-05, "loss": 2.7734, "step": 34349 }, { "epoch": 2.1323483766838414, "grad_norm": 0.5169675860804339, "learning_rate": 2.3388567718315186e-05, "loss": 2.6898, "step": 34350 }, { "epoch": 2.1324104537835993, "grad_norm": 0.5340111726456441, "learning_rate": 2.3385510269001366e-05, "loss": 2.7532, "step": 34351 }, { "epoch": 2.132472530883357, "grad_norm": 0.5151641404505365, "learning_rate": 2.33824529585418e-05, "loss": 2.7065, "step": 34352 }, { "epoch": 2.132534607983115, "grad_norm": 0.48796987141954573, "learning_rate": 2.337939578695244e-05, "loss": 2.6737, "step": 34353 }, { "epoch": 2.132596685082873, "grad_norm": 0.5473542157391422, "learning_rate": 2.3376338754249215e-05, "loss": 2.722, "step": 34354 }, { "epoch": 2.132658762182631, "grad_norm": 0.5460093049559506, "learning_rate": 2.337328186044811e-05, "loss": 2.6477, "step": 34355 }, { "epoch": 2.132720839282389, "grad_norm": 0.5227888272723867, "learning_rate": 2.337022510556506e-05, "loss": 2.7083, "step": 34356 }, { "epoch": 2.132782916382147, "grad_norm": 0.5190061913251468, "learning_rate": 2.336716848961601e-05, "loss": 2.6768, "step": 34357 }, { "epoch": 2.1328449934819047, "grad_norm": 0.49516809265766404, "learning_rate": 2.336411201261689e-05, "loss": 2.7233, "step": 34358 }, { "epoch": 2.1329070705816626, "grad_norm": 0.517127333768798, "learning_rate": 2.336105567458366e-05, "loss": 2.7005, "step": 34359 }, { "epoch": 2.13296914768142, "grad_norm": 0.5430674566112308, "learning_rate": 2.335799947553229e-05, "loss": 2.7272, "step": 34360 }, { "epoch": 2.1330312247811785, "grad_norm": 0.46110290856930086, "learning_rate": 2.3354943415478703e-05, "loss": 2.6411, "step": 34361 }, { "epoch": 2.133093301880936, "grad_norm": 0.5148500919675009, "learning_rate": 2.3351887494438844e-05, "loss": 2.6068, "step": 34362 }, { "epoch": 2.133155378980694, "grad_norm": 0.47702305848891235, "learning_rate": 2.334883171242866e-05, "loss": 2.778, "step": 34363 }, { "epoch": 2.1332174560804518, "grad_norm": 0.49625385133992794, "learning_rate": 2.334577606946407e-05, "loss": 2.6388, "step": 34364 }, { "epoch": 2.1332795331802097, "grad_norm": 0.5181637312122009, "learning_rate": 2.334272056556106e-05, "loss": 2.7169, "step": 34365 }, { "epoch": 2.1333416102799676, "grad_norm": 0.4711757835209626, "learning_rate": 2.3339665200735543e-05, "loss": 2.5933, "step": 34366 }, { "epoch": 2.1334036873797255, "grad_norm": 0.5080446739565588, "learning_rate": 2.3336609975003465e-05, "loss": 2.6759, "step": 34367 }, { "epoch": 2.1334657644794834, "grad_norm": 0.5244255135704937, "learning_rate": 2.333355488838075e-05, "loss": 2.7388, "step": 34368 }, { "epoch": 2.1335278415792414, "grad_norm": 0.5219007679865095, "learning_rate": 2.3330499940883365e-05, "loss": 2.7088, "step": 34369 }, { "epoch": 2.1335899186789993, "grad_norm": 0.4727037579329159, "learning_rate": 2.3327445132527242e-05, "loss": 2.7403, "step": 34370 }, { "epoch": 2.133651995778757, "grad_norm": 0.5211811112546263, "learning_rate": 2.3324390463328312e-05, "loss": 2.6, "step": 34371 }, { "epoch": 2.133714072878515, "grad_norm": 0.5119121564577109, "learning_rate": 2.332133593330249e-05, "loss": 2.6761, "step": 34372 }, { "epoch": 2.133776149978273, "grad_norm": 0.4963662908246824, "learning_rate": 2.331828154246576e-05, "loss": 2.6616, "step": 34373 }, { "epoch": 2.133838227078031, "grad_norm": 0.5393444926982142, "learning_rate": 2.3315227290834028e-05, "loss": 2.7025, "step": 34374 }, { "epoch": 2.133900304177789, "grad_norm": 0.557230880725921, "learning_rate": 2.3312173178423234e-05, "loss": 2.7772, "step": 34375 }, { "epoch": 2.133962381277547, "grad_norm": 0.523212128132129, "learning_rate": 2.3309119205249313e-05, "loss": 2.7344, "step": 34376 }, { "epoch": 2.1340244583773047, "grad_norm": 0.5191278540746141, "learning_rate": 2.3306065371328173e-05, "loss": 2.6529, "step": 34377 }, { "epoch": 2.1340865354770626, "grad_norm": 0.4881489841610811, "learning_rate": 2.3303011676675794e-05, "loss": 2.6888, "step": 34378 }, { "epoch": 2.1341486125768205, "grad_norm": 0.505564334799561, "learning_rate": 2.329995812130808e-05, "loss": 2.7299, "step": 34379 }, { "epoch": 2.1342106896765785, "grad_norm": 0.5326115606868964, "learning_rate": 2.3296904705240968e-05, "loss": 2.8004, "step": 34380 }, { "epoch": 2.1342727667763364, "grad_norm": 0.5216075166048046, "learning_rate": 2.329385142849037e-05, "loss": 2.6403, "step": 34381 }, { "epoch": 2.1343348438760943, "grad_norm": 0.4980150399598843, "learning_rate": 2.3290798291072246e-05, "loss": 2.6602, "step": 34382 }, { "epoch": 2.134396920975852, "grad_norm": 0.4946297347126808, "learning_rate": 2.3287745293002517e-05, "loss": 2.6741, "step": 34383 }, { "epoch": 2.13445899807561, "grad_norm": 0.5150365361429109, "learning_rate": 2.32846924342971e-05, "loss": 2.6127, "step": 34384 }, { "epoch": 2.1345210751753676, "grad_norm": 0.4798928818288661, "learning_rate": 2.328163971497193e-05, "loss": 2.6423, "step": 34385 }, { "epoch": 2.1345831522751255, "grad_norm": 0.5274399563664025, "learning_rate": 2.327858713504292e-05, "loss": 2.6576, "step": 34386 }, { "epoch": 2.1346452293748834, "grad_norm": 0.5069503214102092, "learning_rate": 2.3275534694526018e-05, "loss": 2.6894, "step": 34387 }, { "epoch": 2.1347073064746414, "grad_norm": 0.4736703287584177, "learning_rate": 2.327248239343714e-05, "loss": 2.7817, "step": 34388 }, { "epoch": 2.1347693835743993, "grad_norm": 0.5282276234434656, "learning_rate": 2.326943023179221e-05, "loss": 2.599, "step": 34389 }, { "epoch": 2.134831460674157, "grad_norm": 0.5112174104409151, "learning_rate": 2.3266378209607137e-05, "loss": 2.6445, "step": 34390 }, { "epoch": 2.134893537773915, "grad_norm": 0.5206650430270989, "learning_rate": 2.326332632689787e-05, "loss": 2.7719, "step": 34391 }, { "epoch": 2.134955614873673, "grad_norm": 0.47989317681342036, "learning_rate": 2.3260274583680308e-05, "loss": 2.6354, "step": 34392 }, { "epoch": 2.135017691973431, "grad_norm": 0.5544864666338983, "learning_rate": 2.32572229799704e-05, "loss": 2.6973, "step": 34393 }, { "epoch": 2.135079769073189, "grad_norm": 0.5060160037250289, "learning_rate": 2.3254171515784053e-05, "loss": 2.6794, "step": 34394 }, { "epoch": 2.1351418461729468, "grad_norm": 0.5205407051729563, "learning_rate": 2.3251120191137165e-05, "loss": 2.7934, "step": 34395 }, { "epoch": 2.1352039232727047, "grad_norm": 0.47464012705698344, "learning_rate": 2.3248069006045698e-05, "loss": 2.6546, "step": 34396 }, { "epoch": 2.1352660003724626, "grad_norm": 0.5455635148894525, "learning_rate": 2.3245017960525552e-05, "loss": 2.6054, "step": 34397 }, { "epoch": 2.1353280774722205, "grad_norm": 0.5272236440012101, "learning_rate": 2.3241967054592634e-05, "loss": 2.6598, "step": 34398 }, { "epoch": 2.1353901545719784, "grad_norm": 0.4919058129912279, "learning_rate": 2.3238916288262875e-05, "loss": 2.6133, "step": 34399 }, { "epoch": 2.1354522316717364, "grad_norm": 0.5831202361882313, "learning_rate": 2.323586566155217e-05, "loss": 2.6619, "step": 34400 }, { "epoch": 2.1355143087714943, "grad_norm": 0.5214889183275138, "learning_rate": 2.3232815174476464e-05, "loss": 2.8224, "step": 34401 }, { "epoch": 2.135576385871252, "grad_norm": 0.5701375719159252, "learning_rate": 2.3229764827051664e-05, "loss": 2.7796, "step": 34402 }, { "epoch": 2.13563846297101, "grad_norm": 0.5225808549679362, "learning_rate": 2.3226714619293676e-05, "loss": 2.6642, "step": 34403 }, { "epoch": 2.135700540070768, "grad_norm": 0.5243592757548877, "learning_rate": 2.32236645512184e-05, "loss": 2.7737, "step": 34404 }, { "epoch": 2.135762617170526, "grad_norm": 0.5457896497829786, "learning_rate": 2.3220614622841786e-05, "loss": 2.675, "step": 34405 }, { "epoch": 2.135824694270284, "grad_norm": 0.4703239876747744, "learning_rate": 2.321756483417972e-05, "loss": 2.6494, "step": 34406 }, { "epoch": 2.135886771370042, "grad_norm": 0.5636043568949021, "learning_rate": 2.3214515185248132e-05, "loss": 2.6698, "step": 34407 }, { "epoch": 2.1359488484697993, "grad_norm": 0.5183053144423536, "learning_rate": 2.3211465676062895e-05, "loss": 2.6986, "step": 34408 }, { "epoch": 2.136010925569557, "grad_norm": 0.4977893836909762, "learning_rate": 2.3208416306639964e-05, "loss": 2.6773, "step": 34409 }, { "epoch": 2.136073002669315, "grad_norm": 0.5123262366686712, "learning_rate": 2.3205367076995227e-05, "loss": 2.6462, "step": 34410 }, { "epoch": 2.136135079769073, "grad_norm": 0.4703815042054653, "learning_rate": 2.3202317987144593e-05, "loss": 2.7105, "step": 34411 }, { "epoch": 2.136197156868831, "grad_norm": 0.4846338995312437, "learning_rate": 2.3199269037103975e-05, "loss": 2.6745, "step": 34412 }, { "epoch": 2.136259233968589, "grad_norm": 0.5163057919204908, "learning_rate": 2.319622022688926e-05, "loss": 2.7247, "step": 34413 }, { "epoch": 2.1363213110683468, "grad_norm": 0.4967971080930055, "learning_rate": 2.3193171556516385e-05, "loss": 2.781, "step": 34414 }, { "epoch": 2.1363833881681047, "grad_norm": 0.5336465354267987, "learning_rate": 2.3190123026001248e-05, "loss": 2.6611, "step": 34415 }, { "epoch": 2.1364454652678626, "grad_norm": 0.47011504099490165, "learning_rate": 2.318707463535974e-05, "loss": 2.7405, "step": 34416 }, { "epoch": 2.1365075423676205, "grad_norm": 0.54276966314661, "learning_rate": 2.318402638460776e-05, "loss": 2.6265, "step": 34417 }, { "epoch": 2.1365696194673784, "grad_norm": 0.4733683998367009, "learning_rate": 2.318097827376124e-05, "loss": 2.7599, "step": 34418 }, { "epoch": 2.1366316965671364, "grad_norm": 0.5135686502337932, "learning_rate": 2.3177930302836064e-05, "loss": 2.6718, "step": 34419 }, { "epoch": 2.1366937736668943, "grad_norm": 0.49277618564681575, "learning_rate": 2.317488247184814e-05, "loss": 2.7426, "step": 34420 }, { "epoch": 2.136755850766652, "grad_norm": 0.5333980441572149, "learning_rate": 2.317183478081337e-05, "loss": 2.7779, "step": 34421 }, { "epoch": 2.13681792786641, "grad_norm": 0.5125327588288193, "learning_rate": 2.3168787229747623e-05, "loss": 2.7005, "step": 34422 }, { "epoch": 2.136880004966168, "grad_norm": 0.5601940041219479, "learning_rate": 2.3165739818666855e-05, "loss": 2.7273, "step": 34423 }, { "epoch": 2.136942082065926, "grad_norm": 0.5026610017835885, "learning_rate": 2.316269254758691e-05, "loss": 2.6614, "step": 34424 }, { "epoch": 2.137004159165684, "grad_norm": 0.5617699123634142, "learning_rate": 2.3159645416523733e-05, "loss": 2.741, "step": 34425 }, { "epoch": 2.137066236265442, "grad_norm": 0.5410273030807531, "learning_rate": 2.3156598425493204e-05, "loss": 2.6122, "step": 34426 }, { "epoch": 2.1371283133651997, "grad_norm": 0.5348594287888891, "learning_rate": 2.31535515745112e-05, "loss": 2.6645, "step": 34427 }, { "epoch": 2.1371903904649576, "grad_norm": 0.5204395585819384, "learning_rate": 2.3150504863593653e-05, "loss": 2.6616, "step": 34428 }, { "epoch": 2.1372524675647155, "grad_norm": 0.586287508376702, "learning_rate": 2.3147458292756437e-05, "loss": 2.712, "step": 34429 }, { "epoch": 2.1373145446644735, "grad_norm": 0.528414000491158, "learning_rate": 2.3144411862015454e-05, "loss": 2.7246, "step": 34430 }, { "epoch": 2.1373766217642314, "grad_norm": 0.6206629956803033, "learning_rate": 2.314136557138657e-05, "loss": 2.6865, "step": 34431 }, { "epoch": 2.1374386988639893, "grad_norm": 0.5703279937311738, "learning_rate": 2.3138319420885722e-05, "loss": 2.6817, "step": 34432 }, { "epoch": 2.1375007759637468, "grad_norm": 0.5176362490856068, "learning_rate": 2.3135273410528783e-05, "loss": 2.6679, "step": 34433 }, { "epoch": 2.1375628530635047, "grad_norm": 0.5769515113059007, "learning_rate": 2.3132227540331647e-05, "loss": 2.6812, "step": 34434 }, { "epoch": 2.1376249301632626, "grad_norm": 0.5689255034916737, "learning_rate": 2.3129181810310196e-05, "loss": 2.6134, "step": 34435 }, { "epoch": 2.1376870072630205, "grad_norm": 0.5735816275516618, "learning_rate": 2.3126136220480315e-05, "loss": 2.7525, "step": 34436 }, { "epoch": 2.1377490843627784, "grad_norm": 0.4916678061579846, "learning_rate": 2.312309077085792e-05, "loss": 2.7038, "step": 34437 }, { "epoch": 2.1378111614625364, "grad_norm": 0.5706817528954687, "learning_rate": 2.3120045461458883e-05, "loss": 2.6983, "step": 34438 }, { "epoch": 2.1378732385622943, "grad_norm": 0.523109788935566, "learning_rate": 2.3117000292299102e-05, "loss": 2.6455, "step": 34439 }, { "epoch": 2.137935315662052, "grad_norm": 0.5202174709934744, "learning_rate": 2.3113955263394432e-05, "loss": 2.7458, "step": 34440 }, { "epoch": 2.13799739276181, "grad_norm": 0.49467772292827233, "learning_rate": 2.3110910374760803e-05, "loss": 2.6688, "step": 34441 }, { "epoch": 2.138059469861568, "grad_norm": 0.5344254453171912, "learning_rate": 2.3107865626414082e-05, "loss": 2.663, "step": 34442 }, { "epoch": 2.138121546961326, "grad_norm": 0.49121984090756116, "learning_rate": 2.3104821018370153e-05, "loss": 2.6334, "step": 34443 }, { "epoch": 2.138183624061084, "grad_norm": 0.5382743547105902, "learning_rate": 2.3101776550644888e-05, "loss": 2.6757, "step": 34444 }, { "epoch": 2.138245701160842, "grad_norm": 0.48075131116996733, "learning_rate": 2.30987322232542e-05, "loss": 2.6408, "step": 34445 }, { "epoch": 2.1383077782605997, "grad_norm": 0.5132648320235917, "learning_rate": 2.309568803621396e-05, "loss": 2.586, "step": 34446 }, { "epoch": 2.1383698553603576, "grad_norm": 0.4958474907202937, "learning_rate": 2.309264398954004e-05, "loss": 2.6579, "step": 34447 }, { "epoch": 2.1384319324601155, "grad_norm": 0.49909191818813553, "learning_rate": 2.3089600083248336e-05, "loss": 2.7182, "step": 34448 }, { "epoch": 2.1384940095598735, "grad_norm": 0.5373387349863348, "learning_rate": 2.3086556317354697e-05, "loss": 2.7677, "step": 34449 }, { "epoch": 2.1385560866596314, "grad_norm": 0.5276320058162759, "learning_rate": 2.3083512691875047e-05, "loss": 2.7059, "step": 34450 }, { "epoch": 2.1386181637593893, "grad_norm": 0.522005485166034, "learning_rate": 2.3080469206825245e-05, "loss": 2.6693, "step": 34451 }, { "epoch": 2.138680240859147, "grad_norm": 0.4902868812680647, "learning_rate": 2.3077425862221175e-05, "loss": 2.6043, "step": 34452 }, { "epoch": 2.138742317958905, "grad_norm": 0.5059793101819019, "learning_rate": 2.3074382658078685e-05, "loss": 2.6376, "step": 34453 }, { "epoch": 2.138804395058663, "grad_norm": 0.5252769659256754, "learning_rate": 2.30713395944137e-05, "loss": 2.7237, "step": 34454 }, { "epoch": 2.138866472158421, "grad_norm": 0.5203844965705755, "learning_rate": 2.3068296671242067e-05, "loss": 2.7293, "step": 34455 }, { "epoch": 2.1389285492581784, "grad_norm": 0.5098054950658902, "learning_rate": 2.3065253888579675e-05, "loss": 2.6815, "step": 34456 }, { "epoch": 2.1389906263579364, "grad_norm": 0.5277057394281977, "learning_rate": 2.3062211246442374e-05, "loss": 2.7071, "step": 34457 }, { "epoch": 2.1390527034576943, "grad_norm": 0.5026603199583849, "learning_rate": 2.305916874484605e-05, "loss": 2.7275, "step": 34458 }, { "epoch": 2.139114780557452, "grad_norm": 0.5525700355173881, "learning_rate": 2.3056126383806602e-05, "loss": 2.6599, "step": 34459 }, { "epoch": 2.13917685765721, "grad_norm": 0.49559464466320424, "learning_rate": 2.305308416333989e-05, "loss": 2.661, "step": 34460 }, { "epoch": 2.139238934756968, "grad_norm": 0.5131641678259382, "learning_rate": 2.3050042083461774e-05, "loss": 2.7182, "step": 34461 }, { "epoch": 2.139301011856726, "grad_norm": 0.5512973180344788, "learning_rate": 2.3047000144188136e-05, "loss": 2.6506, "step": 34462 }, { "epoch": 2.139363088956484, "grad_norm": 0.47229808446559113, "learning_rate": 2.3043958345534816e-05, "loss": 2.6624, "step": 34463 }, { "epoch": 2.139425166056242, "grad_norm": 0.5451846019209102, "learning_rate": 2.3040916687517733e-05, "loss": 2.6351, "step": 34464 }, { "epoch": 2.1394872431559997, "grad_norm": 0.5701900768830195, "learning_rate": 2.3037875170152727e-05, "loss": 2.7484, "step": 34465 }, { "epoch": 2.1395493202557576, "grad_norm": 0.5226470330627363, "learning_rate": 2.303483379345568e-05, "loss": 2.7489, "step": 34466 }, { "epoch": 2.1396113973555155, "grad_norm": 0.560253930719384, "learning_rate": 2.303179255744243e-05, "loss": 2.8416, "step": 34467 }, { "epoch": 2.1396734744552734, "grad_norm": 0.5392088207744477, "learning_rate": 2.302875146212888e-05, "loss": 2.6894, "step": 34468 }, { "epoch": 2.1397355515550314, "grad_norm": 0.5497239698984411, "learning_rate": 2.302571050753088e-05, "loss": 2.7152, "step": 34469 }, { "epoch": 2.1397976286547893, "grad_norm": 0.5620572679026499, "learning_rate": 2.3022669693664294e-05, "loss": 2.623, "step": 34470 }, { "epoch": 2.139859705754547, "grad_norm": 0.5640032338208631, "learning_rate": 2.3019629020544996e-05, "loss": 2.7074, "step": 34471 }, { "epoch": 2.139921782854305, "grad_norm": 0.5696147674317856, "learning_rate": 2.301658848818882e-05, "loss": 2.6351, "step": 34472 }, { "epoch": 2.139983859954063, "grad_norm": 0.5554580708911038, "learning_rate": 2.3013548096611677e-05, "loss": 2.687, "step": 34473 }, { "epoch": 2.140045937053821, "grad_norm": 0.5768517314065887, "learning_rate": 2.30105078458294e-05, "loss": 2.6652, "step": 34474 }, { "epoch": 2.140108014153579, "grad_norm": 0.5075381491561912, "learning_rate": 2.3007467735857856e-05, "loss": 2.7331, "step": 34475 }, { "epoch": 2.140170091253337, "grad_norm": 0.5958975052677791, "learning_rate": 2.300442776671289e-05, "loss": 2.7293, "step": 34476 }, { "epoch": 2.1402321683530947, "grad_norm": 0.5347363850615673, "learning_rate": 2.300138793841039e-05, "loss": 2.6899, "step": 34477 }, { "epoch": 2.1402942454528526, "grad_norm": 0.6051000887859849, "learning_rate": 2.2998348250966207e-05, "loss": 2.7907, "step": 34478 }, { "epoch": 2.1403563225526105, "grad_norm": 0.5598058021964999, "learning_rate": 2.2995308704396196e-05, "loss": 2.7331, "step": 34479 }, { "epoch": 2.1404183996523685, "grad_norm": 0.5321280852038622, "learning_rate": 2.2992269298716196e-05, "loss": 2.6772, "step": 34480 }, { "epoch": 2.140480476752126, "grad_norm": 0.5305889729377392, "learning_rate": 2.2989230033942107e-05, "loss": 2.6996, "step": 34481 }, { "epoch": 2.140542553851884, "grad_norm": 0.5323966778105554, "learning_rate": 2.2986190910089756e-05, "loss": 2.7183, "step": 34482 }, { "epoch": 2.1406046309516418, "grad_norm": 0.5512665578729877, "learning_rate": 2.2983151927175007e-05, "loss": 2.7723, "step": 34483 }, { "epoch": 2.1406667080513997, "grad_norm": 0.5585360997468137, "learning_rate": 2.2980113085213713e-05, "loss": 2.6686, "step": 34484 }, { "epoch": 2.1407287851511576, "grad_norm": 0.5280159233809685, "learning_rate": 2.297707438422171e-05, "loss": 2.7162, "step": 34485 }, { "epoch": 2.1407908622509155, "grad_norm": 0.5130207591048702, "learning_rate": 2.2974035824214886e-05, "loss": 2.6105, "step": 34486 }, { "epoch": 2.1408529393506734, "grad_norm": 0.5199811321283401, "learning_rate": 2.2970997405209075e-05, "loss": 2.7395, "step": 34487 }, { "epoch": 2.1409150164504314, "grad_norm": 0.48338068149207836, "learning_rate": 2.296795912722014e-05, "loss": 2.6671, "step": 34488 }, { "epoch": 2.1409770935501893, "grad_norm": 0.4707856798390885, "learning_rate": 2.29649209902639e-05, "loss": 2.7285, "step": 34489 }, { "epoch": 2.141039170649947, "grad_norm": 0.5010529695955612, "learning_rate": 2.296188299435623e-05, "loss": 2.6485, "step": 34490 }, { "epoch": 2.141101247749705, "grad_norm": 0.4647514508077498, "learning_rate": 2.295884513951301e-05, "loss": 2.5289, "step": 34491 }, { "epoch": 2.141163324849463, "grad_norm": 0.5126602596834214, "learning_rate": 2.295580742575005e-05, "loss": 2.7106, "step": 34492 }, { "epoch": 2.141225401949221, "grad_norm": 0.5223470973832721, "learning_rate": 2.2952769853083205e-05, "loss": 2.7055, "step": 34493 }, { "epoch": 2.141287479048979, "grad_norm": 0.48014026360754375, "learning_rate": 2.2949732421528307e-05, "loss": 2.6663, "step": 34494 }, { "epoch": 2.141349556148737, "grad_norm": 0.5366792100670602, "learning_rate": 2.2946695131101247e-05, "loss": 2.6771, "step": 34495 }, { "epoch": 2.1414116332484947, "grad_norm": 0.49229702412126025, "learning_rate": 2.2943657981817845e-05, "loss": 2.6994, "step": 34496 }, { "epoch": 2.1414737103482526, "grad_norm": 0.492903672772858, "learning_rate": 2.2940620973693944e-05, "loss": 2.6185, "step": 34497 }, { "epoch": 2.1415357874480105, "grad_norm": 0.5188870559170351, "learning_rate": 2.293758410674539e-05, "loss": 2.6625, "step": 34498 }, { "epoch": 2.1415978645477685, "grad_norm": 0.5047663364307968, "learning_rate": 2.293454738098802e-05, "loss": 2.725, "step": 34499 }, { "epoch": 2.1416599416475264, "grad_norm": 0.49373021629390723, "learning_rate": 2.29315107964377e-05, "loss": 2.7077, "step": 34500 }, { "epoch": 2.1417220187472843, "grad_norm": 0.47613421722287513, "learning_rate": 2.2928474353110256e-05, "loss": 2.6527, "step": 34501 }, { "epoch": 2.141784095847042, "grad_norm": 0.5032119107260455, "learning_rate": 2.2925438051021537e-05, "loss": 2.6989, "step": 34502 }, { "epoch": 2.1418461729468, "grad_norm": 0.49498343683236556, "learning_rate": 2.292240189018736e-05, "loss": 2.6604, "step": 34503 }, { "epoch": 2.1419082500465576, "grad_norm": 0.47441836622538913, "learning_rate": 2.2919365870623605e-05, "loss": 2.7361, "step": 34504 }, { "epoch": 2.1419703271463155, "grad_norm": 0.5519861609634954, "learning_rate": 2.2916329992346093e-05, "loss": 2.7091, "step": 34505 }, { "epoch": 2.1420324042460734, "grad_norm": 0.5030239868644513, "learning_rate": 2.291329425537066e-05, "loss": 2.6493, "step": 34506 }, { "epoch": 2.1420944813458314, "grad_norm": 0.4651635296206801, "learning_rate": 2.2910258659713147e-05, "loss": 2.7427, "step": 34507 }, { "epoch": 2.1421565584455893, "grad_norm": 0.5402187717526228, "learning_rate": 2.2907223205389372e-05, "loss": 2.6526, "step": 34508 }, { "epoch": 2.142218635545347, "grad_norm": 0.5011176492518734, "learning_rate": 2.2904187892415212e-05, "loss": 2.6638, "step": 34509 }, { "epoch": 2.142280712645105, "grad_norm": 0.5429391921526304, "learning_rate": 2.290115272080648e-05, "loss": 2.7189, "step": 34510 }, { "epoch": 2.142342789744863, "grad_norm": 0.4942271643087249, "learning_rate": 2.2898117690579013e-05, "loss": 2.6865, "step": 34511 }, { "epoch": 2.142404866844621, "grad_norm": 0.513343996483213, "learning_rate": 2.2895082801748626e-05, "loss": 2.7017, "step": 34512 }, { "epoch": 2.142466943944379, "grad_norm": 0.4564119528388367, "learning_rate": 2.289204805433119e-05, "loss": 2.7111, "step": 34513 }, { "epoch": 2.142529021044137, "grad_norm": 0.5127499215243617, "learning_rate": 2.288901344834252e-05, "loss": 2.7507, "step": 34514 }, { "epoch": 2.1425910981438947, "grad_norm": 0.4858067640228255, "learning_rate": 2.2885978983798452e-05, "loss": 2.6393, "step": 34515 }, { "epoch": 2.1426531752436526, "grad_norm": 0.5366771276512766, "learning_rate": 2.288294466071479e-05, "loss": 2.6304, "step": 34516 }, { "epoch": 2.1427152523434105, "grad_norm": 0.550715990461475, "learning_rate": 2.2879910479107407e-05, "loss": 2.7253, "step": 34517 }, { "epoch": 2.1427773294431685, "grad_norm": 0.4545611773014952, "learning_rate": 2.2876876438992117e-05, "loss": 2.6472, "step": 34518 }, { "epoch": 2.1428394065429264, "grad_norm": 0.4651071779752376, "learning_rate": 2.287384254038475e-05, "loss": 2.6174, "step": 34519 }, { "epoch": 2.1429014836426843, "grad_norm": 0.4806741105028814, "learning_rate": 2.2870808783301128e-05, "loss": 2.6402, "step": 34520 }, { "epoch": 2.142963560742442, "grad_norm": 0.48073788990056615, "learning_rate": 2.2867775167757067e-05, "loss": 2.7452, "step": 34521 }, { "epoch": 2.1430256378422, "grad_norm": 0.4925653279008777, "learning_rate": 2.286474169376842e-05, "loss": 2.7572, "step": 34522 }, { "epoch": 2.143087714941958, "grad_norm": 0.526120242932276, "learning_rate": 2.286170836135099e-05, "loss": 2.7638, "step": 34523 }, { "epoch": 2.143149792041716, "grad_norm": 0.48296973745145777, "learning_rate": 2.2858675170520638e-05, "loss": 2.7247, "step": 34524 }, { "epoch": 2.143211869141474, "grad_norm": 0.512831460949754, "learning_rate": 2.2855642121293163e-05, "loss": 2.7198, "step": 34525 }, { "epoch": 2.143273946241232, "grad_norm": 0.496164256815087, "learning_rate": 2.2852609213684373e-05, "loss": 2.6189, "step": 34526 }, { "epoch": 2.1433360233409897, "grad_norm": 0.5440845414045682, "learning_rate": 2.284957644771013e-05, "loss": 2.6999, "step": 34527 }, { "epoch": 2.1433981004407476, "grad_norm": 0.5204559071266712, "learning_rate": 2.2846543823386233e-05, "loss": 2.6522, "step": 34528 }, { "epoch": 2.143460177540505, "grad_norm": 0.5689008827029626, "learning_rate": 2.284351134072852e-05, "loss": 2.7654, "step": 34529 }, { "epoch": 2.143522254640263, "grad_norm": 0.5430075542388142, "learning_rate": 2.2840478999752773e-05, "loss": 2.7969, "step": 34530 }, { "epoch": 2.143584331740021, "grad_norm": 0.511823832664189, "learning_rate": 2.283744680047486e-05, "loss": 2.7181, "step": 34531 }, { "epoch": 2.143646408839779, "grad_norm": 0.5556136981685745, "learning_rate": 2.2834414742910583e-05, "loss": 2.6489, "step": 34532 }, { "epoch": 2.143708485939537, "grad_norm": 0.48480013679161754, "learning_rate": 2.283138282707576e-05, "loss": 2.7249, "step": 34533 }, { "epoch": 2.1437705630392947, "grad_norm": 0.5522110419444293, "learning_rate": 2.2828351052986203e-05, "loss": 2.5971, "step": 34534 }, { "epoch": 2.1438326401390526, "grad_norm": 0.5232049226039328, "learning_rate": 2.282531942065772e-05, "loss": 2.6952, "step": 34535 }, { "epoch": 2.1438947172388105, "grad_norm": 0.4420567167462827, "learning_rate": 2.2822287930106162e-05, "loss": 2.6648, "step": 34536 }, { "epoch": 2.1439567943385685, "grad_norm": 0.5365886591653773, "learning_rate": 2.281925658134732e-05, "loss": 2.7257, "step": 34537 }, { "epoch": 2.1440188714383264, "grad_norm": 0.5052698301135661, "learning_rate": 2.281622537439702e-05, "loss": 2.6811, "step": 34538 }, { "epoch": 2.1440809485380843, "grad_norm": 0.48557632430636, "learning_rate": 2.2813194309271052e-05, "loss": 2.7291, "step": 34539 }, { "epoch": 2.144143025637842, "grad_norm": 0.5147780195542528, "learning_rate": 2.2810163385985266e-05, "loss": 2.6423, "step": 34540 }, { "epoch": 2.1442051027376, "grad_norm": 0.5304451623989986, "learning_rate": 2.280713260455546e-05, "loss": 2.7491, "step": 34541 }, { "epoch": 2.144267179837358, "grad_norm": 0.5150109015394262, "learning_rate": 2.2804101964997444e-05, "loss": 2.7833, "step": 34542 }, { "epoch": 2.144329256937116, "grad_norm": 0.5343963629930762, "learning_rate": 2.280107146732701e-05, "loss": 2.6354, "step": 34543 }, { "epoch": 2.144391334036874, "grad_norm": 0.5302241462293078, "learning_rate": 2.279804111156001e-05, "loss": 2.6664, "step": 34544 }, { "epoch": 2.144453411136632, "grad_norm": 0.42827109190157736, "learning_rate": 2.279501089771223e-05, "loss": 2.6231, "step": 34545 }, { "epoch": 2.1445154882363897, "grad_norm": 0.5907761486350789, "learning_rate": 2.2791980825799485e-05, "loss": 2.707, "step": 34546 }, { "epoch": 2.1445775653361476, "grad_norm": 0.4641292755172487, "learning_rate": 2.2788950895837584e-05, "loss": 2.6963, "step": 34547 }, { "epoch": 2.1446396424359055, "grad_norm": 0.5484524999235519, "learning_rate": 2.2785921107842313e-05, "loss": 2.615, "step": 34548 }, { "epoch": 2.1447017195356635, "grad_norm": 0.5319611893960128, "learning_rate": 2.278289146182951e-05, "loss": 2.7595, "step": 34549 }, { "epoch": 2.1447637966354214, "grad_norm": 0.4632909950659515, "learning_rate": 2.2779861957814975e-05, "loss": 2.7267, "step": 34550 }, { "epoch": 2.1448258737351793, "grad_norm": 0.5066403873534222, "learning_rate": 2.27768325958145e-05, "loss": 2.6666, "step": 34551 }, { "epoch": 2.1448879508349368, "grad_norm": 0.49858787922883147, "learning_rate": 2.2773803375843888e-05, "loss": 2.645, "step": 34552 }, { "epoch": 2.1449500279346947, "grad_norm": 0.48803303322323066, "learning_rate": 2.2770774297918968e-05, "loss": 2.6614, "step": 34553 }, { "epoch": 2.1450121050344526, "grad_norm": 0.501774262499784, "learning_rate": 2.276774536205553e-05, "loss": 2.663, "step": 34554 }, { "epoch": 2.1450741821342105, "grad_norm": 0.5163776222408056, "learning_rate": 2.2764716568269356e-05, "loss": 2.7398, "step": 34555 }, { "epoch": 2.1451362592339684, "grad_norm": 0.5940854480210094, "learning_rate": 2.2761687916576284e-05, "loss": 2.7911, "step": 34556 }, { "epoch": 2.1451983363337264, "grad_norm": 0.5639048094399418, "learning_rate": 2.2758659406992105e-05, "loss": 2.5324, "step": 34557 }, { "epoch": 2.1452604134334843, "grad_norm": 0.5358901734211594, "learning_rate": 2.2755631039532587e-05, "loss": 2.7544, "step": 34558 }, { "epoch": 2.145322490533242, "grad_norm": 0.5394632259807116, "learning_rate": 2.275260281421358e-05, "loss": 2.6999, "step": 34559 }, { "epoch": 2.145384567633, "grad_norm": 0.6217061268765791, "learning_rate": 2.2749574731050856e-05, "loss": 2.7586, "step": 34560 }, { "epoch": 2.145446644732758, "grad_norm": 0.498356935812373, "learning_rate": 2.2746546790060213e-05, "loss": 2.5938, "step": 34561 }, { "epoch": 2.145508721832516, "grad_norm": 0.6056517293757505, "learning_rate": 2.2743518991257433e-05, "loss": 2.7495, "step": 34562 }, { "epoch": 2.145570798932274, "grad_norm": 0.5486978396605908, "learning_rate": 2.2740491334658352e-05, "loss": 2.704, "step": 34563 }, { "epoch": 2.145632876032032, "grad_norm": 0.5300468677365231, "learning_rate": 2.2737463820278747e-05, "loss": 2.6539, "step": 34564 }, { "epoch": 2.1456949531317897, "grad_norm": 0.6132599650777399, "learning_rate": 2.273443644813441e-05, "loss": 2.7433, "step": 34565 }, { "epoch": 2.1457570302315476, "grad_norm": 0.5743395619880278, "learning_rate": 2.2731409218241124e-05, "loss": 2.7755, "step": 34566 }, { "epoch": 2.1458191073313055, "grad_norm": 0.502577367539193, "learning_rate": 2.2728382130614705e-05, "loss": 2.7428, "step": 34567 }, { "epoch": 2.1458811844310635, "grad_norm": 0.5560075070591337, "learning_rate": 2.2725355185270942e-05, "loss": 2.653, "step": 34568 }, { "epoch": 2.1459432615308214, "grad_norm": 0.5211466006082024, "learning_rate": 2.272232838222562e-05, "loss": 2.7278, "step": 34569 }, { "epoch": 2.1460053386305793, "grad_norm": 0.4737868854788143, "learning_rate": 2.2719301721494534e-05, "loss": 2.7705, "step": 34570 }, { "epoch": 2.146067415730337, "grad_norm": 0.5903800648220219, "learning_rate": 2.2716275203093455e-05, "loss": 2.7157, "step": 34571 }, { "epoch": 2.146129492830095, "grad_norm": 0.549416711762215, "learning_rate": 2.271324882703821e-05, "loss": 2.5957, "step": 34572 }, { "epoch": 2.146191569929853, "grad_norm": 0.5907040820588894, "learning_rate": 2.2710222593344565e-05, "loss": 2.7013, "step": 34573 }, { "epoch": 2.146253647029611, "grad_norm": 0.5363810647174002, "learning_rate": 2.270719650202832e-05, "loss": 2.5854, "step": 34574 }, { "epoch": 2.1463157241293684, "grad_norm": 0.5838084896935335, "learning_rate": 2.2704170553105237e-05, "loss": 2.6996, "step": 34575 }, { "epoch": 2.146377801229127, "grad_norm": 0.572533606579823, "learning_rate": 2.2701144746591142e-05, "loss": 2.7132, "step": 34576 }, { "epoch": 2.1464398783288843, "grad_norm": 0.6063687188619975, "learning_rate": 2.26981190825018e-05, "loss": 2.7816, "step": 34577 }, { "epoch": 2.146501955428642, "grad_norm": 0.5504401212465037, "learning_rate": 2.2695093560853e-05, "loss": 2.707, "step": 34578 }, { "epoch": 2.1465640325284, "grad_norm": 0.5493574203438644, "learning_rate": 2.26920681816605e-05, "loss": 2.6842, "step": 34579 }, { "epoch": 2.146626109628158, "grad_norm": 0.5523688789489504, "learning_rate": 2.2689042944940138e-05, "loss": 2.7022, "step": 34580 }, { "epoch": 2.146688186727916, "grad_norm": 0.49704676424317673, "learning_rate": 2.2686017850707663e-05, "loss": 2.6882, "step": 34581 }, { "epoch": 2.146750263827674, "grad_norm": 0.4720734084394369, "learning_rate": 2.268299289897886e-05, "loss": 2.731, "step": 34582 }, { "epoch": 2.146812340927432, "grad_norm": 0.4666676841310354, "learning_rate": 2.267996808976952e-05, "loss": 2.626, "step": 34583 }, { "epoch": 2.1468744180271897, "grad_norm": 0.5437314556511208, "learning_rate": 2.2676943423095404e-05, "loss": 2.7307, "step": 34584 }, { "epoch": 2.1469364951269476, "grad_norm": 0.46367082153817674, "learning_rate": 2.267391889897232e-05, "loss": 2.7135, "step": 34585 }, { "epoch": 2.1469985722267055, "grad_norm": 0.5210810898554035, "learning_rate": 2.2670894517416036e-05, "loss": 2.7172, "step": 34586 }, { "epoch": 2.1470606493264635, "grad_norm": 0.5539758173160952, "learning_rate": 2.2667870278442327e-05, "loss": 2.7392, "step": 34587 }, { "epoch": 2.1471227264262214, "grad_norm": 0.5485970495999944, "learning_rate": 2.2664846182066957e-05, "loss": 2.7797, "step": 34588 }, { "epoch": 2.1471848035259793, "grad_norm": 0.511234112191903, "learning_rate": 2.2661822228305724e-05, "loss": 2.7065, "step": 34589 }, { "epoch": 2.147246880625737, "grad_norm": 0.5590220056818952, "learning_rate": 2.2658798417174415e-05, "loss": 2.795, "step": 34590 }, { "epoch": 2.147308957725495, "grad_norm": 0.5198253286874069, "learning_rate": 2.2655774748688796e-05, "loss": 2.6262, "step": 34591 }, { "epoch": 2.147371034825253, "grad_norm": 0.5421047884159806, "learning_rate": 2.265275122286463e-05, "loss": 2.6715, "step": 34592 }, { "epoch": 2.147433111925011, "grad_norm": 0.5495020845042184, "learning_rate": 2.2649727839717694e-05, "loss": 2.6468, "step": 34593 }, { "epoch": 2.147495189024769, "grad_norm": 0.5317205140619555, "learning_rate": 2.2646704599263774e-05, "loss": 2.7038, "step": 34594 }, { "epoch": 2.147557266124527, "grad_norm": 0.5049431357141818, "learning_rate": 2.2643681501518643e-05, "loss": 2.7161, "step": 34595 }, { "epoch": 2.1476193432242847, "grad_norm": 0.4864093299079534, "learning_rate": 2.2640658546498068e-05, "loss": 2.7107, "step": 34596 }, { "epoch": 2.1476814203240426, "grad_norm": 0.4961285015760182, "learning_rate": 2.263763573421781e-05, "loss": 2.7011, "step": 34597 }, { "epoch": 2.1477434974238006, "grad_norm": 0.5547019258154529, "learning_rate": 2.263461306469364e-05, "loss": 2.6711, "step": 34598 }, { "epoch": 2.1478055745235585, "grad_norm": 0.5113822762627409, "learning_rate": 2.263159053794135e-05, "loss": 2.7828, "step": 34599 }, { "epoch": 2.147867651623316, "grad_norm": 0.5315867086254428, "learning_rate": 2.2628568153976696e-05, "loss": 2.6943, "step": 34600 }, { "epoch": 2.147929728723074, "grad_norm": 0.5294958493715737, "learning_rate": 2.262554591281545e-05, "loss": 2.7851, "step": 34601 }, { "epoch": 2.147991805822832, "grad_norm": 0.5450627134707846, "learning_rate": 2.2622523814473355e-05, "loss": 2.6771, "step": 34602 }, { "epoch": 2.1480538829225897, "grad_norm": 0.5520126159528945, "learning_rate": 2.261950185896622e-05, "loss": 2.6227, "step": 34603 }, { "epoch": 2.1481159600223476, "grad_norm": 0.5439457782079049, "learning_rate": 2.261648004630979e-05, "loss": 2.6982, "step": 34604 }, { "epoch": 2.1481780371221055, "grad_norm": 0.5491005163867324, "learning_rate": 2.261345837651983e-05, "loss": 2.649, "step": 34605 }, { "epoch": 2.1482401142218635, "grad_norm": 0.5637773210287008, "learning_rate": 2.2610436849612103e-05, "loss": 2.6001, "step": 34606 }, { "epoch": 2.1483021913216214, "grad_norm": 0.4995331521903373, "learning_rate": 2.2607415465602367e-05, "loss": 2.571, "step": 34607 }, { "epoch": 2.1483642684213793, "grad_norm": 0.614803853855141, "learning_rate": 2.2604394224506405e-05, "loss": 2.7479, "step": 34608 }, { "epoch": 2.148426345521137, "grad_norm": 0.501673163924277, "learning_rate": 2.260137312633997e-05, "loss": 2.6412, "step": 34609 }, { "epoch": 2.148488422620895, "grad_norm": 0.5097955570714261, "learning_rate": 2.2598352171118825e-05, "loss": 2.7011, "step": 34610 }, { "epoch": 2.148550499720653, "grad_norm": 0.5215056054297735, "learning_rate": 2.2595331358858707e-05, "loss": 2.6615, "step": 34611 }, { "epoch": 2.148612576820411, "grad_norm": 0.4709614246824604, "learning_rate": 2.259231068957542e-05, "loss": 2.6752, "step": 34612 }, { "epoch": 2.148674653920169, "grad_norm": 0.53590457692798, "learning_rate": 2.25892901632847e-05, "loss": 2.6811, "step": 34613 }, { "epoch": 2.148736731019927, "grad_norm": 0.48897856702035475, "learning_rate": 2.2586269780002307e-05, "loss": 2.6699, "step": 34614 }, { "epoch": 2.1487988081196847, "grad_norm": 0.4401850729467429, "learning_rate": 2.2583249539743978e-05, "loss": 2.5897, "step": 34615 }, { "epoch": 2.1488608852194426, "grad_norm": 0.49634539612116013, "learning_rate": 2.2580229442525513e-05, "loss": 2.6815, "step": 34616 }, { "epoch": 2.1489229623192005, "grad_norm": 0.49942949649382806, "learning_rate": 2.257720948836265e-05, "loss": 2.746, "step": 34617 }, { "epoch": 2.1489850394189585, "grad_norm": 0.528217665313201, "learning_rate": 2.257418967727114e-05, "loss": 2.7314, "step": 34618 }, { "epoch": 2.1490471165187164, "grad_norm": 0.5024328351564092, "learning_rate": 2.257117000926674e-05, "loss": 2.6893, "step": 34619 }, { "epoch": 2.1491091936184743, "grad_norm": 0.5636673576189826, "learning_rate": 2.2568150484365184e-05, "loss": 2.6274, "step": 34620 }, { "epoch": 2.149171270718232, "grad_norm": 0.49496722536727084, "learning_rate": 2.256513110258225e-05, "loss": 2.6353, "step": 34621 }, { "epoch": 2.14923334781799, "grad_norm": 0.5758870342082766, "learning_rate": 2.2562111863933706e-05, "loss": 2.6398, "step": 34622 }, { "epoch": 2.1492954249177476, "grad_norm": 0.5034924071931781, "learning_rate": 2.255909276843528e-05, "loss": 2.6843, "step": 34623 }, { "epoch": 2.149357502017506, "grad_norm": 0.5492743853285319, "learning_rate": 2.2556073816102735e-05, "loss": 2.6974, "step": 34624 }, { "epoch": 2.1494195791172634, "grad_norm": 0.4529196668588093, "learning_rate": 2.2553055006951795e-05, "loss": 2.752, "step": 34625 }, { "epoch": 2.1494816562170214, "grad_norm": 0.5561733760992992, "learning_rate": 2.2550036340998244e-05, "loss": 2.7066, "step": 34626 }, { "epoch": 2.1495437333167793, "grad_norm": 0.44624370119046525, "learning_rate": 2.254701781825782e-05, "loss": 2.7215, "step": 34627 }, { "epoch": 2.149605810416537, "grad_norm": 0.5472196310512344, "learning_rate": 2.2543999438746272e-05, "loss": 2.6666, "step": 34628 }, { "epoch": 2.149667887516295, "grad_norm": 0.5619776207516973, "learning_rate": 2.254098120247932e-05, "loss": 2.6849, "step": 34629 }, { "epoch": 2.149729964616053, "grad_norm": 0.5064865995820451, "learning_rate": 2.253796310947276e-05, "loss": 2.6548, "step": 34630 }, { "epoch": 2.149792041715811, "grad_norm": 0.5752440142131524, "learning_rate": 2.2534945159742306e-05, "loss": 2.7351, "step": 34631 }, { "epoch": 2.149854118815569, "grad_norm": 0.5035026767625467, "learning_rate": 2.253192735330371e-05, "loss": 2.5685, "step": 34632 }, { "epoch": 2.149916195915327, "grad_norm": 0.5249516237768088, "learning_rate": 2.2528909690172723e-05, "loss": 2.7153, "step": 34633 }, { "epoch": 2.1499782730150847, "grad_norm": 0.5445963663274866, "learning_rate": 2.2525892170365064e-05, "loss": 2.6791, "step": 34634 }, { "epoch": 2.1500403501148426, "grad_norm": 0.536528227860736, "learning_rate": 2.252287479389651e-05, "loss": 2.6666, "step": 34635 }, { "epoch": 2.1501024272146005, "grad_norm": 0.5407389956116035, "learning_rate": 2.251985756078279e-05, "loss": 2.7089, "step": 34636 }, { "epoch": 2.1501645043143585, "grad_norm": 0.5269917761651418, "learning_rate": 2.2516840471039647e-05, "loss": 2.5883, "step": 34637 }, { "epoch": 2.1502265814141164, "grad_norm": 0.6235185948071615, "learning_rate": 2.2513823524682796e-05, "loss": 2.6766, "step": 34638 }, { "epoch": 2.1502886585138743, "grad_norm": 0.5562978391756576, "learning_rate": 2.2510806721728017e-05, "loss": 2.6629, "step": 34639 }, { "epoch": 2.150350735613632, "grad_norm": 0.6437217483987723, "learning_rate": 2.2507790062191037e-05, "loss": 2.6361, "step": 34640 }, { "epoch": 2.15041281271339, "grad_norm": 0.6618321606439854, "learning_rate": 2.2504773546087594e-05, "loss": 2.7435, "step": 34641 }, { "epoch": 2.150474889813148, "grad_norm": 0.5537859077091986, "learning_rate": 2.2501757173433412e-05, "loss": 2.6021, "step": 34642 }, { "epoch": 2.150536966912906, "grad_norm": 0.6813471252430588, "learning_rate": 2.249874094424423e-05, "loss": 2.672, "step": 34643 }, { "epoch": 2.150599044012664, "grad_norm": 0.5092324578171672, "learning_rate": 2.2495724858535805e-05, "loss": 2.7054, "step": 34644 }, { "epoch": 2.150661121112422, "grad_norm": 0.6286175936150533, "learning_rate": 2.2492708916323862e-05, "loss": 2.7921, "step": 34645 }, { "epoch": 2.1507231982121797, "grad_norm": 0.5236877298779136, "learning_rate": 2.2489693117624135e-05, "loss": 2.692, "step": 34646 }, { "epoch": 2.1507852753119376, "grad_norm": 0.4911326806770548, "learning_rate": 2.248667746245233e-05, "loss": 2.7278, "step": 34647 }, { "epoch": 2.150847352411695, "grad_norm": 0.5612141429092419, "learning_rate": 2.2483661950824237e-05, "loss": 2.7357, "step": 34648 }, { "epoch": 2.150909429511453, "grad_norm": 0.5817515321931879, "learning_rate": 2.248064658275556e-05, "loss": 2.6226, "step": 34649 }, { "epoch": 2.150971506611211, "grad_norm": 0.5313065162216734, "learning_rate": 2.2477631358262026e-05, "loss": 2.6557, "step": 34650 }, { "epoch": 2.151033583710969, "grad_norm": 0.4908888653972171, "learning_rate": 2.247461627735935e-05, "loss": 2.618, "step": 34651 }, { "epoch": 2.151095660810727, "grad_norm": 0.5806178398224644, "learning_rate": 2.2471601340063304e-05, "loss": 2.7711, "step": 34652 }, { "epoch": 2.1511577379104847, "grad_norm": 0.47911807691537045, "learning_rate": 2.2468586546389596e-05, "loss": 2.5876, "step": 34653 }, { "epoch": 2.1512198150102426, "grad_norm": 0.5320296176879759, "learning_rate": 2.246557189635394e-05, "loss": 2.8015, "step": 34654 }, { "epoch": 2.1512818921100005, "grad_norm": 0.486711328122296, "learning_rate": 2.2462557389972095e-05, "loss": 2.6796, "step": 34655 }, { "epoch": 2.1513439692097585, "grad_norm": 0.5365240063020283, "learning_rate": 2.245954302725977e-05, "loss": 2.7296, "step": 34656 }, { "epoch": 2.1514060463095164, "grad_norm": 0.4924924904594494, "learning_rate": 2.2456528808232686e-05, "loss": 2.5639, "step": 34657 }, { "epoch": 2.1514681234092743, "grad_norm": 0.5709446056072047, "learning_rate": 2.245351473290659e-05, "loss": 2.7196, "step": 34658 }, { "epoch": 2.151530200509032, "grad_norm": 0.5412842121720439, "learning_rate": 2.245050080129719e-05, "loss": 2.7345, "step": 34659 }, { "epoch": 2.15159227760879, "grad_norm": 0.4975470933717538, "learning_rate": 2.244748701342022e-05, "loss": 2.6613, "step": 34660 }, { "epoch": 2.151654354708548, "grad_norm": 0.580276070161572, "learning_rate": 2.2444473369291387e-05, "loss": 2.6624, "step": 34661 }, { "epoch": 2.151716431808306, "grad_norm": 0.4914843971690238, "learning_rate": 2.2441459868926435e-05, "loss": 2.6397, "step": 34662 }, { "epoch": 2.151778508908064, "grad_norm": 0.5641516438047179, "learning_rate": 2.2438446512341076e-05, "loss": 2.7088, "step": 34663 }, { "epoch": 2.151840586007822, "grad_norm": 0.5333571997408597, "learning_rate": 2.243543329955104e-05, "loss": 2.666, "step": 34664 }, { "epoch": 2.1519026631075797, "grad_norm": 0.5635877264899709, "learning_rate": 2.2432420230572015e-05, "loss": 2.7067, "step": 34665 }, { "epoch": 2.1519647402073376, "grad_norm": 0.49232223725238905, "learning_rate": 2.242940730541977e-05, "loss": 2.7223, "step": 34666 }, { "epoch": 2.1520268173070956, "grad_norm": 0.5313526503179776, "learning_rate": 2.2426394524109994e-05, "loss": 2.7131, "step": 34667 }, { "epoch": 2.1520888944068535, "grad_norm": 0.580966769707681, "learning_rate": 2.2423381886658417e-05, "loss": 2.7244, "step": 34668 }, { "epoch": 2.1521509715066114, "grad_norm": 0.5113395567243374, "learning_rate": 2.242036939308075e-05, "loss": 2.6242, "step": 34669 }, { "epoch": 2.1522130486063693, "grad_norm": 0.5126695026985635, "learning_rate": 2.241735704339269e-05, "loss": 2.6551, "step": 34670 }, { "epoch": 2.152275125706127, "grad_norm": 0.5008019929538474, "learning_rate": 2.241434483760999e-05, "loss": 2.6965, "step": 34671 }, { "epoch": 2.152337202805885, "grad_norm": 0.5208796734857686, "learning_rate": 2.2411332775748356e-05, "loss": 2.6708, "step": 34672 }, { "epoch": 2.1523992799056426, "grad_norm": 0.5171196400357931, "learning_rate": 2.2408320857823495e-05, "loss": 2.6892, "step": 34673 }, { "epoch": 2.1524613570054005, "grad_norm": 0.5022030516176623, "learning_rate": 2.24053090838511e-05, "loss": 2.7025, "step": 34674 }, { "epoch": 2.1525234341051585, "grad_norm": 0.5858765888343137, "learning_rate": 2.2402297453846927e-05, "loss": 2.7305, "step": 34675 }, { "epoch": 2.1525855112049164, "grad_norm": 0.5171285728237113, "learning_rate": 2.2399285967826672e-05, "loss": 2.6007, "step": 34676 }, { "epoch": 2.1526475883046743, "grad_norm": 0.5608539604098873, "learning_rate": 2.2396274625806034e-05, "loss": 2.76, "step": 34677 }, { "epoch": 2.152709665404432, "grad_norm": 0.5711001091849613, "learning_rate": 2.239326342780072e-05, "loss": 2.6397, "step": 34678 }, { "epoch": 2.15277174250419, "grad_norm": 0.5465367497288437, "learning_rate": 2.2390252373826472e-05, "loss": 2.7538, "step": 34679 }, { "epoch": 2.152833819603948, "grad_norm": 0.5367926478701808, "learning_rate": 2.238724146389898e-05, "loss": 2.6349, "step": 34680 }, { "epoch": 2.152895896703706, "grad_norm": 0.5167470629755793, "learning_rate": 2.2384230698033947e-05, "loss": 2.6425, "step": 34681 }, { "epoch": 2.152957973803464, "grad_norm": 0.4674078801116399, "learning_rate": 2.238122007624709e-05, "loss": 2.6668, "step": 34682 }, { "epoch": 2.153020050903222, "grad_norm": 0.5294064904476931, "learning_rate": 2.237820959855409e-05, "loss": 2.5933, "step": 34683 }, { "epoch": 2.1530821280029797, "grad_norm": 0.5055157587989001, "learning_rate": 2.23751992649707e-05, "loss": 2.6365, "step": 34684 }, { "epoch": 2.1531442051027376, "grad_norm": 0.5079370786003993, "learning_rate": 2.2372189075512597e-05, "loss": 2.6817, "step": 34685 }, { "epoch": 2.1532062822024955, "grad_norm": 0.5367258263000698, "learning_rate": 2.2369179030195476e-05, "loss": 2.7681, "step": 34686 }, { "epoch": 2.1532683593022535, "grad_norm": 0.46891737175081666, "learning_rate": 2.236616912903507e-05, "loss": 2.6227, "step": 34687 }, { "epoch": 2.1533304364020114, "grad_norm": 0.5141443030946452, "learning_rate": 2.2363159372047056e-05, "loss": 2.6582, "step": 34688 }, { "epoch": 2.1533925135017693, "grad_norm": 0.5527439723175737, "learning_rate": 2.236014975924716e-05, "loss": 2.7531, "step": 34689 }, { "epoch": 2.153454590601527, "grad_norm": 0.49726697407649595, "learning_rate": 2.2357140290651074e-05, "loss": 2.7775, "step": 34690 }, { "epoch": 2.153516667701285, "grad_norm": 0.48371365220876816, "learning_rate": 2.23541309662745e-05, "loss": 2.6978, "step": 34691 }, { "epoch": 2.153578744801043, "grad_norm": 0.5612077704771932, "learning_rate": 2.2351121786133133e-05, "loss": 2.6756, "step": 34692 }, { "epoch": 2.153640821900801, "grad_norm": 0.5497283340476516, "learning_rate": 2.234811275024266e-05, "loss": 2.6806, "step": 34693 }, { "epoch": 2.153702899000559, "grad_norm": 0.49196377468107316, "learning_rate": 2.2345103858618815e-05, "loss": 2.651, "step": 34694 }, { "epoch": 2.153764976100317, "grad_norm": 0.5456986445554984, "learning_rate": 2.2342095111277273e-05, "loss": 2.6415, "step": 34695 }, { "epoch": 2.1538270532000743, "grad_norm": 0.5137849400465947, "learning_rate": 2.2339086508233737e-05, "loss": 2.6864, "step": 34696 }, { "epoch": 2.153889130299832, "grad_norm": 0.527131497088659, "learning_rate": 2.2336078049503883e-05, "loss": 2.6867, "step": 34697 }, { "epoch": 2.15395120739959, "grad_norm": 0.5315059998730751, "learning_rate": 2.2333069735103445e-05, "loss": 2.7174, "step": 34698 }, { "epoch": 2.154013284499348, "grad_norm": 0.525972774136425, "learning_rate": 2.2330061565048098e-05, "loss": 2.6554, "step": 34699 }, { "epoch": 2.154075361599106, "grad_norm": 0.5288957434192467, "learning_rate": 2.2327053539353538e-05, "loss": 2.6539, "step": 34700 }, { "epoch": 2.154137438698864, "grad_norm": 0.5218453461679561, "learning_rate": 2.2324045658035436e-05, "loss": 2.652, "step": 34701 }, { "epoch": 2.154199515798622, "grad_norm": 0.5623984245141211, "learning_rate": 2.232103792110953e-05, "loss": 2.6371, "step": 34702 }, { "epoch": 2.1542615928983797, "grad_norm": 0.5086457678566767, "learning_rate": 2.2318030328591484e-05, "loss": 2.6965, "step": 34703 }, { "epoch": 2.1543236699981376, "grad_norm": 0.5524086413648139, "learning_rate": 2.2315022880496996e-05, "loss": 2.6803, "step": 34704 }, { "epoch": 2.1543857470978955, "grad_norm": 0.5224583705945997, "learning_rate": 2.231201557684175e-05, "loss": 2.7038, "step": 34705 }, { "epoch": 2.1544478241976535, "grad_norm": 0.47019597190943163, "learning_rate": 2.2309008417641424e-05, "loss": 2.672, "step": 34706 }, { "epoch": 2.1545099012974114, "grad_norm": 0.5840575768766902, "learning_rate": 2.2306001402911742e-05, "loss": 2.6763, "step": 34707 }, { "epoch": 2.1545719783971693, "grad_norm": 0.49570104119093567, "learning_rate": 2.2302994532668375e-05, "loss": 2.6654, "step": 34708 }, { "epoch": 2.154634055496927, "grad_norm": 0.46312627146104907, "learning_rate": 2.229998780692701e-05, "loss": 2.6608, "step": 34709 }, { "epoch": 2.154696132596685, "grad_norm": 0.6163431500336845, "learning_rate": 2.229698122570331e-05, "loss": 2.7315, "step": 34710 }, { "epoch": 2.154758209696443, "grad_norm": 0.4387933652862619, "learning_rate": 2.2293974789013006e-05, "loss": 2.7535, "step": 34711 }, { "epoch": 2.154820286796201, "grad_norm": 0.4795439589262209, "learning_rate": 2.2290968496871763e-05, "loss": 2.6773, "step": 34712 }, { "epoch": 2.154882363895959, "grad_norm": 0.5184582406851005, "learning_rate": 2.228796234929526e-05, "loss": 2.6802, "step": 34713 }, { "epoch": 2.154944440995717, "grad_norm": 0.48763393982977693, "learning_rate": 2.228495634629917e-05, "loss": 2.6245, "step": 34714 }, { "epoch": 2.1550065180954747, "grad_norm": 0.5714231154487904, "learning_rate": 2.2281950487899205e-05, "loss": 2.7116, "step": 34715 }, { "epoch": 2.1550685951952326, "grad_norm": 0.51992392937235, "learning_rate": 2.227894477411104e-05, "loss": 2.6534, "step": 34716 }, { "epoch": 2.1551306722949906, "grad_norm": 0.503255675506046, "learning_rate": 2.227593920495034e-05, "loss": 2.6862, "step": 34717 }, { "epoch": 2.1551927493947485, "grad_norm": 0.5306690349212514, "learning_rate": 2.22729337804328e-05, "loss": 2.6378, "step": 34718 }, { "epoch": 2.155254826494506, "grad_norm": 0.505151496390511, "learning_rate": 2.2269928500574085e-05, "loss": 2.642, "step": 34719 }, { "epoch": 2.1553169035942643, "grad_norm": 0.561000945772132, "learning_rate": 2.2266923365389876e-05, "loss": 2.7482, "step": 34720 }, { "epoch": 2.155378980694022, "grad_norm": 0.5298951768680459, "learning_rate": 2.2263918374895876e-05, "loss": 2.7445, "step": 34721 }, { "epoch": 2.1554410577937797, "grad_norm": 0.5138844967416704, "learning_rate": 2.226091352910775e-05, "loss": 2.767, "step": 34722 }, { "epoch": 2.1555031348935376, "grad_norm": 0.5444881394778575, "learning_rate": 2.2257908828041175e-05, "loss": 2.6369, "step": 34723 }, { "epoch": 2.1555652119932955, "grad_norm": 0.5029438794550256, "learning_rate": 2.22549042717118e-05, "loss": 2.7196, "step": 34724 }, { "epoch": 2.1556272890930535, "grad_norm": 0.5196949120209174, "learning_rate": 2.225189986013535e-05, "loss": 2.6678, "step": 34725 }, { "epoch": 2.1556893661928114, "grad_norm": 0.49528068837438405, "learning_rate": 2.2248895593327473e-05, "loss": 2.7429, "step": 34726 }, { "epoch": 2.1557514432925693, "grad_norm": 0.47229644384549674, "learning_rate": 2.2245891471303842e-05, "loss": 2.7014, "step": 34727 }, { "epoch": 2.155813520392327, "grad_norm": 0.5218373507778888, "learning_rate": 2.2242887494080133e-05, "loss": 2.7188, "step": 34728 }, { "epoch": 2.155875597492085, "grad_norm": 0.5454594957784507, "learning_rate": 2.2239883661672e-05, "loss": 2.6262, "step": 34729 }, { "epoch": 2.155937674591843, "grad_norm": 0.5078455694141487, "learning_rate": 2.2236879974095153e-05, "loss": 2.7043, "step": 34730 }, { "epoch": 2.155999751691601, "grad_norm": 0.5150844457057134, "learning_rate": 2.2233876431365242e-05, "loss": 2.7705, "step": 34731 }, { "epoch": 2.156061828791359, "grad_norm": 0.5736386749207001, "learning_rate": 2.223087303349794e-05, "loss": 2.7064, "step": 34732 }, { "epoch": 2.156123905891117, "grad_norm": 0.45333280402332615, "learning_rate": 2.2227869780508893e-05, "loss": 2.7083, "step": 34733 }, { "epoch": 2.1561859829908747, "grad_norm": 0.6071751820349953, "learning_rate": 2.2224866672413814e-05, "loss": 2.7257, "step": 34734 }, { "epoch": 2.1562480600906326, "grad_norm": 0.5155097948322249, "learning_rate": 2.222186370922834e-05, "loss": 2.752, "step": 34735 }, { "epoch": 2.1563101371903906, "grad_norm": 0.5285002565933659, "learning_rate": 2.2218860890968152e-05, "loss": 2.6866, "step": 34736 }, { "epoch": 2.1563722142901485, "grad_norm": 0.5290879685411221, "learning_rate": 2.2215858217648894e-05, "loss": 2.5962, "step": 34737 }, { "epoch": 2.1564342913899064, "grad_norm": 0.5669903867182451, "learning_rate": 2.2212855689286267e-05, "loss": 2.6593, "step": 34738 }, { "epoch": 2.1564963684896643, "grad_norm": 0.594593528062247, "learning_rate": 2.220985330589591e-05, "loss": 2.738, "step": 34739 }, { "epoch": 2.1565584455894222, "grad_norm": 0.5011174708367184, "learning_rate": 2.22068510674935e-05, "loss": 2.681, "step": 34740 }, { "epoch": 2.15662052268918, "grad_norm": 0.6099231162241222, "learning_rate": 2.22038489740947e-05, "loss": 2.702, "step": 34741 }, { "epoch": 2.156682599788938, "grad_norm": 0.5113753119975695, "learning_rate": 2.2200847025715142e-05, "loss": 2.666, "step": 34742 }, { "epoch": 2.156744676888696, "grad_norm": 0.5395960171754017, "learning_rate": 2.2197845222370532e-05, "loss": 2.6834, "step": 34743 }, { "epoch": 2.1568067539884535, "grad_norm": 0.5656699557906545, "learning_rate": 2.2194843564076516e-05, "loss": 2.7108, "step": 34744 }, { "epoch": 2.1568688310882114, "grad_norm": 0.49157899752492434, "learning_rate": 2.2191842050848748e-05, "loss": 2.6755, "step": 34745 }, { "epoch": 2.1569309081879693, "grad_norm": 0.5463790279389192, "learning_rate": 2.2188840682702876e-05, "loss": 2.7391, "step": 34746 }, { "epoch": 2.156992985287727, "grad_norm": 0.596131317834697, "learning_rate": 2.218583945965459e-05, "loss": 2.6933, "step": 34747 }, { "epoch": 2.157055062387485, "grad_norm": 0.5603675156586418, "learning_rate": 2.2182838381719536e-05, "loss": 2.716, "step": 34748 }, { "epoch": 2.157117139487243, "grad_norm": 0.514489657901439, "learning_rate": 2.217983744891336e-05, "loss": 2.6805, "step": 34749 }, { "epoch": 2.157179216587001, "grad_norm": 0.5625498802533478, "learning_rate": 2.217683666125171e-05, "loss": 2.6815, "step": 34750 }, { "epoch": 2.157241293686759, "grad_norm": 0.5011395343624379, "learning_rate": 2.217383601875027e-05, "loss": 2.7062, "step": 34751 }, { "epoch": 2.157303370786517, "grad_norm": 0.4732449388601165, "learning_rate": 2.217083552142467e-05, "loss": 2.7104, "step": 34752 }, { "epoch": 2.1573654478862747, "grad_norm": 0.5722261747230926, "learning_rate": 2.21678351692906e-05, "loss": 2.7422, "step": 34753 }, { "epoch": 2.1574275249860326, "grad_norm": 0.466691472112361, "learning_rate": 2.2164834962363684e-05, "loss": 2.5468, "step": 34754 }, { "epoch": 2.1574896020857905, "grad_norm": 0.4712653383748851, "learning_rate": 2.2161834900659584e-05, "loss": 2.6356, "step": 34755 }, { "epoch": 2.1575516791855485, "grad_norm": 0.5203179161545948, "learning_rate": 2.2158834984193926e-05, "loss": 2.7496, "step": 34756 }, { "epoch": 2.1576137562853064, "grad_norm": 0.5196132165023774, "learning_rate": 2.215583521298241e-05, "loss": 2.6737, "step": 34757 }, { "epoch": 2.1576758333850643, "grad_norm": 0.48227899574569133, "learning_rate": 2.2152835587040655e-05, "loss": 2.6246, "step": 34758 }, { "epoch": 2.157737910484822, "grad_norm": 0.5446553826287704, "learning_rate": 2.214983610638432e-05, "loss": 2.6279, "step": 34759 }, { "epoch": 2.15779998758458, "grad_norm": 0.4981654904803552, "learning_rate": 2.2146836771029035e-05, "loss": 2.8294, "step": 34760 }, { "epoch": 2.157862064684338, "grad_norm": 0.5116889013073423, "learning_rate": 2.2143837580990477e-05, "loss": 2.657, "step": 34761 }, { "epoch": 2.157924141784096, "grad_norm": 0.5091830026323323, "learning_rate": 2.2140838536284288e-05, "loss": 2.7812, "step": 34762 }, { "epoch": 2.157986218883854, "grad_norm": 0.5317839295521007, "learning_rate": 2.21378396369261e-05, "loss": 2.7304, "step": 34763 }, { "epoch": 2.158048295983612, "grad_norm": 0.5295692109216594, "learning_rate": 2.213484088293155e-05, "loss": 2.672, "step": 34764 }, { "epoch": 2.1581103730833697, "grad_norm": 0.5330165494689896, "learning_rate": 2.2131842274316318e-05, "loss": 2.698, "step": 34765 }, { "epoch": 2.1581724501831276, "grad_norm": 0.5072410596113236, "learning_rate": 2.2128843811096034e-05, "loss": 2.6944, "step": 34766 }, { "epoch": 2.158234527282885, "grad_norm": 0.5076194983294362, "learning_rate": 2.2125845493286334e-05, "loss": 2.6831, "step": 34767 }, { "epoch": 2.158296604382643, "grad_norm": 0.5222645328028122, "learning_rate": 2.2122847320902863e-05, "loss": 2.7153, "step": 34768 }, { "epoch": 2.158358681482401, "grad_norm": 0.5518417969648647, "learning_rate": 2.2119849293961248e-05, "loss": 2.7394, "step": 34769 }, { "epoch": 2.158420758582159, "grad_norm": 0.5338031062076876, "learning_rate": 2.211685141247717e-05, "loss": 2.7006, "step": 34770 }, { "epoch": 2.158482835681917, "grad_norm": 0.5115500601605799, "learning_rate": 2.211385367646624e-05, "loss": 2.6592, "step": 34771 }, { "epoch": 2.1585449127816747, "grad_norm": 0.49627212261307224, "learning_rate": 2.2110856085944108e-05, "loss": 2.7299, "step": 34772 }, { "epoch": 2.1586069898814326, "grad_norm": 0.5550795767843655, "learning_rate": 2.2107858640926393e-05, "loss": 2.6315, "step": 34773 }, { "epoch": 2.1586690669811905, "grad_norm": 0.5054456572880904, "learning_rate": 2.210486134142877e-05, "loss": 2.7151, "step": 34774 }, { "epoch": 2.1587311440809485, "grad_norm": 0.5471580590258455, "learning_rate": 2.2101864187466857e-05, "loss": 2.6969, "step": 34775 }, { "epoch": 2.1587932211807064, "grad_norm": 0.6064193892044545, "learning_rate": 2.2098867179056286e-05, "loss": 2.778, "step": 34776 }, { "epoch": 2.1588552982804643, "grad_norm": 0.5132861121192702, "learning_rate": 2.20958703162127e-05, "loss": 2.6598, "step": 34777 }, { "epoch": 2.158917375380222, "grad_norm": 0.4582043614415546, "learning_rate": 2.2092873598951718e-05, "loss": 2.6209, "step": 34778 }, { "epoch": 2.15897945247998, "grad_norm": 0.540141394303024, "learning_rate": 2.2089877027289007e-05, "loss": 2.6529, "step": 34779 }, { "epoch": 2.159041529579738, "grad_norm": 0.49907937828096016, "learning_rate": 2.2086880601240183e-05, "loss": 2.7235, "step": 34780 }, { "epoch": 2.159103606679496, "grad_norm": 0.48819544228942724, "learning_rate": 2.2083884320820874e-05, "loss": 2.7204, "step": 34781 }, { "epoch": 2.159165683779254, "grad_norm": 0.5389404158630023, "learning_rate": 2.2080888186046705e-05, "loss": 2.756, "step": 34782 }, { "epoch": 2.159227760879012, "grad_norm": 0.5511500970010673, "learning_rate": 2.2077892196933336e-05, "loss": 2.5766, "step": 34783 }, { "epoch": 2.1592898379787697, "grad_norm": 0.5324789306340805, "learning_rate": 2.2074896353496383e-05, "loss": 2.6293, "step": 34784 }, { "epoch": 2.1593519150785276, "grad_norm": 0.5187376170231586, "learning_rate": 2.2071900655751455e-05, "loss": 2.6515, "step": 34785 }, { "epoch": 2.1594139921782856, "grad_norm": 0.517629628048151, "learning_rate": 2.2068905103714216e-05, "loss": 2.5514, "step": 34786 }, { "epoch": 2.1594760692780435, "grad_norm": 0.49001857673697713, "learning_rate": 2.2065909697400267e-05, "loss": 2.6251, "step": 34787 }, { "epoch": 2.1595381463778014, "grad_norm": 0.48486762990854987, "learning_rate": 2.2062914436825262e-05, "loss": 2.5988, "step": 34788 }, { "epoch": 2.1596002234775593, "grad_norm": 0.49338037260410045, "learning_rate": 2.2059919322004817e-05, "loss": 2.637, "step": 34789 }, { "epoch": 2.1596623005773172, "grad_norm": 0.49699913969291665, "learning_rate": 2.2056924352954554e-05, "loss": 2.6894, "step": 34790 }, { "epoch": 2.159724377677075, "grad_norm": 0.46659672807669617, "learning_rate": 2.2053929529690104e-05, "loss": 2.6195, "step": 34791 }, { "epoch": 2.1597864547768326, "grad_norm": 0.48186024703858527, "learning_rate": 2.2050934852227067e-05, "loss": 2.6569, "step": 34792 }, { "epoch": 2.1598485318765905, "grad_norm": 0.5412914247936834, "learning_rate": 2.2047940320581105e-05, "loss": 2.7131, "step": 34793 }, { "epoch": 2.1599106089763485, "grad_norm": 0.5498474267097901, "learning_rate": 2.2044945934767823e-05, "loss": 2.6973, "step": 34794 }, { "epoch": 2.1599726860761064, "grad_norm": 0.4731442005640534, "learning_rate": 2.2041951694802847e-05, "loss": 2.7193, "step": 34795 }, { "epoch": 2.1600347631758643, "grad_norm": 0.44836316107271257, "learning_rate": 2.2038957600701776e-05, "loss": 2.6476, "step": 34796 }, { "epoch": 2.160096840275622, "grad_norm": 0.532523508090279, "learning_rate": 2.203596365248027e-05, "loss": 2.679, "step": 34797 }, { "epoch": 2.16015891737538, "grad_norm": 0.5082774747842581, "learning_rate": 2.2032969850153927e-05, "loss": 2.6832, "step": 34798 }, { "epoch": 2.160220994475138, "grad_norm": 0.49705854600571, "learning_rate": 2.202997619373838e-05, "loss": 2.6833, "step": 34799 }, { "epoch": 2.160283071574896, "grad_norm": 0.5272372950727481, "learning_rate": 2.2026982683249208e-05, "loss": 2.6119, "step": 34800 }, { "epoch": 2.160345148674654, "grad_norm": 0.5373544030050034, "learning_rate": 2.202398931870208e-05, "loss": 2.7523, "step": 34801 }, { "epoch": 2.160407225774412, "grad_norm": 0.4728486008612412, "learning_rate": 2.202099610011259e-05, "loss": 2.6584, "step": 34802 }, { "epoch": 2.1604693028741697, "grad_norm": 0.5251695529441166, "learning_rate": 2.201800302749636e-05, "loss": 2.7197, "step": 34803 }, { "epoch": 2.1605313799739276, "grad_norm": 0.5210960456763262, "learning_rate": 2.2015010100868992e-05, "loss": 2.6341, "step": 34804 }, { "epoch": 2.1605934570736856, "grad_norm": 0.5185264967674847, "learning_rate": 2.2012017320246097e-05, "loss": 2.7114, "step": 34805 }, { "epoch": 2.1606555341734435, "grad_norm": 0.5367416232592096, "learning_rate": 2.2009024685643326e-05, "loss": 2.7708, "step": 34806 }, { "epoch": 2.1607176112732014, "grad_norm": 0.5439395322098761, "learning_rate": 2.200603219707626e-05, "loss": 2.8018, "step": 34807 }, { "epoch": 2.1607796883729593, "grad_norm": 0.5150630952553676, "learning_rate": 2.2003039854560525e-05, "loss": 2.6173, "step": 34808 }, { "epoch": 2.1608417654727172, "grad_norm": 0.4826002679172362, "learning_rate": 2.2000047658111706e-05, "loss": 2.6602, "step": 34809 }, { "epoch": 2.160903842572475, "grad_norm": 0.5053350315956644, "learning_rate": 2.1997055607745455e-05, "loss": 2.6901, "step": 34810 }, { "epoch": 2.160965919672233, "grad_norm": 0.47308672068558383, "learning_rate": 2.1994063703477362e-05, "loss": 2.6945, "step": 34811 }, { "epoch": 2.161027996771991, "grad_norm": 0.508480542265448, "learning_rate": 2.1991071945323038e-05, "loss": 2.7387, "step": 34812 }, { "epoch": 2.161090073871749, "grad_norm": 0.5268926598518165, "learning_rate": 2.198808033329809e-05, "loss": 2.7311, "step": 34813 }, { "epoch": 2.161152150971507, "grad_norm": 0.4867449739595401, "learning_rate": 2.1985088867418107e-05, "loss": 2.6546, "step": 34814 }, { "epoch": 2.1612142280712643, "grad_norm": 0.4781488998523291, "learning_rate": 2.1982097547698737e-05, "loss": 2.7146, "step": 34815 }, { "epoch": 2.161276305171022, "grad_norm": 0.4836742038390384, "learning_rate": 2.197910637415557e-05, "loss": 2.68, "step": 34816 }, { "epoch": 2.16133838227078, "grad_norm": 0.5149749646536684, "learning_rate": 2.19761153468042e-05, "loss": 2.6755, "step": 34817 }, { "epoch": 2.161400459370538, "grad_norm": 0.5166681315288492, "learning_rate": 2.197312446566023e-05, "loss": 2.6643, "step": 34818 }, { "epoch": 2.161462536470296, "grad_norm": 0.48893773007744046, "learning_rate": 2.1970133730739263e-05, "loss": 2.707, "step": 34819 }, { "epoch": 2.161524613570054, "grad_norm": 0.5230012470653538, "learning_rate": 2.1967143142056935e-05, "loss": 2.7439, "step": 34820 }, { "epoch": 2.161586690669812, "grad_norm": 0.5145882964055752, "learning_rate": 2.1964152699628826e-05, "loss": 2.7267, "step": 34821 }, { "epoch": 2.1616487677695697, "grad_norm": 0.544331302356182, "learning_rate": 2.1961162403470543e-05, "loss": 2.681, "step": 34822 }, { "epoch": 2.1617108448693276, "grad_norm": 0.5063813804350674, "learning_rate": 2.195817225359766e-05, "loss": 2.7013, "step": 34823 }, { "epoch": 2.1617729219690855, "grad_norm": 0.522957140152861, "learning_rate": 2.195518225002582e-05, "loss": 2.5974, "step": 34824 }, { "epoch": 2.1618349990688435, "grad_norm": 0.5798073338858967, "learning_rate": 2.19521923927706e-05, "loss": 2.735, "step": 34825 }, { "epoch": 2.1618970761686014, "grad_norm": 0.49472063123518695, "learning_rate": 2.1949202681847603e-05, "loss": 2.6384, "step": 34826 }, { "epoch": 2.1619591532683593, "grad_norm": 0.5831892539519121, "learning_rate": 2.1946213117272425e-05, "loss": 2.6487, "step": 34827 }, { "epoch": 2.162021230368117, "grad_norm": 0.5452203240957018, "learning_rate": 2.1943223699060645e-05, "loss": 2.7107, "step": 34828 }, { "epoch": 2.162083307467875, "grad_norm": 0.5382794760797999, "learning_rate": 2.194023442722789e-05, "loss": 2.7121, "step": 34829 }, { "epoch": 2.162145384567633, "grad_norm": 0.49663073095571847, "learning_rate": 2.193724530178975e-05, "loss": 2.6485, "step": 34830 }, { "epoch": 2.162207461667391, "grad_norm": 0.5066662395887499, "learning_rate": 2.1934256322761814e-05, "loss": 2.6919, "step": 34831 }, { "epoch": 2.162269538767149, "grad_norm": 0.5512400941110094, "learning_rate": 2.1931267490159658e-05, "loss": 2.7141, "step": 34832 }, { "epoch": 2.162331615866907, "grad_norm": 0.46892962193166476, "learning_rate": 2.1928278803998908e-05, "loss": 2.684, "step": 34833 }, { "epoch": 2.1623936929666647, "grad_norm": 0.5343838986553772, "learning_rate": 2.1925290264295145e-05, "loss": 2.6311, "step": 34834 }, { "epoch": 2.1624557700664226, "grad_norm": 0.5087782260084893, "learning_rate": 2.192230187106396e-05, "loss": 2.6837, "step": 34835 }, { "epoch": 2.1625178471661806, "grad_norm": 0.4613931795743436, "learning_rate": 2.191931362432092e-05, "loss": 2.6638, "step": 34836 }, { "epoch": 2.1625799242659385, "grad_norm": 0.5384544592171923, "learning_rate": 2.191632552408166e-05, "loss": 2.6687, "step": 34837 }, { "epoch": 2.1626420013656964, "grad_norm": 0.5003262885705587, "learning_rate": 2.1913337570361748e-05, "loss": 2.6561, "step": 34838 }, { "epoch": 2.1627040784654543, "grad_norm": 0.49821412513269353, "learning_rate": 2.1910349763176763e-05, "loss": 2.7317, "step": 34839 }, { "epoch": 2.162766155565212, "grad_norm": 0.5282716287968308, "learning_rate": 2.1907362102542312e-05, "loss": 2.7165, "step": 34840 }, { "epoch": 2.1628282326649697, "grad_norm": 0.512060211638407, "learning_rate": 2.1904374588473947e-05, "loss": 2.576, "step": 34841 }, { "epoch": 2.1628903097647276, "grad_norm": 0.5626617569933304, "learning_rate": 2.1901387220987306e-05, "loss": 2.697, "step": 34842 }, { "epoch": 2.1629523868644855, "grad_norm": 0.5482099044530652, "learning_rate": 2.1898400000097948e-05, "loss": 2.7421, "step": 34843 }, { "epoch": 2.1630144639642435, "grad_norm": 0.461686253574001, "learning_rate": 2.189541292582145e-05, "loss": 2.6761, "step": 34844 }, { "epoch": 2.1630765410640014, "grad_norm": 0.5609024196270601, "learning_rate": 2.1892425998173392e-05, "loss": 2.6647, "step": 34845 }, { "epoch": 2.1631386181637593, "grad_norm": 0.5268976090400675, "learning_rate": 2.1889439217169392e-05, "loss": 2.7357, "step": 34846 }, { "epoch": 2.163200695263517, "grad_norm": 0.516685060845341, "learning_rate": 2.188645258282501e-05, "loss": 2.6934, "step": 34847 }, { "epoch": 2.163262772363275, "grad_norm": 0.5172458949933302, "learning_rate": 2.1883466095155836e-05, "loss": 2.6628, "step": 34848 }, { "epoch": 2.163324849463033, "grad_norm": 0.5504052522338702, "learning_rate": 2.188047975417742e-05, "loss": 2.6546, "step": 34849 }, { "epoch": 2.163386926562791, "grad_norm": 0.5035056954433913, "learning_rate": 2.1877493559905366e-05, "loss": 2.6618, "step": 34850 }, { "epoch": 2.163449003662549, "grad_norm": 0.5292904279815939, "learning_rate": 2.1874507512355276e-05, "loss": 2.7148, "step": 34851 }, { "epoch": 2.163511080762307, "grad_norm": 0.5205182330321237, "learning_rate": 2.1871521611542707e-05, "loss": 2.6824, "step": 34852 }, { "epoch": 2.1635731578620647, "grad_norm": 0.4847824321233967, "learning_rate": 2.186853585748324e-05, "loss": 2.6355, "step": 34853 }, { "epoch": 2.1636352349618226, "grad_norm": 0.5566144702856887, "learning_rate": 2.1865550250192444e-05, "loss": 2.6512, "step": 34854 }, { "epoch": 2.1636973120615806, "grad_norm": 0.5231205279433723, "learning_rate": 2.186256478968589e-05, "loss": 2.7005, "step": 34855 }, { "epoch": 2.1637593891613385, "grad_norm": 0.48142847492108704, "learning_rate": 2.1859579475979186e-05, "loss": 2.6898, "step": 34856 }, { "epoch": 2.1638214662610964, "grad_norm": 0.5490412179625124, "learning_rate": 2.1856594309087886e-05, "loss": 2.6749, "step": 34857 }, { "epoch": 2.1638835433608543, "grad_norm": 0.5036551862898224, "learning_rate": 2.1853609289027565e-05, "loss": 2.6669, "step": 34858 }, { "epoch": 2.1639456204606122, "grad_norm": 0.48584031914922565, "learning_rate": 2.1850624415813776e-05, "loss": 2.6422, "step": 34859 }, { "epoch": 2.16400769756037, "grad_norm": 0.5314833465166976, "learning_rate": 2.1847639689462136e-05, "loss": 2.6668, "step": 34860 }, { "epoch": 2.164069774660128, "grad_norm": 0.4960384938863418, "learning_rate": 2.184465510998819e-05, "loss": 2.7673, "step": 34861 }, { "epoch": 2.164131851759886, "grad_norm": 0.4948785571716993, "learning_rate": 2.184167067740751e-05, "loss": 2.6185, "step": 34862 }, { "epoch": 2.1641939288596435, "grad_norm": 0.5532092962310028, "learning_rate": 2.183868639173568e-05, "loss": 2.6951, "step": 34863 }, { "epoch": 2.1642560059594014, "grad_norm": 0.5140467389134196, "learning_rate": 2.1835702252988233e-05, "loss": 2.6662, "step": 34864 }, { "epoch": 2.1643180830591593, "grad_norm": 0.5967682118779133, "learning_rate": 2.1832718261180784e-05, "loss": 2.6292, "step": 34865 }, { "epoch": 2.164380160158917, "grad_norm": 0.4663243635616746, "learning_rate": 2.1829734416328886e-05, "loss": 2.6578, "step": 34866 }, { "epoch": 2.164442237258675, "grad_norm": 0.5259040111050036, "learning_rate": 2.1826750718448097e-05, "loss": 2.6872, "step": 34867 }, { "epoch": 2.164504314358433, "grad_norm": 0.5708472214236818, "learning_rate": 2.182376716755397e-05, "loss": 2.6925, "step": 34868 }, { "epoch": 2.164566391458191, "grad_norm": 0.48128365981062987, "learning_rate": 2.1820783763662113e-05, "loss": 2.7583, "step": 34869 }, { "epoch": 2.164628468557949, "grad_norm": 0.594519969216596, "learning_rate": 2.1817800506788066e-05, "loss": 2.6789, "step": 34870 }, { "epoch": 2.164690545657707, "grad_norm": 0.5231302982549678, "learning_rate": 2.1814817396947394e-05, "loss": 2.8197, "step": 34871 }, { "epoch": 2.1647526227574647, "grad_norm": 0.528540856096756, "learning_rate": 2.181183443415564e-05, "loss": 2.8006, "step": 34872 }, { "epoch": 2.1648146998572226, "grad_norm": 0.5387168550142123, "learning_rate": 2.1808851618428418e-05, "loss": 2.6714, "step": 34873 }, { "epoch": 2.1648767769569806, "grad_norm": 0.5633382115021192, "learning_rate": 2.180586894978125e-05, "loss": 2.6992, "step": 34874 }, { "epoch": 2.1649388540567385, "grad_norm": 0.5166318484336293, "learning_rate": 2.1802886428229713e-05, "loss": 2.6629, "step": 34875 }, { "epoch": 2.1650009311564964, "grad_norm": 0.5650274133783544, "learning_rate": 2.179990405378936e-05, "loss": 2.6889, "step": 34876 }, { "epoch": 2.1650630082562543, "grad_norm": 0.5111734819394261, "learning_rate": 2.1796921826475734e-05, "loss": 2.6873, "step": 34877 }, { "epoch": 2.1651250853560122, "grad_norm": 0.49443909146444925, "learning_rate": 2.1793939746304438e-05, "loss": 2.7543, "step": 34878 }, { "epoch": 2.16518716245577, "grad_norm": 0.5213043816646307, "learning_rate": 2.1790957813291e-05, "loss": 2.6834, "step": 34879 }, { "epoch": 2.165249239555528, "grad_norm": 0.5194970313254484, "learning_rate": 2.1787976027450986e-05, "loss": 2.6715, "step": 34880 }, { "epoch": 2.165311316655286, "grad_norm": 0.5075404685661635, "learning_rate": 2.178499438879993e-05, "loss": 2.6152, "step": 34881 }, { "epoch": 2.165373393755044, "grad_norm": 0.5177129030185889, "learning_rate": 2.1782012897353425e-05, "loss": 2.726, "step": 34882 }, { "epoch": 2.165435470854802, "grad_norm": 0.5729253305878251, "learning_rate": 2.1779031553126993e-05, "loss": 2.7206, "step": 34883 }, { "epoch": 2.1654975479545597, "grad_norm": 0.5710142860496064, "learning_rate": 2.1776050356136225e-05, "loss": 2.6738, "step": 34884 }, { "epoch": 2.1655596250543176, "grad_norm": 0.5237953613867493, "learning_rate": 2.1773069306396647e-05, "loss": 2.701, "step": 34885 }, { "epoch": 2.1656217021540756, "grad_norm": 0.5170161237028607, "learning_rate": 2.1770088403923806e-05, "loss": 2.6449, "step": 34886 }, { "epoch": 2.1656837792538335, "grad_norm": 0.5261033056675554, "learning_rate": 2.176710764873328e-05, "loss": 2.7053, "step": 34887 }, { "epoch": 2.165745856353591, "grad_norm": 0.472406116220867, "learning_rate": 2.176412704084061e-05, "loss": 2.6444, "step": 34888 }, { "epoch": 2.165807933453349, "grad_norm": 0.5363430868508214, "learning_rate": 2.1761146580261343e-05, "loss": 2.6994, "step": 34889 }, { "epoch": 2.165870010553107, "grad_norm": 0.49788432741155414, "learning_rate": 2.175816626701103e-05, "loss": 2.7416, "step": 34890 }, { "epoch": 2.1659320876528647, "grad_norm": 0.431330435618337, "learning_rate": 2.1755186101105207e-05, "loss": 2.6752, "step": 34891 }, { "epoch": 2.1659941647526226, "grad_norm": 0.5666844967065906, "learning_rate": 2.1752206082559445e-05, "loss": 2.7099, "step": 34892 }, { "epoch": 2.1660562418523805, "grad_norm": 0.4924667111392165, "learning_rate": 2.1749226211389286e-05, "loss": 2.6312, "step": 34893 }, { "epoch": 2.1661183189521385, "grad_norm": 0.5089008113393615, "learning_rate": 2.1746246487610265e-05, "loss": 2.7485, "step": 34894 }, { "epoch": 2.1661803960518964, "grad_norm": 0.5299073531039284, "learning_rate": 2.1743266911237925e-05, "loss": 2.6585, "step": 34895 }, { "epoch": 2.1662424731516543, "grad_norm": 0.4590806413705345, "learning_rate": 2.1740287482287835e-05, "loss": 2.6538, "step": 34896 }, { "epoch": 2.166304550251412, "grad_norm": 0.5901776562344174, "learning_rate": 2.1737308200775524e-05, "loss": 2.7583, "step": 34897 }, { "epoch": 2.16636662735117, "grad_norm": 0.4625343045838135, "learning_rate": 2.1734329066716542e-05, "loss": 2.6335, "step": 34898 }, { "epoch": 2.166428704450928, "grad_norm": 0.5319156683627025, "learning_rate": 2.1731350080126427e-05, "loss": 2.7179, "step": 34899 }, { "epoch": 2.166490781550686, "grad_norm": 0.49760288569228467, "learning_rate": 2.1728371241020702e-05, "loss": 2.5799, "step": 34900 }, { "epoch": 2.166552858650444, "grad_norm": 0.5405589355499426, "learning_rate": 2.172539254941494e-05, "loss": 2.7569, "step": 34901 }, { "epoch": 2.166614935750202, "grad_norm": 0.47895688322903324, "learning_rate": 2.1722414005324675e-05, "loss": 2.6793, "step": 34902 }, { "epoch": 2.1666770128499597, "grad_norm": 0.5068312922211559, "learning_rate": 2.171943560876544e-05, "loss": 2.7676, "step": 34903 }, { "epoch": 2.1667390899497176, "grad_norm": 0.5510616806039756, "learning_rate": 2.1716457359752752e-05, "loss": 2.7077, "step": 34904 }, { "epoch": 2.1668011670494756, "grad_norm": 0.48222020988250786, "learning_rate": 2.1713479258302193e-05, "loss": 2.7573, "step": 34905 }, { "epoch": 2.1668632441492335, "grad_norm": 0.5063201102228839, "learning_rate": 2.171050130442928e-05, "loss": 2.6634, "step": 34906 }, { "epoch": 2.1669253212489914, "grad_norm": 0.5199650552360329, "learning_rate": 2.170752349814955e-05, "loss": 2.6486, "step": 34907 }, { "epoch": 2.1669873983487493, "grad_norm": 0.5056745861810489, "learning_rate": 2.170454583947852e-05, "loss": 2.7005, "step": 34908 }, { "epoch": 2.1670494754485072, "grad_norm": 0.5321913961289185, "learning_rate": 2.1701568328431758e-05, "loss": 2.7263, "step": 34909 }, { "epoch": 2.167111552548265, "grad_norm": 0.5287241916805977, "learning_rate": 2.169859096502479e-05, "loss": 2.7444, "step": 34910 }, { "epoch": 2.1671736296480226, "grad_norm": 0.5111197831822191, "learning_rate": 2.1695613749273136e-05, "loss": 2.5841, "step": 34911 }, { "epoch": 2.1672357067477805, "grad_norm": 0.5200984984423337, "learning_rate": 2.169263668119234e-05, "loss": 2.6219, "step": 34912 }, { "epoch": 2.1672977838475385, "grad_norm": 0.5095520156610864, "learning_rate": 2.1689659760797913e-05, "loss": 2.6512, "step": 34913 }, { "epoch": 2.1673598609472964, "grad_norm": 0.5013460577631266, "learning_rate": 2.168668298810542e-05, "loss": 2.6126, "step": 34914 }, { "epoch": 2.1674219380470543, "grad_norm": 0.49804771273673415, "learning_rate": 2.168370636313038e-05, "loss": 2.6977, "step": 34915 }, { "epoch": 2.167484015146812, "grad_norm": 0.4883893770660284, "learning_rate": 2.1680729885888297e-05, "loss": 2.6667, "step": 34916 }, { "epoch": 2.16754609224657, "grad_norm": 0.5049108206028626, "learning_rate": 2.1677753556394736e-05, "loss": 2.654, "step": 34917 }, { "epoch": 2.167608169346328, "grad_norm": 0.4933173064284442, "learning_rate": 2.1674777374665195e-05, "loss": 2.6899, "step": 34918 }, { "epoch": 2.167670246446086, "grad_norm": 0.48182151287307196, "learning_rate": 2.167180134071523e-05, "loss": 2.6174, "step": 34919 }, { "epoch": 2.167732323545844, "grad_norm": 0.5048563376592675, "learning_rate": 2.166882545456036e-05, "loss": 2.7364, "step": 34920 }, { "epoch": 2.167794400645602, "grad_norm": 0.4873154510960566, "learning_rate": 2.16658497162161e-05, "loss": 2.7376, "step": 34921 }, { "epoch": 2.1678564777453597, "grad_norm": 0.5000154373646846, "learning_rate": 2.166287412569797e-05, "loss": 2.7058, "step": 34922 }, { "epoch": 2.1679185548451176, "grad_norm": 0.48211171917553297, "learning_rate": 2.1659898683021518e-05, "loss": 2.7368, "step": 34923 }, { "epoch": 2.1679806319448756, "grad_norm": 0.5026264646917222, "learning_rate": 2.1656923388202256e-05, "loss": 2.6788, "step": 34924 }, { "epoch": 2.1680427090446335, "grad_norm": 0.4291488565835717, "learning_rate": 2.16539482412557e-05, "loss": 2.6609, "step": 34925 }, { "epoch": 2.1681047861443914, "grad_norm": 0.49965456116250334, "learning_rate": 2.165097324219738e-05, "loss": 2.7473, "step": 34926 }, { "epoch": 2.1681668632441493, "grad_norm": 0.49313399723678364, "learning_rate": 2.16479983910428e-05, "loss": 2.6052, "step": 34927 }, { "epoch": 2.1682289403439072, "grad_norm": 0.5419482514040059, "learning_rate": 2.1645023687807507e-05, "loss": 2.6637, "step": 34928 }, { "epoch": 2.168291017443665, "grad_norm": 0.5215157774424899, "learning_rate": 2.1642049132507013e-05, "loss": 2.743, "step": 34929 }, { "epoch": 2.168353094543423, "grad_norm": 0.520763456107571, "learning_rate": 2.1639074725156832e-05, "loss": 2.631, "step": 34930 }, { "epoch": 2.168415171643181, "grad_norm": 0.5915815257435179, "learning_rate": 2.163610046577247e-05, "loss": 2.6574, "step": 34931 }, { "epoch": 2.168477248742939, "grad_norm": 0.5984848860263996, "learning_rate": 2.163312635436947e-05, "loss": 2.654, "step": 34932 }, { "epoch": 2.168539325842697, "grad_norm": 0.5091263720097964, "learning_rate": 2.1630152390963338e-05, "loss": 2.6504, "step": 34933 }, { "epoch": 2.1686014029424547, "grad_norm": 0.5425261195878297, "learning_rate": 2.1627178575569586e-05, "loss": 2.5649, "step": 34934 }, { "epoch": 2.1686634800422127, "grad_norm": 0.5063425356275855, "learning_rate": 2.1624204908203715e-05, "loss": 2.6187, "step": 34935 }, { "epoch": 2.16872555714197, "grad_norm": 0.5418720833986982, "learning_rate": 2.162123138888127e-05, "loss": 2.6948, "step": 34936 }, { "epoch": 2.168787634241728, "grad_norm": 0.5514891654045794, "learning_rate": 2.1618258017617755e-05, "loss": 2.6637, "step": 34937 }, { "epoch": 2.168849711341486, "grad_norm": 0.4860301349572897, "learning_rate": 2.1615284794428677e-05, "loss": 2.6705, "step": 34938 }, { "epoch": 2.168911788441244, "grad_norm": 0.5318255952690357, "learning_rate": 2.1612311719329544e-05, "loss": 2.6438, "step": 34939 }, { "epoch": 2.168973865541002, "grad_norm": 0.5145828234197687, "learning_rate": 2.160933879233586e-05, "loss": 2.7089, "step": 34940 }, { "epoch": 2.1690359426407597, "grad_norm": 0.5481219924705056, "learning_rate": 2.160636601346316e-05, "loss": 2.7597, "step": 34941 }, { "epoch": 2.1690980197405176, "grad_norm": 0.5083674699494376, "learning_rate": 2.1603393382726943e-05, "loss": 2.7057, "step": 34942 }, { "epoch": 2.1691600968402756, "grad_norm": 0.4629641768210004, "learning_rate": 2.160042090014272e-05, "loss": 2.6692, "step": 34943 }, { "epoch": 2.1692221739400335, "grad_norm": 0.567819424063707, "learning_rate": 2.1597448565725976e-05, "loss": 2.6755, "step": 34944 }, { "epoch": 2.1692842510397914, "grad_norm": 0.47000158431415157, "learning_rate": 2.159447637949225e-05, "loss": 2.6424, "step": 34945 }, { "epoch": 2.1693463281395493, "grad_norm": 0.5384122271149823, "learning_rate": 2.1591504341457042e-05, "loss": 2.7272, "step": 34946 }, { "epoch": 2.1694084052393072, "grad_norm": 0.46109321356354044, "learning_rate": 2.1588532451635853e-05, "loss": 2.7215, "step": 34947 }, { "epoch": 2.169470482339065, "grad_norm": 0.5339687007453732, "learning_rate": 2.1585560710044185e-05, "loss": 2.7958, "step": 34948 }, { "epoch": 2.169532559438823, "grad_norm": 0.5248656362073972, "learning_rate": 2.1582589116697527e-05, "loss": 2.6989, "step": 34949 }, { "epoch": 2.169594636538581, "grad_norm": 0.5014838017313175, "learning_rate": 2.1579617671611403e-05, "loss": 2.6994, "step": 34950 }, { "epoch": 2.169656713638339, "grad_norm": 0.5385888271477542, "learning_rate": 2.157664637480133e-05, "loss": 2.6234, "step": 34951 }, { "epoch": 2.169718790738097, "grad_norm": 0.47355350734957286, "learning_rate": 2.1573675226282787e-05, "loss": 2.6309, "step": 34952 }, { "epoch": 2.1697808678378547, "grad_norm": 0.49828336000519097, "learning_rate": 2.1570704226071285e-05, "loss": 2.6703, "step": 34953 }, { "epoch": 2.1698429449376126, "grad_norm": 0.560039223932154, "learning_rate": 2.1567733374182304e-05, "loss": 2.6917, "step": 34954 }, { "epoch": 2.1699050220373706, "grad_norm": 0.5358982094022431, "learning_rate": 2.1564762670631377e-05, "loss": 2.6844, "step": 34955 }, { "epoch": 2.1699670991371285, "grad_norm": 0.4959259951049648, "learning_rate": 2.1561792115433988e-05, "loss": 2.7069, "step": 34956 }, { "epoch": 2.1700291762368864, "grad_norm": 0.4902370042464432, "learning_rate": 2.155882170860563e-05, "loss": 2.6317, "step": 34957 }, { "epoch": 2.1700912533366443, "grad_norm": 0.5102651202416061, "learning_rate": 2.1555851450161786e-05, "loss": 2.6538, "step": 34958 }, { "epoch": 2.170153330436402, "grad_norm": 0.48038533114625886, "learning_rate": 2.1552881340117985e-05, "loss": 2.6888, "step": 34959 }, { "epoch": 2.1702154075361597, "grad_norm": 0.5056924139379783, "learning_rate": 2.154991137848972e-05, "loss": 2.5789, "step": 34960 }, { "epoch": 2.1702774846359176, "grad_norm": 0.5232212417015535, "learning_rate": 2.154694156529246e-05, "loss": 2.7195, "step": 34961 }, { "epoch": 2.1703395617356755, "grad_norm": 0.4498046371453307, "learning_rate": 2.154397190054172e-05, "loss": 2.6834, "step": 34962 }, { "epoch": 2.1704016388354335, "grad_norm": 0.4878121067756666, "learning_rate": 2.1541002384252968e-05, "loss": 2.6891, "step": 34963 }, { "epoch": 2.1704637159351914, "grad_norm": 0.5115978593205128, "learning_rate": 2.153803301644173e-05, "loss": 2.7747, "step": 34964 }, { "epoch": 2.1705257930349493, "grad_norm": 0.5185784803201533, "learning_rate": 2.1535063797123484e-05, "loss": 2.6435, "step": 34965 }, { "epoch": 2.170587870134707, "grad_norm": 0.501013384501285, "learning_rate": 2.1532094726313724e-05, "loss": 2.7666, "step": 34966 }, { "epoch": 2.170649947234465, "grad_norm": 0.4873467161596501, "learning_rate": 2.1529125804027918e-05, "loss": 2.7078, "step": 34967 }, { "epoch": 2.170712024334223, "grad_norm": 0.544971039991272, "learning_rate": 2.1526157030281587e-05, "loss": 2.7053, "step": 34968 }, { "epoch": 2.170774101433981, "grad_norm": 0.5212362320543064, "learning_rate": 2.1523188405090212e-05, "loss": 2.7732, "step": 34969 }, { "epoch": 2.170836178533739, "grad_norm": 0.4687654651633817, "learning_rate": 2.1520219928469276e-05, "loss": 2.6928, "step": 34970 }, { "epoch": 2.170898255633497, "grad_norm": 0.5422837031870671, "learning_rate": 2.1517251600434247e-05, "loss": 2.6037, "step": 34971 }, { "epoch": 2.1709603327332547, "grad_norm": 0.5507745133657583, "learning_rate": 2.151428342100065e-05, "loss": 2.6954, "step": 34972 }, { "epoch": 2.1710224098330126, "grad_norm": 0.4994691922010961, "learning_rate": 2.1511315390183945e-05, "loss": 2.7672, "step": 34973 }, { "epoch": 2.1710844869327706, "grad_norm": 0.4900726416323789, "learning_rate": 2.1508347507999632e-05, "loss": 2.6266, "step": 34974 }, { "epoch": 2.1711465640325285, "grad_norm": 0.5619034252678103, "learning_rate": 2.150537977446318e-05, "loss": 2.8357, "step": 34975 }, { "epoch": 2.1712086411322864, "grad_norm": 0.48429262322154665, "learning_rate": 2.1502412189590065e-05, "loss": 2.6826, "step": 34976 }, { "epoch": 2.1712707182320443, "grad_norm": 0.5323650774813659, "learning_rate": 2.1499444753395803e-05, "loss": 2.7419, "step": 34977 }, { "epoch": 2.1713327953318022, "grad_norm": 0.5396704631649889, "learning_rate": 2.1496477465895853e-05, "loss": 2.6323, "step": 34978 }, { "epoch": 2.17139487243156, "grad_norm": 0.5535445190296855, "learning_rate": 2.1493510327105704e-05, "loss": 2.7635, "step": 34979 }, { "epoch": 2.171456949531318, "grad_norm": 0.5237273652004149, "learning_rate": 2.1490543337040808e-05, "loss": 2.6751, "step": 34980 }, { "epoch": 2.171519026631076, "grad_norm": 0.46027413818906915, "learning_rate": 2.1487576495716673e-05, "loss": 2.7073, "step": 34981 }, { "epoch": 2.1715811037308335, "grad_norm": 0.5036967529571642, "learning_rate": 2.1484609803148786e-05, "loss": 2.6843, "step": 34982 }, { "epoch": 2.171643180830592, "grad_norm": 0.5028388376492084, "learning_rate": 2.1481643259352617e-05, "loss": 2.6902, "step": 34983 }, { "epoch": 2.1717052579303493, "grad_norm": 0.5148749821036147, "learning_rate": 2.147867686434364e-05, "loss": 2.6997, "step": 34984 }, { "epoch": 2.171767335030107, "grad_norm": 0.5429165492590745, "learning_rate": 2.147571061813731e-05, "loss": 2.6975, "step": 34985 }, { "epoch": 2.171829412129865, "grad_norm": 0.5284382235512498, "learning_rate": 2.1472744520749133e-05, "loss": 2.7066, "step": 34986 }, { "epoch": 2.171891489229623, "grad_norm": 0.48573920199554654, "learning_rate": 2.1469778572194577e-05, "loss": 2.672, "step": 34987 }, { "epoch": 2.171953566329381, "grad_norm": 0.521604945499577, "learning_rate": 2.1466812772489113e-05, "loss": 2.6791, "step": 34988 }, { "epoch": 2.172015643429139, "grad_norm": 0.4935572501487005, "learning_rate": 2.1463847121648217e-05, "loss": 2.8211, "step": 34989 }, { "epoch": 2.172077720528897, "grad_norm": 0.507291203471866, "learning_rate": 2.1460881619687333e-05, "loss": 2.6741, "step": 34990 }, { "epoch": 2.1721397976286547, "grad_norm": 0.5388463734239523, "learning_rate": 2.145791626662198e-05, "loss": 2.6316, "step": 34991 }, { "epoch": 2.1722018747284126, "grad_norm": 0.4521054274203921, "learning_rate": 2.145495106246761e-05, "loss": 2.6754, "step": 34992 }, { "epoch": 2.1722639518281706, "grad_norm": 0.6110662319822026, "learning_rate": 2.1451986007239684e-05, "loss": 2.6366, "step": 34993 }, { "epoch": 2.1723260289279285, "grad_norm": 0.4880166977415727, "learning_rate": 2.1449021100953664e-05, "loss": 2.7281, "step": 34994 }, { "epoch": 2.1723881060276864, "grad_norm": 0.5515224798883029, "learning_rate": 2.1446056343625042e-05, "loss": 2.6424, "step": 34995 }, { "epoch": 2.1724501831274443, "grad_norm": 0.5183102018810675, "learning_rate": 2.1443091735269284e-05, "loss": 2.7235, "step": 34996 }, { "epoch": 2.1725122602272022, "grad_norm": 0.5344554182521165, "learning_rate": 2.1440127275901845e-05, "loss": 2.6221, "step": 34997 }, { "epoch": 2.17257433732696, "grad_norm": 0.5590469815790853, "learning_rate": 2.1437162965538198e-05, "loss": 2.6797, "step": 34998 }, { "epoch": 2.172636414426718, "grad_norm": 0.4899959820465865, "learning_rate": 2.1434198804193785e-05, "loss": 2.7253, "step": 34999 }, { "epoch": 2.172698491526476, "grad_norm": 0.4757937186976636, "learning_rate": 2.1431234791884108e-05, "loss": 2.6603, "step": 35000 }, { "epoch": 2.172760568626234, "grad_norm": 0.5905510659339748, "learning_rate": 2.1428270928624616e-05, "loss": 2.627, "step": 35001 }, { "epoch": 2.172822645725992, "grad_norm": 0.4441026515378609, "learning_rate": 2.1425307214430774e-05, "loss": 2.6116, "step": 35002 }, { "epoch": 2.1728847228257497, "grad_norm": 0.525112858630783, "learning_rate": 2.1422343649318023e-05, "loss": 2.6112, "step": 35003 }, { "epoch": 2.1729467999255077, "grad_norm": 0.5412651891096825, "learning_rate": 2.1419380233301855e-05, "loss": 2.7645, "step": 35004 }, { "epoch": 2.1730088770252656, "grad_norm": 0.5217083796327049, "learning_rate": 2.1416416966397722e-05, "loss": 2.6775, "step": 35005 }, { "epoch": 2.1730709541250235, "grad_norm": 0.5382370918221526, "learning_rate": 2.141345384862108e-05, "loss": 2.6466, "step": 35006 }, { "epoch": 2.173133031224781, "grad_norm": 0.5470928030896491, "learning_rate": 2.141049087998737e-05, "loss": 2.7072, "step": 35007 }, { "epoch": 2.173195108324539, "grad_norm": 0.5770336882609199, "learning_rate": 2.140752806051209e-05, "loss": 2.7074, "step": 35008 }, { "epoch": 2.173257185424297, "grad_norm": 0.5284260416621508, "learning_rate": 2.1404565390210673e-05, "loss": 2.6796, "step": 35009 }, { "epoch": 2.1733192625240547, "grad_norm": 0.5047128121845922, "learning_rate": 2.1401602869098587e-05, "loss": 2.7385, "step": 35010 }, { "epoch": 2.1733813396238126, "grad_norm": 0.531539557588121, "learning_rate": 2.1398640497191276e-05, "loss": 2.703, "step": 35011 }, { "epoch": 2.1734434167235706, "grad_norm": 0.47685725055324973, "learning_rate": 2.1395678274504183e-05, "loss": 2.6589, "step": 35012 }, { "epoch": 2.1735054938233285, "grad_norm": 0.5361394273804981, "learning_rate": 2.1392716201052797e-05, "loss": 2.7479, "step": 35013 }, { "epoch": 2.1735675709230864, "grad_norm": 0.5031389116058029, "learning_rate": 2.1389754276852536e-05, "loss": 2.6649, "step": 35014 }, { "epoch": 2.1736296480228443, "grad_norm": 0.4942882905241478, "learning_rate": 2.1386792501918895e-05, "loss": 2.6833, "step": 35015 }, { "epoch": 2.1736917251226022, "grad_norm": 0.5004521054475424, "learning_rate": 2.1383830876267302e-05, "loss": 2.6755, "step": 35016 }, { "epoch": 2.17375380222236, "grad_norm": 0.5321374661125908, "learning_rate": 2.1380869399913195e-05, "loss": 2.6672, "step": 35017 }, { "epoch": 2.173815879322118, "grad_norm": 0.4883934807543037, "learning_rate": 2.1377908072872056e-05, "loss": 2.6699, "step": 35018 }, { "epoch": 2.173877956421876, "grad_norm": 0.5041896703164329, "learning_rate": 2.1374946895159313e-05, "loss": 2.7653, "step": 35019 }, { "epoch": 2.173940033521634, "grad_norm": 0.5061835056952291, "learning_rate": 2.1371985866790428e-05, "loss": 2.5881, "step": 35020 }, { "epoch": 2.174002110621392, "grad_norm": 0.5111374210183434, "learning_rate": 2.136902498778082e-05, "loss": 2.7368, "step": 35021 }, { "epoch": 2.1740641877211497, "grad_norm": 0.5342890542053803, "learning_rate": 2.1366064258145986e-05, "loss": 2.7198, "step": 35022 }, { "epoch": 2.1741262648209077, "grad_norm": 0.503847192659892, "learning_rate": 2.1363103677901336e-05, "loss": 2.6337, "step": 35023 }, { "epoch": 2.1741883419206656, "grad_norm": 0.4898366948886683, "learning_rate": 2.136014324706233e-05, "loss": 2.6535, "step": 35024 }, { "epoch": 2.1742504190204235, "grad_norm": 0.49667218045506617, "learning_rate": 2.1357182965644408e-05, "loss": 2.6198, "step": 35025 }, { "epoch": 2.1743124961201814, "grad_norm": 0.47519967374670025, "learning_rate": 2.1354222833663003e-05, "loss": 2.6312, "step": 35026 }, { "epoch": 2.1743745732199393, "grad_norm": 0.5139129131160062, "learning_rate": 2.135126285113358e-05, "loss": 2.6493, "step": 35027 }, { "epoch": 2.1744366503196972, "grad_norm": 0.48841842250457385, "learning_rate": 2.1348303018071582e-05, "loss": 2.7798, "step": 35028 }, { "epoch": 2.174498727419455, "grad_norm": 0.49530301190262854, "learning_rate": 2.1345343334492435e-05, "loss": 2.7381, "step": 35029 }, { "epoch": 2.1745608045192126, "grad_norm": 0.4898335007077442, "learning_rate": 2.1342383800411574e-05, "loss": 2.6707, "step": 35030 }, { "epoch": 2.174622881618971, "grad_norm": 0.5092695314241548, "learning_rate": 2.1339424415844473e-05, "loss": 2.6769, "step": 35031 }, { "epoch": 2.1746849587187285, "grad_norm": 0.4971208527957697, "learning_rate": 2.1336465180806554e-05, "loss": 2.7638, "step": 35032 }, { "epoch": 2.1747470358184864, "grad_norm": 0.5059493290517961, "learning_rate": 2.1333506095313248e-05, "loss": 2.6771, "step": 35033 }, { "epoch": 2.1748091129182443, "grad_norm": 0.46811918092072535, "learning_rate": 2.1330547159380003e-05, "loss": 2.707, "step": 35034 }, { "epoch": 2.174871190018002, "grad_norm": 0.493106151239921, "learning_rate": 2.1327588373022234e-05, "loss": 2.6096, "step": 35035 }, { "epoch": 2.17493326711776, "grad_norm": 0.4894960778851746, "learning_rate": 2.132462973625542e-05, "loss": 2.7172, "step": 35036 }, { "epoch": 2.174995344217518, "grad_norm": 0.5566173883239348, "learning_rate": 2.1321671249094965e-05, "loss": 2.6397, "step": 35037 }, { "epoch": 2.175057421317276, "grad_norm": 0.48140162333562236, "learning_rate": 2.131871291155632e-05, "loss": 2.551, "step": 35038 }, { "epoch": 2.175119498417034, "grad_norm": 0.5212086437935067, "learning_rate": 2.131575472365489e-05, "loss": 2.6854, "step": 35039 }, { "epoch": 2.175181575516792, "grad_norm": 0.5343666098309662, "learning_rate": 2.1312796685406155e-05, "loss": 2.7486, "step": 35040 }, { "epoch": 2.1752436526165497, "grad_norm": 0.48067652853960163, "learning_rate": 2.130983879682552e-05, "loss": 2.6436, "step": 35041 }, { "epoch": 2.1753057297163076, "grad_norm": 0.5473383167028754, "learning_rate": 2.130688105792842e-05, "loss": 2.6097, "step": 35042 }, { "epoch": 2.1753678068160656, "grad_norm": 0.5031089093022436, "learning_rate": 2.130392346873027e-05, "loss": 2.7371, "step": 35043 }, { "epoch": 2.1754298839158235, "grad_norm": 0.5219163284383903, "learning_rate": 2.130096602924654e-05, "loss": 2.7098, "step": 35044 }, { "epoch": 2.1754919610155814, "grad_norm": 0.46679365085142754, "learning_rate": 2.129800873949263e-05, "loss": 2.6364, "step": 35045 }, { "epoch": 2.1755540381153393, "grad_norm": 0.508646968981165, "learning_rate": 2.129505159948398e-05, "loss": 2.6797, "step": 35046 }, { "epoch": 2.1756161152150972, "grad_norm": 0.5403931269326502, "learning_rate": 2.1292094609235992e-05, "loss": 2.6862, "step": 35047 }, { "epoch": 2.175678192314855, "grad_norm": 0.5324432212037704, "learning_rate": 2.128913776876414e-05, "loss": 2.7447, "step": 35048 }, { "epoch": 2.175740269414613, "grad_norm": 0.5302982219151903, "learning_rate": 2.12861810780838e-05, "loss": 2.7057, "step": 35049 }, { "epoch": 2.175802346514371, "grad_norm": 0.5296596395991893, "learning_rate": 2.128322453721045e-05, "loss": 2.6402, "step": 35050 }, { "epoch": 2.175864423614129, "grad_norm": 0.5307240406970555, "learning_rate": 2.128026814615948e-05, "loss": 2.713, "step": 35051 }, { "epoch": 2.175926500713887, "grad_norm": 0.4875853347673132, "learning_rate": 2.127731190494633e-05, "loss": 2.7235, "step": 35052 }, { "epoch": 2.1759885778136447, "grad_norm": 0.5397316745953004, "learning_rate": 2.12743558135864e-05, "loss": 2.6635, "step": 35053 }, { "epoch": 2.1760506549134027, "grad_norm": 0.4591395014511727, "learning_rate": 2.1271399872095143e-05, "loss": 2.7749, "step": 35054 }, { "epoch": 2.17611273201316, "grad_norm": 0.47923281116335537, "learning_rate": 2.1268444080487966e-05, "loss": 2.6589, "step": 35055 }, { "epoch": 2.176174809112918, "grad_norm": 0.5239897419620663, "learning_rate": 2.126548843878029e-05, "loss": 2.6418, "step": 35056 }, { "epoch": 2.176236886212676, "grad_norm": 0.44065147617958095, "learning_rate": 2.1262532946987524e-05, "loss": 2.5045, "step": 35057 }, { "epoch": 2.176298963312434, "grad_norm": 0.4856584680039124, "learning_rate": 2.1259577605125114e-05, "loss": 2.6594, "step": 35058 }, { "epoch": 2.176361040412192, "grad_norm": 0.44406414732335553, "learning_rate": 2.1256622413208465e-05, "loss": 2.715, "step": 35059 }, { "epoch": 2.1764231175119497, "grad_norm": 0.5006362324191548, "learning_rate": 2.1253667371252995e-05, "loss": 2.6284, "step": 35060 }, { "epoch": 2.1764851946117076, "grad_norm": 0.48453408366841355, "learning_rate": 2.125071247927412e-05, "loss": 2.6621, "step": 35061 }, { "epoch": 2.1765472717114656, "grad_norm": 0.5337964344597339, "learning_rate": 2.1247757737287245e-05, "loss": 2.6214, "step": 35062 }, { "epoch": 2.1766093488112235, "grad_norm": 0.4840751612062072, "learning_rate": 2.1244803145307806e-05, "loss": 2.7779, "step": 35063 }, { "epoch": 2.1766714259109814, "grad_norm": 0.5176837453497609, "learning_rate": 2.1241848703351214e-05, "loss": 2.6537, "step": 35064 }, { "epoch": 2.1767335030107393, "grad_norm": 0.5438284894045746, "learning_rate": 2.123889441143288e-05, "loss": 2.6711, "step": 35065 }, { "epoch": 2.1767955801104972, "grad_norm": 0.48178249723044503, "learning_rate": 2.12359402695682e-05, "loss": 2.6423, "step": 35066 }, { "epoch": 2.176857657210255, "grad_norm": 0.535748731550245, "learning_rate": 2.1232986277772617e-05, "loss": 2.6769, "step": 35067 }, { "epoch": 2.176919734310013, "grad_norm": 0.5325892061633842, "learning_rate": 2.1230032436061525e-05, "loss": 2.7182, "step": 35068 }, { "epoch": 2.176981811409771, "grad_norm": 0.5600025045815726, "learning_rate": 2.122707874445034e-05, "loss": 2.6631, "step": 35069 }, { "epoch": 2.177043888509529, "grad_norm": 0.48801257561774697, "learning_rate": 2.1224125202954455e-05, "loss": 2.7292, "step": 35070 }, { "epoch": 2.177105965609287, "grad_norm": 0.5171733496997685, "learning_rate": 2.1221171811589314e-05, "loss": 2.6233, "step": 35071 }, { "epoch": 2.1771680427090447, "grad_norm": 0.540027390641909, "learning_rate": 2.12182185703703e-05, "loss": 2.7578, "step": 35072 }, { "epoch": 2.1772301198088027, "grad_norm": 0.5495382039202567, "learning_rate": 2.121526547931283e-05, "loss": 2.6887, "step": 35073 }, { "epoch": 2.1772921969085606, "grad_norm": 0.5760359264269425, "learning_rate": 2.1212312538432305e-05, "loss": 2.7597, "step": 35074 }, { "epoch": 2.1773542740083185, "grad_norm": 0.5301861770237438, "learning_rate": 2.1209359747744116e-05, "loss": 2.72, "step": 35075 }, { "epoch": 2.1774163511080764, "grad_norm": 0.5500997832121888, "learning_rate": 2.120640710726371e-05, "loss": 2.6976, "step": 35076 }, { "epoch": 2.1774784282078343, "grad_norm": 0.5340508788637406, "learning_rate": 2.120345461700646e-05, "loss": 2.6422, "step": 35077 }, { "epoch": 2.177540505307592, "grad_norm": 0.5015349371636282, "learning_rate": 2.1200502276987783e-05, "loss": 2.666, "step": 35078 }, { "epoch": 2.17760258240735, "grad_norm": 0.5338270568167038, "learning_rate": 2.119755008722306e-05, "loss": 2.7014, "step": 35079 }, { "epoch": 2.1776646595071076, "grad_norm": 0.5185483080800212, "learning_rate": 2.119459804772771e-05, "loss": 2.7155, "step": 35080 }, { "epoch": 2.1777267366068656, "grad_norm": 0.5232017042924817, "learning_rate": 2.119164615851715e-05, "loss": 2.6926, "step": 35081 }, { "epoch": 2.1777888137066235, "grad_norm": 0.5286782510022607, "learning_rate": 2.1188694419606762e-05, "loss": 2.6354, "step": 35082 }, { "epoch": 2.1778508908063814, "grad_norm": 0.5466510041491697, "learning_rate": 2.1185742831011957e-05, "loss": 2.6918, "step": 35083 }, { "epoch": 2.1779129679061393, "grad_norm": 0.5366380878010315, "learning_rate": 2.1182791392748126e-05, "loss": 2.7123, "step": 35084 }, { "epoch": 2.1779750450058972, "grad_norm": 0.5871641162725248, "learning_rate": 2.1179840104830644e-05, "loss": 2.6963, "step": 35085 }, { "epoch": 2.178037122105655, "grad_norm": 0.539388585632311, "learning_rate": 2.1176888967274954e-05, "loss": 2.6099, "step": 35086 }, { "epoch": 2.178099199205413, "grad_norm": 0.5931070510753207, "learning_rate": 2.117393798009643e-05, "loss": 2.7467, "step": 35087 }, { "epoch": 2.178161276305171, "grad_norm": 0.5752860808846492, "learning_rate": 2.1170987143310472e-05, "loss": 2.6136, "step": 35088 }, { "epoch": 2.178223353404929, "grad_norm": 0.5537624974272127, "learning_rate": 2.1168036456932455e-05, "loss": 2.7074, "step": 35089 }, { "epoch": 2.178285430504687, "grad_norm": 0.5516179270356787, "learning_rate": 2.1165085920977806e-05, "loss": 2.7743, "step": 35090 }, { "epoch": 2.1783475076044447, "grad_norm": 0.5247537375939937, "learning_rate": 2.1162135535461907e-05, "loss": 2.7029, "step": 35091 }, { "epoch": 2.1784095847042027, "grad_norm": 0.5558533036201334, "learning_rate": 2.115918530040014e-05, "loss": 2.7339, "step": 35092 }, { "epoch": 2.1784716618039606, "grad_norm": 0.5301471370112792, "learning_rate": 2.1156235215807896e-05, "loss": 2.7446, "step": 35093 }, { "epoch": 2.1785337389037185, "grad_norm": 0.5275248746295192, "learning_rate": 2.1153285281700592e-05, "loss": 2.668, "step": 35094 }, { "epoch": 2.1785958160034764, "grad_norm": 0.5773839324884564, "learning_rate": 2.1150335498093593e-05, "loss": 2.7231, "step": 35095 }, { "epoch": 2.1786578931032343, "grad_norm": 0.5140079225827596, "learning_rate": 2.1147385865002307e-05, "loss": 2.7719, "step": 35096 }, { "epoch": 2.1787199702029922, "grad_norm": 0.5525945479849924, "learning_rate": 2.1144436382442107e-05, "loss": 2.679, "step": 35097 }, { "epoch": 2.17878204730275, "grad_norm": 0.4843108380799055, "learning_rate": 2.1141487050428373e-05, "loss": 2.6388, "step": 35098 }, { "epoch": 2.178844124402508, "grad_norm": 0.5775103523515502, "learning_rate": 2.113853786897652e-05, "loss": 2.72, "step": 35099 }, { "epoch": 2.178906201502266, "grad_norm": 0.49684413718094456, "learning_rate": 2.113558883810193e-05, "loss": 2.6934, "step": 35100 }, { "epoch": 2.178968278602024, "grad_norm": 0.5477436251737277, "learning_rate": 2.1132639957819973e-05, "loss": 2.6935, "step": 35101 }, { "epoch": 2.179030355701782, "grad_norm": 0.5258633497057058, "learning_rate": 2.112969122814602e-05, "loss": 2.7398, "step": 35102 }, { "epoch": 2.1790924328015393, "grad_norm": 0.5139993414299453, "learning_rate": 2.11267426490955e-05, "loss": 2.7454, "step": 35103 }, { "epoch": 2.179154509901297, "grad_norm": 0.5617858273668274, "learning_rate": 2.1123794220683768e-05, "loss": 2.5946, "step": 35104 }, { "epoch": 2.179216587001055, "grad_norm": 0.49707365503292755, "learning_rate": 2.1120845942926214e-05, "loss": 2.7134, "step": 35105 }, { "epoch": 2.179278664100813, "grad_norm": 0.5132582707247358, "learning_rate": 2.1117897815838206e-05, "loss": 2.6376, "step": 35106 }, { "epoch": 2.179340741200571, "grad_norm": 0.5172462406971363, "learning_rate": 2.111494983943515e-05, "loss": 2.5723, "step": 35107 }, { "epoch": 2.179402818300329, "grad_norm": 0.4849645503105822, "learning_rate": 2.1112002013732406e-05, "loss": 2.5991, "step": 35108 }, { "epoch": 2.179464895400087, "grad_norm": 0.5319567355099106, "learning_rate": 2.110905433874537e-05, "loss": 2.6783, "step": 35109 }, { "epoch": 2.1795269724998447, "grad_norm": 0.5012134787461119, "learning_rate": 2.1106106814489407e-05, "loss": 2.7514, "step": 35110 }, { "epoch": 2.1795890495996026, "grad_norm": 0.5085767423771588, "learning_rate": 2.110315944097988e-05, "loss": 2.6888, "step": 35111 }, { "epoch": 2.1796511266993606, "grad_norm": 0.4905189812669065, "learning_rate": 2.1100212218232185e-05, "loss": 2.7027, "step": 35112 }, { "epoch": 2.1797132037991185, "grad_norm": 0.5156083248794402, "learning_rate": 2.109726514626172e-05, "loss": 2.6922, "step": 35113 }, { "epoch": 2.1797752808988764, "grad_norm": 0.551822840816774, "learning_rate": 2.1094318225083835e-05, "loss": 2.7112, "step": 35114 }, { "epoch": 2.1798373579986343, "grad_norm": 0.5082484501441183, "learning_rate": 2.109137145471391e-05, "loss": 2.71, "step": 35115 }, { "epoch": 2.1798994350983922, "grad_norm": 0.526461789152907, "learning_rate": 2.10884248351673e-05, "loss": 2.6285, "step": 35116 }, { "epoch": 2.17996151219815, "grad_norm": 0.5107617347142256, "learning_rate": 2.1085478366459416e-05, "loss": 2.6371, "step": 35117 }, { "epoch": 2.180023589297908, "grad_norm": 0.501560933713755, "learning_rate": 2.1082532048605608e-05, "loss": 2.6366, "step": 35118 }, { "epoch": 2.180085666397666, "grad_norm": 0.5080855589101453, "learning_rate": 2.1079585881621256e-05, "loss": 2.6748, "step": 35119 }, { "epoch": 2.180147743497424, "grad_norm": 0.5359330769348022, "learning_rate": 2.107663986552172e-05, "loss": 2.6304, "step": 35120 }, { "epoch": 2.180209820597182, "grad_norm": 0.5882208267529505, "learning_rate": 2.107369400032236e-05, "loss": 2.7594, "step": 35121 }, { "epoch": 2.1802718976969397, "grad_norm": 0.5349158700856933, "learning_rate": 2.107074828603857e-05, "loss": 2.8254, "step": 35122 }, { "epoch": 2.1803339747966977, "grad_norm": 0.49263882002166665, "learning_rate": 2.106780272268572e-05, "loss": 2.6617, "step": 35123 }, { "epoch": 2.1803960518964556, "grad_norm": 0.573071534003943, "learning_rate": 2.106485731027916e-05, "loss": 2.6432, "step": 35124 }, { "epoch": 2.1804581289962135, "grad_norm": 0.4974080645137533, "learning_rate": 2.1061912048834247e-05, "loss": 2.5987, "step": 35125 }, { "epoch": 2.180520206095971, "grad_norm": 0.5221070133642715, "learning_rate": 2.1058966938366385e-05, "loss": 2.6531, "step": 35126 }, { "epoch": 2.180582283195729, "grad_norm": 0.5003849276382847, "learning_rate": 2.1056021978890916e-05, "loss": 2.5864, "step": 35127 }, { "epoch": 2.180644360295487, "grad_norm": 0.4868673682441046, "learning_rate": 2.105307717042321e-05, "loss": 2.7029, "step": 35128 }, { "epoch": 2.1807064373952447, "grad_norm": 0.5098595660530404, "learning_rate": 2.1050132512978604e-05, "loss": 2.692, "step": 35129 }, { "epoch": 2.1807685144950026, "grad_norm": 0.5293148642998526, "learning_rate": 2.1047188006572506e-05, "loss": 2.665, "step": 35130 }, { "epoch": 2.1808305915947606, "grad_norm": 0.48374992461783045, "learning_rate": 2.1044243651220253e-05, "loss": 2.7412, "step": 35131 }, { "epoch": 2.1808926686945185, "grad_norm": 0.5235368150465263, "learning_rate": 2.1041299446937214e-05, "loss": 2.7144, "step": 35132 }, { "epoch": 2.1809547457942764, "grad_norm": 0.5323417049063007, "learning_rate": 2.103835539373874e-05, "loss": 2.6501, "step": 35133 }, { "epoch": 2.1810168228940343, "grad_norm": 0.4666172623737506, "learning_rate": 2.1035411491640185e-05, "loss": 2.7066, "step": 35134 }, { "epoch": 2.1810788999937922, "grad_norm": 0.5189486226761253, "learning_rate": 2.1032467740656934e-05, "loss": 2.6092, "step": 35135 }, { "epoch": 2.18114097709355, "grad_norm": 0.5741293441820727, "learning_rate": 2.1029524140804335e-05, "loss": 2.7306, "step": 35136 }, { "epoch": 2.181203054193308, "grad_norm": 0.4546684582318157, "learning_rate": 2.1026580692097734e-05, "loss": 2.7232, "step": 35137 }, { "epoch": 2.181265131293066, "grad_norm": 0.5665793318859889, "learning_rate": 2.1023637394552488e-05, "loss": 2.7664, "step": 35138 }, { "epoch": 2.181327208392824, "grad_norm": 0.5135979544824646, "learning_rate": 2.102069424818397e-05, "loss": 2.7156, "step": 35139 }, { "epoch": 2.181389285492582, "grad_norm": 0.46372384516846604, "learning_rate": 2.1017751253007528e-05, "loss": 2.6764, "step": 35140 }, { "epoch": 2.1814513625923397, "grad_norm": 0.5100198455296531, "learning_rate": 2.101480840903851e-05, "loss": 2.6369, "step": 35141 }, { "epoch": 2.1815134396920977, "grad_norm": 0.5073266566257445, "learning_rate": 2.1011865716292255e-05, "loss": 2.6375, "step": 35142 }, { "epoch": 2.1815755167918556, "grad_norm": 0.5503523218324443, "learning_rate": 2.1008923174784155e-05, "loss": 2.7549, "step": 35143 }, { "epoch": 2.1816375938916135, "grad_norm": 0.517382746833633, "learning_rate": 2.100598078452954e-05, "loss": 2.7233, "step": 35144 }, { "epoch": 2.1816996709913714, "grad_norm": 0.5321733920634817, "learning_rate": 2.1003038545543746e-05, "loss": 2.6717, "step": 35145 }, { "epoch": 2.1817617480911293, "grad_norm": 0.47085404252051416, "learning_rate": 2.100009645784215e-05, "loss": 2.6373, "step": 35146 }, { "epoch": 2.1818238251908872, "grad_norm": 0.5368713128759317, "learning_rate": 2.09971545214401e-05, "loss": 2.6727, "step": 35147 }, { "epoch": 2.181885902290645, "grad_norm": 0.5087711675209691, "learning_rate": 2.099421273635291e-05, "loss": 2.6372, "step": 35148 }, { "epoch": 2.181947979390403, "grad_norm": 0.4704986856413836, "learning_rate": 2.099127110259598e-05, "loss": 2.7761, "step": 35149 }, { "epoch": 2.182010056490161, "grad_norm": 0.5474681362477674, "learning_rate": 2.098832962018462e-05, "loss": 2.7005, "step": 35150 }, { "epoch": 2.1820721335899185, "grad_norm": 0.4872042565831619, "learning_rate": 2.0985388289134196e-05, "loss": 2.7569, "step": 35151 }, { "epoch": 2.1821342106896764, "grad_norm": 0.5058426725362312, "learning_rate": 2.0982447109460017e-05, "loss": 2.6719, "step": 35152 }, { "epoch": 2.1821962877894343, "grad_norm": 0.4862689093269293, "learning_rate": 2.097950608117748e-05, "loss": 2.5972, "step": 35153 }, { "epoch": 2.1822583648891922, "grad_norm": 0.4972166734021163, "learning_rate": 2.0976565204301907e-05, "loss": 2.6546, "step": 35154 }, { "epoch": 2.18232044198895, "grad_norm": 0.49712648805007587, "learning_rate": 2.0973624478848634e-05, "loss": 2.6307, "step": 35155 }, { "epoch": 2.182382519088708, "grad_norm": 0.5567959485100501, "learning_rate": 2.0970683904832994e-05, "loss": 2.695, "step": 35156 }, { "epoch": 2.182444596188466, "grad_norm": 0.5241103900280587, "learning_rate": 2.0967743482270362e-05, "loss": 2.7437, "step": 35157 }, { "epoch": 2.182506673288224, "grad_norm": 0.5424223421477469, "learning_rate": 2.096480321117606e-05, "loss": 2.6858, "step": 35158 }, { "epoch": 2.182568750387982, "grad_norm": 0.493148652393055, "learning_rate": 2.0961863091565426e-05, "loss": 2.6458, "step": 35159 }, { "epoch": 2.1826308274877397, "grad_norm": 0.5541422407125907, "learning_rate": 2.0958923123453798e-05, "loss": 2.7075, "step": 35160 }, { "epoch": 2.1826929045874977, "grad_norm": 0.5051422043088662, "learning_rate": 2.0955983306856507e-05, "loss": 2.6899, "step": 35161 }, { "epoch": 2.1827549816872556, "grad_norm": 0.5643311206999948, "learning_rate": 2.0953043641788917e-05, "loss": 2.6891, "step": 35162 }, { "epoch": 2.1828170587870135, "grad_norm": 0.4993155831759746, "learning_rate": 2.095010412826635e-05, "loss": 2.6828, "step": 35163 }, { "epoch": 2.1828791358867714, "grad_norm": 0.5656999327621298, "learning_rate": 2.0947164766304138e-05, "loss": 2.7225, "step": 35164 }, { "epoch": 2.1829412129865293, "grad_norm": 0.4961950544115451, "learning_rate": 2.0944225555917607e-05, "loss": 2.6213, "step": 35165 }, { "epoch": 2.1830032900862872, "grad_norm": 0.5298067290664665, "learning_rate": 2.094128649712212e-05, "loss": 2.7428, "step": 35166 }, { "epoch": 2.183065367186045, "grad_norm": 0.5478898849275606, "learning_rate": 2.0938347589933e-05, "loss": 2.708, "step": 35167 }, { "epoch": 2.183127444285803, "grad_norm": 0.5087530668855121, "learning_rate": 2.093540883436557e-05, "loss": 2.5945, "step": 35168 }, { "epoch": 2.183189521385561, "grad_norm": 0.5737431171727444, "learning_rate": 2.0932470230435176e-05, "loss": 2.7352, "step": 35169 }, { "epoch": 2.183251598485319, "grad_norm": 0.5346226781884313, "learning_rate": 2.0929531778157112e-05, "loss": 2.7397, "step": 35170 }, { "epoch": 2.183313675585077, "grad_norm": 0.48985859728249487, "learning_rate": 2.0926593477546762e-05, "loss": 2.6812, "step": 35171 }, { "epoch": 2.1833757526848347, "grad_norm": 0.5546499964970673, "learning_rate": 2.0923655328619433e-05, "loss": 2.6408, "step": 35172 }, { "epoch": 2.1834378297845927, "grad_norm": 0.47098779578099786, "learning_rate": 2.092071733139045e-05, "loss": 2.6231, "step": 35173 }, { "epoch": 2.18349990688435, "grad_norm": 0.46648763844260993, "learning_rate": 2.0917779485875128e-05, "loss": 2.647, "step": 35174 }, { "epoch": 2.183561983984108, "grad_norm": 0.5734408552606944, "learning_rate": 2.0914841792088824e-05, "loss": 2.7636, "step": 35175 }, { "epoch": 2.183624061083866, "grad_norm": 0.4771709466497536, "learning_rate": 2.0911904250046853e-05, "loss": 2.6532, "step": 35176 }, { "epoch": 2.183686138183624, "grad_norm": 0.6424138517179121, "learning_rate": 2.0908966859764547e-05, "loss": 2.7666, "step": 35177 }, { "epoch": 2.183748215283382, "grad_norm": 0.5554900743385744, "learning_rate": 2.0906029621257195e-05, "loss": 2.6967, "step": 35178 }, { "epoch": 2.1838102923831397, "grad_norm": 0.5114056451958828, "learning_rate": 2.0903092534540153e-05, "loss": 2.6355, "step": 35179 }, { "epoch": 2.1838723694828976, "grad_norm": 0.4823670761552864, "learning_rate": 2.0900155599628756e-05, "loss": 2.582, "step": 35180 }, { "epoch": 2.1839344465826556, "grad_norm": 0.5048980120320735, "learning_rate": 2.0897218816538317e-05, "loss": 2.6149, "step": 35181 }, { "epoch": 2.1839965236824135, "grad_norm": 0.514234836295459, "learning_rate": 2.089428218528415e-05, "loss": 2.6724, "step": 35182 }, { "epoch": 2.1840586007821714, "grad_norm": 0.5221996188626602, "learning_rate": 2.0891345705881572e-05, "loss": 2.6736, "step": 35183 }, { "epoch": 2.1841206778819293, "grad_norm": 0.5330709216650931, "learning_rate": 2.0888409378345897e-05, "loss": 2.793, "step": 35184 }, { "epoch": 2.1841827549816872, "grad_norm": 0.5444922992890364, "learning_rate": 2.088547320269248e-05, "loss": 2.757, "step": 35185 }, { "epoch": 2.184244832081445, "grad_norm": 0.527071762586202, "learning_rate": 2.088253717893661e-05, "loss": 2.7393, "step": 35186 }, { "epoch": 2.184306909181203, "grad_norm": 0.5543503072200188, "learning_rate": 2.0879601307093616e-05, "loss": 2.6445, "step": 35187 }, { "epoch": 2.184368986280961, "grad_norm": 0.5252226663873457, "learning_rate": 2.087666558717879e-05, "loss": 2.6638, "step": 35188 }, { "epoch": 2.184431063380719, "grad_norm": 0.5368450277219912, "learning_rate": 2.0873730019207487e-05, "loss": 2.7067, "step": 35189 }, { "epoch": 2.184493140480477, "grad_norm": 0.5553714672373788, "learning_rate": 2.087079460319501e-05, "loss": 2.7069, "step": 35190 }, { "epoch": 2.1845552175802347, "grad_norm": 0.5118067716649912, "learning_rate": 2.086785933915667e-05, "loss": 2.7753, "step": 35191 }, { "epoch": 2.1846172946799927, "grad_norm": 0.534535514415333, "learning_rate": 2.0864924227107758e-05, "loss": 2.7266, "step": 35192 }, { "epoch": 2.1846793717797506, "grad_norm": 0.5629452956134117, "learning_rate": 2.0861989267063624e-05, "loss": 2.7884, "step": 35193 }, { "epoch": 2.1847414488795085, "grad_norm": 0.5356326801724247, "learning_rate": 2.0859054459039574e-05, "loss": 2.6654, "step": 35194 }, { "epoch": 2.1848035259792664, "grad_norm": 0.5101161643106434, "learning_rate": 2.0856119803050904e-05, "loss": 2.6433, "step": 35195 }, { "epoch": 2.1848656030790243, "grad_norm": 0.5563318031681214, "learning_rate": 2.0853185299112932e-05, "loss": 2.6975, "step": 35196 }, { "epoch": 2.1849276801787822, "grad_norm": 0.5494178508759476, "learning_rate": 2.0850250947240952e-05, "loss": 2.801, "step": 35197 }, { "epoch": 2.18498975727854, "grad_norm": 0.5234100847039448, "learning_rate": 2.0847316747450306e-05, "loss": 2.6415, "step": 35198 }, { "epoch": 2.1850518343782976, "grad_norm": 0.5459789018921338, "learning_rate": 2.0844382699756287e-05, "loss": 2.6532, "step": 35199 }, { "epoch": 2.1851139114780556, "grad_norm": 0.46656081483591544, "learning_rate": 2.0841448804174203e-05, "loss": 2.5992, "step": 35200 }, { "epoch": 2.1851759885778135, "grad_norm": 0.5702878868330984, "learning_rate": 2.0838515060719338e-05, "loss": 2.7552, "step": 35201 }, { "epoch": 2.1852380656775714, "grad_norm": 0.5146409962530558, "learning_rate": 2.083558146940704e-05, "loss": 2.6384, "step": 35202 }, { "epoch": 2.1853001427773293, "grad_norm": 0.46414285599859895, "learning_rate": 2.0832648030252587e-05, "loss": 2.7227, "step": 35203 }, { "epoch": 2.1853622198770872, "grad_norm": 0.5736147759848764, "learning_rate": 2.08297147432713e-05, "loss": 2.7323, "step": 35204 }, { "epoch": 2.185424296976845, "grad_norm": 0.446802290756719, "learning_rate": 2.0826781608478464e-05, "loss": 2.6749, "step": 35205 }, { "epoch": 2.185486374076603, "grad_norm": 0.47085529257898806, "learning_rate": 2.0823848625889375e-05, "loss": 2.6518, "step": 35206 }, { "epoch": 2.185548451176361, "grad_norm": 0.5390023460549447, "learning_rate": 2.0820915795519368e-05, "loss": 2.7096, "step": 35207 }, { "epoch": 2.185610528276119, "grad_norm": 0.5499555472269599, "learning_rate": 2.0817983117383726e-05, "loss": 2.7339, "step": 35208 }, { "epoch": 2.185672605375877, "grad_norm": 0.46509731012852995, "learning_rate": 2.0815050591497754e-05, "loss": 2.6873, "step": 35209 }, { "epoch": 2.1857346824756347, "grad_norm": 0.5183728432648412, "learning_rate": 2.081211821787673e-05, "loss": 2.6465, "step": 35210 }, { "epoch": 2.1857967595753927, "grad_norm": 0.5224387190898458, "learning_rate": 2.080918599653597e-05, "loss": 2.6817, "step": 35211 }, { "epoch": 2.1858588366751506, "grad_norm": 0.5359011290195059, "learning_rate": 2.0806253927490787e-05, "loss": 2.7005, "step": 35212 }, { "epoch": 2.1859209137749085, "grad_norm": 0.5674161917606555, "learning_rate": 2.0803322010756466e-05, "loss": 2.6882, "step": 35213 }, { "epoch": 2.1859829908746664, "grad_norm": 0.5151759995946231, "learning_rate": 2.0800390246348305e-05, "loss": 2.6631, "step": 35214 }, { "epoch": 2.1860450679744243, "grad_norm": 0.4665016294044095, "learning_rate": 2.0797458634281575e-05, "loss": 2.7163, "step": 35215 }, { "epoch": 2.1861071450741822, "grad_norm": 0.49591645504408455, "learning_rate": 2.079452717457161e-05, "loss": 2.7356, "step": 35216 }, { "epoch": 2.18616922217394, "grad_norm": 0.48937163104557785, "learning_rate": 2.0791595867233686e-05, "loss": 2.6669, "step": 35217 }, { "epoch": 2.186231299273698, "grad_norm": 0.4383169448651099, "learning_rate": 2.0788664712283102e-05, "loss": 2.6569, "step": 35218 }, { "epoch": 2.186293376373456, "grad_norm": 0.5154313388464097, "learning_rate": 2.0785733709735144e-05, "loss": 2.7111, "step": 35219 }, { "epoch": 2.186355453473214, "grad_norm": 0.5181930631652506, "learning_rate": 2.0782802859605084e-05, "loss": 2.5938, "step": 35220 }, { "epoch": 2.186417530572972, "grad_norm": 0.4988432855774184, "learning_rate": 2.0779872161908255e-05, "loss": 2.6105, "step": 35221 }, { "epoch": 2.1864796076727293, "grad_norm": 0.5408271433687144, "learning_rate": 2.0776941616659928e-05, "loss": 2.7966, "step": 35222 }, { "epoch": 2.1865416847724872, "grad_norm": 0.5705676096105423, "learning_rate": 2.0774011223875383e-05, "loss": 2.7204, "step": 35223 }, { "epoch": 2.186603761872245, "grad_norm": 0.48273217110131533, "learning_rate": 2.077108098356991e-05, "loss": 2.7051, "step": 35224 }, { "epoch": 2.186665838972003, "grad_norm": 0.5544084196113289, "learning_rate": 2.0768150895758808e-05, "loss": 2.6914, "step": 35225 }, { "epoch": 2.186727916071761, "grad_norm": 0.550788879341691, "learning_rate": 2.076522096045737e-05, "loss": 2.6997, "step": 35226 }, { "epoch": 2.186789993171519, "grad_norm": 0.5160783450916658, "learning_rate": 2.0762291177680866e-05, "loss": 2.6423, "step": 35227 }, { "epoch": 2.186852070271277, "grad_norm": 0.573373444848855, "learning_rate": 2.0759361547444573e-05, "loss": 2.5577, "step": 35228 }, { "epoch": 2.1869141473710347, "grad_norm": 0.5689551256029555, "learning_rate": 2.07564320697638e-05, "loss": 2.6698, "step": 35229 }, { "epoch": 2.1869762244707927, "grad_norm": 0.5734349565163731, "learning_rate": 2.0753502744653826e-05, "loss": 2.7702, "step": 35230 }, { "epoch": 2.1870383015705506, "grad_norm": 0.5642398576530554, "learning_rate": 2.0750573572129922e-05, "loss": 2.6734, "step": 35231 }, { "epoch": 2.1871003786703085, "grad_norm": 0.5137005932108503, "learning_rate": 2.0747644552207383e-05, "loss": 2.665, "step": 35232 }, { "epoch": 2.1871624557700664, "grad_norm": 0.547315828578751, "learning_rate": 2.0744715684901462e-05, "loss": 2.6872, "step": 35233 }, { "epoch": 2.1872245328698243, "grad_norm": 0.5247911912199893, "learning_rate": 2.074178697022748e-05, "loss": 2.683, "step": 35234 }, { "epoch": 2.1872866099695822, "grad_norm": 0.4882949522003218, "learning_rate": 2.0738858408200696e-05, "loss": 2.7089, "step": 35235 }, { "epoch": 2.18734868706934, "grad_norm": 0.5706126063610601, "learning_rate": 2.0735929998836395e-05, "loss": 2.7195, "step": 35236 }, { "epoch": 2.187410764169098, "grad_norm": 0.4964254126663902, "learning_rate": 2.0733001742149833e-05, "loss": 2.6809, "step": 35237 }, { "epoch": 2.187472841268856, "grad_norm": 0.6155980806225976, "learning_rate": 2.0730073638156323e-05, "loss": 2.6513, "step": 35238 }, { "epoch": 2.187534918368614, "grad_norm": 0.5136552525140646, "learning_rate": 2.072714568687112e-05, "loss": 2.7225, "step": 35239 }, { "epoch": 2.187596995468372, "grad_norm": 0.5254064791531461, "learning_rate": 2.072421788830951e-05, "loss": 2.7344, "step": 35240 }, { "epoch": 2.1876590725681297, "grad_norm": 0.6380562104258841, "learning_rate": 2.072129024248674e-05, "loss": 2.705, "step": 35241 }, { "epoch": 2.1877211496678877, "grad_norm": 0.5709570025624527, "learning_rate": 2.0718362749418123e-05, "loss": 2.7252, "step": 35242 }, { "epoch": 2.1877832267676456, "grad_norm": 0.493658788320129, "learning_rate": 2.0715435409118906e-05, "loss": 2.642, "step": 35243 }, { "epoch": 2.1878453038674035, "grad_norm": 0.6337401807652604, "learning_rate": 2.0712508221604382e-05, "loss": 2.6909, "step": 35244 }, { "epoch": 2.1879073809671614, "grad_norm": 0.5470866474952848, "learning_rate": 2.0709581186889814e-05, "loss": 2.6967, "step": 35245 }, { "epoch": 2.1879694580669193, "grad_norm": 0.5329192806872735, "learning_rate": 2.0706654304990476e-05, "loss": 2.6713, "step": 35246 }, { "epoch": 2.188031535166677, "grad_norm": 0.5594406110762067, "learning_rate": 2.070372757592161e-05, "loss": 2.7225, "step": 35247 }, { "epoch": 2.1880936122664347, "grad_norm": 0.49391420929386853, "learning_rate": 2.0700800999698534e-05, "loss": 2.7418, "step": 35248 }, { "epoch": 2.1881556893661926, "grad_norm": 0.556557418226017, "learning_rate": 2.0697874576336484e-05, "loss": 2.7612, "step": 35249 }, { "epoch": 2.1882177664659506, "grad_norm": 0.4974802122062635, "learning_rate": 2.0694948305850738e-05, "loss": 2.6622, "step": 35250 }, { "epoch": 2.1882798435657085, "grad_norm": 0.5373837411812435, "learning_rate": 2.0692022188256543e-05, "loss": 2.6653, "step": 35251 }, { "epoch": 2.1883419206654664, "grad_norm": 0.5962942223773894, "learning_rate": 2.0689096223569204e-05, "loss": 2.6907, "step": 35252 }, { "epoch": 2.1884039977652243, "grad_norm": 0.5199370842707417, "learning_rate": 2.0686170411803963e-05, "loss": 2.7156, "step": 35253 }, { "epoch": 2.1884660748649822, "grad_norm": 0.5771839111580053, "learning_rate": 2.0683244752976087e-05, "loss": 2.6378, "step": 35254 }, { "epoch": 2.18852815196474, "grad_norm": 0.5122009509719397, "learning_rate": 2.0680319247100837e-05, "loss": 2.7169, "step": 35255 }, { "epoch": 2.188590229064498, "grad_norm": 0.5310000785425275, "learning_rate": 2.067739389419347e-05, "loss": 2.6978, "step": 35256 }, { "epoch": 2.188652306164256, "grad_norm": 0.5541504153638944, "learning_rate": 2.0674468694269267e-05, "loss": 2.61, "step": 35257 }, { "epoch": 2.188714383264014, "grad_norm": 0.6039208838502145, "learning_rate": 2.0671543647343484e-05, "loss": 2.6662, "step": 35258 }, { "epoch": 2.188776460363772, "grad_norm": 0.522475368977818, "learning_rate": 2.0668618753431375e-05, "loss": 2.8262, "step": 35259 }, { "epoch": 2.1888385374635297, "grad_norm": 0.5385558470206293, "learning_rate": 2.0665694012548188e-05, "loss": 2.6766, "step": 35260 }, { "epoch": 2.1889006145632877, "grad_norm": 0.5652654801399714, "learning_rate": 2.066276942470921e-05, "loss": 2.6698, "step": 35261 }, { "epoch": 2.1889626916630456, "grad_norm": 0.48733680940304613, "learning_rate": 2.0659844989929688e-05, "loss": 2.6518, "step": 35262 }, { "epoch": 2.1890247687628035, "grad_norm": 0.5260040219374376, "learning_rate": 2.0656920708224874e-05, "loss": 2.6096, "step": 35263 }, { "epoch": 2.1890868458625614, "grad_norm": 0.5308966022207403, "learning_rate": 2.0653996579610014e-05, "loss": 2.7123, "step": 35264 }, { "epoch": 2.1891489229623193, "grad_norm": 0.4722861898407855, "learning_rate": 2.0651072604100398e-05, "loss": 2.6867, "step": 35265 }, { "epoch": 2.1892110000620773, "grad_norm": 0.5955066794750311, "learning_rate": 2.0648148781711258e-05, "loss": 2.7311, "step": 35266 }, { "epoch": 2.189273077161835, "grad_norm": 0.4956282135954263, "learning_rate": 2.064522511245785e-05, "loss": 2.744, "step": 35267 }, { "epoch": 2.189335154261593, "grad_norm": 0.5054373729937259, "learning_rate": 2.064230159635543e-05, "loss": 2.6721, "step": 35268 }, { "epoch": 2.189397231361351, "grad_norm": 0.5348073281299648, "learning_rate": 2.0639378233419225e-05, "loss": 2.7143, "step": 35269 }, { "epoch": 2.1894593084611085, "grad_norm": 0.5292105797947045, "learning_rate": 2.063645502366454e-05, "loss": 2.6844, "step": 35270 }, { "epoch": 2.1895213855608664, "grad_norm": 0.4967749116964407, "learning_rate": 2.063353196710659e-05, "loss": 2.7023, "step": 35271 }, { "epoch": 2.1895834626606243, "grad_norm": 0.552773895468356, "learning_rate": 2.0630609063760635e-05, "loss": 2.7146, "step": 35272 }, { "epoch": 2.1896455397603822, "grad_norm": 0.5340005689396489, "learning_rate": 2.062768631364191e-05, "loss": 2.6699, "step": 35273 }, { "epoch": 2.18970761686014, "grad_norm": 0.5213291637307979, "learning_rate": 2.062476371676569e-05, "loss": 2.7567, "step": 35274 }, { "epoch": 2.189769693959898, "grad_norm": 0.5105773192874853, "learning_rate": 2.0621841273147202e-05, "loss": 2.7089, "step": 35275 }, { "epoch": 2.189831771059656, "grad_norm": 0.49182537926445147, "learning_rate": 2.061891898280169e-05, "loss": 2.6613, "step": 35276 }, { "epoch": 2.189893848159414, "grad_norm": 0.5021262853566575, "learning_rate": 2.0615996845744423e-05, "loss": 2.6576, "step": 35277 }, { "epoch": 2.189955925259172, "grad_norm": 0.553362728476579, "learning_rate": 2.0613074861990617e-05, "loss": 2.7607, "step": 35278 }, { "epoch": 2.1900180023589297, "grad_norm": 0.5088027688667843, "learning_rate": 2.061015303155555e-05, "loss": 2.6418, "step": 35279 }, { "epoch": 2.1900800794586877, "grad_norm": 0.5313609231822033, "learning_rate": 2.0607231354454455e-05, "loss": 2.6659, "step": 35280 }, { "epoch": 2.1901421565584456, "grad_norm": 0.5301502288377501, "learning_rate": 2.060430983070257e-05, "loss": 2.682, "step": 35281 }, { "epoch": 2.1902042336582035, "grad_norm": 0.47067966640155207, "learning_rate": 2.060138846031513e-05, "loss": 2.6938, "step": 35282 }, { "epoch": 2.1902663107579614, "grad_norm": 0.5217435385683677, "learning_rate": 2.0598467243307367e-05, "loss": 2.7457, "step": 35283 }, { "epoch": 2.1903283878577193, "grad_norm": 0.46908107827517004, "learning_rate": 2.059554617969456e-05, "loss": 2.7209, "step": 35284 }, { "epoch": 2.1903904649574772, "grad_norm": 0.4867579693859433, "learning_rate": 2.0592625269491923e-05, "loss": 2.6927, "step": 35285 }, { "epoch": 2.190452542057235, "grad_norm": 0.49671092653308274, "learning_rate": 2.0589704512714696e-05, "loss": 2.7431, "step": 35286 }, { "epoch": 2.190514619156993, "grad_norm": 0.4865039829206768, "learning_rate": 2.0586783909378104e-05, "loss": 2.7496, "step": 35287 }, { "epoch": 2.190576696256751, "grad_norm": 0.5213252764257728, "learning_rate": 2.058386345949742e-05, "loss": 2.5949, "step": 35288 }, { "epoch": 2.190638773356509, "grad_norm": 0.5019296425399251, "learning_rate": 2.058094316308786e-05, "loss": 2.7284, "step": 35289 }, { "epoch": 2.190700850456267, "grad_norm": 0.5470216282050377, "learning_rate": 2.0578023020164668e-05, "loss": 2.661, "step": 35290 }, { "epoch": 2.1907629275560248, "grad_norm": 0.5384571435835451, "learning_rate": 2.0575103030743048e-05, "loss": 2.6741, "step": 35291 }, { "epoch": 2.1908250046557827, "grad_norm": 0.5358658979090906, "learning_rate": 2.057218319483828e-05, "loss": 2.5833, "step": 35292 }, { "epoch": 2.1908870817555406, "grad_norm": 0.47938704260430065, "learning_rate": 2.056926351246557e-05, "loss": 2.6043, "step": 35293 }, { "epoch": 2.1909491588552985, "grad_norm": 0.5133064333516456, "learning_rate": 2.0566343983640162e-05, "loss": 2.6768, "step": 35294 }, { "epoch": 2.191011235955056, "grad_norm": 0.543669653031602, "learning_rate": 2.056342460837728e-05, "loss": 2.7019, "step": 35295 }, { "epoch": 2.191073313054814, "grad_norm": 0.4955519473486681, "learning_rate": 2.056050538669214e-05, "loss": 2.7282, "step": 35296 }, { "epoch": 2.191135390154572, "grad_norm": 0.525340308293562, "learning_rate": 2.0557586318600008e-05, "loss": 2.7408, "step": 35297 }, { "epoch": 2.1911974672543297, "grad_norm": 0.5367797672508642, "learning_rate": 2.05546674041161e-05, "loss": 2.692, "step": 35298 }, { "epoch": 2.1912595443540877, "grad_norm": 0.4576427900289233, "learning_rate": 2.0551748643255638e-05, "loss": 2.7014, "step": 35299 }, { "epoch": 2.1913216214538456, "grad_norm": 0.4891242417920166, "learning_rate": 2.0548830036033833e-05, "loss": 2.7277, "step": 35300 }, { "epoch": 2.1913836985536035, "grad_norm": 0.521539806605951, "learning_rate": 2.054591158246595e-05, "loss": 2.6347, "step": 35301 }, { "epoch": 2.1914457756533614, "grad_norm": 0.4503455087342708, "learning_rate": 2.054299328256719e-05, "loss": 2.6914, "step": 35302 }, { "epoch": 2.1915078527531193, "grad_norm": 0.4723631209964098, "learning_rate": 2.0540075136352794e-05, "loss": 2.7299, "step": 35303 }, { "epoch": 2.1915699298528772, "grad_norm": 0.5401592983063834, "learning_rate": 2.0537157143837974e-05, "loss": 2.7575, "step": 35304 }, { "epoch": 2.191632006952635, "grad_norm": 0.4862603902564732, "learning_rate": 2.0534239305037937e-05, "loss": 2.6485, "step": 35305 }, { "epoch": 2.191694084052393, "grad_norm": 0.4502269377947238, "learning_rate": 2.0531321619967953e-05, "loss": 2.7372, "step": 35306 }, { "epoch": 2.191756161152151, "grad_norm": 0.503558460245432, "learning_rate": 2.052840408864321e-05, "loss": 2.6607, "step": 35307 }, { "epoch": 2.191818238251909, "grad_norm": 0.45964181767690293, "learning_rate": 2.052548671107894e-05, "loss": 2.7572, "step": 35308 }, { "epoch": 2.191880315351667, "grad_norm": 0.5303681571765365, "learning_rate": 2.0522569487290344e-05, "loss": 2.6175, "step": 35309 }, { "epoch": 2.1919423924514247, "grad_norm": 0.521422393201406, "learning_rate": 2.0519652417292657e-05, "loss": 2.5806, "step": 35310 }, { "epoch": 2.1920044695511827, "grad_norm": 0.48132109055024885, "learning_rate": 2.051673550110112e-05, "loss": 2.7242, "step": 35311 }, { "epoch": 2.1920665466509406, "grad_norm": 0.5143514279217134, "learning_rate": 2.0513818738730928e-05, "loss": 2.7074, "step": 35312 }, { "epoch": 2.1921286237506985, "grad_norm": 0.5523301388629552, "learning_rate": 2.0510902130197308e-05, "loss": 2.6424, "step": 35313 }, { "epoch": 2.1921907008504564, "grad_norm": 0.4855176154210552, "learning_rate": 2.0507985675515446e-05, "loss": 2.6509, "step": 35314 }, { "epoch": 2.1922527779502143, "grad_norm": 0.5381837953177074, "learning_rate": 2.0505069374700603e-05, "loss": 2.7732, "step": 35315 }, { "epoch": 2.1923148550499723, "grad_norm": 0.520135112687061, "learning_rate": 2.0502153227767972e-05, "loss": 2.6155, "step": 35316 }, { "epoch": 2.19237693214973, "grad_norm": 0.5253968362547433, "learning_rate": 2.0499237234732765e-05, "loss": 2.8307, "step": 35317 }, { "epoch": 2.1924390092494876, "grad_norm": 0.5722681768054507, "learning_rate": 2.0496321395610207e-05, "loss": 2.7067, "step": 35318 }, { "epoch": 2.1925010863492456, "grad_norm": 0.5408049355659761, "learning_rate": 2.049340571041548e-05, "loss": 2.7047, "step": 35319 }, { "epoch": 2.1925631634490035, "grad_norm": 0.5590381588442562, "learning_rate": 2.0490490179163835e-05, "loss": 2.7875, "step": 35320 }, { "epoch": 2.1926252405487614, "grad_norm": 0.5427256252538428, "learning_rate": 2.0487574801870468e-05, "loss": 2.7, "step": 35321 }, { "epoch": 2.1926873176485193, "grad_norm": 0.5572273241018498, "learning_rate": 2.0484659578550586e-05, "loss": 2.7318, "step": 35322 }, { "epoch": 2.1927493947482772, "grad_norm": 0.5064441438154631, "learning_rate": 2.0481744509219387e-05, "loss": 2.7464, "step": 35323 }, { "epoch": 2.192811471848035, "grad_norm": 0.5828874677862219, "learning_rate": 2.0478829593892107e-05, "loss": 2.7826, "step": 35324 }, { "epoch": 2.192873548947793, "grad_norm": 0.5033693928412193, "learning_rate": 2.0475914832583937e-05, "loss": 2.6679, "step": 35325 }, { "epoch": 2.192935626047551, "grad_norm": 0.5205581096803414, "learning_rate": 2.0473000225310092e-05, "loss": 2.6602, "step": 35326 }, { "epoch": 2.192997703147309, "grad_norm": 0.4859850182559209, "learning_rate": 2.0470085772085752e-05, "loss": 2.6905, "step": 35327 }, { "epoch": 2.193059780247067, "grad_norm": 0.5534782391658115, "learning_rate": 2.0467171472926162e-05, "loss": 2.789, "step": 35328 }, { "epoch": 2.1931218573468247, "grad_norm": 0.4980937262545321, "learning_rate": 2.046425732784651e-05, "loss": 2.668, "step": 35329 }, { "epoch": 2.1931839344465827, "grad_norm": 0.5765905895472165, "learning_rate": 2.0461343336861992e-05, "loss": 2.7992, "step": 35330 }, { "epoch": 2.1932460115463406, "grad_norm": 0.4959651900591455, "learning_rate": 2.0458429499987817e-05, "loss": 2.65, "step": 35331 }, { "epoch": 2.1933080886460985, "grad_norm": 0.48539527987844244, "learning_rate": 2.0455515817239175e-05, "loss": 2.6962, "step": 35332 }, { "epoch": 2.1933701657458564, "grad_norm": 0.5518255951693236, "learning_rate": 2.0452602288631288e-05, "loss": 2.6695, "step": 35333 }, { "epoch": 2.1934322428456143, "grad_norm": 0.4818082141980034, "learning_rate": 2.0449688914179355e-05, "loss": 2.678, "step": 35334 }, { "epoch": 2.1934943199453723, "grad_norm": 0.5309566360985747, "learning_rate": 2.044677569389857e-05, "loss": 2.6924, "step": 35335 }, { "epoch": 2.19355639704513, "grad_norm": 0.4861313142059888, "learning_rate": 2.0443862627804105e-05, "loss": 2.738, "step": 35336 }, { "epoch": 2.193618474144888, "grad_norm": 0.490640100015933, "learning_rate": 2.0440949715911207e-05, "loss": 2.8106, "step": 35337 }, { "epoch": 2.193680551244646, "grad_norm": 0.48234662538591394, "learning_rate": 2.0438036958235046e-05, "loss": 2.6099, "step": 35338 }, { "epoch": 2.193742628344404, "grad_norm": 0.48394978224903834, "learning_rate": 2.043512435479083e-05, "loss": 2.6181, "step": 35339 }, { "epoch": 2.193804705444162, "grad_norm": 0.48776266406740265, "learning_rate": 2.043221190559374e-05, "loss": 2.7135, "step": 35340 }, { "epoch": 2.1938667825439193, "grad_norm": 0.5004110434875567, "learning_rate": 2.0429299610658966e-05, "loss": 2.6622, "step": 35341 }, { "epoch": 2.1939288596436777, "grad_norm": 0.5179428542900606, "learning_rate": 2.0426387470001705e-05, "loss": 2.6891, "step": 35342 }, { "epoch": 2.193990936743435, "grad_norm": 0.5283994774023153, "learning_rate": 2.042347548363719e-05, "loss": 2.6213, "step": 35343 }, { "epoch": 2.194053013843193, "grad_norm": 0.507974804103214, "learning_rate": 2.0420563651580577e-05, "loss": 2.6783, "step": 35344 }, { "epoch": 2.194115090942951, "grad_norm": 0.5521687825370908, "learning_rate": 2.0417651973847063e-05, "loss": 2.7236, "step": 35345 }, { "epoch": 2.194177168042709, "grad_norm": 0.47749281583244824, "learning_rate": 2.0414740450451826e-05, "loss": 2.6895, "step": 35346 }, { "epoch": 2.194239245142467, "grad_norm": 0.49846866175423754, "learning_rate": 2.0411829081410084e-05, "loss": 2.7234, "step": 35347 }, { "epoch": 2.1943013222422247, "grad_norm": 0.4966085271227673, "learning_rate": 2.0408917866737014e-05, "loss": 2.6164, "step": 35348 }, { "epoch": 2.1943633993419827, "grad_norm": 0.4764752699017542, "learning_rate": 2.04060068064478e-05, "loss": 2.7211, "step": 35349 }, { "epoch": 2.1944254764417406, "grad_norm": 0.4971424043010112, "learning_rate": 2.0403095900557624e-05, "loss": 2.6531, "step": 35350 }, { "epoch": 2.1944875535414985, "grad_norm": 0.5051165690767483, "learning_rate": 2.0400185149081692e-05, "loss": 2.6606, "step": 35351 }, { "epoch": 2.1945496306412564, "grad_norm": 0.449924975339975, "learning_rate": 2.0397274552035184e-05, "loss": 2.6899, "step": 35352 }, { "epoch": 2.1946117077410143, "grad_norm": 0.4833354099683984, "learning_rate": 2.0394364109433273e-05, "loss": 2.6848, "step": 35353 }, { "epoch": 2.1946737848407722, "grad_norm": 0.5371479922196994, "learning_rate": 2.0391453821291156e-05, "loss": 2.7132, "step": 35354 }, { "epoch": 2.19473586194053, "grad_norm": 0.4329625596010366, "learning_rate": 2.0388543687623997e-05, "loss": 2.7427, "step": 35355 }, { "epoch": 2.194797939040288, "grad_norm": 0.5240757595106795, "learning_rate": 2.038563370844701e-05, "loss": 2.7636, "step": 35356 }, { "epoch": 2.194860016140046, "grad_norm": 0.56435720823343, "learning_rate": 2.0382723883775363e-05, "loss": 2.663, "step": 35357 }, { "epoch": 2.194922093239804, "grad_norm": 0.4771709466497536, "learning_rate": 2.0379814213624233e-05, "loss": 2.7235, "step": 35358 }, { "epoch": 2.194984170339562, "grad_norm": 0.5504655135939628, "learning_rate": 2.0376904698008785e-05, "loss": 2.7501, "step": 35359 }, { "epoch": 2.1950462474393198, "grad_norm": 0.5483038899836983, "learning_rate": 2.0373995336944236e-05, "loss": 2.6967, "step": 35360 }, { "epoch": 2.1951083245390777, "grad_norm": 0.5261999936934253, "learning_rate": 2.0371086130445744e-05, "loss": 2.666, "step": 35361 }, { "epoch": 2.1951704016388356, "grad_norm": 0.5218968278351757, "learning_rate": 2.0368177078528482e-05, "loss": 2.6254, "step": 35362 }, { "epoch": 2.1952324787385935, "grad_norm": 0.5478412538806351, "learning_rate": 2.0365268181207624e-05, "loss": 2.625, "step": 35363 }, { "epoch": 2.1952945558383514, "grad_norm": 0.5993703061139797, "learning_rate": 2.0362359438498374e-05, "loss": 2.7326, "step": 35364 }, { "epoch": 2.1953566329381093, "grad_norm": 0.49210935305292397, "learning_rate": 2.0359450850415885e-05, "loss": 2.7498, "step": 35365 }, { "epoch": 2.195418710037867, "grad_norm": 0.5439687891607468, "learning_rate": 2.035654241697534e-05, "loss": 2.6154, "step": 35366 }, { "epoch": 2.1954807871376247, "grad_norm": 0.5881162191580724, "learning_rate": 2.0353634138191908e-05, "loss": 2.7515, "step": 35367 }, { "epoch": 2.1955428642373827, "grad_norm": 0.5281133616592762, "learning_rate": 2.0350726014080746e-05, "loss": 2.7362, "step": 35368 }, { "epoch": 2.1956049413371406, "grad_norm": 0.5971497277930683, "learning_rate": 2.0347818044657062e-05, "loss": 2.7728, "step": 35369 }, { "epoch": 2.1956670184368985, "grad_norm": 0.5430541760323323, "learning_rate": 2.034491022993601e-05, "loss": 2.6505, "step": 35370 }, { "epoch": 2.1957290955366564, "grad_norm": 0.4838010402239324, "learning_rate": 2.034200256993276e-05, "loss": 2.7587, "step": 35371 }, { "epoch": 2.1957911726364143, "grad_norm": 0.5321488909959908, "learning_rate": 2.033909506466246e-05, "loss": 2.6327, "step": 35372 }, { "epoch": 2.1958532497361722, "grad_norm": 0.5285468047898082, "learning_rate": 2.0336187714140327e-05, "loss": 2.6465, "step": 35373 }, { "epoch": 2.19591532683593, "grad_norm": 0.4849778085232927, "learning_rate": 2.033328051838148e-05, "loss": 2.6232, "step": 35374 }, { "epoch": 2.195977403935688, "grad_norm": 0.4653790541243528, "learning_rate": 2.0330373477401132e-05, "loss": 2.6912, "step": 35375 }, { "epoch": 2.196039481035446, "grad_norm": 0.5314666642817472, "learning_rate": 2.0327466591214427e-05, "loss": 2.6511, "step": 35376 }, { "epoch": 2.196101558135204, "grad_norm": 0.4929173522852012, "learning_rate": 2.032455985983651e-05, "loss": 2.6444, "step": 35377 }, { "epoch": 2.196163635234962, "grad_norm": 0.523535619620595, "learning_rate": 2.0321653283282594e-05, "loss": 2.6796, "step": 35378 }, { "epoch": 2.1962257123347197, "grad_norm": 0.5142044381601862, "learning_rate": 2.031874686156781e-05, "loss": 2.6606, "step": 35379 }, { "epoch": 2.1962877894344777, "grad_norm": 0.486383136257686, "learning_rate": 2.0315840594707337e-05, "loss": 2.7015, "step": 35380 }, { "epoch": 2.1963498665342356, "grad_norm": 0.5030814525168263, "learning_rate": 2.0312934482716324e-05, "loss": 2.5776, "step": 35381 }, { "epoch": 2.1964119436339935, "grad_norm": 0.4859500472481967, "learning_rate": 2.031002852560992e-05, "loss": 2.7401, "step": 35382 }, { "epoch": 2.1964740207337514, "grad_norm": 0.5455426470994085, "learning_rate": 2.0307122723403326e-05, "loss": 2.6735, "step": 35383 }, { "epoch": 2.1965360978335093, "grad_norm": 0.48199626356671055, "learning_rate": 2.030421707611168e-05, "loss": 2.641, "step": 35384 }, { "epoch": 2.1965981749332673, "grad_norm": 0.5563200713609815, "learning_rate": 2.0301311583750142e-05, "loss": 2.758, "step": 35385 }, { "epoch": 2.196660252033025, "grad_norm": 0.5548473718772592, "learning_rate": 2.0298406246333852e-05, "loss": 2.6966, "step": 35386 }, { "epoch": 2.196722329132783, "grad_norm": 0.4919325455833025, "learning_rate": 2.029550106387801e-05, "loss": 2.684, "step": 35387 }, { "epoch": 2.196784406232541, "grad_norm": 0.5746688376337379, "learning_rate": 2.0292596036397743e-05, "loss": 2.6883, "step": 35388 }, { "epoch": 2.1968464833322985, "grad_norm": 0.47635980892058405, "learning_rate": 2.0289691163908216e-05, "loss": 2.7669, "step": 35389 }, { "epoch": 2.196908560432057, "grad_norm": 0.5159831681537321, "learning_rate": 2.0286786446424583e-05, "loss": 2.6834, "step": 35390 }, { "epoch": 2.1969706375318143, "grad_norm": 0.5012323273185529, "learning_rate": 2.028388188396198e-05, "loss": 2.7408, "step": 35391 }, { "epoch": 2.1970327146315722, "grad_norm": 0.4913356918356425, "learning_rate": 2.02809774765356e-05, "loss": 2.6478, "step": 35392 }, { "epoch": 2.19709479173133, "grad_norm": 0.49086383667587336, "learning_rate": 2.0278073224160572e-05, "loss": 2.737, "step": 35393 }, { "epoch": 2.197156868831088, "grad_norm": 0.4986551945433659, "learning_rate": 2.027516912685205e-05, "loss": 2.663, "step": 35394 }, { "epoch": 2.197218945930846, "grad_norm": 0.4823109734819889, "learning_rate": 2.0272265184625173e-05, "loss": 2.7683, "step": 35395 }, { "epoch": 2.197281023030604, "grad_norm": 0.48168572541856985, "learning_rate": 2.0269361397495123e-05, "loss": 2.7568, "step": 35396 }, { "epoch": 2.197343100130362, "grad_norm": 0.5143046379865239, "learning_rate": 2.0266457765477033e-05, "loss": 2.665, "step": 35397 }, { "epoch": 2.1974051772301197, "grad_norm": 0.48821219909703234, "learning_rate": 2.0263554288586044e-05, "loss": 2.63, "step": 35398 }, { "epoch": 2.1974672543298777, "grad_norm": 0.545756039958668, "learning_rate": 2.0260650966837298e-05, "loss": 2.6826, "step": 35399 }, { "epoch": 2.1975293314296356, "grad_norm": 0.5365674978744408, "learning_rate": 2.0257747800245967e-05, "loss": 2.6564, "step": 35400 }, { "epoch": 2.1975914085293935, "grad_norm": 0.4879462351690808, "learning_rate": 2.0254844788827193e-05, "loss": 2.6406, "step": 35401 }, { "epoch": 2.1976534856291514, "grad_norm": 0.5510858275295876, "learning_rate": 2.0251941932596113e-05, "loss": 2.6717, "step": 35402 }, { "epoch": 2.1977155627289093, "grad_norm": 0.4376219170861395, "learning_rate": 2.0249039231567873e-05, "loss": 2.6396, "step": 35403 }, { "epoch": 2.1977776398286673, "grad_norm": 0.5569538506749157, "learning_rate": 2.0246136685757604e-05, "loss": 2.717, "step": 35404 }, { "epoch": 2.197839716928425, "grad_norm": 0.4820685695254719, "learning_rate": 2.0243234295180475e-05, "loss": 2.6935, "step": 35405 }, { "epoch": 2.197901794028183, "grad_norm": 0.5370768311784915, "learning_rate": 2.024033205985162e-05, "loss": 2.7133, "step": 35406 }, { "epoch": 2.197963871127941, "grad_norm": 0.5228402730401964, "learning_rate": 2.0237429979786154e-05, "loss": 2.7575, "step": 35407 }, { "epoch": 2.198025948227699, "grad_norm": 0.4668261466963854, "learning_rate": 2.0234528054999263e-05, "loss": 2.6459, "step": 35408 }, { "epoch": 2.198088025327457, "grad_norm": 0.5245506634385223, "learning_rate": 2.023162628550605e-05, "loss": 2.6502, "step": 35409 }, { "epoch": 2.1981501024272148, "grad_norm": 0.5255863151421355, "learning_rate": 2.0228724671321682e-05, "loss": 2.7348, "step": 35410 }, { "epoch": 2.1982121795269727, "grad_norm": 0.4452510255912094, "learning_rate": 2.0225823212461286e-05, "loss": 2.6174, "step": 35411 }, { "epoch": 2.1982742566267306, "grad_norm": 0.5217158914483729, "learning_rate": 2.0222921908939997e-05, "loss": 2.5737, "step": 35412 }, { "epoch": 2.1983363337264885, "grad_norm": 0.520267910808048, "learning_rate": 2.022002076077294e-05, "loss": 2.7193, "step": 35413 }, { "epoch": 2.198398410826246, "grad_norm": 0.46683312119732556, "learning_rate": 2.0217119767975275e-05, "loss": 2.7076, "step": 35414 }, { "epoch": 2.198460487926004, "grad_norm": 0.5014982128834092, "learning_rate": 2.021421893056213e-05, "loss": 2.6967, "step": 35415 }, { "epoch": 2.198522565025762, "grad_norm": 0.543700870399066, "learning_rate": 2.0211318248548632e-05, "loss": 2.6712, "step": 35416 }, { "epoch": 2.1985846421255197, "grad_norm": 0.4909316950939877, "learning_rate": 2.0208417721949923e-05, "loss": 2.6518, "step": 35417 }, { "epoch": 2.1986467192252777, "grad_norm": 0.5305492884940957, "learning_rate": 2.0205517350781112e-05, "loss": 2.6954, "step": 35418 }, { "epoch": 2.1987087963250356, "grad_norm": 0.5018026816685291, "learning_rate": 2.0202617135057366e-05, "loss": 2.6496, "step": 35419 }, { "epoch": 2.1987708734247935, "grad_norm": 0.56946003577228, "learning_rate": 2.01997170747938e-05, "loss": 2.6792, "step": 35420 }, { "epoch": 2.1988329505245514, "grad_norm": 0.4940505869259768, "learning_rate": 2.019681717000555e-05, "loss": 2.7249, "step": 35421 }, { "epoch": 2.1988950276243093, "grad_norm": 0.5006287019571309, "learning_rate": 2.0193917420707715e-05, "loss": 2.7122, "step": 35422 }, { "epoch": 2.1989571047240672, "grad_norm": 0.5571611614067521, "learning_rate": 2.019101782691547e-05, "loss": 2.6678, "step": 35423 }, { "epoch": 2.199019181823825, "grad_norm": 0.4982627998972241, "learning_rate": 2.0188118388643917e-05, "loss": 2.7246, "step": 35424 }, { "epoch": 2.199081258923583, "grad_norm": 0.5371850255229029, "learning_rate": 2.0185219105908192e-05, "loss": 2.591, "step": 35425 }, { "epoch": 2.199143336023341, "grad_norm": 0.49196377468107316, "learning_rate": 2.0182319978723413e-05, "loss": 2.6405, "step": 35426 }, { "epoch": 2.199205413123099, "grad_norm": 0.49651071342038694, "learning_rate": 2.0179421007104698e-05, "loss": 2.689, "step": 35427 }, { "epoch": 2.199267490222857, "grad_norm": 0.5090198542539266, "learning_rate": 2.0176522191067193e-05, "loss": 2.6763, "step": 35428 }, { "epoch": 2.1993295673226148, "grad_norm": 0.48649392116939943, "learning_rate": 2.0173623530626018e-05, "loss": 2.7341, "step": 35429 }, { "epoch": 2.1993916444223727, "grad_norm": 0.525658011199303, "learning_rate": 2.017072502579628e-05, "loss": 2.6288, "step": 35430 }, { "epoch": 2.1994537215221306, "grad_norm": 0.4638265009654538, "learning_rate": 2.0167826676593104e-05, "loss": 2.6467, "step": 35431 }, { "epoch": 2.1995157986218885, "grad_norm": 0.4898075205958812, "learning_rate": 2.016492848303163e-05, "loss": 2.7382, "step": 35432 }, { "epoch": 2.1995778757216464, "grad_norm": 0.5404617283528482, "learning_rate": 2.0162030445126967e-05, "loss": 2.7271, "step": 35433 }, { "epoch": 2.1996399528214043, "grad_norm": 0.4957501881339851, "learning_rate": 2.015913256289424e-05, "loss": 2.6256, "step": 35434 }, { "epoch": 2.1997020299211623, "grad_norm": 0.5005944413896479, "learning_rate": 2.0156234836348536e-05, "loss": 2.6644, "step": 35435 }, { "epoch": 2.19976410702092, "grad_norm": 0.4892775784761803, "learning_rate": 2.0153337265505022e-05, "loss": 2.6227, "step": 35436 }, { "epoch": 2.1998261841206777, "grad_norm": 0.49868038501915724, "learning_rate": 2.0150439850378795e-05, "loss": 2.7648, "step": 35437 }, { "epoch": 2.199888261220436, "grad_norm": 0.48847171883238816, "learning_rate": 2.014754259098497e-05, "loss": 2.6169, "step": 35438 }, { "epoch": 2.1999503383201935, "grad_norm": 0.5184432375120763, "learning_rate": 2.0144645487338658e-05, "loss": 2.6945, "step": 35439 }, { "epoch": 2.2000124154199514, "grad_norm": 0.5055538418354197, "learning_rate": 2.0141748539454964e-05, "loss": 2.7417, "step": 35440 }, { "epoch": 2.2000744925197093, "grad_norm": 0.4921467022411433, "learning_rate": 2.0138851747349015e-05, "loss": 2.6664, "step": 35441 }, { "epoch": 2.2001365696194672, "grad_norm": 0.4933461221468892, "learning_rate": 2.013595511103594e-05, "loss": 2.7341, "step": 35442 }, { "epoch": 2.200198646719225, "grad_norm": 0.4736117643418318, "learning_rate": 2.0133058630530842e-05, "loss": 2.6414, "step": 35443 }, { "epoch": 2.200260723818983, "grad_norm": 0.5071259773062564, "learning_rate": 2.0130162305848827e-05, "loss": 2.5959, "step": 35444 }, { "epoch": 2.200322800918741, "grad_norm": 0.4627833342651112, "learning_rate": 2.0127266137004984e-05, "loss": 2.6803, "step": 35445 }, { "epoch": 2.200384878018499, "grad_norm": 0.50992686911385, "learning_rate": 2.0124370124014464e-05, "loss": 2.7164, "step": 35446 }, { "epoch": 2.200446955118257, "grad_norm": 0.4685109800662218, "learning_rate": 2.012147426689236e-05, "loss": 2.6766, "step": 35447 }, { "epoch": 2.2005090322180147, "grad_norm": 0.5147428480265633, "learning_rate": 2.011857856565378e-05, "loss": 2.6352, "step": 35448 }, { "epoch": 2.2005711093177727, "grad_norm": 0.5077492601191109, "learning_rate": 2.0115683020313806e-05, "loss": 2.7309, "step": 35449 }, { "epoch": 2.2006331864175306, "grad_norm": 0.4719330638252134, "learning_rate": 2.011278763088759e-05, "loss": 2.6739, "step": 35450 }, { "epoch": 2.2006952635172885, "grad_norm": 0.5618011849176657, "learning_rate": 2.0109892397390217e-05, "loss": 2.7028, "step": 35451 }, { "epoch": 2.2007573406170464, "grad_norm": 0.4766883527578807, "learning_rate": 2.010699731983679e-05, "loss": 2.7404, "step": 35452 }, { "epoch": 2.2008194177168043, "grad_norm": 0.46058536268419514, "learning_rate": 2.0104102398242418e-05, "loss": 2.7056, "step": 35453 }, { "epoch": 2.2008814948165623, "grad_norm": 0.5218462313643244, "learning_rate": 2.0101207632622182e-05, "loss": 2.6917, "step": 35454 }, { "epoch": 2.20094357191632, "grad_norm": 0.501267852275035, "learning_rate": 2.0098313022991218e-05, "loss": 2.7426, "step": 35455 }, { "epoch": 2.201005649016078, "grad_norm": 0.473988020319995, "learning_rate": 2.0095418569364617e-05, "loss": 2.6682, "step": 35456 }, { "epoch": 2.201067726115836, "grad_norm": 0.5258795012999332, "learning_rate": 2.0092524271757474e-05, "loss": 2.6647, "step": 35457 }, { "epoch": 2.201129803215594, "grad_norm": 0.4864859573831058, "learning_rate": 2.0089630130184878e-05, "loss": 2.7498, "step": 35458 }, { "epoch": 2.201191880315352, "grad_norm": 0.4811068228977003, "learning_rate": 2.0086736144661962e-05, "loss": 2.6139, "step": 35459 }, { "epoch": 2.2012539574151098, "grad_norm": 0.5227764282277834, "learning_rate": 2.00838423152038e-05, "loss": 2.6949, "step": 35460 }, { "epoch": 2.2013160345148677, "grad_norm": 0.5041023582502467, "learning_rate": 2.0080948641825498e-05, "loss": 2.6222, "step": 35461 }, { "epoch": 2.201378111614625, "grad_norm": 0.5131223807227393, "learning_rate": 2.0078055124542128e-05, "loss": 2.7058, "step": 35462 }, { "epoch": 2.201440188714383, "grad_norm": 0.44979378757032723, "learning_rate": 2.0075161763368834e-05, "loss": 2.6756, "step": 35463 }, { "epoch": 2.201502265814141, "grad_norm": 0.49502336865952873, "learning_rate": 2.007226855832068e-05, "loss": 2.6787, "step": 35464 }, { "epoch": 2.201564342913899, "grad_norm": 0.4750453532049984, "learning_rate": 2.0069375509412763e-05, "loss": 2.6436, "step": 35465 }, { "epoch": 2.201626420013657, "grad_norm": 0.5204683614527255, "learning_rate": 2.0066482616660188e-05, "loss": 2.7481, "step": 35466 }, { "epoch": 2.2016884971134147, "grad_norm": 0.45694025878219535, "learning_rate": 2.0063589880078015e-05, "loss": 2.5959, "step": 35467 }, { "epoch": 2.2017505742131727, "grad_norm": 0.49920207668871247, "learning_rate": 2.006069729968138e-05, "loss": 2.6711, "step": 35468 }, { "epoch": 2.2018126513129306, "grad_norm": 0.4679728105015433, "learning_rate": 2.0057804875485357e-05, "loss": 2.6655, "step": 35469 }, { "epoch": 2.2018747284126885, "grad_norm": 0.4896083497358842, "learning_rate": 2.005491260750504e-05, "loss": 2.7063, "step": 35470 }, { "epoch": 2.2019368055124464, "grad_norm": 0.5294674531625481, "learning_rate": 2.0052020495755487e-05, "loss": 2.7402, "step": 35471 }, { "epoch": 2.2019988826122043, "grad_norm": 0.4971685705891258, "learning_rate": 2.004912854025183e-05, "loss": 2.6786, "step": 35472 }, { "epoch": 2.2020609597119623, "grad_norm": 0.497091566429587, "learning_rate": 2.0046236741009135e-05, "loss": 2.6362, "step": 35473 }, { "epoch": 2.20212303681172, "grad_norm": 0.4943205765862489, "learning_rate": 2.00433450980425e-05, "loss": 2.771, "step": 35474 }, { "epoch": 2.202185113911478, "grad_norm": 0.48444444404677633, "learning_rate": 2.004045361136701e-05, "loss": 2.6031, "step": 35475 }, { "epoch": 2.202247191011236, "grad_norm": 0.512181109562517, "learning_rate": 2.0037562280997744e-05, "loss": 2.6458, "step": 35476 }, { "epoch": 2.202309268110994, "grad_norm": 0.45956753958122093, "learning_rate": 2.0034671106949775e-05, "loss": 2.7295, "step": 35477 }, { "epoch": 2.202371345210752, "grad_norm": 0.5330825501655303, "learning_rate": 2.0031780089238216e-05, "loss": 2.738, "step": 35478 }, { "epoch": 2.2024334223105098, "grad_norm": 0.48086017078671583, "learning_rate": 2.002888922787814e-05, "loss": 2.6895, "step": 35479 }, { "epoch": 2.2024954994102677, "grad_norm": 0.49896586285712263, "learning_rate": 2.002599852288462e-05, "loss": 2.6257, "step": 35480 }, { "epoch": 2.2025575765100256, "grad_norm": 0.46332275564940406, "learning_rate": 2.0023107974272726e-05, "loss": 2.6296, "step": 35481 }, { "epoch": 2.2026196536097835, "grad_norm": 0.542947343363148, "learning_rate": 2.002021758205757e-05, "loss": 2.6815, "step": 35482 }, { "epoch": 2.2026817307095414, "grad_norm": 0.5069311857304106, "learning_rate": 2.001732734625422e-05, "loss": 2.7438, "step": 35483 }, { "epoch": 2.2027438078092993, "grad_norm": 0.5247528005091799, "learning_rate": 2.0014437266877754e-05, "loss": 2.6912, "step": 35484 }, { "epoch": 2.202805884909057, "grad_norm": 0.4637502422486364, "learning_rate": 2.0011547343943225e-05, "loss": 2.7065, "step": 35485 }, { "epoch": 2.202867962008815, "grad_norm": 0.5138419852316658, "learning_rate": 2.0008657577465755e-05, "loss": 2.6937, "step": 35486 }, { "epoch": 2.2029300391085727, "grad_norm": 0.5500559258204127, "learning_rate": 2.00057679674604e-05, "loss": 2.6966, "step": 35487 }, { "epoch": 2.2029921162083306, "grad_norm": 0.527589189844681, "learning_rate": 2.000287851394223e-05, "loss": 2.666, "step": 35488 }, { "epoch": 2.2030541933080885, "grad_norm": 0.5835477032043347, "learning_rate": 1.999998921692633e-05, "loss": 2.689, "step": 35489 }, { "epoch": 2.2031162704078464, "grad_norm": 0.5254225597106572, "learning_rate": 1.9997100076427744e-05, "loss": 2.6353, "step": 35490 }, { "epoch": 2.2031783475076043, "grad_norm": 0.5588089855046456, "learning_rate": 1.9994211092461594e-05, "loss": 2.7992, "step": 35491 }, { "epoch": 2.2032404246073622, "grad_norm": 0.5219644344242556, "learning_rate": 1.9991322265042928e-05, "loss": 2.7357, "step": 35492 }, { "epoch": 2.20330250170712, "grad_norm": 0.5134494430443873, "learning_rate": 1.9988433594186816e-05, "loss": 2.7526, "step": 35493 }, { "epoch": 2.203364578806878, "grad_norm": 0.5602880006402948, "learning_rate": 1.998554507990832e-05, "loss": 2.7286, "step": 35494 }, { "epoch": 2.203426655906636, "grad_norm": 0.5946012718882528, "learning_rate": 1.998265672222253e-05, "loss": 2.775, "step": 35495 }, { "epoch": 2.203488733006394, "grad_norm": 0.503298755520139, "learning_rate": 1.9979768521144516e-05, "loss": 2.6854, "step": 35496 }, { "epoch": 2.203550810106152, "grad_norm": 0.5052122594412087, "learning_rate": 1.9976880476689337e-05, "loss": 2.6957, "step": 35497 }, { "epoch": 2.2036128872059098, "grad_norm": 0.5473268005029236, "learning_rate": 1.9973992588872037e-05, "loss": 2.7213, "step": 35498 }, { "epoch": 2.2036749643056677, "grad_norm": 0.5006740616035553, "learning_rate": 1.9971104857707724e-05, "loss": 2.694, "step": 35499 }, { "epoch": 2.2037370414054256, "grad_norm": 0.4818764356054946, "learning_rate": 1.996821728321145e-05, "loss": 2.6902, "step": 35500 }, { "epoch": 2.2037991185051835, "grad_norm": 0.5202031201014946, "learning_rate": 1.996532986539828e-05, "loss": 2.6753, "step": 35501 }, { "epoch": 2.2038611956049414, "grad_norm": 0.5100136515248832, "learning_rate": 1.996244260428327e-05, "loss": 2.6312, "step": 35502 }, { "epoch": 2.2039232727046993, "grad_norm": 0.5243858741842169, "learning_rate": 1.9959555499881472e-05, "loss": 2.6768, "step": 35503 }, { "epoch": 2.2039853498044573, "grad_norm": 0.46929988397173744, "learning_rate": 1.9956668552207986e-05, "loss": 2.6786, "step": 35504 }, { "epoch": 2.204047426904215, "grad_norm": 0.5057618625284901, "learning_rate": 1.9953781761277834e-05, "loss": 2.6568, "step": 35505 }, { "epoch": 2.204109504003973, "grad_norm": 0.4773799108366028, "learning_rate": 1.9950895127106112e-05, "loss": 2.6857, "step": 35506 }, { "epoch": 2.204171581103731, "grad_norm": 0.46972510323724087, "learning_rate": 1.9948008649707865e-05, "loss": 2.6951, "step": 35507 }, { "epoch": 2.204233658203489, "grad_norm": 0.49847162123934086, "learning_rate": 1.994512232909814e-05, "loss": 2.6348, "step": 35508 }, { "epoch": 2.204295735303247, "grad_norm": 0.41282593320464905, "learning_rate": 1.994223616529202e-05, "loss": 2.6222, "step": 35509 }, { "epoch": 2.2043578124030043, "grad_norm": 0.4647560678058245, "learning_rate": 1.9939350158304553e-05, "loss": 2.6612, "step": 35510 }, { "epoch": 2.2044198895027622, "grad_norm": 0.4611407171441952, "learning_rate": 1.9936464308150792e-05, "loss": 2.6801, "step": 35511 }, { "epoch": 2.20448196660252, "grad_norm": 0.5152889948678196, "learning_rate": 1.993357861484578e-05, "loss": 2.7557, "step": 35512 }, { "epoch": 2.204544043702278, "grad_norm": 0.4879247508395943, "learning_rate": 1.9930693078404605e-05, "loss": 2.7265, "step": 35513 }, { "epoch": 2.204606120802036, "grad_norm": 0.4952977620528471, "learning_rate": 1.9927807698842305e-05, "loss": 2.6492, "step": 35514 }, { "epoch": 2.204668197901794, "grad_norm": 0.49907498923901084, "learning_rate": 1.992492247617393e-05, "loss": 2.6878, "step": 35515 }, { "epoch": 2.204730275001552, "grad_norm": 0.4984025919059648, "learning_rate": 1.9922037410414535e-05, "loss": 2.7061, "step": 35516 }, { "epoch": 2.2047923521013097, "grad_norm": 0.4929519348178529, "learning_rate": 1.991915250157915e-05, "loss": 2.6699, "step": 35517 }, { "epoch": 2.2048544292010677, "grad_norm": 0.5046696166658206, "learning_rate": 1.9916267749682875e-05, "loss": 2.6809, "step": 35518 }, { "epoch": 2.2049165063008256, "grad_norm": 0.5135835927081295, "learning_rate": 1.9913383154740734e-05, "loss": 2.6712, "step": 35519 }, { "epoch": 2.2049785834005835, "grad_norm": 0.48087123357310735, "learning_rate": 1.9910498716767774e-05, "loss": 2.7323, "step": 35520 }, { "epoch": 2.2050406605003414, "grad_norm": 0.4863700542223002, "learning_rate": 1.9907614435779028e-05, "loss": 2.7609, "step": 35521 }, { "epoch": 2.2051027376000993, "grad_norm": 0.5112621803977901, "learning_rate": 1.9904730311789577e-05, "loss": 2.7122, "step": 35522 }, { "epoch": 2.2051648146998573, "grad_norm": 0.48113671059255425, "learning_rate": 1.990184634481446e-05, "loss": 2.7254, "step": 35523 }, { "epoch": 2.205226891799615, "grad_norm": 0.47772890151749037, "learning_rate": 1.9898962534868716e-05, "loss": 2.7079, "step": 35524 }, { "epoch": 2.205288968899373, "grad_norm": 0.4713873579306622, "learning_rate": 1.989607888196739e-05, "loss": 2.725, "step": 35525 }, { "epoch": 2.205351045999131, "grad_norm": 0.4858002612988008, "learning_rate": 1.9893195386125513e-05, "loss": 2.6876, "step": 35526 }, { "epoch": 2.205413123098889, "grad_norm": 0.45704277545709715, "learning_rate": 1.9890312047358162e-05, "loss": 2.7101, "step": 35527 }, { "epoch": 2.205475200198647, "grad_norm": 0.43945256551071155, "learning_rate": 1.988742886568036e-05, "loss": 2.6648, "step": 35528 }, { "epoch": 2.2055372772984048, "grad_norm": 0.4938749420853419, "learning_rate": 1.988454584110715e-05, "loss": 2.7063, "step": 35529 }, { "epoch": 2.2055993543981627, "grad_norm": 0.45534816801218486, "learning_rate": 1.9881662973653557e-05, "loss": 2.7492, "step": 35530 }, { "epoch": 2.2056614314979206, "grad_norm": 0.4879177571579594, "learning_rate": 1.9878780263334668e-05, "loss": 2.6289, "step": 35531 }, { "epoch": 2.2057235085976785, "grad_norm": 0.513226856618516, "learning_rate": 1.9875897710165487e-05, "loss": 2.7071, "step": 35532 }, { "epoch": 2.205785585697436, "grad_norm": 0.4793524761951298, "learning_rate": 1.987301531416106e-05, "loss": 2.6617, "step": 35533 }, { "epoch": 2.205847662797194, "grad_norm": 0.5428383485374799, "learning_rate": 1.9870133075336406e-05, "loss": 2.6818, "step": 35534 }, { "epoch": 2.205909739896952, "grad_norm": 0.47164243941995143, "learning_rate": 1.9867250993706604e-05, "loss": 2.6613, "step": 35535 }, { "epoch": 2.2059718169967097, "grad_norm": 0.5139123332057054, "learning_rate": 1.9864369069286666e-05, "loss": 2.727, "step": 35536 }, { "epoch": 2.2060338940964677, "grad_norm": 0.5186225861028662, "learning_rate": 1.986148730209163e-05, "loss": 2.6331, "step": 35537 }, { "epoch": 2.2060959711962256, "grad_norm": 0.45814567394368594, "learning_rate": 1.9858605692136517e-05, "loss": 2.6227, "step": 35538 }, { "epoch": 2.2061580482959835, "grad_norm": 0.5152670745075004, "learning_rate": 1.985572423943639e-05, "loss": 2.6693, "step": 35539 }, { "epoch": 2.2062201253957414, "grad_norm": 0.5224426266322514, "learning_rate": 1.9852842944006252e-05, "loss": 2.7142, "step": 35540 }, { "epoch": 2.2062822024954993, "grad_norm": 0.4966072968813753, "learning_rate": 1.9849961805861167e-05, "loss": 2.6785, "step": 35541 }, { "epoch": 2.2063442795952573, "grad_norm": 0.5718039254802656, "learning_rate": 1.9847080825016146e-05, "loss": 2.6967, "step": 35542 }, { "epoch": 2.206406356695015, "grad_norm": 0.477925025658423, "learning_rate": 1.984420000148623e-05, "loss": 2.6605, "step": 35543 }, { "epoch": 2.206468433794773, "grad_norm": 0.5731208321888844, "learning_rate": 1.9841319335286423e-05, "loss": 2.6408, "step": 35544 }, { "epoch": 2.206530510894531, "grad_norm": 0.5623582025458338, "learning_rate": 1.983843882643179e-05, "loss": 2.7111, "step": 35545 }, { "epoch": 2.206592587994289, "grad_norm": 0.4742094796294267, "learning_rate": 1.983555847493735e-05, "loss": 2.7356, "step": 35546 }, { "epoch": 2.206654665094047, "grad_norm": 0.5840001183878766, "learning_rate": 1.9832678280818124e-05, "loss": 2.783, "step": 35547 }, { "epoch": 2.2067167421938048, "grad_norm": 0.5199267380714118, "learning_rate": 1.9829798244089114e-05, "loss": 2.6977, "step": 35548 }, { "epoch": 2.2067788192935627, "grad_norm": 0.5423417072702776, "learning_rate": 1.9826918364765395e-05, "loss": 2.6726, "step": 35549 }, { "epoch": 2.2068408963933206, "grad_norm": 0.495653693197193, "learning_rate": 1.9824038642861963e-05, "loss": 2.5942, "step": 35550 }, { "epoch": 2.2069029734930785, "grad_norm": 0.4877761975403757, "learning_rate": 1.982115907839385e-05, "loss": 2.6655, "step": 35551 }, { "epoch": 2.2069650505928364, "grad_norm": 0.5156933710256215, "learning_rate": 1.9818279671376073e-05, "loss": 2.7721, "step": 35552 }, { "epoch": 2.2070271276925943, "grad_norm": 0.5024954096741039, "learning_rate": 1.9815400421823644e-05, "loss": 2.6604, "step": 35553 }, { "epoch": 2.2070892047923523, "grad_norm": 0.5182536197688675, "learning_rate": 1.9812521329751615e-05, "loss": 2.6814, "step": 35554 }, { "epoch": 2.20715128189211, "grad_norm": 0.47812342238321764, "learning_rate": 1.9809642395174987e-05, "loss": 2.7124, "step": 35555 }, { "epoch": 2.207213358991868, "grad_norm": 0.4755507344527515, "learning_rate": 1.980676361810879e-05, "loss": 2.7697, "step": 35556 }, { "epoch": 2.207275436091626, "grad_norm": 0.5518611574867023, "learning_rate": 1.980388499856801e-05, "loss": 2.6122, "step": 35557 }, { "epoch": 2.2073375131913835, "grad_norm": 0.48239231405153066, "learning_rate": 1.9801006536567717e-05, "loss": 2.7376, "step": 35558 }, { "epoch": 2.2073995902911414, "grad_norm": 0.516209184728065, "learning_rate": 1.9798128232122904e-05, "loss": 2.7384, "step": 35559 }, { "epoch": 2.2074616673908993, "grad_norm": 0.49897274648227363, "learning_rate": 1.9795250085248584e-05, "loss": 2.6154, "step": 35560 }, { "epoch": 2.2075237444906572, "grad_norm": 0.5162556288645532, "learning_rate": 1.9792372095959784e-05, "loss": 2.7245, "step": 35561 }, { "epoch": 2.207585821590415, "grad_norm": 0.48373975937622565, "learning_rate": 1.9789494264271485e-05, "loss": 2.7142, "step": 35562 }, { "epoch": 2.207647898690173, "grad_norm": 0.5539636590258575, "learning_rate": 1.9786616590198753e-05, "loss": 2.7174, "step": 35563 }, { "epoch": 2.207709975789931, "grad_norm": 0.50143682147126, "learning_rate": 1.9783739073756575e-05, "loss": 2.6194, "step": 35564 }, { "epoch": 2.207772052889689, "grad_norm": 0.5508319750214573, "learning_rate": 1.9780861714959964e-05, "loss": 2.6882, "step": 35565 }, { "epoch": 2.207834129989447, "grad_norm": 0.4904117643240398, "learning_rate": 1.9777984513823917e-05, "loss": 2.7246, "step": 35566 }, { "epoch": 2.2078962070892048, "grad_norm": 0.5277110198392178, "learning_rate": 1.9775107470363473e-05, "loss": 2.6525, "step": 35567 }, { "epoch": 2.2079582841889627, "grad_norm": 0.5078677514169472, "learning_rate": 1.9772230584593638e-05, "loss": 2.6713, "step": 35568 }, { "epoch": 2.2080203612887206, "grad_norm": 0.4591716303213813, "learning_rate": 1.9769353856529415e-05, "loss": 2.5883, "step": 35569 }, { "epoch": 2.2080824383884785, "grad_norm": 0.4596723554283164, "learning_rate": 1.976647728618579e-05, "loss": 2.5863, "step": 35570 }, { "epoch": 2.2081445154882364, "grad_norm": 0.533697855946203, "learning_rate": 1.9763600873577793e-05, "loss": 2.6572, "step": 35571 }, { "epoch": 2.2082065925879943, "grad_norm": 0.4681948712539777, "learning_rate": 1.976072461872045e-05, "loss": 2.6323, "step": 35572 }, { "epoch": 2.2082686696877523, "grad_norm": 0.4667966675059604, "learning_rate": 1.9757848521628748e-05, "loss": 2.7342, "step": 35573 }, { "epoch": 2.20833074678751, "grad_norm": 0.48786988286807653, "learning_rate": 1.975497258231769e-05, "loss": 2.6398, "step": 35574 }, { "epoch": 2.208392823887268, "grad_norm": 0.48978862786928995, "learning_rate": 1.9752096800802278e-05, "loss": 2.7057, "step": 35575 }, { "epoch": 2.208454900987026, "grad_norm": 0.4525018880077618, "learning_rate": 1.9749221177097504e-05, "loss": 2.608, "step": 35576 }, { "epoch": 2.208516978086784, "grad_norm": 0.4926239228342084, "learning_rate": 1.974634571121841e-05, "loss": 2.7462, "step": 35577 }, { "epoch": 2.208579055186542, "grad_norm": 0.45086866156044, "learning_rate": 1.974347040317997e-05, "loss": 2.7343, "step": 35578 }, { "epoch": 2.2086411322862998, "grad_norm": 0.5365892701088888, "learning_rate": 1.974059525299719e-05, "loss": 2.7418, "step": 35579 }, { "epoch": 2.2087032093860577, "grad_norm": 0.49609378003698545, "learning_rate": 1.9737720260685056e-05, "loss": 2.6247, "step": 35580 }, { "epoch": 2.208765286485815, "grad_norm": 0.4990716601100212, "learning_rate": 1.9734845426258596e-05, "loss": 2.6713, "step": 35581 }, { "epoch": 2.208827363585573, "grad_norm": 0.5244553192788313, "learning_rate": 1.9731970749732797e-05, "loss": 2.693, "step": 35582 }, { "epoch": 2.208889440685331, "grad_norm": 0.4961409508785732, "learning_rate": 1.9729096231122656e-05, "loss": 2.744, "step": 35583 }, { "epoch": 2.208951517785089, "grad_norm": 0.5274565682000335, "learning_rate": 1.9726221870443144e-05, "loss": 2.6543, "step": 35584 }, { "epoch": 2.209013594884847, "grad_norm": 0.5119651026716637, "learning_rate": 1.9723347667709306e-05, "loss": 2.734, "step": 35585 }, { "epoch": 2.2090756719846047, "grad_norm": 0.510056482113859, "learning_rate": 1.9720473622936115e-05, "loss": 2.6822, "step": 35586 }, { "epoch": 2.2091377490843627, "grad_norm": 0.5472898544622282, "learning_rate": 1.971759973613856e-05, "loss": 2.6731, "step": 35587 }, { "epoch": 2.2091998261841206, "grad_norm": 0.4737188202151864, "learning_rate": 1.9714726007331637e-05, "loss": 2.755, "step": 35588 }, { "epoch": 2.2092619032838785, "grad_norm": 0.5267247558450511, "learning_rate": 1.971185243653032e-05, "loss": 2.6436, "step": 35589 }, { "epoch": 2.2093239803836364, "grad_norm": 0.5099958871694293, "learning_rate": 1.9708979023749645e-05, "loss": 2.7031, "step": 35590 }, { "epoch": 2.2093860574833943, "grad_norm": 0.5063720814948727, "learning_rate": 1.970610576900458e-05, "loss": 2.6913, "step": 35591 }, { "epoch": 2.2094481345831523, "grad_norm": 0.4539329954268121, "learning_rate": 1.9703232672310114e-05, "loss": 2.6398, "step": 35592 }, { "epoch": 2.20951021168291, "grad_norm": 0.4843545263331922, "learning_rate": 1.9700359733681216e-05, "loss": 2.5957, "step": 35593 }, { "epoch": 2.209572288782668, "grad_norm": 0.4916123102200693, "learning_rate": 1.9697486953132922e-05, "loss": 2.6466, "step": 35594 }, { "epoch": 2.209634365882426, "grad_norm": 0.4666934038347951, "learning_rate": 1.969461433068019e-05, "loss": 2.7643, "step": 35595 }, { "epoch": 2.209696442982184, "grad_norm": 0.5000128744375203, "learning_rate": 1.969174186633801e-05, "loss": 2.727, "step": 35596 }, { "epoch": 2.209758520081942, "grad_norm": 0.5432604548171976, "learning_rate": 1.9688869560121376e-05, "loss": 2.7658, "step": 35597 }, { "epoch": 2.2098205971816998, "grad_norm": 0.5057512558088489, "learning_rate": 1.968599741204524e-05, "loss": 2.6754, "step": 35598 }, { "epoch": 2.2098826742814577, "grad_norm": 0.5051827342308732, "learning_rate": 1.9683125422124642e-05, "loss": 2.6849, "step": 35599 }, { "epoch": 2.2099447513812156, "grad_norm": 0.5363405308188149, "learning_rate": 1.9680253590374537e-05, "loss": 2.6319, "step": 35600 }, { "epoch": 2.2100068284809735, "grad_norm": 0.4947826602183016, "learning_rate": 1.9677381916809912e-05, "loss": 2.6804, "step": 35601 }, { "epoch": 2.2100689055807314, "grad_norm": 0.5412591048921881, "learning_rate": 1.9674510401445722e-05, "loss": 2.6507, "step": 35602 }, { "epoch": 2.2101309826804894, "grad_norm": 0.5166547203477405, "learning_rate": 1.967163904429699e-05, "loss": 2.765, "step": 35603 }, { "epoch": 2.2101930597802473, "grad_norm": 0.5077309469614144, "learning_rate": 1.9668767845378672e-05, "loss": 2.7333, "step": 35604 }, { "epoch": 2.210255136880005, "grad_norm": 0.5397447884552303, "learning_rate": 1.9665896804705762e-05, "loss": 2.7581, "step": 35605 }, { "epoch": 2.2103172139797627, "grad_norm": 0.4751756531536747, "learning_rate": 1.9663025922293237e-05, "loss": 2.696, "step": 35606 }, { "epoch": 2.2103792910795206, "grad_norm": 0.5357735129949445, "learning_rate": 1.9660155198156055e-05, "loss": 2.6686, "step": 35607 }, { "epoch": 2.2104413681792785, "grad_norm": 0.4751343198392581, "learning_rate": 1.965728463230922e-05, "loss": 2.6693, "step": 35608 }, { "epoch": 2.2105034452790364, "grad_norm": 0.47144493437418483, "learning_rate": 1.9654414224767702e-05, "loss": 2.5928, "step": 35609 }, { "epoch": 2.2105655223787943, "grad_norm": 0.49406663242665194, "learning_rate": 1.965154397554647e-05, "loss": 2.706, "step": 35610 }, { "epoch": 2.2106275994785523, "grad_norm": 0.4998952785737137, "learning_rate": 1.9648673884660502e-05, "loss": 2.7058, "step": 35611 }, { "epoch": 2.21068967657831, "grad_norm": 0.5661209703429626, "learning_rate": 1.9645803952124754e-05, "loss": 2.7701, "step": 35612 }, { "epoch": 2.210751753678068, "grad_norm": 0.4912512063115963, "learning_rate": 1.9642934177954227e-05, "loss": 2.6715, "step": 35613 }, { "epoch": 2.210813830777826, "grad_norm": 0.5214485984845594, "learning_rate": 1.9640064562163885e-05, "loss": 2.6985, "step": 35614 }, { "epoch": 2.210875907877584, "grad_norm": 0.5466425537930205, "learning_rate": 1.9637195104768697e-05, "loss": 2.748, "step": 35615 }, { "epoch": 2.210937984977342, "grad_norm": 0.4972096755562881, "learning_rate": 1.9634325805783616e-05, "loss": 2.6206, "step": 35616 }, { "epoch": 2.2110000620770998, "grad_norm": 0.5464992594788081, "learning_rate": 1.9631456665223647e-05, "loss": 2.7254, "step": 35617 }, { "epoch": 2.2110621391768577, "grad_norm": 0.5053842736336169, "learning_rate": 1.962858768310374e-05, "loss": 2.6596, "step": 35618 }, { "epoch": 2.2111242162766156, "grad_norm": 0.5047491254473021, "learning_rate": 1.9625718859438863e-05, "loss": 2.703, "step": 35619 }, { "epoch": 2.2111862933763735, "grad_norm": 0.5211204945165228, "learning_rate": 1.9622850194243965e-05, "loss": 2.7218, "step": 35620 }, { "epoch": 2.2112483704761314, "grad_norm": 0.47984034063481373, "learning_rate": 1.961998168753405e-05, "loss": 2.7187, "step": 35621 }, { "epoch": 2.2113104475758893, "grad_norm": 0.513332907807599, "learning_rate": 1.9617113339324066e-05, "loss": 2.7631, "step": 35622 }, { "epoch": 2.2113725246756473, "grad_norm": 0.5028543360288501, "learning_rate": 1.961424514962898e-05, "loss": 2.6969, "step": 35623 }, { "epoch": 2.211434601775405, "grad_norm": 0.44507672944937376, "learning_rate": 1.961137711846375e-05, "loss": 2.6827, "step": 35624 }, { "epoch": 2.211496678875163, "grad_norm": 0.49054059499240493, "learning_rate": 1.9608509245843332e-05, "loss": 2.7391, "step": 35625 }, { "epoch": 2.211558755974921, "grad_norm": 0.5058694492353212, "learning_rate": 1.960564153178271e-05, "loss": 2.7566, "step": 35626 }, { "epoch": 2.211620833074679, "grad_norm": 0.4550150553339932, "learning_rate": 1.9602773976296834e-05, "loss": 2.6509, "step": 35627 }, { "epoch": 2.211682910174437, "grad_norm": 0.5202557381034411, "learning_rate": 1.9599906579400668e-05, "loss": 2.6742, "step": 35628 }, { "epoch": 2.2117449872741943, "grad_norm": 0.4751792594559404, "learning_rate": 1.9597039341109152e-05, "loss": 2.6437, "step": 35629 }, { "epoch": 2.2118070643739522, "grad_norm": 0.4836895769170545, "learning_rate": 1.9594172261437278e-05, "loss": 2.7541, "step": 35630 }, { "epoch": 2.21186914147371, "grad_norm": 0.4900252515504654, "learning_rate": 1.959130534039999e-05, "loss": 2.6427, "step": 35631 }, { "epoch": 2.211931218573468, "grad_norm": 0.5158277748662986, "learning_rate": 1.958843857801225e-05, "loss": 2.7185, "step": 35632 }, { "epoch": 2.211993295673226, "grad_norm": 0.5181350015427298, "learning_rate": 1.958557197428898e-05, "loss": 2.7228, "step": 35633 }, { "epoch": 2.212055372772984, "grad_norm": 0.5305435588599904, "learning_rate": 1.9582705529245188e-05, "loss": 2.747, "step": 35634 }, { "epoch": 2.212117449872742, "grad_norm": 0.47404199602398905, "learning_rate": 1.9579839242895803e-05, "loss": 2.6922, "step": 35635 }, { "epoch": 2.2121795269724998, "grad_norm": 0.5379331617635813, "learning_rate": 1.9576973115255765e-05, "loss": 2.6511, "step": 35636 }, { "epoch": 2.2122416040722577, "grad_norm": 0.5292585014143384, "learning_rate": 1.9574107146340058e-05, "loss": 2.6878, "step": 35637 }, { "epoch": 2.2123036811720156, "grad_norm": 0.5079815509817732, "learning_rate": 1.9571241336163625e-05, "loss": 2.6868, "step": 35638 }, { "epoch": 2.2123657582717735, "grad_norm": 0.47833162779564903, "learning_rate": 1.9568375684741393e-05, "loss": 2.7615, "step": 35639 }, { "epoch": 2.2124278353715314, "grad_norm": 0.498517535848203, "learning_rate": 1.9565510192088348e-05, "loss": 2.7135, "step": 35640 }, { "epoch": 2.2124899124712893, "grad_norm": 0.5302238370905745, "learning_rate": 1.9562644858219432e-05, "loss": 2.7312, "step": 35641 }, { "epoch": 2.2125519895710473, "grad_norm": 0.49657949561387876, "learning_rate": 1.9559779683149583e-05, "loss": 2.6188, "step": 35642 }, { "epoch": 2.212614066670805, "grad_norm": 0.5146504642801351, "learning_rate": 1.9556914666893733e-05, "loss": 2.7197, "step": 35643 }, { "epoch": 2.212676143770563, "grad_norm": 0.4780340145530145, "learning_rate": 1.9554049809466867e-05, "loss": 2.7456, "step": 35644 }, { "epoch": 2.212738220870321, "grad_norm": 0.4766684242474002, "learning_rate": 1.9551185110883913e-05, "loss": 2.633, "step": 35645 }, { "epoch": 2.212800297970079, "grad_norm": 0.4941852082054266, "learning_rate": 1.9548320571159824e-05, "loss": 2.7586, "step": 35646 }, { "epoch": 2.212862375069837, "grad_norm": 0.4585368307945631, "learning_rate": 1.9545456190309537e-05, "loss": 2.7048, "step": 35647 }, { "epoch": 2.2129244521695948, "grad_norm": 0.5096440955975797, "learning_rate": 1.9542591968347978e-05, "loss": 2.7263, "step": 35648 }, { "epoch": 2.2129865292693527, "grad_norm": 0.47419906273688495, "learning_rate": 1.9539727905290128e-05, "loss": 2.7237, "step": 35649 }, { "epoch": 2.2130486063691106, "grad_norm": 0.49080184373517055, "learning_rate": 1.9536864001150912e-05, "loss": 2.7383, "step": 35650 }, { "epoch": 2.2131106834688685, "grad_norm": 0.4481184545541718, "learning_rate": 1.9534000255945272e-05, "loss": 2.6597, "step": 35651 }, { "epoch": 2.2131727605686264, "grad_norm": 0.4896539086004627, "learning_rate": 1.9531136669688133e-05, "loss": 2.7423, "step": 35652 }, { "epoch": 2.2132348376683844, "grad_norm": 0.4821547877006254, "learning_rate": 1.9528273242394467e-05, "loss": 2.67, "step": 35653 }, { "epoch": 2.213296914768142, "grad_norm": 0.4712141120460283, "learning_rate": 1.9525409974079194e-05, "loss": 2.714, "step": 35654 }, { "epoch": 2.2133589918678997, "grad_norm": 0.46836229821343606, "learning_rate": 1.952254686475726e-05, "loss": 2.6838, "step": 35655 }, { "epoch": 2.2134210689676577, "grad_norm": 0.4504125075305572, "learning_rate": 1.9519683914443582e-05, "loss": 2.7246, "step": 35656 }, { "epoch": 2.2134831460674156, "grad_norm": 0.465688194876462, "learning_rate": 1.9516821123153122e-05, "loss": 2.6685, "step": 35657 }, { "epoch": 2.2135452231671735, "grad_norm": 0.44516438815175147, "learning_rate": 1.9513958490900813e-05, "loss": 2.6022, "step": 35658 }, { "epoch": 2.2136073002669314, "grad_norm": 0.5207357283370408, "learning_rate": 1.951109601770158e-05, "loss": 2.6764, "step": 35659 }, { "epoch": 2.2136693773666893, "grad_norm": 0.48220947161176775, "learning_rate": 1.950823370357036e-05, "loss": 2.6099, "step": 35660 }, { "epoch": 2.2137314544664473, "grad_norm": 0.46524661902419345, "learning_rate": 1.9505371548522072e-05, "loss": 2.6526, "step": 35661 }, { "epoch": 2.213793531566205, "grad_norm": 0.4952392879013116, "learning_rate": 1.950250955257168e-05, "loss": 2.6827, "step": 35662 }, { "epoch": 2.213855608665963, "grad_norm": 0.5226451228258817, "learning_rate": 1.94996477157341e-05, "loss": 2.6867, "step": 35663 }, { "epoch": 2.213917685765721, "grad_norm": 0.45182993404278105, "learning_rate": 1.9496786038024255e-05, "loss": 2.6372, "step": 35664 }, { "epoch": 2.213979762865479, "grad_norm": 0.5452695441610895, "learning_rate": 1.9493924519457075e-05, "loss": 2.774, "step": 35665 }, { "epoch": 2.214041839965237, "grad_norm": 0.45531737296822544, "learning_rate": 1.9491063160047508e-05, "loss": 2.6074, "step": 35666 }, { "epoch": 2.2141039170649948, "grad_norm": 0.45333180147968727, "learning_rate": 1.948820195981047e-05, "loss": 2.657, "step": 35667 }, { "epoch": 2.2141659941647527, "grad_norm": 0.49235492462386965, "learning_rate": 1.948534091876089e-05, "loss": 2.7162, "step": 35668 }, { "epoch": 2.2142280712645106, "grad_norm": 0.45885113344550293, "learning_rate": 1.948248003691368e-05, "loss": 2.739, "step": 35669 }, { "epoch": 2.2142901483642685, "grad_norm": 0.48617994203392423, "learning_rate": 1.9479619314283777e-05, "loss": 2.6349, "step": 35670 }, { "epoch": 2.2143522254640264, "grad_norm": 0.5122828395662701, "learning_rate": 1.9476758750886127e-05, "loss": 2.7152, "step": 35671 }, { "epoch": 2.2144143025637844, "grad_norm": 0.4615012132457189, "learning_rate": 1.9473898346735635e-05, "loss": 2.6351, "step": 35672 }, { "epoch": 2.2144763796635423, "grad_norm": 0.5330986507473832, "learning_rate": 1.9471038101847232e-05, "loss": 2.6701, "step": 35673 }, { "epoch": 2.2145384567633, "grad_norm": 0.4941173442437609, "learning_rate": 1.9468178016235827e-05, "loss": 2.7213, "step": 35674 }, { "epoch": 2.214600533863058, "grad_norm": 0.4553746906607219, "learning_rate": 1.9465318089916336e-05, "loss": 2.6379, "step": 35675 }, { "epoch": 2.214662610962816, "grad_norm": 0.504769022826387, "learning_rate": 1.946245832290371e-05, "loss": 2.6917, "step": 35676 }, { "epoch": 2.2147246880625735, "grad_norm": 0.4927470946707733, "learning_rate": 1.9459598715212852e-05, "loss": 2.7165, "step": 35677 }, { "epoch": 2.2147867651623314, "grad_norm": 0.5013534882846884, "learning_rate": 1.945673926685868e-05, "loss": 2.7033, "step": 35678 }, { "epoch": 2.2148488422620893, "grad_norm": 0.48564872624825145, "learning_rate": 1.9453879977856098e-05, "loss": 2.757, "step": 35679 }, { "epoch": 2.2149109193618473, "grad_norm": 0.4902860150742394, "learning_rate": 1.945102084822006e-05, "loss": 2.7166, "step": 35680 }, { "epoch": 2.214972996461605, "grad_norm": 0.49035498666581967, "learning_rate": 1.9448161877965453e-05, "loss": 2.6378, "step": 35681 }, { "epoch": 2.215035073561363, "grad_norm": 0.48997036029252533, "learning_rate": 1.944530306710721e-05, "loss": 2.6081, "step": 35682 }, { "epoch": 2.215097150661121, "grad_norm": 0.4691447344368912, "learning_rate": 1.9442444415660215e-05, "loss": 2.7376, "step": 35683 }, { "epoch": 2.215159227760879, "grad_norm": 0.466271364014048, "learning_rate": 1.9439585923639432e-05, "loss": 2.6529, "step": 35684 }, { "epoch": 2.215221304860637, "grad_norm": 0.4587737878665873, "learning_rate": 1.943672759105974e-05, "loss": 2.6924, "step": 35685 }, { "epoch": 2.2152833819603948, "grad_norm": 0.5167668444153952, "learning_rate": 1.9433869417936064e-05, "loss": 2.6028, "step": 35686 }, { "epoch": 2.2153454590601527, "grad_norm": 0.47666254713150125, "learning_rate": 1.943101140428331e-05, "loss": 2.7078, "step": 35687 }, { "epoch": 2.2154075361599106, "grad_norm": 0.47692782048263593, "learning_rate": 1.9428153550116372e-05, "loss": 2.6872, "step": 35688 }, { "epoch": 2.2154696132596685, "grad_norm": 0.49309456212780545, "learning_rate": 1.94252958554502e-05, "loss": 2.697, "step": 35689 }, { "epoch": 2.2155316903594264, "grad_norm": 0.5506452839002295, "learning_rate": 1.9422438320299674e-05, "loss": 2.6989, "step": 35690 }, { "epoch": 2.2155937674591843, "grad_norm": 0.5049507493845777, "learning_rate": 1.9419580944679717e-05, "loss": 2.6999, "step": 35691 }, { "epoch": 2.2156558445589423, "grad_norm": 0.518820570546235, "learning_rate": 1.9416723728605208e-05, "loss": 2.713, "step": 35692 }, { "epoch": 2.2157179216587, "grad_norm": 0.5535964982117719, "learning_rate": 1.9413866672091097e-05, "loss": 2.7316, "step": 35693 }, { "epoch": 2.215779998758458, "grad_norm": 0.4617201416201202, "learning_rate": 1.9411009775152267e-05, "loss": 2.6384, "step": 35694 }, { "epoch": 2.215842075858216, "grad_norm": 0.5371897134518916, "learning_rate": 1.9408153037803622e-05, "loss": 2.7811, "step": 35695 }, { "epoch": 2.215904152957974, "grad_norm": 0.5163789496657449, "learning_rate": 1.9405296460060075e-05, "loss": 2.7122, "step": 35696 }, { "epoch": 2.215966230057732, "grad_norm": 0.48898636831696113, "learning_rate": 1.9402440041936498e-05, "loss": 2.6569, "step": 35697 }, { "epoch": 2.2160283071574898, "grad_norm": 0.506988531820019, "learning_rate": 1.9399583783447844e-05, "loss": 2.714, "step": 35698 }, { "epoch": 2.2160903842572477, "grad_norm": 0.4334156254733434, "learning_rate": 1.939672768460899e-05, "loss": 2.6521, "step": 35699 }, { "epoch": 2.216152461357005, "grad_norm": 0.49649698284957305, "learning_rate": 1.9393871745434833e-05, "loss": 2.6153, "step": 35700 }, { "epoch": 2.2162145384567635, "grad_norm": 0.48716775197151724, "learning_rate": 1.9391015965940258e-05, "loss": 2.6503, "step": 35701 }, { "epoch": 2.216276615556521, "grad_norm": 0.5418118838518947, "learning_rate": 1.938816034614019e-05, "loss": 2.6495, "step": 35702 }, { "epoch": 2.216338692656279, "grad_norm": 0.5234378700824397, "learning_rate": 1.938530488604953e-05, "loss": 2.6717, "step": 35703 }, { "epoch": 2.216400769756037, "grad_norm": 0.460322924027827, "learning_rate": 1.938244958568317e-05, "loss": 2.7186, "step": 35704 }, { "epoch": 2.2164628468557948, "grad_norm": 0.5416751518563038, "learning_rate": 1.9379594445056003e-05, "loss": 2.692, "step": 35705 }, { "epoch": 2.2165249239555527, "grad_norm": 0.5173114593484331, "learning_rate": 1.937673946418291e-05, "loss": 2.6939, "step": 35706 }, { "epoch": 2.2165870010553106, "grad_norm": 0.49383678820349197, "learning_rate": 1.9373884643078815e-05, "loss": 2.6843, "step": 35707 }, { "epoch": 2.2166490781550685, "grad_norm": 0.4933412743371774, "learning_rate": 1.9371029981758603e-05, "loss": 2.5498, "step": 35708 }, { "epoch": 2.2167111552548264, "grad_norm": 0.5060443908044334, "learning_rate": 1.936817548023716e-05, "loss": 2.5847, "step": 35709 }, { "epoch": 2.2167732323545843, "grad_norm": 0.5201032257784642, "learning_rate": 1.936532113852938e-05, "loss": 2.6571, "step": 35710 }, { "epoch": 2.2168353094543423, "grad_norm": 0.492523109213372, "learning_rate": 1.9362466956650144e-05, "loss": 2.7017, "step": 35711 }, { "epoch": 2.2168973865541, "grad_norm": 0.5283900583313074, "learning_rate": 1.9359612934614368e-05, "loss": 2.7377, "step": 35712 }, { "epoch": 2.216959463653858, "grad_norm": 0.47311287883314124, "learning_rate": 1.9356759072436935e-05, "loss": 2.703, "step": 35713 }, { "epoch": 2.217021540753616, "grad_norm": 0.5270693312231327, "learning_rate": 1.9353905370132725e-05, "loss": 2.7028, "step": 35714 }, { "epoch": 2.217083617853374, "grad_norm": 0.49740863370818517, "learning_rate": 1.9351051827716616e-05, "loss": 2.6974, "step": 35715 }, { "epoch": 2.217145694953132, "grad_norm": 0.47735598434528154, "learning_rate": 1.934819844520352e-05, "loss": 2.6089, "step": 35716 }, { "epoch": 2.2172077720528898, "grad_norm": 0.5323262812438546, "learning_rate": 1.9345345222608318e-05, "loss": 2.6889, "step": 35717 }, { "epoch": 2.2172698491526477, "grad_norm": 0.4796018610998111, "learning_rate": 1.9342492159945897e-05, "loss": 2.6595, "step": 35718 }, { "epoch": 2.2173319262524056, "grad_norm": 0.4946117040386205, "learning_rate": 1.9339639257231113e-05, "loss": 2.5365, "step": 35719 }, { "epoch": 2.2173940033521635, "grad_norm": 0.5648549903183874, "learning_rate": 1.9336786514478887e-05, "loss": 2.7071, "step": 35720 }, { "epoch": 2.2174560804519214, "grad_norm": 0.5401912426503871, "learning_rate": 1.93339339317041e-05, "loss": 2.6932, "step": 35721 }, { "epoch": 2.2175181575516794, "grad_norm": 0.5278476426941597, "learning_rate": 1.9331081508921616e-05, "loss": 2.7207, "step": 35722 }, { "epoch": 2.2175802346514373, "grad_norm": 0.5122765856593169, "learning_rate": 1.932822924614633e-05, "loss": 2.7064, "step": 35723 }, { "epoch": 2.217642311751195, "grad_norm": 0.5251214568107803, "learning_rate": 1.9325377143393096e-05, "loss": 2.6473, "step": 35724 }, { "epoch": 2.2177043888509527, "grad_norm": 0.5220904881909811, "learning_rate": 1.932252520067684e-05, "loss": 2.7973, "step": 35725 }, { "epoch": 2.2177664659507106, "grad_norm": 0.4841114834800275, "learning_rate": 1.9319673418012413e-05, "loss": 2.646, "step": 35726 }, { "epoch": 2.2178285430504685, "grad_norm": 0.5435528255188138, "learning_rate": 1.9316821795414697e-05, "loss": 2.6656, "step": 35727 }, { "epoch": 2.2178906201502264, "grad_norm": 0.4737044604665425, "learning_rate": 1.9313970332898558e-05, "loss": 2.6102, "step": 35728 }, { "epoch": 2.2179526972499843, "grad_norm": 0.4640230900634472, "learning_rate": 1.9311119030478896e-05, "loss": 2.6723, "step": 35729 }, { "epoch": 2.2180147743497423, "grad_norm": 0.5394494423267531, "learning_rate": 1.930826788817058e-05, "loss": 2.7136, "step": 35730 }, { "epoch": 2.2180768514495, "grad_norm": 0.45308762429093824, "learning_rate": 1.9305416905988477e-05, "loss": 2.7003, "step": 35731 }, { "epoch": 2.218138928549258, "grad_norm": 0.5302209143154614, "learning_rate": 1.930256608394747e-05, "loss": 2.7203, "step": 35732 }, { "epoch": 2.218201005649016, "grad_norm": 0.5256910351561915, "learning_rate": 1.9299715422062408e-05, "loss": 2.8033, "step": 35733 }, { "epoch": 2.218263082748774, "grad_norm": 0.4364916728023968, "learning_rate": 1.9296864920348202e-05, "loss": 2.7288, "step": 35734 }, { "epoch": 2.218325159848532, "grad_norm": 0.5323556166105263, "learning_rate": 1.9294014578819687e-05, "loss": 2.7057, "step": 35735 }, { "epoch": 2.2183872369482898, "grad_norm": 0.49860966617879005, "learning_rate": 1.9291164397491767e-05, "loss": 2.7233, "step": 35736 }, { "epoch": 2.2184493140480477, "grad_norm": 0.4943812240662115, "learning_rate": 1.92883143763793e-05, "loss": 2.5875, "step": 35737 }, { "epoch": 2.2185113911478056, "grad_norm": 0.5439766510131976, "learning_rate": 1.928546451549714e-05, "loss": 2.6603, "step": 35738 }, { "epoch": 2.2185734682475635, "grad_norm": 0.4823020755421815, "learning_rate": 1.928261481486018e-05, "loss": 2.7085, "step": 35739 }, { "epoch": 2.2186355453473214, "grad_norm": 0.5464127085268305, "learning_rate": 1.927976527448327e-05, "loss": 2.6951, "step": 35740 }, { "epoch": 2.2186976224470794, "grad_norm": 0.4862435508633771, "learning_rate": 1.9276915894381285e-05, "loss": 2.6531, "step": 35741 }, { "epoch": 2.2187596995468373, "grad_norm": 0.4959413639824278, "learning_rate": 1.9274066674569073e-05, "loss": 2.6277, "step": 35742 }, { "epoch": 2.218821776646595, "grad_norm": 0.5504471597544004, "learning_rate": 1.927121761506154e-05, "loss": 2.6087, "step": 35743 }, { "epoch": 2.218883853746353, "grad_norm": 0.4906124161971325, "learning_rate": 1.9268368715873513e-05, "loss": 2.6714, "step": 35744 }, { "epoch": 2.218945930846111, "grad_norm": 0.542301838716277, "learning_rate": 1.926551997701987e-05, "loss": 2.7144, "step": 35745 }, { "epoch": 2.219008007945869, "grad_norm": 0.533203907501429, "learning_rate": 1.9262671398515468e-05, "loss": 2.5028, "step": 35746 }, { "epoch": 2.219070085045627, "grad_norm": 0.536283378095813, "learning_rate": 1.925982298037516e-05, "loss": 2.6347, "step": 35747 }, { "epoch": 2.2191321621453843, "grad_norm": 0.5221822976036418, "learning_rate": 1.925697472261383e-05, "loss": 2.6957, "step": 35748 }, { "epoch": 2.2191942392451427, "grad_norm": 0.48205357751807704, "learning_rate": 1.925412662524633e-05, "loss": 2.6266, "step": 35749 }, { "epoch": 2.2192563163449, "grad_norm": 0.57149500599045, "learning_rate": 1.9251278688287504e-05, "loss": 2.6723, "step": 35750 }, { "epoch": 2.219318393444658, "grad_norm": 0.4913272606175905, "learning_rate": 1.9248430911752213e-05, "loss": 2.6598, "step": 35751 }, { "epoch": 2.219380470544416, "grad_norm": 0.5495007015064125, "learning_rate": 1.9245583295655344e-05, "loss": 2.6778, "step": 35752 }, { "epoch": 2.219442547644174, "grad_norm": 0.5025844537166622, "learning_rate": 1.924273584001173e-05, "loss": 2.6812, "step": 35753 }, { "epoch": 2.219504624743932, "grad_norm": 0.49717035392289494, "learning_rate": 1.9239888544836226e-05, "loss": 2.7299, "step": 35754 }, { "epoch": 2.2195667018436898, "grad_norm": 0.4971519058501003, "learning_rate": 1.9237041410143674e-05, "loss": 2.6679, "step": 35755 }, { "epoch": 2.2196287789434477, "grad_norm": 0.5435053691633901, "learning_rate": 1.9234194435948966e-05, "loss": 2.6594, "step": 35756 }, { "epoch": 2.2196908560432056, "grad_norm": 0.45283916929277807, "learning_rate": 1.9231347622266932e-05, "loss": 2.6315, "step": 35757 }, { "epoch": 2.2197529331429635, "grad_norm": 0.501231019238622, "learning_rate": 1.922850096911243e-05, "loss": 2.6682, "step": 35758 }, { "epoch": 2.2198150102427214, "grad_norm": 0.5116169952094437, "learning_rate": 1.9225654476500304e-05, "loss": 2.7278, "step": 35759 }, { "epoch": 2.2198770873424793, "grad_norm": 0.5044991309987382, "learning_rate": 1.9222808144445393e-05, "loss": 2.756, "step": 35760 }, { "epoch": 2.2199391644422373, "grad_norm": 0.49733075287015654, "learning_rate": 1.921996197296258e-05, "loss": 2.6789, "step": 35761 }, { "epoch": 2.220001241541995, "grad_norm": 0.5207088003686337, "learning_rate": 1.92171159620667e-05, "loss": 2.7146, "step": 35762 }, { "epoch": 2.220063318641753, "grad_norm": 0.5074909365835933, "learning_rate": 1.92142701117726e-05, "loss": 2.6823, "step": 35763 }, { "epoch": 2.220125395741511, "grad_norm": 0.48817444205362365, "learning_rate": 1.9211424422095108e-05, "loss": 2.7, "step": 35764 }, { "epoch": 2.220187472841269, "grad_norm": 0.5270292686608373, "learning_rate": 1.9208578893049107e-05, "loss": 2.644, "step": 35765 }, { "epoch": 2.220249549941027, "grad_norm": 0.49851938908346627, "learning_rate": 1.920573352464942e-05, "loss": 2.7316, "step": 35766 }, { "epoch": 2.2203116270407848, "grad_norm": 0.4944485394823798, "learning_rate": 1.920288831691088e-05, "loss": 2.6962, "step": 35767 }, { "epoch": 2.2203737041405427, "grad_norm": 0.605926635865137, "learning_rate": 1.9200043269848373e-05, "loss": 2.7172, "step": 35768 }, { "epoch": 2.2204357812403006, "grad_norm": 0.4629364634806495, "learning_rate": 1.919719838347669e-05, "loss": 2.7109, "step": 35769 }, { "epoch": 2.2204978583400585, "grad_norm": 0.51637286079719, "learning_rate": 1.9194353657810725e-05, "loss": 2.6896, "step": 35770 }, { "epoch": 2.2205599354398164, "grad_norm": 0.5057662819293567, "learning_rate": 1.919150909286529e-05, "loss": 2.6268, "step": 35771 }, { "epoch": 2.2206220125395744, "grad_norm": 0.5130478872791967, "learning_rate": 1.9188664688655235e-05, "loss": 2.7335, "step": 35772 }, { "epoch": 2.220684089639332, "grad_norm": 0.5378382228571532, "learning_rate": 1.9185820445195397e-05, "loss": 2.7401, "step": 35773 }, { "epoch": 2.2207461667390898, "grad_norm": 0.4710968713356476, "learning_rate": 1.9182976362500594e-05, "loss": 2.603, "step": 35774 }, { "epoch": 2.2208082438388477, "grad_norm": 0.4844101923801888, "learning_rate": 1.9180132440585702e-05, "loss": 2.69, "step": 35775 }, { "epoch": 2.2208703209386056, "grad_norm": 0.45310395290559086, "learning_rate": 1.9177288679465545e-05, "loss": 2.661, "step": 35776 }, { "epoch": 2.2209323980383635, "grad_norm": 0.47916736982819197, "learning_rate": 1.9174445079154945e-05, "loss": 2.649, "step": 35777 }, { "epoch": 2.2209944751381214, "grad_norm": 0.4888258833892477, "learning_rate": 1.9171601639668742e-05, "loss": 2.6232, "step": 35778 }, { "epoch": 2.2210565522378793, "grad_norm": 0.4927214647630692, "learning_rate": 1.9168758361021787e-05, "loss": 2.6613, "step": 35779 }, { "epoch": 2.2211186293376373, "grad_norm": 0.4653935746987491, "learning_rate": 1.916591524322891e-05, "loss": 2.6713, "step": 35780 }, { "epoch": 2.221180706437395, "grad_norm": 0.49380620564083705, "learning_rate": 1.9163072286304935e-05, "loss": 2.6653, "step": 35781 }, { "epoch": 2.221242783537153, "grad_norm": 0.4514723763733804, "learning_rate": 1.91602294902647e-05, "loss": 2.6793, "step": 35782 }, { "epoch": 2.221304860636911, "grad_norm": 0.48655405886169123, "learning_rate": 1.9157386855123012e-05, "loss": 2.5884, "step": 35783 }, { "epoch": 2.221366937736669, "grad_norm": 0.4902507885620685, "learning_rate": 1.9154544380894746e-05, "loss": 2.676, "step": 35784 }, { "epoch": 2.221429014836427, "grad_norm": 0.4850649227115746, "learning_rate": 1.9151702067594707e-05, "loss": 2.8035, "step": 35785 }, { "epoch": 2.2214910919361848, "grad_norm": 0.48249292815663997, "learning_rate": 1.914885991523773e-05, "loss": 2.7206, "step": 35786 }, { "epoch": 2.2215531690359427, "grad_norm": 0.5049372335651165, "learning_rate": 1.914601792383862e-05, "loss": 2.6703, "step": 35787 }, { "epoch": 2.2216152461357006, "grad_norm": 0.4945206970969121, "learning_rate": 1.9143176093412245e-05, "loss": 2.7192, "step": 35788 }, { "epoch": 2.2216773232354585, "grad_norm": 0.5301436797657805, "learning_rate": 1.914033442397341e-05, "loss": 2.7887, "step": 35789 }, { "epoch": 2.2217394003352164, "grad_norm": 0.44652339200795577, "learning_rate": 1.9137492915536943e-05, "loss": 2.6501, "step": 35790 }, { "epoch": 2.2218014774349744, "grad_norm": 0.5118748084732834, "learning_rate": 1.913465156811765e-05, "loss": 2.6375, "step": 35791 }, { "epoch": 2.2218635545347323, "grad_norm": 0.5439950314165161, "learning_rate": 1.9131810381730386e-05, "loss": 2.737, "step": 35792 }, { "epoch": 2.22192563163449, "grad_norm": 0.4984884212157377, "learning_rate": 1.9128969356389964e-05, "loss": 2.6678, "step": 35793 }, { "epoch": 2.221987708734248, "grad_norm": 0.45735642692218126, "learning_rate": 1.91261284921112e-05, "loss": 2.633, "step": 35794 }, { "epoch": 2.222049785834006, "grad_norm": 0.5455416364662751, "learning_rate": 1.912328778890892e-05, "loss": 2.7466, "step": 35795 }, { "epoch": 2.2221118629337635, "grad_norm": 0.4800773825457351, "learning_rate": 1.912044724679793e-05, "loss": 2.6431, "step": 35796 }, { "epoch": 2.222173940033522, "grad_norm": 0.4930237523270637, "learning_rate": 1.9117606865793075e-05, "loss": 2.8012, "step": 35797 }, { "epoch": 2.2222360171332793, "grad_norm": 0.5217228033657225, "learning_rate": 1.9114766645909164e-05, "loss": 2.6638, "step": 35798 }, { "epoch": 2.2222980942330373, "grad_norm": 0.5393672579212994, "learning_rate": 1.9111926587161017e-05, "loss": 2.7111, "step": 35799 }, { "epoch": 2.222360171332795, "grad_norm": 0.5155867938134213, "learning_rate": 1.9109086689563425e-05, "loss": 2.6064, "step": 35800 }, { "epoch": 2.222422248432553, "grad_norm": 0.5543787459598495, "learning_rate": 1.910624695313123e-05, "loss": 2.7198, "step": 35801 }, { "epoch": 2.222484325532311, "grad_norm": 0.49016797045337906, "learning_rate": 1.9103407377879257e-05, "loss": 2.6593, "step": 35802 }, { "epoch": 2.222546402632069, "grad_norm": 0.5381150142615991, "learning_rate": 1.9100567963822312e-05, "loss": 2.7493, "step": 35803 }, { "epoch": 2.222608479731827, "grad_norm": 0.5178654862657236, "learning_rate": 1.9097728710975205e-05, "loss": 2.6554, "step": 35804 }, { "epoch": 2.2226705568315848, "grad_norm": 0.5206797246117767, "learning_rate": 1.909488961935273e-05, "loss": 2.5551, "step": 35805 }, { "epoch": 2.2227326339313427, "grad_norm": 0.5457098401365759, "learning_rate": 1.9092050688969738e-05, "loss": 2.6592, "step": 35806 }, { "epoch": 2.2227947110311006, "grad_norm": 0.4833007097196035, "learning_rate": 1.9089211919841016e-05, "loss": 2.6519, "step": 35807 }, { "epoch": 2.2228567881308585, "grad_norm": 0.5852568678683095, "learning_rate": 1.908637331198138e-05, "loss": 2.7225, "step": 35808 }, { "epoch": 2.2229188652306164, "grad_norm": 0.5208326117192673, "learning_rate": 1.9083534865405645e-05, "loss": 2.6181, "step": 35809 }, { "epoch": 2.2229809423303744, "grad_norm": 0.5048571936122903, "learning_rate": 1.9080696580128594e-05, "loss": 2.728, "step": 35810 }, { "epoch": 2.2230430194301323, "grad_norm": 0.5215522636199883, "learning_rate": 1.907785845616507e-05, "loss": 2.5901, "step": 35811 }, { "epoch": 2.22310509652989, "grad_norm": 0.5441183366339033, "learning_rate": 1.9075020493529867e-05, "loss": 2.735, "step": 35812 }, { "epoch": 2.223167173629648, "grad_norm": 0.49612474720151445, "learning_rate": 1.9072182692237782e-05, "loss": 2.6685, "step": 35813 }, { "epoch": 2.223229250729406, "grad_norm": 0.5118424360748779, "learning_rate": 1.906934505230362e-05, "loss": 2.6689, "step": 35814 }, { "epoch": 2.223291327829164, "grad_norm": 0.5663997518232443, "learning_rate": 1.906650757374221e-05, "loss": 2.6187, "step": 35815 }, { "epoch": 2.223353404928922, "grad_norm": 0.539060703219308, "learning_rate": 1.9063670256568333e-05, "loss": 2.7022, "step": 35816 }, { "epoch": 2.22341548202868, "grad_norm": 0.568272351291514, "learning_rate": 1.9060833100796805e-05, "loss": 2.6923, "step": 35817 }, { "epoch": 2.2234775591284377, "grad_norm": 0.4983907073726742, "learning_rate": 1.9057996106442418e-05, "loss": 2.6991, "step": 35818 }, { "epoch": 2.2235396362281956, "grad_norm": 0.5451429458704776, "learning_rate": 1.9055159273519964e-05, "loss": 2.6766, "step": 35819 }, { "epoch": 2.2236017133279535, "grad_norm": 0.5461838851129871, "learning_rate": 1.9052322602044264e-05, "loss": 2.6886, "step": 35820 }, { "epoch": 2.223663790427711, "grad_norm": 0.5720187491181803, "learning_rate": 1.9049486092030116e-05, "loss": 2.6494, "step": 35821 }, { "epoch": 2.223725867527469, "grad_norm": 0.5775357413688275, "learning_rate": 1.9046649743492316e-05, "loss": 2.7882, "step": 35822 }, { "epoch": 2.223787944627227, "grad_norm": 0.51833992801572, "learning_rate": 1.9043813556445638e-05, "loss": 2.6517, "step": 35823 }, { "epoch": 2.2238500217269848, "grad_norm": 0.5498177508196361, "learning_rate": 1.9040977530904913e-05, "loss": 2.5938, "step": 35824 }, { "epoch": 2.2239120988267427, "grad_norm": 0.5505995213895547, "learning_rate": 1.903814166688493e-05, "loss": 2.6691, "step": 35825 }, { "epoch": 2.2239741759265006, "grad_norm": 0.5287426795633087, "learning_rate": 1.9035305964400473e-05, "loss": 2.6767, "step": 35826 }, { "epoch": 2.2240362530262585, "grad_norm": 0.5759425854678321, "learning_rate": 1.9032470423466325e-05, "loss": 2.6785, "step": 35827 }, { "epoch": 2.2240983301260164, "grad_norm": 0.5327593175573594, "learning_rate": 1.9029635044097315e-05, "loss": 2.6567, "step": 35828 }, { "epoch": 2.2241604072257743, "grad_norm": 0.5832909897921753, "learning_rate": 1.9026799826308222e-05, "loss": 2.7341, "step": 35829 }, { "epoch": 2.2242224843255323, "grad_norm": 0.5129282976015042, "learning_rate": 1.902396477011383e-05, "loss": 2.6495, "step": 35830 }, { "epoch": 2.22428456142529, "grad_norm": 0.5421187794253493, "learning_rate": 1.902112987552893e-05, "loss": 2.7336, "step": 35831 }, { "epoch": 2.224346638525048, "grad_norm": 0.5487994529321018, "learning_rate": 1.90182951425683e-05, "loss": 2.7433, "step": 35832 }, { "epoch": 2.224408715624806, "grad_norm": 0.5766446000858735, "learning_rate": 1.9015460571246746e-05, "loss": 2.7858, "step": 35833 }, { "epoch": 2.224470792724564, "grad_norm": 0.5311319837194094, "learning_rate": 1.9012626161579073e-05, "loss": 2.7116, "step": 35834 }, { "epoch": 2.224532869824322, "grad_norm": 0.5294857743795478, "learning_rate": 1.9009791913580054e-05, "loss": 2.7858, "step": 35835 }, { "epoch": 2.2245949469240798, "grad_norm": 0.586103568385114, "learning_rate": 1.900695782726447e-05, "loss": 2.6495, "step": 35836 }, { "epoch": 2.2246570240238377, "grad_norm": 0.5550533485117782, "learning_rate": 1.9004123902647097e-05, "loss": 2.6524, "step": 35837 }, { "epoch": 2.2247191011235956, "grad_norm": 0.5136246169945966, "learning_rate": 1.900129013974275e-05, "loss": 2.7783, "step": 35838 }, { "epoch": 2.2247811782233535, "grad_norm": 0.5163605674124787, "learning_rate": 1.8998456538566195e-05, "loss": 2.7089, "step": 35839 }, { "epoch": 2.2248432553231114, "grad_norm": 0.4930457398134527, "learning_rate": 1.8995623099132224e-05, "loss": 2.7366, "step": 35840 }, { "epoch": 2.2249053324228694, "grad_norm": 0.5281899621048701, "learning_rate": 1.8992789821455594e-05, "loss": 2.8156, "step": 35841 }, { "epoch": 2.2249674095226273, "grad_norm": 0.48582509080733566, "learning_rate": 1.8989956705551122e-05, "loss": 2.6039, "step": 35842 }, { "epoch": 2.225029486622385, "grad_norm": 0.5097870219333054, "learning_rate": 1.8987123751433573e-05, "loss": 2.7222, "step": 35843 }, { "epoch": 2.2250915637221427, "grad_norm": 0.4854435670408944, "learning_rate": 1.8984290959117734e-05, "loss": 2.7118, "step": 35844 }, { "epoch": 2.225153640821901, "grad_norm": 0.512764451950828, "learning_rate": 1.8981458328618375e-05, "loss": 2.7526, "step": 35845 }, { "epoch": 2.2252157179216585, "grad_norm": 0.5037539052220276, "learning_rate": 1.8978625859950254e-05, "loss": 2.7052, "step": 35846 }, { "epoch": 2.2252777950214164, "grad_norm": 0.5182535047582837, "learning_rate": 1.89757935531282e-05, "loss": 2.7123, "step": 35847 }, { "epoch": 2.2253398721211743, "grad_norm": 0.48789543167812915, "learning_rate": 1.8972961408166955e-05, "loss": 2.6428, "step": 35848 }, { "epoch": 2.2254019492209323, "grad_norm": 0.5626674773651463, "learning_rate": 1.8970129425081305e-05, "loss": 2.7189, "step": 35849 }, { "epoch": 2.22546402632069, "grad_norm": 0.5131675652379124, "learning_rate": 1.896729760388601e-05, "loss": 2.6673, "step": 35850 }, { "epoch": 2.225526103420448, "grad_norm": 0.47444338309407297, "learning_rate": 1.896446594459587e-05, "loss": 2.6995, "step": 35851 }, { "epoch": 2.225588180520206, "grad_norm": 0.5079959538009533, "learning_rate": 1.8961634447225645e-05, "loss": 2.6437, "step": 35852 }, { "epoch": 2.225650257619964, "grad_norm": 0.5174065935452041, "learning_rate": 1.8958803111790108e-05, "loss": 2.6684, "step": 35853 }, { "epoch": 2.225712334719722, "grad_norm": 0.47258128998362364, "learning_rate": 1.8955971938304014e-05, "loss": 2.7692, "step": 35854 }, { "epoch": 2.2257744118194798, "grad_norm": 0.5013703700203488, "learning_rate": 1.895314092678217e-05, "loss": 2.7896, "step": 35855 }, { "epoch": 2.2258364889192377, "grad_norm": 0.5568957898404646, "learning_rate": 1.8950310077239326e-05, "loss": 2.7087, "step": 35856 }, { "epoch": 2.2258985660189956, "grad_norm": 0.4606374475202551, "learning_rate": 1.8947479389690247e-05, "loss": 2.6744, "step": 35857 }, { "epoch": 2.2259606431187535, "grad_norm": 0.4982079937191351, "learning_rate": 1.894464886414971e-05, "loss": 2.5939, "step": 35858 }, { "epoch": 2.2260227202185114, "grad_norm": 0.5350605120261572, "learning_rate": 1.8941818500632462e-05, "loss": 2.7591, "step": 35859 }, { "epoch": 2.2260847973182694, "grad_norm": 0.4752341502695651, "learning_rate": 1.8938988299153304e-05, "loss": 2.6521, "step": 35860 }, { "epoch": 2.2261468744180273, "grad_norm": 0.5350304337339278, "learning_rate": 1.893615825972698e-05, "loss": 2.72, "step": 35861 }, { "epoch": 2.226208951517785, "grad_norm": 0.5298697830324932, "learning_rate": 1.8933328382368264e-05, "loss": 2.7466, "step": 35862 }, { "epoch": 2.226271028617543, "grad_norm": 0.4985160263530991, "learning_rate": 1.8930498667091896e-05, "loss": 2.703, "step": 35863 }, { "epoch": 2.226333105717301, "grad_norm": 0.5128382311031182, "learning_rate": 1.8927669113912677e-05, "loss": 2.7203, "step": 35864 }, { "epoch": 2.226395182817059, "grad_norm": 0.44485450080446115, "learning_rate": 1.892483972284535e-05, "loss": 2.6554, "step": 35865 }, { "epoch": 2.226457259916817, "grad_norm": 0.5313856007916892, "learning_rate": 1.8922010493904664e-05, "loss": 2.7163, "step": 35866 }, { "epoch": 2.226519337016575, "grad_norm": 0.47924444009722766, "learning_rate": 1.8919181427105408e-05, "loss": 2.6592, "step": 35867 }, { "epoch": 2.2265814141163327, "grad_norm": 0.5496794048014686, "learning_rate": 1.891635252246233e-05, "loss": 2.7811, "step": 35868 }, { "epoch": 2.22664349121609, "grad_norm": 0.5083097810708873, "learning_rate": 1.8913523779990168e-05, "loss": 2.7226, "step": 35869 }, { "epoch": 2.226705568315848, "grad_norm": 0.5096407039357022, "learning_rate": 1.8910695199703716e-05, "loss": 2.7137, "step": 35870 }, { "epoch": 2.226767645415606, "grad_norm": 0.47662385952020475, "learning_rate": 1.8907866781617717e-05, "loss": 2.6888, "step": 35871 }, { "epoch": 2.226829722515364, "grad_norm": 0.5116943761247763, "learning_rate": 1.890503852574692e-05, "loss": 2.6097, "step": 35872 }, { "epoch": 2.226891799615122, "grad_norm": 0.5074770479696628, "learning_rate": 1.890221043210607e-05, "loss": 2.6475, "step": 35873 }, { "epoch": 2.2269538767148798, "grad_norm": 0.5511388226716116, "learning_rate": 1.8899382500709955e-05, "loss": 2.6203, "step": 35874 }, { "epoch": 2.2270159538146377, "grad_norm": 0.5092865896437193, "learning_rate": 1.8896554731573312e-05, "loss": 2.6634, "step": 35875 }, { "epoch": 2.2270780309143956, "grad_norm": 0.5485026796305045, "learning_rate": 1.8893727124710896e-05, "loss": 2.7752, "step": 35876 }, { "epoch": 2.2271401080141535, "grad_norm": 0.5094329215976461, "learning_rate": 1.8890899680137437e-05, "loss": 2.6424, "step": 35877 }, { "epoch": 2.2272021851139114, "grad_norm": 0.5232963656420463, "learning_rate": 1.8888072397867723e-05, "loss": 2.7669, "step": 35878 }, { "epoch": 2.2272642622136694, "grad_norm": 0.5775188928874293, "learning_rate": 1.8885245277916487e-05, "loss": 2.7386, "step": 35879 }, { "epoch": 2.2273263393134273, "grad_norm": 0.4770820787016295, "learning_rate": 1.8882418320298485e-05, "loss": 2.6661, "step": 35880 }, { "epoch": 2.227388416413185, "grad_norm": 0.5582513893783233, "learning_rate": 1.8879591525028456e-05, "loss": 2.695, "step": 35881 }, { "epoch": 2.227450493512943, "grad_norm": 0.531838119564, "learning_rate": 1.8876764892121135e-05, "loss": 2.6818, "step": 35882 }, { "epoch": 2.227512570612701, "grad_norm": 0.47383264475001, "learning_rate": 1.887393842159131e-05, "loss": 2.6313, "step": 35883 }, { "epoch": 2.227574647712459, "grad_norm": 0.5489647854661674, "learning_rate": 1.88711121134537e-05, "loss": 2.6232, "step": 35884 }, { "epoch": 2.227636724812217, "grad_norm": 0.5576083154189664, "learning_rate": 1.8868285967723053e-05, "loss": 2.6683, "step": 35885 }, { "epoch": 2.227698801911975, "grad_norm": 0.4994639712247497, "learning_rate": 1.8865459984414097e-05, "loss": 2.7195, "step": 35886 }, { "epoch": 2.2277608790117327, "grad_norm": 0.5547323880367541, "learning_rate": 1.8862634163541616e-05, "loss": 2.7318, "step": 35887 }, { "epoch": 2.2278229561114906, "grad_norm": 0.5269776663074247, "learning_rate": 1.885980850512033e-05, "loss": 2.6914, "step": 35888 }, { "epoch": 2.2278850332112485, "grad_norm": 0.5046405321593523, "learning_rate": 1.8856983009164987e-05, "loss": 2.6407, "step": 35889 }, { "epoch": 2.2279471103110065, "grad_norm": 0.5433613023648335, "learning_rate": 1.8854157675690297e-05, "loss": 2.675, "step": 35890 }, { "epoch": 2.2280091874107644, "grad_norm": 0.48896336022897263, "learning_rate": 1.8851332504711054e-05, "loss": 2.6776, "step": 35891 }, { "epoch": 2.228071264510522, "grad_norm": 0.519459253356189, "learning_rate": 1.8848507496241967e-05, "loss": 2.7029, "step": 35892 }, { "epoch": 2.2281333416102798, "grad_norm": 0.5129228940575894, "learning_rate": 1.884568265029778e-05, "loss": 2.7174, "step": 35893 }, { "epoch": 2.2281954187100377, "grad_norm": 0.5288229083574438, "learning_rate": 1.884285796689323e-05, "loss": 2.6848, "step": 35894 }, { "epoch": 2.2282574958097956, "grad_norm": 0.4902042820421674, "learning_rate": 1.8840033446043033e-05, "loss": 2.7313, "step": 35895 }, { "epoch": 2.2283195729095535, "grad_norm": 0.5344051185284975, "learning_rate": 1.8837209087761966e-05, "loss": 2.6274, "step": 35896 }, { "epoch": 2.2283816500093114, "grad_norm": 0.556037844157262, "learning_rate": 1.8834384892064742e-05, "loss": 2.6827, "step": 35897 }, { "epoch": 2.2284437271090693, "grad_norm": 0.5655540896062373, "learning_rate": 1.883156085896608e-05, "loss": 2.7449, "step": 35898 }, { "epoch": 2.2285058042088273, "grad_norm": 0.5379843780436921, "learning_rate": 1.8828736988480754e-05, "loss": 2.7346, "step": 35899 }, { "epoch": 2.228567881308585, "grad_norm": 0.5044196123532613, "learning_rate": 1.8825913280623453e-05, "loss": 2.7316, "step": 35900 }, { "epoch": 2.228629958408343, "grad_norm": 0.49595380295973074, "learning_rate": 1.882308973540895e-05, "loss": 2.7197, "step": 35901 }, { "epoch": 2.228692035508101, "grad_norm": 0.5007262915389915, "learning_rate": 1.882026635285195e-05, "loss": 2.7416, "step": 35902 }, { "epoch": 2.228754112607859, "grad_norm": 0.5208478035506159, "learning_rate": 1.8817443132967193e-05, "loss": 2.7267, "step": 35903 }, { "epoch": 2.228816189707617, "grad_norm": 0.4849362659362438, "learning_rate": 1.881462007576939e-05, "loss": 2.6287, "step": 35904 }, { "epoch": 2.2288782668073748, "grad_norm": 0.47657811811525996, "learning_rate": 1.88117971812733e-05, "loss": 2.6372, "step": 35905 }, { "epoch": 2.2289403439071327, "grad_norm": 0.5475886456913955, "learning_rate": 1.8808974449493637e-05, "loss": 2.6711, "step": 35906 }, { "epoch": 2.2290024210068906, "grad_norm": 0.4712743814571272, "learning_rate": 1.8806151880445123e-05, "loss": 2.7969, "step": 35907 }, { "epoch": 2.2290644981066485, "grad_norm": 0.5769741614967665, "learning_rate": 1.880332947414249e-05, "loss": 2.7488, "step": 35908 }, { "epoch": 2.2291265752064064, "grad_norm": 0.46599223804865325, "learning_rate": 1.8800507230600444e-05, "loss": 2.6387, "step": 35909 }, { "epoch": 2.2291886523061644, "grad_norm": 0.5653196501187439, "learning_rate": 1.879768514983374e-05, "loss": 2.7863, "step": 35910 }, { "epoch": 2.2292507294059223, "grad_norm": 0.579144712644936, "learning_rate": 1.879486323185709e-05, "loss": 2.714, "step": 35911 }, { "epoch": 2.22931280650568, "grad_norm": 0.5117449862910567, "learning_rate": 1.8792041476685214e-05, "loss": 2.6171, "step": 35912 }, { "epoch": 2.229374883605438, "grad_norm": 0.5341349134911192, "learning_rate": 1.878921988433281e-05, "loss": 2.6432, "step": 35913 }, { "epoch": 2.229436960705196, "grad_norm": 0.5159962213695773, "learning_rate": 1.8786398454814647e-05, "loss": 2.7038, "step": 35914 }, { "epoch": 2.229499037804954, "grad_norm": 0.5585680602241799, "learning_rate": 1.878357718814542e-05, "loss": 2.657, "step": 35915 }, { "epoch": 2.229561114904712, "grad_norm": 0.5405503623906143, "learning_rate": 1.8780756084339844e-05, "loss": 2.6863, "step": 35916 }, { "epoch": 2.2296231920044693, "grad_norm": 0.49126854135985865, "learning_rate": 1.8777935143412643e-05, "loss": 2.7408, "step": 35917 }, { "epoch": 2.2296852691042273, "grad_norm": 0.6096001233673972, "learning_rate": 1.877511436537852e-05, "loss": 2.6044, "step": 35918 }, { "epoch": 2.229747346203985, "grad_norm": 0.4762318668017197, "learning_rate": 1.8772293750252224e-05, "loss": 2.6752, "step": 35919 }, { "epoch": 2.229809423303743, "grad_norm": 0.5499360654257076, "learning_rate": 1.8769473298048444e-05, "loss": 2.6046, "step": 35920 }, { "epoch": 2.229871500403501, "grad_norm": 0.5417392724845002, "learning_rate": 1.8766653008781908e-05, "loss": 2.7181, "step": 35921 }, { "epoch": 2.229933577503259, "grad_norm": 0.48170405431969177, "learning_rate": 1.8763832882467313e-05, "loss": 2.7433, "step": 35922 }, { "epoch": 2.229995654603017, "grad_norm": 0.5069230726895956, "learning_rate": 1.8761012919119396e-05, "loss": 2.6239, "step": 35923 }, { "epoch": 2.2300577317027748, "grad_norm": 0.5385745230956615, "learning_rate": 1.875819311875286e-05, "loss": 2.687, "step": 35924 }, { "epoch": 2.2301198088025327, "grad_norm": 0.5007472712613364, "learning_rate": 1.8755373481382417e-05, "loss": 2.7143, "step": 35925 }, { "epoch": 2.2301818859022906, "grad_norm": 0.4974391445722737, "learning_rate": 1.8752554007022755e-05, "loss": 2.6893, "step": 35926 }, { "epoch": 2.2302439630020485, "grad_norm": 0.5403175946376607, "learning_rate": 1.8749734695688625e-05, "loss": 2.673, "step": 35927 }, { "epoch": 2.2303060401018064, "grad_norm": 0.49813880819904305, "learning_rate": 1.874691554739471e-05, "loss": 2.7046, "step": 35928 }, { "epoch": 2.2303681172015644, "grad_norm": 0.523823666820287, "learning_rate": 1.8744096562155728e-05, "loss": 2.6744, "step": 35929 }, { "epoch": 2.2304301943013223, "grad_norm": 0.5207784210503645, "learning_rate": 1.874127773998638e-05, "loss": 2.6369, "step": 35930 }, { "epoch": 2.23049227140108, "grad_norm": 0.47789287916459117, "learning_rate": 1.8738459080901355e-05, "loss": 2.6597, "step": 35931 }, { "epoch": 2.230554348500838, "grad_norm": 0.5203305456108002, "learning_rate": 1.8735640584915383e-05, "loss": 2.7142, "step": 35932 }, { "epoch": 2.230616425600596, "grad_norm": 0.4843048229448409, "learning_rate": 1.8732822252043176e-05, "loss": 2.7365, "step": 35933 }, { "epoch": 2.230678502700354, "grad_norm": 0.4814439351833985, "learning_rate": 1.8730004082299428e-05, "loss": 2.6336, "step": 35934 }, { "epoch": 2.230740579800112, "grad_norm": 0.5242248136849058, "learning_rate": 1.8727186075698844e-05, "loss": 2.6897, "step": 35935 }, { "epoch": 2.23080265689987, "grad_norm": 0.481397537429947, "learning_rate": 1.87243682322561e-05, "loss": 2.6629, "step": 35936 }, { "epoch": 2.2308647339996277, "grad_norm": 0.486148356721371, "learning_rate": 1.8721550551985934e-05, "loss": 2.7206, "step": 35937 }, { "epoch": 2.2309268110993856, "grad_norm": 0.5360796403487382, "learning_rate": 1.8718733034903034e-05, "loss": 2.6204, "step": 35938 }, { "epoch": 2.2309888881991435, "grad_norm": 0.460296330397807, "learning_rate": 1.8715915681022102e-05, "loss": 2.5811, "step": 35939 }, { "epoch": 2.231050965298901, "grad_norm": 0.46420836099037277, "learning_rate": 1.8713098490357812e-05, "loss": 2.6734, "step": 35940 }, { "epoch": 2.231113042398659, "grad_norm": 0.49398475570941563, "learning_rate": 1.87102814629249e-05, "loss": 2.7348, "step": 35941 }, { "epoch": 2.231175119498417, "grad_norm": 0.45463540476822556, "learning_rate": 1.8707464598738044e-05, "loss": 2.6455, "step": 35942 }, { "epoch": 2.2312371965981748, "grad_norm": 0.4787092478865117, "learning_rate": 1.8704647897811945e-05, "loss": 2.6696, "step": 35943 }, { "epoch": 2.2312992736979327, "grad_norm": 0.44791713056614235, "learning_rate": 1.8701831360161288e-05, "loss": 2.6559, "step": 35944 }, { "epoch": 2.2313613507976906, "grad_norm": 0.490991031270556, "learning_rate": 1.8699014985800763e-05, "loss": 2.5913, "step": 35945 }, { "epoch": 2.2314234278974485, "grad_norm": 0.4719032721228792, "learning_rate": 1.8696198774745092e-05, "loss": 2.7109, "step": 35946 }, { "epoch": 2.2314855049972064, "grad_norm": 0.4510145532359729, "learning_rate": 1.869338272700895e-05, "loss": 2.7219, "step": 35947 }, { "epoch": 2.2315475820969644, "grad_norm": 0.4689778727738866, "learning_rate": 1.869056684260703e-05, "loss": 2.5792, "step": 35948 }, { "epoch": 2.2316096591967223, "grad_norm": 0.5239694084781721, "learning_rate": 1.868775112155401e-05, "loss": 2.6434, "step": 35949 }, { "epoch": 2.23167173629648, "grad_norm": 0.4772782656990244, "learning_rate": 1.868493556386461e-05, "loss": 2.7518, "step": 35950 }, { "epoch": 2.231733813396238, "grad_norm": 0.5411981212870236, "learning_rate": 1.8682120169553502e-05, "loss": 2.7848, "step": 35951 }, { "epoch": 2.231795890495996, "grad_norm": 0.5087095992990412, "learning_rate": 1.8679304938635373e-05, "loss": 2.7544, "step": 35952 }, { "epoch": 2.231857967595754, "grad_norm": 0.48456761929472025, "learning_rate": 1.8676489871124912e-05, "loss": 2.556, "step": 35953 }, { "epoch": 2.231920044695512, "grad_norm": 0.510556159777809, "learning_rate": 1.8673674967036797e-05, "loss": 2.6261, "step": 35954 }, { "epoch": 2.23198212179527, "grad_norm": 0.5450003782760987, "learning_rate": 1.867086022638574e-05, "loss": 2.7214, "step": 35955 }, { "epoch": 2.2320441988950277, "grad_norm": 0.48512187412122804, "learning_rate": 1.866804564918641e-05, "loss": 2.6263, "step": 35956 }, { "epoch": 2.2321062759947856, "grad_norm": 0.4808898878913959, "learning_rate": 1.866523123545349e-05, "loss": 2.5518, "step": 35957 }, { "epoch": 2.2321683530945435, "grad_norm": 0.46362950710719264, "learning_rate": 1.8662416985201646e-05, "loss": 2.6556, "step": 35958 }, { "epoch": 2.2322304301943015, "grad_norm": 0.5631191237266888, "learning_rate": 1.8659602898445606e-05, "loss": 2.7006, "step": 35959 }, { "epoch": 2.2322925072940594, "grad_norm": 0.5326066086393116, "learning_rate": 1.8656788975200018e-05, "loss": 2.6273, "step": 35960 }, { "epoch": 2.2323545843938173, "grad_norm": 0.5353733930899878, "learning_rate": 1.8653975215479574e-05, "loss": 2.6479, "step": 35961 }, { "epoch": 2.232416661493575, "grad_norm": 0.5701391662145764, "learning_rate": 1.8651161619298936e-05, "loss": 2.7532, "step": 35962 }, { "epoch": 2.232478738593333, "grad_norm": 0.5352486857754903, "learning_rate": 1.864834818667281e-05, "loss": 2.6211, "step": 35963 }, { "epoch": 2.232540815693091, "grad_norm": 0.48908981515087496, "learning_rate": 1.8645534917615846e-05, "loss": 2.7156, "step": 35964 }, { "epoch": 2.2326028927928485, "grad_norm": 0.49198911089690495, "learning_rate": 1.8642721812142757e-05, "loss": 2.6525, "step": 35965 }, { "epoch": 2.2326649698926064, "grad_norm": 0.5531977406317717, "learning_rate": 1.8639908870268195e-05, "loss": 2.6785, "step": 35966 }, { "epoch": 2.2327270469923644, "grad_norm": 0.5022470229835823, "learning_rate": 1.863709609200685e-05, "loss": 2.7085, "step": 35967 }, { "epoch": 2.2327891240921223, "grad_norm": 0.5660958327387486, "learning_rate": 1.8634283477373365e-05, "loss": 2.6907, "step": 35968 }, { "epoch": 2.23285120119188, "grad_norm": 0.49804086120565016, "learning_rate": 1.8631471026382457e-05, "loss": 2.7134, "step": 35969 }, { "epoch": 2.232913278291638, "grad_norm": 0.5186706240949617, "learning_rate": 1.862865873904878e-05, "loss": 2.7342, "step": 35970 }, { "epoch": 2.232975355391396, "grad_norm": 0.5674917671261819, "learning_rate": 1.8625846615387003e-05, "loss": 2.6481, "step": 35971 }, { "epoch": 2.233037432491154, "grad_norm": 0.5257736568271069, "learning_rate": 1.862303465541179e-05, "loss": 2.6865, "step": 35972 }, { "epoch": 2.233099509590912, "grad_norm": 0.5164083252834801, "learning_rate": 1.8620222859137836e-05, "loss": 2.725, "step": 35973 }, { "epoch": 2.2331615866906698, "grad_norm": 0.5458660627943357, "learning_rate": 1.8617411226579795e-05, "loss": 2.7274, "step": 35974 }, { "epoch": 2.2332236637904277, "grad_norm": 0.4940267740804916, "learning_rate": 1.8614599757752338e-05, "loss": 2.7251, "step": 35975 }, { "epoch": 2.2332857408901856, "grad_norm": 0.5112335583904072, "learning_rate": 1.8611788452670122e-05, "loss": 2.7378, "step": 35976 }, { "epoch": 2.2333478179899435, "grad_norm": 0.47385984662097563, "learning_rate": 1.8608977311347835e-05, "loss": 2.6656, "step": 35977 }, { "epoch": 2.2334098950897014, "grad_norm": 0.5022217740396359, "learning_rate": 1.860616633380014e-05, "loss": 2.6374, "step": 35978 }, { "epoch": 2.2334719721894594, "grad_norm": 0.4991291180590526, "learning_rate": 1.8603355520041694e-05, "loss": 2.6061, "step": 35979 }, { "epoch": 2.2335340492892173, "grad_norm": 0.49296068586956904, "learning_rate": 1.860054487008716e-05, "loss": 2.6622, "step": 35980 }, { "epoch": 2.233596126388975, "grad_norm": 0.47175774412161764, "learning_rate": 1.859773438395119e-05, "loss": 2.7238, "step": 35981 }, { "epoch": 2.233658203488733, "grad_norm": 0.5159050729913657, "learning_rate": 1.8594924061648484e-05, "loss": 2.7336, "step": 35982 }, { "epoch": 2.233720280588491, "grad_norm": 0.5459010033726315, "learning_rate": 1.859211390319368e-05, "loss": 2.6876, "step": 35983 }, { "epoch": 2.233782357688249, "grad_norm": 0.4969572077227979, "learning_rate": 1.8589303908601448e-05, "loss": 2.6825, "step": 35984 }, { "epoch": 2.233844434788007, "grad_norm": 0.49782949453346115, "learning_rate": 1.8586494077886417e-05, "loss": 2.6928, "step": 35985 }, { "epoch": 2.233906511887765, "grad_norm": 0.5359259868308269, "learning_rate": 1.858368441106329e-05, "loss": 2.6369, "step": 35986 }, { "epoch": 2.2339685889875227, "grad_norm": 0.5030288154167283, "learning_rate": 1.8580874908146712e-05, "loss": 2.642, "step": 35987 }, { "epoch": 2.23403066608728, "grad_norm": 0.47343251820577603, "learning_rate": 1.8578065569151336e-05, "loss": 2.6977, "step": 35988 }, { "epoch": 2.234092743187038, "grad_norm": 0.48084493967138303, "learning_rate": 1.857525639409182e-05, "loss": 2.6851, "step": 35989 }, { "epoch": 2.234154820286796, "grad_norm": 0.5298551311074919, "learning_rate": 1.8572447382982803e-05, "loss": 2.7078, "step": 35990 }, { "epoch": 2.234216897386554, "grad_norm": 0.4859130344180969, "learning_rate": 1.8569638535838978e-05, "loss": 2.6862, "step": 35991 }, { "epoch": 2.234278974486312, "grad_norm": 0.5387792534199052, "learning_rate": 1.8566829852674972e-05, "loss": 2.7055, "step": 35992 }, { "epoch": 2.2343410515860698, "grad_norm": 0.5300834132628592, "learning_rate": 1.8564021333505448e-05, "loss": 2.7967, "step": 35993 }, { "epoch": 2.2344031286858277, "grad_norm": 0.5129180424400903, "learning_rate": 1.8561212978345037e-05, "loss": 2.6677, "step": 35994 }, { "epoch": 2.2344652057855856, "grad_norm": 0.581734293510495, "learning_rate": 1.8558404787208433e-05, "loss": 2.7388, "step": 35995 }, { "epoch": 2.2345272828853435, "grad_norm": 0.5123450835950224, "learning_rate": 1.855559676011026e-05, "loss": 2.6659, "step": 35996 }, { "epoch": 2.2345893599851014, "grad_norm": 0.5406629328567437, "learning_rate": 1.8552788897065154e-05, "loss": 2.6569, "step": 35997 }, { "epoch": 2.2346514370848594, "grad_norm": 0.5221323851943702, "learning_rate": 1.85499811980878e-05, "loss": 2.6926, "step": 35998 }, { "epoch": 2.2347135141846173, "grad_norm": 0.5224242865996093, "learning_rate": 1.8547173663192812e-05, "loss": 2.7285, "step": 35999 }, { "epoch": 2.234775591284375, "grad_norm": 0.5038676875536562, "learning_rate": 1.8544366292394877e-05, "loss": 2.6849, "step": 36000 }, { "epoch": 2.234837668384133, "grad_norm": 0.4807809265254646, "learning_rate": 1.8541559085708614e-05, "loss": 2.7347, "step": 36001 }, { "epoch": 2.234899745483891, "grad_norm": 0.4842928693728476, "learning_rate": 1.8538752043148673e-05, "loss": 2.5889, "step": 36002 }, { "epoch": 2.234961822583649, "grad_norm": 0.4827439781131432, "learning_rate": 1.85359451647297e-05, "loss": 2.6344, "step": 36003 }, { "epoch": 2.235023899683407, "grad_norm": 0.4783327025510149, "learning_rate": 1.8533138450466326e-05, "loss": 2.6712, "step": 36004 }, { "epoch": 2.235085976783165, "grad_norm": 0.4718774574417899, "learning_rate": 1.853033190037322e-05, "loss": 2.6292, "step": 36005 }, { "epoch": 2.2351480538829227, "grad_norm": 0.4898407408212713, "learning_rate": 1.8527525514465017e-05, "loss": 2.6779, "step": 36006 }, { "epoch": 2.2352101309826806, "grad_norm": 0.5023206560156941, "learning_rate": 1.852471929275635e-05, "loss": 2.6627, "step": 36007 }, { "epoch": 2.2352722080824385, "grad_norm": 0.479540525280199, "learning_rate": 1.8521913235261844e-05, "loss": 2.6733, "step": 36008 }, { "epoch": 2.2353342851821965, "grad_norm": 0.5236618642918656, "learning_rate": 1.851910734199618e-05, "loss": 2.8003, "step": 36009 }, { "epoch": 2.2353963622819544, "grad_norm": 0.5125620350858275, "learning_rate": 1.8516301612973973e-05, "loss": 2.7462, "step": 36010 }, { "epoch": 2.2354584393817123, "grad_norm": 0.50972335464786, "learning_rate": 1.8513496048209867e-05, "loss": 2.6302, "step": 36011 }, { "epoch": 2.23552051648147, "grad_norm": 0.5633529448152165, "learning_rate": 1.851069064771847e-05, "loss": 2.6861, "step": 36012 }, { "epoch": 2.2355825935812277, "grad_norm": 0.48286754084550143, "learning_rate": 1.850788541151447e-05, "loss": 2.7215, "step": 36013 }, { "epoch": 2.2356446706809856, "grad_norm": 0.5343767879969259, "learning_rate": 1.8505080339612475e-05, "loss": 2.5889, "step": 36014 }, { "epoch": 2.2357067477807435, "grad_norm": 0.528698797970858, "learning_rate": 1.850227543202712e-05, "loss": 2.6837, "step": 36015 }, { "epoch": 2.2357688248805014, "grad_norm": 0.5147318763339926, "learning_rate": 1.849947068877304e-05, "loss": 2.7791, "step": 36016 }, { "epoch": 2.2358309019802594, "grad_norm": 0.4847672627657291, "learning_rate": 1.849666610986485e-05, "loss": 2.6475, "step": 36017 }, { "epoch": 2.2358929790800173, "grad_norm": 0.5534440192813836, "learning_rate": 1.8493861695317223e-05, "loss": 2.6678, "step": 36018 }, { "epoch": 2.235955056179775, "grad_norm": 0.4957873231580166, "learning_rate": 1.849105744514476e-05, "loss": 2.6567, "step": 36019 }, { "epoch": 2.236017133279533, "grad_norm": 0.4776513839339049, "learning_rate": 1.8488253359362105e-05, "loss": 2.6992, "step": 36020 }, { "epoch": 2.236079210379291, "grad_norm": 0.4885335804311408, "learning_rate": 1.8485449437983866e-05, "loss": 2.8, "step": 36021 }, { "epoch": 2.236141287479049, "grad_norm": 0.48901217875107483, "learning_rate": 1.8482645681024703e-05, "loss": 2.673, "step": 36022 }, { "epoch": 2.236203364578807, "grad_norm": 0.49407049292047184, "learning_rate": 1.8479842088499226e-05, "loss": 2.7161, "step": 36023 }, { "epoch": 2.236265441678565, "grad_norm": 0.519238353569251, "learning_rate": 1.847703866042207e-05, "loss": 2.6937, "step": 36024 }, { "epoch": 2.2363275187783227, "grad_norm": 0.4920936525276559, "learning_rate": 1.8474235396807833e-05, "loss": 2.6847, "step": 36025 }, { "epoch": 2.2363895958780806, "grad_norm": 0.5531346250224647, "learning_rate": 1.847143229767118e-05, "loss": 2.7169, "step": 36026 }, { "epoch": 2.2364516729778385, "grad_norm": 0.5165577457068421, "learning_rate": 1.8468629363026724e-05, "loss": 2.6391, "step": 36027 }, { "epoch": 2.2365137500775965, "grad_norm": 0.48013534453216744, "learning_rate": 1.846582659288908e-05, "loss": 2.5745, "step": 36028 }, { "epoch": 2.2365758271773544, "grad_norm": 0.5233412981591478, "learning_rate": 1.846302398727286e-05, "loss": 2.6957, "step": 36029 }, { "epoch": 2.2366379042771123, "grad_norm": 0.44919911217266334, "learning_rate": 1.8460221546192713e-05, "loss": 2.641, "step": 36030 }, { "epoch": 2.23669998137687, "grad_norm": 0.5045973599839217, "learning_rate": 1.8457419269663228e-05, "loss": 2.6164, "step": 36031 }, { "epoch": 2.236762058476628, "grad_norm": 0.5182156073810759, "learning_rate": 1.8454617157699067e-05, "loss": 2.6351, "step": 36032 }, { "epoch": 2.236824135576386, "grad_norm": 0.4495658197194205, "learning_rate": 1.845181521031482e-05, "loss": 2.6482, "step": 36033 }, { "epoch": 2.236886212676144, "grad_norm": 0.48012679422128207, "learning_rate": 1.8449013427525113e-05, "loss": 2.6991, "step": 36034 }, { "epoch": 2.236948289775902, "grad_norm": 0.4781241080331897, "learning_rate": 1.8446211809344548e-05, "loss": 2.6903, "step": 36035 }, { "epoch": 2.2370103668756594, "grad_norm": 0.4500908885669656, "learning_rate": 1.844341035578777e-05, "loss": 2.6641, "step": 36036 }, { "epoch": 2.2370724439754173, "grad_norm": 0.4672032586441316, "learning_rate": 1.8440609066869384e-05, "loss": 2.6876, "step": 36037 }, { "epoch": 2.237134521075175, "grad_norm": 0.4794901053389269, "learning_rate": 1.8437807942604e-05, "loss": 2.6663, "step": 36038 }, { "epoch": 2.237196598174933, "grad_norm": 0.45479604467619633, "learning_rate": 1.8435006983006237e-05, "loss": 2.7455, "step": 36039 }, { "epoch": 2.237258675274691, "grad_norm": 0.44757109626022606, "learning_rate": 1.8432206188090685e-05, "loss": 2.6831, "step": 36040 }, { "epoch": 2.237320752374449, "grad_norm": 0.49269699792007343, "learning_rate": 1.8429405557871993e-05, "loss": 2.6529, "step": 36041 }, { "epoch": 2.237382829474207, "grad_norm": 0.5036636478254497, "learning_rate": 1.8426605092364756e-05, "loss": 2.6973, "step": 36042 }, { "epoch": 2.2374449065739648, "grad_norm": 0.47408040713824634, "learning_rate": 1.8423804791583588e-05, "loss": 2.5955, "step": 36043 }, { "epoch": 2.2375069836737227, "grad_norm": 0.4665249627120431, "learning_rate": 1.8421004655543077e-05, "loss": 2.6666, "step": 36044 }, { "epoch": 2.2375690607734806, "grad_norm": 0.4695709986055352, "learning_rate": 1.8418204684257867e-05, "loss": 2.7439, "step": 36045 }, { "epoch": 2.2376311378732385, "grad_norm": 0.47937164034405966, "learning_rate": 1.8415404877742548e-05, "loss": 2.76, "step": 36046 }, { "epoch": 2.2376932149729964, "grad_norm": 0.5236692342644645, "learning_rate": 1.8412605236011732e-05, "loss": 2.6382, "step": 36047 }, { "epoch": 2.2377552920727544, "grad_norm": 0.4585052749315795, "learning_rate": 1.8409805759080005e-05, "loss": 2.637, "step": 36048 }, { "epoch": 2.2378173691725123, "grad_norm": 0.46487817716936003, "learning_rate": 1.8407006446962012e-05, "loss": 2.6704, "step": 36049 }, { "epoch": 2.23787944627227, "grad_norm": 0.5253257285953807, "learning_rate": 1.8404207299672327e-05, "loss": 2.6434, "step": 36050 }, { "epoch": 2.237941523372028, "grad_norm": 0.47816935875758515, "learning_rate": 1.840140831722557e-05, "loss": 2.5905, "step": 36051 }, { "epoch": 2.238003600471786, "grad_norm": 0.4321713701494371, "learning_rate": 1.839860949963633e-05, "loss": 2.6679, "step": 36052 }, { "epoch": 2.238065677571544, "grad_norm": 0.47397104358048986, "learning_rate": 1.8395810846919203e-05, "loss": 2.6107, "step": 36053 }, { "epoch": 2.238127754671302, "grad_norm": 0.5155914469119013, "learning_rate": 1.8393012359088824e-05, "loss": 2.6711, "step": 36054 }, { "epoch": 2.23818983177106, "grad_norm": 0.4576769123511907, "learning_rate": 1.8390214036159762e-05, "loss": 2.749, "step": 36055 }, { "epoch": 2.2382519088708177, "grad_norm": 0.5352985721876236, "learning_rate": 1.8387415878146636e-05, "loss": 2.7022, "step": 36056 }, { "epoch": 2.2383139859705756, "grad_norm": 0.5002667192508559, "learning_rate": 1.8384617885064015e-05, "loss": 2.6543, "step": 36057 }, { "epoch": 2.2383760630703335, "grad_norm": 0.4318337677237297, "learning_rate": 1.8381820056926535e-05, "loss": 2.6775, "step": 36058 }, { "epoch": 2.238438140170091, "grad_norm": 0.4989584415756883, "learning_rate": 1.8379022393748773e-05, "loss": 2.6547, "step": 36059 }, { "epoch": 2.2385002172698494, "grad_norm": 0.46337530458294696, "learning_rate": 1.837622489554533e-05, "loss": 2.673, "step": 36060 }, { "epoch": 2.238562294369607, "grad_norm": 0.49350255548207816, "learning_rate": 1.8373427562330787e-05, "loss": 2.6609, "step": 36061 }, { "epoch": 2.2386243714693648, "grad_norm": 0.49898007796124433, "learning_rate": 1.8370630394119743e-05, "loss": 2.7168, "step": 36062 }, { "epoch": 2.2386864485691227, "grad_norm": 0.4497167397775494, "learning_rate": 1.8367833390926814e-05, "loss": 2.7548, "step": 36063 }, { "epoch": 2.2387485256688806, "grad_norm": 0.5072470230834903, "learning_rate": 1.836503655276658e-05, "loss": 2.6804, "step": 36064 }, { "epoch": 2.2388106027686385, "grad_norm": 0.4503813750085559, "learning_rate": 1.836223987965363e-05, "loss": 2.728, "step": 36065 }, { "epoch": 2.2388726798683964, "grad_norm": 0.4718468253459084, "learning_rate": 1.8359443371602553e-05, "loss": 2.6962, "step": 36066 }, { "epoch": 2.2389347569681544, "grad_norm": 0.4814464422061812, "learning_rate": 1.8356647028627926e-05, "loss": 2.6403, "step": 36067 }, { "epoch": 2.2389968340679123, "grad_norm": 0.4687832661586575, "learning_rate": 1.8353850850744368e-05, "loss": 2.6878, "step": 36068 }, { "epoch": 2.23905891116767, "grad_norm": 0.4596292389048382, "learning_rate": 1.8351054837966458e-05, "loss": 2.6488, "step": 36069 }, { "epoch": 2.239120988267428, "grad_norm": 0.45392698808526605, "learning_rate": 1.8348258990308774e-05, "loss": 2.6455, "step": 36070 }, { "epoch": 2.239183065367186, "grad_norm": 0.4707886550883073, "learning_rate": 1.834546330778589e-05, "loss": 2.6583, "step": 36071 }, { "epoch": 2.239245142466944, "grad_norm": 0.5076740956596758, "learning_rate": 1.8342667790412426e-05, "loss": 2.5988, "step": 36072 }, { "epoch": 2.239307219566702, "grad_norm": 0.4941743982406058, "learning_rate": 1.8339872438202948e-05, "loss": 2.7396, "step": 36073 }, { "epoch": 2.23936929666646, "grad_norm": 0.5077670737269145, "learning_rate": 1.833707725117204e-05, "loss": 2.6815, "step": 36074 }, { "epoch": 2.2394313737662177, "grad_norm": 0.523469980854771, "learning_rate": 1.833428222933427e-05, "loss": 2.6525, "step": 36075 }, { "epoch": 2.2394934508659756, "grad_norm": 0.5213584319202846, "learning_rate": 1.8331487372704258e-05, "loss": 2.7307, "step": 36076 }, { "epoch": 2.2395555279657335, "grad_norm": 0.5302906630723287, "learning_rate": 1.832869268129656e-05, "loss": 2.6742, "step": 36077 }, { "epoch": 2.2396176050654915, "grad_norm": 0.5195099675688146, "learning_rate": 1.8325898155125758e-05, "loss": 2.691, "step": 36078 }, { "epoch": 2.2396796821652494, "grad_norm": 0.6140494910650642, "learning_rate": 1.8323103794206442e-05, "loss": 2.7275, "step": 36079 }, { "epoch": 2.2397417592650073, "grad_norm": 0.4671234201736827, "learning_rate": 1.8320309598553158e-05, "loss": 2.6881, "step": 36080 }, { "epoch": 2.239803836364765, "grad_norm": 0.5327173332624717, "learning_rate": 1.8317515568180532e-05, "loss": 2.6422, "step": 36081 }, { "epoch": 2.239865913464523, "grad_norm": 0.511445096058932, "learning_rate": 1.831472170310312e-05, "loss": 2.7213, "step": 36082 }, { "epoch": 2.239927990564281, "grad_norm": 0.4828397662910886, "learning_rate": 1.8311928003335494e-05, "loss": 2.6709, "step": 36083 }, { "epoch": 2.2399900676640385, "grad_norm": 0.43887513191288524, "learning_rate": 1.8309134468892215e-05, "loss": 2.6492, "step": 36084 }, { "epoch": 2.2400521447637964, "grad_norm": 0.5585207858149347, "learning_rate": 1.8306341099787894e-05, "loss": 2.7518, "step": 36085 }, { "epoch": 2.2401142218635544, "grad_norm": 0.4728608122231728, "learning_rate": 1.8303547896037083e-05, "loss": 2.7265, "step": 36086 }, { "epoch": 2.2401762989633123, "grad_norm": 0.4804141471043489, "learning_rate": 1.8300754857654358e-05, "loss": 2.6473, "step": 36087 }, { "epoch": 2.24023837606307, "grad_norm": 0.49533652539153145, "learning_rate": 1.829796198465429e-05, "loss": 2.7494, "step": 36088 }, { "epoch": 2.240300453162828, "grad_norm": 0.5121953942910688, "learning_rate": 1.8295169277051432e-05, "loss": 2.6927, "step": 36089 }, { "epoch": 2.240362530262586, "grad_norm": 0.5019283956529225, "learning_rate": 1.829237673486039e-05, "loss": 2.7905, "step": 36090 }, { "epoch": 2.240424607362344, "grad_norm": 0.5040763154001947, "learning_rate": 1.828958435809572e-05, "loss": 2.6767, "step": 36091 }, { "epoch": 2.240486684462102, "grad_norm": 0.4817479789742964, "learning_rate": 1.828679214677198e-05, "loss": 2.7157, "step": 36092 }, { "epoch": 2.24054876156186, "grad_norm": 0.48385231981311666, "learning_rate": 1.828400010090373e-05, "loss": 2.7884, "step": 36093 }, { "epoch": 2.2406108386616177, "grad_norm": 0.5187758783986448, "learning_rate": 1.8281208220505562e-05, "loss": 2.7709, "step": 36094 }, { "epoch": 2.2406729157613756, "grad_norm": 0.5095135588091216, "learning_rate": 1.8278416505592023e-05, "loss": 2.7973, "step": 36095 }, { "epoch": 2.2407349928611335, "grad_norm": 0.4713791389243159, "learning_rate": 1.82756249561777e-05, "loss": 2.7353, "step": 36096 }, { "epoch": 2.2407970699608915, "grad_norm": 0.5035584010619877, "learning_rate": 1.8272833572277142e-05, "loss": 2.7285, "step": 36097 }, { "epoch": 2.2408591470606494, "grad_norm": 0.47289022849315415, "learning_rate": 1.8270042353904898e-05, "loss": 2.6535, "step": 36098 }, { "epoch": 2.2409212241604073, "grad_norm": 0.49781018785225895, "learning_rate": 1.826725130107556e-05, "loss": 2.7151, "step": 36099 }, { "epoch": 2.240983301260165, "grad_norm": 0.5152174755272021, "learning_rate": 1.8264460413803674e-05, "loss": 2.6922, "step": 36100 }, { "epoch": 2.241045378359923, "grad_norm": 0.46349724724640406, "learning_rate": 1.8261669692103804e-05, "loss": 2.6591, "step": 36101 }, { "epoch": 2.241107455459681, "grad_norm": 0.49270770420359583, "learning_rate": 1.8258879135990508e-05, "loss": 2.7075, "step": 36102 }, { "epoch": 2.241169532559439, "grad_norm": 0.49270605593498434, "learning_rate": 1.825608874547833e-05, "loss": 2.7005, "step": 36103 }, { "epoch": 2.241231609659197, "grad_norm": 0.49992513096078905, "learning_rate": 1.8253298520581857e-05, "loss": 2.7288, "step": 36104 }, { "epoch": 2.241293686758955, "grad_norm": 0.504401975947903, "learning_rate": 1.8250508461315637e-05, "loss": 2.6415, "step": 36105 }, { "epoch": 2.2413557638587127, "grad_norm": 0.5012583098352982, "learning_rate": 1.8247718567694218e-05, "loss": 2.6235, "step": 36106 }, { "epoch": 2.24141784095847, "grad_norm": 0.517967164280278, "learning_rate": 1.8244928839732145e-05, "loss": 2.6541, "step": 36107 }, { "epoch": 2.2414799180582285, "grad_norm": 0.4984400225967143, "learning_rate": 1.8242139277443998e-05, "loss": 2.6726, "step": 36108 }, { "epoch": 2.241541995157986, "grad_norm": 0.4922080262763183, "learning_rate": 1.8239349880844325e-05, "loss": 2.6832, "step": 36109 }, { "epoch": 2.241604072257744, "grad_norm": 0.4924086876031922, "learning_rate": 1.8236560649947677e-05, "loss": 2.7197, "step": 36110 }, { "epoch": 2.241666149357502, "grad_norm": 0.5034377707463489, "learning_rate": 1.8233771584768577e-05, "loss": 2.6739, "step": 36111 }, { "epoch": 2.2417282264572598, "grad_norm": 0.4979100207626128, "learning_rate": 1.8230982685321625e-05, "loss": 2.7507, "step": 36112 }, { "epoch": 2.2417903035570177, "grad_norm": 0.47114077281181027, "learning_rate": 1.8228193951621347e-05, "loss": 2.6607, "step": 36113 }, { "epoch": 2.2418523806567756, "grad_norm": 0.5194922124207377, "learning_rate": 1.8225405383682297e-05, "loss": 2.666, "step": 36114 }, { "epoch": 2.2419144577565335, "grad_norm": 0.468815512847475, "learning_rate": 1.822261698151902e-05, "loss": 2.7648, "step": 36115 }, { "epoch": 2.2419765348562914, "grad_norm": 0.4865109662969444, "learning_rate": 1.8219828745146045e-05, "loss": 2.7349, "step": 36116 }, { "epoch": 2.2420386119560494, "grad_norm": 0.46433672781910956, "learning_rate": 1.8217040674577952e-05, "loss": 2.6545, "step": 36117 }, { "epoch": 2.2421006890558073, "grad_norm": 0.495209815034213, "learning_rate": 1.8214252769829277e-05, "loss": 2.6683, "step": 36118 }, { "epoch": 2.242162766155565, "grad_norm": 0.572122809911182, "learning_rate": 1.8211465030914566e-05, "loss": 2.6799, "step": 36119 }, { "epoch": 2.242224843255323, "grad_norm": 0.447926378884109, "learning_rate": 1.8208677457848333e-05, "loss": 2.638, "step": 36120 }, { "epoch": 2.242286920355081, "grad_norm": 0.577849270819877, "learning_rate": 1.8205890050645175e-05, "loss": 2.6325, "step": 36121 }, { "epoch": 2.242348997454839, "grad_norm": 0.5055359502398057, "learning_rate": 1.82031028093196e-05, "loss": 2.6585, "step": 36122 }, { "epoch": 2.242411074554597, "grad_norm": 0.48997574325291826, "learning_rate": 1.8200315733886154e-05, "loss": 2.6264, "step": 36123 }, { "epoch": 2.242473151654355, "grad_norm": 0.5383510594161012, "learning_rate": 1.8197528824359383e-05, "loss": 2.6917, "step": 36124 }, { "epoch": 2.2425352287541127, "grad_norm": 0.48399277957641684, "learning_rate": 1.8194742080753813e-05, "loss": 2.6875, "step": 36125 }, { "epoch": 2.2425973058538706, "grad_norm": 0.5077246369864489, "learning_rate": 1.8191955503084006e-05, "loss": 2.6835, "step": 36126 }, { "epoch": 2.2426593829536285, "grad_norm": 0.5231155715056254, "learning_rate": 1.8189169091364488e-05, "loss": 2.6081, "step": 36127 }, { "epoch": 2.2427214600533865, "grad_norm": 0.49753794379242194, "learning_rate": 1.8186382845609783e-05, "loss": 2.5789, "step": 36128 }, { "epoch": 2.2427835371531444, "grad_norm": 0.48270103933393554, "learning_rate": 1.818359676583446e-05, "loss": 2.5662, "step": 36129 }, { "epoch": 2.2428456142529023, "grad_norm": 0.4783222975715017, "learning_rate": 1.8180810852053015e-05, "loss": 2.7633, "step": 36130 }, { "epoch": 2.24290769135266, "grad_norm": 0.46781811271873086, "learning_rate": 1.817802510428002e-05, "loss": 2.6616, "step": 36131 }, { "epoch": 2.2429697684524177, "grad_norm": 0.47910070656390114, "learning_rate": 1.8175239522530002e-05, "loss": 2.6812, "step": 36132 }, { "epoch": 2.2430318455521756, "grad_norm": 0.46039080122823434, "learning_rate": 1.817245410681748e-05, "loss": 2.6728, "step": 36133 }, { "epoch": 2.2430939226519335, "grad_norm": 0.48374411813477025, "learning_rate": 1.8169668857156975e-05, "loss": 2.5576, "step": 36134 }, { "epoch": 2.2431559997516914, "grad_norm": 0.4907414067420132, "learning_rate": 1.8166883773563048e-05, "loss": 2.6371, "step": 36135 }, { "epoch": 2.2432180768514494, "grad_norm": 0.5110320514137778, "learning_rate": 1.8164098856050215e-05, "loss": 2.8029, "step": 36136 }, { "epoch": 2.2432801539512073, "grad_norm": 0.48520034798482287, "learning_rate": 1.8161314104633014e-05, "loss": 2.6634, "step": 36137 }, { "epoch": 2.243342231050965, "grad_norm": 0.5291495311645951, "learning_rate": 1.815852951932596e-05, "loss": 2.5957, "step": 36138 }, { "epoch": 2.243404308150723, "grad_norm": 0.4671643618209158, "learning_rate": 1.8155745100143573e-05, "loss": 2.6243, "step": 36139 }, { "epoch": 2.243466385250481, "grad_norm": 0.5120971096266388, "learning_rate": 1.8152960847100413e-05, "loss": 2.6532, "step": 36140 }, { "epoch": 2.243528462350239, "grad_norm": 0.5046177653506795, "learning_rate": 1.8150176760210985e-05, "loss": 2.7212, "step": 36141 }, { "epoch": 2.243590539449997, "grad_norm": 0.49241980868012974, "learning_rate": 1.8147392839489817e-05, "loss": 2.7427, "step": 36142 }, { "epoch": 2.243652616549755, "grad_norm": 0.5262933512974234, "learning_rate": 1.8144609084951418e-05, "loss": 2.6498, "step": 36143 }, { "epoch": 2.2437146936495127, "grad_norm": 0.4938553902863596, "learning_rate": 1.8141825496610343e-05, "loss": 2.5917, "step": 36144 }, { "epoch": 2.2437767707492706, "grad_norm": 0.4615909179999372, "learning_rate": 1.8139042074481093e-05, "loss": 2.6372, "step": 36145 }, { "epoch": 2.2438388478490285, "grad_norm": 0.5062874415174279, "learning_rate": 1.81362588185782e-05, "loss": 2.7739, "step": 36146 }, { "epoch": 2.2439009249487865, "grad_norm": 0.5179455874228119, "learning_rate": 1.8133475728916165e-05, "loss": 2.7022, "step": 36147 }, { "epoch": 2.2439630020485444, "grad_norm": 0.5462538597703829, "learning_rate": 1.8130692805509536e-05, "loss": 2.6974, "step": 36148 }, { "epoch": 2.2440250791483023, "grad_norm": 0.5363810091555508, "learning_rate": 1.812791004837282e-05, "loss": 2.712, "step": 36149 }, { "epoch": 2.24408715624806, "grad_norm": 0.5237005341507273, "learning_rate": 1.8125127457520534e-05, "loss": 2.6552, "step": 36150 }, { "epoch": 2.244149233347818, "grad_norm": 0.5248088238818952, "learning_rate": 1.8122345032967197e-05, "loss": 2.7554, "step": 36151 }, { "epoch": 2.244211310447576, "grad_norm": 0.49892321511469406, "learning_rate": 1.8119562774727312e-05, "loss": 2.6587, "step": 36152 }, { "epoch": 2.244273387547334, "grad_norm": 0.5694124618584191, "learning_rate": 1.8116780682815415e-05, "loss": 2.6418, "step": 36153 }, { "epoch": 2.244335464647092, "grad_norm": 0.5240935584039459, "learning_rate": 1.811399875724602e-05, "loss": 2.6973, "step": 36154 }, { "epoch": 2.2443975417468494, "grad_norm": 0.5860467427460266, "learning_rate": 1.811121699803363e-05, "loss": 2.706, "step": 36155 }, { "epoch": 2.2444596188466077, "grad_norm": 0.5601424509474295, "learning_rate": 1.810843540519275e-05, "loss": 2.6664, "step": 36156 }, { "epoch": 2.244521695946365, "grad_norm": 0.5300086609114745, "learning_rate": 1.8105653978737917e-05, "loss": 2.6784, "step": 36157 }, { "epoch": 2.244583773046123, "grad_norm": 0.5947417709836393, "learning_rate": 1.8102872718683633e-05, "loss": 2.6947, "step": 36158 }, { "epoch": 2.244645850145881, "grad_norm": 0.5040763154001947, "learning_rate": 1.81000916250444e-05, "loss": 2.7185, "step": 36159 }, { "epoch": 2.244707927245639, "grad_norm": 0.535806886866723, "learning_rate": 1.8097310697834725e-05, "loss": 2.6939, "step": 36160 }, { "epoch": 2.244770004345397, "grad_norm": 0.532454079810582, "learning_rate": 1.8094529937069114e-05, "loss": 2.6554, "step": 36161 }, { "epoch": 2.244832081445155, "grad_norm": 0.526614440710696, "learning_rate": 1.809174934276211e-05, "loss": 2.6925, "step": 36162 }, { "epoch": 2.2448941585449127, "grad_norm": 0.5114678794399872, "learning_rate": 1.8088968914928196e-05, "loss": 2.7949, "step": 36163 }, { "epoch": 2.2449562356446706, "grad_norm": 0.5829755048889594, "learning_rate": 1.808618865358187e-05, "loss": 2.6762, "step": 36164 }, { "epoch": 2.2450183127444285, "grad_norm": 0.5284215297801057, "learning_rate": 1.8083408558737652e-05, "loss": 2.6401, "step": 36165 }, { "epoch": 2.2450803898441865, "grad_norm": 0.5457486406048119, "learning_rate": 1.8080628630410025e-05, "loss": 2.7233, "step": 36166 }, { "epoch": 2.2451424669439444, "grad_norm": 0.5452520265904491, "learning_rate": 1.8077848868613517e-05, "loss": 2.7722, "step": 36167 }, { "epoch": 2.2452045440437023, "grad_norm": 0.5115674792625904, "learning_rate": 1.8075069273362626e-05, "loss": 2.7281, "step": 36168 }, { "epoch": 2.24526662114346, "grad_norm": 0.5168361599869589, "learning_rate": 1.8072289844671846e-05, "loss": 2.6788, "step": 36169 }, { "epoch": 2.245328698243218, "grad_norm": 0.5570171488171486, "learning_rate": 1.8069510582555665e-05, "loss": 2.7303, "step": 36170 }, { "epoch": 2.245390775342976, "grad_norm": 0.49073240355874415, "learning_rate": 1.8066731487028616e-05, "loss": 2.6148, "step": 36171 }, { "epoch": 2.245452852442734, "grad_norm": 0.4731480742992101, "learning_rate": 1.8063952558105178e-05, "loss": 2.6919, "step": 36172 }, { "epoch": 2.245514929542492, "grad_norm": 0.527854559009171, "learning_rate": 1.8061173795799853e-05, "loss": 2.6698, "step": 36173 }, { "epoch": 2.24557700664225, "grad_norm": 0.47408383318302016, "learning_rate": 1.805839520012714e-05, "loss": 2.6799, "step": 36174 }, { "epoch": 2.2456390837420077, "grad_norm": 0.49748164996134636, "learning_rate": 1.8055616771101513e-05, "loss": 2.6856, "step": 36175 }, { "epoch": 2.2457011608417656, "grad_norm": 0.5148518574702918, "learning_rate": 1.8052838508737507e-05, "loss": 2.6863, "step": 36176 }, { "epoch": 2.2457632379415235, "grad_norm": 0.49236463961871063, "learning_rate": 1.8050060413049602e-05, "loss": 2.5884, "step": 36177 }, { "epoch": 2.2458253150412815, "grad_norm": 0.49718355638821, "learning_rate": 1.8047282484052282e-05, "loss": 2.5812, "step": 36178 }, { "epoch": 2.2458873921410394, "grad_norm": 0.47651875404566146, "learning_rate": 1.804450472176003e-05, "loss": 2.6713, "step": 36179 }, { "epoch": 2.245949469240797, "grad_norm": 0.5052308998442595, "learning_rate": 1.8041727126187375e-05, "loss": 2.7158, "step": 36180 }, { "epoch": 2.2460115463405548, "grad_norm": 0.48652196184851076, "learning_rate": 1.8038949697348783e-05, "loss": 2.6549, "step": 36181 }, { "epoch": 2.2460736234403127, "grad_norm": 0.494056468326008, "learning_rate": 1.803617243525875e-05, "loss": 2.8017, "step": 36182 }, { "epoch": 2.2461357005400706, "grad_norm": 0.5174975926347843, "learning_rate": 1.8033395339931752e-05, "loss": 2.7387, "step": 36183 }, { "epoch": 2.2461977776398285, "grad_norm": 0.511965335518125, "learning_rate": 1.8030618411382306e-05, "loss": 2.7249, "step": 36184 }, { "epoch": 2.2462598547395864, "grad_norm": 0.4831367169497829, "learning_rate": 1.802784164962488e-05, "loss": 2.4927, "step": 36185 }, { "epoch": 2.2463219318393444, "grad_norm": 0.4737214467595778, "learning_rate": 1.8025065054673973e-05, "loss": 2.7437, "step": 36186 }, { "epoch": 2.2463840089391023, "grad_norm": 0.504923569825303, "learning_rate": 1.802228862654406e-05, "loss": 2.6619, "step": 36187 }, { "epoch": 2.24644608603886, "grad_norm": 0.4765208335547024, "learning_rate": 1.8019512365249615e-05, "loss": 2.631, "step": 36188 }, { "epoch": 2.246508163138618, "grad_norm": 0.5111814692204825, "learning_rate": 1.8016736270805153e-05, "loss": 2.7033, "step": 36189 }, { "epoch": 2.246570240238376, "grad_norm": 0.47867057004426644, "learning_rate": 1.8013960343225138e-05, "loss": 2.6873, "step": 36190 }, { "epoch": 2.246632317338134, "grad_norm": 0.45460783925889386, "learning_rate": 1.8011184582524064e-05, "loss": 2.7138, "step": 36191 }, { "epoch": 2.246694394437892, "grad_norm": 0.5019804357082677, "learning_rate": 1.8008408988716386e-05, "loss": 2.724, "step": 36192 }, { "epoch": 2.24675647153765, "grad_norm": 0.47412478664543245, "learning_rate": 1.8005633561816604e-05, "loss": 2.6056, "step": 36193 }, { "epoch": 2.2468185486374077, "grad_norm": 0.5011403669292944, "learning_rate": 1.8002858301839213e-05, "loss": 2.7175, "step": 36194 }, { "epoch": 2.2468806257371656, "grad_norm": 0.4590662781707286, "learning_rate": 1.800008320879868e-05, "loss": 2.5987, "step": 36195 }, { "epoch": 2.2469427028369235, "grad_norm": 0.462593498315983, "learning_rate": 1.799730828270948e-05, "loss": 2.6604, "step": 36196 }, { "epoch": 2.2470047799366815, "grad_norm": 0.44586054222765614, "learning_rate": 1.7994533523586076e-05, "loss": 2.6294, "step": 36197 }, { "epoch": 2.2470668570364394, "grad_norm": 0.5041714347529818, "learning_rate": 1.799175893144298e-05, "loss": 2.8126, "step": 36198 }, { "epoch": 2.2471289341361973, "grad_norm": 0.501958498202409, "learning_rate": 1.798898450629464e-05, "loss": 2.6524, "step": 36199 }, { "epoch": 2.247191011235955, "grad_norm": 0.47891408717851003, "learning_rate": 1.798621024815555e-05, "loss": 2.6547, "step": 36200 }, { "epoch": 2.247253088335713, "grad_norm": 0.5156641569586113, "learning_rate": 1.7983436157040162e-05, "loss": 2.633, "step": 36201 }, { "epoch": 2.247315165435471, "grad_norm": 0.48596567029346954, "learning_rate": 1.7980662232962953e-05, "loss": 2.7787, "step": 36202 }, { "epoch": 2.2473772425352285, "grad_norm": 0.48374288598085974, "learning_rate": 1.7977888475938408e-05, "loss": 2.6996, "step": 36203 }, { "epoch": 2.247439319634987, "grad_norm": 0.48039299282371983, "learning_rate": 1.7975114885980998e-05, "loss": 2.6247, "step": 36204 }, { "epoch": 2.2475013967347444, "grad_norm": 0.43561702214711673, "learning_rate": 1.797234146310519e-05, "loss": 2.6587, "step": 36205 }, { "epoch": 2.2475634738345023, "grad_norm": 0.4892095058448663, "learning_rate": 1.7969568207325433e-05, "loss": 2.6697, "step": 36206 }, { "epoch": 2.24762555093426, "grad_norm": 0.4997983019034415, "learning_rate": 1.7966795118656233e-05, "loss": 2.6373, "step": 36207 }, { "epoch": 2.247687628034018, "grad_norm": 0.4758946632799828, "learning_rate": 1.7964022197112034e-05, "loss": 2.6775, "step": 36208 }, { "epoch": 2.247749705133776, "grad_norm": 0.48821135974421437, "learning_rate": 1.796124944270731e-05, "loss": 2.6462, "step": 36209 }, { "epoch": 2.247811782233534, "grad_norm": 0.4992532965895573, "learning_rate": 1.795847685545653e-05, "loss": 2.647, "step": 36210 }, { "epoch": 2.247873859333292, "grad_norm": 0.4887033088020979, "learning_rate": 1.795570443537413e-05, "loss": 2.627, "step": 36211 }, { "epoch": 2.24793593643305, "grad_norm": 0.4682136168848735, "learning_rate": 1.7952932182474624e-05, "loss": 2.6942, "step": 36212 }, { "epoch": 2.2479980135328077, "grad_norm": 0.5041052846612267, "learning_rate": 1.7950160096772444e-05, "loss": 2.6638, "step": 36213 }, { "epoch": 2.2480600906325656, "grad_norm": 0.44992098445708156, "learning_rate": 1.7947388178282066e-05, "loss": 2.7741, "step": 36214 }, { "epoch": 2.2481221677323235, "grad_norm": 0.45699598714310413, "learning_rate": 1.7944616427017923e-05, "loss": 2.6751, "step": 36215 }, { "epoch": 2.2481842448320815, "grad_norm": 0.5234623518780452, "learning_rate": 1.7941844842994516e-05, "loss": 2.6924, "step": 36216 }, { "epoch": 2.2482463219318394, "grad_norm": 0.4823692385726592, "learning_rate": 1.7939073426226283e-05, "loss": 2.6217, "step": 36217 }, { "epoch": 2.2483083990315973, "grad_norm": 0.4414152127596606, "learning_rate": 1.7936302176727694e-05, "loss": 2.7124, "step": 36218 }, { "epoch": 2.248370476131355, "grad_norm": 0.5301936811459848, "learning_rate": 1.7933531094513178e-05, "loss": 2.7211, "step": 36219 }, { "epoch": 2.248432553231113, "grad_norm": 0.5435669985661018, "learning_rate": 1.7930760179597234e-05, "loss": 2.7319, "step": 36220 }, { "epoch": 2.248494630330871, "grad_norm": 0.46185765315144345, "learning_rate": 1.7927989431994296e-05, "loss": 2.6241, "step": 36221 }, { "epoch": 2.248556707430629, "grad_norm": 0.5282872836635212, "learning_rate": 1.7925218851718827e-05, "loss": 2.6264, "step": 36222 }, { "epoch": 2.248618784530387, "grad_norm": 0.5291048947521902, "learning_rate": 1.7922448438785277e-05, "loss": 2.7308, "step": 36223 }, { "epoch": 2.248680861630145, "grad_norm": 0.4857239704834474, "learning_rate": 1.7919678193208084e-05, "loss": 2.6508, "step": 36224 }, { "epoch": 2.2487429387299027, "grad_norm": 0.5175087360437128, "learning_rate": 1.791690811500173e-05, "loss": 2.714, "step": 36225 }, { "epoch": 2.2488050158296606, "grad_norm": 0.48248389459324975, "learning_rate": 1.7914138204180642e-05, "loss": 2.7347, "step": 36226 }, { "epoch": 2.2488670929294186, "grad_norm": 0.46932644368132886, "learning_rate": 1.7911368460759305e-05, "loss": 2.6943, "step": 36227 }, { "epoch": 2.248929170029176, "grad_norm": 0.48307064772190433, "learning_rate": 1.7908598884752142e-05, "loss": 2.7284, "step": 36228 }, { "epoch": 2.248991247128934, "grad_norm": 0.5371305426848043, "learning_rate": 1.790582947617359e-05, "loss": 2.7051, "step": 36229 }, { "epoch": 2.249053324228692, "grad_norm": 0.4725099132143803, "learning_rate": 1.790306023503814e-05, "loss": 2.7123, "step": 36230 }, { "epoch": 2.24911540132845, "grad_norm": 0.5042845023459828, "learning_rate": 1.790029116136021e-05, "loss": 2.6604, "step": 36231 }, { "epoch": 2.2491774784282077, "grad_norm": 0.4494058136595368, "learning_rate": 1.789752225515426e-05, "loss": 2.632, "step": 36232 }, { "epoch": 2.2492395555279656, "grad_norm": 0.484893229348156, "learning_rate": 1.789475351643471e-05, "loss": 2.6804, "step": 36233 }, { "epoch": 2.2493016326277235, "grad_norm": 0.5229335465562733, "learning_rate": 1.7891984945216043e-05, "loss": 2.6863, "step": 36234 }, { "epoch": 2.2493637097274815, "grad_norm": 0.4462712657894257, "learning_rate": 1.788921654151268e-05, "loss": 2.6519, "step": 36235 }, { "epoch": 2.2494257868272394, "grad_norm": 0.4397563312176683, "learning_rate": 1.7886448305339077e-05, "loss": 2.686, "step": 36236 }, { "epoch": 2.2494878639269973, "grad_norm": 0.5244159662933174, "learning_rate": 1.7883680236709664e-05, "loss": 2.7158, "step": 36237 }, { "epoch": 2.249549941026755, "grad_norm": 0.4727454140792231, "learning_rate": 1.7880912335638872e-05, "loss": 2.74, "step": 36238 }, { "epoch": 2.249612018126513, "grad_norm": 0.4650174305325354, "learning_rate": 1.7878144602141178e-05, "loss": 2.7002, "step": 36239 }, { "epoch": 2.249674095226271, "grad_norm": 0.4620852808342298, "learning_rate": 1.7875377036230996e-05, "loss": 2.6945, "step": 36240 }, { "epoch": 2.249736172326029, "grad_norm": 0.48574578222916415, "learning_rate": 1.7872609637922773e-05, "loss": 2.7589, "step": 36241 }, { "epoch": 2.249798249425787, "grad_norm": 0.5002880160972496, "learning_rate": 1.786984240723093e-05, "loss": 2.6282, "step": 36242 }, { "epoch": 2.249860326525545, "grad_norm": 0.4980562097011246, "learning_rate": 1.7867075344169933e-05, "loss": 2.6702, "step": 36243 }, { "epoch": 2.2499224036253027, "grad_norm": 0.5426043378459006, "learning_rate": 1.7864308448754203e-05, "loss": 2.6508, "step": 36244 }, { "epoch": 2.2499844807250606, "grad_norm": 0.4952099053060092, "learning_rate": 1.7861541720998182e-05, "loss": 2.6836, "step": 36245 }, { "epoch": 2.2500465578248185, "grad_norm": 0.6183862030899067, "learning_rate": 1.785877516091628e-05, "loss": 2.6888, "step": 36246 }, { "epoch": 2.2501086349245765, "grad_norm": 0.4581243045588233, "learning_rate": 1.7856008768522963e-05, "loss": 2.73, "step": 36247 }, { "epoch": 2.2501707120243344, "grad_norm": 0.5257462783224641, "learning_rate": 1.785324254383266e-05, "loss": 2.7012, "step": 36248 }, { "epoch": 2.2502327891240923, "grad_norm": 0.5700734303675689, "learning_rate": 1.7850476486859785e-05, "loss": 2.6757, "step": 36249 }, { "epoch": 2.25029486622385, "grad_norm": 0.5342769499386657, "learning_rate": 1.784771059761879e-05, "loss": 2.6627, "step": 36250 }, { "epoch": 2.2503569433236077, "grad_norm": 0.547722405153372, "learning_rate": 1.784494487612407e-05, "loss": 2.7929, "step": 36251 }, { "epoch": 2.250419020423366, "grad_norm": 0.529999439131242, "learning_rate": 1.7842179322390097e-05, "loss": 2.6568, "step": 36252 }, { "epoch": 2.2504810975231235, "grad_norm": 0.5294836636725729, "learning_rate": 1.7839413936431277e-05, "loss": 2.6369, "step": 36253 }, { "epoch": 2.2505431746228814, "grad_norm": 0.5120703093804978, "learning_rate": 1.7836648718262044e-05, "loss": 2.7723, "step": 36254 }, { "epoch": 2.2506052517226394, "grad_norm": 0.55356984973278, "learning_rate": 1.783388366789681e-05, "loss": 2.6371, "step": 36255 }, { "epoch": 2.2506673288223973, "grad_norm": 0.5307185656456199, "learning_rate": 1.7831118785350025e-05, "loss": 2.676, "step": 36256 }, { "epoch": 2.250729405922155, "grad_norm": 0.4934214911542583, "learning_rate": 1.7828354070636105e-05, "loss": 2.6434, "step": 36257 }, { "epoch": 2.250791483021913, "grad_norm": 0.547177367407958, "learning_rate": 1.7825589523769475e-05, "loss": 2.7775, "step": 36258 }, { "epoch": 2.250853560121671, "grad_norm": 0.5393820382252443, "learning_rate": 1.782282514476453e-05, "loss": 2.7537, "step": 36259 }, { "epoch": 2.250915637221429, "grad_norm": 0.5427243073565989, "learning_rate": 1.782006093363574e-05, "loss": 2.6856, "step": 36260 }, { "epoch": 2.250977714321187, "grad_norm": 0.5015961383222992, "learning_rate": 1.781729689039749e-05, "loss": 2.6076, "step": 36261 }, { "epoch": 2.251039791420945, "grad_norm": 0.5465034312483417, "learning_rate": 1.7814533015064227e-05, "loss": 2.7311, "step": 36262 }, { "epoch": 2.2511018685207027, "grad_norm": 0.5697340971638435, "learning_rate": 1.7811769307650362e-05, "loss": 2.6632, "step": 36263 }, { "epoch": 2.2511639456204606, "grad_norm": 0.5033428087932057, "learning_rate": 1.7809005768170306e-05, "loss": 2.7323, "step": 36264 }, { "epoch": 2.2512260227202185, "grad_norm": 0.5738498514839147, "learning_rate": 1.780624239663847e-05, "loss": 2.7884, "step": 36265 }, { "epoch": 2.2512880998199765, "grad_norm": 0.4597527423567346, "learning_rate": 1.7803479193069294e-05, "loss": 2.6906, "step": 36266 }, { "epoch": 2.2513501769197344, "grad_norm": 0.5106569587390254, "learning_rate": 1.780071615747718e-05, "loss": 2.6328, "step": 36267 }, { "epoch": 2.2514122540194923, "grad_norm": 0.5878843643007193, "learning_rate": 1.7797953289876558e-05, "loss": 2.7028, "step": 36268 }, { "epoch": 2.25147433111925, "grad_norm": 0.5090945858617478, "learning_rate": 1.7795190590281806e-05, "loss": 2.634, "step": 36269 }, { "epoch": 2.251536408219008, "grad_norm": 0.5047396488233611, "learning_rate": 1.7792428058707382e-05, "loss": 2.6838, "step": 36270 }, { "epoch": 2.251598485318766, "grad_norm": 0.5032402190969639, "learning_rate": 1.778966569516768e-05, "loss": 2.6002, "step": 36271 }, { "epoch": 2.251660562418524, "grad_norm": 0.4952077538237219, "learning_rate": 1.778690349967711e-05, "loss": 2.6291, "step": 36272 }, { "epoch": 2.251722639518282, "grad_norm": 0.5096023707396069, "learning_rate": 1.778414147225008e-05, "loss": 2.572, "step": 36273 }, { "epoch": 2.2517847166180394, "grad_norm": 0.497825528493392, "learning_rate": 1.7781379612900994e-05, "loss": 2.7147, "step": 36274 }, { "epoch": 2.2518467937177977, "grad_norm": 0.48731028337648447, "learning_rate": 1.777861792164429e-05, "loss": 2.6455, "step": 36275 }, { "epoch": 2.251908870817555, "grad_norm": 0.5324739213834857, "learning_rate": 1.7775856398494355e-05, "loss": 2.7128, "step": 36276 }, { "epoch": 2.251970947917313, "grad_norm": 0.5161847342091309, "learning_rate": 1.7773095043465598e-05, "loss": 2.6574, "step": 36277 }, { "epoch": 2.252033025017071, "grad_norm": 0.5379561806084837, "learning_rate": 1.7770333856572417e-05, "loss": 2.6546, "step": 36278 }, { "epoch": 2.252095102116829, "grad_norm": 0.5219371416019657, "learning_rate": 1.776757283782924e-05, "loss": 2.6569, "step": 36279 }, { "epoch": 2.252157179216587, "grad_norm": 0.49325994182743166, "learning_rate": 1.7764811987250463e-05, "loss": 2.6831, "step": 36280 }, { "epoch": 2.252219256316345, "grad_norm": 0.5581378276585374, "learning_rate": 1.7762051304850485e-05, "loss": 2.6255, "step": 36281 }, { "epoch": 2.2522813334161027, "grad_norm": 0.5158726646494339, "learning_rate": 1.7759290790643694e-05, "loss": 2.6027, "step": 36282 }, { "epoch": 2.2523434105158606, "grad_norm": 0.5720305497040952, "learning_rate": 1.7756530444644532e-05, "loss": 2.7499, "step": 36283 }, { "epoch": 2.2524054876156185, "grad_norm": 0.47864926087567355, "learning_rate": 1.775377026686737e-05, "loss": 2.6155, "step": 36284 }, { "epoch": 2.2524675647153765, "grad_norm": 0.5528506638554584, "learning_rate": 1.775101025732662e-05, "loss": 2.7013, "step": 36285 }, { "epoch": 2.2525296418151344, "grad_norm": 0.5164756693616278, "learning_rate": 1.7748250416036682e-05, "loss": 2.6915, "step": 36286 }, { "epoch": 2.2525917189148923, "grad_norm": 0.5279309147500505, "learning_rate": 1.774549074301193e-05, "loss": 2.7284, "step": 36287 }, { "epoch": 2.25265379601465, "grad_norm": 0.5283272226273057, "learning_rate": 1.7742731238266797e-05, "loss": 2.7265, "step": 36288 }, { "epoch": 2.252715873114408, "grad_norm": 0.46944911003459106, "learning_rate": 1.773997190181567e-05, "loss": 2.7916, "step": 36289 }, { "epoch": 2.252777950214166, "grad_norm": 0.5180931551802346, "learning_rate": 1.7737212733672942e-05, "loss": 2.6327, "step": 36290 }, { "epoch": 2.252840027313924, "grad_norm": 0.5073608737660381, "learning_rate": 1.773445373385299e-05, "loss": 2.6543, "step": 36291 }, { "epoch": 2.252902104413682, "grad_norm": 0.45536634625556077, "learning_rate": 1.773169490237022e-05, "loss": 2.6755, "step": 36292 }, { "epoch": 2.25296418151344, "grad_norm": 0.5107924835939558, "learning_rate": 1.7728936239239053e-05, "loss": 2.6922, "step": 36293 }, { "epoch": 2.2530262586131977, "grad_norm": 0.48991155444105466, "learning_rate": 1.772617774447386e-05, "loss": 2.7287, "step": 36294 }, { "epoch": 2.2530883357129556, "grad_norm": 0.4869087919767852, "learning_rate": 1.772341941808903e-05, "loss": 2.6713, "step": 36295 }, { "epoch": 2.2531504128127136, "grad_norm": 0.4942345963336762, "learning_rate": 1.772066126009894e-05, "loss": 2.7537, "step": 36296 }, { "epoch": 2.2532124899124715, "grad_norm": 0.5213199313340339, "learning_rate": 1.7717903270518017e-05, "loss": 2.6856, "step": 36297 }, { "epoch": 2.2532745670122294, "grad_norm": 0.45146196295536783, "learning_rate": 1.7715145449360627e-05, "loss": 2.656, "step": 36298 }, { "epoch": 2.253336644111987, "grad_norm": 0.46164000044538517, "learning_rate": 1.7712387796641163e-05, "loss": 2.6381, "step": 36299 }, { "epoch": 2.2533987212117452, "grad_norm": 0.5049044458709195, "learning_rate": 1.7709630312374004e-05, "loss": 2.5824, "step": 36300 }, { "epoch": 2.2534607983115027, "grad_norm": 0.5017044995332546, "learning_rate": 1.7706872996573526e-05, "loss": 2.6752, "step": 36301 }, { "epoch": 2.2535228754112606, "grad_norm": 0.4874122974783424, "learning_rate": 1.770411584925415e-05, "loss": 2.6011, "step": 36302 }, { "epoch": 2.2535849525110185, "grad_norm": 0.5175050503949654, "learning_rate": 1.7701358870430245e-05, "loss": 2.7158, "step": 36303 }, { "epoch": 2.2536470296107765, "grad_norm": 0.47584802193470527, "learning_rate": 1.7698602060116187e-05, "loss": 2.6304, "step": 36304 }, { "epoch": 2.2537091067105344, "grad_norm": 0.49616652428011293, "learning_rate": 1.769584541832635e-05, "loss": 2.6722, "step": 36305 }, { "epoch": 2.2537711838102923, "grad_norm": 0.4999510025812476, "learning_rate": 1.7693088945075143e-05, "loss": 2.7444, "step": 36306 }, { "epoch": 2.25383326091005, "grad_norm": 0.5047124578959893, "learning_rate": 1.7690332640376932e-05, "loss": 2.7211, "step": 36307 }, { "epoch": 2.253895338009808, "grad_norm": 0.4783033717665282, "learning_rate": 1.7687576504246107e-05, "loss": 2.6452, "step": 36308 }, { "epoch": 2.253957415109566, "grad_norm": 0.44498583875583014, "learning_rate": 1.7684820536697035e-05, "loss": 2.7256, "step": 36309 }, { "epoch": 2.254019492209324, "grad_norm": 0.5201287813001065, "learning_rate": 1.768206473774408e-05, "loss": 2.6841, "step": 36310 }, { "epoch": 2.254081569309082, "grad_norm": 0.49194935683766594, "learning_rate": 1.767930910740166e-05, "loss": 2.7356, "step": 36311 }, { "epoch": 2.25414364640884, "grad_norm": 0.4713906612986089, "learning_rate": 1.767655364568413e-05, "loss": 2.5521, "step": 36312 }, { "epoch": 2.2542057235085977, "grad_norm": 0.5324675408172803, "learning_rate": 1.7673798352605865e-05, "loss": 2.7188, "step": 36313 }, { "epoch": 2.2542678006083556, "grad_norm": 0.5024274670267892, "learning_rate": 1.767104322818122e-05, "loss": 2.5993, "step": 36314 }, { "epoch": 2.2543298777081135, "grad_norm": 0.4899615406250726, "learning_rate": 1.766828827242461e-05, "loss": 2.7905, "step": 36315 }, { "epoch": 2.2543919548078715, "grad_norm": 0.5553416035527875, "learning_rate": 1.7665533485350393e-05, "loss": 2.7029, "step": 36316 }, { "epoch": 2.2544540319076294, "grad_norm": 0.4539976103558017, "learning_rate": 1.7662778866972936e-05, "loss": 2.5799, "step": 36317 }, { "epoch": 2.2545161090073873, "grad_norm": 0.49386092702404666, "learning_rate": 1.7660024417306593e-05, "loss": 2.6507, "step": 36318 }, { "epoch": 2.254578186107145, "grad_norm": 0.5765334465749646, "learning_rate": 1.765727013636577e-05, "loss": 2.7686, "step": 36319 }, { "epoch": 2.254640263206903, "grad_norm": 0.4873025164049362, "learning_rate": 1.7654516024164826e-05, "loss": 2.7323, "step": 36320 }, { "epoch": 2.254702340306661, "grad_norm": 0.46834075865283725, "learning_rate": 1.7651762080718115e-05, "loss": 2.6653, "step": 36321 }, { "epoch": 2.2547644174064185, "grad_norm": 0.5691245784343228, "learning_rate": 1.7649008306040022e-05, "loss": 2.6788, "step": 36322 }, { "epoch": 2.254826494506177, "grad_norm": 0.503323683988317, "learning_rate": 1.7646254700144883e-05, "loss": 2.6087, "step": 36323 }, { "epoch": 2.2548885716059344, "grad_norm": 0.4566424256379821, "learning_rate": 1.7643501263047092e-05, "loss": 2.6296, "step": 36324 }, { "epoch": 2.2549506487056923, "grad_norm": 0.5063886782179047, "learning_rate": 1.764074799476102e-05, "loss": 2.7334, "step": 36325 }, { "epoch": 2.25501272580545, "grad_norm": 0.5192282230363674, "learning_rate": 1.763799489530103e-05, "loss": 2.6244, "step": 36326 }, { "epoch": 2.255074802905208, "grad_norm": 0.48349952026213205, "learning_rate": 1.763524196468147e-05, "loss": 2.7368, "step": 36327 }, { "epoch": 2.255136880004966, "grad_norm": 0.4807648715408473, "learning_rate": 1.7632489202916698e-05, "loss": 2.7541, "step": 36328 }, { "epoch": 2.255198957104724, "grad_norm": 0.5121823605831752, "learning_rate": 1.7629736610021097e-05, "loss": 2.7564, "step": 36329 }, { "epoch": 2.255261034204482, "grad_norm": 0.49844014217908117, "learning_rate": 1.7626984186009027e-05, "loss": 2.6533, "step": 36330 }, { "epoch": 2.25532311130424, "grad_norm": 0.5005524087156131, "learning_rate": 1.762423193089483e-05, "loss": 2.6932, "step": 36331 }, { "epoch": 2.2553851884039977, "grad_norm": 0.49208196386533787, "learning_rate": 1.7621479844692863e-05, "loss": 2.772, "step": 36332 }, { "epoch": 2.2554472655037556, "grad_norm": 0.5161868992959141, "learning_rate": 1.7618727927417518e-05, "loss": 2.666, "step": 36333 }, { "epoch": 2.2555093426035135, "grad_norm": 0.5103070829142505, "learning_rate": 1.761597617908312e-05, "loss": 2.6741, "step": 36334 }, { "epoch": 2.2555714197032715, "grad_norm": 0.5323967337882166, "learning_rate": 1.761322459970404e-05, "loss": 2.7599, "step": 36335 }, { "epoch": 2.2556334968030294, "grad_norm": 0.47565818410196614, "learning_rate": 1.7610473189294628e-05, "loss": 2.7026, "step": 36336 }, { "epoch": 2.2556955739027873, "grad_norm": 0.5008683710594954, "learning_rate": 1.7607721947869226e-05, "loss": 2.6886, "step": 36337 }, { "epoch": 2.255757651002545, "grad_norm": 0.5197999015823423, "learning_rate": 1.760497087544222e-05, "loss": 2.6382, "step": 36338 }, { "epoch": 2.255819728102303, "grad_norm": 0.49488472983557735, "learning_rate": 1.7602219972027943e-05, "loss": 2.6523, "step": 36339 }, { "epoch": 2.255881805202061, "grad_norm": 0.5184232038627312, "learning_rate": 1.7599469237640746e-05, "loss": 2.6909, "step": 36340 }, { "epoch": 2.255943882301819, "grad_norm": 0.4745020803982759, "learning_rate": 1.759671867229497e-05, "loss": 2.659, "step": 36341 }, { "epoch": 2.256005959401577, "grad_norm": 0.48013044092421997, "learning_rate": 1.7593968276004995e-05, "loss": 2.6308, "step": 36342 }, { "epoch": 2.256068036501335, "grad_norm": 0.44424679040044557, "learning_rate": 1.759121804878516e-05, "loss": 2.6473, "step": 36343 }, { "epoch": 2.2561301136010927, "grad_norm": 0.46386116425468565, "learning_rate": 1.7588467990649805e-05, "loss": 2.6339, "step": 36344 }, { "epoch": 2.2561921907008506, "grad_norm": 0.48226612681194814, "learning_rate": 1.7585718101613273e-05, "loss": 2.6571, "step": 36345 }, { "epoch": 2.2562542678006086, "grad_norm": 0.46788936126278446, "learning_rate": 1.758296838168991e-05, "loss": 2.7392, "step": 36346 }, { "epoch": 2.256316344900366, "grad_norm": 0.5049154245258789, "learning_rate": 1.7580218830894084e-05, "loss": 2.7135, "step": 36347 }, { "epoch": 2.2563784220001244, "grad_norm": 0.5017945747970133, "learning_rate": 1.7577469449240127e-05, "loss": 2.6123, "step": 36348 }, { "epoch": 2.256440499099882, "grad_norm": 0.47347787112094597, "learning_rate": 1.7574720236742383e-05, "loss": 2.6229, "step": 36349 }, { "epoch": 2.25650257619964, "grad_norm": 0.46353297993382103, "learning_rate": 1.7571971193415177e-05, "loss": 2.662, "step": 36350 }, { "epoch": 2.2565646532993977, "grad_norm": 0.4872072386153394, "learning_rate": 1.756922231927289e-05, "loss": 2.6768, "step": 36351 }, { "epoch": 2.2566267303991556, "grad_norm": 0.4918212587645808, "learning_rate": 1.7566473614329837e-05, "loss": 2.6808, "step": 36352 }, { "epoch": 2.2566888074989135, "grad_norm": 0.49637768347136574, "learning_rate": 1.756372507860037e-05, "loss": 2.7178, "step": 36353 }, { "epoch": 2.2567508845986715, "grad_norm": 0.4802089391885219, "learning_rate": 1.7560976712098804e-05, "loss": 2.6579, "step": 36354 }, { "epoch": 2.2568129616984294, "grad_norm": 0.48887812944725895, "learning_rate": 1.7558228514839514e-05, "loss": 2.6615, "step": 36355 }, { "epoch": 2.2568750387981873, "grad_norm": 0.4893119614535194, "learning_rate": 1.755548048683682e-05, "loss": 2.6638, "step": 36356 }, { "epoch": 2.256937115897945, "grad_norm": 0.49196836346485884, "learning_rate": 1.755273262810505e-05, "loss": 2.6731, "step": 36357 }, { "epoch": 2.256999192997703, "grad_norm": 0.5140858421707627, "learning_rate": 1.7549984938658564e-05, "loss": 2.6906, "step": 36358 }, { "epoch": 2.257061270097461, "grad_norm": 0.5103761663393198, "learning_rate": 1.754723741851168e-05, "loss": 2.7075, "step": 36359 }, { "epoch": 2.257123347197219, "grad_norm": 0.4512037951254437, "learning_rate": 1.7544490067678726e-05, "loss": 2.6016, "step": 36360 }, { "epoch": 2.257185424296977, "grad_norm": 0.5055559935003114, "learning_rate": 1.7541742886174055e-05, "loss": 2.7108, "step": 36361 }, { "epoch": 2.257247501396735, "grad_norm": 0.46998766583633267, "learning_rate": 1.7538995874012e-05, "loss": 2.5975, "step": 36362 }, { "epoch": 2.2573095784964927, "grad_norm": 0.47706961618995936, "learning_rate": 1.753624903120688e-05, "loss": 2.6985, "step": 36363 }, { "epoch": 2.2573716555962506, "grad_norm": 0.4783520165836432, "learning_rate": 1.753350235777301e-05, "loss": 2.6548, "step": 36364 }, { "epoch": 2.2574337326960086, "grad_norm": 0.4956362259063696, "learning_rate": 1.753075585372476e-05, "loss": 2.6971, "step": 36365 }, { "epoch": 2.2574958097957665, "grad_norm": 0.49896398141593595, "learning_rate": 1.7528009519076437e-05, "loss": 2.7495, "step": 36366 }, { "epoch": 2.2575578868955244, "grad_norm": 0.4897102959006423, "learning_rate": 1.7525263353842374e-05, "loss": 2.6791, "step": 36367 }, { "epoch": 2.2576199639952823, "grad_norm": 0.5444751398630837, "learning_rate": 1.7522517358036876e-05, "loss": 2.6802, "step": 36368 }, { "epoch": 2.2576820410950402, "grad_norm": 0.47698060440021184, "learning_rate": 1.7519771531674306e-05, "loss": 2.5933, "step": 36369 }, { "epoch": 2.2577441181947977, "grad_norm": 0.4720087108262875, "learning_rate": 1.7517025874768977e-05, "loss": 2.6727, "step": 36370 }, { "epoch": 2.257806195294556, "grad_norm": 0.5076212302704147, "learning_rate": 1.751428038733521e-05, "loss": 2.6772, "step": 36371 }, { "epoch": 2.2578682723943135, "grad_norm": 0.47932808859119586, "learning_rate": 1.751153506938732e-05, "loss": 2.714, "step": 36372 }, { "epoch": 2.2579303494940715, "grad_norm": 0.45665616350845983, "learning_rate": 1.750878992093963e-05, "loss": 2.6475, "step": 36373 }, { "epoch": 2.2579924265938294, "grad_norm": 0.4994401927164599, "learning_rate": 1.7506044942006478e-05, "loss": 2.6649, "step": 36374 }, { "epoch": 2.2580545036935873, "grad_norm": 0.4688560048086864, "learning_rate": 1.7503300132602184e-05, "loss": 2.7195, "step": 36375 }, { "epoch": 2.258116580793345, "grad_norm": 0.5088943694627943, "learning_rate": 1.750055549274106e-05, "loss": 2.7268, "step": 36376 }, { "epoch": 2.258178657893103, "grad_norm": 0.5057035228174308, "learning_rate": 1.7497811022437415e-05, "loss": 2.7365, "step": 36377 }, { "epoch": 2.258240734992861, "grad_norm": 0.4596898602309016, "learning_rate": 1.7495066721705595e-05, "loss": 2.7131, "step": 36378 }, { "epoch": 2.258302812092619, "grad_norm": 0.46218019202427657, "learning_rate": 1.74923225905599e-05, "loss": 2.7072, "step": 36379 }, { "epoch": 2.258364889192377, "grad_norm": 0.5246966575564082, "learning_rate": 1.748957862901465e-05, "loss": 2.6508, "step": 36380 }, { "epoch": 2.258426966292135, "grad_norm": 0.43879266911792625, "learning_rate": 1.748683483708415e-05, "loss": 2.7182, "step": 36381 }, { "epoch": 2.2584890433918927, "grad_norm": 0.5002452725592672, "learning_rate": 1.7484091214782734e-05, "loss": 2.5949, "step": 36382 }, { "epoch": 2.2585511204916506, "grad_norm": 0.426478034519317, "learning_rate": 1.748134776212471e-05, "loss": 2.6548, "step": 36383 }, { "epoch": 2.2586131975914086, "grad_norm": 0.4655956950342604, "learning_rate": 1.747860447912439e-05, "loss": 2.6996, "step": 36384 }, { "epoch": 2.2586752746911665, "grad_norm": 0.510978162821467, "learning_rate": 1.7475861365796087e-05, "loss": 2.8204, "step": 36385 }, { "epoch": 2.2587373517909244, "grad_norm": 0.46664292099121374, "learning_rate": 1.747311842215409e-05, "loss": 2.6664, "step": 36386 }, { "epoch": 2.2587994288906823, "grad_norm": 0.4650931931240954, "learning_rate": 1.747037564821275e-05, "loss": 2.6105, "step": 36387 }, { "epoch": 2.25886150599044, "grad_norm": 0.5011116128371792, "learning_rate": 1.7467633043986352e-05, "loss": 2.6671, "step": 36388 }, { "epoch": 2.258923583090198, "grad_norm": 0.4540345993711375, "learning_rate": 1.7464890609489214e-05, "loss": 2.6724, "step": 36389 }, { "epoch": 2.258985660189956, "grad_norm": 0.5075717648845591, "learning_rate": 1.7462148344735628e-05, "loss": 2.6601, "step": 36390 }, { "epoch": 2.259047737289714, "grad_norm": 0.4669150358547526, "learning_rate": 1.7459406249739906e-05, "loss": 2.6606, "step": 36391 }, { "epoch": 2.259109814389472, "grad_norm": 0.5005202269244482, "learning_rate": 1.745666432451638e-05, "loss": 2.6823, "step": 36392 }, { "epoch": 2.25917189148923, "grad_norm": 0.49338615628703875, "learning_rate": 1.745392256907934e-05, "loss": 2.6389, "step": 36393 }, { "epoch": 2.2592339685889877, "grad_norm": 0.4787839643168528, "learning_rate": 1.7451180983443082e-05, "loss": 2.7181, "step": 36394 }, { "epoch": 2.259296045688745, "grad_norm": 0.5394562375188114, "learning_rate": 1.7448439567621915e-05, "loss": 2.6814, "step": 36395 }, { "epoch": 2.2593581227885036, "grad_norm": 0.5198316637703194, "learning_rate": 1.7445698321630122e-05, "loss": 2.6629, "step": 36396 }, { "epoch": 2.259420199888261, "grad_norm": 0.47826132731924653, "learning_rate": 1.7442957245482045e-05, "loss": 2.6414, "step": 36397 }, { "epoch": 2.259482276988019, "grad_norm": 0.5025210303560325, "learning_rate": 1.7440216339191962e-05, "loss": 2.6377, "step": 36398 }, { "epoch": 2.259544354087777, "grad_norm": 0.5131303086279192, "learning_rate": 1.7437475602774173e-05, "loss": 2.5904, "step": 36399 }, { "epoch": 2.259606431187535, "grad_norm": 0.49102612880982766, "learning_rate": 1.7434735036242968e-05, "loss": 2.645, "step": 36400 }, { "epoch": 2.2596685082872927, "grad_norm": 0.4967374755267279, "learning_rate": 1.7431994639612663e-05, "loss": 2.6123, "step": 36401 }, { "epoch": 2.2597305853870506, "grad_norm": 0.5361223341399051, "learning_rate": 1.7429254412897557e-05, "loss": 2.7535, "step": 36402 }, { "epoch": 2.2597926624868085, "grad_norm": 0.46479276169095496, "learning_rate": 1.7426514356111928e-05, "loss": 2.7223, "step": 36403 }, { "epoch": 2.2598547395865665, "grad_norm": 0.5139301941426704, "learning_rate": 1.7423774469270073e-05, "loss": 2.7303, "step": 36404 }, { "epoch": 2.2599168166863244, "grad_norm": 0.5060097018129766, "learning_rate": 1.7421034752386307e-05, "loss": 2.6524, "step": 36405 }, { "epoch": 2.2599788937860823, "grad_norm": 0.4603020927511324, "learning_rate": 1.741829520547491e-05, "loss": 2.7068, "step": 36406 }, { "epoch": 2.26004097088584, "grad_norm": 0.43287673294531254, "learning_rate": 1.741555582855018e-05, "loss": 2.6733, "step": 36407 }, { "epoch": 2.260103047985598, "grad_norm": 0.4828716143124766, "learning_rate": 1.7412816621626405e-05, "loss": 2.6574, "step": 36408 }, { "epoch": 2.260165125085356, "grad_norm": 0.45263948408843785, "learning_rate": 1.741007758471786e-05, "loss": 2.6435, "step": 36409 }, { "epoch": 2.260227202185114, "grad_norm": 0.4842628226052197, "learning_rate": 1.7407338717838867e-05, "loss": 2.7301, "step": 36410 }, { "epoch": 2.260289279284872, "grad_norm": 0.4879130997379444, "learning_rate": 1.7404600021003703e-05, "loss": 2.7808, "step": 36411 }, { "epoch": 2.26035135638463, "grad_norm": 0.4680820633585384, "learning_rate": 1.7401861494226652e-05, "loss": 2.5756, "step": 36412 }, { "epoch": 2.2604134334843877, "grad_norm": 0.4998471503277736, "learning_rate": 1.7399123137521983e-05, "loss": 2.7371, "step": 36413 }, { "epoch": 2.2604755105841456, "grad_norm": 0.4557563737286514, "learning_rate": 1.739638495090402e-05, "loss": 2.7202, "step": 36414 }, { "epoch": 2.2605375876839036, "grad_norm": 0.5137054374376605, "learning_rate": 1.7393646934387036e-05, "loss": 2.6714, "step": 36415 }, { "epoch": 2.2605996647836615, "grad_norm": 0.48343553497072905, "learning_rate": 1.7390909087985308e-05, "loss": 2.8078, "step": 36416 }, { "epoch": 2.2606617418834194, "grad_norm": 0.47539466285714715, "learning_rate": 1.7388171411713105e-05, "loss": 2.704, "step": 36417 }, { "epoch": 2.260723818983177, "grad_norm": 0.5214876039097865, "learning_rate": 1.7385433905584752e-05, "loss": 2.6717, "step": 36418 }, { "epoch": 2.2607858960829352, "grad_norm": 0.5193650689765058, "learning_rate": 1.7382696569614503e-05, "loss": 2.6832, "step": 36419 }, { "epoch": 2.2608479731826927, "grad_norm": 0.49012536284460767, "learning_rate": 1.7379959403816643e-05, "loss": 2.6844, "step": 36420 }, { "epoch": 2.2609100502824506, "grad_norm": 0.4869647627415554, "learning_rate": 1.7377222408205456e-05, "loss": 2.6752, "step": 36421 }, { "epoch": 2.2609721273822085, "grad_norm": 0.4878659274933652, "learning_rate": 1.7374485582795202e-05, "loss": 2.7095, "step": 36422 }, { "epoch": 2.2610342044819665, "grad_norm": 0.49835810193361035, "learning_rate": 1.737174892760018e-05, "loss": 2.7815, "step": 36423 }, { "epoch": 2.2610962815817244, "grad_norm": 0.45647881204981516, "learning_rate": 1.7369012442634673e-05, "loss": 2.6505, "step": 36424 }, { "epoch": 2.2611583586814823, "grad_norm": 0.4516279054416815, "learning_rate": 1.7366276127912957e-05, "loss": 2.6792, "step": 36425 }, { "epoch": 2.26122043578124, "grad_norm": 0.4674856299981431, "learning_rate": 1.7363539983449295e-05, "loss": 2.6613, "step": 36426 }, { "epoch": 2.261282512880998, "grad_norm": 0.4902257272633917, "learning_rate": 1.7360804009257957e-05, "loss": 2.6947, "step": 36427 }, { "epoch": 2.261344589980756, "grad_norm": 0.44107738836132954, "learning_rate": 1.735806820535324e-05, "loss": 2.5854, "step": 36428 }, { "epoch": 2.261406667080514, "grad_norm": 0.4870871017980322, "learning_rate": 1.7355332571749412e-05, "loss": 2.6786, "step": 36429 }, { "epoch": 2.261468744180272, "grad_norm": 0.5040934015534985, "learning_rate": 1.7352597108460728e-05, "loss": 2.6604, "step": 36430 }, { "epoch": 2.26153082128003, "grad_norm": 0.5077480568710102, "learning_rate": 1.734986181550148e-05, "loss": 2.6061, "step": 36431 }, { "epoch": 2.2615928983797877, "grad_norm": 0.48902619565333255, "learning_rate": 1.7347126692885907e-05, "loss": 2.6301, "step": 36432 }, { "epoch": 2.2616549754795456, "grad_norm": 0.5706271518528822, "learning_rate": 1.7344391740628318e-05, "loss": 2.6786, "step": 36433 }, { "epoch": 2.2617170525793036, "grad_norm": 0.4619522078976107, "learning_rate": 1.7341656958742963e-05, "loss": 2.7395, "step": 36434 }, { "epoch": 2.2617791296790615, "grad_norm": 0.47464417694969924, "learning_rate": 1.733892234724412e-05, "loss": 2.7197, "step": 36435 }, { "epoch": 2.2618412067788194, "grad_norm": 0.5514650901819553, "learning_rate": 1.7336187906146024e-05, "loss": 2.6893, "step": 36436 }, { "epoch": 2.2619032838785773, "grad_norm": 0.4787651034480212, "learning_rate": 1.7333453635462983e-05, "loss": 2.6316, "step": 36437 }, { "epoch": 2.2619653609783352, "grad_norm": 0.5471481186198661, "learning_rate": 1.7330719535209245e-05, "loss": 2.6293, "step": 36438 }, { "epoch": 2.262027438078093, "grad_norm": 0.5701505613997179, "learning_rate": 1.7327985605399072e-05, "loss": 2.7891, "step": 36439 }, { "epoch": 2.262089515177851, "grad_norm": 0.4984422348658566, "learning_rate": 1.7325251846046714e-05, "loss": 2.7358, "step": 36440 }, { "epoch": 2.262151592277609, "grad_norm": 0.5569145198032696, "learning_rate": 1.7322518257166465e-05, "loss": 2.6942, "step": 36441 }, { "epoch": 2.262213669377367, "grad_norm": 0.5013388351906204, "learning_rate": 1.7319784838772573e-05, "loss": 2.6343, "step": 36442 }, { "epoch": 2.2622757464771244, "grad_norm": 0.4859146750630715, "learning_rate": 1.7317051590879298e-05, "loss": 2.6733, "step": 36443 }, { "epoch": 2.2623378235768827, "grad_norm": 0.46469307709498053, "learning_rate": 1.7314318513500897e-05, "loss": 2.6594, "step": 36444 }, { "epoch": 2.26239990067664, "grad_norm": 0.4705223116099498, "learning_rate": 1.7311585606651615e-05, "loss": 2.644, "step": 36445 }, { "epoch": 2.262461977776398, "grad_norm": 0.5020813240179902, "learning_rate": 1.7308852870345742e-05, "loss": 2.7357, "step": 36446 }, { "epoch": 2.262524054876156, "grad_norm": 0.5128298337751084, "learning_rate": 1.730612030459752e-05, "loss": 2.5753, "step": 36447 }, { "epoch": 2.262586131975914, "grad_norm": 0.48719913356219297, "learning_rate": 1.7303387909421205e-05, "loss": 2.7253, "step": 36448 }, { "epoch": 2.262648209075672, "grad_norm": 0.4656143532886288, "learning_rate": 1.7300655684831036e-05, "loss": 2.6492, "step": 36449 }, { "epoch": 2.26271028617543, "grad_norm": 0.45849147871495904, "learning_rate": 1.72979236308413e-05, "loss": 2.5921, "step": 36450 }, { "epoch": 2.2627723632751877, "grad_norm": 0.490088938991014, "learning_rate": 1.729519174746624e-05, "loss": 2.7086, "step": 36451 }, { "epoch": 2.2628344403749456, "grad_norm": 0.46375139899441264, "learning_rate": 1.72924600347201e-05, "loss": 2.6708, "step": 36452 }, { "epoch": 2.2628965174747036, "grad_norm": 0.46259886161867814, "learning_rate": 1.7289728492617118e-05, "loss": 2.6982, "step": 36453 }, { "epoch": 2.2629585945744615, "grad_norm": 0.540965191525156, "learning_rate": 1.7286997121171584e-05, "loss": 2.7507, "step": 36454 }, { "epoch": 2.2630206716742194, "grad_norm": 0.45362382595260364, "learning_rate": 1.7284265920397707e-05, "loss": 2.6133, "step": 36455 }, { "epoch": 2.2630827487739773, "grad_norm": 0.4091628442179145, "learning_rate": 1.7281534890309775e-05, "loss": 2.6288, "step": 36456 }, { "epoch": 2.263144825873735, "grad_norm": 0.4353572963167584, "learning_rate": 1.7278804030922014e-05, "loss": 2.7191, "step": 36457 }, { "epoch": 2.263206902973493, "grad_norm": 0.46946439346397645, "learning_rate": 1.7276073342248682e-05, "loss": 2.6214, "step": 36458 }, { "epoch": 2.263268980073251, "grad_norm": 0.46938524118215047, "learning_rate": 1.7273342824304e-05, "loss": 2.6451, "step": 36459 }, { "epoch": 2.263331057173009, "grad_norm": 0.4846874121709531, "learning_rate": 1.7270612477102248e-05, "loss": 2.7745, "step": 36460 }, { "epoch": 2.263393134272767, "grad_norm": 0.4740629464979087, "learning_rate": 1.726788230065766e-05, "loss": 2.7104, "step": 36461 }, { "epoch": 2.263455211372525, "grad_norm": 0.5146742349199529, "learning_rate": 1.7265152294984473e-05, "loss": 2.7145, "step": 36462 }, { "epoch": 2.2635172884722827, "grad_norm": 0.4542694109735729, "learning_rate": 1.7262422460096916e-05, "loss": 2.7382, "step": 36463 }, { "epoch": 2.2635793655720406, "grad_norm": 0.49232291826143065, "learning_rate": 1.725969279600927e-05, "loss": 2.6902, "step": 36464 }, { "epoch": 2.2636414426717986, "grad_norm": 0.4858176374871704, "learning_rate": 1.725696330273575e-05, "loss": 2.6962, "step": 36465 }, { "epoch": 2.263703519771556, "grad_norm": 0.4430392113407726, "learning_rate": 1.72542339802906e-05, "loss": 2.6008, "step": 36466 }, { "epoch": 2.2637655968713144, "grad_norm": 0.49631684460844233, "learning_rate": 1.7251504828688047e-05, "loss": 2.6627, "step": 36467 }, { "epoch": 2.263827673971072, "grad_norm": 0.5289200007624721, "learning_rate": 1.724877584794236e-05, "loss": 2.7446, "step": 36468 }, { "epoch": 2.26388975107083, "grad_norm": 0.4815253912696617, "learning_rate": 1.7246047038067757e-05, "loss": 2.6777, "step": 36469 }, { "epoch": 2.2639518281705877, "grad_norm": 0.48276272983252694, "learning_rate": 1.7243318399078483e-05, "loss": 2.6893, "step": 36470 }, { "epoch": 2.2640139052703456, "grad_norm": 0.5077560686887932, "learning_rate": 1.7240589930988765e-05, "loss": 2.6838, "step": 36471 }, { "epoch": 2.2640759823701035, "grad_norm": 0.4504929425494366, "learning_rate": 1.723786163381283e-05, "loss": 2.6661, "step": 36472 }, { "epoch": 2.2641380594698615, "grad_norm": 0.492347418826812, "learning_rate": 1.723513350756493e-05, "loss": 2.5977, "step": 36473 }, { "epoch": 2.2642001365696194, "grad_norm": 0.49349031137967986, "learning_rate": 1.7232405552259307e-05, "loss": 2.7083, "step": 36474 }, { "epoch": 2.2642622136693773, "grad_norm": 0.4841392312453779, "learning_rate": 1.722967776791017e-05, "loss": 2.6819, "step": 36475 }, { "epoch": 2.264324290769135, "grad_norm": 0.4975514359937341, "learning_rate": 1.722695015453174e-05, "loss": 2.6526, "step": 36476 }, { "epoch": 2.264386367868893, "grad_norm": 0.49875131790923954, "learning_rate": 1.722422271213829e-05, "loss": 2.736, "step": 36477 }, { "epoch": 2.264448444968651, "grad_norm": 0.43271696089749856, "learning_rate": 1.7221495440744023e-05, "loss": 2.6462, "step": 36478 }, { "epoch": 2.264510522068409, "grad_norm": 0.5248999591152228, "learning_rate": 1.7218768340363172e-05, "loss": 2.7357, "step": 36479 }, { "epoch": 2.264572599168167, "grad_norm": 0.500047741041258, "learning_rate": 1.721604141100997e-05, "loss": 2.6696, "step": 36480 }, { "epoch": 2.264634676267925, "grad_norm": 0.48425474519920575, "learning_rate": 1.7213314652698614e-05, "loss": 2.7225, "step": 36481 }, { "epoch": 2.2646967533676827, "grad_norm": 0.5265277058682865, "learning_rate": 1.721058806544338e-05, "loss": 2.6521, "step": 36482 }, { "epoch": 2.2647588304674406, "grad_norm": 0.4897767929237491, "learning_rate": 1.720786164925846e-05, "loss": 2.6776, "step": 36483 }, { "epoch": 2.2648209075671986, "grad_norm": 0.5278418555037976, "learning_rate": 1.720513540415809e-05, "loss": 2.7126, "step": 36484 }, { "epoch": 2.2648829846669565, "grad_norm": 0.48951452529249406, "learning_rate": 1.7202409330156473e-05, "loss": 2.6577, "step": 36485 }, { "epoch": 2.2649450617667144, "grad_norm": 0.5048422880160416, "learning_rate": 1.719968342726786e-05, "loss": 2.6562, "step": 36486 }, { "epoch": 2.2650071388664723, "grad_norm": 0.48045110282433934, "learning_rate": 1.7196957695506465e-05, "loss": 2.607, "step": 36487 }, { "epoch": 2.2650692159662302, "grad_norm": 0.48980432622470155, "learning_rate": 1.719423213488649e-05, "loss": 2.7202, "step": 36488 }, { "epoch": 2.265131293065988, "grad_norm": 0.5204684759738811, "learning_rate": 1.7191506745422186e-05, "loss": 2.6363, "step": 36489 }, { "epoch": 2.265193370165746, "grad_norm": 0.4542602917938713, "learning_rate": 1.718878152712774e-05, "loss": 2.6504, "step": 36490 }, { "epoch": 2.2652554472655035, "grad_norm": 0.48308167531041146, "learning_rate": 1.7186056480017397e-05, "loss": 2.6861, "step": 36491 }, { "epoch": 2.265317524365262, "grad_norm": 0.4951400751545829, "learning_rate": 1.718333160410537e-05, "loss": 2.6235, "step": 36492 }, { "epoch": 2.2653796014650194, "grad_norm": 0.4316184439505065, "learning_rate": 1.7180606899405864e-05, "loss": 2.5851, "step": 36493 }, { "epoch": 2.2654416785647773, "grad_norm": 0.5174902787424231, "learning_rate": 1.71778823659331e-05, "loss": 2.6563, "step": 36494 }, { "epoch": 2.265503755664535, "grad_norm": 0.4364223832625325, "learning_rate": 1.717515800370128e-05, "loss": 2.7079, "step": 36495 }, { "epoch": 2.265565832764293, "grad_norm": 0.4504648423728334, "learning_rate": 1.7172433812724642e-05, "loss": 2.7258, "step": 36496 }, { "epoch": 2.265627909864051, "grad_norm": 0.5627969381922852, "learning_rate": 1.7169709793017386e-05, "loss": 2.8019, "step": 36497 }, { "epoch": 2.265689986963809, "grad_norm": 0.4887975480836816, "learning_rate": 1.7166985944593723e-05, "loss": 2.6577, "step": 36498 }, { "epoch": 2.265752064063567, "grad_norm": 0.5741055694936513, "learning_rate": 1.7164262267467845e-05, "loss": 2.7845, "step": 36499 }, { "epoch": 2.265814141163325, "grad_norm": 0.5140659285781757, "learning_rate": 1.7161538761654e-05, "loss": 2.6987, "step": 36500 }, { "epoch": 2.2658762182630827, "grad_norm": 0.5618485280930753, "learning_rate": 1.7158815427166386e-05, "loss": 2.6784, "step": 36501 }, { "epoch": 2.2659382953628406, "grad_norm": 0.5351532706296085, "learning_rate": 1.7156092264019195e-05, "loss": 2.6578, "step": 36502 }, { "epoch": 2.2660003724625986, "grad_norm": 0.5753592757770459, "learning_rate": 1.7153369272226634e-05, "loss": 2.7224, "step": 36503 }, { "epoch": 2.2660624495623565, "grad_norm": 0.5407765545402835, "learning_rate": 1.7150646451802933e-05, "loss": 2.5522, "step": 36504 }, { "epoch": 2.2661245266621144, "grad_norm": 0.5282446054308639, "learning_rate": 1.7147923802762284e-05, "loss": 2.6764, "step": 36505 }, { "epoch": 2.2661866037618723, "grad_norm": 0.5626053711425605, "learning_rate": 1.714520132511889e-05, "loss": 2.6725, "step": 36506 }, { "epoch": 2.2662486808616302, "grad_norm": 0.5157192317514505, "learning_rate": 1.7142479018886964e-05, "loss": 2.749, "step": 36507 }, { "epoch": 2.266310757961388, "grad_norm": 0.4861644178257193, "learning_rate": 1.7139756884080675e-05, "loss": 2.6891, "step": 36508 }, { "epoch": 2.266372835061146, "grad_norm": 0.6352750641263023, "learning_rate": 1.7137034920714275e-05, "loss": 2.7045, "step": 36509 }, { "epoch": 2.266434912160904, "grad_norm": 0.45916354963533496, "learning_rate": 1.713431312880194e-05, "loss": 2.6705, "step": 36510 }, { "epoch": 2.266496989260662, "grad_norm": 0.4953028013080066, "learning_rate": 1.7131591508357868e-05, "loss": 2.7373, "step": 36511 }, { "epoch": 2.26655906636042, "grad_norm": 0.5456026261888712, "learning_rate": 1.712887005939625e-05, "loss": 2.7405, "step": 36512 }, { "epoch": 2.2666211434601777, "grad_norm": 0.46105776042345653, "learning_rate": 1.7126148781931313e-05, "loss": 2.7155, "step": 36513 }, { "epoch": 2.266683220559935, "grad_norm": 0.4931328640880048, "learning_rate": 1.712342767597724e-05, "loss": 2.5925, "step": 36514 }, { "epoch": 2.2667452976596936, "grad_norm": 0.5188369126646247, "learning_rate": 1.712070674154822e-05, "loss": 2.7572, "step": 36515 }, { "epoch": 2.266807374759451, "grad_norm": 0.48299300020291225, "learning_rate": 1.711798597865846e-05, "loss": 2.6887, "step": 36516 }, { "epoch": 2.266869451859209, "grad_norm": 0.512471787908987, "learning_rate": 1.711526538732213e-05, "loss": 2.7646, "step": 36517 }, { "epoch": 2.266931528958967, "grad_norm": 0.4955030062889338, "learning_rate": 1.7112544967553457e-05, "loss": 2.6593, "step": 36518 }, { "epoch": 2.266993606058725, "grad_norm": 0.5293061090205825, "learning_rate": 1.7109824719366623e-05, "loss": 2.7146, "step": 36519 }, { "epoch": 2.2670556831584827, "grad_norm": 0.534475910407569, "learning_rate": 1.7107104642775823e-05, "loss": 2.6782, "step": 36520 }, { "epoch": 2.2671177602582406, "grad_norm": 0.48885507054234856, "learning_rate": 1.710438473779522e-05, "loss": 2.6903, "step": 36521 }, { "epoch": 2.2671798373579986, "grad_norm": 0.5084582697700466, "learning_rate": 1.7101665004439025e-05, "loss": 2.6191, "step": 36522 }, { "epoch": 2.2672419144577565, "grad_norm": 0.5952427320072684, "learning_rate": 1.7098945442721458e-05, "loss": 2.7087, "step": 36523 }, { "epoch": 2.2673039915575144, "grad_norm": 0.4980447357730176, "learning_rate": 1.7096226052656672e-05, "loss": 2.662, "step": 36524 }, { "epoch": 2.2673660686572723, "grad_norm": 0.5389528854157688, "learning_rate": 1.7093506834258865e-05, "loss": 2.7283, "step": 36525 }, { "epoch": 2.2674281457570302, "grad_norm": 0.5400689170269868, "learning_rate": 1.70907877875422e-05, "loss": 2.7277, "step": 36526 }, { "epoch": 2.267490222856788, "grad_norm": 0.5248038833890262, "learning_rate": 1.7088068912520906e-05, "loss": 2.7389, "step": 36527 }, { "epoch": 2.267552299956546, "grad_norm": 0.5632427927293469, "learning_rate": 1.7085350209209146e-05, "loss": 2.6727, "step": 36528 }, { "epoch": 2.267614377056304, "grad_norm": 0.5152723088735413, "learning_rate": 1.7082631677621096e-05, "loss": 2.621, "step": 36529 }, { "epoch": 2.267676454156062, "grad_norm": 0.5059288006718519, "learning_rate": 1.7079913317770957e-05, "loss": 2.5244, "step": 36530 }, { "epoch": 2.26773853125582, "grad_norm": 0.5370429812923886, "learning_rate": 1.707719512967288e-05, "loss": 2.7776, "step": 36531 }, { "epoch": 2.2678006083555777, "grad_norm": 0.46979715685233797, "learning_rate": 1.7074477113341087e-05, "loss": 2.6415, "step": 36532 }, { "epoch": 2.2678626854553356, "grad_norm": 0.5480586185950836, "learning_rate": 1.7071759268789744e-05, "loss": 2.6648, "step": 36533 }, { "epoch": 2.2679247625550936, "grad_norm": 0.5140967406954066, "learning_rate": 1.706904159603302e-05, "loss": 2.605, "step": 36534 }, { "epoch": 2.2679868396548515, "grad_norm": 0.47887816416791124, "learning_rate": 1.7066324095085085e-05, "loss": 2.6947, "step": 36535 }, { "epoch": 2.2680489167546094, "grad_norm": 0.51571076575717, "learning_rate": 1.706360676596015e-05, "loss": 2.7632, "step": 36536 }, { "epoch": 2.268110993854367, "grad_norm": 0.48933222770122087, "learning_rate": 1.7060889608672375e-05, "loss": 2.7707, "step": 36537 }, { "epoch": 2.2681730709541252, "grad_norm": 0.46732609967081096, "learning_rate": 1.7058172623235935e-05, "loss": 2.6163, "step": 36538 }, { "epoch": 2.2682351480538827, "grad_norm": 0.5550803552914295, "learning_rate": 1.7055455809664984e-05, "loss": 2.6355, "step": 36539 }, { "epoch": 2.2682972251536406, "grad_norm": 0.4292959510022703, "learning_rate": 1.7052739167973742e-05, "loss": 2.7003, "step": 36540 }, { "epoch": 2.2683593022533985, "grad_norm": 0.48959112330252114, "learning_rate": 1.7050022698176347e-05, "loss": 2.7678, "step": 36541 }, { "epoch": 2.2684213793531565, "grad_norm": 0.4876239625117401, "learning_rate": 1.7047306400286993e-05, "loss": 2.6757, "step": 36542 }, { "epoch": 2.2684834564529144, "grad_norm": 0.43088299296606686, "learning_rate": 1.7044590274319838e-05, "loss": 2.7421, "step": 36543 }, { "epoch": 2.2685455335526723, "grad_norm": 0.4735268071107865, "learning_rate": 1.7041874320289035e-05, "loss": 2.6885, "step": 36544 }, { "epoch": 2.26860761065243, "grad_norm": 0.5136589367763699, "learning_rate": 1.70391585382088e-05, "loss": 2.7052, "step": 36545 }, { "epoch": 2.268669687752188, "grad_norm": 0.45778683209640436, "learning_rate": 1.7036442928093272e-05, "loss": 2.7269, "step": 36546 }, { "epoch": 2.268731764851946, "grad_norm": 0.4459875578342239, "learning_rate": 1.703372748995662e-05, "loss": 2.6332, "step": 36547 }, { "epoch": 2.268793841951704, "grad_norm": 0.4800218659377651, "learning_rate": 1.7031012223813e-05, "loss": 2.6725, "step": 36548 }, { "epoch": 2.268855919051462, "grad_norm": 0.47395145669776256, "learning_rate": 1.7028297129676613e-05, "loss": 2.8107, "step": 36549 }, { "epoch": 2.26891799615122, "grad_norm": 0.49731060283146084, "learning_rate": 1.7025582207561598e-05, "loss": 2.6747, "step": 36550 }, { "epoch": 2.2689800732509777, "grad_norm": 0.4832417397885927, "learning_rate": 1.7022867457482128e-05, "loss": 2.6994, "step": 36551 }, { "epoch": 2.2690421503507356, "grad_norm": 0.47182876095105347, "learning_rate": 1.7020152879452346e-05, "loss": 2.6137, "step": 36552 }, { "epoch": 2.2691042274504936, "grad_norm": 0.5366297296830552, "learning_rate": 1.7017438473486435e-05, "loss": 2.7412, "step": 36553 }, { "epoch": 2.2691663045502515, "grad_norm": 0.504314611996171, "learning_rate": 1.7014724239598574e-05, "loss": 2.6805, "step": 36554 }, { "epoch": 2.2692283816500094, "grad_norm": 0.46881142849552443, "learning_rate": 1.7012010177802895e-05, "loss": 2.768, "step": 36555 }, { "epoch": 2.2692904587497673, "grad_norm": 0.5350886114214942, "learning_rate": 1.7009296288113574e-05, "loss": 2.7261, "step": 36556 }, { "epoch": 2.2693525358495252, "grad_norm": 0.47240263069362715, "learning_rate": 1.700658257054476e-05, "loss": 2.6689, "step": 36557 }, { "epoch": 2.269414612949283, "grad_norm": 0.47569519605213695, "learning_rate": 1.7003869025110597e-05, "loss": 2.6474, "step": 36558 }, { "epoch": 2.269476690049041, "grad_norm": 0.520454704624241, "learning_rate": 1.700115565182528e-05, "loss": 2.7614, "step": 36559 }, { "epoch": 2.269538767148799, "grad_norm": 0.4975122761529659, "learning_rate": 1.6998442450702946e-05, "loss": 2.6696, "step": 36560 }, { "epoch": 2.269600844248557, "grad_norm": 0.5473784173238968, "learning_rate": 1.6995729421757745e-05, "loss": 2.7027, "step": 36561 }, { "epoch": 2.2696629213483144, "grad_norm": 0.5231271934235414, "learning_rate": 1.6993016565003815e-05, "loss": 2.7389, "step": 36562 }, { "epoch": 2.2697249984480727, "grad_norm": 0.5495896943885646, "learning_rate": 1.6990303880455353e-05, "loss": 2.7053, "step": 36563 }, { "epoch": 2.26978707554783, "grad_norm": 0.48764822552699477, "learning_rate": 1.6987591368126487e-05, "loss": 2.6413, "step": 36564 }, { "epoch": 2.269849152647588, "grad_norm": 0.522364270545763, "learning_rate": 1.698487902803137e-05, "loss": 2.7366, "step": 36565 }, { "epoch": 2.269911229747346, "grad_norm": 0.5545256203854463, "learning_rate": 1.6982166860184155e-05, "loss": 2.6497, "step": 36566 }, { "epoch": 2.269973306847104, "grad_norm": 0.5195930558921164, "learning_rate": 1.6979454864598977e-05, "loss": 2.7442, "step": 36567 }, { "epoch": 2.270035383946862, "grad_norm": 0.47538369204278025, "learning_rate": 1.697674304129001e-05, "loss": 2.6883, "step": 36568 }, { "epoch": 2.27009746104662, "grad_norm": 0.5038403312682107, "learning_rate": 1.69740313902714e-05, "loss": 2.6877, "step": 36569 }, { "epoch": 2.2701595381463777, "grad_norm": 0.5230063185519849, "learning_rate": 1.6971319911557278e-05, "loss": 2.6345, "step": 36570 }, { "epoch": 2.2702216152461356, "grad_norm": 0.45917416159137425, "learning_rate": 1.6968608605161785e-05, "loss": 2.6458, "step": 36571 }, { "epoch": 2.2702836923458936, "grad_norm": 0.4861490463789301, "learning_rate": 1.696589747109909e-05, "loss": 2.6749, "step": 36572 }, { "epoch": 2.2703457694456515, "grad_norm": 0.4847920529695006, "learning_rate": 1.6963186509383333e-05, "loss": 2.6633, "step": 36573 }, { "epoch": 2.2704078465454094, "grad_norm": 0.49814102180589354, "learning_rate": 1.696047572002865e-05, "loss": 2.6515, "step": 36574 }, { "epoch": 2.2704699236451673, "grad_norm": 0.49513307805491125, "learning_rate": 1.695776510304917e-05, "loss": 2.721, "step": 36575 }, { "epoch": 2.2705320007449252, "grad_norm": 0.4753612951228559, "learning_rate": 1.695505465845907e-05, "loss": 2.6026, "step": 36576 }, { "epoch": 2.270594077844683, "grad_norm": 0.5031595834254761, "learning_rate": 1.6952344386272466e-05, "loss": 2.7142, "step": 36577 }, { "epoch": 2.270656154944441, "grad_norm": 0.5181963988508435, "learning_rate": 1.6949634286503508e-05, "loss": 2.7348, "step": 36578 }, { "epoch": 2.270718232044199, "grad_norm": 0.5135626150917759, "learning_rate": 1.694692435916633e-05, "loss": 2.7247, "step": 36579 }, { "epoch": 2.270780309143957, "grad_norm": 0.58078104195169, "learning_rate": 1.6944214604275054e-05, "loss": 2.6813, "step": 36580 }, { "epoch": 2.270842386243715, "grad_norm": 0.4877701640406843, "learning_rate": 1.6941505021843856e-05, "loss": 2.6409, "step": 36581 }, { "epoch": 2.2709044633434727, "grad_norm": 0.49499514729115196, "learning_rate": 1.6938795611886848e-05, "loss": 2.6596, "step": 36582 }, { "epoch": 2.2709665404432307, "grad_norm": 0.46990644535852905, "learning_rate": 1.693608637441817e-05, "loss": 2.7111, "step": 36583 }, { "epoch": 2.2710286175429886, "grad_norm": 0.5322727568291525, "learning_rate": 1.6933377309451937e-05, "loss": 2.6585, "step": 36584 }, { "epoch": 2.271090694642746, "grad_norm": 0.48006987102915166, "learning_rate": 1.6930668417002322e-05, "loss": 2.7304, "step": 36585 }, { "epoch": 2.2711527717425044, "grad_norm": 0.5096024292211258, "learning_rate": 1.692795969708342e-05, "loss": 2.7026, "step": 36586 }, { "epoch": 2.271214848842262, "grad_norm": 0.4945483730539703, "learning_rate": 1.6925251149709392e-05, "loss": 2.654, "step": 36587 }, { "epoch": 2.27127692594202, "grad_norm": 0.4834170097097122, "learning_rate": 1.692254277489436e-05, "loss": 2.7044, "step": 36588 }, { "epoch": 2.2713390030417777, "grad_norm": 0.49357325130644225, "learning_rate": 1.6919834572652443e-05, "loss": 2.7009, "step": 36589 }, { "epoch": 2.2714010801415356, "grad_norm": 0.4357426977218725, "learning_rate": 1.691712654299779e-05, "loss": 2.6661, "step": 36590 }, { "epoch": 2.2714631572412936, "grad_norm": 0.517886807503458, "learning_rate": 1.6914418685944517e-05, "loss": 2.7511, "step": 36591 }, { "epoch": 2.2715252343410515, "grad_norm": 0.48068069807434916, "learning_rate": 1.691171100150676e-05, "loss": 2.6612, "step": 36592 }, { "epoch": 2.2715873114408094, "grad_norm": 0.44560173747055254, "learning_rate": 1.6909003489698632e-05, "loss": 2.6267, "step": 36593 }, { "epoch": 2.2716493885405673, "grad_norm": 0.5181149847353412, "learning_rate": 1.6906296150534255e-05, "loss": 2.5611, "step": 36594 }, { "epoch": 2.2717114656403252, "grad_norm": 0.48588025102361404, "learning_rate": 1.690358898402778e-05, "loss": 2.6672, "step": 36595 }, { "epoch": 2.271773542740083, "grad_norm": 0.4484285951853389, "learning_rate": 1.6900881990193314e-05, "loss": 2.6204, "step": 36596 }, { "epoch": 2.271835619839841, "grad_norm": 0.49483501512137384, "learning_rate": 1.6898175169044983e-05, "loss": 2.7187, "step": 36597 }, { "epoch": 2.271897696939599, "grad_norm": 0.4820432992229876, "learning_rate": 1.6895468520596892e-05, "loss": 2.705, "step": 36598 }, { "epoch": 2.271959774039357, "grad_norm": 0.48452273546184405, "learning_rate": 1.6892762044863193e-05, "loss": 2.6566, "step": 36599 }, { "epoch": 2.272021851139115, "grad_norm": 0.5049057739467115, "learning_rate": 1.689005574185799e-05, "loss": 2.7523, "step": 36600 }, { "epoch": 2.2720839282388727, "grad_norm": 0.46405929602855944, "learning_rate": 1.688734961159541e-05, "loss": 2.6267, "step": 36601 }, { "epoch": 2.2721460053386306, "grad_norm": 0.4492249861574614, "learning_rate": 1.688464365408956e-05, "loss": 2.6154, "step": 36602 }, { "epoch": 2.2722080824383886, "grad_norm": 0.48063943513838475, "learning_rate": 1.688193786935455e-05, "loss": 2.653, "step": 36603 }, { "epoch": 2.2722701595381465, "grad_norm": 0.48137760266551616, "learning_rate": 1.6879232257404527e-05, "loss": 2.6225, "step": 36604 }, { "epoch": 2.2723322366379044, "grad_norm": 0.46278500861015126, "learning_rate": 1.6876526818253585e-05, "loss": 2.6923, "step": 36605 }, { "epoch": 2.2723943137376623, "grad_norm": 0.5092548134959719, "learning_rate": 1.687382155191585e-05, "loss": 2.7776, "step": 36606 }, { "epoch": 2.2724563908374202, "grad_norm": 0.48067525752181645, "learning_rate": 1.6871116458405407e-05, "loss": 2.6592, "step": 36607 }, { "epoch": 2.272518467937178, "grad_norm": 0.5543064366618815, "learning_rate": 1.6868411537736406e-05, "loss": 2.6923, "step": 36608 }, { "epoch": 2.272580545036936, "grad_norm": 0.5071225688030742, "learning_rate": 1.6865706789922953e-05, "loss": 2.6139, "step": 36609 }, { "epoch": 2.2726426221366935, "grad_norm": 0.46019696701551527, "learning_rate": 1.6863002214979147e-05, "loss": 2.7858, "step": 36610 }, { "epoch": 2.272704699236452, "grad_norm": 0.5107425376059388, "learning_rate": 1.6860297812919085e-05, "loss": 2.6694, "step": 36611 }, { "epoch": 2.2727667763362094, "grad_norm": 0.48932152369451337, "learning_rate": 1.6857593583756913e-05, "loss": 2.6217, "step": 36612 }, { "epoch": 2.2728288534359673, "grad_norm": 0.45565044449766007, "learning_rate": 1.6854889527506722e-05, "loss": 2.7241, "step": 36613 }, { "epoch": 2.272890930535725, "grad_norm": 0.5397356778073813, "learning_rate": 1.6852185644182617e-05, "loss": 2.6572, "step": 36614 }, { "epoch": 2.272953007635483, "grad_norm": 0.5310327983468389, "learning_rate": 1.6849481933798706e-05, "loss": 2.6723, "step": 36615 }, { "epoch": 2.273015084735241, "grad_norm": 0.4525298452282414, "learning_rate": 1.6846778396369077e-05, "loss": 2.7472, "step": 36616 }, { "epoch": 2.273077161834999, "grad_norm": 0.5386151655303688, "learning_rate": 1.6844075031907875e-05, "loss": 2.6482, "step": 36617 }, { "epoch": 2.273139238934757, "grad_norm": 0.54745733046934, "learning_rate": 1.6841371840429175e-05, "loss": 2.7215, "step": 36618 }, { "epoch": 2.273201316034515, "grad_norm": 0.5446412652637883, "learning_rate": 1.6838668821947078e-05, "loss": 2.6674, "step": 36619 }, { "epoch": 2.2732633931342727, "grad_norm": 0.5360427530660442, "learning_rate": 1.683596597647571e-05, "loss": 2.6432, "step": 36620 }, { "epoch": 2.2733254702340306, "grad_norm": 0.521670333338558, "learning_rate": 1.683326330402914e-05, "loss": 2.7842, "step": 36621 }, { "epoch": 2.2733875473337886, "grad_norm": 0.5164853345648271, "learning_rate": 1.6830560804621508e-05, "loss": 2.7314, "step": 36622 }, { "epoch": 2.2734496244335465, "grad_norm": 0.5412415401158795, "learning_rate": 1.6827858478266888e-05, "loss": 2.6305, "step": 36623 }, { "epoch": 2.2735117015333044, "grad_norm": 0.4814600603734856, "learning_rate": 1.6825156324979385e-05, "loss": 2.665, "step": 36624 }, { "epoch": 2.2735737786330623, "grad_norm": 0.505370386095519, "learning_rate": 1.682245434477308e-05, "loss": 2.674, "step": 36625 }, { "epoch": 2.2736358557328202, "grad_norm": 0.4817158865309944, "learning_rate": 1.6819752537662098e-05, "loss": 2.7114, "step": 36626 }, { "epoch": 2.273697932832578, "grad_norm": 0.4878692872731631, "learning_rate": 1.6817050903660524e-05, "loss": 2.6703, "step": 36627 }, { "epoch": 2.273760009932336, "grad_norm": 0.46844642663737535, "learning_rate": 1.681434944278245e-05, "loss": 2.6621, "step": 36628 }, { "epoch": 2.273822087032094, "grad_norm": 0.4707083167585179, "learning_rate": 1.681164815504197e-05, "loss": 2.6386, "step": 36629 }, { "epoch": 2.273884164131852, "grad_norm": 0.5205864193876092, "learning_rate": 1.6808947040453165e-05, "loss": 2.722, "step": 36630 }, { "epoch": 2.27394624123161, "grad_norm": 0.47687042169135546, "learning_rate": 1.6806246099030153e-05, "loss": 2.6834, "step": 36631 }, { "epoch": 2.2740083183313677, "grad_norm": 0.5091421472709017, "learning_rate": 1.680354533078702e-05, "loss": 2.7312, "step": 36632 }, { "epoch": 2.274070395431125, "grad_norm": 0.4901878517234248, "learning_rate": 1.6800844735737843e-05, "loss": 2.7102, "step": 36633 }, { "epoch": 2.2741324725308836, "grad_norm": 0.4489873041097567, "learning_rate": 1.6798144313896707e-05, "loss": 2.7154, "step": 36634 }, { "epoch": 2.274194549630641, "grad_norm": 0.48211046739552466, "learning_rate": 1.679544406527772e-05, "loss": 2.6248, "step": 36635 }, { "epoch": 2.274256626730399, "grad_norm": 0.5035694978361983, "learning_rate": 1.679274398989497e-05, "loss": 2.6574, "step": 36636 }, { "epoch": 2.274318703830157, "grad_norm": 0.525598591198913, "learning_rate": 1.6790044087762534e-05, "loss": 2.7, "step": 36637 }, { "epoch": 2.274380780929915, "grad_norm": 0.46424139087667005, "learning_rate": 1.6787344358894485e-05, "loss": 2.7547, "step": 36638 }, { "epoch": 2.2744428580296727, "grad_norm": 0.4732190399239569, "learning_rate": 1.6784644803304938e-05, "loss": 2.6967, "step": 36639 }, { "epoch": 2.2745049351294306, "grad_norm": 0.49475855130925017, "learning_rate": 1.678194542100796e-05, "loss": 2.7293, "step": 36640 }, { "epoch": 2.2745670122291886, "grad_norm": 0.4639371959660102, "learning_rate": 1.6779246212017642e-05, "loss": 2.6532, "step": 36641 }, { "epoch": 2.2746290893289465, "grad_norm": 0.4683943670975756, "learning_rate": 1.6776547176348058e-05, "loss": 2.6759, "step": 36642 }, { "epoch": 2.2746911664287044, "grad_norm": 0.4838882122921341, "learning_rate": 1.6773848314013274e-05, "loss": 2.715, "step": 36643 }, { "epoch": 2.2747532435284623, "grad_norm": 0.4746190607785861, "learning_rate": 1.677114962502741e-05, "loss": 2.7451, "step": 36644 }, { "epoch": 2.2748153206282202, "grad_norm": 0.45351079433868297, "learning_rate": 1.676845110940452e-05, "loss": 2.5773, "step": 36645 }, { "epoch": 2.274877397727978, "grad_norm": 0.4592657647041002, "learning_rate": 1.6765752767158692e-05, "loss": 2.7221, "step": 36646 }, { "epoch": 2.274939474827736, "grad_norm": 0.504523291313885, "learning_rate": 1.6763054598303985e-05, "loss": 2.765, "step": 36647 }, { "epoch": 2.275001551927494, "grad_norm": 0.44909649719582967, "learning_rate": 1.6760356602854503e-05, "loss": 2.661, "step": 36648 }, { "epoch": 2.275063629027252, "grad_norm": 0.465855387200895, "learning_rate": 1.6757658780824308e-05, "loss": 2.6351, "step": 36649 }, { "epoch": 2.27512570612701, "grad_norm": 0.5285671313352327, "learning_rate": 1.6754961132227482e-05, "loss": 2.6766, "step": 36650 }, { "epoch": 2.2751877832267677, "grad_norm": 0.4548933284292517, "learning_rate": 1.675226365707809e-05, "loss": 2.6725, "step": 36651 }, { "epoch": 2.2752498603265257, "grad_norm": 0.4953970936159021, "learning_rate": 1.67495663553902e-05, "loss": 2.7155, "step": 36652 }, { "epoch": 2.2753119374262836, "grad_norm": 0.4830964503421004, "learning_rate": 1.6746869227177885e-05, "loss": 2.7401, "step": 36653 }, { "epoch": 2.2753740145260415, "grad_norm": 0.4847116530971564, "learning_rate": 1.674417227245525e-05, "loss": 2.6553, "step": 36654 }, { "epoch": 2.2754360916257994, "grad_norm": 0.4876557272603009, "learning_rate": 1.6741475491236335e-05, "loss": 2.7508, "step": 36655 }, { "epoch": 2.2754981687255573, "grad_norm": 0.4466374579559722, "learning_rate": 1.6738778883535222e-05, "loss": 2.7433, "step": 36656 }, { "epoch": 2.2755602458253152, "grad_norm": 0.46843839460800285, "learning_rate": 1.6736082449365952e-05, "loss": 2.6441, "step": 36657 }, { "epoch": 2.2756223229250727, "grad_norm": 0.473627511220936, "learning_rate": 1.673338618874264e-05, "loss": 2.6849, "step": 36658 }, { "epoch": 2.275684400024831, "grad_norm": 0.46079313312677406, "learning_rate": 1.673069010167932e-05, "loss": 2.6941, "step": 36659 }, { "epoch": 2.2757464771245886, "grad_norm": 0.4802615021118082, "learning_rate": 1.6727994188190067e-05, "loss": 2.6542, "step": 36660 }, { "epoch": 2.2758085542243465, "grad_norm": 0.4890367690072394, "learning_rate": 1.6725298448288933e-05, "loss": 2.6757, "step": 36661 }, { "epoch": 2.2758706313241044, "grad_norm": 0.514344996368717, "learning_rate": 1.6722602881990007e-05, "loss": 2.7038, "step": 36662 }, { "epoch": 2.2759327084238623, "grad_norm": 0.5019794561107292, "learning_rate": 1.671990748930734e-05, "loss": 2.625, "step": 36663 }, { "epoch": 2.2759947855236202, "grad_norm": 0.47426855149429903, "learning_rate": 1.6717212270255e-05, "loss": 2.7432, "step": 36664 }, { "epoch": 2.276056862623378, "grad_norm": 0.5320409608646507, "learning_rate": 1.6714517224847038e-05, "loss": 2.6893, "step": 36665 }, { "epoch": 2.276118939723136, "grad_norm": 0.4966939164320011, "learning_rate": 1.6711822353097507e-05, "loss": 2.6978, "step": 36666 }, { "epoch": 2.276181016822894, "grad_norm": 0.49343520160934595, "learning_rate": 1.6709127655020495e-05, "loss": 2.7602, "step": 36667 }, { "epoch": 2.276243093922652, "grad_norm": 0.5115712950736325, "learning_rate": 1.6706433130630045e-05, "loss": 2.74, "step": 36668 }, { "epoch": 2.27630517102241, "grad_norm": 0.4615272854367319, "learning_rate": 1.6703738779940214e-05, "loss": 2.7483, "step": 36669 }, { "epoch": 2.2763672481221677, "grad_norm": 0.4865338759678843, "learning_rate": 1.6701044602965044e-05, "loss": 2.6776, "step": 36670 }, { "epoch": 2.2764293252219256, "grad_norm": 0.45198959295774177, "learning_rate": 1.6698350599718625e-05, "loss": 2.6437, "step": 36671 }, { "epoch": 2.2764914023216836, "grad_norm": 0.4573668201650249, "learning_rate": 1.6695656770215e-05, "loss": 2.7136, "step": 36672 }, { "epoch": 2.2765534794214415, "grad_norm": 0.4998232707973523, "learning_rate": 1.6692963114468212e-05, "loss": 2.6345, "step": 36673 }, { "epoch": 2.2766155565211994, "grad_norm": 0.457305483583973, "learning_rate": 1.6690269632492313e-05, "loss": 2.7668, "step": 36674 }, { "epoch": 2.2766776336209573, "grad_norm": 0.4815665164064608, "learning_rate": 1.6687576324301374e-05, "loss": 2.6333, "step": 36675 }, { "epoch": 2.2767397107207152, "grad_norm": 0.5015102169083876, "learning_rate": 1.6684883189909432e-05, "loss": 2.7642, "step": 36676 }, { "epoch": 2.276801787820473, "grad_norm": 0.4787094657809958, "learning_rate": 1.6682190229330552e-05, "loss": 2.7619, "step": 36677 }, { "epoch": 2.276863864920231, "grad_norm": 0.4510883732928832, "learning_rate": 1.6679497442578767e-05, "loss": 2.6753, "step": 36678 }, { "epoch": 2.276925942019989, "grad_norm": 0.4894384785353634, "learning_rate": 1.6676804829668118e-05, "loss": 2.6776, "step": 36679 }, { "epoch": 2.276988019119747, "grad_norm": 0.4956168187537713, "learning_rate": 1.6674112390612686e-05, "loss": 2.7751, "step": 36680 }, { "epoch": 2.2770500962195044, "grad_norm": 0.5094351153794686, "learning_rate": 1.6671420125426497e-05, "loss": 2.7101, "step": 36681 }, { "epoch": 2.2771121733192627, "grad_norm": 0.48681055981828186, "learning_rate": 1.6668728034123598e-05, "loss": 2.6879, "step": 36682 }, { "epoch": 2.27717425041902, "grad_norm": 0.5072582447966865, "learning_rate": 1.666603611671802e-05, "loss": 2.6708, "step": 36683 }, { "epoch": 2.277236327518778, "grad_norm": 0.5136219769163979, "learning_rate": 1.666334437322383e-05, "loss": 2.7372, "step": 36684 }, { "epoch": 2.277298404618536, "grad_norm": 0.5467542515010744, "learning_rate": 1.6660652803655075e-05, "loss": 2.7095, "step": 36685 }, { "epoch": 2.277360481718294, "grad_norm": 0.5214017023854766, "learning_rate": 1.665796140802579e-05, "loss": 2.6104, "step": 36686 }, { "epoch": 2.277422558818052, "grad_norm": 0.5205170593012935, "learning_rate": 1.665527018635002e-05, "loss": 2.7591, "step": 36687 }, { "epoch": 2.27748463591781, "grad_norm": 0.5367412623496914, "learning_rate": 1.6652579138641776e-05, "loss": 2.7768, "step": 36688 }, { "epoch": 2.2775467130175677, "grad_norm": 0.51912736610261, "learning_rate": 1.6649888264915143e-05, "loss": 2.6163, "step": 36689 }, { "epoch": 2.2776087901173256, "grad_norm": 0.5204413623870352, "learning_rate": 1.6647197565184137e-05, "loss": 2.7659, "step": 36690 }, { "epoch": 2.2776708672170836, "grad_norm": 0.5476361563529436, "learning_rate": 1.6644507039462796e-05, "loss": 2.7411, "step": 36691 }, { "epoch": 2.2777329443168415, "grad_norm": 0.5592027292357574, "learning_rate": 1.664181668776516e-05, "loss": 2.6494, "step": 36692 }, { "epoch": 2.2777950214165994, "grad_norm": 0.5021576223642501, "learning_rate": 1.6639126510105253e-05, "loss": 2.6778, "step": 36693 }, { "epoch": 2.2778570985163573, "grad_norm": 0.5172138299607172, "learning_rate": 1.6636436506497134e-05, "loss": 2.6685, "step": 36694 }, { "epoch": 2.2779191756161152, "grad_norm": 0.543659456979803, "learning_rate": 1.663374667695482e-05, "loss": 2.7909, "step": 36695 }, { "epoch": 2.277981252715873, "grad_norm": 0.4712726107973432, "learning_rate": 1.6631057021492356e-05, "loss": 2.6827, "step": 36696 }, { "epoch": 2.278043329815631, "grad_norm": 0.4846703336566919, "learning_rate": 1.662836754012375e-05, "loss": 2.7206, "step": 36697 }, { "epoch": 2.278105406915389, "grad_norm": 0.5118054032666989, "learning_rate": 1.6625678232863068e-05, "loss": 2.6145, "step": 36698 }, { "epoch": 2.278167484015147, "grad_norm": 0.47781941116354526, "learning_rate": 1.6622989099724323e-05, "loss": 2.7162, "step": 36699 }, { "epoch": 2.278229561114905, "grad_norm": 0.5457881208675879, "learning_rate": 1.6620300140721552e-05, "loss": 2.6917, "step": 36700 }, { "epoch": 2.2782916382146627, "grad_norm": 0.519833097036287, "learning_rate": 1.6617611355868772e-05, "loss": 2.735, "step": 36701 }, { "epoch": 2.2783537153144207, "grad_norm": 0.5041219560118396, "learning_rate": 1.6614922745180006e-05, "loss": 2.7572, "step": 36702 }, { "epoch": 2.2784157924141786, "grad_norm": 0.5176159251103183, "learning_rate": 1.6612234308669305e-05, "loss": 2.661, "step": 36703 }, { "epoch": 2.2784778695139365, "grad_norm": 0.4590950528075003, "learning_rate": 1.6609546046350686e-05, "loss": 2.6016, "step": 36704 }, { "epoch": 2.2785399466136944, "grad_norm": 0.5135014762652905, "learning_rate": 1.660685795823817e-05, "loss": 2.7112, "step": 36705 }, { "epoch": 2.278602023713452, "grad_norm": 0.48171437078682305, "learning_rate": 1.6604170044345767e-05, "loss": 2.7043, "step": 36706 }, { "epoch": 2.2786641008132102, "grad_norm": 0.4923696181009705, "learning_rate": 1.6601482304687534e-05, "loss": 2.6478, "step": 36707 }, { "epoch": 2.2787261779129677, "grad_norm": 0.46207383277913644, "learning_rate": 1.659879473927747e-05, "loss": 2.829, "step": 36708 }, { "epoch": 2.2787882550127256, "grad_norm": 0.5339224298764598, "learning_rate": 1.6596107348129604e-05, "loss": 2.7278, "step": 36709 }, { "epoch": 2.2788503321124836, "grad_norm": 0.4937164234864962, "learning_rate": 1.659342013125794e-05, "loss": 2.7036, "step": 36710 }, { "epoch": 2.2789124092122415, "grad_norm": 0.4516669855374739, "learning_rate": 1.659073308867653e-05, "loss": 2.7329, "step": 36711 }, { "epoch": 2.2789744863119994, "grad_norm": 0.4947876896614729, "learning_rate": 1.6588046220399372e-05, "loss": 2.6576, "step": 36712 }, { "epoch": 2.2790365634117573, "grad_norm": 0.5340076567033053, "learning_rate": 1.658535952644049e-05, "loss": 2.6418, "step": 36713 }, { "epoch": 2.2790986405115152, "grad_norm": 0.4555860149256246, "learning_rate": 1.6582673006813905e-05, "loss": 2.7025, "step": 36714 }, { "epoch": 2.279160717611273, "grad_norm": 0.5433673356197751, "learning_rate": 1.65799866615336e-05, "loss": 2.7058, "step": 36715 }, { "epoch": 2.279222794711031, "grad_norm": 0.554897671493523, "learning_rate": 1.657730049061364e-05, "loss": 2.7033, "step": 36716 }, { "epoch": 2.279284871810789, "grad_norm": 0.47459691035225116, "learning_rate": 1.6574614494068003e-05, "loss": 2.7003, "step": 36717 }, { "epoch": 2.279346948910547, "grad_norm": 0.52299164529586, "learning_rate": 1.657192867191073e-05, "loss": 2.6185, "step": 36718 }, { "epoch": 2.279409026010305, "grad_norm": 0.5458434321170588, "learning_rate": 1.6569243024155816e-05, "loss": 2.6698, "step": 36719 }, { "epoch": 2.2794711031100627, "grad_norm": 0.4628935222293627, "learning_rate": 1.656655755081727e-05, "loss": 2.746, "step": 36720 }, { "epoch": 2.2795331802098207, "grad_norm": 0.5719838669342436, "learning_rate": 1.656387225190912e-05, "loss": 2.7688, "step": 36721 }, { "epoch": 2.2795952573095786, "grad_norm": 0.5631752729255969, "learning_rate": 1.6561187127445364e-05, "loss": 2.6234, "step": 36722 }, { "epoch": 2.2796573344093365, "grad_norm": 0.4696788803294414, "learning_rate": 1.6558502177440016e-05, "loss": 2.7413, "step": 36723 }, { "epoch": 2.2797194115090944, "grad_norm": 0.5465295790653794, "learning_rate": 1.655581740190706e-05, "loss": 2.7219, "step": 36724 }, { "epoch": 2.2797814886088523, "grad_norm": 0.5327282422248802, "learning_rate": 1.6553132800860544e-05, "loss": 2.6983, "step": 36725 }, { "epoch": 2.2798435657086102, "grad_norm": 0.48506429295257025, "learning_rate": 1.655044837431445e-05, "loss": 2.7365, "step": 36726 }, { "epoch": 2.279905642808368, "grad_norm": 0.4826831804927123, "learning_rate": 1.6547764122282788e-05, "loss": 2.6116, "step": 36727 }, { "epoch": 2.279967719908126, "grad_norm": 0.5525208185338631, "learning_rate": 1.654508004477956e-05, "loss": 2.6838, "step": 36728 }, { "epoch": 2.2800297970078836, "grad_norm": 0.4675733739727852, "learning_rate": 1.6542396141818762e-05, "loss": 2.6485, "step": 36729 }, { "epoch": 2.280091874107642, "grad_norm": 0.496871593751568, "learning_rate": 1.6539712413414416e-05, "loss": 2.7363, "step": 36730 }, { "epoch": 2.2801539512073994, "grad_norm": 0.5382426288186823, "learning_rate": 1.653702885958051e-05, "loss": 2.6151, "step": 36731 }, { "epoch": 2.2802160283071573, "grad_norm": 0.5080966447939774, "learning_rate": 1.6534345480331054e-05, "loss": 2.7404, "step": 36732 }, { "epoch": 2.2802781054069152, "grad_norm": 0.5411353133803388, "learning_rate": 1.653166227568002e-05, "loss": 2.7889, "step": 36733 }, { "epoch": 2.280340182506673, "grad_norm": 0.49653968893079126, "learning_rate": 1.6528979245641445e-05, "loss": 2.633, "step": 36734 }, { "epoch": 2.280402259606431, "grad_norm": 0.4570065190071488, "learning_rate": 1.652629639022932e-05, "loss": 2.7664, "step": 36735 }, { "epoch": 2.280464336706189, "grad_norm": 0.4793438342033604, "learning_rate": 1.652361370945762e-05, "loss": 2.6053, "step": 36736 }, { "epoch": 2.280526413805947, "grad_norm": 0.5180602796877227, "learning_rate": 1.652093120334036e-05, "loss": 2.7114, "step": 36737 }, { "epoch": 2.280588490905705, "grad_norm": 0.4838250637712162, "learning_rate": 1.6518248871891513e-05, "loss": 2.6937, "step": 36738 }, { "epoch": 2.2806505680054627, "grad_norm": 0.49476760169953066, "learning_rate": 1.6515566715125108e-05, "loss": 2.6703, "step": 36739 }, { "epoch": 2.2807126451052206, "grad_norm": 0.48875726004181635, "learning_rate": 1.651288473305512e-05, "loss": 2.701, "step": 36740 }, { "epoch": 2.2807747222049786, "grad_norm": 0.4307413702192758, "learning_rate": 1.651020292569554e-05, "loss": 2.6444, "step": 36741 }, { "epoch": 2.2808367993047365, "grad_norm": 0.5086460608141306, "learning_rate": 1.650752129306034e-05, "loss": 2.622, "step": 36742 }, { "epoch": 2.2808988764044944, "grad_norm": 0.5001968056069238, "learning_rate": 1.6504839835163554e-05, "loss": 2.6585, "step": 36743 }, { "epoch": 2.2809609535042523, "grad_norm": 0.4540225544769698, "learning_rate": 1.650215855201915e-05, "loss": 2.6083, "step": 36744 }, { "epoch": 2.2810230306040102, "grad_norm": 0.5150369990585592, "learning_rate": 1.6499477443641116e-05, "loss": 2.7071, "step": 36745 }, { "epoch": 2.281085107703768, "grad_norm": 0.4666690571615891, "learning_rate": 1.649679651004342e-05, "loss": 2.7012, "step": 36746 }, { "epoch": 2.281147184803526, "grad_norm": 0.5031408070515696, "learning_rate": 1.649411575124009e-05, "loss": 2.7044, "step": 36747 }, { "epoch": 2.281209261903284, "grad_norm": 0.4921848811730847, "learning_rate": 1.6491435167245095e-05, "loss": 2.6836, "step": 36748 }, { "epoch": 2.281271339003042, "grad_norm": 0.44182780905440205, "learning_rate": 1.6488754758072412e-05, "loss": 2.6518, "step": 36749 }, { "epoch": 2.2813334161028, "grad_norm": 0.48600315435488695, "learning_rate": 1.6486074523736016e-05, "loss": 2.7008, "step": 36750 }, { "epoch": 2.2813954932025577, "grad_norm": 0.4942697650293584, "learning_rate": 1.6483394464249925e-05, "loss": 2.6907, "step": 36751 }, { "epoch": 2.2814575703023157, "grad_norm": 0.4711844646357607, "learning_rate": 1.6480714579628076e-05, "loss": 2.6989, "step": 36752 }, { "epoch": 2.2815196474020736, "grad_norm": 0.4984180339013597, "learning_rate": 1.64780348698845e-05, "loss": 2.6588, "step": 36753 }, { "epoch": 2.281581724501831, "grad_norm": 0.502141894754512, "learning_rate": 1.6475355335033154e-05, "loss": 2.6826, "step": 36754 }, { "epoch": 2.2816438016015894, "grad_norm": 0.4949498843823275, "learning_rate": 1.6472675975088016e-05, "loss": 2.6235, "step": 36755 }, { "epoch": 2.281705878701347, "grad_norm": 0.5433464109079507, "learning_rate": 1.646999679006305e-05, "loss": 2.7952, "step": 36756 }, { "epoch": 2.281767955801105, "grad_norm": 0.4244110037091728, "learning_rate": 1.6467317779972262e-05, "loss": 2.712, "step": 36757 }, { "epoch": 2.2818300329008627, "grad_norm": 0.4551480448304426, "learning_rate": 1.6464638944829625e-05, "loss": 2.6032, "step": 36758 }, { "epoch": 2.2818921100006206, "grad_norm": 0.4809044668838945, "learning_rate": 1.64619602846491e-05, "loss": 2.712, "step": 36759 }, { "epoch": 2.2819541871003786, "grad_norm": 0.4536424838912183, "learning_rate": 1.645928179944466e-05, "loss": 2.621, "step": 36760 }, { "epoch": 2.2820162642001365, "grad_norm": 0.501893095590483, "learning_rate": 1.6456603489230305e-05, "loss": 2.7429, "step": 36761 }, { "epoch": 2.2820783412998944, "grad_norm": 0.4943893922188993, "learning_rate": 1.645392535401999e-05, "loss": 2.7115, "step": 36762 }, { "epoch": 2.2821404183996523, "grad_norm": 0.4737712068013471, "learning_rate": 1.645124739382769e-05, "loss": 2.6623, "step": 36763 }, { "epoch": 2.2822024954994102, "grad_norm": 0.49785610360027127, "learning_rate": 1.6448569608667382e-05, "loss": 2.6469, "step": 36764 }, { "epoch": 2.282264572599168, "grad_norm": 0.5099450449693366, "learning_rate": 1.644589199855301e-05, "loss": 2.6326, "step": 36765 }, { "epoch": 2.282326649698926, "grad_norm": 0.48176867166669984, "learning_rate": 1.6443214563498582e-05, "loss": 2.6526, "step": 36766 }, { "epoch": 2.282388726798684, "grad_norm": 0.49740936766846483, "learning_rate": 1.644053730351805e-05, "loss": 2.6958, "step": 36767 }, { "epoch": 2.282450803898442, "grad_norm": 0.47876950749626096, "learning_rate": 1.6437860218625378e-05, "loss": 2.6662, "step": 36768 }, { "epoch": 2.2825128809982, "grad_norm": 0.476283429466894, "learning_rate": 1.643518330883453e-05, "loss": 2.6938, "step": 36769 }, { "epoch": 2.2825749580979577, "grad_norm": 0.4582269140443833, "learning_rate": 1.6432506574159483e-05, "loss": 2.6175, "step": 36770 }, { "epoch": 2.2826370351977157, "grad_norm": 0.4662269720573064, "learning_rate": 1.6429830014614207e-05, "loss": 2.6612, "step": 36771 }, { "epoch": 2.2826991122974736, "grad_norm": 0.5646474958983956, "learning_rate": 1.6427153630212654e-05, "loss": 2.6911, "step": 36772 }, { "epoch": 2.2827611893972315, "grad_norm": 0.4732433015225374, "learning_rate": 1.6424477420968777e-05, "loss": 2.7576, "step": 36773 }, { "epoch": 2.2828232664969894, "grad_norm": 0.468709848591668, "learning_rate": 1.642180138689656e-05, "loss": 2.6471, "step": 36774 }, { "epoch": 2.2828853435967473, "grad_norm": 0.536903924662441, "learning_rate": 1.6419125528009967e-05, "loss": 2.6974, "step": 36775 }, { "epoch": 2.2829474206965052, "grad_norm": 0.4285572097894638, "learning_rate": 1.6416449844322945e-05, "loss": 2.6531, "step": 36776 }, { "epoch": 2.2830094977962627, "grad_norm": 0.5416936378611017, "learning_rate": 1.6413774335849452e-05, "loss": 2.6806, "step": 36777 }, { "epoch": 2.283071574896021, "grad_norm": 0.5125036262756081, "learning_rate": 1.641109900260344e-05, "loss": 2.7093, "step": 36778 }, { "epoch": 2.2831336519957786, "grad_norm": 0.4659858585389436, "learning_rate": 1.6408423844598893e-05, "loss": 2.657, "step": 36779 }, { "epoch": 2.2831957290955365, "grad_norm": 0.5389780448323991, "learning_rate": 1.6405748861849758e-05, "loss": 2.6362, "step": 36780 }, { "epoch": 2.2832578061952944, "grad_norm": 0.48714250147554433, "learning_rate": 1.6403074054369982e-05, "loss": 2.6355, "step": 36781 }, { "epoch": 2.2833198832950523, "grad_norm": 0.47488224364316683, "learning_rate": 1.6400399422173507e-05, "loss": 2.6303, "step": 36782 }, { "epoch": 2.2833819603948102, "grad_norm": 0.5043867319248166, "learning_rate": 1.6397724965274308e-05, "loss": 2.6644, "step": 36783 }, { "epoch": 2.283444037494568, "grad_norm": 0.45534070670493376, "learning_rate": 1.6395050683686348e-05, "loss": 2.7051, "step": 36784 }, { "epoch": 2.283506114594326, "grad_norm": 0.4468784072052504, "learning_rate": 1.639237657742357e-05, "loss": 2.7196, "step": 36785 }, { "epoch": 2.283568191694084, "grad_norm": 0.47282754930849624, "learning_rate": 1.6389702646499922e-05, "loss": 2.6483, "step": 36786 }, { "epoch": 2.283630268793842, "grad_norm": 0.47541561640902874, "learning_rate": 1.638702889092935e-05, "loss": 2.6888, "step": 36787 }, { "epoch": 2.2836923458936, "grad_norm": 0.46150762245954774, "learning_rate": 1.6384355310725796e-05, "loss": 2.8029, "step": 36788 }, { "epoch": 2.2837544229933577, "grad_norm": 0.47646182200604786, "learning_rate": 1.6381681905903235e-05, "loss": 2.7688, "step": 36789 }, { "epoch": 2.2838165000931157, "grad_norm": 0.46991456328147513, "learning_rate": 1.6379008676475598e-05, "loss": 2.7078, "step": 36790 }, { "epoch": 2.2838785771928736, "grad_norm": 0.4501422014993724, "learning_rate": 1.637633562245684e-05, "loss": 2.6272, "step": 36791 }, { "epoch": 2.2839406542926315, "grad_norm": 0.5127133611499465, "learning_rate": 1.6373662743860883e-05, "loss": 2.7467, "step": 36792 }, { "epoch": 2.2840027313923894, "grad_norm": 0.476418082445937, "learning_rate": 1.6370990040701707e-05, "loss": 2.6583, "step": 36793 }, { "epoch": 2.2840648084921473, "grad_norm": 0.4748306857311194, "learning_rate": 1.6368317512993243e-05, "loss": 2.6894, "step": 36794 }, { "epoch": 2.2841268855919052, "grad_norm": 0.525661526294233, "learning_rate": 1.6365645160749426e-05, "loss": 2.7072, "step": 36795 }, { "epoch": 2.284188962691663, "grad_norm": 0.49326711654352995, "learning_rate": 1.6362972983984188e-05, "loss": 2.7995, "step": 36796 }, { "epoch": 2.284251039791421, "grad_norm": 0.5316779152309046, "learning_rate": 1.63603009827115e-05, "loss": 2.7098, "step": 36797 }, { "epoch": 2.284313116891179, "grad_norm": 0.5153905161969815, "learning_rate": 1.635762915694529e-05, "loss": 2.6787, "step": 36798 }, { "epoch": 2.284375193990937, "grad_norm": 0.5139686974278236, "learning_rate": 1.6354957506699498e-05, "loss": 2.7585, "step": 36799 }, { "epoch": 2.284437271090695, "grad_norm": 0.421553171109758, "learning_rate": 1.635228603198806e-05, "loss": 2.6811, "step": 36800 }, { "epoch": 2.2844993481904527, "grad_norm": 0.4648000552373548, "learning_rate": 1.6349614732824898e-05, "loss": 2.7235, "step": 36801 }, { "epoch": 2.2845614252902102, "grad_norm": 0.499243909652536, "learning_rate": 1.634694360922398e-05, "loss": 2.7142, "step": 36802 }, { "epoch": 2.2846235023899686, "grad_norm": 0.5053830647548062, "learning_rate": 1.6344272661199223e-05, "loss": 2.7664, "step": 36803 }, { "epoch": 2.284685579489726, "grad_norm": 0.4939622519391434, "learning_rate": 1.634160188876457e-05, "loss": 2.7432, "step": 36804 }, { "epoch": 2.284747656589484, "grad_norm": 0.4880439028856287, "learning_rate": 1.6338931291933936e-05, "loss": 2.6927, "step": 36805 }, { "epoch": 2.284809733689242, "grad_norm": 0.47225246048241404, "learning_rate": 1.6336260870721283e-05, "loss": 2.6229, "step": 36806 }, { "epoch": 2.284871810789, "grad_norm": 0.49049159441267826, "learning_rate": 1.6333590625140525e-05, "loss": 2.6083, "step": 36807 }, { "epoch": 2.2849338878887577, "grad_norm": 0.5060389726466834, "learning_rate": 1.6330920555205602e-05, "loss": 2.6535, "step": 36808 }, { "epoch": 2.2849959649885156, "grad_norm": 0.49389159671101696, "learning_rate": 1.632825066093042e-05, "loss": 2.7317, "step": 36809 }, { "epoch": 2.2850580420882736, "grad_norm": 0.4946774518214865, "learning_rate": 1.632558094232895e-05, "loss": 2.6944, "step": 36810 }, { "epoch": 2.2851201191880315, "grad_norm": 0.5366480285160492, "learning_rate": 1.6322911399415096e-05, "loss": 2.671, "step": 36811 }, { "epoch": 2.2851821962877894, "grad_norm": 0.4782325374368428, "learning_rate": 1.632024203220279e-05, "loss": 2.677, "step": 36812 }, { "epoch": 2.2852442733875473, "grad_norm": 0.4881263029298476, "learning_rate": 1.6317572840705958e-05, "loss": 2.623, "step": 36813 }, { "epoch": 2.2853063504873052, "grad_norm": 0.5200150610503289, "learning_rate": 1.631490382493851e-05, "loss": 2.7319, "step": 36814 }, { "epoch": 2.285368427587063, "grad_norm": 0.4695152394235616, "learning_rate": 1.6312234984914382e-05, "loss": 2.776, "step": 36815 }, { "epoch": 2.285430504686821, "grad_norm": 0.5142673768263537, "learning_rate": 1.6309566320647518e-05, "loss": 2.6652, "step": 36816 }, { "epoch": 2.285492581786579, "grad_norm": 0.47014725972863564, "learning_rate": 1.630689783215183e-05, "loss": 2.66, "step": 36817 }, { "epoch": 2.285554658886337, "grad_norm": 0.5074940196281563, "learning_rate": 1.6304229519441232e-05, "loss": 2.6545, "step": 36818 }, { "epoch": 2.285616735986095, "grad_norm": 0.466137823821008, "learning_rate": 1.6301561382529633e-05, "loss": 2.7432, "step": 36819 }, { "epoch": 2.2856788130858527, "grad_norm": 0.43796926263622016, "learning_rate": 1.6298893421430982e-05, "loss": 2.6586, "step": 36820 }, { "epoch": 2.2857408901856107, "grad_norm": 0.4662587084225402, "learning_rate": 1.629622563615919e-05, "loss": 2.6282, "step": 36821 }, { "epoch": 2.2858029672853686, "grad_norm": 0.48419076738379024, "learning_rate": 1.6293558026728166e-05, "loss": 2.6407, "step": 36822 }, { "epoch": 2.2858650443851265, "grad_norm": 0.44318759632123, "learning_rate": 1.6290890593151836e-05, "loss": 2.6778, "step": 36823 }, { "epoch": 2.2859271214848844, "grad_norm": 0.4406950800201458, "learning_rate": 1.6288223335444096e-05, "loss": 2.7012, "step": 36824 }, { "epoch": 2.285989198584642, "grad_norm": 0.4614853593419602, "learning_rate": 1.6285556253618895e-05, "loss": 2.7828, "step": 36825 }, { "epoch": 2.2860512756844003, "grad_norm": 0.47825685627694337, "learning_rate": 1.6282889347690134e-05, "loss": 2.6424, "step": 36826 }, { "epoch": 2.2861133527841577, "grad_norm": 0.483598733115213, "learning_rate": 1.628022261767172e-05, "loss": 2.6218, "step": 36827 }, { "epoch": 2.2861754298839156, "grad_norm": 0.44240570009152524, "learning_rate": 1.6277556063577552e-05, "loss": 2.6409, "step": 36828 }, { "epoch": 2.2862375069836736, "grad_norm": 0.47001516949238525, "learning_rate": 1.6274889685421584e-05, "loss": 2.691, "step": 36829 }, { "epoch": 2.2862995840834315, "grad_norm": 0.5276303960437936, "learning_rate": 1.6272223483217703e-05, "loss": 2.6659, "step": 36830 }, { "epoch": 2.2863616611831894, "grad_norm": 0.4461652888775626, "learning_rate": 1.6269557456979818e-05, "loss": 2.7109, "step": 36831 }, { "epoch": 2.2864237382829473, "grad_norm": 0.47991388738096474, "learning_rate": 1.6266891606721816e-05, "loss": 2.6331, "step": 36832 }, { "epoch": 2.2864858153827052, "grad_norm": 0.4844808924197317, "learning_rate": 1.6264225932457654e-05, "loss": 2.7253, "step": 36833 }, { "epoch": 2.286547892482463, "grad_norm": 0.482026760754323, "learning_rate": 1.6261560434201216e-05, "loss": 2.6446, "step": 36834 }, { "epoch": 2.286609969582221, "grad_norm": 0.4733621197356654, "learning_rate": 1.6258895111966406e-05, "loss": 2.7032, "step": 36835 }, { "epoch": 2.286672046681979, "grad_norm": 0.47480019708892857, "learning_rate": 1.6256229965767127e-05, "loss": 2.6668, "step": 36836 }, { "epoch": 2.286734123781737, "grad_norm": 0.5172208020416923, "learning_rate": 1.6253564995617276e-05, "loss": 2.6716, "step": 36837 }, { "epoch": 2.286796200881495, "grad_norm": 0.5096377215936773, "learning_rate": 1.6250900201530782e-05, "loss": 2.6616, "step": 36838 }, { "epoch": 2.2868582779812527, "grad_norm": 0.46513670016906816, "learning_rate": 1.6248235583521537e-05, "loss": 2.6456, "step": 36839 }, { "epoch": 2.2869203550810107, "grad_norm": 0.48685806386968133, "learning_rate": 1.6245571141603437e-05, "loss": 2.6303, "step": 36840 }, { "epoch": 2.2869824321807686, "grad_norm": 0.5375353147305237, "learning_rate": 1.6242906875790376e-05, "loss": 2.7641, "step": 36841 }, { "epoch": 2.2870445092805265, "grad_norm": 0.5225359711288601, "learning_rate": 1.6240242786096278e-05, "loss": 2.707, "step": 36842 }, { "epoch": 2.2871065863802844, "grad_norm": 0.4419044619967085, "learning_rate": 1.6237578872535024e-05, "loss": 2.6379, "step": 36843 }, { "epoch": 2.2871686634800423, "grad_norm": 0.5013242411154638, "learning_rate": 1.6234915135120526e-05, "loss": 2.7411, "step": 36844 }, { "epoch": 2.2872307405798002, "grad_norm": 0.4842836539981821, "learning_rate": 1.623225157386665e-05, "loss": 2.6194, "step": 36845 }, { "epoch": 2.287292817679558, "grad_norm": 0.46430149025366707, "learning_rate": 1.6229588188787332e-05, "loss": 2.7229, "step": 36846 }, { "epoch": 2.287354894779316, "grad_norm": 0.4727793604147157, "learning_rate": 1.6226924979896453e-05, "loss": 2.668, "step": 36847 }, { "epoch": 2.287416971879074, "grad_norm": 0.49785053646659855, "learning_rate": 1.6224261947207885e-05, "loss": 2.7383, "step": 36848 }, { "epoch": 2.287479048978832, "grad_norm": 0.48420175407447263, "learning_rate": 1.6221599090735563e-05, "loss": 2.6929, "step": 36849 }, { "epoch": 2.2875411260785894, "grad_norm": 0.49120417262248134, "learning_rate": 1.6218936410493356e-05, "loss": 2.6915, "step": 36850 }, { "epoch": 2.2876032031783478, "grad_norm": 0.48383941573155664, "learning_rate": 1.6216273906495143e-05, "loss": 2.748, "step": 36851 }, { "epoch": 2.2876652802781052, "grad_norm": 0.5297004596846482, "learning_rate": 1.6213611578754845e-05, "loss": 2.6626, "step": 36852 }, { "epoch": 2.287727357377863, "grad_norm": 0.43622958130038486, "learning_rate": 1.6210949427286338e-05, "loss": 2.7078, "step": 36853 }, { "epoch": 2.287789434477621, "grad_norm": 0.44320985398613394, "learning_rate": 1.620828745210351e-05, "loss": 2.7473, "step": 36854 }, { "epoch": 2.287851511577379, "grad_norm": 0.4769045430902693, "learning_rate": 1.6205625653220236e-05, "loss": 2.6851, "step": 36855 }, { "epoch": 2.287913588677137, "grad_norm": 0.4415446208025763, "learning_rate": 1.620296403065043e-05, "loss": 2.6157, "step": 36856 }, { "epoch": 2.287975665776895, "grad_norm": 0.4777532459939248, "learning_rate": 1.6200302584407966e-05, "loss": 2.6603, "step": 36857 }, { "epoch": 2.2880377428766527, "grad_norm": 0.45206841240561596, "learning_rate": 1.619764131450673e-05, "loss": 2.7592, "step": 36858 }, { "epoch": 2.2880998199764107, "grad_norm": 0.49237219054928205, "learning_rate": 1.619498022096058e-05, "loss": 2.7611, "step": 36859 }, { "epoch": 2.2881618970761686, "grad_norm": 0.4904903184487447, "learning_rate": 1.6192319303783453e-05, "loss": 2.7656, "step": 36860 }, { "epoch": 2.2882239741759265, "grad_norm": 0.48941374100455265, "learning_rate": 1.6189658562989197e-05, "loss": 2.7197, "step": 36861 }, { "epoch": 2.2882860512756844, "grad_norm": 0.49021744413889506, "learning_rate": 1.6186997998591703e-05, "loss": 2.7164, "step": 36862 }, { "epoch": 2.2883481283754423, "grad_norm": 0.5281757714058674, "learning_rate": 1.618433761060485e-05, "loss": 2.7618, "step": 36863 }, { "epoch": 2.2884102054752002, "grad_norm": 0.49589669827144606, "learning_rate": 1.6181677399042495e-05, "loss": 2.6819, "step": 36864 }, { "epoch": 2.288472282574958, "grad_norm": 0.5105325476482885, "learning_rate": 1.617901736391856e-05, "loss": 2.6249, "step": 36865 }, { "epoch": 2.288534359674716, "grad_norm": 0.4961443597404749, "learning_rate": 1.61763575052469e-05, "loss": 2.6449, "step": 36866 }, { "epoch": 2.288596436774474, "grad_norm": 0.4567283050184917, "learning_rate": 1.617369782304139e-05, "loss": 2.6948, "step": 36867 }, { "epoch": 2.288658513874232, "grad_norm": 0.5049138013526836, "learning_rate": 1.61710383173159e-05, "loss": 2.7225, "step": 36868 }, { "epoch": 2.28872059097399, "grad_norm": 0.5200172101895824, "learning_rate": 1.6168378988084325e-05, "loss": 2.7137, "step": 36869 }, { "epoch": 2.2887826680737478, "grad_norm": 0.4466104164336941, "learning_rate": 1.6165719835360537e-05, "loss": 2.7465, "step": 36870 }, { "epoch": 2.2888447451735057, "grad_norm": 0.5087383340158803, "learning_rate": 1.6163060859158395e-05, "loss": 2.6759, "step": 36871 }, { "epoch": 2.2889068222732636, "grad_norm": 0.4569007980796822, "learning_rate": 1.6160402059491787e-05, "loss": 2.6237, "step": 36872 }, { "epoch": 2.288968899373021, "grad_norm": 0.4418387361993817, "learning_rate": 1.6157743436374556e-05, "loss": 2.6909, "step": 36873 }, { "epoch": 2.2890309764727794, "grad_norm": 0.46207767032779834, "learning_rate": 1.6155084989820608e-05, "loss": 2.6388, "step": 36874 }, { "epoch": 2.289093053572537, "grad_norm": 0.4800750856481801, "learning_rate": 1.61524267198438e-05, "loss": 2.6874, "step": 36875 }, { "epoch": 2.289155130672295, "grad_norm": 0.5482811425057559, "learning_rate": 1.6149768626457994e-05, "loss": 2.7353, "step": 36876 }, { "epoch": 2.2892172077720527, "grad_norm": 0.5053091112541039, "learning_rate": 1.614711070967705e-05, "loss": 2.6422, "step": 36877 }, { "epoch": 2.2892792848718106, "grad_norm": 0.5353193939352977, "learning_rate": 1.6144452969514863e-05, "loss": 2.6509, "step": 36878 }, { "epoch": 2.2893413619715686, "grad_norm": 0.4897074051823883, "learning_rate": 1.6141795405985282e-05, "loss": 2.7098, "step": 36879 }, { "epoch": 2.2894034390713265, "grad_norm": 0.49904644459580727, "learning_rate": 1.613913801910218e-05, "loss": 2.7515, "step": 36880 }, { "epoch": 2.2894655161710844, "grad_norm": 0.49300274618362766, "learning_rate": 1.613648080887939e-05, "loss": 2.6322, "step": 36881 }, { "epoch": 2.2895275932708423, "grad_norm": 0.47699239758165196, "learning_rate": 1.6133823775330803e-05, "loss": 2.6667, "step": 36882 }, { "epoch": 2.2895896703706002, "grad_norm": 0.5178142081159415, "learning_rate": 1.6131166918470296e-05, "loss": 2.638, "step": 36883 }, { "epoch": 2.289651747470358, "grad_norm": 0.45296159626107146, "learning_rate": 1.6128510238311717e-05, "loss": 2.658, "step": 36884 }, { "epoch": 2.289713824570116, "grad_norm": 0.49122428496714154, "learning_rate": 1.6125853734868918e-05, "loss": 2.7149, "step": 36885 }, { "epoch": 2.289775901669874, "grad_norm": 0.5239263216355409, "learning_rate": 1.6123197408155765e-05, "loss": 2.7312, "step": 36886 }, { "epoch": 2.289837978769632, "grad_norm": 0.4960863007714589, "learning_rate": 1.61205412581861e-05, "loss": 2.6613, "step": 36887 }, { "epoch": 2.28990005586939, "grad_norm": 0.48351942917940155, "learning_rate": 1.611788528497381e-05, "loss": 2.6265, "step": 36888 }, { "epoch": 2.2899621329691477, "grad_norm": 0.46849927554690307, "learning_rate": 1.6115229488532734e-05, "loss": 2.6199, "step": 36889 }, { "epoch": 2.2900242100689057, "grad_norm": 0.4748370091724593, "learning_rate": 1.611257386887674e-05, "loss": 2.708, "step": 36890 }, { "epoch": 2.2900862871686636, "grad_norm": 0.45609774988661106, "learning_rate": 1.6109918426019648e-05, "loss": 2.6373, "step": 36891 }, { "epoch": 2.2901483642684215, "grad_norm": 0.5019672555084865, "learning_rate": 1.6107263159975365e-05, "loss": 2.8164, "step": 36892 }, { "epoch": 2.2902104413681794, "grad_norm": 0.5157814655459513, "learning_rate": 1.6104608070757714e-05, "loss": 2.6475, "step": 36893 }, { "epoch": 2.2902725184679373, "grad_norm": 0.5267788156884363, "learning_rate": 1.610195315838055e-05, "loss": 2.7123, "step": 36894 }, { "epoch": 2.2903345955676953, "grad_norm": 0.4558143390810381, "learning_rate": 1.609929842285771e-05, "loss": 2.6712, "step": 36895 }, { "epoch": 2.290396672667453, "grad_norm": 0.45093915147274083, "learning_rate": 1.6096643864203076e-05, "loss": 2.6562, "step": 36896 }, { "epoch": 2.290458749767211, "grad_norm": 0.48869753068132665, "learning_rate": 1.6093989482430483e-05, "loss": 2.585, "step": 36897 }, { "epoch": 2.2905208268669686, "grad_norm": 0.464335299754834, "learning_rate": 1.609133527755377e-05, "loss": 2.6649, "step": 36898 }, { "epoch": 2.290582903966727, "grad_norm": 0.4921577535450438, "learning_rate": 1.60886812495868e-05, "loss": 2.6487, "step": 36899 }, { "epoch": 2.2906449810664844, "grad_norm": 0.4934074028118123, "learning_rate": 1.6086027398543397e-05, "loss": 2.7548, "step": 36900 }, { "epoch": 2.2907070581662423, "grad_norm": 0.46595039390770815, "learning_rate": 1.6083373724437432e-05, "loss": 2.7018, "step": 36901 }, { "epoch": 2.2907691352660002, "grad_norm": 0.47707033458919873, "learning_rate": 1.6080720227282746e-05, "loss": 2.6462, "step": 36902 }, { "epoch": 2.290831212365758, "grad_norm": 0.46742421854315686, "learning_rate": 1.6078066907093175e-05, "loss": 2.583, "step": 36903 }, { "epoch": 2.290893289465516, "grad_norm": 0.4933687144261575, "learning_rate": 1.6075413763882546e-05, "loss": 2.739, "step": 36904 }, { "epoch": 2.290955366565274, "grad_norm": 0.4817794506917544, "learning_rate": 1.6072760797664737e-05, "loss": 2.6926, "step": 36905 }, { "epoch": 2.291017443665032, "grad_norm": 0.5167404305449093, "learning_rate": 1.607010800845357e-05, "loss": 2.6756, "step": 36906 }, { "epoch": 2.29107952076479, "grad_norm": 0.4870015120521979, "learning_rate": 1.6067455396262885e-05, "loss": 2.6916, "step": 36907 }, { "epoch": 2.2911415978645477, "grad_norm": 0.5157756874225964, "learning_rate": 1.6064802961106525e-05, "loss": 2.6724, "step": 36908 }, { "epoch": 2.2912036749643057, "grad_norm": 0.4708199573687362, "learning_rate": 1.6062150702998308e-05, "loss": 2.6821, "step": 36909 }, { "epoch": 2.2912657520640636, "grad_norm": 0.49298575926634686, "learning_rate": 1.6059498621952108e-05, "loss": 2.737, "step": 36910 }, { "epoch": 2.2913278291638215, "grad_norm": 0.4939686321336354, "learning_rate": 1.605684671798174e-05, "loss": 2.7393, "step": 36911 }, { "epoch": 2.2913899062635794, "grad_norm": 0.4567153197351213, "learning_rate": 1.6054194991101047e-05, "loss": 2.5716, "step": 36912 }, { "epoch": 2.2914519833633373, "grad_norm": 0.46823999953194356, "learning_rate": 1.6051543441323835e-05, "loss": 2.6147, "step": 36913 }, { "epoch": 2.2915140604630952, "grad_norm": 0.5403367337936267, "learning_rate": 1.6048892068663966e-05, "loss": 2.7227, "step": 36914 }, { "epoch": 2.291576137562853, "grad_norm": 0.4978615359852149, "learning_rate": 1.604624087313529e-05, "loss": 2.6859, "step": 36915 }, { "epoch": 2.291638214662611, "grad_norm": 0.5340866482064822, "learning_rate": 1.6043589854751613e-05, "loss": 2.6726, "step": 36916 }, { "epoch": 2.291700291762369, "grad_norm": 0.5004220611679594, "learning_rate": 1.6040939013526775e-05, "loss": 2.7174, "step": 36917 }, { "epoch": 2.291762368862127, "grad_norm": 0.4807598658713327, "learning_rate": 1.603828834947458e-05, "loss": 2.6445, "step": 36918 }, { "epoch": 2.291824445961885, "grad_norm": 0.481827589936115, "learning_rate": 1.6035637862608898e-05, "loss": 2.6758, "step": 36919 }, { "epoch": 2.2918865230616428, "grad_norm": 0.4747981100438866, "learning_rate": 1.603298755294354e-05, "loss": 2.6922, "step": 36920 }, { "epoch": 2.2919486001614002, "grad_norm": 0.4470078004235365, "learning_rate": 1.603033742049233e-05, "loss": 2.6897, "step": 36921 }, { "epoch": 2.2920106772611586, "grad_norm": 0.48280242242758475, "learning_rate": 1.60276874652691e-05, "loss": 2.6585, "step": 36922 }, { "epoch": 2.292072754360916, "grad_norm": 0.4732271482539504, "learning_rate": 1.6025037687287654e-05, "loss": 2.7041, "step": 36923 }, { "epoch": 2.292134831460674, "grad_norm": 0.48062264683696776, "learning_rate": 1.6022388086561846e-05, "loss": 2.6322, "step": 36924 }, { "epoch": 2.292196908560432, "grad_norm": 0.45419553379162564, "learning_rate": 1.6019738663105488e-05, "loss": 2.7059, "step": 36925 }, { "epoch": 2.29225898566019, "grad_norm": 0.4965665621678378, "learning_rate": 1.6017089416932408e-05, "loss": 2.7336, "step": 36926 }, { "epoch": 2.2923210627599477, "grad_norm": 0.503118801671107, "learning_rate": 1.60144403480564e-05, "loss": 2.6691, "step": 36927 }, { "epoch": 2.2923831398597057, "grad_norm": 0.4355244652153329, "learning_rate": 1.6011791456491327e-05, "loss": 2.6592, "step": 36928 }, { "epoch": 2.2924452169594636, "grad_norm": 0.4634838085998648, "learning_rate": 1.6009142742250988e-05, "loss": 2.7264, "step": 36929 }, { "epoch": 2.2925072940592215, "grad_norm": 0.4526791516674082, "learning_rate": 1.60064942053492e-05, "loss": 2.6071, "step": 36930 }, { "epoch": 2.2925693711589794, "grad_norm": 0.46037911683436905, "learning_rate": 1.600384584579977e-05, "loss": 2.8326, "step": 36931 }, { "epoch": 2.2926314482587373, "grad_norm": 0.6223143334012611, "learning_rate": 1.6001197663616545e-05, "loss": 2.6508, "step": 36932 }, { "epoch": 2.2926935253584952, "grad_norm": 0.4274177598763865, "learning_rate": 1.599854965881333e-05, "loss": 2.6745, "step": 36933 }, { "epoch": 2.292755602458253, "grad_norm": 0.4674521121064382, "learning_rate": 1.5995901831403925e-05, "loss": 2.6823, "step": 36934 }, { "epoch": 2.292817679558011, "grad_norm": 0.4745200587244535, "learning_rate": 1.599325418140216e-05, "loss": 2.6633, "step": 36935 }, { "epoch": 2.292879756657769, "grad_norm": 0.41957404346403626, "learning_rate": 1.599060670882183e-05, "loss": 2.6688, "step": 36936 }, { "epoch": 2.292941833757527, "grad_norm": 0.44698509847414625, "learning_rate": 1.5987959413676778e-05, "loss": 2.7104, "step": 36937 }, { "epoch": 2.293003910857285, "grad_norm": 0.49353587420783956, "learning_rate": 1.5985312295980797e-05, "loss": 2.7186, "step": 36938 }, { "epoch": 2.2930659879570428, "grad_norm": 0.4676609262372197, "learning_rate": 1.5982665355747695e-05, "loss": 2.7397, "step": 36939 }, { "epoch": 2.2931280650568007, "grad_norm": 0.43896453823465104, "learning_rate": 1.5980018592991275e-05, "loss": 2.731, "step": 36940 }, { "epoch": 2.2931901421565586, "grad_norm": 0.4635221945140999, "learning_rate": 1.597737200772537e-05, "loss": 2.5962, "step": 36941 }, { "epoch": 2.2932522192563165, "grad_norm": 0.48222390255632835, "learning_rate": 1.5974725599963773e-05, "loss": 2.6846, "step": 36942 }, { "epoch": 2.2933142963560744, "grad_norm": 0.49480956865015524, "learning_rate": 1.5972079369720295e-05, "loss": 2.7057, "step": 36943 }, { "epoch": 2.293376373455832, "grad_norm": 0.43819510213329343, "learning_rate": 1.5969433317008725e-05, "loss": 2.7005, "step": 36944 }, { "epoch": 2.2934384505555903, "grad_norm": 0.47367879113469896, "learning_rate": 1.5966787441842894e-05, "loss": 2.7445, "step": 36945 }, { "epoch": 2.2935005276553477, "grad_norm": 0.5054412350837945, "learning_rate": 1.5964141744236582e-05, "loss": 2.7752, "step": 36946 }, { "epoch": 2.2935626047551056, "grad_norm": 0.40299132001015203, "learning_rate": 1.596149622420362e-05, "loss": 2.7013, "step": 36947 }, { "epoch": 2.2936246818548636, "grad_norm": 0.4598786266243007, "learning_rate": 1.59588508817578e-05, "loss": 2.6359, "step": 36948 }, { "epoch": 2.2936867589546215, "grad_norm": 0.46429813644800105, "learning_rate": 1.595620571691291e-05, "loss": 2.5948, "step": 36949 }, { "epoch": 2.2937488360543794, "grad_norm": 0.43467051175979915, "learning_rate": 1.595356072968275e-05, "loss": 2.6232, "step": 36950 }, { "epoch": 2.2938109131541373, "grad_norm": 0.4663959521206316, "learning_rate": 1.5950915920081137e-05, "loss": 2.5937, "step": 36951 }, { "epoch": 2.2938729902538952, "grad_norm": 0.4520578314214318, "learning_rate": 1.5948271288121867e-05, "loss": 2.6069, "step": 36952 }, { "epoch": 2.293935067353653, "grad_norm": 0.4537071237719915, "learning_rate": 1.594562683381873e-05, "loss": 2.6403, "step": 36953 }, { "epoch": 2.293997144453411, "grad_norm": 0.4639468476007587, "learning_rate": 1.5942982557185514e-05, "loss": 2.6669, "step": 36954 }, { "epoch": 2.294059221553169, "grad_norm": 0.4695905778061489, "learning_rate": 1.5940338458236044e-05, "loss": 2.7166, "step": 36955 }, { "epoch": 2.294121298652927, "grad_norm": 0.4847959258440648, "learning_rate": 1.5937694536984088e-05, "loss": 2.7566, "step": 36956 }, { "epoch": 2.294183375752685, "grad_norm": 0.4843561414923696, "learning_rate": 1.5935050793443452e-05, "loss": 2.6585, "step": 36957 }, { "epoch": 2.2942454528524427, "grad_norm": 0.44310558298072245, "learning_rate": 1.593240722762792e-05, "loss": 2.6744, "step": 36958 }, { "epoch": 2.2943075299522007, "grad_norm": 0.4958163559702504, "learning_rate": 1.592976383955128e-05, "loss": 2.6176, "step": 36959 }, { "epoch": 2.2943696070519586, "grad_norm": 0.5208562146564856, "learning_rate": 1.592712062922735e-05, "loss": 2.6097, "step": 36960 }, { "epoch": 2.2944316841517165, "grad_norm": 0.453780062496664, "learning_rate": 1.59244775966699e-05, "loss": 2.6553, "step": 36961 }, { "epoch": 2.2944937612514744, "grad_norm": 0.479008261440127, "learning_rate": 1.5921834741892723e-05, "loss": 2.6493, "step": 36962 }, { "epoch": 2.2945558383512323, "grad_norm": 0.5422769709044393, "learning_rate": 1.5919192064909588e-05, "loss": 2.7143, "step": 36963 }, { "epoch": 2.2946179154509903, "grad_norm": 0.4824308788370533, "learning_rate": 1.5916549565734322e-05, "loss": 2.6537, "step": 36964 }, { "epoch": 2.294679992550748, "grad_norm": 0.4694757406520095, "learning_rate": 1.5913907244380687e-05, "loss": 2.6441, "step": 36965 }, { "epoch": 2.294742069650506, "grad_norm": 0.5957934201706034, "learning_rate": 1.5911265100862477e-05, "loss": 2.7582, "step": 36966 }, { "epoch": 2.294804146750264, "grad_norm": 0.47134101350532814, "learning_rate": 1.590862313519345e-05, "loss": 2.6116, "step": 36967 }, { "epoch": 2.294866223850022, "grad_norm": 0.4444625338790421, "learning_rate": 1.5905981347387434e-05, "loss": 2.6055, "step": 36968 }, { "epoch": 2.2949283009497794, "grad_norm": 0.5042209677445018, "learning_rate": 1.590333973745819e-05, "loss": 2.6998, "step": 36969 }, { "epoch": 2.2949903780495378, "grad_norm": 0.4590703518470611, "learning_rate": 1.5900698305419497e-05, "loss": 2.6062, "step": 36970 }, { "epoch": 2.2950524551492952, "grad_norm": 0.4344609669480269, "learning_rate": 1.5898057051285143e-05, "loss": 2.6412, "step": 36971 }, { "epoch": 2.295114532249053, "grad_norm": 0.5155543076577285, "learning_rate": 1.5895415975068884e-05, "loss": 2.6499, "step": 36972 }, { "epoch": 2.295176609348811, "grad_norm": 0.5042669794158244, "learning_rate": 1.5892775076784532e-05, "loss": 2.6743, "step": 36973 }, { "epoch": 2.295238686448569, "grad_norm": 0.4520098512966419, "learning_rate": 1.5890134356445856e-05, "loss": 2.7031, "step": 36974 }, { "epoch": 2.295300763548327, "grad_norm": 0.4494488168447294, "learning_rate": 1.5887493814066632e-05, "loss": 2.6354, "step": 36975 }, { "epoch": 2.295362840648085, "grad_norm": 0.4876236263657706, "learning_rate": 1.5884853449660612e-05, "loss": 2.6823, "step": 36976 }, { "epoch": 2.2954249177478427, "grad_norm": 0.46391761907186585, "learning_rate": 1.588221326324161e-05, "loss": 2.633, "step": 36977 }, { "epoch": 2.2954869948476007, "grad_norm": 0.4821501673290576, "learning_rate": 1.5879573254823393e-05, "loss": 2.6841, "step": 36978 }, { "epoch": 2.2955490719473586, "grad_norm": 0.5132650062162237, "learning_rate": 1.58769334244197e-05, "loss": 2.6895, "step": 36979 }, { "epoch": 2.2956111490471165, "grad_norm": 0.45943051022269715, "learning_rate": 1.587429377204435e-05, "loss": 2.7525, "step": 36980 }, { "epoch": 2.2956732261468744, "grad_norm": 0.43227329751325255, "learning_rate": 1.5871654297711075e-05, "loss": 2.7721, "step": 36981 }, { "epoch": 2.2957353032466323, "grad_norm": 0.5171139921289699, "learning_rate": 1.586901500143368e-05, "loss": 2.7521, "step": 36982 }, { "epoch": 2.2957973803463902, "grad_norm": 0.4984248802476896, "learning_rate": 1.586637588322592e-05, "loss": 2.6875, "step": 36983 }, { "epoch": 2.295859457446148, "grad_norm": 0.4582133858679004, "learning_rate": 1.5863736943101564e-05, "loss": 2.7369, "step": 36984 }, { "epoch": 2.295921534545906, "grad_norm": 0.49212799015255876, "learning_rate": 1.586109818107438e-05, "loss": 2.6633, "step": 36985 }, { "epoch": 2.295983611645664, "grad_norm": 0.5055554629536545, "learning_rate": 1.5858459597158115e-05, "loss": 2.6686, "step": 36986 }, { "epoch": 2.296045688745422, "grad_norm": 0.4752932201891944, "learning_rate": 1.585582119136657e-05, "loss": 2.6615, "step": 36987 }, { "epoch": 2.29610776584518, "grad_norm": 0.477166168720391, "learning_rate": 1.5853182963713504e-05, "loss": 2.7043, "step": 36988 }, { "epoch": 2.2961698429449378, "grad_norm": 0.4746207247660323, "learning_rate": 1.5850544914212662e-05, "loss": 2.6635, "step": 36989 }, { "epoch": 2.2962319200446957, "grad_norm": 0.469365002592975, "learning_rate": 1.58479070428778e-05, "loss": 2.702, "step": 36990 }, { "epoch": 2.2962939971444536, "grad_norm": 0.5122832758825108, "learning_rate": 1.5845269349722723e-05, "loss": 2.6039, "step": 36991 }, { "epoch": 2.296356074244211, "grad_norm": 0.5001326325455555, "learning_rate": 1.5842631834761162e-05, "loss": 2.7426, "step": 36992 }, { "epoch": 2.2964181513439694, "grad_norm": 0.47130448165566063, "learning_rate": 1.5839994498006883e-05, "loss": 2.6341, "step": 36993 }, { "epoch": 2.296480228443727, "grad_norm": 0.5155232647031512, "learning_rate": 1.583735733947363e-05, "loss": 2.7038, "step": 36994 }, { "epoch": 2.296542305543485, "grad_norm": 0.50272381466887, "learning_rate": 1.583472035917519e-05, "loss": 2.6792, "step": 36995 }, { "epoch": 2.2966043826432427, "grad_norm": 0.472844724672677, "learning_rate": 1.583208355712531e-05, "loss": 2.7667, "step": 36996 }, { "epoch": 2.2966664597430007, "grad_norm": 0.4784252003459767, "learning_rate": 1.5829446933337745e-05, "loss": 2.6997, "step": 36997 }, { "epoch": 2.2967285368427586, "grad_norm": 0.4892621068708896, "learning_rate": 1.582681048782625e-05, "loss": 2.5901, "step": 36998 }, { "epoch": 2.2967906139425165, "grad_norm": 0.47260403935873047, "learning_rate": 1.5824174220604567e-05, "loss": 2.577, "step": 36999 }, { "epoch": 2.2968526910422744, "grad_norm": 0.4781911099922988, "learning_rate": 1.5821538131686476e-05, "loss": 2.7262, "step": 37000 }, { "epoch": 2.2969147681420323, "grad_norm": 0.47570985593527326, "learning_rate": 1.5818902221085717e-05, "loss": 2.7308, "step": 37001 }, { "epoch": 2.2969768452417902, "grad_norm": 0.5049696060042153, "learning_rate": 1.5816266488816046e-05, "loss": 2.6746, "step": 37002 }, { "epoch": 2.297038922341548, "grad_norm": 0.46545573428608206, "learning_rate": 1.5813630934891195e-05, "loss": 2.6856, "step": 37003 }, { "epoch": 2.297100999441306, "grad_norm": 0.4247601715967055, "learning_rate": 1.5810995559324943e-05, "loss": 2.6496, "step": 37004 }, { "epoch": 2.297163076541064, "grad_norm": 0.454368168492233, "learning_rate": 1.580836036213103e-05, "loss": 2.6524, "step": 37005 }, { "epoch": 2.297225153640822, "grad_norm": 0.460967547843565, "learning_rate": 1.58057253433232e-05, "loss": 2.7293, "step": 37006 }, { "epoch": 2.29728723074058, "grad_norm": 0.4618452153906962, "learning_rate": 1.58030905029152e-05, "loss": 2.7185, "step": 37007 }, { "epoch": 2.2973493078403378, "grad_norm": 0.4761934571229638, "learning_rate": 1.580045584092077e-05, "loss": 2.6203, "step": 37008 }, { "epoch": 2.2974113849400957, "grad_norm": 0.4727356426302082, "learning_rate": 1.579782135735367e-05, "loss": 2.562, "step": 37009 }, { "epoch": 2.2974734620398536, "grad_norm": 0.4676976152422457, "learning_rate": 1.5795187052227644e-05, "loss": 2.6451, "step": 37010 }, { "epoch": 2.2975355391396115, "grad_norm": 0.42071947156340883, "learning_rate": 1.5792552925556432e-05, "loss": 2.7836, "step": 37011 }, { "epoch": 2.2975976162393694, "grad_norm": 0.460738058100004, "learning_rate": 1.5789918977353756e-05, "loss": 2.69, "step": 37012 }, { "epoch": 2.2976596933391273, "grad_norm": 0.4761221993550009, "learning_rate": 1.578728520763338e-05, "loss": 2.6409, "step": 37013 }, { "epoch": 2.2977217704388853, "grad_norm": 0.46212363781365823, "learning_rate": 1.5784651616409058e-05, "loss": 2.7134, "step": 37014 }, { "epoch": 2.297783847538643, "grad_norm": 0.508104856403839, "learning_rate": 1.5782018203694515e-05, "loss": 2.7, "step": 37015 }, { "epoch": 2.297845924638401, "grad_norm": 0.4202974458572378, "learning_rate": 1.5779384969503493e-05, "loss": 2.6467, "step": 37016 }, { "epoch": 2.2979080017381586, "grad_norm": 0.4177892245911826, "learning_rate": 1.5776751913849702e-05, "loss": 2.6708, "step": 37017 }, { "epoch": 2.297970078837917, "grad_norm": 0.5385873884595047, "learning_rate": 1.5774119036746926e-05, "loss": 2.6676, "step": 37018 }, { "epoch": 2.2980321559376744, "grad_norm": 0.47991044085675433, "learning_rate": 1.577148633820888e-05, "loss": 2.6026, "step": 37019 }, { "epoch": 2.2980942330374323, "grad_norm": 0.4806456821590169, "learning_rate": 1.5768853818249296e-05, "loss": 2.7143, "step": 37020 }, { "epoch": 2.2981563101371902, "grad_norm": 0.548709625809604, "learning_rate": 1.5766221476881917e-05, "loss": 2.7651, "step": 37021 }, { "epoch": 2.298218387236948, "grad_norm": 0.5107433253434236, "learning_rate": 1.5763589314120447e-05, "loss": 2.6522, "step": 37022 }, { "epoch": 2.298280464336706, "grad_norm": 0.45253351674902753, "learning_rate": 1.5760957329978665e-05, "loss": 2.7373, "step": 37023 }, { "epoch": 2.298342541436464, "grad_norm": 0.5063546896641906, "learning_rate": 1.575832552447028e-05, "loss": 2.5588, "step": 37024 }, { "epoch": 2.298404618536222, "grad_norm": 0.5277518494689715, "learning_rate": 1.575569389760902e-05, "loss": 2.6675, "step": 37025 }, { "epoch": 2.29846669563598, "grad_norm": 0.4777477409133718, "learning_rate": 1.5753062449408607e-05, "loss": 2.6941, "step": 37026 }, { "epoch": 2.2985287727357377, "grad_norm": 0.4971372188364265, "learning_rate": 1.575043117988279e-05, "loss": 2.6565, "step": 37027 }, { "epoch": 2.2985908498354957, "grad_norm": 0.47867760544166116, "learning_rate": 1.5747800089045296e-05, "loss": 2.6721, "step": 37028 }, { "epoch": 2.2986529269352536, "grad_norm": 0.4819347532310568, "learning_rate": 1.5745169176909836e-05, "loss": 2.6793, "step": 37029 }, { "epoch": 2.2987150040350115, "grad_norm": 0.5100765230232845, "learning_rate": 1.5742538443490135e-05, "loss": 2.6164, "step": 37030 }, { "epoch": 2.2987770811347694, "grad_norm": 0.4895416776372627, "learning_rate": 1.5739907888799936e-05, "loss": 2.6198, "step": 37031 }, { "epoch": 2.2988391582345273, "grad_norm": 0.49259644125660806, "learning_rate": 1.573727751285296e-05, "loss": 2.624, "step": 37032 }, { "epoch": 2.2989012353342853, "grad_norm": 0.5157440221572276, "learning_rate": 1.5734647315662927e-05, "loss": 2.6846, "step": 37033 }, { "epoch": 2.298963312434043, "grad_norm": 0.49323195188461694, "learning_rate": 1.573201729724355e-05, "loss": 2.7662, "step": 37034 }, { "epoch": 2.299025389533801, "grad_norm": 0.5019147390687171, "learning_rate": 1.5729387457608547e-05, "loss": 2.6334, "step": 37035 }, { "epoch": 2.299087466633559, "grad_norm": 0.506878507439723, "learning_rate": 1.5726757796771665e-05, "loss": 2.7178, "step": 37036 }, { "epoch": 2.299149543733317, "grad_norm": 0.5528826834731014, "learning_rate": 1.5724128314746605e-05, "loss": 2.7122, "step": 37037 }, { "epoch": 2.299211620833075, "grad_norm": 0.49952488916360643, "learning_rate": 1.5721499011547088e-05, "loss": 2.6766, "step": 37038 }, { "epoch": 2.2992736979328328, "grad_norm": 0.47520566302659567, "learning_rate": 1.5718869887186817e-05, "loss": 2.6853, "step": 37039 }, { "epoch": 2.2993357750325902, "grad_norm": 0.5211111440429235, "learning_rate": 1.5716240941679538e-05, "loss": 2.6523, "step": 37040 }, { "epoch": 2.2993978521323486, "grad_norm": 0.5525940086684711, "learning_rate": 1.5713612175038956e-05, "loss": 2.7725, "step": 37041 }, { "epoch": 2.299459929232106, "grad_norm": 0.42909070966189516, "learning_rate": 1.5710983587278784e-05, "loss": 2.6437, "step": 37042 }, { "epoch": 2.299522006331864, "grad_norm": 0.5952786793951431, "learning_rate": 1.5708355178412732e-05, "loss": 2.7207, "step": 37043 }, { "epoch": 2.299584083431622, "grad_norm": 0.5023849055160087, "learning_rate": 1.5705726948454496e-05, "loss": 2.6493, "step": 37044 }, { "epoch": 2.29964616053138, "grad_norm": 0.5315484442806434, "learning_rate": 1.570309889741781e-05, "loss": 2.6592, "step": 37045 }, { "epoch": 2.2997082376311377, "grad_norm": 0.5111398115542547, "learning_rate": 1.5700471025316406e-05, "loss": 2.7139, "step": 37046 }, { "epoch": 2.2997703147308957, "grad_norm": 0.47856580510068664, "learning_rate": 1.5697843332163965e-05, "loss": 2.686, "step": 37047 }, { "epoch": 2.2998323918306536, "grad_norm": 0.4952200005980652, "learning_rate": 1.56952158179742e-05, "loss": 2.6494, "step": 37048 }, { "epoch": 2.2998944689304115, "grad_norm": 0.5586237332493369, "learning_rate": 1.569258848276081e-05, "loss": 2.6579, "step": 37049 }, { "epoch": 2.2999565460301694, "grad_norm": 0.4678236868673208, "learning_rate": 1.5689961326537527e-05, "loss": 2.6826, "step": 37050 }, { "epoch": 2.3000186231299273, "grad_norm": 0.5296668417173896, "learning_rate": 1.568733434931805e-05, "loss": 2.6633, "step": 37051 }, { "epoch": 2.3000807002296852, "grad_norm": 0.5672171214541999, "learning_rate": 1.5684707551116073e-05, "loss": 2.5897, "step": 37052 }, { "epoch": 2.300142777329443, "grad_norm": 0.4872087831479076, "learning_rate": 1.56820809319453e-05, "loss": 2.6989, "step": 37053 }, { "epoch": 2.300204854429201, "grad_norm": 0.5467525072515601, "learning_rate": 1.5679454491819444e-05, "loss": 2.6002, "step": 37054 }, { "epoch": 2.300266931528959, "grad_norm": 0.5580267796482358, "learning_rate": 1.5676828230752217e-05, "loss": 2.641, "step": 37055 }, { "epoch": 2.300329008628717, "grad_norm": 0.5421044860522636, "learning_rate": 1.5674202148757304e-05, "loss": 2.7363, "step": 37056 }, { "epoch": 2.300391085728475, "grad_norm": 0.5223506632857166, "learning_rate": 1.567157624584841e-05, "loss": 2.6743, "step": 37057 }, { "epoch": 2.3004531628282328, "grad_norm": 0.5101727142714965, "learning_rate": 1.5668950522039228e-05, "loss": 2.7912, "step": 37058 }, { "epoch": 2.3005152399279907, "grad_norm": 0.5042998676521511, "learning_rate": 1.5666324977343476e-05, "loss": 2.6236, "step": 37059 }, { "epoch": 2.3005773170277486, "grad_norm": 0.5226439823826402, "learning_rate": 1.5663699611774846e-05, "loss": 2.7013, "step": 37060 }, { "epoch": 2.3006393941275065, "grad_norm": 0.4706340121739963, "learning_rate": 1.566107442534702e-05, "loss": 2.7286, "step": 37061 }, { "epoch": 2.3007014712272644, "grad_norm": 0.4880337965501228, "learning_rate": 1.56584494180737e-05, "loss": 2.7035, "step": 37062 }, { "epoch": 2.3007635483270223, "grad_norm": 0.48110240927247383, "learning_rate": 1.5655824589968598e-05, "loss": 2.7058, "step": 37063 }, { "epoch": 2.3008256254267803, "grad_norm": 0.4541969937368653, "learning_rate": 1.5653199941045393e-05, "loss": 2.7025, "step": 37064 }, { "epoch": 2.3008877025265377, "grad_norm": 0.49837336590998693, "learning_rate": 1.5650575471317786e-05, "loss": 2.7667, "step": 37065 }, { "epoch": 2.300949779626296, "grad_norm": 0.4812616458325933, "learning_rate": 1.5647951180799446e-05, "loss": 2.7453, "step": 37066 }, { "epoch": 2.3010118567260536, "grad_norm": 0.4804772166925029, "learning_rate": 1.5645327069504106e-05, "loss": 2.6543, "step": 37067 }, { "epoch": 2.3010739338258115, "grad_norm": 0.4964431223324627, "learning_rate": 1.564270313744543e-05, "loss": 2.7102, "step": 37068 }, { "epoch": 2.3011360109255694, "grad_norm": 0.4577546385827456, "learning_rate": 1.5640079384637115e-05, "loss": 2.6839, "step": 37069 }, { "epoch": 2.3011980880253273, "grad_norm": 0.47734234274299325, "learning_rate": 1.5637455811092842e-05, "loss": 2.77, "step": 37070 }, { "epoch": 2.3012601651250852, "grad_norm": 0.4895493938640216, "learning_rate": 1.5634832416826288e-05, "loss": 2.6799, "step": 37071 }, { "epoch": 2.301322242224843, "grad_norm": 0.4846754065455385, "learning_rate": 1.5632209201851177e-05, "loss": 2.6445, "step": 37072 }, { "epoch": 2.301384319324601, "grad_norm": 0.5210633310478605, "learning_rate": 1.562958616618117e-05, "loss": 2.7096, "step": 37073 }, { "epoch": 2.301446396424359, "grad_norm": 0.49910836887428384, "learning_rate": 1.5626963309829954e-05, "loss": 2.6307, "step": 37074 }, { "epoch": 2.301508473524117, "grad_norm": 0.5082049393969671, "learning_rate": 1.5624340632811208e-05, "loss": 2.6823, "step": 37075 }, { "epoch": 2.301570550623875, "grad_norm": 0.4809552652549272, "learning_rate": 1.562171813513863e-05, "loss": 2.6561, "step": 37076 }, { "epoch": 2.3016326277236328, "grad_norm": 0.4835590134379477, "learning_rate": 1.561909581682588e-05, "loss": 2.693, "step": 37077 }, { "epoch": 2.3016947048233907, "grad_norm": 0.47296652580068427, "learning_rate": 1.5616473677886666e-05, "loss": 2.7165, "step": 37078 }, { "epoch": 2.3017567819231486, "grad_norm": 0.5358052738440624, "learning_rate": 1.561385171833466e-05, "loss": 2.6671, "step": 37079 }, { "epoch": 2.3018188590229065, "grad_norm": 0.5079775321996838, "learning_rate": 1.5611229938183518e-05, "loss": 2.6772, "step": 37080 }, { "epoch": 2.3018809361226644, "grad_norm": 0.46462172326817536, "learning_rate": 1.5608608337446956e-05, "loss": 2.608, "step": 37081 }, { "epoch": 2.3019430132224223, "grad_norm": 0.5435611594250384, "learning_rate": 1.5605986916138636e-05, "loss": 2.657, "step": 37082 }, { "epoch": 2.3020050903221803, "grad_norm": 0.4828344889451299, "learning_rate": 1.5603365674272224e-05, "loss": 2.6504, "step": 37083 }, { "epoch": 2.302067167421938, "grad_norm": 0.49320391503386635, "learning_rate": 1.5600744611861414e-05, "loss": 2.6818, "step": 37084 }, { "epoch": 2.302129244521696, "grad_norm": 0.5283879714485424, "learning_rate": 1.5598123728919846e-05, "loss": 2.6666, "step": 37085 }, { "epoch": 2.302191321621454, "grad_norm": 0.469065162884945, "learning_rate": 1.5595503025461233e-05, "loss": 2.7166, "step": 37086 }, { "epoch": 2.302253398721212, "grad_norm": 0.4627042950667662, "learning_rate": 1.5592882501499235e-05, "loss": 2.7283, "step": 37087 }, { "epoch": 2.3023154758209694, "grad_norm": 0.47583856472317443, "learning_rate": 1.5590262157047526e-05, "loss": 2.662, "step": 37088 }, { "epoch": 2.3023775529207278, "grad_norm": 0.5002396128623181, "learning_rate": 1.558764199211975e-05, "loss": 2.7196, "step": 37089 }, { "epoch": 2.3024396300204852, "grad_norm": 0.45139432765259857, "learning_rate": 1.5585022006729617e-05, "loss": 2.7172, "step": 37090 }, { "epoch": 2.302501707120243, "grad_norm": 0.5326274516925013, "learning_rate": 1.5582402200890778e-05, "loss": 2.6021, "step": 37091 }, { "epoch": 2.302563784220001, "grad_norm": 0.5205703898132074, "learning_rate": 1.557978257461691e-05, "loss": 2.7504, "step": 37092 }, { "epoch": 2.302625861319759, "grad_norm": 0.4924821577132063, "learning_rate": 1.5577163127921663e-05, "loss": 2.5908, "step": 37093 }, { "epoch": 2.302687938419517, "grad_norm": 0.4689320369749496, "learning_rate": 1.5574543860818695e-05, "loss": 2.5916, "step": 37094 }, { "epoch": 2.302750015519275, "grad_norm": 0.4989452412771809, "learning_rate": 1.5571924773321712e-05, "loss": 2.6685, "step": 37095 }, { "epoch": 2.3028120926190327, "grad_norm": 0.4941838965486603, "learning_rate": 1.556930586544435e-05, "loss": 2.7271, "step": 37096 }, { "epoch": 2.3028741697187907, "grad_norm": 0.43454975551031694, "learning_rate": 1.556668713720028e-05, "loss": 2.6564, "step": 37097 }, { "epoch": 2.3029362468185486, "grad_norm": 0.5072078038897622, "learning_rate": 1.556406858860315e-05, "loss": 2.6927, "step": 37098 }, { "epoch": 2.3029983239183065, "grad_norm": 0.5131046659132182, "learning_rate": 1.556145021966664e-05, "loss": 2.6098, "step": 37099 }, { "epoch": 2.3030604010180644, "grad_norm": 0.46223440212157735, "learning_rate": 1.5558832030404412e-05, "loss": 2.745, "step": 37100 }, { "epoch": 2.3031224781178223, "grad_norm": 0.5230764880639744, "learning_rate": 1.555621402083012e-05, "loss": 2.6743, "step": 37101 }, { "epoch": 2.3031845552175803, "grad_norm": 0.493148622176686, "learning_rate": 1.5553596190957397e-05, "loss": 2.6337, "step": 37102 }, { "epoch": 2.303246632317338, "grad_norm": 0.4936676023032453, "learning_rate": 1.5550978540799943e-05, "loss": 2.657, "step": 37103 }, { "epoch": 2.303308709417096, "grad_norm": 0.48437512305473485, "learning_rate": 1.5548361070371404e-05, "loss": 2.6009, "step": 37104 }, { "epoch": 2.303370786516854, "grad_norm": 0.5030213503960992, "learning_rate": 1.5545743779685422e-05, "loss": 2.6443, "step": 37105 }, { "epoch": 2.303432863616612, "grad_norm": 0.5001956139816454, "learning_rate": 1.5543126668755658e-05, "loss": 2.7204, "step": 37106 }, { "epoch": 2.30349494071637, "grad_norm": 0.4652927858468954, "learning_rate": 1.5540509737595754e-05, "loss": 2.6579, "step": 37107 }, { "epoch": 2.3035570178161278, "grad_norm": 0.4859308358867018, "learning_rate": 1.5537892986219394e-05, "loss": 2.7326, "step": 37108 }, { "epoch": 2.3036190949158857, "grad_norm": 0.46123843956496535, "learning_rate": 1.553527641464021e-05, "loss": 2.6836, "step": 37109 }, { "epoch": 2.3036811720156436, "grad_norm": 0.4982879504007189, "learning_rate": 1.5532660022871836e-05, "loss": 2.6219, "step": 37110 }, { "epoch": 2.3037432491154015, "grad_norm": 0.4624657270259384, "learning_rate": 1.5530043810927964e-05, "loss": 2.691, "step": 37111 }, { "epoch": 2.3038053262151594, "grad_norm": 0.4622354498306785, "learning_rate": 1.55274277788222e-05, "loss": 2.7147, "step": 37112 }, { "epoch": 2.303867403314917, "grad_norm": 0.4388294286913978, "learning_rate": 1.5524811926568234e-05, "loss": 2.7444, "step": 37113 }, { "epoch": 2.3039294804146753, "grad_norm": 0.4673412133916123, "learning_rate": 1.5522196254179687e-05, "loss": 2.6407, "step": 37114 }, { "epoch": 2.3039915575144327, "grad_norm": 0.5213330510068375, "learning_rate": 1.5519580761670215e-05, "loss": 2.6863, "step": 37115 }, { "epoch": 2.3040536346141907, "grad_norm": 0.4663593843770465, "learning_rate": 1.551696544905345e-05, "loss": 2.6672, "step": 37116 }, { "epoch": 2.3041157117139486, "grad_norm": 0.5347469463109609, "learning_rate": 1.5514350316343057e-05, "loss": 2.7234, "step": 37117 }, { "epoch": 2.3041777888137065, "grad_norm": 0.5798907764128212, "learning_rate": 1.5511735363552672e-05, "loss": 2.6824, "step": 37118 }, { "epoch": 2.3042398659134644, "grad_norm": 0.4782606107090072, "learning_rate": 1.5509120590695937e-05, "loss": 2.6753, "step": 37119 }, { "epoch": 2.3043019430132223, "grad_norm": 0.4981523439519716, "learning_rate": 1.550650599778649e-05, "loss": 2.7136, "step": 37120 }, { "epoch": 2.3043640201129802, "grad_norm": 0.5319017171369083, "learning_rate": 1.5503891584837965e-05, "loss": 2.7338, "step": 37121 }, { "epoch": 2.304426097212738, "grad_norm": 0.4956580373761754, "learning_rate": 1.550127735186402e-05, "loss": 2.6811, "step": 37122 }, { "epoch": 2.304488174312496, "grad_norm": 0.4804322143557713, "learning_rate": 1.549866329887829e-05, "loss": 2.6783, "step": 37123 }, { "epoch": 2.304550251412254, "grad_norm": 0.5011307625918575, "learning_rate": 1.549604942589441e-05, "loss": 2.6429, "step": 37124 }, { "epoch": 2.304612328512012, "grad_norm": 0.5262626303473223, "learning_rate": 1.5493435732925997e-05, "loss": 2.6366, "step": 37125 }, { "epoch": 2.30467440561177, "grad_norm": 0.4841583136441487, "learning_rate": 1.549082221998673e-05, "loss": 2.7628, "step": 37126 }, { "epoch": 2.3047364827115278, "grad_norm": 0.47004859972338336, "learning_rate": 1.548820888709021e-05, "loss": 2.7095, "step": 37127 }, { "epoch": 2.3047985598112857, "grad_norm": 0.5302138321397152, "learning_rate": 1.5485595734250086e-05, "loss": 2.698, "step": 37128 }, { "epoch": 2.3048606369110436, "grad_norm": 0.46943985720781484, "learning_rate": 1.5482982761479992e-05, "loss": 2.6369, "step": 37129 }, { "epoch": 2.3049227140108015, "grad_norm": 0.5213125853163993, "learning_rate": 1.5480369968793534e-05, "loss": 2.6529, "step": 37130 }, { "epoch": 2.3049847911105594, "grad_norm": 0.5244740712872628, "learning_rate": 1.5477757356204387e-05, "loss": 2.7238, "step": 37131 }, { "epoch": 2.3050468682103173, "grad_norm": 0.5171968891409437, "learning_rate": 1.547514492372616e-05, "loss": 2.7036, "step": 37132 }, { "epoch": 2.3051089453100753, "grad_norm": 0.49821457376974737, "learning_rate": 1.5472532671372482e-05, "loss": 2.7227, "step": 37133 }, { "epoch": 2.305171022409833, "grad_norm": 0.5321243888003759, "learning_rate": 1.546992059915697e-05, "loss": 2.6191, "step": 37134 }, { "epoch": 2.305233099509591, "grad_norm": 0.47082149236131354, "learning_rate": 1.5467308707093274e-05, "loss": 2.6501, "step": 37135 }, { "epoch": 2.3052951766093486, "grad_norm": 0.49235454631022396, "learning_rate": 1.5464696995195016e-05, "loss": 2.6696, "step": 37136 }, { "epoch": 2.305357253709107, "grad_norm": 0.4920706080337173, "learning_rate": 1.5462085463475817e-05, "loss": 2.7388, "step": 37137 }, { "epoch": 2.3054193308088644, "grad_norm": 0.5072521639515014, "learning_rate": 1.5459474111949284e-05, "loss": 2.6832, "step": 37138 }, { "epoch": 2.3054814079086223, "grad_norm": 0.4729731104548634, "learning_rate": 1.545686294062908e-05, "loss": 2.7244, "step": 37139 }, { "epoch": 2.3055434850083802, "grad_norm": 0.45478541247438664, "learning_rate": 1.5454251949528804e-05, "loss": 2.6159, "step": 37140 }, { "epoch": 2.305605562108138, "grad_norm": 0.47973392087997346, "learning_rate": 1.545164113866208e-05, "loss": 2.6418, "step": 37141 }, { "epoch": 2.305667639207896, "grad_norm": 0.4797000939272291, "learning_rate": 1.5449030508042538e-05, "loss": 2.7165, "step": 37142 }, { "epoch": 2.305729716307654, "grad_norm": 0.4208922418903959, "learning_rate": 1.544642005768377e-05, "loss": 2.7112, "step": 37143 }, { "epoch": 2.305791793407412, "grad_norm": 0.4647889465595548, "learning_rate": 1.5443809787599418e-05, "loss": 2.6306, "step": 37144 }, { "epoch": 2.30585387050717, "grad_norm": 0.4687296704016432, "learning_rate": 1.5441199697803117e-05, "loss": 2.7445, "step": 37145 }, { "epoch": 2.3059159476069278, "grad_norm": 0.49846401323458567, "learning_rate": 1.5438589788308466e-05, "loss": 2.5901, "step": 37146 }, { "epoch": 2.3059780247066857, "grad_norm": 0.47075932907183576, "learning_rate": 1.543598005912908e-05, "loss": 2.6686, "step": 37147 }, { "epoch": 2.3060401018064436, "grad_norm": 0.48691133206883824, "learning_rate": 1.543337051027856e-05, "loss": 2.7077, "step": 37148 }, { "epoch": 2.3061021789062015, "grad_norm": 0.5020440164053515, "learning_rate": 1.5430761141770555e-05, "loss": 2.6981, "step": 37149 }, { "epoch": 2.3061642560059594, "grad_norm": 0.5012992726697197, "learning_rate": 1.5428151953618662e-05, "loss": 2.7127, "step": 37150 }, { "epoch": 2.3062263331057173, "grad_norm": 0.435134059127884, "learning_rate": 1.542554294583649e-05, "loss": 2.8133, "step": 37151 }, { "epoch": 2.3062884102054753, "grad_norm": 0.5109609860963117, "learning_rate": 1.5422934118437643e-05, "loss": 2.6579, "step": 37152 }, { "epoch": 2.306350487305233, "grad_norm": 0.5008790514301128, "learning_rate": 1.5420325471435754e-05, "loss": 2.6343, "step": 37153 }, { "epoch": 2.306412564404991, "grad_norm": 0.46981788429584137, "learning_rate": 1.541771700484442e-05, "loss": 2.6611, "step": 37154 }, { "epoch": 2.306474641504749, "grad_norm": 0.5334939966748866, "learning_rate": 1.5415108718677263e-05, "loss": 2.6056, "step": 37155 }, { "epoch": 2.306536718604507, "grad_norm": 0.4909135134183474, "learning_rate": 1.5412500612947866e-05, "loss": 2.6637, "step": 37156 }, { "epoch": 2.306598795704265, "grad_norm": 0.4866176187879984, "learning_rate": 1.5409892687669842e-05, "loss": 2.7217, "step": 37157 }, { "epoch": 2.3066608728040228, "grad_norm": 0.4984306352913947, "learning_rate": 1.5407284942856816e-05, "loss": 2.6503, "step": 37158 }, { "epoch": 2.3067229499037807, "grad_norm": 0.5031241920417316, "learning_rate": 1.5404677378522387e-05, "loss": 2.6948, "step": 37159 }, { "epoch": 2.3067850270035386, "grad_norm": 0.5106748459951933, "learning_rate": 1.5402069994680153e-05, "loss": 2.661, "step": 37160 }, { "epoch": 2.306847104103296, "grad_norm": 0.4702973100357378, "learning_rate": 1.53994627913437e-05, "loss": 2.7303, "step": 37161 }, { "epoch": 2.3069091812030544, "grad_norm": 0.48296666754844736, "learning_rate": 1.5396855768526663e-05, "loss": 2.7242, "step": 37162 }, { "epoch": 2.306971258302812, "grad_norm": 0.49812087462953364, "learning_rate": 1.5394248926242634e-05, "loss": 2.6241, "step": 37163 }, { "epoch": 2.30703333540257, "grad_norm": 0.4949878169978407, "learning_rate": 1.5391642264505206e-05, "loss": 2.6745, "step": 37164 }, { "epoch": 2.3070954125023277, "grad_norm": 0.47612011810487953, "learning_rate": 1.5389035783327972e-05, "loss": 2.6572, "step": 37165 }, { "epoch": 2.3071574896020857, "grad_norm": 0.527171382284566, "learning_rate": 1.5386429482724553e-05, "loss": 2.6818, "step": 37166 }, { "epoch": 2.3072195667018436, "grad_norm": 0.509946096928278, "learning_rate": 1.5383823362708534e-05, "loss": 2.5964, "step": 37167 }, { "epoch": 2.3072816438016015, "grad_norm": 0.4971579603702522, "learning_rate": 1.538121742329351e-05, "loss": 2.738, "step": 37168 }, { "epoch": 2.3073437209013594, "grad_norm": 0.5536640288651659, "learning_rate": 1.5378611664493077e-05, "loss": 2.6948, "step": 37169 }, { "epoch": 2.3074057980011173, "grad_norm": 0.5048262012847299, "learning_rate": 1.5376006086320822e-05, "loss": 2.7377, "step": 37170 }, { "epoch": 2.3074678751008753, "grad_norm": 0.502522839174255, "learning_rate": 1.5373400688790358e-05, "loss": 2.7181, "step": 37171 }, { "epoch": 2.307529952200633, "grad_norm": 0.521225339868795, "learning_rate": 1.5370795471915274e-05, "loss": 2.7145, "step": 37172 }, { "epoch": 2.307592029300391, "grad_norm": 0.4919460097778712, "learning_rate": 1.5368190435709153e-05, "loss": 2.7175, "step": 37173 }, { "epoch": 2.307654106400149, "grad_norm": 0.4912479758281598, "learning_rate": 1.5365585580185578e-05, "loss": 2.5338, "step": 37174 }, { "epoch": 2.307716183499907, "grad_norm": 0.522665963987848, "learning_rate": 1.536298090535816e-05, "loss": 2.6655, "step": 37175 }, { "epoch": 2.307778260599665, "grad_norm": 0.4666452999701487, "learning_rate": 1.5360376411240466e-05, "loss": 2.7555, "step": 37176 }, { "epoch": 2.3078403376994228, "grad_norm": 0.5192959761837762, "learning_rate": 1.535777209784612e-05, "loss": 2.7598, "step": 37177 }, { "epoch": 2.3079024147991807, "grad_norm": 0.5188510428687352, "learning_rate": 1.5355167965188678e-05, "loss": 2.6864, "step": 37178 }, { "epoch": 2.3079644918989386, "grad_norm": 0.49193590788013536, "learning_rate": 1.5352564013281735e-05, "loss": 2.6928, "step": 37179 }, { "epoch": 2.3080265689986965, "grad_norm": 0.5019000726710037, "learning_rate": 1.5349960242138867e-05, "loss": 2.5914, "step": 37180 }, { "epoch": 2.3080886460984544, "grad_norm": 0.4744204549409798, "learning_rate": 1.5347356651773677e-05, "loss": 2.69, "step": 37181 }, { "epoch": 2.3081507231982124, "grad_norm": 0.4990906940682171, "learning_rate": 1.5344753242199744e-05, "loss": 2.6605, "step": 37182 }, { "epoch": 2.3082128002979703, "grad_norm": 0.4859023164235052, "learning_rate": 1.5342150013430646e-05, "loss": 2.6887, "step": 37183 }, { "epoch": 2.3082748773977277, "grad_norm": 0.5007526276453317, "learning_rate": 1.5339546965479944e-05, "loss": 2.7188, "step": 37184 }, { "epoch": 2.308336954497486, "grad_norm": 0.49742809077879224, "learning_rate": 1.533694409836126e-05, "loss": 2.7064, "step": 37185 }, { "epoch": 2.3083990315972436, "grad_norm": 0.48877108608916964, "learning_rate": 1.5334341412088155e-05, "loss": 2.6913, "step": 37186 }, { "epoch": 2.3084611086970015, "grad_norm": 0.5243973827018136, "learning_rate": 1.5331738906674202e-05, "loss": 2.5702, "step": 37187 }, { "epoch": 2.3085231857967594, "grad_norm": 0.4470637169747659, "learning_rate": 1.5329136582132975e-05, "loss": 2.6349, "step": 37188 }, { "epoch": 2.3085852628965173, "grad_norm": 0.42377141054465234, "learning_rate": 1.532653443847807e-05, "loss": 2.7864, "step": 37189 }, { "epoch": 2.3086473399962752, "grad_norm": 0.49795773774374935, "learning_rate": 1.5323932475723046e-05, "loss": 2.6725, "step": 37190 }, { "epoch": 2.308709417096033, "grad_norm": 0.5120492697680101, "learning_rate": 1.5321330693881493e-05, "loss": 2.7008, "step": 37191 }, { "epoch": 2.308771494195791, "grad_norm": 0.456584875458013, "learning_rate": 1.531872909296697e-05, "loss": 2.6467, "step": 37192 }, { "epoch": 2.308833571295549, "grad_norm": 0.47717934694551095, "learning_rate": 1.5316127672993048e-05, "loss": 2.5523, "step": 37193 }, { "epoch": 2.308895648395307, "grad_norm": 0.4919412390424219, "learning_rate": 1.531352643397332e-05, "loss": 2.599, "step": 37194 }, { "epoch": 2.308957725495065, "grad_norm": 0.49420308860448936, "learning_rate": 1.531092537592134e-05, "loss": 2.6955, "step": 37195 }, { "epoch": 2.3090198025948228, "grad_norm": 0.5011342713185144, "learning_rate": 1.530832449885069e-05, "loss": 2.5872, "step": 37196 }, { "epoch": 2.3090818796945807, "grad_norm": 0.4303896801949083, "learning_rate": 1.5305723802774914e-05, "loss": 2.6261, "step": 37197 }, { "epoch": 2.3091439567943386, "grad_norm": 0.4802152228395608, "learning_rate": 1.5303123287707614e-05, "loss": 2.7253, "step": 37198 }, { "epoch": 2.3092060338940965, "grad_norm": 0.4721743120716129, "learning_rate": 1.5300522953662344e-05, "loss": 2.6112, "step": 37199 }, { "epoch": 2.3092681109938544, "grad_norm": 0.5058240841371338, "learning_rate": 1.529792280065267e-05, "loss": 2.6503, "step": 37200 }, { "epoch": 2.3093301880936123, "grad_norm": 0.4856011806081296, "learning_rate": 1.5295322828692142e-05, "loss": 2.6637, "step": 37201 }, { "epoch": 2.3093922651933703, "grad_norm": 0.48496178493663844, "learning_rate": 1.5292723037794356e-05, "loss": 2.7152, "step": 37202 }, { "epoch": 2.309454342293128, "grad_norm": 0.46296229391137983, "learning_rate": 1.5290123427972857e-05, "loss": 2.7748, "step": 37203 }, { "epoch": 2.309516419392886, "grad_norm": 0.45624815862101165, "learning_rate": 1.528752399924121e-05, "loss": 2.6157, "step": 37204 }, { "epoch": 2.309578496492644, "grad_norm": 0.49251141561552086, "learning_rate": 1.5284924751612977e-05, "loss": 2.6013, "step": 37205 }, { "epoch": 2.309640573592402, "grad_norm": 0.4415744865988115, "learning_rate": 1.5282325685101708e-05, "loss": 2.7054, "step": 37206 }, { "epoch": 2.30970265069216, "grad_norm": 0.47355226441370507, "learning_rate": 1.5279726799720983e-05, "loss": 2.6236, "step": 37207 }, { "epoch": 2.3097647277919178, "grad_norm": 0.5197641811353662, "learning_rate": 1.527712809548434e-05, "loss": 2.7205, "step": 37208 }, { "epoch": 2.3098268048916752, "grad_norm": 0.4593164581832146, "learning_rate": 1.5274529572405367e-05, "loss": 2.6297, "step": 37209 }, { "epoch": 2.3098888819914336, "grad_norm": 0.5046816928871928, "learning_rate": 1.52719312304976e-05, "loss": 2.6912, "step": 37210 }, { "epoch": 2.309950959091191, "grad_norm": 0.4613819081273226, "learning_rate": 1.5269333069774576e-05, "loss": 2.6142, "step": 37211 }, { "epoch": 2.310013036190949, "grad_norm": 0.49731409356577577, "learning_rate": 1.5266735090249897e-05, "loss": 2.6504, "step": 37212 }, { "epoch": 2.310075113290707, "grad_norm": 0.477553821795898, "learning_rate": 1.526413729193709e-05, "loss": 2.7356, "step": 37213 }, { "epoch": 2.310137190390465, "grad_norm": 0.4900991093627661, "learning_rate": 1.5261539674849712e-05, "loss": 2.6663, "step": 37214 }, { "epoch": 2.3101992674902228, "grad_norm": 0.5141178994053632, "learning_rate": 1.5258942239001302e-05, "loss": 2.5939, "step": 37215 }, { "epoch": 2.3102613445899807, "grad_norm": 0.5085002057235437, "learning_rate": 1.5256344984405435e-05, "loss": 2.6936, "step": 37216 }, { "epoch": 2.3103234216897386, "grad_norm": 0.48646024265227733, "learning_rate": 1.525374791107565e-05, "loss": 2.6009, "step": 37217 }, { "epoch": 2.3103854987894965, "grad_norm": 0.46171183121043047, "learning_rate": 1.52511510190255e-05, "loss": 2.5832, "step": 37218 }, { "epoch": 2.3104475758892544, "grad_norm": 0.5242333411561025, "learning_rate": 1.5248554308268531e-05, "loss": 2.6926, "step": 37219 }, { "epoch": 2.3105096529890123, "grad_norm": 0.5213461417673975, "learning_rate": 1.524595777881827e-05, "loss": 2.7119, "step": 37220 }, { "epoch": 2.3105717300887703, "grad_norm": 0.47485425306718926, "learning_rate": 1.5243361430688307e-05, "loss": 2.659, "step": 37221 }, { "epoch": 2.310633807188528, "grad_norm": 0.5200839723588102, "learning_rate": 1.524076526389216e-05, "loss": 2.7333, "step": 37222 }, { "epoch": 2.310695884288286, "grad_norm": 0.4802608815677606, "learning_rate": 1.5238169278443382e-05, "loss": 2.6181, "step": 37223 }, { "epoch": 2.310757961388044, "grad_norm": 0.5219060214753494, "learning_rate": 1.523557347435549e-05, "loss": 2.7207, "step": 37224 }, { "epoch": 2.310820038487802, "grad_norm": 0.43490030853649114, "learning_rate": 1.5232977851642077e-05, "loss": 2.7402, "step": 37225 }, { "epoch": 2.31088211558756, "grad_norm": 0.46679361892905313, "learning_rate": 1.523038241031665e-05, "loss": 2.6123, "step": 37226 }, { "epoch": 2.3109441926873178, "grad_norm": 0.5199374568447165, "learning_rate": 1.5227787150392764e-05, "loss": 2.7002, "step": 37227 }, { "epoch": 2.3110062697870757, "grad_norm": 0.4661541588180086, "learning_rate": 1.5225192071883953e-05, "loss": 2.7133, "step": 37228 }, { "epoch": 2.3110683468868336, "grad_norm": 0.5190508349334848, "learning_rate": 1.5222597174803738e-05, "loss": 2.5466, "step": 37229 }, { "epoch": 2.3111304239865915, "grad_norm": 0.4731230361827171, "learning_rate": 1.5220002459165694e-05, "loss": 2.7414, "step": 37230 }, { "epoch": 2.3111925010863494, "grad_norm": 0.4841937679753184, "learning_rate": 1.521740792498334e-05, "loss": 2.6725, "step": 37231 }, { "epoch": 2.311254578186107, "grad_norm": 0.49837883750561346, "learning_rate": 1.5214813572270214e-05, "loss": 2.6563, "step": 37232 }, { "epoch": 2.3113166552858653, "grad_norm": 0.4482828425169374, "learning_rate": 1.521221940103983e-05, "loss": 2.7268, "step": 37233 }, { "epoch": 2.3113787323856227, "grad_norm": 0.5050784235670914, "learning_rate": 1.5209625411305761e-05, "loss": 2.717, "step": 37234 }, { "epoch": 2.3114408094853807, "grad_norm": 0.477344122104531, "learning_rate": 1.5207031603081522e-05, "loss": 2.5834, "step": 37235 }, { "epoch": 2.3115028865851386, "grad_norm": 0.4421134617157036, "learning_rate": 1.5204437976380647e-05, "loss": 2.7685, "step": 37236 }, { "epoch": 2.3115649636848965, "grad_norm": 0.4779867246552037, "learning_rate": 1.5201844531216646e-05, "loss": 2.6802, "step": 37237 }, { "epoch": 2.3116270407846544, "grad_norm": 0.49536332853391835, "learning_rate": 1.5199251267603088e-05, "loss": 2.6526, "step": 37238 }, { "epoch": 2.3116891178844123, "grad_norm": 0.4419507407167427, "learning_rate": 1.5196658185553487e-05, "loss": 2.6416, "step": 37239 }, { "epoch": 2.3117511949841703, "grad_norm": 0.464986891305385, "learning_rate": 1.5194065285081366e-05, "loss": 2.7076, "step": 37240 }, { "epoch": 2.311813272083928, "grad_norm": 0.46460682577200674, "learning_rate": 1.5191472566200238e-05, "loss": 2.6094, "step": 37241 }, { "epoch": 2.311875349183686, "grad_norm": 0.470758110412081, "learning_rate": 1.518888002892367e-05, "loss": 2.8384, "step": 37242 }, { "epoch": 2.311937426283444, "grad_norm": 0.4757982128858978, "learning_rate": 1.5186287673265143e-05, "loss": 2.7164, "step": 37243 }, { "epoch": 2.311999503383202, "grad_norm": 0.4394323895118856, "learning_rate": 1.518369549923822e-05, "loss": 2.6428, "step": 37244 }, { "epoch": 2.31206158048296, "grad_norm": 0.4372194957681806, "learning_rate": 1.5181103506856414e-05, "loss": 2.6746, "step": 37245 }, { "epoch": 2.3121236575827178, "grad_norm": 0.4369776536675117, "learning_rate": 1.517851169613324e-05, "loss": 2.641, "step": 37246 }, { "epoch": 2.3121857346824757, "grad_norm": 0.4817353433132385, "learning_rate": 1.517592006708221e-05, "loss": 2.7147, "step": 37247 }, { "epoch": 2.3122478117822336, "grad_norm": 0.47038248624984985, "learning_rate": 1.5173328619716876e-05, "loss": 2.7178, "step": 37248 }, { "epoch": 2.3123098888819915, "grad_norm": 0.4530344576646093, "learning_rate": 1.517073735405074e-05, "loss": 2.6285, "step": 37249 }, { "epoch": 2.3123719659817494, "grad_norm": 0.4923138683310246, "learning_rate": 1.5168146270097322e-05, "loss": 2.7381, "step": 37250 }, { "epoch": 2.3124340430815074, "grad_norm": 0.4761167693073926, "learning_rate": 1.5165555367870127e-05, "loss": 2.6575, "step": 37251 }, { "epoch": 2.3124961201812653, "grad_norm": 0.4891876810028923, "learning_rate": 1.51629646473827e-05, "loss": 2.6494, "step": 37252 }, { "epoch": 2.312558197281023, "grad_norm": 0.5031117230197683, "learning_rate": 1.516037410864854e-05, "loss": 2.6307, "step": 37253 }, { "epoch": 2.312620274380781, "grad_norm": 0.4679974555125716, "learning_rate": 1.515778375168117e-05, "loss": 2.639, "step": 37254 }, { "epoch": 2.312682351480539, "grad_norm": 0.4864129448272794, "learning_rate": 1.5155193576494098e-05, "loss": 2.6343, "step": 37255 }, { "epoch": 2.312744428580297, "grad_norm": 0.4604629563095591, "learning_rate": 1.5152603583100828e-05, "loss": 2.6072, "step": 37256 }, { "epoch": 2.3128065056800544, "grad_norm": 0.49866259045558353, "learning_rate": 1.5150013771514898e-05, "loss": 2.6132, "step": 37257 }, { "epoch": 2.3128685827798128, "grad_norm": 0.46915578762075616, "learning_rate": 1.5147424141749806e-05, "loss": 2.6768, "step": 37258 }, { "epoch": 2.3129306598795702, "grad_norm": 0.49505197972178355, "learning_rate": 1.514483469381906e-05, "loss": 2.6984, "step": 37259 }, { "epoch": 2.312992736979328, "grad_norm": 0.49648671843713266, "learning_rate": 1.5142245427736163e-05, "loss": 2.6898, "step": 37260 }, { "epoch": 2.313054814079086, "grad_norm": 0.46921200237065364, "learning_rate": 1.5139656343514647e-05, "loss": 2.7681, "step": 37261 }, { "epoch": 2.313116891178844, "grad_norm": 0.500638613569346, "learning_rate": 1.5137067441168008e-05, "loss": 2.6519, "step": 37262 }, { "epoch": 2.313178968278602, "grad_norm": 0.4979886189217348, "learning_rate": 1.5134478720709749e-05, "loss": 2.7979, "step": 37263 }, { "epoch": 2.31324104537836, "grad_norm": 0.4942022292746449, "learning_rate": 1.5131890182153374e-05, "loss": 2.6214, "step": 37264 }, { "epoch": 2.3133031224781178, "grad_norm": 0.46068585525651745, "learning_rate": 1.5129301825512382e-05, "loss": 2.6943, "step": 37265 }, { "epoch": 2.3133651995778757, "grad_norm": 0.49845815394548537, "learning_rate": 1.5126713650800305e-05, "loss": 2.6815, "step": 37266 }, { "epoch": 2.3134272766776336, "grad_norm": 0.482837112194483, "learning_rate": 1.5124125658030625e-05, "loss": 2.7228, "step": 37267 }, { "epoch": 2.3134893537773915, "grad_norm": 0.4656686435957641, "learning_rate": 1.5121537847216849e-05, "loss": 2.7279, "step": 37268 }, { "epoch": 2.3135514308771494, "grad_norm": 0.49928015807328996, "learning_rate": 1.5118950218372463e-05, "loss": 2.6839, "step": 37269 }, { "epoch": 2.3136135079769073, "grad_norm": 0.46038686870218776, "learning_rate": 1.5116362771510994e-05, "loss": 2.746, "step": 37270 }, { "epoch": 2.3136755850766653, "grad_norm": 0.4720772751786724, "learning_rate": 1.5113775506645933e-05, "loss": 2.7081, "step": 37271 }, { "epoch": 2.313737662176423, "grad_norm": 0.47175804419297435, "learning_rate": 1.5111188423790772e-05, "loss": 2.6624, "step": 37272 }, { "epoch": 2.313799739276181, "grad_norm": 0.49070074687272786, "learning_rate": 1.5108601522958993e-05, "loss": 2.5796, "step": 37273 }, { "epoch": 2.313861816375939, "grad_norm": 0.47672342491295794, "learning_rate": 1.5106014804164114e-05, "loss": 2.6206, "step": 37274 }, { "epoch": 2.313923893475697, "grad_norm": 0.45857637817649044, "learning_rate": 1.510342826741964e-05, "loss": 2.6069, "step": 37275 }, { "epoch": 2.313985970575455, "grad_norm": 0.4954406915357113, "learning_rate": 1.510084191273905e-05, "loss": 2.6318, "step": 37276 }, { "epoch": 2.3140480476752128, "grad_norm": 0.4707432963217592, "learning_rate": 1.5098255740135842e-05, "loss": 2.6603, "step": 37277 }, { "epoch": 2.3141101247749707, "grad_norm": 0.49317419967052717, "learning_rate": 1.5095669749623508e-05, "loss": 2.7047, "step": 37278 }, { "epoch": 2.3141722018747286, "grad_norm": 0.5014527494909168, "learning_rate": 1.509308394121552e-05, "loss": 2.694, "step": 37279 }, { "epoch": 2.314234278974486, "grad_norm": 0.4934757720571332, "learning_rate": 1.5090498314925405e-05, "loss": 2.7032, "step": 37280 }, { "epoch": 2.3142963560742444, "grad_norm": 0.49508566069439847, "learning_rate": 1.5087912870766635e-05, "loss": 2.6391, "step": 37281 }, { "epoch": 2.314358433174002, "grad_norm": 0.4896258951178465, "learning_rate": 1.5085327608752698e-05, "loss": 2.5638, "step": 37282 }, { "epoch": 2.31442051027376, "grad_norm": 0.5178125965990841, "learning_rate": 1.5082742528897064e-05, "loss": 2.6784, "step": 37283 }, { "epoch": 2.3144825873735178, "grad_norm": 0.43518946399324987, "learning_rate": 1.5080157631213254e-05, "loss": 2.6929, "step": 37284 }, { "epoch": 2.3145446644732757, "grad_norm": 0.4641776080214564, "learning_rate": 1.5077572915714744e-05, "loss": 2.5803, "step": 37285 }, { "epoch": 2.3146067415730336, "grad_norm": 0.4907714971273457, "learning_rate": 1.5074988382415012e-05, "loss": 2.6815, "step": 37286 }, { "epoch": 2.3146688186727915, "grad_norm": 0.49831333886045676, "learning_rate": 1.5072404031327525e-05, "loss": 2.693, "step": 37287 }, { "epoch": 2.3147308957725494, "grad_norm": 0.4721082710784727, "learning_rate": 1.5069819862465806e-05, "loss": 2.6266, "step": 37288 }, { "epoch": 2.3147929728723073, "grad_norm": 0.48013689629640266, "learning_rate": 1.506723587584331e-05, "loss": 2.6732, "step": 37289 }, { "epoch": 2.3148550499720653, "grad_norm": 0.46213540707654144, "learning_rate": 1.5064652071473528e-05, "loss": 2.6267, "step": 37290 }, { "epoch": 2.314917127071823, "grad_norm": 0.5731241081838211, "learning_rate": 1.5062068449369937e-05, "loss": 2.7144, "step": 37291 }, { "epoch": 2.314979204171581, "grad_norm": 0.4765589510676082, "learning_rate": 1.5059485009546003e-05, "loss": 2.7158, "step": 37292 }, { "epoch": 2.315041281271339, "grad_norm": 0.5027130845666502, "learning_rate": 1.5056901752015235e-05, "loss": 2.6927, "step": 37293 }, { "epoch": 2.315103358371097, "grad_norm": 0.4928846719805274, "learning_rate": 1.5054318676791085e-05, "loss": 2.6408, "step": 37294 }, { "epoch": 2.315165435470855, "grad_norm": 0.4910795669376707, "learning_rate": 1.5051735783887044e-05, "loss": 2.7253, "step": 37295 }, { "epoch": 2.3152275125706128, "grad_norm": 0.4859072231170372, "learning_rate": 1.5049153073316563e-05, "loss": 2.769, "step": 37296 }, { "epoch": 2.3152895896703707, "grad_norm": 0.4401047516617658, "learning_rate": 1.5046570545093153e-05, "loss": 2.7339, "step": 37297 }, { "epoch": 2.3153516667701286, "grad_norm": 0.48851526374748633, "learning_rate": 1.504398819923027e-05, "loss": 2.7168, "step": 37298 }, { "epoch": 2.3154137438698865, "grad_norm": 0.4637027007142576, "learning_rate": 1.5041406035741385e-05, "loss": 2.6375, "step": 37299 }, { "epoch": 2.3154758209696444, "grad_norm": 0.4449131161813111, "learning_rate": 1.5038824054639972e-05, "loss": 2.6741, "step": 37300 }, { "epoch": 2.3155378980694024, "grad_norm": 0.45130792853577584, "learning_rate": 1.5036242255939487e-05, "loss": 2.6324, "step": 37301 }, { "epoch": 2.3155999751691603, "grad_norm": 0.44252288150356767, "learning_rate": 1.5033660639653424e-05, "loss": 2.7355, "step": 37302 }, { "epoch": 2.3156620522689177, "grad_norm": 0.4666928131436598, "learning_rate": 1.5031079205795245e-05, "loss": 2.7884, "step": 37303 }, { "epoch": 2.315724129368676, "grad_norm": 0.4203871522429293, "learning_rate": 1.5028497954378417e-05, "loss": 2.5989, "step": 37304 }, { "epoch": 2.3157862064684336, "grad_norm": 0.4310294437319436, "learning_rate": 1.5025916885416386e-05, "loss": 2.683, "step": 37305 }, { "epoch": 2.3158482835681915, "grad_norm": 0.4484999503536473, "learning_rate": 1.5023335998922655e-05, "loss": 2.7161, "step": 37306 }, { "epoch": 2.3159103606679494, "grad_norm": 0.47523868110457995, "learning_rate": 1.5020755294910649e-05, "loss": 2.6444, "step": 37307 }, { "epoch": 2.3159724377677073, "grad_norm": 0.4177457624616583, "learning_rate": 1.5018174773393878e-05, "loss": 2.6713, "step": 37308 }, { "epoch": 2.3160345148674653, "grad_norm": 0.45751663255943736, "learning_rate": 1.5015594434385778e-05, "loss": 2.7276, "step": 37309 }, { "epoch": 2.316096591967223, "grad_norm": 0.4809898094200122, "learning_rate": 1.50130142778998e-05, "loss": 2.7752, "step": 37310 }, { "epoch": 2.316158669066981, "grad_norm": 0.45050738062830026, "learning_rate": 1.5010434303949434e-05, "loss": 2.7425, "step": 37311 }, { "epoch": 2.316220746166739, "grad_norm": 0.4512062720195525, "learning_rate": 1.5007854512548125e-05, "loss": 2.7695, "step": 37312 }, { "epoch": 2.316282823266497, "grad_norm": 0.45937300117213514, "learning_rate": 1.5005274903709337e-05, "loss": 2.6796, "step": 37313 }, { "epoch": 2.316344900366255, "grad_norm": 0.457142270289517, "learning_rate": 1.5002695477446522e-05, "loss": 2.6138, "step": 37314 }, { "epoch": 2.3164069774660128, "grad_norm": 0.44449299143188736, "learning_rate": 1.5000116233773126e-05, "loss": 2.6559, "step": 37315 }, { "epoch": 2.3164690545657707, "grad_norm": 0.4843895202429208, "learning_rate": 1.4997537172702636e-05, "loss": 2.6938, "step": 37316 }, { "epoch": 2.3165311316655286, "grad_norm": 0.45614900777646766, "learning_rate": 1.4994958294248495e-05, "loss": 2.6959, "step": 37317 }, { "epoch": 2.3165932087652865, "grad_norm": 0.4503027566389132, "learning_rate": 1.499237959842415e-05, "loss": 2.718, "step": 37318 }, { "epoch": 2.3166552858650444, "grad_norm": 0.4948944704436743, "learning_rate": 1.4989801085243044e-05, "loss": 2.7785, "step": 37319 }, { "epoch": 2.3167173629648024, "grad_norm": 0.47816022793011614, "learning_rate": 1.4987222754718661e-05, "loss": 2.6625, "step": 37320 }, { "epoch": 2.3167794400645603, "grad_norm": 0.4403886120371064, "learning_rate": 1.4984644606864433e-05, "loss": 2.6203, "step": 37321 }, { "epoch": 2.316841517164318, "grad_norm": 0.5144207503227094, "learning_rate": 1.498206664169382e-05, "loss": 2.7068, "step": 37322 }, { "epoch": 2.316903594264076, "grad_norm": 0.4498002310899418, "learning_rate": 1.4979488859220247e-05, "loss": 2.6471, "step": 37323 }, { "epoch": 2.316965671363834, "grad_norm": 0.4663555341332298, "learning_rate": 1.4976911259457206e-05, "loss": 2.6592, "step": 37324 }, { "epoch": 2.317027748463592, "grad_norm": 0.45437392403896293, "learning_rate": 1.4974333842418109e-05, "loss": 2.6578, "step": 37325 }, { "epoch": 2.31708982556335, "grad_norm": 0.433115841849911, "learning_rate": 1.4971756608116421e-05, "loss": 2.6494, "step": 37326 }, { "epoch": 2.3171519026631078, "grad_norm": 0.5009563062710862, "learning_rate": 1.4969179556565583e-05, "loss": 2.7085, "step": 37327 }, { "epoch": 2.3172139797628653, "grad_norm": 0.44665223751615823, "learning_rate": 1.4966602687779024e-05, "loss": 2.6167, "step": 37328 }, { "epoch": 2.3172760568626236, "grad_norm": 0.4515971866735858, "learning_rate": 1.4964026001770215e-05, "loss": 2.6584, "step": 37329 }, { "epoch": 2.317338133962381, "grad_norm": 0.5108088783500767, "learning_rate": 1.4961449498552593e-05, "loss": 2.7144, "step": 37330 }, { "epoch": 2.317400211062139, "grad_norm": 0.48559275722616985, "learning_rate": 1.4958873178139594e-05, "loss": 2.6768, "step": 37331 }, { "epoch": 2.317462288161897, "grad_norm": 0.49735120170300817, "learning_rate": 1.4956297040544636e-05, "loss": 2.6407, "step": 37332 }, { "epoch": 2.317524365261655, "grad_norm": 0.4929982425917588, "learning_rate": 1.4953721085781208e-05, "loss": 2.6358, "step": 37333 }, { "epoch": 2.3175864423614128, "grad_norm": 0.47875719784636966, "learning_rate": 1.4951145313862724e-05, "loss": 2.6733, "step": 37334 }, { "epoch": 2.3176485194611707, "grad_norm": 0.5089724570787136, "learning_rate": 1.494856972480262e-05, "loss": 2.6416, "step": 37335 }, { "epoch": 2.3177105965609286, "grad_norm": 0.5277878763113896, "learning_rate": 1.4945994318614326e-05, "loss": 2.6757, "step": 37336 }, { "epoch": 2.3177726736606865, "grad_norm": 0.46358561753923944, "learning_rate": 1.49434190953113e-05, "loss": 2.709, "step": 37337 }, { "epoch": 2.3178347507604444, "grad_norm": 0.4646935420615141, "learning_rate": 1.4940844054906966e-05, "loss": 2.6593, "step": 37338 }, { "epoch": 2.3178968278602023, "grad_norm": 0.5206762045034725, "learning_rate": 1.4938269197414751e-05, "loss": 2.6839, "step": 37339 }, { "epoch": 2.3179589049599603, "grad_norm": 0.46445474455914404, "learning_rate": 1.4935694522848104e-05, "loss": 2.6399, "step": 37340 }, { "epoch": 2.318020982059718, "grad_norm": 0.48348911860548216, "learning_rate": 1.4933120031220454e-05, "loss": 2.6753, "step": 37341 }, { "epoch": 2.318083059159476, "grad_norm": 0.539815791022977, "learning_rate": 1.493054572254522e-05, "loss": 2.6424, "step": 37342 }, { "epoch": 2.318145136259234, "grad_norm": 0.4596403263971705, "learning_rate": 1.492797159683585e-05, "loss": 2.6814, "step": 37343 }, { "epoch": 2.318207213358992, "grad_norm": 0.49703379798579034, "learning_rate": 1.4925397654105772e-05, "loss": 2.725, "step": 37344 }, { "epoch": 2.31826929045875, "grad_norm": 0.5287167512526826, "learning_rate": 1.4922823894368404e-05, "loss": 2.6861, "step": 37345 }, { "epoch": 2.3183313675585078, "grad_norm": 0.5123687866924921, "learning_rate": 1.4920250317637169e-05, "loss": 2.6945, "step": 37346 }, { "epoch": 2.3183934446582657, "grad_norm": 0.5287455259634344, "learning_rate": 1.4917676923925516e-05, "loss": 2.7296, "step": 37347 }, { "epoch": 2.3184555217580236, "grad_norm": 0.49178468781536383, "learning_rate": 1.4915103713246858e-05, "loss": 2.6966, "step": 37348 }, { "epoch": 2.3185175988577815, "grad_norm": 0.47574836730075504, "learning_rate": 1.4912530685614623e-05, "loss": 2.5848, "step": 37349 }, { "epoch": 2.3185796759575394, "grad_norm": 0.5449603214168832, "learning_rate": 1.4909957841042232e-05, "loss": 2.6645, "step": 37350 }, { "epoch": 2.318641753057297, "grad_norm": 0.4934989773972573, "learning_rate": 1.4907385179543093e-05, "loss": 2.7297, "step": 37351 }, { "epoch": 2.3187038301570553, "grad_norm": 0.4771850459467271, "learning_rate": 1.4904812701130661e-05, "loss": 2.6127, "step": 37352 }, { "epoch": 2.3187659072568128, "grad_norm": 0.5233529720094087, "learning_rate": 1.4902240405818346e-05, "loss": 2.6508, "step": 37353 }, { "epoch": 2.3188279843565707, "grad_norm": 0.4594412133122972, "learning_rate": 1.4899668293619556e-05, "loss": 2.6583, "step": 37354 }, { "epoch": 2.3188900614563286, "grad_norm": 0.4946217512872884, "learning_rate": 1.4897096364547703e-05, "loss": 2.6637, "step": 37355 }, { "epoch": 2.3189521385560865, "grad_norm": 0.5066601516384235, "learning_rate": 1.489452461861624e-05, "loss": 2.6505, "step": 37356 }, { "epoch": 2.3190142156558444, "grad_norm": 0.47413608515538574, "learning_rate": 1.4891953055838559e-05, "loss": 2.6391, "step": 37357 }, { "epoch": 2.3190762927556023, "grad_norm": 0.5286778845881662, "learning_rate": 1.488938167622808e-05, "loss": 2.7071, "step": 37358 }, { "epoch": 2.3191383698553603, "grad_norm": 0.486286192162422, "learning_rate": 1.4886810479798213e-05, "loss": 2.707, "step": 37359 }, { "epoch": 2.319200446955118, "grad_norm": 0.4928170371224708, "learning_rate": 1.4884239466562389e-05, "loss": 2.6636, "step": 37360 }, { "epoch": 2.319262524054876, "grad_norm": 0.5112255719231811, "learning_rate": 1.4881668636534012e-05, "loss": 2.7475, "step": 37361 }, { "epoch": 2.319324601154634, "grad_norm": 0.477955673502796, "learning_rate": 1.4879097989726497e-05, "loss": 2.6469, "step": 37362 }, { "epoch": 2.319386678254392, "grad_norm": 0.47026728796120676, "learning_rate": 1.4876527526153256e-05, "loss": 2.6716, "step": 37363 }, { "epoch": 2.31944875535415, "grad_norm": 0.4911548134989737, "learning_rate": 1.487395724582768e-05, "loss": 2.6087, "step": 37364 }, { "epoch": 2.3195108324539078, "grad_norm": 0.42681414758294706, "learning_rate": 1.4871387148763211e-05, "loss": 2.7245, "step": 37365 }, { "epoch": 2.3195729095536657, "grad_norm": 0.4873220100458272, "learning_rate": 1.4868817234973242e-05, "loss": 2.7354, "step": 37366 }, { "epoch": 2.3196349866534236, "grad_norm": 0.4803644082507053, "learning_rate": 1.4866247504471182e-05, "loss": 2.6349, "step": 37367 }, { "epoch": 2.3196970637531815, "grad_norm": 0.4637100756751834, "learning_rate": 1.4863677957270423e-05, "loss": 2.6751, "step": 37368 }, { "epoch": 2.3197591408529394, "grad_norm": 0.5208898291146492, "learning_rate": 1.4861108593384404e-05, "loss": 2.7009, "step": 37369 }, { "epoch": 2.3198212179526974, "grad_norm": 0.4983176000475968, "learning_rate": 1.4858539412826511e-05, "loss": 2.7044, "step": 37370 }, { "epoch": 2.3198832950524553, "grad_norm": 0.5166938281098001, "learning_rate": 1.4855970415610144e-05, "loss": 2.7528, "step": 37371 }, { "epoch": 2.319945372152213, "grad_norm": 0.46184358603535675, "learning_rate": 1.4853401601748701e-05, "loss": 2.7266, "step": 37372 }, { "epoch": 2.320007449251971, "grad_norm": 0.478867117572695, "learning_rate": 1.485083297125559e-05, "loss": 2.7167, "step": 37373 }, { "epoch": 2.320069526351729, "grad_norm": 0.496910399249596, "learning_rate": 1.4848264524144235e-05, "loss": 2.6794, "step": 37374 }, { "epoch": 2.320131603451487, "grad_norm": 0.4717832810903414, "learning_rate": 1.484569626042801e-05, "loss": 2.7618, "step": 37375 }, { "epoch": 2.3201936805512444, "grad_norm": 0.44829997768487134, "learning_rate": 1.484312818012033e-05, "loss": 2.7025, "step": 37376 }, { "epoch": 2.320255757651003, "grad_norm": 0.46013241302977054, "learning_rate": 1.4840560283234584e-05, "loss": 2.6537, "step": 37377 }, { "epoch": 2.3203178347507603, "grad_norm": 0.5167526572204835, "learning_rate": 1.4837992569784148e-05, "loss": 2.6958, "step": 37378 }, { "epoch": 2.320379911850518, "grad_norm": 0.4281678463734112, "learning_rate": 1.4835425039782463e-05, "loss": 2.6941, "step": 37379 }, { "epoch": 2.320441988950276, "grad_norm": 0.4075526663031588, "learning_rate": 1.4832857693242902e-05, "loss": 2.6233, "step": 37380 }, { "epoch": 2.320504066050034, "grad_norm": 0.4958051457794357, "learning_rate": 1.4830290530178854e-05, "loss": 2.5964, "step": 37381 }, { "epoch": 2.320566143149792, "grad_norm": 0.5031853775853371, "learning_rate": 1.4827723550603706e-05, "loss": 2.6949, "step": 37382 }, { "epoch": 2.32062822024955, "grad_norm": 0.462546707707396, "learning_rate": 1.4825156754530877e-05, "loss": 2.7026, "step": 37383 }, { "epoch": 2.3206902973493078, "grad_norm": 0.4642334305359245, "learning_rate": 1.4822590141973747e-05, "loss": 2.6198, "step": 37384 }, { "epoch": 2.3207523744490657, "grad_norm": 0.46732952740351147, "learning_rate": 1.4820023712945702e-05, "loss": 2.7065, "step": 37385 }, { "epoch": 2.3208144515488236, "grad_norm": 0.46149956653157187, "learning_rate": 1.4817457467460122e-05, "loss": 2.748, "step": 37386 }, { "epoch": 2.3208765286485815, "grad_norm": 0.4602926722371944, "learning_rate": 1.4814891405530418e-05, "loss": 2.7531, "step": 37387 }, { "epoch": 2.3209386057483394, "grad_norm": 0.47562702792583744, "learning_rate": 1.4812325527169968e-05, "loss": 2.5982, "step": 37388 }, { "epoch": 2.3210006828480974, "grad_norm": 0.49842607610639633, "learning_rate": 1.4809759832392157e-05, "loss": 2.729, "step": 37389 }, { "epoch": 2.3210627599478553, "grad_norm": 0.45259383738166015, "learning_rate": 1.4807194321210372e-05, "loss": 2.6391, "step": 37390 }, { "epoch": 2.321124837047613, "grad_norm": 0.4376038530247663, "learning_rate": 1.4804628993637981e-05, "loss": 2.6631, "step": 37391 }, { "epoch": 2.321186914147371, "grad_norm": 0.43700810436267196, "learning_rate": 1.4802063849688403e-05, "loss": 2.72, "step": 37392 }, { "epoch": 2.321248991247129, "grad_norm": 0.45251578450992913, "learning_rate": 1.4799498889375002e-05, "loss": 2.6957, "step": 37393 }, { "epoch": 2.321311068346887, "grad_norm": 0.43596268379454084, "learning_rate": 1.4796934112711158e-05, "loss": 2.6971, "step": 37394 }, { "epoch": 2.321373145446645, "grad_norm": 0.45869779365092916, "learning_rate": 1.4794369519710238e-05, "loss": 2.6991, "step": 37395 }, { "epoch": 2.3214352225464028, "grad_norm": 0.4537742173218872, "learning_rate": 1.4791805110385659e-05, "loss": 2.7513, "step": 37396 }, { "epoch": 2.3214972996461607, "grad_norm": 0.42045226201929986, "learning_rate": 1.4789240884750777e-05, "loss": 2.6823, "step": 37397 }, { "epoch": 2.3215593767459186, "grad_norm": 0.4577300768922429, "learning_rate": 1.4786676842818975e-05, "loss": 2.6417, "step": 37398 }, { "epoch": 2.321621453845676, "grad_norm": 0.47544684912178786, "learning_rate": 1.4784112984603626e-05, "loss": 2.624, "step": 37399 }, { "epoch": 2.3216835309454344, "grad_norm": 0.46560987281548677, "learning_rate": 1.4781549310118093e-05, "loss": 2.7611, "step": 37400 }, { "epoch": 2.321745608045192, "grad_norm": 0.48974338580188226, "learning_rate": 1.4778985819375784e-05, "loss": 2.7135, "step": 37401 }, { "epoch": 2.32180768514495, "grad_norm": 0.4116359359581294, "learning_rate": 1.4776422512390054e-05, "loss": 2.5705, "step": 37402 }, { "epoch": 2.3218697622447078, "grad_norm": 0.4683868591033003, "learning_rate": 1.4773859389174282e-05, "loss": 2.6244, "step": 37403 }, { "epoch": 2.3219318393444657, "grad_norm": 0.4564490889857011, "learning_rate": 1.4771296449741818e-05, "loss": 2.6562, "step": 37404 }, { "epoch": 2.3219939164442236, "grad_norm": 0.44799556887406466, "learning_rate": 1.4768733694106058e-05, "loss": 2.6701, "step": 37405 }, { "epoch": 2.3220559935439815, "grad_norm": 0.4372996995717257, "learning_rate": 1.4766171122280375e-05, "loss": 2.6477, "step": 37406 }, { "epoch": 2.3221180706437394, "grad_norm": 0.516675485899729, "learning_rate": 1.4763608734278134e-05, "loss": 2.7222, "step": 37407 }, { "epoch": 2.3221801477434973, "grad_norm": 0.4556478282458111, "learning_rate": 1.47610465301127e-05, "loss": 2.6615, "step": 37408 }, { "epoch": 2.3222422248432553, "grad_norm": 0.4473953891732702, "learning_rate": 1.4758484509797427e-05, "loss": 2.7712, "step": 37409 }, { "epoch": 2.322304301943013, "grad_norm": 0.4691881833836904, "learning_rate": 1.4755922673345706e-05, "loss": 2.727, "step": 37410 }, { "epoch": 2.322366379042771, "grad_norm": 0.4533316206928935, "learning_rate": 1.4753361020770901e-05, "loss": 2.667, "step": 37411 }, { "epoch": 2.322428456142529, "grad_norm": 0.49407317715230253, "learning_rate": 1.4750799552086358e-05, "loss": 2.5857, "step": 37412 }, { "epoch": 2.322490533242287, "grad_norm": 0.4828771689847916, "learning_rate": 1.474823826730546e-05, "loss": 2.6285, "step": 37413 }, { "epoch": 2.322552610342045, "grad_norm": 0.4614670346503888, "learning_rate": 1.4745677166441541e-05, "loss": 2.6903, "step": 37414 }, { "epoch": 2.3226146874418028, "grad_norm": 0.4789797653352237, "learning_rate": 1.4743116249507993e-05, "loss": 2.6606, "step": 37415 }, { "epoch": 2.3226767645415607, "grad_norm": 0.46897226468269776, "learning_rate": 1.4740555516518173e-05, "loss": 2.6413, "step": 37416 }, { "epoch": 2.3227388416413186, "grad_norm": 0.4452283679391592, "learning_rate": 1.473799496748543e-05, "loss": 2.6596, "step": 37417 }, { "epoch": 2.3228009187410765, "grad_norm": 0.4355258679990427, "learning_rate": 1.4735434602423114e-05, "loss": 2.7216, "step": 37418 }, { "epoch": 2.3228629958408344, "grad_norm": 0.4514189699941773, "learning_rate": 1.4732874421344616e-05, "loss": 2.7493, "step": 37419 }, { "epoch": 2.3229250729405924, "grad_norm": 0.47607194948878234, "learning_rate": 1.4730314424263269e-05, "loss": 2.6222, "step": 37420 }, { "epoch": 2.3229871500403503, "grad_norm": 0.43386306646712813, "learning_rate": 1.4727754611192434e-05, "loss": 2.6757, "step": 37421 }, { "epoch": 2.323049227140108, "grad_norm": 0.43913751076810703, "learning_rate": 1.4725194982145452e-05, "loss": 2.6958, "step": 37422 }, { "epoch": 2.323111304239866, "grad_norm": 0.4807544417124026, "learning_rate": 1.4722635537135704e-05, "loss": 2.7091, "step": 37423 }, { "epoch": 2.3231733813396236, "grad_norm": 0.42789975109052986, "learning_rate": 1.4720076276176532e-05, "loss": 2.6589, "step": 37424 }, { "epoch": 2.323235458439382, "grad_norm": 0.4552106214185019, "learning_rate": 1.4717517199281288e-05, "loss": 2.6871, "step": 37425 }, { "epoch": 2.3232975355391394, "grad_norm": 0.45524245475432173, "learning_rate": 1.4714958306463318e-05, "loss": 2.771, "step": 37426 }, { "epoch": 2.3233596126388973, "grad_norm": 0.43729555939130166, "learning_rate": 1.4712399597735966e-05, "loss": 2.7779, "step": 37427 }, { "epoch": 2.3234216897386553, "grad_norm": 0.5029411240545848, "learning_rate": 1.47098410731126e-05, "loss": 2.6912, "step": 37428 }, { "epoch": 2.323483766838413, "grad_norm": 0.45567202800225853, "learning_rate": 1.4707282732606565e-05, "loss": 2.7218, "step": 37429 }, { "epoch": 2.323545843938171, "grad_norm": 0.48507469155141986, "learning_rate": 1.4704724576231205e-05, "loss": 2.6265, "step": 37430 }, { "epoch": 2.323607921037929, "grad_norm": 0.481756175737532, "learning_rate": 1.4702166603999845e-05, "loss": 2.6655, "step": 37431 }, { "epoch": 2.323669998137687, "grad_norm": 0.48624724362825267, "learning_rate": 1.469960881592587e-05, "loss": 2.7537, "step": 37432 }, { "epoch": 2.323732075237445, "grad_norm": 0.5197530287433725, "learning_rate": 1.4697051212022606e-05, "loss": 2.6766, "step": 37433 }, { "epoch": 2.3237941523372028, "grad_norm": 0.4926335569043634, "learning_rate": 1.4694493792303393e-05, "loss": 2.7315, "step": 37434 }, { "epoch": 2.3238562294369607, "grad_norm": 0.48432476029265825, "learning_rate": 1.4691936556781578e-05, "loss": 2.6323, "step": 37435 }, { "epoch": 2.3239183065367186, "grad_norm": 0.4983106625172722, "learning_rate": 1.468937950547048e-05, "loss": 2.6758, "step": 37436 }, { "epoch": 2.3239803836364765, "grad_norm": 0.46502611445858444, "learning_rate": 1.4686822638383486e-05, "loss": 2.7208, "step": 37437 }, { "epoch": 2.3240424607362344, "grad_norm": 0.49320939865968355, "learning_rate": 1.4684265955533889e-05, "loss": 2.6756, "step": 37438 }, { "epoch": 2.3241045378359924, "grad_norm": 0.5101870844403656, "learning_rate": 1.4681709456935066e-05, "loss": 2.6386, "step": 37439 }, { "epoch": 2.3241666149357503, "grad_norm": 0.49411389124026245, "learning_rate": 1.4679153142600338e-05, "loss": 2.6548, "step": 37440 }, { "epoch": 2.324228692035508, "grad_norm": 0.5345802827462114, "learning_rate": 1.4676597012543031e-05, "loss": 2.7948, "step": 37441 }, { "epoch": 2.324290769135266, "grad_norm": 0.4987592651108934, "learning_rate": 1.4674041066776506e-05, "loss": 2.701, "step": 37442 }, { "epoch": 2.324352846235024, "grad_norm": 0.5146548652609451, "learning_rate": 1.4671485305314087e-05, "loss": 2.7328, "step": 37443 }, { "epoch": 2.324414923334782, "grad_norm": 0.4988948111293528, "learning_rate": 1.46689297281691e-05, "loss": 2.7002, "step": 37444 }, { "epoch": 2.32447700043454, "grad_norm": 0.4651369084036066, "learning_rate": 1.4666374335354872e-05, "loss": 2.6789, "step": 37445 }, { "epoch": 2.324539077534298, "grad_norm": 0.50234984517293, "learning_rate": 1.4663819126884758e-05, "loss": 2.6963, "step": 37446 }, { "epoch": 2.3246011546340553, "grad_norm": 0.521874785280469, "learning_rate": 1.4661264102772083e-05, "loss": 2.7632, "step": 37447 }, { "epoch": 2.3246632317338136, "grad_norm": 0.44533195787717555, "learning_rate": 1.4658709263030168e-05, "loss": 2.6705, "step": 37448 }, { "epoch": 2.324725308833571, "grad_norm": 0.543391879307255, "learning_rate": 1.4656154607672346e-05, "loss": 2.6832, "step": 37449 }, { "epoch": 2.324787385933329, "grad_norm": 0.4493878253703531, "learning_rate": 1.465360013671193e-05, "loss": 2.6546, "step": 37450 }, { "epoch": 2.324849463033087, "grad_norm": 0.5086018516258559, "learning_rate": 1.4651045850162282e-05, "loss": 2.588, "step": 37451 }, { "epoch": 2.324911540132845, "grad_norm": 0.49790223957319646, "learning_rate": 1.4648491748036708e-05, "loss": 2.6887, "step": 37452 }, { "epoch": 2.3249736172326028, "grad_norm": 0.46764387912874134, "learning_rate": 1.4645937830348532e-05, "loss": 2.7402, "step": 37453 }, { "epoch": 2.3250356943323607, "grad_norm": 0.4466044273741686, "learning_rate": 1.4643384097111068e-05, "loss": 2.6465, "step": 37454 }, { "epoch": 2.3250977714321186, "grad_norm": 0.5061473247817255, "learning_rate": 1.4640830548337669e-05, "loss": 2.7591, "step": 37455 }, { "epoch": 2.3251598485318765, "grad_norm": 0.48899820715628234, "learning_rate": 1.4638277184041637e-05, "loss": 2.5755, "step": 37456 }, { "epoch": 2.3252219256316344, "grad_norm": 0.4447392572024339, "learning_rate": 1.4635724004236295e-05, "loss": 2.6569, "step": 37457 }, { "epoch": 2.3252840027313924, "grad_norm": 0.45124423284887205, "learning_rate": 1.4633171008934954e-05, "loss": 2.6543, "step": 37458 }, { "epoch": 2.3253460798311503, "grad_norm": 0.5465907584334176, "learning_rate": 1.4630618198150964e-05, "loss": 2.6876, "step": 37459 }, { "epoch": 2.325408156930908, "grad_norm": 0.45313483260437387, "learning_rate": 1.462806557189762e-05, "loss": 2.6122, "step": 37460 }, { "epoch": 2.325470234030666, "grad_norm": 0.48591495105887805, "learning_rate": 1.4625513130188245e-05, "loss": 2.6732, "step": 37461 }, { "epoch": 2.325532311130424, "grad_norm": 0.5174403169096587, "learning_rate": 1.4622960873036162e-05, "loss": 2.6493, "step": 37462 }, { "epoch": 2.325594388230182, "grad_norm": 0.5228917137464094, "learning_rate": 1.4620408800454655e-05, "loss": 2.6639, "step": 37463 }, { "epoch": 2.32565646532994, "grad_norm": 0.49072009033746916, "learning_rate": 1.4617856912457083e-05, "loss": 2.6332, "step": 37464 }, { "epoch": 2.3257185424296978, "grad_norm": 0.5382920993547581, "learning_rate": 1.4615305209056745e-05, "loss": 2.655, "step": 37465 }, { "epoch": 2.3257806195294557, "grad_norm": 0.4593805916115016, "learning_rate": 1.4612753690266945e-05, "loss": 2.6935, "step": 37466 }, { "epoch": 2.3258426966292136, "grad_norm": 0.4836017835994357, "learning_rate": 1.4610202356100987e-05, "loss": 2.725, "step": 37467 }, { "epoch": 2.3259047737289715, "grad_norm": 0.47543841821534977, "learning_rate": 1.460765120657221e-05, "loss": 2.6269, "step": 37468 }, { "epoch": 2.3259668508287294, "grad_norm": 0.4786369792657549, "learning_rate": 1.460510024169391e-05, "loss": 2.6826, "step": 37469 }, { "epoch": 2.3260289279284874, "grad_norm": 0.5279681712133572, "learning_rate": 1.4602549461479376e-05, "loss": 2.608, "step": 37470 }, { "epoch": 2.3260910050282453, "grad_norm": 0.5510426705777326, "learning_rate": 1.459999886594195e-05, "loss": 2.6475, "step": 37471 }, { "epoch": 2.3261530821280028, "grad_norm": 0.4274478806407763, "learning_rate": 1.4597448455094904e-05, "loss": 2.636, "step": 37472 }, { "epoch": 2.326215159227761, "grad_norm": 0.506359810163696, "learning_rate": 1.4594898228951587e-05, "loss": 2.7171, "step": 37473 }, { "epoch": 2.3262772363275186, "grad_norm": 0.5150961331112928, "learning_rate": 1.4592348187525274e-05, "loss": 2.7256, "step": 37474 }, { "epoch": 2.3263393134272765, "grad_norm": 0.44678693250361756, "learning_rate": 1.458979833082928e-05, "loss": 2.6805, "step": 37475 }, { "epoch": 2.3264013905270344, "grad_norm": 0.4612234812507476, "learning_rate": 1.4587248658876907e-05, "loss": 2.7164, "step": 37476 }, { "epoch": 2.3264634676267923, "grad_norm": 0.4451306791717224, "learning_rate": 1.4584699171681432e-05, "loss": 2.6853, "step": 37477 }, { "epoch": 2.3265255447265503, "grad_norm": 0.44017958888265435, "learning_rate": 1.4582149869256194e-05, "loss": 2.6885, "step": 37478 }, { "epoch": 2.326587621826308, "grad_norm": 0.48091748070150825, "learning_rate": 1.4579600751614486e-05, "loss": 2.699, "step": 37479 }, { "epoch": 2.326649698926066, "grad_norm": 0.498334345412529, "learning_rate": 1.4577051818769593e-05, "loss": 2.6907, "step": 37480 }, { "epoch": 2.326711776025824, "grad_norm": 0.4242819891734448, "learning_rate": 1.4574503070734807e-05, "loss": 2.6272, "step": 37481 }, { "epoch": 2.326773853125582, "grad_norm": 0.43238698720402857, "learning_rate": 1.4571954507523456e-05, "loss": 2.6583, "step": 37482 }, { "epoch": 2.32683593022534, "grad_norm": 0.46619966043450106, "learning_rate": 1.4569406129148816e-05, "loss": 2.6858, "step": 37483 }, { "epoch": 2.3268980073250978, "grad_norm": 0.4850089016815175, "learning_rate": 1.4566857935624184e-05, "loss": 2.7221, "step": 37484 }, { "epoch": 2.3269600844248557, "grad_norm": 0.4797968624836883, "learning_rate": 1.456430992696286e-05, "loss": 2.7494, "step": 37485 }, { "epoch": 2.3270221615246136, "grad_norm": 0.4420374855095222, "learning_rate": 1.4561762103178111e-05, "loss": 2.6821, "step": 37486 }, { "epoch": 2.3270842386243715, "grad_norm": 0.49278652734716843, "learning_rate": 1.4559214464283278e-05, "loss": 2.6295, "step": 37487 }, { "epoch": 2.3271463157241294, "grad_norm": 0.5014171484799654, "learning_rate": 1.4556667010291618e-05, "loss": 2.665, "step": 37488 }, { "epoch": 2.3272083928238874, "grad_norm": 0.5025757664643374, "learning_rate": 1.4554119741216432e-05, "loss": 2.6799, "step": 37489 }, { "epoch": 2.3272704699236453, "grad_norm": 0.4532264727394997, "learning_rate": 1.4551572657070994e-05, "loss": 2.7604, "step": 37490 }, { "epoch": 2.327332547023403, "grad_norm": 0.48057227841382383, "learning_rate": 1.4549025757868622e-05, "loss": 2.6192, "step": 37491 }, { "epoch": 2.327394624123161, "grad_norm": 0.5336995032587111, "learning_rate": 1.4546479043622591e-05, "loss": 2.658, "step": 37492 }, { "epoch": 2.327456701222919, "grad_norm": 0.4868510701685569, "learning_rate": 1.4543932514346182e-05, "loss": 2.664, "step": 37493 }, { "epoch": 2.327518778322677, "grad_norm": 0.4707301911609315, "learning_rate": 1.4541386170052668e-05, "loss": 2.7578, "step": 37494 }, { "epoch": 2.3275808554224344, "grad_norm": 0.49849626798581503, "learning_rate": 1.4538840010755367e-05, "loss": 2.6455, "step": 37495 }, { "epoch": 2.327642932522193, "grad_norm": 0.47002691552016496, "learning_rate": 1.4536294036467545e-05, "loss": 2.7175, "step": 37496 }, { "epoch": 2.3277050096219503, "grad_norm": 0.484139539031971, "learning_rate": 1.4533748247202489e-05, "loss": 2.66, "step": 37497 }, { "epoch": 2.327767086721708, "grad_norm": 0.47687337460122137, "learning_rate": 1.4531202642973479e-05, "loss": 2.7239, "step": 37498 }, { "epoch": 2.327829163821466, "grad_norm": 0.46849760572072047, "learning_rate": 1.452865722379378e-05, "loss": 2.6386, "step": 37499 }, { "epoch": 2.327891240921224, "grad_norm": 0.42785811702820703, "learning_rate": 1.4526111989676699e-05, "loss": 2.7388, "step": 37500 }, { "epoch": 2.327953318020982, "grad_norm": 0.4971232058219579, "learning_rate": 1.4523566940635502e-05, "loss": 2.6365, "step": 37501 }, { "epoch": 2.32801539512074, "grad_norm": 0.4225541582044456, "learning_rate": 1.4521022076683472e-05, "loss": 2.7221, "step": 37502 }, { "epoch": 2.3280774722204978, "grad_norm": 0.43017027350200315, "learning_rate": 1.4518477397833868e-05, "loss": 2.6413, "step": 37503 }, { "epoch": 2.3281395493202557, "grad_norm": 0.4640836351250624, "learning_rate": 1.4515932904099976e-05, "loss": 2.6551, "step": 37504 }, { "epoch": 2.3282016264200136, "grad_norm": 0.44220536401472355, "learning_rate": 1.4513388595495098e-05, "loss": 2.6183, "step": 37505 }, { "epoch": 2.3282637035197715, "grad_norm": 0.44981831883171314, "learning_rate": 1.4510844472032482e-05, "loss": 2.7126, "step": 37506 }, { "epoch": 2.3283257806195294, "grad_norm": 0.5107080801959314, "learning_rate": 1.4508300533725406e-05, "loss": 2.7949, "step": 37507 }, { "epoch": 2.3283878577192874, "grad_norm": 0.44606050611309306, "learning_rate": 1.4505756780587131e-05, "loss": 2.6945, "step": 37508 }, { "epoch": 2.3284499348190453, "grad_norm": 0.48641380260080286, "learning_rate": 1.4503213212630951e-05, "loss": 2.6461, "step": 37509 }, { "epoch": 2.328512011918803, "grad_norm": 0.42697644252920947, "learning_rate": 1.4500669829870122e-05, "loss": 2.7249, "step": 37510 }, { "epoch": 2.328574089018561, "grad_norm": 0.43414819339774785, "learning_rate": 1.449812663231792e-05, "loss": 2.7939, "step": 37511 }, { "epoch": 2.328636166118319, "grad_norm": 0.431971720755583, "learning_rate": 1.4495583619987613e-05, "loss": 2.7392, "step": 37512 }, { "epoch": 2.328698243218077, "grad_norm": 0.44653085047882657, "learning_rate": 1.4493040792892442e-05, "loss": 2.7206, "step": 37513 }, { "epoch": 2.328760320317835, "grad_norm": 0.4362258067158398, "learning_rate": 1.4490498151045718e-05, "loss": 2.6263, "step": 37514 }, { "epoch": 2.328822397417593, "grad_norm": 0.43654975555757364, "learning_rate": 1.4487955694460681e-05, "loss": 2.6426, "step": 37515 }, { "epoch": 2.3288844745173507, "grad_norm": 0.4821209297038031, "learning_rate": 1.4485413423150607e-05, "loss": 2.7227, "step": 37516 }, { "epoch": 2.3289465516171086, "grad_norm": 0.4286684436636431, "learning_rate": 1.4482871337128729e-05, "loss": 2.7083, "step": 37517 }, { "epoch": 2.3290086287168665, "grad_norm": 0.496616508614846, "learning_rate": 1.4480329436408353e-05, "loss": 2.7205, "step": 37518 }, { "epoch": 2.3290707058166245, "grad_norm": 0.4339669314225696, "learning_rate": 1.4477787721002722e-05, "loss": 2.6161, "step": 37519 }, { "epoch": 2.329132782916382, "grad_norm": 0.4417268888184521, "learning_rate": 1.4475246190925096e-05, "loss": 2.7298, "step": 37520 }, { "epoch": 2.3291948600161403, "grad_norm": 0.48642524454224684, "learning_rate": 1.4472704846188729e-05, "loss": 2.7574, "step": 37521 }, { "epoch": 2.3292569371158978, "grad_norm": 0.44365450409101465, "learning_rate": 1.4470163686806877e-05, "loss": 2.6646, "step": 37522 }, { "epoch": 2.3293190142156557, "grad_norm": 0.4706216163817579, "learning_rate": 1.4467622712792816e-05, "loss": 2.7161, "step": 37523 }, { "epoch": 2.3293810913154136, "grad_norm": 0.44677519250570713, "learning_rate": 1.4465081924159796e-05, "loss": 2.7171, "step": 37524 }, { "epoch": 2.3294431684151715, "grad_norm": 0.455247348216261, "learning_rate": 1.446254132092107e-05, "loss": 2.75, "step": 37525 }, { "epoch": 2.3295052455149294, "grad_norm": 0.44775262497580276, "learning_rate": 1.4460000903089876e-05, "loss": 2.7201, "step": 37526 }, { "epoch": 2.3295673226146874, "grad_norm": 0.4518870345764726, "learning_rate": 1.44574606706795e-05, "loss": 2.6525, "step": 37527 }, { "epoch": 2.3296293997144453, "grad_norm": 0.4598291455307797, "learning_rate": 1.4454920623703182e-05, "loss": 2.612, "step": 37528 }, { "epoch": 2.329691476814203, "grad_norm": 0.48341082932743784, "learning_rate": 1.4452380762174172e-05, "loss": 2.736, "step": 37529 }, { "epoch": 2.329753553913961, "grad_norm": 0.440568026788999, "learning_rate": 1.4449841086105702e-05, "loss": 2.6872, "step": 37530 }, { "epoch": 2.329815631013719, "grad_norm": 0.47859301817038163, "learning_rate": 1.4447301595511065e-05, "loss": 2.629, "step": 37531 }, { "epoch": 2.329877708113477, "grad_norm": 0.46227008736838604, "learning_rate": 1.444476229040348e-05, "loss": 2.748, "step": 37532 }, { "epoch": 2.329939785213235, "grad_norm": 0.496031224072651, "learning_rate": 1.4442223170796205e-05, "loss": 2.692, "step": 37533 }, { "epoch": 2.3300018623129928, "grad_norm": 0.4753013871879568, "learning_rate": 1.4439684236702477e-05, "loss": 2.6901, "step": 37534 }, { "epoch": 2.3300639394127507, "grad_norm": 0.4806014862311009, "learning_rate": 1.4437145488135539e-05, "loss": 2.7427, "step": 37535 }, { "epoch": 2.3301260165125086, "grad_norm": 0.47570565849352525, "learning_rate": 1.4434606925108645e-05, "loss": 2.7043, "step": 37536 }, { "epoch": 2.3301880936122665, "grad_norm": 0.45054718631411494, "learning_rate": 1.4432068547635052e-05, "loss": 2.7636, "step": 37537 }, { "epoch": 2.3302501707120244, "grad_norm": 0.46603897049041115, "learning_rate": 1.4429530355728e-05, "loss": 2.7538, "step": 37538 }, { "epoch": 2.3303122478117824, "grad_norm": 0.48502435534565985, "learning_rate": 1.4426992349400714e-05, "loss": 2.6276, "step": 37539 }, { "epoch": 2.3303743249115403, "grad_norm": 0.4471006463968702, "learning_rate": 1.4424454528666437e-05, "loss": 2.7029, "step": 37540 }, { "epoch": 2.330436402011298, "grad_norm": 0.4381223577563923, "learning_rate": 1.442191689353843e-05, "loss": 2.685, "step": 37541 }, { "epoch": 2.330498479111056, "grad_norm": 0.4926197485161092, "learning_rate": 1.4419379444029918e-05, "loss": 2.7009, "step": 37542 }, { "epoch": 2.3305605562108136, "grad_norm": 0.42480110956197975, "learning_rate": 1.4416842180154139e-05, "loss": 2.6596, "step": 37543 }, { "epoch": 2.330622633310572, "grad_norm": 0.43274039414960114, "learning_rate": 1.4414305101924319e-05, "loss": 2.7007, "step": 37544 }, { "epoch": 2.3306847104103294, "grad_norm": 0.48762923386140705, "learning_rate": 1.4411768209353726e-05, "loss": 2.6606, "step": 37545 }, { "epoch": 2.3307467875100873, "grad_norm": 0.4288091699070319, "learning_rate": 1.4409231502455573e-05, "loss": 2.5816, "step": 37546 }, { "epoch": 2.3308088646098453, "grad_norm": 0.4281769122372516, "learning_rate": 1.4406694981243101e-05, "loss": 2.6096, "step": 37547 }, { "epoch": 2.330870941709603, "grad_norm": 0.47876557031017825, "learning_rate": 1.4404158645729543e-05, "loss": 2.702, "step": 37548 }, { "epoch": 2.330933018809361, "grad_norm": 0.47275176541277686, "learning_rate": 1.4401622495928114e-05, "loss": 2.6293, "step": 37549 }, { "epoch": 2.330995095909119, "grad_norm": 0.45967102633274437, "learning_rate": 1.439908653185208e-05, "loss": 2.6163, "step": 37550 }, { "epoch": 2.331057173008877, "grad_norm": 0.4573399406042472, "learning_rate": 1.4396550753514654e-05, "loss": 2.7828, "step": 37551 }, { "epoch": 2.331119250108635, "grad_norm": 0.5006366491212558, "learning_rate": 1.4394015160929064e-05, "loss": 2.586, "step": 37552 }, { "epoch": 2.3311813272083928, "grad_norm": 0.44040065762750513, "learning_rate": 1.4391479754108522e-05, "loss": 2.6908, "step": 37553 }, { "epoch": 2.3312434043081507, "grad_norm": 0.46121648653778824, "learning_rate": 1.4388944533066295e-05, "loss": 2.6908, "step": 37554 }, { "epoch": 2.3313054814079086, "grad_norm": 0.4517592532757099, "learning_rate": 1.438640949781559e-05, "loss": 2.6373, "step": 37555 }, { "epoch": 2.3313675585076665, "grad_norm": 0.47440844077450073, "learning_rate": 1.4383874648369628e-05, "loss": 2.6999, "step": 37556 }, { "epoch": 2.3314296356074244, "grad_norm": 0.42499851759483753, "learning_rate": 1.4381339984741621e-05, "loss": 2.695, "step": 37557 }, { "epoch": 2.3314917127071824, "grad_norm": 0.44510230742117624, "learning_rate": 1.4378805506944832e-05, "loss": 2.6724, "step": 37558 }, { "epoch": 2.3315537898069403, "grad_norm": 0.47544589320868397, "learning_rate": 1.437627121499246e-05, "loss": 2.7423, "step": 37559 }, { "epoch": 2.331615866906698, "grad_norm": 0.45181707182373343, "learning_rate": 1.4373737108897722e-05, "loss": 2.7508, "step": 37560 }, { "epoch": 2.331677944006456, "grad_norm": 0.48682650724930027, "learning_rate": 1.4371203188673854e-05, "loss": 2.6733, "step": 37561 }, { "epoch": 2.331740021106214, "grad_norm": 0.48533791536298715, "learning_rate": 1.436866945433405e-05, "loss": 2.7198, "step": 37562 }, { "epoch": 2.331802098205972, "grad_norm": 0.4039916392809127, "learning_rate": 1.4366135905891565e-05, "loss": 2.6645, "step": 37563 }, { "epoch": 2.33186417530573, "grad_norm": 0.43677576610722235, "learning_rate": 1.4363602543359594e-05, "loss": 2.6989, "step": 37564 }, { "epoch": 2.331926252405488, "grad_norm": 0.46903363229869793, "learning_rate": 1.4361069366751367e-05, "loss": 2.6526, "step": 37565 }, { "epoch": 2.3319883295052457, "grad_norm": 0.45303204010397674, "learning_rate": 1.4358536376080072e-05, "loss": 2.6508, "step": 37566 }, { "epoch": 2.3320504066050036, "grad_norm": 0.4971472000582645, "learning_rate": 1.4356003571358962e-05, "loss": 2.7253, "step": 37567 }, { "epoch": 2.332112483704761, "grad_norm": 0.4752562709958493, "learning_rate": 1.4353470952601234e-05, "loss": 2.6226, "step": 37568 }, { "epoch": 2.3321745608045195, "grad_norm": 0.45449829818755294, "learning_rate": 1.4350938519820084e-05, "loss": 2.6394, "step": 37569 }, { "epoch": 2.332236637904277, "grad_norm": 0.4945730194723951, "learning_rate": 1.4348406273028758e-05, "loss": 2.6709, "step": 37570 }, { "epoch": 2.332298715004035, "grad_norm": 0.4906776521457617, "learning_rate": 1.4345874212240456e-05, "loss": 2.6451, "step": 37571 }, { "epoch": 2.3323607921037928, "grad_norm": 0.47438913894711554, "learning_rate": 1.4343342337468364e-05, "loss": 2.668, "step": 37572 }, { "epoch": 2.3324228692035507, "grad_norm": 0.5150858053927658, "learning_rate": 1.4340810648725728e-05, "loss": 2.7262, "step": 37573 }, { "epoch": 2.3324849463033086, "grad_norm": 0.5099423566199606, "learning_rate": 1.433827914602574e-05, "loss": 2.7125, "step": 37574 }, { "epoch": 2.3325470234030665, "grad_norm": 0.45919695861865073, "learning_rate": 1.4335747829381601e-05, "loss": 2.6603, "step": 37575 }, { "epoch": 2.3326091005028244, "grad_norm": 0.5033242168874593, "learning_rate": 1.4333216698806517e-05, "loss": 2.6932, "step": 37576 }, { "epoch": 2.3326711776025824, "grad_norm": 0.4895976517659951, "learning_rate": 1.4330685754313711e-05, "loss": 2.6689, "step": 37577 }, { "epoch": 2.3327332547023403, "grad_norm": 0.47157735089578245, "learning_rate": 1.4328154995916376e-05, "loss": 2.6863, "step": 37578 }, { "epoch": 2.332795331802098, "grad_norm": 0.5253172472229268, "learning_rate": 1.4325624423627716e-05, "loss": 2.7512, "step": 37579 }, { "epoch": 2.332857408901856, "grad_norm": 0.5161687700222387, "learning_rate": 1.4323094037460915e-05, "loss": 2.6376, "step": 37580 }, { "epoch": 2.332919486001614, "grad_norm": 0.4801594894155907, "learning_rate": 1.4320563837429213e-05, "loss": 2.7634, "step": 37581 }, { "epoch": 2.332981563101372, "grad_norm": 0.44453995600590623, "learning_rate": 1.431803382354579e-05, "loss": 2.6526, "step": 37582 }, { "epoch": 2.33304364020113, "grad_norm": 0.4915956844664973, "learning_rate": 1.4315503995823848e-05, "loss": 2.6364, "step": 37583 }, { "epoch": 2.333105717300888, "grad_norm": 0.47271312024699497, "learning_rate": 1.431297435427658e-05, "loss": 2.6503, "step": 37584 }, { "epoch": 2.3331677944006457, "grad_norm": 0.4624913582123337, "learning_rate": 1.431044489891718e-05, "loss": 2.6939, "step": 37585 }, { "epoch": 2.3332298715004036, "grad_norm": 0.4461511945732649, "learning_rate": 1.430791562975886e-05, "loss": 2.6735, "step": 37586 }, { "epoch": 2.3332919486001615, "grad_norm": 0.5148854876952974, "learning_rate": 1.4305386546814815e-05, "loss": 2.6731, "step": 37587 }, { "epoch": 2.3333540256999195, "grad_norm": 0.513118983011594, "learning_rate": 1.4302857650098233e-05, "loss": 2.6917, "step": 37588 }, { "epoch": 2.3334161027996774, "grad_norm": 0.4572675208347841, "learning_rate": 1.4300328939622292e-05, "loss": 2.7419, "step": 37589 }, { "epoch": 2.3334781798994353, "grad_norm": 0.5162758909387998, "learning_rate": 1.4297800415400215e-05, "loss": 2.5024, "step": 37590 }, { "epoch": 2.3335402569991928, "grad_norm": 0.5235629144990028, "learning_rate": 1.4295272077445181e-05, "loss": 2.6513, "step": 37591 }, { "epoch": 2.333602334098951, "grad_norm": 0.40940528422734485, "learning_rate": 1.429274392577038e-05, "loss": 2.6544, "step": 37592 }, { "epoch": 2.3336644111987086, "grad_norm": 0.4703306251657886, "learning_rate": 1.4290215960388987e-05, "loss": 2.6717, "step": 37593 }, { "epoch": 2.3337264882984665, "grad_norm": 0.5239716835924978, "learning_rate": 1.4287688181314224e-05, "loss": 2.5682, "step": 37594 }, { "epoch": 2.3337885653982244, "grad_norm": 0.470946284758326, "learning_rate": 1.428516058855926e-05, "loss": 2.6868, "step": 37595 }, { "epoch": 2.3338506424979824, "grad_norm": 0.46240588854485387, "learning_rate": 1.4282633182137278e-05, "loss": 2.7176, "step": 37596 }, { "epoch": 2.3339127195977403, "grad_norm": 0.5216731897632855, "learning_rate": 1.4280105962061474e-05, "loss": 2.5763, "step": 37597 }, { "epoch": 2.333974796697498, "grad_norm": 0.5120943161811197, "learning_rate": 1.4277578928345015e-05, "loss": 2.7715, "step": 37598 }, { "epoch": 2.334036873797256, "grad_norm": 0.4797266369488698, "learning_rate": 1.427505208100111e-05, "loss": 2.695, "step": 37599 }, { "epoch": 2.334098950897014, "grad_norm": 0.4774916455651458, "learning_rate": 1.4272525420042937e-05, "loss": 2.6694, "step": 37600 }, { "epoch": 2.334161027996772, "grad_norm": 0.5617224033963767, "learning_rate": 1.4269998945483648e-05, "loss": 2.7542, "step": 37601 }, { "epoch": 2.33422310509653, "grad_norm": 0.5303349468041617, "learning_rate": 1.4267472657336473e-05, "loss": 2.6574, "step": 37602 }, { "epoch": 2.3342851821962878, "grad_norm": 0.5085842137534076, "learning_rate": 1.4264946555614545e-05, "loss": 2.6511, "step": 37603 }, { "epoch": 2.3343472592960457, "grad_norm": 0.5151061424112052, "learning_rate": 1.4262420640331081e-05, "loss": 2.7399, "step": 37604 }, { "epoch": 2.3344093363958036, "grad_norm": 0.42894443071033456, "learning_rate": 1.425989491149925e-05, "loss": 2.6629, "step": 37605 }, { "epoch": 2.3344714134955615, "grad_norm": 0.49891280648128405, "learning_rate": 1.4257369369132217e-05, "loss": 2.7401, "step": 37606 }, { "epoch": 2.3345334905953194, "grad_norm": 0.5012127652215592, "learning_rate": 1.4254844013243157e-05, "loss": 2.637, "step": 37607 }, { "epoch": 2.3345955676950774, "grad_norm": 0.4371350162406853, "learning_rate": 1.4252318843845264e-05, "loss": 2.6264, "step": 37608 }, { "epoch": 2.3346576447948353, "grad_norm": 0.4571438675069926, "learning_rate": 1.4249793860951699e-05, "loss": 2.6542, "step": 37609 }, { "epoch": 2.334719721894593, "grad_norm": 0.5472755327764169, "learning_rate": 1.424726906457564e-05, "loss": 2.7171, "step": 37610 }, { "epoch": 2.334781798994351, "grad_norm": 0.4762490132589456, "learning_rate": 1.4244744454730257e-05, "loss": 2.7076, "step": 37611 }, { "epoch": 2.334843876094109, "grad_norm": 0.4678685006085925, "learning_rate": 1.424222003142871e-05, "loss": 2.6923, "step": 37612 }, { "epoch": 2.334905953193867, "grad_norm": 0.5309186070107912, "learning_rate": 1.4239695794684193e-05, "loss": 2.647, "step": 37613 }, { "epoch": 2.334968030293625, "grad_norm": 0.47121888708116044, "learning_rate": 1.4237171744509869e-05, "loss": 2.6789, "step": 37614 }, { "epoch": 2.335030107393383, "grad_norm": 0.5483238102376785, "learning_rate": 1.42346478809189e-05, "loss": 2.72, "step": 37615 }, { "epoch": 2.3350921844931403, "grad_norm": 0.5457532276699436, "learning_rate": 1.4232124203924435e-05, "loss": 2.7358, "step": 37616 }, { "epoch": 2.3351542615928986, "grad_norm": 0.49448315053550695, "learning_rate": 1.422960071353968e-05, "loss": 2.7219, "step": 37617 }, { "epoch": 2.335216338692656, "grad_norm": 0.506368815054467, "learning_rate": 1.4227077409777784e-05, "loss": 2.6586, "step": 37618 }, { "epoch": 2.335278415792414, "grad_norm": 0.5053771087730808, "learning_rate": 1.422455429265191e-05, "loss": 2.7242, "step": 37619 }, { "epoch": 2.335340492892172, "grad_norm": 0.479156376548516, "learning_rate": 1.4222031362175215e-05, "loss": 2.7223, "step": 37620 }, { "epoch": 2.33540256999193, "grad_norm": 0.5115216873098221, "learning_rate": 1.4219508618360855e-05, "loss": 2.7286, "step": 37621 }, { "epoch": 2.3354646470916878, "grad_norm": 0.520017869257174, "learning_rate": 1.4216986061222021e-05, "loss": 2.6954, "step": 37622 }, { "epoch": 2.3355267241914457, "grad_norm": 0.520238037050657, "learning_rate": 1.4214463690771856e-05, "loss": 2.6996, "step": 37623 }, { "epoch": 2.3355888012912036, "grad_norm": 0.537008102579631, "learning_rate": 1.4211941507023525e-05, "loss": 2.6201, "step": 37624 }, { "epoch": 2.3356508783909615, "grad_norm": 0.48364382593955796, "learning_rate": 1.4209419509990162e-05, "loss": 2.6739, "step": 37625 }, { "epoch": 2.3357129554907194, "grad_norm": 0.4691983780581237, "learning_rate": 1.4206897699684967e-05, "loss": 2.7088, "step": 37626 }, { "epoch": 2.3357750325904774, "grad_norm": 0.4463106980703422, "learning_rate": 1.4204376076121072e-05, "loss": 2.6557, "step": 37627 }, { "epoch": 2.3358371096902353, "grad_norm": 0.4915346781380524, "learning_rate": 1.420185463931164e-05, "loss": 2.6827, "step": 37628 }, { "epoch": 2.335899186789993, "grad_norm": 0.4737172474271427, "learning_rate": 1.4199333389269803e-05, "loss": 2.6831, "step": 37629 }, { "epoch": 2.335961263889751, "grad_norm": 0.5126816520687574, "learning_rate": 1.4196812326008752e-05, "loss": 2.6351, "step": 37630 }, { "epoch": 2.336023340989509, "grad_norm": 0.4734463039372768, "learning_rate": 1.4194291449541625e-05, "loss": 2.63, "step": 37631 }, { "epoch": 2.336085418089267, "grad_norm": 0.497951064524397, "learning_rate": 1.4191770759881568e-05, "loss": 2.6957, "step": 37632 }, { "epoch": 2.336147495189025, "grad_norm": 0.41930342087559064, "learning_rate": 1.4189250257041736e-05, "loss": 2.6996, "step": 37633 }, { "epoch": 2.336209572288783, "grad_norm": 0.5136472456796672, "learning_rate": 1.4186729941035265e-05, "loss": 2.7508, "step": 37634 }, { "epoch": 2.3362716493885407, "grad_norm": 0.49089891295426336, "learning_rate": 1.4184209811875315e-05, "loss": 2.6304, "step": 37635 }, { "epoch": 2.3363337264882986, "grad_norm": 0.42241344705241324, "learning_rate": 1.4181689869575054e-05, "loss": 2.6841, "step": 37636 }, { "epoch": 2.3363958035880565, "grad_norm": 0.5170444832448763, "learning_rate": 1.4179170114147611e-05, "loss": 2.7654, "step": 37637 }, { "epoch": 2.3364578806878145, "grad_norm": 0.4556471087739191, "learning_rate": 1.4176650545606135e-05, "loss": 2.5893, "step": 37638 }, { "epoch": 2.336519957787572, "grad_norm": 0.480593238768283, "learning_rate": 1.4174131163963756e-05, "loss": 2.7652, "step": 37639 }, { "epoch": 2.3365820348873303, "grad_norm": 0.5100030163600955, "learning_rate": 1.4171611969233645e-05, "loss": 2.6738, "step": 37640 }, { "epoch": 2.3366441119870878, "grad_norm": 0.46543684556509607, "learning_rate": 1.4169092961428932e-05, "loss": 2.6953, "step": 37641 }, { "epoch": 2.3367061890868457, "grad_norm": 0.46108704097033204, "learning_rate": 1.4166574140562761e-05, "loss": 2.6612, "step": 37642 }, { "epoch": 2.3367682661866036, "grad_norm": 0.49141901057282034, "learning_rate": 1.4164055506648255e-05, "loss": 2.6487, "step": 37643 }, { "epoch": 2.3368303432863615, "grad_norm": 0.4800731767317151, "learning_rate": 1.4161537059698582e-05, "loss": 2.6726, "step": 37644 }, { "epoch": 2.3368924203861194, "grad_norm": 0.462965045853627, "learning_rate": 1.4159018799726876e-05, "loss": 2.7268, "step": 37645 }, { "epoch": 2.3369544974858774, "grad_norm": 0.4701100011703414, "learning_rate": 1.4156500726746268e-05, "loss": 2.682, "step": 37646 }, { "epoch": 2.3370165745856353, "grad_norm": 0.4696287184104298, "learning_rate": 1.4153982840769897e-05, "loss": 2.6493, "step": 37647 }, { "epoch": 2.337078651685393, "grad_norm": 0.45615428351657683, "learning_rate": 1.4151465141810882e-05, "loss": 2.7166, "step": 37648 }, { "epoch": 2.337140728785151, "grad_norm": 0.4731287682961449, "learning_rate": 1.4148947629882398e-05, "loss": 2.699, "step": 37649 }, { "epoch": 2.337202805884909, "grad_norm": 0.5003989237589281, "learning_rate": 1.414643030499755e-05, "loss": 2.6112, "step": 37650 }, { "epoch": 2.337264882984667, "grad_norm": 0.4517774934683084, "learning_rate": 1.4143913167169481e-05, "loss": 2.683, "step": 37651 }, { "epoch": 2.337326960084425, "grad_norm": 0.47078908238319356, "learning_rate": 1.4141396216411306e-05, "loss": 2.6513, "step": 37652 }, { "epoch": 2.337389037184183, "grad_norm": 0.503153630739259, "learning_rate": 1.4138879452736187e-05, "loss": 2.7041, "step": 37653 }, { "epoch": 2.3374511142839407, "grad_norm": 0.4585168120849343, "learning_rate": 1.4136362876157238e-05, "loss": 2.654, "step": 37654 }, { "epoch": 2.3375131913836986, "grad_norm": 0.44039991324663824, "learning_rate": 1.4133846486687597e-05, "loss": 2.6981, "step": 37655 }, { "epoch": 2.3375752684834565, "grad_norm": 0.48879161864839155, "learning_rate": 1.4131330284340377e-05, "loss": 2.6578, "step": 37656 }, { "epoch": 2.3376373455832145, "grad_norm": 0.46569496243623104, "learning_rate": 1.4128814269128704e-05, "loss": 2.6942, "step": 37657 }, { "epoch": 2.3376994226829724, "grad_norm": 0.41896065992862974, "learning_rate": 1.4126298441065728e-05, "loss": 2.7091, "step": 37658 }, { "epoch": 2.3377614997827303, "grad_norm": 0.4239408457411857, "learning_rate": 1.4123782800164564e-05, "loss": 2.7242, "step": 37659 }, { "epoch": 2.337823576882488, "grad_norm": 0.47801822579623704, "learning_rate": 1.4121267346438333e-05, "loss": 2.7074, "step": 37660 }, { "epoch": 2.337885653982246, "grad_norm": 0.4724334159994505, "learning_rate": 1.4118752079900143e-05, "loss": 2.6517, "step": 37661 }, { "epoch": 2.3379477310820036, "grad_norm": 0.4460842572486574, "learning_rate": 1.4116237000563153e-05, "loss": 2.6901, "step": 37662 }, { "epoch": 2.338009808181762, "grad_norm": 0.48520659770575986, "learning_rate": 1.4113722108440463e-05, "loss": 2.5445, "step": 37663 }, { "epoch": 2.3380718852815194, "grad_norm": 0.4593904362894137, "learning_rate": 1.41112074035452e-05, "loss": 2.6478, "step": 37664 }, { "epoch": 2.3381339623812774, "grad_norm": 0.4556705073797295, "learning_rate": 1.4108692885890462e-05, "loss": 2.619, "step": 37665 }, { "epoch": 2.3381960394810353, "grad_norm": 0.5272504088127538, "learning_rate": 1.4106178555489403e-05, "loss": 2.6959, "step": 37666 }, { "epoch": 2.338258116580793, "grad_norm": 0.4914632797504364, "learning_rate": 1.4103664412355105e-05, "loss": 2.6946, "step": 37667 }, { "epoch": 2.338320193680551, "grad_norm": 0.48649402837332567, "learning_rate": 1.4101150456500722e-05, "loss": 2.7003, "step": 37668 }, { "epoch": 2.338382270780309, "grad_norm": 0.45744308435539366, "learning_rate": 1.4098636687939354e-05, "loss": 2.7338, "step": 37669 }, { "epoch": 2.338444347880067, "grad_norm": 0.4436459896263223, "learning_rate": 1.4096123106684111e-05, "loss": 2.6474, "step": 37670 }, { "epoch": 2.338506424979825, "grad_norm": 0.501398396012944, "learning_rate": 1.4093609712748096e-05, "loss": 2.7119, "step": 37671 }, { "epoch": 2.338568502079583, "grad_norm": 0.516250837431591, "learning_rate": 1.4091096506144447e-05, "loss": 2.6817, "step": 37672 }, { "epoch": 2.3386305791793407, "grad_norm": 0.45630782492530864, "learning_rate": 1.408858348688627e-05, "loss": 2.7236, "step": 37673 }, { "epoch": 2.3386926562790986, "grad_norm": 0.4560841912277148, "learning_rate": 1.4086070654986667e-05, "loss": 2.6951, "step": 37674 }, { "epoch": 2.3387547333788565, "grad_norm": 0.5143754441362506, "learning_rate": 1.4083558010458736e-05, "loss": 2.6491, "step": 37675 }, { "epoch": 2.3388168104786144, "grad_norm": 0.43299854081086403, "learning_rate": 1.4081045553315619e-05, "loss": 2.6616, "step": 37676 }, { "epoch": 2.3388788875783724, "grad_norm": 0.48282038487782764, "learning_rate": 1.4078533283570406e-05, "loss": 2.6719, "step": 37677 }, { "epoch": 2.3389409646781303, "grad_norm": 0.4528395806185709, "learning_rate": 1.4076021201236205e-05, "loss": 2.6023, "step": 37678 }, { "epoch": 2.339003041777888, "grad_norm": 0.4959204513753885, "learning_rate": 1.407350930632611e-05, "loss": 2.7043, "step": 37679 }, { "epoch": 2.339065118877646, "grad_norm": 0.46866368452731366, "learning_rate": 1.4070997598853252e-05, "loss": 2.6184, "step": 37680 }, { "epoch": 2.339127195977404, "grad_norm": 0.4697507665667489, "learning_rate": 1.4068486078830723e-05, "loss": 2.7144, "step": 37681 }, { "epoch": 2.339189273077162, "grad_norm": 0.5152304035348946, "learning_rate": 1.406597474627162e-05, "loss": 2.6795, "step": 37682 }, { "epoch": 2.33925135017692, "grad_norm": 0.43249180959652883, "learning_rate": 1.4063463601189053e-05, "loss": 2.6553, "step": 37683 }, { "epoch": 2.339313427276678, "grad_norm": 0.4654740620156646, "learning_rate": 1.4060952643596103e-05, "loss": 2.7324, "step": 37684 }, { "epoch": 2.3393755043764357, "grad_norm": 0.48080671256016255, "learning_rate": 1.4058441873505906e-05, "loss": 2.6083, "step": 37685 }, { "epoch": 2.3394375814761936, "grad_norm": 0.47281714924343693, "learning_rate": 1.4055931290931545e-05, "loss": 2.5871, "step": 37686 }, { "epoch": 2.339499658575951, "grad_norm": 0.4737963050164809, "learning_rate": 1.4053420895886111e-05, "loss": 2.5987, "step": 37687 }, { "epoch": 2.3395617356757095, "grad_norm": 0.47581335501253913, "learning_rate": 1.4050910688382696e-05, "loss": 2.7295, "step": 37688 }, { "epoch": 2.339623812775467, "grad_norm": 0.474401027960532, "learning_rate": 1.404840066843442e-05, "loss": 2.6944, "step": 37689 }, { "epoch": 2.339685889875225, "grad_norm": 0.5131334739477866, "learning_rate": 1.404589083605436e-05, "loss": 2.707, "step": 37690 }, { "epoch": 2.3397479669749828, "grad_norm": 0.49172646258584346, "learning_rate": 1.4043381191255623e-05, "loss": 2.7665, "step": 37691 }, { "epoch": 2.3398100440747407, "grad_norm": 0.4444267766970755, "learning_rate": 1.4040871734051292e-05, "loss": 2.6528, "step": 37692 }, { "epoch": 2.3398721211744986, "grad_norm": 0.4535513549188127, "learning_rate": 1.4038362464454447e-05, "loss": 2.7678, "step": 37693 }, { "epoch": 2.3399341982742565, "grad_norm": 0.4722825614473309, "learning_rate": 1.4035853382478209e-05, "loss": 2.755, "step": 37694 }, { "epoch": 2.3399962753740144, "grad_norm": 0.4848789854089953, "learning_rate": 1.4033344488135659e-05, "loss": 2.7686, "step": 37695 }, { "epoch": 2.3400583524737724, "grad_norm": 0.4318599747722931, "learning_rate": 1.4030835781439877e-05, "loss": 2.7011, "step": 37696 }, { "epoch": 2.3401204295735303, "grad_norm": 0.4407354845360378, "learning_rate": 1.4028327262403946e-05, "loss": 2.6625, "step": 37697 }, { "epoch": 2.340182506673288, "grad_norm": 0.45029989422130423, "learning_rate": 1.4025818931040973e-05, "loss": 2.6255, "step": 37698 }, { "epoch": 2.340244583773046, "grad_norm": 0.4343733266070949, "learning_rate": 1.402331078736404e-05, "loss": 2.6834, "step": 37699 }, { "epoch": 2.340306660872804, "grad_norm": 0.4174740261004614, "learning_rate": 1.4020802831386214e-05, "loss": 2.5749, "step": 37700 }, { "epoch": 2.340368737972562, "grad_norm": 0.46410527594090256, "learning_rate": 1.4018295063120613e-05, "loss": 2.7601, "step": 37701 }, { "epoch": 2.34043081507232, "grad_norm": 0.4109674152755499, "learning_rate": 1.4015787482580278e-05, "loss": 2.7244, "step": 37702 }, { "epoch": 2.340492892172078, "grad_norm": 0.44953530809296466, "learning_rate": 1.4013280089778335e-05, "loss": 2.6489, "step": 37703 }, { "epoch": 2.3405549692718357, "grad_norm": 0.42964383683740176, "learning_rate": 1.4010772884727841e-05, "loss": 2.6783, "step": 37704 }, { "epoch": 2.3406170463715936, "grad_norm": 0.4412285607454455, "learning_rate": 1.4008265867441888e-05, "loss": 2.6614, "step": 37705 }, { "epoch": 2.3406791234713515, "grad_norm": 0.4351280319680273, "learning_rate": 1.4005759037933547e-05, "loss": 2.6875, "step": 37706 }, { "epoch": 2.3407412005711095, "grad_norm": 0.4480755897373362, "learning_rate": 1.4003252396215882e-05, "loss": 2.5632, "step": 37707 }, { "epoch": 2.3408032776708674, "grad_norm": 0.4783474529350065, "learning_rate": 1.4000745942302002e-05, "loss": 2.7823, "step": 37708 }, { "epoch": 2.3408653547706253, "grad_norm": 0.4562457580891775, "learning_rate": 1.3998239676204971e-05, "loss": 2.7079, "step": 37709 }, { "epoch": 2.3409274318703828, "grad_norm": 0.49012160809001626, "learning_rate": 1.3995733597937865e-05, "loss": 2.7932, "step": 37710 }, { "epoch": 2.340989508970141, "grad_norm": 0.4675615981773651, "learning_rate": 1.399322770751374e-05, "loss": 2.6028, "step": 37711 }, { "epoch": 2.3410515860698986, "grad_norm": 0.44069754835203817, "learning_rate": 1.3990722004945705e-05, "loss": 2.7529, "step": 37712 }, { "epoch": 2.3411136631696565, "grad_norm": 0.5184856016758832, "learning_rate": 1.3988216490246813e-05, "loss": 2.775, "step": 37713 }, { "epoch": 2.3411757402694144, "grad_norm": 0.47211631957909855, "learning_rate": 1.3985711163430137e-05, "loss": 2.6093, "step": 37714 }, { "epoch": 2.3412378173691724, "grad_norm": 0.4453854919309101, "learning_rate": 1.3983206024508732e-05, "loss": 2.6568, "step": 37715 }, { "epoch": 2.3412998944689303, "grad_norm": 0.500884763401152, "learning_rate": 1.3980701073495706e-05, "loss": 2.6445, "step": 37716 }, { "epoch": 2.341361971568688, "grad_norm": 0.4748627257250698, "learning_rate": 1.39781963104041e-05, "loss": 2.6569, "step": 37717 }, { "epoch": 2.341424048668446, "grad_norm": 0.4449630001577224, "learning_rate": 1.3975691735246993e-05, "loss": 2.7052, "step": 37718 }, { "epoch": 2.341486125768204, "grad_norm": 0.464091389331615, "learning_rate": 1.3973187348037442e-05, "loss": 2.6008, "step": 37719 }, { "epoch": 2.341548202867962, "grad_norm": 0.4662689192106399, "learning_rate": 1.397068314878851e-05, "loss": 2.6233, "step": 37720 }, { "epoch": 2.34161027996772, "grad_norm": 0.44707361622560854, "learning_rate": 1.3968179137513282e-05, "loss": 2.6977, "step": 37721 }, { "epoch": 2.341672357067478, "grad_norm": 0.4667681441782598, "learning_rate": 1.3965675314224808e-05, "loss": 2.6135, "step": 37722 }, { "epoch": 2.3417344341672357, "grad_norm": 0.48420423143106495, "learning_rate": 1.3963171678936155e-05, "loss": 2.6853, "step": 37723 }, { "epoch": 2.3417965112669936, "grad_norm": 0.4890446912551662, "learning_rate": 1.3960668231660373e-05, "loss": 2.6035, "step": 37724 }, { "epoch": 2.3418585883667515, "grad_norm": 0.4638184531862634, "learning_rate": 1.3958164972410542e-05, "loss": 2.6456, "step": 37725 }, { "epoch": 2.3419206654665095, "grad_norm": 0.4557681112452113, "learning_rate": 1.3955661901199717e-05, "loss": 2.5571, "step": 37726 }, { "epoch": 2.3419827425662674, "grad_norm": 0.5991417120360222, "learning_rate": 1.3953159018040956e-05, "loss": 2.6571, "step": 37727 }, { "epoch": 2.3420448196660253, "grad_norm": 0.4318710506109573, "learning_rate": 1.3950656322947298e-05, "loss": 2.6387, "step": 37728 }, { "epoch": 2.342106896765783, "grad_norm": 0.4535148686313154, "learning_rate": 1.394815381593183e-05, "loss": 2.6758, "step": 37729 }, { "epoch": 2.342168973865541, "grad_norm": 0.49225178556084515, "learning_rate": 1.39456514970076e-05, "loss": 2.5244, "step": 37730 }, { "epoch": 2.342231050965299, "grad_norm": 0.49452481017055333, "learning_rate": 1.3943149366187659e-05, "loss": 2.6049, "step": 37731 }, { "epoch": 2.342293128065057, "grad_norm": 0.5009728742021272, "learning_rate": 1.3940647423485042e-05, "loss": 2.7234, "step": 37732 }, { "epoch": 2.342355205164815, "grad_norm": 0.4978240767651653, "learning_rate": 1.3938145668912833e-05, "loss": 2.6347, "step": 37733 }, { "epoch": 2.342417282264573, "grad_norm": 0.5010232885034263, "learning_rate": 1.3935644102484064e-05, "loss": 2.7451, "step": 37734 }, { "epoch": 2.3424793593643303, "grad_norm": 0.46805021185171775, "learning_rate": 1.3933142724211807e-05, "loss": 2.6458, "step": 37735 }, { "epoch": 2.3425414364640886, "grad_norm": 0.49824527453905876, "learning_rate": 1.39306415341091e-05, "loss": 2.6746, "step": 37736 }, { "epoch": 2.342603513563846, "grad_norm": 0.5091435228270873, "learning_rate": 1.3928140532188994e-05, "loss": 2.6542, "step": 37737 }, { "epoch": 2.342665590663604, "grad_norm": 0.4934542415984701, "learning_rate": 1.3925639718464518e-05, "loss": 2.6401, "step": 37738 }, { "epoch": 2.342727667763362, "grad_norm": 0.5554951442880542, "learning_rate": 1.3923139092948756e-05, "loss": 2.6936, "step": 37739 }, { "epoch": 2.34278974486312, "grad_norm": 0.5092420556633712, "learning_rate": 1.3920638655654732e-05, "loss": 2.6509, "step": 37740 }, { "epoch": 2.342851821962878, "grad_norm": 0.5372636330775503, "learning_rate": 1.3918138406595493e-05, "loss": 2.7056, "step": 37741 }, { "epoch": 2.3429138990626357, "grad_norm": 0.4864259338072327, "learning_rate": 1.3915638345784087e-05, "loss": 2.6799, "step": 37742 }, { "epoch": 2.3429759761623936, "grad_norm": 0.5140723636225858, "learning_rate": 1.391313847323354e-05, "loss": 2.7181, "step": 37743 }, { "epoch": 2.3430380532621515, "grad_norm": 0.5396754056175073, "learning_rate": 1.3910638788956925e-05, "loss": 2.6806, "step": 37744 }, { "epoch": 2.3431001303619095, "grad_norm": 0.5162021412498702, "learning_rate": 1.390813929296727e-05, "loss": 2.7198, "step": 37745 }, { "epoch": 2.3431622074616674, "grad_norm": 0.5430378218153984, "learning_rate": 1.3905639985277613e-05, "loss": 2.6151, "step": 37746 }, { "epoch": 2.3432242845614253, "grad_norm": 0.5425964561095685, "learning_rate": 1.3903140865900982e-05, "loss": 2.6652, "step": 37747 }, { "epoch": 2.343286361661183, "grad_norm": 0.5372335394480672, "learning_rate": 1.3900641934850439e-05, "loss": 2.7273, "step": 37748 }, { "epoch": 2.343348438760941, "grad_norm": 0.5566288193906399, "learning_rate": 1.3898143192139013e-05, "loss": 2.7115, "step": 37749 }, { "epoch": 2.343410515860699, "grad_norm": 0.535146587869337, "learning_rate": 1.3895644637779737e-05, "loss": 2.6023, "step": 37750 }, { "epoch": 2.343472592960457, "grad_norm": 0.48306662220379976, "learning_rate": 1.3893146271785634e-05, "loss": 2.6136, "step": 37751 }, { "epoch": 2.343534670060215, "grad_norm": 0.5318862246842938, "learning_rate": 1.3890648094169762e-05, "loss": 2.6977, "step": 37752 }, { "epoch": 2.343596747159973, "grad_norm": 0.48684528536829624, "learning_rate": 1.3888150104945153e-05, "loss": 2.6515, "step": 37753 }, { "epoch": 2.3436588242597307, "grad_norm": 0.5135891923972025, "learning_rate": 1.3885652304124824e-05, "loss": 2.545, "step": 37754 }, { "epoch": 2.3437209013594886, "grad_norm": 0.5142359084561733, "learning_rate": 1.3883154691721816e-05, "loss": 2.7624, "step": 37755 }, { "epoch": 2.3437829784592465, "grad_norm": 0.5208010536363377, "learning_rate": 1.3880657267749142e-05, "loss": 2.6806, "step": 37756 }, { "epoch": 2.3438450555590045, "grad_norm": 0.45614585537012775, "learning_rate": 1.3878160032219867e-05, "loss": 2.6969, "step": 37757 }, { "epoch": 2.343907132658762, "grad_norm": 0.508921454076308, "learning_rate": 1.3875662985146991e-05, "loss": 2.6789, "step": 37758 }, { "epoch": 2.3439692097585203, "grad_norm": 0.46420202116610465, "learning_rate": 1.3873166126543558e-05, "loss": 2.5703, "step": 37759 }, { "epoch": 2.3440312868582778, "grad_norm": 0.48648416551322343, "learning_rate": 1.3870669456422575e-05, "loss": 2.7459, "step": 37760 }, { "epoch": 2.3440933639580357, "grad_norm": 0.45875657292576305, "learning_rate": 1.3868172974797089e-05, "loss": 2.6684, "step": 37761 }, { "epoch": 2.3441554410577936, "grad_norm": 0.442160459978968, "learning_rate": 1.3865676681680118e-05, "loss": 2.6348, "step": 37762 }, { "epoch": 2.3442175181575515, "grad_norm": 0.4598403578233571, "learning_rate": 1.3863180577084684e-05, "loss": 2.6119, "step": 37763 }, { "epoch": 2.3442795952573094, "grad_norm": 0.5097914356698325, "learning_rate": 1.3860684661023793e-05, "loss": 2.7252, "step": 37764 }, { "epoch": 2.3443416723570674, "grad_norm": 0.4785243909383228, "learning_rate": 1.3858188933510486e-05, "loss": 2.7399, "step": 37765 }, { "epoch": 2.3444037494568253, "grad_norm": 0.41440812566810387, "learning_rate": 1.3855693394557789e-05, "loss": 2.7233, "step": 37766 }, { "epoch": 2.344465826556583, "grad_norm": 0.4542077545367859, "learning_rate": 1.3853198044178722e-05, "loss": 2.6141, "step": 37767 }, { "epoch": 2.344527903656341, "grad_norm": 0.4631587833175423, "learning_rate": 1.3850702882386291e-05, "loss": 2.7243, "step": 37768 }, { "epoch": 2.344589980756099, "grad_norm": 0.44197958449104885, "learning_rate": 1.3848207909193522e-05, "loss": 2.677, "step": 37769 }, { "epoch": 2.344652057855857, "grad_norm": 0.48717874799284927, "learning_rate": 1.3845713124613413e-05, "loss": 2.7163, "step": 37770 }, { "epoch": 2.344714134955615, "grad_norm": 0.43315265316780865, "learning_rate": 1.3843218528659008e-05, "loss": 2.6045, "step": 37771 }, { "epoch": 2.344776212055373, "grad_norm": 0.45339620629592536, "learning_rate": 1.3840724121343313e-05, "loss": 2.6662, "step": 37772 }, { "epoch": 2.3448382891551307, "grad_norm": 0.4886969360951014, "learning_rate": 1.3838229902679334e-05, "loss": 2.673, "step": 37773 }, { "epoch": 2.3449003662548886, "grad_norm": 0.4351433736652763, "learning_rate": 1.3835735872680078e-05, "loss": 2.5894, "step": 37774 }, { "epoch": 2.3449624433546465, "grad_norm": 0.4835436362089877, "learning_rate": 1.383324203135858e-05, "loss": 2.6753, "step": 37775 }, { "epoch": 2.3450245204544045, "grad_norm": 0.5290297787997555, "learning_rate": 1.3830748378727832e-05, "loss": 2.7952, "step": 37776 }, { "epoch": 2.3450865975541624, "grad_norm": 0.45805100031983864, "learning_rate": 1.3828254914800859e-05, "loss": 2.6381, "step": 37777 }, { "epoch": 2.3451486746539203, "grad_norm": 0.4987276846929541, "learning_rate": 1.382576163959064e-05, "loss": 2.6539, "step": 37778 }, { "epoch": 2.345210751753678, "grad_norm": 0.5200660362403077, "learning_rate": 1.3823268553110225e-05, "loss": 2.7024, "step": 37779 }, { "epoch": 2.345272828853436, "grad_norm": 0.4655588243810286, "learning_rate": 1.3820775655372591e-05, "loss": 2.6782, "step": 37780 }, { "epoch": 2.345334905953194, "grad_norm": 0.45764568792041477, "learning_rate": 1.3818282946390764e-05, "loss": 2.6612, "step": 37781 }, { "epoch": 2.345396983052952, "grad_norm": 0.4582816893207722, "learning_rate": 1.3815790426177733e-05, "loss": 2.6523, "step": 37782 }, { "epoch": 2.3454590601527094, "grad_norm": 0.461137679646717, "learning_rate": 1.3813298094746491e-05, "loss": 2.7185, "step": 37783 }, { "epoch": 2.345521137252468, "grad_norm": 0.463237519797598, "learning_rate": 1.3810805952110073e-05, "loss": 2.7813, "step": 37784 }, { "epoch": 2.3455832143522253, "grad_norm": 0.4580780008050884, "learning_rate": 1.380831399828147e-05, "loss": 2.6107, "step": 37785 }, { "epoch": 2.345645291451983, "grad_norm": 0.48245912477568004, "learning_rate": 1.3805822233273674e-05, "loss": 2.6869, "step": 37786 }, { "epoch": 2.345707368551741, "grad_norm": 0.4825438526712218, "learning_rate": 1.3803330657099677e-05, "loss": 2.6318, "step": 37787 }, { "epoch": 2.345769445651499, "grad_norm": 0.461143011410942, "learning_rate": 1.3800839269772503e-05, "loss": 2.7282, "step": 37788 }, { "epoch": 2.345831522751257, "grad_norm": 0.4987930691797352, "learning_rate": 1.3798348071305139e-05, "loss": 2.7161, "step": 37789 }, { "epoch": 2.345893599851015, "grad_norm": 0.5073732677316428, "learning_rate": 1.3795857061710587e-05, "loss": 2.6805, "step": 37790 }, { "epoch": 2.345955676950773, "grad_norm": 0.447167631516972, "learning_rate": 1.3793366241001832e-05, "loss": 2.7165, "step": 37791 }, { "epoch": 2.3460177540505307, "grad_norm": 0.4724195849396412, "learning_rate": 1.3790875609191861e-05, "loss": 2.66, "step": 37792 }, { "epoch": 2.3460798311502886, "grad_norm": 0.4484580689873407, "learning_rate": 1.37883851662937e-05, "loss": 2.7898, "step": 37793 }, { "epoch": 2.3461419082500465, "grad_norm": 0.4636615819519677, "learning_rate": 1.3785894912320319e-05, "loss": 2.6294, "step": 37794 }, { "epoch": 2.3462039853498045, "grad_norm": 0.5114480387266054, "learning_rate": 1.378340484728472e-05, "loss": 2.7428, "step": 37795 }, { "epoch": 2.3462660624495624, "grad_norm": 0.4448200983553848, "learning_rate": 1.3780914971199872e-05, "loss": 2.7632, "step": 37796 }, { "epoch": 2.3463281395493203, "grad_norm": 0.43664711175271526, "learning_rate": 1.37784252840788e-05, "loss": 2.6653, "step": 37797 }, { "epoch": 2.346390216649078, "grad_norm": 0.4961202719499888, "learning_rate": 1.3775935785934457e-05, "loss": 2.7498, "step": 37798 }, { "epoch": 2.346452293748836, "grad_norm": 0.4767975774588263, "learning_rate": 1.3773446476779872e-05, "loss": 2.7087, "step": 37799 }, { "epoch": 2.346514370848594, "grad_norm": 0.4470231510883388, "learning_rate": 1.3770957356628006e-05, "loss": 2.6533, "step": 37800 }, { "epoch": 2.346576447948352, "grad_norm": 0.5033218188368758, "learning_rate": 1.3768468425491837e-05, "loss": 2.6495, "step": 37801 }, { "epoch": 2.34663852504811, "grad_norm": 0.46014973841465645, "learning_rate": 1.3765979683384379e-05, "loss": 2.6273, "step": 37802 }, { "epoch": 2.346700602147868, "grad_norm": 0.4508245047405837, "learning_rate": 1.37634911303186e-05, "loss": 2.7528, "step": 37803 }, { "epoch": 2.3467626792476257, "grad_norm": 0.4521663159652505, "learning_rate": 1.3761002766307485e-05, "loss": 2.7366, "step": 37804 }, { "epoch": 2.3468247563473836, "grad_norm": 0.4483147522941415, "learning_rate": 1.375851459136402e-05, "loss": 2.5829, "step": 37805 }, { "epoch": 2.346886833447141, "grad_norm": 0.4440977191582588, "learning_rate": 1.375602660550116e-05, "loss": 2.7503, "step": 37806 }, { "epoch": 2.3469489105468995, "grad_norm": 0.44808149262236374, "learning_rate": 1.375353880873193e-05, "loss": 2.5618, "step": 37807 }, { "epoch": 2.347010987646657, "grad_norm": 0.4548940982294881, "learning_rate": 1.3751051201069281e-05, "loss": 2.6472, "step": 37808 }, { "epoch": 2.347073064746415, "grad_norm": 0.4352324680991195, "learning_rate": 1.3748563782526202e-05, "loss": 2.7622, "step": 37809 }, { "epoch": 2.347135141846173, "grad_norm": 0.4560027491137388, "learning_rate": 1.374607655311565e-05, "loss": 2.7039, "step": 37810 }, { "epoch": 2.3471972189459307, "grad_norm": 0.4859814614775041, "learning_rate": 1.374358951285063e-05, "loss": 2.5861, "step": 37811 }, { "epoch": 2.3472592960456886, "grad_norm": 0.49662944076015925, "learning_rate": 1.3741102661744109e-05, "loss": 2.6506, "step": 37812 }, { "epoch": 2.3473213731454465, "grad_norm": 0.47605309072708596, "learning_rate": 1.3738615999809056e-05, "loss": 2.753, "step": 37813 }, { "epoch": 2.3473834502452045, "grad_norm": 0.46835925982730775, "learning_rate": 1.3736129527058433e-05, "loss": 2.7024, "step": 37814 }, { "epoch": 2.3474455273449624, "grad_norm": 0.5100407643482242, "learning_rate": 1.3733643243505234e-05, "loss": 2.7088, "step": 37815 }, { "epoch": 2.3475076044447203, "grad_norm": 0.4716786449728871, "learning_rate": 1.3731157149162426e-05, "loss": 2.662, "step": 37816 }, { "epoch": 2.347569681544478, "grad_norm": 0.47025445472193195, "learning_rate": 1.3728671244042979e-05, "loss": 2.6423, "step": 37817 }, { "epoch": 2.347631758644236, "grad_norm": 0.48904636709885924, "learning_rate": 1.3726185528159857e-05, "loss": 2.6488, "step": 37818 }, { "epoch": 2.347693835743994, "grad_norm": 0.49047821181626355, "learning_rate": 1.3723700001526013e-05, "loss": 2.8382, "step": 37819 }, { "epoch": 2.347755912843752, "grad_norm": 0.4656002556543287, "learning_rate": 1.3721214664154453e-05, "loss": 2.6293, "step": 37820 }, { "epoch": 2.34781798994351, "grad_norm": 0.46197238416828645, "learning_rate": 1.371872951605812e-05, "loss": 2.7087, "step": 37821 }, { "epoch": 2.347880067043268, "grad_norm": 0.483992133028354, "learning_rate": 1.371624455724998e-05, "loss": 2.6723, "step": 37822 }, { "epoch": 2.3479421441430257, "grad_norm": 0.4674345792038249, "learning_rate": 1.3713759787742991e-05, "loss": 2.7051, "step": 37823 }, { "epoch": 2.3480042212427836, "grad_norm": 0.47268124979283443, "learning_rate": 1.371127520755014e-05, "loss": 2.72, "step": 37824 }, { "epoch": 2.3480662983425415, "grad_norm": 0.5251436752180356, "learning_rate": 1.370879081668437e-05, "loss": 2.7512, "step": 37825 }, { "epoch": 2.3481283754422995, "grad_norm": 0.44962105358874555, "learning_rate": 1.3706306615158655e-05, "loss": 2.7105, "step": 37826 }, { "epoch": 2.3481904525420574, "grad_norm": 0.45212548282102877, "learning_rate": 1.3703822602985944e-05, "loss": 2.5948, "step": 37827 }, { "epoch": 2.3482525296418153, "grad_norm": 0.5473446328136699, "learning_rate": 1.370133878017919e-05, "loss": 2.6537, "step": 37828 }, { "epoch": 2.348314606741573, "grad_norm": 0.4845482147356714, "learning_rate": 1.3698855146751377e-05, "loss": 2.6998, "step": 37829 }, { "epoch": 2.348376683841331, "grad_norm": 0.5037855550841035, "learning_rate": 1.3696371702715449e-05, "loss": 2.6469, "step": 37830 }, { "epoch": 2.3484387609410886, "grad_norm": 0.5398421247488198, "learning_rate": 1.3693888448084346e-05, "loss": 2.7485, "step": 37831 }, { "epoch": 2.348500838040847, "grad_norm": 0.47192826444116487, "learning_rate": 1.3691405382871058e-05, "loss": 2.6741, "step": 37832 }, { "epoch": 2.3485629151406044, "grad_norm": 0.5308204765749072, "learning_rate": 1.3688922507088509e-05, "loss": 2.6706, "step": 37833 }, { "epoch": 2.3486249922403624, "grad_norm": 0.4909215267941061, "learning_rate": 1.3686439820749674e-05, "loss": 2.6687, "step": 37834 }, { "epoch": 2.3486870693401203, "grad_norm": 0.4655297930880899, "learning_rate": 1.3683957323867508e-05, "loss": 2.6674, "step": 37835 }, { "epoch": 2.348749146439878, "grad_norm": 0.4750463412918964, "learning_rate": 1.3681475016454948e-05, "loss": 2.6984, "step": 37836 }, { "epoch": 2.348811223539636, "grad_norm": 0.47742406166361184, "learning_rate": 1.3678992898524934e-05, "loss": 2.6459, "step": 37837 }, { "epoch": 2.348873300639394, "grad_norm": 0.5460544697573969, "learning_rate": 1.3676510970090445e-05, "loss": 2.6717, "step": 37838 }, { "epoch": 2.348935377739152, "grad_norm": 0.46844386594593557, "learning_rate": 1.3674029231164415e-05, "loss": 2.81, "step": 37839 }, { "epoch": 2.34899745483891, "grad_norm": 0.461347107008201, "learning_rate": 1.3671547681759794e-05, "loss": 2.6526, "step": 37840 }, { "epoch": 2.349059531938668, "grad_norm": 0.49659622457169905, "learning_rate": 1.3669066321889535e-05, "loss": 2.7225, "step": 37841 }, { "epoch": 2.3491216090384257, "grad_norm": 0.4422524872866781, "learning_rate": 1.3666585151566552e-05, "loss": 2.6319, "step": 37842 }, { "epoch": 2.3491836861381836, "grad_norm": 0.4556925967847985, "learning_rate": 1.3664104170803832e-05, "loss": 2.6406, "step": 37843 }, { "epoch": 2.3492457632379415, "grad_norm": 0.49567255778115793, "learning_rate": 1.3661623379614303e-05, "loss": 2.8078, "step": 37844 }, { "epoch": 2.3493078403376995, "grad_norm": 0.4803926361086555, "learning_rate": 1.3659142778010908e-05, "loss": 2.634, "step": 37845 }, { "epoch": 2.3493699174374574, "grad_norm": 0.4636428611824089, "learning_rate": 1.3656662366006567e-05, "loss": 2.7219, "step": 37846 }, { "epoch": 2.3494319945372153, "grad_norm": 0.4604831654726045, "learning_rate": 1.3654182143614263e-05, "loss": 2.6686, "step": 37847 }, { "epoch": 2.349494071636973, "grad_norm": 0.5140561599108577, "learning_rate": 1.365170211084691e-05, "loss": 2.6185, "step": 37848 }, { "epoch": 2.349556148736731, "grad_norm": 0.5205478617020298, "learning_rate": 1.3649222267717448e-05, "loss": 2.7425, "step": 37849 }, { "epoch": 2.349618225836489, "grad_norm": 0.45153498384266605, "learning_rate": 1.3646742614238806e-05, "loss": 2.6562, "step": 37850 }, { "epoch": 2.349680302936247, "grad_norm": 0.5413729592749237, "learning_rate": 1.3644263150423952e-05, "loss": 2.6513, "step": 37851 }, { "epoch": 2.349742380036005, "grad_norm": 0.5252775049211688, "learning_rate": 1.3641783876285796e-05, "loss": 2.6358, "step": 37852 }, { "epoch": 2.349804457135763, "grad_norm": 0.4705472664385717, "learning_rate": 1.3639304791837281e-05, "loss": 2.6799, "step": 37853 }, { "epoch": 2.3498665342355203, "grad_norm": 0.5402933249923844, "learning_rate": 1.3636825897091344e-05, "loss": 2.6281, "step": 37854 }, { "epoch": 2.3499286113352786, "grad_norm": 0.5267877544058766, "learning_rate": 1.3634347192060898e-05, "loss": 2.7345, "step": 37855 }, { "epoch": 2.349990688435036, "grad_norm": 0.4655975032854568, "learning_rate": 1.3631868676758903e-05, "loss": 2.7105, "step": 37856 }, { "epoch": 2.350052765534794, "grad_norm": 0.49918633058735057, "learning_rate": 1.3629390351198279e-05, "loss": 2.7049, "step": 37857 }, { "epoch": 2.350114842634552, "grad_norm": 0.4990245124820246, "learning_rate": 1.3626912215391952e-05, "loss": 2.6146, "step": 37858 }, { "epoch": 2.35017691973431, "grad_norm": 0.4739986933476912, "learning_rate": 1.3624434269352843e-05, "loss": 2.7264, "step": 37859 }, { "epoch": 2.350238996834068, "grad_norm": 0.5115933446609339, "learning_rate": 1.3621956513093903e-05, "loss": 2.7664, "step": 37860 }, { "epoch": 2.3503010739338257, "grad_norm": 0.5260953183401426, "learning_rate": 1.361947894662805e-05, "loss": 2.6983, "step": 37861 }, { "epoch": 2.3503631510335836, "grad_norm": 0.43017223066581495, "learning_rate": 1.3617001569968202e-05, "loss": 2.646, "step": 37862 }, { "epoch": 2.3504252281333415, "grad_norm": 0.4858210421043954, "learning_rate": 1.361452438312728e-05, "loss": 2.7377, "step": 37863 }, { "epoch": 2.3504873052330995, "grad_norm": 0.5053074303669364, "learning_rate": 1.361204738611821e-05, "loss": 2.5832, "step": 37864 }, { "epoch": 2.3505493823328574, "grad_norm": 0.4551369788624162, "learning_rate": 1.3609570578953945e-05, "loss": 2.6191, "step": 37865 }, { "epoch": 2.3506114594326153, "grad_norm": 0.47558042299805997, "learning_rate": 1.3607093961647382e-05, "loss": 2.6812, "step": 37866 }, { "epoch": 2.350673536532373, "grad_norm": 0.444484828274899, "learning_rate": 1.3604617534211439e-05, "loss": 2.6616, "step": 37867 }, { "epoch": 2.350735613632131, "grad_norm": 0.4315073834690393, "learning_rate": 1.360214129665905e-05, "loss": 2.6516, "step": 37868 }, { "epoch": 2.350797690731889, "grad_norm": 0.47664321153405076, "learning_rate": 1.3599665249003108e-05, "loss": 2.6891, "step": 37869 }, { "epoch": 2.350859767831647, "grad_norm": 0.4232802123560109, "learning_rate": 1.3597189391256559e-05, "loss": 2.6387, "step": 37870 }, { "epoch": 2.350921844931405, "grad_norm": 0.4896619273740398, "learning_rate": 1.3594713723432317e-05, "loss": 2.7098, "step": 37871 }, { "epoch": 2.350983922031163, "grad_norm": 0.48123851613796753, "learning_rate": 1.3592238245543287e-05, "loss": 2.6468, "step": 37872 }, { "epoch": 2.3510459991309207, "grad_norm": 0.4677120160451332, "learning_rate": 1.3589762957602376e-05, "loss": 2.7035, "step": 37873 }, { "epoch": 2.3511080762306786, "grad_norm": 0.46645286684920717, "learning_rate": 1.3587287859622522e-05, "loss": 2.6954, "step": 37874 }, { "epoch": 2.3511701533304366, "grad_norm": 0.5115400686688326, "learning_rate": 1.3584812951616627e-05, "loss": 2.6395, "step": 37875 }, { "epoch": 2.3512322304301945, "grad_norm": 0.4912388909275294, "learning_rate": 1.3582338233597608e-05, "loss": 2.7097, "step": 37876 }, { "epoch": 2.3512943075299524, "grad_norm": 0.4282964210091739, "learning_rate": 1.3579863705578366e-05, "loss": 2.5798, "step": 37877 }, { "epoch": 2.3513563846297103, "grad_norm": 0.49237400638705714, "learning_rate": 1.3577389367571802e-05, "loss": 2.6988, "step": 37878 }, { "epoch": 2.351418461729468, "grad_norm": 0.548413510354885, "learning_rate": 1.3574915219590856e-05, "loss": 2.6754, "step": 37879 }, { "epoch": 2.351480538829226, "grad_norm": 0.45459636679454946, "learning_rate": 1.3572441261648416e-05, "loss": 2.5775, "step": 37880 }, { "epoch": 2.3515426159289836, "grad_norm": 0.49055189511348424, "learning_rate": 1.3569967493757401e-05, "loss": 2.6177, "step": 37881 }, { "epoch": 2.3516046930287415, "grad_norm": 0.49313899816895773, "learning_rate": 1.356749391593069e-05, "loss": 2.6541, "step": 37882 }, { "epoch": 2.3516667701284995, "grad_norm": 0.4861622109878643, "learning_rate": 1.356502052818122e-05, "loss": 2.6705, "step": 37883 }, { "epoch": 2.3517288472282574, "grad_norm": 0.47392287663171595, "learning_rate": 1.3562547330521885e-05, "loss": 2.7327, "step": 37884 }, { "epoch": 2.3517909243280153, "grad_norm": 0.5564230778202776, "learning_rate": 1.3560074322965589e-05, "loss": 2.7134, "step": 37885 }, { "epoch": 2.351853001427773, "grad_norm": 0.443311478636622, "learning_rate": 1.3557601505525214e-05, "loss": 2.722, "step": 37886 }, { "epoch": 2.351915078527531, "grad_norm": 0.49587184719028665, "learning_rate": 1.355512887821369e-05, "loss": 2.6457, "step": 37887 }, { "epoch": 2.351977155627289, "grad_norm": 0.4912293508426774, "learning_rate": 1.3552656441043909e-05, "loss": 2.5952, "step": 37888 }, { "epoch": 2.352039232727047, "grad_norm": 0.5033520452808536, "learning_rate": 1.355018419402877e-05, "loss": 2.7374, "step": 37889 }, { "epoch": 2.352101309826805, "grad_norm": 0.5102930665389308, "learning_rate": 1.3547712137181162e-05, "loss": 2.6826, "step": 37890 }, { "epoch": 2.352163386926563, "grad_norm": 0.46497362390180524, "learning_rate": 1.3545240270513981e-05, "loss": 2.6307, "step": 37891 }, { "epoch": 2.3522254640263207, "grad_norm": 0.48715481335876354, "learning_rate": 1.354276859404014e-05, "loss": 2.6507, "step": 37892 }, { "epoch": 2.3522875411260786, "grad_norm": 0.505380204712936, "learning_rate": 1.3540297107772532e-05, "loss": 2.7125, "step": 37893 }, { "epoch": 2.3523496182258365, "grad_norm": 0.48159485945259134, "learning_rate": 1.3537825811724036e-05, "loss": 2.6058, "step": 37894 }, { "epoch": 2.3524116953255945, "grad_norm": 0.5291289735370459, "learning_rate": 1.353535470590755e-05, "loss": 2.7046, "step": 37895 }, { "epoch": 2.3524737724253524, "grad_norm": 0.4881538377466217, "learning_rate": 1.3532883790335966e-05, "loss": 2.7323, "step": 37896 }, { "epoch": 2.3525358495251103, "grad_norm": 0.49810168394804677, "learning_rate": 1.353041306502219e-05, "loss": 2.6717, "step": 37897 }, { "epoch": 2.352597926624868, "grad_norm": 0.5270311912810278, "learning_rate": 1.352794252997911e-05, "loss": 2.7482, "step": 37898 }, { "epoch": 2.352660003724626, "grad_norm": 0.4963651350343226, "learning_rate": 1.3525472185219601e-05, "loss": 2.6365, "step": 37899 }, { "epoch": 2.352722080824384, "grad_norm": 0.5149045592822787, "learning_rate": 1.3523002030756549e-05, "loss": 2.6765, "step": 37900 }, { "epoch": 2.352784157924142, "grad_norm": 0.5538609753871117, "learning_rate": 1.352053206660286e-05, "loss": 2.6392, "step": 37901 }, { "epoch": 2.3528462350238994, "grad_norm": 0.49548703737654254, "learning_rate": 1.3518062292771417e-05, "loss": 2.7035, "step": 37902 }, { "epoch": 2.352908312123658, "grad_norm": 0.5391532710954939, "learning_rate": 1.3515592709275093e-05, "loss": 2.6839, "step": 37903 }, { "epoch": 2.3529703892234153, "grad_norm": 0.4377946882960842, "learning_rate": 1.351312331612678e-05, "loss": 2.6868, "step": 37904 }, { "epoch": 2.353032466323173, "grad_norm": 0.5128636256459822, "learning_rate": 1.351065411333935e-05, "loss": 2.7332, "step": 37905 }, { "epoch": 2.353094543422931, "grad_norm": 0.5022416231986317, "learning_rate": 1.3508185100925708e-05, "loss": 2.5706, "step": 37906 }, { "epoch": 2.353156620522689, "grad_norm": 0.445731200421353, "learning_rate": 1.3505716278898722e-05, "loss": 2.6041, "step": 37907 }, { "epoch": 2.353218697622447, "grad_norm": 0.5031290196323349, "learning_rate": 1.3503247647271272e-05, "loss": 2.7447, "step": 37908 }, { "epoch": 2.353280774722205, "grad_norm": 0.45381977813709734, "learning_rate": 1.350077920605623e-05, "loss": 2.6817, "step": 37909 }, { "epoch": 2.353342851821963, "grad_norm": 0.4692722272552325, "learning_rate": 1.3498310955266492e-05, "loss": 2.6283, "step": 37910 }, { "epoch": 2.3534049289217207, "grad_norm": 0.4659995447983381, "learning_rate": 1.3495842894914929e-05, "loss": 2.5986, "step": 37911 }, { "epoch": 2.3534670060214786, "grad_norm": 0.5028791976324921, "learning_rate": 1.3493375025014415e-05, "loss": 2.6618, "step": 37912 }, { "epoch": 2.3535290831212365, "grad_norm": 0.45779371646195244, "learning_rate": 1.349090734557783e-05, "loss": 2.6773, "step": 37913 }, { "epoch": 2.3535911602209945, "grad_norm": 0.4341017522290571, "learning_rate": 1.3488439856618024e-05, "loss": 2.5697, "step": 37914 }, { "epoch": 2.3536532373207524, "grad_norm": 0.49137252364387385, "learning_rate": 1.3485972558147908e-05, "loss": 2.6346, "step": 37915 }, { "epoch": 2.3537153144205103, "grad_norm": 0.5107000563332231, "learning_rate": 1.3483505450180333e-05, "loss": 2.686, "step": 37916 }, { "epoch": 2.353777391520268, "grad_norm": 0.4533013132330459, "learning_rate": 1.348103853272818e-05, "loss": 2.6653, "step": 37917 }, { "epoch": 2.353839468620026, "grad_norm": 0.4597790757434248, "learning_rate": 1.3478571805804297e-05, "loss": 2.6056, "step": 37918 }, { "epoch": 2.353901545719784, "grad_norm": 0.5044717203405026, "learning_rate": 1.3476105269421585e-05, "loss": 2.6453, "step": 37919 }, { "epoch": 2.353963622819542, "grad_norm": 0.48115560237461624, "learning_rate": 1.3473638923592901e-05, "loss": 2.6568, "step": 37920 }, { "epoch": 2.3540256999193, "grad_norm": 0.4744232660549648, "learning_rate": 1.3471172768331109e-05, "loss": 2.8301, "step": 37921 }, { "epoch": 2.354087777019058, "grad_norm": 0.5215348065739763, "learning_rate": 1.3468706803649062e-05, "loss": 2.7251, "step": 37922 }, { "epoch": 2.3541498541188157, "grad_norm": 0.46762507879322307, "learning_rate": 1.3466241029559651e-05, "loss": 2.6886, "step": 37923 }, { "epoch": 2.3542119312185736, "grad_norm": 0.49230966111609625, "learning_rate": 1.3463775446075732e-05, "loss": 2.7952, "step": 37924 }, { "epoch": 2.3542740083183316, "grad_norm": 0.5030528982399943, "learning_rate": 1.346131005321017e-05, "loss": 2.7214, "step": 37925 }, { "epoch": 2.3543360854180895, "grad_norm": 0.47807386598604645, "learning_rate": 1.3458844850975815e-05, "loss": 2.6498, "step": 37926 }, { "epoch": 2.354398162517847, "grad_norm": 0.4917792034534428, "learning_rate": 1.345637983938553e-05, "loss": 2.6837, "step": 37927 }, { "epoch": 2.3544602396176053, "grad_norm": 0.525816620633171, "learning_rate": 1.3453915018452195e-05, "loss": 2.7066, "step": 37928 }, { "epoch": 2.354522316717363, "grad_norm": 0.5061311028815181, "learning_rate": 1.3451450388188642e-05, "loss": 2.6884, "step": 37929 }, { "epoch": 2.3545843938171207, "grad_norm": 0.5188976525820957, "learning_rate": 1.3448985948607756e-05, "loss": 2.7361, "step": 37930 }, { "epoch": 2.3546464709168786, "grad_norm": 0.4890926790523668, "learning_rate": 1.3446521699722386e-05, "loss": 2.6567, "step": 37931 }, { "epoch": 2.3547085480166365, "grad_norm": 0.4924855162577276, "learning_rate": 1.3444057641545376e-05, "loss": 2.725, "step": 37932 }, { "epoch": 2.3547706251163945, "grad_norm": 0.5035792627952316, "learning_rate": 1.34415937740896e-05, "loss": 2.7568, "step": 37933 }, { "epoch": 2.3548327022161524, "grad_norm": 0.4711212738499057, "learning_rate": 1.3439130097367908e-05, "loss": 2.6338, "step": 37934 }, { "epoch": 2.3548947793159103, "grad_norm": 0.4864461671286715, "learning_rate": 1.3436666611393145e-05, "loss": 2.6398, "step": 37935 }, { "epoch": 2.354956856415668, "grad_norm": 0.5065031631329533, "learning_rate": 1.3434203316178157e-05, "loss": 2.7538, "step": 37936 }, { "epoch": 2.355018933515426, "grad_norm": 0.5108958609409784, "learning_rate": 1.3431740211735821e-05, "loss": 2.6918, "step": 37937 }, { "epoch": 2.355081010615184, "grad_norm": 0.49206149289660295, "learning_rate": 1.3429277298078979e-05, "loss": 2.7032, "step": 37938 }, { "epoch": 2.355143087714942, "grad_norm": 0.5182822278588276, "learning_rate": 1.3426814575220469e-05, "loss": 2.7058, "step": 37939 }, { "epoch": 2.3552051648147, "grad_norm": 0.5136622148773068, "learning_rate": 1.3424352043173144e-05, "loss": 2.7442, "step": 37940 }, { "epoch": 2.355267241914458, "grad_norm": 0.4858189870680239, "learning_rate": 1.3421889701949842e-05, "loss": 2.7287, "step": 37941 }, { "epoch": 2.3553293190142157, "grad_norm": 0.45371431636003745, "learning_rate": 1.3419427551563435e-05, "loss": 2.6791, "step": 37942 }, { "epoch": 2.3553913961139736, "grad_norm": 0.47243291133847953, "learning_rate": 1.341696559202676e-05, "loss": 2.6991, "step": 37943 }, { "epoch": 2.3554534732137316, "grad_norm": 0.5009108829387563, "learning_rate": 1.3414503823352647e-05, "loss": 2.691, "step": 37944 }, { "epoch": 2.3555155503134895, "grad_norm": 0.4772639037685232, "learning_rate": 1.3412042245553941e-05, "loss": 2.6765, "step": 37945 }, { "epoch": 2.3555776274132474, "grad_norm": 0.47139236829306935, "learning_rate": 1.3409580858643505e-05, "loss": 2.6768, "step": 37946 }, { "epoch": 2.3556397045130053, "grad_norm": 0.4966810909705935, "learning_rate": 1.340711966263417e-05, "loss": 2.7149, "step": 37947 }, { "epoch": 2.3557017816127632, "grad_norm": 0.4663519874074786, "learning_rate": 1.340465865753877e-05, "loss": 2.6781, "step": 37948 }, { "epoch": 2.355763858712521, "grad_norm": 0.5129807032140338, "learning_rate": 1.340219784337014e-05, "loss": 2.6644, "step": 37949 }, { "epoch": 2.3558259358122786, "grad_norm": 0.5523257143348478, "learning_rate": 1.339973722014114e-05, "loss": 2.7645, "step": 37950 }, { "epoch": 2.355888012912037, "grad_norm": 0.5215433780280133, "learning_rate": 1.3397276787864598e-05, "loss": 2.6561, "step": 37951 }, { "epoch": 2.3559500900117945, "grad_norm": 0.5015292623089389, "learning_rate": 1.3394816546553352e-05, "loss": 2.6275, "step": 37952 }, { "epoch": 2.3560121671115524, "grad_norm": 0.47867592442690665, "learning_rate": 1.3392356496220226e-05, "loss": 2.715, "step": 37953 }, { "epoch": 2.3560742442113103, "grad_norm": 0.45587613778564623, "learning_rate": 1.3389896636878052e-05, "loss": 2.6936, "step": 37954 }, { "epoch": 2.356136321311068, "grad_norm": 0.44317295339583807, "learning_rate": 1.338743696853969e-05, "loss": 2.6653, "step": 37955 }, { "epoch": 2.356198398410826, "grad_norm": 0.5282230252366726, "learning_rate": 1.3384977491217959e-05, "loss": 2.6539, "step": 37956 }, { "epoch": 2.356260475510584, "grad_norm": 0.39298098565954886, "learning_rate": 1.3382518204925686e-05, "loss": 2.5579, "step": 37957 }, { "epoch": 2.356322552610342, "grad_norm": 0.4617928795600531, "learning_rate": 1.3380059109675692e-05, "loss": 2.6079, "step": 37958 }, { "epoch": 2.3563846297101, "grad_norm": 0.520303195721257, "learning_rate": 1.3377600205480834e-05, "loss": 2.6665, "step": 37959 }, { "epoch": 2.356446706809858, "grad_norm": 0.4567618921231964, "learning_rate": 1.3375141492353927e-05, "loss": 2.6876, "step": 37960 }, { "epoch": 2.3565087839096157, "grad_norm": 0.4809121512759216, "learning_rate": 1.3372682970307792e-05, "loss": 2.6194, "step": 37961 }, { "epoch": 2.3565708610093736, "grad_norm": 0.5204218637814204, "learning_rate": 1.3370224639355256e-05, "loss": 2.6832, "step": 37962 }, { "epoch": 2.3566329381091315, "grad_norm": 0.48818805568157, "learning_rate": 1.3367766499509154e-05, "loss": 2.6431, "step": 37963 }, { "epoch": 2.3566950152088895, "grad_norm": 0.4291447940209314, "learning_rate": 1.3365308550782302e-05, "loss": 2.6913, "step": 37964 }, { "epoch": 2.3567570923086474, "grad_norm": 0.5281324352215035, "learning_rate": 1.3362850793187537e-05, "loss": 2.6975, "step": 37965 }, { "epoch": 2.3568191694084053, "grad_norm": 0.47840947122029487, "learning_rate": 1.3360393226737678e-05, "loss": 2.6396, "step": 37966 }, { "epoch": 2.356881246508163, "grad_norm": 0.44020903953619567, "learning_rate": 1.335793585144554e-05, "loss": 2.6855, "step": 37967 }, { "epoch": 2.356943323607921, "grad_norm": 0.5051546232112952, "learning_rate": 1.335547866732393e-05, "loss": 2.6978, "step": 37968 }, { "epoch": 2.357005400707679, "grad_norm": 0.44989082813533265, "learning_rate": 1.3353021674385701e-05, "loss": 2.6215, "step": 37969 }, { "epoch": 2.357067477807437, "grad_norm": 0.46688077483888935, "learning_rate": 1.3350564872643651e-05, "loss": 2.703, "step": 37970 }, { "epoch": 2.357129554907195, "grad_norm": 0.485219296522208, "learning_rate": 1.3348108262110604e-05, "loss": 2.7783, "step": 37971 }, { "epoch": 2.357191632006953, "grad_norm": 0.43192518356729137, "learning_rate": 1.3345651842799356e-05, "loss": 2.7655, "step": 37972 }, { "epoch": 2.3572537091067107, "grad_norm": 0.4082150252218358, "learning_rate": 1.334319561472276e-05, "loss": 2.5703, "step": 37973 }, { "epoch": 2.3573157862064686, "grad_norm": 0.4820694041183688, "learning_rate": 1.3340739577893608e-05, "loss": 2.6666, "step": 37974 }, { "epoch": 2.357377863306226, "grad_norm": 0.48811244335459975, "learning_rate": 1.3338283732324719e-05, "loss": 2.7165, "step": 37975 }, { "epoch": 2.3574399404059845, "grad_norm": 0.4431697086867279, "learning_rate": 1.33358280780289e-05, "loss": 2.6744, "step": 37976 }, { "epoch": 2.357502017505742, "grad_norm": 0.49365258523423866, "learning_rate": 1.3333372615018953e-05, "loss": 2.6718, "step": 37977 }, { "epoch": 2.3575640946055, "grad_norm": 0.481461282894663, "learning_rate": 1.333091734330772e-05, "loss": 2.737, "step": 37978 }, { "epoch": 2.357626171705258, "grad_norm": 0.44234660086728755, "learning_rate": 1.3328462262907992e-05, "loss": 2.7514, "step": 37979 }, { "epoch": 2.3576882488050157, "grad_norm": 0.4666457789577292, "learning_rate": 1.3326007373832577e-05, "loss": 2.6463, "step": 37980 }, { "epoch": 2.3577503259047736, "grad_norm": 0.468351846718754, "learning_rate": 1.332355267609427e-05, "loss": 2.6649, "step": 37981 }, { "epoch": 2.3578124030045315, "grad_norm": 0.5298875560330234, "learning_rate": 1.3321098169705904e-05, "loss": 2.631, "step": 37982 }, { "epoch": 2.3578744801042895, "grad_norm": 0.48564774439117153, "learning_rate": 1.331864385468028e-05, "loss": 2.7067, "step": 37983 }, { "epoch": 2.3579365572040474, "grad_norm": 0.4647179761999051, "learning_rate": 1.3316189731030187e-05, "loss": 2.6366, "step": 37984 }, { "epoch": 2.3579986343038053, "grad_norm": 0.48795455684779765, "learning_rate": 1.3313735798768429e-05, "loss": 2.7182, "step": 37985 }, { "epoch": 2.358060711403563, "grad_norm": 0.5184468015235307, "learning_rate": 1.331128205790783e-05, "loss": 2.8115, "step": 37986 }, { "epoch": 2.358122788503321, "grad_norm": 0.488573200683021, "learning_rate": 1.330882850846118e-05, "loss": 2.6499, "step": 37987 }, { "epoch": 2.358184865603079, "grad_norm": 0.4660091057550831, "learning_rate": 1.330637515044128e-05, "loss": 2.6789, "step": 37988 }, { "epoch": 2.358246942702837, "grad_norm": 0.4733066812447482, "learning_rate": 1.3303921983860922e-05, "loss": 2.5766, "step": 37989 }, { "epoch": 2.358309019802595, "grad_norm": 0.5488250297920672, "learning_rate": 1.3301469008732903e-05, "loss": 2.6276, "step": 37990 }, { "epoch": 2.358371096902353, "grad_norm": 0.43265152702357434, "learning_rate": 1.3299016225070044e-05, "loss": 2.7174, "step": 37991 }, { "epoch": 2.3584331740021107, "grad_norm": 0.47939964691124437, "learning_rate": 1.3296563632885123e-05, "loss": 2.7437, "step": 37992 }, { "epoch": 2.3584952511018686, "grad_norm": 0.5122407187556095, "learning_rate": 1.3294111232190942e-05, "loss": 2.6754, "step": 37993 }, { "epoch": 2.3585573282016266, "grad_norm": 0.4593566034487014, "learning_rate": 1.329165902300028e-05, "loss": 2.6681, "step": 37994 }, { "epoch": 2.3586194053013845, "grad_norm": 0.4842045237890813, "learning_rate": 1.3289207005325944e-05, "loss": 2.6013, "step": 37995 }, { "epoch": 2.3586814824011424, "grad_norm": 0.46656578121577313, "learning_rate": 1.3286755179180738e-05, "loss": 2.7483, "step": 37996 }, { "epoch": 2.3587435595009003, "grad_norm": 0.42714064686477193, "learning_rate": 1.3284303544577442e-05, "loss": 2.7069, "step": 37997 }, { "epoch": 2.358805636600658, "grad_norm": 0.472891457413729, "learning_rate": 1.3281852101528851e-05, "loss": 2.6451, "step": 37998 }, { "epoch": 2.358867713700416, "grad_norm": 0.4479366915314826, "learning_rate": 1.3279400850047734e-05, "loss": 2.6884, "step": 37999 }, { "epoch": 2.3589297908001736, "grad_norm": 0.43519751046679167, "learning_rate": 1.3276949790146915e-05, "loss": 2.7152, "step": 38000 }, { "epoch": 2.3589918678999315, "grad_norm": 0.4475366196220686, "learning_rate": 1.3274498921839163e-05, "loss": 2.698, "step": 38001 }, { "epoch": 2.3590539449996895, "grad_norm": 0.4547830861351469, "learning_rate": 1.3272048245137265e-05, "loss": 2.681, "step": 38002 }, { "epoch": 2.3591160220994474, "grad_norm": 0.47444352442856674, "learning_rate": 1.326959776005401e-05, "loss": 2.6198, "step": 38003 }, { "epoch": 2.3591780991992053, "grad_norm": 0.43954873634376307, "learning_rate": 1.3267147466602159e-05, "loss": 2.7247, "step": 38004 }, { "epoch": 2.359240176298963, "grad_norm": 0.451012455241729, "learning_rate": 1.3264697364794538e-05, "loss": 2.7033, "step": 38005 }, { "epoch": 2.359302253398721, "grad_norm": 0.4675310179676219, "learning_rate": 1.32622474546439e-05, "loss": 2.7341, "step": 38006 }, { "epoch": 2.359364330498479, "grad_norm": 0.4483642245780004, "learning_rate": 1.3259797736163043e-05, "loss": 2.5343, "step": 38007 }, { "epoch": 2.359426407598237, "grad_norm": 0.49825485973026623, "learning_rate": 1.3257348209364722e-05, "loss": 2.6186, "step": 38008 }, { "epoch": 2.359488484697995, "grad_norm": 0.4459846844267299, "learning_rate": 1.325489887426175e-05, "loss": 2.7135, "step": 38009 }, { "epoch": 2.359550561797753, "grad_norm": 0.44768155011488125, "learning_rate": 1.3252449730866894e-05, "loss": 2.669, "step": 38010 }, { "epoch": 2.3596126388975107, "grad_norm": 0.4649577121323099, "learning_rate": 1.3250000779192928e-05, "loss": 2.6904, "step": 38011 }, { "epoch": 2.3596747159972686, "grad_norm": 0.45781153721106, "learning_rate": 1.3247552019252629e-05, "loss": 2.6967, "step": 38012 }, { "epoch": 2.3597367930970266, "grad_norm": 0.4525878122650451, "learning_rate": 1.3245103451058754e-05, "loss": 2.7734, "step": 38013 }, { "epoch": 2.3597988701967845, "grad_norm": 0.4575931160131834, "learning_rate": 1.3242655074624116e-05, "loss": 2.7841, "step": 38014 }, { "epoch": 2.3598609472965424, "grad_norm": 0.44615977812439006, "learning_rate": 1.324020688996147e-05, "loss": 2.6741, "step": 38015 }, { "epoch": 2.3599230243963003, "grad_norm": 0.48189890078632786, "learning_rate": 1.3237758897083586e-05, "loss": 2.6515, "step": 38016 }, { "epoch": 2.3599851014960582, "grad_norm": 0.5084706076807928, "learning_rate": 1.3235311096003223e-05, "loss": 2.6843, "step": 38017 }, { "epoch": 2.360047178595816, "grad_norm": 0.4922323811973261, "learning_rate": 1.3232863486733182e-05, "loss": 2.6323, "step": 38018 }, { "epoch": 2.360109255695574, "grad_norm": 0.49541968259970864, "learning_rate": 1.3230416069286223e-05, "loss": 2.7827, "step": 38019 }, { "epoch": 2.360171332795332, "grad_norm": 0.4424799460915447, "learning_rate": 1.3227968843675104e-05, "loss": 2.6625, "step": 38020 }, { "epoch": 2.36023340989509, "grad_norm": 0.5135457569395314, "learning_rate": 1.3225521809912584e-05, "loss": 2.6842, "step": 38021 }, { "epoch": 2.360295486994848, "grad_norm": 0.45668750453775125, "learning_rate": 1.3223074968011456e-05, "loss": 2.6509, "step": 38022 }, { "epoch": 2.3603575640946053, "grad_norm": 0.45289653740656516, "learning_rate": 1.322062831798448e-05, "loss": 2.7583, "step": 38023 }, { "epoch": 2.3604196411943636, "grad_norm": 0.4666644910211029, "learning_rate": 1.3218181859844408e-05, "loss": 2.5809, "step": 38024 }, { "epoch": 2.360481718294121, "grad_norm": 0.4815919354927941, "learning_rate": 1.3215735593604017e-05, "loss": 2.7172, "step": 38025 }, { "epoch": 2.360543795393879, "grad_norm": 0.4327123980667193, "learning_rate": 1.3213289519276045e-05, "loss": 2.7732, "step": 38026 }, { "epoch": 2.360605872493637, "grad_norm": 0.5028140036855041, "learning_rate": 1.3210843636873272e-05, "loss": 2.6448, "step": 38027 }, { "epoch": 2.360667949593395, "grad_norm": 0.46112454385546164, "learning_rate": 1.3208397946408474e-05, "loss": 2.7799, "step": 38028 }, { "epoch": 2.360730026693153, "grad_norm": 0.4582397101610665, "learning_rate": 1.3205952447894393e-05, "loss": 2.6597, "step": 38029 }, { "epoch": 2.3607921037929107, "grad_norm": 0.4678448361756261, "learning_rate": 1.3203507141343796e-05, "loss": 2.685, "step": 38030 }, { "epoch": 2.3608541808926686, "grad_norm": 0.47335577659164707, "learning_rate": 1.3201062026769417e-05, "loss": 2.6462, "step": 38031 }, { "epoch": 2.3609162579924265, "grad_norm": 0.4694212081605291, "learning_rate": 1.3198617104184042e-05, "loss": 2.6099, "step": 38032 }, { "epoch": 2.3609783350921845, "grad_norm": 0.5193303228779916, "learning_rate": 1.3196172373600418e-05, "loss": 2.704, "step": 38033 }, { "epoch": 2.3610404121919424, "grad_norm": 0.4712758991602185, "learning_rate": 1.31937278350313e-05, "loss": 2.621, "step": 38034 }, { "epoch": 2.3611024892917003, "grad_norm": 0.48883438823309533, "learning_rate": 1.319128348848942e-05, "loss": 2.6952, "step": 38035 }, { "epoch": 2.361164566391458, "grad_norm": 0.48199647997524486, "learning_rate": 1.3188839333987568e-05, "loss": 2.7152, "step": 38036 }, { "epoch": 2.361226643491216, "grad_norm": 0.5049811143931856, "learning_rate": 1.3186395371538474e-05, "loss": 2.6646, "step": 38037 }, { "epoch": 2.361288720590974, "grad_norm": 0.4842128174508544, "learning_rate": 1.3183951601154892e-05, "loss": 2.5951, "step": 38038 }, { "epoch": 2.361350797690732, "grad_norm": 0.4598953783487364, "learning_rate": 1.3181508022849575e-05, "loss": 2.7057, "step": 38039 }, { "epoch": 2.36141287479049, "grad_norm": 0.5286821124276634, "learning_rate": 1.3179064636635252e-05, "loss": 2.6862, "step": 38040 }, { "epoch": 2.361474951890248, "grad_norm": 0.48069512841541784, "learning_rate": 1.31766214425247e-05, "loss": 2.7276, "step": 38041 }, { "epoch": 2.3615370289900057, "grad_norm": 0.4331065181130273, "learning_rate": 1.317417844053066e-05, "loss": 2.7067, "step": 38042 }, { "epoch": 2.3615991060897636, "grad_norm": 0.4616095767229757, "learning_rate": 1.3171735630665866e-05, "loss": 2.6234, "step": 38043 }, { "epoch": 2.3616611831895216, "grad_norm": 0.49010429328270905, "learning_rate": 1.3169293012943052e-05, "loss": 2.6321, "step": 38044 }, { "epoch": 2.3617232602892795, "grad_norm": 0.44811579432801557, "learning_rate": 1.316685058737499e-05, "loss": 2.7393, "step": 38045 }, { "epoch": 2.361785337389037, "grad_norm": 0.5145061665435126, "learning_rate": 1.3164408353974417e-05, "loss": 2.6502, "step": 38046 }, { "epoch": 2.3618474144887953, "grad_norm": 0.5052593900175104, "learning_rate": 1.3161966312754065e-05, "loss": 2.6502, "step": 38047 }, { "epoch": 2.361909491588553, "grad_norm": 0.44267224744441974, "learning_rate": 1.3159524463726675e-05, "loss": 2.7142, "step": 38048 }, { "epoch": 2.3619715686883107, "grad_norm": 0.47009473873055496, "learning_rate": 1.3157082806904974e-05, "loss": 2.7103, "step": 38049 }, { "epoch": 2.3620336457880686, "grad_norm": 0.5489813702676022, "learning_rate": 1.3154641342301733e-05, "loss": 2.7325, "step": 38050 }, { "epoch": 2.3620957228878265, "grad_norm": 0.4824950745682016, "learning_rate": 1.3152200069929677e-05, "loss": 2.5741, "step": 38051 }, { "epoch": 2.3621577999875845, "grad_norm": 0.47268984021085814, "learning_rate": 1.314975898980153e-05, "loss": 2.5559, "step": 38052 }, { "epoch": 2.3622198770873424, "grad_norm": 0.5589588279404397, "learning_rate": 1.3147318101930028e-05, "loss": 2.6139, "step": 38053 }, { "epoch": 2.3622819541871003, "grad_norm": 0.46578353960612756, "learning_rate": 1.3144877406327921e-05, "loss": 2.6974, "step": 38054 }, { "epoch": 2.362344031286858, "grad_norm": 0.49623808675922665, "learning_rate": 1.3142436903007944e-05, "loss": 2.7427, "step": 38055 }, { "epoch": 2.362406108386616, "grad_norm": 0.4855961480769156, "learning_rate": 1.3139996591982818e-05, "loss": 2.6767, "step": 38056 }, { "epoch": 2.362468185486374, "grad_norm": 0.45552259032094305, "learning_rate": 1.313755647326526e-05, "loss": 2.7215, "step": 38057 }, { "epoch": 2.362530262586132, "grad_norm": 0.4494812405812086, "learning_rate": 1.3135116546868036e-05, "loss": 2.6173, "step": 38058 }, { "epoch": 2.36259233968589, "grad_norm": 0.5187786933133236, "learning_rate": 1.3132676812803856e-05, "loss": 2.7179, "step": 38059 }, { "epoch": 2.362654416785648, "grad_norm": 0.4410713072799242, "learning_rate": 1.3130237271085439e-05, "loss": 2.6343, "step": 38060 }, { "epoch": 2.3627164938854057, "grad_norm": 0.4739660604776255, "learning_rate": 1.3127797921725537e-05, "loss": 2.6556, "step": 38061 }, { "epoch": 2.3627785709851636, "grad_norm": 0.4875954045572821, "learning_rate": 1.3125358764736867e-05, "loss": 2.6991, "step": 38062 }, { "epoch": 2.3628406480849216, "grad_norm": 0.4900912649690877, "learning_rate": 1.3122919800132138e-05, "loss": 2.7035, "step": 38063 }, { "epoch": 2.3629027251846795, "grad_norm": 0.4983184373292077, "learning_rate": 1.3120481027924097e-05, "loss": 2.7553, "step": 38064 }, { "epoch": 2.3629648022844374, "grad_norm": 0.5263897213332349, "learning_rate": 1.3118042448125468e-05, "loss": 2.696, "step": 38065 }, { "epoch": 2.3630268793841953, "grad_norm": 0.48652863869170565, "learning_rate": 1.311560406074896e-05, "loss": 2.6899, "step": 38066 }, { "epoch": 2.3630889564839532, "grad_norm": 0.5259246097918481, "learning_rate": 1.3113165865807286e-05, "loss": 2.6542, "step": 38067 }, { "epoch": 2.363151033583711, "grad_norm": 0.49896999901637734, "learning_rate": 1.3110727863313193e-05, "loss": 2.7503, "step": 38068 }, { "epoch": 2.3632131106834686, "grad_norm": 0.4863493121943802, "learning_rate": 1.3108290053279388e-05, "loss": 2.8013, "step": 38069 }, { "epoch": 2.363275187783227, "grad_norm": 0.4924088691738121, "learning_rate": 1.3105852435718591e-05, "loss": 2.6968, "step": 38070 }, { "epoch": 2.3633372648829845, "grad_norm": 0.48380559862973105, "learning_rate": 1.3103415010643504e-05, "loss": 2.6084, "step": 38071 }, { "epoch": 2.3633993419827424, "grad_norm": 0.5596616979942278, "learning_rate": 1.310097777806687e-05, "loss": 2.6275, "step": 38072 }, { "epoch": 2.3634614190825003, "grad_norm": 0.5119840502061384, "learning_rate": 1.3098540738001397e-05, "loss": 2.7308, "step": 38073 }, { "epoch": 2.363523496182258, "grad_norm": 0.5288876292164055, "learning_rate": 1.3096103890459793e-05, "loss": 2.8363, "step": 38074 }, { "epoch": 2.363585573282016, "grad_norm": 0.4717616134481589, "learning_rate": 1.309366723545477e-05, "loss": 2.7395, "step": 38075 }, { "epoch": 2.363647650381774, "grad_norm": 0.5406224719302654, "learning_rate": 1.3091230772999035e-05, "loss": 2.7272, "step": 38076 }, { "epoch": 2.363709727481532, "grad_norm": 0.5002772933228754, "learning_rate": 1.3088794503105323e-05, "loss": 2.6691, "step": 38077 }, { "epoch": 2.36377180458129, "grad_norm": 0.48705507050688496, "learning_rate": 1.3086358425786326e-05, "loss": 2.6338, "step": 38078 }, { "epoch": 2.363833881681048, "grad_norm": 0.5278715529983063, "learning_rate": 1.3083922541054765e-05, "loss": 2.6933, "step": 38079 }, { "epoch": 2.3638959587808057, "grad_norm": 0.4811170127970791, "learning_rate": 1.3081486848923325e-05, "loss": 2.6238, "step": 38080 }, { "epoch": 2.3639580358805636, "grad_norm": 0.4706395213101942, "learning_rate": 1.3079051349404742e-05, "loss": 2.7123, "step": 38081 }, { "epoch": 2.3640201129803216, "grad_norm": 0.49188059366188974, "learning_rate": 1.3076616042511713e-05, "loss": 2.707, "step": 38082 }, { "epoch": 2.3640821900800795, "grad_norm": 0.47418408903496795, "learning_rate": 1.307418092825694e-05, "loss": 2.666, "step": 38083 }, { "epoch": 2.3641442671798374, "grad_norm": 0.5132972598663593, "learning_rate": 1.3071746006653113e-05, "loss": 2.7677, "step": 38084 }, { "epoch": 2.3642063442795953, "grad_norm": 0.45123729809454544, "learning_rate": 1.306931127771297e-05, "loss": 2.6683, "step": 38085 }, { "epoch": 2.3642684213793532, "grad_norm": 0.4871970996314661, "learning_rate": 1.3066876741449197e-05, "loss": 2.7135, "step": 38086 }, { "epoch": 2.364330498479111, "grad_norm": 0.4517388682469015, "learning_rate": 1.3064442397874487e-05, "loss": 2.6181, "step": 38087 }, { "epoch": 2.364392575578869, "grad_norm": 0.5264892436610609, "learning_rate": 1.3062008247001556e-05, "loss": 2.7042, "step": 38088 }, { "epoch": 2.364454652678627, "grad_norm": 0.5075221772351562, "learning_rate": 1.3059574288843073e-05, "loss": 2.6524, "step": 38089 }, { "epoch": 2.364516729778385, "grad_norm": 0.4761073330831075, "learning_rate": 1.3057140523411775e-05, "loss": 2.6122, "step": 38090 }, { "epoch": 2.364578806878143, "grad_norm": 0.4627760572335976, "learning_rate": 1.3054706950720346e-05, "loss": 2.6672, "step": 38091 }, { "epoch": 2.3646408839779007, "grad_norm": 0.45959484004086415, "learning_rate": 1.3052273570781476e-05, "loss": 2.6932, "step": 38092 }, { "epoch": 2.3647029610776586, "grad_norm": 0.4639186308594041, "learning_rate": 1.3049840383607853e-05, "loss": 2.6876, "step": 38093 }, { "epoch": 2.364765038177416, "grad_norm": 0.46419325762088015, "learning_rate": 1.304740738921218e-05, "loss": 2.6796, "step": 38094 }, { "epoch": 2.3648271152771745, "grad_norm": 0.47843667760936537, "learning_rate": 1.3044974587607167e-05, "loss": 2.7095, "step": 38095 }, { "epoch": 2.364889192376932, "grad_norm": 0.454920893017382, "learning_rate": 1.3042541978805495e-05, "loss": 2.6345, "step": 38096 }, { "epoch": 2.36495126947669, "grad_norm": 0.4533917858444021, "learning_rate": 1.3040109562819853e-05, "loss": 2.6383, "step": 38097 }, { "epoch": 2.365013346576448, "grad_norm": 0.5066150926496094, "learning_rate": 1.3037677339662935e-05, "loss": 2.7262, "step": 38098 }, { "epoch": 2.3650754236762057, "grad_norm": 0.44961228755004246, "learning_rate": 1.3035245309347405e-05, "loss": 2.7399, "step": 38099 }, { "epoch": 2.3651375007759636, "grad_norm": 0.5135804301624524, "learning_rate": 1.3032813471885996e-05, "loss": 2.6941, "step": 38100 }, { "epoch": 2.3651995778757215, "grad_norm": 0.5015509214877081, "learning_rate": 1.3030381827291372e-05, "loss": 2.627, "step": 38101 }, { "epoch": 2.3652616549754795, "grad_norm": 0.48160904582033753, "learning_rate": 1.3027950375576214e-05, "loss": 2.6193, "step": 38102 }, { "epoch": 2.3653237320752374, "grad_norm": 0.527972940979801, "learning_rate": 1.3025519116753205e-05, "loss": 2.6384, "step": 38103 }, { "epoch": 2.3653858091749953, "grad_norm": 0.4822908601966965, "learning_rate": 1.3023088050835053e-05, "loss": 2.5835, "step": 38104 }, { "epoch": 2.365447886274753, "grad_norm": 0.48042270780946433, "learning_rate": 1.3020657177834428e-05, "loss": 2.6711, "step": 38105 }, { "epoch": 2.365509963374511, "grad_norm": 0.4959265960475374, "learning_rate": 1.3018226497764008e-05, "loss": 2.7559, "step": 38106 }, { "epoch": 2.365572040474269, "grad_norm": 0.499648715834741, "learning_rate": 1.3015796010636466e-05, "loss": 2.714, "step": 38107 }, { "epoch": 2.365634117574027, "grad_norm": 0.4478397599073736, "learning_rate": 1.3013365716464505e-05, "loss": 2.735, "step": 38108 }, { "epoch": 2.365696194673785, "grad_norm": 0.5388875483903675, "learning_rate": 1.30109356152608e-05, "loss": 2.6977, "step": 38109 }, { "epoch": 2.365758271773543, "grad_norm": 0.46129888172150557, "learning_rate": 1.3008505707038021e-05, "loss": 2.7768, "step": 38110 }, { "epoch": 2.3658203488733007, "grad_norm": 0.46415089815306215, "learning_rate": 1.3006075991808848e-05, "loss": 2.7332, "step": 38111 }, { "epoch": 2.3658824259730586, "grad_norm": 0.5027663771277561, "learning_rate": 1.300364646958594e-05, "loss": 2.6718, "step": 38112 }, { "epoch": 2.3659445030728166, "grad_norm": 0.48210039122078385, "learning_rate": 1.3001217140382004e-05, "loss": 2.724, "step": 38113 }, { "epoch": 2.3660065801725745, "grad_norm": 0.459303464963913, "learning_rate": 1.2998788004209705e-05, "loss": 2.6904, "step": 38114 }, { "epoch": 2.3660686572723324, "grad_norm": 0.5151190153578252, "learning_rate": 1.2996359061081708e-05, "loss": 2.6679, "step": 38115 }, { "epoch": 2.3661307343720903, "grad_norm": 0.4862127971636591, "learning_rate": 1.299393031101067e-05, "loss": 2.7423, "step": 38116 }, { "epoch": 2.366192811471848, "grad_norm": 0.4886151812238525, "learning_rate": 1.2991501754009299e-05, "loss": 2.6313, "step": 38117 }, { "epoch": 2.366254888571606, "grad_norm": 0.5271852608335255, "learning_rate": 1.298907339009025e-05, "loss": 2.7318, "step": 38118 }, { "epoch": 2.3663169656713636, "grad_norm": 0.4830892942245187, "learning_rate": 1.2986645219266191e-05, "loss": 2.6942, "step": 38119 }, { "epoch": 2.3663790427711215, "grad_norm": 0.5295090197424419, "learning_rate": 1.2984217241549773e-05, "loss": 2.6509, "step": 38120 }, { "epoch": 2.3664411198708795, "grad_norm": 0.4642151822235843, "learning_rate": 1.2981789456953692e-05, "loss": 2.5712, "step": 38121 }, { "epoch": 2.3665031969706374, "grad_norm": 0.5124778068203226, "learning_rate": 1.2979361865490603e-05, "loss": 2.7347, "step": 38122 }, { "epoch": 2.3665652740703953, "grad_norm": 0.5054585993836864, "learning_rate": 1.2976934467173169e-05, "loss": 2.6421, "step": 38123 }, { "epoch": 2.366627351170153, "grad_norm": 0.4990048936955565, "learning_rate": 1.2974507262014057e-05, "loss": 2.6685, "step": 38124 }, { "epoch": 2.366689428269911, "grad_norm": 0.46224952113971074, "learning_rate": 1.2972080250025914e-05, "loss": 2.653, "step": 38125 }, { "epoch": 2.366751505369669, "grad_norm": 0.5152359564050522, "learning_rate": 1.2969653431221418e-05, "loss": 2.6916, "step": 38126 }, { "epoch": 2.366813582469427, "grad_norm": 0.5546899446245508, "learning_rate": 1.2967226805613242e-05, "loss": 2.7584, "step": 38127 }, { "epoch": 2.366875659569185, "grad_norm": 0.4653443437930766, "learning_rate": 1.2964800373214036e-05, "loss": 2.7004, "step": 38128 }, { "epoch": 2.366937736668943, "grad_norm": 0.492156890643945, "learning_rate": 1.2962374134036454e-05, "loss": 2.6385, "step": 38129 }, { "epoch": 2.3669998137687007, "grad_norm": 0.5929924750325043, "learning_rate": 1.2959948088093143e-05, "loss": 2.679, "step": 38130 }, { "epoch": 2.3670618908684586, "grad_norm": 0.47774398245083366, "learning_rate": 1.2957522235396796e-05, "loss": 2.6532, "step": 38131 }, { "epoch": 2.3671239679682166, "grad_norm": 0.45180979956092826, "learning_rate": 1.295509657596004e-05, "loss": 2.6637, "step": 38132 }, { "epoch": 2.3671860450679745, "grad_norm": 0.4956983657740715, "learning_rate": 1.2952671109795544e-05, "loss": 2.7939, "step": 38133 }, { "epoch": 2.3672481221677324, "grad_norm": 0.48228457269195696, "learning_rate": 1.2950245836915953e-05, "loss": 2.6287, "step": 38134 }, { "epoch": 2.3673101992674903, "grad_norm": 0.502783537434184, "learning_rate": 1.2947820757333906e-05, "loss": 2.687, "step": 38135 }, { "epoch": 2.3673722763672482, "grad_norm": 0.4198027863303717, "learning_rate": 1.2945395871062094e-05, "loss": 2.6832, "step": 38136 }, { "epoch": 2.367434353467006, "grad_norm": 0.42888844493047984, "learning_rate": 1.2942971178113145e-05, "loss": 2.6304, "step": 38137 }, { "epoch": 2.367496430566764, "grad_norm": 0.4488071546311947, "learning_rate": 1.294054667849971e-05, "loss": 2.6107, "step": 38138 }, { "epoch": 2.367558507666522, "grad_norm": 0.4217474532703419, "learning_rate": 1.2938122372234423e-05, "loss": 2.6499, "step": 38139 }, { "epoch": 2.36762058476628, "grad_norm": 0.4382716084336057, "learning_rate": 1.2935698259329965e-05, "loss": 2.7065, "step": 38140 }, { "epoch": 2.367682661866038, "grad_norm": 0.4875577372814801, "learning_rate": 1.2933274339798967e-05, "loss": 2.7088, "step": 38141 }, { "epoch": 2.3677447389657953, "grad_norm": 0.4523232865522642, "learning_rate": 1.2930850613654078e-05, "loss": 2.6553, "step": 38142 }, { "epoch": 2.3678068160655537, "grad_norm": 0.45308530567860844, "learning_rate": 1.2928427080907918e-05, "loss": 2.6347, "step": 38143 }, { "epoch": 2.367868893165311, "grad_norm": 0.48234460188651723, "learning_rate": 1.2926003741573172e-05, "loss": 2.6542, "step": 38144 }, { "epoch": 2.367930970265069, "grad_norm": 0.4386684618007797, "learning_rate": 1.2923580595662466e-05, "loss": 2.6947, "step": 38145 }, { "epoch": 2.367993047364827, "grad_norm": 0.4449269482839974, "learning_rate": 1.2921157643188443e-05, "loss": 2.6297, "step": 38146 }, { "epoch": 2.368055124464585, "grad_norm": 0.46451925924962617, "learning_rate": 1.2918734884163736e-05, "loss": 2.671, "step": 38147 }, { "epoch": 2.368117201564343, "grad_norm": 0.4607592254114566, "learning_rate": 1.2916312318600982e-05, "loss": 2.7614, "step": 38148 }, { "epoch": 2.3681792786641007, "grad_norm": 0.4464893017060672, "learning_rate": 1.2913889946512837e-05, "loss": 2.7131, "step": 38149 }, { "epoch": 2.3682413557638586, "grad_norm": 0.4817239601083644, "learning_rate": 1.2911467767911938e-05, "loss": 2.6877, "step": 38150 }, { "epoch": 2.3683034328636166, "grad_norm": 0.4844460435265773, "learning_rate": 1.2909045782810909e-05, "loss": 2.7354, "step": 38151 }, { "epoch": 2.3683655099633745, "grad_norm": 0.44559857732864183, "learning_rate": 1.2906623991222382e-05, "loss": 2.6071, "step": 38152 }, { "epoch": 2.3684275870631324, "grad_norm": 0.4902348461376829, "learning_rate": 1.2904202393159016e-05, "loss": 2.6677, "step": 38153 }, { "epoch": 2.3684896641628903, "grad_norm": 0.4634455963477033, "learning_rate": 1.2901780988633433e-05, "loss": 2.653, "step": 38154 }, { "epoch": 2.3685517412626482, "grad_norm": 0.4311477816417046, "learning_rate": 1.289935977765826e-05, "loss": 2.7253, "step": 38155 }, { "epoch": 2.368613818362406, "grad_norm": 0.501219811232006, "learning_rate": 1.2896938760246118e-05, "loss": 2.7268, "step": 38156 }, { "epoch": 2.368675895462164, "grad_norm": 0.5072883835620375, "learning_rate": 1.2894517936409673e-05, "loss": 2.7379, "step": 38157 }, { "epoch": 2.368737972561922, "grad_norm": 0.42450382496488837, "learning_rate": 1.289209730616152e-05, "loss": 2.6157, "step": 38158 }, { "epoch": 2.36880004966168, "grad_norm": 0.49458050655798746, "learning_rate": 1.2889676869514317e-05, "loss": 2.617, "step": 38159 }, { "epoch": 2.368862126761438, "grad_norm": 0.5010866040587704, "learning_rate": 1.2887256626480681e-05, "loss": 2.6796, "step": 38160 }, { "epoch": 2.3689242038611957, "grad_norm": 0.45189648193328774, "learning_rate": 1.2884836577073234e-05, "loss": 2.6905, "step": 38161 }, { "epoch": 2.3689862809609536, "grad_norm": 0.4691118909690014, "learning_rate": 1.288241672130459e-05, "loss": 2.6422, "step": 38162 }, { "epoch": 2.3690483580607116, "grad_norm": 0.47614384068742743, "learning_rate": 1.2879997059187403e-05, "loss": 2.6754, "step": 38163 }, { "epoch": 2.3691104351604695, "grad_norm": 0.472114173325681, "learning_rate": 1.2877577590734285e-05, "loss": 2.5999, "step": 38164 }, { "epoch": 2.369172512260227, "grad_norm": 0.5141120446166, "learning_rate": 1.2875158315957858e-05, "loss": 2.6265, "step": 38165 }, { "epoch": 2.3692345893599853, "grad_norm": 0.457194062927033, "learning_rate": 1.2872739234870723e-05, "loss": 2.6471, "step": 38166 }, { "epoch": 2.369296666459743, "grad_norm": 0.5150936162914067, "learning_rate": 1.2870320347485537e-05, "loss": 2.6788, "step": 38167 }, { "epoch": 2.3693587435595007, "grad_norm": 0.49897506091191407, "learning_rate": 1.2867901653814907e-05, "loss": 2.7122, "step": 38168 }, { "epoch": 2.3694208206592586, "grad_norm": 0.4824687765079325, "learning_rate": 1.2865483153871444e-05, "loss": 2.7508, "step": 38169 }, { "epoch": 2.3694828977590165, "grad_norm": 0.5345409503715597, "learning_rate": 1.2863064847667755e-05, "loss": 2.7266, "step": 38170 }, { "epoch": 2.3695449748587745, "grad_norm": 0.4799016691857299, "learning_rate": 1.2860646735216486e-05, "loss": 2.6902, "step": 38171 }, { "epoch": 2.3696070519585324, "grad_norm": 0.4543632327658862, "learning_rate": 1.2858228816530244e-05, "loss": 2.6474, "step": 38172 }, { "epoch": 2.3696691290582903, "grad_norm": 0.46428712809263584, "learning_rate": 1.2855811091621633e-05, "loss": 2.753, "step": 38173 }, { "epoch": 2.369731206158048, "grad_norm": 0.4989696107862808, "learning_rate": 1.2853393560503274e-05, "loss": 2.5532, "step": 38174 }, { "epoch": 2.369793283257806, "grad_norm": 0.5106679304418954, "learning_rate": 1.2850976223187766e-05, "loss": 2.6684, "step": 38175 }, { "epoch": 2.369855360357564, "grad_norm": 0.45525086689384164, "learning_rate": 1.2848559079687744e-05, "loss": 2.6757, "step": 38176 }, { "epoch": 2.369917437457322, "grad_norm": 0.4661575791852157, "learning_rate": 1.2846142130015814e-05, "loss": 2.7534, "step": 38177 }, { "epoch": 2.36997951455708, "grad_norm": 0.43199636722070406, "learning_rate": 1.2843725374184574e-05, "loss": 2.6706, "step": 38178 }, { "epoch": 2.370041591656838, "grad_norm": 0.47429165978816945, "learning_rate": 1.2841308812206626e-05, "loss": 2.6971, "step": 38179 }, { "epoch": 2.3701036687565957, "grad_norm": 0.4278573159985791, "learning_rate": 1.2838892444094609e-05, "loss": 2.6995, "step": 38180 }, { "epoch": 2.3701657458563536, "grad_norm": 0.4386482157657991, "learning_rate": 1.2836476269861108e-05, "loss": 2.7945, "step": 38181 }, { "epoch": 2.3702278229561116, "grad_norm": 0.4689063765040389, "learning_rate": 1.283406028951873e-05, "loss": 2.7382, "step": 38182 }, { "epoch": 2.3702899000558695, "grad_norm": 0.5107700201733095, "learning_rate": 1.2831644503080087e-05, "loss": 2.6681, "step": 38183 }, { "epoch": 2.3703519771556274, "grad_norm": 0.44794956537419994, "learning_rate": 1.2829228910557756e-05, "loss": 2.6594, "step": 38184 }, { "epoch": 2.3704140542553853, "grad_norm": 0.43046817952583855, "learning_rate": 1.282681351196438e-05, "loss": 2.646, "step": 38185 }, { "epoch": 2.3704761313551432, "grad_norm": 0.4975135940103446, "learning_rate": 1.2824398307312547e-05, "loss": 2.7217, "step": 38186 }, { "epoch": 2.370538208454901, "grad_norm": 0.4651711857415719, "learning_rate": 1.2821983296614842e-05, "loss": 2.7162, "step": 38187 }, { "epoch": 2.370600285554659, "grad_norm": 0.45104101686085796, "learning_rate": 1.281956847988387e-05, "loss": 2.6748, "step": 38188 }, { "epoch": 2.370662362654417, "grad_norm": 0.48256591623334927, "learning_rate": 1.2817153857132248e-05, "loss": 2.6416, "step": 38189 }, { "epoch": 2.3707244397541745, "grad_norm": 0.4613263865800706, "learning_rate": 1.2814739428372557e-05, "loss": 2.6862, "step": 38190 }, { "epoch": 2.370786516853933, "grad_norm": 0.508157788775372, "learning_rate": 1.2812325193617386e-05, "loss": 2.7191, "step": 38191 }, { "epoch": 2.3708485939536903, "grad_norm": 0.43833887194021914, "learning_rate": 1.2809911152879356e-05, "loss": 2.7347, "step": 38192 }, { "epoch": 2.370910671053448, "grad_norm": 0.462890930819511, "learning_rate": 1.2807497306171029e-05, "loss": 2.7097, "step": 38193 }, { "epoch": 2.370972748153206, "grad_norm": 0.43766492051262645, "learning_rate": 1.2805083653505035e-05, "loss": 2.7143, "step": 38194 }, { "epoch": 2.371034825252964, "grad_norm": 0.45749945174116624, "learning_rate": 1.2802670194893951e-05, "loss": 2.6089, "step": 38195 }, { "epoch": 2.371096902352722, "grad_norm": 0.4451763547278504, "learning_rate": 1.2800256930350362e-05, "loss": 2.6265, "step": 38196 }, { "epoch": 2.37115897945248, "grad_norm": 0.432244874727477, "learning_rate": 1.2797843859886871e-05, "loss": 2.6047, "step": 38197 }, { "epoch": 2.371221056552238, "grad_norm": 0.44974306440598405, "learning_rate": 1.2795430983516038e-05, "loss": 2.6288, "step": 38198 }, { "epoch": 2.3712831336519957, "grad_norm": 0.4733403669621602, "learning_rate": 1.2793018301250492e-05, "loss": 2.7522, "step": 38199 }, { "epoch": 2.3713452107517536, "grad_norm": 0.4076421615780426, "learning_rate": 1.2790605813102796e-05, "loss": 2.7197, "step": 38200 }, { "epoch": 2.3714072878515116, "grad_norm": 0.5135278826315135, "learning_rate": 1.2788193519085551e-05, "loss": 2.7526, "step": 38201 }, { "epoch": 2.3714693649512695, "grad_norm": 0.45455943988287695, "learning_rate": 1.2785781419211312e-05, "loss": 2.7353, "step": 38202 }, { "epoch": 2.3715314420510274, "grad_norm": 0.41916492459268234, "learning_rate": 1.2783369513492705e-05, "loss": 2.628, "step": 38203 }, { "epoch": 2.3715935191507853, "grad_norm": 0.4451146941305563, "learning_rate": 1.278095780194229e-05, "loss": 2.6072, "step": 38204 }, { "epoch": 2.3716555962505432, "grad_norm": 0.44757890349320584, "learning_rate": 1.2778546284572657e-05, "loss": 2.7165, "step": 38205 }, { "epoch": 2.371717673350301, "grad_norm": 0.451049243064603, "learning_rate": 1.2776134961396364e-05, "loss": 2.6839, "step": 38206 }, { "epoch": 2.371779750450059, "grad_norm": 0.4671791139698433, "learning_rate": 1.2773723832426032e-05, "loss": 2.7052, "step": 38207 }, { "epoch": 2.371841827549817, "grad_norm": 0.4016125525366024, "learning_rate": 1.2771312897674215e-05, "loss": 2.6588, "step": 38208 }, { "epoch": 2.371903904649575, "grad_norm": 0.48864284096753513, "learning_rate": 1.2768902157153495e-05, "loss": 2.6985, "step": 38209 }, { "epoch": 2.371965981749333, "grad_norm": 0.4375449395940839, "learning_rate": 1.2766491610876458e-05, "loss": 2.6944, "step": 38210 }, { "epoch": 2.3720280588490907, "grad_norm": 0.439014368371726, "learning_rate": 1.2764081258855654e-05, "loss": 2.6826, "step": 38211 }, { "epoch": 2.3720901359488487, "grad_norm": 0.4449216733803095, "learning_rate": 1.276167110110369e-05, "loss": 2.6988, "step": 38212 }, { "epoch": 2.372152213048606, "grad_norm": 0.42915451632895135, "learning_rate": 1.2759261137633133e-05, "loss": 2.5786, "step": 38213 }, { "epoch": 2.3722142901483645, "grad_norm": 0.46174942857059464, "learning_rate": 1.2756851368456546e-05, "loss": 2.6606, "step": 38214 }, { "epoch": 2.372276367248122, "grad_norm": 0.4254825489217515, "learning_rate": 1.2754441793586491e-05, "loss": 2.7077, "step": 38215 }, { "epoch": 2.37233844434788, "grad_norm": 0.4423139067958028, "learning_rate": 1.2752032413035574e-05, "loss": 2.6339, "step": 38216 }, { "epoch": 2.372400521447638, "grad_norm": 0.47936368257474743, "learning_rate": 1.2749623226816343e-05, "loss": 2.6064, "step": 38217 }, { "epoch": 2.3724625985473957, "grad_norm": 0.43979472129311514, "learning_rate": 1.2747214234941373e-05, "loss": 2.6322, "step": 38218 }, { "epoch": 2.3725246756471536, "grad_norm": 0.44981035170175276, "learning_rate": 1.2744805437423229e-05, "loss": 2.6832, "step": 38219 }, { "epoch": 2.3725867527469116, "grad_norm": 0.4384486947836183, "learning_rate": 1.2742396834274467e-05, "loss": 2.7088, "step": 38220 }, { "epoch": 2.3726488298466695, "grad_norm": 0.4514147612501163, "learning_rate": 1.2739988425507676e-05, "loss": 2.6861, "step": 38221 }, { "epoch": 2.3727109069464274, "grad_norm": 0.5000832607563976, "learning_rate": 1.2737580211135414e-05, "loss": 2.7078, "step": 38222 }, { "epoch": 2.3727729840461853, "grad_norm": 0.4410113027815283, "learning_rate": 1.2735172191170237e-05, "loss": 2.6272, "step": 38223 }, { "epoch": 2.3728350611459432, "grad_norm": 0.44086798197625393, "learning_rate": 1.27327643656247e-05, "loss": 2.6896, "step": 38224 }, { "epoch": 2.372897138245701, "grad_norm": 0.48262342497601474, "learning_rate": 1.2730356734511378e-05, "loss": 2.6518, "step": 38225 }, { "epoch": 2.372959215345459, "grad_norm": 0.44158705662654263, "learning_rate": 1.272794929784285e-05, "loss": 2.6727, "step": 38226 }, { "epoch": 2.373021292445217, "grad_norm": 0.45007973136807217, "learning_rate": 1.2725542055631655e-05, "loss": 2.657, "step": 38227 }, { "epoch": 2.373083369544975, "grad_norm": 0.47721430495245776, "learning_rate": 1.2723135007890359e-05, "loss": 2.6704, "step": 38228 }, { "epoch": 2.373145446644733, "grad_norm": 0.43747750292565096, "learning_rate": 1.27207281546315e-05, "loss": 2.6456, "step": 38229 }, { "epoch": 2.3732075237444907, "grad_norm": 0.42642658220113794, "learning_rate": 1.2718321495867669e-05, "loss": 2.6574, "step": 38230 }, { "epoch": 2.3732696008442487, "grad_norm": 0.4666218449434577, "learning_rate": 1.2715915031611402e-05, "loss": 2.6825, "step": 38231 }, { "epoch": 2.3733316779440066, "grad_norm": 0.5048070441231391, "learning_rate": 1.271350876187526e-05, "loss": 2.6827, "step": 38232 }, { "epoch": 2.3733937550437645, "grad_norm": 0.48333203408735315, "learning_rate": 1.2711102686671794e-05, "loss": 2.7085, "step": 38233 }, { "epoch": 2.3734558321435224, "grad_norm": 0.4765755854826905, "learning_rate": 1.2708696806013542e-05, "loss": 2.7225, "step": 38234 }, { "epoch": 2.3735179092432803, "grad_norm": 0.48090218942960455, "learning_rate": 1.2706291119913088e-05, "loss": 2.6761, "step": 38235 }, { "epoch": 2.3735799863430382, "grad_norm": 0.4959159141847421, "learning_rate": 1.2703885628382967e-05, "loss": 2.6309, "step": 38236 }, { "epoch": 2.373642063442796, "grad_norm": 0.49037370563543375, "learning_rate": 1.2701480331435722e-05, "loss": 2.6854, "step": 38237 }, { "epoch": 2.3737041405425536, "grad_norm": 0.5201474887503094, "learning_rate": 1.2699075229083901e-05, "loss": 2.7829, "step": 38238 }, { "epoch": 2.373766217642312, "grad_norm": 0.4545532441284558, "learning_rate": 1.2696670321340071e-05, "loss": 2.6636, "step": 38239 }, { "epoch": 2.3738282947420695, "grad_norm": 0.470360659037791, "learning_rate": 1.269426560821677e-05, "loss": 2.6454, "step": 38240 }, { "epoch": 2.3738903718418274, "grad_norm": 0.4790478296391397, "learning_rate": 1.2691861089726542e-05, "loss": 2.6534, "step": 38241 }, { "epoch": 2.3739524489415853, "grad_norm": 0.4790541751782749, "learning_rate": 1.2689456765881913e-05, "loss": 2.6214, "step": 38242 }, { "epoch": 2.374014526041343, "grad_norm": 0.4855560393937831, "learning_rate": 1.2687052636695462e-05, "loss": 2.7292, "step": 38243 }, { "epoch": 2.374076603141101, "grad_norm": 0.5187820826801185, "learning_rate": 1.2684648702179719e-05, "loss": 2.7687, "step": 38244 }, { "epoch": 2.374138680240859, "grad_norm": 0.5058706274953535, "learning_rate": 1.2682244962347217e-05, "loss": 2.6536, "step": 38245 }, { "epoch": 2.374200757340617, "grad_norm": 0.46675508703147534, "learning_rate": 1.2679841417210504e-05, "loss": 2.6846, "step": 38246 }, { "epoch": 2.374262834440375, "grad_norm": 0.49805212579091024, "learning_rate": 1.2677438066782104e-05, "loss": 2.7356, "step": 38247 }, { "epoch": 2.374324911540133, "grad_norm": 0.4941020543420729, "learning_rate": 1.2675034911074585e-05, "loss": 2.6957, "step": 38248 }, { "epoch": 2.3743869886398907, "grad_norm": 0.47077426926748234, "learning_rate": 1.2672631950100466e-05, "loss": 2.6045, "step": 38249 }, { "epoch": 2.3744490657396486, "grad_norm": 0.46571796820423345, "learning_rate": 1.2670229183872295e-05, "loss": 2.7036, "step": 38250 }, { "epoch": 2.3745111428394066, "grad_norm": 0.4714176089619217, "learning_rate": 1.2667826612402578e-05, "loss": 2.6279, "step": 38251 }, { "epoch": 2.3745732199391645, "grad_norm": 0.48392879791988447, "learning_rate": 1.2665424235703894e-05, "loss": 2.5867, "step": 38252 }, { "epoch": 2.3746352970389224, "grad_norm": 0.5100331974444627, "learning_rate": 1.2663022053788748e-05, "loss": 2.727, "step": 38253 }, { "epoch": 2.3746973741386803, "grad_norm": 0.4402707271917263, "learning_rate": 1.2660620066669687e-05, "loss": 2.7311, "step": 38254 }, { "epoch": 2.3747594512384382, "grad_norm": 0.5423559669167093, "learning_rate": 1.2658218274359218e-05, "loss": 2.6981, "step": 38255 }, { "epoch": 2.374821528338196, "grad_norm": 0.4605240826713944, "learning_rate": 1.2655816676869887e-05, "loss": 2.6083, "step": 38256 }, { "epoch": 2.374883605437954, "grad_norm": 0.45498972349299277, "learning_rate": 1.2653415274214243e-05, "loss": 2.595, "step": 38257 }, { "epoch": 2.374945682537712, "grad_norm": 0.4565676106213254, "learning_rate": 1.2651014066404798e-05, "loss": 2.6664, "step": 38258 }, { "epoch": 2.37500775963747, "grad_norm": 0.4617646602777628, "learning_rate": 1.2648613053454078e-05, "loss": 2.6607, "step": 38259 }, { "epoch": 2.375069836737228, "grad_norm": 0.4751384282431249, "learning_rate": 1.264621223537461e-05, "loss": 2.6819, "step": 38260 }, { "epoch": 2.3751319138369853, "grad_norm": 0.4810370828843274, "learning_rate": 1.2643811612178908e-05, "loss": 2.733, "step": 38261 }, { "epoch": 2.3751939909367437, "grad_norm": 0.5028634629581982, "learning_rate": 1.2641411183879525e-05, "loss": 2.708, "step": 38262 }, { "epoch": 2.375256068036501, "grad_norm": 0.5145805936681221, "learning_rate": 1.2639010950488966e-05, "loss": 2.7425, "step": 38263 }, { "epoch": 2.375318145136259, "grad_norm": 0.4346408572426899, "learning_rate": 1.2636610912019759e-05, "loss": 2.7067, "step": 38264 }, { "epoch": 2.375380222236017, "grad_norm": 0.47334244469427217, "learning_rate": 1.2634211068484408e-05, "loss": 2.6734, "step": 38265 }, { "epoch": 2.375442299335775, "grad_norm": 0.4977477128854318, "learning_rate": 1.2631811419895462e-05, "loss": 2.694, "step": 38266 }, { "epoch": 2.375504376435533, "grad_norm": 0.45838055764733493, "learning_rate": 1.2629411966265431e-05, "loss": 2.7167, "step": 38267 }, { "epoch": 2.3755664535352907, "grad_norm": 0.4418146556510795, "learning_rate": 1.2627012707606822e-05, "loss": 2.6159, "step": 38268 }, { "epoch": 2.3756285306350486, "grad_norm": 0.4269346311704859, "learning_rate": 1.262461364393217e-05, "loss": 2.6909, "step": 38269 }, { "epoch": 2.3756906077348066, "grad_norm": 0.46515077981753294, "learning_rate": 1.262221477525396e-05, "loss": 2.7166, "step": 38270 }, { "epoch": 2.3757526848345645, "grad_norm": 0.5071532738889929, "learning_rate": 1.2619816101584747e-05, "loss": 2.6297, "step": 38271 }, { "epoch": 2.3758147619343224, "grad_norm": 0.4766017241048183, "learning_rate": 1.2617417622937027e-05, "loss": 2.7451, "step": 38272 }, { "epoch": 2.3758768390340803, "grad_norm": 0.44319847312945687, "learning_rate": 1.2615019339323314e-05, "loss": 2.7229, "step": 38273 }, { "epoch": 2.3759389161338382, "grad_norm": 0.4545502937395252, "learning_rate": 1.2612621250756107e-05, "loss": 2.6494, "step": 38274 }, { "epoch": 2.376000993233596, "grad_norm": 0.5329230554608829, "learning_rate": 1.2610223357247946e-05, "loss": 2.7262, "step": 38275 }, { "epoch": 2.376063070333354, "grad_norm": 0.4571152960289975, "learning_rate": 1.2607825658811323e-05, "loss": 2.6112, "step": 38276 }, { "epoch": 2.376125147433112, "grad_norm": 0.4641538838276335, "learning_rate": 1.2605428155458753e-05, "loss": 2.7094, "step": 38277 }, { "epoch": 2.37618722453287, "grad_norm": 0.5400542107056195, "learning_rate": 1.260303084720273e-05, "loss": 2.7646, "step": 38278 }, { "epoch": 2.376249301632628, "grad_norm": 0.4131154980915756, "learning_rate": 1.2600633734055789e-05, "loss": 2.5806, "step": 38279 }, { "epoch": 2.3763113787323857, "grad_norm": 0.48736065866112266, "learning_rate": 1.2598236816030417e-05, "loss": 2.6701, "step": 38280 }, { "epoch": 2.3763734558321437, "grad_norm": 0.4474271291134874, "learning_rate": 1.2595840093139121e-05, "loss": 2.6795, "step": 38281 }, { "epoch": 2.3764355329319016, "grad_norm": 0.46924083760624063, "learning_rate": 1.2593443565394414e-05, "loss": 2.7439, "step": 38282 }, { "epoch": 2.3764976100316595, "grad_norm": 0.4626123259724658, "learning_rate": 1.2591047232808773e-05, "loss": 2.7546, "step": 38283 }, { "epoch": 2.3765596871314174, "grad_norm": 0.47496211126699983, "learning_rate": 1.2588651095394732e-05, "loss": 2.7201, "step": 38284 }, { "epoch": 2.3766217642311753, "grad_norm": 0.4684015886564865, "learning_rate": 1.2586255153164789e-05, "loss": 2.6679, "step": 38285 }, { "epoch": 2.376683841330933, "grad_norm": 0.46668720952299936, "learning_rate": 1.2583859406131427e-05, "loss": 2.6899, "step": 38286 }, { "epoch": 2.376745918430691, "grad_norm": 0.44974773607961804, "learning_rate": 1.2581463854307146e-05, "loss": 2.6831, "step": 38287 }, { "epoch": 2.3768079955304486, "grad_norm": 0.45419294196716786, "learning_rate": 1.2579068497704461e-05, "loss": 2.699, "step": 38288 }, { "epoch": 2.3768700726302066, "grad_norm": 0.44744766062466945, "learning_rate": 1.2576673336335849e-05, "loss": 2.6518, "step": 38289 }, { "epoch": 2.3769321497299645, "grad_norm": 0.48545851574814314, "learning_rate": 1.2574278370213826e-05, "loss": 2.6704, "step": 38290 }, { "epoch": 2.3769942268297224, "grad_norm": 0.4879902544758469, "learning_rate": 1.257188359935088e-05, "loss": 2.7309, "step": 38291 }, { "epoch": 2.3770563039294803, "grad_norm": 0.4430812517907279, "learning_rate": 1.2569489023759495e-05, "loss": 2.6964, "step": 38292 }, { "epoch": 2.3771183810292382, "grad_norm": 0.5046006969527226, "learning_rate": 1.2567094643452182e-05, "loss": 2.6551, "step": 38293 }, { "epoch": 2.377180458128996, "grad_norm": 0.4970604496485132, "learning_rate": 1.2564700458441424e-05, "loss": 2.6819, "step": 38294 }, { "epoch": 2.377242535228754, "grad_norm": 0.4563960200974974, "learning_rate": 1.256230646873971e-05, "loss": 2.6758, "step": 38295 }, { "epoch": 2.377304612328512, "grad_norm": 0.49846271283708404, "learning_rate": 1.255991267435953e-05, "loss": 2.6493, "step": 38296 }, { "epoch": 2.37736668942827, "grad_norm": 0.48296049683009784, "learning_rate": 1.2557519075313367e-05, "loss": 2.6654, "step": 38297 }, { "epoch": 2.377428766528028, "grad_norm": 0.46972971893328613, "learning_rate": 1.2555125671613726e-05, "loss": 2.6868, "step": 38298 }, { "epoch": 2.3774908436277857, "grad_norm": 0.5057602420718299, "learning_rate": 1.2552732463273087e-05, "loss": 2.6175, "step": 38299 }, { "epoch": 2.3775529207275437, "grad_norm": 0.4628077404645513, "learning_rate": 1.2550339450303927e-05, "loss": 2.6615, "step": 38300 }, { "epoch": 2.3776149978273016, "grad_norm": 0.4826040811893358, "learning_rate": 1.2547946632718732e-05, "loss": 2.6163, "step": 38301 }, { "epoch": 2.3776770749270595, "grad_norm": 0.5028651816471386, "learning_rate": 1.2545554010529997e-05, "loss": 2.7151, "step": 38302 }, { "epoch": 2.3777391520268174, "grad_norm": 0.4358047611408908, "learning_rate": 1.2543161583750207e-05, "loss": 2.6206, "step": 38303 }, { "epoch": 2.3778012291265753, "grad_norm": 0.5202798827432492, "learning_rate": 1.2540769352391829e-05, "loss": 2.7097, "step": 38304 }, { "epoch": 2.3778633062263332, "grad_norm": 0.5138301822991791, "learning_rate": 1.2538377316467342e-05, "loss": 2.6065, "step": 38305 }, { "epoch": 2.377925383326091, "grad_norm": 0.4831576277361183, "learning_rate": 1.2535985475989242e-05, "loss": 2.7487, "step": 38306 }, { "epoch": 2.377987460425849, "grad_norm": 0.4776754360070607, "learning_rate": 1.2533593830970008e-05, "loss": 2.6309, "step": 38307 }, { "epoch": 2.378049537525607, "grad_norm": 0.5250493185402364, "learning_rate": 1.2531202381422107e-05, "loss": 2.6671, "step": 38308 }, { "epoch": 2.3781116146253645, "grad_norm": 0.48615592757521653, "learning_rate": 1.252881112735802e-05, "loss": 2.6614, "step": 38309 }, { "epoch": 2.378173691725123, "grad_norm": 0.48127240525146064, "learning_rate": 1.2526420068790206e-05, "loss": 2.6298, "step": 38310 }, { "epoch": 2.3782357688248803, "grad_norm": 0.44324034718966615, "learning_rate": 1.2524029205731169e-05, "loss": 2.7126, "step": 38311 }, { "epoch": 2.378297845924638, "grad_norm": 0.47360083088604593, "learning_rate": 1.2521638538193375e-05, "loss": 2.633, "step": 38312 }, { "epoch": 2.378359923024396, "grad_norm": 0.43458291367680935, "learning_rate": 1.2519248066189287e-05, "loss": 2.6204, "step": 38313 }, { "epoch": 2.378422000124154, "grad_norm": 0.4390860147390099, "learning_rate": 1.2516857789731363e-05, "loss": 2.6765, "step": 38314 }, { "epoch": 2.378484077223912, "grad_norm": 0.44803913977812776, "learning_rate": 1.2514467708832112e-05, "loss": 2.7253, "step": 38315 }, { "epoch": 2.37854615432367, "grad_norm": 0.46783774935277206, "learning_rate": 1.2512077823503977e-05, "loss": 2.7389, "step": 38316 }, { "epoch": 2.378608231423428, "grad_norm": 0.441860589678661, "learning_rate": 1.2509688133759434e-05, "loss": 2.6961, "step": 38317 }, { "epoch": 2.3786703085231857, "grad_norm": 0.4396084320291453, "learning_rate": 1.2507298639610954e-05, "loss": 2.6429, "step": 38318 }, { "epoch": 2.3787323856229436, "grad_norm": 0.4206763475940702, "learning_rate": 1.250490934107098e-05, "loss": 2.629, "step": 38319 }, { "epoch": 2.3787944627227016, "grad_norm": 0.4675069220943319, "learning_rate": 1.250252023815201e-05, "loss": 2.6582, "step": 38320 }, { "epoch": 2.3788565398224595, "grad_norm": 0.44777849934803515, "learning_rate": 1.2500131330866494e-05, "loss": 2.6986, "step": 38321 }, { "epoch": 2.3789186169222174, "grad_norm": 0.42244935674574363, "learning_rate": 1.2497742619226887e-05, "loss": 2.6636, "step": 38322 }, { "epoch": 2.3789806940219753, "grad_norm": 0.4277598229768018, "learning_rate": 1.249535410324567e-05, "loss": 2.7039, "step": 38323 }, { "epoch": 2.3790427711217332, "grad_norm": 0.4479610251115572, "learning_rate": 1.2492965782935284e-05, "loss": 2.7075, "step": 38324 }, { "epoch": 2.379104848221491, "grad_norm": 0.4359302287708799, "learning_rate": 1.2490577658308212e-05, "loss": 2.7205, "step": 38325 }, { "epoch": 2.379166925321249, "grad_norm": 0.409896641269584, "learning_rate": 1.2488189729376903e-05, "loss": 2.6357, "step": 38326 }, { "epoch": 2.379229002421007, "grad_norm": 0.47536044875249606, "learning_rate": 1.2485801996153817e-05, "loss": 2.6865, "step": 38327 }, { "epoch": 2.379291079520765, "grad_norm": 0.47280126506620845, "learning_rate": 1.2483414458651393e-05, "loss": 2.7284, "step": 38328 }, { "epoch": 2.379353156620523, "grad_norm": 0.46999746271219606, "learning_rate": 1.2481027116882115e-05, "loss": 2.6801, "step": 38329 }, { "epoch": 2.3794152337202807, "grad_norm": 0.43466806062497276, "learning_rate": 1.247863997085843e-05, "loss": 2.6247, "step": 38330 }, { "epoch": 2.3794773108200387, "grad_norm": 0.44693672380792077, "learning_rate": 1.2476253020592788e-05, "loss": 2.6646, "step": 38331 }, { "epoch": 2.3795393879197966, "grad_norm": 0.467764310807486, "learning_rate": 1.2473866266097645e-05, "loss": 2.6536, "step": 38332 }, { "epoch": 2.3796014650195545, "grad_norm": 0.4521544025596163, "learning_rate": 1.2471479707385436e-05, "loss": 2.6058, "step": 38333 }, { "epoch": 2.379663542119312, "grad_norm": 0.44668331303228104, "learning_rate": 1.2469093344468641e-05, "loss": 2.6753, "step": 38334 }, { "epoch": 2.3797256192190703, "grad_norm": 0.43286654344957254, "learning_rate": 1.2466707177359698e-05, "loss": 2.7047, "step": 38335 }, { "epoch": 2.379787696318828, "grad_norm": 0.48396834869575034, "learning_rate": 1.2464321206071055e-05, "loss": 2.7054, "step": 38336 }, { "epoch": 2.3798497734185857, "grad_norm": 0.4610752449471834, "learning_rate": 1.2461935430615146e-05, "loss": 2.616, "step": 38337 }, { "epoch": 2.3799118505183436, "grad_norm": 0.45808491332404083, "learning_rate": 1.2459549851004443e-05, "loss": 2.6312, "step": 38338 }, { "epoch": 2.3799739276181016, "grad_norm": 0.5087988444319684, "learning_rate": 1.2457164467251387e-05, "loss": 2.7025, "step": 38339 }, { "epoch": 2.3800360047178595, "grad_norm": 0.43170529739787067, "learning_rate": 1.2454779279368417e-05, "loss": 2.6964, "step": 38340 }, { "epoch": 2.3800980818176174, "grad_norm": 0.4533997557722712, "learning_rate": 1.2452394287367964e-05, "loss": 2.7263, "step": 38341 }, { "epoch": 2.3801601589173753, "grad_norm": 0.5005655964005317, "learning_rate": 1.2450009491262493e-05, "loss": 2.6489, "step": 38342 }, { "epoch": 2.3802222360171332, "grad_norm": 0.5121249559536059, "learning_rate": 1.2447624891064447e-05, "loss": 2.7192, "step": 38343 }, { "epoch": 2.380284313116891, "grad_norm": 0.4849984094996771, "learning_rate": 1.2445240486786252e-05, "loss": 2.6483, "step": 38344 }, { "epoch": 2.380346390216649, "grad_norm": 0.4526470886820313, "learning_rate": 1.2442856278440357e-05, "loss": 2.66, "step": 38345 }, { "epoch": 2.380408467316407, "grad_norm": 0.5345844639057307, "learning_rate": 1.2440472266039182e-05, "loss": 2.6498, "step": 38346 }, { "epoch": 2.380470544416165, "grad_norm": 0.46690576472409423, "learning_rate": 1.2438088449595198e-05, "loss": 2.6019, "step": 38347 }, { "epoch": 2.380532621515923, "grad_norm": 0.45279425026794906, "learning_rate": 1.243570482912082e-05, "loss": 2.6402, "step": 38348 }, { "epoch": 2.3805946986156807, "grad_norm": 0.4681737695864493, "learning_rate": 1.2433321404628495e-05, "loss": 2.7879, "step": 38349 }, { "epoch": 2.3806567757154387, "grad_norm": 0.5104335632901066, "learning_rate": 1.2430938176130636e-05, "loss": 2.6134, "step": 38350 }, { "epoch": 2.3807188528151966, "grad_norm": 0.46653831375807847, "learning_rate": 1.2428555143639703e-05, "loss": 2.6975, "step": 38351 }, { "epoch": 2.3807809299149545, "grad_norm": 0.47616535585993214, "learning_rate": 1.2426172307168121e-05, "loss": 2.6915, "step": 38352 }, { "epoch": 2.3808430070147124, "grad_norm": 0.518107765840644, "learning_rate": 1.242378966672832e-05, "loss": 2.7277, "step": 38353 }, { "epoch": 2.3809050841144703, "grad_norm": 0.5057849313756955, "learning_rate": 1.2421407222332732e-05, "loss": 2.7216, "step": 38354 }, { "epoch": 2.3809671612142282, "grad_norm": 0.4771753185693454, "learning_rate": 1.2419024973993765e-05, "loss": 2.5914, "step": 38355 }, { "epoch": 2.381029238313986, "grad_norm": 0.5225904642850497, "learning_rate": 1.2416642921723876e-05, "loss": 2.6555, "step": 38356 }, { "epoch": 2.3810913154137436, "grad_norm": 0.46542107769826147, "learning_rate": 1.2414261065535494e-05, "loss": 2.6811, "step": 38357 }, { "epoch": 2.381153392513502, "grad_norm": 0.4392642314572487, "learning_rate": 1.2411879405441034e-05, "loss": 2.5972, "step": 38358 }, { "epoch": 2.3812154696132595, "grad_norm": 0.460486563243186, "learning_rate": 1.240949794145293e-05, "loss": 2.7097, "step": 38359 }, { "epoch": 2.3812775467130174, "grad_norm": 0.487126793813792, "learning_rate": 1.240711667358358e-05, "loss": 2.6743, "step": 38360 }, { "epoch": 2.3813396238127753, "grad_norm": 0.48251517934097043, "learning_rate": 1.2404735601845446e-05, "loss": 2.7119, "step": 38361 }, { "epoch": 2.3814017009125332, "grad_norm": 0.4502382899227842, "learning_rate": 1.2402354726250937e-05, "loss": 2.7254, "step": 38362 }, { "epoch": 2.381463778012291, "grad_norm": 0.5068548415816891, "learning_rate": 1.2399974046812462e-05, "loss": 2.6879, "step": 38363 }, { "epoch": 2.381525855112049, "grad_norm": 0.4798076392212402, "learning_rate": 1.2397593563542442e-05, "loss": 2.7405, "step": 38364 }, { "epoch": 2.381587932211807, "grad_norm": 0.508189369619522, "learning_rate": 1.2395213276453316e-05, "loss": 2.7265, "step": 38365 }, { "epoch": 2.381650009311565, "grad_norm": 0.5046182968832149, "learning_rate": 1.2392833185557495e-05, "loss": 2.684, "step": 38366 }, { "epoch": 2.381712086411323, "grad_norm": 0.4863944410587191, "learning_rate": 1.2390453290867388e-05, "loss": 2.6825, "step": 38367 }, { "epoch": 2.3817741635110807, "grad_norm": 0.5375536658975213, "learning_rate": 1.2388073592395416e-05, "loss": 2.7272, "step": 38368 }, { "epoch": 2.3818362406108387, "grad_norm": 0.5374788158145557, "learning_rate": 1.2385694090153981e-05, "loss": 2.7045, "step": 38369 }, { "epoch": 2.3818983177105966, "grad_norm": 0.47590300782542294, "learning_rate": 1.238331478415553e-05, "loss": 2.6912, "step": 38370 }, { "epoch": 2.3819603948103545, "grad_norm": 0.4888854903888103, "learning_rate": 1.2380935674412453e-05, "loss": 2.5114, "step": 38371 }, { "epoch": 2.3820224719101124, "grad_norm": 0.5063803799243838, "learning_rate": 1.2378556760937171e-05, "loss": 2.7147, "step": 38372 }, { "epoch": 2.3820845490098703, "grad_norm": 0.4949378115670714, "learning_rate": 1.2376178043742076e-05, "loss": 2.7122, "step": 38373 }, { "epoch": 2.3821466261096282, "grad_norm": 0.44364729955460336, "learning_rate": 1.2373799522839602e-05, "loss": 2.6428, "step": 38374 }, { "epoch": 2.382208703209386, "grad_norm": 0.5063261434018619, "learning_rate": 1.2371421198242156e-05, "loss": 2.6801, "step": 38375 }, { "epoch": 2.382270780309144, "grad_norm": 0.4620597722445012, "learning_rate": 1.2369043069962139e-05, "loss": 2.6289, "step": 38376 }, { "epoch": 2.382332857408902, "grad_norm": 0.44680050649166003, "learning_rate": 1.2366665138011946e-05, "loss": 2.6856, "step": 38377 }, { "epoch": 2.38239493450866, "grad_norm": 0.4642334305359245, "learning_rate": 1.2364287402404012e-05, "loss": 2.6579, "step": 38378 }, { "epoch": 2.382457011608418, "grad_norm": 0.4914511970863274, "learning_rate": 1.2361909863150723e-05, "loss": 2.6532, "step": 38379 }, { "epoch": 2.3825190887081757, "grad_norm": 0.419146722782997, "learning_rate": 1.2359532520264489e-05, "loss": 2.638, "step": 38380 }, { "epoch": 2.3825811658079337, "grad_norm": 0.4635186743302427, "learning_rate": 1.2357155373757712e-05, "loss": 2.7222, "step": 38381 }, { "epoch": 2.382643242907691, "grad_norm": 0.4876360941702725, "learning_rate": 1.2354778423642777e-05, "loss": 2.7415, "step": 38382 }, { "epoch": 2.3827053200074495, "grad_norm": 0.4606261899303869, "learning_rate": 1.2352401669932117e-05, "loss": 2.6738, "step": 38383 }, { "epoch": 2.382767397107207, "grad_norm": 0.4776703043758349, "learning_rate": 1.235002511263812e-05, "loss": 2.7582, "step": 38384 }, { "epoch": 2.382829474206965, "grad_norm": 0.46169456444408313, "learning_rate": 1.2347648751773177e-05, "loss": 2.8022, "step": 38385 }, { "epoch": 2.382891551306723, "grad_norm": 0.48068934703110466, "learning_rate": 1.2345272587349681e-05, "loss": 2.7005, "step": 38386 }, { "epoch": 2.3829536284064807, "grad_norm": 0.4620726718337662, "learning_rate": 1.2342896619380035e-05, "loss": 2.7429, "step": 38387 }, { "epoch": 2.3830157055062386, "grad_norm": 0.44502412928837914, "learning_rate": 1.2340520847876658e-05, "loss": 2.5975, "step": 38388 }, { "epoch": 2.3830777826059966, "grad_norm": 0.5141516935660985, "learning_rate": 1.2338145272851919e-05, "loss": 2.7291, "step": 38389 }, { "epoch": 2.3831398597057545, "grad_norm": 0.5114893216951011, "learning_rate": 1.2335769894318228e-05, "loss": 2.6657, "step": 38390 }, { "epoch": 2.3832019368055124, "grad_norm": 0.4568579744696983, "learning_rate": 1.2333394712287949e-05, "loss": 2.6779, "step": 38391 }, { "epoch": 2.3832640139052703, "grad_norm": 0.5071970217522567, "learning_rate": 1.2331019726773508e-05, "loss": 2.6949, "step": 38392 }, { "epoch": 2.3833260910050282, "grad_norm": 0.508773569175017, "learning_rate": 1.2328644937787282e-05, "loss": 2.6605, "step": 38393 }, { "epoch": 2.383388168104786, "grad_norm": 0.47920042589376466, "learning_rate": 1.2326270345341655e-05, "loss": 2.6439, "step": 38394 }, { "epoch": 2.383450245204544, "grad_norm": 0.5002069343071751, "learning_rate": 1.2323895949449033e-05, "loss": 2.7092, "step": 38395 }, { "epoch": 2.383512322304302, "grad_norm": 0.512521797972672, "learning_rate": 1.2321521750121768e-05, "loss": 2.6803, "step": 38396 }, { "epoch": 2.38357439940406, "grad_norm": 0.4296704549009583, "learning_rate": 1.231914774737229e-05, "loss": 2.7148, "step": 38397 }, { "epoch": 2.383636476503818, "grad_norm": 0.4894494083199494, "learning_rate": 1.2316773941212962e-05, "loss": 2.6455, "step": 38398 }, { "epoch": 2.3836985536035757, "grad_norm": 0.49578255933029064, "learning_rate": 1.2314400331656173e-05, "loss": 2.7351, "step": 38399 }, { "epoch": 2.3837606307033337, "grad_norm": 0.49574297420892266, "learning_rate": 1.2312026918714292e-05, "loss": 2.6987, "step": 38400 }, { "epoch": 2.3838227078030916, "grad_norm": 0.5087751800341848, "learning_rate": 1.2309653702399732e-05, "loss": 2.6352, "step": 38401 }, { "epoch": 2.3838847849028495, "grad_norm": 0.5127464632222001, "learning_rate": 1.230728068272486e-05, "loss": 2.6784, "step": 38402 }, { "epoch": 2.3839468620026074, "grad_norm": 0.4922092826503864, "learning_rate": 1.2304907859702053e-05, "loss": 2.71, "step": 38403 }, { "epoch": 2.3840089391023653, "grad_norm": 0.5175224418196933, "learning_rate": 1.2302535233343688e-05, "loss": 2.7016, "step": 38404 }, { "epoch": 2.384071016202123, "grad_norm": 0.5271682729872477, "learning_rate": 1.2300162803662135e-05, "loss": 2.5484, "step": 38405 }, { "epoch": 2.384133093301881, "grad_norm": 0.5172687685637197, "learning_rate": 1.2297790570669804e-05, "loss": 2.7055, "step": 38406 }, { "epoch": 2.3841951704016386, "grad_norm": 0.47925457631743446, "learning_rate": 1.229541853437905e-05, "loss": 2.7263, "step": 38407 }, { "epoch": 2.3842572475013966, "grad_norm": 0.5225373399433342, "learning_rate": 1.2293046694802251e-05, "loss": 2.674, "step": 38408 }, { "epoch": 2.3843193246011545, "grad_norm": 0.5186104897110158, "learning_rate": 1.229067505195176e-05, "loss": 2.6712, "step": 38409 }, { "epoch": 2.3843814017009124, "grad_norm": 0.4775444919768988, "learning_rate": 1.2288303605839996e-05, "loss": 2.6715, "step": 38410 }, { "epoch": 2.3844434788006703, "grad_norm": 0.4855619929960749, "learning_rate": 1.22859323564793e-05, "loss": 2.6507, "step": 38411 }, { "epoch": 2.3845055559004282, "grad_norm": 0.4970355668130806, "learning_rate": 1.2283561303882052e-05, "loss": 2.6622, "step": 38412 }, { "epoch": 2.384567633000186, "grad_norm": 0.43940324288292787, "learning_rate": 1.2281190448060608e-05, "loss": 2.7134, "step": 38413 }, { "epoch": 2.384629710099944, "grad_norm": 0.4997795781413084, "learning_rate": 1.2278819789027362e-05, "loss": 2.6435, "step": 38414 }, { "epoch": 2.384691787199702, "grad_norm": 0.4665974306647765, "learning_rate": 1.2276449326794676e-05, "loss": 2.7145, "step": 38415 }, { "epoch": 2.38475386429946, "grad_norm": 0.4761318543587478, "learning_rate": 1.2274079061374905e-05, "loss": 2.6549, "step": 38416 }, { "epoch": 2.384815941399218, "grad_norm": 0.47210183217917884, "learning_rate": 1.2271708992780422e-05, "loss": 2.6391, "step": 38417 }, { "epoch": 2.3848780184989757, "grad_norm": 0.49657580466478685, "learning_rate": 1.2269339121023583e-05, "loss": 2.6353, "step": 38418 }, { "epoch": 2.3849400955987337, "grad_norm": 0.5047608159905856, "learning_rate": 1.2266969446116771e-05, "loss": 2.6719, "step": 38419 }, { "epoch": 2.3850021726984916, "grad_norm": 0.4590731595790272, "learning_rate": 1.2264599968072327e-05, "loss": 2.6608, "step": 38420 }, { "epoch": 2.3850642497982495, "grad_norm": 0.4773365207505539, "learning_rate": 1.2262230686902637e-05, "loss": 2.6467, "step": 38421 }, { "epoch": 2.3851263268980074, "grad_norm": 0.4848404770450843, "learning_rate": 1.2259861602620055e-05, "loss": 2.7348, "step": 38422 }, { "epoch": 2.3851884039977653, "grad_norm": 0.4524233416963523, "learning_rate": 1.2257492715236918e-05, "loss": 2.6034, "step": 38423 }, { "epoch": 2.3852504810975232, "grad_norm": 0.42645920141989285, "learning_rate": 1.2255124024765623e-05, "loss": 2.6909, "step": 38424 }, { "epoch": 2.385312558197281, "grad_norm": 0.4401171774511657, "learning_rate": 1.2252755531218508e-05, "loss": 2.6621, "step": 38425 }, { "epoch": 2.385374635297039, "grad_norm": 0.4522669329997469, "learning_rate": 1.225038723460793e-05, "loss": 2.6919, "step": 38426 }, { "epoch": 2.385436712396797, "grad_norm": 0.462134214041627, "learning_rate": 1.2248019134946225e-05, "loss": 2.5638, "step": 38427 }, { "epoch": 2.3854987894965545, "grad_norm": 0.4537240869305557, "learning_rate": 1.2245651232245792e-05, "loss": 2.5956, "step": 38428 }, { "epoch": 2.385560866596313, "grad_norm": 0.46183022833128784, "learning_rate": 1.2243283526518957e-05, "loss": 2.7693, "step": 38429 }, { "epoch": 2.3856229436960703, "grad_norm": 0.46988007699669804, "learning_rate": 1.2240916017778082e-05, "loss": 2.6157, "step": 38430 }, { "epoch": 2.3856850207958282, "grad_norm": 0.47065014362926716, "learning_rate": 1.2238548706035507e-05, "loss": 2.6966, "step": 38431 }, { "epoch": 2.385747097895586, "grad_norm": 0.49572560024187623, "learning_rate": 1.223618159130358e-05, "loss": 2.6721, "step": 38432 }, { "epoch": 2.385809174995344, "grad_norm": 0.4295905610755157, "learning_rate": 1.2233814673594679e-05, "loss": 2.6335, "step": 38433 }, { "epoch": 2.385871252095102, "grad_norm": 0.4913751164748918, "learning_rate": 1.2231447952921132e-05, "loss": 2.6292, "step": 38434 }, { "epoch": 2.38593332919486, "grad_norm": 0.47959932889978896, "learning_rate": 1.222908142929529e-05, "loss": 2.7422, "step": 38435 }, { "epoch": 2.385995406294618, "grad_norm": 0.4267541985609726, "learning_rate": 1.2226715102729486e-05, "loss": 2.6601, "step": 38436 }, { "epoch": 2.3860574833943757, "grad_norm": 0.44431707321198094, "learning_rate": 1.2224348973236088e-05, "loss": 2.6734, "step": 38437 }, { "epoch": 2.3861195604941337, "grad_norm": 0.4583904725639594, "learning_rate": 1.222198304082744e-05, "loss": 2.6798, "step": 38438 }, { "epoch": 2.3861816375938916, "grad_norm": 0.4980973311652778, "learning_rate": 1.2219617305515874e-05, "loss": 2.686, "step": 38439 }, { "epoch": 2.3862437146936495, "grad_norm": 0.4266633337264467, "learning_rate": 1.2217251767313732e-05, "loss": 2.6328, "step": 38440 }, { "epoch": 2.3863057917934074, "grad_norm": 0.48412978210175467, "learning_rate": 1.2214886426233351e-05, "loss": 2.715, "step": 38441 }, { "epoch": 2.3863678688931653, "grad_norm": 0.44076466139362463, "learning_rate": 1.2212521282287092e-05, "loss": 2.682, "step": 38442 }, { "epoch": 2.3864299459929232, "grad_norm": 0.4471265918243713, "learning_rate": 1.2210156335487282e-05, "loss": 2.7445, "step": 38443 }, { "epoch": 2.386492023092681, "grad_norm": 0.46582674225982085, "learning_rate": 1.2207791585846263e-05, "loss": 2.7055, "step": 38444 }, { "epoch": 2.386554100192439, "grad_norm": 0.4476357473634757, "learning_rate": 1.2205427033376348e-05, "loss": 2.6528, "step": 38445 }, { "epoch": 2.386616177292197, "grad_norm": 0.4638865898491595, "learning_rate": 1.2203062678089916e-05, "loss": 2.6669, "step": 38446 }, { "epoch": 2.386678254391955, "grad_norm": 0.4604107708215445, "learning_rate": 1.2200698519999282e-05, "loss": 2.6072, "step": 38447 }, { "epoch": 2.386740331491713, "grad_norm": 0.47753584847022007, "learning_rate": 1.2198334559116782e-05, "loss": 2.712, "step": 38448 }, { "epoch": 2.3868024085914707, "grad_norm": 0.4504165602244846, "learning_rate": 1.2195970795454725e-05, "loss": 2.6306, "step": 38449 }, { "epoch": 2.3868644856912287, "grad_norm": 0.43837628141770113, "learning_rate": 1.2193607229025483e-05, "loss": 2.6978, "step": 38450 }, { "epoch": 2.3869265627909866, "grad_norm": 0.44529453040558203, "learning_rate": 1.2191243859841373e-05, "loss": 2.665, "step": 38451 }, { "epoch": 2.3869886398907445, "grad_norm": 0.443352468159522, "learning_rate": 1.2188880687914723e-05, "loss": 2.6752, "step": 38452 }, { "epoch": 2.387050716990502, "grad_norm": 0.4618153859847365, "learning_rate": 1.218651771325784e-05, "loss": 2.6169, "step": 38453 }, { "epoch": 2.3871127940902603, "grad_norm": 0.43386086836176735, "learning_rate": 1.218415493588309e-05, "loss": 2.6897, "step": 38454 }, { "epoch": 2.387174871190018, "grad_norm": 0.43706669807899945, "learning_rate": 1.2181792355802773e-05, "loss": 2.704, "step": 38455 }, { "epoch": 2.3872369482897757, "grad_norm": 0.4530828555009119, "learning_rate": 1.2179429973029244e-05, "loss": 2.7092, "step": 38456 }, { "epoch": 2.3872990253895336, "grad_norm": 0.45027578631751713, "learning_rate": 1.2177067787574802e-05, "loss": 2.6456, "step": 38457 }, { "epoch": 2.3873611024892916, "grad_norm": 0.44747406879074325, "learning_rate": 1.2174705799451785e-05, "loss": 2.6674, "step": 38458 }, { "epoch": 2.3874231795890495, "grad_norm": 0.46436666802059073, "learning_rate": 1.2172344008672493e-05, "loss": 2.7097, "step": 38459 }, { "epoch": 2.3874852566888074, "grad_norm": 0.4859607795217226, "learning_rate": 1.2169982415249276e-05, "loss": 2.6956, "step": 38460 }, { "epoch": 2.3875473337885653, "grad_norm": 0.46454578755429055, "learning_rate": 1.2167621019194447e-05, "loss": 2.6541, "step": 38461 }, { "epoch": 2.3876094108883232, "grad_norm": 0.4378472210541834, "learning_rate": 1.2165259820520324e-05, "loss": 2.678, "step": 38462 }, { "epoch": 2.387671487988081, "grad_norm": 0.47846928586214155, "learning_rate": 1.2162898819239216e-05, "loss": 2.6948, "step": 38463 }, { "epoch": 2.387733565087839, "grad_norm": 0.4470962803516334, "learning_rate": 1.216053801536346e-05, "loss": 2.7286, "step": 38464 }, { "epoch": 2.387795642187597, "grad_norm": 0.4546120512195724, "learning_rate": 1.2158177408905364e-05, "loss": 2.6592, "step": 38465 }, { "epoch": 2.387857719287355, "grad_norm": 0.4902836444306817, "learning_rate": 1.2155816999877239e-05, "loss": 2.6919, "step": 38466 }, { "epoch": 2.387919796387113, "grad_norm": 0.40506970903561745, "learning_rate": 1.2153456788291407e-05, "loss": 2.6161, "step": 38467 }, { "epoch": 2.3879818734868707, "grad_norm": 0.4669097380883467, "learning_rate": 1.2151096774160164e-05, "loss": 2.7172, "step": 38468 }, { "epoch": 2.3880439505866287, "grad_norm": 0.5195962105200255, "learning_rate": 1.2148736957495848e-05, "loss": 2.7445, "step": 38469 }, { "epoch": 2.3881060276863866, "grad_norm": 0.45332014880163024, "learning_rate": 1.2146377338310766e-05, "loss": 2.7354, "step": 38470 }, { "epoch": 2.3881681047861445, "grad_norm": 0.47340428453012623, "learning_rate": 1.2144017916617218e-05, "loss": 2.7147, "step": 38471 }, { "epoch": 2.3882301818859024, "grad_norm": 0.5486738592256161, "learning_rate": 1.214165869242751e-05, "loss": 2.7036, "step": 38472 }, { "epoch": 2.3882922589856603, "grad_norm": 0.47245300273689933, "learning_rate": 1.2139299665753968e-05, "loss": 2.6389, "step": 38473 }, { "epoch": 2.3883543360854183, "grad_norm": 0.4722295679505909, "learning_rate": 1.213694083660889e-05, "loss": 2.7077, "step": 38474 }, { "epoch": 2.388416413185176, "grad_norm": 0.4946630528253528, "learning_rate": 1.2134582205004585e-05, "loss": 2.6409, "step": 38475 }, { "epoch": 2.3884784902849336, "grad_norm": 0.44715761771283835, "learning_rate": 1.2132223770953344e-05, "loss": 2.7037, "step": 38476 }, { "epoch": 2.388540567384692, "grad_norm": 0.5043258103181251, "learning_rate": 1.21298655344675e-05, "loss": 2.6852, "step": 38477 }, { "epoch": 2.3886026444844495, "grad_norm": 0.4795318556030069, "learning_rate": 1.2127507495559337e-05, "loss": 2.6071, "step": 38478 }, { "epoch": 2.3886647215842074, "grad_norm": 0.4682436274274608, "learning_rate": 1.2125149654241158e-05, "loss": 2.6849, "step": 38479 }, { "epoch": 2.3887267986839653, "grad_norm": 0.45848613236628755, "learning_rate": 1.2122792010525274e-05, "loss": 2.706, "step": 38480 }, { "epoch": 2.3887888757837232, "grad_norm": 0.4459311053506597, "learning_rate": 1.2120434564423965e-05, "loss": 2.6381, "step": 38481 }, { "epoch": 2.388850952883481, "grad_norm": 0.4652520638409344, "learning_rate": 1.2118077315949554e-05, "loss": 2.6837, "step": 38482 }, { "epoch": 2.388913029983239, "grad_norm": 0.44837736863041594, "learning_rate": 1.2115720265114333e-05, "loss": 2.5757, "step": 38483 }, { "epoch": 2.388975107082997, "grad_norm": 0.46532791632226855, "learning_rate": 1.2113363411930595e-05, "loss": 2.7141, "step": 38484 }, { "epoch": 2.389037184182755, "grad_norm": 0.4406373238700105, "learning_rate": 1.2111006756410625e-05, "loss": 2.725, "step": 38485 }, { "epoch": 2.389099261282513, "grad_norm": 0.5015413250122445, "learning_rate": 1.2108650298566727e-05, "loss": 2.7181, "step": 38486 }, { "epoch": 2.3891613383822707, "grad_norm": 0.4359058218216945, "learning_rate": 1.2106294038411214e-05, "loss": 2.6773, "step": 38487 }, { "epoch": 2.3892234154820287, "grad_norm": 0.4605876435410094, "learning_rate": 1.210393797595637e-05, "loss": 2.648, "step": 38488 }, { "epoch": 2.3892854925817866, "grad_norm": 0.4895690415372812, "learning_rate": 1.2101582111214478e-05, "loss": 2.6636, "step": 38489 }, { "epoch": 2.3893475696815445, "grad_norm": 0.43259090576250947, "learning_rate": 1.2099226444197825e-05, "loss": 2.7018, "step": 38490 }, { "epoch": 2.3894096467813024, "grad_norm": 0.5038151621736968, "learning_rate": 1.20968709749187e-05, "loss": 2.6791, "step": 38491 }, { "epoch": 2.3894717238810603, "grad_norm": 0.5010187082985422, "learning_rate": 1.2094515703389419e-05, "loss": 2.7565, "step": 38492 }, { "epoch": 2.3895338009808182, "grad_norm": 0.4090664326746123, "learning_rate": 1.2092160629622246e-05, "loss": 2.6528, "step": 38493 }, { "epoch": 2.389595878080576, "grad_norm": 0.4710469236872114, "learning_rate": 1.2089805753629473e-05, "loss": 2.7275, "step": 38494 }, { "epoch": 2.389657955180334, "grad_norm": 0.4669562829740392, "learning_rate": 1.2087451075423373e-05, "loss": 2.7268, "step": 38495 }, { "epoch": 2.389720032280092, "grad_norm": 0.44780385644614285, "learning_rate": 1.2085096595016259e-05, "loss": 2.6654, "step": 38496 }, { "epoch": 2.38978210937985, "grad_norm": 0.451314003756334, "learning_rate": 1.2082742312420403e-05, "loss": 2.6049, "step": 38497 }, { "epoch": 2.389844186479608, "grad_norm": 0.5031031337274969, "learning_rate": 1.2080388227648081e-05, "loss": 2.7157, "step": 38498 }, { "epoch": 2.3899062635793658, "grad_norm": 0.45428720596795613, "learning_rate": 1.207803434071157e-05, "loss": 2.676, "step": 38499 }, { "epoch": 2.3899683406791237, "grad_norm": 0.4413849141863748, "learning_rate": 1.2075680651623167e-05, "loss": 2.6587, "step": 38500 }, { "epoch": 2.390030417778881, "grad_norm": 0.4974835819381958, "learning_rate": 1.2073327160395149e-05, "loss": 2.7087, "step": 38501 }, { "epoch": 2.3900924948786395, "grad_norm": 0.4767528841544549, "learning_rate": 1.2070973867039787e-05, "loss": 2.7152, "step": 38502 }, { "epoch": 2.390154571978397, "grad_norm": 0.4863699929473018, "learning_rate": 1.2068620771569367e-05, "loss": 2.5498, "step": 38503 }, { "epoch": 2.390216649078155, "grad_norm": 0.4820350764291201, "learning_rate": 1.206626787399614e-05, "loss": 2.7278, "step": 38504 }, { "epoch": 2.390278726177913, "grad_norm": 0.46145731500322373, "learning_rate": 1.2063915174332419e-05, "loss": 2.7205, "step": 38505 }, { "epoch": 2.3903408032776707, "grad_norm": 0.47065245486216695, "learning_rate": 1.2061562672590465e-05, "loss": 2.6734, "step": 38506 }, { "epoch": 2.3904028803774287, "grad_norm": 0.4545816816176988, "learning_rate": 1.2059210368782548e-05, "loss": 2.7306, "step": 38507 }, { "epoch": 2.3904649574771866, "grad_norm": 0.43943267774696254, "learning_rate": 1.2056858262920922e-05, "loss": 2.6127, "step": 38508 }, { "epoch": 2.3905270345769445, "grad_norm": 0.4420023581109642, "learning_rate": 1.2054506355017898e-05, "loss": 2.6429, "step": 38509 }, { "epoch": 2.3905891116767024, "grad_norm": 0.4751866130918419, "learning_rate": 1.2052154645085717e-05, "loss": 2.7094, "step": 38510 }, { "epoch": 2.3906511887764603, "grad_norm": 0.4816794764185947, "learning_rate": 1.2049803133136667e-05, "loss": 2.7445, "step": 38511 }, { "epoch": 2.3907132658762182, "grad_norm": 0.4491798218155403, "learning_rate": 1.2047451819182987e-05, "loss": 2.6023, "step": 38512 }, { "epoch": 2.390775342975976, "grad_norm": 0.4243959236043178, "learning_rate": 1.2045100703236978e-05, "loss": 2.6811, "step": 38513 }, { "epoch": 2.390837420075734, "grad_norm": 0.472920840268757, "learning_rate": 1.2042749785310898e-05, "loss": 2.6888, "step": 38514 }, { "epoch": 2.390899497175492, "grad_norm": 0.44720500218697934, "learning_rate": 1.2040399065417001e-05, "loss": 2.6282, "step": 38515 }, { "epoch": 2.39096157427525, "grad_norm": 0.4571261674179179, "learning_rate": 1.203804854356756e-05, "loss": 2.7063, "step": 38516 }, { "epoch": 2.391023651375008, "grad_norm": 0.4995081389626125, "learning_rate": 1.2035698219774816e-05, "loss": 2.7263, "step": 38517 }, { "epoch": 2.3910857284747657, "grad_norm": 0.452213818167146, "learning_rate": 1.2033348094051056e-05, "loss": 2.7499, "step": 38518 }, { "epoch": 2.3911478055745237, "grad_norm": 0.45071678824633576, "learning_rate": 1.2030998166408547e-05, "loss": 2.6308, "step": 38519 }, { "epoch": 2.3912098826742816, "grad_norm": 0.44992759175021235, "learning_rate": 1.2028648436859541e-05, "loss": 2.6099, "step": 38520 }, { "epoch": 2.3912719597740395, "grad_norm": 0.4853264324352395, "learning_rate": 1.2026298905416294e-05, "loss": 2.7211, "step": 38521 }, { "epoch": 2.3913340368737974, "grad_norm": 0.4808998344995409, "learning_rate": 1.2023949572091048e-05, "loss": 2.6859, "step": 38522 }, { "epoch": 2.3913961139735553, "grad_norm": 0.4830940906857326, "learning_rate": 1.2021600436896091e-05, "loss": 2.7374, "step": 38523 }, { "epoch": 2.391458191073313, "grad_norm": 0.44701671754054895, "learning_rate": 1.2019251499843664e-05, "loss": 2.6327, "step": 38524 }, { "epoch": 2.391520268173071, "grad_norm": 0.49356404315639135, "learning_rate": 1.201690276094602e-05, "loss": 2.6851, "step": 38525 }, { "epoch": 2.3915823452728286, "grad_norm": 0.4657857150299216, "learning_rate": 1.2014554220215418e-05, "loss": 2.6018, "step": 38526 }, { "epoch": 2.3916444223725866, "grad_norm": 0.45660430985700423, "learning_rate": 1.201220587766409e-05, "loss": 2.6247, "step": 38527 }, { "epoch": 2.3917064994723445, "grad_norm": 0.5207613100480133, "learning_rate": 1.2009857733304324e-05, "loss": 2.6656, "step": 38528 }, { "epoch": 2.3917685765721024, "grad_norm": 0.4885799257416723, "learning_rate": 1.2007509787148353e-05, "loss": 2.6461, "step": 38529 }, { "epoch": 2.3918306536718603, "grad_norm": 0.4736300753484859, "learning_rate": 1.2005162039208423e-05, "loss": 2.6121, "step": 38530 }, { "epoch": 2.3918927307716182, "grad_norm": 0.4957500679027609, "learning_rate": 1.2002814489496772e-05, "loss": 2.5913, "step": 38531 }, { "epoch": 2.391954807871376, "grad_norm": 0.5068252062534305, "learning_rate": 1.2000467138025678e-05, "loss": 2.6552, "step": 38532 }, { "epoch": 2.392016884971134, "grad_norm": 0.4593047464588089, "learning_rate": 1.199811998480737e-05, "loss": 2.752, "step": 38533 }, { "epoch": 2.392078962070892, "grad_norm": 0.5094830252184694, "learning_rate": 1.1995773029854102e-05, "loss": 2.7494, "step": 38534 }, { "epoch": 2.39214103917065, "grad_norm": 0.4915656748461188, "learning_rate": 1.1993426273178087e-05, "loss": 2.6595, "step": 38535 }, { "epoch": 2.392203116270408, "grad_norm": 0.4681400304044237, "learning_rate": 1.1991079714791614e-05, "loss": 2.6649, "step": 38536 }, { "epoch": 2.3922651933701657, "grad_norm": 0.4743127248996991, "learning_rate": 1.1988733354706905e-05, "loss": 2.6754, "step": 38537 }, { "epoch": 2.3923272704699237, "grad_norm": 0.4527590523577861, "learning_rate": 1.1986387192936205e-05, "loss": 2.6998, "step": 38538 }, { "epoch": 2.3923893475696816, "grad_norm": 0.4402623842090792, "learning_rate": 1.1984041229491744e-05, "loss": 2.7154, "step": 38539 }, { "epoch": 2.3924514246694395, "grad_norm": 0.482283058735082, "learning_rate": 1.1981695464385756e-05, "loss": 2.7238, "step": 38540 }, { "epoch": 2.3925135017691974, "grad_norm": 0.47022614113931194, "learning_rate": 1.197934989763051e-05, "loss": 2.6662, "step": 38541 }, { "epoch": 2.3925755788689553, "grad_norm": 0.4437326212958857, "learning_rate": 1.1977004529238223e-05, "loss": 2.5944, "step": 38542 }, { "epoch": 2.3926376559687133, "grad_norm": 0.450418115125945, "learning_rate": 1.1974659359221135e-05, "loss": 2.675, "step": 38543 }, { "epoch": 2.392699733068471, "grad_norm": 0.4453827986525597, "learning_rate": 1.1972314387591466e-05, "loss": 2.6333, "step": 38544 }, { "epoch": 2.392761810168229, "grad_norm": 0.45069706691056693, "learning_rate": 1.1969969614361476e-05, "loss": 2.6279, "step": 38545 }, { "epoch": 2.392823887267987, "grad_norm": 0.4781211316818629, "learning_rate": 1.196762503954339e-05, "loss": 2.659, "step": 38546 }, { "epoch": 2.392885964367745, "grad_norm": 0.4419623559989471, "learning_rate": 1.196528066314943e-05, "loss": 2.7305, "step": 38547 }, { "epoch": 2.392948041467503, "grad_norm": 0.44943142707762973, "learning_rate": 1.1962936485191828e-05, "loss": 2.7307, "step": 38548 }, { "epoch": 2.3930101185672603, "grad_norm": 0.49752057260600424, "learning_rate": 1.1960592505682832e-05, "loss": 2.6968, "step": 38549 }, { "epoch": 2.3930721956670187, "grad_norm": 0.45990375398213984, "learning_rate": 1.1958248724634658e-05, "loss": 2.6092, "step": 38550 }, { "epoch": 2.393134272766776, "grad_norm": 0.45031366013239155, "learning_rate": 1.195590514205952e-05, "loss": 2.6149, "step": 38551 }, { "epoch": 2.393196349866534, "grad_norm": 0.4881753272562185, "learning_rate": 1.1953561757969678e-05, "loss": 2.7363, "step": 38552 }, { "epoch": 2.393258426966292, "grad_norm": 0.4455728773529136, "learning_rate": 1.1951218572377338e-05, "loss": 2.6152, "step": 38553 }, { "epoch": 2.39332050406605, "grad_norm": 0.46977961631201054, "learning_rate": 1.1948875585294711e-05, "loss": 2.6777, "step": 38554 }, { "epoch": 2.393382581165808, "grad_norm": 0.4530446211457371, "learning_rate": 1.1946532796734056e-05, "loss": 2.6876, "step": 38555 }, { "epoch": 2.3934446582655657, "grad_norm": 0.45524815014982334, "learning_rate": 1.1944190206707573e-05, "loss": 2.6108, "step": 38556 }, { "epoch": 2.3935067353653237, "grad_norm": 0.4916871418838293, "learning_rate": 1.1941847815227492e-05, "loss": 2.675, "step": 38557 }, { "epoch": 2.3935688124650816, "grad_norm": 0.4801511412577486, "learning_rate": 1.1939505622306013e-05, "loss": 2.6547, "step": 38558 }, { "epoch": 2.3936308895648395, "grad_norm": 0.43770759629967193, "learning_rate": 1.1937163627955389e-05, "loss": 2.6531, "step": 38559 }, { "epoch": 2.3936929666645974, "grad_norm": 0.477765862214053, "learning_rate": 1.193482183218782e-05, "loss": 2.5918, "step": 38560 }, { "epoch": 2.3937550437643553, "grad_norm": 0.4902375665689966, "learning_rate": 1.193248023501553e-05, "loss": 2.6888, "step": 38561 }, { "epoch": 2.3938171208641132, "grad_norm": 0.4719879059525152, "learning_rate": 1.1930138836450716e-05, "loss": 2.6688, "step": 38562 }, { "epoch": 2.393879197963871, "grad_norm": 0.49007752177515884, "learning_rate": 1.1927797636505622e-05, "loss": 2.7075, "step": 38563 }, { "epoch": 2.393941275063629, "grad_norm": 0.4463235687338465, "learning_rate": 1.1925456635192451e-05, "loss": 2.7163, "step": 38564 }, { "epoch": 2.394003352163387, "grad_norm": 0.4337391480674139, "learning_rate": 1.1923115832523419e-05, "loss": 2.6635, "step": 38565 }, { "epoch": 2.394065429263145, "grad_norm": 0.4961965409365873, "learning_rate": 1.1920775228510728e-05, "loss": 2.6311, "step": 38566 }, { "epoch": 2.394127506362903, "grad_norm": 0.47353171616569517, "learning_rate": 1.1918434823166585e-05, "loss": 2.6696, "step": 38567 }, { "epoch": 2.3941895834626608, "grad_norm": 0.4140049156597415, "learning_rate": 1.1916094616503232e-05, "loss": 2.7373, "step": 38568 }, { "epoch": 2.3942516605624187, "grad_norm": 0.4467846979255133, "learning_rate": 1.191375460853285e-05, "loss": 2.7104, "step": 38569 }, { "epoch": 2.3943137376621766, "grad_norm": 0.46459734821420823, "learning_rate": 1.1911414799267662e-05, "loss": 2.6209, "step": 38570 }, { "epoch": 2.3943758147619345, "grad_norm": 0.447966397282277, "learning_rate": 1.1909075188719853e-05, "loss": 2.6533, "step": 38571 }, { "epoch": 2.394437891861692, "grad_norm": 0.4489828070617667, "learning_rate": 1.1906735776901662e-05, "loss": 2.7331, "step": 38572 }, { "epoch": 2.3944999689614503, "grad_norm": 0.4459066609958126, "learning_rate": 1.1904396563825277e-05, "loss": 2.7842, "step": 38573 }, { "epoch": 2.394562046061208, "grad_norm": 0.49554718125821146, "learning_rate": 1.1902057549502899e-05, "loss": 2.7498, "step": 38574 }, { "epoch": 2.3946241231609657, "grad_norm": 0.43895420150671, "learning_rate": 1.189971873394674e-05, "loss": 2.6798, "step": 38575 }, { "epoch": 2.3946862002607237, "grad_norm": 0.475594694773671, "learning_rate": 1.1897380117168983e-05, "loss": 2.7205, "step": 38576 }, { "epoch": 2.3947482773604816, "grad_norm": 0.47121310011183953, "learning_rate": 1.1895041699181853e-05, "loss": 2.6188, "step": 38577 }, { "epoch": 2.3948103544602395, "grad_norm": 0.4501773061068203, "learning_rate": 1.1892703479997547e-05, "loss": 2.6657, "step": 38578 }, { "epoch": 2.3948724315599974, "grad_norm": 0.48091393292186096, "learning_rate": 1.1890365459628256e-05, "loss": 2.6498, "step": 38579 }, { "epoch": 2.3949345086597553, "grad_norm": 0.48476538769564786, "learning_rate": 1.1888027638086163e-05, "loss": 2.7001, "step": 38580 }, { "epoch": 2.3949965857595132, "grad_norm": 0.4494364003876332, "learning_rate": 1.1885690015383499e-05, "loss": 2.6482, "step": 38581 }, { "epoch": 2.395058662859271, "grad_norm": 0.46620383159334466, "learning_rate": 1.1883352591532442e-05, "loss": 2.7799, "step": 38582 }, { "epoch": 2.395120739959029, "grad_norm": 0.4805444177109838, "learning_rate": 1.188101536654519e-05, "loss": 2.6875, "step": 38583 }, { "epoch": 2.395182817058787, "grad_norm": 0.4762160808769547, "learning_rate": 1.1878678340433914e-05, "loss": 2.7066, "step": 38584 }, { "epoch": 2.395244894158545, "grad_norm": 0.4428275190995126, "learning_rate": 1.1876341513210831e-05, "loss": 2.499, "step": 38585 }, { "epoch": 2.395306971258303, "grad_norm": 0.4399923449630671, "learning_rate": 1.1874004884888145e-05, "loss": 2.6946, "step": 38586 }, { "epoch": 2.3953690483580607, "grad_norm": 0.4538355714853924, "learning_rate": 1.187166845547803e-05, "loss": 2.5927, "step": 38587 }, { "epoch": 2.3954311254578187, "grad_norm": 0.4546025127935967, "learning_rate": 1.1869332224992674e-05, "loss": 2.6952, "step": 38588 }, { "epoch": 2.3954932025575766, "grad_norm": 0.436929398780381, "learning_rate": 1.1866996193444274e-05, "loss": 2.5971, "step": 38589 }, { "epoch": 2.3955552796573345, "grad_norm": 0.4556244284753132, "learning_rate": 1.1864660360844993e-05, "loss": 2.7635, "step": 38590 }, { "epoch": 2.3956173567570924, "grad_norm": 0.4354967165406018, "learning_rate": 1.1862324727207052e-05, "loss": 2.7638, "step": 38591 }, { "epoch": 2.3956794338568503, "grad_norm": 0.42712714583047956, "learning_rate": 1.1859989292542617e-05, "loss": 2.598, "step": 38592 }, { "epoch": 2.3957415109566083, "grad_norm": 0.44493282596015543, "learning_rate": 1.1857654056863882e-05, "loss": 2.6378, "step": 38593 }, { "epoch": 2.395803588056366, "grad_norm": 0.4597254998937639, "learning_rate": 1.185531902018301e-05, "loss": 2.7623, "step": 38594 }, { "epoch": 2.395865665156124, "grad_norm": 0.4922318211536365, "learning_rate": 1.1852984182512211e-05, "loss": 2.6511, "step": 38595 }, { "epoch": 2.395927742255882, "grad_norm": 0.42695555481214936, "learning_rate": 1.1850649543863657e-05, "loss": 2.5175, "step": 38596 }, { "epoch": 2.3959898193556395, "grad_norm": 0.4510430321294659, "learning_rate": 1.1848315104249524e-05, "loss": 2.6173, "step": 38597 }, { "epoch": 2.396051896455398, "grad_norm": 0.4802568324981655, "learning_rate": 1.1845980863681983e-05, "loss": 2.7054, "step": 38598 }, { "epoch": 2.3961139735551553, "grad_norm": 0.4646499133477458, "learning_rate": 1.1843646822173227e-05, "loss": 2.7258, "step": 38599 }, { "epoch": 2.3961760506549132, "grad_norm": 0.44787805597836816, "learning_rate": 1.1841312979735437e-05, "loss": 2.6478, "step": 38600 }, { "epoch": 2.396238127754671, "grad_norm": 0.47182413420720287, "learning_rate": 1.1838979336380774e-05, "loss": 2.6554, "step": 38601 }, { "epoch": 2.396300204854429, "grad_norm": 0.48869341430031965, "learning_rate": 1.1836645892121423e-05, "loss": 2.632, "step": 38602 }, { "epoch": 2.396362281954187, "grad_norm": 0.46287988899753935, "learning_rate": 1.1834312646969543e-05, "loss": 2.6311, "step": 38603 }, { "epoch": 2.396424359053945, "grad_norm": 0.4786084921892443, "learning_rate": 1.1831979600937327e-05, "loss": 2.6414, "step": 38604 }, { "epoch": 2.396486436153703, "grad_norm": 0.491050542350099, "learning_rate": 1.1829646754036944e-05, "loss": 2.7052, "step": 38605 }, { "epoch": 2.3965485132534607, "grad_norm": 0.46233701798364935, "learning_rate": 1.1827314106280557e-05, "loss": 2.7103, "step": 38606 }, { "epoch": 2.3966105903532187, "grad_norm": 0.4799771003178075, "learning_rate": 1.1824981657680323e-05, "loss": 2.7566, "step": 38607 }, { "epoch": 2.3966726674529766, "grad_norm": 0.45092568554539286, "learning_rate": 1.1822649408248449e-05, "loss": 2.6921, "step": 38608 }, { "epoch": 2.3967347445527345, "grad_norm": 0.4176854751243297, "learning_rate": 1.1820317357997073e-05, "loss": 2.59, "step": 38609 }, { "epoch": 2.3967968216524924, "grad_norm": 0.4678005935336144, "learning_rate": 1.1817985506938374e-05, "loss": 2.5908, "step": 38610 }, { "epoch": 2.3968588987522503, "grad_norm": 0.47365143729488846, "learning_rate": 1.1815653855084513e-05, "loss": 2.6994, "step": 38611 }, { "epoch": 2.3969209758520083, "grad_norm": 0.40296750651622903, "learning_rate": 1.1813322402447641e-05, "loss": 2.611, "step": 38612 }, { "epoch": 2.396983052951766, "grad_norm": 0.43116411169285784, "learning_rate": 1.1810991149039951e-05, "loss": 2.7012, "step": 38613 }, { "epoch": 2.397045130051524, "grad_norm": 0.44664147817821465, "learning_rate": 1.1808660094873585e-05, "loss": 2.7131, "step": 38614 }, { "epoch": 2.397107207151282, "grad_norm": 0.4360068761661748, "learning_rate": 1.1806329239960717e-05, "loss": 2.6441, "step": 38615 }, { "epoch": 2.39716928425104, "grad_norm": 0.43942940541997805, "learning_rate": 1.1803998584313485e-05, "loss": 2.7125, "step": 38616 }, { "epoch": 2.397231361350798, "grad_norm": 0.43406421466046563, "learning_rate": 1.1801668127944066e-05, "loss": 2.5915, "step": 38617 }, { "epoch": 2.3972934384505558, "grad_norm": 0.45008151918709677, "learning_rate": 1.1799337870864634e-05, "loss": 2.6203, "step": 38618 }, { "epoch": 2.3973555155503137, "grad_norm": 0.4704294160777019, "learning_rate": 1.1797007813087324e-05, "loss": 2.7006, "step": 38619 }, { "epoch": 2.397417592650071, "grad_norm": 0.4154605654928573, "learning_rate": 1.17946779546243e-05, "loss": 2.6587, "step": 38620 }, { "epoch": 2.3974796697498295, "grad_norm": 0.4381456889656432, "learning_rate": 1.1792348295487705e-05, "loss": 2.5985, "step": 38621 }, { "epoch": 2.397541746849587, "grad_norm": 0.41599889730573275, "learning_rate": 1.1790018835689714e-05, "loss": 2.6417, "step": 38622 }, { "epoch": 2.397603823949345, "grad_norm": 0.41870624327855266, "learning_rate": 1.1787689575242477e-05, "loss": 2.7133, "step": 38623 }, { "epoch": 2.397665901049103, "grad_norm": 0.42168489340766263, "learning_rate": 1.1785360514158133e-05, "loss": 2.6863, "step": 38624 }, { "epoch": 2.3977279781488607, "grad_norm": 0.45167981903990073, "learning_rate": 1.1783031652448844e-05, "loss": 2.5919, "step": 38625 }, { "epoch": 2.3977900552486187, "grad_norm": 0.4338001241222554, "learning_rate": 1.178070299012674e-05, "loss": 2.6465, "step": 38626 }, { "epoch": 2.3978521323483766, "grad_norm": 0.4037514292098221, "learning_rate": 1.1778374527204e-05, "loss": 2.6512, "step": 38627 }, { "epoch": 2.3979142094481345, "grad_norm": 0.45223960204583696, "learning_rate": 1.1776046263692764e-05, "loss": 2.5809, "step": 38628 }, { "epoch": 2.3979762865478924, "grad_norm": 0.4305544170839715, "learning_rate": 1.1773718199605165e-05, "loss": 2.682, "step": 38629 }, { "epoch": 2.3980383636476503, "grad_norm": 0.4601113140399438, "learning_rate": 1.1771390334953353e-05, "loss": 2.7863, "step": 38630 }, { "epoch": 2.3981004407474082, "grad_norm": 0.43478962368463087, "learning_rate": 1.1769062669749487e-05, "loss": 2.6684, "step": 38631 }, { "epoch": 2.398162517847166, "grad_norm": 0.4749185473474063, "learning_rate": 1.1766735204005703e-05, "loss": 2.7405, "step": 38632 }, { "epoch": 2.398224594946924, "grad_norm": 0.43510210737977484, "learning_rate": 1.176440793773414e-05, "loss": 2.7548, "step": 38633 }, { "epoch": 2.398286672046682, "grad_norm": 0.4481722376146044, "learning_rate": 1.1762080870946934e-05, "loss": 2.7337, "step": 38634 }, { "epoch": 2.39834874914644, "grad_norm": 0.4725142967241024, "learning_rate": 1.1759754003656243e-05, "loss": 2.6726, "step": 38635 }, { "epoch": 2.398410826246198, "grad_norm": 0.4765693476450756, "learning_rate": 1.1757427335874204e-05, "loss": 2.6549, "step": 38636 }, { "epoch": 2.3984729033459558, "grad_norm": 0.44155663485725233, "learning_rate": 1.175510086761295e-05, "loss": 2.6257, "step": 38637 }, { "epoch": 2.3985349804457137, "grad_norm": 0.4546053645089332, "learning_rate": 1.1752774598884614e-05, "loss": 2.636, "step": 38638 }, { "epoch": 2.3985970575454716, "grad_norm": 0.49254775108692955, "learning_rate": 1.175044852970133e-05, "loss": 2.6209, "step": 38639 }, { "epoch": 2.3986591346452295, "grad_norm": 0.4774094856313867, "learning_rate": 1.1748122660075251e-05, "loss": 2.683, "step": 38640 }, { "epoch": 2.3987212117449874, "grad_norm": 0.5052648755186044, "learning_rate": 1.1745796990018503e-05, "loss": 2.6799, "step": 38641 }, { "epoch": 2.3987832888447453, "grad_norm": 0.4478273986606079, "learning_rate": 1.174347151954322e-05, "loss": 2.7065, "step": 38642 }, { "epoch": 2.3988453659445033, "grad_norm": 0.4916368461983769, "learning_rate": 1.1741146248661517e-05, "loss": 2.6827, "step": 38643 }, { "epoch": 2.398907443044261, "grad_norm": 0.47733113573454655, "learning_rate": 1.1738821177385556e-05, "loss": 2.8258, "step": 38644 }, { "epoch": 2.3989695201440187, "grad_norm": 0.4275802961866253, "learning_rate": 1.1736496305727451e-05, "loss": 2.5979, "step": 38645 }, { "epoch": 2.399031597243777, "grad_norm": 0.5145229063449369, "learning_rate": 1.1734171633699337e-05, "loss": 2.6569, "step": 38646 }, { "epoch": 2.3990936743435345, "grad_norm": 0.517578009839315, "learning_rate": 1.1731847161313325e-05, "loss": 2.6819, "step": 38647 }, { "epoch": 2.3991557514432924, "grad_norm": 0.4476546382375731, "learning_rate": 1.1729522888581574e-05, "loss": 2.6798, "step": 38648 }, { "epoch": 2.3992178285430503, "grad_norm": 0.5264458535600557, "learning_rate": 1.1727198815516172e-05, "loss": 2.6592, "step": 38649 }, { "epoch": 2.3992799056428082, "grad_norm": 0.47226131112247993, "learning_rate": 1.1724874942129283e-05, "loss": 2.5692, "step": 38650 }, { "epoch": 2.399341982742566, "grad_norm": 0.4794831284795735, "learning_rate": 1.1722551268433008e-05, "loss": 2.6726, "step": 38651 }, { "epoch": 2.399404059842324, "grad_norm": 0.46238353977015656, "learning_rate": 1.1720227794439481e-05, "loss": 2.6582, "step": 38652 }, { "epoch": 2.399466136942082, "grad_norm": 0.4868710409745347, "learning_rate": 1.17179045201608e-05, "loss": 2.7133, "step": 38653 }, { "epoch": 2.39952821404184, "grad_norm": 0.4713915938241265, "learning_rate": 1.1715581445609119e-05, "loss": 2.6299, "step": 38654 }, { "epoch": 2.399590291141598, "grad_norm": 0.46363747781689363, "learning_rate": 1.1713258570796542e-05, "loss": 2.593, "step": 38655 }, { "epoch": 2.3996523682413557, "grad_norm": 0.47882411121709556, "learning_rate": 1.1710935895735193e-05, "loss": 2.6689, "step": 38656 }, { "epoch": 2.3997144453411137, "grad_norm": 0.44979363849037474, "learning_rate": 1.1708613420437176e-05, "loss": 2.6864, "step": 38657 }, { "epoch": 2.3997765224408716, "grad_norm": 0.5026426811153677, "learning_rate": 1.1706291144914627e-05, "loss": 2.6897, "step": 38658 }, { "epoch": 2.3998385995406295, "grad_norm": 0.45746728686334487, "learning_rate": 1.1703969069179654e-05, "loss": 2.5933, "step": 38659 }, { "epoch": 2.3999006766403874, "grad_norm": 0.4688421158879389, "learning_rate": 1.170164719324437e-05, "loss": 2.6571, "step": 38660 }, { "epoch": 2.3999627537401453, "grad_norm": 0.47990912123507595, "learning_rate": 1.169932551712089e-05, "loss": 2.6929, "step": 38661 }, { "epoch": 2.4000248308399033, "grad_norm": 0.4481707580433324, "learning_rate": 1.1697004040821314e-05, "loss": 2.6823, "step": 38662 }, { "epoch": 2.400086907939661, "grad_norm": 0.48851049001103597, "learning_rate": 1.1694682764357779e-05, "loss": 2.626, "step": 38663 }, { "epoch": 2.400148985039419, "grad_norm": 0.5472319664090273, "learning_rate": 1.1692361687742381e-05, "loss": 2.6357, "step": 38664 }, { "epoch": 2.400211062139177, "grad_norm": 0.4665501153813527, "learning_rate": 1.1690040810987236e-05, "loss": 2.7016, "step": 38665 }, { "epoch": 2.400273139238935, "grad_norm": 0.49815982211077636, "learning_rate": 1.1687720134104434e-05, "loss": 2.6773, "step": 38666 }, { "epoch": 2.400335216338693, "grad_norm": 0.4921011621952965, "learning_rate": 1.1685399657106111e-05, "loss": 2.7431, "step": 38667 }, { "epoch": 2.4003972934384503, "grad_norm": 0.4677076353168447, "learning_rate": 1.1683079380004353e-05, "loss": 2.6776, "step": 38668 }, { "epoch": 2.4004593705382087, "grad_norm": 0.4131488256177631, "learning_rate": 1.1680759302811278e-05, "loss": 2.672, "step": 38669 }, { "epoch": 2.400521447637966, "grad_norm": 0.5042374875110106, "learning_rate": 1.1678439425538968e-05, "loss": 2.7135, "step": 38670 }, { "epoch": 2.400583524737724, "grad_norm": 0.5044204690473649, "learning_rate": 1.1676119748199554e-05, "loss": 2.7872, "step": 38671 }, { "epoch": 2.400645601837482, "grad_norm": 0.4805756271708665, "learning_rate": 1.167380027080513e-05, "loss": 2.7009, "step": 38672 }, { "epoch": 2.40070767893724, "grad_norm": 0.5142953953844112, "learning_rate": 1.1671480993367794e-05, "loss": 2.728, "step": 38673 }, { "epoch": 2.400769756036998, "grad_norm": 0.428092719343802, "learning_rate": 1.1669161915899646e-05, "loss": 2.6063, "step": 38674 }, { "epoch": 2.4008318331367557, "grad_norm": 0.45683757232073413, "learning_rate": 1.1666843038412772e-05, "loss": 2.5893, "step": 38675 }, { "epoch": 2.4008939102365137, "grad_norm": 0.4596176162024959, "learning_rate": 1.1664524360919293e-05, "loss": 2.6272, "step": 38676 }, { "epoch": 2.4009559873362716, "grad_norm": 0.44811496300410303, "learning_rate": 1.1662205883431303e-05, "loss": 2.6609, "step": 38677 }, { "epoch": 2.4010180644360295, "grad_norm": 0.49975630662802106, "learning_rate": 1.165988760596089e-05, "loss": 2.7299, "step": 38678 }, { "epoch": 2.4010801415357874, "grad_norm": 0.5094244973876558, "learning_rate": 1.1657569528520135e-05, "loss": 2.6847, "step": 38679 }, { "epoch": 2.4011422186355453, "grad_norm": 0.40963923247787765, "learning_rate": 1.1655251651121158e-05, "loss": 2.5721, "step": 38680 }, { "epoch": 2.4012042957353033, "grad_norm": 0.46151154543568523, "learning_rate": 1.1652933973776048e-05, "loss": 2.6307, "step": 38681 }, { "epoch": 2.401266372835061, "grad_norm": 0.4915654323366738, "learning_rate": 1.1650616496496875e-05, "loss": 2.6529, "step": 38682 }, { "epoch": 2.401328449934819, "grad_norm": 0.47052601691563245, "learning_rate": 1.1648299219295755e-05, "loss": 2.7009, "step": 38683 }, { "epoch": 2.401390527034577, "grad_norm": 0.4536029662570141, "learning_rate": 1.1645982142184759e-05, "loss": 2.6258, "step": 38684 }, { "epoch": 2.401452604134335, "grad_norm": 0.4351266107789657, "learning_rate": 1.1643665265176001e-05, "loss": 2.6814, "step": 38685 }, { "epoch": 2.401514681234093, "grad_norm": 0.45661264795596845, "learning_rate": 1.1641348588281547e-05, "loss": 2.714, "step": 38686 }, { "epoch": 2.4015767583338508, "grad_norm": 0.512824719764039, "learning_rate": 1.1639032111513493e-05, "loss": 2.6728, "step": 38687 }, { "epoch": 2.4016388354336087, "grad_norm": 0.4778700697442726, "learning_rate": 1.163671583488392e-05, "loss": 2.658, "step": 38688 }, { "epoch": 2.4017009125333666, "grad_norm": 0.43245816375722235, "learning_rate": 1.1634399758404895e-05, "loss": 2.6773, "step": 38689 }, { "epoch": 2.4017629896331245, "grad_norm": 0.48496445813170486, "learning_rate": 1.1632083882088541e-05, "loss": 2.7181, "step": 38690 }, { "epoch": 2.4018250667328824, "grad_norm": 0.4801793351040918, "learning_rate": 1.1629768205946917e-05, "loss": 2.6965, "step": 38691 }, { "epoch": 2.4018871438326403, "grad_norm": 0.4725891727739673, "learning_rate": 1.1627452729992105e-05, "loss": 2.7316, "step": 38692 }, { "epoch": 2.401949220932398, "grad_norm": 0.4816419188396219, "learning_rate": 1.162513745423618e-05, "loss": 2.6695, "step": 38693 }, { "epoch": 2.402011298032156, "grad_norm": 0.4232243222988953, "learning_rate": 1.1622822378691234e-05, "loss": 2.6741, "step": 38694 }, { "epoch": 2.4020733751319137, "grad_norm": 0.4364875590964615, "learning_rate": 1.1620507503369343e-05, "loss": 2.6619, "step": 38695 }, { "epoch": 2.4021354522316716, "grad_norm": 0.4470888979720448, "learning_rate": 1.1618192828282587e-05, "loss": 2.7013, "step": 38696 }, { "epoch": 2.4021975293314295, "grad_norm": 0.4361992129026673, "learning_rate": 1.1615878353443017e-05, "loss": 2.7498, "step": 38697 }, { "epoch": 2.4022596064311874, "grad_norm": 0.4684788555495329, "learning_rate": 1.1613564078862738e-05, "loss": 2.6375, "step": 38698 }, { "epoch": 2.4023216835309453, "grad_norm": 0.44884458798858345, "learning_rate": 1.161125000455382e-05, "loss": 2.6906, "step": 38699 }, { "epoch": 2.4023837606307032, "grad_norm": 0.44279694698805233, "learning_rate": 1.1608936130528331e-05, "loss": 2.7258, "step": 38700 }, { "epoch": 2.402445837730461, "grad_norm": 0.5050584203618819, "learning_rate": 1.1606622456798338e-05, "loss": 2.7546, "step": 38701 }, { "epoch": 2.402507914830219, "grad_norm": 0.46303776523053275, "learning_rate": 1.16043089833759e-05, "loss": 2.6976, "step": 38702 }, { "epoch": 2.402569991929977, "grad_norm": 0.48025829078770016, "learning_rate": 1.1601995710273123e-05, "loss": 2.6784, "step": 38703 }, { "epoch": 2.402632069029735, "grad_norm": 0.4506189004376797, "learning_rate": 1.1599682637502052e-05, "loss": 2.6816, "step": 38704 }, { "epoch": 2.402694146129493, "grad_norm": 0.45185864185466496, "learning_rate": 1.1597369765074761e-05, "loss": 2.6793, "step": 38705 }, { "epoch": 2.4027562232292508, "grad_norm": 0.4356378537256097, "learning_rate": 1.1595057093003297e-05, "loss": 2.7386, "step": 38706 }, { "epoch": 2.4028183003290087, "grad_norm": 0.45531058206985553, "learning_rate": 1.1592744621299762e-05, "loss": 2.5809, "step": 38707 }, { "epoch": 2.4028803774287666, "grad_norm": 0.44840722795981025, "learning_rate": 1.1590432349976193e-05, "loss": 2.7114, "step": 38708 }, { "epoch": 2.4029424545285245, "grad_norm": 0.45143055621630446, "learning_rate": 1.1588120279044674e-05, "loss": 2.7214, "step": 38709 }, { "epoch": 2.4030045316282824, "grad_norm": 0.48554232127969005, "learning_rate": 1.1585808408517252e-05, "loss": 2.7069, "step": 38710 }, { "epoch": 2.4030666087280403, "grad_norm": 0.46591443084672046, "learning_rate": 1.1583496738405975e-05, "loss": 2.7028, "step": 38711 }, { "epoch": 2.4031286858277983, "grad_norm": 0.49452230918479667, "learning_rate": 1.1581185268722938e-05, "loss": 2.7012, "step": 38712 }, { "epoch": 2.403190762927556, "grad_norm": 0.4554344059606231, "learning_rate": 1.1578873999480183e-05, "loss": 2.5809, "step": 38713 }, { "epoch": 2.403252840027314, "grad_norm": 0.48347213645697346, "learning_rate": 1.1576562930689771e-05, "loss": 2.63, "step": 38714 }, { "epoch": 2.403314917127072, "grad_norm": 0.46679192704008443, "learning_rate": 1.1574252062363743e-05, "loss": 2.6982, "step": 38715 }, { "epoch": 2.4033769942268295, "grad_norm": 0.46046132206305707, "learning_rate": 1.1571941394514175e-05, "loss": 2.6661, "step": 38716 }, { "epoch": 2.403439071326588, "grad_norm": 0.45892960241879427, "learning_rate": 1.1569630927153125e-05, "loss": 2.6592, "step": 38717 }, { "epoch": 2.4035011484263453, "grad_norm": 0.5241381667121147, "learning_rate": 1.1567320660292642e-05, "loss": 2.6785, "step": 38718 }, { "epoch": 2.4035632255261032, "grad_norm": 0.46222719704236187, "learning_rate": 1.156501059394478e-05, "loss": 2.7107, "step": 38719 }, { "epoch": 2.403625302625861, "grad_norm": 0.49597271621749767, "learning_rate": 1.1562700728121567e-05, "loss": 2.6981, "step": 38720 }, { "epoch": 2.403687379725619, "grad_norm": 0.44507110477674233, "learning_rate": 1.1560391062835097e-05, "loss": 2.7637, "step": 38721 }, { "epoch": 2.403749456825377, "grad_norm": 0.5020797807190546, "learning_rate": 1.1558081598097393e-05, "loss": 2.6276, "step": 38722 }, { "epoch": 2.403811533925135, "grad_norm": 0.47901689393209884, "learning_rate": 1.1555772333920512e-05, "loss": 2.6143, "step": 38723 }, { "epoch": 2.403873611024893, "grad_norm": 0.4628536354447993, "learning_rate": 1.15534632703165e-05, "loss": 2.7572, "step": 38724 }, { "epoch": 2.4039356881246507, "grad_norm": 0.46850377609640553, "learning_rate": 1.1551154407297388e-05, "loss": 2.633, "step": 38725 }, { "epoch": 2.4039977652244087, "grad_norm": 0.4653926141455282, "learning_rate": 1.1548845744875253e-05, "loss": 2.7648, "step": 38726 }, { "epoch": 2.4040598423241666, "grad_norm": 0.4042918071337055, "learning_rate": 1.154653728306212e-05, "loss": 2.7124, "step": 38727 }, { "epoch": 2.4041219194239245, "grad_norm": 0.5057130992417892, "learning_rate": 1.154422902187004e-05, "loss": 2.6651, "step": 38728 }, { "epoch": 2.4041839965236824, "grad_norm": 0.5536391061952508, "learning_rate": 1.1541920961311036e-05, "loss": 2.7061, "step": 38729 }, { "epoch": 2.4042460736234403, "grad_norm": 0.4077665558873919, "learning_rate": 1.1539613101397183e-05, "loss": 2.7009, "step": 38730 }, { "epoch": 2.4043081507231983, "grad_norm": 0.49890360727873656, "learning_rate": 1.1537305442140506e-05, "loss": 2.6512, "step": 38731 }, { "epoch": 2.404370227822956, "grad_norm": 0.5133736038151869, "learning_rate": 1.1534997983553043e-05, "loss": 2.6705, "step": 38732 }, { "epoch": 2.404432304922714, "grad_norm": 0.5025416682887801, "learning_rate": 1.1532690725646817e-05, "loss": 2.748, "step": 38733 }, { "epoch": 2.404494382022472, "grad_norm": 0.4751180740469863, "learning_rate": 1.1530383668433898e-05, "loss": 2.726, "step": 38734 }, { "epoch": 2.40455645912223, "grad_norm": 0.4869955454501216, "learning_rate": 1.1528076811926303e-05, "loss": 2.7026, "step": 38735 }, { "epoch": 2.404618536221988, "grad_norm": 0.49284279811806636, "learning_rate": 1.1525770156136074e-05, "loss": 2.662, "step": 38736 }, { "epoch": 2.4046806133217458, "grad_norm": 0.4534623600239646, "learning_rate": 1.1523463701075238e-05, "loss": 2.7361, "step": 38737 }, { "epoch": 2.4047426904215037, "grad_norm": 0.48111956798116384, "learning_rate": 1.152115744675582e-05, "loss": 2.7761, "step": 38738 }, { "epoch": 2.4048047675212616, "grad_norm": 0.44056134677259257, "learning_rate": 1.1518851393189884e-05, "loss": 2.6875, "step": 38739 }, { "epoch": 2.4048668446210195, "grad_norm": 0.481496579920624, "learning_rate": 1.1516545540389435e-05, "loss": 2.6415, "step": 38740 }, { "epoch": 2.404928921720777, "grad_norm": 0.4945002216168901, "learning_rate": 1.1514239888366513e-05, "loss": 2.6747, "step": 38741 }, { "epoch": 2.4049909988205354, "grad_norm": 0.4392457260701177, "learning_rate": 1.1511934437133132e-05, "loss": 2.6229, "step": 38742 }, { "epoch": 2.405053075920293, "grad_norm": 0.4525478403345673, "learning_rate": 1.1509629186701342e-05, "loss": 2.6675, "step": 38743 }, { "epoch": 2.4051151530200507, "grad_norm": 0.4608646270561325, "learning_rate": 1.1507324137083165e-05, "loss": 2.6107, "step": 38744 }, { "epoch": 2.4051772301198087, "grad_norm": 0.4649570551391849, "learning_rate": 1.150501928829062e-05, "loss": 2.6948, "step": 38745 }, { "epoch": 2.4052393072195666, "grad_norm": 0.42697281299091044, "learning_rate": 1.1502714640335738e-05, "loss": 2.7136, "step": 38746 }, { "epoch": 2.4053013843193245, "grad_norm": 0.46720311511937135, "learning_rate": 1.1500410193230521e-05, "loss": 2.7067, "step": 38747 }, { "epoch": 2.4053634614190824, "grad_norm": 0.4442976044269368, "learning_rate": 1.1498105946987009e-05, "loss": 2.6412, "step": 38748 }, { "epoch": 2.4054255385188403, "grad_norm": 0.4524127854817645, "learning_rate": 1.1495801901617243e-05, "loss": 2.6849, "step": 38749 }, { "epoch": 2.4054876156185983, "grad_norm": 0.4513048908948287, "learning_rate": 1.1493498057133223e-05, "loss": 2.7043, "step": 38750 }, { "epoch": 2.405549692718356, "grad_norm": 0.42984879674136295, "learning_rate": 1.1491194413546968e-05, "loss": 2.6276, "step": 38751 }, { "epoch": 2.405611769818114, "grad_norm": 0.49425312314665154, "learning_rate": 1.1488890970870491e-05, "loss": 2.6431, "step": 38752 }, { "epoch": 2.405673846917872, "grad_norm": 0.4847855212657199, "learning_rate": 1.1486587729115833e-05, "loss": 2.6935, "step": 38753 }, { "epoch": 2.40573592401763, "grad_norm": 0.44655037206113823, "learning_rate": 1.1484284688294994e-05, "loss": 2.6309, "step": 38754 }, { "epoch": 2.405798001117388, "grad_norm": 0.504745494238594, "learning_rate": 1.1481981848419993e-05, "loss": 2.6557, "step": 38755 }, { "epoch": 2.4058600782171458, "grad_norm": 0.46560361608263634, "learning_rate": 1.147967920950283e-05, "loss": 2.6611, "step": 38756 }, { "epoch": 2.4059221553169037, "grad_norm": 0.42806354903099414, "learning_rate": 1.1477376771555548e-05, "loss": 2.6826, "step": 38757 }, { "epoch": 2.4059842324166616, "grad_norm": 0.43801328657848276, "learning_rate": 1.1475074534590135e-05, "loss": 2.6931, "step": 38758 }, { "epoch": 2.4060463095164195, "grad_norm": 0.4749149390654036, "learning_rate": 1.1472772498618617e-05, "loss": 2.7605, "step": 38759 }, { "epoch": 2.4061083866161774, "grad_norm": 0.5069795379365599, "learning_rate": 1.1470470663652999e-05, "loss": 2.6628, "step": 38760 }, { "epoch": 2.4061704637159353, "grad_norm": 0.4655826370118433, "learning_rate": 1.1468169029705273e-05, "loss": 2.6709, "step": 38761 }, { "epoch": 2.4062325408156933, "grad_norm": 0.4645822414533617, "learning_rate": 1.1465867596787473e-05, "loss": 2.6948, "step": 38762 }, { "epoch": 2.406294617915451, "grad_norm": 0.4956411414626718, "learning_rate": 1.1463566364911599e-05, "loss": 2.6749, "step": 38763 }, { "epoch": 2.4063566950152087, "grad_norm": 0.45881881974636457, "learning_rate": 1.1461265334089655e-05, "loss": 2.6147, "step": 38764 }, { "epoch": 2.406418772114967, "grad_norm": 0.47774950316985687, "learning_rate": 1.145896450433363e-05, "loss": 2.6476, "step": 38765 }, { "epoch": 2.4064808492147245, "grad_norm": 0.5112897807549097, "learning_rate": 1.145666387565556e-05, "loss": 2.743, "step": 38766 }, { "epoch": 2.4065429263144824, "grad_norm": 0.4415433383841728, "learning_rate": 1.145436344806743e-05, "loss": 2.6642, "step": 38767 }, { "epoch": 2.4066050034142403, "grad_norm": 0.46003764620440896, "learning_rate": 1.1452063221581244e-05, "loss": 2.5611, "step": 38768 }, { "epoch": 2.4066670805139982, "grad_norm": 0.46995359711665013, "learning_rate": 1.1449763196208984e-05, "loss": 2.7109, "step": 38769 }, { "epoch": 2.406729157613756, "grad_norm": 0.44015648397345347, "learning_rate": 1.1447463371962686e-05, "loss": 2.7194, "step": 38770 }, { "epoch": 2.406791234713514, "grad_norm": 0.44446960786110273, "learning_rate": 1.1445163748854326e-05, "loss": 2.6888, "step": 38771 }, { "epoch": 2.406853311813272, "grad_norm": 0.48412964359502064, "learning_rate": 1.1442864326895913e-05, "loss": 2.6825, "step": 38772 }, { "epoch": 2.40691538891303, "grad_norm": 0.4513670429013267, "learning_rate": 1.1440565106099433e-05, "loss": 2.6657, "step": 38773 }, { "epoch": 2.406977466012788, "grad_norm": 0.4070949021429483, "learning_rate": 1.1438266086476874e-05, "loss": 2.6584, "step": 38774 }, { "epoch": 2.4070395431125458, "grad_norm": 0.42241397619579035, "learning_rate": 1.1435967268040249e-05, "loss": 2.6011, "step": 38775 }, { "epoch": 2.4071016202123037, "grad_norm": 0.4615586346908018, "learning_rate": 1.1433668650801549e-05, "loss": 2.6784, "step": 38776 }, { "epoch": 2.4071636973120616, "grad_norm": 0.4269017341307346, "learning_rate": 1.1431370234772764e-05, "loss": 2.666, "step": 38777 }, { "epoch": 2.4072257744118195, "grad_norm": 0.4532019122389763, "learning_rate": 1.142907201996587e-05, "loss": 2.6275, "step": 38778 }, { "epoch": 2.4072878515115774, "grad_norm": 0.45245280227783996, "learning_rate": 1.1426774006392882e-05, "loss": 2.6977, "step": 38779 }, { "epoch": 2.4073499286113353, "grad_norm": 0.4773068789962458, "learning_rate": 1.1424476194065765e-05, "loss": 2.7157, "step": 38780 }, { "epoch": 2.4074120057110933, "grad_norm": 0.4792022139037994, "learning_rate": 1.1422178582996528e-05, "loss": 2.7331, "step": 38781 }, { "epoch": 2.407474082810851, "grad_norm": 0.42999518389000824, "learning_rate": 1.141988117319716e-05, "loss": 2.6455, "step": 38782 }, { "epoch": 2.407536159910609, "grad_norm": 0.477967349097452, "learning_rate": 1.1417583964679617e-05, "loss": 2.661, "step": 38783 }, { "epoch": 2.407598237010367, "grad_norm": 0.5160773922542177, "learning_rate": 1.1415286957455924e-05, "loss": 2.7955, "step": 38784 }, { "epoch": 2.407660314110125, "grad_norm": 0.4921765553302055, "learning_rate": 1.1412990151538038e-05, "loss": 2.707, "step": 38785 }, { "epoch": 2.407722391209883, "grad_norm": 0.4819857521780681, "learning_rate": 1.1410693546937956e-05, "loss": 2.665, "step": 38786 }, { "epoch": 2.4077844683096403, "grad_norm": 0.4502561118554626, "learning_rate": 1.1408397143667654e-05, "loss": 2.7793, "step": 38787 }, { "epoch": 2.4078465454093987, "grad_norm": 0.42640299417350735, "learning_rate": 1.1406100941739095e-05, "loss": 2.6984, "step": 38788 }, { "epoch": 2.407908622509156, "grad_norm": 0.48316293239345687, "learning_rate": 1.1403804941164291e-05, "loss": 2.7067, "step": 38789 }, { "epoch": 2.407970699608914, "grad_norm": 0.5388815479362997, "learning_rate": 1.1401509141955208e-05, "loss": 2.6577, "step": 38790 }, { "epoch": 2.408032776708672, "grad_norm": 0.4660365244678787, "learning_rate": 1.1399213544123821e-05, "loss": 2.7434, "step": 38791 }, { "epoch": 2.40809485380843, "grad_norm": 0.5159375081793024, "learning_rate": 1.1396918147682095e-05, "loss": 2.7718, "step": 38792 }, { "epoch": 2.408156930908188, "grad_norm": 0.46168188022112505, "learning_rate": 1.1394622952642031e-05, "loss": 2.6638, "step": 38793 }, { "epoch": 2.4082190080079457, "grad_norm": 0.4682368330563038, "learning_rate": 1.1392327959015592e-05, "loss": 2.6303, "step": 38794 }, { "epoch": 2.4082810851077037, "grad_norm": 0.4657265270837792, "learning_rate": 1.1390033166814751e-05, "loss": 2.7759, "step": 38795 }, { "epoch": 2.4083431622074616, "grad_norm": 0.44949083795640765, "learning_rate": 1.1387738576051476e-05, "loss": 2.6334, "step": 38796 }, { "epoch": 2.4084052393072195, "grad_norm": 0.4564609882370965, "learning_rate": 1.1385444186737732e-05, "loss": 2.7389, "step": 38797 }, { "epoch": 2.4084673164069774, "grad_norm": 0.4441859400839767, "learning_rate": 1.1383149998885511e-05, "loss": 2.7065, "step": 38798 }, { "epoch": 2.4085293935067353, "grad_norm": 0.4492913726485635, "learning_rate": 1.1380856012506775e-05, "loss": 2.6223, "step": 38799 }, { "epoch": 2.4085914706064933, "grad_norm": 0.49662834559153124, "learning_rate": 1.1378562227613481e-05, "loss": 2.6768, "step": 38800 }, { "epoch": 2.408653547706251, "grad_norm": 0.4671239784205053, "learning_rate": 1.1376268644217597e-05, "loss": 2.664, "step": 38801 }, { "epoch": 2.408715624806009, "grad_norm": 0.4164889512949302, "learning_rate": 1.1373975262331104e-05, "loss": 2.6384, "step": 38802 }, { "epoch": 2.408777701905767, "grad_norm": 0.4512783932140748, "learning_rate": 1.1371682081965963e-05, "loss": 2.6344, "step": 38803 }, { "epoch": 2.408839779005525, "grad_norm": 0.4413281430706845, "learning_rate": 1.1369389103134132e-05, "loss": 2.7225, "step": 38804 }, { "epoch": 2.408901856105283, "grad_norm": 0.43973174693343586, "learning_rate": 1.1367096325847565e-05, "loss": 2.6894, "step": 38805 }, { "epoch": 2.4089639332050408, "grad_norm": 0.45710206094568046, "learning_rate": 1.1364803750118241e-05, "loss": 2.6922, "step": 38806 }, { "epoch": 2.4090260103047987, "grad_norm": 0.4144216455410706, "learning_rate": 1.1362511375958123e-05, "loss": 2.6915, "step": 38807 }, { "epoch": 2.4090880874045566, "grad_norm": 0.4489405226512101, "learning_rate": 1.1360219203379157e-05, "loss": 2.7547, "step": 38808 }, { "epoch": 2.4091501645043145, "grad_norm": 0.4085432579396, "learning_rate": 1.1357927232393312e-05, "loss": 2.6376, "step": 38809 }, { "epoch": 2.4092122416040724, "grad_norm": 0.423386814279213, "learning_rate": 1.1355635463012527e-05, "loss": 2.6702, "step": 38810 }, { "epoch": 2.4092743187038304, "grad_norm": 0.45319321546421376, "learning_rate": 1.1353343895248786e-05, "loss": 2.7252, "step": 38811 }, { "epoch": 2.409336395803588, "grad_norm": 0.4319915725671897, "learning_rate": 1.135105252911403e-05, "loss": 2.7265, "step": 38812 }, { "epoch": 2.409398472903346, "grad_norm": 0.4122346985801263, "learning_rate": 1.1348761364620203e-05, "loss": 2.5916, "step": 38813 }, { "epoch": 2.4094605500031037, "grad_norm": 0.43020460051732873, "learning_rate": 1.1346470401779286e-05, "loss": 2.6058, "step": 38814 }, { "epoch": 2.4095226271028616, "grad_norm": 0.4420910477332188, "learning_rate": 1.1344179640603203e-05, "loss": 2.6973, "step": 38815 }, { "epoch": 2.4095847042026195, "grad_norm": 0.4680777815979152, "learning_rate": 1.1341889081103929e-05, "loss": 2.6335, "step": 38816 }, { "epoch": 2.4096467813023774, "grad_norm": 0.40942421027696274, "learning_rate": 1.1339598723293404e-05, "loss": 2.6293, "step": 38817 }, { "epoch": 2.4097088584021353, "grad_norm": 0.4253550154648903, "learning_rate": 1.1337308567183586e-05, "loss": 2.6713, "step": 38818 }, { "epoch": 2.4097709355018933, "grad_norm": 0.443357644098743, "learning_rate": 1.133501861278639e-05, "loss": 2.6322, "step": 38819 }, { "epoch": 2.409833012601651, "grad_norm": 0.46852901335926417, "learning_rate": 1.1332728860113807e-05, "loss": 2.7878, "step": 38820 }, { "epoch": 2.409895089701409, "grad_norm": 0.45053685073456495, "learning_rate": 1.1330439309177765e-05, "loss": 2.6612, "step": 38821 }, { "epoch": 2.409957166801167, "grad_norm": 0.4321603537286118, "learning_rate": 1.1328149959990208e-05, "loss": 2.6719, "step": 38822 }, { "epoch": 2.410019243900925, "grad_norm": 0.4283019528539734, "learning_rate": 1.1325860812563082e-05, "loss": 2.6315, "step": 38823 }, { "epoch": 2.410081321000683, "grad_norm": 0.486854727718438, "learning_rate": 1.1323571866908311e-05, "loss": 2.6832, "step": 38824 }, { "epoch": 2.4101433981004408, "grad_norm": 0.47717783240497297, "learning_rate": 1.1321283123037873e-05, "loss": 2.6962, "step": 38825 }, { "epoch": 2.4102054752001987, "grad_norm": 0.47678026320727945, "learning_rate": 1.1318994580963688e-05, "loss": 2.7237, "step": 38826 }, { "epoch": 2.4102675522999566, "grad_norm": 0.45162901075046147, "learning_rate": 1.1316706240697694e-05, "loss": 2.6455, "step": 38827 }, { "epoch": 2.4103296293997145, "grad_norm": 0.512732135721897, "learning_rate": 1.1314418102251827e-05, "loss": 2.7355, "step": 38828 }, { "epoch": 2.4103917064994724, "grad_norm": 0.4808154367287545, "learning_rate": 1.1312130165638047e-05, "loss": 2.6729, "step": 38829 }, { "epoch": 2.4104537835992303, "grad_norm": 0.44450329992723575, "learning_rate": 1.1309842430868267e-05, "loss": 2.7789, "step": 38830 }, { "epoch": 2.4105158606989883, "grad_norm": 0.4984063141732064, "learning_rate": 1.1307554897954436e-05, "loss": 2.6525, "step": 38831 }, { "epoch": 2.410577937798746, "grad_norm": 0.49230981245538613, "learning_rate": 1.1305267566908489e-05, "loss": 2.7477, "step": 38832 }, { "epoch": 2.410640014898504, "grad_norm": 0.49239251241847887, "learning_rate": 1.1302980437742333e-05, "loss": 2.6268, "step": 38833 }, { "epoch": 2.410702091998262, "grad_norm": 0.5086305631331931, "learning_rate": 1.130069351046794e-05, "loss": 2.7077, "step": 38834 }, { "epoch": 2.4107641690980195, "grad_norm": 0.49695083591250017, "learning_rate": 1.1298406785097227e-05, "loss": 2.7008, "step": 38835 }, { "epoch": 2.410826246197778, "grad_norm": 0.4689525325653073, "learning_rate": 1.1296120261642113e-05, "loss": 2.6364, "step": 38836 }, { "epoch": 2.4108883232975353, "grad_norm": 0.4743999285926653, "learning_rate": 1.1293833940114528e-05, "loss": 2.7156, "step": 38837 }, { "epoch": 2.4109504003972932, "grad_norm": 0.5195570630996286, "learning_rate": 1.1291547820526415e-05, "loss": 2.5897, "step": 38838 }, { "epoch": 2.411012477497051, "grad_norm": 0.47959365858744646, "learning_rate": 1.1289261902889704e-05, "loss": 2.6097, "step": 38839 }, { "epoch": 2.411074554596809, "grad_norm": 0.4534410163958626, "learning_rate": 1.1286976187216302e-05, "loss": 2.664, "step": 38840 }, { "epoch": 2.411136631696567, "grad_norm": 0.48044800132052284, "learning_rate": 1.1284690673518133e-05, "loss": 2.6007, "step": 38841 }, { "epoch": 2.411198708796325, "grad_norm": 0.5551859268429005, "learning_rate": 1.1282405361807141e-05, "loss": 2.6066, "step": 38842 }, { "epoch": 2.411260785896083, "grad_norm": 0.4810244750162423, "learning_rate": 1.1280120252095244e-05, "loss": 2.6641, "step": 38843 }, { "epoch": 2.4113228629958408, "grad_norm": 0.45790082580731867, "learning_rate": 1.1277835344394355e-05, "loss": 2.7392, "step": 38844 }, { "epoch": 2.4113849400955987, "grad_norm": 0.4578178190631887, "learning_rate": 1.1275550638716403e-05, "loss": 2.69, "step": 38845 }, { "epoch": 2.4114470171953566, "grad_norm": 0.540687323709895, "learning_rate": 1.1273266135073291e-05, "loss": 2.6354, "step": 38846 }, { "epoch": 2.4115090942951145, "grad_norm": 0.4748293990652508, "learning_rate": 1.1270981833476946e-05, "loss": 2.7438, "step": 38847 }, { "epoch": 2.4115711713948724, "grad_norm": 0.4414053891603971, "learning_rate": 1.126869773393931e-05, "loss": 2.587, "step": 38848 }, { "epoch": 2.4116332484946303, "grad_norm": 0.5112881778182167, "learning_rate": 1.1266413836472278e-05, "loss": 2.6846, "step": 38849 }, { "epoch": 2.4116953255943883, "grad_norm": 0.468436120166744, "learning_rate": 1.1264130141087764e-05, "loss": 2.5885, "step": 38850 }, { "epoch": 2.411757402694146, "grad_norm": 0.41637376583423785, "learning_rate": 1.1261846647797674e-05, "loss": 2.6081, "step": 38851 }, { "epoch": 2.411819479793904, "grad_norm": 0.44789204603735294, "learning_rate": 1.1259563356613951e-05, "loss": 2.6137, "step": 38852 }, { "epoch": 2.411881556893662, "grad_norm": 0.42301541296855294, "learning_rate": 1.1257280267548492e-05, "loss": 2.7088, "step": 38853 }, { "epoch": 2.41194363399342, "grad_norm": 0.45849630500049215, "learning_rate": 1.1254997380613203e-05, "loss": 2.6055, "step": 38854 }, { "epoch": 2.412005711093178, "grad_norm": 0.41404859051643583, "learning_rate": 1.1252714695819988e-05, "loss": 2.6784, "step": 38855 }, { "epoch": 2.4120677881929358, "grad_norm": 0.4762556776824691, "learning_rate": 1.1250432213180779e-05, "loss": 2.7585, "step": 38856 }, { "epoch": 2.4121298652926937, "grad_norm": 0.4521095473370076, "learning_rate": 1.124814993270747e-05, "loss": 2.6676, "step": 38857 }, { "epoch": 2.4121919423924516, "grad_norm": 0.4929094015872138, "learning_rate": 1.1245867854411968e-05, "loss": 2.7095, "step": 38858 }, { "epoch": 2.4122540194922095, "grad_norm": 0.4469673628157116, "learning_rate": 1.1243585978306187e-05, "loss": 2.5786, "step": 38859 }, { "epoch": 2.412316096591967, "grad_norm": 0.47866046815345864, "learning_rate": 1.1241304304402011e-05, "loss": 2.6721, "step": 38860 }, { "epoch": 2.4123781736917254, "grad_norm": 0.4295559768525586, "learning_rate": 1.123902283271137e-05, "loss": 2.701, "step": 38861 }, { "epoch": 2.412440250791483, "grad_norm": 0.45519744553046165, "learning_rate": 1.1236741563246162e-05, "loss": 2.6648, "step": 38862 }, { "epoch": 2.4125023278912407, "grad_norm": 0.45290370996618784, "learning_rate": 1.1234460496018279e-05, "loss": 2.637, "step": 38863 }, { "epoch": 2.4125644049909987, "grad_norm": 0.4232630676555029, "learning_rate": 1.1232179631039613e-05, "loss": 2.6948, "step": 38864 }, { "epoch": 2.4126264820907566, "grad_norm": 0.45668526946009, "learning_rate": 1.1229898968322094e-05, "loss": 2.656, "step": 38865 }, { "epoch": 2.4126885591905145, "grad_norm": 0.4459786034335451, "learning_rate": 1.1227618507877603e-05, "loss": 2.642, "step": 38866 }, { "epoch": 2.4127506362902724, "grad_norm": 0.4532959714153833, "learning_rate": 1.1225338249718032e-05, "loss": 2.6885, "step": 38867 }, { "epoch": 2.4128127133900303, "grad_norm": 0.41828202602013814, "learning_rate": 1.1223058193855278e-05, "loss": 2.7208, "step": 38868 }, { "epoch": 2.4128747904897883, "grad_norm": 0.47553306144654617, "learning_rate": 1.1220778340301252e-05, "loss": 2.5369, "step": 38869 }, { "epoch": 2.412936867589546, "grad_norm": 0.4190177054647486, "learning_rate": 1.121849868906784e-05, "loss": 2.6407, "step": 38870 }, { "epoch": 2.412998944689304, "grad_norm": 0.44369677170672434, "learning_rate": 1.1216219240166942e-05, "loss": 2.6806, "step": 38871 }, { "epoch": 2.413061021789062, "grad_norm": 0.4483841648350934, "learning_rate": 1.1213939993610435e-05, "loss": 2.6945, "step": 38872 }, { "epoch": 2.41312309888882, "grad_norm": 0.4388696314799255, "learning_rate": 1.1211660949410208e-05, "loss": 2.6298, "step": 38873 }, { "epoch": 2.413185175988578, "grad_norm": 0.4926666924325414, "learning_rate": 1.1209382107578176e-05, "loss": 2.676, "step": 38874 }, { "epoch": 2.4132472530883358, "grad_norm": 0.4362173692911087, "learning_rate": 1.1207103468126212e-05, "loss": 2.6715, "step": 38875 }, { "epoch": 2.4133093301880937, "grad_norm": 0.4695541319135106, "learning_rate": 1.1204825031066207e-05, "loss": 2.7703, "step": 38876 }, { "epoch": 2.4133714072878516, "grad_norm": 0.4667053292499657, "learning_rate": 1.1202546796410036e-05, "loss": 2.6494, "step": 38877 }, { "epoch": 2.4134334843876095, "grad_norm": 0.4865264641335708, "learning_rate": 1.1200268764169614e-05, "loss": 2.6455, "step": 38878 }, { "epoch": 2.4134955614873674, "grad_norm": 0.43683740986218844, "learning_rate": 1.1197990934356789e-05, "loss": 2.593, "step": 38879 }, { "epoch": 2.4135576385871254, "grad_norm": 0.45645073759734145, "learning_rate": 1.1195713306983479e-05, "loss": 2.6914, "step": 38880 }, { "epoch": 2.4136197156868833, "grad_norm": 0.43530720159150066, "learning_rate": 1.1193435882061554e-05, "loss": 2.6783, "step": 38881 }, { "epoch": 2.413681792786641, "grad_norm": 0.4845667428773511, "learning_rate": 1.1191158659602897e-05, "loss": 2.6025, "step": 38882 }, { "epoch": 2.4137438698863987, "grad_norm": 0.4992566095840149, "learning_rate": 1.118888163961937e-05, "loss": 2.6783, "step": 38883 }, { "epoch": 2.413805946986157, "grad_norm": 0.487938188199828, "learning_rate": 1.1186604822122882e-05, "loss": 2.714, "step": 38884 }, { "epoch": 2.4138680240859145, "grad_norm": 0.47387200047824873, "learning_rate": 1.11843282071253e-05, "loss": 2.7286, "step": 38885 }, { "epoch": 2.4139301011856724, "grad_norm": 0.4553754269244245, "learning_rate": 1.1182051794638498e-05, "loss": 2.7193, "step": 38886 }, { "epoch": 2.4139921782854303, "grad_norm": 0.4883587890094225, "learning_rate": 1.1179775584674345e-05, "loss": 2.6981, "step": 38887 }, { "epoch": 2.4140542553851883, "grad_norm": 0.4756269182725438, "learning_rate": 1.1177499577244744e-05, "loss": 2.6042, "step": 38888 }, { "epoch": 2.414116332484946, "grad_norm": 0.4422674133825271, "learning_rate": 1.1175223772361549e-05, "loss": 2.7255, "step": 38889 }, { "epoch": 2.414178409584704, "grad_norm": 0.4459654388016122, "learning_rate": 1.1172948170036635e-05, "loss": 2.7329, "step": 38890 }, { "epoch": 2.414240486684462, "grad_norm": 0.4349577987139692, "learning_rate": 1.1170672770281858e-05, "loss": 2.7001, "step": 38891 }, { "epoch": 2.41430256378422, "grad_norm": 0.42896878215717144, "learning_rate": 1.116839757310913e-05, "loss": 2.6823, "step": 38892 }, { "epoch": 2.414364640883978, "grad_norm": 0.42737210404585574, "learning_rate": 1.1166122578530291e-05, "loss": 2.5827, "step": 38893 }, { "epoch": 2.4144267179837358, "grad_norm": 0.4591613103839172, "learning_rate": 1.1163847786557224e-05, "loss": 2.7844, "step": 38894 }, { "epoch": 2.4144887950834937, "grad_norm": 0.4365314593847476, "learning_rate": 1.1161573197201785e-05, "loss": 2.6466, "step": 38895 }, { "epoch": 2.4145508721832516, "grad_norm": 0.4563785848620422, "learning_rate": 1.1159298810475838e-05, "loss": 2.7273, "step": 38896 }, { "epoch": 2.4146129492830095, "grad_norm": 0.4539479641943193, "learning_rate": 1.1157024626391272e-05, "loss": 2.6471, "step": 38897 }, { "epoch": 2.4146750263827674, "grad_norm": 0.4350004831673142, "learning_rate": 1.1154750644959933e-05, "loss": 2.6574, "step": 38898 }, { "epoch": 2.4147371034825253, "grad_norm": 0.4398812306833582, "learning_rate": 1.1152476866193695e-05, "loss": 2.6057, "step": 38899 }, { "epoch": 2.4147991805822833, "grad_norm": 0.4730737434398742, "learning_rate": 1.1150203290104399e-05, "loss": 2.6711, "step": 38900 }, { "epoch": 2.414861257682041, "grad_norm": 0.42945059400995506, "learning_rate": 1.1147929916703936e-05, "loss": 2.6299, "step": 38901 }, { "epoch": 2.414923334781799, "grad_norm": 0.46326441098982135, "learning_rate": 1.1145656746004157e-05, "loss": 2.6975, "step": 38902 }, { "epoch": 2.414985411881557, "grad_norm": 0.440111303177556, "learning_rate": 1.1143383778016919e-05, "loss": 2.7121, "step": 38903 }, { "epoch": 2.415047488981315, "grad_norm": 0.4403145379478883, "learning_rate": 1.1141111012754069e-05, "loss": 2.6347, "step": 38904 }, { "epoch": 2.415109566081073, "grad_norm": 0.4476940984520603, "learning_rate": 1.1138838450227485e-05, "loss": 2.739, "step": 38905 }, { "epoch": 2.4151716431808308, "grad_norm": 0.44381189989146896, "learning_rate": 1.1136566090449019e-05, "loss": 2.7739, "step": 38906 }, { "epoch": 2.4152337202805887, "grad_norm": 0.44874448204367073, "learning_rate": 1.1134293933430518e-05, "loss": 2.6253, "step": 38907 }, { "epoch": 2.415295797380346, "grad_norm": 0.44761112977701056, "learning_rate": 1.1132021979183843e-05, "loss": 2.7209, "step": 38908 }, { "epoch": 2.4153578744801045, "grad_norm": 0.4093754921247713, "learning_rate": 1.1129750227720832e-05, "loss": 2.5917, "step": 38909 }, { "epoch": 2.415419951579862, "grad_norm": 0.49520382696831444, "learning_rate": 1.112747867905336e-05, "loss": 2.6656, "step": 38910 }, { "epoch": 2.41548202867962, "grad_norm": 0.44627483854435684, "learning_rate": 1.112520733319326e-05, "loss": 2.682, "step": 38911 }, { "epoch": 2.415544105779378, "grad_norm": 0.45696375421823116, "learning_rate": 1.11229361901524e-05, "loss": 2.7092, "step": 38912 }, { "epoch": 2.4156061828791358, "grad_norm": 0.42220101298879753, "learning_rate": 1.1120665249942624e-05, "loss": 2.7512, "step": 38913 }, { "epoch": 2.4156682599788937, "grad_norm": 0.45617452023237853, "learning_rate": 1.1118394512575758e-05, "loss": 2.6393, "step": 38914 }, { "epoch": 2.4157303370786516, "grad_norm": 0.46873280175766263, "learning_rate": 1.111612397806368e-05, "loss": 2.6337, "step": 38915 }, { "epoch": 2.4157924141784095, "grad_norm": 0.4462833362437359, "learning_rate": 1.1113853646418227e-05, "loss": 2.6483, "step": 38916 }, { "epoch": 2.4158544912781674, "grad_norm": 0.46881408253372, "learning_rate": 1.1111583517651237e-05, "loss": 2.6089, "step": 38917 }, { "epoch": 2.4159165683779253, "grad_norm": 0.413482926166128, "learning_rate": 1.110931359177454e-05, "loss": 2.7068, "step": 38918 }, { "epoch": 2.4159786454776833, "grad_norm": 0.4212707501403134, "learning_rate": 1.110704386880001e-05, "loss": 2.6626, "step": 38919 }, { "epoch": 2.416040722577441, "grad_norm": 0.4224811543826838, "learning_rate": 1.1104774348739478e-05, "loss": 2.7242, "step": 38920 }, { "epoch": 2.416102799677199, "grad_norm": 0.48420500079388684, "learning_rate": 1.1102505031604776e-05, "loss": 2.6459, "step": 38921 }, { "epoch": 2.416164876776957, "grad_norm": 0.4239703172806722, "learning_rate": 1.1100235917407748e-05, "loss": 2.706, "step": 38922 }, { "epoch": 2.416226953876715, "grad_norm": 0.45071220927044076, "learning_rate": 1.109796700616022e-05, "loss": 2.6722, "step": 38923 }, { "epoch": 2.416289030976473, "grad_norm": 0.4823822746548785, "learning_rate": 1.1095698297874053e-05, "loss": 2.6886, "step": 38924 }, { "epoch": 2.4163511080762308, "grad_norm": 0.44621875634865643, "learning_rate": 1.1093429792561077e-05, "loss": 2.7069, "step": 38925 }, { "epoch": 2.4164131851759887, "grad_norm": 0.4311018640985805, "learning_rate": 1.109116149023312e-05, "loss": 2.6341, "step": 38926 }, { "epoch": 2.4164752622757466, "grad_norm": 0.5248884048385436, "learning_rate": 1.1088893390902006e-05, "loss": 2.6328, "step": 38927 }, { "epoch": 2.4165373393755045, "grad_norm": 0.5042567254120573, "learning_rate": 1.1086625494579595e-05, "loss": 2.6841, "step": 38928 }, { "epoch": 2.4165994164752624, "grad_norm": 0.4590682419793698, "learning_rate": 1.10843578012777e-05, "loss": 2.72, "step": 38929 }, { "epoch": 2.4166614935750204, "grad_norm": 0.4876703942669875, "learning_rate": 1.1082090311008165e-05, "loss": 2.6555, "step": 38930 }, { "epoch": 2.416723570674778, "grad_norm": 0.4455250182753128, "learning_rate": 1.107982302378281e-05, "loss": 2.6271, "step": 38931 }, { "epoch": 2.416785647774536, "grad_norm": 0.48628536480663653, "learning_rate": 1.1077555939613454e-05, "loss": 2.6073, "step": 38932 }, { "epoch": 2.4168477248742937, "grad_norm": 0.48425137572715915, "learning_rate": 1.107528905851195e-05, "loss": 2.6953, "step": 38933 }, { "epoch": 2.4169098019740516, "grad_norm": 0.4951936561175812, "learning_rate": 1.1073022380490111e-05, "loss": 2.7182, "step": 38934 }, { "epoch": 2.4169718790738095, "grad_norm": 0.4644437078318517, "learning_rate": 1.1070755905559766e-05, "loss": 2.6818, "step": 38935 }, { "epoch": 2.4170339561735674, "grad_norm": 0.49858659409779676, "learning_rate": 1.106848963373272e-05, "loss": 2.7252, "step": 38936 }, { "epoch": 2.4170960332733253, "grad_norm": 0.4354614550387034, "learning_rate": 1.1066223565020833e-05, "loss": 2.6677, "step": 38937 }, { "epoch": 2.4171581103730833, "grad_norm": 0.45803481553110886, "learning_rate": 1.1063957699435906e-05, "loss": 2.5804, "step": 38938 }, { "epoch": 2.417220187472841, "grad_norm": 0.45342645811784643, "learning_rate": 1.1061692036989769e-05, "loss": 2.689, "step": 38939 }, { "epoch": 2.417282264572599, "grad_norm": 0.4636023317422806, "learning_rate": 1.1059426577694216e-05, "loss": 2.7554, "step": 38940 }, { "epoch": 2.417344341672357, "grad_norm": 0.4571471271171783, "learning_rate": 1.1057161321561105e-05, "loss": 2.7137, "step": 38941 }, { "epoch": 2.417406418772115, "grad_norm": 0.4581683595286558, "learning_rate": 1.1054896268602239e-05, "loss": 2.6992, "step": 38942 }, { "epoch": 2.417468495871873, "grad_norm": 0.4305905649178259, "learning_rate": 1.1052631418829435e-05, "loss": 2.6808, "step": 38943 }, { "epoch": 2.4175305729716308, "grad_norm": 0.4527911896887262, "learning_rate": 1.1050366772254489e-05, "loss": 2.6436, "step": 38944 }, { "epoch": 2.4175926500713887, "grad_norm": 0.4355206160904821, "learning_rate": 1.1048102328889248e-05, "loss": 2.6055, "step": 38945 }, { "epoch": 2.4176547271711466, "grad_norm": 0.44295498390942056, "learning_rate": 1.1045838088745503e-05, "loss": 2.6583, "step": 38946 }, { "epoch": 2.4177168042709045, "grad_norm": 0.45138123844053907, "learning_rate": 1.1043574051835087e-05, "loss": 2.6607, "step": 38947 }, { "epoch": 2.4177788813706624, "grad_norm": 0.4546567578754372, "learning_rate": 1.104131021816981e-05, "loss": 2.7658, "step": 38948 }, { "epoch": 2.4178409584704204, "grad_norm": 0.43935040439011014, "learning_rate": 1.1039046587761464e-05, "loss": 2.6755, "step": 38949 }, { "epoch": 2.4179030355701783, "grad_norm": 0.4464027543777258, "learning_rate": 1.1036783160621866e-05, "loss": 2.6928, "step": 38950 }, { "epoch": 2.417965112669936, "grad_norm": 0.4805993778706134, "learning_rate": 1.1034519936762838e-05, "loss": 2.6762, "step": 38951 }, { "epoch": 2.418027189769694, "grad_norm": 0.4316270403213069, "learning_rate": 1.1032256916196182e-05, "loss": 2.6463, "step": 38952 }, { "epoch": 2.418089266869452, "grad_norm": 0.4897368288542498, "learning_rate": 1.1029994098933699e-05, "loss": 2.6832, "step": 38953 }, { "epoch": 2.41815134396921, "grad_norm": 0.452852660583939, "learning_rate": 1.102773148498718e-05, "loss": 2.5581, "step": 38954 }, { "epoch": 2.418213421068968, "grad_norm": 0.4424464030592611, "learning_rate": 1.1025469074368466e-05, "loss": 2.6569, "step": 38955 }, { "epoch": 2.4182754981687253, "grad_norm": 0.4536357500406078, "learning_rate": 1.1023206867089337e-05, "loss": 2.6318, "step": 38956 }, { "epoch": 2.4183375752684837, "grad_norm": 0.43745691223499894, "learning_rate": 1.1020944863161603e-05, "loss": 2.7331, "step": 38957 }, { "epoch": 2.418399652368241, "grad_norm": 0.4696755966485278, "learning_rate": 1.1018683062597063e-05, "loss": 2.6845, "step": 38958 }, { "epoch": 2.418461729467999, "grad_norm": 0.4551813229622692, "learning_rate": 1.10164214654075e-05, "loss": 2.6318, "step": 38959 }, { "epoch": 2.418523806567757, "grad_norm": 0.46396756341001005, "learning_rate": 1.101416007160474e-05, "loss": 2.6734, "step": 38960 }, { "epoch": 2.418585883667515, "grad_norm": 0.4734294651470173, "learning_rate": 1.1011898881200577e-05, "loss": 2.688, "step": 38961 }, { "epoch": 2.418647960767273, "grad_norm": 0.41412401642065116, "learning_rate": 1.1009637894206804e-05, "loss": 2.5821, "step": 38962 }, { "epoch": 2.4187100378670308, "grad_norm": 0.5109270100581232, "learning_rate": 1.1007377110635193e-05, "loss": 2.6926, "step": 38963 }, { "epoch": 2.4187721149667887, "grad_norm": 0.48325797458153036, "learning_rate": 1.1005116530497583e-05, "loss": 2.6522, "step": 38964 }, { "epoch": 2.4188341920665466, "grad_norm": 0.4212891431444964, "learning_rate": 1.100285615380574e-05, "loss": 2.6429, "step": 38965 }, { "epoch": 2.4188962691663045, "grad_norm": 0.464984840330535, "learning_rate": 1.1000595980571472e-05, "loss": 2.6636, "step": 38966 }, { "epoch": 2.4189583462660624, "grad_norm": 0.4424175055165307, "learning_rate": 1.0998336010806554e-05, "loss": 2.5102, "step": 38967 }, { "epoch": 2.4190204233658203, "grad_norm": 0.48438564411896656, "learning_rate": 1.0996076244522775e-05, "loss": 2.6718, "step": 38968 }, { "epoch": 2.4190825004655783, "grad_norm": 0.453730129799957, "learning_rate": 1.099381668173195e-05, "loss": 2.7688, "step": 38969 }, { "epoch": 2.419144577565336, "grad_norm": 0.46940670107014837, "learning_rate": 1.099155732244585e-05, "loss": 2.5723, "step": 38970 }, { "epoch": 2.419206654665094, "grad_norm": 0.4767599478505411, "learning_rate": 1.0989298166676265e-05, "loss": 2.651, "step": 38971 }, { "epoch": 2.419268731764852, "grad_norm": 0.49793366286134116, "learning_rate": 1.098703921443497e-05, "loss": 2.713, "step": 38972 }, { "epoch": 2.41933080886461, "grad_norm": 0.5255553260374386, "learning_rate": 1.0984780465733769e-05, "loss": 2.6728, "step": 38973 }, { "epoch": 2.419392885964368, "grad_norm": 0.4764970515601251, "learning_rate": 1.098252192058445e-05, "loss": 2.6489, "step": 38974 }, { "epoch": 2.4194549630641258, "grad_norm": 0.4979826792314998, "learning_rate": 1.098026357899878e-05, "loss": 2.6579, "step": 38975 }, { "epoch": 2.4195170401638837, "grad_norm": 0.46505054714436056, "learning_rate": 1.0978005440988532e-05, "loss": 2.593, "step": 38976 }, { "epoch": 2.4195791172636416, "grad_norm": 0.5113850735977845, "learning_rate": 1.0975747506565504e-05, "loss": 2.624, "step": 38977 }, { "epoch": 2.4196411943633995, "grad_norm": 0.4876850608325676, "learning_rate": 1.0973489775741492e-05, "loss": 2.7111, "step": 38978 }, { "epoch": 2.419703271463157, "grad_norm": 0.4809163807376921, "learning_rate": 1.0971232248528252e-05, "loss": 2.6898, "step": 38979 }, { "epoch": 2.4197653485629154, "grad_norm": 0.44080855827564985, "learning_rate": 1.096897492493757e-05, "loss": 2.6544, "step": 38980 }, { "epoch": 2.419827425662673, "grad_norm": 0.48045643736407007, "learning_rate": 1.0966717804981225e-05, "loss": 2.7707, "step": 38981 }, { "epoch": 2.4198895027624308, "grad_norm": 0.5175605339083891, "learning_rate": 1.0964460888670968e-05, "loss": 2.7478, "step": 38982 }, { "epoch": 2.4199515798621887, "grad_norm": 0.4769538616952666, "learning_rate": 1.0962204176018614e-05, "loss": 2.6657, "step": 38983 }, { "epoch": 2.4200136569619466, "grad_norm": 0.45947685604628646, "learning_rate": 1.0959947667035919e-05, "loss": 2.7774, "step": 38984 }, { "epoch": 2.4200757340617045, "grad_norm": 0.43021959824241884, "learning_rate": 1.0957691361734652e-05, "loss": 2.6658, "step": 38985 }, { "epoch": 2.4201378111614624, "grad_norm": 0.4961420321060298, "learning_rate": 1.0955435260126574e-05, "loss": 2.7484, "step": 38986 }, { "epoch": 2.4201998882612203, "grad_norm": 0.4471378893627417, "learning_rate": 1.0953179362223482e-05, "loss": 2.6305, "step": 38987 }, { "epoch": 2.4202619653609783, "grad_norm": 0.4639371959660102, "learning_rate": 1.0950923668037127e-05, "loss": 2.7677, "step": 38988 }, { "epoch": 2.420324042460736, "grad_norm": 0.45848577485746567, "learning_rate": 1.0948668177579285e-05, "loss": 2.6669, "step": 38989 }, { "epoch": 2.420386119560494, "grad_norm": 0.4932383566312085, "learning_rate": 1.0946412890861712e-05, "loss": 2.6752, "step": 38990 }, { "epoch": 2.420448196660252, "grad_norm": 0.467969976560485, "learning_rate": 1.0944157807896193e-05, "loss": 2.7367, "step": 38991 }, { "epoch": 2.42051027376001, "grad_norm": 0.45850655865736273, "learning_rate": 1.0941902928694486e-05, "loss": 2.5766, "step": 38992 }, { "epoch": 2.420572350859768, "grad_norm": 0.4875933569992311, "learning_rate": 1.0939648253268353e-05, "loss": 2.663, "step": 38993 }, { "epoch": 2.4206344279595258, "grad_norm": 0.4435542850332983, "learning_rate": 1.0937393781629556e-05, "loss": 2.7522, "step": 38994 }, { "epoch": 2.4206965050592837, "grad_norm": 0.43589267773740914, "learning_rate": 1.093513951378985e-05, "loss": 2.6858, "step": 38995 }, { "epoch": 2.4207585821590416, "grad_norm": 0.4488402389073516, "learning_rate": 1.093288544976101e-05, "loss": 2.6154, "step": 38996 }, { "epoch": 2.4208206592587995, "grad_norm": 0.5259202181174798, "learning_rate": 1.0930631589554796e-05, "loss": 2.7017, "step": 38997 }, { "epoch": 2.4208827363585574, "grad_norm": 0.44616996461475666, "learning_rate": 1.092837793318296e-05, "loss": 2.6299, "step": 38998 }, { "epoch": 2.4209448134583154, "grad_norm": 0.4067361380530926, "learning_rate": 1.0926124480657246e-05, "loss": 2.5777, "step": 38999 }, { "epoch": 2.4210068905580733, "grad_norm": 0.46917211282402466, "learning_rate": 1.092387123198944e-05, "loss": 2.6643, "step": 39000 }, { "epoch": 2.421068967657831, "grad_norm": 0.4550838551958677, "learning_rate": 1.0921618187191291e-05, "loss": 2.6844, "step": 39001 }, { "epoch": 2.421131044757589, "grad_norm": 0.41977348367138756, "learning_rate": 1.091936534627454e-05, "loss": 2.6683, "step": 39002 }, { "epoch": 2.421193121857347, "grad_norm": 0.45702347382018904, "learning_rate": 1.0917112709250948e-05, "loss": 2.705, "step": 39003 }, { "epoch": 2.4212551989571045, "grad_norm": 0.4403073126038592, "learning_rate": 1.091486027613225e-05, "loss": 2.6616, "step": 39004 }, { "epoch": 2.421317276056863, "grad_norm": 0.4206728408100178, "learning_rate": 1.0912608046930235e-05, "loss": 2.7203, "step": 39005 }, { "epoch": 2.4213793531566203, "grad_norm": 0.4386623133553986, "learning_rate": 1.0910356021656626e-05, "loss": 2.7056, "step": 39006 }, { "epoch": 2.4214414302563783, "grad_norm": 0.48353448359807877, "learning_rate": 1.0908104200323183e-05, "loss": 2.7589, "step": 39007 }, { "epoch": 2.421503507356136, "grad_norm": 0.4004204343538466, "learning_rate": 1.0905852582941634e-05, "loss": 2.6514, "step": 39008 }, { "epoch": 2.421565584455894, "grad_norm": 0.42633119117510815, "learning_rate": 1.0903601169523752e-05, "loss": 2.5942, "step": 39009 }, { "epoch": 2.421627661555652, "grad_norm": 0.4652708159427226, "learning_rate": 1.090134996008127e-05, "loss": 2.6859, "step": 39010 }, { "epoch": 2.42168973865541, "grad_norm": 0.46485886429008566, "learning_rate": 1.0899098954625942e-05, "loss": 2.6169, "step": 39011 }, { "epoch": 2.421751815755168, "grad_norm": 0.45943981868141676, "learning_rate": 1.0896848153169508e-05, "loss": 2.6514, "step": 39012 }, { "epoch": 2.4218138928549258, "grad_norm": 0.4381112698948999, "learning_rate": 1.0894597555723701e-05, "loss": 2.6418, "step": 39013 }, { "epoch": 2.4218759699546837, "grad_norm": 0.4169830035375321, "learning_rate": 1.0892347162300282e-05, "loss": 2.6929, "step": 39014 }, { "epoch": 2.4219380470544416, "grad_norm": 0.46457930664347413, "learning_rate": 1.0890096972910985e-05, "loss": 2.7032, "step": 39015 }, { "epoch": 2.4220001241541995, "grad_norm": 0.5155117026035784, "learning_rate": 1.088784698756754e-05, "loss": 2.7113, "step": 39016 }, { "epoch": 2.4220622012539574, "grad_norm": 0.4768072344126327, "learning_rate": 1.0885597206281701e-05, "loss": 2.6887, "step": 39017 }, { "epoch": 2.4221242783537154, "grad_norm": 0.4650828604038178, "learning_rate": 1.0883347629065178e-05, "loss": 2.7012, "step": 39018 }, { "epoch": 2.4221863554534733, "grad_norm": 0.5113301731007402, "learning_rate": 1.0881098255929739e-05, "loss": 2.7621, "step": 39019 }, { "epoch": 2.422248432553231, "grad_norm": 0.5066852676381951, "learning_rate": 1.0878849086887106e-05, "loss": 2.6504, "step": 39020 }, { "epoch": 2.422310509652989, "grad_norm": 0.47085502357934356, "learning_rate": 1.0876600121949016e-05, "loss": 2.6917, "step": 39021 }, { "epoch": 2.422372586752747, "grad_norm": 0.4537961527574317, "learning_rate": 1.087435136112719e-05, "loss": 2.7409, "step": 39022 }, { "epoch": 2.422434663852505, "grad_norm": 0.47056004419191355, "learning_rate": 1.0872102804433376e-05, "loss": 2.6352, "step": 39023 }, { "epoch": 2.422496740952263, "grad_norm": 0.49220445391807166, "learning_rate": 1.0869854451879307e-05, "loss": 2.7152, "step": 39024 }, { "epoch": 2.422558818052021, "grad_norm": 0.4373418147353312, "learning_rate": 1.0867606303476708e-05, "loss": 2.6903, "step": 39025 }, { "epoch": 2.4226208951517787, "grad_norm": 0.4481129678204207, "learning_rate": 1.0865358359237287e-05, "loss": 2.6041, "step": 39026 }, { "epoch": 2.422682972251536, "grad_norm": 0.5286575059275448, "learning_rate": 1.0863110619172812e-05, "loss": 2.6828, "step": 39027 }, { "epoch": 2.4227450493512945, "grad_norm": 0.4504388907410954, "learning_rate": 1.0860863083294986e-05, "loss": 2.7002, "step": 39028 }, { "epoch": 2.422807126451052, "grad_norm": 0.4457574261310444, "learning_rate": 1.0858615751615542e-05, "loss": 2.6559, "step": 39029 }, { "epoch": 2.42286920355081, "grad_norm": 0.48594293315340087, "learning_rate": 1.0856368624146196e-05, "loss": 2.6925, "step": 39030 }, { "epoch": 2.422931280650568, "grad_norm": 0.493658788320129, "learning_rate": 1.0854121700898667e-05, "loss": 2.6663, "step": 39031 }, { "epoch": 2.4229933577503258, "grad_norm": 0.41941810375027805, "learning_rate": 1.0851874981884701e-05, "loss": 2.5816, "step": 39032 }, { "epoch": 2.4230554348500837, "grad_norm": 0.4442698335207757, "learning_rate": 1.0849628467116007e-05, "loss": 2.6362, "step": 39033 }, { "epoch": 2.4231175119498416, "grad_norm": 0.5003963627959023, "learning_rate": 1.0847382156604308e-05, "loss": 2.6613, "step": 39034 }, { "epoch": 2.4231795890495995, "grad_norm": 0.5242568762568857, "learning_rate": 1.0845136050361303e-05, "loss": 2.6501, "step": 39035 }, { "epoch": 2.4232416661493574, "grad_norm": 0.5227476955182413, "learning_rate": 1.0842890148398743e-05, "loss": 2.8176, "step": 39036 }, { "epoch": 2.4233037432491153, "grad_norm": 0.46401438736602013, "learning_rate": 1.0840644450728332e-05, "loss": 2.7049, "step": 39037 }, { "epoch": 2.4233658203488733, "grad_norm": 0.4909046348094989, "learning_rate": 1.0838398957361784e-05, "loss": 2.7306, "step": 39038 }, { "epoch": 2.423427897448631, "grad_norm": 0.4787784555260384, "learning_rate": 1.08361536683108e-05, "loss": 2.6995, "step": 39039 }, { "epoch": 2.423489974548389, "grad_norm": 0.4382391033885237, "learning_rate": 1.0833908583587127e-05, "loss": 2.6107, "step": 39040 }, { "epoch": 2.423552051648147, "grad_norm": 0.4815801157071827, "learning_rate": 1.083166370320246e-05, "loss": 2.6279, "step": 39041 }, { "epoch": 2.423614128747905, "grad_norm": 0.45915087662644, "learning_rate": 1.0829419027168497e-05, "loss": 2.6866, "step": 39042 }, { "epoch": 2.423676205847663, "grad_norm": 0.46016845557406133, "learning_rate": 1.0827174555496978e-05, "loss": 2.6836, "step": 39043 }, { "epoch": 2.4237382829474208, "grad_norm": 0.4768277977596885, "learning_rate": 1.0824930288199603e-05, "loss": 2.7267, "step": 39044 }, { "epoch": 2.4238003600471787, "grad_norm": 0.5065597340192356, "learning_rate": 1.0822686225288059e-05, "loss": 2.7034, "step": 39045 }, { "epoch": 2.4238624371469366, "grad_norm": 0.4780691905890675, "learning_rate": 1.0820442366774081e-05, "loss": 2.6274, "step": 39046 }, { "epoch": 2.4239245142466945, "grad_norm": 0.4646039552630672, "learning_rate": 1.0818198712669375e-05, "loss": 2.6584, "step": 39047 }, { "epoch": 2.4239865913464524, "grad_norm": 0.4703200114718081, "learning_rate": 1.0815955262985634e-05, "loss": 2.7052, "step": 39048 }, { "epoch": 2.4240486684462104, "grad_norm": 0.4039502708308552, "learning_rate": 1.0813712017734551e-05, "loss": 2.6498, "step": 39049 }, { "epoch": 2.4241107455459683, "grad_norm": 0.469542199513348, "learning_rate": 1.0811468976927863e-05, "loss": 2.6795, "step": 39050 }, { "epoch": 2.424172822645726, "grad_norm": 0.494465024108485, "learning_rate": 1.080922614057725e-05, "loss": 2.7044, "step": 39051 }, { "epoch": 2.4242348997454837, "grad_norm": 0.46830108127090386, "learning_rate": 1.0806983508694424e-05, "loss": 2.6218, "step": 39052 }, { "epoch": 2.424296976845242, "grad_norm": 0.44378035468392696, "learning_rate": 1.0804741081291076e-05, "loss": 2.7209, "step": 39053 }, { "epoch": 2.4243590539449995, "grad_norm": 0.4565387746036876, "learning_rate": 1.080249885837889e-05, "loss": 2.6881, "step": 39054 }, { "epoch": 2.4244211310447574, "grad_norm": 0.43701765173608786, "learning_rate": 1.0800256839969602e-05, "loss": 2.6464, "step": 39055 }, { "epoch": 2.4244832081445153, "grad_norm": 0.43859144715077286, "learning_rate": 1.0798015026074893e-05, "loss": 2.6557, "step": 39056 }, { "epoch": 2.4245452852442733, "grad_norm": 0.48473695336698147, "learning_rate": 1.0795773416706451e-05, "loss": 2.7133, "step": 39057 }, { "epoch": 2.424607362344031, "grad_norm": 0.4844464433957023, "learning_rate": 1.0793532011875968e-05, "loss": 2.6929, "step": 39058 }, { "epoch": 2.424669439443789, "grad_norm": 0.42877257649117845, "learning_rate": 1.0791290811595156e-05, "loss": 2.7119, "step": 39059 }, { "epoch": 2.424731516543547, "grad_norm": 0.4626695451046856, "learning_rate": 1.0789049815875696e-05, "loss": 2.7291, "step": 39060 }, { "epoch": 2.424793593643305, "grad_norm": 0.467069809420376, "learning_rate": 1.0786809024729289e-05, "loss": 2.7156, "step": 39061 }, { "epoch": 2.424855670743063, "grad_norm": 0.4231301462601469, "learning_rate": 1.0784568438167597e-05, "loss": 2.6095, "step": 39062 }, { "epoch": 2.4249177478428208, "grad_norm": 0.45194890870128085, "learning_rate": 1.078232805620235e-05, "loss": 2.534, "step": 39063 }, { "epoch": 2.4249798249425787, "grad_norm": 0.44818714942540727, "learning_rate": 1.078008787884522e-05, "loss": 2.6312, "step": 39064 }, { "epoch": 2.4250419020423366, "grad_norm": 0.43608144277375077, "learning_rate": 1.0777847906107885e-05, "loss": 2.6551, "step": 39065 }, { "epoch": 2.4251039791420945, "grad_norm": 0.4516847181085409, "learning_rate": 1.077560813800204e-05, "loss": 2.6767, "step": 39066 }, { "epoch": 2.4251660562418524, "grad_norm": 0.47598957580884743, "learning_rate": 1.0773368574539362e-05, "loss": 2.7232, "step": 39067 }, { "epoch": 2.4252281333416104, "grad_norm": 0.42509454068339825, "learning_rate": 1.0771129215731552e-05, "loss": 2.6398, "step": 39068 }, { "epoch": 2.4252902104413683, "grad_norm": 0.4147945267089973, "learning_rate": 1.0768890061590287e-05, "loss": 2.5904, "step": 39069 }, { "epoch": 2.425352287541126, "grad_norm": 0.4700069106385284, "learning_rate": 1.0766651112127246e-05, "loss": 2.7734, "step": 39070 }, { "epoch": 2.425414364640884, "grad_norm": 0.4602641181614219, "learning_rate": 1.0764412367354092e-05, "loss": 2.6857, "step": 39071 }, { "epoch": 2.425476441740642, "grad_norm": 0.4148982987397263, "learning_rate": 1.076217382728254e-05, "loss": 2.7616, "step": 39072 }, { "epoch": 2.4255385188404, "grad_norm": 0.42073804804258713, "learning_rate": 1.0759935491924256e-05, "loss": 2.6886, "step": 39073 }, { "epoch": 2.425600595940158, "grad_norm": 0.43773942594866766, "learning_rate": 1.0757697361290908e-05, "loss": 2.6831, "step": 39074 }, { "epoch": 2.4256626730399153, "grad_norm": 0.42183450221903673, "learning_rate": 1.0755459435394171e-05, "loss": 2.7026, "step": 39075 }, { "epoch": 2.4257247501396737, "grad_norm": 0.5022168783865444, "learning_rate": 1.0753221714245726e-05, "loss": 2.6448, "step": 39076 }, { "epoch": 2.425786827239431, "grad_norm": 0.445432429309821, "learning_rate": 1.0750984197857266e-05, "loss": 2.6554, "step": 39077 }, { "epoch": 2.425848904339189, "grad_norm": 0.4773929739138209, "learning_rate": 1.0748746886240446e-05, "loss": 2.6772, "step": 39078 }, { "epoch": 2.425910981438947, "grad_norm": 0.43627405403628794, "learning_rate": 1.0746509779406949e-05, "loss": 2.705, "step": 39079 }, { "epoch": 2.425973058538705, "grad_norm": 0.44698011455851727, "learning_rate": 1.0744272877368433e-05, "loss": 2.6211, "step": 39080 }, { "epoch": 2.426035135638463, "grad_norm": 0.46789721163452663, "learning_rate": 1.074203618013656e-05, "loss": 2.5989, "step": 39081 }, { "epoch": 2.4260972127382208, "grad_norm": 0.46462515491650297, "learning_rate": 1.0739799687723035e-05, "loss": 2.7241, "step": 39082 }, { "epoch": 2.4261592898379787, "grad_norm": 0.4600632507179208, "learning_rate": 1.0737563400139505e-05, "loss": 2.6775, "step": 39083 }, { "epoch": 2.4262213669377366, "grad_norm": 0.4298049592819263, "learning_rate": 1.0735327317397637e-05, "loss": 2.7094, "step": 39084 }, { "epoch": 2.4262834440374945, "grad_norm": 0.4025983467849334, "learning_rate": 1.073309143950908e-05, "loss": 2.6269, "step": 39085 }, { "epoch": 2.4263455211372524, "grad_norm": 0.4521312174803282, "learning_rate": 1.0730855766485536e-05, "loss": 2.75, "step": 39086 }, { "epoch": 2.4264075982370104, "grad_norm": 0.47224783788966596, "learning_rate": 1.0728620298338649e-05, "loss": 2.6988, "step": 39087 }, { "epoch": 2.4264696753367683, "grad_norm": 0.4360113190802465, "learning_rate": 1.0726385035080083e-05, "loss": 2.6896, "step": 39088 }, { "epoch": 2.426531752436526, "grad_norm": 0.4199476899484657, "learning_rate": 1.0724149976721488e-05, "loss": 2.6402, "step": 39089 }, { "epoch": 2.426593829536284, "grad_norm": 0.47618331832842103, "learning_rate": 1.0721915123274545e-05, "loss": 2.6304, "step": 39090 }, { "epoch": 2.426655906636042, "grad_norm": 0.43471824614047494, "learning_rate": 1.071968047475091e-05, "loss": 2.7258, "step": 39091 }, { "epoch": 2.4267179837358, "grad_norm": 0.44542461790809357, "learning_rate": 1.0717446031162236e-05, "loss": 2.7275, "step": 39092 }, { "epoch": 2.426780060835558, "grad_norm": 0.43306844697145513, "learning_rate": 1.0715211792520186e-05, "loss": 2.6955, "step": 39093 }, { "epoch": 2.426842137935316, "grad_norm": 0.4380897906164146, "learning_rate": 1.0712977758836402e-05, "loss": 2.5989, "step": 39094 }, { "epoch": 2.4269042150350737, "grad_norm": 0.43500507338075584, "learning_rate": 1.0710743930122558e-05, "loss": 2.7132, "step": 39095 }, { "epoch": 2.4269662921348316, "grad_norm": 0.5001394852149716, "learning_rate": 1.0708510306390301e-05, "loss": 2.7074, "step": 39096 }, { "epoch": 2.4270283692345895, "grad_norm": 0.41210472863637926, "learning_rate": 1.0706276887651284e-05, "loss": 2.639, "step": 39097 }, { "epoch": 2.4270904463343475, "grad_norm": 0.4141005552546084, "learning_rate": 1.0704043673917147e-05, "loss": 2.7279, "step": 39098 }, { "epoch": 2.4271525234341054, "grad_norm": 0.433245304073412, "learning_rate": 1.0701810665199568e-05, "loss": 2.7257, "step": 39099 }, { "epoch": 2.427214600533863, "grad_norm": 0.4515657068422486, "learning_rate": 1.0699577861510179e-05, "loss": 2.6673, "step": 39100 }, { "epoch": 2.427276677633621, "grad_norm": 0.39699831195803925, "learning_rate": 1.0697345262860636e-05, "loss": 2.6824, "step": 39101 }, { "epoch": 2.4273387547333787, "grad_norm": 0.4463040539249907, "learning_rate": 1.0695112869262586e-05, "loss": 2.6182, "step": 39102 }, { "epoch": 2.4274008318331366, "grad_norm": 0.4608261006111342, "learning_rate": 1.0692880680727656e-05, "loss": 2.6518, "step": 39103 }, { "epoch": 2.4274629089328945, "grad_norm": 0.40226870827707323, "learning_rate": 1.0690648697267524e-05, "loss": 2.6727, "step": 39104 }, { "epoch": 2.4275249860326524, "grad_norm": 0.40806677699465904, "learning_rate": 1.0688416918893823e-05, "loss": 2.7206, "step": 39105 }, { "epoch": 2.4275870631324103, "grad_norm": 0.44870174345073516, "learning_rate": 1.0686185345618187e-05, "loss": 2.6288, "step": 39106 }, { "epoch": 2.4276491402321683, "grad_norm": 0.4617356324699773, "learning_rate": 1.068395397745226e-05, "loss": 2.7408, "step": 39107 }, { "epoch": 2.427711217331926, "grad_norm": 0.41958194547131344, "learning_rate": 1.0681722814407686e-05, "loss": 2.6938, "step": 39108 }, { "epoch": 2.427773294431684, "grad_norm": 0.4229613903745516, "learning_rate": 1.0679491856496127e-05, "loss": 2.7018, "step": 39109 }, { "epoch": 2.427835371531442, "grad_norm": 0.4290864381829087, "learning_rate": 1.0677261103729197e-05, "loss": 2.6706, "step": 39110 }, { "epoch": 2.4278974486312, "grad_norm": 0.42868766634612937, "learning_rate": 1.0675030556118549e-05, "loss": 2.6459, "step": 39111 }, { "epoch": 2.427959525730958, "grad_norm": 0.4110205490261107, "learning_rate": 1.0672800213675793e-05, "loss": 2.701, "step": 39112 }, { "epoch": 2.4280216028307158, "grad_norm": 0.4493023504335813, "learning_rate": 1.0670570076412601e-05, "loss": 2.7189, "step": 39113 }, { "epoch": 2.4280836799304737, "grad_norm": 0.4001391854123814, "learning_rate": 1.0668340144340594e-05, "loss": 2.665, "step": 39114 }, { "epoch": 2.4281457570302316, "grad_norm": 0.41389834550889537, "learning_rate": 1.0666110417471403e-05, "loss": 2.6116, "step": 39115 }, { "epoch": 2.4282078341299895, "grad_norm": 0.436826204104334, "learning_rate": 1.0663880895816663e-05, "loss": 2.7433, "step": 39116 }, { "epoch": 2.4282699112297474, "grad_norm": 0.39930647168077427, "learning_rate": 1.0661651579387994e-05, "loss": 2.5836, "step": 39117 }, { "epoch": 2.4283319883295054, "grad_norm": 0.44180116457420854, "learning_rate": 1.065942246819705e-05, "loss": 2.6186, "step": 39118 }, { "epoch": 2.4283940654292633, "grad_norm": 0.477638031546895, "learning_rate": 1.0657193562255452e-05, "loss": 2.6262, "step": 39119 }, { "epoch": 2.428456142529021, "grad_norm": 0.4233810070401208, "learning_rate": 1.0654964861574823e-05, "loss": 2.6523, "step": 39120 }, { "epoch": 2.428518219628779, "grad_norm": 0.4504733437294989, "learning_rate": 1.0652736366166782e-05, "loss": 2.7109, "step": 39121 }, { "epoch": 2.428580296728537, "grad_norm": 0.4499260848322501, "learning_rate": 1.0650508076042975e-05, "loss": 2.6795, "step": 39122 }, { "epoch": 2.4286423738282945, "grad_norm": 0.44036269259897576, "learning_rate": 1.0648279991215026e-05, "loss": 2.6614, "step": 39123 }, { "epoch": 2.428704450928053, "grad_norm": 0.4411996847594701, "learning_rate": 1.0646052111694555e-05, "loss": 2.663, "step": 39124 }, { "epoch": 2.4287665280278103, "grad_norm": 0.46906089009841656, "learning_rate": 1.0643824437493166e-05, "loss": 2.6906, "step": 39125 }, { "epoch": 2.4288286051275683, "grad_norm": 0.45661463863418444, "learning_rate": 1.0641596968622513e-05, "loss": 2.7178, "step": 39126 }, { "epoch": 2.428890682227326, "grad_norm": 0.49734822057020445, "learning_rate": 1.0639369705094203e-05, "loss": 2.7285, "step": 39127 }, { "epoch": 2.428952759327084, "grad_norm": 0.4512928722019041, "learning_rate": 1.0637142646919857e-05, "loss": 2.6151, "step": 39128 }, { "epoch": 2.429014836426842, "grad_norm": 0.49950429066255736, "learning_rate": 1.0634915794111094e-05, "loss": 2.7544, "step": 39129 }, { "epoch": 2.4290769135266, "grad_norm": 0.4586541631570172, "learning_rate": 1.0632689146679514e-05, "loss": 2.6337, "step": 39130 }, { "epoch": 2.429138990626358, "grad_norm": 0.4569655314088691, "learning_rate": 1.0630462704636768e-05, "loss": 2.6525, "step": 39131 }, { "epoch": 2.4292010677261158, "grad_norm": 0.4794554686617247, "learning_rate": 1.0628236467994456e-05, "loss": 2.6538, "step": 39132 }, { "epoch": 2.4292631448258737, "grad_norm": 0.4517490938617715, "learning_rate": 1.0626010436764195e-05, "loss": 2.7475, "step": 39133 }, { "epoch": 2.4293252219256316, "grad_norm": 0.48808083045000333, "learning_rate": 1.0623784610957576e-05, "loss": 2.624, "step": 39134 }, { "epoch": 2.4293872990253895, "grad_norm": 0.470534599178272, "learning_rate": 1.062155899058625e-05, "loss": 2.6289, "step": 39135 }, { "epoch": 2.4294493761251474, "grad_norm": 0.4311288933084607, "learning_rate": 1.0619333575661804e-05, "loss": 2.7057, "step": 39136 }, { "epoch": 2.4295114532249054, "grad_norm": 0.4554775106330185, "learning_rate": 1.061710836619586e-05, "loss": 2.6747, "step": 39137 }, { "epoch": 2.4295735303246633, "grad_norm": 0.4307091790958227, "learning_rate": 1.0614883362200024e-05, "loss": 2.7229, "step": 39138 }, { "epoch": 2.429635607424421, "grad_norm": 0.4694860877788087, "learning_rate": 1.0612658563685884e-05, "loss": 2.679, "step": 39139 }, { "epoch": 2.429697684524179, "grad_norm": 0.46358618004625607, "learning_rate": 1.0610433970665085e-05, "loss": 2.6066, "step": 39140 }, { "epoch": 2.429759761623937, "grad_norm": 0.4304361929931117, "learning_rate": 1.0608209583149198e-05, "loss": 2.6223, "step": 39141 }, { "epoch": 2.429821838723695, "grad_norm": 0.46187305873313195, "learning_rate": 1.0605985401149854e-05, "loss": 2.6842, "step": 39142 }, { "epoch": 2.429883915823453, "grad_norm": 0.47147733087342264, "learning_rate": 1.0603761424678654e-05, "loss": 2.7977, "step": 39143 }, { "epoch": 2.429945992923211, "grad_norm": 0.44960120133196196, "learning_rate": 1.0601537653747174e-05, "loss": 2.6927, "step": 39144 }, { "epoch": 2.4300080700229687, "grad_norm": 0.41586953047012665, "learning_rate": 1.0599314088367056e-05, "loss": 2.6209, "step": 39145 }, { "epoch": 2.4300701471227266, "grad_norm": 0.4892131762215997, "learning_rate": 1.059709072854988e-05, "loss": 2.6683, "step": 39146 }, { "epoch": 2.4301322242224845, "grad_norm": 0.44910127513511655, "learning_rate": 1.0594867574307243e-05, "loss": 2.6943, "step": 39147 }, { "epoch": 2.430194301322242, "grad_norm": 0.4697171723821045, "learning_rate": 1.0592644625650739e-05, "loss": 2.7518, "step": 39148 }, { "epoch": 2.430256378422, "grad_norm": 0.47375378194522316, "learning_rate": 1.0590421882591984e-05, "loss": 2.6817, "step": 39149 }, { "epoch": 2.430318455521758, "grad_norm": 0.4732015789901673, "learning_rate": 1.0588199345142569e-05, "loss": 2.6938, "step": 39150 }, { "epoch": 2.4303805326215158, "grad_norm": 0.42734783596059067, "learning_rate": 1.058597701331408e-05, "loss": 2.6492, "step": 39151 }, { "epoch": 2.4304426097212737, "grad_norm": 0.45281467005470977, "learning_rate": 1.0583754887118125e-05, "loss": 2.714, "step": 39152 }, { "epoch": 2.4305046868210316, "grad_norm": 0.4282955860075262, "learning_rate": 1.0581532966566272e-05, "loss": 2.6266, "step": 39153 }, { "epoch": 2.4305667639207895, "grad_norm": 0.3963275467717821, "learning_rate": 1.0579311251670144e-05, "loss": 2.7016, "step": 39154 }, { "epoch": 2.4306288410205474, "grad_norm": 0.43295786169011324, "learning_rate": 1.0577089742441315e-05, "loss": 2.6416, "step": 39155 }, { "epoch": 2.4306909181203054, "grad_norm": 0.44366600759542296, "learning_rate": 1.0574868438891384e-05, "loss": 2.6866, "step": 39156 }, { "epoch": 2.4307529952200633, "grad_norm": 0.45579204307229315, "learning_rate": 1.0572647341031922e-05, "loss": 2.6969, "step": 39157 }, { "epoch": 2.430815072319821, "grad_norm": 0.41391510409270205, "learning_rate": 1.0570426448874537e-05, "loss": 2.6089, "step": 39158 }, { "epoch": 2.430877149419579, "grad_norm": 0.410832311127315, "learning_rate": 1.0568205762430816e-05, "loss": 2.6304, "step": 39159 }, { "epoch": 2.430939226519337, "grad_norm": 0.4278652043337116, "learning_rate": 1.0565985281712337e-05, "loss": 2.606, "step": 39160 }, { "epoch": 2.431001303619095, "grad_norm": 0.42079508280265315, "learning_rate": 1.0563765006730669e-05, "loss": 2.6589, "step": 39161 }, { "epoch": 2.431063380718853, "grad_norm": 0.3965802757902541, "learning_rate": 1.056154493749743e-05, "loss": 2.7541, "step": 39162 }, { "epoch": 2.431125457818611, "grad_norm": 0.4447120667492114, "learning_rate": 1.0559325074024185e-05, "loss": 2.6936, "step": 39163 }, { "epoch": 2.4311875349183687, "grad_norm": 0.47625979205700525, "learning_rate": 1.055710541632251e-05, "loss": 2.6676, "step": 39164 }, { "epoch": 2.4312496120181266, "grad_norm": 0.4189622426558021, "learning_rate": 1.0554885964403994e-05, "loss": 2.6802, "step": 39165 }, { "epoch": 2.4313116891178845, "grad_norm": 0.44494556903126753, "learning_rate": 1.0552666718280202e-05, "loss": 2.6099, "step": 39166 }, { "epoch": 2.4313737662176425, "grad_norm": 0.46422660957084916, "learning_rate": 1.0550447677962738e-05, "loss": 2.6884, "step": 39167 }, { "epoch": 2.4314358433174004, "grad_norm": 0.4389673218110845, "learning_rate": 1.0548228843463164e-05, "loss": 2.6588, "step": 39168 }, { "epoch": 2.4314979204171583, "grad_norm": 0.4252758000759254, "learning_rate": 1.0546010214793057e-05, "loss": 2.7083, "step": 39169 }, { "epoch": 2.431559997516916, "grad_norm": 0.43722433533549016, "learning_rate": 1.0543791791963975e-05, "loss": 2.6404, "step": 39170 }, { "epoch": 2.4316220746166737, "grad_norm": 0.4396359042360154, "learning_rate": 1.0541573574987529e-05, "loss": 2.7531, "step": 39171 }, { "epoch": 2.431684151716432, "grad_norm": 0.43505001380757297, "learning_rate": 1.0539355563875269e-05, "loss": 2.7062, "step": 39172 }, { "epoch": 2.4317462288161895, "grad_norm": 0.3934156868109082, "learning_rate": 1.053713775863876e-05, "loss": 2.697, "step": 39173 }, { "epoch": 2.4318083059159474, "grad_norm": 0.44538611088077606, "learning_rate": 1.0534920159289597e-05, "loss": 2.6624, "step": 39174 }, { "epoch": 2.4318703830157054, "grad_norm": 0.4523433817069707, "learning_rate": 1.0532702765839319e-05, "loss": 2.7245, "step": 39175 }, { "epoch": 2.4319324601154633, "grad_norm": 0.414330485730553, "learning_rate": 1.0530485578299526e-05, "loss": 2.7331, "step": 39176 }, { "epoch": 2.431994537215221, "grad_norm": 0.4174882854599906, "learning_rate": 1.0528268596681773e-05, "loss": 2.6453, "step": 39177 }, { "epoch": 2.432056614314979, "grad_norm": 0.4784909767497928, "learning_rate": 1.0526051820997623e-05, "loss": 2.6507, "step": 39178 }, { "epoch": 2.432118691414737, "grad_norm": 0.420701089073079, "learning_rate": 1.0523835251258646e-05, "loss": 2.649, "step": 39179 }, { "epoch": 2.432180768514495, "grad_norm": 0.4336221874637394, "learning_rate": 1.0521618887476387e-05, "loss": 2.705, "step": 39180 }, { "epoch": 2.432242845614253, "grad_norm": 0.46615933730812686, "learning_rate": 1.0519402729662441e-05, "loss": 2.6211, "step": 39181 }, { "epoch": 2.4323049227140108, "grad_norm": 0.4407944955225096, "learning_rate": 1.0517186777828358e-05, "loss": 2.6526, "step": 39182 }, { "epoch": 2.4323669998137687, "grad_norm": 0.47155687458988094, "learning_rate": 1.0514971031985693e-05, "loss": 2.7931, "step": 39183 }, { "epoch": 2.4324290769135266, "grad_norm": 0.43830904065046233, "learning_rate": 1.0512755492145997e-05, "loss": 2.6494, "step": 39184 }, { "epoch": 2.4324911540132845, "grad_norm": 0.4780505352999169, "learning_rate": 1.0510540158320852e-05, "loss": 2.6884, "step": 39185 }, { "epoch": 2.4325532311130424, "grad_norm": 0.4371186876711181, "learning_rate": 1.050832503052181e-05, "loss": 2.6566, "step": 39186 }, { "epoch": 2.4326153082128004, "grad_norm": 0.4312018498892816, "learning_rate": 1.050611010876042e-05, "loss": 2.6347, "step": 39187 }, { "epoch": 2.4326773853125583, "grad_norm": 0.42668381667066496, "learning_rate": 1.050389539304824e-05, "loss": 2.6438, "step": 39188 }, { "epoch": 2.432739462412316, "grad_norm": 0.4221134218498172, "learning_rate": 1.0501680883396819e-05, "loss": 2.5815, "step": 39189 }, { "epoch": 2.432801539512074, "grad_norm": 0.45124167360670714, "learning_rate": 1.0499466579817724e-05, "loss": 2.6524, "step": 39190 }, { "epoch": 2.432863616611832, "grad_norm": 0.4328529283942484, "learning_rate": 1.0497252482322506e-05, "loss": 2.6443, "step": 39191 }, { "epoch": 2.43292569371159, "grad_norm": 0.47677138706558786, "learning_rate": 1.0495038590922706e-05, "loss": 2.694, "step": 39192 }, { "epoch": 2.432987770811348, "grad_norm": 0.4262072156034417, "learning_rate": 1.0492824905629866e-05, "loss": 2.7193, "step": 39193 }, { "epoch": 2.4330498479111053, "grad_norm": 0.45616645178663784, "learning_rate": 1.049061142645557e-05, "loss": 2.6645, "step": 39194 }, { "epoch": 2.4331119250108637, "grad_norm": 0.4495488156681203, "learning_rate": 1.0488398153411338e-05, "loss": 2.7335, "step": 39195 }, { "epoch": 2.433174002110621, "grad_norm": 0.4469991331486756, "learning_rate": 1.0486185086508732e-05, "loss": 2.6798, "step": 39196 }, { "epoch": 2.433236079210379, "grad_norm": 0.45480302345936685, "learning_rate": 1.048397222575927e-05, "loss": 2.655, "step": 39197 }, { "epoch": 2.433298156310137, "grad_norm": 0.4813045891851107, "learning_rate": 1.0481759571174531e-05, "loss": 2.6792, "step": 39198 }, { "epoch": 2.433360233409895, "grad_norm": 0.445955631900634, "learning_rate": 1.0479547122766054e-05, "loss": 2.7297, "step": 39199 }, { "epoch": 2.433422310509653, "grad_norm": 0.43767639418863824, "learning_rate": 1.0477334880545364e-05, "loss": 2.6676, "step": 39200 }, { "epoch": 2.4334843876094108, "grad_norm": 0.4920437618056707, "learning_rate": 1.0475122844524021e-05, "loss": 2.7183, "step": 39201 }, { "epoch": 2.4335464647091687, "grad_norm": 0.4952890522939382, "learning_rate": 1.0472911014713538e-05, "loss": 2.5756, "step": 39202 }, { "epoch": 2.4336085418089266, "grad_norm": 0.4747159704745123, "learning_rate": 1.0470699391125488e-05, "loss": 2.6308, "step": 39203 }, { "epoch": 2.4336706189086845, "grad_norm": 0.4658714282382782, "learning_rate": 1.0468487973771396e-05, "loss": 2.648, "step": 39204 }, { "epoch": 2.4337326960084424, "grad_norm": 0.539340624725006, "learning_rate": 1.0466276762662802e-05, "loss": 2.7315, "step": 39205 }, { "epoch": 2.4337947731082004, "grad_norm": 0.515041483532362, "learning_rate": 1.046406575781122e-05, "loss": 2.7181, "step": 39206 }, { "epoch": 2.4338568502079583, "grad_norm": 0.45058581447723656, "learning_rate": 1.0461854959228207e-05, "loss": 2.5777, "step": 39207 }, { "epoch": 2.433918927307716, "grad_norm": 0.4921316539022926, "learning_rate": 1.0459644366925309e-05, "loss": 2.6298, "step": 39208 }, { "epoch": 2.433981004407474, "grad_norm": 0.4922146562623179, "learning_rate": 1.0457433980914044e-05, "loss": 2.5937, "step": 39209 }, { "epoch": 2.434043081507232, "grad_norm": 0.48595464682301653, "learning_rate": 1.0455223801205948e-05, "loss": 2.6615, "step": 39210 }, { "epoch": 2.43410515860699, "grad_norm": 0.4764116392396934, "learning_rate": 1.045301382781253e-05, "loss": 2.7499, "step": 39211 }, { "epoch": 2.434167235706748, "grad_norm": 0.5184378627065318, "learning_rate": 1.0450804060745362e-05, "loss": 2.7122, "step": 39212 }, { "epoch": 2.434229312806506, "grad_norm": 0.4853871752685284, "learning_rate": 1.044859450001594e-05, "loss": 2.6263, "step": 39213 }, { "epoch": 2.4342913899062637, "grad_norm": 0.48257613707977953, "learning_rate": 1.0446385145635807e-05, "loss": 2.6414, "step": 39214 }, { "epoch": 2.4343534670060216, "grad_norm": 0.4455973232978628, "learning_rate": 1.0444175997616485e-05, "loss": 2.5777, "step": 39215 }, { "epoch": 2.4344155441057795, "grad_norm": 0.4992524757992441, "learning_rate": 1.0441967055969487e-05, "loss": 2.6791, "step": 39216 }, { "epoch": 2.4344776212055375, "grad_norm": 0.4458182792112503, "learning_rate": 1.043975832070636e-05, "loss": 2.6752, "step": 39217 }, { "epoch": 2.4345396983052954, "grad_norm": 0.4822293565374849, "learning_rate": 1.0437549791838619e-05, "loss": 2.7373, "step": 39218 }, { "epoch": 2.434601775405053, "grad_norm": 0.46530157672058264, "learning_rate": 1.0435341469377785e-05, "loss": 2.6678, "step": 39219 }, { "epoch": 2.434663852504811, "grad_norm": 0.5084223093531397, "learning_rate": 1.0433133353335368e-05, "loss": 2.6417, "step": 39220 }, { "epoch": 2.4347259296045687, "grad_norm": 0.5362884906924407, "learning_rate": 1.0430925443722906e-05, "loss": 2.6214, "step": 39221 }, { "epoch": 2.4347880067043266, "grad_norm": 0.5143324227400206, "learning_rate": 1.042871774055192e-05, "loss": 2.7101, "step": 39222 }, { "epoch": 2.4348500838040845, "grad_norm": 0.49809873722297743, "learning_rate": 1.0426510243833914e-05, "loss": 2.6359, "step": 39223 }, { "epoch": 2.4349121609038424, "grad_norm": 0.5053058969211344, "learning_rate": 1.0424302953580412e-05, "loss": 2.5816, "step": 39224 }, { "epoch": 2.4349742380036004, "grad_norm": 0.50012911679176, "learning_rate": 1.0422095869802917e-05, "loss": 2.664, "step": 39225 }, { "epoch": 2.4350363151033583, "grad_norm": 0.5335358360664078, "learning_rate": 1.0419888992512966e-05, "loss": 2.7124, "step": 39226 }, { "epoch": 2.435098392203116, "grad_norm": 0.4774693007112407, "learning_rate": 1.0417682321722066e-05, "loss": 2.6252, "step": 39227 }, { "epoch": 2.435160469302874, "grad_norm": 0.47587099090262175, "learning_rate": 1.0415475857441726e-05, "loss": 2.6635, "step": 39228 }, { "epoch": 2.435222546402632, "grad_norm": 0.43804577436286757, "learning_rate": 1.0413269599683446e-05, "loss": 2.7364, "step": 39229 }, { "epoch": 2.43528462350239, "grad_norm": 0.4845002227857489, "learning_rate": 1.0411063548458754e-05, "loss": 2.6809, "step": 39230 }, { "epoch": 2.435346700602148, "grad_norm": 0.47749420454592595, "learning_rate": 1.0408857703779163e-05, "loss": 2.6946, "step": 39231 }, { "epoch": 2.435408777701906, "grad_norm": 0.49646392287921504, "learning_rate": 1.040665206565617e-05, "loss": 2.5843, "step": 39232 }, { "epoch": 2.4354708548016637, "grad_norm": 0.4636001621441891, "learning_rate": 1.0404446634101267e-05, "loss": 2.5466, "step": 39233 }, { "epoch": 2.4355329319014216, "grad_norm": 0.47218501031649696, "learning_rate": 1.0402241409125996e-05, "loss": 2.7249, "step": 39234 }, { "epoch": 2.4355950090011795, "grad_norm": 0.47908803218579393, "learning_rate": 1.0400036390741847e-05, "loss": 2.6325, "step": 39235 }, { "epoch": 2.4356570861009375, "grad_norm": 0.44419940905040023, "learning_rate": 1.0397831578960315e-05, "loss": 2.7198, "step": 39236 }, { "epoch": 2.4357191632006954, "grad_norm": 0.4390700132602774, "learning_rate": 1.039562697379291e-05, "loss": 2.6247, "step": 39237 }, { "epoch": 2.4357812403004533, "grad_norm": 0.5138309073028474, "learning_rate": 1.0393422575251121e-05, "loss": 2.6963, "step": 39238 }, { "epoch": 2.435843317400211, "grad_norm": 0.4562069232410325, "learning_rate": 1.0391218383346462e-05, "loss": 2.7003, "step": 39239 }, { "epoch": 2.435905394499969, "grad_norm": 0.4012234284168064, "learning_rate": 1.0389014398090447e-05, "loss": 2.6466, "step": 39240 }, { "epoch": 2.435967471599727, "grad_norm": 0.4323838683313714, "learning_rate": 1.0386810619494559e-05, "loss": 2.6526, "step": 39241 }, { "epoch": 2.4360295486994845, "grad_norm": 0.4285118840070811, "learning_rate": 1.0384607047570293e-05, "loss": 2.7454, "step": 39242 }, { "epoch": 2.436091625799243, "grad_norm": 0.43692314060075493, "learning_rate": 1.038240368232914e-05, "loss": 2.7207, "step": 39243 }, { "epoch": 2.4361537028990004, "grad_norm": 0.43586314055967823, "learning_rate": 1.0380200523782619e-05, "loss": 2.6749, "step": 39244 }, { "epoch": 2.4362157799987583, "grad_norm": 0.45843214537672966, "learning_rate": 1.0377997571942206e-05, "loss": 2.7021, "step": 39245 }, { "epoch": 2.436277857098516, "grad_norm": 0.4321912472687274, "learning_rate": 1.0375794826819402e-05, "loss": 2.5795, "step": 39246 }, { "epoch": 2.436339934198274, "grad_norm": 0.3987394198745003, "learning_rate": 1.0373592288425676e-05, "loss": 2.6451, "step": 39247 }, { "epoch": 2.436402011298032, "grad_norm": 0.4242068238101144, "learning_rate": 1.037138995677256e-05, "loss": 2.6328, "step": 39248 }, { "epoch": 2.43646408839779, "grad_norm": 0.450568468592757, "learning_rate": 1.0369187831871518e-05, "loss": 2.7238, "step": 39249 }, { "epoch": 2.436526165497548, "grad_norm": 0.4224832882482212, "learning_rate": 1.036698591373404e-05, "loss": 2.6128, "step": 39250 }, { "epoch": 2.4365882425973058, "grad_norm": 0.4223077814759642, "learning_rate": 1.0364784202371625e-05, "loss": 2.6415, "step": 39251 }, { "epoch": 2.4366503196970637, "grad_norm": 0.4730882010769153, "learning_rate": 1.0362582697795736e-05, "loss": 2.6988, "step": 39252 }, { "epoch": 2.4367123967968216, "grad_norm": 0.46419632327547883, "learning_rate": 1.0360381400017887e-05, "loss": 2.5893, "step": 39253 }, { "epoch": 2.4367744738965795, "grad_norm": 0.41481855930203804, "learning_rate": 1.0358180309049553e-05, "loss": 2.7345, "step": 39254 }, { "epoch": 2.4368365509963374, "grad_norm": 0.45549408071756636, "learning_rate": 1.0355979424902218e-05, "loss": 2.7282, "step": 39255 }, { "epoch": 2.4368986280960954, "grad_norm": 0.44133997731931723, "learning_rate": 1.0353778747587344e-05, "loss": 2.6326, "step": 39256 }, { "epoch": 2.4369607051958533, "grad_norm": 0.4229118533244212, "learning_rate": 1.0351578277116447e-05, "loss": 2.6415, "step": 39257 }, { "epoch": 2.437022782295611, "grad_norm": 0.4111820834757349, "learning_rate": 1.0349378013500993e-05, "loss": 2.6791, "step": 39258 }, { "epoch": 2.437084859395369, "grad_norm": 0.4544719539084269, "learning_rate": 1.0347177956752457e-05, "loss": 2.6438, "step": 39259 }, { "epoch": 2.437146936495127, "grad_norm": 0.4136550806254796, "learning_rate": 1.0344978106882309e-05, "loss": 2.534, "step": 39260 }, { "epoch": 2.437209013594885, "grad_norm": 0.4485480567676573, "learning_rate": 1.0342778463902048e-05, "loss": 2.7016, "step": 39261 }, { "epoch": 2.437271090694643, "grad_norm": 0.4310001091569731, "learning_rate": 1.0340579027823133e-05, "loss": 2.7045, "step": 39262 }, { "epoch": 2.437333167794401, "grad_norm": 0.43292075844863487, "learning_rate": 1.0338379798657056e-05, "loss": 2.7084, "step": 39263 }, { "epoch": 2.4373952448941587, "grad_norm": 0.43463254332997353, "learning_rate": 1.0336180776415271e-05, "loss": 2.6654, "step": 39264 }, { "epoch": 2.4374573219939166, "grad_norm": 0.44002404320656563, "learning_rate": 1.0333981961109252e-05, "loss": 2.6959, "step": 39265 }, { "epoch": 2.4375193990936745, "grad_norm": 0.4523438264257073, "learning_rate": 1.0331783352750484e-05, "loss": 2.6683, "step": 39266 }, { "epoch": 2.437581476193432, "grad_norm": 0.43618271265055103, "learning_rate": 1.032958495135044e-05, "loss": 2.6987, "step": 39267 }, { "epoch": 2.4376435532931904, "grad_norm": 0.4236185812769177, "learning_rate": 1.0327386756920582e-05, "loss": 2.7015, "step": 39268 }, { "epoch": 2.437705630392948, "grad_norm": 0.4290273450270419, "learning_rate": 1.032518876947236e-05, "loss": 2.6818, "step": 39269 }, { "epoch": 2.4377677074927058, "grad_norm": 0.45758572387638224, "learning_rate": 1.0322990989017272e-05, "loss": 2.6593, "step": 39270 }, { "epoch": 2.4378297845924637, "grad_norm": 0.45563059331132044, "learning_rate": 1.0320793415566777e-05, "loss": 2.7102, "step": 39271 }, { "epoch": 2.4378918616922216, "grad_norm": 0.40512350595181684, "learning_rate": 1.0318596049132318e-05, "loss": 2.7021, "step": 39272 }, { "epoch": 2.4379539387919795, "grad_norm": 0.4089370955252859, "learning_rate": 1.0316398889725393e-05, "loss": 2.6992, "step": 39273 }, { "epoch": 2.4380160158917374, "grad_norm": 0.45057753020579766, "learning_rate": 1.0314201937357449e-05, "loss": 2.7639, "step": 39274 }, { "epoch": 2.4380780929914954, "grad_norm": 0.4607274013085568, "learning_rate": 1.031200519203993e-05, "loss": 2.6789, "step": 39275 }, { "epoch": 2.4381401700912533, "grad_norm": 0.4510675945567778, "learning_rate": 1.0309808653784331e-05, "loss": 2.6718, "step": 39276 }, { "epoch": 2.438202247191011, "grad_norm": 0.43570386512040943, "learning_rate": 1.0307612322602095e-05, "loss": 2.7493, "step": 39277 }, { "epoch": 2.438264324290769, "grad_norm": 0.4553087329643214, "learning_rate": 1.030541619850468e-05, "loss": 2.6804, "step": 39278 }, { "epoch": 2.438326401390527, "grad_norm": 0.4688176742022035, "learning_rate": 1.0303220281503529e-05, "loss": 2.7004, "step": 39279 }, { "epoch": 2.438388478490285, "grad_norm": 0.43463489181215637, "learning_rate": 1.030102457161013e-05, "loss": 2.6069, "step": 39280 }, { "epoch": 2.438450555590043, "grad_norm": 0.4719282328661082, "learning_rate": 1.0298829068835924e-05, "loss": 2.6311, "step": 39281 }, { "epoch": 2.438512632689801, "grad_norm": 0.4494090630801567, "learning_rate": 1.0296633773192367e-05, "loss": 2.6913, "step": 39282 }, { "epoch": 2.4385747097895587, "grad_norm": 0.4844660827153847, "learning_rate": 1.0294438684690894e-05, "loss": 2.6061, "step": 39283 }, { "epoch": 2.4386367868893166, "grad_norm": 0.44839966777224316, "learning_rate": 1.0292243803342988e-05, "loss": 2.7031, "step": 39284 }, { "epoch": 2.4386988639890745, "grad_norm": 0.46563371483756694, "learning_rate": 1.0290049129160084e-05, "loss": 2.6749, "step": 39285 }, { "epoch": 2.4387609410888325, "grad_norm": 0.4768863890601527, "learning_rate": 1.0287854662153628e-05, "loss": 2.6869, "step": 39286 }, { "epoch": 2.4388230181885904, "grad_norm": 0.4693173471897235, "learning_rate": 1.0285660402335084e-05, "loss": 2.7318, "step": 39287 }, { "epoch": 2.4388850952883483, "grad_norm": 0.482210244157214, "learning_rate": 1.0283466349715876e-05, "loss": 2.6803, "step": 39288 }, { "epoch": 2.438947172388106, "grad_norm": 0.44722151226411205, "learning_rate": 1.0281272504307477e-05, "loss": 2.6961, "step": 39289 }, { "epoch": 2.4390092494878637, "grad_norm": 0.4618257595445265, "learning_rate": 1.0279078866121317e-05, "loss": 2.6429, "step": 39290 }, { "epoch": 2.439071326587622, "grad_norm": 0.4354378260105753, "learning_rate": 1.0276885435168849e-05, "loss": 2.5838, "step": 39291 }, { "epoch": 2.4391334036873795, "grad_norm": 0.4691445756248805, "learning_rate": 1.02746922114615e-05, "loss": 2.6681, "step": 39292 }, { "epoch": 2.4391954807871374, "grad_norm": 0.45695212893834813, "learning_rate": 1.0272499195010738e-05, "loss": 2.7571, "step": 39293 }, { "epoch": 2.4392575578868954, "grad_norm": 0.406935223987409, "learning_rate": 1.0270306385827994e-05, "loss": 2.7108, "step": 39294 }, { "epoch": 2.4393196349866533, "grad_norm": 0.43476196520585386, "learning_rate": 1.0268113783924705e-05, "loss": 2.6416, "step": 39295 }, { "epoch": 2.439381712086411, "grad_norm": 0.4367681751705586, "learning_rate": 1.02659213893123e-05, "loss": 2.6896, "step": 39296 }, { "epoch": 2.439443789186169, "grad_norm": 0.41073240956899815, "learning_rate": 1.0263729202002237e-05, "loss": 2.7181, "step": 39297 }, { "epoch": 2.439505866285927, "grad_norm": 0.42597178474415964, "learning_rate": 1.0261537222005946e-05, "loss": 2.7318, "step": 39298 }, { "epoch": 2.439567943385685, "grad_norm": 0.44556677401790057, "learning_rate": 1.0259345449334868e-05, "loss": 2.7311, "step": 39299 }, { "epoch": 2.439630020485443, "grad_norm": 0.411235660546476, "learning_rate": 1.0257153884000425e-05, "loss": 2.696, "step": 39300 }, { "epoch": 2.439692097585201, "grad_norm": 0.4355443090525768, "learning_rate": 1.025496252601405e-05, "loss": 2.6898, "step": 39301 }, { "epoch": 2.4397541746849587, "grad_norm": 0.44033986801852215, "learning_rate": 1.0252771375387193e-05, "loss": 2.6851, "step": 39302 }, { "epoch": 2.4398162517847166, "grad_norm": 0.4505746033908538, "learning_rate": 1.0250580432131285e-05, "loss": 2.6793, "step": 39303 }, { "epoch": 2.4398783288844745, "grad_norm": 0.40727209847444484, "learning_rate": 1.0248389696257727e-05, "loss": 2.6374, "step": 39304 }, { "epoch": 2.4399404059842325, "grad_norm": 0.44840807535750526, "learning_rate": 1.0246199167777987e-05, "loss": 2.7522, "step": 39305 }, { "epoch": 2.4400024830839904, "grad_norm": 0.4677888393852945, "learning_rate": 1.0244008846703462e-05, "loss": 2.6982, "step": 39306 }, { "epoch": 2.4400645601837483, "grad_norm": 0.43465571903682043, "learning_rate": 1.0241818733045604e-05, "loss": 2.7058, "step": 39307 }, { "epoch": 2.440126637283506, "grad_norm": 0.46202168410573596, "learning_rate": 1.0239628826815834e-05, "loss": 2.6214, "step": 39308 }, { "epoch": 2.440188714383264, "grad_norm": 0.41045737338990523, "learning_rate": 1.0237439128025572e-05, "loss": 2.6648, "step": 39309 }, { "epoch": 2.440250791483022, "grad_norm": 0.4496045487689204, "learning_rate": 1.0235249636686229e-05, "loss": 2.6397, "step": 39310 }, { "epoch": 2.44031286858278, "grad_norm": 0.4363776183168639, "learning_rate": 1.0233060352809254e-05, "loss": 2.7538, "step": 39311 }, { "epoch": 2.440374945682538, "grad_norm": 0.3918449045333568, "learning_rate": 1.0230871276406056e-05, "loss": 2.6841, "step": 39312 }, { "epoch": 2.440437022782296, "grad_norm": 0.39001948155138194, "learning_rate": 1.022868240748806e-05, "loss": 2.6329, "step": 39313 }, { "epoch": 2.4404990998820537, "grad_norm": 0.46105981270781454, "learning_rate": 1.022649374606668e-05, "loss": 2.6783, "step": 39314 }, { "epoch": 2.440561176981811, "grad_norm": 0.43734242803299533, "learning_rate": 1.0224305292153324e-05, "loss": 2.6144, "step": 39315 }, { "epoch": 2.4406232540815695, "grad_norm": 0.38924291250113396, "learning_rate": 1.0222117045759439e-05, "loss": 2.6403, "step": 39316 }, { "epoch": 2.440685331181327, "grad_norm": 0.41416033902441957, "learning_rate": 1.0219929006896423e-05, "loss": 2.6405, "step": 39317 }, { "epoch": 2.440747408281085, "grad_norm": 0.42814922681883666, "learning_rate": 1.0217741175575696e-05, "loss": 2.7047, "step": 39318 }, { "epoch": 2.440809485380843, "grad_norm": 0.41737005547184935, "learning_rate": 1.021555355180866e-05, "loss": 2.664, "step": 39319 }, { "epoch": 2.4408715624806008, "grad_norm": 0.4185465624541759, "learning_rate": 1.0213366135606751e-05, "loss": 2.6656, "step": 39320 }, { "epoch": 2.4409336395803587, "grad_norm": 0.3953795425197508, "learning_rate": 1.0211178926981363e-05, "loss": 2.6719, "step": 39321 }, { "epoch": 2.4409957166801166, "grad_norm": 0.41750329384675117, "learning_rate": 1.0208991925943923e-05, "loss": 2.6602, "step": 39322 }, { "epoch": 2.4410577937798745, "grad_norm": 0.43382698521944363, "learning_rate": 1.0206805132505825e-05, "loss": 2.6852, "step": 39323 }, { "epoch": 2.4411198708796324, "grad_norm": 0.4049791485516785, "learning_rate": 1.0204618546678474e-05, "loss": 2.6976, "step": 39324 }, { "epoch": 2.4411819479793904, "grad_norm": 0.4461613812596061, "learning_rate": 1.0202432168473303e-05, "loss": 2.6166, "step": 39325 }, { "epoch": 2.4412440250791483, "grad_norm": 0.4295876647103092, "learning_rate": 1.0200245997901703e-05, "loss": 2.6947, "step": 39326 }, { "epoch": 2.441306102178906, "grad_norm": 0.42764192392021727, "learning_rate": 1.0198060034975082e-05, "loss": 2.6799, "step": 39327 }, { "epoch": 2.441368179278664, "grad_norm": 0.4601724223483078, "learning_rate": 1.0195874279704832e-05, "loss": 2.6088, "step": 39328 }, { "epoch": 2.441430256378422, "grad_norm": 0.4488871636002788, "learning_rate": 1.019368873210238e-05, "loss": 2.652, "step": 39329 }, { "epoch": 2.44149233347818, "grad_norm": 0.47689557552320955, "learning_rate": 1.019150339217912e-05, "loss": 2.6742, "step": 39330 }, { "epoch": 2.441554410577938, "grad_norm": 0.45754074975706294, "learning_rate": 1.0189318259946446e-05, "loss": 2.6528, "step": 39331 }, { "epoch": 2.441616487677696, "grad_norm": 0.47086212828308904, "learning_rate": 1.018713333541575e-05, "loss": 2.7222, "step": 39332 }, { "epoch": 2.4416785647774537, "grad_norm": 0.46217301834147123, "learning_rate": 1.018494861859846e-05, "loss": 2.7325, "step": 39333 }, { "epoch": 2.4417406418772116, "grad_norm": 0.427152298577139, "learning_rate": 1.018276410950596e-05, "loss": 2.7496, "step": 39334 }, { "epoch": 2.4418027189769695, "grad_norm": 0.46171610745934066, "learning_rate": 1.0180579808149644e-05, "loss": 2.777, "step": 39335 }, { "epoch": 2.4418647960767275, "grad_norm": 0.44254992025657164, "learning_rate": 1.0178395714540911e-05, "loss": 2.6773, "step": 39336 }, { "epoch": 2.4419268731764854, "grad_norm": 0.42773094350056723, "learning_rate": 1.017621182869114e-05, "loss": 2.6398, "step": 39337 }, { "epoch": 2.441988950276243, "grad_norm": 0.45845010382573825, "learning_rate": 1.0174028150611736e-05, "loss": 2.7797, "step": 39338 }, { "epoch": 2.442051027376001, "grad_norm": 0.4728220183932844, "learning_rate": 1.0171844680314114e-05, "loss": 2.7351, "step": 39339 }, { "epoch": 2.4421131044757587, "grad_norm": 0.47059297657867655, "learning_rate": 1.0169661417809645e-05, "loss": 2.6107, "step": 39340 }, { "epoch": 2.4421751815755166, "grad_norm": 0.46324819926175936, "learning_rate": 1.0167478363109723e-05, "loss": 2.7651, "step": 39341 }, { "epoch": 2.4422372586752745, "grad_norm": 0.43008756220110295, "learning_rate": 1.0165295516225721e-05, "loss": 2.6337, "step": 39342 }, { "epoch": 2.4422993357750324, "grad_norm": 0.4554128929946272, "learning_rate": 1.016311287716905e-05, "loss": 2.7543, "step": 39343 }, { "epoch": 2.4423614128747904, "grad_norm": 0.47650366560675855, "learning_rate": 1.01609304459511e-05, "loss": 2.7343, "step": 39344 }, { "epoch": 2.4424234899745483, "grad_norm": 0.42952975065067467, "learning_rate": 1.0158748222583243e-05, "loss": 2.6379, "step": 39345 }, { "epoch": 2.442485567074306, "grad_norm": 0.4652358573166096, "learning_rate": 1.015656620707685e-05, "loss": 2.6545, "step": 39346 }, { "epoch": 2.442547644174064, "grad_norm": 0.4685809943385969, "learning_rate": 1.0154384399443345e-05, "loss": 2.7638, "step": 39347 }, { "epoch": 2.442609721273822, "grad_norm": 0.4440270660351466, "learning_rate": 1.0152202799694088e-05, "loss": 2.6515, "step": 39348 }, { "epoch": 2.44267179837358, "grad_norm": 0.44574632761239974, "learning_rate": 1.015002140784046e-05, "loss": 2.7077, "step": 39349 }, { "epoch": 2.442733875473338, "grad_norm": 0.46554386084807214, "learning_rate": 1.0147840223893845e-05, "loss": 2.6753, "step": 39350 }, { "epoch": 2.442795952573096, "grad_norm": 0.46184986144168366, "learning_rate": 1.0145659247865608e-05, "loss": 2.697, "step": 39351 }, { "epoch": 2.4428580296728537, "grad_norm": 0.6618769189808578, "learning_rate": 1.0143478479767154e-05, "loss": 2.6924, "step": 39352 }, { "epoch": 2.4429201067726116, "grad_norm": 0.4486759223204126, "learning_rate": 1.014129791960985e-05, "loss": 2.7369, "step": 39353 }, { "epoch": 2.4429821838723695, "grad_norm": 0.476853906932416, "learning_rate": 1.0139117567405071e-05, "loss": 2.6115, "step": 39354 }, { "epoch": 2.4430442609721275, "grad_norm": 0.4840845498517214, "learning_rate": 1.0136937423164178e-05, "loss": 2.6359, "step": 39355 }, { "epoch": 2.4431063380718854, "grad_norm": 0.47181767563906324, "learning_rate": 1.0134757486898572e-05, "loss": 2.6448, "step": 39356 }, { "epoch": 2.4431684151716433, "grad_norm": 0.47288039701358164, "learning_rate": 1.0132577758619615e-05, "loss": 2.5826, "step": 39357 }, { "epoch": 2.443230492271401, "grad_norm": 0.5169443532779774, "learning_rate": 1.0130398238338678e-05, "loss": 2.5959, "step": 39358 }, { "epoch": 2.443292569371159, "grad_norm": 0.4566504693581389, "learning_rate": 1.0128218926067135e-05, "loss": 2.6396, "step": 39359 }, { "epoch": 2.443354646470917, "grad_norm": 0.49278554459257834, "learning_rate": 1.0126039821816335e-05, "loss": 2.7743, "step": 39360 }, { "epoch": 2.443416723570675, "grad_norm": 0.4584476335649784, "learning_rate": 1.0123860925597678e-05, "loss": 2.6275, "step": 39361 }, { "epoch": 2.443478800670433, "grad_norm": 0.4479833616089782, "learning_rate": 1.012168223742252e-05, "loss": 2.6766, "step": 39362 }, { "epoch": 2.4435408777701904, "grad_norm": 0.44975051918127046, "learning_rate": 1.011950375730223e-05, "loss": 2.5864, "step": 39363 }, { "epoch": 2.4436029548699487, "grad_norm": 0.44012983988381643, "learning_rate": 1.0117325485248152e-05, "loss": 2.7139, "step": 39364 }, { "epoch": 2.443665031969706, "grad_norm": 0.4393149942958703, "learning_rate": 1.0115147421271687e-05, "loss": 2.7751, "step": 39365 }, { "epoch": 2.443727109069464, "grad_norm": 0.4366102196370203, "learning_rate": 1.0112969565384179e-05, "loss": 2.7024, "step": 39366 }, { "epoch": 2.443789186169222, "grad_norm": 0.4529260165586822, "learning_rate": 1.011079191759699e-05, "loss": 2.692, "step": 39367 }, { "epoch": 2.44385126326898, "grad_norm": 0.44938772589394094, "learning_rate": 1.010861447792147e-05, "loss": 2.6813, "step": 39368 }, { "epoch": 2.443913340368738, "grad_norm": 0.4469719468174264, "learning_rate": 1.0106437246369004e-05, "loss": 2.6284, "step": 39369 }, { "epoch": 2.443975417468496, "grad_norm": 0.42352741314417985, "learning_rate": 1.0104260222950928e-05, "loss": 2.6821, "step": 39370 }, { "epoch": 2.4440374945682537, "grad_norm": 0.425629912612729, "learning_rate": 1.0102083407678626e-05, "loss": 2.7159, "step": 39371 }, { "epoch": 2.4440995716680116, "grad_norm": 0.40038392596339434, "learning_rate": 1.0099906800563435e-05, "loss": 2.6211, "step": 39372 }, { "epoch": 2.4441616487677695, "grad_norm": 0.43792184859614747, "learning_rate": 1.0097730401616723e-05, "loss": 2.6306, "step": 39373 }, { "epoch": 2.4442237258675275, "grad_norm": 0.4513961102649761, "learning_rate": 1.0095554210849822e-05, "loss": 2.7257, "step": 39374 }, { "epoch": 2.4442858029672854, "grad_norm": 0.41787445929968997, "learning_rate": 1.0093378228274114e-05, "loss": 2.6254, "step": 39375 }, { "epoch": 2.4443478800670433, "grad_norm": 0.43707406222666084, "learning_rate": 1.0091202453900944e-05, "loss": 2.707, "step": 39376 }, { "epoch": 2.444409957166801, "grad_norm": 0.44173977497207706, "learning_rate": 1.0089026887741655e-05, "loss": 2.6207, "step": 39377 }, { "epoch": 2.444472034266559, "grad_norm": 0.46894211011346854, "learning_rate": 1.008685152980759e-05, "loss": 2.7921, "step": 39378 }, { "epoch": 2.444534111366317, "grad_norm": 0.4592832363366289, "learning_rate": 1.0084676380110119e-05, "loss": 2.6501, "step": 39379 }, { "epoch": 2.444596188466075, "grad_norm": 0.506606709823796, "learning_rate": 1.0082501438660586e-05, "loss": 2.7522, "step": 39380 }, { "epoch": 2.444658265565833, "grad_norm": 0.45613786808762297, "learning_rate": 1.0080326705470333e-05, "loss": 2.6744, "step": 39381 }, { "epoch": 2.444720342665591, "grad_norm": 0.46112822773135437, "learning_rate": 1.0078152180550687e-05, "loss": 2.6469, "step": 39382 }, { "epoch": 2.4447824197653487, "grad_norm": 0.46531621179579746, "learning_rate": 1.0075977863913033e-05, "loss": 2.6813, "step": 39383 }, { "epoch": 2.4448444968651066, "grad_norm": 0.4473836651421478, "learning_rate": 1.007380375556869e-05, "loss": 2.6316, "step": 39384 }, { "epoch": 2.4449065739648645, "grad_norm": 0.4499135325062941, "learning_rate": 1.0071629855529007e-05, "loss": 2.6273, "step": 39385 }, { "epoch": 2.444968651064622, "grad_norm": 0.4658600732240589, "learning_rate": 1.0069456163805324e-05, "loss": 2.6878, "step": 39386 }, { "epoch": 2.4450307281643804, "grad_norm": 0.4651037498833931, "learning_rate": 1.0067282680408968e-05, "loss": 2.6058, "step": 39387 }, { "epoch": 2.445092805264138, "grad_norm": 0.4829312311246708, "learning_rate": 1.006510940535131e-05, "loss": 2.7234, "step": 39388 }, { "epoch": 2.4451548823638958, "grad_norm": 0.5190261737621877, "learning_rate": 1.0062936338643664e-05, "loss": 2.6891, "step": 39389 }, { "epoch": 2.4452169594636537, "grad_norm": 0.4634759960018382, "learning_rate": 1.0060763480297375e-05, "loss": 2.6586, "step": 39390 }, { "epoch": 2.4452790365634116, "grad_norm": 0.4471692143773002, "learning_rate": 1.0058590830323767e-05, "loss": 2.7395, "step": 39391 }, { "epoch": 2.4453411136631695, "grad_norm": 0.49089276605934407, "learning_rate": 1.00564183887342e-05, "loss": 2.6628, "step": 39392 }, { "epoch": 2.4454031907629274, "grad_norm": 0.45459320362131916, "learning_rate": 1.0054246155539998e-05, "loss": 2.7333, "step": 39393 }, { "epoch": 2.4454652678626854, "grad_norm": 0.4505065702558329, "learning_rate": 1.0052074130752487e-05, "loss": 2.6867, "step": 39394 }, { "epoch": 2.4455273449624433, "grad_norm": 0.4693506048849842, "learning_rate": 1.0049902314383003e-05, "loss": 2.7218, "step": 39395 }, { "epoch": 2.445589422062201, "grad_norm": 0.42662046129532316, "learning_rate": 1.0047730706442865e-05, "loss": 2.6318, "step": 39396 }, { "epoch": 2.445651499161959, "grad_norm": 0.4565418590185376, "learning_rate": 1.0045559306943426e-05, "loss": 2.7204, "step": 39397 }, { "epoch": 2.445713576261717, "grad_norm": 0.44272946886959386, "learning_rate": 1.0043388115896008e-05, "loss": 2.6441, "step": 39398 }, { "epoch": 2.445775653361475, "grad_norm": 0.47719016718290796, "learning_rate": 1.0041217133311925e-05, "loss": 2.7037, "step": 39399 }, { "epoch": 2.445837730461233, "grad_norm": 0.47547695156599284, "learning_rate": 1.0039046359202508e-05, "loss": 2.6728, "step": 39400 }, { "epoch": 2.445899807560991, "grad_norm": 0.44500636570374286, "learning_rate": 1.0036875793579099e-05, "loss": 2.5442, "step": 39401 }, { "epoch": 2.4459618846607487, "grad_norm": 0.48649931196613017, "learning_rate": 1.0034705436453007e-05, "loss": 2.6615, "step": 39402 }, { "epoch": 2.4460239617605066, "grad_norm": 0.47573975380524747, "learning_rate": 1.0032535287835548e-05, "loss": 2.6866, "step": 39403 }, { "epoch": 2.4460860388602645, "grad_norm": 0.4036376111413449, "learning_rate": 1.0030365347738062e-05, "loss": 2.6629, "step": 39404 }, { "epoch": 2.4461481159600225, "grad_norm": 0.4268831640722297, "learning_rate": 1.0028195616171859e-05, "loss": 2.6395, "step": 39405 }, { "epoch": 2.4462101930597804, "grad_norm": 0.486604037781154, "learning_rate": 1.0026026093148272e-05, "loss": 2.6199, "step": 39406 }, { "epoch": 2.4462722701595383, "grad_norm": 0.45264198604675265, "learning_rate": 1.0023856778678608e-05, "loss": 2.6388, "step": 39407 }, { "epoch": 2.446334347259296, "grad_norm": 0.464052087172492, "learning_rate": 1.0021687672774194e-05, "loss": 2.6284, "step": 39408 }, { "epoch": 2.446396424359054, "grad_norm": 0.4895780661144044, "learning_rate": 1.0019518775446334e-05, "loss": 2.6428, "step": 39409 }, { "epoch": 2.446458501458812, "grad_norm": 0.4884306564632628, "learning_rate": 1.001735008670634e-05, "loss": 2.6497, "step": 39410 }, { "epoch": 2.4465205785585695, "grad_norm": 0.44021415088574833, "learning_rate": 1.0015181606565555e-05, "loss": 2.7002, "step": 39411 }, { "epoch": 2.446582655658328, "grad_norm": 0.4600384559831413, "learning_rate": 1.0013013335035266e-05, "loss": 2.5977, "step": 39412 }, { "epoch": 2.4466447327580854, "grad_norm": 0.4781747499151469, "learning_rate": 1.0010845272126796e-05, "loss": 2.7182, "step": 39413 }, { "epoch": 2.4467068098578433, "grad_norm": 0.4463943590951164, "learning_rate": 1.0008677417851443e-05, "loss": 2.6788, "step": 39414 }, { "epoch": 2.446768886957601, "grad_norm": 0.44614922400709206, "learning_rate": 1.000650977222054e-05, "loss": 2.756, "step": 39415 }, { "epoch": 2.446830964057359, "grad_norm": 0.4101107799212236, "learning_rate": 1.0004342335245381e-05, "loss": 2.7158, "step": 39416 }, { "epoch": 2.446893041157117, "grad_norm": 0.4668845888350643, "learning_rate": 1.0002175106937283e-05, "loss": 2.6559, "step": 39417 }, { "epoch": 2.446955118256875, "grad_norm": 0.4524602947413323, "learning_rate": 1.0000008087307532e-05, "loss": 2.686, "step": 39418 }, { "epoch": 2.447017195356633, "grad_norm": 0.38664296881565957, "learning_rate": 9.997841276367464e-06, "loss": 2.6673, "step": 39419 }, { "epoch": 2.447079272456391, "grad_norm": 0.42748997294506746, "learning_rate": 9.995674674128363e-06, "loss": 2.654, "step": 39420 }, { "epoch": 2.4471413495561487, "grad_norm": 0.45938325147938186, "learning_rate": 9.993508280601543e-06, "loss": 2.7166, "step": 39421 }, { "epoch": 2.4472034266559066, "grad_norm": 0.4753559817728543, "learning_rate": 9.9913420957983e-06, "loss": 2.6527, "step": 39422 }, { "epoch": 2.4472655037556645, "grad_norm": 0.39461138827290476, "learning_rate": 9.989176119729932e-06, "loss": 2.592, "step": 39423 }, { "epoch": 2.4473275808554225, "grad_norm": 0.42617487429954154, "learning_rate": 9.98701035240775e-06, "loss": 2.559, "step": 39424 }, { "epoch": 2.4473896579551804, "grad_norm": 0.4613431018849858, "learning_rate": 9.984844793843051e-06, "loss": 2.6602, "step": 39425 }, { "epoch": 2.4474517350549383, "grad_norm": 0.48611119039474326, "learning_rate": 9.982679444047132e-06, "loss": 2.6742, "step": 39426 }, { "epoch": 2.447513812154696, "grad_norm": 0.4335312497569241, "learning_rate": 9.980514303031274e-06, "loss": 2.601, "step": 39427 }, { "epoch": 2.447575889254454, "grad_norm": 0.48080229618001047, "learning_rate": 9.9783493708068e-06, "loss": 2.6073, "step": 39428 }, { "epoch": 2.447637966354212, "grad_norm": 0.47961163247026595, "learning_rate": 9.976184647384995e-06, "loss": 2.643, "step": 39429 }, { "epoch": 2.44770004345397, "grad_norm": 0.47329602409569155, "learning_rate": 9.97402013277715e-06, "loss": 2.7087, "step": 39430 }, { "epoch": 2.447762120553728, "grad_norm": 0.4648809177702081, "learning_rate": 9.971855826994547e-06, "loss": 2.6348, "step": 39431 }, { "epoch": 2.447824197653486, "grad_norm": 0.5129004658367252, "learning_rate": 9.969691730048503e-06, "loss": 2.685, "step": 39432 }, { "epoch": 2.4478862747532437, "grad_norm": 0.47864613213043794, "learning_rate": 9.967527841950292e-06, "loss": 2.6719, "step": 39433 }, { "epoch": 2.447948351853001, "grad_norm": 0.4673456772662809, "learning_rate": 9.965364162711205e-06, "loss": 2.6656, "step": 39434 }, { "epoch": 2.4480104289527596, "grad_norm": 0.48319494414747616, "learning_rate": 9.963200692342523e-06, "loss": 2.6383, "step": 39435 }, { "epoch": 2.448072506052517, "grad_norm": 0.4696996764421313, "learning_rate": 9.961037430855553e-06, "loss": 2.7092, "step": 39436 }, { "epoch": 2.448134583152275, "grad_norm": 0.47724879205330056, "learning_rate": 9.958874378261557e-06, "loss": 2.7023, "step": 39437 }, { "epoch": 2.448196660252033, "grad_norm": 0.46825762956761025, "learning_rate": 9.95671153457185e-06, "loss": 2.6833, "step": 39438 }, { "epoch": 2.448258737351791, "grad_norm": 0.46157374356056524, "learning_rate": 9.9545488997977e-06, "loss": 2.5609, "step": 39439 }, { "epoch": 2.4483208144515487, "grad_norm": 0.5304853321568037, "learning_rate": 9.952386473950387e-06, "loss": 2.7529, "step": 39440 }, { "epoch": 2.4483828915513066, "grad_norm": 0.44210372104187595, "learning_rate": 9.95022425704118e-06, "loss": 2.6325, "step": 39441 }, { "epoch": 2.4484449686510645, "grad_norm": 0.4550127956666188, "learning_rate": 9.948062249081392e-06, "loss": 2.6369, "step": 39442 }, { "epoch": 2.4485070457508225, "grad_norm": 0.4833233707596058, "learning_rate": 9.945900450082286e-06, "loss": 2.6611, "step": 39443 }, { "epoch": 2.4485691228505804, "grad_norm": 0.4935096813796312, "learning_rate": 9.94373886005514e-06, "loss": 2.7181, "step": 39444 }, { "epoch": 2.4486311999503383, "grad_norm": 0.4454606796812768, "learning_rate": 9.941577479011233e-06, "loss": 2.658, "step": 39445 }, { "epoch": 2.448693277050096, "grad_norm": 0.4819143768777044, "learning_rate": 9.93941630696183e-06, "loss": 2.5934, "step": 39446 }, { "epoch": 2.448755354149854, "grad_norm": 0.4541728794332077, "learning_rate": 9.937255343918229e-06, "loss": 2.6331, "step": 39447 }, { "epoch": 2.448817431249612, "grad_norm": 0.47111657689905506, "learning_rate": 9.93509458989169e-06, "loss": 2.6139, "step": 39448 }, { "epoch": 2.44887950834937, "grad_norm": 0.4635126947785805, "learning_rate": 9.932934044893489e-06, "loss": 2.6072, "step": 39449 }, { "epoch": 2.448941585449128, "grad_norm": 0.44064956555978685, "learning_rate": 9.930773708934887e-06, "loss": 2.678, "step": 39450 }, { "epoch": 2.449003662548886, "grad_norm": 0.4739907711233172, "learning_rate": 9.92861358202718e-06, "loss": 2.6732, "step": 39451 }, { "epoch": 2.4490657396486437, "grad_norm": 0.4698598438648907, "learning_rate": 9.926453664181618e-06, "loss": 2.635, "step": 39452 }, { "epoch": 2.4491278167484016, "grad_norm": 0.48453900421738105, "learning_rate": 9.92429395540948e-06, "loss": 2.6988, "step": 39453 }, { "epoch": 2.4491898938481595, "grad_norm": 0.46211931696795494, "learning_rate": 9.922134455722016e-06, "loss": 2.6709, "step": 39454 }, { "epoch": 2.4492519709479175, "grad_norm": 0.48652268160436246, "learning_rate": 9.919975165130513e-06, "loss": 2.6951, "step": 39455 }, { "epoch": 2.4493140480476754, "grad_norm": 0.4509381601298385, "learning_rate": 9.917816083646237e-06, "loss": 2.7743, "step": 39456 }, { "epoch": 2.4493761251474333, "grad_norm": 0.4753366087016637, "learning_rate": 9.915657211280439e-06, "loss": 2.6437, "step": 39457 }, { "epoch": 2.449438202247191, "grad_norm": 0.44919055352659965, "learning_rate": 9.913498548044392e-06, "loss": 2.6208, "step": 39458 }, { "epoch": 2.4495002793469487, "grad_norm": 0.4316344627527376, "learning_rate": 9.91134009394934e-06, "loss": 2.7341, "step": 39459 }, { "epoch": 2.449562356446707, "grad_norm": 0.4443798336941151, "learning_rate": 9.909181849006571e-06, "loss": 2.7021, "step": 39460 }, { "epoch": 2.4496244335464645, "grad_norm": 0.43378573111088353, "learning_rate": 9.907023813227334e-06, "loss": 2.7064, "step": 39461 }, { "epoch": 2.4496865106462224, "grad_norm": 0.44683689074977784, "learning_rate": 9.904865986622886e-06, "loss": 2.6752, "step": 39462 }, { "epoch": 2.4497485877459804, "grad_norm": 0.45044989018120807, "learning_rate": 9.902708369204471e-06, "loss": 2.6449, "step": 39463 }, { "epoch": 2.4498106648457383, "grad_norm": 0.46332368833236515, "learning_rate": 9.90055096098338e-06, "loss": 2.6889, "step": 39464 }, { "epoch": 2.449872741945496, "grad_norm": 0.4175316316362899, "learning_rate": 9.898393761970842e-06, "loss": 2.7417, "step": 39465 }, { "epoch": 2.449934819045254, "grad_norm": 0.4418095122266132, "learning_rate": 9.896236772178119e-06, "loss": 2.7004, "step": 39466 }, { "epoch": 2.449996896145012, "grad_norm": 0.44259512145628466, "learning_rate": 9.894079991616457e-06, "loss": 2.6738, "step": 39467 }, { "epoch": 2.45005897324477, "grad_norm": 0.45172136871412755, "learning_rate": 9.891923420297112e-06, "loss": 2.703, "step": 39468 }, { "epoch": 2.450121050344528, "grad_norm": 0.500095537118234, "learning_rate": 9.889767058231348e-06, "loss": 2.65, "step": 39469 }, { "epoch": 2.450183127444286, "grad_norm": 0.4535805286947083, "learning_rate": 9.887610905430411e-06, "loss": 2.7634, "step": 39470 }, { "epoch": 2.4502452045440437, "grad_norm": 0.4071055902697897, "learning_rate": 9.885454961905544e-06, "loss": 2.6794, "step": 39471 }, { "epoch": 2.4503072816438016, "grad_norm": 0.4702979278841631, "learning_rate": 9.883299227667997e-06, "loss": 2.7112, "step": 39472 }, { "epoch": 2.4503693587435595, "grad_norm": 0.4715736538507686, "learning_rate": 9.881143702729006e-06, "loss": 2.6412, "step": 39473 }, { "epoch": 2.4504314358433175, "grad_norm": 0.4821361668692742, "learning_rate": 9.87898838709984e-06, "loss": 2.6441, "step": 39474 }, { "epoch": 2.4504935129430754, "grad_norm": 0.4765582318974714, "learning_rate": 9.87683328079173e-06, "loss": 2.6985, "step": 39475 }, { "epoch": 2.4505555900428333, "grad_norm": 0.4500931232842397, "learning_rate": 9.874678383815921e-06, "loss": 2.7346, "step": 39476 }, { "epoch": 2.450617667142591, "grad_norm": 0.47184208824632634, "learning_rate": 9.872523696183639e-06, "loss": 2.7176, "step": 39477 }, { "epoch": 2.450679744242349, "grad_norm": 0.4889053476141809, "learning_rate": 9.870369217906161e-06, "loss": 2.72, "step": 39478 }, { "epoch": 2.450741821342107, "grad_norm": 0.5084391908435807, "learning_rate": 9.86821494899471e-06, "loss": 2.6256, "step": 39479 }, { "epoch": 2.450803898441865, "grad_norm": 0.4683398677781855, "learning_rate": 9.866060889460515e-06, "loss": 2.7504, "step": 39480 }, { "epoch": 2.450865975541623, "grad_norm": 0.4592498822998577, "learning_rate": 9.863907039314819e-06, "loss": 2.7576, "step": 39481 }, { "epoch": 2.4509280526413804, "grad_norm": 0.4864362420314936, "learning_rate": 9.861753398568874e-06, "loss": 2.7598, "step": 39482 }, { "epoch": 2.4509901297411387, "grad_norm": 0.5018925611718057, "learning_rate": 9.8595999672339e-06, "loss": 2.6303, "step": 39483 }, { "epoch": 2.451052206840896, "grad_norm": 0.48017132865450396, "learning_rate": 9.857446745321142e-06, "loss": 2.6796, "step": 39484 }, { "epoch": 2.451114283940654, "grad_norm": 0.44049969981548115, "learning_rate": 9.85529373284183e-06, "loss": 2.7259, "step": 39485 }, { "epoch": 2.451176361040412, "grad_norm": 0.532255063474982, "learning_rate": 9.85314092980718e-06, "loss": 2.6795, "step": 39486 }, { "epoch": 2.45123843814017, "grad_norm": 0.49674448002463417, "learning_rate": 9.850988336228456e-06, "loss": 2.7077, "step": 39487 }, { "epoch": 2.451300515239928, "grad_norm": 0.5068124460645339, "learning_rate": 9.84883595211687e-06, "loss": 2.7512, "step": 39488 }, { "epoch": 2.451362592339686, "grad_norm": 0.47767290919249555, "learning_rate": 9.846683777483657e-06, "loss": 2.7366, "step": 39489 }, { "epoch": 2.4514246694394437, "grad_norm": 0.44273165660082164, "learning_rate": 9.844531812340025e-06, "loss": 2.6654, "step": 39490 }, { "epoch": 2.4514867465392016, "grad_norm": 0.4872881441232291, "learning_rate": 9.842380056697237e-06, "loss": 2.7116, "step": 39491 }, { "epoch": 2.4515488236389595, "grad_norm": 0.4975723848687114, "learning_rate": 9.840228510566496e-06, "loss": 2.6831, "step": 39492 }, { "epoch": 2.4516109007387175, "grad_norm": 0.4275845652868386, "learning_rate": 9.838077173959031e-06, "loss": 2.6731, "step": 39493 }, { "epoch": 2.4516729778384754, "grad_norm": 0.42982050831123586, "learning_rate": 9.835926046886074e-06, "loss": 2.7545, "step": 39494 }, { "epoch": 2.4517350549382333, "grad_norm": 0.466859581857432, "learning_rate": 9.833775129358825e-06, "loss": 2.6553, "step": 39495 }, { "epoch": 2.451797132037991, "grad_norm": 0.4814927578774607, "learning_rate": 9.83162442138853e-06, "loss": 2.6034, "step": 39496 }, { "epoch": 2.451859209137749, "grad_norm": 0.4378123530755896, "learning_rate": 9.829473922986404e-06, "loss": 2.704, "step": 39497 }, { "epoch": 2.451921286237507, "grad_norm": 0.40895781043639967, "learning_rate": 9.827323634163665e-06, "loss": 2.6185, "step": 39498 }, { "epoch": 2.451983363337265, "grad_norm": 0.47396186331248363, "learning_rate": 9.825173554931521e-06, "loss": 2.619, "step": 39499 }, { "epoch": 2.452045440437023, "grad_norm": 0.44718127725511514, "learning_rate": 9.823023685301208e-06, "loss": 2.6844, "step": 39500 }, { "epoch": 2.452107517536781, "grad_norm": 0.43072291380815275, "learning_rate": 9.820874025283922e-06, "loss": 2.6002, "step": 39501 }, { "epoch": 2.4521695946365387, "grad_norm": 0.4434093161532116, "learning_rate": 9.818724574890903e-06, "loss": 2.69, "step": 39502 }, { "epoch": 2.4522316717362966, "grad_norm": 0.4210661268640035, "learning_rate": 9.816575334133354e-06, "loss": 2.6675, "step": 39503 }, { "epoch": 2.4522937488360546, "grad_norm": 0.4257341586328138, "learning_rate": 9.814426303022472e-06, "loss": 2.7107, "step": 39504 }, { "epoch": 2.4523558259358125, "grad_norm": 0.4454224768575766, "learning_rate": 9.8122774815695e-06, "loss": 2.6538, "step": 39505 }, { "epoch": 2.4524179030355704, "grad_norm": 0.4398331418968749, "learning_rate": 9.810128869785629e-06, "loss": 2.6643, "step": 39506 }, { "epoch": 2.452479980135328, "grad_norm": 0.4044084625313318, "learning_rate": 9.807980467682071e-06, "loss": 2.6809, "step": 39507 }, { "epoch": 2.4525420572350862, "grad_norm": 0.42387333641974484, "learning_rate": 9.805832275270032e-06, "loss": 2.6301, "step": 39508 }, { "epoch": 2.4526041343348437, "grad_norm": 0.43597281801727816, "learning_rate": 9.803684292560716e-06, "loss": 2.6185, "step": 39509 }, { "epoch": 2.4526662114346016, "grad_norm": 0.4107708096111536, "learning_rate": 9.801536519565352e-06, "loss": 2.649, "step": 39510 }, { "epoch": 2.4527282885343595, "grad_norm": 0.41836610961394527, "learning_rate": 9.799388956295124e-06, "loss": 2.7094, "step": 39511 }, { "epoch": 2.4527903656341175, "grad_norm": 0.3932233224733435, "learning_rate": 9.797241602761248e-06, "loss": 2.6827, "step": 39512 }, { "epoch": 2.4528524427338754, "grad_norm": 0.4368461593636804, "learning_rate": 9.795094458974907e-06, "loss": 2.6334, "step": 39513 }, { "epoch": 2.4529145198336333, "grad_norm": 0.43823811731914414, "learning_rate": 9.792947524947332e-06, "loss": 2.7197, "step": 39514 }, { "epoch": 2.452976596933391, "grad_norm": 0.4218573036720874, "learning_rate": 9.790800800689709e-06, "loss": 2.622, "step": 39515 }, { "epoch": 2.453038674033149, "grad_norm": 0.4368847881088355, "learning_rate": 9.788654286213245e-06, "loss": 2.6935, "step": 39516 }, { "epoch": 2.453100751132907, "grad_norm": 0.4603511345804303, "learning_rate": 9.786507981529114e-06, "loss": 2.7391, "step": 39517 }, { "epoch": 2.453162828232665, "grad_norm": 0.4021731079597237, "learning_rate": 9.784361886648547e-06, "loss": 2.6033, "step": 39518 }, { "epoch": 2.453224905332423, "grad_norm": 0.4456956954724654, "learning_rate": 9.782216001582728e-06, "loss": 2.7319, "step": 39519 }, { "epoch": 2.453286982432181, "grad_norm": 0.43014273370416245, "learning_rate": 9.78007032634285e-06, "loss": 2.6671, "step": 39520 }, { "epoch": 2.4533490595319387, "grad_norm": 0.4704781148808929, "learning_rate": 9.777924860940113e-06, "loss": 2.6259, "step": 39521 }, { "epoch": 2.4534111366316966, "grad_norm": 0.4615054107257909, "learning_rate": 9.775779605385687e-06, "loss": 2.6897, "step": 39522 }, { "epoch": 2.4534732137314545, "grad_norm": 0.4112738143263812, "learning_rate": 9.773634559690803e-06, "loss": 2.5846, "step": 39523 }, { "epoch": 2.4535352908312125, "grad_norm": 0.4061155280056254, "learning_rate": 9.771489723866628e-06, "loss": 2.5571, "step": 39524 }, { "epoch": 2.4535973679309704, "grad_norm": 0.4358099754333682, "learning_rate": 9.769345097924359e-06, "loss": 2.6224, "step": 39525 }, { "epoch": 2.4536594450307283, "grad_norm": 0.43659675545668586, "learning_rate": 9.767200681875167e-06, "loss": 2.7066, "step": 39526 }, { "epoch": 2.453721522130486, "grad_norm": 0.4345584482009313, "learning_rate": 9.765056475730272e-06, "loss": 2.7555, "step": 39527 }, { "epoch": 2.453783599230244, "grad_norm": 0.41474807411836623, "learning_rate": 9.76291247950084e-06, "loss": 2.612, "step": 39528 }, { "epoch": 2.453845676330002, "grad_norm": 0.4283193655216657, "learning_rate": 9.76076869319807e-06, "loss": 2.7064, "step": 39529 }, { "epoch": 2.4539077534297595, "grad_norm": 0.44743222461291576, "learning_rate": 9.758625116833132e-06, "loss": 2.5946, "step": 39530 }, { "epoch": 2.453969830529518, "grad_norm": 0.4390748663684607, "learning_rate": 9.756481750417207e-06, "loss": 2.6356, "step": 39531 }, { "epoch": 2.4540319076292754, "grad_norm": 0.44338222898489543, "learning_rate": 9.754338593961493e-06, "loss": 2.6528, "step": 39532 }, { "epoch": 2.4540939847290333, "grad_norm": 0.42527665852609053, "learning_rate": 9.75219564747717e-06, "loss": 2.7172, "step": 39533 }, { "epoch": 2.454156061828791, "grad_norm": 0.4436394567211244, "learning_rate": 9.750052910975399e-06, "loss": 2.6743, "step": 39534 }, { "epoch": 2.454218138928549, "grad_norm": 0.4322065380644735, "learning_rate": 9.747910384467385e-06, "loss": 2.7218, "step": 39535 }, { "epoch": 2.454280216028307, "grad_norm": 0.43597944871343414, "learning_rate": 9.745768067964279e-06, "loss": 2.7065, "step": 39536 }, { "epoch": 2.454342293128065, "grad_norm": 0.45398069024088183, "learning_rate": 9.743625961477288e-06, "loss": 2.5921, "step": 39537 }, { "epoch": 2.454404370227823, "grad_norm": 0.38387157921740434, "learning_rate": 9.741484065017569e-06, "loss": 2.6357, "step": 39538 }, { "epoch": 2.454466447327581, "grad_norm": 0.4375110011761204, "learning_rate": 9.739342378596306e-06, "loss": 2.6276, "step": 39539 }, { "epoch": 2.4545285244273387, "grad_norm": 0.4638494227169845, "learning_rate": 9.737200902224652e-06, "loss": 2.6299, "step": 39540 }, { "epoch": 2.4545906015270966, "grad_norm": 0.440956865962754, "learning_rate": 9.735059635913813e-06, "loss": 2.7683, "step": 39541 }, { "epoch": 2.4546526786268545, "grad_norm": 0.446734133439177, "learning_rate": 9.732918579674938e-06, "loss": 2.6906, "step": 39542 }, { "epoch": 2.4547147557266125, "grad_norm": 0.4461632014790661, "learning_rate": 9.730777733519203e-06, "loss": 2.7437, "step": 39543 }, { "epoch": 2.4547768328263704, "grad_norm": 0.43302789485821136, "learning_rate": 9.728637097457771e-06, "loss": 2.6039, "step": 39544 }, { "epoch": 2.4548389099261283, "grad_norm": 0.40353402670956123, "learning_rate": 9.726496671501811e-06, "loss": 2.656, "step": 39545 }, { "epoch": 2.454900987025886, "grad_norm": 0.4893921840812549, "learning_rate": 9.724356455662503e-06, "loss": 2.7659, "step": 39546 }, { "epoch": 2.454963064125644, "grad_norm": 0.4546366010931538, "learning_rate": 9.722216449951005e-06, "loss": 2.6789, "step": 39547 }, { "epoch": 2.455025141225402, "grad_norm": 0.42034245210588583, "learning_rate": 9.72007665437848e-06, "loss": 2.6946, "step": 39548 }, { "epoch": 2.45508721832516, "grad_norm": 0.4712302078320343, "learning_rate": 9.717937068956085e-06, "loss": 2.627, "step": 39549 }, { "epoch": 2.455149295424918, "grad_norm": 0.4332367742074893, "learning_rate": 9.715797693695001e-06, "loss": 2.5733, "step": 39550 }, { "epoch": 2.455211372524676, "grad_norm": 0.42390435935139115, "learning_rate": 9.71365852860638e-06, "loss": 2.693, "step": 39551 }, { "epoch": 2.4552734496244337, "grad_norm": 0.4646356421332797, "learning_rate": 9.711519573701384e-06, "loss": 2.6999, "step": 39552 }, { "epoch": 2.455335526724191, "grad_norm": 0.4642422735673377, "learning_rate": 9.709380828991155e-06, "loss": 2.6155, "step": 39553 }, { "epoch": 2.4553976038239496, "grad_norm": 0.4506071445350758, "learning_rate": 9.707242294486878e-06, "loss": 2.6921, "step": 39554 }, { "epoch": 2.455459680923707, "grad_norm": 0.43211581970054647, "learning_rate": 9.705103970199703e-06, "loss": 2.6248, "step": 39555 }, { "epoch": 2.455521758023465, "grad_norm": 0.4670822675705842, "learning_rate": 9.702965856140784e-06, "loss": 2.6142, "step": 39556 }, { "epoch": 2.455583835123223, "grad_norm": 0.43880221162184774, "learning_rate": 9.700827952321273e-06, "loss": 2.6223, "step": 39557 }, { "epoch": 2.455645912222981, "grad_norm": 0.46072953592081933, "learning_rate": 9.698690258752313e-06, "loss": 2.6702, "step": 39558 }, { "epoch": 2.4557079893227387, "grad_norm": 0.4668582093872117, "learning_rate": 9.696552775445084e-06, "loss": 2.6967, "step": 39559 }, { "epoch": 2.4557700664224966, "grad_norm": 0.4134433543799222, "learning_rate": 9.694415502410719e-06, "loss": 2.5493, "step": 39560 }, { "epoch": 2.4558321435222545, "grad_norm": 0.4260543857096688, "learning_rate": 9.692278439660375e-06, "loss": 2.6768, "step": 39561 }, { "epoch": 2.4558942206220125, "grad_norm": 0.45107162484750574, "learning_rate": 9.690141587205187e-06, "loss": 2.7397, "step": 39562 }, { "epoch": 2.4559562977217704, "grad_norm": 0.4419521905364201, "learning_rate": 9.688004945056328e-06, "loss": 2.6875, "step": 39563 }, { "epoch": 2.4560183748215283, "grad_norm": 0.47174138204711363, "learning_rate": 9.685868513224932e-06, "loss": 2.6796, "step": 39564 }, { "epoch": 2.456080451921286, "grad_norm": 0.44376566415997915, "learning_rate": 9.683732291722148e-06, "loss": 2.6547, "step": 39565 }, { "epoch": 2.456142529021044, "grad_norm": 0.4449294266298506, "learning_rate": 9.681596280559108e-06, "loss": 2.7024, "step": 39566 }, { "epoch": 2.456204606120802, "grad_norm": 0.41451019699521224, "learning_rate": 9.679460479746966e-06, "loss": 2.7091, "step": 39567 }, { "epoch": 2.45626668322056, "grad_norm": 0.43846544961079387, "learning_rate": 9.67732488929688e-06, "loss": 2.6324, "step": 39568 }, { "epoch": 2.456328760320318, "grad_norm": 0.4679311434707772, "learning_rate": 9.675189509219979e-06, "loss": 2.6902, "step": 39569 }, { "epoch": 2.456390837420076, "grad_norm": 0.4482738342416365, "learning_rate": 9.673054339527404e-06, "loss": 2.6769, "step": 39570 }, { "epoch": 2.4564529145198337, "grad_norm": 0.4395926529527674, "learning_rate": 9.670919380230293e-06, "loss": 2.6724, "step": 39571 }, { "epoch": 2.4565149916195916, "grad_norm": 0.4377196986474854, "learning_rate": 9.668784631339772e-06, "loss": 2.6898, "step": 39572 }, { "epoch": 2.4565770687193496, "grad_norm": 0.44842817981352007, "learning_rate": 9.666650092867008e-06, "loss": 2.6529, "step": 39573 }, { "epoch": 2.4566391458191075, "grad_norm": 0.42493889102856797, "learning_rate": 9.664515764823117e-06, "loss": 2.6106, "step": 39574 }, { "epoch": 2.4567012229188654, "grad_norm": 0.44141641115629265, "learning_rate": 9.662381647219243e-06, "loss": 2.6693, "step": 39575 }, { "epoch": 2.4567633000186233, "grad_norm": 0.4553436028847664, "learning_rate": 9.660247740066508e-06, "loss": 2.5916, "step": 39576 }, { "epoch": 2.4568253771183812, "grad_norm": 0.44679912243067893, "learning_rate": 9.658114043376059e-06, "loss": 2.6286, "step": 39577 }, { "epoch": 2.4568874542181387, "grad_norm": 0.4785664122741534, "learning_rate": 9.655980557159022e-06, "loss": 2.759, "step": 39578 }, { "epoch": 2.456949531317897, "grad_norm": 0.4299200988987943, "learning_rate": 9.653847281426532e-06, "loss": 2.6902, "step": 39579 }, { "epoch": 2.4570116084176545, "grad_norm": 0.4649747295332294, "learning_rate": 9.651714216189716e-06, "loss": 2.7775, "step": 39580 }, { "epoch": 2.4570736855174125, "grad_norm": 0.42509645110892125, "learning_rate": 9.649581361459687e-06, "loss": 2.6407, "step": 39581 }, { "epoch": 2.4571357626171704, "grad_norm": 0.42198202753552894, "learning_rate": 9.647448717247598e-06, "loss": 2.6179, "step": 39582 }, { "epoch": 2.4571978397169283, "grad_norm": 0.43722506808220923, "learning_rate": 9.64531628356457e-06, "loss": 2.75, "step": 39583 }, { "epoch": 2.457259916816686, "grad_norm": 0.4471465372855156, "learning_rate": 9.643184060421717e-06, "loss": 2.6254, "step": 39584 }, { "epoch": 2.457321993916444, "grad_norm": 0.45735790935840737, "learning_rate": 9.641052047830162e-06, "loss": 2.7099, "step": 39585 }, { "epoch": 2.457384071016202, "grad_norm": 0.49891205979909076, "learning_rate": 9.638920245801048e-06, "loss": 2.6353, "step": 39586 }, { "epoch": 2.45744614811596, "grad_norm": 0.4386842231677018, "learning_rate": 9.636788654345486e-06, "loss": 2.6937, "step": 39587 }, { "epoch": 2.457508225215718, "grad_norm": 0.4474909019344525, "learning_rate": 9.634657273474596e-06, "loss": 2.67, "step": 39588 }, { "epoch": 2.457570302315476, "grad_norm": 0.43211838876457576, "learning_rate": 9.63252610319948e-06, "loss": 2.6375, "step": 39589 }, { "epoch": 2.4576323794152337, "grad_norm": 0.464637213591854, "learning_rate": 9.63039514353129e-06, "loss": 2.6542, "step": 39590 }, { "epoch": 2.4576944565149916, "grad_norm": 0.4562338858854966, "learning_rate": 9.628264394481135e-06, "loss": 2.6157, "step": 39591 }, { "epoch": 2.4577565336147496, "grad_norm": 0.41792584136610755, "learning_rate": 9.626133856060116e-06, "loss": 2.67, "step": 39592 }, { "epoch": 2.4578186107145075, "grad_norm": 0.4350156238369976, "learning_rate": 9.624003528279363e-06, "loss": 2.6181, "step": 39593 }, { "epoch": 2.4578806878142654, "grad_norm": 0.477317805607514, "learning_rate": 9.621873411149973e-06, "loss": 2.7112, "step": 39594 }, { "epoch": 2.4579427649140233, "grad_norm": 0.47964736074697556, "learning_rate": 9.619743504683082e-06, "loss": 2.6924, "step": 39595 }, { "epoch": 2.458004842013781, "grad_norm": 0.41207889245278967, "learning_rate": 9.617613808889791e-06, "loss": 2.5931, "step": 39596 }, { "epoch": 2.458066919113539, "grad_norm": 0.4974155987978249, "learning_rate": 9.615484323781215e-06, "loss": 2.6455, "step": 39597 }, { "epoch": 2.458128996213297, "grad_norm": 0.48792975935348704, "learning_rate": 9.613355049368445e-06, "loss": 2.7126, "step": 39598 }, { "epoch": 2.458191073313055, "grad_norm": 0.43386126333464514, "learning_rate": 9.611225985662609e-06, "loss": 2.6532, "step": 39599 }, { "epoch": 2.458253150412813, "grad_norm": 0.4825868671916714, "learning_rate": 9.609097132674822e-06, "loss": 2.6323, "step": 39600 }, { "epoch": 2.4583152275125704, "grad_norm": 0.4693040593275113, "learning_rate": 9.60696849041618e-06, "loss": 2.6614, "step": 39601 }, { "epoch": 2.4583773046123287, "grad_norm": 0.49484992102631703, "learning_rate": 9.60484005889779e-06, "loss": 2.7447, "step": 39602 }, { "epoch": 2.458439381712086, "grad_norm": 0.43210625023368104, "learning_rate": 9.602711838130745e-06, "loss": 2.6744, "step": 39603 }, { "epoch": 2.458501458811844, "grad_norm": 0.4613422782464658, "learning_rate": 9.60058382812617e-06, "loss": 2.6322, "step": 39604 }, { "epoch": 2.458563535911602, "grad_norm": 0.4518102118238542, "learning_rate": 9.598456028895158e-06, "loss": 2.6764, "step": 39605 }, { "epoch": 2.45862561301136, "grad_norm": 0.4303028036062567, "learning_rate": 9.59632844044881e-06, "loss": 2.6012, "step": 39606 }, { "epoch": 2.458687690111118, "grad_norm": 0.414401617274111, "learning_rate": 9.59420106279822e-06, "loss": 2.6747, "step": 39607 }, { "epoch": 2.458749767210876, "grad_norm": 0.4213695854160511, "learning_rate": 9.592073895954484e-06, "loss": 2.6583, "step": 39608 }, { "epoch": 2.4588118443106337, "grad_norm": 0.4889864292641009, "learning_rate": 9.589946939928718e-06, "loss": 2.6618, "step": 39609 }, { "epoch": 2.4588739214103916, "grad_norm": 0.4597624170131468, "learning_rate": 9.587820194732005e-06, "loss": 2.6311, "step": 39610 }, { "epoch": 2.4589359985101495, "grad_norm": 0.4332191980267852, "learning_rate": 9.58569366037545e-06, "loss": 2.5962, "step": 39611 }, { "epoch": 2.4589980756099075, "grad_norm": 0.43253859587542315, "learning_rate": 9.58356733687013e-06, "loss": 2.6524, "step": 39612 }, { "epoch": 2.4590601527096654, "grad_norm": 0.44488590285962626, "learning_rate": 9.581441224227161e-06, "loss": 2.7206, "step": 39613 }, { "epoch": 2.4591222298094233, "grad_norm": 0.46531265714807973, "learning_rate": 9.57931532245762e-06, "loss": 2.7642, "step": 39614 }, { "epoch": 2.459184306909181, "grad_norm": 0.43438456133741205, "learning_rate": 9.577189631572614e-06, "loss": 2.6887, "step": 39615 }, { "epoch": 2.459246384008939, "grad_norm": 0.45428169533452595, "learning_rate": 9.575064151583213e-06, "loss": 2.6648, "step": 39616 }, { "epoch": 2.459308461108697, "grad_norm": 0.4330539952325405, "learning_rate": 9.572938882500509e-06, "loss": 2.6641, "step": 39617 }, { "epoch": 2.459370538208455, "grad_norm": 0.4395079684473624, "learning_rate": 9.570813824335606e-06, "loss": 2.6614, "step": 39618 }, { "epoch": 2.459432615308213, "grad_norm": 0.44561815647151043, "learning_rate": 9.568688977099589e-06, "loss": 2.6359, "step": 39619 }, { "epoch": 2.459494692407971, "grad_norm": 0.4364247391851126, "learning_rate": 9.56656434080353e-06, "loss": 2.7137, "step": 39620 }, { "epoch": 2.4595567695077287, "grad_norm": 0.44084770175014176, "learning_rate": 9.564439915458506e-06, "loss": 2.6867, "step": 39621 }, { "epoch": 2.4596188466074866, "grad_norm": 0.4619284500558921, "learning_rate": 9.56231570107563e-06, "loss": 2.6696, "step": 39622 }, { "epoch": 2.4596809237072446, "grad_norm": 0.449349823778355, "learning_rate": 9.56019169766597e-06, "loss": 2.6953, "step": 39623 }, { "epoch": 2.4597430008070025, "grad_norm": 0.43982717913077657, "learning_rate": 9.558067905240608e-06, "loss": 2.6186, "step": 39624 }, { "epoch": 2.4598050779067604, "grad_norm": 0.4880623288529836, "learning_rate": 9.555944323810612e-06, "loss": 2.6915, "step": 39625 }, { "epoch": 2.459867155006518, "grad_norm": 0.4456850969311042, "learning_rate": 9.553820953387083e-06, "loss": 2.599, "step": 39626 }, { "epoch": 2.4599292321062762, "grad_norm": 0.4659921581054629, "learning_rate": 9.55169779398109e-06, "loss": 2.6735, "step": 39627 }, { "epoch": 2.4599913092060337, "grad_norm": 0.42255482822941803, "learning_rate": 9.54957484560371e-06, "loss": 2.6441, "step": 39628 }, { "epoch": 2.4600533863057916, "grad_norm": 0.4556666649238449, "learning_rate": 9.547452108266013e-06, "loss": 2.6814, "step": 39629 }, { "epoch": 2.4601154634055495, "grad_norm": 0.4187697107031704, "learning_rate": 9.545329581979068e-06, "loss": 2.6142, "step": 39630 }, { "epoch": 2.4601775405053075, "grad_norm": 0.4506205042430929, "learning_rate": 9.543207266753973e-06, "loss": 2.7827, "step": 39631 }, { "epoch": 2.4602396176050654, "grad_norm": 0.4646722012216, "learning_rate": 9.541085162601776e-06, "loss": 2.6934, "step": 39632 }, { "epoch": 2.4603016947048233, "grad_norm": 0.47622740799591806, "learning_rate": 9.538963269533568e-06, "loss": 2.6233, "step": 39633 }, { "epoch": 2.460363771804581, "grad_norm": 0.4677545945865995, "learning_rate": 9.536841587560413e-06, "loss": 2.7159, "step": 39634 }, { "epoch": 2.460425848904339, "grad_norm": 0.45144664769818155, "learning_rate": 9.534720116693363e-06, "loss": 2.6748, "step": 39635 }, { "epoch": 2.460487926004097, "grad_norm": 0.4312308079139284, "learning_rate": 9.53259885694352e-06, "loss": 2.723, "step": 39636 }, { "epoch": 2.460550003103855, "grad_norm": 0.4682715517350183, "learning_rate": 9.530477808321925e-06, "loss": 2.6225, "step": 39637 }, { "epoch": 2.460612080203613, "grad_norm": 0.45132937303885257, "learning_rate": 9.528356970839658e-06, "loss": 2.6318, "step": 39638 }, { "epoch": 2.460674157303371, "grad_norm": 0.41081849177146496, "learning_rate": 9.52623634450776e-06, "loss": 2.6316, "step": 39639 }, { "epoch": 2.4607362344031287, "grad_norm": 0.46395754285498664, "learning_rate": 9.52411592933733e-06, "loss": 2.6498, "step": 39640 }, { "epoch": 2.4607983115028866, "grad_norm": 0.4619532562467278, "learning_rate": 9.521995725339411e-06, "loss": 2.7198, "step": 39641 }, { "epoch": 2.4608603886026446, "grad_norm": 0.4592887680706059, "learning_rate": 9.519875732525064e-06, "loss": 2.7121, "step": 39642 }, { "epoch": 2.4609224657024025, "grad_norm": 0.4519681962648207, "learning_rate": 9.517755950905356e-06, "loss": 2.6672, "step": 39643 }, { "epoch": 2.4609845428021604, "grad_norm": 0.4224382278892137, "learning_rate": 9.515636380491328e-06, "loss": 2.6777, "step": 39644 }, { "epoch": 2.4610466199019183, "grad_norm": 0.4455847660750572, "learning_rate": 9.513517021294066e-06, "loss": 2.6086, "step": 39645 }, { "epoch": 2.4611086970016762, "grad_norm": 0.4415118673442579, "learning_rate": 9.511397873324613e-06, "loss": 2.7512, "step": 39646 }, { "epoch": 2.461170774101434, "grad_norm": 0.44038933951890835, "learning_rate": 9.509278936594029e-06, "loss": 2.635, "step": 39647 }, { "epoch": 2.461232851201192, "grad_norm": 0.4636789200849328, "learning_rate": 9.507160211113353e-06, "loss": 2.6672, "step": 39648 }, { "epoch": 2.4612949283009495, "grad_norm": 0.494847542134225, "learning_rate": 9.505041696893662e-06, "loss": 2.6925, "step": 39649 }, { "epoch": 2.461357005400708, "grad_norm": 0.45556128724913225, "learning_rate": 9.502923393946001e-06, "loss": 2.7218, "step": 39650 }, { "epoch": 2.4614190825004654, "grad_norm": 0.4730884688067678, "learning_rate": 9.500805302281418e-06, "loss": 2.7645, "step": 39651 }, { "epoch": 2.4614811596002233, "grad_norm": 0.46606349395445434, "learning_rate": 9.498687421910952e-06, "loss": 2.6951, "step": 39652 }, { "epoch": 2.461543236699981, "grad_norm": 0.47649437777053405, "learning_rate": 9.49656975284568e-06, "loss": 2.684, "step": 39653 }, { "epoch": 2.461605313799739, "grad_norm": 0.47757494582957843, "learning_rate": 9.49445229509664e-06, "loss": 2.6733, "step": 39654 }, { "epoch": 2.461667390899497, "grad_norm": 0.5209805344150392, "learning_rate": 9.492335048674872e-06, "loss": 2.6495, "step": 39655 }, { "epoch": 2.461729467999255, "grad_norm": 0.5252158572717126, "learning_rate": 9.490218013591423e-06, "loss": 2.6832, "step": 39656 }, { "epoch": 2.461791545099013, "grad_norm": 0.43242875380588497, "learning_rate": 9.488101189857328e-06, "loss": 2.7348, "step": 39657 }, { "epoch": 2.461853622198771, "grad_norm": 0.48219144004947556, "learning_rate": 9.485984577483665e-06, "loss": 2.734, "step": 39658 }, { "epoch": 2.4619156992985287, "grad_norm": 0.5157306158386634, "learning_rate": 9.483868176481447e-06, "loss": 2.5559, "step": 39659 }, { "epoch": 2.4619777763982866, "grad_norm": 0.47695526759910617, "learning_rate": 9.481751986861726e-06, "loss": 2.6802, "step": 39660 }, { "epoch": 2.4620398534980446, "grad_norm": 0.493764138321114, "learning_rate": 9.479636008635528e-06, "loss": 2.7563, "step": 39661 }, { "epoch": 2.4621019305978025, "grad_norm": 0.4701350571231908, "learning_rate": 9.47752024181392e-06, "loss": 2.5958, "step": 39662 }, { "epoch": 2.4621640076975604, "grad_norm": 0.4507172345700179, "learning_rate": 9.47540468640793e-06, "loss": 2.6724, "step": 39663 }, { "epoch": 2.4622260847973183, "grad_norm": 0.4473389311439982, "learning_rate": 9.473289342428581e-06, "loss": 2.6498, "step": 39664 }, { "epoch": 2.462288161897076, "grad_norm": 0.49034887853571796, "learning_rate": 9.471174209886913e-06, "loss": 2.6401, "step": 39665 }, { "epoch": 2.462350238996834, "grad_norm": 0.4650230062084852, "learning_rate": 9.469059288793986e-06, "loss": 2.6844, "step": 39666 }, { "epoch": 2.462412316096592, "grad_norm": 0.4532187299141179, "learning_rate": 9.466944579160796e-06, "loss": 2.666, "step": 39667 }, { "epoch": 2.46247439319635, "grad_norm": 0.45453419739156736, "learning_rate": 9.46483008099841e-06, "loss": 2.6412, "step": 39668 }, { "epoch": 2.462536470296108, "grad_norm": 0.4851294763521215, "learning_rate": 9.462715794317844e-06, "loss": 2.5859, "step": 39669 }, { "epoch": 2.462598547395866, "grad_norm": 0.4822067058889232, "learning_rate": 9.460601719130135e-06, "loss": 2.6463, "step": 39670 }, { "epoch": 2.4626606244956237, "grad_norm": 0.4465992390153366, "learning_rate": 9.45848785544629e-06, "loss": 2.6804, "step": 39671 }, { "epoch": 2.4627227015953816, "grad_norm": 0.4713592862717038, "learning_rate": 9.456374203277374e-06, "loss": 2.6701, "step": 39672 }, { "epoch": 2.4627847786951396, "grad_norm": 0.4352938853855112, "learning_rate": 9.454260762634392e-06, "loss": 2.7202, "step": 39673 }, { "epoch": 2.462846855794897, "grad_norm": 0.46485436050743123, "learning_rate": 9.452147533528377e-06, "loss": 2.5727, "step": 39674 }, { "epoch": 2.4629089328946554, "grad_norm": 0.49820664778898555, "learning_rate": 9.450034515970336e-06, "loss": 2.6063, "step": 39675 }, { "epoch": 2.462971009994413, "grad_norm": 0.4325969166021471, "learning_rate": 9.447921709971325e-06, "loss": 2.6889, "step": 39676 }, { "epoch": 2.463033087094171, "grad_norm": 0.46545919179946776, "learning_rate": 9.445809115542348e-06, "loss": 2.7417, "step": 39677 }, { "epoch": 2.4630951641939287, "grad_norm": 0.4781804682152664, "learning_rate": 9.44369673269443e-06, "loss": 2.6262, "step": 39678 }, { "epoch": 2.4631572412936866, "grad_norm": 0.4232903157865473, "learning_rate": 9.441584561438593e-06, "loss": 2.6866, "step": 39679 }, { "epoch": 2.4632193183934445, "grad_norm": 0.4212936705300586, "learning_rate": 9.439472601785843e-06, "loss": 2.6459, "step": 39680 }, { "epoch": 2.4632813954932025, "grad_norm": 0.44524249145585254, "learning_rate": 9.437360853747224e-06, "loss": 2.6957, "step": 39681 }, { "epoch": 2.4633434725929604, "grad_norm": 0.4388824657162646, "learning_rate": 9.43524931733374e-06, "loss": 2.7145, "step": 39682 }, { "epoch": 2.4634055496927183, "grad_norm": 0.42903465614075403, "learning_rate": 9.433137992556402e-06, "loss": 2.7113, "step": 39683 }, { "epoch": 2.463467626792476, "grad_norm": 0.44112333160024353, "learning_rate": 9.431026879426224e-06, "loss": 2.6967, "step": 39684 }, { "epoch": 2.463529703892234, "grad_norm": 0.49882014962006554, "learning_rate": 9.428915977954239e-06, "loss": 2.7666, "step": 39685 }, { "epoch": 2.463591780991992, "grad_norm": 0.4232241814642326, "learning_rate": 9.426805288151447e-06, "loss": 2.7724, "step": 39686 }, { "epoch": 2.46365385809175, "grad_norm": 0.4393953921224033, "learning_rate": 9.424694810028861e-06, "loss": 2.6974, "step": 39687 }, { "epoch": 2.463715935191508, "grad_norm": 0.44129601511510014, "learning_rate": 9.422584543597485e-06, "loss": 2.6788, "step": 39688 }, { "epoch": 2.463778012291266, "grad_norm": 0.47600997100031645, "learning_rate": 9.420474488868341e-06, "loss": 2.6872, "step": 39689 }, { "epoch": 2.4638400893910237, "grad_norm": 0.4410384004183941, "learning_rate": 9.41836464585244e-06, "loss": 2.6028, "step": 39690 }, { "epoch": 2.4639021664907816, "grad_norm": 0.43690010222435766, "learning_rate": 9.416255014560776e-06, "loss": 2.6014, "step": 39691 }, { "epoch": 2.4639642435905396, "grad_norm": 0.4372472714294831, "learning_rate": 9.414145595004364e-06, "loss": 2.5734, "step": 39692 }, { "epoch": 2.4640263206902975, "grad_norm": 0.432693905195727, "learning_rate": 9.412036387194196e-06, "loss": 2.548, "step": 39693 }, { "epoch": 2.4640883977900554, "grad_norm": 0.45657857664238743, "learning_rate": 9.409927391141294e-06, "loss": 2.7276, "step": 39694 }, { "epoch": 2.4641504748898133, "grad_norm": 0.4801861622037155, "learning_rate": 9.407818606856661e-06, "loss": 2.6896, "step": 39695 }, { "epoch": 2.4642125519895712, "grad_norm": 0.4240838962293085, "learning_rate": 9.40571003435129e-06, "loss": 2.61, "step": 39696 }, { "epoch": 2.4642746290893287, "grad_norm": 0.4543786956762747, "learning_rate": 9.40360167363617e-06, "loss": 2.6855, "step": 39697 }, { "epoch": 2.464336706189087, "grad_norm": 0.4526363072398598, "learning_rate": 9.401493524722315e-06, "loss": 2.6283, "step": 39698 }, { "epoch": 2.4643987832888445, "grad_norm": 0.4669691908984326, "learning_rate": 9.399385587620741e-06, "loss": 2.7098, "step": 39699 }, { "epoch": 2.4644608603886025, "grad_norm": 0.4310927041992687, "learning_rate": 9.397277862342423e-06, "loss": 2.6512, "step": 39700 }, { "epoch": 2.4645229374883604, "grad_norm": 0.43623313382070994, "learning_rate": 9.395170348898369e-06, "loss": 2.6366, "step": 39701 }, { "epoch": 2.4645850145881183, "grad_norm": 0.42341495186430267, "learning_rate": 9.393063047299555e-06, "loss": 2.6453, "step": 39702 }, { "epoch": 2.464647091687876, "grad_norm": 0.43803594323524897, "learning_rate": 9.390955957556997e-06, "loss": 2.753, "step": 39703 }, { "epoch": 2.464709168787634, "grad_norm": 0.44660302602274066, "learning_rate": 9.388849079681683e-06, "loss": 2.6478, "step": 39704 }, { "epoch": 2.464771245887392, "grad_norm": 0.4465207389647884, "learning_rate": 9.386742413684602e-06, "loss": 2.6619, "step": 39705 }, { "epoch": 2.46483332298715, "grad_norm": 0.4199685536544571, "learning_rate": 9.384635959576743e-06, "loss": 2.6875, "step": 39706 }, { "epoch": 2.464895400086908, "grad_norm": 0.49099660030827325, "learning_rate": 9.382529717369087e-06, "loss": 2.6634, "step": 39707 }, { "epoch": 2.464957477186666, "grad_norm": 0.41876545849065877, "learning_rate": 9.380423687072642e-06, "loss": 2.6652, "step": 39708 }, { "epoch": 2.4650195542864237, "grad_norm": 0.39162098748409835, "learning_rate": 9.378317868698394e-06, "loss": 2.5263, "step": 39709 }, { "epoch": 2.4650816313861816, "grad_norm": 0.45094999001297104, "learning_rate": 9.376212262257323e-06, "loss": 2.681, "step": 39710 }, { "epoch": 2.4651437084859396, "grad_norm": 0.4848267079492169, "learning_rate": 9.374106867760396e-06, "loss": 2.623, "step": 39711 }, { "epoch": 2.4652057855856975, "grad_norm": 0.41308674913691873, "learning_rate": 9.372001685218635e-06, "loss": 2.6826, "step": 39712 }, { "epoch": 2.4652678626854554, "grad_norm": 0.438549282101483, "learning_rate": 9.369896714642996e-06, "loss": 2.6883, "step": 39713 }, { "epoch": 2.4653299397852133, "grad_norm": 0.48020062292146776, "learning_rate": 9.367791956044476e-06, "loss": 2.6984, "step": 39714 }, { "epoch": 2.4653920168849712, "grad_norm": 0.4456756683638612, "learning_rate": 9.365687409434043e-06, "loss": 2.6935, "step": 39715 }, { "epoch": 2.465454093984729, "grad_norm": 0.42238704195606513, "learning_rate": 9.363583074822674e-06, "loss": 2.6996, "step": 39716 }, { "epoch": 2.465516171084487, "grad_norm": 0.44685218059190857, "learning_rate": 9.36147895222137e-06, "loss": 2.6466, "step": 39717 }, { "epoch": 2.465578248184245, "grad_norm": 0.48139482895063207, "learning_rate": 9.359375041641099e-06, "loss": 2.7005, "step": 39718 }, { "epoch": 2.465640325284003, "grad_norm": 0.42355149554012345, "learning_rate": 9.357271343092826e-06, "loss": 2.6525, "step": 39719 }, { "epoch": 2.465702402383761, "grad_norm": 0.42418594022716727, "learning_rate": 9.355167856587526e-06, "loss": 2.6676, "step": 39720 }, { "epoch": 2.4657644794835187, "grad_norm": 0.4243908499677826, "learning_rate": 9.3530645821362e-06, "loss": 2.688, "step": 39721 }, { "epoch": 2.465826556583276, "grad_norm": 0.4394682987560469, "learning_rate": 9.350961519749795e-06, "loss": 2.7552, "step": 39722 }, { "epoch": 2.4658886336830346, "grad_norm": 0.39493951352477435, "learning_rate": 9.348858669439298e-06, "loss": 2.6801, "step": 39723 }, { "epoch": 2.465950710782792, "grad_norm": 0.4205895903818347, "learning_rate": 9.346756031215659e-06, "loss": 2.5773, "step": 39724 }, { "epoch": 2.46601278788255, "grad_norm": 0.4001147366416746, "learning_rate": 9.344653605089875e-06, "loss": 2.6018, "step": 39725 }, { "epoch": 2.466074864982308, "grad_norm": 0.442334153486616, "learning_rate": 9.342551391072901e-06, "loss": 2.7701, "step": 39726 }, { "epoch": 2.466136942082066, "grad_norm": 0.4144808617028551, "learning_rate": 9.34044938917571e-06, "loss": 2.7211, "step": 39727 }, { "epoch": 2.4661990191818237, "grad_norm": 0.3966309976301618, "learning_rate": 9.338347599409264e-06, "loss": 2.5887, "step": 39728 }, { "epoch": 2.4662610962815816, "grad_norm": 0.4324824551788799, "learning_rate": 9.336246021784516e-06, "loss": 2.7556, "step": 39729 }, { "epoch": 2.4663231733813396, "grad_norm": 0.44813990205064247, "learning_rate": 9.334144656312444e-06, "loss": 2.732, "step": 39730 }, { "epoch": 2.4663852504810975, "grad_norm": 0.45124837713927207, "learning_rate": 9.332043503004023e-06, "loss": 2.6967, "step": 39731 }, { "epoch": 2.4664473275808554, "grad_norm": 0.46738419235669826, "learning_rate": 9.329942561870203e-06, "loss": 2.7193, "step": 39732 }, { "epoch": 2.4665094046806133, "grad_norm": 0.4269479288830296, "learning_rate": 9.327841832921947e-06, "loss": 2.6863, "step": 39733 }, { "epoch": 2.4665714817803712, "grad_norm": 0.43685045730022865, "learning_rate": 9.3257413161702e-06, "loss": 2.6265, "step": 39734 }, { "epoch": 2.466633558880129, "grad_norm": 0.4570614731405568, "learning_rate": 9.32364101162595e-06, "loss": 2.5732, "step": 39735 }, { "epoch": 2.466695635979887, "grad_norm": 0.48879829497402866, "learning_rate": 9.321540919300136e-06, "loss": 2.6792, "step": 39736 }, { "epoch": 2.466757713079645, "grad_norm": 0.43527606704850175, "learning_rate": 9.319441039203719e-06, "loss": 2.6384, "step": 39737 }, { "epoch": 2.466819790179403, "grad_norm": 0.454478658975613, "learning_rate": 9.317341371347643e-06, "loss": 2.7361, "step": 39738 }, { "epoch": 2.466881867279161, "grad_norm": 0.4370421500239538, "learning_rate": 9.315241915742883e-06, "loss": 2.6971, "step": 39739 }, { "epoch": 2.4669439443789187, "grad_norm": 0.42484158761251767, "learning_rate": 9.313142672400389e-06, "loss": 2.6108, "step": 39740 }, { "epoch": 2.4670060214786766, "grad_norm": 0.44391837149494934, "learning_rate": 9.311043641331096e-06, "loss": 2.5838, "step": 39741 }, { "epoch": 2.4670680985784346, "grad_norm": 0.4564828272082797, "learning_rate": 9.308944822545974e-06, "loss": 2.6091, "step": 39742 }, { "epoch": 2.4671301756781925, "grad_norm": 0.4255955142256241, "learning_rate": 9.306846216055948e-06, "loss": 2.7039, "step": 39743 }, { "epoch": 2.4671922527779504, "grad_norm": 0.45270732835054506, "learning_rate": 9.304747821872e-06, "loss": 2.664, "step": 39744 }, { "epoch": 2.467254329877708, "grad_norm": 0.46607207846061194, "learning_rate": 9.302649640005063e-06, "loss": 2.6896, "step": 39745 }, { "epoch": 2.4673164069774662, "grad_norm": 0.47051342825826015, "learning_rate": 9.30055167046608e-06, "loss": 2.5595, "step": 39746 }, { "epoch": 2.4673784840772237, "grad_norm": 0.45386842054611115, "learning_rate": 9.298453913265981e-06, "loss": 2.7071, "step": 39747 }, { "epoch": 2.4674405611769816, "grad_norm": 0.40635463394155125, "learning_rate": 9.296356368415748e-06, "loss": 2.6777, "step": 39748 }, { "epoch": 2.4675026382767395, "grad_norm": 0.4482244018483462, "learning_rate": 9.294259035926306e-06, "loss": 2.7111, "step": 39749 }, { "epoch": 2.4675647153764975, "grad_norm": 0.476600410954654, "learning_rate": 9.29216191580859e-06, "loss": 2.7029, "step": 39750 }, { "epoch": 2.4676267924762554, "grad_norm": 0.43169230156096716, "learning_rate": 9.29006500807355e-06, "loss": 2.7202, "step": 39751 }, { "epoch": 2.4676888695760133, "grad_norm": 0.48347646680739886, "learning_rate": 9.287968312732115e-06, "loss": 2.6848, "step": 39752 }, { "epoch": 2.467750946675771, "grad_norm": 0.4498248945062764, "learning_rate": 9.28587182979524e-06, "loss": 2.6018, "step": 39753 }, { "epoch": 2.467813023775529, "grad_norm": 0.43539716944626483, "learning_rate": 9.28377555927386e-06, "loss": 2.666, "step": 39754 }, { "epoch": 2.467875100875287, "grad_norm": 0.44964793062318215, "learning_rate": 9.281679501178902e-06, "loss": 2.6159, "step": 39755 }, { "epoch": 2.467937177975045, "grad_norm": 0.4858140028019085, "learning_rate": 9.279583655521295e-06, "loss": 2.6595, "step": 39756 }, { "epoch": 2.467999255074803, "grad_norm": 0.43687895564306345, "learning_rate": 9.277488022312003e-06, "loss": 2.7358, "step": 39757 }, { "epoch": 2.468061332174561, "grad_norm": 0.45318335124547215, "learning_rate": 9.27539260156194e-06, "loss": 2.6298, "step": 39758 }, { "epoch": 2.4681234092743187, "grad_norm": 0.49183059043543864, "learning_rate": 9.273297393282037e-06, "loss": 2.6733, "step": 39759 }, { "epoch": 2.4681854863740766, "grad_norm": 0.4313588668406266, "learning_rate": 9.271202397483215e-06, "loss": 2.646, "step": 39760 }, { "epoch": 2.4682475634738346, "grad_norm": 0.4373882184959773, "learning_rate": 9.26910761417643e-06, "loss": 2.7193, "step": 39761 }, { "epoch": 2.4683096405735925, "grad_norm": 0.4161442481700258, "learning_rate": 9.267013043372602e-06, "loss": 2.6857, "step": 39762 }, { "epoch": 2.4683717176733504, "grad_norm": 0.45603877495199885, "learning_rate": 9.26491868508264e-06, "loss": 2.6153, "step": 39763 }, { "epoch": 2.4684337947731083, "grad_norm": 0.45808336818179785, "learning_rate": 9.2628245393175e-06, "loss": 2.6661, "step": 39764 }, { "epoch": 2.4684958718728662, "grad_norm": 0.44378537453434513, "learning_rate": 9.260730606088098e-06, "loss": 2.7536, "step": 39765 }, { "epoch": 2.468557948972624, "grad_norm": 0.44406255339502043, "learning_rate": 9.258636885405337e-06, "loss": 2.6968, "step": 39766 }, { "epoch": 2.468620026072382, "grad_norm": 0.49514970540963676, "learning_rate": 9.256543377280168e-06, "loss": 2.6925, "step": 39767 }, { "epoch": 2.46868210317214, "grad_norm": 0.4523299411121975, "learning_rate": 9.254450081723503e-06, "loss": 2.682, "step": 39768 }, { "epoch": 2.468744180271898, "grad_norm": 0.43433202146694083, "learning_rate": 9.252356998746265e-06, "loss": 2.7584, "step": 39769 }, { "epoch": 2.4688062573716554, "grad_norm": 0.4507794014877949, "learning_rate": 9.25026412835936e-06, "loss": 2.6531, "step": 39770 }, { "epoch": 2.4688683344714137, "grad_norm": 0.4442181273840782, "learning_rate": 9.248171470573736e-06, "loss": 2.7138, "step": 39771 }, { "epoch": 2.468930411571171, "grad_norm": 0.48112594813914744, "learning_rate": 9.246079025400289e-06, "loss": 2.5927, "step": 39772 }, { "epoch": 2.468992488670929, "grad_norm": 0.46689200928227736, "learning_rate": 9.24398679284994e-06, "loss": 2.7357, "step": 39773 }, { "epoch": 2.469054565770687, "grad_norm": 0.4130918714309042, "learning_rate": 9.241894772933596e-06, "loss": 2.6242, "step": 39774 }, { "epoch": 2.469116642870445, "grad_norm": 0.42079901350725235, "learning_rate": 9.239802965662191e-06, "loss": 2.6981, "step": 39775 }, { "epoch": 2.469178719970203, "grad_norm": 0.4600983918477993, "learning_rate": 9.237711371046636e-06, "loss": 2.6491, "step": 39776 }, { "epoch": 2.469240797069961, "grad_norm": 0.4380269113301149, "learning_rate": 9.235619989097832e-06, "loss": 2.6394, "step": 39777 }, { "epoch": 2.4693028741697187, "grad_norm": 0.41569352553076605, "learning_rate": 9.233528819826692e-06, "loss": 2.6765, "step": 39778 }, { "epoch": 2.4693649512694766, "grad_norm": 0.40107673846712577, "learning_rate": 9.231437863244118e-06, "loss": 2.6304, "step": 39779 }, { "epoch": 2.4694270283692346, "grad_norm": 0.43174768216317744, "learning_rate": 9.229347119361042e-06, "loss": 2.6331, "step": 39780 }, { "epoch": 2.4694891054689925, "grad_norm": 0.4551306108979082, "learning_rate": 9.22725658818836e-06, "loss": 2.6583, "step": 39781 }, { "epoch": 2.4695511825687504, "grad_norm": 0.4381073074468898, "learning_rate": 9.22516626973698e-06, "loss": 2.6871, "step": 39782 }, { "epoch": 2.4696132596685083, "grad_norm": 0.42806356643630805, "learning_rate": 9.22307616401779e-06, "loss": 2.6988, "step": 39783 }, { "epoch": 2.4696753367682662, "grad_norm": 0.44391291676784966, "learning_rate": 9.220986271041726e-06, "loss": 2.6202, "step": 39784 }, { "epoch": 2.469737413868024, "grad_norm": 0.47172287137131863, "learning_rate": 9.218896590819675e-06, "loss": 2.6996, "step": 39785 }, { "epoch": 2.469799490967782, "grad_norm": 0.44690001424391207, "learning_rate": 9.216807123362536e-06, "loss": 2.6679, "step": 39786 }, { "epoch": 2.46986156806754, "grad_norm": 0.4112799555585166, "learning_rate": 9.214717868681205e-06, "loss": 2.7195, "step": 39787 }, { "epoch": 2.469923645167298, "grad_norm": 0.4722355633453027, "learning_rate": 9.212628826786607e-06, "loss": 2.6125, "step": 39788 }, { "epoch": 2.469985722267056, "grad_norm": 0.4185020929954175, "learning_rate": 9.210539997689627e-06, "loss": 2.7334, "step": 39789 }, { "epoch": 2.4700477993668137, "grad_norm": 0.47375187901146903, "learning_rate": 9.208451381401156e-06, "loss": 2.6273, "step": 39790 }, { "epoch": 2.4701098764665717, "grad_norm": 0.4258833771461983, "learning_rate": 9.206362977932104e-06, "loss": 2.6759, "step": 39791 }, { "epoch": 2.4701719535663296, "grad_norm": 0.436495957160309, "learning_rate": 9.204274787293343e-06, "loss": 2.6793, "step": 39792 }, { "epoch": 2.470234030666087, "grad_norm": 0.4787535095584209, "learning_rate": 9.202186809495794e-06, "loss": 2.7233, "step": 39793 }, { "epoch": 2.4702961077658454, "grad_norm": 0.41682716695720684, "learning_rate": 9.200099044550342e-06, "loss": 2.5946, "step": 39794 }, { "epoch": 2.470358184865603, "grad_norm": 0.47037298250842796, "learning_rate": 9.198011492467879e-06, "loss": 2.6594, "step": 39795 }, { "epoch": 2.470420261965361, "grad_norm": 0.4165227959520697, "learning_rate": 9.195924153259284e-06, "loss": 2.7631, "step": 39796 }, { "epoch": 2.4704823390651187, "grad_norm": 0.42640444443803727, "learning_rate": 9.193837026935454e-06, "loss": 2.6622, "step": 39797 }, { "epoch": 2.4705444161648766, "grad_norm": 0.4444080336212853, "learning_rate": 9.191750113507298e-06, "loss": 2.599, "step": 39798 }, { "epoch": 2.4706064932646346, "grad_norm": 0.47202864665450306, "learning_rate": 9.189663412985689e-06, "loss": 2.6362, "step": 39799 }, { "epoch": 2.4706685703643925, "grad_norm": 0.4289500757836523, "learning_rate": 9.187576925381509e-06, "loss": 2.6788, "step": 39800 }, { "epoch": 2.4707306474641504, "grad_norm": 0.4556870704463639, "learning_rate": 9.185490650705653e-06, "loss": 2.7121, "step": 39801 }, { "epoch": 2.4707927245639083, "grad_norm": 0.47314262586589867, "learning_rate": 9.18340458896898e-06, "loss": 2.6857, "step": 39802 }, { "epoch": 2.4708548016636662, "grad_norm": 0.4410425392500232, "learning_rate": 9.181318740182415e-06, "loss": 2.681, "step": 39803 }, { "epoch": 2.470916878763424, "grad_norm": 0.4455482126771397, "learning_rate": 9.179233104356816e-06, "loss": 2.67, "step": 39804 }, { "epoch": 2.470978955863182, "grad_norm": 0.49561723967606497, "learning_rate": 9.177147681503062e-06, "loss": 2.6641, "step": 39805 }, { "epoch": 2.47104103296294, "grad_norm": 0.45078027748254235, "learning_rate": 9.175062471632035e-06, "loss": 2.6584, "step": 39806 }, { "epoch": 2.471103110062698, "grad_norm": 0.4436072106383329, "learning_rate": 9.172977474754624e-06, "loss": 2.6439, "step": 39807 }, { "epoch": 2.471165187162456, "grad_norm": 0.4605325439397673, "learning_rate": 9.170892690881705e-06, "loss": 2.6607, "step": 39808 }, { "epoch": 2.4712272642622137, "grad_norm": 0.4455048497029163, "learning_rate": 9.168808120024146e-06, "loss": 2.6605, "step": 39809 }, { "epoch": 2.4712893413619716, "grad_norm": 0.45114069579088256, "learning_rate": 9.166723762192819e-06, "loss": 2.7222, "step": 39810 }, { "epoch": 2.4713514184617296, "grad_norm": 0.4599892188446631, "learning_rate": 9.164639617398617e-06, "loss": 2.6511, "step": 39811 }, { "epoch": 2.4714134955614875, "grad_norm": 0.47405898593915313, "learning_rate": 9.162555685652403e-06, "loss": 2.6178, "step": 39812 }, { "epoch": 2.4714755726612454, "grad_norm": 0.5093401013454343, "learning_rate": 9.160471966965051e-06, "loss": 2.6795, "step": 39813 }, { "epoch": 2.4715376497610033, "grad_norm": 0.5334237728576232, "learning_rate": 9.158388461347427e-06, "loss": 2.6657, "step": 39814 }, { "epoch": 2.4715997268607612, "grad_norm": 0.4547432088643685, "learning_rate": 9.156305168810398e-06, "loss": 2.602, "step": 39815 }, { "epoch": 2.471661803960519, "grad_norm": 0.4659805022415385, "learning_rate": 9.15422208936485e-06, "loss": 2.635, "step": 39816 }, { "epoch": 2.471723881060277, "grad_norm": 0.5848991858187282, "learning_rate": 9.152139223021639e-06, "loss": 2.6984, "step": 39817 }, { "epoch": 2.4717859581600345, "grad_norm": 0.4761421663750508, "learning_rate": 9.150056569791637e-06, "loss": 2.585, "step": 39818 }, { "epoch": 2.471848035259793, "grad_norm": 0.44620371196473246, "learning_rate": 9.147974129685694e-06, "loss": 2.5995, "step": 39819 }, { "epoch": 2.4719101123595504, "grad_norm": 0.5214875467611145, "learning_rate": 9.145891902714698e-06, "loss": 2.662, "step": 39820 }, { "epoch": 2.4719721894593083, "grad_norm": 0.5218714445529821, "learning_rate": 9.143809888889499e-06, "loss": 2.6781, "step": 39821 }, { "epoch": 2.472034266559066, "grad_norm": 0.42818098398311244, "learning_rate": 9.14172808822097e-06, "loss": 2.7129, "step": 39822 }, { "epoch": 2.472096343658824, "grad_norm": 0.41015605018247553, "learning_rate": 9.139646500719944e-06, "loss": 2.7001, "step": 39823 }, { "epoch": 2.472158420758582, "grad_norm": 0.4854481560667259, "learning_rate": 9.137565126397319e-06, "loss": 2.7366, "step": 39824 }, { "epoch": 2.47222049785834, "grad_norm": 0.43004542961885334, "learning_rate": 9.13548396526393e-06, "loss": 2.6373, "step": 39825 }, { "epoch": 2.472282574958098, "grad_norm": 0.4452060438367102, "learning_rate": 9.13340301733065e-06, "loss": 2.6483, "step": 39826 }, { "epoch": 2.472344652057856, "grad_norm": 0.4162042396181911, "learning_rate": 9.131322282608323e-06, "loss": 2.5796, "step": 39827 }, { "epoch": 2.4724067291576137, "grad_norm": 0.4729286386288085, "learning_rate": 9.129241761107793e-06, "loss": 2.6897, "step": 39828 }, { "epoch": 2.4724688062573716, "grad_norm": 0.47768711845126166, "learning_rate": 9.127161452839932e-06, "loss": 2.7011, "step": 39829 }, { "epoch": 2.4725308833571296, "grad_norm": 0.422044490393252, "learning_rate": 9.125081357815607e-06, "loss": 2.4924, "step": 39830 }, { "epoch": 2.4725929604568875, "grad_norm": 0.4587278095408176, "learning_rate": 9.123001476045655e-06, "loss": 2.5302, "step": 39831 }, { "epoch": 2.4726550375566454, "grad_norm": 0.4378115872762289, "learning_rate": 9.120921807540927e-06, "loss": 2.6166, "step": 39832 }, { "epoch": 2.4727171146564033, "grad_norm": 0.4216616232228628, "learning_rate": 9.118842352312263e-06, "loss": 2.6954, "step": 39833 }, { "epoch": 2.4727791917561612, "grad_norm": 0.4157489763798961, "learning_rate": 9.11676311037053e-06, "loss": 2.6817, "step": 39834 }, { "epoch": 2.472841268855919, "grad_norm": 0.4816144448843209, "learning_rate": 9.114684081726576e-06, "loss": 2.6437, "step": 39835 }, { "epoch": 2.472903345955677, "grad_norm": 0.4434298320696031, "learning_rate": 9.112605266391234e-06, "loss": 2.7108, "step": 39836 }, { "epoch": 2.472965423055435, "grad_norm": 0.43111429014145347, "learning_rate": 9.110526664375357e-06, "loss": 2.5972, "step": 39837 }, { "epoch": 2.473027500155193, "grad_norm": 0.47526805993470456, "learning_rate": 9.108448275689779e-06, "loss": 2.7061, "step": 39838 }, { "epoch": 2.473089577254951, "grad_norm": 0.43853137516924345, "learning_rate": 9.106370100345363e-06, "loss": 2.6857, "step": 39839 }, { "epoch": 2.4731516543547087, "grad_norm": 0.392576492837578, "learning_rate": 9.104292138352944e-06, "loss": 2.6032, "step": 39840 }, { "epoch": 2.473213731454466, "grad_norm": 0.4824743049331941, "learning_rate": 9.102214389723356e-06, "loss": 2.6702, "step": 39841 }, { "epoch": 2.4732758085542246, "grad_norm": 0.4078918986290565, "learning_rate": 9.10013685446744e-06, "loss": 2.6783, "step": 39842 }, { "epoch": 2.473337885653982, "grad_norm": 0.367853575092638, "learning_rate": 9.098059532596043e-06, "loss": 2.6254, "step": 39843 }, { "epoch": 2.47339996275374, "grad_norm": 0.3974830565050385, "learning_rate": 9.095982424120003e-06, "loss": 2.608, "step": 39844 }, { "epoch": 2.473462039853498, "grad_norm": 0.4580713809538648, "learning_rate": 9.093905529050151e-06, "loss": 2.7418, "step": 39845 }, { "epoch": 2.473524116953256, "grad_norm": 0.4159396823106431, "learning_rate": 9.09182884739731e-06, "loss": 2.6454, "step": 39846 }, { "epoch": 2.4735861940530137, "grad_norm": 0.40710509613312584, "learning_rate": 9.089752379172345e-06, "loss": 2.6332, "step": 39847 }, { "epoch": 2.4736482711527716, "grad_norm": 0.4254108883013521, "learning_rate": 9.087676124386069e-06, "loss": 2.6906, "step": 39848 }, { "epoch": 2.4737103482525296, "grad_norm": 0.46261918684412945, "learning_rate": 9.085600083049322e-06, "loss": 2.6559, "step": 39849 }, { "epoch": 2.4737724253522875, "grad_norm": 0.42714492035906065, "learning_rate": 9.083524255172932e-06, "loss": 2.6546, "step": 39850 }, { "epoch": 2.4738345024520454, "grad_norm": 0.41468179921627824, "learning_rate": 9.081448640767714e-06, "loss": 2.7164, "step": 39851 }, { "epoch": 2.4738965795518033, "grad_norm": 0.4267487514066392, "learning_rate": 9.079373239844524e-06, "loss": 2.652, "step": 39852 }, { "epoch": 2.4739586566515612, "grad_norm": 0.4367837321842797, "learning_rate": 9.07729805241418e-06, "loss": 2.7245, "step": 39853 }, { "epoch": 2.474020733751319, "grad_norm": 0.4214018888604985, "learning_rate": 9.075223078487505e-06, "loss": 2.6323, "step": 39854 }, { "epoch": 2.474082810851077, "grad_norm": 0.4597542656829735, "learning_rate": 9.073148318075315e-06, "loss": 2.6292, "step": 39855 }, { "epoch": 2.474144887950835, "grad_norm": 0.4372256645030507, "learning_rate": 9.071073771188455e-06, "loss": 2.6868, "step": 39856 }, { "epoch": 2.474206965050593, "grad_norm": 0.4213415057524228, "learning_rate": 9.068999437837739e-06, "loss": 2.6792, "step": 39857 }, { "epoch": 2.474269042150351, "grad_norm": 0.4511291351551637, "learning_rate": 9.066925318033992e-06, "loss": 2.6976, "step": 39858 }, { "epoch": 2.4743311192501087, "grad_norm": 0.4264976703718809, "learning_rate": 9.064851411788016e-06, "loss": 2.6745, "step": 39859 }, { "epoch": 2.4743931963498667, "grad_norm": 0.4318507101843929, "learning_rate": 9.062777719110665e-06, "loss": 2.685, "step": 39860 }, { "epoch": 2.4744552734496246, "grad_norm": 0.45619861039025716, "learning_rate": 9.060704240012724e-06, "loss": 2.6999, "step": 39861 }, { "epoch": 2.4745173505493825, "grad_norm": 0.4176188457107486, "learning_rate": 9.058630974505045e-06, "loss": 2.6959, "step": 39862 }, { "epoch": 2.4745794276491404, "grad_norm": 0.4259332682958398, "learning_rate": 9.056557922598418e-06, "loss": 2.6417, "step": 39863 }, { "epoch": 2.4746415047488983, "grad_norm": 0.47404531232722846, "learning_rate": 9.054485084303677e-06, "loss": 2.7086, "step": 39864 }, { "epoch": 2.4747035818486562, "grad_norm": 0.419463985882987, "learning_rate": 9.05241245963161e-06, "loss": 2.7025, "step": 39865 }, { "epoch": 2.4747656589484137, "grad_norm": 0.4429569518655901, "learning_rate": 9.050340048593059e-06, "loss": 2.665, "step": 39866 }, { "epoch": 2.474827736048172, "grad_norm": 0.4244577503847547, "learning_rate": 9.048267851198827e-06, "loss": 2.694, "step": 39867 }, { "epoch": 2.4748898131479296, "grad_norm": 0.4498832266347651, "learning_rate": 9.046195867459722e-06, "loss": 2.6123, "step": 39868 }, { "epoch": 2.4749518902476875, "grad_norm": 0.4298602710517778, "learning_rate": 9.044124097386547e-06, "loss": 2.6934, "step": 39869 }, { "epoch": 2.4750139673474454, "grad_norm": 0.4588263056852313, "learning_rate": 9.042052540990126e-06, "loss": 2.7024, "step": 39870 }, { "epoch": 2.4750760444472033, "grad_norm": 0.4169145283981918, "learning_rate": 9.039981198281261e-06, "loss": 2.5933, "step": 39871 }, { "epoch": 2.4751381215469612, "grad_norm": 0.42981619208190835, "learning_rate": 9.03791006927076e-06, "loss": 2.7035, "step": 39872 }, { "epoch": 2.475200198646719, "grad_norm": 0.42857814113718856, "learning_rate": 9.035839153969411e-06, "loss": 2.7172, "step": 39873 }, { "epoch": 2.475262275746477, "grad_norm": 0.4031375905222083, "learning_rate": 9.033768452388048e-06, "loss": 2.6589, "step": 39874 }, { "epoch": 2.475324352846235, "grad_norm": 0.45044749182994603, "learning_rate": 9.031697964537461e-06, "loss": 2.7174, "step": 39875 }, { "epoch": 2.475386429945993, "grad_norm": 0.42159366057020825, "learning_rate": 9.029627690428444e-06, "loss": 2.7075, "step": 39876 }, { "epoch": 2.475448507045751, "grad_norm": 0.4235155033744014, "learning_rate": 9.027557630071814e-06, "loss": 2.6569, "step": 39877 }, { "epoch": 2.4755105841455087, "grad_norm": 0.4455446340955059, "learning_rate": 9.025487783478342e-06, "loss": 2.6731, "step": 39878 }, { "epoch": 2.4755726612452666, "grad_norm": 0.4534688828569141, "learning_rate": 9.023418150658864e-06, "loss": 2.6266, "step": 39879 }, { "epoch": 2.4756347383450246, "grad_norm": 0.4426609537539227, "learning_rate": 9.021348731624164e-06, "loss": 2.6036, "step": 39880 }, { "epoch": 2.4756968154447825, "grad_norm": 0.42589159944193794, "learning_rate": 9.019279526385028e-06, "loss": 2.6586, "step": 39881 }, { "epoch": 2.4757588925445404, "grad_norm": 0.4060265953838262, "learning_rate": 9.017210534952253e-06, "loss": 2.6941, "step": 39882 }, { "epoch": 2.4758209696442983, "grad_norm": 0.4699115824979366, "learning_rate": 9.015141757336647e-06, "loss": 2.6589, "step": 39883 }, { "epoch": 2.4758830467440562, "grad_norm": 0.4857265934667043, "learning_rate": 9.013073193548998e-06, "loss": 2.6255, "step": 39884 }, { "epoch": 2.475945123843814, "grad_norm": 0.4563919878503789, "learning_rate": 9.011004843600102e-06, "loss": 2.6213, "step": 39885 }, { "epoch": 2.476007200943572, "grad_norm": 0.42788066713824424, "learning_rate": 9.008936707500737e-06, "loss": 2.6755, "step": 39886 }, { "epoch": 2.47606927804333, "grad_norm": 0.4950714090164756, "learning_rate": 9.00686878526169e-06, "loss": 2.5844, "step": 39887 }, { "epoch": 2.476131355143088, "grad_norm": 0.4983657863053246, "learning_rate": 9.004801076893771e-06, "loss": 2.709, "step": 39888 }, { "epoch": 2.4761934322428454, "grad_norm": 0.43179770670949263, "learning_rate": 9.002733582407757e-06, "loss": 2.6675, "step": 39889 }, { "epoch": 2.4762555093426037, "grad_norm": 0.4621074828030952, "learning_rate": 9.000666301814436e-06, "loss": 2.7037, "step": 39890 }, { "epoch": 2.476317586442361, "grad_norm": 0.5137353430000703, "learning_rate": 8.998599235124577e-06, "loss": 2.7513, "step": 39891 }, { "epoch": 2.476379663542119, "grad_norm": 0.4652563395788258, "learning_rate": 8.996532382348994e-06, "loss": 2.727, "step": 39892 }, { "epoch": 2.476441740641877, "grad_norm": 0.47101753466053403, "learning_rate": 8.994465743498454e-06, "loss": 2.7523, "step": 39893 }, { "epoch": 2.476503817741635, "grad_norm": 0.48443421647897283, "learning_rate": 8.992399318583728e-06, "loss": 2.6441, "step": 39894 }, { "epoch": 2.476565894841393, "grad_norm": 0.4421232021749288, "learning_rate": 8.990333107615622e-06, "loss": 2.5723, "step": 39895 }, { "epoch": 2.476627971941151, "grad_norm": 0.4881992881649771, "learning_rate": 8.988267110604893e-06, "loss": 2.6477, "step": 39896 }, { "epoch": 2.4766900490409087, "grad_norm": 0.49111324726741035, "learning_rate": 8.986201327562343e-06, "loss": 2.7317, "step": 39897 }, { "epoch": 2.4767521261406666, "grad_norm": 0.41908896590939704, "learning_rate": 8.984135758498736e-06, "loss": 2.6384, "step": 39898 }, { "epoch": 2.4768142032404246, "grad_norm": 0.4604487980408835, "learning_rate": 8.982070403424847e-06, "loss": 2.693, "step": 39899 }, { "epoch": 2.4768762803401825, "grad_norm": 0.5060176822584607, "learning_rate": 8.98000526235146e-06, "loss": 2.6446, "step": 39900 }, { "epoch": 2.4769383574399404, "grad_norm": 0.4533358445109677, "learning_rate": 8.977940335289326e-06, "loss": 2.633, "step": 39901 }, { "epoch": 2.4770004345396983, "grad_norm": 0.41433974646868094, "learning_rate": 8.975875622249246e-06, "loss": 2.6093, "step": 39902 }, { "epoch": 2.4770625116394562, "grad_norm": 0.48964607229068574, "learning_rate": 8.973811123241988e-06, "loss": 2.7606, "step": 39903 }, { "epoch": 2.477124588739214, "grad_norm": 0.4918613715749775, "learning_rate": 8.971746838278316e-06, "loss": 2.7652, "step": 39904 }, { "epoch": 2.477186665838972, "grad_norm": 0.4866008223801847, "learning_rate": 8.969682767368986e-06, "loss": 2.7542, "step": 39905 }, { "epoch": 2.47724874293873, "grad_norm": 0.4652283143642468, "learning_rate": 8.967618910524795e-06, "loss": 2.6656, "step": 39906 }, { "epoch": 2.477310820038488, "grad_norm": 0.45248794168138645, "learning_rate": 8.965555267756498e-06, "loss": 2.7027, "step": 39907 }, { "epoch": 2.477372897138246, "grad_norm": 0.4453327609355231, "learning_rate": 8.963491839074861e-06, "loss": 2.6222, "step": 39908 }, { "epoch": 2.4774349742380037, "grad_norm": 0.49823450780174033, "learning_rate": 8.961428624490631e-06, "loss": 2.6771, "step": 39909 }, { "epoch": 2.4774970513377617, "grad_norm": 0.4722578876325357, "learning_rate": 8.959365624014609e-06, "loss": 2.658, "step": 39910 }, { "epoch": 2.4775591284375196, "grad_norm": 0.4521332114109163, "learning_rate": 8.957302837657538e-06, "loss": 2.7366, "step": 39911 }, { "epoch": 2.477621205537277, "grad_norm": 0.46598696176867305, "learning_rate": 8.955240265430181e-06, "loss": 2.6103, "step": 39912 }, { "epoch": 2.4776832826370354, "grad_norm": 0.503776858964975, "learning_rate": 8.953177907343297e-06, "loss": 2.6291, "step": 39913 }, { "epoch": 2.477745359736793, "grad_norm": 0.4297054980583437, "learning_rate": 8.951115763407642e-06, "loss": 2.5712, "step": 39914 }, { "epoch": 2.477807436836551, "grad_norm": 0.44388627995301366, "learning_rate": 8.949053833633986e-06, "loss": 2.7264, "step": 39915 }, { "epoch": 2.4778695139363087, "grad_norm": 0.4351553932332177, "learning_rate": 8.946992118033088e-06, "loss": 2.6227, "step": 39916 }, { "epoch": 2.4779315910360666, "grad_norm": 0.4677168746230641, "learning_rate": 8.944930616615693e-06, "loss": 2.7449, "step": 39917 }, { "epoch": 2.4779936681358246, "grad_norm": 0.45355059926680275, "learning_rate": 8.942869329392555e-06, "loss": 2.6454, "step": 39918 }, { "epoch": 2.4780557452355825, "grad_norm": 0.44512518908617976, "learning_rate": 8.940808256374445e-06, "loss": 2.7388, "step": 39919 }, { "epoch": 2.4781178223353404, "grad_norm": 0.4719052298759447, "learning_rate": 8.9387473975721e-06, "loss": 2.6912, "step": 39920 }, { "epoch": 2.4781798994350983, "grad_norm": 0.45004981745366673, "learning_rate": 8.936686752996282e-06, "loss": 2.6821, "step": 39921 }, { "epoch": 2.4782419765348562, "grad_norm": 0.4557087503128862, "learning_rate": 8.93462632265774e-06, "loss": 2.7506, "step": 39922 }, { "epoch": 2.478304053634614, "grad_norm": 0.4275449742814226, "learning_rate": 8.93256610656721e-06, "loss": 2.6814, "step": 39923 }, { "epoch": 2.478366130734372, "grad_norm": 0.4525250870246987, "learning_rate": 8.930506104735458e-06, "loss": 2.7181, "step": 39924 }, { "epoch": 2.47842820783413, "grad_norm": 0.44748404221346033, "learning_rate": 8.928446317173229e-06, "loss": 2.5962, "step": 39925 }, { "epoch": 2.478490284933888, "grad_norm": 0.43375689212479296, "learning_rate": 8.926386743891268e-06, "loss": 2.5703, "step": 39926 }, { "epoch": 2.478552362033646, "grad_norm": 0.4846413403093853, "learning_rate": 8.9243273849003e-06, "loss": 2.6553, "step": 39927 }, { "epoch": 2.4786144391334037, "grad_norm": 0.4193150416064197, "learning_rate": 8.922268240211095e-06, "loss": 2.6489, "step": 39928 }, { "epoch": 2.4786765162331617, "grad_norm": 0.42142784298124925, "learning_rate": 8.920209309834399e-06, "loss": 2.6679, "step": 39929 }, { "epoch": 2.4787385933329196, "grad_norm": 0.4425133518855503, "learning_rate": 8.918150593780938e-06, "loss": 2.657, "step": 39930 }, { "epoch": 2.4788006704326775, "grad_norm": 0.4501230674253004, "learning_rate": 8.916092092061462e-06, "loss": 2.6527, "step": 39931 }, { "epoch": 2.4788627475324354, "grad_norm": 0.4705513357235733, "learning_rate": 8.914033804686694e-06, "loss": 2.7197, "step": 39932 }, { "epoch": 2.4789248246321933, "grad_norm": 0.3856447120834162, "learning_rate": 8.911975731667399e-06, "loss": 2.7295, "step": 39933 }, { "epoch": 2.4789869017319512, "grad_norm": 0.4316875898262528, "learning_rate": 8.909917873014301e-06, "loss": 2.7366, "step": 39934 }, { "epoch": 2.479048978831709, "grad_norm": 0.4686774356624715, "learning_rate": 8.907860228738141e-06, "loss": 2.649, "step": 39935 }, { "epoch": 2.479111055931467, "grad_norm": 0.41749992102205363, "learning_rate": 8.905802798849643e-06, "loss": 2.5938, "step": 39936 }, { "epoch": 2.4791731330312246, "grad_norm": 0.42183224143461306, "learning_rate": 8.903745583359535e-06, "loss": 2.5941, "step": 39937 }, { "epoch": 2.479235210130983, "grad_norm": 0.4389982794226121, "learning_rate": 8.901688582278578e-06, "loss": 2.6888, "step": 39938 }, { "epoch": 2.4792972872307404, "grad_norm": 0.4593076176432545, "learning_rate": 8.899631795617492e-06, "loss": 2.5941, "step": 39939 }, { "epoch": 2.4793593643304983, "grad_norm": 0.42285566790484147, "learning_rate": 8.897575223387e-06, "loss": 2.6713, "step": 39940 }, { "epoch": 2.4794214414302562, "grad_norm": 0.4157607323000509, "learning_rate": 8.895518865597823e-06, "loss": 2.7062, "step": 39941 }, { "epoch": 2.479483518530014, "grad_norm": 0.4494091625518459, "learning_rate": 8.893462722260715e-06, "loss": 2.755, "step": 39942 }, { "epoch": 2.479545595629772, "grad_norm": 0.4483437018040481, "learning_rate": 8.891406793386392e-06, "loss": 2.6223, "step": 39943 }, { "epoch": 2.47960767272953, "grad_norm": 0.42949580338063187, "learning_rate": 8.889351078985581e-06, "loss": 2.7059, "step": 39944 }, { "epoch": 2.479669749829288, "grad_norm": 0.43102145773429557, "learning_rate": 8.887295579068989e-06, "loss": 2.7256, "step": 39945 }, { "epoch": 2.479731826929046, "grad_norm": 0.41120284836340126, "learning_rate": 8.885240293647373e-06, "loss": 2.5735, "step": 39946 }, { "epoch": 2.4797939040288037, "grad_norm": 0.4239813004848445, "learning_rate": 8.883185222731439e-06, "loss": 2.6858, "step": 39947 }, { "epoch": 2.4798559811285616, "grad_norm": 0.42402473846615507, "learning_rate": 8.881130366331908e-06, "loss": 2.6117, "step": 39948 }, { "epoch": 2.4799180582283196, "grad_norm": 0.4220618258165409, "learning_rate": 8.879075724459502e-06, "loss": 2.7313, "step": 39949 }, { "epoch": 2.4799801353280775, "grad_norm": 0.41711500802445706, "learning_rate": 8.877021297124922e-06, "loss": 2.6111, "step": 39950 }, { "epoch": 2.4800422124278354, "grad_norm": 0.4102181524157739, "learning_rate": 8.874967084338925e-06, "loss": 2.5569, "step": 39951 }, { "epoch": 2.4801042895275933, "grad_norm": 0.44495817778184216, "learning_rate": 8.872913086112205e-06, "loss": 2.7041, "step": 39952 }, { "epoch": 2.4801663666273512, "grad_norm": 0.41946876387307297, "learning_rate": 8.87085930245548e-06, "loss": 2.6546, "step": 39953 }, { "epoch": 2.480228443727109, "grad_norm": 0.3840038096728193, "learning_rate": 8.86880573337946e-06, "loss": 2.6844, "step": 39954 }, { "epoch": 2.480290520826867, "grad_norm": 0.3905903610130081, "learning_rate": 8.866752378894871e-06, "loss": 2.7026, "step": 39955 }, { "epoch": 2.480352597926625, "grad_norm": 0.4016809095961178, "learning_rate": 8.864699239012425e-06, "loss": 2.6438, "step": 39956 }, { "epoch": 2.480414675026383, "grad_norm": 0.40764674914376375, "learning_rate": 8.862646313742828e-06, "loss": 2.6152, "step": 39957 }, { "epoch": 2.480476752126141, "grad_norm": 0.41777130169208826, "learning_rate": 8.860593603096778e-06, "loss": 2.6296, "step": 39958 }, { "epoch": 2.4805388292258987, "grad_norm": 0.4323646204260325, "learning_rate": 8.858541107085e-06, "loss": 2.7398, "step": 39959 }, { "epoch": 2.480600906325656, "grad_norm": 0.4093228001110401, "learning_rate": 8.856488825718211e-06, "loss": 2.5695, "step": 39960 }, { "epoch": 2.4806629834254146, "grad_norm": 0.4168667571126481, "learning_rate": 8.854436759007106e-06, "loss": 2.7076, "step": 39961 }, { "epoch": 2.480725060525172, "grad_norm": 0.465457591102019, "learning_rate": 8.852384906962397e-06, "loss": 2.697, "step": 39962 }, { "epoch": 2.48078713762493, "grad_norm": 0.4433797083825952, "learning_rate": 8.850333269594786e-06, "loss": 2.6752, "step": 39963 }, { "epoch": 2.480849214724688, "grad_norm": 0.4368847540010749, "learning_rate": 8.848281846914958e-06, "loss": 2.6249, "step": 39964 }, { "epoch": 2.480911291824446, "grad_norm": 0.41943464179124557, "learning_rate": 8.846230638933651e-06, "loss": 2.7324, "step": 39965 }, { "epoch": 2.4809733689242037, "grad_norm": 0.42258608900292693, "learning_rate": 8.844179645661543e-06, "loss": 2.6205, "step": 39966 }, { "epoch": 2.4810354460239616, "grad_norm": 0.4772010964665602, "learning_rate": 8.842128867109345e-06, "loss": 2.6461, "step": 39967 }, { "epoch": 2.4810975231237196, "grad_norm": 0.42372015707318456, "learning_rate": 8.840078303287741e-06, "loss": 2.6845, "step": 39968 }, { "epoch": 2.4811596002234775, "grad_norm": 0.49157097971029373, "learning_rate": 8.838027954207451e-06, "loss": 2.6803, "step": 39969 }, { "epoch": 2.4812216773232354, "grad_norm": 0.456836724250054, "learning_rate": 8.83597781987916e-06, "loss": 2.7061, "step": 39970 }, { "epoch": 2.4812837544229933, "grad_norm": 0.4351109602695137, "learning_rate": 8.833927900313565e-06, "loss": 2.6692, "step": 39971 }, { "epoch": 2.4813458315227512, "grad_norm": 0.4716964939827388, "learning_rate": 8.831878195521365e-06, "loss": 2.6788, "step": 39972 }, { "epoch": 2.481407908622509, "grad_norm": 0.437261073402487, "learning_rate": 8.829828705513238e-06, "loss": 2.6459, "step": 39973 }, { "epoch": 2.481469985722267, "grad_norm": 0.4091526286571597, "learning_rate": 8.827779430299898e-06, "loss": 2.6408, "step": 39974 }, { "epoch": 2.481532062822025, "grad_norm": 0.4425531863452906, "learning_rate": 8.825730369892032e-06, "loss": 2.6103, "step": 39975 }, { "epoch": 2.481594139921783, "grad_norm": 0.4734058426196691, "learning_rate": 8.82368152430032e-06, "loss": 2.6418, "step": 39976 }, { "epoch": 2.481656217021541, "grad_norm": 0.4331481809316548, "learning_rate": 8.821632893535447e-06, "loss": 2.7395, "step": 39977 }, { "epoch": 2.4817182941212987, "grad_norm": 0.4163943255007163, "learning_rate": 8.819584477608129e-06, "loss": 2.6394, "step": 39978 }, { "epoch": 2.4817803712210567, "grad_norm": 0.4457704464806337, "learning_rate": 8.817536276529032e-06, "loss": 2.7209, "step": 39979 }, { "epoch": 2.4818424483208146, "grad_norm": 0.44603780606110766, "learning_rate": 8.815488290308848e-06, "loss": 2.7536, "step": 39980 }, { "epoch": 2.4819045254205725, "grad_norm": 0.3930374608725573, "learning_rate": 8.813440518958244e-06, "loss": 2.655, "step": 39981 }, { "epoch": 2.4819666025203304, "grad_norm": 0.41883677964108856, "learning_rate": 8.81139296248793e-06, "loss": 2.7155, "step": 39982 }, { "epoch": 2.4820286796200883, "grad_norm": 0.43638521604727837, "learning_rate": 8.809345620908583e-06, "loss": 2.6792, "step": 39983 }, { "epoch": 2.4820907567198462, "grad_norm": 0.4584359809054616, "learning_rate": 8.807298494230876e-06, "loss": 2.7144, "step": 39984 }, { "epoch": 2.4821528338196037, "grad_norm": 0.45983010150280224, "learning_rate": 8.805251582465496e-06, "loss": 2.6631, "step": 39985 }, { "epoch": 2.482214910919362, "grad_norm": 0.3991841631574135, "learning_rate": 8.8032048856231e-06, "loss": 2.5744, "step": 39986 }, { "epoch": 2.4822769880191196, "grad_norm": 0.4214127091650732, "learning_rate": 8.801158403714405e-06, "loss": 2.76, "step": 39987 }, { "epoch": 2.4823390651188775, "grad_norm": 0.42814002115113703, "learning_rate": 8.799112136750059e-06, "loss": 2.6515, "step": 39988 }, { "epoch": 2.4824011422186354, "grad_norm": 0.453948259625553, "learning_rate": 8.797066084740752e-06, "loss": 2.713, "step": 39989 }, { "epoch": 2.4824632193183933, "grad_norm": 0.4094588390432749, "learning_rate": 8.795020247697144e-06, "loss": 2.6688, "step": 39990 }, { "epoch": 2.4825252964181512, "grad_norm": 0.4617201738932643, "learning_rate": 8.79297462562993e-06, "loss": 2.6938, "step": 39991 }, { "epoch": 2.482587373517909, "grad_norm": 0.4521412529590846, "learning_rate": 8.790929218549754e-06, "loss": 2.6599, "step": 39992 }, { "epoch": 2.482649450617667, "grad_norm": 0.43794151575319645, "learning_rate": 8.78888402646732e-06, "loss": 2.7651, "step": 39993 }, { "epoch": 2.482711527717425, "grad_norm": 0.4405141440739316, "learning_rate": 8.78683904939328e-06, "loss": 2.6812, "step": 39994 }, { "epoch": 2.482773604817183, "grad_norm": 0.47202328000605626, "learning_rate": 8.784794287338293e-06, "loss": 2.6896, "step": 39995 }, { "epoch": 2.482835681916941, "grad_norm": 0.4220576067719509, "learning_rate": 8.782749740313051e-06, "loss": 2.6156, "step": 39996 }, { "epoch": 2.4828977590166987, "grad_norm": 0.4345909713806607, "learning_rate": 8.780705408328217e-06, "loss": 2.7269, "step": 39997 }, { "epoch": 2.4829598361164567, "grad_norm": 0.4481582895520731, "learning_rate": 8.778661291394441e-06, "loss": 2.6715, "step": 39998 }, { "epoch": 2.4830219132162146, "grad_norm": 0.49851683341045594, "learning_rate": 8.776617389522395e-06, "loss": 2.7214, "step": 39999 }, { "epoch": 2.4830839903159725, "grad_norm": 0.4423616753722138, "learning_rate": 8.774573702722738e-06, "loss": 2.6839, "step": 40000 }, { "epoch": 2.4831460674157304, "grad_norm": 0.43125867558474124, "learning_rate": 8.772530231006149e-06, "loss": 2.6313, "step": 40001 }, { "epoch": 2.4832081445154883, "grad_norm": 0.4768244851831555, "learning_rate": 8.770486974383274e-06, "loss": 2.6355, "step": 40002 }, { "epoch": 2.4832702216152462, "grad_norm": 0.48840940700972457, "learning_rate": 8.768443932864778e-06, "loss": 2.6233, "step": 40003 }, { "epoch": 2.483332298715004, "grad_norm": 0.44573763580846526, "learning_rate": 8.766401106461309e-06, "loss": 2.6509, "step": 40004 }, { "epoch": 2.483394375814762, "grad_norm": 0.4242470600102734, "learning_rate": 8.764358495183544e-06, "loss": 2.7807, "step": 40005 }, { "epoch": 2.48345645291452, "grad_norm": 0.5199726208943324, "learning_rate": 8.762316099042139e-06, "loss": 2.6376, "step": 40006 }, { "epoch": 2.483518530014278, "grad_norm": 0.4575007220038465, "learning_rate": 8.760273918047735e-06, "loss": 2.7239, "step": 40007 }, { "epoch": 2.4835806071140354, "grad_norm": 0.4260164538651222, "learning_rate": 8.758231952210982e-06, "loss": 2.6904, "step": 40008 }, { "epoch": 2.4836426842137937, "grad_norm": 0.5276305654937599, "learning_rate": 8.756190201542557e-06, "loss": 2.7377, "step": 40009 }, { "epoch": 2.4837047613135512, "grad_norm": 0.46533235147631297, "learning_rate": 8.754148666053102e-06, "loss": 2.7516, "step": 40010 }, { "epoch": 2.483766838413309, "grad_norm": 0.4148949765691292, "learning_rate": 8.752107345753263e-06, "loss": 2.7182, "step": 40011 }, { "epoch": 2.483828915513067, "grad_norm": 0.46129190430342026, "learning_rate": 8.750066240653693e-06, "loss": 2.6358, "step": 40012 }, { "epoch": 2.483890992612825, "grad_norm": 0.48509797623168355, "learning_rate": 8.748025350765032e-06, "loss": 2.6267, "step": 40013 }, { "epoch": 2.483953069712583, "grad_norm": 0.4507852854953005, "learning_rate": 8.745984676097951e-06, "loss": 2.6928, "step": 40014 }, { "epoch": 2.484015146812341, "grad_norm": 0.4411571947846845, "learning_rate": 8.74394421666308e-06, "loss": 2.6397, "step": 40015 }, { "epoch": 2.4840772239120987, "grad_norm": 0.42031157397767344, "learning_rate": 8.741903972471066e-06, "loss": 2.6588, "step": 40016 }, { "epoch": 2.4841393010118566, "grad_norm": 0.4617983650979848, "learning_rate": 8.739863943532544e-06, "loss": 2.6462, "step": 40017 }, { "epoch": 2.4842013781116146, "grad_norm": 0.44380399281479316, "learning_rate": 8.73782412985818e-06, "loss": 2.7124, "step": 40018 }, { "epoch": 2.4842634552113725, "grad_norm": 0.43395063816792373, "learning_rate": 8.735784531458602e-06, "loss": 2.5989, "step": 40019 }, { "epoch": 2.4843255323111304, "grad_norm": 0.3812975588042325, "learning_rate": 8.733745148344458e-06, "loss": 2.5499, "step": 40020 }, { "epoch": 2.4843876094108883, "grad_norm": 0.4334752032459842, "learning_rate": 8.731705980526378e-06, "loss": 2.6345, "step": 40021 }, { "epoch": 2.4844496865106462, "grad_norm": 0.4247227556866199, "learning_rate": 8.729667028014998e-06, "loss": 2.6453, "step": 40022 }, { "epoch": 2.484511763610404, "grad_norm": 0.4427972162068311, "learning_rate": 8.727628290820972e-06, "loss": 2.6648, "step": 40023 }, { "epoch": 2.484573840710162, "grad_norm": 0.40807137804485827, "learning_rate": 8.725589768954934e-06, "loss": 2.627, "step": 40024 }, { "epoch": 2.48463591780992, "grad_norm": 0.40412167424875656, "learning_rate": 8.723551462427498e-06, "loss": 2.6466, "step": 40025 }, { "epoch": 2.484697994909678, "grad_norm": 0.43360710123487384, "learning_rate": 8.72151337124933e-06, "loss": 2.6964, "step": 40026 }, { "epoch": 2.484760072009436, "grad_norm": 0.4451414080763596, "learning_rate": 8.71947549543103e-06, "loss": 2.724, "step": 40027 }, { "epoch": 2.4848221491091937, "grad_norm": 0.3916003448391293, "learning_rate": 8.717437834983262e-06, "loss": 2.6404, "step": 40028 }, { "epoch": 2.4848842262089517, "grad_norm": 0.4645098761431483, "learning_rate": 8.715400389916645e-06, "loss": 2.6389, "step": 40029 }, { "epoch": 2.4849463033087096, "grad_norm": 0.4695394067801579, "learning_rate": 8.7133631602418e-06, "loss": 2.6285, "step": 40030 }, { "epoch": 2.4850083804084675, "grad_norm": 0.4994097295102457, "learning_rate": 8.711326145969356e-06, "loss": 2.7181, "step": 40031 }, { "epoch": 2.4850704575082254, "grad_norm": 0.44653744119392247, "learning_rate": 8.709289347109956e-06, "loss": 2.779, "step": 40032 }, { "epoch": 2.485132534607983, "grad_norm": 0.4238386372611746, "learning_rate": 8.707252763674218e-06, "loss": 2.768, "step": 40033 }, { "epoch": 2.4851946117077413, "grad_norm": 0.4588775023358647, "learning_rate": 8.705216395672766e-06, "loss": 2.6521, "step": 40034 }, { "epoch": 2.4852566888074987, "grad_norm": 0.4300607100866541, "learning_rate": 8.703180243116227e-06, "loss": 2.6209, "step": 40035 }, { "epoch": 2.4853187659072566, "grad_norm": 0.4367725932823269, "learning_rate": 8.701144306015207e-06, "loss": 2.7425, "step": 40036 }, { "epoch": 2.4853808430070146, "grad_norm": 0.4071548362155195, "learning_rate": 8.699108584380356e-06, "loss": 2.6204, "step": 40037 }, { "epoch": 2.4854429201067725, "grad_norm": 0.4248433062658767, "learning_rate": 8.69707307822228e-06, "loss": 2.7634, "step": 40038 }, { "epoch": 2.4855049972065304, "grad_norm": 0.4266372615148644, "learning_rate": 8.695037787551602e-06, "loss": 2.6848, "step": 40039 }, { "epoch": 2.4855670743062883, "grad_norm": 0.41608118606261363, "learning_rate": 8.693002712378929e-06, "loss": 2.6996, "step": 40040 }, { "epoch": 2.4856291514060462, "grad_norm": 0.4318525734694347, "learning_rate": 8.690967852714899e-06, "loss": 2.6991, "step": 40041 }, { "epoch": 2.485691228505804, "grad_norm": 0.4838476232600202, "learning_rate": 8.688933208570116e-06, "loss": 2.7633, "step": 40042 }, { "epoch": 2.485753305605562, "grad_norm": 0.47633791149685323, "learning_rate": 8.6868987799552e-06, "loss": 2.7104, "step": 40043 }, { "epoch": 2.48581538270532, "grad_norm": 0.4071089028000861, "learning_rate": 8.684864566880746e-06, "loss": 2.5943, "step": 40044 }, { "epoch": 2.485877459805078, "grad_norm": 0.43981581238397416, "learning_rate": 8.682830569357398e-06, "loss": 2.6848, "step": 40045 }, { "epoch": 2.485939536904836, "grad_norm": 0.4524281997778725, "learning_rate": 8.68079678739575e-06, "loss": 2.7467, "step": 40046 }, { "epoch": 2.4860016140045937, "grad_norm": 0.43357565558541844, "learning_rate": 8.67876322100642e-06, "loss": 2.5873, "step": 40047 }, { "epoch": 2.4860636911043517, "grad_norm": 0.47459745980835644, "learning_rate": 8.676729870200012e-06, "loss": 2.6624, "step": 40048 }, { "epoch": 2.4861257682041096, "grad_norm": 0.43386811519821394, "learning_rate": 8.67469673498712e-06, "loss": 2.6562, "step": 40049 }, { "epoch": 2.4861878453038675, "grad_norm": 0.4195996312224801, "learning_rate": 8.672663815378385e-06, "loss": 2.6971, "step": 40050 }, { "epoch": 2.4862499224036254, "grad_norm": 0.4213369435144881, "learning_rate": 8.670631111384392e-06, "loss": 2.6817, "step": 40051 }, { "epoch": 2.4863119995033833, "grad_norm": 0.44511266875975825, "learning_rate": 8.66859862301575e-06, "loss": 2.7033, "step": 40052 }, { "epoch": 2.4863740766031412, "grad_norm": 0.4248503211054442, "learning_rate": 8.666566350283052e-06, "loss": 2.6824, "step": 40053 }, { "epoch": 2.486436153702899, "grad_norm": 0.4175481551774933, "learning_rate": 8.664534293196924e-06, "loss": 2.6283, "step": 40054 }, { "epoch": 2.486498230802657, "grad_norm": 0.4363940599803442, "learning_rate": 8.66250245176795e-06, "loss": 2.6826, "step": 40055 }, { "epoch": 2.4865603079024146, "grad_norm": 0.41913333754945764, "learning_rate": 8.660470826006744e-06, "loss": 2.5932, "step": 40056 }, { "epoch": 2.486622385002173, "grad_norm": 0.4198206402487909, "learning_rate": 8.65843941592389e-06, "loss": 2.7411, "step": 40057 }, { "epoch": 2.4866844621019304, "grad_norm": 0.4283541191595425, "learning_rate": 8.656408221529988e-06, "loss": 2.6453, "step": 40058 }, { "epoch": 2.4867465392016883, "grad_norm": 0.41100445191090734, "learning_rate": 8.654377242835632e-06, "loss": 2.6696, "step": 40059 }, { "epoch": 2.4868086163014462, "grad_norm": 0.3933049396325937, "learning_rate": 8.652346479851447e-06, "loss": 2.6627, "step": 40060 }, { "epoch": 2.486870693401204, "grad_norm": 0.4240119113699752, "learning_rate": 8.650315932588005e-06, "loss": 2.6772, "step": 40061 }, { "epoch": 2.486932770500962, "grad_norm": 0.41048478184159615, "learning_rate": 8.6482856010559e-06, "loss": 2.5877, "step": 40062 }, { "epoch": 2.48699484760072, "grad_norm": 0.3960667181478147, "learning_rate": 8.64625548526572e-06, "loss": 2.6095, "step": 40063 }, { "epoch": 2.487056924700478, "grad_norm": 0.4413762040166387, "learning_rate": 8.644225585228071e-06, "loss": 2.725, "step": 40064 }, { "epoch": 2.487119001800236, "grad_norm": 0.4071788439705246, "learning_rate": 8.642195900953543e-06, "loss": 2.7281, "step": 40065 }, { "epoch": 2.4871810788999937, "grad_norm": 0.40575593469126, "learning_rate": 8.640166432452718e-06, "loss": 2.7085, "step": 40066 }, { "epoch": 2.4872431559997517, "grad_norm": 0.4029821127857293, "learning_rate": 8.638137179736167e-06, "loss": 2.6661, "step": 40067 }, { "epoch": 2.4873052330995096, "grad_norm": 0.4623555821396651, "learning_rate": 8.636108142814514e-06, "loss": 2.6381, "step": 40068 }, { "epoch": 2.4873673101992675, "grad_norm": 0.455147897504233, "learning_rate": 8.634079321698324e-06, "loss": 2.618, "step": 40069 }, { "epoch": 2.4874293872990254, "grad_norm": 0.4538123081187431, "learning_rate": 8.632050716398183e-06, "loss": 2.5828, "step": 40070 }, { "epoch": 2.4874914643987833, "grad_norm": 0.4430510671515141, "learning_rate": 8.630022326924676e-06, "loss": 2.728, "step": 40071 }, { "epoch": 2.4875535414985412, "grad_norm": 0.4007056694820162, "learning_rate": 8.627994153288372e-06, "loss": 2.5409, "step": 40072 }, { "epoch": 2.487615618598299, "grad_norm": 0.4551234898037936, "learning_rate": 8.625966195499879e-06, "loss": 2.68, "step": 40073 }, { "epoch": 2.487677695698057, "grad_norm": 0.42906480231504374, "learning_rate": 8.623938453569768e-06, "loss": 2.7076, "step": 40074 }, { "epoch": 2.487739772797815, "grad_norm": 0.4501394042676778, "learning_rate": 8.62191092750861e-06, "loss": 2.6524, "step": 40075 }, { "epoch": 2.487801849897573, "grad_norm": 0.40991049170387883, "learning_rate": 8.619883617326979e-06, "loss": 2.6418, "step": 40076 }, { "epoch": 2.487863926997331, "grad_norm": 0.43182313952725687, "learning_rate": 8.617856523035466e-06, "loss": 2.7198, "step": 40077 }, { "epoch": 2.4879260040970888, "grad_norm": 0.4282769894031362, "learning_rate": 8.615829644644646e-06, "loss": 2.5806, "step": 40078 }, { "epoch": 2.4879880811968467, "grad_norm": 0.46792507700213526, "learning_rate": 8.613802982165093e-06, "loss": 2.6362, "step": 40079 }, { "epoch": 2.4880501582966046, "grad_norm": 0.4305807885172978, "learning_rate": 8.611776535607358e-06, "loss": 2.6526, "step": 40080 }, { "epoch": 2.488112235396362, "grad_norm": 0.4760376588526101, "learning_rate": 8.609750304982051e-06, "loss": 2.6144, "step": 40081 }, { "epoch": 2.4881743124961204, "grad_norm": 0.4329978869465902, "learning_rate": 8.607724290299717e-06, "loss": 2.6154, "step": 40082 }, { "epoch": 2.488236389595878, "grad_norm": 0.4406048918931277, "learning_rate": 8.605698491570935e-06, "loss": 2.7604, "step": 40083 }, { "epoch": 2.488298466695636, "grad_norm": 0.47694477008372016, "learning_rate": 8.603672908806276e-06, "loss": 2.6539, "step": 40084 }, { "epoch": 2.4883605437953937, "grad_norm": 0.43072076886934224, "learning_rate": 8.601647542016294e-06, "loss": 2.5319, "step": 40085 }, { "epoch": 2.4884226208951516, "grad_norm": 0.47209968585989887, "learning_rate": 8.599622391211576e-06, "loss": 2.7031, "step": 40086 }, { "epoch": 2.4884846979949096, "grad_norm": 0.44814633609942794, "learning_rate": 8.59759745640268e-06, "loss": 2.5782, "step": 40087 }, { "epoch": 2.4885467750946675, "grad_norm": 0.43457490725713516, "learning_rate": 8.595572737600166e-06, "loss": 2.7087, "step": 40088 }, { "epoch": 2.4886088521944254, "grad_norm": 0.41037308547066903, "learning_rate": 8.593548234814592e-06, "loss": 2.6587, "step": 40089 }, { "epoch": 2.4886709292941833, "grad_norm": 0.4752674955772748, "learning_rate": 8.591523948056524e-06, "loss": 2.6475, "step": 40090 }, { "epoch": 2.4887330063939412, "grad_norm": 0.4414486316150728, "learning_rate": 8.589499877336544e-06, "loss": 2.7153, "step": 40091 }, { "epoch": 2.488795083493699, "grad_norm": 0.40346840271023693, "learning_rate": 8.587476022665192e-06, "loss": 2.6289, "step": 40092 }, { "epoch": 2.488857160593457, "grad_norm": 0.45576662363797404, "learning_rate": 8.585452384053034e-06, "loss": 2.7152, "step": 40093 }, { "epoch": 2.488919237693215, "grad_norm": 0.44118521228245244, "learning_rate": 8.583428961510609e-06, "loss": 2.5827, "step": 40094 }, { "epoch": 2.488981314792973, "grad_norm": 0.4117169613591236, "learning_rate": 8.581405755048506e-06, "loss": 2.6994, "step": 40095 }, { "epoch": 2.489043391892731, "grad_norm": 0.4597164727370405, "learning_rate": 8.579382764677257e-06, "loss": 2.6237, "step": 40096 }, { "epoch": 2.4891054689924887, "grad_norm": 0.44542179105015667, "learning_rate": 8.577359990407429e-06, "loss": 2.6706, "step": 40097 }, { "epoch": 2.4891675460922467, "grad_norm": 0.4245827807790039, "learning_rate": 8.57533743224957e-06, "loss": 2.7029, "step": 40098 }, { "epoch": 2.4892296231920046, "grad_norm": 0.4537392267939672, "learning_rate": 8.573315090214217e-06, "loss": 2.6896, "step": 40099 }, { "epoch": 2.4892917002917625, "grad_norm": 0.49307129242308734, "learning_rate": 8.57129296431195e-06, "loss": 2.7252, "step": 40100 }, { "epoch": 2.4893537773915204, "grad_norm": 0.37790090367516277, "learning_rate": 8.569271054553308e-06, "loss": 2.6672, "step": 40101 }, { "epoch": 2.4894158544912783, "grad_norm": 0.46531901386382263, "learning_rate": 8.567249360948832e-06, "loss": 2.6171, "step": 40102 }, { "epoch": 2.4894779315910363, "grad_norm": 0.4402584411216947, "learning_rate": 8.565227883509059e-06, "loss": 2.7214, "step": 40103 }, { "epoch": 2.4895400086907937, "grad_norm": 0.45263176414304734, "learning_rate": 8.563206622244568e-06, "loss": 2.6118, "step": 40104 }, { "epoch": 2.489602085790552, "grad_norm": 0.4180604887492087, "learning_rate": 8.561185577165886e-06, "loss": 2.7244, "step": 40105 }, { "epoch": 2.4896641628903096, "grad_norm": 0.4185906712047531, "learning_rate": 8.55916474828356e-06, "loss": 2.7257, "step": 40106 }, { "epoch": 2.4897262399900675, "grad_norm": 0.4939984656138278, "learning_rate": 8.55714413560813e-06, "loss": 2.6689, "step": 40107 }, { "epoch": 2.4897883170898254, "grad_norm": 0.4394717572818732, "learning_rate": 8.555123739150123e-06, "loss": 2.6213, "step": 40108 }, { "epoch": 2.4898503941895833, "grad_norm": 0.43910956622592845, "learning_rate": 8.553103558920117e-06, "loss": 2.6995, "step": 40109 }, { "epoch": 2.4899124712893412, "grad_norm": 0.5130070492462752, "learning_rate": 8.551083594928627e-06, "loss": 2.6364, "step": 40110 }, { "epoch": 2.489974548389099, "grad_norm": 0.44100344685853154, "learning_rate": 8.549063847186195e-06, "loss": 2.6221, "step": 40111 }, { "epoch": 2.490036625488857, "grad_norm": 0.46695312374907033, "learning_rate": 8.54704431570335e-06, "loss": 2.612, "step": 40112 }, { "epoch": 2.490098702588615, "grad_norm": 0.4609423653297333, "learning_rate": 8.545025000490648e-06, "loss": 2.6053, "step": 40113 }, { "epoch": 2.490160779688373, "grad_norm": 0.4518565477795699, "learning_rate": 8.543005901558614e-06, "loss": 2.5834, "step": 40114 }, { "epoch": 2.490222856788131, "grad_norm": 0.4932919176003684, "learning_rate": 8.540987018917784e-06, "loss": 2.581, "step": 40115 }, { "epoch": 2.4902849338878887, "grad_norm": 0.5318653525834638, "learning_rate": 8.538968352578674e-06, "loss": 2.63, "step": 40116 }, { "epoch": 2.4903470109876467, "grad_norm": 0.4514441556190375, "learning_rate": 8.536949902551849e-06, "loss": 2.6446, "step": 40117 }, { "epoch": 2.4904090880874046, "grad_norm": 0.432081989392306, "learning_rate": 8.534931668847817e-06, "loss": 2.6024, "step": 40118 }, { "epoch": 2.4904711651871625, "grad_norm": 0.48030329390807486, "learning_rate": 8.532913651477114e-06, "loss": 2.642, "step": 40119 }, { "epoch": 2.4905332422869204, "grad_norm": 0.48937294037581586, "learning_rate": 8.530895850450266e-06, "loss": 2.7306, "step": 40120 }, { "epoch": 2.4905953193866783, "grad_norm": 0.4486173667789773, "learning_rate": 8.52887826577779e-06, "loss": 2.6672, "step": 40121 }, { "epoch": 2.4906573964864362, "grad_norm": 0.4960198384891753, "learning_rate": 8.526860897470234e-06, "loss": 2.7077, "step": 40122 }, { "epoch": 2.490719473586194, "grad_norm": 0.4754158828282003, "learning_rate": 8.524843745538103e-06, "loss": 2.6332, "step": 40123 }, { "epoch": 2.490781550685952, "grad_norm": 0.46742984520736613, "learning_rate": 8.522826809991941e-06, "loss": 2.6624, "step": 40124 }, { "epoch": 2.49084362778571, "grad_norm": 0.4351727884872046, "learning_rate": 8.520810090842263e-06, "loss": 2.5848, "step": 40125 }, { "epoch": 2.490905704885468, "grad_norm": 0.4581874828481717, "learning_rate": 8.518793588099571e-06, "loss": 2.6491, "step": 40126 }, { "epoch": 2.490967781985226, "grad_norm": 0.4348947064352759, "learning_rate": 8.51677730177442e-06, "loss": 2.738, "step": 40127 }, { "epoch": 2.4910298590849838, "grad_norm": 0.43346181357136565, "learning_rate": 8.51476123187731e-06, "loss": 2.7339, "step": 40128 }, { "epoch": 2.4910919361847412, "grad_norm": 0.4342623701831787, "learning_rate": 8.512745378418763e-06, "loss": 2.604, "step": 40129 }, { "epoch": 2.4911540132844996, "grad_norm": 0.4468922951826044, "learning_rate": 8.510729741409285e-06, "loss": 2.5917, "step": 40130 }, { "epoch": 2.491216090384257, "grad_norm": 0.44382658888502824, "learning_rate": 8.508714320859406e-06, "loss": 2.7265, "step": 40131 }, { "epoch": 2.491278167484015, "grad_norm": 0.45822099551631645, "learning_rate": 8.506699116779643e-06, "loss": 2.6555, "step": 40132 }, { "epoch": 2.491340244583773, "grad_norm": 0.45114053064102955, "learning_rate": 8.504684129180501e-06, "loss": 2.6502, "step": 40133 }, { "epoch": 2.491402321683531, "grad_norm": 0.41537076373469906, "learning_rate": 8.502669358072497e-06, "loss": 2.6522, "step": 40134 }, { "epoch": 2.4914643987832887, "grad_norm": 0.41642263140492874, "learning_rate": 8.50065480346613e-06, "loss": 2.7882, "step": 40135 }, { "epoch": 2.4915264758830467, "grad_norm": 0.4235440721864153, "learning_rate": 8.498640465371932e-06, "loss": 2.5917, "step": 40136 }, { "epoch": 2.4915885529828046, "grad_norm": 0.39976441330061674, "learning_rate": 8.496626343800401e-06, "loss": 2.6632, "step": 40137 }, { "epoch": 2.4916506300825625, "grad_norm": 0.4482837566306981, "learning_rate": 8.494612438762045e-06, "loss": 2.6681, "step": 40138 }, { "epoch": 2.4917127071823204, "grad_norm": 0.4201838009143917, "learning_rate": 8.492598750267361e-06, "loss": 2.6023, "step": 40139 }, { "epoch": 2.4917747842820783, "grad_norm": 0.4183517733185156, "learning_rate": 8.490585278326873e-06, "loss": 2.7749, "step": 40140 }, { "epoch": 2.4918368613818362, "grad_norm": 0.42845777192351514, "learning_rate": 8.488572022951085e-06, "loss": 2.7624, "step": 40141 }, { "epoch": 2.491898938481594, "grad_norm": 0.45768939827131144, "learning_rate": 8.48655898415049e-06, "loss": 2.6954, "step": 40142 }, { "epoch": 2.491961015581352, "grad_norm": 0.44394680212587434, "learning_rate": 8.484546161935597e-06, "loss": 2.6621, "step": 40143 }, { "epoch": 2.49202309268111, "grad_norm": 0.4355701560054941, "learning_rate": 8.48253355631689e-06, "loss": 2.5582, "step": 40144 }, { "epoch": 2.492085169780868, "grad_norm": 0.47083244288481285, "learning_rate": 8.480521167304894e-06, "loss": 2.6691, "step": 40145 }, { "epoch": 2.492147246880626, "grad_norm": 0.4578196092125215, "learning_rate": 8.478508994910106e-06, "loss": 2.5888, "step": 40146 }, { "epoch": 2.4922093239803838, "grad_norm": 0.43682202532348197, "learning_rate": 8.476497039143011e-06, "loss": 2.7607, "step": 40147 }, { "epoch": 2.4922714010801417, "grad_norm": 0.41906071565093134, "learning_rate": 8.474485300014102e-06, "loss": 2.6218, "step": 40148 }, { "epoch": 2.4923334781798996, "grad_norm": 0.4937247686337373, "learning_rate": 8.47247377753389e-06, "loss": 2.6043, "step": 40149 }, { "epoch": 2.4923955552796575, "grad_norm": 0.4482189496567901, "learning_rate": 8.47046247171287e-06, "loss": 2.7001, "step": 40150 }, { "epoch": 2.4924576323794154, "grad_norm": 0.401771063764587, "learning_rate": 8.468451382561526e-06, "loss": 2.6323, "step": 40151 }, { "epoch": 2.492519709479173, "grad_norm": 0.4270287533131211, "learning_rate": 8.46644051009034e-06, "loss": 2.7319, "step": 40152 }, { "epoch": 2.4925817865789313, "grad_norm": 0.49965494886039585, "learning_rate": 8.46442985430983e-06, "loss": 2.6972, "step": 40153 }, { "epoch": 2.4926438636786887, "grad_norm": 0.47513293990937877, "learning_rate": 8.462419415230476e-06, "loss": 2.7561, "step": 40154 }, { "epoch": 2.4927059407784466, "grad_norm": 0.45359670816127734, "learning_rate": 8.460409192862762e-06, "loss": 2.5839, "step": 40155 }, { "epoch": 2.4927680178782046, "grad_norm": 0.420779465856782, "learning_rate": 8.458399187217159e-06, "loss": 2.7661, "step": 40156 }, { "epoch": 2.4928300949779625, "grad_norm": 0.4767571348867637, "learning_rate": 8.456389398304188e-06, "loss": 2.6478, "step": 40157 }, { "epoch": 2.4928921720777204, "grad_norm": 0.43374857844578746, "learning_rate": 8.454379826134306e-06, "loss": 2.7753, "step": 40158 }, { "epoch": 2.4929542491774783, "grad_norm": 0.39648837759318356, "learning_rate": 8.452370470718023e-06, "loss": 2.6638, "step": 40159 }, { "epoch": 2.4930163262772362, "grad_norm": 0.42905160494180683, "learning_rate": 8.450361332065814e-06, "loss": 2.7388, "step": 40160 }, { "epoch": 2.493078403376994, "grad_norm": 0.4399298392245429, "learning_rate": 8.448352410188148e-06, "loss": 2.6389, "step": 40161 }, { "epoch": 2.493140480476752, "grad_norm": 0.4183192521488901, "learning_rate": 8.446343705095511e-06, "loss": 2.7512, "step": 40162 }, { "epoch": 2.49320255757651, "grad_norm": 0.44280833816457005, "learning_rate": 8.444335216798393e-06, "loss": 2.5883, "step": 40163 }, { "epoch": 2.493264634676268, "grad_norm": 0.4276380561133443, "learning_rate": 8.442326945307267e-06, "loss": 2.657, "step": 40164 }, { "epoch": 2.493326711776026, "grad_norm": 0.4234763412679097, "learning_rate": 8.44031889063261e-06, "loss": 2.6619, "step": 40165 }, { "epoch": 2.4933887888757837, "grad_norm": 0.43858945960653173, "learning_rate": 8.438311052784886e-06, "loss": 2.692, "step": 40166 }, { "epoch": 2.4934508659755417, "grad_norm": 0.3930076602609042, "learning_rate": 8.436303431774595e-06, "loss": 2.6898, "step": 40167 }, { "epoch": 2.4935129430752996, "grad_norm": 0.46113893988746096, "learning_rate": 8.434296027612199e-06, "loss": 2.7545, "step": 40168 }, { "epoch": 2.4935750201750575, "grad_norm": 0.4289203036981518, "learning_rate": 8.432288840308173e-06, "loss": 2.6974, "step": 40169 }, { "epoch": 2.4936370972748154, "grad_norm": 0.40518772183961926, "learning_rate": 8.43028186987298e-06, "loss": 2.6122, "step": 40170 }, { "epoch": 2.4936991743745733, "grad_norm": 0.40779310377626565, "learning_rate": 8.428275116317092e-06, "loss": 2.6554, "step": 40171 }, { "epoch": 2.4937612514743313, "grad_norm": 0.4093775123053185, "learning_rate": 8.426268579650997e-06, "loss": 2.7326, "step": 40172 }, { "epoch": 2.493823328574089, "grad_norm": 0.42783326699766966, "learning_rate": 8.424262259885146e-06, "loss": 2.6436, "step": 40173 }, { "epoch": 2.493885405673847, "grad_norm": 0.448962926582979, "learning_rate": 8.422256157030017e-06, "loss": 2.7611, "step": 40174 }, { "epoch": 2.493947482773605, "grad_norm": 0.42080799024899945, "learning_rate": 8.420250271096059e-06, "loss": 2.6692, "step": 40175 }, { "epoch": 2.494009559873363, "grad_norm": 0.4107591285422562, "learning_rate": 8.418244602093755e-06, "loss": 2.6634, "step": 40176 }, { "epoch": 2.4940716369731204, "grad_norm": 0.5008412079765925, "learning_rate": 8.416239150033573e-06, "loss": 2.6355, "step": 40177 }, { "epoch": 2.4941337140728788, "grad_norm": 0.4318099574396411, "learning_rate": 8.414233914925957e-06, "loss": 2.6228, "step": 40178 }, { "epoch": 2.4941957911726362, "grad_norm": 0.4433640467191747, "learning_rate": 8.412228896781371e-06, "loss": 2.6651, "step": 40179 }, { "epoch": 2.494257868272394, "grad_norm": 0.440642244258573, "learning_rate": 8.410224095610293e-06, "loss": 2.7265, "step": 40180 }, { "epoch": 2.494319945372152, "grad_norm": 0.46125346200569, "learning_rate": 8.408219511423171e-06, "loss": 2.752, "step": 40181 }, { "epoch": 2.49438202247191, "grad_norm": 0.4155280071750818, "learning_rate": 8.406215144230467e-06, "loss": 2.621, "step": 40182 }, { "epoch": 2.494444099571668, "grad_norm": 0.4118289083528455, "learning_rate": 8.404210994042633e-06, "loss": 2.6902, "step": 40183 }, { "epoch": 2.494506176671426, "grad_norm": 0.4199354302635915, "learning_rate": 8.40220706087012e-06, "loss": 2.704, "step": 40184 }, { "epoch": 2.4945682537711837, "grad_norm": 0.45275375351068714, "learning_rate": 8.4002033447234e-06, "loss": 2.6804, "step": 40185 }, { "epoch": 2.4946303308709417, "grad_norm": 0.436091864679868, "learning_rate": 8.39819984561292e-06, "loss": 2.6759, "step": 40186 }, { "epoch": 2.4946924079706996, "grad_norm": 0.42734042624901813, "learning_rate": 8.39619656354913e-06, "loss": 2.6797, "step": 40187 }, { "epoch": 2.4947544850704575, "grad_norm": 0.45688015323034015, "learning_rate": 8.394193498542469e-06, "loss": 2.6168, "step": 40188 }, { "epoch": 2.4948165621702154, "grad_norm": 0.4656783393416557, "learning_rate": 8.392190650603398e-06, "loss": 2.7796, "step": 40189 }, { "epoch": 2.4948786392699733, "grad_norm": 0.3965825678059232, "learning_rate": 8.390188019742385e-06, "loss": 2.7255, "step": 40190 }, { "epoch": 2.4949407163697312, "grad_norm": 0.4480605079286265, "learning_rate": 8.388185605969862e-06, "loss": 2.6031, "step": 40191 }, { "epoch": 2.495002793469489, "grad_norm": 0.44452881033015074, "learning_rate": 8.386183409296277e-06, "loss": 2.7202, "step": 40192 }, { "epoch": 2.495064870569247, "grad_norm": 0.4297684679906441, "learning_rate": 8.384181429732074e-06, "loss": 2.7167, "step": 40193 }, { "epoch": 2.495126947669005, "grad_norm": 0.4074631879420903, "learning_rate": 8.382179667287687e-06, "loss": 2.7269, "step": 40194 }, { "epoch": 2.495189024768763, "grad_norm": 0.4484394445609754, "learning_rate": 8.380178121973581e-06, "loss": 2.6568, "step": 40195 }, { "epoch": 2.495251101868521, "grad_norm": 0.4463000974396914, "learning_rate": 8.378176793800196e-06, "loss": 2.5682, "step": 40196 }, { "epoch": 2.4953131789682788, "grad_norm": 0.3806997502657064, "learning_rate": 8.376175682777959e-06, "loss": 2.6709, "step": 40197 }, { "epoch": 2.4953752560680367, "grad_norm": 0.45917805582565463, "learning_rate": 8.374174788917311e-06, "loss": 2.6675, "step": 40198 }, { "epoch": 2.4954373331677946, "grad_norm": 0.46130437313374506, "learning_rate": 8.37217411222871e-06, "loss": 2.7313, "step": 40199 }, { "epoch": 2.495499410267552, "grad_norm": 0.4428370251321628, "learning_rate": 8.370173652722579e-06, "loss": 2.5424, "step": 40200 }, { "epoch": 2.4955614873673104, "grad_norm": 0.46546084051208503, "learning_rate": 8.368173410409364e-06, "loss": 2.7047, "step": 40201 }, { "epoch": 2.495623564467068, "grad_norm": 0.45550432017232084, "learning_rate": 8.366173385299474e-06, "loss": 2.6148, "step": 40202 }, { "epoch": 2.495685641566826, "grad_norm": 0.43685727933490637, "learning_rate": 8.364173577403383e-06, "loss": 2.7079, "step": 40203 }, { "epoch": 2.4957477186665837, "grad_norm": 0.47777632609443776, "learning_rate": 8.362173986731508e-06, "loss": 2.6482, "step": 40204 }, { "epoch": 2.4958097957663417, "grad_norm": 0.4233697266904916, "learning_rate": 8.360174613294275e-06, "loss": 2.6582, "step": 40205 }, { "epoch": 2.4958718728660996, "grad_norm": 0.45188621019041786, "learning_rate": 8.358175457102118e-06, "loss": 2.6381, "step": 40206 }, { "epoch": 2.4959339499658575, "grad_norm": 0.47544764832635045, "learning_rate": 8.356176518165459e-06, "loss": 2.7671, "step": 40207 }, { "epoch": 2.4959960270656154, "grad_norm": 0.4642486609879731, "learning_rate": 8.35417779649475e-06, "loss": 2.6618, "step": 40208 }, { "epoch": 2.4960581041653733, "grad_norm": 0.44658687679849385, "learning_rate": 8.352179292100403e-06, "loss": 2.7299, "step": 40209 }, { "epoch": 2.4961201812651312, "grad_norm": 0.45360983199500965, "learning_rate": 8.350181004992846e-06, "loss": 2.7204, "step": 40210 }, { "epoch": 2.496182258364889, "grad_norm": 0.4896416899889874, "learning_rate": 8.348182935182497e-06, "loss": 2.6406, "step": 40211 }, { "epoch": 2.496244335464647, "grad_norm": 0.47321247443651954, "learning_rate": 8.346185082679798e-06, "loss": 2.7338, "step": 40212 }, { "epoch": 2.496306412564405, "grad_norm": 0.4661673286923365, "learning_rate": 8.344187447495161e-06, "loss": 2.5889, "step": 40213 }, { "epoch": 2.496368489664163, "grad_norm": 0.4466013076929643, "learning_rate": 8.342190029639013e-06, "loss": 2.7385, "step": 40214 }, { "epoch": 2.496430566763921, "grad_norm": 0.4866691221159997, "learning_rate": 8.340192829121763e-06, "loss": 2.7689, "step": 40215 }, { "epoch": 2.4964926438636788, "grad_norm": 0.4326674903575, "learning_rate": 8.33819584595385e-06, "loss": 2.6843, "step": 40216 }, { "epoch": 2.4965547209634367, "grad_norm": 0.4701542483509339, "learning_rate": 8.336199080145685e-06, "loss": 2.7682, "step": 40217 }, { "epoch": 2.4966167980631946, "grad_norm": 0.44727382069370886, "learning_rate": 8.33420253170768e-06, "loss": 2.6367, "step": 40218 }, { "epoch": 2.4966788751629525, "grad_norm": 0.45058402865914554, "learning_rate": 8.332206200650256e-06, "loss": 2.7471, "step": 40219 }, { "epoch": 2.4967409522627104, "grad_norm": 0.44311350252007464, "learning_rate": 8.33021008698382e-06, "loss": 2.6632, "step": 40220 }, { "epoch": 2.4968030293624683, "grad_norm": 0.4611599757051838, "learning_rate": 8.328214190718797e-06, "loss": 2.6951, "step": 40221 }, { "epoch": 2.4968651064622263, "grad_norm": 0.4350056728453533, "learning_rate": 8.32621851186559e-06, "loss": 2.6074, "step": 40222 }, { "epoch": 2.496927183561984, "grad_norm": 0.45704434041895886, "learning_rate": 8.324223050434632e-06, "loss": 2.7891, "step": 40223 }, { "epoch": 2.496989260661742, "grad_norm": 0.4652289709743014, "learning_rate": 8.322227806436317e-06, "loss": 2.6837, "step": 40224 }, { "epoch": 2.4970513377614996, "grad_norm": 0.4495927331994739, "learning_rate": 8.320232779881044e-06, "loss": 2.719, "step": 40225 }, { "epoch": 2.497113414861258, "grad_norm": 0.4569755422435244, "learning_rate": 8.318237970779247e-06, "loss": 2.7103, "step": 40226 }, { "epoch": 2.4971754919610154, "grad_norm": 0.44384466827024704, "learning_rate": 8.316243379141325e-06, "loss": 2.647, "step": 40227 }, { "epoch": 2.4972375690607733, "grad_norm": 0.453533120355907, "learning_rate": 8.314249004977675e-06, "loss": 2.6157, "step": 40228 }, { "epoch": 2.4972996461605312, "grad_norm": 0.4674339256920147, "learning_rate": 8.312254848298712e-06, "loss": 2.6152, "step": 40229 }, { "epoch": 2.497361723260289, "grad_norm": 0.46573678153420883, "learning_rate": 8.310260909114819e-06, "loss": 2.6391, "step": 40230 }, { "epoch": 2.497423800360047, "grad_norm": 0.4429206189577044, "learning_rate": 8.30826718743643e-06, "loss": 2.6942, "step": 40231 }, { "epoch": 2.497485877459805, "grad_norm": 0.466132932800569, "learning_rate": 8.306273683273935e-06, "loss": 2.616, "step": 40232 }, { "epoch": 2.497547954559563, "grad_norm": 0.4522536713507548, "learning_rate": 8.304280396637726e-06, "loss": 2.5784, "step": 40233 }, { "epoch": 2.497610031659321, "grad_norm": 0.4089806010920172, "learning_rate": 8.3022873275382e-06, "loss": 2.6912, "step": 40234 }, { "epoch": 2.4976721087590787, "grad_norm": 0.43407360364802167, "learning_rate": 8.300294475985776e-06, "loss": 2.7045, "step": 40235 }, { "epoch": 2.4977341858588367, "grad_norm": 0.4605105896197107, "learning_rate": 8.298301841990835e-06, "loss": 2.668, "step": 40236 }, { "epoch": 2.4977962629585946, "grad_norm": 0.44845328419516634, "learning_rate": 8.296309425563781e-06, "loss": 2.6934, "step": 40237 }, { "epoch": 2.4978583400583525, "grad_norm": 0.4381619962670919, "learning_rate": 8.294317226714986e-06, "loss": 2.6965, "step": 40238 }, { "epoch": 2.4979204171581104, "grad_norm": 0.4435774145178769, "learning_rate": 8.29232524545488e-06, "loss": 2.6638, "step": 40239 }, { "epoch": 2.4979824942578683, "grad_norm": 0.4389236651530715, "learning_rate": 8.290333481793837e-06, "loss": 2.6397, "step": 40240 }, { "epoch": 2.4980445713576263, "grad_norm": 0.4407694282285164, "learning_rate": 8.288341935742249e-06, "loss": 2.6847, "step": 40241 }, { "epoch": 2.498106648457384, "grad_norm": 0.4714244681211757, "learning_rate": 8.286350607310506e-06, "loss": 2.6423, "step": 40242 }, { "epoch": 2.498168725557142, "grad_norm": 0.43389085093266405, "learning_rate": 8.284359496508987e-06, "loss": 2.6353, "step": 40243 }, { "epoch": 2.4982308026569, "grad_norm": 0.42905056302663047, "learning_rate": 8.282368603348107e-06, "loss": 2.7079, "step": 40244 }, { "epoch": 2.498292879756658, "grad_norm": 0.44338876568009555, "learning_rate": 8.280377927838234e-06, "loss": 2.6318, "step": 40245 }, { "epoch": 2.498354956856416, "grad_norm": 0.45880746881646606, "learning_rate": 8.278387469989758e-06, "loss": 2.683, "step": 40246 }, { "epoch": 2.4984170339561738, "grad_norm": 0.42766579204446603, "learning_rate": 8.276397229813048e-06, "loss": 2.6185, "step": 40247 }, { "epoch": 2.4984791110559312, "grad_norm": 0.43671546143196116, "learning_rate": 8.274407207318518e-06, "loss": 2.6796, "step": 40248 }, { "epoch": 2.4985411881556896, "grad_norm": 0.46957385461758244, "learning_rate": 8.27241740251653e-06, "loss": 2.6564, "step": 40249 }, { "epoch": 2.498603265255447, "grad_norm": 0.41177385240151365, "learning_rate": 8.270427815417475e-06, "loss": 2.7074, "step": 40250 }, { "epoch": 2.498665342355205, "grad_norm": 0.4766905878251346, "learning_rate": 8.268438446031718e-06, "loss": 2.6992, "step": 40251 }, { "epoch": 2.498727419454963, "grad_norm": 0.46399995208916744, "learning_rate": 8.266449294369655e-06, "loss": 2.6438, "step": 40252 }, { "epoch": 2.498789496554721, "grad_norm": 0.44013806688412455, "learning_rate": 8.264460360441661e-06, "loss": 2.6179, "step": 40253 }, { "epoch": 2.4988515736544787, "grad_norm": 0.4485079573730209, "learning_rate": 8.262471644258096e-06, "loss": 2.6954, "step": 40254 }, { "epoch": 2.4989136507542367, "grad_norm": 0.4584105292969645, "learning_rate": 8.260483145829356e-06, "loss": 2.7039, "step": 40255 }, { "epoch": 2.4989757278539946, "grad_norm": 0.4579875754499351, "learning_rate": 8.258494865165816e-06, "loss": 2.709, "step": 40256 }, { "epoch": 2.4990378049537525, "grad_norm": 0.46081694950080504, "learning_rate": 8.256506802277825e-06, "loss": 2.6269, "step": 40257 }, { "epoch": 2.4990998820535104, "grad_norm": 0.46453866644108965, "learning_rate": 8.254518957175783e-06, "loss": 2.6925, "step": 40258 }, { "epoch": 2.4991619591532683, "grad_norm": 0.44131517736694903, "learning_rate": 8.25253132987005e-06, "loss": 2.683, "step": 40259 }, { "epoch": 2.4992240362530262, "grad_norm": 0.4304394125244404, "learning_rate": 8.250543920370995e-06, "loss": 2.6755, "step": 40260 }, { "epoch": 2.499286113352784, "grad_norm": 0.45386896226481604, "learning_rate": 8.248556728688977e-06, "loss": 2.7055, "step": 40261 }, { "epoch": 2.499348190452542, "grad_norm": 0.45835417642306275, "learning_rate": 8.246569754834389e-06, "loss": 2.7677, "step": 40262 }, { "epoch": 2.4994102675523, "grad_norm": 0.42843499133532376, "learning_rate": 8.24458299881758e-06, "loss": 2.677, "step": 40263 }, { "epoch": 2.499472344652058, "grad_norm": 0.490286714107972, "learning_rate": 8.242596460648916e-06, "loss": 2.6915, "step": 40264 }, { "epoch": 2.499534421751816, "grad_norm": 0.49000137993073295, "learning_rate": 8.24061014033875e-06, "loss": 2.7151, "step": 40265 }, { "epoch": 2.4995964988515738, "grad_norm": 0.42869062093156796, "learning_rate": 8.238624037897474e-06, "loss": 2.6603, "step": 40266 }, { "epoch": 2.4996585759513317, "grad_norm": 0.4676418079423841, "learning_rate": 8.236638153335429e-06, "loss": 2.6957, "step": 40267 }, { "epoch": 2.4997206530510896, "grad_norm": 0.4340737237982831, "learning_rate": 8.234652486662991e-06, "loss": 2.6876, "step": 40268 }, { "epoch": 2.4997827301508475, "grad_norm": 0.4779632026572422, "learning_rate": 8.2326670378905e-06, "loss": 2.6831, "step": 40269 }, { "epoch": 2.4998448072506054, "grad_norm": 0.4593853112445375, "learning_rate": 8.230681807028318e-06, "loss": 2.6969, "step": 40270 }, { "epoch": 2.4999068843503633, "grad_norm": 0.4568812295251556, "learning_rate": 8.22869679408682e-06, "loss": 2.7565, "step": 40271 }, { "epoch": 2.4999689614501213, "grad_norm": 0.4202772898621104, "learning_rate": 8.226711999076353e-06, "loss": 2.629, "step": 40272 }, { "epoch": 2.5000310385498787, "grad_norm": 0.4447120667492114, "learning_rate": 8.224727422007267e-06, "loss": 2.6778, "step": 40273 }, { "epoch": 2.500093115649637, "grad_norm": 0.4108868585864995, "learning_rate": 8.222743062889909e-06, "loss": 2.6031, "step": 40274 }, { "epoch": 2.5001551927493946, "grad_norm": 0.46817114374680197, "learning_rate": 8.22075892173465e-06, "loss": 2.6784, "step": 40275 }, { "epoch": 2.5002172698491525, "grad_norm": 0.43470391780692813, "learning_rate": 8.218774998551836e-06, "loss": 2.7055, "step": 40276 }, { "epoch": 2.5002793469489104, "grad_norm": 0.4379660474222274, "learning_rate": 8.216791293351822e-06, "loss": 2.6005, "step": 40277 }, { "epoch": 2.5003414240486683, "grad_norm": 0.42821139904784555, "learning_rate": 8.214807806144941e-06, "loss": 2.5906, "step": 40278 }, { "epoch": 2.5004035011484262, "grad_norm": 0.4008095945035553, "learning_rate": 8.212824536941549e-06, "loss": 2.6149, "step": 40279 }, { "epoch": 2.500465578248184, "grad_norm": 0.4065435889241869, "learning_rate": 8.210841485752002e-06, "loss": 2.6339, "step": 40280 }, { "epoch": 2.500527655347942, "grad_norm": 0.44107446607434053, "learning_rate": 8.208858652586643e-06, "loss": 2.6444, "step": 40281 }, { "epoch": 2.5005897324477, "grad_norm": 0.42265969167496203, "learning_rate": 8.206876037455807e-06, "loss": 2.6139, "step": 40282 }, { "epoch": 2.500651809547458, "grad_norm": 0.46618169685656063, "learning_rate": 8.204893640369838e-06, "loss": 2.6941, "step": 40283 }, { "epoch": 2.500713886647216, "grad_norm": 0.4223069875610597, "learning_rate": 8.202911461339096e-06, "loss": 2.6952, "step": 40284 }, { "epoch": 2.5007759637469738, "grad_norm": 0.4578660692765018, "learning_rate": 8.200929500373911e-06, "loss": 2.7335, "step": 40285 }, { "epoch": 2.5008380408467317, "grad_norm": 0.434646136923254, "learning_rate": 8.198947757484626e-06, "loss": 2.7147, "step": 40286 }, { "epoch": 2.5009001179464896, "grad_norm": 0.43671745750495305, "learning_rate": 8.196966232681563e-06, "loss": 2.6655, "step": 40287 }, { "epoch": 2.5009621950462475, "grad_norm": 0.42336604863320354, "learning_rate": 8.194984925975074e-06, "loss": 2.67, "step": 40288 }, { "epoch": 2.5010242721460054, "grad_norm": 0.4639174584704664, "learning_rate": 8.193003837375512e-06, "loss": 2.714, "step": 40289 }, { "epoch": 2.5010863492457633, "grad_norm": 0.4537705230012442, "learning_rate": 8.191022966893197e-06, "loss": 2.67, "step": 40290 }, { "epoch": 2.5011484263455213, "grad_norm": 0.4076636733420629, "learning_rate": 8.189042314538464e-06, "loss": 2.5957, "step": 40291 }, { "epoch": 2.501210503445279, "grad_norm": 0.4010728373943771, "learning_rate": 8.18706188032165e-06, "loss": 2.6263, "step": 40292 }, { "epoch": 2.501272580545037, "grad_norm": 0.46553403426640694, "learning_rate": 8.18508166425307e-06, "loss": 2.7293, "step": 40293 }, { "epoch": 2.5013346576447946, "grad_norm": 0.4690201616589822, "learning_rate": 8.183101666343085e-06, "loss": 2.7617, "step": 40294 }, { "epoch": 2.501396734744553, "grad_norm": 0.4403843317128653, "learning_rate": 8.181121886602006e-06, "loss": 2.6669, "step": 40295 }, { "epoch": 2.5014588118443104, "grad_norm": 0.42931693310516766, "learning_rate": 8.179142325040168e-06, "loss": 2.656, "step": 40296 }, { "epoch": 2.5015208889440688, "grad_norm": 0.43307421032805454, "learning_rate": 8.177162981667885e-06, "loss": 2.6029, "step": 40297 }, { "epoch": 2.5015829660438262, "grad_norm": 0.45215117286859885, "learning_rate": 8.175183856495511e-06, "loss": 2.6141, "step": 40298 }, { "epoch": 2.5016450431435846, "grad_norm": 0.4506927687681239, "learning_rate": 8.173204949533347e-06, "loss": 2.7402, "step": 40299 }, { "epoch": 2.501707120243342, "grad_norm": 0.3990094063374307, "learning_rate": 8.171226260791736e-06, "loss": 2.6361, "step": 40300 }, { "epoch": 2.5017691973431, "grad_norm": 0.4748494361368788, "learning_rate": 8.169247790280976e-06, "loss": 2.6385, "step": 40301 }, { "epoch": 2.501831274442858, "grad_norm": 0.45289877473093104, "learning_rate": 8.167269538011419e-06, "loss": 2.6287, "step": 40302 }, { "epoch": 2.501893351542616, "grad_norm": 0.414604282408973, "learning_rate": 8.165291503993367e-06, "loss": 2.7173, "step": 40303 }, { "epoch": 2.5019554286423737, "grad_norm": 0.443515094852853, "learning_rate": 8.163313688237145e-06, "loss": 2.6507, "step": 40304 }, { "epoch": 2.5020175057421317, "grad_norm": 0.473585334836476, "learning_rate": 8.16133609075308e-06, "loss": 2.6958, "step": 40305 }, { "epoch": 2.5020795828418896, "grad_norm": 0.416646988721939, "learning_rate": 8.159358711551463e-06, "loss": 2.6414, "step": 40306 }, { "epoch": 2.5021416599416475, "grad_norm": 0.4465129299160574, "learning_rate": 8.15738155064264e-06, "loss": 2.7202, "step": 40307 }, { "epoch": 2.5022037370414054, "grad_norm": 0.5139645804974061, "learning_rate": 8.155404608036915e-06, "loss": 2.6748, "step": 40308 }, { "epoch": 2.5022658141411633, "grad_norm": 0.4587932756992947, "learning_rate": 8.153427883744608e-06, "loss": 2.6044, "step": 40309 }, { "epoch": 2.5023278912409213, "grad_norm": 0.4314591039039488, "learning_rate": 8.151451377776009e-06, "loss": 2.7019, "step": 40310 }, { "epoch": 2.502389968340679, "grad_norm": 0.45089506765198745, "learning_rate": 8.149475090141462e-06, "loss": 2.5727, "step": 40311 }, { "epoch": 2.502452045440437, "grad_norm": 0.4441466546634908, "learning_rate": 8.147499020851263e-06, "loss": 2.5596, "step": 40312 }, { "epoch": 2.502514122540195, "grad_norm": 0.46242163032825595, "learning_rate": 8.14552316991572e-06, "loss": 2.6851, "step": 40313 }, { "epoch": 2.502576199639953, "grad_norm": 0.4331132098928575, "learning_rate": 8.143547537345141e-06, "loss": 2.626, "step": 40314 }, { "epoch": 2.502638276739711, "grad_norm": 0.45454304881434604, "learning_rate": 8.141572123149827e-06, "loss": 2.7029, "step": 40315 }, { "epoch": 2.5027003538394688, "grad_norm": 0.40055408397635867, "learning_rate": 8.139596927340104e-06, "loss": 2.6728, "step": 40316 }, { "epoch": 2.5027624309392267, "grad_norm": 0.4338012576804302, "learning_rate": 8.137621949926267e-06, "loss": 2.6716, "step": 40317 }, { "epoch": 2.5028245080389846, "grad_norm": 0.4471704306766929, "learning_rate": 8.135647190918617e-06, "loss": 2.6255, "step": 40318 }, { "epoch": 2.502886585138742, "grad_norm": 0.43159255023281595, "learning_rate": 8.133672650327445e-06, "loss": 2.6428, "step": 40319 }, { "epoch": 2.5029486622385004, "grad_norm": 0.4220965474813321, "learning_rate": 8.131698328163068e-06, "loss": 2.5958, "step": 40320 }, { "epoch": 2.503010739338258, "grad_norm": 0.4284205745968198, "learning_rate": 8.129724224435798e-06, "loss": 2.6798, "step": 40321 }, { "epoch": 2.5030728164380163, "grad_norm": 0.43234542336259, "learning_rate": 8.127750339155921e-06, "loss": 2.6664, "step": 40322 }, { "epoch": 2.5031348935377737, "grad_norm": 0.3964738515575423, "learning_rate": 8.125776672333729e-06, "loss": 2.6557, "step": 40323 }, { "epoch": 2.5031969706375317, "grad_norm": 0.43289053657557, "learning_rate": 8.12380322397952e-06, "loss": 2.7149, "step": 40324 }, { "epoch": 2.5032590477372896, "grad_norm": 0.4808865258293481, "learning_rate": 8.121829994103608e-06, "loss": 2.6902, "step": 40325 }, { "epoch": 2.5033211248370475, "grad_norm": 0.410310171401076, "learning_rate": 8.119856982716273e-06, "loss": 2.6618, "step": 40326 }, { "epoch": 2.5033832019368054, "grad_norm": 0.4045084708273449, "learning_rate": 8.117884189827812e-06, "loss": 2.6236, "step": 40327 }, { "epoch": 2.5034452790365633, "grad_norm": 0.4372377631501903, "learning_rate": 8.115911615448513e-06, "loss": 2.59, "step": 40328 }, { "epoch": 2.5035073561363212, "grad_norm": 0.48541033751600576, "learning_rate": 8.11393925958866e-06, "loss": 2.6651, "step": 40329 }, { "epoch": 2.503569433236079, "grad_norm": 0.44353150710724215, "learning_rate": 8.111967122258568e-06, "loss": 2.7218, "step": 40330 }, { "epoch": 2.503631510335837, "grad_norm": 0.40472319603867335, "learning_rate": 8.109995203468512e-06, "loss": 2.7225, "step": 40331 }, { "epoch": 2.503693587435595, "grad_norm": 0.47676724585139885, "learning_rate": 8.108023503228778e-06, "loss": 2.6879, "step": 40332 }, { "epoch": 2.503755664535353, "grad_norm": 0.46766728289711545, "learning_rate": 8.106052021549649e-06, "loss": 2.7022, "step": 40333 }, { "epoch": 2.503817741635111, "grad_norm": 0.4189416133832904, "learning_rate": 8.104080758441424e-06, "loss": 2.686, "step": 40334 }, { "epoch": 2.5038798187348688, "grad_norm": 0.4048598051549933, "learning_rate": 8.10210971391438e-06, "loss": 2.6322, "step": 40335 }, { "epoch": 2.5039418958346267, "grad_norm": 0.4478934433557457, "learning_rate": 8.100138887978802e-06, "loss": 2.6096, "step": 40336 }, { "epoch": 2.5040039729343846, "grad_norm": 0.44839551376877046, "learning_rate": 8.098168280644957e-06, "loss": 2.7588, "step": 40337 }, { "epoch": 2.5040660500341425, "grad_norm": 0.4528857783541403, "learning_rate": 8.096197891923151e-06, "loss": 2.6633, "step": 40338 }, { "epoch": 2.5041281271339004, "grad_norm": 0.41415103829458755, "learning_rate": 8.094227721823661e-06, "loss": 2.6794, "step": 40339 }, { "epoch": 2.5041902042336583, "grad_norm": 0.4539310914692446, "learning_rate": 8.092257770356748e-06, "loss": 2.6959, "step": 40340 }, { "epoch": 2.5042522813334163, "grad_norm": 0.42283258549700925, "learning_rate": 8.090288037532706e-06, "loss": 2.6536, "step": 40341 }, { "epoch": 2.5043143584331737, "grad_norm": 0.43046357555014103, "learning_rate": 8.088318523361793e-06, "loss": 2.683, "step": 40342 }, { "epoch": 2.504376435532932, "grad_norm": 0.44876971815136657, "learning_rate": 8.086349227854306e-06, "loss": 2.6516, "step": 40343 }, { "epoch": 2.5044385126326896, "grad_norm": 0.449334734994917, "learning_rate": 8.08438015102051e-06, "loss": 2.6868, "step": 40344 }, { "epoch": 2.504500589732448, "grad_norm": 0.42056276960700173, "learning_rate": 8.082411292870678e-06, "loss": 2.6627, "step": 40345 }, { "epoch": 2.5045626668322054, "grad_norm": 0.45640576590799986, "learning_rate": 8.080442653415071e-06, "loss": 2.6056, "step": 40346 }, { "epoch": 2.5046247439319638, "grad_norm": 0.4509666273262589, "learning_rate": 8.078474232663985e-06, "loss": 2.691, "step": 40347 }, { "epoch": 2.5046868210317212, "grad_norm": 0.42809882814923733, "learning_rate": 8.076506030627672e-06, "loss": 2.6191, "step": 40348 }, { "epoch": 2.504748898131479, "grad_norm": 0.4616031689247495, "learning_rate": 8.07453804731641e-06, "loss": 2.6288, "step": 40349 }, { "epoch": 2.504810975231237, "grad_norm": 0.46756604401319063, "learning_rate": 8.072570282740444e-06, "loss": 2.7811, "step": 40350 }, { "epoch": 2.504873052330995, "grad_norm": 0.4562523880987506, "learning_rate": 8.070602736910071e-06, "loss": 2.6952, "step": 40351 }, { "epoch": 2.504935129430753, "grad_norm": 0.4373156976783278, "learning_rate": 8.06863540983554e-06, "loss": 2.6434, "step": 40352 }, { "epoch": 2.504997206530511, "grad_norm": 0.43598144815614537, "learning_rate": 8.066668301527108e-06, "loss": 2.5879, "step": 40353 }, { "epoch": 2.5050592836302688, "grad_norm": 0.42040503451143924, "learning_rate": 8.064701411995057e-06, "loss": 2.5896, "step": 40354 }, { "epoch": 2.5051213607300267, "grad_norm": 0.46154647943978305, "learning_rate": 8.06273474124964e-06, "loss": 2.6767, "step": 40355 }, { "epoch": 2.5051834378297846, "grad_norm": 0.451202639236873, "learning_rate": 8.060768289301108e-06, "loss": 2.6074, "step": 40356 }, { "epoch": 2.5052455149295425, "grad_norm": 0.41744286440312317, "learning_rate": 8.058802056159737e-06, "loss": 2.5951, "step": 40357 }, { "epoch": 2.5053075920293004, "grad_norm": 0.4315891494091465, "learning_rate": 8.056836041835774e-06, "loss": 2.6746, "step": 40358 }, { "epoch": 2.5053696691290583, "grad_norm": 0.4021994137425847, "learning_rate": 8.054870246339486e-06, "loss": 2.6393, "step": 40359 }, { "epoch": 2.5054317462288163, "grad_norm": 0.4103525509429749, "learning_rate": 8.052904669681105e-06, "loss": 2.7143, "step": 40360 }, { "epoch": 2.505493823328574, "grad_norm": 0.4626824598995019, "learning_rate": 8.05093931187092e-06, "loss": 2.7209, "step": 40361 }, { "epoch": 2.505555900428332, "grad_norm": 0.4228871178600509, "learning_rate": 8.048974172919167e-06, "loss": 2.7563, "step": 40362 }, { "epoch": 2.50561797752809, "grad_norm": 0.42099806807906087, "learning_rate": 8.047009252836096e-06, "loss": 2.6185, "step": 40363 }, { "epoch": 2.505680054627848, "grad_norm": 0.41626143021684214, "learning_rate": 8.045044551631964e-06, "loss": 2.6786, "step": 40364 }, { "epoch": 2.505742131727606, "grad_norm": 0.43731522063995115, "learning_rate": 8.04308006931701e-06, "loss": 2.5746, "step": 40365 }, { "epoch": 2.5058042088273638, "grad_norm": 0.46743673100454786, "learning_rate": 8.0411158059015e-06, "loss": 2.5852, "step": 40366 }, { "epoch": 2.5058662859271212, "grad_norm": 0.4275832409990878, "learning_rate": 8.03915176139568e-06, "loss": 2.6781, "step": 40367 }, { "epoch": 2.5059283630268796, "grad_norm": 0.4111801627612398, "learning_rate": 8.037187935809787e-06, "loss": 2.6468, "step": 40368 }, { "epoch": 2.505990440126637, "grad_norm": 0.4413249860919675, "learning_rate": 8.035224329154062e-06, "loss": 2.7142, "step": 40369 }, { "epoch": 2.5060525172263954, "grad_norm": 0.42731674918003304, "learning_rate": 8.03326094143877e-06, "loss": 2.6652, "step": 40370 }, { "epoch": 2.506114594326153, "grad_norm": 0.40898191274577617, "learning_rate": 8.031297772674146e-06, "loss": 2.6459, "step": 40371 }, { "epoch": 2.506176671425911, "grad_norm": 0.4529698204825293, "learning_rate": 8.029334822870427e-06, "loss": 2.725, "step": 40372 }, { "epoch": 2.5062387485256687, "grad_norm": 0.43774596180406683, "learning_rate": 8.027372092037843e-06, "loss": 2.6505, "step": 40373 }, { "epoch": 2.5063008256254267, "grad_norm": 0.42086944126166476, "learning_rate": 8.025409580186666e-06, "loss": 2.6846, "step": 40374 }, { "epoch": 2.5063629027251846, "grad_norm": 0.40576919199904765, "learning_rate": 8.023447287327113e-06, "loss": 2.7029, "step": 40375 }, { "epoch": 2.5064249798249425, "grad_norm": 0.46095573259921246, "learning_rate": 8.021485213469426e-06, "loss": 2.7026, "step": 40376 }, { "epoch": 2.5064870569247004, "grad_norm": 0.42663181286750096, "learning_rate": 8.019523358623843e-06, "loss": 2.6899, "step": 40377 }, { "epoch": 2.5065491340244583, "grad_norm": 0.4457971712699538, "learning_rate": 8.017561722800582e-06, "loss": 2.7349, "step": 40378 }, { "epoch": 2.5066112111242163, "grad_norm": 0.4142937644166976, "learning_rate": 8.015600306009907e-06, "loss": 2.6772, "step": 40379 }, { "epoch": 2.506673288223974, "grad_norm": 0.44709351405663245, "learning_rate": 8.01363910826204e-06, "loss": 2.7004, "step": 40380 }, { "epoch": 2.506735365323732, "grad_norm": 0.4414063175167606, "learning_rate": 8.011678129567208e-06, "loss": 2.6852, "step": 40381 }, { "epoch": 2.50679744242349, "grad_norm": 0.4552884414013773, "learning_rate": 8.009717369935631e-06, "loss": 2.6011, "step": 40382 }, { "epoch": 2.506859519523248, "grad_norm": 0.4212618716913068, "learning_rate": 8.007756829377566e-06, "loss": 2.6338, "step": 40383 }, { "epoch": 2.506921596623006, "grad_norm": 0.4293373762191144, "learning_rate": 8.005796507903223e-06, "loss": 2.7494, "step": 40384 }, { "epoch": 2.5069836737227638, "grad_norm": 0.4216349943363577, "learning_rate": 8.00383640552283e-06, "loss": 2.5616, "step": 40385 }, { "epoch": 2.5070457508225217, "grad_norm": 0.41646066778857194, "learning_rate": 8.001876522246627e-06, "loss": 2.6408, "step": 40386 }, { "epoch": 2.5071078279222796, "grad_norm": 0.42740146100830323, "learning_rate": 7.999916858084811e-06, "loss": 2.6411, "step": 40387 }, { "epoch": 2.5071699050220375, "grad_norm": 0.4488357569856552, "learning_rate": 7.99795741304764e-06, "loss": 2.7263, "step": 40388 }, { "epoch": 2.5072319821217954, "grad_norm": 0.4413194317804764, "learning_rate": 7.995998187145327e-06, "loss": 2.6674, "step": 40389 }, { "epoch": 2.507294059221553, "grad_norm": 0.4347419485549809, "learning_rate": 7.994039180388085e-06, "loss": 2.683, "step": 40390 }, { "epoch": 2.5073561363213113, "grad_norm": 0.443395739168989, "learning_rate": 7.992080392786133e-06, "loss": 2.7196, "step": 40391 }, { "epoch": 2.5074182134210687, "grad_norm": 0.4586652579761837, "learning_rate": 7.990121824349689e-06, "loss": 2.6803, "step": 40392 }, { "epoch": 2.507480290520827, "grad_norm": 0.45130470929608096, "learning_rate": 7.988163475088989e-06, "loss": 2.6162, "step": 40393 }, { "epoch": 2.5075423676205846, "grad_norm": 0.43516600852995924, "learning_rate": 7.986205345014241e-06, "loss": 2.648, "step": 40394 }, { "epoch": 2.507604444720343, "grad_norm": 0.46204501789776714, "learning_rate": 7.984247434135655e-06, "loss": 2.7775, "step": 40395 }, { "epoch": 2.5076665218201004, "grad_norm": 0.44980332858451405, "learning_rate": 7.98228974246344e-06, "loss": 2.6569, "step": 40396 }, { "epoch": 2.5077285989198583, "grad_norm": 0.4452413870263816, "learning_rate": 7.980332270007828e-06, "loss": 2.7269, "step": 40397 }, { "epoch": 2.5077906760196162, "grad_norm": 0.4126863102217651, "learning_rate": 7.978375016779028e-06, "loss": 2.67, "step": 40398 }, { "epoch": 2.507852753119374, "grad_norm": 0.4226361578433395, "learning_rate": 7.976417982787243e-06, "loss": 2.6386, "step": 40399 }, { "epoch": 2.507914830219132, "grad_norm": 0.428148774370985, "learning_rate": 7.974461168042675e-06, "loss": 2.648, "step": 40400 }, { "epoch": 2.50797690731889, "grad_norm": 0.43818484927414775, "learning_rate": 7.972504572555556e-06, "loss": 2.7231, "step": 40401 }, { "epoch": 2.508038984418648, "grad_norm": 0.4551024369061195, "learning_rate": 7.970548196336087e-06, "loss": 2.7135, "step": 40402 }, { "epoch": 2.508101061518406, "grad_norm": 0.4593919445986933, "learning_rate": 7.968592039394467e-06, "loss": 2.6257, "step": 40403 }, { "epoch": 2.5081631386181638, "grad_norm": 0.44204768271997913, "learning_rate": 7.96663610174091e-06, "loss": 2.6664, "step": 40404 }, { "epoch": 2.5082252157179217, "grad_norm": 0.41243792269066964, "learning_rate": 7.964680383385603e-06, "loss": 2.7016, "step": 40405 }, { "epoch": 2.5082872928176796, "grad_norm": 0.41791674923285865, "learning_rate": 7.962724884338779e-06, "loss": 2.6095, "step": 40406 }, { "epoch": 2.5083493699174375, "grad_norm": 0.4841219794939775, "learning_rate": 7.960769604610619e-06, "loss": 2.6387, "step": 40407 }, { "epoch": 2.5084114470171954, "grad_norm": 0.4587998364167225, "learning_rate": 7.958814544211329e-06, "loss": 2.6966, "step": 40408 }, { "epoch": 2.5084735241169533, "grad_norm": 0.4193559248291367, "learning_rate": 7.956859703151099e-06, "loss": 2.5848, "step": 40409 }, { "epoch": 2.5085356012167113, "grad_norm": 0.43360414578459094, "learning_rate": 7.95490508144015e-06, "loss": 2.6708, "step": 40410 }, { "epoch": 2.508597678316469, "grad_norm": 0.43815422529298687, "learning_rate": 7.952950679088667e-06, "loss": 2.6047, "step": 40411 }, { "epoch": 2.508659755416227, "grad_norm": 0.40455339190671574, "learning_rate": 7.95099649610685e-06, "loss": 2.725, "step": 40412 }, { "epoch": 2.508721832515985, "grad_norm": 0.4187812750794775, "learning_rate": 7.949042532504892e-06, "loss": 2.6105, "step": 40413 }, { "epoch": 2.508783909615743, "grad_norm": 0.42639864335032174, "learning_rate": 7.947088788292972e-06, "loss": 2.5925, "step": 40414 }, { "epoch": 2.5088459867155004, "grad_norm": 0.43948030175840574, "learning_rate": 7.945135263481313e-06, "loss": 2.659, "step": 40415 }, { "epoch": 2.5089080638152588, "grad_norm": 0.42074075741286227, "learning_rate": 7.943181958080087e-06, "loss": 2.638, "step": 40416 }, { "epoch": 2.5089701409150162, "grad_norm": 0.45140096289633835, "learning_rate": 7.9412288720995e-06, "loss": 2.6233, "step": 40417 }, { "epoch": 2.5090322180147746, "grad_norm": 0.44169401393052904, "learning_rate": 7.939276005549712e-06, "loss": 2.6404, "step": 40418 }, { "epoch": 2.509094295114532, "grad_norm": 0.47823169614798305, "learning_rate": 7.937323358440935e-06, "loss": 2.7453, "step": 40419 }, { "epoch": 2.50915637221429, "grad_norm": 0.3947630286419175, "learning_rate": 7.935370930783365e-06, "loss": 2.6651, "step": 40420 }, { "epoch": 2.509218449314048, "grad_norm": 0.41711529381913337, "learning_rate": 7.933418722587177e-06, "loss": 2.672, "step": 40421 }, { "epoch": 2.509280526413806, "grad_norm": 0.46477843074114095, "learning_rate": 7.931466733862558e-06, "loss": 2.6809, "step": 40422 }, { "epoch": 2.5093426035135638, "grad_norm": 0.44246556600657266, "learning_rate": 7.929514964619671e-06, "loss": 2.6721, "step": 40423 }, { "epoch": 2.5094046806133217, "grad_norm": 0.4292724165053758, "learning_rate": 7.927563414868732e-06, "loss": 2.6046, "step": 40424 }, { "epoch": 2.5094667577130796, "grad_norm": 0.45542467208287624, "learning_rate": 7.92561208461991e-06, "loss": 2.7054, "step": 40425 }, { "epoch": 2.5095288348128375, "grad_norm": 0.4488334330086374, "learning_rate": 7.923660973883384e-06, "loss": 2.6671, "step": 40426 }, { "epoch": 2.5095909119125954, "grad_norm": 0.4426807806379433, "learning_rate": 7.921710082669331e-06, "loss": 2.6775, "step": 40427 }, { "epoch": 2.5096529890123533, "grad_norm": 0.4402919985436394, "learning_rate": 7.91975941098792e-06, "loss": 2.6747, "step": 40428 }, { "epoch": 2.5097150661121113, "grad_norm": 0.4338939418015492, "learning_rate": 7.917808958849354e-06, "loss": 2.6715, "step": 40429 }, { "epoch": 2.509777143211869, "grad_norm": 0.4714944922144726, "learning_rate": 7.915858726263791e-06, "loss": 2.6768, "step": 40430 }, { "epoch": 2.509839220311627, "grad_norm": 0.43357720214609163, "learning_rate": 7.913908713241407e-06, "loss": 2.6368, "step": 40431 }, { "epoch": 2.509901297411385, "grad_norm": 0.40621474920069733, "learning_rate": 7.911958919792373e-06, "loss": 2.6601, "step": 40432 }, { "epoch": 2.509963374511143, "grad_norm": 0.4108149371113315, "learning_rate": 7.910009345926872e-06, "loss": 2.6256, "step": 40433 }, { "epoch": 2.510025451610901, "grad_norm": 0.4914710567660604, "learning_rate": 7.908059991655076e-06, "loss": 2.7021, "step": 40434 }, { "epoch": 2.5100875287106588, "grad_norm": 0.44857501470173206, "learning_rate": 7.906110856987142e-06, "loss": 2.7092, "step": 40435 }, { "epoch": 2.5101496058104167, "grad_norm": 0.41152027927340074, "learning_rate": 7.904161941933241e-06, "loss": 2.7233, "step": 40436 }, { "epoch": 2.5102116829101746, "grad_norm": 0.4300503499194521, "learning_rate": 7.902213246503553e-06, "loss": 2.6648, "step": 40437 }, { "epoch": 2.510273760009932, "grad_norm": 0.46015495209438007, "learning_rate": 7.900264770708243e-06, "loss": 2.6519, "step": 40438 }, { "epoch": 2.5103358371096904, "grad_norm": 0.44308779292612294, "learning_rate": 7.89831651455747e-06, "loss": 2.6304, "step": 40439 }, { "epoch": 2.510397914209448, "grad_norm": 0.453319343457384, "learning_rate": 7.8963684780614e-06, "loss": 2.6398, "step": 40440 }, { "epoch": 2.5104599913092063, "grad_norm": 0.47089458070027607, "learning_rate": 7.894420661230183e-06, "loss": 2.7892, "step": 40441 }, { "epoch": 2.5105220684089637, "grad_norm": 0.443234447069761, "learning_rate": 7.892473064074008e-06, "loss": 2.6873, "step": 40442 }, { "epoch": 2.510584145508722, "grad_norm": 0.46990888709228734, "learning_rate": 7.890525686603023e-06, "loss": 2.6973, "step": 40443 }, { "epoch": 2.5106462226084796, "grad_norm": 0.4908562625420223, "learning_rate": 7.88857852882739e-06, "loss": 2.6516, "step": 40444 }, { "epoch": 2.5107082997082375, "grad_norm": 0.4441971111360825, "learning_rate": 7.886631590757254e-06, "loss": 2.6267, "step": 40445 }, { "epoch": 2.5107703768079954, "grad_norm": 0.4587474130146977, "learning_rate": 7.884684872402792e-06, "loss": 2.7089, "step": 40446 }, { "epoch": 2.5108324539077533, "grad_norm": 0.4587562318685228, "learning_rate": 7.882738373774156e-06, "loss": 2.6998, "step": 40447 }, { "epoch": 2.5108945310075113, "grad_norm": 0.43722399452264427, "learning_rate": 7.880792094881496e-06, "loss": 2.6115, "step": 40448 }, { "epoch": 2.510956608107269, "grad_norm": 0.4425193963137135, "learning_rate": 7.87884603573497e-06, "loss": 2.5969, "step": 40449 }, { "epoch": 2.511018685207027, "grad_norm": 0.3983460676712816, "learning_rate": 7.876900196344716e-06, "loss": 2.6534, "step": 40450 }, { "epoch": 2.511080762306785, "grad_norm": 0.44418976443974745, "learning_rate": 7.874954576720906e-06, "loss": 2.7023, "step": 40451 }, { "epoch": 2.511142839406543, "grad_norm": 0.47195243452729196, "learning_rate": 7.873009176873687e-06, "loss": 2.695, "step": 40452 }, { "epoch": 2.511204916506301, "grad_norm": 0.43015801071543813, "learning_rate": 7.871063996813211e-06, "loss": 2.6652, "step": 40453 }, { "epoch": 2.5112669936060588, "grad_norm": 0.4436124507862857, "learning_rate": 7.869119036549622e-06, "loss": 2.7102, "step": 40454 }, { "epoch": 2.5113290707058167, "grad_norm": 0.4846266431251754, "learning_rate": 7.867174296093049e-06, "loss": 2.7546, "step": 40455 }, { "epoch": 2.5113911478055746, "grad_norm": 0.44589220761973614, "learning_rate": 7.865229775453676e-06, "loss": 2.7249, "step": 40456 }, { "epoch": 2.5114532249053325, "grad_norm": 0.41574641368853765, "learning_rate": 7.86328547464162e-06, "loss": 2.6524, "step": 40457 }, { "epoch": 2.5115153020050904, "grad_norm": 0.4467716237520196, "learning_rate": 7.861341393667037e-06, "loss": 2.6379, "step": 40458 }, { "epoch": 2.5115773791048484, "grad_norm": 0.42420589294058864, "learning_rate": 7.859397532540053e-06, "loss": 2.7954, "step": 40459 }, { "epoch": 2.5116394562046063, "grad_norm": 0.43597447570077147, "learning_rate": 7.857453891270832e-06, "loss": 2.6305, "step": 40460 }, { "epoch": 2.5117015333043637, "grad_norm": 0.4253901514027939, "learning_rate": 7.85551046986951e-06, "loss": 2.6227, "step": 40461 }, { "epoch": 2.511763610404122, "grad_norm": 0.4111408948670183, "learning_rate": 7.853567268346213e-06, "loss": 2.6575, "step": 40462 }, { "epoch": 2.5118256875038796, "grad_norm": 0.4027576354716063, "learning_rate": 7.85162428671109e-06, "loss": 2.7055, "step": 40463 }, { "epoch": 2.511887764603638, "grad_norm": 0.4187556729003621, "learning_rate": 7.849681524974267e-06, "loss": 2.6771, "step": 40464 }, { "epoch": 2.5119498417033954, "grad_norm": 0.44765951478310195, "learning_rate": 7.847738983145897e-06, "loss": 2.7351, "step": 40465 }, { "epoch": 2.5120119188031538, "grad_norm": 0.42480987896432687, "learning_rate": 7.845796661236104e-06, "loss": 2.6327, "step": 40466 }, { "epoch": 2.5120739959029112, "grad_norm": 0.4696769767490998, "learning_rate": 7.84385455925502e-06, "loss": 2.7416, "step": 40467 }, { "epoch": 2.512136073002669, "grad_norm": 0.4343996034144862, "learning_rate": 7.841912677212777e-06, "loss": 2.7216, "step": 40468 }, { "epoch": 2.512198150102427, "grad_norm": 0.4165472117692772, "learning_rate": 7.839971015119518e-06, "loss": 2.7078, "step": 40469 }, { "epoch": 2.512260227202185, "grad_norm": 0.43572022961084383, "learning_rate": 7.838029572985362e-06, "loss": 2.6965, "step": 40470 }, { "epoch": 2.512322304301943, "grad_norm": 0.40982683668239883, "learning_rate": 7.836088350820442e-06, "loss": 2.7054, "step": 40471 }, { "epoch": 2.512384381401701, "grad_norm": 0.43874651578731544, "learning_rate": 7.834147348634879e-06, "loss": 2.6449, "step": 40472 }, { "epoch": 2.5124464585014588, "grad_norm": 0.42295291734310053, "learning_rate": 7.83220656643881e-06, "loss": 2.6924, "step": 40473 }, { "epoch": 2.5125085356012167, "grad_norm": 0.42128726850966836, "learning_rate": 7.83026600424236e-06, "loss": 2.613, "step": 40474 }, { "epoch": 2.5125706127009746, "grad_norm": 0.39917623065781, "learning_rate": 7.828325662055647e-06, "loss": 2.6056, "step": 40475 }, { "epoch": 2.5126326898007325, "grad_norm": 0.4248274523017578, "learning_rate": 7.826385539888798e-06, "loss": 2.6584, "step": 40476 }, { "epoch": 2.5126947669004904, "grad_norm": 0.41680169500479475, "learning_rate": 7.82444563775192e-06, "loss": 2.6352, "step": 40477 }, { "epoch": 2.5127568440002483, "grad_norm": 0.4103963966334134, "learning_rate": 7.822505955655162e-06, "loss": 2.5956, "step": 40478 }, { "epoch": 2.5128189211000063, "grad_norm": 0.38892095817686817, "learning_rate": 7.820566493608628e-06, "loss": 2.5878, "step": 40479 }, { "epoch": 2.512880998199764, "grad_norm": 0.40595346412073374, "learning_rate": 7.818627251622435e-06, "loss": 2.6931, "step": 40480 }, { "epoch": 2.512943075299522, "grad_norm": 0.4239931093085621, "learning_rate": 7.816688229706699e-06, "loss": 2.6712, "step": 40481 }, { "epoch": 2.51300515239928, "grad_norm": 0.3994514615706739, "learning_rate": 7.814749427871543e-06, "loss": 2.627, "step": 40482 }, { "epoch": 2.513067229499038, "grad_norm": 0.431144601961921, "learning_rate": 7.81281084612709e-06, "loss": 2.6757, "step": 40483 }, { "epoch": 2.513129306598796, "grad_norm": 0.4221409559817954, "learning_rate": 7.810872484483422e-06, "loss": 2.7298, "step": 40484 }, { "epoch": 2.5131913836985538, "grad_norm": 0.42645076294319856, "learning_rate": 7.808934342950696e-06, "loss": 2.6236, "step": 40485 }, { "epoch": 2.5132534607983112, "grad_norm": 0.45197385049708905, "learning_rate": 7.806996421538981e-06, "loss": 2.6563, "step": 40486 }, { "epoch": 2.5133155378980696, "grad_norm": 0.41909844149784675, "learning_rate": 7.805058720258428e-06, "loss": 2.6661, "step": 40487 }, { "epoch": 2.513377614997827, "grad_norm": 0.4387625800065434, "learning_rate": 7.803121239119121e-06, "loss": 2.685, "step": 40488 }, { "epoch": 2.5134396920975854, "grad_norm": 0.38712301220628326, "learning_rate": 7.801183978131177e-06, "loss": 2.7051, "step": 40489 }, { "epoch": 2.513501769197343, "grad_norm": 0.4183610340872671, "learning_rate": 7.7992469373047e-06, "loss": 2.6747, "step": 40490 }, { "epoch": 2.513563846297101, "grad_norm": 0.4521936843016825, "learning_rate": 7.797310116649786e-06, "loss": 2.6704, "step": 40491 }, { "epoch": 2.5136259233968588, "grad_norm": 0.43755537772515424, "learning_rate": 7.795373516176563e-06, "loss": 2.7284, "step": 40492 }, { "epoch": 2.5136880004966167, "grad_norm": 0.3936273398912792, "learning_rate": 7.793437135895115e-06, "loss": 2.6797, "step": 40493 }, { "epoch": 2.5137500775963746, "grad_norm": 0.47735061515703736, "learning_rate": 7.791500975815557e-06, "loss": 2.7363, "step": 40494 }, { "epoch": 2.5138121546961325, "grad_norm": 0.42906869198970343, "learning_rate": 7.789565035947972e-06, "loss": 2.6659, "step": 40495 }, { "epoch": 2.5138742317958904, "grad_norm": 0.4331487485641866, "learning_rate": 7.787629316302486e-06, "loss": 2.6846, "step": 40496 }, { "epoch": 2.5139363088956483, "grad_norm": 0.4353583744794546, "learning_rate": 7.785693816889184e-06, "loss": 2.6054, "step": 40497 }, { "epoch": 2.5139983859954063, "grad_norm": 0.4278052110239145, "learning_rate": 7.783758537718167e-06, "loss": 2.7062, "step": 40498 }, { "epoch": 2.514060463095164, "grad_norm": 0.4396267187786154, "learning_rate": 7.781823478799528e-06, "loss": 2.6361, "step": 40499 }, { "epoch": 2.514122540194922, "grad_norm": 0.4182751860266314, "learning_rate": 7.779888640143351e-06, "loss": 2.6559, "step": 40500 }, { "epoch": 2.51418461729468, "grad_norm": 0.42704640980962083, "learning_rate": 7.77795402175976e-06, "loss": 2.6811, "step": 40501 }, { "epoch": 2.514246694394438, "grad_norm": 0.443187646755269, "learning_rate": 7.776019623658831e-06, "loss": 2.7062, "step": 40502 }, { "epoch": 2.514308771494196, "grad_norm": 0.4048828816908369, "learning_rate": 7.774085445850653e-06, "loss": 2.6467, "step": 40503 }, { "epoch": 2.5143708485939538, "grad_norm": 0.40408601652602283, "learning_rate": 7.77215148834532e-06, "loss": 2.5785, "step": 40504 }, { "epoch": 2.5144329256937117, "grad_norm": 0.4137732377110256, "learning_rate": 7.770217751152931e-06, "loss": 2.7141, "step": 40505 }, { "epoch": 2.5144950027934696, "grad_norm": 0.4349587237023833, "learning_rate": 7.768284234283563e-06, "loss": 2.685, "step": 40506 }, { "epoch": 2.5145570798932275, "grad_norm": 0.43182272543638567, "learning_rate": 7.766350937747314e-06, "loss": 2.6908, "step": 40507 }, { "epoch": 2.5146191569929854, "grad_norm": 0.4198313415874339, "learning_rate": 7.764417861554251e-06, "loss": 2.6822, "step": 40508 }, { "epoch": 2.514681234092743, "grad_norm": 0.406563546152257, "learning_rate": 7.762485005714487e-06, "loss": 2.7171, "step": 40509 }, { "epoch": 2.5147433111925013, "grad_norm": 0.40948638711038243, "learning_rate": 7.76055237023809e-06, "loss": 2.6488, "step": 40510 }, { "epoch": 2.5148053882922587, "grad_norm": 0.40399938502005234, "learning_rate": 7.758619955135143e-06, "loss": 2.6994, "step": 40511 }, { "epoch": 2.514867465392017, "grad_norm": 0.4279025544096338, "learning_rate": 7.756687760415737e-06, "loss": 2.673, "step": 40512 }, { "epoch": 2.5149295424917746, "grad_norm": 0.41927538053838653, "learning_rate": 7.754755786089929e-06, "loss": 2.6231, "step": 40513 }, { "epoch": 2.514991619591533, "grad_norm": 0.40359797860594937, "learning_rate": 7.752824032167827e-06, "loss": 2.609, "step": 40514 }, { "epoch": 2.5150536966912904, "grad_norm": 0.3968093742719753, "learning_rate": 7.750892498659501e-06, "loss": 2.734, "step": 40515 }, { "epoch": 2.5151157737910483, "grad_norm": 0.3781756209545846, "learning_rate": 7.74896118557501e-06, "loss": 2.6728, "step": 40516 }, { "epoch": 2.5151778508908063, "grad_norm": 0.39271015585047253, "learning_rate": 7.747030092924457e-06, "loss": 2.6806, "step": 40517 }, { "epoch": 2.515239927990564, "grad_norm": 0.3766453924789281, "learning_rate": 7.745099220717894e-06, "loss": 2.6707, "step": 40518 }, { "epoch": 2.515302005090322, "grad_norm": 0.39505791212428487, "learning_rate": 7.74316856896542e-06, "loss": 2.6071, "step": 40519 }, { "epoch": 2.51536408219008, "grad_norm": 0.4447816394982971, "learning_rate": 7.741238137677093e-06, "loss": 2.7321, "step": 40520 }, { "epoch": 2.515426159289838, "grad_norm": 0.40309256711502817, "learning_rate": 7.739307926862987e-06, "loss": 2.5838, "step": 40521 }, { "epoch": 2.515488236389596, "grad_norm": 0.4179688391284313, "learning_rate": 7.737377936533158e-06, "loss": 2.6774, "step": 40522 }, { "epoch": 2.5155503134893538, "grad_norm": 0.44612414027714, "learning_rate": 7.735448166697695e-06, "loss": 2.7583, "step": 40523 }, { "epoch": 2.5156123905891117, "grad_norm": 0.44366399240452076, "learning_rate": 7.733518617366664e-06, "loss": 2.602, "step": 40524 }, { "epoch": 2.5156744676888696, "grad_norm": 0.4178466261948092, "learning_rate": 7.73158928855013e-06, "loss": 2.5306, "step": 40525 }, { "epoch": 2.5157365447886275, "grad_norm": 0.4256563440900676, "learning_rate": 7.729660180258148e-06, "loss": 2.7098, "step": 40526 }, { "epoch": 2.5157986218883854, "grad_norm": 0.4363885794946804, "learning_rate": 7.727731292500784e-06, "loss": 2.5825, "step": 40527 }, { "epoch": 2.5158606989881434, "grad_norm": 0.41680401882638457, "learning_rate": 7.725802625288114e-06, "loss": 2.672, "step": 40528 }, { "epoch": 2.5159227760879013, "grad_norm": 0.4508345197278887, "learning_rate": 7.723874178630197e-06, "loss": 2.772, "step": 40529 }, { "epoch": 2.515984853187659, "grad_norm": 0.40637553549584293, "learning_rate": 7.72194595253709e-06, "loss": 2.6794, "step": 40530 }, { "epoch": 2.516046930287417, "grad_norm": 0.43805924503065774, "learning_rate": 7.720017947018848e-06, "loss": 2.8377, "step": 40531 }, { "epoch": 2.516109007387175, "grad_norm": 0.47901463861137117, "learning_rate": 7.718090162085546e-06, "loss": 2.7065, "step": 40532 }, { "epoch": 2.516171084486933, "grad_norm": 0.42634913868956353, "learning_rate": 7.71616259774723e-06, "loss": 2.6356, "step": 40533 }, { "epoch": 2.5162331615866904, "grad_norm": 0.41029381034685947, "learning_rate": 7.714235254013957e-06, "loss": 2.6385, "step": 40534 }, { "epoch": 2.5162952386864488, "grad_norm": 0.4690022107948626, "learning_rate": 7.712308130895785e-06, "loss": 2.6886, "step": 40535 }, { "epoch": 2.5163573157862063, "grad_norm": 0.4463942923327929, "learning_rate": 7.710381228402757e-06, "loss": 2.732, "step": 40536 }, { "epoch": 2.5164193928859646, "grad_norm": 0.4031150424821619, "learning_rate": 7.708454546544946e-06, "loss": 2.6615, "step": 40537 }, { "epoch": 2.516481469985722, "grad_norm": 0.40338625611262424, "learning_rate": 7.706528085332393e-06, "loss": 2.6673, "step": 40538 }, { "epoch": 2.51654354708548, "grad_norm": 0.39095607079201516, "learning_rate": 7.704601844775156e-06, "loss": 2.693, "step": 40539 }, { "epoch": 2.516605624185238, "grad_norm": 0.4478907651583301, "learning_rate": 7.70267582488326e-06, "loss": 2.7563, "step": 40540 }, { "epoch": 2.516667701284996, "grad_norm": 0.4423816167213161, "learning_rate": 7.700750025666791e-06, "loss": 2.6502, "step": 40541 }, { "epoch": 2.5167297783847538, "grad_norm": 0.4424175055165307, "learning_rate": 7.698824447135771e-06, "loss": 2.7907, "step": 40542 }, { "epoch": 2.5167918554845117, "grad_norm": 0.43025490837356234, "learning_rate": 7.696899089300257e-06, "loss": 2.6947, "step": 40543 }, { "epoch": 2.5168539325842696, "grad_norm": 0.3898929603111386, "learning_rate": 7.694973952170276e-06, "loss": 2.6937, "step": 40544 }, { "epoch": 2.5169160096840275, "grad_norm": 0.39382189064108486, "learning_rate": 7.6930490357559e-06, "loss": 2.5562, "step": 40545 }, { "epoch": 2.5169780867837854, "grad_norm": 0.40470447357401956, "learning_rate": 7.691124340067158e-06, "loss": 2.6137, "step": 40546 }, { "epoch": 2.5170401638835433, "grad_norm": 0.40707085283150396, "learning_rate": 7.689199865114088e-06, "loss": 2.7015, "step": 40547 }, { "epoch": 2.5171022409833013, "grad_norm": 0.397673360310332, "learning_rate": 7.687275610906736e-06, "loss": 2.7101, "step": 40548 }, { "epoch": 2.517164318083059, "grad_norm": 0.4288615524821002, "learning_rate": 7.685351577455124e-06, "loss": 2.6524, "step": 40549 }, { "epoch": 2.517226395182817, "grad_norm": 0.44588226543011505, "learning_rate": 7.683427764769308e-06, "loss": 2.6731, "step": 40550 }, { "epoch": 2.517288472282575, "grad_norm": 0.39278776350409195, "learning_rate": 7.681504172859333e-06, "loss": 2.7034, "step": 40551 }, { "epoch": 2.517350549382333, "grad_norm": 0.4555644927643948, "learning_rate": 7.679580801735225e-06, "loss": 2.7488, "step": 40552 }, { "epoch": 2.517412626482091, "grad_norm": 0.4288625079920403, "learning_rate": 7.677657651407017e-06, "loss": 2.6455, "step": 40553 }, { "epoch": 2.5174747035818488, "grad_norm": 0.4392806329222833, "learning_rate": 7.67573472188473e-06, "loss": 2.6468, "step": 40554 }, { "epoch": 2.5175367806816067, "grad_norm": 0.46583944157443935, "learning_rate": 7.673812013178422e-06, "loss": 2.5806, "step": 40555 }, { "epoch": 2.5175988577813646, "grad_norm": 0.42099244026115473, "learning_rate": 7.67188952529811e-06, "loss": 2.6946, "step": 40556 }, { "epoch": 2.517660934881122, "grad_norm": 0.4564781428533041, "learning_rate": 7.66996725825383e-06, "loss": 2.6462, "step": 40557 }, { "epoch": 2.5177230119808804, "grad_norm": 0.46444681995883014, "learning_rate": 7.668045212055591e-06, "loss": 2.67, "step": 40558 }, { "epoch": 2.517785089080638, "grad_norm": 0.444660730393923, "learning_rate": 7.666123386713447e-06, "loss": 2.6803, "step": 40559 }, { "epoch": 2.5178471661803963, "grad_norm": 0.41904773660550804, "learning_rate": 7.664201782237418e-06, "loss": 2.6586, "step": 40560 }, { "epoch": 2.5179092432801538, "grad_norm": 0.4428030885007192, "learning_rate": 7.662280398637528e-06, "loss": 2.6371, "step": 40561 }, { "epoch": 2.517971320379912, "grad_norm": 0.49610543425009096, "learning_rate": 7.660359235923797e-06, "loss": 2.6839, "step": 40562 }, { "epoch": 2.5180333974796696, "grad_norm": 0.470221989816018, "learning_rate": 7.658438294106234e-06, "loss": 2.6531, "step": 40563 }, { "epoch": 2.5180954745794275, "grad_norm": 0.4142960663380609, "learning_rate": 7.656517573194893e-06, "loss": 2.7987, "step": 40564 }, { "epoch": 2.5181575516791854, "grad_norm": 0.4237454768870617, "learning_rate": 7.654597073199775e-06, "loss": 2.6733, "step": 40565 }, { "epoch": 2.5182196287789433, "grad_norm": 0.49728483356255454, "learning_rate": 7.652676794130908e-06, "loss": 2.7482, "step": 40566 }, { "epoch": 2.5182817058787013, "grad_norm": 0.4443265473767411, "learning_rate": 7.6507567359983e-06, "loss": 2.5569, "step": 40567 }, { "epoch": 2.518343782978459, "grad_norm": 0.45889265077291813, "learning_rate": 7.648836898811979e-06, "loss": 2.595, "step": 40568 }, { "epoch": 2.518405860078217, "grad_norm": 0.42524349305894976, "learning_rate": 7.646917282581956e-06, "loss": 2.6579, "step": 40569 }, { "epoch": 2.518467937177975, "grad_norm": 0.466486360708293, "learning_rate": 7.64499788731825e-06, "loss": 2.7026, "step": 40570 }, { "epoch": 2.518530014277733, "grad_norm": 0.4277762823608539, "learning_rate": 7.64307871303086e-06, "loss": 2.7198, "step": 40571 }, { "epoch": 2.518592091377491, "grad_norm": 0.4475798856301419, "learning_rate": 7.64115975972982e-06, "loss": 2.7673, "step": 40572 }, { "epoch": 2.5186541684772488, "grad_norm": 0.4216539545586895, "learning_rate": 7.639241027425136e-06, "loss": 2.6341, "step": 40573 }, { "epoch": 2.5187162455770067, "grad_norm": 0.4719411468880248, "learning_rate": 7.637322516126817e-06, "loss": 2.7042, "step": 40574 }, { "epoch": 2.5187783226767646, "grad_norm": 0.45551714370155616, "learning_rate": 7.635404225844861e-06, "loss": 2.6878, "step": 40575 }, { "epoch": 2.5188403997765225, "grad_norm": 0.42633732522072904, "learning_rate": 7.63348615658928e-06, "loss": 2.6882, "step": 40576 }, { "epoch": 2.5189024768762804, "grad_norm": 0.4275379513743092, "learning_rate": 7.631568308370102e-06, "loss": 2.7086, "step": 40577 }, { "epoch": 2.5189645539760384, "grad_norm": 0.48384999464006845, "learning_rate": 7.629650681197309e-06, "loss": 2.6848, "step": 40578 }, { "epoch": 2.5190266310757963, "grad_norm": 0.4084035208124392, "learning_rate": 7.6277332750809225e-06, "loss": 2.6505, "step": 40579 }, { "epoch": 2.519088708175554, "grad_norm": 0.4466274155829825, "learning_rate": 7.625816090030918e-06, "loss": 2.7032, "step": 40580 }, { "epoch": 2.519150785275312, "grad_norm": 0.45143714141514585, "learning_rate": 7.623899126057338e-06, "loss": 2.6977, "step": 40581 }, { "epoch": 2.5192128623750696, "grad_norm": 0.4338871762043398, "learning_rate": 7.621982383170145e-06, "loss": 2.6769, "step": 40582 }, { "epoch": 2.519274939474828, "grad_norm": 0.4775986897505706, "learning_rate": 7.6200658613793715e-06, "loss": 2.6974, "step": 40583 }, { "epoch": 2.5193370165745854, "grad_norm": 0.39525166446762394, "learning_rate": 7.618149560695003e-06, "loss": 2.5883, "step": 40584 }, { "epoch": 2.519399093674344, "grad_norm": 0.43383854322325355, "learning_rate": 7.616233481127039e-06, "loss": 2.6682, "step": 40585 }, { "epoch": 2.5194611707741013, "grad_norm": 0.4189619759044504, "learning_rate": 7.614317622685457e-06, "loss": 2.6859, "step": 40586 }, { "epoch": 2.519523247873859, "grad_norm": 0.40774214419030874, "learning_rate": 7.612401985380286e-06, "loss": 2.7559, "step": 40587 }, { "epoch": 2.519585324973617, "grad_norm": 0.4256283721846257, "learning_rate": 7.6104865692214975e-06, "loss": 2.7245, "step": 40588 }, { "epoch": 2.519647402073375, "grad_norm": 0.40540010242333835, "learning_rate": 7.608571374219098e-06, "loss": 2.6687, "step": 40589 }, { "epoch": 2.519709479173133, "grad_norm": 0.43773883022782806, "learning_rate": 7.606656400383055e-06, "loss": 2.6195, "step": 40590 }, { "epoch": 2.519771556272891, "grad_norm": 0.445292121019355, "learning_rate": 7.60474164772339e-06, "loss": 2.7052, "step": 40591 }, { "epoch": 2.5198336333726488, "grad_norm": 0.44361177897590476, "learning_rate": 7.602827116250083e-06, "loss": 2.7414, "step": 40592 }, { "epoch": 2.5198957104724067, "grad_norm": 0.44634306599698736, "learning_rate": 7.600912805973121e-06, "loss": 2.678, "step": 40593 }, { "epoch": 2.5199577875721646, "grad_norm": 0.4402003907267801, "learning_rate": 7.5989987169024725e-06, "loss": 2.6962, "step": 40594 }, { "epoch": 2.5200198646719225, "grad_norm": 0.4265727988596276, "learning_rate": 7.597084849048153e-06, "loss": 2.7076, "step": 40595 }, { "epoch": 2.5200819417716804, "grad_norm": 0.47804662336332576, "learning_rate": 7.595171202420137e-06, "loss": 2.6593, "step": 40596 }, { "epoch": 2.5201440188714384, "grad_norm": 0.4434683241558201, "learning_rate": 7.593257777028406e-06, "loss": 2.6221, "step": 40597 }, { "epoch": 2.5202060959711963, "grad_norm": 0.4475193053881659, "learning_rate": 7.591344572882947e-06, "loss": 2.7241, "step": 40598 }, { "epoch": 2.520268173070954, "grad_norm": 0.47422797179957327, "learning_rate": 7.589431589993723e-06, "loss": 2.6985, "step": 40599 }, { "epoch": 2.520330250170712, "grad_norm": 0.47470757367280264, "learning_rate": 7.587518828370743e-06, "loss": 2.6452, "step": 40600 }, { "epoch": 2.52039232727047, "grad_norm": 0.4623121356887898, "learning_rate": 7.585606288023972e-06, "loss": 2.6986, "step": 40601 }, { "epoch": 2.520454404370228, "grad_norm": 0.4548406677246493, "learning_rate": 7.58369396896339e-06, "loss": 2.6739, "step": 40602 }, { "epoch": 2.520516481469986, "grad_norm": 0.4442707223508218, "learning_rate": 7.581781871198962e-06, "loss": 2.6255, "step": 40603 }, { "epoch": 2.5205785585697438, "grad_norm": 0.4163471745424132, "learning_rate": 7.579869994740691e-06, "loss": 2.6373, "step": 40604 }, { "epoch": 2.5206406356695013, "grad_norm": 0.4536100290960014, "learning_rate": 7.577958339598529e-06, "loss": 2.6803, "step": 40605 }, { "epoch": 2.5207027127692596, "grad_norm": 0.4494604538536012, "learning_rate": 7.576046905782458e-06, "loss": 2.6532, "step": 40606 }, { "epoch": 2.520764789869017, "grad_norm": 0.43710881864701523, "learning_rate": 7.574135693302442e-06, "loss": 2.6195, "step": 40607 }, { "epoch": 2.5208268669687754, "grad_norm": 0.417915786524816, "learning_rate": 7.572224702168468e-06, "loss": 2.6883, "step": 40608 }, { "epoch": 2.520888944068533, "grad_norm": 0.44067573863959986, "learning_rate": 7.5703139323904994e-06, "loss": 2.7088, "step": 40609 }, { "epoch": 2.5209510211682913, "grad_norm": 0.42225379186207146, "learning_rate": 7.5684033839785004e-06, "loss": 2.6165, "step": 40610 }, { "epoch": 2.5210130982680488, "grad_norm": 0.4149333505079696, "learning_rate": 7.566493056942442e-06, "loss": 2.6734, "step": 40611 }, { "epoch": 2.5210751753678067, "grad_norm": 0.39180592367689915, "learning_rate": 7.5645829512922815e-06, "loss": 2.6747, "step": 40612 }, { "epoch": 2.5211372524675646, "grad_norm": 0.4024010219366187, "learning_rate": 7.562673067038001e-06, "loss": 2.6818, "step": 40613 }, { "epoch": 2.5211993295673225, "grad_norm": 0.4423760251499272, "learning_rate": 7.560763404189558e-06, "loss": 2.6695, "step": 40614 }, { "epoch": 2.5212614066670804, "grad_norm": 0.41848975535470134, "learning_rate": 7.558853962756906e-06, "loss": 2.6529, "step": 40615 }, { "epoch": 2.5213234837668383, "grad_norm": 0.40795125902624263, "learning_rate": 7.556944742750028e-06, "loss": 2.6988, "step": 40616 }, { "epoch": 2.5213855608665963, "grad_norm": 0.43140228733894165, "learning_rate": 7.555035744178857e-06, "loss": 2.6518, "step": 40617 }, { "epoch": 2.521447637966354, "grad_norm": 0.435045835091704, "learning_rate": 7.553126967053381e-06, "loss": 2.6971, "step": 40618 }, { "epoch": 2.521509715066112, "grad_norm": 0.4226322442251599, "learning_rate": 7.551218411383548e-06, "loss": 2.6195, "step": 40619 }, { "epoch": 2.52157179216587, "grad_norm": 0.40917473473658283, "learning_rate": 7.549310077179312e-06, "loss": 2.7593, "step": 40620 }, { "epoch": 2.521633869265628, "grad_norm": 0.43991905096330125, "learning_rate": 7.547401964450618e-06, "loss": 2.6007, "step": 40621 }, { "epoch": 2.521695946365386, "grad_norm": 0.4505679559778139, "learning_rate": 7.5454940732074485e-06, "loss": 2.7085, "step": 40622 }, { "epoch": 2.5217580234651438, "grad_norm": 0.4070383821766813, "learning_rate": 7.543586403459741e-06, "loss": 2.7058, "step": 40623 }, { "epoch": 2.5218201005649017, "grad_norm": 0.4042488106431138, "learning_rate": 7.541678955217451e-06, "loss": 2.5805, "step": 40624 }, { "epoch": 2.5218821776646596, "grad_norm": 0.43873060381909806, "learning_rate": 7.5397717284905246e-06, "loss": 2.6013, "step": 40625 }, { "epoch": 2.5219442547644175, "grad_norm": 0.4282503369213678, "learning_rate": 7.537864723288906e-06, "loss": 2.7103, "step": 40626 }, { "epoch": 2.5220063318641754, "grad_norm": 0.45176347530261085, "learning_rate": 7.535957939622573e-06, "loss": 2.6746, "step": 40627 }, { "epoch": 2.5220684089639334, "grad_norm": 0.4197415341653286, "learning_rate": 7.53405137750145e-06, "loss": 2.6667, "step": 40628 }, { "epoch": 2.5221304860636913, "grad_norm": 0.4057550349413788, "learning_rate": 7.532145036935489e-06, "loss": 2.6554, "step": 40629 }, { "epoch": 2.5221925631634488, "grad_norm": 0.3967785612965971, "learning_rate": 7.530238917934629e-06, "loss": 2.6075, "step": 40630 }, { "epoch": 2.522254640263207, "grad_norm": 0.4324386779497855, "learning_rate": 7.528333020508832e-06, "loss": 2.6965, "step": 40631 }, { "epoch": 2.5223167173629646, "grad_norm": 0.45520720063453424, "learning_rate": 7.526427344668036e-06, "loss": 2.6716, "step": 40632 }, { "epoch": 2.522378794462723, "grad_norm": 0.4460652831660072, "learning_rate": 7.524521890422176e-06, "loss": 2.643, "step": 40633 }, { "epoch": 2.5224408715624804, "grad_norm": 0.41164035231674817, "learning_rate": 7.522616657781195e-06, "loss": 2.7143, "step": 40634 }, { "epoch": 2.5225029486622383, "grad_norm": 0.40705346469700243, "learning_rate": 7.520711646755024e-06, "loss": 2.6758, "step": 40635 }, { "epoch": 2.5225650257619963, "grad_norm": 0.428423044084651, "learning_rate": 7.5188068573536275e-06, "loss": 2.6458, "step": 40636 }, { "epoch": 2.522627102861754, "grad_norm": 0.44253843824820877, "learning_rate": 7.5169022895869245e-06, "loss": 2.641, "step": 40637 }, { "epoch": 2.522689179961512, "grad_norm": 0.40195029195634224, "learning_rate": 7.5149979434648575e-06, "loss": 2.5411, "step": 40638 }, { "epoch": 2.52275125706127, "grad_norm": 0.4053819442175412, "learning_rate": 7.513093818997347e-06, "loss": 2.6412, "step": 40639 }, { "epoch": 2.522813334161028, "grad_norm": 0.4059902791459159, "learning_rate": 7.51118991619435e-06, "loss": 2.5588, "step": 40640 }, { "epoch": 2.522875411260786, "grad_norm": 0.40182459771757933, "learning_rate": 7.509286235065794e-06, "loss": 2.7197, "step": 40641 }, { "epoch": 2.5229374883605438, "grad_norm": 0.4266317779400498, "learning_rate": 7.507382775621602e-06, "loss": 2.647, "step": 40642 }, { "epoch": 2.5229995654603017, "grad_norm": 0.42447716375085137, "learning_rate": 7.505479537871701e-06, "loss": 2.6526, "step": 40643 }, { "epoch": 2.5230616425600596, "grad_norm": 0.4168053058604594, "learning_rate": 7.503576521826045e-06, "loss": 2.7038, "step": 40644 }, { "epoch": 2.5231237196598175, "grad_norm": 0.413166642478064, "learning_rate": 7.50167372749454e-06, "loss": 2.6387, "step": 40645 }, { "epoch": 2.5231857967595754, "grad_norm": 0.3900136741661834, "learning_rate": 7.4997711548871196e-06, "loss": 2.6069, "step": 40646 }, { "epoch": 2.5232478738593334, "grad_norm": 0.4397204793831388, "learning_rate": 7.497868804013702e-06, "loss": 2.7053, "step": 40647 }, { "epoch": 2.5233099509590913, "grad_norm": 0.40292950924653526, "learning_rate": 7.49596667488423e-06, "loss": 2.581, "step": 40648 }, { "epoch": 2.523372028058849, "grad_norm": 0.41173040673911504, "learning_rate": 7.4940647675086074e-06, "loss": 2.6621, "step": 40649 }, { "epoch": 2.523434105158607, "grad_norm": 0.4151187595816161, "learning_rate": 7.49216308189678e-06, "loss": 2.685, "step": 40650 }, { "epoch": 2.523496182258365, "grad_norm": 0.3956966184964227, "learning_rate": 7.490261618058658e-06, "loss": 2.7122, "step": 40651 }, { "epoch": 2.523558259358123, "grad_norm": 0.4106982872939082, "learning_rate": 7.488360376004155e-06, "loss": 2.6102, "step": 40652 }, { "epoch": 2.5236203364578804, "grad_norm": 0.3940164966034317, "learning_rate": 7.486459355743192e-06, "loss": 2.6383, "step": 40653 }, { "epoch": 2.523682413557639, "grad_norm": 0.41619720436327706, "learning_rate": 7.484558557285698e-06, "loss": 2.617, "step": 40654 }, { "epoch": 2.5237444906573963, "grad_norm": 0.3944121077017688, "learning_rate": 7.482657980641583e-06, "loss": 2.6147, "step": 40655 }, { "epoch": 2.5238065677571546, "grad_norm": 0.46223628799624955, "learning_rate": 7.480757625820767e-06, "loss": 2.6571, "step": 40656 }, { "epoch": 2.523868644856912, "grad_norm": 0.431186281555551, "learning_rate": 7.478857492833147e-06, "loss": 2.6755, "step": 40657 }, { "epoch": 2.5239307219566705, "grad_norm": 0.37773054134786477, "learning_rate": 7.47695758168866e-06, "loss": 2.5936, "step": 40658 }, { "epoch": 2.523992799056428, "grad_norm": 0.43545601414513163, "learning_rate": 7.475057892397208e-06, "loss": 2.6953, "step": 40659 }, { "epoch": 2.524054876156186, "grad_norm": 0.421712879537207, "learning_rate": 7.4731584249687014e-06, "loss": 2.7416, "step": 40660 }, { "epoch": 2.5241169532559438, "grad_norm": 0.43727710699941574, "learning_rate": 7.471259179413048e-06, "loss": 2.6639, "step": 40661 }, { "epoch": 2.5241790303557017, "grad_norm": 0.4146571118475714, "learning_rate": 7.469360155740157e-06, "loss": 2.6628, "step": 40662 }, { "epoch": 2.5242411074554596, "grad_norm": 0.4150840645122285, "learning_rate": 7.467461353959942e-06, "loss": 2.7159, "step": 40663 }, { "epoch": 2.5243031845552175, "grad_norm": 0.44445276087026686, "learning_rate": 7.465562774082313e-06, "loss": 2.6724, "step": 40664 }, { "epoch": 2.5243652616549754, "grad_norm": 0.4531530174275289, "learning_rate": 7.463664416117161e-06, "loss": 2.6982, "step": 40665 }, { "epoch": 2.5244273387547334, "grad_norm": 0.4146306260928751, "learning_rate": 7.461766280074395e-06, "loss": 2.6106, "step": 40666 }, { "epoch": 2.5244894158544913, "grad_norm": 0.4707523019514561, "learning_rate": 7.459868365963923e-06, "loss": 2.6534, "step": 40667 }, { "epoch": 2.524551492954249, "grad_norm": 0.4236207270008565, "learning_rate": 7.4579706737956505e-06, "loss": 2.7737, "step": 40668 }, { "epoch": 2.524613570054007, "grad_norm": 0.42558660345213606, "learning_rate": 7.456073203579472e-06, "loss": 2.687, "step": 40669 }, { "epoch": 2.524675647153765, "grad_norm": 0.4022903406933597, "learning_rate": 7.454175955325288e-06, "loss": 2.6739, "step": 40670 }, { "epoch": 2.524737724253523, "grad_norm": 0.4653226965561372, "learning_rate": 7.452278929042983e-06, "loss": 2.6221, "step": 40671 }, { "epoch": 2.524799801353281, "grad_norm": 0.45373809378480096, "learning_rate": 7.4503821247424835e-06, "loss": 2.6622, "step": 40672 }, { "epoch": 2.5248618784530388, "grad_norm": 0.45800193994331967, "learning_rate": 7.448485542433664e-06, "loss": 2.6875, "step": 40673 }, { "epoch": 2.5249239555527967, "grad_norm": 0.43374562395950017, "learning_rate": 7.446589182126429e-06, "loss": 2.7506, "step": 40674 }, { "epoch": 2.5249860326525546, "grad_norm": 0.4640217413166001, "learning_rate": 7.4446930438306525e-06, "loss": 2.7398, "step": 40675 }, { "epoch": 2.5250481097523125, "grad_norm": 0.42123866657439785, "learning_rate": 7.442797127556261e-06, "loss": 2.5871, "step": 40676 }, { "epoch": 2.5251101868520704, "grad_norm": 0.44259837037772043, "learning_rate": 7.440901433313124e-06, "loss": 2.732, "step": 40677 }, { "epoch": 2.525172263951828, "grad_norm": 0.4465110276911936, "learning_rate": 7.439005961111134e-06, "loss": 2.6245, "step": 40678 }, { "epoch": 2.5252343410515863, "grad_norm": 0.41496347973316017, "learning_rate": 7.437110710960177e-06, "loss": 2.6595, "step": 40679 }, { "epoch": 2.5252964181513438, "grad_norm": 0.470095657977712, "learning_rate": 7.4352156828701395e-06, "loss": 2.6637, "step": 40680 }, { "epoch": 2.525358495251102, "grad_norm": 0.4204582337503552, "learning_rate": 7.43332087685093e-06, "loss": 2.6616, "step": 40681 }, { "epoch": 2.5254205723508596, "grad_norm": 0.4136735420805828, "learning_rate": 7.4314262929124135e-06, "loss": 2.7618, "step": 40682 }, { "epoch": 2.5254826494506175, "grad_norm": 0.404460081615752, "learning_rate": 7.429531931064488e-06, "loss": 2.7022, "step": 40683 }, { "epoch": 2.5255447265503754, "grad_norm": 0.40757579145445344, "learning_rate": 7.427637791317022e-06, "loss": 2.5498, "step": 40684 }, { "epoch": 2.5256068036501333, "grad_norm": 0.4431935811204563, "learning_rate": 7.4257438736798965e-06, "loss": 2.6419, "step": 40685 }, { "epoch": 2.5256688807498913, "grad_norm": 0.4220516047111548, "learning_rate": 7.423850178163011e-06, "loss": 2.7207, "step": 40686 }, { "epoch": 2.525730957849649, "grad_norm": 0.42163098307101365, "learning_rate": 7.421956704776234e-06, "loss": 2.6349, "step": 40687 }, { "epoch": 2.525793034949407, "grad_norm": 0.40086736441757753, "learning_rate": 7.420063453529447e-06, "loss": 2.6264, "step": 40688 }, { "epoch": 2.525855112049165, "grad_norm": 0.4198857846386888, "learning_rate": 7.4181704244325135e-06, "loss": 2.7653, "step": 40689 }, { "epoch": 2.525917189148923, "grad_norm": 0.4191346707592169, "learning_rate": 7.416277617495332e-06, "loss": 2.6861, "step": 40690 }, { "epoch": 2.525979266248681, "grad_norm": 0.41428527597112913, "learning_rate": 7.414385032727766e-06, "loss": 2.7222, "step": 40691 }, { "epoch": 2.5260413433484388, "grad_norm": 0.38504732537766667, "learning_rate": 7.412492670139698e-06, "loss": 2.7311, "step": 40692 }, { "epoch": 2.5261034204481967, "grad_norm": 0.42049920072831115, "learning_rate": 7.41060052974098e-06, "loss": 2.6876, "step": 40693 }, { "epoch": 2.5261654975479546, "grad_norm": 0.4339557030491712, "learning_rate": 7.408708611541504e-06, "loss": 2.6434, "step": 40694 }, { "epoch": 2.5262275746477125, "grad_norm": 0.4212504638476888, "learning_rate": 7.40681691555114e-06, "loss": 2.6824, "step": 40695 }, { "epoch": 2.5262896517474704, "grad_norm": 0.44249094132132905, "learning_rate": 7.4049254417797534e-06, "loss": 2.6533, "step": 40696 }, { "epoch": 2.5263517288472284, "grad_norm": 0.44682365135805086, "learning_rate": 7.403034190237207e-06, "loss": 2.7076, "step": 40697 }, { "epoch": 2.5264138059469863, "grad_norm": 0.4005441696882173, "learning_rate": 7.40114316093336e-06, "loss": 2.6865, "step": 40698 }, { "epoch": 2.526475883046744, "grad_norm": 0.45639368564298144, "learning_rate": 7.399252353878105e-06, "loss": 2.727, "step": 40699 }, { "epoch": 2.526537960146502, "grad_norm": 0.44656253507706095, "learning_rate": 7.39736176908129e-06, "loss": 2.6752, "step": 40700 }, { "epoch": 2.5266000372462596, "grad_norm": 0.3918414059222121, "learning_rate": 7.395471406552784e-06, "loss": 2.6227, "step": 40701 }, { "epoch": 2.526662114346018, "grad_norm": 0.412216010015105, "learning_rate": 7.393581266302429e-06, "loss": 2.6153, "step": 40702 }, { "epoch": 2.5267241914457754, "grad_norm": 0.48072546016907763, "learning_rate": 7.391691348340119e-06, "loss": 2.6381, "step": 40703 }, { "epoch": 2.526786268545534, "grad_norm": 0.4404217535348904, "learning_rate": 7.389801652675699e-06, "loss": 2.6122, "step": 40704 }, { "epoch": 2.5268483456452913, "grad_norm": 0.4111505717521625, "learning_rate": 7.3879121793190284e-06, "loss": 2.6646, "step": 40705 }, { "epoch": 2.5269104227450496, "grad_norm": 0.3804381559945046, "learning_rate": 7.386022928279962e-06, "loss": 2.7046, "step": 40706 }, { "epoch": 2.526972499844807, "grad_norm": 0.4178627806761764, "learning_rate": 7.384133899568352e-06, "loss": 2.6105, "step": 40707 }, { "epoch": 2.527034576944565, "grad_norm": 0.4588850197899625, "learning_rate": 7.382245093194068e-06, "loss": 2.6477, "step": 40708 }, { "epoch": 2.527096654044323, "grad_norm": 0.41030704814255264, "learning_rate": 7.380356509166958e-06, "loss": 2.7166, "step": 40709 }, { "epoch": 2.527158731144081, "grad_norm": 0.37985921752143115, "learning_rate": 7.378468147496875e-06, "loss": 2.7327, "step": 40710 }, { "epoch": 2.5272208082438388, "grad_norm": 0.3830463512587463, "learning_rate": 7.3765800081936564e-06, "loss": 2.6842, "step": 40711 }, { "epoch": 2.5272828853435967, "grad_norm": 0.42035845749900996, "learning_rate": 7.374692091267182e-06, "loss": 2.6411, "step": 40712 }, { "epoch": 2.5273449624433546, "grad_norm": 0.40280683958872265, "learning_rate": 7.372804396727273e-06, "loss": 2.7146, "step": 40713 }, { "epoch": 2.5274070395431125, "grad_norm": 0.383976199120284, "learning_rate": 7.370916924583804e-06, "loss": 2.597, "step": 40714 }, { "epoch": 2.5274691166428704, "grad_norm": 0.4282584789781795, "learning_rate": 7.369029674846606e-06, "loss": 2.6693, "step": 40715 }, { "epoch": 2.5275311937426284, "grad_norm": 0.40283066258053474, "learning_rate": 7.367142647525521e-06, "loss": 2.68, "step": 40716 }, { "epoch": 2.5275932708423863, "grad_norm": 0.4414484797169082, "learning_rate": 7.365255842630414e-06, "loss": 2.665, "step": 40717 }, { "epoch": 2.527655347942144, "grad_norm": 0.40631944723018254, "learning_rate": 7.363369260171121e-06, "loss": 2.5969, "step": 40718 }, { "epoch": 2.527717425041902, "grad_norm": 0.4252001620617573, "learning_rate": 7.3614829001574795e-06, "loss": 2.5906, "step": 40719 }, { "epoch": 2.52777950214166, "grad_norm": 0.40982238260081544, "learning_rate": 7.35959676259933e-06, "loss": 2.6409, "step": 40720 }, { "epoch": 2.527841579241418, "grad_norm": 0.42882312184488885, "learning_rate": 7.3577108475065045e-06, "loss": 2.7014, "step": 40721 }, { "epoch": 2.527903656341176, "grad_norm": 0.4355812914469441, "learning_rate": 7.355825154888862e-06, "loss": 2.6813, "step": 40722 }, { "epoch": 2.527965733440934, "grad_norm": 0.39906258262687194, "learning_rate": 7.353939684756234e-06, "loss": 2.6054, "step": 40723 }, { "epoch": 2.5280278105406917, "grad_norm": 0.43252838118821135, "learning_rate": 7.352054437118455e-06, "loss": 2.7726, "step": 40724 }, { "epoch": 2.5280898876404496, "grad_norm": 0.4188596019789083, "learning_rate": 7.350169411985352e-06, "loss": 2.6787, "step": 40725 }, { "epoch": 2.528151964740207, "grad_norm": 0.4388490891300574, "learning_rate": 7.348284609366779e-06, "loss": 2.7295, "step": 40726 }, { "epoch": 2.5282140418399655, "grad_norm": 0.407443585644849, "learning_rate": 7.34640002927256e-06, "loss": 2.6775, "step": 40727 }, { "epoch": 2.528276118939723, "grad_norm": 0.41327889988020106, "learning_rate": 7.344515671712526e-06, "loss": 2.6468, "step": 40728 }, { "epoch": 2.5283381960394813, "grad_norm": 0.44648334439643467, "learning_rate": 7.342631536696498e-06, "loss": 2.6645, "step": 40729 }, { "epoch": 2.5284002731392388, "grad_norm": 0.44742747880647926, "learning_rate": 7.340747624234323e-06, "loss": 2.6689, "step": 40730 }, { "epoch": 2.5284623502389967, "grad_norm": 0.3890212707428547, "learning_rate": 7.3388639343358325e-06, "loss": 2.6526, "step": 40731 }, { "epoch": 2.5285244273387546, "grad_norm": 0.3990692104949376, "learning_rate": 7.33698046701084e-06, "loss": 2.5791, "step": 40732 }, { "epoch": 2.5285865044385125, "grad_norm": 0.4737586100155738, "learning_rate": 7.3350972222691775e-06, "loss": 2.6348, "step": 40733 }, { "epoch": 2.5286485815382704, "grad_norm": 0.4221004307639629, "learning_rate": 7.333214200120658e-06, "loss": 2.655, "step": 40734 }, { "epoch": 2.5287106586380284, "grad_norm": 0.3910422576157351, "learning_rate": 7.331331400575131e-06, "loss": 2.6633, "step": 40735 }, { "epoch": 2.5287727357377863, "grad_norm": 0.4014130169397198, "learning_rate": 7.3294488236424044e-06, "loss": 2.5605, "step": 40736 }, { "epoch": 2.528834812837544, "grad_norm": 0.44302775883084433, "learning_rate": 7.327566469332303e-06, "loss": 2.6618, "step": 40737 }, { "epoch": 2.528896889937302, "grad_norm": 0.4606514544246763, "learning_rate": 7.325684337654631e-06, "loss": 2.6947, "step": 40738 }, { "epoch": 2.52895896703706, "grad_norm": 0.4220621259143402, "learning_rate": 7.323802428619242e-06, "loss": 2.7216, "step": 40739 }, { "epoch": 2.529021044136818, "grad_norm": 0.41921645062555696, "learning_rate": 7.321920742235933e-06, "loss": 2.7393, "step": 40740 }, { "epoch": 2.529083121236576, "grad_norm": 0.39353174185299933, "learning_rate": 7.320039278514523e-06, "loss": 2.5007, "step": 40741 }, { "epoch": 2.5291451983363338, "grad_norm": 0.41713613844036, "learning_rate": 7.318158037464818e-06, "loss": 2.6821, "step": 40742 }, { "epoch": 2.5292072754360917, "grad_norm": 0.41858565179923335, "learning_rate": 7.316277019096651e-06, "loss": 2.5827, "step": 40743 }, { "epoch": 2.5292693525358496, "grad_norm": 0.44937693257241257, "learning_rate": 7.314396223419834e-06, "loss": 2.6974, "step": 40744 }, { "epoch": 2.5293314296356075, "grad_norm": 0.40932154415644934, "learning_rate": 7.312515650444174e-06, "loss": 2.6637, "step": 40745 }, { "epoch": 2.5293935067353654, "grad_norm": 0.42873641440218146, "learning_rate": 7.310635300179469e-06, "loss": 2.6964, "step": 40746 }, { "epoch": 2.5294555838351234, "grad_norm": 0.43723597393473534, "learning_rate": 7.308755172635557e-06, "loss": 2.7077, "step": 40747 }, { "epoch": 2.5295176609348813, "grad_norm": 0.42659639496332974, "learning_rate": 7.306875267822222e-06, "loss": 2.6256, "step": 40748 }, { "epoch": 2.5295797380346388, "grad_norm": 0.40958762944058474, "learning_rate": 7.304995585749291e-06, "loss": 2.6036, "step": 40749 }, { "epoch": 2.529641815134397, "grad_norm": 0.44454139738116144, "learning_rate": 7.303116126426562e-06, "loss": 2.66, "step": 40750 }, { "epoch": 2.5297038922341546, "grad_norm": 0.4211979837874712, "learning_rate": 7.301236889863844e-06, "loss": 2.5317, "step": 40751 }, { "epoch": 2.529765969333913, "grad_norm": 0.40973127237479295, "learning_rate": 7.299357876070922e-06, "loss": 2.6102, "step": 40752 }, { "epoch": 2.5298280464336704, "grad_norm": 0.38273473845116357, "learning_rate": 7.297479085057635e-06, "loss": 2.6272, "step": 40753 }, { "epoch": 2.529890123533429, "grad_norm": 0.4647453748966372, "learning_rate": 7.2956005168337615e-06, "loss": 2.6861, "step": 40754 }, { "epoch": 2.5299522006331863, "grad_norm": 0.43359771932898616, "learning_rate": 7.293722171409106e-06, "loss": 2.6741, "step": 40755 }, { "epoch": 2.530014277732944, "grad_norm": 0.4565944703780273, "learning_rate": 7.2918440487934724e-06, "loss": 2.72, "step": 40756 }, { "epoch": 2.530076354832702, "grad_norm": 0.41591668183628533, "learning_rate": 7.289966148996646e-06, "loss": 2.6581, "step": 40757 }, { "epoch": 2.53013843193246, "grad_norm": 0.43281287251811057, "learning_rate": 7.288088472028443e-06, "loss": 2.6429, "step": 40758 }, { "epoch": 2.530200509032218, "grad_norm": 0.4227135236931272, "learning_rate": 7.286211017898653e-06, "loss": 2.6839, "step": 40759 }, { "epoch": 2.530262586131976, "grad_norm": 0.4131465894393282, "learning_rate": 7.284333786617065e-06, "loss": 2.6225, "step": 40760 }, { "epoch": 2.5303246632317338, "grad_norm": 0.41049028146277045, "learning_rate": 7.28245677819347e-06, "loss": 2.5869, "step": 40761 }, { "epoch": 2.5303867403314917, "grad_norm": 0.4466792598266956, "learning_rate": 7.280579992637676e-06, "loss": 2.6617, "step": 40762 }, { "epoch": 2.5304488174312496, "grad_norm": 0.42818837914900704, "learning_rate": 7.278703429959471e-06, "loss": 2.6747, "step": 40763 }, { "epoch": 2.5305108945310075, "grad_norm": 0.4217040810678601, "learning_rate": 7.276827090168637e-06, "loss": 2.7272, "step": 40764 }, { "epoch": 2.5305729716307654, "grad_norm": 0.4227358371171427, "learning_rate": 7.274950973274958e-06, "loss": 2.6023, "step": 40765 }, { "epoch": 2.5306350487305234, "grad_norm": 0.4052299568107508, "learning_rate": 7.273075079288244e-06, "loss": 2.6243, "step": 40766 }, { "epoch": 2.5306971258302813, "grad_norm": 0.42161610394730437, "learning_rate": 7.2711994082182655e-06, "loss": 2.6208, "step": 40767 }, { "epoch": 2.530759202930039, "grad_norm": 0.46781078659385994, "learning_rate": 7.26932396007482e-06, "loss": 2.8321, "step": 40768 }, { "epoch": 2.530821280029797, "grad_norm": 0.4154222940594993, "learning_rate": 7.267448734867677e-06, "loss": 2.746, "step": 40769 }, { "epoch": 2.530883357129555, "grad_norm": 0.40917027355645036, "learning_rate": 7.265573732606623e-06, "loss": 2.6671, "step": 40770 }, { "epoch": 2.530945434229313, "grad_norm": 0.41890492277920466, "learning_rate": 7.263698953301451e-06, "loss": 2.7292, "step": 40771 }, { "epoch": 2.531007511329071, "grad_norm": 0.4273613997585752, "learning_rate": 7.261824396961936e-06, "loss": 2.6913, "step": 40772 }, { "epoch": 2.531069588428829, "grad_norm": 0.42849134934550104, "learning_rate": 7.259950063597859e-06, "loss": 2.7721, "step": 40773 }, { "epoch": 2.5311316655285863, "grad_norm": 0.3707499266525988, "learning_rate": 7.2580759532189906e-06, "loss": 2.6366, "step": 40774 }, { "epoch": 2.5311937426283446, "grad_norm": 0.380112762284288, "learning_rate": 7.256202065835121e-06, "loss": 2.7141, "step": 40775 }, { "epoch": 2.531255819728102, "grad_norm": 0.4308923821070917, "learning_rate": 7.254328401456028e-06, "loss": 2.6649, "step": 40776 }, { "epoch": 2.5313178968278605, "grad_norm": 0.4273688788409386, "learning_rate": 7.252454960091476e-06, "loss": 2.6869, "step": 40777 }, { "epoch": 2.531379973927618, "grad_norm": 0.4130268460477876, "learning_rate": 7.250581741751239e-06, "loss": 2.5941, "step": 40778 }, { "epoch": 2.531442051027376, "grad_norm": 0.39394017123043407, "learning_rate": 7.248708746445088e-06, "loss": 2.6255, "step": 40779 }, { "epoch": 2.5315041281271338, "grad_norm": 0.4104637446568443, "learning_rate": 7.246835974182814e-06, "loss": 2.6546, "step": 40780 }, { "epoch": 2.5315662052268917, "grad_norm": 0.42973089865951947, "learning_rate": 7.244963424974177e-06, "loss": 2.6676, "step": 40781 }, { "epoch": 2.5316282823266496, "grad_norm": 0.39332293556635156, "learning_rate": 7.243091098828947e-06, "loss": 2.6998, "step": 40782 }, { "epoch": 2.5316903594264075, "grad_norm": 0.40184957284414596, "learning_rate": 7.241218995756887e-06, "loss": 2.658, "step": 40783 }, { "epoch": 2.5317524365261654, "grad_norm": 0.47015326582981176, "learning_rate": 7.2393471157677526e-06, "loss": 2.691, "step": 40784 }, { "epoch": 2.5318145136259234, "grad_norm": 0.4229400225079742, "learning_rate": 7.237475458871346e-06, "loss": 2.703, "step": 40785 }, { "epoch": 2.5318765907256813, "grad_norm": 0.41111516121782393, "learning_rate": 7.235604025077403e-06, "loss": 2.7227, "step": 40786 }, { "epoch": 2.531938667825439, "grad_norm": 0.41808629388516905, "learning_rate": 7.2337328143957e-06, "loss": 2.7024, "step": 40787 }, { "epoch": 2.532000744925197, "grad_norm": 0.44773262331101904, "learning_rate": 7.231861826835979e-06, "loss": 2.6703, "step": 40788 }, { "epoch": 2.532062822024955, "grad_norm": 0.4119229550439539, "learning_rate": 7.229991062408031e-06, "loss": 2.6571, "step": 40789 }, { "epoch": 2.532124899124713, "grad_norm": 0.4326630303267292, "learning_rate": 7.2281205211216006e-06, "loss": 2.7533, "step": 40790 }, { "epoch": 2.532186976224471, "grad_norm": 0.4453277919887565, "learning_rate": 7.226250202986451e-06, "loss": 2.7023, "step": 40791 }, { "epoch": 2.532249053324229, "grad_norm": 0.4469273716575523, "learning_rate": 7.224380108012324e-06, "loss": 2.7005, "step": 40792 }, { "epoch": 2.5323111304239867, "grad_norm": 0.3861165029258302, "learning_rate": 7.222510236209001e-06, "loss": 2.6659, "step": 40793 }, { "epoch": 2.5323732075237446, "grad_norm": 0.4311070661434237, "learning_rate": 7.2206405875862306e-06, "loss": 2.7332, "step": 40794 }, { "epoch": 2.5324352846235025, "grad_norm": 0.39309521682068926, "learning_rate": 7.218771162153765e-06, "loss": 2.6611, "step": 40795 }, { "epoch": 2.5324973617232605, "grad_norm": 0.40956834711668366, "learning_rate": 7.216901959921352e-06, "loss": 2.6642, "step": 40796 }, { "epoch": 2.532559438823018, "grad_norm": 0.4430251353067166, "learning_rate": 7.2150329808987335e-06, "loss": 2.5955, "step": 40797 }, { "epoch": 2.5326215159227763, "grad_norm": 0.40833701899221053, "learning_rate": 7.213164225095687e-06, "loss": 2.685, "step": 40798 }, { "epoch": 2.5326835930225338, "grad_norm": 0.40672531198842155, "learning_rate": 7.211295692521952e-06, "loss": 2.6455, "step": 40799 }, { "epoch": 2.532745670122292, "grad_norm": 0.4384935541407767, "learning_rate": 7.209427383187273e-06, "loss": 2.6772, "step": 40800 }, { "epoch": 2.5328077472220496, "grad_norm": 0.41186100132818393, "learning_rate": 7.20755929710139e-06, "loss": 2.7256, "step": 40801 }, { "epoch": 2.532869824321808, "grad_norm": 0.39810941315325904, "learning_rate": 7.20569143427407e-06, "loss": 2.5864, "step": 40802 }, { "epoch": 2.5329319014215654, "grad_norm": 0.40213170065379744, "learning_rate": 7.2038237947150424e-06, "loss": 2.7419, "step": 40803 }, { "epoch": 2.5329939785213234, "grad_norm": 0.38725528680924765, "learning_rate": 7.20195637843406e-06, "loss": 2.6682, "step": 40804 }, { "epoch": 2.5330560556210813, "grad_norm": 0.41066369047159856, "learning_rate": 7.200089185440856e-06, "loss": 2.6367, "step": 40805 }, { "epoch": 2.533118132720839, "grad_norm": 0.4394890325499309, "learning_rate": 7.198222215745165e-06, "loss": 2.7448, "step": 40806 }, { "epoch": 2.533180209820597, "grad_norm": 0.4177013682817675, "learning_rate": 7.196355469356758e-06, "loss": 2.7426, "step": 40807 }, { "epoch": 2.533242286920355, "grad_norm": 0.3855100295250175, "learning_rate": 7.194488946285349e-06, "loss": 2.6476, "step": 40808 }, { "epoch": 2.533304364020113, "grad_norm": 0.42674311213194754, "learning_rate": 7.192622646540686e-06, "loss": 2.6405, "step": 40809 }, { "epoch": 2.533366441119871, "grad_norm": 0.4482305687289865, "learning_rate": 7.19075657013249e-06, "loss": 2.7091, "step": 40810 }, { "epoch": 2.5334285182196288, "grad_norm": 0.42980634606324025, "learning_rate": 7.188890717070507e-06, "loss": 2.6589, "step": 40811 }, { "epoch": 2.5334905953193867, "grad_norm": 0.4066795132645536, "learning_rate": 7.187025087364485e-06, "loss": 2.7127, "step": 40812 }, { "epoch": 2.5335526724191446, "grad_norm": 0.41628178065638227, "learning_rate": 7.185159681024151e-06, "loss": 2.6028, "step": 40813 }, { "epoch": 2.5336147495189025, "grad_norm": 0.470621014789664, "learning_rate": 7.183294498059228e-06, "loss": 2.62, "step": 40814 }, { "epoch": 2.5336768266186604, "grad_norm": 0.39372263086657294, "learning_rate": 7.181429538479439e-06, "loss": 2.6949, "step": 40815 }, { "epoch": 2.5337389037184184, "grad_norm": 0.40627888430101844, "learning_rate": 7.179564802294542e-06, "loss": 2.7422, "step": 40816 }, { "epoch": 2.5338009808181763, "grad_norm": 0.4118503280682801, "learning_rate": 7.177700289514249e-06, "loss": 2.6449, "step": 40817 }, { "epoch": 2.533863057917934, "grad_norm": 0.46871202632950215, "learning_rate": 7.175836000148295e-06, "loss": 2.6876, "step": 40818 }, { "epoch": 2.533925135017692, "grad_norm": 0.43871418179105626, "learning_rate": 7.173971934206392e-06, "loss": 2.6321, "step": 40819 }, { "epoch": 2.53398721211745, "grad_norm": 0.39471754076055726, "learning_rate": 7.172108091698265e-06, "loss": 2.665, "step": 40820 }, { "epoch": 2.534049289217208, "grad_norm": 0.4272441231818395, "learning_rate": 7.1702444726336585e-06, "loss": 2.6219, "step": 40821 }, { "epoch": 2.5341113663169654, "grad_norm": 0.4139003076123175, "learning_rate": 7.168381077022285e-06, "loss": 2.6799, "step": 40822 }, { "epoch": 2.534173443416724, "grad_norm": 0.4288051388765773, "learning_rate": 7.166517904873865e-06, "loss": 2.6541, "step": 40823 }, { "epoch": 2.5342355205164813, "grad_norm": 0.41517749945916754, "learning_rate": 7.164654956198108e-06, "loss": 2.6984, "step": 40824 }, { "epoch": 2.5342975976162396, "grad_norm": 0.4409585555988381, "learning_rate": 7.162792231004756e-06, "loss": 2.6324, "step": 40825 }, { "epoch": 2.534359674715997, "grad_norm": 0.4306761378803302, "learning_rate": 7.160929729303517e-06, "loss": 2.6476, "step": 40826 }, { "epoch": 2.534421751815755, "grad_norm": 0.4195671712603108, "learning_rate": 7.159067451104107e-06, "loss": 2.6914, "step": 40827 }, { "epoch": 2.534483828915513, "grad_norm": 0.43397790198387304, "learning_rate": 7.157205396416228e-06, "loss": 2.6489, "step": 40828 }, { "epoch": 2.534545906015271, "grad_norm": 0.44701296737282503, "learning_rate": 7.155343565249623e-06, "loss": 2.6672, "step": 40829 }, { "epoch": 2.5346079831150288, "grad_norm": 0.4140628238892638, "learning_rate": 7.153481957613989e-06, "loss": 2.6698, "step": 40830 }, { "epoch": 2.5346700602147867, "grad_norm": 0.4450654465521969, "learning_rate": 7.151620573519041e-06, "loss": 2.6856, "step": 40831 }, { "epoch": 2.5347321373145446, "grad_norm": 0.4239578927371323, "learning_rate": 7.149759412974494e-06, "loss": 2.6614, "step": 40832 }, { "epoch": 2.5347942144143025, "grad_norm": 0.42736459015237466, "learning_rate": 7.1478984759900395e-06, "loss": 2.633, "step": 40833 }, { "epoch": 2.5348562915140604, "grad_norm": 0.4355646993642042, "learning_rate": 7.1460377625754134e-06, "loss": 2.601, "step": 40834 }, { "epoch": 2.5349183686138184, "grad_norm": 0.430127404184892, "learning_rate": 7.14417727274031e-06, "loss": 2.6928, "step": 40835 }, { "epoch": 2.5349804457135763, "grad_norm": 0.39565357300928594, "learning_rate": 7.142317006494442e-06, "loss": 2.6744, "step": 40836 }, { "epoch": 2.535042522813334, "grad_norm": 0.4437154272876428, "learning_rate": 7.1404569638474915e-06, "loss": 2.7138, "step": 40837 }, { "epoch": 2.535104599913092, "grad_norm": 0.425975702661821, "learning_rate": 7.138597144809201e-06, "loss": 2.7279, "step": 40838 }, { "epoch": 2.53516667701285, "grad_norm": 0.4170138601348441, "learning_rate": 7.136737549389249e-06, "loss": 2.6774, "step": 40839 }, { "epoch": 2.535228754112608, "grad_norm": 0.4117948769361887, "learning_rate": 7.134878177597343e-06, "loss": 2.7388, "step": 40840 }, { "epoch": 2.535290831212366, "grad_norm": 0.3922470176794925, "learning_rate": 7.1330190294431884e-06, "loss": 2.6157, "step": 40841 }, { "epoch": 2.535352908312124, "grad_norm": 0.4185587382153954, "learning_rate": 7.131160104936463e-06, "loss": 2.6634, "step": 40842 }, { "epoch": 2.5354149854118817, "grad_norm": 0.46716314973175355, "learning_rate": 7.129301404086896e-06, "loss": 2.6515, "step": 40843 }, { "epoch": 2.5354770625116396, "grad_norm": 0.3983442533996662, "learning_rate": 7.12744292690416e-06, "loss": 2.5875, "step": 40844 }, { "epoch": 2.535539139611397, "grad_norm": 0.408058816324961, "learning_rate": 7.125584673397967e-06, "loss": 2.5706, "step": 40845 }, { "epoch": 2.5356012167111555, "grad_norm": 0.4180423279644906, "learning_rate": 7.1237266435780156e-06, "loss": 2.7419, "step": 40846 }, { "epoch": 2.535663293810913, "grad_norm": 0.42148625158646985, "learning_rate": 7.1218688374539765e-06, "loss": 2.66, "step": 40847 }, { "epoch": 2.5357253709106713, "grad_norm": 0.4238662175230797, "learning_rate": 7.120011255035564e-06, "loss": 2.6089, "step": 40848 }, { "epoch": 2.5357874480104288, "grad_norm": 0.39883961675006113, "learning_rate": 7.118153896332469e-06, "loss": 2.6555, "step": 40849 }, { "epoch": 2.5358495251101867, "grad_norm": 0.4098803364184419, "learning_rate": 7.1162967613543685e-06, "loss": 2.6691, "step": 40850 }, { "epoch": 2.5359116022099446, "grad_norm": 0.40138378248226153, "learning_rate": 7.114439850110949e-06, "loss": 2.6919, "step": 40851 }, { "epoch": 2.5359736793097025, "grad_norm": 0.41475739738738043, "learning_rate": 7.1125831626119246e-06, "loss": 2.7186, "step": 40852 }, { "epoch": 2.5360357564094604, "grad_norm": 0.4115536273587888, "learning_rate": 7.11072669886696e-06, "loss": 2.7076, "step": 40853 }, { "epoch": 2.5360978335092184, "grad_norm": 0.42929544769688516, "learning_rate": 7.108870458885747e-06, "loss": 2.6443, "step": 40854 }, { "epoch": 2.5361599106089763, "grad_norm": 0.39807059654402543, "learning_rate": 7.107014442677967e-06, "loss": 2.7065, "step": 40855 }, { "epoch": 2.536221987708734, "grad_norm": 0.4183622272867944, "learning_rate": 7.105158650253297e-06, "loss": 2.7329, "step": 40856 }, { "epoch": 2.536284064808492, "grad_norm": 0.4273524037658296, "learning_rate": 7.1033030816214315e-06, "loss": 2.6918, "step": 40857 }, { "epoch": 2.53634614190825, "grad_norm": 0.43510114844860415, "learning_rate": 7.101447736792055e-06, "loss": 2.6347, "step": 40858 }, { "epoch": 2.536408219008008, "grad_norm": 0.39253261169484094, "learning_rate": 7.099592615774836e-06, "loss": 2.7234, "step": 40859 }, { "epoch": 2.536470296107766, "grad_norm": 0.4233981293677954, "learning_rate": 7.097737718579451e-06, "loss": 2.6459, "step": 40860 }, { "epoch": 2.536532373207524, "grad_norm": 0.4433625679069191, "learning_rate": 7.09588304521559e-06, "loss": 2.6509, "step": 40861 }, { "epoch": 2.5365944503072817, "grad_norm": 0.4118713667887322, "learning_rate": 7.094028595692919e-06, "loss": 2.7663, "step": 40862 }, { "epoch": 2.5366565274070396, "grad_norm": 0.4009619381741993, "learning_rate": 7.09217437002112e-06, "loss": 2.7267, "step": 40863 }, { "epoch": 2.5367186045067975, "grad_norm": 0.4290875147385311, "learning_rate": 7.090320368209857e-06, "loss": 2.5169, "step": 40864 }, { "epoch": 2.5367806816065555, "grad_norm": 0.42589784478177195, "learning_rate": 7.088466590268816e-06, "loss": 2.6951, "step": 40865 }, { "epoch": 2.5368427587063134, "grad_norm": 0.42022743640624327, "learning_rate": 7.086613036207662e-06, "loss": 2.6286, "step": 40866 }, { "epoch": 2.5369048358060713, "grad_norm": 0.4191569436230456, "learning_rate": 7.084759706036065e-06, "loss": 2.6947, "step": 40867 }, { "epoch": 2.5369669129058288, "grad_norm": 0.4140347165557095, "learning_rate": 7.0829065997636945e-06, "loss": 2.7259, "step": 40868 }, { "epoch": 2.537028990005587, "grad_norm": 0.43913549176242545, "learning_rate": 7.08105371740021e-06, "loss": 2.7599, "step": 40869 }, { "epoch": 2.5370910671053446, "grad_norm": 0.4370568619752702, "learning_rate": 7.079201058955298e-06, "loss": 2.7023, "step": 40870 }, { "epoch": 2.537153144205103, "grad_norm": 0.4711641293867702, "learning_rate": 7.0773486244386125e-06, "loss": 2.5909, "step": 40871 }, { "epoch": 2.5372152213048604, "grad_norm": 0.4016248892027972, "learning_rate": 7.075496413859817e-06, "loss": 2.6963, "step": 40872 }, { "epoch": 2.537277298404619, "grad_norm": 0.40580943867891606, "learning_rate": 7.073644427228571e-06, "loss": 2.6702, "step": 40873 }, { "epoch": 2.5373393755043763, "grad_norm": 0.4493488123482757, "learning_rate": 7.071792664554549e-06, "loss": 2.6852, "step": 40874 }, { "epoch": 2.537401452604134, "grad_norm": 0.4175143757932031, "learning_rate": 7.069941125847407e-06, "loss": 2.6539, "step": 40875 }, { "epoch": 2.537463529703892, "grad_norm": 0.41855929003244086, "learning_rate": 7.068089811116807e-06, "loss": 2.6314, "step": 40876 }, { "epoch": 2.53752560680365, "grad_norm": 0.4322242416133358, "learning_rate": 7.066238720372387e-06, "loss": 2.6716, "step": 40877 }, { "epoch": 2.537587683903408, "grad_norm": 0.4394975597381113, "learning_rate": 7.0643878536238274e-06, "loss": 2.6853, "step": 40878 }, { "epoch": 2.537649761003166, "grad_norm": 0.38911018385774543, "learning_rate": 7.062537210880788e-06, "loss": 2.6681, "step": 40879 }, { "epoch": 2.537711838102924, "grad_norm": 0.42598556725920356, "learning_rate": 7.060686792152915e-06, "loss": 2.6319, "step": 40880 }, { "epoch": 2.5377739152026817, "grad_norm": 0.38855018819441883, "learning_rate": 7.058836597449864e-06, "loss": 2.6595, "step": 40881 }, { "epoch": 2.5378359923024396, "grad_norm": 0.4062717505281113, "learning_rate": 7.056986626781292e-06, "loss": 2.6151, "step": 40882 }, { "epoch": 2.5378980694021975, "grad_norm": 0.4259330408950218, "learning_rate": 7.055136880156826e-06, "loss": 2.7098, "step": 40883 }, { "epoch": 2.5379601465019554, "grad_norm": 0.42623473000232837, "learning_rate": 7.053287357586152e-06, "loss": 2.6171, "step": 40884 }, { "epoch": 2.5380222236017134, "grad_norm": 0.423164340096341, "learning_rate": 7.051438059078902e-06, "loss": 2.6197, "step": 40885 }, { "epoch": 2.5380843007014713, "grad_norm": 0.41616678849583744, "learning_rate": 7.0495889846447285e-06, "loss": 2.6723, "step": 40886 }, { "epoch": 2.538146377801229, "grad_norm": 0.4156221669322811, "learning_rate": 7.047740134293257e-06, "loss": 2.7607, "step": 40887 }, { "epoch": 2.538208454900987, "grad_norm": 0.4455523263382735, "learning_rate": 7.045891508034169e-06, "loss": 2.7358, "step": 40888 }, { "epoch": 2.538270532000745, "grad_norm": 0.39994303178426616, "learning_rate": 7.0440431058770965e-06, "loss": 2.7562, "step": 40889 }, { "epoch": 2.538332609100503, "grad_norm": 0.3874512295641793, "learning_rate": 7.0421949278316746e-06, "loss": 2.6008, "step": 40890 }, { "epoch": 2.538394686200261, "grad_norm": 0.41419405017080996, "learning_rate": 7.040346973907547e-06, "loss": 2.5816, "step": 40891 }, { "epoch": 2.538456763300019, "grad_norm": 0.4324910688305006, "learning_rate": 7.038499244114349e-06, "loss": 2.7682, "step": 40892 }, { "epoch": 2.5385188403997763, "grad_norm": 0.4175755085876543, "learning_rate": 7.036651738461741e-06, "loss": 2.6507, "step": 40893 }, { "epoch": 2.5385809174995346, "grad_norm": 0.42276192082743014, "learning_rate": 7.034804456959354e-06, "loss": 2.5837, "step": 40894 }, { "epoch": 2.538642994599292, "grad_norm": 0.42127709935463015, "learning_rate": 7.032957399616818e-06, "loss": 2.6126, "step": 40895 }, { "epoch": 2.5387050716990505, "grad_norm": 0.4897291307944144, "learning_rate": 7.031110566443766e-06, "loss": 2.6279, "step": 40896 }, { "epoch": 2.538767148798808, "grad_norm": 0.43529462138182445, "learning_rate": 7.02926395744985e-06, "loss": 2.6367, "step": 40897 }, { "epoch": 2.538829225898566, "grad_norm": 0.41933844200917925, "learning_rate": 7.027417572644701e-06, "loss": 2.6405, "step": 40898 }, { "epoch": 2.5388913029983238, "grad_norm": 0.4161514096449435, "learning_rate": 7.0255714120379405e-06, "loss": 2.6452, "step": 40899 }, { "epoch": 2.5389533800980817, "grad_norm": 0.4792898026794111, "learning_rate": 7.023725475639198e-06, "loss": 2.6838, "step": 40900 }, { "epoch": 2.5390154571978396, "grad_norm": 0.41842163363539187, "learning_rate": 7.021879763458117e-06, "loss": 2.6582, "step": 40901 }, { "epoch": 2.5390775342975975, "grad_norm": 0.4165123673683093, "learning_rate": 7.020034275504328e-06, "loss": 2.6555, "step": 40902 }, { "epoch": 2.5391396113973554, "grad_norm": 0.4378300852106313, "learning_rate": 7.018189011787457e-06, "loss": 2.7195, "step": 40903 }, { "epoch": 2.5392016884971134, "grad_norm": 0.46643438589832, "learning_rate": 7.016343972317124e-06, "loss": 2.6546, "step": 40904 }, { "epoch": 2.5392637655968713, "grad_norm": 0.4392438771847534, "learning_rate": 7.01449915710295e-06, "loss": 2.708, "step": 40905 }, { "epoch": 2.539325842696629, "grad_norm": 0.3904353635145428, "learning_rate": 7.0126545661545805e-06, "loss": 2.6353, "step": 40906 }, { "epoch": 2.539387919796387, "grad_norm": 0.4186196117094216, "learning_rate": 7.010810199481621e-06, "loss": 2.7098, "step": 40907 }, { "epoch": 2.539449996896145, "grad_norm": 0.4277699251076883, "learning_rate": 7.008966057093707e-06, "loss": 2.7331, "step": 40908 }, { "epoch": 2.539512073995903, "grad_norm": 0.4137492164304753, "learning_rate": 7.007122139000444e-06, "loss": 2.5979, "step": 40909 }, { "epoch": 2.539574151095661, "grad_norm": 0.3806179748788847, "learning_rate": 7.005278445211455e-06, "loss": 2.625, "step": 40910 }, { "epoch": 2.539636228195419, "grad_norm": 0.4181377924685667, "learning_rate": 7.003434975736383e-06, "loss": 2.7183, "step": 40911 }, { "epoch": 2.5396983052951767, "grad_norm": 0.43109775080938056, "learning_rate": 7.001591730584827e-06, "loss": 2.6932, "step": 40912 }, { "epoch": 2.5397603823949346, "grad_norm": 0.4158005135885776, "learning_rate": 6.999748709766402e-06, "loss": 2.6371, "step": 40913 }, { "epoch": 2.5398224594946925, "grad_norm": 0.40294167615886956, "learning_rate": 6.99790591329072e-06, "loss": 2.6402, "step": 40914 }, { "epoch": 2.5398845365944505, "grad_norm": 0.4248303811206558, "learning_rate": 6.996063341167414e-06, "loss": 2.7266, "step": 40915 }, { "epoch": 2.539946613694208, "grad_norm": 0.42934060399022994, "learning_rate": 6.994220993406081e-06, "loss": 2.747, "step": 40916 }, { "epoch": 2.5400086907939663, "grad_norm": 0.42495751098407125, "learning_rate": 6.992378870016336e-06, "loss": 2.7379, "step": 40917 }, { "epoch": 2.5400707678937238, "grad_norm": 0.4091765920298896, "learning_rate": 6.990536971007794e-06, "loss": 2.6044, "step": 40918 }, { "epoch": 2.540132844993482, "grad_norm": 0.4082937550664402, "learning_rate": 6.988695296390046e-06, "loss": 2.7807, "step": 40919 }, { "epoch": 2.5401949220932396, "grad_norm": 0.4194743588480468, "learning_rate": 6.986853846172736e-06, "loss": 2.6408, "step": 40920 }, { "epoch": 2.540256999192998, "grad_norm": 0.4223740769257948, "learning_rate": 6.985012620365444e-06, "loss": 2.7303, "step": 40921 }, { "epoch": 2.5403190762927554, "grad_norm": 0.41433275148366727, "learning_rate": 6.983171618977785e-06, "loss": 2.7063, "step": 40922 }, { "epoch": 2.5403811533925134, "grad_norm": 0.4250602041406024, "learning_rate": 6.9813308420193516e-06, "loss": 2.817, "step": 40923 }, { "epoch": 2.5404432304922713, "grad_norm": 0.4054264377280347, "learning_rate": 6.979490289499769e-06, "loss": 2.6034, "step": 40924 }, { "epoch": 2.540505307592029, "grad_norm": 0.4178997232356345, "learning_rate": 6.9776499614286285e-06, "loss": 2.7202, "step": 40925 }, { "epoch": 2.540567384691787, "grad_norm": 0.42569731838769764, "learning_rate": 6.975809857815535e-06, "loss": 2.6549, "step": 40926 }, { "epoch": 2.540629461791545, "grad_norm": 0.41569718185847876, "learning_rate": 6.973969978670086e-06, "loss": 2.6994, "step": 40927 }, { "epoch": 2.540691538891303, "grad_norm": 0.38696735752548, "learning_rate": 6.9721303240018686e-06, "loss": 2.7168, "step": 40928 }, { "epoch": 2.540753615991061, "grad_norm": 0.4295776747020373, "learning_rate": 6.970290893820502e-06, "loss": 2.6747, "step": 40929 }, { "epoch": 2.540815693090819, "grad_norm": 0.4446301502769221, "learning_rate": 6.968451688135575e-06, "loss": 2.7914, "step": 40930 }, { "epoch": 2.5408777701905767, "grad_norm": 0.40341780179347414, "learning_rate": 6.966612706956682e-06, "loss": 2.5433, "step": 40931 }, { "epoch": 2.5409398472903346, "grad_norm": 0.40432182640444, "learning_rate": 6.9647739502934016e-06, "loss": 2.6299, "step": 40932 }, { "epoch": 2.5410019243900925, "grad_norm": 0.4299707346205756, "learning_rate": 6.962935418155359e-06, "loss": 2.7486, "step": 40933 }, { "epoch": 2.5410640014898505, "grad_norm": 0.4257137700075005, "learning_rate": 6.961097110552128e-06, "loss": 2.619, "step": 40934 }, { "epoch": 2.5411260785896084, "grad_norm": 0.3683741043721972, "learning_rate": 6.959259027493304e-06, "loss": 2.6029, "step": 40935 }, { "epoch": 2.5411881556893663, "grad_norm": 0.4130554187945171, "learning_rate": 6.95742116898846e-06, "loss": 2.6547, "step": 40936 }, { "epoch": 2.541250232789124, "grad_norm": 0.4157500516303071, "learning_rate": 6.955583535047211e-06, "loss": 2.6542, "step": 40937 }, { "epoch": 2.541312309888882, "grad_norm": 0.4026301207407753, "learning_rate": 6.953746125679134e-06, "loss": 2.6381, "step": 40938 }, { "epoch": 2.54137438698864, "grad_norm": 0.40042880736862624, "learning_rate": 6.951908940893809e-06, "loss": 2.6029, "step": 40939 }, { "epoch": 2.541436464088398, "grad_norm": 0.39729748113358454, "learning_rate": 6.9500719807008285e-06, "loss": 2.6623, "step": 40940 }, { "epoch": 2.5414985411881554, "grad_norm": 0.43084097275641575, "learning_rate": 6.948235245109763e-06, "loss": 2.6574, "step": 40941 }, { "epoch": 2.541560618287914, "grad_norm": 0.4243722227059051, "learning_rate": 6.946398734130199e-06, "loss": 2.6233, "step": 40942 }, { "epoch": 2.5416226953876713, "grad_norm": 0.39209408122393996, "learning_rate": 6.94456244777174e-06, "loss": 2.6461, "step": 40943 }, { "epoch": 2.5416847724874296, "grad_norm": 0.39710215653352904, "learning_rate": 6.942726386043952e-06, "loss": 2.6051, "step": 40944 }, { "epoch": 2.541746849587187, "grad_norm": 0.425660825025498, "learning_rate": 6.940890548956414e-06, "loss": 2.6008, "step": 40945 }, { "epoch": 2.541808926686945, "grad_norm": 0.45160982418562595, "learning_rate": 6.939054936518691e-06, "loss": 2.6755, "step": 40946 }, { "epoch": 2.541871003786703, "grad_norm": 0.4044574473965957, "learning_rate": 6.937219548740381e-06, "loss": 2.6418, "step": 40947 }, { "epoch": 2.541933080886461, "grad_norm": 0.4302057608680693, "learning_rate": 6.93538438563105e-06, "loss": 2.6365, "step": 40948 }, { "epoch": 2.541995157986219, "grad_norm": 0.40202455914619506, "learning_rate": 6.933549447200277e-06, "loss": 2.6968, "step": 40949 }, { "epoch": 2.5420572350859767, "grad_norm": 0.4173432063160236, "learning_rate": 6.931714733457617e-06, "loss": 2.6359, "step": 40950 }, { "epoch": 2.5421193121857346, "grad_norm": 0.43029972157173446, "learning_rate": 6.929880244412668e-06, "loss": 2.6603, "step": 40951 }, { "epoch": 2.5421813892854925, "grad_norm": 0.40056645324132384, "learning_rate": 6.928045980074988e-06, "loss": 2.6682, "step": 40952 }, { "epoch": 2.5422434663852504, "grad_norm": 0.39850154062856386, "learning_rate": 6.926211940454147e-06, "loss": 2.7634, "step": 40953 }, { "epoch": 2.5423055434850084, "grad_norm": 0.4189025394683726, "learning_rate": 6.924378125559716e-06, "loss": 2.7592, "step": 40954 }, { "epoch": 2.5423676205847663, "grad_norm": 0.41211525067665067, "learning_rate": 6.922544535401254e-06, "loss": 2.6407, "step": 40955 }, { "epoch": 2.542429697684524, "grad_norm": 0.38782030296006764, "learning_rate": 6.9207111699883424e-06, "loss": 2.5816, "step": 40956 }, { "epoch": 2.542491774784282, "grad_norm": 0.4056586950734036, "learning_rate": 6.918878029330539e-06, "loss": 2.7179, "step": 40957 }, { "epoch": 2.54255385188404, "grad_norm": 0.40543797841109275, "learning_rate": 6.917045113437409e-06, "loss": 2.7343, "step": 40958 }, { "epoch": 2.542615928983798, "grad_norm": 0.4420230404633201, "learning_rate": 6.915212422318501e-06, "loss": 2.7068, "step": 40959 }, { "epoch": 2.542678006083556, "grad_norm": 0.4165321331588629, "learning_rate": 6.913379955983401e-06, "loss": 2.6679, "step": 40960 }, { "epoch": 2.542740083183314, "grad_norm": 0.4222831694198637, "learning_rate": 6.911547714441657e-06, "loss": 2.6747, "step": 40961 }, { "epoch": 2.5428021602830717, "grad_norm": 0.42263633413159374, "learning_rate": 6.909715697702824e-06, "loss": 2.6467, "step": 40962 }, { "epoch": 2.5428642373828296, "grad_norm": 0.40970054015030194, "learning_rate": 6.907883905776458e-06, "loss": 2.7253, "step": 40963 }, { "epoch": 2.542926314482587, "grad_norm": 0.4265305985758724, "learning_rate": 6.906052338672136e-06, "loss": 2.6756, "step": 40964 }, { "epoch": 2.5429883915823455, "grad_norm": 0.42867815939480436, "learning_rate": 6.904220996399397e-06, "loss": 2.7038, "step": 40965 }, { "epoch": 2.543050468682103, "grad_norm": 0.4523531489474636, "learning_rate": 6.902389878967796e-06, "loss": 2.6643, "step": 40966 }, { "epoch": 2.5431125457818613, "grad_norm": 0.40724822428148155, "learning_rate": 6.9005589863868956e-06, "loss": 2.6908, "step": 40967 }, { "epoch": 2.5431746228816188, "grad_norm": 0.45051500467340994, "learning_rate": 6.8987283186662254e-06, "loss": 2.6295, "step": 40968 }, { "epoch": 2.543236699981377, "grad_norm": 0.4189920643860708, "learning_rate": 6.896897875815367e-06, "loss": 2.7049, "step": 40969 }, { "epoch": 2.5432987770811346, "grad_norm": 0.44661598835567495, "learning_rate": 6.895067657843857e-06, "loss": 2.6169, "step": 40970 }, { "epoch": 2.5433608541808925, "grad_norm": 0.43001753530122705, "learning_rate": 6.893237664761243e-06, "loss": 2.7166, "step": 40971 }, { "epoch": 2.5434229312806504, "grad_norm": 0.39639568746325543, "learning_rate": 6.891407896577057e-06, "loss": 2.6463, "step": 40972 }, { "epoch": 2.5434850083804084, "grad_norm": 0.40047621393708555, "learning_rate": 6.8895783533008794e-06, "loss": 2.7684, "step": 40973 }, { "epoch": 2.5435470854801663, "grad_norm": 0.40877855618203646, "learning_rate": 6.887749034942232e-06, "loss": 2.7359, "step": 40974 }, { "epoch": 2.543609162579924, "grad_norm": 0.4648640091271979, "learning_rate": 6.885919941510655e-06, "loss": 2.6236, "step": 40975 }, { "epoch": 2.543671239679682, "grad_norm": 0.4533144456187095, "learning_rate": 6.884091073015708e-06, "loss": 2.7336, "step": 40976 }, { "epoch": 2.54373331677944, "grad_norm": 0.4075285160290956, "learning_rate": 6.882262429466924e-06, "loss": 2.7542, "step": 40977 }, { "epoch": 2.543795393879198, "grad_norm": 0.4121702248725266, "learning_rate": 6.880434010873837e-06, "loss": 2.627, "step": 40978 }, { "epoch": 2.543857470978956, "grad_norm": 0.42852381139427614, "learning_rate": 6.878605817246004e-06, "loss": 2.6738, "step": 40979 }, { "epoch": 2.543919548078714, "grad_norm": 0.40363506384808395, "learning_rate": 6.876777848592947e-06, "loss": 2.6774, "step": 40980 }, { "epoch": 2.5439816251784717, "grad_norm": 0.41178282686087786, "learning_rate": 6.874950104924216e-06, "loss": 2.7037, "step": 40981 }, { "epoch": 2.5440437022782296, "grad_norm": 0.41531002399863265, "learning_rate": 6.87312258624932e-06, "loss": 2.701, "step": 40982 }, { "epoch": 2.5441057793779875, "grad_norm": 0.43014683880471904, "learning_rate": 6.871295292577828e-06, "loss": 2.6368, "step": 40983 }, { "epoch": 2.5441678564777455, "grad_norm": 0.4011019087624052, "learning_rate": 6.869468223919262e-06, "loss": 2.715, "step": 40984 }, { "epoch": 2.5442299335775034, "grad_norm": 0.4487170527697271, "learning_rate": 6.867641380283146e-06, "loss": 2.6238, "step": 40985 }, { "epoch": 2.5442920106772613, "grad_norm": 0.4359950680475654, "learning_rate": 6.865814761679007e-06, "loss": 2.6628, "step": 40986 }, { "epoch": 2.544354087777019, "grad_norm": 0.426969584769864, "learning_rate": 6.8639883681163976e-06, "loss": 2.6252, "step": 40987 }, { "epoch": 2.544416164876777, "grad_norm": 0.4179348085116259, "learning_rate": 6.862162199604827e-06, "loss": 2.6813, "step": 40988 }, { "epoch": 2.5444782419765346, "grad_norm": 0.44115042235574625, "learning_rate": 6.860336256153827e-06, "loss": 2.7019, "step": 40989 }, { "epoch": 2.544540319076293, "grad_norm": 0.4038652520982448, "learning_rate": 6.858510537772933e-06, "loss": 2.7236, "step": 40990 }, { "epoch": 2.5446023961760504, "grad_norm": 0.418029994577452, "learning_rate": 6.856685044471644e-06, "loss": 2.6105, "step": 40991 }, { "epoch": 2.544664473275809, "grad_norm": 0.3975194189738455, "learning_rate": 6.854859776259515e-06, "loss": 2.7358, "step": 40992 }, { "epoch": 2.5447265503755663, "grad_norm": 0.44207309887526935, "learning_rate": 6.853034733146057e-06, "loss": 2.6571, "step": 40993 }, { "epoch": 2.544788627475324, "grad_norm": 0.47491705697338055, "learning_rate": 6.851209915140794e-06, "loss": 2.6959, "step": 40994 }, { "epoch": 2.544850704575082, "grad_norm": 0.42362471942549007, "learning_rate": 6.849385322253232e-06, "loss": 2.6809, "step": 40995 }, { "epoch": 2.54491278167484, "grad_norm": 0.41289615104102706, "learning_rate": 6.8475609544929096e-06, "loss": 2.6731, "step": 40996 }, { "epoch": 2.544974858774598, "grad_norm": 0.4600324959783095, "learning_rate": 6.8457368118693414e-06, "loss": 2.7333, "step": 40997 }, { "epoch": 2.545036935874356, "grad_norm": 0.46537410709972726, "learning_rate": 6.843912894392035e-06, "loss": 2.6244, "step": 40998 }, { "epoch": 2.545099012974114, "grad_norm": 0.4444120907762821, "learning_rate": 6.842089202070507e-06, "loss": 2.6183, "step": 40999 }, { "epoch": 2.5451610900738717, "grad_norm": 0.43643155080276896, "learning_rate": 6.8402657349142806e-06, "loss": 2.6965, "step": 41000 }, { "epoch": 2.5452231671736296, "grad_norm": 0.42886761559088815, "learning_rate": 6.838442492932867e-06, "loss": 2.6646, "step": 41001 }, { "epoch": 2.5452852442733875, "grad_norm": 0.4008022890275893, "learning_rate": 6.83661947613578e-06, "loss": 2.6259, "step": 41002 }, { "epoch": 2.5453473213731455, "grad_norm": 0.4051097677263287, "learning_rate": 6.834796684532529e-06, "loss": 2.6525, "step": 41003 }, { "epoch": 2.5454093984729034, "grad_norm": 0.43402761798431844, "learning_rate": 6.832974118132606e-06, "loss": 2.717, "step": 41004 }, { "epoch": 2.5454714755726613, "grad_norm": 0.43512606284941296, "learning_rate": 6.831151776945549e-06, "loss": 2.6692, "step": 41005 }, { "epoch": 2.545533552672419, "grad_norm": 0.4461524804464985, "learning_rate": 6.829329660980854e-06, "loss": 2.7639, "step": 41006 }, { "epoch": 2.545595629772177, "grad_norm": 0.3891814260760793, "learning_rate": 6.827507770248026e-06, "loss": 2.7306, "step": 41007 }, { "epoch": 2.545657706871935, "grad_norm": 0.4436004756136805, "learning_rate": 6.825686104756556e-06, "loss": 2.7562, "step": 41008 }, { "epoch": 2.545719783971693, "grad_norm": 0.43931048303483833, "learning_rate": 6.8238646645159665e-06, "loss": 2.7131, "step": 41009 }, { "epoch": 2.545781861071451, "grad_norm": 0.44500986490137817, "learning_rate": 6.82204344953577e-06, "loss": 2.6337, "step": 41010 }, { "epoch": 2.545843938171209, "grad_norm": 0.39719847089981175, "learning_rate": 6.820222459825449e-06, "loss": 2.67, "step": 41011 }, { "epoch": 2.5459060152709663, "grad_norm": 0.45846399879450833, "learning_rate": 6.81840169539451e-06, "loss": 2.616, "step": 41012 }, { "epoch": 2.5459680923707246, "grad_norm": 0.44820162853505136, "learning_rate": 6.816581156252444e-06, "loss": 2.661, "step": 41013 }, { "epoch": 2.546030169470482, "grad_norm": 0.42325829729192943, "learning_rate": 6.814760842408768e-06, "loss": 2.6034, "step": 41014 }, { "epoch": 2.5460922465702405, "grad_norm": 0.44501343104099356, "learning_rate": 6.812940753872965e-06, "loss": 2.7322, "step": 41015 }, { "epoch": 2.546154323669998, "grad_norm": 0.42643365835499825, "learning_rate": 6.811120890654538e-06, "loss": 2.6725, "step": 41016 }, { "epoch": 2.5462164007697563, "grad_norm": 0.431125989996309, "learning_rate": 6.8093012527629785e-06, "loss": 2.7302, "step": 41017 }, { "epoch": 2.546278477869514, "grad_norm": 0.41926971182663314, "learning_rate": 6.8074818402077634e-06, "loss": 2.6291, "step": 41018 }, { "epoch": 2.5463405549692717, "grad_norm": 0.4187031648549507, "learning_rate": 6.805662652998418e-06, "loss": 2.5876, "step": 41019 }, { "epoch": 2.5464026320690296, "grad_norm": 0.4205701392901886, "learning_rate": 6.8038436911444125e-06, "loss": 2.73, "step": 41020 }, { "epoch": 2.5464647091687875, "grad_norm": 0.4253222064579053, "learning_rate": 6.802024954655239e-06, "loss": 2.7046, "step": 41021 }, { "epoch": 2.5465267862685455, "grad_norm": 0.4081620921268925, "learning_rate": 6.80020644354038e-06, "loss": 2.7024, "step": 41022 }, { "epoch": 2.5465888633683034, "grad_norm": 0.4235559107906946, "learning_rate": 6.7983881578093376e-06, "loss": 2.708, "step": 41023 }, { "epoch": 2.5466509404680613, "grad_norm": 0.4847357237367674, "learning_rate": 6.7965700974716e-06, "loss": 2.6512, "step": 41024 }, { "epoch": 2.546713017567819, "grad_norm": 0.41141089234302697, "learning_rate": 6.794752262536636e-06, "loss": 2.6228, "step": 41025 }, { "epoch": 2.546775094667577, "grad_norm": 0.4382259612815908, "learning_rate": 6.792934653013944e-06, "loss": 2.7305, "step": 41026 }, { "epoch": 2.546837171767335, "grad_norm": 0.4338238766993067, "learning_rate": 6.7911172689129885e-06, "loss": 2.6297, "step": 41027 }, { "epoch": 2.546899248867093, "grad_norm": 0.4578805026693027, "learning_rate": 6.789300110243269e-06, "loss": 2.6485, "step": 41028 }, { "epoch": 2.546961325966851, "grad_norm": 0.41681084721177925, "learning_rate": 6.787483177014264e-06, "loss": 2.6509, "step": 41029 }, { "epoch": 2.547023403066609, "grad_norm": 0.42175722244220004, "learning_rate": 6.785666469235452e-06, "loss": 2.5769, "step": 41030 }, { "epoch": 2.5470854801663667, "grad_norm": 0.4140258808857491, "learning_rate": 6.7838499869162965e-06, "loss": 2.6598, "step": 41031 }, { "epoch": 2.5471475572661246, "grad_norm": 0.4559893673631762, "learning_rate": 6.782033730066295e-06, "loss": 2.7212, "step": 41032 }, { "epoch": 2.5472096343658825, "grad_norm": 0.45365238739721647, "learning_rate": 6.780217698694918e-06, "loss": 2.6697, "step": 41033 }, { "epoch": 2.5472717114656405, "grad_norm": 0.42603572625244635, "learning_rate": 6.778401892811637e-06, "loss": 2.5963, "step": 41034 }, { "epoch": 2.5473337885653984, "grad_norm": 0.42676236916210963, "learning_rate": 6.776586312425914e-06, "loss": 2.7089, "step": 41035 }, { "epoch": 2.5473958656651563, "grad_norm": 0.46740874088228346, "learning_rate": 6.774770957547244e-06, "loss": 2.7136, "step": 41036 }, { "epoch": 2.5474579427649138, "grad_norm": 0.4269528325281424, "learning_rate": 6.77295582818509e-06, "loss": 2.6713, "step": 41037 }, { "epoch": 2.547520019864672, "grad_norm": 0.4634307575056887, "learning_rate": 6.7711409243489175e-06, "loss": 2.5991, "step": 41038 }, { "epoch": 2.5475820969644296, "grad_norm": 0.46458225749052323, "learning_rate": 6.769326246048191e-06, "loss": 2.7249, "step": 41039 }, { "epoch": 2.547644174064188, "grad_norm": 0.38450863771322724, "learning_rate": 6.7675117932923804e-06, "loss": 2.7162, "step": 41040 }, { "epoch": 2.5477062511639454, "grad_norm": 0.44290235045303095, "learning_rate": 6.765697566090956e-06, "loss": 2.6555, "step": 41041 }, { "epoch": 2.5477683282637034, "grad_norm": 0.44750458776264085, "learning_rate": 6.763883564453389e-06, "loss": 2.6142, "step": 41042 }, { "epoch": 2.5478304053634613, "grad_norm": 0.46030554041680444, "learning_rate": 6.7620697883891415e-06, "loss": 2.6033, "step": 41043 }, { "epoch": 2.547892482463219, "grad_norm": 0.4299455041507694, "learning_rate": 6.760256237907669e-06, "loss": 2.7924, "step": 41044 }, { "epoch": 2.547954559562977, "grad_norm": 0.405566282063352, "learning_rate": 6.758442913018426e-06, "loss": 2.6856, "step": 41045 }, { "epoch": 2.548016636662735, "grad_norm": 0.40404493433865657, "learning_rate": 6.756629813730897e-06, "loss": 2.7324, "step": 41046 }, { "epoch": 2.548078713762493, "grad_norm": 0.4410110155782058, "learning_rate": 6.754816940054526e-06, "loss": 2.8059, "step": 41047 }, { "epoch": 2.548140790862251, "grad_norm": 0.43135691506085133, "learning_rate": 6.753004291998777e-06, "loss": 2.6331, "step": 41048 }, { "epoch": 2.548202867962009, "grad_norm": 0.4462169363558173, "learning_rate": 6.751191869573087e-06, "loss": 2.5567, "step": 41049 }, { "epoch": 2.5482649450617667, "grad_norm": 0.44595209000263764, "learning_rate": 6.749379672786943e-06, "loss": 2.7316, "step": 41050 }, { "epoch": 2.5483270221615246, "grad_norm": 0.42228532192932183, "learning_rate": 6.747567701649782e-06, "loss": 2.7108, "step": 41051 }, { "epoch": 2.5483890992612825, "grad_norm": 0.41634874930956106, "learning_rate": 6.745755956171057e-06, "loss": 2.6498, "step": 41052 }, { "epoch": 2.5484511763610405, "grad_norm": 0.4516583911758163, "learning_rate": 6.743944436360228e-06, "loss": 2.7127, "step": 41053 }, { "epoch": 2.5485132534607984, "grad_norm": 0.4426646229725985, "learning_rate": 6.742133142226731e-06, "loss": 2.6901, "step": 41054 }, { "epoch": 2.5485753305605563, "grad_norm": 0.44407332488361856, "learning_rate": 6.7403220737800364e-06, "loss": 2.7591, "step": 41055 }, { "epoch": 2.548637407660314, "grad_norm": 0.44122371444000447, "learning_rate": 6.7385112310295815e-06, "loss": 2.5851, "step": 41056 }, { "epoch": 2.548699484760072, "grad_norm": 0.44441812614493587, "learning_rate": 6.73670061398482e-06, "loss": 2.7253, "step": 41057 }, { "epoch": 2.54876156185983, "grad_norm": 0.4322996157945597, "learning_rate": 6.7348902226551825e-06, "loss": 2.6608, "step": 41058 }, { "epoch": 2.548823638959588, "grad_norm": 0.4407052913856649, "learning_rate": 6.733080057050134e-06, "loss": 2.7104, "step": 41059 }, { "epoch": 2.5488857160593454, "grad_norm": 0.44956185879416993, "learning_rate": 6.731270117179111e-06, "loss": 2.6311, "step": 41060 }, { "epoch": 2.548947793159104, "grad_norm": 0.4903744197378194, "learning_rate": 6.729460403051557e-06, "loss": 2.6872, "step": 41061 }, { "epoch": 2.5490098702588613, "grad_norm": 0.39374416513509664, "learning_rate": 6.7276509146769085e-06, "loss": 2.5828, "step": 41062 }, { "epoch": 2.5490719473586196, "grad_norm": 0.40504197094513483, "learning_rate": 6.7258416520646015e-06, "loss": 2.7103, "step": 41063 }, { "epoch": 2.549134024458377, "grad_norm": 0.44451835157724834, "learning_rate": 6.724032615224096e-06, "loss": 2.7267, "step": 41064 }, { "epoch": 2.5491961015581355, "grad_norm": 0.43300681727066576, "learning_rate": 6.7222238041648175e-06, "loss": 2.6799, "step": 41065 }, { "epoch": 2.549258178657893, "grad_norm": 0.4463374906679387, "learning_rate": 6.720415218896198e-06, "loss": 2.6358, "step": 41066 }, { "epoch": 2.549320255757651, "grad_norm": 0.4052709556411676, "learning_rate": 6.718606859427673e-06, "loss": 2.6504, "step": 41067 }, { "epoch": 2.549382332857409, "grad_norm": 0.3857262137280507, "learning_rate": 6.716798725768697e-06, "loss": 2.6249, "step": 41068 }, { "epoch": 2.5494444099571667, "grad_norm": 0.388969173498284, "learning_rate": 6.7149908179286845e-06, "loss": 2.5949, "step": 41069 }, { "epoch": 2.5495064870569246, "grad_norm": 0.408836639741346, "learning_rate": 6.713183135917073e-06, "loss": 2.6815, "step": 41070 }, { "epoch": 2.5495685641566825, "grad_norm": 0.4411746236000481, "learning_rate": 6.711375679743282e-06, "loss": 2.6879, "step": 41071 }, { "epoch": 2.5496306412564405, "grad_norm": 0.42506905583734683, "learning_rate": 6.709568449416764e-06, "loss": 2.6607, "step": 41072 }, { "epoch": 2.5496927183561984, "grad_norm": 0.42414185123824955, "learning_rate": 6.70776144494693e-06, "loss": 2.7097, "step": 41073 }, { "epoch": 2.5497547954559563, "grad_norm": 0.38630125841424096, "learning_rate": 6.705954666343222e-06, "loss": 2.7418, "step": 41074 }, { "epoch": 2.549816872555714, "grad_norm": 0.40652119314542434, "learning_rate": 6.7041481136150554e-06, "loss": 2.7106, "step": 41075 }, { "epoch": 2.549878949655472, "grad_norm": 0.427720039162275, "learning_rate": 6.7023417867718596e-06, "loss": 2.7201, "step": 41076 }, { "epoch": 2.54994102675523, "grad_norm": 0.4028655252114817, "learning_rate": 6.70053568582305e-06, "loss": 2.6922, "step": 41077 }, { "epoch": 2.550003103854988, "grad_norm": 0.4167628852692117, "learning_rate": 6.698729810778065e-06, "loss": 2.6657, "step": 41078 }, { "epoch": 2.550065180954746, "grad_norm": 0.4110232318173538, "learning_rate": 6.696924161646323e-06, "loss": 2.6486, "step": 41079 }, { "epoch": 2.550127258054504, "grad_norm": 0.3766979480444482, "learning_rate": 6.695118738437234e-06, "loss": 2.6507, "step": 41080 }, { "epoch": 2.5501893351542617, "grad_norm": 0.42114480722915504, "learning_rate": 6.693313541160218e-06, "loss": 2.7435, "step": 41081 }, { "epoch": 2.5502514122540196, "grad_norm": 0.4315827274761427, "learning_rate": 6.691508569824706e-06, "loss": 2.7873, "step": 41082 }, { "epoch": 2.5503134893537776, "grad_norm": 0.4034775434399551, "learning_rate": 6.689703824440102e-06, "loss": 2.6874, "step": 41083 }, { "epoch": 2.5503755664535355, "grad_norm": 0.4421733840468558, "learning_rate": 6.687899305015832e-06, "loss": 2.6284, "step": 41084 }, { "epoch": 2.550437643553293, "grad_norm": 0.4427457083090752, "learning_rate": 6.6860950115613e-06, "loss": 2.7066, "step": 41085 }, { "epoch": 2.5504997206530513, "grad_norm": 0.4086627289294249, "learning_rate": 6.6842909440859305e-06, "loss": 2.6605, "step": 41086 }, { "epoch": 2.550561797752809, "grad_norm": 0.3970005452587121, "learning_rate": 6.6824871025991276e-06, "loss": 2.6618, "step": 41087 }, { "epoch": 2.550623874852567, "grad_norm": 0.44517054721291094, "learning_rate": 6.680683487110312e-06, "loss": 2.5963, "step": 41088 }, { "epoch": 2.5506859519523246, "grad_norm": 0.416676618536675, "learning_rate": 6.678880097628881e-06, "loss": 2.6889, "step": 41089 }, { "epoch": 2.5507480290520825, "grad_norm": 0.43194677965399053, "learning_rate": 6.677076934164239e-06, "loss": 2.6484, "step": 41090 }, { "epoch": 2.5508101061518405, "grad_norm": 0.4289803147423428, "learning_rate": 6.675273996725817e-06, "loss": 2.688, "step": 41091 }, { "epoch": 2.5508721832515984, "grad_norm": 0.417940103137428, "learning_rate": 6.673471285323007e-06, "loss": 2.5956, "step": 41092 }, { "epoch": 2.5509342603513563, "grad_norm": 0.4123643384724875, "learning_rate": 6.6716687999652135e-06, "loss": 2.6394, "step": 41093 }, { "epoch": 2.550996337451114, "grad_norm": 0.4678623059298837, "learning_rate": 6.669866540661835e-06, "loss": 2.6959, "step": 41094 }, { "epoch": 2.551058414550872, "grad_norm": 0.4556046088617727, "learning_rate": 6.668064507422289e-06, "loss": 2.6636, "step": 41095 }, { "epoch": 2.55112049165063, "grad_norm": 0.4566139043686396, "learning_rate": 6.6662627002559766e-06, "loss": 2.71, "step": 41096 }, { "epoch": 2.551182568750388, "grad_norm": 0.4526388585966983, "learning_rate": 6.664461119172283e-06, "loss": 2.7464, "step": 41097 }, { "epoch": 2.551244645850146, "grad_norm": 0.41653925219122845, "learning_rate": 6.662659764180618e-06, "loss": 2.6456, "step": 41098 }, { "epoch": 2.551306722949904, "grad_norm": 0.4213847738245418, "learning_rate": 6.660858635290368e-06, "loss": 2.6859, "step": 41099 }, { "epoch": 2.5513688000496617, "grad_norm": 0.4095720399338981, "learning_rate": 6.659057732510948e-06, "loss": 2.6344, "step": 41100 }, { "epoch": 2.5514308771494196, "grad_norm": 0.4440570165468748, "learning_rate": 6.657257055851751e-06, "loss": 2.6824, "step": 41101 }, { "epoch": 2.5514929542491775, "grad_norm": 0.4594925035979716, "learning_rate": 6.655456605322158e-06, "loss": 2.7682, "step": 41102 }, { "epoch": 2.5515550313489355, "grad_norm": 0.390234465880665, "learning_rate": 6.653656380931561e-06, "loss": 2.666, "step": 41103 }, { "epoch": 2.5516171084486934, "grad_norm": 0.3829149965361216, "learning_rate": 6.651856382689376e-06, "loss": 2.7642, "step": 41104 }, { "epoch": 2.5516791855484513, "grad_norm": 0.3945195413495031, "learning_rate": 6.650056610604976e-06, "loss": 2.7306, "step": 41105 }, { "epoch": 2.551741262648209, "grad_norm": 0.43337716888047856, "learning_rate": 6.648257064687746e-06, "loss": 2.712, "step": 41106 }, { "epoch": 2.551803339747967, "grad_norm": 0.41171431928128976, "learning_rate": 6.646457744947094e-06, "loss": 2.6821, "step": 41107 }, { "epoch": 2.5518654168477246, "grad_norm": 0.386570179461321, "learning_rate": 6.644658651392382e-06, "loss": 2.6468, "step": 41108 }, { "epoch": 2.551927493947483, "grad_norm": 0.419419489346945, "learning_rate": 6.6428597840330235e-06, "loss": 2.6224, "step": 41109 }, { "epoch": 2.5519895710472404, "grad_norm": 0.4104840195120544, "learning_rate": 6.641061142878391e-06, "loss": 2.6751, "step": 41110 }, { "epoch": 2.552051648146999, "grad_norm": 0.41276330274982864, "learning_rate": 6.639262727937867e-06, "loss": 2.633, "step": 41111 }, { "epoch": 2.5521137252467563, "grad_norm": 0.4215411878872614, "learning_rate": 6.637464539220839e-06, "loss": 2.687, "step": 41112 }, { "epoch": 2.5521758023465146, "grad_norm": 0.4192327478142718, "learning_rate": 6.635666576736671e-06, "loss": 2.6755, "step": 41113 }, { "epoch": 2.552237879446272, "grad_norm": 0.40989031571871826, "learning_rate": 6.633868840494772e-06, "loss": 2.6683, "step": 41114 }, { "epoch": 2.55229995654603, "grad_norm": 0.4408493073020478, "learning_rate": 6.6320713305045025e-06, "loss": 2.5921, "step": 41115 }, { "epoch": 2.552362033645788, "grad_norm": 0.4571282862492207, "learning_rate": 6.630274046775248e-06, "loss": 2.8069, "step": 41116 }, { "epoch": 2.552424110745546, "grad_norm": 0.4587338839462408, "learning_rate": 6.628476989316368e-06, "loss": 2.682, "step": 41117 }, { "epoch": 2.552486187845304, "grad_norm": 0.4555073952351955, "learning_rate": 6.626680158137266e-06, "loss": 2.5662, "step": 41118 }, { "epoch": 2.5525482649450617, "grad_norm": 0.4068452069439356, "learning_rate": 6.624883553247302e-06, "loss": 2.7011, "step": 41119 }, { "epoch": 2.5526103420448196, "grad_norm": 0.42009564729781546, "learning_rate": 6.623087174655851e-06, "loss": 2.6522, "step": 41120 }, { "epoch": 2.5526724191445775, "grad_norm": 0.43091875014757075, "learning_rate": 6.621291022372272e-06, "loss": 2.6857, "step": 41121 }, { "epoch": 2.5527344962443355, "grad_norm": 0.44500979793134676, "learning_rate": 6.619495096405959e-06, "loss": 2.6043, "step": 41122 }, { "epoch": 2.5527965733440934, "grad_norm": 0.46832373629044016, "learning_rate": 6.617699396766269e-06, "loss": 2.6724, "step": 41123 }, { "epoch": 2.5528586504438513, "grad_norm": 0.424153901512633, "learning_rate": 6.6159039234625795e-06, "loss": 2.7156, "step": 41124 }, { "epoch": 2.552920727543609, "grad_norm": 0.4279618899243696, "learning_rate": 6.614108676504243e-06, "loss": 2.5833, "step": 41125 }, { "epoch": 2.552982804643367, "grad_norm": 0.4050519222955562, "learning_rate": 6.612313655900632e-06, "loss": 2.6272, "step": 41126 }, { "epoch": 2.553044881743125, "grad_norm": 0.441241680963783, "learning_rate": 6.610518861661114e-06, "loss": 2.7375, "step": 41127 }, { "epoch": 2.553106958842883, "grad_norm": 0.46263667677496134, "learning_rate": 6.60872429379506e-06, "loss": 2.7593, "step": 41128 }, { "epoch": 2.553169035942641, "grad_norm": 0.393819752827234, "learning_rate": 6.606929952311825e-06, "loss": 2.5998, "step": 41129 }, { "epoch": 2.553231113042399, "grad_norm": 0.4315317802075108, "learning_rate": 6.605135837220755e-06, "loss": 2.6588, "step": 41130 }, { "epoch": 2.5532931901421567, "grad_norm": 0.4481282307496386, "learning_rate": 6.603341948531239e-06, "loss": 2.6622, "step": 41131 }, { "epoch": 2.5533552672419146, "grad_norm": 0.4350705814101469, "learning_rate": 6.601548286252624e-06, "loss": 2.7033, "step": 41132 }, { "epoch": 2.553417344341672, "grad_norm": 0.4019803192756118, "learning_rate": 6.599754850394263e-06, "loss": 2.5062, "step": 41133 }, { "epoch": 2.5534794214414305, "grad_norm": 0.3973141335847884, "learning_rate": 6.597961640965511e-06, "loss": 2.7775, "step": 41134 }, { "epoch": 2.553541498541188, "grad_norm": 0.4283032053372157, "learning_rate": 6.596168657975738e-06, "loss": 2.7823, "step": 41135 }, { "epoch": 2.5536035756409463, "grad_norm": 0.43519144994522885, "learning_rate": 6.594375901434285e-06, "loss": 2.7193, "step": 41136 }, { "epoch": 2.553665652740704, "grad_norm": 0.421317809870078, "learning_rate": 6.592583371350513e-06, "loss": 2.7688, "step": 41137 }, { "epoch": 2.5537277298404617, "grad_norm": 0.4247077392796772, "learning_rate": 6.590791067733776e-06, "loss": 2.6881, "step": 41138 }, { "epoch": 2.5537898069402196, "grad_norm": 0.42258954464868587, "learning_rate": 6.588998990593403e-06, "loss": 2.5763, "step": 41139 }, { "epoch": 2.5538518840399775, "grad_norm": 0.4126641214321251, "learning_rate": 6.587207139938767e-06, "loss": 2.6354, "step": 41140 }, { "epoch": 2.5539139611397355, "grad_norm": 0.41240491710440796, "learning_rate": 6.585415515779214e-06, "loss": 2.6562, "step": 41141 }, { "epoch": 2.5539760382394934, "grad_norm": 0.40563472592136224, "learning_rate": 6.583624118124093e-06, "loss": 2.6545, "step": 41142 }, { "epoch": 2.5540381153392513, "grad_norm": 0.4150816233634892, "learning_rate": 6.581832946982741e-06, "loss": 2.638, "step": 41143 }, { "epoch": 2.554100192439009, "grad_norm": 0.42258243938718343, "learning_rate": 6.580042002364495e-06, "loss": 2.6374, "step": 41144 }, { "epoch": 2.554162269538767, "grad_norm": 0.43568107008796303, "learning_rate": 6.578251284278725e-06, "loss": 2.7179, "step": 41145 }, { "epoch": 2.554224346638525, "grad_norm": 0.4034084750193975, "learning_rate": 6.576460792734757e-06, "loss": 2.6502, "step": 41146 }, { "epoch": 2.554286423738283, "grad_norm": 0.41952812008959983, "learning_rate": 6.5746705277419384e-06, "loss": 2.6322, "step": 41147 }, { "epoch": 2.554348500838041, "grad_norm": 0.4050429458571803, "learning_rate": 6.572880489309602e-06, "loss": 2.7406, "step": 41148 }, { "epoch": 2.554410577937799, "grad_norm": 0.3938109365873266, "learning_rate": 6.571090677447078e-06, "loss": 2.7159, "step": 41149 }, { "epoch": 2.5544726550375567, "grad_norm": 0.40060841289467075, "learning_rate": 6.5693010921637265e-06, "loss": 2.625, "step": 41150 }, { "epoch": 2.5545347321373146, "grad_norm": 0.42375966587126573, "learning_rate": 6.567511733468879e-06, "loss": 2.7472, "step": 41151 }, { "epoch": 2.5545968092370726, "grad_norm": 0.4173587911759632, "learning_rate": 6.565722601371865e-06, "loss": 2.696, "step": 41152 }, { "epoch": 2.5546588863368305, "grad_norm": 0.38744174919376695, "learning_rate": 6.563933695882008e-06, "loss": 2.6765, "step": 41153 }, { "epoch": 2.5547209634365884, "grad_norm": 0.40764376997349544, "learning_rate": 6.562145017008664e-06, "loss": 2.6598, "step": 41154 }, { "epoch": 2.5547830405363463, "grad_norm": 0.4336850010159021, "learning_rate": 6.560356564761155e-06, "loss": 2.685, "step": 41155 }, { "epoch": 2.554845117636104, "grad_norm": 0.39775252834861574, "learning_rate": 6.5585683391488125e-06, "loss": 2.6569, "step": 41156 }, { "epoch": 2.554907194735862, "grad_norm": 0.4299558148581893, "learning_rate": 6.5567803401809515e-06, "loss": 2.687, "step": 41157 }, { "epoch": 2.5549692718356196, "grad_norm": 0.41541049269077396, "learning_rate": 6.55499256786693e-06, "loss": 2.686, "step": 41158 }, { "epoch": 2.555031348935378, "grad_norm": 0.48602026268334986, "learning_rate": 6.553205022216053e-06, "loss": 2.7106, "step": 41159 }, { "epoch": 2.5550934260351355, "grad_norm": 0.4460727325939825, "learning_rate": 6.5514177032376554e-06, "loss": 2.6712, "step": 41160 }, { "epoch": 2.555155503134894, "grad_norm": 0.397654380868892, "learning_rate": 6.54963061094106e-06, "loss": 2.6971, "step": 41161 }, { "epoch": 2.5552175802346513, "grad_norm": 0.4313943082433972, "learning_rate": 6.547843745335575e-06, "loss": 2.7031, "step": 41162 }, { "epoch": 2.555279657334409, "grad_norm": 0.41873569179401743, "learning_rate": 6.546057106430548e-06, "loss": 2.5967, "step": 41163 }, { "epoch": 2.555341734434167, "grad_norm": 0.41595300910632554, "learning_rate": 6.544270694235294e-06, "loss": 2.5856, "step": 41164 }, { "epoch": 2.555403811533925, "grad_norm": 0.41060241773569883, "learning_rate": 6.542484508759128e-06, "loss": 2.7913, "step": 41165 }, { "epoch": 2.555465888633683, "grad_norm": 0.4163607746077409, "learning_rate": 6.5406985500113595e-06, "loss": 2.6293, "step": 41166 }, { "epoch": 2.555527965733441, "grad_norm": 0.4046487428818858, "learning_rate": 6.538912818001325e-06, "loss": 2.6819, "step": 41167 }, { "epoch": 2.555590042833199, "grad_norm": 0.43959214448761064, "learning_rate": 6.537127312738334e-06, "loss": 2.6976, "step": 41168 }, { "epoch": 2.5556521199329567, "grad_norm": 0.43813529890855485, "learning_rate": 6.5353420342317005e-06, "loss": 2.6102, "step": 41169 }, { "epoch": 2.5557141970327146, "grad_norm": 0.427701974974187, "learning_rate": 6.53355698249073e-06, "loss": 2.5719, "step": 41170 }, { "epoch": 2.5557762741324725, "grad_norm": 0.39810067319580705, "learning_rate": 6.531772157524741e-06, "loss": 2.7073, "step": 41171 }, { "epoch": 2.5558383512322305, "grad_norm": 0.4239032520563603, "learning_rate": 6.529987559343065e-06, "loss": 2.6409, "step": 41172 }, { "epoch": 2.5559004283319884, "grad_norm": 0.4339932843383107, "learning_rate": 6.528203187954995e-06, "loss": 2.7201, "step": 41173 }, { "epoch": 2.5559625054317463, "grad_norm": 0.42852619335457276, "learning_rate": 6.526419043369841e-06, "loss": 2.6729, "step": 41174 }, { "epoch": 2.556024582531504, "grad_norm": 0.41844526205349053, "learning_rate": 6.524635125596912e-06, "loss": 2.7548, "step": 41175 }, { "epoch": 2.556086659631262, "grad_norm": 0.36885160404160083, "learning_rate": 6.522851434645505e-06, "loss": 2.6518, "step": 41176 }, { "epoch": 2.55614873673102, "grad_norm": 0.4313127644793312, "learning_rate": 6.521067970524947e-06, "loss": 2.8098, "step": 41177 }, { "epoch": 2.556210813830778, "grad_norm": 0.41582352048855753, "learning_rate": 6.519284733244529e-06, "loss": 2.6288, "step": 41178 }, { "epoch": 2.556272890930536, "grad_norm": 0.3859810005413876, "learning_rate": 6.517501722813562e-06, "loss": 2.7658, "step": 41179 }, { "epoch": 2.556334968030294, "grad_norm": 0.4233602235105902, "learning_rate": 6.515718939241333e-06, "loss": 2.6664, "step": 41180 }, { "epoch": 2.5563970451300513, "grad_norm": 0.4192741899375739, "learning_rate": 6.513936382537167e-06, "loss": 2.6199, "step": 41181 }, { "epoch": 2.5564591222298096, "grad_norm": 0.45625021620967393, "learning_rate": 6.5121540527103505e-06, "loss": 2.5439, "step": 41182 }, { "epoch": 2.556521199329567, "grad_norm": 0.4075210019123262, "learning_rate": 6.5103719497701835e-06, "loss": 2.6399, "step": 41183 }, { "epoch": 2.5565832764293255, "grad_norm": 0.4072296544970763, "learning_rate": 6.508590073725951e-06, "loss": 2.6914, "step": 41184 }, { "epoch": 2.556645353529083, "grad_norm": 0.4060335132679954, "learning_rate": 6.5068084245869745e-06, "loss": 2.6806, "step": 41185 }, { "epoch": 2.556707430628841, "grad_norm": 0.43513024079482526, "learning_rate": 6.505027002362535e-06, "loss": 2.6762, "step": 41186 }, { "epoch": 2.556769507728599, "grad_norm": 0.41933101514353405, "learning_rate": 6.5032458070619305e-06, "loss": 2.6303, "step": 41187 }, { "epoch": 2.5568315848283567, "grad_norm": 0.48351543821601506, "learning_rate": 6.5014648386944485e-06, "loss": 2.6996, "step": 41188 }, { "epoch": 2.5568936619281146, "grad_norm": 0.4164912768614799, "learning_rate": 6.499684097269376e-06, "loss": 2.5837, "step": 41189 }, { "epoch": 2.5569557390278725, "grad_norm": 0.41175789330310425, "learning_rate": 6.497903582796022e-06, "loss": 2.6872, "step": 41190 }, { "epoch": 2.5570178161276305, "grad_norm": 0.4083197030238339, "learning_rate": 6.496123295283663e-06, "loss": 2.6345, "step": 41191 }, { "epoch": 2.5570798932273884, "grad_norm": 0.42064538770242105, "learning_rate": 6.494343234741596e-06, "loss": 2.6357, "step": 41192 }, { "epoch": 2.5571419703271463, "grad_norm": 0.42720980244862056, "learning_rate": 6.492563401179086e-06, "loss": 2.7312, "step": 41193 }, { "epoch": 2.557204047426904, "grad_norm": 0.40771922948614214, "learning_rate": 6.490783794605443e-06, "loss": 2.6599, "step": 41194 }, { "epoch": 2.557266124526662, "grad_norm": 0.4382327449021703, "learning_rate": 6.489004415029948e-06, "loss": 2.5799, "step": 41195 }, { "epoch": 2.55732820162642, "grad_norm": 0.3971085544527422, "learning_rate": 6.487225262461877e-06, "loss": 2.691, "step": 41196 }, { "epoch": 2.557390278726178, "grad_norm": 0.4187142862206855, "learning_rate": 6.4854463369105165e-06, "loss": 2.6504, "step": 41197 }, { "epoch": 2.557452355825936, "grad_norm": 0.39989524527597015, "learning_rate": 6.483667638385138e-06, "loss": 2.6877, "step": 41198 }, { "epoch": 2.557514432925694, "grad_norm": 0.42788270442555704, "learning_rate": 6.481889166895033e-06, "loss": 2.6645, "step": 41199 }, { "epoch": 2.5575765100254517, "grad_norm": 0.4169672438153968, "learning_rate": 6.4801109224494836e-06, "loss": 2.6643, "step": 41200 }, { "epoch": 2.5576385871252096, "grad_norm": 0.39649570618020635, "learning_rate": 6.478332905057755e-06, "loss": 2.6418, "step": 41201 }, { "epoch": 2.5577006642249676, "grad_norm": 0.44846009586307684, "learning_rate": 6.476555114729121e-06, "loss": 2.6437, "step": 41202 }, { "epoch": 2.5577627413247255, "grad_norm": 0.46064763734282616, "learning_rate": 6.474777551472877e-06, "loss": 2.7126, "step": 41203 }, { "epoch": 2.557824818424483, "grad_norm": 0.4000671568761754, "learning_rate": 6.473000215298269e-06, "loss": 2.6612, "step": 41204 }, { "epoch": 2.5578868955242413, "grad_norm": 0.4039965818173206, "learning_rate": 6.471223106214602e-06, "loss": 2.7013, "step": 41205 }, { "epoch": 2.557948972623999, "grad_norm": 0.4033001393591056, "learning_rate": 6.469446224231124e-06, "loss": 2.6468, "step": 41206 }, { "epoch": 2.558011049723757, "grad_norm": 0.44641914391781246, "learning_rate": 6.46766956935711e-06, "loss": 2.6729, "step": 41207 }, { "epoch": 2.5580731268235146, "grad_norm": 0.4236646591886045, "learning_rate": 6.465893141601836e-06, "loss": 2.6114, "step": 41208 }, { "epoch": 2.558135203923273, "grad_norm": 0.4333834094983012, "learning_rate": 6.464116940974569e-06, "loss": 2.6644, "step": 41209 }, { "epoch": 2.5581972810230305, "grad_norm": 0.41810995907380794, "learning_rate": 6.462340967484571e-06, "loss": 2.6764, "step": 41210 }, { "epoch": 2.5582593581227884, "grad_norm": 0.43455565352843695, "learning_rate": 6.460565221141107e-06, "loss": 2.7203, "step": 41211 }, { "epoch": 2.5583214352225463, "grad_norm": 0.3834609359195, "learning_rate": 6.458789701953439e-06, "loss": 2.7689, "step": 41212 }, { "epoch": 2.558383512322304, "grad_norm": 0.4354666221136877, "learning_rate": 6.4570144099308395e-06, "loss": 2.7048, "step": 41213 }, { "epoch": 2.558445589422062, "grad_norm": 0.4213623358080374, "learning_rate": 6.45523934508257e-06, "loss": 2.7043, "step": 41214 }, { "epoch": 2.55850766652182, "grad_norm": 0.40912354660009126, "learning_rate": 6.453464507417889e-06, "loss": 2.7404, "step": 41215 }, { "epoch": 2.558569743621578, "grad_norm": 0.40373721985327876, "learning_rate": 6.4516898969460394e-06, "loss": 2.7142, "step": 41216 }, { "epoch": 2.558631820721336, "grad_norm": 0.4145538545636134, "learning_rate": 6.449915513676308e-06, "loss": 2.6696, "step": 41217 }, { "epoch": 2.558693897821094, "grad_norm": 0.4081819702215446, "learning_rate": 6.448141357617943e-06, "loss": 2.6995, "step": 41218 }, { "epoch": 2.5587559749208517, "grad_norm": 0.4178299539725083, "learning_rate": 6.4463674287801925e-06, "loss": 2.6474, "step": 41219 }, { "epoch": 2.5588180520206096, "grad_norm": 0.4094915180515595, "learning_rate": 6.44459372717231e-06, "loss": 2.6434, "step": 41220 }, { "epoch": 2.5588801291203676, "grad_norm": 0.4284284177801091, "learning_rate": 6.44282025280356e-06, "loss": 2.6863, "step": 41221 }, { "epoch": 2.5589422062201255, "grad_norm": 0.40674007639769294, "learning_rate": 6.441047005683198e-06, "loss": 2.5752, "step": 41222 }, { "epoch": 2.5590042833198834, "grad_norm": 0.3968095244817167, "learning_rate": 6.439273985820465e-06, "loss": 2.5998, "step": 41223 }, { "epoch": 2.5590663604196413, "grad_norm": 0.4042434104177114, "learning_rate": 6.4375011932246155e-06, "loss": 2.6777, "step": 41224 }, { "epoch": 2.5591284375193992, "grad_norm": 0.42485079460294206, "learning_rate": 6.43572862790488e-06, "loss": 2.6258, "step": 41225 }, { "epoch": 2.559190514619157, "grad_norm": 0.4312226528674206, "learning_rate": 6.4339562898705405e-06, "loss": 2.5942, "step": 41226 }, { "epoch": 2.5592525917189146, "grad_norm": 0.38943707080071127, "learning_rate": 6.432184179130829e-06, "loss": 2.6556, "step": 41227 }, { "epoch": 2.559314668818673, "grad_norm": 0.40475804292181566, "learning_rate": 6.430412295694988e-06, "loss": 2.5937, "step": 41228 }, { "epoch": 2.5593767459184305, "grad_norm": 0.4453929360006398, "learning_rate": 6.428640639572247e-06, "loss": 2.6549, "step": 41229 }, { "epoch": 2.559438823018189, "grad_norm": 0.4239538683019609, "learning_rate": 6.42686921077188e-06, "loss": 2.6074, "step": 41230 }, { "epoch": 2.5595009001179463, "grad_norm": 0.40463600125587484, "learning_rate": 6.42509800930311e-06, "loss": 2.6665, "step": 41231 }, { "epoch": 2.5595629772177046, "grad_norm": 0.41845818856612316, "learning_rate": 6.423327035175186e-06, "loss": 2.6745, "step": 41232 }, { "epoch": 2.559625054317462, "grad_norm": 0.4193300556823972, "learning_rate": 6.421556288397346e-06, "loss": 2.6005, "step": 41233 }, { "epoch": 2.55968713141722, "grad_norm": 0.4487451959793817, "learning_rate": 6.419785768978808e-06, "loss": 2.6587, "step": 41234 }, { "epoch": 2.559749208516978, "grad_norm": 0.3982197970858742, "learning_rate": 6.418015476928846e-06, "loss": 2.715, "step": 41235 }, { "epoch": 2.559811285616736, "grad_norm": 0.4165170718903938, "learning_rate": 6.416245412256672e-06, "loss": 2.6747, "step": 41236 }, { "epoch": 2.559873362716494, "grad_norm": 0.40881342188782666, "learning_rate": 6.414475574971513e-06, "loss": 2.6685, "step": 41237 }, { "epoch": 2.5599354398162517, "grad_norm": 0.42869371453449817, "learning_rate": 6.412705965082632e-06, "loss": 2.7218, "step": 41238 }, { "epoch": 2.5599975169160096, "grad_norm": 0.40208595312908874, "learning_rate": 6.410936582599231e-06, "loss": 2.6764, "step": 41239 }, { "epoch": 2.5600595940157675, "grad_norm": 0.4233804087139912, "learning_rate": 6.4091674275305654e-06, "loss": 2.6508, "step": 41240 }, { "epoch": 2.5601216711155255, "grad_norm": 0.3973725804149958, "learning_rate": 6.407398499885858e-06, "loss": 2.6255, "step": 41241 }, { "epoch": 2.5601837482152834, "grad_norm": 0.4293520224696292, "learning_rate": 6.405629799674334e-06, "loss": 2.6617, "step": 41242 }, { "epoch": 2.5602458253150413, "grad_norm": 0.4170737264989535, "learning_rate": 6.403861326905214e-06, "loss": 2.6335, "step": 41243 }, { "epoch": 2.560307902414799, "grad_norm": 0.42832091366664377, "learning_rate": 6.40209308158774e-06, "loss": 2.6521, "step": 41244 }, { "epoch": 2.560369979514557, "grad_norm": 0.41980024838569713, "learning_rate": 6.400325063731133e-06, "loss": 2.6389, "step": 41245 }, { "epoch": 2.560432056614315, "grad_norm": 0.4410116913504315, "learning_rate": 6.398557273344613e-06, "loss": 2.6315, "step": 41246 }, { "epoch": 2.560494133714073, "grad_norm": 0.41602418559404847, "learning_rate": 6.396789710437401e-06, "loss": 2.7232, "step": 41247 }, { "epoch": 2.560556210813831, "grad_norm": 0.389389047353982, "learning_rate": 6.395022375018716e-06, "loss": 2.6324, "step": 41248 }, { "epoch": 2.560618287913589, "grad_norm": 0.4084255397420164, "learning_rate": 6.3932552670977915e-06, "loss": 2.6429, "step": 41249 }, { "epoch": 2.5606803650133467, "grad_norm": 0.4325890801041351, "learning_rate": 6.391488386683842e-06, "loss": 2.594, "step": 41250 }, { "epoch": 2.5607424421131046, "grad_norm": 0.44492137195535275, "learning_rate": 6.389721733786086e-06, "loss": 2.5871, "step": 41251 }, { "epoch": 2.560804519212862, "grad_norm": 0.4146524041911931, "learning_rate": 6.387955308413729e-06, "loss": 2.6642, "step": 41252 }, { "epoch": 2.5608665963126205, "grad_norm": 0.42826303707060287, "learning_rate": 6.3861891105760015e-06, "loss": 2.6415, "step": 41253 }, { "epoch": 2.560928673412378, "grad_norm": 0.43094679359777227, "learning_rate": 6.384423140282114e-06, "loss": 2.7145, "step": 41254 }, { "epoch": 2.5609907505121363, "grad_norm": 0.4119911747503415, "learning_rate": 6.3826573975412806e-06, "loss": 2.6691, "step": 41255 }, { "epoch": 2.561052827611894, "grad_norm": 0.41879420901127673, "learning_rate": 6.380891882362705e-06, "loss": 2.6222, "step": 41256 }, { "epoch": 2.5611149047116517, "grad_norm": 0.39166609310892736, "learning_rate": 6.379126594755614e-06, "loss": 2.649, "step": 41257 }, { "epoch": 2.5611769818114096, "grad_norm": 0.4225541934689443, "learning_rate": 6.377361534729204e-06, "loss": 2.6625, "step": 41258 }, { "epoch": 2.5612390589111675, "grad_norm": 0.38819789260344323, "learning_rate": 6.375596702292691e-06, "loss": 2.7381, "step": 41259 }, { "epoch": 2.5613011360109255, "grad_norm": 0.375322064540683, "learning_rate": 6.373832097455284e-06, "loss": 2.7528, "step": 41260 }, { "epoch": 2.5613632131106834, "grad_norm": 0.4020323983808737, "learning_rate": 6.3720677202261715e-06, "loss": 2.5492, "step": 41261 }, { "epoch": 2.5614252902104413, "grad_norm": 0.4776079717114991, "learning_rate": 6.370303570614583e-06, "loss": 2.6349, "step": 41262 }, { "epoch": 2.561487367310199, "grad_norm": 0.42374524831217614, "learning_rate": 6.368539648629718e-06, "loss": 2.7662, "step": 41263 }, { "epoch": 2.561549444409957, "grad_norm": 0.4210527318676716, "learning_rate": 6.366775954280768e-06, "loss": 2.549, "step": 41264 }, { "epoch": 2.561611521509715, "grad_norm": 0.46182385586049784, "learning_rate": 6.365012487576927e-06, "loss": 2.6631, "step": 41265 }, { "epoch": 2.561673598609473, "grad_norm": 0.4337405222701028, "learning_rate": 6.363249248527426e-06, "loss": 2.6348, "step": 41266 }, { "epoch": 2.561735675709231, "grad_norm": 0.4080346594889567, "learning_rate": 6.36148623714144e-06, "loss": 2.645, "step": 41267 }, { "epoch": 2.561797752808989, "grad_norm": 0.44039911811114063, "learning_rate": 6.359723453428179e-06, "loss": 2.6156, "step": 41268 }, { "epoch": 2.5618598299087467, "grad_norm": 0.4431212537792772, "learning_rate": 6.3579608973968255e-06, "loss": 2.695, "step": 41269 }, { "epoch": 2.5619219070085046, "grad_norm": 0.44265208354460883, "learning_rate": 6.356198569056581e-06, "loss": 2.7193, "step": 41270 }, { "epoch": 2.5619839841082626, "grad_norm": 0.42871602952579074, "learning_rate": 6.354436468416652e-06, "loss": 2.6497, "step": 41271 }, { "epoch": 2.5620460612080205, "grad_norm": 0.4102481375808331, "learning_rate": 6.3526745954862245e-06, "loss": 2.6302, "step": 41272 }, { "epoch": 2.5621081383077784, "grad_norm": 0.44026258728558487, "learning_rate": 6.350912950274496e-06, "loss": 2.7466, "step": 41273 }, { "epoch": 2.5621702154075363, "grad_norm": 0.42220793055549627, "learning_rate": 6.349151532790643e-06, "loss": 2.7258, "step": 41274 }, { "epoch": 2.562232292507294, "grad_norm": 0.462010411865605, "learning_rate": 6.3473903430438575e-06, "loss": 2.6874, "step": 41275 }, { "epoch": 2.562294369607052, "grad_norm": 0.4106643436111247, "learning_rate": 6.345629381043339e-06, "loss": 2.7357, "step": 41276 }, { "epoch": 2.5623564467068096, "grad_norm": 0.41822197632014396, "learning_rate": 6.343868646798273e-06, "loss": 2.6525, "step": 41277 }, { "epoch": 2.562418523806568, "grad_norm": 0.4443097452681171, "learning_rate": 6.3421081403178424e-06, "loss": 2.537, "step": 41278 }, { "epoch": 2.5624806009063255, "grad_norm": 0.4071793563157227, "learning_rate": 6.340347861611218e-06, "loss": 2.6584, "step": 41279 }, { "epoch": 2.562542678006084, "grad_norm": 0.41180063045137566, "learning_rate": 6.338587810687613e-06, "loss": 2.7498, "step": 41280 }, { "epoch": 2.5626047551058413, "grad_norm": 0.4389070805955621, "learning_rate": 6.336827987556193e-06, "loss": 2.6943, "step": 41281 }, { "epoch": 2.562666832205599, "grad_norm": 0.42228410452777954, "learning_rate": 6.33506839222614e-06, "loss": 2.6588, "step": 41282 }, { "epoch": 2.562728909305357, "grad_norm": 0.3994259447787967, "learning_rate": 6.333309024706635e-06, "loss": 2.657, "step": 41283 }, { "epoch": 2.562790986405115, "grad_norm": 0.40171935880801085, "learning_rate": 6.331549885006849e-06, "loss": 2.6679, "step": 41284 }, { "epoch": 2.562853063504873, "grad_norm": 0.43811327661455624, "learning_rate": 6.329790973135974e-06, "loss": 2.6937, "step": 41285 }, { "epoch": 2.562915140604631, "grad_norm": 0.43273290460009256, "learning_rate": 6.328032289103186e-06, "loss": 2.6441, "step": 41286 }, { "epoch": 2.562977217704389, "grad_norm": 0.41679910303465967, "learning_rate": 6.326273832917651e-06, "loss": 2.6338, "step": 41287 }, { "epoch": 2.5630392948041467, "grad_norm": 0.3968225925115618, "learning_rate": 6.324515604588538e-06, "loss": 2.7124, "step": 41288 }, { "epoch": 2.5631013719039046, "grad_norm": 0.40519651119503897, "learning_rate": 6.3227576041250456e-06, "loss": 2.6873, "step": 41289 }, { "epoch": 2.5631634490036626, "grad_norm": 0.43099360929915637, "learning_rate": 6.320999831536323e-06, "loss": 2.7458, "step": 41290 }, { "epoch": 2.5632255261034205, "grad_norm": 0.4043001184010904, "learning_rate": 6.319242286831551e-06, "loss": 2.6188, "step": 41291 }, { "epoch": 2.5632876032031784, "grad_norm": 0.4200425085626856, "learning_rate": 6.317484970019882e-06, "loss": 2.6796, "step": 41292 }, { "epoch": 2.5633496803029363, "grad_norm": 0.3963133720360682, "learning_rate": 6.315727881110512e-06, "loss": 2.5379, "step": 41293 }, { "epoch": 2.5634117574026942, "grad_norm": 0.41364767780083256, "learning_rate": 6.3139710201125964e-06, "loss": 2.6867, "step": 41294 }, { "epoch": 2.563473834502452, "grad_norm": 0.3839426097489486, "learning_rate": 6.3122143870352926e-06, "loss": 2.6008, "step": 41295 }, { "epoch": 2.56353591160221, "grad_norm": 0.42738392377799045, "learning_rate": 6.310457981887779e-06, "loss": 2.6278, "step": 41296 }, { "epoch": 2.563597988701968, "grad_norm": 0.4153947271173554, "learning_rate": 6.3087018046791955e-06, "loss": 2.7318, "step": 41297 }, { "epoch": 2.563660065801726, "grad_norm": 0.39276710630347644, "learning_rate": 6.306945855418733e-06, "loss": 2.5212, "step": 41298 }, { "epoch": 2.563722142901484, "grad_norm": 0.43125579042279044, "learning_rate": 6.3051901341155465e-06, "loss": 2.6794, "step": 41299 }, { "epoch": 2.5637842200012413, "grad_norm": 0.42811835485717, "learning_rate": 6.303434640778782e-06, "loss": 2.6496, "step": 41300 }, { "epoch": 2.5638462971009996, "grad_norm": 0.44039517624800595, "learning_rate": 6.301679375417602e-06, "loss": 2.6812, "step": 41301 }, { "epoch": 2.563908374200757, "grad_norm": 0.4438832922332544, "learning_rate": 6.299924338041169e-06, "loss": 2.6827, "step": 41302 }, { "epoch": 2.5639704513005155, "grad_norm": 0.49730315684849385, "learning_rate": 6.298169528658648e-06, "loss": 2.6954, "step": 41303 }, { "epoch": 2.564032528400273, "grad_norm": 0.4244898188215918, "learning_rate": 6.296414947279183e-06, "loss": 2.6272, "step": 41304 }, { "epoch": 2.564094605500031, "grad_norm": 0.4208452762035904, "learning_rate": 6.294660593911933e-06, "loss": 2.662, "step": 41305 }, { "epoch": 2.564156682599789, "grad_norm": 0.46197431949565565, "learning_rate": 6.29290646856604e-06, "loss": 2.6871, "step": 41306 }, { "epoch": 2.5642187596995467, "grad_norm": 0.4125254189722347, "learning_rate": 6.291152571250674e-06, "loss": 2.6758, "step": 41307 }, { "epoch": 2.5642808367993046, "grad_norm": 0.42297819501169653, "learning_rate": 6.289398901974974e-06, "loss": 2.7378, "step": 41308 }, { "epoch": 2.5643429138990625, "grad_norm": 0.4777522946955868, "learning_rate": 6.287645460748093e-06, "loss": 2.6707, "step": 41309 }, { "epoch": 2.5644049909988205, "grad_norm": 0.4179848462864659, "learning_rate": 6.285892247579173e-06, "loss": 2.6128, "step": 41310 }, { "epoch": 2.5644670680985784, "grad_norm": 0.43358957443288737, "learning_rate": 6.2841392624773564e-06, "loss": 2.6568, "step": 41311 }, { "epoch": 2.5645291451983363, "grad_norm": 0.43799877685763655, "learning_rate": 6.28238650545181e-06, "loss": 2.6879, "step": 41312 }, { "epoch": 2.564591222298094, "grad_norm": 0.414608810910333, "learning_rate": 6.280633976511668e-06, "loss": 2.6688, "step": 41313 }, { "epoch": 2.564653299397852, "grad_norm": 0.4431448261899402, "learning_rate": 6.278881675666065e-06, "loss": 2.6581, "step": 41314 }, { "epoch": 2.56471537649761, "grad_norm": 0.4451928396240811, "learning_rate": 6.277129602924148e-06, "loss": 2.7828, "step": 41315 }, { "epoch": 2.564777453597368, "grad_norm": 0.40662956845718906, "learning_rate": 6.275377758295065e-06, "loss": 2.6788, "step": 41316 }, { "epoch": 2.564839530697126, "grad_norm": 0.4255970197602846, "learning_rate": 6.273626141787953e-06, "loss": 2.6975, "step": 41317 }, { "epoch": 2.564901607796884, "grad_norm": 0.43564804681156005, "learning_rate": 6.271874753411944e-06, "loss": 2.7376, "step": 41318 }, { "epoch": 2.5649636848966417, "grad_norm": 0.4060361923108403, "learning_rate": 6.27012359317618e-06, "loss": 2.7478, "step": 41319 }, { "epoch": 2.5650257619963996, "grad_norm": 0.432433250711809, "learning_rate": 6.268372661089794e-06, "loss": 2.5975, "step": 41320 }, { "epoch": 2.5650878390961576, "grad_norm": 0.42270504569324197, "learning_rate": 6.2666219571619275e-06, "loss": 2.7277, "step": 41321 }, { "epoch": 2.5651499161959155, "grad_norm": 0.41202104896529096, "learning_rate": 6.264871481401713e-06, "loss": 2.6047, "step": 41322 }, { "epoch": 2.565211993295673, "grad_norm": 0.4148848123576742, "learning_rate": 6.263121233818281e-06, "loss": 2.6773, "step": 41323 }, { "epoch": 2.5652740703954313, "grad_norm": 0.4330287035288888, "learning_rate": 6.2613712144207525e-06, "loss": 2.6743, "step": 41324 }, { "epoch": 2.565336147495189, "grad_norm": 0.40321121402838983, "learning_rate": 6.259621423218282e-06, "loss": 2.737, "step": 41325 }, { "epoch": 2.565398224594947, "grad_norm": 0.39040851316144914, "learning_rate": 6.257871860219983e-06, "loss": 2.6478, "step": 41326 }, { "epoch": 2.5654603016947046, "grad_norm": 0.4365244445106446, "learning_rate": 6.256122525434982e-06, "loss": 2.6561, "step": 41327 }, { "epoch": 2.565522378794463, "grad_norm": 0.4187032538270487, "learning_rate": 6.2543734188724004e-06, "loss": 2.669, "step": 41328 }, { "epoch": 2.5655844558942205, "grad_norm": 0.43238628072103286, "learning_rate": 6.252624540541385e-06, "loss": 2.663, "step": 41329 }, { "epoch": 2.5656465329939784, "grad_norm": 0.42678778780976734, "learning_rate": 6.250875890451047e-06, "loss": 2.6294, "step": 41330 }, { "epoch": 2.5657086100937363, "grad_norm": 0.4377623861141105, "learning_rate": 6.249127468610505e-06, "loss": 2.7233, "step": 41331 }, { "epoch": 2.565770687193494, "grad_norm": 0.4129801044209389, "learning_rate": 6.2473792750288914e-06, "loss": 2.6721, "step": 41332 }, { "epoch": 2.565832764293252, "grad_norm": 0.4208996589094173, "learning_rate": 6.245631309715311e-06, "loss": 2.6848, "step": 41333 }, { "epoch": 2.56589484139301, "grad_norm": 0.41247379768844195, "learning_rate": 6.243883572678899e-06, "loss": 2.6667, "step": 41334 }, { "epoch": 2.565956918492768, "grad_norm": 0.41493144715806124, "learning_rate": 6.242136063928761e-06, "loss": 2.6551, "step": 41335 }, { "epoch": 2.566018995592526, "grad_norm": 0.45018354553600004, "learning_rate": 6.240388783474033e-06, "loss": 2.6648, "step": 41336 }, { "epoch": 2.566081072692284, "grad_norm": 0.4416785286564046, "learning_rate": 6.238641731323819e-06, "loss": 2.6548, "step": 41337 }, { "epoch": 2.5661431497920417, "grad_norm": 0.3851834463611522, "learning_rate": 6.236894907487223e-06, "loss": 2.6415, "step": 41338 }, { "epoch": 2.5662052268917996, "grad_norm": 0.40556769661315767, "learning_rate": 6.2351483119733826e-06, "loss": 2.7449, "step": 41339 }, { "epoch": 2.5662673039915576, "grad_norm": 0.4048088444127591, "learning_rate": 6.2334019447914e-06, "loss": 2.5795, "step": 41340 }, { "epoch": 2.5663293810913155, "grad_norm": 0.41698157410883996, "learning_rate": 6.231655805950382e-06, "loss": 2.7059, "step": 41341 }, { "epoch": 2.5663914581910734, "grad_norm": 0.41873138585638103, "learning_rate": 6.229909895459429e-06, "loss": 2.5998, "step": 41342 }, { "epoch": 2.5664535352908313, "grad_norm": 0.4153627098623216, "learning_rate": 6.228164213327669e-06, "loss": 2.6855, "step": 41343 }, { "epoch": 2.5665156123905892, "grad_norm": 0.4044047962572625, "learning_rate": 6.226418759564206e-06, "loss": 2.6519, "step": 41344 }, { "epoch": 2.566577689490347, "grad_norm": 0.41794482724187454, "learning_rate": 6.2246735341781435e-06, "loss": 2.6941, "step": 41345 }, { "epoch": 2.566639766590105, "grad_norm": 0.4112828902742354, "learning_rate": 6.222928537178585e-06, "loss": 2.6411, "step": 41346 }, { "epoch": 2.566701843689863, "grad_norm": 0.39733331679351075, "learning_rate": 6.221183768574629e-06, "loss": 2.6496, "step": 41347 }, { "epoch": 2.5667639207896205, "grad_norm": 0.41681540536489964, "learning_rate": 6.21943922837539e-06, "loss": 2.6928, "step": 41348 }, { "epoch": 2.566825997889379, "grad_norm": 0.40337437966969597, "learning_rate": 6.2176949165899665e-06, "loss": 2.6653, "step": 41349 }, { "epoch": 2.5668880749891363, "grad_norm": 0.40435302270778595, "learning_rate": 6.215950833227457e-06, "loss": 2.683, "step": 41350 }, { "epoch": 2.5669501520888947, "grad_norm": 0.3881585646903444, "learning_rate": 6.214206978296949e-06, "loss": 2.7407, "step": 41351 }, { "epoch": 2.567012229188652, "grad_norm": 0.4136438052242106, "learning_rate": 6.212463351807568e-06, "loss": 2.6137, "step": 41352 }, { "epoch": 2.56707430628841, "grad_norm": 0.40721946362498834, "learning_rate": 6.2107199537683905e-06, "loss": 2.6172, "step": 41353 }, { "epoch": 2.567136383388168, "grad_norm": 0.40094971118625944, "learning_rate": 6.20897678418852e-06, "loss": 2.5801, "step": 41354 }, { "epoch": 2.567198460487926, "grad_norm": 0.4185380890546033, "learning_rate": 6.207233843077037e-06, "loss": 2.6425, "step": 41355 }, { "epoch": 2.567260537587684, "grad_norm": 0.4237587516015978, "learning_rate": 6.205491130443053e-06, "loss": 2.6785, "step": 41356 }, { "epoch": 2.5673226146874417, "grad_norm": 0.4387179010011985, "learning_rate": 6.2037486462956595e-06, "loss": 2.7289, "step": 41357 }, { "epoch": 2.5673846917871996, "grad_norm": 0.4164445483718902, "learning_rate": 6.202006390643939e-06, "loss": 2.565, "step": 41358 }, { "epoch": 2.5674467688869576, "grad_norm": 0.45388239011576287, "learning_rate": 6.200264363496982e-06, "loss": 2.6533, "step": 41359 }, { "epoch": 2.5675088459867155, "grad_norm": 0.41281772139522566, "learning_rate": 6.198522564863868e-06, "loss": 2.6684, "step": 41360 }, { "epoch": 2.5675709230864734, "grad_norm": 0.4174191970924927, "learning_rate": 6.1967809947537035e-06, "loss": 2.6946, "step": 41361 }, { "epoch": 2.5676330001862313, "grad_norm": 0.39929124580763703, "learning_rate": 6.195039653175572e-06, "loss": 2.6944, "step": 41362 }, { "epoch": 2.5676950772859892, "grad_norm": 0.39150044569815917, "learning_rate": 6.193298540138548e-06, "loss": 2.7185, "step": 41363 }, { "epoch": 2.567757154385747, "grad_norm": 0.4120881314700629, "learning_rate": 6.191557655651708e-06, "loss": 2.6994, "step": 41364 }, { "epoch": 2.567819231485505, "grad_norm": 0.3882346833395531, "learning_rate": 6.189816999724163e-06, "loss": 2.6221, "step": 41365 }, { "epoch": 2.567881308585263, "grad_norm": 0.41595714677457185, "learning_rate": 6.188076572364976e-06, "loss": 2.6336, "step": 41366 }, { "epoch": 2.567943385685021, "grad_norm": 0.4158024846345392, "learning_rate": 6.186336373583229e-06, "loss": 2.6663, "step": 41367 }, { "epoch": 2.568005462784779, "grad_norm": 0.3944830345680719, "learning_rate": 6.1845964033879875e-06, "loss": 2.651, "step": 41368 }, { "epoch": 2.5680675398845367, "grad_norm": 0.4207035861637064, "learning_rate": 6.1828566617883555e-06, "loss": 2.6149, "step": 41369 }, { "epoch": 2.5681296169842946, "grad_norm": 0.3992095834235828, "learning_rate": 6.181117148793386e-06, "loss": 2.674, "step": 41370 }, { "epoch": 2.568191694084052, "grad_norm": 0.3975657294468598, "learning_rate": 6.179377864412178e-06, "loss": 2.6618, "step": 41371 }, { "epoch": 2.5682537711838105, "grad_norm": 0.44638910153157146, "learning_rate": 6.177638808653796e-06, "loss": 2.6563, "step": 41372 }, { "epoch": 2.568315848283568, "grad_norm": 0.41300184327284567, "learning_rate": 6.175899981527305e-06, "loss": 2.666, "step": 41373 }, { "epoch": 2.5683779253833263, "grad_norm": 0.4152346520409508, "learning_rate": 6.174161383041777e-06, "loss": 2.553, "step": 41374 }, { "epoch": 2.568440002483084, "grad_norm": 0.4170570948379279, "learning_rate": 6.1724230132062965e-06, "loss": 2.6759, "step": 41375 }, { "epoch": 2.568502079582842, "grad_norm": 0.41350411605746756, "learning_rate": 6.170684872029925e-06, "loss": 2.6933, "step": 41376 }, { "epoch": 2.5685641566825996, "grad_norm": 0.4170038726384396, "learning_rate": 6.168946959521732e-06, "loss": 2.702, "step": 41377 }, { "epoch": 2.5686262337823575, "grad_norm": 0.42558756631484357, "learning_rate": 6.167209275690766e-06, "loss": 2.6894, "step": 41378 }, { "epoch": 2.5686883108821155, "grad_norm": 0.43545904257541423, "learning_rate": 6.165471820546126e-06, "loss": 2.6408, "step": 41379 }, { "epoch": 2.5687503879818734, "grad_norm": 0.4054348727500287, "learning_rate": 6.16373459409686e-06, "loss": 2.6652, "step": 41380 }, { "epoch": 2.5688124650816313, "grad_norm": 0.4080519145177449, "learning_rate": 6.161997596352032e-06, "loss": 2.6714, "step": 41381 }, { "epoch": 2.568874542181389, "grad_norm": 0.39928329677554286, "learning_rate": 6.1602608273207074e-06, "loss": 2.7122, "step": 41382 }, { "epoch": 2.568936619281147, "grad_norm": 0.387778496646321, "learning_rate": 6.15852428701193e-06, "loss": 2.7158, "step": 41383 }, { "epoch": 2.568998696380905, "grad_norm": 0.41015786670184534, "learning_rate": 6.156787975434786e-06, "loss": 2.654, "step": 41384 }, { "epoch": 2.569060773480663, "grad_norm": 0.40984063493807904, "learning_rate": 6.155051892598324e-06, "loss": 2.6691, "step": 41385 }, { "epoch": 2.569122850580421, "grad_norm": 0.386102435741038, "learning_rate": 6.153316038511598e-06, "loss": 2.6993, "step": 41386 }, { "epoch": 2.569184927680179, "grad_norm": 0.39572318535555634, "learning_rate": 6.151580413183655e-06, "loss": 2.7154, "step": 41387 }, { "epoch": 2.5692470047799367, "grad_norm": 0.4024324782227491, "learning_rate": 6.149845016623574e-06, "loss": 2.6083, "step": 41388 }, { "epoch": 2.5693090818796946, "grad_norm": 0.44572242474373397, "learning_rate": 6.1481098488403945e-06, "loss": 2.7179, "step": 41389 }, { "epoch": 2.5693711589794526, "grad_norm": 0.43542773066917967, "learning_rate": 6.146374909843172e-06, "loss": 2.681, "step": 41390 }, { "epoch": 2.5694332360792105, "grad_norm": 0.453138647202808, "learning_rate": 6.14464019964095e-06, "loss": 2.6394, "step": 41391 }, { "epoch": 2.5694953131789684, "grad_norm": 0.3893899657862012, "learning_rate": 6.142905718242792e-06, "loss": 2.5604, "step": 41392 }, { "epoch": 2.5695573902787263, "grad_norm": 0.42324275363097374, "learning_rate": 6.141171465657747e-06, "loss": 2.7554, "step": 41393 }, { "epoch": 2.5696194673784842, "grad_norm": 0.41193850984190583, "learning_rate": 6.139437441894852e-06, "loss": 2.6432, "step": 41394 }, { "epoch": 2.569681544478242, "grad_norm": 0.4322827771050306, "learning_rate": 6.137703646963167e-06, "loss": 2.7027, "step": 41395 }, { "epoch": 2.5697436215779996, "grad_norm": 0.43631749767724765, "learning_rate": 6.135970080871711e-06, "loss": 2.6594, "step": 41396 }, { "epoch": 2.569805698677758, "grad_norm": 0.40484154910067577, "learning_rate": 6.134236743629562e-06, "loss": 2.6869, "step": 41397 }, { "epoch": 2.5698677757775155, "grad_norm": 0.41120380866878853, "learning_rate": 6.132503635245751e-06, "loss": 2.6492, "step": 41398 }, { "epoch": 2.569929852877274, "grad_norm": 0.4208042366840298, "learning_rate": 6.1307707557293146e-06, "loss": 2.6267, "step": 41399 }, { "epoch": 2.5699919299770313, "grad_norm": 0.407143380796961, "learning_rate": 6.1290381050892855e-06, "loss": 2.637, "step": 41400 }, { "epoch": 2.570054007076789, "grad_norm": 0.40826857192760596, "learning_rate": 6.127305683334716e-06, "loss": 2.7143, "step": 41401 }, { "epoch": 2.570116084176547, "grad_norm": 0.42003303653350443, "learning_rate": 6.125573490474651e-06, "loss": 2.6231, "step": 41402 }, { "epoch": 2.570178161276305, "grad_norm": 0.42021649692998325, "learning_rate": 6.123841526518126e-06, "loss": 2.7193, "step": 41403 }, { "epoch": 2.570240238376063, "grad_norm": 0.4319946425313235, "learning_rate": 6.122109791474168e-06, "loss": 2.7723, "step": 41404 }, { "epoch": 2.570302315475821, "grad_norm": 0.41839143290072917, "learning_rate": 6.120378285351802e-06, "loss": 2.6293, "step": 41405 }, { "epoch": 2.570364392575579, "grad_norm": 0.4254719371865402, "learning_rate": 6.118647008160089e-06, "loss": 2.5779, "step": 41406 }, { "epoch": 2.5704264696753367, "grad_norm": 0.4790301612245872, "learning_rate": 6.116915959908043e-06, "loss": 2.6928, "step": 41407 }, { "epoch": 2.5704885467750946, "grad_norm": 0.4514018541903056, "learning_rate": 6.115185140604701e-06, "loss": 2.6545, "step": 41408 }, { "epoch": 2.5705506238748526, "grad_norm": 0.41162541976592215, "learning_rate": 6.113454550259096e-06, "loss": 2.6317, "step": 41409 }, { "epoch": 2.5706127009746105, "grad_norm": 0.45223146339562353, "learning_rate": 6.111724188880235e-06, "loss": 2.7262, "step": 41410 }, { "epoch": 2.5706747780743684, "grad_norm": 0.44592178223463697, "learning_rate": 6.1099940564771756e-06, "loss": 2.6566, "step": 41411 }, { "epoch": 2.5707368551741263, "grad_norm": 0.4410268621643361, "learning_rate": 6.108264153058934e-06, "loss": 2.7817, "step": 41412 }, { "epoch": 2.5707989322738842, "grad_norm": 0.3940596074747627, "learning_rate": 6.106534478634535e-06, "loss": 2.6202, "step": 41413 }, { "epoch": 2.570861009373642, "grad_norm": 0.4135384211985395, "learning_rate": 6.104805033212985e-06, "loss": 2.6423, "step": 41414 }, { "epoch": 2.5709230864734, "grad_norm": 0.42250012953603894, "learning_rate": 6.103075816803338e-06, "loss": 2.6776, "step": 41415 }, { "epoch": 2.570985163573158, "grad_norm": 0.4521361610937594, "learning_rate": 6.1013468294146005e-06, "loss": 2.6507, "step": 41416 }, { "epoch": 2.571047240672916, "grad_norm": 0.3833174780835409, "learning_rate": 6.099618071055791e-06, "loss": 2.6763, "step": 41417 }, { "epoch": 2.571109317772674, "grad_norm": 0.4216203627525026, "learning_rate": 6.097889541735935e-06, "loss": 2.6542, "step": 41418 }, { "epoch": 2.5711713948724313, "grad_norm": 0.48290208707572907, "learning_rate": 6.096161241464032e-06, "loss": 2.7133, "step": 41419 }, { "epoch": 2.5712334719721897, "grad_norm": 0.3997348905240262, "learning_rate": 6.094433170249125e-06, "loss": 2.589, "step": 41420 }, { "epoch": 2.571295549071947, "grad_norm": 0.41954147496896116, "learning_rate": 6.0927053281002235e-06, "loss": 2.6501, "step": 41421 }, { "epoch": 2.5713576261717055, "grad_norm": 0.484029985423126, "learning_rate": 6.09097771502633e-06, "loss": 2.7048, "step": 41422 }, { "epoch": 2.571419703271463, "grad_norm": 0.4540436902649465, "learning_rate": 6.089250331036461e-06, "loss": 2.7139, "step": 41423 }, { "epoch": 2.5714817803712213, "grad_norm": 0.44456491125003406, "learning_rate": 6.087523176139643e-06, "loss": 2.6622, "step": 41424 }, { "epoch": 2.571543857470979, "grad_norm": 0.4232494605354436, "learning_rate": 6.085796250344872e-06, "loss": 2.6916, "step": 41425 }, { "epoch": 2.5716059345707367, "grad_norm": 0.45072639237318457, "learning_rate": 6.084069553661165e-06, "loss": 2.6706, "step": 41426 }, { "epoch": 2.5716680116704946, "grad_norm": 0.47660613251105155, "learning_rate": 6.082343086097514e-06, "loss": 2.7446, "step": 41427 }, { "epoch": 2.5717300887702526, "grad_norm": 0.4626784985419146, "learning_rate": 6.080616847662956e-06, "loss": 2.7595, "step": 41428 }, { "epoch": 2.5717921658700105, "grad_norm": 0.4229012475546248, "learning_rate": 6.07889083836648e-06, "loss": 2.6918, "step": 41429 }, { "epoch": 2.5718542429697684, "grad_norm": 0.4441710784590419, "learning_rate": 6.077165058217094e-06, "loss": 2.6696, "step": 41430 }, { "epoch": 2.5719163200695263, "grad_norm": 0.4879197728144771, "learning_rate": 6.075439507223796e-06, "loss": 2.625, "step": 41431 }, { "epoch": 2.5719783971692842, "grad_norm": 0.42345821921132365, "learning_rate": 6.073714185395591e-06, "loss": 2.6887, "step": 41432 }, { "epoch": 2.572040474269042, "grad_norm": 0.40251630027414936, "learning_rate": 6.071989092741476e-06, "loss": 2.6905, "step": 41433 }, { "epoch": 2.5721025513688, "grad_norm": 0.421952452437471, "learning_rate": 6.070264229270467e-06, "loss": 2.7024, "step": 41434 }, { "epoch": 2.572164628468558, "grad_norm": 0.4229069909114855, "learning_rate": 6.068539594991557e-06, "loss": 2.645, "step": 41435 }, { "epoch": 2.572226705568316, "grad_norm": 0.43143570467473, "learning_rate": 6.0668151899137385e-06, "loss": 2.5975, "step": 41436 }, { "epoch": 2.572288782668074, "grad_norm": 0.413763262026202, "learning_rate": 6.065091014046004e-06, "loss": 2.627, "step": 41437 }, { "epoch": 2.5723508597678317, "grad_norm": 0.41832562834963155, "learning_rate": 6.063367067397363e-06, "loss": 2.5807, "step": 41438 }, { "epoch": 2.5724129368675897, "grad_norm": 0.44586727653361224, "learning_rate": 6.061643349976804e-06, "loss": 2.6755, "step": 41439 }, { "epoch": 2.5724750139673476, "grad_norm": 0.428811289657844, "learning_rate": 6.059919861793318e-06, "loss": 2.6635, "step": 41440 }, { "epoch": 2.5725370910671055, "grad_norm": 0.41056924644661213, "learning_rate": 6.058196602855881e-06, "loss": 2.7384, "step": 41441 }, { "epoch": 2.5725991681668634, "grad_norm": 0.3974132462365865, "learning_rate": 6.056473573173516e-06, "loss": 2.6179, "step": 41442 }, { "epoch": 2.5726612452666213, "grad_norm": 0.4195171268579326, "learning_rate": 6.0547507727551915e-06, "loss": 2.6116, "step": 41443 }, { "epoch": 2.572723322366379, "grad_norm": 0.4171642332696957, "learning_rate": 6.053028201609901e-06, "loss": 2.6122, "step": 41444 }, { "epoch": 2.572785399466137, "grad_norm": 0.43315225754877984, "learning_rate": 6.051305859746637e-06, "loss": 2.6186, "step": 41445 }, { "epoch": 2.5728474765658946, "grad_norm": 0.40298185394505487, "learning_rate": 6.049583747174364e-06, "loss": 2.6378, "step": 41446 }, { "epoch": 2.572909553665653, "grad_norm": 0.4319427089087294, "learning_rate": 6.047861863902088e-06, "loss": 2.6729, "step": 41447 }, { "epoch": 2.5729716307654105, "grad_norm": 0.4562213111375455, "learning_rate": 6.046140209938795e-06, "loss": 2.7202, "step": 41448 }, { "epoch": 2.5730337078651684, "grad_norm": 0.42142400653955325, "learning_rate": 6.0444187852934545e-06, "loss": 2.6831, "step": 41449 }, { "epoch": 2.5730957849649263, "grad_norm": 0.3865895488961875, "learning_rate": 6.0426975899750385e-06, "loss": 2.6864, "step": 41450 }, { "epoch": 2.573157862064684, "grad_norm": 0.4049829752016383, "learning_rate": 6.040976623992556e-06, "loss": 2.663, "step": 41451 }, { "epoch": 2.573219939164442, "grad_norm": 0.4699105360481676, "learning_rate": 6.039255887354967e-06, "loss": 2.7579, "step": 41452 }, { "epoch": 2.5732820162642, "grad_norm": 0.4176600198400769, "learning_rate": 6.037535380071252e-06, "loss": 2.6728, "step": 41453 }, { "epoch": 2.573344093363958, "grad_norm": 0.4199089758022894, "learning_rate": 6.035815102150388e-06, "loss": 2.743, "step": 41454 }, { "epoch": 2.573406170463716, "grad_norm": 0.3826894173007941, "learning_rate": 6.034095053601341e-06, "loss": 2.7172, "step": 41455 }, { "epoch": 2.573468247563474, "grad_norm": 0.3998474716901775, "learning_rate": 6.032375234433096e-06, "loss": 2.6358, "step": 41456 }, { "epoch": 2.5735303246632317, "grad_norm": 0.4189415955989965, "learning_rate": 6.030655644654631e-06, "loss": 2.6641, "step": 41457 }, { "epoch": 2.5735924017629896, "grad_norm": 0.42283043577351426, "learning_rate": 6.028936284274905e-06, "loss": 2.6525, "step": 41458 }, { "epoch": 2.5736544788627476, "grad_norm": 0.4094065217265526, "learning_rate": 6.027217153302883e-06, "loss": 2.7735, "step": 41459 }, { "epoch": 2.5737165559625055, "grad_norm": 0.41787882755590855, "learning_rate": 6.025498251747558e-06, "loss": 2.6053, "step": 41460 }, { "epoch": 2.5737786330622634, "grad_norm": 0.4372305892040484, "learning_rate": 6.023779579617883e-06, "loss": 2.65, "step": 41461 }, { "epoch": 2.5738407101620213, "grad_norm": 0.36739921553485955, "learning_rate": 6.022061136922819e-06, "loss": 2.6382, "step": 41462 }, { "epoch": 2.5739027872617792, "grad_norm": 0.4023817840672645, "learning_rate": 6.020342923671335e-06, "loss": 2.6267, "step": 41463 }, { "epoch": 2.573964864361537, "grad_norm": 0.4072423515643685, "learning_rate": 6.018624939872403e-06, "loss": 2.5976, "step": 41464 }, { "epoch": 2.574026941461295, "grad_norm": 0.3789496052900635, "learning_rate": 6.016907185534987e-06, "loss": 2.682, "step": 41465 }, { "epoch": 2.574089018561053, "grad_norm": 0.3779033681295443, "learning_rate": 6.01518966066803e-06, "loss": 2.6141, "step": 41466 }, { "epoch": 2.5741510956608105, "grad_norm": 0.40695626049477907, "learning_rate": 6.013472365280515e-06, "loss": 2.5745, "step": 41467 }, { "epoch": 2.574213172760569, "grad_norm": 0.42126139415986225, "learning_rate": 6.0117552993814e-06, "loss": 2.7127, "step": 41468 }, { "epoch": 2.5742752498603263, "grad_norm": 0.4023260464523285, "learning_rate": 6.0100384629796165e-06, "loss": 2.7011, "step": 41469 }, { "epoch": 2.5743373269600847, "grad_norm": 0.4130861719730514, "learning_rate": 6.008321856084159e-06, "loss": 2.7289, "step": 41470 }, { "epoch": 2.574399404059842, "grad_norm": 0.3935869075942118, "learning_rate": 6.006605478703958e-06, "loss": 2.7727, "step": 41471 }, { "epoch": 2.5744614811596005, "grad_norm": 0.38966759656110445, "learning_rate": 6.004889330847985e-06, "loss": 2.7044, "step": 41472 }, { "epoch": 2.574523558259358, "grad_norm": 0.39905154838261614, "learning_rate": 6.003173412525171e-06, "loss": 2.6137, "step": 41473 }, { "epoch": 2.574585635359116, "grad_norm": 0.40610193341923806, "learning_rate": 6.001457723744486e-06, "loss": 2.7288, "step": 41474 }, { "epoch": 2.574647712458874, "grad_norm": 0.41189146382330477, "learning_rate": 5.9997422645148846e-06, "loss": 2.7074, "step": 41475 }, { "epoch": 2.5747097895586317, "grad_norm": 0.40964472526293466, "learning_rate": 5.998027034845305e-06, "loss": 2.6842, "step": 41476 }, { "epoch": 2.5747718666583896, "grad_norm": 0.4163034544808079, "learning_rate": 5.996312034744694e-06, "loss": 2.6706, "step": 41477 }, { "epoch": 2.5748339437581476, "grad_norm": 0.4176935912342686, "learning_rate": 5.994597264222013e-06, "loss": 2.7497, "step": 41478 }, { "epoch": 2.5748960208579055, "grad_norm": 0.40404205769150203, "learning_rate": 5.992882723286197e-06, "loss": 2.7261, "step": 41479 }, { "epoch": 2.5749580979576634, "grad_norm": 0.37939109457023146, "learning_rate": 5.991168411946202e-06, "loss": 2.6013, "step": 41480 }, { "epoch": 2.5750201750574213, "grad_norm": 0.3700892858128907, "learning_rate": 5.989454330210964e-06, "loss": 2.6702, "step": 41481 }, { "epoch": 2.5750822521571792, "grad_norm": 0.3866761116643237, "learning_rate": 5.987740478089409e-06, "loss": 2.5594, "step": 41482 }, { "epoch": 2.575144329256937, "grad_norm": 0.4154344537817567, "learning_rate": 5.9860268555905154e-06, "loss": 2.7314, "step": 41483 }, { "epoch": 2.575206406356695, "grad_norm": 0.4053592453164468, "learning_rate": 5.984313462723201e-06, "loss": 2.6849, "step": 41484 }, { "epoch": 2.575268483456453, "grad_norm": 0.3793717111133881, "learning_rate": 5.98260029949641e-06, "loss": 2.6735, "step": 41485 }, { "epoch": 2.575330560556211, "grad_norm": 0.39180179718363856, "learning_rate": 5.980887365919063e-06, "loss": 2.6129, "step": 41486 }, { "epoch": 2.575392637655969, "grad_norm": 0.39020296186112946, "learning_rate": 5.979174662000131e-06, "loss": 2.5721, "step": 41487 }, { "epoch": 2.5754547147557267, "grad_norm": 0.38869386160306796, "learning_rate": 5.977462187748522e-06, "loss": 2.6748, "step": 41488 }, { "epoch": 2.5755167918554847, "grad_norm": 0.40758553470239894, "learning_rate": 5.975749943173187e-06, "loss": 2.6814, "step": 41489 }, { "epoch": 2.5755788689552426, "grad_norm": 0.40654049170914885, "learning_rate": 5.974037928283038e-06, "loss": 2.6506, "step": 41490 }, { "epoch": 2.5756409460550005, "grad_norm": 0.41351885464428845, "learning_rate": 5.972326143087032e-06, "loss": 2.6996, "step": 41491 }, { "epoch": 2.575703023154758, "grad_norm": 0.4330801284412836, "learning_rate": 5.9706145875940925e-06, "loss": 2.7619, "step": 41492 }, { "epoch": 2.5757651002545163, "grad_norm": 0.39834075575993255, "learning_rate": 5.968903261813141e-06, "loss": 2.666, "step": 41493 }, { "epoch": 2.575827177354274, "grad_norm": 0.41112128669729675, "learning_rate": 5.96719216575311e-06, "loss": 2.7004, "step": 41494 }, { "epoch": 2.575889254454032, "grad_norm": 0.3984183138547538, "learning_rate": 5.965481299422915e-06, "loss": 2.685, "step": 41495 }, { "epoch": 2.5759513315537896, "grad_norm": 0.44366269931887076, "learning_rate": 5.963770662831509e-06, "loss": 2.6025, "step": 41496 }, { "epoch": 2.5760134086535476, "grad_norm": 0.4253782412405225, "learning_rate": 5.9620602559878e-06, "loss": 2.6878, "step": 41497 }, { "epoch": 2.5760754857533055, "grad_norm": 0.4145984959185875, "learning_rate": 5.960350078900711e-06, "loss": 2.7002, "step": 41498 }, { "epoch": 2.5761375628530634, "grad_norm": 0.38170545451009974, "learning_rate": 5.958640131579157e-06, "loss": 2.7265, "step": 41499 }, { "epoch": 2.5761996399528213, "grad_norm": 0.45327603355228024, "learning_rate": 5.95693041403207e-06, "loss": 2.6395, "step": 41500 }, { "epoch": 2.5762617170525792, "grad_norm": 0.43038886656569003, "learning_rate": 5.955220926268379e-06, "loss": 2.661, "step": 41501 }, { "epoch": 2.576323794152337, "grad_norm": 0.3973176777664149, "learning_rate": 5.95351166829699e-06, "loss": 2.7536, "step": 41502 }, { "epoch": 2.576385871252095, "grad_norm": 0.41400218020431756, "learning_rate": 5.951802640126824e-06, "loss": 2.61, "step": 41503 }, { "epoch": 2.576447948351853, "grad_norm": 0.4571184254508094, "learning_rate": 5.950093841766802e-06, "loss": 2.6522, "step": 41504 }, { "epoch": 2.576510025451611, "grad_norm": 0.43180403917255705, "learning_rate": 5.948385273225815e-06, "loss": 2.6057, "step": 41505 }, { "epoch": 2.576572102551369, "grad_norm": 0.46430999501110126, "learning_rate": 5.946676934512813e-06, "loss": 2.7324, "step": 41506 }, { "epoch": 2.5766341796511267, "grad_norm": 0.41575180786666466, "learning_rate": 5.944968825636687e-06, "loss": 2.6092, "step": 41507 }, { "epoch": 2.5766962567508847, "grad_norm": 0.4241841486524033, "learning_rate": 5.94326094660636e-06, "loss": 2.6942, "step": 41508 }, { "epoch": 2.5767583338506426, "grad_norm": 0.4212097291346243, "learning_rate": 5.941553297430718e-06, "loss": 2.6589, "step": 41509 }, { "epoch": 2.5768204109504005, "grad_norm": 0.4179460394562934, "learning_rate": 5.939845878118705e-06, "loss": 2.691, "step": 41510 }, { "epoch": 2.5768824880501584, "grad_norm": 0.43947141821191715, "learning_rate": 5.938138688679207e-06, "loss": 2.6708, "step": 41511 }, { "epoch": 2.5769445651499163, "grad_norm": 0.4284268352424939, "learning_rate": 5.936431729121139e-06, "loss": 2.6482, "step": 41512 }, { "epoch": 2.5770066422496742, "grad_norm": 0.4357029759141907, "learning_rate": 5.9347249994533895e-06, "loss": 2.6646, "step": 41513 }, { "epoch": 2.577068719349432, "grad_norm": 0.4647244532757641, "learning_rate": 5.933018499684895e-06, "loss": 2.6548, "step": 41514 }, { "epoch": 2.5771307964491896, "grad_norm": 0.43233930562736644, "learning_rate": 5.931312229824532e-06, "loss": 2.5895, "step": 41515 }, { "epoch": 2.577192873548948, "grad_norm": 0.41883674406359744, "learning_rate": 5.929606189881215e-06, "loss": 2.5753, "step": 41516 }, { "epoch": 2.5772549506487055, "grad_norm": 0.41696243716028836, "learning_rate": 5.9279003798638375e-06, "loss": 2.6607, "step": 41517 }, { "epoch": 2.577317027748464, "grad_norm": 0.4151549413141392, "learning_rate": 5.9261947997812985e-06, "loss": 2.6678, "step": 41518 }, { "epoch": 2.5773791048482213, "grad_norm": 0.4080993301357127, "learning_rate": 5.924489449642507e-06, "loss": 2.5998, "step": 41519 }, { "epoch": 2.5774411819479797, "grad_norm": 0.4040063007330804, "learning_rate": 5.922784329456355e-06, "loss": 2.6916, "step": 41520 }, { "epoch": 2.577503259047737, "grad_norm": 0.38525951463608266, "learning_rate": 5.921079439231736e-06, "loss": 2.6955, "step": 41521 }, { "epoch": 2.577565336147495, "grad_norm": 0.46397164222780357, "learning_rate": 5.919374778977538e-06, "loss": 2.6916, "step": 41522 }, { "epoch": 2.577627413247253, "grad_norm": 0.3984061210094126, "learning_rate": 5.9176703487026694e-06, "loss": 2.5703, "step": 41523 }, { "epoch": 2.577689490347011, "grad_norm": 0.4022977302676412, "learning_rate": 5.915966148416019e-06, "loss": 2.72, "step": 41524 }, { "epoch": 2.577751567446769, "grad_norm": 0.41941702013943855, "learning_rate": 5.914262178126473e-06, "loss": 2.6645, "step": 41525 }, { "epoch": 2.5778136445465267, "grad_norm": 0.4064258414961443, "learning_rate": 5.912558437842914e-06, "loss": 2.5927, "step": 41526 }, { "epoch": 2.5778757216462846, "grad_norm": 0.40613169047773123, "learning_rate": 5.9108549275742495e-06, "loss": 2.7199, "step": 41527 }, { "epoch": 2.5779377987460426, "grad_norm": 0.40925824997567134, "learning_rate": 5.909151647329358e-06, "loss": 2.7491, "step": 41528 }, { "epoch": 2.5779998758458005, "grad_norm": 0.38677654412482676, "learning_rate": 5.907448597117126e-06, "loss": 2.6295, "step": 41529 }, { "epoch": 2.5780619529455584, "grad_norm": 0.39295449885166445, "learning_rate": 5.905745776946442e-06, "loss": 2.6545, "step": 41530 }, { "epoch": 2.5781240300453163, "grad_norm": 0.4124040679931728, "learning_rate": 5.904043186826175e-06, "loss": 2.6299, "step": 41531 }, { "epoch": 2.5781861071450742, "grad_norm": 0.42453938234561855, "learning_rate": 5.902340826765213e-06, "loss": 2.7237, "step": 41532 }, { "epoch": 2.578248184244832, "grad_norm": 0.40718632780595804, "learning_rate": 5.900638696772459e-06, "loss": 2.661, "step": 41533 }, { "epoch": 2.57831026134459, "grad_norm": 0.4092192892425701, "learning_rate": 5.89893679685678e-06, "loss": 2.6811, "step": 41534 }, { "epoch": 2.578372338444348, "grad_norm": 0.3987188281009782, "learning_rate": 5.897235127027051e-06, "loss": 2.6203, "step": 41535 }, { "epoch": 2.578434415544106, "grad_norm": 0.4455977747493498, "learning_rate": 5.895533687292143e-06, "loss": 2.6377, "step": 41536 }, { "epoch": 2.578496492643864, "grad_norm": 0.3938115609198221, "learning_rate": 5.893832477660954e-06, "loss": 2.6317, "step": 41537 }, { "epoch": 2.5785585697436217, "grad_norm": 0.41547078735035664, "learning_rate": 5.89213149814235e-06, "loss": 2.654, "step": 41538 }, { "epoch": 2.5786206468433797, "grad_norm": 0.3949444372899269, "learning_rate": 5.8904307487452015e-06, "loss": 2.7071, "step": 41539 }, { "epoch": 2.578682723943137, "grad_norm": 0.41465317682548614, "learning_rate": 5.888730229478378e-06, "loss": 2.6884, "step": 41540 }, { "epoch": 2.5787448010428955, "grad_norm": 0.40942186276641435, "learning_rate": 5.8870299403507524e-06, "loss": 2.6811, "step": 41541 }, { "epoch": 2.578806878142653, "grad_norm": 0.43062203822896017, "learning_rate": 5.88532988137121e-06, "loss": 2.6031, "step": 41542 }, { "epoch": 2.5788689552424113, "grad_norm": 0.38168579818148807, "learning_rate": 5.8836300525486165e-06, "loss": 2.5897, "step": 41543 }, { "epoch": 2.578931032342169, "grad_norm": 0.38692397634946707, "learning_rate": 5.881930453891826e-06, "loss": 2.6654, "step": 41544 }, { "epoch": 2.5789931094419267, "grad_norm": 0.43168893604141884, "learning_rate": 5.88023108540971e-06, "loss": 2.7096, "step": 41545 }, { "epoch": 2.5790551865416846, "grad_norm": 0.41986876750725627, "learning_rate": 5.878531947111143e-06, "loss": 2.6816, "step": 41546 }, { "epoch": 2.5791172636414426, "grad_norm": 0.4179522609132031, "learning_rate": 5.876833039004992e-06, "loss": 2.6761, "step": 41547 }, { "epoch": 2.5791793407412005, "grad_norm": 0.4003065439655175, "learning_rate": 5.8751343611001095e-06, "loss": 2.6547, "step": 41548 }, { "epoch": 2.5792414178409584, "grad_norm": 0.43598143106693027, "learning_rate": 5.873435913405357e-06, "loss": 2.6401, "step": 41549 }, { "epoch": 2.5793034949407163, "grad_norm": 0.39235893697145663, "learning_rate": 5.871737695929608e-06, "loss": 2.6302, "step": 41550 }, { "epoch": 2.5793655720404742, "grad_norm": 0.4238862731703456, "learning_rate": 5.8700397086817185e-06, "loss": 2.6767, "step": 41551 }, { "epoch": 2.579427649140232, "grad_norm": 0.4272868981630813, "learning_rate": 5.868341951670541e-06, "loss": 2.6857, "step": 41552 }, { "epoch": 2.57948972623999, "grad_norm": 0.3897274386596925, "learning_rate": 5.866644424904938e-06, "loss": 2.6548, "step": 41553 }, { "epoch": 2.579551803339748, "grad_norm": 0.4349971946264109, "learning_rate": 5.864947128393755e-06, "loss": 2.7632, "step": 41554 }, { "epoch": 2.579613880439506, "grad_norm": 0.43632388407422934, "learning_rate": 5.86325006214587e-06, "loss": 2.6377, "step": 41555 }, { "epoch": 2.579675957539264, "grad_norm": 0.40965458299049007, "learning_rate": 5.861553226170124e-06, "loss": 2.6735, "step": 41556 }, { "epoch": 2.5797380346390217, "grad_norm": 0.4253630377841543, "learning_rate": 5.859856620475368e-06, "loss": 2.6557, "step": 41557 }, { "epoch": 2.5798001117387797, "grad_norm": 0.42996827402177934, "learning_rate": 5.858160245070443e-06, "loss": 2.7214, "step": 41558 }, { "epoch": 2.5798621888385376, "grad_norm": 0.44695817800171656, "learning_rate": 5.856464099964226e-06, "loss": 2.6598, "step": 41559 }, { "epoch": 2.5799242659382955, "grad_norm": 0.44892273147340384, "learning_rate": 5.854768185165549e-06, "loss": 2.6837, "step": 41560 }, { "epoch": 2.5799863430380534, "grad_norm": 0.4385262102255974, "learning_rate": 5.853072500683266e-06, "loss": 2.7232, "step": 41561 }, { "epoch": 2.5800484201378113, "grad_norm": 0.41923070403581786, "learning_rate": 5.851377046526208e-06, "loss": 2.6406, "step": 41562 }, { "epoch": 2.580110497237569, "grad_norm": 0.42498279214130374, "learning_rate": 5.84968182270324e-06, "loss": 2.6221, "step": 41563 }, { "epoch": 2.580172574337327, "grad_norm": 0.4744261870831352, "learning_rate": 5.847986829223195e-06, "loss": 2.6806, "step": 41564 }, { "epoch": 2.5802346514370846, "grad_norm": 0.4215432027892035, "learning_rate": 5.846292066094933e-06, "loss": 2.6451, "step": 41565 }, { "epoch": 2.580296728536843, "grad_norm": 0.4460880486279333, "learning_rate": 5.8445975333272785e-06, "loss": 2.6636, "step": 41566 }, { "epoch": 2.5803588056366005, "grad_norm": 0.41260087051493927, "learning_rate": 5.842903230929081e-06, "loss": 2.6768, "step": 41567 }, { "epoch": 2.580420882736359, "grad_norm": 0.42495358367905267, "learning_rate": 5.841209158909167e-06, "loss": 2.7609, "step": 41568 }, { "epoch": 2.5804829598361163, "grad_norm": 0.4192997249181597, "learning_rate": 5.839515317276395e-06, "loss": 2.6115, "step": 41569 }, { "epoch": 2.5805450369358742, "grad_norm": 0.4598840539906259, "learning_rate": 5.837821706039587e-06, "loss": 2.7378, "step": 41570 }, { "epoch": 2.580607114035632, "grad_norm": 0.4341159459737053, "learning_rate": 5.83612832520759e-06, "loss": 2.6806, "step": 41571 }, { "epoch": 2.58066919113539, "grad_norm": 0.42680459889952527, "learning_rate": 5.8344351747892216e-06, "loss": 2.7026, "step": 41572 }, { "epoch": 2.580731268235148, "grad_norm": 0.42041233607135775, "learning_rate": 5.832742254793333e-06, "loss": 2.5991, "step": 41573 }, { "epoch": 2.580793345334906, "grad_norm": 0.42401767482557995, "learning_rate": 5.831049565228752e-06, "loss": 2.6741, "step": 41574 }, { "epoch": 2.580855422434664, "grad_norm": 0.41195386510703297, "learning_rate": 5.829357106104311e-06, "loss": 2.671, "step": 41575 }, { "epoch": 2.5809174995344217, "grad_norm": 0.41189359828422395, "learning_rate": 5.827664877428824e-06, "loss": 2.7103, "step": 41576 }, { "epoch": 2.5809795766341797, "grad_norm": 0.4151624967243657, "learning_rate": 5.825972879211139e-06, "loss": 2.6466, "step": 41577 }, { "epoch": 2.5810416537339376, "grad_norm": 0.40320489446572905, "learning_rate": 5.824281111460084e-06, "loss": 2.5915, "step": 41578 }, { "epoch": 2.5811037308336955, "grad_norm": 0.4191931856019218, "learning_rate": 5.822589574184472e-06, "loss": 2.8002, "step": 41579 }, { "epoch": 2.5811658079334534, "grad_norm": 0.42018708126938653, "learning_rate": 5.820898267393138e-06, "loss": 2.6306, "step": 41580 }, { "epoch": 2.5812278850332113, "grad_norm": 0.3907867668888078, "learning_rate": 5.8192071910948944e-06, "loss": 2.6592, "step": 41581 }, { "epoch": 2.5812899621329692, "grad_norm": 0.4127156647530551, "learning_rate": 5.817516345298579e-06, "loss": 2.7099, "step": 41582 }, { "epoch": 2.581352039232727, "grad_norm": 0.4349218254164703, "learning_rate": 5.815825730013008e-06, "loss": 2.6575, "step": 41583 }, { "epoch": 2.581414116332485, "grad_norm": 0.43297272962918426, "learning_rate": 5.814135345246996e-06, "loss": 2.7239, "step": 41584 }, { "epoch": 2.581476193432243, "grad_norm": 0.43935820508089124, "learning_rate": 5.812445191009364e-06, "loss": 2.7349, "step": 41585 }, { "epoch": 2.5815382705320005, "grad_norm": 0.39762101007886336, "learning_rate": 5.8107552673089374e-06, "loss": 2.6472, "step": 41586 }, { "epoch": 2.581600347631759, "grad_norm": 0.41310860862474724, "learning_rate": 5.809065574154527e-06, "loss": 2.7401, "step": 41587 }, { "epoch": 2.5816624247315163, "grad_norm": 0.4131550471795652, "learning_rate": 5.807376111554952e-06, "loss": 2.8061, "step": 41588 }, { "epoch": 2.5817245018312747, "grad_norm": 0.41545968676003486, "learning_rate": 5.805686879519023e-06, "loss": 2.697, "step": 41589 }, { "epoch": 2.581786578931032, "grad_norm": 0.4187451219986295, "learning_rate": 5.803997878055545e-06, "loss": 2.7147, "step": 41590 }, { "epoch": 2.5818486560307905, "grad_norm": 0.41995414787298774, "learning_rate": 5.802309107173348e-06, "loss": 2.6622, "step": 41591 }, { "epoch": 2.581910733130548, "grad_norm": 0.4458458199787543, "learning_rate": 5.800620566881232e-06, "loss": 2.7176, "step": 41592 }, { "epoch": 2.581972810230306, "grad_norm": 0.4178501032054646, "learning_rate": 5.798932257188011e-06, "loss": 2.6513, "step": 41593 }, { "epoch": 2.582034887330064, "grad_norm": 0.4376508078190807, "learning_rate": 5.797244178102479e-06, "loss": 2.7082, "step": 41594 }, { "epoch": 2.5820969644298217, "grad_norm": 0.41555542184209343, "learning_rate": 5.795556329633467e-06, "loss": 2.7115, "step": 41595 }, { "epoch": 2.5821590415295796, "grad_norm": 0.4082776234374827, "learning_rate": 5.793868711789769e-06, "loss": 2.7345, "step": 41596 }, { "epoch": 2.5822211186293376, "grad_norm": 0.4124672045820007, "learning_rate": 5.792181324580176e-06, "loss": 2.6897, "step": 41597 }, { "epoch": 2.5822831957290955, "grad_norm": 0.39121636450246133, "learning_rate": 5.790494168013516e-06, "loss": 2.6896, "step": 41598 }, { "epoch": 2.5823452728288534, "grad_norm": 0.4009844028990327, "learning_rate": 5.78880724209857e-06, "loss": 2.6373, "step": 41599 }, { "epoch": 2.5824073499286113, "grad_norm": 0.4136210913820755, "learning_rate": 5.7871205468441605e-06, "loss": 2.7797, "step": 41600 }, { "epoch": 2.5824694270283692, "grad_norm": 0.4018746576559884, "learning_rate": 5.785434082259072e-06, "loss": 2.6833, "step": 41601 }, { "epoch": 2.582531504128127, "grad_norm": 0.3902297690790827, "learning_rate": 5.783747848352111e-06, "loss": 2.7353, "step": 41602 }, { "epoch": 2.582593581227885, "grad_norm": 0.42841384430067997, "learning_rate": 5.782061845132075e-06, "loss": 2.7416, "step": 41603 }, { "epoch": 2.582655658327643, "grad_norm": 0.4118746229005051, "learning_rate": 5.78037607260774e-06, "loss": 2.6572, "step": 41604 }, { "epoch": 2.582717735427401, "grad_norm": 0.41902736047121203, "learning_rate": 5.778690530787933e-06, "loss": 2.6786, "step": 41605 }, { "epoch": 2.582779812527159, "grad_norm": 0.3875155092027484, "learning_rate": 5.777005219681431e-06, "loss": 2.6912, "step": 41606 }, { "epoch": 2.5828418896269167, "grad_norm": 0.43621072513127307, "learning_rate": 5.775320139297025e-06, "loss": 2.7242, "step": 41607 }, { "epoch": 2.5829039667266747, "grad_norm": 0.39498275000896144, "learning_rate": 5.773635289643503e-06, "loss": 2.5858, "step": 41608 }, { "epoch": 2.5829660438264326, "grad_norm": 0.46756652205753646, "learning_rate": 5.77195067072967e-06, "loss": 2.6945, "step": 41609 }, { "epoch": 2.5830281209261905, "grad_norm": 0.40052060134279094, "learning_rate": 5.770266282564307e-06, "loss": 2.6433, "step": 41610 }, { "epoch": 2.583090198025948, "grad_norm": 0.48892546309076496, "learning_rate": 5.768582125156203e-06, "loss": 2.6839, "step": 41611 }, { "epoch": 2.5831522751257063, "grad_norm": 0.4545922530260991, "learning_rate": 5.7668981985141365e-06, "loss": 2.665, "step": 41612 }, { "epoch": 2.583214352225464, "grad_norm": 0.4085294341020211, "learning_rate": 5.765214502646904e-06, "loss": 2.7045, "step": 41613 }, { "epoch": 2.583276429325222, "grad_norm": 0.4226532222054102, "learning_rate": 5.763531037563291e-06, "loss": 2.6724, "step": 41614 }, { "epoch": 2.5833385064249796, "grad_norm": 0.44765755085647935, "learning_rate": 5.761847803272075e-06, "loss": 2.6312, "step": 41615 }, { "epoch": 2.5834005835247376, "grad_norm": 0.43819231365140543, "learning_rate": 5.760164799782036e-06, "loss": 2.6975, "step": 41616 }, { "epoch": 2.5834626606244955, "grad_norm": 0.4205927258561313, "learning_rate": 5.7584820271019465e-06, "loss": 2.5969, "step": 41617 }, { "epoch": 2.5835247377242534, "grad_norm": 0.4513483899443301, "learning_rate": 5.756799485240605e-06, "loss": 2.6993, "step": 41618 }, { "epoch": 2.5835868148240113, "grad_norm": 0.4337039669960618, "learning_rate": 5.755117174206787e-06, "loss": 2.7249, "step": 41619 }, { "epoch": 2.5836488919237692, "grad_norm": 0.4529886204919715, "learning_rate": 5.7534350940092575e-06, "loss": 2.6708, "step": 41620 }, { "epoch": 2.583710969023527, "grad_norm": 0.418536059685546, "learning_rate": 5.751753244656788e-06, "loss": 2.7185, "step": 41621 }, { "epoch": 2.583773046123285, "grad_norm": 0.4389448490192994, "learning_rate": 5.750071626158177e-06, "loss": 2.7078, "step": 41622 }, { "epoch": 2.583835123223043, "grad_norm": 0.4664703248450313, "learning_rate": 5.748390238522183e-06, "loss": 2.6716, "step": 41623 }, { "epoch": 2.583897200322801, "grad_norm": 0.42953098220789093, "learning_rate": 5.746709081757584e-06, "loss": 2.6698, "step": 41624 }, { "epoch": 2.583959277422559, "grad_norm": 0.4005367663636739, "learning_rate": 5.745028155873139e-06, "loss": 2.63, "step": 41625 }, { "epoch": 2.5840213545223167, "grad_norm": 0.44190365270821547, "learning_rate": 5.743347460877618e-06, "loss": 2.6728, "step": 41626 }, { "epoch": 2.5840834316220747, "grad_norm": 0.41227653692832356, "learning_rate": 5.74166699677981e-06, "loss": 2.6399, "step": 41627 }, { "epoch": 2.5841455087218326, "grad_norm": 0.4162157141681245, "learning_rate": 5.739986763588467e-06, "loss": 2.6167, "step": 41628 }, { "epoch": 2.5842075858215905, "grad_norm": 0.4080986728910233, "learning_rate": 5.738306761312356e-06, "loss": 2.6906, "step": 41629 }, { "epoch": 2.5842696629213484, "grad_norm": 0.4218979228820991, "learning_rate": 5.736626989960236e-06, "loss": 2.6219, "step": 41630 }, { "epoch": 2.5843317400211063, "grad_norm": 0.3806736419865954, "learning_rate": 5.734947449540878e-06, "loss": 2.6569, "step": 41631 }, { "epoch": 2.5843938171208642, "grad_norm": 0.4073603202279474, "learning_rate": 5.733268140063058e-06, "loss": 2.6837, "step": 41632 }, { "epoch": 2.584455894220622, "grad_norm": 0.4270296954766295, "learning_rate": 5.731589061535519e-06, "loss": 2.6989, "step": 41633 }, { "epoch": 2.5845179713203796, "grad_norm": 0.42388692351287394, "learning_rate": 5.729910213967027e-06, "loss": 2.6618, "step": 41634 }, { "epoch": 2.584580048420138, "grad_norm": 0.4204793024741263, "learning_rate": 5.728231597366335e-06, "loss": 2.7624, "step": 41635 }, { "epoch": 2.5846421255198955, "grad_norm": 0.4021990802997657, "learning_rate": 5.7265532117422085e-06, "loss": 2.6153, "step": 41636 }, { "epoch": 2.584704202619654, "grad_norm": 0.4098488336815736, "learning_rate": 5.724875057103407e-06, "loss": 2.6581, "step": 41637 }, { "epoch": 2.5847662797194113, "grad_norm": 0.40095706972215567, "learning_rate": 5.7231971334586795e-06, "loss": 2.6972, "step": 41638 }, { "epoch": 2.5848283568191697, "grad_norm": 0.3884790603611092, "learning_rate": 5.721519440816781e-06, "loss": 2.7151, "step": 41639 }, { "epoch": 2.584890433918927, "grad_norm": 0.4277564961883768, "learning_rate": 5.719841979186452e-06, "loss": 2.6486, "step": 41640 }, { "epoch": 2.584952511018685, "grad_norm": 0.45389988840605616, "learning_rate": 5.718164748576471e-06, "loss": 2.705, "step": 41641 }, { "epoch": 2.585014588118443, "grad_norm": 0.41171560412947744, "learning_rate": 5.716487748995575e-06, "loss": 2.6796, "step": 41642 }, { "epoch": 2.585076665218201, "grad_norm": 0.41723448885204795, "learning_rate": 5.7148109804525065e-06, "loss": 2.6989, "step": 41643 }, { "epoch": 2.585138742317959, "grad_norm": 0.41761568790844855, "learning_rate": 5.713134442956014e-06, "loss": 2.6778, "step": 41644 }, { "epoch": 2.5852008194177167, "grad_norm": 0.4014852679603268, "learning_rate": 5.711458136514863e-06, "loss": 2.7113, "step": 41645 }, { "epoch": 2.5852628965174747, "grad_norm": 0.41378024215466025, "learning_rate": 5.709782061137781e-06, "loss": 2.6518, "step": 41646 }, { "epoch": 2.5853249736172326, "grad_norm": 0.3853436112973226, "learning_rate": 5.7081062168335195e-06, "loss": 2.6068, "step": 41647 }, { "epoch": 2.5853870507169905, "grad_norm": 0.37876307348469346, "learning_rate": 5.706430603610808e-06, "loss": 2.6538, "step": 41648 }, { "epoch": 2.5854491278167484, "grad_norm": 0.38402501586800786, "learning_rate": 5.704755221478414e-06, "loss": 2.6828, "step": 41649 }, { "epoch": 2.5855112049165063, "grad_norm": 0.4039605810638318, "learning_rate": 5.703080070445066e-06, "loss": 2.7014, "step": 41650 }, { "epoch": 2.5855732820162642, "grad_norm": 0.38332566101578497, "learning_rate": 5.701405150519501e-06, "loss": 2.67, "step": 41651 }, { "epoch": 2.585635359116022, "grad_norm": 0.42228964455861745, "learning_rate": 5.699730461710457e-06, "loss": 2.7764, "step": 41652 }, { "epoch": 2.58569743621578, "grad_norm": 0.4147047783501133, "learning_rate": 5.698056004026664e-06, "loss": 2.685, "step": 41653 }, { "epoch": 2.585759513315538, "grad_norm": 0.3747656407772758, "learning_rate": 5.696381777476878e-06, "loss": 2.6239, "step": 41654 }, { "epoch": 2.585821590415296, "grad_norm": 0.3756345902867356, "learning_rate": 5.694707782069825e-06, "loss": 2.6557, "step": 41655 }, { "epoch": 2.585883667515054, "grad_norm": 0.4167106088673377, "learning_rate": 5.693034017814236e-06, "loss": 2.7604, "step": 41656 }, { "epoch": 2.5859457446148117, "grad_norm": 0.40000245421371966, "learning_rate": 5.691360484718838e-06, "loss": 2.7116, "step": 41657 }, { "epoch": 2.5860078217145697, "grad_norm": 0.4244361769861041, "learning_rate": 5.689687182792375e-06, "loss": 2.7408, "step": 41658 }, { "epoch": 2.586069898814327, "grad_norm": 0.3822160277866573, "learning_rate": 5.688014112043571e-06, "loss": 2.7133, "step": 41659 }, { "epoch": 2.5861319759140855, "grad_norm": 0.41945141009504006, "learning_rate": 5.686341272481155e-06, "loss": 2.6812, "step": 41660 }, { "epoch": 2.586194053013843, "grad_norm": 0.433201638331842, "learning_rate": 5.68466866411384e-06, "loss": 2.616, "step": 41661 }, { "epoch": 2.5862561301136013, "grad_norm": 0.4465088584775118, "learning_rate": 5.682996286950382e-06, "loss": 2.7118, "step": 41662 }, { "epoch": 2.586318207213359, "grad_norm": 0.39963322155334385, "learning_rate": 5.681324140999478e-06, "loss": 2.6194, "step": 41663 }, { "epoch": 2.5863802843131167, "grad_norm": 0.3821235217663864, "learning_rate": 5.679652226269877e-06, "loss": 2.7034, "step": 41664 }, { "epoch": 2.5864423614128746, "grad_norm": 0.38385560524134776, "learning_rate": 5.677980542770284e-06, "loss": 2.6297, "step": 41665 }, { "epoch": 2.5865044385126326, "grad_norm": 0.40382318804751277, "learning_rate": 5.676309090509429e-06, "loss": 2.6622, "step": 41666 }, { "epoch": 2.5865665156123905, "grad_norm": 0.3998552790742764, "learning_rate": 5.674637869496024e-06, "loss": 2.6405, "step": 41667 }, { "epoch": 2.5866285927121484, "grad_norm": 0.38552065898345383, "learning_rate": 5.6729668797388e-06, "loss": 2.7344, "step": 41668 }, { "epoch": 2.5866906698119063, "grad_norm": 0.3871523227983585, "learning_rate": 5.671296121246466e-06, "loss": 2.6406, "step": 41669 }, { "epoch": 2.5867527469116642, "grad_norm": 0.40215384066093657, "learning_rate": 5.669625594027744e-06, "loss": 2.6091, "step": 41670 }, { "epoch": 2.586814824011422, "grad_norm": 0.3842862817677285, "learning_rate": 5.667955298091337e-06, "loss": 2.6586, "step": 41671 }, { "epoch": 2.58687690111118, "grad_norm": 0.39410222209535306, "learning_rate": 5.6662852334459785e-06, "loss": 2.6579, "step": 41672 }, { "epoch": 2.586938978210938, "grad_norm": 0.38009733600425566, "learning_rate": 5.664615400100376e-06, "loss": 2.6446, "step": 41673 }, { "epoch": 2.587001055310696, "grad_norm": 0.3828871323372312, "learning_rate": 5.662945798063229e-06, "loss": 2.6069, "step": 41674 }, { "epoch": 2.587063132410454, "grad_norm": 0.40400069440413744, "learning_rate": 5.661276427343265e-06, "loss": 2.6477, "step": 41675 }, { "epoch": 2.5871252095102117, "grad_norm": 0.4013273121872168, "learning_rate": 5.6596072879491706e-06, "loss": 2.6395, "step": 41676 }, { "epoch": 2.5871872866099697, "grad_norm": 0.40238941266833367, "learning_rate": 5.657938379889682e-06, "loss": 2.6015, "step": 41677 }, { "epoch": 2.5872493637097276, "grad_norm": 0.3646516486150647, "learning_rate": 5.656269703173495e-06, "loss": 2.7085, "step": 41678 }, { "epoch": 2.5873114408094855, "grad_norm": 0.3785738912955996, "learning_rate": 5.654601257809311e-06, "loss": 2.6954, "step": 41679 }, { "epoch": 2.5873735179092434, "grad_norm": 0.4211381906474354, "learning_rate": 5.652933043805825e-06, "loss": 2.7088, "step": 41680 }, { "epoch": 2.5874355950090013, "grad_norm": 0.3775896223051467, "learning_rate": 5.651265061171768e-06, "loss": 2.7173, "step": 41681 }, { "epoch": 2.587497672108759, "grad_norm": 0.4195691423690158, "learning_rate": 5.649597309915822e-06, "loss": 2.5776, "step": 41682 }, { "epoch": 2.587559749208517, "grad_norm": 0.39031993397702963, "learning_rate": 5.6479297900466965e-06, "loss": 2.6375, "step": 41683 }, { "epoch": 2.5876218263082746, "grad_norm": 0.3902265996607049, "learning_rate": 5.6462625015730795e-06, "loss": 2.6738, "step": 41684 }, { "epoch": 2.587683903408033, "grad_norm": 0.42270707267333846, "learning_rate": 5.6445954445036866e-06, "loss": 2.6526, "step": 41685 }, { "epoch": 2.5877459805077905, "grad_norm": 0.41573333117180916, "learning_rate": 5.64292861884721e-06, "loss": 2.6266, "step": 41686 }, { "epoch": 2.587808057607549, "grad_norm": 0.40249517979904675, "learning_rate": 5.641262024612337e-06, "loss": 2.7074, "step": 41687 }, { "epoch": 2.5878701347073063, "grad_norm": 0.3967481778480978, "learning_rate": 5.6395956618077726e-06, "loss": 2.7308, "step": 41688 }, { "epoch": 2.5879322118070642, "grad_norm": 0.3906945929523422, "learning_rate": 5.637929530442193e-06, "loss": 2.6214, "step": 41689 }, { "epoch": 2.587994288906822, "grad_norm": 0.4087239278708458, "learning_rate": 5.636263630524319e-06, "loss": 2.6689, "step": 41690 }, { "epoch": 2.58805636600658, "grad_norm": 0.40754975957075973, "learning_rate": 5.6345979620628204e-06, "loss": 2.7085, "step": 41691 }, { "epoch": 2.588118443106338, "grad_norm": 0.431134250559322, "learning_rate": 5.632932525066398e-06, "loss": 2.6117, "step": 41692 }, { "epoch": 2.588180520206096, "grad_norm": 0.4385178680335371, "learning_rate": 5.631267319543732e-06, "loss": 2.711, "step": 41693 }, { "epoch": 2.588242597305854, "grad_norm": 0.3922885377076749, "learning_rate": 5.629602345503521e-06, "loss": 2.7544, "step": 41694 }, { "epoch": 2.5883046744056117, "grad_norm": 0.38846810908491397, "learning_rate": 5.6279376029544365e-06, "loss": 2.6345, "step": 41695 }, { "epoch": 2.5883667515053697, "grad_norm": 0.4331576585773584, "learning_rate": 5.626273091905182e-06, "loss": 2.6528, "step": 41696 }, { "epoch": 2.5884288286051276, "grad_norm": 0.40571483795694013, "learning_rate": 5.624608812364429e-06, "loss": 2.6915, "step": 41697 }, { "epoch": 2.5884909057048855, "grad_norm": 0.3966103151742564, "learning_rate": 5.62294476434086e-06, "loss": 2.7158, "step": 41698 }, { "epoch": 2.5885529828046434, "grad_norm": 0.404108768507666, "learning_rate": 5.621280947843172e-06, "loss": 2.7456, "step": 41699 }, { "epoch": 2.5886150599044013, "grad_norm": 0.4246107338301912, "learning_rate": 5.61961736288003e-06, "loss": 2.7181, "step": 41700 }, { "epoch": 2.5886771370041592, "grad_norm": 0.4091244025198364, "learning_rate": 5.617954009460119e-06, "loss": 2.6802, "step": 41701 }, { "epoch": 2.588739214103917, "grad_norm": 0.39080924457394167, "learning_rate": 5.616290887592118e-06, "loss": 2.7053, "step": 41702 }, { "epoch": 2.588801291203675, "grad_norm": 0.3991675887200066, "learning_rate": 5.614627997284694e-06, "loss": 2.6812, "step": 41703 }, { "epoch": 2.588863368303433, "grad_norm": 0.396668020477843, "learning_rate": 5.612965338546538e-06, "loss": 2.7429, "step": 41704 }, { "epoch": 2.588925445403191, "grad_norm": 0.4286475862518968, "learning_rate": 5.611302911386318e-06, "loss": 2.6928, "step": 41705 }, { "epoch": 2.588987522502949, "grad_norm": 0.3866410225484442, "learning_rate": 5.609640715812703e-06, "loss": 2.7332, "step": 41706 }, { "epoch": 2.5890495996027063, "grad_norm": 0.3836933434006163, "learning_rate": 5.607978751834364e-06, "loss": 2.6177, "step": 41707 }, { "epoch": 2.5891116767024647, "grad_norm": 0.42681303038056523, "learning_rate": 5.606317019459984e-06, "loss": 2.7307, "step": 41708 }, { "epoch": 2.589173753802222, "grad_norm": 0.43305060588585953, "learning_rate": 5.604655518698221e-06, "loss": 2.7079, "step": 41709 }, { "epoch": 2.5892358309019805, "grad_norm": 0.4351284429133968, "learning_rate": 5.602994249557753e-06, "loss": 2.6799, "step": 41710 }, { "epoch": 2.589297908001738, "grad_norm": 0.37126264242359147, "learning_rate": 5.601333212047227e-06, "loss": 2.6905, "step": 41711 }, { "epoch": 2.589359985101496, "grad_norm": 0.4546476628526109, "learning_rate": 5.599672406175338e-06, "loss": 2.6908, "step": 41712 }, { "epoch": 2.589422062201254, "grad_norm": 0.4050997258370686, "learning_rate": 5.598011831950733e-06, "loss": 2.654, "step": 41713 }, { "epoch": 2.5894841393010117, "grad_norm": 0.42656422289825263, "learning_rate": 5.596351489382085e-06, "loss": 2.6958, "step": 41714 }, { "epoch": 2.5895462164007697, "grad_norm": 0.40151744545248036, "learning_rate": 5.594691378478045e-06, "loss": 2.6315, "step": 41715 }, { "epoch": 2.5896082935005276, "grad_norm": 0.4043511985365755, "learning_rate": 5.593031499247276e-06, "loss": 2.7003, "step": 41716 }, { "epoch": 2.5896703706002855, "grad_norm": 0.4228519148968292, "learning_rate": 5.591371851698446e-06, "loss": 2.7178, "step": 41717 }, { "epoch": 2.5897324477000434, "grad_norm": 0.418520180423021, "learning_rate": 5.5897124358402104e-06, "loss": 2.7114, "step": 41718 }, { "epoch": 2.5897945247998013, "grad_norm": 0.3740997037495855, "learning_rate": 5.588053251681224e-06, "loss": 2.704, "step": 41719 }, { "epoch": 2.5898566018995592, "grad_norm": 0.3869343744113821, "learning_rate": 5.586394299230141e-06, "loss": 2.716, "step": 41720 }, { "epoch": 2.589918678999317, "grad_norm": 0.40857239953708363, "learning_rate": 5.584735578495625e-06, "loss": 2.6622, "step": 41721 }, { "epoch": 2.589980756099075, "grad_norm": 0.412936984141983, "learning_rate": 5.583077089486333e-06, "loss": 2.6254, "step": 41722 }, { "epoch": 2.590042833198833, "grad_norm": 0.40634253255029085, "learning_rate": 5.5814188322109054e-06, "loss": 2.6688, "step": 41723 }, { "epoch": 2.590104910298591, "grad_norm": 0.4317247817908671, "learning_rate": 5.579760806677997e-06, "loss": 2.6675, "step": 41724 }, { "epoch": 2.590166987398349, "grad_norm": 0.3978092438893437, "learning_rate": 5.578103012896252e-06, "loss": 2.5785, "step": 41725 }, { "epoch": 2.5902290644981067, "grad_norm": 0.4305560264106281, "learning_rate": 5.5764454508743404e-06, "loss": 2.7137, "step": 41726 }, { "epoch": 2.5902911415978647, "grad_norm": 0.3979006685000161, "learning_rate": 5.574788120620894e-06, "loss": 2.7259, "step": 41727 }, { "epoch": 2.5903532186976226, "grad_norm": 0.42551711402615444, "learning_rate": 5.5731310221445506e-06, "loss": 2.7623, "step": 41728 }, { "epoch": 2.5904152957973805, "grad_norm": 0.3779089475948826, "learning_rate": 5.571474155453982e-06, "loss": 2.6688, "step": 41729 }, { "epoch": 2.590477372897138, "grad_norm": 0.4475797358127825, "learning_rate": 5.569817520557802e-06, "loss": 2.6275, "step": 41730 }, { "epoch": 2.5905394499968963, "grad_norm": 0.42460701388182504, "learning_rate": 5.568161117464687e-06, "loss": 2.7041, "step": 41731 }, { "epoch": 2.590601527096654, "grad_norm": 0.41982056926055916, "learning_rate": 5.56650494618326e-06, "loss": 2.5674, "step": 41732 }, { "epoch": 2.590663604196412, "grad_norm": 0.39073601098991295, "learning_rate": 5.564849006722161e-06, "loss": 2.6784, "step": 41733 }, { "epoch": 2.5907256812961696, "grad_norm": 0.43621134001920914, "learning_rate": 5.563193299090019e-06, "loss": 2.6969, "step": 41734 }, { "epoch": 2.590787758395928, "grad_norm": 0.38287892057333817, "learning_rate": 5.561537823295498e-06, "loss": 2.6603, "step": 41735 }, { "epoch": 2.5908498354956855, "grad_norm": 0.4200423666612027, "learning_rate": 5.559882579347225e-06, "loss": 2.66, "step": 41736 }, { "epoch": 2.5909119125954434, "grad_norm": 0.40553922095927025, "learning_rate": 5.558227567253832e-06, "loss": 2.7166, "step": 41737 }, { "epoch": 2.5909739896952013, "grad_norm": 0.394521618716077, "learning_rate": 5.556572787023951e-06, "loss": 2.6313, "step": 41738 }, { "epoch": 2.5910360667949592, "grad_norm": 0.45484351794683475, "learning_rate": 5.554918238666212e-06, "loss": 2.7924, "step": 41739 }, { "epoch": 2.591098143894717, "grad_norm": 0.4114036302378079, "learning_rate": 5.553263922189261e-06, "loss": 2.6587, "step": 41740 }, { "epoch": 2.591160220994475, "grad_norm": 0.3831913105332346, "learning_rate": 5.551609837601723e-06, "loss": 2.6665, "step": 41741 }, { "epoch": 2.591222298094233, "grad_norm": 0.4119168595734107, "learning_rate": 5.5499559849122255e-06, "loss": 2.4993, "step": 41742 }, { "epoch": 2.591284375193991, "grad_norm": 0.40216067695241225, "learning_rate": 5.548302364129393e-06, "loss": 2.6279, "step": 41743 }, { "epoch": 2.591346452293749, "grad_norm": 0.4046009413007232, "learning_rate": 5.546648975261864e-06, "loss": 2.607, "step": 41744 }, { "epoch": 2.5914085293935067, "grad_norm": 0.4308963763205252, "learning_rate": 5.54499581831826e-06, "loss": 2.5763, "step": 41745 }, { "epoch": 2.5914706064932647, "grad_norm": 0.40542542698377754, "learning_rate": 5.5433428933072015e-06, "loss": 2.6104, "step": 41746 }, { "epoch": 2.5915326835930226, "grad_norm": 0.3924722101615017, "learning_rate": 5.54169020023731e-06, "loss": 2.7128, "step": 41747 }, { "epoch": 2.5915947606927805, "grad_norm": 0.39953797896786464, "learning_rate": 5.540037739117221e-06, "loss": 2.662, "step": 41748 }, { "epoch": 2.5916568377925384, "grad_norm": 0.4406108610441374, "learning_rate": 5.5383855099555455e-06, "loss": 2.6723, "step": 41749 }, { "epoch": 2.5917189148922963, "grad_norm": 0.41366473471843346, "learning_rate": 5.536733512760911e-06, "loss": 2.6495, "step": 41750 }, { "epoch": 2.5917809919920543, "grad_norm": 0.41872745353580726, "learning_rate": 5.535081747541931e-06, "loss": 2.611, "step": 41751 }, { "epoch": 2.591843069091812, "grad_norm": 0.42336421839183325, "learning_rate": 5.533430214307211e-06, "loss": 2.6278, "step": 41752 }, { "epoch": 2.59190514619157, "grad_norm": 0.3714180382706175, "learning_rate": 5.531778913065394e-06, "loss": 2.7253, "step": 41753 }, { "epoch": 2.591967223291328, "grad_norm": 0.39123272351886446, "learning_rate": 5.530127843825078e-06, "loss": 2.6692, "step": 41754 }, { "epoch": 2.5920293003910855, "grad_norm": 0.38691396313684795, "learning_rate": 5.528477006594884e-06, "loss": 2.6454, "step": 41755 }, { "epoch": 2.592091377490844, "grad_norm": 0.4005187411144096, "learning_rate": 5.526826401383406e-06, "loss": 2.6242, "step": 41756 }, { "epoch": 2.5921534545906013, "grad_norm": 0.3855867676353542, "learning_rate": 5.525176028199286e-06, "loss": 2.6682, "step": 41757 }, { "epoch": 2.5922155316903597, "grad_norm": 0.4022156224351852, "learning_rate": 5.523525887051118e-06, "loss": 2.7618, "step": 41758 }, { "epoch": 2.592277608790117, "grad_norm": 0.40274723894533204, "learning_rate": 5.5218759779475106e-06, "loss": 2.655, "step": 41759 }, { "epoch": 2.592339685889875, "grad_norm": 0.41882390039181555, "learning_rate": 5.520226300897074e-06, "loss": 2.7554, "step": 41760 }, { "epoch": 2.592401762989633, "grad_norm": 0.3923176522882141, "learning_rate": 5.5185768559084085e-06, "loss": 2.6137, "step": 41761 }, { "epoch": 2.592463840089391, "grad_norm": 0.3912328758697021, "learning_rate": 5.516927642990122e-06, "loss": 2.7286, "step": 41762 }, { "epoch": 2.592525917189149, "grad_norm": 0.4010642734777882, "learning_rate": 5.51527866215083e-06, "loss": 2.6597, "step": 41763 }, { "epoch": 2.5925879942889067, "grad_norm": 0.4020509487741977, "learning_rate": 5.513629913399132e-06, "loss": 2.5943, "step": 41764 }, { "epoch": 2.5926500713886647, "grad_norm": 0.3898405208721474, "learning_rate": 5.511981396743626e-06, "loss": 2.6995, "step": 41765 }, { "epoch": 2.5927121484884226, "grad_norm": 0.41791257748199045, "learning_rate": 5.510333112192906e-06, "loss": 2.678, "step": 41766 }, { "epoch": 2.5927742255881805, "grad_norm": 0.4050961393877203, "learning_rate": 5.508685059755581e-06, "loss": 2.5889, "step": 41767 }, { "epoch": 2.5928363026879384, "grad_norm": 0.4014336746724875, "learning_rate": 5.507037239440255e-06, "loss": 2.6773, "step": 41768 }, { "epoch": 2.5928983797876963, "grad_norm": 0.42716211856095737, "learning_rate": 5.5053896512555106e-06, "loss": 2.6734, "step": 41769 }, { "epoch": 2.5929604568874542, "grad_norm": 0.3903622124794831, "learning_rate": 5.503742295209941e-06, "loss": 2.6206, "step": 41770 }, { "epoch": 2.593022533987212, "grad_norm": 0.3706832020172263, "learning_rate": 5.50209517131216e-06, "loss": 2.6637, "step": 41771 }, { "epoch": 2.59308461108697, "grad_norm": 0.3952668009039791, "learning_rate": 5.500448279570752e-06, "loss": 2.7082, "step": 41772 }, { "epoch": 2.593146688186728, "grad_norm": 0.46922961176426436, "learning_rate": 5.498801619994309e-06, "loss": 2.6232, "step": 41773 }, { "epoch": 2.593208765286486, "grad_norm": 0.4323330154713119, "learning_rate": 5.4971551925914185e-06, "loss": 2.6274, "step": 41774 }, { "epoch": 2.593270842386244, "grad_norm": 0.40066636057980587, "learning_rate": 5.495508997370663e-06, "loss": 2.6421, "step": 41775 }, { "epoch": 2.5933329194860018, "grad_norm": 0.42066991846764323, "learning_rate": 5.493863034340652e-06, "loss": 2.7228, "step": 41776 }, { "epoch": 2.5933949965857597, "grad_norm": 0.40627235571165227, "learning_rate": 5.492217303509961e-06, "loss": 2.6187, "step": 41777 }, { "epoch": 2.593457073685517, "grad_norm": 0.42856332935900004, "learning_rate": 5.490571804887173e-06, "loss": 2.6955, "step": 41778 }, { "epoch": 2.5935191507852755, "grad_norm": 0.4059165356268476, "learning_rate": 5.488926538480876e-06, "loss": 2.5649, "step": 41779 }, { "epoch": 2.593581227885033, "grad_norm": 0.3991452830698588, "learning_rate": 5.487281504299657e-06, "loss": 2.6384, "step": 41780 }, { "epoch": 2.5936433049847913, "grad_norm": 0.42636207021804384, "learning_rate": 5.485636702352098e-06, "loss": 2.614, "step": 41781 }, { "epoch": 2.593705382084549, "grad_norm": 0.4029655466473785, "learning_rate": 5.483992132646781e-06, "loss": 2.7074, "step": 41782 }, { "epoch": 2.593767459184307, "grad_norm": 0.39906990128048614, "learning_rate": 5.482347795192272e-06, "loss": 2.6398, "step": 41783 }, { "epoch": 2.5938295362840647, "grad_norm": 0.4244961374396917, "learning_rate": 5.480703689997174e-06, "loss": 2.6441, "step": 41784 }, { "epoch": 2.5938916133838226, "grad_norm": 0.43880513206200106, "learning_rate": 5.479059817070048e-06, "loss": 2.5667, "step": 41785 }, { "epoch": 2.5939536904835805, "grad_norm": 0.43402584986788345, "learning_rate": 5.477416176419475e-06, "loss": 2.5827, "step": 41786 }, { "epoch": 2.5940157675833384, "grad_norm": 0.4170567911387034, "learning_rate": 5.475772768054033e-06, "loss": 2.6288, "step": 41787 }, { "epoch": 2.5940778446830963, "grad_norm": 0.39251249153766726, "learning_rate": 5.474129591982286e-06, "loss": 2.6423, "step": 41788 }, { "epoch": 2.5941399217828542, "grad_norm": 0.4343799645668478, "learning_rate": 5.472486648212816e-06, "loss": 2.7457, "step": 41789 }, { "epoch": 2.594201998882612, "grad_norm": 0.41535824339362026, "learning_rate": 5.470843936754199e-06, "loss": 2.6441, "step": 41790 }, { "epoch": 2.59426407598237, "grad_norm": 0.4277198301307104, "learning_rate": 5.469201457614997e-06, "loss": 2.7145, "step": 41791 }, { "epoch": 2.594326153082128, "grad_norm": 0.39286979354578766, "learning_rate": 5.467559210803769e-06, "loss": 2.6415, "step": 41792 }, { "epoch": 2.594388230181886, "grad_norm": 0.4273335045906166, "learning_rate": 5.465917196329107e-06, "loss": 2.7043, "step": 41793 }, { "epoch": 2.594450307281644, "grad_norm": 0.4272671067656301, "learning_rate": 5.4642754141995545e-06, "loss": 2.7321, "step": 41794 }, { "epoch": 2.5945123843814017, "grad_norm": 0.44623520273631206, "learning_rate": 5.4626338644236995e-06, "loss": 2.7011, "step": 41795 }, { "epoch": 2.5945744614811597, "grad_norm": 0.44380076950744884, "learning_rate": 5.460992547010097e-06, "loss": 2.6601, "step": 41796 }, { "epoch": 2.5946365385809176, "grad_norm": 0.40880084651870574, "learning_rate": 5.4593514619672995e-06, "loss": 2.7035, "step": 41797 }, { "epoch": 2.5946986156806755, "grad_norm": 0.4066777544907265, "learning_rate": 5.457710609303884e-06, "loss": 2.6167, "step": 41798 }, { "epoch": 2.5947606927804334, "grad_norm": 0.44938122672063174, "learning_rate": 5.456069989028412e-06, "loss": 2.659, "step": 41799 }, { "epoch": 2.5948227698801913, "grad_norm": 0.4309258216951878, "learning_rate": 5.45442960114943e-06, "loss": 2.6716, "step": 41800 }, { "epoch": 2.5948848469799493, "grad_norm": 0.40238493180853285, "learning_rate": 5.452789445675505e-06, "loss": 2.7222, "step": 41801 }, { "epoch": 2.594946924079707, "grad_norm": 0.3847077999135393, "learning_rate": 5.451149522615179e-06, "loss": 2.6343, "step": 41802 }, { "epoch": 2.5950090011794646, "grad_norm": 0.4182061920253086, "learning_rate": 5.4495098319770346e-06, "loss": 2.6853, "step": 41803 }, { "epoch": 2.595071078279223, "grad_norm": 0.3991151171284151, "learning_rate": 5.447870373769609e-06, "loss": 2.66, "step": 41804 }, { "epoch": 2.5951331553789805, "grad_norm": 0.4000403197650282, "learning_rate": 5.446231148001462e-06, "loss": 2.681, "step": 41805 }, { "epoch": 2.595195232478739, "grad_norm": 0.4099319207843703, "learning_rate": 5.444592154681128e-06, "loss": 2.6526, "step": 41806 }, { "epoch": 2.5952573095784963, "grad_norm": 0.3859600948561363, "learning_rate": 5.44295339381719e-06, "loss": 2.7174, "step": 41807 }, { "epoch": 2.5953193866782542, "grad_norm": 0.4226560250667742, "learning_rate": 5.441314865418173e-06, "loss": 2.7505, "step": 41808 }, { "epoch": 2.595381463778012, "grad_norm": 0.4366416344432182, "learning_rate": 5.439676569492641e-06, "loss": 2.6542, "step": 41809 }, { "epoch": 2.59544354087777, "grad_norm": 0.40516766006521077, "learning_rate": 5.438038506049131e-06, "loss": 2.6421, "step": 41810 }, { "epoch": 2.595505617977528, "grad_norm": 0.397728569784155, "learning_rate": 5.436400675096176e-06, "loss": 2.6837, "step": 41811 }, { "epoch": 2.595567695077286, "grad_norm": 0.37599166245296733, "learning_rate": 5.434763076642352e-06, "loss": 2.5993, "step": 41812 }, { "epoch": 2.595629772177044, "grad_norm": 0.40088344111415675, "learning_rate": 5.433125710696191e-06, "loss": 2.6648, "step": 41813 }, { "epoch": 2.5956918492768017, "grad_norm": 0.41735047218224747, "learning_rate": 5.431488577266231e-06, "loss": 2.6499, "step": 41814 }, { "epoch": 2.5957539263765597, "grad_norm": 0.37751686358522, "learning_rate": 5.429851676360998e-06, "loss": 2.6356, "step": 41815 }, { "epoch": 2.5958160034763176, "grad_norm": 0.4195208031403232, "learning_rate": 5.4282150079890625e-06, "loss": 2.6681, "step": 41816 }, { "epoch": 2.5958780805760755, "grad_norm": 0.4351572423680494, "learning_rate": 5.426578572158952e-06, "loss": 2.6311, "step": 41817 }, { "epoch": 2.5959401576758334, "grad_norm": 0.4118863989565211, "learning_rate": 5.424942368879199e-06, "loss": 2.6652, "step": 41818 }, { "epoch": 2.5960022347755913, "grad_norm": 0.42437281963366935, "learning_rate": 5.423306398158334e-06, "loss": 2.6772, "step": 41819 }, { "epoch": 2.5960643118753493, "grad_norm": 0.42944901523910217, "learning_rate": 5.421670660004907e-06, "loss": 2.6556, "step": 41820 }, { "epoch": 2.596126388975107, "grad_norm": 0.4182193931523907, "learning_rate": 5.420035154427449e-06, "loss": 2.6706, "step": 41821 }, { "epoch": 2.596188466074865, "grad_norm": 0.41766508604693364, "learning_rate": 5.4183998814344885e-06, "loss": 2.5974, "step": 41822 }, { "epoch": 2.596250543174623, "grad_norm": 0.4074663147138612, "learning_rate": 5.416764841034561e-06, "loss": 2.6381, "step": 41823 }, { "epoch": 2.596312620274381, "grad_norm": 0.41077155327044196, "learning_rate": 5.4151300332361825e-06, "loss": 2.6636, "step": 41824 }, { "epoch": 2.596374697374139, "grad_norm": 0.4187540716001787, "learning_rate": 5.413495458047907e-06, "loss": 2.5916, "step": 41825 }, { "epoch": 2.5964367744738963, "grad_norm": 0.41030141895377953, "learning_rate": 5.411861115478234e-06, "loss": 2.6192, "step": 41826 }, { "epoch": 2.5964988515736547, "grad_norm": 0.40830006885626086, "learning_rate": 5.410227005535718e-06, "loss": 2.7417, "step": 41827 }, { "epoch": 2.596560928673412, "grad_norm": 0.3989843467949924, "learning_rate": 5.408593128228872e-06, "loss": 2.6497, "step": 41828 }, { "epoch": 2.5966230057731705, "grad_norm": 0.38232096430731927, "learning_rate": 5.406959483566215e-06, "loss": 2.6604, "step": 41829 }, { "epoch": 2.596685082872928, "grad_norm": 0.4153640013623919, "learning_rate": 5.405326071556283e-06, "loss": 2.6473, "step": 41830 }, { "epoch": 2.5967471599726863, "grad_norm": 0.4173815336729959, "learning_rate": 5.4036928922075846e-06, "loss": 2.6583, "step": 41831 }, { "epoch": 2.596809237072444, "grad_norm": 0.41089748434316514, "learning_rate": 5.402059945528654e-06, "loss": 2.6122, "step": 41832 }, { "epoch": 2.5968713141722017, "grad_norm": 0.39264598658453753, "learning_rate": 5.400427231527988e-06, "loss": 2.6128, "step": 41833 }, { "epoch": 2.5969333912719597, "grad_norm": 0.43329856034595887, "learning_rate": 5.398794750214136e-06, "loss": 2.7298, "step": 41834 }, { "epoch": 2.5969954683717176, "grad_norm": 0.4133675158918734, "learning_rate": 5.397162501595593e-06, "loss": 2.6131, "step": 41835 }, { "epoch": 2.5970575454714755, "grad_norm": 0.40572919841500166, "learning_rate": 5.395530485680883e-06, "loss": 2.7016, "step": 41836 }, { "epoch": 2.5971196225712334, "grad_norm": 0.39083063435113996, "learning_rate": 5.393898702478523e-06, "loss": 2.6884, "step": 41837 }, { "epoch": 2.5971816996709913, "grad_norm": 0.41790571361322765, "learning_rate": 5.392267151997004e-06, "loss": 2.7179, "step": 41838 }, { "epoch": 2.5972437767707492, "grad_norm": 0.3927585699529281, "learning_rate": 5.390635834244873e-06, "loss": 2.6737, "step": 41839 }, { "epoch": 2.597305853870507, "grad_norm": 0.3988503206150143, "learning_rate": 5.38900474923062e-06, "loss": 2.7428, "step": 41840 }, { "epoch": 2.597367930970265, "grad_norm": 0.37461471633364973, "learning_rate": 5.387373896962761e-06, "loss": 2.6208, "step": 41841 }, { "epoch": 2.597430008070023, "grad_norm": 0.4189929890569571, "learning_rate": 5.385743277449795e-06, "loss": 2.7896, "step": 41842 }, { "epoch": 2.597492085169781, "grad_norm": 0.4375808164339771, "learning_rate": 5.384112890700244e-06, "loss": 2.6204, "step": 41843 }, { "epoch": 2.597554162269539, "grad_norm": 0.4200744351776036, "learning_rate": 5.382482736722605e-06, "loss": 2.6434, "step": 41844 }, { "epoch": 2.5976162393692968, "grad_norm": 0.39314094926535625, "learning_rate": 5.380852815525389e-06, "loss": 2.7768, "step": 41845 }, { "epoch": 2.5976783164690547, "grad_norm": 0.4275027828237519, "learning_rate": 5.3792231271170886e-06, "loss": 2.6983, "step": 41846 }, { "epoch": 2.5977403935688126, "grad_norm": 0.41905408394233007, "learning_rate": 5.377593671506209e-06, "loss": 2.6653, "step": 41847 }, { "epoch": 2.5978024706685705, "grad_norm": 0.41530417558121857, "learning_rate": 5.375964448701265e-06, "loss": 2.6194, "step": 41848 }, { "epoch": 2.5978645477683284, "grad_norm": 0.4287006316108491, "learning_rate": 5.37433545871075e-06, "loss": 2.5987, "step": 41849 }, { "epoch": 2.5979266248680863, "grad_norm": 0.40592556616528747, "learning_rate": 5.372706701543156e-06, "loss": 2.6866, "step": 41850 }, { "epoch": 2.597988701967844, "grad_norm": 0.378260681935647, "learning_rate": 5.3710781772069726e-06, "loss": 2.6707, "step": 41851 }, { "epoch": 2.598050779067602, "grad_norm": 0.40987451959217114, "learning_rate": 5.369449885710725e-06, "loss": 2.6801, "step": 41852 }, { "epoch": 2.5981128561673597, "grad_norm": 0.4000065706130622, "learning_rate": 5.367821827062886e-06, "loss": 2.7092, "step": 41853 }, { "epoch": 2.598174933267118, "grad_norm": 0.4164653192245588, "learning_rate": 5.366194001271957e-06, "loss": 2.7128, "step": 41854 }, { "epoch": 2.5982370103668755, "grad_norm": 0.4366241611960823, "learning_rate": 5.364566408346422e-06, "loss": 2.5774, "step": 41855 }, { "epoch": 2.5982990874666334, "grad_norm": 0.4122811813556837, "learning_rate": 5.362939048294785e-06, "loss": 2.6494, "step": 41856 }, { "epoch": 2.5983611645663913, "grad_norm": 0.4263990277621903, "learning_rate": 5.361311921125534e-06, "loss": 2.6935, "step": 41857 }, { "epoch": 2.5984232416661492, "grad_norm": 0.4094608952047575, "learning_rate": 5.359685026847155e-06, "loss": 2.6578, "step": 41858 }, { "epoch": 2.598485318765907, "grad_norm": 0.43632777733989586, "learning_rate": 5.358058365468127e-06, "loss": 2.6831, "step": 41859 }, { "epoch": 2.598547395865665, "grad_norm": 0.4141804868911128, "learning_rate": 5.356431936996953e-06, "loss": 2.6526, "step": 41860 }, { "epoch": 2.598609472965423, "grad_norm": 0.4056419076319968, "learning_rate": 5.354805741442104e-06, "loss": 2.6538, "step": 41861 }, { "epoch": 2.598671550065181, "grad_norm": 0.41059020564172966, "learning_rate": 5.353179778812078e-06, "loss": 2.634, "step": 41862 }, { "epoch": 2.598733627164939, "grad_norm": 0.420118277108071, "learning_rate": 5.351554049115353e-06, "loss": 2.6593, "step": 41863 }, { "epoch": 2.5987957042646967, "grad_norm": 0.42596835653664544, "learning_rate": 5.349928552360411e-06, "loss": 2.6189, "step": 41864 }, { "epoch": 2.5988577813644547, "grad_norm": 0.42025551959241925, "learning_rate": 5.348303288555722e-06, "loss": 2.7009, "step": 41865 }, { "epoch": 2.5989198584642126, "grad_norm": 0.41863527363278585, "learning_rate": 5.346678257709781e-06, "loss": 2.544, "step": 41866 }, { "epoch": 2.5989819355639705, "grad_norm": 0.37782643014044887, "learning_rate": 5.345053459831062e-06, "loss": 2.6588, "step": 41867 }, { "epoch": 2.5990440126637284, "grad_norm": 0.4217215541535778, "learning_rate": 5.343428894928038e-06, "loss": 2.7158, "step": 41868 }, { "epoch": 2.5991060897634863, "grad_norm": 0.43637802808547455, "learning_rate": 5.3418045630091725e-06, "loss": 2.5744, "step": 41869 }, { "epoch": 2.5991681668632443, "grad_norm": 0.36377346369518754, "learning_rate": 5.340180464082972e-06, "loss": 2.7278, "step": 41870 }, { "epoch": 2.599230243963002, "grad_norm": 0.39951097576451783, "learning_rate": 5.338556598157885e-06, "loss": 2.7704, "step": 41871 }, { "epoch": 2.59929232106276, "grad_norm": 0.4098458341602964, "learning_rate": 5.336932965242392e-06, "loss": 2.661, "step": 41872 }, { "epoch": 2.599354398162518, "grad_norm": 0.44548371018598315, "learning_rate": 5.335309565344965e-06, "loss": 2.5993, "step": 41873 }, { "epoch": 2.5994164752622755, "grad_norm": 0.3925090178589644, "learning_rate": 5.3336863984740524e-06, "loss": 2.5788, "step": 41874 }, { "epoch": 2.599478552362034, "grad_norm": 0.39492053475987343, "learning_rate": 5.3320634646381595e-06, "loss": 2.7125, "step": 41875 }, { "epoch": 2.5995406294617913, "grad_norm": 0.40319888894158884, "learning_rate": 5.330440763845734e-06, "loss": 2.6638, "step": 41876 }, { "epoch": 2.5996027065615497, "grad_norm": 0.40398230731223245, "learning_rate": 5.328818296105237e-06, "loss": 2.7074, "step": 41877 }, { "epoch": 2.599664783661307, "grad_norm": 0.4105584851627731, "learning_rate": 5.327196061425138e-06, "loss": 2.6146, "step": 41878 }, { "epoch": 2.5997268607610655, "grad_norm": 0.41570309643816117, "learning_rate": 5.325574059813903e-06, "loss": 2.6603, "step": 41879 }, { "epoch": 2.599788937860823, "grad_norm": 0.4204597399599274, "learning_rate": 5.323952291279999e-06, "loss": 2.6972, "step": 41880 }, { "epoch": 2.599851014960581, "grad_norm": 0.39565990019861186, "learning_rate": 5.322330755831878e-06, "loss": 2.6368, "step": 41881 }, { "epoch": 2.599913092060339, "grad_norm": 0.3923097898551625, "learning_rate": 5.320709453477995e-06, "loss": 2.6648, "step": 41882 }, { "epoch": 2.5999751691600967, "grad_norm": 0.3882766899900333, "learning_rate": 5.319088384226828e-06, "loss": 2.6689, "step": 41883 }, { "epoch": 2.6000372462598547, "grad_norm": 0.3935678257417534, "learning_rate": 5.317467548086824e-06, "loss": 2.7013, "step": 41884 }, { "epoch": 2.6000993233596126, "grad_norm": 0.3886861558907199, "learning_rate": 5.315846945066438e-06, "loss": 2.6753, "step": 41885 }, { "epoch": 2.6001614004593705, "grad_norm": 0.40606522025814323, "learning_rate": 5.314226575174125e-06, "loss": 2.7438, "step": 41886 }, { "epoch": 2.6002234775591284, "grad_norm": 0.39884591207715536, "learning_rate": 5.312606438418333e-06, "loss": 2.6939, "step": 41887 }, { "epoch": 2.6002855546588863, "grad_norm": 0.39901121758464975, "learning_rate": 5.3109865348075275e-06, "loss": 2.7087, "step": 41888 }, { "epoch": 2.6003476317586443, "grad_norm": 0.40982240078083865, "learning_rate": 5.309366864350151e-06, "loss": 2.731, "step": 41889 }, { "epoch": 2.600409708858402, "grad_norm": 0.3902238693579691, "learning_rate": 5.307747427054666e-06, "loss": 2.6435, "step": 41890 }, { "epoch": 2.60047178595816, "grad_norm": 0.38109946874978384, "learning_rate": 5.306128222929496e-06, "loss": 2.5781, "step": 41891 }, { "epoch": 2.600533863057918, "grad_norm": 0.42224678682128963, "learning_rate": 5.304509251983103e-06, "loss": 2.6055, "step": 41892 }, { "epoch": 2.600595940157676, "grad_norm": 0.4330560942055092, "learning_rate": 5.302890514223952e-06, "loss": 2.7071, "step": 41893 }, { "epoch": 2.600658017257434, "grad_norm": 0.3881684306320156, "learning_rate": 5.301272009660469e-06, "loss": 2.6792, "step": 41894 }, { "epoch": 2.6007200943571918, "grad_norm": 0.4204231464005098, "learning_rate": 5.299653738301097e-06, "loss": 2.7668, "step": 41895 }, { "epoch": 2.6007821714569497, "grad_norm": 0.436927284310098, "learning_rate": 5.298035700154286e-06, "loss": 2.7033, "step": 41896 }, { "epoch": 2.6008442485567076, "grad_norm": 0.4085130929202734, "learning_rate": 5.296417895228462e-06, "loss": 2.6172, "step": 41897 }, { "epoch": 2.6009063256564655, "grad_norm": 0.41456783695428945, "learning_rate": 5.29480032353209e-06, "loss": 2.6013, "step": 41898 }, { "epoch": 2.600968402756223, "grad_norm": 0.4078690654066868, "learning_rate": 5.293182985073597e-06, "loss": 2.6009, "step": 41899 }, { "epoch": 2.6010304798559813, "grad_norm": 0.4447269102957148, "learning_rate": 5.2915658798614264e-06, "loss": 2.6552, "step": 41900 }, { "epoch": 2.601092556955739, "grad_norm": 0.4108140847137682, "learning_rate": 5.289949007903994e-06, "loss": 2.6267, "step": 41901 }, { "epoch": 2.601154634055497, "grad_norm": 0.4272573240711534, "learning_rate": 5.288332369209764e-06, "loss": 2.6478, "step": 41902 }, { "epoch": 2.6012167111552547, "grad_norm": 0.412541926305132, "learning_rate": 5.286715963787153e-06, "loss": 2.7306, "step": 41903 }, { "epoch": 2.6012787882550126, "grad_norm": 0.41823661990747346, "learning_rate": 5.285099791644604e-06, "loss": 2.6393, "step": 41904 }, { "epoch": 2.6013408653547705, "grad_norm": 0.4314352556734206, "learning_rate": 5.283483852790538e-06, "loss": 2.6439, "step": 41905 }, { "epoch": 2.6014029424545284, "grad_norm": 0.4288910333265352, "learning_rate": 5.281868147233393e-06, "loss": 2.7289, "step": 41906 }, { "epoch": 2.6014650195542863, "grad_norm": 0.4215067916721155, "learning_rate": 5.280252674981606e-06, "loss": 2.6922, "step": 41907 }, { "epoch": 2.6015270966540442, "grad_norm": 0.39051040875975196, "learning_rate": 5.278637436043593e-06, "loss": 2.6022, "step": 41908 }, { "epoch": 2.601589173753802, "grad_norm": 0.4373872816111816, "learning_rate": 5.277022430427786e-06, "loss": 2.6742, "step": 41909 }, { "epoch": 2.60165125085356, "grad_norm": 0.4540371264609681, "learning_rate": 5.2754076581425946e-06, "loss": 2.7194, "step": 41910 }, { "epoch": 2.601713327953318, "grad_norm": 0.4206751255362796, "learning_rate": 5.273793119196474e-06, "loss": 2.5424, "step": 41911 }, { "epoch": 2.601775405053076, "grad_norm": 0.4130500254791385, "learning_rate": 5.2721788135978335e-06, "loss": 2.7081, "step": 41912 }, { "epoch": 2.601837482152834, "grad_norm": 0.41278848241782634, "learning_rate": 5.270564741355094e-06, "loss": 2.7905, "step": 41913 }, { "epoch": 2.6018995592525918, "grad_norm": 0.4036635999973895, "learning_rate": 5.26895090247666e-06, "loss": 2.6911, "step": 41914 }, { "epoch": 2.6019616363523497, "grad_norm": 0.41259315984860423, "learning_rate": 5.267337296970987e-06, "loss": 2.6571, "step": 41915 }, { "epoch": 2.6020237134521076, "grad_norm": 0.40468149733049863, "learning_rate": 5.2657239248464665e-06, "loss": 2.6733, "step": 41916 }, { "epoch": 2.6020857905518655, "grad_norm": 0.42988526388229387, "learning_rate": 5.2641107861115315e-06, "loss": 2.5976, "step": 41917 }, { "epoch": 2.6021478676516234, "grad_norm": 0.39917690259329897, "learning_rate": 5.262497880774575e-06, "loss": 2.6473, "step": 41918 }, { "epoch": 2.6022099447513813, "grad_norm": 0.3788211569653637, "learning_rate": 5.260885208844041e-06, "loss": 2.6843, "step": 41919 }, { "epoch": 2.6022720218511393, "grad_norm": 0.4021931894310466, "learning_rate": 5.259272770328328e-06, "loss": 2.7012, "step": 41920 }, { "epoch": 2.602334098950897, "grad_norm": 0.41668173246167056, "learning_rate": 5.25766056523585e-06, "loss": 2.6493, "step": 41921 }, { "epoch": 2.6023961760506547, "grad_norm": 0.4080774214089586, "learning_rate": 5.256048593575019e-06, "loss": 2.7294, "step": 41922 }, { "epoch": 2.602458253150413, "grad_norm": 0.4135689222769064, "learning_rate": 5.254436855354233e-06, "loss": 2.6601, "step": 41923 }, { "epoch": 2.6025203302501705, "grad_norm": 0.4070355632941503, "learning_rate": 5.252825350581925e-06, "loss": 2.7458, "step": 41924 }, { "epoch": 2.602582407349929, "grad_norm": 0.42755546486080725, "learning_rate": 5.2512140792664754e-06, "loss": 2.7131, "step": 41925 }, { "epoch": 2.6026444844496863, "grad_norm": 0.40830630955834574, "learning_rate": 5.2496030414163176e-06, "loss": 2.7351, "step": 41926 }, { "epoch": 2.6027065615494447, "grad_norm": 0.39900380448987044, "learning_rate": 5.2479922370398445e-06, "loss": 2.7009, "step": 41927 }, { "epoch": 2.602768638649202, "grad_norm": 0.4043320718823154, "learning_rate": 5.2463816661454504e-06, "loss": 2.6349, "step": 41928 }, { "epoch": 2.60283071574896, "grad_norm": 0.43846451502776135, "learning_rate": 5.24477132874156e-06, "loss": 2.7537, "step": 41929 }, { "epoch": 2.602892792848718, "grad_norm": 0.41094023845035754, "learning_rate": 5.243161224836557e-06, "loss": 2.6241, "step": 41930 }, { "epoch": 2.602954869948476, "grad_norm": 0.39993673509732386, "learning_rate": 5.241551354438851e-06, "loss": 2.6744, "step": 41931 }, { "epoch": 2.603016947048234, "grad_norm": 0.4144178880674424, "learning_rate": 5.239941717556834e-06, "loss": 2.6479, "step": 41932 }, { "epoch": 2.6030790241479917, "grad_norm": 0.42742537742299547, "learning_rate": 5.238332314198896e-06, "loss": 2.7438, "step": 41933 }, { "epoch": 2.6031411012477497, "grad_norm": 0.3999153293462196, "learning_rate": 5.236723144373456e-06, "loss": 2.7552, "step": 41934 }, { "epoch": 2.6032031783475076, "grad_norm": 0.41985315159733155, "learning_rate": 5.235114208088904e-06, "loss": 2.6964, "step": 41935 }, { "epoch": 2.6032652554472655, "grad_norm": 0.3763282023234384, "learning_rate": 5.23350550535362e-06, "loss": 2.6346, "step": 41936 }, { "epoch": 2.6033273325470234, "grad_norm": 0.3810633773243024, "learning_rate": 5.231897036175998e-06, "loss": 2.6912, "step": 41937 }, { "epoch": 2.6033894096467813, "grad_norm": 0.43027951465250935, "learning_rate": 5.230288800564448e-06, "loss": 2.6759, "step": 41938 }, { "epoch": 2.6034514867465393, "grad_norm": 0.42302388474823693, "learning_rate": 5.228680798527347e-06, "loss": 2.6506, "step": 41939 }, { "epoch": 2.603513563846297, "grad_norm": 0.4203083832266189, "learning_rate": 5.227073030073093e-06, "loss": 2.6326, "step": 41940 }, { "epoch": 2.603575640946055, "grad_norm": 0.4119527980340868, "learning_rate": 5.225465495210052e-06, "loss": 2.6198, "step": 41941 }, { "epoch": 2.603637718045813, "grad_norm": 0.42583906148702116, "learning_rate": 5.22385819394664e-06, "loss": 2.6838, "step": 41942 }, { "epoch": 2.603699795145571, "grad_norm": 0.39863426361492715, "learning_rate": 5.222251126291228e-06, "loss": 2.5724, "step": 41943 }, { "epoch": 2.603761872245329, "grad_norm": 0.4351983322259295, "learning_rate": 5.220644292252202e-06, "loss": 2.5785, "step": 41944 }, { "epoch": 2.6038239493450868, "grad_norm": 0.4149428850822285, "learning_rate": 5.219037691837947e-06, "loss": 2.6669, "step": 41945 }, { "epoch": 2.6038860264448447, "grad_norm": 0.3750508194502898, "learning_rate": 5.217431325056832e-06, "loss": 2.6674, "step": 41946 }, { "epoch": 2.603948103544602, "grad_norm": 0.406888148802903, "learning_rate": 5.215825191917256e-06, "loss": 2.6946, "step": 41947 }, { "epoch": 2.6040101806443605, "grad_norm": 0.41661998566758274, "learning_rate": 5.214219292427597e-06, "loss": 2.728, "step": 41948 }, { "epoch": 2.604072257744118, "grad_norm": 0.4188267644580103, "learning_rate": 5.212613626596225e-06, "loss": 2.7064, "step": 41949 }, { "epoch": 2.6041343348438764, "grad_norm": 0.4494706981505057, "learning_rate": 5.21100819443151e-06, "loss": 2.6852, "step": 41950 }, { "epoch": 2.604196411943634, "grad_norm": 0.39471603069949757, "learning_rate": 5.209402995941848e-06, "loss": 2.6332, "step": 41951 }, { "epoch": 2.6042584890433917, "grad_norm": 0.3938836362915516, "learning_rate": 5.207798031135608e-06, "loss": 2.5851, "step": 41952 }, { "epoch": 2.6043205661431497, "grad_norm": 0.3867120067894772, "learning_rate": 5.206193300021156e-06, "loss": 2.7397, "step": 41953 }, { "epoch": 2.6043826432429076, "grad_norm": 0.42890599012801284, "learning_rate": 5.2045888026068575e-06, "loss": 2.7638, "step": 41954 }, { "epoch": 2.6044447203426655, "grad_norm": 0.4328194827676738, "learning_rate": 5.202984538901107e-06, "loss": 2.6259, "step": 41955 }, { "epoch": 2.6045067974424234, "grad_norm": 0.37658887272206404, "learning_rate": 5.201380508912257e-06, "loss": 2.6623, "step": 41956 }, { "epoch": 2.6045688745421813, "grad_norm": 0.3913073302425419, "learning_rate": 5.199776712648669e-06, "loss": 2.6095, "step": 41957 }, { "epoch": 2.6046309516419393, "grad_norm": 0.39559151976360746, "learning_rate": 5.198173150118734e-06, "loss": 2.627, "step": 41958 }, { "epoch": 2.604693028741697, "grad_norm": 0.4105987523310807, "learning_rate": 5.1965698213308045e-06, "loss": 2.5786, "step": 41959 }, { "epoch": 2.604755105841455, "grad_norm": 0.40830034257326464, "learning_rate": 5.194966726293238e-06, "loss": 2.6645, "step": 41960 }, { "epoch": 2.604817182941213, "grad_norm": 0.36675866428690346, "learning_rate": 5.193363865014417e-06, "loss": 2.5877, "step": 41961 }, { "epoch": 2.604879260040971, "grad_norm": 0.4006191996840384, "learning_rate": 5.191761237502696e-06, "loss": 2.7443, "step": 41962 }, { "epoch": 2.604941337140729, "grad_norm": 0.3992600832031936, "learning_rate": 5.190158843766429e-06, "loss": 2.7071, "step": 41963 }, { "epoch": 2.6050034142404868, "grad_norm": 0.43594998577654726, "learning_rate": 5.188556683813978e-06, "loss": 2.6416, "step": 41964 }, { "epoch": 2.6050654913402447, "grad_norm": 0.4299972111589349, "learning_rate": 5.186954757653712e-06, "loss": 2.6261, "step": 41965 }, { "epoch": 2.6051275684400026, "grad_norm": 0.4211498138248431, "learning_rate": 5.1853530652939796e-06, "loss": 2.6506, "step": 41966 }, { "epoch": 2.6051896455397605, "grad_norm": 0.4307334999626057, "learning_rate": 5.183751606743142e-06, "loss": 2.6454, "step": 41967 }, { "epoch": 2.6052517226395184, "grad_norm": 0.4513110487000275, "learning_rate": 5.1821503820095375e-06, "loss": 2.7885, "step": 41968 }, { "epoch": 2.6053137997392763, "grad_norm": 0.41426545693171063, "learning_rate": 5.1805493911015465e-06, "loss": 2.6195, "step": 41969 }, { "epoch": 2.605375876839034, "grad_norm": 0.3829554271520627, "learning_rate": 5.178948634027514e-06, "loss": 2.6781, "step": 41970 }, { "epoch": 2.605437953938792, "grad_norm": 0.4553943893949787, "learning_rate": 5.177348110795782e-06, "loss": 2.6692, "step": 41971 }, { "epoch": 2.6055000310385497, "grad_norm": 0.4104968518706772, "learning_rate": 5.175747821414706e-06, "loss": 2.7314, "step": 41972 }, { "epoch": 2.605562108138308, "grad_norm": 0.4360560532845542, "learning_rate": 5.174147765892628e-06, "loss": 2.687, "step": 41973 }, { "epoch": 2.6056241852380655, "grad_norm": 0.43036392029353365, "learning_rate": 5.17254794423791e-06, "loss": 2.6484, "step": 41974 }, { "epoch": 2.6056862623378234, "grad_norm": 0.4019218936868702, "learning_rate": 5.170948356458893e-06, "loss": 2.6913, "step": 41975 }, { "epoch": 2.6057483394375813, "grad_norm": 0.3854816184429874, "learning_rate": 5.169349002563923e-06, "loss": 2.5582, "step": 41976 }, { "epoch": 2.6058104165373392, "grad_norm": 0.4200800575579388, "learning_rate": 5.167749882561335e-06, "loss": 2.668, "step": 41977 }, { "epoch": 2.605872493637097, "grad_norm": 0.41345450910553744, "learning_rate": 5.166150996459484e-06, "loss": 2.6212, "step": 41978 }, { "epoch": 2.605934570736855, "grad_norm": 0.40723567376132574, "learning_rate": 5.164552344266704e-06, "loss": 2.6282, "step": 41979 }, { "epoch": 2.605996647836613, "grad_norm": 0.41085026471427866, "learning_rate": 5.162953925991343e-06, "loss": 2.664, "step": 41980 }, { "epoch": 2.606058724936371, "grad_norm": 0.3911736831028991, "learning_rate": 5.161355741641738e-06, "loss": 2.6282, "step": 41981 }, { "epoch": 2.606120802036129, "grad_norm": 0.4061227195595882, "learning_rate": 5.15975779122621e-06, "loss": 2.7099, "step": 41982 }, { "epoch": 2.6061828791358868, "grad_norm": 0.4153612927951241, "learning_rate": 5.1581600747531265e-06, "loss": 2.6828, "step": 41983 }, { "epoch": 2.6062449562356447, "grad_norm": 0.5101987672061897, "learning_rate": 5.156562592230802e-06, "loss": 2.6667, "step": 41984 }, { "epoch": 2.6063070333354026, "grad_norm": 0.4119234614884866, "learning_rate": 5.154965343667584e-06, "loss": 2.6335, "step": 41985 }, { "epoch": 2.6063691104351605, "grad_norm": 0.40970197679677295, "learning_rate": 5.153368329071778e-06, "loss": 2.6782, "step": 41986 }, { "epoch": 2.6064311875349184, "grad_norm": 0.38885077954886527, "learning_rate": 5.151771548451756e-06, "loss": 2.5897, "step": 41987 }, { "epoch": 2.6064932646346763, "grad_norm": 0.4192250880367613, "learning_rate": 5.150175001815821e-06, "loss": 2.6453, "step": 41988 }, { "epoch": 2.6065553417344343, "grad_norm": 0.39172017115855173, "learning_rate": 5.148578689172317e-06, "loss": 2.5884, "step": 41989 }, { "epoch": 2.606617418834192, "grad_norm": 0.4030397191884818, "learning_rate": 5.146982610529555e-06, "loss": 2.6672, "step": 41990 }, { "epoch": 2.60667949593395, "grad_norm": 0.3973358482328187, "learning_rate": 5.145386765895871e-06, "loss": 2.6042, "step": 41991 }, { "epoch": 2.606741573033708, "grad_norm": 0.43173035599484716, "learning_rate": 5.143791155279604e-06, "loss": 2.644, "step": 41992 }, { "epoch": 2.6068036501334655, "grad_norm": 0.44085150436360154, "learning_rate": 5.142195778689068e-06, "loss": 2.6182, "step": 41993 }, { "epoch": 2.606865727233224, "grad_norm": 0.39532803817789686, "learning_rate": 5.140600636132587e-06, "loss": 2.5528, "step": 41994 }, { "epoch": 2.6069278043329813, "grad_norm": 0.4094628421800068, "learning_rate": 5.139005727618479e-06, "loss": 2.6316, "step": 41995 }, { "epoch": 2.6069898814327397, "grad_norm": 0.4304561329288951, "learning_rate": 5.1374110531550616e-06, "loss": 2.7067, "step": 41996 }, { "epoch": 2.607051958532497, "grad_norm": 0.43138411827476314, "learning_rate": 5.135816612750672e-06, "loss": 2.6725, "step": 41997 }, { "epoch": 2.6071140356322555, "grad_norm": 0.40601336483145545, "learning_rate": 5.134222406413619e-06, "loss": 2.6436, "step": 41998 }, { "epoch": 2.607176112732013, "grad_norm": 0.4065069889638481, "learning_rate": 5.132628434152215e-06, "loss": 2.6863, "step": 41999 }, { "epoch": 2.607238189831771, "grad_norm": 0.37846391952615827, "learning_rate": 5.131034695974774e-06, "loss": 2.6263, "step": 42000 }, { "epoch": 2.607300266931529, "grad_norm": 0.43193644747566506, "learning_rate": 5.129441191889628e-06, "loss": 2.6695, "step": 42001 }, { "epoch": 2.6073623440312867, "grad_norm": 0.4172382388174416, "learning_rate": 5.127847921905077e-06, "loss": 2.7167, "step": 42002 }, { "epoch": 2.6074244211310447, "grad_norm": 0.391035855710253, "learning_rate": 5.126254886029436e-06, "loss": 2.6509, "step": 42003 }, { "epoch": 2.6074864982308026, "grad_norm": 0.4028560561905354, "learning_rate": 5.124662084271009e-06, "loss": 2.6049, "step": 42004 }, { "epoch": 2.6075485753305605, "grad_norm": 0.4446258605142223, "learning_rate": 5.123069516638123e-06, "loss": 2.6532, "step": 42005 }, { "epoch": 2.6076106524303184, "grad_norm": 0.4343097548574399, "learning_rate": 5.1214771831390785e-06, "loss": 2.6589, "step": 42006 }, { "epoch": 2.6076727295300763, "grad_norm": 0.39378804371227777, "learning_rate": 5.119885083782183e-06, "loss": 2.6235, "step": 42007 }, { "epoch": 2.6077348066298343, "grad_norm": 0.3911082521036681, "learning_rate": 5.118293218575737e-06, "loss": 2.5947, "step": 42008 }, { "epoch": 2.607796883729592, "grad_norm": 0.3738969036038869, "learning_rate": 5.116701587528044e-06, "loss": 2.7784, "step": 42009 }, { "epoch": 2.60785896082935, "grad_norm": 0.4172658089567892, "learning_rate": 5.1151101906474265e-06, "loss": 2.6269, "step": 42010 }, { "epoch": 2.607921037929108, "grad_norm": 0.41058247536183184, "learning_rate": 5.113519027942177e-06, "loss": 2.6596, "step": 42011 }, { "epoch": 2.607983115028866, "grad_norm": 0.3987131100514405, "learning_rate": 5.111928099420588e-06, "loss": 2.7603, "step": 42012 }, { "epoch": 2.608045192128624, "grad_norm": 0.3921375364416112, "learning_rate": 5.1103374050909595e-06, "loss": 2.5692, "step": 42013 }, { "epoch": 2.6081072692283818, "grad_norm": 0.3837042173604713, "learning_rate": 5.108746944961612e-06, "loss": 2.5926, "step": 42014 }, { "epoch": 2.6081693463281397, "grad_norm": 0.42952001948989643, "learning_rate": 5.107156719040829e-06, "loss": 2.6867, "step": 42015 }, { "epoch": 2.6082314234278976, "grad_norm": 0.39398809379412586, "learning_rate": 5.105566727336902e-06, "loss": 2.6553, "step": 42016 }, { "epoch": 2.6082935005276555, "grad_norm": 0.40351475053602004, "learning_rate": 5.103976969858137e-06, "loss": 2.6967, "step": 42017 }, { "epoch": 2.608355577627413, "grad_norm": 0.39821929192313443, "learning_rate": 5.102387446612811e-06, "loss": 2.7145, "step": 42018 }, { "epoch": 2.6084176547271714, "grad_norm": 0.38535449669286875, "learning_rate": 5.100798157609238e-06, "loss": 2.6663, "step": 42019 }, { "epoch": 2.608479731826929, "grad_norm": 0.4183641328371914, "learning_rate": 5.099209102855701e-06, "loss": 2.6062, "step": 42020 }, { "epoch": 2.608541808926687, "grad_norm": 0.4043025140783705, "learning_rate": 5.097620282360494e-06, "loss": 2.6192, "step": 42021 }, { "epoch": 2.6086038860264447, "grad_norm": 0.38871956537466174, "learning_rate": 5.096031696131892e-06, "loss": 2.6773, "step": 42022 }, { "epoch": 2.6086659631262026, "grad_norm": 0.38118196177364627, "learning_rate": 5.09444334417819e-06, "loss": 2.6847, "step": 42023 }, { "epoch": 2.6087280402259605, "grad_norm": 0.39637565061647545, "learning_rate": 5.092855226507687e-06, "loss": 2.6222, "step": 42024 }, { "epoch": 2.6087901173257184, "grad_norm": 0.4173993840585011, "learning_rate": 5.091267343128664e-06, "loss": 2.7807, "step": 42025 }, { "epoch": 2.6088521944254763, "grad_norm": 0.3973247660348245, "learning_rate": 5.089679694049398e-06, "loss": 2.6748, "step": 42026 }, { "epoch": 2.6089142715252343, "grad_norm": 0.37904591296857904, "learning_rate": 5.088092279278167e-06, "loss": 2.6337, "step": 42027 }, { "epoch": 2.608976348624992, "grad_norm": 0.42778601835448105, "learning_rate": 5.0865050988232745e-06, "loss": 2.682, "step": 42028 }, { "epoch": 2.60903842572475, "grad_norm": 0.413221098158603, "learning_rate": 5.0849181526929805e-06, "loss": 2.6447, "step": 42029 }, { "epoch": 2.609100502824508, "grad_norm": 0.3917516102225216, "learning_rate": 5.083331440895578e-06, "loss": 2.7061, "step": 42030 }, { "epoch": 2.609162579924266, "grad_norm": 0.42555426750674313, "learning_rate": 5.081744963439339e-06, "loss": 2.6711, "step": 42031 }, { "epoch": 2.609224657024024, "grad_norm": 0.39121381251040044, "learning_rate": 5.0801587203325285e-06, "loss": 2.7234, "step": 42032 }, { "epoch": 2.6092867341237818, "grad_norm": 0.45891278297000593, "learning_rate": 5.078572711583446e-06, "loss": 2.7709, "step": 42033 }, { "epoch": 2.6093488112235397, "grad_norm": 0.39157945375508935, "learning_rate": 5.0769869372003565e-06, "loss": 2.6577, "step": 42034 }, { "epoch": 2.6094108883232976, "grad_norm": 0.40101450254252374, "learning_rate": 5.075401397191526e-06, "loss": 2.649, "step": 42035 }, { "epoch": 2.6094729654230555, "grad_norm": 0.36411890239697653, "learning_rate": 5.0738160915652256e-06, "loss": 2.5516, "step": 42036 }, { "epoch": 2.6095350425228134, "grad_norm": 0.417010268951693, "learning_rate": 5.072231020329743e-06, "loss": 2.6857, "step": 42037 }, { "epoch": 2.6095971196225713, "grad_norm": 0.3919918750254208, "learning_rate": 5.070646183493338e-06, "loss": 2.637, "step": 42038 }, { "epoch": 2.6096591967223293, "grad_norm": 0.3919730196506398, "learning_rate": 5.069061581064277e-06, "loss": 2.6594, "step": 42039 }, { "epoch": 2.609721273822087, "grad_norm": 0.39157321285075486, "learning_rate": 5.06747721305082e-06, "loss": 2.7073, "step": 42040 }, { "epoch": 2.6097833509218447, "grad_norm": 0.3743288471514153, "learning_rate": 5.0658930794612535e-06, "loss": 2.663, "step": 42041 }, { "epoch": 2.609845428021603, "grad_norm": 0.37881472553857554, "learning_rate": 5.064309180303833e-06, "loss": 2.5879, "step": 42042 }, { "epoch": 2.6099075051213605, "grad_norm": 0.4137315687240488, "learning_rate": 5.0627255155868136e-06, "loss": 2.6312, "step": 42043 }, { "epoch": 2.609969582221119, "grad_norm": 0.36647179023507553, "learning_rate": 5.061142085318471e-06, "loss": 2.68, "step": 42044 }, { "epoch": 2.6100316593208763, "grad_norm": 0.370944344410021, "learning_rate": 5.059558889507043e-06, "loss": 2.539, "step": 42045 }, { "epoch": 2.6100937364206347, "grad_norm": 0.4002939433160194, "learning_rate": 5.057975928160819e-06, "loss": 2.7261, "step": 42046 }, { "epoch": 2.610155813520392, "grad_norm": 0.38033565801798236, "learning_rate": 5.056393201288045e-06, "loss": 2.7105, "step": 42047 }, { "epoch": 2.61021789062015, "grad_norm": 0.3950081388624906, "learning_rate": 5.054810708896979e-06, "loss": 2.7132, "step": 42048 }, { "epoch": 2.610279967719908, "grad_norm": 0.37203312485249307, "learning_rate": 5.0532284509958675e-06, "loss": 2.7044, "step": 42049 }, { "epoch": 2.610342044819666, "grad_norm": 0.3858070029357317, "learning_rate": 5.0516464275929764e-06, "loss": 2.6889, "step": 42050 }, { "epoch": 2.610404121919424, "grad_norm": 0.42740452907801796, "learning_rate": 5.050064638696567e-06, "loss": 2.6857, "step": 42051 }, { "epoch": 2.6104661990191818, "grad_norm": 0.41796670004083486, "learning_rate": 5.048483084314875e-06, "loss": 2.7018, "step": 42052 }, { "epoch": 2.6105282761189397, "grad_norm": 0.3775759083489221, "learning_rate": 5.046901764456153e-06, "loss": 2.487, "step": 42053 }, { "epoch": 2.6105903532186976, "grad_norm": 0.40940732245934086, "learning_rate": 5.045320679128668e-06, "loss": 2.7209, "step": 42054 }, { "epoch": 2.6106524303184555, "grad_norm": 0.4055461655288866, "learning_rate": 5.04373982834066e-06, "loss": 2.5798, "step": 42055 }, { "epoch": 2.6107145074182134, "grad_norm": 0.4173692700157524, "learning_rate": 5.042159212100361e-06, "loss": 2.6916, "step": 42056 }, { "epoch": 2.6107765845179713, "grad_norm": 0.44813155593758414, "learning_rate": 5.040578830416043e-06, "loss": 2.7, "step": 42057 }, { "epoch": 2.6108386616177293, "grad_norm": 0.38937955676086733, "learning_rate": 5.038998683295937e-06, "loss": 2.6956, "step": 42058 }, { "epoch": 2.610900738717487, "grad_norm": 0.4237855986605582, "learning_rate": 5.037418770748281e-06, "loss": 2.7884, "step": 42059 }, { "epoch": 2.610962815817245, "grad_norm": 0.3993928339383611, "learning_rate": 5.035839092781336e-06, "loss": 2.6547, "step": 42060 }, { "epoch": 2.611024892917003, "grad_norm": 0.40384656365517024, "learning_rate": 5.034259649403333e-06, "loss": 2.637, "step": 42061 }, { "epoch": 2.611086970016761, "grad_norm": 0.4190383309912444, "learning_rate": 5.032680440622517e-06, "loss": 2.7271, "step": 42062 }, { "epoch": 2.611149047116519, "grad_norm": 0.40773699123510226, "learning_rate": 5.031101466447108e-06, "loss": 2.6795, "step": 42063 }, { "epoch": 2.6112111242162768, "grad_norm": 0.4206295349307412, "learning_rate": 5.029522726885372e-06, "loss": 2.5684, "step": 42064 }, { "epoch": 2.6112732013160347, "grad_norm": 0.4275215525241479, "learning_rate": 5.027944221945535e-06, "loss": 2.7711, "step": 42065 }, { "epoch": 2.611335278415792, "grad_norm": 0.44108071602871773, "learning_rate": 5.026365951635825e-06, "loss": 2.6298, "step": 42066 }, { "epoch": 2.6113973555155505, "grad_norm": 0.3893367696825719, "learning_rate": 5.024787915964485e-06, "loss": 2.6101, "step": 42067 }, { "epoch": 2.611459432615308, "grad_norm": 0.43244036640993094, "learning_rate": 5.023210114939736e-06, "loss": 2.626, "step": 42068 }, { "epoch": 2.6115215097150664, "grad_norm": 0.4345849709771018, "learning_rate": 5.021632548569826e-06, "loss": 2.7049, "step": 42069 }, { "epoch": 2.611583586814824, "grad_norm": 0.4137435620432267, "learning_rate": 5.020055216862979e-06, "loss": 2.7162, "step": 42070 }, { "epoch": 2.6116456639145817, "grad_norm": 0.37385266342066353, "learning_rate": 5.018478119827424e-06, "loss": 2.6604, "step": 42071 }, { "epoch": 2.6117077410143397, "grad_norm": 0.37642389413691824, "learning_rate": 5.016901257471379e-06, "loss": 2.6605, "step": 42072 }, { "epoch": 2.6117698181140976, "grad_norm": 0.4066465533114802, "learning_rate": 5.015324629803087e-06, "loss": 2.6245, "step": 42073 }, { "epoch": 2.6118318952138555, "grad_norm": 0.4209805472664024, "learning_rate": 5.013748236830767e-06, "loss": 2.7166, "step": 42074 }, { "epoch": 2.6118939723136134, "grad_norm": 0.36865084818353505, "learning_rate": 5.012172078562644e-06, "loss": 2.6327, "step": 42075 }, { "epoch": 2.6119560494133713, "grad_norm": 0.36649346198252136, "learning_rate": 5.01059615500693e-06, "loss": 2.6961, "step": 42076 }, { "epoch": 2.6120181265131293, "grad_norm": 0.407337402335263, "learning_rate": 5.009020466171871e-06, "loss": 2.7098, "step": 42077 }, { "epoch": 2.612080203612887, "grad_norm": 0.3903661633290522, "learning_rate": 5.00744501206567e-06, "loss": 2.7058, "step": 42078 }, { "epoch": 2.612142280712645, "grad_norm": 0.38100504901374, "learning_rate": 5.005869792696549e-06, "loss": 2.656, "step": 42079 }, { "epoch": 2.612204357812403, "grad_norm": 0.40255013514971216, "learning_rate": 5.004294808072735e-06, "loss": 2.5761, "step": 42080 }, { "epoch": 2.612266434912161, "grad_norm": 0.391379295684964, "learning_rate": 5.002720058202426e-06, "loss": 2.726, "step": 42081 }, { "epoch": 2.612328512011919, "grad_norm": 0.39442291283082753, "learning_rate": 5.001145543093855e-06, "loss": 2.6989, "step": 42082 }, { "epoch": 2.6123905891116768, "grad_norm": 0.40092586934561, "learning_rate": 4.999571262755237e-06, "loss": 2.6869, "step": 42083 }, { "epoch": 2.6124526662114347, "grad_norm": 0.4083771403162892, "learning_rate": 4.9979972171947776e-06, "loss": 2.7252, "step": 42084 }, { "epoch": 2.6125147433111926, "grad_norm": 0.4210798222321003, "learning_rate": 4.99642340642068e-06, "loss": 2.7135, "step": 42085 }, { "epoch": 2.6125768204109505, "grad_norm": 0.39901315953129773, "learning_rate": 4.9948498304411785e-06, "loss": 2.6283, "step": 42086 }, { "epoch": 2.6126388975107084, "grad_norm": 0.38740205603690553, "learning_rate": 4.99327648926447e-06, "loss": 2.6655, "step": 42087 }, { "epoch": 2.6127009746104664, "grad_norm": 0.4158516860815819, "learning_rate": 4.991703382898755e-06, "loss": 2.7201, "step": 42088 }, { "epoch": 2.612763051710224, "grad_norm": 0.4114669565512422, "learning_rate": 4.990130511352253e-06, "loss": 2.6634, "step": 42089 }, { "epoch": 2.612825128809982, "grad_norm": 0.39926347948536234, "learning_rate": 4.988557874633165e-06, "loss": 2.6468, "step": 42090 }, { "epoch": 2.6128872059097397, "grad_norm": 0.39549730075006706, "learning_rate": 4.986985472749706e-06, "loss": 2.6487, "step": 42091 }, { "epoch": 2.612949283009498, "grad_norm": 0.3967421121468204, "learning_rate": 4.985413305710069e-06, "loss": 2.6717, "step": 42092 }, { "epoch": 2.6130113601092555, "grad_norm": 0.4278776721341552, "learning_rate": 4.983841373522457e-06, "loss": 2.7387, "step": 42093 }, { "epoch": 2.613073437209014, "grad_norm": 0.3855943034179139, "learning_rate": 4.982269676195078e-06, "loss": 2.5835, "step": 42094 }, { "epoch": 2.6131355143087713, "grad_norm": 0.39572194272153927, "learning_rate": 4.980698213736112e-06, "loss": 2.7262, "step": 42095 }, { "epoch": 2.6131975914085293, "grad_norm": 0.386775754329993, "learning_rate": 4.979126986153781e-06, "loss": 2.6753, "step": 42096 }, { "epoch": 2.613259668508287, "grad_norm": 0.42656747165339765, "learning_rate": 4.977555993456278e-06, "loss": 2.6914, "step": 42097 }, { "epoch": 2.613321745608045, "grad_norm": 0.40454501217402933, "learning_rate": 4.975985235651792e-06, "loss": 2.6459, "step": 42098 }, { "epoch": 2.613383822707803, "grad_norm": 0.4063773689142303, "learning_rate": 4.974414712748516e-06, "loss": 2.5588, "step": 42099 }, { "epoch": 2.613445899807561, "grad_norm": 0.41647961312002724, "learning_rate": 4.972844424754652e-06, "loss": 2.6426, "step": 42100 }, { "epoch": 2.613507976907319, "grad_norm": 0.4211876355822099, "learning_rate": 4.971274371678392e-06, "loss": 2.6338, "step": 42101 }, { "epoch": 2.6135700540070768, "grad_norm": 0.4265362406602253, "learning_rate": 4.969704553527926e-06, "loss": 2.6632, "step": 42102 }, { "epoch": 2.6136321311068347, "grad_norm": 0.4238944990509827, "learning_rate": 4.968134970311433e-06, "loss": 2.594, "step": 42103 }, { "epoch": 2.6136942082065926, "grad_norm": 0.4794234403284987, "learning_rate": 4.966565622037123e-06, "loss": 2.7402, "step": 42104 }, { "epoch": 2.6137562853063505, "grad_norm": 0.4415378036936006, "learning_rate": 4.964996508713166e-06, "loss": 2.7531, "step": 42105 }, { "epoch": 2.6138183624061084, "grad_norm": 0.43556312564931365, "learning_rate": 4.9634276303477615e-06, "loss": 2.6984, "step": 42106 }, { "epoch": 2.6138804395058663, "grad_norm": 0.43265318021170873, "learning_rate": 4.961858986949086e-06, "loss": 2.6318, "step": 42107 }, { "epoch": 2.6139425166056243, "grad_norm": 0.4132212063415538, "learning_rate": 4.960290578525312e-06, "loss": 2.6752, "step": 42108 }, { "epoch": 2.614004593705382, "grad_norm": 0.41808559887820745, "learning_rate": 4.958722405084648e-06, "loss": 2.7506, "step": 42109 }, { "epoch": 2.61406667080514, "grad_norm": 0.4046169433062693, "learning_rate": 4.957154466635267e-06, "loss": 2.6978, "step": 42110 }, { "epoch": 2.614128747904898, "grad_norm": 0.41127377809465104, "learning_rate": 4.9555867631853435e-06, "loss": 2.7497, "step": 42111 }, { "epoch": 2.614190825004656, "grad_norm": 0.39908095368664703, "learning_rate": 4.9540192947430455e-06, "loss": 2.6452, "step": 42112 }, { "epoch": 2.614252902104414, "grad_norm": 0.43824695786194423, "learning_rate": 4.952452061316576e-06, "loss": 2.7078, "step": 42113 }, { "epoch": 2.6143149792041713, "grad_norm": 0.4152235630813257, "learning_rate": 4.9508850629141016e-06, "loss": 2.6192, "step": 42114 }, { "epoch": 2.6143770563039297, "grad_norm": 0.40631697176057013, "learning_rate": 4.949318299543792e-06, "loss": 2.6352, "step": 42115 }, { "epoch": 2.614439133403687, "grad_norm": 0.42727934789670147, "learning_rate": 4.9477517712138326e-06, "loss": 2.6305, "step": 42116 }, { "epoch": 2.6145012105034455, "grad_norm": 0.3789506473289605, "learning_rate": 4.946185477932375e-06, "loss": 2.6254, "step": 42117 }, { "epoch": 2.614563287603203, "grad_norm": 0.3737032642934445, "learning_rate": 4.944619419707613e-06, "loss": 2.5927, "step": 42118 }, { "epoch": 2.614625364702961, "grad_norm": 0.391897836229698, "learning_rate": 4.943053596547708e-06, "loss": 2.5912, "step": 42119 }, { "epoch": 2.614687441802719, "grad_norm": 0.38773024855629085, "learning_rate": 4.941488008460837e-06, "loss": 2.6514, "step": 42120 }, { "epoch": 2.6147495189024768, "grad_norm": 0.38448858211167497, "learning_rate": 4.939922655455148e-06, "loss": 2.599, "step": 42121 }, { "epoch": 2.6148115960022347, "grad_norm": 0.39819592276846766, "learning_rate": 4.9383575375388184e-06, "loss": 2.6733, "step": 42122 }, { "epoch": 2.6148736731019926, "grad_norm": 0.4028667273173607, "learning_rate": 4.936792654720029e-06, "loss": 2.6538, "step": 42123 }, { "epoch": 2.6149357502017505, "grad_norm": 0.3786878841440381, "learning_rate": 4.9352280070069315e-06, "loss": 2.6858, "step": 42124 }, { "epoch": 2.6149978273015084, "grad_norm": 0.39611765621958006, "learning_rate": 4.933663594407689e-06, "loss": 2.701, "step": 42125 }, { "epoch": 2.6150599044012663, "grad_norm": 0.4329623875172868, "learning_rate": 4.932099416930452e-06, "loss": 2.6979, "step": 42126 }, { "epoch": 2.6151219815010243, "grad_norm": 0.43193668896512444, "learning_rate": 4.9305354745834085e-06, "loss": 2.7397, "step": 42127 }, { "epoch": 2.615184058600782, "grad_norm": 0.37454215551484416, "learning_rate": 4.928971767374696e-06, "loss": 2.7015, "step": 42128 }, { "epoch": 2.61524613570054, "grad_norm": 0.39581804706102497, "learning_rate": 4.9274082953124854e-06, "loss": 2.6997, "step": 42129 }, { "epoch": 2.615308212800298, "grad_norm": 0.4021911702140508, "learning_rate": 4.925845058404921e-06, "loss": 2.6506, "step": 42130 }, { "epoch": 2.615370289900056, "grad_norm": 0.39087825195062553, "learning_rate": 4.924282056660162e-06, "loss": 2.687, "step": 42131 }, { "epoch": 2.615432366999814, "grad_norm": 0.43139890228612865, "learning_rate": 4.922719290086369e-06, "loss": 2.7039, "step": 42132 }, { "epoch": 2.6154944440995718, "grad_norm": 0.4201643842347396, "learning_rate": 4.921156758691698e-06, "loss": 2.6635, "step": 42133 }, { "epoch": 2.6155565211993297, "grad_norm": 0.3817963250337546, "learning_rate": 4.919594462484289e-06, "loss": 2.6402, "step": 42134 }, { "epoch": 2.6156185982990876, "grad_norm": 0.4290974119781773, "learning_rate": 4.918032401472295e-06, "loss": 2.7533, "step": 42135 }, { "epoch": 2.6156806753988455, "grad_norm": 0.42690059970483857, "learning_rate": 4.916470575663878e-06, "loss": 2.6424, "step": 42136 }, { "epoch": 2.615742752498603, "grad_norm": 0.41382905389250496, "learning_rate": 4.914908985067179e-06, "loss": 2.7124, "step": 42137 }, { "epoch": 2.6158048295983614, "grad_norm": 0.38909143776880994, "learning_rate": 4.913347629690346e-06, "loss": 2.704, "step": 42138 }, { "epoch": 2.615866906698119, "grad_norm": 0.4149979873976106, "learning_rate": 4.911786509541511e-06, "loss": 2.617, "step": 42139 }, { "epoch": 2.615928983797877, "grad_norm": 0.4153430139886553, "learning_rate": 4.910225624628845e-06, "loss": 2.6682, "step": 42140 }, { "epoch": 2.6159910608976347, "grad_norm": 0.39936986925891854, "learning_rate": 4.9086649749604755e-06, "loss": 2.6571, "step": 42141 }, { "epoch": 2.616053137997393, "grad_norm": 0.41237792535133355, "learning_rate": 4.907104560544545e-06, "loss": 2.7339, "step": 42142 }, { "epoch": 2.6161152150971505, "grad_norm": 0.40455499416550605, "learning_rate": 4.905544381389199e-06, "loss": 2.5459, "step": 42143 }, { "epoch": 2.6161772921969084, "grad_norm": 0.41269577033563176, "learning_rate": 4.903984437502563e-06, "loss": 2.6039, "step": 42144 }, { "epoch": 2.6162393692966663, "grad_norm": 0.40919334370624, "learning_rate": 4.902424728892801e-06, "loss": 2.6055, "step": 42145 }, { "epoch": 2.6163014463964243, "grad_norm": 0.419529434285417, "learning_rate": 4.900865255568032e-06, "loss": 2.6555, "step": 42146 }, { "epoch": 2.616363523496182, "grad_norm": 0.37523318034578945, "learning_rate": 4.899306017536404e-06, "loss": 2.6714, "step": 42147 }, { "epoch": 2.61642560059594, "grad_norm": 0.4179588031476503, "learning_rate": 4.897747014806031e-06, "loss": 2.651, "step": 42148 }, { "epoch": 2.616487677695698, "grad_norm": 0.44367908932045913, "learning_rate": 4.89618824738507e-06, "loss": 2.6612, "step": 42149 }, { "epoch": 2.616549754795456, "grad_norm": 0.4344736570388335, "learning_rate": 4.894629715281645e-06, "loss": 2.7171, "step": 42150 }, { "epoch": 2.616611831895214, "grad_norm": 0.37126161894293175, "learning_rate": 4.8930714185038864e-06, "loss": 2.7805, "step": 42151 }, { "epoch": 2.6166739089949718, "grad_norm": 0.37942626506199656, "learning_rate": 4.891513357059924e-06, "loss": 2.696, "step": 42152 }, { "epoch": 2.6167359860947297, "grad_norm": 0.3884444026379747, "learning_rate": 4.889955530957879e-06, "loss": 2.6543, "step": 42153 }, { "epoch": 2.6167980631944876, "grad_norm": 0.42598665165136274, "learning_rate": 4.888397940205885e-06, "loss": 2.6576, "step": 42154 }, { "epoch": 2.6168601402942455, "grad_norm": 0.38243668462270813, "learning_rate": 4.88684058481208e-06, "loss": 2.7002, "step": 42155 }, { "epoch": 2.6169222173940034, "grad_norm": 0.41117202680506326, "learning_rate": 4.885283464784579e-06, "loss": 2.6161, "step": 42156 }, { "epoch": 2.6169842944937614, "grad_norm": 0.39622139313020277, "learning_rate": 4.883726580131509e-06, "loss": 2.6931, "step": 42157 }, { "epoch": 2.6170463715935193, "grad_norm": 0.3843967392829296, "learning_rate": 4.882169930860981e-06, "loss": 2.6535, "step": 42158 }, { "epoch": 2.617108448693277, "grad_norm": 0.41319028289508636, "learning_rate": 4.880613516981131e-06, "loss": 2.6535, "step": 42159 }, { "epoch": 2.617170525793035, "grad_norm": 0.4178929304714367, "learning_rate": 4.879057338500076e-06, "loss": 2.6922, "step": 42160 }, { "epoch": 2.617232602892793, "grad_norm": 0.4162475405153579, "learning_rate": 4.877501395425927e-06, "loss": 2.6492, "step": 42161 }, { "epoch": 2.6172946799925505, "grad_norm": 0.40153188180528626, "learning_rate": 4.875945687766803e-06, "loss": 2.6555, "step": 42162 }, { "epoch": 2.617356757092309, "grad_norm": 0.3585174113611708, "learning_rate": 4.874390215530833e-06, "loss": 2.6729, "step": 42163 }, { "epoch": 2.6174188341920663, "grad_norm": 0.3973798739460476, "learning_rate": 4.8728349787261205e-06, "loss": 2.6882, "step": 42164 }, { "epoch": 2.6174809112918247, "grad_norm": 0.4243319983739798, "learning_rate": 4.871279977360788e-06, "loss": 2.596, "step": 42165 }, { "epoch": 2.617542988391582, "grad_norm": 0.39804480405199505, "learning_rate": 4.869725211442938e-06, "loss": 2.6757, "step": 42166 }, { "epoch": 2.61760506549134, "grad_norm": 0.3763507912691469, "learning_rate": 4.868170680980683e-06, "loss": 2.7335, "step": 42167 }, { "epoch": 2.617667142591098, "grad_norm": 0.41151464857538966, "learning_rate": 4.866616385982148e-06, "loss": 2.7403, "step": 42168 }, { "epoch": 2.617729219690856, "grad_norm": 0.4028327710693964, "learning_rate": 4.865062326455428e-06, "loss": 2.6175, "step": 42169 }, { "epoch": 2.617791296790614, "grad_norm": 0.40531211576539167, "learning_rate": 4.863508502408631e-06, "loss": 2.74, "step": 42170 }, { "epoch": 2.6178533738903718, "grad_norm": 0.4374980074973374, "learning_rate": 4.861954913849864e-06, "loss": 2.6941, "step": 42171 }, { "epoch": 2.6179154509901297, "grad_norm": 0.3957649807724255, "learning_rate": 4.8604015607872475e-06, "loss": 2.6751, "step": 42172 }, { "epoch": 2.6179775280898876, "grad_norm": 0.42139432155429546, "learning_rate": 4.85884844322887e-06, "loss": 2.6619, "step": 42173 }, { "epoch": 2.6180396051896455, "grad_norm": 0.38190717105865224, "learning_rate": 4.857295561182834e-06, "loss": 2.666, "step": 42174 }, { "epoch": 2.6181016822894034, "grad_norm": 0.42195548949826633, "learning_rate": 4.855742914657246e-06, "loss": 2.6991, "step": 42175 }, { "epoch": 2.6181637593891613, "grad_norm": 0.43553447278074336, "learning_rate": 4.8541905036602105e-06, "loss": 2.5963, "step": 42176 }, { "epoch": 2.6182258364889193, "grad_norm": 0.4287544871392959, "learning_rate": 4.852638328199821e-06, "loss": 2.5994, "step": 42177 }, { "epoch": 2.618287913588677, "grad_norm": 0.39296741068253455, "learning_rate": 4.8510863882841814e-06, "loss": 2.6373, "step": 42178 }, { "epoch": 2.618349990688435, "grad_norm": 0.37249217309985555, "learning_rate": 4.849534683921381e-06, "loss": 2.6676, "step": 42179 }, { "epoch": 2.618412067788193, "grad_norm": 0.40035742645331407, "learning_rate": 4.847983215119512e-06, "loss": 2.6701, "step": 42180 }, { "epoch": 2.618474144887951, "grad_norm": 0.39440732841564907, "learning_rate": 4.846431981886684e-06, "loss": 2.6229, "step": 42181 }, { "epoch": 2.618536221987709, "grad_norm": 0.40691407653275385, "learning_rate": 4.844880984230982e-06, "loss": 2.5899, "step": 42182 }, { "epoch": 2.6185982990874668, "grad_norm": 0.39493570275251677, "learning_rate": 4.843330222160497e-06, "loss": 2.8336, "step": 42183 }, { "epoch": 2.6186603761872247, "grad_norm": 0.38388132243413886, "learning_rate": 4.841779695683313e-06, "loss": 2.6595, "step": 42184 }, { "epoch": 2.618722453286982, "grad_norm": 0.40056717864500613, "learning_rate": 4.8402294048075335e-06, "loss": 2.6118, "step": 42185 }, { "epoch": 2.6187845303867405, "grad_norm": 0.420509991106413, "learning_rate": 4.838679349541242e-06, "loss": 2.5553, "step": 42186 }, { "epoch": 2.618846607486498, "grad_norm": 0.39462654926792223, "learning_rate": 4.837129529892515e-06, "loss": 2.7085, "step": 42187 }, { "epoch": 2.6189086845862564, "grad_norm": 0.39728934218165585, "learning_rate": 4.835579945869451e-06, "loss": 2.5437, "step": 42188 }, { "epoch": 2.618970761686014, "grad_norm": 0.3749897279921982, "learning_rate": 4.834030597480127e-06, "loss": 2.6685, "step": 42189 }, { "epoch": 2.619032838785772, "grad_norm": 0.3823551248308976, "learning_rate": 4.832481484732637e-06, "loss": 2.675, "step": 42190 }, { "epoch": 2.6190949158855297, "grad_norm": 0.4343997920803324, "learning_rate": 4.830932607635053e-06, "loss": 2.72, "step": 42191 }, { "epoch": 2.6191569929852876, "grad_norm": 0.4054235157513858, "learning_rate": 4.829383966195461e-06, "loss": 2.6735, "step": 42192 }, { "epoch": 2.6192190700850455, "grad_norm": 0.3961969658713301, "learning_rate": 4.827835560421939e-06, "loss": 2.6484, "step": 42193 }, { "epoch": 2.6192811471848034, "grad_norm": 0.4009269843502739, "learning_rate": 4.8262873903225595e-06, "loss": 2.6403, "step": 42194 }, { "epoch": 2.6193432242845613, "grad_norm": 0.43478151825322364, "learning_rate": 4.824739455905408e-06, "loss": 2.68, "step": 42195 }, { "epoch": 2.6194053013843193, "grad_norm": 0.4023694150572048, "learning_rate": 4.823191757178563e-06, "loss": 2.6283, "step": 42196 }, { "epoch": 2.619467378484077, "grad_norm": 0.42387268605637, "learning_rate": 4.8216442941500905e-06, "loss": 2.6172, "step": 42197 }, { "epoch": 2.619529455583835, "grad_norm": 0.4080470211022218, "learning_rate": 4.820097066828055e-06, "loss": 2.6496, "step": 42198 }, { "epoch": 2.619591532683593, "grad_norm": 0.43128127247301923, "learning_rate": 4.8185500752205515e-06, "loss": 2.696, "step": 42199 }, { "epoch": 2.619653609783351, "grad_norm": 0.42930636408474077, "learning_rate": 4.817003319335645e-06, "loss": 2.6622, "step": 42200 }, { "epoch": 2.619715686883109, "grad_norm": 0.4203386943837265, "learning_rate": 4.815456799181395e-06, "loss": 2.6146, "step": 42201 }, { "epoch": 2.6197777639828668, "grad_norm": 0.4170224717017762, "learning_rate": 4.813910514765879e-06, "loss": 2.6078, "step": 42202 }, { "epoch": 2.6198398410826247, "grad_norm": 0.4003560307142023, "learning_rate": 4.8123644660971525e-06, "loss": 2.5973, "step": 42203 }, { "epoch": 2.6199019181823826, "grad_norm": 0.42089500337337077, "learning_rate": 4.810818653183297e-06, "loss": 2.6877, "step": 42204 }, { "epoch": 2.6199639952821405, "grad_norm": 0.42269840065554504, "learning_rate": 4.809273076032372e-06, "loss": 2.7339, "step": 42205 }, { "epoch": 2.6200260723818984, "grad_norm": 0.3773206786690562, "learning_rate": 4.80772773465244e-06, "loss": 2.6747, "step": 42206 }, { "epoch": 2.6200881494816564, "grad_norm": 0.3852346630633021, "learning_rate": 4.806182629051553e-06, "loss": 2.7239, "step": 42207 }, { "epoch": 2.6201502265814143, "grad_norm": 0.4374027484705407, "learning_rate": 4.804637759237795e-06, "loss": 2.691, "step": 42208 }, { "epoch": 2.620212303681172, "grad_norm": 0.3684368185630754, "learning_rate": 4.80309312521921e-06, "loss": 2.6046, "step": 42209 }, { "epoch": 2.6202743807809297, "grad_norm": 0.38675963060487045, "learning_rate": 4.801548727003862e-06, "loss": 2.7049, "step": 42210 }, { "epoch": 2.620336457880688, "grad_norm": 0.38236925193628846, "learning_rate": 4.800004564599797e-06, "loss": 2.6264, "step": 42211 }, { "epoch": 2.6203985349804455, "grad_norm": 0.42726375870289207, "learning_rate": 4.7984606380150895e-06, "loss": 2.7034, "step": 42212 }, { "epoch": 2.620460612080204, "grad_norm": 0.39360057478102917, "learning_rate": 4.796916947257785e-06, "loss": 2.6821, "step": 42213 }, { "epoch": 2.6205226891799613, "grad_norm": 0.4153034041580496, "learning_rate": 4.7953734923359365e-06, "loss": 2.5855, "step": 42214 }, { "epoch": 2.6205847662797193, "grad_norm": 0.38138870233272043, "learning_rate": 4.7938302732576045e-06, "loss": 2.6439, "step": 42215 }, { "epoch": 2.620646843379477, "grad_norm": 0.41451952561449895, "learning_rate": 4.792287290030817e-06, "loss": 2.6648, "step": 42216 }, { "epoch": 2.620708920479235, "grad_norm": 0.43590554834673506, "learning_rate": 4.790744542663655e-06, "loss": 2.6801, "step": 42217 }, { "epoch": 2.620770997578993, "grad_norm": 0.42284426781967516, "learning_rate": 4.7892020311641585e-06, "loss": 2.6774, "step": 42218 }, { "epoch": 2.620833074678751, "grad_norm": 0.40493569146088193, "learning_rate": 4.7876597555403525e-06, "loss": 2.6128, "step": 42219 }, { "epoch": 2.620895151778509, "grad_norm": 0.4054251329486114, "learning_rate": 4.786117715800315e-06, "loss": 2.6037, "step": 42220 }, { "epoch": 2.6209572288782668, "grad_norm": 0.42335746050899997, "learning_rate": 4.784575911952066e-06, "loss": 2.6158, "step": 42221 }, { "epoch": 2.6210193059780247, "grad_norm": 0.39525392648748026, "learning_rate": 4.783034344003673e-06, "loss": 2.6906, "step": 42222 }, { "epoch": 2.6210813830777826, "grad_norm": 0.444416231718601, "learning_rate": 4.781493011963162e-06, "loss": 2.7204, "step": 42223 }, { "epoch": 2.6211434601775405, "grad_norm": 0.4019424325838746, "learning_rate": 4.779951915838582e-06, "loss": 2.7045, "step": 42224 }, { "epoch": 2.6212055372772984, "grad_norm": 0.39067887888312136, "learning_rate": 4.7784110556379605e-06, "loss": 2.6059, "step": 42225 }, { "epoch": 2.6212676143770564, "grad_norm": 0.40812558254416514, "learning_rate": 4.776870431369357e-06, "loss": 2.7085, "step": 42226 }, { "epoch": 2.6213296914768143, "grad_norm": 0.4118475240303927, "learning_rate": 4.775330043040799e-06, "loss": 2.6724, "step": 42227 }, { "epoch": 2.621391768576572, "grad_norm": 0.40591180002029237, "learning_rate": 4.773789890660324e-06, "loss": 2.6451, "step": 42228 }, { "epoch": 2.62145384567633, "grad_norm": 0.3886156089349882, "learning_rate": 4.772249974235965e-06, "loss": 2.6612, "step": 42229 }, { "epoch": 2.621515922776088, "grad_norm": 0.4202136777967571, "learning_rate": 4.770710293775743e-06, "loss": 2.6315, "step": 42230 }, { "epoch": 2.621577999875846, "grad_norm": 0.4143592203165071, "learning_rate": 4.769170849287719e-06, "loss": 2.6905, "step": 42231 }, { "epoch": 2.621640076975604, "grad_norm": 0.4163322139574268, "learning_rate": 4.767631640779913e-06, "loss": 2.618, "step": 42232 }, { "epoch": 2.6217021540753613, "grad_norm": 0.39602815284261433, "learning_rate": 4.766092668260353e-06, "loss": 2.6885, "step": 42233 }, { "epoch": 2.6217642311751197, "grad_norm": 0.4122838740152906, "learning_rate": 4.764553931737053e-06, "loss": 2.6069, "step": 42234 }, { "epoch": 2.621826308274877, "grad_norm": 0.397799729429175, "learning_rate": 4.763015431218071e-06, "loss": 2.6148, "step": 42235 }, { "epoch": 2.6218883853746355, "grad_norm": 0.41433852370346835, "learning_rate": 4.761477166711414e-06, "loss": 2.652, "step": 42236 }, { "epoch": 2.621950462474393, "grad_norm": 0.4117546724576639, "learning_rate": 4.759939138225117e-06, "loss": 2.625, "step": 42237 }, { "epoch": 2.6220125395741514, "grad_norm": 0.40606202765248395, "learning_rate": 4.758401345767199e-06, "loss": 2.614, "step": 42238 }, { "epoch": 2.622074616673909, "grad_norm": 0.3985773103046103, "learning_rate": 4.756863789345678e-06, "loss": 2.6312, "step": 42239 }, { "epoch": 2.6221366937736668, "grad_norm": 0.4082689916593155, "learning_rate": 4.7553264689685905e-06, "loss": 2.6514, "step": 42240 }, { "epoch": 2.6221987708734247, "grad_norm": 0.39617958944486875, "learning_rate": 4.753789384643942e-06, "loss": 2.6664, "step": 42241 }, { "epoch": 2.6222608479731826, "grad_norm": 0.39039990615423437, "learning_rate": 4.752252536379765e-06, "loss": 2.6752, "step": 42242 }, { "epoch": 2.6223229250729405, "grad_norm": 0.43932396580138, "learning_rate": 4.750715924184057e-06, "loss": 2.6639, "step": 42243 }, { "epoch": 2.6223850021726984, "grad_norm": 0.3873040749934091, "learning_rate": 4.749179548064858e-06, "loss": 2.6825, "step": 42244 }, { "epoch": 2.6224470792724563, "grad_norm": 0.3935911289504894, "learning_rate": 4.747643408030178e-06, "loss": 2.6561, "step": 42245 }, { "epoch": 2.6225091563722143, "grad_norm": 0.3973709866950163, "learning_rate": 4.7461075040880265e-06, "loss": 2.7458, "step": 42246 }, { "epoch": 2.622571233471972, "grad_norm": 0.428784427122555, "learning_rate": 4.744571836246409e-06, "loss": 2.7116, "step": 42247 }, { "epoch": 2.62263331057173, "grad_norm": 0.38549619146363184, "learning_rate": 4.743036404513352e-06, "loss": 2.6286, "step": 42248 }, { "epoch": 2.622695387671488, "grad_norm": 0.3987359443865965, "learning_rate": 4.741501208896864e-06, "loss": 2.7309, "step": 42249 }, { "epoch": 2.622757464771246, "grad_norm": 0.40262060917304976, "learning_rate": 4.739966249404948e-06, "loss": 2.6018, "step": 42250 }, { "epoch": 2.622819541871004, "grad_norm": 0.4205613700612286, "learning_rate": 4.7384315260456166e-06, "loss": 2.6925, "step": 42251 }, { "epoch": 2.6228816189707618, "grad_norm": 0.396850492066132, "learning_rate": 4.736897038826871e-06, "loss": 2.5912, "step": 42252 }, { "epoch": 2.6229436960705197, "grad_norm": 0.3807678308262984, "learning_rate": 4.735362787756714e-06, "loss": 2.6212, "step": 42253 }, { "epoch": 2.6230057731702776, "grad_norm": 0.3967658486040214, "learning_rate": 4.733828772843174e-06, "loss": 2.6788, "step": 42254 }, { "epoch": 2.6230678502700355, "grad_norm": 0.42521325118669157, "learning_rate": 4.732294994094233e-06, "loss": 2.617, "step": 42255 }, { "epoch": 2.6231299273697934, "grad_norm": 0.39397207612722723, "learning_rate": 4.730761451517901e-06, "loss": 2.6798, "step": 42256 }, { "epoch": 2.6231920044695514, "grad_norm": 0.4423606479641685, "learning_rate": 4.729228145122166e-06, "loss": 2.7406, "step": 42257 }, { "epoch": 2.623254081569309, "grad_norm": 0.40711649772637, "learning_rate": 4.727695074915045e-06, "loss": 2.6925, "step": 42258 }, { "epoch": 2.623316158669067, "grad_norm": 0.4027008767908714, "learning_rate": 4.72616224090453e-06, "loss": 2.7442, "step": 42259 }, { "epoch": 2.6233782357688247, "grad_norm": 0.39002907118896674, "learning_rate": 4.7246296430986155e-06, "loss": 2.6042, "step": 42260 }, { "epoch": 2.623440312868583, "grad_norm": 0.43115716501652995, "learning_rate": 4.723097281505295e-06, "loss": 2.6198, "step": 42261 }, { "epoch": 2.6235023899683405, "grad_norm": 0.404262356966178, "learning_rate": 4.721565156132573e-06, "loss": 2.6492, "step": 42262 }, { "epoch": 2.6235644670680984, "grad_norm": 0.3950078559347785, "learning_rate": 4.720033266988438e-06, "loss": 2.6877, "step": 42263 }, { "epoch": 2.6236265441678563, "grad_norm": 0.3898299709782307, "learning_rate": 4.718501614080883e-06, "loss": 2.6462, "step": 42264 }, { "epoch": 2.6236886212676143, "grad_norm": 0.3804123627160489, "learning_rate": 4.7169701974178914e-06, "loss": 2.6352, "step": 42265 }, { "epoch": 2.623750698367372, "grad_norm": 0.41304468620792545, "learning_rate": 4.715439017007456e-06, "loss": 2.6234, "step": 42266 }, { "epoch": 2.62381277546713, "grad_norm": 0.41763019219316183, "learning_rate": 4.7139080728575755e-06, "loss": 2.669, "step": 42267 }, { "epoch": 2.623874852566888, "grad_norm": 0.4167752025228716, "learning_rate": 4.712377364976234e-06, "loss": 2.6416, "step": 42268 }, { "epoch": 2.623936929666646, "grad_norm": 0.39222300775286895, "learning_rate": 4.710846893371407e-06, "loss": 2.6485, "step": 42269 }, { "epoch": 2.623999006766404, "grad_norm": 0.3897761085554679, "learning_rate": 4.709316658051083e-06, "loss": 2.6333, "step": 42270 }, { "epoch": 2.6240610838661618, "grad_norm": 0.3677673428476662, "learning_rate": 4.707786659023255e-06, "loss": 2.5776, "step": 42271 }, { "epoch": 2.6241231609659197, "grad_norm": 0.38606356982564144, "learning_rate": 4.706256896295902e-06, "loss": 2.7042, "step": 42272 }, { "epoch": 2.6241852380656776, "grad_norm": 0.3790534215407634, "learning_rate": 4.704727369876999e-06, "loss": 2.5787, "step": 42273 }, { "epoch": 2.6242473151654355, "grad_norm": 0.37111729998902354, "learning_rate": 4.703198079774518e-06, "loss": 2.6593, "step": 42274 }, { "epoch": 2.6243093922651934, "grad_norm": 0.3796915980302352, "learning_rate": 4.701669025996463e-06, "loss": 2.6603, "step": 42275 }, { "epoch": 2.6243714693649514, "grad_norm": 0.39520930575484253, "learning_rate": 4.70014020855079e-06, "loss": 2.7192, "step": 42276 }, { "epoch": 2.6244335464647093, "grad_norm": 0.3929042884633883, "learning_rate": 4.698611627445493e-06, "loss": 2.7073, "step": 42277 }, { "epoch": 2.624495623564467, "grad_norm": 0.3841969698037807, "learning_rate": 4.697083282688531e-06, "loss": 2.5927, "step": 42278 }, { "epoch": 2.624557700664225, "grad_norm": 0.4002698390482882, "learning_rate": 4.695555174287869e-06, "loss": 2.6648, "step": 42279 }, { "epoch": 2.624619777763983, "grad_norm": 0.3973902984007039, "learning_rate": 4.694027302251508e-06, "loss": 2.6429, "step": 42280 }, { "epoch": 2.6246818548637405, "grad_norm": 0.4309981211789507, "learning_rate": 4.692499666587408e-06, "loss": 2.6459, "step": 42281 }, { "epoch": 2.624743931963499, "grad_norm": 0.36665536996749964, "learning_rate": 4.690972267303528e-06, "loss": 2.6967, "step": 42282 }, { "epoch": 2.6248060090632563, "grad_norm": 0.38771304995609135, "learning_rate": 4.689445104407841e-06, "loss": 2.643, "step": 42283 }, { "epoch": 2.6248680861630147, "grad_norm": 0.40959594238886615, "learning_rate": 4.687918177908329e-06, "loss": 2.5604, "step": 42284 }, { "epoch": 2.624930163262772, "grad_norm": 0.3979618000311809, "learning_rate": 4.6863914878129355e-06, "loss": 2.7044, "step": 42285 }, { "epoch": 2.6249922403625305, "grad_norm": 0.4070463811071802, "learning_rate": 4.684865034129648e-06, "loss": 2.6565, "step": 42286 }, { "epoch": 2.625054317462288, "grad_norm": 0.3893297082144895, "learning_rate": 4.683338816866417e-06, "loss": 2.6154, "step": 42287 }, { "epoch": 2.625116394562046, "grad_norm": 0.41422015013974356, "learning_rate": 4.681812836031213e-06, "loss": 2.644, "step": 42288 }, { "epoch": 2.625178471661804, "grad_norm": 0.3810799962275879, "learning_rate": 4.6802870916319854e-06, "loss": 2.6286, "step": 42289 }, { "epoch": 2.6252405487615618, "grad_norm": 0.397517657157706, "learning_rate": 4.678761583676705e-06, "loss": 2.6987, "step": 42290 }, { "epoch": 2.6253026258613197, "grad_norm": 0.3786538846901553, "learning_rate": 4.677236312173333e-06, "loss": 2.6161, "step": 42291 }, { "epoch": 2.6253647029610776, "grad_norm": 0.4078844277393882, "learning_rate": 4.675711277129818e-06, "loss": 2.5957, "step": 42292 }, { "epoch": 2.6254267800608355, "grad_norm": 0.4146265650338143, "learning_rate": 4.67418647855411e-06, "loss": 2.6868, "step": 42293 }, { "epoch": 2.6254888571605934, "grad_norm": 0.3981132871201355, "learning_rate": 4.6726619164541855e-06, "loss": 2.6581, "step": 42294 }, { "epoch": 2.6255509342603514, "grad_norm": 0.4147612775300646, "learning_rate": 4.671137590837987e-06, "loss": 2.671, "step": 42295 }, { "epoch": 2.6256130113601093, "grad_norm": 0.4318380982937145, "learning_rate": 4.669613501713471e-06, "loss": 2.6095, "step": 42296 }, { "epoch": 2.625675088459867, "grad_norm": 0.38834986934259386, "learning_rate": 4.6680896490885694e-06, "loss": 2.6999, "step": 42297 }, { "epoch": 2.625737165559625, "grad_norm": 0.41211456367878246, "learning_rate": 4.666566032971265e-06, "loss": 2.5475, "step": 42298 }, { "epoch": 2.625799242659383, "grad_norm": 0.4390890011646342, "learning_rate": 4.665042653369489e-06, "loss": 2.6093, "step": 42299 }, { "epoch": 2.625861319759141, "grad_norm": 0.4278527535838413, "learning_rate": 4.663519510291186e-06, "loss": 2.7374, "step": 42300 }, { "epoch": 2.625923396858899, "grad_norm": 0.3847985810238813, "learning_rate": 4.661996603744312e-06, "loss": 2.6492, "step": 42301 }, { "epoch": 2.625985473958657, "grad_norm": 0.38804253468429545, "learning_rate": 4.660473933736797e-06, "loss": 2.6596, "step": 42302 }, { "epoch": 2.6260475510584147, "grad_norm": 0.403715406595195, "learning_rate": 4.658951500276609e-06, "loss": 2.6123, "step": 42303 }, { "epoch": 2.6261096281581726, "grad_norm": 0.4227929195767572, "learning_rate": 4.6574293033716744e-06, "loss": 2.6456, "step": 42304 }, { "epoch": 2.6261717052579305, "grad_norm": 0.3982490954284069, "learning_rate": 4.655907343029936e-06, "loss": 2.6537, "step": 42305 }, { "epoch": 2.626233782357688, "grad_norm": 0.39704952457650755, "learning_rate": 4.654385619259333e-06, "loss": 2.7263, "step": 42306 }, { "epoch": 2.6262958594574464, "grad_norm": 0.3777691206785517, "learning_rate": 4.652864132067813e-06, "loss": 2.6129, "step": 42307 }, { "epoch": 2.626357936557204, "grad_norm": 0.42508775772709256, "learning_rate": 4.651342881463311e-06, "loss": 2.6733, "step": 42308 }, { "epoch": 2.626420013656962, "grad_norm": 0.38134367057876367, "learning_rate": 4.649821867453763e-06, "loss": 2.6365, "step": 42309 }, { "epoch": 2.6264820907567197, "grad_norm": 0.38722425232506674, "learning_rate": 4.648301090047097e-06, "loss": 2.6542, "step": 42310 }, { "epoch": 2.6265441678564776, "grad_norm": 0.399053900886677, "learning_rate": 4.646780549251256e-06, "loss": 2.6736, "step": 42311 }, { "epoch": 2.6266062449562355, "grad_norm": 0.37867480022480304, "learning_rate": 4.645260245074174e-06, "loss": 2.5781, "step": 42312 }, { "epoch": 2.6266683220559934, "grad_norm": 0.40649747646669904, "learning_rate": 4.643740177523781e-06, "loss": 2.6596, "step": 42313 }, { "epoch": 2.6267303991557513, "grad_norm": 0.4372613460296647, "learning_rate": 4.642220346608006e-06, "loss": 2.6907, "step": 42314 }, { "epoch": 2.6267924762555093, "grad_norm": 0.3976168877206713, "learning_rate": 4.640700752334765e-06, "loss": 2.7041, "step": 42315 }, { "epoch": 2.626854553355267, "grad_norm": 0.3905431089393267, "learning_rate": 4.639181394712011e-06, "loss": 2.7405, "step": 42316 }, { "epoch": 2.626916630455025, "grad_norm": 0.37758182810335533, "learning_rate": 4.637662273747662e-06, "loss": 2.6804, "step": 42317 }, { "epoch": 2.626978707554783, "grad_norm": 0.3971535246009132, "learning_rate": 4.636143389449632e-06, "loss": 2.6436, "step": 42318 }, { "epoch": 2.627040784654541, "grad_norm": 0.3742522692551041, "learning_rate": 4.63462474182586e-06, "loss": 2.7101, "step": 42319 }, { "epoch": 2.627102861754299, "grad_norm": 0.37265137483298916, "learning_rate": 4.633106330884257e-06, "loss": 2.7531, "step": 42320 }, { "epoch": 2.6271649388540568, "grad_norm": 0.3609370821995815, "learning_rate": 4.63158815663276e-06, "loss": 2.5915, "step": 42321 }, { "epoch": 2.6272270159538147, "grad_norm": 0.39942766087148274, "learning_rate": 4.63007021907928e-06, "loss": 2.6465, "step": 42322 }, { "epoch": 2.6272890930535726, "grad_norm": 0.3782709242094104, "learning_rate": 4.628552518231738e-06, "loss": 2.6908, "step": 42323 }, { "epoch": 2.6273511701533305, "grad_norm": 0.39793169415976143, "learning_rate": 4.627035054098044e-06, "loss": 2.6177, "step": 42324 }, { "epoch": 2.6274132472530884, "grad_norm": 0.3959236690270281, "learning_rate": 4.6255178266861264e-06, "loss": 2.7058, "step": 42325 }, { "epoch": 2.6274753243528464, "grad_norm": 0.3756713817605732, "learning_rate": 4.624000836003906e-06, "loss": 2.5587, "step": 42326 }, { "epoch": 2.6275374014526043, "grad_norm": 0.3732649081115213, "learning_rate": 4.622484082059281e-06, "loss": 2.658, "step": 42327 }, { "epoch": 2.627599478552362, "grad_norm": 0.3717129009439266, "learning_rate": 4.620967564860179e-06, "loss": 2.6927, "step": 42328 }, { "epoch": 2.6276615556521197, "grad_norm": 0.4186000512799716, "learning_rate": 4.619451284414494e-06, "loss": 2.6851, "step": 42329 }, { "epoch": 2.627723632751878, "grad_norm": 0.4023666004988184, "learning_rate": 4.617935240730154e-06, "loss": 2.6434, "step": 42330 }, { "epoch": 2.6277857098516355, "grad_norm": 0.36662246977787816, "learning_rate": 4.616419433815067e-06, "loss": 2.7127, "step": 42331 }, { "epoch": 2.627847786951394, "grad_norm": 0.37776302635307923, "learning_rate": 4.614903863677133e-06, "loss": 2.5864, "step": 42332 }, { "epoch": 2.6279098640511513, "grad_norm": 0.4103412392655447, "learning_rate": 4.613388530324259e-06, "loss": 2.6651, "step": 42333 }, { "epoch": 2.6279719411509093, "grad_norm": 0.4179797839553266, "learning_rate": 4.611873433764358e-06, "loss": 2.7149, "step": 42334 }, { "epoch": 2.628034018250667, "grad_norm": 0.37785594920681437, "learning_rate": 4.610358574005335e-06, "loss": 2.7296, "step": 42335 }, { "epoch": 2.628096095350425, "grad_norm": 0.39075324814264645, "learning_rate": 4.6088439510550916e-06, "loss": 2.654, "step": 42336 }, { "epoch": 2.628158172450183, "grad_norm": 0.4095462622832773, "learning_rate": 4.607329564921525e-06, "loss": 2.6685, "step": 42337 }, { "epoch": 2.628220249549941, "grad_norm": 0.42471933493967917, "learning_rate": 4.605815415612536e-06, "loss": 2.554, "step": 42338 }, { "epoch": 2.628282326649699, "grad_norm": 0.41987994672241086, "learning_rate": 4.604301503136033e-06, "loss": 2.6211, "step": 42339 }, { "epoch": 2.6283444037494568, "grad_norm": 0.43764319801139406, "learning_rate": 4.602787827499905e-06, "loss": 2.7476, "step": 42340 }, { "epoch": 2.6284064808492147, "grad_norm": 0.41050941158429344, "learning_rate": 4.601274388712057e-06, "loss": 2.6383, "step": 42341 }, { "epoch": 2.6284685579489726, "grad_norm": 0.38986939783526364, "learning_rate": 4.5997611867803765e-06, "loss": 2.5684, "step": 42342 }, { "epoch": 2.6285306350487305, "grad_norm": 0.3951888128388845, "learning_rate": 4.5982482217127695e-06, "loss": 2.6725, "step": 42343 }, { "epoch": 2.6285927121484884, "grad_norm": 0.3821156250624443, "learning_rate": 4.596735493517123e-06, "loss": 2.5875, "step": 42344 }, { "epoch": 2.6286547892482464, "grad_norm": 0.42042527298991883, "learning_rate": 4.595223002201327e-06, "loss": 2.657, "step": 42345 }, { "epoch": 2.6287168663480043, "grad_norm": 0.4259846227863642, "learning_rate": 4.593710747773266e-06, "loss": 2.5935, "step": 42346 }, { "epoch": 2.628778943447762, "grad_norm": 0.4009051296935982, "learning_rate": 4.592198730240854e-06, "loss": 2.7325, "step": 42347 }, { "epoch": 2.62884102054752, "grad_norm": 0.3965152296265818, "learning_rate": 4.590686949611955e-06, "loss": 2.7182, "step": 42348 }, { "epoch": 2.628903097647278, "grad_norm": 0.3888453762465275, "learning_rate": 4.589175405894475e-06, "loss": 2.5875, "step": 42349 }, { "epoch": 2.628965174747036, "grad_norm": 0.3896849383321229, "learning_rate": 4.5876640990962785e-06, "loss": 2.6118, "step": 42350 }, { "epoch": 2.629027251846794, "grad_norm": 0.3986669141732493, "learning_rate": 4.586153029225271e-06, "loss": 2.7045, "step": 42351 }, { "epoch": 2.6290893289465513, "grad_norm": 0.40259481207806624, "learning_rate": 4.584642196289318e-06, "loss": 2.7122, "step": 42352 }, { "epoch": 2.6291514060463097, "grad_norm": 0.37407795481357087, "learning_rate": 4.58313160029632e-06, "loss": 2.6777, "step": 42353 }, { "epoch": 2.629213483146067, "grad_norm": 0.4111133308053593, "learning_rate": 4.581621241254153e-06, "loss": 2.6467, "step": 42354 }, { "epoch": 2.6292755602458255, "grad_norm": 0.41410562902432474, "learning_rate": 4.580111119170688e-06, "loss": 2.7039, "step": 42355 }, { "epoch": 2.629337637345583, "grad_norm": 0.39457785453873223, "learning_rate": 4.578601234053803e-06, "loss": 2.7058, "step": 42356 }, { "epoch": 2.6293997144453414, "grad_norm": 0.40245561988869744, "learning_rate": 4.577091585911391e-06, "loss": 2.7216, "step": 42357 }, { "epoch": 2.629461791545099, "grad_norm": 0.36382585127932815, "learning_rate": 4.57558217475132e-06, "loss": 2.6585, "step": 42358 }, { "epoch": 2.6295238686448568, "grad_norm": 0.39476710530812625, "learning_rate": 4.574073000581464e-06, "loss": 2.6791, "step": 42359 }, { "epoch": 2.6295859457446147, "grad_norm": 0.4042558142409652, "learning_rate": 4.57256406340969e-06, "loss": 2.6189, "step": 42360 }, { "epoch": 2.6296480228443726, "grad_norm": 0.3875331972114954, "learning_rate": 4.571055363243881e-06, "loss": 2.6397, "step": 42361 }, { "epoch": 2.6297100999441305, "grad_norm": 0.3809552585508735, "learning_rate": 4.5695469000919085e-06, "loss": 2.6328, "step": 42362 }, { "epoch": 2.6297721770438884, "grad_norm": 0.3837908290987391, "learning_rate": 4.568038673961639e-06, "loss": 2.6802, "step": 42363 }, { "epoch": 2.6298342541436464, "grad_norm": 0.4129854084455292, "learning_rate": 4.566530684860937e-06, "loss": 2.6805, "step": 42364 }, { "epoch": 2.6298963312434043, "grad_norm": 0.40245101017497575, "learning_rate": 4.565022932797664e-06, "loss": 2.6498, "step": 42365 }, { "epoch": 2.629958408343162, "grad_norm": 0.3818876226620947, "learning_rate": 4.5635154177797065e-06, "loss": 2.6656, "step": 42366 }, { "epoch": 2.63002048544292, "grad_norm": 0.38462069730115545, "learning_rate": 4.562008139814922e-06, "loss": 2.7276, "step": 42367 }, { "epoch": 2.630082562542678, "grad_norm": 0.41263198258305506, "learning_rate": 4.560501098911169e-06, "loss": 2.5614, "step": 42368 }, { "epoch": 2.630144639642436, "grad_norm": 0.4187787487352999, "learning_rate": 4.5589942950763076e-06, "loss": 2.629, "step": 42369 }, { "epoch": 2.630206716742194, "grad_norm": 0.40867430583950676, "learning_rate": 4.557487728318205e-06, "loss": 2.7299, "step": 42370 }, { "epoch": 2.630268793841952, "grad_norm": 0.3842150820711782, "learning_rate": 4.5559813986447265e-06, "loss": 2.721, "step": 42371 }, { "epoch": 2.6303308709417097, "grad_norm": 0.40553620793440387, "learning_rate": 4.554475306063727e-06, "loss": 2.609, "step": 42372 }, { "epoch": 2.6303929480414676, "grad_norm": 0.41840254474526123, "learning_rate": 4.552969450583055e-06, "loss": 2.5895, "step": 42373 }, { "epoch": 2.6304550251412255, "grad_norm": 0.4142567880519806, "learning_rate": 4.551463832210573e-06, "loss": 2.6584, "step": 42374 }, { "epoch": 2.6305171022409835, "grad_norm": 0.39586952529219643, "learning_rate": 4.549958450954139e-06, "loss": 2.6233, "step": 42375 }, { "epoch": 2.6305791793407414, "grad_norm": 0.38814345818087503, "learning_rate": 4.54845330682161e-06, "loss": 2.5663, "step": 42376 }, { "epoch": 2.630641256440499, "grad_norm": 0.37388062302809827, "learning_rate": 4.546948399820833e-06, "loss": 2.6319, "step": 42377 }, { "epoch": 2.630703333540257, "grad_norm": 0.3999874038501141, "learning_rate": 4.545443729959648e-06, "loss": 2.6823, "step": 42378 }, { "epoch": 2.6307654106400147, "grad_norm": 0.39456053898935983, "learning_rate": 4.5439392972459305e-06, "loss": 2.7175, "step": 42379 }, { "epoch": 2.630827487739773, "grad_norm": 0.3903266721218705, "learning_rate": 4.542435101687515e-06, "loss": 2.593, "step": 42380 }, { "epoch": 2.6308895648395305, "grad_norm": 0.3580122694038801, "learning_rate": 4.540931143292248e-06, "loss": 2.6234, "step": 42381 }, { "epoch": 2.6309516419392884, "grad_norm": 0.383867658570376, "learning_rate": 4.539427422067971e-06, "loss": 2.641, "step": 42382 }, { "epoch": 2.6310137190390464, "grad_norm": 0.4237473933947128, "learning_rate": 4.5379239380225425e-06, "loss": 2.7192, "step": 42383 }, { "epoch": 2.6310757961388043, "grad_norm": 0.4031818882036163, "learning_rate": 4.536420691163806e-06, "loss": 2.6316, "step": 42384 }, { "epoch": 2.631137873238562, "grad_norm": 0.3858892039918374, "learning_rate": 4.534917681499601e-06, "loss": 2.7186, "step": 42385 }, { "epoch": 2.63119995033832, "grad_norm": 0.39743926717932493, "learning_rate": 4.533414909037764e-06, "loss": 2.7181, "step": 42386 }, { "epoch": 2.631262027438078, "grad_norm": 0.40795012669332814, "learning_rate": 4.53191237378614e-06, "loss": 2.5884, "step": 42387 }, { "epoch": 2.631324104537836, "grad_norm": 0.38425760569891176, "learning_rate": 4.530410075752562e-06, "loss": 2.6378, "step": 42388 }, { "epoch": 2.631386181637594, "grad_norm": 0.41250851362285884, "learning_rate": 4.528908014944877e-06, "loss": 2.662, "step": 42389 }, { "epoch": 2.6314482587373518, "grad_norm": 0.3891519045463479, "learning_rate": 4.52740619137092e-06, "loss": 2.605, "step": 42390 }, { "epoch": 2.6315103358371097, "grad_norm": 0.3795058039383492, "learning_rate": 4.525904605038522e-06, "loss": 2.7551, "step": 42391 }, { "epoch": 2.6315724129368676, "grad_norm": 0.41043344849035707, "learning_rate": 4.5244032559555115e-06, "loss": 2.6811, "step": 42392 }, { "epoch": 2.6316344900366255, "grad_norm": 0.396402754627255, "learning_rate": 4.522902144129737e-06, "loss": 2.6621, "step": 42393 }, { "epoch": 2.6316965671363834, "grad_norm": 0.3688467965533898, "learning_rate": 4.52140126956902e-06, "loss": 2.7234, "step": 42394 }, { "epoch": 2.6317586442361414, "grad_norm": 0.4192797875089913, "learning_rate": 4.519900632281193e-06, "loss": 2.6722, "step": 42395 }, { "epoch": 2.6318207213358993, "grad_norm": 0.41004293783907125, "learning_rate": 4.5184002322740785e-06, "loss": 2.7188, "step": 42396 }, { "epoch": 2.631882798435657, "grad_norm": 0.42424616435289036, "learning_rate": 4.516900069555519e-06, "loss": 2.6611, "step": 42397 }, { "epoch": 2.631944875535415, "grad_norm": 0.38104728562494855, "learning_rate": 4.515400144133336e-06, "loss": 2.5703, "step": 42398 }, { "epoch": 2.632006952635173, "grad_norm": 0.3870440954449136, "learning_rate": 4.513900456015347e-06, "loss": 2.6534, "step": 42399 }, { "epoch": 2.6320690297349305, "grad_norm": 0.39911847730898575, "learning_rate": 4.512401005209388e-06, "loss": 2.6672, "step": 42400 }, { "epoch": 2.632131106834689, "grad_norm": 0.399175726705451, "learning_rate": 4.51090179172326e-06, "loss": 2.7574, "step": 42401 }, { "epoch": 2.6321931839344463, "grad_norm": 0.39719124907038045, "learning_rate": 4.509402815564812e-06, "loss": 2.6714, "step": 42402 }, { "epoch": 2.6322552610342047, "grad_norm": 0.40336978045557004, "learning_rate": 4.507904076741853e-06, "loss": 2.605, "step": 42403 }, { "epoch": 2.632317338133962, "grad_norm": 0.41302643115117393, "learning_rate": 4.506405575262202e-06, "loss": 2.6436, "step": 42404 }, { "epoch": 2.6323794152337205, "grad_norm": 0.4020881581716997, "learning_rate": 4.504907311133666e-06, "loss": 2.6773, "step": 42405 }, { "epoch": 2.632441492333478, "grad_norm": 0.4288179268426163, "learning_rate": 4.503409284364085e-06, "loss": 2.7048, "step": 42406 }, { "epoch": 2.632503569433236, "grad_norm": 0.3957871569084635, "learning_rate": 4.501911494961264e-06, "loss": 2.6115, "step": 42407 }, { "epoch": 2.632565646532994, "grad_norm": 0.3905495952306195, "learning_rate": 4.500413942933013e-06, "loss": 2.5554, "step": 42408 }, { "epoch": 2.6326277236327518, "grad_norm": 0.40793591747441615, "learning_rate": 4.4989166282871476e-06, "loss": 2.5954, "step": 42409 }, { "epoch": 2.6326898007325097, "grad_norm": 0.3965502717076959, "learning_rate": 4.497419551031468e-06, "loss": 2.6313, "step": 42410 }, { "epoch": 2.6327518778322676, "grad_norm": 0.4045779960089077, "learning_rate": 4.4959227111738115e-06, "loss": 2.6577, "step": 42411 }, { "epoch": 2.6328139549320255, "grad_norm": 0.3899942264374057, "learning_rate": 4.494426108721966e-06, "loss": 2.6505, "step": 42412 }, { "epoch": 2.6328760320317834, "grad_norm": 0.37590125184252815, "learning_rate": 4.4929297436837544e-06, "loss": 2.6276, "step": 42413 }, { "epoch": 2.6329381091315414, "grad_norm": 0.41277501735274813, "learning_rate": 4.491433616066959e-06, "loss": 2.6501, "step": 42414 }, { "epoch": 2.6330001862312993, "grad_norm": 0.38543236760643623, "learning_rate": 4.489937725879417e-06, "loss": 2.6708, "step": 42415 }, { "epoch": 2.633062263331057, "grad_norm": 0.40106906632131845, "learning_rate": 4.488442073128907e-06, "loss": 2.6647, "step": 42416 }, { "epoch": 2.633124340430815, "grad_norm": 0.4042771376062799, "learning_rate": 4.486946657823249e-06, "loss": 2.6922, "step": 42417 }, { "epoch": 2.633186417530573, "grad_norm": 0.39598798449546957, "learning_rate": 4.485451479970243e-06, "loss": 2.7649, "step": 42418 }, { "epoch": 2.633248494630331, "grad_norm": 0.40059446402951154, "learning_rate": 4.483956539577672e-06, "loss": 2.6456, "step": 42419 }, { "epoch": 2.633310571730089, "grad_norm": 0.40180915204064227, "learning_rate": 4.482461836653362e-06, "loss": 2.7036, "step": 42420 }, { "epoch": 2.633372648829847, "grad_norm": 0.40464312704125643, "learning_rate": 4.480967371205097e-06, "loss": 2.681, "step": 42421 }, { "epoch": 2.6334347259296047, "grad_norm": 0.37009772096033733, "learning_rate": 4.479473143240676e-06, "loss": 2.5823, "step": 42422 }, { "epoch": 2.6334968030293626, "grad_norm": 0.4073398350219123, "learning_rate": 4.477979152767897e-06, "loss": 2.6922, "step": 42423 }, { "epoch": 2.6335588801291205, "grad_norm": 0.42846724898703, "learning_rate": 4.476485399794539e-06, "loss": 2.7375, "step": 42424 }, { "epoch": 2.633620957228878, "grad_norm": 0.3861419537725567, "learning_rate": 4.474991884328417e-06, "loss": 2.7514, "step": 42425 }, { "epoch": 2.6336830343286364, "grad_norm": 0.4013584813568292, "learning_rate": 4.473498606377313e-06, "loss": 2.7282, "step": 42426 }, { "epoch": 2.633745111428394, "grad_norm": 0.39991093254880333, "learning_rate": 4.472005565949022e-06, "loss": 2.6071, "step": 42427 }, { "epoch": 2.633807188528152, "grad_norm": 0.4178171150163775, "learning_rate": 4.470512763051321e-06, "loss": 2.6699, "step": 42428 }, { "epoch": 2.6338692656279097, "grad_norm": 0.4072690250936073, "learning_rate": 4.469020197692014e-06, "loss": 2.6409, "step": 42429 }, { "epoch": 2.6339313427276676, "grad_norm": 0.4034508223812332, "learning_rate": 4.467527869878885e-06, "loss": 2.6496, "step": 42430 }, { "epoch": 2.6339934198274255, "grad_norm": 0.3913564128708043, "learning_rate": 4.466035779619709e-06, "loss": 2.6982, "step": 42431 }, { "epoch": 2.6340554969271834, "grad_norm": 0.40638692089022066, "learning_rate": 4.464543926922276e-06, "loss": 2.7589, "step": 42432 }, { "epoch": 2.6341175740269414, "grad_norm": 0.4097034679941108, "learning_rate": 4.46305231179438e-06, "loss": 2.7121, "step": 42433 }, { "epoch": 2.6341796511266993, "grad_norm": 0.41569167943151647, "learning_rate": 4.461560934243791e-06, "loss": 2.7087, "step": 42434 }, { "epoch": 2.634241728226457, "grad_norm": 0.3828920943366849, "learning_rate": 4.460069794278299e-06, "loss": 2.687, "step": 42435 }, { "epoch": 2.634303805326215, "grad_norm": 0.38284829030790446, "learning_rate": 4.458578891905674e-06, "loss": 2.6469, "step": 42436 }, { "epoch": 2.634365882425973, "grad_norm": 0.4020331767362585, "learning_rate": 4.457088227133688e-06, "loss": 2.6957, "step": 42437 }, { "epoch": 2.634427959525731, "grad_norm": 0.40771868127170435, "learning_rate": 4.45559779997014e-06, "loss": 2.5079, "step": 42438 }, { "epoch": 2.634490036625489, "grad_norm": 0.38800653222204157, "learning_rate": 4.454107610422797e-06, "loss": 2.6914, "step": 42439 }, { "epoch": 2.634552113725247, "grad_norm": 0.40831807904393075, "learning_rate": 4.452617658499425e-06, "loss": 2.6796, "step": 42440 }, { "epoch": 2.6346141908250047, "grad_norm": 0.41721747072841037, "learning_rate": 4.4511279442078e-06, "loss": 2.6841, "step": 42441 }, { "epoch": 2.6346762679247626, "grad_norm": 0.42269820676693504, "learning_rate": 4.449638467555706e-06, "loss": 2.693, "step": 42442 }, { "epoch": 2.6347383450245205, "grad_norm": 0.38210096210771927, "learning_rate": 4.448149228550908e-06, "loss": 2.7329, "step": 42443 }, { "epoch": 2.6348004221242785, "grad_norm": 0.4011412307044643, "learning_rate": 4.446660227201166e-06, "loss": 2.5739, "step": 42444 }, { "epoch": 2.6348624992240364, "grad_norm": 0.4236633402340907, "learning_rate": 4.445171463514253e-06, "loss": 2.7209, "step": 42445 }, { "epoch": 2.6349245763237943, "grad_norm": 0.39245787719493236, "learning_rate": 4.443682937497951e-06, "loss": 2.6029, "step": 42446 }, { "epoch": 2.634986653423552, "grad_norm": 0.41753723472461396, "learning_rate": 4.442194649160009e-06, "loss": 2.569, "step": 42447 }, { "epoch": 2.6350487305233097, "grad_norm": 0.4295000187470828, "learning_rate": 4.440706598508199e-06, "loss": 2.6977, "step": 42448 }, { "epoch": 2.635110807623068, "grad_norm": 0.42478960383152736, "learning_rate": 4.4392187855502695e-06, "loss": 2.7402, "step": 42449 }, { "epoch": 2.6351728847228255, "grad_norm": 0.40420997533427677, "learning_rate": 4.4377312102940104e-06, "loss": 2.6851, "step": 42450 }, { "epoch": 2.635234961822584, "grad_norm": 0.40276342559141715, "learning_rate": 4.436243872747159e-06, "loss": 2.6588, "step": 42451 }, { "epoch": 2.6352970389223414, "grad_norm": 0.3985062895090643, "learning_rate": 4.434756772917486e-06, "loss": 2.6351, "step": 42452 }, { "epoch": 2.6353591160220997, "grad_norm": 0.3937506743834034, "learning_rate": 4.433269910812759e-06, "loss": 2.6013, "step": 42453 }, { "epoch": 2.635421193121857, "grad_norm": 0.39981120914419604, "learning_rate": 4.431783286440716e-06, "loss": 2.6483, "step": 42454 }, { "epoch": 2.635483270221615, "grad_norm": 0.391237103581783, "learning_rate": 4.430296899809117e-06, "loss": 2.7527, "step": 42455 }, { "epoch": 2.635545347321373, "grad_norm": 0.37636009569915096, "learning_rate": 4.4288107509257326e-06, "loss": 2.6461, "step": 42456 }, { "epoch": 2.635607424421131, "grad_norm": 0.4154503613176013, "learning_rate": 4.427324839798297e-06, "loss": 2.7461, "step": 42457 }, { "epoch": 2.635669501520889, "grad_norm": 0.3854057099958649, "learning_rate": 4.425839166434581e-06, "loss": 2.5487, "step": 42458 }, { "epoch": 2.6357315786206468, "grad_norm": 0.39125210968643165, "learning_rate": 4.4243537308423175e-06, "loss": 2.641, "step": 42459 }, { "epoch": 2.6357936557204047, "grad_norm": 0.38643107648011843, "learning_rate": 4.422868533029262e-06, "loss": 2.6923, "step": 42460 }, { "epoch": 2.6358557328201626, "grad_norm": 0.4197333156491524, "learning_rate": 4.421383573003168e-06, "loss": 2.8524, "step": 42461 }, { "epoch": 2.6359178099199205, "grad_norm": 0.37677046940942127, "learning_rate": 4.4198988507717864e-06, "loss": 2.6652, "step": 42462 }, { "epoch": 2.6359798870196784, "grad_norm": 0.38001441147104154, "learning_rate": 4.418414366342855e-06, "loss": 2.6105, "step": 42463 }, { "epoch": 2.6360419641194364, "grad_norm": 0.4154775121194563, "learning_rate": 4.416930119724111e-06, "loss": 2.6674, "step": 42464 }, { "epoch": 2.6361040412191943, "grad_norm": 0.401103320481264, "learning_rate": 4.415446110923316e-06, "loss": 2.6933, "step": 42465 }, { "epoch": 2.636166118318952, "grad_norm": 0.3970441577889407, "learning_rate": 4.413962339948208e-06, "loss": 2.5694, "step": 42466 }, { "epoch": 2.63622819541871, "grad_norm": 0.41642474264052326, "learning_rate": 4.4124788068065245e-06, "loss": 2.7429, "step": 42467 }, { "epoch": 2.636290272518468, "grad_norm": 0.39460647923312003, "learning_rate": 4.410995511505994e-06, "loss": 2.7428, "step": 42468 }, { "epoch": 2.636352349618226, "grad_norm": 0.4131366346916838, "learning_rate": 4.409512454054382e-06, "loss": 2.731, "step": 42469 }, { "epoch": 2.636414426717984, "grad_norm": 0.3889699013754966, "learning_rate": 4.408029634459404e-06, "loss": 2.7376, "step": 42470 }, { "epoch": 2.636476503817742, "grad_norm": 0.3802052676281939, "learning_rate": 4.406547052728804e-06, "loss": 2.6457, "step": 42471 }, { "epoch": 2.6365385809174997, "grad_norm": 0.38654577837390297, "learning_rate": 4.405064708870321e-06, "loss": 2.6892, "step": 42472 }, { "epoch": 2.636600658017257, "grad_norm": 0.38541131617294644, "learning_rate": 4.403582602891671e-06, "loss": 2.7459, "step": 42473 }, { "epoch": 2.6366627351170155, "grad_norm": 0.40831931983814684, "learning_rate": 4.402100734800613e-06, "loss": 2.7116, "step": 42474 }, { "epoch": 2.636724812216773, "grad_norm": 0.4220736353865647, "learning_rate": 4.400619104604858e-06, "loss": 2.7062, "step": 42475 }, { "epoch": 2.6367868893165314, "grad_norm": 0.40428461986893804, "learning_rate": 4.3991377123121515e-06, "loss": 2.6596, "step": 42476 }, { "epoch": 2.636848966416289, "grad_norm": 0.4248476379429882, "learning_rate": 4.3976565579301965e-06, "loss": 2.6168, "step": 42477 }, { "epoch": 2.6369110435160468, "grad_norm": 0.41637798879026555, "learning_rate": 4.396175641466749e-06, "loss": 2.6275, "step": 42478 }, { "epoch": 2.6369731206158047, "grad_norm": 0.40854692354893885, "learning_rate": 4.394694962929524e-06, "loss": 2.6906, "step": 42479 }, { "epoch": 2.6370351977155626, "grad_norm": 0.4319794822153902, "learning_rate": 4.39321452232625e-06, "loss": 2.6926, "step": 42480 }, { "epoch": 2.6370972748153205, "grad_norm": 0.41310895129727854, "learning_rate": 4.391734319664637e-06, "loss": 2.6287, "step": 42481 }, { "epoch": 2.6371593519150784, "grad_norm": 0.42619931406277395, "learning_rate": 4.390254354952411e-06, "loss": 2.6905, "step": 42482 }, { "epoch": 2.6372214290148364, "grad_norm": 0.39118322539961425, "learning_rate": 4.388774628197317e-06, "loss": 2.7196, "step": 42483 }, { "epoch": 2.6372835061145943, "grad_norm": 0.4043505536255927, "learning_rate": 4.387295139407055e-06, "loss": 2.6684, "step": 42484 }, { "epoch": 2.637345583214352, "grad_norm": 0.42573223356830586, "learning_rate": 4.385815888589345e-06, "loss": 2.7228, "step": 42485 }, { "epoch": 2.63740766031411, "grad_norm": 0.4190679161808432, "learning_rate": 4.38433687575191e-06, "loss": 2.6462, "step": 42486 }, { "epoch": 2.637469737413868, "grad_norm": 0.4079516608210374, "learning_rate": 4.3828581009024535e-06, "loss": 2.7194, "step": 42487 }, { "epoch": 2.637531814513626, "grad_norm": 0.39808336115046367, "learning_rate": 4.3813795640487096e-06, "loss": 2.6938, "step": 42488 }, { "epoch": 2.637593891613384, "grad_norm": 0.4078774681811602, "learning_rate": 4.379901265198383e-06, "loss": 2.7266, "step": 42489 }, { "epoch": 2.637655968713142, "grad_norm": 0.4434107780066725, "learning_rate": 4.378423204359189e-06, "loss": 2.6628, "step": 42490 }, { "epoch": 2.6377180458128997, "grad_norm": 0.4045998180061361, "learning_rate": 4.376945381538822e-06, "loss": 2.6588, "step": 42491 }, { "epoch": 2.6377801229126576, "grad_norm": 0.39256432732015745, "learning_rate": 4.375467796745014e-06, "loss": 2.7325, "step": 42492 }, { "epoch": 2.6378422000124155, "grad_norm": 0.40082017140710835, "learning_rate": 4.3739904499854715e-06, "loss": 2.637, "step": 42493 }, { "epoch": 2.6379042771121735, "grad_norm": 0.4087876328517594, "learning_rate": 4.372513341267898e-06, "loss": 2.6702, "step": 42494 }, { "epoch": 2.6379663542119314, "grad_norm": 0.3948291749003913, "learning_rate": 4.371036470599982e-06, "loss": 2.6054, "step": 42495 }, { "epoch": 2.638028431311689, "grad_norm": 0.39972583196225264, "learning_rate": 4.369559837989462e-06, "loss": 2.6822, "step": 42496 }, { "epoch": 2.638090508411447, "grad_norm": 0.3949041587108297, "learning_rate": 4.368083443444021e-06, "loss": 2.6305, "step": 42497 }, { "epoch": 2.6381525855112047, "grad_norm": 0.3717970558033003, "learning_rate": 4.366607286971369e-06, "loss": 2.6151, "step": 42498 }, { "epoch": 2.638214662610963, "grad_norm": 0.3877287881461374, "learning_rate": 4.3651313685792064e-06, "loss": 2.6496, "step": 42499 }, { "epoch": 2.6382767397107205, "grad_norm": 0.37164711100331616, "learning_rate": 4.36365568827522e-06, "loss": 2.6108, "step": 42500 }, { "epoch": 2.638338816810479, "grad_norm": 0.38634276174313825, "learning_rate": 4.3621802460671266e-06, "loss": 2.7335, "step": 42501 }, { "epoch": 2.6384008939102364, "grad_norm": 0.38700977125695774, "learning_rate": 4.36070504196262e-06, "loss": 2.6793, "step": 42502 }, { "epoch": 2.6384629710099943, "grad_norm": 0.41457682281762354, "learning_rate": 4.359230075969395e-06, "loss": 2.6891, "step": 42503 }, { "epoch": 2.638525048109752, "grad_norm": 0.4137600747982243, "learning_rate": 4.357755348095138e-06, "loss": 2.6517, "step": 42504 }, { "epoch": 2.63858712520951, "grad_norm": 0.40099293136074327, "learning_rate": 4.356280858347561e-06, "loss": 2.5961, "step": 42505 }, { "epoch": 2.638649202309268, "grad_norm": 0.38366580760261965, "learning_rate": 4.354806606734346e-06, "loss": 2.7244, "step": 42506 }, { "epoch": 2.638711279409026, "grad_norm": 0.39121017494070204, "learning_rate": 4.353332593263188e-06, "loss": 2.603, "step": 42507 }, { "epoch": 2.638773356508784, "grad_norm": 0.37866128297933277, "learning_rate": 4.3518588179417674e-06, "loss": 2.6707, "step": 42508 }, { "epoch": 2.638835433608542, "grad_norm": 0.38381546356791374, "learning_rate": 4.35038528077778e-06, "loss": 2.6404, "step": 42509 }, { "epoch": 2.6388975107082997, "grad_norm": 0.3799666089206938, "learning_rate": 4.348911981778919e-06, "loss": 2.6858, "step": 42510 }, { "epoch": 2.6389595878080576, "grad_norm": 0.4428673925949602, "learning_rate": 4.347438920952868e-06, "loss": 2.6553, "step": 42511 }, { "epoch": 2.6390216649078155, "grad_norm": 0.4085272638264618, "learning_rate": 4.345966098307314e-06, "loss": 2.6962, "step": 42512 }, { "epoch": 2.6390837420075735, "grad_norm": 0.39840617711231624, "learning_rate": 4.344493513849923e-06, "loss": 2.5983, "step": 42513 }, { "epoch": 2.6391458191073314, "grad_norm": 0.3915292573230062, "learning_rate": 4.343021167588396e-06, "loss": 2.5555, "step": 42514 }, { "epoch": 2.6392078962070893, "grad_norm": 0.3675153769954527, "learning_rate": 4.34154905953042e-06, "loss": 2.6008, "step": 42515 }, { "epoch": 2.639269973306847, "grad_norm": 0.3883781089707388, "learning_rate": 4.340077189683667e-06, "loss": 2.6309, "step": 42516 }, { "epoch": 2.639332050406605, "grad_norm": 0.4175850363606384, "learning_rate": 4.338605558055814e-06, "loss": 2.7013, "step": 42517 }, { "epoch": 2.639394127506363, "grad_norm": 0.4018781245393448, "learning_rate": 4.337134164654527e-06, "loss": 2.5947, "step": 42518 }, { "epoch": 2.639456204606121, "grad_norm": 0.38082897358380624, "learning_rate": 4.335663009487511e-06, "loss": 2.6263, "step": 42519 }, { "epoch": 2.639518281705879, "grad_norm": 0.40695210454340136, "learning_rate": 4.334192092562428e-06, "loss": 2.6466, "step": 42520 }, { "epoch": 2.6395803588056364, "grad_norm": 0.3746392780099482, "learning_rate": 4.332721413886947e-06, "loss": 2.5796, "step": 42521 }, { "epoch": 2.6396424359053947, "grad_norm": 0.3723632180218627, "learning_rate": 4.3312509734687475e-06, "loss": 2.7334, "step": 42522 }, { "epoch": 2.639704513005152, "grad_norm": 0.4108789525551309, "learning_rate": 4.329780771315484e-06, "loss": 2.6329, "step": 42523 }, { "epoch": 2.6397665901049105, "grad_norm": 0.40899918246129224, "learning_rate": 4.3283108074348545e-06, "loss": 2.6716, "step": 42524 }, { "epoch": 2.639828667204668, "grad_norm": 0.3927659301950875, "learning_rate": 4.326841081834515e-06, "loss": 2.6956, "step": 42525 }, { "epoch": 2.639890744304426, "grad_norm": 0.3978388533811885, "learning_rate": 4.325371594522132e-06, "loss": 2.6127, "step": 42526 }, { "epoch": 2.639952821404184, "grad_norm": 0.39362796451497706, "learning_rate": 4.32390234550536e-06, "loss": 2.6211, "step": 42527 }, { "epoch": 2.6400148985039418, "grad_norm": 0.3874124220356198, "learning_rate": 4.322433334791892e-06, "loss": 2.719, "step": 42528 }, { "epoch": 2.6400769756036997, "grad_norm": 0.43378981891207163, "learning_rate": 4.320964562389379e-06, "loss": 2.677, "step": 42529 }, { "epoch": 2.6401390527034576, "grad_norm": 0.40558474430590974, "learning_rate": 4.319496028305481e-06, "loss": 2.6436, "step": 42530 }, { "epoch": 2.6402011298032155, "grad_norm": 0.4327032894794993, "learning_rate": 4.318027732547847e-06, "loss": 2.6607, "step": 42531 }, { "epoch": 2.6402632069029734, "grad_norm": 0.4200293115196909, "learning_rate": 4.3165596751241655e-06, "loss": 2.7069, "step": 42532 }, { "epoch": 2.6403252840027314, "grad_norm": 0.4110205852801631, "learning_rate": 4.315091856042086e-06, "loss": 2.6735, "step": 42533 }, { "epoch": 2.6403873611024893, "grad_norm": 0.39684820159787226, "learning_rate": 4.313624275309258e-06, "loss": 2.614, "step": 42534 }, { "epoch": 2.640449438202247, "grad_norm": 0.41972777737902756, "learning_rate": 4.312156932933342e-06, "loss": 2.7099, "step": 42535 }, { "epoch": 2.640511515302005, "grad_norm": 0.4412288309211858, "learning_rate": 4.310689828921988e-06, "loss": 2.6649, "step": 42536 }, { "epoch": 2.640573592401763, "grad_norm": 0.44071071819059154, "learning_rate": 4.309222963282866e-06, "loss": 2.78, "step": 42537 }, { "epoch": 2.640635669501521, "grad_norm": 0.40824287622547173, "learning_rate": 4.307756336023622e-06, "loss": 2.6628, "step": 42538 }, { "epoch": 2.640697746601279, "grad_norm": 0.4120451890759707, "learning_rate": 4.306289947151898e-06, "loss": 2.7281, "step": 42539 }, { "epoch": 2.640759823701037, "grad_norm": 0.40810957206202253, "learning_rate": 4.3048237966753506e-06, "loss": 2.706, "step": 42540 }, { "epoch": 2.6408219008007947, "grad_norm": 0.380215516305596, "learning_rate": 4.303357884601633e-06, "loss": 2.6792, "step": 42541 }, { "epoch": 2.6408839779005526, "grad_norm": 0.44337377650864035, "learning_rate": 4.301892210938396e-06, "loss": 2.772, "step": 42542 }, { "epoch": 2.6409460550003105, "grad_norm": 0.4088592913685087, "learning_rate": 4.3004267756932784e-06, "loss": 2.6961, "step": 42543 }, { "epoch": 2.641008132100068, "grad_norm": 0.4212766218404467, "learning_rate": 4.298961578873928e-06, "loss": 2.6969, "step": 42544 }, { "epoch": 2.6410702091998264, "grad_norm": 0.41812139914406177, "learning_rate": 4.297496620487978e-06, "loss": 2.6532, "step": 42545 }, { "epoch": 2.641132286299584, "grad_norm": 0.40813029246381866, "learning_rate": 4.296031900543096e-06, "loss": 2.5818, "step": 42546 }, { "epoch": 2.641194363399342, "grad_norm": 0.4035372393156307, "learning_rate": 4.294567419046902e-06, "loss": 2.6439, "step": 42547 }, { "epoch": 2.6412564404990997, "grad_norm": 0.41270902135189247, "learning_rate": 4.293103176007052e-06, "loss": 2.6233, "step": 42548 }, { "epoch": 2.641318517598858, "grad_norm": 0.3826513924529868, "learning_rate": 4.2916391714311775e-06, "loss": 2.6441, "step": 42549 }, { "epoch": 2.6413805946986155, "grad_norm": 0.3969778363217148, "learning_rate": 4.290175405326907e-06, "loss": 2.6198, "step": 42550 }, { "epoch": 2.6414426717983734, "grad_norm": 0.43869487196036216, "learning_rate": 4.2887118777019005e-06, "loss": 2.6537, "step": 42551 }, { "epoch": 2.6415047488981314, "grad_norm": 0.4336597289327738, "learning_rate": 4.28724858856378e-06, "loss": 2.6857, "step": 42552 }, { "epoch": 2.6415668259978893, "grad_norm": 0.4045815133892944, "learning_rate": 4.285785537920184e-06, "loss": 2.6597, "step": 42553 }, { "epoch": 2.641628903097647, "grad_norm": 0.40103900792945624, "learning_rate": 4.284322725778727e-06, "loss": 2.6706, "step": 42554 }, { "epoch": 2.641690980197405, "grad_norm": 0.44238975132096137, "learning_rate": 4.2828601521470724e-06, "loss": 2.6357, "step": 42555 }, { "epoch": 2.641753057297163, "grad_norm": 0.4208071757990673, "learning_rate": 4.281397817032829e-06, "loss": 2.6965, "step": 42556 }, { "epoch": 2.641815134396921, "grad_norm": 0.4495990304560576, "learning_rate": 4.279935720443634e-06, "loss": 2.6542, "step": 42557 }, { "epoch": 2.641877211496679, "grad_norm": 0.42947387586886515, "learning_rate": 4.278473862387111e-06, "loss": 2.755, "step": 42558 }, { "epoch": 2.641939288596437, "grad_norm": 0.4035281922561243, "learning_rate": 4.277012242870881e-06, "loss": 2.7539, "step": 42559 }, { "epoch": 2.6420013656961947, "grad_norm": 0.4144294300550646, "learning_rate": 4.275550861902589e-06, "loss": 2.7001, "step": 42560 }, { "epoch": 2.6420634427959526, "grad_norm": 0.42000234852997065, "learning_rate": 4.274089719489843e-06, "loss": 2.7672, "step": 42561 }, { "epoch": 2.6421255198957105, "grad_norm": 0.40330246708070655, "learning_rate": 4.2726288156402786e-06, "loss": 2.6806, "step": 42562 }, { "epoch": 2.6421875969954685, "grad_norm": 0.3804054881302683, "learning_rate": 4.271168150361499e-06, "loss": 2.6571, "step": 42563 }, { "epoch": 2.6422496740952264, "grad_norm": 0.3849086979568832, "learning_rate": 4.269707723661142e-06, "loss": 2.6509, "step": 42564 }, { "epoch": 2.6423117511949843, "grad_norm": 0.46117151106881926, "learning_rate": 4.268247535546821e-06, "loss": 2.6945, "step": 42565 }, { "epoch": 2.642373828294742, "grad_norm": 0.4095073107578019, "learning_rate": 4.2667875860261556e-06, "loss": 2.6685, "step": 42566 }, { "epoch": 2.6424359053945, "grad_norm": 0.3815092167059696, "learning_rate": 4.265327875106756e-06, "loss": 2.6668, "step": 42567 }, { "epoch": 2.642497982494258, "grad_norm": 0.3980948339738438, "learning_rate": 4.26386840279625e-06, "loss": 2.7281, "step": 42568 }, { "epoch": 2.6425600595940155, "grad_norm": 0.4009088837244717, "learning_rate": 4.262409169102244e-06, "loss": 2.7933, "step": 42569 }, { "epoch": 2.642622136693774, "grad_norm": 0.4085394828747099, "learning_rate": 4.260950174032352e-06, "loss": 2.7317, "step": 42570 }, { "epoch": 2.6426842137935314, "grad_norm": 0.3638380152607434, "learning_rate": 4.259491417594191e-06, "loss": 2.6746, "step": 42571 }, { "epoch": 2.6427462908932897, "grad_norm": 0.38518027410579947, "learning_rate": 4.258032899795355e-06, "loss": 2.8047, "step": 42572 }, { "epoch": 2.642808367993047, "grad_norm": 0.4079266574728457, "learning_rate": 4.256574620643472e-06, "loss": 2.6972, "step": 42573 }, { "epoch": 2.642870445092805, "grad_norm": 0.4067485574475143, "learning_rate": 4.255116580146151e-06, "loss": 2.6375, "step": 42574 }, { "epoch": 2.642932522192563, "grad_norm": 0.38416603735143345, "learning_rate": 4.253658778310987e-06, "loss": 2.6478, "step": 42575 }, { "epoch": 2.642994599292321, "grad_norm": 0.40279213447671225, "learning_rate": 4.252201215145579e-06, "loss": 2.6972, "step": 42576 }, { "epoch": 2.643056676392079, "grad_norm": 0.4094667361027323, "learning_rate": 4.250743890657555e-06, "loss": 2.6832, "step": 42577 }, { "epoch": 2.643118753491837, "grad_norm": 0.39469241636812985, "learning_rate": 4.249286804854507e-06, "loss": 2.641, "step": 42578 }, { "epoch": 2.6431808305915947, "grad_norm": 0.37545869509806146, "learning_rate": 4.2478299577440375e-06, "loss": 2.5294, "step": 42579 }, { "epoch": 2.6432429076913526, "grad_norm": 0.38870737498347857, "learning_rate": 4.246373349333732e-06, "loss": 2.6544, "step": 42580 }, { "epoch": 2.6433049847911105, "grad_norm": 0.38752027735415534, "learning_rate": 4.244916979631209e-06, "loss": 2.6844, "step": 42581 }, { "epoch": 2.6433670618908685, "grad_norm": 0.39146632192289127, "learning_rate": 4.243460848644066e-06, "loss": 2.6264, "step": 42582 }, { "epoch": 2.6434291389906264, "grad_norm": 0.42066529581188655, "learning_rate": 4.242004956379897e-06, "loss": 2.6802, "step": 42583 }, { "epoch": 2.6434912160903843, "grad_norm": 0.3836935764172728, "learning_rate": 4.2405493028462975e-06, "loss": 2.6379, "step": 42584 }, { "epoch": 2.643553293190142, "grad_norm": 0.38720085456607045, "learning_rate": 4.2390938880508606e-06, "loss": 2.6857, "step": 42585 }, { "epoch": 2.6436153702899, "grad_norm": 0.3682761186218778, "learning_rate": 4.237638712001169e-06, "loss": 2.7057, "step": 42586 }, { "epoch": 2.643677447389658, "grad_norm": 0.42004727997214186, "learning_rate": 4.236183774704833e-06, "loss": 2.6528, "step": 42587 }, { "epoch": 2.643739524489416, "grad_norm": 0.3906042284211218, "learning_rate": 4.234729076169442e-06, "loss": 2.6812, "step": 42588 }, { "epoch": 2.643801601589174, "grad_norm": 0.39917934768233754, "learning_rate": 4.233274616402572e-06, "loss": 2.725, "step": 42589 }, { "epoch": 2.643863678688932, "grad_norm": 0.4079112967305568, "learning_rate": 4.2318203954118175e-06, "loss": 2.6448, "step": 42590 }, { "epoch": 2.6439257557886897, "grad_norm": 0.4008962276679951, "learning_rate": 4.230366413204767e-06, "loss": 2.6083, "step": 42591 }, { "epoch": 2.643987832888447, "grad_norm": 0.4043599876494391, "learning_rate": 4.228912669789015e-06, "loss": 2.736, "step": 42592 }, { "epoch": 2.6440499099882055, "grad_norm": 0.3862836682975574, "learning_rate": 4.227459165172132e-06, "loss": 2.689, "step": 42593 }, { "epoch": 2.644111987087963, "grad_norm": 0.45024811937679676, "learning_rate": 4.226005899361701e-06, "loss": 2.714, "step": 42594 }, { "epoch": 2.6441740641877214, "grad_norm": 0.42015519868003476, "learning_rate": 4.224552872365306e-06, "loss": 2.7064, "step": 42595 }, { "epoch": 2.644236141287479, "grad_norm": 0.3767683139442933, "learning_rate": 4.22310008419054e-06, "loss": 2.687, "step": 42596 }, { "epoch": 2.644298218387237, "grad_norm": 0.387598789604608, "learning_rate": 4.221647534844975e-06, "loss": 2.5871, "step": 42597 }, { "epoch": 2.6443602954869947, "grad_norm": 0.4132231175690144, "learning_rate": 4.220195224336188e-06, "loss": 2.6672, "step": 42598 }, { "epoch": 2.6444223725867526, "grad_norm": 0.4197598699291977, "learning_rate": 4.218743152671744e-06, "loss": 2.6736, "step": 42599 }, { "epoch": 2.6444844496865105, "grad_norm": 0.4031157263347082, "learning_rate": 4.21729131985924e-06, "loss": 2.6431, "step": 42600 }, { "epoch": 2.6445465267862684, "grad_norm": 0.41269075144726547, "learning_rate": 4.21583972590624e-06, "loss": 2.6767, "step": 42601 }, { "epoch": 2.6446086038860264, "grad_norm": 0.415083849117329, "learning_rate": 4.2143883708203215e-06, "loss": 2.7653, "step": 42602 }, { "epoch": 2.6446706809857843, "grad_norm": 0.4527814154423555, "learning_rate": 4.212937254609045e-06, "loss": 2.6548, "step": 42603 }, { "epoch": 2.644732758085542, "grad_norm": 0.461825695012993, "learning_rate": 4.21148637728e-06, "loss": 2.7112, "step": 42604 }, { "epoch": 2.6447948351853, "grad_norm": 0.37146288932977317, "learning_rate": 4.210035738840746e-06, "loss": 2.5979, "step": 42605 }, { "epoch": 2.644856912285058, "grad_norm": 0.39937567118318185, "learning_rate": 4.208585339298849e-06, "loss": 2.5643, "step": 42606 }, { "epoch": 2.644918989384816, "grad_norm": 0.41338721577209314, "learning_rate": 4.207135178661875e-06, "loss": 2.6445, "step": 42607 }, { "epoch": 2.644981066484574, "grad_norm": 0.40917349653636137, "learning_rate": 4.205685256937392e-06, "loss": 2.7217, "step": 42608 }, { "epoch": 2.645043143584332, "grad_norm": 0.40721735955320704, "learning_rate": 4.20423557413297e-06, "loss": 2.6698, "step": 42609 }, { "epoch": 2.6451052206840897, "grad_norm": 0.4105962482239644, "learning_rate": 4.20278613025617e-06, "loss": 2.6676, "step": 42610 }, { "epoch": 2.6451672977838476, "grad_norm": 0.3960710823750927, "learning_rate": 4.201336925314553e-06, "loss": 2.6671, "step": 42611 }, { "epoch": 2.6452293748836055, "grad_norm": 0.36631157064588454, "learning_rate": 4.199887959315668e-06, "loss": 2.7357, "step": 42612 }, { "epoch": 2.6452914519833635, "grad_norm": 0.3872027018089748, "learning_rate": 4.198439232267088e-06, "loss": 2.6361, "step": 42613 }, { "epoch": 2.6453535290831214, "grad_norm": 0.43188329926064484, "learning_rate": 4.196990744176377e-06, "loss": 2.6805, "step": 42614 }, { "epoch": 2.6454156061828793, "grad_norm": 0.41812405419349613, "learning_rate": 4.195542495051086e-06, "loss": 2.696, "step": 42615 }, { "epoch": 2.645477683282637, "grad_norm": 0.3828709032310927, "learning_rate": 4.19409448489877e-06, "loss": 2.6489, "step": 42616 }, { "epoch": 2.6455397603823947, "grad_norm": 0.4127901249091126, "learning_rate": 4.192646713726972e-06, "loss": 2.6626, "step": 42617 }, { "epoch": 2.645601837482153, "grad_norm": 0.3949055737188293, "learning_rate": 4.191199181543265e-06, "loss": 2.7022, "step": 42618 }, { "epoch": 2.6456639145819105, "grad_norm": 0.40213824088317784, "learning_rate": 4.189751888355192e-06, "loss": 2.5821, "step": 42619 }, { "epoch": 2.645725991681669, "grad_norm": 0.3988130333087884, "learning_rate": 4.188304834170303e-06, "loss": 2.6833, "step": 42620 }, { "epoch": 2.6457880687814264, "grad_norm": 0.38654015010037174, "learning_rate": 4.186858018996154e-06, "loss": 2.6615, "step": 42621 }, { "epoch": 2.6458501458811843, "grad_norm": 0.3929086119667952, "learning_rate": 4.185411442840276e-06, "loss": 2.7422, "step": 42622 }, { "epoch": 2.645912222980942, "grad_norm": 0.3635522795314935, "learning_rate": 4.1839651057102355e-06, "loss": 2.6342, "step": 42623 }, { "epoch": 2.6459743000807, "grad_norm": 0.3806231230578317, "learning_rate": 4.182519007613572e-06, "loss": 2.8021, "step": 42624 }, { "epoch": 2.646036377180458, "grad_norm": 0.3968415929596156, "learning_rate": 4.181073148557829e-06, "loss": 2.6122, "step": 42625 }, { "epoch": 2.646098454280216, "grad_norm": 0.3954038695337069, "learning_rate": 4.179627528550545e-06, "loss": 2.6684, "step": 42626 }, { "epoch": 2.646160531379974, "grad_norm": 0.4054490409867958, "learning_rate": 4.178182147599275e-06, "loss": 2.6547, "step": 42627 }, { "epoch": 2.646222608479732, "grad_norm": 0.4072358567162927, "learning_rate": 4.1767370057115505e-06, "loss": 2.6201, "step": 42628 }, { "epoch": 2.6462846855794897, "grad_norm": 0.4004315425155165, "learning_rate": 4.175292102894912e-06, "loss": 2.6988, "step": 42629 }, { "epoch": 2.6463467626792476, "grad_norm": 0.39092721692289945, "learning_rate": 4.173847439156903e-06, "loss": 2.6542, "step": 42630 }, { "epoch": 2.6464088397790055, "grad_norm": 0.39884391327486773, "learning_rate": 4.1724030145050444e-06, "loss": 2.5686, "step": 42631 }, { "epoch": 2.6464709168787635, "grad_norm": 0.3655577059708788, "learning_rate": 4.170958828946897e-06, "loss": 2.699, "step": 42632 }, { "epoch": 2.6465329939785214, "grad_norm": 0.4115416426359142, "learning_rate": 4.169514882489977e-06, "loss": 2.7283, "step": 42633 }, { "epoch": 2.6465950710782793, "grad_norm": 0.39519799424909574, "learning_rate": 4.16807117514183e-06, "loss": 2.6967, "step": 42634 }, { "epoch": 2.646657148178037, "grad_norm": 0.3782830767014276, "learning_rate": 4.1666277069099704e-06, "loss": 2.6621, "step": 42635 }, { "epoch": 2.646719225277795, "grad_norm": 0.3857066270879038, "learning_rate": 4.165184477801948e-06, "loss": 2.6424, "step": 42636 }, { "epoch": 2.646781302377553, "grad_norm": 0.3829636956406262, "learning_rate": 4.163741487825284e-06, "loss": 2.5858, "step": 42637 }, { "epoch": 2.646843379477311, "grad_norm": 0.40725854249522286, "learning_rate": 4.162298736987513e-06, "loss": 2.659, "step": 42638 }, { "epoch": 2.646905456577069, "grad_norm": 0.39466404332449495, "learning_rate": 4.160856225296145e-06, "loss": 2.6533, "step": 42639 }, { "epoch": 2.6469675336768264, "grad_norm": 0.40062436980080396, "learning_rate": 4.159413952758723e-06, "loss": 2.6727, "step": 42640 }, { "epoch": 2.6470296107765847, "grad_norm": 0.38450602182452315, "learning_rate": 4.157971919382769e-06, "loss": 2.6777, "step": 42641 }, { "epoch": 2.647091687876342, "grad_norm": 0.4142643598428498, "learning_rate": 4.156530125175806e-06, "loss": 2.688, "step": 42642 }, { "epoch": 2.6471537649761006, "grad_norm": 0.38021857321898045, "learning_rate": 4.155088570145354e-06, "loss": 2.6325, "step": 42643 }, { "epoch": 2.647215842075858, "grad_norm": 0.3846431091751526, "learning_rate": 4.1536472542989256e-06, "loss": 2.7065, "step": 42644 }, { "epoch": 2.6472779191756164, "grad_norm": 0.4209174131997169, "learning_rate": 4.152206177644047e-06, "loss": 2.7078, "step": 42645 }, { "epoch": 2.647339996275374, "grad_norm": 0.4030459304269073, "learning_rate": 4.150765340188251e-06, "loss": 2.7008, "step": 42646 }, { "epoch": 2.647402073375132, "grad_norm": 0.38626727327901156, "learning_rate": 4.149324741939037e-06, "loss": 2.724, "step": 42647 }, { "epoch": 2.6474641504748897, "grad_norm": 0.37654589857806436, "learning_rate": 4.147884382903932e-06, "loss": 2.6467, "step": 42648 }, { "epoch": 2.6475262275746476, "grad_norm": 0.38974890692375913, "learning_rate": 4.146444263090432e-06, "loss": 2.6187, "step": 42649 }, { "epoch": 2.6475883046744055, "grad_norm": 0.394308858454673, "learning_rate": 4.1450043825060735e-06, "loss": 2.6532, "step": 42650 }, { "epoch": 2.6476503817741635, "grad_norm": 0.38620903640484433, "learning_rate": 4.1435647411583625e-06, "loss": 2.6983, "step": 42651 }, { "epoch": 2.6477124588739214, "grad_norm": 0.40290269636467146, "learning_rate": 4.142125339054803e-06, "loss": 2.725, "step": 42652 }, { "epoch": 2.6477745359736793, "grad_norm": 0.41412126376063946, "learning_rate": 4.1406861762029015e-06, "loss": 2.6965, "step": 42653 }, { "epoch": 2.647836613073437, "grad_norm": 0.42995411664103483, "learning_rate": 4.139247252610179e-06, "loss": 2.7283, "step": 42654 }, { "epoch": 2.647898690173195, "grad_norm": 0.38896084112262364, "learning_rate": 4.137808568284135e-06, "loss": 2.7002, "step": 42655 }, { "epoch": 2.647960767272953, "grad_norm": 0.403747775423045, "learning_rate": 4.13637012323228e-06, "loss": 2.5986, "step": 42656 }, { "epoch": 2.648022844372711, "grad_norm": 0.4170232935432668, "learning_rate": 4.1349319174621205e-06, "loss": 2.675, "step": 42657 }, { "epoch": 2.648084921472469, "grad_norm": 0.39465179112684134, "learning_rate": 4.1334939509811375e-06, "loss": 2.6296, "step": 42658 }, { "epoch": 2.648146998572227, "grad_norm": 0.3933735849641721, "learning_rate": 4.132056223796865e-06, "loss": 2.5828, "step": 42659 }, { "epoch": 2.6482090756719847, "grad_norm": 0.377004748352463, "learning_rate": 4.130618735916791e-06, "loss": 2.5643, "step": 42660 }, { "epoch": 2.6482711527717426, "grad_norm": 0.39900470079076517, "learning_rate": 4.1291814873484106e-06, "loss": 2.6926, "step": 42661 }, { "epoch": 2.6483332298715005, "grad_norm": 0.4257646259947253, "learning_rate": 4.1277444780992215e-06, "loss": 2.6788, "step": 42662 }, { "epoch": 2.6483953069712585, "grad_norm": 0.39378675713104117, "learning_rate": 4.12630770817673e-06, "loss": 2.5792, "step": 42663 }, { "epoch": 2.6484573840710164, "grad_norm": 0.3872678499694652, "learning_rate": 4.12487117758843e-06, "loss": 2.5841, "step": 42664 }, { "epoch": 2.648519461170774, "grad_norm": 0.38185909815642866, "learning_rate": 4.123434886341809e-06, "loss": 2.6718, "step": 42665 }, { "epoch": 2.648581538270532, "grad_norm": 0.4054594049865236, "learning_rate": 4.121998834444363e-06, "loss": 2.7659, "step": 42666 }, { "epoch": 2.6486436153702897, "grad_norm": 0.4135753176695941, "learning_rate": 4.12056302190359e-06, "loss": 2.7186, "step": 42667 }, { "epoch": 2.648705692470048, "grad_norm": 0.3903534326706094, "learning_rate": 4.1191274487269784e-06, "loss": 2.623, "step": 42668 }, { "epoch": 2.6487677695698055, "grad_norm": 0.3840035186372918, "learning_rate": 4.117692114922017e-06, "loss": 2.6269, "step": 42669 }, { "epoch": 2.6488298466695634, "grad_norm": 0.3632953394957093, "learning_rate": 4.116257020496195e-06, "loss": 2.6284, "step": 42670 }, { "epoch": 2.6488919237693214, "grad_norm": 0.39323273924495206, "learning_rate": 4.114822165456994e-06, "loss": 2.6604, "step": 42671 }, { "epoch": 2.6489540008690793, "grad_norm": 0.398868178430921, "learning_rate": 4.113387549811909e-06, "loss": 2.5716, "step": 42672 }, { "epoch": 2.649016077968837, "grad_norm": 0.38388347677736245, "learning_rate": 4.111953173568417e-06, "loss": 2.6754, "step": 42673 }, { "epoch": 2.649078155068595, "grad_norm": 0.3797453016074607, "learning_rate": 4.1105190367340115e-06, "loss": 2.6545, "step": 42674 }, { "epoch": 2.649140232168353, "grad_norm": 0.379505313129952, "learning_rate": 4.109085139316155e-06, "loss": 2.6627, "step": 42675 }, { "epoch": 2.649202309268111, "grad_norm": 0.38055288257998227, "learning_rate": 4.1076514813223555e-06, "loss": 2.7024, "step": 42676 }, { "epoch": 2.649264386367869, "grad_norm": 0.42018163763924704, "learning_rate": 4.106218062760075e-06, "loss": 2.6376, "step": 42677 }, { "epoch": 2.649326463467627, "grad_norm": 0.38223337626438814, "learning_rate": 4.104784883636786e-06, "loss": 2.6946, "step": 42678 }, { "epoch": 2.6493885405673847, "grad_norm": 0.3992854613163929, "learning_rate": 4.1033519439599876e-06, "loss": 2.6701, "step": 42679 }, { "epoch": 2.6494506176671426, "grad_norm": 0.3644880124777844, "learning_rate": 4.101919243737146e-06, "loss": 2.6047, "step": 42680 }, { "epoch": 2.6495126947669005, "grad_norm": 0.3891126730559719, "learning_rate": 4.100486782975721e-06, "loss": 2.6564, "step": 42681 }, { "epoch": 2.6495747718666585, "grad_norm": 0.37142425676409985, "learning_rate": 4.0990545616832075e-06, "loss": 2.6761, "step": 42682 }, { "epoch": 2.6496368489664164, "grad_norm": 0.38830980852128494, "learning_rate": 4.097622579867072e-06, "loss": 2.5942, "step": 42683 }, { "epoch": 2.6496989260661743, "grad_norm": 0.41455543614302603, "learning_rate": 4.0961908375347806e-06, "loss": 2.7262, "step": 42684 }, { "epoch": 2.649761003165932, "grad_norm": 0.42426693960504624, "learning_rate": 4.094759334693793e-06, "loss": 2.6922, "step": 42685 }, { "epoch": 2.64982308026569, "grad_norm": 0.41260520431566566, "learning_rate": 4.093328071351604e-06, "loss": 2.7376, "step": 42686 }, { "epoch": 2.649885157365448, "grad_norm": 0.39031770062856885, "learning_rate": 4.091897047515664e-06, "loss": 2.5391, "step": 42687 }, { "epoch": 2.6499472344652055, "grad_norm": 0.35817185393337825, "learning_rate": 4.090466263193443e-06, "loss": 2.7028, "step": 42688 }, { "epoch": 2.650009311564964, "grad_norm": 0.40889999918374437, "learning_rate": 4.089035718392398e-06, "loss": 2.6891, "step": 42689 }, { "epoch": 2.6500713886647214, "grad_norm": 0.4231596566549016, "learning_rate": 4.087605413120005e-06, "loss": 2.6073, "step": 42690 }, { "epoch": 2.6501334657644797, "grad_norm": 0.41788214383421884, "learning_rate": 4.086175347383719e-06, "loss": 2.707, "step": 42691 }, { "epoch": 2.650195542864237, "grad_norm": 0.40501805727345963, "learning_rate": 4.084745521191002e-06, "loss": 2.5546, "step": 42692 }, { "epoch": 2.6502576199639956, "grad_norm": 0.36782657525057055, "learning_rate": 4.08331593454932e-06, "loss": 2.7072, "step": 42693 }, { "epoch": 2.650319697063753, "grad_norm": 0.39289292955888927, "learning_rate": 4.081886587466111e-06, "loss": 2.6389, "step": 42694 }, { "epoch": 2.650381774163511, "grad_norm": 0.4059935457216915, "learning_rate": 4.080457479948857e-06, "loss": 2.5776, "step": 42695 }, { "epoch": 2.650443851263269, "grad_norm": 0.3985425025358207, "learning_rate": 4.079028612005004e-06, "loss": 2.6732, "step": 42696 }, { "epoch": 2.650505928363027, "grad_norm": 0.3905532007989006, "learning_rate": 4.077599983642005e-06, "loss": 2.7454, "step": 42697 }, { "epoch": 2.6505680054627847, "grad_norm": 0.3957532897964298, "learning_rate": 4.076171594867306e-06, "loss": 2.6854, "step": 42698 }, { "epoch": 2.6506300825625426, "grad_norm": 0.39660605080990885, "learning_rate": 4.074743445688378e-06, "loss": 2.6589, "step": 42699 }, { "epoch": 2.6506921596623005, "grad_norm": 0.39958899661371083, "learning_rate": 4.07331553611266e-06, "loss": 2.697, "step": 42700 }, { "epoch": 2.6507542367620585, "grad_norm": 0.4092938939768912, "learning_rate": 4.071887866147605e-06, "loss": 2.6515, "step": 42701 }, { "epoch": 2.6508163138618164, "grad_norm": 0.3986227128629244, "learning_rate": 4.0704604358006545e-06, "loss": 2.5697, "step": 42702 }, { "epoch": 2.6508783909615743, "grad_norm": 0.40145230840897544, "learning_rate": 4.069033245079268e-06, "loss": 2.6145, "step": 42703 }, { "epoch": 2.650940468061332, "grad_norm": 0.387352087647453, "learning_rate": 4.0676062939908885e-06, "loss": 2.7188, "step": 42704 }, { "epoch": 2.65100254516109, "grad_norm": 0.3793765227021731, "learning_rate": 4.066179582542956e-06, "loss": 2.6649, "step": 42705 }, { "epoch": 2.651064622260848, "grad_norm": 0.3796186141330188, "learning_rate": 4.064753110742914e-06, "loss": 2.5878, "step": 42706 }, { "epoch": 2.651126699360606, "grad_norm": 0.3710272277637145, "learning_rate": 4.063326878598195e-06, "loss": 2.6727, "step": 42707 }, { "epoch": 2.651188776460364, "grad_norm": 0.40027058360337986, "learning_rate": 4.0619008861162655e-06, "loss": 2.5953, "step": 42708 }, { "epoch": 2.651250853560122, "grad_norm": 0.4259413147081708, "learning_rate": 4.060475133304548e-06, "loss": 2.6564, "step": 42709 }, { "epoch": 2.6513129306598797, "grad_norm": 0.38743459549498377, "learning_rate": 4.059049620170479e-06, "loss": 2.5824, "step": 42710 }, { "epoch": 2.651375007759637, "grad_norm": 0.41339186574410375, "learning_rate": 4.0576243467215e-06, "loss": 2.6789, "step": 42711 }, { "epoch": 2.6514370848593956, "grad_norm": 0.37706130456948345, "learning_rate": 4.056199312965042e-06, "loss": 2.5886, "step": 42712 }, { "epoch": 2.651499161959153, "grad_norm": 0.40353923333412023, "learning_rate": 4.05477451890856e-06, "loss": 2.6568, "step": 42713 }, { "epoch": 2.6515612390589114, "grad_norm": 0.4200521932256105, "learning_rate": 4.053349964559466e-06, "loss": 2.6473, "step": 42714 }, { "epoch": 2.651623316158669, "grad_norm": 0.3987832159001823, "learning_rate": 4.051925649925203e-06, "loss": 2.7203, "step": 42715 }, { "epoch": 2.6516853932584272, "grad_norm": 0.3819863103790319, "learning_rate": 4.050501575013193e-06, "loss": 2.6732, "step": 42716 }, { "epoch": 2.6517474703581847, "grad_norm": 0.38546752808883217, "learning_rate": 4.0490777398308755e-06, "loss": 2.6567, "step": 42717 }, { "epoch": 2.6518095474579426, "grad_norm": 0.39522210621835824, "learning_rate": 4.047654144385676e-06, "loss": 2.7173, "step": 42718 }, { "epoch": 2.6518716245577005, "grad_norm": 0.38733462219439674, "learning_rate": 4.046230788685018e-06, "loss": 2.7194, "step": 42719 }, { "epoch": 2.6519337016574585, "grad_norm": 0.41292419152792426, "learning_rate": 4.044807672736328e-06, "loss": 2.6518, "step": 42720 }, { "epoch": 2.6519957787572164, "grad_norm": 0.4284647623649614, "learning_rate": 4.043384796547029e-06, "loss": 2.7807, "step": 42721 }, { "epoch": 2.6520578558569743, "grad_norm": 0.412121921727975, "learning_rate": 4.0419621601245515e-06, "loss": 2.6734, "step": 42722 }, { "epoch": 2.652119932956732, "grad_norm": 0.37770268919792627, "learning_rate": 4.04053976347632e-06, "loss": 2.6855, "step": 42723 }, { "epoch": 2.65218201005649, "grad_norm": 0.4102365142760675, "learning_rate": 4.039117606609744e-06, "loss": 2.7019, "step": 42724 }, { "epoch": 2.652244087156248, "grad_norm": 0.3750053445117298, "learning_rate": 4.037695689532245e-06, "loss": 2.6359, "step": 42725 }, { "epoch": 2.652306164256006, "grad_norm": 0.4051142000561418, "learning_rate": 4.036274012251251e-06, "loss": 2.6582, "step": 42726 }, { "epoch": 2.652368241355764, "grad_norm": 0.4269962648478503, "learning_rate": 4.034852574774173e-06, "loss": 2.5939, "step": 42727 }, { "epoch": 2.652430318455522, "grad_norm": 0.3909494959563438, "learning_rate": 4.033431377108426e-06, "loss": 2.6929, "step": 42728 }, { "epoch": 2.6524923955552797, "grad_norm": 0.3772999052871618, "learning_rate": 4.0320104192614286e-06, "loss": 2.7156, "step": 42729 }, { "epoch": 2.6525544726550376, "grad_norm": 0.4153674453429473, "learning_rate": 4.030589701240583e-06, "loss": 2.5486, "step": 42730 }, { "epoch": 2.6526165497547955, "grad_norm": 0.3889533514096415, "learning_rate": 4.029169223053319e-06, "loss": 2.7106, "step": 42731 }, { "epoch": 2.6526786268545535, "grad_norm": 0.4112092805548024, "learning_rate": 4.0277489847070415e-06, "loss": 2.6512, "step": 42732 }, { "epoch": 2.6527407039543114, "grad_norm": 0.38196557617491794, "learning_rate": 4.026328986209154e-06, "loss": 2.6161, "step": 42733 }, { "epoch": 2.6528027810540693, "grad_norm": 0.3771103644649469, "learning_rate": 4.024909227567064e-06, "loss": 2.6703, "step": 42734 }, { "epoch": 2.652864858153827, "grad_norm": 0.38346538532463686, "learning_rate": 4.023489708788186e-06, "loss": 2.6595, "step": 42735 }, { "epoch": 2.6529269352535847, "grad_norm": 0.3824716724552459, "learning_rate": 4.022070429879932e-06, "loss": 2.6635, "step": 42736 }, { "epoch": 2.652989012353343, "grad_norm": 0.3915659443550812, "learning_rate": 4.020651390849689e-06, "loss": 2.6744, "step": 42737 }, { "epoch": 2.6530510894531005, "grad_norm": 0.3858283224391513, "learning_rate": 4.019232591704869e-06, "loss": 2.6398, "step": 42738 }, { "epoch": 2.653113166552859, "grad_norm": 0.3662005003395496, "learning_rate": 4.017814032452877e-06, "loss": 2.6967, "step": 42739 }, { "epoch": 2.6531752436526164, "grad_norm": 0.38410622108107173, "learning_rate": 4.016395713101112e-06, "loss": 2.7017, "step": 42740 }, { "epoch": 2.6532373207523743, "grad_norm": 0.4084534675961819, "learning_rate": 4.01497763365698e-06, "loss": 2.6597, "step": 42741 }, { "epoch": 2.653299397852132, "grad_norm": 0.3857604782983033, "learning_rate": 4.013559794127869e-06, "loss": 2.6118, "step": 42742 }, { "epoch": 2.65336147495189, "grad_norm": 0.3971961824382039, "learning_rate": 4.012142194521168e-06, "loss": 2.713, "step": 42743 }, { "epoch": 2.653423552051648, "grad_norm": 0.38041908049609735, "learning_rate": 4.010724834844287e-06, "loss": 2.7158, "step": 42744 }, { "epoch": 2.653485629151406, "grad_norm": 0.38779548100484124, "learning_rate": 4.009307715104632e-06, "loss": 2.6237, "step": 42745 }, { "epoch": 2.653547706251164, "grad_norm": 0.42192730755955604, "learning_rate": 4.0078908353095805e-06, "loss": 2.704, "step": 42746 }, { "epoch": 2.653609783350922, "grad_norm": 0.40598378261935053, "learning_rate": 4.006474195466526e-06, "loss": 2.7388, "step": 42747 }, { "epoch": 2.6536718604506797, "grad_norm": 0.3888736756883742, "learning_rate": 4.0050577955828505e-06, "loss": 2.7955, "step": 42748 }, { "epoch": 2.6537339375504376, "grad_norm": 0.3731283528290721, "learning_rate": 4.0036416356659666e-06, "loss": 2.5711, "step": 42749 }, { "epoch": 2.6537960146501955, "grad_norm": 0.37570908184568524, "learning_rate": 4.0022257157232516e-06, "loss": 2.634, "step": 42750 }, { "epoch": 2.6538580917499535, "grad_norm": 0.3735031413401978, "learning_rate": 4.000810035762087e-06, "loss": 2.6378, "step": 42751 }, { "epoch": 2.6539201688497114, "grad_norm": 0.39622376243777846, "learning_rate": 3.999394595789857e-06, "loss": 2.6731, "step": 42752 }, { "epoch": 2.6539822459494693, "grad_norm": 0.37268734128476205, "learning_rate": 3.997979395813961e-06, "loss": 2.5672, "step": 42753 }, { "epoch": 2.654044323049227, "grad_norm": 0.38907357165657735, "learning_rate": 3.996564435841771e-06, "loss": 2.6225, "step": 42754 }, { "epoch": 2.654106400148985, "grad_norm": 0.3852889283171862, "learning_rate": 3.995149715880675e-06, "loss": 2.6666, "step": 42755 }, { "epoch": 2.654168477248743, "grad_norm": 0.3677207240465593, "learning_rate": 3.993735235938051e-06, "loss": 2.6528, "step": 42756 }, { "epoch": 2.654230554348501, "grad_norm": 0.3789881981393596, "learning_rate": 3.99232099602127e-06, "loss": 2.6584, "step": 42757 }, { "epoch": 2.654292631448259, "grad_norm": 0.4043174038157838, "learning_rate": 3.990906996137727e-06, "loss": 2.5816, "step": 42758 }, { "epoch": 2.6543547085480164, "grad_norm": 0.3859870809297984, "learning_rate": 3.989493236294795e-06, "loss": 2.6561, "step": 42759 }, { "epoch": 2.6544167856477747, "grad_norm": 0.4113268903838754, "learning_rate": 3.988079716499843e-06, "loss": 2.6962, "step": 42760 }, { "epoch": 2.654478862747532, "grad_norm": 0.3650274863736924, "learning_rate": 3.986666436760239e-06, "loss": 2.5594, "step": 42761 }, { "epoch": 2.6545409398472906, "grad_norm": 0.38410734611633035, "learning_rate": 3.985253397083377e-06, "loss": 2.6883, "step": 42762 }, { "epoch": 2.654603016947048, "grad_norm": 0.3820518605107932, "learning_rate": 3.983840597476618e-06, "loss": 2.664, "step": 42763 }, { "epoch": 2.6546650940468064, "grad_norm": 0.4025386782537456, "learning_rate": 3.982428037947334e-06, "loss": 2.6069, "step": 42764 }, { "epoch": 2.654727171146564, "grad_norm": 0.37872042474377193, "learning_rate": 3.9810157185028895e-06, "loss": 2.6649, "step": 42765 }, { "epoch": 2.654789248246322, "grad_norm": 0.37706146264625123, "learning_rate": 3.9796036391506534e-06, "loss": 2.7371, "step": 42766 }, { "epoch": 2.6548513253460797, "grad_norm": 0.3682623410782352, "learning_rate": 3.978191799898006e-06, "loss": 2.6493, "step": 42767 }, { "epoch": 2.6549134024458376, "grad_norm": 0.37643454265240345, "learning_rate": 3.9767802007523e-06, "loss": 2.7164, "step": 42768 }, { "epoch": 2.6549754795455955, "grad_norm": 0.3760973451487714, "learning_rate": 3.9753688417209055e-06, "loss": 2.6639, "step": 42769 }, { "epoch": 2.6550375566453535, "grad_norm": 0.39474085159524164, "learning_rate": 3.973957722811178e-06, "loss": 2.6546, "step": 42770 }, { "epoch": 2.6550996337451114, "grad_norm": 0.398716753916016, "learning_rate": 3.972546844030495e-06, "loss": 2.6779, "step": 42771 }, { "epoch": 2.6551617108448693, "grad_norm": 0.39026207276681935, "learning_rate": 3.971136205386211e-06, "loss": 2.6341, "step": 42772 }, { "epoch": 2.655223787944627, "grad_norm": 0.37262120351997613, "learning_rate": 3.969725806885677e-06, "loss": 2.6899, "step": 42773 }, { "epoch": 2.655285865044385, "grad_norm": 0.3751360130015436, "learning_rate": 3.968315648536253e-06, "loss": 2.7107, "step": 42774 }, { "epoch": 2.655347942144143, "grad_norm": 0.41031140617114403, "learning_rate": 3.96690573034531e-06, "loss": 2.7047, "step": 42775 }, { "epoch": 2.655410019243901, "grad_norm": 0.39268234160169746, "learning_rate": 3.965496052320184e-06, "loss": 2.6535, "step": 42776 }, { "epoch": 2.655472096343659, "grad_norm": 0.3812055921487628, "learning_rate": 3.964086614468254e-06, "loss": 2.6663, "step": 42777 }, { "epoch": 2.655534173443417, "grad_norm": 0.4208376634803056, "learning_rate": 3.962677416796856e-06, "loss": 2.6853, "step": 42778 }, { "epoch": 2.6555962505431747, "grad_norm": 0.42147720090401297, "learning_rate": 3.961268459313344e-06, "loss": 2.652, "step": 42779 }, { "epoch": 2.6556583276429326, "grad_norm": 0.4014354564217405, "learning_rate": 3.959859742025063e-06, "loss": 2.6544, "step": 42780 }, { "epoch": 2.6557204047426906, "grad_norm": 0.36737206061538163, "learning_rate": 3.958451264939378e-06, "loss": 2.6097, "step": 42781 }, { "epoch": 2.6557824818424485, "grad_norm": 0.39836338700292073, "learning_rate": 3.9570430280636336e-06, "loss": 2.6837, "step": 42782 }, { "epoch": 2.6558445589422064, "grad_norm": 0.3760952650672815, "learning_rate": 3.955635031405169e-06, "loss": 2.644, "step": 42783 }, { "epoch": 2.655906636041964, "grad_norm": 0.36496249192356056, "learning_rate": 3.954227274971328e-06, "loss": 2.6445, "step": 42784 }, { "epoch": 2.6559687131417222, "grad_norm": 0.38581159908770996, "learning_rate": 3.952819758769471e-06, "loss": 2.699, "step": 42785 }, { "epoch": 2.6560307902414797, "grad_norm": 0.3942059223690132, "learning_rate": 3.951412482806927e-06, "loss": 2.7038, "step": 42786 }, { "epoch": 2.656092867341238, "grad_norm": 0.3954524060281642, "learning_rate": 3.950005447091049e-06, "loss": 2.7014, "step": 42787 }, { "epoch": 2.6561549444409955, "grad_norm": 0.3635334452088391, "learning_rate": 3.948598651629159e-06, "loss": 2.6179, "step": 42788 }, { "epoch": 2.6562170215407535, "grad_norm": 0.3815576656450823, "learning_rate": 3.94719209642862e-06, "loss": 2.7064, "step": 42789 }, { "epoch": 2.6562790986405114, "grad_norm": 0.35676727385961265, "learning_rate": 3.945785781496758e-06, "loss": 2.6755, "step": 42790 }, { "epoch": 2.6563411757402693, "grad_norm": 0.36934476586678305, "learning_rate": 3.944379706840912e-06, "loss": 2.7013, "step": 42791 }, { "epoch": 2.656403252840027, "grad_norm": 0.41577691406718653, "learning_rate": 3.942973872468414e-06, "loss": 2.5954, "step": 42792 }, { "epoch": 2.656465329939785, "grad_norm": 0.4081261667231989, "learning_rate": 3.9415682783865985e-06, "loss": 2.758, "step": 42793 }, { "epoch": 2.656527407039543, "grad_norm": 0.39902221559999373, "learning_rate": 3.9401629246028085e-06, "loss": 2.7355, "step": 42794 }, { "epoch": 2.656589484139301, "grad_norm": 0.3847231187917099, "learning_rate": 3.938757811124371e-06, "loss": 2.5934, "step": 42795 }, { "epoch": 2.656651561239059, "grad_norm": 0.39769571104780027, "learning_rate": 3.9373529379586165e-06, "loss": 2.6268, "step": 42796 }, { "epoch": 2.656713638338817, "grad_norm": 0.3963849924993977, "learning_rate": 3.935948305112863e-06, "loss": 2.6976, "step": 42797 }, { "epoch": 2.6567757154385747, "grad_norm": 0.4019530167500727, "learning_rate": 3.934543912594457e-06, "loss": 2.6529, "step": 42798 }, { "epoch": 2.6568377925383326, "grad_norm": 0.41742110694787465, "learning_rate": 3.93313976041072e-06, "loss": 2.7516, "step": 42799 }, { "epoch": 2.6568998696380905, "grad_norm": 0.40055343295183227, "learning_rate": 3.9317358485689796e-06, "loss": 2.6831, "step": 42800 }, { "epoch": 2.6569619467378485, "grad_norm": 0.3858876786912477, "learning_rate": 3.930332177076551e-06, "loss": 2.616, "step": 42801 }, { "epoch": 2.6570240238376064, "grad_norm": 0.43145520124245657, "learning_rate": 3.928928745940757e-06, "loss": 2.7458, "step": 42802 }, { "epoch": 2.6570861009373643, "grad_norm": 0.4360750357406261, "learning_rate": 3.927525555168937e-06, "loss": 2.7056, "step": 42803 }, { "epoch": 2.657148178037122, "grad_norm": 0.419019750282108, "learning_rate": 3.926122604768395e-06, "loss": 2.6625, "step": 42804 }, { "epoch": 2.65721025513688, "grad_norm": 0.4014587112362952, "learning_rate": 3.924719894746459e-06, "loss": 2.647, "step": 42805 }, { "epoch": 2.657272332236638, "grad_norm": 0.4158518652459148, "learning_rate": 3.92331742511044e-06, "loss": 2.5566, "step": 42806 }, { "epoch": 2.6573344093363955, "grad_norm": 0.4178146363391341, "learning_rate": 3.921915195867665e-06, "loss": 2.7072, "step": 42807 }, { "epoch": 2.657396486436154, "grad_norm": 0.3829318853009948, "learning_rate": 3.920513207025445e-06, "loss": 2.5825, "step": 42808 }, { "epoch": 2.6574585635359114, "grad_norm": 0.38608867680525305, "learning_rate": 3.919111458591079e-06, "loss": 2.6806, "step": 42809 }, { "epoch": 2.6575206406356697, "grad_norm": 0.39436569139403904, "learning_rate": 3.917709950571913e-06, "loss": 2.5844, "step": 42810 }, { "epoch": 2.657582717735427, "grad_norm": 0.4064714305135579, "learning_rate": 3.916308682975228e-06, "loss": 2.7486, "step": 42811 }, { "epoch": 2.6576447948351856, "grad_norm": 0.3821671359297351, "learning_rate": 3.914907655808353e-06, "loss": 2.7021, "step": 42812 }, { "epoch": 2.657706871934943, "grad_norm": 0.40573946344370687, "learning_rate": 3.913506869078598e-06, "loss": 2.6825, "step": 42813 }, { "epoch": 2.657768949034701, "grad_norm": 0.3778795904031522, "learning_rate": 3.912106322793263e-06, "loss": 2.5311, "step": 42814 }, { "epoch": 2.657831026134459, "grad_norm": 0.4112971468977002, "learning_rate": 3.910706016959659e-06, "loss": 2.7007, "step": 42815 }, { "epoch": 2.657893103234217, "grad_norm": 0.3574058028813119, "learning_rate": 3.90930595158508e-06, "loss": 2.7011, "step": 42816 }, { "epoch": 2.6579551803339747, "grad_norm": 0.3755716695074015, "learning_rate": 3.907906126676847e-06, "loss": 2.6586, "step": 42817 }, { "epoch": 2.6580172574337326, "grad_norm": 0.38209113449333393, "learning_rate": 3.906506542242261e-06, "loss": 2.6448, "step": 42818 }, { "epoch": 2.6580793345334905, "grad_norm": 0.4301766818813288, "learning_rate": 3.9051071982886216e-06, "loss": 2.6661, "step": 42819 }, { "epoch": 2.6581414116332485, "grad_norm": 0.3980997748595413, "learning_rate": 3.903708094823216e-06, "loss": 2.6242, "step": 42820 }, { "epoch": 2.6582034887330064, "grad_norm": 0.3989784644835492, "learning_rate": 3.902309231853363e-06, "loss": 2.6375, "step": 42821 }, { "epoch": 2.6582655658327643, "grad_norm": 0.39800455843648874, "learning_rate": 3.900910609386355e-06, "loss": 2.6188, "step": 42822 }, { "epoch": 2.658327642932522, "grad_norm": 0.37258990994846747, "learning_rate": 3.899512227429486e-06, "loss": 2.6937, "step": 42823 }, { "epoch": 2.65838972003228, "grad_norm": 0.4204319185124929, "learning_rate": 3.898114085990045e-06, "loss": 2.6496, "step": 42824 }, { "epoch": 2.658451797132038, "grad_norm": 0.38024289048149046, "learning_rate": 3.8967161850753445e-06, "loss": 2.6634, "step": 42825 }, { "epoch": 2.658513874231796, "grad_norm": 0.36614957168139395, "learning_rate": 3.895318524692665e-06, "loss": 2.637, "step": 42826 }, { "epoch": 2.658575951331554, "grad_norm": 0.4112169809333092, "learning_rate": 3.893921104849308e-06, "loss": 2.6731, "step": 42827 }, { "epoch": 2.658638028431312, "grad_norm": 0.38951957694327416, "learning_rate": 3.892523925552549e-06, "loss": 2.6383, "step": 42828 }, { "epoch": 2.6587001055310697, "grad_norm": 0.4017802801975741, "learning_rate": 3.891126986809679e-06, "loss": 2.7582, "step": 42829 }, { "epoch": 2.6587621826308276, "grad_norm": 0.39401349001146174, "learning_rate": 3.889730288628002e-06, "loss": 2.6419, "step": 42830 }, { "epoch": 2.6588242597305856, "grad_norm": 0.3966769234783092, "learning_rate": 3.888333831014801e-06, "loss": 2.6892, "step": 42831 }, { "epoch": 2.658886336830343, "grad_norm": 0.38889750923336086, "learning_rate": 3.886937613977348e-06, "loss": 2.6416, "step": 42832 }, { "epoch": 2.6589484139301014, "grad_norm": 0.40898067396178084, "learning_rate": 3.885541637522932e-06, "loss": 2.6258, "step": 42833 }, { "epoch": 2.659010491029859, "grad_norm": 0.35980238586888985, "learning_rate": 3.884145901658848e-06, "loss": 2.5454, "step": 42834 }, { "epoch": 2.6590725681296172, "grad_norm": 0.3864989185700579, "learning_rate": 3.882750406392371e-06, "loss": 2.6426, "step": 42835 }, { "epoch": 2.6591346452293747, "grad_norm": 0.38714027550909375, "learning_rate": 3.881355151730775e-06, "loss": 2.6452, "step": 42836 }, { "epoch": 2.6591967223291326, "grad_norm": 0.37771639855104916, "learning_rate": 3.879960137681343e-06, "loss": 2.7288, "step": 42837 }, { "epoch": 2.6592587994288905, "grad_norm": 0.3947782781774904, "learning_rate": 3.878565364251358e-06, "loss": 2.6759, "step": 42838 }, { "epoch": 2.6593208765286485, "grad_norm": 0.4043462787610717, "learning_rate": 3.877170831448096e-06, "loss": 2.7801, "step": 42839 }, { "epoch": 2.6593829536284064, "grad_norm": 0.3891176514046543, "learning_rate": 3.8757765392788305e-06, "loss": 2.6422, "step": 42840 }, { "epoch": 2.6594450307281643, "grad_norm": 0.3787424185595241, "learning_rate": 3.874382487750838e-06, "loss": 2.6878, "step": 42841 }, { "epoch": 2.659507107827922, "grad_norm": 0.37312627616421273, "learning_rate": 3.872988676871381e-06, "loss": 2.5842, "step": 42842 }, { "epoch": 2.65956918492768, "grad_norm": 0.35977875792217545, "learning_rate": 3.8715951066477356e-06, "loss": 2.6544, "step": 42843 }, { "epoch": 2.659631262027438, "grad_norm": 0.37444972354482925, "learning_rate": 3.87020177708719e-06, "loss": 2.6249, "step": 42844 }, { "epoch": 2.659693339127196, "grad_norm": 0.4024634137022375, "learning_rate": 3.868808688196996e-06, "loss": 2.7539, "step": 42845 }, { "epoch": 2.659755416226954, "grad_norm": 0.39580839061787376, "learning_rate": 3.8674158399844285e-06, "loss": 2.7281, "step": 42846 }, { "epoch": 2.659817493326712, "grad_norm": 0.3707296492611062, "learning_rate": 3.866023232456739e-06, "loss": 2.6274, "step": 42847 }, { "epoch": 2.6598795704264697, "grad_norm": 0.4229533753491402, "learning_rate": 3.8646308656212214e-06, "loss": 2.6659, "step": 42848 }, { "epoch": 2.6599416475262276, "grad_norm": 0.37924087142393786, "learning_rate": 3.86323873948512e-06, "loss": 2.6613, "step": 42849 }, { "epoch": 2.6600037246259856, "grad_norm": 0.4044034144870546, "learning_rate": 3.861846854055701e-06, "loss": 2.7414, "step": 42850 }, { "epoch": 2.6600658017257435, "grad_norm": 0.3866009966431888, "learning_rate": 3.8604552093402304e-06, "loss": 2.6061, "step": 42851 }, { "epoch": 2.6601278788255014, "grad_norm": 0.3660405080310517, "learning_rate": 3.859063805345953e-06, "loss": 2.5466, "step": 42852 }, { "epoch": 2.6601899559252593, "grad_norm": 0.3735516313896354, "learning_rate": 3.857672642080151e-06, "loss": 2.6405, "step": 42853 }, { "epoch": 2.6602520330250172, "grad_norm": 0.4024447157014893, "learning_rate": 3.856281719550075e-06, "loss": 2.7134, "step": 42854 }, { "epoch": 2.6603141101247747, "grad_norm": 0.3897920883899543, "learning_rate": 3.854891037762981e-06, "loss": 2.6241, "step": 42855 }, { "epoch": 2.660376187224533, "grad_norm": 0.39018064020112847, "learning_rate": 3.853500596726106e-06, "loss": 2.7375, "step": 42856 }, { "epoch": 2.6604382643242905, "grad_norm": 0.377581571582587, "learning_rate": 3.8521103964467344e-06, "loss": 2.5693, "step": 42857 }, { "epoch": 2.660500341424049, "grad_norm": 0.3971825451845726, "learning_rate": 3.850720436932104e-06, "loss": 2.6329, "step": 42858 }, { "epoch": 2.6605624185238064, "grad_norm": 0.40573402796890695, "learning_rate": 3.84933071818947e-06, "loss": 2.6609, "step": 42859 }, { "epoch": 2.6606244956235647, "grad_norm": 0.389692605175604, "learning_rate": 3.847941240226072e-06, "loss": 2.6179, "step": 42860 }, { "epoch": 2.660686572723322, "grad_norm": 0.3903261376557055, "learning_rate": 3.8465520030491756e-06, "loss": 2.7458, "step": 42861 }, { "epoch": 2.66074864982308, "grad_norm": 0.3913171358368562, "learning_rate": 3.845163006666025e-06, "loss": 2.6155, "step": 42862 }, { "epoch": 2.660810726922838, "grad_norm": 0.4034620502427697, "learning_rate": 3.843774251083859e-06, "loss": 2.6245, "step": 42863 }, { "epoch": 2.660872804022596, "grad_norm": 0.4175901748445951, "learning_rate": 3.842385736309933e-06, "loss": 2.7558, "step": 42864 }, { "epoch": 2.660934881122354, "grad_norm": 0.39805781281092933, "learning_rate": 3.84099746235147e-06, "loss": 2.5381, "step": 42865 }, { "epoch": 2.660996958222112, "grad_norm": 0.39109390725145216, "learning_rate": 3.839609429215746e-06, "loss": 2.6419, "step": 42866 }, { "epoch": 2.6610590353218697, "grad_norm": 0.39907225367635796, "learning_rate": 3.838221636909978e-06, "loss": 2.6367, "step": 42867 }, { "epoch": 2.6611211124216276, "grad_norm": 0.4047860213432685, "learning_rate": 3.836834085441415e-06, "loss": 2.7495, "step": 42868 }, { "epoch": 2.6611831895213856, "grad_norm": 0.3803803586015135, "learning_rate": 3.835446774817292e-06, "loss": 2.7327, "step": 42869 }, { "epoch": 2.6612452666211435, "grad_norm": 0.4324786306625645, "learning_rate": 3.834059705044851e-06, "loss": 2.5947, "step": 42870 }, { "epoch": 2.6613073437209014, "grad_norm": 0.37649280726260737, "learning_rate": 3.8326728761313325e-06, "loss": 2.6536, "step": 42871 }, { "epoch": 2.6613694208206593, "grad_norm": 0.3786388712281856, "learning_rate": 3.8312862880839624e-06, "loss": 2.6696, "step": 42872 }, { "epoch": 2.661431497920417, "grad_norm": 0.38687479358106786, "learning_rate": 3.829899940909976e-06, "loss": 2.665, "step": 42873 }, { "epoch": 2.661493575020175, "grad_norm": 0.3987598050565221, "learning_rate": 3.828513834616609e-06, "loss": 2.6359, "step": 42874 }, { "epoch": 2.661555652119933, "grad_norm": 0.405723928075751, "learning_rate": 3.827127969211103e-06, "loss": 2.5883, "step": 42875 }, { "epoch": 2.661617729219691, "grad_norm": 0.3672666464383435, "learning_rate": 3.825742344700678e-06, "loss": 2.6861, "step": 42876 }, { "epoch": 2.661679806319449, "grad_norm": 0.39149058760123323, "learning_rate": 3.824356961092562e-06, "loss": 2.6529, "step": 42877 }, { "epoch": 2.661741883419207, "grad_norm": 0.4086403034305062, "learning_rate": 3.822971818393989e-06, "loss": 2.5628, "step": 42878 }, { "epoch": 2.6618039605189647, "grad_norm": 0.40487114116152384, "learning_rate": 3.821586916612174e-06, "loss": 2.6551, "step": 42879 }, { "epoch": 2.661866037618722, "grad_norm": 0.38668042773322253, "learning_rate": 3.820202255754357e-06, "loss": 2.6426, "step": 42880 }, { "epoch": 2.6619281147184806, "grad_norm": 0.3643347119523118, "learning_rate": 3.818817835827754e-06, "loss": 2.6778, "step": 42881 }, { "epoch": 2.661990191818238, "grad_norm": 0.3862335745087381, "learning_rate": 3.817433656839586e-06, "loss": 2.7446, "step": 42882 }, { "epoch": 2.6620522689179964, "grad_norm": 0.3840526423108735, "learning_rate": 3.816049718797076e-06, "loss": 2.6884, "step": 42883 }, { "epoch": 2.662114346017754, "grad_norm": 0.41469150127328913, "learning_rate": 3.8146660217074525e-06, "loss": 2.7306, "step": 42884 }, { "epoch": 2.662176423117512, "grad_norm": 0.400043895667282, "learning_rate": 3.8132825655779303e-06, "loss": 2.7746, "step": 42885 }, { "epoch": 2.6622385002172697, "grad_norm": 0.3844628280224495, "learning_rate": 3.8118993504157207e-06, "loss": 2.6703, "step": 42886 }, { "epoch": 2.6623005773170276, "grad_norm": 0.3714442755941131, "learning_rate": 3.8105163762280406e-06, "loss": 2.648, "step": 42887 }, { "epoch": 2.6623626544167855, "grad_norm": 0.4061666366435997, "learning_rate": 3.809133643022111e-06, "loss": 2.652, "step": 42888 }, { "epoch": 2.6624247315165435, "grad_norm": 0.37102562128393746, "learning_rate": 3.80775115080515e-06, "loss": 2.6107, "step": 42889 }, { "epoch": 2.6624868086163014, "grad_norm": 0.3829155024316057, "learning_rate": 3.8063688995843672e-06, "loss": 2.6092, "step": 42890 }, { "epoch": 2.6625488857160593, "grad_norm": 0.38888302535601815, "learning_rate": 3.8049868893669683e-06, "loss": 2.6244, "step": 42891 }, { "epoch": 2.662610962815817, "grad_norm": 0.4088569588373497, "learning_rate": 3.8036051201601585e-06, "loss": 2.617, "step": 42892 }, { "epoch": 2.662673039915575, "grad_norm": 0.3857478274162458, "learning_rate": 3.80222359197116e-06, "loss": 2.6719, "step": 42893 }, { "epoch": 2.662735117015333, "grad_norm": 0.36466913803016915, "learning_rate": 3.800842304807184e-06, "loss": 2.614, "step": 42894 }, { "epoch": 2.662797194115091, "grad_norm": 0.4129956013627689, "learning_rate": 3.799461258675424e-06, "loss": 2.6703, "step": 42895 }, { "epoch": 2.662859271214849, "grad_norm": 0.44003293252832826, "learning_rate": 3.7980804535830804e-06, "loss": 2.746, "step": 42896 }, { "epoch": 2.662921348314607, "grad_norm": 0.40135734898744213, "learning_rate": 3.7966998895373805e-06, "loss": 2.7037, "step": 42897 }, { "epoch": 2.6629834254143647, "grad_norm": 0.38732290758914456, "learning_rate": 3.7953195665455078e-06, "loss": 2.6915, "step": 42898 }, { "epoch": 2.6630455025141226, "grad_norm": 0.39887423048001963, "learning_rate": 3.7939394846146726e-06, "loss": 2.668, "step": 42899 }, { "epoch": 2.6631075796138806, "grad_norm": 0.3579570120290239, "learning_rate": 3.7925596437520748e-06, "loss": 2.568, "step": 42900 }, { "epoch": 2.6631696567136385, "grad_norm": 0.4009491165511986, "learning_rate": 3.791180043964898e-06, "loss": 2.5823, "step": 42901 }, { "epoch": 2.6632317338133964, "grad_norm": 0.4004047856501707, "learning_rate": 3.789800685260364e-06, "loss": 2.6659, "step": 42902 }, { "epoch": 2.663293810913154, "grad_norm": 0.42152231096971415, "learning_rate": 3.788421567645656e-06, "loss": 2.7101, "step": 42903 }, { "epoch": 2.6633558880129122, "grad_norm": 0.4124682522607519, "learning_rate": 3.787042691127968e-06, "loss": 2.6459, "step": 42904 }, { "epoch": 2.6634179651126697, "grad_norm": 0.392899452909796, "learning_rate": 3.785664055714494e-06, "loss": 2.7286, "step": 42905 }, { "epoch": 2.663480042212428, "grad_norm": 0.38956502150612016, "learning_rate": 3.7842856614124346e-06, "loss": 2.6812, "step": 42906 }, { "epoch": 2.6635421193121855, "grad_norm": 0.42225172741628647, "learning_rate": 3.7829075082289665e-06, "loss": 2.5932, "step": 42907 }, { "epoch": 2.663604196411944, "grad_norm": 0.37567231389666506, "learning_rate": 3.7815295961713017e-06, "loss": 2.6742, "step": 42908 }, { "epoch": 2.6636662735117014, "grad_norm": 0.4120360033428705, "learning_rate": 3.780151925246611e-06, "loss": 2.6415, "step": 42909 }, { "epoch": 2.6637283506114593, "grad_norm": 0.41630884144127117, "learning_rate": 3.778774495462084e-06, "loss": 2.7324, "step": 42910 }, { "epoch": 2.663790427711217, "grad_norm": 0.406275454970808, "learning_rate": 3.77739730682492e-06, "loss": 2.5942, "step": 42911 }, { "epoch": 2.663852504810975, "grad_norm": 0.3843778408534818, "learning_rate": 3.7760203593422972e-06, "loss": 2.6005, "step": 42912 }, { "epoch": 2.663914581910733, "grad_norm": 0.4054207775277388, "learning_rate": 3.7746436530213923e-06, "loss": 2.7052, "step": 42913 }, { "epoch": 2.663976659010491, "grad_norm": 0.40085935370728265, "learning_rate": 3.7732671878693946e-06, "loss": 2.7073, "step": 42914 }, { "epoch": 2.664038736110249, "grad_norm": 0.3757838956934054, "learning_rate": 3.7718909638934764e-06, "loss": 2.5734, "step": 42915 }, { "epoch": 2.664100813210007, "grad_norm": 0.4068110333391488, "learning_rate": 3.770514981100831e-06, "loss": 2.6772, "step": 42916 }, { "epoch": 2.6641628903097647, "grad_norm": 0.39342229618331914, "learning_rate": 3.769139239498637e-06, "loss": 2.7175, "step": 42917 }, { "epoch": 2.6642249674095226, "grad_norm": 0.40090927399299087, "learning_rate": 3.7677637390940656e-06, "loss": 2.6542, "step": 42918 }, { "epoch": 2.6642870445092806, "grad_norm": 0.43359569171092716, "learning_rate": 3.766388479894278e-06, "loss": 2.6466, "step": 42919 }, { "epoch": 2.6643491216090385, "grad_norm": 0.418749694681768, "learning_rate": 3.7650134619064802e-06, "loss": 2.6274, "step": 42920 }, { "epoch": 2.6644111987087964, "grad_norm": 0.3979841158491136, "learning_rate": 3.7636386851378325e-06, "loss": 2.6012, "step": 42921 }, { "epoch": 2.6644732758085543, "grad_norm": 0.4167668182489829, "learning_rate": 3.7622641495955013e-06, "loss": 2.6919, "step": 42922 }, { "epoch": 2.6645353529083122, "grad_norm": 0.39631190565516183, "learning_rate": 3.760889855286659e-06, "loss": 2.7144, "step": 42923 }, { "epoch": 2.66459743000807, "grad_norm": 0.4102459582362743, "learning_rate": 3.759515802218483e-06, "loss": 2.7134, "step": 42924 }, { "epoch": 2.664659507107828, "grad_norm": 0.3982094504767241, "learning_rate": 3.7581419903981395e-06, "loss": 2.6771, "step": 42925 }, { "epoch": 2.664721584207586, "grad_norm": 0.4307495689710961, "learning_rate": 3.756768419832801e-06, "loss": 2.7473, "step": 42926 }, { "epoch": 2.664783661307344, "grad_norm": 0.4324495149926165, "learning_rate": 3.7553950905296178e-06, "loss": 2.7759, "step": 42927 }, { "epoch": 2.6648457384071014, "grad_norm": 0.3695314750106159, "learning_rate": 3.7540220024957662e-06, "loss": 2.6439, "step": 42928 }, { "epoch": 2.6649078155068597, "grad_norm": 0.38018057560733204, "learning_rate": 3.752649155738408e-06, "loss": 2.7657, "step": 42929 }, { "epoch": 2.664969892606617, "grad_norm": 0.4174307631856987, "learning_rate": 3.7512765502647095e-06, "loss": 2.7024, "step": 42930 }, { "epoch": 2.6650319697063756, "grad_norm": 0.38602107143207753, "learning_rate": 3.749904186081832e-06, "loss": 2.6719, "step": 42931 }, { "epoch": 2.665094046806133, "grad_norm": 0.4073765064702092, "learning_rate": 3.7485320631969188e-06, "loss": 2.7093, "step": 42932 }, { "epoch": 2.665156123905891, "grad_norm": 0.3812535324871068, "learning_rate": 3.747160181617149e-06, "loss": 2.6596, "step": 42933 }, { "epoch": 2.665218201005649, "grad_norm": 0.40439396305223946, "learning_rate": 3.7457885413496764e-06, "loss": 2.7299, "step": 42934 }, { "epoch": 2.665280278105407, "grad_norm": 0.3522319248312062, "learning_rate": 3.744417142401652e-06, "loss": 2.6399, "step": 42935 }, { "epoch": 2.6653423552051647, "grad_norm": 0.39972499319614896, "learning_rate": 3.743045984780236e-06, "loss": 2.6772, "step": 42936 }, { "epoch": 2.6654044323049226, "grad_norm": 0.3764188072862128, "learning_rate": 3.7416750684925673e-06, "loss": 2.6306, "step": 42937 }, { "epoch": 2.6654665094046806, "grad_norm": 0.38657191407714214, "learning_rate": 3.7403043935458125e-06, "loss": 2.6529, "step": 42938 }, { "epoch": 2.6655285865044385, "grad_norm": 0.42577939514316093, "learning_rate": 3.738933959947127e-06, "loss": 2.6569, "step": 42939 }, { "epoch": 2.6655906636041964, "grad_norm": 0.37921682391092104, "learning_rate": 3.7375637677036446e-06, "loss": 2.6237, "step": 42940 }, { "epoch": 2.6656527407039543, "grad_norm": 0.38287886219523926, "learning_rate": 3.7361938168225252e-06, "loss": 2.652, "step": 42941 }, { "epoch": 2.6657148178037122, "grad_norm": 0.3997974933672113, "learning_rate": 3.734824107310908e-06, "loss": 2.7349, "step": 42942 }, { "epoch": 2.66577689490347, "grad_norm": 0.4152584438579335, "learning_rate": 3.733454639175954e-06, "loss": 2.7698, "step": 42943 }, { "epoch": 2.665838972003228, "grad_norm": 0.3914167962395205, "learning_rate": 3.7320854124248017e-06, "loss": 2.7026, "step": 42944 }, { "epoch": 2.665901049102986, "grad_norm": 0.3695365558590141, "learning_rate": 3.73071642706459e-06, "loss": 2.6543, "step": 42945 }, { "epoch": 2.665963126202744, "grad_norm": 0.3710466254579612, "learning_rate": 3.729347683102452e-06, "loss": 2.6995, "step": 42946 }, { "epoch": 2.666025203302502, "grad_norm": 0.4176441964673213, "learning_rate": 3.727979180545549e-06, "loss": 2.6283, "step": 42947 }, { "epoch": 2.6660872804022597, "grad_norm": 0.38628748727254125, "learning_rate": 3.7266109194010133e-06, "loss": 2.702, "step": 42948 }, { "epoch": 2.6661493575020176, "grad_norm": 0.35562574671447345, "learning_rate": 3.725242899675979e-06, "loss": 2.7317, "step": 42949 }, { "epoch": 2.6662114346017756, "grad_norm": 0.3889523170133938, "learning_rate": 3.7238751213775847e-06, "loss": 2.6766, "step": 42950 }, { "epoch": 2.666273511701533, "grad_norm": 0.36504612115247326, "learning_rate": 3.722507584512963e-06, "loss": 2.6295, "step": 42951 }, { "epoch": 2.6663355888012914, "grad_norm": 0.37394259303694166, "learning_rate": 3.721140289089259e-06, "loss": 2.6433, "step": 42952 }, { "epoch": 2.666397665901049, "grad_norm": 0.3983645465857031, "learning_rate": 3.719773235113605e-06, "loss": 2.6646, "step": 42953 }, { "epoch": 2.6664597430008072, "grad_norm": 0.3660480594818573, "learning_rate": 3.7184064225931294e-06, "loss": 2.6173, "step": 42954 }, { "epoch": 2.6665218201005647, "grad_norm": 0.38148044906805373, "learning_rate": 3.717039851534948e-06, "loss": 2.666, "step": 42955 }, { "epoch": 2.666583897200323, "grad_norm": 0.34534317070264575, "learning_rate": 3.715673521946217e-06, "loss": 2.6476, "step": 42956 }, { "epoch": 2.6666459743000805, "grad_norm": 0.3926057758367483, "learning_rate": 3.714307433834052e-06, "loss": 2.7447, "step": 42957 }, { "epoch": 2.6667080513998385, "grad_norm": 0.38365623369222324, "learning_rate": 3.7129415872055874e-06, "loss": 2.6527, "step": 42958 }, { "epoch": 2.6667701284995964, "grad_norm": 0.3547394785836315, "learning_rate": 3.7115759820679275e-06, "loss": 2.713, "step": 42959 }, { "epoch": 2.6668322055993543, "grad_norm": 0.3583637189549853, "learning_rate": 3.7102106184282224e-06, "loss": 2.6595, "step": 42960 }, { "epoch": 2.666894282699112, "grad_norm": 0.3606516924389077, "learning_rate": 3.708845496293589e-06, "loss": 2.5134, "step": 42961 }, { "epoch": 2.66695635979887, "grad_norm": 0.39092195667497903, "learning_rate": 3.7074806156711438e-06, "loss": 2.6064, "step": 42962 }, { "epoch": 2.667018436898628, "grad_norm": 0.4122611936815758, "learning_rate": 3.7061159765680088e-06, "loss": 2.6996, "step": 42963 }, { "epoch": 2.667080513998386, "grad_norm": 0.38004630917158566, "learning_rate": 3.7047515789913e-06, "loss": 2.6821, "step": 42964 }, { "epoch": 2.667142591098144, "grad_norm": 0.3959873071488293, "learning_rate": 3.7033874229481457e-06, "loss": 2.6894, "step": 42965 }, { "epoch": 2.667204668197902, "grad_norm": 0.3767238769751965, "learning_rate": 3.702023508445662e-06, "loss": 2.6865, "step": 42966 }, { "epoch": 2.6672667452976597, "grad_norm": 0.37132430694434243, "learning_rate": 3.7006598354909548e-06, "loss": 2.7294, "step": 42967 }, { "epoch": 2.6673288223974176, "grad_norm": 0.38427311702279593, "learning_rate": 3.6992964040911403e-06, "loss": 2.6267, "step": 42968 }, { "epoch": 2.6673908994971756, "grad_norm": 0.38506528155376796, "learning_rate": 3.697933214253346e-06, "loss": 2.6389, "step": 42969 }, { "epoch": 2.6674529765969335, "grad_norm": 0.36700349620796063, "learning_rate": 3.6965702659846723e-06, "loss": 2.6356, "step": 42970 }, { "epoch": 2.6675150536966914, "grad_norm": 0.3769413842462996, "learning_rate": 3.6952075592922355e-06, "loss": 2.6874, "step": 42971 }, { "epoch": 2.6675771307964493, "grad_norm": 0.3880289405470775, "learning_rate": 3.6938450941831294e-06, "loss": 2.7533, "step": 42972 }, { "epoch": 2.6676392078962072, "grad_norm": 0.37323378828450304, "learning_rate": 3.6924828706644767e-06, "loss": 2.6019, "step": 42973 }, { "epoch": 2.667701284995965, "grad_norm": 0.36380363150913225, "learning_rate": 3.691120888743388e-06, "loss": 2.6485, "step": 42974 }, { "epoch": 2.667763362095723, "grad_norm": 0.3757004157222695, "learning_rate": 3.689759148426969e-06, "loss": 2.6116, "step": 42975 }, { "epoch": 2.6678254391954805, "grad_norm": 0.3742982736082467, "learning_rate": 3.6883976497223137e-06, "loss": 2.5857, "step": 42976 }, { "epoch": 2.667887516295239, "grad_norm": 0.3581370926191789, "learning_rate": 3.6870363926365337e-06, "loss": 2.6416, "step": 42977 }, { "epoch": 2.6679495933949964, "grad_norm": 0.38988241183618955, "learning_rate": 3.685675377176717e-06, "loss": 2.5789, "step": 42978 }, { "epoch": 2.6680116704947547, "grad_norm": 0.3719393081518525, "learning_rate": 3.684314603349981e-06, "loss": 2.6583, "step": 42979 }, { "epoch": 2.668073747594512, "grad_norm": 0.3718699118322202, "learning_rate": 3.682954071163425e-06, "loss": 2.6385, "step": 42980 }, { "epoch": 2.66813582469427, "grad_norm": 0.37542668545500785, "learning_rate": 3.6815937806241374e-06, "loss": 2.6818, "step": 42981 }, { "epoch": 2.668197901794028, "grad_norm": 0.40042718859942134, "learning_rate": 3.6802337317392133e-06, "loss": 2.5666, "step": 42982 }, { "epoch": 2.668259978893786, "grad_norm": 0.38215703707146126, "learning_rate": 3.678873924515758e-06, "loss": 2.6231, "step": 42983 }, { "epoch": 2.668322055993544, "grad_norm": 0.3889986897424436, "learning_rate": 3.677514358960865e-06, "loss": 2.637, "step": 42984 }, { "epoch": 2.668384133093302, "grad_norm": 0.40894329006778274, "learning_rate": 3.67615503508163e-06, "loss": 2.6276, "step": 42985 }, { "epoch": 2.6684462101930597, "grad_norm": 0.3823252710788319, "learning_rate": 3.674795952885135e-06, "loss": 2.6165, "step": 42986 }, { "epoch": 2.6685082872928176, "grad_norm": 0.42867108553004996, "learning_rate": 3.673437112378464e-06, "loss": 2.6732, "step": 42987 }, { "epoch": 2.6685703643925756, "grad_norm": 0.37921949593211696, "learning_rate": 3.672078513568733e-06, "loss": 2.6792, "step": 42988 }, { "epoch": 2.6686324414923335, "grad_norm": 0.35657576172728656, "learning_rate": 3.670720156463009e-06, "loss": 2.6414, "step": 42989 }, { "epoch": 2.6686945185920914, "grad_norm": 0.3735247043319174, "learning_rate": 3.669362041068386e-06, "loss": 2.5729, "step": 42990 }, { "epoch": 2.6687565956918493, "grad_norm": 0.4047078425769913, "learning_rate": 3.668004167391942e-06, "loss": 2.6922, "step": 42991 }, { "epoch": 2.6688186727916072, "grad_norm": 0.4135852798730219, "learning_rate": 3.6666465354407766e-06, "loss": 2.6246, "step": 42992 }, { "epoch": 2.668880749891365, "grad_norm": 0.38005340589882447, "learning_rate": 3.665289145221962e-06, "loss": 2.6484, "step": 42993 }, { "epoch": 2.668942826991123, "grad_norm": 0.44494891800385045, "learning_rate": 3.6639319967425813e-06, "loss": 2.643, "step": 42994 }, { "epoch": 2.669004904090881, "grad_norm": 0.44713893911951197, "learning_rate": 3.662575090009712e-06, "loss": 2.6801, "step": 42995 }, { "epoch": 2.669066981190639, "grad_norm": 0.4337050664477387, "learning_rate": 3.661218425030438e-06, "loss": 2.7074, "step": 42996 }, { "epoch": 2.669129058290397, "grad_norm": 0.382856969780593, "learning_rate": 3.6598620018118425e-06, "loss": 2.6887, "step": 42997 }, { "epoch": 2.6691911353901547, "grad_norm": 0.3784354322772313, "learning_rate": 3.6585058203609967e-06, "loss": 2.6297, "step": 42998 }, { "epoch": 2.669253212489912, "grad_norm": 0.40396836428543814, "learning_rate": 3.6571498806849737e-06, "loss": 2.6102, "step": 42999 }, { "epoch": 2.6693152895896706, "grad_norm": 0.4204168551771928, "learning_rate": 3.655794182790839e-06, "loss": 2.5517, "step": 43000 }, { "epoch": 2.669377366689428, "grad_norm": 0.39445530755384955, "learning_rate": 3.654438726685688e-06, "loss": 2.7281, "step": 43001 }, { "epoch": 2.6694394437891864, "grad_norm": 0.38059342709086064, "learning_rate": 3.6530835123765817e-06, "loss": 2.6087, "step": 43002 }, { "epoch": 2.669501520888944, "grad_norm": 0.37476906023166784, "learning_rate": 3.6517285398705915e-06, "loss": 2.7373, "step": 43003 }, { "epoch": 2.6695635979887022, "grad_norm": 0.40583666535066465, "learning_rate": 3.650373809174773e-06, "loss": 2.6488, "step": 43004 }, { "epoch": 2.6696256750884597, "grad_norm": 0.408794467560402, "learning_rate": 3.6490193202962044e-06, "loss": 2.7331, "step": 43005 }, { "epoch": 2.6696877521882176, "grad_norm": 0.3951515571252223, "learning_rate": 3.647665073241968e-06, "loss": 2.7158, "step": 43006 }, { "epoch": 2.6697498292879756, "grad_norm": 0.38906995237635633, "learning_rate": 3.646311068019109e-06, "loss": 2.7133, "step": 43007 }, { "epoch": 2.6698119063877335, "grad_norm": 0.37885775698216667, "learning_rate": 3.644957304634705e-06, "loss": 2.7101, "step": 43008 }, { "epoch": 2.6698739834874914, "grad_norm": 0.3967056033047169, "learning_rate": 3.6436037830958003e-06, "loss": 2.7273, "step": 43009 }, { "epoch": 2.6699360605872493, "grad_norm": 0.3981210536551601, "learning_rate": 3.642250503409472e-06, "loss": 2.5813, "step": 43010 }, { "epoch": 2.6699981376870072, "grad_norm": 0.38762385480921313, "learning_rate": 3.640897465582782e-06, "loss": 2.6879, "step": 43011 }, { "epoch": 2.670060214786765, "grad_norm": 0.38043392578170676, "learning_rate": 3.6395446696227854e-06, "loss": 2.7284, "step": 43012 }, { "epoch": 2.670122291886523, "grad_norm": 0.37520631439425495, "learning_rate": 3.638192115536532e-06, "loss": 2.6395, "step": 43013 }, { "epoch": 2.670184368986281, "grad_norm": 0.40248610931618745, "learning_rate": 3.6368398033310824e-06, "loss": 2.6367, "step": 43014 }, { "epoch": 2.670246446086039, "grad_norm": 0.4002278997465979, "learning_rate": 3.6354877330134984e-06, "loss": 2.761, "step": 43015 }, { "epoch": 2.670308523185797, "grad_norm": 0.4036432778882587, "learning_rate": 3.634135904590835e-06, "loss": 2.5994, "step": 43016 }, { "epoch": 2.6703706002855547, "grad_norm": 0.3635518491606941, "learning_rate": 3.632784318070137e-06, "loss": 2.6566, "step": 43017 }, { "epoch": 2.6704326773853126, "grad_norm": 0.37399220149431517, "learning_rate": 3.631432973458454e-06, "loss": 2.6602, "step": 43018 }, { "epoch": 2.6704947544850706, "grad_norm": 0.3930112811937316, "learning_rate": 3.630081870762847e-06, "loss": 2.6275, "step": 43019 }, { "epoch": 2.6705568315848285, "grad_norm": 0.42134815450684937, "learning_rate": 3.6287310099903605e-06, "loss": 2.669, "step": 43020 }, { "epoch": 2.6706189086845864, "grad_norm": 0.3998031586364824, "learning_rate": 3.6273803911480443e-06, "loss": 2.6287, "step": 43021 }, { "epoch": 2.6706809857843443, "grad_norm": 0.38757697156178167, "learning_rate": 3.626030014242937e-06, "loss": 2.5986, "step": 43022 }, { "epoch": 2.6707430628841022, "grad_norm": 0.3892412280709488, "learning_rate": 3.6246798792820833e-06, "loss": 2.6738, "step": 43023 }, { "epoch": 2.6708051399838597, "grad_norm": 0.4208568012505126, "learning_rate": 3.623329986272539e-06, "loss": 2.6843, "step": 43024 }, { "epoch": 2.670867217083618, "grad_norm": 0.3841609560458613, "learning_rate": 3.621980335221342e-06, "loss": 2.634, "step": 43025 }, { "epoch": 2.6709292941833755, "grad_norm": 0.3600349379182487, "learning_rate": 3.620630926135532e-06, "loss": 2.7249, "step": 43026 }, { "epoch": 2.670991371283134, "grad_norm": 0.4155618225209174, "learning_rate": 3.6192817590221417e-06, "loss": 2.7009, "step": 43027 }, { "epoch": 2.6710534483828914, "grad_norm": 0.3924716786171208, "learning_rate": 3.6179328338882213e-06, "loss": 2.6585, "step": 43028 }, { "epoch": 2.6711155254826493, "grad_norm": 0.39086077250494444, "learning_rate": 3.6165841507408092e-06, "loss": 2.7354, "step": 43029 }, { "epoch": 2.671177602582407, "grad_norm": 0.41728502126700917, "learning_rate": 3.615235709586939e-06, "loss": 2.625, "step": 43030 }, { "epoch": 2.671239679682165, "grad_norm": 0.3566069563049171, "learning_rate": 3.6138875104336335e-06, "loss": 2.7157, "step": 43031 }, { "epoch": 2.671301756781923, "grad_norm": 0.3856085630342087, "learning_rate": 3.612539553287947e-06, "loss": 2.7188, "step": 43032 }, { "epoch": 2.671363833881681, "grad_norm": 0.39054869860265135, "learning_rate": 3.611191838156902e-06, "loss": 2.7211, "step": 43033 }, { "epoch": 2.671425910981439, "grad_norm": 0.4126490273181189, "learning_rate": 3.609844365047532e-06, "loss": 2.7264, "step": 43034 }, { "epoch": 2.671487988081197, "grad_norm": 0.36845602909440184, "learning_rate": 3.608497133966865e-06, "loss": 2.5953, "step": 43035 }, { "epoch": 2.6715500651809547, "grad_norm": 0.3843620429917756, "learning_rate": 3.6071501449219225e-06, "loss": 2.6638, "step": 43036 }, { "epoch": 2.6716121422807126, "grad_norm": 0.3863244406560128, "learning_rate": 3.6058033979197437e-06, "loss": 2.6491, "step": 43037 }, { "epoch": 2.6716742193804706, "grad_norm": 0.3732463043959977, "learning_rate": 3.6044568929673505e-06, "loss": 2.7267, "step": 43038 }, { "epoch": 2.6717362964802285, "grad_norm": 0.39125290948803654, "learning_rate": 3.603110630071771e-06, "loss": 2.6809, "step": 43039 }, { "epoch": 2.6717983735799864, "grad_norm": 0.41302969619586855, "learning_rate": 3.601764609240027e-06, "loss": 2.7305, "step": 43040 }, { "epoch": 2.6718604506797443, "grad_norm": 0.3819815511692349, "learning_rate": 3.6004188304791354e-06, "loss": 2.7021, "step": 43041 }, { "epoch": 2.6719225277795022, "grad_norm": 0.39085667415755515, "learning_rate": 3.5990732937961292e-06, "loss": 2.6727, "step": 43042 }, { "epoch": 2.67198460487926, "grad_norm": 0.3704871188484429, "learning_rate": 3.5977279991980197e-06, "loss": 2.6476, "step": 43043 }, { "epoch": 2.672046681979018, "grad_norm": 0.39993185416527427, "learning_rate": 3.5963829466918342e-06, "loss": 2.6456, "step": 43044 }, { "epoch": 2.672108759078776, "grad_norm": 0.3660571780072728, "learning_rate": 3.5950381362845677e-06, "loss": 2.6723, "step": 43045 }, { "epoch": 2.672170836178534, "grad_norm": 0.42721665635194817, "learning_rate": 3.593693567983264e-06, "loss": 2.6046, "step": 43046 }, { "epoch": 2.6722329132782914, "grad_norm": 0.376296880437484, "learning_rate": 3.592349241794929e-06, "loss": 2.6361, "step": 43047 }, { "epoch": 2.6722949903780497, "grad_norm": 0.36398870023019075, "learning_rate": 3.591005157726568e-06, "loss": 2.6563, "step": 43048 }, { "epoch": 2.672357067477807, "grad_norm": 0.38460955867573104, "learning_rate": 3.5896613157852034e-06, "loss": 2.7609, "step": 43049 }, { "epoch": 2.6724191445775656, "grad_norm": 0.39697515245241666, "learning_rate": 3.5883177159778347e-06, "loss": 2.6689, "step": 43050 }, { "epoch": 2.672481221677323, "grad_norm": 0.3811887441734603, "learning_rate": 3.5869743583114846e-06, "loss": 2.6898, "step": 43051 }, { "epoch": 2.6725432987770814, "grad_norm": 0.3881466638044543, "learning_rate": 3.5856312427931583e-06, "loss": 2.7776, "step": 43052 }, { "epoch": 2.672605375876839, "grad_norm": 0.410741678873383, "learning_rate": 3.5842883694298614e-06, "loss": 2.7279, "step": 43053 }, { "epoch": 2.672667452976597, "grad_norm": 0.379255684247353, "learning_rate": 3.5829457382285936e-06, "loss": 2.6688, "step": 43054 }, { "epoch": 2.6727295300763547, "grad_norm": 0.3989719658291234, "learning_rate": 3.581603349196372e-06, "loss": 2.6378, "step": 43055 }, { "epoch": 2.6727916071761126, "grad_norm": 0.40085593377292134, "learning_rate": 3.5802612023401904e-06, "loss": 2.7286, "step": 43056 }, { "epoch": 2.6728536842758706, "grad_norm": 0.4067140459806431, "learning_rate": 3.5789192976670606e-06, "loss": 2.5807, "step": 43057 }, { "epoch": 2.6729157613756285, "grad_norm": 0.3767706078333659, "learning_rate": 3.577577635183965e-06, "loss": 2.6213, "step": 43058 }, { "epoch": 2.6729778384753864, "grad_norm": 0.39732577863427443, "learning_rate": 3.5762362148979325e-06, "loss": 2.6919, "step": 43059 }, { "epoch": 2.6730399155751443, "grad_norm": 0.3820075895631507, "learning_rate": 3.5748950368159397e-06, "loss": 2.6575, "step": 43060 }, { "epoch": 2.6731019926749022, "grad_norm": 0.4100233680183249, "learning_rate": 3.5735541009449867e-06, "loss": 2.7109, "step": 43061 }, { "epoch": 2.67316406977466, "grad_norm": 0.37629725663236613, "learning_rate": 3.5722134072920742e-06, "loss": 2.6653, "step": 43062 }, { "epoch": 2.673226146874418, "grad_norm": 0.37998678554597126, "learning_rate": 3.5708729558641907e-06, "loss": 2.727, "step": 43063 }, { "epoch": 2.673288223974176, "grad_norm": 0.3720460217974646, "learning_rate": 3.5695327466683415e-06, "loss": 2.6052, "step": 43064 }, { "epoch": 2.673350301073934, "grad_norm": 0.36400454309531316, "learning_rate": 3.56819277971151e-06, "loss": 2.6843, "step": 43065 }, { "epoch": 2.673412378173692, "grad_norm": 0.37658291757401263, "learning_rate": 3.566853055000685e-06, "loss": 2.6961, "step": 43066 }, { "epoch": 2.6734744552734497, "grad_norm": 0.39711069332356513, "learning_rate": 3.5655135725428555e-06, "loss": 2.6106, "step": 43067 }, { "epoch": 2.6735365323732077, "grad_norm": 0.37941742856168226, "learning_rate": 3.5641743323450217e-06, "loss": 2.6985, "step": 43068 }, { "epoch": 2.6735986094729656, "grad_norm": 0.3554359462818781, "learning_rate": 3.562835334414166e-06, "loss": 2.6419, "step": 43069 }, { "epoch": 2.6736606865727235, "grad_norm": 0.3999345368223231, "learning_rate": 3.5614965787572676e-06, "loss": 2.701, "step": 43070 }, { "epoch": 2.6737227636724814, "grad_norm": 0.3827437514250407, "learning_rate": 3.5601580653813085e-06, "loss": 2.596, "step": 43071 }, { "epoch": 2.673784840772239, "grad_norm": 0.3721651972535117, "learning_rate": 3.558819794293283e-06, "loss": 2.6185, "step": 43072 }, { "epoch": 2.6738469178719972, "grad_norm": 0.40998566122684516, "learning_rate": 3.5574817655001645e-06, "loss": 2.7409, "step": 43073 }, { "epoch": 2.6739089949717547, "grad_norm": 0.389308044591822, "learning_rate": 3.5561439790089414e-06, "loss": 2.6839, "step": 43074 }, { "epoch": 2.673971072071513, "grad_norm": 0.38917786523211817, "learning_rate": 3.5548064348265962e-06, "loss": 2.6535, "step": 43075 }, { "epoch": 2.6740331491712706, "grad_norm": 0.37603759270290177, "learning_rate": 3.553469132960097e-06, "loss": 2.7419, "step": 43076 }, { "epoch": 2.6740952262710285, "grad_norm": 0.3926719819031916, "learning_rate": 3.552132073416414e-06, "loss": 2.6686, "step": 43077 }, { "epoch": 2.6741573033707864, "grad_norm": 0.37218443554594355, "learning_rate": 3.550795256202544e-06, "loss": 2.6959, "step": 43078 }, { "epoch": 2.6742193804705443, "grad_norm": 0.4118752741197707, "learning_rate": 3.5494586813254517e-06, "loss": 2.6521, "step": 43079 }, { "epoch": 2.6742814575703022, "grad_norm": 0.36760391809691645, "learning_rate": 3.54812234879211e-06, "loss": 2.6494, "step": 43080 }, { "epoch": 2.67434353467006, "grad_norm": 0.3892308150679888, "learning_rate": 3.546786258609486e-06, "loss": 2.725, "step": 43081 }, { "epoch": 2.674405611769818, "grad_norm": 0.37898684165653573, "learning_rate": 3.5454504107845567e-06, "loss": 2.6836, "step": 43082 }, { "epoch": 2.674467688869576, "grad_norm": 0.36548807645127535, "learning_rate": 3.5441148053242945e-06, "loss": 2.6236, "step": 43083 }, { "epoch": 2.674529765969334, "grad_norm": 0.36597282297823824, "learning_rate": 3.542779442235661e-06, "loss": 2.6099, "step": 43084 }, { "epoch": 2.674591843069092, "grad_norm": 0.382348207230304, "learning_rate": 3.5414443215256223e-06, "loss": 2.6734, "step": 43085 }, { "epoch": 2.6746539201688497, "grad_norm": 0.3862446277207762, "learning_rate": 3.540109443201145e-06, "loss": 2.6857, "step": 43086 }, { "epoch": 2.6747159972686076, "grad_norm": 0.361052392450649, "learning_rate": 3.538774807269196e-06, "loss": 2.7202, "step": 43087 }, { "epoch": 2.6747780743683656, "grad_norm": 0.3831471129067583, "learning_rate": 3.5374404137367422e-06, "loss": 2.6643, "step": 43088 }, { "epoch": 2.6748401514681235, "grad_norm": 0.39201472074617316, "learning_rate": 3.5361062626107387e-06, "loss": 2.6797, "step": 43089 }, { "epoch": 2.6749022285678814, "grad_norm": 0.35408444244861653, "learning_rate": 3.534772353898147e-06, "loss": 2.621, "step": 43090 }, { "epoch": 2.6749643056676393, "grad_norm": 0.37601149755891927, "learning_rate": 3.5334386876059277e-06, "loss": 2.6668, "step": 43091 }, { "epoch": 2.6750263827673972, "grad_norm": 0.3907783016747953, "learning_rate": 3.532105263741042e-06, "loss": 2.6314, "step": 43092 }, { "epoch": 2.675088459867155, "grad_norm": 0.37723761777863013, "learning_rate": 3.5307720823104463e-06, "loss": 2.6306, "step": 43093 }, { "epoch": 2.675150536966913, "grad_norm": 0.3955417571395041, "learning_rate": 3.5294391433210783e-06, "loss": 2.6804, "step": 43094 }, { "epoch": 2.6752126140666705, "grad_norm": 0.38878005166133933, "learning_rate": 3.5281064467799217e-06, "loss": 2.6946, "step": 43095 }, { "epoch": 2.675274691166429, "grad_norm": 0.39399168680589663, "learning_rate": 3.52677399269391e-06, "loss": 2.6694, "step": 43096 }, { "epoch": 2.6753367682661864, "grad_norm": 0.38062962180026616, "learning_rate": 3.5254417810700047e-06, "loss": 2.678, "step": 43097 }, { "epoch": 2.6753988453659447, "grad_norm": 0.3591365022818911, "learning_rate": 3.5241098119151496e-06, "loss": 2.6051, "step": 43098 }, { "epoch": 2.675460922465702, "grad_norm": 0.395551646123023, "learning_rate": 3.522778085236289e-06, "loss": 2.6636, "step": 43099 }, { "epoch": 2.67552299956546, "grad_norm": 0.3701738701648619, "learning_rate": 3.5214466010403847e-06, "loss": 2.7164, "step": 43100 }, { "epoch": 2.675585076665218, "grad_norm": 0.38538427044918777, "learning_rate": 3.5201153593343805e-06, "loss": 2.6006, "step": 43101 }, { "epoch": 2.675647153764976, "grad_norm": 0.37987075041226315, "learning_rate": 3.5187843601252157e-06, "loss": 2.6299, "step": 43102 }, { "epoch": 2.675709230864734, "grad_norm": 0.3715241596406222, "learning_rate": 3.5174536034198236e-06, "loss": 2.6009, "step": 43103 }, { "epoch": 2.675771307964492, "grad_norm": 0.38997921014050496, "learning_rate": 3.516123089225165e-06, "loss": 2.5511, "step": 43104 }, { "epoch": 2.6758333850642497, "grad_norm": 0.3807423142730104, "learning_rate": 3.5147928175481848e-06, "loss": 2.6853, "step": 43105 }, { "epoch": 2.6758954621640076, "grad_norm": 0.36769405892017715, "learning_rate": 3.513462788395816e-06, "loss": 2.6059, "step": 43106 }, { "epoch": 2.6759575392637656, "grad_norm": 0.4079360453230653, "learning_rate": 3.512133001774992e-06, "loss": 2.6867, "step": 43107 }, { "epoch": 2.6760196163635235, "grad_norm": 0.39556437900245806, "learning_rate": 3.510803457692652e-06, "loss": 2.6228, "step": 43108 }, { "epoch": 2.6760816934632814, "grad_norm": 0.3649213540905073, "learning_rate": 3.5094741561557455e-06, "loss": 2.6731, "step": 43109 }, { "epoch": 2.6761437705630393, "grad_norm": 0.3872145932242064, "learning_rate": 3.508145097171195e-06, "loss": 2.5958, "step": 43110 }, { "epoch": 2.6762058476627972, "grad_norm": 0.3806720762157938, "learning_rate": 3.506816280745934e-06, "loss": 2.587, "step": 43111 }, { "epoch": 2.676267924762555, "grad_norm": 0.401767410512317, "learning_rate": 3.5054877068869064e-06, "loss": 2.6209, "step": 43112 }, { "epoch": 2.676330001862313, "grad_norm": 0.3962475863542663, "learning_rate": 3.5041593756010236e-06, "loss": 2.6327, "step": 43113 }, { "epoch": 2.676392078962071, "grad_norm": 0.38464742868616003, "learning_rate": 3.5028312868952363e-06, "loss": 2.7375, "step": 43114 }, { "epoch": 2.676454156061829, "grad_norm": 0.3966384175036112, "learning_rate": 3.5015034407764713e-06, "loss": 2.7628, "step": 43115 }, { "epoch": 2.676516233161587, "grad_norm": 0.3826561238610717, "learning_rate": 3.5001758372516457e-06, "loss": 2.6461, "step": 43116 }, { "epoch": 2.6765783102613447, "grad_norm": 0.3716946605366601, "learning_rate": 3.498848476327682e-06, "loss": 2.6281, "step": 43117 }, { "epoch": 2.676640387361102, "grad_norm": 0.38271665346380146, "learning_rate": 3.497521358011524e-06, "loss": 2.5998, "step": 43118 }, { "epoch": 2.6767024644608606, "grad_norm": 0.371449209929553, "learning_rate": 3.4961944823100888e-06, "loss": 2.6147, "step": 43119 }, { "epoch": 2.676764541560618, "grad_norm": 0.3908932146501316, "learning_rate": 3.4948678492302934e-06, "loss": 2.651, "step": 43120 }, { "epoch": 2.6768266186603764, "grad_norm": 0.3708091447563634, "learning_rate": 3.493541458779065e-06, "loss": 2.7018, "step": 43121 }, { "epoch": 2.676888695760134, "grad_norm": 0.41353031362264536, "learning_rate": 3.4922153109633094e-06, "loss": 2.7776, "step": 43122 }, { "epoch": 2.6769507728598922, "grad_norm": 0.36111328336274273, "learning_rate": 3.4908894057899653e-06, "loss": 2.6492, "step": 43123 }, { "epoch": 2.6770128499596497, "grad_norm": 0.4058838992011129, "learning_rate": 3.489563743265939e-06, "loss": 2.7334, "step": 43124 }, { "epoch": 2.6770749270594076, "grad_norm": 0.3713983991833905, "learning_rate": 3.4882383233981518e-06, "loss": 2.6986, "step": 43125 }, { "epoch": 2.6771370041591656, "grad_norm": 0.3781949672122542, "learning_rate": 3.4869131461935102e-06, "loss": 2.6412, "step": 43126 }, { "epoch": 2.6771990812589235, "grad_norm": 0.38594239266572283, "learning_rate": 3.485588211658941e-06, "loss": 2.5962, "step": 43127 }, { "epoch": 2.6772611583586814, "grad_norm": 0.3709228121927532, "learning_rate": 3.4842635198013505e-06, "loss": 2.6825, "step": 43128 }, { "epoch": 2.6773232354584393, "grad_norm": 0.39499944345821336, "learning_rate": 3.48293907062765e-06, "loss": 2.6711, "step": 43129 }, { "epoch": 2.6773853125581972, "grad_norm": 0.3871958323298737, "learning_rate": 3.481614864144739e-06, "loss": 2.6278, "step": 43130 }, { "epoch": 2.677447389657955, "grad_norm": 0.4012984428100355, "learning_rate": 3.4802909003595396e-06, "loss": 2.8391, "step": 43131 }, { "epoch": 2.677509466757713, "grad_norm": 0.3953058174382678, "learning_rate": 3.4789671792789636e-06, "loss": 2.7268, "step": 43132 }, { "epoch": 2.677571543857471, "grad_norm": 0.39310717637817044, "learning_rate": 3.477643700909905e-06, "loss": 2.6405, "step": 43133 }, { "epoch": 2.677633620957229, "grad_norm": 0.3875861795023321, "learning_rate": 3.476320465259275e-06, "loss": 2.6802, "step": 43134 }, { "epoch": 2.677695698056987, "grad_norm": 0.4394067867008006, "learning_rate": 3.474997472333963e-06, "loss": 2.7393, "step": 43135 }, { "epoch": 2.6777577751567447, "grad_norm": 0.3960137416232657, "learning_rate": 3.4736747221408907e-06, "loss": 2.6661, "step": 43136 }, { "epoch": 2.6778198522565027, "grad_norm": 0.3634800520805757, "learning_rate": 3.4723522146869527e-06, "loss": 2.6131, "step": 43137 }, { "epoch": 2.6778819293562606, "grad_norm": 0.4046611526926388, "learning_rate": 3.4710299499790545e-06, "loss": 2.6809, "step": 43138 }, { "epoch": 2.6779440064560185, "grad_norm": 0.40473792303256373, "learning_rate": 3.4697079280240853e-06, "loss": 2.6867, "step": 43139 }, { "epoch": 2.6780060835557764, "grad_norm": 0.4160922163542196, "learning_rate": 3.468386148828934e-06, "loss": 2.6894, "step": 43140 }, { "epoch": 2.6780681606555343, "grad_norm": 0.3728769324710092, "learning_rate": 3.4670646124005223e-06, "loss": 2.6474, "step": 43141 }, { "epoch": 2.6781302377552922, "grad_norm": 0.3788705396046709, "learning_rate": 3.4657433187457235e-06, "loss": 2.6306, "step": 43142 }, { "epoch": 2.6781923148550497, "grad_norm": 0.39482909941882327, "learning_rate": 3.464422267871442e-06, "loss": 2.6977, "step": 43143 }, { "epoch": 2.678254391954808, "grad_norm": 0.40470447357401956, "learning_rate": 3.4631014597845624e-06, "loss": 2.6694, "step": 43144 }, { "epoch": 2.6783164690545656, "grad_norm": 0.36859870166694797, "learning_rate": 3.461780894491984e-06, "loss": 2.5595, "step": 43145 }, { "epoch": 2.678378546154324, "grad_norm": 0.361698785317406, "learning_rate": 3.4604605720005902e-06, "loss": 2.6039, "step": 43146 }, { "epoch": 2.6784406232540814, "grad_norm": 0.39234518855297196, "learning_rate": 3.4591404923172756e-06, "loss": 2.6603, "step": 43147 }, { "epoch": 2.6785027003538393, "grad_norm": 0.37492546691890893, "learning_rate": 3.4578206554489235e-06, "loss": 2.7758, "step": 43148 }, { "epoch": 2.6785647774535972, "grad_norm": 0.40685679894563576, "learning_rate": 3.456501061402406e-06, "loss": 2.6788, "step": 43149 }, { "epoch": 2.678626854553355, "grad_norm": 0.37178012214586403, "learning_rate": 3.4551817101846353e-06, "loss": 2.6476, "step": 43150 }, { "epoch": 2.678688931653113, "grad_norm": 0.387134347951211, "learning_rate": 3.4538626018024767e-06, "loss": 2.7294, "step": 43151 }, { "epoch": 2.678751008752871, "grad_norm": 0.3878949513727111, "learning_rate": 3.4525437362628197e-06, "loss": 2.6785, "step": 43152 }, { "epoch": 2.678813085852629, "grad_norm": 0.40608081595947026, "learning_rate": 3.4512251135725316e-06, "loss": 2.7007, "step": 43153 }, { "epoch": 2.678875162952387, "grad_norm": 0.3752408843435893, "learning_rate": 3.4499067337385115e-06, "loss": 2.6805, "step": 43154 }, { "epoch": 2.6789372400521447, "grad_norm": 0.38900334394900515, "learning_rate": 3.4485885967676266e-06, "loss": 2.6764, "step": 43155 }, { "epoch": 2.6789993171519026, "grad_norm": 0.38167939550435265, "learning_rate": 3.4472707026667604e-06, "loss": 2.7183, "step": 43156 }, { "epoch": 2.6790613942516606, "grad_norm": 0.3815721932694423, "learning_rate": 3.4459530514427795e-06, "loss": 2.6884, "step": 43157 }, { "epoch": 2.6791234713514185, "grad_norm": 0.38097383784283273, "learning_rate": 3.444635643102556e-06, "loss": 2.5783, "step": 43158 }, { "epoch": 2.6791855484511764, "grad_norm": 0.3788867237526237, "learning_rate": 3.4433184776529736e-06, "loss": 2.6056, "step": 43159 }, { "epoch": 2.6792476255509343, "grad_norm": 0.3942429083740896, "learning_rate": 3.44200155510091e-06, "loss": 2.7131, "step": 43160 }, { "epoch": 2.6793097026506922, "grad_norm": 0.37619244211287095, "learning_rate": 3.4406848754532205e-06, "loss": 2.6173, "step": 43161 }, { "epoch": 2.67937177975045, "grad_norm": 0.3591071664967752, "learning_rate": 3.4393684387167725e-06, "loss": 2.6472, "step": 43162 }, { "epoch": 2.679433856850208, "grad_norm": 0.3837569711095493, "learning_rate": 3.4380522448984544e-06, "loss": 2.7109, "step": 43163 }, { "epoch": 2.679495933949966, "grad_norm": 0.4087405887301915, "learning_rate": 3.4367362940051163e-06, "loss": 2.6578, "step": 43164 }, { "epoch": 2.679558011049724, "grad_norm": 0.3932753297938382, "learning_rate": 3.4354205860436308e-06, "loss": 2.7226, "step": 43165 }, { "epoch": 2.6796200881494814, "grad_norm": 0.4041637994068074, "learning_rate": 3.4341051210208532e-06, "loss": 2.6994, "step": 43166 }, { "epoch": 2.6796821652492397, "grad_norm": 0.36878057611715986, "learning_rate": 3.4327898989436557e-06, "loss": 2.6973, "step": 43167 }, { "epoch": 2.679744242348997, "grad_norm": 0.3633654917183893, "learning_rate": 3.4314749198189054e-06, "loss": 2.6671, "step": 43168 }, { "epoch": 2.6798063194487556, "grad_norm": 0.4002311016591516, "learning_rate": 3.4301601836534413e-06, "loss": 2.6525, "step": 43169 }, { "epoch": 2.679868396548513, "grad_norm": 0.3712944291242175, "learning_rate": 3.428845690454152e-06, "loss": 2.6122, "step": 43170 }, { "epoch": 2.6799304736482714, "grad_norm": 0.40284625403987445, "learning_rate": 3.4275314402278713e-06, "loss": 2.6287, "step": 43171 }, { "epoch": 2.679992550748029, "grad_norm": 0.4048479719394572, "learning_rate": 3.4262174329814656e-06, "loss": 2.7309, "step": 43172 }, { "epoch": 2.680054627847787, "grad_norm": 0.3772779656163541, "learning_rate": 3.4249036687217904e-06, "loss": 2.7334, "step": 43173 }, { "epoch": 2.6801167049475447, "grad_norm": 0.36657802239973447, "learning_rate": 3.423590147455702e-06, "loss": 2.6313, "step": 43174 }, { "epoch": 2.6801787820473026, "grad_norm": 0.37439898170501845, "learning_rate": 3.4222768691900552e-06, "loss": 2.7222, "step": 43175 }, { "epoch": 2.6802408591470606, "grad_norm": 0.4094757429316364, "learning_rate": 3.420963833931684e-06, "loss": 2.6897, "step": 43176 }, { "epoch": 2.6803029362468185, "grad_norm": 0.4167501564620559, "learning_rate": 3.4196510416874606e-06, "loss": 2.6595, "step": 43177 }, { "epoch": 2.6803650133465764, "grad_norm": 0.39328620404354064, "learning_rate": 3.418338492464224e-06, "loss": 2.6364, "step": 43178 }, { "epoch": 2.6804270904463343, "grad_norm": 0.39418033063376967, "learning_rate": 3.417026186268829e-06, "loss": 2.6856, "step": 43179 }, { "epoch": 2.6804891675460922, "grad_norm": 0.4014137036918545, "learning_rate": 3.4157141231081045e-06, "loss": 2.6582, "step": 43180 }, { "epoch": 2.68055124464585, "grad_norm": 0.3831499130836337, "learning_rate": 3.4144023029889173e-06, "loss": 2.6762, "step": 43181 }, { "epoch": 2.680613321745608, "grad_norm": 0.38779002457257344, "learning_rate": 3.4130907259181e-06, "loss": 2.6951, "step": 43182 }, { "epoch": 2.680675398845366, "grad_norm": 0.39540451019393963, "learning_rate": 3.411779391902503e-06, "loss": 2.7047, "step": 43183 }, { "epoch": 2.680737475945124, "grad_norm": 0.3711612238527013, "learning_rate": 3.4104683009489547e-06, "loss": 2.6402, "step": 43184 }, { "epoch": 2.680799553044882, "grad_norm": 0.35826255862710465, "learning_rate": 3.4091574530643042e-06, "loss": 2.7565, "step": 43185 }, { "epoch": 2.6808616301446397, "grad_norm": 0.3764520981903316, "learning_rate": 3.4078468482553918e-06, "loss": 2.686, "step": 43186 }, { "epoch": 2.6809237072443977, "grad_norm": 0.35448756359437, "learning_rate": 3.406536486529055e-06, "loss": 2.6237, "step": 43187 }, { "epoch": 2.6809857843441556, "grad_norm": 0.40133973187843547, "learning_rate": 3.405226367892128e-06, "loss": 2.6878, "step": 43188 }, { "epoch": 2.6810478614439135, "grad_norm": 0.37893440692538344, "learning_rate": 3.403916492351439e-06, "loss": 2.6118, "step": 43189 }, { "epoch": 2.6811099385436714, "grad_norm": 0.4014808326827697, "learning_rate": 3.4026068599138315e-06, "loss": 2.7176, "step": 43190 }, { "epoch": 2.681172015643429, "grad_norm": 0.36279594656336994, "learning_rate": 3.4012974705861454e-06, "loss": 2.7191, "step": 43191 }, { "epoch": 2.6812340927431872, "grad_norm": 0.3529724774917405, "learning_rate": 3.399988324375197e-06, "loss": 2.5927, "step": 43192 }, { "epoch": 2.6812961698429447, "grad_norm": 0.37937259487916847, "learning_rate": 3.398679421287815e-06, "loss": 2.7212, "step": 43193 }, { "epoch": 2.681358246942703, "grad_norm": 0.40352004972914185, "learning_rate": 3.3973707613308426e-06, "loss": 2.5729, "step": 43194 }, { "epoch": 2.6814203240424606, "grad_norm": 0.3741354911421703, "learning_rate": 3.396062344511103e-06, "loss": 2.7225, "step": 43195 }, { "epoch": 2.6814824011422185, "grad_norm": 0.34803420320984046, "learning_rate": 3.394754170835418e-06, "loss": 2.6568, "step": 43196 }, { "epoch": 2.6815444782419764, "grad_norm": 0.40531832894491926, "learning_rate": 3.3934462403106104e-06, "loss": 2.7204, "step": 43197 }, { "epoch": 2.6816065553417343, "grad_norm": 0.3932399390389309, "learning_rate": 3.3921385529435026e-06, "loss": 2.7314, "step": 43198 }, { "epoch": 2.6816686324414922, "grad_norm": 0.37750048255257995, "learning_rate": 3.3908311087409327e-06, "loss": 2.6635, "step": 43199 }, { "epoch": 2.68173070954125, "grad_norm": 0.39240323626396956, "learning_rate": 3.389523907709713e-06, "loss": 2.757, "step": 43200 }, { "epoch": 2.681792786641008, "grad_norm": 0.3867809168621625, "learning_rate": 3.3882169498566595e-06, "loss": 2.8274, "step": 43201 }, { "epoch": 2.681854863740766, "grad_norm": 0.3714845105801725, "learning_rate": 3.386910235188584e-06, "loss": 2.6759, "step": 43202 }, { "epoch": 2.681916940840524, "grad_norm": 0.4030121926375338, "learning_rate": 3.385603763712314e-06, "loss": 2.6559, "step": 43203 }, { "epoch": 2.681979017940282, "grad_norm": 0.36802304347209885, "learning_rate": 3.384297535434672e-06, "loss": 2.7006, "step": 43204 }, { "epoch": 2.6820410950400397, "grad_norm": 0.3762648628949183, "learning_rate": 3.382991550362469e-06, "loss": 2.7419, "step": 43205 }, { "epoch": 2.6821031721397977, "grad_norm": 0.3684449882166982, "learning_rate": 3.381685808502516e-06, "loss": 2.6521, "step": 43206 }, { "epoch": 2.6821652492395556, "grad_norm": 0.3581804241279597, "learning_rate": 3.380380309861625e-06, "loss": 2.6304, "step": 43207 }, { "epoch": 2.6822273263393135, "grad_norm": 0.3570426012567765, "learning_rate": 3.379075054446601e-06, "loss": 2.7056, "step": 43208 }, { "epoch": 2.6822894034390714, "grad_norm": 0.40748732380370695, "learning_rate": 3.377770042264261e-06, "loss": 2.7439, "step": 43209 }, { "epoch": 2.6823514805388293, "grad_norm": 0.3756667607113253, "learning_rate": 3.3764652733214163e-06, "loss": 2.6056, "step": 43210 }, { "epoch": 2.6824135576385872, "grad_norm": 0.3827165366580915, "learning_rate": 3.3751607476248726e-06, "loss": 2.7172, "step": 43211 }, { "epoch": 2.682475634738345, "grad_norm": 0.3743387392280166, "learning_rate": 3.3738564651814242e-06, "loss": 2.7089, "step": 43212 }, { "epoch": 2.682537711838103, "grad_norm": 0.38934970580678124, "learning_rate": 3.3725524259978936e-06, "loss": 2.7016, "step": 43213 }, { "epoch": 2.6825997889378606, "grad_norm": 0.35788267688465103, "learning_rate": 3.371248630081081e-06, "loss": 2.6014, "step": 43214 }, { "epoch": 2.682661866037619, "grad_norm": 0.38791942124595785, "learning_rate": 3.369945077437775e-06, "loss": 2.6216, "step": 43215 }, { "epoch": 2.6827239431373764, "grad_norm": 0.3627666602472575, "learning_rate": 3.3686417680747817e-06, "loss": 2.5822, "step": 43216 }, { "epoch": 2.6827860202371347, "grad_norm": 0.3799358223063694, "learning_rate": 3.3673387019989123e-06, "loss": 2.6114, "step": 43217 }, { "epoch": 2.6828480973368922, "grad_norm": 0.41070236905503504, "learning_rate": 3.366035879216961e-06, "loss": 2.6768, "step": 43218 }, { "epoch": 2.6829101744366506, "grad_norm": 0.35369143671870557, "learning_rate": 3.364733299735717e-06, "loss": 2.5442, "step": 43219 }, { "epoch": 2.682972251536408, "grad_norm": 0.3978601461316982, "learning_rate": 3.363430963561981e-06, "loss": 2.6611, "step": 43220 }, { "epoch": 2.683034328636166, "grad_norm": 0.38022825328256044, "learning_rate": 3.3621288707025413e-06, "loss": 2.6144, "step": 43221 }, { "epoch": 2.683096405735924, "grad_norm": 0.3817790152421967, "learning_rate": 3.3608270211642034e-06, "loss": 2.7121, "step": 43222 }, { "epoch": 2.683158482835682, "grad_norm": 0.38095965899138384, "learning_rate": 3.3595254149537513e-06, "loss": 2.6157, "step": 43223 }, { "epoch": 2.6832205599354397, "grad_norm": 0.3953146380340296, "learning_rate": 3.3582240520779794e-06, "loss": 2.641, "step": 43224 }, { "epoch": 2.6832826370351976, "grad_norm": 0.37785484499468286, "learning_rate": 3.3569229325436657e-06, "loss": 2.5775, "step": 43225 }, { "epoch": 2.6833447141349556, "grad_norm": 0.34517368357240136, "learning_rate": 3.3556220563576157e-06, "loss": 2.5056, "step": 43226 }, { "epoch": 2.6834067912347135, "grad_norm": 0.3833262246792712, "learning_rate": 3.3543214235266128e-06, "loss": 2.7495, "step": 43227 }, { "epoch": 2.6834688683344714, "grad_norm": 0.37849309360144434, "learning_rate": 3.3530210340574354e-06, "loss": 2.6926, "step": 43228 }, { "epoch": 2.6835309454342293, "grad_norm": 0.3949742992707587, "learning_rate": 3.3517208879568606e-06, "loss": 2.7197, "step": 43229 }, { "epoch": 2.6835930225339872, "grad_norm": 0.3632853107840708, "learning_rate": 3.3504209852316948e-06, "loss": 2.7219, "step": 43230 }, { "epoch": 2.683655099633745, "grad_norm": 0.37572916982531074, "learning_rate": 3.3491213258887044e-06, "loss": 2.6573, "step": 43231 }, { "epoch": 2.683717176733503, "grad_norm": 0.3967610976704838, "learning_rate": 3.3478219099346785e-06, "loss": 2.7093, "step": 43232 }, { "epoch": 2.683779253833261, "grad_norm": 0.39723836679632035, "learning_rate": 3.346522737376384e-06, "loss": 2.672, "step": 43233 }, { "epoch": 2.683841330933019, "grad_norm": 0.3842723414714721, "learning_rate": 3.3452238082206044e-06, "loss": 2.617, "step": 43234 }, { "epoch": 2.683903408032777, "grad_norm": 0.37113621117453294, "learning_rate": 3.343925122474112e-06, "loss": 2.676, "step": 43235 }, { "epoch": 2.6839654851325347, "grad_norm": 0.3798285791277549, "learning_rate": 3.342626680143701e-06, "loss": 2.6391, "step": 43236 }, { "epoch": 2.6840275622322927, "grad_norm": 0.3979646644637188, "learning_rate": 3.3413284812361278e-06, "loss": 2.6576, "step": 43237 }, { "epoch": 2.6840896393320506, "grad_norm": 0.3696707496492111, "learning_rate": 3.3400305257581753e-06, "loss": 2.6642, "step": 43238 }, { "epoch": 2.684151716431808, "grad_norm": 0.4005925297485307, "learning_rate": 3.3387328137165994e-06, "loss": 2.7523, "step": 43239 }, { "epoch": 2.6842137935315664, "grad_norm": 0.37153837772154436, "learning_rate": 3.337435345118195e-06, "loss": 2.6896, "step": 43240 }, { "epoch": 2.684275870631324, "grad_norm": 0.4196322839790533, "learning_rate": 3.3361381199697117e-06, "loss": 2.6549, "step": 43241 }, { "epoch": 2.6843379477310823, "grad_norm": 0.3667140302447817, "learning_rate": 3.3348411382779278e-06, "loss": 2.64, "step": 43242 }, { "epoch": 2.6844000248308397, "grad_norm": 0.35379184027271343, "learning_rate": 3.333544400049604e-06, "loss": 2.7273, "step": 43243 }, { "epoch": 2.6844621019305976, "grad_norm": 0.39253025806934266, "learning_rate": 3.3322479052915022e-06, "loss": 2.7109, "step": 43244 }, { "epoch": 2.6845241790303556, "grad_norm": 0.38903507917359614, "learning_rate": 3.3309516540104004e-06, "loss": 2.6453, "step": 43245 }, { "epoch": 2.6845862561301135, "grad_norm": 0.38619157713481783, "learning_rate": 3.3296556462130477e-06, "loss": 2.6216, "step": 43246 }, { "epoch": 2.6846483332298714, "grad_norm": 0.378522226061507, "learning_rate": 3.328359881906218e-06, "loss": 2.7125, "step": 43247 }, { "epoch": 2.6847104103296293, "grad_norm": 0.40610145640794276, "learning_rate": 3.3270643610966546e-06, "loss": 2.6675, "step": 43248 }, { "epoch": 2.6847724874293872, "grad_norm": 0.40924313947744834, "learning_rate": 3.3257690837911302e-06, "loss": 2.7157, "step": 43249 }, { "epoch": 2.684834564529145, "grad_norm": 0.3639915659157926, "learning_rate": 3.3244740499964067e-06, "loss": 2.684, "step": 43250 }, { "epoch": 2.684896641628903, "grad_norm": 0.3848983419574894, "learning_rate": 3.323179259719228e-06, "loss": 2.6567, "step": 43251 }, { "epoch": 2.684958718728661, "grad_norm": 0.3935800738228291, "learning_rate": 3.3218847129663442e-06, "loss": 2.6155, "step": 43252 }, { "epoch": 2.685020795828419, "grad_norm": 0.40426171191346794, "learning_rate": 3.3205904097445338e-06, "loss": 2.7395, "step": 43253 }, { "epoch": 2.685082872928177, "grad_norm": 0.36205229489368707, "learning_rate": 3.31929635006053e-06, "loss": 2.715, "step": 43254 }, { "epoch": 2.6851449500279347, "grad_norm": 0.3855105706681638, "learning_rate": 3.3180025339210886e-06, "loss": 2.641, "step": 43255 }, { "epoch": 2.6852070271276927, "grad_norm": 0.3944087074250983, "learning_rate": 3.316708961332965e-06, "loss": 2.6821, "step": 43256 }, { "epoch": 2.6852691042274506, "grad_norm": 0.35445117975970397, "learning_rate": 3.315415632302893e-06, "loss": 2.6416, "step": 43257 }, { "epoch": 2.6853311813272085, "grad_norm": 0.3842881624087604, "learning_rate": 3.314122546837639e-06, "loss": 2.7976, "step": 43258 }, { "epoch": 2.6853932584269664, "grad_norm": 0.39043954261451874, "learning_rate": 3.312829704943943e-06, "loss": 2.6146, "step": 43259 }, { "epoch": 2.6854553355267243, "grad_norm": 0.40670803731555766, "learning_rate": 3.3115371066285426e-06, "loss": 2.6876, "step": 43260 }, { "epoch": 2.6855174126264822, "grad_norm": 0.4120008316645052, "learning_rate": 3.310244751898184e-06, "loss": 2.6404, "step": 43261 }, { "epoch": 2.6855794897262397, "grad_norm": 0.3847858017021998, "learning_rate": 3.308952640759616e-06, "loss": 2.6727, "step": 43262 }, { "epoch": 2.685641566825998, "grad_norm": 0.36364715361029937, "learning_rate": 3.307660773219584e-06, "loss": 2.6599, "step": 43263 }, { "epoch": 2.6857036439257556, "grad_norm": 0.38008863271868837, "learning_rate": 3.3063691492848157e-06, "loss": 2.661, "step": 43264 }, { "epoch": 2.685765721025514, "grad_norm": 0.37658242295615146, "learning_rate": 3.3050777689620504e-06, "loss": 2.6649, "step": 43265 }, { "epoch": 2.6858277981252714, "grad_norm": 0.3925434685588026, "learning_rate": 3.3037866322580323e-06, "loss": 2.6986, "step": 43266 }, { "epoch": 2.6858898752250298, "grad_norm": 0.3957966820998488, "learning_rate": 3.3024957391794897e-06, "loss": 2.6825, "step": 43267 }, { "epoch": 2.6859519523247872, "grad_norm": 0.3976350632515082, "learning_rate": 3.3012050897331726e-06, "loss": 2.6468, "step": 43268 }, { "epoch": 2.686014029424545, "grad_norm": 0.35414280764304584, "learning_rate": 3.299914683925803e-06, "loss": 2.5718, "step": 43269 }, { "epoch": 2.686076106524303, "grad_norm": 0.3711631308496958, "learning_rate": 3.298624521764121e-06, "loss": 2.7022, "step": 43270 }, { "epoch": 2.686138183624061, "grad_norm": 0.3766170644205812, "learning_rate": 3.297334603254837e-06, "loss": 2.5855, "step": 43271 }, { "epoch": 2.686200260723819, "grad_norm": 0.37085073444529804, "learning_rate": 3.2960449284047066e-06, "loss": 2.644, "step": 43272 }, { "epoch": 2.686262337823577, "grad_norm": 0.42994564278376957, "learning_rate": 3.2947554972204475e-06, "loss": 2.6719, "step": 43273 }, { "epoch": 2.6863244149233347, "grad_norm": 0.39006110498209146, "learning_rate": 3.293466309708787e-06, "loss": 2.6501, "step": 43274 }, { "epoch": 2.6863864920230927, "grad_norm": 0.3586309027280775, "learning_rate": 3.2921773658764423e-06, "loss": 2.6184, "step": 43275 }, { "epoch": 2.6864485691228506, "grad_norm": 0.3782990299475603, "learning_rate": 3.2908886657301576e-06, "loss": 2.7066, "step": 43276 }, { "epoch": 2.6865106462226085, "grad_norm": 0.4126614854185128, "learning_rate": 3.2896002092766444e-06, "loss": 2.6603, "step": 43277 }, { "epoch": 2.6865727233223664, "grad_norm": 0.39539701063558147, "learning_rate": 3.288311996522625e-06, "loss": 2.6897, "step": 43278 }, { "epoch": 2.6866348004221243, "grad_norm": 0.4037835368231985, "learning_rate": 3.2870240274748167e-06, "loss": 2.7663, "step": 43279 }, { "epoch": 2.6866968775218822, "grad_norm": 0.37040866053968424, "learning_rate": 3.2857363021399467e-06, "loss": 2.5557, "step": 43280 }, { "epoch": 2.68675895462164, "grad_norm": 0.36017796746765374, "learning_rate": 3.284448820524738e-06, "loss": 2.7525, "step": 43281 }, { "epoch": 2.686821031721398, "grad_norm": 0.37704632649510833, "learning_rate": 3.2831615826358964e-06, "loss": 2.6502, "step": 43282 }, { "epoch": 2.686883108821156, "grad_norm": 0.3995955038786208, "learning_rate": 3.281874588480138e-06, "loss": 2.5657, "step": 43283 }, { "epoch": 2.686945185920914, "grad_norm": 0.37984409478194275, "learning_rate": 3.2805878380641807e-06, "loss": 2.6796, "step": 43284 }, { "epoch": 2.687007263020672, "grad_norm": 0.4013690994848297, "learning_rate": 3.2793013313947406e-06, "loss": 2.7049, "step": 43285 }, { "epoch": 2.6870693401204298, "grad_norm": 0.39728478504589465, "learning_rate": 3.2780150684785294e-06, "loss": 2.6781, "step": 43286 }, { "epoch": 2.6871314172201872, "grad_norm": 0.37007680385773345, "learning_rate": 3.2767290493222526e-06, "loss": 2.6643, "step": 43287 }, { "epoch": 2.6871934943199456, "grad_norm": 0.3755321898663297, "learning_rate": 3.275443273932621e-06, "loss": 2.6081, "step": 43288 }, { "epoch": 2.687255571419703, "grad_norm": 0.3791317612786857, "learning_rate": 3.274157742316347e-06, "loss": 2.6314, "step": 43289 }, { "epoch": 2.6873176485194614, "grad_norm": 0.38001072551623805, "learning_rate": 3.272872454480136e-06, "loss": 2.6501, "step": 43290 }, { "epoch": 2.687379725619219, "grad_norm": 0.370839121948382, "learning_rate": 3.271587410430693e-06, "loss": 2.6018, "step": 43291 }, { "epoch": 2.687441802718977, "grad_norm": 0.3914578904370531, "learning_rate": 3.270302610174725e-06, "loss": 2.6126, "step": 43292 }, { "epoch": 2.6875038798187347, "grad_norm": 0.396560755546226, "learning_rate": 3.2690180537189253e-06, "loss": 2.6932, "step": 43293 }, { "epoch": 2.6875659569184926, "grad_norm": 0.3555561628911838, "learning_rate": 3.267733741070006e-06, "loss": 2.6734, "step": 43294 }, { "epoch": 2.6876280340182506, "grad_norm": 0.36120795236226333, "learning_rate": 3.266449672234667e-06, "loss": 2.7343, "step": 43295 }, { "epoch": 2.6876901111180085, "grad_norm": 0.39475536590538535, "learning_rate": 3.2651658472196034e-06, "loss": 2.592, "step": 43296 }, { "epoch": 2.6877521882177664, "grad_norm": 0.36107238792177715, "learning_rate": 3.2638822660315092e-06, "loss": 2.6652, "step": 43297 }, { "epoch": 2.6878142653175243, "grad_norm": 0.35535273912060356, "learning_rate": 3.262598928677091e-06, "loss": 2.6565, "step": 43298 }, { "epoch": 2.6878763424172822, "grad_norm": 0.35992296570202137, "learning_rate": 3.2613158351630424e-06, "loss": 2.6647, "step": 43299 }, { "epoch": 2.68793841951704, "grad_norm": 0.37863875316452666, "learning_rate": 3.2600329854960477e-06, "loss": 2.7041, "step": 43300 }, { "epoch": 2.688000496616798, "grad_norm": 0.37978697207686973, "learning_rate": 3.258750379682818e-06, "loss": 2.6609, "step": 43301 }, { "epoch": 2.688062573716556, "grad_norm": 0.3676932686627648, "learning_rate": 3.2574680177300197e-06, "loss": 2.7197, "step": 43302 }, { "epoch": 2.688124650816314, "grad_norm": 0.37144124676225065, "learning_rate": 3.2561858996443706e-06, "loss": 2.5724, "step": 43303 }, { "epoch": 2.688186727916072, "grad_norm": 0.37148972516464124, "learning_rate": 3.254904025432548e-06, "loss": 2.675, "step": 43304 }, { "epoch": 2.6882488050158297, "grad_norm": 0.36822893699340276, "learning_rate": 3.2536223951012358e-06, "loss": 2.6779, "step": 43305 }, { "epoch": 2.6883108821155877, "grad_norm": 0.3586971067520347, "learning_rate": 3.252341008657128e-06, "loss": 2.685, "step": 43306 }, { "epoch": 2.6883729592153456, "grad_norm": 0.41278669552427916, "learning_rate": 3.2510598661068926e-06, "loss": 2.6946, "step": 43307 }, { "epoch": 2.6884350363151035, "grad_norm": 0.4127131373847809, "learning_rate": 3.249778967457234e-06, "loss": 2.6759, "step": 43308 }, { "epoch": 2.6884971134148614, "grad_norm": 0.39945271125472104, "learning_rate": 3.2484983127148316e-06, "loss": 2.7449, "step": 43309 }, { "epoch": 2.688559190514619, "grad_norm": 0.3652804136584045, "learning_rate": 3.2472179018863624e-06, "loss": 2.6596, "step": 43310 }, { "epoch": 2.6886212676143773, "grad_norm": 0.3716715280075208, "learning_rate": 3.2459377349784992e-06, "loss": 2.6178, "step": 43311 }, { "epoch": 2.6886833447141347, "grad_norm": 0.3874915906547396, "learning_rate": 3.2446578119979365e-06, "loss": 2.7133, "step": 43312 }, { "epoch": 2.688745421813893, "grad_norm": 0.3817916415148371, "learning_rate": 3.2433781329513413e-06, "loss": 2.66, "step": 43313 }, { "epoch": 2.6888074989136506, "grad_norm": 0.3807220407002608, "learning_rate": 3.242098697845397e-06, "loss": 2.7199, "step": 43314 }, { "epoch": 2.688869576013409, "grad_norm": 0.40493267394527055, "learning_rate": 3.240819506686765e-06, "loss": 2.6484, "step": 43315 }, { "epoch": 2.6889316531131664, "grad_norm": 0.3541392311053612, "learning_rate": 3.239540559482135e-06, "loss": 2.7618, "step": 43316 }, { "epoch": 2.6889937302129243, "grad_norm": 0.36554502851554266, "learning_rate": 3.2382618562381726e-06, "loss": 2.6665, "step": 43317 }, { "epoch": 2.6890558073126822, "grad_norm": 0.3683196935386526, "learning_rate": 3.2369833969615514e-06, "loss": 2.6898, "step": 43318 }, { "epoch": 2.68911788441244, "grad_norm": 0.38802563794466216, "learning_rate": 3.2357051816589436e-06, "loss": 2.6839, "step": 43319 }, { "epoch": 2.689179961512198, "grad_norm": 0.39073019518607116, "learning_rate": 3.2344272103370045e-06, "loss": 2.6674, "step": 43320 }, { "epoch": 2.689242038611956, "grad_norm": 0.3886602773805079, "learning_rate": 3.233149483002418e-06, "loss": 2.6171, "step": 43321 }, { "epoch": 2.689304115711714, "grad_norm": 0.3637902375542968, "learning_rate": 3.231871999661845e-06, "loss": 2.7041, "step": 43322 }, { "epoch": 2.689366192811472, "grad_norm": 0.385105273604855, "learning_rate": 3.2305947603219476e-06, "loss": 2.6049, "step": 43323 }, { "epoch": 2.6894282699112297, "grad_norm": 0.3701299096490105, "learning_rate": 3.2293177649893813e-06, "loss": 2.5515, "step": 43324 }, { "epoch": 2.6894903470109877, "grad_norm": 0.4019213375649568, "learning_rate": 3.2280410136708296e-06, "loss": 2.6197, "step": 43325 }, { "epoch": 2.6895524241107456, "grad_norm": 0.38876450934732387, "learning_rate": 3.226764506372942e-06, "loss": 2.6945, "step": 43326 }, { "epoch": 2.6896145012105035, "grad_norm": 0.3736446245143973, "learning_rate": 3.225488243102376e-06, "loss": 2.6647, "step": 43327 }, { "epoch": 2.6896765783102614, "grad_norm": 0.36659812295607974, "learning_rate": 3.2242122238657913e-06, "loss": 2.6957, "step": 43328 }, { "epoch": 2.6897386554100193, "grad_norm": 0.3773510073325244, "learning_rate": 3.2229364486698444e-06, "loss": 2.6432, "step": 43329 }, { "epoch": 2.6898007325097772, "grad_norm": 0.39610400064360207, "learning_rate": 3.2216609175211963e-06, "loss": 2.5578, "step": 43330 }, { "epoch": 2.689862809609535, "grad_norm": 0.38442445529889974, "learning_rate": 3.2203856304265034e-06, "loss": 2.7004, "step": 43331 }, { "epoch": 2.689924886709293, "grad_norm": 0.37089047134627007, "learning_rate": 3.2191105873924153e-06, "loss": 2.6638, "step": 43332 }, { "epoch": 2.689986963809051, "grad_norm": 0.38856339977663984, "learning_rate": 3.2178357884255715e-06, "loss": 2.5785, "step": 43333 }, { "epoch": 2.690049040908809, "grad_norm": 0.38844645495231483, "learning_rate": 3.216561233532639e-06, "loss": 2.6211, "step": 43334 }, { "epoch": 2.6901111180085664, "grad_norm": 0.3457829334788619, "learning_rate": 3.2152869227202674e-06, "loss": 2.6204, "step": 43335 }, { "epoch": 2.6901731951083248, "grad_norm": 0.3815632502543004, "learning_rate": 3.214012855995102e-06, "loss": 2.6912, "step": 43336 }, { "epoch": 2.6902352722080822, "grad_norm": 0.37932396506184224, "learning_rate": 3.212739033363793e-06, "loss": 2.7457, "step": 43337 }, { "epoch": 2.6902973493078406, "grad_norm": 0.3792573540916898, "learning_rate": 3.211465454832974e-06, "loss": 2.633, "step": 43338 }, { "epoch": 2.690359426407598, "grad_norm": 0.3747793979272984, "learning_rate": 3.2101921204093054e-06, "loss": 2.6717, "step": 43339 }, { "epoch": 2.690421503507356, "grad_norm": 0.36156785590079776, "learning_rate": 3.2089190300994223e-06, "loss": 2.6784, "step": 43340 }, { "epoch": 2.690483580607114, "grad_norm": 0.38455913056661784, "learning_rate": 3.2076461839099623e-06, "loss": 2.7332, "step": 43341 }, { "epoch": 2.690545657706872, "grad_norm": 0.41788763524795813, "learning_rate": 3.2063735818475773e-06, "loss": 2.6204, "step": 43342 }, { "epoch": 2.6906077348066297, "grad_norm": 0.3778734584250696, "learning_rate": 3.205101223918894e-06, "loss": 2.727, "step": 43343 }, { "epoch": 2.6906698119063877, "grad_norm": 0.3749100656750051, "learning_rate": 3.203829110130563e-06, "loss": 2.6134, "step": 43344 }, { "epoch": 2.6907318890061456, "grad_norm": 0.37566717720307324, "learning_rate": 3.202557240489218e-06, "loss": 2.7069, "step": 43345 }, { "epoch": 2.6907939661059035, "grad_norm": 0.39126422079280754, "learning_rate": 3.201285615001487e-06, "loss": 2.6227, "step": 43346 }, { "epoch": 2.6908560432056614, "grad_norm": 0.3902420073747052, "learning_rate": 3.2000142336740092e-06, "loss": 2.6602, "step": 43347 }, { "epoch": 2.6909181203054193, "grad_norm": 0.38061997151928995, "learning_rate": 3.198743096513418e-06, "loss": 2.5784, "step": 43348 }, { "epoch": 2.6909801974051772, "grad_norm": 0.40746269423909104, "learning_rate": 3.1974722035263526e-06, "loss": 2.6834, "step": 43349 }, { "epoch": 2.691042274504935, "grad_norm": 0.3535146449797308, "learning_rate": 3.1962015547194303e-06, "loss": 2.6109, "step": 43350 }, { "epoch": 2.691104351604693, "grad_norm": 0.3882366983808826, "learning_rate": 3.194931150099284e-06, "loss": 2.6952, "step": 43351 }, { "epoch": 2.691166428704451, "grad_norm": 0.40541971163702323, "learning_rate": 3.193660989672548e-06, "loss": 2.7023, "step": 43352 }, { "epoch": 2.691228505804209, "grad_norm": 0.3818139657725993, "learning_rate": 3.1923910734458496e-06, "loss": 2.6188, "step": 43353 }, { "epoch": 2.691290582903967, "grad_norm": 0.38046232214164416, "learning_rate": 3.1911214014258063e-06, "loss": 2.7155, "step": 43354 }, { "epoch": 2.6913526600037248, "grad_norm": 0.38189770913536997, "learning_rate": 3.189851973619046e-06, "loss": 2.7238, "step": 43355 }, { "epoch": 2.6914147371034827, "grad_norm": 0.3703436045721623, "learning_rate": 3.188582790032185e-06, "loss": 2.6474, "step": 43356 }, { "epoch": 2.6914768142032406, "grad_norm": 0.39195019050824176, "learning_rate": 3.187313850671858e-06, "loss": 2.6893, "step": 43357 }, { "epoch": 2.691538891302998, "grad_norm": 0.38545531216606693, "learning_rate": 3.1860451555446812e-06, "loss": 2.7085, "step": 43358 }, { "epoch": 2.6916009684027564, "grad_norm": 0.3669865037803806, "learning_rate": 3.184776704657272e-06, "loss": 2.7027, "step": 43359 }, { "epoch": 2.691663045502514, "grad_norm": 0.36471036559845094, "learning_rate": 3.1835084980162357e-06, "loss": 2.5644, "step": 43360 }, { "epoch": 2.6917251226022723, "grad_norm": 0.3902920257476027, "learning_rate": 3.1822405356282113e-06, "loss": 2.5947, "step": 43361 }, { "epoch": 2.6917871997020297, "grad_norm": 0.37104335241918285, "learning_rate": 3.1809728174998054e-06, "loss": 2.7467, "step": 43362 }, { "epoch": 2.691849276801788, "grad_norm": 0.4140534130035002, "learning_rate": 3.1797053436376234e-06, "loss": 2.5792, "step": 43363 }, { "epoch": 2.6919113539015456, "grad_norm": 0.3742071950319128, "learning_rate": 3.178438114048282e-06, "loss": 2.6907, "step": 43364 }, { "epoch": 2.6919734310013035, "grad_norm": 0.3715068726090023, "learning_rate": 3.177171128738404e-06, "loss": 2.6679, "step": 43365 }, { "epoch": 2.6920355081010614, "grad_norm": 0.39855534549828187, "learning_rate": 3.175904387714579e-06, "loss": 2.6039, "step": 43366 }, { "epoch": 2.6920975852008193, "grad_norm": 0.36430473127735225, "learning_rate": 3.1746378909834396e-06, "loss": 2.6681, "step": 43367 }, { "epoch": 2.6921596623005772, "grad_norm": 0.38433319810413874, "learning_rate": 3.1733716385515754e-06, "loss": 2.7857, "step": 43368 }, { "epoch": 2.692221739400335, "grad_norm": 0.3858552018731788, "learning_rate": 3.1721056304256035e-06, "loss": 2.6864, "step": 43369 }, { "epoch": 2.692283816500093, "grad_norm": 0.4242353811920531, "learning_rate": 3.1708398666121187e-06, "loss": 2.7384, "step": 43370 }, { "epoch": 2.692345893599851, "grad_norm": 0.4203912285454995, "learning_rate": 3.169574347117732e-06, "loss": 2.7658, "step": 43371 }, { "epoch": 2.692407970699609, "grad_norm": 0.37166846093498795, "learning_rate": 3.1683090719490494e-06, "loss": 2.6907, "step": 43372 }, { "epoch": 2.692470047799367, "grad_norm": 0.35490611896813096, "learning_rate": 3.1670440411126655e-06, "loss": 2.6061, "step": 43373 }, { "epoch": 2.6925321248991247, "grad_norm": 0.3843598913309232, "learning_rate": 3.1657792546151756e-06, "loss": 2.766, "step": 43374 }, { "epoch": 2.6925942019988827, "grad_norm": 0.38064796253211747, "learning_rate": 3.16451471246319e-06, "loss": 2.6562, "step": 43375 }, { "epoch": 2.6926562790986406, "grad_norm": 0.3752499581793775, "learning_rate": 3.1632504146633045e-06, "loss": 2.5728, "step": 43376 }, { "epoch": 2.6927183561983985, "grad_norm": 0.3862160391830869, "learning_rate": 3.161986361222108e-06, "loss": 2.7044, "step": 43377 }, { "epoch": 2.6927804332981564, "grad_norm": 0.36870381583315476, "learning_rate": 3.1607225521462e-06, "loss": 2.6667, "step": 43378 }, { "epoch": 2.6928425103979143, "grad_norm": 0.38104466552694466, "learning_rate": 3.1594589874421656e-06, "loss": 2.697, "step": 43379 }, { "epoch": 2.6929045874976723, "grad_norm": 0.375691570914219, "learning_rate": 3.1581956671166146e-06, "loss": 2.6255, "step": 43380 }, { "epoch": 2.69296666459743, "grad_norm": 0.3604826649930379, "learning_rate": 3.156932591176126e-06, "loss": 2.6883, "step": 43381 }, { "epoch": 2.693028741697188, "grad_norm": 0.3658170489131527, "learning_rate": 3.1556697596272887e-06, "loss": 2.715, "step": 43382 }, { "epoch": 2.6930908187969456, "grad_norm": 0.4014862515089635, "learning_rate": 3.1544071724766864e-06, "loss": 2.679, "step": 43383 }, { "epoch": 2.693152895896704, "grad_norm": 0.3786025845972437, "learning_rate": 3.1531448297309253e-06, "loss": 2.6665, "step": 43384 }, { "epoch": 2.6932149729964614, "grad_norm": 0.3614661494389322, "learning_rate": 3.1518827313965825e-06, "loss": 2.6624, "step": 43385 }, { "epoch": 2.6932770500962198, "grad_norm": 0.3615267851853314, "learning_rate": 3.150620877480237e-06, "loss": 2.623, "step": 43386 }, { "epoch": 2.6933391271959772, "grad_norm": 0.3788893784375041, "learning_rate": 3.1493592679884663e-06, "loss": 2.6002, "step": 43387 }, { "epoch": 2.693401204295735, "grad_norm": 0.366433343109335, "learning_rate": 3.148097902927871e-06, "loss": 2.6272, "step": 43388 }, { "epoch": 2.693463281395493, "grad_norm": 0.3903881308502839, "learning_rate": 3.1468367823050237e-06, "loss": 2.7049, "step": 43389 }, { "epoch": 2.693525358495251, "grad_norm": 0.3903156008875917, "learning_rate": 3.1455759061265023e-06, "loss": 2.6757, "step": 43390 }, { "epoch": 2.693587435595009, "grad_norm": 0.37781891690366776, "learning_rate": 3.1443152743988846e-06, "loss": 2.6607, "step": 43391 }, { "epoch": 2.693649512694767, "grad_norm": 0.3837346045608607, "learning_rate": 3.143054887128738e-06, "loss": 2.6872, "step": 43392 }, { "epoch": 2.6937115897945247, "grad_norm": 0.37166956348031555, "learning_rate": 3.1417947443226625e-06, "loss": 2.6787, "step": 43393 }, { "epoch": 2.6937736668942827, "grad_norm": 0.3517935311643307, "learning_rate": 3.1405348459872143e-06, "loss": 2.682, "step": 43394 }, { "epoch": 2.6938357439940406, "grad_norm": 0.40018539751541066, "learning_rate": 3.1392751921289764e-06, "loss": 2.7136, "step": 43395 }, { "epoch": 2.6938978210937985, "grad_norm": 0.36739706592805504, "learning_rate": 3.1380157827545053e-06, "loss": 2.6473, "step": 43396 }, { "epoch": 2.6939598981935564, "grad_norm": 0.3826374315375245, "learning_rate": 3.13675661787039e-06, "loss": 2.6652, "step": 43397 }, { "epoch": 2.6940219752933143, "grad_norm": 0.364507042759227, "learning_rate": 3.1354976974831805e-06, "loss": 2.6365, "step": 43398 }, { "epoch": 2.6940840523930722, "grad_norm": 0.38871174516767415, "learning_rate": 3.1342390215994723e-06, "loss": 2.6372, "step": 43399 }, { "epoch": 2.69414612949283, "grad_norm": 0.3935044022232684, "learning_rate": 3.1329805902258093e-06, "loss": 2.7, "step": 43400 }, { "epoch": 2.694208206592588, "grad_norm": 0.3868042437137173, "learning_rate": 3.131722403368759e-06, "loss": 2.7322, "step": 43401 }, { "epoch": 2.694270283692346, "grad_norm": 0.37638668134533854, "learning_rate": 3.1304644610348987e-06, "loss": 2.5711, "step": 43402 }, { "epoch": 2.694332360792104, "grad_norm": 0.3920805705530903, "learning_rate": 3.1292067632307797e-06, "loss": 2.5842, "step": 43403 }, { "epoch": 2.694394437891862, "grad_norm": 0.38753927247170833, "learning_rate": 3.1279493099629687e-06, "loss": 2.6319, "step": 43404 }, { "epoch": 2.6944565149916198, "grad_norm": 0.38257599356215116, "learning_rate": 3.126692101238027e-06, "loss": 2.5915, "step": 43405 }, { "epoch": 2.6945185920913772, "grad_norm": 0.40818330269464614, "learning_rate": 3.1254351370625047e-06, "loss": 2.6651, "step": 43406 }, { "epoch": 2.6945806691911356, "grad_norm": 0.38200452746209845, "learning_rate": 3.124178417442969e-06, "loss": 2.6355, "step": 43407 }, { "epoch": 2.694642746290893, "grad_norm": 0.4506017211742351, "learning_rate": 3.1229219423859755e-06, "loss": 2.6482, "step": 43408 }, { "epoch": 2.6947048233906514, "grad_norm": 0.4058546196344141, "learning_rate": 3.1216657118980806e-06, "loss": 2.6835, "step": 43409 }, { "epoch": 2.694766900490409, "grad_norm": 0.3909060611478908, "learning_rate": 3.120409725985829e-06, "loss": 2.7089, "step": 43410 }, { "epoch": 2.6948289775901673, "grad_norm": 0.38940003013067026, "learning_rate": 3.119153984655787e-06, "loss": 2.6396, "step": 43411 }, { "epoch": 2.6948910546899247, "grad_norm": 0.40019917446672815, "learning_rate": 3.1178984879144944e-06, "loss": 2.6366, "step": 43412 }, { "epoch": 2.6949531317896827, "grad_norm": 0.4054399079518432, "learning_rate": 3.1166432357685128e-06, "loss": 2.6876, "step": 43413 }, { "epoch": 2.6950152088894406, "grad_norm": 0.3586376337862252, "learning_rate": 3.115388228224375e-06, "loss": 2.6484, "step": 43414 }, { "epoch": 2.6950772859891985, "grad_norm": 0.40414127175032144, "learning_rate": 3.1141334652886435e-06, "loss": 2.7509, "step": 43415 }, { "epoch": 2.6951393630889564, "grad_norm": 0.3679543664473855, "learning_rate": 3.112878946967862e-06, "loss": 2.6893, "step": 43416 }, { "epoch": 2.6952014401887143, "grad_norm": 0.3993165286486805, "learning_rate": 3.111624673268576e-06, "loss": 2.7067, "step": 43417 }, { "epoch": 2.6952635172884722, "grad_norm": 0.3912828627852383, "learning_rate": 3.1103706441973245e-06, "loss": 2.5865, "step": 43418 }, { "epoch": 2.69532559438823, "grad_norm": 0.372595129058981, "learning_rate": 3.1091168597606414e-06, "loss": 2.6925, "step": 43419 }, { "epoch": 2.695387671487988, "grad_norm": 0.3617030286565111, "learning_rate": 3.1078633199650877e-06, "loss": 2.7266, "step": 43420 }, { "epoch": 2.695449748587746, "grad_norm": 0.3666404544968923, "learning_rate": 3.106610024817197e-06, "loss": 2.6245, "step": 43421 }, { "epoch": 2.695511825687504, "grad_norm": 0.3604240654680237, "learning_rate": 3.1053569743235034e-06, "loss": 2.6124, "step": 43422 }, { "epoch": 2.695573902787262, "grad_norm": 0.4094203159497818, "learning_rate": 3.1041041684905404e-06, "loss": 2.721, "step": 43423 }, { "epoch": 2.6956359798870198, "grad_norm": 0.37377329696334655, "learning_rate": 3.102851607324858e-06, "loss": 2.7221, "step": 43424 }, { "epoch": 2.6956980569867777, "grad_norm": 0.3963122064516875, "learning_rate": 3.1015992908329847e-06, "loss": 2.7328, "step": 43425 }, { "epoch": 2.6957601340865356, "grad_norm": 0.39628084718509127, "learning_rate": 3.100347219021449e-06, "loss": 2.6647, "step": 43426 }, { "epoch": 2.6958222111862935, "grad_norm": 0.3830940223479413, "learning_rate": 3.099095391896789e-06, "loss": 2.7215, "step": 43427 }, { "epoch": 2.6958842882860514, "grad_norm": 0.3708810296977958, "learning_rate": 3.0978438094655283e-06, "loss": 2.6971, "step": 43428 }, { "epoch": 2.6959463653858093, "grad_norm": 0.39010726951244534, "learning_rate": 3.0965924717342052e-06, "loss": 2.7792, "step": 43429 }, { "epoch": 2.6960084424855673, "grad_norm": 0.37361692643081185, "learning_rate": 3.095341378709349e-06, "loss": 2.6517, "step": 43430 }, { "epoch": 2.6960705195853247, "grad_norm": 0.39103770389015635, "learning_rate": 3.0940905303974756e-06, "loss": 2.6154, "step": 43431 }, { "epoch": 2.696132596685083, "grad_norm": 0.4165784941204804, "learning_rate": 3.0928399268051243e-06, "loss": 2.6595, "step": 43432 }, { "epoch": 2.6961946737848406, "grad_norm": 0.38745576775270246, "learning_rate": 3.091589567938813e-06, "loss": 2.7852, "step": 43433 }, { "epoch": 2.696256750884599, "grad_norm": 0.36524596170416274, "learning_rate": 3.090339453805069e-06, "loss": 2.7332, "step": 43434 }, { "epoch": 2.6963188279843564, "grad_norm": 0.3785998885428118, "learning_rate": 3.0890895844104096e-06, "loss": 2.6889, "step": 43435 }, { "epoch": 2.6963809050841143, "grad_norm": 0.3481808037607843, "learning_rate": 3.087839959761363e-06, "loss": 2.6905, "step": 43436 }, { "epoch": 2.6964429821838722, "grad_norm": 0.38919752604451996, "learning_rate": 3.086590579864429e-06, "loss": 2.7296, "step": 43437 }, { "epoch": 2.69650505928363, "grad_norm": 0.3575584900965733, "learning_rate": 3.085341444726153e-06, "loss": 2.6952, "step": 43438 }, { "epoch": 2.696567136383388, "grad_norm": 0.36898473445205254, "learning_rate": 3.084092554353041e-06, "loss": 2.5947, "step": 43439 }, { "epoch": 2.696629213483146, "grad_norm": 0.36843143943480583, "learning_rate": 3.0828439087516037e-06, "loss": 2.5956, "step": 43440 }, { "epoch": 2.696691290582904, "grad_norm": 0.3585494136619854, "learning_rate": 3.081595507928359e-06, "loss": 2.7091, "step": 43441 }, { "epoch": 2.696753367682662, "grad_norm": 0.35822000661358516, "learning_rate": 3.0803473518898173e-06, "loss": 2.6821, "step": 43442 }, { "epoch": 2.6968154447824197, "grad_norm": 0.3738074812176684, "learning_rate": 3.0790994406424965e-06, "loss": 2.6608, "step": 43443 }, { "epoch": 2.6968775218821777, "grad_norm": 0.3868090591561155, "learning_rate": 3.077851774192908e-06, "loss": 2.6093, "step": 43444 }, { "epoch": 2.6969395989819356, "grad_norm": 0.36065088675027807, "learning_rate": 3.0766043525475573e-06, "loss": 2.6435, "step": 43445 }, { "epoch": 2.6970016760816935, "grad_norm": 0.39331778313049565, "learning_rate": 3.0753571757129396e-06, "loss": 2.6254, "step": 43446 }, { "epoch": 2.6970637531814514, "grad_norm": 0.3847492427533427, "learning_rate": 3.074110243695588e-06, "loss": 2.707, "step": 43447 }, { "epoch": 2.6971258302812093, "grad_norm": 0.3755938080326527, "learning_rate": 3.0728635565019926e-06, "loss": 2.697, "step": 43448 }, { "epoch": 2.6971879073809673, "grad_norm": 0.3883143942366318, "learning_rate": 3.071617114138664e-06, "loss": 2.6206, "step": 43449 }, { "epoch": 2.697249984480725, "grad_norm": 0.38774133596449567, "learning_rate": 3.070370916612092e-06, "loss": 2.7009, "step": 43450 }, { "epoch": 2.697312061580483, "grad_norm": 0.3826106570875832, "learning_rate": 3.0691249639287932e-06, "loss": 2.6398, "step": 43451 }, { "epoch": 2.697374138680241, "grad_norm": 0.3436261409082827, "learning_rate": 3.067879256095263e-06, "loss": 2.7066, "step": 43452 }, { "epoch": 2.697436215779999, "grad_norm": 0.39605883608100484, "learning_rate": 3.0666337931179956e-06, "loss": 2.7424, "step": 43453 }, { "epoch": 2.6974982928797564, "grad_norm": 0.3794001280618491, "learning_rate": 3.065388575003497e-06, "loss": 2.6675, "step": 43454 }, { "epoch": 2.6975603699795148, "grad_norm": 0.3928980306768252, "learning_rate": 3.064143601758257e-06, "loss": 2.6276, "step": 43455 }, { "epoch": 2.6976224470792722, "grad_norm": 0.39699547808783964, "learning_rate": 3.0628988733887754e-06, "loss": 2.5845, "step": 43456 }, { "epoch": 2.6976845241790306, "grad_norm": 0.3690410055737695, "learning_rate": 3.061654389901547e-06, "loss": 2.7258, "step": 43457 }, { "epoch": 2.697746601278788, "grad_norm": 0.35261928881848104, "learning_rate": 3.0604101513030612e-06, "loss": 2.6451, "step": 43458 }, { "epoch": 2.697808678378546, "grad_norm": 0.3709915020270187, "learning_rate": 3.059166157599802e-06, "loss": 2.7358, "step": 43459 }, { "epoch": 2.697870755478304, "grad_norm": 0.35064002977209097, "learning_rate": 3.0579224087982806e-06, "loss": 2.6445, "step": 43460 }, { "epoch": 2.697932832578062, "grad_norm": 0.3824760554492026, "learning_rate": 3.056678904904969e-06, "loss": 2.6603, "step": 43461 }, { "epoch": 2.6979949096778197, "grad_norm": 0.38470107955577865, "learning_rate": 3.0554356459263576e-06, "loss": 2.6776, "step": 43462 }, { "epoch": 2.6980569867775777, "grad_norm": 0.3933770888884933, "learning_rate": 3.0541926318689406e-06, "loss": 2.7419, "step": 43463 }, { "epoch": 2.6981190638773356, "grad_norm": 0.376378783050853, "learning_rate": 3.052949862739185e-06, "loss": 2.6587, "step": 43464 }, { "epoch": 2.6981811409770935, "grad_norm": 0.3824111430422545, "learning_rate": 3.051707338543591e-06, "loss": 2.627, "step": 43465 }, { "epoch": 2.6982432180768514, "grad_norm": 0.40075966180135664, "learning_rate": 3.050465059288643e-06, "loss": 2.7585, "step": 43466 }, { "epoch": 2.6983052951766093, "grad_norm": 0.3901008904611946, "learning_rate": 3.049223024980813e-06, "loss": 2.7199, "step": 43467 }, { "epoch": 2.6983673722763672, "grad_norm": 0.3887860116194061, "learning_rate": 3.0479812356265847e-06, "loss": 2.6919, "step": 43468 }, { "epoch": 2.698429449376125, "grad_norm": 0.374951458809275, "learning_rate": 3.046739691232431e-06, "loss": 2.6689, "step": 43469 }, { "epoch": 2.698491526475883, "grad_norm": 0.38811775464898557, "learning_rate": 3.045498391804846e-06, "loss": 2.5803, "step": 43470 }, { "epoch": 2.698553603575641, "grad_norm": 0.38607709807823987, "learning_rate": 3.044257337350287e-06, "loss": 2.6339, "step": 43471 }, { "epoch": 2.698615680675399, "grad_norm": 0.4370889435630204, "learning_rate": 3.043016527875242e-06, "loss": 2.7963, "step": 43472 }, { "epoch": 2.698677757775157, "grad_norm": 0.4346266120703014, "learning_rate": 3.0417759633861674e-06, "loss": 2.6827, "step": 43473 }, { "epoch": 2.6987398348749148, "grad_norm": 0.3699368160658461, "learning_rate": 3.0405356438895584e-06, "loss": 2.7588, "step": 43474 }, { "epoch": 2.6988019119746727, "grad_norm": 0.37697514286864536, "learning_rate": 3.0392955693918702e-06, "loss": 2.799, "step": 43475 }, { "epoch": 2.6988639890744306, "grad_norm": 0.41459420092474725, "learning_rate": 3.0380557398995814e-06, "loss": 2.5954, "step": 43476 }, { "epoch": 2.698926066174188, "grad_norm": 0.3789653731949298, "learning_rate": 3.036816155419153e-06, "loss": 2.6293, "step": 43477 }, { "epoch": 2.6989881432739464, "grad_norm": 0.4158271935907301, "learning_rate": 3.0355768159570473e-06, "loss": 2.718, "step": 43478 }, { "epoch": 2.699050220373704, "grad_norm": 0.40120736534378765, "learning_rate": 3.0343377215197477e-06, "loss": 2.6325, "step": 43479 }, { "epoch": 2.6991122974734623, "grad_norm": 0.39950344139593447, "learning_rate": 3.03309887211371e-06, "loss": 2.6095, "step": 43480 }, { "epoch": 2.6991743745732197, "grad_norm": 0.3542588572098175, "learning_rate": 3.0318602677453955e-06, "loss": 2.582, "step": 43481 }, { "epoch": 2.699236451672978, "grad_norm": 0.3524445911748725, "learning_rate": 3.030621908421266e-06, "loss": 2.6156, "step": 43482 }, { "epoch": 2.6992985287727356, "grad_norm": 0.3800789490991736, "learning_rate": 3.029383794147783e-06, "loss": 2.6786, "step": 43483 }, { "epoch": 2.6993606058724935, "grad_norm": 0.41110852819925836, "learning_rate": 3.0281459249314135e-06, "loss": 2.6313, "step": 43484 }, { "epoch": 2.6994226829722514, "grad_norm": 0.4070801506028689, "learning_rate": 3.026908300778608e-06, "loss": 2.6926, "step": 43485 }, { "epoch": 2.6994847600720093, "grad_norm": 0.4003088146459961, "learning_rate": 3.0256709216958168e-06, "loss": 2.6312, "step": 43486 }, { "epoch": 2.6995468371717672, "grad_norm": 0.375968120529717, "learning_rate": 3.0244337876895124e-06, "loss": 2.6931, "step": 43487 }, { "epoch": 2.699608914271525, "grad_norm": 0.38563845241582717, "learning_rate": 3.0231968987661397e-06, "loss": 2.667, "step": 43488 }, { "epoch": 2.699670991371283, "grad_norm": 0.4049520114147867, "learning_rate": 3.0219602549321547e-06, "loss": 2.7041, "step": 43489 }, { "epoch": 2.699733068471041, "grad_norm": 0.43041285932591977, "learning_rate": 3.020723856194002e-06, "loss": 2.7137, "step": 43490 }, { "epoch": 2.699795145570799, "grad_norm": 0.3949101017103566, "learning_rate": 3.0194877025581326e-06, "loss": 2.712, "step": 43491 }, { "epoch": 2.699857222670557, "grad_norm": 0.40726492721641266, "learning_rate": 3.018251794031013e-06, "loss": 2.6299, "step": 43492 }, { "epoch": 2.6999192997703148, "grad_norm": 0.3876356179813777, "learning_rate": 3.017016130619077e-06, "loss": 2.7335, "step": 43493 }, { "epoch": 2.6999813768700727, "grad_norm": 0.3793061888756329, "learning_rate": 3.0157807123287694e-06, "loss": 2.6383, "step": 43494 }, { "epoch": 2.7000434539698306, "grad_norm": 0.3397163174117258, "learning_rate": 3.0145455391665354e-06, "loss": 2.573, "step": 43495 }, { "epoch": 2.7001055310695885, "grad_norm": 0.38402251309347696, "learning_rate": 3.013310611138831e-06, "loss": 2.6826, "step": 43496 }, { "epoch": 2.7001676081693464, "grad_norm": 0.40858173607984677, "learning_rate": 3.012075928252078e-06, "loss": 2.7138, "step": 43497 }, { "epoch": 2.7002296852691043, "grad_norm": 0.40317878364362475, "learning_rate": 3.0108414905127446e-06, "loss": 2.7034, "step": 43498 }, { "epoch": 2.7002917623688623, "grad_norm": 0.38140327546218983, "learning_rate": 3.0096072979272585e-06, "loss": 2.6483, "step": 43499 }, { "epoch": 2.70035383946862, "grad_norm": 0.38878152728647697, "learning_rate": 3.008373350502053e-06, "loss": 2.6831, "step": 43500 }, { "epoch": 2.700415916568378, "grad_norm": 0.3805183840419674, "learning_rate": 3.007139648243573e-06, "loss": 2.7279, "step": 43501 }, { "epoch": 2.7004779936681356, "grad_norm": 0.39119286269257203, "learning_rate": 3.005906191158259e-06, "loss": 2.6959, "step": 43502 }, { "epoch": 2.700540070767894, "grad_norm": 0.376843431551886, "learning_rate": 3.0046729792525375e-06, "loss": 2.676, "step": 43503 }, { "epoch": 2.7006021478676514, "grad_norm": 0.37113488621840535, "learning_rate": 3.0034400125328432e-06, "loss": 2.609, "step": 43504 }, { "epoch": 2.7006642249674098, "grad_norm": 0.37601066533774147, "learning_rate": 3.00220729100561e-06, "loss": 2.7375, "step": 43505 }, { "epoch": 2.7007263020671672, "grad_norm": 0.361424778563524, "learning_rate": 3.0009748146772766e-06, "loss": 2.6235, "step": 43506 }, { "epoch": 2.700788379166925, "grad_norm": 0.3506411878143495, "learning_rate": 2.9997425835542605e-06, "loss": 2.6606, "step": 43507 }, { "epoch": 2.700850456266683, "grad_norm": 0.3803606533826218, "learning_rate": 2.9985105976430005e-06, "loss": 2.6553, "step": 43508 }, { "epoch": 2.700912533366441, "grad_norm": 0.379648955472657, "learning_rate": 2.9972788569499145e-06, "loss": 2.6664, "step": 43509 }, { "epoch": 2.700974610466199, "grad_norm": 0.38676845346724265, "learning_rate": 2.9960473614814467e-06, "loss": 2.6651, "step": 43510 }, { "epoch": 2.701036687565957, "grad_norm": 0.34508457990485847, "learning_rate": 2.9948161112440033e-06, "loss": 2.7105, "step": 43511 }, { "epoch": 2.7010987646657147, "grad_norm": 0.3761374587559336, "learning_rate": 2.993585106244018e-06, "loss": 2.7187, "step": 43512 }, { "epoch": 2.7011608417654727, "grad_norm": 0.4128234967507642, "learning_rate": 2.9923543464879076e-06, "loss": 2.6714, "step": 43513 }, { "epoch": 2.7012229188652306, "grad_norm": 0.3928430337199138, "learning_rate": 2.9911238319820955e-06, "loss": 2.6601, "step": 43514 }, { "epoch": 2.7012849959649885, "grad_norm": 0.3859284156625923, "learning_rate": 2.9898935627330038e-06, "loss": 2.636, "step": 43515 }, { "epoch": 2.7013470730647464, "grad_norm": 0.3765704727911692, "learning_rate": 2.9886635387470495e-06, "loss": 2.7157, "step": 43516 }, { "epoch": 2.7014091501645043, "grad_norm": 0.39983385032810137, "learning_rate": 2.987433760030656e-06, "loss": 2.6622, "step": 43517 }, { "epoch": 2.7014712272642623, "grad_norm": 0.410104639348436, "learning_rate": 2.986204226590217e-06, "loss": 2.6833, "step": 43518 }, { "epoch": 2.70153330436402, "grad_norm": 0.39992740166003743, "learning_rate": 2.984974938432178e-06, "loss": 2.7302, "step": 43519 }, { "epoch": 2.701595381463778, "grad_norm": 0.36424594890512196, "learning_rate": 2.983745895562934e-06, "loss": 2.6438, "step": 43520 }, { "epoch": 2.701657458563536, "grad_norm": 0.37309139046025536, "learning_rate": 2.9825170979889015e-06, "loss": 2.7181, "step": 43521 }, { "epoch": 2.701719535663294, "grad_norm": 0.3896480550740213, "learning_rate": 2.9812885457164874e-06, "loss": 2.6933, "step": 43522 }, { "epoch": 2.701781612763052, "grad_norm": 0.36652556064953906, "learning_rate": 2.980060238752108e-06, "loss": 2.6648, "step": 43523 }, { "epoch": 2.7018436898628098, "grad_norm": 0.39999591631592923, "learning_rate": 2.97883217710217e-06, "loss": 2.7251, "step": 43524 }, { "epoch": 2.7019057669625672, "grad_norm": 0.37152630542253434, "learning_rate": 2.9776043607730787e-06, "loss": 2.6481, "step": 43525 }, { "epoch": 2.7019678440623256, "grad_norm": 0.3945345925671079, "learning_rate": 2.976376789771246e-06, "loss": 2.714, "step": 43526 }, { "epoch": 2.702029921162083, "grad_norm": 0.40177295528316687, "learning_rate": 2.975149464103061e-06, "loss": 2.7911, "step": 43527 }, { "epoch": 2.7020919982618414, "grad_norm": 0.37452480885135586, "learning_rate": 2.9739223837749407e-06, "loss": 2.6023, "step": 43528 }, { "epoch": 2.702154075361599, "grad_norm": 0.37392185116463705, "learning_rate": 2.972695548793275e-06, "loss": 2.6367, "step": 43529 }, { "epoch": 2.7022161524613573, "grad_norm": 0.38306974992754916, "learning_rate": 2.971468959164486e-06, "loss": 2.6994, "step": 43530 }, { "epoch": 2.7022782295611147, "grad_norm": 0.38084673738297853, "learning_rate": 2.9702426148949582e-06, "loss": 2.6061, "step": 43531 }, { "epoch": 2.7023403066608727, "grad_norm": 0.3797396510222727, "learning_rate": 2.96901651599108e-06, "loss": 2.6131, "step": 43532 }, { "epoch": 2.7024023837606306, "grad_norm": 0.3640629142136735, "learning_rate": 2.9677906624592744e-06, "loss": 2.7278, "step": 43533 }, { "epoch": 2.7024644608603885, "grad_norm": 0.37654866869838893, "learning_rate": 2.966565054305914e-06, "loss": 2.7653, "step": 43534 }, { "epoch": 2.7025265379601464, "grad_norm": 0.3919799574696091, "learning_rate": 2.965339691537411e-06, "loss": 2.6711, "step": 43535 }, { "epoch": 2.7025886150599043, "grad_norm": 0.376826190824417, "learning_rate": 2.964114574160132e-06, "loss": 2.6363, "step": 43536 }, { "epoch": 2.7026506921596622, "grad_norm": 0.37486093644705826, "learning_rate": 2.962889702180499e-06, "loss": 2.606, "step": 43537 }, { "epoch": 2.70271276925942, "grad_norm": 0.38140220105476763, "learning_rate": 2.9616650756048857e-06, "loss": 2.6323, "step": 43538 }, { "epoch": 2.702774846359178, "grad_norm": 0.38689260716961, "learning_rate": 2.960440694439687e-06, "loss": 2.6997, "step": 43539 }, { "epoch": 2.702836923458936, "grad_norm": 0.3825293875917643, "learning_rate": 2.9592165586912912e-06, "loss": 2.6489, "step": 43540 }, { "epoch": 2.702899000558694, "grad_norm": 0.37154114507037, "learning_rate": 2.957992668366072e-06, "loss": 2.6162, "step": 43541 }, { "epoch": 2.702961077658452, "grad_norm": 0.3844709671987223, "learning_rate": 2.956769023470429e-06, "loss": 2.7325, "step": 43542 }, { "epoch": 2.7030231547582098, "grad_norm": 0.36704825743939856, "learning_rate": 2.955545624010747e-06, "loss": 2.7349, "step": 43543 }, { "epoch": 2.7030852318579677, "grad_norm": 0.40972898117604073, "learning_rate": 2.9543224699934035e-06, "loss": 2.6603, "step": 43544 }, { "epoch": 2.7031473089577256, "grad_norm": 0.36959298487163394, "learning_rate": 2.9530995614247715e-06, "loss": 2.662, "step": 43545 }, { "epoch": 2.7032093860574835, "grad_norm": 0.4167447036939809, "learning_rate": 2.9518768983112455e-06, "loss": 2.6341, "step": 43546 }, { "epoch": 2.7032714631572414, "grad_norm": 0.36821582541607006, "learning_rate": 2.9506544806591983e-06, "loss": 2.6261, "step": 43547 }, { "epoch": 2.7033335402569993, "grad_norm": 0.37000878993753983, "learning_rate": 2.9494323084750144e-06, "loss": 2.4799, "step": 43548 }, { "epoch": 2.7033956173567573, "grad_norm": 0.41280967185865897, "learning_rate": 2.9482103817650595e-06, "loss": 2.6973, "step": 43549 }, { "epoch": 2.7034576944565147, "grad_norm": 0.37808612670956765, "learning_rate": 2.9469887005357077e-06, "loss": 2.6207, "step": 43550 }, { "epoch": 2.703519771556273, "grad_norm": 0.3902597436305576, "learning_rate": 2.945767264793342e-06, "loss": 2.7216, "step": 43551 }, { "epoch": 2.7035818486560306, "grad_norm": 0.36504159011088194, "learning_rate": 2.9445460745443355e-06, "loss": 2.6538, "step": 43552 }, { "epoch": 2.703643925755789, "grad_norm": 0.3757783838111239, "learning_rate": 2.9433251297950546e-06, "loss": 2.7685, "step": 43553 }, { "epoch": 2.7037060028555464, "grad_norm": 0.4107295797592486, "learning_rate": 2.942104430551862e-06, "loss": 2.7049, "step": 43554 }, { "epoch": 2.7037680799553043, "grad_norm": 0.3853027738223807, "learning_rate": 2.9408839768211404e-06, "loss": 2.6825, "step": 43555 }, { "epoch": 2.7038301570550622, "grad_norm": 0.38372466345408257, "learning_rate": 2.939663768609252e-06, "loss": 2.6898, "step": 43556 }, { "epoch": 2.70389223415482, "grad_norm": 0.36241113379140716, "learning_rate": 2.938443805922564e-06, "loss": 2.5846, "step": 43557 }, { "epoch": 2.703954311254578, "grad_norm": 0.390136604165835, "learning_rate": 2.937224088767432e-06, "loss": 2.605, "step": 43558 }, { "epoch": 2.704016388354336, "grad_norm": 0.3992980378129717, "learning_rate": 2.936004617150234e-06, "loss": 2.7508, "step": 43559 }, { "epoch": 2.704078465454094, "grad_norm": 0.3734402405067665, "learning_rate": 2.9347853910773217e-06, "loss": 2.6294, "step": 43560 }, { "epoch": 2.704140542553852, "grad_norm": 0.37681236999276635, "learning_rate": 2.9335664105550608e-06, "loss": 2.7444, "step": 43561 }, { "epoch": 2.7042026196536098, "grad_norm": 0.3682948722472553, "learning_rate": 2.9323476755898027e-06, "loss": 2.6745, "step": 43562 }, { "epoch": 2.7042646967533677, "grad_norm": 0.38224536379740115, "learning_rate": 2.9311291861879197e-06, "loss": 2.6524, "step": 43563 }, { "epoch": 2.7043267738531256, "grad_norm": 0.37459940176089657, "learning_rate": 2.929910942355757e-06, "loss": 2.6359, "step": 43564 }, { "epoch": 2.7043888509528835, "grad_norm": 0.4067785053123945, "learning_rate": 2.928692944099681e-06, "loss": 2.6874, "step": 43565 }, { "epoch": 2.7044509280526414, "grad_norm": 0.3745523403374179, "learning_rate": 2.9274751914260434e-06, "loss": 2.6596, "step": 43566 }, { "epoch": 2.7045130051523993, "grad_norm": 0.3824720620567426, "learning_rate": 2.9262576843411883e-06, "loss": 2.6446, "step": 43567 }, { "epoch": 2.7045750822521573, "grad_norm": 0.35531447280027445, "learning_rate": 2.9250404228514717e-06, "loss": 2.6819, "step": 43568 }, { "epoch": 2.704637159351915, "grad_norm": 0.34610322116955416, "learning_rate": 2.9238234069632555e-06, "loss": 2.6027, "step": 43569 }, { "epoch": 2.704699236451673, "grad_norm": 0.3866893294731825, "learning_rate": 2.9226066366828785e-06, "loss": 2.7257, "step": 43570 }, { "epoch": 2.704761313551431, "grad_norm": 0.3769125249473324, "learning_rate": 2.921390112016692e-06, "loss": 2.7556, "step": 43571 }, { "epoch": 2.704823390651189, "grad_norm": 0.40320781404207473, "learning_rate": 2.9201738329710347e-06, "loss": 2.6595, "step": 43572 }, { "epoch": 2.7048854677509464, "grad_norm": 0.36745313410453395, "learning_rate": 2.9189577995522687e-06, "loss": 2.6876, "step": 43573 }, { "epoch": 2.7049475448507048, "grad_norm": 0.3995385384072, "learning_rate": 2.9177420117667275e-06, "loss": 2.6078, "step": 43574 }, { "epoch": 2.7050096219504622, "grad_norm": 0.40028506492448496, "learning_rate": 2.9165264696207563e-06, "loss": 2.6248, "step": 43575 }, { "epoch": 2.7050716990502206, "grad_norm": 0.41111409196796805, "learning_rate": 2.9153111731206938e-06, "loss": 2.6001, "step": 43576 }, { "epoch": 2.705133776149978, "grad_norm": 0.385844851967979, "learning_rate": 2.91409612227288e-06, "loss": 2.6277, "step": 43577 }, { "epoch": 2.7051958532497364, "grad_norm": 0.3810788036010211, "learning_rate": 2.91288131708366e-06, "loss": 2.6808, "step": 43578 }, { "epoch": 2.705257930349494, "grad_norm": 0.4058378953729523, "learning_rate": 2.9116667575593725e-06, "loss": 2.7084, "step": 43579 }, { "epoch": 2.705320007449252, "grad_norm": 0.3944991259196549, "learning_rate": 2.9104524437063517e-06, "loss": 2.6606, "step": 43580 }, { "epoch": 2.7053820845490097, "grad_norm": 0.38999021450628707, "learning_rate": 2.9092383755309207e-06, "loss": 2.6902, "step": 43581 }, { "epoch": 2.7054441616487677, "grad_norm": 0.3888297598974782, "learning_rate": 2.9080245530394344e-06, "loss": 2.6709, "step": 43582 }, { "epoch": 2.7055062387485256, "grad_norm": 0.3968378380024849, "learning_rate": 2.906810976238217e-06, "loss": 2.7475, "step": 43583 }, { "epoch": 2.7055683158482835, "grad_norm": 0.3828860815526257, "learning_rate": 2.9055976451335953e-06, "loss": 2.6059, "step": 43584 }, { "epoch": 2.7056303929480414, "grad_norm": 0.40675062730971756, "learning_rate": 2.9043845597318984e-06, "loss": 2.7837, "step": 43585 }, { "epoch": 2.7056924700477993, "grad_norm": 0.3584834109821567, "learning_rate": 2.9031717200394657e-06, "loss": 2.7368, "step": 43586 }, { "epoch": 2.7057545471475573, "grad_norm": 0.3745577111284178, "learning_rate": 2.90195912606262e-06, "loss": 2.6002, "step": 43587 }, { "epoch": 2.705816624247315, "grad_norm": 0.3864454980177103, "learning_rate": 2.9007467778076893e-06, "loss": 2.7997, "step": 43588 }, { "epoch": 2.705878701347073, "grad_norm": 0.3877359364170884, "learning_rate": 2.899534675280996e-06, "loss": 2.6704, "step": 43589 }, { "epoch": 2.705940778446831, "grad_norm": 0.375513202450378, "learning_rate": 2.8983228184888577e-06, "loss": 2.7101, "step": 43590 }, { "epoch": 2.706002855546589, "grad_norm": 0.3979461857030696, "learning_rate": 2.8971112074376084e-06, "loss": 2.742, "step": 43591 }, { "epoch": 2.706064932646347, "grad_norm": 0.36650594400214964, "learning_rate": 2.895899842133565e-06, "loss": 2.6951, "step": 43592 }, { "epoch": 2.7061270097461048, "grad_norm": 0.37269543775606945, "learning_rate": 2.8946887225830445e-06, "loss": 2.7018, "step": 43593 }, { "epoch": 2.7061890868458627, "grad_norm": 0.38858130851925426, "learning_rate": 2.893477848792364e-06, "loss": 2.7125, "step": 43594 }, { "epoch": 2.7062511639456206, "grad_norm": 0.34937911735797245, "learning_rate": 2.892267220767847e-06, "loss": 2.6369, "step": 43595 }, { "epoch": 2.7063132410453785, "grad_norm": 0.37756529202350064, "learning_rate": 2.8910568385158097e-06, "loss": 2.7449, "step": 43596 }, { "epoch": 2.7063753181451364, "grad_norm": 0.39568498198931645, "learning_rate": 2.8898467020425702e-06, "loss": 2.652, "step": 43597 }, { "epoch": 2.706437395244894, "grad_norm": 0.3550734575256501, "learning_rate": 2.8886368113544337e-06, "loss": 2.634, "step": 43598 }, { "epoch": 2.7064994723446523, "grad_norm": 0.3826116696826879, "learning_rate": 2.887427166457718e-06, "loss": 2.6989, "step": 43599 }, { "epoch": 2.7065615494444097, "grad_norm": 0.3814474602876621, "learning_rate": 2.8862177673587233e-06, "loss": 2.7135, "step": 43600 }, { "epoch": 2.706623626544168, "grad_norm": 0.37650909361679613, "learning_rate": 2.8850086140637723e-06, "loss": 2.7188, "step": 43601 }, { "epoch": 2.7066857036439256, "grad_norm": 0.3954289864060298, "learning_rate": 2.8837997065791713e-06, "loss": 2.5877, "step": 43602 }, { "epoch": 2.7067477807436835, "grad_norm": 0.39163786227585207, "learning_rate": 2.882591044911226e-06, "loss": 2.7185, "step": 43603 }, { "epoch": 2.7068098578434414, "grad_norm": 0.3807407683532636, "learning_rate": 2.8813826290662317e-06, "loss": 2.6974, "step": 43604 }, { "epoch": 2.7068719349431993, "grad_norm": 0.3786550259250426, "learning_rate": 2.880174459050511e-06, "loss": 2.6548, "step": 43605 }, { "epoch": 2.7069340120429572, "grad_norm": 0.3818585518386718, "learning_rate": 2.8789665348703588e-06, "loss": 2.6383, "step": 43606 }, { "epoch": 2.706996089142715, "grad_norm": 0.38508660342540896, "learning_rate": 2.877758856532081e-06, "loss": 2.6678, "step": 43607 }, { "epoch": 2.707058166242473, "grad_norm": 0.37616025722543545, "learning_rate": 2.876551424041962e-06, "loss": 2.662, "step": 43608 }, { "epoch": 2.707120243342231, "grad_norm": 0.3549819379563604, "learning_rate": 2.8753442374063243e-06, "loss": 2.688, "step": 43609 }, { "epoch": 2.707182320441989, "grad_norm": 0.37737614115504264, "learning_rate": 2.8741372966314515e-06, "loss": 2.6406, "step": 43610 }, { "epoch": 2.707244397541747, "grad_norm": 0.40096723394231437, "learning_rate": 2.8729306017236503e-06, "loss": 2.6329, "step": 43611 }, { "epoch": 2.7073064746415048, "grad_norm": 0.358471688832589, "learning_rate": 2.871724152689209e-06, "loss": 2.7178, "step": 43612 }, { "epoch": 2.7073685517412627, "grad_norm": 0.3844701339100213, "learning_rate": 2.8705179495344127e-06, "loss": 2.719, "step": 43613 }, { "epoch": 2.7074306288410206, "grad_norm": 0.3553060222025175, "learning_rate": 2.8693119922655775e-06, "loss": 2.6218, "step": 43614 }, { "epoch": 2.7074927059407785, "grad_norm": 0.37864235408956276, "learning_rate": 2.8681062808889768e-06, "loss": 2.6467, "step": 43615 }, { "epoch": 2.7075547830405364, "grad_norm": 0.3614622331165202, "learning_rate": 2.866900815410911e-06, "loss": 2.6267, "step": 43616 }, { "epoch": 2.7076168601402943, "grad_norm": 0.35419527340341067, "learning_rate": 2.8656955958376586e-06, "loss": 2.6152, "step": 43617 }, { "epoch": 2.7076789372400523, "grad_norm": 0.3511737581569602, "learning_rate": 2.864490622175525e-06, "loss": 2.693, "step": 43618 }, { "epoch": 2.70774101433981, "grad_norm": 0.3732473024729442, "learning_rate": 2.863285894430784e-06, "loss": 2.6798, "step": 43619 }, { "epoch": 2.707803091439568, "grad_norm": 0.3483947240120669, "learning_rate": 2.8620814126097185e-06, "loss": 2.7121, "step": 43620 }, { "epoch": 2.7078651685393256, "grad_norm": 0.3836991881923424, "learning_rate": 2.8608771767186183e-06, "loss": 2.6196, "step": 43621 }, { "epoch": 2.707927245639084, "grad_norm": 0.35920844157437476, "learning_rate": 2.8596731867637673e-06, "loss": 2.655, "step": 43622 }, { "epoch": 2.7079893227388414, "grad_norm": 0.38449811591938077, "learning_rate": 2.8584694427514437e-06, "loss": 2.772, "step": 43623 }, { "epoch": 2.7080513998385998, "grad_norm": 0.39361270827081, "learning_rate": 2.8572659446879312e-06, "loss": 2.7149, "step": 43624 }, { "epoch": 2.7081134769383572, "grad_norm": 0.3653605646015275, "learning_rate": 2.856062692579509e-06, "loss": 2.6057, "step": 43625 }, { "epoch": 2.7081755540381156, "grad_norm": 0.3839323441266207, "learning_rate": 2.854859686432443e-06, "loss": 2.7005, "step": 43626 }, { "epoch": 2.708237631137873, "grad_norm": 0.391280825348194, "learning_rate": 2.8536569262530233e-06, "loss": 2.6773, "step": 43627 }, { "epoch": 2.708299708237631, "grad_norm": 0.3545959993215275, "learning_rate": 2.8524544120475173e-06, "loss": 2.7367, "step": 43628 }, { "epoch": 2.708361785337389, "grad_norm": 0.3654883006894016, "learning_rate": 2.8512521438222084e-06, "loss": 2.6363, "step": 43629 }, { "epoch": 2.708423862437147, "grad_norm": 0.37096236061273274, "learning_rate": 2.850050121583364e-06, "loss": 2.6904, "step": 43630 }, { "epoch": 2.7084859395369048, "grad_norm": 0.3384585046105987, "learning_rate": 2.8488483453372407e-06, "loss": 2.6708, "step": 43631 }, { "epoch": 2.7085480166366627, "grad_norm": 0.3399388191280311, "learning_rate": 2.847646815090138e-06, "loss": 2.6388, "step": 43632 }, { "epoch": 2.7086100937364206, "grad_norm": 0.3626997405620517, "learning_rate": 2.846445530848302e-06, "loss": 2.6359, "step": 43633 }, { "epoch": 2.7086721708361785, "grad_norm": 0.37803175383360244, "learning_rate": 2.84524449261801e-06, "loss": 2.5836, "step": 43634 }, { "epoch": 2.7087342479359364, "grad_norm": 0.37091997997041376, "learning_rate": 2.844043700405513e-06, "loss": 2.5957, "step": 43635 }, { "epoch": 2.7087963250356943, "grad_norm": 0.35479271710311805, "learning_rate": 2.8428431542171006e-06, "loss": 2.6753, "step": 43636 }, { "epoch": 2.7088584021354523, "grad_norm": 0.4025765643542062, "learning_rate": 2.8416428540590177e-06, "loss": 2.5797, "step": 43637 }, { "epoch": 2.70892047923521, "grad_norm": 0.39546079002879075, "learning_rate": 2.840442799937537e-06, "loss": 2.5989, "step": 43638 }, { "epoch": 2.708982556334968, "grad_norm": 0.3718822534639138, "learning_rate": 2.8392429918589148e-06, "loss": 2.6829, "step": 43639 }, { "epoch": 2.709044633434726, "grad_norm": 0.3573675687630393, "learning_rate": 2.838043429829401e-06, "loss": 2.6014, "step": 43640 }, { "epoch": 2.709106710534484, "grad_norm": 0.38521712094907345, "learning_rate": 2.8368441138552694e-06, "loss": 2.6903, "step": 43641 }, { "epoch": 2.709168787634242, "grad_norm": 0.34358558190573985, "learning_rate": 2.8356450439427752e-06, "loss": 2.6863, "step": 43642 }, { "epoch": 2.7092308647339998, "grad_norm": 0.35995873436587345, "learning_rate": 2.8344462200981693e-06, "loss": 2.6364, "step": 43643 }, { "epoch": 2.7092929418337577, "grad_norm": 0.3924390442099872, "learning_rate": 2.8332476423276968e-06, "loss": 2.6462, "step": 43644 }, { "epoch": 2.7093550189335156, "grad_norm": 0.37673657378692266, "learning_rate": 2.83204931063763e-06, "loss": 2.6709, "step": 43645 }, { "epoch": 2.709417096033273, "grad_norm": 0.3938855467715665, "learning_rate": 2.8308512250342145e-06, "loss": 2.6073, "step": 43646 }, { "epoch": 2.7094791731330314, "grad_norm": 0.3841410956419629, "learning_rate": 2.829653385523695e-06, "loss": 2.7362, "step": 43647 }, { "epoch": 2.709541250232789, "grad_norm": 0.3822189907284988, "learning_rate": 2.828455792112328e-06, "loss": 2.647, "step": 43648 }, { "epoch": 2.7096033273325473, "grad_norm": 0.387392574463106, "learning_rate": 2.827258444806352e-06, "loss": 2.6587, "step": 43649 }, { "epoch": 2.7096654044323047, "grad_norm": 0.3881697358350894, "learning_rate": 2.826061343612024e-06, "loss": 2.5961, "step": 43650 }, { "epoch": 2.7097274815320627, "grad_norm": 0.3806384888812933, "learning_rate": 2.824864488535589e-06, "loss": 2.7008, "step": 43651 }, { "epoch": 2.7097895586318206, "grad_norm": 0.4107102784855752, "learning_rate": 2.823667879583286e-06, "loss": 2.6513, "step": 43652 }, { "epoch": 2.7098516357315785, "grad_norm": 0.36852204502215685, "learning_rate": 2.822471516761355e-06, "loss": 2.6271, "step": 43653 }, { "epoch": 2.7099137128313364, "grad_norm": 0.35538406201648853, "learning_rate": 2.821275400076051e-06, "loss": 2.6289, "step": 43654 }, { "epoch": 2.7099757899310943, "grad_norm": 0.3770629248532115, "learning_rate": 2.8200795295336037e-06, "loss": 2.6583, "step": 43655 }, { "epoch": 2.7100378670308523, "grad_norm": 0.37719110280159623, "learning_rate": 2.818883905140257e-06, "loss": 2.6616, "step": 43656 }, { "epoch": 2.71009994413061, "grad_norm": 0.38091739310445194, "learning_rate": 2.8176885269022347e-06, "loss": 2.7077, "step": 43657 }, { "epoch": 2.710162021230368, "grad_norm": 0.3967475581367407, "learning_rate": 2.816493394825798e-06, "loss": 2.7003, "step": 43658 }, { "epoch": 2.710224098330126, "grad_norm": 0.37542513749236384, "learning_rate": 2.8152985089171635e-06, "loss": 2.6092, "step": 43659 }, { "epoch": 2.710286175429884, "grad_norm": 0.39218991571419504, "learning_rate": 2.8141038691825714e-06, "loss": 2.678, "step": 43660 }, { "epoch": 2.710348252529642, "grad_norm": 0.3780334093754806, "learning_rate": 2.8129094756282557e-06, "loss": 2.6811, "step": 43661 }, { "epoch": 2.7104103296293998, "grad_norm": 0.39046335690587475, "learning_rate": 2.81171532826045e-06, "loss": 2.7904, "step": 43662 }, { "epoch": 2.7104724067291577, "grad_norm": 0.36105144320570515, "learning_rate": 2.810521427085372e-06, "loss": 2.5891, "step": 43663 }, { "epoch": 2.7105344838289156, "grad_norm": 0.3995703692151353, "learning_rate": 2.809327772109266e-06, "loss": 2.621, "step": 43664 }, { "epoch": 2.7105965609286735, "grad_norm": 0.39940038904788117, "learning_rate": 2.8081343633383495e-06, "loss": 2.6147, "step": 43665 }, { "epoch": 2.7106586380284314, "grad_norm": 0.3707092501714127, "learning_rate": 2.8069412007788567e-06, "loss": 2.5851, "step": 43666 }, { "epoch": 2.7107207151281894, "grad_norm": 0.3696942693650024, "learning_rate": 2.8057482844369986e-06, "loss": 2.7052, "step": 43667 }, { "epoch": 2.7107827922279473, "grad_norm": 0.3718014444005077, "learning_rate": 2.804555614319021e-06, "loss": 2.6482, "step": 43668 }, { "epoch": 2.7108448693277047, "grad_norm": 0.3891642724636745, "learning_rate": 2.80336319043113e-06, "loss": 2.7549, "step": 43669 }, { "epoch": 2.710906946427463, "grad_norm": 0.3526699848061017, "learning_rate": 2.8021710127795476e-06, "loss": 2.6584, "step": 43670 }, { "epoch": 2.7109690235272206, "grad_norm": 0.38075863410617955, "learning_rate": 2.800979081370497e-06, "loss": 2.5842, "step": 43671 }, { "epoch": 2.711031100626979, "grad_norm": 0.40296197818244606, "learning_rate": 2.799787396210196e-06, "loss": 2.7535, "step": 43672 }, { "epoch": 2.7110931777267364, "grad_norm": 0.3844115084856049, "learning_rate": 2.798595957304867e-06, "loss": 2.7031, "step": 43673 }, { "epoch": 2.7111552548264948, "grad_norm": 0.3680213024087107, "learning_rate": 2.7974047646607217e-06, "loss": 2.6894, "step": 43674 }, { "epoch": 2.7112173319262522, "grad_norm": 0.35229287044485663, "learning_rate": 2.7962138182839716e-06, "loss": 2.6652, "step": 43675 }, { "epoch": 2.71127940902601, "grad_norm": 0.3880949099021391, "learning_rate": 2.7950231181808285e-06, "loss": 2.6741, "step": 43676 }, { "epoch": 2.711341486125768, "grad_norm": 0.37450561120575027, "learning_rate": 2.7938326643575154e-06, "loss": 2.6971, "step": 43677 }, { "epoch": 2.711403563225526, "grad_norm": 0.3827776990369677, "learning_rate": 2.792642456820238e-06, "loss": 2.6756, "step": 43678 }, { "epoch": 2.711465640325284, "grad_norm": 0.35444741715125005, "learning_rate": 2.791452495575203e-06, "loss": 2.6745, "step": 43679 }, { "epoch": 2.711527717425042, "grad_norm": 0.38925535408924233, "learning_rate": 2.790262780628611e-06, "loss": 2.6299, "step": 43680 }, { "epoch": 2.7115897945247998, "grad_norm": 0.41159928197302253, "learning_rate": 2.7890733119866896e-06, "loss": 2.6824, "step": 43681 }, { "epoch": 2.7116518716245577, "grad_norm": 0.393624671033405, "learning_rate": 2.7878840896556346e-06, "loss": 2.5558, "step": 43682 }, { "epoch": 2.7117139487243156, "grad_norm": 0.39567846690224706, "learning_rate": 2.7866951136416464e-06, "loss": 2.7135, "step": 43683 }, { "epoch": 2.7117760258240735, "grad_norm": 0.38527341921358704, "learning_rate": 2.785506383950931e-06, "loss": 2.6347, "step": 43684 }, { "epoch": 2.7118381029238314, "grad_norm": 0.37045707294880553, "learning_rate": 2.784317900589689e-06, "loss": 2.6677, "step": 43685 }, { "epoch": 2.7119001800235893, "grad_norm": 0.3655045881635692, "learning_rate": 2.7831296635641267e-06, "loss": 2.6435, "step": 43686 }, { "epoch": 2.7119622571233473, "grad_norm": 0.36729370772598297, "learning_rate": 2.7819416728804335e-06, "loss": 2.6309, "step": 43687 }, { "epoch": 2.712024334223105, "grad_norm": 0.3730065888226877, "learning_rate": 2.7807539285448216e-06, "loss": 2.7628, "step": 43688 }, { "epoch": 2.712086411322863, "grad_norm": 0.38678226527339465, "learning_rate": 2.7795664305634684e-06, "loss": 2.6319, "step": 43689 }, { "epoch": 2.712148488422621, "grad_norm": 0.3998418815840446, "learning_rate": 2.7783791789425863e-06, "loss": 2.6361, "step": 43690 }, { "epoch": 2.712210565522379, "grad_norm": 0.4079235159649944, "learning_rate": 2.7771921736883643e-06, "loss": 2.7085, "step": 43691 }, { "epoch": 2.712272642622137, "grad_norm": 0.36787510467927037, "learning_rate": 2.7760054148069926e-06, "loss": 2.6552, "step": 43692 }, { "epoch": 2.7123347197218948, "grad_norm": 0.41063228412014363, "learning_rate": 2.77481890230466e-06, "loss": 2.6026, "step": 43693 }, { "epoch": 2.7123967968216522, "grad_norm": 0.3851566361121666, "learning_rate": 2.773632636187562e-06, "loss": 2.6595, "step": 43694 }, { "epoch": 2.7124588739214106, "grad_norm": 0.38042450555550034, "learning_rate": 2.7724466164618935e-06, "loss": 2.7017, "step": 43695 }, { "epoch": 2.712520951021168, "grad_norm": 0.37913254734555046, "learning_rate": 2.7712608431338384e-06, "loss": 2.6521, "step": 43696 }, { "epoch": 2.7125830281209264, "grad_norm": 0.35362621284781187, "learning_rate": 2.7700753162095748e-06, "loss": 2.6007, "step": 43697 }, { "epoch": 2.712645105220684, "grad_norm": 0.3566833955118536, "learning_rate": 2.7688900356952987e-06, "loss": 2.6797, "step": 43698 }, { "epoch": 2.712707182320442, "grad_norm": 0.38432953418321825, "learning_rate": 2.7677050015971815e-06, "loss": 2.722, "step": 43699 }, { "epoch": 2.7127692594201998, "grad_norm": 0.3647139406139621, "learning_rate": 2.76652021392142e-06, "loss": 2.6604, "step": 43700 }, { "epoch": 2.7128313365199577, "grad_norm": 0.3830864179329491, "learning_rate": 2.765335672674185e-06, "loss": 2.7015, "step": 43701 }, { "epoch": 2.7128934136197156, "grad_norm": 0.3830573797370777, "learning_rate": 2.764151377861668e-06, "loss": 2.6168, "step": 43702 }, { "epoch": 2.7129554907194735, "grad_norm": 0.3503028449226144, "learning_rate": 2.7629673294900293e-06, "loss": 2.745, "step": 43703 }, { "epoch": 2.7130175678192314, "grad_norm": 0.38065099639937544, "learning_rate": 2.761783527565459e-06, "loss": 2.6676, "step": 43704 }, { "epoch": 2.7130796449189893, "grad_norm": 0.3581365101147953, "learning_rate": 2.7605999720941356e-06, "loss": 2.5884, "step": 43705 }, { "epoch": 2.7131417220187473, "grad_norm": 0.37795742435841556, "learning_rate": 2.7594166630822314e-06, "loss": 2.6492, "step": 43706 }, { "epoch": 2.713203799118505, "grad_norm": 0.376305374430308, "learning_rate": 2.7582336005359144e-06, "loss": 2.7066, "step": 43707 }, { "epoch": 2.713265876218263, "grad_norm": 0.37304431854490533, "learning_rate": 2.7570507844613623e-06, "loss": 2.638, "step": 43708 }, { "epoch": 2.713327953318021, "grad_norm": 0.35315179807541963, "learning_rate": 2.7558682148647485e-06, "loss": 2.6902, "step": 43709 }, { "epoch": 2.713390030417779, "grad_norm": 0.36390099934008047, "learning_rate": 2.7546858917522345e-06, "loss": 2.7031, "step": 43710 }, { "epoch": 2.713452107517537, "grad_norm": 0.3784573442215961, "learning_rate": 2.7535038151299988e-06, "loss": 2.6562, "step": 43711 }, { "epoch": 2.7135141846172948, "grad_norm": 0.3463213183187838, "learning_rate": 2.7523219850041915e-06, "loss": 2.6294, "step": 43712 }, { "epoch": 2.7135762617170527, "grad_norm": 0.3803185756013116, "learning_rate": 2.7511404013810028e-06, "loss": 2.749, "step": 43713 }, { "epoch": 2.7136383388168106, "grad_norm": 0.367521904792101, "learning_rate": 2.7499590642665774e-06, "loss": 2.659, "step": 43714 }, { "epoch": 2.7137004159165685, "grad_norm": 0.36907112637999767, "learning_rate": 2.748777973667094e-06, "loss": 2.7548, "step": 43715 }, { "epoch": 2.7137624930163264, "grad_norm": 0.384104708097425, "learning_rate": 2.7475971295886915e-06, "loss": 2.66, "step": 43716 }, { "epoch": 2.713824570116084, "grad_norm": 0.3969015172675156, "learning_rate": 2.7464165320375603e-06, "loss": 2.6897, "step": 43717 }, { "epoch": 2.7138866472158423, "grad_norm": 0.3592413156066313, "learning_rate": 2.745236181019839e-06, "loss": 2.6569, "step": 43718 }, { "epoch": 2.7139487243155997, "grad_norm": 0.3730545842038219, "learning_rate": 2.744056076541696e-06, "loss": 2.7477, "step": 43719 }, { "epoch": 2.714010801415358, "grad_norm": 0.38922333055003794, "learning_rate": 2.7428762186092805e-06, "loss": 2.737, "step": 43720 }, { "epoch": 2.7140728785151156, "grad_norm": 0.3808621528719809, "learning_rate": 2.7416966072287497e-06, "loss": 2.6632, "step": 43721 }, { "epoch": 2.714134955614874, "grad_norm": 0.37122864530136296, "learning_rate": 2.7405172424062653e-06, "loss": 2.7571, "step": 43722 }, { "epoch": 2.7141970327146314, "grad_norm": 0.3949652258438617, "learning_rate": 2.739338124147972e-06, "loss": 2.68, "step": 43723 }, { "epoch": 2.7142591098143893, "grad_norm": 0.3818115655872315, "learning_rate": 2.7381592524600264e-06, "loss": 2.7236, "step": 43724 }, { "epoch": 2.7143211869141473, "grad_norm": 0.37892492975812064, "learning_rate": 2.736980627348573e-06, "loss": 2.6566, "step": 43725 }, { "epoch": 2.714383264013905, "grad_norm": 0.38193073705679254, "learning_rate": 2.7358022488197634e-06, "loss": 2.6149, "step": 43726 }, { "epoch": 2.714445341113663, "grad_norm": 0.4107383230828412, "learning_rate": 2.7346241168797524e-06, "loss": 2.6769, "step": 43727 }, { "epoch": 2.714507418213421, "grad_norm": 0.3853223229832235, "learning_rate": 2.733446231534681e-06, "loss": 2.6488, "step": 43728 }, { "epoch": 2.714569495313179, "grad_norm": 0.37341129019468505, "learning_rate": 2.732268592790699e-06, "loss": 2.6224, "step": 43729 }, { "epoch": 2.714631572412937, "grad_norm": 0.3827487736812508, "learning_rate": 2.7310912006539347e-06, "loss": 2.6252, "step": 43730 }, { "epoch": 2.7146936495126948, "grad_norm": 0.36212739951036965, "learning_rate": 2.729914055130556e-06, "loss": 2.6691, "step": 43731 }, { "epoch": 2.7147557266124527, "grad_norm": 0.40932429268974546, "learning_rate": 2.728737156226685e-06, "loss": 2.6874, "step": 43732 }, { "epoch": 2.7148178037122106, "grad_norm": 0.38866591329171685, "learning_rate": 2.7275605039484676e-06, "loss": 2.7256, "step": 43733 }, { "epoch": 2.7148798808119685, "grad_norm": 0.35594867963373195, "learning_rate": 2.7263840983020485e-06, "loss": 2.5627, "step": 43734 }, { "epoch": 2.7149419579117264, "grad_norm": 0.393838822473594, "learning_rate": 2.7252079392935446e-06, "loss": 2.7717, "step": 43735 }, { "epoch": 2.7150040350114844, "grad_norm": 0.3812969921418113, "learning_rate": 2.7240320269291185e-06, "loss": 2.6499, "step": 43736 }, { "epoch": 2.7150661121112423, "grad_norm": 0.355949600623585, "learning_rate": 2.7228563612148984e-06, "loss": 2.7063, "step": 43737 }, { "epoch": 2.715128189211, "grad_norm": 0.3826153500538627, "learning_rate": 2.7216809421570067e-06, "loss": 2.6602, "step": 43738 }, { "epoch": 2.715190266310758, "grad_norm": 0.37975538612348203, "learning_rate": 2.7205057697615776e-06, "loss": 2.6733, "step": 43739 }, { "epoch": 2.715252343410516, "grad_norm": 0.36485391067635015, "learning_rate": 2.7193308440347563e-06, "loss": 2.6628, "step": 43740 }, { "epoch": 2.715314420510274, "grad_norm": 0.3683804349249185, "learning_rate": 2.71815616498266e-06, "loss": 2.7193, "step": 43741 }, { "epoch": 2.7153764976100314, "grad_norm": 0.3792858778265846, "learning_rate": 2.7169817326114233e-06, "loss": 2.6727, "step": 43742 }, { "epoch": 2.7154385747097898, "grad_norm": 0.3716039864148167, "learning_rate": 2.715807546927168e-06, "loss": 2.604, "step": 43743 }, { "epoch": 2.7155006518095473, "grad_norm": 0.3717129009439266, "learning_rate": 2.7146336079360234e-06, "loss": 2.6883, "step": 43744 }, { "epoch": 2.7155627289093056, "grad_norm": 0.3836711867537372, "learning_rate": 2.713459915644123e-06, "loss": 2.5826, "step": 43745 }, { "epoch": 2.715624806009063, "grad_norm": 0.372081165706106, "learning_rate": 2.7122864700575734e-06, "loss": 2.7039, "step": 43746 }, { "epoch": 2.715686883108821, "grad_norm": 0.3564282974989652, "learning_rate": 2.7111132711825082e-06, "loss": 2.6283, "step": 43747 }, { "epoch": 2.715748960208579, "grad_norm": 0.38564341765431576, "learning_rate": 2.7099403190250395e-06, "loss": 2.7158, "step": 43748 }, { "epoch": 2.715811037308337, "grad_norm": 0.3623879020936325, "learning_rate": 2.7087676135912954e-06, "loss": 2.6983, "step": 43749 }, { "epoch": 2.7158731144080948, "grad_norm": 0.37400630583011524, "learning_rate": 2.7075951548873933e-06, "loss": 2.6615, "step": 43750 }, { "epoch": 2.7159351915078527, "grad_norm": 0.36440155629376325, "learning_rate": 2.7064229429194454e-06, "loss": 2.7062, "step": 43751 }, { "epoch": 2.7159972686076106, "grad_norm": 0.37121144486281943, "learning_rate": 2.705250977693563e-06, "loss": 2.7348, "step": 43752 }, { "epoch": 2.7160593457073685, "grad_norm": 0.3808750247185364, "learning_rate": 2.7040792592158747e-06, "loss": 2.7125, "step": 43753 }, { "epoch": 2.7161214228071264, "grad_norm": 0.3605729536913975, "learning_rate": 2.7029077874924925e-06, "loss": 2.7542, "step": 43754 }, { "epoch": 2.7161834999068843, "grad_norm": 0.374173843779402, "learning_rate": 2.7017365625295165e-06, "loss": 2.593, "step": 43755 }, { "epoch": 2.7162455770066423, "grad_norm": 0.39258351488351456, "learning_rate": 2.7005655843330535e-06, "loss": 2.6482, "step": 43756 }, { "epoch": 2.7163076541064, "grad_norm": 0.3749010631065426, "learning_rate": 2.6993948529092373e-06, "loss": 2.6867, "step": 43757 }, { "epoch": 2.716369731206158, "grad_norm": 0.38900026030396917, "learning_rate": 2.6982243682641516e-06, "loss": 2.6778, "step": 43758 }, { "epoch": 2.716431808305916, "grad_norm": 0.3943944070517647, "learning_rate": 2.6970541304039087e-06, "loss": 2.61, "step": 43759 }, { "epoch": 2.716493885405674, "grad_norm": 0.36818607982904894, "learning_rate": 2.6958841393346255e-06, "loss": 2.6332, "step": 43760 }, { "epoch": 2.716555962505432, "grad_norm": 0.36300808880991564, "learning_rate": 2.694714395062403e-06, "loss": 2.6332, "step": 43761 }, { "epoch": 2.7166180396051898, "grad_norm": 0.4031916267575289, "learning_rate": 2.6935448975933252e-06, "loss": 2.6023, "step": 43762 }, { "epoch": 2.7166801167049477, "grad_norm": 0.3688047788633094, "learning_rate": 2.692375646933515e-06, "loss": 2.6663, "step": 43763 }, { "epoch": 2.7167421938047056, "grad_norm": 0.37577862173576965, "learning_rate": 2.6912066430890725e-06, "loss": 2.6724, "step": 43764 }, { "epoch": 2.716804270904463, "grad_norm": 0.39032232001895817, "learning_rate": 2.6900378860660826e-06, "loss": 2.6412, "step": 43765 }, { "epoch": 2.7168663480042214, "grad_norm": 0.4196237969925278, "learning_rate": 2.688869375870645e-06, "loss": 2.6867, "step": 43766 }, { "epoch": 2.716928425103979, "grad_norm": 0.4052615428236116, "learning_rate": 2.687701112508867e-06, "loss": 2.6894, "step": 43767 }, { "epoch": 2.7169905022037373, "grad_norm": 0.39815000371701836, "learning_rate": 2.6865330959868373e-06, "loss": 2.7257, "step": 43768 }, { "epoch": 2.7170525793034948, "grad_norm": 0.41501875883816225, "learning_rate": 2.6853653263106514e-06, "loss": 2.7234, "step": 43769 }, { "epoch": 2.717114656403253, "grad_norm": 0.3738303819227501, "learning_rate": 2.6841978034864047e-06, "loss": 2.6335, "step": 43770 }, { "epoch": 2.7171767335030106, "grad_norm": 0.3787065353255739, "learning_rate": 2.6830305275201696e-06, "loss": 2.6744, "step": 43771 }, { "epoch": 2.7172388106027685, "grad_norm": 0.3788219043412116, "learning_rate": 2.6818634984180633e-06, "loss": 2.6892, "step": 43772 }, { "epoch": 2.7173008877025264, "grad_norm": 0.414769001780129, "learning_rate": 2.6806967161861596e-06, "loss": 2.742, "step": 43773 }, { "epoch": 2.7173629648022843, "grad_norm": 0.40999663744105824, "learning_rate": 2.6795301808305473e-06, "loss": 2.7318, "step": 43774 }, { "epoch": 2.7174250419020423, "grad_norm": 0.4064800821341946, "learning_rate": 2.6783638923573106e-06, "loss": 2.6875, "step": 43775 }, { "epoch": 2.7174871190018, "grad_norm": 0.3900113626499892, "learning_rate": 2.6771978507725394e-06, "loss": 2.7408, "step": 43776 }, { "epoch": 2.717549196101558, "grad_norm": 0.3529247910582655, "learning_rate": 2.6760320560823173e-06, "loss": 2.6612, "step": 43777 }, { "epoch": 2.717611273201316, "grad_norm": 0.36501199408642737, "learning_rate": 2.6748665082927283e-06, "loss": 2.6889, "step": 43778 }, { "epoch": 2.717673350301074, "grad_norm": 0.39242452017188606, "learning_rate": 2.6737012074098345e-06, "loss": 2.6198, "step": 43779 }, { "epoch": 2.717735427400832, "grad_norm": 0.3744935151889639, "learning_rate": 2.672536153439742e-06, "loss": 2.6052, "step": 43780 }, { "epoch": 2.7177975045005898, "grad_norm": 0.35613297665286675, "learning_rate": 2.6713713463885183e-06, "loss": 2.6294, "step": 43781 }, { "epoch": 2.7178595816003477, "grad_norm": 0.35263195557547966, "learning_rate": 2.670206786262236e-06, "loss": 2.6216, "step": 43782 }, { "epoch": 2.7179216587001056, "grad_norm": 0.38792356983124493, "learning_rate": 2.6690424730669796e-06, "loss": 2.629, "step": 43783 }, { "epoch": 2.7179837357998635, "grad_norm": 0.4019090285362737, "learning_rate": 2.6678784068088103e-06, "loss": 2.6608, "step": 43784 }, { "epoch": 2.7180458128996214, "grad_norm": 0.38938879861154985, "learning_rate": 2.6667145874938125e-06, "loss": 2.6575, "step": 43785 }, { "epoch": 2.7181078899993794, "grad_norm": 0.378893920855157, "learning_rate": 2.6655510151280593e-06, "loss": 2.6875, "step": 43786 }, { "epoch": 2.7181699670991373, "grad_norm": 0.359541957883105, "learning_rate": 2.6643876897176122e-06, "loss": 2.6781, "step": 43787 }, { "epoch": 2.718232044198895, "grad_norm": 0.3732554866032366, "learning_rate": 2.6632246112685444e-06, "loss": 2.6035, "step": 43788 }, { "epoch": 2.718294121298653, "grad_norm": 0.37384405391163866, "learning_rate": 2.6620617797869286e-06, "loss": 2.6269, "step": 43789 }, { "epoch": 2.7183561983984106, "grad_norm": 0.3474945699658599, "learning_rate": 2.660899195278832e-06, "loss": 2.6426, "step": 43790 }, { "epoch": 2.718418275498169, "grad_norm": 0.38714248869737133, "learning_rate": 2.6597368577503055e-06, "loss": 2.6812, "step": 43791 }, { "epoch": 2.7184803525979264, "grad_norm": 0.3934007632349966, "learning_rate": 2.6585747672074336e-06, "loss": 2.6841, "step": 43792 }, { "epoch": 2.718542429697685, "grad_norm": 0.3639018183070222, "learning_rate": 2.6574129236562605e-06, "loss": 2.7001, "step": 43793 }, { "epoch": 2.7186045067974423, "grad_norm": 0.3676477145441603, "learning_rate": 2.6562513271028654e-06, "loss": 2.6542, "step": 43794 }, { "epoch": 2.7186665838972, "grad_norm": 0.3793921550347381, "learning_rate": 2.655089977553299e-06, "loss": 2.7251, "step": 43795 }, { "epoch": 2.718728660996958, "grad_norm": 0.3780106056191392, "learning_rate": 2.6539288750136228e-06, "loss": 2.6682, "step": 43796 }, { "epoch": 2.718790738096716, "grad_norm": 0.3578514685920598, "learning_rate": 2.652768019489893e-06, "loss": 2.7358, "step": 43797 }, { "epoch": 2.718852815196474, "grad_norm": 0.36914169472837105, "learning_rate": 2.6516074109881606e-06, "loss": 2.7023, "step": 43798 }, { "epoch": 2.718914892296232, "grad_norm": 0.3884738436796042, "learning_rate": 2.650447049514493e-06, "loss": 2.6214, "step": 43799 }, { "epoch": 2.7189769693959898, "grad_norm": 0.35385821271186596, "learning_rate": 2.649286935074935e-06, "loss": 2.6855, "step": 43800 }, { "epoch": 2.7190390464957477, "grad_norm": 0.3781942382978803, "learning_rate": 2.6481270676755432e-06, "loss": 2.6818, "step": 43801 }, { "epoch": 2.7191011235955056, "grad_norm": 0.42825111981817254, "learning_rate": 2.6469674473223625e-06, "loss": 2.7082, "step": 43802 }, { "epoch": 2.7191632006952635, "grad_norm": 0.3740032977466234, "learning_rate": 2.6458080740214553e-06, "loss": 2.6928, "step": 43803 }, { "epoch": 2.7192252777950214, "grad_norm": 0.38379570176300465, "learning_rate": 2.6446489477788606e-06, "loss": 2.6497, "step": 43804 }, { "epoch": 2.7192873548947794, "grad_norm": 0.40338479697284746, "learning_rate": 2.643490068600629e-06, "loss": 2.5963, "step": 43805 }, { "epoch": 2.7193494319945373, "grad_norm": 0.41092437389572317, "learning_rate": 2.642331436492801e-06, "loss": 2.6269, "step": 43806 }, { "epoch": 2.719411509094295, "grad_norm": 0.3779601841334806, "learning_rate": 2.6411730514614318e-06, "loss": 2.613, "step": 43807 }, { "epoch": 2.719473586194053, "grad_norm": 0.36886822777209266, "learning_rate": 2.640014913512556e-06, "loss": 2.6442, "step": 43808 }, { "epoch": 2.719535663293811, "grad_norm": 0.3807652088053275, "learning_rate": 2.6388570226522247e-06, "loss": 2.6972, "step": 43809 }, { "epoch": 2.719597740393569, "grad_norm": 0.3837902467044489, "learning_rate": 2.637699378886471e-06, "loss": 2.6341, "step": 43810 }, { "epoch": 2.719659817493327, "grad_norm": 0.3917723590030325, "learning_rate": 2.63654198222133e-06, "loss": 2.6686, "step": 43811 }, { "epoch": 2.7197218945930848, "grad_norm": 0.3695887716417185, "learning_rate": 2.635384832662852e-06, "loss": 2.6895, "step": 43812 }, { "epoch": 2.7197839716928423, "grad_norm": 0.38258508817300324, "learning_rate": 2.634227930217076e-06, "loss": 2.6002, "step": 43813 }, { "epoch": 2.7198460487926006, "grad_norm": 0.35467518268321563, "learning_rate": 2.6330712748900254e-06, "loss": 2.6229, "step": 43814 }, { "epoch": 2.719908125892358, "grad_norm": 0.37817508901702324, "learning_rate": 2.6319148666877345e-06, "loss": 2.6747, "step": 43815 }, { "epoch": 2.7199702029921164, "grad_norm": 0.39291117191349345, "learning_rate": 2.630758705616254e-06, "loss": 2.7105, "step": 43816 }, { "epoch": 2.720032280091874, "grad_norm": 0.3699622521927386, "learning_rate": 2.6296027916816013e-06, "loss": 2.6857, "step": 43817 }, { "epoch": 2.7200943571916323, "grad_norm": 0.3908904508825212, "learning_rate": 2.62844712488981e-06, "loss": 2.7029, "step": 43818 }, { "epoch": 2.7201564342913898, "grad_norm": 0.4060761556597062, "learning_rate": 2.627291705246915e-06, "loss": 2.7839, "step": 43819 }, { "epoch": 2.7202185113911477, "grad_norm": 0.38869319071378544, "learning_rate": 2.6261365327589272e-06, "loss": 2.7744, "step": 43820 }, { "epoch": 2.7202805884909056, "grad_norm": 0.3977541392714588, "learning_rate": 2.624981607431892e-06, "loss": 2.6642, "step": 43821 }, { "epoch": 2.7203426655906635, "grad_norm": 0.3990270142994766, "learning_rate": 2.623826929271833e-06, "loss": 2.6235, "step": 43822 }, { "epoch": 2.7204047426904214, "grad_norm": 0.3710713429964246, "learning_rate": 2.6226724982847673e-06, "loss": 2.6569, "step": 43823 }, { "epoch": 2.7204668197901793, "grad_norm": 0.3997201283180392, "learning_rate": 2.621518314476712e-06, "loss": 2.668, "step": 43824 }, { "epoch": 2.7205288968899373, "grad_norm": 0.386500711336752, "learning_rate": 2.620364377853696e-06, "loss": 2.5965, "step": 43825 }, { "epoch": 2.720590973989695, "grad_norm": 0.36463879662934073, "learning_rate": 2.6192106884217526e-06, "loss": 2.6519, "step": 43826 }, { "epoch": 2.720653051089453, "grad_norm": 0.350261123353985, "learning_rate": 2.6180572461868835e-06, "loss": 2.6838, "step": 43827 }, { "epoch": 2.720715128189211, "grad_norm": 0.3846796394741213, "learning_rate": 2.616904051155117e-06, "loss": 2.6521, "step": 43828 }, { "epoch": 2.720777205288969, "grad_norm": 0.39238986915178664, "learning_rate": 2.6157511033324534e-06, "loss": 2.6114, "step": 43829 }, { "epoch": 2.720839282388727, "grad_norm": 0.38099568201982364, "learning_rate": 2.6145984027249326e-06, "loss": 2.6919, "step": 43830 }, { "epoch": 2.7209013594884848, "grad_norm": 0.37259534901988756, "learning_rate": 2.6134459493385497e-06, "loss": 2.625, "step": 43831 }, { "epoch": 2.7209634365882427, "grad_norm": 0.37704658338002656, "learning_rate": 2.6122937431793226e-06, "loss": 2.5372, "step": 43832 }, { "epoch": 2.7210255136880006, "grad_norm": 0.3678208631020092, "learning_rate": 2.611141784253268e-06, "loss": 2.6513, "step": 43833 }, { "epoch": 2.7210875907877585, "grad_norm": 0.37924745278958394, "learning_rate": 2.609990072566376e-06, "loss": 2.7499, "step": 43834 }, { "epoch": 2.7211496678875164, "grad_norm": 0.3944346243497074, "learning_rate": 2.608838608124681e-06, "loss": 2.685, "step": 43835 }, { "epoch": 2.721211744987274, "grad_norm": 0.41128260042670134, "learning_rate": 2.6076873909341827e-06, "loss": 2.6325, "step": 43836 }, { "epoch": 2.7212738220870323, "grad_norm": 0.3665415581121989, "learning_rate": 2.6065364210008824e-06, "loss": 2.6413, "step": 43837 }, { "epoch": 2.7213358991867898, "grad_norm": 0.3638309503715497, "learning_rate": 2.6053856983307758e-06, "loss": 2.6527, "step": 43838 }, { "epoch": 2.721397976286548, "grad_norm": 0.38709235206844933, "learning_rate": 2.6042352229298906e-06, "loss": 2.7777, "step": 43839 }, { "epoch": 2.7214600533863056, "grad_norm": 0.3832580736993253, "learning_rate": 2.6030849948042113e-06, "loss": 2.6491, "step": 43840 }, { "epoch": 2.721522130486064, "grad_norm": 0.38953941177527246, "learning_rate": 2.601935013959744e-06, "loss": 2.75, "step": 43841 }, { "epoch": 2.7215842075858214, "grad_norm": 0.3839193030914695, "learning_rate": 2.6007852804024848e-06, "loss": 2.6022, "step": 43842 }, { "epoch": 2.7216462846855793, "grad_norm": 0.3676345619714359, "learning_rate": 2.5996357941384387e-06, "loss": 2.6105, "step": 43843 }, { "epoch": 2.7217083617853373, "grad_norm": 0.36548363243093807, "learning_rate": 2.5984865551736013e-06, "loss": 2.7133, "step": 43844 }, { "epoch": 2.721770438885095, "grad_norm": 0.37743621467286836, "learning_rate": 2.5973375635139627e-06, "loss": 2.6822, "step": 43845 }, { "epoch": 2.721832515984853, "grad_norm": 0.3962521930236781, "learning_rate": 2.5961888191655293e-06, "loss": 2.7118, "step": 43846 }, { "epoch": 2.721894593084611, "grad_norm": 0.3776035528068376, "learning_rate": 2.5950403221342734e-06, "loss": 2.6594, "step": 43847 }, { "epoch": 2.721956670184369, "grad_norm": 0.34146332951060343, "learning_rate": 2.5938920724262074e-06, "loss": 2.6392, "step": 43848 }, { "epoch": 2.722018747284127, "grad_norm": 0.3591029339791853, "learning_rate": 2.592744070047315e-06, "loss": 2.6958, "step": 43849 }, { "epoch": 2.7220808243838848, "grad_norm": 0.36418705475444296, "learning_rate": 2.591596315003586e-06, "loss": 2.6804, "step": 43850 }, { "epoch": 2.7221429014836427, "grad_norm": 0.3615495776386983, "learning_rate": 2.590448807300999e-06, "loss": 2.6538, "step": 43851 }, { "epoch": 2.7222049785834006, "grad_norm": 0.3549608017879314, "learning_rate": 2.5893015469455605e-06, "loss": 2.6057, "step": 43852 }, { "epoch": 2.7222670556831585, "grad_norm": 0.37068030766488247, "learning_rate": 2.5881545339432434e-06, "loss": 2.7144, "step": 43853 }, { "epoch": 2.7223291327829164, "grad_norm": 0.3769358102251096, "learning_rate": 2.5870077683000315e-06, "loss": 2.6647, "step": 43854 }, { "epoch": 2.7223912098826744, "grad_norm": 0.34938740210110564, "learning_rate": 2.5858612500219148e-06, "loss": 2.6153, "step": 43855 }, { "epoch": 2.7224532869824323, "grad_norm": 0.3462753839690479, "learning_rate": 2.5847149791148604e-06, "loss": 2.6715, "step": 43856 }, { "epoch": 2.72251536408219, "grad_norm": 0.36557849441550744, "learning_rate": 2.5835689555848587e-06, "loss": 2.6685, "step": 43857 }, { "epoch": 2.722577441181948, "grad_norm": 0.3613387647557229, "learning_rate": 2.5824231794378927e-06, "loss": 2.7113, "step": 43858 }, { "epoch": 2.722639518281706, "grad_norm": 0.3669160894944879, "learning_rate": 2.581277650679942e-06, "loss": 2.6841, "step": 43859 }, { "epoch": 2.722701595381464, "grad_norm": 0.3841576201903912, "learning_rate": 2.5801323693169734e-06, "loss": 2.5966, "step": 43860 }, { "epoch": 2.7227636724812214, "grad_norm": 0.36015322636392194, "learning_rate": 2.57898733535496e-06, "loss": 2.6836, "step": 43861 }, { "epoch": 2.72282574958098, "grad_norm": 0.36514291281087274, "learning_rate": 2.5778425487998857e-06, "loss": 2.634, "step": 43862 }, { "epoch": 2.7228878266807373, "grad_norm": 0.3549191344531333, "learning_rate": 2.5766980096577187e-06, "loss": 2.685, "step": 43863 }, { "epoch": 2.7229499037804956, "grad_norm": 0.371077065343724, "learning_rate": 2.575553717934431e-06, "loss": 2.5963, "step": 43864 }, { "epoch": 2.723011980880253, "grad_norm": 0.3619690854797943, "learning_rate": 2.5744096736359914e-06, "loss": 2.6872, "step": 43865 }, { "epoch": 2.723074057980011, "grad_norm": 0.37570559162322525, "learning_rate": 2.5732658767683714e-06, "loss": 2.7112, "step": 43866 }, { "epoch": 2.723136135079769, "grad_norm": 0.35443699094372205, "learning_rate": 2.572122327337534e-06, "loss": 2.6182, "step": 43867 }, { "epoch": 2.723198212179527, "grad_norm": 0.3759699436942768, "learning_rate": 2.5709790253494516e-06, "loss": 2.775, "step": 43868 }, { "epoch": 2.7232602892792848, "grad_norm": 0.3592237900774035, "learning_rate": 2.569835970810086e-06, "loss": 2.7561, "step": 43869 }, { "epoch": 2.7233223663790427, "grad_norm": 0.3484349263528108, "learning_rate": 2.5686931637253943e-06, "loss": 2.6857, "step": 43870 }, { "epoch": 2.7233844434788006, "grad_norm": 0.39803946940203166, "learning_rate": 2.5675506041013485e-06, "loss": 2.6633, "step": 43871 }, { "epoch": 2.7234465205785585, "grad_norm": 0.35529568410115436, "learning_rate": 2.5664082919439058e-06, "loss": 2.6467, "step": 43872 }, { "epoch": 2.7235085976783164, "grad_norm": 0.3857041352250884, "learning_rate": 2.5652662272590276e-06, "loss": 2.6672, "step": 43873 }, { "epoch": 2.7235706747780744, "grad_norm": 0.37748471267011174, "learning_rate": 2.564124410052665e-06, "loss": 2.7287, "step": 43874 }, { "epoch": 2.7236327518778323, "grad_norm": 0.36565256911452, "learning_rate": 2.5629828403307853e-06, "loss": 2.6546, "step": 43875 }, { "epoch": 2.72369482897759, "grad_norm": 0.35476243408085306, "learning_rate": 2.5618415180993394e-06, "loss": 2.6119, "step": 43876 }, { "epoch": 2.723756906077348, "grad_norm": 0.3702620170080386, "learning_rate": 2.5607004433642834e-06, "loss": 2.6969, "step": 43877 }, { "epoch": 2.723818983177106, "grad_norm": 0.3752993739720812, "learning_rate": 2.5595596161315625e-06, "loss": 2.6466, "step": 43878 }, { "epoch": 2.723881060276864, "grad_norm": 0.37056517853189436, "learning_rate": 2.558419036407139e-06, "loss": 2.661, "step": 43879 }, { "epoch": 2.723943137376622, "grad_norm": 0.37651566336828235, "learning_rate": 2.5572787041969694e-06, "loss": 2.7136, "step": 43880 }, { "epoch": 2.7240052144763798, "grad_norm": 0.3986139654687963, "learning_rate": 2.556138619506987e-06, "loss": 2.6765, "step": 43881 }, { "epoch": 2.7240672915761377, "grad_norm": 0.38745507549009234, "learning_rate": 2.554998782343149e-06, "loss": 2.7418, "step": 43882 }, { "epoch": 2.7241293686758956, "grad_norm": 0.37342413954198, "learning_rate": 2.5538591927113887e-06, "loss": 2.6555, "step": 43883 }, { "epoch": 2.724191445775653, "grad_norm": 0.3692321462053443, "learning_rate": 2.552719850617674e-06, "loss": 2.6576, "step": 43884 }, { "epoch": 2.7242535228754114, "grad_norm": 0.3669212065587759, "learning_rate": 2.5515807560679396e-06, "loss": 2.6607, "step": 43885 }, { "epoch": 2.724315599975169, "grad_norm": 0.36502232235099563, "learning_rate": 2.5504419090681244e-06, "loss": 2.6224, "step": 43886 }, { "epoch": 2.7243776770749273, "grad_norm": 0.3731220429269749, "learning_rate": 2.5493033096241624e-06, "loss": 2.7048, "step": 43887 }, { "epoch": 2.7244397541746848, "grad_norm": 0.37316551119566005, "learning_rate": 2.5481649577420164e-06, "loss": 2.6392, "step": 43888 }, { "epoch": 2.724501831274443, "grad_norm": 0.3820971597868949, "learning_rate": 2.5470268534276142e-06, "loss": 2.767, "step": 43889 }, { "epoch": 2.7245639083742006, "grad_norm": 0.38279219982272994, "learning_rate": 2.5458889966868793e-06, "loss": 2.5922, "step": 43890 }, { "epoch": 2.7246259854739585, "grad_norm": 0.39325445190718006, "learning_rate": 2.5447513875257734e-06, "loss": 2.6834, "step": 43891 }, { "epoch": 2.7246880625737164, "grad_norm": 0.37224442620776815, "learning_rate": 2.5436140259502084e-06, "loss": 2.6754, "step": 43892 }, { "epoch": 2.7247501396734743, "grad_norm": 0.3775338162187424, "learning_rate": 2.5424769119661406e-06, "loss": 2.6414, "step": 43893 }, { "epoch": 2.7248122167732323, "grad_norm": 0.3992629009997815, "learning_rate": 2.5413400455794934e-06, "loss": 2.5879, "step": 43894 }, { "epoch": 2.72487429387299, "grad_norm": 0.3828945072072997, "learning_rate": 2.5402034267961896e-06, "loss": 2.5912, "step": 43895 }, { "epoch": 2.724936370972748, "grad_norm": 0.36145609058853306, "learning_rate": 2.5390670556221686e-06, "loss": 2.7026, "step": 43896 }, { "epoch": 2.724998448072506, "grad_norm": 0.3784259623021664, "learning_rate": 2.537930932063348e-06, "loss": 2.7215, "step": 43897 }, { "epoch": 2.725060525172264, "grad_norm": 0.381498045823828, "learning_rate": 2.5367950561256737e-06, "loss": 2.6625, "step": 43898 }, { "epoch": 2.725122602272022, "grad_norm": 0.3753290520860607, "learning_rate": 2.5356594278150626e-06, "loss": 2.6709, "step": 43899 }, { "epoch": 2.7251846793717798, "grad_norm": 0.3768517946023157, "learning_rate": 2.5345240471374376e-06, "loss": 2.6374, "step": 43900 }, { "epoch": 2.7252467564715377, "grad_norm": 0.36069389561145543, "learning_rate": 2.533388914098711e-06, "loss": 2.6406, "step": 43901 }, { "epoch": 2.7253088335712956, "grad_norm": 0.4097317997123745, "learning_rate": 2.532254028704828e-06, "loss": 2.6298, "step": 43902 }, { "epoch": 2.7253709106710535, "grad_norm": 0.3758359770072158, "learning_rate": 2.5311193909617005e-06, "loss": 2.793, "step": 43903 }, { "epoch": 2.7254329877708114, "grad_norm": 0.38319972947664604, "learning_rate": 2.5299850008752456e-06, "loss": 2.6542, "step": 43904 }, { "epoch": 2.7254950648705694, "grad_norm": 0.3609871776461941, "learning_rate": 2.5288508584513816e-06, "loss": 2.6536, "step": 43905 }, { "epoch": 2.7255571419703273, "grad_norm": 0.3779163013226567, "learning_rate": 2.5277169636960194e-06, "loss": 2.617, "step": 43906 }, { "epoch": 2.725619219070085, "grad_norm": 0.38981240630045505, "learning_rate": 2.5265833166150943e-06, "loss": 2.6645, "step": 43907 }, { "epoch": 2.725681296169843, "grad_norm": 0.3835742727955875, "learning_rate": 2.5254499172145006e-06, "loss": 2.6441, "step": 43908 }, { "epoch": 2.7257433732696006, "grad_norm": 0.3979938692658251, "learning_rate": 2.5243167655001677e-06, "loss": 2.6236, "step": 43909 }, { "epoch": 2.725805450369359, "grad_norm": 0.33783304683084736, "learning_rate": 2.5231838614779846e-06, "loss": 2.657, "step": 43910 }, { "epoch": 2.7258675274691164, "grad_norm": 0.37232315803732535, "learning_rate": 2.522051205153886e-06, "loss": 2.7461, "step": 43911 }, { "epoch": 2.725929604568875, "grad_norm": 0.38882642577361276, "learning_rate": 2.5209187965337776e-06, "loss": 2.6499, "step": 43912 }, { "epoch": 2.7259916816686323, "grad_norm": 0.39605584499311514, "learning_rate": 2.5197866356235555e-06, "loss": 2.688, "step": 43913 }, { "epoch": 2.72605375876839, "grad_norm": 0.3962406480449743, "learning_rate": 2.5186547224291255e-06, "loss": 2.703, "step": 43914 }, { "epoch": 2.726115835868148, "grad_norm": 0.36324199341353347, "learning_rate": 2.5175230569564114e-06, "loss": 2.6345, "step": 43915 }, { "epoch": 2.726177912967906, "grad_norm": 0.377864979963491, "learning_rate": 2.5163916392113017e-06, "loss": 2.5488, "step": 43916 }, { "epoch": 2.726239990067664, "grad_norm": 0.3749241752259839, "learning_rate": 2.5152604691997096e-06, "loss": 2.6962, "step": 43917 }, { "epoch": 2.726302067167422, "grad_norm": 0.4078361650521126, "learning_rate": 2.5141295469275295e-06, "loss": 2.6162, "step": 43918 }, { "epoch": 2.7263641442671798, "grad_norm": 0.37757817759909124, "learning_rate": 2.5129988724006513e-06, "loss": 2.7149, "step": 43919 }, { "epoch": 2.7264262213669377, "grad_norm": 0.39189156237194644, "learning_rate": 2.5118684456249987e-06, "loss": 2.735, "step": 43920 }, { "epoch": 2.7264882984666956, "grad_norm": 0.3543027261168847, "learning_rate": 2.5107382666064494e-06, "loss": 2.6812, "step": 43921 }, { "epoch": 2.7265503755664535, "grad_norm": 0.3717523252277216, "learning_rate": 2.509608335350905e-06, "loss": 2.6175, "step": 43922 }, { "epoch": 2.7266124526662114, "grad_norm": 0.381897767663451, "learning_rate": 2.5084786518642713e-06, "loss": 2.5445, "step": 43923 }, { "epoch": 2.7266745297659694, "grad_norm": 0.3515095670905251, "learning_rate": 2.5073492161524214e-06, "loss": 2.6199, "step": 43924 }, { "epoch": 2.7267366068657273, "grad_norm": 0.37994554879294123, "learning_rate": 2.5062200282212734e-06, "loss": 2.7057, "step": 43925 }, { "epoch": 2.726798683965485, "grad_norm": 0.38429054712590666, "learning_rate": 2.5050910880767e-06, "loss": 2.67, "step": 43926 }, { "epoch": 2.726860761065243, "grad_norm": 0.387940144523259, "learning_rate": 2.503962395724596e-06, "loss": 2.6261, "step": 43927 }, { "epoch": 2.726922838165001, "grad_norm": 0.363967923334807, "learning_rate": 2.5028339511708466e-06, "loss": 2.71, "step": 43928 }, { "epoch": 2.726984915264759, "grad_norm": 0.38058238594271926, "learning_rate": 2.5017057544213464e-06, "loss": 2.7077, "step": 43929 }, { "epoch": 2.727046992364517, "grad_norm": 0.38760176906580573, "learning_rate": 2.5005778054819796e-06, "loss": 2.6279, "step": 43930 }, { "epoch": 2.727109069464275, "grad_norm": 0.37482094463575993, "learning_rate": 2.4994501043586306e-06, "loss": 2.6426, "step": 43931 }, { "epoch": 2.7271711465640323, "grad_norm": 0.3946997216719252, "learning_rate": 2.4983226510571777e-06, "loss": 2.6666, "step": 43932 }, { "epoch": 2.7272332236637906, "grad_norm": 0.37294424352850875, "learning_rate": 2.4971954455835056e-06, "loss": 2.5842, "step": 43933 }, { "epoch": 2.727295300763548, "grad_norm": 0.3827440434184851, "learning_rate": 2.496068487943498e-06, "loss": 2.6589, "step": 43934 }, { "epoch": 2.7273573778633065, "grad_norm": 0.3804812192342275, "learning_rate": 2.494941778143034e-06, "loss": 2.5838, "step": 43935 }, { "epoch": 2.727419454963064, "grad_norm": 0.3643671643770726, "learning_rate": 2.4938153161879975e-06, "loss": 2.6447, "step": 43936 }, { "epoch": 2.7274815320628223, "grad_norm": 0.36071992153812593, "learning_rate": 2.4926891020842446e-06, "loss": 2.7058, "step": 43937 }, { "epoch": 2.7275436091625798, "grad_norm": 0.35199433401950003, "learning_rate": 2.491563135837677e-06, "loss": 2.6229, "step": 43938 }, { "epoch": 2.7276056862623377, "grad_norm": 0.3528594677074523, "learning_rate": 2.49043741745415e-06, "loss": 2.7072, "step": 43939 }, { "epoch": 2.7276677633620956, "grad_norm": 0.34291359625230106, "learning_rate": 2.4893119469395543e-06, "loss": 2.6974, "step": 43940 }, { "epoch": 2.7277298404618535, "grad_norm": 0.34916752859902955, "learning_rate": 2.4881867242997458e-06, "loss": 2.554, "step": 43941 }, { "epoch": 2.7277919175616114, "grad_norm": 0.3614209236368989, "learning_rate": 2.4870617495405924e-06, "loss": 2.6882, "step": 43942 }, { "epoch": 2.7278539946613694, "grad_norm": 0.36768025956586475, "learning_rate": 2.4859370226679833e-06, "loss": 2.5518, "step": 43943 }, { "epoch": 2.7279160717611273, "grad_norm": 0.3512983601426796, "learning_rate": 2.4848125436877757e-06, "loss": 2.6619, "step": 43944 }, { "epoch": 2.727978148860885, "grad_norm": 0.3773132935863653, "learning_rate": 2.4836883126058365e-06, "loss": 2.6341, "step": 43945 }, { "epoch": 2.728040225960643, "grad_norm": 0.3603690331832082, "learning_rate": 2.4825643294280166e-06, "loss": 2.7631, "step": 43946 }, { "epoch": 2.728102303060401, "grad_norm": 0.4016982891698733, "learning_rate": 2.481440594160206e-06, "loss": 2.6512, "step": 43947 }, { "epoch": 2.728164380160159, "grad_norm": 0.36476037177516735, "learning_rate": 2.480317106808255e-06, "loss": 2.6137, "step": 43948 }, { "epoch": 2.728226457259917, "grad_norm": 0.3467297585017463, "learning_rate": 2.4791938673780268e-06, "loss": 2.7117, "step": 43949 }, { "epoch": 2.7282885343596748, "grad_norm": 0.37489506127413214, "learning_rate": 2.4780708758753712e-06, "loss": 2.6331, "step": 43950 }, { "epoch": 2.7283506114594327, "grad_norm": 0.3605257352539632, "learning_rate": 2.476948132306162e-06, "loss": 2.7002, "step": 43951 }, { "epoch": 2.7284126885591906, "grad_norm": 0.3548662718223235, "learning_rate": 2.4758256366762545e-06, "loss": 2.7049, "step": 43952 }, { "epoch": 2.7284747656589485, "grad_norm": 0.3469301420492251, "learning_rate": 2.474703388991495e-06, "loss": 2.6473, "step": 43953 }, { "epoch": 2.7285368427587064, "grad_norm": 0.3896074010517118, "learning_rate": 2.473581389257751e-06, "loss": 2.6047, "step": 43954 }, { "epoch": 2.7285989198584644, "grad_norm": 0.3516888603311591, "learning_rate": 2.472459637480862e-06, "loss": 2.5924, "step": 43955 }, { "epoch": 2.7286609969582223, "grad_norm": 0.35463724242422184, "learning_rate": 2.4713381336666896e-06, "loss": 2.5964, "step": 43956 }, { "epoch": 2.7287230740579798, "grad_norm": 0.35073211937569615, "learning_rate": 2.470216877821091e-06, "loss": 2.5747, "step": 43957 }, { "epoch": 2.728785151157738, "grad_norm": 0.34793323027230144, "learning_rate": 2.4690958699499055e-06, "loss": 2.5811, "step": 43958 }, { "epoch": 2.7288472282574956, "grad_norm": 0.36608734079661576, "learning_rate": 2.46797511005899e-06, "loss": 2.7085, "step": 43959 }, { "epoch": 2.728909305357254, "grad_norm": 0.3968616813767772, "learning_rate": 2.466854598154178e-06, "loss": 2.7543, "step": 43960 }, { "epoch": 2.7289713824570114, "grad_norm": 0.36297812171821675, "learning_rate": 2.4657343342413317e-06, "loss": 2.7232, "step": 43961 }, { "epoch": 2.7290334595567693, "grad_norm": 0.36389606502525074, "learning_rate": 2.4646143183262916e-06, "loss": 2.7108, "step": 43962 }, { "epoch": 2.7290955366565273, "grad_norm": 0.371501497822716, "learning_rate": 2.4634945504148964e-06, "loss": 2.6917, "step": 43963 }, { "epoch": 2.729157613756285, "grad_norm": 0.36810205073947355, "learning_rate": 2.462375030512981e-06, "loss": 2.6692, "step": 43964 }, { "epoch": 2.729219690856043, "grad_norm": 0.3369932772626642, "learning_rate": 2.461255758626402e-06, "loss": 2.6897, "step": 43965 }, { "epoch": 2.729281767955801, "grad_norm": 0.4010260029826081, "learning_rate": 2.4601367347609926e-06, "loss": 2.6802, "step": 43966 }, { "epoch": 2.729343845055559, "grad_norm": 0.36181228779496455, "learning_rate": 2.4590179589225936e-06, "loss": 2.5665, "step": 43967 }, { "epoch": 2.729405922155317, "grad_norm": 0.35859564570837976, "learning_rate": 2.4578994311170387e-06, "loss": 2.6987, "step": 43968 }, { "epoch": 2.7294679992550748, "grad_norm": 0.3693707066978702, "learning_rate": 2.4567811513501514e-06, "loss": 2.6637, "step": 43969 }, { "epoch": 2.7295300763548327, "grad_norm": 0.3486682435739159, "learning_rate": 2.4556631196277935e-06, "loss": 2.6853, "step": 43970 }, { "epoch": 2.7295921534545906, "grad_norm": 0.36454994412072383, "learning_rate": 2.454545335955777e-06, "loss": 2.6719, "step": 43971 }, { "epoch": 2.7296542305543485, "grad_norm": 0.353058398135319, "learning_rate": 2.453427800339941e-06, "loss": 2.6146, "step": 43972 }, { "epoch": 2.7297163076541064, "grad_norm": 0.3705437449243235, "learning_rate": 2.4523105127861044e-06, "loss": 2.6483, "step": 43973 }, { "epoch": 2.7297783847538644, "grad_norm": 0.3543690659431531, "learning_rate": 2.451193473300117e-06, "loss": 2.6499, "step": 43974 }, { "epoch": 2.7298404618536223, "grad_norm": 0.37936746900898355, "learning_rate": 2.4500766818877973e-06, "loss": 2.6309, "step": 43975 }, { "epoch": 2.72990253895338, "grad_norm": 0.36031115959295323, "learning_rate": 2.4489601385549677e-06, "loss": 2.594, "step": 43976 }, { "epoch": 2.729964616053138, "grad_norm": 0.35299450311352304, "learning_rate": 2.4478438433074514e-06, "loss": 2.5919, "step": 43977 }, { "epoch": 2.730026693152896, "grad_norm": 0.3523865367175436, "learning_rate": 2.4467277961510826e-06, "loss": 2.693, "step": 43978 }, { "epoch": 2.730088770252654, "grad_norm": 0.35970415714779147, "learning_rate": 2.4456119970916847e-06, "loss": 2.6313, "step": 43979 }, { "epoch": 2.7301508473524114, "grad_norm": 0.3681061190635194, "learning_rate": 2.4444964461350637e-06, "loss": 2.6786, "step": 43980 }, { "epoch": 2.73021292445217, "grad_norm": 0.3911829587517696, "learning_rate": 2.4433811432870545e-06, "loss": 2.663, "step": 43981 }, { "epoch": 2.7302750015519273, "grad_norm": 0.3799011108384707, "learning_rate": 2.4422660885534633e-06, "loss": 2.6759, "step": 43982 }, { "epoch": 2.7303370786516856, "grad_norm": 0.3959155771076838, "learning_rate": 2.4411512819401238e-06, "loss": 2.8015, "step": 43983 }, { "epoch": 2.730399155751443, "grad_norm": 0.39904848637243023, "learning_rate": 2.440036723452843e-06, "loss": 2.607, "step": 43984 }, { "epoch": 2.7304612328512015, "grad_norm": 0.3395854582261769, "learning_rate": 2.438922413097433e-06, "loss": 2.6648, "step": 43985 }, { "epoch": 2.730523309950959, "grad_norm": 0.3561318887705077, "learning_rate": 2.4378083508797057e-06, "loss": 2.694, "step": 43986 }, { "epoch": 2.730585387050717, "grad_norm": 0.35202139468563226, "learning_rate": 2.436694536805484e-06, "loss": 2.5859, "step": 43987 }, { "epoch": 2.7306474641504748, "grad_norm": 0.38277147034288406, "learning_rate": 2.435580970880569e-06, "loss": 2.6659, "step": 43988 }, { "epoch": 2.7307095412502327, "grad_norm": 0.377694246355904, "learning_rate": 2.434467653110778e-06, "loss": 2.5494, "step": 43989 }, { "epoch": 2.7307716183499906, "grad_norm": 0.3569420475617485, "learning_rate": 2.433354583501918e-06, "loss": 2.6226, "step": 43990 }, { "epoch": 2.7308336954497485, "grad_norm": 0.338604718230024, "learning_rate": 2.432241762059795e-06, "loss": 2.5856, "step": 43991 }, { "epoch": 2.7308957725495064, "grad_norm": 0.3737672172204057, "learning_rate": 2.43112918879021e-06, "loss": 2.6161, "step": 43992 }, { "epoch": 2.7309578496492644, "grad_norm": 0.36186990072185704, "learning_rate": 2.430016863698975e-06, "loss": 2.6848, "step": 43993 }, { "epoch": 2.7310199267490223, "grad_norm": 0.3798838912046361, "learning_rate": 2.4289047867918915e-06, "loss": 2.6245, "step": 43994 }, { "epoch": 2.73108200384878, "grad_norm": 0.36139752520540885, "learning_rate": 2.427792958074754e-06, "loss": 2.6814, "step": 43995 }, { "epoch": 2.731144080948538, "grad_norm": 0.36874348101268506, "learning_rate": 2.4266813775533693e-06, "loss": 2.6424, "step": 43996 }, { "epoch": 2.731206158048296, "grad_norm": 0.3699799134627318, "learning_rate": 2.425570045233544e-06, "loss": 2.6464, "step": 43997 }, { "epoch": 2.731268235148054, "grad_norm": 0.39142996817049724, "learning_rate": 2.424458961121062e-06, "loss": 2.6999, "step": 43998 }, { "epoch": 2.731330312247812, "grad_norm": 0.36353528974490973, "learning_rate": 2.423348125221736e-06, "loss": 2.597, "step": 43999 }, { "epoch": 2.73139238934757, "grad_norm": 0.37726848633598037, "learning_rate": 2.4222375375413387e-06, "loss": 2.5551, "step": 44000 }, { "epoch": 2.7314544664473277, "grad_norm": 0.3886869609722909, "learning_rate": 2.4211271980856877e-06, "loss": 2.663, "step": 44001 }, { "epoch": 2.7315165435470856, "grad_norm": 0.38038562751788757, "learning_rate": 2.420017106860567e-06, "loss": 2.6861, "step": 44002 }, { "epoch": 2.7315786206468435, "grad_norm": 0.3817011212257266, "learning_rate": 2.4189072638717668e-06, "loss": 2.6739, "step": 44003 }, { "epoch": 2.7316406977466015, "grad_norm": 0.3865522353675975, "learning_rate": 2.417797669125077e-06, "loss": 2.6537, "step": 44004 }, { "epoch": 2.731702774846359, "grad_norm": 0.41946261819550834, "learning_rate": 2.4166883226262816e-06, "loss": 2.6609, "step": 44005 }, { "epoch": 2.7317648519461173, "grad_norm": 0.3762395756250093, "learning_rate": 2.4155792243811816e-06, "loss": 2.5792, "step": 44006 }, { "epoch": 2.7318269290458748, "grad_norm": 0.39047669456432227, "learning_rate": 2.4144703743955553e-06, "loss": 2.6984, "step": 44007 }, { "epoch": 2.731889006145633, "grad_norm": 0.37983038376233014, "learning_rate": 2.4133617726751876e-06, "loss": 2.6434, "step": 44008 }, { "epoch": 2.7319510832453906, "grad_norm": 0.3802698316437425, "learning_rate": 2.412253419225863e-06, "loss": 2.7329, "step": 44009 }, { "epoch": 2.7320131603451485, "grad_norm": 0.3486373110355394, "learning_rate": 2.4111453140533647e-06, "loss": 2.6407, "step": 44010 }, { "epoch": 2.7320752374449064, "grad_norm": 0.3865541435331468, "learning_rate": 2.410037457163478e-06, "loss": 2.7, "step": 44011 }, { "epoch": 2.7321373145446644, "grad_norm": 0.3832479647067896, "learning_rate": 2.408929848561975e-06, "loss": 2.6811, "step": 44012 }, { "epoch": 2.7321993916444223, "grad_norm": 0.34380066021503014, "learning_rate": 2.4078224882546354e-06, "loss": 2.632, "step": 44013 }, { "epoch": 2.73226146874418, "grad_norm": 0.3422277840543642, "learning_rate": 2.4067153762472426e-06, "loss": 2.5917, "step": 44014 }, { "epoch": 2.732323545843938, "grad_norm": 0.36498588635971, "learning_rate": 2.405608512545571e-06, "loss": 2.722, "step": 44015 }, { "epoch": 2.732385622943696, "grad_norm": 0.37317443584062926, "learning_rate": 2.4045018971553924e-06, "loss": 2.6654, "step": 44016 }, { "epoch": 2.732447700043454, "grad_norm": 0.372448066124957, "learning_rate": 2.403395530082486e-06, "loss": 2.7699, "step": 44017 }, { "epoch": 2.732509777143212, "grad_norm": 0.3998596019516906, "learning_rate": 2.4022894113326087e-06, "loss": 2.6282, "step": 44018 }, { "epoch": 2.7325718542429698, "grad_norm": 0.3767136124173951, "learning_rate": 2.40118354091155e-06, "loss": 2.7165, "step": 44019 }, { "epoch": 2.7326339313427277, "grad_norm": 0.3812073120857914, "learning_rate": 2.400077918825072e-06, "loss": 2.6202, "step": 44020 }, { "epoch": 2.7326960084424856, "grad_norm": 0.3581153728862138, "learning_rate": 2.3989725450789313e-06, "loss": 2.7179, "step": 44021 }, { "epoch": 2.7327580855422435, "grad_norm": 0.34439310690879993, "learning_rate": 2.397867419678923e-06, "loss": 2.7562, "step": 44022 }, { "epoch": 2.7328201626420014, "grad_norm": 0.365586035029719, "learning_rate": 2.3967625426307816e-06, "loss": 2.6761, "step": 44023 }, { "epoch": 2.7328822397417594, "grad_norm": 0.3589213867051326, "learning_rate": 2.3956579139402967e-06, "loss": 2.5874, "step": 44024 }, { "epoch": 2.7329443168415173, "grad_norm": 0.358613347367812, "learning_rate": 2.3945535336132195e-06, "loss": 2.7251, "step": 44025 }, { "epoch": 2.733006393941275, "grad_norm": 0.3917025390727744, "learning_rate": 2.3934494016553123e-06, "loss": 2.6954, "step": 44026 }, { "epoch": 2.733068471041033, "grad_norm": 0.39550585334198407, "learning_rate": 2.392345518072331e-06, "loss": 2.6895, "step": 44027 }, { "epoch": 2.7331305481407906, "grad_norm": 0.36504369235789313, "learning_rate": 2.391241882870043e-06, "loss": 2.6781, "step": 44028 }, { "epoch": 2.733192625240549, "grad_norm": 0.3529223844007975, "learning_rate": 2.390138496054212e-06, "loss": 2.7112, "step": 44029 }, { "epoch": 2.7332547023403064, "grad_norm": 0.365774601709304, "learning_rate": 2.3890353576305755e-06, "loss": 2.6328, "step": 44030 }, { "epoch": 2.733316779440065, "grad_norm": 0.37340861651715984, "learning_rate": 2.3879324676049088e-06, "loss": 2.7799, "step": 44031 }, { "epoch": 2.7333788565398223, "grad_norm": 0.3871133120862991, "learning_rate": 2.3868298259829447e-06, "loss": 2.757, "step": 44032 }, { "epoch": 2.7334409336395806, "grad_norm": 0.3956749645278692, "learning_rate": 2.3857274327704572e-06, "loss": 2.6924, "step": 44033 }, { "epoch": 2.733503010739338, "grad_norm": 0.380266422467607, "learning_rate": 2.3846252879731857e-06, "loss": 2.6721, "step": 44034 }, { "epoch": 2.733565087839096, "grad_norm": 0.37256383329734294, "learning_rate": 2.3835233915968867e-06, "loss": 2.694, "step": 44035 }, { "epoch": 2.733627164938854, "grad_norm": 0.3801533929330981, "learning_rate": 2.382421743647295e-06, "loss": 2.5503, "step": 44036 }, { "epoch": 2.733689242038612, "grad_norm": 0.37652714036992097, "learning_rate": 2.381320344130178e-06, "loss": 2.6894, "step": 44037 }, { "epoch": 2.7337513191383698, "grad_norm": 0.39410553049114133, "learning_rate": 2.3802191930512753e-06, "loss": 2.745, "step": 44038 }, { "epoch": 2.7338133962381277, "grad_norm": 0.3600081588231194, "learning_rate": 2.3791182904163267e-06, "loss": 2.6822, "step": 44039 }, { "epoch": 2.7338754733378856, "grad_norm": 0.35982387953311923, "learning_rate": 2.3780176362310722e-06, "loss": 2.6462, "step": 44040 }, { "epoch": 2.7339375504376435, "grad_norm": 0.35801141615332255, "learning_rate": 2.376917230501263e-06, "loss": 2.7221, "step": 44041 }, { "epoch": 2.7339996275374014, "grad_norm": 0.3772338849411996, "learning_rate": 2.375817073232639e-06, "loss": 2.6737, "step": 44042 }, { "epoch": 2.7340617046371594, "grad_norm": 0.40042069485027854, "learning_rate": 2.3747171644309398e-06, "loss": 2.6158, "step": 44043 }, { "epoch": 2.7341237817369173, "grad_norm": 0.376965814090286, "learning_rate": 2.3736175041018993e-06, "loss": 2.681, "step": 44044 }, { "epoch": 2.734185858836675, "grad_norm": 0.37476810596846444, "learning_rate": 2.372518092251258e-06, "loss": 2.6923, "step": 44045 }, { "epoch": 2.734247935936433, "grad_norm": 0.36299058093968395, "learning_rate": 2.3714189288847555e-06, "loss": 2.6042, "step": 44046 }, { "epoch": 2.734310013036191, "grad_norm": 0.3671741280251418, "learning_rate": 2.3703200140081205e-06, "loss": 2.6603, "step": 44047 }, { "epoch": 2.734372090135949, "grad_norm": 0.35860038285780177, "learning_rate": 2.3692213476270874e-06, "loss": 2.6043, "step": 44048 }, { "epoch": 2.734434167235707, "grad_norm": 0.3856069400150457, "learning_rate": 2.368122929747385e-06, "loss": 2.635, "step": 44049 }, { "epoch": 2.734496244335465, "grad_norm": 0.36420274579051615, "learning_rate": 2.3670247603747587e-06, "loss": 2.6482, "step": 44050 }, { "epoch": 2.7345583214352227, "grad_norm": 0.40494479906981584, "learning_rate": 2.365926839514926e-06, "loss": 2.7262, "step": 44051 }, { "epoch": 2.7346203985349806, "grad_norm": 0.3474943662776802, "learning_rate": 2.3648291671736102e-06, "loss": 2.6618, "step": 44052 }, { "epoch": 2.734682475634738, "grad_norm": 0.3531717978331703, "learning_rate": 2.3637317433565454e-06, "loss": 2.5697, "step": 44053 }, { "epoch": 2.7347445527344965, "grad_norm": 0.3840352984139042, "learning_rate": 2.3626345680694606e-06, "loss": 2.6857, "step": 44054 }, { "epoch": 2.734806629834254, "grad_norm": 0.41091086587787223, "learning_rate": 2.361537641318068e-06, "loss": 2.7954, "step": 44055 }, { "epoch": 2.7348687069340123, "grad_norm": 0.3876767478360798, "learning_rate": 2.3604409631081017e-06, "loss": 2.7262, "step": 44056 }, { "epoch": 2.7349307840337698, "grad_norm": 0.34741091912307637, "learning_rate": 2.3593445334452845e-06, "loss": 2.7129, "step": 44057 }, { "epoch": 2.7349928611335277, "grad_norm": 0.3749057333401243, "learning_rate": 2.3582483523353295e-06, "loss": 2.6829, "step": 44058 }, { "epoch": 2.7350549382332856, "grad_norm": 0.35680392267338523, "learning_rate": 2.3571524197839535e-06, "loss": 2.6586, "step": 44059 }, { "epoch": 2.7351170153330435, "grad_norm": 0.37601991874102697, "learning_rate": 2.3560567357968853e-06, "loss": 2.7747, "step": 44060 }, { "epoch": 2.7351790924328014, "grad_norm": 0.38970386617596237, "learning_rate": 2.354961300379832e-06, "loss": 2.6148, "step": 44061 }, { "epoch": 2.7352411695325594, "grad_norm": 0.35766301593238686, "learning_rate": 2.3538661135385167e-06, "loss": 2.5838, "step": 44062 }, { "epoch": 2.7353032466323173, "grad_norm": 0.377611780637951, "learning_rate": 2.35277117527864e-06, "loss": 2.6239, "step": 44063 }, { "epoch": 2.735365323732075, "grad_norm": 0.37110511360839293, "learning_rate": 2.3516764856059305e-06, "loss": 2.6655, "step": 44064 }, { "epoch": 2.735427400831833, "grad_norm": 0.3522317979162718, "learning_rate": 2.3505820445260897e-06, "loss": 2.6877, "step": 44065 }, { "epoch": 2.735489477931591, "grad_norm": 0.3700862458916235, "learning_rate": 2.3494878520448293e-06, "loss": 2.5513, "step": 44066 }, { "epoch": 2.735551555031349, "grad_norm": 0.36285498436874736, "learning_rate": 2.3483939081678618e-06, "loss": 2.6697, "step": 44067 }, { "epoch": 2.735613632131107, "grad_norm": 0.3825370420334212, "learning_rate": 2.347300212900877e-06, "loss": 2.6968, "step": 44068 }, { "epoch": 2.735675709230865, "grad_norm": 0.36908901196895305, "learning_rate": 2.346206766249609e-06, "loss": 2.6917, "step": 44069 }, { "epoch": 2.7357377863306227, "grad_norm": 0.3859031438014413, "learning_rate": 2.3451135682197478e-06, "loss": 2.7125, "step": 44070 }, { "epoch": 2.7357998634303806, "grad_norm": 0.3491424340615018, "learning_rate": 2.3440206188169946e-06, "loss": 2.5833, "step": 44071 }, { "epoch": 2.7358619405301385, "grad_norm": 0.35796627421794647, "learning_rate": 2.3429279180470496e-06, "loss": 2.5489, "step": 44072 }, { "epoch": 2.7359240176298965, "grad_norm": 0.36415704146321853, "learning_rate": 2.341835465915626e-06, "loss": 2.7349, "step": 44073 }, { "epoch": 2.7359860947296544, "grad_norm": 0.3691234281863635, "learning_rate": 2.3407432624284186e-06, "loss": 2.6598, "step": 44074 }, { "epoch": 2.7360481718294123, "grad_norm": 0.3747156177498323, "learning_rate": 2.3396513075911175e-06, "loss": 2.7208, "step": 44075 }, { "epoch": 2.7361102489291698, "grad_norm": 0.36344655695591244, "learning_rate": 2.3385596014094293e-06, "loss": 2.6744, "step": 44076 }, { "epoch": 2.736172326028928, "grad_norm": 0.38677022572099895, "learning_rate": 2.337468143889038e-06, "loss": 2.6067, "step": 44077 }, { "epoch": 2.7362344031286856, "grad_norm": 0.37110158008546407, "learning_rate": 2.3363769350356555e-06, "loss": 2.7443, "step": 44078 }, { "epoch": 2.736296480228444, "grad_norm": 0.35417167108403425, "learning_rate": 2.3352859748549617e-06, "loss": 2.6133, "step": 44079 }, { "epoch": 2.7363585573282014, "grad_norm": 0.3741758548943704, "learning_rate": 2.334195263352651e-06, "loss": 2.6883, "step": 44080 }, { "epoch": 2.73642063442796, "grad_norm": 0.4077746501619177, "learning_rate": 2.3331048005344136e-06, "loss": 2.6743, "step": 44081 }, { "epoch": 2.7364827115277173, "grad_norm": 0.35938101224431185, "learning_rate": 2.3320145864059393e-06, "loss": 2.666, "step": 44082 }, { "epoch": 2.736544788627475, "grad_norm": 0.3731322665121988, "learning_rate": 2.330924620972924e-06, "loss": 2.6604, "step": 44083 }, { "epoch": 2.736606865727233, "grad_norm": 0.35622811166546214, "learning_rate": 2.3298349042410407e-06, "loss": 2.5992, "step": 44084 }, { "epoch": 2.736668942826991, "grad_norm": 0.37374963524331756, "learning_rate": 2.328745436215979e-06, "loss": 2.7647, "step": 44085 }, { "epoch": 2.736731019926749, "grad_norm": 0.3741948703481557, "learning_rate": 2.3276562169034235e-06, "loss": 2.6328, "step": 44086 }, { "epoch": 2.736793097026507, "grad_norm": 0.3731537511026455, "learning_rate": 2.3265672463090636e-06, "loss": 2.6869, "step": 44087 }, { "epoch": 2.736855174126265, "grad_norm": 0.35954647534297524, "learning_rate": 2.325478524438579e-06, "loss": 2.6066, "step": 44088 }, { "epoch": 2.7369172512260227, "grad_norm": 0.38562204928637134, "learning_rate": 2.3243900512976425e-06, "loss": 2.6918, "step": 44089 }, { "epoch": 2.7369793283257806, "grad_norm": 0.355952656618298, "learning_rate": 2.3233018268919383e-06, "loss": 2.6289, "step": 44090 }, { "epoch": 2.7370414054255385, "grad_norm": 0.35871931051115274, "learning_rate": 2.3222138512271394e-06, "loss": 2.7499, "step": 44091 }, { "epoch": 2.7371034825252964, "grad_norm": 0.36851416012745974, "learning_rate": 2.3211261243089256e-06, "loss": 2.6429, "step": 44092 }, { "epoch": 2.7371655596250544, "grad_norm": 0.34433061159616296, "learning_rate": 2.3200386461429745e-06, "loss": 2.6434, "step": 44093 }, { "epoch": 2.7372276367248123, "grad_norm": 0.3591711253835743, "learning_rate": 2.3189514167349545e-06, "loss": 2.6699, "step": 44094 }, { "epoch": 2.73728971382457, "grad_norm": 0.37915355437875337, "learning_rate": 2.3178644360905335e-06, "loss": 2.63, "step": 44095 }, { "epoch": 2.737351790924328, "grad_norm": 0.3676248137672365, "learning_rate": 2.3167777042154005e-06, "loss": 2.7085, "step": 44096 }, { "epoch": 2.737413868024086, "grad_norm": 0.37373288973424607, "learning_rate": 2.315691221115207e-06, "loss": 2.6539, "step": 44097 }, { "epoch": 2.737475945123844, "grad_norm": 0.35548134666173414, "learning_rate": 2.314604986795632e-06, "loss": 2.6454, "step": 44098 }, { "epoch": 2.737538022223602, "grad_norm": 0.3952671967432295, "learning_rate": 2.313519001262332e-06, "loss": 2.692, "step": 44099 }, { "epoch": 2.73760009932336, "grad_norm": 0.3897583120596307, "learning_rate": 2.3124332645209855e-06, "loss": 2.7833, "step": 44100 }, { "epoch": 2.7376621764231173, "grad_norm": 0.33237942784341573, "learning_rate": 2.3113477765772494e-06, "loss": 2.6249, "step": 44101 }, { "epoch": 2.7377242535228756, "grad_norm": 0.34141474479739276, "learning_rate": 2.3102625374367913e-06, "loss": 2.6293, "step": 44102 }, { "epoch": 2.737786330622633, "grad_norm": 0.38443997927111073, "learning_rate": 2.3091775471052733e-06, "loss": 2.6554, "step": 44103 }, { "epoch": 2.7378484077223915, "grad_norm": 0.3721196099410872, "learning_rate": 2.308092805588341e-06, "loss": 2.7128, "step": 44104 }, { "epoch": 2.737910484822149, "grad_norm": 0.38483376069139485, "learning_rate": 2.3070083128916785e-06, "loss": 2.673, "step": 44105 }, { "epoch": 2.737972561921907, "grad_norm": 0.3768516759787845, "learning_rate": 2.305924069020926e-06, "loss": 2.6503, "step": 44106 }, { "epoch": 2.7380346390216648, "grad_norm": 0.3658646841272882, "learning_rate": 2.3048400739817454e-06, "loss": 2.6728, "step": 44107 }, { "epoch": 2.7380967161214227, "grad_norm": 0.3653605238167153, "learning_rate": 2.3037563277797937e-06, "loss": 2.7101, "step": 44108 }, { "epoch": 2.7381587932211806, "grad_norm": 0.36171427532419115, "learning_rate": 2.302672830420721e-06, "loss": 2.4619, "step": 44109 }, { "epoch": 2.7382208703209385, "grad_norm": 0.3606097735821731, "learning_rate": 2.3015895819101906e-06, "loss": 2.6104, "step": 44110 }, { "epoch": 2.7382829474206964, "grad_norm": 0.40019893244329774, "learning_rate": 2.300506582253842e-06, "loss": 2.7048, "step": 44111 }, { "epoch": 2.7383450245204544, "grad_norm": 0.36459251346837834, "learning_rate": 2.2994238314573314e-06, "loss": 2.6435, "step": 44112 }, { "epoch": 2.7384071016202123, "grad_norm": 0.41160360821818975, "learning_rate": 2.298341329526299e-06, "loss": 2.686, "step": 44113 }, { "epoch": 2.73846917871997, "grad_norm": 0.36754134558658713, "learning_rate": 2.297259076466407e-06, "loss": 2.568, "step": 44114 }, { "epoch": 2.738531255819728, "grad_norm": 0.35113173691413707, "learning_rate": 2.2961770722832952e-06, "loss": 2.7477, "step": 44115 }, { "epoch": 2.738593332919486, "grad_norm": 0.3784300180793361, "learning_rate": 2.2950953169826097e-06, "loss": 2.6378, "step": 44116 }, { "epoch": 2.738655410019244, "grad_norm": 0.35439654444577073, "learning_rate": 2.2940138105699836e-06, "loss": 2.659, "step": 44117 }, { "epoch": 2.738717487119002, "grad_norm": 0.3712157601002397, "learning_rate": 2.2929325530510747e-06, "loss": 2.6712, "step": 44118 }, { "epoch": 2.73877956421876, "grad_norm": 0.3763431099918926, "learning_rate": 2.291851544431511e-06, "loss": 2.6078, "step": 44119 }, { "epoch": 2.7388416413185177, "grad_norm": 0.36279549475877637, "learning_rate": 2.2907707847169503e-06, "loss": 2.6637, "step": 44120 }, { "epoch": 2.7389037184182756, "grad_norm": 0.3531255413996504, "learning_rate": 2.2896902739130143e-06, "loss": 2.7388, "step": 44121 }, { "epoch": 2.7389657955180335, "grad_norm": 0.3638032014356765, "learning_rate": 2.288610012025344e-06, "loss": 2.6756, "step": 44122 }, { "epoch": 2.7390278726177915, "grad_norm": 0.37956147712332255, "learning_rate": 2.2875299990595843e-06, "loss": 2.6715, "step": 44123 }, { "epoch": 2.739089949717549, "grad_norm": 0.3959400405440984, "learning_rate": 2.286450235021359e-06, "loss": 2.5956, "step": 44124 }, { "epoch": 2.7391520268173073, "grad_norm": 0.3586734267666335, "learning_rate": 2.2853707199163134e-06, "loss": 2.5643, "step": 44125 }, { "epoch": 2.7392141039170648, "grad_norm": 0.36113727787572547, "learning_rate": 2.2842914537500647e-06, "loss": 2.6624, "step": 44126 }, { "epoch": 2.739276181016823, "grad_norm": 0.3597553148148779, "learning_rate": 2.283212436528248e-06, "loss": 2.6495, "step": 44127 }, { "epoch": 2.7393382581165806, "grad_norm": 0.3559685850426867, "learning_rate": 2.2821336682564975e-06, "loss": 2.655, "step": 44128 }, { "epoch": 2.739400335216339, "grad_norm": 0.38206367823105364, "learning_rate": 2.281055148940442e-06, "loss": 2.6844, "step": 44129 }, { "epoch": 2.7394624123160964, "grad_norm": 0.3909308380669278, "learning_rate": 2.27997687858571e-06, "loss": 2.7492, "step": 44130 }, { "epoch": 2.7395244894158544, "grad_norm": 0.3636771065471926, "learning_rate": 2.2788988571979142e-06, "loss": 2.7143, "step": 44131 }, { "epoch": 2.7395865665156123, "grad_norm": 0.3753226005217816, "learning_rate": 2.277821084782689e-06, "loss": 2.7422, "step": 44132 }, { "epoch": 2.73964864361537, "grad_norm": 0.3686037953727288, "learning_rate": 2.2767435613456624e-06, "loss": 2.7505, "step": 44133 }, { "epoch": 2.739710720715128, "grad_norm": 0.38265978433418407, "learning_rate": 2.275666286892447e-06, "loss": 2.6405, "step": 44134 }, { "epoch": 2.739772797814886, "grad_norm": 0.37647739097575145, "learning_rate": 2.274589261428661e-06, "loss": 2.6344, "step": 44135 }, { "epoch": 2.739834874914644, "grad_norm": 0.40062412803424086, "learning_rate": 2.273512484959933e-06, "loss": 2.6633, "step": 44136 }, { "epoch": 2.739896952014402, "grad_norm": 0.3704676315245684, "learning_rate": 2.272435957491881e-06, "loss": 2.7378, "step": 44137 }, { "epoch": 2.73995902911416, "grad_norm": 0.3749794954416147, "learning_rate": 2.2713596790301107e-06, "loss": 2.5963, "step": 44138 }, { "epoch": 2.7400211062139177, "grad_norm": 0.3583160846277384, "learning_rate": 2.2702836495802514e-06, "loss": 2.6566, "step": 44139 }, { "epoch": 2.7400831833136756, "grad_norm": 0.3647989543722041, "learning_rate": 2.2692078691478935e-06, "loss": 2.6117, "step": 44140 }, { "epoch": 2.7401452604134335, "grad_norm": 0.39349576826134663, "learning_rate": 2.2681323377386765e-06, "loss": 2.6186, "step": 44141 }, { "epoch": 2.7402073375131915, "grad_norm": 0.35480283886844616, "learning_rate": 2.2670570553582014e-06, "loss": 2.6398, "step": 44142 }, { "epoch": 2.7402694146129494, "grad_norm": 0.37020460319392984, "learning_rate": 2.2659820220120755e-06, "loss": 2.6499, "step": 44143 }, { "epoch": 2.7403314917127073, "grad_norm": 0.37304777374958203, "learning_rate": 2.2649072377059046e-06, "loss": 2.7457, "step": 44144 }, { "epoch": 2.740393568812465, "grad_norm": 0.3747244259453169, "learning_rate": 2.2638327024453065e-06, "loss": 2.7114, "step": 44145 }, { "epoch": 2.740455645912223, "grad_norm": 0.36931026940439565, "learning_rate": 2.262758416235877e-06, "loss": 2.6497, "step": 44146 }, { "epoch": 2.740517723011981, "grad_norm": 0.3552224280431307, "learning_rate": 2.2616843790832287e-06, "loss": 2.6742, "step": 44147 }, { "epoch": 2.740579800111739, "grad_norm": 0.3668609343826685, "learning_rate": 2.2606105909929565e-06, "loss": 2.6333, "step": 44148 }, { "epoch": 2.7406418772114964, "grad_norm": 0.3524010195149873, "learning_rate": 2.259537051970678e-06, "loss": 2.6962, "step": 44149 }, { "epoch": 2.740703954311255, "grad_norm": 0.38212174745946786, "learning_rate": 2.2584637620219785e-06, "loss": 2.6398, "step": 44150 }, { "epoch": 2.7407660314110123, "grad_norm": 0.378107822437883, "learning_rate": 2.2573907211524637e-06, "loss": 2.7155, "step": 44151 }, { "epoch": 2.7408281085107706, "grad_norm": 0.3448145702722383, "learning_rate": 2.2563179293677237e-06, "loss": 2.5939, "step": 44152 }, { "epoch": 2.740890185610528, "grad_norm": 0.3665186695324185, "learning_rate": 2.255245386673377e-06, "loss": 2.7479, "step": 44153 }, { "epoch": 2.740952262710286, "grad_norm": 0.3522182706373221, "learning_rate": 2.2541730930749904e-06, "loss": 2.744, "step": 44154 }, { "epoch": 2.741014339810044, "grad_norm": 0.3730493515491666, "learning_rate": 2.2531010485781822e-06, "loss": 2.5988, "step": 44155 }, { "epoch": 2.741076416909802, "grad_norm": 0.36401336485698194, "learning_rate": 2.2520292531885422e-06, "loss": 2.6259, "step": 44156 }, { "epoch": 2.74113849400956, "grad_norm": 0.3754063034356074, "learning_rate": 2.2509577069116495e-06, "loss": 2.6353, "step": 44157 }, { "epoch": 2.7412005711093177, "grad_norm": 0.38611955171312073, "learning_rate": 2.249886409753099e-06, "loss": 2.677, "step": 44158 }, { "epoch": 2.7412626482090756, "grad_norm": 0.38814232564885043, "learning_rate": 2.2488153617184924e-06, "loss": 2.644, "step": 44159 }, { "epoch": 2.7413247253088335, "grad_norm": 0.38274011122140056, "learning_rate": 2.247744562813403e-06, "loss": 2.6225, "step": 44160 }, { "epoch": 2.7413868024085914, "grad_norm": 0.3482018700419358, "learning_rate": 2.246674013043426e-06, "loss": 2.6128, "step": 44161 }, { "epoch": 2.7414488795083494, "grad_norm": 0.36756741369290935, "learning_rate": 2.2456037124141403e-06, "loss": 2.636, "step": 44162 }, { "epoch": 2.7415109566081073, "grad_norm": 0.4036603514818892, "learning_rate": 2.244533660931125e-06, "loss": 2.6316, "step": 44163 }, { "epoch": 2.741573033707865, "grad_norm": 0.38754671261318036, "learning_rate": 2.243463858599981e-06, "loss": 2.648, "step": 44164 }, { "epoch": 2.741635110807623, "grad_norm": 0.3691120843097155, "learning_rate": 2.2423943054262764e-06, "loss": 2.6871, "step": 44165 }, { "epoch": 2.741697187907381, "grad_norm": 0.35503100595599324, "learning_rate": 2.241325001415595e-06, "loss": 2.722, "step": 44166 }, { "epoch": 2.741759265007139, "grad_norm": 0.35961677876802056, "learning_rate": 2.240255946573511e-06, "loss": 2.6451, "step": 44167 }, { "epoch": 2.741821342106897, "grad_norm": 0.3751104510090832, "learning_rate": 2.239187140905613e-06, "loss": 2.6059, "step": 44168 }, { "epoch": 2.741883419206655, "grad_norm": 0.3967737166257481, "learning_rate": 2.2381185844174647e-06, "loss": 2.6899, "step": 44169 }, { "epoch": 2.7419454963064127, "grad_norm": 0.4021869094476758, "learning_rate": 2.2370502771146494e-06, "loss": 2.6536, "step": 44170 }, { "epoch": 2.7420075734061706, "grad_norm": 0.38183769362258013, "learning_rate": 2.2359822190027302e-06, "loss": 2.6611, "step": 44171 }, { "epoch": 2.742069650505928, "grad_norm": 0.3593238089217195, "learning_rate": 2.2349144100872964e-06, "loss": 2.6109, "step": 44172 }, { "epoch": 2.7421317276056865, "grad_norm": 0.3673781447897381, "learning_rate": 2.2338468503739053e-06, "loss": 2.701, "step": 44173 }, { "epoch": 2.742193804705444, "grad_norm": 0.3945904300230101, "learning_rate": 2.232779539868135e-06, "loss": 2.6886, "step": 44174 }, { "epoch": 2.7422558818052023, "grad_norm": 0.3523733008229729, "learning_rate": 2.2317124785755427e-06, "loss": 2.574, "step": 44175 }, { "epoch": 2.7423179589049598, "grad_norm": 0.3508612845179245, "learning_rate": 2.2306456665017016e-06, "loss": 2.6179, "step": 44176 }, { "epoch": 2.742380036004718, "grad_norm": 0.3689701554601263, "learning_rate": 2.229579103652185e-06, "loss": 2.6041, "step": 44177 }, { "epoch": 2.7424421131044756, "grad_norm": 0.3812298464529377, "learning_rate": 2.2285127900325497e-06, "loss": 2.7195, "step": 44178 }, { "epoch": 2.7425041902042335, "grad_norm": 0.3726384988391674, "learning_rate": 2.2274467256483635e-06, "loss": 2.6557, "step": 44179 }, { "epoch": 2.7425662673039914, "grad_norm": 0.3656217183674971, "learning_rate": 2.2263809105051713e-06, "loss": 2.7147, "step": 44180 }, { "epoch": 2.7426283444037494, "grad_norm": 0.3623911505090774, "learning_rate": 2.2253153446085583e-06, "loss": 2.683, "step": 44181 }, { "epoch": 2.7426904215035073, "grad_norm": 0.3511683055454412, "learning_rate": 2.2242500279640752e-06, "loss": 2.7603, "step": 44182 }, { "epoch": 2.742752498603265, "grad_norm": 0.3697706426992922, "learning_rate": 2.2231849605772735e-06, "loss": 2.6639, "step": 44183 }, { "epoch": 2.742814575703023, "grad_norm": 0.39173346602525494, "learning_rate": 2.2221201424537097e-06, "loss": 2.691, "step": 44184 }, { "epoch": 2.742876652802781, "grad_norm": 0.41008269240774914, "learning_rate": 2.2210555735989458e-06, "loss": 2.641, "step": 44185 }, { "epoch": 2.742938729902539, "grad_norm": 0.3786221058200609, "learning_rate": 2.219991254018533e-06, "loss": 2.6714, "step": 44186 }, { "epoch": 2.743000807002297, "grad_norm": 0.3832395468230017, "learning_rate": 2.2189271837180337e-06, "loss": 2.7087, "step": 44187 }, { "epoch": 2.743062884102055, "grad_norm": 0.35982868333647566, "learning_rate": 2.217863362702982e-06, "loss": 2.6573, "step": 44188 }, { "epoch": 2.7431249612018127, "grad_norm": 0.35252416267503883, "learning_rate": 2.2167997909789405e-06, "loss": 2.584, "step": 44189 }, { "epoch": 2.7431870383015706, "grad_norm": 0.3609772086491172, "learning_rate": 2.2157364685514493e-06, "loss": 2.7107, "step": 44190 }, { "epoch": 2.7432491154013285, "grad_norm": 0.36512622149902085, "learning_rate": 2.2146733954260647e-06, "loss": 2.67, "step": 44191 }, { "epoch": 2.7433111925010865, "grad_norm": 0.38273203257092897, "learning_rate": 2.213610571608332e-06, "loss": 2.7228, "step": 44192 }, { "epoch": 2.7433732696008444, "grad_norm": 0.3614379097429777, "learning_rate": 2.212547997103792e-06, "loss": 2.6555, "step": 44193 }, { "epoch": 2.7434353467006023, "grad_norm": 0.36558680946237143, "learning_rate": 2.2114856719179844e-06, "loss": 2.6565, "step": 44194 }, { "epoch": 2.74349742380036, "grad_norm": 0.3622006989411762, "learning_rate": 2.21042359605646e-06, "loss": 2.6869, "step": 44195 }, { "epoch": 2.743559500900118, "grad_norm": 0.37047913500172486, "learning_rate": 2.209361769524765e-06, "loss": 2.677, "step": 44196 }, { "epoch": 2.7436215779998756, "grad_norm": 0.366971155104982, "learning_rate": 2.2083001923284217e-06, "loss": 2.7276, "step": 44197 }, { "epoch": 2.743683655099634, "grad_norm": 0.39677272139534264, "learning_rate": 2.207238864472977e-06, "loss": 2.7665, "step": 44198 }, { "epoch": 2.7437457321993914, "grad_norm": 0.4158525998188726, "learning_rate": 2.2061777859639755e-06, "loss": 2.6341, "step": 44199 }, { "epoch": 2.74380780929915, "grad_norm": 0.37615778135080036, "learning_rate": 2.2051169568069462e-06, "loss": 2.7077, "step": 44200 }, { "epoch": 2.7438698863989073, "grad_norm": 0.36424126472419255, "learning_rate": 2.204056377007424e-06, "loss": 2.7439, "step": 44201 }, { "epoch": 2.743931963498665, "grad_norm": 0.37162221122385625, "learning_rate": 2.202996046570943e-06, "loss": 2.7696, "step": 44202 }, { "epoch": 2.743994040598423, "grad_norm": 0.38112404259110727, "learning_rate": 2.201935965503027e-06, "loss": 2.6604, "step": 44203 }, { "epoch": 2.744056117698181, "grad_norm": 0.40486839920203294, "learning_rate": 2.200876133809221e-06, "loss": 2.732, "step": 44204 }, { "epoch": 2.744118194797939, "grad_norm": 0.3649612874573748, "learning_rate": 2.1998165514950485e-06, "loss": 2.6164, "step": 44205 }, { "epoch": 2.744180271897697, "grad_norm": 0.3635939616033433, "learning_rate": 2.1987572185660387e-06, "loss": 2.6977, "step": 44206 }, { "epoch": 2.744242348997455, "grad_norm": 0.38585508601756746, "learning_rate": 2.1976981350277093e-06, "loss": 2.6324, "step": 44207 }, { "epoch": 2.7443044260972127, "grad_norm": 0.3622144396475023, "learning_rate": 2.1966393008856e-06, "loss": 2.6074, "step": 44208 }, { "epoch": 2.7443665031969706, "grad_norm": 0.38688629066125696, "learning_rate": 2.195580716145229e-06, "loss": 2.598, "step": 44209 }, { "epoch": 2.7444285802967285, "grad_norm": 0.3792173347399582, "learning_rate": 2.1945223808121142e-06, "loss": 2.7439, "step": 44210 }, { "epoch": 2.7444906573964865, "grad_norm": 0.3774205408090498, "learning_rate": 2.193464294891784e-06, "loss": 2.7058, "step": 44211 }, { "epoch": 2.7445527344962444, "grad_norm": 0.38457211115752127, "learning_rate": 2.1924064583897507e-06, "loss": 2.6567, "step": 44212 }, { "epoch": 2.7446148115960023, "grad_norm": 0.33695721557242797, "learning_rate": 2.191348871311544e-06, "loss": 2.6815, "step": 44213 }, { "epoch": 2.74467688869576, "grad_norm": 0.34944843871575254, "learning_rate": 2.190291533662675e-06, "loss": 2.676, "step": 44214 }, { "epoch": 2.744738965795518, "grad_norm": 0.3644822480074288, "learning_rate": 2.1892344454486624e-06, "loss": 2.6249, "step": 44215 }, { "epoch": 2.744801042895276, "grad_norm": 0.3735228692321482, "learning_rate": 2.1881776066750128e-06, "loss": 2.6679, "step": 44216 }, { "epoch": 2.744863119995034, "grad_norm": 0.3728716573600552, "learning_rate": 2.187121017347249e-06, "loss": 2.7023, "step": 44217 }, { "epoch": 2.744925197094792, "grad_norm": 0.3862548704673573, "learning_rate": 2.1860646774708836e-06, "loss": 2.7219, "step": 44218 }, { "epoch": 2.74498727419455, "grad_norm": 0.36484057570062917, "learning_rate": 2.185008587051429e-06, "loss": 2.6687, "step": 44219 }, { "epoch": 2.7450493512943073, "grad_norm": 0.3373424736376916, "learning_rate": 2.1839527460943977e-06, "loss": 2.6959, "step": 44220 }, { "epoch": 2.7451114283940656, "grad_norm": 0.3720039048092491, "learning_rate": 2.182897154605279e-06, "loss": 2.6356, "step": 44221 }, { "epoch": 2.745173505493823, "grad_norm": 0.3512477313338353, "learning_rate": 2.1818418125896023e-06, "loss": 2.6918, "step": 44222 }, { "epoch": 2.7452355825935815, "grad_norm": 0.3964501352042214, "learning_rate": 2.180786720052863e-06, "loss": 2.685, "step": 44223 }, { "epoch": 2.745297659693339, "grad_norm": 0.35989802075101757, "learning_rate": 2.179731877000574e-06, "loss": 2.6469, "step": 44224 }, { "epoch": 2.745359736793097, "grad_norm": 0.36820433216799564, "learning_rate": 2.1786772834382297e-06, "loss": 2.6215, "step": 44225 }, { "epoch": 2.745421813892855, "grad_norm": 0.37776144851816434, "learning_rate": 2.1776229393713265e-06, "loss": 2.633, "step": 44226 }, { "epoch": 2.7454838909926127, "grad_norm": 0.36963090177977476, "learning_rate": 2.1765688448053823e-06, "loss": 2.5811, "step": 44227 }, { "epoch": 2.7455459680923706, "grad_norm": 0.35024369092333, "learning_rate": 2.1755149997458867e-06, "loss": 2.5871, "step": 44228 }, { "epoch": 2.7456080451921285, "grad_norm": 0.35488126226410943, "learning_rate": 2.1744614041983356e-06, "loss": 2.7089, "step": 44229 }, { "epoch": 2.7456701222918865, "grad_norm": 0.3708305630241099, "learning_rate": 2.1734080581682305e-06, "loss": 2.6867, "step": 44230 }, { "epoch": 2.7457321993916444, "grad_norm": 0.37740229987594476, "learning_rate": 2.172354961661066e-06, "loss": 2.7102, "step": 44231 }, { "epoch": 2.7457942764914023, "grad_norm": 0.39310124402825836, "learning_rate": 2.1713021146823385e-06, "loss": 2.7349, "step": 44232 }, { "epoch": 2.74585635359116, "grad_norm": 0.3744969570229821, "learning_rate": 2.1702495172375436e-06, "loss": 2.6172, "step": 44233 }, { "epoch": 2.745918430690918, "grad_norm": 0.3764695143750071, "learning_rate": 2.1691971693321544e-06, "loss": 2.6269, "step": 44234 }, { "epoch": 2.745980507790676, "grad_norm": 0.40345791369855644, "learning_rate": 2.1681450709716834e-06, "loss": 2.6977, "step": 44235 }, { "epoch": 2.746042584890434, "grad_norm": 0.3554485231450687, "learning_rate": 2.167093222161615e-06, "loss": 2.6776, "step": 44236 }, { "epoch": 2.746104661990192, "grad_norm": 0.36530747931483604, "learning_rate": 2.1660416229074278e-06, "loss": 2.6959, "step": 44237 }, { "epoch": 2.74616673908995, "grad_norm": 0.34734624263745917, "learning_rate": 2.1649902732146176e-06, "loss": 2.6779, "step": 44238 }, { "epoch": 2.7462288161897077, "grad_norm": 0.3761447876935335, "learning_rate": 2.163939173088653e-06, "loss": 2.6142, "step": 44239 }, { "epoch": 2.7462908932894656, "grad_norm": 0.3878320217521587, "learning_rate": 2.1628883225350447e-06, "loss": 2.5969, "step": 44240 }, { "epoch": 2.7463529703892235, "grad_norm": 0.37927454323836973, "learning_rate": 2.1618377215592566e-06, "loss": 2.6967, "step": 44241 }, { "epoch": 2.7464150474889815, "grad_norm": 0.36968985575347, "learning_rate": 2.1607873701667726e-06, "loss": 2.7265, "step": 44242 }, { "epoch": 2.746477124588739, "grad_norm": 0.3988142663115195, "learning_rate": 2.1597372683630713e-06, "loss": 2.703, "step": 44243 }, { "epoch": 2.7465392016884973, "grad_norm": 0.3501873353729324, "learning_rate": 2.1586874161536377e-06, "loss": 2.5854, "step": 44244 }, { "epoch": 2.7466012787882548, "grad_norm": 0.38599217680059517, "learning_rate": 2.1576378135439503e-06, "loss": 2.7201, "step": 44245 }, { "epoch": 2.746663355888013, "grad_norm": 0.3454637506917771, "learning_rate": 2.156588460539477e-06, "loss": 2.6879, "step": 44246 }, { "epoch": 2.7467254329877706, "grad_norm": 0.409039056687572, "learning_rate": 2.155539357145697e-06, "loss": 2.6298, "step": 44247 }, { "epoch": 2.746787510087529, "grad_norm": 0.3567071658311674, "learning_rate": 2.154490503368073e-06, "loss": 2.7047, "step": 44248 }, { "epoch": 2.7468495871872864, "grad_norm": 0.3545087910276391, "learning_rate": 2.1534418992120996e-06, "loss": 2.6842, "step": 44249 }, { "epoch": 2.7469116642870444, "grad_norm": 0.33194950443842197, "learning_rate": 2.1523935446832234e-06, "loss": 2.616, "step": 44250 }, { "epoch": 2.7469737413868023, "grad_norm": 0.3470210154680375, "learning_rate": 2.1513454397869337e-06, "loss": 2.6802, "step": 44251 }, { "epoch": 2.74703581848656, "grad_norm": 0.3823828134664367, "learning_rate": 2.150297584528693e-06, "loss": 2.7023, "step": 44252 }, { "epoch": 2.747097895586318, "grad_norm": 0.36870133030286456, "learning_rate": 2.1492499789139586e-06, "loss": 2.639, "step": 44253 }, { "epoch": 2.747159972686076, "grad_norm": 0.38097420941943, "learning_rate": 2.148202622948209e-06, "loss": 2.6151, "step": 44254 }, { "epoch": 2.747222049785834, "grad_norm": 0.36048183825823327, "learning_rate": 2.1471555166369005e-06, "loss": 2.7065, "step": 44255 }, { "epoch": 2.747284126885592, "grad_norm": 0.36314816241822956, "learning_rate": 2.1461086599854964e-06, "loss": 2.6897, "step": 44256 }, { "epoch": 2.74734620398535, "grad_norm": 0.4003960399726726, "learning_rate": 2.1450620529994582e-06, "loss": 2.7253, "step": 44257 }, { "epoch": 2.7474082810851077, "grad_norm": 0.34983799028787854, "learning_rate": 2.144015695684254e-06, "loss": 2.6028, "step": 44258 }, { "epoch": 2.7474703581848656, "grad_norm": 0.4058239977400593, "learning_rate": 2.1429695880453403e-06, "loss": 2.6257, "step": 44259 }, { "epoch": 2.7475324352846235, "grad_norm": 0.3481073132239053, "learning_rate": 2.141923730088169e-06, "loss": 2.6268, "step": 44260 }, { "epoch": 2.7475945123843815, "grad_norm": 0.36482685221519023, "learning_rate": 2.140878121818196e-06, "loss": 2.6253, "step": 44261 }, { "epoch": 2.7476565894841394, "grad_norm": 0.36485097008040723, "learning_rate": 2.1398327632408786e-06, "loss": 2.7484, "step": 44262 }, { "epoch": 2.7477186665838973, "grad_norm": 0.3765954015137053, "learning_rate": 2.1387876543616793e-06, "loss": 2.6741, "step": 44263 }, { "epoch": 2.747780743683655, "grad_norm": 0.3766152246056778, "learning_rate": 2.1377427951860375e-06, "loss": 2.6821, "step": 44264 }, { "epoch": 2.747842820783413, "grad_norm": 0.3883111516234379, "learning_rate": 2.136698185719416e-06, "loss": 2.7704, "step": 44265 }, { "epoch": 2.747904897883171, "grad_norm": 0.3694128415366986, "learning_rate": 2.1356538259672543e-06, "loss": 2.6962, "step": 44266 }, { "epoch": 2.747966974982929, "grad_norm": 0.34541988084092967, "learning_rate": 2.134609715935004e-06, "loss": 2.7134, "step": 44267 }, { "epoch": 2.7480290520826864, "grad_norm": 0.3572779923464596, "learning_rate": 2.1335658556281225e-06, "loss": 2.642, "step": 44268 }, { "epoch": 2.748091129182445, "grad_norm": 0.3329944887600157, "learning_rate": 2.1325222450520434e-06, "loss": 2.6936, "step": 44269 }, { "epoch": 2.7481532062822023, "grad_norm": 0.3464201694594683, "learning_rate": 2.1314788842122126e-06, "loss": 2.6502, "step": 44270 }, { "epoch": 2.7482152833819606, "grad_norm": 0.3862306230740052, "learning_rate": 2.130435773114081e-06, "loss": 2.6723, "step": 44271 }, { "epoch": 2.748277360481718, "grad_norm": 0.3684595273254598, "learning_rate": 2.1293929117630896e-06, "loss": 2.7687, "step": 44272 }, { "epoch": 2.748339437581476, "grad_norm": 0.3566949884401629, "learning_rate": 2.1283503001646776e-06, "loss": 2.7076, "step": 44273 }, { "epoch": 2.748401514681234, "grad_norm": 0.3631346827347017, "learning_rate": 2.12730793832428e-06, "loss": 2.6129, "step": 44274 }, { "epoch": 2.748463591780992, "grad_norm": 0.35942531316424764, "learning_rate": 2.1262658262473313e-06, "loss": 2.6543, "step": 44275 }, { "epoch": 2.74852566888075, "grad_norm": 0.37142116759002863, "learning_rate": 2.1252239639392822e-06, "loss": 2.7074, "step": 44276 }, { "epoch": 2.7485877459805077, "grad_norm": 0.3784642738806451, "learning_rate": 2.1241823514055627e-06, "loss": 2.7075, "step": 44277 }, { "epoch": 2.7486498230802656, "grad_norm": 0.3484990907982852, "learning_rate": 2.1231409886516064e-06, "loss": 2.6265, "step": 44278 }, { "epoch": 2.7487119001800235, "grad_norm": 0.35787984555741775, "learning_rate": 2.1220998756828424e-06, "loss": 2.6778, "step": 44279 }, { "epoch": 2.7487739772797815, "grad_norm": 0.34103840187100876, "learning_rate": 2.1210590125047057e-06, "loss": 2.6819, "step": 44280 }, { "epoch": 2.7488360543795394, "grad_norm": 0.36474823855340155, "learning_rate": 2.120018399122631e-06, "loss": 2.6252, "step": 44281 }, { "epoch": 2.7488981314792973, "grad_norm": 0.3722051941695744, "learning_rate": 2.1189780355420465e-06, "loss": 2.7029, "step": 44282 }, { "epoch": 2.748960208579055, "grad_norm": 0.3366652744960013, "learning_rate": 2.1179379217683647e-06, "loss": 2.6958, "step": 44283 }, { "epoch": 2.749022285678813, "grad_norm": 0.3825259790760901, "learning_rate": 2.1168980578070264e-06, "loss": 2.6417, "step": 44284 }, { "epoch": 2.749084362778571, "grad_norm": 0.3716964445280709, "learning_rate": 2.1158584436634653e-06, "loss": 2.6377, "step": 44285 }, { "epoch": 2.749146439878329, "grad_norm": 0.3752346893887431, "learning_rate": 2.1148190793430944e-06, "loss": 2.6292, "step": 44286 }, { "epoch": 2.749208516978087, "grad_norm": 0.3563333832048168, "learning_rate": 2.1137799648513314e-06, "loss": 2.7037, "step": 44287 }, { "epoch": 2.749270594077845, "grad_norm": 0.3709811190288883, "learning_rate": 2.1127411001936048e-06, "loss": 2.6992, "step": 44288 }, { "epoch": 2.7493326711776027, "grad_norm": 0.369924631078296, "learning_rate": 2.1117024853753276e-06, "loss": 2.6625, "step": 44289 }, { "epoch": 2.7493947482773606, "grad_norm": 0.3636052112551591, "learning_rate": 2.110664120401934e-06, "loss": 2.6629, "step": 44290 }, { "epoch": 2.749456825377118, "grad_norm": 0.36557649714568935, "learning_rate": 2.1096260052788253e-06, "loss": 2.705, "step": 44291 }, { "epoch": 2.7495189024768765, "grad_norm": 0.3657412148670194, "learning_rate": 2.1085881400114305e-06, "loss": 2.7064, "step": 44292 }, { "epoch": 2.749580979576634, "grad_norm": 0.3664137621924891, "learning_rate": 2.1075505246051454e-06, "loss": 2.6606, "step": 44293 }, { "epoch": 2.7496430566763923, "grad_norm": 0.3709447662040339, "learning_rate": 2.106513159065404e-06, "loss": 2.6173, "step": 44294 }, { "epoch": 2.74970513377615, "grad_norm": 0.34222603149860537, "learning_rate": 2.1054760433976083e-06, "loss": 2.5367, "step": 44295 }, { "epoch": 2.749767210875908, "grad_norm": 0.3600872486735614, "learning_rate": 2.1044391776071704e-06, "loss": 2.634, "step": 44296 }, { "epoch": 2.7498292879756656, "grad_norm": 0.3727872054607416, "learning_rate": 2.103402561699502e-06, "loss": 2.6088, "step": 44297 }, { "epoch": 2.7498913650754235, "grad_norm": 0.3825727607361885, "learning_rate": 2.1023661956800055e-06, "loss": 2.6787, "step": 44298 }, { "epoch": 2.7499534421751815, "grad_norm": 0.3569631081627598, "learning_rate": 2.1013300795540926e-06, "loss": 2.6258, "step": 44299 }, { "epoch": 2.7500155192749394, "grad_norm": 0.38498344645892735, "learning_rate": 2.100294213327175e-06, "loss": 2.651, "step": 44300 }, { "epoch": 2.7500775963746973, "grad_norm": 0.3749733359076054, "learning_rate": 2.0992585970046443e-06, "loss": 2.7282, "step": 44301 }, { "epoch": 2.750139673474455, "grad_norm": 0.40719348215084183, "learning_rate": 2.0982232305919056e-06, "loss": 2.6677, "step": 44302 }, { "epoch": 2.750201750574213, "grad_norm": 0.35395877945130727, "learning_rate": 2.097188114094373e-06, "loss": 2.6861, "step": 44303 }, { "epoch": 2.750263827673971, "grad_norm": 0.3472633391778162, "learning_rate": 2.0961532475174404e-06, "loss": 2.6857, "step": 44304 }, { "epoch": 2.750325904773729, "grad_norm": 0.37092660854196763, "learning_rate": 2.095118630866505e-06, "loss": 2.5604, "step": 44305 }, { "epoch": 2.750387981873487, "grad_norm": 0.3698948416253026, "learning_rate": 2.094084264146956e-06, "loss": 2.6963, "step": 44306 }, { "epoch": 2.750450058973245, "grad_norm": 0.36530417525022346, "learning_rate": 2.093050147364206e-06, "loss": 2.7033, "step": 44307 }, { "epoch": 2.7505121360730027, "grad_norm": 0.38336167550353895, "learning_rate": 2.09201628052364e-06, "loss": 2.7951, "step": 44308 }, { "epoch": 2.7505742131727606, "grad_norm": 0.3388920425841979, "learning_rate": 2.0909826636306582e-06, "loss": 2.6375, "step": 44309 }, { "epoch": 2.7506362902725185, "grad_norm": 0.370246944982841, "learning_rate": 2.089949296690652e-06, "loss": 2.6618, "step": 44310 }, { "epoch": 2.7506983673722765, "grad_norm": 0.3683371301530109, "learning_rate": 2.088916179709005e-06, "loss": 2.7516, "step": 44311 }, { "epoch": 2.7507604444720344, "grad_norm": 0.37777880433977806, "learning_rate": 2.0878833126911136e-06, "loss": 2.5926, "step": 44312 }, { "epoch": 2.7508225215717923, "grad_norm": 0.3636629294044032, "learning_rate": 2.0868506956423727e-06, "loss": 2.7071, "step": 44313 }, { "epoch": 2.75088459867155, "grad_norm": 0.3639531433828673, "learning_rate": 2.085818328568162e-06, "loss": 2.6784, "step": 44314 }, { "epoch": 2.750946675771308, "grad_norm": 0.3607809511681563, "learning_rate": 2.0847862114738603e-06, "loss": 2.7499, "step": 44315 }, { "epoch": 2.7510087528710656, "grad_norm": 0.37121170578557666, "learning_rate": 2.0837543443648576e-06, "loss": 2.7324, "step": 44316 }, { "epoch": 2.751070829970824, "grad_norm": 0.3825341205088774, "learning_rate": 2.08272272724655e-06, "loss": 2.7065, "step": 44317 }, { "epoch": 2.7511329070705814, "grad_norm": 0.3830574380879705, "learning_rate": 2.0816913601243105e-06, "loss": 2.6502, "step": 44318 }, { "epoch": 2.75119498417034, "grad_norm": 0.36221773075726854, "learning_rate": 2.0806602430035183e-06, "loss": 2.5958, "step": 44319 }, { "epoch": 2.7512570612700973, "grad_norm": 0.3760764645707782, "learning_rate": 2.0796293758895526e-06, "loss": 2.6529, "step": 44320 }, { "epoch": 2.751319138369855, "grad_norm": 0.3489019279536868, "learning_rate": 2.0785987587877976e-06, "loss": 2.6891, "step": 44321 }, { "epoch": 2.751381215469613, "grad_norm": 0.35063779867071876, "learning_rate": 2.0775683917036214e-06, "loss": 2.6254, "step": 44322 }, { "epoch": 2.751443292569371, "grad_norm": 0.3687664941557899, "learning_rate": 2.076538274642409e-06, "loss": 2.6401, "step": 44323 }, { "epoch": 2.751505369669129, "grad_norm": 0.37052977018791944, "learning_rate": 2.0755084076095334e-06, "loss": 2.6541, "step": 44324 }, { "epoch": 2.751567446768887, "grad_norm": 0.3646397365355408, "learning_rate": 2.0744787906103513e-06, "loss": 2.6508, "step": 44325 }, { "epoch": 2.751629523868645, "grad_norm": 0.3476807767810134, "learning_rate": 2.073449423650259e-06, "loss": 2.5943, "step": 44326 }, { "epoch": 2.7516916009684027, "grad_norm": 0.3702451942538931, "learning_rate": 2.0724203067346126e-06, "loss": 2.7032, "step": 44327 }, { "epoch": 2.7517536780681606, "grad_norm": 0.36146363475257476, "learning_rate": 2.071391439868786e-06, "loss": 2.6562, "step": 44328 }, { "epoch": 2.7518157551679185, "grad_norm": 0.3796648317014434, "learning_rate": 2.070362823058136e-06, "loss": 2.6909, "step": 44329 }, { "epoch": 2.7518778322676765, "grad_norm": 0.3536724985975427, "learning_rate": 2.0693344563080475e-06, "loss": 2.6372, "step": 44330 }, { "epoch": 2.7519399093674344, "grad_norm": 0.3789847774342272, "learning_rate": 2.0683063396238766e-06, "loss": 2.7043, "step": 44331 }, { "epoch": 2.7520019864671923, "grad_norm": 0.35537915620308963, "learning_rate": 2.0672784730109918e-06, "loss": 2.6824, "step": 44332 }, { "epoch": 2.75206406356695, "grad_norm": 0.36144085749320537, "learning_rate": 2.0662508564747442e-06, "loss": 2.614, "step": 44333 }, { "epoch": 2.752126140666708, "grad_norm": 0.38006763816237155, "learning_rate": 2.0652234900205016e-06, "loss": 2.6604, "step": 44334 }, { "epoch": 2.752188217766466, "grad_norm": 0.38038059364766535, "learning_rate": 2.0641963736536265e-06, "loss": 2.6261, "step": 44335 }, { "epoch": 2.752250294866224, "grad_norm": 0.35714596380858044, "learning_rate": 2.0631695073794754e-06, "loss": 2.6795, "step": 44336 }, { "epoch": 2.752312371965982, "grad_norm": 0.3634171385567749, "learning_rate": 2.0621428912034056e-06, "loss": 2.653, "step": 44337 }, { "epoch": 2.75237444906574, "grad_norm": 0.3670859703758742, "learning_rate": 2.061116525130774e-06, "loss": 2.7234, "step": 44338 }, { "epoch": 2.7524365261654973, "grad_norm": 0.4013945113114248, "learning_rate": 2.060090409166937e-06, "loss": 2.6726, "step": 44339 }, { "epoch": 2.7524986032652556, "grad_norm": 0.3752029783547033, "learning_rate": 2.0590645433172407e-06, "loss": 2.7105, "step": 44340 }, { "epoch": 2.752560680365013, "grad_norm": 0.3989493504335475, "learning_rate": 2.0580389275870526e-06, "loss": 2.6496, "step": 44341 }, { "epoch": 2.7526227574647715, "grad_norm": 0.3849681960011412, "learning_rate": 2.057013561981702e-06, "loss": 2.635, "step": 44342 }, { "epoch": 2.752684834564529, "grad_norm": 0.382243804464784, "learning_rate": 2.055988446506557e-06, "loss": 2.6258, "step": 44343 }, { "epoch": 2.7527469116642873, "grad_norm": 0.37346781204612733, "learning_rate": 2.0549635811669575e-06, "loss": 2.6787, "step": 44344 }, { "epoch": 2.752808988764045, "grad_norm": 0.3661043139051671, "learning_rate": 2.053938965968255e-06, "loss": 2.5969, "step": 44345 }, { "epoch": 2.7528710658638027, "grad_norm": 0.3848312244552999, "learning_rate": 2.0529146009157895e-06, "loss": 2.6757, "step": 44346 }, { "epoch": 2.7529331429635606, "grad_norm": 0.3792190244003328, "learning_rate": 2.0518904860149067e-06, "loss": 2.647, "step": 44347 }, { "epoch": 2.7529952200633185, "grad_norm": 0.35162370996614895, "learning_rate": 2.0508666212709473e-06, "loss": 2.6206, "step": 44348 }, { "epoch": 2.7530572971630765, "grad_norm": 0.38336146172001634, "learning_rate": 2.0498430066892614e-06, "loss": 2.6426, "step": 44349 }, { "epoch": 2.7531193742628344, "grad_norm": 0.35401868065882747, "learning_rate": 2.0488196422751905e-06, "loss": 2.6893, "step": 44350 }, { "epoch": 2.7531814513625923, "grad_norm": 0.33952889162061717, "learning_rate": 2.047796528034063e-06, "loss": 2.6172, "step": 44351 }, { "epoch": 2.75324352846235, "grad_norm": 0.367424786779411, "learning_rate": 2.046773663971219e-06, "loss": 2.688, "step": 44352 }, { "epoch": 2.753305605562108, "grad_norm": 0.3691547128723998, "learning_rate": 2.0457510500919984e-06, "loss": 2.6144, "step": 44353 }, { "epoch": 2.753367682661866, "grad_norm": 0.3771323533977081, "learning_rate": 2.0447286864017424e-06, "loss": 2.6937, "step": 44354 }, { "epoch": 2.753429759761624, "grad_norm": 0.3846149827423197, "learning_rate": 2.0437065729057736e-06, "loss": 2.6778, "step": 44355 }, { "epoch": 2.753491836861382, "grad_norm": 0.35988646889339104, "learning_rate": 2.042684709609427e-06, "loss": 2.6895, "step": 44356 }, { "epoch": 2.75355391396114, "grad_norm": 0.3666892628137281, "learning_rate": 2.0416630965180426e-06, "loss": 2.6287, "step": 44357 }, { "epoch": 2.7536159910608977, "grad_norm": 0.3623370544288726, "learning_rate": 2.0406417336369388e-06, "loss": 2.6205, "step": 44358 }, { "epoch": 2.7536780681606556, "grad_norm": 0.3579250608413921, "learning_rate": 2.039620620971455e-06, "loss": 2.5321, "step": 44359 }, { "epoch": 2.7537401452604136, "grad_norm": 0.37493041505524616, "learning_rate": 2.03859975852691e-06, "loss": 2.6498, "step": 44360 }, { "epoch": 2.7538022223601715, "grad_norm": 0.35259997615643734, "learning_rate": 2.0375791463086268e-06, "loss": 2.6459, "step": 44361 }, { "epoch": 2.7538642994599294, "grad_norm": 0.3511494435241996, "learning_rate": 2.036558784321946e-06, "loss": 2.6755, "step": 44362 }, { "epoch": 2.7539263765596873, "grad_norm": 0.35894341053481116, "learning_rate": 2.035538672572179e-06, "loss": 2.6299, "step": 44363 }, { "epoch": 2.753988453659445, "grad_norm": 0.3681455248306712, "learning_rate": 2.0345188110646505e-06, "loss": 2.7053, "step": 44364 }, { "epoch": 2.754050530759203, "grad_norm": 0.39305842610663805, "learning_rate": 2.033499199804678e-06, "loss": 2.6793, "step": 44365 }, { "epoch": 2.7541126078589606, "grad_norm": 0.3544779793069918, "learning_rate": 2.0324798387975907e-06, "loss": 2.5774, "step": 44366 }, { "epoch": 2.754174684958719, "grad_norm": 0.3464069636935056, "learning_rate": 2.0314607280486953e-06, "loss": 2.6775, "step": 44367 }, { "epoch": 2.7542367620584765, "grad_norm": 0.3710198178177909, "learning_rate": 2.030441867563321e-06, "loss": 2.5982, "step": 44368 }, { "epoch": 2.7542988391582344, "grad_norm": 0.35379731562783734, "learning_rate": 2.029423257346763e-06, "loss": 2.6229, "step": 44369 }, { "epoch": 2.7543609162579923, "grad_norm": 0.36545949515652143, "learning_rate": 2.0284048974043623e-06, "loss": 2.5727, "step": 44370 }, { "epoch": 2.75442299335775, "grad_norm": 0.35950324631261166, "learning_rate": 2.027386787741414e-06, "loss": 2.7042, "step": 44371 }, { "epoch": 2.754485070457508, "grad_norm": 0.3697567193437635, "learning_rate": 2.026368928363237e-06, "loss": 2.6963, "step": 44372 }, { "epoch": 2.754547147557266, "grad_norm": 0.3534647551971756, "learning_rate": 2.0253513192751373e-06, "loss": 2.6658, "step": 44373 }, { "epoch": 2.754609224657024, "grad_norm": 0.3689853805990285, "learning_rate": 2.024333960482422e-06, "loss": 2.7283, "step": 44374 }, { "epoch": 2.754671301756782, "grad_norm": 0.34015378293153387, "learning_rate": 2.0233168519904034e-06, "loss": 2.5805, "step": 44375 }, { "epoch": 2.75473337885654, "grad_norm": 0.3663661576970492, "learning_rate": 2.022299993804394e-06, "loss": 2.6186, "step": 44376 }, { "epoch": 2.7547954559562977, "grad_norm": 0.36145707999606697, "learning_rate": 2.0212833859296896e-06, "loss": 2.6222, "step": 44377 }, { "epoch": 2.7548575330560556, "grad_norm": 0.381807467674903, "learning_rate": 2.020267028371592e-06, "loss": 2.6583, "step": 44378 }, { "epoch": 2.7549196101558135, "grad_norm": 0.3517634241481508, "learning_rate": 2.0192509211354127e-06, "loss": 2.6184, "step": 44379 }, { "epoch": 2.7549816872555715, "grad_norm": 0.39922604414216706, "learning_rate": 2.0182350642264536e-06, "loss": 2.7334, "step": 44380 }, { "epoch": 2.7550437643553294, "grad_norm": 0.3497200582890722, "learning_rate": 2.017219457649999e-06, "loss": 2.651, "step": 44381 }, { "epoch": 2.7551058414550873, "grad_norm": 0.37938331773994166, "learning_rate": 2.0162041014113677e-06, "loss": 2.6529, "step": 44382 }, { "epoch": 2.755167918554845, "grad_norm": 0.3447357371030645, "learning_rate": 2.0151889955158488e-06, "loss": 2.6385, "step": 44383 }, { "epoch": 2.755229995654603, "grad_norm": 0.37232287788239715, "learning_rate": 2.014174139968733e-06, "loss": 2.623, "step": 44384 }, { "epoch": 2.755292072754361, "grad_norm": 0.37823745852225316, "learning_rate": 2.013159534775322e-06, "loss": 2.5721, "step": 44385 }, { "epoch": 2.755354149854119, "grad_norm": 0.3674729840410767, "learning_rate": 2.0121451799409107e-06, "loss": 2.5917, "step": 44386 }, { "epoch": 2.7554162269538764, "grad_norm": 0.3827866526071469, "learning_rate": 2.0111310754707846e-06, "loss": 2.7004, "step": 44387 }, { "epoch": 2.755478304053635, "grad_norm": 0.37006738171597425, "learning_rate": 2.010117221370228e-06, "loss": 2.7072, "step": 44388 }, { "epoch": 2.7555403811533923, "grad_norm": 0.3457317986962351, "learning_rate": 2.0091036176445534e-06, "loss": 2.6253, "step": 44389 }, { "epoch": 2.7556024582531506, "grad_norm": 0.35874381820988294, "learning_rate": 2.0080902642990286e-06, "loss": 2.5859, "step": 44390 }, { "epoch": 2.755664535352908, "grad_norm": 0.36242558603428165, "learning_rate": 2.0070771613389494e-06, "loss": 2.5879, "step": 44391 }, { "epoch": 2.7557266124526665, "grad_norm": 0.3643081875655632, "learning_rate": 2.0060643087695954e-06, "loss": 2.5704, "step": 44392 }, { "epoch": 2.755788689552424, "grad_norm": 0.38337252000218647, "learning_rate": 2.0050517065962617e-06, "loss": 2.7043, "step": 44393 }, { "epoch": 2.755850766652182, "grad_norm": 0.37933749797548394, "learning_rate": 2.004039354824222e-06, "loss": 2.6727, "step": 44394 }, { "epoch": 2.75591284375194, "grad_norm": 0.37808754554183505, "learning_rate": 2.0030272534587613e-06, "loss": 2.7039, "step": 44395 }, { "epoch": 2.7559749208516977, "grad_norm": 0.36224337981936094, "learning_rate": 2.0020154025051584e-06, "loss": 2.7763, "step": 44396 }, { "epoch": 2.7560369979514556, "grad_norm": 0.39170524004785523, "learning_rate": 2.0010038019686873e-06, "loss": 2.7211, "step": 44397 }, { "epoch": 2.7560990750512135, "grad_norm": 0.3784949439730258, "learning_rate": 1.9999924518546378e-06, "loss": 2.7623, "step": 44398 }, { "epoch": 2.7561611521509715, "grad_norm": 0.3782834706168405, "learning_rate": 1.9989813521682774e-06, "loss": 2.5388, "step": 44399 }, { "epoch": 2.7562232292507294, "grad_norm": 0.36840126632722336, "learning_rate": 1.9979705029148863e-06, "loss": 2.5817, "step": 44400 }, { "epoch": 2.7562853063504873, "grad_norm": 0.3764113253341931, "learning_rate": 1.996959904099732e-06, "loss": 2.6769, "step": 44401 }, { "epoch": 2.756347383450245, "grad_norm": 0.34914067353516925, "learning_rate": 1.995949555728094e-06, "loss": 2.7353, "step": 44402 }, { "epoch": 2.756409460550003, "grad_norm": 0.3414748282266441, "learning_rate": 1.9949394578052395e-06, "loss": 2.7354, "step": 44403 }, { "epoch": 2.756471537649761, "grad_norm": 0.3883213974212573, "learning_rate": 1.9939296103364425e-06, "loss": 2.6893, "step": 44404 }, { "epoch": 2.756533614749519, "grad_norm": 0.3539808594666547, "learning_rate": 1.99292001332696e-06, "loss": 2.6666, "step": 44405 }, { "epoch": 2.756595691849277, "grad_norm": 0.3905991354962542, "learning_rate": 1.9919106667820764e-06, "loss": 2.7059, "step": 44406 }, { "epoch": 2.756657768949035, "grad_norm": 0.3434716202861785, "learning_rate": 1.9909015707070487e-06, "loss": 2.7392, "step": 44407 }, { "epoch": 2.7567198460487927, "grad_norm": 0.37550619849030076, "learning_rate": 1.98989272510714e-06, "loss": 2.6131, "step": 44408 }, { "epoch": 2.7567819231485506, "grad_norm": 0.3706998240059619, "learning_rate": 1.988884129987617e-06, "loss": 2.6477, "step": 44409 }, { "epoch": 2.7568440002483086, "grad_norm": 0.368496024229375, "learning_rate": 1.9878757853537377e-06, "loss": 2.6371, "step": 44410 }, { "epoch": 2.7569060773480665, "grad_norm": 0.36826517350815235, "learning_rate": 1.98686769121077e-06, "loss": 2.6939, "step": 44411 }, { "epoch": 2.756968154447824, "grad_norm": 0.3661064711024099, "learning_rate": 1.98585984756397e-06, "loss": 2.6655, "step": 44412 }, { "epoch": 2.7570302315475823, "grad_norm": 0.35691528690842456, "learning_rate": 1.9848522544185898e-06, "loss": 2.6602, "step": 44413 }, { "epoch": 2.75709230864734, "grad_norm": 0.36144091933386174, "learning_rate": 1.983844911779892e-06, "loss": 2.6749, "step": 44414 }, { "epoch": 2.757154385747098, "grad_norm": 0.3682487653438384, "learning_rate": 1.982837819653127e-06, "loss": 2.575, "step": 44415 }, { "epoch": 2.7572164628468556, "grad_norm": 0.36874562277054757, "learning_rate": 1.9818309780435584e-06, "loss": 2.6532, "step": 44416 }, { "epoch": 2.7572785399466135, "grad_norm": 0.35390175233387167, "learning_rate": 1.9808243869564424e-06, "loss": 2.6255, "step": 44417 }, { "epoch": 2.7573406170463715, "grad_norm": 0.3472369054497753, "learning_rate": 1.9798180463970138e-06, "loss": 2.658, "step": 44418 }, { "epoch": 2.7574026941461294, "grad_norm": 0.36085846766415397, "learning_rate": 1.9788119563705297e-06, "loss": 2.644, "step": 44419 }, { "epoch": 2.7574647712458873, "grad_norm": 0.37497645542618446, "learning_rate": 1.977806116882247e-06, "loss": 2.6688, "step": 44420 }, { "epoch": 2.757526848345645, "grad_norm": 0.35845327347601497, "learning_rate": 1.976800527937411e-06, "loss": 2.712, "step": 44421 }, { "epoch": 2.757588925445403, "grad_norm": 0.3650907958444303, "learning_rate": 1.9757951895412575e-06, "loss": 2.6603, "step": 44422 }, { "epoch": 2.757651002545161, "grad_norm": 0.3628918601549889, "learning_rate": 1.974790101699042e-06, "loss": 2.7302, "step": 44423 }, { "epoch": 2.757713079644919, "grad_norm": 0.35525342686749933, "learning_rate": 1.9737852644160003e-06, "loss": 2.7244, "step": 44424 }, { "epoch": 2.757775156744677, "grad_norm": 0.38548770670826976, "learning_rate": 1.972780677697389e-06, "loss": 2.6732, "step": 44425 }, { "epoch": 2.757837233844435, "grad_norm": 0.3620147779040642, "learning_rate": 1.9717763415484313e-06, "loss": 2.6822, "step": 44426 }, { "epoch": 2.7578993109441927, "grad_norm": 0.37322596300634453, "learning_rate": 1.9707722559743844e-06, "loss": 2.5865, "step": 44427 }, { "epoch": 2.7579613880439506, "grad_norm": 0.3559052857514795, "learning_rate": 1.969768420980467e-06, "loss": 2.6331, "step": 44428 }, { "epoch": 2.7580234651437086, "grad_norm": 0.37336391937114805, "learning_rate": 1.9687648365719345e-06, "loss": 2.783, "step": 44429 }, { "epoch": 2.7580855422434665, "grad_norm": 0.3592637760703912, "learning_rate": 1.9677615027540176e-06, "loss": 2.7645, "step": 44430 }, { "epoch": 2.7581476193432244, "grad_norm": 0.34891414243908564, "learning_rate": 1.9667584195319443e-06, "loss": 2.6429, "step": 44431 }, { "epoch": 2.7582096964429823, "grad_norm": 0.343040709886468, "learning_rate": 1.9657555869109555e-06, "loss": 2.689, "step": 44432 }, { "epoch": 2.7582717735427402, "grad_norm": 0.3592662439396975, "learning_rate": 1.964753004896275e-06, "loss": 2.6419, "step": 44433 }, { "epoch": 2.758333850642498, "grad_norm": 0.36631437748100965, "learning_rate": 1.9637506734931475e-06, "loss": 2.6213, "step": 44434 }, { "epoch": 2.7583959277422556, "grad_norm": 0.3827877036646933, "learning_rate": 1.962748592706792e-06, "loss": 2.7378, "step": 44435 }, { "epoch": 2.758458004842014, "grad_norm": 0.3691850665754771, "learning_rate": 1.961746762542438e-06, "loss": 2.6743, "step": 44436 }, { "epoch": 2.7585200819417715, "grad_norm": 0.3623093967183104, "learning_rate": 1.9607451830053027e-06, "loss": 2.7109, "step": 44437 }, { "epoch": 2.75858215904153, "grad_norm": 0.3612860990261078, "learning_rate": 1.959743854100632e-06, "loss": 2.6458, "step": 44438 }, { "epoch": 2.7586442361412873, "grad_norm": 0.36073582534673626, "learning_rate": 1.958742775833644e-06, "loss": 2.7035, "step": 44439 }, { "epoch": 2.7587063132410456, "grad_norm": 0.3646136021537618, "learning_rate": 1.9577419482095517e-06, "loss": 2.7292, "step": 44440 }, { "epoch": 2.758768390340803, "grad_norm": 0.3699273097959276, "learning_rate": 1.956741371233578e-06, "loss": 2.6749, "step": 44441 }, { "epoch": 2.758830467440561, "grad_norm": 0.3486269675226999, "learning_rate": 1.9557410449109526e-06, "loss": 2.6341, "step": 44442 }, { "epoch": 2.758892544540319, "grad_norm": 0.36137808377043057, "learning_rate": 1.9547409692468876e-06, "loss": 2.6364, "step": 44443 }, { "epoch": 2.758954621640077, "grad_norm": 0.34133955742743327, "learning_rate": 1.953741144246607e-06, "loss": 2.5638, "step": 44444 }, { "epoch": 2.759016698739835, "grad_norm": 0.3838895321638352, "learning_rate": 1.9527415699153172e-06, "loss": 2.679, "step": 44445 }, { "epoch": 2.7590787758395927, "grad_norm": 0.36423158934753946, "learning_rate": 1.951742246258237e-06, "loss": 2.6762, "step": 44446 }, { "epoch": 2.7591408529393506, "grad_norm": 0.3617436674363884, "learning_rate": 1.950743173280578e-06, "loss": 2.6815, "step": 44447 }, { "epoch": 2.7592029300391085, "grad_norm": 0.3572514445553959, "learning_rate": 1.9497443509875645e-06, "loss": 2.8107, "step": 44448 }, { "epoch": 2.7592650071388665, "grad_norm": 0.35269593740169825, "learning_rate": 1.9487457793844033e-06, "loss": 2.6739, "step": 44449 }, { "epoch": 2.7593270842386244, "grad_norm": 0.35495666675972704, "learning_rate": 1.9477474584762955e-06, "loss": 2.6574, "step": 44450 }, { "epoch": 2.7593891613383823, "grad_norm": 0.3469094925132601, "learning_rate": 1.9467493882684484e-06, "loss": 2.7399, "step": 44451 }, { "epoch": 2.75945123843814, "grad_norm": 0.3612888417927425, "learning_rate": 1.94575156876608e-06, "loss": 2.7314, "step": 44452 }, { "epoch": 2.759513315537898, "grad_norm": 0.369957680666225, "learning_rate": 1.9447539999743912e-06, "loss": 2.6817, "step": 44453 }, { "epoch": 2.759575392637656, "grad_norm": 0.3480832873990448, "learning_rate": 1.9437566818985895e-06, "loss": 2.6525, "step": 44454 }, { "epoch": 2.759637469737414, "grad_norm": 0.33278420287710087, "learning_rate": 1.9427596145438656e-06, "loss": 2.6645, "step": 44455 }, { "epoch": 2.759699546837172, "grad_norm": 0.3596166958955641, "learning_rate": 1.9417627979154417e-06, "loss": 2.586, "step": 44456 }, { "epoch": 2.75976162393693, "grad_norm": 0.3636575411257107, "learning_rate": 1.940766232018504e-06, "loss": 2.6983, "step": 44457 }, { "epoch": 2.7598237010366877, "grad_norm": 0.3740105091380476, "learning_rate": 1.9397699168582584e-06, "loss": 2.737, "step": 44458 }, { "epoch": 2.7598857781364456, "grad_norm": 0.34183376658119397, "learning_rate": 1.9387738524398955e-06, "loss": 2.6231, "step": 44459 }, { "epoch": 2.759947855236203, "grad_norm": 0.36372516540748023, "learning_rate": 1.9377780387686172e-06, "loss": 2.5985, "step": 44460 }, { "epoch": 2.7600099323359615, "grad_norm": 0.3498474993724192, "learning_rate": 1.936782475849619e-06, "loss": 2.6292, "step": 44461 }, { "epoch": 2.760072009435719, "grad_norm": 0.3670971333114082, "learning_rate": 1.9357871636880964e-06, "loss": 2.6825, "step": 44462 }, { "epoch": 2.7601340865354773, "grad_norm": 0.34733632186586605, "learning_rate": 1.93479210228924e-06, "loss": 2.631, "step": 44463 }, { "epoch": 2.760196163635235, "grad_norm": 0.35875644525194444, "learning_rate": 1.9337972916582404e-06, "loss": 2.6796, "step": 44464 }, { "epoch": 2.7602582407349927, "grad_norm": 0.36305775484283037, "learning_rate": 1.932802731800287e-06, "loss": 2.644, "step": 44465 }, { "epoch": 2.7603203178347506, "grad_norm": 0.37844314983633, "learning_rate": 1.931808422720577e-06, "loss": 2.5997, "step": 44466 }, { "epoch": 2.7603823949345085, "grad_norm": 0.35826116526388907, "learning_rate": 1.9308143644242883e-06, "loss": 2.6731, "step": 44467 }, { "epoch": 2.7604444720342665, "grad_norm": 0.3341447657002368, "learning_rate": 1.9298205569166063e-06, "loss": 2.6179, "step": 44468 }, { "epoch": 2.7605065491340244, "grad_norm": 0.34939035555920117, "learning_rate": 1.928827000202721e-06, "loss": 2.6207, "step": 44469 }, { "epoch": 2.7605686262337823, "grad_norm": 0.34004268085585737, "learning_rate": 1.927833694287817e-06, "loss": 2.6784, "step": 44470 }, { "epoch": 2.76063070333354, "grad_norm": 0.34814387852418976, "learning_rate": 1.9268406391770744e-06, "loss": 2.6548, "step": 44471 }, { "epoch": 2.760692780433298, "grad_norm": 0.351269069710567, "learning_rate": 1.925847834875677e-06, "loss": 2.6613, "step": 44472 }, { "epoch": 2.760754857533056, "grad_norm": 0.36930141277523326, "learning_rate": 1.924855281388793e-06, "loss": 2.7348, "step": 44473 }, { "epoch": 2.760816934632814, "grad_norm": 0.355645502602286, "learning_rate": 1.9238629787216133e-06, "loss": 2.7368, "step": 44474 }, { "epoch": 2.760879011732572, "grad_norm": 0.3452623434449876, "learning_rate": 1.9228709268793164e-06, "loss": 2.6811, "step": 44475 }, { "epoch": 2.76094108883233, "grad_norm": 0.3657917319078768, "learning_rate": 1.921879125867071e-06, "loss": 2.6522, "step": 44476 }, { "epoch": 2.7610031659320877, "grad_norm": 0.36761763924808166, "learning_rate": 1.9208875756900502e-06, "loss": 2.6923, "step": 44477 }, { "epoch": 2.7610652430318456, "grad_norm": 0.33782392736580963, "learning_rate": 1.9198962763534335e-06, "loss": 2.6468, "step": 44478 }, { "epoch": 2.7611273201316036, "grad_norm": 0.38156735078944654, "learning_rate": 1.9189052278623833e-06, "loss": 2.5568, "step": 44479 }, { "epoch": 2.7611893972313615, "grad_norm": 0.3410759543662143, "learning_rate": 1.9179144302220843e-06, "loss": 2.5696, "step": 44480 }, { "epoch": 2.7612514743311194, "grad_norm": 0.3683445536118516, "learning_rate": 1.9169238834376934e-06, "loss": 2.5933, "step": 44481 }, { "epoch": 2.7613135514308773, "grad_norm": 0.34826830232768535, "learning_rate": 1.91593358751439e-06, "loss": 2.633, "step": 44482 }, { "epoch": 2.761375628530635, "grad_norm": 0.3637757985644043, "learning_rate": 1.914943542457326e-06, "loss": 2.6059, "step": 44483 }, { "epoch": 2.761437705630393, "grad_norm": 0.3671703334610246, "learning_rate": 1.9139537482716796e-06, "loss": 2.7099, "step": 44484 }, { "epoch": 2.7614997827301506, "grad_norm": 0.3801948814973629, "learning_rate": 1.9129642049626084e-06, "loss": 2.6098, "step": 44485 }, { "epoch": 2.761561859829909, "grad_norm": 0.3788090414038204, "learning_rate": 1.9119749125352805e-06, "loss": 2.6457, "step": 44486 }, { "epoch": 2.7616239369296665, "grad_norm": 0.37222611176576287, "learning_rate": 1.9109858709948412e-06, "loss": 2.656, "step": 44487 }, { "epoch": 2.761686014029425, "grad_norm": 0.3389889940666736, "learning_rate": 1.909997080346476e-06, "loss": 2.6117, "step": 44488 }, { "epoch": 2.7617480911291823, "grad_norm": 0.3598884149312199, "learning_rate": 1.909008540595325e-06, "loss": 2.743, "step": 44489 }, { "epoch": 2.76181016822894, "grad_norm": 0.369627334006832, "learning_rate": 1.9080202517465508e-06, "loss": 2.6826, "step": 44490 }, { "epoch": 2.761872245328698, "grad_norm": 0.3907179531128034, "learning_rate": 1.9070322138053043e-06, "loss": 2.672, "step": 44491 }, { "epoch": 2.761934322428456, "grad_norm": 0.36718267072383176, "learning_rate": 1.906044426776754e-06, "loss": 2.6605, "step": 44492 }, { "epoch": 2.761996399528214, "grad_norm": 0.3676485048994908, "learning_rate": 1.905056890666046e-06, "loss": 2.6269, "step": 44493 }, { "epoch": 2.762058476627972, "grad_norm": 0.3761177293385976, "learning_rate": 1.9040696054783257e-06, "loss": 2.6077, "step": 44494 }, { "epoch": 2.76212055372773, "grad_norm": 0.4102304663950504, "learning_rate": 1.903082571218756e-06, "loss": 2.6902, "step": 44495 }, { "epoch": 2.7621826308274877, "grad_norm": 0.35928219135868167, "learning_rate": 1.9020957878924716e-06, "loss": 2.5744, "step": 44496 }, { "epoch": 2.7622447079272456, "grad_norm": 0.34936005210631704, "learning_rate": 1.901109255504635e-06, "loss": 2.6651, "step": 44497 }, { "epoch": 2.7623067850270036, "grad_norm": 0.3610709641322768, "learning_rate": 1.9001229740603921e-06, "loss": 2.6396, "step": 44498 }, { "epoch": 2.7623688621267615, "grad_norm": 0.35623974032435624, "learning_rate": 1.8991369435648776e-06, "loss": 2.7374, "step": 44499 }, { "epoch": 2.7624309392265194, "grad_norm": 0.3547405497331925, "learning_rate": 1.898151164023243e-06, "loss": 2.6237, "step": 44500 }, { "epoch": 2.7624930163262773, "grad_norm": 0.37100849176081324, "learning_rate": 1.8971656354406341e-06, "loss": 2.6713, "step": 44501 }, { "epoch": 2.7625550934260352, "grad_norm": 0.3688640062638735, "learning_rate": 1.896180357822186e-06, "loss": 2.6485, "step": 44502 }, { "epoch": 2.762617170525793, "grad_norm": 0.3492581505278366, "learning_rate": 1.8951953311730497e-06, "loss": 2.6128, "step": 44503 }, { "epoch": 2.762679247625551, "grad_norm": 0.3675920003447062, "learning_rate": 1.8942105554983547e-06, "loss": 2.6175, "step": 44504 }, { "epoch": 2.762741324725309, "grad_norm": 0.3832342977025639, "learning_rate": 1.89322603080323e-06, "loss": 2.6173, "step": 44505 }, { "epoch": 2.762803401825067, "grad_norm": 0.3606298349145636, "learning_rate": 1.892241757092833e-06, "loss": 2.6771, "step": 44506 }, { "epoch": 2.762865478924825, "grad_norm": 0.38076620674081035, "learning_rate": 1.8912577343722926e-06, "loss": 2.6398, "step": 44507 }, { "epoch": 2.7629275560245823, "grad_norm": 0.33717033480419684, "learning_rate": 1.8902739626467382e-06, "loss": 2.7259, "step": 44508 }, { "epoch": 2.7629896331243406, "grad_norm": 0.37527757544095974, "learning_rate": 1.8892904419212932e-06, "loss": 2.7191, "step": 44509 }, { "epoch": 2.763051710224098, "grad_norm": 0.3533984142753363, "learning_rate": 1.8883071722011093e-06, "loss": 2.6257, "step": 44510 }, { "epoch": 2.7631137873238565, "grad_norm": 0.4080140072729839, "learning_rate": 1.8873241534913045e-06, "loss": 2.6712, "step": 44511 }, { "epoch": 2.763175864423614, "grad_norm": 0.390866757921132, "learning_rate": 1.8863413857970025e-06, "loss": 2.6363, "step": 44512 }, { "epoch": 2.763237941523372, "grad_norm": 0.3885210981200213, "learning_rate": 1.8853588691233493e-06, "loss": 2.6347, "step": 44513 }, { "epoch": 2.76330001862313, "grad_norm": 0.3621560380451011, "learning_rate": 1.8843766034754462e-06, "loss": 2.6811, "step": 44514 }, { "epoch": 2.7633620957228877, "grad_norm": 0.3653576280834096, "learning_rate": 1.8833945888584392e-06, "loss": 2.7143, "step": 44515 }, { "epoch": 2.7634241728226456, "grad_norm": 0.36346379687621067, "learning_rate": 1.8824128252774465e-06, "loss": 2.7364, "step": 44516 }, { "epoch": 2.7634862499224035, "grad_norm": 0.3466123918578211, "learning_rate": 1.8814313127375804e-06, "loss": 2.6125, "step": 44517 }, { "epoch": 2.7635483270221615, "grad_norm": 0.34592055902119195, "learning_rate": 1.880450051243976e-06, "loss": 2.6326, "step": 44518 }, { "epoch": 2.7636104041219194, "grad_norm": 0.3722307354968071, "learning_rate": 1.8794690408017346e-06, "loss": 2.6938, "step": 44519 }, { "epoch": 2.7636724812216773, "grad_norm": 0.3459180282481583, "learning_rate": 1.878488281415991e-06, "loss": 2.6101, "step": 44520 }, { "epoch": 2.763734558321435, "grad_norm": 0.36176537528379726, "learning_rate": 1.8775077730918577e-06, "loss": 2.7382, "step": 44521 }, { "epoch": 2.763796635421193, "grad_norm": 0.3325006420803324, "learning_rate": 1.8765275158344474e-06, "loss": 2.6419, "step": 44522 }, { "epoch": 2.763858712520951, "grad_norm": 0.37308951328925677, "learning_rate": 1.8755475096488672e-06, "loss": 2.6944, "step": 44523 }, { "epoch": 2.763920789620709, "grad_norm": 0.3463455093946259, "learning_rate": 1.8745677545402462e-06, "loss": 2.6248, "step": 44524 }, { "epoch": 2.763982866720467, "grad_norm": 0.3515206841925729, "learning_rate": 1.8735882505136915e-06, "loss": 2.7204, "step": 44525 }, { "epoch": 2.764044943820225, "grad_norm": 0.3511201724986591, "learning_rate": 1.8726089975743044e-06, "loss": 2.7224, "step": 44526 }, { "epoch": 2.7641070209199827, "grad_norm": 0.3703363016583559, "learning_rate": 1.8716299957271978e-06, "loss": 2.7559, "step": 44527 }, { "epoch": 2.7641690980197406, "grad_norm": 0.36113395628843425, "learning_rate": 1.8706512449774837e-06, "loss": 2.7453, "step": 44528 }, { "epoch": 2.7642311751194986, "grad_norm": 0.39626511021513666, "learning_rate": 1.8696727453302643e-06, "loss": 2.5771, "step": 44529 }, { "epoch": 2.7642932522192565, "grad_norm": 0.354452398920356, "learning_rate": 1.8686944967906516e-06, "loss": 2.5776, "step": 44530 }, { "epoch": 2.764355329319014, "grad_norm": 0.360070695459017, "learning_rate": 1.867716499363742e-06, "loss": 2.6555, "step": 44531 }, { "epoch": 2.7644174064187723, "grad_norm": 0.3924884219193141, "learning_rate": 1.866738753054631e-06, "loss": 2.6536, "step": 44532 }, { "epoch": 2.76447948351853, "grad_norm": 0.3792993335270979, "learning_rate": 1.865761257868437e-06, "loss": 2.611, "step": 44533 }, { "epoch": 2.764541560618288, "grad_norm": 0.38469598594992843, "learning_rate": 1.8647840138102557e-06, "loss": 2.5921, "step": 44534 }, { "epoch": 2.7646036377180456, "grad_norm": 0.39179044434888444, "learning_rate": 1.8638070208851777e-06, "loss": 2.6125, "step": 44535 }, { "epoch": 2.764665714817804, "grad_norm": 0.35706841340085277, "learning_rate": 1.8628302790982933e-06, "loss": 2.651, "step": 44536 }, { "epoch": 2.7647277919175615, "grad_norm": 0.36491990448426437, "learning_rate": 1.8618537884547205e-06, "loss": 2.6802, "step": 44537 }, { "epoch": 2.7647898690173194, "grad_norm": 0.3864623867278185, "learning_rate": 1.8608775489595443e-06, "loss": 2.6765, "step": 44538 }, { "epoch": 2.7648519461170773, "grad_norm": 0.41196482505222093, "learning_rate": 1.8599015606178495e-06, "loss": 2.6494, "step": 44539 }, { "epoch": 2.764914023216835, "grad_norm": 0.3655278459892522, "learning_rate": 1.8589258234347317e-06, "loss": 2.5978, "step": 44540 }, { "epoch": 2.764976100316593, "grad_norm": 0.39732876016211344, "learning_rate": 1.8579503374152928e-06, "loss": 2.6747, "step": 44541 }, { "epoch": 2.765038177416351, "grad_norm": 0.37918898268762347, "learning_rate": 1.8569751025646121e-06, "loss": 2.6892, "step": 44542 }, { "epoch": 2.765100254516109, "grad_norm": 0.37048860700235925, "learning_rate": 1.8560001188877796e-06, "loss": 2.5032, "step": 44543 }, { "epoch": 2.765162331615867, "grad_norm": 0.35565907759213206, "learning_rate": 1.8550253863898859e-06, "loss": 2.6669, "step": 44544 }, { "epoch": 2.765224408715625, "grad_norm": 0.34582355793768305, "learning_rate": 1.854050905075999e-06, "loss": 2.6371, "step": 44545 }, { "epoch": 2.7652864858153827, "grad_norm": 0.3676510380782029, "learning_rate": 1.8530766749512207e-06, "loss": 2.6847, "step": 44546 }, { "epoch": 2.7653485629151406, "grad_norm": 0.3444078825744142, "learning_rate": 1.8521026960206355e-06, "loss": 2.6797, "step": 44547 }, { "epoch": 2.7654106400148986, "grad_norm": 0.3631722891869653, "learning_rate": 1.8511289682893174e-06, "loss": 2.6546, "step": 44548 }, { "epoch": 2.7654727171146565, "grad_norm": 0.3548305567477024, "learning_rate": 1.8501554917623454e-06, "loss": 2.621, "step": 44549 }, { "epoch": 2.7655347942144144, "grad_norm": 0.3720256948877426, "learning_rate": 1.8491822664447989e-06, "loss": 2.6846, "step": 44550 }, { "epoch": 2.7655968713141723, "grad_norm": 0.3720354479377055, "learning_rate": 1.8482092923417627e-06, "loss": 2.6675, "step": 44551 }, { "epoch": 2.7656589484139302, "grad_norm": 0.3755027262236963, "learning_rate": 1.8472365694583105e-06, "loss": 2.6765, "step": 44552 }, { "epoch": 2.765721025513688, "grad_norm": 0.38145234335393197, "learning_rate": 1.8462640977995105e-06, "loss": 2.7069, "step": 44553 }, { "epoch": 2.765783102613446, "grad_norm": 0.3654240610330751, "learning_rate": 1.845291877370442e-06, "loss": 2.7425, "step": 44554 }, { "epoch": 2.765845179713204, "grad_norm": 0.3876601234436252, "learning_rate": 1.8443199081761675e-06, "loss": 2.7593, "step": 44555 }, { "epoch": 2.7659072568129615, "grad_norm": 0.35177590992413854, "learning_rate": 1.8433481902217775e-06, "loss": 2.698, "step": 44556 }, { "epoch": 2.76596933391272, "grad_norm": 0.3638388753248436, "learning_rate": 1.8423767235123235e-06, "loss": 2.679, "step": 44557 }, { "epoch": 2.7660314110124773, "grad_norm": 0.34987530104749237, "learning_rate": 1.8414055080528846e-06, "loss": 2.7317, "step": 44558 }, { "epoch": 2.7660934881122357, "grad_norm": 0.33907222865689984, "learning_rate": 1.8404345438485183e-06, "loss": 2.6784, "step": 44559 }, { "epoch": 2.766155565211993, "grad_norm": 0.39203918053704356, "learning_rate": 1.8394638309043034e-06, "loss": 2.6269, "step": 44560 }, { "epoch": 2.766217642311751, "grad_norm": 0.3789568208591225, "learning_rate": 1.838493369225297e-06, "loss": 2.6467, "step": 44561 }, { "epoch": 2.766279719411509, "grad_norm": 0.3645117644045426, "learning_rate": 1.8375231588165564e-06, "loss": 2.6429, "step": 44562 }, { "epoch": 2.766341796511267, "grad_norm": 0.34964007237527905, "learning_rate": 1.8365531996831497e-06, "loss": 2.7267, "step": 44563 }, { "epoch": 2.766403873611025, "grad_norm": 0.3579542645471194, "learning_rate": 1.8355834918301395e-06, "loss": 2.6337, "step": 44564 }, { "epoch": 2.7664659507107827, "grad_norm": 0.36796052198666607, "learning_rate": 1.8346140352625884e-06, "loss": 2.6117, "step": 44565 }, { "epoch": 2.7665280278105406, "grad_norm": 0.3400157953372955, "learning_rate": 1.8336448299855425e-06, "loss": 2.6895, "step": 44566 }, { "epoch": 2.7665901049102986, "grad_norm": 0.3468074831679646, "learning_rate": 1.832675876004064e-06, "loss": 2.6569, "step": 44567 }, { "epoch": 2.7666521820100565, "grad_norm": 0.37977335708741927, "learning_rate": 1.831707173323205e-06, "loss": 2.6607, "step": 44568 }, { "epoch": 2.7667142591098144, "grad_norm": 0.3661364670374414, "learning_rate": 1.8307387219480277e-06, "loss": 2.6288, "step": 44569 }, { "epoch": 2.7667763362095723, "grad_norm": 0.3736940532134603, "learning_rate": 1.8297705218835836e-06, "loss": 2.6776, "step": 44570 }, { "epoch": 2.7668384133093302, "grad_norm": 0.3572268761851924, "learning_rate": 1.8288025731349134e-06, "loss": 2.7055, "step": 44571 }, { "epoch": 2.766900490409088, "grad_norm": 0.3788243234687141, "learning_rate": 1.827834875707074e-06, "loss": 2.6419, "step": 44572 }, { "epoch": 2.766962567508846, "grad_norm": 0.3583425951285651, "learning_rate": 1.8268674296051169e-06, "loss": 2.6289, "step": 44573 }, { "epoch": 2.767024644608604, "grad_norm": 0.34567123740163763, "learning_rate": 1.8259002348340881e-06, "loss": 2.6947, "step": 44574 }, { "epoch": 2.767086721708362, "grad_norm": 0.3719947918519635, "learning_rate": 1.8249332913990337e-06, "loss": 2.6955, "step": 44575 }, { "epoch": 2.76714879880812, "grad_norm": 0.37117302700306365, "learning_rate": 1.8239665993049938e-06, "loss": 2.6617, "step": 44576 }, { "epoch": 2.7672108759078777, "grad_norm": 0.3446975349978341, "learning_rate": 1.823000158557009e-06, "loss": 2.6737, "step": 44577 }, { "epoch": 2.7672729530076356, "grad_norm": 0.3547198192643164, "learning_rate": 1.8220339691601362e-06, "loss": 2.6427, "step": 44578 }, { "epoch": 2.767335030107393, "grad_norm": 0.36572762703668826, "learning_rate": 1.8210680311194107e-06, "loss": 2.7367, "step": 44579 }, { "epoch": 2.7673971072071515, "grad_norm": 0.3743390974874552, "learning_rate": 1.8201023444398667e-06, "loss": 2.6424, "step": 44580 }, { "epoch": 2.767459184306909, "grad_norm": 0.3682027740768227, "learning_rate": 1.8191369091265453e-06, "loss": 2.7121, "step": 44581 }, { "epoch": 2.7675212614066673, "grad_norm": 0.34290626321603285, "learning_rate": 1.8181717251844754e-06, "loss": 2.6044, "step": 44582 }, { "epoch": 2.767583338506425, "grad_norm": 0.37540535079224696, "learning_rate": 1.8172067926187087e-06, "loss": 2.6806, "step": 44583 }, { "epoch": 2.7676454156061827, "grad_norm": 0.3762572194670965, "learning_rate": 1.8162421114342688e-06, "loss": 2.7456, "step": 44584 }, { "epoch": 2.7677074927059406, "grad_norm": 0.3789347018541717, "learning_rate": 1.8152776816361904e-06, "loss": 2.5866, "step": 44585 }, { "epoch": 2.7677695698056985, "grad_norm": 0.3857817038081093, "learning_rate": 1.8143135032294978e-06, "loss": 2.6119, "step": 44586 }, { "epoch": 2.7678316469054565, "grad_norm": 0.35077867018466774, "learning_rate": 1.8133495762192421e-06, "loss": 2.578, "step": 44587 }, { "epoch": 2.7678937240052144, "grad_norm": 0.36851941674266964, "learning_rate": 1.8123859006104305e-06, "loss": 2.6894, "step": 44588 }, { "epoch": 2.7679558011049723, "grad_norm": 0.3551725683730367, "learning_rate": 1.8114224764081034e-06, "loss": 2.6176, "step": 44589 }, { "epoch": 2.76801787820473, "grad_norm": 0.36894796278624137, "learning_rate": 1.810459303617279e-06, "loss": 2.6408, "step": 44590 }, { "epoch": 2.768079955304488, "grad_norm": 0.396421436895512, "learning_rate": 1.8094963822429922e-06, "loss": 2.6596, "step": 44591 }, { "epoch": 2.768142032404246, "grad_norm": 0.38787338049473985, "learning_rate": 1.8085337122902613e-06, "loss": 2.6513, "step": 44592 }, { "epoch": 2.768204109504004, "grad_norm": 0.3752735650086268, "learning_rate": 1.80757129376411e-06, "loss": 2.6156, "step": 44593 }, { "epoch": 2.768266186603762, "grad_norm": 0.34836732816375726, "learning_rate": 1.806609126669556e-06, "loss": 2.6338, "step": 44594 }, { "epoch": 2.76832826370352, "grad_norm": 0.36378333558381887, "learning_rate": 1.8056472110116185e-06, "loss": 2.651, "step": 44595 }, { "epoch": 2.7683903408032777, "grad_norm": 0.36032133311604797, "learning_rate": 1.8046855467953262e-06, "loss": 2.5595, "step": 44596 }, { "epoch": 2.7684524179030356, "grad_norm": 0.3719070757327935, "learning_rate": 1.8037241340256861e-06, "loss": 2.7241, "step": 44597 }, { "epoch": 2.7685144950027936, "grad_norm": 0.3732572431736644, "learning_rate": 1.8027629727077166e-06, "loss": 2.6426, "step": 44598 }, { "epoch": 2.7685765721025515, "grad_norm": 0.35511130921728384, "learning_rate": 1.801802062846425e-06, "loss": 2.6544, "step": 44599 }, { "epoch": 2.7686386492023094, "grad_norm": 0.3681549354453984, "learning_rate": 1.8008414044468457e-06, "loss": 2.6321, "step": 44600 }, { "epoch": 2.7687007263020673, "grad_norm": 0.3654847944047756, "learning_rate": 1.7998809975139698e-06, "loss": 2.6793, "step": 44601 }, { "epoch": 2.768762803401825, "grad_norm": 0.3512948500857339, "learning_rate": 1.7989208420528204e-06, "loss": 2.6247, "step": 44602 }, { "epoch": 2.768824880501583, "grad_norm": 0.3679391188951208, "learning_rate": 1.7979609380683993e-06, "loss": 2.6064, "step": 44603 }, { "epoch": 2.7688869576013406, "grad_norm": 0.3655450896618808, "learning_rate": 1.7970012855657136e-06, "loss": 2.5982, "step": 44604 }, { "epoch": 2.768949034701099, "grad_norm": 0.36581315880342513, "learning_rate": 1.7960418845497817e-06, "loss": 2.6728, "step": 44605 }, { "epoch": 2.7690111118008565, "grad_norm": 0.36606019032736253, "learning_rate": 1.7950827350255994e-06, "loss": 2.8109, "step": 44606 }, { "epoch": 2.769073188900615, "grad_norm": 0.34763991928320426, "learning_rate": 1.7941238369981739e-06, "loss": 2.6381, "step": 44607 }, { "epoch": 2.7691352660003723, "grad_norm": 0.3544787990259145, "learning_rate": 1.7931651904724955e-06, "loss": 2.6234, "step": 44608 }, { "epoch": 2.76919734310013, "grad_norm": 0.35105884290192424, "learning_rate": 1.7922067954535882e-06, "loss": 2.6782, "step": 44609 }, { "epoch": 2.769259420199888, "grad_norm": 0.34985300451375784, "learning_rate": 1.791248651946431e-06, "loss": 2.6506, "step": 44610 }, { "epoch": 2.769321497299646, "grad_norm": 0.35704884057983227, "learning_rate": 1.7902907599560426e-06, "loss": 2.6321, "step": 44611 }, { "epoch": 2.769383574399404, "grad_norm": 0.3927665941276754, "learning_rate": 1.7893331194874075e-06, "loss": 2.7129, "step": 44612 }, { "epoch": 2.769445651499162, "grad_norm": 0.35634137035961294, "learning_rate": 1.7883757305455218e-06, "loss": 2.6151, "step": 44613 }, { "epoch": 2.76950772859892, "grad_norm": 0.3570844172450514, "learning_rate": 1.787418593135387e-06, "loss": 2.7379, "step": 44614 }, { "epoch": 2.7695698056986777, "grad_norm": 0.36996311815835753, "learning_rate": 1.7864617072619937e-06, "loss": 2.5519, "step": 44615 }, { "epoch": 2.7696318827984356, "grad_norm": 0.3622713304983107, "learning_rate": 1.785505072930338e-06, "loss": 2.6534, "step": 44616 }, { "epoch": 2.7696939598981936, "grad_norm": 0.37636039264525695, "learning_rate": 1.7845486901454045e-06, "loss": 2.7225, "step": 44617 }, { "epoch": 2.7697560369979515, "grad_norm": 0.3471049963902252, "learning_rate": 1.7835925589121783e-06, "loss": 2.5338, "step": 44618 }, { "epoch": 2.7698181140977094, "grad_norm": 0.34938427801898125, "learning_rate": 1.7826366792356607e-06, "loss": 2.633, "step": 44619 }, { "epoch": 2.7698801911974673, "grad_norm": 0.3432890445690732, "learning_rate": 1.7816810511208316e-06, "loss": 2.7506, "step": 44620 }, { "epoch": 2.7699422682972252, "grad_norm": 0.3464742884839668, "learning_rate": 1.7807256745726808e-06, "loss": 2.628, "step": 44621 }, { "epoch": 2.770004345396983, "grad_norm": 0.3554405368885147, "learning_rate": 1.7797705495961825e-06, "loss": 2.6212, "step": 44622 }, { "epoch": 2.770066422496741, "grad_norm": 0.36407228710460693, "learning_rate": 1.7788156761963326e-06, "loss": 2.7235, "step": 44623 }, { "epoch": 2.770128499596499, "grad_norm": 0.3609059522308646, "learning_rate": 1.7778610543781048e-06, "loss": 2.6876, "step": 44624 }, { "epoch": 2.770190576696257, "grad_norm": 0.3624543243439668, "learning_rate": 1.7769066841464844e-06, "loss": 2.6349, "step": 44625 }, { "epoch": 2.770252653796015, "grad_norm": 0.35764062159878796, "learning_rate": 1.7759525655064446e-06, "loss": 2.7145, "step": 44626 }, { "epoch": 2.7703147308957723, "grad_norm": 0.3558929762485631, "learning_rate": 1.7749986984629707e-06, "loss": 2.5396, "step": 44627 }, { "epoch": 2.7703768079955307, "grad_norm": 0.31843656060601544, "learning_rate": 1.7740450830210365e-06, "loss": 2.5928, "step": 44628 }, { "epoch": 2.770438885095288, "grad_norm": 0.3621905370655711, "learning_rate": 1.7730917191856156e-06, "loss": 2.6511, "step": 44629 }, { "epoch": 2.7705009621950465, "grad_norm": 0.35637758212456994, "learning_rate": 1.7721386069616818e-06, "loss": 2.6869, "step": 44630 }, { "epoch": 2.770563039294804, "grad_norm": 0.3756527583867012, "learning_rate": 1.7711857463542091e-06, "loss": 2.6729, "step": 44631 }, { "epoch": 2.770625116394562, "grad_norm": 0.36139193821931137, "learning_rate": 1.7702331373681657e-06, "loss": 2.598, "step": 44632 }, { "epoch": 2.77068719349432, "grad_norm": 0.3525249023975172, "learning_rate": 1.769280780008531e-06, "loss": 2.6062, "step": 44633 }, { "epoch": 2.7707492705940777, "grad_norm": 0.35730457986938685, "learning_rate": 1.768328674280262e-06, "loss": 2.7278, "step": 44634 }, { "epoch": 2.7708113476938356, "grad_norm": 0.35088601194088404, "learning_rate": 1.7673768201883268e-06, "loss": 2.6898, "step": 44635 }, { "epoch": 2.7708734247935936, "grad_norm": 0.3509359392486405, "learning_rate": 1.7664252177376995e-06, "loss": 2.6621, "step": 44636 }, { "epoch": 2.7709355018933515, "grad_norm": 0.3523185864786135, "learning_rate": 1.7654738669333426e-06, "loss": 2.5572, "step": 44637 }, { "epoch": 2.7709975789931094, "grad_norm": 0.3656329463845116, "learning_rate": 1.7645227677802189e-06, "loss": 2.6044, "step": 44638 }, { "epoch": 2.7710596560928673, "grad_norm": 0.34954762046584076, "learning_rate": 1.7635719202832912e-06, "loss": 2.6565, "step": 44639 }, { "epoch": 2.7711217331926252, "grad_norm": 0.37440136971049937, "learning_rate": 1.762621324447511e-06, "loss": 2.677, "step": 44640 }, { "epoch": 2.771183810292383, "grad_norm": 0.35904892350468276, "learning_rate": 1.7616709802778465e-06, "loss": 2.6385, "step": 44641 }, { "epoch": 2.771245887392141, "grad_norm": 0.37093793709913175, "learning_rate": 1.7607208877792604e-06, "loss": 2.752, "step": 44642 }, { "epoch": 2.771307964491899, "grad_norm": 0.33791905763103997, "learning_rate": 1.7597710469566986e-06, "loss": 2.5927, "step": 44643 }, { "epoch": 2.771370041591657, "grad_norm": 0.3634018851531537, "learning_rate": 1.7588214578151242e-06, "loss": 2.6473, "step": 44644 }, { "epoch": 2.771432118691415, "grad_norm": 0.34723818212493174, "learning_rate": 1.7578721203594827e-06, "loss": 2.5937, "step": 44645 }, { "epoch": 2.7714941957911727, "grad_norm": 0.3710425291349457, "learning_rate": 1.7569230345947429e-06, "loss": 2.7683, "step": 44646 }, { "epoch": 2.7715562728909307, "grad_norm": 0.37702433254505874, "learning_rate": 1.755974200525845e-06, "loss": 2.69, "step": 44647 }, { "epoch": 2.7716183499906886, "grad_norm": 0.3773560026402826, "learning_rate": 1.7550256181577407e-06, "loss": 2.6194, "step": 44648 }, { "epoch": 2.7716804270904465, "grad_norm": 0.37379125654445916, "learning_rate": 1.754077287495376e-06, "loss": 2.607, "step": 44649 }, { "epoch": 2.771742504190204, "grad_norm": 0.3825301861539024, "learning_rate": 1.753129208543708e-06, "loss": 2.6454, "step": 44650 }, { "epoch": 2.7718045812899623, "grad_norm": 0.3449430137369162, "learning_rate": 1.7521813813076771e-06, "loss": 2.5689, "step": 44651 }, { "epoch": 2.77186665838972, "grad_norm": 0.3523160170744487, "learning_rate": 1.7512338057922295e-06, "loss": 2.6799, "step": 44652 }, { "epoch": 2.771928735489478, "grad_norm": 0.37678917595584543, "learning_rate": 1.7502864820023057e-06, "loss": 2.6477, "step": 44653 }, { "epoch": 2.7719908125892356, "grad_norm": 0.34755423474237856, "learning_rate": 1.7493394099428407e-06, "loss": 2.7328, "step": 44654 }, { "epoch": 2.772052889688994, "grad_norm": 0.3446876676840906, "learning_rate": 1.748392589618797e-06, "loss": 2.6646, "step": 44655 }, { "epoch": 2.7721149667887515, "grad_norm": 0.38267477633207486, "learning_rate": 1.747446021035104e-06, "loss": 2.7214, "step": 44656 }, { "epoch": 2.7721770438885094, "grad_norm": 0.33812237107458915, "learning_rate": 1.7464997041966913e-06, "loss": 2.7489, "step": 44657 }, { "epoch": 2.7722391209882673, "grad_norm": 0.39454624411917855, "learning_rate": 1.7455536391085048e-06, "loss": 2.7582, "step": 44658 }, { "epoch": 2.772301198088025, "grad_norm": 0.33430844473469706, "learning_rate": 1.744607825775485e-06, "loss": 2.6011, "step": 44659 }, { "epoch": 2.772363275187783, "grad_norm": 0.3790847908252519, "learning_rate": 1.7436622642025557e-06, "loss": 2.7085, "step": 44660 }, { "epoch": 2.772425352287541, "grad_norm": 0.33490258868074974, "learning_rate": 1.7427169543946576e-06, "loss": 2.6861, "step": 44661 }, { "epoch": 2.772487429387299, "grad_norm": 0.35585453771324654, "learning_rate": 1.7417718963567143e-06, "loss": 2.6314, "step": 44662 }, { "epoch": 2.772549506487057, "grad_norm": 0.35302633071690753, "learning_rate": 1.7408270900936663e-06, "loss": 2.6659, "step": 44663 }, { "epoch": 2.772611583586815, "grad_norm": 0.37801277371046565, "learning_rate": 1.7398825356104432e-06, "loss": 2.7062, "step": 44664 }, { "epoch": 2.7726736606865727, "grad_norm": 0.3504964135077639, "learning_rate": 1.738938232911963e-06, "loss": 2.5978, "step": 44665 }, { "epoch": 2.7727357377863306, "grad_norm": 0.35797353810405413, "learning_rate": 1.737994182003161e-06, "loss": 2.724, "step": 44666 }, { "epoch": 2.7727978148860886, "grad_norm": 0.3704317512523104, "learning_rate": 1.7370503828889495e-06, "loss": 2.697, "step": 44667 }, { "epoch": 2.7728598919858465, "grad_norm": 0.3755840283611653, "learning_rate": 1.736106835574275e-06, "loss": 2.705, "step": 44668 }, { "epoch": 2.7729219690856044, "grad_norm": 0.3440775286261938, "learning_rate": 1.7351635400640442e-06, "loss": 2.693, "step": 44669 }, { "epoch": 2.7729840461853623, "grad_norm": 0.3896597380182574, "learning_rate": 1.7342204963631813e-06, "loss": 2.6309, "step": 44670 }, { "epoch": 2.7730461232851202, "grad_norm": 0.3718003021678651, "learning_rate": 1.7332777044765992e-06, "loss": 2.6704, "step": 44671 }, { "epoch": 2.773108200384878, "grad_norm": 0.3646545090907992, "learning_rate": 1.7323351644092323e-06, "loss": 2.7507, "step": 44672 }, { "epoch": 2.773170277484636, "grad_norm": 0.36202990453056005, "learning_rate": 1.7313928761659937e-06, "loss": 2.6696, "step": 44673 }, { "epoch": 2.773232354584394, "grad_norm": 0.3515421437622, "learning_rate": 1.7304508397517905e-06, "loss": 2.7143, "step": 44674 }, { "epoch": 2.7732944316841515, "grad_norm": 0.3787808948292046, "learning_rate": 1.729509055171541e-06, "loss": 2.6843, "step": 44675 }, { "epoch": 2.77335650878391, "grad_norm": 0.36244406679680347, "learning_rate": 1.7285675224301578e-06, "loss": 2.6403, "step": 44676 }, { "epoch": 2.7734185858836673, "grad_norm": 0.3586562474127731, "learning_rate": 1.727626241532565e-06, "loss": 2.6559, "step": 44677 }, { "epoch": 2.7734806629834257, "grad_norm": 0.3560347191028777, "learning_rate": 1.7266852124836641e-06, "loss": 2.679, "step": 44678 }, { "epoch": 2.773542740083183, "grad_norm": 0.3377476317967164, "learning_rate": 1.7257444352883622e-06, "loss": 2.6405, "step": 44679 }, { "epoch": 2.773604817182941, "grad_norm": 0.38387291843914484, "learning_rate": 1.7248039099515724e-06, "loss": 2.6715, "step": 44680 }, { "epoch": 2.773666894282699, "grad_norm": 0.367420143124982, "learning_rate": 1.7238636364781958e-06, "loss": 2.7478, "step": 44681 }, { "epoch": 2.773728971382457, "grad_norm": 0.3514593502886218, "learning_rate": 1.7229236148731454e-06, "loss": 2.6939, "step": 44682 }, { "epoch": 2.773791048482215, "grad_norm": 0.3720805249352034, "learning_rate": 1.7219838451413227e-06, "loss": 2.7301, "step": 44683 }, { "epoch": 2.7738531255819727, "grad_norm": 0.33935680800798007, "learning_rate": 1.7210443272876298e-06, "loss": 2.6732, "step": 44684 }, { "epoch": 2.7739152026817306, "grad_norm": 0.36621675613867033, "learning_rate": 1.7201050613169679e-06, "loss": 2.6297, "step": 44685 }, { "epoch": 2.7739772797814886, "grad_norm": 0.33434448013775026, "learning_rate": 1.7191660472342387e-06, "loss": 2.6599, "step": 44686 }, { "epoch": 2.7740393568812465, "grad_norm": 0.3752192611239712, "learning_rate": 1.7182272850443383e-06, "loss": 2.6559, "step": 44687 }, { "epoch": 2.7741014339810044, "grad_norm": 0.3607468543295541, "learning_rate": 1.7172887747521737e-06, "loss": 2.6864, "step": 44688 }, { "epoch": 2.7741635110807623, "grad_norm": 0.3710906781521194, "learning_rate": 1.7163505163626303e-06, "loss": 2.6357, "step": 44689 }, { "epoch": 2.7742255881805202, "grad_norm": 0.36836476003486934, "learning_rate": 1.715412509880604e-06, "loss": 2.5662, "step": 44690 }, { "epoch": 2.774287665280278, "grad_norm": 0.3717001127082258, "learning_rate": 1.7144747553109963e-06, "loss": 2.6893, "step": 44691 }, { "epoch": 2.774349742380036, "grad_norm": 0.34152310988004614, "learning_rate": 1.713537252658698e-06, "loss": 2.6586, "step": 44692 }, { "epoch": 2.774411819479794, "grad_norm": 0.3536776808760803, "learning_rate": 1.7126000019285938e-06, "loss": 2.602, "step": 44693 }, { "epoch": 2.774473896579552, "grad_norm": 0.36166936895720403, "learning_rate": 1.7116630031255743e-06, "loss": 2.6669, "step": 44694 }, { "epoch": 2.77453597367931, "grad_norm": 0.37475572020686543, "learning_rate": 1.7107262562545357e-06, "loss": 2.6002, "step": 44695 }, { "epoch": 2.7745980507790677, "grad_norm": 0.33919921131697434, "learning_rate": 1.7097897613203628e-06, "loss": 2.5755, "step": 44696 }, { "epoch": 2.7746601278788257, "grad_norm": 0.39816435632851094, "learning_rate": 1.7088535183279409e-06, "loss": 2.6472, "step": 44697 }, { "epoch": 2.774722204978583, "grad_norm": 0.37505362047854196, "learning_rate": 1.7079175272821434e-06, "loss": 2.6381, "step": 44698 }, { "epoch": 2.7747842820783415, "grad_norm": 0.38314557668991805, "learning_rate": 1.7069817881878725e-06, "loss": 2.6686, "step": 44699 }, { "epoch": 2.774846359178099, "grad_norm": 0.35495465170314544, "learning_rate": 1.7060463010500016e-06, "loss": 2.7646, "step": 44700 }, { "epoch": 2.7749084362778573, "grad_norm": 0.35352754309359796, "learning_rate": 1.7051110658734049e-06, "loss": 2.7253, "step": 44701 }, { "epoch": 2.774970513377615, "grad_norm": 0.35879581883215267, "learning_rate": 1.7041760826629727e-06, "loss": 2.6297, "step": 44702 }, { "epoch": 2.775032590477373, "grad_norm": 0.38562193336071204, "learning_rate": 1.7032413514235735e-06, "loss": 2.6978, "step": 44703 }, { "epoch": 2.7750946675771306, "grad_norm": 0.3531586863055074, "learning_rate": 1.7023068721600921e-06, "loss": 2.6706, "step": 44704 }, { "epoch": 2.7751567446768886, "grad_norm": 0.36455574839549837, "learning_rate": 1.7013726448774025e-06, "loss": 2.7309, "step": 44705 }, { "epoch": 2.7752188217766465, "grad_norm": 0.35676082077270055, "learning_rate": 1.700438669580373e-06, "loss": 2.6626, "step": 44706 }, { "epoch": 2.7752808988764044, "grad_norm": 0.3599914984891328, "learning_rate": 1.6995049462738777e-06, "loss": 2.6357, "step": 44707 }, { "epoch": 2.7753429759761623, "grad_norm": 0.3690229157579097, "learning_rate": 1.6985714749627902e-06, "loss": 2.754, "step": 44708 }, { "epoch": 2.7754050530759202, "grad_norm": 0.3594684064800959, "learning_rate": 1.6976382556519844e-06, "loss": 2.695, "step": 44709 }, { "epoch": 2.775467130175678, "grad_norm": 0.35735095211589146, "learning_rate": 1.6967052883463286e-06, "loss": 2.6962, "step": 44710 }, { "epoch": 2.775529207275436, "grad_norm": 0.3527931819151259, "learning_rate": 1.6957725730506857e-06, "loss": 2.6123, "step": 44711 }, { "epoch": 2.775591284375194, "grad_norm": 0.37374054491720393, "learning_rate": 1.694840109769913e-06, "loss": 2.6875, "step": 44712 }, { "epoch": 2.775653361474952, "grad_norm": 0.3753573979529452, "learning_rate": 1.6939078985089008e-06, "loss": 2.6991, "step": 44713 }, { "epoch": 2.77571543857471, "grad_norm": 0.34938162306033804, "learning_rate": 1.6929759392724899e-06, "loss": 2.6404, "step": 44714 }, { "epoch": 2.7757775156744677, "grad_norm": 0.35897564467481086, "learning_rate": 1.6920442320655538e-06, "loss": 2.6897, "step": 44715 }, { "epoch": 2.7758395927742257, "grad_norm": 0.35182296852696865, "learning_rate": 1.6911127768929501e-06, "loss": 2.6662, "step": 44716 }, { "epoch": 2.7759016698739836, "grad_norm": 0.3674988948065819, "learning_rate": 1.690181573759536e-06, "loss": 2.6737, "step": 44717 }, { "epoch": 2.7759637469737415, "grad_norm": 0.34741026501893163, "learning_rate": 1.6892506226701743e-06, "loss": 2.6746, "step": 44718 }, { "epoch": 2.7760258240734994, "grad_norm": 0.35495721250224926, "learning_rate": 1.6883199236297164e-06, "loss": 2.6723, "step": 44719 }, { "epoch": 2.7760879011732573, "grad_norm": 0.3590205974381441, "learning_rate": 1.6873894766430254e-06, "loss": 2.6846, "step": 44720 }, { "epoch": 2.7761499782730152, "grad_norm": 0.3430585300530676, "learning_rate": 1.6864592817149471e-06, "loss": 2.5231, "step": 44721 }, { "epoch": 2.776212055372773, "grad_norm": 0.36559428876698064, "learning_rate": 1.6855293388503445e-06, "loss": 2.6535, "step": 44722 }, { "epoch": 2.7762741324725306, "grad_norm": 0.35338807307654946, "learning_rate": 1.6845996480540637e-06, "loss": 2.6954, "step": 44723 }, { "epoch": 2.776336209572289, "grad_norm": 0.3819626502793931, "learning_rate": 1.6836702093309564e-06, "loss": 2.678, "step": 44724 }, { "epoch": 2.7763982866720465, "grad_norm": 0.352496411393208, "learning_rate": 1.6827410226858686e-06, "loss": 2.6655, "step": 44725 }, { "epoch": 2.776460363771805, "grad_norm": 0.3570794096003737, "learning_rate": 1.6818120881236466e-06, "loss": 2.6759, "step": 44726 }, { "epoch": 2.7765224408715623, "grad_norm": 0.3954666304635138, "learning_rate": 1.6808834056491473e-06, "loss": 2.7296, "step": 44727 }, { "epoch": 2.77658451797132, "grad_norm": 0.36385908624596064, "learning_rate": 1.6799549752672061e-06, "loss": 2.6574, "step": 44728 }, { "epoch": 2.776646595071078, "grad_norm": 0.36147579577842714, "learning_rate": 1.6790267969826744e-06, "loss": 2.6417, "step": 44729 }, { "epoch": 2.776708672170836, "grad_norm": 0.36552169024951436, "learning_rate": 1.6780988708003765e-06, "loss": 2.6209, "step": 44730 }, { "epoch": 2.776770749270594, "grad_norm": 0.3588069697413862, "learning_rate": 1.6771711967251802e-06, "loss": 2.6175, "step": 44731 }, { "epoch": 2.776832826370352, "grad_norm": 0.36932519809353753, "learning_rate": 1.6762437747619096e-06, "loss": 2.6568, "step": 44732 }, { "epoch": 2.77689490347011, "grad_norm": 0.3727844673482864, "learning_rate": 1.6753166049154056e-06, "loss": 2.6336, "step": 44733 }, { "epoch": 2.7769569805698677, "grad_norm": 0.36295671220540804, "learning_rate": 1.674389687190503e-06, "loss": 2.7554, "step": 44734 }, { "epoch": 2.7770190576696256, "grad_norm": 0.33433535465617986, "learning_rate": 1.6734630215920476e-06, "loss": 2.6563, "step": 44735 }, { "epoch": 2.7770811347693836, "grad_norm": 0.3382353485011645, "learning_rate": 1.672536608124864e-06, "loss": 2.6466, "step": 44736 }, { "epoch": 2.7771432118691415, "grad_norm": 0.37411207139923863, "learning_rate": 1.6716104467937866e-06, "loss": 2.7008, "step": 44737 }, { "epoch": 2.7772052889688994, "grad_norm": 0.43076527416117133, "learning_rate": 1.670684537603656e-06, "loss": 2.6357, "step": 44738 }, { "epoch": 2.7772673660686573, "grad_norm": 0.3623170464970564, "learning_rate": 1.6697588805592857e-06, "loss": 2.6349, "step": 44739 }, { "epoch": 2.7773294431684152, "grad_norm": 0.3610864191701844, "learning_rate": 1.6688334756655267e-06, "loss": 2.6344, "step": 44740 }, { "epoch": 2.777391520268173, "grad_norm": 0.35325634125770294, "learning_rate": 1.6679083229271863e-06, "loss": 2.6456, "step": 44741 }, { "epoch": 2.777453597367931, "grad_norm": 0.3478601586352265, "learning_rate": 1.6669834223491054e-06, "loss": 2.7365, "step": 44742 }, { "epoch": 2.777515674467689, "grad_norm": 0.3809961709082919, "learning_rate": 1.6660587739361079e-06, "loss": 2.6738, "step": 44743 }, { "epoch": 2.777577751567447, "grad_norm": 0.3657178076306806, "learning_rate": 1.6651343776930117e-06, "loss": 2.6632, "step": 44744 }, { "epoch": 2.777639828667205, "grad_norm": 0.3651116516862474, "learning_rate": 1.6642102336246468e-06, "loss": 2.6526, "step": 44745 }, { "epoch": 2.7777019057669623, "grad_norm": 0.36890430050843354, "learning_rate": 1.6632863417358312e-06, "loss": 2.6892, "step": 44746 }, { "epoch": 2.7777639828667207, "grad_norm": 0.3561479765381341, "learning_rate": 1.6623627020313836e-06, "loss": 2.6071, "step": 44747 }, { "epoch": 2.777826059966478, "grad_norm": 0.36917596476293124, "learning_rate": 1.6614393145161166e-06, "loss": 2.638, "step": 44748 }, { "epoch": 2.7778881370662365, "grad_norm": 0.34151473253176007, "learning_rate": 1.6605161791948654e-06, "loss": 2.7247, "step": 44749 }, { "epoch": 2.777950214165994, "grad_norm": 0.35467152748610176, "learning_rate": 1.659593296072437e-06, "loss": 2.5564, "step": 44750 }, { "epoch": 2.7780122912657523, "grad_norm": 0.3466679208397339, "learning_rate": 1.6586706651536387e-06, "loss": 2.6526, "step": 44751 }, { "epoch": 2.77807436836551, "grad_norm": 0.38645777903651596, "learning_rate": 1.6577482864432948e-06, "loss": 2.7181, "step": 44752 }, { "epoch": 2.7781364454652677, "grad_norm": 0.33322084162883026, "learning_rate": 1.6568261599462121e-06, "loss": 2.7238, "step": 44753 }, { "epoch": 2.7781985225650256, "grad_norm": 0.35463135985008715, "learning_rate": 1.6559042856672035e-06, "loss": 2.5499, "step": 44754 }, { "epoch": 2.7782605996647836, "grad_norm": 0.34094091866489906, "learning_rate": 1.6549826636110766e-06, "loss": 2.6528, "step": 44755 }, { "epoch": 2.7783226767645415, "grad_norm": 0.34526925959790744, "learning_rate": 1.6540612937826439e-06, "loss": 2.6833, "step": 44756 }, { "epoch": 2.7783847538642994, "grad_norm": 0.3745764486184414, "learning_rate": 1.6531401761867072e-06, "loss": 2.6275, "step": 44757 }, { "epoch": 2.7784468309640573, "grad_norm": 0.34858661634052973, "learning_rate": 1.6522193108280793e-06, "loss": 2.6421, "step": 44758 }, { "epoch": 2.7785089080638152, "grad_norm": 0.3410750369025176, "learning_rate": 1.6512986977115564e-06, "loss": 2.7042, "step": 44759 }, { "epoch": 2.778570985163573, "grad_norm": 0.3523040156998295, "learning_rate": 1.6503783368419455e-06, "loss": 2.6788, "step": 44760 }, { "epoch": 2.778633062263331, "grad_norm": 0.3761986807109002, "learning_rate": 1.6494582282240433e-06, "loss": 2.7116, "step": 44761 }, { "epoch": 2.778695139363089, "grad_norm": 0.3679939706829339, "learning_rate": 1.6485383718626623e-06, "loss": 2.6818, "step": 44762 }, { "epoch": 2.778757216462847, "grad_norm": 0.36947306039756517, "learning_rate": 1.6476187677625932e-06, "loss": 2.6742, "step": 44763 }, { "epoch": 2.778819293562605, "grad_norm": 0.3577279828132186, "learning_rate": 1.6466994159286375e-06, "loss": 2.6277, "step": 44764 }, { "epoch": 2.7788813706623627, "grad_norm": 0.36058068163453405, "learning_rate": 1.6457803163655861e-06, "loss": 2.6588, "step": 44765 }, { "epoch": 2.7789434477621207, "grad_norm": 0.3634386029476837, "learning_rate": 1.644861469078235e-06, "loss": 2.6193, "step": 44766 }, { "epoch": 2.7790055248618786, "grad_norm": 0.3746958930035585, "learning_rate": 1.6439428740713802e-06, "loss": 2.6915, "step": 44767 }, { "epoch": 2.7790676019616365, "grad_norm": 0.3624059324875783, "learning_rate": 1.6430245313498182e-06, "loss": 2.6481, "step": 44768 }, { "epoch": 2.7791296790613944, "grad_norm": 0.36112395007996595, "learning_rate": 1.642106440918334e-06, "loss": 2.6631, "step": 44769 }, { "epoch": 2.7791917561611523, "grad_norm": 0.3374192693194155, "learning_rate": 1.641188602781718e-06, "loss": 2.5927, "step": 44770 }, { "epoch": 2.77925383326091, "grad_norm": 0.34814464895452407, "learning_rate": 1.640271016944761e-06, "loss": 2.6805, "step": 44771 }, { "epoch": 2.779315910360668, "grad_norm": 0.3557629049377185, "learning_rate": 1.6393536834122536e-06, "loss": 2.6822, "step": 44772 }, { "epoch": 2.7793779874604256, "grad_norm": 0.36183207654288135, "learning_rate": 1.6384366021889752e-06, "loss": 2.5989, "step": 44773 }, { "epoch": 2.779440064560184, "grad_norm": 0.3587178358419777, "learning_rate": 1.6375197732797109e-06, "loss": 2.67, "step": 44774 }, { "epoch": 2.7795021416599415, "grad_norm": 0.369809427864604, "learning_rate": 1.6366031966892515e-06, "loss": 2.6299, "step": 44775 }, { "epoch": 2.7795642187596994, "grad_norm": 0.360090310934843, "learning_rate": 1.635686872422365e-06, "loss": 2.6428, "step": 44776 }, { "epoch": 2.7796262958594573, "grad_norm": 0.3669077233467772, "learning_rate": 1.6347708004838536e-06, "loss": 2.7554, "step": 44777 }, { "epoch": 2.7796883729592152, "grad_norm": 0.34785169829548535, "learning_rate": 1.6338549808784743e-06, "loss": 2.6869, "step": 44778 }, { "epoch": 2.779750450058973, "grad_norm": 0.3601438135412008, "learning_rate": 1.6329394136110233e-06, "loss": 2.7485, "step": 44779 }, { "epoch": 2.779812527158731, "grad_norm": 0.3572416215812413, "learning_rate": 1.6320240986862578e-06, "loss": 2.674, "step": 44780 }, { "epoch": 2.779874604258489, "grad_norm": 0.33769285334612276, "learning_rate": 1.6311090361089743e-06, "loss": 2.6355, "step": 44781 }, { "epoch": 2.779936681358247, "grad_norm": 0.34972855864660013, "learning_rate": 1.6301942258839297e-06, "loss": 2.6697, "step": 44782 }, { "epoch": 2.779998758458005, "grad_norm": 0.34252467461619196, "learning_rate": 1.6292796680159096e-06, "loss": 2.6444, "step": 44783 }, { "epoch": 2.7800608355577627, "grad_norm": 0.3625913036395088, "learning_rate": 1.6283653625096761e-06, "loss": 2.6685, "step": 44784 }, { "epoch": 2.7801229126575207, "grad_norm": 0.3621554414320598, "learning_rate": 1.6274513093700039e-06, "loss": 2.6153, "step": 44785 }, { "epoch": 2.7801849897572786, "grad_norm": 0.3434700693058821, "learning_rate": 1.6265375086016665e-06, "loss": 2.6171, "step": 44786 }, { "epoch": 2.7802470668570365, "grad_norm": 0.34417620159977674, "learning_rate": 1.6256239602094215e-06, "loss": 2.7357, "step": 44787 }, { "epoch": 2.7803091439567944, "grad_norm": 0.34924065736673043, "learning_rate": 1.6247106641980425e-06, "loss": 2.6259, "step": 44788 }, { "epoch": 2.7803712210565523, "grad_norm": 0.348479774288398, "learning_rate": 1.6237976205722816e-06, "loss": 2.7302, "step": 44789 }, { "epoch": 2.7804332981563102, "grad_norm": 0.3744739976012428, "learning_rate": 1.622884829336918e-06, "loss": 2.7042, "step": 44790 }, { "epoch": 2.780495375256068, "grad_norm": 0.33434298709675736, "learning_rate": 1.621972290496715e-06, "loss": 2.6584, "step": 44791 }, { "epoch": 2.780557452355826, "grad_norm": 0.35822225288729054, "learning_rate": 1.6210600040564183e-06, "loss": 2.6994, "step": 44792 }, { "epoch": 2.780619529455584, "grad_norm": 0.3621320493115359, "learning_rate": 1.6201479700207967e-06, "loss": 2.7164, "step": 44793 }, { "epoch": 2.7806816065553415, "grad_norm": 0.3468600166796364, "learning_rate": 1.6192361883946128e-06, "loss": 2.6744, "step": 44794 }, { "epoch": 2.7807436836551, "grad_norm": 0.36818043396670297, "learning_rate": 1.618324659182613e-06, "loss": 2.5424, "step": 44795 }, { "epoch": 2.7808057607548573, "grad_norm": 0.3744817172172745, "learning_rate": 1.6174133823895655e-06, "loss": 2.6703, "step": 44796 }, { "epoch": 2.7808678378546157, "grad_norm": 0.3935232978210607, "learning_rate": 1.6165023580202055e-06, "loss": 2.6919, "step": 44797 }, { "epoch": 2.780929914954373, "grad_norm": 0.34684431438638996, "learning_rate": 1.6155915860793069e-06, "loss": 2.6342, "step": 44798 }, { "epoch": 2.7809919920541315, "grad_norm": 0.3468329292422105, "learning_rate": 1.614681066571616e-06, "loss": 2.6331, "step": 44799 }, { "epoch": 2.781054069153889, "grad_norm": 0.35899347287542926, "learning_rate": 1.6137707995018735e-06, "loss": 2.6422, "step": 44800 }, { "epoch": 2.781116146253647, "grad_norm": 0.37385302214590355, "learning_rate": 1.6128607848748422e-06, "loss": 2.6343, "step": 44801 }, { "epoch": 2.781178223353405, "grad_norm": 0.36889878681910937, "learning_rate": 1.6119510226952516e-06, "loss": 2.6525, "step": 44802 }, { "epoch": 2.7812403004531627, "grad_norm": 0.38283040530311635, "learning_rate": 1.6110415129678647e-06, "loss": 2.7034, "step": 44803 }, { "epoch": 2.7813023775529206, "grad_norm": 0.3670658559387594, "learning_rate": 1.6101322556974218e-06, "loss": 2.6746, "step": 44804 }, { "epoch": 2.7813644546526786, "grad_norm": 0.38021496762623314, "learning_rate": 1.6092232508886695e-06, "loss": 2.7014, "step": 44805 }, { "epoch": 2.7814265317524365, "grad_norm": 0.35363247031057615, "learning_rate": 1.6083144985463373e-06, "loss": 2.7052, "step": 44806 }, { "epoch": 2.7814886088521944, "grad_norm": 0.36170819887134514, "learning_rate": 1.6074059986751765e-06, "loss": 2.5715, "step": 44807 }, { "epoch": 2.7815506859519523, "grad_norm": 0.3553912948229757, "learning_rate": 1.606497751279934e-06, "loss": 2.6629, "step": 44808 }, { "epoch": 2.7816127630517102, "grad_norm": 0.3756672168689299, "learning_rate": 1.605589756365339e-06, "loss": 2.6991, "step": 44809 }, { "epoch": 2.781674840151468, "grad_norm": 0.37787032338818693, "learning_rate": 1.6046820139361319e-06, "loss": 2.6721, "step": 44810 }, { "epoch": 2.781736917251226, "grad_norm": 0.3811007981628072, "learning_rate": 1.603774523997037e-06, "loss": 2.6957, "step": 44811 }, { "epoch": 2.781798994350984, "grad_norm": 0.3829193355410565, "learning_rate": 1.6028672865528117e-06, "loss": 2.6592, "step": 44812 }, { "epoch": 2.781861071450742, "grad_norm": 0.38560628307677564, "learning_rate": 1.6019603016081686e-06, "loss": 2.6099, "step": 44813 }, { "epoch": 2.7819231485505, "grad_norm": 0.36504148805974806, "learning_rate": 1.601053569167854e-06, "loss": 2.632, "step": 44814 }, { "epoch": 2.7819852256502577, "grad_norm": 0.3499267991186453, "learning_rate": 1.6001470892365866e-06, "loss": 2.6679, "step": 44815 }, { "epoch": 2.7820473027500157, "grad_norm": 0.3750016093219557, "learning_rate": 1.5992408618191014e-06, "loss": 2.7136, "step": 44816 }, { "epoch": 2.7821093798497736, "grad_norm": 0.3517285908322617, "learning_rate": 1.5983348869201276e-06, "loss": 2.6704, "step": 44817 }, { "epoch": 2.7821714569495315, "grad_norm": 0.3737808715866874, "learning_rate": 1.5974291645443951e-06, "loss": 2.6376, "step": 44818 }, { "epoch": 2.782233534049289, "grad_norm": 0.41128499166274873, "learning_rate": 1.5965236946966222e-06, "loss": 2.6911, "step": 44819 }, { "epoch": 2.7822956111490473, "grad_norm": 0.3571262909248688, "learning_rate": 1.5956184773815274e-06, "loss": 2.6388, "step": 44820 }, { "epoch": 2.782357688248805, "grad_norm": 0.3458877221488122, "learning_rate": 1.5947135126038515e-06, "loss": 2.6329, "step": 44821 }, { "epoch": 2.782419765348563, "grad_norm": 0.3691411094056842, "learning_rate": 1.5938088003683016e-06, "loss": 2.6626, "step": 44822 }, { "epoch": 2.7824818424483206, "grad_norm": 0.3722731270289363, "learning_rate": 1.5929043406796075e-06, "loss": 2.6629, "step": 44823 }, { "epoch": 2.7825439195480786, "grad_norm": 0.3765226683384685, "learning_rate": 1.592000133542476e-06, "loss": 2.6811, "step": 44824 }, { "epoch": 2.7826059966478365, "grad_norm": 0.37554157409672473, "learning_rate": 1.5910961789616319e-06, "loss": 2.6355, "step": 44825 }, { "epoch": 2.7826680737475944, "grad_norm": 0.3310375828107971, "learning_rate": 1.5901924769417931e-06, "loss": 2.6359, "step": 44826 }, { "epoch": 2.7827301508473523, "grad_norm": 0.36423011654039356, "learning_rate": 1.5892890274876725e-06, "loss": 2.5664, "step": 44827 }, { "epoch": 2.7827922279471102, "grad_norm": 0.3312715028585281, "learning_rate": 1.5883858306039834e-06, "loss": 2.6202, "step": 44828 }, { "epoch": 2.782854305046868, "grad_norm": 0.36214896091673937, "learning_rate": 1.5874828862954328e-06, "loss": 2.6229, "step": 44829 }, { "epoch": 2.782916382146626, "grad_norm": 0.37678905731260015, "learning_rate": 1.5865801945667446e-06, "loss": 2.7086, "step": 44830 }, { "epoch": 2.782978459246384, "grad_norm": 0.383681750643304, "learning_rate": 1.5856777554226154e-06, "loss": 2.6988, "step": 44831 }, { "epoch": 2.783040536346142, "grad_norm": 0.3557974165777087, "learning_rate": 1.5847755688677579e-06, "loss": 2.5715, "step": 44832 }, { "epoch": 2.7831026134459, "grad_norm": 0.3809817191003502, "learning_rate": 1.5838736349068795e-06, "loss": 2.6447, "step": 44833 }, { "epoch": 2.7831646905456577, "grad_norm": 0.3674800397356962, "learning_rate": 1.5829719535446874e-06, "loss": 2.669, "step": 44834 }, { "epoch": 2.7832267676454157, "grad_norm": 0.3816443740832377, "learning_rate": 1.5820705247858836e-06, "loss": 2.7005, "step": 44835 }, { "epoch": 2.7832888447451736, "grad_norm": 0.34409463469059254, "learning_rate": 1.5811693486351754e-06, "loss": 2.5927, "step": 44836 }, { "epoch": 2.7833509218449315, "grad_norm": 0.36990583922806564, "learning_rate": 1.5802684250972587e-06, "loss": 2.6481, "step": 44837 }, { "epoch": 2.7834129989446894, "grad_norm": 0.3702635463114237, "learning_rate": 1.5793677541768304e-06, "loss": 2.6718, "step": 44838 }, { "epoch": 2.7834750760444473, "grad_norm": 0.3361471542144884, "learning_rate": 1.5784673358785917e-06, "loss": 2.6544, "step": 44839 }, { "epoch": 2.7835371531442052, "grad_norm": 0.34853169232948106, "learning_rate": 1.5775671702072504e-06, "loss": 2.5827, "step": 44840 }, { "epoch": 2.783599230243963, "grad_norm": 0.34047594133741743, "learning_rate": 1.5766672571675022e-06, "loss": 2.5873, "step": 44841 }, { "epoch": 2.7836613073437206, "grad_norm": 0.3537599762362425, "learning_rate": 1.5757675967640273e-06, "loss": 2.5684, "step": 44842 }, { "epoch": 2.783723384443479, "grad_norm": 0.3765837287458986, "learning_rate": 1.5748681890015326e-06, "loss": 2.6647, "step": 44843 }, { "epoch": 2.7837854615432365, "grad_norm": 0.3534807008806031, "learning_rate": 1.5739690338847036e-06, "loss": 2.6058, "step": 44844 }, { "epoch": 2.783847538642995, "grad_norm": 0.34672263510729084, "learning_rate": 1.573070131418236e-06, "loss": 2.6571, "step": 44845 }, { "epoch": 2.7839096157427523, "grad_norm": 0.3380472669076819, "learning_rate": 1.5721714816068211e-06, "loss": 2.5913, "step": 44846 }, { "epoch": 2.7839716928425107, "grad_norm": 0.38256352950445116, "learning_rate": 1.5712730844551327e-06, "loss": 2.6434, "step": 44847 }, { "epoch": 2.784033769942268, "grad_norm": 0.395027735828877, "learning_rate": 1.570374939967878e-06, "loss": 2.6369, "step": 44848 }, { "epoch": 2.784095847042026, "grad_norm": 0.35463051947438795, "learning_rate": 1.5694770481497312e-06, "loss": 2.7052, "step": 44849 }, { "epoch": 2.784157924141784, "grad_norm": 0.35643356512908786, "learning_rate": 1.568579409005383e-06, "loss": 2.7008, "step": 44850 }, { "epoch": 2.784220001241542, "grad_norm": 0.35226431837281624, "learning_rate": 1.5676820225395128e-06, "loss": 2.684, "step": 44851 }, { "epoch": 2.7842820783413, "grad_norm": 0.37819392309122946, "learning_rate": 1.566784888756795e-06, "loss": 2.6259, "step": 44852 }, { "epoch": 2.7843441554410577, "grad_norm": 0.37010985998786305, "learning_rate": 1.5658880076619253e-06, "loss": 2.7, "step": 44853 }, { "epoch": 2.7844062325408157, "grad_norm": 0.3371935473015584, "learning_rate": 1.5649913792595783e-06, "loss": 2.5997, "step": 44854 }, { "epoch": 2.7844683096405736, "grad_norm": 0.3487889770890502, "learning_rate": 1.5640950035544277e-06, "loss": 2.6581, "step": 44855 }, { "epoch": 2.7845303867403315, "grad_norm": 0.3472299319312035, "learning_rate": 1.5631988805511422e-06, "loss": 2.7259, "step": 44856 }, { "epoch": 2.7845924638400894, "grad_norm": 0.3837276147470369, "learning_rate": 1.5623030102544122e-06, "loss": 2.645, "step": 44857 }, { "epoch": 2.7846545409398473, "grad_norm": 0.3766487553052694, "learning_rate": 1.5614073926689122e-06, "loss": 2.6895, "step": 44858 }, { "epoch": 2.7847166180396052, "grad_norm": 0.3588709821300082, "learning_rate": 1.5605120277993046e-06, "loss": 2.629, "step": 44859 }, { "epoch": 2.784778695139363, "grad_norm": 0.3924804679906674, "learning_rate": 1.5596169156502637e-06, "loss": 2.7104, "step": 44860 }, { "epoch": 2.784840772239121, "grad_norm": 0.35397622888167063, "learning_rate": 1.5587220562264582e-06, "loss": 2.5869, "step": 44861 }, { "epoch": 2.784902849338879, "grad_norm": 0.3375105291066384, "learning_rate": 1.5578274495325617e-06, "loss": 2.748, "step": 44862 }, { "epoch": 2.784964926438637, "grad_norm": 0.3710618657920626, "learning_rate": 1.556933095573243e-06, "loss": 2.6355, "step": 44863 }, { "epoch": 2.785027003538395, "grad_norm": 0.37082961871754827, "learning_rate": 1.5560389943531594e-06, "loss": 2.74, "step": 44864 }, { "epoch": 2.7850890806381527, "grad_norm": 0.3448626976183241, "learning_rate": 1.5551451458769795e-06, "loss": 2.6396, "step": 44865 }, { "epoch": 2.7851511577379107, "grad_norm": 0.364266607674377, "learning_rate": 1.5542515501493716e-06, "loss": 2.5853, "step": 44866 }, { "epoch": 2.785213234837668, "grad_norm": 0.37473025156808076, "learning_rate": 1.5533582071749931e-06, "loss": 2.537, "step": 44867 }, { "epoch": 2.7852753119374265, "grad_norm": 0.35147046897904, "learning_rate": 1.552465116958507e-06, "loss": 2.6938, "step": 44868 }, { "epoch": 2.785337389037184, "grad_norm": 0.3561458008617302, "learning_rate": 1.551572279504565e-06, "loss": 2.635, "step": 44869 }, { "epoch": 2.7853994661369423, "grad_norm": 0.36620768227719896, "learning_rate": 1.550679694817836e-06, "loss": 2.7169, "step": 44870 }, { "epoch": 2.7854615432367, "grad_norm": 0.346660796174601, "learning_rate": 1.5497873629029768e-06, "loss": 2.5941, "step": 44871 }, { "epoch": 2.7855236203364577, "grad_norm": 0.36519212529408157, "learning_rate": 1.5488952837646288e-06, "loss": 2.749, "step": 44872 }, { "epoch": 2.7855856974362156, "grad_norm": 0.3494789263766619, "learning_rate": 1.5480034574074654e-06, "loss": 2.6548, "step": 44873 }, { "epoch": 2.7856477745359736, "grad_norm": 0.34277299190829186, "learning_rate": 1.5471118838361276e-06, "loss": 2.663, "step": 44874 }, { "epoch": 2.7857098516357315, "grad_norm": 0.36122379343218236, "learning_rate": 1.5462205630552618e-06, "loss": 2.6917, "step": 44875 }, { "epoch": 2.7857719287354894, "grad_norm": 0.3420307340308733, "learning_rate": 1.5453294950695363e-06, "loss": 2.6117, "step": 44876 }, { "epoch": 2.7858340058352473, "grad_norm": 0.3443326022743381, "learning_rate": 1.5444386798835864e-06, "loss": 2.603, "step": 44877 }, { "epoch": 2.7858960829350052, "grad_norm": 0.3493551896608613, "learning_rate": 1.5435481175020638e-06, "loss": 2.6864, "step": 44878 }, { "epoch": 2.785958160034763, "grad_norm": 0.3547597458651965, "learning_rate": 1.5426578079296094e-06, "loss": 2.6574, "step": 44879 }, { "epoch": 2.786020237134521, "grad_norm": 0.3637531456041166, "learning_rate": 1.541767751170875e-06, "loss": 2.5985, "step": 44880 }, { "epoch": 2.786082314234279, "grad_norm": 0.33960514900521466, "learning_rate": 1.5408779472305013e-06, "loss": 2.649, "step": 44881 }, { "epoch": 2.786144391334037, "grad_norm": 0.34764587729763846, "learning_rate": 1.5399883961131345e-06, "loss": 2.7327, "step": 44882 }, { "epoch": 2.786206468433795, "grad_norm": 0.35848060518337765, "learning_rate": 1.5390990978234043e-06, "loss": 2.6049, "step": 44883 }, { "epoch": 2.7862685455335527, "grad_norm": 0.33983313883734867, "learning_rate": 1.5382100523659682e-06, "loss": 2.6088, "step": 44884 }, { "epoch": 2.7863306226333107, "grad_norm": 0.353215875578984, "learning_rate": 1.53732125974545e-06, "loss": 2.6492, "step": 44885 }, { "epoch": 2.7863926997330686, "grad_norm": 0.34090808295693364, "learning_rate": 1.5364327199664964e-06, "loss": 2.5669, "step": 44886 }, { "epoch": 2.7864547768328265, "grad_norm": 0.383488389285811, "learning_rate": 1.5355444330337366e-06, "loss": 2.7206, "step": 44887 }, { "epoch": 2.7865168539325844, "grad_norm": 0.36714137578983985, "learning_rate": 1.5346563989518005e-06, "loss": 2.728, "step": 44888 }, { "epoch": 2.7865789310323423, "grad_norm": 0.3494655803573178, "learning_rate": 1.5337686177253285e-06, "loss": 2.751, "step": 44889 }, { "epoch": 2.7866410081321, "grad_norm": 0.33158224825423077, "learning_rate": 1.5328810893589563e-06, "loss": 2.6261, "step": 44890 }, { "epoch": 2.786703085231858, "grad_norm": 0.38012052419097864, "learning_rate": 1.5319938138573075e-06, "loss": 2.7115, "step": 44891 }, { "epoch": 2.7867651623316156, "grad_norm": 0.3591784893636131, "learning_rate": 1.5311067912250065e-06, "loss": 2.6832, "step": 44892 }, { "epoch": 2.786827239431374, "grad_norm": 0.3402199143184565, "learning_rate": 1.5302200214666996e-06, "loss": 2.6038, "step": 44893 }, { "epoch": 2.7868893165311315, "grad_norm": 0.35437369139495306, "learning_rate": 1.529333504586994e-06, "loss": 2.678, "step": 44894 }, { "epoch": 2.78695139363089, "grad_norm": 0.331165463978779, "learning_rate": 1.5284472405905248e-06, "loss": 2.6664, "step": 44895 }, { "epoch": 2.7870134707306473, "grad_norm": 0.33966327128118684, "learning_rate": 1.5275612294819053e-06, "loss": 2.6356, "step": 44896 }, { "epoch": 2.7870755478304052, "grad_norm": 0.34990081147974444, "learning_rate": 1.526675471265776e-06, "loss": 2.6408, "step": 44897 }, { "epoch": 2.787137624930163, "grad_norm": 0.35525932011334643, "learning_rate": 1.5257899659467446e-06, "loss": 2.7528, "step": 44898 }, { "epoch": 2.787199702029921, "grad_norm": 0.35722126568370705, "learning_rate": 1.5249047135294348e-06, "loss": 2.6389, "step": 44899 }, { "epoch": 2.787261779129679, "grad_norm": 0.3432281607968647, "learning_rate": 1.5240197140184654e-06, "loss": 2.6674, "step": 44900 }, { "epoch": 2.787323856229437, "grad_norm": 0.3652988315763044, "learning_rate": 1.5231349674184493e-06, "loss": 2.6873, "step": 44901 }, { "epoch": 2.787385933329195, "grad_norm": 0.35520180960040676, "learning_rate": 1.5222504737340104e-06, "loss": 2.7148, "step": 44902 }, { "epoch": 2.7874480104289527, "grad_norm": 0.3561737280059226, "learning_rate": 1.521366232969762e-06, "loss": 2.6054, "step": 44903 }, { "epoch": 2.7875100875287107, "grad_norm": 0.3692411860987606, "learning_rate": 1.5204822451303113e-06, "loss": 2.6462, "step": 44904 }, { "epoch": 2.7875721646284686, "grad_norm": 0.3731531121735646, "learning_rate": 1.519598510220266e-06, "loss": 2.6036, "step": 44905 }, { "epoch": 2.7876342417282265, "grad_norm": 0.35046348452173465, "learning_rate": 1.5187150282442441e-06, "loss": 2.6261, "step": 44906 }, { "epoch": 2.7876963188279844, "grad_norm": 0.33530566212987023, "learning_rate": 1.5178317992068647e-06, "loss": 2.7265, "step": 44907 }, { "epoch": 2.7877583959277423, "grad_norm": 0.3665295041804009, "learning_rate": 1.5169488231127238e-06, "loss": 2.675, "step": 44908 }, { "epoch": 2.7878204730275002, "grad_norm": 0.3846049287869505, "learning_rate": 1.5160660999664345e-06, "loss": 2.7058, "step": 44909 }, { "epoch": 2.787882550127258, "grad_norm": 0.3855649903282392, "learning_rate": 1.515183629772593e-06, "loss": 2.6954, "step": 44910 }, { "epoch": 2.787944627227016, "grad_norm": 0.3740017638139391, "learning_rate": 1.5143014125358069e-06, "loss": 2.7215, "step": 44911 }, { "epoch": 2.788006704326774, "grad_norm": 0.38393166491697467, "learning_rate": 1.5134194482606833e-06, "loss": 2.7289, "step": 44912 }, { "epoch": 2.788068781426532, "grad_norm": 0.35914517392922013, "learning_rate": 1.5125377369518245e-06, "loss": 2.6892, "step": 44913 }, { "epoch": 2.78813085852629, "grad_norm": 0.3573493258529545, "learning_rate": 1.5116562786138211e-06, "loss": 2.6787, "step": 44914 }, { "epoch": 2.7881929356260473, "grad_norm": 0.35461545540218387, "learning_rate": 1.5107750732512805e-06, "loss": 2.6706, "step": 44915 }, { "epoch": 2.7882550127258057, "grad_norm": 0.38212553003790384, "learning_rate": 1.509894120868799e-06, "loss": 2.6932, "step": 44916 }, { "epoch": 2.788317089825563, "grad_norm": 0.35679256298783135, "learning_rate": 1.5090134214709672e-06, "loss": 2.6258, "step": 44917 }, { "epoch": 2.7883791669253215, "grad_norm": 0.36090108018905315, "learning_rate": 1.5081329750623875e-06, "loss": 2.72, "step": 44918 }, { "epoch": 2.788441244025079, "grad_norm": 0.35457990417585467, "learning_rate": 1.5072527816476444e-06, "loss": 2.6748, "step": 44919 }, { "epoch": 2.788503321124837, "grad_norm": 0.3546832491918258, "learning_rate": 1.5063728412313405e-06, "loss": 2.7262, "step": 44920 }, { "epoch": 2.788565398224595, "grad_norm": 0.35302093837725596, "learning_rate": 1.5054931538180662e-06, "loss": 2.6958, "step": 44921 }, { "epoch": 2.7886274753243527, "grad_norm": 0.3498265428387619, "learning_rate": 1.5046137194124011e-06, "loss": 2.5505, "step": 44922 }, { "epoch": 2.7886895524241107, "grad_norm": 0.3605314183254201, "learning_rate": 1.5037345380189416e-06, "loss": 2.678, "step": 44923 }, { "epoch": 2.7887516295238686, "grad_norm": 0.3618707036964076, "learning_rate": 1.502855609642262e-06, "loss": 2.6317, "step": 44924 }, { "epoch": 2.7888137066236265, "grad_norm": 0.3879442160657897, "learning_rate": 1.5019769342869694e-06, "loss": 2.7126, "step": 44925 }, { "epoch": 2.7888757837233844, "grad_norm": 0.3443289995711641, "learning_rate": 1.5010985119576327e-06, "loss": 2.6345, "step": 44926 }, { "epoch": 2.7889378608231423, "grad_norm": 0.3658930708694339, "learning_rate": 1.5002203426588423e-06, "loss": 2.6901, "step": 44927 }, { "epoch": 2.7889999379229002, "grad_norm": 0.3330101393453348, "learning_rate": 1.4993424263951673e-06, "loss": 2.6734, "step": 44928 }, { "epoch": 2.789062015022658, "grad_norm": 0.3807011789279597, "learning_rate": 1.4984647631712035e-06, "loss": 2.6407, "step": 44929 }, { "epoch": 2.789124092122416, "grad_norm": 0.35148385545802846, "learning_rate": 1.4975873529915195e-06, "loss": 2.6925, "step": 44930 }, { "epoch": 2.789186169222174, "grad_norm": 0.366174050062787, "learning_rate": 1.4967101958607011e-06, "loss": 2.6589, "step": 44931 }, { "epoch": 2.789248246321932, "grad_norm": 0.36670183975766246, "learning_rate": 1.4958332917833108e-06, "loss": 2.649, "step": 44932 }, { "epoch": 2.78931032342169, "grad_norm": 0.35574129157489165, "learning_rate": 1.4949566407639449e-06, "loss": 2.5661, "step": 44933 }, { "epoch": 2.7893724005214477, "grad_norm": 0.3367619267725534, "learning_rate": 1.494080242807161e-06, "loss": 2.6216, "step": 44934 }, { "epoch": 2.7894344776212057, "grad_norm": 0.3614782897687655, "learning_rate": 1.4932040979175333e-06, "loss": 2.684, "step": 44935 }, { "epoch": 2.7894965547209636, "grad_norm": 0.34840082950671214, "learning_rate": 1.4923282060996413e-06, "loss": 2.7177, "step": 44936 }, { "epoch": 2.7895586318207215, "grad_norm": 0.3497802274278866, "learning_rate": 1.491452567358037e-06, "loss": 2.6551, "step": 44937 }, { "epoch": 2.789620708920479, "grad_norm": 0.36021284209048277, "learning_rate": 1.4905771816973058e-06, "loss": 2.6821, "step": 44938 }, { "epoch": 2.7896827860202373, "grad_norm": 0.3508845574348299, "learning_rate": 1.4897020491220105e-06, "loss": 2.7243, "step": 44939 }, { "epoch": 2.789744863119995, "grad_norm": 0.3574966394159782, "learning_rate": 1.4888271696367196e-06, "loss": 2.7285, "step": 44940 }, { "epoch": 2.789806940219753, "grad_norm": 0.35989191363234, "learning_rate": 1.4879525432459907e-06, "loss": 2.6163, "step": 44941 }, { "epoch": 2.7898690173195106, "grad_norm": 0.3603034258689919, "learning_rate": 1.4870781699543868e-06, "loss": 2.7217, "step": 44942 }, { "epoch": 2.789931094419269, "grad_norm": 0.358109089729305, "learning_rate": 1.4862040497664763e-06, "loss": 2.7512, "step": 44943 }, { "epoch": 2.7899931715190265, "grad_norm": 0.3560436755563227, "learning_rate": 1.485330182686817e-06, "loss": 2.6559, "step": 44944 }, { "epoch": 2.7900552486187844, "grad_norm": 0.3533615176844148, "learning_rate": 1.4844565687199718e-06, "loss": 2.6696, "step": 44945 }, { "epoch": 2.7901173257185423, "grad_norm": 0.3558713707854402, "learning_rate": 1.483583207870487e-06, "loss": 2.7232, "step": 44946 }, { "epoch": 2.7901794028183002, "grad_norm": 0.3458564547580702, "learning_rate": 1.4827101001429256e-06, "loss": 2.5561, "step": 44947 }, { "epoch": 2.790241479918058, "grad_norm": 0.3485482912161801, "learning_rate": 1.4818372455418506e-06, "loss": 2.704, "step": 44948 }, { "epoch": 2.790303557017816, "grad_norm": 0.3518726781934496, "learning_rate": 1.4809646440718027e-06, "loss": 2.6835, "step": 44949 }, { "epoch": 2.790365634117574, "grad_norm": 0.3635793098955988, "learning_rate": 1.4800922957373453e-06, "loss": 2.6689, "step": 44950 }, { "epoch": 2.790427711217332, "grad_norm": 0.34518322401905327, "learning_rate": 1.4792202005430188e-06, "loss": 2.6151, "step": 44951 }, { "epoch": 2.79048978831709, "grad_norm": 0.3563603548153622, "learning_rate": 1.4783483584933865e-06, "loss": 2.6766, "step": 44952 }, { "epoch": 2.7905518654168477, "grad_norm": 0.3527143893066534, "learning_rate": 1.4774767695929893e-06, "loss": 2.5904, "step": 44953 }, { "epoch": 2.7906139425166057, "grad_norm": 0.3789292554655279, "learning_rate": 1.476605433846373e-06, "loss": 2.6325, "step": 44954 }, { "epoch": 2.7906760196163636, "grad_norm": 0.3500906865105842, "learning_rate": 1.475734351258079e-06, "loss": 2.7458, "step": 44955 }, { "epoch": 2.7907380967161215, "grad_norm": 0.3291128252858514, "learning_rate": 1.47486352183267e-06, "loss": 2.6277, "step": 44956 }, { "epoch": 2.7908001738158794, "grad_norm": 0.35387442489922244, "learning_rate": 1.4739929455746705e-06, "loss": 2.7492, "step": 44957 }, { "epoch": 2.7908622509156373, "grad_norm": 0.38660340563768464, "learning_rate": 1.473122622488632e-06, "loss": 2.6756, "step": 44958 }, { "epoch": 2.7909243280153953, "grad_norm": 0.3502386173805687, "learning_rate": 1.4722525525790954e-06, "loss": 2.6714, "step": 44959 }, { "epoch": 2.790986405115153, "grad_norm": 0.3401901409068388, "learning_rate": 1.4713827358505904e-06, "loss": 2.6444, "step": 44960 }, { "epoch": 2.7910484822149106, "grad_norm": 0.3639538189339265, "learning_rate": 1.4705131723076693e-06, "loss": 2.6319, "step": 44961 }, { "epoch": 2.791110559314669, "grad_norm": 0.325418058065565, "learning_rate": 1.4696438619548614e-06, "loss": 2.68, "step": 44962 }, { "epoch": 2.7911726364144265, "grad_norm": 0.3420327598778939, "learning_rate": 1.4687748047967021e-06, "loss": 2.5519, "step": 44963 }, { "epoch": 2.791234713514185, "grad_norm": 0.37977439686607595, "learning_rate": 1.467906000837721e-06, "loss": 2.6946, "step": 44964 }, { "epoch": 2.7912967906139423, "grad_norm": 0.3339232227044239, "learning_rate": 1.4670374500824646e-06, "loss": 2.6862, "step": 44965 }, { "epoch": 2.7913588677137007, "grad_norm": 0.379446922033204, "learning_rate": 1.4661691525354514e-06, "loss": 2.6709, "step": 44966 }, { "epoch": 2.791420944813458, "grad_norm": 0.38834676132330026, "learning_rate": 1.4653011082012169e-06, "loss": 2.6745, "step": 44967 }, { "epoch": 2.791483021913216, "grad_norm": 0.3420950760415365, "learning_rate": 1.464433317084285e-06, "loss": 2.6731, "step": 44968 }, { "epoch": 2.791545099012974, "grad_norm": 0.3665669249570785, "learning_rate": 1.4635657791891856e-06, "loss": 2.6697, "step": 44969 }, { "epoch": 2.791607176112732, "grad_norm": 0.35402739348269524, "learning_rate": 1.462698494520448e-06, "loss": 2.6446, "step": 44970 }, { "epoch": 2.79166925321249, "grad_norm": 0.3433036181733144, "learning_rate": 1.461831463082597e-06, "loss": 2.5925, "step": 44971 }, { "epoch": 2.7917313303122477, "grad_norm": 0.3626636053717544, "learning_rate": 1.4609646848801562e-06, "loss": 2.6338, "step": 44972 }, { "epoch": 2.7917934074120057, "grad_norm": 0.35260313513796027, "learning_rate": 1.4600981599176444e-06, "loss": 2.6962, "step": 44973 }, { "epoch": 2.7918554845117636, "grad_norm": 0.3844131559336644, "learning_rate": 1.4592318881995804e-06, "loss": 2.6505, "step": 44974 }, { "epoch": 2.7919175616115215, "grad_norm": 0.32943743537449155, "learning_rate": 1.4583658697304935e-06, "loss": 2.6749, "step": 44975 }, { "epoch": 2.7919796387112794, "grad_norm": 0.36641951660995326, "learning_rate": 1.4575001045148918e-06, "loss": 2.6552, "step": 44976 }, { "epoch": 2.7920417158110373, "grad_norm": 0.3680030612377675, "learning_rate": 1.4566345925572988e-06, "loss": 2.6918, "step": 44977 }, { "epoch": 2.7921037929107952, "grad_norm": 0.366615031771795, "learning_rate": 1.4557693338622225e-06, "loss": 2.7412, "step": 44978 }, { "epoch": 2.792165870010553, "grad_norm": 0.36051779946975204, "learning_rate": 1.4549043284341868e-06, "loss": 2.581, "step": 44979 }, { "epoch": 2.792227947110311, "grad_norm": 0.371281927482681, "learning_rate": 1.4540395762776993e-06, "loss": 2.7263, "step": 44980 }, { "epoch": 2.792290024210069, "grad_norm": 0.3721605526884174, "learning_rate": 1.453175077397273e-06, "loss": 2.6778, "step": 44981 }, { "epoch": 2.792352101309827, "grad_norm": 0.3617859697603274, "learning_rate": 1.4523108317974098e-06, "loss": 2.6908, "step": 44982 }, { "epoch": 2.792414178409585, "grad_norm": 0.3604070522518055, "learning_rate": 1.4514468394826342e-06, "loss": 2.7214, "step": 44983 }, { "epoch": 2.7924762555093428, "grad_norm": 0.37942221993406045, "learning_rate": 1.4505831004574478e-06, "loss": 2.6697, "step": 44984 }, { "epoch": 2.7925383326091007, "grad_norm": 0.3918110959941702, "learning_rate": 1.4497196147263526e-06, "loss": 2.8155, "step": 44985 }, { "epoch": 2.792600409708858, "grad_norm": 0.35580744694228006, "learning_rate": 1.448856382293856e-06, "loss": 2.6118, "step": 44986 }, { "epoch": 2.7926624868086165, "grad_norm": 0.3775476501088095, "learning_rate": 1.4479934031644549e-06, "loss": 2.6672, "step": 44987 }, { "epoch": 2.792724563908374, "grad_norm": 0.3574373210633723, "learning_rate": 1.4471306773426675e-06, "loss": 2.6313, "step": 44988 }, { "epoch": 2.7927866410081323, "grad_norm": 0.37793270376022015, "learning_rate": 1.4462682048329846e-06, "loss": 2.6994, "step": 44989 }, { "epoch": 2.79284871810789, "grad_norm": 0.37671238618947656, "learning_rate": 1.4454059856399082e-06, "loss": 2.6859, "step": 44990 }, { "epoch": 2.7929107952076477, "grad_norm": 0.37480683119868, "learning_rate": 1.4445440197679294e-06, "loss": 2.6638, "step": 44991 }, { "epoch": 2.7929728723074057, "grad_norm": 0.3572795146658762, "learning_rate": 1.4436823072215556e-06, "loss": 2.7128, "step": 44992 }, { "epoch": 2.7930349494071636, "grad_norm": 0.3500967943508121, "learning_rate": 1.4428208480052774e-06, "loss": 2.7312, "step": 44993 }, { "epoch": 2.7930970265069215, "grad_norm": 0.3593003361305705, "learning_rate": 1.441959642123597e-06, "loss": 2.6705, "step": 44994 }, { "epoch": 2.7931591036066794, "grad_norm": 0.3609534305508301, "learning_rate": 1.4410986895809942e-06, "loss": 2.6832, "step": 44995 }, { "epoch": 2.7932211807064373, "grad_norm": 0.3488443518235906, "learning_rate": 1.4402379903819652e-06, "loss": 2.7168, "step": 44996 }, { "epoch": 2.7932832578061952, "grad_norm": 0.3450717656501964, "learning_rate": 1.439377544531012e-06, "loss": 2.6925, "step": 44997 }, { "epoch": 2.793345334905953, "grad_norm": 0.3621109395574619, "learning_rate": 1.438517352032609e-06, "loss": 2.7217, "step": 44998 }, { "epoch": 2.793407412005711, "grad_norm": 0.3749079790061973, "learning_rate": 1.4376574128912524e-06, "loss": 2.5396, "step": 44999 }, { "epoch": 2.793469489105469, "grad_norm": 0.3415040969302676, "learning_rate": 1.436797727111422e-06, "loss": 2.5762, "step": 45000 }, { "epoch": 2.793531566205227, "grad_norm": 0.3602328841625927, "learning_rate": 1.4359382946976086e-06, "loss": 2.6749, "step": 45001 }, { "epoch": 2.793593643304985, "grad_norm": 0.34149346099753597, "learning_rate": 1.4350791156542976e-06, "loss": 2.6875, "step": 45002 }, { "epoch": 2.7936557204047427, "grad_norm": 0.35615536118681235, "learning_rate": 1.4342201899859576e-06, "loss": 2.6302, "step": 45003 }, { "epoch": 2.7937177975045007, "grad_norm": 0.3446063407499735, "learning_rate": 1.4333615176970905e-06, "loss": 2.6819, "step": 45004 }, { "epoch": 2.7937798746042586, "grad_norm": 0.362256357923042, "learning_rate": 1.4325030987921595e-06, "loss": 2.7152, "step": 45005 }, { "epoch": 2.7938419517040165, "grad_norm": 0.3340672284247516, "learning_rate": 1.431644933275661e-06, "loss": 2.7613, "step": 45006 }, { "epoch": 2.7939040288037744, "grad_norm": 0.36186648291281204, "learning_rate": 1.4307870211520525e-06, "loss": 2.6483, "step": 45007 }, { "epoch": 2.7939661059035323, "grad_norm": 0.34870884170492134, "learning_rate": 1.4299293624258247e-06, "loss": 2.7289, "step": 45008 }, { "epoch": 2.79402818300329, "grad_norm": 0.34815987530423176, "learning_rate": 1.4290719571014411e-06, "loss": 2.7345, "step": 45009 }, { "epoch": 2.794090260103048, "grad_norm": 0.3407447643766406, "learning_rate": 1.4282148051833811e-06, "loss": 2.7014, "step": 45010 }, { "epoch": 2.7941523372028056, "grad_norm": 0.3265845149892377, "learning_rate": 1.427357906676119e-06, "loss": 2.6575, "step": 45011 }, { "epoch": 2.794214414302564, "grad_norm": 0.36476850121647486, "learning_rate": 1.4265012615841178e-06, "loss": 2.6116, "step": 45012 }, { "epoch": 2.7942764914023215, "grad_norm": 0.36210873798203225, "learning_rate": 1.4256448699118574e-06, "loss": 2.6726, "step": 45013 }, { "epoch": 2.79433856850208, "grad_norm": 0.3652661355636729, "learning_rate": 1.4247887316637899e-06, "loss": 2.6684, "step": 45014 }, { "epoch": 2.7944006456018373, "grad_norm": 0.3469684419126396, "learning_rate": 1.4239328468444001e-06, "loss": 2.6927, "step": 45015 }, { "epoch": 2.7944627227015952, "grad_norm": 0.3529486984089506, "learning_rate": 1.4230772154581407e-06, "loss": 2.6269, "step": 45016 }, { "epoch": 2.794524799801353, "grad_norm": 0.3336227868908242, "learning_rate": 1.4222218375094798e-06, "loss": 2.6518, "step": 45017 }, { "epoch": 2.794586876901111, "grad_norm": 0.3508323080035802, "learning_rate": 1.4213667130028807e-06, "loss": 2.6669, "step": 45018 }, { "epoch": 2.794648954000869, "grad_norm": 0.36695980560543334, "learning_rate": 1.4205118419428066e-06, "loss": 2.642, "step": 45019 }, { "epoch": 2.794711031100627, "grad_norm": 0.3688198088177244, "learning_rate": 1.4196572243337147e-06, "loss": 2.6247, "step": 45020 }, { "epoch": 2.794773108200385, "grad_norm": 0.3485944603947369, "learning_rate": 1.418802860180063e-06, "loss": 2.7122, "step": 45021 }, { "epoch": 2.7948351853001427, "grad_norm": 0.3490920048045917, "learning_rate": 1.4179487494863141e-06, "loss": 2.6342, "step": 45022 }, { "epoch": 2.7948972623999007, "grad_norm": 0.36345463378614074, "learning_rate": 1.4170948922569094e-06, "loss": 2.5881, "step": 45023 }, { "epoch": 2.7949593394996586, "grad_norm": 0.3557511559760479, "learning_rate": 1.4162412884963227e-06, "loss": 2.6284, "step": 45024 }, { "epoch": 2.7950214165994165, "grad_norm": 0.35660693541194044, "learning_rate": 1.4153879382090008e-06, "loss": 2.6479, "step": 45025 }, { "epoch": 2.7950834936991744, "grad_norm": 0.36991020997939295, "learning_rate": 1.4145348413993953e-06, "loss": 2.6949, "step": 45026 }, { "epoch": 2.7951455707989323, "grad_norm": 0.3507082202564184, "learning_rate": 1.4136819980719474e-06, "loss": 2.7145, "step": 45027 }, { "epoch": 2.7952076478986903, "grad_norm": 0.37356972129317556, "learning_rate": 1.4128294082311254e-06, "loss": 2.7199, "step": 45028 }, { "epoch": 2.795269724998448, "grad_norm": 0.3608668088838922, "learning_rate": 1.4119770718813597e-06, "loss": 2.6486, "step": 45029 }, { "epoch": 2.795331802098206, "grad_norm": 0.36960344718222454, "learning_rate": 1.4111249890271127e-06, "loss": 2.6589, "step": 45030 }, { "epoch": 2.795393879197964, "grad_norm": 0.3631457619632606, "learning_rate": 1.4102731596728148e-06, "loss": 2.6319, "step": 45031 }, { "epoch": 2.795455956297722, "grad_norm": 0.36972975772337463, "learning_rate": 1.4094215838229176e-06, "loss": 2.6075, "step": 45032 }, { "epoch": 2.79551803339748, "grad_norm": 0.38188397429764337, "learning_rate": 1.4085702614818675e-06, "loss": 2.6905, "step": 45033 }, { "epoch": 2.7955801104972373, "grad_norm": 0.35049059960891804, "learning_rate": 1.4077191926541e-06, "loss": 2.7546, "step": 45034 }, { "epoch": 2.7956421875969957, "grad_norm": 0.3494987525890394, "learning_rate": 1.4068683773440616e-06, "loss": 2.7088, "step": 45035 }, { "epoch": 2.795704264696753, "grad_norm": 0.3887569966867188, "learning_rate": 1.4060178155561766e-06, "loss": 2.6976, "step": 45036 }, { "epoch": 2.7957663417965115, "grad_norm": 0.36788454246393726, "learning_rate": 1.4051675072948967e-06, "loss": 2.6913, "step": 45037 }, { "epoch": 2.795828418896269, "grad_norm": 0.3581162466940059, "learning_rate": 1.404317452564663e-06, "loss": 2.6508, "step": 45038 }, { "epoch": 2.795890495996027, "grad_norm": 0.339222965855628, "learning_rate": 1.4034676513698996e-06, "loss": 2.7005, "step": 45039 }, { "epoch": 2.795952573095785, "grad_norm": 0.3608075283586195, "learning_rate": 1.4026181037150422e-06, "loss": 2.6382, "step": 45040 }, { "epoch": 2.7960146501955427, "grad_norm": 0.3951794426787577, "learning_rate": 1.4017688096045146e-06, "loss": 2.653, "step": 45041 }, { "epoch": 2.7960767272953007, "grad_norm": 0.37322610274489326, "learning_rate": 1.4009197690427688e-06, "loss": 2.6254, "step": 45042 }, { "epoch": 2.7961388043950586, "grad_norm": 0.3589758522258955, "learning_rate": 1.4000709820342184e-06, "loss": 2.6127, "step": 45043 }, { "epoch": 2.7962008814948165, "grad_norm": 0.3646602095433842, "learning_rate": 1.3992224485832982e-06, "loss": 2.5581, "step": 45044 }, { "epoch": 2.7962629585945744, "grad_norm": 0.34528833494600214, "learning_rate": 1.3983741686944274e-06, "loss": 2.657, "step": 45045 }, { "epoch": 2.7963250356943323, "grad_norm": 0.34704125046007583, "learning_rate": 1.3975261423720354e-06, "loss": 2.6124, "step": 45046 }, { "epoch": 2.7963871127940902, "grad_norm": 0.3480108039051714, "learning_rate": 1.3966783696205576e-06, "loss": 2.6207, "step": 45047 }, { "epoch": 2.796449189893848, "grad_norm": 0.3770121987656023, "learning_rate": 1.3958308504444017e-06, "loss": 2.7142, "step": 45048 }, { "epoch": 2.796511266993606, "grad_norm": 0.37790975592048515, "learning_rate": 1.3949835848479974e-06, "loss": 2.6092, "step": 45049 }, { "epoch": 2.796573344093364, "grad_norm": 0.3727729350772891, "learning_rate": 1.3941365728357581e-06, "loss": 2.6836, "step": 45050 }, { "epoch": 2.796635421193122, "grad_norm": 0.37432020878620686, "learning_rate": 1.393289814412113e-06, "loss": 2.7011, "step": 45051 }, { "epoch": 2.79669749829288, "grad_norm": 0.35579448489374377, "learning_rate": 1.3924433095814704e-06, "loss": 2.6818, "step": 45052 }, { "epoch": 2.7967595753926378, "grad_norm": 0.3424006986760761, "learning_rate": 1.3915970583482596e-06, "loss": 2.6314, "step": 45053 }, { "epoch": 2.7968216524923957, "grad_norm": 0.34730988296640375, "learning_rate": 1.3907510607168717e-06, "loss": 2.7186, "step": 45054 }, { "epoch": 2.7968837295921536, "grad_norm": 0.3720898560525036, "learning_rate": 1.3899053166917475e-06, "loss": 2.7373, "step": 45055 }, { "epoch": 2.7969458066919115, "grad_norm": 0.36829260648778767, "learning_rate": 1.389059826277289e-06, "loss": 2.6206, "step": 45056 }, { "epoch": 2.797007883791669, "grad_norm": 0.3392724683513525, "learning_rate": 1.3882145894779042e-06, "loss": 2.6633, "step": 45057 }, { "epoch": 2.7970699608914273, "grad_norm": 0.3945164630498298, "learning_rate": 1.387369606298e-06, "loss": 2.6947, "step": 45058 }, { "epoch": 2.797132037991185, "grad_norm": 0.34878576220090507, "learning_rate": 1.3865248767419903e-06, "loss": 2.6078, "step": 45059 }, { "epoch": 2.797194115090943, "grad_norm": 0.35000222792938124, "learning_rate": 1.3856804008142877e-06, "loss": 2.7047, "step": 45060 }, { "epoch": 2.7972561921907007, "grad_norm": 0.3699382460132778, "learning_rate": 1.3848361785192888e-06, "loss": 2.6771, "step": 45061 }, { "epoch": 2.797318269290459, "grad_norm": 0.36490371342037325, "learning_rate": 1.3839922098614012e-06, "loss": 2.637, "step": 45062 }, { "epoch": 2.7973803463902165, "grad_norm": 0.32563259749234863, "learning_rate": 1.3831484948450213e-06, "loss": 2.6045, "step": 45063 }, { "epoch": 2.7974424234899744, "grad_norm": 0.3657729314210964, "learning_rate": 1.382305033474568e-06, "loss": 2.6658, "step": 45064 }, { "epoch": 2.7975045005897323, "grad_norm": 0.3558643780386982, "learning_rate": 1.3814618257544266e-06, "loss": 2.7165, "step": 45065 }, { "epoch": 2.7975665776894902, "grad_norm": 0.36314816241822956, "learning_rate": 1.3806188716890045e-06, "loss": 2.6587, "step": 45066 }, { "epoch": 2.797628654789248, "grad_norm": 0.3378791807143953, "learning_rate": 1.379776171282693e-06, "loss": 2.6393, "step": 45067 }, { "epoch": 2.797690731889006, "grad_norm": 0.36596241974455634, "learning_rate": 1.3789337245398937e-06, "loss": 2.6644, "step": 45068 }, { "epoch": 2.797752808988764, "grad_norm": 0.359128452814209, "learning_rate": 1.3780915314649979e-06, "loss": 2.6737, "step": 45069 }, { "epoch": 2.797814886088522, "grad_norm": 0.3427274841631506, "learning_rate": 1.3772495920624073e-06, "loss": 2.6401, "step": 45070 }, { "epoch": 2.79787696318828, "grad_norm": 0.3391752244637478, "learning_rate": 1.3764079063365077e-06, "loss": 2.5751, "step": 45071 }, { "epoch": 2.7979390402880377, "grad_norm": 0.34957983648728364, "learning_rate": 1.3755664742916897e-06, "loss": 2.6883, "step": 45072 }, { "epoch": 2.7980011173877957, "grad_norm": 0.3922203103433293, "learning_rate": 1.3747252959323443e-06, "loss": 2.7014, "step": 45073 }, { "epoch": 2.7980631944875536, "grad_norm": 0.3530959172214826, "learning_rate": 1.3738843712628623e-06, "loss": 2.5943, "step": 45074 }, { "epoch": 2.7981252715873115, "grad_norm": 0.35540458603139624, "learning_rate": 1.373043700287635e-06, "loss": 2.6599, "step": 45075 }, { "epoch": 2.7981873486870694, "grad_norm": 0.3484525348002678, "learning_rate": 1.372203283011042e-06, "loss": 2.7003, "step": 45076 }, { "epoch": 2.7982494257868273, "grad_norm": 0.3473724750098959, "learning_rate": 1.3713631194374576e-06, "loss": 2.7835, "step": 45077 }, { "epoch": 2.7983115028865853, "grad_norm": 0.3522253040410964, "learning_rate": 1.3705232095712895e-06, "loss": 2.6561, "step": 45078 }, { "epoch": 2.798373579986343, "grad_norm": 0.3765907918028676, "learning_rate": 1.3696835534169062e-06, "loss": 2.8123, "step": 45079 }, { "epoch": 2.798435657086101, "grad_norm": 0.33663398047365894, "learning_rate": 1.3688441509786875e-06, "loss": 2.7258, "step": 45080 }, { "epoch": 2.798497734185859, "grad_norm": 0.37334851357912685, "learning_rate": 1.3680050022610136e-06, "loss": 2.604, "step": 45081 }, { "epoch": 2.7985598112856165, "grad_norm": 0.3480621604260189, "learning_rate": 1.3671661072682585e-06, "loss": 2.6473, "step": 45082 }, { "epoch": 2.798621888385375, "grad_norm": 0.34631547735669604, "learning_rate": 1.3663274660048076e-06, "loss": 2.6431, "step": 45083 }, { "epoch": 2.7986839654851323, "grad_norm": 0.3454420321453487, "learning_rate": 1.3654890784750352e-06, "loss": 2.7729, "step": 45084 }, { "epoch": 2.7987460425848907, "grad_norm": 0.37380512928165743, "learning_rate": 1.3646509446833156e-06, "loss": 2.6992, "step": 45085 }, { "epoch": 2.798808119684648, "grad_norm": 0.3444376915779059, "learning_rate": 1.3638130646340064e-06, "loss": 2.7934, "step": 45086 }, { "epoch": 2.798870196784406, "grad_norm": 0.36487441250609054, "learning_rate": 1.362975438331504e-06, "loss": 2.6916, "step": 45087 }, { "epoch": 2.798932273884164, "grad_norm": 0.37346819109073287, "learning_rate": 1.3621380657801607e-06, "loss": 2.5841, "step": 45088 }, { "epoch": 2.798994350983922, "grad_norm": 0.35078057117014155, "learning_rate": 1.3613009469843508e-06, "loss": 2.6736, "step": 45089 }, { "epoch": 2.79905642808368, "grad_norm": 0.3630439434421586, "learning_rate": 1.3604640819484372e-06, "loss": 2.6757, "step": 45090 }, { "epoch": 2.7991185051834377, "grad_norm": 0.3540178177829584, "learning_rate": 1.3596274706767941e-06, "loss": 2.6377, "step": 45091 }, { "epoch": 2.7991805822831957, "grad_norm": 0.34810036784404397, "learning_rate": 1.3587911131737852e-06, "loss": 2.6075, "step": 45092 }, { "epoch": 2.7992426593829536, "grad_norm": 0.3471546733901783, "learning_rate": 1.3579550094437677e-06, "loss": 2.6181, "step": 45093 }, { "epoch": 2.7993047364827115, "grad_norm": 0.34445232473459114, "learning_rate": 1.3571191594911104e-06, "loss": 2.6251, "step": 45094 }, { "epoch": 2.7993668135824694, "grad_norm": 0.3980419589142452, "learning_rate": 1.3562835633201654e-06, "loss": 2.671, "step": 45095 }, { "epoch": 2.7994288906822273, "grad_norm": 0.3897970962893833, "learning_rate": 1.3554482209353014e-06, "loss": 2.6442, "step": 45096 }, { "epoch": 2.7994909677819853, "grad_norm": 0.3614947166966941, "learning_rate": 1.3546131323408705e-06, "loss": 2.6729, "step": 45097 }, { "epoch": 2.799553044881743, "grad_norm": 0.33188146690849635, "learning_rate": 1.353778297541236e-06, "loss": 2.6321, "step": 45098 }, { "epoch": 2.799615121981501, "grad_norm": 0.34018176358537827, "learning_rate": 1.3529437165407443e-06, "loss": 2.6907, "step": 45099 }, { "epoch": 2.799677199081259, "grad_norm": 0.37866336864112016, "learning_rate": 1.3521093893437586e-06, "loss": 2.6865, "step": 45100 }, { "epoch": 2.799739276181017, "grad_norm": 0.39074447712012444, "learning_rate": 1.35127531595462e-06, "loss": 2.6443, "step": 45101 }, { "epoch": 2.799801353280775, "grad_norm": 0.3682889852896971, "learning_rate": 1.3504414963776968e-06, "loss": 2.686, "step": 45102 }, { "epoch": 2.7998634303805328, "grad_norm": 0.3426355698725041, "learning_rate": 1.3496079306173305e-06, "loss": 2.6227, "step": 45103 }, { "epoch": 2.7999255074802907, "grad_norm": 0.36636876075114805, "learning_rate": 1.3487746186778616e-06, "loss": 2.6211, "step": 45104 }, { "epoch": 2.799987584580048, "grad_norm": 0.34486732094688693, "learning_rate": 1.3479415605636537e-06, "loss": 2.6388, "step": 45105 }, { "epoch": 2.8000496616798065, "grad_norm": 0.3477591030264959, "learning_rate": 1.3471087562790419e-06, "loss": 2.6912, "step": 45106 }, { "epoch": 2.800111738779564, "grad_norm": 0.3340928866248856, "learning_rate": 1.3462762058283784e-06, "loss": 2.6381, "step": 45107 }, { "epoch": 2.8001738158793223, "grad_norm": 0.3521999408213627, "learning_rate": 1.3454439092160042e-06, "loss": 2.7017, "step": 45108 }, { "epoch": 2.80023589297908, "grad_norm": 0.3575357557969256, "learning_rate": 1.3446118664462547e-06, "loss": 2.6219, "step": 45109 }, { "epoch": 2.800297970078838, "grad_norm": 0.34623116505689977, "learning_rate": 1.3437800775234765e-06, "loss": 2.6485, "step": 45110 }, { "epoch": 2.8003600471785957, "grad_norm": 0.36676930901859706, "learning_rate": 1.3429485424520162e-06, "loss": 2.6529, "step": 45111 }, { "epoch": 2.8004221242783536, "grad_norm": 0.3545463667732232, "learning_rate": 1.3421172612362033e-06, "loss": 2.7287, "step": 45112 }, { "epoch": 2.8004842013781115, "grad_norm": 0.3208678711016411, "learning_rate": 1.3412862338803678e-06, "loss": 2.6722, "step": 45113 }, { "epoch": 2.8005462784778694, "grad_norm": 0.35616730600260105, "learning_rate": 1.340455460388862e-06, "loss": 2.6469, "step": 45114 }, { "epoch": 2.8006083555776273, "grad_norm": 0.3424578896098951, "learning_rate": 1.3396249407660156e-06, "loss": 2.6217, "step": 45115 }, { "epoch": 2.8006704326773852, "grad_norm": 0.3690099737271935, "learning_rate": 1.3387946750161528e-06, "loss": 2.6715, "step": 45116 }, { "epoch": 2.800732509777143, "grad_norm": 0.35761197563969516, "learning_rate": 1.3379646631436093e-06, "loss": 2.6587, "step": 45117 }, { "epoch": 2.800794586876901, "grad_norm": 0.3509678580043769, "learning_rate": 1.3371349051527204e-06, "loss": 2.635, "step": 45118 }, { "epoch": 2.800856663976659, "grad_norm": 0.3780022287858115, "learning_rate": 1.3363054010478159e-06, "loss": 2.6394, "step": 45119 }, { "epoch": 2.800918741076417, "grad_norm": 0.3535416735752325, "learning_rate": 1.3354761508332147e-06, "loss": 2.6706, "step": 45120 }, { "epoch": 2.800980818176175, "grad_norm": 0.34934497403965115, "learning_rate": 1.3346471545132523e-06, "loss": 2.7226, "step": 45121 }, { "epoch": 2.8010428952759328, "grad_norm": 0.3510744521945395, "learning_rate": 1.3338184120922414e-06, "loss": 2.5872, "step": 45122 }, { "epoch": 2.8011049723756907, "grad_norm": 0.3450255677693609, "learning_rate": 1.3329899235745236e-06, "loss": 2.6956, "step": 45123 }, { "epoch": 2.8011670494754486, "grad_norm": 0.3946089715224888, "learning_rate": 1.3321616889644062e-06, "loss": 2.5478, "step": 45124 }, { "epoch": 2.8012291265752065, "grad_norm": 0.35304441714122387, "learning_rate": 1.3313337082662192e-06, "loss": 2.6933, "step": 45125 }, { "epoch": 2.8012912036749644, "grad_norm": 0.3450058084323762, "learning_rate": 1.3305059814842758e-06, "loss": 2.5704, "step": 45126 }, { "epoch": 2.8013532807747223, "grad_norm": 0.34632756793155695, "learning_rate": 1.3296785086229003e-06, "loss": 2.6406, "step": 45127 }, { "epoch": 2.8014153578744803, "grad_norm": 0.3886723734023383, "learning_rate": 1.3288512896864058e-06, "loss": 2.6963, "step": 45128 }, { "epoch": 2.801477434974238, "grad_norm": 0.3632966520264384, "learning_rate": 1.3280243246791114e-06, "loss": 2.7024, "step": 45129 }, { "epoch": 2.8015395120739957, "grad_norm": 0.3729897700036992, "learning_rate": 1.3271976136053244e-06, "loss": 2.6681, "step": 45130 }, { "epoch": 2.801601589173754, "grad_norm": 0.3432797770409981, "learning_rate": 1.3263711564693637e-06, "loss": 2.6326, "step": 45131 }, { "epoch": 2.8016636662735115, "grad_norm": 0.366254452981015, "learning_rate": 1.3255449532755483e-06, "loss": 2.6491, "step": 45132 }, { "epoch": 2.80172574337327, "grad_norm": 0.38541609102498275, "learning_rate": 1.3247190040281743e-06, "loss": 2.7297, "step": 45133 }, { "epoch": 2.8017878204730273, "grad_norm": 0.35259729258468303, "learning_rate": 1.3238933087315553e-06, "loss": 2.6519, "step": 45134 }, { "epoch": 2.8018498975727852, "grad_norm": 0.3445081370033587, "learning_rate": 1.3230678673900043e-06, "loss": 2.7456, "step": 45135 }, { "epoch": 2.801911974672543, "grad_norm": 0.32818520084535424, "learning_rate": 1.3222426800078237e-06, "loss": 2.6202, "step": 45136 }, { "epoch": 2.801974051772301, "grad_norm": 0.3621591856769443, "learning_rate": 1.3214177465893263e-06, "loss": 2.7023, "step": 45137 }, { "epoch": 2.802036128872059, "grad_norm": 0.3782420087690009, "learning_rate": 1.3205930671388033e-06, "loss": 2.6988, "step": 45138 }, { "epoch": 2.802098205971817, "grad_norm": 0.36789527613076684, "learning_rate": 1.319768641660568e-06, "loss": 2.7356, "step": 45139 }, { "epoch": 2.802160283071575, "grad_norm": 0.3488716781783362, "learning_rate": 1.3189444701589116e-06, "loss": 2.7301, "step": 45140 }, { "epoch": 2.8022223601713327, "grad_norm": 0.36404547753874844, "learning_rate": 1.3181205526381467e-06, "loss": 2.7364, "step": 45141 }, { "epoch": 2.8022844372710907, "grad_norm": 0.35282945164788826, "learning_rate": 1.3172968891025595e-06, "loss": 2.6689, "step": 45142 }, { "epoch": 2.8023465143708486, "grad_norm": 0.3267264078009957, "learning_rate": 1.316473479556457e-06, "loss": 2.605, "step": 45143 }, { "epoch": 2.8024085914706065, "grad_norm": 0.3553069448582006, "learning_rate": 1.3156503240041307e-06, "loss": 2.7322, "step": 45144 }, { "epoch": 2.8024706685703644, "grad_norm": 0.3474945163637189, "learning_rate": 1.3148274224498658e-06, "loss": 2.5904, "step": 45145 }, { "epoch": 2.8025327456701223, "grad_norm": 0.3773181511703941, "learning_rate": 1.3140047748979756e-06, "loss": 2.6588, "step": 45146 }, { "epoch": 2.8025948227698803, "grad_norm": 0.35685211378826165, "learning_rate": 1.31318238135274e-06, "loss": 2.6769, "step": 45147 }, { "epoch": 2.802656899869638, "grad_norm": 0.3619339066163853, "learning_rate": 1.31236024181845e-06, "loss": 2.6423, "step": 45148 }, { "epoch": 2.802718976969396, "grad_norm": 0.356055770596161, "learning_rate": 1.311538356299391e-06, "loss": 2.7433, "step": 45149 }, { "epoch": 2.802781054069154, "grad_norm": 0.35694695276813065, "learning_rate": 1.3107167247998597e-06, "loss": 2.6905, "step": 45150 }, { "epoch": 2.802843131168912, "grad_norm": 0.3463191131789264, "learning_rate": 1.3098953473241416e-06, "loss": 2.6171, "step": 45151 }, { "epoch": 2.80290520826867, "grad_norm": 0.37025913948586114, "learning_rate": 1.3090742238765163e-06, "loss": 2.7033, "step": 45152 }, { "epoch": 2.8029672853684273, "grad_norm": 0.3833496840072052, "learning_rate": 1.308253354461264e-06, "loss": 2.665, "step": 45153 }, { "epoch": 2.8030293624681857, "grad_norm": 0.3474219743691181, "learning_rate": 1.3074327390826868e-06, "loss": 2.7375, "step": 45154 }, { "epoch": 2.803091439567943, "grad_norm": 0.3760191856103148, "learning_rate": 1.3066123777450479e-06, "loss": 2.7035, "step": 45155 }, { "epoch": 2.8031535166677015, "grad_norm": 0.370295378534651, "learning_rate": 1.3057922704526326e-06, "loss": 2.6989, "step": 45156 }, { "epoch": 2.803215593767459, "grad_norm": 0.33056303667748693, "learning_rate": 1.3049724172097211e-06, "loss": 2.6878, "step": 45157 }, { "epoch": 2.8032776708672174, "grad_norm": 0.3480384632982195, "learning_rate": 1.3041528180205874e-06, "loss": 2.6685, "step": 45158 }, { "epoch": 2.803339747966975, "grad_norm": 0.389530651675062, "learning_rate": 1.3033334728895118e-06, "loss": 2.6885, "step": 45159 }, { "epoch": 2.8034018250667327, "grad_norm": 0.3229341450954943, "learning_rate": 1.3025143818207686e-06, "loss": 2.7146, "step": 45160 }, { "epoch": 2.8034639021664907, "grad_norm": 0.3435433156537019, "learning_rate": 1.3016955448186318e-06, "loss": 2.7005, "step": 45161 }, { "epoch": 2.8035259792662486, "grad_norm": 0.33597460253555966, "learning_rate": 1.3008769618873596e-06, "loss": 2.6301, "step": 45162 }, { "epoch": 2.8035880563660065, "grad_norm": 0.3531556377792081, "learning_rate": 1.3000586330312481e-06, "loss": 2.5873, "step": 45163 }, { "epoch": 2.8036501334657644, "grad_norm": 0.3573288092848746, "learning_rate": 1.2992405582545441e-06, "loss": 2.714, "step": 45164 }, { "epoch": 2.8037122105655223, "grad_norm": 0.35102188081004126, "learning_rate": 1.2984227375615277e-06, "loss": 2.766, "step": 45165 }, { "epoch": 2.8037742876652803, "grad_norm": 0.34409269676778304, "learning_rate": 1.2976051709564675e-06, "loss": 2.6683, "step": 45166 }, { "epoch": 2.803836364765038, "grad_norm": 0.3563302677477356, "learning_rate": 1.2967878584436155e-06, "loss": 2.6523, "step": 45167 }, { "epoch": 2.803898441864796, "grad_norm": 0.34030782868022763, "learning_rate": 1.2959708000272407e-06, "loss": 2.6918, "step": 45168 }, { "epoch": 2.803960518964554, "grad_norm": 0.37605532527151103, "learning_rate": 1.2951539957116176e-06, "loss": 2.6239, "step": 45169 }, { "epoch": 2.804022596064312, "grad_norm": 0.3365788436860016, "learning_rate": 1.2943374455010037e-06, "loss": 2.677, "step": 45170 }, { "epoch": 2.80408467316407, "grad_norm": 0.3629743653945635, "learning_rate": 1.2935211493996513e-06, "loss": 2.6943, "step": 45171 }, { "epoch": 2.8041467502638278, "grad_norm": 0.3609594165043477, "learning_rate": 1.292705107411818e-06, "loss": 2.5839, "step": 45172 }, { "epoch": 2.8042088273635857, "grad_norm": 0.34107479661399875, "learning_rate": 1.291889319541767e-06, "loss": 2.5898, "step": 45173 }, { "epoch": 2.8042709044633436, "grad_norm": 0.3555599975846559, "learning_rate": 1.2910737857937617e-06, "loss": 2.587, "step": 45174 }, { "epoch": 2.8043329815631015, "grad_norm": 0.3650820001175694, "learning_rate": 1.290258506172043e-06, "loss": 2.718, "step": 45175 }, { "epoch": 2.8043950586628594, "grad_norm": 0.36717995169165474, "learning_rate": 1.2894434806808686e-06, "loss": 2.6882, "step": 45176 }, { "epoch": 2.8044571357626173, "grad_norm": 0.3442726713276718, "learning_rate": 1.2886287093244965e-06, "loss": 2.645, "step": 45177 }, { "epoch": 2.804519212862375, "grad_norm": 0.3655929641056524, "learning_rate": 1.287814192107173e-06, "loss": 2.7021, "step": 45178 }, { "epoch": 2.804581289962133, "grad_norm": 0.38119179328218233, "learning_rate": 1.2869999290331503e-06, "loss": 2.6953, "step": 45179 }, { "epoch": 2.8046433670618907, "grad_norm": 0.35026803652429306, "learning_rate": 1.2861859201066694e-06, "loss": 2.6688, "step": 45180 }, { "epoch": 2.804705444161649, "grad_norm": 0.3397507375831261, "learning_rate": 1.2853721653319829e-06, "loss": 2.6036, "step": 45181 }, { "epoch": 2.8047675212614065, "grad_norm": 0.36389487750380917, "learning_rate": 1.2845586647133368e-06, "loss": 2.6578, "step": 45182 }, { "epoch": 2.8048295983611644, "grad_norm": 0.3577787566619978, "learning_rate": 1.283745418254978e-06, "loss": 2.6665, "step": 45183 }, { "epoch": 2.8048916754609223, "grad_norm": 0.36466691103676574, "learning_rate": 1.2829324259611419e-06, "loss": 2.6723, "step": 45184 }, { "epoch": 2.8049537525606802, "grad_norm": 0.3280995790534384, "learning_rate": 1.2821196878360697e-06, "loss": 2.64, "step": 45185 }, { "epoch": 2.805015829660438, "grad_norm": 0.3530607194543295, "learning_rate": 1.2813072038840134e-06, "loss": 2.5619, "step": 45186 }, { "epoch": 2.805077906760196, "grad_norm": 0.3648417601439229, "learning_rate": 1.2804949741092031e-06, "loss": 2.6623, "step": 45187 }, { "epoch": 2.805139983859954, "grad_norm": 0.3873964209713248, "learning_rate": 1.2796829985158744e-06, "loss": 2.7068, "step": 45188 }, { "epoch": 2.805202060959712, "grad_norm": 0.3493458058028622, "learning_rate": 1.2788712771082623e-06, "loss": 2.6883, "step": 45189 }, { "epoch": 2.80526413805947, "grad_norm": 0.3706223153568706, "learning_rate": 1.2780598098906137e-06, "loss": 2.6992, "step": 45190 }, { "epoch": 2.8053262151592278, "grad_norm": 0.3390337509940663, "learning_rate": 1.2772485968671588e-06, "loss": 2.6971, "step": 45191 }, { "epoch": 2.8053882922589857, "grad_norm": 0.3417297624366809, "learning_rate": 1.276437638042116e-06, "loss": 2.6698, "step": 45192 }, { "epoch": 2.8054503693587436, "grad_norm": 0.32195815243737935, "learning_rate": 1.2756269334197324e-06, "loss": 2.6303, "step": 45193 }, { "epoch": 2.8055124464585015, "grad_norm": 0.3556248248859102, "learning_rate": 1.2748164830042263e-06, "loss": 2.6412, "step": 45194 }, { "epoch": 2.8055745235582594, "grad_norm": 0.3363580953999241, "learning_rate": 1.2740062867998337e-06, "loss": 2.6098, "step": 45195 }, { "epoch": 2.8056366006580173, "grad_norm": 0.35282829022951645, "learning_rate": 1.2731963448107842e-06, "loss": 2.6489, "step": 45196 }, { "epoch": 2.8056986777577753, "grad_norm": 0.3556173244646932, "learning_rate": 1.2723866570412912e-06, "loss": 2.6032, "step": 45197 }, { "epoch": 2.805760754857533, "grad_norm": 0.3540296032227764, "learning_rate": 1.2715772234955847e-06, "loss": 2.7018, "step": 45198 }, { "epoch": 2.805822831957291, "grad_norm": 0.35147028879334336, "learning_rate": 1.2707680441778946e-06, "loss": 2.5765, "step": 45199 }, { "epoch": 2.805884909057049, "grad_norm": 0.3639525497157498, "learning_rate": 1.2699591190924286e-06, "loss": 2.74, "step": 45200 }, { "epoch": 2.8059469861568065, "grad_norm": 0.3682733874317207, "learning_rate": 1.2691504482434225e-06, "loss": 2.7828, "step": 45201 }, { "epoch": 2.806009063256565, "grad_norm": 0.36152845448248944, "learning_rate": 1.268342031635089e-06, "loss": 2.762, "step": 45202 }, { "epoch": 2.8060711403563223, "grad_norm": 0.36087289950423784, "learning_rate": 1.267533869271642e-06, "loss": 2.6583, "step": 45203 }, { "epoch": 2.8061332174560807, "grad_norm": 0.3662321364306973, "learning_rate": 1.2667259611573056e-06, "loss": 2.7417, "step": 45204 }, { "epoch": 2.806195294555838, "grad_norm": 0.36144232105256896, "learning_rate": 1.2659183072962877e-06, "loss": 2.6239, "step": 45205 }, { "epoch": 2.8062573716555965, "grad_norm": 0.3521404601858112, "learning_rate": 1.2651109076928014e-06, "loss": 2.6946, "step": 45206 }, { "epoch": 2.806319448755354, "grad_norm": 0.3629443544621141, "learning_rate": 1.2643037623510712e-06, "loss": 2.6709, "step": 45207 }, { "epoch": 2.806381525855112, "grad_norm": 0.3502144824779209, "learning_rate": 1.263496871275288e-06, "loss": 2.5901, "step": 45208 }, { "epoch": 2.80644360295487, "grad_norm": 0.3276553766932967, "learning_rate": 1.2626902344696767e-06, "loss": 2.633, "step": 45209 }, { "epoch": 2.8065056800546277, "grad_norm": 0.33453938730419686, "learning_rate": 1.2618838519384447e-06, "loss": 2.6576, "step": 45210 }, { "epoch": 2.8065677571543857, "grad_norm": 0.3559946634208721, "learning_rate": 1.261077723685794e-06, "loss": 2.7106, "step": 45211 }, { "epoch": 2.8066298342541436, "grad_norm": 0.3340581065211064, "learning_rate": 1.2602718497159327e-06, "loss": 2.7282, "step": 45212 }, { "epoch": 2.8066919113539015, "grad_norm": 0.352608745165822, "learning_rate": 1.2594662300330685e-06, "loss": 2.7339, "step": 45213 }, { "epoch": 2.8067539884536594, "grad_norm": 0.3632645551885715, "learning_rate": 1.2586608646413977e-06, "loss": 2.6841, "step": 45214 }, { "epoch": 2.8068160655534173, "grad_norm": 0.3437670465057391, "learning_rate": 1.2578557535451286e-06, "loss": 2.6593, "step": 45215 }, { "epoch": 2.8068781426531753, "grad_norm": 0.3740972341538221, "learning_rate": 1.2570508967484573e-06, "loss": 2.6819, "step": 45216 }, { "epoch": 2.806940219752933, "grad_norm": 0.3614933564022779, "learning_rate": 1.2562462942555752e-06, "loss": 2.7221, "step": 45217 }, { "epoch": 2.807002296852691, "grad_norm": 0.3368039379649082, "learning_rate": 1.2554419460707012e-06, "loss": 2.7603, "step": 45218 }, { "epoch": 2.807064373952449, "grad_norm": 0.3694564437152492, "learning_rate": 1.254637852198015e-06, "loss": 2.6464, "step": 45219 }, { "epoch": 2.807126451052207, "grad_norm": 0.363357146343843, "learning_rate": 1.2538340126417192e-06, "loss": 2.6925, "step": 45220 }, { "epoch": 2.807188528151965, "grad_norm": 0.33991288983571644, "learning_rate": 1.253030427405999e-06, "loss": 2.6062, "step": 45221 }, { "epoch": 2.8072506052517228, "grad_norm": 0.35746637703099365, "learning_rate": 1.2522270964950568e-06, "loss": 2.6135, "step": 45222 }, { "epoch": 2.8073126823514807, "grad_norm": 0.36275723308289964, "learning_rate": 1.2514240199130833e-06, "loss": 2.6634, "step": 45223 }, { "epoch": 2.8073747594512386, "grad_norm": 0.340987320344702, "learning_rate": 1.250621197664259e-06, "loss": 2.701, "step": 45224 }, { "epoch": 2.8074368365509965, "grad_norm": 0.36229971087963037, "learning_rate": 1.2498186297527802e-06, "loss": 2.6242, "step": 45225 }, { "epoch": 2.807498913650754, "grad_norm": 0.36115487558938936, "learning_rate": 1.2490163161828328e-06, "loss": 2.6326, "step": 45226 }, { "epoch": 2.8075609907505124, "grad_norm": 0.3483249576640037, "learning_rate": 1.248214256958602e-06, "loss": 2.7052, "step": 45227 }, { "epoch": 2.80762306785027, "grad_norm": 0.35923872312660565, "learning_rate": 1.2474124520842734e-06, "loss": 2.6122, "step": 45228 }, { "epoch": 2.807685144950028, "grad_norm": 0.37394968605492473, "learning_rate": 1.246610901564027e-06, "loss": 2.6931, "step": 45229 }, { "epoch": 2.8077472220497857, "grad_norm": 0.35976987374413544, "learning_rate": 1.245809605402043e-06, "loss": 2.7396, "step": 45230 }, { "epoch": 2.8078092991495436, "grad_norm": 0.33408814764459244, "learning_rate": 1.245008563602512e-06, "loss": 2.6722, "step": 45231 }, { "epoch": 2.8078713762493015, "grad_norm": 0.33968666436268935, "learning_rate": 1.2442077761696037e-06, "loss": 2.7045, "step": 45232 }, { "epoch": 2.8079334533490594, "grad_norm": 0.3551487372997167, "learning_rate": 1.2434072431075028e-06, "loss": 2.6486, "step": 45233 }, { "epoch": 2.8079955304488173, "grad_norm": 0.3446171616682262, "learning_rate": 1.2426069644203787e-06, "loss": 2.616, "step": 45234 }, { "epoch": 2.8080576075485753, "grad_norm": 0.33065938839253584, "learning_rate": 1.2418069401124111e-06, "loss": 2.6632, "step": 45235 }, { "epoch": 2.808119684648333, "grad_norm": 0.3387175237740962, "learning_rate": 1.2410071701877745e-06, "loss": 2.6784, "step": 45236 }, { "epoch": 2.808181761748091, "grad_norm": 0.3555932298625129, "learning_rate": 1.240207654650638e-06, "loss": 2.6617, "step": 45237 }, { "epoch": 2.808243838847849, "grad_norm": 0.34575029902607307, "learning_rate": 1.2394083935051815e-06, "loss": 2.6848, "step": 45238 }, { "epoch": 2.808305915947607, "grad_norm": 0.3623612557640744, "learning_rate": 1.238609386755557e-06, "loss": 2.7053, "step": 45239 }, { "epoch": 2.808367993047365, "grad_norm": 0.34800827762813746, "learning_rate": 1.2378106344059503e-06, "loss": 2.6232, "step": 45240 }, { "epoch": 2.8084300701471228, "grad_norm": 0.3328198212727223, "learning_rate": 1.2370121364605303e-06, "loss": 2.6134, "step": 45241 }, { "epoch": 2.8084921472468807, "grad_norm": 0.37235623484821706, "learning_rate": 1.236213892923449e-06, "loss": 2.6505, "step": 45242 }, { "epoch": 2.8085542243466386, "grad_norm": 0.36560678110621286, "learning_rate": 1.2354159037988756e-06, "loss": 2.7183, "step": 45243 }, { "epoch": 2.8086163014463965, "grad_norm": 0.3467284477222002, "learning_rate": 1.2346181690909787e-06, "loss": 2.5869, "step": 45244 }, { "epoch": 2.8086783785461544, "grad_norm": 0.345656871395382, "learning_rate": 1.233820688803916e-06, "loss": 2.6984, "step": 45245 }, { "epoch": 2.8087404556459123, "grad_norm": 0.3470711445624017, "learning_rate": 1.2330234629418514e-06, "loss": 2.6809, "step": 45246 }, { "epoch": 2.8088025327456703, "grad_norm": 0.37276849795597233, "learning_rate": 1.2322264915089365e-06, "loss": 2.6833, "step": 45247 }, { "epoch": 2.808864609845428, "grad_norm": 0.3592805530998888, "learning_rate": 1.231429774509335e-06, "loss": 2.637, "step": 45248 }, { "epoch": 2.8089266869451857, "grad_norm": 0.35305387151934947, "learning_rate": 1.2306333119472102e-06, "loss": 2.6696, "step": 45249 }, { "epoch": 2.808988764044944, "grad_norm": 0.36942125179317425, "learning_rate": 1.2298371038267087e-06, "loss": 2.585, "step": 45250 }, { "epoch": 2.8090508411447015, "grad_norm": 0.3495211036959089, "learning_rate": 1.2290411501519827e-06, "loss": 2.6596, "step": 45251 }, { "epoch": 2.80911291824446, "grad_norm": 0.36100261563254343, "learning_rate": 1.22824545092719e-06, "loss": 2.688, "step": 45252 }, { "epoch": 2.8091749953442173, "grad_norm": 0.33406740684581115, "learning_rate": 1.2274500061564775e-06, "loss": 2.6391, "step": 45253 }, { "epoch": 2.8092370724439757, "grad_norm": 0.34887590668443264, "learning_rate": 1.2266548158440028e-06, "loss": 2.6906, "step": 45254 }, { "epoch": 2.809299149543733, "grad_norm": 0.35564634057962746, "learning_rate": 1.2258598799939125e-06, "loss": 2.6299, "step": 45255 }, { "epoch": 2.809361226643491, "grad_norm": 0.3774880680194273, "learning_rate": 1.225065198610348e-06, "loss": 2.6938, "step": 45256 }, { "epoch": 2.809423303743249, "grad_norm": 0.3348418488406254, "learning_rate": 1.2242707716974554e-06, "loss": 2.6919, "step": 45257 }, { "epoch": 2.809485380843007, "grad_norm": 0.3625840500699073, "learning_rate": 1.2234765992593877e-06, "loss": 2.6222, "step": 45258 }, { "epoch": 2.809547457942765, "grad_norm": 0.37591492779351493, "learning_rate": 1.2226826813002856e-06, "loss": 2.6016, "step": 45259 }, { "epoch": 2.8096095350425228, "grad_norm": 0.35505205400729895, "learning_rate": 1.2218890178242848e-06, "loss": 2.6611, "step": 45260 }, { "epoch": 2.8096716121422807, "grad_norm": 0.35891457793389187, "learning_rate": 1.2210956088355262e-06, "loss": 2.6987, "step": 45261 }, { "epoch": 2.8097336892420386, "grad_norm": 0.36540481344084347, "learning_rate": 1.2203024543381624e-06, "loss": 2.6734, "step": 45262 }, { "epoch": 2.8097957663417965, "grad_norm": 0.3569983177105351, "learning_rate": 1.2195095543363177e-06, "loss": 2.6104, "step": 45263 }, { "epoch": 2.8098578434415544, "grad_norm": 0.36074375633472777, "learning_rate": 1.2187169088341333e-06, "loss": 2.6741, "step": 45264 }, { "epoch": 2.8099199205413123, "grad_norm": 0.3434348721008838, "learning_rate": 1.2179245178357447e-06, "loss": 2.6567, "step": 45265 }, { "epoch": 2.8099819976410703, "grad_norm": 0.35345548044368186, "learning_rate": 1.2171323813452873e-06, "loss": 2.6059, "step": 45266 }, { "epoch": 2.810044074740828, "grad_norm": 0.33897629006637797, "learning_rate": 1.216340499366886e-06, "loss": 2.6823, "step": 45267 }, { "epoch": 2.810106151840586, "grad_norm": 0.3592583010729203, "learning_rate": 1.215548871904687e-06, "loss": 2.6448, "step": 45268 }, { "epoch": 2.810168228940344, "grad_norm": 0.35263031811373585, "learning_rate": 1.2147574989628151e-06, "loss": 2.6377, "step": 45269 }, { "epoch": 2.810230306040102, "grad_norm": 0.3525615589857603, "learning_rate": 1.2139663805453894e-06, "loss": 2.7361, "step": 45270 }, { "epoch": 2.81029238313986, "grad_norm": 0.34032538693481074, "learning_rate": 1.213175516656545e-06, "loss": 2.6923, "step": 45271 }, { "epoch": 2.8103544602396178, "grad_norm": 0.34189531271179724, "learning_rate": 1.2123849073004123e-06, "loss": 2.698, "step": 45272 }, { "epoch": 2.8104165373393757, "grad_norm": 0.339528353995179, "learning_rate": 1.2115945524811102e-06, "loss": 2.6468, "step": 45273 }, { "epoch": 2.810478614439133, "grad_norm": 0.33928924573049996, "learning_rate": 1.2108044522027572e-06, "loss": 2.61, "step": 45274 }, { "epoch": 2.8105406915388915, "grad_norm": 0.36591852339251274, "learning_rate": 1.210014606469484e-06, "loss": 2.5702, "step": 45275 }, { "epoch": 2.810602768638649, "grad_norm": 0.3375713846866385, "learning_rate": 1.209225015285409e-06, "loss": 2.661, "step": 45276 }, { "epoch": 2.8106648457384074, "grad_norm": 0.35918915130941903, "learning_rate": 1.208435678654657e-06, "loss": 2.6517, "step": 45277 }, { "epoch": 2.810726922838165, "grad_norm": 0.3599302107994775, "learning_rate": 1.2076465965813356e-06, "loss": 2.6253, "step": 45278 }, { "epoch": 2.8107889999379227, "grad_norm": 0.3644070357999646, "learning_rate": 1.2068577690695692e-06, "loss": 2.6824, "step": 45279 }, { "epoch": 2.8108510770376807, "grad_norm": 0.34745263984468927, "learning_rate": 1.2060691961234661e-06, "loss": 2.6786, "step": 45280 }, { "epoch": 2.8109131541374386, "grad_norm": 0.36121075760234694, "learning_rate": 1.2052808777471503e-06, "loss": 2.7451, "step": 45281 }, { "epoch": 2.8109752312371965, "grad_norm": 0.32901937211174626, "learning_rate": 1.2044928139447242e-06, "loss": 2.7168, "step": 45282 }, { "epoch": 2.8110373083369544, "grad_norm": 0.344669758814795, "learning_rate": 1.2037050047203125e-06, "loss": 2.6185, "step": 45283 }, { "epoch": 2.8110993854367123, "grad_norm": 0.3545820684481397, "learning_rate": 1.202917450078006e-06, "loss": 2.6849, "step": 45284 }, { "epoch": 2.8111614625364703, "grad_norm": 0.36673060865696067, "learning_rate": 1.202130150021935e-06, "loss": 2.5814, "step": 45285 }, { "epoch": 2.811223539636228, "grad_norm": 0.35356757272335293, "learning_rate": 1.2013431045561908e-06, "loss": 2.6549, "step": 45286 }, { "epoch": 2.811285616735986, "grad_norm": 0.3516024672784046, "learning_rate": 1.2005563136848918e-06, "loss": 2.715, "step": 45287 }, { "epoch": 2.811347693835744, "grad_norm": 0.3357974581137166, "learning_rate": 1.1997697774121297e-06, "loss": 2.582, "step": 45288 }, { "epoch": 2.811409770935502, "grad_norm": 0.3462643351496635, "learning_rate": 1.1989834957420231e-06, "loss": 2.8107, "step": 45289 }, { "epoch": 2.81147184803526, "grad_norm": 0.3264698335139027, "learning_rate": 1.1981974686786635e-06, "loss": 2.6366, "step": 45290 }, { "epoch": 2.8115339251350178, "grad_norm": 0.35414295491147035, "learning_rate": 1.1974116962261528e-06, "loss": 2.6501, "step": 45291 }, { "epoch": 2.8115960022347757, "grad_norm": 0.34216347788619184, "learning_rate": 1.1966261783885935e-06, "loss": 2.7186, "step": 45292 }, { "epoch": 2.8116580793345336, "grad_norm": 0.33512022718497675, "learning_rate": 1.1958409151700822e-06, "loss": 2.6559, "step": 45293 }, { "epoch": 2.8117201564342915, "grad_norm": 0.34715038099417755, "learning_rate": 1.1950559065747158e-06, "loss": 2.6688, "step": 45294 }, { "epoch": 2.8117822335340494, "grad_norm": 0.3568030247702277, "learning_rate": 1.1942711526065964e-06, "loss": 2.7331, "step": 45295 }, { "epoch": 2.8118443106338074, "grad_norm": 0.3763423180987251, "learning_rate": 1.1934866532698097e-06, "loss": 2.6761, "step": 45296 }, { "epoch": 2.811906387733565, "grad_norm": 0.34368667235953765, "learning_rate": 1.1927024085684469e-06, "loss": 2.6861, "step": 45297 }, { "epoch": 2.811968464833323, "grad_norm": 0.3467060670084432, "learning_rate": 1.191918418506599e-06, "loss": 2.5728, "step": 45298 }, { "epoch": 2.8120305419330807, "grad_norm": 0.3657889006937718, "learning_rate": 1.191134683088374e-06, "loss": 2.7321, "step": 45299 }, { "epoch": 2.812092619032839, "grad_norm": 0.3589344641656572, "learning_rate": 1.190351202317841e-06, "loss": 2.6521, "step": 45300 }, { "epoch": 2.8121546961325965, "grad_norm": 0.3489849225158017, "learning_rate": 1.189567976199102e-06, "loss": 2.6852, "step": 45301 }, { "epoch": 2.812216773232355, "grad_norm": 0.34330711227389543, "learning_rate": 1.1887850047362315e-06, "loss": 2.66, "step": 45302 }, { "epoch": 2.8122788503321123, "grad_norm": 0.35782881537142175, "learning_rate": 1.1880022879333151e-06, "loss": 2.7426, "step": 45303 }, { "epoch": 2.8123409274318703, "grad_norm": 0.37489551837070917, "learning_rate": 1.1872198257944445e-06, "loss": 2.6458, "step": 45304 }, { "epoch": 2.812403004531628, "grad_norm": 0.3615317106201973, "learning_rate": 1.1864376183236992e-06, "loss": 2.6605, "step": 45305 }, { "epoch": 2.812465081631386, "grad_norm": 0.3479421168953448, "learning_rate": 1.1856556655251538e-06, "loss": 2.6954, "step": 45306 }, { "epoch": 2.812527158731144, "grad_norm": 0.3375968317358902, "learning_rate": 1.1848739674028942e-06, "loss": 2.5902, "step": 45307 }, { "epoch": 2.812589235830902, "grad_norm": 0.39231601904214075, "learning_rate": 1.184092523961e-06, "loss": 2.632, "step": 45308 }, { "epoch": 2.81265131293066, "grad_norm": 0.38475421946704813, "learning_rate": 1.1833113352035464e-06, "loss": 2.7293, "step": 45309 }, { "epoch": 2.8127133900304178, "grad_norm": 0.35416094223706895, "learning_rate": 1.1825304011346073e-06, "loss": 2.6862, "step": 45310 }, { "epoch": 2.8127754671301757, "grad_norm": 0.35687182266370854, "learning_rate": 1.181749721758252e-06, "loss": 2.6332, "step": 45311 }, { "epoch": 2.8128375442299336, "grad_norm": 0.3516123630342641, "learning_rate": 1.1809692970785658e-06, "loss": 2.6432, "step": 45312 }, { "epoch": 2.8128996213296915, "grad_norm": 0.35498600972506633, "learning_rate": 1.1801891270996123e-06, "loss": 2.6429, "step": 45313 }, { "epoch": 2.8129616984294494, "grad_norm": 0.36108734768944734, "learning_rate": 1.1794092118254662e-06, "loss": 2.7059, "step": 45314 }, { "epoch": 2.8130237755292073, "grad_norm": 0.35349601353996873, "learning_rate": 1.1786295512601907e-06, "loss": 2.6447, "step": 45315 }, { "epoch": 2.8130858526289653, "grad_norm": 0.34401041700356905, "learning_rate": 1.177850145407855e-06, "loss": 2.6103, "step": 45316 }, { "epoch": 2.813147929728723, "grad_norm": 0.3437735484467022, "learning_rate": 1.1770709942725332e-06, "loss": 2.658, "step": 45317 }, { "epoch": 2.813210006828481, "grad_norm": 0.35018012273964083, "learning_rate": 1.176292097858278e-06, "loss": 2.707, "step": 45318 }, { "epoch": 2.813272083928239, "grad_norm": 0.3443494684824823, "learning_rate": 1.1755134561691638e-06, "loss": 2.6775, "step": 45319 }, { "epoch": 2.813334161027997, "grad_norm": 0.3437945272043648, "learning_rate": 1.174735069209243e-06, "loss": 2.7008, "step": 45320 }, { "epoch": 2.813396238127755, "grad_norm": 0.3516565514962133, "learning_rate": 1.1739569369825897e-06, "loss": 2.6514, "step": 45321 }, { "epoch": 2.8134583152275123, "grad_norm": 0.3297112022485432, "learning_rate": 1.1731790594932512e-06, "loss": 2.6747, "step": 45322 }, { "epoch": 2.8135203923272707, "grad_norm": 0.34568735938033995, "learning_rate": 1.1724014367452906e-06, "loss": 2.6737, "step": 45323 }, { "epoch": 2.813582469427028, "grad_norm": 0.3659181568887439, "learning_rate": 1.171624068742766e-06, "loss": 2.7267, "step": 45324 }, { "epoch": 2.8136445465267865, "grad_norm": 0.3458335436490963, "learning_rate": 1.1708469554897295e-06, "loss": 2.6871, "step": 45325 }, { "epoch": 2.813706623626544, "grad_norm": 0.365227153523089, "learning_rate": 1.1700700969902446e-06, "loss": 2.5872, "step": 45326 }, { "epoch": 2.813768700726302, "grad_norm": 0.33785226637243965, "learning_rate": 1.1692934932483524e-06, "loss": 2.719, "step": 45327 }, { "epoch": 2.81383077782606, "grad_norm": 0.3504686929921254, "learning_rate": 1.168517144268111e-06, "loss": 2.6179, "step": 45328 }, { "epoch": 2.8138928549258178, "grad_norm": 0.340170768722327, "learning_rate": 1.1677410500535669e-06, "loss": 2.6711, "step": 45329 }, { "epoch": 2.8139549320255757, "grad_norm": 0.3401512202047398, "learning_rate": 1.166965210608778e-06, "loss": 2.7076, "step": 45330 }, { "epoch": 2.8140170091253336, "grad_norm": 0.3610561274989161, "learning_rate": 1.1661896259377801e-06, "loss": 2.6928, "step": 45331 }, { "epoch": 2.8140790862250915, "grad_norm": 0.36363600770163457, "learning_rate": 1.1654142960446313e-06, "loss": 2.7789, "step": 45332 }, { "epoch": 2.8141411633248494, "grad_norm": 0.33516505614407033, "learning_rate": 1.1646392209333723e-06, "loss": 2.6508, "step": 45333 }, { "epoch": 2.8142032404246073, "grad_norm": 0.36172127856695224, "learning_rate": 1.163864400608039e-06, "loss": 2.7052, "step": 45334 }, { "epoch": 2.8142653175243653, "grad_norm": 0.3516726533194132, "learning_rate": 1.1630898350726837e-06, "loss": 2.7192, "step": 45335 }, { "epoch": 2.814327394624123, "grad_norm": 0.35896161394323073, "learning_rate": 1.1623155243313477e-06, "loss": 2.6732, "step": 45336 }, { "epoch": 2.814389471723881, "grad_norm": 0.34206492131493366, "learning_rate": 1.1615414683880665e-06, "loss": 2.6751, "step": 45337 }, { "epoch": 2.814451548823639, "grad_norm": 0.354602428770811, "learning_rate": 1.1607676672468704e-06, "loss": 2.7234, "step": 45338 }, { "epoch": 2.814513625923397, "grad_norm": 0.3558644617849811, "learning_rate": 1.159994120911817e-06, "loss": 2.6611, "step": 45339 }, { "epoch": 2.814575703023155, "grad_norm": 0.36263441109633154, "learning_rate": 1.1592208293869256e-06, "loss": 2.6933, "step": 45340 }, { "epoch": 2.8146377801229128, "grad_norm": 0.3650499377970217, "learning_rate": 1.1584477926762372e-06, "loss": 2.6326, "step": 45341 }, { "epoch": 2.8146998572226707, "grad_norm": 0.3709645498129724, "learning_rate": 1.157675010783782e-06, "loss": 2.7003, "step": 45342 }, { "epoch": 2.8147619343224286, "grad_norm": 0.38532071809443175, "learning_rate": 1.15690248371359e-06, "loss": 2.7473, "step": 45343 }, { "epoch": 2.8148240114221865, "grad_norm": 0.3668066647069197, "learning_rate": 1.1561302114697025e-06, "loss": 2.6797, "step": 45344 }, { "epoch": 2.814886088521944, "grad_norm": 0.35889396400663837, "learning_rate": 1.1553581940561388e-06, "loss": 2.6101, "step": 45345 }, { "epoch": 2.8149481656217024, "grad_norm": 0.3676080122015373, "learning_rate": 1.154586431476923e-06, "loss": 2.5847, "step": 45346 }, { "epoch": 2.81501024272146, "grad_norm": 0.3682077923209174, "learning_rate": 1.1538149237360906e-06, "loss": 2.6956, "step": 45347 }, { "epoch": 2.815072319821218, "grad_norm": 0.3497595757427188, "learning_rate": 1.153043670837667e-06, "loss": 2.6449, "step": 45348 }, { "epoch": 2.8151343969209757, "grad_norm": 0.3597837072875155, "learning_rate": 1.1522726727856702e-06, "loss": 2.65, "step": 45349 }, { "epoch": 2.8151964740207336, "grad_norm": 0.3588752173832016, "learning_rate": 1.1515019295841256e-06, "loss": 2.6803, "step": 45350 }, { "epoch": 2.8152585511204915, "grad_norm": 0.3867006971981352, "learning_rate": 1.1507314412370518e-06, "loss": 2.6351, "step": 45351 }, { "epoch": 2.8153206282202494, "grad_norm": 0.35812529672074467, "learning_rate": 1.1499612077484677e-06, "loss": 2.6862, "step": 45352 }, { "epoch": 2.8153827053200073, "grad_norm": 0.3480226643166003, "learning_rate": 1.1491912291224039e-06, "loss": 2.697, "step": 45353 }, { "epoch": 2.8154447824197653, "grad_norm": 0.3577723009941308, "learning_rate": 1.1484215053628622e-06, "loss": 2.6355, "step": 45354 }, { "epoch": 2.815506859519523, "grad_norm": 0.3714085498525164, "learning_rate": 1.1476520364738674e-06, "loss": 2.6502, "step": 45355 }, { "epoch": 2.815568936619281, "grad_norm": 0.3610384217894395, "learning_rate": 1.1468828224594274e-06, "loss": 2.6305, "step": 45356 }, { "epoch": 2.815631013719039, "grad_norm": 0.3594846350910624, "learning_rate": 1.1461138633235612e-06, "loss": 2.6804, "step": 45357 }, { "epoch": 2.815693090818797, "grad_norm": 0.36389502082556807, "learning_rate": 1.145345159070277e-06, "loss": 2.6377, "step": 45358 }, { "epoch": 2.815755167918555, "grad_norm": 0.34750467918182165, "learning_rate": 1.1445767097035931e-06, "loss": 2.6099, "step": 45359 }, { "epoch": 2.8158172450183128, "grad_norm": 0.35522729407275866, "learning_rate": 1.1438085152275012e-06, "loss": 2.5732, "step": 45360 }, { "epoch": 2.8158793221180707, "grad_norm": 0.35486910619809714, "learning_rate": 1.1430405756460316e-06, "loss": 2.6647, "step": 45361 }, { "epoch": 2.8159413992178286, "grad_norm": 0.3327693029138549, "learning_rate": 1.1422728909631753e-06, "loss": 2.6078, "step": 45362 }, { "epoch": 2.8160034763175865, "grad_norm": 0.3563514272292186, "learning_rate": 1.14150546118294e-06, "loss": 2.6861, "step": 45363 }, { "epoch": 2.8160655534173444, "grad_norm": 0.34790182548360904, "learning_rate": 1.140738286309334e-06, "loss": 2.6005, "step": 45364 }, { "epoch": 2.8161276305171024, "grad_norm": 0.3379977392743196, "learning_rate": 1.1399713663463596e-06, "loss": 2.5782, "step": 45365 }, { "epoch": 2.8161897076168603, "grad_norm": 0.35719796760286315, "learning_rate": 1.1392047012980078e-06, "loss": 2.7159, "step": 45366 }, { "epoch": 2.816251784716618, "grad_norm": 0.33090895424293626, "learning_rate": 1.1384382911682979e-06, "loss": 2.6637, "step": 45367 }, { "epoch": 2.8163138618163757, "grad_norm": 0.34507533900747195, "learning_rate": 1.1376721359612098e-06, "loss": 2.6514, "step": 45368 }, { "epoch": 2.816375938916134, "grad_norm": 0.3462599779233759, "learning_rate": 1.1369062356807514e-06, "loss": 2.6488, "step": 45369 }, { "epoch": 2.8164380160158915, "grad_norm": 0.3639955164310973, "learning_rate": 1.1361405903309085e-06, "loss": 2.6967, "step": 45370 }, { "epoch": 2.81650009311565, "grad_norm": 0.35244885080306837, "learning_rate": 1.135375199915689e-06, "loss": 2.6866, "step": 45371 }, { "epoch": 2.8165621702154073, "grad_norm": 0.34150096618241993, "learning_rate": 1.134610064439079e-06, "loss": 2.7431, "step": 45372 }, { "epoch": 2.8166242473151657, "grad_norm": 0.35141396033621675, "learning_rate": 1.133845183905069e-06, "loss": 2.6476, "step": 45373 }, { "epoch": 2.816686324414923, "grad_norm": 0.33528792988969647, "learning_rate": 1.1330805583176452e-06, "loss": 2.6478, "step": 45374 }, { "epoch": 2.816748401514681, "grad_norm": 0.34549921563096725, "learning_rate": 1.1323161876808096e-06, "loss": 2.703, "step": 45375 }, { "epoch": 2.816810478614439, "grad_norm": 0.34719913949005554, "learning_rate": 1.1315520719985485e-06, "loss": 2.6762, "step": 45376 }, { "epoch": 2.816872555714197, "grad_norm": 0.32509335360053215, "learning_rate": 1.1307882112748359e-06, "loss": 2.5532, "step": 45377 }, { "epoch": 2.816934632813955, "grad_norm": 0.33465639065967423, "learning_rate": 1.1300246055136687e-06, "loss": 2.6441, "step": 45378 }, { "epoch": 2.8169967099137128, "grad_norm": 0.35759972488544217, "learning_rate": 1.1292612547190217e-06, "loss": 2.6539, "step": 45379 }, { "epoch": 2.8170587870134707, "grad_norm": 0.3677776545091892, "learning_rate": 1.1284981588948862e-06, "loss": 2.6361, "step": 45380 }, { "epoch": 2.8171208641132286, "grad_norm": 0.362880834769745, "learning_rate": 1.1277353180452422e-06, "loss": 2.6186, "step": 45381 }, { "epoch": 2.8171829412129865, "grad_norm": 0.33186462933919303, "learning_rate": 1.1269727321740642e-06, "loss": 2.6049, "step": 45382 }, { "epoch": 2.8172450183127444, "grad_norm": 0.3294423543333218, "learning_rate": 1.1262104012853324e-06, "loss": 2.6649, "step": 45383 }, { "epoch": 2.8173070954125023, "grad_norm": 0.36999312362209336, "learning_rate": 1.1254483253830272e-06, "loss": 2.7309, "step": 45384 }, { "epoch": 2.8173691725122603, "grad_norm": 0.35164216516475705, "learning_rate": 1.124686504471123e-06, "loss": 2.6319, "step": 45385 }, { "epoch": 2.817431249612018, "grad_norm": 0.3660511940005895, "learning_rate": 1.1239249385535945e-06, "loss": 2.6545, "step": 45386 }, { "epoch": 2.817493326711776, "grad_norm": 0.35948589935821834, "learning_rate": 1.123163627634416e-06, "loss": 2.6659, "step": 45387 }, { "epoch": 2.817555403811534, "grad_norm": 0.3491130161017839, "learning_rate": 1.1224025717175512e-06, "loss": 2.7367, "step": 45388 }, { "epoch": 2.817617480911292, "grad_norm": 0.3365641781261265, "learning_rate": 1.1216417708069805e-06, "loss": 2.7016, "step": 45389 }, { "epoch": 2.81767955801105, "grad_norm": 0.3425784412239033, "learning_rate": 1.1208812249066725e-06, "loss": 2.7726, "step": 45390 }, { "epoch": 2.8177416351108078, "grad_norm": 0.361076329106911, "learning_rate": 1.1201209340205965e-06, "loss": 2.5863, "step": 45391 }, { "epoch": 2.8178037122105657, "grad_norm": 0.3639609428505823, "learning_rate": 1.1193608981527104e-06, "loss": 2.7046, "step": 45392 }, { "epoch": 2.817865789310323, "grad_norm": 0.343078586135209, "learning_rate": 1.1186011173069833e-06, "loss": 2.6845, "step": 45393 }, { "epoch": 2.8179278664100815, "grad_norm": 0.3507631114526575, "learning_rate": 1.1178415914873841e-06, "loss": 2.5672, "step": 45394 }, { "epoch": 2.817989943509839, "grad_norm": 0.3612089837030618, "learning_rate": 1.117082320697871e-06, "loss": 2.6649, "step": 45395 }, { "epoch": 2.8180520206095974, "grad_norm": 0.35200109673526986, "learning_rate": 1.1163233049424072e-06, "loss": 2.6462, "step": 45396 }, { "epoch": 2.818114097709355, "grad_norm": 0.36609947034051965, "learning_rate": 1.1155645442249451e-06, "loss": 2.6876, "step": 45397 }, { "epoch": 2.8181761748091128, "grad_norm": 0.36825657499287395, "learning_rate": 1.1148060385494542e-06, "loss": 2.6765, "step": 45398 }, { "epoch": 2.8182382519088707, "grad_norm": 0.3570667858510914, "learning_rate": 1.1140477879198918e-06, "loss": 2.6172, "step": 45399 }, { "epoch": 2.8183003290086286, "grad_norm": 0.3329865680976961, "learning_rate": 1.1132897923402053e-06, "loss": 2.6408, "step": 45400 }, { "epoch": 2.8183624061083865, "grad_norm": 0.3414092672638064, "learning_rate": 1.1125320518143579e-06, "loss": 2.6839, "step": 45401 }, { "epoch": 2.8184244832081444, "grad_norm": 0.3746257941473409, "learning_rate": 1.111774566346291e-06, "loss": 2.6868, "step": 45402 }, { "epoch": 2.8184865603079023, "grad_norm": 0.3577122368744909, "learning_rate": 1.1110173359399733e-06, "loss": 2.5982, "step": 45403 }, { "epoch": 2.8185486374076603, "grad_norm": 0.3367052398591147, "learning_rate": 1.1102603605993412e-06, "loss": 2.6561, "step": 45404 }, { "epoch": 2.818610714507418, "grad_norm": 0.33772426986313264, "learning_rate": 1.1095036403283521e-06, "loss": 2.6306, "step": 45405 }, { "epoch": 2.818672791607176, "grad_norm": 0.3371757817640651, "learning_rate": 1.108747175130942e-06, "loss": 2.6359, "step": 45406 }, { "epoch": 2.818734868706934, "grad_norm": 0.3676046477426631, "learning_rate": 1.1079909650110743e-06, "loss": 2.5949, "step": 45407 }, { "epoch": 2.818796945806692, "grad_norm": 0.3367494595678364, "learning_rate": 1.1072350099726903e-06, "loss": 2.5922, "step": 45408 }, { "epoch": 2.81885902290645, "grad_norm": 0.3529381434868941, "learning_rate": 1.1064793100197257e-06, "loss": 2.6261, "step": 45409 }, { "epoch": 2.8189211000062078, "grad_norm": 0.36706544998519025, "learning_rate": 1.1057238651561219e-06, "loss": 2.5754, "step": 45410 }, { "epoch": 2.8189831771059657, "grad_norm": 0.34735694602785244, "learning_rate": 1.1049686753858368e-06, "loss": 2.689, "step": 45411 }, { "epoch": 2.8190452542057236, "grad_norm": 0.3402214910649588, "learning_rate": 1.104213740712795e-06, "loss": 2.7672, "step": 45412 }, { "epoch": 2.8191073313054815, "grad_norm": 0.37312397983865236, "learning_rate": 1.1034590611409435e-06, "loss": 2.7504, "step": 45413 }, { "epoch": 2.8191694084052394, "grad_norm": 0.3757046397387453, "learning_rate": 1.102704636674212e-06, "loss": 2.6935, "step": 45414 }, { "epoch": 2.8192314855049974, "grad_norm": 0.34140942002489844, "learning_rate": 1.101950467316537e-06, "loss": 2.6559, "step": 45415 }, { "epoch": 2.819293562604755, "grad_norm": 0.3495274772832649, "learning_rate": 1.1011965530718648e-06, "loss": 2.6159, "step": 45416 }, { "epoch": 2.819355639704513, "grad_norm": 0.34898710013666967, "learning_rate": 1.1004428939441202e-06, "loss": 2.6128, "step": 45417 }, { "epoch": 2.8194177168042707, "grad_norm": 0.37091933719353853, "learning_rate": 1.0996894899372334e-06, "loss": 2.721, "step": 45418 }, { "epoch": 2.819479793904029, "grad_norm": 0.3593540392858177, "learning_rate": 1.0989363410551346e-06, "loss": 2.7229, "step": 45419 }, { "epoch": 2.8195418710037865, "grad_norm": 0.3419408767350782, "learning_rate": 1.0981834473017594e-06, "loss": 2.7084, "step": 45420 }, { "epoch": 2.819603948103545, "grad_norm": 0.32635733762495894, "learning_rate": 1.0974308086810325e-06, "loss": 2.6984, "step": 45421 }, { "epoch": 2.8196660252033023, "grad_norm": 0.3351516736658481, "learning_rate": 1.0966784251968786e-06, "loss": 2.6588, "step": 45422 }, { "epoch": 2.8197281023030603, "grad_norm": 0.35429980309400433, "learning_rate": 1.095926296853228e-06, "loss": 2.6374, "step": 45423 }, { "epoch": 2.819790179402818, "grad_norm": 0.3540788242384863, "learning_rate": 1.0951744236539941e-06, "loss": 2.6994, "step": 45424 }, { "epoch": 2.819852256502576, "grad_norm": 0.366152400106669, "learning_rate": 1.094422805603118e-06, "loss": 2.6433, "step": 45425 }, { "epoch": 2.819914333602334, "grad_norm": 0.3667014334011124, "learning_rate": 1.0936714427045025e-06, "loss": 2.5916, "step": 45426 }, { "epoch": 2.819976410702092, "grad_norm": 0.33183109758158136, "learning_rate": 1.0929203349620831e-06, "loss": 2.6385, "step": 45427 }, { "epoch": 2.82003848780185, "grad_norm": 0.3585410808704001, "learning_rate": 1.0921694823797623e-06, "loss": 2.675, "step": 45428 }, { "epoch": 2.8201005649016078, "grad_norm": 0.35092314757893395, "learning_rate": 1.091418884961465e-06, "loss": 2.6318, "step": 45429 }, { "epoch": 2.8201626420013657, "grad_norm": 0.3722806921488594, "learning_rate": 1.090668542711115e-06, "loss": 2.6989, "step": 45430 }, { "epoch": 2.8202247191011236, "grad_norm": 0.35198336946556574, "learning_rate": 1.0899184556326214e-06, "loss": 2.5976, "step": 45431 }, { "epoch": 2.8202867962008815, "grad_norm": 0.35607462380947863, "learning_rate": 1.089168623729897e-06, "loss": 2.6557, "step": 45432 }, { "epoch": 2.8203488733006394, "grad_norm": 0.3603681234880628, "learning_rate": 1.0884190470068444e-06, "loss": 2.7247, "step": 45433 }, { "epoch": 2.8204109504003974, "grad_norm": 0.3395783275789174, "learning_rate": 1.0876697254673885e-06, "loss": 2.6495, "step": 45434 }, { "epoch": 2.8204730275001553, "grad_norm": 0.3438932272158187, "learning_rate": 1.0869206591154368e-06, "loss": 2.6958, "step": 45435 }, { "epoch": 2.820535104599913, "grad_norm": 0.3449495151111619, "learning_rate": 1.0861718479548921e-06, "loss": 2.6922, "step": 45436 }, { "epoch": 2.820597181699671, "grad_norm": 0.35829737015205415, "learning_rate": 1.0854232919896624e-06, "loss": 2.6373, "step": 45437 }, { "epoch": 2.820659258799429, "grad_norm": 0.3588569680682693, "learning_rate": 1.0846749912236554e-06, "loss": 2.5755, "step": 45438 }, { "epoch": 2.820721335899187, "grad_norm": 0.3678625273409657, "learning_rate": 1.0839269456607737e-06, "loss": 2.6309, "step": 45439 }, { "epoch": 2.820783412998945, "grad_norm": 0.366121611827256, "learning_rate": 1.0831791553049142e-06, "loss": 2.6831, "step": 45440 }, { "epoch": 2.8208454900987023, "grad_norm": 0.34404948583272604, "learning_rate": 1.0824316201599905e-06, "loss": 2.7664, "step": 45441 }, { "epoch": 2.8209075671984607, "grad_norm": 0.3485160760421246, "learning_rate": 1.0816843402298937e-06, "loss": 2.6149, "step": 45442 }, { "epoch": 2.820969644298218, "grad_norm": 0.34811451529734844, "learning_rate": 1.0809373155185265e-06, "loss": 2.609, "step": 45443 }, { "epoch": 2.8210317213979765, "grad_norm": 0.3474988366697585, "learning_rate": 1.0801905460297856e-06, "loss": 2.657, "step": 45444 }, { "epoch": 2.821093798497734, "grad_norm": 0.36717041462858196, "learning_rate": 1.0794440317675626e-06, "loss": 2.7476, "step": 45445 }, { "epoch": 2.821155875597492, "grad_norm": 0.3709193974539179, "learning_rate": 1.078697772735754e-06, "loss": 2.685, "step": 45446 }, { "epoch": 2.82121795269725, "grad_norm": 0.35920728003934466, "learning_rate": 1.0779517689382623e-06, "loss": 2.599, "step": 45447 }, { "epoch": 2.8212800297970078, "grad_norm": 0.3295576856960723, "learning_rate": 1.0772060203789681e-06, "loss": 2.5806, "step": 45448 }, { "epoch": 2.8213421068967657, "grad_norm": 0.32645520450786214, "learning_rate": 1.0764605270617678e-06, "loss": 2.6012, "step": 45449 }, { "epoch": 2.8214041839965236, "grad_norm": 0.3639330195747789, "learning_rate": 1.075715288990553e-06, "loss": 2.706, "step": 45450 }, { "epoch": 2.8214662610962815, "grad_norm": 0.37755432019137, "learning_rate": 1.0749703061691984e-06, "loss": 2.6592, "step": 45451 }, { "epoch": 2.8215283381960394, "grad_norm": 0.34324097875935994, "learning_rate": 1.074225578601612e-06, "loss": 2.6063, "step": 45452 }, { "epoch": 2.8215904152957973, "grad_norm": 0.3521347051683917, "learning_rate": 1.0734811062916627e-06, "loss": 2.7068, "step": 45453 }, { "epoch": 2.8216524923955553, "grad_norm": 0.37010833004943955, "learning_rate": 1.072736889243242e-06, "loss": 2.6037, "step": 45454 }, { "epoch": 2.821714569495313, "grad_norm": 0.35695548975381003, "learning_rate": 1.0719929274602247e-06, "loss": 2.665, "step": 45455 }, { "epoch": 2.821776646595071, "grad_norm": 0.3359321105879578, "learning_rate": 1.071249220946502e-06, "loss": 2.5269, "step": 45456 }, { "epoch": 2.821838723694829, "grad_norm": 0.3510704942274889, "learning_rate": 1.0705057697059484e-06, "loss": 2.6337, "step": 45457 }, { "epoch": 2.821900800794587, "grad_norm": 0.34594607033317787, "learning_rate": 1.0697625737424445e-06, "loss": 2.6725, "step": 45458 }, { "epoch": 2.821962877894345, "grad_norm": 0.3367443707802037, "learning_rate": 1.06901963305987e-06, "loss": 2.6512, "step": 45459 }, { "epoch": 2.8220249549941028, "grad_norm": 0.3673728921251547, "learning_rate": 1.0682769476620947e-06, "loss": 2.6737, "step": 45460 }, { "epoch": 2.8220870320938607, "grad_norm": 0.35746174991162266, "learning_rate": 1.0675345175530038e-06, "loss": 2.7219, "step": 45461 }, { "epoch": 2.8221491091936186, "grad_norm": 0.35631699015878376, "learning_rate": 1.0667923427364556e-06, "loss": 2.6366, "step": 45462 }, { "epoch": 2.8222111862933765, "grad_norm": 0.34780234578228875, "learning_rate": 1.066050423216336e-06, "loss": 2.6757, "step": 45463 }, { "epoch": 2.822273263393134, "grad_norm": 0.3654462638288216, "learning_rate": 1.0653087589965082e-06, "loss": 2.6717, "step": 45464 }, { "epoch": 2.8223353404928924, "grad_norm": 0.37345257014428135, "learning_rate": 1.0645673500808418e-06, "loss": 2.6433, "step": 45465 }, { "epoch": 2.82239741759265, "grad_norm": 0.3506124056119818, "learning_rate": 1.063826196473211e-06, "loss": 2.6353, "step": 45466 }, { "epoch": 2.822459494692408, "grad_norm": 0.35356788881189616, "learning_rate": 1.0630852981774797e-06, "loss": 2.7313, "step": 45467 }, { "epoch": 2.8225215717921657, "grad_norm": 0.3397811086516313, "learning_rate": 1.0623446551975114e-06, "loss": 2.6908, "step": 45468 }, { "epoch": 2.822583648891924, "grad_norm": 0.3324165016573801, "learning_rate": 1.061604267537164e-06, "loss": 2.6652, "step": 45469 }, { "epoch": 2.8226457259916815, "grad_norm": 0.3313550683498656, "learning_rate": 1.060864135200318e-06, "loss": 2.6559, "step": 45470 }, { "epoch": 2.8227078030914394, "grad_norm": 0.35428169663794706, "learning_rate": 1.06012425819082e-06, "loss": 2.6153, "step": 45471 }, { "epoch": 2.8227698801911973, "grad_norm": 0.3404699891528706, "learning_rate": 1.0593846365125336e-06, "loss": 2.6278, "step": 45472 }, { "epoch": 2.8228319572909553, "grad_norm": 0.33603060341510915, "learning_rate": 1.0586452701693172e-06, "loss": 2.705, "step": 45473 }, { "epoch": 2.822894034390713, "grad_norm": 0.3434576937506661, "learning_rate": 1.0579061591650286e-06, "loss": 2.6636, "step": 45474 }, { "epoch": 2.822956111490471, "grad_norm": 0.35436809879559966, "learning_rate": 1.0571673035035311e-06, "loss": 2.6996, "step": 45475 }, { "epoch": 2.823018188590229, "grad_norm": 0.3484342420964543, "learning_rate": 1.0564287031886666e-06, "loss": 2.6581, "step": 45476 }, { "epoch": 2.823080265689987, "grad_norm": 0.3440144345407194, "learning_rate": 1.0556903582242982e-06, "loss": 2.5963, "step": 45477 }, { "epoch": 2.823142342789745, "grad_norm": 0.36471351161395105, "learning_rate": 1.0549522686142733e-06, "loss": 2.6199, "step": 45478 }, { "epoch": 2.8232044198895028, "grad_norm": 0.3395674557962276, "learning_rate": 1.0542144343624438e-06, "loss": 2.8073, "step": 45479 }, { "epoch": 2.8232664969892607, "grad_norm": 0.38404975171577166, "learning_rate": 1.0534768554726626e-06, "loss": 2.7106, "step": 45480 }, { "epoch": 2.8233285740890186, "grad_norm": 0.3257882609173297, "learning_rate": 1.052739531948771e-06, "loss": 2.6049, "step": 45481 }, { "epoch": 2.8233906511887765, "grad_norm": 0.3631458645471308, "learning_rate": 1.0520024637946213e-06, "loss": 2.5891, "step": 45482 }, { "epoch": 2.8234527282885344, "grad_norm": 0.34692844546182894, "learning_rate": 1.051265651014055e-06, "loss": 2.7439, "step": 45483 }, { "epoch": 2.8235148053882924, "grad_norm": 0.3640695857721457, "learning_rate": 1.0505290936109248e-06, "loss": 2.6539, "step": 45484 }, { "epoch": 2.8235768824880503, "grad_norm": 0.36615667322571754, "learning_rate": 1.0497927915890605e-06, "loss": 2.6904, "step": 45485 }, { "epoch": 2.823638959587808, "grad_norm": 0.3434959686742377, "learning_rate": 1.0490567449523147e-06, "loss": 2.6471, "step": 45486 }, { "epoch": 2.823701036687566, "grad_norm": 0.37319470018153217, "learning_rate": 1.048320953704518e-06, "loss": 2.7137, "step": 45487 }, { "epoch": 2.823763113787324, "grad_norm": 0.35373899874076337, "learning_rate": 1.0475854178495227e-06, "loss": 2.6866, "step": 45488 }, { "epoch": 2.8238251908870815, "grad_norm": 0.34304559667875134, "learning_rate": 1.0468501373911532e-06, "loss": 2.6542, "step": 45489 }, { "epoch": 2.82388726798684, "grad_norm": 0.3565661081972361, "learning_rate": 1.0461151123332514e-06, "loss": 2.7073, "step": 45490 }, { "epoch": 2.8239493450865973, "grad_norm": 0.3368425044486091, "learning_rate": 1.0453803426796472e-06, "loss": 2.637, "step": 45491 }, { "epoch": 2.8240114221863557, "grad_norm": 0.34863704390322114, "learning_rate": 1.044645828434182e-06, "loss": 2.7248, "step": 45492 }, { "epoch": 2.824073499286113, "grad_norm": 0.3362834723556645, "learning_rate": 1.0439115696006806e-06, "loss": 2.5632, "step": 45493 }, { "epoch": 2.824135576385871, "grad_norm": 0.35580139525302373, "learning_rate": 1.0431775661829735e-06, "loss": 2.6388, "step": 45494 }, { "epoch": 2.824197653485629, "grad_norm": 0.3685014024148805, "learning_rate": 1.0424438181849017e-06, "loss": 2.7786, "step": 45495 }, { "epoch": 2.824259730585387, "grad_norm": 0.3563102152613478, "learning_rate": 1.041710325610279e-06, "loss": 2.6289, "step": 45496 }, { "epoch": 2.824321807685145, "grad_norm": 0.3469540222472469, "learning_rate": 1.0409770884629412e-06, "loss": 2.677, "step": 45497 }, { "epoch": 2.8243838847849028, "grad_norm": 0.3323490641339002, "learning_rate": 1.0402441067467183e-06, "loss": 2.6811, "step": 45498 }, { "epoch": 2.8244459618846607, "grad_norm": 0.349759842017693, "learning_rate": 1.039511380465419e-06, "loss": 2.6918, "step": 45499 }, { "epoch": 2.8245080389844186, "grad_norm": 0.34700658725180655, "learning_rate": 1.0387789096228783e-06, "loss": 2.6977, "step": 45500 }, { "epoch": 2.8245701160841765, "grad_norm": 0.34155413045863714, "learning_rate": 1.0380466942229105e-06, "loss": 2.6608, "step": 45501 }, { "epoch": 2.8246321931839344, "grad_norm": 0.36883908020181605, "learning_rate": 1.03731473426934e-06, "loss": 2.6423, "step": 45502 }, { "epoch": 2.8246942702836924, "grad_norm": 0.3695034484405551, "learning_rate": 1.0365830297659862e-06, "loss": 2.7003, "step": 45503 }, { "epoch": 2.8247563473834503, "grad_norm": 0.35151288424166643, "learning_rate": 1.0358515807166625e-06, "loss": 2.6341, "step": 45504 }, { "epoch": 2.824818424483208, "grad_norm": 0.3530080956241257, "learning_rate": 1.035120387125188e-06, "loss": 2.7522, "step": 45505 }, { "epoch": 2.824880501582966, "grad_norm": 0.3449796444651353, "learning_rate": 1.0343894489953764e-06, "loss": 2.7685, "step": 45506 }, { "epoch": 2.824942578682724, "grad_norm": 0.3548982044801735, "learning_rate": 1.033658766331047e-06, "loss": 2.7188, "step": 45507 }, { "epoch": 2.825004655782482, "grad_norm": 0.3468972718761108, "learning_rate": 1.0329283391360022e-06, "loss": 2.6195, "step": 45508 }, { "epoch": 2.82506673288224, "grad_norm": 0.3706886690655496, "learning_rate": 1.0321981674140558e-06, "loss": 2.6705, "step": 45509 }, { "epoch": 2.825128809981998, "grad_norm": 0.34393071712825635, "learning_rate": 1.0314682511690266e-06, "loss": 2.6647, "step": 45510 }, { "epoch": 2.8251908870817557, "grad_norm": 0.37526496824442207, "learning_rate": 1.0307385904047119e-06, "loss": 2.7225, "step": 45511 }, { "epoch": 2.825252964181513, "grad_norm": 0.35606792799896997, "learning_rate": 1.0300091851249194e-06, "loss": 2.6939, "step": 45512 }, { "epoch": 2.8253150412812715, "grad_norm": 0.35421130188934763, "learning_rate": 1.0292800353334575e-06, "loss": 2.6468, "step": 45513 }, { "epoch": 2.825377118381029, "grad_norm": 0.34374756161951997, "learning_rate": 1.0285511410341285e-06, "loss": 2.6337, "step": 45514 }, { "epoch": 2.8254391954807874, "grad_norm": 0.3518825451768633, "learning_rate": 1.0278225022307352e-06, "loss": 2.6373, "step": 45515 }, { "epoch": 2.825501272580545, "grad_norm": 0.3466854685986403, "learning_rate": 1.0270941189270855e-06, "loss": 2.6646, "step": 45516 }, { "epoch": 2.825563349680303, "grad_norm": 0.3455194534919764, "learning_rate": 1.0263659911269707e-06, "loss": 2.7023, "step": 45517 }, { "epoch": 2.8256254267800607, "grad_norm": 0.35105744217008006, "learning_rate": 1.0256381188341934e-06, "loss": 2.6026, "step": 45518 }, { "epoch": 2.8256875038798186, "grad_norm": 0.36753445324322065, "learning_rate": 1.0249105020525507e-06, "loss": 2.6858, "step": 45519 }, { "epoch": 2.8257495809795765, "grad_norm": 0.37445702582400187, "learning_rate": 1.0241831407858394e-06, "loss": 2.6804, "step": 45520 }, { "epoch": 2.8258116580793344, "grad_norm": 0.35629570316482834, "learning_rate": 1.0234560350378563e-06, "loss": 2.7077, "step": 45521 }, { "epoch": 2.8258737351790923, "grad_norm": 0.3524404794750041, "learning_rate": 1.0227291848123933e-06, "loss": 2.8255, "step": 45522 }, { "epoch": 2.8259358122788503, "grad_norm": 0.3417596524270048, "learning_rate": 1.0220025901132358e-06, "loss": 2.6035, "step": 45523 }, { "epoch": 2.825997889378608, "grad_norm": 0.3548096215496054, "learning_rate": 1.0212762509441865e-06, "loss": 2.6533, "step": 45524 }, { "epoch": 2.826059966478366, "grad_norm": 0.3360160913133789, "learning_rate": 1.0205501673090256e-06, "loss": 2.6606, "step": 45525 }, { "epoch": 2.826122043578124, "grad_norm": 0.3621990121718427, "learning_rate": 1.0198243392115503e-06, "loss": 2.6094, "step": 45526 }, { "epoch": 2.826184120677882, "grad_norm": 0.34037268243954, "learning_rate": 1.0190987666555297e-06, "loss": 2.6874, "step": 45527 }, { "epoch": 2.82624619777764, "grad_norm": 0.354415233655954, "learning_rate": 1.0183734496447661e-06, "loss": 2.6258, "step": 45528 }, { "epoch": 2.8263082748773978, "grad_norm": 0.348991305895372, "learning_rate": 1.0176483881830457e-06, "loss": 2.5694, "step": 45529 }, { "epoch": 2.8263703519771557, "grad_norm": 0.3455995630868329, "learning_rate": 1.0169235822741375e-06, "loss": 2.6581, "step": 45530 }, { "epoch": 2.8264324290769136, "grad_norm": 0.3534946119270142, "learning_rate": 1.0161990319218383e-06, "loss": 2.7324, "step": 45531 }, { "epoch": 2.8264945061766715, "grad_norm": 0.32904351059189785, "learning_rate": 1.0154747371299067e-06, "loss": 2.6247, "step": 45532 }, { "epoch": 2.8265565832764294, "grad_norm": 0.36742237370867004, "learning_rate": 1.014750697902145e-06, "loss": 2.5918, "step": 45533 }, { "epoch": 2.8266186603761874, "grad_norm": 0.3657922818533986, "learning_rate": 1.014026914242322e-06, "loss": 2.6017, "step": 45534 }, { "epoch": 2.8266807374759453, "grad_norm": 0.35657918845865144, "learning_rate": 1.0133033861542073e-06, "loss": 2.5863, "step": 45535 }, { "epoch": 2.826742814575703, "grad_norm": 0.34630665656424625, "learning_rate": 1.0125801136415868e-06, "loss": 2.6459, "step": 45536 }, { "epoch": 2.8268048916754607, "grad_norm": 0.34754589558165694, "learning_rate": 1.0118570967082187e-06, "loss": 2.6438, "step": 45537 }, { "epoch": 2.826866968775219, "grad_norm": 0.35407356366129683, "learning_rate": 1.0111343353578884e-06, "loss": 2.5961, "step": 45538 }, { "epoch": 2.8269290458749765, "grad_norm": 0.35080394501536694, "learning_rate": 1.0104118295943655e-06, "loss": 2.719, "step": 45539 }, { "epoch": 2.826991122974735, "grad_norm": 0.34371611037958844, "learning_rate": 1.0096895794214133e-06, "loss": 2.6286, "step": 45540 }, { "epoch": 2.8270532000744923, "grad_norm": 0.3467634610249642, "learning_rate": 1.0089675848428015e-06, "loss": 2.6503, "step": 45541 }, { "epoch": 2.8271152771742503, "grad_norm": 0.37000450089655423, "learning_rate": 1.0082458458623044e-06, "loss": 2.6721, "step": 45542 }, { "epoch": 2.827177354274008, "grad_norm": 0.3711794904739773, "learning_rate": 1.0075243624836805e-06, "loss": 2.6843, "step": 45543 }, { "epoch": 2.827239431373766, "grad_norm": 0.37437301116185817, "learning_rate": 1.0068031347106932e-06, "loss": 2.6466, "step": 45544 }, { "epoch": 2.827301508473524, "grad_norm": 0.35558790787808114, "learning_rate": 1.0060821625471063e-06, "loss": 2.6391, "step": 45545 }, { "epoch": 2.827363585573282, "grad_norm": 0.3495021740957646, "learning_rate": 1.0053614459966888e-06, "loss": 2.6924, "step": 45546 }, { "epoch": 2.82742566267304, "grad_norm": 0.3454498613318892, "learning_rate": 1.0046409850631878e-06, "loss": 2.5445, "step": 45547 }, { "epoch": 2.8274877397727978, "grad_norm": 0.3248616474280013, "learning_rate": 1.0039207797503724e-06, "loss": 2.6859, "step": 45548 }, { "epoch": 2.8275498168725557, "grad_norm": 0.3341854783876854, "learning_rate": 1.0032008300619954e-06, "loss": 2.6288, "step": 45549 }, { "epoch": 2.8276118939723136, "grad_norm": 0.3702454961043023, "learning_rate": 1.0024811360018149e-06, "loss": 2.693, "step": 45550 }, { "epoch": 2.8276739710720715, "grad_norm": 0.3441418668311031, "learning_rate": 1.0017616975735833e-06, "loss": 2.6869, "step": 45551 }, { "epoch": 2.8277360481718294, "grad_norm": 0.3486982332284681, "learning_rate": 1.0010425147810587e-06, "loss": 2.6706, "step": 45552 }, { "epoch": 2.8277981252715874, "grad_norm": 0.3687807377434373, "learning_rate": 1.0003235876279882e-06, "loss": 2.7658, "step": 45553 }, { "epoch": 2.8278602023713453, "grad_norm": 0.3435309535812727, "learning_rate": 9.996049161181187e-07, "loss": 2.7099, "step": 45554 }, { "epoch": 2.827922279471103, "grad_norm": 0.3513311472587387, "learning_rate": 9.988865002552138e-07, "loss": 2.6157, "step": 45555 }, { "epoch": 2.827984356570861, "grad_norm": 0.3635047102272737, "learning_rate": 9.981683400430097e-07, "loss": 2.6418, "step": 45556 }, { "epoch": 2.828046433670619, "grad_norm": 0.35520720029558867, "learning_rate": 9.974504354852642e-07, "loss": 2.724, "step": 45557 }, { "epoch": 2.828108510770377, "grad_norm": 0.34785130204668097, "learning_rate": 9.967327865857079e-07, "loss": 2.6495, "step": 45558 }, { "epoch": 2.828170587870135, "grad_norm": 0.3596292094202079, "learning_rate": 9.960153933480932e-07, "loss": 2.7109, "step": 45559 }, { "epoch": 2.8282326649698923, "grad_norm": 0.3344128409034635, "learning_rate": 9.95298255776156e-07, "loss": 2.6517, "step": 45560 }, { "epoch": 2.8282947420696507, "grad_norm": 0.3378441949597432, "learning_rate": 9.945813738736542e-07, "loss": 2.5878, "step": 45561 }, { "epoch": 2.828356819169408, "grad_norm": 0.35021248268462074, "learning_rate": 9.938647476443185e-07, "loss": 2.6713, "step": 45562 }, { "epoch": 2.8284188962691665, "grad_norm": 0.32688486669685457, "learning_rate": 9.931483770918848e-07, "loss": 2.6275, "step": 45563 }, { "epoch": 2.828480973368924, "grad_norm": 0.3481889993268642, "learning_rate": 9.924322622200888e-07, "loss": 2.6863, "step": 45564 }, { "epoch": 2.8285430504686824, "grad_norm": 0.32489311221199035, "learning_rate": 9.91716403032672e-07, "loss": 2.6233, "step": 45565 }, { "epoch": 2.82860512756844, "grad_norm": 0.33796926917231557, "learning_rate": 9.910007995333702e-07, "loss": 2.6283, "step": 45566 }, { "epoch": 2.8286672046681978, "grad_norm": 0.34369939735063976, "learning_rate": 9.902854517259141e-07, "loss": 2.6789, "step": 45567 }, { "epoch": 2.8287292817679557, "grad_norm": 0.3693575347928136, "learning_rate": 9.895703596140338e-07, "loss": 2.6861, "step": 45568 }, { "epoch": 2.8287913588677136, "grad_norm": 0.3377189089648608, "learning_rate": 9.888555232014652e-07, "loss": 2.6796, "step": 45569 }, { "epoch": 2.8288534359674715, "grad_norm": 0.339281109683621, "learning_rate": 9.881409424919385e-07, "loss": 2.5389, "step": 45570 }, { "epoch": 2.8289155130672294, "grad_norm": 0.36481206621566115, "learning_rate": 9.874266174891733e-07, "loss": 2.6711, "step": 45571 }, { "epoch": 2.8289775901669874, "grad_norm": 0.34103208810190666, "learning_rate": 9.867125481969053e-07, "loss": 2.6969, "step": 45572 }, { "epoch": 2.8290396672667453, "grad_norm": 0.3603235249350402, "learning_rate": 9.859987346188537e-07, "loss": 2.634, "step": 45573 }, { "epoch": 2.829101744366503, "grad_norm": 0.33347526027232594, "learning_rate": 9.85285176758749e-07, "loss": 2.6536, "step": 45574 }, { "epoch": 2.829163821466261, "grad_norm": 0.34288966281535976, "learning_rate": 9.845718746203048e-07, "loss": 2.6473, "step": 45575 }, { "epoch": 2.829225898566019, "grad_norm": 0.34491109906792283, "learning_rate": 9.838588282072513e-07, "loss": 2.7499, "step": 45576 }, { "epoch": 2.829287975665777, "grad_norm": 0.34374279318103973, "learning_rate": 9.831460375233027e-07, "loss": 2.7601, "step": 45577 }, { "epoch": 2.829350052765535, "grad_norm": 0.3374949657983084, "learning_rate": 9.824335025721831e-07, "loss": 2.6899, "step": 45578 }, { "epoch": 2.829412129865293, "grad_norm": 0.31963798400500937, "learning_rate": 9.817212233576068e-07, "loss": 2.6335, "step": 45579 }, { "epoch": 2.8294742069650507, "grad_norm": 0.3644051343388642, "learning_rate": 9.810091998832925e-07, "loss": 2.668, "step": 45580 }, { "epoch": 2.8295362840648086, "grad_norm": 0.3500803433743631, "learning_rate": 9.80297432152949e-07, "loss": 2.6555, "step": 45581 }, { "epoch": 2.8295983611645665, "grad_norm": 0.3444552123572579, "learning_rate": 9.79585920170295e-07, "loss": 2.6505, "step": 45582 }, { "epoch": 2.8296604382643245, "grad_norm": 0.34254547972730587, "learning_rate": 9.788746639390444e-07, "loss": 2.7275, "step": 45583 }, { "epoch": 2.8297225153640824, "grad_norm": 0.3336743706297475, "learning_rate": 9.781636634629e-07, "loss": 2.5789, "step": 45584 }, { "epoch": 2.82978459246384, "grad_norm": 0.3458311522752091, "learning_rate": 9.774529187455805e-07, "loss": 2.7165, "step": 45585 }, { "epoch": 2.829846669563598, "grad_norm": 0.34665000681016717, "learning_rate": 9.767424297907834e-07, "loss": 2.67, "step": 45586 }, { "epoch": 2.8299087466633557, "grad_norm": 0.35182779686642235, "learning_rate": 9.760321966022279e-07, "loss": 2.5965, "step": 45587 }, { "epoch": 2.829970823763114, "grad_norm": 0.33572157862569824, "learning_rate": 9.753222191836164e-07, "loss": 2.6593, "step": 45588 }, { "epoch": 2.8300329008628715, "grad_norm": 0.3433948112316288, "learning_rate": 9.746124975386461e-07, "loss": 2.6589, "step": 45589 }, { "epoch": 2.8300949779626294, "grad_norm": 0.33227619784871637, "learning_rate": 9.739030316710195e-07, "loss": 2.7472, "step": 45590 }, { "epoch": 2.8301570550623874, "grad_norm": 0.36522915270330625, "learning_rate": 9.731938215844449e-07, "loss": 2.6524, "step": 45591 }, { "epoch": 2.8302191321621453, "grad_norm": 0.34572032303785427, "learning_rate": 9.724848672826247e-07, "loss": 2.6249, "step": 45592 }, { "epoch": 2.830281209261903, "grad_norm": 0.3226585420430127, "learning_rate": 9.71776168769245e-07, "loss": 2.653, "step": 45593 }, { "epoch": 2.830343286361661, "grad_norm": 0.3516629605339039, "learning_rate": 9.710677260480138e-07, "loss": 2.617, "step": 45594 }, { "epoch": 2.830405363461419, "grad_norm": 0.3599739059951937, "learning_rate": 9.703595391226228e-07, "loss": 2.6168, "step": 45595 }, { "epoch": 2.830467440561177, "grad_norm": 0.3324123663612573, "learning_rate": 9.696516079967743e-07, "loss": 2.7366, "step": 45596 }, { "epoch": 2.830529517660935, "grad_norm": 0.35577961669600583, "learning_rate": 9.689439326741544e-07, "loss": 2.6755, "step": 45597 }, { "epoch": 2.8305915947606928, "grad_norm": 0.3340992311779666, "learning_rate": 9.682365131584548e-07, "loss": 2.6318, "step": 45598 }, { "epoch": 2.8306536718604507, "grad_norm": 0.3823219581819744, "learning_rate": 9.675293494533722e-07, "loss": 2.7611, "step": 45599 }, { "epoch": 2.8307157489602086, "grad_norm": 0.3526271593757294, "learning_rate": 9.668224415625815e-07, "loss": 2.585, "step": 45600 }, { "epoch": 2.8307778260599665, "grad_norm": 0.36767861819850906, "learning_rate": 9.66115789489791e-07, "loss": 2.6893, "step": 45601 }, { "epoch": 2.8308399031597244, "grad_norm": 0.32138550537028693, "learning_rate": 9.654093932386754e-07, "loss": 2.5983, "step": 45602 }, { "epoch": 2.8309019802594824, "grad_norm": 0.34859508021710867, "learning_rate": 9.647032528129263e-07, "loss": 2.6537, "step": 45603 }, { "epoch": 2.8309640573592403, "grad_norm": 0.33655711629275376, "learning_rate": 9.63997368216213e-07, "loss": 2.7655, "step": 45604 }, { "epoch": 2.831026134458998, "grad_norm": 0.3618072014382956, "learning_rate": 9.632917394522378e-07, "loss": 2.6509, "step": 45605 }, { "epoch": 2.831088211558756, "grad_norm": 0.3492367852864147, "learning_rate": 9.62586366524676e-07, "loss": 2.7088, "step": 45606 }, { "epoch": 2.831150288658514, "grad_norm": 0.3561389808668717, "learning_rate": 9.61881249437202e-07, "loss": 2.6949, "step": 45607 }, { "epoch": 2.8312123657582715, "grad_norm": 0.3761904021678953, "learning_rate": 9.611763881934965e-07, "loss": 2.6938, "step": 45608 }, { "epoch": 2.83127444285803, "grad_norm": 0.348455164765223, "learning_rate": 9.604717827972398e-07, "loss": 2.6687, "step": 45609 }, { "epoch": 2.8313365199577873, "grad_norm": 0.33224165364683006, "learning_rate": 9.597674332521067e-07, "loss": 2.6733, "step": 45610 }, { "epoch": 2.8313985970575457, "grad_norm": 0.3306965198573661, "learning_rate": 9.59063339561772e-07, "loss": 2.6468, "step": 45611 }, { "epoch": 2.831460674157303, "grad_norm": 0.32638728857775356, "learning_rate": 9.583595017299052e-07, "loss": 2.5957, "step": 45612 }, { "epoch": 2.8315227512570615, "grad_norm": 0.3760170258385573, "learning_rate": 9.576559197601865e-07, "loss": 2.6406, "step": 45613 }, { "epoch": 2.831584828356819, "grad_norm": 0.3393514180113175, "learning_rate": 9.569525936562796e-07, "loss": 2.6201, "step": 45614 }, { "epoch": 2.831646905456577, "grad_norm": 0.34871763378271986, "learning_rate": 9.562495234218593e-07, "loss": 2.7144, "step": 45615 }, { "epoch": 2.831708982556335, "grad_norm": 0.3551813577754886, "learning_rate": 9.555467090605897e-07, "loss": 2.5961, "step": 45616 }, { "epoch": 2.8317710596560928, "grad_norm": 0.3455892472288143, "learning_rate": 9.548441505761395e-07, "loss": 2.7756, "step": 45617 }, { "epoch": 2.8318331367558507, "grad_norm": 0.32379844722799606, "learning_rate": 9.541418479721731e-07, "loss": 2.6472, "step": 45618 }, { "epoch": 2.8318952138556086, "grad_norm": 0.34935661854439853, "learning_rate": 9.534398012523538e-07, "loss": 2.7014, "step": 45619 }, { "epoch": 2.8319572909553665, "grad_norm": 0.35433332177963905, "learning_rate": 9.527380104203454e-07, "loss": 2.6266, "step": 45620 }, { "epoch": 2.8320193680551244, "grad_norm": 0.3737864727225949, "learning_rate": 9.520364754798117e-07, "loss": 2.6369, "step": 45621 }, { "epoch": 2.8320814451548824, "grad_norm": 0.3353029179159395, "learning_rate": 9.513351964344108e-07, "loss": 2.623, "step": 45622 }, { "epoch": 2.8321435222546403, "grad_norm": 0.3601646455460633, "learning_rate": 9.50634173287801e-07, "loss": 2.6756, "step": 45623 }, { "epoch": 2.832205599354398, "grad_norm": 0.3542584786433549, "learning_rate": 9.499334060436349e-07, "loss": 2.6829, "step": 45624 }, { "epoch": 2.832267676454156, "grad_norm": 0.38315936352950625, "learning_rate": 9.492328947055818e-07, "loss": 2.7343, "step": 45625 }, { "epoch": 2.832329753553914, "grad_norm": 0.3312634285418165, "learning_rate": 9.485326392772775e-07, "loss": 2.6853, "step": 45626 }, { "epoch": 2.832391830653672, "grad_norm": 0.349871521171646, "learning_rate": 9.478326397623916e-07, "loss": 2.7504, "step": 45627 }, { "epoch": 2.83245390775343, "grad_norm": 0.3361450707318892, "learning_rate": 9.471328961645709e-07, "loss": 2.6738, "step": 45628 }, { "epoch": 2.832515984853188, "grad_norm": 0.34797326111149723, "learning_rate": 9.464334084874682e-07, "loss": 2.6765, "step": 45629 }, { "epoch": 2.8325780619529457, "grad_norm": 0.3567616143624988, "learning_rate": 9.457341767347306e-07, "loss": 2.6855, "step": 45630 }, { "epoch": 2.8326401390527036, "grad_norm": 0.3627051430696477, "learning_rate": 9.450352009099994e-07, "loss": 2.6754, "step": 45631 }, { "epoch": 2.8327022161524615, "grad_norm": 0.3443440484505127, "learning_rate": 9.44336481016933e-07, "loss": 2.6571, "step": 45632 }, { "epoch": 2.832764293252219, "grad_norm": 0.34692733945418563, "learning_rate": 9.43638017059173e-07, "loss": 2.6661, "step": 45633 }, { "epoch": 2.8328263703519774, "grad_norm": 0.3523716304431204, "learning_rate": 9.429398090403607e-07, "loss": 2.6972, "step": 45634 }, { "epoch": 2.832888447451735, "grad_norm": 0.35946973298409407, "learning_rate": 9.422418569641434e-07, "loss": 2.662, "step": 45635 }, { "epoch": 2.832950524551493, "grad_norm": 0.3514605162307223, "learning_rate": 9.415441608341569e-07, "loss": 2.6953, "step": 45636 }, { "epoch": 2.8330126016512507, "grad_norm": 0.3348113301434569, "learning_rate": 9.408467206540428e-07, "loss": 2.7027, "step": 45637 }, { "epoch": 2.8330746787510086, "grad_norm": 0.33947673798884603, "learning_rate": 9.401495364274426e-07, "loss": 2.6928, "step": 45638 }, { "epoch": 2.8331367558507665, "grad_norm": 0.365785152865857, "learning_rate": 9.394526081579924e-07, "loss": 2.6945, "step": 45639 }, { "epoch": 2.8331988329505244, "grad_norm": 0.3476308320429685, "learning_rate": 9.387559358493225e-07, "loss": 2.6814, "step": 45640 }, { "epoch": 2.8332609100502824, "grad_norm": 0.3523623057721364, "learning_rate": 9.380595195050801e-07, "loss": 2.6741, "step": 45641 }, { "epoch": 2.8333229871500403, "grad_norm": 0.32920634245881064, "learning_rate": 9.373633591288899e-07, "loss": 2.7006, "step": 45642 }, { "epoch": 2.833385064249798, "grad_norm": 0.3486936072761539, "learning_rate": 9.366674547243825e-07, "loss": 2.6351, "step": 45643 }, { "epoch": 2.833447141349556, "grad_norm": 0.3578395591732293, "learning_rate": 9.359718062951939e-07, "loss": 2.622, "step": 45644 }, { "epoch": 2.833509218449314, "grad_norm": 0.3507612316137343, "learning_rate": 9.352764138449488e-07, "loss": 2.7245, "step": 45645 }, { "epoch": 2.833571295549072, "grad_norm": 0.3603548499476887, "learning_rate": 9.345812773772778e-07, "loss": 2.6312, "step": 45646 }, { "epoch": 2.83363337264883, "grad_norm": 0.37363573104109365, "learning_rate": 9.338863968958111e-07, "loss": 2.7345, "step": 45647 }, { "epoch": 2.833695449748588, "grad_norm": 0.35207772118800007, "learning_rate": 9.331917724041684e-07, "loss": 2.6323, "step": 45648 }, { "epoch": 2.8337575268483457, "grad_norm": 0.3441993312579003, "learning_rate": 9.324974039059686e-07, "loss": 2.6291, "step": 45649 }, { "epoch": 2.8338196039481036, "grad_norm": 0.34185901621915615, "learning_rate": 9.318032914048536e-07, "loss": 2.6257, "step": 45650 }, { "epoch": 2.8338816810478615, "grad_norm": 0.36424335113536543, "learning_rate": 9.311094349044258e-07, "loss": 2.7636, "step": 45651 }, { "epoch": 2.8339437581476195, "grad_norm": 0.32612076090436865, "learning_rate": 9.304158344083103e-07, "loss": 2.7673, "step": 45652 }, { "epoch": 2.8340058352473774, "grad_norm": 0.34669452689204944, "learning_rate": 9.297224899201262e-07, "loss": 2.6768, "step": 45653 }, { "epoch": 2.8340679123471353, "grad_norm": 0.35444125815847466, "learning_rate": 9.290294014434986e-07, "loss": 2.6992, "step": 45654 }, { "epoch": 2.834129989446893, "grad_norm": 0.37042360530286145, "learning_rate": 9.283365689820356e-07, "loss": 2.769, "step": 45655 }, { "epoch": 2.8341920665466507, "grad_norm": 0.33776688928610593, "learning_rate": 9.276439925393509e-07, "loss": 2.6402, "step": 45656 }, { "epoch": 2.834254143646409, "grad_norm": 0.34892072996918777, "learning_rate": 9.26951672119064e-07, "loss": 2.6028, "step": 45657 }, { "epoch": 2.8343162207461665, "grad_norm": 0.3680223753906099, "learning_rate": 9.262596077247777e-07, "loss": 2.6482, "step": 45658 }, { "epoch": 2.834378297845925, "grad_norm": 0.3351349225850084, "learning_rate": 9.255677993601053e-07, "loss": 2.6773, "step": 45659 }, { "epoch": 2.8344403749456824, "grad_norm": 0.3648286493676577, "learning_rate": 9.248762470286665e-07, "loss": 2.6754, "step": 45660 }, { "epoch": 2.8345024520454407, "grad_norm": 0.3664138028600674, "learning_rate": 9.241849507340639e-07, "loss": 2.6615, "step": 45661 }, { "epoch": 2.834564529145198, "grad_norm": 0.36106091490759634, "learning_rate": 9.234939104798945e-07, "loss": 2.6945, "step": 45662 }, { "epoch": 2.834626606244956, "grad_norm": 0.3222363892211912, "learning_rate": 9.228031262697723e-07, "loss": 2.6853, "step": 45663 }, { "epoch": 2.834688683344714, "grad_norm": 0.34849464392685164, "learning_rate": 9.221125981073053e-07, "loss": 2.6527, "step": 45664 }, { "epoch": 2.834750760444472, "grad_norm": 0.34341734265570484, "learning_rate": 9.214223259960852e-07, "loss": 2.628, "step": 45665 }, { "epoch": 2.83481283754423, "grad_norm": 0.33080324960240304, "learning_rate": 9.207323099397258e-07, "loss": 2.698, "step": 45666 }, { "epoch": 2.8348749146439878, "grad_norm": 0.3546887738041124, "learning_rate": 9.20042549941813e-07, "loss": 2.7235, "step": 45667 }, { "epoch": 2.8349369917437457, "grad_norm": 0.37779381253808386, "learning_rate": 9.193530460059551e-07, "loss": 2.6529, "step": 45668 }, { "epoch": 2.8349990688435036, "grad_norm": 0.3579405476363951, "learning_rate": 9.186637981357494e-07, "loss": 2.6907, "step": 45669 }, { "epoch": 2.8350611459432615, "grad_norm": 0.3275541039358688, "learning_rate": 9.17974806334787e-07, "loss": 2.6812, "step": 45670 }, { "epoch": 2.8351232230430194, "grad_norm": 0.35729575928697227, "learning_rate": 9.172860706066655e-07, "loss": 2.7355, "step": 45671 }, { "epoch": 2.8351853001427774, "grad_norm": 0.36305529222443994, "learning_rate": 9.165975909549763e-07, "loss": 2.611, "step": 45672 }, { "epoch": 2.8352473772425353, "grad_norm": 0.36528137231048513, "learning_rate": 9.159093673833164e-07, "loss": 2.7464, "step": 45673 }, { "epoch": 2.835309454342293, "grad_norm": 0.3532541372262099, "learning_rate": 9.152213998952718e-07, "loss": 2.6776, "step": 45674 }, { "epoch": 2.835371531442051, "grad_norm": 0.33218151507005755, "learning_rate": 9.145336884944344e-07, "loss": 2.7286, "step": 45675 }, { "epoch": 2.835433608541809, "grad_norm": 0.3437855225328674, "learning_rate": 9.138462331843844e-07, "loss": 2.6095, "step": 45676 }, { "epoch": 2.835495685641567, "grad_norm": 0.36799943719048583, "learning_rate": 9.131590339687246e-07, "loss": 2.5924, "step": 45677 }, { "epoch": 2.835557762741325, "grad_norm": 0.351248261627294, "learning_rate": 9.124720908510243e-07, "loss": 2.6143, "step": 45678 }, { "epoch": 2.835619839841083, "grad_norm": 0.3618508964707214, "learning_rate": 9.117854038348749e-07, "loss": 2.682, "step": 45679 }, { "epoch": 2.8356819169408407, "grad_norm": 0.34219518069352284, "learning_rate": 9.110989729238517e-07, "loss": 2.7232, "step": 45680 }, { "epoch": 2.835743994040598, "grad_norm": 0.3337164802720562, "learning_rate": 9.104127981215516e-07, "loss": 2.7052, "step": 45681 }, { "epoch": 2.8358060711403565, "grad_norm": 0.3528966595849691, "learning_rate": 9.097268794315383e-07, "loss": 2.5497, "step": 45682 }, { "epoch": 2.835868148240114, "grad_norm": 0.34669491371781547, "learning_rate": 9.090412168574036e-07, "loss": 2.6403, "step": 45683 }, { "epoch": 2.8359302253398724, "grad_norm": 0.35956543562201454, "learning_rate": 9.083558104027112e-07, "loss": 2.6057, "step": 45684 }, { "epoch": 2.83599230243963, "grad_norm": 0.33607667435346233, "learning_rate": 9.076706600710416e-07, "loss": 2.6268, "step": 45685 }, { "epoch": 2.8360543795393878, "grad_norm": 0.363515593715054, "learning_rate": 9.069857658659808e-07, "loss": 2.7334, "step": 45686 }, { "epoch": 2.8361164566391457, "grad_norm": 0.3397494547020618, "learning_rate": 9.063011277910871e-07, "loss": 2.6963, "step": 45687 }, { "epoch": 2.8361785337389036, "grad_norm": 0.35619270047944523, "learning_rate": 9.05616745849941e-07, "loss": 2.673, "step": 45688 }, { "epoch": 2.8362406108386615, "grad_norm": 0.3339755520017722, "learning_rate": 9.049326200461061e-07, "loss": 2.6923, "step": 45689 }, { "epoch": 2.8363026879384194, "grad_norm": 0.3424501116818861, "learning_rate": 9.042487503831632e-07, "loss": 2.7026, "step": 45690 }, { "epoch": 2.8363647650381774, "grad_norm": 0.34623322012417035, "learning_rate": 9.035651368646648e-07, "loss": 2.7185, "step": 45691 }, { "epoch": 2.8364268421379353, "grad_norm": 0.34339320566207715, "learning_rate": 9.028817794941913e-07, "loss": 2.6804, "step": 45692 }, { "epoch": 2.836488919237693, "grad_norm": 0.3449433485282835, "learning_rate": 9.021986782753012e-07, "loss": 2.6843, "step": 45693 }, { "epoch": 2.836550996337451, "grad_norm": 0.32665914127665446, "learning_rate": 9.01515833211558e-07, "loss": 2.5558, "step": 45694 }, { "epoch": 2.836613073437209, "grad_norm": 0.35131758530076185, "learning_rate": 9.008332443065259e-07, "loss": 2.5873, "step": 45695 }, { "epoch": 2.836675150536967, "grad_norm": 0.35411179575622675, "learning_rate": 9.001509115637685e-07, "loss": 2.6829, "step": 45696 }, { "epoch": 2.836737227636725, "grad_norm": 0.3497094804879892, "learning_rate": 8.994688349868385e-07, "loss": 2.6353, "step": 45697 }, { "epoch": 2.836799304736483, "grad_norm": 0.35327995200511336, "learning_rate": 8.987870145793054e-07, "loss": 2.7361, "step": 45698 }, { "epoch": 2.8368613818362407, "grad_norm": 0.34750467918182165, "learning_rate": 8.981054503447162e-07, "loss": 2.7058, "step": 45699 }, { "epoch": 2.8369234589359986, "grad_norm": 0.34042686553153134, "learning_rate": 8.974241422866292e-07, "loss": 2.6831, "step": 45700 }, { "epoch": 2.8369855360357565, "grad_norm": 0.36106211174984804, "learning_rate": 8.967430904086027e-07, "loss": 2.5838, "step": 45701 }, { "epoch": 2.8370476131355145, "grad_norm": 0.3612523592307591, "learning_rate": 8.960622947141895e-07, "loss": 2.7156, "step": 45702 }, { "epoch": 2.8371096902352724, "grad_norm": 0.35567759572424795, "learning_rate": 8.953817552069365e-07, "loss": 2.5709, "step": 45703 }, { "epoch": 2.83717176733503, "grad_norm": 0.36493564567480263, "learning_rate": 8.947014718904023e-07, "loss": 2.5618, "step": 45704 }, { "epoch": 2.837233844434788, "grad_norm": 0.3389419341306402, "learning_rate": 8.940214447681283e-07, "loss": 2.6165, "step": 45705 }, { "epoch": 2.8372959215345457, "grad_norm": 0.36502677198755495, "learning_rate": 8.933416738436673e-07, "loss": 2.6975, "step": 45706 }, { "epoch": 2.837357998634304, "grad_norm": 0.3386820526529718, "learning_rate": 8.926621591205608e-07, "loss": 2.6854, "step": 45707 }, { "epoch": 2.8374200757340615, "grad_norm": 0.36616147533808996, "learning_rate": 8.919829006023562e-07, "loss": 2.6919, "step": 45708 }, { "epoch": 2.8374821528338194, "grad_norm": 0.33395797223583046, "learning_rate": 8.913038982926003e-07, "loss": 2.7027, "step": 45709 }, { "epoch": 2.8375442299335774, "grad_norm": 0.36905404745786247, "learning_rate": 8.906251521948294e-07, "loss": 2.6769, "step": 45710 }, { "epoch": 2.8376063070333353, "grad_norm": 0.3454402419758431, "learning_rate": 8.899466623125963e-07, "loss": 2.6251, "step": 45711 }, { "epoch": 2.837668384133093, "grad_norm": 0.32173519987317656, "learning_rate": 8.892684286494257e-07, "loss": 2.5981, "step": 45712 }, { "epoch": 2.837730461232851, "grad_norm": 0.3521441522338589, "learning_rate": 8.885904512088705e-07, "loss": 2.6467, "step": 45713 }, { "epoch": 2.837792538332609, "grad_norm": 0.34346831224281915, "learning_rate": 8.879127299944557e-07, "loss": 2.6415, "step": 45714 }, { "epoch": 2.837854615432367, "grad_norm": 0.34757932604879965, "learning_rate": 8.872352650097283e-07, "loss": 2.6199, "step": 45715 }, { "epoch": 2.837916692532125, "grad_norm": 0.33999446289378216, "learning_rate": 8.865580562582076e-07, "loss": 2.7127, "step": 45716 }, { "epoch": 2.837978769631883, "grad_norm": 0.3511522760746033, "learning_rate": 8.858811037434467e-07, "loss": 2.5702, "step": 45717 }, { "epoch": 2.8380408467316407, "grad_norm": 0.35885412366596436, "learning_rate": 8.852044074689647e-07, "loss": 2.7191, "step": 45718 }, { "epoch": 2.8381029238313986, "grad_norm": 0.34213361228706707, "learning_rate": 8.845279674382922e-07, "loss": 2.7104, "step": 45719 }, { "epoch": 2.8381650009311565, "grad_norm": 0.3486686068413098, "learning_rate": 8.838517836549598e-07, "loss": 2.627, "step": 45720 }, { "epoch": 2.8382270780309145, "grad_norm": 0.3450973936836979, "learning_rate": 8.831758561224923e-07, "loss": 2.6979, "step": 45721 }, { "epoch": 2.8382891551306724, "grad_norm": 0.3512044566891637, "learning_rate": 8.825001848444258e-07, "loss": 2.5702, "step": 45722 }, { "epoch": 2.8383512322304303, "grad_norm": 0.3257166146577395, "learning_rate": 8.818247698242799e-07, "loss": 2.6226, "step": 45723 }, { "epoch": 2.838413309330188, "grad_norm": 0.39179628244737447, "learning_rate": 8.811496110655737e-07, "loss": 2.7261, "step": 45724 }, { "epoch": 2.838475386429946, "grad_norm": 0.3516049677312345, "learning_rate": 8.804747085718379e-07, "loss": 2.6961, "step": 45725 }, { "epoch": 2.838537463529704, "grad_norm": 0.3392699099477294, "learning_rate": 8.798000623465863e-07, "loss": 2.6975, "step": 45726 }, { "epoch": 2.8385995406294615, "grad_norm": 0.36646251937762586, "learning_rate": 8.791256723933438e-07, "loss": 2.7818, "step": 45727 }, { "epoch": 2.83866161772922, "grad_norm": 0.3533580913769189, "learning_rate": 8.7845153871563e-07, "loss": 2.6741, "step": 45728 }, { "epoch": 2.8387236948289774, "grad_norm": 0.3360806093790522, "learning_rate": 8.777776613169586e-07, "loss": 2.6014, "step": 45729 }, { "epoch": 2.8387857719287357, "grad_norm": 0.35856641112520576, "learning_rate": 8.77104040200838e-07, "loss": 2.7094, "step": 45730 }, { "epoch": 2.838847849028493, "grad_norm": 0.3335281624741793, "learning_rate": 8.764306753707985e-07, "loss": 2.6878, "step": 45731 }, { "epoch": 2.8389099261282515, "grad_norm": 0.34743058456439246, "learning_rate": 8.757575668303431e-07, "loss": 2.6582, "step": 45732 }, { "epoch": 2.838972003228009, "grad_norm": 0.3468436592136989, "learning_rate": 8.75084714582991e-07, "loss": 2.6551, "step": 45733 }, { "epoch": 2.839034080327767, "grad_norm": 0.37638282129731243, "learning_rate": 8.744121186322396e-07, "loss": 2.729, "step": 45734 }, { "epoch": 2.839096157427525, "grad_norm": 0.3434674445540406, "learning_rate": 8.737397789816081e-07, "loss": 2.7204, "step": 45735 }, { "epoch": 2.8391582345272828, "grad_norm": 0.3439167441916527, "learning_rate": 8.730676956346051e-07, "loss": 2.7613, "step": 45736 }, { "epoch": 2.8392203116270407, "grad_norm": 0.3512303478709774, "learning_rate": 8.723958685947331e-07, "loss": 2.6152, "step": 45737 }, { "epoch": 2.8392823887267986, "grad_norm": 0.34149953715556935, "learning_rate": 8.717242978654949e-07, "loss": 2.6355, "step": 45738 }, { "epoch": 2.8393444658265565, "grad_norm": 0.3271027490397116, "learning_rate": 8.710529834503989e-07, "loss": 2.6453, "step": 45739 }, { "epoch": 2.8394065429263144, "grad_norm": 0.3734538469766857, "learning_rate": 8.703819253529421e-07, "loss": 2.6927, "step": 45740 }, { "epoch": 2.8394686200260724, "grad_norm": 0.3385225571838851, "learning_rate": 8.697111235766387e-07, "loss": 2.6516, "step": 45741 }, { "epoch": 2.8395306971258303, "grad_norm": 0.3340830517719542, "learning_rate": 8.690405781249744e-07, "loss": 2.6163, "step": 45742 }, { "epoch": 2.839592774225588, "grad_norm": 0.3283188905520626, "learning_rate": 8.683702890014523e-07, "loss": 2.7229, "step": 45743 }, { "epoch": 2.839654851325346, "grad_norm": 0.33748370678744255, "learning_rate": 8.677002562095637e-07, "loss": 2.6482, "step": 45744 }, { "epoch": 2.839716928425104, "grad_norm": 0.32817500732464294, "learning_rate": 8.670304797528173e-07, "loss": 2.6423, "step": 45745 }, { "epoch": 2.839779005524862, "grad_norm": 0.3500720537472635, "learning_rate": 8.663609596346989e-07, "loss": 2.6497, "step": 45746 }, { "epoch": 2.83984108262462, "grad_norm": 0.3442643608873749, "learning_rate": 8.656916958587058e-07, "loss": 2.6475, "step": 45747 }, { "epoch": 2.839903159724378, "grad_norm": 0.33763849630995124, "learning_rate": 8.650226884283185e-07, "loss": 2.6614, "step": 45748 }, { "epoch": 2.8399652368241357, "grad_norm": 0.336179790062337, "learning_rate": 8.643539373470399e-07, "loss": 2.7334, "step": 45749 }, { "epoch": 2.8400273139238936, "grad_norm": 0.36321731743148106, "learning_rate": 8.636854426183616e-07, "loss": 2.6662, "step": 45750 }, { "epoch": 2.8400893910236515, "grad_norm": 0.34971666021867137, "learning_rate": 8.630172042457585e-07, "loss": 2.6521, "step": 45751 }, { "epoch": 2.840151468123409, "grad_norm": 0.3459875908435876, "learning_rate": 8.623492222327167e-07, "loss": 2.6787, "step": 45752 }, { "epoch": 2.8402135452231674, "grad_norm": 0.34170466677892514, "learning_rate": 8.616814965827391e-07, "loss": 2.6731, "step": 45753 }, { "epoch": 2.840275622322925, "grad_norm": 0.32907849239457276, "learning_rate": 8.61014027299295e-07, "loss": 2.6809, "step": 45754 }, { "epoch": 2.840337699422683, "grad_norm": 0.347862782369674, "learning_rate": 8.60346814385865e-07, "loss": 2.7271, "step": 45755 }, { "epoch": 2.8403997765224407, "grad_norm": 0.35071584690785373, "learning_rate": 8.596798578459409e-07, "loss": 2.6501, "step": 45756 }, { "epoch": 2.8404618536221986, "grad_norm": 0.3553371814754402, "learning_rate": 8.590131576829863e-07, "loss": 2.7163, "step": 45757 }, { "epoch": 2.8405239307219565, "grad_norm": 0.35266327717024487, "learning_rate": 8.583467139004931e-07, "loss": 2.6437, "step": 45758 }, { "epoch": 2.8405860078217144, "grad_norm": 0.35196071959503594, "learning_rate": 8.576805265019416e-07, "loss": 2.6708, "step": 45759 }, { "epoch": 2.8406480849214724, "grad_norm": 0.349096444066704, "learning_rate": 8.57014595490796e-07, "loss": 2.7232, "step": 45760 }, { "epoch": 2.8407101620212303, "grad_norm": 0.3495840777403249, "learning_rate": 8.563489208705366e-07, "loss": 2.6714, "step": 45761 }, { "epoch": 2.840772239120988, "grad_norm": 0.3591552767566574, "learning_rate": 8.55683502644633e-07, "loss": 2.6251, "step": 45762 }, { "epoch": 2.840834316220746, "grad_norm": 0.3582960393076676, "learning_rate": 8.550183408165602e-07, "loss": 2.6966, "step": 45763 }, { "epoch": 2.840896393320504, "grad_norm": 0.37551367863555546, "learning_rate": 8.543534353897875e-07, "loss": 2.623, "step": 45764 }, { "epoch": 2.840958470420262, "grad_norm": 0.36481280144526573, "learning_rate": 8.536887863677845e-07, "loss": 2.6632, "step": 45765 }, { "epoch": 2.84102054752002, "grad_norm": 0.34478786236872194, "learning_rate": 8.530243937540095e-07, "loss": 2.6801, "step": 45766 }, { "epoch": 2.841082624619778, "grad_norm": 0.35527853015774924, "learning_rate": 8.523602575519429e-07, "loss": 2.7114, "step": 45767 }, { "epoch": 2.8411447017195357, "grad_norm": 0.37841580298019356, "learning_rate": 8.516963777650489e-07, "loss": 2.7451, "step": 45768 }, { "epoch": 2.8412067788192936, "grad_norm": 0.3285476256216971, "learning_rate": 8.5103275439678e-07, "loss": 2.6144, "step": 45769 }, { "epoch": 2.8412688559190515, "grad_norm": 0.37423490915470975, "learning_rate": 8.503693874506058e-07, "loss": 2.693, "step": 45770 }, { "epoch": 2.8413309330188095, "grad_norm": 0.36615065014828463, "learning_rate": 8.497062769299846e-07, "loss": 2.6643, "step": 45771 }, { "epoch": 2.8413930101185674, "grad_norm": 0.3493531956121265, "learning_rate": 8.490434228383803e-07, "loss": 2.6768, "step": 45772 }, { "epoch": 2.8414550872183253, "grad_norm": 0.34649830761134254, "learning_rate": 8.483808251792457e-07, "loss": 2.6547, "step": 45773 }, { "epoch": 2.841517164318083, "grad_norm": 0.3726676091476242, "learning_rate": 8.477184839560448e-07, "loss": 2.6994, "step": 45774 }, { "epoch": 2.8415792414178407, "grad_norm": 0.355951735636352, "learning_rate": 8.470563991722247e-07, "loss": 2.6775, "step": 45775 }, { "epoch": 2.841641318517599, "grad_norm": 0.35905411118683767, "learning_rate": 8.463945708312438e-07, "loss": 2.7261, "step": 45776 }, { "epoch": 2.8417033956173565, "grad_norm": 0.35608452082378755, "learning_rate": 8.457329989365548e-07, "loss": 2.7239, "step": 45777 }, { "epoch": 2.841765472717115, "grad_norm": 0.3489386982399279, "learning_rate": 8.450716834916106e-07, "loss": 2.6636, "step": 45778 }, { "epoch": 2.8418275498168724, "grad_norm": 0.337065576954864, "learning_rate": 8.44410624499864e-07, "loss": 2.6493, "step": 45779 }, { "epoch": 2.8418896269166307, "grad_norm": 0.3433981199830425, "learning_rate": 8.437498219647511e-07, "loss": 2.5965, "step": 45780 }, { "epoch": 2.841951704016388, "grad_norm": 0.3563292222862352, "learning_rate": 8.430892758897358e-07, "loss": 2.6467, "step": 45781 }, { "epoch": 2.842013781116146, "grad_norm": 0.33710992620783325, "learning_rate": 8.424289862782542e-07, "loss": 2.6784, "step": 45782 }, { "epoch": 2.842075858215904, "grad_norm": 0.3405141575729643, "learning_rate": 8.417689531337536e-07, "loss": 2.6441, "step": 45783 }, { "epoch": 2.842137935315662, "grad_norm": 0.3590370953090934, "learning_rate": 8.411091764596757e-07, "loss": 2.6476, "step": 45784 }, { "epoch": 2.84220001241542, "grad_norm": 0.36899099395326956, "learning_rate": 8.404496562594677e-07, "loss": 2.7064, "step": 45785 }, { "epoch": 2.842262089515178, "grad_norm": 0.3659696267128646, "learning_rate": 8.397903925365713e-07, "loss": 2.554, "step": 45786 }, { "epoch": 2.8423241666149357, "grad_norm": 0.3472653666818214, "learning_rate": 8.391313852944172e-07, "loss": 2.7205, "step": 45787 }, { "epoch": 2.8423862437146936, "grad_norm": 0.3290308415072247, "learning_rate": 8.384726345364469e-07, "loss": 2.7708, "step": 45788 }, { "epoch": 2.8424483208144515, "grad_norm": 0.3599252013476514, "learning_rate": 8.378141402661022e-07, "loss": 2.6756, "step": 45789 }, { "epoch": 2.8425103979142095, "grad_norm": 0.33549049489972754, "learning_rate": 8.371559024868192e-07, "loss": 2.6899, "step": 45790 }, { "epoch": 2.8425724750139674, "grad_norm": 0.3547535502906019, "learning_rate": 8.36497921202023e-07, "loss": 2.6799, "step": 45791 }, { "epoch": 2.8426345521137253, "grad_norm": 0.3625748236484726, "learning_rate": 8.358401964151552e-07, "loss": 2.7144, "step": 45792 }, { "epoch": 2.842696629213483, "grad_norm": 0.34129142445177785, "learning_rate": 8.351827281296465e-07, "loss": 2.6139, "step": 45793 }, { "epoch": 2.842758706313241, "grad_norm": 0.3259692790667402, "learning_rate": 8.345255163489163e-07, "loss": 2.6811, "step": 45794 }, { "epoch": 2.842820783412999, "grad_norm": 0.32446122581680564, "learning_rate": 8.338685610764119e-07, "loss": 2.5668, "step": 45795 }, { "epoch": 2.842882860512757, "grad_norm": 0.3398095913963031, "learning_rate": 8.332118623155472e-07, "loss": 2.6558, "step": 45796 }, { "epoch": 2.842944937612515, "grad_norm": 0.35280606416858445, "learning_rate": 8.325554200697527e-07, "loss": 2.69, "step": 45797 }, { "epoch": 2.843007014712273, "grad_norm": 0.3664646744662892, "learning_rate": 8.318992343424481e-07, "loss": 2.6554, "step": 45798 }, { "epoch": 2.8430690918120307, "grad_norm": 0.345704676737778, "learning_rate": 8.312433051370638e-07, "loss": 2.6327, "step": 45799 }, { "epoch": 2.843131168911788, "grad_norm": 0.35003355069796854, "learning_rate": 8.305876324570194e-07, "loss": 2.7668, "step": 45800 }, { "epoch": 2.8431932460115465, "grad_norm": 0.342506674402179, "learning_rate": 8.299322163057344e-07, "loss": 2.6736, "step": 45801 }, { "epoch": 2.843255323111304, "grad_norm": 0.3367666613132275, "learning_rate": 8.292770566866281e-07, "loss": 2.6479, "step": 45802 }, { "epoch": 2.8433174002110624, "grad_norm": 0.3338133725531851, "learning_rate": 8.286221536031258e-07, "loss": 2.5753, "step": 45803 }, { "epoch": 2.84337947731082, "grad_norm": 0.3237248646062641, "learning_rate": 8.279675070586357e-07, "loss": 2.6527, "step": 45804 }, { "epoch": 2.8434415544105778, "grad_norm": 0.3572479408503799, "learning_rate": 8.273131170565718e-07, "loss": 2.7047, "step": 45805 }, { "epoch": 2.8435036315103357, "grad_norm": 0.3264662961388362, "learning_rate": 8.266589836003591e-07, "loss": 2.642, "step": 45806 }, { "epoch": 2.8435657086100936, "grad_norm": 0.3485636709343044, "learning_rate": 8.260051066934004e-07, "loss": 2.6284, "step": 45807 }, { "epoch": 2.8436277857098515, "grad_norm": 0.33061563866167615, "learning_rate": 8.253514863391099e-07, "loss": 2.5446, "step": 45808 }, { "epoch": 2.8436898628096094, "grad_norm": 0.3408142023713859, "learning_rate": 8.246981225409011e-07, "loss": 2.6024, "step": 45809 }, { "epoch": 2.8437519399093674, "grad_norm": 0.3441518255656156, "learning_rate": 8.240450153021773e-07, "loss": 2.7682, "step": 45810 }, { "epoch": 2.8438140170091253, "grad_norm": 0.33054998630329774, "learning_rate": 8.233921646263465e-07, "loss": 2.6039, "step": 45811 }, { "epoch": 2.843876094108883, "grad_norm": 0.320728868759197, "learning_rate": 8.227395705168173e-07, "loss": 2.6545, "step": 45812 }, { "epoch": 2.843938171208641, "grad_norm": 0.3489292604757127, "learning_rate": 8.22087232976998e-07, "loss": 2.6561, "step": 45813 }, { "epoch": 2.844000248308399, "grad_norm": 0.3464869863590204, "learning_rate": 8.214351520102859e-07, "loss": 2.768, "step": 45814 }, { "epoch": 2.844062325408157, "grad_norm": 0.35710480180715065, "learning_rate": 8.207833276200893e-07, "loss": 2.7004, "step": 45815 }, { "epoch": 2.844124402507915, "grad_norm": 0.3386491849309647, "learning_rate": 8.201317598097946e-07, "loss": 2.6672, "step": 45816 }, { "epoch": 2.844186479607673, "grad_norm": 0.3839526422375169, "learning_rate": 8.194804485828211e-07, "loss": 2.7128, "step": 45817 }, { "epoch": 2.8442485567074307, "grad_norm": 0.3521043945630613, "learning_rate": 8.188293939425551e-07, "loss": 2.6945, "step": 45818 }, { "epoch": 2.8443106338071886, "grad_norm": 0.3264052532650131, "learning_rate": 8.181785958923938e-07, "loss": 2.6605, "step": 45819 }, { "epoch": 2.8443727109069465, "grad_norm": 0.35925823885656044, "learning_rate": 8.175280544357289e-07, "loss": 2.6305, "step": 45820 }, { "epoch": 2.8444347880067045, "grad_norm": 0.3497478700945163, "learning_rate": 8.168777695759689e-07, "loss": 2.6804, "step": 45821 }, { "epoch": 2.8444968651064624, "grad_norm": 0.3323247397814132, "learning_rate": 8.162277413164887e-07, "loss": 2.7272, "step": 45822 }, { "epoch": 2.84455894220622, "grad_norm": 0.3357431270814789, "learning_rate": 8.155779696606913e-07, "loss": 2.6481, "step": 45823 }, { "epoch": 2.844621019305978, "grad_norm": 0.3491345916486239, "learning_rate": 8.149284546119684e-07, "loss": 2.6222, "step": 45824 }, { "epoch": 2.8446830964057357, "grad_norm": 0.35351803815660854, "learning_rate": 8.142791961736951e-07, "loss": 2.7263, "step": 45825 }, { "epoch": 2.844745173505494, "grad_norm": 0.3566969101143964, "learning_rate": 8.136301943492741e-07, "loss": 2.6722, "step": 45826 }, { "epoch": 2.8448072506052515, "grad_norm": 0.3523431693366229, "learning_rate": 8.129814491420862e-07, "loss": 2.671, "step": 45827 }, { "epoch": 2.84486932770501, "grad_norm": 0.35398704751747984, "learning_rate": 8.123329605555119e-07, "loss": 2.6828, "step": 45828 }, { "epoch": 2.8449314048047674, "grad_norm": 0.36284607284258397, "learning_rate": 8.116847285929374e-07, "loss": 2.5983, "step": 45829 }, { "epoch": 2.8449934819045253, "grad_norm": 0.33104633782133847, "learning_rate": 8.110367532577379e-07, "loss": 2.4711, "step": 45830 }, { "epoch": 2.845055559004283, "grad_norm": 0.32436388283349926, "learning_rate": 8.103890345533105e-07, "loss": 2.7662, "step": 45831 }, { "epoch": 2.845117636104041, "grad_norm": 0.3463613741227554, "learning_rate": 8.097415724830192e-07, "loss": 2.6908, "step": 45832 }, { "epoch": 2.845179713203799, "grad_norm": 0.33401706602488174, "learning_rate": 8.090943670502505e-07, "loss": 2.6389, "step": 45833 }, { "epoch": 2.845241790303557, "grad_norm": 0.3308077766330618, "learning_rate": 8.084474182583734e-07, "loss": 2.6099, "step": 45834 }, { "epoch": 2.845303867403315, "grad_norm": 0.3622758344898507, "learning_rate": 8.078007261107746e-07, "loss": 2.6969, "step": 45835 }, { "epoch": 2.845365944503073, "grad_norm": 0.3464176961051347, "learning_rate": 8.071542906108176e-07, "loss": 2.6452, "step": 45836 }, { "epoch": 2.8454280216028307, "grad_norm": 0.3610071767294409, "learning_rate": 8.065081117618834e-07, "loss": 2.6927, "step": 45837 }, { "epoch": 2.8454900987025886, "grad_norm": 0.3363767901046564, "learning_rate": 8.058621895673302e-07, "loss": 2.6744, "step": 45838 }, { "epoch": 2.8455521758023465, "grad_norm": 0.34005716353458515, "learning_rate": 8.052165240305387e-07, "loss": 2.6646, "step": 45839 }, { "epoch": 2.8456142529021045, "grad_norm": 0.3461694105125588, "learning_rate": 8.045711151548785e-07, "loss": 2.617, "step": 45840 }, { "epoch": 2.8456763300018624, "grad_norm": 0.3188373160912786, "learning_rate": 8.039259629437135e-07, "loss": 2.6377, "step": 45841 }, { "epoch": 2.8457384071016203, "grad_norm": 0.350558787280149, "learning_rate": 8.032810674004077e-07, "loss": 2.6337, "step": 45842 }, { "epoch": 2.845800484201378, "grad_norm": 0.31517120966964723, "learning_rate": 8.02636428528325e-07, "loss": 2.6544, "step": 45843 }, { "epoch": 2.845862561301136, "grad_norm": 0.35138508240623717, "learning_rate": 8.01992046330835e-07, "loss": 2.6229, "step": 45844 }, { "epoch": 2.845924638400894, "grad_norm": 0.3529081024510441, "learning_rate": 8.013479208112906e-07, "loss": 2.6916, "step": 45845 }, { "epoch": 2.845986715500652, "grad_norm": 0.327293546008747, "learning_rate": 8.007040519730613e-07, "loss": 2.6204, "step": 45846 }, { "epoch": 2.84604879260041, "grad_norm": 0.3464446654675556, "learning_rate": 8.000604398195e-07, "loss": 2.6639, "step": 45847 }, { "epoch": 2.8461108697001674, "grad_norm": 0.36651129040494895, "learning_rate": 7.994170843539706e-07, "loss": 2.6927, "step": 45848 }, { "epoch": 2.8461729467999257, "grad_norm": 0.35467390126782594, "learning_rate": 7.98773985579826e-07, "loss": 2.7346, "step": 45849 }, { "epoch": 2.846235023899683, "grad_norm": 0.35178446650197326, "learning_rate": 7.981311435004247e-07, "loss": 2.6614, "step": 45850 }, { "epoch": 2.8462971009994416, "grad_norm": 0.34398582345560963, "learning_rate": 7.974885581191083e-07, "loss": 2.6437, "step": 45851 }, { "epoch": 2.846359178099199, "grad_norm": 0.34049936611197756, "learning_rate": 7.968462294392409e-07, "loss": 2.5939, "step": 45852 }, { "epoch": 2.846421255198957, "grad_norm": 0.3502337671326113, "learning_rate": 7.962041574641755e-07, "loss": 2.7014, "step": 45853 }, { "epoch": 2.846483332298715, "grad_norm": 0.3434857740135562, "learning_rate": 7.955623421972591e-07, "loss": 2.6815, "step": 45854 }, { "epoch": 2.846545409398473, "grad_norm": 0.32588946514268, "learning_rate": 7.949207836418282e-07, "loss": 2.5385, "step": 45855 }, { "epoch": 2.8466074864982307, "grad_norm": 0.3453044855905237, "learning_rate": 7.942794818012523e-07, "loss": 2.6675, "step": 45856 }, { "epoch": 2.8466695635979886, "grad_norm": 0.3471647495812642, "learning_rate": 7.936384366788562e-07, "loss": 2.7257, "step": 45857 }, { "epoch": 2.8467316406977465, "grad_norm": 0.3430531656396084, "learning_rate": 7.929976482779989e-07, "loss": 2.656, "step": 45858 }, { "epoch": 2.8467937177975045, "grad_norm": 0.3562817969269909, "learning_rate": 7.923571166020161e-07, "loss": 2.7024, "step": 45859 }, { "epoch": 2.8468557948972624, "grad_norm": 0.35748676065458995, "learning_rate": 7.917168416542553e-07, "loss": 2.5765, "step": 45860 }, { "epoch": 2.8469178719970203, "grad_norm": 0.33979729082718235, "learning_rate": 7.910768234380472e-07, "loss": 2.7313, "step": 45861 }, { "epoch": 2.846979949096778, "grad_norm": 0.35196584240358386, "learning_rate": 7.904370619567447e-07, "loss": 2.7266, "step": 45862 }, { "epoch": 2.847042026196536, "grad_norm": 0.3422519596996726, "learning_rate": 7.897975572136729e-07, "loss": 2.658, "step": 45863 }, { "epoch": 2.847104103296294, "grad_norm": 0.3521512188539775, "learning_rate": 7.891583092121791e-07, "loss": 2.6687, "step": 45864 }, { "epoch": 2.847166180396052, "grad_norm": 0.34648767446088785, "learning_rate": 7.885193179555883e-07, "loss": 2.7117, "step": 45865 }, { "epoch": 2.84722825749581, "grad_norm": 0.3574407186908549, "learning_rate": 7.878805834472369e-07, "loss": 2.5694, "step": 45866 }, { "epoch": 2.847290334595568, "grad_norm": 0.34240871707911436, "learning_rate": 7.87242105690461e-07, "loss": 2.6886, "step": 45867 }, { "epoch": 2.8473524116953257, "grad_norm": 0.36514813634095716, "learning_rate": 7.866038846885914e-07, "loss": 2.5937, "step": 45868 }, { "epoch": 2.8474144887950836, "grad_norm": 0.332333898053045, "learning_rate": 7.859659204449588e-07, "loss": 2.7125, "step": 45869 }, { "epoch": 2.8474765658948415, "grad_norm": 0.34711858341796387, "learning_rate": 7.853282129628825e-07, "loss": 2.7471, "step": 45870 }, { "epoch": 2.847538642994599, "grad_norm": 0.3232687740398417, "learning_rate": 7.846907622456989e-07, "loss": 2.619, "step": 45871 }, { "epoch": 2.8476007200943574, "grad_norm": 0.34336536736407464, "learning_rate": 7.840535682967332e-07, "loss": 2.6187, "step": 45872 }, { "epoch": 2.847662797194115, "grad_norm": 0.3389537601752509, "learning_rate": 7.834166311193048e-07, "loss": 2.5494, "step": 45873 }, { "epoch": 2.847724874293873, "grad_norm": 0.33711674440181566, "learning_rate": 7.827799507167388e-07, "loss": 2.647, "step": 45874 }, { "epoch": 2.8477869513936307, "grad_norm": 0.33032536699033155, "learning_rate": 7.821435270923605e-07, "loss": 2.6031, "step": 45875 }, { "epoch": 2.847849028493389, "grad_norm": 0.3470537343785935, "learning_rate": 7.815073602494894e-07, "loss": 2.6861, "step": 45876 }, { "epoch": 2.8479111055931465, "grad_norm": 0.3427284406813872, "learning_rate": 7.808714501914394e-07, "loss": 2.6377, "step": 45877 }, { "epoch": 2.8479731826929044, "grad_norm": 0.35359930660316824, "learning_rate": 7.802357969215301e-07, "loss": 2.5425, "step": 45878 }, { "epoch": 2.8480352597926624, "grad_norm": 0.35228440023824276, "learning_rate": 7.796004004430812e-07, "loss": 2.6646, "step": 45879 }, { "epoch": 2.8480973368924203, "grad_norm": 0.3436146057689688, "learning_rate": 7.789652607594067e-07, "loss": 2.6146, "step": 45880 }, { "epoch": 2.848159413992178, "grad_norm": 0.3579562627178686, "learning_rate": 7.783303778738149e-07, "loss": 2.6338, "step": 45881 }, { "epoch": 2.848221491091936, "grad_norm": 0.3664276499080676, "learning_rate": 7.77695751789631e-07, "loss": 2.7199, "step": 45882 }, { "epoch": 2.848283568191694, "grad_norm": 0.3462659381647976, "learning_rate": 7.770613825101469e-07, "loss": 2.5825, "step": 45883 }, { "epoch": 2.848345645291452, "grad_norm": 0.33356654929551044, "learning_rate": 7.764272700386876e-07, "loss": 2.6501, "step": 45884 }, { "epoch": 2.84840772239121, "grad_norm": 0.3498682949354523, "learning_rate": 7.757934143785562e-07, "loss": 2.6263, "step": 45885 }, { "epoch": 2.848469799490968, "grad_norm": 0.35124340410929084, "learning_rate": 7.751598155330663e-07, "loss": 2.6595, "step": 45886 }, { "epoch": 2.8485318765907257, "grad_norm": 0.34906605106552163, "learning_rate": 7.745264735055047e-07, "loss": 2.5582, "step": 45887 }, { "epoch": 2.8485939536904836, "grad_norm": 0.34759279807894194, "learning_rate": 7.738933882991961e-07, "loss": 2.5527, "step": 45888 }, { "epoch": 2.8486560307902415, "grad_norm": 0.3459758760174766, "learning_rate": 7.732605599174325e-07, "loss": 2.6307, "step": 45889 }, { "epoch": 2.8487181078899995, "grad_norm": 0.3347277928166255, "learning_rate": 7.726279883635223e-07, "loss": 2.6374, "step": 45890 }, { "epoch": 2.8487801849897574, "grad_norm": 0.3711469110256797, "learning_rate": 7.71995673640763e-07, "loss": 2.6268, "step": 45891 }, { "epoch": 2.8488422620895153, "grad_norm": 0.363710621409729, "learning_rate": 7.713636157524517e-07, "loss": 2.8165, "step": 45892 }, { "epoch": 2.848904339189273, "grad_norm": 0.35896871239964256, "learning_rate": 7.70731814701886e-07, "loss": 2.7115, "step": 45893 }, { "epoch": 2.848966416289031, "grad_norm": 0.35581456644492415, "learning_rate": 7.701002704923632e-07, "loss": 2.7049, "step": 45894 }, { "epoch": 2.849028493388789, "grad_norm": 0.3535551186098398, "learning_rate": 7.694689831271807e-07, "loss": 2.6338, "step": 45895 }, { "epoch": 2.8490905704885465, "grad_norm": 0.3696557542934798, "learning_rate": 7.688379526096301e-07, "loss": 2.7005, "step": 45896 }, { "epoch": 2.849152647588305, "grad_norm": 0.3295666834853097, "learning_rate": 7.682071789429979e-07, "loss": 2.6122, "step": 45897 }, { "epoch": 2.8492147246880624, "grad_norm": 0.33202084917720026, "learning_rate": 7.675766621305814e-07, "loss": 2.6739, "step": 45898 }, { "epoch": 2.8492768017878207, "grad_norm": 0.3646872395087115, "learning_rate": 7.669464021756723e-07, "loss": 2.5633, "step": 45899 }, { "epoch": 2.849338878887578, "grad_norm": 0.3434155636273462, "learning_rate": 7.66316399081557e-07, "loss": 2.6883, "step": 45900 }, { "epoch": 2.849400955987336, "grad_norm": 0.33727550192573563, "learning_rate": 7.656866528515161e-07, "loss": 2.7568, "step": 45901 }, { "epoch": 2.849463033087094, "grad_norm": 0.33358344613808, "learning_rate": 7.650571634888415e-07, "loss": 2.7603, "step": 45902 }, { "epoch": 2.849525110186852, "grad_norm": 0.35005223872124497, "learning_rate": 7.644279309968194e-07, "loss": 2.5764, "step": 45903 }, { "epoch": 2.84958718728661, "grad_norm": 0.35421407840375874, "learning_rate": 7.637989553787251e-07, "loss": 2.6819, "step": 45904 }, { "epoch": 2.849649264386368, "grad_norm": 0.3400071507641606, "learning_rate": 7.631702366378501e-07, "loss": 2.7333, "step": 45905 }, { "epoch": 2.8497113414861257, "grad_norm": 0.34536347164058934, "learning_rate": 7.625417747774589e-07, "loss": 2.6958, "step": 45906 }, { "epoch": 2.8497734185858836, "grad_norm": 0.36534967489498255, "learning_rate": 7.619135698008428e-07, "loss": 2.6265, "step": 45907 }, { "epoch": 2.8498354956856415, "grad_norm": 0.360187151859558, "learning_rate": 7.61285621711283e-07, "loss": 2.6949, "step": 45908 }, { "epoch": 2.8498975727853995, "grad_norm": 0.3470037315962277, "learning_rate": 7.606579305120431e-07, "loss": 2.6073, "step": 45909 }, { "epoch": 2.8499596498851574, "grad_norm": 0.34040409239226416, "learning_rate": 7.600304962064043e-07, "loss": 2.6334, "step": 45910 }, { "epoch": 2.8500217269849153, "grad_norm": 0.3520087906216644, "learning_rate": 7.594033187976412e-07, "loss": 2.6032, "step": 45911 }, { "epoch": 2.850083804084673, "grad_norm": 0.34584218261879235, "learning_rate": 7.587763982890294e-07, "loss": 2.686, "step": 45912 }, { "epoch": 2.850145881184431, "grad_norm": 0.357520022231731, "learning_rate": 7.581497346838274e-07, "loss": 2.6629, "step": 45913 }, { "epoch": 2.850207958284189, "grad_norm": 0.3647047680752044, "learning_rate": 7.575233279853156e-07, "loss": 2.6292, "step": 45914 }, { "epoch": 2.850270035383947, "grad_norm": 0.33727267433135477, "learning_rate": 7.568971781967582e-07, "loss": 2.6807, "step": 45915 }, { "epoch": 2.850332112483705, "grad_norm": 0.34554269807530447, "learning_rate": 7.562712853214193e-07, "loss": 2.659, "step": 45916 }, { "epoch": 2.850394189583463, "grad_norm": 0.3396397230314728, "learning_rate": 7.556456493625741e-07, "loss": 2.6292, "step": 45917 }, { "epoch": 2.8504562666832207, "grad_norm": 0.3747130130293343, "learning_rate": 7.550202703234754e-07, "loss": 2.7221, "step": 45918 }, { "epoch": 2.850518343782978, "grad_norm": 0.34145873646172187, "learning_rate": 7.543951482073874e-07, "loss": 2.7044, "step": 45919 }, { "epoch": 2.8505804208827366, "grad_norm": 0.33856406382297627, "learning_rate": 7.537702830175797e-07, "loss": 2.6065, "step": 45920 }, { "epoch": 2.850642497982494, "grad_norm": 0.3404615202403552, "learning_rate": 7.531456747573052e-07, "loss": 2.7145, "step": 45921 }, { "epoch": 2.8507045750822524, "grad_norm": 0.3414804356194406, "learning_rate": 7.525213234298279e-07, "loss": 2.5736, "step": 45922 }, { "epoch": 2.85076665218201, "grad_norm": 0.3368085613094763, "learning_rate": 7.518972290384007e-07, "loss": 2.6558, "step": 45923 }, { "epoch": 2.8508287292817682, "grad_norm": 0.3525828072836935, "learning_rate": 7.512733915862769e-07, "loss": 2.6984, "step": 45924 }, { "epoch": 2.8508908063815257, "grad_norm": 0.3531994538600518, "learning_rate": 7.506498110767201e-07, "loss": 2.6672, "step": 45925 }, { "epoch": 2.8509528834812836, "grad_norm": 0.3400171648629349, "learning_rate": 7.500264875129781e-07, "loss": 2.7279, "step": 45926 }, { "epoch": 2.8510149605810415, "grad_norm": 0.33137866571313407, "learning_rate": 7.494034208983036e-07, "loss": 2.6514, "step": 45927 }, { "epoch": 2.8510770376807995, "grad_norm": 0.33946487529333924, "learning_rate": 7.487806112359497e-07, "loss": 2.7152, "step": 45928 }, { "epoch": 2.8511391147805574, "grad_norm": 0.32797864647300873, "learning_rate": 7.481580585291581e-07, "loss": 2.6912, "step": 45929 }, { "epoch": 2.8512011918803153, "grad_norm": 0.35067623537566267, "learning_rate": 7.475357627811874e-07, "loss": 2.6782, "step": 45930 }, { "epoch": 2.851263268980073, "grad_norm": 0.3355427685132172, "learning_rate": 7.469137239952795e-07, "loss": 2.7032, "step": 45931 }, { "epoch": 2.851325346079831, "grad_norm": 0.361222514621586, "learning_rate": 7.462919421746761e-07, "loss": 2.6699, "step": 45932 }, { "epoch": 2.851387423179589, "grad_norm": 0.34393295924429845, "learning_rate": 7.456704173226192e-07, "loss": 2.6446, "step": 45933 }, { "epoch": 2.851449500279347, "grad_norm": 0.3439281608815071, "learning_rate": 7.450491494423672e-07, "loss": 2.6735, "step": 45934 }, { "epoch": 2.851511577379105, "grad_norm": 0.34824672659645206, "learning_rate": 7.444281385371454e-07, "loss": 2.6515, "step": 45935 }, { "epoch": 2.851573654478863, "grad_norm": 0.3715905327645232, "learning_rate": 7.438073846102011e-07, "loss": 2.7078, "step": 45936 }, { "epoch": 2.8516357315786207, "grad_norm": 0.3280297776972606, "learning_rate": 7.431868876647652e-07, "loss": 2.7053, "step": 45937 }, { "epoch": 2.8516978086783786, "grad_norm": 0.3400148202316783, "learning_rate": 7.425666477040904e-07, "loss": 2.6839, "step": 45938 }, { "epoch": 2.8517598857781365, "grad_norm": 0.34559720243176023, "learning_rate": 7.419466647313967e-07, "loss": 2.5968, "step": 45939 }, { "epoch": 2.8518219628778945, "grad_norm": 0.34195262085300276, "learning_rate": 7.413269387499256e-07, "loss": 2.6325, "step": 45940 }, { "epoch": 2.8518840399776524, "grad_norm": 0.3302990777672953, "learning_rate": 7.407074697629135e-07, "loss": 2.7635, "step": 45941 }, { "epoch": 2.8519461170774103, "grad_norm": 0.32837459744028336, "learning_rate": 7.400882577735857e-07, "loss": 2.7004, "step": 45942 }, { "epoch": 2.852008194177168, "grad_norm": 0.32242343391058226, "learning_rate": 7.394693027851729e-07, "loss": 2.6818, "step": 45943 }, { "epoch": 2.8520702712769257, "grad_norm": 0.32598808958450143, "learning_rate": 7.388506048009114e-07, "loss": 2.7317, "step": 45944 }, { "epoch": 2.852132348376684, "grad_norm": 0.3235482901484525, "learning_rate": 7.382321638240263e-07, "loss": 2.6989, "step": 45945 }, { "epoch": 2.8521944254764415, "grad_norm": 0.3483696272736521, "learning_rate": 7.376139798577375e-07, "loss": 2.5861, "step": 45946 }, { "epoch": 2.8522565025762, "grad_norm": 0.3465735044529516, "learning_rate": 7.369960529052755e-07, "loss": 2.643, "step": 45947 }, { "epoch": 2.8523185796759574, "grad_norm": 0.3400529896942242, "learning_rate": 7.363783829698711e-07, "loss": 2.7794, "step": 45948 }, { "epoch": 2.8523806567757153, "grad_norm": 0.3506023010005399, "learning_rate": 7.35760970054733e-07, "loss": 2.6871, "step": 45949 }, { "epoch": 2.852442733875473, "grad_norm": 0.3568948288935454, "learning_rate": 7.351438141630917e-07, "loss": 2.7041, "step": 45950 }, { "epoch": 2.852504810975231, "grad_norm": 0.33423003146912467, "learning_rate": 7.345269152981616e-07, "loss": 2.6435, "step": 45951 }, { "epoch": 2.852566888074989, "grad_norm": 0.3345950271733133, "learning_rate": 7.339102734631675e-07, "loss": 2.6502, "step": 45952 }, { "epoch": 2.852628965174747, "grad_norm": 0.33906516412488197, "learning_rate": 7.332938886613238e-07, "loss": 2.5571, "step": 45953 }, { "epoch": 2.852691042274505, "grad_norm": 0.3225790291623623, "learning_rate": 7.326777608958446e-07, "loss": 2.6971, "step": 45954 }, { "epoch": 2.852753119374263, "grad_norm": 0.3298440588620329, "learning_rate": 7.320618901699439e-07, "loss": 2.6301, "step": 45955 }, { "epoch": 2.8528151964740207, "grad_norm": 0.34764689529348797, "learning_rate": 7.314462764868357e-07, "loss": 2.6516, "step": 45956 }, { "epoch": 2.8528772735737786, "grad_norm": 0.3643838086707482, "learning_rate": 7.308309198497398e-07, "loss": 2.6613, "step": 45957 }, { "epoch": 2.8529393506735365, "grad_norm": 0.35046631198669426, "learning_rate": 7.302158202618537e-07, "loss": 2.6752, "step": 45958 }, { "epoch": 2.8530014277732945, "grad_norm": 0.3438331763467016, "learning_rate": 7.296009777263968e-07, "loss": 2.7144, "step": 45959 }, { "epoch": 2.8530635048730524, "grad_norm": 0.3548880434464272, "learning_rate": 7.289863922465667e-07, "loss": 2.7023, "step": 45960 }, { "epoch": 2.8531255819728103, "grad_norm": 0.34856298693068355, "learning_rate": 7.283720638255776e-07, "loss": 2.6105, "step": 45961 }, { "epoch": 2.853187659072568, "grad_norm": 0.3365359631120773, "learning_rate": 7.277579924666323e-07, "loss": 2.675, "step": 45962 }, { "epoch": 2.853249736172326, "grad_norm": 0.3658665169102904, "learning_rate": 7.271441781729393e-07, "loss": 2.6608, "step": 45963 }, { "epoch": 2.853311813272084, "grad_norm": 0.3431003348547909, "learning_rate": 7.265306209476908e-07, "loss": 2.7257, "step": 45964 }, { "epoch": 2.853373890371842, "grad_norm": 0.35608337002230356, "learning_rate": 7.25917320794095e-07, "loss": 2.688, "step": 45965 }, { "epoch": 2.8534359674716, "grad_norm": 0.3411510578755237, "learning_rate": 7.253042777153496e-07, "loss": 2.6418, "step": 45966 }, { "epoch": 2.8534980445713574, "grad_norm": 0.338063774506361, "learning_rate": 7.246914917146574e-07, "loss": 2.7191, "step": 45967 }, { "epoch": 2.8535601216711157, "grad_norm": 0.337114158587591, "learning_rate": 7.240789627952105e-07, "loss": 2.6691, "step": 45968 }, { "epoch": 2.853622198770873, "grad_norm": 0.3555927479538165, "learning_rate": 7.234666909602006e-07, "loss": 2.6858, "step": 45969 }, { "epoch": 2.8536842758706316, "grad_norm": 0.36636046345172796, "learning_rate": 7.228546762128308e-07, "loss": 2.6486, "step": 45970 }, { "epoch": 2.853746352970389, "grad_norm": 0.34163346880421175, "learning_rate": 7.222429185562929e-07, "loss": 2.7205, "step": 45971 }, { "epoch": 2.8538084300701474, "grad_norm": 0.3312889665500721, "learning_rate": 7.216314179937733e-07, "loss": 2.687, "step": 45972 }, { "epoch": 2.853870507169905, "grad_norm": 0.3451567281233592, "learning_rate": 7.210201745284639e-07, "loss": 2.6193, "step": 45973 }, { "epoch": 2.853932584269663, "grad_norm": 0.3095223904169304, "learning_rate": 7.204091881635567e-07, "loss": 2.6422, "step": 45974 }, { "epoch": 2.8539946613694207, "grad_norm": 0.3478617650056046, "learning_rate": 7.197984589022433e-07, "loss": 2.6599, "step": 45975 }, { "epoch": 2.8540567384691786, "grad_norm": 0.3587961718457643, "learning_rate": 7.191879867476991e-07, "loss": 2.6144, "step": 45976 }, { "epoch": 2.8541188155689365, "grad_norm": 0.3376661558344298, "learning_rate": 7.185777717031162e-07, "loss": 2.6928, "step": 45977 }, { "epoch": 2.8541808926686945, "grad_norm": 0.32605615721769277, "learning_rate": 7.17967813771675e-07, "loss": 2.6569, "step": 45978 }, { "epoch": 2.8542429697684524, "grad_norm": 0.35810026815740476, "learning_rate": 7.173581129565565e-07, "loss": 2.6579, "step": 45979 }, { "epoch": 2.8543050468682103, "grad_norm": 0.30833747942602724, "learning_rate": 7.167486692609526e-07, "loss": 2.6818, "step": 45980 }, { "epoch": 2.854367123967968, "grad_norm": 0.352928010464744, "learning_rate": 7.161394826880275e-07, "loss": 2.7298, "step": 45981 }, { "epoch": 2.854429201067726, "grad_norm": 0.33703374533686714, "learning_rate": 7.155305532409673e-07, "loss": 2.6888, "step": 45982 }, { "epoch": 2.854491278167484, "grad_norm": 0.3370180384204501, "learning_rate": 7.14921880922953e-07, "loss": 2.7164, "step": 45983 }, { "epoch": 2.854553355267242, "grad_norm": 0.35714792477895213, "learning_rate": 7.143134657371542e-07, "loss": 2.7026, "step": 45984 }, { "epoch": 2.854615432367, "grad_norm": 0.3564820362142253, "learning_rate": 7.137053076867462e-07, "loss": 2.6646, "step": 45985 }, { "epoch": 2.854677509466758, "grad_norm": 0.3427701553231059, "learning_rate": 7.130974067748986e-07, "loss": 2.6616, "step": 45986 }, { "epoch": 2.8547395865665157, "grad_norm": 0.3232097667097423, "learning_rate": 7.124897630047867e-07, "loss": 2.6792, "step": 45987 }, { "epoch": 2.8548016636662736, "grad_norm": 0.3337140690467825, "learning_rate": 7.118823763795857e-07, "loss": 2.6783, "step": 45988 }, { "epoch": 2.8548637407660316, "grad_norm": 0.31582000958882883, "learning_rate": 7.112752469024597e-07, "loss": 2.6143, "step": 45989 }, { "epoch": 2.8549258178657895, "grad_norm": 0.34739753655049965, "learning_rate": 7.10668374576573e-07, "loss": 2.6598, "step": 45990 }, { "epoch": 2.8549878949655474, "grad_norm": 0.34320053705724185, "learning_rate": 7.100617594050952e-07, "loss": 2.6351, "step": 45991 }, { "epoch": 2.855049972065305, "grad_norm": 0.33843818574661566, "learning_rate": 7.094554013911902e-07, "loss": 2.6048, "step": 45992 }, { "epoch": 2.8551120491650632, "grad_norm": 0.35715374504250286, "learning_rate": 7.088493005380225e-07, "loss": 2.7857, "step": 45993 }, { "epoch": 2.8551741262648207, "grad_norm": 0.33182335123011625, "learning_rate": 7.082434568487562e-07, "loss": 2.6124, "step": 45994 }, { "epoch": 2.855236203364579, "grad_norm": 0.3470019387478219, "learning_rate": 7.076378703265496e-07, "loss": 2.7274, "step": 45995 }, { "epoch": 2.8552982804643365, "grad_norm": 0.34383741264393464, "learning_rate": 7.070325409745559e-07, "loss": 2.6535, "step": 45996 }, { "epoch": 2.8553603575640945, "grad_norm": 0.34201380793682934, "learning_rate": 7.064274687959449e-07, "loss": 2.6174, "step": 45997 }, { "epoch": 2.8554224346638524, "grad_norm": 0.327566113660037, "learning_rate": 7.058226537938694e-07, "loss": 2.6172, "step": 45998 }, { "epoch": 2.8554845117636103, "grad_norm": 0.3615955497255803, "learning_rate": 7.052180959714827e-07, "loss": 2.7877, "step": 45999 }, { "epoch": 2.855546588863368, "grad_norm": 0.32590070176621105, "learning_rate": 7.046137953319376e-07, "loss": 2.57, "step": 46000 }, { "epoch": 2.855608665963126, "grad_norm": 0.35288492076115374, "learning_rate": 7.040097518783928e-07, "loss": 2.6011, "step": 46001 }, { "epoch": 2.855670743062884, "grad_norm": 0.3614153782425258, "learning_rate": 7.034059656139902e-07, "loss": 2.5947, "step": 46002 }, { "epoch": 2.855732820162642, "grad_norm": 0.33340477426202153, "learning_rate": 7.028024365418939e-07, "loss": 2.6591, "step": 46003 }, { "epoch": 2.8557948972624, "grad_norm": 0.359968772964351, "learning_rate": 7.021991646652404e-07, "loss": 2.7154, "step": 46004 }, { "epoch": 2.855856974362158, "grad_norm": 0.35406828196316986, "learning_rate": 7.015961499871771e-07, "loss": 2.6444, "step": 46005 }, { "epoch": 2.8559190514619157, "grad_norm": 0.3281717948243275, "learning_rate": 7.009933925108625e-07, "loss": 2.5625, "step": 46006 }, { "epoch": 2.8559811285616736, "grad_norm": 0.3480794987657198, "learning_rate": 7.003908922394275e-07, "loss": 2.6562, "step": 46007 }, { "epoch": 2.8560432056614315, "grad_norm": 0.337093891305382, "learning_rate": 6.997886491760253e-07, "loss": 2.6514, "step": 46008 }, { "epoch": 2.8561052827611895, "grad_norm": 0.3527371597602561, "learning_rate": 6.991866633237864e-07, "loss": 2.6554, "step": 46009 }, { "epoch": 2.8561673598609474, "grad_norm": 0.33155387894175375, "learning_rate": 6.985849346858641e-07, "loss": 2.7314, "step": 46010 }, { "epoch": 2.8562294369607053, "grad_norm": 0.3461767174707706, "learning_rate": 6.979834632653948e-07, "loss": 2.6597, "step": 46011 }, { "epoch": 2.856291514060463, "grad_norm": 0.3485804179167175, "learning_rate": 6.973822490655147e-07, "loss": 2.7433, "step": 46012 }, { "epoch": 2.856353591160221, "grad_norm": 0.33022513868285824, "learning_rate": 6.967812920893546e-07, "loss": 2.564, "step": 46013 }, { "epoch": 2.856415668259979, "grad_norm": 0.3643834201762448, "learning_rate": 6.961805923400566e-07, "loss": 2.6758, "step": 46014 }, { "epoch": 2.8564777453597365, "grad_norm": 0.35548677503556897, "learning_rate": 6.955801498207571e-07, "loss": 2.6705, "step": 46015 }, { "epoch": 2.856539822459495, "grad_norm": 0.3672316910144469, "learning_rate": 6.949799645345811e-07, "loss": 2.646, "step": 46016 }, { "epoch": 2.8566018995592524, "grad_norm": 0.3370533530849013, "learning_rate": 6.943800364846653e-07, "loss": 2.7465, "step": 46017 }, { "epoch": 2.8566639766590107, "grad_norm": 0.33500894586162294, "learning_rate": 6.937803656741348e-07, "loss": 2.5879, "step": 46018 }, { "epoch": 2.856726053758768, "grad_norm": 0.35362893074928126, "learning_rate": 6.931809521061261e-07, "loss": 2.6863, "step": 46019 }, { "epoch": 2.8567881308585266, "grad_norm": 0.3918415580363897, "learning_rate": 6.925817957837588e-07, "loss": 2.61, "step": 46020 }, { "epoch": 2.856850207958284, "grad_norm": 0.3455742633547076, "learning_rate": 6.919828967101639e-07, "loss": 2.7088, "step": 46021 }, { "epoch": 2.856912285058042, "grad_norm": 0.3528860186533151, "learning_rate": 6.913842548884663e-07, "loss": 2.716, "step": 46022 }, { "epoch": 2.8569743621578, "grad_norm": 0.34183967322382247, "learning_rate": 6.907858703217806e-07, "loss": 2.643, "step": 46023 }, { "epoch": 2.857036439257558, "grad_norm": 0.35835563133784193, "learning_rate": 6.90187743013243e-07, "loss": 2.522, "step": 46024 }, { "epoch": 2.8570985163573157, "grad_norm": 0.3657037299725101, "learning_rate": 6.895898729659622e-07, "loss": 2.6994, "step": 46025 }, { "epoch": 2.8571605934570736, "grad_norm": 0.35048040645295203, "learning_rate": 6.889922601830634e-07, "loss": 2.6394, "step": 46026 }, { "epoch": 2.8572226705568315, "grad_norm": 0.3569029496132772, "learning_rate": 6.883949046676663e-07, "loss": 2.7494, "step": 46027 }, { "epoch": 2.8572847476565895, "grad_norm": 0.32896560894435994, "learning_rate": 6.877978064228741e-07, "loss": 2.6046, "step": 46028 }, { "epoch": 2.8573468247563474, "grad_norm": 0.3847558267188121, "learning_rate": 6.872009654518174e-07, "loss": 2.7191, "step": 46029 }, { "epoch": 2.8574089018561053, "grad_norm": 0.33991462143927, "learning_rate": 6.866043817576051e-07, "loss": 2.6029, "step": 46030 }, { "epoch": 2.857470978955863, "grad_norm": 0.33440405149016095, "learning_rate": 6.860080553433512e-07, "loss": 2.6613, "step": 46031 }, { "epoch": 2.857533056055621, "grad_norm": 0.3328877117796747, "learning_rate": 6.854119862121533e-07, "loss": 2.7448, "step": 46032 }, { "epoch": 2.857595133155379, "grad_norm": 0.34013319295698097, "learning_rate": 6.848161743671422e-07, "loss": 2.7238, "step": 46033 }, { "epoch": 2.857657210255137, "grad_norm": 0.34790926737921546, "learning_rate": 6.842206198114154e-07, "loss": 2.6657, "step": 46034 }, { "epoch": 2.857719287354895, "grad_norm": 0.37591774220285623, "learning_rate": 6.836253225480815e-07, "loss": 2.6278, "step": 46035 }, { "epoch": 2.857781364454653, "grad_norm": 0.33186373131150115, "learning_rate": 6.830302825802492e-07, "loss": 2.6674, "step": 46036 }, { "epoch": 2.8578434415544107, "grad_norm": 0.3284682907676154, "learning_rate": 6.824354999110161e-07, "loss": 2.596, "step": 46037 }, { "epoch": 2.8579055186541686, "grad_norm": 0.336306214366533, "learning_rate": 6.818409745434906e-07, "loss": 2.5999, "step": 46038 }, { "epoch": 2.8579675957539266, "grad_norm": 0.35832899704608606, "learning_rate": 6.812467064807703e-07, "loss": 2.7536, "step": 46039 }, { "epoch": 2.858029672853684, "grad_norm": 0.34628740063917757, "learning_rate": 6.80652695725964e-07, "loss": 2.6213, "step": 46040 }, { "epoch": 2.8580917499534424, "grad_norm": 0.3529897329469134, "learning_rate": 6.800589422821579e-07, "loss": 2.7136, "step": 46041 }, { "epoch": 2.8581538270532, "grad_norm": 0.348481538153589, "learning_rate": 6.794654461524608e-07, "loss": 2.7214, "step": 46042 }, { "epoch": 2.8582159041529582, "grad_norm": 0.33567848866210215, "learning_rate": 6.788722073399645e-07, "loss": 2.641, "step": 46043 }, { "epoch": 2.8582779812527157, "grad_norm": 0.34145042299522893, "learning_rate": 6.782792258477666e-07, "loss": 2.6782, "step": 46044 }, { "epoch": 2.8583400583524736, "grad_norm": 0.3418279906142413, "learning_rate": 6.776865016789591e-07, "loss": 2.6907, "step": 46045 }, { "epoch": 2.8584021354522315, "grad_norm": 0.338914367733329, "learning_rate": 6.770940348366339e-07, "loss": 2.7614, "step": 46046 }, { "epoch": 2.8584642125519895, "grad_norm": 0.354094942160409, "learning_rate": 6.765018253238831e-07, "loss": 2.7373, "step": 46047 }, { "epoch": 2.8585262896517474, "grad_norm": 0.36014867513969806, "learning_rate": 6.759098731437985e-07, "loss": 2.6433, "step": 46048 }, { "epoch": 2.8585883667515053, "grad_norm": 0.34915160999443134, "learning_rate": 6.753181782994611e-07, "loss": 2.6758, "step": 46049 }, { "epoch": 2.858650443851263, "grad_norm": 0.3488644063093598, "learning_rate": 6.747267407939628e-07, "loss": 2.7322, "step": 46050 }, { "epoch": 2.858712520951021, "grad_norm": 0.34227493640917733, "learning_rate": 6.7413556063039e-07, "loss": 2.647, "step": 46051 }, { "epoch": 2.858774598050779, "grad_norm": 0.3534132561602673, "learning_rate": 6.735446378118293e-07, "loss": 2.5934, "step": 46052 }, { "epoch": 2.858836675150537, "grad_norm": 0.34104195194935427, "learning_rate": 6.729539723413613e-07, "loss": 2.6455, "step": 46053 }, { "epoch": 2.858898752250295, "grad_norm": 0.3489129573110696, "learning_rate": 6.723635642220616e-07, "loss": 2.6653, "step": 46054 }, { "epoch": 2.858960829350053, "grad_norm": 0.3625490130799718, "learning_rate": 6.717734134570164e-07, "loss": 2.6292, "step": 46055 }, { "epoch": 2.8590229064498107, "grad_norm": 0.3784405510840049, "learning_rate": 6.711835200493122e-07, "loss": 2.6091, "step": 46056 }, { "epoch": 2.8590849835495686, "grad_norm": 0.3510278769349467, "learning_rate": 6.705938840020132e-07, "loss": 2.6701, "step": 46057 }, { "epoch": 2.8591470606493266, "grad_norm": 0.3158839943023803, "learning_rate": 6.700045053182003e-07, "loss": 2.7297, "step": 46058 }, { "epoch": 2.8592091377490845, "grad_norm": 0.3550347413941585, "learning_rate": 6.694153840009487e-07, "loss": 2.6666, "step": 46059 }, { "epoch": 2.8592712148488424, "grad_norm": 0.3165909381192217, "learning_rate": 6.688265200533339e-07, "loss": 2.6675, "step": 46060 }, { "epoch": 2.8593332919486003, "grad_norm": 0.3223613481252744, "learning_rate": 6.682379134784256e-07, "loss": 2.6135, "step": 46061 }, { "epoch": 2.8593953690483582, "grad_norm": 0.3224547784726582, "learning_rate": 6.67649564279299e-07, "loss": 2.6879, "step": 46062 }, { "epoch": 2.8594574461481157, "grad_norm": 0.3438304026813644, "learning_rate": 6.670614724590185e-07, "loss": 2.6596, "step": 46063 }, { "epoch": 2.859519523247874, "grad_norm": 0.36248523912302555, "learning_rate": 6.664736380206482e-07, "loss": 2.7325, "step": 46064 }, { "epoch": 2.8595816003476315, "grad_norm": 0.32865577865077844, "learning_rate": 6.65886060967269e-07, "loss": 2.758, "step": 46065 }, { "epoch": 2.85964367744739, "grad_norm": 0.3228449498288039, "learning_rate": 6.652987413019396e-07, "loss": 2.6712, "step": 46066 }, { "epoch": 2.8597057545471474, "grad_norm": 0.3475011844065948, "learning_rate": 6.647116790277186e-07, "loss": 2.6282, "step": 46067 }, { "epoch": 2.8597678316469057, "grad_norm": 0.32592537982675973, "learning_rate": 6.641248741476702e-07, "loss": 2.6888, "step": 46068 }, { "epoch": 2.859829908746663, "grad_norm": 0.32905590748137925, "learning_rate": 6.635383266648643e-07, "loss": 2.6053, "step": 46069 }, { "epoch": 2.859891985846421, "grad_norm": 0.33687972847516434, "learning_rate": 6.629520365823538e-07, "loss": 2.6901, "step": 46070 }, { "epoch": 2.859954062946179, "grad_norm": 0.35461946834866004, "learning_rate": 6.623660039032031e-07, "loss": 2.7034, "step": 46071 }, { "epoch": 2.860016140045937, "grad_norm": 0.3263558422881277, "learning_rate": 6.617802286304598e-07, "loss": 2.634, "step": 46072 }, { "epoch": 2.860078217145695, "grad_norm": 0.35564240206895453, "learning_rate": 6.611947107671934e-07, "loss": 2.699, "step": 46073 }, { "epoch": 2.860140294245453, "grad_norm": 0.3336676049054153, "learning_rate": 6.606094503164462e-07, "loss": 2.6858, "step": 46074 }, { "epoch": 2.8602023713452107, "grad_norm": 0.34484785501848125, "learning_rate": 6.600244472812822e-07, "loss": 2.6858, "step": 46075 }, { "epoch": 2.8602644484449686, "grad_norm": 0.340501160376696, "learning_rate": 6.594397016647435e-07, "loss": 2.6516, "step": 46076 }, { "epoch": 2.8603265255447266, "grad_norm": 0.35727336278728067, "learning_rate": 6.588552134698889e-07, "loss": 2.6392, "step": 46077 }, { "epoch": 2.8603886026444845, "grad_norm": 0.33930704333909606, "learning_rate": 6.582709826997602e-07, "loss": 2.7123, "step": 46078 }, { "epoch": 2.8604506797442424, "grad_norm": 0.3418716239936856, "learning_rate": 6.576870093574161e-07, "loss": 2.6414, "step": 46079 }, { "epoch": 2.8605127568440003, "grad_norm": 0.35010093357568617, "learning_rate": 6.571032934458932e-07, "loss": 2.7091, "step": 46080 }, { "epoch": 2.860574833943758, "grad_norm": 0.3465952595808665, "learning_rate": 6.565198349682388e-07, "loss": 2.5658, "step": 46081 }, { "epoch": 2.860636911043516, "grad_norm": 0.34960603978787863, "learning_rate": 6.559366339275008e-07, "loss": 2.5664, "step": 46082 }, { "epoch": 2.860698988143274, "grad_norm": 0.32219168074787824, "learning_rate": 6.55353690326721e-07, "loss": 2.6579, "step": 46083 }, { "epoch": 2.860761065243032, "grad_norm": 0.35673596798835727, "learning_rate": 6.547710041689415e-07, "loss": 2.7737, "step": 46084 }, { "epoch": 2.86082314234279, "grad_norm": 0.3526704284562587, "learning_rate": 6.541885754571986e-07, "loss": 2.7298, "step": 46085 }, { "epoch": 2.8608852194425474, "grad_norm": 0.3376626695572562, "learning_rate": 6.536064041945345e-07, "loss": 2.6561, "step": 46086 }, { "epoch": 2.8609472965423057, "grad_norm": 0.3368534199414641, "learning_rate": 6.5302449038398e-07, "loss": 2.5769, "step": 46087 }, { "epoch": 2.861009373642063, "grad_norm": 0.3278220344771508, "learning_rate": 6.524428340285826e-07, "loss": 2.6043, "step": 46088 }, { "epoch": 2.8610714507418216, "grad_norm": 0.34224086806902243, "learning_rate": 6.518614351313679e-07, "loss": 2.5736, "step": 46089 }, { "epoch": 2.861133527841579, "grad_norm": 0.3344719100164178, "learning_rate": 6.512802936953721e-07, "loss": 2.66, "step": 46090 }, { "epoch": 2.8611956049413374, "grad_norm": 0.3242187844701542, "learning_rate": 6.506994097236207e-07, "loss": 2.6173, "step": 46091 }, { "epoch": 2.861257682041095, "grad_norm": 0.36300324498189523, "learning_rate": 6.501187832191613e-07, "loss": 2.5927, "step": 46092 }, { "epoch": 2.861319759140853, "grad_norm": 0.3414563799090986, "learning_rate": 6.495384141850081e-07, "loss": 2.7275, "step": 46093 }, { "epoch": 2.8613818362406107, "grad_norm": 0.3537839429766499, "learning_rate": 6.489583026241919e-07, "loss": 2.6969, "step": 46094 }, { "epoch": 2.8614439133403686, "grad_norm": 0.3419871645242118, "learning_rate": 6.483784485397382e-07, "loss": 2.7444, "step": 46095 }, { "epoch": 2.8615059904401265, "grad_norm": 0.316197008246722, "learning_rate": 6.477988519346778e-07, "loss": 2.6564, "step": 46096 }, { "epoch": 2.8615680675398845, "grad_norm": 0.35531956822486865, "learning_rate": 6.472195128120307e-07, "loss": 2.5675, "step": 46097 }, { "epoch": 2.8616301446396424, "grad_norm": 0.31881544292178265, "learning_rate": 6.46640431174822e-07, "loss": 2.5798, "step": 46098 }, { "epoch": 2.8616922217394003, "grad_norm": 0.33663788685246127, "learning_rate": 6.460616070260717e-07, "loss": 2.6186, "step": 46099 }, { "epoch": 2.861754298839158, "grad_norm": 0.3373161238878041, "learning_rate": 6.454830403687939e-07, "loss": 2.5401, "step": 46100 }, { "epoch": 2.861816375938916, "grad_norm": 0.34370711449223446, "learning_rate": 6.449047312060142e-07, "loss": 2.6335, "step": 46101 }, { "epoch": 2.861878453038674, "grad_norm": 0.36365114884364863, "learning_rate": 6.443266795407521e-07, "loss": 2.6849, "step": 46102 }, { "epoch": 2.861940530138432, "grad_norm": 0.32929807959140545, "learning_rate": 6.437488853760165e-07, "loss": 2.6834, "step": 46103 }, { "epoch": 2.86200260723819, "grad_norm": 0.3604563118880853, "learning_rate": 6.431713487148217e-07, "loss": 2.7555, "step": 46104 }, { "epoch": 2.862064684337948, "grad_norm": 0.351889786435008, "learning_rate": 6.425940695601873e-07, "loss": 2.6802, "step": 46105 }, { "epoch": 2.8621267614377057, "grad_norm": 0.3514283878109117, "learning_rate": 6.420170479151222e-07, "loss": 2.6814, "step": 46106 }, { "epoch": 2.8621888385374636, "grad_norm": 0.33047729823823113, "learning_rate": 6.41440283782635e-07, "loss": 2.5648, "step": 46107 }, { "epoch": 2.8622509156372216, "grad_norm": 0.3331298828125, "learning_rate": 6.408637771657345e-07, "loss": 2.6937, "step": 46108 }, { "epoch": 2.8623129927369795, "grad_norm": 0.34132039236650924, "learning_rate": 6.402875280674292e-07, "loss": 2.6143, "step": 46109 }, { "epoch": 2.8623750698367374, "grad_norm": 0.35855319556335125, "learning_rate": 6.397115364907335e-07, "loss": 2.7407, "step": 46110 }, { "epoch": 2.862437146936495, "grad_norm": 0.3351771265921738, "learning_rate": 6.391358024386396e-07, "loss": 2.6775, "step": 46111 }, { "epoch": 2.8624992240362532, "grad_norm": 0.3585886437544941, "learning_rate": 6.38560325914156e-07, "loss": 2.5982, "step": 46112 }, { "epoch": 2.8625613011360107, "grad_norm": 0.32216498219475126, "learning_rate": 6.379851069202858e-07, "loss": 2.6324, "step": 46113 }, { "epoch": 2.862623378235769, "grad_norm": 0.35622999402808353, "learning_rate": 6.374101454600323e-07, "loss": 2.6153, "step": 46114 }, { "epoch": 2.8626854553355265, "grad_norm": 0.33314692480570746, "learning_rate": 6.368354415363987e-07, "loss": 2.6719, "step": 46115 }, { "epoch": 2.8627475324352845, "grad_norm": 0.3254458290156621, "learning_rate": 6.362609951523713e-07, "loss": 2.6121, "step": 46116 }, { "epoch": 2.8628096095350424, "grad_norm": 0.35196367262807404, "learning_rate": 6.356868063109533e-07, "loss": 2.5834, "step": 46117 }, { "epoch": 2.8628716866348003, "grad_norm": 0.3247491106574705, "learning_rate": 6.35112875015137e-07, "loss": 2.6052, "step": 46118 }, { "epoch": 2.862933763734558, "grad_norm": 0.34309092107500033, "learning_rate": 6.345392012679307e-07, "loss": 2.6453, "step": 46119 }, { "epoch": 2.862995840834316, "grad_norm": 0.35796194495497663, "learning_rate": 6.339657850723102e-07, "loss": 2.6626, "step": 46120 }, { "epoch": 2.863057917934074, "grad_norm": 0.33596319277785935, "learning_rate": 6.333926264312783e-07, "loss": 2.6587, "step": 46121 }, { "epoch": 2.863119995033832, "grad_norm": 0.3404631068097692, "learning_rate": 6.328197253478163e-07, "loss": 2.586, "step": 46122 }, { "epoch": 2.86318207213359, "grad_norm": 0.35512637323599405, "learning_rate": 6.322470818249215e-07, "loss": 2.714, "step": 46123 }, { "epoch": 2.863244149233348, "grad_norm": 0.3285517982226159, "learning_rate": 6.316746958655806e-07, "loss": 2.6128, "step": 46124 }, { "epoch": 2.8633062263331057, "grad_norm": 0.33255833694843295, "learning_rate": 6.311025674727744e-07, "loss": 2.6537, "step": 46125 }, { "epoch": 2.8633683034328636, "grad_norm": 0.3434160300807951, "learning_rate": 6.30530696649495e-07, "loss": 2.7291, "step": 46126 }, { "epoch": 2.8634303805326216, "grad_norm": 0.3284682340606101, "learning_rate": 6.299590833987123e-07, "loss": 2.6413, "step": 46127 }, { "epoch": 2.8634924576323795, "grad_norm": 0.3167949572974322, "learning_rate": 6.29387727723424e-07, "loss": 2.7904, "step": 46128 }, { "epoch": 2.8635545347321374, "grad_norm": 0.34001839195322225, "learning_rate": 6.288166296266051e-07, "loss": 2.751, "step": 46129 }, { "epoch": 2.8636166118318953, "grad_norm": 0.3525369173867011, "learning_rate": 6.28245789111237e-07, "loss": 2.6512, "step": 46130 }, { "epoch": 2.8636786889316532, "grad_norm": 0.3349451111371675, "learning_rate": 6.276752061802949e-07, "loss": 2.6645, "step": 46131 }, { "epoch": 2.863740766031411, "grad_norm": 0.33476704360145, "learning_rate": 6.271048808367541e-07, "loss": 2.6896, "step": 46132 }, { "epoch": 2.863802843131169, "grad_norm": 0.3485328573769439, "learning_rate": 6.265348130835957e-07, "loss": 2.7256, "step": 46133 }, { "epoch": 2.8638649202309265, "grad_norm": 0.334180027277655, "learning_rate": 6.259650029237896e-07, "loss": 2.8355, "step": 46134 }, { "epoch": 2.863926997330685, "grad_norm": 0.3364041436355778, "learning_rate": 6.25395450360311e-07, "loss": 2.6997, "step": 46135 }, { "epoch": 2.8639890744304424, "grad_norm": 0.3470372572288328, "learning_rate": 6.248261553961298e-07, "loss": 2.689, "step": 46136 }, { "epoch": 2.8640511515302007, "grad_norm": 0.35277330853271155, "learning_rate": 6.242571180342161e-07, "loss": 2.7161, "step": 46137 }, { "epoch": 2.864113228629958, "grad_norm": 0.34398886660992617, "learning_rate": 6.236883382775393e-07, "loss": 2.589, "step": 46138 }, { "epoch": 2.8641753057297166, "grad_norm": 0.3561617206482354, "learning_rate": 6.231198161290696e-07, "loss": 2.6405, "step": 46139 }, { "epoch": 2.864237382829474, "grad_norm": 0.3331153114405764, "learning_rate": 6.225515515917657e-07, "loss": 2.7307, "step": 46140 }, { "epoch": 2.864299459929232, "grad_norm": 0.33346906024617073, "learning_rate": 6.219835446686029e-07, "loss": 2.6224, "step": 46141 }, { "epoch": 2.86436153702899, "grad_norm": 0.35243717105442207, "learning_rate": 6.214157953625343e-07, "loss": 2.7367, "step": 46142 }, { "epoch": 2.864423614128748, "grad_norm": 0.3441991472658171, "learning_rate": 6.2084830367653e-07, "loss": 2.6274, "step": 46143 }, { "epoch": 2.8644856912285057, "grad_norm": 0.3426280351773654, "learning_rate": 6.202810696135431e-07, "loss": 2.5728, "step": 46144 }, { "epoch": 2.8645477683282636, "grad_norm": 0.3348622079266345, "learning_rate": 6.197140931765377e-07, "loss": 2.7488, "step": 46145 }, { "epoch": 2.8646098454280216, "grad_norm": 0.3388135466321082, "learning_rate": 6.191473743684783e-07, "loss": 2.6497, "step": 46146 }, { "epoch": 2.8646719225277795, "grad_norm": 0.33239891788809184, "learning_rate": 6.185809131923071e-07, "loss": 2.6161, "step": 46147 }, { "epoch": 2.8647339996275374, "grad_norm": 0.3493122457843592, "learning_rate": 6.180147096509936e-07, "loss": 2.6469, "step": 46148 }, { "epoch": 2.8647960767272953, "grad_norm": 0.33324447455765543, "learning_rate": 6.174487637474802e-07, "loss": 2.7491, "step": 46149 }, { "epoch": 2.8648581538270532, "grad_norm": 0.33188971700562536, "learning_rate": 6.168830754847199e-07, "loss": 2.6106, "step": 46150 }, { "epoch": 2.864920230926811, "grad_norm": 0.3576398091268377, "learning_rate": 6.163176448656771e-07, "loss": 2.7058, "step": 46151 }, { "epoch": 2.864982308026569, "grad_norm": 0.35737968154087785, "learning_rate": 6.157524718932883e-07, "loss": 2.666, "step": 46152 }, { "epoch": 2.865044385126327, "grad_norm": 0.33032703607671227, "learning_rate": 6.151875565705123e-07, "loss": 2.6805, "step": 46153 }, { "epoch": 2.865106462226085, "grad_norm": 0.34195610696992745, "learning_rate": 6.146228989002855e-07, "loss": 2.6236, "step": 46154 }, { "epoch": 2.865168539325843, "grad_norm": 0.3259919864029428, "learning_rate": 6.140584988855669e-07, "loss": 2.7436, "step": 46155 }, { "epoch": 2.8652306164256007, "grad_norm": 0.36542112502521346, "learning_rate": 6.134943565292872e-07, "loss": 2.643, "step": 46156 }, { "epoch": 2.8652926935253586, "grad_norm": 0.3371292643243766, "learning_rate": 6.129304718344054e-07, "loss": 2.6306, "step": 46157 }, { "epoch": 2.8653547706251166, "grad_norm": 0.34703838435520895, "learning_rate": 6.123668448038411e-07, "loss": 2.671, "step": 46158 }, { "epoch": 2.865416847724874, "grad_norm": 0.3524420226893183, "learning_rate": 6.118034754405588e-07, "loss": 2.6462, "step": 46159 }, { "epoch": 2.8654789248246324, "grad_norm": 0.3420419958455909, "learning_rate": 6.11240363747484e-07, "loss": 2.6517, "step": 46160 }, { "epoch": 2.86554100192439, "grad_norm": 0.34083750552994696, "learning_rate": 6.106775097275585e-07, "loss": 2.6485, "step": 46161 }, { "epoch": 2.8656030790241482, "grad_norm": 0.3420287953058218, "learning_rate": 6.101149133837191e-07, "loss": 2.6138, "step": 46162 }, { "epoch": 2.8656651561239057, "grad_norm": 0.3387852000673124, "learning_rate": 6.095525747188968e-07, "loss": 2.6252, "step": 46163 }, { "epoch": 2.8657272332236636, "grad_norm": 0.35069420930436984, "learning_rate": 6.089904937360336e-07, "loss": 2.7528, "step": 46164 }, { "epoch": 2.8657893103234215, "grad_norm": 0.33067694073928994, "learning_rate": 6.084286704380493e-07, "loss": 2.6741, "step": 46165 }, { "epoch": 2.8658513874231795, "grad_norm": 0.3562856238104816, "learning_rate": 6.078671048278916e-07, "loss": 2.6794, "step": 46166 }, { "epoch": 2.8659134645229374, "grad_norm": 0.36454924923656373, "learning_rate": 6.07305796908475e-07, "loss": 2.7151, "step": 46167 }, { "epoch": 2.8659755416226953, "grad_norm": 0.3343842661099563, "learning_rate": 6.067447466827358e-07, "loss": 2.6083, "step": 46168 }, { "epoch": 2.866037618722453, "grad_norm": 0.3423273601987457, "learning_rate": 6.061839541536052e-07, "loss": 2.6005, "step": 46169 }, { "epoch": 2.866099695822211, "grad_norm": 0.33723807856200716, "learning_rate": 6.056234193239973e-07, "loss": 2.7431, "step": 46170 }, { "epoch": 2.866161772921969, "grad_norm": 0.33283236851733033, "learning_rate": 6.050631421968433e-07, "loss": 2.7533, "step": 46171 }, { "epoch": 2.866223850021727, "grad_norm": 0.3393312075059013, "learning_rate": 6.045031227750686e-07, "loss": 2.5629, "step": 46172 }, { "epoch": 2.866285927121485, "grad_norm": 0.3627752040451577, "learning_rate": 6.03943361061593e-07, "loss": 2.7182, "step": 46173 }, { "epoch": 2.866348004221243, "grad_norm": 0.33508699897654376, "learning_rate": 6.033838570593309e-07, "loss": 2.6352, "step": 46174 }, { "epoch": 2.8664100813210007, "grad_norm": 0.3319439043832497, "learning_rate": 6.028246107712132e-07, "loss": 2.6459, "step": 46175 }, { "epoch": 2.8664721584207586, "grad_norm": 0.3221055876115759, "learning_rate": 6.022656222001489e-07, "loss": 2.6879, "step": 46176 }, { "epoch": 2.8665342355205166, "grad_norm": 0.3261982455986143, "learning_rate": 6.01706891349052e-07, "loss": 2.7197, "step": 46177 }, { "epoch": 2.8665963126202745, "grad_norm": 0.3269349383693387, "learning_rate": 6.011484182208482e-07, "loss": 2.6776, "step": 46178 }, { "epoch": 2.8666583897200324, "grad_norm": 0.3135610567583946, "learning_rate": 6.005902028184463e-07, "loss": 2.6472, "step": 46179 }, { "epoch": 2.8667204668197903, "grad_norm": 0.34599531078330786, "learning_rate": 6.000322451447493e-07, "loss": 2.7125, "step": 46180 }, { "epoch": 2.8667825439195482, "grad_norm": 0.3662319126480102, "learning_rate": 5.994745452026828e-07, "loss": 2.6292, "step": 46181 }, { "epoch": 2.8668446210193057, "grad_norm": 0.35586586453229013, "learning_rate": 5.989171029951446e-07, "loss": 2.607, "step": 46182 }, { "epoch": 2.866906698119064, "grad_norm": 0.3373429484885705, "learning_rate": 5.983599185250544e-07, "loss": 2.6117, "step": 46183 }, { "epoch": 2.8669687752188215, "grad_norm": 0.3377127096307162, "learning_rate": 5.978029917953099e-07, "loss": 2.6375, "step": 46184 }, { "epoch": 2.86703085231858, "grad_norm": 0.3494141422482736, "learning_rate": 5.972463228088199e-07, "loss": 2.739, "step": 46185 }, { "epoch": 2.8670929294183374, "grad_norm": 0.33124605347423675, "learning_rate": 5.966899115684876e-07, "loss": 2.623, "step": 46186 }, { "epoch": 2.8671550065180957, "grad_norm": 0.35199162466342654, "learning_rate": 5.961337580772219e-07, "loss": 2.7897, "step": 46187 }, { "epoch": 2.867217083617853, "grad_norm": 0.35424053833547337, "learning_rate": 5.955778623379149e-07, "loss": 2.6762, "step": 46188 }, { "epoch": 2.867279160717611, "grad_norm": 0.3675696028431939, "learning_rate": 5.950222243534753e-07, "loss": 2.6464, "step": 46189 }, { "epoch": 2.867341237817369, "grad_norm": 0.35942541680997386, "learning_rate": 5.944668441267953e-07, "loss": 2.7397, "step": 46190 }, { "epoch": 2.867403314917127, "grad_norm": 0.33321329528385124, "learning_rate": 5.939117216607781e-07, "loss": 2.5653, "step": 46191 }, { "epoch": 2.867465392016885, "grad_norm": 0.34875163556470834, "learning_rate": 5.933568569583159e-07, "loss": 2.6447, "step": 46192 }, { "epoch": 2.867527469116643, "grad_norm": 0.34584775151603736, "learning_rate": 5.928022500223063e-07, "loss": 2.6747, "step": 46193 }, { "epoch": 2.8675895462164007, "grad_norm": 0.34400814290514126, "learning_rate": 5.922479008556359e-07, "loss": 2.6693, "step": 46194 }, { "epoch": 2.8676516233161586, "grad_norm": 0.3289612490774616, "learning_rate": 5.916938094612079e-07, "loss": 2.6462, "step": 46195 }, { "epoch": 2.8677137004159166, "grad_norm": 0.3425065003773662, "learning_rate": 5.911399758419089e-07, "loss": 2.6838, "step": 46196 }, { "epoch": 2.8677757775156745, "grad_norm": 0.33416015059562376, "learning_rate": 5.9058640000062e-07, "loss": 2.6776, "step": 46197 }, { "epoch": 2.8678378546154324, "grad_norm": 0.35456418654001887, "learning_rate": 5.900330819402444e-07, "loss": 2.6651, "step": 46198 }, { "epoch": 2.8678999317151903, "grad_norm": 0.3407329348945981, "learning_rate": 5.89480021663652e-07, "loss": 2.7056, "step": 46199 }, { "epoch": 2.8679620088149482, "grad_norm": 0.3295235009003547, "learning_rate": 5.889272191737461e-07, "loss": 2.7029, "step": 46200 }, { "epoch": 2.868024085914706, "grad_norm": 0.3367758094077004, "learning_rate": 5.883746744733964e-07, "loss": 2.5432, "step": 46201 }, { "epoch": 2.868086163014464, "grad_norm": 0.34750595487322494, "learning_rate": 5.878223875654953e-07, "loss": 2.7393, "step": 46202 }, { "epoch": 2.868148240114222, "grad_norm": 0.3484940559948911, "learning_rate": 5.872703584529127e-07, "loss": 2.5898, "step": 46203 }, { "epoch": 2.86821031721398, "grad_norm": 0.32826173294928934, "learning_rate": 5.867185871385405e-07, "loss": 2.5777, "step": 46204 }, { "epoch": 2.868272394313738, "grad_norm": 0.3354043944504845, "learning_rate": 5.861670736252545e-07, "loss": 2.6969, "step": 46205 }, { "epoch": 2.8683344714134957, "grad_norm": 0.3298951381009124, "learning_rate": 5.856158179159299e-07, "loss": 2.6816, "step": 46206 }, { "epoch": 2.868396548513253, "grad_norm": 0.33332315320125844, "learning_rate": 5.850648200134423e-07, "loss": 2.6847, "step": 46207 }, { "epoch": 2.8684586256130116, "grad_norm": 0.3178579559189484, "learning_rate": 5.845140799206672e-07, "loss": 2.6682, "step": 46208 }, { "epoch": 2.868520702712769, "grad_norm": 0.33232141045732627, "learning_rate": 5.839635976404801e-07, "loss": 2.6333, "step": 46209 }, { "epoch": 2.8685827798125274, "grad_norm": 0.35095524794826954, "learning_rate": 5.834133731757508e-07, "loss": 2.6328, "step": 46210 }, { "epoch": 2.868644856912285, "grad_norm": 0.3581641988289085, "learning_rate": 5.828634065293492e-07, "loss": 2.6513, "step": 46211 }, { "epoch": 2.868706934012043, "grad_norm": 0.34509578524052165, "learning_rate": 5.823136977041399e-07, "loss": 2.6567, "step": 46212 }, { "epoch": 2.8687690111118007, "grad_norm": 0.3487125914517246, "learning_rate": 5.817642467030038e-07, "loss": 2.6621, "step": 46213 }, { "epoch": 2.8688310882115586, "grad_norm": 0.33240502580376213, "learning_rate": 5.812150535287997e-07, "loss": 2.6436, "step": 46214 }, { "epoch": 2.8688931653113166, "grad_norm": 0.357160837708575, "learning_rate": 5.80666118184392e-07, "loss": 2.7227, "step": 46215 }, { "epoch": 2.8689552424110745, "grad_norm": 0.3280728730167728, "learning_rate": 5.801174406726451e-07, "loss": 2.7086, "step": 46216 }, { "epoch": 2.8690173195108324, "grad_norm": 0.35265409753740024, "learning_rate": 5.795690209964233e-07, "loss": 2.6505, "step": 46217 }, { "epoch": 2.8690793966105903, "grad_norm": 0.3366420919687812, "learning_rate": 5.790208591585855e-07, "loss": 2.7532, "step": 46218 }, { "epoch": 2.8691414737103482, "grad_norm": 0.3479731968774128, "learning_rate": 5.784729551619961e-07, "loss": 2.6162, "step": 46219 }, { "epoch": 2.869203550810106, "grad_norm": 0.33681315140936374, "learning_rate": 5.779253090095138e-07, "loss": 2.6497, "step": 46220 }, { "epoch": 2.869265627909864, "grad_norm": 0.34034365575304165, "learning_rate": 5.773779207039865e-07, "loss": 2.6951, "step": 46221 }, { "epoch": 2.869327705009622, "grad_norm": 0.3416181478514566, "learning_rate": 5.768307902482784e-07, "loss": 2.6905, "step": 46222 }, { "epoch": 2.86938978210938, "grad_norm": 0.3409902482351371, "learning_rate": 5.762839176452428e-07, "loss": 2.6543, "step": 46223 }, { "epoch": 2.869451859209138, "grad_norm": 0.36391553572931845, "learning_rate": 5.757373028977331e-07, "loss": 2.6311, "step": 46224 }, { "epoch": 2.8695139363088957, "grad_norm": 0.3278543059617084, "learning_rate": 5.751909460086025e-07, "loss": 2.6369, "step": 46225 }, { "epoch": 2.8695760134086536, "grad_norm": 0.3307596991054706, "learning_rate": 5.746448469806931e-07, "loss": 2.6626, "step": 46226 }, { "epoch": 2.8696380905084116, "grad_norm": 0.3515560043582588, "learning_rate": 5.740990058168583e-07, "loss": 2.5727, "step": 46227 }, { "epoch": 2.8697001676081695, "grad_norm": 0.34428226920374594, "learning_rate": 5.735534225199568e-07, "loss": 2.7301, "step": 46228 }, { "epoch": 2.8697622447079274, "grad_norm": 0.34388092105458923, "learning_rate": 5.730080970928198e-07, "loss": 2.6438, "step": 46229 }, { "epoch": 2.869824321807685, "grad_norm": 0.3289330614266343, "learning_rate": 5.724630295382949e-07, "loss": 2.6423, "step": 46230 }, { "epoch": 2.8698863989074432, "grad_norm": 0.33263880177290767, "learning_rate": 5.719182198592354e-07, "loss": 2.6532, "step": 46231 }, { "epoch": 2.8699484760072007, "grad_norm": 0.32720145180293847, "learning_rate": 5.71373668058478e-07, "loss": 2.6006, "step": 46232 }, { "epoch": 2.870010553106959, "grad_norm": 0.3325750273832912, "learning_rate": 5.708293741388592e-07, "loss": 2.6675, "step": 46233 }, { "epoch": 2.8700726302067165, "grad_norm": 0.32540246591627175, "learning_rate": 5.702853381032269e-07, "loss": 2.6424, "step": 46234 }, { "epoch": 2.870134707306475, "grad_norm": 0.33865362907987256, "learning_rate": 5.697415599544121e-07, "loss": 2.6974, "step": 46235 }, { "epoch": 2.8701967844062324, "grad_norm": 0.3200320318514401, "learning_rate": 5.691980396952568e-07, "loss": 2.6985, "step": 46236 }, { "epoch": 2.8702588615059903, "grad_norm": 0.35486935814151477, "learning_rate": 5.686547773285922e-07, "loss": 2.7499, "step": 46237 }, { "epoch": 2.870320938605748, "grad_norm": 0.3516284562865552, "learning_rate": 5.681117728572605e-07, "loss": 2.6589, "step": 46238 }, { "epoch": 2.870383015705506, "grad_norm": 0.3459613288384612, "learning_rate": 5.675690262840872e-07, "loss": 2.6579, "step": 46239 }, { "epoch": 2.870445092805264, "grad_norm": 0.3314919523784151, "learning_rate": 5.670265376119033e-07, "loss": 2.6356, "step": 46240 }, { "epoch": 2.870507169905022, "grad_norm": 0.3327670639436144, "learning_rate": 5.66484306843551e-07, "loss": 2.5951, "step": 46241 }, { "epoch": 2.87056924700478, "grad_norm": 0.3355446114868983, "learning_rate": 5.659423339818448e-07, "loss": 2.6864, "step": 46242 }, { "epoch": 2.870631324104538, "grad_norm": 0.35834118128469905, "learning_rate": 5.654006190296157e-07, "loss": 2.6316, "step": 46243 }, { "epoch": 2.8706934012042957, "grad_norm": 0.34433013556272035, "learning_rate": 5.648591619896948e-07, "loss": 2.6677, "step": 46244 }, { "epoch": 2.8707554783040536, "grad_norm": 0.3478545791178175, "learning_rate": 5.64317962864902e-07, "loss": 2.7419, "step": 46245 }, { "epoch": 2.8708175554038116, "grad_norm": 0.3418041446457661, "learning_rate": 5.637770216580684e-07, "loss": 2.694, "step": 46246 }, { "epoch": 2.8708796325035695, "grad_norm": 0.33819219328796674, "learning_rate": 5.632363383720085e-07, "loss": 2.69, "step": 46247 }, { "epoch": 2.8709417096033274, "grad_norm": 0.33884182486039566, "learning_rate": 5.626959130095422e-07, "loss": 2.636, "step": 46248 }, { "epoch": 2.8710037867030853, "grad_norm": 0.3208598484605185, "learning_rate": 5.62155745573495e-07, "loss": 2.6999, "step": 46249 }, { "epoch": 2.8710658638028432, "grad_norm": 0.3353731049609593, "learning_rate": 5.616158360666868e-07, "loss": 2.6361, "step": 46250 }, { "epoch": 2.871127940902601, "grad_norm": 0.32227128919767756, "learning_rate": 5.610761844919266e-07, "loss": 2.6324, "step": 46251 }, { "epoch": 2.871190018002359, "grad_norm": 0.33401742292049574, "learning_rate": 5.605367908520343e-07, "loss": 2.6438, "step": 46252 }, { "epoch": 2.871252095102117, "grad_norm": 0.32736432368190654, "learning_rate": 5.599976551498187e-07, "loss": 2.6038, "step": 46253 }, { "epoch": 2.871314172201875, "grad_norm": 0.33581813643379343, "learning_rate": 5.594587773881055e-07, "loss": 2.6177, "step": 46254 }, { "epoch": 2.8713762493016324, "grad_norm": 0.3351896967349104, "learning_rate": 5.589201575696922e-07, "loss": 2.6608, "step": 46255 }, { "epoch": 2.8714383264013907, "grad_norm": 0.3524842259245958, "learning_rate": 5.583817956973991e-07, "loss": 2.6723, "step": 46256 }, { "epoch": 2.871500403501148, "grad_norm": 0.3462511557170709, "learning_rate": 5.578436917740293e-07, "loss": 2.6838, "step": 46257 }, { "epoch": 2.8715624806009066, "grad_norm": 0.3276667574033162, "learning_rate": 5.573058458023916e-07, "loss": 2.703, "step": 46258 }, { "epoch": 2.871624557700664, "grad_norm": 0.33419087367283845, "learning_rate": 5.567682577852895e-07, "loss": 2.7048, "step": 46259 }, { "epoch": 2.871686634800422, "grad_norm": 0.3383679959298378, "learning_rate": 5.562309277255317e-07, "loss": 2.687, "step": 46260 }, { "epoch": 2.87174871190018, "grad_norm": 0.33473199967041606, "learning_rate": 5.556938556259216e-07, "loss": 2.6462, "step": 46261 }, { "epoch": 2.871810788999938, "grad_norm": 0.3446226098341274, "learning_rate": 5.55157041489257e-07, "loss": 2.641, "step": 46262 }, { "epoch": 2.8718728660996957, "grad_norm": 0.3250366717519859, "learning_rate": 5.546204853183412e-07, "loss": 2.7507, "step": 46263 }, { "epoch": 2.8719349431994536, "grad_norm": 0.33711219158741973, "learning_rate": 5.540841871159774e-07, "loss": 2.683, "step": 46264 }, { "epoch": 2.8719970202992116, "grad_norm": 0.33935007873206563, "learning_rate": 5.535481468849579e-07, "loss": 2.6618, "step": 46265 }, { "epoch": 2.8720590973989695, "grad_norm": 0.33168879407404184, "learning_rate": 5.530123646280749e-07, "loss": 2.6009, "step": 46266 }, { "epoch": 2.8721211744987274, "grad_norm": 0.34110057205388755, "learning_rate": 5.524768403481373e-07, "loss": 2.6589, "step": 46267 }, { "epoch": 2.8721832515984853, "grad_norm": 0.3383451722959814, "learning_rate": 5.519415740479317e-07, "loss": 2.7531, "step": 46268 }, { "epoch": 2.8722453286982432, "grad_norm": 0.3364495987343971, "learning_rate": 5.514065657302503e-07, "loss": 2.6695, "step": 46269 }, { "epoch": 2.872307405798001, "grad_norm": 0.3450510480853211, "learning_rate": 5.508718153978854e-07, "loss": 2.7079, "step": 46270 }, { "epoch": 2.872369482897759, "grad_norm": 0.33298581853373516, "learning_rate": 5.503373230536236e-07, "loss": 2.7035, "step": 46271 }, { "epoch": 2.872431559997517, "grad_norm": 0.3442364956423935, "learning_rate": 5.49803088700257e-07, "loss": 2.7805, "step": 46272 }, { "epoch": 2.872493637097275, "grad_norm": 0.3433920774195228, "learning_rate": 5.492691123405724e-07, "loss": 2.6302, "step": 46273 }, { "epoch": 2.872555714197033, "grad_norm": 0.34382071634643474, "learning_rate": 5.487353939773565e-07, "loss": 2.6778, "step": 46274 }, { "epoch": 2.8726177912967907, "grad_norm": 0.3246881120949521, "learning_rate": 5.482019336133904e-07, "loss": 2.6583, "step": 46275 }, { "epoch": 2.8726798683965487, "grad_norm": 0.3487064807325438, "learning_rate": 5.476687312514606e-07, "loss": 2.7082, "step": 46276 }, { "epoch": 2.8727419454963066, "grad_norm": 0.3489112810420474, "learning_rate": 5.471357868943538e-07, "loss": 2.6466, "step": 46277 }, { "epoch": 2.872804022596064, "grad_norm": 0.3506520987219825, "learning_rate": 5.466031005448402e-07, "loss": 2.7505, "step": 46278 }, { "epoch": 2.8728660996958224, "grad_norm": 0.3540443976034584, "learning_rate": 5.460706722057008e-07, "loss": 2.724, "step": 46279 }, { "epoch": 2.87292817679558, "grad_norm": 0.3335605296583702, "learning_rate": 5.455385018797166e-07, "loss": 2.6286, "step": 46280 }, { "epoch": 2.8729902538953382, "grad_norm": 0.3588527118374697, "learning_rate": 5.450065895696633e-07, "loss": 2.5687, "step": 46281 }, { "epoch": 2.8730523309950957, "grad_norm": 0.3278300458287176, "learning_rate": 5.444749352783219e-07, "loss": 2.6635, "step": 46282 }, { "epoch": 2.873114408094854, "grad_norm": 0.33562308390832296, "learning_rate": 5.439435390084568e-07, "loss": 2.6823, "step": 46283 }, { "epoch": 2.8731764851946116, "grad_norm": 0.36248651347887556, "learning_rate": 5.434124007628438e-07, "loss": 2.638, "step": 46284 }, { "epoch": 2.8732385622943695, "grad_norm": 0.36376486141612485, "learning_rate": 5.428815205442528e-07, "loss": 2.6643, "step": 46285 }, { "epoch": 2.8733006393941274, "grad_norm": 0.3502516786792398, "learning_rate": 5.423508983554537e-07, "loss": 2.7296, "step": 46286 }, { "epoch": 2.8733627164938853, "grad_norm": 0.34358686130586585, "learning_rate": 5.418205341992221e-07, "loss": 2.6387, "step": 46287 }, { "epoch": 2.8734247935936432, "grad_norm": 0.33934834424780214, "learning_rate": 5.412904280783171e-07, "loss": 2.6687, "step": 46288 }, { "epoch": 2.873486870693401, "grad_norm": 0.33970337742018414, "learning_rate": 5.407605799955029e-07, "loss": 2.6748, "step": 46289 }, { "epoch": 2.873548947793159, "grad_norm": 0.35411705576546026, "learning_rate": 5.402309899535496e-07, "loss": 2.7316, "step": 46290 }, { "epoch": 2.873611024892917, "grad_norm": 0.32943608971548827, "learning_rate": 5.397016579552217e-07, "loss": 2.7209, "step": 46291 }, { "epoch": 2.873673101992675, "grad_norm": 0.33078681888856887, "learning_rate": 5.391725840032724e-07, "loss": 2.7441, "step": 46292 }, { "epoch": 2.873735179092433, "grad_norm": 0.34579998749903385, "learning_rate": 5.386437681004663e-07, "loss": 2.7112, "step": 46293 }, { "epoch": 2.8737972561921907, "grad_norm": 0.33162844290673393, "learning_rate": 5.381152102495679e-07, "loss": 2.574, "step": 46294 }, { "epoch": 2.8738593332919486, "grad_norm": 0.35330459450110496, "learning_rate": 5.375869104533248e-07, "loss": 2.6937, "step": 46295 }, { "epoch": 2.8739214103917066, "grad_norm": 0.3439254421405982, "learning_rate": 5.370588687145017e-07, "loss": 2.6488, "step": 46296 }, { "epoch": 2.8739834874914645, "grad_norm": 0.3462934787423459, "learning_rate": 5.365310850358518e-07, "loss": 2.6476, "step": 46297 }, { "epoch": 2.8740455645912224, "grad_norm": 0.3317419587010934, "learning_rate": 5.36003559420123e-07, "loss": 2.6945, "step": 46298 }, { "epoch": 2.8741076416909803, "grad_norm": 0.330603300258655, "learning_rate": 5.354762918700684e-07, "loss": 2.6877, "step": 46299 }, { "epoch": 2.8741697187907382, "grad_norm": 0.31432166816992324, "learning_rate": 5.349492823884472e-07, "loss": 2.6345, "step": 46300 }, { "epoch": 2.874231795890496, "grad_norm": 0.3469687425396436, "learning_rate": 5.344225309780015e-07, "loss": 2.6876, "step": 46301 }, { "epoch": 2.874293872990254, "grad_norm": 0.32611045716006404, "learning_rate": 5.338960376414847e-07, "loss": 2.7265, "step": 46302 }, { "epoch": 2.8743559500900115, "grad_norm": 0.33253818409231384, "learning_rate": 5.333698023816392e-07, "loss": 2.7591, "step": 46303 }, { "epoch": 2.87441802718977, "grad_norm": 0.33773311626227115, "learning_rate": 5.328438252012069e-07, "loss": 2.6944, "step": 46304 }, { "epoch": 2.8744801042895274, "grad_norm": 0.3439483937044784, "learning_rate": 5.323181061029414e-07, "loss": 2.6738, "step": 46305 }, { "epoch": 2.8745421813892857, "grad_norm": 0.34742796828852135, "learning_rate": 5.317926450895849e-07, "loss": 2.6878, "step": 46306 }, { "epoch": 2.874604258489043, "grad_norm": 0.35083136296153145, "learning_rate": 5.312674421638686e-07, "loss": 2.6203, "step": 46307 }, { "epoch": 2.874666335588801, "grad_norm": 0.3561919892913921, "learning_rate": 5.307424973285402e-07, "loss": 2.7242, "step": 46308 }, { "epoch": 2.874728412688559, "grad_norm": 0.35151255570732703, "learning_rate": 5.302178105863365e-07, "loss": 2.6748, "step": 46309 }, { "epoch": 2.874790489788317, "grad_norm": 0.35821012700183386, "learning_rate": 5.296933819399996e-07, "loss": 2.6031, "step": 46310 }, { "epoch": 2.874852566888075, "grad_norm": 0.3428682701453078, "learning_rate": 5.291692113922553e-07, "loss": 2.6329, "step": 46311 }, { "epoch": 2.874914643987833, "grad_norm": 0.34287131235348767, "learning_rate": 5.286452989458512e-07, "loss": 2.681, "step": 46312 }, { "epoch": 2.8749767210875907, "grad_norm": 0.3553269281536176, "learning_rate": 5.281216446035075e-07, "loss": 2.7199, "step": 46313 }, { "epoch": 2.8750387981873486, "grad_norm": 0.3592026753456593, "learning_rate": 5.27598248367972e-07, "loss": 2.7193, "step": 46314 }, { "epoch": 2.8751008752871066, "grad_norm": 0.3206769219309752, "learning_rate": 5.270751102419591e-07, "loss": 2.6176, "step": 46315 }, { "epoch": 2.8751629523868645, "grad_norm": 0.34330290198681607, "learning_rate": 5.265522302282055e-07, "loss": 2.5659, "step": 46316 }, { "epoch": 2.8752250294866224, "grad_norm": 0.34412906077393485, "learning_rate": 5.260296083294425e-07, "loss": 2.6218, "step": 46317 }, { "epoch": 2.8752871065863803, "grad_norm": 0.3359317446370917, "learning_rate": 5.2550724454839e-07, "loss": 2.7332, "step": 46318 }, { "epoch": 2.8753491836861382, "grad_norm": 0.35250529922736157, "learning_rate": 5.249851388877791e-07, "loss": 2.6519, "step": 46319 }, { "epoch": 2.875411260785896, "grad_norm": 0.3460091997808842, "learning_rate": 5.244632913503355e-07, "loss": 2.722, "step": 46320 }, { "epoch": 2.875473337885654, "grad_norm": 0.337607623530534, "learning_rate": 5.239417019387683e-07, "loss": 2.7497, "step": 46321 }, { "epoch": 2.875535414985412, "grad_norm": 0.35026317604802637, "learning_rate": 5.234203706558138e-07, "loss": 2.6474, "step": 46322 }, { "epoch": 2.87559749208517, "grad_norm": 0.3459290558216018, "learning_rate": 5.228992975041868e-07, "loss": 2.7189, "step": 46323 }, { "epoch": 2.875659569184928, "grad_norm": 0.3406921626730508, "learning_rate": 5.223784824866018e-07, "loss": 2.6121, "step": 46324 }, { "epoch": 2.8757216462846857, "grad_norm": 0.3299333717715158, "learning_rate": 5.218579256057788e-07, "loss": 2.6186, "step": 46325 }, { "epoch": 2.875783723384443, "grad_norm": 0.3322920504281789, "learning_rate": 5.213376268644377e-07, "loss": 2.5697, "step": 46326 }, { "epoch": 2.8758458004842016, "grad_norm": 0.3443676536351106, "learning_rate": 5.208175862652875e-07, "loss": 2.6708, "step": 46327 }, { "epoch": 2.875907877583959, "grad_norm": 0.32462480694277035, "learning_rate": 5.202978038110428e-07, "loss": 2.6795, "step": 46328 }, { "epoch": 2.8759699546837174, "grad_norm": 0.358688382728768, "learning_rate": 5.197782795044126e-07, "loss": 2.5808, "step": 46329 }, { "epoch": 2.876032031783475, "grad_norm": 0.3290243426096157, "learning_rate": 5.192590133481168e-07, "loss": 2.7483, "step": 46330 }, { "epoch": 2.8760941088832332, "grad_norm": 0.3704576763040988, "learning_rate": 5.187400053448587e-07, "loss": 2.7812, "step": 46331 }, { "epoch": 2.8761561859829907, "grad_norm": 0.3351118231718767, "learning_rate": 5.182212554973476e-07, "loss": 2.6796, "step": 46332 }, { "epoch": 2.8762182630827486, "grad_norm": 0.33678912735011274, "learning_rate": 5.17702763808281e-07, "loss": 2.6949, "step": 46333 }, { "epoch": 2.8762803401825066, "grad_norm": 0.32043767437112614, "learning_rate": 5.171845302803735e-07, "loss": 2.555, "step": 46334 }, { "epoch": 2.8763424172822645, "grad_norm": 0.3283422295983645, "learning_rate": 5.166665549163286e-07, "loss": 2.7388, "step": 46335 }, { "epoch": 2.8764044943820224, "grad_norm": 0.3416650027238445, "learning_rate": 5.161488377188495e-07, "loss": 2.684, "step": 46336 }, { "epoch": 2.8764665714817803, "grad_norm": 0.34619826080035815, "learning_rate": 5.156313786906341e-07, "loss": 2.7355, "step": 46337 }, { "epoch": 2.8765286485815382, "grad_norm": 0.32779025986961424, "learning_rate": 5.151141778343749e-07, "loss": 2.7565, "step": 46338 }, { "epoch": 2.876590725681296, "grad_norm": 0.3338621596313385, "learning_rate": 5.145972351527862e-07, "loss": 2.6606, "step": 46339 }, { "epoch": 2.876652802781054, "grad_norm": 0.32672983974424835, "learning_rate": 5.140805506485546e-07, "loss": 2.6358, "step": 46340 }, { "epoch": 2.876714879880812, "grad_norm": 0.33308455626088496, "learning_rate": 5.135641243243783e-07, "loss": 2.6956, "step": 46341 }, { "epoch": 2.87677695698057, "grad_norm": 0.3260859530457661, "learning_rate": 5.130479561829493e-07, "loss": 2.5853, "step": 46342 }, { "epoch": 2.876839034080328, "grad_norm": 0.3368522255582645, "learning_rate": 5.1253204622696e-07, "loss": 2.6367, "step": 46343 }, { "epoch": 2.8769011111800857, "grad_norm": 0.31994557880903496, "learning_rate": 5.120163944591083e-07, "loss": 2.72, "step": 46344 }, { "epoch": 2.8769631882798437, "grad_norm": 0.3402527616943743, "learning_rate": 5.115010008820809e-07, "loss": 2.6352, "step": 46345 }, { "epoch": 2.8770252653796016, "grad_norm": 0.33044725578153916, "learning_rate": 5.109858654985644e-07, "loss": 2.5884, "step": 46346 }, { "epoch": 2.8770873424793595, "grad_norm": 0.3265515590276842, "learning_rate": 5.104709883112513e-07, "loss": 2.6381, "step": 46347 }, { "epoch": 2.8771494195791174, "grad_norm": 0.34532123960249417, "learning_rate": 5.099563693228171e-07, "loss": 2.6597, "step": 46348 }, { "epoch": 2.8772114966788753, "grad_norm": 0.3593204291001804, "learning_rate": 5.094420085359653e-07, "loss": 2.5875, "step": 46349 }, { "epoch": 2.8772735737786332, "grad_norm": 0.32996881243689535, "learning_rate": 5.089279059533658e-07, "loss": 2.6778, "step": 46350 }, { "epoch": 2.8773356508783907, "grad_norm": 0.33310401623179103, "learning_rate": 5.084140615777e-07, "loss": 2.6797, "step": 46351 }, { "epoch": 2.877397727978149, "grad_norm": 0.35081511633498996, "learning_rate": 5.079004754116545e-07, "loss": 2.6576, "step": 46352 }, { "epoch": 2.8774598050779066, "grad_norm": 0.3557081569128875, "learning_rate": 5.07387147457905e-07, "loss": 2.6869, "step": 46353 }, { "epoch": 2.877521882177665, "grad_norm": 0.3270835014917869, "learning_rate": 5.068740777191383e-07, "loss": 2.7503, "step": 46354 }, { "epoch": 2.8775839592774224, "grad_norm": 0.33849701464673404, "learning_rate": 5.063612661980189e-07, "loss": 2.6715, "step": 46355 }, { "epoch": 2.8776460363771803, "grad_norm": 0.3234657367414482, "learning_rate": 5.058487128972278e-07, "loss": 2.6588, "step": 46356 }, { "epoch": 2.8777081134769382, "grad_norm": 0.34321388790448654, "learning_rate": 5.053364178194409e-07, "loss": 2.684, "step": 46357 }, { "epoch": 2.877770190576696, "grad_norm": 0.32619237549824354, "learning_rate": 5.048243809673281e-07, "loss": 2.6626, "step": 46358 }, { "epoch": 2.877832267676454, "grad_norm": 0.33046883251649084, "learning_rate": 5.043126023435651e-07, "loss": 2.6246, "step": 46359 }, { "epoch": 2.877894344776212, "grad_norm": 0.36438354285876434, "learning_rate": 5.038010819508166e-07, "loss": 2.6966, "step": 46360 }, { "epoch": 2.87795642187597, "grad_norm": 0.3136400289746981, "learning_rate": 5.032898197917468e-07, "loss": 2.6205, "step": 46361 }, { "epoch": 2.878018498975728, "grad_norm": 0.3485906986904953, "learning_rate": 5.027788158690372e-07, "loss": 2.6944, "step": 46362 }, { "epoch": 2.8780805760754857, "grad_norm": 0.3604478164765462, "learning_rate": 5.022680701853466e-07, "loss": 2.6639, "step": 46363 }, { "epoch": 2.8781426531752436, "grad_norm": 0.3344529306205785, "learning_rate": 5.017575827433396e-07, "loss": 2.6773, "step": 46364 }, { "epoch": 2.8782047302750016, "grad_norm": 0.3426442785994153, "learning_rate": 5.012473535456752e-07, "loss": 2.6744, "step": 46365 }, { "epoch": 2.8782668073747595, "grad_norm": 0.3129815206526578, "learning_rate": 5.007373825950235e-07, "loss": 2.7088, "step": 46366 }, { "epoch": 2.8783288844745174, "grad_norm": 0.3477816837639545, "learning_rate": 5.002276698940378e-07, "loss": 2.6885, "step": 46367 }, { "epoch": 2.8783909615742753, "grad_norm": 0.3522656508543234, "learning_rate": 4.997182154453828e-07, "loss": 2.6726, "step": 46368 }, { "epoch": 2.8784530386740332, "grad_norm": 0.3410526238085097, "learning_rate": 4.992090192517174e-07, "loss": 2.6105, "step": 46369 }, { "epoch": 2.878515115773791, "grad_norm": 0.32765469452029883, "learning_rate": 4.987000813156895e-07, "loss": 2.5909, "step": 46370 }, { "epoch": 2.878577192873549, "grad_norm": 0.336871501058692, "learning_rate": 4.981914016399636e-07, "loss": 2.6753, "step": 46371 }, { "epoch": 2.878639269973307, "grad_norm": 0.33182567515454353, "learning_rate": 4.976829802271876e-07, "loss": 2.7045, "step": 46372 }, { "epoch": 2.878701347073065, "grad_norm": 0.33617583403493456, "learning_rate": 4.971748170800206e-07, "loss": 2.5851, "step": 46373 }, { "epoch": 2.8787634241728224, "grad_norm": 0.3445404458062196, "learning_rate": 4.966669122011048e-07, "loss": 2.744, "step": 46374 }, { "epoch": 2.8788255012725807, "grad_norm": 0.34919372016592676, "learning_rate": 4.961592655930991e-07, "loss": 2.687, "step": 46375 }, { "epoch": 2.878887578372338, "grad_norm": 0.33280777726382016, "learning_rate": 4.956518772586516e-07, "loss": 2.6328, "step": 46376 }, { "epoch": 2.8789496554720966, "grad_norm": 0.33657648617118036, "learning_rate": 4.951447472003989e-07, "loss": 2.6809, "step": 46377 }, { "epoch": 2.879011732571854, "grad_norm": 0.34111061957910155, "learning_rate": 4.946378754209946e-07, "loss": 2.7587, "step": 46378 }, { "epoch": 2.8790738096716124, "grad_norm": 0.3265784693194566, "learning_rate": 4.941312619230809e-07, "loss": 2.6547, "step": 46379 }, { "epoch": 2.87913588677137, "grad_norm": 0.33133113200400177, "learning_rate": 4.936249067093058e-07, "loss": 2.7429, "step": 46380 }, { "epoch": 2.879197963871128, "grad_norm": 0.32653326013154255, "learning_rate": 4.931188097823059e-07, "loss": 2.6491, "step": 46381 }, { "epoch": 2.8792600409708857, "grad_norm": 0.3364098355482688, "learning_rate": 4.926129711447291e-07, "loss": 2.7098, "step": 46382 }, { "epoch": 2.8793221180706436, "grad_norm": 0.32666152474839466, "learning_rate": 4.921073907992014e-07, "loss": 2.659, "step": 46383 }, { "epoch": 2.8793841951704016, "grad_norm": 0.334421217923944, "learning_rate": 4.916020687483702e-07, "loss": 2.6766, "step": 46384 }, { "epoch": 2.8794462722701595, "grad_norm": 0.34308132244534156, "learning_rate": 4.910970049948727e-07, "loss": 2.7008, "step": 46385 }, { "epoch": 2.8795083493699174, "grad_norm": 0.3208626813697295, "learning_rate": 4.905921995413343e-07, "loss": 2.6227, "step": 46386 }, { "epoch": 2.8795704264696753, "grad_norm": 0.3262641113890662, "learning_rate": 4.90087652390403e-07, "loss": 2.6558, "step": 46387 }, { "epoch": 2.8796325035694332, "grad_norm": 0.3051707152899834, "learning_rate": 4.895833635446989e-07, "loss": 2.5988, "step": 46388 }, { "epoch": 2.879694580669191, "grad_norm": 0.34526306635669063, "learning_rate": 4.890793330068588e-07, "loss": 2.7142, "step": 46389 }, { "epoch": 2.879756657768949, "grad_norm": 0.35424602778757813, "learning_rate": 4.885755607795084e-07, "loss": 2.6972, "step": 46390 }, { "epoch": 2.879818734868707, "grad_norm": 0.3428692805960185, "learning_rate": 4.880720468652788e-07, "loss": 2.6269, "step": 46391 }, { "epoch": 2.879880811968465, "grad_norm": 0.3308460399548075, "learning_rate": 4.875687912668015e-07, "loss": 2.6055, "step": 46392 }, { "epoch": 2.879942889068223, "grad_norm": 0.3285488161789926, "learning_rate": 4.870657939866908e-07, "loss": 2.6071, "step": 46393 }, { "epoch": 2.8800049661679807, "grad_norm": 0.3295022240485527, "learning_rate": 4.865630550275835e-07, "loss": 2.7332, "step": 46394 }, { "epoch": 2.8800670432677387, "grad_norm": 0.3409334120780702, "learning_rate": 4.860605743920943e-07, "loss": 2.6174, "step": 46395 }, { "epoch": 2.8801291203674966, "grad_norm": 0.3421905974679045, "learning_rate": 4.855583520828488e-07, "loss": 2.7215, "step": 46396 }, { "epoch": 2.8801911974672545, "grad_norm": 0.3266343817927399, "learning_rate": 4.850563881024617e-07, "loss": 2.6334, "step": 46397 }, { "epoch": 2.8802532745670124, "grad_norm": 0.3503665077984924, "learning_rate": 4.845546824535585e-07, "loss": 2.6338, "step": 46398 }, { "epoch": 2.88031535166677, "grad_norm": 0.34599676430917436, "learning_rate": 4.840532351387539e-07, "loss": 2.7076, "step": 46399 }, { "epoch": 2.8803774287665282, "grad_norm": 0.3295217486079387, "learning_rate": 4.835520461606624e-07, "loss": 2.6981, "step": 46400 }, { "epoch": 2.8804395058662857, "grad_norm": 0.3502266086395471, "learning_rate": 4.830511155218986e-07, "loss": 2.6727, "step": 46401 }, { "epoch": 2.880501582966044, "grad_norm": 0.3522738782872276, "learning_rate": 4.825504432250772e-07, "loss": 2.6898, "step": 46402 }, { "epoch": 2.8805636600658016, "grad_norm": 0.34236832940064593, "learning_rate": 4.820500292728181e-07, "loss": 2.6309, "step": 46403 }, { "epoch": 2.8806257371655595, "grad_norm": 0.3354354811163944, "learning_rate": 4.815498736677193e-07, "loss": 2.656, "step": 46404 }, { "epoch": 2.8806878142653174, "grad_norm": 0.3284693341747653, "learning_rate": 4.81049976412401e-07, "loss": 2.6469, "step": 46405 }, { "epoch": 2.8807498913650753, "grad_norm": 0.33626044076358724, "learning_rate": 4.805503375094555e-07, "loss": 2.677, "step": 46406 }, { "epoch": 2.8808119684648332, "grad_norm": 0.33311799539426784, "learning_rate": 4.800509569615086e-07, "loss": 2.6981, "step": 46407 }, { "epoch": 2.880874045564591, "grad_norm": 0.3331493401357845, "learning_rate": 4.795518347711581e-07, "loss": 2.6293, "step": 46408 }, { "epoch": 2.880936122664349, "grad_norm": 0.3231776077897424, "learning_rate": 4.790529709410019e-07, "loss": 2.6255, "step": 46409 }, { "epoch": 2.880998199764107, "grad_norm": 0.361240066932944, "learning_rate": 4.785543654736491e-07, "loss": 2.6829, "step": 46410 }, { "epoch": 2.881060276863865, "grad_norm": 0.3213148715427558, "learning_rate": 4.780560183717031e-07, "loss": 2.5224, "step": 46411 }, { "epoch": 2.881122353963623, "grad_norm": 0.3517637206772304, "learning_rate": 4.775579296377564e-07, "loss": 2.5336, "step": 46412 }, { "epoch": 2.8811844310633807, "grad_norm": 0.3387467007986632, "learning_rate": 4.770600992744179e-07, "loss": 2.6751, "step": 46413 }, { "epoch": 2.8812465081631387, "grad_norm": 0.36404877256127044, "learning_rate": 4.765625272842744e-07, "loss": 2.6759, "step": 46414 }, { "epoch": 2.8813085852628966, "grad_norm": 0.31941277049203254, "learning_rate": 4.7606521366992397e-07, "loss": 2.628, "step": 46415 }, { "epoch": 2.8813706623626545, "grad_norm": 0.3338049691184092, "learning_rate": 4.7556815843397e-07, "loss": 2.6818, "step": 46416 }, { "epoch": 2.8814327394624124, "grad_norm": 0.3391254551185285, "learning_rate": 4.750713615789992e-07, "loss": 2.6322, "step": 46417 }, { "epoch": 2.8814948165621703, "grad_norm": 0.33951324527504606, "learning_rate": 4.745748231075986e-07, "loss": 2.6198, "step": 46418 }, { "epoch": 2.8815568936619282, "grad_norm": 0.3329690368092598, "learning_rate": 4.7407854302237155e-07, "loss": 2.6526, "step": 46419 }, { "epoch": 2.881618970761686, "grad_norm": 0.35330798969531163, "learning_rate": 4.735825213258882e-07, "loss": 2.6543, "step": 46420 }, { "epoch": 2.881681047861444, "grad_norm": 0.3449381862252707, "learning_rate": 4.7308675802075766e-07, "loss": 2.6902, "step": 46421 }, { "epoch": 2.8817431249612016, "grad_norm": 0.3211739632499877, "learning_rate": 4.7259125310955e-07, "loss": 2.6798, "step": 46422 }, { "epoch": 2.88180520206096, "grad_norm": 0.33758363393937957, "learning_rate": 4.720960065948632e-07, "loss": 2.6079, "step": 46423 }, { "epoch": 2.8818672791607174, "grad_norm": 0.33860236381836767, "learning_rate": 4.716010184792674e-07, "loss": 2.6937, "step": 46424 }, { "epoch": 2.8819293562604757, "grad_norm": 0.3516920804589307, "learning_rate": 4.7110628876536057e-07, "loss": 2.6302, "step": 46425 }, { "epoch": 2.8819914333602332, "grad_norm": 0.33199083418861486, "learning_rate": 4.706118174557128e-07, "loss": 2.6293, "step": 46426 }, { "epoch": 2.8820535104599916, "grad_norm": 0.34296649810468827, "learning_rate": 4.7011760455290544e-07, "loss": 2.7081, "step": 46427 }, { "epoch": 2.882115587559749, "grad_norm": 0.3564701019446071, "learning_rate": 4.696236500595197e-07, "loss": 2.7042, "step": 46428 }, { "epoch": 2.882177664659507, "grad_norm": 0.35596299657022457, "learning_rate": 4.691299539781313e-07, "loss": 2.6818, "step": 46429 }, { "epoch": 2.882239741759265, "grad_norm": 0.33142876659628917, "learning_rate": 4.68636516311316e-07, "loss": 2.6674, "step": 46430 }, { "epoch": 2.882301818859023, "grad_norm": 0.33128723484070866, "learning_rate": 4.681433370616439e-07, "loss": 2.6584, "step": 46431 }, { "epoch": 2.8823638959587807, "grad_norm": 0.3251065886379882, "learning_rate": 4.676504162317019e-07, "loss": 2.5375, "step": 46432 }, { "epoch": 2.8824259730585386, "grad_norm": 0.34195795895508846, "learning_rate": 4.671577538240435e-07, "loss": 2.6491, "step": 46433 }, { "epoch": 2.8824880501582966, "grad_norm": 0.3433384924777986, "learning_rate": 4.666653498412499e-07, "loss": 2.5661, "step": 46434 }, { "epoch": 2.8825501272580545, "grad_norm": 0.3241418609448469, "learning_rate": 4.661732042858857e-07, "loss": 2.6623, "step": 46435 }, { "epoch": 2.8826122043578124, "grad_norm": 0.34689245008273145, "learning_rate": 4.6568131716052676e-07, "loss": 2.7076, "step": 46436 }, { "epoch": 2.8826742814575703, "grad_norm": 0.32983952989864757, "learning_rate": 4.651896884677265e-07, "loss": 2.7377, "step": 46437 }, { "epoch": 2.8827363585573282, "grad_norm": 0.3308245216036099, "learning_rate": 4.646983182100606e-07, "loss": 2.7193, "step": 46438 }, { "epoch": 2.882798435657086, "grad_norm": 0.3301726550112472, "learning_rate": 4.642072063900882e-07, "loss": 2.6152, "step": 46439 }, { "epoch": 2.882860512756844, "grad_norm": 0.3354131021172014, "learning_rate": 4.6371635301037386e-07, "loss": 2.58, "step": 46440 }, { "epoch": 2.882922589856602, "grad_norm": 0.3339350256795172, "learning_rate": 4.632257580734711e-07, "loss": 2.6568, "step": 46441 }, { "epoch": 2.88298466695636, "grad_norm": 0.33024078514532523, "learning_rate": 4.6273542158195016e-07, "loss": 2.6864, "step": 46442 }, { "epoch": 2.883046744056118, "grad_norm": 0.34985870123225865, "learning_rate": 4.622453435383589e-07, "loss": 2.5791, "step": 46443 }, { "epoch": 2.8831088211558757, "grad_norm": 0.31195467814061967, "learning_rate": 4.61755523945262e-07, "loss": 2.6066, "step": 46444 }, { "epoch": 2.8831708982556337, "grad_norm": 0.3453005585828075, "learning_rate": 4.6126596280521294e-07, "loss": 2.5856, "step": 46445 }, { "epoch": 2.8832329753553916, "grad_norm": 0.3369604990926399, "learning_rate": 4.607766601207653e-07, "loss": 2.6607, "step": 46446 }, { "epoch": 2.883295052455149, "grad_norm": 0.3288311059245233, "learning_rate": 4.6028761589447244e-07, "loss": 2.6185, "step": 46447 }, { "epoch": 2.8833571295549074, "grad_norm": 0.3192800887147309, "learning_rate": 4.597988301288825e-07, "loss": 2.5924, "step": 46448 }, { "epoch": 2.883419206654665, "grad_norm": 0.3267141619688094, "learning_rate": 4.5931030282655443e-07, "loss": 2.6486, "step": 46449 }, { "epoch": 2.8834812837544233, "grad_norm": 0.3226574913906204, "learning_rate": 4.588220339900251e-07, "loss": 2.6697, "step": 46450 }, { "epoch": 2.8835433608541807, "grad_norm": 0.33990359601528447, "learning_rate": 4.5833402362184806e-07, "loss": 2.657, "step": 46451 }, { "epoch": 2.8836054379539386, "grad_norm": 0.34977263362867383, "learning_rate": 4.5784627172457127e-07, "loss": 2.7094, "step": 46452 }, { "epoch": 2.8836675150536966, "grad_norm": 0.3599711739072435, "learning_rate": 4.573587783007427e-07, "loss": 2.7009, "step": 46453 }, { "epoch": 2.8837295921534545, "grad_norm": 0.3406437085639709, "learning_rate": 4.568715433528936e-07, "loss": 2.7015, "step": 46454 }, { "epoch": 2.8837916692532124, "grad_norm": 0.332339166467048, "learning_rate": 4.56384566883572e-07, "loss": 2.6555, "step": 46455 }, { "epoch": 2.8838537463529703, "grad_norm": 0.32453149625633865, "learning_rate": 4.5589784889532026e-07, "loss": 2.6259, "step": 46456 }, { "epoch": 2.8839158234527282, "grad_norm": 0.33438149205260365, "learning_rate": 4.554113893906753e-07, "loss": 2.6654, "step": 46457 }, { "epoch": 2.883977900552486, "grad_norm": 0.34218087560085586, "learning_rate": 4.549251883721795e-07, "loss": 2.6251, "step": 46458 }, { "epoch": 2.884039977652244, "grad_norm": 0.3324792084073099, "learning_rate": 4.544392458423696e-07, "loss": 2.6066, "step": 46459 }, { "epoch": 2.884102054752002, "grad_norm": 0.34563649068773805, "learning_rate": 4.539535618037716e-07, "loss": 2.6642, "step": 46460 }, { "epoch": 2.88416413185176, "grad_norm": 0.33704036611891725, "learning_rate": 4.534681362589277e-07, "loss": 2.7123, "step": 46461 }, { "epoch": 2.884226208951518, "grad_norm": 0.3542734106804404, "learning_rate": 4.529829692103693e-07, "loss": 2.679, "step": 46462 }, { "epoch": 2.8842882860512757, "grad_norm": 0.34589681210343315, "learning_rate": 4.5249806066062774e-07, "loss": 2.6743, "step": 46463 }, { "epoch": 2.8843503631510337, "grad_norm": 0.3306744397557866, "learning_rate": 4.5201341061222315e-07, "loss": 2.6785, "step": 46464 }, { "epoch": 2.8844124402507916, "grad_norm": 0.33107495316223745, "learning_rate": 4.515290190677035e-07, "loss": 2.6688, "step": 46465 }, { "epoch": 2.8844745173505495, "grad_norm": 0.33506121677493794, "learning_rate": 4.5104488602957795e-07, "loss": 2.6332, "step": 46466 }, { "epoch": 2.8845365944503074, "grad_norm": 0.34515014430629803, "learning_rate": 4.5056101150038333e-07, "loss": 2.7424, "step": 46467 }, { "epoch": 2.8845986715500653, "grad_norm": 0.35712808510339356, "learning_rate": 4.500773954826454e-07, "loss": 2.6413, "step": 46468 }, { "epoch": 2.8846607486498232, "grad_norm": 0.34785714934808687, "learning_rate": 4.495940379788732e-07, "loss": 2.6366, "step": 46469 }, { "epoch": 2.8847228257495807, "grad_norm": 0.3390880491589412, "learning_rate": 4.4911093899160373e-07, "loss": 2.6409, "step": 46470 }, { "epoch": 2.884784902849339, "grad_norm": 0.33978534065000815, "learning_rate": 4.48628098523346e-07, "loss": 2.6697, "step": 46471 }, { "epoch": 2.8848469799490966, "grad_norm": 0.32230225557161757, "learning_rate": 4.4814551657663127e-07, "loss": 2.6237, "step": 46472 }, { "epoch": 2.884909057048855, "grad_norm": 0.3225746407185558, "learning_rate": 4.4766319315396324e-07, "loss": 2.6792, "step": 46473 }, { "epoch": 2.8849711341486124, "grad_norm": 0.32944166455271795, "learning_rate": 4.471811282578675e-07, "loss": 2.6901, "step": 46474 }, { "epoch": 2.8850332112483703, "grad_norm": 0.3262417655577973, "learning_rate": 4.466993218908644e-07, "loss": 2.6779, "step": 46475 }, { "epoch": 2.8850952883481282, "grad_norm": 0.33931448709591505, "learning_rate": 4.4621777405545187e-07, "loss": 2.6648, "step": 46476 }, { "epoch": 2.885157365447886, "grad_norm": 0.3371083459584254, "learning_rate": 4.4573648475415565e-07, "loss": 2.7677, "step": 46477 }, { "epoch": 2.885219442547644, "grad_norm": 0.31395283582633526, "learning_rate": 4.452554539894793e-07, "loss": 2.5525, "step": 46478 }, { "epoch": 2.885281519647402, "grad_norm": 0.33225047784000244, "learning_rate": 4.44774681763932e-07, "loss": 2.5662, "step": 46479 }, { "epoch": 2.88534359674716, "grad_norm": 0.34873851808494877, "learning_rate": 4.442941680800283e-07, "loss": 2.7458, "step": 46480 }, { "epoch": 2.885405673846918, "grad_norm": 0.3421121832293939, "learning_rate": 4.438139129402774e-07, "loss": 2.7027, "step": 46481 }, { "epoch": 2.8854677509466757, "grad_norm": 0.3344806753714786, "learning_rate": 4.433339163471717e-07, "loss": 2.6952, "step": 46482 }, { "epoch": 2.8855298280464337, "grad_norm": 0.33679512246458376, "learning_rate": 4.428541783032258e-07, "loss": 2.6803, "step": 46483 }, { "epoch": 2.8855919051461916, "grad_norm": 0.31720518570250283, "learning_rate": 4.423746988109434e-07, "loss": 2.7424, "step": 46484 }, { "epoch": 2.8856539822459495, "grad_norm": 0.3274784416463154, "learning_rate": 4.418954778728168e-07, "loss": 2.5633, "step": 46485 }, { "epoch": 2.8857160593457074, "grad_norm": 0.3480032678380814, "learning_rate": 4.4141651549136074e-07, "loss": 2.6717, "step": 46486 }, { "epoch": 2.8857781364454653, "grad_norm": 0.33326181250439374, "learning_rate": 4.409378116690566e-07, "loss": 2.6511, "step": 46487 }, { "epoch": 2.8858402135452232, "grad_norm": 0.3368245102550882, "learning_rate": 4.4045936640841334e-07, "loss": 2.6262, "step": 46488 }, { "epoch": 2.885902290644981, "grad_norm": 0.3473838531948687, "learning_rate": 4.399811797119291e-07, "loss": 2.5969, "step": 46489 }, { "epoch": 2.885964367744739, "grad_norm": 0.33282038092796645, "learning_rate": 4.395032515820907e-07, "loss": 2.5816, "step": 46490 }, { "epoch": 2.886026444844497, "grad_norm": 0.3495890222517728, "learning_rate": 4.390255820213962e-07, "loss": 2.6791, "step": 46491 }, { "epoch": 2.886088521944255, "grad_norm": 0.3344159377519725, "learning_rate": 4.3854817103233247e-07, "loss": 2.683, "step": 46492 }, { "epoch": 2.8861505990440124, "grad_norm": 0.34538739544373087, "learning_rate": 4.3807101861739754e-07, "loss": 2.6567, "step": 46493 }, { "epoch": 2.8862126761437708, "grad_norm": 0.357047463347331, "learning_rate": 4.375941247790782e-07, "loss": 2.6211, "step": 46494 }, { "epoch": 2.8862747532435282, "grad_norm": 0.34712743723751716, "learning_rate": 4.3711748951986153e-07, "loss": 2.7014, "step": 46495 }, { "epoch": 2.8863368303432866, "grad_norm": 0.34690776360919273, "learning_rate": 4.3664111284222877e-07, "loss": 2.6242, "step": 46496 }, { "epoch": 2.886398907443044, "grad_norm": 0.3300561615447801, "learning_rate": 4.361649947486779e-07, "loss": 2.6409, "step": 46497 }, { "epoch": 2.8864609845428024, "grad_norm": 0.3399590263852115, "learning_rate": 4.3568913524168475e-07, "loss": 2.6891, "step": 46498 }, { "epoch": 2.88652306164256, "grad_norm": 0.3392863580600826, "learning_rate": 4.3521353432373626e-07, "loss": 2.6405, "step": 46499 }, { "epoch": 2.886585138742318, "grad_norm": 0.3182701149529718, "learning_rate": 4.347381919973081e-07, "loss": 2.6711, "step": 46500 }, { "epoch": 2.8866472158420757, "grad_norm": 0.34186726526510425, "learning_rate": 4.3426310826488184e-07, "loss": 2.6333, "step": 46501 }, { "epoch": 2.8867092929418336, "grad_norm": 0.34448623925465616, "learning_rate": 4.3378828312894413e-07, "loss": 2.647, "step": 46502 }, { "epoch": 2.8867713700415916, "grad_norm": 0.3515180983584613, "learning_rate": 4.3331371659195987e-07, "loss": 2.6336, "step": 46503 }, { "epoch": 2.8868334471413495, "grad_norm": 0.32987919292191986, "learning_rate": 4.3283940865641027e-07, "loss": 2.6434, "step": 46504 }, { "epoch": 2.8868955242411074, "grad_norm": 0.3615577586661847, "learning_rate": 4.3236535932477117e-07, "loss": 2.7201, "step": 46505 }, { "epoch": 2.8869576013408653, "grad_norm": 0.32981774256972496, "learning_rate": 4.318915685995184e-07, "loss": 2.649, "step": 46506 }, { "epoch": 2.8870196784406232, "grad_norm": 0.3374310714482387, "learning_rate": 4.3141803648311663e-07, "loss": 2.6858, "step": 46507 }, { "epoch": 2.887081755540381, "grad_norm": 0.3469005794372679, "learning_rate": 4.3094476297804166e-07, "loss": 2.6275, "step": 46508 }, { "epoch": 2.887143832640139, "grad_norm": 0.34714660364179356, "learning_rate": 4.3047174808675815e-07, "loss": 2.6534, "step": 46509 }, { "epoch": 2.887205909739897, "grad_norm": 0.33608782532035236, "learning_rate": 4.299989918117364e-07, "loss": 2.658, "step": 46510 }, { "epoch": 2.887267986839655, "grad_norm": 0.34115693265995645, "learning_rate": 4.2952649415544664e-07, "loss": 2.8064, "step": 46511 }, { "epoch": 2.887330063939413, "grad_norm": 0.3314627661349661, "learning_rate": 4.2905425512035356e-07, "loss": 2.6328, "step": 46512 }, { "epoch": 2.8873921410391707, "grad_norm": 0.3253460211252745, "learning_rate": 4.285822747089163e-07, "loss": 2.6891, "step": 46513 }, { "epoch": 2.8874542181389287, "grad_norm": 0.3488132960637326, "learning_rate": 4.28110552923594e-07, "loss": 2.7499, "step": 46514 }, { "epoch": 2.8875162952386866, "grad_norm": 0.34637150565704106, "learning_rate": 4.2763908976685695e-07, "loss": 2.6597, "step": 46515 }, { "epoch": 2.8875783723384445, "grad_norm": 0.3233178848188364, "learning_rate": 4.2716788524116424e-07, "loss": 2.6488, "step": 46516 }, { "epoch": 2.8876404494382024, "grad_norm": 0.3433122882519438, "learning_rate": 4.2669693934896946e-07, "loss": 2.7036, "step": 46517 }, { "epoch": 2.88770252653796, "grad_norm": 0.34108122973457833, "learning_rate": 4.2622625209272627e-07, "loss": 2.652, "step": 46518 }, { "epoch": 2.8877646036377183, "grad_norm": 0.3482171580786538, "learning_rate": 4.257558234748993e-07, "loss": 2.7338, "step": 46519 }, { "epoch": 2.8878266807374757, "grad_norm": 0.3407624094397382, "learning_rate": 4.252856534979366e-07, "loss": 2.6903, "step": 46520 }, { "epoch": 2.887888757837234, "grad_norm": 0.3347364401809483, "learning_rate": 4.248157421642973e-07, "loss": 2.6942, "step": 46521 }, { "epoch": 2.8879508349369916, "grad_norm": 0.33835315468742133, "learning_rate": 4.2434608947642396e-07, "loss": 2.5362, "step": 46522 }, { "epoch": 2.8880129120367495, "grad_norm": 0.3269790666468041, "learning_rate": 4.238766954367701e-07, "loss": 2.7172, "step": 46523 }, { "epoch": 2.8880749891365074, "grad_norm": 0.33603033734707355, "learning_rate": 4.2340756004779494e-07, "loss": 2.6532, "step": 46524 }, { "epoch": 2.8881370662362653, "grad_norm": 0.34132945115697716, "learning_rate": 4.2293868331193533e-07, "loss": 2.6995, "step": 46525 }, { "epoch": 2.8881991433360232, "grad_norm": 0.31793215819937415, "learning_rate": 4.224700652316338e-07, "loss": 2.6662, "step": 46526 }, { "epoch": 2.888261220435781, "grad_norm": 0.3344641357227618, "learning_rate": 4.220017058093495e-07, "loss": 2.6455, "step": 46527 }, { "epoch": 2.888323297535539, "grad_norm": 0.3523441843338888, "learning_rate": 4.2153360504750827e-07, "loss": 2.6604, "step": 46528 }, { "epoch": 2.888385374635297, "grad_norm": 0.3427934341901326, "learning_rate": 4.2106576294856925e-07, "loss": 2.6545, "step": 46529 }, { "epoch": 2.888447451735055, "grad_norm": 0.33836382327241854, "learning_rate": 4.2059817951496385e-07, "loss": 2.6268, "step": 46530 }, { "epoch": 2.888509528834813, "grad_norm": 0.3399343808222676, "learning_rate": 4.201308547491345e-07, "loss": 2.6651, "step": 46531 }, { "epoch": 2.8885716059345707, "grad_norm": 0.34004860765446665, "learning_rate": 4.196637886535182e-07, "loss": 2.6207, "step": 46532 }, { "epoch": 2.8886336830343287, "grad_norm": 0.324606123998638, "learning_rate": 4.1919698123055184e-07, "loss": 2.6823, "step": 46533 }, { "epoch": 2.8886957601340866, "grad_norm": 0.3462710053651743, "learning_rate": 4.1873043248266687e-07, "loss": 2.6973, "step": 46534 }, { "epoch": 2.8887578372338445, "grad_norm": 0.34917203091451354, "learning_rate": 4.182641424123057e-07, "loss": 2.7004, "step": 46535 }, { "epoch": 2.8888199143336024, "grad_norm": 0.36935479142729705, "learning_rate": 4.1779811102189424e-07, "loss": 2.7317, "step": 46536 }, { "epoch": 2.8888819914333603, "grad_norm": 0.34569164838976063, "learning_rate": 4.1733233831386944e-07, "loss": 2.6909, "step": 46537 }, { "epoch": 2.8889440685331182, "grad_norm": 0.34241569086992435, "learning_rate": 4.168668242906626e-07, "loss": 2.6759, "step": 46538 }, { "epoch": 2.889006145632876, "grad_norm": 0.32872191131461337, "learning_rate": 4.164015689546885e-07, "loss": 2.7979, "step": 46539 }, { "epoch": 2.889068222732634, "grad_norm": 0.34338665312472044, "learning_rate": 4.1593657230838967e-07, "loss": 2.6794, "step": 46540 }, { "epoch": 2.8891302998323916, "grad_norm": 0.33173161618710423, "learning_rate": 4.1547183435418633e-07, "loss": 2.6542, "step": 46541 }, { "epoch": 2.88919237693215, "grad_norm": 0.33871712783789654, "learning_rate": 4.150073550945044e-07, "loss": 2.6571, "step": 46542 }, { "epoch": 2.8892544540319074, "grad_norm": 0.34300929157931714, "learning_rate": 4.145431345317641e-07, "loss": 2.6482, "step": 46543 }, { "epoch": 2.8893165311316658, "grad_norm": 0.3392500569583167, "learning_rate": 4.1407917266839124e-07, "loss": 2.5998, "step": 46544 }, { "epoch": 2.8893786082314232, "grad_norm": 0.32410654171524234, "learning_rate": 4.136154695068006e-07, "loss": 2.6001, "step": 46545 }, { "epoch": 2.8894406853311816, "grad_norm": 0.3125634486636552, "learning_rate": 4.131520250494181e-07, "loss": 2.7003, "step": 46546 }, { "epoch": 2.889502762430939, "grad_norm": 0.3314434907915971, "learning_rate": 4.126888392986583e-07, "loss": 2.6363, "step": 46547 }, { "epoch": 2.889564839530697, "grad_norm": 0.3594982309112208, "learning_rate": 4.1222591225694165e-07, "loss": 2.6372, "step": 46548 }, { "epoch": 2.889626916630455, "grad_norm": 0.3246577978454346, "learning_rate": 4.117632439266827e-07, "loss": 2.6355, "step": 46549 }, { "epoch": 2.889688993730213, "grad_norm": 0.3320090454755129, "learning_rate": 4.113008343102853e-07, "loss": 2.706, "step": 46550 }, { "epoch": 2.8897510708299707, "grad_norm": 0.3491546508027261, "learning_rate": 4.1083868341017515e-07, "loss": 2.7314, "step": 46551 }, { "epoch": 2.8898131479297287, "grad_norm": 0.36295948340237727, "learning_rate": 4.103767912287559e-07, "loss": 2.6624, "step": 46552 }, { "epoch": 2.8898752250294866, "grad_norm": 0.32622677232554814, "learning_rate": 4.0991515776843683e-07, "loss": 2.6231, "step": 46553 }, { "epoch": 2.8899373021292445, "grad_norm": 0.32643747080876373, "learning_rate": 4.094537830316325e-07, "loss": 2.6655, "step": 46554 }, { "epoch": 2.8899993792290024, "grad_norm": 0.3457863055659866, "learning_rate": 4.089926670207467e-07, "loss": 2.7316, "step": 46555 }, { "epoch": 2.8900614563287603, "grad_norm": 0.32769322369990095, "learning_rate": 4.0853180973818293e-07, "loss": 2.6776, "step": 46556 }, { "epoch": 2.8901235334285182, "grad_norm": 0.3427595478080276, "learning_rate": 4.0807121118634494e-07, "loss": 2.6824, "step": 46557 }, { "epoch": 2.890185610528276, "grad_norm": 0.319427908622604, "learning_rate": 4.0761087136764186e-07, "loss": 2.6882, "step": 46558 }, { "epoch": 2.890247687628034, "grad_norm": 0.3554062631201702, "learning_rate": 4.071507902844718e-07, "loss": 2.6906, "step": 46559 }, { "epoch": 2.890309764727792, "grad_norm": 0.3344492660474487, "learning_rate": 4.0669096793923276e-07, "loss": 2.7158, "step": 46560 }, { "epoch": 2.89037184182755, "grad_norm": 0.3173870145079408, "learning_rate": 4.0623140433432847e-07, "loss": 2.6398, "step": 46561 }, { "epoch": 2.890433918927308, "grad_norm": 0.33606749614288633, "learning_rate": 4.0577209947215147e-07, "loss": 2.5929, "step": 46562 }, { "epoch": 2.8904959960270658, "grad_norm": 0.3255419444550034, "learning_rate": 4.0531305335510525e-07, "loss": 2.6412, "step": 46563 }, { "epoch": 2.8905580731268237, "grad_norm": 0.3276044713166988, "learning_rate": 4.04854265985577e-07, "loss": 2.6122, "step": 46564 }, { "epoch": 2.8906201502265816, "grad_norm": 0.3358757827310985, "learning_rate": 4.043957373659646e-07, "loss": 2.6402, "step": 46565 }, { "epoch": 2.890682227326339, "grad_norm": 0.321155671138006, "learning_rate": 4.0393746749866066e-07, "loss": 2.7046, "step": 46566 }, { "epoch": 2.8907443044260974, "grad_norm": 0.3318717461227129, "learning_rate": 4.034794563860522e-07, "loss": 2.7232, "step": 46567 }, { "epoch": 2.890806381525855, "grad_norm": 0.33777874544167846, "learning_rate": 4.0302170403053173e-07, "loss": 2.6275, "step": 46568 }, { "epoch": 2.8908684586256133, "grad_norm": 0.3626812933706776, "learning_rate": 4.0256421043448624e-07, "loss": 2.7091, "step": 46569 }, { "epoch": 2.8909305357253707, "grad_norm": 0.3310226154827211, "learning_rate": 4.021069756003082e-07, "loss": 2.7289, "step": 46570 }, { "epoch": 2.8909926128251286, "grad_norm": 0.33632460181399065, "learning_rate": 4.0164999953037905e-07, "loss": 2.6683, "step": 46571 }, { "epoch": 2.8910546899248866, "grad_norm": 0.3386967145001056, "learning_rate": 4.0119328222707476e-07, "loss": 2.6666, "step": 46572 }, { "epoch": 2.8911167670246445, "grad_norm": 0.31918054702189275, "learning_rate": 4.0073682369278776e-07, "loss": 2.735, "step": 46573 }, { "epoch": 2.8911788441244024, "grad_norm": 0.32118168807760816, "learning_rate": 4.002806239298995e-07, "loss": 2.7152, "step": 46574 }, { "epoch": 2.8912409212241603, "grad_norm": 0.33308350494273137, "learning_rate": 3.998246829407859e-07, "loss": 2.6631, "step": 46575 }, { "epoch": 2.8913029983239182, "grad_norm": 0.3541708296175866, "learning_rate": 3.993690007278339e-07, "loss": 2.6614, "step": 46576 }, { "epoch": 2.891365075423676, "grad_norm": 0.34104466090735636, "learning_rate": 3.9891357729340826e-07, "loss": 2.6433, "step": 46577 }, { "epoch": 2.891427152523434, "grad_norm": 0.35835087016167533, "learning_rate": 3.984584126398905e-07, "loss": 2.6365, "step": 46578 }, { "epoch": 2.891489229623192, "grad_norm": 0.314836824781109, "learning_rate": 3.980035067696619e-07, "loss": 2.6709, "step": 46579 }, { "epoch": 2.89155130672295, "grad_norm": 0.3231315884573306, "learning_rate": 3.975488596850874e-07, "loss": 2.6233, "step": 46580 }, { "epoch": 2.891613383822708, "grad_norm": 0.3415032678848509, "learning_rate": 3.9709447138854273e-07, "loss": 2.6503, "step": 46581 }, { "epoch": 2.8916754609224657, "grad_norm": 0.33998217997063057, "learning_rate": 3.9664034188239827e-07, "loss": 2.7233, "step": 46582 }, { "epoch": 2.8917375380222237, "grad_norm": 0.3321670927524147, "learning_rate": 3.9618647116902443e-07, "loss": 2.6871, "step": 46583 }, { "epoch": 2.8917996151219816, "grad_norm": 0.3356349491974745, "learning_rate": 3.9573285925078587e-07, "loss": 2.7368, "step": 46584 }, { "epoch": 2.8918616922217395, "grad_norm": 0.3574397181658714, "learning_rate": 3.952795061300474e-07, "loss": 2.5956, "step": 46585 }, { "epoch": 2.8919237693214974, "grad_norm": 0.35524216441399603, "learning_rate": 3.9482641180917936e-07, "loss": 2.5879, "step": 46586 }, { "epoch": 2.8919858464212553, "grad_norm": 0.344338476871641, "learning_rate": 3.943735762905465e-07, "loss": 2.6652, "step": 46587 }, { "epoch": 2.8920479235210133, "grad_norm": 0.3285731366195324, "learning_rate": 3.939209995765081e-07, "loss": 2.6791, "step": 46588 }, { "epoch": 2.8921100006207707, "grad_norm": 0.33354515061156337, "learning_rate": 3.934686816694233e-07, "loss": 2.6823, "step": 46589 }, { "epoch": 2.892172077720529, "grad_norm": 0.3317328402365272, "learning_rate": 3.930166225716625e-07, "loss": 2.6246, "step": 46590 }, { "epoch": 2.8922341548202866, "grad_norm": 0.32814115529934584, "learning_rate": 3.9256482228556825e-07, "loss": 2.7277, "step": 46591 }, { "epoch": 2.892296231920045, "grad_norm": 0.3373687550764257, "learning_rate": 3.921132808135053e-07, "loss": 2.7185, "step": 46592 }, { "epoch": 2.8923583090198024, "grad_norm": 0.34795877602529673, "learning_rate": 3.9166199815783844e-07, "loss": 2.8145, "step": 46593 }, { "epoch": 2.8924203861195608, "grad_norm": 0.34951221458568293, "learning_rate": 3.912109743209047e-07, "loss": 2.6825, "step": 46594 }, { "epoch": 2.8924824632193182, "grad_norm": 0.341107572589135, "learning_rate": 3.9076020930506887e-07, "loss": 2.6484, "step": 46595 }, { "epoch": 2.892544540319076, "grad_norm": 0.3575158751237487, "learning_rate": 3.9030970311268454e-07, "loss": 2.7008, "step": 46596 }, { "epoch": 2.892606617418834, "grad_norm": 0.3301317746956106, "learning_rate": 3.8985945574609437e-07, "loss": 2.6224, "step": 46597 }, { "epoch": 2.892668694518592, "grad_norm": 0.35813996351979915, "learning_rate": 3.8940946720765205e-07, "loss": 2.6735, "step": 46598 }, { "epoch": 2.89273077161835, "grad_norm": 0.35060280039406594, "learning_rate": 3.8895973749970006e-07, "loss": 2.6253, "step": 46599 }, { "epoch": 2.892792848718108, "grad_norm": 0.3298830663613426, "learning_rate": 3.8851026662459213e-07, "loss": 2.6449, "step": 46600 }, { "epoch": 2.8928549258178657, "grad_norm": 0.33645916511543056, "learning_rate": 3.880610545846708e-07, "loss": 2.6074, "step": 46601 }, { "epoch": 2.8929170029176237, "grad_norm": 0.32767903584795954, "learning_rate": 3.8761210138227863e-07, "loss": 2.7298, "step": 46602 }, { "epoch": 2.8929790800173816, "grad_norm": 0.3310962639641654, "learning_rate": 3.8716340701975827e-07, "loss": 2.6954, "step": 46603 }, { "epoch": 2.8930411571171395, "grad_norm": 0.3271164266383354, "learning_rate": 3.867149714994467e-07, "loss": 2.6444, "step": 46604 }, { "epoch": 2.8931032342168974, "grad_norm": 0.3386359181395365, "learning_rate": 3.8626679482368644e-07, "loss": 2.6176, "step": 46605 }, { "epoch": 2.8931653113166553, "grad_norm": 0.32802522368454357, "learning_rate": 3.8581887699482013e-07, "loss": 2.5574, "step": 46606 }, { "epoch": 2.8932273884164132, "grad_norm": 0.35102146691445457, "learning_rate": 3.853712180151792e-07, "loss": 2.661, "step": 46607 }, { "epoch": 2.893289465516171, "grad_norm": 0.3267795703710163, "learning_rate": 3.8492381788710063e-07, "loss": 2.6085, "step": 46608 }, { "epoch": 2.893351542615929, "grad_norm": 0.33094477446488424, "learning_rate": 3.84476676612916e-07, "loss": 2.7471, "step": 46609 }, { "epoch": 2.893413619715687, "grad_norm": 0.33282804811004973, "learning_rate": 3.8402979419496774e-07, "loss": 2.7235, "step": 46610 }, { "epoch": 2.893475696815445, "grad_norm": 0.3260256959377549, "learning_rate": 3.8358317063557637e-07, "loss": 2.7107, "step": 46611 }, { "epoch": 2.893537773915203, "grad_norm": 0.342688145038054, "learning_rate": 3.831368059370788e-07, "loss": 2.7411, "step": 46612 }, { "epoch": 2.8935998510149608, "grad_norm": 0.3258249756630618, "learning_rate": 3.8269070010180096e-07, "loss": 2.6966, "step": 46613 }, { "epoch": 2.8936619281147182, "grad_norm": 0.3291841737199988, "learning_rate": 3.8224485313206884e-07, "loss": 2.6391, "step": 46614 }, { "epoch": 2.8937240052144766, "grad_norm": 0.3418547227135733, "learning_rate": 3.817992650302138e-07, "loss": 2.5743, "step": 46615 }, { "epoch": 2.893786082314234, "grad_norm": 0.33096653262513864, "learning_rate": 3.813539357985563e-07, "loss": 2.6876, "step": 46616 }, { "epoch": 2.8938481594139924, "grad_norm": 0.32622605290713286, "learning_rate": 3.8090886543942216e-07, "loss": 2.6696, "step": 46617 }, { "epoch": 2.89391023651375, "grad_norm": 0.3473646033612846, "learning_rate": 3.804640539551263e-07, "loss": 2.6878, "step": 46618 }, { "epoch": 2.893972313613508, "grad_norm": 0.32530707652530005, "learning_rate": 3.800195013480001e-07, "loss": 2.6938, "step": 46619 }, { "epoch": 2.8940343907132657, "grad_norm": 0.31857848087331886, "learning_rate": 3.7957520762035845e-07, "loss": 2.6546, "step": 46620 }, { "epoch": 2.8940964678130237, "grad_norm": 0.3427967704808609, "learning_rate": 3.791311727745162e-07, "loss": 2.6267, "step": 46621 }, { "epoch": 2.8941585449127816, "grad_norm": 0.34379473308458947, "learning_rate": 3.78687396812788e-07, "loss": 2.6718, "step": 46622 }, { "epoch": 2.8942206220125395, "grad_norm": 0.335775081050648, "learning_rate": 3.7824387973749985e-07, "loss": 2.6765, "step": 46623 }, { "epoch": 2.8942826991122974, "grad_norm": 0.33935362451724593, "learning_rate": 3.778006215509611e-07, "loss": 2.5935, "step": 46624 }, { "epoch": 2.8943447762120553, "grad_norm": 0.3300505293647567, "learning_rate": 3.773576222554809e-07, "loss": 2.7094, "step": 46625 }, { "epoch": 2.8944068533118132, "grad_norm": 0.3349924990184302, "learning_rate": 3.769148818533685e-07, "loss": 2.6393, "step": 46626 }, { "epoch": 2.894468930411571, "grad_norm": 0.3268918981615309, "learning_rate": 3.764724003469389e-07, "loss": 2.5827, "step": 46627 }, { "epoch": 2.894531007511329, "grad_norm": 0.34530299678516224, "learning_rate": 3.7603017773850115e-07, "loss": 2.7356, "step": 46628 }, { "epoch": 2.894593084611087, "grad_norm": 0.31833299866920045, "learning_rate": 3.7558821403035906e-07, "loss": 2.6232, "step": 46629 }, { "epoch": 2.894655161710845, "grad_norm": 0.33601668999154377, "learning_rate": 3.751465092248219e-07, "loss": 2.5658, "step": 46630 }, { "epoch": 2.894717238810603, "grad_norm": 0.33171146925681866, "learning_rate": 3.7470506332418776e-07, "loss": 2.5954, "step": 46631 }, { "epoch": 2.8947793159103608, "grad_norm": 0.3326482761658337, "learning_rate": 3.7426387633077155e-07, "loss": 2.6782, "step": 46632 }, { "epoch": 2.8948413930101187, "grad_norm": 0.3367646037873741, "learning_rate": 3.738229482468658e-07, "loss": 2.7088, "step": 46633 }, { "epoch": 2.8949034701098766, "grad_norm": 0.3617470040213765, "learning_rate": 3.7338227907476874e-07, "loss": 2.6948, "step": 46634 }, { "epoch": 2.8949655472096345, "grad_norm": 0.3262235178010238, "learning_rate": 3.729418688167841e-07, "loss": 2.6614, "step": 46635 }, { "epoch": 2.8950276243093924, "grad_norm": 0.35017878232239413, "learning_rate": 3.7250171747521546e-07, "loss": 2.5869, "step": 46636 }, { "epoch": 2.89508970140915, "grad_norm": 0.33356780011557025, "learning_rate": 3.720618250523444e-07, "loss": 2.6635, "step": 46637 }, { "epoch": 2.8951517785089083, "grad_norm": 0.3610946932243464, "learning_rate": 3.716221915504803e-07, "loss": 2.7004, "step": 46638 }, { "epoch": 2.8952138556086657, "grad_norm": 0.3509779520914785, "learning_rate": 3.7118281697191e-07, "loss": 2.6992, "step": 46639 }, { "epoch": 2.895275932708424, "grad_norm": 0.34661131708491577, "learning_rate": 3.707437013189263e-07, "loss": 2.6512, "step": 46640 }, { "epoch": 2.8953380098081816, "grad_norm": 0.32741057867735207, "learning_rate": 3.7030484459381623e-07, "loss": 2.653, "step": 46641 }, { "epoch": 2.89540008690794, "grad_norm": 0.3419111987063424, "learning_rate": 3.698662467988778e-07, "loss": 2.5719, "step": 46642 }, { "epoch": 2.8954621640076974, "grad_norm": 0.3230581883544784, "learning_rate": 3.6942790793639823e-07, "loss": 2.6436, "step": 46643 }, { "epoch": 2.8955242411074553, "grad_norm": 0.3509020854238429, "learning_rate": 3.68989828008659e-07, "loss": 2.6393, "step": 46644 }, { "epoch": 2.8955863182072132, "grad_norm": 0.34109544988637625, "learning_rate": 3.6855200701794156e-07, "loss": 2.692, "step": 46645 }, { "epoch": 2.895648395306971, "grad_norm": 0.3427692097894941, "learning_rate": 3.681144449665441e-07, "loss": 2.6734, "step": 46646 }, { "epoch": 2.895710472406729, "grad_norm": 0.3332983522636814, "learning_rate": 3.676771418567426e-07, "loss": 2.6468, "step": 46647 }, { "epoch": 2.895772549506487, "grad_norm": 0.33450443089049464, "learning_rate": 3.6724009769081856e-07, "loss": 2.7004, "step": 46648 }, { "epoch": 2.895834626606245, "grad_norm": 0.34691050193152667, "learning_rate": 3.668033124710479e-07, "loss": 2.6968, "step": 46649 }, { "epoch": 2.895896703706003, "grad_norm": 0.3207291939811699, "learning_rate": 3.663667861997122e-07, "loss": 2.6674, "step": 46650 }, { "epoch": 2.8959587808057607, "grad_norm": 0.32811978880967546, "learning_rate": 3.659305188790929e-07, "loss": 2.6064, "step": 46651 }, { "epoch": 2.8960208579055187, "grad_norm": 0.3407650003664948, "learning_rate": 3.65494510511466e-07, "loss": 2.6722, "step": 46652 }, { "epoch": 2.8960829350052766, "grad_norm": 0.3502931567729427, "learning_rate": 3.6505876109910187e-07, "loss": 2.6498, "step": 46653 }, { "epoch": 2.8961450121050345, "grad_norm": 0.3502499450006925, "learning_rate": 3.6462327064427094e-07, "loss": 2.6225, "step": 46654 }, { "epoch": 2.8962070892047924, "grad_norm": 0.3159283690826331, "learning_rate": 3.641880391492547e-07, "loss": 2.7401, "step": 46655 }, { "epoch": 2.8962691663045503, "grad_norm": 0.35735284941329737, "learning_rate": 3.63753066616318e-07, "loss": 2.6977, "step": 46656 }, { "epoch": 2.8963312434043083, "grad_norm": 0.34444704691974487, "learning_rate": 3.633183530477313e-07, "loss": 2.6273, "step": 46657 }, { "epoch": 2.896393320504066, "grad_norm": 0.35940580650642534, "learning_rate": 3.6288389844575944e-07, "loss": 2.6018, "step": 46658 }, { "epoch": 2.896455397603824, "grad_norm": 0.3324598687376284, "learning_rate": 3.624497028126728e-07, "loss": 2.6974, "step": 46659 }, { "epoch": 2.896517474703582, "grad_norm": 0.3421093847218591, "learning_rate": 3.620157661507362e-07, "loss": 2.7231, "step": 46660 }, { "epoch": 2.89657955180334, "grad_norm": 0.33196869429169773, "learning_rate": 3.6158208846221454e-07, "loss": 2.6517, "step": 46661 }, { "epoch": 2.8966416289030974, "grad_norm": 0.33391225605027064, "learning_rate": 3.6114866974937266e-07, "loss": 2.7145, "step": 46662 }, { "epoch": 2.8967037060028558, "grad_norm": 0.3394112301029013, "learning_rate": 3.6071551001446434e-07, "loss": 2.6648, "step": 46663 }, { "epoch": 2.8967657831026132, "grad_norm": 0.3299349863860252, "learning_rate": 3.6028260925975444e-07, "loss": 2.6579, "step": 46664 }, { "epoch": 2.8968278602023716, "grad_norm": 0.3417408161512986, "learning_rate": 3.5984996748749665e-07, "loss": 2.7036, "step": 46665 }, { "epoch": 2.896889937302129, "grad_norm": 0.3356289222601758, "learning_rate": 3.5941758469996145e-07, "loss": 2.6931, "step": 46666 }, { "epoch": 2.896952014401887, "grad_norm": 0.33824485336329857, "learning_rate": 3.589854608993859e-07, "loss": 2.7351, "step": 46667 }, { "epoch": 2.897014091501645, "grad_norm": 0.32682618168672906, "learning_rate": 3.585535960880404e-07, "loss": 2.6672, "step": 46668 }, { "epoch": 2.897076168601403, "grad_norm": 0.34813913820003256, "learning_rate": 3.581219902681676e-07, "loss": 2.7352, "step": 46669 }, { "epoch": 2.8971382457011607, "grad_norm": 0.32179173392504606, "learning_rate": 3.5769064344202687e-07, "loss": 2.6338, "step": 46670 }, { "epoch": 2.8972003228009187, "grad_norm": 0.3186471492660767, "learning_rate": 3.5725955561186075e-07, "loss": 2.7561, "step": 46671 }, { "epoch": 2.8972623999006766, "grad_norm": 0.32298916597620586, "learning_rate": 3.568287267799231e-07, "loss": 2.6573, "step": 46672 }, { "epoch": 2.8973244770004345, "grad_norm": 0.34789967319347626, "learning_rate": 3.5639815694846755e-07, "loss": 2.6767, "step": 46673 }, { "epoch": 2.8973865541001924, "grad_norm": 0.33941006667219126, "learning_rate": 3.559678461197258e-07, "loss": 2.6596, "step": 46674 }, { "epoch": 2.8974486311999503, "grad_norm": 0.32870837985226325, "learning_rate": 3.5553779429595703e-07, "loss": 2.6446, "step": 46675 }, { "epoch": 2.8975107082997082, "grad_norm": 0.35971267012249336, "learning_rate": 3.5510800147939837e-07, "loss": 2.7279, "step": 46676 }, { "epoch": 2.897572785399466, "grad_norm": 0.3581087984544567, "learning_rate": 3.5467846767229253e-07, "loss": 2.6394, "step": 46677 }, { "epoch": 2.897634862499224, "grad_norm": 0.33724963297162963, "learning_rate": 3.5424919287687653e-07, "loss": 2.6865, "step": 46678 }, { "epoch": 2.897696939598982, "grad_norm": 0.3441743831985609, "learning_rate": 3.538201770953986e-07, "loss": 2.6211, "step": 46679 }, { "epoch": 2.89775901669874, "grad_norm": 0.32276466340160304, "learning_rate": 3.5339142033009587e-07, "loss": 2.6764, "step": 46680 }, { "epoch": 2.897821093798498, "grad_norm": 0.33447945024036196, "learning_rate": 3.5296292258319983e-07, "loss": 2.6854, "step": 46681 }, { "epoch": 2.8978831708982558, "grad_norm": 0.3218701797420708, "learning_rate": 3.5253468385694767e-07, "loss": 2.6248, "step": 46682 }, { "epoch": 2.8979452479980137, "grad_norm": 0.34383336053462316, "learning_rate": 3.521067041535764e-07, "loss": 2.728, "step": 46683 }, { "epoch": 2.8980073250977716, "grad_norm": 0.3240754603679212, "learning_rate": 3.51678983475312e-07, "loss": 2.7504, "step": 46684 }, { "epoch": 2.898069402197529, "grad_norm": 0.327731282307124, "learning_rate": 3.512515218243917e-07, "loss": 2.6418, "step": 46685 }, { "epoch": 2.8981314792972874, "grad_norm": 0.3413327035315268, "learning_rate": 3.5082431920304694e-07, "loss": 2.709, "step": 46686 }, { "epoch": 2.898193556397045, "grad_norm": 0.3353476224744486, "learning_rate": 3.5039737561350927e-07, "loss": 2.659, "step": 46687 }, { "epoch": 2.8982556334968033, "grad_norm": 0.3368600553265681, "learning_rate": 3.499706910579936e-07, "loss": 2.6308, "step": 46688 }, { "epoch": 2.8983177105965607, "grad_norm": 0.3406321708554343, "learning_rate": 3.495442655387371e-07, "loss": 2.6483, "step": 46689 }, { "epoch": 2.898379787696319, "grad_norm": 0.35072767957359463, "learning_rate": 3.491180990579601e-07, "loss": 2.6615, "step": 46690 }, { "epoch": 2.8984418647960766, "grad_norm": 0.31270018364699936, "learning_rate": 3.486921916178887e-07, "loss": 2.6858, "step": 46691 }, { "epoch": 2.8985039418958345, "grad_norm": 0.3106669187948087, "learning_rate": 3.4826654322074323e-07, "loss": 2.6625, "step": 46692 }, { "epoch": 2.8985660189955924, "grad_norm": 0.3146953242370053, "learning_rate": 3.478411538687443e-07, "loss": 2.6117, "step": 46693 }, { "epoch": 2.8986280960953503, "grad_norm": 0.34101536369532204, "learning_rate": 3.474160235641122e-07, "loss": 2.6924, "step": 46694 }, { "epoch": 2.8986901731951082, "grad_norm": 0.32052612159407207, "learning_rate": 3.469911523090619e-07, "loss": 2.6366, "step": 46695 }, { "epoch": 2.898752250294866, "grad_norm": 0.35479706403771194, "learning_rate": 3.4656654010581936e-07, "loss": 2.6774, "step": 46696 }, { "epoch": 2.898814327394624, "grad_norm": 0.3229678854264054, "learning_rate": 3.461421869565884e-07, "loss": 2.6229, "step": 46697 }, { "epoch": 2.898876404494382, "grad_norm": 0.34223350973298117, "learning_rate": 3.4571809286358393e-07, "loss": 2.7198, "step": 46698 }, { "epoch": 2.89893848159414, "grad_norm": 0.32327550387910753, "learning_rate": 3.452942578290264e-07, "loss": 2.6501, "step": 46699 }, { "epoch": 2.899000558693898, "grad_norm": 0.3167721904540331, "learning_rate": 3.448706818551195e-07, "loss": 2.6324, "step": 46700 }, { "epoch": 2.8990626357936558, "grad_norm": 0.35719204376614694, "learning_rate": 3.4444736494407826e-07, "loss": 2.7131, "step": 46701 }, { "epoch": 2.8991247128934137, "grad_norm": 0.32936486371565477, "learning_rate": 3.4402430709811197e-07, "loss": 2.71, "step": 46702 }, { "epoch": 2.8991867899931716, "grad_norm": 0.33914792966676116, "learning_rate": 3.436015083194188e-07, "loss": 2.6409, "step": 46703 }, { "epoch": 2.8992488670929295, "grad_norm": 0.3511087350232185, "learning_rate": 3.4317896861021937e-07, "loss": 2.7017, "step": 46704 }, { "epoch": 2.8993109441926874, "grad_norm": 0.3281004760296922, "learning_rate": 3.427566879727062e-07, "loss": 2.6082, "step": 46705 }, { "epoch": 2.8993730212924453, "grad_norm": 0.3353392797052516, "learning_rate": 3.423346664090832e-07, "loss": 2.722, "step": 46706 }, { "epoch": 2.8994350983922033, "grad_norm": 0.3316599396561765, "learning_rate": 3.419129039215541e-07, "loss": 2.6464, "step": 46707 }, { "epoch": 2.899497175491961, "grad_norm": 0.31323093523964607, "learning_rate": 3.4149140051232266e-07, "loss": 2.6964, "step": 46708 }, { "epoch": 2.899559252591719, "grad_norm": 0.3468900015740209, "learning_rate": 3.410701561835872e-07, "loss": 2.6813, "step": 46709 }, { "epoch": 2.8996213296914766, "grad_norm": 0.30172682420939456, "learning_rate": 3.406491709375459e-07, "loss": 2.6911, "step": 46710 }, { "epoch": 2.899683406791235, "grad_norm": 0.32504346813603596, "learning_rate": 3.402284447763915e-07, "loss": 2.7024, "step": 46711 }, { "epoch": 2.8997454838909924, "grad_norm": 0.33736309038478274, "learning_rate": 3.398079777023222e-07, "loss": 2.6973, "step": 46712 }, { "epoch": 2.8998075609907508, "grad_norm": 0.3424047786196754, "learning_rate": 3.393877697175252e-07, "loss": 2.631, "step": 46713 }, { "epoch": 2.8998696380905082, "grad_norm": 0.3578407459687607, "learning_rate": 3.3896782082419864e-07, "loss": 2.6939, "step": 46714 }, { "epoch": 2.899931715190266, "grad_norm": 0.33517464807337827, "learning_rate": 3.385481310245353e-07, "loss": 2.542, "step": 46715 }, { "epoch": 2.899993792290024, "grad_norm": 0.3425216945856782, "learning_rate": 3.381287003207223e-07, "loss": 2.6644, "step": 46716 }, { "epoch": 2.900055869389782, "grad_norm": 0.33755046061062144, "learning_rate": 3.3770952871494124e-07, "loss": 2.6484, "step": 46717 }, { "epoch": 2.90011794648954, "grad_norm": 0.3531477157071966, "learning_rate": 3.372906162093903e-07, "loss": 2.712, "step": 46718 }, { "epoch": 2.900180023589298, "grad_norm": 0.3346918543364064, "learning_rate": 3.368719628062511e-07, "loss": 2.6287, "step": 46719 }, { "epoch": 2.9002421006890557, "grad_norm": 0.32537627125527246, "learning_rate": 3.364535685077108e-07, "loss": 2.7063, "step": 46720 }, { "epoch": 2.9003041777888137, "grad_norm": 0.3553396556478474, "learning_rate": 3.360354333159399e-07, "loss": 2.6578, "step": 46721 }, { "epoch": 2.9003662548885716, "grad_norm": 0.30142855659076756, "learning_rate": 3.3561755723313657e-07, "loss": 2.6696, "step": 46722 }, { "epoch": 2.9004283319883295, "grad_norm": 0.32971500987040453, "learning_rate": 3.351999402614658e-07, "loss": 2.6613, "step": 46723 }, { "epoch": 2.9004904090880874, "grad_norm": 0.32346969849649476, "learning_rate": 3.347825824031203e-07, "loss": 2.5806, "step": 46724 }, { "epoch": 2.9005524861878453, "grad_norm": 0.358132245321697, "learning_rate": 3.34365483660265e-07, "loss": 2.758, "step": 46725 }, { "epoch": 2.9006145632876033, "grad_norm": 0.32864731134887376, "learning_rate": 3.339486440350814e-07, "loss": 2.684, "step": 46726 }, { "epoch": 2.900676640387361, "grad_norm": 0.32055542038440515, "learning_rate": 3.335320635297512e-07, "loss": 2.762, "step": 46727 }, { "epoch": 2.900738717487119, "grad_norm": 0.34416863568868805, "learning_rate": 3.331157421464337e-07, "loss": 2.6083, "step": 46728 }, { "epoch": 2.900800794586877, "grad_norm": 0.3690591752550984, "learning_rate": 3.326996798873161e-07, "loss": 2.7725, "step": 46729 }, { "epoch": 2.900862871686635, "grad_norm": 0.3521374134235994, "learning_rate": 3.322838767545522e-07, "loss": 2.6367, "step": 46730 }, { "epoch": 2.900924948786393, "grad_norm": 0.3321325484190803, "learning_rate": 3.3186833275032916e-07, "loss": 2.7315, "step": 46731 }, { "epoch": 2.9009870258861508, "grad_norm": 0.3316180743850581, "learning_rate": 3.3145304787680076e-07, "loss": 2.6342, "step": 46732 }, { "epoch": 2.9010491029859082, "grad_norm": 0.32040002836517206, "learning_rate": 3.310380221361431e-07, "loss": 2.6239, "step": 46733 }, { "epoch": 2.9011111800856666, "grad_norm": 0.33070966582990874, "learning_rate": 3.3062325553051556e-07, "loss": 2.7053, "step": 46734 }, { "epoch": 2.901173257185424, "grad_norm": 0.3444181364704788, "learning_rate": 3.30208748062083e-07, "loss": 2.6495, "step": 46735 }, { "epoch": 2.9012353342851824, "grad_norm": 0.3444576889881257, "learning_rate": 3.2979449973301046e-07, "loss": 2.6366, "step": 46736 }, { "epoch": 2.90129741138494, "grad_norm": 0.3602047959913897, "learning_rate": 3.2938051054545724e-07, "loss": 2.7606, "step": 46737 }, { "epoch": 2.9013594884846983, "grad_norm": 0.322133482225922, "learning_rate": 3.289667805015884e-07, "loss": 2.6117, "step": 46738 }, { "epoch": 2.9014215655844557, "grad_norm": 0.3480577186753369, "learning_rate": 3.28553309603552e-07, "loss": 2.7707, "step": 46739 }, { "epoch": 2.9014836426842137, "grad_norm": 0.34192186522539375, "learning_rate": 3.281400978535076e-07, "loss": 2.6603, "step": 46740 }, { "epoch": 2.9015457197839716, "grad_norm": 0.3465460828774397, "learning_rate": 3.2772714525362014e-07, "loss": 2.7516, "step": 46741 }, { "epoch": 2.9016077968837295, "grad_norm": 0.3176660887455929, "learning_rate": 3.2731445180604337e-07, "loss": 2.6814, "step": 46742 }, { "epoch": 2.9016698739834874, "grad_norm": 0.3349173714913587, "learning_rate": 3.2690201751292007e-07, "loss": 2.6045, "step": 46743 }, { "epoch": 2.9017319510832453, "grad_norm": 0.33534410098466927, "learning_rate": 3.26489842376404e-07, "loss": 2.6348, "step": 46744 }, { "epoch": 2.9017940281830032, "grad_norm": 0.32753300624245896, "learning_rate": 3.2607792639865464e-07, "loss": 2.706, "step": 46745 }, { "epoch": 2.901856105282761, "grad_norm": 0.3372505056117278, "learning_rate": 3.256662695818147e-07, "loss": 2.6744, "step": 46746 }, { "epoch": 2.901918182382519, "grad_norm": 0.3394211849554562, "learning_rate": 3.252548719280324e-07, "loss": 2.6347, "step": 46747 }, { "epoch": 2.901980259482277, "grad_norm": 0.32569411688279576, "learning_rate": 3.248437334394561e-07, "loss": 2.6619, "step": 46748 }, { "epoch": 2.902042336582035, "grad_norm": 0.3330316842659704, "learning_rate": 3.244328541182229e-07, "loss": 2.6609, "step": 46749 }, { "epoch": 2.902104413681793, "grad_norm": 0.3296536758210969, "learning_rate": 3.240222339664867e-07, "loss": 2.6529, "step": 46750 }, { "epoch": 2.9021664907815508, "grad_norm": 0.33305344031294737, "learning_rate": 3.236118729863902e-07, "loss": 2.6805, "step": 46751 }, { "epoch": 2.9022285678813087, "grad_norm": 0.33629221264797965, "learning_rate": 3.23201771180065e-07, "loss": 2.6729, "step": 46752 }, { "epoch": 2.9022906449810666, "grad_norm": 0.33377033759240776, "learning_rate": 3.227919285496539e-07, "loss": 2.6714, "step": 46753 }, { "epoch": 2.9023527220808245, "grad_norm": 0.31894535127769186, "learning_rate": 3.2238234509729956e-07, "loss": 2.7214, "step": 46754 }, { "epoch": 2.9024147991805824, "grad_norm": 0.34396261445717413, "learning_rate": 3.219730208251392e-07, "loss": 2.7222, "step": 46755 }, { "epoch": 2.9024768762803403, "grad_norm": 0.33001259147579487, "learning_rate": 3.2156395573529895e-07, "loss": 2.6866, "step": 46756 }, { "epoch": 2.9025389533800983, "grad_norm": 0.33125308233302525, "learning_rate": 3.2115514982992143e-07, "loss": 2.6787, "step": 46757 }, { "epoch": 2.9026010304798557, "grad_norm": 0.34295058495593594, "learning_rate": 3.2074660311113837e-07, "loss": 2.6603, "step": 46758 }, { "epoch": 2.902663107579614, "grad_norm": 0.3587015933099523, "learning_rate": 3.203383155810813e-07, "loss": 2.7502, "step": 46759 }, { "epoch": 2.9027251846793716, "grad_norm": 0.3235645242863143, "learning_rate": 3.199302872418819e-07, "loss": 2.6993, "step": 46760 }, { "epoch": 2.90278726177913, "grad_norm": 0.3374561095838986, "learning_rate": 3.195225180956607e-07, "loss": 2.6671, "step": 46761 }, { "epoch": 2.9028493388788874, "grad_norm": 0.34700539560891464, "learning_rate": 3.1911500814455487e-07, "loss": 2.6065, "step": 46762 }, { "epoch": 2.9029114159786453, "grad_norm": 0.319229808285339, "learning_rate": 3.1870775739068494e-07, "loss": 2.6809, "step": 46763 }, { "epoch": 2.9029734930784032, "grad_norm": 0.3196433800935083, "learning_rate": 3.1830076583617696e-07, "loss": 2.6408, "step": 46764 }, { "epoch": 2.903035570178161, "grad_norm": 0.32489806557382095, "learning_rate": 3.1789403348315703e-07, "loss": 2.6345, "step": 46765 }, { "epoch": 2.903097647277919, "grad_norm": 0.32880047118824357, "learning_rate": 3.1748756033374015e-07, "loss": 2.6, "step": 46766 }, { "epoch": 2.903159724377677, "grad_norm": 0.3425892935729824, "learning_rate": 3.1708134639005236e-07, "loss": 2.6318, "step": 46767 }, { "epoch": 2.903221801477435, "grad_norm": 0.34741613046620445, "learning_rate": 3.166753916542142e-07, "loss": 2.6919, "step": 46768 }, { "epoch": 2.903283878577193, "grad_norm": 0.349233841196618, "learning_rate": 3.1626969612834065e-07, "loss": 2.6557, "step": 46769 }, { "epoch": 2.9033459556769508, "grad_norm": 0.33389818741414096, "learning_rate": 3.158642598145467e-07, "loss": 2.6663, "step": 46770 }, { "epoch": 2.9034080327767087, "grad_norm": 0.32674613243154876, "learning_rate": 3.154590827149584e-07, "loss": 2.7222, "step": 46771 }, { "epoch": 2.9034701098764666, "grad_norm": 0.34698752047444803, "learning_rate": 3.1505416483167404e-07, "loss": 2.6619, "step": 46772 }, { "epoch": 2.9035321869762245, "grad_norm": 0.35003700954744954, "learning_rate": 3.1464950616681423e-07, "loss": 2.7792, "step": 46773 }, { "epoch": 2.9035942640759824, "grad_norm": 0.34036284297662595, "learning_rate": 3.142451067224883e-07, "loss": 2.6854, "step": 46774 }, { "epoch": 2.9036563411757403, "grad_norm": 0.3328246454700194, "learning_rate": 3.138409665008113e-07, "loss": 2.5912, "step": 46775 }, { "epoch": 2.9037184182754983, "grad_norm": 0.3239643523627119, "learning_rate": 3.134370855038871e-07, "loss": 2.6367, "step": 46776 }, { "epoch": 2.903780495375256, "grad_norm": 0.3364700044707952, "learning_rate": 3.130334637338195e-07, "loss": 2.6893, "step": 46777 }, { "epoch": 2.903842572475014, "grad_norm": 0.34208201912074965, "learning_rate": 3.126301011927235e-07, "loss": 2.6863, "step": 46778 }, { "epoch": 2.903904649574772, "grad_norm": 0.34508026175930995, "learning_rate": 3.122269978826975e-07, "loss": 2.7118, "step": 46779 }, { "epoch": 2.90396672667453, "grad_norm": 0.32962503881096206, "learning_rate": 3.1182415380583975e-07, "loss": 2.6657, "step": 46780 }, { "epoch": 2.9040288037742874, "grad_norm": 0.3310374027567365, "learning_rate": 3.114215689642652e-07, "loss": 2.7091, "step": 46781 }, { "epoch": 2.9040908808740458, "grad_norm": 0.3305582696316174, "learning_rate": 3.1101924336006115e-07, "loss": 2.6531, "step": 46782 }, { "epoch": 2.9041529579738032, "grad_norm": 0.3331061299188025, "learning_rate": 3.1061717699533697e-07, "loss": 2.6817, "step": 46783 }, { "epoch": 2.9042150350735616, "grad_norm": 0.3452936106905172, "learning_rate": 3.1021536987218545e-07, "loss": 2.7101, "step": 46784 }, { "epoch": 2.904277112173319, "grad_norm": 0.3396880352152913, "learning_rate": 3.098138219926938e-07, "loss": 2.7197, "step": 46785 }, { "epoch": 2.9043391892730774, "grad_norm": 0.3313094652530214, "learning_rate": 3.09412533358977e-07, "loss": 2.6713, "step": 46786 }, { "epoch": 2.904401266372835, "grad_norm": 0.33054032779305276, "learning_rate": 3.090115039731112e-07, "loss": 2.7376, "step": 46787 }, { "epoch": 2.904463343472593, "grad_norm": 0.33706715740478294, "learning_rate": 3.086107338372002e-07, "loss": 2.6275, "step": 46788 }, { "epoch": 2.9045254205723507, "grad_norm": 0.33602770991390957, "learning_rate": 3.0821022295332567e-07, "loss": 2.7099, "step": 46789 }, { "epoch": 2.9045874976721087, "grad_norm": 0.34117925155069173, "learning_rate": 3.0780997132358046e-07, "loss": 2.5768, "step": 46790 }, { "epoch": 2.9046495747718666, "grad_norm": 0.34165233280577934, "learning_rate": 3.074099789500573e-07, "loss": 2.6015, "step": 46791 }, { "epoch": 2.9047116518716245, "grad_norm": 0.33994337792004203, "learning_rate": 3.070102458348434e-07, "loss": 2.661, "step": 46792 }, { "epoch": 2.9047737289713824, "grad_norm": 0.3609890145535057, "learning_rate": 3.066107719800093e-07, "loss": 2.6654, "step": 46793 }, { "epoch": 2.9048358060711403, "grad_norm": 0.31756764825147554, "learning_rate": 3.0621155738765895e-07, "loss": 2.6705, "step": 46794 }, { "epoch": 2.9048978831708983, "grad_norm": 0.3160251630138227, "learning_rate": 3.0581260205986837e-07, "loss": 2.6008, "step": 46795 }, { "epoch": 2.904959960270656, "grad_norm": 0.3248029065384384, "learning_rate": 3.054139059987138e-07, "loss": 2.6643, "step": 46796 }, { "epoch": 2.905022037370414, "grad_norm": 0.3370697767355332, "learning_rate": 3.0501546920627676e-07, "loss": 2.5514, "step": 46797 }, { "epoch": 2.905084114470172, "grad_norm": 0.3418816052728001, "learning_rate": 3.0461729168463905e-07, "loss": 2.6087, "step": 46798 }, { "epoch": 2.90514619156993, "grad_norm": 0.33466291376486623, "learning_rate": 3.042193734358767e-07, "loss": 2.5962, "step": 46799 }, { "epoch": 2.905208268669688, "grad_norm": 0.34067648227352665, "learning_rate": 3.0382171446206584e-07, "loss": 2.6761, "step": 46800 }, { "epoch": 2.9052703457694458, "grad_norm": 0.3274446427537726, "learning_rate": 3.0342431476528265e-07, "loss": 2.545, "step": 46801 }, { "epoch": 2.9053324228692037, "grad_norm": 0.33101207042493497, "learning_rate": 3.0302717434759765e-07, "loss": 2.6473, "step": 46802 }, { "epoch": 2.9053944999689616, "grad_norm": 0.33359030290738945, "learning_rate": 3.0263029321108136e-07, "loss": 2.6069, "step": 46803 }, { "epoch": 2.9054565770687195, "grad_norm": 0.3390940146149452, "learning_rate": 3.0223367135781e-07, "loss": 2.6813, "step": 46804 }, { "epoch": 2.9055186541684774, "grad_norm": 0.33082163887149907, "learning_rate": 3.018373087898541e-07, "loss": 2.6266, "step": 46805 }, { "epoch": 2.905580731268235, "grad_norm": 0.3176077295184621, "learning_rate": 3.01441205509273e-07, "loss": 2.5414, "step": 46806 }, { "epoch": 2.9056428083679933, "grad_norm": 0.325349181376398, "learning_rate": 3.0104536151813743e-07, "loss": 2.5704, "step": 46807 }, { "epoch": 2.9057048854677507, "grad_norm": 0.3414189565463219, "learning_rate": 3.0064977681851236e-07, "loss": 2.6059, "step": 46808 }, { "epoch": 2.905766962567509, "grad_norm": 0.3191948792041747, "learning_rate": 3.002544514124683e-07, "loss": 2.6761, "step": 46809 }, { "epoch": 2.9058290396672666, "grad_norm": 0.3541290693335409, "learning_rate": 2.9985938530205923e-07, "loss": 2.6704, "step": 46810 }, { "epoch": 2.9058911167670245, "grad_norm": 0.3179131756891256, "learning_rate": 2.9946457848934464e-07, "loss": 2.673, "step": 46811 }, { "epoch": 2.9059531938667824, "grad_norm": 0.3229427967984507, "learning_rate": 2.99070030976395e-07, "loss": 2.6229, "step": 46812 }, { "epoch": 2.9060152709665403, "grad_norm": 0.34469461698265885, "learning_rate": 2.986757427652587e-07, "loss": 2.6731, "step": 46813 }, { "epoch": 2.9060773480662982, "grad_norm": 0.33756378109806623, "learning_rate": 2.982817138579952e-07, "loss": 2.6641, "step": 46814 }, { "epoch": 2.906139425166056, "grad_norm": 0.3235659289039279, "learning_rate": 2.97887944256664e-07, "loss": 2.6926, "step": 46815 }, { "epoch": 2.906201502265814, "grad_norm": 0.33112466123267675, "learning_rate": 2.9749443396331346e-07, "loss": 2.6878, "step": 46816 }, { "epoch": 2.906263579365572, "grad_norm": 0.34135934345990626, "learning_rate": 2.97101182980003e-07, "loss": 2.7112, "step": 46817 }, { "epoch": 2.90632565646533, "grad_norm": 0.346711546817477, "learning_rate": 2.9670819130878657e-07, "loss": 2.687, "step": 46818 }, { "epoch": 2.906387733565088, "grad_norm": 0.3097770794630351, "learning_rate": 2.9631545895170143e-07, "loss": 2.6469, "step": 46819 }, { "epoch": 2.9064498106648458, "grad_norm": 0.34154989856254403, "learning_rate": 2.95922985910807e-07, "loss": 2.7193, "step": 46820 }, { "epoch": 2.9065118877646037, "grad_norm": 0.3472507555152139, "learning_rate": 2.955307721881462e-07, "loss": 2.7329, "step": 46821 }, { "epoch": 2.9065739648643616, "grad_norm": 0.35108766278030784, "learning_rate": 2.951388177857728e-07, "loss": 2.7076, "step": 46822 }, { "epoch": 2.9066360419641195, "grad_norm": 0.33208599761187685, "learning_rate": 2.9474712270572414e-07, "loss": 2.6531, "step": 46823 }, { "epoch": 2.9066981190638774, "grad_norm": 0.32795912091939255, "learning_rate": 2.943556869500486e-07, "loss": 2.7258, "step": 46824 }, { "epoch": 2.9067601961636353, "grad_norm": 0.3584516729996651, "learning_rate": 2.939645105207778e-07, "loss": 2.6453, "step": 46825 }, { "epoch": 2.9068222732633933, "grad_norm": 0.3465514469705979, "learning_rate": 2.935735934199657e-07, "loss": 2.5986, "step": 46826 }, { "epoch": 2.906884350363151, "grad_norm": 0.32423400843001526, "learning_rate": 2.931829356496496e-07, "loss": 2.6373, "step": 46827 }, { "epoch": 2.906946427462909, "grad_norm": 0.3219575623296964, "learning_rate": 2.9279253721186116e-07, "loss": 2.7188, "step": 46828 }, { "epoch": 2.9070085045626666, "grad_norm": 0.32428964735931076, "learning_rate": 2.924023981086432e-07, "loss": 2.5983, "step": 46829 }, { "epoch": 2.907070581662425, "grad_norm": 0.33818611278471494, "learning_rate": 2.9201251834202746e-07, "loss": 2.677, "step": 46830 }, { "epoch": 2.9071326587621824, "grad_norm": 0.322642920451968, "learning_rate": 2.916228979140512e-07, "loss": 2.6935, "step": 46831 }, { "epoch": 2.9071947358619408, "grad_norm": 0.3477970331151298, "learning_rate": 2.9123353682674604e-07, "loss": 2.6581, "step": 46832 }, { "epoch": 2.9072568129616982, "grad_norm": 0.3356389670955495, "learning_rate": 2.9084443508213824e-07, "loss": 2.6207, "step": 46833 }, { "epoch": 2.907318890061456, "grad_norm": 0.33326608257592266, "learning_rate": 2.9045559268226494e-07, "loss": 2.6843, "step": 46834 }, { "epoch": 2.907380967161214, "grad_norm": 0.35272404263762774, "learning_rate": 2.9006700962915247e-07, "loss": 2.6428, "step": 46835 }, { "epoch": 2.907443044260972, "grad_norm": 0.33394643780033123, "learning_rate": 2.896786859248268e-07, "loss": 2.6193, "step": 46836 }, { "epoch": 2.90750512136073, "grad_norm": 0.33878368261255815, "learning_rate": 2.8929062157131424e-07, "loss": 2.8192, "step": 46837 }, { "epoch": 2.907567198460488, "grad_norm": 0.3168693144512053, "learning_rate": 2.889028165706409e-07, "loss": 2.5576, "step": 46838 }, { "epoch": 2.9076292755602458, "grad_norm": 0.3325062103516113, "learning_rate": 2.8851527092482735e-07, "loss": 2.6533, "step": 46839 }, { "epoch": 2.9076913526600037, "grad_norm": 0.3294326068078377, "learning_rate": 2.881279846359053e-07, "loss": 2.6966, "step": 46840 }, { "epoch": 2.9077534297597616, "grad_norm": 0.33647162093309463, "learning_rate": 2.877409577058787e-07, "loss": 2.6039, "step": 46841 }, { "epoch": 2.9078155068595195, "grad_norm": 0.3351052976883069, "learning_rate": 2.8735419013677935e-07, "loss": 2.5985, "step": 46842 }, { "epoch": 2.9078775839592774, "grad_norm": 0.3463280519760334, "learning_rate": 2.8696768193062217e-07, "loss": 2.7332, "step": 46843 }, { "epoch": 2.9079396610590353, "grad_norm": 0.3424347946302417, "learning_rate": 2.865814330894223e-07, "loss": 2.6406, "step": 46844 }, { "epoch": 2.9080017381587933, "grad_norm": 0.33618772413945214, "learning_rate": 2.861954436151948e-07, "loss": 2.6773, "step": 46845 }, { "epoch": 2.908063815258551, "grad_norm": 0.33164003549620014, "learning_rate": 2.8580971350995466e-07, "loss": 2.7529, "step": 46846 }, { "epoch": 2.908125892358309, "grad_norm": 0.32293224168973783, "learning_rate": 2.85424242775717e-07, "loss": 2.7082, "step": 46847 }, { "epoch": 2.908187969458067, "grad_norm": 0.35745924874107843, "learning_rate": 2.850390314144857e-07, "loss": 2.6863, "step": 46848 }, { "epoch": 2.908250046557825, "grad_norm": 0.33420315765419223, "learning_rate": 2.846540794282759e-07, "loss": 2.6312, "step": 46849 }, { "epoch": 2.908312123657583, "grad_norm": 0.34172111761583746, "learning_rate": 2.8426938681909154e-07, "loss": 2.6373, "step": 46850 }, { "epoch": 2.9083742007573408, "grad_norm": 0.342581148906311, "learning_rate": 2.838849535889476e-07, "loss": 2.6706, "step": 46851 }, { "epoch": 2.9084362778570982, "grad_norm": 0.3408944888130475, "learning_rate": 2.83500779739837e-07, "loss": 2.5723, "step": 46852 }, { "epoch": 2.9084983549568566, "grad_norm": 0.33605489232208813, "learning_rate": 2.8311686527378034e-07, "loss": 2.6013, "step": 46853 }, { "epoch": 2.908560432056614, "grad_norm": 0.3385519050614968, "learning_rate": 2.8273321019277045e-07, "loss": 2.6792, "step": 46854 }, { "epoch": 2.9086225091563724, "grad_norm": 0.32192650494011554, "learning_rate": 2.8234981449881124e-07, "loss": 2.6387, "step": 46855 }, { "epoch": 2.90868458625613, "grad_norm": 0.32446709279251906, "learning_rate": 2.819666781938957e-07, "loss": 2.6909, "step": 46856 }, { "epoch": 2.9087466633558883, "grad_norm": 0.3424914142653255, "learning_rate": 2.8158380128003315e-07, "loss": 2.6686, "step": 46857 }, { "epoch": 2.9088087404556457, "grad_norm": 0.3374518704534781, "learning_rate": 2.812011837592221e-07, "loss": 2.6231, "step": 46858 }, { "epoch": 2.9088708175554037, "grad_norm": 0.35114318424939256, "learning_rate": 2.8081882563344985e-07, "loss": 2.7653, "step": 46859 }, { "epoch": 2.9089328946551616, "grad_norm": 0.32890483738104315, "learning_rate": 2.8043672690472036e-07, "loss": 2.6188, "step": 46860 }, { "epoch": 2.9089949717549195, "grad_norm": 0.32216161725691456, "learning_rate": 2.800548875750153e-07, "loss": 2.6906, "step": 46861 }, { "epoch": 2.9090570488546774, "grad_norm": 0.31410798981723626, "learning_rate": 2.796733076463387e-07, "loss": 2.6799, "step": 46862 }, { "epoch": 2.9091191259544353, "grad_norm": 0.33524563982245337, "learning_rate": 2.792919871206723e-07, "loss": 2.6985, "step": 46863 }, { "epoch": 2.9091812030541933, "grad_norm": 0.3207463025149786, "learning_rate": 2.789109260000144e-07, "loss": 2.5828, "step": 46864 }, { "epoch": 2.909243280153951, "grad_norm": 0.34052408018263114, "learning_rate": 2.785301242863414e-07, "loss": 2.712, "step": 46865 }, { "epoch": 2.909305357253709, "grad_norm": 0.32770520558999744, "learning_rate": 2.781495819816515e-07, "loss": 2.7067, "step": 46866 }, { "epoch": 2.909367434353467, "grad_norm": 0.34963554412250003, "learning_rate": 2.777692990879266e-07, "loss": 2.6732, "step": 46867 }, { "epoch": 2.909429511453225, "grad_norm": 0.33848371987939124, "learning_rate": 2.773892756071539e-07, "loss": 2.651, "step": 46868 }, { "epoch": 2.909491588552983, "grad_norm": 0.3357947179188884, "learning_rate": 2.7700951154130407e-07, "loss": 2.6858, "step": 46869 }, { "epoch": 2.9095536656527408, "grad_norm": 0.3268725242302172, "learning_rate": 2.7663000689236994e-07, "loss": 2.6578, "step": 46870 }, { "epoch": 2.9096157427524987, "grad_norm": 0.32725155472076484, "learning_rate": 2.7625076166232223e-07, "loss": 2.669, "step": 46871 }, { "epoch": 2.9096778198522566, "grad_norm": 0.32377567809867996, "learning_rate": 2.758717758531537e-07, "loss": 2.6985, "step": 46872 }, { "epoch": 2.9097398969520145, "grad_norm": 0.3364448818769615, "learning_rate": 2.7549304946682954e-07, "loss": 2.596, "step": 46873 }, { "epoch": 2.9098019740517724, "grad_norm": 0.3342546183923999, "learning_rate": 2.751145825053314e-07, "loss": 2.6961, "step": 46874 }, { "epoch": 2.9098640511515304, "grad_norm": 0.33004003225972606, "learning_rate": 2.7473637497062444e-07, "loss": 2.6571, "step": 46875 }, { "epoch": 2.9099261282512883, "grad_norm": 0.3219806567173746, "learning_rate": 2.7435842686469595e-07, "loss": 2.6552, "step": 46876 }, { "epoch": 2.9099882053510457, "grad_norm": 0.3225821241341996, "learning_rate": 2.7398073818951097e-07, "loss": 2.6507, "step": 46877 }, { "epoch": 2.910050282450804, "grad_norm": 0.3224071307608896, "learning_rate": 2.7360330894703466e-07, "loss": 2.6641, "step": 46878 }, { "epoch": 2.9101123595505616, "grad_norm": 0.3324910962554708, "learning_rate": 2.732261391392432e-07, "loss": 2.6792, "step": 46879 }, { "epoch": 2.91017443665032, "grad_norm": 0.3418210702234296, "learning_rate": 2.728492287681073e-07, "loss": 2.695, "step": 46880 }, { "epoch": 2.9102365137500774, "grad_norm": 0.32840260615220757, "learning_rate": 2.724725778355808e-07, "loss": 2.5015, "step": 46881 }, { "epoch": 2.9102985908498353, "grad_norm": 0.32330298644308025, "learning_rate": 2.7209618634364554e-07, "loss": 2.5291, "step": 46882 }, { "epoch": 2.9103606679495932, "grad_norm": 0.3190533849103507, "learning_rate": 2.7172005429424997e-07, "loss": 2.6389, "step": 46883 }, { "epoch": 2.910422745049351, "grad_norm": 0.3377189972107887, "learning_rate": 2.713441816893647e-07, "loss": 2.6396, "step": 46884 }, { "epoch": 2.910484822149109, "grad_norm": 0.3458706186008086, "learning_rate": 2.709685685309493e-07, "loss": 2.6723, "step": 46885 }, { "epoch": 2.910546899248867, "grad_norm": 0.3477912812045554, "learning_rate": 2.7059321482096334e-07, "loss": 2.6957, "step": 46886 }, { "epoch": 2.910608976348625, "grad_norm": 0.33872472753247057, "learning_rate": 2.702181205613663e-07, "loss": 2.6479, "step": 46887 }, { "epoch": 2.910671053448383, "grad_norm": 0.33751100372107673, "learning_rate": 2.6984328575410665e-07, "loss": 2.6451, "step": 46888 }, { "epoch": 2.9107331305481408, "grad_norm": 0.3302671579658451, "learning_rate": 2.694687104011551e-07, "loss": 2.6611, "step": 46889 }, { "epoch": 2.9107952076478987, "grad_norm": 0.31801918201578, "learning_rate": 2.690943945044544e-07, "loss": 2.6461, "step": 46890 }, { "epoch": 2.9108572847476566, "grad_norm": 0.3391406580062249, "learning_rate": 2.6872033806595866e-07, "loss": 2.7399, "step": 46891 }, { "epoch": 2.9109193618474145, "grad_norm": 0.33860477323944643, "learning_rate": 2.683465410876218e-07, "loss": 2.6475, "step": 46892 }, { "epoch": 2.9109814389471724, "grad_norm": 0.3316372047850037, "learning_rate": 2.6797300357139786e-07, "loss": 2.6947, "step": 46893 }, { "epoch": 2.9110435160469303, "grad_norm": 0.32042440925109744, "learning_rate": 2.6759972551922976e-07, "loss": 2.6721, "step": 46894 }, { "epoch": 2.9111055931466883, "grad_norm": 0.3406512762136502, "learning_rate": 2.672267069330658e-07, "loss": 2.7142, "step": 46895 }, { "epoch": 2.911167670246446, "grad_norm": 0.3482524495904708, "learning_rate": 2.668539478148546e-07, "loss": 2.6923, "step": 46896 }, { "epoch": 2.911229747346204, "grad_norm": 0.32944199248138184, "learning_rate": 2.6648144816653896e-07, "loss": 2.7184, "step": 46897 }, { "epoch": 2.911291824445962, "grad_norm": 0.3303277240085094, "learning_rate": 2.661092079900618e-07, "loss": 2.6582, "step": 46898 }, { "epoch": 2.91135390154572, "grad_norm": 0.31667604835106644, "learning_rate": 2.657372272873715e-07, "loss": 2.6191, "step": 46899 }, { "epoch": 2.9114159786454774, "grad_norm": 0.32649240334656493, "learning_rate": 2.653655060603999e-07, "loss": 2.6191, "step": 46900 }, { "epoch": 2.9114780557452358, "grad_norm": 0.3615661043599062, "learning_rate": 2.6499404431108434e-07, "loss": 2.6824, "step": 46901 }, { "epoch": 2.9115401328449932, "grad_norm": 0.3264597347771499, "learning_rate": 2.646228420413732e-07, "loss": 2.623, "step": 46902 }, { "epoch": 2.9116022099447516, "grad_norm": 0.32172531147176325, "learning_rate": 2.642518992531984e-07, "loss": 2.6028, "step": 46903 }, { "epoch": 2.911664287044509, "grad_norm": 0.3225450864591768, "learning_rate": 2.6388121594849714e-07, "loss": 2.6211, "step": 46904 }, { "epoch": 2.9117263641442674, "grad_norm": 0.3434501553981472, "learning_rate": 2.635107921292013e-07, "loss": 2.6187, "step": 46905 }, { "epoch": 2.911788441244025, "grad_norm": 0.32406279248451564, "learning_rate": 2.63140627797237e-07, "loss": 2.5566, "step": 46906 }, { "epoch": 2.911850518343783, "grad_norm": 0.3438852759260748, "learning_rate": 2.627707229545473e-07, "loss": 2.6319, "step": 46907 }, { "epoch": 2.9119125954435408, "grad_norm": 0.3391279267326467, "learning_rate": 2.624010776030583e-07, "loss": 2.6972, "step": 46908 }, { "epoch": 2.9119746725432987, "grad_norm": 0.43783721530049147, "learning_rate": 2.6203169174469635e-07, "loss": 2.6994, "step": 46909 }, { "epoch": 2.9120367496430566, "grad_norm": 0.3362366541975761, "learning_rate": 2.6166256538138756e-07, "loss": 2.7907, "step": 46910 }, { "epoch": 2.9120988267428145, "grad_norm": 0.3368554105707192, "learning_rate": 2.612936985150638e-07, "loss": 2.7258, "step": 46911 }, { "epoch": 2.9121609038425724, "grad_norm": 0.3442502283498634, "learning_rate": 2.609250911476402e-07, "loss": 2.7049, "step": 46912 }, { "epoch": 2.9122229809423303, "grad_norm": 0.3260088530330082, "learning_rate": 2.605567432810485e-07, "loss": 2.7359, "step": 46913 }, { "epoch": 2.9122850580420883, "grad_norm": 0.33120271867123463, "learning_rate": 2.601886549172039e-07, "loss": 2.6673, "step": 46914 }, { "epoch": 2.912347135141846, "grad_norm": 0.3394816541370904, "learning_rate": 2.598208260580326e-07, "loss": 2.7047, "step": 46915 }, { "epoch": 2.912409212241604, "grad_norm": 0.3342383461977057, "learning_rate": 2.594532567054497e-07, "loss": 2.6597, "step": 46916 }, { "epoch": 2.912471289341362, "grad_norm": 0.33446111728784295, "learning_rate": 2.5908594686137045e-07, "loss": 2.7666, "step": 46917 }, { "epoch": 2.91253336644112, "grad_norm": 0.32110703016231235, "learning_rate": 2.5871889652772095e-07, "loss": 2.6555, "step": 46918 }, { "epoch": 2.912595443540878, "grad_norm": 0.3610696847804007, "learning_rate": 2.5835210570641086e-07, "loss": 2.648, "step": 46919 }, { "epoch": 2.9126575206406358, "grad_norm": 0.3253555590360872, "learning_rate": 2.5798557439934425e-07, "loss": 2.65, "step": 46920 }, { "epoch": 2.9127195977403937, "grad_norm": 0.3435816460941259, "learning_rate": 2.5761930260844723e-07, "loss": 2.7344, "step": 46921 }, { "epoch": 2.9127816748401516, "grad_norm": 0.34339248966242436, "learning_rate": 2.5725329033562953e-07, "loss": 2.6918, "step": 46922 }, { "epoch": 2.9128437519399095, "grad_norm": 0.331637710270918, "learning_rate": 2.5688753758279507e-07, "loss": 2.6082, "step": 46923 }, { "epoch": 2.9129058290396674, "grad_norm": 0.35005894317890457, "learning_rate": 2.565220443518479e-07, "loss": 2.681, "step": 46924 }, { "epoch": 2.912967906139425, "grad_norm": 0.33019262506124, "learning_rate": 2.5615681064470876e-07, "loss": 2.7033, "step": 46925 }, { "epoch": 2.9130299832391833, "grad_norm": 0.33114305515249887, "learning_rate": 2.557918364632705e-07, "loss": 2.6133, "step": 46926 }, { "epoch": 2.9130920603389407, "grad_norm": 0.32580030143157734, "learning_rate": 2.5542712180944283e-07, "loss": 2.7161, "step": 46927 }, { "epoch": 2.913154137438699, "grad_norm": 0.3471033543200453, "learning_rate": 2.550626666851241e-07, "loss": 2.6069, "step": 46928 }, { "epoch": 2.9132162145384566, "grad_norm": 0.3292213131515599, "learning_rate": 2.54698471092224e-07, "loss": 2.5394, "step": 46929 }, { "epoch": 2.9132782916382145, "grad_norm": 0.35746977438230926, "learning_rate": 2.543345350326409e-07, "loss": 2.7414, "step": 46930 }, { "epoch": 2.9133403687379724, "grad_norm": 0.3507542856808358, "learning_rate": 2.5397085850826787e-07, "loss": 2.6512, "step": 46931 }, { "epoch": 2.9134024458377303, "grad_norm": 0.3384054924895396, "learning_rate": 2.5360744152100323e-07, "loss": 2.6226, "step": 46932 }, { "epoch": 2.9134645229374883, "grad_norm": 0.3534410619345429, "learning_rate": 2.532442840727456e-07, "loss": 2.683, "step": 46933 }, { "epoch": 2.913526600037246, "grad_norm": 0.3235478411075307, "learning_rate": 2.5288138616539337e-07, "loss": 2.6995, "step": 46934 }, { "epoch": 2.913588677137004, "grad_norm": 0.3374493424050405, "learning_rate": 2.525187478008284e-07, "loss": 2.6552, "step": 46935 }, { "epoch": 2.913650754236762, "grad_norm": 0.3357886938492147, "learning_rate": 2.5215636898095474e-07, "loss": 2.6699, "step": 46936 }, { "epoch": 2.91371283133652, "grad_norm": 0.3454249820218341, "learning_rate": 2.517942497076542e-07, "loss": 2.7308, "step": 46937 }, { "epoch": 2.913774908436278, "grad_norm": 0.31744798431312315, "learning_rate": 2.514323899828197e-07, "loss": 2.6581, "step": 46938 }, { "epoch": 2.9138369855360358, "grad_norm": 0.3255768104743078, "learning_rate": 2.510707898083442e-07, "loss": 2.6647, "step": 46939 }, { "epoch": 2.9138990626357937, "grad_norm": 0.3394441008602924, "learning_rate": 2.507094491861095e-07, "loss": 2.6238, "step": 46940 }, { "epoch": 2.9139611397355516, "grad_norm": 0.33354991964382985, "learning_rate": 2.5034836811799743e-07, "loss": 2.6685, "step": 46941 }, { "epoch": 2.9140232168353095, "grad_norm": 0.33119103203736777, "learning_rate": 2.499875466058954e-07, "loss": 2.7384, "step": 46942 }, { "epoch": 2.9140852939350674, "grad_norm": 0.34814715284133585, "learning_rate": 2.496269846516852e-07, "loss": 2.6584, "step": 46943 }, { "epoch": 2.9141473710348254, "grad_norm": 0.34264301742420566, "learning_rate": 2.492666822572487e-07, "loss": 2.7166, "step": 46944 }, { "epoch": 2.9142094481345833, "grad_norm": 0.3269687899502286, "learning_rate": 2.489066394244677e-07, "loss": 2.5729, "step": 46945 }, { "epoch": 2.914271525234341, "grad_norm": 0.3555124696108603, "learning_rate": 2.4854685615521843e-07, "loss": 2.6612, "step": 46946 }, { "epoch": 2.914333602334099, "grad_norm": 0.3435274942901196, "learning_rate": 2.481873324513717e-07, "loss": 2.6349, "step": 46947 }, { "epoch": 2.9143956794338566, "grad_norm": 0.33281118007630994, "learning_rate": 2.478280683148149e-07, "loss": 2.586, "step": 46948 }, { "epoch": 2.914457756533615, "grad_norm": 0.3502296507510596, "learning_rate": 2.4746906374741305e-07, "loss": 2.5631, "step": 46949 }, { "epoch": 2.9145198336333724, "grad_norm": 0.3488392579287433, "learning_rate": 2.4711031875104816e-07, "loss": 2.6952, "step": 46950 }, { "epoch": 2.9145819107331308, "grad_norm": 0.3486281643084799, "learning_rate": 2.4675183332757976e-07, "loss": 2.6851, "step": 46951 }, { "epoch": 2.9146439878328883, "grad_norm": 0.3406643551655831, "learning_rate": 2.4639360747888975e-07, "loss": 2.6341, "step": 46952 }, { "epoch": 2.9147060649326466, "grad_norm": 0.33111735963197025, "learning_rate": 2.460356412068432e-07, "loss": 2.6567, "step": 46953 }, { "epoch": 2.914768142032404, "grad_norm": 0.33853426582854484, "learning_rate": 2.4567793451330534e-07, "loss": 2.6594, "step": 46954 }, { "epoch": 2.914830219132162, "grad_norm": 0.3261126618688719, "learning_rate": 2.453204874001469e-07, "loss": 2.6011, "step": 46955 }, { "epoch": 2.91489229623192, "grad_norm": 0.33114969246291487, "learning_rate": 2.44963299869222e-07, "loss": 2.7672, "step": 46956 }, { "epoch": 2.914954373331678, "grad_norm": 0.35310488491663133, "learning_rate": 2.4460637192240674e-07, "loss": 2.6477, "step": 46957 }, { "epoch": 2.9150164504314358, "grad_norm": 0.32951033021626297, "learning_rate": 2.4424970356156094e-07, "loss": 2.6637, "step": 46958 }, { "epoch": 2.9150785275311937, "grad_norm": 0.32909554043022804, "learning_rate": 2.4389329478853863e-07, "loss": 2.7242, "step": 46959 }, { "epoch": 2.9151406046309516, "grad_norm": 0.3486481884896853, "learning_rate": 2.435371456052049e-07, "loss": 2.6394, "step": 46960 }, { "epoch": 2.9152026817307095, "grad_norm": 0.32088852470629314, "learning_rate": 2.431812560134195e-07, "loss": 2.6228, "step": 46961 }, { "epoch": 2.9152647588304674, "grad_norm": 0.34598324089930804, "learning_rate": 2.4282562601503634e-07, "loss": 2.6816, "step": 46962 }, { "epoch": 2.9153268359302253, "grad_norm": 0.3270916903798948, "learning_rate": 2.424702556119096e-07, "loss": 2.6429, "step": 46963 }, { "epoch": 2.9153889130299833, "grad_norm": 0.3277043643710262, "learning_rate": 2.4211514480589337e-07, "loss": 2.7099, "step": 46964 }, { "epoch": 2.915450990129741, "grad_norm": 0.33858654262756493, "learning_rate": 2.4176029359884167e-07, "loss": 2.5703, "step": 46965 }, { "epoch": 2.915513067229499, "grad_norm": 0.32741126135897397, "learning_rate": 2.414057019926086e-07, "loss": 2.5706, "step": 46966 }, { "epoch": 2.915575144329257, "grad_norm": 0.3388628011194235, "learning_rate": 2.410513699890371e-07, "loss": 2.6676, "step": 46967 }, { "epoch": 2.915637221429015, "grad_norm": 0.31183364873058356, "learning_rate": 2.406972975899813e-07, "loss": 2.7315, "step": 46968 }, { "epoch": 2.915699298528773, "grad_norm": 0.3449473336005732, "learning_rate": 2.4034348479728965e-07, "loss": 2.4934, "step": 46969 }, { "epoch": 2.9157613756285308, "grad_norm": 0.3319774584112314, "learning_rate": 2.3998993161279957e-07, "loss": 2.5943, "step": 46970 }, { "epoch": 2.9158234527282887, "grad_norm": 0.3103166242442587, "learning_rate": 2.396366380383652e-07, "loss": 2.6389, "step": 46971 }, { "epoch": 2.9158855298280466, "grad_norm": 0.3344309871362365, "learning_rate": 2.392836040758295e-07, "loss": 2.641, "step": 46972 }, { "epoch": 2.915947606927804, "grad_norm": 0.3333786161281676, "learning_rate": 2.3893082972702984e-07, "loss": 2.6859, "step": 46973 }, { "epoch": 2.9160096840275624, "grad_norm": 0.33425497503437096, "learning_rate": 2.385783149938037e-07, "loss": 2.6758, "step": 46974 }, { "epoch": 2.91607176112732, "grad_norm": 0.33268162477828317, "learning_rate": 2.382260598779995e-07, "loss": 2.7116, "step": 46975 }, { "epoch": 2.9161338382270783, "grad_norm": 0.3339964769635876, "learning_rate": 2.3787406438144921e-07, "loss": 2.6573, "step": 46976 }, { "epoch": 2.9161959153268358, "grad_norm": 0.33993473150532527, "learning_rate": 2.3752232850599022e-07, "loss": 2.7763, "step": 46977 }, { "epoch": 2.9162579924265937, "grad_norm": 0.3236564678233026, "learning_rate": 2.3717085225345438e-07, "loss": 2.5842, "step": 46978 }, { "epoch": 2.9163200695263516, "grad_norm": 0.35786974836247637, "learning_rate": 2.368196356256791e-07, "loss": 2.7533, "step": 46979 }, { "epoch": 2.9163821466261095, "grad_norm": 0.3264980626757104, "learning_rate": 2.364686786244963e-07, "loss": 2.6676, "step": 46980 }, { "epoch": 2.9164442237258674, "grad_norm": 0.3449600012888238, "learning_rate": 2.3611798125173778e-07, "loss": 2.7491, "step": 46981 }, { "epoch": 2.9165063008256253, "grad_norm": 0.34133184132786965, "learning_rate": 2.3576754350923547e-07, "loss": 2.609, "step": 46982 }, { "epoch": 2.9165683779253833, "grad_norm": 0.3170261080509613, "learning_rate": 2.354173653988101e-07, "loss": 2.6734, "step": 46983 }, { "epoch": 2.916630455025141, "grad_norm": 0.34811932016392866, "learning_rate": 2.3506744692229354e-07, "loss": 2.6815, "step": 46984 }, { "epoch": 2.916692532124899, "grad_norm": 0.33620658344274035, "learning_rate": 2.3471778808151212e-07, "loss": 2.6119, "step": 46985 }, { "epoch": 2.916754609224657, "grad_norm": 0.32667824277429486, "learning_rate": 2.3436838887829214e-07, "loss": 2.7393, "step": 46986 }, { "epoch": 2.916816686324415, "grad_norm": 0.32424569305244794, "learning_rate": 2.3401924931444886e-07, "loss": 2.6574, "step": 46987 }, { "epoch": 2.916878763424173, "grad_norm": 0.3306379029617257, "learning_rate": 2.3367036939180853e-07, "loss": 2.6431, "step": 46988 }, { "epoch": 2.9169408405239308, "grad_norm": 0.34990876447709846, "learning_rate": 2.3332174911219197e-07, "loss": 2.6907, "step": 46989 }, { "epoch": 2.9170029176236887, "grad_norm": 0.3421901184574084, "learning_rate": 2.3297338847741989e-07, "loss": 2.7228, "step": 46990 }, { "epoch": 2.9170649947234466, "grad_norm": 0.3335764998860472, "learning_rate": 2.3262528748930202e-07, "loss": 2.6638, "step": 46991 }, { "epoch": 2.9171270718232045, "grad_norm": 0.3416212775254067, "learning_rate": 2.3227744614966462e-07, "loss": 2.6859, "step": 46992 }, { "epoch": 2.9171891489229624, "grad_norm": 0.32876365814556147, "learning_rate": 2.3192986446031185e-07, "loss": 2.6443, "step": 46993 }, { "epoch": 2.9172512260227204, "grad_norm": 0.3249308796887105, "learning_rate": 2.3158254242306998e-07, "loss": 2.6765, "step": 46994 }, { "epoch": 2.9173133031224783, "grad_norm": 0.3612977504850772, "learning_rate": 2.312354800397376e-07, "loss": 2.7163, "step": 46995 }, { "epoch": 2.9173753802222357, "grad_norm": 0.33672839591245246, "learning_rate": 2.3088867731212994e-07, "loss": 2.714, "step": 46996 }, { "epoch": 2.917437457321994, "grad_norm": 0.33366930193166866, "learning_rate": 2.3054213424206217e-07, "loss": 2.6038, "step": 46997 }, { "epoch": 2.9174995344217516, "grad_norm": 0.3503429559030534, "learning_rate": 2.3019585083133842e-07, "loss": 2.6397, "step": 46998 }, { "epoch": 2.91756161152151, "grad_norm": 0.33620017893311016, "learning_rate": 2.2984982708176284e-07, "loss": 2.7717, "step": 46999 }, { "epoch": 2.9176236886212674, "grad_norm": 0.33271593295793245, "learning_rate": 2.2950406299514505e-07, "loss": 2.6813, "step": 47000 }, { "epoch": 2.917685765721026, "grad_norm": 0.32226991361596585, "learning_rate": 2.291585585732836e-07, "loss": 2.6143, "step": 47001 }, { "epoch": 2.9177478428207833, "grad_norm": 0.33643508256032734, "learning_rate": 2.2881331381798265e-07, "loss": 2.6551, "step": 47002 }, { "epoch": 2.917809919920541, "grad_norm": 0.32750451366939565, "learning_rate": 2.2846832873105185e-07, "loss": 2.6688, "step": 47003 }, { "epoch": 2.917871997020299, "grad_norm": 0.3361831587436504, "learning_rate": 2.2812360331427863e-07, "loss": 2.682, "step": 47004 }, { "epoch": 2.917934074120057, "grad_norm": 0.34457424354019056, "learning_rate": 2.2777913756946712e-07, "loss": 2.7095, "step": 47005 }, { "epoch": 2.917996151219815, "grad_norm": 0.3347922253361717, "learning_rate": 2.2743493149841588e-07, "loss": 2.7099, "step": 47006 }, { "epoch": 2.918058228319573, "grad_norm": 0.3486920047379451, "learning_rate": 2.2709098510292348e-07, "loss": 2.6115, "step": 47007 }, { "epoch": 2.9181203054193308, "grad_norm": 0.3479953355229939, "learning_rate": 2.2674729838477738e-07, "loss": 2.6564, "step": 47008 }, { "epoch": 2.9181823825190887, "grad_norm": 0.3319566530689725, "learning_rate": 2.2640387134577058e-07, "loss": 2.6632, "step": 47009 }, { "epoch": 2.9182444596188466, "grad_norm": 0.3445852060086443, "learning_rate": 2.2606070398769607e-07, "loss": 2.7188, "step": 47010 }, { "epoch": 2.9183065367186045, "grad_norm": 0.31365742917365114, "learning_rate": 2.2571779631235246e-07, "loss": 2.6876, "step": 47011 }, { "epoch": 2.9183686138183624, "grad_norm": 0.31960624663454434, "learning_rate": 2.2537514832152164e-07, "loss": 2.6172, "step": 47012 }, { "epoch": 2.9184306909181204, "grad_norm": 0.33564406154356635, "learning_rate": 2.2503276001699103e-07, "loss": 2.674, "step": 47013 }, { "epoch": 2.9184927680178783, "grad_norm": 0.34555068669210215, "learning_rate": 2.2469063140054258e-07, "loss": 2.6915, "step": 47014 }, { "epoch": 2.918554845117636, "grad_norm": 0.34046080901718334, "learning_rate": 2.243487624739693e-07, "loss": 2.6429, "step": 47015 }, { "epoch": 2.918616922217394, "grad_norm": 0.3522196138696448, "learning_rate": 2.2400715323905308e-07, "loss": 2.7341, "step": 47016 }, { "epoch": 2.918678999317152, "grad_norm": 0.34948304094118027, "learning_rate": 2.2366580369757585e-07, "loss": 2.6746, "step": 47017 }, { "epoch": 2.91874107641691, "grad_norm": 0.34483217993453186, "learning_rate": 2.2332471385131948e-07, "loss": 2.7182, "step": 47018 }, { "epoch": 2.918803153516668, "grad_norm": 0.3561083311181374, "learning_rate": 2.2298388370205482e-07, "loss": 2.7186, "step": 47019 }, { "epoch": 2.9188652306164258, "grad_norm": 0.3503167013823278, "learning_rate": 2.2264331325157483e-07, "loss": 2.7072, "step": 47020 }, { "epoch": 2.9189273077161833, "grad_norm": 0.32247245393702134, "learning_rate": 2.2230300250164482e-07, "loss": 2.5849, "step": 47021 }, { "epoch": 2.9189893848159416, "grad_norm": 0.3340641283436299, "learning_rate": 2.2196295145404666e-07, "loss": 2.7011, "step": 47022 }, { "epoch": 2.919051461915699, "grad_norm": 0.32604837649370727, "learning_rate": 2.2162316011054563e-07, "loss": 2.6732, "step": 47023 }, { "epoch": 2.9191135390154574, "grad_norm": 0.32766383552047706, "learning_rate": 2.212836284729236e-07, "loss": 2.7158, "step": 47024 }, { "epoch": 2.919175616115215, "grad_norm": 0.35868424912937225, "learning_rate": 2.2094435654295142e-07, "loss": 2.5991, "step": 47025 }, { "epoch": 2.919237693214973, "grad_norm": 0.33462444119198365, "learning_rate": 2.2060534432239433e-07, "loss": 2.5827, "step": 47026 }, { "epoch": 2.9192997703147308, "grad_norm": 0.34850147455456276, "learning_rate": 2.2026659181302311e-07, "loss": 2.7234, "step": 47027 }, { "epoch": 2.9193618474144887, "grad_norm": 0.33471628490940286, "learning_rate": 2.199280990166086e-07, "loss": 2.6889, "step": 47028 }, { "epoch": 2.9194239245142466, "grad_norm": 0.3399856424631059, "learning_rate": 2.1958986593491604e-07, "loss": 2.7251, "step": 47029 }, { "epoch": 2.9194860016140045, "grad_norm": 0.3303094200236093, "learning_rate": 2.1925189256970514e-07, "loss": 2.7195, "step": 47030 }, { "epoch": 2.9195480787137624, "grad_norm": 0.3482752764477508, "learning_rate": 2.1891417892274113e-07, "loss": 2.7906, "step": 47031 }, { "epoch": 2.9196101558135203, "grad_norm": 0.33542937285285357, "learning_rate": 2.1857672499578376e-07, "loss": 2.6723, "step": 47032 }, { "epoch": 2.9196722329132783, "grad_norm": 0.32113286543524244, "learning_rate": 2.1823953079059823e-07, "loss": 2.598, "step": 47033 }, { "epoch": 2.919734310013036, "grad_norm": 0.34558257463418063, "learning_rate": 2.1790259630893873e-07, "loss": 2.6902, "step": 47034 }, { "epoch": 2.919796387112794, "grad_norm": 0.32197693118172244, "learning_rate": 2.175659215525705e-07, "loss": 2.6397, "step": 47035 }, { "epoch": 2.919858464212552, "grad_norm": 0.3387455680775434, "learning_rate": 2.172295065232477e-07, "loss": 2.7675, "step": 47036 }, { "epoch": 2.91992054131231, "grad_norm": 0.3178284201859045, "learning_rate": 2.1689335122271893e-07, "loss": 2.6077, "step": 47037 }, { "epoch": 2.919982618412068, "grad_norm": 0.32505437872859133, "learning_rate": 2.1655745565274943e-07, "loss": 2.7118, "step": 47038 }, { "epoch": 2.9200446955118258, "grad_norm": 0.3220818082054485, "learning_rate": 2.1622181981508228e-07, "loss": 2.6331, "step": 47039 }, { "epoch": 2.9201067726115837, "grad_norm": 0.3434878021244181, "learning_rate": 2.1588644371146604e-07, "loss": 2.6658, "step": 47040 }, { "epoch": 2.9201688497113416, "grad_norm": 0.3364269771471624, "learning_rate": 2.1555132734366045e-07, "loss": 2.6703, "step": 47041 }, { "epoch": 2.9202309268110995, "grad_norm": 0.3399443641891159, "learning_rate": 2.1521647071340855e-07, "loss": 2.681, "step": 47042 }, { "epoch": 2.9202930039108574, "grad_norm": 0.34829676484454525, "learning_rate": 2.1488187382245895e-07, "loss": 2.719, "step": 47043 }, { "epoch": 2.920355081010615, "grad_norm": 0.33053124380242427, "learning_rate": 2.1454753667255468e-07, "loss": 2.5841, "step": 47044 }, { "epoch": 2.9204171581103733, "grad_norm": 0.3293543447572488, "learning_rate": 2.1421345926544433e-07, "loss": 2.7268, "step": 47045 }, { "epoch": 2.9204792352101308, "grad_norm": 0.3217620267102318, "learning_rate": 2.1387964160286543e-07, "loss": 2.7033, "step": 47046 }, { "epoch": 2.920541312309889, "grad_norm": 0.3401087243133544, "learning_rate": 2.135460836865666e-07, "loss": 2.7347, "step": 47047 }, { "epoch": 2.9206033894096466, "grad_norm": 0.33348387309034266, "learning_rate": 2.1321278551827972e-07, "loss": 2.667, "step": 47048 }, { "epoch": 2.920665466509405, "grad_norm": 0.3457401385131846, "learning_rate": 2.128797470997479e-07, "loss": 2.6447, "step": 47049 }, { "epoch": 2.9207275436091624, "grad_norm": 0.33960166069414943, "learning_rate": 2.125469684327086e-07, "loss": 2.65, "step": 47050 }, { "epoch": 2.9207896207089203, "grad_norm": 0.34382840909558265, "learning_rate": 2.1221444951889936e-07, "loss": 2.6624, "step": 47051 }, { "epoch": 2.9208516978086783, "grad_norm": 0.34188756557016403, "learning_rate": 2.1188219036005763e-07, "loss": 2.6539, "step": 47052 }, { "epoch": 2.920913774908436, "grad_norm": 0.34059497412186757, "learning_rate": 2.1155019095790984e-07, "loss": 2.7029, "step": 47053 }, { "epoch": 2.920975852008194, "grad_norm": 0.3388354481720837, "learning_rate": 2.1121845131418795e-07, "loss": 2.7398, "step": 47054 }, { "epoch": 2.921037929107952, "grad_norm": 0.33303093480355395, "learning_rate": 2.108869714306294e-07, "loss": 2.748, "step": 47055 }, { "epoch": 2.92110000620771, "grad_norm": 0.3245827802833739, "learning_rate": 2.1055575130896066e-07, "loss": 2.6096, "step": 47056 }, { "epoch": 2.921162083307468, "grad_norm": 0.3379679023707356, "learning_rate": 2.102247909509081e-07, "loss": 2.7141, "step": 47057 }, { "epoch": 2.9212241604072258, "grad_norm": 0.32920420373284254, "learning_rate": 2.0989409035819808e-07, "loss": 2.6282, "step": 47058 }, { "epoch": 2.9212862375069837, "grad_norm": 0.33047735460049865, "learning_rate": 2.0956364953256258e-07, "loss": 2.6525, "step": 47059 }, { "epoch": 2.9213483146067416, "grad_norm": 0.32956406103944014, "learning_rate": 2.092334684757169e-07, "loss": 2.7539, "step": 47060 }, { "epoch": 2.9214103917064995, "grad_norm": 0.3331675775489899, "learning_rate": 2.0890354718938744e-07, "loss": 2.6152, "step": 47061 }, { "epoch": 2.9214724688062574, "grad_norm": 0.3462669817344203, "learning_rate": 2.08573885675295e-07, "loss": 2.6865, "step": 47062 }, { "epoch": 2.9215345459060154, "grad_norm": 0.3323702484098231, "learning_rate": 2.082444839351605e-07, "loss": 2.6865, "step": 47063 }, { "epoch": 2.9215966230057733, "grad_norm": 0.34028342733212946, "learning_rate": 2.0791534197070472e-07, "loss": 2.7877, "step": 47064 }, { "epoch": 2.921658700105531, "grad_norm": 0.32289156380366224, "learning_rate": 2.0758645978363744e-07, "loss": 2.6659, "step": 47065 }, { "epoch": 2.921720777205289, "grad_norm": 0.312105895919411, "learning_rate": 2.072578373756795e-07, "loss": 2.6656, "step": 47066 }, { "epoch": 2.921782854305047, "grad_norm": 0.34514686314402004, "learning_rate": 2.069294747485462e-07, "loss": 2.6544, "step": 47067 }, { "epoch": 2.921844931404805, "grad_norm": 0.3310660413747459, "learning_rate": 2.066013719039528e-07, "loss": 2.7419, "step": 47068 }, { "epoch": 2.9219070085045624, "grad_norm": 0.32248967791827754, "learning_rate": 2.0627352884360352e-07, "loss": 2.6969, "step": 47069 }, { "epoch": 2.921969085604321, "grad_norm": 0.3261100344939786, "learning_rate": 2.0594594556921365e-07, "loss": 2.6694, "step": 47070 }, { "epoch": 2.9220311627040783, "grad_norm": 0.32827937944942376, "learning_rate": 2.0561862208249295e-07, "loss": 2.6288, "step": 47071 }, { "epoch": 2.9220932398038366, "grad_norm": 0.3344334934445889, "learning_rate": 2.0529155838514558e-07, "loss": 2.682, "step": 47072 }, { "epoch": 2.922155316903594, "grad_norm": 0.3565031866878169, "learning_rate": 2.0496475447888132e-07, "loss": 2.6775, "step": 47073 }, { "epoch": 2.922217394003352, "grad_norm": 0.33710485391248346, "learning_rate": 2.0463821036540433e-07, "loss": 2.6733, "step": 47074 }, { "epoch": 2.92227947110311, "grad_norm": 0.3180373616905288, "learning_rate": 2.043119260464188e-07, "loss": 2.7353, "step": 47075 }, { "epoch": 2.922341548202868, "grad_norm": 0.3224531726144869, "learning_rate": 2.0398590152362896e-07, "loss": 2.6191, "step": 47076 }, { "epoch": 2.9224036253026258, "grad_norm": 0.32055380501113967, "learning_rate": 2.0366013679872785e-07, "loss": 2.6683, "step": 47077 }, { "epoch": 2.9224657024023837, "grad_norm": 0.3523991695557724, "learning_rate": 2.0333463187342528e-07, "loss": 2.6804, "step": 47078 }, { "epoch": 2.9225277795021416, "grad_norm": 0.32301535815892823, "learning_rate": 2.0300938674941427e-07, "loss": 2.6735, "step": 47079 }, { "epoch": 2.9225898566018995, "grad_norm": 0.330703143600366, "learning_rate": 2.0268440142838796e-07, "loss": 2.6885, "step": 47080 }, { "epoch": 2.9226519337016574, "grad_norm": 0.3494719549572954, "learning_rate": 2.0235967591204496e-07, "loss": 2.717, "step": 47081 }, { "epoch": 2.9227140108014154, "grad_norm": 0.32098460047762734, "learning_rate": 2.0203521020208395e-07, "loss": 2.5781, "step": 47082 }, { "epoch": 2.9227760879011733, "grad_norm": 0.34128961251105266, "learning_rate": 2.017110043001924e-07, "loss": 2.5766, "step": 47083 }, { "epoch": 2.922838165000931, "grad_norm": 0.31467599256778256, "learning_rate": 2.0138705820806348e-07, "loss": 2.5885, "step": 47084 }, { "epoch": 2.922900242100689, "grad_norm": 0.3330854509971254, "learning_rate": 2.0106337192739022e-07, "loss": 2.604, "step": 47085 }, { "epoch": 2.922962319200447, "grad_norm": 0.3269535566028586, "learning_rate": 2.0073994545985463e-07, "loss": 2.6444, "step": 47086 }, { "epoch": 2.923024396300205, "grad_norm": 0.32809800082344376, "learning_rate": 2.0041677880714982e-07, "loss": 2.7105, "step": 47087 }, { "epoch": 2.923086473399963, "grad_norm": 0.33055928390245337, "learning_rate": 2.0009387197095776e-07, "loss": 2.6076, "step": 47088 }, { "epoch": 2.9231485504997208, "grad_norm": 0.3641779098622842, "learning_rate": 1.997712249529715e-07, "loss": 2.6408, "step": 47089 }, { "epoch": 2.9232106275994787, "grad_norm": 0.3260041222370208, "learning_rate": 1.9944883775486202e-07, "loss": 2.6768, "step": 47090 }, { "epoch": 2.9232727046992366, "grad_norm": 0.3215987733154483, "learning_rate": 1.9912671037831677e-07, "loss": 2.6751, "step": 47091 }, { "epoch": 2.923334781798994, "grad_norm": 0.32614972844575263, "learning_rate": 1.9880484282502332e-07, "loss": 2.6776, "step": 47092 }, { "epoch": 2.9233968588987524, "grad_norm": 0.33091497708440615, "learning_rate": 1.9848323509665257e-07, "loss": 2.6285, "step": 47093 }, { "epoch": 2.92345893599851, "grad_norm": 0.3291136629047775, "learning_rate": 1.9816188719488093e-07, "loss": 2.6412, "step": 47094 }, { "epoch": 2.9235210130982683, "grad_norm": 0.33632913206084825, "learning_rate": 1.9784079912139043e-07, "loss": 2.6889, "step": 47095 }, { "epoch": 2.9235830901980258, "grad_norm": 0.3321294751452649, "learning_rate": 1.975199708778519e-07, "loss": 2.7187, "step": 47096 }, { "epoch": 2.923645167297784, "grad_norm": 0.33719927008010786, "learning_rate": 1.9719940246594737e-07, "loss": 2.7161, "step": 47097 }, { "epoch": 2.9237072443975416, "grad_norm": 0.314849142118336, "learning_rate": 1.9687909388734215e-07, "loss": 2.6584, "step": 47098 }, { "epoch": 2.9237693214972995, "grad_norm": 0.33458230107925313, "learning_rate": 1.965590451437016e-07, "loss": 2.6706, "step": 47099 }, { "epoch": 2.9238313985970574, "grad_norm": 0.3368618468581374, "learning_rate": 1.962392562367077e-07, "loss": 2.5756, "step": 47100 }, { "epoch": 2.9238934756968153, "grad_norm": 0.3350536562879087, "learning_rate": 1.9591972716802575e-07, "loss": 2.6545, "step": 47101 }, { "epoch": 2.9239555527965733, "grad_norm": 0.33318588104885144, "learning_rate": 1.9560045793932113e-07, "loss": 2.6992, "step": 47102 }, { "epoch": 2.924017629896331, "grad_norm": 0.3337963534251404, "learning_rate": 1.9528144855225915e-07, "loss": 2.6361, "step": 47103 }, { "epoch": 2.924079706996089, "grad_norm": 0.3475577182656924, "learning_rate": 1.9496269900849962e-07, "loss": 2.7283, "step": 47104 }, { "epoch": 2.924141784095847, "grad_norm": 0.33358177101014946, "learning_rate": 1.9464420930971338e-07, "loss": 2.5989, "step": 47105 }, { "epoch": 2.924203861195605, "grad_norm": 0.35374297950571887, "learning_rate": 1.9432597945756026e-07, "loss": 2.6939, "step": 47106 }, { "epoch": 2.924265938295363, "grad_norm": 0.3599896357951765, "learning_rate": 1.9400800945370002e-07, "loss": 2.7154, "step": 47107 }, { "epoch": 2.9243280153951208, "grad_norm": 0.3379088380348119, "learning_rate": 1.9369029929978689e-07, "loss": 2.6909, "step": 47108 }, { "epoch": 2.9243900924948787, "grad_norm": 0.3395996642232668, "learning_rate": 1.933728489974862e-07, "loss": 2.5935, "step": 47109 }, { "epoch": 2.9244521695946366, "grad_norm": 0.3245496210583837, "learning_rate": 1.9305565854844664e-07, "loss": 2.6191, "step": 47110 }, { "epoch": 2.9245142466943945, "grad_norm": 0.3264082663050599, "learning_rate": 1.9273872795432802e-07, "loss": 2.7226, "step": 47111 }, { "epoch": 2.9245763237941524, "grad_norm": 0.3413660876920658, "learning_rate": 1.9242205721678453e-07, "loss": 2.6325, "step": 47112 }, { "epoch": 2.9246384008939104, "grad_norm": 0.3203047542100686, "learning_rate": 1.9210564633746486e-07, "loss": 2.8288, "step": 47113 }, { "epoch": 2.9247004779936683, "grad_norm": 0.3125711121710406, "learning_rate": 1.9178949531802325e-07, "loss": 2.5857, "step": 47114 }, { "epoch": 2.924762555093426, "grad_norm": 0.328845561266055, "learning_rate": 1.9147360416010284e-07, "loss": 2.6371, "step": 47115 }, { "epoch": 2.924824632193184, "grad_norm": 0.3413825330199931, "learning_rate": 1.911579728653634e-07, "loss": 2.6002, "step": 47116 }, { "epoch": 2.9248867092929416, "grad_norm": 0.35254299341266304, "learning_rate": 1.9084260143543697e-07, "loss": 2.6064, "step": 47117 }, { "epoch": 2.9249487863927, "grad_norm": 0.3553332605059181, "learning_rate": 1.9052748987197777e-07, "loss": 2.7723, "step": 47118 }, { "epoch": 2.9250108634924574, "grad_norm": 0.34095794172395416, "learning_rate": 1.9021263817663447e-07, "loss": 2.6307, "step": 47119 }, { "epoch": 2.925072940592216, "grad_norm": 0.3214558339572901, "learning_rate": 1.8989804635103913e-07, "loss": 2.6897, "step": 47120 }, { "epoch": 2.9251350176919733, "grad_norm": 0.31641933037364467, "learning_rate": 1.8958371439683486e-07, "loss": 2.6752, "step": 47121 }, { "epoch": 2.925197094791731, "grad_norm": 0.33860194574355823, "learning_rate": 1.892696423156648e-07, "loss": 2.6128, "step": 47122 }, { "epoch": 2.925259171891489, "grad_norm": 0.34219214336084075, "learning_rate": 1.8895583010917207e-07, "loss": 2.6385, "step": 47123 }, { "epoch": 2.925321248991247, "grad_norm": 0.33058032368560175, "learning_rate": 1.8864227777898314e-07, "loss": 2.6958, "step": 47124 }, { "epoch": 2.925383326091005, "grad_norm": 0.3387139493333002, "learning_rate": 1.8832898532674115e-07, "loss": 2.6714, "step": 47125 }, { "epoch": 2.925445403190763, "grad_norm": 0.34756469592584194, "learning_rate": 1.8801595275408367e-07, "loss": 2.6909, "step": 47126 }, { "epoch": 2.9255074802905208, "grad_norm": 0.33382150794786536, "learning_rate": 1.8770318006263167e-07, "loss": 2.6119, "step": 47127 }, { "epoch": 2.9255695573902787, "grad_norm": 0.34403455400747174, "learning_rate": 1.873906672540282e-07, "loss": 2.6446, "step": 47128 }, { "epoch": 2.9256316344900366, "grad_norm": 0.3351091774204298, "learning_rate": 1.8707841432989982e-07, "loss": 2.6293, "step": 47129 }, { "epoch": 2.9256937115897945, "grad_norm": 0.3493806101189925, "learning_rate": 1.8676642129187849e-07, "loss": 2.571, "step": 47130 }, { "epoch": 2.9257557886895524, "grad_norm": 0.3576125589984772, "learning_rate": 1.8645468814158517e-07, "loss": 2.6965, "step": 47131 }, { "epoch": 2.9258178657893104, "grad_norm": 0.32959743914859313, "learning_rate": 1.861432148806519e-07, "loss": 2.6202, "step": 47132 }, { "epoch": 2.9258799428890683, "grad_norm": 0.3368113374978477, "learning_rate": 1.8583200151070512e-07, "loss": 2.622, "step": 47133 }, { "epoch": 2.925942019988826, "grad_norm": 0.321509543600565, "learning_rate": 1.8552104803336023e-07, "loss": 2.6159, "step": 47134 }, { "epoch": 2.926004097088584, "grad_norm": 0.324900542226417, "learning_rate": 1.8521035445024925e-07, "loss": 2.6439, "step": 47135 }, { "epoch": 2.926066174188342, "grad_norm": 0.34242657013515254, "learning_rate": 1.8489992076298755e-07, "loss": 2.7018, "step": 47136 }, { "epoch": 2.9261282512881, "grad_norm": 0.3340988743697429, "learning_rate": 1.845897469731961e-07, "loss": 2.6458, "step": 47137 }, { "epoch": 2.926190328387858, "grad_norm": 0.3367964055379499, "learning_rate": 1.842798330824902e-07, "loss": 2.6644, "step": 47138 }, { "epoch": 2.926252405487616, "grad_norm": 0.34180063518090714, "learning_rate": 1.8397017909249635e-07, "loss": 2.617, "step": 47139 }, { "epoch": 2.9263144825873733, "grad_norm": 0.31200441163868786, "learning_rate": 1.8366078500481886e-07, "loss": 2.5574, "step": 47140 }, { "epoch": 2.9263765596871316, "grad_norm": 0.33751699705062677, "learning_rate": 1.833516508210731e-07, "loss": 2.6373, "step": 47141 }, { "epoch": 2.926438636786889, "grad_norm": 0.33052647629801934, "learning_rate": 1.8304277654287995e-07, "loss": 2.6377, "step": 47142 }, { "epoch": 2.9265007138866475, "grad_norm": 0.3336079467146833, "learning_rate": 1.827341621718437e-07, "loss": 2.6311, "step": 47143 }, { "epoch": 2.926562790986405, "grad_norm": 0.3426404624429582, "learning_rate": 1.8242580770957973e-07, "loss": 2.7017, "step": 47144 }, { "epoch": 2.9266248680861633, "grad_norm": 0.3340593331974614, "learning_rate": 1.8211771315769232e-07, "loss": 2.661, "step": 47145 }, { "epoch": 2.9266869451859208, "grad_norm": 0.3562715289654499, "learning_rate": 1.8180987851779129e-07, "loss": 2.7343, "step": 47146 }, { "epoch": 2.9267490222856787, "grad_norm": 0.33190971843062655, "learning_rate": 1.8150230379148092e-07, "loss": 2.6755, "step": 47147 }, { "epoch": 2.9268110993854366, "grad_norm": 0.3404033810491052, "learning_rate": 1.8119498898036547e-07, "loss": 2.5671, "step": 47148 }, { "epoch": 2.9268731764851945, "grad_norm": 0.3280866464270431, "learning_rate": 1.808879340860492e-07, "loss": 2.642, "step": 47149 }, { "epoch": 2.9269352535849524, "grad_norm": 0.3221359570045637, "learning_rate": 1.8058113911014197e-07, "loss": 2.6402, "step": 47150 }, { "epoch": 2.9269973306847104, "grad_norm": 0.32576202866455867, "learning_rate": 1.8027460405423135e-07, "loss": 2.6041, "step": 47151 }, { "epoch": 2.9270594077844683, "grad_norm": 0.3336503215412092, "learning_rate": 1.7996832891992165e-07, "loss": 2.7674, "step": 47152 }, { "epoch": 2.927121484884226, "grad_norm": 0.3250529117794734, "learning_rate": 1.796623137088116e-07, "loss": 2.6911, "step": 47153 }, { "epoch": 2.927183561983984, "grad_norm": 0.3094699107150938, "learning_rate": 1.7935655842249987e-07, "loss": 2.6329, "step": 47154 }, { "epoch": 2.927245639083742, "grad_norm": 0.3311657339553932, "learning_rate": 1.7905106306257413e-07, "loss": 2.6888, "step": 47155 }, { "epoch": 2.9273077161835, "grad_norm": 0.32286351545837966, "learning_rate": 1.7874582763063862e-07, "loss": 2.6786, "step": 47156 }, { "epoch": 2.927369793283258, "grad_norm": 0.3447547231092012, "learning_rate": 1.78440852128281e-07, "loss": 2.705, "step": 47157 }, { "epoch": 2.9274318703830158, "grad_norm": 0.3194461480673452, "learning_rate": 1.7813613655708882e-07, "loss": 2.598, "step": 47158 }, { "epoch": 2.9274939474827737, "grad_norm": 0.3449913283308684, "learning_rate": 1.7783168091865533e-07, "loss": 2.6613, "step": 47159 }, { "epoch": 2.9275560245825316, "grad_norm": 0.33356228307038277, "learning_rate": 1.7752748521457363e-07, "loss": 2.7277, "step": 47160 }, { "epoch": 2.9276181016822895, "grad_norm": 0.3387286977826678, "learning_rate": 1.7722354944642028e-07, "loss": 2.6559, "step": 47161 }, { "epoch": 2.9276801787820474, "grad_norm": 0.3337048593454526, "learning_rate": 1.7691987361578844e-07, "loss": 2.601, "step": 47162 }, { "epoch": 2.9277422558818054, "grad_norm": 0.31499672337371915, "learning_rate": 1.7661645772426017e-07, "loss": 2.7242, "step": 47163 }, { "epoch": 2.9278043329815633, "grad_norm": 0.3220295359173966, "learning_rate": 1.763133017734231e-07, "loss": 2.5548, "step": 47164 }, { "epoch": 2.9278664100813208, "grad_norm": 0.3225716034185408, "learning_rate": 1.7601040576484819e-07, "loss": 2.6448, "step": 47165 }, { "epoch": 2.927928487181079, "grad_norm": 0.31775331439338816, "learning_rate": 1.7570776970012303e-07, "loss": 2.6179, "step": 47166 }, { "epoch": 2.9279905642808366, "grad_norm": 0.32091250861830634, "learning_rate": 1.7540539358082976e-07, "loss": 2.6233, "step": 47167 }, { "epoch": 2.928052641380595, "grad_norm": 0.3345090748806116, "learning_rate": 1.7510327740853927e-07, "loss": 2.7369, "step": 47168 }, { "epoch": 2.9281147184803524, "grad_norm": 0.33122822761958376, "learning_rate": 1.7480142118482812e-07, "loss": 2.6815, "step": 47169 }, { "epoch": 2.9281767955801103, "grad_norm": 0.35545686555215894, "learning_rate": 1.7449982491127836e-07, "loss": 2.6211, "step": 47170 }, { "epoch": 2.9282388726798683, "grad_norm": 0.318799259082084, "learning_rate": 1.7419848858945542e-07, "loss": 2.6499, "step": 47171 }, { "epoch": 2.928300949779626, "grad_norm": 0.3462773849833433, "learning_rate": 1.738974122209358e-07, "loss": 2.7453, "step": 47172 }, { "epoch": 2.928363026879384, "grad_norm": 0.3378322969929502, "learning_rate": 1.7359659580729048e-07, "loss": 2.6375, "step": 47173 }, { "epoch": 2.928425103979142, "grad_norm": 0.3222436030406243, "learning_rate": 1.7329603935008486e-07, "loss": 2.6474, "step": 47174 }, { "epoch": 2.9284871810789, "grad_norm": 0.327612578956699, "learning_rate": 1.7299574285088992e-07, "loss": 2.6879, "step": 47175 }, { "epoch": 2.928549258178658, "grad_norm": 0.34688222637038074, "learning_rate": 1.7269570631127663e-07, "loss": 2.5972, "step": 47176 }, { "epoch": 2.9286113352784158, "grad_norm": 0.3303877715508257, "learning_rate": 1.723959297327993e-07, "loss": 2.6215, "step": 47177 }, { "epoch": 2.9286734123781737, "grad_norm": 0.3336713897087053, "learning_rate": 1.720964131170344e-07, "loss": 2.7109, "step": 47178 }, { "epoch": 2.9287354894779316, "grad_norm": 0.33931269753507104, "learning_rate": 1.717971564655363e-07, "loss": 2.7244, "step": 47179 }, { "epoch": 2.9287975665776895, "grad_norm": 0.34632899854994326, "learning_rate": 1.7149815977986484e-07, "loss": 2.728, "step": 47180 }, { "epoch": 2.9288596436774474, "grad_norm": 0.3130993579027603, "learning_rate": 1.7119942306158543e-07, "loss": 2.668, "step": 47181 }, { "epoch": 2.9289217207772054, "grad_norm": 0.3269128776945436, "learning_rate": 1.7090094631225795e-07, "loss": 2.6318, "step": 47182 }, { "epoch": 2.9289837978769633, "grad_norm": 0.34086061037212634, "learning_rate": 1.706027295334367e-07, "loss": 2.6297, "step": 47183 }, { "epoch": 2.929045874976721, "grad_norm": 0.3298321998485797, "learning_rate": 1.70304772726676e-07, "loss": 2.7052, "step": 47184 }, { "epoch": 2.929107952076479, "grad_norm": 0.32799509292513995, "learning_rate": 1.7000707589353016e-07, "loss": 2.7139, "step": 47185 }, { "epoch": 2.929170029176237, "grad_norm": 0.33721938741689295, "learning_rate": 1.6970963903555348e-07, "loss": 2.6464, "step": 47186 }, { "epoch": 2.929232106275995, "grad_norm": 0.3648104323667902, "learning_rate": 1.6941246215430584e-07, "loss": 2.6652, "step": 47187 }, { "epoch": 2.9292941833757524, "grad_norm": 0.3445837465285315, "learning_rate": 1.691155452513249e-07, "loss": 2.6771, "step": 47188 }, { "epoch": 2.929356260475511, "grad_norm": 0.3456270382012254, "learning_rate": 1.6881888832816496e-07, "loss": 2.7439, "step": 47189 }, { "epoch": 2.9294183375752683, "grad_norm": 0.34425103995845513, "learning_rate": 1.6852249138636922e-07, "loss": 2.6353, "step": 47190 }, { "epoch": 2.9294804146750266, "grad_norm": 0.34462389619406286, "learning_rate": 1.6822635442749756e-07, "loss": 2.6282, "step": 47191 }, { "epoch": 2.929542491774784, "grad_norm": 0.33904811195750184, "learning_rate": 1.679304774530821e-07, "loss": 2.7201, "step": 47192 }, { "epoch": 2.9296045688745425, "grad_norm": 0.3393736800103133, "learning_rate": 1.6763486046467158e-07, "loss": 2.6418, "step": 47193 }, { "epoch": 2.9296666459743, "grad_norm": 0.3352819633799848, "learning_rate": 1.6733950346380368e-07, "loss": 2.7246, "step": 47194 }, { "epoch": 2.929728723074058, "grad_norm": 0.3454131186943075, "learning_rate": 1.6704440645202158e-07, "loss": 2.6168, "step": 47195 }, { "epoch": 2.9297908001738158, "grad_norm": 0.2987320689482415, "learning_rate": 1.6674956943087405e-07, "loss": 2.6339, "step": 47196 }, { "epoch": 2.9298528772735737, "grad_norm": 0.3174335380512747, "learning_rate": 1.6645499240188766e-07, "loss": 2.6483, "step": 47197 }, { "epoch": 2.9299149543733316, "grad_norm": 0.32456552964516705, "learning_rate": 1.6616067536660008e-07, "loss": 2.6872, "step": 47198 }, { "epoch": 2.9299770314730895, "grad_norm": 0.33086750061073666, "learning_rate": 1.6586661832654894e-07, "loss": 2.6594, "step": 47199 }, { "epoch": 2.9300391085728474, "grad_norm": 0.33162361254152, "learning_rate": 1.655728212832719e-07, "loss": 2.782, "step": 47200 }, { "epoch": 2.9301011856726054, "grad_norm": 0.3359300479505947, "learning_rate": 1.6527928423830108e-07, "loss": 2.6758, "step": 47201 }, { "epoch": 2.9301632627723633, "grad_norm": 0.34382597127117237, "learning_rate": 1.6498600719316858e-07, "loss": 2.6314, "step": 47202 }, { "epoch": 2.930225339872121, "grad_norm": 0.3185333291570395, "learning_rate": 1.6469299014939543e-07, "loss": 2.5637, "step": 47203 }, { "epoch": 2.930287416971879, "grad_norm": 0.3365433463983972, "learning_rate": 1.6440023310851926e-07, "loss": 2.69, "step": 47204 }, { "epoch": 2.930349494071637, "grad_norm": 0.3155645076455895, "learning_rate": 1.6410773607206665e-07, "loss": 2.7073, "step": 47205 }, { "epoch": 2.930411571171395, "grad_norm": 0.3317144565600904, "learning_rate": 1.6381549904156412e-07, "loss": 2.6079, "step": 47206 }, { "epoch": 2.930473648271153, "grad_norm": 0.33977576924168823, "learning_rate": 1.6352352201852717e-07, "loss": 2.6484, "step": 47207 }, { "epoch": 2.930535725370911, "grad_norm": 0.33156576626002066, "learning_rate": 1.6323180500449343e-07, "loss": 2.6462, "step": 47208 }, { "epoch": 2.9305978024706687, "grad_norm": 0.32625433741715376, "learning_rate": 1.629403480009728e-07, "loss": 2.587, "step": 47209 }, { "epoch": 2.9306598795704266, "grad_norm": 0.3481584201091466, "learning_rate": 1.6264915100949186e-07, "loss": 2.6722, "step": 47210 }, { "epoch": 2.930721956670184, "grad_norm": 0.3304638837451536, "learning_rate": 1.6235821403157158e-07, "loss": 2.5668, "step": 47211 }, { "epoch": 2.9307840337699425, "grad_norm": 0.3526102348192403, "learning_rate": 1.620675370687219e-07, "loss": 2.685, "step": 47212 }, { "epoch": 2.9308461108697, "grad_norm": 0.33554163607858495, "learning_rate": 1.6177712012246936e-07, "loss": 2.7988, "step": 47213 }, { "epoch": 2.9309081879694583, "grad_norm": 0.34410240792490965, "learning_rate": 1.6148696319432388e-07, "loss": 2.7299, "step": 47214 }, { "epoch": 2.9309702650692158, "grad_norm": 0.3229844024844211, "learning_rate": 1.6119706628579533e-07, "loss": 2.6155, "step": 47215 }, { "epoch": 2.931032342168974, "grad_norm": 0.31944817719950774, "learning_rate": 1.6090742939840475e-07, "loss": 2.6163, "step": 47216 }, { "epoch": 2.9310944192687316, "grad_norm": 0.3412832924738316, "learning_rate": 1.6061805253365648e-07, "loss": 2.6265, "step": 47217 }, { "epoch": 2.9311564963684895, "grad_norm": 0.3305938799441625, "learning_rate": 1.603289356930604e-07, "loss": 2.6499, "step": 47218 }, { "epoch": 2.9312185734682474, "grad_norm": 0.31652826146641255, "learning_rate": 1.60040078878132e-07, "loss": 2.5489, "step": 47219 }, { "epoch": 2.9312806505680054, "grad_norm": 0.32283380301496667, "learning_rate": 1.597514820903756e-07, "loss": 2.6682, "step": 47220 }, { "epoch": 2.9313427276677633, "grad_norm": 0.3435130169413544, "learning_rate": 1.5946314533129004e-07, "loss": 2.7306, "step": 47221 }, { "epoch": 2.931404804767521, "grad_norm": 0.32681194479510883, "learning_rate": 1.591750686023852e-07, "loss": 2.6204, "step": 47222 }, { "epoch": 2.931466881867279, "grad_norm": 0.3268650478649596, "learning_rate": 1.588872519051654e-07, "loss": 2.6534, "step": 47223 }, { "epoch": 2.931528958967037, "grad_norm": 0.34351489306665367, "learning_rate": 1.5859969524112395e-07, "loss": 2.6106, "step": 47224 }, { "epoch": 2.931591036066795, "grad_norm": 0.3534171035642171, "learning_rate": 1.5831239861177628e-07, "loss": 2.7167, "step": 47225 }, { "epoch": 2.931653113166553, "grad_norm": 0.3276777284428707, "learning_rate": 1.5802536201860453e-07, "loss": 2.6636, "step": 47226 }, { "epoch": 2.9317151902663108, "grad_norm": 0.33484652152719, "learning_rate": 1.5773858546311858e-07, "loss": 2.5861, "step": 47227 }, { "epoch": 2.9317772673660687, "grad_norm": 0.33313629043583953, "learning_rate": 1.5745206894681174e-07, "loss": 2.6931, "step": 47228 }, { "epoch": 2.9318393444658266, "grad_norm": 0.3440411050330776, "learning_rate": 1.5716581247117168e-07, "loss": 2.6731, "step": 47229 }, { "epoch": 2.9319014215655845, "grad_norm": 0.34375282849968786, "learning_rate": 1.5687981603769718e-07, "loss": 2.6532, "step": 47230 }, { "epoch": 2.9319634986653424, "grad_norm": 0.3224915261781771, "learning_rate": 1.5659407964788153e-07, "loss": 2.6587, "step": 47231 }, { "epoch": 2.9320255757651004, "grad_norm": 0.33939823457486906, "learning_rate": 1.5630860330321796e-07, "loss": 2.6636, "step": 47232 }, { "epoch": 2.9320876528648583, "grad_norm": 0.3373763520184365, "learning_rate": 1.5602338700519413e-07, "loss": 2.7107, "step": 47233 }, { "epoch": 2.932149729964616, "grad_norm": 0.33546966312747906, "learning_rate": 1.5573843075529225e-07, "loss": 2.6968, "step": 47234 }, { "epoch": 2.932211807064374, "grad_norm": 0.33614727611998524, "learning_rate": 1.5545373455500002e-07, "loss": 2.5975, "step": 47235 }, { "epoch": 2.9322738841641316, "grad_norm": 0.37502715886633065, "learning_rate": 1.5516929840581063e-07, "loss": 2.5968, "step": 47236 }, { "epoch": 2.93233596126389, "grad_norm": 0.34511017460328225, "learning_rate": 1.5488512230920627e-07, "loss": 2.6682, "step": 47237 }, { "epoch": 2.9323980383636474, "grad_norm": 0.33588212687984687, "learning_rate": 1.5460120626665796e-07, "loss": 2.6443, "step": 47238 }, { "epoch": 2.932460115463406, "grad_norm": 0.3326141850083423, "learning_rate": 1.5431755027966455e-07, "loss": 2.6786, "step": 47239 }, { "epoch": 2.9325221925631633, "grad_norm": 0.3552179394911829, "learning_rate": 1.5403415434969148e-07, "loss": 2.533, "step": 47240 }, { "epoch": 2.932584269662921, "grad_norm": 0.3314457836608595, "learning_rate": 1.5375101847822093e-07, "loss": 2.7001, "step": 47241 }, { "epoch": 2.932646346762679, "grad_norm": 0.31769248526813876, "learning_rate": 1.5346814266673504e-07, "loss": 2.7205, "step": 47242 }, { "epoch": 2.932708423862437, "grad_norm": 0.33120322482026626, "learning_rate": 1.531855269167104e-07, "loss": 2.6139, "step": 47243 }, { "epoch": 2.932770500962195, "grad_norm": 0.34900677283090376, "learning_rate": 1.5290317122961251e-07, "loss": 2.6526, "step": 47244 }, { "epoch": 2.932832578061953, "grad_norm": 0.32888320334688775, "learning_rate": 1.5262107560691796e-07, "loss": 2.6274, "step": 47245 }, { "epoch": 2.9328946551617108, "grad_norm": 0.3339712017651457, "learning_rate": 1.5233924005010337e-07, "loss": 2.7252, "step": 47246 }, { "epoch": 2.9329567322614687, "grad_norm": 0.3427844684271332, "learning_rate": 1.520576645606342e-07, "loss": 2.6238, "step": 47247 }, { "epoch": 2.9330188093612266, "grad_norm": 0.3360600469761138, "learning_rate": 1.517763491399815e-07, "loss": 2.6906, "step": 47248 }, { "epoch": 2.9330808864609845, "grad_norm": 0.3353091173632419, "learning_rate": 1.5149529378961637e-07, "loss": 2.621, "step": 47249 }, { "epoch": 2.9331429635607424, "grad_norm": 0.3331254991667194, "learning_rate": 1.5121449851099867e-07, "loss": 2.6724, "step": 47250 }, { "epoch": 2.9332050406605004, "grad_norm": 0.32154023576256335, "learning_rate": 1.5093396330559394e-07, "loss": 2.7549, "step": 47251 }, { "epoch": 2.9332671177602583, "grad_norm": 0.3477093302463132, "learning_rate": 1.5065368817487323e-07, "loss": 2.7008, "step": 47252 }, { "epoch": 2.933329194860016, "grad_norm": 0.34628721775654403, "learning_rate": 1.5037367312028537e-07, "loss": 2.8194, "step": 47253 }, { "epoch": 2.933391271959774, "grad_norm": 0.31467941387436144, "learning_rate": 1.5009391814330698e-07, "loss": 2.691, "step": 47254 }, { "epoch": 2.933453349059532, "grad_norm": 0.3319886011894288, "learning_rate": 1.4981442324538687e-07, "loss": 2.6846, "step": 47255 }, { "epoch": 2.93351542615929, "grad_norm": 0.3318929047433277, "learning_rate": 1.4953518842799052e-07, "loss": 2.6759, "step": 47256 }, { "epoch": 2.933577503259048, "grad_norm": 0.3275227241785874, "learning_rate": 1.4925621369256127e-07, "loss": 2.7429, "step": 47257 }, { "epoch": 2.933639580358806, "grad_norm": 0.3254593015371571, "learning_rate": 1.4897749904057567e-07, "loss": 2.676, "step": 47258 }, { "epoch": 2.9337016574585633, "grad_norm": 0.34524997818357717, "learning_rate": 1.486990444734715e-07, "loss": 2.7468, "step": 47259 }, { "epoch": 2.9337637345583216, "grad_norm": 0.33431023879685184, "learning_rate": 1.4842084999270312e-07, "loss": 2.665, "step": 47260 }, { "epoch": 2.933825811658079, "grad_norm": 0.35113844197876715, "learning_rate": 1.4814291559972494e-07, "loss": 2.6551, "step": 47261 }, { "epoch": 2.9338878887578375, "grad_norm": 0.3592289337533247, "learning_rate": 1.4786524129599133e-07, "loss": 2.642, "step": 47262 }, { "epoch": 2.933949965857595, "grad_norm": 0.31471770866054344, "learning_rate": 1.4758782708294561e-07, "loss": 2.6489, "step": 47263 }, { "epoch": 2.9340120429573533, "grad_norm": 0.3191995942114101, "learning_rate": 1.473106729620366e-07, "loss": 2.7355, "step": 47264 }, { "epoch": 2.9340741200571108, "grad_norm": 0.3312517440498072, "learning_rate": 1.470337789347076e-07, "loss": 2.658, "step": 47265 }, { "epoch": 2.9341361971568687, "grad_norm": 0.32400188316869355, "learning_rate": 1.4675714500240744e-07, "loss": 2.6682, "step": 47266 }, { "epoch": 2.9341982742566266, "grad_norm": 0.32637953857573926, "learning_rate": 1.464807711665739e-07, "loss": 2.7047, "step": 47267 }, { "epoch": 2.9342603513563845, "grad_norm": 0.33658754310484384, "learning_rate": 1.4620465742865574e-07, "loss": 2.7436, "step": 47268 }, { "epoch": 2.9343224284561424, "grad_norm": 0.33648217203390285, "learning_rate": 1.4592880379009077e-07, "loss": 2.5903, "step": 47269 }, { "epoch": 2.9343845055559004, "grad_norm": 0.3222019132577753, "learning_rate": 1.456532102523167e-07, "loss": 2.6236, "step": 47270 }, { "epoch": 2.9344465826556583, "grad_norm": 0.3331190689696895, "learning_rate": 1.4537787681677684e-07, "loss": 2.6535, "step": 47271 }, { "epoch": 2.934508659755416, "grad_norm": 0.3296923329690621, "learning_rate": 1.451028034848978e-07, "loss": 2.5815, "step": 47272 }, { "epoch": 2.934570736855174, "grad_norm": 0.32023325381765577, "learning_rate": 1.4482799025812288e-07, "loss": 2.6796, "step": 47273 }, { "epoch": 2.934632813954932, "grad_norm": 0.3410668669979059, "learning_rate": 1.4455343713787873e-07, "loss": 2.6425, "step": 47274 }, { "epoch": 2.93469489105469, "grad_norm": 0.3206899442763585, "learning_rate": 1.4427914412560862e-07, "loss": 2.7439, "step": 47275 }, { "epoch": 2.934756968154448, "grad_norm": 0.3369322067019024, "learning_rate": 1.4400511122273364e-07, "loss": 2.7017, "step": 47276 }, { "epoch": 2.934819045254206, "grad_norm": 0.3360156145874107, "learning_rate": 1.4373133843068599e-07, "loss": 2.7058, "step": 47277 }, { "epoch": 2.9348811223539637, "grad_norm": 0.3473434433997925, "learning_rate": 1.4345782575089784e-07, "loss": 2.753, "step": 47278 }, { "epoch": 2.9349431994537216, "grad_norm": 0.3316032006632982, "learning_rate": 1.4318457318479028e-07, "loss": 2.6922, "step": 47279 }, { "epoch": 2.9350052765534795, "grad_norm": 0.3360369557269028, "learning_rate": 1.429115807337955e-07, "loss": 2.6743, "step": 47280 }, { "epoch": 2.9350673536532375, "grad_norm": 0.331004113579454, "learning_rate": 1.4263884839932907e-07, "loss": 2.6571, "step": 47281 }, { "epoch": 2.9351294307529954, "grad_norm": 0.3329922513038777, "learning_rate": 1.4236637618282312e-07, "loss": 2.7412, "step": 47282 }, { "epoch": 2.9351915078527533, "grad_norm": 0.3622683895055785, "learning_rate": 1.420941640856932e-07, "loss": 2.7475, "step": 47283 }, { "epoch": 2.9352535849525108, "grad_norm": 0.33987041894912656, "learning_rate": 1.418222121093604e-07, "loss": 2.6769, "step": 47284 }, { "epoch": 2.935315662052269, "grad_norm": 0.33929923710010723, "learning_rate": 1.415505202552514e-07, "loss": 2.6812, "step": 47285 }, { "epoch": 2.9353777391520266, "grad_norm": 0.3327902813331717, "learning_rate": 1.4127908852477056e-07, "loss": 2.6901, "step": 47286 }, { "epoch": 2.935439816251785, "grad_norm": 0.34030026433997773, "learning_rate": 1.4100791691933902e-07, "loss": 2.5923, "step": 47287 }, { "epoch": 2.9355018933515424, "grad_norm": 0.32672169879693636, "learning_rate": 1.4073700544037782e-07, "loss": 2.6511, "step": 47288 }, { "epoch": 2.9355639704513004, "grad_norm": 0.3219110908459091, "learning_rate": 1.4046635408929143e-07, "loss": 2.6087, "step": 47289 }, { "epoch": 2.9356260475510583, "grad_norm": 0.32666277920022, "learning_rate": 1.4019596286750093e-07, "loss": 2.7201, "step": 47290 }, { "epoch": 2.935688124650816, "grad_norm": 0.3292511732939349, "learning_rate": 1.3992583177640517e-07, "loss": 2.6463, "step": 47291 }, { "epoch": 2.935750201750574, "grad_norm": 0.3311554521905818, "learning_rate": 1.3965596081742526e-07, "loss": 2.7151, "step": 47292 }, { "epoch": 2.935812278850332, "grad_norm": 0.3362773684173142, "learning_rate": 1.3938634999196565e-07, "loss": 2.6305, "step": 47293 }, { "epoch": 2.93587435595009, "grad_norm": 0.32050096968297326, "learning_rate": 1.3911699930142518e-07, "loss": 2.5959, "step": 47294 }, { "epoch": 2.935936433049848, "grad_norm": 0.3556163816631247, "learning_rate": 1.3884790874721942e-07, "loss": 2.7197, "step": 47295 }, { "epoch": 2.935998510149606, "grad_norm": 0.3535863689641731, "learning_rate": 1.3857907833074723e-07, "loss": 2.6344, "step": 47296 }, { "epoch": 2.9360605872493637, "grad_norm": 0.33832004580142033, "learning_rate": 1.3831050805341307e-07, "loss": 2.6207, "step": 47297 }, { "epoch": 2.9361226643491216, "grad_norm": 0.3410500896818096, "learning_rate": 1.3804219791661576e-07, "loss": 2.6698, "step": 47298 }, { "epoch": 2.9361847414488795, "grad_norm": 0.34069329985666374, "learning_rate": 1.377741479217598e-07, "loss": 2.6099, "step": 47299 }, { "epoch": 2.9362468185486374, "grad_norm": 0.3255181299733124, "learning_rate": 1.3750635807023848e-07, "loss": 2.7405, "step": 47300 }, { "epoch": 2.9363088956483954, "grad_norm": 0.3262552052115512, "learning_rate": 1.3723882836345624e-07, "loss": 2.6244, "step": 47301 }, { "epoch": 2.9363709727481533, "grad_norm": 0.3276805592525262, "learning_rate": 1.3697155880279533e-07, "loss": 2.6538, "step": 47302 }, { "epoch": 2.936433049847911, "grad_norm": 0.34700906714416485, "learning_rate": 1.3670454938966015e-07, "loss": 2.7316, "step": 47303 }, { "epoch": 2.936495126947669, "grad_norm": 0.33873281095678515, "learning_rate": 1.3643780012544405e-07, "loss": 2.7805, "step": 47304 }, { "epoch": 2.936557204047427, "grad_norm": 0.35761730917027984, "learning_rate": 1.3617131101154035e-07, "loss": 2.6897, "step": 47305 }, { "epoch": 2.936619281147185, "grad_norm": 0.34963539495552837, "learning_rate": 1.3590508204933128e-07, "loss": 2.664, "step": 47306 }, { "epoch": 2.9366813582469424, "grad_norm": 0.3461317110447718, "learning_rate": 1.3563911324021018e-07, "loss": 2.6588, "step": 47307 }, { "epoch": 2.936743435346701, "grad_norm": 0.3444167628122382, "learning_rate": 1.353734045855648e-07, "loss": 2.5921, "step": 47308 }, { "epoch": 2.9368055124464583, "grad_norm": 0.32843386781464406, "learning_rate": 1.3510795608678294e-07, "loss": 2.5819, "step": 47309 }, { "epoch": 2.9368675895462166, "grad_norm": 0.33785811030773005, "learning_rate": 1.348427677452524e-07, "loss": 2.6058, "step": 47310 }, { "epoch": 2.936929666645974, "grad_norm": 0.31454149035305135, "learning_rate": 1.3457783956234427e-07, "loss": 2.584, "step": 47311 }, { "epoch": 2.9369917437457325, "grad_norm": 0.3523938310475888, "learning_rate": 1.343131715394519e-07, "loss": 2.7289, "step": 47312 }, { "epoch": 2.93705382084549, "grad_norm": 0.32563209412474636, "learning_rate": 1.3404876367795193e-07, "loss": 2.5746, "step": 47313 }, { "epoch": 2.937115897945248, "grad_norm": 0.32472318451362375, "learning_rate": 1.3378461597922664e-07, "loss": 2.6495, "step": 47314 }, { "epoch": 2.9371779750450058, "grad_norm": 0.31840986307849567, "learning_rate": 1.3352072844465269e-07, "loss": 2.6424, "step": 47315 }, { "epoch": 2.9372400521447637, "grad_norm": 0.3480083632649474, "learning_rate": 1.3325710107560673e-07, "loss": 2.6185, "step": 47316 }, { "epoch": 2.9373021292445216, "grad_norm": 0.34215143614638577, "learning_rate": 1.3299373387346547e-07, "loss": 2.5512, "step": 47317 }, { "epoch": 2.9373642063442795, "grad_norm": 0.3113468231559507, "learning_rate": 1.3273062683959448e-07, "loss": 2.6579, "step": 47318 }, { "epoch": 2.9374262834440374, "grad_norm": 0.340751258390878, "learning_rate": 1.324677799753815e-07, "loss": 2.6001, "step": 47319 }, { "epoch": 2.9374883605437954, "grad_norm": 0.339933339729809, "learning_rate": 1.3220519328218661e-07, "loss": 2.6561, "step": 47320 }, { "epoch": 2.9375504376435533, "grad_norm": 0.33790362338603946, "learning_rate": 1.319428667613809e-07, "loss": 2.6282, "step": 47321 }, { "epoch": 2.937612514743311, "grad_norm": 0.3308591574676051, "learning_rate": 1.3168080041434106e-07, "loss": 2.6386, "step": 47322 }, { "epoch": 2.937674591843069, "grad_norm": 0.30659875766162786, "learning_rate": 1.3141899424242154e-07, "loss": 2.5294, "step": 47323 }, { "epoch": 2.937736668942827, "grad_norm": 0.3311468350608404, "learning_rate": 1.3115744824699904e-07, "loss": 2.6431, "step": 47324 }, { "epoch": 2.937798746042585, "grad_norm": 0.33661487950984803, "learning_rate": 1.308961624294336e-07, "loss": 2.6349, "step": 47325 }, { "epoch": 2.937860823142343, "grad_norm": 0.3186921796528065, "learning_rate": 1.3063513679109073e-07, "loss": 2.6614, "step": 47326 }, { "epoch": 2.937922900242101, "grad_norm": 0.34086215136785525, "learning_rate": 1.3037437133333052e-07, "loss": 2.7174, "step": 47327 }, { "epoch": 2.9379849773418587, "grad_norm": 0.32768103673608184, "learning_rate": 1.301138660575074e-07, "loss": 2.7357, "step": 47328 }, { "epoch": 2.9380470544416166, "grad_norm": 0.33109879551527005, "learning_rate": 1.2985362096499255e-07, "loss": 2.7558, "step": 47329 }, { "epoch": 2.9381091315413745, "grad_norm": 0.3351526295748707, "learning_rate": 1.2959363605713483e-07, "loss": 2.65, "step": 47330 }, { "epoch": 2.9381712086411325, "grad_norm": 0.3473918745376762, "learning_rate": 1.2933391133529427e-07, "loss": 2.657, "step": 47331 }, { "epoch": 2.93823328574089, "grad_norm": 0.3344956327468299, "learning_rate": 1.2907444680082538e-07, "loss": 2.6266, "step": 47332 }, { "epoch": 2.9382953628406483, "grad_norm": 0.33970844381456855, "learning_rate": 1.288152424550826e-07, "loss": 2.5965, "step": 47333 }, { "epoch": 2.9383574399404058, "grad_norm": 0.32814860259044376, "learning_rate": 1.2855629829941484e-07, "loss": 2.6412, "step": 47334 }, { "epoch": 2.938419517040164, "grad_norm": 0.31395599210372876, "learning_rate": 1.2829761433518216e-07, "loss": 2.7108, "step": 47335 }, { "epoch": 2.9384815941399216, "grad_norm": 0.33507561458359664, "learning_rate": 1.2803919056372237e-07, "loss": 2.6306, "step": 47336 }, { "epoch": 2.9385436712396795, "grad_norm": 0.3418506035128627, "learning_rate": 1.2778102698638994e-07, "loss": 2.6353, "step": 47337 }, { "epoch": 2.9386057483394374, "grad_norm": 0.34627032759107584, "learning_rate": 1.2752312360452823e-07, "loss": 2.6562, "step": 47338 }, { "epoch": 2.9386678254391954, "grad_norm": 0.3470437193703848, "learning_rate": 1.2726548041948615e-07, "loss": 2.686, "step": 47339 }, { "epoch": 2.9387299025389533, "grad_norm": 0.3374822165960925, "learning_rate": 1.2700809743261266e-07, "loss": 2.7317, "step": 47340 }, { "epoch": 2.938791979638711, "grad_norm": 0.3444143832338165, "learning_rate": 1.2675097464524e-07, "loss": 2.6749, "step": 47341 }, { "epoch": 2.938854056738469, "grad_norm": 0.327083376208179, "learning_rate": 1.264941120587171e-07, "loss": 2.7697, "step": 47342 }, { "epoch": 2.938916133838227, "grad_norm": 0.3226018827507575, "learning_rate": 1.262375096743762e-07, "loss": 2.6611, "step": 47343 }, { "epoch": 2.938978210937985, "grad_norm": 0.32875200945454, "learning_rate": 1.2598116749356625e-07, "loss": 2.6678, "step": 47344 }, { "epoch": 2.939040288037743, "grad_norm": 0.33690562583201017, "learning_rate": 1.2572508551761952e-07, "loss": 2.6519, "step": 47345 }, { "epoch": 2.939102365137501, "grad_norm": 0.32231772030620803, "learning_rate": 1.2546926374787382e-07, "loss": 2.6561, "step": 47346 }, { "epoch": 2.9391644422372587, "grad_norm": 0.3516145667621057, "learning_rate": 1.2521370218566142e-07, "loss": 2.6395, "step": 47347 }, { "epoch": 2.9392265193370166, "grad_norm": 0.3365048784749744, "learning_rate": 1.2495840083231458e-07, "loss": 2.6883, "step": 47348 }, { "epoch": 2.9392885964367745, "grad_norm": 0.3266243337652899, "learning_rate": 1.2470335968917113e-07, "loss": 2.6097, "step": 47349 }, { "epoch": 2.9393506735365325, "grad_norm": 0.33222028175799634, "learning_rate": 1.2444857875755224e-07, "loss": 2.6205, "step": 47350 }, { "epoch": 2.9394127506362904, "grad_norm": 0.3189883542379065, "learning_rate": 1.241940580387957e-07, "loss": 2.7913, "step": 47351 }, { "epoch": 2.9394748277360483, "grad_norm": 0.3246008792688127, "learning_rate": 1.2393979753422825e-07, "loss": 2.6772, "step": 47352 }, { "epoch": 2.939536904835806, "grad_norm": 0.3433753269209276, "learning_rate": 1.2368579724517103e-07, "loss": 2.7388, "step": 47353 }, { "epoch": 2.939598981935564, "grad_norm": 0.3295444598564096, "learning_rate": 1.2343205717295636e-07, "loss": 2.6964, "step": 47354 }, { "epoch": 2.9396610590353216, "grad_norm": 0.3508055803828618, "learning_rate": 1.2317857731890536e-07, "loss": 2.677, "step": 47355 }, { "epoch": 2.93972313613508, "grad_norm": 0.3267196464252625, "learning_rate": 1.2292535768433366e-07, "loss": 2.7101, "step": 47356 }, { "epoch": 2.9397852132348374, "grad_norm": 0.3295431711564734, "learning_rate": 1.2267239827057352e-07, "loss": 2.6714, "step": 47357 }, { "epoch": 2.939847290334596, "grad_norm": 0.33460995717057007, "learning_rate": 1.2241969907894059e-07, "loss": 2.6644, "step": 47358 }, { "epoch": 2.9399093674343533, "grad_norm": 0.33693689462447796, "learning_rate": 1.221672601107504e-07, "loss": 2.803, "step": 47359 }, { "epoch": 2.9399714445341116, "grad_norm": 0.31935110260304395, "learning_rate": 1.2191508136732422e-07, "loss": 2.6953, "step": 47360 }, { "epoch": 2.940033521633869, "grad_norm": 0.336202118027623, "learning_rate": 1.2166316284997205e-07, "loss": 2.6124, "step": 47361 }, { "epoch": 2.940095598733627, "grad_norm": 0.3294477142186373, "learning_rate": 1.2141150456000948e-07, "loss": 2.6784, "step": 47362 }, { "epoch": 2.940157675833385, "grad_norm": 0.3403072047106219, "learning_rate": 1.2116010649875774e-07, "loss": 2.7124, "step": 47363 }, { "epoch": 2.940219752933143, "grad_norm": 0.3171162938775013, "learning_rate": 1.2090896866751578e-07, "loss": 2.6241, "step": 47364 }, { "epoch": 2.940281830032901, "grad_norm": 0.3371083128062606, "learning_rate": 1.2065809106760472e-07, "loss": 2.7615, "step": 47365 }, { "epoch": 2.9403439071326587, "grad_norm": 0.3319535557220458, "learning_rate": 1.2040747370032911e-07, "loss": 2.6726, "step": 47366 }, { "epoch": 2.9404059842324166, "grad_norm": 0.3396039862388408, "learning_rate": 1.2015711656699346e-07, "loss": 2.6483, "step": 47367 }, { "epoch": 2.9404680613321745, "grad_norm": 0.33248859771380374, "learning_rate": 1.1990701966890783e-07, "loss": 2.6924, "step": 47368 }, { "epoch": 2.9405301384319324, "grad_norm": 0.3239043331319543, "learning_rate": 1.196571830073767e-07, "loss": 2.6739, "step": 47369 }, { "epoch": 2.9405922155316904, "grad_norm": 0.3124541368208199, "learning_rate": 1.1940760658370465e-07, "loss": 2.6516, "step": 47370 }, { "epoch": 2.9406542926314483, "grad_norm": 0.35005990094808814, "learning_rate": 1.1915829039919058e-07, "loss": 2.6926, "step": 47371 }, { "epoch": 2.940716369731206, "grad_norm": 0.33057131967879533, "learning_rate": 1.1890923445513902e-07, "loss": 2.6711, "step": 47372 }, { "epoch": 2.940778446830964, "grad_norm": 0.33144226567357765, "learning_rate": 1.1866043875284338e-07, "loss": 2.6397, "step": 47373 }, { "epoch": 2.940840523930722, "grad_norm": 0.3374401904905215, "learning_rate": 1.1841190329360819e-07, "loss": 2.6393, "step": 47374 }, { "epoch": 2.94090260103048, "grad_norm": 0.3403324800376177, "learning_rate": 1.1816362807872128e-07, "loss": 2.6814, "step": 47375 }, { "epoch": 2.940964678130238, "grad_norm": 0.3236765867015097, "learning_rate": 1.1791561310949273e-07, "loss": 2.6477, "step": 47376 }, { "epoch": 2.941026755229996, "grad_norm": 0.3322080365935638, "learning_rate": 1.176678583871993e-07, "loss": 2.6178, "step": 47377 }, { "epoch": 2.9410888323297537, "grad_norm": 0.3298241241810954, "learning_rate": 1.1742036391315104e-07, "loss": 2.711, "step": 47378 }, { "epoch": 2.9411509094295116, "grad_norm": 0.33015653141730067, "learning_rate": 1.1717312968861916e-07, "loss": 2.696, "step": 47379 }, { "epoch": 2.941212986529269, "grad_norm": 0.31782015672160735, "learning_rate": 1.1692615571491372e-07, "loss": 2.6935, "step": 47380 }, { "epoch": 2.9412750636290275, "grad_norm": 0.323987303702881, "learning_rate": 1.166794419933115e-07, "loss": 2.7267, "step": 47381 }, { "epoch": 2.941337140728785, "grad_norm": 0.33058321979085786, "learning_rate": 1.1643298852510032e-07, "loss": 2.6734, "step": 47382 }, { "epoch": 2.9413992178285433, "grad_norm": 0.3371523581073444, "learning_rate": 1.1618679531156806e-07, "loss": 2.6011, "step": 47383 }, { "epoch": 2.9414612949283008, "grad_norm": 0.3205243084908924, "learning_rate": 1.1594086235400259e-07, "loss": 2.616, "step": 47384 }, { "epoch": 2.9415233720280587, "grad_norm": 0.33136373627599075, "learning_rate": 1.1569518965368065e-07, "loss": 2.6131, "step": 47385 }, { "epoch": 2.9415854491278166, "grad_norm": 0.3233773677784954, "learning_rate": 1.1544977721188454e-07, "loss": 2.6922, "step": 47386 }, { "epoch": 2.9416475262275745, "grad_norm": 0.3399299424585553, "learning_rate": 1.1520462502990215e-07, "loss": 2.7252, "step": 47387 }, { "epoch": 2.9417096033273324, "grad_norm": 0.3323943564907074, "learning_rate": 1.1495973310900465e-07, "loss": 2.6532, "step": 47388 }, { "epoch": 2.9417716804270904, "grad_norm": 0.3407437257601901, "learning_rate": 1.1471510145046882e-07, "loss": 2.6322, "step": 47389 }, { "epoch": 2.9418337575268483, "grad_norm": 0.33184328929891777, "learning_rate": 1.1447073005558251e-07, "loss": 2.6816, "step": 47390 }, { "epoch": 2.941895834626606, "grad_norm": 0.32109880466612917, "learning_rate": 1.1422661892560583e-07, "loss": 2.7259, "step": 47391 }, { "epoch": 2.941957911726364, "grad_norm": 0.32871326438909804, "learning_rate": 1.1398276806182107e-07, "loss": 2.5116, "step": 47392 }, { "epoch": 2.942019988826122, "grad_norm": 0.3333107250215718, "learning_rate": 1.1373917746549944e-07, "loss": 2.689, "step": 47393 }, { "epoch": 2.94208206592588, "grad_norm": 0.331838574323231, "learning_rate": 1.1349584713790663e-07, "loss": 2.6518, "step": 47394 }, { "epoch": 2.942144143025638, "grad_norm": 0.3209847165358314, "learning_rate": 1.1325277708031934e-07, "loss": 2.7449, "step": 47395 }, { "epoch": 2.942206220125396, "grad_norm": 0.3344910776676083, "learning_rate": 1.1300996729400326e-07, "loss": 2.6253, "step": 47396 }, { "epoch": 2.9422682972251537, "grad_norm": 0.34162155014373125, "learning_rate": 1.1276741778022404e-07, "loss": 2.6888, "step": 47397 }, { "epoch": 2.9423303743249116, "grad_norm": 0.32061016394215397, "learning_rate": 1.1252512854024733e-07, "loss": 2.7147, "step": 47398 }, { "epoch": 2.9423924514246695, "grad_norm": 0.326347886067793, "learning_rate": 1.1228309957533878e-07, "loss": 2.6283, "step": 47399 }, { "epoch": 2.9424545285244275, "grad_norm": 0.3414695589389891, "learning_rate": 1.120413308867585e-07, "loss": 2.6791, "step": 47400 }, { "epoch": 2.9425166056241854, "grad_norm": 0.3404474268026782, "learning_rate": 1.1179982247577214e-07, "loss": 2.6212, "step": 47401 }, { "epoch": 2.9425786827239433, "grad_norm": 0.33018186169283004, "learning_rate": 1.1155857434363426e-07, "loss": 2.6824, "step": 47402 }, { "epoch": 2.9426407598237008, "grad_norm": 0.34698656496085317, "learning_rate": 1.1131758649160495e-07, "loss": 2.7605, "step": 47403 }, { "epoch": 2.942702836923459, "grad_norm": 0.3351985544568726, "learning_rate": 1.1107685892094432e-07, "loss": 2.6502, "step": 47404 }, { "epoch": 2.9427649140232166, "grad_norm": 0.3324450214723484, "learning_rate": 1.1083639163290694e-07, "loss": 2.6376, "step": 47405 }, { "epoch": 2.942826991122975, "grad_norm": 0.3196637049309227, "learning_rate": 1.1059618462874733e-07, "loss": 2.6728, "step": 47406 }, { "epoch": 2.9428890682227324, "grad_norm": 0.32396477782819766, "learning_rate": 1.1035623790972005e-07, "loss": 2.6291, "step": 47407 }, { "epoch": 2.942951145322491, "grad_norm": 0.3207188796379796, "learning_rate": 1.1011655147707412e-07, "loss": 2.6459, "step": 47408 }, { "epoch": 2.9430132224222483, "grad_norm": 0.3364776992047048, "learning_rate": 1.0987712533205852e-07, "loss": 2.7535, "step": 47409 }, { "epoch": 2.943075299522006, "grad_norm": 0.3275620081125783, "learning_rate": 1.096379594759278e-07, "loss": 2.622, "step": 47410 }, { "epoch": 2.943137376621764, "grad_norm": 0.3412351297275608, "learning_rate": 1.09399053909931e-07, "loss": 2.7079, "step": 47411 }, { "epoch": 2.943199453721522, "grad_norm": 0.3335926815257058, "learning_rate": 1.0916040863530596e-07, "loss": 2.6356, "step": 47412 }, { "epoch": 2.94326153082128, "grad_norm": 0.3215584133173962, "learning_rate": 1.0892202365330173e-07, "loss": 2.6519, "step": 47413 }, { "epoch": 2.943323607921038, "grad_norm": 0.3137662386966348, "learning_rate": 1.0868389896516729e-07, "loss": 2.5587, "step": 47414 }, { "epoch": 2.943385685020796, "grad_norm": 0.331010674895853, "learning_rate": 1.0844603457213498e-07, "loss": 2.6632, "step": 47415 }, { "epoch": 2.9434477621205537, "grad_norm": 0.33351582012228886, "learning_rate": 1.0820843047545382e-07, "loss": 2.6589, "step": 47416 }, { "epoch": 2.9435098392203116, "grad_norm": 0.33626657824436357, "learning_rate": 1.0797108667636169e-07, "loss": 2.6061, "step": 47417 }, { "epoch": 2.9435719163200695, "grad_norm": 0.32651865678544245, "learning_rate": 1.077340031760965e-07, "loss": 2.6254, "step": 47418 }, { "epoch": 2.9436339934198275, "grad_norm": 0.3232699033719668, "learning_rate": 1.0749717997589059e-07, "loss": 2.6708, "step": 47419 }, { "epoch": 2.9436960705195854, "grad_norm": 0.34280520343578685, "learning_rate": 1.0726061707698743e-07, "loss": 2.7225, "step": 47420 }, { "epoch": 2.9437581476193433, "grad_norm": 0.33833909454201094, "learning_rate": 1.0702431448061378e-07, "loss": 2.6606, "step": 47421 }, { "epoch": 2.943820224719101, "grad_norm": 0.32712796275277073, "learning_rate": 1.0678827218801312e-07, "loss": 2.5626, "step": 47422 }, { "epoch": 2.943882301818859, "grad_norm": 0.35654663318097735, "learning_rate": 1.065524902004067e-07, "loss": 2.5126, "step": 47423 }, { "epoch": 2.943944378918617, "grad_norm": 0.32458800235168367, "learning_rate": 1.0631696851902684e-07, "loss": 2.7122, "step": 47424 }, { "epoch": 2.944006456018375, "grad_norm": 0.34676688803463096, "learning_rate": 1.0608170714510035e-07, "loss": 2.7198, "step": 47425 }, { "epoch": 2.944068533118133, "grad_norm": 0.3170393156038076, "learning_rate": 1.0584670607986514e-07, "loss": 2.6543, "step": 47426 }, { "epoch": 2.944130610217891, "grad_norm": 0.3494843520487121, "learning_rate": 1.0561196532453133e-07, "loss": 2.6165, "step": 47427 }, { "epoch": 2.9441926873176483, "grad_norm": 0.3185673017126851, "learning_rate": 1.0537748488034239e-07, "loss": 2.6266, "step": 47428 }, { "epoch": 2.9442547644174066, "grad_norm": 0.33890390334629356, "learning_rate": 1.051432647485029e-07, "loss": 2.71, "step": 47429 }, { "epoch": 2.944316841517164, "grad_norm": 0.32317325052862117, "learning_rate": 1.0490930493024521e-07, "loss": 2.6592, "step": 47430 }, { "epoch": 2.9443789186169225, "grad_norm": 0.3458134211223392, "learning_rate": 1.0467560542679056e-07, "loss": 2.6775, "step": 47431 }, { "epoch": 2.94444099571668, "grad_norm": 0.32936775919792854, "learning_rate": 1.0444216623935466e-07, "loss": 2.5314, "step": 47432 }, { "epoch": 2.944503072816438, "grad_norm": 0.32858916785092285, "learning_rate": 1.0420898736915874e-07, "loss": 2.675, "step": 47433 }, { "epoch": 2.944565149916196, "grad_norm": 0.3245915142888354, "learning_rate": 1.0397606881741296e-07, "loss": 2.7278, "step": 47434 }, { "epoch": 2.9446272270159537, "grad_norm": 0.3439186506102449, "learning_rate": 1.0374341058533854e-07, "loss": 2.6217, "step": 47435 }, { "epoch": 2.9446893041157116, "grad_norm": 0.3308814729339142, "learning_rate": 1.035110126741512e-07, "loss": 2.6256, "step": 47436 }, { "epoch": 2.9447513812154695, "grad_norm": 0.31924963435269343, "learning_rate": 1.0327887508505552e-07, "loss": 2.6434, "step": 47437 }, { "epoch": 2.9448134583152275, "grad_norm": 0.32164555630854297, "learning_rate": 1.0304699781926719e-07, "loss": 2.6302, "step": 47438 }, { "epoch": 2.9448755354149854, "grad_norm": 0.33212760200457514, "learning_rate": 1.0281538087799636e-07, "loss": 2.7031, "step": 47439 }, { "epoch": 2.9449376125147433, "grad_norm": 0.3187301793200281, "learning_rate": 1.0258402426244762e-07, "loss": 2.6559, "step": 47440 }, { "epoch": 2.944999689614501, "grad_norm": 0.31949698910391444, "learning_rate": 1.0235292797383667e-07, "loss": 2.6732, "step": 47441 }, { "epoch": 2.945061766714259, "grad_norm": 0.3262965712065238, "learning_rate": 1.0212209201335699e-07, "loss": 2.6459, "step": 47442 }, { "epoch": 2.945123843814017, "grad_norm": 0.3328727493016659, "learning_rate": 1.0189151638222427e-07, "loss": 2.6672, "step": 47443 }, { "epoch": 2.945185920913775, "grad_norm": 0.3340541142159653, "learning_rate": 1.0166120108163757e-07, "loss": 2.5817, "step": 47444 }, { "epoch": 2.945247998013533, "grad_norm": 0.3137567046767515, "learning_rate": 1.0143114611279591e-07, "loss": 2.7015, "step": 47445 }, { "epoch": 2.945310075113291, "grad_norm": 0.3312344246027573, "learning_rate": 1.0120135147689835e-07, "loss": 2.726, "step": 47446 }, { "epoch": 2.9453721522130487, "grad_norm": 0.3223092829869397, "learning_rate": 1.0097181717514947e-07, "loss": 2.5992, "step": 47447 }, { "epoch": 2.9454342293128066, "grad_norm": 0.3257286235100089, "learning_rate": 1.0074254320874277e-07, "loss": 2.698, "step": 47448 }, { "epoch": 2.9454963064125645, "grad_norm": 0.326814680514974, "learning_rate": 1.0051352957887728e-07, "loss": 2.6552, "step": 47449 }, { "epoch": 2.9455583835123225, "grad_norm": 0.33300883049683455, "learning_rate": 1.0028477628674648e-07, "loss": 2.655, "step": 47450 }, { "epoch": 2.94562046061208, "grad_norm": 0.3454795804303731, "learning_rate": 1.0005628333354389e-07, "loss": 2.5346, "step": 47451 }, { "epoch": 2.9456825377118383, "grad_norm": 0.3322825770893788, "learning_rate": 9.982805072045742e-08, "loss": 2.6477, "step": 47452 }, { "epoch": 2.9457446148115958, "grad_norm": 0.3344093875493631, "learning_rate": 9.960007844868613e-08, "loss": 2.6422, "step": 47453 }, { "epoch": 2.945806691911354, "grad_norm": 0.33039277783890136, "learning_rate": 9.93723665194124e-08, "loss": 2.582, "step": 47454 }, { "epoch": 2.9458687690111116, "grad_norm": 0.31283503692263354, "learning_rate": 9.914491493382971e-08, "loss": 2.6961, "step": 47455 }, { "epoch": 2.94593084611087, "grad_norm": 0.3501920373341154, "learning_rate": 9.891772369312047e-08, "loss": 2.6682, "step": 47456 }, { "epoch": 2.9459929232106274, "grad_norm": 0.3347815875918138, "learning_rate": 9.869079279846705e-08, "loss": 2.6847, "step": 47457 }, { "epoch": 2.9460550003103854, "grad_norm": 0.3416713157004115, "learning_rate": 9.846412225106294e-08, "loss": 2.6824, "step": 47458 }, { "epoch": 2.9461170774101433, "grad_norm": 0.32315021831526325, "learning_rate": 9.823771205208498e-08, "loss": 2.6827, "step": 47459 }, { "epoch": 2.946179154509901, "grad_norm": 0.3279859725152158, "learning_rate": 9.801156220271557e-08, "loss": 2.6795, "step": 47460 }, { "epoch": 2.946241231609659, "grad_norm": 0.3309117799261361, "learning_rate": 9.778567270413153e-08, "loss": 2.6425, "step": 47461 }, { "epoch": 2.946303308709417, "grad_norm": 0.3355391047402932, "learning_rate": 9.756004355751525e-08, "loss": 2.6225, "step": 47462 }, { "epoch": 2.946365385809175, "grad_norm": 0.33364155670237444, "learning_rate": 9.733467476403801e-08, "loss": 2.572, "step": 47463 }, { "epoch": 2.946427462908933, "grad_norm": 0.31909647838190824, "learning_rate": 9.710956632488777e-08, "loss": 2.6283, "step": 47464 }, { "epoch": 2.946489540008691, "grad_norm": 0.3043654280142444, "learning_rate": 9.688471824122469e-08, "loss": 2.5695, "step": 47465 }, { "epoch": 2.9465516171084487, "grad_norm": 0.3502633249476443, "learning_rate": 9.666013051423118e-08, "loss": 2.7353, "step": 47466 }, { "epoch": 2.9466136942082066, "grad_norm": 0.3360630843051217, "learning_rate": 9.64358031450785e-08, "loss": 2.618, "step": 47467 }, { "epoch": 2.9466757713079645, "grad_norm": 0.32416247833205397, "learning_rate": 9.62117361349324e-08, "loss": 2.6438, "step": 47468 }, { "epoch": 2.9467378484077225, "grad_norm": 0.31542655777928635, "learning_rate": 9.598792948496415e-08, "loss": 2.6977, "step": 47469 }, { "epoch": 2.9467999255074804, "grad_norm": 0.3294913476776111, "learning_rate": 9.576438319633951e-08, "loss": 2.6657, "step": 47470 }, { "epoch": 2.9468620026072383, "grad_norm": 0.3469602497265203, "learning_rate": 9.554109727022976e-08, "loss": 2.6372, "step": 47471 }, { "epoch": 2.946924079706996, "grad_norm": 0.32373118220692265, "learning_rate": 9.531807170779505e-08, "loss": 2.7339, "step": 47472 }, { "epoch": 2.946986156806754, "grad_norm": 0.3350389906803217, "learning_rate": 9.509530651020115e-08, "loss": 2.5628, "step": 47473 }, { "epoch": 2.947048233906512, "grad_norm": 0.3354287731264365, "learning_rate": 9.487280167861378e-08, "loss": 2.6237, "step": 47474 }, { "epoch": 2.94711031100627, "grad_norm": 0.32725498115676926, "learning_rate": 9.465055721418758e-08, "loss": 2.7261, "step": 47475 }, { "epoch": 2.9471723881060274, "grad_norm": 0.3422527325084325, "learning_rate": 9.442857311808274e-08, "loss": 2.7444, "step": 47476 }, { "epoch": 2.947234465205786, "grad_norm": 0.33247500666279795, "learning_rate": 9.420684939145941e-08, "loss": 2.6396, "step": 47477 }, { "epoch": 2.9472965423055433, "grad_norm": 0.3507748575626389, "learning_rate": 9.398538603547225e-08, "loss": 2.7062, "step": 47478 }, { "epoch": 2.9473586194053016, "grad_norm": 0.33808475496381707, "learning_rate": 9.376418305128143e-08, "loss": 2.6217, "step": 47479 }, { "epoch": 2.947420696505059, "grad_norm": 0.3400414538532201, "learning_rate": 9.354324044003604e-08, "loss": 2.6968, "step": 47480 }, { "epoch": 2.947482773604817, "grad_norm": 0.3228186630406661, "learning_rate": 9.332255820289071e-08, "loss": 2.6595, "step": 47481 }, { "epoch": 2.947544850704575, "grad_norm": 0.33023786347574724, "learning_rate": 9.310213634100007e-08, "loss": 2.6755, "step": 47482 }, { "epoch": 2.947606927804333, "grad_norm": 0.3397875991560685, "learning_rate": 9.288197485550765e-08, "loss": 2.6496, "step": 47483 }, { "epoch": 2.947669004904091, "grad_norm": 0.33720767730642165, "learning_rate": 9.266207374756808e-08, "loss": 2.636, "step": 47484 }, { "epoch": 2.9477310820038487, "grad_norm": 0.3258300977901554, "learning_rate": 9.244243301832489e-08, "loss": 2.6601, "step": 47485 }, { "epoch": 2.9477931591036066, "grad_norm": 0.3357550546999878, "learning_rate": 9.222305266892162e-08, "loss": 2.6818, "step": 47486 }, { "epoch": 2.9478552362033645, "grad_norm": 0.3188222667639151, "learning_rate": 9.200393270050734e-08, "loss": 2.5885, "step": 47487 }, { "epoch": 2.9479173133031225, "grad_norm": 0.3446994695202038, "learning_rate": 9.178507311422557e-08, "loss": 2.5844, "step": 47488 }, { "epoch": 2.9479793904028804, "grad_norm": 0.3236718793643893, "learning_rate": 9.156647391121987e-08, "loss": 2.7036, "step": 47489 }, { "epoch": 2.9480414675026383, "grad_norm": 0.3382458335714314, "learning_rate": 9.134813509262263e-08, "loss": 2.5837, "step": 47490 }, { "epoch": 2.948103544602396, "grad_norm": 0.32259162828375354, "learning_rate": 9.113005665958297e-08, "loss": 2.628, "step": 47491 }, { "epoch": 2.948165621702154, "grad_norm": 0.31698406112699734, "learning_rate": 9.091223861322773e-08, "loss": 2.7115, "step": 47492 }, { "epoch": 2.948227698801912, "grad_norm": 0.33264926166638836, "learning_rate": 9.069468095470047e-08, "loss": 2.6432, "step": 47493 }, { "epoch": 2.94828977590167, "grad_norm": 0.32021716490169233, "learning_rate": 9.047738368513914e-08, "loss": 2.7102, "step": 47494 }, { "epoch": 2.948351853001428, "grad_norm": 0.35573863170015774, "learning_rate": 9.026034680567063e-08, "loss": 2.7086, "step": 47495 }, { "epoch": 2.948413930101186, "grad_norm": 0.314787302092147, "learning_rate": 9.004357031742738e-08, "loss": 2.5915, "step": 47496 }, { "epoch": 2.9484760072009437, "grad_norm": 0.3410087217700595, "learning_rate": 8.982705422154736e-08, "loss": 2.6429, "step": 47497 }, { "epoch": 2.9485380843007016, "grad_norm": 0.32715162588814767, "learning_rate": 8.961079851915188e-08, "loss": 2.7326, "step": 47498 }, { "epoch": 2.948600161400459, "grad_norm": 0.3383544648826451, "learning_rate": 8.939480321137894e-08, "loss": 2.669, "step": 47499 }, { "epoch": 2.9486622385002175, "grad_norm": 0.3502550715590509, "learning_rate": 8.91790682993443e-08, "loss": 2.6495, "step": 47500 }, { "epoch": 2.948724315599975, "grad_norm": 0.3265846974982832, "learning_rate": 8.89635937841804e-08, "loss": 2.6006, "step": 47501 }, { "epoch": 2.9487863926997333, "grad_norm": 0.3527110200840839, "learning_rate": 8.874837966700856e-08, "loss": 2.6653, "step": 47502 }, { "epoch": 2.948848469799491, "grad_norm": 0.3247229206528471, "learning_rate": 8.853342594895564e-08, "loss": 2.6485, "step": 47503 }, { "epoch": 2.948910546899249, "grad_norm": 0.32689252494601573, "learning_rate": 8.831873263113744e-08, "loss": 2.6236, "step": 47504 }, { "epoch": 2.9489726239990066, "grad_norm": 0.32702911250495337, "learning_rate": 8.810429971467527e-08, "loss": 2.7645, "step": 47505 }, { "epoch": 2.9490347010987645, "grad_norm": 0.3241767281566022, "learning_rate": 8.789012720069045e-08, "loss": 2.6413, "step": 47506 }, { "epoch": 2.9490967781985225, "grad_norm": 0.32876705749121315, "learning_rate": 8.767621509030432e-08, "loss": 2.6605, "step": 47507 }, { "epoch": 2.9491588552982804, "grad_norm": 0.3349001748716741, "learning_rate": 8.746256338462155e-08, "loss": 2.7569, "step": 47508 }, { "epoch": 2.9492209323980383, "grad_norm": 0.3224614213261709, "learning_rate": 8.724917208476901e-08, "loss": 2.7314, "step": 47509 }, { "epoch": 2.949283009497796, "grad_norm": 0.3222980482956963, "learning_rate": 8.703604119185138e-08, "loss": 2.6161, "step": 47510 }, { "epoch": 2.949345086597554, "grad_norm": 0.3089704266858879, "learning_rate": 8.682317070698443e-08, "loss": 2.6491, "step": 47511 }, { "epoch": 2.949407163697312, "grad_norm": 0.31933122450590823, "learning_rate": 8.661056063127282e-08, "loss": 2.6627, "step": 47512 }, { "epoch": 2.94946924079707, "grad_norm": 0.32702481795254074, "learning_rate": 8.639821096583789e-08, "loss": 2.6707, "step": 47513 }, { "epoch": 2.949531317896828, "grad_norm": 0.3129363708772854, "learning_rate": 8.618612171177876e-08, "loss": 2.6495, "step": 47514 }, { "epoch": 2.949593394996586, "grad_norm": 0.32658785716996663, "learning_rate": 8.597429287020009e-08, "loss": 2.6909, "step": 47515 }, { "epoch": 2.9496554720963437, "grad_norm": 0.32696628338987094, "learning_rate": 8.576272444221212e-08, "loss": 2.6832, "step": 47516 }, { "epoch": 2.9497175491961016, "grad_norm": 0.3245510443691939, "learning_rate": 8.555141642891951e-08, "loss": 2.6897, "step": 47517 }, { "epoch": 2.9497796262958595, "grad_norm": 0.3263041062633567, "learning_rate": 8.534036883142138e-08, "loss": 2.6262, "step": 47518 }, { "epoch": 2.9498417033956175, "grad_norm": 0.31599354613781633, "learning_rate": 8.512958165081686e-08, "loss": 2.6742, "step": 47519 }, { "epoch": 2.9499037804953754, "grad_norm": 0.3468751920235592, "learning_rate": 8.491905488821061e-08, "loss": 2.7451, "step": 47520 }, { "epoch": 2.9499658575951333, "grad_norm": 0.32694473757177467, "learning_rate": 8.470878854470176e-08, "loss": 2.6446, "step": 47521 }, { "epoch": 2.950027934694891, "grad_norm": 0.34073790944956217, "learning_rate": 8.449878262138388e-08, "loss": 2.7085, "step": 47522 }, { "epoch": 2.950090011794649, "grad_norm": 0.32875564688195025, "learning_rate": 8.428903711935054e-08, "loss": 2.6461, "step": 47523 }, { "epoch": 2.9501520888944066, "grad_norm": 0.34692916490180126, "learning_rate": 8.407955203969531e-08, "loss": 2.6362, "step": 47524 }, { "epoch": 2.950214165994165, "grad_norm": 0.326290371779294, "learning_rate": 8.387032738352285e-08, "loss": 2.6441, "step": 47525 }, { "epoch": 2.9502762430939224, "grad_norm": 0.32957327341881704, "learning_rate": 8.366136315191008e-08, "loss": 2.6867, "step": 47526 }, { "epoch": 2.950338320193681, "grad_norm": 0.324520797671369, "learning_rate": 8.345265934595614e-08, "loss": 2.688, "step": 47527 }, { "epoch": 2.9504003972934383, "grad_norm": 0.3244717771118356, "learning_rate": 8.324421596674902e-08, "loss": 2.6696, "step": 47528 }, { "epoch": 2.950462474393196, "grad_norm": 0.31618171542732, "learning_rate": 8.303603301537122e-08, "loss": 2.7689, "step": 47529 }, { "epoch": 2.950524551492954, "grad_norm": 0.33992224915750613, "learning_rate": 8.282811049291073e-08, "loss": 2.7532, "step": 47530 }, { "epoch": 2.950586628592712, "grad_norm": 0.3311162008115067, "learning_rate": 8.262044840046113e-08, "loss": 2.6711, "step": 47531 }, { "epoch": 2.95064870569247, "grad_norm": 0.32329455179142375, "learning_rate": 8.241304673909378e-08, "loss": 2.5954, "step": 47532 }, { "epoch": 2.950710782792228, "grad_norm": 0.33412486462174684, "learning_rate": 8.220590550989671e-08, "loss": 2.6996, "step": 47533 }, { "epoch": 2.950772859891986, "grad_norm": 0.3298125016496379, "learning_rate": 8.199902471394683e-08, "loss": 2.6579, "step": 47534 }, { "epoch": 2.9508349369917437, "grad_norm": 0.3172988898022222, "learning_rate": 8.17924043523266e-08, "loss": 2.6949, "step": 47535 }, { "epoch": 2.9508970140915016, "grad_norm": 0.3189534571133922, "learning_rate": 8.158604442611295e-08, "loss": 2.6872, "step": 47536 }, { "epoch": 2.9509590911912595, "grad_norm": 0.3405572810796132, "learning_rate": 8.137994493638835e-08, "loss": 2.5938, "step": 47537 }, { "epoch": 2.9510211682910175, "grad_norm": 0.3349484477481103, "learning_rate": 8.117410588421859e-08, "loss": 2.7201, "step": 47538 }, { "epoch": 2.9510832453907754, "grad_norm": 0.3186842542050234, "learning_rate": 8.096852727068061e-08, "loss": 2.6279, "step": 47539 }, { "epoch": 2.9511453224905333, "grad_norm": 0.35236862796828766, "learning_rate": 8.07632090968513e-08, "loss": 2.6911, "step": 47540 }, { "epoch": 2.951207399590291, "grad_norm": 0.3282157227573515, "learning_rate": 8.055815136379651e-08, "loss": 2.573, "step": 47541 }, { "epoch": 2.951269476690049, "grad_norm": 0.33451127991809304, "learning_rate": 8.035335407258759e-08, "loss": 2.6439, "step": 47542 }, { "epoch": 2.951331553789807, "grad_norm": 0.3346856211939916, "learning_rate": 8.01488172242959e-08, "loss": 2.708, "step": 47543 }, { "epoch": 2.951393630889565, "grad_norm": 0.3381588483146028, "learning_rate": 7.994454081998726e-08, "loss": 2.7267, "step": 47544 }, { "epoch": 2.951455707989323, "grad_norm": 0.3337084092967963, "learning_rate": 7.974052486072192e-08, "loss": 2.6903, "step": 47545 }, { "epoch": 2.951517785089081, "grad_norm": 0.33396694070920796, "learning_rate": 7.953676934757126e-08, "loss": 2.5792, "step": 47546 }, { "epoch": 2.9515798621888383, "grad_norm": 0.32473465651400046, "learning_rate": 7.933327428159554e-08, "loss": 2.6803, "step": 47547 }, { "epoch": 2.9516419392885966, "grad_norm": 0.34201398221233575, "learning_rate": 7.913003966385502e-08, "loss": 2.6989, "step": 47548 }, { "epoch": 2.951704016388354, "grad_norm": 0.316125274044173, "learning_rate": 7.89270654954155e-08, "loss": 2.6425, "step": 47549 }, { "epoch": 2.9517660934881125, "grad_norm": 0.3105672913563848, "learning_rate": 7.872435177733173e-08, "loss": 2.62, "step": 47550 }, { "epoch": 2.95182817058787, "grad_norm": 0.3199973416776807, "learning_rate": 7.852189851065838e-08, "loss": 2.5734, "step": 47551 }, { "epoch": 2.9518902476876283, "grad_norm": 0.349368870447672, "learning_rate": 7.83197056964613e-08, "loss": 2.6504, "step": 47552 }, { "epoch": 2.951952324787386, "grad_norm": 0.35338601745052145, "learning_rate": 7.811777333578408e-08, "loss": 2.6892, "step": 47553 }, { "epoch": 2.9520144018871437, "grad_norm": 0.33830532358332327, "learning_rate": 7.791610142968697e-08, "loss": 2.728, "step": 47554 }, { "epoch": 2.9520764789869016, "grad_norm": 0.3351425257009141, "learning_rate": 7.771468997921916e-08, "loss": 2.6649, "step": 47555 }, { "epoch": 2.9521385560866595, "grad_norm": 0.3268362007126376, "learning_rate": 7.751353898543534e-08, "loss": 2.6733, "step": 47556 }, { "epoch": 2.9522006331864175, "grad_norm": 0.33110365603913294, "learning_rate": 7.731264844937913e-08, "loss": 2.6755, "step": 47557 }, { "epoch": 2.9522627102861754, "grad_norm": 0.3331856462516158, "learning_rate": 7.711201837210525e-08, "loss": 2.6332, "step": 47558 }, { "epoch": 2.9523247873859333, "grad_norm": 0.3215392857282542, "learning_rate": 7.691164875465728e-08, "loss": 2.5541, "step": 47559 }, { "epoch": 2.952386864485691, "grad_norm": 0.3238214563435462, "learning_rate": 7.671153959807886e-08, "loss": 2.6426, "step": 47560 }, { "epoch": 2.952448941585449, "grad_norm": 0.3429804771695051, "learning_rate": 7.65116909034136e-08, "loss": 2.6732, "step": 47561 }, { "epoch": 2.952511018685207, "grad_norm": 0.34542846544880634, "learning_rate": 7.631210267171063e-08, "loss": 2.6612, "step": 47562 }, { "epoch": 2.952573095784965, "grad_norm": 0.3221996702269187, "learning_rate": 7.611277490400248e-08, "loss": 2.6324, "step": 47563 }, { "epoch": 2.952635172884723, "grad_norm": 0.3428929438483262, "learning_rate": 7.591370760133832e-08, "loss": 2.5734, "step": 47564 }, { "epoch": 2.952697249984481, "grad_norm": 0.332629887085873, "learning_rate": 7.571490076475063e-08, "loss": 2.6288, "step": 47565 }, { "epoch": 2.9527593270842387, "grad_norm": 0.3327266927254575, "learning_rate": 7.551635439527193e-08, "loss": 2.6844, "step": 47566 }, { "epoch": 2.9528214041839966, "grad_norm": 0.34051823824016864, "learning_rate": 7.53180684939514e-08, "loss": 2.7866, "step": 47567 }, { "epoch": 2.9528834812837546, "grad_norm": 0.32964885042184805, "learning_rate": 7.512004306181597e-08, "loss": 2.6667, "step": 47568 }, { "epoch": 2.9529455583835125, "grad_norm": 0.3436782935519361, "learning_rate": 7.492227809989261e-08, "loss": 2.6643, "step": 47569 }, { "epoch": 2.9530076354832704, "grad_norm": 0.31891402390689766, "learning_rate": 7.472477360922491e-08, "loss": 2.5299, "step": 47570 }, { "epoch": 2.9530697125830283, "grad_norm": 0.3313709200273702, "learning_rate": 7.452752959083431e-08, "loss": 2.7241, "step": 47571 }, { "epoch": 2.953131789682786, "grad_norm": 0.32600659048697056, "learning_rate": 7.433054604575884e-08, "loss": 2.7505, "step": 47572 }, { "epoch": 2.953193866782544, "grad_norm": 0.3209609933665583, "learning_rate": 7.413382297501437e-08, "loss": 2.6994, "step": 47573 }, { "epoch": 2.9532559438823016, "grad_norm": 0.3526236731102991, "learning_rate": 7.393736037963894e-08, "loss": 2.6631, "step": 47574 }, { "epoch": 2.95331802098206, "grad_norm": 0.3190895903554101, "learning_rate": 7.374115826064843e-08, "loss": 2.6669, "step": 47575 }, { "epoch": 2.9533800980818175, "grad_norm": 0.32528671492259886, "learning_rate": 7.354521661906977e-08, "loss": 2.704, "step": 47576 }, { "epoch": 2.9534421751815754, "grad_norm": 0.32052189100403206, "learning_rate": 7.334953545592994e-08, "loss": 2.6118, "step": 47577 }, { "epoch": 2.9535042522813333, "grad_norm": 0.3239039190882916, "learning_rate": 7.315411477223921e-08, "loss": 2.631, "step": 47578 }, { "epoch": 2.953566329381091, "grad_norm": 0.34590855114635194, "learning_rate": 7.295895456903012e-08, "loss": 2.7177, "step": 47579 }, { "epoch": 2.953628406480849, "grad_norm": 0.33946293288834456, "learning_rate": 7.276405484731297e-08, "loss": 2.6828, "step": 47580 }, { "epoch": 2.953690483580607, "grad_norm": 0.33965280803130077, "learning_rate": 7.25694156081036e-08, "loss": 2.646, "step": 47581 }, { "epoch": 2.953752560680365, "grad_norm": 0.32914340831377475, "learning_rate": 7.237503685242342e-08, "loss": 2.5336, "step": 47582 }, { "epoch": 2.953814637780123, "grad_norm": 0.32506635474523393, "learning_rate": 7.218091858128274e-08, "loss": 2.6811, "step": 47583 }, { "epoch": 2.953876714879881, "grad_norm": 0.32676937860333904, "learning_rate": 7.198706079569185e-08, "loss": 2.6382, "step": 47584 }, { "epoch": 2.9539387919796387, "grad_norm": 0.3443406838767725, "learning_rate": 7.179346349666661e-08, "loss": 2.6404, "step": 47585 }, { "epoch": 2.9540008690793966, "grad_norm": 0.3194068340030814, "learning_rate": 7.160012668521176e-08, "loss": 2.6886, "step": 47586 }, { "epoch": 2.9540629461791545, "grad_norm": 0.33076748162951863, "learning_rate": 7.140705036234318e-08, "loss": 2.72, "step": 47587 }, { "epoch": 2.9541250232789125, "grad_norm": 0.3377395578834279, "learning_rate": 7.121423452906561e-08, "loss": 2.6017, "step": 47588 }, { "epoch": 2.9541871003786704, "grad_norm": 0.33889448292319885, "learning_rate": 7.10216791863838e-08, "loss": 2.7019, "step": 47589 }, { "epoch": 2.9542491774784283, "grad_norm": 0.34030827749976295, "learning_rate": 7.082938433529696e-08, "loss": 2.6656, "step": 47590 }, { "epoch": 2.954311254578186, "grad_norm": 0.32998797071452163, "learning_rate": 7.063734997682092e-08, "loss": 2.6873, "step": 47591 }, { "epoch": 2.954373331677944, "grad_norm": 0.3364300997474894, "learning_rate": 7.04455761119438e-08, "loss": 2.5596, "step": 47592 }, { "epoch": 2.954435408777702, "grad_norm": 0.3166987868172861, "learning_rate": 7.025406274168145e-08, "loss": 2.6506, "step": 47593 }, { "epoch": 2.95449748587746, "grad_norm": 0.3319953674673994, "learning_rate": 7.00628098670164e-08, "loss": 2.6738, "step": 47594 }, { "epoch": 2.9545595629772174, "grad_norm": 0.3313517292828073, "learning_rate": 6.9871817488959e-08, "loss": 2.6954, "step": 47595 }, { "epoch": 2.954621640076976, "grad_norm": 0.3239021018904018, "learning_rate": 6.968108560850284e-08, "loss": 2.6653, "step": 47596 }, { "epoch": 2.9546837171767333, "grad_norm": 0.3178281740432641, "learning_rate": 6.949061422663605e-08, "loss": 2.6604, "step": 47597 }, { "epoch": 2.9547457942764916, "grad_norm": 0.3179855023594152, "learning_rate": 6.930040334436339e-08, "loss": 2.6321, "step": 47598 }, { "epoch": 2.954807871376249, "grad_norm": 0.34108683267622397, "learning_rate": 6.911045296266738e-08, "loss": 2.6591, "step": 47599 }, { "epoch": 2.954869948476007, "grad_norm": 0.3313486824952908, "learning_rate": 6.892076308254725e-08, "loss": 2.6751, "step": 47600 }, { "epoch": 2.954932025575765, "grad_norm": 0.31512117208617063, "learning_rate": 6.873133370498552e-08, "loss": 2.6361, "step": 47601 }, { "epoch": 2.954994102675523, "grad_norm": 0.3285845195545846, "learning_rate": 6.854216483097587e-08, "loss": 2.6976, "step": 47602 }, { "epoch": 2.955056179775281, "grad_norm": 0.3203774944517892, "learning_rate": 6.835325646150081e-08, "loss": 2.5781, "step": 47603 }, { "epoch": 2.9551182568750387, "grad_norm": 0.32241037759138136, "learning_rate": 6.816460859754847e-08, "loss": 2.6468, "step": 47604 }, { "epoch": 2.9551803339747966, "grad_norm": 0.33254101833654237, "learning_rate": 6.79762212401014e-08, "loss": 2.6784, "step": 47605 }, { "epoch": 2.9552424110745545, "grad_norm": 0.34676492207397946, "learning_rate": 6.778809439014766e-08, "loss": 2.7183, "step": 47606 }, { "epoch": 2.9553044881743125, "grad_norm": 0.31991645705930083, "learning_rate": 6.760022804866428e-08, "loss": 2.759, "step": 47607 }, { "epoch": 2.9553665652740704, "grad_norm": 0.3258784110363017, "learning_rate": 6.741262221662825e-08, "loss": 2.6503, "step": 47608 }, { "epoch": 2.9554286423738283, "grad_norm": 0.33364671514495614, "learning_rate": 6.722527689502766e-08, "loss": 2.6233, "step": 47609 }, { "epoch": 2.955490719473586, "grad_norm": 0.3284499058435717, "learning_rate": 6.703819208482842e-08, "loss": 2.5927, "step": 47610 }, { "epoch": 2.955552796573344, "grad_norm": 0.3369616267588668, "learning_rate": 6.685136778701861e-08, "loss": 2.672, "step": 47611 }, { "epoch": 2.955614873673102, "grad_norm": 0.32965043252774523, "learning_rate": 6.666480400256414e-08, "loss": 2.6967, "step": 47612 }, { "epoch": 2.95567695077286, "grad_norm": 0.3114001469605037, "learning_rate": 6.647850073243644e-08, "loss": 2.6304, "step": 47613 }, { "epoch": 2.955739027872618, "grad_norm": 0.33351474782399904, "learning_rate": 6.629245797761807e-08, "loss": 2.6218, "step": 47614 }, { "epoch": 2.955801104972376, "grad_norm": 0.31909745903799636, "learning_rate": 6.610667573907492e-08, "loss": 2.6263, "step": 47615 }, { "epoch": 2.9558631820721337, "grad_norm": 0.3326567424163112, "learning_rate": 6.592115401777288e-08, "loss": 2.6382, "step": 47616 }, { "epoch": 2.9559252591718916, "grad_norm": 0.3263696425029983, "learning_rate": 6.573589281467784e-08, "loss": 2.7026, "step": 47617 }, { "epoch": 2.955987336271649, "grad_norm": 0.34665950662680034, "learning_rate": 6.555089213076682e-08, "loss": 2.7758, "step": 47618 }, { "epoch": 2.9560494133714075, "grad_norm": 0.3375821993651908, "learning_rate": 6.536615196699458e-08, "loss": 2.6867, "step": 47619 }, { "epoch": 2.956111490471165, "grad_norm": 0.3233968589422573, "learning_rate": 6.518167232433259e-08, "loss": 2.728, "step": 47620 }, { "epoch": 2.9561735675709233, "grad_norm": 0.32926663976040565, "learning_rate": 6.499745320374117e-08, "loss": 2.7435, "step": 47621 }, { "epoch": 2.956235644670681, "grad_norm": 0.34346881116287425, "learning_rate": 6.481349460617514e-08, "loss": 2.6373, "step": 47622 }, { "epoch": 2.956297721770439, "grad_norm": 0.3216625697775993, "learning_rate": 6.462979653260037e-08, "loss": 2.7701, "step": 47623 }, { "epoch": 2.9563597988701966, "grad_norm": 0.3508677505516167, "learning_rate": 6.444635898397722e-08, "loss": 2.7122, "step": 47624 }, { "epoch": 2.9564218759699545, "grad_norm": 0.3212681679467999, "learning_rate": 6.426318196125491e-08, "loss": 2.5985, "step": 47625 }, { "epoch": 2.9564839530697125, "grad_norm": 0.3237552316898359, "learning_rate": 6.408026546539936e-08, "loss": 2.6727, "step": 47626 }, { "epoch": 2.9565460301694704, "grad_norm": 0.32358622609909415, "learning_rate": 6.389760949735423e-08, "loss": 2.6824, "step": 47627 }, { "epoch": 2.9566081072692283, "grad_norm": 0.3351623107788977, "learning_rate": 6.371521405807989e-08, "loss": 2.7195, "step": 47628 }, { "epoch": 2.956670184368986, "grad_norm": 0.3270758363245228, "learning_rate": 6.353307914852557e-08, "loss": 2.6465, "step": 47629 }, { "epoch": 2.956732261468744, "grad_norm": 0.3226873932372232, "learning_rate": 6.335120476964607e-08, "loss": 2.5895, "step": 47630 }, { "epoch": 2.956794338568502, "grad_norm": 0.32080835249282325, "learning_rate": 6.31695909223795e-08, "loss": 2.7002, "step": 47631 }, { "epoch": 2.95685641566826, "grad_norm": 0.3257408492105085, "learning_rate": 6.298823760768069e-08, "loss": 2.6506, "step": 47632 }, { "epoch": 2.956918492768018, "grad_norm": 0.3324684182204929, "learning_rate": 6.280714482649885e-08, "loss": 2.7016, "step": 47633 }, { "epoch": 2.956980569867776, "grad_norm": 0.32468613865689044, "learning_rate": 6.262631257977214e-08, "loss": 2.6845, "step": 47634 }, { "epoch": 2.9570426469675337, "grad_norm": 0.33289998789549785, "learning_rate": 6.244574086844979e-08, "loss": 2.6844, "step": 47635 }, { "epoch": 2.9571047240672916, "grad_norm": 0.31128075692825474, "learning_rate": 6.226542969346438e-08, "loss": 2.7081, "step": 47636 }, { "epoch": 2.9571668011670496, "grad_norm": 0.34670803330282435, "learning_rate": 6.208537905577072e-08, "loss": 2.6426, "step": 47637 }, { "epoch": 2.9572288782668075, "grad_norm": 0.3350020958884707, "learning_rate": 6.190558895629583e-08, "loss": 2.7738, "step": 47638 }, { "epoch": 2.9572909553665654, "grad_norm": 0.34355127485791287, "learning_rate": 6.172605939598341e-08, "loss": 2.6589, "step": 47639 }, { "epoch": 2.9573530324663233, "grad_norm": 0.3140680194009147, "learning_rate": 6.154679037576604e-08, "loss": 2.7043, "step": 47640 }, { "epoch": 2.9574151095660812, "grad_norm": 0.33560406971246737, "learning_rate": 6.136778189658743e-08, "loss": 2.6901, "step": 47641 }, { "epoch": 2.957477186665839, "grad_norm": 0.31534490243037533, "learning_rate": 6.118903395936903e-08, "loss": 2.611, "step": 47642 }, { "epoch": 2.9575392637655966, "grad_norm": 0.31698089974379123, "learning_rate": 6.101054656505456e-08, "loss": 2.6319, "step": 47643 }, { "epoch": 2.957601340865355, "grad_norm": 0.3404603275729622, "learning_rate": 6.083231971457659e-08, "loss": 2.6685, "step": 47644 }, { "epoch": 2.9576634179651125, "grad_norm": 0.34531452946476476, "learning_rate": 6.065435340885106e-08, "loss": 2.7228, "step": 47645 }, { "epoch": 2.957725495064871, "grad_norm": 0.3199270650887469, "learning_rate": 6.04766476488161e-08, "loss": 2.6633, "step": 47646 }, { "epoch": 2.9577875721646283, "grad_norm": 0.3272505757324598, "learning_rate": 6.029920243540432e-08, "loss": 2.6579, "step": 47647 }, { "epoch": 2.957849649264386, "grad_norm": 0.3450741191041632, "learning_rate": 6.012201776953164e-08, "loss": 2.7126, "step": 47648 }, { "epoch": 2.957911726364144, "grad_norm": 0.3175929151706295, "learning_rate": 5.994509365211953e-08, "loss": 2.6345, "step": 47649 }, { "epoch": 2.957973803463902, "grad_norm": 0.32698354408653313, "learning_rate": 5.976843008410616e-08, "loss": 2.6981, "step": 47650 }, { "epoch": 2.95803588056366, "grad_norm": 0.3198016559125853, "learning_rate": 5.959202706640188e-08, "loss": 2.5876, "step": 47651 }, { "epoch": 2.958097957663418, "grad_norm": 0.3490814613308179, "learning_rate": 5.9415884599928195e-08, "loss": 2.7519, "step": 47652 }, { "epoch": 2.958160034763176, "grad_norm": 0.3298969561637333, "learning_rate": 5.9240002685601034e-08, "loss": 2.5523, "step": 47653 }, { "epoch": 2.9582221118629337, "grad_norm": 0.3235680127925963, "learning_rate": 5.906438132434744e-08, "loss": 2.643, "step": 47654 }, { "epoch": 2.9582841889626916, "grad_norm": 0.3363446939354658, "learning_rate": 5.8889020517077786e-08, "loss": 2.6483, "step": 47655 }, { "epoch": 2.9583462660624495, "grad_norm": 0.32534042190932205, "learning_rate": 5.871392026470801e-08, "loss": 2.6242, "step": 47656 }, { "epoch": 2.9584083431622075, "grad_norm": 0.3220579923421281, "learning_rate": 5.85390805681485e-08, "loss": 2.604, "step": 47657 }, { "epoch": 2.9584704202619654, "grad_norm": 0.33358949886355965, "learning_rate": 5.836450142832073e-08, "loss": 2.6498, "step": 47658 }, { "epoch": 2.9585324973617233, "grad_norm": 0.32759496476795047, "learning_rate": 5.819018284612399e-08, "loss": 2.5739, "step": 47659 }, { "epoch": 2.958594574461481, "grad_norm": 0.3178099120718316, "learning_rate": 5.801612482247421e-08, "loss": 2.5983, "step": 47660 }, { "epoch": 2.958656651561239, "grad_norm": 0.3455071729068648, "learning_rate": 5.7842327358276214e-08, "loss": 2.6416, "step": 47661 }, { "epoch": 2.958718728660997, "grad_norm": 0.3340403192077489, "learning_rate": 5.76687904544404e-08, "loss": 2.7224, "step": 47662 }, { "epoch": 2.958780805760755, "grad_norm": 0.3364776327761004, "learning_rate": 5.7495514111871597e-08, "loss": 2.6645, "step": 47663 }, { "epoch": 2.958842882860513, "grad_norm": 0.32584821895506555, "learning_rate": 5.7322498331469074e-08, "loss": 2.6612, "step": 47664 }, { "epoch": 2.958904959960271, "grad_norm": 0.31812873630165733, "learning_rate": 5.714974311414323e-08, "loss": 2.6594, "step": 47665 }, { "epoch": 2.9589670370600283, "grad_norm": 0.3243394190507108, "learning_rate": 5.697724846078778e-08, "loss": 2.7428, "step": 47666 }, { "epoch": 2.9590291141597866, "grad_norm": 0.3251264688289247, "learning_rate": 5.6805014372307564e-08, "loss": 2.6192, "step": 47667 }, { "epoch": 2.959091191259544, "grad_norm": 0.3377866750474379, "learning_rate": 5.663304084960186e-08, "loss": 2.6489, "step": 47668 }, { "epoch": 2.9591532683593025, "grad_norm": 0.32788276813829176, "learning_rate": 5.64613278935644e-08, "loss": 2.6498, "step": 47669 }, { "epoch": 2.95921534545906, "grad_norm": 0.3462030599809005, "learning_rate": 5.628987550508891e-08, "loss": 2.6909, "step": 47670 }, { "epoch": 2.9592774225588183, "grad_norm": 0.31865623301062407, "learning_rate": 5.611868368508022e-08, "loss": 2.6933, "step": 47671 }, { "epoch": 2.959339499658576, "grad_norm": 0.3468573423994784, "learning_rate": 5.5947752434420964e-08, "loss": 2.6721, "step": 47672 }, { "epoch": 2.9594015767583337, "grad_norm": 0.32076025116015716, "learning_rate": 5.577708175400487e-08, "loss": 2.6428, "step": 47673 }, { "epoch": 2.9594636538580916, "grad_norm": 0.3394474042228222, "learning_rate": 5.560667164472566e-08, "loss": 2.7069, "step": 47674 }, { "epoch": 2.9595257309578495, "grad_norm": 0.30936947634611367, "learning_rate": 5.543652210747152e-08, "loss": 2.5699, "step": 47675 }, { "epoch": 2.9595878080576075, "grad_norm": 0.32028219033098676, "learning_rate": 5.526663314312508e-08, "loss": 2.6274, "step": 47676 }, { "epoch": 2.9596498851573654, "grad_norm": 0.32453233422073907, "learning_rate": 5.509700475258561e-08, "loss": 2.6367, "step": 47677 }, { "epoch": 2.9597119622571233, "grad_norm": 0.33479609756924245, "learning_rate": 5.492763693672465e-08, "loss": 2.6582, "step": 47678 }, { "epoch": 2.959774039356881, "grad_norm": 0.35822763972702487, "learning_rate": 5.475852969643036e-08, "loss": 2.6878, "step": 47679 }, { "epoch": 2.959836116456639, "grad_norm": 0.3065210826166417, "learning_rate": 5.458968303258538e-08, "loss": 2.7086, "step": 47680 }, { "epoch": 2.959898193556397, "grad_norm": 0.31941488147624536, "learning_rate": 5.442109694607233e-08, "loss": 2.6283, "step": 47681 }, { "epoch": 2.959960270656155, "grad_norm": 0.32274812355574545, "learning_rate": 5.42527714377683e-08, "loss": 2.5888, "step": 47682 }, { "epoch": 2.960022347755913, "grad_norm": 0.34576180601202394, "learning_rate": 5.408470650855035e-08, "loss": 2.7127, "step": 47683 }, { "epoch": 2.960084424855671, "grad_norm": 0.3465922823045875, "learning_rate": 5.3916902159301117e-08, "loss": 2.7244, "step": 47684 }, { "epoch": 2.9601465019554287, "grad_norm": 0.3404433452867045, "learning_rate": 5.374935839089212e-08, "loss": 2.6098, "step": 47685 }, { "epoch": 2.9602085790551866, "grad_norm": 0.32964626253226237, "learning_rate": 5.358207520419489e-08, "loss": 2.7608, "step": 47686 }, { "epoch": 2.9602706561549446, "grad_norm": 0.33589735456607744, "learning_rate": 5.34150526000865e-08, "loss": 2.6226, "step": 47687 }, { "epoch": 2.9603327332547025, "grad_norm": 0.3505436757577006, "learning_rate": 5.3248290579432926e-08, "loss": 2.6921, "step": 47688 }, { "epoch": 2.9603948103544604, "grad_norm": 0.32341766204441635, "learning_rate": 5.308178914311124e-08, "loss": 2.6402, "step": 47689 }, { "epoch": 2.9604568874542183, "grad_norm": 0.317201826871329, "learning_rate": 5.2915548291987415e-08, "loss": 2.6789, "step": 47690 }, { "epoch": 2.960518964553976, "grad_norm": 0.32750296669628226, "learning_rate": 5.274956802692743e-08, "loss": 2.6797, "step": 47691 }, { "epoch": 2.960581041653734, "grad_norm": 0.3136994825971469, "learning_rate": 5.258384834879726e-08, "loss": 2.676, "step": 47692 }, { "epoch": 2.9606431187534916, "grad_norm": 0.3343511207467656, "learning_rate": 5.2418389258462874e-08, "loss": 2.7029, "step": 47693 }, { "epoch": 2.96070519585325, "grad_norm": 0.34370782983647225, "learning_rate": 5.225319075679025e-08, "loss": 2.7486, "step": 47694 }, { "epoch": 2.9607672729530075, "grad_norm": 0.3215383009336293, "learning_rate": 5.208825284463425e-08, "loss": 2.6343, "step": 47695 }, { "epoch": 2.9608293500527654, "grad_norm": 0.33437243442798953, "learning_rate": 5.192357552286087e-08, "loss": 2.7153, "step": 47696 }, { "epoch": 2.9608914271525233, "grad_norm": 0.3413442939566078, "learning_rate": 5.175915879232496e-08, "loss": 2.7238, "step": 47697 }, { "epoch": 2.960953504252281, "grad_norm": 0.3236089970465512, "learning_rate": 5.159500265388695e-08, "loss": 2.6428, "step": 47698 }, { "epoch": 2.961015581352039, "grad_norm": 0.3352288934370165, "learning_rate": 5.1431107108407264e-08, "loss": 2.6239, "step": 47699 }, { "epoch": 2.961077658451797, "grad_norm": 0.32957405335076895, "learning_rate": 5.126747215673522e-08, "loss": 2.573, "step": 47700 }, { "epoch": 2.961139735551555, "grad_norm": 0.32000330213616224, "learning_rate": 5.1104097799725694e-08, "loss": 2.6839, "step": 47701 }, { "epoch": 2.961201812651313, "grad_norm": 0.31922802282860907, "learning_rate": 5.0940984038228e-08, "loss": 2.6006, "step": 47702 }, { "epoch": 2.961263889751071, "grad_norm": 0.33775251796469963, "learning_rate": 5.077813087310257e-08, "loss": 2.6103, "step": 47703 }, { "epoch": 2.9613259668508287, "grad_norm": 0.33836144516495903, "learning_rate": 5.0615538305193164e-08, "loss": 2.6899, "step": 47704 }, { "epoch": 2.9613880439505866, "grad_norm": 0.3467862032739128, "learning_rate": 5.045320633534356e-08, "loss": 2.6286, "step": 47705 }, { "epoch": 2.9614501210503446, "grad_norm": 0.32553395689895065, "learning_rate": 5.0291134964408625e-08, "loss": 2.6631, "step": 47706 }, { "epoch": 2.9615121981501025, "grad_norm": 0.32208269880823664, "learning_rate": 5.012932419322658e-08, "loss": 2.7485, "step": 47707 }, { "epoch": 2.9615742752498604, "grad_norm": 0.3169447237536951, "learning_rate": 4.996777402265229e-08, "loss": 2.7268, "step": 47708 }, { "epoch": 2.9616363523496183, "grad_norm": 0.31039336146913415, "learning_rate": 4.980648445351843e-08, "loss": 2.6786, "step": 47709 }, { "epoch": 2.9616984294493762, "grad_norm": 0.32202088281489966, "learning_rate": 4.964545548666877e-08, "loss": 2.6747, "step": 47710 }, { "epoch": 2.961760506549134, "grad_norm": 0.320660739100304, "learning_rate": 4.948468712294707e-08, "loss": 2.616, "step": 47711 }, { "epoch": 2.961822583648892, "grad_norm": 0.34502923877835745, "learning_rate": 4.932417936318601e-08, "loss": 2.632, "step": 47712 }, { "epoch": 2.96188466074865, "grad_norm": 0.3251498307965158, "learning_rate": 4.9163932208229346e-08, "loss": 2.6365, "step": 47713 }, { "epoch": 2.9619467378484075, "grad_norm": 0.3178896334584153, "learning_rate": 4.9003945658909754e-08, "loss": 2.698, "step": 47714 }, { "epoch": 2.962008814948166, "grad_norm": 0.32349929495680096, "learning_rate": 4.88442197160599e-08, "loss": 2.6168, "step": 47715 }, { "epoch": 2.9620708920479233, "grad_norm": 0.3292867213230261, "learning_rate": 4.868475438051801e-08, "loss": 2.6859, "step": 47716 }, { "epoch": 2.9621329691476816, "grad_norm": 0.3348727541140164, "learning_rate": 4.852554965311673e-08, "loss": 2.666, "step": 47717 }, { "epoch": 2.962195046247439, "grad_norm": 0.32673981611876524, "learning_rate": 4.83666055346832e-08, "loss": 2.5848, "step": 47718 }, { "epoch": 2.9622571233471975, "grad_norm": 0.326877208268272, "learning_rate": 4.8207922026044515e-08, "loss": 2.7, "step": 47719 }, { "epoch": 2.962319200446955, "grad_norm": 0.3419233578578014, "learning_rate": 4.804949912803336e-08, "loss": 2.7164, "step": 47720 }, { "epoch": 2.962381277546713, "grad_norm": 0.32818308951851016, "learning_rate": 4.789133684147129e-08, "loss": 2.7964, "step": 47721 }, { "epoch": 2.962443354646471, "grad_norm": 0.3300227508144054, "learning_rate": 4.773343516718542e-08, "loss": 2.5853, "step": 47722 }, { "epoch": 2.9625054317462287, "grad_norm": 0.3470269733733953, "learning_rate": 4.757579410600843e-08, "loss": 2.7165, "step": 47723 }, { "epoch": 2.9625675088459866, "grad_norm": 0.3171586404083195, "learning_rate": 4.7418413658745217e-08, "loss": 2.6355, "step": 47724 }, { "epoch": 2.9626295859457445, "grad_norm": 0.30898341192006173, "learning_rate": 4.726129382623401e-08, "loss": 2.6203, "step": 47725 }, { "epoch": 2.9626916630455025, "grad_norm": 0.3519528553037637, "learning_rate": 4.710443460928527e-08, "loss": 2.65, "step": 47726 }, { "epoch": 2.9627537401452604, "grad_norm": 0.33994120811800727, "learning_rate": 4.6947836008714995e-08, "loss": 2.6509, "step": 47727 }, { "epoch": 2.9628158172450183, "grad_norm": 0.3159157282707092, "learning_rate": 4.6791498025350325e-08, "loss": 2.6459, "step": 47728 }, { "epoch": 2.962877894344776, "grad_norm": 0.3361152688678717, "learning_rate": 4.6635420659996156e-08, "loss": 2.6563, "step": 47729 }, { "epoch": 2.962939971444534, "grad_norm": 0.3199250389939422, "learning_rate": 4.647960391347406e-08, "loss": 2.6411, "step": 47730 }, { "epoch": 2.963002048544292, "grad_norm": 0.3339062314888106, "learning_rate": 4.6324047786594496e-08, "loss": 2.714, "step": 47731 }, { "epoch": 2.96306412564405, "grad_norm": 0.3334382483162031, "learning_rate": 4.616875228016793e-08, "loss": 2.649, "step": 47732 }, { "epoch": 2.963126202743808, "grad_norm": 0.3162925539487125, "learning_rate": 4.601371739501037e-08, "loss": 2.6517, "step": 47733 }, { "epoch": 2.963188279843566, "grad_norm": 0.3262395845592236, "learning_rate": 4.5858943131921185e-08, "loss": 2.7386, "step": 47734 }, { "epoch": 2.9632503569433237, "grad_norm": 0.3435517519705633, "learning_rate": 4.570442949171083e-08, "loss": 2.687, "step": 47735 }, { "epoch": 2.9633124340430816, "grad_norm": 0.32182330207083426, "learning_rate": 4.5550176475189776e-08, "loss": 2.754, "step": 47736 }, { "epoch": 2.9633745111428396, "grad_norm": 0.3353766483617562, "learning_rate": 4.539618408316293e-08, "loss": 2.7092, "step": 47737 }, { "epoch": 2.9634365882425975, "grad_norm": 0.3316445960356406, "learning_rate": 4.524245231642965e-08, "loss": 2.6542, "step": 47738 }, { "epoch": 2.963498665342355, "grad_norm": 0.3255529413282856, "learning_rate": 4.508898117579485e-08, "loss": 2.6406, "step": 47739 }, { "epoch": 2.9635607424421133, "grad_norm": 0.3070183996369988, "learning_rate": 4.4935770662052346e-08, "loss": 2.6343, "step": 47740 }, { "epoch": 2.963622819541871, "grad_norm": 0.31761592813602696, "learning_rate": 4.478282077601259e-08, "loss": 2.6478, "step": 47741 }, { "epoch": 2.963684896641629, "grad_norm": 0.31786877328052077, "learning_rate": 4.4630131518463846e-08, "loss": 2.6277, "step": 47742 }, { "epoch": 2.9637469737413866, "grad_norm": 0.30943876774897466, "learning_rate": 4.447770289021103e-08, "loss": 2.6276, "step": 47743 }, { "epoch": 2.9638090508411445, "grad_norm": 0.31215206088284464, "learning_rate": 4.4325534892042385e-08, "loss": 2.6571, "step": 47744 }, { "epoch": 2.9638711279409025, "grad_norm": 0.3180183503176561, "learning_rate": 4.417362752475174e-08, "loss": 2.6391, "step": 47745 }, { "epoch": 2.9639332050406604, "grad_norm": 0.3299756652897867, "learning_rate": 4.402198078913844e-08, "loss": 2.755, "step": 47746 }, { "epoch": 2.9639952821404183, "grad_norm": 0.3485699017179855, "learning_rate": 4.3870594685990754e-08, "loss": 2.6995, "step": 47747 }, { "epoch": 2.964057359240176, "grad_norm": 0.32255295169063997, "learning_rate": 4.371946921609138e-08, "loss": 2.6355, "step": 47748 }, { "epoch": 2.964119436339934, "grad_norm": 0.34456318341828673, "learning_rate": 4.3568604380239685e-08, "loss": 2.6495, "step": 47749 }, { "epoch": 2.964181513439692, "grad_norm": 0.3459982608960937, "learning_rate": 4.341800017921838e-08, "loss": 2.6254, "step": 47750 }, { "epoch": 2.96424359053945, "grad_norm": 0.33618063223582995, "learning_rate": 4.326765661381016e-08, "loss": 2.7483, "step": 47751 }, { "epoch": 2.964305667639208, "grad_norm": 0.3593126533164259, "learning_rate": 4.311757368480329e-08, "loss": 2.6835, "step": 47752 }, { "epoch": 2.964367744738966, "grad_norm": 0.3196123192874646, "learning_rate": 4.296775139298048e-08, "loss": 2.6945, "step": 47753 }, { "epoch": 2.9644298218387237, "grad_norm": 0.3237560716634097, "learning_rate": 4.281818973911888e-08, "loss": 2.6271, "step": 47754 }, { "epoch": 2.9644918989384816, "grad_norm": 0.32216143224213095, "learning_rate": 4.266888872400676e-08, "loss": 2.7402, "step": 47755 }, { "epoch": 2.9645539760382396, "grad_norm": 0.3357694116620575, "learning_rate": 4.251984834841571e-08, "loss": 2.6079, "step": 47756 }, { "epoch": 2.9646160531379975, "grad_norm": 0.32957706002768805, "learning_rate": 4.237106861312845e-08, "loss": 2.6462, "step": 47757 }, { "epoch": 2.9646781302377554, "grad_norm": 0.329615974803239, "learning_rate": 4.2222549518916576e-08, "loss": 2.7749, "step": 47758 }, { "epoch": 2.9647402073375133, "grad_norm": 0.33664524577155025, "learning_rate": 4.20742910665628e-08, "loss": 2.7115, "step": 47759 }, { "epoch": 2.9648022844372712, "grad_norm": 0.33673289860076977, "learning_rate": 4.192629325682762e-08, "loss": 2.6903, "step": 47760 }, { "epoch": 2.964864361537029, "grad_norm": 0.3231439585261315, "learning_rate": 4.177855609049375e-08, "loss": 2.621, "step": 47761 }, { "epoch": 2.9649264386367866, "grad_norm": 0.3338891836222118, "learning_rate": 4.163107956833279e-08, "loss": 2.6952, "step": 47762 }, { "epoch": 2.964988515736545, "grad_norm": 0.3382068543416883, "learning_rate": 4.148386369110524e-08, "loss": 2.6334, "step": 47763 }, { "epoch": 2.9650505928363025, "grad_norm": 0.3220965895797318, "learning_rate": 4.1336908459588264e-08, "loss": 2.6436, "step": 47764 }, { "epoch": 2.965112669936061, "grad_norm": 0.333111553869078, "learning_rate": 4.119021387454236e-08, "loss": 2.6753, "step": 47765 }, { "epoch": 2.9651747470358183, "grad_norm": 0.3334504706479674, "learning_rate": 4.1043779936733583e-08, "loss": 2.6723, "step": 47766 }, { "epoch": 2.9652368241355767, "grad_norm": 0.35437209351842974, "learning_rate": 4.089760664692799e-08, "loss": 2.6222, "step": 47767 }, { "epoch": 2.965298901235334, "grad_norm": 0.341052175968523, "learning_rate": 4.075169400588607e-08, "loss": 2.6288, "step": 47768 }, { "epoch": 2.965360978335092, "grad_norm": 0.32690822835414235, "learning_rate": 4.0606042014373904e-08, "loss": 2.6514, "step": 47769 }, { "epoch": 2.96542305543485, "grad_norm": 0.3382564504752687, "learning_rate": 4.0460650673146416e-08, "loss": 2.6001, "step": 47770 }, { "epoch": 2.965485132534608, "grad_norm": 0.3331781885580878, "learning_rate": 4.0315519982964124e-08, "loss": 2.7241, "step": 47771 }, { "epoch": 2.965547209634366, "grad_norm": 0.31714960773927287, "learning_rate": 4.017064994458197e-08, "loss": 2.6607, "step": 47772 }, { "epoch": 2.9656092867341237, "grad_norm": 0.34054216331308185, "learning_rate": 4.002604055876047e-08, "loss": 2.6684, "step": 47773 }, { "epoch": 2.9656713638338816, "grad_norm": 0.3493915817167805, "learning_rate": 3.988169182624901e-08, "loss": 2.7582, "step": 47774 }, { "epoch": 2.9657334409336396, "grad_norm": 0.32797116124692044, "learning_rate": 3.9737603747808104e-08, "loss": 2.6241, "step": 47775 }, { "epoch": 2.9657955180333975, "grad_norm": 0.3366574955252122, "learning_rate": 3.9593776324181596e-08, "loss": 2.6815, "step": 47776 }, { "epoch": 2.9658575951331554, "grad_norm": 0.32210967018323017, "learning_rate": 3.945020955611889e-08, "loss": 2.6362, "step": 47777 }, { "epoch": 2.9659196722329133, "grad_norm": 0.3395962965232576, "learning_rate": 3.930690344437493e-08, "loss": 2.6676, "step": 47778 }, { "epoch": 2.9659817493326712, "grad_norm": 0.3187917452945761, "learning_rate": 3.9163857989699127e-08, "loss": 2.7031, "step": 47779 }, { "epoch": 2.966043826432429, "grad_norm": 0.30796853240153976, "learning_rate": 3.902107319282977e-08, "loss": 2.544, "step": 47780 }, { "epoch": 2.966105903532187, "grad_norm": 0.34510118267121415, "learning_rate": 3.8878549054516263e-08, "loss": 2.6436, "step": 47781 }, { "epoch": 2.966167980631945, "grad_norm": 0.3069453213836452, "learning_rate": 3.873628557550246e-08, "loss": 2.6979, "step": 47782 }, { "epoch": 2.966230057731703, "grad_norm": 0.33522537070000735, "learning_rate": 3.859428275653221e-08, "loss": 2.6156, "step": 47783 }, { "epoch": 2.966292134831461, "grad_norm": 0.35534185722745376, "learning_rate": 3.845254059833825e-08, "loss": 2.7527, "step": 47784 }, { "epoch": 2.9663542119312187, "grad_norm": 0.33229196074103057, "learning_rate": 3.831105910166999e-08, "loss": 2.6244, "step": 47785 }, { "epoch": 2.9664162890309766, "grad_norm": 0.34384900531325385, "learning_rate": 3.816983826726572e-08, "loss": 2.5988, "step": 47786 }, { "epoch": 2.966478366130734, "grad_norm": 0.34792154883577864, "learning_rate": 3.802887809585265e-08, "loss": 2.6627, "step": 47787 }, { "epoch": 2.9665404432304925, "grad_norm": 0.3236051175797652, "learning_rate": 3.788817858816906e-08, "loss": 2.6556, "step": 47788 }, { "epoch": 2.96660252033025, "grad_norm": 0.3265290503276751, "learning_rate": 3.774773974495882e-08, "loss": 2.6534, "step": 47789 }, { "epoch": 2.9666645974300083, "grad_norm": 0.34122088264422834, "learning_rate": 3.760756156694356e-08, "loss": 2.6001, "step": 47790 }, { "epoch": 2.966726674529766, "grad_norm": 0.3271680797544606, "learning_rate": 3.746764405485603e-08, "loss": 2.5969, "step": 47791 }, { "epoch": 2.9667887516295237, "grad_norm": 0.3416922381860843, "learning_rate": 3.7327987209434535e-08, "loss": 2.7269, "step": 47792 }, { "epoch": 2.9668508287292816, "grad_norm": 0.3234421609920481, "learning_rate": 3.718859103139516e-08, "loss": 2.553, "step": 47793 }, { "epoch": 2.9669129058290395, "grad_norm": 0.3293424794213928, "learning_rate": 3.7049455521476204e-08, "loss": 2.667, "step": 47794 }, { "epoch": 2.9669749829287975, "grad_norm": 0.3228869719336736, "learning_rate": 3.691058068039932e-08, "loss": 2.6993, "step": 47795 }, { "epoch": 2.9670370600285554, "grad_norm": 0.32665077047734503, "learning_rate": 3.677196650888615e-08, "loss": 2.654, "step": 47796 }, { "epoch": 2.9670991371283133, "grad_norm": 0.3259863868544881, "learning_rate": 3.663361300766388e-08, "loss": 2.6885, "step": 47797 }, { "epoch": 2.967161214228071, "grad_norm": 0.32685321753518465, "learning_rate": 3.649552017745417e-08, "loss": 2.6394, "step": 47798 }, { "epoch": 2.967223291327829, "grad_norm": 0.3117348243756411, "learning_rate": 3.63576880189731e-08, "loss": 2.7295, "step": 47799 }, { "epoch": 2.967285368427587, "grad_norm": 0.3433203396182473, "learning_rate": 3.6220116532947876e-08, "loss": 2.6857, "step": 47800 }, { "epoch": 2.967347445527345, "grad_norm": 0.33756097799027596, "learning_rate": 3.608280572008904e-08, "loss": 2.736, "step": 47801 }, { "epoch": 2.967409522627103, "grad_norm": 0.3358536992788075, "learning_rate": 3.5945755581112685e-08, "loss": 2.6399, "step": 47802 }, { "epoch": 2.967471599726861, "grad_norm": 0.3283521229471581, "learning_rate": 3.580896611674045e-08, "loss": 2.6926, "step": 47803 }, { "epoch": 2.9675336768266187, "grad_norm": 0.3304437722580734, "learning_rate": 3.567243732768288e-08, "loss": 2.6934, "step": 47804 }, { "epoch": 2.9675957539263766, "grad_norm": 0.3282274926285268, "learning_rate": 3.553616921465053e-08, "loss": 2.6145, "step": 47805 }, { "epoch": 2.9676578310261346, "grad_norm": 0.3193538205800253, "learning_rate": 3.540016177835392e-08, "loss": 2.6731, "step": 47806 }, { "epoch": 2.9677199081258925, "grad_norm": 0.3317313466710282, "learning_rate": 3.5264415019503616e-08, "loss": 2.6869, "step": 47807 }, { "epoch": 2.9677819852256504, "grad_norm": 0.33385026481161434, "learning_rate": 3.512892893880459e-08, "loss": 2.6673, "step": 47808 }, { "epoch": 2.9678440623254083, "grad_norm": 0.32364508421980726, "learning_rate": 3.4993703536972954e-08, "loss": 2.6012, "step": 47809 }, { "epoch": 2.967906139425166, "grad_norm": 0.3480795308729551, "learning_rate": 3.4858738814708136e-08, "loss": 2.6503, "step": 47810 }, { "epoch": 2.967968216524924, "grad_norm": 0.34369280729481383, "learning_rate": 3.472403477271513e-08, "loss": 2.7231, "step": 47811 }, { "epoch": 2.9680302936246816, "grad_norm": 0.3350984163337381, "learning_rate": 3.4589591411693375e-08, "loss": 2.6674, "step": 47812 }, { "epoch": 2.96809237072444, "grad_norm": 0.324520946902984, "learning_rate": 3.4455408732347873e-08, "loss": 2.6306, "step": 47813 }, { "epoch": 2.9681544478241975, "grad_norm": 0.3167159601586449, "learning_rate": 3.43214867353836e-08, "loss": 2.6372, "step": 47814 }, { "epoch": 2.968216524923956, "grad_norm": 0.3207162313071553, "learning_rate": 3.4187825421488904e-08, "loss": 2.6949, "step": 47815 }, { "epoch": 2.9682786020237133, "grad_norm": 0.33788400985328154, "learning_rate": 3.405442479136323e-08, "loss": 2.7269, "step": 47816 }, { "epoch": 2.968340679123471, "grad_norm": 0.3315625753771151, "learning_rate": 3.392128484571156e-08, "loss": 2.6205, "step": 47817 }, { "epoch": 2.968402756223229, "grad_norm": 0.3215055808655263, "learning_rate": 3.3788405585216676e-08, "loss": 2.5749, "step": 47818 }, { "epoch": 2.968464833322987, "grad_norm": 0.33661285425720194, "learning_rate": 3.3655787010578034e-08, "loss": 2.6803, "step": 47819 }, { "epoch": 2.968526910422745, "grad_norm": 0.336458589368709, "learning_rate": 3.352342912248952e-08, "loss": 2.6652, "step": 47820 }, { "epoch": 2.968588987522503, "grad_norm": 0.3272177893163929, "learning_rate": 3.339133192163946e-08, "loss": 2.7708, "step": 47821 }, { "epoch": 2.968651064622261, "grad_norm": 0.3563540825299644, "learning_rate": 3.32594954087162e-08, "loss": 2.8334, "step": 47822 }, { "epoch": 2.9687131417220187, "grad_norm": 0.3280280514953919, "learning_rate": 3.3127919584402535e-08, "loss": 2.6204, "step": 47823 }, { "epoch": 2.9687752188217766, "grad_norm": 0.3381066926946474, "learning_rate": 3.2996604449397897e-08, "loss": 2.7626, "step": 47824 }, { "epoch": 2.9688372959215346, "grad_norm": 0.3464099748245146, "learning_rate": 3.2865550004373966e-08, "loss": 2.6573, "step": 47825 }, { "epoch": 2.9688993730212925, "grad_norm": 0.34112590872355947, "learning_rate": 3.273475625002465e-08, "loss": 2.6923, "step": 47826 }, { "epoch": 2.9689614501210504, "grad_norm": 0.33417016154568013, "learning_rate": 3.260422318702716e-08, "loss": 2.5837, "step": 47827 }, { "epoch": 2.9690235272208083, "grad_norm": 0.3356302097917832, "learning_rate": 3.2473950816064305e-08, "loss": 2.6465, "step": 47828 }, { "epoch": 2.9690856043205662, "grad_norm": 0.34564522080482807, "learning_rate": 3.2343939137813305e-08, "loss": 2.6534, "step": 47829 }, { "epoch": 2.969147681420324, "grad_norm": 0.3127083441492664, "learning_rate": 3.221418815295696e-08, "loss": 2.6487, "step": 47830 }, { "epoch": 2.969209758520082, "grad_norm": 0.34188172513355153, "learning_rate": 3.208469786216694e-08, "loss": 2.6271, "step": 47831 }, { "epoch": 2.96927183561984, "grad_norm": 0.3369696419247803, "learning_rate": 3.19554682661205e-08, "loss": 2.6305, "step": 47832 }, { "epoch": 2.969333912719598, "grad_norm": 0.3419152518048734, "learning_rate": 3.182649936548932e-08, "loss": 2.6723, "step": 47833 }, { "epoch": 2.969395989819356, "grad_norm": 0.3245446968139587, "learning_rate": 3.169779116095617e-08, "loss": 2.6227, "step": 47834 }, { "epoch": 2.9694580669191133, "grad_norm": 0.32492056114186124, "learning_rate": 3.156934365318165e-08, "loss": 2.5818, "step": 47835 }, { "epoch": 2.9695201440188717, "grad_norm": 0.3546133123441032, "learning_rate": 3.1441156842842986e-08, "loss": 2.6653, "step": 47836 }, { "epoch": 2.969582221118629, "grad_norm": 0.32859838488797727, "learning_rate": 3.1313230730600775e-08, "loss": 2.6955, "step": 47837 }, { "epoch": 2.9696442982183875, "grad_norm": 0.33256745518242364, "learning_rate": 3.1185565317132235e-08, "loss": 2.6525, "step": 47838 }, { "epoch": 2.969706375318145, "grad_norm": 0.3149571969265418, "learning_rate": 3.1058160603097964e-08, "loss": 2.7001, "step": 47839 }, { "epoch": 2.969768452417903, "grad_norm": 0.331010674895853, "learning_rate": 3.093101658916409e-08, "loss": 2.6803, "step": 47840 }, { "epoch": 2.969830529517661, "grad_norm": 0.3299643416357268, "learning_rate": 3.080413327599674e-08, "loss": 2.6541, "step": 47841 }, { "epoch": 2.9698926066174187, "grad_norm": 0.32944211686803104, "learning_rate": 3.0677510664250954e-08, "loss": 2.734, "step": 47842 }, { "epoch": 2.9699546837171766, "grad_norm": 0.33063709173824485, "learning_rate": 3.055114875458731e-08, "loss": 2.6559, "step": 47843 }, { "epoch": 2.9700167608169346, "grad_norm": 0.3254845166610989, "learning_rate": 3.0425047547677496e-08, "loss": 2.7, "step": 47844 }, { "epoch": 2.9700788379166925, "grad_norm": 0.33226045663250575, "learning_rate": 3.0299207044165445e-08, "loss": 2.647, "step": 47845 }, { "epoch": 2.9701409150164504, "grad_norm": 0.3334583914628543, "learning_rate": 3.0173627244711736e-08, "loss": 2.7236, "step": 47846 }, { "epoch": 2.9702029921162083, "grad_norm": 0.33123267011474966, "learning_rate": 3.004830814997694e-08, "loss": 2.6474, "step": 47847 }, { "epoch": 2.9702650692159662, "grad_norm": 0.32651537094256017, "learning_rate": 2.992324976060501e-08, "loss": 2.5258, "step": 47848 }, { "epoch": 2.970327146315724, "grad_norm": 0.3259681705151171, "learning_rate": 2.9798452077262062e-08, "loss": 2.5841, "step": 47849 }, { "epoch": 2.970389223415482, "grad_norm": 0.3231374565173179, "learning_rate": 2.967391510058648e-08, "loss": 2.6027, "step": 47850 }, { "epoch": 2.97045130051524, "grad_norm": 0.3255728171484761, "learning_rate": 2.9549638831233295e-08, "loss": 2.5678, "step": 47851 }, { "epoch": 2.970513377614998, "grad_norm": 0.3387262232554598, "learning_rate": 2.9425623269846437e-08, "loss": 2.6705, "step": 47852 }, { "epoch": 2.970575454714756, "grad_norm": 0.33419203297894035, "learning_rate": 2.9301868417080937e-08, "loss": 2.6561, "step": 47853 }, { "epoch": 2.9706375318145137, "grad_norm": 0.3355669373097921, "learning_rate": 2.917837427358072e-08, "loss": 2.6405, "step": 47854 }, { "epoch": 2.9706996089142716, "grad_norm": 0.32383713611908194, "learning_rate": 2.9055140839984173e-08, "loss": 2.7236, "step": 47855 }, { "epoch": 2.9707616860140296, "grad_norm": 0.3383343269520687, "learning_rate": 2.893216811694077e-08, "loss": 2.6356, "step": 47856 }, { "epoch": 2.9708237631137875, "grad_norm": 0.3429286746548831, "learning_rate": 2.8809456105083344e-08, "loss": 2.7112, "step": 47857 }, { "epoch": 2.970885840213545, "grad_norm": 0.3352251039928588, "learning_rate": 2.868700480506692e-08, "loss": 2.6963, "step": 47858 }, { "epoch": 2.9709479173133033, "grad_norm": 0.3244210152337449, "learning_rate": 2.8564814217513225e-08, "loss": 2.6285, "step": 47859 }, { "epoch": 2.971009994413061, "grad_norm": 0.35760622533792163, "learning_rate": 2.844288434307174e-08, "loss": 2.6118, "step": 47860 }, { "epoch": 2.971072071512819, "grad_norm": 0.33675399516156396, "learning_rate": 2.8321215182375294e-08, "loss": 2.6869, "step": 47861 }, { "epoch": 2.9711341486125766, "grad_norm": 0.3353953311286981, "learning_rate": 2.819980673605671e-08, "loss": 2.6782, "step": 47862 }, { "epoch": 2.971196225712335, "grad_norm": 0.34751909743805975, "learning_rate": 2.807865900474882e-08, "loss": 2.6819, "step": 47863 }, { "epoch": 2.9712583028120925, "grad_norm": 0.3293899608196589, "learning_rate": 2.795777198909e-08, "loss": 2.7255, "step": 47864 }, { "epoch": 2.9713203799118504, "grad_norm": 0.3238830897087613, "learning_rate": 2.783714568970197e-08, "loss": 2.6271, "step": 47865 }, { "epoch": 2.9713824570116083, "grad_norm": 0.31794661695600485, "learning_rate": 2.7716780107223117e-08, "loss": 2.6275, "step": 47866 }, { "epoch": 2.971444534111366, "grad_norm": 0.3383831998214141, "learning_rate": 2.759667524227516e-08, "loss": 2.6725, "step": 47867 }, { "epoch": 2.971506611211124, "grad_norm": 0.31921333035775845, "learning_rate": 2.7476831095485376e-08, "loss": 2.5764, "step": 47868 }, { "epoch": 2.971568688310882, "grad_norm": 0.34092223384115705, "learning_rate": 2.7357247667481045e-08, "loss": 2.6076, "step": 47869 }, { "epoch": 2.97163076541064, "grad_norm": 0.3256907197781769, "learning_rate": 2.7237924958883887e-08, "loss": 2.6831, "step": 47870 }, { "epoch": 2.971692842510398, "grad_norm": 0.32973244301964283, "learning_rate": 2.711886297032118e-08, "loss": 2.6206, "step": 47871 }, { "epoch": 2.971754919610156, "grad_norm": 0.3281367731252915, "learning_rate": 2.7000061702409096e-08, "loss": 2.7638, "step": 47872 }, { "epoch": 2.9718169967099137, "grad_norm": 0.3341883098085236, "learning_rate": 2.6881521155769364e-08, "loss": 2.6396, "step": 47873 }, { "epoch": 2.9718790738096716, "grad_norm": 0.32718720846882743, "learning_rate": 2.6763241331018153e-08, "loss": 2.6069, "step": 47874 }, { "epoch": 2.9719411509094296, "grad_norm": 0.32548037341125974, "learning_rate": 2.664522222877719e-08, "loss": 2.7219, "step": 47875 }, { "epoch": 2.9720032280091875, "grad_norm": 0.34436255842743757, "learning_rate": 2.6527463849657097e-08, "loss": 2.7206, "step": 47876 }, { "epoch": 2.9720653051089454, "grad_norm": 0.3309744566075261, "learning_rate": 2.64099661942796e-08, "loss": 2.6871, "step": 47877 }, { "epoch": 2.9721273822087033, "grad_norm": 0.34577640467996845, "learning_rate": 2.6292729263249772e-08, "loss": 2.6815, "step": 47878 }, { "epoch": 2.9721894593084612, "grad_norm": 0.3192285246246466, "learning_rate": 2.617575305717823e-08, "loss": 2.6581, "step": 47879 }, { "epoch": 2.972251536408219, "grad_norm": 0.33254545449640566, "learning_rate": 2.6059037576681157e-08, "loss": 2.5933, "step": 47880 }, { "epoch": 2.972313613507977, "grad_norm": 0.3340650539104304, "learning_rate": 2.5942582822369166e-08, "loss": 2.7527, "step": 47881 }, { "epoch": 2.972375690607735, "grad_norm": 0.32923773147329816, "learning_rate": 2.5826388794847334e-08, "loss": 2.726, "step": 47882 }, { "epoch": 2.9724377677074925, "grad_norm": 0.32847050233318537, "learning_rate": 2.571045549471518e-08, "loss": 2.6918, "step": 47883 }, { "epoch": 2.972499844807251, "grad_norm": 0.33739506759195487, "learning_rate": 2.5594782922588877e-08, "loss": 2.5842, "step": 47884 }, { "epoch": 2.9725619219070083, "grad_norm": 0.3297256190061995, "learning_rate": 2.5479371079062398e-08, "loss": 2.6185, "step": 47885 }, { "epoch": 2.9726239990067667, "grad_norm": 0.3340128947791192, "learning_rate": 2.536421996474081e-08, "loss": 2.7124, "step": 47886 }, { "epoch": 2.972686076106524, "grad_norm": 0.31400117322830134, "learning_rate": 2.5249329580229185e-08, "loss": 2.5495, "step": 47887 }, { "epoch": 2.972748153206282, "grad_norm": 0.33352664344013266, "learning_rate": 2.5134699926121495e-08, "loss": 2.5612, "step": 47888 }, { "epoch": 2.97281023030604, "grad_norm": 0.32803559220514444, "learning_rate": 2.5020331003022814e-08, "loss": 2.688, "step": 47889 }, { "epoch": 2.972872307405798, "grad_norm": 0.33528864097502054, "learning_rate": 2.4906222811516e-08, "loss": 2.69, "step": 47890 }, { "epoch": 2.972934384505556, "grad_norm": 0.31289020257318867, "learning_rate": 2.4792375352211683e-08, "loss": 2.7114, "step": 47891 }, { "epoch": 2.9729964616053137, "grad_norm": 0.31767466111312564, "learning_rate": 2.467878862569828e-08, "loss": 2.5447, "step": 47892 }, { "epoch": 2.9730585387050716, "grad_norm": 0.3399451422438106, "learning_rate": 2.4565462632564205e-08, "loss": 2.7424, "step": 47893 }, { "epoch": 2.9731206158048296, "grad_norm": 0.3262479316528936, "learning_rate": 2.4452397373403434e-08, "loss": 2.6362, "step": 47894 }, { "epoch": 2.9731826929045875, "grad_norm": 0.33891281788126276, "learning_rate": 2.4339592848809934e-08, "loss": 2.8339, "step": 47895 }, { "epoch": 2.9732447700043454, "grad_norm": 0.33914118526177833, "learning_rate": 2.422704905936657e-08, "loss": 2.6665, "step": 47896 }, { "epoch": 2.9733068471041033, "grad_norm": 0.3370834256613329, "learning_rate": 2.411476600566176e-08, "loss": 2.6323, "step": 47897 }, { "epoch": 2.9733689242038612, "grad_norm": 0.33694628132999677, "learning_rate": 2.4002743688283926e-08, "loss": 2.6474, "step": 47898 }, { "epoch": 2.973431001303619, "grad_norm": 0.34250977420434353, "learning_rate": 2.3890982107815928e-08, "loss": 2.5637, "step": 47899 }, { "epoch": 2.973493078403377, "grad_norm": 0.3488820785028517, "learning_rate": 2.3779481264840642e-08, "loss": 2.7076, "step": 47900 }, { "epoch": 2.973555155503135, "grad_norm": 0.31963080461860827, "learning_rate": 2.3668241159940928e-08, "loss": 2.5928, "step": 47901 }, { "epoch": 2.973617232602893, "grad_norm": 0.30963329071692525, "learning_rate": 2.355726179369411e-08, "loss": 2.5639, "step": 47902 }, { "epoch": 2.973679309702651, "grad_norm": 0.3314171441359066, "learning_rate": 2.3446543166683042e-08, "loss": 2.6271, "step": 47903 }, { "epoch": 2.9737413868024087, "grad_norm": 0.316062269316716, "learning_rate": 2.3336085279485054e-08, "loss": 2.6209, "step": 47904 }, { "epoch": 2.9738034639021667, "grad_norm": 0.32514631343252165, "learning_rate": 2.3225888132671902e-08, "loss": 2.6479, "step": 47905 }, { "epoch": 2.973865541001924, "grad_norm": 0.34261187797088255, "learning_rate": 2.3115951726820907e-08, "loss": 2.6951, "step": 47906 }, { "epoch": 2.9739276181016825, "grad_norm": 0.320240885009261, "learning_rate": 2.300627606250938e-08, "loss": 2.6848, "step": 47907 }, { "epoch": 2.97398969520144, "grad_norm": 0.33524957349040346, "learning_rate": 2.289686114030354e-08, "loss": 2.6288, "step": 47908 }, { "epoch": 2.9740517723011983, "grad_norm": 0.3166151652158467, "learning_rate": 2.2787706960780698e-08, "loss": 2.753, "step": 47909 }, { "epoch": 2.974113849400956, "grad_norm": 0.33570805188017644, "learning_rate": 2.267881352450707e-08, "loss": 2.6146, "step": 47910 }, { "epoch": 2.974175926500714, "grad_norm": 0.31680214215473135, "learning_rate": 2.257018083204332e-08, "loss": 2.5932, "step": 47911 }, { "epoch": 2.9742380036004716, "grad_norm": 0.31331651826812357, "learning_rate": 2.2461808883972313e-08, "loss": 2.6747, "step": 47912 }, { "epoch": 2.9743000807002296, "grad_norm": 0.34988608677856586, "learning_rate": 2.235369768084361e-08, "loss": 2.6563, "step": 47913 }, { "epoch": 2.9743621577999875, "grad_norm": 0.3216654303610141, "learning_rate": 2.2245847223234528e-08, "loss": 2.6619, "step": 47914 }, { "epoch": 2.9744242348997454, "grad_norm": 0.3340407318395421, "learning_rate": 2.2138257511694627e-08, "loss": 2.7333, "step": 47915 }, { "epoch": 2.9744863119995033, "grad_norm": 0.30950015984979423, "learning_rate": 2.203092854679567e-08, "loss": 2.6875, "step": 47916 }, { "epoch": 2.9745483890992612, "grad_norm": 0.343709889152481, "learning_rate": 2.192386032909277e-08, "loss": 2.624, "step": 47917 }, { "epoch": 2.974610466199019, "grad_norm": 0.3454413419606141, "learning_rate": 2.1817052859146593e-08, "loss": 2.6872, "step": 47918 }, { "epoch": 2.974672543298777, "grad_norm": 0.3194380430718129, "learning_rate": 2.1710506137512242e-08, "loss": 2.5713, "step": 47919 }, { "epoch": 2.974734620398535, "grad_norm": 0.33836191858583203, "learning_rate": 2.160422016475039e-08, "loss": 2.6627, "step": 47920 }, { "epoch": 2.974796697498293, "grad_norm": 0.3271301947674276, "learning_rate": 2.1498194941405038e-08, "loss": 2.6454, "step": 47921 }, { "epoch": 2.974858774598051, "grad_norm": 0.3357219115165836, "learning_rate": 2.13924304680424e-08, "loss": 2.5771, "step": 47922 }, { "epoch": 2.9749208516978087, "grad_norm": 0.33581811424743707, "learning_rate": 2.128692674520649e-08, "loss": 2.7353, "step": 47923 }, { "epoch": 2.9749829287975667, "grad_norm": 0.3411843724600969, "learning_rate": 2.118168377345242e-08, "loss": 2.6077, "step": 47924 }, { "epoch": 2.9750450058973246, "grad_norm": 0.32709670155940057, "learning_rate": 2.1076701553318646e-08, "loss": 2.6582, "step": 47925 }, { "epoch": 2.9751070829970825, "grad_norm": 0.3421072395468874, "learning_rate": 2.0971980085365828e-08, "loss": 2.6766, "step": 47926 }, { "epoch": 2.9751691600968404, "grad_norm": 0.32932725406493013, "learning_rate": 2.0867519370137977e-08, "loss": 2.6703, "step": 47927 }, { "epoch": 2.9752312371965983, "grad_norm": 0.3151522618099573, "learning_rate": 2.0763319408173555e-08, "loss": 2.686, "step": 47928 }, { "epoch": 2.9752933142963562, "grad_norm": 0.33425468526279845, "learning_rate": 2.0659380200016565e-08, "loss": 2.6488, "step": 47929 }, { "epoch": 2.975355391396114, "grad_norm": 0.347094027643358, "learning_rate": 2.055570174621657e-08, "loss": 2.6642, "step": 47930 }, { "epoch": 2.9754174684958716, "grad_norm": 0.3293076161615926, "learning_rate": 2.0452284047312032e-08, "loss": 2.606, "step": 47931 }, { "epoch": 2.97547954559563, "grad_norm": 0.3253450936544444, "learning_rate": 2.0349127103835853e-08, "loss": 2.6545, "step": 47932 }, { "epoch": 2.9755416226953875, "grad_norm": 0.3393082290812053, "learning_rate": 2.0246230916332043e-08, "loss": 2.6731, "step": 47933 }, { "epoch": 2.975603699795146, "grad_norm": 0.32759800098036934, "learning_rate": 2.014359548533906e-08, "loss": 2.7028, "step": 47934 }, { "epoch": 2.9756657768949033, "grad_norm": 0.3207308897761144, "learning_rate": 2.0041220811389815e-08, "loss": 2.589, "step": 47935 }, { "epoch": 2.975727853994661, "grad_norm": 0.3256835365581319, "learning_rate": 1.9939106895017213e-08, "loss": 2.6023, "step": 47936 }, { "epoch": 2.975789931094419, "grad_norm": 0.31680563457152067, "learning_rate": 1.983725373675971e-08, "loss": 2.6985, "step": 47937 }, { "epoch": 2.975852008194177, "grad_norm": 0.32649793716745834, "learning_rate": 1.973566133713911e-08, "loss": 2.6664, "step": 47938 }, { "epoch": 2.975914085293935, "grad_norm": 0.317963840062429, "learning_rate": 1.9634329696693875e-08, "loss": 2.6507, "step": 47939 }, { "epoch": 2.975976162393693, "grad_norm": 0.3148444566174474, "learning_rate": 1.953325881594581e-08, "loss": 2.7154, "step": 47940 }, { "epoch": 2.976038239493451, "grad_norm": 0.33846652835472196, "learning_rate": 1.943244869542782e-08, "loss": 2.678, "step": 47941 }, { "epoch": 2.9761003165932087, "grad_norm": 0.32849633685265794, "learning_rate": 1.9331899335661708e-08, "loss": 2.6931, "step": 47942 }, { "epoch": 2.9761623936929666, "grad_norm": 0.33810782755611296, "learning_rate": 1.9231610737180382e-08, "loss": 2.6844, "step": 47943 }, { "epoch": 2.9762244707927246, "grad_norm": 0.3516427584271738, "learning_rate": 1.9131582900494548e-08, "loss": 2.633, "step": 47944 }, { "epoch": 2.9762865478924825, "grad_norm": 0.32027476949143774, "learning_rate": 1.9031815826137113e-08, "loss": 2.6977, "step": 47945 }, { "epoch": 2.9763486249922404, "grad_norm": 0.34533199496927347, "learning_rate": 1.8932309514624323e-08, "loss": 2.754, "step": 47946 }, { "epoch": 2.9764107020919983, "grad_norm": 0.32467994290131186, "learning_rate": 1.8833063966472442e-08, "loss": 2.6898, "step": 47947 }, { "epoch": 2.9764727791917562, "grad_norm": 0.326297964063472, "learning_rate": 1.8734079182203268e-08, "loss": 2.6718, "step": 47948 }, { "epoch": 2.976534856291514, "grad_norm": 0.3294271222844559, "learning_rate": 1.8635355162327507e-08, "loss": 2.6676, "step": 47949 }, { "epoch": 2.976596933391272, "grad_norm": 0.3165478095348649, "learning_rate": 1.853689190737251e-08, "loss": 2.5749, "step": 47950 }, { "epoch": 2.97665901049103, "grad_norm": 0.3327128538761363, "learning_rate": 1.8438689417837886e-08, "loss": 2.6577, "step": 47951 }, { "epoch": 2.976721087590788, "grad_norm": 0.32716500539247767, "learning_rate": 1.834074769423988e-08, "loss": 2.6212, "step": 47952 }, { "epoch": 2.976783164690546, "grad_norm": 0.32365140337776055, "learning_rate": 1.824306673709475e-08, "loss": 2.6824, "step": 47953 }, { "epoch": 2.9768452417903033, "grad_norm": 0.3375462005923519, "learning_rate": 1.81456465469132e-08, "loss": 2.6948, "step": 47954 }, { "epoch": 2.9769073188900617, "grad_norm": 0.3330965343430189, "learning_rate": 1.8048487124194823e-08, "loss": 2.7236, "step": 47955 }, { "epoch": 2.976969395989819, "grad_norm": 0.31588748507701747, "learning_rate": 1.795158846945033e-08, "loss": 2.631, "step": 47956 }, { "epoch": 2.9770314730895775, "grad_norm": 0.309965128590854, "learning_rate": 1.785495058319042e-08, "loss": 2.6372, "step": 47957 }, { "epoch": 2.977093550189335, "grad_norm": 0.32105755799232466, "learning_rate": 1.7758573465909146e-08, "loss": 2.6502, "step": 47958 }, { "epoch": 2.977155627289093, "grad_norm": 0.31654522045608235, "learning_rate": 1.7662457118117202e-08, "loss": 2.6293, "step": 47959 }, { "epoch": 2.977217704388851, "grad_norm": 0.3271751392911619, "learning_rate": 1.756660154031975e-08, "loss": 2.628, "step": 47960 }, { "epoch": 2.9772797814886087, "grad_norm": 0.3392347711217561, "learning_rate": 1.747100673300528e-08, "loss": 2.6804, "step": 47961 }, { "epoch": 2.9773418585883666, "grad_norm": 0.31846809870986736, "learning_rate": 1.7375672696673397e-08, "loss": 2.6863, "step": 47962 }, { "epoch": 2.9774039356881246, "grad_norm": 0.3254671421276855, "learning_rate": 1.72805994318348e-08, "loss": 2.6779, "step": 47963 }, { "epoch": 2.9774660127878825, "grad_norm": 0.31979394434553704, "learning_rate": 1.7185786938972438e-08, "loss": 2.6085, "step": 47964 }, { "epoch": 2.9775280898876404, "grad_norm": 0.33500000528435203, "learning_rate": 1.709123521858591e-08, "loss": 2.6534, "step": 47965 }, { "epoch": 2.9775901669873983, "grad_norm": 0.30508720638420284, "learning_rate": 1.699694427116927e-08, "loss": 2.6576, "step": 47966 }, { "epoch": 2.9776522440871562, "grad_norm": 0.34348806241585417, "learning_rate": 1.690291409721656e-08, "loss": 2.6921, "step": 47967 }, { "epoch": 2.977714321186914, "grad_norm": 0.3293470717755858, "learning_rate": 1.6809144697210734e-08, "loss": 2.7147, "step": 47968 }, { "epoch": 2.977776398286672, "grad_norm": 0.3376162633383382, "learning_rate": 1.6715636071645834e-08, "loss": 2.7343, "step": 47969 }, { "epoch": 2.97783847538643, "grad_norm": 0.32955145718300544, "learning_rate": 1.6622388221010364e-08, "loss": 2.7514, "step": 47970 }, { "epoch": 2.977900552486188, "grad_norm": 0.33023691590168436, "learning_rate": 1.6529401145792824e-08, "loss": 2.652, "step": 47971 }, { "epoch": 2.977962629585946, "grad_norm": 0.3312597624221882, "learning_rate": 1.6436674846470602e-08, "loss": 2.6324, "step": 47972 }, { "epoch": 2.9780247066857037, "grad_norm": 0.34140816519961686, "learning_rate": 1.63442093235322e-08, "loss": 2.6757, "step": 47973 }, { "epoch": 2.9780867837854617, "grad_norm": 0.32299774700447864, "learning_rate": 1.6252004577460566e-08, "loss": 2.5542, "step": 47974 }, { "epoch": 2.9781488608852196, "grad_norm": 0.3336959731381443, "learning_rate": 1.6160060608738647e-08, "loss": 2.6374, "step": 47975 }, { "epoch": 2.9782109379849775, "grad_norm": 0.3464408481614352, "learning_rate": 1.6068377417843837e-08, "loss": 2.7047, "step": 47976 }, { "epoch": 2.978273015084735, "grad_norm": 0.32619702362144576, "learning_rate": 1.5976955005253537e-08, "loss": 2.6697, "step": 47977 }, { "epoch": 2.9783350921844933, "grad_norm": 0.33107996031131715, "learning_rate": 1.5885793371439582e-08, "loss": 2.6543, "step": 47978 }, { "epoch": 2.978397169284251, "grad_norm": 0.347036688296986, "learning_rate": 1.579489251689048e-08, "loss": 2.677, "step": 47979 }, { "epoch": 2.978459246384009, "grad_norm": 0.3270244648170743, "learning_rate": 1.570425244207252e-08, "loss": 2.7071, "step": 47980 }, { "epoch": 2.9785213234837666, "grad_norm": 0.3324354292261269, "learning_rate": 1.5613873147457546e-08, "loss": 2.652, "step": 47981 }, { "epoch": 2.978583400583525, "grad_norm": 0.32586106893702305, "learning_rate": 1.552375463352296e-08, "loss": 2.7092, "step": 47982 }, { "epoch": 2.9786454776832825, "grad_norm": 0.32185295733186103, "learning_rate": 1.5433896900729494e-08, "loss": 2.713, "step": 47983 }, { "epoch": 2.9787075547830404, "grad_norm": 0.3224354383251503, "learning_rate": 1.5344299949554554e-08, "loss": 2.678, "step": 47984 }, { "epoch": 2.9787696318827983, "grad_norm": 0.3351836173348195, "learning_rate": 1.5254963780464427e-08, "loss": 2.6546, "step": 47985 }, { "epoch": 2.9788317089825562, "grad_norm": 0.32453205872583246, "learning_rate": 1.5165888393919858e-08, "loss": 2.6514, "step": 47986 }, { "epoch": 2.978893786082314, "grad_norm": 0.3284887046534069, "learning_rate": 1.507707379039269e-08, "loss": 2.5712, "step": 47987 }, { "epoch": 2.978955863182072, "grad_norm": 0.33053574074710257, "learning_rate": 1.4988519970338122e-08, "loss": 2.6865, "step": 47988 }, { "epoch": 2.97901794028183, "grad_norm": 0.326099936038657, "learning_rate": 1.490022693422799e-08, "loss": 2.7493, "step": 47989 }, { "epoch": 2.979080017381588, "grad_norm": 0.33696022270328296, "learning_rate": 1.4812194682511937e-08, "loss": 2.687, "step": 47990 }, { "epoch": 2.979142094481346, "grad_norm": 0.31410330512060713, "learning_rate": 1.4724423215661809e-08, "loss": 2.6378, "step": 47991 }, { "epoch": 2.9792041715811037, "grad_norm": 0.32732969358577235, "learning_rate": 1.4636912534121694e-08, "loss": 2.6636, "step": 47992 }, { "epoch": 2.9792662486808617, "grad_norm": 0.32082834807564536, "learning_rate": 1.4549662638357887e-08, "loss": 2.6709, "step": 47993 }, { "epoch": 2.9793283257806196, "grad_norm": 0.332637222681184, "learning_rate": 1.4462673528825577e-08, "loss": 2.6927, "step": 47994 }, { "epoch": 2.9793904028803775, "grad_norm": 0.32828518953642183, "learning_rate": 1.4375945205974406e-08, "loss": 2.6897, "step": 47995 }, { "epoch": 2.9794524799801354, "grad_norm": 0.31799781488396583, "learning_rate": 1.4289477670254014e-08, "loss": 2.6971, "step": 47996 }, { "epoch": 2.9795145570798933, "grad_norm": 0.342229101186636, "learning_rate": 1.4203270922125144e-08, "loss": 2.7787, "step": 47997 }, { "epoch": 2.9795766341796512, "grad_norm": 0.346719529992112, "learning_rate": 1.4117324962031886e-08, "loss": 2.6463, "step": 47998 }, { "epoch": 2.979638711279409, "grad_norm": 0.3276222328133213, "learning_rate": 1.4031639790423878e-08, "loss": 2.5921, "step": 47999 }, { "epoch": 2.979700788379167, "grad_norm": 0.33265697758685986, "learning_rate": 1.3946215407745212e-08, "loss": 2.7142, "step": 48000 }, { "epoch": 2.979762865478925, "grad_norm": 0.31654602071905813, "learning_rate": 1.3861051814439974e-08, "loss": 2.6526, "step": 48001 }, { "epoch": 2.9798249425786825, "grad_norm": 0.3404244142912926, "learning_rate": 1.3776149010963357e-08, "loss": 2.6691, "step": 48002 }, { "epoch": 2.979887019678441, "grad_norm": 0.3286587030497496, "learning_rate": 1.369150699774835e-08, "loss": 2.6739, "step": 48003 }, { "epoch": 2.9799490967781983, "grad_norm": 0.33282824958106594, "learning_rate": 1.3607125775239038e-08, "loss": 2.6408, "step": 48004 }, { "epoch": 2.9800111738779567, "grad_norm": 0.3109274517525189, "learning_rate": 1.3523005343873962e-08, "loss": 2.6526, "step": 48005 }, { "epoch": 2.980073250977714, "grad_norm": 0.3253332080534657, "learning_rate": 1.3439145704097211e-08, "loss": 2.7407, "step": 48006 }, { "epoch": 2.980135328077472, "grad_norm": 0.3303514736933666, "learning_rate": 1.3355546856341771e-08, "loss": 2.6259, "step": 48007 }, { "epoch": 2.98019740517723, "grad_norm": 0.31476712394806233, "learning_rate": 1.3272208801040631e-08, "loss": 2.6303, "step": 48008 }, { "epoch": 2.980259482276988, "grad_norm": 0.31734150544372475, "learning_rate": 1.3189131538637878e-08, "loss": 2.6416, "step": 48009 }, { "epoch": 2.980321559376746, "grad_norm": 0.31415507010391347, "learning_rate": 1.3106315069560947e-08, "loss": 2.7207, "step": 48010 }, { "epoch": 2.9803836364765037, "grad_norm": 0.32727778141030756, "learning_rate": 1.302375939424838e-08, "loss": 2.5814, "step": 48011 }, { "epoch": 2.9804457135762616, "grad_norm": 0.33487898377330394, "learning_rate": 1.2941464513116508e-08, "loss": 2.6664, "step": 48012 }, { "epoch": 2.9805077906760196, "grad_norm": 0.33284414302139065, "learning_rate": 1.285943042660942e-08, "loss": 2.6104, "step": 48013 }, { "epoch": 2.9805698677757775, "grad_norm": 0.3280951054896862, "learning_rate": 1.2777657135149002e-08, "loss": 2.6948, "step": 48014 }, { "epoch": 2.9806319448755354, "grad_norm": 0.34361397696265095, "learning_rate": 1.2696144639162689e-08, "loss": 2.7129, "step": 48015 }, { "epoch": 2.9806940219752933, "grad_norm": 0.3280234974587097, "learning_rate": 1.2614892939072364e-08, "loss": 2.7314, "step": 48016 }, { "epoch": 2.9807560990750512, "grad_norm": 0.3165220590274476, "learning_rate": 1.2533902035311019e-08, "loss": 2.6318, "step": 48017 }, { "epoch": 2.980818176174809, "grad_norm": 0.31409174132983375, "learning_rate": 1.245317192828943e-08, "loss": 2.5421, "step": 48018 }, { "epoch": 2.980880253274567, "grad_norm": 0.3163477761196026, "learning_rate": 1.2372702618435039e-08, "loss": 2.6457, "step": 48019 }, { "epoch": 2.980942330374325, "grad_norm": 0.3354692189391456, "learning_rate": 1.2292494106169727e-08, "loss": 2.6901, "step": 48020 }, { "epoch": 2.981004407474083, "grad_norm": 0.32426151315341795, "learning_rate": 1.2212546391909829e-08, "loss": 2.6538, "step": 48021 }, { "epoch": 2.981066484573841, "grad_norm": 0.33640065535355074, "learning_rate": 1.2132859476071678e-08, "loss": 2.7034, "step": 48022 }, { "epoch": 2.9811285616735987, "grad_norm": 0.34598508209526996, "learning_rate": 1.2053433359066057e-08, "loss": 2.6784, "step": 48023 }, { "epoch": 2.9811906387733567, "grad_norm": 0.32672009110688593, "learning_rate": 1.1974268041320403e-08, "loss": 2.6968, "step": 48024 }, { "epoch": 2.981252715873114, "grad_norm": 0.32818028574627545, "learning_rate": 1.1895363523234393e-08, "loss": 2.7736, "step": 48025 }, { "epoch": 2.9813147929728725, "grad_norm": 0.3194768985327395, "learning_rate": 1.1816719805229915e-08, "loss": 2.6369, "step": 48026 }, { "epoch": 2.98137687007263, "grad_norm": 0.334812309277188, "learning_rate": 1.1738336887706647e-08, "loss": 2.5958, "step": 48027 }, { "epoch": 2.9814389471723883, "grad_norm": 0.32934287531651785, "learning_rate": 1.1660214771086475e-08, "loss": 2.6492, "step": 48028 }, { "epoch": 2.981501024272146, "grad_norm": 0.3044211005522278, "learning_rate": 1.1582353455763528e-08, "loss": 2.6663, "step": 48029 }, { "epoch": 2.981563101371904, "grad_norm": 0.3252036863972932, "learning_rate": 1.150475294215414e-08, "loss": 2.558, "step": 48030 }, { "epoch": 2.9816251784716616, "grad_norm": 0.335807442440114, "learning_rate": 1.142741323065799e-08, "loss": 2.7432, "step": 48031 }, { "epoch": 2.9816872555714196, "grad_norm": 0.3282127376602462, "learning_rate": 1.1350334321680311e-08, "loss": 2.6954, "step": 48032 }, { "epoch": 2.9817493326711775, "grad_norm": 0.32779661276459665, "learning_rate": 1.1273516215626334e-08, "loss": 2.54, "step": 48033 }, { "epoch": 2.9818114097709354, "grad_norm": 0.32873675675061725, "learning_rate": 1.1196958912890188e-08, "loss": 2.6822, "step": 48034 }, { "epoch": 2.9818734868706933, "grad_norm": 0.3349803107083739, "learning_rate": 1.1120662413877104e-08, "loss": 2.7101, "step": 48035 }, { "epoch": 2.9819355639704512, "grad_norm": 0.31662586031848455, "learning_rate": 1.104462671898121e-08, "loss": 2.6744, "step": 48036 }, { "epoch": 2.981997641070209, "grad_norm": 0.3382200608625036, "learning_rate": 1.096885182860219e-08, "loss": 2.6062, "step": 48037 }, { "epoch": 2.982059718169967, "grad_norm": 0.3329336022166218, "learning_rate": 1.0893337743134169e-08, "loss": 2.7097, "step": 48038 }, { "epoch": 2.982121795269725, "grad_norm": 0.32728189052321327, "learning_rate": 1.0818084462965728e-08, "loss": 2.6001, "step": 48039 }, { "epoch": 2.982183872369483, "grad_norm": 0.31886497088354315, "learning_rate": 1.0743091988502097e-08, "loss": 2.5654, "step": 48040 }, { "epoch": 2.982245949469241, "grad_norm": 0.3377206297562958, "learning_rate": 1.0668360320120752e-08, "loss": 2.628, "step": 48041 }, { "epoch": 2.9823080265689987, "grad_norm": 0.3204808839298448, "learning_rate": 1.0593889458221374e-08, "loss": 2.6334, "step": 48042 }, { "epoch": 2.9823701036687567, "grad_norm": 0.3341278749450004, "learning_rate": 1.0519679403186988e-08, "loss": 2.716, "step": 48043 }, { "epoch": 2.9824321807685146, "grad_norm": 0.3306555803791675, "learning_rate": 1.0445730155411725e-08, "loss": 2.6136, "step": 48044 }, { "epoch": 2.9824942578682725, "grad_norm": 0.33342051731825173, "learning_rate": 1.0372041715267511e-08, "loss": 2.6372, "step": 48045 }, { "epoch": 2.9825563349680304, "grad_norm": 0.3256016046133325, "learning_rate": 1.0298614083154023e-08, "loss": 2.563, "step": 48046 }, { "epoch": 2.9826184120677883, "grad_norm": 0.34058284411166934, "learning_rate": 1.0225447259448739e-08, "loss": 2.6341, "step": 48047 }, { "epoch": 2.9826804891675462, "grad_norm": 0.34035318930053354, "learning_rate": 1.0152541244529135e-08, "loss": 2.7516, "step": 48048 }, { "epoch": 2.982742566267304, "grad_norm": 0.34245514831836393, "learning_rate": 1.0079896038783787e-08, "loss": 2.6848, "step": 48049 }, { "epoch": 2.9828046433670616, "grad_norm": 0.3305625633568576, "learning_rate": 1.0007511642584622e-08, "loss": 2.6781, "step": 48050 }, { "epoch": 2.98286672046682, "grad_norm": 0.3126168271079007, "learning_rate": 9.935388056309115e-09, "loss": 2.7336, "step": 48051 }, { "epoch": 2.9829287975665775, "grad_norm": 0.3260279240723312, "learning_rate": 9.863525280340292e-09, "loss": 2.6509, "step": 48052 }, { "epoch": 2.982990874666336, "grad_norm": 0.33632415875478744, "learning_rate": 9.791923315044527e-09, "loss": 2.6556, "step": 48053 }, { "epoch": 2.9830529517660933, "grad_norm": 0.3418886987776509, "learning_rate": 9.720582160804848e-09, "loss": 2.6465, "step": 48054 }, { "epoch": 2.9831150288658512, "grad_norm": 0.31585705743317316, "learning_rate": 9.649501817987627e-09, "loss": 2.6457, "step": 48055 }, { "epoch": 2.983177105965609, "grad_norm": 0.31506919929542127, "learning_rate": 9.57868228695924e-09, "loss": 2.6301, "step": 48056 }, { "epoch": 2.983239183065367, "grad_norm": 0.32476499800154685, "learning_rate": 9.50812356810271e-09, "loss": 2.7186, "step": 48057 }, { "epoch": 2.983301260165125, "grad_norm": 0.3338740205539746, "learning_rate": 9.437825661773314e-09, "loss": 2.6104, "step": 48058 }, { "epoch": 2.983363337264883, "grad_norm": 0.3303028334986919, "learning_rate": 9.367788568342973e-09, "loss": 2.667, "step": 48059 }, { "epoch": 2.983425414364641, "grad_norm": 0.3322994719558226, "learning_rate": 9.298012288172509e-09, "loss": 2.5839, "step": 48060 }, { "epoch": 2.9834874914643987, "grad_norm": 0.33540988119539805, "learning_rate": 9.228496821633848e-09, "loss": 2.6673, "step": 48061 }, { "epoch": 2.9835495685641567, "grad_norm": 0.3172569025151512, "learning_rate": 9.159242169087811e-09, "loss": 2.621, "step": 48062 }, { "epoch": 2.9836116456639146, "grad_norm": 0.3288608202514413, "learning_rate": 9.090248330889672e-09, "loss": 2.7282, "step": 48063 }, { "epoch": 2.9836737227636725, "grad_norm": 0.31454160878860626, "learning_rate": 9.021515307405803e-09, "loss": 2.6742, "step": 48064 }, { "epoch": 2.9837357998634304, "grad_norm": 0.32266205188993646, "learning_rate": 8.953043098991476e-09, "loss": 2.6633, "step": 48065 }, { "epoch": 2.9837978769631883, "grad_norm": 0.3269856859444186, "learning_rate": 8.884831706007512e-09, "loss": 2.6166, "step": 48066 }, { "epoch": 2.9838599540629462, "grad_norm": 0.33656919215924297, "learning_rate": 8.816881128803634e-09, "loss": 2.6924, "step": 48067 }, { "epoch": 2.983922031162704, "grad_norm": 0.3352825633690033, "learning_rate": 8.749191367740662e-09, "loss": 2.7023, "step": 48068 }, { "epoch": 2.983984108262462, "grad_norm": 0.33989220854688845, "learning_rate": 8.681762423168316e-09, "loss": 2.673, "step": 48069 }, { "epoch": 2.98404618536222, "grad_norm": 0.3516080085121978, "learning_rate": 8.614594295436318e-09, "loss": 2.5332, "step": 48070 }, { "epoch": 2.984108262461978, "grad_norm": 0.3415494295605227, "learning_rate": 8.547686984899938e-09, "loss": 2.6406, "step": 48071 }, { "epoch": 2.984170339561736, "grad_norm": 0.32068678457655775, "learning_rate": 8.481040491903347e-09, "loss": 2.7086, "step": 48072 }, { "epoch": 2.9842324166614933, "grad_norm": 0.3405280403937436, "learning_rate": 8.414654816801815e-09, "loss": 2.7157, "step": 48073 }, { "epoch": 2.9842944937612517, "grad_norm": 0.3187210509300117, "learning_rate": 8.34852995993396e-09, "loss": 2.6752, "step": 48074 }, { "epoch": 2.984356570861009, "grad_norm": 0.32840305989863405, "learning_rate": 8.282665921649502e-09, "loss": 2.5782, "step": 48075 }, { "epoch": 2.9844186479607675, "grad_norm": 0.34154117283807317, "learning_rate": 8.217062702287059e-09, "loss": 2.748, "step": 48076 }, { "epoch": 2.984480725060525, "grad_norm": 0.33871618198843495, "learning_rate": 8.151720302196352e-09, "loss": 2.7175, "step": 48077 }, { "epoch": 2.9845428021602833, "grad_norm": 0.3100785857235885, "learning_rate": 8.08663872171045e-09, "loss": 2.6747, "step": 48078 }, { "epoch": 2.984604879260041, "grad_norm": 0.33349904272694153, "learning_rate": 8.021817961173517e-09, "loss": 2.6863, "step": 48079 }, { "epoch": 2.9846669563597987, "grad_norm": 0.33855158595689877, "learning_rate": 7.957258020924175e-09, "loss": 2.6621, "step": 48080 }, { "epoch": 2.9847290334595566, "grad_norm": 0.32695752167109776, "learning_rate": 7.892958901295489e-09, "loss": 2.6435, "step": 48081 }, { "epoch": 2.9847911105593146, "grad_norm": 0.33122781148407315, "learning_rate": 7.828920602626078e-09, "loss": 2.6797, "step": 48082 }, { "epoch": 2.9848531876590725, "grad_norm": 0.3270925559527491, "learning_rate": 7.76514312525456e-09, "loss": 2.6633, "step": 48083 }, { "epoch": 2.9849152647588304, "grad_norm": 0.3149273418272821, "learning_rate": 7.701626469502898e-09, "loss": 2.6943, "step": 48084 }, { "epoch": 2.9849773418585883, "grad_norm": 0.33008041606762556, "learning_rate": 7.638370635709712e-09, "loss": 2.635, "step": 48085 }, { "epoch": 2.9850394189583462, "grad_norm": 0.3301434536903484, "learning_rate": 7.575375624196967e-09, "loss": 2.6853, "step": 48086 }, { "epoch": 2.985101496058104, "grad_norm": 0.3416109069627949, "learning_rate": 7.512641435308832e-09, "loss": 2.6476, "step": 48087 }, { "epoch": 2.985163573157862, "grad_norm": 0.3304896977055372, "learning_rate": 7.450168069356167e-09, "loss": 2.6704, "step": 48088 }, { "epoch": 2.98522565025762, "grad_norm": 0.34083181104965193, "learning_rate": 7.387955526677593e-09, "loss": 2.6773, "step": 48089 }, { "epoch": 2.985287727357378, "grad_norm": 0.33309451006620017, "learning_rate": 7.326003807589521e-09, "loss": 2.7483, "step": 48090 }, { "epoch": 2.985349804457136, "grad_norm": 0.335551294957139, "learning_rate": 7.264312912419469e-09, "loss": 2.7274, "step": 48091 }, { "epoch": 2.9854118815568937, "grad_norm": 0.33828670241827075, "learning_rate": 7.202882841483849e-09, "loss": 2.6116, "step": 48092 }, { "epoch": 2.9854739586566517, "grad_norm": 0.3134914881875521, "learning_rate": 7.141713595110177e-09, "loss": 2.6421, "step": 48093 }, { "epoch": 2.9855360357564096, "grad_norm": 0.34916389042965273, "learning_rate": 7.080805173614868e-09, "loss": 2.6854, "step": 48094 }, { "epoch": 2.9855981128561675, "grad_norm": 0.3194160710855119, "learning_rate": 7.020157577314335e-09, "loss": 2.6025, "step": 48095 }, { "epoch": 2.9856601899559254, "grad_norm": 0.32497325108563235, "learning_rate": 6.95977080652499e-09, "loss": 2.5941, "step": 48096 }, { "epoch": 2.9857222670556833, "grad_norm": 0.3274924106846939, "learning_rate": 6.899644861563248e-09, "loss": 2.6571, "step": 48097 }, { "epoch": 2.985784344155441, "grad_norm": 0.3200411228489354, "learning_rate": 6.839779742739971e-09, "loss": 2.6076, "step": 48098 }, { "epoch": 2.985846421255199, "grad_norm": 0.33146061949038413, "learning_rate": 6.780175450371573e-09, "loss": 2.6632, "step": 48099 }, { "epoch": 2.9859084983549566, "grad_norm": 0.32332224013078964, "learning_rate": 6.720831984768916e-09, "loss": 2.641, "step": 48100 }, { "epoch": 2.985970575454715, "grad_norm": 0.3461416663417358, "learning_rate": 6.661749346237312e-09, "loss": 2.6232, "step": 48101 }, { "epoch": 2.9860326525544725, "grad_norm": 0.33261725380359825, "learning_rate": 6.6029275350876215e-09, "loss": 2.6953, "step": 48102 }, { "epoch": 2.9860947296542304, "grad_norm": 0.32681054272980214, "learning_rate": 6.54436655163071e-09, "loss": 2.5423, "step": 48103 }, { "epoch": 2.9861568067539883, "grad_norm": 0.33966436803924077, "learning_rate": 6.4860663961663345e-09, "loss": 2.7367, "step": 48104 }, { "epoch": 2.9862188838537462, "grad_norm": 0.3399848316294224, "learning_rate": 6.428027068999809e-09, "loss": 2.7317, "step": 48105 }, { "epoch": 2.986280960953504, "grad_norm": 0.3320366129722562, "learning_rate": 6.370248570436443e-09, "loss": 2.6284, "step": 48106 }, { "epoch": 2.986343038053262, "grad_norm": 0.330420615476092, "learning_rate": 6.312730900770447e-09, "loss": 2.6359, "step": 48107 }, { "epoch": 2.98640511515302, "grad_norm": 0.33017909745878354, "learning_rate": 6.2554740603126825e-09, "loss": 2.6874, "step": 48108 }, { "epoch": 2.986467192252778, "grad_norm": 0.32698613026424417, "learning_rate": 6.198478049357359e-09, "loss": 2.7567, "step": 48109 }, { "epoch": 2.986529269352536, "grad_norm": 0.33260801373216314, "learning_rate": 6.141742868193134e-09, "loss": 2.7023, "step": 48110 }, { "epoch": 2.9865913464522937, "grad_norm": 0.3422989890096458, "learning_rate": 6.085268517130871e-09, "loss": 2.7453, "step": 48111 }, { "epoch": 2.9866534235520517, "grad_norm": 0.3364531972313992, "learning_rate": 6.029054996459227e-09, "loss": 2.6537, "step": 48112 }, { "epoch": 2.9867155006518096, "grad_norm": 0.3159592143142118, "learning_rate": 5.97310230646686e-09, "loss": 2.6024, "step": 48113 }, { "epoch": 2.9867775777515675, "grad_norm": 0.3295125687044257, "learning_rate": 5.917410447447979e-09, "loss": 2.6559, "step": 48114 }, { "epoch": 2.9868396548513254, "grad_norm": 0.32222670117755725, "learning_rate": 5.861979419696795e-09, "loss": 2.5668, "step": 48115 }, { "epoch": 2.9869017319510833, "grad_norm": 0.3185883500154634, "learning_rate": 5.806809223501963e-09, "loss": 2.6188, "step": 48116 }, { "epoch": 2.9869638090508412, "grad_norm": 0.3413293092741605, "learning_rate": 5.751899859146592e-09, "loss": 2.6365, "step": 48117 }, { "epoch": 2.987025886150599, "grad_norm": 0.340337340041231, "learning_rate": 5.697251326919339e-09, "loss": 2.6478, "step": 48118 }, { "epoch": 2.987087963250357, "grad_norm": 0.3283053878668867, "learning_rate": 5.6428636271088635e-09, "loss": 2.6611, "step": 48119 }, { "epoch": 2.987150040350115, "grad_norm": 0.32529718218238796, "learning_rate": 5.5887367599982695e-09, "loss": 2.6401, "step": 48120 }, { "epoch": 2.9872121174498725, "grad_norm": 0.34428823122202346, "learning_rate": 5.534870725865115e-09, "loss": 2.6653, "step": 48121 }, { "epoch": 2.987274194549631, "grad_norm": 0.3261025292238803, "learning_rate": 5.481265524992507e-09, "loss": 2.6411, "step": 48122 }, { "epoch": 2.9873362716493883, "grad_norm": 0.3419207648133878, "learning_rate": 5.4279211576580004e-09, "loss": 2.6242, "step": 48123 }, { "epoch": 2.9873983487491467, "grad_norm": 0.3506734096033065, "learning_rate": 5.374837624144702e-09, "loss": 2.7494, "step": 48124 }, { "epoch": 2.987460425848904, "grad_norm": 0.3424752397358314, "learning_rate": 5.322014924730168e-09, "loss": 2.5766, "step": 48125 }, { "epoch": 2.9875225029486625, "grad_norm": 0.3278096363826222, "learning_rate": 5.269453059686402e-09, "loss": 2.6697, "step": 48126 }, { "epoch": 2.98758458004842, "grad_norm": 0.324420923370625, "learning_rate": 5.217152029285411e-09, "loss": 2.6601, "step": 48127 }, { "epoch": 2.987646657148178, "grad_norm": 0.3195255428261545, "learning_rate": 5.165111833804748e-09, "loss": 2.6136, "step": 48128 }, { "epoch": 2.987708734247936, "grad_norm": 0.33854839489437594, "learning_rate": 5.113332473516419e-09, "loss": 2.7352, "step": 48129 }, { "epoch": 2.9877708113476937, "grad_norm": 0.32425192008336023, "learning_rate": 5.061813948686878e-09, "loss": 2.6644, "step": 48130 }, { "epoch": 2.9878328884474517, "grad_norm": 0.3425629124150876, "learning_rate": 5.0105562595881286e-09, "loss": 2.6628, "step": 48131 }, { "epoch": 2.9878949655472096, "grad_norm": 0.3267839821450204, "learning_rate": 4.959559406486625e-09, "loss": 2.7575, "step": 48132 }, { "epoch": 2.9879570426469675, "grad_norm": 0.3157240495022711, "learning_rate": 4.908823389643269e-09, "loss": 2.6737, "step": 48133 }, { "epoch": 2.9880191197467254, "grad_norm": 0.33367976301476, "learning_rate": 4.858348209330066e-09, "loss": 2.7211, "step": 48134 }, { "epoch": 2.9880811968464833, "grad_norm": 0.3400305859222953, "learning_rate": 4.808133865807917e-09, "loss": 2.6806, "step": 48135 }, { "epoch": 2.9881432739462412, "grad_norm": 0.3221504121074382, "learning_rate": 4.7581803593377275e-09, "loss": 2.6737, "step": 48136 }, { "epoch": 2.988205351045999, "grad_norm": 0.3221642652693613, "learning_rate": 4.7084876901803965e-09, "loss": 2.6398, "step": 48137 }, { "epoch": 2.988267428145757, "grad_norm": 0.32315621284312745, "learning_rate": 4.659055858596828e-09, "loss": 2.5696, "step": 48138 }, { "epoch": 2.988329505245515, "grad_norm": 0.3192992466371109, "learning_rate": 4.609884864842373e-09, "loss": 2.6383, "step": 48139 }, { "epoch": 2.988391582345273, "grad_norm": 0.3246294889863126, "learning_rate": 4.560974709177934e-09, "loss": 2.6506, "step": 48140 }, { "epoch": 2.988453659445031, "grad_norm": 0.3247791181582639, "learning_rate": 4.512325391853311e-09, "loss": 2.6754, "step": 48141 }, { "epoch": 2.9885157365447887, "grad_norm": 0.3157837949836175, "learning_rate": 4.4639369131238566e-09, "loss": 2.7248, "step": 48142 }, { "epoch": 2.9885778136445467, "grad_norm": 0.31604524905027503, "learning_rate": 4.41580927324492e-09, "loss": 2.6658, "step": 48143 }, { "epoch": 2.9886398907443046, "grad_norm": 0.3273463888676509, "learning_rate": 4.367942472466302e-09, "loss": 2.6798, "step": 48144 }, { "epoch": 2.9887019678440625, "grad_norm": 0.3294174987041974, "learning_rate": 4.320336511032252e-09, "loss": 2.7402, "step": 48145 }, { "epoch": 2.98876404494382, "grad_norm": 0.33520224417914496, "learning_rate": 4.2729913892036736e-09, "loss": 2.6742, "step": 48146 }, { "epoch": 2.9888261220435783, "grad_norm": 0.3397405401896027, "learning_rate": 4.225907107213712e-09, "loss": 2.7082, "step": 48147 }, { "epoch": 2.988888199143336, "grad_norm": 0.3338331582912787, "learning_rate": 4.179083665312167e-09, "loss": 2.6883, "step": 48148 }, { "epoch": 2.988950276243094, "grad_norm": 0.3192820255605779, "learning_rate": 4.1325210637488395e-09, "loss": 2.7246, "step": 48149 }, { "epoch": 2.9890123533428516, "grad_norm": 0.33206090233732755, "learning_rate": 4.086219302762428e-09, "loss": 2.6096, "step": 48150 }, { "epoch": 2.9890744304426096, "grad_norm": 0.3194463229930393, "learning_rate": 4.040178382597182e-09, "loss": 2.5628, "step": 48151 }, { "epoch": 2.9891365075423675, "grad_norm": 0.33125835669071046, "learning_rate": 3.994398303491798e-09, "loss": 2.6194, "step": 48152 }, { "epoch": 2.9891985846421254, "grad_norm": 0.3238728988119948, "learning_rate": 3.9488790656794226e-09, "loss": 2.5855, "step": 48153 }, { "epoch": 2.9892606617418833, "grad_norm": 0.32815632216501167, "learning_rate": 3.903620669409857e-09, "loss": 2.7103, "step": 48154 }, { "epoch": 2.9893227388416412, "grad_norm": 0.32054355475509605, "learning_rate": 3.858623114905147e-09, "loss": 2.6413, "step": 48155 }, { "epoch": 2.989384815941399, "grad_norm": 0.33294537314058553, "learning_rate": 3.813886402409539e-09, "loss": 2.6534, "step": 48156 }, { "epoch": 2.989446893041157, "grad_norm": 0.3142677850842013, "learning_rate": 3.769410532156181e-09, "loss": 2.6489, "step": 48157 }, { "epoch": 2.989508970140915, "grad_norm": 0.3184112319369343, "learning_rate": 3.7251955043726696e-09, "loss": 2.6863, "step": 48158 }, { "epoch": 2.989571047240673, "grad_norm": 0.3228360416399354, "learning_rate": 3.68124131929215e-09, "loss": 2.7002, "step": 48159 }, { "epoch": 2.989633124340431, "grad_norm": 0.321059495718402, "learning_rate": 3.6375479771422194e-09, "loss": 2.7425, "step": 48160 }, { "epoch": 2.9896952014401887, "grad_norm": 0.32041918908396666, "learning_rate": 3.594115478156024e-09, "loss": 2.7731, "step": 48161 }, { "epoch": 2.9897572785399467, "grad_norm": 0.32789472037960105, "learning_rate": 3.550943822550057e-09, "loss": 2.5439, "step": 48162 }, { "epoch": 2.9898193556397046, "grad_norm": 0.3133912371441025, "learning_rate": 3.5080330105630167e-09, "loss": 2.6614, "step": 48163 }, { "epoch": 2.9898814327394625, "grad_norm": 0.33854609510997696, "learning_rate": 3.4653830424058455e-09, "loss": 2.6679, "step": 48164 }, { "epoch": 2.9899435098392204, "grad_norm": 0.3267450949228479, "learning_rate": 3.42299391831169e-09, "loss": 2.6703, "step": 48165 }, { "epoch": 2.9900055869389783, "grad_norm": 0.33303007347899233, "learning_rate": 3.3808656384914928e-09, "loss": 2.6628, "step": 48166 }, { "epoch": 2.9900676640387363, "grad_norm": 0.33586811857340026, "learning_rate": 3.3389982031728493e-09, "loss": 2.6609, "step": 48167 }, { "epoch": 2.990129741138494, "grad_norm": 0.3384637877556906, "learning_rate": 3.2973916125722536e-09, "loss": 2.7504, "step": 48168 }, { "epoch": 2.9901918182382516, "grad_norm": 0.3364492112016551, "learning_rate": 3.2560458669061987e-09, "loss": 2.6238, "step": 48169 }, { "epoch": 2.99025389533801, "grad_norm": 0.33569636673013703, "learning_rate": 3.2149609663911783e-09, "loss": 2.603, "step": 48170 }, { "epoch": 2.9903159724377675, "grad_norm": 0.32901105003946146, "learning_rate": 3.174136911243686e-09, "loss": 2.632, "step": 48171 }, { "epoch": 2.990378049537526, "grad_norm": 0.3392365281494989, "learning_rate": 3.1335737016691127e-09, "loss": 2.6204, "step": 48172 }, { "epoch": 2.9904401266372833, "grad_norm": 0.3317125473844103, "learning_rate": 3.093271337883952e-09, "loss": 2.6771, "step": 48173 }, { "epoch": 2.9905022037370417, "grad_norm": 0.30784055078563644, "learning_rate": 3.053229820099146e-09, "loss": 2.6604, "step": 48174 }, { "epoch": 2.990564280836799, "grad_norm": 0.3092644561404744, "learning_rate": 3.0134491485256377e-09, "loss": 2.6351, "step": 48175 }, { "epoch": 2.990626357936557, "grad_norm": 0.33162090525955135, "learning_rate": 2.973929323363267e-09, "loss": 2.6285, "step": 48176 }, { "epoch": 2.990688435036315, "grad_norm": 0.3471097615695117, "learning_rate": 2.9346703448285275e-09, "loss": 2.5955, "step": 48177 }, { "epoch": 2.990750512136073, "grad_norm": 0.33020159426690715, "learning_rate": 2.8956722131157076e-09, "loss": 2.6646, "step": 48178 }, { "epoch": 2.990812589235831, "grad_norm": 0.32749572084994627, "learning_rate": 2.8569349284357504e-09, "loss": 2.6295, "step": 48179 }, { "epoch": 2.9908746663355887, "grad_norm": 0.3367804552613801, "learning_rate": 2.818458490988496e-09, "loss": 2.5703, "step": 48180 }, { "epoch": 2.9909367434353467, "grad_norm": 0.3373367201588135, "learning_rate": 2.780242900973784e-09, "loss": 2.6356, "step": 48181 }, { "epoch": 2.9909988205351046, "grad_norm": 0.3144383397387804, "learning_rate": 2.742288158591455e-09, "loss": 2.6015, "step": 48182 }, { "epoch": 2.9910608976348625, "grad_norm": 0.34829700015066106, "learning_rate": 2.704594264041349e-09, "loss": 2.7285, "step": 48183 }, { "epoch": 2.9911229747346204, "grad_norm": 0.36637768827276596, "learning_rate": 2.6671612175177552e-09, "loss": 2.6648, "step": 48184 }, { "epoch": 2.9911850518343783, "grad_norm": 0.33964362774471135, "learning_rate": 2.6299890192205135e-09, "loss": 2.6505, "step": 48185 }, { "epoch": 2.9912471289341362, "grad_norm": 0.32857504136123633, "learning_rate": 2.5930776693328107e-09, "loss": 2.6615, "step": 48186 }, { "epoch": 2.991309206033894, "grad_norm": 0.318714411936693, "learning_rate": 2.556427168060038e-09, "loss": 2.7563, "step": 48187 }, { "epoch": 2.991371283133652, "grad_norm": 0.3431124084364968, "learning_rate": 2.520037515585383e-09, "loss": 2.6198, "step": 48188 }, { "epoch": 2.99143336023341, "grad_norm": 0.3290920312786057, "learning_rate": 2.4839087121031334e-09, "loss": 2.7076, "step": 48189 }, { "epoch": 2.991495437333168, "grad_norm": 0.3443567166953013, "learning_rate": 2.448040757796477e-09, "loss": 2.6007, "step": 48190 }, { "epoch": 2.991557514432926, "grad_norm": 0.3144906418496679, "learning_rate": 2.4124336528597024e-09, "loss": 2.7242, "step": 48191 }, { "epoch": 2.9916195915326838, "grad_norm": 0.3158831333959486, "learning_rate": 2.3770873974704454e-09, "loss": 2.7467, "step": 48192 }, { "epoch": 2.9916816686324417, "grad_norm": 0.31642538177670343, "learning_rate": 2.342001991817444e-09, "loss": 2.6559, "step": 48193 }, { "epoch": 2.991743745732199, "grad_norm": 0.3239787488843963, "learning_rate": 2.307177436083885e-09, "loss": 2.7311, "step": 48194 }, { "epoch": 2.9918058228319575, "grad_norm": 0.3277072290538137, "learning_rate": 2.272613730452955e-09, "loss": 2.7442, "step": 48195 }, { "epoch": 2.991867899931715, "grad_norm": 0.31667557780167244, "learning_rate": 2.2383108751022897e-09, "loss": 2.5751, "step": 48196 }, { "epoch": 2.9919299770314733, "grad_norm": 0.32104220661131955, "learning_rate": 2.204268870215076e-09, "loss": 2.6919, "step": 48197 }, { "epoch": 2.991992054131231, "grad_norm": 0.32073176090016847, "learning_rate": 2.1704877159633985e-09, "loss": 2.5883, "step": 48198 }, { "epoch": 2.9920541312309887, "grad_norm": 0.33892699711470026, "learning_rate": 2.1369674125193417e-09, "loss": 2.6413, "step": 48199 }, { "epoch": 2.9921162083307467, "grad_norm": 0.33152577688507145, "learning_rate": 2.1037079600716437e-09, "loss": 2.607, "step": 48200 }, { "epoch": 2.9921782854305046, "grad_norm": 0.32236799290405305, "learning_rate": 2.0707093587868377e-09, "loss": 2.6813, "step": 48201 }, { "epoch": 2.9922403625302625, "grad_norm": 0.33850380490177595, "learning_rate": 2.037971608831457e-09, "loss": 2.6948, "step": 48202 }, { "epoch": 2.9923024396300204, "grad_norm": 0.32085463537779674, "learning_rate": 2.005494710383138e-09, "loss": 2.6537, "step": 48203 }, { "epoch": 2.9923645167297783, "grad_norm": 0.3526018884529807, "learning_rate": 1.9732786636139644e-09, "loss": 2.6038, "step": 48204 }, { "epoch": 2.9924265938295362, "grad_norm": 0.32456013504582976, "learning_rate": 1.941323468679368e-09, "loss": 2.6786, "step": 48205 }, { "epoch": 2.992488670929294, "grad_norm": 0.3320695181840372, "learning_rate": 1.9096291257625355e-09, "loss": 2.6859, "step": 48206 }, { "epoch": 2.992550748029052, "grad_norm": 0.3454856619626975, "learning_rate": 1.8781956350133466e-09, "loss": 2.7226, "step": 48207 }, { "epoch": 2.99261282512881, "grad_norm": 0.3259595076829022, "learning_rate": 1.8470229966094378e-09, "loss": 2.5824, "step": 48208 }, { "epoch": 2.992674902228568, "grad_norm": 0.3415170341414108, "learning_rate": 1.8161112107062394e-09, "loss": 2.7145, "step": 48209 }, { "epoch": 2.992736979328326, "grad_norm": 0.33753002086211115, "learning_rate": 1.7854602774591834e-09, "loss": 2.7362, "step": 48210 }, { "epoch": 2.9927990564280837, "grad_norm": 0.33544087850760984, "learning_rate": 1.7550701970403538e-09, "loss": 2.6938, "step": 48211 }, { "epoch": 2.9928611335278417, "grad_norm": 0.335545599583077, "learning_rate": 1.7249409696051822e-09, "loss": 2.7385, "step": 48212 }, { "epoch": 2.9929232106275996, "grad_norm": 0.33154064284395046, "learning_rate": 1.6950725953035485e-09, "loss": 2.639, "step": 48213 }, { "epoch": 2.9929852877273575, "grad_norm": 0.31851607836573403, "learning_rate": 1.665465074296435e-09, "loss": 2.7399, "step": 48214 }, { "epoch": 2.9930473648271154, "grad_norm": 0.33452641410129635, "learning_rate": 1.6361184067392731e-09, "loss": 2.7097, "step": 48215 }, { "epoch": 2.9931094419268733, "grad_norm": 0.31667297800366734, "learning_rate": 1.6070325927874942e-09, "loss": 2.6522, "step": 48216 }, { "epoch": 2.993171519026631, "grad_norm": 0.3329821266269177, "learning_rate": 1.5782076325854267e-09, "loss": 2.6233, "step": 48217 }, { "epoch": 2.993233596126389, "grad_norm": 0.3197537992570838, "learning_rate": 1.5496435262885023e-09, "loss": 2.7211, "step": 48218 }, { "epoch": 2.9932956732261466, "grad_norm": 0.323573769321384, "learning_rate": 1.5213402740466009e-09, "loss": 2.6491, "step": 48219 }, { "epoch": 2.993357750325905, "grad_norm": 0.31487396464418027, "learning_rate": 1.4932978760040516e-09, "loss": 2.6533, "step": 48220 }, { "epoch": 2.9934198274256625, "grad_norm": 0.3128174599817351, "learning_rate": 1.4655163323051835e-09, "loss": 2.6887, "step": 48221 }, { "epoch": 2.993481904525421, "grad_norm": 0.3406500185956548, "learning_rate": 1.4379956430998765e-09, "loss": 2.7006, "step": 48222 }, { "epoch": 2.9935439816251783, "grad_norm": 0.31198311019383546, "learning_rate": 1.4107358085324596e-09, "loss": 2.622, "step": 48223 }, { "epoch": 2.9936060587249362, "grad_norm": 0.3292290302042099, "learning_rate": 1.3837368287417108e-09, "loss": 2.6459, "step": 48224 }, { "epoch": 2.993668135824694, "grad_norm": 0.3343632540021231, "learning_rate": 1.356998703871959e-09, "loss": 2.6396, "step": 48225 }, { "epoch": 2.993730212924452, "grad_norm": 0.3285109997037036, "learning_rate": 1.3305214340564309e-09, "loss": 2.6758, "step": 48226 }, { "epoch": 2.99379229002421, "grad_norm": 0.3259473701877131, "learning_rate": 1.3043050194394556e-09, "loss": 2.6276, "step": 48227 }, { "epoch": 2.993854367123968, "grad_norm": 0.3289953790625681, "learning_rate": 1.27834946015426e-09, "loss": 2.5847, "step": 48228 }, { "epoch": 2.993916444223726, "grad_norm": 0.32027965470295533, "learning_rate": 1.2526547563340707e-09, "loss": 2.6866, "step": 48229 }, { "epoch": 2.9939785213234837, "grad_norm": 0.3285905169782795, "learning_rate": 1.2272209081232167e-09, "loss": 2.5951, "step": 48230 }, { "epoch": 2.9940405984232417, "grad_norm": 0.3235646163925741, "learning_rate": 1.2020479156438225e-09, "loss": 2.678, "step": 48231 }, { "epoch": 2.9941026755229996, "grad_norm": 0.342574863570553, "learning_rate": 1.177135779029115e-09, "loss": 2.6709, "step": 48232 }, { "epoch": 2.9941647526227575, "grad_norm": 0.3148422558278109, "learning_rate": 1.1524844984178718e-09, "loss": 2.6933, "step": 48233 }, { "epoch": 2.9942268297225154, "grad_norm": 0.3262666119259642, "learning_rate": 1.1280940739266666e-09, "loss": 2.6275, "step": 48234 }, { "epoch": 2.9942889068222733, "grad_norm": 0.3360840123076942, "learning_rate": 1.103964505688726e-09, "loss": 2.6658, "step": 48235 }, { "epoch": 2.9943509839220313, "grad_norm": 0.3417594453208218, "learning_rate": 1.0800957938261747e-09, "loss": 2.6631, "step": 48236 }, { "epoch": 2.994413061021789, "grad_norm": 0.35580656746442063, "learning_rate": 1.0564879384722392e-09, "loss": 2.6146, "step": 48237 }, { "epoch": 2.994475138121547, "grad_norm": 0.33481818401944446, "learning_rate": 1.0331409397379422e-09, "loss": 2.6703, "step": 48238 }, { "epoch": 2.994537215221305, "grad_norm": 0.3375867899811362, "learning_rate": 1.0100547977565101e-09, "loss": 2.7306, "step": 48239 }, { "epoch": 2.994599292321063, "grad_norm": 0.33040926197208564, "learning_rate": 9.872295126389653e-10, "loss": 2.7048, "step": 48240 }, { "epoch": 2.994661369420821, "grad_norm": 0.3247193413028688, "learning_rate": 9.646650845129834e-10, "loss": 2.6249, "step": 48241 }, { "epoch": 2.9947234465205783, "grad_norm": 0.3418865522179802, "learning_rate": 9.423615134895869e-10, "loss": 2.6135, "step": 48242 }, { "epoch": 2.9947855236203367, "grad_norm": 0.336621021594999, "learning_rate": 9.203187996909002e-10, "loss": 2.6843, "step": 48243 }, { "epoch": 2.994847600720094, "grad_norm": 0.3378525971639286, "learning_rate": 8.985369432223945e-10, "loss": 2.7406, "step": 48244 }, { "epoch": 2.9949096778198525, "grad_norm": 0.33475154192756695, "learning_rate": 8.770159442117453e-10, "loss": 2.5738, "step": 48245 }, { "epoch": 2.99497175491961, "grad_norm": 0.3649799256160735, "learning_rate": 8.55755802758873e-10, "loss": 2.6826, "step": 48246 }, { "epoch": 2.995033832019368, "grad_norm": 0.31729898372728443, "learning_rate": 8.347565189803508e-10, "loss": 2.6815, "step": 48247 }, { "epoch": 2.995095909119126, "grad_norm": 0.32981255812551374, "learning_rate": 8.140180929816499e-10, "loss": 2.6841, "step": 48248 }, { "epoch": 2.9951579862188837, "grad_norm": 0.3246953173367288, "learning_rate": 7.935405248737926e-10, "loss": 2.6026, "step": 48249 }, { "epoch": 2.9952200633186417, "grad_norm": 0.33416184512050545, "learning_rate": 7.733238147678012e-10, "loss": 2.6367, "step": 48250 }, { "epoch": 2.9952821404183996, "grad_norm": 0.33520970129504857, "learning_rate": 7.533679627635959e-10, "loss": 2.7083, "step": 48251 }, { "epoch": 2.9953442175181575, "grad_norm": 0.322459907922886, "learning_rate": 7.336729689666477e-10, "loss": 2.6713, "step": 48252 }, { "epoch": 2.9954062946179154, "grad_norm": 0.33672720108578824, "learning_rate": 7.14238833482428e-10, "loss": 2.6605, "step": 48253 }, { "epoch": 2.9954683717176733, "grad_norm": 0.33109879551527005, "learning_rate": 6.950655564053054e-10, "loss": 2.6611, "step": 48254 }, { "epoch": 2.9955304488174312, "grad_norm": 0.3274948335924391, "learning_rate": 6.761531378407515e-10, "loss": 2.6863, "step": 48255 }, { "epoch": 2.995592525917189, "grad_norm": 0.3240491124535784, "learning_rate": 6.575015778886862e-10, "loss": 2.6523, "step": 48256 }, { "epoch": 2.995654603016947, "grad_norm": 0.3260828799056844, "learning_rate": 6.391108766434784e-10, "loss": 2.7564, "step": 48257 }, { "epoch": 2.995716680116705, "grad_norm": 0.328978959969566, "learning_rate": 6.20981034199497e-10, "loss": 2.6333, "step": 48258 }, { "epoch": 2.995778757216463, "grad_norm": 0.3183469828434667, "learning_rate": 6.031120506566623e-10, "loss": 2.6697, "step": 48259 }, { "epoch": 2.995840834316221, "grad_norm": 0.326925036318763, "learning_rate": 5.85503926103792e-10, "loss": 2.643, "step": 48260 }, { "epoch": 2.9959029114159788, "grad_norm": 0.3278220003858747, "learning_rate": 5.681566606352551e-10, "loss": 2.7057, "step": 48261 }, { "epoch": 2.9959649885157367, "grad_norm": 0.33977103277496024, "learning_rate": 5.510702543398694e-10, "loss": 2.5641, "step": 48262 }, { "epoch": 2.9960270656154946, "grad_norm": 0.31995549892776803, "learning_rate": 5.342447073064527e-10, "loss": 2.6277, "step": 48263 }, { "epoch": 2.9960891427152525, "grad_norm": 0.32578477331584427, "learning_rate": 5.17680019623823e-10, "loss": 2.6771, "step": 48264 }, { "epoch": 2.99615121981501, "grad_norm": 0.3300991390243045, "learning_rate": 5.013761913752469e-10, "loss": 2.6307, "step": 48265 }, { "epoch": 2.9962132969147683, "grad_norm": 0.318995115990157, "learning_rate": 4.853332226550933e-10, "loss": 2.697, "step": 48266 }, { "epoch": 2.996275374014526, "grad_norm": 0.32596042197685254, "learning_rate": 4.695511135355269e-10, "loss": 2.5888, "step": 48267 }, { "epoch": 2.996337451114284, "grad_norm": 0.34249788603861187, "learning_rate": 4.540298641053653e-10, "loss": 2.7095, "step": 48268 }, { "epoch": 2.9963995282140417, "grad_norm": 0.32038330830663553, "learning_rate": 4.387694744478754e-10, "loss": 2.6981, "step": 48269 }, { "epoch": 2.9964616053138, "grad_norm": 0.33797943183007373, "learning_rate": 4.2376994463522166e-10, "loss": 2.6373, "step": 48270 }, { "epoch": 2.9965236824135575, "grad_norm": 0.3397645638827141, "learning_rate": 4.0903127475067083e-10, "loss": 2.6801, "step": 48271 }, { "epoch": 2.9965857595133154, "grad_norm": 0.3232251649736504, "learning_rate": 3.9455346487193846e-10, "loss": 2.625, "step": 48272 }, { "epoch": 2.9966478366130733, "grad_norm": 0.338048335848694, "learning_rate": 3.8033651507118906e-10, "loss": 2.6736, "step": 48273 }, { "epoch": 2.9967099137128312, "grad_norm": 0.332256969695258, "learning_rate": 3.6638042542058713e-10, "loss": 2.6547, "step": 48274 }, { "epoch": 2.996771990812589, "grad_norm": 0.31477644984344877, "learning_rate": 3.5268519600339945e-10, "loss": 2.7212, "step": 48275 }, { "epoch": 2.996834067912347, "grad_norm": 0.3264829785202331, "learning_rate": 3.3925082688068823e-10, "loss": 2.6318, "step": 48276 }, { "epoch": 2.996896145012105, "grad_norm": 0.31141791155810183, "learning_rate": 3.2607731812461795e-10, "loss": 2.6873, "step": 48277 }, { "epoch": 2.996958222111863, "grad_norm": 0.33430856731067865, "learning_rate": 3.1316466980735315e-10, "loss": 2.7227, "step": 48278 }, { "epoch": 2.997020299211621, "grad_norm": 0.32000932068959576, "learning_rate": 3.005128819899561e-10, "loss": 2.7504, "step": 48279 }, { "epoch": 2.9970823763113787, "grad_norm": 0.31404429568011505, "learning_rate": 2.8812195474459124e-10, "loss": 2.6425, "step": 48280 }, { "epoch": 2.9971444534111367, "grad_norm": 0.32615085922425113, "learning_rate": 2.7599188813787203e-10, "loss": 2.679, "step": 48281 }, { "epoch": 2.9972065305108946, "grad_norm": 0.3166542023581536, "learning_rate": 2.6412268222530957e-10, "loss": 2.5645, "step": 48282 }, { "epoch": 2.9972686076106525, "grad_norm": 0.32092547497233886, "learning_rate": 2.5251433707351723e-10, "loss": 2.6426, "step": 48283 }, { "epoch": 2.9973306847104104, "grad_norm": 0.33633521290752655, "learning_rate": 2.4116685274355733e-10, "loss": 2.7324, "step": 48284 }, { "epoch": 2.9973927618101683, "grad_norm": 0.32796391438576383, "learning_rate": 2.3008022929094097e-10, "loss": 2.676, "step": 48285 }, { "epoch": 2.9974548389099263, "grad_norm": 0.3274468157259126, "learning_rate": 2.1925446677673045e-10, "loss": 2.6689, "step": 48286 }, { "epoch": 2.997516916009684, "grad_norm": 0.3251497162249983, "learning_rate": 2.086895652564369e-10, "loss": 2.6757, "step": 48287 }, { "epoch": 2.997578993109442, "grad_norm": 0.32035949408819175, "learning_rate": 1.983855247855715e-10, "loss": 2.6647, "step": 48288 }, { "epoch": 2.9976410702092, "grad_norm": 0.32425212688307314, "learning_rate": 1.8834234541964534e-10, "loss": 2.6621, "step": 48289 }, { "epoch": 2.9977031473089575, "grad_norm": 0.3159355854387691, "learning_rate": 1.7856002720861854e-10, "loss": 2.65, "step": 48290 }, { "epoch": 2.997765224408716, "grad_norm": 0.3148739054889139, "learning_rate": 1.6903857020245107e-10, "loss": 2.6332, "step": 48291 }, { "epoch": 2.9978273015084733, "grad_norm": 0.29713210467076046, "learning_rate": 1.59777974451103e-10, "loss": 2.5869, "step": 48292 }, { "epoch": 2.9978893786082317, "grad_norm": 0.3402175053860788, "learning_rate": 1.5077824000453433e-10, "loss": 2.7317, "step": 48293 }, { "epoch": 2.997951455707989, "grad_norm": 0.3364001238026374, "learning_rate": 1.4203936691270513e-10, "loss": 2.6557, "step": 48294 }, { "epoch": 2.998013532807747, "grad_norm": 0.3312312980086782, "learning_rate": 1.3356135521447321e-10, "loss": 2.6089, "step": 48295 }, { "epoch": 2.998075609907505, "grad_norm": 0.3273397654834864, "learning_rate": 1.2534420495424747e-10, "loss": 2.5988, "step": 48296 }, { "epoch": 2.998137687007263, "grad_norm": 0.3538839623715325, "learning_rate": 1.1738791618198796e-10, "loss": 2.6317, "step": 48297 }, { "epoch": 2.998199764107021, "grad_norm": 0.30971773891074167, "learning_rate": 1.0969248893655249e-10, "loss": 2.6722, "step": 48298 }, { "epoch": 2.9982618412067787, "grad_norm": 0.3238150139559945, "learning_rate": 1.0225792325679884e-10, "loss": 2.6064, "step": 48299 }, { "epoch": 2.9983239183065367, "grad_norm": 0.3343836645091112, "learning_rate": 9.508421917603372e-11, "loss": 2.7756, "step": 48300 }, { "epoch": 2.9983859954062946, "grad_norm": 0.33283785289138684, "learning_rate": 8.817137674421716e-11, "loss": 2.5037, "step": 48301 }, { "epoch": 2.9984480725060525, "grad_norm": 0.3384442286998087, "learning_rate": 8.151939598355363e-11, "loss": 2.7026, "step": 48302 }, { "epoch": 2.9985101496058104, "grad_norm": 0.3543885134587845, "learning_rate": 7.512827693845203e-11, "loss": 2.7068, "step": 48303 }, { "epoch": 2.9985722267055683, "grad_norm": 0.3477730715493817, "learning_rate": 6.899801964221908e-11, "loss": 2.6417, "step": 48304 }, { "epoch": 2.9986343038053263, "grad_norm": 0.32902792041581885, "learning_rate": 6.312862412261033e-11, "loss": 2.616, "step": 48305 }, { "epoch": 2.998696380905084, "grad_norm": 0.3433791349104418, "learning_rate": 5.752009040738138e-11, "loss": 2.7049, "step": 48306 }, { "epoch": 2.998758458004842, "grad_norm": 0.31866929117040427, "learning_rate": 5.217241853539001e-11, "loss": 2.8564, "step": 48307 }, { "epoch": 2.9988205351046, "grad_norm": 0.3054445959624735, "learning_rate": 4.7085608523289584e-11, "loss": 2.6695, "step": 48308 }, { "epoch": 2.998882612204358, "grad_norm": 0.31744205802203246, "learning_rate": 4.22596604099379e-11, "loss": 2.6195, "step": 48309 }, { "epoch": 2.998944689304116, "grad_norm": 0.34095918728027164, "learning_rate": 3.76945742119883e-11, "loss": 2.6602, "step": 48310 }, { "epoch": 2.9990067664038738, "grad_norm": 0.339450356370107, "learning_rate": 3.339034995164525e-11, "loss": 2.7157, "step": 48311 }, { "epoch": 2.9990688435036317, "grad_norm": 0.32193446628531236, "learning_rate": 2.934698765666433e-11, "loss": 2.7049, "step": 48312 }, { "epoch": 2.999130920603389, "grad_norm": 0.33763735987097676, "learning_rate": 2.5564487349249987e-11, "loss": 2.6002, "step": 48313 }, { "epoch": 2.9991929977031475, "grad_norm": 0.3176394085564083, "learning_rate": 2.204284904605558e-11, "loss": 2.7029, "step": 48314 }, { "epoch": 2.999255074802905, "grad_norm": 0.33431684665703293, "learning_rate": 1.878207276373445e-11, "loss": 2.6348, "step": 48315 }, { "epoch": 2.9993171519026633, "grad_norm": 0.3329052361605665, "learning_rate": 1.5782158518939937e-11, "loss": 2.8755, "step": 48316 }, { "epoch": 2.999379229002421, "grad_norm": 0.3314495713562207, "learning_rate": 1.3043106333876509e-11, "loss": 2.6503, "step": 48317 }, { "epoch": 2.999441306102179, "grad_norm": 0.3646149712413653, "learning_rate": 1.0564916219646392e-11, "loss": 2.7513, "step": 48318 }, { "epoch": 2.9995033832019367, "grad_norm": 0.33994876948937797, "learning_rate": 8.347588187351818e-12, "loss": 2.6825, "step": 48319 }, { "epoch": 2.9995654603016946, "grad_norm": 0.35664871891860705, "learning_rate": 6.391122248095016e-12, "loss": 2.7345, "step": 48320 }, { "epoch": 2.9996275374014525, "grad_norm": 0.3216824196176597, "learning_rate": 4.695518412978217e-12, "loss": 2.7442, "step": 48321 }, { "epoch": 2.9996896145012104, "grad_norm": 0.3142019536344022, "learning_rate": 3.260776687552536e-12, "loss": 2.6969, "step": 48322 }, { "epoch": 2.9997516916009683, "grad_norm": 0.32386962063369656, "learning_rate": 2.086897088471318e-12, "loss": 2.7293, "step": 48323 }, { "epoch": 2.9998137687007262, "grad_norm": 0.31880214535773305, "learning_rate": 1.1738796157345634e-12, "loss": 2.6531, "step": 48324 }, { "epoch": 2.999875845800484, "grad_norm": 0.3251154116969691, "learning_rate": 5.217242748933871e-13, "loss": 2.7007, "step": 48325 }, { "epoch": 2.999937922900242, "grad_norm": 0.325085698804795, "learning_rate": 1.3043107149890433e-13, "loss": 2.657, "step": 48326 }, { "epoch": 3.0, "grad_norm": 0.3313838256558564, "learning_rate": 0.0, "loss": 2.5818, "step": 48327 } ], "logging_steps": 1, "max_steps": 48327, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 12000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 5137128241496064.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }